KCNE1 and KCNE3 modulate KCNQ1 channels by affecting different gating transitions

Significance

Regulatory β-subunits associate with voltage-gated K⁺ channels to modulate their biophysical properties and physiological roles. KCNE1 and KCNE3 β-subunits turn voltage-dependent KCNQ1 channels into delayed activating KCNQ1/KCNE1 channels and apparent voltage-independent KCNQ1/KCNE3 channels, respectively, which are important for cardiac action potentials and transport of water and salts across epithelial cells. Mutations in KCNQ1/KCNE1 and KCNQ1/KCNE3 channels are associated with diseases, such as cardiac arrhythmias, congenital deafness, secretory diarrhea, and tinnitus. Therefore, KCNQ1, KCNE1, and KCNE3 are potential drug targets. We here propose a model for how KCNE1 and KCNE3 differentially modulate KCNQ1 that will allow for a better understanding of how mutations in KCNQ1, KCNE1, and KCNE3 cause diseases and how to design drugs to treat these diseases.

Abstract

KCNE β-subunits assemble with and modulate the properties of voltage-gated K⁺ channels. In the heart, KCNE1 associates with the α-subunit KCNQ1 to generate the slowly activating, voltage-dependent potassium current (I_Ks) in the heart that controls the repolarization phase of cardiac action potentials. By contrast, in epithelial cells from the colon, stomach, and kidney, KCNE3 coassembles with KCNQ1 to form K⁺ channels that are voltage-independent K⁺ channels in the physiological voltage range and important for controlling water and salt secretion and absorption. How KCNE1 and KCNE3 subunits modify KCNQ1 channel gating so differently is largely unknown. Here, we use voltage clamp fluorometry to determine how KCNE1 and KCNE3 affect the voltage sensor and the gate of KCNQ1. By separating S4 movement and gate opening by mutations or phosphatidylinositol 4,5-bisphosphate depletion, we show that KCNE1 affects both the S4 movement and the gate, whereas KCNE3 affects the S4 movement and only affects the gate in KCNQ1 if an intact S4-to-gate coupling is present. Further, we show that a triple mutation in the middle of the transmembrane (TM) segment of KCNE3 introduces KCNE1-like effects on the second S4 movement and the gate. In addition, we show that differences in two residues at the external end of the KCNE TM segments underlie differences in the effects of the different KCNEs on the first S4 movement and the voltage sensor-to-gate coupling.

Voltage-gated K⁺ (Kv) channels are mainly expressed in excitable cells, where changes in the voltage across the membrane, such as action potentials, demand rapid channel activation and deactivation. Among Kv channels, the KCNQ1 channel (also called Kv7.1 or KvLQT1) differs from most other Kv channels in that KCNQ1 plays key physiological roles in nonexcitable cells, such as in epithelia, in addition to its roles in excitable cells, such as cardiomyocytes. The KCNQ1 channels display dramatically different biophysical properties in various cell types, differences that are thought to be mainly due to the KCNQ1 channel’s ability to associate with five tissue-specific KCNE β-subunits to form different K⁺ channel complexes (1–7).

KCNQ1 subunits expressed by themselves form voltage-dependent K⁺ channels that open at negative voltages (1, 2) (Fig. 1 A and D, black squares). However, coassembly of KCNQ1 with KCNE1 (also called MinK) produces a much slower activating potassium current (I_Ks) in the heart that activates at positive voltages (Fig. 1 B and D, black triangles) and shapes the repolarization phase of cardiac action potentials (1, 2). Mutations in the KCNQ1/KCNE1 complex are linked to life-threatening cardiac arrhythmias, such as torsade de pointes (8, 9). Association of KCNQ1 with KCNE3 (also called MiRP2) produces channels that are voltage-independent in the physiological voltage range (Fig. 1 C and D, black circles) and that are crucial in regulating the transport of water and salt in several epithelial tissues, including the colon, small intestine, and airways (3, 10, 11). Therefore, the KCNQ1/KCNE3 channel could be a potential drug target to treat life-threatening diseases, such as human colonic secretory diarrhea (12, 13). How the different accessory β-subunits KCNE1 and KCNE3 mediate such different effects on the activation of KCNQ1 channels is still debated. Here, we simultaneously track changes in voltage sensor movement and in gate opening of KCNQ1/KCNE1 and KCNQ1/KCNE3 channels using voltage clamp fluorometry (VCF) to determine the differences in the molecular mechanisms by which KCNE1 and KCNE3 alter KCNQ1 channel gating.

Fig. 1.

KCNQ1 channel with and without KCNE β-subunit. Representative current (black) and fluorescence (red) traces from KCNQ1 (A), KCNQ1/KCNE1 (B), and KCNQ1/KCNE3 (C) channels for voltage steps between −200 mV (KCNE1: −180 mV) and +80 mV (KCNE3: +20 mV) in steps of 20 mV, from a holding potential of −80 mV followed by a −40-mV tail potential. In B, a prepulse to −140 mV was used to deactivate the S4s more completely. The dashed-dotted line represents zero current. (D) Normalized G(V) (black) and F(V) (red) from KCNQ1 (squares), KCNQ1/KCNE1 (triangles), and KCNQ1/KCNE3 (circles) channels (mean ± SEM; n = 5). Black and red lines are single Boltzmann fits from G(V) and F(V) of KCNQ1 and KCNQ1/KCNE3 and G(V) of KCNQ1/KCNE1; the F(V) of KCNQ1/KCNE1 was fit with a double Boltzmann equation. Black arrows indicate left (KCNQ1/KCNE3) and right (KCNQ1/KCNE1) shifts of the G(V) relative to the G(V) of KCNQ1 alone. (E) Topology of KCNQ1, KCNE3, and KCNE1. Residues mutated in this study are indicated.

Pore-forming KCNQ1 channel α-subunits form tetramers. Each α-subunit has six transmembrane (TM) segments (S1–S6) divided into two functional domains: a voltage sensor domain comprising four peripheral TM helices (S1–S4) and a centrally located pore domain (S5–S6) (14). The fourth TM segment, S4, has several positively charged amino acids and has been shown to move and function as the voltage sensor (15–19). In Kv channels, it is thought that the S4 movement upon membrane depolarization causes a lateral pull on the S4–S5 linker, which, in turn, is coupled to S6 and thereby opens the S6 gate (14, 20–22). All five KCNE regulatory β-subunits have only a single TM segment (23). Disulfide cross-linking studies suggest that KCNE1 localizes between the voltage sensor domain and the pore domain of KCNQ1 channels (24–26), such that KCNE1 could affect the voltage sensor movements, the pore, or the coupling between the voltage sensor and the pore. Because of the high homology within the KCNEs family, it is likely that KCNE3 is positioned in the KCNQ1 channel structure in a similar location. However, how KCNE1 and KCNE3 differently affect the S4 and the pore in KCNQ1 channel to produce voltage-dependent KCNQ1/KCNE1 channels and voltage-insensitive KCNQ1/KCNE3 channels is still unclear. Using VCF, we recently showed that KCNE3 shifts the equilibrium of the S4 movement of KCNQ1 to very negative voltages (Fig. 1D, red circles), thereby making KCNQ1/KCNE3 channels voltage-independent in the physiological voltage range (27) (Fig. 1D, black circles). We also showed that two negatively charged residues at the extracellular end of the KCNE3 TM segment are necessary for this voltage shift (27, 28). The effect of KCNE3 on the KCNQ1 gate is only present with an intact voltage sensor-to-gate coupling (27). We refer to effects by KCNE β-subunits that require an intact voltage sensor-to-gate coupling as indirect effects and effects that remain even in the absence of an intact voltage sensor-to-gate coupling as direct effects. We therefore concluded that the effect of KCNE3 on the KCNQ1 gate is only indirect and that the effect of KCNE3 is transmitted from the voltage sensor to the gate by the voltage sensor-to-gate coupling (27). We showed in other studies that KCNE1 shifts a component of the S4 movement of KCNQ1 to more negative voltages relative to that of KCNQ1 alone (although to a lesser extent than KCNE3) (Fig. 1D, red triangles), but shifts the gate opening to more positive voltages (29, 30) (Fig. 1D, black triangles). In KCNQ1/KCNE1 channels, the S4 movement occurs in two steps (Fig. 1D, red triangles; also see Fig. 3E, Top): The first step occurs at negative voltages and correlates with the main gating charge movement, whereas the second step occurs at positive voltages and correlates with channel opening (29). The molecular mechanism of how KCNE1 affects KCNQ1 is not completely understood. However, a recent study proposed that all of the effects of KCNE1 are on the voltage sensor-to-gate coupling in KCNQ1 (31). For example, it was shown that the KCNQ1 channel opens from an intermediate S4 position (after the first S4 step), whereas KCNQ1/KCNE1 channels open from a fully activated S4 position (31).

