Multiple modalities converge on a common gate to control K2P channel function.

Members of the K(2P) potassium channel family regulate neuronal excitability and are implicated in pain, anaesthetic responses, thermosensation, neuroprotection, and mood. Unlike other potassium channels, K(2P)s are gated by remarkably diverse stimuli that include chemical, thermal, and mechanical modalities. It has remained unclear whether the various gating inputs act through separate or common channel elements. Here, we show that protons, heat, and pressure affect activity of the prototypical, polymodal K(2P), K(2P)2.1 (KCNK2/TREK-1), at a common molecular gate that comprises elements of the pore-forming segments and the N-terminal end of the M4 transmembrane segment. We further demonstrate that the M4 gating element is conserved among K(2P)s and is employed regardless of whether the gating stimuli are inhibitory or activating. Our results define a unique gating mechanism shared by K(2P) family members and suggest that their diverse sensory properties are achieved by coupling different molecular sensors to a conserved core gating apparatus.

Introduction

K2P (KCNK) potassium channels produce ‘background' currents that stabilize the membrane resting potential and that have a critical role in control of cell excitability (Bayliss and Barrett, 2008; Enyedi and Czirjak, 2010). K2Ps are members of the superfamily of voltage-gated ion channels (Yu et al, 2005) but have a unique topology; the channels are dimers of subunits that each has four transmembrane segments and two pore regions per polypeptide (Goldstein et al, 2005). Although well known as ‘leak' channels that are constitutively open at rest, many K2Ps are polymodal and respond to a wide range of diverse regulatory inputs that include extracellular and intracellular pH, temperature, membrane stretch, polyunsaturated fatty acids, volatile anaesthetics, and noxious chemicals (Honore, 2007; Enyedi and Czirjak, 2010). Numerous studies have established that K2ps have major roles in the brain, cardiovascular system, and somatosensory neurons (Duprat et al, 2007; Honore, 2007; Bayliss and Barrett, 2008; Folgering et al, 2008; Sabbadini and Yost, 2009; Enyedi and Czirjak, 2010). Further, mutations in K2Ps have been linked to mental retardation (Barel et al, 2008) and migraine (Lafreniere et al, 2010). Nevertheless, our understanding of K2p gating mechanisms lags far behind other potassium channel classes (Cohen et al, 2009; Mathie et al, 2010) and presents a barrier to explaining how such diverse types of inputs modulate K2P function. Thus, defining K2p functional mechanisms remain an important objective for establishing how these channels influence excitation.

K2P2.1 (KCNK2/TREK-1) (Fink et al, 1996; Honore, 2007) is the best-studied polymodal K2P and responds to both extracellular (Cohen et al, 2008; Sandoz et al, 2009) and intracellular (Maingret et al, 1999b; Honore et al, 2002) acidosis, heat (Maingret et al, 2000; Noel et al, 2009), mechanical forces (Maingret et al, 1999a; Noel et al, 2009), and anaesthetics (Patel et al, 1999; Heurteaux et al, 2004). K2P2.1 (TREK-1) is important in sensory neuron pain responses (Alloui et al, 2006; Noel et al, 2009) and vasodilation (Bryan et al, 2006, 2007; Blondeau et al, 2007; Garry et al, 2007). Its activity is linked to chronic pain, thermosensation (Alloui et al, 2006; Noel et al, 2009), response to general anaesthetics (Heurteaux et al, 2004), and depression (Gordon and Hen, 2006; Heurteaux et al, 2006; Perlis et al, 2008; Dillon et al, 2010). Consequently, K2P2.1 (TREK-1) along with other K2Ps present attractive targets for the development of new agents directed at treating ischaemic injury, pain, and depression (Honore, 2007; Bayliss and Barrett, 2008).

In general, potassium channels contain two major points of control or ‘gates' that are used to varied degrees to control activity depending on the particular channel (Yellen, 2002). One gate, called the ‘outer' or ‘C-type' gate involves the selectivity filter, which makes the direct contacts with the permeant ions, and is sensitive to external potassium concentration. The other gate, a constriction of the pore-lining transmembrane segments, is known as the ‘inner gate' and can block access to the channel pore from the cytoplasmic side. Both K2P2.1 (TREK-1) and the drosophila K2P, KCNK0, have been shown to have a C-type gate (Zilberberg et al, 2001; Cohen et al, 2008). Because of the extracellular placement of key proton sensing residues, this type of gate has been implicated in the response of a number of K2Ps to external pH (Kim et al, 2000; Rajan et al, 2000; Lopes et al, 2001; Morton et al, 2005; Sandoz et al, 2009). Recent studies also suggest the presence of an inner gate in mutant KCNK0 channels (Ben-Abu et al, 2009), although the extent to which such a gate operates in native K2Ps is unclear. Nevertheless, the gating mechanisms that control K2P function remain obscure and many fundamental questions remain unanswered regarding the generality of these gating elements within the K2P family and how such elements are coupled to the diverse types of gating inputs (Cohen et al, 2009; Mathie et al, 2010).

Such questions are magnified in the archetypal polymodal K2P K2P2.1 (TREK-1). This channel is controlled by physically diverse signals that act on different channel elements. For example, a histidine in the first extracellular loop is the main sensor for extracellular protons (Cohen et al, 2008; Sandoz et al, 2009), whereas the C-terminal cytosolic domain contains elements that are involved in the response to temperature and mechanical inputs (Patel et al, 1998; Maingret et al, 1999b, 2000; Honore et al, 2002). It has remained unclear whether these diverse modulatory inputs and sensors control K2P2.1 (TREK-1) via separate or common gating mechanisms. Additionally, because of the high sequence divergence in the K2P family, the extent to which gating mechanisms are conserved among K2Ps has remained unresolved. Here, we examine these issues with regard to external proton, heat, and pressure-evoked K2P2.1 (TREK-1) gating and demonstrate that these three gating inputs act via a common gate that has the characteristics of a C-type gate. Further, we find that regardless of whether the gating signal activates or inhibits channel function, this mechanism is conserved among diverse K2P channels.

