A unique alkaline pH-regulated and fatty acid-activated tandem pore domain potassium channel (K_2P) from a marine sponge

Tandem pore domain K⁺ channels (K_2P; KCNK) are a family of ion channels involved in the control of background or leak membrane conductances (Bayliss and Barrett, 2008; Enyedi and Czirják, 2010). Background K⁺ conductances are a major determinant of resting membrane potential and input resistance, two key components of electrical signaling in animal cells. In mammals, there are 15 K_2P genes that encode for six subfamilies that group together functionally, based on the mechanisms of channel activation and inhibition (Enyedi and Czirják, 2010). Members of the K_2P family of channels are regulated in variable ways by diverse chemical, physical and pharmacologic agents, including arachidonic acid (AA), internal and/or external protons, membrane stretch, temperature and anesthetic agents (Bayliss and Barrett, 2008; Dedman et al., 2009). A structural understanding for the distinct mechanisms of activation is still incomplete, although the highly variable C-terminal region is implicated in the regulation of some K_2P channels by AA, internal pH and mechanical stimuli (Kim et al., 2001b) and specific residues near the channel pore regulate external or internal proton sensitivity in some channels (Rajan et al., 2000; Kim et al., 2001a; Morton et al., 2003; Niemeyer et al., 2007; Sandoz et al., 2009; Niemeyer et al., 2010). A better understanding of the molecular determinants of K_2P channel activation might be achieved by the isolation of more diverse members of this important family of ion channels.

We report the first identification of a novel K_2P channel cloned from the marine sponge Amphimedon queenslandica (Hooper and Van Soest, 2006) a demosponge from the Great Barrier Reef. Sponges are sessile, aquatic, multicellular animals that lack a nervous system, but display contractions to expel water (Nickel, 2004; Elliott and Leys, 2007). There is limited information available on membrane conductances in sponges (Zocchi et al., 2001; Tompkins-MacDonald et al., 2009), but hexactinellid (glass) sponges are known to coordinate flagellar movement and demonstrate calcium-dependent action potentials (Leys et al., 1999). Studying ion channels in native sponges using fine electrodes is impeded by the complex glycocalyx and high membrane fluidity (Carpaneto et al., 2003) and thus requires an alternative approach such as the cloning and functional expression of ion channels. Because the phylum Porifera may share a common ancestor with all other metazoans (Philippe et al., 2009), insight into the structure and function of ion channels from this unusual animal (A. queenslandica) may be a key to understanding the genetic repertoire present at the origin of metazoan evolution. Moreover, an understanding of the activation mechanisms for channels cloned from diverse species can provide comparative data that will be useful in determining structure–function relationships of ion channel proteins and understanding their physiological roles in different organisms.

As shown in this study, a K⁺ channel subunit cloned from A. queenslandica (AquK_2P) possesses two pore-forming domains and four transmembrane segments, features common to all K_2P channels. The deduced amino acid sequence for AquK_2P fails to classify the channel into a known subfamily of K_2P channels. We find that the sponge channel is activated by AA, a polyunsaturated fatty acid, in an external pH-dependent fashion. The sponge K_2P channel is also activated by internal alkalinization with or without AA but its gating is insensitive to temperature changes as well as membrane stretch. Overall, the conservation in sponge and some mammalian K_2P channels of voltage-independent outward rectification and sensitivity to pH and AA suggest that these are ancient elements of animal K_2P channel gating.

Amphimedon queenslandica larvae were procured from Heron Island Reef, Queensland, Australia, and preserved in RNAlater stabilization solution (Qiagen, Valencia, CA, USA). RNA extraction and cDNA preparation and cloning methods have been described previously (Tompkins-MacDonald et al., 2009). The primer pair used for amplification of a predicted full-length K_2P subunit was based on sequence data in the A. queenslandica genome (Srivastava et al., 2010) and was composed of 5′-GGCTCGAGCCACCATGGAGAAAGAGGTCGAG-3′ (forward) and 5′-GGACAGTTCACTCCTTCTCTACTTCAG-3′ (reverse). After cloning into the PCR-4-TOPO vector (Invitrogen, Carlsbad, CA, USA), the AquK_2P cDNA was ligated into a pXT7 expression plasmid. To ensure the integrity of the sequence data and that the cDNA was complete, we subjected two clones to complete sequencing of both DNA stands in both directions and obtained identical results. We compared our sequence data with the genomic database for A. queenslandica and detected only a single K_2P channel sequence at contig 13452:297098..298220; the open reading frame was interrupted by a single intron and the coding region of 1059 nucleotides predicted a 353 amino acid peptide. The exon sequences in the genomic database and two cDNA clones were identical except for the codon predicting residue 350. At this position, our results predicted a valine whereas the genomic database predicts a glycine.

