The Polysite Pharmacology of TREK K_2P Channels

Abstract

K_2P (KCNK) potassium channels form “back-ground” or “leak” currents that have critical roles in cell excitability control in the brain, cardiovascular system, and somatosensory neurons. Similar to many ion channel families, studies of K_2Ps have been limited by poor pharmacology. Of six K_2P subfamilies, the thermo- and mechanosensitive TREK subfamily comprising K_2P2.1 (TREK-1), K_2P4.1 (TRAAK), and K_2P10.1 (TREK-2) are the first to have structures determined for each subfamily member. These structural studies have revealed key architectural features that underlie K_2P function and have uncovered sites residing at every level of the channel structure with respect to the membrane where small molecules or lipids can control channel function. This polysite pharmacology within a relatively small (~70 kDa) ion channel comprises four structurally defined modulator binding sites that occur above (Keystone inhibitor site), at the level of (K_2P modulator pocket), and below (Fenestration and Modulatory lipid sites) the C-type selectivity filter gate that is at the heart of K_2P function. Uncovering this rich structural landscape provides the framework for understanding and developing subtype-selective modulators to probe K_2P function that may provide leads for drugs for anesthesia, pain, arrhythmia, ischemia, and migraine.

4.1. Introduction

Ion channel proteins facilitate the flow of bioelectricity that underlies the physiology of thought, movement, mood, and sensation [1]. The K_2P (KCNK) potassium channel family comprises a set of 15 members (Fig. 4.1a) of the voltage-gated ion channel (VGIC) superfamily [2] that have central roles in controlling cell excitability by producing “leak” potassium currents that are largely time and voltage-independent [3–5]. The 15 K_2P subtypes comprise six subfamilies (TREK, TWIK, TASK, TALK, THIK, and TRESK) that within each subfamily are related by sequence similarities and regulation by shared types of signals [6] (Fig. 4.1a). A diverse range of stimuli affect K_2Ps depending on the subfamily type and include physical gating commands such as pressure and temperature (TREK) [4, 7], external and internal pH (TREK, TASK, TALK, and TWIK) [4, 8], chemicals such as volatile anesthetics (TREK, TASK, THIK, TRESK) and antidepressants (TREK) [4, 9], lipids and polyunsaturated fatty acids (PUFAs) (TREK, THIK) [4, 9], and protein–protein interactions with partners such as 14–3–3, G-proteins, Protein Kinase A, and Protein Kinase C (TREK, TASK, TALK, TWIK, TRESK) [4, 9].

Fig. 4.1.

K_2P channel relationships and architecture (a) K_2P channel dendrogram. Subfamilies are indicated. Asterisks indicate structurally characterized K_2Ps. (b) K_2P subunit diagram. Pore domains 1 and 2 (PD1 and PD2), transmembrane helices (M1–M4), pore helices (P1 and P2), selectivity filters (SF1 and SF2), and CAP domain are indicated. (c and d) Cartoon diagram of the K_2P2.1 (TREK-1) structure (PDB:6CQ6) [11]. Chains are colored marine and orange. Potassium ions are purple. (c), side view, (d), cytoplasmic view. Channel elements are labeled as in “b”

K_2Ps are named for their unique architecture. Each subunit bears two pore-forming domains, PD1 and PD2, each comprising two transmembrane helices (M1–M2 and M3–M4) bridged by a pore helix (P1 and P2) and selectivity filter (SF1 and SF2) (Fig. 4.1b). K_2P subunits dimerize to create a channel in which the pore is intrinsically heterotetrameric due to sequence differences between PD1 and PD2. Structures have been determined for exemplars of five of the fifteen K_2Ps subtypes (Fig. 4.1a), K_2P1.1 (TWIK-1), [10] K_2P2.1 (TREK-1), [11] K_2P3.1 (TASK-1), [12] K_2P4.1 (TRAAK), [13] and K_2P10.1 (TREK-2) [14] (Fig. 4.1c, d). These structures reveal a common protein scaffold that defines the K_2P family.

