Cardiac Calcium Release Channel (Ryanodine Receptor 2) Regulation by Halogenated Anesthetics

RYANODINE receptor 2 receptors (RyR2) are the Ca²⁺ release channel in cardiac sarcoplasmic reticulum (SR). RyR2 can be activated by Ca²⁺ binding to cytoplasmic or luminal sides of the channel^1–3  and can be inhibited by cytoplasmic and luminal Mg²⁺, which acts as a competitive Ca²⁺ antagonist.^4,5  RyR2 also possesses cytoplasmic facing adenine nucleotide–binding sites⁶  by which physiologic concentrations of adenosine triphosphate (ATP; 5 mM) strongly activates channel opening.⁷ 

Many ion transporters in cardiomyocytes are modulated by halogenated anesthetics such as halothane, isoflurane, enflurane, desflurane, and sevoflurane.⁸  Such modulation includes activation of RyR2-mediated SR Ca²⁺ release and inhibition of SERCA2a-driven Ca²⁺ uptake into the SR, both leading to SR Ca²⁺ depletion and reducing cardiac function,⁹  contributing to protecting against injury due to ischemia and reperfusion.¹⁰  (During ischemia, there is Ca²⁺ overload of the SR because of RyR2 inhibition by acidosis, reduced [ATP], and elevated [Mg²⁺]. Upon reperfusion, there is a massive release of SR Ca²⁺ as RyR2 recovers from these ischemic conditions.¹¹ )

Halothane has been proposed to act as a surrogate for ATP, activating RyR2 during periods of ischemia, thus ameliorating Ca²⁺ overloading of the SR.¹²  This proposal was based on the findings that (1) halothane had a more pronounced effect on Ca²⁺ release when ATP levels were low and (2) adenosine appeared to antagonize halothane activation, supporting the notion that halothane and ATP act at a common adenine nucleotide–binding site on RyR2. However, a previous study using permeabilized rat ventricular myocytes found no competitive action between halothane and ATP on SR Ca²⁺ release.¹³  Given the difficulty in precisely controlling the RyR2 environment in cell studies, unequivocal determination of the nature of these molecular interactions requires direct measurements of RyR2 activation in single-channel studies. Although single-channel measurements of RyR2 in artificial lipid bilayers find that clinically relevant concentrations of halothane cause substantial activation of RyR2,^14,15  there has been no examination of competition between halothane and adenine nucleotides in single channels. In this study, we use single-channel measurements to test the hypothesis that halothane and adenine nucleotides share a common activation site on RyR2.

Dantrolene is a well-known inhibitor of Ca²⁺ release in skeletal and cardiac muscles^16–20  that has been used for treatment for halothane-induced malignant hyperthermia in skeletal muscle and arrhythmias in cardiac muscle. Our recent finding that calmodulin is required for dantrolene inhibition of ryanodine receptor 1 (RyR1) and RyR2 explains why dantrolene had no effect on the activity of RyR2 in previous single-channel studies,^15,21–23  where calmodulin would have been absent (calmodulin readily dissociates from the ryanodine receptor (RyR) complex²⁴ ). By adding exogenous calmodulin, we have for the first time employed single-channel recording to probe the mechanism for dantrolene inhibition of halothane-induced Ca²⁺ release by RyR2.

Overall, the mechanism of activation by halogenated anesthetics is poorly defined. Here, we make the first measurements of sevoflurane and desflurane activation of RyR2, the half-activating concentrations (K_a) for other anesthetics and how they alter RyR2 regulation by intracellular ATP, Ca²⁺, and Mg²⁺.

Animal tissues were obtained with approval from the Animal Care and Ethics Committee of the University of Newcastle (Newcastle, New South Wales, Australia; approval number A-2009-153). Sheep hearts were obtained from ewes anesthetized with 2% propofol intravenous (IV) followed by oxygen or isoflurane and euthanized by barbiturate overdose (pentobarbitone, 150 mg/kg IV) before the heart was removed. RyR2 was isolated from the heart muscle as described elsewhere.²⁵ 

