Report

K2P channel gating mechanisms revealed by structures of TREK-2 and a complex with Prozac

See allHide authors and affiliations

Science  13 Mar 2015:
Vol. 347, Issue 6227, pp. 1256-1259
DOI: 10.1126/science.1261512

A sensitive regulator of cellular potassium

A class of potassium channels called K2P channels modulates resting membrane potential in most cells. The channels are regulated by multiple ligands, including the antidepressant drug Prozac, as well as factors such as mechanical stretch and voltage. Dong et al. determined the structure of the human K2P channel, TREK-2, in two conformations and bound to a metabolite of Prozac. The structures show how ligand binding or mechanical stretch might induce switching between the states. Although both states have open channels, one appears primed for gating. A Prozac metabolite binds to the primed state and prevents conformational switching. K2P channels are not a target of Prozac, but their inhibition may contribute to side effects.

Science, this issue p. 1256

Abstract

TREK-2 (KCNK10/K2P10), a two-pore domain potassium (K2P) channel, is gated by multiple stimuli such as stretch, fatty acids, and pH and by several drugs. However, the mechanisms that control channel gating are unclear. Here we present crystal structures of the human TREK-2 channel (up to 3.4 angstrom resolution) in two conformations and in complex with norfluoxetine, the active metabolite of fluoxetine (Prozac) and a state-dependent blocker of TREK channels. Norfluoxetine binds within intramembrane fenestrations found in only one of these two conformations. Channel activation by arachidonic acid and mechanical stretch involves conversion between these states through movement of the pore-lining helices. These results provide an explanation for TREK channel mechanosensitivity, regulation by diverse stimuli, and possible off-target effects of the serotonin reuptake inhibitor Prozac.

Two-pore domain potassium (K2P) channels contribute to the background leak potassium currents in nearly all cells and exhibit versatile, polymodal patterns of regulation. This functional diversity contributes to regulation of the resting membrane potential in many excitable and nonexcitable tissues. K2P channels represent important clinical targets for the treatment of cardiovascular disease and several neurological disorders, including pain and depression (1).

The archetypal polymodal K2P channels TREK-1 and TREK-2 are regulated by physical factors such as mechanical stretch, voltage, and temperature; by natural ligands including polyunsaturated fatty acids such as arachidonic acid (AA); and by intra- and extracellular pH (pHint and pHext) (13). Their activity can also be modulated by diverse pharmacological agents such as volatile anesthetics, neuroprotective drugs, and antidepressants such as fluoxetine (Prozac) (16). Such diverse regulation allows these channels to couple cellular electrical activity to a variety of signaling pathways; consequently, they represent important pharmacological targets (6). In particular, TREK channels are inhibited in vitro by fluoxetine and its active metabolite norfluoxetine at physiologically relevant concentrations (4, 5, 7). This selective serotonin reuptake inhibitor is used in the treatment of a range of depressive and anxiety disorders. In addition to its principal effect of directly inhibiting serotonin transporters, fluoxetine also inhibits several G protein–coupled receptors and ion channels (4, 8). TREK-1 knockout mice appear resistant to depression, suggesting that TREK channel inhibition by fluoxetine may contribute to its antidepressant effects (5, 8). Inhibition of TREK channels in the cardiovascular system may also contribute to some of the drug’s known side effects (9). Norfluoxetine is a state-dependent blocker of TREK channels (4) and is used here as a tool compound to probe the structural basis of TREK channel inhibition.

The molecular and structural mechanisms that allow K2P channels to sense such diverse stimuli are poorly understood. Structures of two members of the K2P channel family (TRAAK and TWIK-1) reveal that they share many basic structural features with classical tetrameric K+ channels but assemble as dimers with a pseudotetrameric pore (1012). Also, they do not appear to gate via constriction of the cytoplasmic entrance to the pore. Instead, this lower part of the conduction pathway remains open even when the channel is closed, and gating occurs primarily within the selectivity filter (2, 13, 14). However, the mechanisms that relay regulatory stimuli to the pore, and how drugs modulate this process, remain unclear.

