Anionic Phospholipids Bind to and Modulate the Activity of Human TRESK Background K+ Channel

The background K+ channel TRESK regulates sensory neuron excitability, and changes in its function/expression contribute to neuronal hyperexcitability after injury/inflammation, making it an attractive therapeutic target for pain-related disorders. Factors that change lipid bilayer composition/properties (including volatile anesthetics, chloroform, chlorpromazine, shear stress, and cell swelling/shrinkage) modify TRESK current, but despite the importance of anionic phospholipids (e.g., PIP2) in the regulation of many ion channels, it remains unknown if membrane lipids affect TRESK function. We describe that both human and rat TRESK contain potential anionic phospholipid binding sites (apbs) in the large cytoplasmic loop, but only the human channel is able to bind to multilamellar vesicles (MLVs), enriched with anionic phospholipids, suggesting an electrostatically mediated interaction. We mapped the apbs to a short stretch of 14 amino acids in the loop, located at the membrane-cytosol interface. Disruption of electrostatic lipid-TRESK interactions inhibited hTRESK currents, while subsequent application of Folch Fraction MLVs or a PIP2 analog activated hTRESK, an effect that was absent in the rat ortholog. Strikingly, channel activation by anionic phospholipids was conferred to rTRESK by replacing the equivalent rat sequence with the human apbs. Finally, in the presence of a calcineurin inhibitor, stimulation of a Gq/11-linked GPCR reduced hTRESK current, revealing a likely inhibitory effect of membrane lipid hydrolysis on hTRESK activity. This novel regulation of hTRESK by anionic phospholipids is a characteristic of the human channel that is not present in rodent orthologs. This must be considered when extrapolating results from animal models and may open the door to the development of novel channel modulators as analgesics.


Abstract:
The background K+ channel TRESK regulates sensory neuron excitability and changes in its function/expression contribute to neuronal hyperexcitability after injury/inflammation, making it an attractive therapeutic target for pain-related disorders.
Factors that change the plasma membrane bilayer composition/properties (including volatile anesthetics, chloroform, chlorpromazine, shear stress and cell swelling/shrinkage) modify TRESK current but despite the importance of anionic phospholipids (e.g. PIP2) in the regulation of many ion channels, it remains unknown if membrane lipids affect TRESK function. We describe that both human and rat TRESK contain potential anionic phospholipid binding sites (apbs) in the large cytoplasmic loop, but only the human channel is able to bind to multilamellar vesicles (MLVs), enriched with anionic phospholipids, suggesting an electrostatically-mediated interaction. We mapped the apbs to a short stretch of 14 amino acids in the loop, located at the membrane-cytosol interface. Disruption of electrostatic lipid-TRESK interactions inhibited hTRESK currents, whilst subsequent application of Folch Fraction MLVs or a PIP2 analog activated hTRESK, an effect that was absent in the rat ortholog. Strikingly, channel activation by anionic phospholipids was conferred to rTRESK by replacing the equivalent rat sequence with the human apbs. Finally, stimulation of a Gq/11-linked GPCR reduced hTRESK current when Ca2+/calcineurin is blocked, while in physiological conditions, the Ca2+-mediated stimulation is prominent. This novel regulation of hTRESK by anionic phospholipids is a characteristic of the human channel that is not present in rodent orthologs. This must be considered when extrapolating results from animal models and may open the door to the development of novel channel modulators as analgesics.

