Norfluoxetine inhibits TREK-2 K2P channels by multiple mechanisms including state-independent effects on the selectivity filter gate

Abstract

The TREK subfamily of two-pore domain K+ (K2P) channels are inhibited by fluoxetine and its metabolite, norfluoxetine (NFx). Although not the principal targets of this antidepressant, TREK channel inhibition by NFx has provided important insights into the conformational changes associated with channel gating and highlighted the role of the selectivity filter in this process. However, despite the availability of TREK-2 crystal structures with NFx bound, the precise mechanisms underlying NFx inhibition remain elusive. NFx has previously been proposed to be a state-dependent inhibitor, but its binding site suggests many possible ways in which this positively charged drug might inhibit channel activity. Here we show that NFx exerts multiple effects on single-channel behavior that influence both the open and closed states of the channel and that the channel can become highly activated by 2-APB while remaining in the down conformation. We also show that the inhibitory effects of NFx are unrelated to its positive charge but can be influenced by agonists which alter filter stability, such as ML335, as well as by an intrinsic voltage-dependent gating process within the filter. NFx therefore not only inhibits channel activity by altering the equilibrium between up and down conformations but also can directly influence filter gating. These results provide further insight into the complex allosteric mechanisms that modulate filter gating in TREK K2P channels and highlight the different ways in which filter gating can be regulated to permit polymodal regulation.

Figure 1.

The current model for TREK channel gating and NFx-binding sites. (A) The TM helices exist in two states (up and down) but it is unclear whether opening of the filter gate requires movement to the up conformation (via route a), whether it can open independently in the down state (via route b), or even whether both options are possible. Current models also suggest that openings from the down state may result in a lower-activity channel than when it is in the up state, because many activatory mechanisms (e.g., membrane stretch) promote movement to the up state. Binding sites for NFx do not exist in the up state, and NFx binding will alter the equilibrium between these two conformations of the TM helices, but is unclear whether NFx binding is state dependent and only stabilizes the closed state of the channel. The presence of positively charged NFx bound within the inner pore may also cause direct pore block and/or allosteric effects on the filter gating mechanism itself. (B) Left: A view of the structure of TREK-2 in the down state showing NFx (as vdW spheres) bound within the fenestrations (PDB accession no. 4XDK. K⁺ ions in the filter are shown as purple spheres. Right: Expanded views of other drug-binding sites near the filter. The top panel (rotated by 90°) shows pore-helix 1 (PH1) and the position of the ML335 (orange) which does not overlap with that of TPA (green) below the filter. In the bottom panel is the inner cavity below the filter showing the predicted positions of NFx (yellow), TPA (green) and BL1249 (purple) when bound to channel. The position of pore-helix 2 (PH2) is also shown. The binding sites for all three ligands are in close proximity and exhibit partial overlap, but not with ML335. (C) Representative traces of macroscopic TREK-2 currents elicited by voltage ramps between −80 and +80 mV in giant excised patches from Xenopus oocytes measured in control solution and various bath concentrations of NFx, as indicated. (D) Similar representative traces of macroscopic TREK-2 currents showing reduced inhibition by NFx in the presence of 80 µM TPA. Block by 80 µM TPA alone shown in red. (E) NFx inhibition of TREK-2 currents at +40 mV in Xenopus oocytes on its own (IC₅₀ = 2.7 µM; h = 1.0, n = 19) and in the presence of 100 mM tetraethylammonium (TEA; IC₅₀ = 3.8 µM; h = 0.8, n = 7) or 80 µM TPA (IC₅₀ = 65 µM; h = 1.2, a = 0.05; n = 12), as indicated. (F) BL1249 activation of TREK-2 currents in Xenopus oocytes on its own (IC₅₀ = 2.5 µM; h = 1.9, n = 13) or in the presence of NFx (IC₅₀ = 9.9 µM; h = 1.4, n = 17). For comparison, the previously reported shift in the presence of 5 µM THexA (Schewe et al., 2019) is also shown as a dotted green line.