Fig. 3.

Decoupling S4 and gate by PIP₂ depletion: KCNE1 still left-shifts the S4 movement. (A) Representative current (black) and fluorescence (red) traces from KCNQ1/KCNE1 channels for the indicated voltage protocol. (B) Currents from KCNQ1/KCNE1 channels during PIP₂ depletion by activation of the CiVSP by stepping 18-fold to +40 mV for 5 s from a holding voltage of −80 mV, followed by a pulse to −40 mV. (C) Representative current (black) and fluorescence (blue) traces from KCNQ1/KCNE1 channels after activation of VSP. In A and C, currents are only shown for voltage steps down to −160 mV. (D) Normalized G(V) and F(V) in the presence of VSP after PIP₂ depletion (filled black and blue squares) compared with G(V) and F(V) in the absence of VSP (open black and red circles) (mean ± SEM; n = 5). The midpoints of the first component of the F(V) fits of KCNQ1/KCNE1 with (red) and without (blue) PIP₂ are −103.01 ± 4 mV and −102.8 ± 3 mV, respectively (n = 8). Dashed lines show the G(V) (black) and F(V) (red) curves of KCNQ1 expressed alone for comparison. (E) Cartoon showing that even in the case where S4 movement is decoupled from gate opening by removing PIP₂ (represented by the interrupted connection between S4 and S6), KCNE1 still shifts the first S4 movement to very negative voltages. S4 (red), KCNE1 (pink), S6 (purple), and Alexa Fluor 488 5-maleimide (green) are illustrated. For simplicity, only two of the four subunits in the tetrameric channel are shown.

Using VCF, we here find that KCNE1 affects both the S4 movement and the gate, independent of the S4-to-gate coupling, in contrast to KCNE3, which primarily affects the S4 segment and only affects the gate indirectly through the S4-to-gate coupling.

Results

KCNE1 Shifts both the Equilibrium of S4 Movement and Gate Opening, Whereas KCNE3 only Shifts the S4 Movement.

Using VCF, we simultaneously monitor gate opening (by ionic current) and S4 movement (by fluorescence) in KCNQ1, KCNQ1/KCNE1, and KCNQ1/KCNE3 channels by labeling a cysteine introduced at position 219 close to the S4 of KCNQ1 with Alexa Fluor 488 5-maleimide (32) (Fig. 1). We will refer to the labeled KCNQ1 subunit simply as KCNQ1. We use the mutation F351A (Fig. 1E) in KCNQ1 to elucidate the differences in the effects of KCNE1 and KCNE3 on the S4 movement and the gate. As we have shown earlier (27), the F351A mutation separates the fluorescence versus voltage [F(V)] curve and the conductance versus voltage [G(V)] curve by 47 mV (Fig. 2B, compare red and black triangles). Because the mutation F351A separates the fluorescence change (due to S4 movement) from the ionic current change (due to movement of the gate) in both time and voltage (30, 31, 33), F351A makes it easier to distinguish effects of the different KCNE β-subunits on the S4 movement and/or the gate. We have previously shown (27) that KCNE3 does not modify the G(V) relation of KCNQ1-F351A (Fig. 2B, compare black circles and triangles), but left-shifts a large fraction of the F(V) relation of KCNQ1-F351A (Fig. 2B, red circles and red arrow). Note that the F(V) relation of KCNQ1-F351A/KCNE3 channels exhibits two components: one fluorescence component that is similarly shifted to negative voltages as the F(V) in KCNQ1/KCNE3 channels and a second fluorescence component at positive voltages that concurs with the G(V) relation of KCNQ1-F351A/KCNE3 channels (Fig. 2B). KCNE3 apparently affects S4 movement preceding the opening of the gate, but not gate opening or a minor component of S4 movement associated with gate opening in KCNQ1-F351A (Fig. 2B).

Fig. 2.

KCNE1 shifts both S4 and gate, and KCNE3 shifts only the S4 movement in F351A mutants. Representative current (black) and fluorescence (red) traces from KCNQ1-F351A/KCNE3 (A; Q1-F351A/E3) and KCNQ1-F351A/KCNE1 (C; Q1-F351A/E1) channels for the indicated voltage protocol are illustrated. Currents are only shown for voltage steps down to −140 mV in A and −160 mV in C. The dashed-dotted line represents zero current. (B) Normalized G(V) (black circles) and F(V) (red circles) of KCNQ1-F351A/KCNE3 channels (mean ± SEM; n = 7). Fig. 2B reprinted with permission from ref. 27. (D) Relative G(V) (black squares) and F(V) (red squares) of KCNQ1-F351A/KCNE1 (mean ± SEM; n = 5). The G(V) is scaled so that the slope of the G(V) is similar to the slope of the G(V) of KCNQ1-F351A/KCNE3 in B, and the F(V) is scaled to make the first component similar to the F(V) of KCNQ1-F351A/KCNE3 in B. For comparison, the G(V) (black) and F(V) (red) curves of KCNQ1/KCNE3 (Q1/E3) channels (dashed-dotted lines) and KCNQ1-F351A (Q1-F351A) channels expressed alone (triangles) are shown. Black and red arrows indicate the shifts in the G(V) and F(V), respectively, relative to those of KCNQ1-F351A alone.

We here similarly test whether KCNE1 affects the S4 movement, the gate, or both in the mutant F351A, which clearly separates the S4 movement from the ionic currents. In contrast to KCNE3, KCNE1 has effects on both the F(V) and the G(V) in KCNQ1-F351A channels (Fig. 2 C and D). KCNE1 right-shifts the G(V) (Fig. 2D, black arrow) and left-shifts the main component of the F(V) (Fig. 2D, red arrow) compared with homomeric KCNQ1-F351A channels. Because KCNE1 shifts the G(V) and the small component of the F(V) to the right and the main component of the F(V) to the left, this suggests that KCNE1 has multiple effects on KCNQ1 channels. Three alternatives may explain the effects of KCNE1: (i) KCNE1 directly affects both S4 and the gate of KCNQ1 channels, but with opposite effects on their voltage dependences, and that the effect on the gate indirectly shifts the second, smaller F(V) component; (ii) KCNE1 directly shifts the main component of the F(V) to the left and the second component of the F(V) to the right, and that the shift of the second component of the F(V) then indirectly shifts the voltage dependence of the gate opening; or (iii) the shifts of the F(V) and G(V) relations of KCNQ1/KCNE1 channels relative to KCNQ1 alone (Fig. 1D) are due to indirect effects on S4 movement and the gate caused by effects of KCNE1 on the voltage sensor-to-gate coupling in KCNQ1, as suggested recently (31).

KCNE1 Acts Directly on the Voltage Sensor.

We test these alternative possibilities by depleting phosphatidylinositol 4,5-bisphosphate (PIP₂) in KCNQ1/KCNE1 channels. PIP₂ depletion has been suggested to decouple S4 movement from gate opening in KCNQ1 channels because after PIP₂ has been depleted, S4 movement still occurs but without opening the gate (34). Therefore, by depleting PIP₂, we can measure any potential direct effect of KCNE1 on S4. We previously used this approach to show that KCNE3 directly affects the S4 movement in KCNQ1 (27).