INTRODUCTION
TWIK-Related Spinal cord K + channel (TRESK) or K2P18.1 (encoded by the KCNK18 gene) is a member of the Two-pore domain potassium channel (K2P) family, which contains 15 members with a shared molecular architecture. K2P channel proteins are comprised of four transmembrane domains and two poreloop forming domains with intracellular N-and C-termini [1]. Four pore-loops are required to form a functional K + selectivity filter and therefore K2P channels function as dimers, unlike the other K + channel subfamilies (Voltage-gated, Calcium-activated and Inwardly rectifying K + channels), which function as tetramers [2,3]. K2P channels play an important role in the maintenance and stabilization of the resting membrane potential and, in excitable cells such as neurons and cardiac myocytes, they modulate the shape and frequency of action potentials. They have been implicated in a wide range of physiological processes including nociception, somatosensation, nutrient and chemo-sensing, hormone secretion, sleep and anesthesia [4][5][6][7].
The coding sequence of the human KCNK18 gene was originally identified by a homology search of the draft human genome using the amino acid sequence of a previously identified K2P channel, K2P1.1 (also known as TWIK1, Tandem of Pore domains in a Weak Inwardly-rectifying K + channel), which led to its cloning from human spinal cord mRNA [8]. TRESK is selectively expressed in a subpopulation of sensory neurons of the dorsal root (DRG) and trigeminal ganglia (TG), which innervate peripheral regions of the body and are responsible for the detection of both innocuous (e.g. touch) and noxious chemical, thermal and mechanical stimuli [9][10][11][12][13][14]. Studies on primary cultures demonstrated that TRESK accounts for a significant proportion of the "resting" or background K + conductance in DRG neurons from mouse and rat [11,15]. The first evidence of a role for TRESK in the regulation of sensory neuron excitability came when DRG neurons from a mutagenized mouse strain carrying a loss-of-function mutation in the Kcnk18 gene were compared with wild-type DRG neurons, revealing that the TRESKmutant neurons were easier to excite by depolarizing stimuli [11]. Subsequent studies have demonstrated that pharmacological inhibition of TRESK by alkylamides, such as hydroxy-α-sanshool and isobutylalkylamide (IBA) or pyrethroids produce sensory neuron activation and contributes to associated somatosensory and nocifensive behaviors [9,10,16,17]. A dominant negative loss-of-function mutation in the KCNK18 gene is linked to familial migraine with aura and over-expression of mouse TRESK with the equivalent mutation in trigeminal neurons was shown to cause hyperexcitability, suggesting that TRESK is required for correct regulation of trigeminal neuron excitability [18,19]. Changes in TRESK functional expression may be a contributing factor to the changes in sensory neuron excitability observed during inflammation and neuropathic pain since Kcnk18 mRNA and protein expression are decreased in rat models of neuropathic pain [10,20] and TRESK activity is inhibited in the presence of inflammatory mediators such as arachidonic acid [8,21]. Increased TRESK activity as a consequence of overexpression was shown to dampen the excitability of TG and DRG neurons and could attenuate nerveinjury induced allodynia, leading to the suggestion that increasing TRESK activity may be a viable therapeutic strategy for the treatment of pain-related conditions [20,[22][23][24]. A better understanding of the underlying mechanisms that influence TRESK activity will be crucial for the rational design of novel channel modulators.
A feature of TRESK is the modulation of its activity by stimuli that produce changes in the composition and/or properties of the plasma membrane. In a recent study from our laboratory we demonstrated that TRESK activity can be modulated by experimentally applied shear stress, changes in osmotic pressure and by compounds that affect membrane bilayer properties, such as chloroform and chlorpromazine [21].
Volatile (e.g. isofluorane) and local (e.g. bupivacaine) anesthetics are potent enhancers and inhibitors of TRESK currents, respectively [25]. Although the precise molecular mechanisms of action are unclear, volatile anesthetics appear to modify ion channel activity via the combined effects of binding to amphiphilic sites on channel proteins and by incorporation into lipid bilayers [26].
TRESK currents are strongly potentiated by increases in intracellular Ca 2+ -concentration via Ca 2+dependent binding of the activated phosphatase calcineurin to the channel, which mediates dephosphorylation of inhibitory phosphorylation sites [27,28]. This mechanism underlies the enhancement of TRESK activity observed upon Gq/11-coupled receptor stimulation [27,29,30]. However, Gq-coupled receptor activation can also produce large decreases in the plasma membrane concentration of the phosphoinositide Phosphatidyl-inositol-4,5-bisphosphate (PIP2) via the stimulation of phospholipase C (PLC) activity, which hydrolyses PIP2 to produce diacylglycerol (DAG) and IP3 [31][32][33]. It is also of note that large influxes of Ca 2+ via TRP channels that co-express with TRESK in sensory neurons have also been shown to have significant effects on global PIP2 levels via activation of Ca 2+ -activated PLC isoforms, for example the PLCδs, which are expressed in sensory neurons [34,35]. Given the importance of PIP2 in the regulation of the activity of a wide variety of ion channels and transporters [36] and, in particular, the evidence for the role of PIP2 in the regulation of other K2P channels [37][38][39], it is feasible that changes in plasma membrane PIP2 concentration may also contribute to physiological modulation of TRESK activity.
It is clear that the membrane environment and changes in lipid composition play an important role in the modulation of TRESK. However, there is no information to date regarding interactions between TRESK and membrane phospholipids and the consequences for channel function. In this study, we have investigated whether TRESK contains binding sites for membrane phospholipids and whether these interactions play a role in the modulation of channel activity.

