TWIK-related K⁺ channels TREK1, TREK2 and TRAAK are two-pore-domain potassium (K_2P) channels (Enyedi and Czirjak, 2010; Feliciangeli et al., 2015). Active as homo or heterodimers (Blin et al., 2016; Lengyel et al., 2016; Levitz et al., 2016), they play important roles in the nervous system, regulating cognitive function, pain perception, neuroprotection as well as anesthesia (Heurteaux et al., 2004; Noel et al., 2011; Wang et al., 2020). They produce inhibitory background currents that are sensitive to a wide range of physical and biological stimuli including mechanical forces, temperature, pH and lipids (Maingret et al., 1999; Maingret et al., 2000a; Maingret et al., 2000b; Honore et al., 2002; Sandoz et al., 2009). They are sensitive to antidepressants and volatile anesthetics (Patel et al., 1999; Heurteaux et al., 2006). These stimuli converge on a gate that involves the selectivity filter (SF) of these channels (Bagriantsev et al., 2011). The use of quaternary ammonium (QA) ions demonstrated that unlike most K⁺ channels, in a closed state, TREK1 has no inner gate preventing the binding of QA ions deep within the pore just below SF (Piechotta et al., 2011). Intracellular acidification (Sandoz et al., 2009; Piechotta et al., 2011), mechanical forces (Bagriantsev et al., 2011) or binding of the cytoplasmic protein AKAP150 (Sandoz et al., 2006) all act indirectly on the SF gate through the fourth membrane-spanning segment (M4) and the cytoplasmic C-ter of the channel (Bagriantsev et al., 2012). TREK1, TREK2 and TRAAK are also sensitive to the extracellular pH (Sandoz et al., 2009). A single extracellular residue, H126 in TREK1 and H151 in TREK2, is involved in proton sensing. TREK1 and TRAAK are inhibited by acidification whereas TREK2 is activated. This differential effect of acidification involves charged residues close to M4 (Sandoz et al., 2009). Structural modeling and site-directed mutagenesis suggested that attraction or repulsion between the protonated side chain of histidine and these closely located negatively or positively charged residues control the gating of these channels.

The recent crystallographic 3-D structures of these channels have shown that TREK/TRAAK channels can adopt two main conformational states which show differences in the positions of the membrane-spanning segments M2 and M4, and described as the “up” and “down” conformations (Brohawn et al., 2012; Lolicato et al., 2014). Although initially attributed to open and closed states, both conformations are in fact associated with channel activity. More recently, we showed that a cytoplasmic domain, immediately following M4 (the proximal C-ter domain, named pCt), exerts opposite effects in TREK1 and TRAAK (Soussia et al., 2018). In basal conditions, pCt favors activity in TREK1 whereas it impairs activity in TRAAK. Using the conformation-dependent binding of the antidepressant fluoxetine, we gathered data suggesting that TREK1 and TRAAK conformations at rest are different, “down” for TREK1 and “up” for TRAAK, and under the influence of pCt. The differential regulation of TREK1 and TRAAK is related to a phosphatidylinositol-4, 5-bisphosphate (PIP₂)-binding site (R329, R330, R331) present within the pCt of TREK1 and absent in TRAAK. The binding of PIP₂ to this site results in a stabilization of the conductive conformation of TREK1 (Soussia et al., 2018). Uncoupling pCt and M4 by introducing glycine residues at the junction of these domains prevents the activating property of TREK1 pCt and the inhibitory effect of TRAAK pCt, suggesting a physical coupling between pCt and the SF gate of the channel.

Despite all these data related to the 3D-structures and regulations of these channels, the nature of the conformational changes that control the activity of these channels remains an open question. Here, by using domain swapping and the conformation-dependent binding of fluoxetine, we evaluated how pCt affects the sensitivity of TREK1 and TRAAK to external pH and openers including ML67, BL1249, and halothane, a volatile anesthetic.