Results

A yeast potassium uptake selection identifies mammalian K2P2.1 (TREK-1) gain-of-function mutations

Complementation of potassium-uptake-deficient yeast has been a fruitful approach for studying a variety inward rectifier and viral potassium channels (Minor, 2009) but has not yet been applied to other types of potassium channels. Because of their ability to conduct ‘leak' currents, we reasoned that K2P channels such as K2P2.1 (KCNK2/TREK-1) might be functional in this system. We found that unlike the inward rectifier Kir2.1 (Minor et al, 1999; Chatelain et al, 2005), mammalian K2P2.1 (TREK-1) complemented growth of a potassium-uptake-deficient yeast strain SGY1528 (Tang et al, 1995) under the mild potassium limited conditions (1 mM KCl) but not under the most stringent complementation conditions (<1 mM KCl) (Figure 1A). To identify key K2P2.1 (TREK-1) gating elements, and following precedents set with G-protein activated inward rectifiers (Sadja et al, 2001; Yi et al, 2001), we constructed a randomly mutagenized K2P2.1 (TREK-1) library that had mutations throughout the channel sequence (Supplementary Figure S1A) and selected for gain-of-function (GOF) mutants that would rescue growth in the presence of 0.5 mM KCl. The selections yielded GOF mutations in four channel regions (Figure 1B): the extracellular portion of the first selectivity filter (I148T), the second P-loop (L267P), the N-terminal part of transmembrane segment 4 (M4) (W275S and F276L), and the C-terminal cytoplasmic domain (E306G, E309A, S333G, and S333R). We also isolated a double mutant that combined the two changes in the extracellular vestibule (I148T/L267P). The identified C-terminal domain positions have been shown previously to cause K2P2.1 (TREK-1) GOF (Maingret et al, 2000; Honore et al, 2002; Murbartian et al, 2005) and thus, validate the general selection strategy. As the GOF positions in the extracellular region and the M4 positions had not been previously implicated in K2P function, we turned our attention to these.

Figure 1

Functional selection identifies GOF mutations in a mammalian K2P2.1 (TREK-1). (A) Growth of potassium-transport-deficient yeast (SGY1528) expressing the yeast potassium transporter TRK1, Kir2.1, an inactive Kir2.1 mutant (Kir2.1*), K2P2.1 (TREK-1), and two exemplar K2P2.1 (TREK-1) GOF mutants, L267P and W275S under non-selective conditions (100 mM KCl), and two different selective conditions (1 mM and 0.5 mM KCl). Rows indicated with ‘Ba2+' show growth in the presence of the Kir2.1 and K2P2.1 (TREK-1) inhibitor 8 mM BaCl2. (B) K2P2.1 (TREK-1) subunit topology diagram. Locations of GOF mutations are indicated in yellow. Transmembrane segments M1, M2, M3, and M4 and the two P-loop domains are labelled. (C) Exemplar current–voltage traces from whole cell recordings of Xenopus oocytes injected with 0.3 ng of K2P2.1 (TREK-1) or GOF mutant mRNA. Currents were elicited in solutions containing 2 mM potassium (ND96) by a ramp protocol from −150 to +50 mV from a −80 mV holding potential. Values for the average current (in μA, mean±s.e., n=5) at 0 mV were K2P2.1 (TREK-1) (0.52±0.40), I148T (1.61±0.106), L267P (1.77±0.121), I148T/L267P (3.11±0.092), W275S (4.86±0.099), F276L (1.70±0.097). (D) Quantification of normalized current amplitudes at 0 mV from Xenopus oocytes injected with 0.3 ng of mRNA for the indicated channels. (E) Cell-attached mode single channel recordings of K2P2.1 (TREK-1), I148T/L267P, and W275S expressed in COS7 cells. O1, O2, and O3, indicate the first, second, and third open states, respectively. C indicates the closed state. (F) Open channel probabilities from single channel analyses calculated on recordings of ∼30 s duration. K2P2.1 (TREK-1) (n=7), I148T (n=7), L267P (n=5), I148T/L267P (n=9), W275S (n=9), F276L (n=8). Data represent mean±s.e.

Measurement of current–voltage relationships in Xenopus oocytes injected with equivalent amounts of mRNA for each of the GOF mutants or wild-type K2P2.1 (TREK-1) revealed that all tested GOF mutants had increased activity relative to wild type (Figure 1C and D). The effects of the I148T/L267P double mutant were additive compared with the individual mutants and suggest that the selectivity filter and second P-loop sites act independently. Importantly, the GOF mutants retained potassium selectivity (Supplementary Figure S1B). Three parameters affect whole cell current (I) as expressed by the equation: I=N × PO × i, where N is the active channel number, PO is single channel open probability, and i is single channel current amplitude (Hille, 2001). Therefore, we examined the effects of the GOF mutations on each of these parameters. Single channel analysis, under conditions in which similar numbers of channels were observed (Supplementary Figure S1C), demonstrated that all of the GOF mutations increase single channel open probability (mean±s.e., n=5–9): wild type (0.04±0.01), I148T (0.10±0.03), L267P (0.12±0.07), I148T/L267P (0.17±0.037), W275S (0.17±0.08), F276L (0.16±0.07) (Figure 1E and F), and that all but F276L cause a slight increase in single channel conductance (Supplementary Figure S1D and E). Furthermore, none of the tested GOF mutants increased surface expression, as judged either by biotinylation of surface proteins in COS7 cells (Supplementary Figure S2A) or by immuno-detection of the channels on the surface of the oocytes (Supplementary Figure S2B). Thus, taken together, the data indicate that the GOF mutations affected the gating machinery of the channel.

Protons, heat, and mechanical force control K2P2.1 (TREK-1) activity via a common gate

The K2P2.1 (TREK-1) gating apparatus includes a C-type-like outer gate that encompasses the selectivity filter that closes in response to extracellular acidosis (Cohen et al, 2008; Sandoz et al, 2009) as well as sensory elements in the C-terminal cytoplasmic tail that respond to temperature and mechanical inputs (Maingret et al, 1999b, 2000). It has remained unclear, however, whether activating stimuli such as increased temperature (Maingret et al, 2000) and mechanical force (Chemin et al, 2005) act at the same outer gate or elsewhere.