In silico search for putative anionic phospholipid binding sites
The primary sequences of human (Genbank accession number NP_862823.1) and rat (Genbank accession number AAS68516.1) TRESK were analyzed using the BH search program. The threshold BH value for residues forming part of a potential membrane binding site was 0.6 and the search was performed with a window size of 10 to obtain scores for N-and C-terminal residues (Brzeska et al., 2010) [40]. A search for the most common putative lipid-binding domains (the PH, PKC C1, PKC C2, PX, FYVE, GLA, GRAM, F-BAR and ENTH domains) in human and rat TRESK was undertaken using the SMART program [41,42].

Molecular Biology
Rat TRESK in the pcDNA3.    Liposome binding assay -All steps were performed at room temperature. 5g of purified GST fusion protein was incubated with 100 g liposomes in a total volume of 150l for 30 minutes followed by centifugation at 21,000g for 1 hour. After centrifugation, the supernatant was removed and 150l 2x Laemmli loading buffer (250mM Tris HCl pH6.8, 40% (v/v) glycerol, 8% (w/v) SDS, 0.008% Bromophenol blue, 20% (v/v) -mercaptoethanol) was added. The pellet was resuspended in 150l liposome binding buffer followed by addition of 150l 2x Laemmli loading buffer. Samples were incubated at 95 o C for 3 minutes before equivalent amounts of supernatant and pellet were analyzed by SDS-PAGE followed by Coomassie staining.

Statistical analysis
Data are presented as mean±s.e.m. Statistical differences between different sets of data were assessed by performing paired or unpaired Student's t-test as indicated. Statistical significance was set at *p<0.05; **p<0.01 and ***p<0.001.

Identification of putative binding sites for anionic phospholipids in TRESK using an in silico approach
In general, protein regions that associate with biological membranes do so via binding to anionic/acidic phospholipids, displaying a wide variation in binding mechanisms [44]. In initial studies, we used an in silico approach to search for putative lipid-binding domains in both the human and rat isoforms of TRESK. A search for the most common putative lipid-binding domains (the PH, PKC C1, PKC C2, PX, FYVE, GLA, GRAM, F-BAR and ENTH domains) in TRESK using the SMART program [41,42] yielded no positive results, indicating that it is unlikely that TRESK contains lipid-binding domains with a well-defined tertiary structure. However, there are numerous examples of K + channels lacking these structurally defined lipid-binding domains that can still bind to acidic/anionic phospholipids via clusters of basic and hydrophobic amino acids [37,[45][46][47][48]. With this in mind, we analyzed the sequences of human and rat TRESK using BH search, a program that identifies putative membrane-binding sites on the basis of basic and hydrophobic amino acid content (available at http://helixweb.nih.gov/bhsearch) [40].
The results of this analysis are shown in Figure

The intracellular loop of human TRESK contains a site for binding to anionic phospholipids
To analyze whether the anionic phospholipid binding sites identified in our in silico analysis could mediate interaction of TRESK with membrane lipids, we assayed for the ability of these regions of human and rat TRESK containing the putative sites to bind to multilamellar vesicles composed of Folch fraction I lipids, an organic extract of bovine brain enriched in phosphoinositides and phosphatidylserine [49]. GST-rTRESK(185-267) (12.38±3.06%, n=4; p<0.001, unpaired two-tailed t-test). It should also be noted that both fusion proteins displayed significantly more pelleting than GST alone (GST pelleting = 1.4±1.2%, n=9; GST vs. GST-hTRESK p<0.0001; GST vs. GST-rTRESK p=0.0017, unpaired two tailed t-tests).
These in vitro observations demonstrate that the intracellular loop domain of human TRESK can interact with anionic phospholipids.