RyRs from sheep heart were incorporated into artificial lipid bilayers that were formed from phosphatidylethanolamine and phosphatidylcholine (8:2 wt/wt) in n-decane (50 mg/ml). For single-channel recording, the cis (cytoplasmic) and trans (luminal) solutions contained 250 mM Cs⁺ (230 mM cesium methane sulfonate plus 20 mM cesium chloride). Before experiments, halogenated anesthetic concentration in the cis solutions was adjusted by adding liquid anesthetic and sealing the solutions in glass-teflon syringes. During recordings, these cis solutions were applied to the bilayer using a syringe pump (constructed in-house), providing continuous local perfusion via a tube placed in close proximity to the bilayer (fig. 1). Thus, switching between any of eight syringe pumps could produce solution exchange at the bilayer within 3 s. The composition of trans solution was altered by means of aliquot additions directly to the bath.²⁶ 

Halogenated anesthetics were obtained from ICI Pharmaceuticals (Australia). Solutions were pH buffered using 10 mM TES (N-tris[hydroxymethyl] methyl-2-aminoethanesulfonic acid, ICN Biomedicals [Australia]) and titrated to pH 7.4 using CsOH (ICN Biomedicals). A Ca²⁺ electrode (Radiometer) was used to determine the purity of Ca²⁺ buffers and Ca²⁺ stock solutions as well as free [Ca²⁺] greater than 100 nM. Free Ca²⁺ was adjusted with CaCl₂ and buffered using (1) 4.5 mM BAPTA (1,2-bis(o-aminophenoxy)ethane- N,N,N′,N′-tetraacetic acid, obtained from Invitrogen [Australia]) for free [Ca²⁺] less than 1 μM; (2) dibromo BAPTA (up to 2 mM) for free [Ca²⁺] between 1 and 10 μM; or (3) sodium citrate (up to 6 mM) for free [Ca²⁺] between 10 and 50 μM. In solutions containing ATP (ATP chelates Ca²⁺ and Mg²⁺), free levels of Mg²⁺ (added as MgCl₂) were calculated using estimates of ATP purity and effective Mg²⁺ binding constants that were determined previously under our experimental conditions.²⁷  Cesium salts were obtained from Sigma-Aldrich Chemical Company (Australia) and CaCl₂ and MgCl₂ from BDH Chemicals (USA).

An Axopatch 200B amplifier (Axon Instruments Pty, Ltd, USA) controlled the bilayer potential and recorded unitary currents. Electrical potential differences are expressed as cytoplasmic potential relative to luminal potential. The channel currents were recorded during the experiments using a 50-kHz sampling rate and 5-kHz low-pass filtering. Before analysis, the current signal was redigitized at 5 kHz and low-pass filtered at 1 kHz with a Gaussian digital filter. Single-channel open probability (P_o), mean open time (T_o), mean closed times (T_c), and channel opening rate (k_o =1/T_c) were measured using a threshold discriminator at 50% of channel amplitude. These parameters were measured from single-channel records using Channel3 software (Nicholas W. Laver). Unless otherwise stated, measurements were performed with the cis solution voltage clamped at −40 mV.

Data are presented as means ± SD or 95% CIs as indicated in figure legends. Individual readings of channel properties were derived from 30 to 100 s of RyR2 recording depending on experimental conditions as described previously.²⁵  Hill equations were fitted to the dose–response data by the method of least-squares, and errors on the Hill equation parameters were derived using the simplex method.²⁸  Hypothesis testing was carried out using either two-tailed Student’s t test or ANOVA (6.0; Graphpad Software Inc., USA) as described in the text. In the case of figure 3, B and C, statistical analyses were programmed using SAS v9.4 (SAS Institute, USA). Opening rate (k_o) and opening duration (T_o) were regressed in separate models on dose, anesthetic, and a dose–anesthetic interaction using mixed-effect regression models, with a random effect for channel and fixed effect for anesthetic. A negative binomial distribution was utilized for the k_o outcome, and a γ distribution was utilized for the T_o outcome. Normal and Poisson distributions were initially trialed for the k_o outcome; however, nonnormality of the residuals in the former and overdispersion in the latter models led to the final choice of the negative binomial distribution. A normal distribution was originally considered for the T_o outcome; however, nonnormality of the residuals led to the final choice of a γ distribution instead.

The data analysis for figure 6 compared the effect of adenosine in the presence and absence of ATP. A two-way ANOVA with repeated measures was used for the no-halothane data; a mixed model was used for the with-halothane data to handle missing values in the repeated measures. Unpaired t tests for interactions from both models were then used to compare combinations of adenosine and ATP in the absence of halothane and in the presence of halothane. Post hoc pairwise comparisons were made if significant differences were detected. Recommended adjustments were made for multiple comparisons (e.g., Holm–Sidak or Tukey–Kramer). In this study, it was not feasible to randomly assign RyR2 to different experimental conditions or to impose blinding methods. The number of experiments was chosen based on the previous experience of experimental scatter.

