Molecular Determinants of Cardiac KATP Channel Activation by Epoxyeicosatrienoic Acids*

We have previously reported that epoxyeicosatrienoic acids (EETs), the cytochrome P450 epoxygenase metabolites of arachidonic acid, are potent stereospecific activators of the cardiac KATP channel. The epoxide group in EET is critical for reducing channel sensitivity to ATP, thereby activating the channel. This study is to identify the molecular sites on the KATP channels for EET-mediated activation. We investigated the effects of EETs on Kir6.2ΔC26 with or without the coexpression of SUR2A and on Kir6.2 mutants of positively charged residues known to affect channel activity coexpressed with SUR2A in HEK293 cells. The ATP IC50 values were significantly increased in Kir6.2 R27A, R50A, K185A, and R201A but not in R16A, K47A, R54A, K67A, R192A, R195A, K207A, K222A, and R314A mutants. Similar to native cardiac KATP channel, 5 μm 11,12-EET increased the ATP IC50 by 9.6-fold in Kir6.2/SUR2A wild type and 8.4-fold in Kir6.2ΔC26. 8,9- and 14,15-EET regioisomers activated the Kir6.2 channel as potently as 11,12-EET. 8,9- and 11,12-EET failed to change the ATP sensitivity of Kir6.2 K185A, R195A, and R201A, whereas their effects were intact in the other mutants. 14,15-EET had a similar effect with K185A and R201A mutants, but instead of R195A, it failed to activate Kir6.2R192A. These results indicate that activation of Kir6.2 by EETs does not require the SUR2A subunit, and the region in the Kir6.2 C terminus from Lys-185 to Arg-201 plays a critical role in EET-mediated Kir6.2 channel activation. Based on computer modeling of the Kir6.2 structure, we infer that the EET-Kir6.2 interaction may allosterically change the ATP binding site on Kir6.2, reducing the channel sensitivity to ATP.

The cardiac ATP-sensitive K ϩ (K ATP ) 1 channels are biological sensors that respond to intracellular metabolic changes and play important roles in regulating cardiac functions (1). Although the role of the cardiac K ATP channels under normal physiological conditions remains unclear, it is crucial in ischemic preconditioning, and activation of K ATP channels occurs during cardiac ischemia and hypoxia, leading to reduced Ca 2ϩ influx and intracellular Ca 2ϩ overload (2). Some important insights were provided by studies using the Kir6.2 knock-out mice, which have compromised ability in modulating cardiac electrophysiological properties, contractility (3,4), and in handling cardiac ischemia and stress (3,5,6). Recently, a form of human dilated cardiomyopathy was found to be associated with mutations of the cardiac K ATP channels (7).
The cardiac K ATP channel is a heterooctamer containing four inward rectifier K ϩ channel (Kir6.2) and four sulfonylurea receptor (SUR2A) subunits. SUR contains two cytoplasmic nucleotide binding folds that were initially thought to be important for channel regulation by ATP (8). However, a truncation mutant of the Kir6.2 involving deletion of the C-terminal 26 residues (Kir6.2⌬C26) gives rise to channels that retain much of the ATP sensitivity in the absence of SUR (9). Also, mutations in the SUR produced only small effects on ATP inhibition of Kir6.2/SUR currents (10). Previous studies have demonstrated that both the N and C termini of Kir6.2 contribute to the site(s) that regulates ATP sensitivity, and they include Arg-50, Cys-166, Ile-167, Thr-171, Arg-176, Arg-177, Glu-179, Ile-182, Lys-185, Arg-192, Arg-201, and Gly-344 residues. Of these, Arg-50, Lys-185, and Arg-201 residues are particularly crucial for ATP sensitivity and are implicated for interaction with the ␣, ␤, and ␥ phosphates of ATP (11)(12)(13). However, the mechanism of Kir6.2 channel inhibition by ATP remains to be elucidated.
