Polymodal activation of the TREK-2 K2P channel produces structurally distinct open states

Abstract

The TREK subfamily of two-pore domain (K2P) K(+) channels exhibit polymodal gating by a wide range of physical and chemical stimuli. Crystal structures now exist for these channels in two main states referred to as the "up" and "down" conformations. However, recent studies have resulted in contradictory and mutually exclusive conclusions about the functional (i.e., conductive) status of these two conformations. To address this problem, we have used the state-dependent TREK-2 inhibitor norfluoxetine that can only bind to the down state, thereby allowing us to distinguish between these two conformations when activated by different stimuli. Our results reconcile these previously contradictory gating models by demonstrating that activation by pressure, temperature, voltage, and pH produce more than one structurally distinct open state and reveal that channel activation does not simply involve switching between the up and down conformations. These results also highlight the diversity of structural mechanisms that K2P channels use to integrate polymodal gating signals.

Figure 1.

The down state can be conductive. (A, left) Structure of TREK-2 in the down state showing NFx (as colored spheres) bound within the fenestrations (Dong et al., 2015). K⁺ ions within the filter are shown in magenta. (right) Cartoon model illustrating how activation by membrane stretch reduces inhibition by NFx (red pentagon). In the unbound down state (middle), the filter gate is closed. The NFx-binding site is only found in the down state; thus, channel activation by membrane stretch (up state, blue) reduces availability of the NFx-binding site. This binary two-state model of gating assumes the down state is always closed and the up state is always open. (B) Relative position of the filter gate mutations (G167I and W306S) that directly activate TREK-2 and the Y315A mutation in TM4, which affects movement between the down and up states. (C) All three mutations produce similar increases in basal whole-cell currents recorded at 100 mV (n ≥ 20; left) and similar reductions in fold activation by membrane stretch (right); currents were recorded in inside-out patches at 0 mV before and after application of negative pressure. Error bars shown are mean ± SEM (n = 5). (D) Example traces showing inhibition by 30 µM NFx inhibition for WT and mutant channels. The voltage protocol involves a step to 0 mV from a holding potential of −80 mV, followed by a ramp from −100 to 60 mV. (E) Dose–response curves showing NFx inhibition is unaffected by the filter gate mutations (G167I and W306S) but reduced by the Y315A mutation. WT IC₅₀ 4.9 ± 0.5 µM, n = 16 compared with G167I (IC₅₀ 7.0 ± 0.6 µM, n = 10), W306S (IC₅₀ 5.0 ± 0.4 µM, n = 10), and Y315A (IC₅₀ > 30 µM, n = 8). Error bars shown are mean ± SEM (n ≥ 6). (F) Revised gating model showing that channels activated by the filter gate mutations are in the down state, whereas activation by the Y315A mutation affects movement between states to reduce NFx inhibition.

Figure 2.

Dynamic interconversion between structurally distinct open conformations. (A) Example showing WT TREK-2 currents from an inside-out patch; at pH_i 7.4, channels can be inhibited by 10 µM NFx (effect magnified within inset dashed box). Channels were then stretch activated by application of negative pressure (−11 mmHg suction to the pipette), whereupon NFx inhibition is markedly reduced. Currents were recorded from voltage steps to 0 mV from a holding potential of −80 mV. The dashed line indicates the zero current level. (B) Similar dynamic changes in NFx inhibition can also be observed for channels activated by intracellular acidification (pH_i 6.0), demonstrating that these two activated states are structurally distinct. (C) Summary of dynamic changes in NFx inhibition seen for stretch activation but not pH_i activation. (D) Repeated NFx application produces consistent inhibition of the G167I mutant channel. The red arrows indicate distinct structural states of the channel. (E) Example experiment (similar to that shown in A and B) showing a dynamic change in NFx inhibition for the G167I mutant as the channel is converted from the down state to the up by membrane stretch. (C and E) Error bars shown are mean ± SEM (n ≥ 6). (F) Cartoon illustrating conversion between different structural states of the channel. The numbers refer to states shown by the red arrows in D. (left) The G167I mutation directly activates the filter gate in the down state, and the NFx-binding site remains available, but when the same channel is subjected to membrane stretch (right), it is converted into the up state, which exhibits reduced NFx sensitivity as the result of closure of the fenestration binding site.

Figure 3.

Direct activation of the filter gate without major conformational changes. (A) Example traces from inside-out patches showing activation of TREK-2 currents upon exchange of intracellular K⁺ for Rb⁺. (B) Summary dose–response curves showing NFx inhibition of TREK-2 does not change when Rb⁺ is used as the permeant ion to directly activate the filter gate. (C) Example showing activation of whole-cell currents upon extracellular acidification. Bar chart shows the effect of the H156A and E103Q mutations that abolish pH_ext activation by interfering with the filter gating mechanism (Dong et al., 2015). (D) Dose–response curves from two-electrode voltage clamp recordings of whole-cell currents showing that neither mutation reduces NFx inhibition (WT IC₅₀ = 23.4 ± 3.9 µM, n = 10; E103Q IC₅₀ = 18.3 ± 2.3 µM, n = 7; H156A IC₅₀ 19.1 ± 2.1 µM, n = 11). Note that when applied extracellularly, higher concentrations of NFx are required to inhibit channel activity. (B–D) Error bars shown are mean ± SEM (n ≥ 6).

Figure 4.

Temperature activation involves conformational changes in the TM helices. (A) Example traces showing that temperature-dependent activation of whole-cell TREK-2 currents from 22°C (left) to 37°C (right) markedly decreases inhibition by 100 µM NFx. For comparison, the currents at 22°C are also shown in gray in the right panel. Currents were recorded by voltage protocols similar to those used in Fig. 1 D. (B) Summary dose–response curves showing reduced NFx sensitivity of whole-cell TREK-2 currents at 22°C, 37°C, and 42°C (data shown are percent inhibition of basal currents recorded at 0 mV; mean ± SEM, n ≥ 6). IC₅₀ at 22°C, 23.4 ± 3.1 µM; but at 37°C and 42°C, >100 µM. (C) A truncated version of TREK-2 (TREK2 ΔN/ΔC) that no longer responds to temperature. Fold activation represents change in current between 22°C and 37°C. (D) The NFx sensitivity of this temperature-insensitive mutant is unaffected by temperature, indicating a specific effect on WT TREK-2. (C and D) Error bars shown are mean ± SEM (n ≥ 6).

Figure 5.

Expanded gating model for TREK-2 activation. This cartoon illustrates how structurally distinct open states (e.g., O₁ and O₂) can be produced by different activatory stimuli. The model allows for the filter to gate independently in both the up and down conformations. The dotted line illustrates how the major conformational changes in the TM helices caused by, e.g., membrane stretch and temperature can also influence the filter gating mechanism. The model also shows how the channel can interconvert between different open states in response to membrane stretch.