To understand the mechanisms of polymodal K2P channel gating and inhibition by drugs, we solved the crystal structure of human TREK-2 in two conformations at 3.4 and 3.9 Å resolution (15) (figs. S1 and S2 and table S1). The truncated protein used for crystallization retains many functional properties exhibited by wild-type TREK-2, including activation by stretch and AA and inhibition by norfluoxetine (fig. S3). The two TREK-2 structures show the classic K2P channel fold (1012), with four transmembrane helices (M1 to M4), two pore-forming regions per chain (P1 and P2), and an extracellular cap domain (Fig. 1, A and B). TREK-2 exhibits the domain swap seen in TRAAK (10) (fig. S4).

Fig. 1 Crystal structures of two conformations of TREK-2.

TREK-2 overall fold viewed parallel to the plane of the membrane, with views of (A) the up-state 3.4 Å structure, with chains A and B in yellow and purple; (B) the down-state 3.9 Å structure, with chains A and B in orange and blue; (C) a superposition of the states; and (D) the superposition viewed from the cytoplasmic side of the membrane. Potassium ions are colored green and the cap disulfide bond in red. (E) Fenestration (indicated with an arrow) in the down state, with the aliphatic chain that binds in the vestibule and the fenestration (Lip3) and lipids (Lip1, Lip2) on the channel surface (green sticks). (F) View of the up state, in the same orientation as in (E), showing absence of the fenestration due to elevation of M4 and insertion of the side chains of Phe316 and Leu320 into the fenestration. The hinge point around Gly312 in M4 is indicated by an open triangle. N, N terminus; C, C terminus; CAP, extracellular CAP domain; TMD. transmembrane domain.

Differences between the two conformations are centered around the lower sections of the M2, M3, and M4 helices (Fig. 1, C and D; fig. S5; and movie S1). In the 3.9 Å structure, these regions project further into the cytoplasm (the “down state”), whereas in the 3.4 Å structure, they move further up into the membrane (the “up state”). In both structures there are two copies of the dimer in the asymmetric unit, and each of the four chains adopts a similar conformation, despite different crystal-packing arrangements (fig. S6), suggesting that the two conformations are not imposed by crystal packing. Profiles of the inner pore show that the cytoplasmic entrance to the vestibule remains open in both conformations (Fig. 1D and fig. S5, D to F). In the up state, M4 is kinked, with a hinge around a conserved glycine (Gly312). In addition, there are hinges in M2 at Gly201/Gly206 and in M3 at Gly248, allowing movement of all three helices (Fig. 1F). The down state is similar to that previously observed in the K2P channels TRAAK and TWIK-1 (1012). We observe substantial movement of all three helices (M2, M3, and M4) between the two conformations (movie S1 and fig. S5). While this manuscript was under review, two studies of TRAAK also observed movement of M4 in both chains (16, 17) or of M2 in one of the chains (17), and these movements were associated with the regulation of channel activity. In TREK-2, we observe movement of all three helices (M2, M3, and M4) in both chains. This coordinated movement of all three helices is supported by molecular dynamics (MD) simulations of the structures within a bilayer. These simulations exhibit downward movement of M2, M3, and M4 in the up state to adopt a conformation similar to the down state, thus indicating that movement between states can occur (fig. S7).

A key feature of K2P channel structures is the intramembrane-side fenestrations just below the selectivity filter (11, 12). These are present only in the down state (Fig. 1E). In the up state, the fenestration is closed by the upward movement and rotation of M4, which places the side chains of Phe316 and Leu320 in the fenestration (Fig. 1F). In the TREK-2 and TWIK-1 down states, there is density for lipid-like molecules extending across the top of the inner vestibule below the filter (Fig. 1E and fig. S5, A and B), which may represent copurified lipids or polyethylene glycol molecules.