EETs are known to regulate vital physiological functions including vasoreactivity (23,24), inflammation (25), and cell proliferation (26,27). EETs have been considered to be candidates of endothelium-derived hyperpolarizing factors in human coronary vessel bed (23,28). EETs are potent modulators of the cardiac Na ϩ channel (29), the L-type Ca 2ϩ channels (30), and the coronary arterial smooth muscle Ca 2ϩ -activated K ϩ channels (31,32). We have previously reported that EETs are potent activators of the cardiac K ATP channels, reducing the channel sensitivity to ATP, and had an EC 50 of about 30 nM (17,18). Single channel kinetic studies showed that EETs reverse the effect of ATP on the K ATP channel properties, prolong the channel open time, and shorten the channel closed times (17). Structural determinant studies showed that only the 11(S),12(R)-EET, but not the 11(R),12(S)-EET enantiomer, could activate the cardiac K ATP channels (18). In addition, the carboxylic group, the carbon chain length, and the number of double bonds on the EET molecule all were not important. Only the presence of the epoxide group is critical for K ATP channel activation (18). These results suggest that there is a specific and direct interaction between EETs and cardiac K ATP channels. In this study we examined the role of SUR2A and the positively charged residues on the Kir6.2 subunit that modulate channel ATP sensitivity and open probability (P o ) on channel sensitivity to EETs.

Kir6.2 Mutagenesis and Expression-Mutations and deletions of mouse
Kir6.2 (GI:1622948) were performed using the QuikChange site-directed mutagenesis kit (Stratagene, La Jolla, CA). Oligonucleotide primers containing the desired mutations were synthesized by IDATE Co. (Coralville, IA). The orientation of the constructs and the correct mutations were verified by DNA sequencing (DNA facilities core of Mayo Clinic). Kir6.2 and SUR2A in pcDNA3.1 were kindly provided by Dr. Andre Terzic. Kir6.2, SUR2A, and green fluorescent protein inserted in pcDNA3.1 were transfected into HEK293 cells in ratios of 1:4:1 (w/w/w) using a FuGENE 6 transfection kit (Roche Applied Science). The transfected cells were detected by the presence of green fluorescent protein expression under an ultraviolet microscope (Olympus, IX70) 48 h after transfection. HEK293 cells were cultured in Dulbecco's modified Eagle's medium with 5 mM glucose and 10% (v/v) fetal calf serum, penicillin (100 units/ml), and streptomycin (100 units/ml).
Electrophysiology-Inside-out single channel recordings were recorded with an Axopatch 200B amplifier (Axon Instruments, Forest City, CA), and output of the amplifier was filtered through an 8-pole Bessel filter at 5 kHz and digitized at 40 kHz as described previously (17).
Macroscopic currents of Kir6.2 channels were elicited from excised inside-out patches using a continuous voltage ramp protocol (ϩ100 mV to Ϫ100 mV) over 100 ms at 30-s intervals with a holding potential of 0 mV. The pipette resistance was 0. The effects of EETs on the ATP sensitivity of the channel were determined by measuring the percentage changes before and after application of 5 M 11,12-EET in the presence of ATP concentrations at IC 50 . To obtain the ATP IC 50 , the relationship between ATP and channel maximal conductance (G) or P o at various ATP concentrations was plotted and fitted using a Hill equation of the form, G ϭ G max /[1 ϩ ([ATP]/IC 50 ) H ], where G max represents channel maximal conductance at zero ATP concentration, [ATP] represents the concentration of ATP, IC 50 is the concentration at half-maximal inhibition, and H is the Hill coefficient. All experiments were conducted at room temperature (23°C).
Computer Modeling of Kir6.2 Structure-Using the crystal structure of a related K ATP -sensitive channel, KirBac1.1 (33), as the template to locate the residues of interest in this study, we generated a putative three-dimensional structural model of Kir6.2. The KirBac1.1 coordinates were downloaded from the Protein Data Bank (file 1P7B). The amino acid sequences of mouse Kir6.2-and KirBac1.1-crystallized protein were aligned with 50% identity according to the conserved domain data base of the National Center for Biotechnology Information (www. ncbi.nlm.nih.gov). The KirBac1.1-crystallized protein has a more complete N terminus than that of GIRK1 (G-protein-activated inward rectifier potassium channel 1). The predicted Kir6.2 structure was created by PyMol software, and the estimated distance between two residues was measured from the model.
Statistical Analysis-Data are present as the mean Ϯ S.E. Student's t test was used to compare data between two groups. A one-way analysis of variance followed by a Tukey test analysis was employed to compare data from multiple groups using SigmaStat software (Jandel, San Rafael, CA).