We coexpress KCNQ1 and KCNE1 with the voltage-sensing phosphatase from Ciona intestinalis (CiVSP) that, when activated, can deplete PIP₂. We record the ionic current during (Fig. 3B) and the ionic current and fluorescence after (Fig. 3C) PIP₂ has been depleted by repeated depolarizing pulses to +40 mV to activate CiVSP and compare these with the ionic current and fluorescence from cells expressing only KCNQ1/KCNE1 channels (Fig. 3A). Activation of CiVSP eliminates KCNQ1/KCNE1 ionic current, but does not eliminate the fluorescence change from the KCNQ1/KCNE1 voltage sensors (compare Fig. 3 A and C, red and blue traces). The F(V) relations from KCNQ1/KCNE1 channels are similar in the presence and absence of PIP₂: The first components of both F(V) relations are left-shifted to a similar extent compared with the F(V) relation of KCNQ1 channels alone (Fig. 3D).

We also measured the effect of depleting PIP₂ in KCNQ1-F351A/KCNE1 channels. Similar to wild-type (wt) KCNQ1/KCNE1, activation of CiVSP by a train of depolarizing pulses to +40 mV eliminates the ionic current of KCNQ1-F351A/KCNE1 channels, but does not eliminate the fluorescence changes reporting on S4 movement (Fig. S1 A–C). The KCNE1-induced left shift of the first fluorescence component and right shift of the second fluorescence component in the KCNQ1-F351A mutant are maintained after PIP₂ depletion (Fig. S1D, red and blue arrows, respectively).

Fig. S1.

Decoupling S4 and gate by PIP₂ depletion in KCNQ1-F351A/KCNE1 channels: KCNE1 still shifts the S4. (A) Representative current (black) and fluorescence (red) traces from KCNQ1-F351A/KCNE1 channels for the indicated voltage protocol. (B) Currents from KCNQ1-F351A/KCNE1 channels during activation of the VSP by stepping 10-fold to +40 mV for 5 s from a holding voltage of −80 mV, followed by a pulse to −40 mV. (C) Representative current (black) and fluorescence (blue) traces from KCNQ1-F351A/KCNE1 channels after activation of VSP. In A and C, currents are only shown for voltage steps down to −160 mV. (D) Relative G(V) and F(V) in the presence of CiVSP after PIP₂ depletion (filled black and blue circles) compared with G(V) and F(V) in the absence of CiVSP (open black and red squares (from Fig. 2D) (mean ± SEM; n = 7). In the presence of CiVSP, the G(V) is scaled using the current amplitude of KCNQ1-F351A/KCNE1 in the absence of CiVSP in Fig. 2D, and the F(V) is scaled to make the first component similar to the F(V) of KCNQ1-F351A/KCNE1 in the absence of CiVSP in Fig. 2D. For comparison, the G(V) (open black triangles) and F(V) (open red triangles) of KCNQ1-F351A (Q1-F351A) channels expressed alone are shown. Black and red arrows indicate the shifts in the G(V) and F(V), respectively, relative to those of KCNQ1-F351A alone.

The persisted shifts in the F(V) relations even in the absence of PIP₂ show that KCNE1 does not act on the S4 movement indirectly by acting on the voltage sensor-to-gate coupling or through a mechanism dependent on the voltage sensor-to-gate coupling (Fig. 3E). These data also show that the second fluorescence component in KCNQ1/KCNE1 channels is not dependent on gate opening or on voltage sensor-to-gate coupling. We propose that the second fluorescence component in KCNQ1/KCNE1 channels reports on a second S4 movement that precedes and triggers channel opening, as we and others have previously suggested for KCNQ1/KCNE1 channels (30, 31) and Shaker K channels (35). Therefore, we propose that the observed KCNE1-induced right shift of the G(V) is, at least partly, due to the KCNE1-induced right shift of the second fluorescent component. All these data show that the effects of KCNE1 on the S4 movement are preserved in the absence of an intact voltage sensor-to-gate coupling and gate opening, as if KCNE1 acts directly on the S4 movement.

KCNE1 Acts Directly on the Gate.

Does KCNE1 also directly affect the gate of KCNQ1 channels? We test this by using the KCNQ1-I268A mutation that causes large constitutive currents at negative voltages (Fig. 4A, red arrow and Fig. 4E, black circles) in the absence of KCNE1 (36). At these negative voltages, KCNQ1-I268A channels have their voltage sensors (S4s) in their resting position (32). Two alternatives may explain why the I268A mutation causes channel opening at these negative voltages: The I268A mutation might (i) directly affect the gate or (ii) affect the S4-to-gate coupling, so that the gate can open even if the S4s are in their resting position. To distinguish between these alternatives, we decouple the S4 movement from the gate in KCNQ1-I268A channel by PIP₂ depletion. We record the ionic current of KCNQ1-I268A channel after PIP₂ has been depleted by CiVSP (Fig. 4 B and E, green) and compare this with the ionic current from cells expressing only KCNQ1-I268A channel without CiVSP (Fig. 4 A and E, black). In wt KCNQ1, PIP₂ depletion removes the ionic currents (i.e., closes the channel) without altering the S4 movement, as if PIP₂ depletion removes the coupling between S4 movement and channel opening (34, 37). In KCNQ1-I268A channels, PIP₂ depletion eliminates the voltage-dependent part of the currents (Fig. 4A, Inset, black arrow) without eliminating the constitutive currents (Fig. 4A, red arrow) at negative voltages (compare Fig. 4 A and B and compare Fig. 4E, black and green). These data show that the I268A mutation (in the absence of KCNE1) destabilizes the closed state relative to the open state when S4 is in the resting state and when S4 is decoupled from the gate (Fig. 4F, Top Right).

Fig. 4.

KCNE1 directly affects the gate. Representative current traces from KCNQ1-I268A channels without (A) and with (C) KCNE1 for the indicated voltage protocol are shown. Currents from KCNQ1-I268A channels without (B) and with (D) KCNE1 for the indicated voltage protocol after activation of VSP are shown. We stepped 18-fold to +40 mV for 5 s from a holding voltage of −80 mV to deplete PIP₂. (A, Inset) Tail currents during channel closing at higher magnification. Red-dotted lines represent zero current. Experiments are performed in 98 mM extracellular potassium (high K⁺) (Methods); cells are held at −80 mV and pulsed to voltages between −160 and +60 mV for 2 s, followed by 2-s pulses to −100 mV, to measure tail currents. Red arrows in A and B indicate the constitutive currents at negative voltages. (E) Normalized G(V)s from tail currents of KCNQ1-I268A (filled black circles, n = 4) and KCNQ1-I268A/KCNE1 (open blue circles, n = 5) in the presence of PIP₂ (without CiVSP) and relative G(V)s of KCNQ1-I268A (filled green squares, n = 4) and KCNQ1-I268A/KCNE1 (filled maroon squares, n = 5) after PIP₂ depletion (after activation of CiVSP). The G(V)s of KCNQ1-I268A and KCNQ1-I268A/KCNE1 after PIP₂ depletion were scaled by normalizing the currents to the maximum tail current amplitude in A and C, respectively. (F) Cartoon of KCNQ1-I268A channel with and without KCNE1 in the presence and absence of PIP₂. (Top Left) In the presence of PIP₂, KCNQ1-I268A channels are leaky when S4 is in its resting position at negative voltages, but open more at depolarized voltages. (Top Right) In the absence of PIP₂, KCNQ1-I268A channels are not voltage-dependent and are leaky at both negative and positive voltages. (Bottom Left) In the presence of PIP₂, KCNQ1-I268A/KCNE1 channels are closed when S4 is in its resting position at negative voltages, but open more at depolarized voltages. (Bottom Right) In the absence of PIP₂, KCNQ1-I268A/KCNE1 channels are closed at both negative and positive voltages. S4 (red), KCNE1 (pink), and S6 (purple) are illustrated. For simplicity, only two of the four subunits in the tetrameric channel are shown.

