Volatile anesthetics (VAs) have been found to induce a delayed protective response called preconditioning to subsequent hypoxic/ischemic injury. VA preconditioning has been primarily studied in canine and rodent heart. A more genetically tractable model of VA preconditioning would be extremely useful. Here, the authors report the development of the nematode Caenorhabditis elegans as a model of VA preconditioning.
Wild-type and mutant C. elegans were exposed to isoflurane, halothane, or air under otherwise identical conditions. After varying recovery periods, the animals were challenged with hypoxic, azide, or hyperthermic incubations. After recovery from these incubations, mortality was scored.
Isoflurane- and halothane-preconditioned animals had significantly reduced mortality to all three types of injuries compared with air controls. Concentrations as low as 1 vol% isoflurane (0.64 mm) and halothane (0.71 mm) induced significant protection. The onset and duration of protection after anesthetic were 6 and 9 h, respectively. A mutation that blocks inhibition of neurotransmitter release by isoflurane did not attenuate the preconditioning effect. A loss-of-function mutation of the Apaf-1 homolog CED-4 blocked the preconditioning effect of isoflurane, but mutation of the downstream caspase CED-3 did not.
Volatile anesthetic preconditioning extends beyond the vertebrate subphylum. This markedly broadens the scope of VA preconditioning and suggests that its mechanisms are widespread across species and is a fundamental and evolutionarily conserved cellular response. C. elegans offers a means to dissect genetically the mechanism for VA preconditioning as illustrated by the novel finding of the requirement for the Apaf-1 homolog CED-4.
PRECONDITIONING is a cellular adaptive response where exposure to a potentially injurious agent induces a protective response to subsequent noxious exposure. Hypoxia and ischemia are the best characterized preconditioning agents and have been well-studied in cardiac myocytes and cerebral neurons.1,2Both of these cell types exhibit a robust hypoxic and ischemic preconditioning response. Indeed, circumstantial evidence suggests that both angina and transient ischemic attacks may induce a protection from subsequent ischemia in humans by a preconditioning mechanism.3–5
Hypoxia and/or ischemia is inherently dangerous and difficult to titrate and therefore unsuitable as a means to protect against hypoxia and ischemia in the clinical setting. However, other agents besides hypoxia/ischemia have been subsequently found to induce a preconditioning-like protective response. Volatile anesthetics (VAs) are one such preconditioning drug.6,7Isoflurane in particular has been well documented to induce preconditioning in a variety of cell types. As for hypoxic/ischemic preconditioning, cardiac myocytes and cerebral neurons have been widely shown to be protected by anesthetic preconditioning. Limited human trials administering VAs before ischemia in patients undergoing coronary artery bypass surgery suggest that VAs can also act to protect these human cell types from hypoxia/ischemia.8
A number of molecules have been implicated in preconditioning mechanisms.1,6,7,9Reactive oxygen species have been consistently found to be important, if not essential, for both hypoxic and anesthetic preconditioning.8,10Adenosine triphosphate-inhibited potassium (KATP) channels have also been implicated in several different models of preconditioning, including hypoxic and anesthetic preconditioning.7,11Apoptosis has also been suggested by some experiments to be inhibited by preconditioning or to induce preconditioning.9,12,13For neuronal protection, in addition to the aforementioned mechanisms, inhibition of glutamate release has been suggested as a mechanism for VA protection and preconditioning.6These are only a subset of molecules with some evidence for a role in preconditioning.
Given the complexity and likelihood of thus far unidentified components of the preconditioning mechanism, a genetically tractable model for the study of preconditioning would have great potential. Not only could such a genetic model test hypotheses in ways not always feasible by other methods, it might discover components of the preconditioning mechanism not previously identified. We have developed the nematode Caenorhabditis elegans as a model to define molecular mediators of hypoxic cell death.14,15The strengths of the model are many. One can screen through large numbers of genes for those that control hypoxic injury without a requirement for a preexisting bias as to what one will find. This ability to go from phenotype (hypoxia resistance for example) to genotype (the gene variation responsible for the phenotype) enables the discovery of completely novel mechanisms. On the other hand, a large collection of readily available mutants and a nearly whole-genome library of RNAi plasmids that can knock down the expression of a particular gene16allows for rapid and affordable hypothesis testing. Finally, C. elegans has a well-characterized and relatively simple apoptotic cell death pathway and has evolutionarily conserved genes that encode proteins that regulate necrotic and autophagic death pathways.17–27Of course, a limitation of C. elegans for understanding hypoxic injury in humans is its evolutionary distance. It is likely that some mechanisms controlling hypoxic injury in C. elegans will translate to humans and some will not.
