Molecular Mechanisms of the Inhibitory Effects of Propofol and Thiamylal on Sarcolemmal Adenosine Triphosphate–sensitive Potassium Channels

The human Kir6.2, rat Kir6.1, rat SUR1, rat SUR2A, and rat SUR2B complementary DNAs (cDNAs) were kindly provided by Susumu Seino, M.D., Ph.D. (Professor and Chairman, Department of Cellular and Molecular Medicine, Chiba University, Chiba, Japan). A truncated form of human Kir6.2 lacking the last 36 amino acids at the C-terminus was obtained by polymerase chain reaction amplification. Polymerase chain reaction products were cloned into the pCR3.1 vector by using the TA cloning system (Invitrogen Corp., Carlsbad, CA) and then cloned into the pcDNA3.1 (−) vector (Invitrogen Corp.) for mammalian expression. Site-directed mutagenesis was performed by using the Site-Directed Mutagenesis system (Invitrogen Corp.). All Kir6.2ΔC36 DNA products were sequenced by using the BigDye terminator cycle sequencing kit and an ABI PRISM 377 DNA sequencer (Applied Biosystems, Foster City, CA) to confirm the sequence.

COS-7 cells were plated at a density of 3 × 10⁵/dish (35 mm in diameter) and cultured in Dulbecco’s modified Eagle’s medium supplemented with 10% fetal calf serum. A full-length Kir cDNA and a full-length SUR cDNA were subcloned into the mammalian expression vector pCMV6c. For electrophysiologic recordings, either wild-type or mutated pCMV6c Kir alone (1 μg), or pCMV6c Kir (1 μg) plus pCMV6c SUR (1 μg) were transfected into COS-7 cells with green fluorescent protein cDNA as a reporter gene by using lipofectamine and Opti-MEN 1 reagents (Life Technologies Inc., Rockville, MD) according to the manufacturer’s instructions. After transfection, cells were cultured for 48–72 h before being subjected to electrophysiologic recordings.

Glibenclamide, diazoxide, or pinacidil was diluted in superfusate and directly applied to cultured cells in the glass-bottom plastic cell bath (2-ml volume) at a rate of 2–2.5 ml/min using a plastic syringe (50-ml volume), vinyl chloride tubing (0.8-mm ID, 50-cm length), and a syringe pump (Terumo STC-525, Tokyo, Japan). When the dose-dependent effects of propofol or thiamylal were studied, the superfusion was stopped for approximately 1 min at each concentration, and these drugs were injected into the cell bath using a glass syringe to five final concentrations in a cumulative manner (total volume injected was approximately 20 μl). Therefore, the superfusion was stopped for approximately 5 min; preliminary studies showed that the stopping of superfusion for approximately 5 min had no significant effects on electrophysiologic measurements. The average percent recovery of K^ATPchannel activities after washout of propofol or thiamylal was 95 ± 7% of the NP⁰measured before drug treatment.

Membrane currents were recorded in the inside-out configurations using a patch clamp amplifier as described previously. 24,25Transfected cells were identified by their green fluorescence under a microscope. The intracellular solution contained 140 mm KCl, 2 mm EGTA, 2 mm MgCl², and 10 mm HEPES (pH = 7.3). The pipette solution contained 140 mm KCl, 1 mm CaCl², 1 mm MgCl², and 10 mm HEPES (pH = 7.4). Recordings were made at 36°± 0.5°C. Patch pipettes were pulled with an electrode puller (PP-830; Narishige, Tokyo, Japan) and coated with Sylgard (Dow Corning, Midland, MI). The resistance of pipettes filled with internal solution and immersed in the Tyrode’s solution was 3–4 MΩ. The sampling frequency of the single-channel data was 5 KHz with a low-pass filter (1 KHz).