To probe how the gating changes affected K2P2.1 (TREK-1) responses to both inhibitory and activating inputs, we challenged each of the GOF mutants by three different stimuli: extracellular acidosis, temperature, and mechanical stress. Strikingly, the GOF mutants I148T, L267P, I148T/L267P, and W275S affected the responses to each of the three stimuli to some degree (Figure 2; Supplementary Figure S3 and Table S1). The largest effects were caused by I148T/L267P and W275S. I148T/L267P blunted the response to all three modalities (Figure 2B, D, and F), whereas W275S had large effects on extracellular pH (pHO) and temperature responses (Figure 2B and D). In stark contrast, the reaction of the F276L GOF mutant to pHO, temperature, and mechanical stress was not significantly different from that of wild type. Even though I148T, L267P, I148T/L267P, and W275S could not be completely inhibited by pHO under the limits of our experimental conditions, their sensitivity to external magnesium inhibition (Maingret et al, 2002) remained similar to wild type (Supplementary Figure S4). This result indicates that the GOF mutations did not create channels that were generally resistant to inhibition. The fact that neither the open probability nor the surface expression of F276L exceeds that of I148T/L267P or W275S eliminates the possibility that the blunted responses of I148T/L267P and W275S arise simply because of the increased open probability or differences in expression levels. Instead, these data support the idea that the I148T, L267P, and W275S mutants act by uncoupling the sensors for the various stimuli from the gating apparatus. Further, when taken together, our data indicate that despite the radically different physical natures of the stimuli, the actions of all three modalities converge on the parts of K2P2.1 (TREK-1) in which these GOF mutations reside, the extracellular vestibule and the outer M4 region, to control gating.

Figure 2

K2P2.1 (TREK-1) GOF mutations affect response to extracellular acidosis, heat, and pressure. (A) Exemplar two-electrode voltage-clamp recordings of the response of K2P2.1 (TREK-1) and the W275S GOF mutant to external pH (pHO) changes in 2 mM [K+]O solutions. (B) Normalized pHO responses (at 0 mV) for the indicated channels. (C) Exemplar two-electrode voltage-clamp recordings of the response of K2P2.1 (TREK-1) and the W275S GOF mutant to temperature in 2 mM [K+]O, pH 7.4 solutions. (D) Normalized temperature responses (at 0 mV) for the indicated channels. (E) Exemplar mechanical force (cell-attached mode, 150 mM KCl pH 7.2 in the bath, 5 mM KCl pH 7.4 in the pipette) responses of K2P2.1 (TREK-1) and the I148T/L267P GOF mutant stimulated by negative pressure applied to the extracellular side of the plasma membrane through the patch pipette. (F) Normalized pressure responses for the indicated channels. In (A, C) currents were elicited by a ramp from −150 to +50 mV from a holding potential of −80 mV. Lines for (B, D) show fits to the equations I=Imin+(Imax–Imin)/(1+([H+]O/K1/2)) and I=Imin+(Imax–Imin)/(1+e1/2(), respectively. Data in (B, D, F) show mean±s.e. (n=8–30). N⩾2 for all experiments.

The effect of pHO on K2P2.1 (TREK-1) is antagonized by increases in the permeant potassium ion concentration (Cohen et al, 2008; Sandoz et al, 2009) and suggests a gating mechanism that relies on conformational changes in the selectivity filter. To probe whether the GOF mutations affected this process, we measured how an increase in the extracellular potassium concentration, [K+]O, from 2 to 90 mM affected pHO inhibition. We found that I148T, I148T/L267P, and W275S had significantly diminished responses to changes in [K+]O in comparison with wild-type K2P2.1 (TREK-1) (Figure 3A and B; Supplementary Figure S5). Notably, the two GOF mutants that had the largest effects on eliminating the responses to the various gating inputs, I148T/L267P and W275S, also had the largest impact on [K+]O sensitivity (Figure 3D). Further, the pHO response of these mutants in both high and low [K+]O conditions was very similar to that of wild-type K2P2.1 (TREK-1) under high [K+]O conditions. In contrast, the F276L GOF mutant was not different than wild type (Figure 3C and D). Thus, three properties indicate that I148T/L267P and W275S GOF mutations act by stabilizing the core gating mechanism of the channel to promote the conductive conformation of the selectivity filter: the large increase in channel activity, the resistance to changes in [K+]O, and the similarity of the pHO response to that of the wild-type channel in high external potassium. These properties are consistent with a C-type gating mechanism that involves changes in the conformation of the selectivity filter.

Figure 3

Extracellular loops and extracellular proximal portion of M4 control K2P2.1 (TREK-1) gating. (A–C) Normalized response to pHO in low (2 mM, data are from Figure 2B) and high (90 mM) [K+]O (2 K and 90 K, respectively) for K2P2.1 (TREK-1) and the indicated GOF mutants. Whole cell currents were elicited in Xenopus oocytes by a ramp from −150 to +50 mV from a holding potential of −80 mV (2 K) or 0 mV (90 K). (D) Quantification of the effect of high (90 mM) external potassium on the pHO response from the curves in (A–C) I6.5norm(90 K)/I6.5norm(2 K). Lines for (A–C) show fits to the Hill equation ((I=Imin+(Imax–Imin)/(1+([H+]O/K1/2))). Data represent mean±s.e. (n=6–30). Statistical analysis: t-test. ***P<0.001, NS, not significant (P>0.05). N⩾2 for all experiments.