Figure 2.

Direct and allosteric interactions of NFx with TREK-2. (A) Representative traces of macroscopic TREK-2 currents elicited by voltage ramps between −80 and +80 mV in giant excised patches from Xenopus oocytes measured in control solution, in the presence of 50 µM ML335 alone, and with various added bath concentrations of NFx, as indicated. (B) Similar representative traces of TREK-2 currents in the absence (control) or presence of the activator, 1 mM 2-APB alone, and with added concentrations of NFx, as indicated. (C) Dose–response curves for NFx inhibition of TREK-2 currents on their own (IC₅₀ = 2.7 µM; h = 1.7, n = 19) or in the presence of 1 mM 2-APB (IC₅₀ = 3.8 µM; h = 0.6, n = 11) or 50 µM ML335 (IC₅₀ = 164 µM; h = 0.8, n = 7), as indicated. Note the large shift in NFx sensitivity that results from ML335 activation, but not 2-APB activation. (D) Modified gating cartoon indicating gating modes with different activities rather than distinct open/closed states. The red arrow shows that 2-APB promotes a highly active state with unaltered NFx sensitivity, suggesting NFx inhibition is not state dependent. Other factors such as membrane stretch promote formation of various NFx-insensitive up conformations.

Figure S1.

Two types of TREK-2 behavior in lipid bilayers. (A and B) Single-channel recordings of TREK-2 incorporated into a bilayer at +80mV (top trace) and −80 mV (bottom trace). The dotted line represents the closed-channel level. (C and D) Macroscopic current–voltage relationships simulated for 100 TREK-2 channels using values of single-channel open probability (P_O) and single-channel current amplitude (i) obtained from single-channel recordings of TREK-2 with standard (C) and high-P_O behavior (D). The lines are fit by hand.

Figure S2.

Comparison of single-channel kinetics of TREK channels in standard and high-P_O mode. (A) Outward currents. Top traces: Single-channel recordings of TREK-2 in the standard (top trace, P_O = 0.14) and high-P_O mode (bottom trace, P_O = 0.92) at +60 mV. Dotted line represents the closed-channel level. Bottom panels: Distributions of single-channel openings (left) and closures (right) obtained from recordings of TREK-2 in standard and high-P_O mode at +60 mV. (B) Inward currents. Top traces: Single-channel recordings of TREK-2 in standard (top trace, P_O = 0.013) and high-P_O mode (bottom trace, P_O = 0.70) at −60 mV. Dotted line represents the closed-channel level. Bottom panels: Distributions of single-channel openings (left) and closures (right) obtained from recordings of TREK-2 in standard and high-P_O mode at −60 mV. The number of exponential components in the dwell-time distributions in both A and B was determined by the least-squares method.

Figure S3.

Effects of NFx on the physical properties of the membrane. The relationship between the modulus of elasticity in the perpendicular direction of the membrane in the presence of NFx (E_⊥NFX), normalized to that in control solution (E_⊥(0)). The number of experimental values is shown above each point. The line is fit by hand.

Figure 3.

Effect of 100 µM NFx on a single TREK-2 channel. Single-channel recordings of TREK-2 in in the high-P_o gating mode at different membrane voltages between +100 mV and −100 mV in the absence (left) and presence (right) of 100 µM NFx. Dotted lines represent the closed-channel levels.

Figure 4.

Summary of the effects of NFx on TREK-2 single channels. (A) The mean single-channel conductance of TREK-2 in the absence (open circles; n = 3) and presence of 100 µM NFx (filled circles; n = 3). (B) Histograms of open single-channel current level in the absence (black line) and presence (gray line) of 100 µM NFx at +100 mV (top) and −100 mV (bottom). (C) Mean single-channel open probability of TREK-2 in the absence (open circles; n = 3) and presence of 100 µM NFx (filled circles; n = 3). (D) Mean single-channel current conductance (filled squares) and mean single-channel open probability (filled circles) in the presence of 100 µM NFx normalized to values obtained in the absence of NFx (n = 3).