Channel currents were recorded with a patch clamp amplifier (CEZ 2200; Nihon Kohden, Tokyo, Japan) and stored in a personal computer (Aptiva; International Business Machine Corporation, Armonk, NY) with an analog-to-digital converter (DigiData 1200; Axon Instruments, Foster, CA). pClamp version 7 software (Axon Instruments) was used for data acquisition and analysis. The open probability (P^o) was determined from current amplitude histograms and was calculated as follows:

where t^jis the time spent at current levels corresponding to j = 0, 1, 2, N channels in the open state; T^dis the duration of the recording; and N is the number of the channels active in the patch. Recordings of 2–3 min were analyzed to determine P^o. The channel activity was expressed as NP^o. Changes of channel activity in the presence of drugs were calculated as the relative channel activity between the values obtained before and after drug treatment.

When the inside-out patches in the ATP-free bath solution are formed, recombinant K^ATPchannel activity gradually decreases with time. This phenomenon is known as run-down . Consequently, data obtained from such experiments with inside-out patches may not accurately represent the relation between a drug and K^ATPchannel activity. To minimize this time-dependent decrease of the channel activity, we determined the effect of a single concentration of propofol and thiamylal from each inside-out patch within 3 min of patch excision. Under these conditions, the average percent recovery of K^ATPchannel activities after drug washout was 96 ± 8% of the NP^omeasured before drug treatment. The drug concentration needed to induce half-maximal inhibition of the channels (EC⁵⁰) and the Hill coefficient were calculated as follows:

where y is the relative NP^o, [D] is the concentration of drug, Kⁱis the EC⁵⁰, and H is the Hill coefficient. To analyze of channel kinetics, unitary events were detected using a 50% threshold level method.

The following drugs were used: propofol (2,6-diisopropylphenol; Aldrich Chemical Co., Milwaukee, WI), thiamylal sodium (Yoshitomi Chemical, St. Louis, MO), glibenclamide, diazoxide, and pinacidil (Sigma-Aldrich Japan, Tokyo, Japan). Propofol, glibenclamide, diazoxide, and pinacidil were dissolved in dimethyl sulfoxide (the final concentration of solvent was 0.01%), which at a twofold higher concentration did not affect K^ATPchannel currents. The thiamylal sodium ampule was opened just before use.

All data are presented as mean ± SD. Differences between data sets were evaluated either by repeated-measure one-way analysis of variance followed by Scheffé F test or by Student t  test. P < 0.05 was considered significant.

Four types of reconstituted recombinant K^ATPchannels were transiently expressed in COS-7 cells, inside-out patches were excised, and the K^ATPchannel currents were recorded. Figure 1ashows representative examples of these currents. We did not observe channel openings in the cell-attached configuration in any case. However, when the patch was excised into a nucleotide-free solution, the K^ATPchannels composed of Kir6.2 in combination with SUR1, SUR2A, or SUR2B showed marked current increases. These currents were blocked by 1 mm ATP, which shows that COS-7 cells cotransfected with wild-type Kir6.2 and SUR express functional ATP-sensitive K⁺channels. The SUR2B/Kir6.1 channel was not spontaneously activated by patch excision in the absence of intracellular ATP. Diazoxide (300 μm), a potent opener of K^ATPchannels, activated the SUR1/Kir6.2, SUR2B/Kir6.2, and SUR2B/Kir6.1 channels with high potency but activated SUR2A/Kir6.2 channels with only lower potency. In all cases, the currents were completely blocked by 3 μm glibenclamide, the sulfonylurea that blocks K^ATPchannels. The current–voltage relations showed inward rectification and a reversal potential of 0 mV (fig. 1b). Single-channel conductance calculations of the SUR1/Kir6.2, SUR2A/Kir6.2, SUR2B/Kir6.2, and SUR2B/Kir6.1 channels were 72 ± 2 (n = 21), 69 ± 3 (n = 18), 62 ± 2 (n = 20), and 26 ± 2 pS (n = 22) at −60 mV membrane potential, respectively.