Folch Multilamellar Vesicles activate human TRESK currents
After identification of the large intracellular loop of TRESK as a potential mediator of the interaction between the channel and anionic phospholipids we went on to examine the functional consequences of disrupting electrostatic interactions between the channel and anionic phospholipids in the lipid bilayer.
As a positive control, we assayed mouse TREK-1 (K2P2.1), another K2P channel that has previously been shown to be modulated by anionic phospholipids [37,38].

Mapping the anionic phospholipid binding site in the cytoplasmic loop of human TRESK
The in silico analysis of human TRESK predicted that the site responsible for binding to anionic phospholipids resides between residues 163 and 191 (Fig. 1). Consistent with this, a GST-fusion protein containing this site showed significant interaction with Folch MLVs (Fig 2C). On closer examination of the amino acid sequence of the putative anionic phospholipid-binding site, two separate clusters of positively charged amino acids could be distinguished. One of these clusters lies between amino acids 163 and 177 ( 163 YNRFRKFPFFTRPLL 177 , denoted "Cluster 1" in Fig. 4A) and the other between amino acids 177 and 191 ( 177 LSKWCPKSLFKKKPD 191 , denoted "Cluster 2" in Fig. 4A). We therefore designed GST fusion proteins to test whether one or both of these stretches were necessary for interaction with Folch MLVs. As previously shown (Fig. 2C) MLVs.

Testing the role of the phospholipid-binding cluster in the cytoplasmic loop of human TRESK using human-rat TRESK chimeric constructs
The liposome binding data shown in Figure 4 strongly suggested that a stretch of amino acids lying between residues 163 and 177 of human TRESK is essential to mediate a physical interaction with anionic phospholipids. We then attempted to investigate the role of this stretch of amino acids in the modulation of TRESK activity by designing a series of chimeras between human and rat TRESK. It is interesting to note that alignment of the anionic phospholipid binding stretch we have identified in human TRESK with the corresponding sequence in rat TRESK revealed significant differences (Fig. 5A). These differences may explain why the loop of human TRESK displays a much stronger binding to Folch MLVs compared to the rat loop (see Fig. 2). Of the four positively charged residues in the human stretch, only two are conserved in the rat sequence (R186 and R195). Furthermore, there is an aspartate residue in the rat sequence (D196), which would be negatively charged at physiological intracellular pH and therefore likely to interfere with anionic phospholipid-based interactions. As shown in Figure 5B, two chimeras between human and rat TRESK were initially constructed. In the first chimera, the large intracellular loop, the 3th and 4th transmembrane domains of hTRESK were replaced with the corresponding sequence from rTRESK (hTRESK-rat-loop-rTRESK). The complementary second chimera was constructed in the same way (rTRESK-human-loop-hTRESK). The effects of poly-L-lysine followed by Folch MLV application to inside-out patches of HEK293 cells expressing each chimera were tested. In both cases, no significant effects of either poly-L-lysine or Folch MLVs on channel currents were observed (Fig. 5C). The results of this analysis suggest that, despite the requirement for the stretch of amino acids between 163 and 177 in hTRESK to mediate binding to anionic phospholipids in a liposome binding assay, this region appears to be necessary but not sufficient to confer modulation by anionic phospholipids on hTRESK, suggesting that other parts of the human channel are also required.
A second set of chimeras was also tested where only the anionic phospholipid binding site (apbs) from human TRESK replaced the equivalent sequence in rat TRESK (rTRESK-h-apbs) or vice versa (hTRESK-r-apbs). The effects of poly-L-lysine followed by Folch MLV application to inside-out patches of HEK293 cells expressing each chimera were tested. As expected from liposome binding data in Figure   2, replacement of the apbs in hTRESK by the rat sequence produced a chimeric channel unresponsive to poly-lysine (-6.0±7.2%, n=7) or Folch MLV application (5.7±5.2%, n=7; Fig. 5D, E). In contrast, when the apbs in rTRESK was replaced by the human one, the channel acquired responsiveness to poly-lysine (-10.0±8.9% n=11; p<0.05) and was potentiated by Folch MLV application (128.7±45.2% n=11; p<0.05; Fig. 5D, E), similarly to what was observed in the wild-type human channel.