The translated sponge K_2P sequence was aligned and compared with amino acid sequences of K_2P (KCNK) sequences obtained from NCBI. Evolutionary analyses were conducted in MEGA5 (Tamura et al., 2011). Alignments were performed in MAFFT (http://mafft.cbrc.jp/alignment/server/) using default parameters and viewed in BioEdit version 7.0.5.3 (Ibis Biosciences, Carlsbad, CA, USA). Membrane topology was predicted by the ExPASy TMHMM server v2.0 (Swiss Institute of Bioinformatics, Lausanne, Switzerland) and adjusted by alignment with channels with published topology profiles. Sequence data for AquK_2P have been deposited in GenBank (accession number JN165777).

RNA preparation, Xenopus laevis oocyte isolation and oocyte injection were carried out using methods previously described in detail (Boland et al., 2009; Tompkins-MacDonald et al., 2009). Animal protocols were approved by the University of Richmond and the Virginia Commonwealth University Institutional Animal Care and Use Committees. Individual oocytes (stages V and VI) were injected with 40–60 nl RNA (0.25–1.0 ng nl^–1) in RNAse-free water. Oocytes were maintained at 17–19°C for 1–5 days in an ND-96 solution containing (in mmol l^–1): 96 NaCl, 1 KCl, 1 CaCl₂, 2 MgCl₂, 10 Hepes, 5 sucrose and 2 Na pyruvate, with 50 U ml^–1 penicillin G, 50 μg ml^–1 streptomycin and, in most cases, 50 μg ml^–1 tetracycline. Solution pH was adjusted with NaOH to 7.4 using an Orion 310 pH meter (Thermo Fisher Scientific, Pittsburgh, PA, USA).

Two-electrode voltage clamp (TEVC) methods were used to record whole oocyte potassium currents using standard methods (Boland et al., 2009) with a Geneclamp 500B (Molecular Devices, Sunnyvale, CA, USA) or an OC-725C amplifier (Warner Instruments, Hamden, CT, USA). Currents were sampled at 5–10 kHz and filtered at 1–2 kHz. Experiments were performed at room temperature (21–23°C), and a small volume bath chamber was perfused continuously during recordings. Bath temperature was changed by cooling or heating the solution from the reservoir to the recording chamber and was measured with a calibrated digital thermometer containing a small probe that was placed next to the oocyte in the recording chamber. The standard TEVC external recording solution contained (in mmol l^–1): 2 KCl, 98 NaCl, 2 MgCl₂, 0.3 CaCl₂ and 5 Hepes with the pH adjusted with NaOH to 6.5–9.0. Solution osmolality was adjusted by adding sucrose or eliminating some of the NaCl and was measured using a Fiske 210 osmometer (Advanced Instruments, Norwood, MA, USA).

For inside-out patch recordings, we used a Dagan 3900A amplifier (Dagan Corporation, Minneapolis, MN, USA). Patch recording pipettes had resistances of approximately 0.5–1.5 Mohm after fire-polishing and were backfilled with a standard extracellular solution that was composed of (in mmol l^–1): 96 Na methanesulfonate (MES), 2 KMES, 1 MgCl₂, 1 CaCl₂ and 5 Hepes with the pH adjusted to 7.4 with NaOH. The bath solution (facing the internal surface of the membrane) was composed of (in mmol l^–1): 96 KMES, 10 Hepes, 5 EGTA and 5 EDTA. The pH of the bath solution for patches was modified with the addition of KOH, which summed to 3 mmol l^–1 K⁺ at the highest pH values tested. Inside-out macropatch recordings were obtained at room temperature (20–23°C) with continual perfusion of the internal surface of the membrane patch.

Electrophysiological data were recorded on Pentium computers equipped with Digidata A/D hardware (Molecular Devices). Molecular Device's Clampex and Clampfit analysis software were used. Data were also transferred to Microsoft Excel (Microsoft, Redmond, WA, USA) or Microcal Origin (OriginLab, Northampton, MA, USA) for additional analysis and the production of figures. Results are expressed as means ± s.e.m.; N is the number of cells or patches tested. Statistical significance was evaluated by paired or unpaired Student's two-tailed t-tests with P<0.05 considered to be significant.