C-type gating of K2P potassium channels is accompanied by an increase of the relative permeability of sodium over potassium (Yuill et al, 2007; Cohen et al, 2008). To apply a further test to the idea that the conductive conformation of the pore is stabilized by the I148T/L267P and W275S GOF mutations, we measured how pHO changes affected the ion selectivity of the mutants in comparison with wild-type K2P2.1 (TREK-1) (Figure 4). Lowering pHO from 9.0 to 6.9 in a recording solution containing 100 mM sodium caused a significant right shift in the reversal potential values measured in oocytes expressing either wild-type or F276L channels (ΔErev=31.1 and 21.3 mV, respectively; Figure 4A–D) and had no effect on the reversal potential recorded in 100 mM potassium (Figure 4E). Of note, the change in the reversal potential in 100 mM sodium at pHO 6.9 did not simply arise from a decrease in the number of active potassium channels (Supplementary Figure S6). These results are consistent with prior studies of wild-type K2P2.1 (TREK-1) (Cohen et al, 2008) and are indicative of C-type gating. In contrast, under the same recording conditions, I148T/L267P and W275S were significantly less sensitive to the pHO change (ΔErev=8.8 and 12.7 mV, respectively). Accordingly, pHO changes from 9.0 to 6.9 steadily increased Na+/K+ permeability ratio of wild-type and F276L channels, but had a minimal effect on I148T/L267P and W275S (Figure 4F). These data provide further support for the idea that I148T/L267P and W275S mutations stabilize the potassium-selective conductive conformation of the selectivity filter and act on a C-type-like gate.

Figure 4

Extracellular loops and extracellular proximal portion of M4 control ion selectivity of K2P2.1 (TREK-1). (A–D) Exemplar two-electrode voltage-clamp recordings of the response of the wild-type and mutant K2P2.1 (TREK-1) channel to pHO changes in 100 mM external sodium solutions. Currents were evoked by 60 ms long pulses from −150 to −50 mV in 10 mV increments from a −80 mV holding potential. Cells were injected with different amounts of mRNA to yield comparable current amplitudes. (E) Exemplar two-electrode voltage-clamp recordings of the response of the wild-type K2P2.1(TREK-1) channel to pHO changes in 100 mM external potassium solutions. Currents were evoked by 60 ms long pulses from −50 to 60 mV in 10 mV increments from a 0-mV holding potential. (F) Quantification (mean±s.e., n=8–11) of apparent permeability ratios at different pHO using the equation pNa/pK=eFΔErev/RT, where pNa and pK are permeabilities for sodium and potassium, respectively, and ΔErev is a difference between the reversal potentials measured in 100 mM sodium and 100 mM potassium solutions. N⩾2 for all experiments.

The extracellular region of M4 is a key element of the K2P2.1 (TREK-1) gating apparatus

Of all the GOF mutations we isolated, W275S stood out. In contrast to the Ile148 site, which is in a region that is similar to a portion of voltage-gated potassium channels that is crucial for control of the C-type gate (Lopez-Barneo et al, 1993) and the Leu267 site, which is not conserved within the K2P family, Trp275 is largely conserved in K2Ps (Supplementary Figure S7). More strikingly, it occurs in a channel element that is rather intolerant to amino-acid changes in other potassium channel classes (Collins et al, 1997; Minor et al, 1999; Irizarry et al, 2002). These observations, together with our data demonstrating that a single change at this site blunted different gating modality responses and [K+]O sensitivity, strongly suggested that this M4 region might have a special role in K2P function. As the Trp → Ser substitution causes a dramatic sidechain volume change, we tested the tolerance of the K2P2.1 (TREK-1) Trp275 position to substitution with a set of amino acids having diverse physicochemical properties. All non-aromatic substitutions examined, W275A/T/L/N/Q/D/E, resulted in channels with increased activity as assessed by two-electrode voltage-clamp (Figure 5A). In contrast, aromatic substitutions, W275Y and W275F, resulted in channels that were similar to wild type. Accordingly, investigation of how the Trp275 substitutions affect gating by pHO and temperature revealed that all non-aromatic substitutions produced channels having reduced responses to both gating stimuli (Figure 5B and C; Supplementary Table S1). These results further support the integral role of the Trp275 position in K2P2.1 (TREK-1) gating and the key role of the extracellular proximal portion of M4.

Figure 5

Tests of the impact of amino-acid changes at Trp275 on K2P2.1 (TREK-1) function. (A) Normalized whole cell current amplitude (at 0 mV) recorded in 2 mM [K+]O pH 7.4 from Xenopus oocytes injected with equivalent amounts of mRNA for the indicated K2P2.1 (TREK-1) Trp275 substitutions. Statistical analysis: t-test. ***P<0.001; NS, not significant (P>0.05). (B) Normalized pHO responses in 2 mM [K+]O for the indicated Trp275 mutants. (C) Normalized temperature responses in 2 mM [K+]O for the indicated W275 mutants. Curves show fits to the Hill equation I=Imin+(Imax–Imin)/(1+([H+]O/K1/2)) or I=Imin+(Imax–Imin)/(1+e1/2(). Data represent mean±s.e. (n=4–30). N⩾2 for all experiments.

The extracellular region of M4 is functionally conserved across the K2P family

To test whether the Trp275 equivalent position might have a general role in controlling gating within the K2P family, we examined how aromatic → Ser mutations at the analogous M4 position in other pHO-sensitive K2P channels affected function (Figure 6A). Strikingly, equivalent changes at the Trp275 homologous position in K2P10.1 (KCNK10/TREK-2), W301S; K2P9.1 (KCNK9/TASK-3), F225S; and K2P3.1 (KCNK3/TASK-1), F225S, profoundly affected pHO gating (Figure 6B–D). This effect occurred even though these pHO-gated K2P channels use different extracellular elements as [H+]O sensors (Kim et al, 2000; Rajan et al, 2000; Lopes et al, 2001; Morton et al, 2005; Sandoz et al, 2009) and was independent of whether [H+]O is inhibitory, as for K2P9.1 (TASK-3) and K2P3.1 (TASK-1), or activating, as for K2P10.1 (TREK-2). Therefore, despite diverse [H+]O-sensor placement and the non-uniform direction of gating responses to [H+]O, our data indicate that this gating stimulus converges on a common site in M4 and strongly suggests that the gates of K2P2.1 (TREK-1), K2P9.1 (TASK-3), K2P3.1 (TASK-1), and K2P10.1 (TREK-2) work by a shared mechanism. As a further test, we examined substitution of a homologous position that is not natively an aromatic residue, Glu228, in the [H+]O-gated channel K2P5.1 (KCNK5/TASK-2). E228S resulted in non-functional channels; however, introduction of an aromatic residue, E228W, profoundly inhibited K2P5.1 (TASK-2) pHO gating (Figure 6E). In accord with our K2P2.1 results, M4 mutations that make the channel less responsive to pHO gating (K2P10.1 (TREK-2) W301S and K2P5.1 (TASK-2) E228W) increase basal activity (Supplementary Figure S8). In contrast, mutations that facilitate pHO responses (K2P3.1 (TASK-1) F225S and K2P3.1 (TASK-3) F225S) suppress macro-current amplitude at pHO 7.4. Together, our studies of the effects of mutations at the K2P2.1 (TREK-1) W275 equivalent M4 position provide further evidence for the general role of this element in K2P channel gating and establish that various K2P channels utilize a conserved gating apparatus to respond to extracellular acidosis regardless of the location of the sensor or direction of the response.