These functional properties of the reconstituted SUR1/Kir6.2, SUR2A/Kir6.2, SUR2B/Kir6.2, and SUR2B/Kir6.1 channels do represent those of the native K^ATPchannels that are present on pancreatic β cells, sarcolemmal cardiomyocytes, nonvascular smooth muscle cells, and vascular smooth muscle cells, respectively. 2,4–6Consequently, these recombinant channels were used as experimental models to characterize the function of the native K^ATPchannels in greater detail.

To assess the effects of propofol or thiamylal on recombinant K^ATPchannels, we measured single-channel currents on inside-out patch configurations in the presence of these drugs. The SUR1/Kir6.2, SUR2A/Kir6.2, and SUR2B/Kir6.2 channel currents were inhibited by application of 100 μm propofol to the intracellular membrane surface, with relative channel activities decreasing to 0.40 ± 0.11, 0.39 ± 0.07, and 0.37 ± 0.09 of control, respectively (fig. 2a). However, propofol did not significantly inhibit the SUR2B/Kir6.1 channel currents. Thiamylal at 300 μm blocked the SUR1/Kir6.2, SUR2A/Kir6.2, SUR2B/Kir6.2, and SUR2B/Kir6.1 channel currents, with relative channel activities decreasing to 0.71 ± 0.12, 0.28 ± 0.11, 0.22 ± 0.08, and 0.24 ± 0.06 of control, respectively (fig. 2b). The inhibitory effects of thiamylal and propofol on K^ATPchannel activities were reversible because the channel activities recovered after washout (fig. 2).

The dose-dependent effects of thiamylal and propofol on K^ATPchannel currents are shown in figures 3 and 4, respectively. The EC⁵⁰s and Hill coefficients of propofol and thiamylal for different types of K^ATPchannels are summarized in table 1. Propofol inhibited the SUR1/Kir6.2, SUR2A/Kir6.2, and SUR2B/Kir6.2 channel activities with equivalent potencies, whereas even high concentrations of propofol had no significant inhibitory effects on the SUR2B/Kir6.1 channels. Thiamylal inhibits the SUR2A/Kir6.2, SUR2B/Kir6.2, and SUR2B/Kir6.1 channel activities with high affinity, but inhibits the SUR1/Kir6.2 channel activities with lower potency. In all cases, the blockades by thiamylal and propofol did not significantly change the conductance of the K^ATPchannels, and the Hill coefficients were close to unity, which indicates that only a single propofol or thiamylal molecule has to interact with the channel to inhibit it. Table 1also indicates that the inhibitory effects of propofol and thiamylal for the SUR2A/Kir6.2 channels are similar to those previously reported for the native rat cardiac K^ATPchannels. 24,25 

It has been previously reported that although wild-type Kir6.2 alone does not show functional channel activity, removal of the last 26 or 36 amino acids at the C-terminus of Kir6.2 (Kir6.2ΔC26 or Kir6.2ΔC36) results in channels that show significant currents in the absence of SUR. 7We confirmed this observation by using a Kir6.2ΔC36 mutant, which showed single-channel currents (fig. 5a). Coexpression of SUR1 enhanced the Kir6.2ΔC36 currents. The Kir6.2ΔC36 currents were blocked by ATP, which confirms that Kir6.2 bears an intrinsic ATP-inhibitory site. The current–voltage relation for the Kir6.2ΔC36 currents was the same as those of SUR1/Kir6.2ΔC36.

Propofol at 100 μm and thiamylal at 1,000 μm inhibited the Kir6.2ΔC36 currents, with relative channel activities decreasing to 0.35 ± 0.12 ms, and 0.27 ± 0.09 of control, respectively (fig. 5b). The dose-dependent effects of thiamylal and propofol on Kir6.2ΔC36 currents are shown in figure 5c. The half-maximal blocks (EC⁵⁰) of propofol and thiamylal are summarized in table 1. These observations suggest that both propofol and thiamylal target the Kir6.2 subunit. SUR does not enhance the propofol sensitivity of Kir6.2, but the inhibitory effect of thiamylal was enhanced by coexpression with SUR, especially coexpression with the SUR2 subunit. The Hill coefficients of propofol and thiamylal for Kir6.2ΔC36 did not change in comparison with the other reconstituted K^ATPchannels (table 1).

