Isoflurane Preconditions Hippocampal Neurons against Oxygen–Glucose Deprivation: Role of Intracellular Ca2+and Mitogen-activated Protein Kinase Signaling

Figure 1shows the basic study design. Slice cultures of hippocampus were exposed to gas-phase 0.5 or 1.5% isoflurane (isoflurane in air plus 5% CO²) for 2 h and returned to standard culture conditions for the next 24 h. At that time, the cultures were exposed to combined oxygen and glucose deprivation for 1 h (simulated ischemia). The percentage of dead and living neurons remaining in CA1, CA3, and dentate regions of the slices were determined serially at 2 and 3 days after the simulated ischemia. Measurements of [Ca²⁺]ⁱin CA1 neurons within the cultures were made in separate groups of slices during simulated preconditioning conditions. Other cultures were processed as above and frozen at the times indicated in figure 1for determination of expression levels of MAP kinases p42/44, the antiapoptotic protein Akt, and the Akt substrate protein GSK-3β.

All studies were approved by the University of California at San Francisco (UCSF) Committee on Animal Research and conform to relevant National Institutes of Health guidelines.

Organotypic cultures of the hippocampus were prepared by standard methods21,22as modified by Sullivan et al.  23Briefly, Sprague-Dawley rats (aged 8–12 days; Simonsen Laboratories, Gilroy, CA) were anesthetized with 1–2% halothane and given an intraperitoneal injection of ketamine (10 mg/kg) and diazepam (0.2 mg/kg). The rats were decapitated, and the hippocampi were removed and placed in 4°C Gey’s Balanced Salt Solution (UCSF Cell Culture Facility). Next, the hippocampi were transversely sliced (400 μm thick) with a tissue slicer (Siskiyou Design Instruments, Grants Pass, OR), and stored in Gey’s Balanced Salt Solution containing 0.038 mg/ml ketamine at 4°C for 1 h.24The slices were then transferred onto 30-mm-diameter membrane inserts (Millicell-CM; Millipore, Bedford, MA), and put into six-well culture trays with 1.5 ml slice culture medium per well. The slice culture medium consisted of 50% Minimal Essential Medium (Eagles with Earle’s balanced salt solution; UCSF Cell Culture Facility), 25% Earles balanced salt solution (UCSF Cell Culture Facility), 25% heat-inactivated horse serum (Hyclone Laboratories, South San Francisco, CA) with 6.5 mg/ml glucose, and 5 mm KCl. Slices were kept in culture for 7–14 days before study.

In vitro  ischemia was simulated by anoxia combined with glucose-free media (OGD). Before hypoxia, the slices were washed three times with glucose-free Hank’s balanced salt solution. The cultures were then placed into a 2-l airtight Billups-Rothenberg Modular Incubator chamber (Del Mar, CA) through which 95% N²–5% CO²gas, preheated to 37°C, was passed at 5–10 l/min. The temperature of the chamber was kept at 37°C both by passing preheated gas through the chamber and by placing a heat lamp over the chamber. The temperature inside the chamber was monitored with a thermocouple thermometer. After 10 min of gas flow, the chamber was sealed and placed in a 37°C incubator. The partial pressure of oxygen was approximately 0–0.2 mmHg, measured with a Clark-type oxygen electrode. After the injury, the culture tray was removed from the chamber, the anoxic glucose-free Hank’s balanced salt solution was aspirated from the wells, and standard (oxygenated) slice culture media was added.

Cell viability was assessed with propidium iodide (PI) fluorescence (Molecular Probes, Eugene, OR). PI, a highly polar fluorescent dye, penetrates damaged plasma membranes and binds to DNA. Before imaging, slice culture media containing 2.3 μm PI was added to the wells of the culture trays. After 15 min, the slices were examined with a Nikon Diaphot 200 inverted microscope (Nikon Corporation, Tokyo, Japan), and fluorescent digital images were taken using a SPOT Jr. Digital Camera (Diagnostic Instruments Inc., Sterling Heights, MI). Excitation light wavelength was 490 nm, and emission was 590 nm. The sensitivity of the camera and intensity of the excitation light was standardized so as to be identical from day to day. PI fluorescence was measured in the dentate gyrus, CA1, and CA3 regions of the hippocampal slices. Slices were discarded if they showed more than slight PI fluorescence in these regions after 7–10 days in culture. Slices were imaged before OGD (signal assumed to represent 0% cell death) and at 2 and 3 days after OGD. In previous studies, we found that maximum post-OGD death consistently occurs at approximately day 2 or 3 and declines over the next 11 days.23Serial measurements of PI fluorescence intensity were made in predefined areas (manually outlining CA1, CA3, and dentate separately) for each slice using NIH Image-J software (National Institutes of Health, Bethesda, MD). Thus, cell death was followed in the same regions of each slice after simulated ischemia. After the measurement of PI fluorescence on the third post-OGD day, all of the neurons in the slice were killed to produce a fluorescence signal equal to 100% neuron death in the regions of interest. This was done by adding 100 μm potassium cyanide and 2 mm sodium iodoacetate to the cultures for at least 20 min. Twelve to 24 h later, final images of PI fluorescence (equated to 100% cell death) were acquired. Percentage of dead cells at 0, 2, and 3 days after OGD were then calculated based on these values. A linear relation exists between cell death and PI fluorescence intensity.22,24 