Figure 6

Importance of the Trp275 position is functionally conserved among K2P channels for both pHO and temperature induced gating. (A) Amino-acid alignment of the M4 region from the indicated K2P channels. Residues conserved in three or more of the indicated sequences are highlighted in blue. The K2P2.1 (TREK-1) Trp275 homologous position is indicated in orange. (B–E) Normalized pHO responses and exemplar two-electrode voltage-clamp recordings of Xenopus oocytes for the indicated channels and mutants (2 mM [K+]O). Injected mRNA amounts for the exemplar traces in ng are as follows for wild type and mutant, respectively: K2P10.1 (TREK-2) 5.0, 1.0; K2P9.1 (TASK-3) 0.08, 0.20; K2P3.1 (TASK-1) 1.5, 5.0; K2P5.1 (TASK-2) 2.0, 1.0. (F–I) Normalized temperature responses for the indicated channels and mutants (2 mM [K+]O, pH 7.4). Exemplar two-electrode voltage-clamp records are shown only for K2P10.1 (TREK-2). Data represent mean±s.e. (n=6–15) and is fitted to I=Imin+(Imax–Imin)/(1+([H+]O/K1/2)) or I=Imin+(Imax–Imin)/(1+e1/2(). Dashed lines connect data points. N⩾2 for all experiments.

Our data show that in K2P2.1 (TREK-1) different modalities act on the same gate even though the gating inputs produce opposite effects (i.e. [H+]O inhibition versus temperature activation (Figure 2)). To test whether this property is a shared feature of the core gating apparatus within the K2P family, we examined how the Trp275 equivalent changes affected temperature responses of other K2P channels. We found that in K2P10.1 (TREK-2), in which both [H+]O and increased temperature activate the channels (Maingret et al, 2000; Kang et al, 2005), the M4 mutant W301S profoundly diminished temperature-induced gating (Figure 6F) in addition to reducing the [H+]O response (Figure 6B). In contrast, aromatic → Ser substitution at the Trp275 position in channels that are [H+]O gated but that lack a strong temperature response, K2P9.1 (TASK-3), K2P3.1 (TASK-1), K2P5.1 (TASK-2), had no effect on temperature sensitivity (Figure 6G–I). These results suggest that gating machinery alterations are not sufficient to affect the temperature response of a K2P channel that lacks strong intrinsic temperature sensitivity. Importantly, the data demonstrate that [H+]O and increased temperature both act on a common gate in channels that respond to both modalities regardless of the direction of the response.

The C-type gate controls diverse gating inputs in multiple K2P family members

In K2Ps that respond to both pHO and temperature, the above experiments support the idea that these diverse inputs act at a common gate that is affected by Trp275 or its equivalent. Because of these observations and the fact that the K2P2.1 (TREK-1) W275S GOF mutant reduces the sensitivity of pHO gating to extracellular potassium concentration (Figure 3) in a way that supports the presence of a C-type gate, we decided to test whether the K2P temperature response is also sensitive to extracellular potassium. In line with the idea that the C-type-like gate responds to both pHO and temperature, we found that high extracellular potassium significantly blunted the temperature response of both K2P2.1 (TREK-1) (Figure 7A) and K2P10.1 (TREK-2) (Figure 7B). The K2P2.1 (TREK-1) temperature response in high external potassium was very similar as that caused by the W275S mutation. In contrast, the temperature response of the GOF mutant F276L, which has a pHO and [K+]O response like wild type (Figure 3C), followed the changes in external potassium in a way that resembled wild type. Together, these results strongly support the idea that the C-type gate controls both proton and temperature responses in the two channels and that the M4 Trp275 position has an important role in this process.

Figure 7

Temperature gating responds to changes in external potassium concentration. Normalized temperature responses of (A) K2P2.1 (TREK-1), K2P2.1 (TREK-1) W275S, and K2P2.1 (TREK-1) F276L and (B) K2P10.1 (TREK-2) from two-electrode voltage-clamp recordings in 2 and 90 mM [K+]O (2 K and 90 K, respectively) elicited by a ramp from −150 to +50 mV from a holding potential of −80 mV (2 K) or 0 mV (90 K) at indicated temperatures at pH 7.4. The K2P2.1 and K2P10.1 responses to heat in 2 mM [K+]O are the same as in Figures 2D and 6F, respectively. Data (mean±s.e., n=8–15) were taken at 0 mV (2 K) or +40 mV (90 K) and fitted with the equation I=Imin+(Imax–Imin)/(1+e1/2(). N⩾2 for all experiments.

Discussion

K2P channels are the most diverse potassium channel family in terms of both sequence (Goldstein et al, 2005; Yu et al, 2005) and gating inputs (Cohen et al, 2009; Mathie et al, 2010). This high sequence diversity together with the radically different physical natures of the gating commands, such as pH, temperature, and pressure, have posed a challenge to the identification of unifying models that could explain the mechanisms by which K2Ps act. To address this question in an unbiased manner, we sought to identify the key gating elements in the archetypal polymodal K2P channel, K2P2.1 (TREK-1), by exploiting a GOF selection in potassium-transport-deficient yeast. This initial effort was based on the premise that changes in components of the channel that contribute to gating should be particularly sensitive to mutations. The rationale builds on the ample precedents in which both functional screens and selection experiments have provided unexpected insights into the mechanisms of permeation and gating of a variety of other ion channel types such as inwardly rectifying potassium channels (Minor et al, 1999; Sadja et al, 2001; Yi et al, 2001; Bichet et al, 2004; Chatelain et al, 2005, 2009), voltage-gated potassium channels (Lai et al, 2005), and TRP channels (Su et al, 2007; Zhou et al, 2007; Grandl et al, 2008, 2010; Myers et al, 2008; Hu et al, 2009).