We therefore reason that if KCNE1 directly affects the gate of KCNQ1 channels, it might modify these constitutive currents at negative voltages in KCNQ1-I268A. Coexpressing KCNE1 with KCNQ1-I268A removes the constitutive currents of KCNQ1-I268A channels at negative voltages (i.e., KCNE1 closes the channel) (Fig. 4 C and E, blue). KCNE1 could do this either by directly acting on the channel gate or by indirectly acting on the gate through the S4-to-gate coupling. To distinguish between these alternatives, we decouple the S4 movement from the gate in KCNQ1-I268A/KCNE1 channel by PIP₂ depletion (Fig. 4 D and E, maroon) and compare the ionic current with the ionic current from cells expressing only KCNQ1-I268A channels (Fig. 4 B and E, green). Coexpression with KCNE1 closes KCNQ1-I268A channels even in the case when S4 movement is decoupled from the gate, as if KCNE1 directly affects the gate of KCNQ1 channels (Fig. 4 D and E, maroon and Fig. 4F, Bottom Right).

“KCNE1-Like” KCNE3 also Affects the Second Component of S4 Movement and the Gate.

It has previously been shown that substituting three residues in the TM segment of KCNE3 by the homologous KCNE1 residues (i.e., the triple mutation T71F-V72T-G73L, which we will refer to as KCNE3-FTL) converts apparently constitutively conducting KCNQ1/KCNE3 channels into voltage-gated channels with ionic currents similar to those in KCNQ1/KCNE1 channels (38). The mechanism underlying these changes in channel gating caused by this triple mutation is not completely understood. It is not known whether the FTL mutations affect S4 movement, gate movement, or the coupling between S4 and the gate.

We measure ionic current and fluorescence of cells expressing KCNQ1/KCNE3-FTL channels (Fig. 5 A and B) to determine whether the effects of KCNE3-FTL on voltage sensor movement and gate opening resemble the effects of KCNE3 or KCNE1. Similar to the F(V) of KCNQ1/KCNE1 channels (29) (Fig. 5B, dashed red line), the F(V) relation in KCNQ1/KCNE3-FTL channels comprises two components, with a first, major fluorescence component at more negative voltages and a second, minor fluorescence component at more depolarized voltages, in a similar voltage range as gate opening (Fig. 5B, red circles). In KCNQ1/KCNE1 channels, the first fluorescence component correlates with the gating currents (29). In KCNQ1/KCNE3-FTL, the first fluorescence component is left-shifted compared with KCNQ1/KCNE1 channels (KCNQ1/KCNE3-FTL: FV_1/2 = −127 ± 2 mV and KCNQ1/KCNE1: FV_1/2 = −103 ± 7 mV; Fig. 5B, compare red dashed and solid lines), and the F(V) relation is somewhere between that of KCNQ1/KCNE3 and KCNQ1/KCNE1 channels (Fig. 5B, red circles between red dashed-dotted and dashed lines). In contrast, the G(V) relation of KCNQ1/KCNE3-FTL channels is shifted toward that of KCNQ1/KCNE1 channels (KCNQ1/KCNE3-FTL: GV_1/2 = −15 ± 3 mV and KCNQ1/KCNE1: GV_1/2 = +26 ± 6 mV; Fig. 5B, black arrow, compare black solid and dashed lines). Similar results are obtained from coexpression of KCNE3-FTL with KCNQ1-F351A (Fig. 5 C and D): The G(V) and both components of the F(V) relation of KCNQ1-F351A/KCNE3 are right-shifted by the FTL mutation (Fig. 5D, black and blue arrows). The effect of the mutated β-subunit KCNE3-FTL on KCNQ1-F351A is very similar to the effect of KCNE1 on KCNQ1-F351A (compare dashed arrows in Figs. 2D and 5D). The KCNE3-FTL–induced right shift of the second fluorescence component in KCNQ1 and KCNQ1-F351A is maintained after PIP₂ depletion (Fig. S2, blue arrows), just as for the KCNE1-induced shifts of the second fluorescence component. These results show that the FTL residues introduce an effect on the voltage sensor movement of KCNQ1 channels, even when the coupling of the voltage sensor and the gate are eliminated by PIP₂ depletion.

Fig. 5.

Interconversion of KCNE3 into KCNE1. Representative current (black) and fluorescence (red) traces from KCNQ1/KCNE3-FTL (A; Q1/E3-FTL), KCNQ1-F351A/KCNE3-FTL (C; Q1-F351A/E3-FTL), KCNQ1-I268A/KCNE3-FTL (E; Q1-I268A/E3-FTL), and KCNQ1/KCNE3-FTLNN (G; Q1/E3-FTLNN) channels are shown. Cells are held at −80 mV and stepped between −200 and +100 mV for 5 s, followed by a pulse to −40 mV to record tail currents. Currents are only shown for voltage steps down to −120 mV (A), −160 mV (C and E), and −140 mV (G). The dashed line in E depicts zero current. Normalized G(V) (black) and F(V) (red) of KCNQ1/KCNE3-FTL (B; n = 7), KCNQ1-F351A/KCNE3-FTL (D), KCNQ1-I268A/KCNE3-FTL (F; n = 5), and KCNQ1/KCNE3-FTLNN (H; n = 5) are shown (mean ± SEM). For comparison, the G(V) (black) and F(V) (red) curves of KCNQ1/KCNE3 (Q1/E3) channels (dashed-dotted lines in B and H), KCNQ1/KCNE1 (Q1/E1) channels (dashed lines in B and H), and KCNQ1/KCNE3-FTL (Q1/E3-FTL) channels (solid thin lines in H) are shown. Thick black and red arrows indicate the shifts of the G(V) and F(V) induced by the mutations FTL (B) and NN (H), relative to the G(V) and F(V) of KCNQ1/KCNE3 (B) and KCNQ1/KCNE3-FTL (H). Thin arrows in H show the shifts induced by the mutation FTL relative to KCNQ1/KCNE3 as in B. In D, the G(V) is scaled so that the slope of the G(V) is similar to the slope of the G(V) of KCNQ1-F351A/KCNE3, and the F(V) is scaled to make the first component similar to the F(V) of KCNQ1-F351A/KCNE3. For comparison, the G(V) (black) and F(V) (red) curves of KCNQ1-F351A channels (open triangles) and KCNQ1-F351A/KCNE3 (Q1-F351A/E3) channels (solid thin lines) are shown in D. Solid arrows in D indicate the shifts of the G(V) (black), the major component of the F(V) (red), and the minor component of the F(V) (blue) induced by the mutation FTL, relative to the G(V) and F(V) of KCNQ1-F351A/KCNE3. Black and red dashed arrows indicate the shifts in the G(V) and F(V), respectively, induced by the mutated β-subunit KCNE3-FTL relative to KCNQ1-F351A. (F) Normalized G(V) from tail currents of KCNQ1-I268A/KCNE3-FTL (filled black circles; n = 6). For comparison, the G(V)s (dashed lines) of KCNQ1-I268A (gray) and KCNQ1-I268A/KCNE1 (blue) in the presence of PIP₂ (without CiVSP) and KCNQ1-I268A (green) and KCNQ1-I268A/KCNE1 (maroon) after PIP₂ depletion (after activation of CiVSP) are shown.

Fig. S2.