Similar to other VGICs [15, 16], the two transmembrane segments of the pore domains form outer (M1 and M3) and inner (M2 and M4) helices that define the pore and support the pore helices and selectivity filter [10–14]. The pore helices and selectivity filter coordinate a set of four potassium ions on the channel central axis. K_2Ps have a unique structural feature, the CAP domain (Fig. 4.1b, c). This extracellular structural element forms an arch directly over the channel pore and creates the bifurcated extracellular ion pathway (EIP) from which the ions exit the channel after passing through the selectivity filter [10, 13]. The M1 helix is domain-swapped between the two subunits, but how this structural intertwining of the subunits impacts function, assembly, or biogenesis is not clear. Each subunit also bears sequences at both the N- and C-termini that are likely to be unstructured on their own but that provide sites for protein–protein interactions that impact function [4, 9].

Unlike other potassium channels, the K_2P principal gate is the selectivity filter “C-type” gate [11, 17–22]. In line with this mechanism of control, the structures of K_2P1.1 (TWIK-1), [10] K_2P2.1 (TREK-1) [11], K_2P4.1 (TRAAK), [13] and K_2P10.1 (TREK-2) [14] show that these channels lack the inner gate that is present in most other VGIC superfamily members (Fig. 4.1d). The K_2P structures have shown that the M4 helix is mobile and can adopt conformations that range between an “up” state and a “down” state [13, 14, 23–25]. The “down” state creates a fenestration just below the P2 pore helix that is open to the center of the membrane bilayer [10, 13, 24, 25]. These M4 conformational changes are linked to C-type gate control [17, 18, 25–27] but do not impede access from the intracellular side [26]. Intriguingly, the recent K_2P3.1 (TASK-1) [12] structure shows a “down” state in which M4 creates an intracellular barrier, termed the “X-gate” that appears to be a special feature of the TASK subfamily [12] and that highlights structural diversification of the M4 segment within the K_2P family.

K_2Ps have roles in a multitude of physiological responses and pathological conditions such as action potential propagation [28, 29], anesthetic responses [30, 31], microglial surveillance [32], sleep duration [33], pain, [34–36] arrhythmia [37], ischemia [30, 38, 39], cardiac fibrosis [40], depression [41], migraine [42], intraocular pressure regulation [43], pulmonary hypertension [44], acute respiratory distress syndrome [45], and cancer [46]. Despite these clear physiological roles, the pharmacology of K_2Ps is generally poor [9, 47] and has been a barrier to understanding K_2P function. The paucity of reagents to probe K_2P activity has motivated multiple efforts that have begun to define new K_2P modulators [11, 36, 48–52] and key structural aspects of K_2P channel pharmacology [11, 14, 53, 54].

K_2Ps are thought to be potential therapeutic targets for pain [47, 55, 56], anesthesia [9, 47], arrhythmia [57, 58], ischemia [59], depression [60], and migraine [9, 61]. Although there are no approved drugs that target K_2Ps specifically, recent advances in the discovery of new classes of a variety of K_2P modulators should enable elaboration of a suite of new pharmacological tools directed at this channel family [11, 12, 36, 48, 49, 51, 62–64].

4.2. The TREK Subfamily: Model Polymodal Ion Channels

The TREK subfamily comprising K_2P2.1 (TREK-1), K_2P10.1 (TREK-2), and K_2P4.1 (TRAAK) is both the most extensively studied K_2P family and the only subfamily for which structures of each subtype are known [11, 13, 14] (Fig. 4.1a). TREK channels are polymodal ion channels that respond to diverse physical and chemical gating cues including temperature, pressure, pH, and modulatory lipids [7, 65]. The sensors for these signals reside in different parts of the channel. The intracellular C-terminal tail is key to modulation by temperature [17, 18, 66–68], pressure [18, 69, 70], intracellular pH [70, 71], responses to lipids such as phosphoinositol [4, 5] bis-phosphate (PIP₂) [68, 72, 73], and control by phosphorylation [17, 74, 75]. The sensor for extracellular pH is a histidine [19, 76] located in the loop that connects the P1 helix to the CAP domain [11]. Gating cues from the extracellular pH sensor [19, 76] and the C-terminal tail [17, 18] converge on the selectivity filter C-type gate and make this channel element the nexus of signal integration and functional control [11, 17, 18, 21, 22].