To form a fully conductive channel, all four K+ binding sites within the filter must be occupied (18, 19). The TREK-2 up state has electron density for four K+ ions in the filter, suggesting that this state is conductive. By contrast, the down state has density for only three ions (Fig. 1, A and B, and fig. S8), implying that it may represent a nonconductive state, even though the inner pore is open. However, the available TRAAK and TWIK-1 structures (1012, 16, 17) have similar occupancy for all four sites within the filter in both states.

To probe the functional importance of these two states, we examined TREK-2 inhibition by fluoxetine and norfluoxetine because both exhibit state-dependent inhibition of TREK channel activity via a selective interaction with the closed state (4). We solved the structure of TREK-2 in complex with norfluoxetine and brominated fluoxetine {3-[2-bromo-4-(trifluoromethyl) phenoxy]-N-methyl-3-phenylpropan-1-amine (hereafter referred to as Br-fluoxetine)} at 3.7 and 3.64 Å resolution (Fig. 2 and fig. S9). These structures were in the down state with clear peaks in anomalous difference maps for bromine in Br-fluoxetine in the fenestrations (Fig. 2A and fig. S9), unequivocally identifying the binding site for Br-fluoxetine. Norfluoxetine was also found bound within the fenestration (Fig. 2, C to E, and movie S1), but neither ligand extended into the vestibule to block the ion path directly (Fig. 2, D and E). The fenestration provides a hydrophobic environment close to the selectivity filter in which both Br-fluoxetine and norfluoxetine interact with residues Ile194 and Pro198 on M2 of chain B, Cys249 and Val253 on M3, Phe316 and Leu320 on M4, and Val276, Leu279, and Thr280 on pore helix 2 (Fig. 2E). There is some flexibility in positioning of fluoxetine derivatives (fig. S9), but this is not unexpected for a relatively low-affinity ligand. Furthermore, a mutation within the binding site [Leu320→Trp320 (L320W) on M4] (20) reduces the inhibition of TREK-2 by norfluoxetine (Fig. 2F).

Fig. 2 Norfluoxetine and Br-fluoxetine bind to TREK-2 in the down-state fenestration adjacent to the pore filter entrance.

(A) Overall fold of the TREK-2 down state shown with the 5 Å anomalous difference map for the Br-fluoxetine/TREK-2 complex shown in pink (contoured at 4.5σ), indicating the location of Br-fluoxetine in the fenestration. (B) Chemical structure of Br-fluoxetine. Norfluoxetine lacks the methyl group on the nitrogen. (C) Cross section of a surface view of TREK-2 in the down state, colored by hydrophobicity [green (most hydrophobic) through yellow to blue (least hydrophobic)]. Yellow sticks represent Lip3. (D) Complex of TREK-2 with norfluoxetine. Norfluoxetine is shown in light blue, dark blue, red, and orange for carbon, nitrogen, oxygen, and fluorine atoms, respectively. For clarity, only one enantiomer of norfluoxetine is shown in (D) and (E). (E) Norfluoxetine binding site, with chains A and B in gold and blue, respectively, and norfluoxetine colored as in (D). (F) Disruption of the binding site by the L320W mutation reduces norfluoxetine inhibition. (G) Stretch activation (–11 mmHg, red) at pHint 7.3 dramatically reduces the efficacy of norfluoxetine inhibition, as does activation by 10 μM AA (blue). Error bars in (F) and (G) denote ± SEM.

The fenestration is present only in the down state; transition to the up state closes the fenestration, thus removing the norfluoxetine binding site. This is consistent with a state-dependent block of the closed channel by norfluoxetine (4) and with the observation that in the down state there are only three K+ ions in the filter, suggesting a nonconductive state. It also predicts that the up state represents a more activated state, which does not bind norfluoxetine. To test this hypothesis, we examined the effect of channel activation on norfluoxetine inhibition. We found that activation either by membrane stretch or AA reduced subsequent norfluoxetine inhibition (Fig. 2G). The mechanisms underlying both stretch and AA activation are thought to be similar (21, 22), suggesting that the up state may represent a conformation activated by stretch and AA. Furthermore, we found that norfluoxetine markedly slows the rate of stretch activation (fig. S10A). These effects are consistent with fluoxetine preventing conversion between the two states. In addition, norfluoxetine’s binding site close to the filter gate may also help to stabilize the nonconductive state.