RyR2 from sheep heart was incorporated into lipid bilayers, and their activity was measured in the presence of a 0.1 μM cytoplasmic Ca²⁺ (i.e., typical of end diastole) and 0.1 mM luminal Ca²⁺, with 2 mM cytoplasmic ATP (vehicle solution; fig. 2). Perfusion of the cytoplasmic bath with vehicle solution containing halothane at mM concentrations caused an increase RyR2 activation (fig. 2A) that was reversed on washout within the time taken for solution exchange at the bilayer (approximately 3 s). The degree of RyR2 activation (fig. 3A) depended on the anesthetic (P < 0.0001, one-way ANOVA) with post hoc multiple comparisons test (Holm–Sidak; P < 0.05), showing that halothane (figs. 2B and 3A), desflurane (fig. 2C), and isoflurane (fig. 2D) caused the largest activation followed by enflurane (fig. 2E) and sevoflurane (fig. 2F). The concentration dependencies of RyR2 activation by these halogenated anesthetics are shown in figure 3, and the parameters of the Hill fits to concentration dependencies of RyR2 open probability (P_o; fig. 3A) are shown in table 1. Most half-maximal activating concentration (K_a) showed no significant dependency on the type of anesthetic with values in the range of 1.3 to 2.1 mM. The exception was isoflurane (K_a = 5.3 mM), which had a significantly higher K_a than that for halothane (K_a = 1.5 mM; P < 0.05, one-way ANOVA with Tukey–Kramer multiple comparisons test). Using regression analysis (see Methods), halothane, enflurane, and desflurane activation could be resolved as increases in both channel opening rate (k_o, the reciprocal of mean closed duration, P = 0.0005, P < 0.0001, and P = 0.08, respectively) and mean open duration (T_o; figs. 3, B and C, respectively; P < 0.0001, P < 0.0001, and P = 0.0006, respectively).

We added halothane and enflurane to the luminal bath in bilayer experiments (5 or 50 mM by aliquot additions of liquid anesthetic) to see if they could activate RyR2 from that side of the membrane. The relative effects of these halogenated anesthetics on P_o are shown in figure 4 where it can be seen that they produced no significant activation of RyR2 from the luminal bath.

Halothane-induced activation of SR Ca²⁺ release in permeabilized ventricular myocytes has been found to require higher halothane concentrations in the presence of ATP.¹²  We looked for similar phenomena in the activation of single RyR2 in bilayers and found that halothane was a potent activator of RyR2 in both the absence and presence of 2 mM ATP (fig. 5; one-way ANOVA, P < 0.012). We also found that the K_a for halothane activation was substantially higher (one-way ANOVA, P < 0.001) in the absence of ATP (table 1), which is the opposite of that expected from competition between ATP and halothane (see Discussion). In permeabilized ventricular myocytes, adenosine (5 mM) has been seen to antagonize halothane activation of SR Ca²⁺ release in the near-absence of ATP (0.05 mM).¹²  Here, we used the same combinations of adenine nucleotide in bilayer experiments to investigate if halothane and adenosine competitively bind to a common site on RyR2 (fig. 6). In the absence of halothane, adenosine (5 mM) did not inhibit RyR2 activity in the absence of ATP but did so in the presence of ATP (two-way ANOVA with repeated measures, interaction P = 0.0077; see Methods). Adenosine caused a two-fold increase in RyR2 activation, whereas 0.05 mM ATP caused a 10-fold increase in RyR2 activity. However, in the presence of 0.05 mM ATP, addition of 5 mM adenosine inhibited channel activity and totally abolished the activating effect of ATP, in accord with these previous studies,^29,30  demonstrating competition between ATP and adenosine for the adenine nucleotide–binding site on RyR2. In the presence of halothane, ATP and adenosine had similar relative effects on RyR2 activity as in the absence of halothane, albeit with higher levels of activation; i.e., adenosine did not inhibit RyR2 activity in the absence of ATP but did so in the presence of ATP (two-way ANOVA with repeated measures, interaction P = 0.0226). These data indicate that adenosine is not a competitive antagonist for halothane, but rather it is a competitive antagonist for ATP. Therefore, the data in figures 5 and 6 do not reveal any competition between halothane and adenine nucleotides on RyR2 activity.