TREK subfamily channels are found throughout the central and peripheral nervous system [4, 28, 29, 34, 47, 77], the eye [43], and the heart [58]. Since their discovery, TREK subfamily channels have been implicated as therapeutic targets for pain, ischemia, and depression [47, 60, 65, 78]. K_2P2.1 (TREK-1) and K_2P10.1 (TREK-2) share ~65% sequence similarity, whereas K_2P2.1 (TREK-1) and K_2P4.1 (TRAAK) share only ~40% similarity [79–81]. Most of this sequence divergence is embedded in the N- and C-terminal cytoplasmic regions. In line with the high degree of conservation in the core elements of the channel, the structures of homodimers of each of the TREK subfamily members are similar to each other [11]. The three subtypes also have been shown to heterodimerize and provide further functional diversity from this subfamily [42, 82–84]. An important consequence of heterodimerization is that it creates a channel in which the PD1 and PD2 domains are all different from each other. How these differences manifest in functional diversification remains to be defined. Understanding how the similarities and differences within the TREK subfamily contribute to function is of critical importance for developing pharmacological tools and potential therapeutics targeted toward this complex subfamily.

Both activation and inhibition of TREK channels have proposed therapeutic benefits. The ability of TREK channels to stabilize the membrane potential and reduce excitability together with their high expression in sensory neurons gives activators of this K_2P subfamily the potential to function as analgesics or anesthetics [35, 47, 85] and as agents against migraine. [42] Interestingly, the original studies of K_2P2.1 (TREK-1) knockout mice indicated that inhibition of this channel might have a role in mitigating depression [41]. Pharmacologically relevant antipsychotics have also been reported to inhibit K_2P2.1 (TREK-1) [86]. Consequently, there has been an effort to explore K_2P2.1 (TREK-1) inhibitors, such as the peptide spadin [87, 88] and small-molecule norfluoxetine [60] as new directions for treating depression. Inhibiting K_2P2.1 (TREK-1) with such agents would facilitate membrane depolarization, but how such effects could result in modulation of the mental disease remains unclear and harder to understand than the links between TREK channel activation and pain suppression.

Because TREK subfamily channels are readily studied in a variety of experimental systems from potassium transport deficient yeast [48] to Xenopus oocytes [11, 22, 52, 54, 75, 80], to transfected mammalian cells [11, 22, 75, 80, 81, 89], to reconstitution assays using purified channels [13, 27, 51, 90], this K_2P subfamily has been a key model for understanding K_2P channel biophysics in general and is leading the way in demonstrating the potential of K_2Ps as pharmacological targets [47].

4.3. The Polysite Pharmacology of TREK Channels

Structural studies of TREK K_2Ps have revealed a strikingly rich structural landscape for functional control, especially given their modest size (~70 kDa). Binding sites for small molecules are found at every layer of the protein starting from its extracellular side through the portion that interacts with the membrane bilayer inner leaflet (Fig. 4.2). This polysite pharmacology comprises four defined binding sites for small molecules or lipids: the Keystone inhibitor site [54], the K_2P modulator pocket, [11] the Fenestration site [14, 53], and the Modulatory lipid site [11]. Each offers a distinct structural environment and mechanism for controlling K_2P function.

Fig. 4.2.