Movement between these two conformations provides a structural mechanism for coupling regulatory signals to the filter gate. In particular, it allows movement of M4 to be coupled to the cytoplasmic domain, which moves on and off the membrane surface in response to various regulatory signals (8, 13, 21, 23). Movement of the helices between states is also associated with reorientation of side chains between M2, M3, and M4. At the cytoplasmic end of M4, Trp326 is packed between the side chains of Met322 on M4 and Arg237 on M3 in the down state (Fig. 3, A and B). However, in the up state, a kinking and 30° rotation of M4 allows the Trp326 side chain to insert into the bilayer, moving the three helices further into the membrane (Fig. 3, A and C). Disruption of these interactions should therefore preferentially destabilize the down state, and we found that mutation of these residues (W326A, M322A, and R237A) reduced stretch activation. (Fig. 3F and fig. S10C).

Fig. 3 Concerted motion of transmembrane helices M2, M3, and M4.

(A) Superposition of the up and down states highlighting movement of M2 to M4. (B) Interactions between helices M2, M3, and M4 in the down state, involving Trp326, Arg237, and Met322. View shown is indicated by black boxed region in (A). The lipid chain that overlays this interface is shown in green. (C) View of corresponding region in the up state, showing disruption of the distal M3-M4 interface due to a shift in M4 orientation. (D and E) Interactions at the top of the M2, M3, and M4 helices, proximal to the pore in the down (D) and up (E) states viewed from the direction indicated by blue boxed region in (A). Transmembrane helix hinge points are indicated by gray spheres. (F) Mutations at the interface between M2, M3, and M4 reduce activation by membrane stretch (–11 mmHg). Error bars denote ± SEM; numbers in parentheses indicate the number of repeats. WT, wild type.

Immediately above Trp326, a set of hydrophobic residues (Phe215, Phe244, Ile245, and Tyr315) form another core of interactions between M2, M3, and M4 (Fig. 3, D and E). These side chains all rearrange between conformations (Fig. 3A). Mutations within this region also reduced stretch activation (Fig. 3F and fig. S10C). In particular, Tyr315 on M4 hydrogen bonds with the backbone carbonyl of Phe244 on M3 in the up state but shifts to Ile245 in the down state. Both the Y315A and Y315F mutations reduced stretch activation, supporting a functional role for this interaction (Fig. 3, D and E).

TREK channels are regulated by many natural lipids in addition to AA (24). We observe lipid-like density on the channel surface in grooves at the top and bottom of M3 and M4: The lower site is found only in the down state, packed against Trp326, potentially stabilizing the down state, whereas the upper site is found in both conformations (Fig. 3B). MD simulations confirm that both of these sites can accommodate lipids (fig. S11).

TREK-2 is also sensitive to changes in both intracellular and extracellular pH. Lower external pH (pHext) affects the filter gate through protonation of a conserved histidine (His156) and also the external P2-M4 linker (25). Our structures reveal that His156 is located within a solvent-accessible extracellular cavity adjacent to both the P1-M2 and P2-M4 linkers (fig. S12, A to C). It is also at the center of an extensive hydrogen bond network. Mutations within either this network or the linkers affected the response to pHext (fig. S12D), which suggests that the structural dynamics of this network may couple external stimuli to the filter gate. Intracellular acidification (pHint) also activates TREK-2, and a glutamate residue within the proximal C terminus (Glu337) is critical for this process (21). However, this region is not resolved in our structures. Furthermore, pHint activation does not reduce norfluoxetine inhibition (fig. S10B), suggesting that the mechanism underlying pHint activation may be different from that of stretch and AA activation. Nevertheless, this distal segment of M4 is still well positioned to affect movement between the up and down states, thereby integrating many regulatory pathways (8, 21).