Second component of the S4 movement is still shifted by KCNE3-FTL when S4 and the gate are uncoupled. (A and B) Normalized G(V) and F(V) in the presence of VSP after PIP₂ depletion (filled black and blue circles) compared with G(V) and F(V) in the absence of VSP (open black and red squares) (mean ± SEM; n = 7). In A, the G(V) (black and gray dashed lines) and F(V) (red dashed and gray solid lines) curves of KCNQ1 and of KCNQ1/KCNE1 channels, respectively, are shown for comparison. In B, the G(V) (open black triangles) and F(V) (open red triangles) of KCNQ1-F351A (Q1-F351A) channels expressed alone are shown for comparison. Black and blue arrows indicate the shifts in the G(V) and the second component of the F(V), respectively, relative to those of KCNQ1 (A) and KCNQ1-F351A (B) channels alone.

Do FTL residues also introduce an effect on the gate? Coexpressing KCNE3-FTL with KCNQ1-I268A removes the constitutive currents of KCNQ1-I268A channels at negative voltages (i.e., KCNE3-FTL closes the channel) (Fig. 5 E and F, circles). KCNE3-FTL closes KCNQ1-I268A channels similar to KCNE1, as if KCNE3-FTL directly affects the gate of KCNQ1 channels (compare Figs. 4E and 5F).

Together, these results show that KCNE1-like effects on the gate and second component of the S4 movement have been introduced by the FTL triple mutation and that both of these effects are independent of an intact voltage sensor-to-gate coupling, as these effects are preserved in PIP₂-depleted conditions.

Converting KCNE3 into KCNE1.

The FTL mutation right-shifts the G(V) and the F(V) of KCNQ1/KCNE3 toward those of KCNQ1/KCNE1 channels (Fig. 5B, arrows). However, the G(V) and the F(V) of KCNQ1/KCNE3-FTL channels are still more left-shifted compared with the F(V) of KCNQ1/KCNE1 (Fig. 5B, compare solid and dashed lines). We have previously shown that the D54N-D55N mutation in KCNE3 shifts the G(V) and F(V) of KCNQ1/KCNE3 channels to more depolarized voltages by removing an electrostatic interaction between D54–D55 and the positive S4 (27). Therefore, we reason that the D54N–D55N mutation (Fig. 1E) will further shift the G(V) and F(V) curves of KCNQ1/KCNE3-FTL toward those of KCNQ1/KCNE1. The G(V) and F(V) curves of KCNQ1/KCNE3-FTL-D54N-D55N channels are shifted toward those of KCNQ1/KCNE1 channels (Fig. 5H, red and black arrows). The midpoints of the major F(V) and G(V) fits for KCNQ1/KCNE1 and KCNQ1/KCNE3-FTLNN channels are similar (the FV_1/2 and GV_1/2 of KCNQ1/KCNE3-FTLNN are −116 ± 7 mV and +8 ± 4 mV, respectively; the FV_1/2 and GV_1/2 of KCNQ1/KCNE1 are −103 ± 7 mV and +26 ± 6 mV, respectively). Therefore, neutralizing D54 and D55 at the external end of the TM segment of KCNE3 more completely converts the effect of KCNE3 into that of KCNE1.

FTL Residues Interact with F339 in S6.

Previous studies showed that mutations in the S6 of KCNQ1 mimic or restore the function of mutations in the FTL residues in KCNE1, suggesting that FTL of KCNE1 interacts with S6 of KCNQ1 channels (38–41). Especially, F339 in S6 of KCNQ1 (Fig. 1E) has been suggested to interact with the FTL residues (41). We therefore introduce the mutation F339A into KCNQ1-I268A channels (KCNQ1-I268A-F339A) to test whether the introduction of the FTL residues in KCNE3 closes the KCNQ1 gate at negative voltages (Fig. 5 E and F) by interacting with F339. KCNQ1-I268A-F339A channels have similar constitutive currents at negative voltages as KCNQ1-I268A channels (Fig. 6 A and C, black circles). We reason that if FTL residues interact with F339 to close the gate of KCNQ1, the KCNE3-FTL mutation will not be able to reduce the constitutive currents of KCNQ1-I268A-F339A channels at negative voltages. Coexpression of KCNQ1-I268A-F339A with KCNE3-FTL does not reduce the large constitutive currents at negative voltages (Fig. 6 B and C, red circles), as if FTL in KCNE3 affects the KCNQ1 gate through interactions with F339 in S6.

Fig. 6.

FTL residues interact with F339 in S6. Representative current traces from KCNQ1-I268A-F339A (A) and KCNQ1-I268A-F339A/KCNE3-FTL (B) channels for the indicated voltage protocol. Dotted lines represent zero current. Experiments are performed in 98 mM extracellular potassium (high K⁺) (Methods), and cells are held at −80 mV and pulsed to voltages between −160 and +60 mV for 2 s, followed by 2-s pulses to −100 mV, to measure tail currents. (C) Normalized G(V)s from tail currents of KCNQ1-I268A-F339A (filled black circles, n = 6) and KCNQ1-I268A-F339A/KCNE3-FTL (filled red circles, n = 5). Black arrows indicate the constitutive currents at negative voltages. For comparison, the G(V)s of KCNQ1-I268A-F339A (dashed line) and KCNQ1-I268A/KCNE3-FTL (solid line) are shown.

Converting KCNE1 Toward KCNE3.

To see whether we can also convert the effect of KCNE1 into that of KCNE3, we substitute the corresponding triple residues in KCNE1 into their KCNE3 counterparts (i.e., the triple mutation F57T-T58V-L59G, which we will refer to as KCNE1-TVG) in an effort to remove the effect of KCNE1 on the second fluorescence component and the gate (Fig. 7 A and B). As previously reported (40), the triple mutation KCNE1-TVG fails to completely convert KCNQ1/KCNE1 channels into constitutively open KCNQ1/KCNE3-like channels (Fig. 7 A and B, black circles). However, both the G(V) and the F(V) relations of KCNQ1/KCNE1-TVG are shifted toward those of KCNQ1/KCNE3 (Fig. 7B, black and red arrows). If we also make the double mutations G40D and K41D in KCNE1-TVG to introduce aspartate residues in KCNE1 at the corresponding positions of D54 and D55 in KCNE3 (Fig. 1E), the G(V) is further shifted toward that of KCNQ1/KCNE3 (Fig. 7 C and D, black circles and black arrow) and the separation of the F(V) and G(V) is decreased and approaching that in KCNQ1 (Fig. 1D) and KCNQ1/KCNE3 channels (Fig. 7D, correspondence of black and red circles). KCNE1 restricts KCNQ1 opening so that KCNQ1/KCNE1 channels open only from the fully activated S4 conformation (i.e., after the second S4 movement), whereas KCNQ1 expressed alone is thought to open from an intermediate S4 conformation (31). However, the G40D/K41D mutation allows gate opening at voltages more negative than those necessary for reaching the fully activated S4 conformation (Fig. 7D). This suggests that these mutations in KCNE1 alter the voltage sensor-to-gate coupling, thereby allowing gate opening from an intermediate S4 conformation.

Fig. 7.