We next identified the regions of Kir6.2 that play critical roles in the inhibition of Kir6.2ΔC36 channel activity mediated by propofol or thiamylal using systematically mutating Kir6.2ΔC36. The effects of propofol (100 μm), thiamylal (1,000 μm) or ATP (1 mm), which all inhibit wild-type Kir6.2ΔC36 currents by less than 50%, were tested on each mutant (fig. 6).

It has been previously reported that the mutations that affect ATP sensitivity are located in two distinct Kir6.2 lesions, namely, the R50D lesion in the N-terminus and a lesion in the C-terminus that includes C166S, T171A, K185Q, and G334D. 26,27We confirmed these observations (fig. 6). Interestingly, propofol sensitivity was also decreased by the C166S, T171A, and K185Q mutations. We identified another mutation, R31E, that also suppressed the ability of propofol to inhibit Kir6.2ΔC36 currents (P < 0.001 vs.  wild-type Kir6.2ΔC36). Several mutations, including C166S and T171A, also caused smaller but nonsignificant shifts in the ability of 1,000 μm thiamylal to inhibit Kir6.2ΔC36 currents (fig. 6).

Recent mutagenesis studies have suggested that C166 or T171 in Kir6.2 plays a role in the intrinsic gating of the channel, possibly by influencing a gate located at the intracellular end of the pore. 26–28We found that thiamylal but not propofol increases the long closed time and decreases the channel P^oof the C166S and T171A mutants (fig. 7and table 2).

We have demonstrated here by using K^ATPchannels reconstituted in COS-7 cells that the intravenous anesthetics propofol and thiamylal specifically inhibit particular K^ATPchannels. Propofol inhibits Kir6.2-containing channels combined with any of the three SUR molecules tested (SUR1, SUR2A, and SUR2B) but has no effect on SUR2B/Kir6.1 channels, whereas thiamylal strongly blocks SUR2A/Kir6.2, SUR2B/Kir6.2, and SUR2B/Kir6.1 channels but has a weaker effect on SUR1/Kir6.2 channels. These observations suggest that propofol and thiamylal could have tissue-specific inhibitory actions in vivo . These observations are also consistent with our previous findings that both propofol and thiamylal inhibit the native rat cardiac K^ATPchannel (SUR2A/Kir6.2) in patch clamp configuration. 24,25 

That propofol selectively blocks Kir6.2-containing channels and also inhibits Kir6.2ΔC36 currents in the concentration range tested supports the notion that Kir6.2 may be the primary target of propofol (table 1). In addition, that propofol does not significantly inhibit SUR2B/Kir6.1 channels suggests that the Kir6.1 does not bear the propofol inhibitory site found on Kir6.2. This makes propofol the first drug reported to selectively block Kir6.2 but not Kir6.1.

The isoforms of Kir6.2 that lack the C-terminal 26 or 36 amino acids retain their sensitivity to ATP as an intrinsic property. 7In the study reported here, we demonstrated that the K185Q mutation in Kir6.2ΔC36 eliminates the abilities of both ATP and propofol to inhibit channel activity without noticeably affecting the single-channel kinetics (fig. 6). This indicates that the site by which propofol mediates K^ATPchannel inhibition is at least partly identical to that involved in the ATP block. Recent studies have also suggested that apart from the C-terminal K185 residue, the distal part of the C-terminal region (amino acids 333–338) and the N-terminal R50 residue participate in ATP sensitivity. 26,28Although mutations of these regions (R50G, G334D) did not abrogate propofol-mediated channel inhibition, we did identify another mutation in the N-terminus, namely, R31E, which significantly reduces the inhibitory effects of propofol (fig. 6). These results indicate that both the N- and C-termini of Kir6.2 participate in the inhibition mediated by propofol as well as that induced by ATP.

