A Non-canonical Voltage-Sensing Mechanism Controls Gating in K2P K(+) Channels

Abstract

Two-pore domain (K2P) K(+) channels are major regulators of excitability that endow cells with an outwardly rectifying background "leak" conductance. In some K2P channels, strong voltage-dependent activation has been observed, but the mechanism remains unresolved because they lack a canonical voltage-sensing domain. Here, we show voltage-dependent gating is common to most K2P channels and that this voltage sensitivity originates from the movement of three to four ions into the high electric field of an inactive selectivity filter. Overall, this ion-flux gating mechanism generates a one-way "check valve" within the filter because outward movement of K(+) induces filter opening, whereas inward movement promotes inactivation. Furthermore, many physiological stimuli switch off this flux gating mode to convert K2P channels into a leak conductance. These findings provide insight into the functional plasticity of a K(+)-selective filter and also refine our understanding of K2P channels and the mechanisms by which ion channels can sense voltage.

Graphical abstract

Figure 1

Voltage Gating Is Common within the K2P Superfamily (A–E) Current responses to a 300-ms voltage step family from −100 to +100 mV from a holding potential of −80 mV recorded from excised inside-out membrane patches in symmetrical K⁺ (120 mM K⁺_int./120 mM K⁺_ext.) expressing the indicated K2P channel with (A) TWIK-1, (B) TRAAK, (C) TREK-2, (D) TASK-3, and (E) TALK-2; the current at +100 mV after K2P channel block by 1 mM TPA is shown in green (Piechotta et al., 2011). The I-V plots indicate currents at the end of the depolarizing steps (gray circles) and the corresponding inward current (tail) amplitudes upon repolarization to −80 mV (blue circles); the insets show the time course of voltage activation and inactivation with higher time resolution; time constants (τ) are obtained with exponential fits. (F and G) Rectification coefficients (currents at +100mV/–100mV) (F) and fold change in tail current amplitudes (G) subsequent to a depolarizing pulse to +100 mV for the indicated K2P channels; data are represented as mean ± SEM. See also Figures S1 and S2.

Figure 2

The Filter Represents the Voltage Gate in K2P Channels (A) The bars represent the rectification coefficients (for WT TREK-1 channels and indicated mutations around the GFG motif in P1 and P2. (B) Sequence alignment of the filter regions for probed K2P channels (and KcsA) is shown, and the critical threonine is highlighted. (C–E) Current responses to voltage families for mutant TREK-1 (C), TRAAK (D), and TASK-3 channels (E) show loss of voltage gating for threonine mutants. (F) I-V plots for WT and mutant channels. (G and H) MD simulations performed on TRAAK show relative ion occupancies for the ion binding sites S1–S4 (G) and their change (H) upon mutation of either one (single T103C) or both positions (double T103C), leading to a loss of binding to S1 and S4. Data are represented as mean ± SEM. See also Figure S3.

Figure 3

Ion-Flux Gating Is Sensitive to the Permeant Ion Species (A) TREK-1 current responses to voltage families in symmetrical K⁺ (120 mM [K⁺]_int./120 mM [K⁺]_ext.) or with K⁺ exchanged by Rb⁺ on the intracellular, extracellular, or both sides of the membrane patches. (B) Normalized tail current amplitudes for the indicated pre-pulse potentials with 120 mM K⁺_int. (black circles) and 120 mM Rb⁺_int. (red circles); data are as mean ± SEM. (C) Circles represent I-V curves with 8 mM or 120 mM K⁺_ext. for TRAAK and TREK-2 indicating the parallel shift of voltage activation and E_rev.. (D) Cartoon depicting the check-valve ion-flux gating behavior with outward ion permeation opening the selectivity filter and inward ion movement closing the filter; the structure above the pore indicates the extracellular cap domain. (E) Voltage-evoked TRAAK currents with 120 mM K⁺_ext. and various intracellular ions (120 mM) as indicated. (F) MD simulations performed on TRAAK show the relative ion occupancies for the S1–S4 sites with Rb⁺ and Cs⁺ permeation in comparison to K⁺ permeation (black dotted line). (G) The IC₅₀ for TREK-1 channel inhibition by TPA⁺ was determined for the respective voltages in symmetrical K⁺ and with intracellular Rb⁺ replacing K⁺; data are as mean ± SEM. See also Figure S4.