Polysite model of TREK subfamily modulation. Central cartoon shows the locations of structurally defined K_2P small molecule binding sites including the Keystone inhibitor site (magenta), K_2P modulator pocket (cyan), fenestration site (green), and modulatory lipid site (grey). CAP and “C-type” SF gates are indicated. Potassium ions are shown (purple). Grey lines denote the lipid bilayer. Black boxes show the details of the individual sites. Grey boxes show modulator chemical structures

4.3.1. The Keystone Inhibitor Site: Block by Polynuclear Ruthenium Amines

The trinuclear oxo-bridged ruthenium amine ruthenium red (RuR) [91] is a polycation with many biological applications [92], including a ~50 year legacy of use as an inhibitor of diverse ion channels. RuR has been shown to inhibit three K_2P channels, two from the TREK subfamily, K_2P4.1 (TRAAK) [93, 94] and K_2P10.1 (TREK-2) [93], as well as K_2P9.1 (TASK-3) [95–97]. Functional studies showed that a negatively charged residue at the base of the K_2P CAP domain comprises a key RuR sensitivity determinant in the natively RuR sensitive channels K_2P9.1 (TASK-3) [95–97] and K_2P10.1 (TREK-2). Further, placing a negatively charged residue at the CAP base is sufficient for rendering a non-RuR sensitive K_2P responsive to RuR inhibition [54, 93]. Hence, this negative residue is both necessary and sufficient for RuR sensitivity in the context of a K_2P channel.

Structural studies of a RuR-sensitive K_2P2.1 (TREK-1) mutant, I110D [54], revealed that RuR inhibits K_2Ps in a 1:1 stoichiometry matching functional studies [93, 94] and places one ruthenium amine moiety directly over the channel pore while the remainder of the RuR molecule occupies one of the two branches of the extracellular ion pathway (EIP). This “finger in the dam” mechanism provides both, and even an electrostatic and physical barrier that prevents the flow of potassium ions through the selectivity filter.

RuR interacts directly with the negatively charged residues that form the RuR-sensitivity determinant and that constitute the “Keystone inhibitor site” at the base of the CAP domain. The principal mode of binding is through a multipronged interaction made by the two acidic residues at the Keystone inhibitor site with multiple RuR elements. This sort of direct engagement of RuR by multiple acidic sidechains is likely to contribute to RuR block of other classes of RuR-sensitive channels where the binding site is thought to be rich in acidic residues such as TRP channels [98–105], the mitochondrial calcium uniporter (MCU) [106–109], CALHM calcium channels [110–112], ryanodine receptors [113, 114], and Piezo channels [115, 116]. The dinuclear ruthenium amine, Ru360 [117], an inhibitor of the mitochondrial calcium uniporter [106, 118, 119] not previously known to affect potassium channels also binds to the Keystone inhibitor site in a similar way, although due to its reduced electrostatic interactions relative to RuR, Ru360 is a weaker blocker (IC₅₀ = 0.287 vs. 11.3 μM, for RuR and Ru360, respectively) [54] (Fig. 4.2).

Once sites of modulator action are known, it is possible to use the structural information to alter the protein or the ligand to create molecules having new properties. Using a structure-based protein engineering approach, our lab-created RuR super-responder K_2P2.1 (TREK-1) mutants having IC₅₀s in the low nanomolar range by placing acidic resides at Asn147 site at the external mouth of the selectivity filter in conjunction with the I110D mutation (IC₅₀ = 12.7 nM) [54]. Because of the shared pore architecture among K_2Ps, this strategy is generalizable to other K_2P members to create subtypes endowed with a high-affinity RuR sensitivity and could provide a means for exploring their functions. The demonstration that compounds such as RuR and Ru360 can block K_2P function by reaching through the EIP raises the possibility of identifying other classes of molecules that could work similarly. Two interesting directions for making subtype-selective modulators directed at the Keystone inhibitor site would be to capitalize on the renewed interest in synthesizing novel polyruthenium amine derivatives [120] or to design compounds having moieties that interact with the Keystone inhibitor site but that also make specific contacts to non-conserved features of CAP exterior. Biologics, such as nanobodies, may be particularly suited to this type of molecular recognition mode as one can envision that a long variable loop from the nanobody could reach through the EIP to block the pore while other parts of the protein recognize subtype-specific features of the CAP exterior and EIP entryway.