Reagents were purchased from Sigma-Aldrich (St Louis, MO, USA) or Nu-Chek Prep Lipids (Elysian, MN, USA). AA was dissolved in ethanol or DMSO at 1000–3000×, stored at –20°C and diluted into recording solution immediately before use. Vehicle controls were determined to be without effect at the highest concentrations of diluents. Fatty-acid-free bovine serum albumin (BSA) was dissolved into recording solutions at 0.5 mg ml^–1.

AquK_2P is the first K_2P (KCNK) channel to be identified from poriferans and the subunit shares the common features of this family of channels: two pore-forming loops (P1 and P2) and four transmembrane domains (TM1–TM4; Fig. 1). The pore-forming loops contain a K⁺ selectivity filter with the conserved residues GYG in the P1 region and GFG in the P2 region. Overall, AquK_2P shows low (18–25%) amino acid identity to human K_2P channels but even the mammalian K_2P channels have only 25–38% overall amino acid identity between subfamilies. Residues that have remained conserved over evolutionary time are likely to be crucial to channel function (black and gray boxes, Fig. 1). For example, the predicted pore regions are 53% (P1) and 82% (P2) conserved among AquK_2P and TREK-1 (Fig. 1).

The six subfamilies of K_2P channels – TREK/TRAAK, TALK/TASK-2, TWIK, THIK, TASK-1,-3,-5 and TRESK – can be discriminated based on functional differences (Bayliss and Barrett, 2008; Enyedi and Czirják, 2010; Mathie et al., 2010). However, phylogenetic comparison of AquK_2P with known human K_2P channels shows that the molecular composition of the sponge variant is sufficiently different that it does not group into these known functional classifications. An unrooted phylogenetic tree shows a separate branch-point for AquK_2P whereas six major branches cluster the human K_2P subfamilies both structurally and functionally (Fig. 2).

Non-inactivating K⁺ channel currents were generated by external application of the polyunsaturated fatty acid AA (5,8,11,14-eicosatetraenoic acid) to oocytes injected with RNA prepared from the AquK_2P cDNA (Fig. 3). We discovered that the activation by AA depended on an external alkaline pH (pH_o; see Fig. 4); therefore, we studied the AA-dependent current activation and assessed K⁺ selectivity using pH_o 8.5 or 9.0. AquK_2P currents are largely time independent, but at strong depolarizations they were not instantaneous (Fig. 3A, see top three traces at +20, +40 and +60 mV and 3C, inset). Resolution of the time dependence of activation was better achieved in inside-out patch recordings (supplementary material Fig. S1A). The rapid current activation was best fit by a two-exponential function, but the fitted time constants (τ₁∼0.8 ms and τ₂∼9 ms) were not voltage or pH dependent (supplementary material Fig. S1B). There was also no saturation of the outward current at strong depolarizations (Fig. 2A, supplementary material Fig. S1A,C).

In TEVC recordings, there was almost no inward current in an external solution containing 2 mmol l^–1 K⁺ and 98 mmol l^–1 Na⁺ (Fig. 3A,B and 3D, inset) and, even in elevated external K⁺, the currents did not show inward rectification (Fig. 3D, inset). Macropatch recordings in symmetrical K⁺ also failed to show significant inward current at pH 6.0 to 8.0; however, at hyperpolarized membrane potentials of –50 to –180 mV and pH 8.5, a small inward current developed (supplementary material Fig. S1B) but the current–voltage curve was still outwardly rectifying. Our data support the conclusion that AquK_2P is an outwardly rectifying channel with no apparent voltage dependence to the activation mechanism. Mechanisms of outward rectification have been revealed in some channel types, including the external Mg²⁺ block of NMDA receptor channels (Nowak et al., 1984), the differential influx or efflux of cations in TRPM7 channels (Penner and Fleig, 2007) and a mechanism intrinsic to the channel protein in TWIK-1 channels (Lesage et al., 1996).