Although calmodulin associates with RyR2 in the cell, it is labile and can readily dissociate when RyR2 is isolated and incorporated into artificial lipid bilayers.^24,31  To investigate the effect of halothane on the RyR2/calmodulin complex, we added 100 nM calmodulin in the cytoplasmic solution (fig. 5). In the absence of halothane, calmodulin caused a moderate (approximately 50%) reduction in the open probability of RyR2 in accord with previous studies.^32,33  Addition of halothane caused RyR2 activation with the same K_a and P_max as seen in the absence of calmodulin (table 1). In the presence of calmodulin, dantrolene (10 μM in the cytoplasmic solution) caused an approximately 40% reduction in RyR2 open probability (fig. 7). There was no significant difference in dantrolene inhibition in the absence of halothane, in the presence of halothane concentrations that produce half activation (2 mM), or at full halothane activation of RyR2 (10 mM; fig. 7B).

The effect of halothane (10 mM) on the activation of RyR2 by cytoplasmic Ca²⁺ (using 0.1 mM luminal Ca²⁺) is shown in figure 8, and the parameters of the Hill fits to the Ca²⁺-dependencies of P_o in figure 8A are given in table 2. Halothane reduced the K_a for RyR2 Ca²⁺ activation from 4.3 to 1.6 μM and caused a four-fold increase in P_o at nM cytoplasmic Ca²⁺ (table 2; P_min). These changes in activity were associated with significant increases in both mean opening rates (Student’s t test, P = 6 × 10⁻⁴) and open time (P = 1.5 × 10⁻⁶; fig. 8, B and C).

In the presence of 0.1 μM cytoplasmic Ca²⁺, luminal Ca²⁺ is a moderate activator of RyR2 (fig. 9). At −40 mV bilayer potential (lumen at ground) and in the absence of halothane, RyR P_o had a bell-shaped response to luminal Ca²⁺ with half-activation at 53 μM and half-inhibition at 0.8 mM (table 2). The activating phase is associated with increases in both mean open time and opening rate, whereas the inhibiting phase was due only to a decrease in mean open time (fig. 9, B and C). A +40 mV bilayer potential, which opposes the flow of Ca²⁺ from lumen to cytoplasm, nearly abolished luminal Ca²⁺ activation at submicromolar concentrations (fig. 9, D–F). The presence of 10 mM halothane had no significant effect in the absence of luminal Ca²⁺, but it did increase peak luminal Ca²⁺ activation by nearly 10-fold (P_max in table 2) along with a three-fold decrease in K_a. Again, these effects were due to increases in mean open time and opening rate.

It is well known that Mg²⁺ in either the cytoplasmic or luminal solutions inhibit RyRs as shown in figure 10.³⁴  Hill fit parameters for cytoplasmic and luminal Mg²⁺ inhibition are shown in table 3. The addition of 10 mM halothane had no significant effect on either of these Mg²⁺ inhibition phenomena.

This study shows that halogenated anesthetics directly and reversibly activate cardiac RyR2 with halothane being the strongest activator, in agreement with previous single-channel studies of RyR2.^14,15  We also (1) demonstrate that the reversibility of activation is rapid and occurs within the time of solution exchange (fig. 2A); (2) demonstrate that luminal application of these anesthetics does not significantly activate RyR2 (fig. 4), (3) make the first direct measurements of the K_a’s for RyR2 activation by halogenated anesthetics (fig. 3; table 1); and (4) determine the effects of the clinically used anesthetics, sevoflurane and desflurane on RyR2. K_a values for anesthetics were in the range of 1.3 to 2.1 mM (1.4 to 2.6 minimum alveolar concentration [MAC]) with the exception of isoflurane (5.3 mM, 6.6 MAC). The effect of anesthetics at clinically used concentrations (1 MAC) was calculated from theoretical Hill fits to their dose–responses in figure 3. Under these conditions, RyR2 activity increases by 120% with halothane, approximately 50% with enflurane and desflurane and is not significantly increased by isoflurane and sevoflurane, consistent with the effects of anesthetics on activation of SR Ca²⁺ release in cardiac muscle.^35,36  Thus, we show that halothane is the most active of the anesthetics on RyR2 activity. Halothane is rarely used anymore, and the modern anesthetics tested had very little effect in inducing RyR2 opening.