Figure 4

Voltage Sensitivity Arises from the Movement of Ions within the Electric Field of the Selectivity Filter (A) TRAAK currents with intracellular Rb⁺ (120 mM [Rb⁺]_int./120 mM [K⁺]_ext.) for different potentials show a maximal P_O achieved for potentials positive to +50 mV as further depolarization does not increase tail current amplitudes. (B) Normalized tail current versus pre-pulse voltage plots for the indicated K2P channels and intracellular ions. (C) Tail current pre-pulse voltage relations as in (B) fitted to a standard Boltzmann function. The table shows the fit parameters V_1/2 ± SEM and z (= equivalent gating charge [e₀]) ± SEM for the indicated K2P channels and intracellular ions. (D) The V_1/2 of voltage activation shifts with E_rev. as shown here for tail current amplitude versus pre-pulse voltage plots for TREK-1 with different K⁺_ext.. (E) V_1/2 of voltage activation (black squares) and measured E_rev. values (gray squares) are plotted for TREK-1, TREK-2, and TRAAK channels with various extracellular and intracellular ion concentrations. See also Figures S5A and S5B. (F) Cartoon depicting the voltage sensor in K2P channels. The structure represents TRAAK (adopted from Brohawn et al., 2012) embedded in a bilayer; the electric potential drop across the selectivity filter (SF) is indicated. The inactivated SF is shown as a structurally distinct ion-depleted state (SF_ion-depleted). Upon depolarization, 3–4 ions are forced into the filter by the high-electric field (SF_ion-occupied). This highly charged state is not stable and then transforms into the permeating state (SF_conductive). See also Figure S5.

Figure 5

K2P Channel Activators Switch the Filter Gating Mode (A) TRAAK currents recorded with increasing arachidonic acid (AA) concentrations in symmetrical K⁺ from inside-out patches. (B and C) From experiments as shown in (A), I-V curves are plotted for different AA concentrations for TRAAK channels (B) and TREK-2 channels (C). (D) I-V plots for TRAAK currents before and after channel activation by 10 μM PIP₂. (E) Representative ramp measurements of TRAAK channels activated by 10 μM AA with either K⁺ or Rb⁺ as the intracellular ion. (F) Representative ramps of TRAAK channels activated by 5 mmHg negative pressure applied via the patch pipette with either K⁺ or Rb⁺ as the intracellular ion. (G) Representative ramps of TREK-2 activated by pH_int. 5.0 with either K⁺ or Rb⁺ as the intracellular ion. (H and I) TRAAK T212C currents lack activation by AA and PIP₂. (J) Fold current change (current_(+activator)/current_{(–activator)}) for TRAAK WT and TRAAK T212C channels upon application of 10 μM AA or 10 μM PIP₂ at −100 mV; data are as mean ± SEM. See also Figure S6.

Figure 6

The Selectivity Filter Senses the Electrochemical Potential (A–C) TRAAK current responses with an inverted physiological K⁺ gradient (4 mM K⁺_int./120 mM K⁺_ext.) (A), symmetrical K⁺ gradient (120 mM K⁺_int./120 mM K⁺_ext.) (B), and physiological K⁺ gradient (120 mM K⁺_int./4 mM K⁺_ext.) (C) before and after AA (10 μM) activation. (D) Bars (mean ± SEM) indicate fold increase of currents at +80 mV upon AA activation for a physiological gradient compared to a symmetrical gradient in TREK-1 and TRAAK channels. (E) Illustration of the electrochemical driving force (i.e., Δμ = V_m − E_rev.) dependence of K2P gating with a positive Δμ, leading to activation but inactivation for a negative Δμ. (F) Cartoon of the proposed mechanism of flux-coupled gating depicting that for a negative Δμ (i.e., all voltages negative to the E_rev.) the filter is ion-depleted and inactive (SF_ion-depleted). For voltages positive to E_rev., the channels start to activate as an increasing fraction of the channels populate the inactive but now ion-occupied state (SF_ion-occupied) that rapidly converts into the active outwardly permeating state (SF_ions-outward). Upon inversion of the driving force (i.e., for potentials negative to the E_rev.), the SF only transiently conducts inward currents as this state is not stable (SF_ions-inward) and finally adopts the initial, structurally distinct, ion-depleted, and inactive state (SF_ion-depleted).