To study the concentration–response relationship for AquK_2P current activation by AA, we applied increasingly higher concentrations of AA without recovery in between the concentrations; the dose–response relationship has a fitted EC₅₀ value of 30.6 μmol l^–1 with a Hill coefficient of 4.0 (Fig. 3C). Because there is a sensitization to prior AA applications (see Fig. 4), the steep slope could be explained by increasing levels of sensitization or perhaps a cooperative effect of AA on channel opening. The channels are mainly permeable to K⁺ because a plot of the current reversal potential versus the log of the K⁺ concentration yielded a linear relationship with a slope of 46 mV decade^–1, which is close to the theoretical value of 58 mV decade^–1 for a K⁺ selective channel (Fig. 3D). Even in elevated external K⁺, the currents did not show inward rectification (Fig. 3D, inset). The dose-dependent activation of AquK_2P by AA (Fig. 3C) did not change the current reversal potential (Fig. 3B).

As noted, the activation of AquK_2P by AA was regulated by external alkalinization (Fig. 4). AquK_2P currents were always activated by 10–100 μmol l^–1 AA when applied in an external recording solution of pH 8.5 or 9.0 and sometimes even at pH 8.0, if the AA was applied for a long duration (>5 min) or if this followed a prior activating event at alkaline pH values. However, only rarely (<5% of cells) were currents detected when up to 200 μmol l^–1 AA was applied at an external pH of 7.4. Independent of pH_o, AquK_2P current activation was reversed completely by external application of bovine serum albumin (BSA), a fatty acid binding protein, confirming that AA was required for activation (Fig. 4A,B).

A sensitization effect was apparent in all cells tested with external AA and alkaline pH. For example, the representative time course in Fig. 4A shows that AA-dependent current activation at pH_o 9.0 was enhanced following prior activation at pH_o 8.5 or 9.0 when compared with the effect of the first application of pH_o 9.0 to the cell. These changes occurred with complete washout of the pH effects and in the constant presence of AA. We also documented this sensitization using three successive applications of AA at pH_o 9.0, with interruptions by non-activating applications of AA at pH_o 7.4. The time course of the response to AA (Fig. 4B) shows that the K⁺ current is activated more quickly and the steady-state current amplitude is significantly increased by each successive application of alkaline pH (Fig. 4C), even with complete washout of the effect in between the applications of alkaline pH. The time course and reversibility of these effects argue against a perfusion artifact as an explanation for the sensitization. However, changes in external pH alone (in the range of 6.5 to 9.0) failed to activate any current in AquK_2P-expressing oocytes (Fig. 4D, N=5). None of the pH_o-regulated, AA-dependent AquK_2P currents showed differences in the current reversal potential or rectification in 2 mmol l^–1 external K⁺, even when the currents were sensitized by prior activation, as for the example shown in Fig. 4E. The most important conclusion is that AquK_2P subunits assemble to make functional K⁺ channels whose AA-dependent activation is regulated by external alkaline pH, but AquK_2P is not activated by external alkalinization alone.

Using inside-out macropatches, we tested the internal pH (pH_i) sensitivity of AquK_2P in the absence of AA. Fig. 5 shows that the channel is regulated by pH_i alone. Activity was very low at pH_i values that are expected in a resting oocyte (∼7.4) and was enhanced at alkaline pH_i values. An internal pH of 6.0 largely blocked the current. Although the alkaline-activated currents showed sensitization (see Fig. 6B), the current amplitude eventually reached a plateau level for each pH_i value (Fig. 5A). The alkaline pH_i-activated currents were outwardly rectifying in 2 mmol l^–1 external K⁺ and there was no change in the reversal potential upon activation by pH_i (Fig. 5B). A concentration–response curve for the pH_i sensitivity, determined after alkaline pH_i-activated currents had stabilized and sensitization was no longer apparent, had a pK_a of 8.18±0.05 (Fig. 5C).

In inside-out patches, AquK_2P channels were also activated by AA applied directly to the internal surface of the membrane (Fig. 6). The patches were first tested at pH 7.4 to mimic the expected internal pH of the oocyte in the TEVC recordings. The time course of current activation in membrane patches showed a delay upon first application of AA with a much shorter delay upon reapplication of AA (Fig. 6A). The second application of AA followed a reversal of the AA effect by BSA and then washout of BSA from the chamber.