Molecular biology - Human TREK1 (KCNK2, GenBank accession number AAH69462.1) and TRAAK (KCNK4, NCBI Reference Sequence: NP_201, 567.1) were cloned into pIRES2-eGFP vector (Clontech). All the chimeras were obtained by overlapping PCR and inserted into the same vector. Substitution of I292G293D294 in TREK1 and I253G254N255 in TRAAK by a repeat of glycines was performed by PCR using Pfu Turbo DNA polymerase (Agilent). All the constructs were verified by DNA sequencing. The exchange of pCt between TREK1 and TRAAK to generate TRAAKpCt_TREK1 and TRAAKpCt_TREK1 has been described elsewhere (Soussia et al., 2018).

Cell culture and transfection - HEK cells were grown in 100-mm tissue-culture dishes (Falcon, Franklin Lakes, NJ) in Dulbecco’s modified Eagle’s medium (Gibco, Life Technologies, Saint Aubin, France) supplemented with 10% fetal calf serum (Hyclone, Thermo Fisher Scientific GMBH, Ulm, Germany) and 1% penicillin-streptomycin (Gibco, Life Technologies, Saint Aubin, France) in a humidified incubator at 37 C (5% CO2). For expression and electrophysiology of WT and mutant channels, 0.8 µg of plasmid was transfected using JET PEI (Polyplus transfection) according to the manufacturer’s instructions. Cells were plated onto 35 mm dishes 24 h before transfection, and experiments were performed over the following 1–2 days.

Electrophysiology - Pipettes were pulled from haematocrit-capillaries (Hirschmann Laborgeraete, Germany) using a vertical puller (PC-10, Narishige International, London, United Kingdom), and had resistances of 2–4 MΩ when filled with internal solution and measured in standard bath solution. Whole cell membrane currents were measured and filtered at 3 kHz by a RK 400 patch clamp amplifier (Bio-Logic Science Instruments), and digitized at 10 kHz using a 12-bit analog-to-digital converter Digidata-1322 (Axon Instrument, Sunnyvale, CA, United States of America). Recordings were done using Clampex 8.2 software (Axon Instrument). The external solution used to record TREK1 and TRAAK currents in HEK cells contained (in mM): 140 NaCl, 10 TE A-Cl, 5 KCl, three MgCl₂, one CaCl₂, 10 HEPES, pH 7.4 with NaOH. The pipette solution contained (in mM): 155 KCl, three MgCl₂, five EGTA, 10 HEPES, pH 7.2 with KOH. Cells were placed in 35 mm dishes through which solution flowed at 1–2 ml/min. Inflow to the dish was by gravity from several reservoirs, selectable by activation of solenoid valves (Warner Scientific). Bath solution exchange was essentially complete by < 30 s. Experiments were performed at room temperature. Cell populations were compared using Kruskall Wallis, Mann-Whitney test or Two-way ANOVA. The data are given as the mean ± SEM.

Many studies have focused on the structural mechanisms controlling the activity and pharmacology of TREK1, TREK2 and TRAAK channels. According to these studies, TREK/TRAAK channels have two main conformations, “up” and “down” (Brohawn et al., 2014; Lolicato et al., 2014) involving a movement of M4 directly coupled to the SF gate (Miller and Long, 2012; Dong et al., 2015; Zhuo et al., 2016). Our previous work has shown that pCt plays a critical role in channel conformation (Soussia et al., 2018). In basal conditions, TRAAK is in the fluoxetine-resistant “up” conformation whereas TREK1 is in the fluoxetine-sensitive “down” conformation (Soussia et al., 2018). In the same conditions, swapping pCt between these two channels leads TREK1pCtTRAAK to adopt the “up” conformation and TRAAKpCtTREK1 the “down” conformation. Here, we show that pCt also affects the sensitivity of TREK1 and TRAAK to extracellular pH and openers by acting on the SF gate. The recent co-crystallisation of TREK2 with a brominated BL1249 has shown that this drug binds to the “fenestration site” (Schewe et al., 2019). This site is located below the P2 pore helix and exposed by the “down” movement of M4 (Brohawn et al., 2012; Miller and Long, 2012; Brohawn et al., 2014; Lolicato et al., 2014). Like BL1249, ML67 also binds to the “fenestration site” (Bagriantsev et al., 2013; Schewe et al., 2019). Since this “fenestration site” is only accessible in the “down” state, we expected TREK1 to be sensitive to BL1249 and ML67, but not TRAAK. Surprisingly, TRAAK is sensitive to both openers like TREK1. However, TRAAK is more sensitive to ML67 than BL1249. Given that pCt affects channel conformation (Soussia et al., 2018), we wondered whether pCt could also affect the sensitivity of TREK1 and TRAAK channels to openers. Replacing TREK1 pCt by that of TRAAK increased the sensitivity of TREK1pCt_TRAAK to ML67 and decreased that to BL1249. In contrast, TRAAKpCt_TREK1 displayed a lower sensitivity to ML67 and a higher sensitivity to BL1249. Similar effects were observed with the 3G-mutation, which decouples pCt from the SF. These data indicate that pCt affects allosterically binding of these openers to the “fenestration site”.