Phosphatidylinositol-4,5-bisphosphate (PIP2)
To determine whether the human TRESK intracellular loop displayed preferential binding for a particular type of anionic phospholipid and to further analyze potential differences between the lipid binding characteristics of the loops from the rat and human channels, we performed pelleting experiments using multilamellar liposomes of defined composition. GST-hTRESK(163-244) pelleted strongly in the presence of liposomes enriched with anionic phospholipids, either 50% phosphatidylcholine (PC)/50% phosphatidylserine (PS) (Fig. 6A, lane 6) or 70% phosphatidylcholine/25% phosphatidylserine/5% phosphatidylinositol-4,5-biphosphate (PIP2; Fig. 6A, lane 8) but did not show significant pelleting with liposomes composed solely of the neutral lipid PC (Fig. 6A, lane 4). Quantification of pelleting from 3 independent experiments showed significant differences between liposomes composed solely of PC and those enriched with anionic phospholipids (PS and PIP2). However, no difference in pelleting was observed between the two types of liposomes enriched with anionic phospholipids (PS vs PIP2) (Fig. 6C).
In contrast, GST-rTRESK(185-267) did not show appreciable specific pelleting with any of the liposomes tested (Fig. 6B, lanes 4, 6 and 8). We attempted to further characterize the lipid specificity of the human TRESK loop using commercially available strips spotted with various types of lipid (PIP strips from Thermo Fisher Scientific). However, no firm conclusions could be drawn due to a high level of GST binding to lipid spots (data not shown). The results of these liposome-pelleting experiments indicate that the intracellular loop of human TRESK contains a binding site for anionic phospholipids that does not display a preference for a specific lipid type.

PIP2 enhances human TRESK currents
It is likely that the anionic phospholipid-binding site in the large cytoplasmic loop of human TRESK will bind preferentially to the phosphoinositide PI(4,5)P2 under physiological conditions in native plasma membranes, on the basis of its trivalency at pH 7 and its concentration in the plasma membrane (approximately 1% of total lipid). In addition, physiologically relevant changes in plasma membrane PI(4,5)P2 concentration can occur during Gq/11-coupled receptor stimulation by agonists such as hormones and neurotransmitters [54]. Therefore, we tested the effect of adding a water-soluble form of PI(4,5)P2 (diC8:0 PI(4,5)P2) to inside-out patches taken from HEK293 cells transiently expressing hTRESK. As displayed in Figure 7, in a similar fashion to Folch MLVs, diC8:0 PI(4,5)P2 (5 M) produced a significant activation of hTRESK previously inhibited by 1μg/ml poly-L-lysine in 7 out of 9 patches tested (% increase: 99.9±39.8%, n=7; p<0.05; Fig. 5C), while no effect was found in two additional patches.
As previously described, changes in membrane PI(4,5)P2 concentration after activation of GPCRs modulate the activity of several ion channels and membrane proteins [31,32,[55][56][57]. It is well known that TRESK is modulated by changes in intracellular Ca 2+ through dephosphorylation by calcineurin [27]. In view of our results, we explored the possibility that a dual regulation of hTRESK by PIP2 and  Fig 7E,F), similarly to what has been previously described [29,58]. This indicates that, at least after stimulation of mGluR5 receptors, the activating effect of Ca 2+ /calcineurin is more important than the lipid-mediated inhibition. Interestingly, in 50% of the recordings (6 of 12) a biphasic effect was observed: an initial and transient decrease in TRESK current followed by a sustained increase, which could reflect an initial hydrolysis of PIP2 from the membrane followed by the stimulatory effect of Ca 2+ /calcineurin on hTRESK.