4.3.2. The K_2P Modulator Pocket: A Cryptic Small Molecule Binding Site for K_2P Control

The K_2P modulator pocket (Fig. 4.2) is unrelated to any previously known small molecule binding pocket in the VGIC superfamily and was discovered in structural studies of K_2P2.1 (TREK-1) with two novel activators, ML335 (N-[(2,4-dichlorophenyl)methyl]-4-(methanesulfonamido) benzamide) and ML402 (N-[2-(4-chloro-2-methylphenoxy)ethyl]thiophene-2-carboxamide (Fig. 4.2) [11]. This L-shaped pocket is found in the P1–M4 interface, an intersubunit interface involved in C-type gating [17, 18]. Both compounds bind in similar ways and act as molecular wedges that stabilize the P1–M4 interface and directly activate the channel selectivity filter C-type gate [11, 22]. In the unliganded structure, the K_2P modulator pocket is occluded by P1–M4 interface interactions that require small movements of few residues to open, making this pocket a cryptic site that relies on conformational change similar to cryptic sites described for soluble proteins [121]. Rigidification of the P1–M4 interface is central to channel activation [11, 22]. The observation that these two compounds stabilize this intersubunit interface highlights the general importance of intersubunit interfaces as sites of channel control.

ML335 and ML402 are remarkably selective, activating K_2P2.1 (TREK-1) and K_2P10.1 (TREK-2) but not K_2P4.1 (TRAAK) [11] (Fig. 4.3). This strong subtype selectivity originates from a single lysine residue on the N-terminal end of M4 that engages in a cation–π interaction with the upper ring of each of the activators (Fig. 4.2). The equivalent residue in K_2P4.1 (TRAAK) is glutamine and exchanging K→Q in K_2P2.1 (TREK-1) and Q→K in K_2P4.1 (TRAAK) at this position is sufficient for rendering the former insensitive to ML335 and ML402 activation and the latter sensitive to activation by both compounds [11]. The importance of a single amino acid difference in an otherwise conserved small molecule binding pocket underscores the potential for exploiting local differences and structural knowledge to develop subtype-selective K_2P modulators.

Fig. 4.3.

Selectivity profiles of TREK subfamily small molecule activators and inhibitors. Asterisks indicate modulators lacking a complete profile of subtype selectivity

The K_2P modulator pocket is unique to K_2Ps [11]. Currently, there are no known natural ligands for this site, but it seems unlikely that such a well-defined binding site is only recognized by two unnatural small molecules. Stabilization of the P1–M4 interface is central for integrating gating cues that arise in other parts of the protein, particularly the C-terminal tail [11, 17, 18, 22]. Because the K_2P modulator pocket is in the center of this interface, it seems very likely that there are natural compounds such as lipids, metabolites, signaling molecules, or regulatory proteins that target this site. Hence, understanding the normal function of this part of the channel and whether Nature has exploited natural compounds to affect K_2P activity through the K_2P modulator pocket remains an important direction for future study.

4.3.3. The Fenestration Site: A Binding Site for Activators and Inhibitors

The K_2P M4 transmembrane helix is a key moving element and serves as a means to convey gating cues from temperature [17, 18, 25, 26], pressure [18, 26, 27], and phosphorylation [17] to the C-type gate. Structural studies of TREK subfamily channels have defined two extreme positions of the M4 helix termed “up” and “down” [13, 14, 23–25]. The “down “ state creates a fenestration just below the P2 pore helix that is open to the center of the membrane bilayer [10, 13, 24, 25], the “Fenestration site” (Fig. 4.2). Structural studies of K_2P10.1 (TREK-2) have shown that this site binds to the K_2P inhibitors fluoxetine and norfluoxetine [14]. These compounds bind to a site defined by the lower part of the P2 pore helix and M4 (Fig. 4.2) and require the M4 helix to adopt the “down” position.

