Modulation of KCa3.1 Channels by Eicosanoids, Omega-3 Fatty Acids, and Molecular Determinants

Background Cytochrome P450- and ω-hydrolase products (epoxyeicosatrienoic acids (EETs), hydroxyeicosatetraeonic acid (20-HETE)), natural omega-3 fatty acids (ω3), and pentacyclic triterpenes have been proposed to contribute to a wide range of vaso-protective and anti-fibrotic/anti-cancer signaling pathways including the modula-tion of membrane ion channels. Here we studied the modulation of intermediate-conductance Ca2+/calmodulin-regulated K+ channels (KCa3.1) by EETs, 20-HETE, ω3, and pentacyclic triterpenes and the structural requirements of these fatty acids to exert channel blockade. Methodology/Principal Findings We studied modulation of cloned human hKCa3.1 and the mutant hKCa3.1V275A in HEK-293 cells, of rKCa3.1 in aortic endothelial cells, and of mKCa3.1 in 3T3-fibroblasts by inside-out and whole-cell patch-clamp experiments, respectively. In inside-out patches, Ca2+-activated hKCa3.1 were inhibited by the ω3, DHA and α-LA, and the ω6, AA, in the lower µmolar range and with similar potencies. 5,6-EET, 8,9-EET, 5,6-DiHETE, and saturated arachidic acid, had no appreciable effects. In contrast, 14,15-EET, its stable derivative, 14,15-EEZE, and 20-HETE produced channel inhibition. 11,12-EET displayed less inhibitory activity. The KCa3.1V275A mutant channel was insensitive to any of the blocking EETs. Non-blocking 5,6-EET antagonized the inhibition caused by AA and augmented cloned hKCa3.1 and rKCa3.1 whole-cell currents. Pentacyclic triterpenes did not modulate KCa3.1 currents. Conclusions/Significance Inhibition of KCa3.1 by EETs (14,15-EET), 20-HETE, and ω3 critically depended on the presence of electron double bonds and hydrophobicity within the 10 carbons preceding the carboxyl-head of the molecules. From the physiological perspective, metabolism of AA to non-blocking 5,6,- and 8,9-EET may cause AA-de-blockade and contribute to cellular signal transduction processes influenced by these fatty acids.

Here, we hypothesized that structurally related omega-3 fatty acids (v3), docosahexaenoic acid (DHA) and a-linolenic acid (a-LA), the cytochrome-P450-epoxygenase (CYP)-generated metabolites of AA, epoxyeicotrienoic acids (5,6-EET, 8,9-EET, 11,12-EET, and 14,15-EET) as well as the v-hydroxylase product, 20-hydroxyeicosatetraeonic acid , are additional lipid modulators of K Ca 3.1. Moreover, epoxygenation of AA to 5,6-EET, 8,9-EET, 11,12-EET, or 14,15-EET may shed light on the structural requirements for channel modulation. In addition, a potential K Ca 3.1-regulation by EETs, 20-HETE, and v3 could be of help to understand the physiological actions of these fatty acids in physiological systems like the vascular endothelium and arteries, in which they have been shown to exert vasodilator or vasoconstrictor actions, respectively (for review see [28,29,30]). Moreover, EETs and v3 have been proposed to have antiinflammatory and anti-atherosclerotic activity and to modulate angiogenesis, cardiac fibrosis and cancer growth [31,32,33,34,35]. In this respect, EETs and K Ca 3.1-functions have overlapping impacts and may be mechanistically linked as components of the same signal transduction pathway(s). Today, several downstream targets such as G-protein-coupled receptors have been proposed to mediate EET-actions but specific receptors for EETs, HETEs, as well as for v3 are still elusive (for review see [30,31]). So far it is unknown whether these fatty acids modulate hK Ca 3.1-functions.
In addition to these fatty acids, we tested whether lipids of the pentacyclic triterpene class, uvaol, erythrodiol, oleanolic acid, and maslinic acid, exert K Ca 3.1-modulatory actions. These natural triterpenes are found in virgin olive oil and have been suggested having antioxidant, antifibrotic, anti-atherosclerotic, and, both, pro-as well as anti-inflammatory activities [35,36,37,38]. However, whether these actions are related to -at least in part -K Ca 3.1modulation has not been studied before.