In contrast with propofol, we found that thiamylal inhibits all four of the recombinant sarcolemmal K^ATPchannels (albeit SUR1/Kir6.2 less potently;fig. 4and table 1) as well as native rat cardiac K^ATPchannels 24and the Kir6.2ΔC36 channels. In addition, although SUR molecules did not enhance the propofol sensitivity of Kir6.2ΔC36 channels, the thiamylal sensitivity of Kir6.2ΔC36 channels was enhanced by coexpression with SUR, especially SUR2, suggesting that thiamylal likely has tissue-specific effects based on differential sensitivities to thiamylal exhibited by the various types of the K^ATPchannels (table 1, EC⁵⁰values). Furthermore, the current study indicates more important findings regarding the molecular mechanisms of thiamylal actions on various types of the K^ATPchannels. One possibility is that thiamylal may bind to both the SUR and Kir molecules. Another plausible possibility is that thiamylal acts on the Kir subunit, but its action is modulated by the SUR, because the Hill coefficients of approximately 1.1–1.2 suggest that the binding of one thiamylal is sufficient to result in the inhibition of channel activity. In addition, the notion that SUR modulates thiamylal sensitivity is also supported from the EC⁵⁰value where the Kir6.2ΔC36, in the absence of SUR, is the least sensitive to thiamylal. Our data also show that for the SUR2B/Kir6.2 and SUR2B/Kir6.1 channels, the EC⁵⁰values are similar despite differences in the Kir subunit. On the other hand, for the SUR1/Kir6.2 and SUR2A/Kir6.2 channels, the EC⁵⁰values are not similar, likely because of the different SURs. Again, this can be accounted for in the absence of thiamylal binding to SUR.

It has been reported that the cytosolic end of the second transmembrane domain of Kir6.2 may play an important role in the gating of the K^ATPchannel pore. 26–28In agreement with these reports, we showed here that mutations in this region, namely Kir6.2ΔC36–C166S and –T171A, markedly increase the channel P^oby reducing the long close time (table 2). The finding that the inhibitory effect of propofol is also reduced by these mutations suggests that these mutations affect the ability of propofol to block channel activity by changing the channel gating kinetics rather than by altering the affinity of propofol for its binding site. In contrast, thiamylal increased the long closed times with all of the Kir6.2ΔC36 mutants. In particular, thiamylal converted the long burst kinetics of Kir6.2ΔC36-T171A currents to the long closed kinetics that were typically observed with the Kir6.2ΔC36 channels (fig. 7band percent long closed time of Kir6.2ΔC36-T171A in table 2). It is therefore possible that thiamylal acts as an open channel blocker of the Kir6 channel.

Recent investigations have established that K^ATPchannel activation plays an important role in ischemic preconditioning of myocardium and neural tissue, during skeletal muscle ischemia, and in the regulation of vascular smooth muscle tone. 10–12In addition, it has been shown that these desirable endogenous effects of K^ATPchannel activation can be induced pharmacologically by K^ATPchannel openers. 29,30These observations suggest new therapeutic intervention strategies that may specifically benefit patients who are at risk for development of untoward ischemic events during cardiac, vascular, or neurologic surgery. In addition, it seems that volatile anesthetics, including isoflurane, desflurane, and sevoflurane, can also protect the myocardium against stunning and infarction by activating K^ATPchannels. 14–16In contrast to the volatile anesthetics, however, we demonstrate here that two representatives intravenous anesthetics, propofol and thiamylal, interact with one or both of the Kir6 subunits to block the K^ATPchannel currents in a concentration-dependent manner. Recent functional studies have provided direct evidence that each Kir6.1 and Kir6.2 play separate physiologic roles. 31–34Kir6.2 forms the pore region of the K^ATPchannels in the heart, brain, and skeletal muscle and activation of these channels has shown to be important for cell protection. 10–12In contrast, the Kir6.1-containing K^ATPchannel is critical in the regulation of vascular tonus, especially in the coronary arteries, and it is known that it protects against vasospasm during and after myocardial ischemia. 31Therefore, our results indicate that intravenous anesthesia with propofol and thiamylal may impair the beneficial effects mediated by K^ATPchannel activation in various organs. However, it is possible that propofol may not significantly inhibit channel activity at the concentrations that are generally used in the clinical setting. Plasma concentrations of propofol up to 50 μm after clinical intravenous induction administration have been reported. 35If protein binding is taken into account, the clinically relevant concentration of propofol is less than 2 μm. 25The concentrations of propofol needed to inhibit K^ATPchannel activity in vitro  are higher than these postulated free plasma concentrations, which suggests that propofol at the concentrations used clinically may not affect K^ATPchannel activity. In the current study, the differential propofol effects on Kir6.1 and Kir6.2 are evident at concentrations greater than 10–30 μm (fig. 3); it is unlikely that this differential effect will be encountered in the clinical setting. However, because propofol is the first drug reported to selectively block Kir6.2 but not Kir6.1, it may be useful in other experimental settings that require modulation of the functions induced by Kir6.2.