RESULTS
We first determined whether the heterologously expressed Kir6.2/SUR2A wild type (Kir6.2 wt) channels are activated by EETs. We measured the ATP sensitivity of Kir6.2 wt in the presence and absence of 5 M 11,12-EET to obtain maximal channel activation. Fig. 2A shows representative recordings of Kir6.2 wt (top panel). ATP at IC 50 concentration (50 M) produced 50% inhibition of the currents, and this effect was reproducible after washout and re-exposure to ATP. However, in the presence of 5 M 11,12-EET, application of ATP at IC 50 concentrations failed to produce the same amount of inhibition in the Kir6.2 wt channels. The ATP inhibitory effect was significantly antagonized by 11,12-EET as observed in native K ATP channels in cardiac ventricular myocytes (17). Single channel recordings confirmed that the effects of ATP concentrations on channel P o were reduced in the presence of 5 M 11,12-EET (  (Fig. 2C). The Hill coefficient (H) remained unchanged (1.23 Ϯ 0.08 at base line versus 1.28 Ϯ 0.10 with EET; n ϭ 6, p ϭ not significant). These results suggest that the Kir6.2 wt channels expressed in HEK293 cells retain the sensitivity to activation by EET, similar to native K ATP channels in cardiac ventricular myocytes.
Kir6.2⌬C26 forms functional channels when expressed in HEK293 cells without the co-expression of SUR2A as previously reported (Fig. 3, A and B)  changing H (Fig. 3C) (Fig. 4B), consistent with previous reports (9,34). These results suggest that the SUR2A subunit modulates the Kir6.2 channel property and ATP sensitivity, but activation of the K ATP channel by EETs does not require the presence of SUR2A. Also, the last 26 amino acids of the Kir6.2 C terminus were not required for sensitivity to EET. Because the SUR2A subunit is not involved with Kir6.2 chan-nel sensitivity to EET, all further studies with Kir6.2 mutant channels were performed with the coexpression of SUR2A.
Because EETs activate the K ATP channels by reducing channel sensitivity to ATP and may share a common mechanism with other lipid metabolites in K ATP channel activation, we tried to identify the molecular sites on the Kir6.2 that are important for EET sensitivity by neutralizing the positively charged residues before Gly-334 and to determine the effects of these mutations on EET sensitivity. We found that the Kir6.2 R27A, R50A, K185A, and R201A mutants significantly reduced the channel ATP sensitivity, consistent with previous reports (11,12,(35)(36)(37)(38).  (Fig. 4B). The R27A mutant produced a small but significant increase in ATP IC 50 (73.4 Ϯ 16.4 M, n ϭ 8, p Ͻ 0.05, versus wt) (see Fig. 6, upper panel). In addition, the G334D mutant was practically ATP-insensitive, with IC 50 Ͼ20 mM, and no currents could be recorded from the R176A, R177A, R206A, and R301A mutants in excised inside-out patches, as previously reported (35,39).
To examine the effects of EETs on ATP inhibition of the Kir6.2 mutants, we determined the effects of 5 M 11,12-EET in the presence of IC 50 concentrations of ATP for each mutant channel. We found that all the mutants of neutralized positively charged residues, including R27A and R50A, on the cytoplasmic N terminus of Kir6.2 were activated by 11,12-EETs, with significant reduction to ATP inhibition ( Fig. 5 and Fig. 6, lower panel). These findings suggest that EET and ATP probably act at different sites on Kir6.2. On the C-terminal side of Kir6.2, sensitivity to 11,12-EET was preserved in R192A, K207A, K222A, and R314A. However, the effects of 11,12-EET were abolished in the K185A, R195A, and R201A mutants, and channel inhibition by ATP was not altered by 11,12-EET (Fig.  5). Summary of ATP sensitivity and 11,12-EET effects on Kir6.2 wt and neutralized positively charged Kir6.2 mutants is shown in Fig. 6.

DISCUSSION
In this study we reported several novel observations on the molecular determinants for activation of the K ATP channels by EETs. First, we found that the SUR2A subunit is not required for EET to activate the Kir6.2 subunit. Second, the N-terminal positively charged residues of Kir6.2 are not important for determining EET sensitivity of the channel. Third, three Cterminal residues Lys-185, Arg-192/Arg-195, and Arg-201 are critical molecular sites for EET sensitivity. These results place the EET sensitivity over a region encompassing 16 amino acid residues on the C terminus of Kir6.2, and this may possibly represent the EET binding pocket. These results also indicate that the molecular determinants for EET are different from ATP and PIP 2 .