Figure S1

Voltage Gating in TREK-1, TRESK, TASK-1, and TASK-2 Channels and K2P Channel Gating Kinetics, Related to Figure 1 (A–D) Current responses to a family of 300 ms voltage steps from −100 to +100 mV from and to a holding potential of −80 mV recorded from inside-out patches excised from Xenopus oocytes in symmetrical K⁺ (120 mM [K⁺]_int./120 mM [K⁺]_ext.) expressing TREK-1, (B) TRESK, (C) TASK-1 and (D) TASK-2 channels. I-V plots indicate the currents at the end of the depolarizing steps (gray circles) and inward current (tail) amplitudes upon repolarization to −80 mV (blue circles); the insets show the time course of voltage activation and deactivation with higher time resolution as indicated by the scale bars; time constants (τ) revealed by exponential fit functions. (E) Summary of time constants for voltage activation obtained for different channels and voltages. Data are represented as mean ± SEM.

Figure S2

Extracellular Divalent Cations Have No Marked Effect on Voltage Gating, Related to Figure 1 (A) I-V plots of TREK-1 currents in symmetrical K⁺ (120 mM [K⁺]_int./120 mM [K⁺]_ext.) obtained from voltage families in the presence of extracellular 3.6 mM Ca²⁺ or 3.6 mM Mg²⁺ and in the absence of divalent cations with 2 mM EDTA added. (B) Protocol as in (A) but with Rb⁺ at the intracellular side (120 mM [Rb⁺]_int./120 mM [K⁺]_ext.). (C) The I-V plots show the normalized inward current (tail) amplitudes subsequent to 300 ms pre-pulse voltage steps to the indicated voltage with Rb⁺ as the intracellular cation (120 mM [Rb⁺]_int./120 mM [K⁺]_ext.) in the presence of extracellular 3.6 mM Ca²⁺ or 3.6 mM Mg²⁺ and in the absence of divalent cations with 2 mM EDTA added as indicated. Data are represented as mean ± SEM.

Figure S3

The Intracellular Pore Entrance Is Not the Voltage Gate in K2P Channels, Related to Figure 2 (A) Current responses of Kv1.2 channels to indicated voltages with and without 1 μM THexA (tetrahexylammonium) showing the typical time course of state-dependent open channel block and the slowing of deactivation resulting in a ‘cross over’ of the tail currents. (B) Same protocol as in (A) but for TREK-1 channels indicating that TREK-1 is inhibited by THexA before voltage activation and that current deactivation is not slowed by THexA. The inset shows the time course of THexA inhibition applied on voltage-activated (at +80 mV) TREK-1 channels via a rapid application system indicating that blocker binding is much slower than activation (note the different scale bars). (C) Time course of MTS-TBAO (20 μM) modification of TREK-1 G186C channels determined at +40 mV and at −80 mV (to assess the fraction of modified channels very briefly (10 ms) stepped to +40 mV). Experiments for TREK-1 channels were performed in solutions containing Rb⁺ on the intracellular side of the membrane (120 mM [Rb⁺]_int./120 mM [K⁺]_ext.) to increase current amplitudes. (D) Cartoon depicting that the intracellular pore entrance is open and accessible to MTS-TBAO modification and THexA (QA⁺) binding in both closed and open states of TREK-1 channels. Data are represented as mean ± SEM.

Figure S4

Ion Activation Profiles for Voltage Activation in Various K2P Channels, Related to Figure 3 (A–E) Depicted voltage family evoked currents with 120 mM [K⁺]_ext. and various intracellular ions (120 mM [Na⁺], [Tl⁺], [Cs⁺], [NH₄⁺] and [Rb⁺]_int.) obtained in same patches for: (A) TREK-1, (B) TREK-2, (C) TALK-2, (D) TASK-3, (E) TRESK.

Figure S5

The V_1/2 of Voltage Activation Shifts with the E_rev., Related to Figure 4 (A) Normalized tail current amplitudes for TRAAK measured with 120 mM [Rb⁺]_int. and different extracellular K⁺ concentrations. (B) Normalized tail current amplitudes for TREK-2 channels measured with 120 mM [K⁺]_ext. and different intracellular Rb⁺ concentrations. (C) Normalized tail current amplitudes for TASK-3 channels measured in symmetrical K⁺ (120 mM [K⁺]_int./120 mM [K⁺]_ext.; black circles) and with low extracellular K⁺ (120 mM [K⁺]_int./4 mM [K⁺]_ext.; blue circles). Data are represented as mean ± SEM.

Figure S6

Arachidonic Acid Activation in TRAAK, TREK-1, and TREK-2 Channels, Related to Figure 5 (A) I-V plots obtained for TREK-1 channels in the absence and the presence of different AA concentrations. (B) Dose-response curves for AA activation of different K2P channels. EC₅₀ values of standard Hill fits are indicated. (C) TREK-1 current at −80 mV upon application of pH_int. 5 and increasing AA concentrations as indicated. Data are represented as mean ± SEM.