DISCUSSION
In this study we present evidence that argues for a species-specific modulation of TRESK by anionic phospholipids, which requires a juxtamembrane region in the large intracellular loop of the channel. It is well established that other members of the K2P family (TREK-1, -2, TASK-1, -3) are activated by anionic phospholipids, with modulation by changes in local PIP2 concentration appearing to be particularly important [37-39, 59, 60]. In the case of TREK-1 (K2P2.1), a short sequence of amino acids acts as a type of "phospholipid sensor", which links the activity of the channel to changes in PIP2 concentration [37]. We used an in silico approach in an attempt to identify regions in TRESK that could fulfill a similar role (Fig. 1). After identifying potential interacting regions, we used a liposome-binding assay to test their ability to bind to multilamellar vesicles enriched with anionic phospholipids (Folch but not to liposomes composed solely of the neutral phospholipid, phosphatidylcholine. Several ion channels and, in particular, potassium channels including Kir, Kv, KCNQ, KCa are regulated by phosphoinositides, including PIP2 (for review see: [64]). In fact, other channels of the K2P family are regulated by different phospholipids as well as by PIP2 [37][38][39][65][66][67]. Here, we found that exogenous PIP2 increased hTRESK activity in isolated inside-out patches. A closer inspection of the putative anionic phospholipid-binding region in hTRESK revealed the presence of two "clusters" of positively-charged amino acids, which may be responsible for mediating binding to anionic phospholipids. We tested the role of these clusters in phospholipid binding using the liposome binding assay and determined that only one of these clusters is necessary to mediate binding to anionic phospholipids (Fig. 4). However, there is a possible smaller contribution from cluster 2 as the GST-hTRESK(177-244) shows significant binding above GST and significantly more than the fusion protein lacking both clusters, GST-hTRESK(187-244). Rat TRESK seems to have cluster 2, which may explain the residual binding activity observed in the pelleting assay (Fig. 2F).
In our experiments, hTRESK is able to bind to PIP2 to increase the channel current. In TREK-1, it has been described that when endogenous phospholipids are "quenched" by poly-cationic poly-lysine, mechano-sensitivity is significantly reduced [37]. Also, PIP2 exerts a dual regulation of TREK-1 activation by pHi or mechanical stimuli. In the presence of PIP2, mechanosensitivity or activation by pHi is reduced but in the presence of poly-lysine, the effects are reverted and PIP2 can promote activation of the channel [37][38][39]. TRESK is not directly activated, but modulated by membrane stretch or by changing membrane curvature [21], in contrast to TREK-1 or -2 that can be directly activated by mechanical stimuli. It remains to be explored if stretch modulation of hTRESK current is enhanced or diminished by the binding to membrane phospholipids. Another possibility is that changes in the intracellular pH (pHi) might produce protonation of some residues to modify the interaction with membrane lipids. In TREK-1 channels at acidic pHi, exogenous phospholipids transform the mechano-gated channel into a K +selective leak conductance [37]. Also, blocking the interaction between the C-terminal domain and membrane lipids by addition of poly-lysine avoids its activation by acidic pHi [37,38]. In hTRESK, acidic pHi produces the opposite effect than in TREK-1 and decreases the current while alkalinization increases it [8]. Whether these effects could be mediated by interfering with the interaction of the channel domain with membrane lipids or if alterations in the pHi affecting the binding to membrane lipids changes the biophysics of the channel remains to be studied.
Other K2P channels are modulated by hormones and neurotransmitters through activation of different GPCRs linked to Gs-cAMP-PKA phosphorylation, Gq-PLC-DAG-PKC phosphorylation or PIP2 membrane depletion (for review see [3]). Interestingly, TRESK is the only member of the family regulated by Ca 2+ /calcineurin dephosphorylation, which increases the channel current [27]. Therefore, TRESK is activated by increasing the cytosolic Ca 2+ through membrane ion channels or after stimulation of Gq/11-coupled receptors and Ca 2+ release from internal stores [27,29,30]. This overall activation is likely to be the sum of two consequences of receptor activation: a Ca 2+ -dependent activation of TRESK by calcineurin and an inhibition due to PIP2 hydrolysis in the vicinity of the channel. Here we show that activation of a membrane GPCR linked to Gq/11 activation and Ca 2+ release (mGluR5) produces a significant decrease in channel current likely due to PIP2 hydrolysis when the Ca 2+ -mediated effects are blocked (calcium buffer + inhibitor of calcineurin). In contrast, when both effects are present, the overall effect is an increase of hTRESK current, indicating that the Ca 2+ /calcineurin effect is more prominent.