The onset of activation of AquK_2P by AA was faster in the membrane patch (Fig. 6A) when compared with the whole cell (Fig. 4A), but the sensitization to a second application of AA was observed both in whole-oocyte recordings with AA applied externally and in excised inside-out patches with AA applied internally. We also tested for sensitization following internal pH-dependent gating in the absence of AA. As shown in Fig. 6B, AquK_2P channels became activated in a solution of pH_i 8.5 (time point 1 in Fig. 6B) and were then de-activated by the return to pH_i 6.0. The second and third application of pH_i 8.5 (time points 3 and 4 in Fig. 6B) yielded a faster onset of current and an increase in current amplitude with each application when compared with the first exposure to the alkaline internal solution. The sensitization was not accompanied by any changes in rectification or reversal potential (Fig. 6C). Most cells eventually reached a stable level of activation by alkaline pH_i and, under these conditions, could then be used for the dose–response relationship already presented in Fig. 5C.

Compared with AA-dependent activation, AquK_2P channels were relatively insensitive to mechanical forces within the membrane that were induced by osmotic shrinking or swelling of the oocytes (Fig. 7A,B), which we confirmed with a dissecting microscope. The small changes in AquK_2P current coincident with cell swelling (142 mOsmkg^–1) or shrinking (400 mOsmkg^–1) are not significantly different from each other and are only a small fraction of the current amplitude measured after 5 min of AA (Fig. 7A), a time point we selected for quantitation of the results. The same results were obtained in nine of nine cells (Fig. 7B) in which we always confirmed the ability to activate the sponge K⁺ channels by application of external AA (pH_o 9.0, 215 mOsm kg^–1) at the end of the experiment, as shown in Fig. 7A. In four additional cells (data not shown), we first reversibly activated AquK_2P currents with external AA (pH_o 9.0, 215 mOsm kg^–1) and then applied the same osmotic challenges at pH_o 7.4 and 9.0. Prior activation of AquK_2P current by AA did not reveal a mechanosensitivity of the sponge K_2P channels. To confirm that the brief osmotic challenges were sufficient to modify the activation of channels that are known to be mechanosensitive, we also tested oocytes expressing TREK-1 or TRAAK (Patel et al., 1998; Maingret et al., 1999a; Maingret et al., 1999b; Bang et al., 2000; Lesage et al., 2000; Dedman et al., 2009) and observed activation of the mammalian K_2P channels in hypoosmotic solutions and deactivation in hyperosmotic solutions (data not shown). Based on these results, we conclude that AquK_2P harbors little or no mechanosensitivity.

AquK_2P channels were not sensitive to temperature in the range of 14–32°C at pH 8.5 or 9.0. An example of the failure to activate AquK_2P currents with cool or warm temperatures is shown in Fig. 7C; the same observations were made in four of four cells in which we always confirmed expression of the K⁺ channels by activation with external AA at alkaline pH_o. For comparison, elevating the temperature from 24°C to 32°C results in a fivefold increase in mammalian TREK-1, TREK-2 and TRAAK currents (Kang et al., 2005).

We completed a physiological study of a structurally distinct sponge K_2P channel to learn about its mechanisms of activation and to compare it with other members of the K_2P channel family. K_2P channels are present throughout the animal kingdom, although we know very little about the functional properties of invertebrate and non-mammalian vertebrate channels. Outside the highly conserved TM1–TM4 domains and the two P regions (Fig. 1), AquK_2P is considerably different from any known K_2P channels and appears to be a distinct class, based on phylogenetic analysis (Fig. 2). The presence of a K_2P subunit in choanoflagellates (http://genome.jgi-psf.org/Monbr1/Monbr1.home.html; data not shown) suggests that these channels were present before the advent of multicellularity. Given the important role of K_2P channels in epithelia (Davis and Cowley, 2006) and the proposal that epithelial cells were the first specialized cells in the evolution of multicellularity (Adams et al., 2010; Fahey and Degnan, 2010), it is not surprising to find K_2P subunits represented in all animal phyla. Presently, functional expression of K_2P channels from non-mammalian species is lacking but could be useful in providing greater insight into the evolutionary relationships of these important ion channels.

Like many of the mammalian K_2P channels (Bayliss and Barrett, 2008; Enyedi and Czirják, 2010), the sponge channel likely plays a major role in setting the resting membrane potential in some poriferan cells. We found that AquK_2P channel activation has a requirement for internal alkaline pH or external alkaline pH plus AA, information that is biologically relevant because seawater has a pH of approximately 8.1 (Orr et al., 2005). Thus, the marine sponge's native environment provides an opportunity for basal activity of this potassium channel in situ and the coupling of changes in the cellular membrane potential to small changes in pH and the presence of polyunsaturated fatty acids. Furthermore, the outwardly rectifying nature of the AquK_2P currents suggests a possible role in the repolarization of action potentials, which could provide a physiological link between pH, AA and the frequency of firing of calcium-dependent action potentials in sponges (Leys et al., 1999).