Remarkably, crystal structures of a K_2P10.1 (TREK-2) complex with a brominated version of an activator, the fenamate BL-1249, although not defining the entire compound, strongly indicate that this molecule and perhaps other activators bind to the Fenestration site created by the “down” M4 position [53]. How can the binding of a small molecule to the same site yield opposite functional outcomes of inhibition and activation? Clearly, the answer cannot be in the stabilization of the M4 “down” state over the “up” state as the binding of both inhibitors and activators to the Fenestration site requires an M4 “down” conformation [14, 53]. Interestingly, it is suggested that activators such as BL-1249 use their tetrazole moiety to create a binding site for potassium ions in the central cavity and thereby stabilize the selectivity filter C-type gate [53]. Given this type of mechanism, it is notable that the norfluoxetine structure places the norfluoxetine amine just below the selectivity filter where its expected positive charge could provide an unfavorable modification to the potassium ion conduction pathway that would lead to channel inhibition (Fig. 4.2). The fenestration site is commonly found in the VGIC superfamily of which K_2Ps are members and serves as the site of action for multiple types of activators of different classes of potassium channels [53]. Understanding the relationship between the occupation of this site, effects on the selectivity filter C-type gate, and the relationship between the properties of activators and inhibitors that can inhabit this site is an important challenge for further development of K_2P modulators.

4.3.4. The Modulatory Lipid Site: PIP₂ and the C-Terminal Tail

PIP₂ is an important modulatory lipid for TREK subfamily channels [68, 72, 73]. The likely site of PIP₂ action has been located in a series of K_2P2.1 (TREK-1) structures [11, 22]. These show the presence of a phospholipid that co-purified with the channel and that was bound to a groove at the M1/M2/M4 interface (Fig. 4.2). The phosphoinositol headgroup contacts an electro-positive patch on the C-tail comprising five residues implicated in PIP₂ modulation (Arg297, Lys301, Lys302, Lys304, and R311) [68, 72] (Fig. 4.2). This same stretch of the C-terminal tail also contains the intracellular proton sensor site, Glu306 [71], and inhibitory phosphorylation site, Ser300 [74]. The key PIP₂-interacting residues are in a portion of the channel that is most affected by movements of M4 between the “up” and “down” positions. Hence, it seems likely that regulatory impacts of the modulatory lipid, intracellular pH sensor, and phosphorylation site on the C-terminal tail are all tightly intertwined with M4 motions [122]. Further study is needed to unravel these interactions, to understand whether other lipids reported to impact TREK channel function, such as phosphatidyl serine and phosphatidic acid [68, 123], compete with PIP₂ at this site, whether this site can be targeted by small molecules, and to define how changes in this lower part of the channel impact the dynamics and function of the C-type gate.

4.4. Subtype Specific Modulators in the TREK Subfamily

The growing progress in developing modulators for the TREK subfamily has been well-reviewed recently elsewhere [9, 47] and is, therefore, not reiterated here. As new modulators are uncovered, one key question is whether it will be possible to create subtype-specific modulators that can not only distinguish among the various K_2P subfamilies but also among the different subtypes within a subfamily. Such a high level of target selectivity would provide powerful tools for delineating the biological functions of these channels and could provide starting points for the development of subtype-selective pharmacology for these therapeutically relevant targets.