Figure S4.

Voltage dependence of NFx block at depolarized potentials. Dose–response relationships were determined at different voltages for NFx inhibition of macroscopic currents in giant excised patches from oocytes expressing WT TREK-2. At saturating concentrations, relatively little voltage dependence is observed, but at depolarized potentials, a small shift is observed. The lines are fit with a Hill inhibition equation (Eq. 1) assuming a = 0. The IC₅₀ values are 4.2 µM, h = 0.72 (0 mV); 2.0 µM, h = 0.74 (+40 mV); and 1.4 µM, h = 0.79 (+80 mV).

Figure S5.

Single-channel properties in the presence of NFx. (A and B) Relative P_o does not define the efficacy of NFx inhibition. Single-channel recordings at +60 mV of a single TREK-2 channel reconstituted in a bilayer with low (A) and high P_o (B) in the absence and presence of 10 µM NFx as indicated. Dotted line represents the closed-channel level. The relative change in P_o is similar in both cases. (C–F) Properties of long closed states and bursts in the presence of NFx. Mean lifetimes (C) and relative areas (D) of three apparent long closed states observed in the presence of 100 µM NFx in single-channel recordings depicted in Fig. 3. The lines through the data are fit by hand. (E and F) The dependence of the mean burst duration and the mean interburst closure on the membrane voltage in the presence of 100 µM NFx. This shows that the increase in NFx inhibition above +60 mV is accompanied by both a decrease in the mean burst length and an increase in the mean interburst close time. The lines are fit by hand.

Figure 5.

Effect of NFx on single-channel kinetics. (A) Mean open time as a function of membrane voltage in the absence (open circles) and presence of 100 µM NFx (filled circles) in single-channel recordings depicted in Fig. 4. The lines through the data are drawn by hand. (B) Relative areas of the shortest (squares) and the second shortest (circles) closed states in the absence (open symbols) and presence (filled symbols) of 100 µM NFx. (C) Mean shortest closed time as a function of membrane voltage in the absence (open circles) and presence of 100 µM NFx (filled circles). (D) Mean second shortest closed time as a function of membrane voltage in the absence (open circles) and presence of 100 µM NFx (filled circles). (E) Distribution of open times in the absence (open bars) and presence (filled bars) of 100 µM NFx at −100 mV. (F) Distribution of open times in the absence (open bars) and presence (filled bars) of 100 µM NFx at +100 mV. (G) Distribution of closed times in the absence (open bars) and presence (filled bars) of 100 µM NFx at −100 mV. (H) Distribution of closed times in the absence (open bars) and presence (filled bars) of 100 µM NFx at +100mV.

Figure 6.

The relative charge of NFx does not contribute to its inhibitory effects. (A) Electrostatic profile of a K⁺ through the pore of the TREK-2 channel in the presence and absence of NFx. The relevant region where NFx binds just below the filter is expanded below and shows two independent calculations in the presence of NFx compared with the Apo structure. A minor increase in the barrier for K⁺ permeation is seen in the presence of two charged NFx molecules bound at their sites below the filter. (B) Single-channel recording of TREK-2 channel at −40 mV in an excised patch from HEK293 cells in either the absence or presence of an uncharged NFx derivative (10 µM desamino chloro-fluoxetine) as indicated. Dotted lines in top traces represent the closed-channel level. A similar outcome was obtained in five separate experiments. (C) Histograms of single-channel currents from recordings shown above. For clarity, both amplitude histograms were scaled to the same open-channel level. The chemical structure of desamino chloro-fluoxetine is shown on the left.

Figure S6.