Prior studies have identified a variety of K2P2.1 (TREK-1) elements that have principal roles in sensing different input types. His126 in the first extracellular loop acts as the external pH sensor (Cohen et al, 2008; Sandoz et al, 2009), whereas heat and mechanical force gating require elements from the cytosolic C-terminal domain (Maingret et al, 2000; Chemin et al, 2005; Kang et al, 2005). Although these different stimuli have different molecular sensors, it has not been obvious whether such components act via a common gate or by independent mechanisms. We expected that some of the GOF mutations from the selection experiment would target such sensor elements. Indeed, we found most of the known GOF mutations that hit key residues in the C-terminal tail (e.g. E306G, E309A, S333G, and S333R). These outcomes validated the concept behind the selection and focused our attention on a set of GOF mutations at K2P2.1 (TREK-1) positions that had not previously been implicated in channel function. These sites included one in the external part of the selectivity filter (I148T), one in the distal part of the second P-loop (L267P), a double mutant combining both (I148T/L267P), and two sites on the N-terminal end of the M4 transmembrane segment (W275S and F276L). All of these mutations increase channel activity and affect channel open probability. The properties of the F276L mutant are important. This mutant affects open probability to a similar degree as the most effective GOF mutants, I148T/L267P and W275S, is expressed on the surface at similar levels as the other GOF mutants, but responds to the various gating inputs in a manner that is similar to wild type. Thus, simply increasing channel open probability is insufficient for causing the change in the responses to the various gating modalities. This result points to another explanation for how I148T/L267P and W275S act and indicates that in addition to increasing the open probability, changes at these sites affect elements central to the integration of the gating commands.

Many K2Ps respond to extracellular pH (pHO) changes (Enyedi and Czirjak, 2010; Mathie et al, 2010). Although, pHO is a common gating modality, the locations of the proton sensors in the K2P extracellular loops are not conserved (Kim et al, 2000; Rajan et al, 2000; Lopes et al, 2001; Morton et al, 2005; Sandoz et al, 2009). Interestingly, the direction of the response to pHO changes is also not shared. This response diversity is best exemplified by two closely related K2Ps that use the equivalent extracellular histidine for proton sensing but that have opposite pHO responses (Sandoz et al, 2009), K2P2.1 (TREK-1), and K2P10.1 (TREK-2). Extracellular protons inhibit K2P2.1 (TREK-1) but activate K2P10.1 (TREK-2). pHO is thought to affect K2Ps by a C-type-like mechanism that involves the channel selectivity filter (Lopes et al, 2001; Cohen et al, 2008; Sandoz et al, 2009). Because the selectivity filter region is the most conserved K2P feature, such a mechanism could provide a general means by which different pHO sensing determinants could affect gating. We found that mutations in the K2P2.1 (TREK-1) extracellular loops and at the Trp275 M4 position reduce the inhibitory effect of low pHO on channel activity. In other types of potassium channels, C-type gating is tightly linked to changes in the ion occupancy of the selectivity filter (Cordero-Morales et al, 2006; Clarke et al, 2010; Cuello et al, 2010). Accordingly, we also found that the sensitivity of the K2P2.1 (TREK-1) pHO response was markedly reduced by increased extracellular potassium (Figure 3). This sensitivity was effectively lost in the I148T/L267P and W275S GOF mutants but not in the F276L GOF mutant. Further, closure of the C-type-like gate has been shown to lead to a loss in ion selectivity (Yuill et al, 2007; Cohen et al, 2008). In accord with the idea that the I148T/L267P and W275S GOF mutants act on a C-type-like gate, these mutants did not display the pHO-induced changes in selectivity that are prominent in both wild-type K2P2.1 (TREK-1) and the F2767L GOF mutant (Figure 4). Thus, we infer that the I148T/L267P and W275S blunt the channel response by favouring a conformation that is equivalent to that of an activated C-type gate.

Discovery of GOF mutants in the N-terminal end of the M4 region is striking. In other potassium channel classes, drastic alterations in the homologous region are generally not tolerated (Collins et al, 1997; Minor et al, 1999; Irizarry et al, 2002). Investigation of the mutational tolerance at the Trp275 position showed that aromatic substitutions produced channels that were very similar to wild type, whereas changes to small or hydrophobic residues increased whole cell currents and reduced responses to both pHO and temperature (Figure 5). Deeper probing of the equivalent position among other K2Ps revealed the general importance of the Trp275 equivalent M4 position in both pHO-mediated and temperature-evoked gating (Figure 6). Explicit tests of the effects of extracellular potassium show that, similar to pHO responses, temperature activation of K2P2.1 (TREK-1) and K2P10.1 (TREK-2) are blunted to a level that is equivalent to that caused by the W275S change. Taken together, these observations provide overwhelming support for a model in which the selectivity filter acts as the common gate used in diverse K2Ps to respond to radically different input modalities and indicate that this gate involves components of the extracellular loops and the N-terminal portion of M4.