Our previous study³⁷  has shown that in the presence of calmodulin, dantrolene inhibits RyR2 by up to 45% with a concentration for half-maximal inhibition of 160 nM. Therefore, at clinical concentrations used here (approximately 10 μM³⁸ ), dantrolene exerts its maximal inhibitory action. Here, we show that dantrolene produces similar relative inhibition (approximately 40%) regardless of the halothane concentration, indicating that dantrolene does not alter the K_a for halothane activation. This degree of inhibition is sufficient to completely negate the activating effects of sevoflurane, isoflurane, desflurane, and enflurane. It also negates halothane activation at 1 MAC and substantially alleviates its activation at 2 MAC.

Our results indicate that halothane activation of RyR2 is different from that seen in the skeletal isoform, RyR1. RyR2 is much less sensitive to halogenated anesthetics than RyR1; the K_a’s for anesthetic activation of RyR2 given in table 1 are an order of magnitude higher than those for RyR1 (K_a approximately 0.1 mM^39,40  and substantially higher than seen in SR Ca²⁺ release through drosophila RyR (K_a approximately 0.6 mM⁴¹ ). Also, halothane activation of RyR1 required the presence of ATP and high concentrations of Ca²⁺ in the cytoplasmic solutions,¹⁵  whereas here and in a previous study,¹⁴  halothane activation of RyR2 does not have this requirement for ATP. However, a common feature of halothane activation of RyR1 and RyR2 is that in both isoforms, halothane strongly potentiates the activating effect of luminal Ca²⁺.²³ 

There is now substantial evidence that Ca²⁺ binding at luminal or cytoplasmic facing sites on the RyR2 protein complex can trigger channel openings.^1–3  Maximally activating concentrations of halothane increased RyR2 sensitivity to activation by cytoplasmic Ca²⁺ indicating that halothane can potentiate the cytoplasmic facing Ca²⁺ activation mechanism of RyR2.

The largest effect of halothane was its potentiation of luminal Ca²⁺ activation of RyR2. Previous studies have found that luminal Ca²⁺ can activate RyR2 by binding to luminal facing sites on the RyR2 molecule and by permeating the open RyR2 pore and binding to cytoplasmic facing sites to sustain channel openings (i.e., Ca²⁺ feed-through).^1–3  Our analysis of the voltage-dependence of RyR2 mean open time and opening rates indicates that halothane potentiates luminal Ca²⁺ activation by a direct action on both the luminal Ca²⁺ sites and cytoplasmic Ca²⁺ sites. Evidence for the former is that halothane decreases RyR2 closed times (fig. 9, B and E; k_o = 1/T_c), during periods where there can be no Ca²⁺ feed-through. Evidence for the latter is that halothane substantially increases RyR2 mean open durations, but only when the bilayer potential difference favors the luminal-to-cytoplasm flow of Ca²⁺ (−40 mV; fig. 9, C and F). Therefore, we envisage that halothane-induced changes in the cytoplasmic Ca²⁺ activation mechanism would account for the increased effect of cytoplasmic Ca²⁺ on mean open durations and opening rates seen in figure 8, B and C together with most of the increased effect of luminal Ca²⁺ on mean open durations in figure 9C.