NFx affects open state of WT TREK-2 channels expressed in HEK cells. (A) Single-channel recordings of TREK-2 in the excised patch at −10 mV in the absence (top trace) and presence of 10 µM NFx (bottom trace). The dotted line represents zero current level. (B) Dwell-time distributions of channel openings in the absence (black bars) and presence (red bars) of 10 µM NFx. The lines are the best fit of the data to a single exponential function.

Figure S7.

Kinetic model of gating at the selectivity filter. (A) Voltage dependence of relative areas of fast and long closed states. Three distinct regions characterized by different voltage dependence are depicted in different shades of gray. (B) Voltage dependence of mean open times, corrected for missed events. Three distinct regions of behavior at these different voltages are highlighted in shades of gray. (C) A kinetic scheme of the TREK-2 selectivity filter gate with three sets of open (O), short (C_F), and long closed states (C_S) affecting three distinct voltage regions depicted in A and B. (D–G) Voltage dependence of intrinsic mean open time (D), short closed time (E), long closed time (F), and relative areas of short and long closed times (G). The lines are a fit of the kinetic model in C to the data shown in Fig. 5 with the following parameters: A₁₂ = 5 ms⁻¹, A₂₁ = 20 ms⁻¹, α₁₂ = 0.46, α₂₁ = −0.46, A₂₃ = 21 ms⁻¹, α₂₃ = 0.95, A₃₂ = 3.5 ms⁻¹, α₃₂ = −0.95, B₂ = 4.2 ms⁻¹, β₂ = 1.3, B₋₂ = 7.2 ms⁻¹, β₋₂ = 1.6 × 10⁻⁵, B₃ = 6.2 × 10⁻³ ms⁻¹, β₃ = 1.1, B₋₃ = 7.5, β₋₃= 2.1 × 10⁻⁵, D₂ = 2.4, δ₂ =1.2 × 10⁻⁵, D₋₂ = 1.6, δ_-2=−0.17, D₃ = 1.7 ms⁻¹, δ₃ = 6.4 × 10⁻⁵, D₋₃ = 4 ms⁻¹, δ₋₃ = −0.30.

Figure 7.

Desensitization of NFx Effects on TREK-2 Y315A. (A) Single-channel recordings of TREK-2 Y315A mutant channels recorded at +40 mV in excised patches from HEK293 cells in the presence and absence of either 10 µM or 100 µM NFx before and after channel desensitization. A similar outcome was obtained in three separate experiments. (B and C) Histograms of single-channel current amplitudes obtained in control solution and in the presence of 10 µM or 100 µM NFx before and after desensitization as indicated. For clarity, all amplitude histograms were scaled to the same open-channel level.

Figure 8.

NFx effects on single TREK-2 channel properties are abolished in the presence of 100 µM ML335. Examples of single-channel recordings of TREK-2 channels at −40 mV in excised patched from HEK293 cells in the presence of 100 µM ML335 having either partial (top two traces) or maximal (bottom two traces) effect on single-channel open probability either in the absence or presence of 10 µM NFx, as indicated. Dotted lines in top traces represent the closed-channel level. Note that NFx has no effect on either P_o or γ in the presence of ML335.

Figure 9.

State-independent inhibition of TREK channels by NFx. (A) Simplified filter gating scheme indicating that NFx interacts with both the open state as well as the long-lived (C_s) and short-lived (C_F) closed states of the filter gate to produce inhibition. The data suggest that multiple (I > 1) inhibited states exist as indicated by the brackets. (B) Summary cartoon indicating the different modes of channel behavior. NFx binding within the fenestrations prevents channels from moving into the up conformation, but we also now show that NFx inhibition affects both the open and closed states to produce multiple distinct closed states. TREK-2 can also adopt a high-activity mode of gating in the down conformation (e.g., when activated by 2-APB or Rb⁺; red arrow). Other stimuli such as intracellular pH, PIP₂, and membrane stretch (red arrow) are thought to promote high-activity modes of gating by stabilization of various forms of the (NFx-insensitive) up conformation.