We therefore conducted an electrophysiological study on cloned hK Ca 3.1 and endothelial rK Ca 3.1 and studied channel modulation by selected v3, the four EETs, and 20-HETE, synthetic stable analogues, and other related fatty acids with structural differences or similarities (for structures see Figure 1). To further study potential binding/interaction sites within the K Ca 3.1 channel, we investigated blocking efficacy of the fatty acids on the AAinsensitive K Ca 3.1-mutant V275A [25]. Moreover, we studied the interactivity of EETs with its precursor, AA. In murine fibroblast, we tested the modulation of mK Ca 3.1 by DHA and by pentacyclic triterpenes.
Our major findings were that the 14,15-EET, 20-HETE, DHA, and a a-LA, were negative modulators of K Ca 3.1 while nonblocking 5,6-EET antagonized AA-mediated inactivation. KCa3.1 blockade critically depended on hydrophobicity of the 10 carbons preceding the carboxyl head and the presence of at least one electron double bond in this part of the carbon chain.

Materials and Methods
Cells, channel clones, and cell culture HEK-293 cells stably expressing hK Ca 3.1 were a kind gift from Dr. Khaled Houamed, University of Chicago and Dr. Heike Wulff, Department of Pharmacology, University of California, Davis. Stably expressing cells were selected with puromycin (1 mg/ ml; Sigma, Deisenhofen, Germany). The hK Ca 3.1 V275A , hK Ca 3.1 T250S , and hK Ca 3.1 T250S/V275A mutants were kind gifts from Dr. Dan Devor, University of Pittsburgh, Department of Cell Biology. The clones were stably expressed in HEK-293 using FuGENE 6 Transfection kit (Roche, Basel, Switzerland) and manufacturer's protocols. Stably expressing HEK-293 cells were selected using geneticin (G-418, 100 ml/10 ml; Sigma, Deisenhofen, Germany). Rat aortic endothelial cells with endogenous rK Ca 3.1 were provided by the BMFZ of the Philipps-University Marburg [39]. Murine 3T3 fibroblasts were obtained from ATCC (3T3-L1, ref# CL-173, ATCC, Rockville, MD). As usual cell culture medium, we used Dulbecco's Modified Eagle Medium (DMEM) supplemented with 10% calf serum and 1% penicillin/ streptomycin (all from Biochrom KG, Berlin, Germany). Before patch-clamp, cells were trypsinized and seeded on cover slips for 4-24 hrs.

Patch-clamp electrophysiology
Membrane currents in excised inside-out patches and whole-cell currents were recorded with an EPC-9 patch-clamp amplifier (HEKA, Lambrecht Pfalz, Germany) using borosilicate glass pipettes with a tip resistance of 2-3 MOhm. Seal resistance was above 1 GOhm. In inside-out experiments, we continuously monitored outward currents at a holding potential of 0 mV prior to patch excision and thereafter. Activation of K Ca 3.1-mediated currents occurred immediately after excision of the patch and exposure of the intracellular side of the patch to the Ca 2+containing bath solution (''intracellular'' solution see below). For conventional whole-cell current measurements, we used voltage ramps (voltage range for recording: 2120 mV to +100 mV; duration, 1 sec; applied every 3 sec; voltage range evaluated: 2 110 to +30 mV). Series resistance was between 7-15 MegaOhms and membrane resistance was .1 GigaOhm. In such experiments, the ''intracellular'' Ca 2+ -containing solution was ''infused'' into the cell via the patch-pipette after seal rupture activating K Ca -currents usually within 2-10 sec. Current amplitudes remained stable thereafter over 5 min and longer in some. The solution was composed of (mM):  [16]. In inside-out experiments, the high Na + solution served as pipette solution and the high K + solution as bath solution; in whole-cell experiments, vice versa. For measurements of rK Ca 3.1 currents in RAEC, we performed the experiments in the presence of the K Ca 2 blocker UCL-1684 (250 nM) [40] to eliminate rK Ca 2.3 currents in these cells.  (Germany). Uvaol ((3b)-Urs-12-ene-3,28-diol), erythrodiol ((3b)-Olean-12-ene-3,28-diol), oleanolic acid ((3b)-3-Hydroxyolean-12-en-28-oic acid), and maslinic acid ((2a,3b)-2,3-Dihydroxyolean-12-en-28-oic acid) were kind gifts from Dr. Jesú s Osada, Department of Biochemistry and Molecular and Cellular Biology, Veterinary School, Health Research Institute of Aragon, CIBEROBN, Zaragoza, Spain. EETs were delivered as ethanol stock solutions. Ethanol was evaporated under nitrogen stream and the EETs were reconstituted in DMSO at a concentration of 10 mM. Stocks were stored at 220uC until use. Stock solutions of the other fatty acids (10 mM) were also prepared with DMSO. Ahead of use stock solutions were diluted 1:10 with the bath buffer and the final DMSO concentration did not exceed 0.2%. Since unsaturated fatty acids are sensitive to oxidative degradation, we minimized exposure times in aqueous solutions and to air and prepared the aqueous pre-dilutions of the compounds immediately before starting the experiments. Bath solutions were not gassed with oxygen.

Statistics
Data are given as mean 6 SEM. For statistical comparison of multiple data sets we used one-way ANOVA and the Tukey post hoc and p-values of ,0.05 were considered significant.

Results
In inside-out experiments on HEK-293 expressing cloned hK Ca 3.1, excision of the patch into the 3 mM Ca 2+ -containing bath solution caused immediate activation of K + -outward currents that were stable over several minutes (Figure 2A). Non-transfected cells did not display these currents. In hK Ca 3.1-HEK-293, K +outward currents were virtually absent in the continuing presence of the classical K Ca 3.1-blocking toxin, charybdotoxin, in the ''extracellular'' pipette solution ( Figure 2A). Likewise, in the continuing presence of the selective small molecule blocker of K Ca 3.1, TRAM-34 [6], in the bath solution prevented K +outward currents, although we observed an initial spike-like outward current (Figure 2A) after excision of the patch.
In the continuing presence of 1 or 10 mM of the v3, docosahexaenoic acid (DHA), arachidonic acid (AA), and alinolenic acid (a-LA), hK Ca 3.1 currents could still be activated by patch-excision but the currents did not last and were inhibited after 30 sec ( Figure 2B and C). The inhibition by 1 mM was less pronounced than inhibition by 10 mM for all v3 tested here ( Figure 2C). However, potencies and kinetics of current inhibition differed between the v3 with the following order of potency and time to full inhibition: DHA$AA.a-LA ( Figure 2D). In contrast, the saturated fatty acid, arachidic acid (ArA), did not produce channel inhibition ( Figure 2B and C).
The single mutants, hK Ca 3.1 V275A and hK Ca 3.1 T250S , and the double mutant, hK Ca 3.1 V275A/T250S , were largely insensitive to AA and TRAM-34 (data shown for hK Ca 3.1 V275A ), although the hK Ca 3.1 T250S mutant appeared to have a smaller impact compared to the virtually complete insensitivity of the hK Ca 3.1 V275A mutant to AA ( Figure 4A and B). With respect to the other hK Ca 3.1-blocking fatty acids, hK Ca 3.1 V275A mutant was also insensitive to 11,12 EET, 14,15-EEZE, and 20-HETE as examples of fully (14,15-EEZE, 20-HETE) or partially (11,12-EET) hK Ca 3.1-blocking fatty acids ( Figure 4A and B).
We next tested the idea whether the 5,6-EET as a non-blocking EET antagonizes AA-mediated channel blockade. These experiments showed that in the presence of both fatty acids, 1 mM 5,6-EET did not significantly prevent channel inhibition by 10 mM AA although the time period to achieve channel inhibition appeared to be increased ( Figure 5A and B). At 1 mM AA we observed a significant antagonism of channel blockade by 1 mM 5,6-EET at a later time point (Figure 5A and B).