In many ways, a C-type gate is well suited to act as a common means to respond to different physical inputs detected by separate channel elements. Structural changes in channel components remote from the selectivity filter have recently been shown to be correlated to selectivity filter ion occupancy changes in inward rectifiers (Clarke et al, 2010). Importantly, such changes do not require large movements of the transmembrane helices. Thus, even though in some channels the action of the inner and outer gates may be coupled (Panyi and Deutsch, 2006, 2007; Ben-Abu et al, 2009), an ‘inner gate' is not an essential element for controlling a C-type, selectivity filter-based gate (Clarke et al, 2010). It seems plausible that conformational changes in diverse channel elements, such as the extracellular loops or intracellular domains, could independently affect the conformation of the channel at the selectivity filter. In this regard, it is striking that a number of mutations have been identified in topologically similar regions that affect the temperature gating of TRPV1 and TRPV3 (Grandl et al, 2008, 2010; Myers et al, 2008). This coincidence is particularly remarkable because the two channel families share very little homology and raises the possibility that a selectivity filter-based gating mechanism is the principal mechanism for gating of other channels within the voltage-gated superfamily to which both K2Ps and TRPs are members (Yu et al, 2005).

It is possible that, in addition to the outer C-type-like gate, K2P2.1 (TREK-1) possesses an intracellular gate formed by the M2 and M4 segments (Treptow and Klein, 2010). Mutations in the intracellular region of the putative pore-lining helices that could contribute to an inner gate, M2 and M4, have been shown to affect activity of K2P channels from the TASK group (Barel et al, 2008; Ashmole et al, 2009), and the fly K2P KCNK0 (Ben-Abu et al, 2009). Of the three modalities affected by our GOF mutants, the pressure response was the least blunted, a result that suggests that other parts of the channel may be important for this response. The general role of a K2P inner gate and other questions, such as whether an inner gate involves the cytoplasmic C-terminus, and how such a gate might couple to the C-type gate, remain open for further study.

The diversity of amino-acid sequences and gating input modalities sets K2p channels apart from other well-studied potassium channel classes such as inward rectifiers and voltage-gated channels (Enyedi and Czirjak, 2010). Our data strongly suggest that despite this exceptional diversity, the various sensing mechanisms act on a conserved core gating machinery that involves the extracellular elements of the pore-forming segments and the extracellular proximal part of the putative fourth transmembrane helix, M4. Although high-resolution structural data for K2P channels is presently lacking, experimentally constrained K2P channel models suggest that the Trp275 position should be intimately involved in supporting the structure of the selectivity filter (Kollewe et al, 2009). Such placement is consistent with the key role of the Trp275 in gating.

Because of their four transmembrane segment/two pore-forming region topology, K2P channels have non-identical pore-lining segments, M2 and M4, that must interact with the selectivity filter elements in different ways. Further, this unusual topology dictates that each of our GOF channels will contain only two copies of the GOF mutations, not four as would be the case in a homomeric potassium channel from the other potassium channel classes. Hence, the unusual K2P architecture may be the reason why, in contrast to homomeric potassium channels classes (Collins et al, 1997; Minor et al, 1999; Irizarry et al, 2002), we find that the upper portion of M4 tolerates changes in diverse K2P family members (Figure 6). Thus, our data raise the interesting question of whether M2 and M4 have different functional roles and set a framework for future analysis of this issue.

Our results strongly suggest that the movement of the upper part of M4 is a central part of the core gating mechanism by which different sensor regions that respond to separate inputs couple to K2P channel gating and that the changes in M4 affect the selectivity filter conformation (Figure 8). This key role for the upper part of M4 is in line with prior studies of other potassium channel classes that have identified gating-sensitive residues in the analogous region of the voltage-gated potassium channel Shaker (Ogielska and Aldrich, 1999; Yifrach and MacKinnon, 2002), demonstrated that changes in the conformation the KcsA pore-lining helix affect selectivity filter ion occupancy (Cuello et al, 2010), and that have shown gating induced changes in the inner part of the selectivity filter in KirBac3.1 (Gupta et al, 2010). Our data suggest that inputs such as pHO, temperature, and pressure act by affecting structural rearrangements in the channel selectivity filter. These rearrangements are coupled to changes in the ion occupancy of the selectivity filter and the conformation of the outer pore and extracellular proximal portion of M4 that involve Ile148, Leu267, and Trp275 (Figure 8). Temperature and mechanical stress have sensing elements located in the intracellular C-terminal domain (Honore, 2007). The impact of the GOF mutations at Ile148, Leu267, and Trp275 on these modalities leads us to propose that changes in the C-terminal domain couple to the C-type gate through movements of the M4 transmembrane segment. These findings set the stage for future studies of the precise details by which the C-terminal tail couples to the C-type-like gate.

Figure 8

Cartoon model of K2P2.1 (TREK-1) C-type-like gating by extracellular acidosis, heat, and pressure. Cartoon model of how extracellular protons ([H+]O), heat, and pressure affect the transition of K2P2.1 (TREK-1) between a low-activity (inhibited) state and a high-activity (activated) state that involves a C-type-like gate. As suggested earlier (Sandoz et al, 2009), external acidification causes structural rearrangements in the pore triggered by electrostatic interactions between the protonated extracellular pH-sensing His126, and the negatively charged region in the P2-loop, which includes Asp263 and Glu265. We propose that Ile148, Leu267, and Trp275 (indicated with stars) are crucial elements of the C-type-like gate. As suggested elsewhere (Honore, 2007), temperature and mechanical stress have their sensing elements located in the intracellular C-terminal domain. We further propose that activated C-terminal domain (indicated with the orange halo) induces movements of the M4 transmembrane segment and affects channel activity through the C-type gate. The blue arrow indicates the putative pathway. For clarity, only one of the two cytoplasmic C-termini is depicted.

Because of the important role of the M4 region in K2P gating, this channel element may prove to be a key site that can be targeted by both natural and synthetic compounds aimed at controlling K2P function. In this regard, demonstration that K2P channels can function in yeast should facilitate development of yeast-based assays for small molecules that overcome the intrinsic challenge posed by ‘background' or ‘leak' nature (Enyedi and Czirjak, 2010) of K2P function.

Materials and methods

Molecular biology

Murine K2P channels were cloned into pGEMHE/pMO, IRES-GFP (Invitrogen), or pYES2-MET25 (Minor et al, 1999) for expression in oocytes, COS7 cells, and yeast, respectively. For expression in COS7 cells, the haemagglutinin (HA) tag (YPYDVPDYA) was added to the N terminus of TREK-1. Cloning was performed using standard molecular biology procedures and verified by DNA sequencing.