Interconversion of KCNE1 into KCNE3. Representative current (black) and fluorescence (red) from KCNQ1/KCNE1-TVG (A; Q1/E1-TVG) and KCNQ1/KCNE1-TVG-G40D-K41D (C; Q1/E1-TVGDD) channels for the indicated voltage protocol. Normalized G(V) (black circles) and F(V) (red circles) of KCNQ1/KCNE1-TVG (B; n = 5) and KCNQ1/KCNE1-TVGDD (D; n = 8) channels (mean ± SEM). For comparison, the G(V) (black) and F(V) (red) curves of KCNQ1/KCNE3 (Q1/E3) (dashed-dotted lines), KCNQ1/KCNE1 (Q1/E1) (dashed lines), and KCNQ1/KCNE1-TVG (thin lines in D) are shown. Thick black and red arrows indicate the shifts of the G(V) and F(V) induced by the mutations TVG (B) and DD (D), relative to the G(V) and F(V) of KCNQ1/KCNE1 (B) and KCNQ1/KCNE1-TVG (D). Thin arrows in D show the shifts induced by the mutation TVG relative to KCNQ1/KCNE1 as in B. (E) Representative currents from cells expressing KCNQ1/KCNE1 (Top), KCNQ1/KCNE1-TVGDD (Middle), and KCNQ1/KCNE1-G40N-K41N (Bottom) in external solutions containing 100 mM Na⁺ (gray), K⁺ (black), or Rb⁺ (red). Cells are held at −80 mV and pulsed to +60 mV for 2 s, followed by a 2-s pulse to −60 mV, to measure tail currents. For control, currents in regular 100 mM Na⁺ ND96 solution (Na⁺, gray) were measured. (F) Summary of Rb⁺/K⁺ permeability ratios calculated by comparing the tail current amplitudes (mean ± SEM; n = 7; *P < 0.05). n.s., not significant. For comparison, the Rb⁺/K⁺ permeability ratio of KCNQ1 channel is shown from an earlier study (31). (G) Proposed effects of KCNE3 and KCNE1 on different domains or residues of KCNQ1 and the functional consequences of these interactions.

To test whether KCNE1-TVGDD alters the voltage sensor-to-gate coupling, we measure the permeability ratios of K⁺ and Rb⁺ for KCNQ1/KCNE1 and KCNQ1/KCNE1-TVGDD channels (Fig. 7 E and F). It has been previously shown that KCNQ1 channels have a higher Rb⁺/K⁺ permeability ratio than KCNQ1/KCNE1 channels (42). This higher Rb⁺/K⁺ permeability ratio in KCNQ1 channels has been suggested to be due to channel opening from an intermediate S4 position, which puts the KCNQ1 pore in a conformation that permeates Rb⁺ better than K⁺ (31). We find that KCNQ1/KCNE1-TVGDD channels have a similar Rb⁺/K⁺ permeability ratio as previously found for KCNQ1 (31) (Fig. 7E, Middle and Fig. 7F, white and hashed bars). However, KCNQ1/KCNE1-TVGDD channels have a significantly increased Rb⁺/K⁺ permeability ratio compared with KCNQ1/KCNE1 channels (Fig. 7E, Top and Middle and Fig. 7F, black and white bars), consistent with that of KCNQ1/KCNE1-TVGDD channels open from an intermediate S4 conformation. This suggests that KCNQ1/KCNE1 and KCNQ1/KCNE1-TVGDD channels open from different S4 conformations, consistent with the finding that the G40D/K41D mutation alters the voltage sensor-to-gate coupling.

To further test whether the effect of the G40D/K41D mutation is due to the removal of the G and K amino acids at positions 40 and 41 or the introduction of the two D amino acids at positions 40 and 41, we measure the effect of the mutation G40N/K41N. Similar to KCNQ1/KCNE1-TVGDD, KCNQ1/KCNE1-NN channels have a significantly increased Rb⁺/K⁺ permeability ratio compared with KCNQ1/KCNE1 channels, as if G40/K41 is necessary for the KCNE1-induced change in voltage sensor-to-gate coupling of KCNQ1 (Fig. 7E, Top and Bottom and Fig. 7F, black and gray bars).

We also measured the effects of coexpressing KCNE1-TVGDD with KCNQ1-F351A (Fig. S3) to see whether KCNE1-TVGDD acts similar to KCNE1 or KCNE3 on KCNQ1-F351A channels. The G(V) and the second component of the F(V) relation of KCNQ1-F351A/KCNE1 are both left-shifted by the TVGDD mutation (Fig. S3B, black and blue arrows) compared with KCNQ1-F351A/KCNE1 channels. The G(V) of KCNQ1-F351A/KCNE1-TVGDD is more similar to the G(V) of KCNQ1-F351A/KCNE3 than to that of KCNQ1-F351A/KCNE1 (Fig. S3B, compare black circles and black dashed-dotted lines), suggesting that the KCNE1-TVGDD acts more similar to KCNE3 than to KCNE1 on KCNQ1 channels. Furthermore, the currents from KCNQ1/KCNE3 channels and KCNQ1/KCNE1-TVGDD channels look very similar in a physiological voltage range, if the holding potential is slightly more positive (Fig. S4). However, additional mutations are clearly necessary to completely convert KCNE1 into KCNE3.

Fig. S3.

KCNE1-TVGDD mutation retains the effect of KCNE1 on the first, major component of S4 movement and reduces effects on the second S4 movement and the gate. (A) Representative current (black) and fluorescence (red) from KCNQ1-F351A/KCNE1-TVG-G40D-K41D (Q1/E1-TVGDD) channels for the indicated voltage protocol. (B) Normalized G(V) (black circles) and F(V) (red circles) of KCNQ1-F351A/KCNE1-TVGDD (n = 5) channels (mean ± SEM). For comparison, the G(V) (black) and F(V) (red) curves of KCNQ1-F351A/KCNE1 (solid lines) and KCNQ1-F351A/KCNE3 (dashed-dotted lines) are shown. Thick black and blue arrows indicate the shifts of the G(V) and the second component of the F(V) induced by the mutation TVGDD relative to the G(V) and the F(V) of KCNQ1-F351A/KCNE1 channel, respectively.

Fig. S4.

KCNQ1/KCNE1-TVGDD channels look similar to KCNQ1/KCNE3 channels in a physiological voltage range. Representative currents from KCNQ1/KCNE3 (A and B) and KCNQ1/KCNE1-TVGDD (C and D) channels for the indicated voltage protocols. Shifting the holding potential by +20 mV makes the KCNQ1/KCNE1-TVGDD channels display similar currents as KCNQ1/KCNE3 channels.

Together, these results show that the effect of KCNE1 on the major component of S4 movement persists in KCNE1-TVGDD, whereas the additional KCNE1-like effects on the second S4 movement and the gate have been removed by the TVGDD mutation.

Discussion

Our results suggest that the main difference between the effects of KCNE1 and KCNE3 on KCNQ1 is that KCNE1 affects both the voltage-sensing domain and the gate, whereas KCNE3 primarily affects the voltage-sensing domain and only indirectly affects the gate through an intact voltage sensor-to-gate coupling. Furthermore, our data suggest that three residues in the middle of the TM segments of KCNE1 and KCNE3 determine whether these KCNE β-subunits affect the gate and the second S4 movement that triggers gate opening, but does not significantly control the effects of these two KCNEs on the first, major S4 movement of KCNQ1 (Figs. 5 and 7). We also show that two residues at the external end of the TM segment of KCNE1 are necessary for the KCNE1-induced changes in voltage sensor-to-gate coupling that only allow for KCNQ1/KCNE1 channel opening from the fully activated S4 position (Fig. 7).

We propose the following model to explain our data. KCNE3 shifts the S4 movement to more negative voltages via an interaction between D54/D55 in KCNE3 and R228 in KCNQ1. The shift in the voltage dependence of S4 movement indirectly shifts the G(V) to more negative voltages via the S4-to-gate coupling (27, 28) (Fig. 7G). In contrast, KCNE1 has several effects on KCNQ1 (Fig. 7G):

• i)We propose that G40/K41 in KCNE1 interacts with residues at the top of S1 in KCNQ1 (e.g., V141; Fig. S5) and that this interaction alters the S4-to-gate coupling in KCNQ1, such that KCNQ1/KCNE1 channels only open from a fully activated S4 conformation.
• ii)In addition, the FTL residues in KCNE1 interact with F339 in KCNQ1 to reduce the open probability at negative voltages. Both of these effects would contribute to the KCNE1-induced G(V) shift to more positive voltages.
• iii)FTL residues also have an effect on the second S4 movement that shifts the second fluorescence component to more positive voltages and indirectly shifts the G(V) to more positive voltages. This effect of FTL on S4 could be direct or indirect through effects on S6 (but not via the S4–S5 linker).
• iv)KCNE1 also has an effect on the first component of S4 movement through a yet unknown mechanism.