We have recently found that C. elegans has a mechanism for hypoxic preconditioning.28We tested the role of the apoptosis pathway in hypoxic preconditioning in C. elegans and found surprisingly that only one component of the pathway, the Apaf-1 homolog CED-4, was required for hypoxic preconditioning. Here, we test whether C. elegans has a mechanism for anesthetic preconditioning and test whether anesthetic and hypoxic preconditioning share a requirement for CED-4.
Materials and Methods
Culture of C. elegans and Strains
All strains were grown at 20° on nematode growth medium (NGM) agar plates seeded with OP50 Escherichia coli bacteria.29The wild-type C. elegans strain used was N2 var Bristol.29,unc-64(md130) has been described previously.30,31,ced-3(n1286) was described and kindly provided by the Horvitz laboratory (Massachusetts Institute of Technology, Cambridge, MA).32 ced-4(n1162) and N2 were obtained from the Caenorhabditis Genetics Center funded by the National Institutes of Health National Center of Research Resources (Bethesda, MD).
For anesthetic preconditioning, uncrowded synchronized populations of young adult animals on uncovered NGM plates with bacteria were placed in small glass chambers.33,34A measured volume of anesthetic (isoflurane: Isothesia, Butler Animal Health Supply, Dublin, OH; halothane: Halocarbon Laboratories, River Edge, NJ) in its liquid phase was deposited through the side port onto the roof of each chamber using a Hamilton syringe, and the side port was quickly sealed. The anesthetic completely volatilizes within 30 s. The volume of anesthetic placed in the chamber was calculated to produce a desired gas phase concentration according to the ideal gas equation, given the known volume of the chamber and the density and molecular weight of the liquid anesthetic. At the end of each assay, the actual VA concentrations in the chambers were measured by gas chromatography (HP 5890A gas chromatograph; Agilent Technologies, Inc., Life Sciences and Chemical Analysis Group, Santa Clara, CA). Chambers containing anesthetic or air control were immediately placed in a 20° temperature-controlled incubator for 4 h. After the 4-h preconditioning incubation, the agar plates with worms were then removed from the glass chambers, and the worms were allowed to recover for a given period before hypoxic, azide, or thermal challenge.
Hypoxic and Azide Killing Incubation
After anesthetic or air control preconditioning, animals were transferred from their NGM plates to a 1.5-ml Eppendorf tube with 1 ml M9 buffer,29the buffer was exchanged three times to remove bacteria, and the final wash was removed down to 100 μl. The tubes were placed into the hypoxia chamber as described previously,14incubated for 20 h at 26°, and then transferred to NGM agar plates seeded with OP50 bacterial to recover in normoxia at 20°. Oxygen tension in the hypoxia chamber was measured by an oxygen meter and electrode (OM4 meter, MI730 electrode; Microelectrodes, Bedford, NH) and was always less than 0.3%. Azide incubations were performed as described previously14except that azide incubations were 70 min in 0.5 m NaN3. Animals were scored 20–24 h after completion of the hypoxic or azide incubations. Animals without pharyngeal pumping and without spontaneous or touched-evoked movement were scored as dead.
After anesthetic or air control preconditioning, young adult animals were transferred from their NGM plates to a 1.5-ml Eppendorf tube with 1 ml M9 buffer. The buffer was exchanged three times to remove bacteria, and the final wash was removed down to 100 μl. The tubes were placed into a 37° circulating water bath and incubated for 1 h. Animals were then transferred to NGM agar plates seeded with OP50 bacterial to recover in normoxia at 20°. Animals were scored 20–24 h later, and those without pharyngeal pumping and spontaneous or touched-evoked movement were scored as dead.
Simultaneously determined experimental and control values were compared for significant differences by paired t test with the significance threshold set at P < 0.01. For determinations of divergence from control or baseline in time course and concentration-response experiments, respectively, all values for a particular time point or concentration were pooled and compared by a Fisher exact chi-square test. The significance threshold was set at P < 0.01 after Bonferroni correction for multiple comparisons. All statistical manipulations and testing were performed using GraphPad Prism Software (GraphPad Software, Inc., San Diego, CA).