The literature is currently divided on whether halothane and ATP share a common activating site on RyR2.^12,13  A key finding in the study by Yang et al.,¹²  supporting the idea that halothane and adenine nucleotides act at a common site was that adenosine was able to reduce halothane activation of SR Ca²⁺ release in the near-absence of ATP (0.05 mM). However, Herland et al.¹³  found that halothane could activate SR Ca²⁺ release regardless of the ATP concentration (0 to 11 mM), indicating that the mechanisms of activation were independent. Our experiments on the actions of ATP and adenosine on halothane activation of RyR2 reconcile these conflicting studies. We found that there was marked halothane activation of RyR2 in both the absence and presence of ATP (2 mM; fig. 6) in agreement with Ca²⁺ release in permeabilized cardiomyocytes¹³  and with a previous single RyR2 channel study.¹⁵  We also found that in the complete absence of ATP, halothane activation was slightly increased by adenosine, rather than inhibited as seen in the study of Yang et al.¹²  However, in that study, adenine effects were actually measured in the presence of low levels of ATP (0.05 mM) because some ATP was required for loading of the SR by SERCA pump. We found that in the presence of 0.05 mM ATP, adenosine did reduce RyR2 activity in the presence of halothane, an effect also seen in the absence of halothane. Thus, our data are consistent with the findings of Yang et al.,¹²  even though we found no evidence of competition between halothane and adenosine nucleotide for RyR2 activation. We conclude that the inhibitory action of adenosine in the presence of both halothane and low concentrations of ATP is due to competition between adenosine and ATP for the nucleotide-binding site on the RyR2, rather than between adenosine and halothane. It is interesting to note that Yang et al.,¹²  reported that halothane had a larger effect on SR Ca²⁺ release at lower [ATP], an effect that could be explained by a common binding site. However, our finding that halothane sensitizes RyR2 to luminal Ca²⁺ (fig. 9), similar to that reported for increasing [ATP],⁴²  it increases an alternative explanation for the finding by Yang et al.¹²  Increasing [ATP]⁴³  or [halothane] decreases the SR luminal [Ca²⁺] “set point” for RyR2 activation where the progressive rise in luminal [Ca²⁺] via SERCA initiates propagating SR Ca²⁺ waves. The minimum attainable SR Ca²⁺ load and the basal luminal Ca²⁺ activation of RyR2 ultimately limit the degree to which this “setpoint” can be lowered. This limitation will produce experimental outcomes that mimic ligand competition at a limited number of binding sites.

Inhibition of the sarcolemmal L-type Ca²⁺ channel, Na/Ca exchanger,⁴⁴  and store-operated Ca²⁺ entry⁴⁵  by volatile anesthetics will inevitably lead to decreased Ca²⁺ loading by cardiomyocytes,⁴⁶  which is likely account for their sustained negative inotropic actions.⁴⁷  Negative inotropy would contribute to cardioprotective properties of anesthetics by depressing cardiac activity and reducing oxygen demand of the myocardium. As previously shown,⁴⁷  halothane, isoflurane, and sevoflurane have similar sustained negative inotropic actions on rat ventricular myocytes, which are largely independent of Ca²⁺ transport by the SR. This is consistent with our data, showing that these three anesthetics have very different actions on the SR Ca²⁺ release channel in spite of their similar inotropic effects.

Our data suggest an additional cardioprotective action of anesthetics whereby they would act as a surrogate agonist for the SR Ca²⁺ release channels in place of ATP when ATP is depleted during ischemia, thus preventing deactivation of RyR2 activity and Ca²⁺ overloading of the SR and the consequent transient aberrant release of SR Ca²⁺ during reperfusion. However, since sevoflurane and isoflurane do not activate RyR2 at clinical concentrations, this mechanism is likely to play a less significant role in their cardioprotective actions than their effects on sarcolemmal transporters or on cell survival signaling pathways in the cytoplasm and mitochondria.^48,49  It is interesting to note that activating RyR2 should potentially lead to anesthetic-induced cardiac arrhythmias.⁵⁰  However, no such concerns exist for these anesthetics. This might be because of compensatory effects of anesthetics on sarcolemmal transporters, a potentially important avenue for future studies.

The authors thank Paul Johnson (School of Biomedical Sciences and Pharmacy, University of Newcastle and Hunter Medical Research Institute, Callaghan, New South Wales, Australia) for his assistance with the experiments and Lucy Leigh, Ph.D. (School of Medicine and Public Health, University of Newcastle, and Hunter Medical Research Institute, Newcastle, New South Wales, Australia) and David Cottee, M.B.B.S., Ph.D., F.A.N.Z.C.A. (School of Biomedical Sciences and Pharmacy, University of Newcastle and Hunter Medical Research Institute, Callaghan, New South Wales, Australia) for their help with the statistics. Thanks to Janice Johnstone, B.Env.Sc. (School of Biomedical Sciences and Pharmacy, University of Newcastle and Hunter Medical Research Institute, Callaghan, New South Wales, Australia) for assistance with the article preparation.

Supported by grant no. APP1082204 from the National Health and Medical Research Council, Canberra, Australian Capital Territory, Australia (to Dr. Laver). Supported by project grant no. 07/017 from the Australian and New Zealand College of Anaesthetists, Melbourne, Victoria, Australia (to Dr. Quail). Supported by infrastructure grant from the Hunter Medical Research Institute, Newcastle, New South Wales, Australia (to Drs. Laver and Quail).

The authors declare no competing interests.