Fig. S5.

Model of the human KCNQ1 channels based on the recent Xenopus KCNQ1 structure (43) with the proposed location of the KCNE1 TM segment. The Xenopus KCNQ1 structure is assumed to be in an uncoupled state (because of the absence of PIP₂ in the structure), with the pore closed and the voltage sensors in their active states (43). S1–S3 are shown in light gray, S4 is shown in red, the pore domain (S5–S6) is shown in white space-filled residues, and KCNE1 is shown in dark gray (only one KCNE1 is shown for simplicity). Yellow (in S6) and green (in S1) space-filled residues are KCNQ1 residues previously shown to interact or form disulfide bonds with space-filled KCNE1 residues (41, 50, 51, 54). R228 is shown as red space-filled residues.

This model of the effects of KCNE1 and KCNE3 on KCNQ1 can explain most of the data we have presented here (Fig. S6) and is consistent with the recent cryo-EM structure of the Xenopus KCNQ1 channel (43) (Fig. S5).

Fig. S6.

Schematic representation of the different effects of KCNE1 and KCNE3 β-subunits on structural domains of KCNQ1 channels. We propose the following model to explain our data. (A) KCNE3 mainly interacts with the voltage sensor S4 of KCNQ1 by an electrostatic interaction between D54 and D55 in KCNE3 and R228 in S4 of KCNQ1 (27, 28). This interaction stabilizes the active state of S4 and shifts the voltage dependence of the S4 movement to very negative voltages, making KCNQ1/KCNE3 channels appear voltage-independent (open) in the physiological voltage range. (B) In KCNE3, changing TVG into FTL (their KCNE1 counterparts) introduces an interaction with KCNQ1 that, independent of the voltage sensor-to-gate coupling (Fig. S2A), shifts the second S4 movement to positive voltages and reduces the open probability at negative voltages [both of these effects will shift the G(V) to more positive voltages], similar to the effect of KCNE1 on KCNQ1 (as shown in D). The effect of FTL is shown here as introducing an additional steric effect on S6 movement (blue hemicircle). When the S4s are fully activated, this steric hindrance is removed and the gate can open. We depict this removal of the steric hindrance to be due to a rotation of KCNE1, because KCNE1 has been shown to rotate relative to KCNQ1 between closed and open channels (54). The effect of KCNE3 on the first S4 movement is mainly retained through interactions between D54/D55 on KCNE3 and the S4 arginines. (C) Introduction of the double mutant, D54N and D55N, in KCNE3-FTL shifts the F(V) and G(V) toward those of KCNQ1/KCNE1 channels by removing the electrostatic interaction between D54/D55 and the S4 arginines. (D) In contrast, KCNE1 interacts with both S4 and S6 (gate) in KCNQ1, as described in Fig. 7G. KCNE1 splits the S4 movement in two steps. Compared with KCNQ1 expressed alone, KCNE1 shifts the first S4 movement to more negative voltages, whereas the second S4 movement is shifted to positive voltages (Fig. 1D). KCNE1 also changes the voltage sensor-to-gate coupling so that the gate only opens after the second S4 movement. The KCNE1 effect on the second S4 movement and the G(V) is partly due to the triplet of residues FTL (pink semicircle). (E) In KCNE1, changing FTL into TVG (their KCNE3 counterparts) removes part of the effect on KCNQ1 S6 and shifts the second S4 movement and the G(V) to more negative voltages. However, some effect of KCNE1 on S6 and the second S4 movement still remains (depicted here as a smaller pink triangle on KCNE1), because the G(V) and the second S4 movement are still at more positive voltages than in KCNQ1 expressed alone. In KCNQ1/KCNE1-TVG (Q1/E1-TVG) channels, the effect of KCNE1 on the first S4 movement is mainly retained. (F) Introduction of the double mutant, G40D and K41D, in KCNE1-TVG further shifts the G(V) toward that of KCNQ1/KCNE3 channels. G40D/K41D allows KCNQ1 gate opening from an intermediate S4 position, similar to what was previously proposed for KCNQ1 channels expressed alone (31). Therefore, the G40/K41 residues in KCNE1 are necessary for the previously shown KCNE1-induced change in voltage sensor-to-gate coupling in KCNQ1 gating that only allows for gate opening from the fully activated S4 position (31). We speculate that G40/K41 residues in KCNE1 interact with residues in the external end of S1 of KCNQ1 (Fig. S5) and that this interaction underlies the KCNE1-induced voltage sensor-to-gate coupling in KCNQ1 channels. S4 (red), S6 (purple), KCNE3 (blue), FTL mutation (blue hemicircle) = D54/D55, KCNE1 (pink), FTL original residues (pink hemicircle) = G40D/K41D, and Alexa Fluor 488 5-maleimide (green) are illustrated. Potassium ions (circles with +).

A previous study suggested that the effects of KCNE1 on KCNQ1 could be explained solely by effects on the voltage sensor-to-gate coupling (31). However, both KCNE1 and KCNE3 appear to directly affect the S4 movement of KCNQ1, because separation of S4 movement and gate opening by the mutation F351A or by removal of PIP₂ still allows KCNE1 and KCNE3 to induce a substantial left shift of the first component of the F(V) relation to negative voltages and allows KCNE1 to induce a substantial right shift of the second component of the F(V) relation to positive voltages. So, in addition to the previous suggested KCNE1-induced change in voltage sensor-to-gate coupling (31), KCNE1 must have an effect on the S4 movement independent of the voltage sensor-to-gate coupling. Therefore, we suggest that the observed KCNE1-induced G(V) shift is, at least partly, due to the KCNE1-induced shift of the second S4 movement (i.e., the second fluorescent component).

We show that KCNE1 eliminates the constitutive currents at negative voltages in KCNQ1-I268A channels, as if KCNE1 also directly affects the gate. In addition, when decoupling the S4 movement from the gate in KCNQ1-I268A channels by PIP₂ depletion, KCNE1 completely removes the remaining constitutive currents at negative voltages (it closes the channel), showing that KCNE1 also directly affects the gate of KCNQ1 channels independent of the voltage sensor-to-gate coupling. In contrast, KCNE3 does not modify the G(V) relation of KCNQ1-F351A channels (27) (Fig. 2B), suggesting that an intact coupling between the voltage sensor and the gate is required for KCNE3 to affect the gate in KCNQ1 (27). We therefore conclude that KCNE1 affects both the S4 movement and the gate movement of KCNQ1, whereas KCNE3 mainly affects the S4 movement and only indirectly affects the gate movement of KCNQ1 through the voltage sensor-to-gate coupling. Our findings that KCNE1 and KCNE3 modulate different regions of KCNQ1 and affect different gating transitions (i.e., KCNE3 affects the S4 segment, whereas KCNE1 affects both the S4 segment and the gate) may explain the dramatically different gating properties conferred by these two related KCNE β-subunits on KCNQ1 channels (Fig. 1).

Several previous studies have sought to determine the regions of KCNEs responsible for the effect on KCNQ1 channel function. One study suggested that the C-terminal region of KCNE1 was the main region responsible for KCNE1 modulation of KCNQ1 gating (44). In contrast, another study suggested that differences in the triplet of residues in the middle of the TM segments of KCNE1 (i.e., FTL) and KCNE3 (i.e., TVG) determine the differences in the effects of these two accessory subunits (38). However, another study concluded that the TM segment of KCNE3 plays an “active” role in KCNQ1 modulation and that the C terminus of KCNE3 is not important (45), whereas the TM segment of KCNE1 plays only a “passive” role and that the C-terminal domain of KCNE1 controls KCNQ1 channel modulation (45). We here show that deleting the C terminus of KCNE3 does not affect KCNQ1/KCNE3-FTL channel gating (Fig. S7), indicating that for a KCNQ1/KCNE1-like channel, such as KCNQ1/KCNE3-FTL, the effect of the KCNE β-subunit is not dependent on the C terminus. This is in contrast to studies showing that the C terminus is necessary for KCNE1 modulation of KCNQ1 (44, 45), but is consistent with another study showing that deletion of the C terminus had no effect (46). Note that even if our study suggests that the C terminus is not necessary for a KCNE1-like effect on KCNQ1, mutations in the C termini of KCNE1 or KCNQ1 could still have effects on KCNQ1/KCNE1 gating, as has been previously shown, for example, for LQTS mutations in these regions (47–49).