The activation of AquK_2P by externally or internally applied AA (Figs 3, 4, 6) shows that the poriferan channel shares a functional feature unique to the TREK/TRAAK subfamily of K_2P channels (Fink et al., 1998; Enyedi and Czirják, 2010). Furthermore, the activity of AquK_2P does not require additional subunits, which differentiates it from K_2P5.1 (Duprat et al., 2007) and the electrically silent members K_2P7.1 (Salinas et al., 1999) and K_2P12.1 (Rajan et al., 2001). Our results are also consistent with prior work that attributes the structural basis for AA activation of mammalian TREK/TRAAK channels to an internally facing region of the channel (Patel et al., 1998; Kim et al., 2001a; Kim et al., 2001b) because the time course of AquK_2P activation was faster when the lipid-soluble AA was applied directly to the internal face of the membrane (Fig. 6) than when it was applied outside the whole cell (Fig. 4). The concentration dependence of AquK_2P channel activation by AA, determined with external application of the fatty acid, showed an EC₅₀ of ∼30 μmol l^–1 (Fig. 3C). The presence and concentration of endogenous AA in sponges is unknown, but free (non-esterified) AA in human tissues ranges from 1 to 100 μmol l^–1 as determined by mass spectrometry (Brash, 2001). Few examples of fitted EC₅₀ values for AA activation of mammalian K_2P channels exist in the literature, but TREK-2 channels are reported to have an EC₅₀ of 7 μmol l^–1 AA (Bang et al., 2000) and AA facilitates the constitutive activation of THIK-1 channels with an EC₅₀ of 1 μmol l^–1 (Rajan et al., 2001). The extracellular AA concentrations used in the present study were necessary to activate the AquK_2P channels and are not unreasonable given the concentrations utilized in other cellular physiology experiments and that be available in biological tissues.

Under the conditions tested, the AA-dependent activation of AquK_2P occurs in the absence of constitutive activation and without significant mechanosensitivity (Fig. 7A,B). We also found no substantial effects of temperature changes on AquK_2P current activation (Fig. 7C). Thus, compared with certain polymodal mammalian K_2P channels, particularly those stimulated by AA (Patel et al., 1998; Duprat et al., 2007; Lotshaw, 2007; Mathie et al., 2010), the mechanism of activation of AquK_2P is more restricted. The internally facing C termini of TREK/TRAAK channels are thought to provide both AA sensitivity and mechanosensitivity (Patel et al., 1998; Kim et al., 2001a; Kim et al., 2001b). Notably, the C-terminal region is structurally divergent among K_2P channels and AquK_2P has one of the shortest C termini (Fig. 1, from Q303 to E353). The identification in this paper of a K_2P channel that shows AA sensitivity but little or no mechanosensitivity could provide useful comparisons for future studies on the unique structural determinants for these two types of stimuli.

Extracellular pH is a common regulator of mammalian K_2P channels (Bayliss and Barrett, 2008; Enyedi and Czirják, 2010). For example, extracellular acidification inhibits TREK-1 with a pK_a of 7.35, inhibits TASK-3 with a pK_a of 5.96 (Rajan et al., 2000), and activates TREK-2 with a pK_a of 7.3 (Sandoz et al., 2009). Despite the differences in the pK_a values for proton inhibition or activation, each of these extracellular pH-regulated K_2P channel displays a resting K⁺ current at standard recording conditions of pH_o 7.4. Because we never observed constitutive activity of AquK_2P at pH_o 7.4 and the sponge channel current was only rarely activated by AA at pH_o 7.4, the effect of pH_o on AquK_2P can be described as alkaline regulation of channel activation by AA. Although TASK-2 (Morton et al., 2003; Zúñiga et al., 2011) and TALK channels (Bayliss and Barrett, 2008) are also activated by external alkalinization, they do not require a polyunsaturated fatty acid, which differs from the mechanism of pH_o regulation of AquK_2P (Fig. 3). The pK_a for external pH regulation of AA-dependent activation of AquK_2P is undetermined because of the sensitization to repeated activation of the channels. However, our data suggest that the midpoint of the pH_o sensitivity of the lipid activation is likely to be greater than 8.0. For example, we always observed activation by AA when pH_o was raised to 8.5, but AA-activated currents at pH_o 8.0 were very small unless tested after an initial sensitizing stimulus. The sponge K_2P channel would exist in a favorable environment of ∼pH 8.1 in seawater (Orr et al., 2005), in which small changes in pH_o could dynamically regulate channel opening by AA.