An increase of intracellular Ca 2+ stimulates Ca 2+ -dependent PLA 2 activity and AA-release. In our fast-whole cell experiments using a pipette solution with 0.3 mM Ca 2+ free , we expected Ca 2+dependent activation of hK Ca 3.1 and also Ca 2+ -dependent PLA-2mediated AA-release. In keeping with the idea that 5,6-EET antagonizes endogenous AA effects, we hypothesized that 5,6-EET augments total hK Ca 3.1-currents in the HEK-293 cells and tested this in a small series of fast-whole cell experiments ( Figure 6). We found that 5,6-EET (at 1 mM) produced significant potentiation by <twofold of the K Ca 3.1 current that was pre-activated by 0.3 mM intracellular Ca 2+ ( Figure 6A). A high concentration of AA (10 mM) abolished these 5,6-EET-potentiated currents. Whole-cell currents produced by the hK Ca 3.1 T250S/V275A mutant did not show potentiation by 5,6-EET ( Figure 6A, right panel).
We performed another series of whole-cell experiments on rat aortic endothelial cells (RAEC) as an established and physiologically relevant cell system involving Ca 2+ -dependent AA and CYP/EETs signaling as well as K Ca 3.1-dependent hyperpolarization as two mechanisms for endothelium-dependent vasodilation besides the nitric oxide pathway [29]. We tested specifically whether 1) AA and 14,15-EET produced a similar inhibition of endogenous rK Ca 3.1 channels in RAEC, 2) rK Ca 3.1 currents displayed a similar sensitivity to inhibition by AA, and 3) 5,6-EET produced potentiation of the current. As shown in figure 6B, these experiments revealed that 14,15-EET at 1 mM abolished calciumactivated rK Ca 3.1 currents in these RAEC, in this regard similar to the findings in hK Ca 3.1-overexpressing HEK-293. With respect to 5,6-EET-potentiation we found that 5,6-EET at 1 mM potentiated by <2.5-fold these endothelial calcium-activated rK Ca 3.1 currents being pre-activated by 0.5 mM and 3 mM intracellular Ca 2+ but not at 0.1 mM, a Ca 2+ -concentration that did not allow channel pre-activation ( Figure 6B). AA at a concentration of 10 mM substantially blocked this 5,6-EET-potentiated current. Similar to the inside-out experiments, we did not see appreciable antagonistic effects at this lower concentration (1 mM) of 5,6-EET in these whole-cell experiments.
The v3, DHA, and pentacyclic triterpenes as e.g. uvaol have been demonstrated experimentally to protect against cardiac fibrosis [35,36], in addition to their documented vaso-protective and anti-inflammatory actions [37,38]. Recently, we reported membrane expression of K Ca 3.1 channels in proliferating murine 3T3-fibroblasts [16]. In the present study, we performed a series of whole-cell experiments and tested whether DHA and pentacyclic triterpenes inhibited mK Ca 3.1 in murine fibroblasts. We found  that DHA at 1-10 mM abolished virtually mK Ca 3.1 ( Figure 6C). In contrast, the pentacyclic triterpenes, uvaol, erythrodiol, maslinic acid, and oelanic acid, did not modulate mK Ca 3.1currents at 1 mM ( Figure 6D).

Discussion
Here we studied modulation of K Ca 3.1 channel by CYPproducts, 5,6-EET, 8,9-EET, 11,12-EET, and 14,15-EET, the vhydrolase product, 20-HETE, and the v3, DHA, and a-LA, and identified structural requirements of these fatty acids for K Ca 3.1modulation. Our major findings were that 14,15-EET and 20-HETE as well as DHA and a-LA produced K Ca 3.1 inhibition with potencies in the lower mmolar range. 11,12-EET was less potent and 5,6-EET and 8,9-EET did not cause inhibition. However, 5,6-EET was able to antagonize AA-induced inhibition. The observation that 14,15-EET and 20-HETE were efficient inhibitors while 5,6 and 8,9-EET not, identified the hydrophobic carbon stretch from C1-10 of the carboxyl head of the molecule as structural requirement for channel inhibition (for schematic overview of structural features of K Ca 3.1-blocking and nonblocking fatty acids see Figure 1).