The mechanisms of ATP inhibition of the cardiac K ATP channels are complex and involve both Kir6.2 and SUR2A subunits. Our results showed that the presence of SUR2A alters the channel IC 50 and H to ATP (Figs. 2 and 3). This is consistent with recent reports that SUR modifies the ATP binding pocket of Kir6.2 by increasing the width of the groove that binds the phosphate tail of ATP without changing the length of the groove and by enhancing the interaction with the adenine ring (40). Previous studies identified Lys-185 and Arg-201 on the C terminus and Arg-50 on the N terminus of Kir6.2 as particularly crucial for ATP binding (12,13,(35)(36)(37). Two processes have been proposed to describe the mechanism of K ATP inhibition by ATP. First, ATP may bind to a channel closed state and stabilizes it, resulting in a decrease in P o . Second, ATP may bind to the channel open state and destabilizes the open channel, reducing P o and shortening the mean open time constant (12). Mutagenesis studies suggested that Lys-185 interacts with the ␤-phosphate and Arg-201 interacts with the ␣ phosphate of ATP; the former interaction destabilizes the channel open state, whereas the latter would favor the channel entrance into stable closed states (12,13). The role of Arg-50 is less clear, and interaction with the ␤ phosphate (12) and the ␥ phosphate (11) of ATP has been suggested. How many ATP binding pockets exist in each Kir6.2 subunit is not known. Studies using maltose-binding fusion proteins of the cytoplasmic regions of Kir6.2 showed that ATP only binds to the cytoplasmic C terminus but not to the N terminus of the channel (41). Our study confirmed that Arg-50, Lys-185, and Arg-201 are critical sites for ATP sensitivity, and Arg-27 also slightly but significantly affects ATP IC 50 . Recently, It has been reported that Arg-50 may approach the vicinity of Lys-185 and Arg-201 of the neighboring Kir6.2 subunit to form an ATP binding site by intramolecular complementation (42).
This study identified Lys-185, Arg-201, and Arg-195 residues as critical sites for determining the channel sensitivity to 8,9and 11,12-EET, whereas Arg-192 rather than Arg-195 is critical for the Kir6.2 channel sensitivity to 14,15-EET. This may be due to the position of the epoxide group and its three-dimensional orientation. Nevertheless, regioisomer-specific interaction with Arg-192 and Arg-195 produced the same effect and potency in Kir6.2 channel activation, indicating that Arg-192/ Arg-195 may represent the epoxide binding site. However, the exact mechanism whereby EETs antagonize the ATP effects is not known. Interestingly, the R27A and R50A mutations altered ATP sensitivity without affecting channel sensitivity to 11,12-EET. In contrast, the R192A or R195A mutation did not change the response to ATP, but channel activation by 11,12-EET was lost. Only the K185A and R201A mutations diminished both the ATP and EET effects on the channel. These results suggest that the EET binding site on Kir6.2 is different from that of ATP.
A direct competition for binding between ATP and PIP 2 at the C terminus of Kir6.2 has been reported (43). Previous studies showed that the basic residues in two regions (176 -222 and 301-314) in the C terminus of Kir6.2 are important for determining the PIP 2 effects (35). Arg-176, Arg-177, Arg-206, Arg-222, and Arg-301 residues in particular are sites critical for PIP 2 sensitivity. Of the N-terminal sites, Arg-54 was identified as a major determinant for PIP 2 modulation of ATP sensitivity in the K ATP channels (44) Interestingly, with the possible exception of Arg-201, none of the PIP 2 -sensitive residues overlaps with those of ATP or EET. Recently, long chain acyl-CoA esters were found to modulate ATP inhibition of the K ATP channels by the same mechanisms as PIP 2 (14), suggesting these lipid metabolites shared a common site of interaction on the channel. Indeed, it has been proposed that there is a novel lipid-interacting motif for inward rectifier K ϩ channels termed KIRLI domain (45). The core of this domain is composed of residues 170 -320, and the structure is composed of antiparallel ␤ strands and an ␣ helix, which closely resembles that of the pleckstrin homology (PH) domain, a known PIP 2 binding motif. According to this model, charged residues including Arg-176, Arg-177, Arg-192, and Arg-195 are clustered on one side of the ␤-strand pocket. Arg-176 and Arg-177 are involved with PIP 2 interaction (45), whereas Arg-192 and Arg-195 are important for EET sensitivity. Whether EETs and PIP 2 interact with the K ATP channels through a similar mechanism that affects ATP sensitivity is at present unknown. However, we would like to emphasize that there are major differences between EET and PIP 2 on K ATP channel activation. First, the EET effects are stereo-specific with an EC 50 in the range of 10 Ϫ8 M and reach steady state within 2 min. Second, the negative charge on the EET molecule is not important. The EET methyl ester is as active as EET in reducing ATP sensitivity of the channel (18). Third, unlike PIP 2 , EETs could not activate the K ATP channels after rundown or from glibenclamide inhibition. Fourth, with the exception of Arg-201, residues on the Kir6.2 channel that are important for PIP 2 sensitivity, such as Arg-54 and Lys-222, are not important for EET sensitivity. Hence, we believe that EETs and fatty acid epoxides constitute a unique class of lipid metabolites that function as K ATP channel activators.