With respect to subtype selectivity, currently characterized TREK subfamily activators mostly fall into two categories. There are many examples of molecules that affect all three members of the TREK subfamily such as Riluzole [89, 124], BL-1249 [52], fenamates [36], ML67–33 [48], 2-Aminoethoxydiphenyl Borate (2-APB) [125, 126], and C3001a [85]. Some of these, such as BL-1249 [52] show limited selectivity for K_2P2.1 (TREK-1) and K_2P10.1 (TREK-2) over K_2P4.1 (TRAAK). The second category of compounds activate K_2P2.1 (TREK-1) and K_2P10.1 (TREK-2) but not K_2P4.1 (TRAAK) and include ML335 [11], ML402 [11], GI-530159 [127], and aristolochic acid (AristA) [128]. Given that K_2P2.1 (TREK-1) and K_2P10.1 (TREK-2) have sequences that are more similar to each other than they are to K_2P4.1 (TRAAK) (Fig. 4.1a), it is not surprising that K_2P4.1 (TRAAK) exhibits different responses to some modulators. There is a report of a K_2P2.1 (TREK-1) selective activator, Compound 36 [36], based on studies of less selective caffeic acid ester activators [36, 56, 63], but a detailed understanding of this high degree of selectivity remains to be defined. The compounds T2A8 and T2A9 are also reported to activate K_2P10.1 (TREK-1) with some selectivity [62]. The fact that there are already compounds showing some degree of selectivity provides an encouraging sign that developing activators having better subtype selectivity is a goal that can be reached.

TREK subfamily inhibitors show slightly more selectivity than activators and follow the same pattern having broadly acting inhibitors and inhibitors showing some selectivity. ML45 [48] and TKDC [129] inhibit all three TREK subfamily channels. There are a set of molecules that inhibit K_2P2.1 (TREK-1) and K_2P10.1 (TREK-2) but not K_2P4.1 (TRAAK), namely, antipsychotics [86], fluoxetine [84], and norfluoxetine [130]. A series of inhibitors (T2I1–10) unable to discriminate between K_2P2.1 (TREK-1) and K_2P10.1 (TREK-2) have also been reported [62], but whether these compounds affect K_2P4.1 (TRAAK) remains to be established. The polyruthenium blockers RuR and Ru360 show an unusual inhibition profile. RuR and Ru360 inhibit K_2P10.1 (TREK-2) but not K_2P2.1 (TREK-1) [54, 93, 94]. K_2P4.1 (TRAAK) is sensitive to RuR but the mechanism of inhibition must be different from the “finger in the dam” mechanism as K_2P4.1 (TRAAK) lacks the defining acidic residue in the Keystone inhibitor site [54]. Ru-TRAAK-1 and Ru-TRAAK-2 inhibit K_2P4.1 (TRAAK) as well as K_2Ps from other subfamilies, such as K_2P1.1 (TWIK-1), K_2P3.1 (TASK-1), and K_2P18.1 (TRESK) [51]. The action of these compounds on other TREK subfamily members has not been reported, but given their ability to inhibit K_2Ps outside of the TREK subfamily, it would be surprising if they did not also show some activity against the more closely related K_2P2.1 (TREK-1) or K_2P10.1 (TREK-2). The peptide spadin has been reported to act as a K_2P2.1 (TREK-1)-selective inhibitor [87, 88], but its mechanism of action remains unclear. T2A3, T2A8, T2A9, and the bioactive lipid 11-deoxyprostaglandin-F2α form an unusual class K_2P2.1 (TREK-1) inhibitors that are reported to also act as K_2P10.1 (TREK-2) activators [62]. How such dual action occurs is not known, but has been proposed to involve a short part of the P2–M4 loop [62]. As with the activators, the growing examples of subtype-selective inhibitors indicate that developing better and more selective inhibitors of the TREK subfamily should also be feasible, especially as more structural data about how such molecules interact with K_2Ps becomes available.

So far, the origins of subtype selectivity for TREK modulators are understood for only two sites at the level of the atomic interactions, the Keystone inhibitor site, and the K_2P modulator pocket. For the Keystone inhibitor site, the negative charge at the Keystone inhibitor site is the principal determinant controlling RuR and Ru360 inhibition [54] (see Sect. 1.3.1). How RuR affects K_2P4.1 (TRAAK) remains unknown, as this channel lacks the negative residue at the Keystone inhibitor site and is inhibited with a stoichiometry higher than the 1:1 interaction of the Keystone inhibitor site [93, 94]. The basis of the subtype selectivity of the ML335 and ML402 activators [11] arises from a single lysine in the K_2P modulator pocket that controls subtype selectivity (see Sect. 1.3.2). Although not yet mapped in atomic detail, BL-1249 shows a ~10-fold selectivity for K_2P2.1 (TREK-1) and K_2P10.1 (TREK-2) over K_2P4.1 (TRAAK) that originates from differences in the M2/M3 helix interface [52]. Understanding the molecular origins of the subtype specificity for this compound as well as GI-530159 [127], aristolochic acid (AristA) [128], and C3001a [85] will require further studies that combines both structural and functional approaches.