Several down-stream targets and receptors for propagation of intracellular or paracrine actions of EETs and v3 have been proposed and, particularly, ion channel modulation by these fatty acid emerged as an additional mechanistic step. Yet, a plethora of channels have been shown to be directly activated by EET or to be a downstream target of EETs [41,42,43,44,45,46,47]. For instance, the TRPV4 channel, a member of the transient receptor potential gene family of cation channels, have been proposed to be activated by 5,6-EET and 8,9-EET and the resulting Ca 2+ -influx into the vascular endothelium caused vasorelaxation [43,44]. TRPA1 channels in afferent neurons were activated by 5,6-EET leading to an increase in nociception in mice [48]. Yet, another TRP channel, the TRPC6 channel, has been shown to be translocated in a PKA-dependent manner to the cell membrane that required 11,12-EET binding to Gs-receptors in endothelial cells [49]. Moreover, 11,12-EET has been proposed to induce hypoxic vasoconstriction in the lung involving TRPC6 mechanism [50]. Other studies showed that 14,15-EET mediates phosphorylation of epithelial sodium channel (ENAC) activity in an ERK1/ 2 dependent mechanism [51].
With respect to K + channels, 8,9-EET, 11,12-EET, and 14,15-EET have been reported to activate ATP-sensitive K + -channels by allosteric interaction with the ATP-binding site of the channel [52]. Two-pore tandem K + channels (K2P) and large-conductance K Ca 1.1 channels were known since long to be activated by v3 and v6 [45,46,53,54,55]. Moreover, 11,12-EET activation of K Ca 1.1 channels was considered a main mechanism in smooth muscle, by which EET produced vasorelaxation [56]. In contrast, 20-HETE has been shown recently to enhance angiotensin-II-induced vasoconstriction by inactivating K Ca 1.1 channels [57]. Interestingly, AA has also been shown to inhibit voltage-gated K + channels such as the T lymphocyte KV1.3 channel [58] and the endogenous KV channels in HEK-293 (unpublished observation by our group). To our knowledge there were no data on direct or downstream modulation of K Ca 3.1 channels by EETs that were not simply linked to EET-mediated increase in intracellular Ca 2+ . Hence, it was well established that K Ca 3.1 channels were inhibited by the v6, AA, that required mechanistically interaction with the lipophilic residues, V275 and T250, lining the channel cavity [25]. The structural requirements of the AA molecule to produce this inhibition remained however unclear. Our present study confirmed AA-mediated inhibition and the requirements of residue V275 and to some extent of T250 ( Figure 4). Moreover, we provided additional insight by showing that the v3, DHA and a-LA produced similar inhibition of the cloned human channel. Moreover, we showed here that AA abolished endogenous rK Ca 3.1 ( Figure 6) that suggested that AA could be an endogenous negative regulator of K Ca 3.1 in the endothelium and could thereby influence the K Ca 3.1-dependent endothelium-derived hyperpolarization (EDH)-mediated type of arterial vasodilation [28,59]. However, this has not been further clarified by the present study. Interestingly, our inside-out experiments showed that K Ca 3.1 could still be activated in the continuous presence of the AA but inactivated rapidly following Ca 2+ -dependent activation (Figure 2). This suggested a major impact of AA on K Ca 3.1-gating unlike charybdotoxin (Figure 2) that obstructs simply the pore and ion flow by binding to the outer vestibule of the channel, independently of gating. However, we cannot exclude that this transient activation seen in the presence of AA reflected a delay of inhibition caused by diffusion of AA and the other compounds from the bath solution towards the excised membrane patch in the patch pipette.