Sensitization of the activation of AquK_2P by AA in alkaline solutions was prominent in TEVC recordings (Fig. 4) and was present in excised patch recordings in which AA was applied rapidly and directly to the internal side of the membrane (Fig. 6A). Repeated applications of alkaline pH (Fig. 6B) to the inside surface of a membrane patch also induced a sensitized response, although current amplitudes in the macropatch recordings tended to reach a stable amplitude over time (often 5–30 min). Because sensitization eventually stabilized following inside-out patch excision, cytoplasmic agents or cytoskeletal elements that are lost upon excision may contribute to this phenomenon. However, the molecular mechanism for the sensitization is not yet defined.

AquK_2P currents are activated by alkaline internal pH with a pK_a of 8.18 (Fig. 5). The sensitivity to internal pH was measured with an external pH of 7.4, which is non-activating for AquK_2P. Under these conditions, internal alkalinization is sufficient to open the channels and AA is not required. Mammalian TREK/TRAAK channels are also regulated by internal pH. Internal alkalinization stimulates TRAAK whereas acidification stimulates TREK-1 and TREK-2 (Maingret et al., 1999b; Kim et al., 2001a; Honoré et al., 2002). However, whereas TREK and TRAAK channels demonstrate constitutive activity in whole oocytes studied by TEVC at pH_o 7.4, AquK_2P currents were negligible under the same conditions and less than 5% of the cells showed AA-dependent current activation at pH_o 7.4. The heterogeneity of cytosolic pH in intact oocytes and the internal pH sensitivity of AquK_2P may explain these observations. Collagenase-treated Xenopus oocytes were shown to have a resting cytosolic pH range of 7.06–7.93, with an average pH of 7.43 (Cicirelli et al., 1983). At an average internal pH of 7.4, we would expect 25% of the maximum current activation based on our inside-out patch clamp recordings (Fig. 5C). Thus, we hypothesize that most of the intact oocytes tested in the TEVC recordings in this study had a resting cytosolic pH less than 7.4, which was insufficient to reveal constitutive channel activation. Further investigation will be needed to determine the mechanisms by which internal alkalinization and AA gate the AquK_2P channel.

The marine sponge AquK_2P channel is AA activated and pH sensitive, properties that are shared by certain mammalian K_2P channels and highlight the conservation of basic gating mechanisms for animal K_2P channels. A requirement for AquK_2P activation by internal alkalinization or the modulatory effects of external alkalinization is also biologically relevant because seawater has a pH of ∼8.1. Thus, the native environment of the marine sponge may provide an opportunity for basal activity of this potassium channel in situ and the coupling of small changes in pH and the presence of polyunsaturated fatty acids to regulation of the cellular membrane potential and action potential repolarization. Finally, AA-dependent channel activation without osmotically induced activation makes AquK_2P unique among K_2P channels and provides an opportunity for the future determination of the structural regions of the channel that are necessary for AA sensitivity in the absence of mechanosensitivity.

 
• AquK_2P

  Amphimedon queenslandica K_2P channel

 
• AA

  arachidonic acid

 
• BSA

  bovine serum albumin

 
• K_2P

  tandem pore domain potassium channel

 
• MES

  methansulfonate

 
• pH_i

  internal/intracellular pH

 
• pH_o

  external/extracellular pH

 
• TEVC

  two-electrode voltage clamp

We thank Doug Bayliss for the gifts of TREK and TRAAK channel cDNAs, Bernard Degnan for supplying larvae from Amphimedon queenslandica, and April Hill for advice on searching the Amphimedon genome. Thanks to Onur Sakarya and Ken Kosik for help with primer design for cloning, undergraduates in the Boland laboratory for help with RNA and oocyte preparations, and Heikki Vaananen and Sophia Gruszecki for oocyte preparation in the Logothetis laboratory.