4.5. Perspectives on K_2P Channel Polysite Pharmacology

Structural data for complexes of K_2P2.1 (TREK-1) and K_2P10.1 (TREK-2) with various modulators have uncovered a surprisingly large number of unique sites for small molecule control present at every level of the channel structure with respect to the membrane. These sites are arranged above (Keystone inhibitor site [54]), at the level of (K_2P modulator pocket [11]), and below (fenestration [14] and modulatory lipid [11] sites) the structure that is at the heart of channel function, the C-type gate [17, 18, 20–22]. It seems likely that there is a fifth site within the channel cavity, as there is good functional evidence that this architectural feature is targeted in TREK channels by alkylammonium pore blockers [9, 53] and this shared K_2P architectural element is the site of crystallographically defined small molecule block of K_2P3.1 (TASK-1) [12]. All of these sites have functions that are crucial for the normal functioning and modulation of K_2Ps. The intersection of small molecule modulators and key sites of channel modulation emphasizes the importance of building an integrated understanding of modulator action and the basic mechanisms of channel function.

The structural pharmacology of the Keystone inhibitor site, K_2P modulator pocket, and Fenestration site has been defined by crystal structures of K_2Ps complexed with nonnatural compounds that exert powerful effects on channel function. These findings highlight two key outstanding questions: To what extent has Nature has capitalized on these control points to influence TREK channel activity? and What are the naturally occurring modulators that target these sites? Answering such questions will be important for developing a better understanding of the roles of K_2Ps in physiological responses. One known natural modulator of great physiological interest whose site remains to be defined on the TREK subfamily is the site of action of the activator arachidonic acid [75, 131]. From a structural perspective, although K_2P4.1 (TRAAK) was the first member of the TREK subfamily structurally characterized [13] and received its name due to its sensitivity to arachidonic acid [131], it remains the only channel in the TREK subfamily lacking any modulator-bound structures. Given the fact that this channel stands apart with respect to the selectivity of many modulators (Fig. 4.3), defining the site of arachidonic acid action in the TREK subfamily along with obtaining structural data for small molecule:K_2P4.1 (TRAAK) complexes will provide important guides for unraveling natural mechanisms of channel modulation and better templates for subtype-selective modulator discovery.

The clear crosstalk between various K_2P moving parts and the C-type gate [17, 18, 52] raises the question of whether the action of modulators at the various sites influence each other. The strength of polyruthenium amine block at the Keystone inhibitor site is not influenced by C-type gate stabilization by compounds occupying the K_2P modulator pocket [54], but the extent to which there may be crosstalk among the other sites remains to be evaluated. Understanding such interactions could provide insight into how to boost the efficacy of current modulators and will refine our understanding of how signals from various parts of the channel impinge on the C-type gate.

The current structural knowledge of modulator sites provides a framework to discover a new chemical matter that can affect K_2P function in novel ways. Such molecules may be engineered to block or enhance the consequences of various physical and chemical stimuli or to modify the channel chemically so that its biogenesis, distribution, and interaction with other cellular components can be followed in complex cell types and tissues. Besides further crystallographic studies of new K_2P: modulator complexes, an obvious key advance will be to understand structural consequences of heterodimer formation [42, 82–84] and to image K_2Ps in lipid membrane environments using cryo-electron microscopy (cryo-EM) so that interactions between the channel and bilayer can be better understood. The initial burst of activity surrounding K_2P structural pharmacology is the first of many waves that will fill out our understanding of this important ion channel family and should lead to new and novel therapeutic strategies for a host of brain, cardiac, and nervous system diseases.