Unlike propofol, thiamylal may well significantly depress K^ATPchannel activity when it is used as an anesthetic. Plasma concentrations of thiamylal up to 0.5 mm after clinical intravenous induction administration have been reported. 36If protein binding is taken into account, the clinically relevant concentrations of thiamylal range from 0.05 to 0.08 mm. 24Thiamylal inhibits all four recombinant K^ATPchannels at these clinical concentrations (fig. 4). Therefore, it is likely that when thiamylal is used as an intravenous anesthetic, it may inhibit the K^ATPchannel activities in the patient. These results may suggest that thiamylal impairs the endogenous organ-protective mechanism mediated by K^ATPchannels against intraoperative ischemic or hypoxic injury. However, there are other well-established mechanisms of organ protection that do not involve K^ATPchannel activities. Thiamylal is a likely candidate for neuroprotection and has been used as such in our country. 37 

Our study has several limitations. First, we combined cDNAs from different species (human and rat) to reconstitute K^ATPchannels. Sequence differences between human and rat cDNAs may induce possible influences on the electrophysiologic properties of K^ATPchannels. However, in most previous studies, 2,4K^ATPchannels were reconstituted by the combination of Kir and SUR cDNAs from different species (rat or mouse), and it has been confirmed that the electrophysiologic properties of all kinds of reconstructed K^ATPchannels are similar to those of the native K^ATPchannels. In addition, although we used the same amount of SUR cDNA and Kir cDNA for transfection, the genomic integration of the various constructs may have been different, and a varying ratio of SUR versus  Kir may affect electrophysiologic findings. Therefore, it might be better for us to establish the level of expression as well as the ratio of SUR versus  Kir subunits by polymerase chain reaction method and Western blot analyses. However, in the current study, we confirmed that the sensitivity to ATP, diazoxide, and glibenclamide and the single-channel conductance of all kinds of reconstituted K^ATPchannels were similar to those of native K^ATPchannels (fig. 1). Therefore, we expect that the reconstituted K^ATPchannels in the current study can be used as experimental models to characterize the function of the native K^ATPchannels and that we can draw conclusions from our experimental model. Second, we studied the effects of propofol and thiamylal on sarcolemmal K^ATPchannels because mitochondrial K^ATPchannels have not been cloned. However, in the heart and brain, mitochondrial rather than sarcolemmal K^ATPchannels might play an important role for the protection of these tissues. In the future, we must study the molecular mechanisms of these anesthetics on reconstituted mitochondrial K^ATPchannels.

In conclusion, propofol inhibits all channels with Kir6.2 but does not inhibit SUR2B/Kir6.1, which is the vascular smooth muscle channel. In contrast, thiamylal inhibits all channels with either Kir6.1 or Kir6.2. These results, as well as site-directed mutagenesis studies, suggest that propofol and thiamylal may act via  the Kir6.2 subunit, albeit by different molecular mechanisms. The N- and C-termini of Kir6.2 participate in the inhibition of K^ATPchannel by propofol. In the case of thiamylal, the SUR subunit seems to modulate anesthetic activity on the Kir subunit.