With respect to eicosanoid-modulation of K Ca 3.1, our study demonstrated that 14,15-EET, the stable analogue, 14,15-EEZE, and 20-HETE were K Ca 3.1-inhibitors with potencies slightly below that of AA. Structurally, this inhibition required apparently hydrophobicity and 2 double electron bonds within the first 10 carbons of the carboxyl head of the molecules. This was concluded from the lack of inhibitory activity of 5,6-EET and 8,9-EET, in which this part of the fatty acid chain was epoxygenated. The partial inhibition caused by 11,12-EET could be explained by the conserved hydrophobicity within carbons 1-10 although 11,12epoxygenation appeared to have efficacy-reducing impact. In respect to channel-eicosanoids interactions, it was likely that epoxygenation as in 5,6,-EET and 8,9-EET did not allow the proper interactions of these molecules with hydrophobic residues of the cavity below the selectivity filter as they have been postulated for AA [25]. The intactness of carboxyl head of the molecule was another structural need since major alterations as in anandamide and 2-arachidonoylglycerol let to a loss of inhibitory Note that at a low intracellular Ca 2+ (0.1 mM) that is below/near the threshold for K Ca 3.1 activation, 5,6-EET did not potentiate the current. In contrast, potentiation occurred at an intracellular Ca 2+ concentration that is near the EC 50 for Ca 2+ -activation of K Ca 3.1 as well as at a saturating Ca 2+ concentration. C) DHA (1 mM) blocked Ca 2+ -pre-activated mK Ca 3.1 in murine fibroblasts. D) Pentacyclic triterpenes did not modulate murine fibroblast mK Ca 3.1 at a concentration of 1 mM. Data are means 6 SEM (% inhibition of K Ca 3.1-current normalized to initial peak amplitude after establishing electrical access (by seal rupture) and stable Ca 2+ -activation of K Ca 3.1-outward currents); Numbers in the graphs indicate the number of whole-cell experiments; *P,0.05 vs. control (peak amplitude of the K Ca 3.1-current in the respective cell); One-way ANOVA and Tukey post hoc test. doi:10.1371/journal.pone.0112081.g006 efficacy (see figure 1 for structures and scheme of blocking efficacy of the fatty acids). However, detailed structural analysis on yet not available crystal structures of the open and closed K Ca 3.1 channel and mapping of AA and EETs interaction/binding will be needed to provide more definite insight into this lipid modulation of K Ca 3.1 channels. In contrast to eicosanoids and v3, the pentacyclic triterpenes studied here did not modulate mK Ca 3.1 channel, which might be explained by their more ''rigid'' and larger structures that may not fit into the internal cavity of the channel.
From the physiological and pharmacological perspective, mircomolar EETs, stable EET-analogues, and 20-HETE have been used to study mechanisms of vasodilation or vasoconstriction. Since K Ca 3.1 has been demonstrated a major component in the EDH-mediated type of endothelium-dependent vasodilation [59] and considering that this channel modulates also functions in several other tissues [1,2], interactions of the different EET and of 20-HETE with K Ca 3.1 channels as described in the present study needs to be taken into account.
An additional interesting observation was that 5,6-EET was capable to antagonize AA-inhibition of K Ca 3.1-activity in isolated patches ( Figure 4). Moreover, at the whole-cell level, 5,6-EET potentiated Ca 2+ -pre-activated K Ca 3.1-currents. While 5,6-EET did not have a direct effect on channel-gating per se as concluded from the inside-out experiments (Figure 3), it was tempting to speculate that 5,6-EET antagonized the -at least partial -channel inhibition caused by endogenous Ca 2+ -dependent PLA 2 -mediated AA-release. This view was fostered by the insensitivity of the hK Ca 3.1 V275A to 5,6-EET-potentiation ( Figure 6). Such a mechanism may represent a novel mechanism of endogenous K Ca 3.1modulation beyond Ca 2+ -regulation of the channel. Moreover, the 5,6-EET-mediated de-blockade of K Ca 3.1 could be a thus far unrecognized mechanism underlying EDH-mediated vasodilation, in which both EETs and K Ca 3.1 have been implicated to play major roles.
In conclusion, the present electrophysiological and structureactivity-relationship study demonstrated modulation of cloned and endogenous K Ca 3.1 channels by selective EETs, 20-HETE, and v3 and revealed major structural determinants of the molecules for channel interaction.