A recent study on volatile anaesthetics, which activate TREK1 (Patel et al., 1999) but not TRAAK, identified a binding site with a glycine as a key element at the M2-M3-M4 interface (Wague et al., 2020). Interestingly, this binding site corresponds to the “fenestration site”, showing that different types of openers bind to this site. Swapping pCt between TREK1 and TRAAK or disrupting pCt/SF coupling altered the sensitivity to halothane. These findings confirm the allosteric effects of pCt on SF gate via the “fenestration site”. However, key questions remain unresolved: how does pCt affect drug binding to the “fenestration site”? And how does the binding of molecules to this “fenestration site” affect the SF state? A recent structural study on the TREK2/BL1249 complex gave new insights. The negatively charged tetrazole moiety of BL1249 promotes the SF conductive state by changing the coordination environment of K⁺ at position S1 and S4, increasing ion permeation (Schewe et al., 2019). ML67 contains a negatively charged moiety like BL1249 (Bagriantsev et al., 2013; Schewe et al., 2019), suggesting that BL1249 and ML67, and possibly halothane, may stabilize the SF gate in a conductive state through a similar mechanism. External pH, which regulates TREK1, may also affect the SF gate by modulating ion permeation. Comparing the structures of the TREK-related TASK2 channel at pH 6.5 and pH 8.5 revealed that external acidification induces a distortion of the SF at the S0 and S1 positions that prevents K⁺ to occupy these positions (Li et al., 2020). TREK1 and TASK2 are similarly sensitive to external pH, suggesting that pCt in TREK1 may also affect the sensitivity to external pH in an allosteric manner. Consistent with this, we found that disrupting pCt/SF coupling with the 3G-mutation changed pH-sensitivity of TREK1 and TRAAK. Not only openers bind to this “fenestration site” but also inhibitors, including fluoxetine and norfluoxetine (Miller and Long, 2012; Dong et al., 2015; Zhuo et al., 2016), indicating that the “fenestration site” is a dual-action site for its ligands. Identifying the precise molecular mechanism by which ligands influence the conformation site of the SF gate is key to the development of novel modulators of TREK/TRAAK channels. It has to be noted that the “fenestration site” is composed of elements from both subunits of the active dimers (Natale et al., 2021), suggesting that homo and heterodimers could be specifically modulated (Blin et al., 2016; Lengyel et al., 2016; Levitz et al., 2016), enhancing the interest in developing drugs specifically targeting this region.

In summary, the allosteric effect of pCt on the stimulatory action of ML67, BL1249, AA and halothane shows that these stimuli converge to the same SF gate. This gate is the same as the gate controlled by extracellular pH. Although the openers act through the same SF gate, they do not affect the conformational states of these channels in the same way. Previously, we showed that AA promotes the shift from the “up” to the “down” conformation of TREK1 and from the “down” to the “up” conformation of TRAAK (Soussia et al., 2018). Here, we showed that the 3G-mutation, which uncouples pCt from SF, changed the conformational state of TREK1 as expected. Surprisingly, and in contrast to AA, ML67 only promotes the transition from the “up” to the “down” conformation of TRAAK under basal conditions. TREK1 remains in the “down” conformation upon ML67 application. Taken collectively, these results provide further evidence that the crystallography identified “up” and “down” conformations do not correspond to open or closed states of these channels.