We tested the ability of two VAs, isoflurane and halothane, to produce protection of the whole C. elegans animal from death after normally lethal injuries. Preliminary experiments indicated that 2% isoflurane (1.28 mm) and 1% halothane (0.71 mm) followed by a 1-h recovery before exposure to the lethal injury might be efficacious. Indeed, under these conditions, both isoflurane and halothane reproducibly produced an approximately twofold reduction in lethality after a hypoxic injury (fig. 1A). Sodium azide is an inhibitor of cytochrome oxidase, the terminal protein complex in the mitochondrial electron chain, and thereby blocks utilization of oxygen for energy production.35Sodium azide has been used a chemical surrogate for hypoxia.36–40Agents or conditions that alter hypoxic sensitivity in general similarly affect azide sensitivity if the agents are in fact altering the consequences of inhibition of oxidative metabolism. Therefore, we tested the effect of isoflurane and halothane pretreatment on azide-mediated death. As for hypoxia, both anesthetics reduced azide lethality (fig. 1B). Finally, to test whether the protective response was specific for hypoxic injury or was more generalized, we measured the effect of isoflurane preconditioning against hyperthermic injury. Isoflurane did produce a significant reduction in hyperthermic death (fig. 1C). This result argues that anesthetic pretreatment in C. elegans induces a protective response to generalized cellular injury. Notably, hypoxic preconditioning does not protect from hyperthermic death.28Therefore, anesthetic preconditioning induces a protective response that is broader than that by hypoxic preconditioning and argues against metabolic suppression as a mechanism for anesthetic preconditioning.
Preconditioning is by definition a delayed protection after the preconditioning stimulus is no longer being applied. To establish that the protection by the anesthetic pretreatment was indeed preconditioning and not due to a direct protective effect of residual drug, we measured the onset and duration of the protection after isoflurane and halothane. Isoflurane significantly protected from hypoxic injury even with no recovery period after the isoflurane preconditioning before hypoxic killing (fig. 2A). The hypoxia protection significantly increased after a 1-h recovery from isoflurane and then gradually waned, returning to baseline by 5 h. For halothane, the protective effect from hypoxia appeared more transient, only significant at 1 h of recovery (fig. 2B). This determination of the time course for protection is limited by the long hypoxic incubation, 20 h, required to kill the majority of C. elegans .14Therefore, the onset of the protective preconditioning response could be at any time during or even after the 20-h hypoxic killing incubation. To get a more precise time course, we used azide killing, which can be accomplished in 70 min. With azide incubation, the protection by isoflurane and halothane was clearly delayed (figs. 2C and D). The earliest time for statistically significant protection after anesthetic exposure was 6 h for both halothane and isoflurane, and the protection lasted up to 9 h. This delayed time course is indicative of a preconditioning action of the anesthetics and not a direct protective effect.