Fig. S7.

KCNE3 C terminus is not necessary for “KCNE1-like” KCNE3 modulation of KCNQ1 channels. (A) Representative current (black) and fluorescence (red) traces from KCNQ1/KCNE3-FTL/Δ84 stop channels for the indicated voltage protocol. (B) Normalized G(V) (black circles) and F(V) (red circles) of KCNQ/KCNE3-FTL/Δ84stop channels (mean ± SEM; n = 8). For comparison, the G(V) and F(V) curves of KCNQ1/KCNE3 (dashed-dotted lines) and KCNQ1/KCNE3-FTL (squares) channels are shown.

Our results are consistent with previous studies that suggested interactions between the FTL residues (F57, T58, and L59) in KCNE1 and S6 of KCNQ1 (38–40). One study showed that mutating F340 in S6 of KCNQ1 increased the speed of activation in a similar manner as in mutations of the FTL residues in KCNE1, whereas another study showed that the F340A mutation restores the function of voltage-independent KCNQ1/KCNE1 channels with an F57W mutation in KCNE1 (38–40). However, another study showed interactions between the KCNE1-T58 and KCNQ1-F339 using mutant cycle analysis (41). However, how these interactions alter the function of the KCNQ1 channel was not clear. We find that the introduction of the FTL in the TM segment of KCNE3 changes KCNQ1/KCNE3 ionic currents into KCNQ1/KCNE1-like currents mainly by introducing effects on the second S4 movement and the gate. These effects are independent of the voltage sensor-to-gate coupling (Fig. 5 and Fig. S2) and likely via F339 for the effects on the gate (Fig. 6).

Other regions of KCNEs and KCNQ1 have also been proposed to affect the properties of KCNQ1/KCNEs, for example, interactions between KCNEs and the external ends of S1, S5, and S6 (50–52). KCNQ1/KCNE1 normally only opens from the fully activated S4 conformation, whereas KCNQ1 expressed alone opens from an intermediate S4 conformation (31). We here show that mutating two residues at the external end of the KCNE1 TM segment allows gate opening at voltages more negative than those necessary for reaching the fully activated S4 conformation (Fig. 7 D–F). This suggests that these mutations alter the voltage sensor-to-gate coupling, thereby allowing gate opening from an intermediate S4 conformation. Our rubidium/potassium permeability experiments further suggest that these mutations in KCNE1 alter the voltage sensor-to-gate coupling of KCNQ1 channels. It is not known how KCNE1 normally alters the voltage sensor-to-gate coupling, thereby allowing KCNQ1/KCNE1 to only open from a fully activated S4 conformation, whereas KCNQ1 expressed alone opens from an intermediate S4 conformation (31). A recent study showed that mutations at the external end of S1 in KCNQ1 also alter the voltage sensor-to-gate coupling in a similar manner as the G40/K41 mutations, thereby allowing gate opening from an intermediate S4 position (53). Residues at the external ends of the KCNQ1 S1 and KCNE1 TM segments have been shown to be close, because when substituted for cysteines, they can form disulfide bonds (41, 50, 51, 54). We therefore propose that interactions between the external ends of the KCNQ1 S1 and KCNE1 TM segments change the voltage sensor-to-gate coupling by altering interactions between the voltage-sensing domain and pore domain, thereby allowing the gate to open from different S4 conformations (31).

In summary, our results suggest that the main difference between the effects of KCNE1 and KCNE3 on KCNQ1 gating is that KCNE1 affects both the voltage-sensing domain and the gate of KCNQ1, whereas KCNE3 primarily affects the voltage-sensing domain of KCNQ1 and only indirectly affects the gate. One molecular difference between KCNE1 and KCNE3 is in a triplet of residues in the middle of the KCNEs’ TM segment that adds an effect on the second, minor S4 movement and the gate of KCNQ1. Another molecular difference is in two residues at the external end of the KCNE TM segments that shift the S4 movement to very negative voltages in the presence of KCNE3 and that alter the coupling between S4 and the gate in the presence of KCNE1. This study shows that KCNE1 and KCNE3 affect KCNQ1 in different structural domains. This knowledge may help future specific drug designs to target KCNQ1-, KCNE1-, and KCNE3-related channelopathies.

Methods

Molecular Biology.

Mutations in human KCNQ1, KCNE3, and KCNE1 were introduced using the QuikChange Site-Directed Mutagenesis Kit (Qiagen). In vitro transcription of cRNA was performed using the mMessage mMachine T7 RNA Transcription Kit (Ambion).

VCF Recordings.

VCF experiments were carried out as previously reported (30). Fifty nanograms of KCNQ1 RNA and 10 ng of KCNE3 or KCNE1 RNA (or their mutated versions) were injected into Xenopus laevis oocytes. VCF experiments were performed 2–7 d after injection: Oocytes were labeled for 30 min with 100 μM Alexa Fluor 488 5-maleimide (Molecular Probes) in high K⁺ ND96 solution [100 mM KCl, 1.8 mM CaCl₂, 1 mM MgCl₂, 5 mM Hepes (pH 7.6) with NaOH] at 4 °C. Following labeling, they were kept on ice to prevent internalization of labeled channels. Oocytes were placed into a recording chamber animal pole “up” in nominally Ca²⁺-free solution [96 mM NaCl, 2 mM KCl, 2.8 mM MgCl₂, 5 mM Hepes (pH 7.6) with NaOH] or in high K⁺ ND96 solution. One hundred micromolar LaCl₃ is used to block endogenous hyperpolarization-activated currents. At this concentration, La³⁺ does not affect G(V) or F(V) curves from KCNQ1 (30).

Data Analysis.

Steady-state voltage dependence of current was calculated from exponential fits of tail currents following different test potentials. Tail currents are measured at −40 mV following five test pulses to voltages between −180 mV and +60 mV (or as specified for each figure). For experiments in high K⁺ solution (Figs. 4 and 6), tail currents are measured at −100 mV following a family of test pulses to voltages between −160 and +60 mV. For Rb⁺/K⁺ permeability experiments (Fig. 7 E and F), tail currents are measured at −60 mV following a test pulse to +60 mV. The Rb⁺ ND96 solution contained 96 mM RbCl, 4 mM KCl, 1.8 mM CaCl₂, 1 mM MgCl₂, and 5 mM Hepes (pH 7.6) with NaOH. Each G(V) experiment was fit with a Boltzmann equation:

$$\mathrm{G}\left(\mathrm{V}\right)=\mathrm{A}0+\left(\mathrm{A}1-\mathrm{A}0\right)/\left(1+\exp \left(\left(\mathrm{V}-{\mathrm{V}}_{1/2}\right)/\mathrm{K}\right)\right),$$

where A0 and A1 are the minimum and maximum, respectively; V_1/2 is the voltage at which there is half-maximal activation; and K is the slope. Data were normalized between the A0 and A1 values of the fit. Fluorescence signals were bleach-subtracted, and data points were averaged over tens of milliseconds at the end of the test pulse to reduce errors from signal noise. The steady-state fluorescence data are fit with a single (or double) Boltzmann curve and normalized between the minimum and maximum fluorescence for each experiment.

Statistics.

All experiments were repeated four or more times from at least four batches of oocytes. Pairwise comparisons were achieved using a Student’s t test, and multiple comparisons were performed using an ANOVA with a Tukey’s test. Data are represented as mean ± SEM, and n represents the number of experiments.