To better understand the role of the EET-sensitive sites on Kir6.2, we constructed a molecular model of Kir6.2 based on the crystal structure of KirBac1.1 (33) (Fig. 8). This ATPsensitive channel shares 50% sequence homology with Kir6.2, and it contains a more complete N terminus than that of Kir3.1, which has been used for creating putative Kir6.2 models (46). -EET, the current inhibition by ATP as IC 50 concentration was reduced, similar to that obtained from 11,12-EET. In contrast, 5 M 14,15-EET had no effect on ATP inhibition. B, recordings of inside-out macroscopic currents of Kir6.2 R195A. 5 M 14,15-EET was able to activate the channel in the presence of IC 50 concentration of ATP, whereas 5 M 8,9-EET had no effect. C, group data in bar graphs showing the effects of 5 M 8,9-, 11,12-, and 14,15-EET on Kir6.2 wt, R185A, R192A, R195A, and R201A mutant channels. The percentage changes in current inhibition by ATP at IC 50 concentrations before and after the application of the EET regioisomers were normalized to that in Kir6.2 wt. Each bar represents the mean Ϯ S.E., and the sample size is indicated above each bar. *, p Ͻ0.05 versus Kir6.2 wt. represent the ATP binding pocket. Based on the three-dimensional structural model created, the estimated distance between Gly-334 to Lys-185 and to the Arg-201 side chain is 8.3 and 12.5 Å, where the distance from Lys-185 to Arg-201 is 6.9 Å, similar to a previous report (13) (Fig. 8B). It has been suggested that Arg-201 interacts with the ␣-phosphate of ATP, and Lys-185 interacts with the ␤-phosphate of ATP. Gly-334 has been proposed to interact with the adenine ring of ATP (13). However, the details of the molecular structure of the Kir6.2 N terminus, in particular the ATP-sensitive sites including Arg-27, has not been completely resolved. The residues from Lys-185 to Arg-201 consist of an anti-parallel ␤-sheet with Arg-192 and Arg-195 forming one end of this structure (Fig.  8B). The estimated distance between Arg-192 to Arg-195 is 5.7 Å, and this distance may accommodate the spatial constraints between the epoxide groups in 11,12-EET and 14,15-EET. Furthermore, the estimated distances from Arg-195 to Lys-185, Arg-201, and Gly-334 are 25.2, 22.0, and 25.4 Å, respectively, whereas those from Arg-192 to these residues are 22.3, 15.7, and 18.6 Å, respectively, and are shorter than those from Arg-195. Because the distance from ␣-phosphate to ␥-phosphate is about 8 Å, Arg-192 and Arg-195 are too far to be involved with ATP binding (40). One working model for the EET effects is that Arg-192 or Arg-195 constitute the EET binding site and upon physical interaction with EETs would result in allosteric changes in the ATP binding site. Hence, Lys-185 and Arg-201 represent sites downstream of EET binding (Fig. 8B), and mutations in these sites produce conformational changes that would preemptively nullify the EET effects. The locations of Arg-192 and Arg-195 are also not in the vicinity of Arg-176 and Arg-177, which are important PIP 2 binding sites, and this may help to explain why EETs and PIP 2 , even though they are both lipid metabolites that activate the cardiac K ATP channels, behave so differently.
In summary, we have provided mechanistic insights on Kir6.2 channel regulation by EETs. Three positively charged residues, Lys-185, Arg-192/Arg-195, and Arg-201, on the C terminus of Kir6.2 are crucial for EET sensitivity. Because Arg-192 and Arg-195 are not ATP-sensitive residues, they may interact with EETs, resulting in allosteric changes in the threedimensional structure of the ATP binding pockets and reducing the channel sensitivity to ATP. Direct binding studies will be helpful to further confirm this finding.