We next measured the concentration dependence of isoflurane and halothane preconditioning. The concentration-effect curve for both anesthetics was biphasic (figs. 3A and B). For isoflurane, significant preconditioning was observed after 1 vol% isoflurane (0.64 mm) and was maximal at 2 vol%. However, the protection was lost with higher concentrations. Similarly, significant preconditioning by halothane was observed after exposure to 1 vol% halothane (0.71 mm) with disappearance of the effect at 3 vol%. The biphasic nature of the curve suggests that at higher concentrations, both halothane and isoflurane have a second action that directly or indirectly opposes the preconditioning mechanism. For general anesthesia in mammals and for behavioral defects in C. elegans , halothane has an aqueous potency higher than that of isoflurane.33,41,42For preconditioning, the relative potencies of the two drugs are unclear because the minimum potencies and EC50s for isoflurane and halothane preconditioning are not well defined, although isoflurane has been shown to precondition at a lower aqueous concentration than the lowest tested for halothane.43Notably, for both isoflurane and halothane, the minimally effective aqueous concentrations for preconditioning in C. elegans are higher than those required in mammalian cells; mammalian halothane and isoflurane EC50s for preconditioning are similar to those that produce anesthesia (0.1–0.3 mm).6,43These concentrations are also higher than required to produce behavioral effects in C. elegans but are lower than required to immobilize wild-type animals.33,34
An important and unresolved question for anesthetic preconditioning is the relation between the mechanism for preconditioning and the mechanism for anesthesia. In particular, inhibition of neurotransmitter release, a likely contributor to anesthesia produced by VAs,44has also been proposed as a mechanism for VA preconditioning in the nervous system.6In C. elegans , the behavioral effects of human clinical concentrations of isoflurane can be blocked by a specific mutation (md130 ) in the gene (unc-64 ) encoding the presynaptic protein syntaxin.31This syntaxin mutant, unc-64(md130) , blocks both the behavioral effects and the inhibition of neurotransmitter release produced by isoflurane acting at clinical concentrations. Therefore, unc-64(md130) offers a means to ask whether inhibition of transmitter release is a necessary component of the preconditioning mechanism of isoflurane in C. elegans . Testing of the hypoxic sensitivity of unc-64(md130) without preconditioning showed that the strain was significantly hypoxia resistant at baseline compared with parallel wild-type controls (figs. 4A and B). We have observed this with some other strains like unc-64(md130) that have reduced locomotion and transmitter release (data not shown); therefore, this was not a surprising result. After isoflurane preconditioning, unc-64(md130) became significantly more hypoxia resistant compared with baseline, and the relative change in hypoxic sensitivity after preconditioning was similar to that of the wild-type strain N2 (figs. 4B and C). Therefore, we can conclude that in C. elegans , the mechanism of isoflurane preconditioning does not require inhibition of neurotransmitter release.
We have recently found that C. elegans has a mechanism for hypoxic preconditioning.28In testing the role in hypoxic preconditioning of the well-characterized C. elegans apoptosis pathway, we found an interesting and surprising result. A functional apoptosis pathway was not required for hypoxic preconditioning, but one component of the pathway was. The Apaf-1 homolog CED-4, which normally serves to activate the caspase CED-3,17was found to be essential for hypoxic preconditioning. However, proteins both upstream and downstream of CED-4 in the apoptosis pathway, including the caspase CED-3, were not essential for hypoxic preconditioning. Given these results, we asked here whether anesthetic preconditioning and hypoxic preconditioning in C. elegans share a requirement for CED-4 (fig. 5). A loss-of-function ced-4 mutant was significantly resistant at baseline compared with the wild-type strain (figs. 5A and B). This result agrees with our previous report and indicates that apoptosis via CED-4 mediates a portion of the cell death after hypoxia in C. elegans .28After isoflurane preconditioning, ced-4 mutant animals actually exhibited a higher proportion of hypoxic death compared with the air controls (figs. 5B and D). ced-3 mutant animals lacking the CED-3 caspase were hypoxia resistant at baseline like the ced-4 mutant, but unlike the ced-4 mutant they exhibited a normal preconditioning response to isoflurane (figs. 5C and D). Therefore, as for hypoxic preconditioning, isoflurane preconditioning requires CED-4 but not CED-3.
The data reported here have shown that the nematode C. elegans has a mechanism for VA preconditioning. The time course of the protection after anesthetic exposure is not immediate and meets the definition of preconditioning. The preconditioning effect is concentration dependent, with the minimally effective concentration for isoflurane in the high human clinical range. The preconditioning mechanism and the anesthetic mechanism acting at these concentrations in C. elegans are genetically separable; unlike the anesthetic mechanism against C. elegans behavior, the anesthetic preconditioning mechanism does not require inhibition of neurotransmitter release. Finally, apoptosis pathway mutants show that anesthetic preconditioning shares with hypoxic preconditioning a requirement for the Apaf-1 homolog CED-4 but that this mechanism is novel, not acting through the downstream caspase.