Library construction and TREK-1 GOF selection in yeast

Standard techniques were used for yeast manipulation (Sherman, 2002). For full-length library construction, the K2P2.1 (TREK-1) coding region along with 45 flanking nucleotides 5′ and 3′ to the gene was amplified from a pYES2-MET25-K2P2.1 (URA3) plasmid using Taq DNA polymerase. Amplification used 29 cycles and conditions conducive to copying errors (100 mM Tris pH 8.3, 500 mM KCl, 70 mM MgCl2, 0.1% gelatin). The PCR product was mixed with empty pYES-MET25 vector linearized at the points of K2P2.1 (TREK-1) insertion (HindIII and XhoI), and transformed into the ura− SGY1528 yeast strain (Tang et al, 1995). Insertion of mutagenized K2P2.1 (TREK-1) into the vector occurred via homologous recombination of the DNA regions flanking the K2P2.1 (TREK-1) gene. Sequence analysis of 25 randomly picked clones from the library showed that 60% of plasmids carried one to three single nucleotide substitutions that were evenly distributed throughout the K2P2.1 (TREK-1) gene. Yeast selection was performed as described previously (Minor et al, 1999). Briefly, yeast transformants (∼50 000 total) were selected on synthetic medium without uracil and methionine (–Ura –Met), containing 100 mM KCl and replica plated onto –Ura –Met with 1.0 mM KCl. Following 3 days of growth at 30°C, colonies were replica plated onto –Ura –Met with 0.5 mM KCl. Plasmids were isolated, retested, and sequenced.

Electrophysiology

Unless indicated otherwise, all electrophysiological recordings were performed at room temperature. For two-electrode voltage-clamp measurements in Xenopus laevis oocytes, defolliculated stages V–VI oocytes were injected with 0.08–5.0 ng mRNA and assayed 24–72 h later. Each oocyte was impaled with two standard microelectrodes (0.3–3.0 MΩ) filled with 3 M KCl, and assayed using the GeneClamp 500B (MDS Analytical Technologies) amplifier controlled by the pClamp software (Molecular Devices). Data were digitized at 1 kHz using Digidata 1332A (MDS Analytical Technologies). For pHO and temperature experiments, currents were evoked from a −80 mV holding potential with a 1-s long ramp from −150 to +50 mV in 2 K (standard ND96, containing 96 mM NaCl, 2 mM KCl, 1.8 mM CaCl2, 2.0 mM MgCl2) or 90 K (90 mM KCl, 8 mM NaCl, 1.8 mM CaCl2, 2.0 mM MgCl2) solutions buffered with 10 mM Tris (pH 9.0, 8.1), 5 mM HEPES (pH 7.8, 7.4, 7.1), or 5 mM MES (pH 6.5, 5.9). The solutions were exchanged consecutively from pH 9.0 to pH 5.9 (22.5°C), or from 12.5 to 32.5°C (pH 7.4). The effects of pH or temperature were reversible in all K2P2.1 (TREK-1) mutants. Dose-response data were fitted with the equations: I=Imin+(Imax–Imin)/(1+([H+]O/K1/2)) where Imax and Imin are maximal and minimal current values, respectively, K1/2 is a half-maximal effective concentration of extracellular protons ([H+]O), and H is the Hill coefficient or I=Imin+(Imax–Imin)/(1+e1/2(), where T1/2 and S are the temperature of half-maximal activation and slope factor, respectively. For oocyte batches exhibiting high background chloride currents at pH⩽6.5, NaCl and KCl were substituted with equimolar amounts of Na- or K-gluconate. Permeability ratios were determined under bionic conditions using the modified Goldman–Hodgkin–Katz equation, pNa/pK=eFΔErev/RT, where pNa and pK are relative permeabilities for sodium and potassium, respectively, ΔErev is the difference between reversal potentials measured in 100 mM Na and 100 mM K solutions (ΔErev=Erev,Na–Erev,K) and F, R, and T have their usual meanings (Hille, 2001). Currents were evoked by 60 ms step pulses from −150 to −50 mV in 10 mV increments from a holding potential of –80 mV in solutions containing 100 mM NaCl or KCl, 1 mM MgCl2, 0.3 mM CaCl2, buffered with 10 mM Tris (pH 9.0), or 5 mM HEPES (pH 6.9–7.9). All reversal potential measurements were corrected for junction potential (2.5 mV). For most figures, 2–6 independent experiments were performed and 8–42 cells were analysed per construct.

Patch-clamp measurements were performed on African green monkey kidney COS7 cells. Cells were cultured in Dulbecco's Modified Eagle's Medium (Gibco BRL Life Technologies) supplemented with 10% fetal calf serum (Hyclone). Cells were transfected 1 day after plating, using a modified DEAE/Dextran protocol as previously described (Chemin et al, 2005). For single channel activity, 0.01 μg of plasmid DNA was transfected per 35 mm dish containing 20 000 cells per dish; for other experiments 0.1 μg was transfected. Experiments were performed between 2 and 4 days after transfection.

Stretch-response experiments were performed in cell-attached mode the pipette solution contained (in mM): 150 NaCl, 5 KCl, 2 CaCl2, and 10 HEPES at pH 7.4 with NaOH, and the bath solution contained (in mM): 155 KCl, 3 MgCl2, 5 EGTA, 10 HEPES at pH 7.2 with KOH. Patch pipettes of about 1.5 MΩ were used. Membrane patches were stimulated with negative pressure pulses, from 0 to −80 mm Hg in −10 mm Hg increments during 300 ms each 3 s, through the recording electrode using a pressure-clamp device (ALA High Speed Pressure-Clamp-1 system; ALA-Scientific). Single channel recordings were obtained in cell-attached mode using patch pipettes of about 4 MΩ and symmetrical solutions (in mM): 150 KCl, 5 EGTA-K, 1 EDTA-K, 10 HEPES, pH 7.2 with KOH. For all experiments, currents were filtered at 1 kHz, digitized at 20 kHz, and analysed with pCLAMP9.2 and ORIGIN6.0 software.