These studies allow us to propose a gating mechanism (Fig. 4) in which movement of the pore-lining helices converts TREK-2 between two functional states. The down state represents a closed or “low-activity” state stabilized by inhibitors such as norfluoxetine binding within the fenestrations. By contrast, activation by membrane stretch or AA stabilizes a “higher-activity” up state that is insensitive to norfluoxetine (Fig. 4). It is possible that these two conformations represent the only open and closed states, but differences in filter-ion occupancy in the down state compared with TRAAK (16, 17) suggest that the filter may be able to gate independently in both conformations, perhaps with a higher open probability in the up state (Fig. 4). Our results provide a mechanism for coupling mechanical forces within the membrane to channel activity through movement of the transmembrane helices (Fig. 4), supporting recent evidence that mechanosensitive K2P channels sense force directly through interactions with lipids (2628). A related mechanism of mechanosensitivity has been proposed for the TRAAK channel (16), but precisely how such conformational changes influence the selectivity filter gate (14, 23), the relative hydration status of the inner pore (29), and whether physiological lipids can sterically occlude the pore in the down state (30) remain to be determined. Nevertheless, our results clearly demonstrate how movement of the pore-lining helices creates a mechanism for regulation of K2P channel gating by diverse stimuli and illustrate how state-dependent inhibition of TREK channels by Prozac may contribute to possible off-target effects of the drug.

Fig. 4 Model of K2P channel gating and inhibition by norfluoxetine.

The down state is shown in orange, fenestrations in yellow, and the up state in blue. Inhibitors such as norfluoxetine are represented by a red triangle-like symbol. (A) Overall scheme for K2P channel gating. Higher- and lower-activity states are shown for both conformations, as the filter may gate independently of these larger changes, though with a different probability. Conformations seen in crystal structures are indicated by asterisks. (B) Schematic of activation of TREK channels by mechanical stretch. The direct interaction of TREK-2 with lipids in the membrane may allow lateral forces to facilitate conversion from the down to the up state. (C) Association of the C-terminal domain (CTD) with the membrane. The diverse cytoplasmic C-terminal extensions of K2P channels provide an additional site for modulation of channel activity. Posttranslational modification of the CTD or protonation due to pH changes could favor association of the CTD with the membrane, thus stabilizing the more active up state.

Supplementary Materials

www.sciencemag.org/content/347/6227/1256/suppl/DC1

Materials and Methods

Supplementary Text

Figs. S1 to S13

Table S1

References (3149)

Movie S1

References and Notes

  1. Materials and methods are available as supplementary materials on Science Online.
  2. Single-letter abbreviations for the amino acid residues are as follows: A, Ala; C, Cys; D, Asp; E, Glu; F, Phe; G, Gly; H, His; I, Ile; K, Lys; L, Leu; M, Met; N, Asn; P, Pro; Q, Gln; R, Arg; S, Ser; T, Thr; V, Val; W, Trp; and Y, Tyr.
  3. Acknowledgments: The Structural Genomics Consortium is a registered charity (no. 1097737) that receives funds from AbbVie, Bayer, Boehringer Ingelheim, Genome Canada through Ontario Genomics Institute grant OGI-055, GlaxoSmithKline, Janssen, Lilly Canada, the Novartis Research Foundation, the Ontario Ministry of Economic Development and Innovation, Pfizer, Takeda, and Wellcome Trust grant 092809/Z/10/Z. M.S.P.S. and S.J.T. are supported by the UK Biotechnology and Biological Sciences Research Council and the Wellcome Trust. Atomic coordinates and structure factors have been deposited with the Protein Data Bank under accession codes 4XDJ (down state), 4BW5 (up state), 4XDL (Br-fluoxetine complex), and 4XDK (norfluoxetine complex). We declare no competing financial interests.
View Abstract

Navigate This Article