The discovery that an invertebrate has a mechanism for isoflurane and halothane preconditioning tremendously increases the phylogenetic breadth of anesthetic preconditioning. Previously, only mammals have been reported to have anesthetic preconditioning. This discovery of anesthetic preconditioning in invertebrates suggests that at least a portion of the mechanisms of anesthetic preconditioning are likely to be ancient and conserved throughout most metazoan phyla. Therefore, vertebrate-specific molecules are not essential or fundamental to preconditioning mechanisms. One such protein is the sarcolemmal KATPchannel. A variety of inhibitors of this channel have suggested its role in both hypoxic and anesthetic preconditioning in cardiac muscle.8However, C. elegans does not have a clear homolog for the sarcolemmal KATPchannel.45Although it is possible that homology searches have not identified an authentic C. elegans homolog, it is more likely that C. elegans does not have such a protein. Of course, one cannot rule out that another potassium channel is regulated by adenosine triphosphate in a manner similar to the mammalian KATPchannel and might serve the same function. One such possibility is the hypothetical mitochondrial KATP, which has been identified only pharmacologically. The amino acid sequence and gene identity are unknown for the mammalian mitochondrial KATP; therefore, C. elegans might well have one. Of course, these data do not rule out a role of the sarcolemmal KATPchannel in mammalian preconditioning; rather, these findings indicate that this channel is not essential for anesthetic preconditioning in all organisms and cell types.
The window of protection produced by anesthetic preconditioning in C. elegans is not typical of either early or late preconditioning. Cytoprotection after early preconditioning has been seen in as little as 15 min after the preconditioning incubation and typically is lost in 4 h.1,2,46Protection after late preconditioning is typically seen at approximately 12 h and lasts at least 24 h after the preconditioning incubation. Hypoxic preconditioning in C. elegans has a time course corresponding well with late preconditioning as defined in other species.28Onset of hypoxic preconditioning in C. elegans is approximately 16 h after preconditioning incubation, and the duration is at least 36 h. In contrast, anesthetic preconditioning in C. elegans begins approximately 6 h after removal of anesthetic and lasts approximately an additional 3 h. Even if one adds the 4-h anesthetic exposure time to the onset and duration (i.e. , 10–13 vs. 6–9 h), the duration of the protective effect is less than typical for hypoxic preconditioning. However, this time course is consistent with induction of new transcription and translation, such as that which has been observed during the delayed window of protection after hypoxic preconditioning.47–49If the protection does require new transcription and translation, the newly synthesized protective molecules must be relatively short-lived compared with hypoxic preconditioning. Alternatively, anesthetics might induce a toxic response that offsets the preconditioning protective response at later time points, and indeed there is some precedent for this.50The hypothesis of a toxic effect of the anesthetics is supported by the fact that in the ced-4 mutant, preexposure to isoflurane actually increases the lethality of hypoxia (figs. 5B and D).
An intriguing finding is the requirement for CED-4, the Apaf-1 homolog, in isoflurane preconditioning. These results agree with those from a more extensive set of experiments examining the role of the apoptosis pathway in hypoxic preconditioning in C. elegans .28Given that CED-4 is required for both isoflurane and hypoxic preconditioning, CED-4 is unlikely to mediate sensing of hypoxia or isoflurane in any direct way. Rather, CED-4 is more likely to be a component of a downstream signal transduction or induced protection mechanism that is common to both anesthetic and hypoxic preconditioning. How might CED-4 transduce or mediate hypoxic and anesthetic preconditioning? Our original hypothesis for hypoxic preconditioning was that hypoxic preconditioning might induce a protective mechanism that acts exclusively or primarily to inhibit apoptotic cell death. If so, mutants that lack essential components of the apoptotic machinery should be both resistant to hypoxia at baseline and incapable of becoming more resistant after hypoxic preconditioning. The data for both hypoxic and anesthetic preconditioning show that although apoptosis is mediating a portion of hypoxic injury in C. elegans as evidenced by the baseline hypoxia resistance, the apoptosis pathway in general is not required for preconditioning. Therefore, the original hypothesis is incorrect. Rather within the apoptosis pathway, CED-4 is uniquely required for hypoxic and anesthetic preconditioning. CED-4 has no known function in the absence of the other components of the apoptosis pathway. Therefore, we have difficulty speculating as to its role in preconditioning. One reasonable hypothesis is that CED-4 activity is induced or in some way increased after hypoxic or anesthetic stress and that this increase in activity of an apoptosis protein serves as a panic signal that induces a counterprotective response. Motivated by this hypothesis, future experiments will attempt to identify genes regulated in a CED-4-dependent manner after hypoxic and anesthetic preconditioning. Such genes, if evolutionarily conserved, could lead to a better understanding of the preconditioning mechanism in higher organisms and potentially to therapies for hypoxia-mediated disease states.