During inflammation or nerve injury, a mix of chemicals is released producing the sensitization of nociceptive sensory neurons and contributing to chronic pain [68]. These chemicals, in many cases, activate pronociceptive GPCRs [68,69]. Because many pronociceptive receptors signal through PLC (and require PIP2 for signaling), the modulation of hTRESK by membrane lipids might be relevant. Enhancing effects of Ca 2+ on TRESK current have been described after stimulation with several neurotransmitters or inflammatory mediators such as bradykinin, 5-HT, glutamate, lysophosphatidic acid (LPA), histamine or muscarinic agonists [29,30,58]. All these studies used rodent (mostly mouse) TRESK orthologs where the Ca 2+ -calcineurin mediated TRESK activation would be more prominent due to the lack of lipid binding and regulation. In contrast, in the human channel, a combination of both effects will be present.
Depending on the receptor activated and the PIP2 hydrolysis in restricted membrane microdomains, the excitatory or inhibitory effects on TRESK could prevail. In fact, several differences have been reported between the rat/mouse and human TRESK channels, including the sensitivity to anesthetics (higher enhancing efficacy to isoflurane and less blocking effect of lidocaine in hTRESK) [70], the docking of calcineurin to the channel and their sensitivity to Ca 2+ regulation [71] or the sensitivity to Zn 2+ ions and pH [70,72]. The fact that human and rodent coding sequences share only about 65% identity and 71% overall amino acid similarity is likely to be the reason for this different modulatory effect, including the lipid regulation described here. These species differences must be taken into consideration when extrapolating pharmacological results from rodent experimental models to human physiology or during drug development.
It has been proposed that in sensory neurons, TRESK counteracts membrane depolarization induced by external stimuli or diverse chemical substances in order to prevent excessive neuronal activation [3,10,11,17,30]. A recent report shows combined enhancement of TRESK and TRPV1 currents by LPA, an inflammatory mediator, with potentiation of hyperpolarizing TRESK currents counteracting to some extent the depolarizing effects derived from TRPV1 [30]. In fact, TRESK seems to be modulated by different stimuli, including Ca 2+ , membrane curvature or arachidonic acid [8,21,27]. As an example, during inflammation, both LPA, arachidonic acid and hypertonic conditions can be present, the first one enhancing TRESK current with the latter two conditions decreasing TRESK current [21,30]. It must be noted that, at least for the human channel, while activation of GPCRs will produce a modulation of TRESK current through membrane lipids and by Ca 2+ /calcineurin, activation of TRESK via Ca 2+ influx through direct activation of membrane channels such as TRPV1 or TRPA1 will only produce the last effect. This would suggest that depending on the stimuli activating the sensory neuron, the role of TRESK counterbalancing the depolarizing effect of the stimulus might be different. Taken together, it appears that depending on the stimuli, TRESK can be modulated negatively or positively and the final amount of hyperpolarizing current counterbalancing depolarization might vary in different situations.
Here we show that membrane lipids also contribute to this regulation, at least in the human channel. In addition, regulation of TRESK expression is another factor to be added into the equation, since injury or inflammation produce a down-regulation of the channel expression [10,73]. Because TRESK represents a common regulation point by all these stimuli, it can be postulated as an interesting way to bypass the membrane receptor diversity and to target a point where different signaling pathways converge. In this way, analgesic drug development to enhance TRESK activity can result as an interesting approach to treat pain.
In summary, here we describe a novel regulation of human TRESK by anionic membrane lipids through interaction with a motif in the intracellular loop of the channel. This regulation is unique in the human channel and not present in rodent orthologs. This finding further expands the different properties of the human channel, which should be taken into account when extrapolating results from rodent models. Also, it opens the door to possible new drug developments that target this channel for the treatment of pain.       Click here to download Figure Fig 3.eps Click here to download Figure Fig 4.  Click here to download Figure Fig 6.tif Click here to download Figure Fig 7.eps