In separate groups of slices, [Ca²⁺]ⁱwas measured before, during, and after 1% isoflurane exposure. Estimates of [Ca²⁺]ⁱin CA1 neurons in slice cultures were made using the indicator fura-2/AM and a dual excitation fluorescence spectrometer (Photon Technology International, South Brunswick, NJ) coupled to a Nikon Diaphot inverted microscope. Slice cultures were incubated with 5–10 μm of the indicator dye for 15–30 min before measurements. Cultures for these measurements were grown on Nunc Anopore (Nalge Nunc, Rochester, NY) culture tray inserts because of their low autofluorescence at fura-2 excitation wavelengths. Slit apertures in the emission light path were adjusted to restrict measurement of light signals to those coming from the CA1 cell body region. Calibration of [Ca²⁺]ⁱwas done by using the dissociation constant of fura-2 determined in vitro  with a Ca²⁺buffer calibration kit (Molecular Probes). The calibration process involved using the same light source, optical path, and filters as used with the slice culture measurements. The dissociation constant for fura-2 was 311 nm, similar to published values.25Background fluorescence (i.e. , fluorescence in the absence of fura) was subtracted from total fluorescence signals before calculation of [Ca²⁺]ⁱas described previously.26Estimates of [Ca²⁺]ⁱwith this technique are accurate to approximately ±10 nm.

Western blots of proteins from culture homogenates were performed with standard methods. Five to eight slices were pooled for each assay, and each study was repeated three or four times. Samples were obtained at the times indicated in figure 1. Protein content in each sample was measured (Bradford protein assay with Coomassie blue) and adjusted so that equal amounts of protein were applied to each lane. Protein bands were visualized after incubation with biotinylated secondary antibodies followed by an enhanced chemiluminescence assay. The intensity of immunostaining was analyzed by scanning the photographic images and using image analysis software (NIH Image) to quantify the staining intensity. In situ  immunostaining of the activated forms of Akt and MAP kinase ERK (p42/44) were done in hippocampal cultures fixed with 4% chilled paraformaldehyde. Antibodies to p-Akt (Ser 473 phosphorylation) and those to MAP kinase p42/44 (Thr 202/Tyr 204 phosphorylation) were obtained from Cell Signaling Technology (Beverly, MA). As with the Western blots, appropriate biotinylated secondary antibodies were used to stain the proteins of interest. Relative protein levels in the in situ  preparations were measured with a microscope, a digital camera, and NIH Image software. This procedure involved outlining regions of interest in each slice and determining the mean gray value within that region as an index of staining intensity.

The percentage survival of neurons in the different regions of the slices is generally not normally distributed. Therefore, the Kruskal–Wallis test followed by the Mann–Whitney U test (JMP; SAS Institute, Cary, NC) was used to compare the means of different treatment groups. T  tests or analyses of variance were used to compare other group means, and allowance was made for multiple comparisons. Differences were considered significant for P < 0.05.

Both a relatively low (0.5%) and relatively high (1.5%) clinical concentration of isoflurane preconditioned neurons in hippocampal slice cultures. Results with 0.5% isoflurane are shown in figure 2, and those for 1.5% isoflurane are shown in figure 3. The degree of preconditioning varied with isoflurane concentration and cell region. With 0.5% isoflurane, protection was statistically significant in CA3 and dentate but not CA1 (P = 0.10). In the slices preconditioned with 1.5% isoflurane, CA1 and dentate neurons were protected, but those in CA 3 were not. The magnitude of the reduction in cell death achieved with preconditioning was approximately 15–40%, which is modest compared with the 70–80% reduction in cell death achieved when isoflurane is present at the same time as the OGD injury.23 

In slice cultures grown on special culture inserts, we measured [Ca²⁺]ⁱin CA1 neurons during application of perfusion solution equilibrated with 1% isoflurane. In this system, approximately 30% of the isoflurane is lost between the reservoir bottle and the microscope stage slice-holding chamber, as measured with a gas chromatograph in the laboratory of Edmund Eger, M.D. (Professor, Department of Anesthesia and Perioperative Care, UCSF). Therefore, the isoflurane concentration reaching the CA1 neurons was approximately 0.7%, between the two concentrations examined in the survival studies. In each of six slices examined, [Ca²⁺]ⁱin CA1 neurons increased during 10 min of isoflurane application and returned to baseline during washout. In figure 4, averaged traces of [Ca²⁺]ⁱare shown. The increase in [Ca²⁺]ⁱafter 10 min of isoflurane was statistically significant (P < 0.001). In several slices, isoflurane perfusion was continued for up to 1 h. Similar increases in [Ca²⁺]ⁱas those shown in figure 4were observed, as was recovery toward baseline (data not shown). The increases in [Ca²⁺]ⁱproduced by isoflurane were almost completely prevented in slices pretreated with 10 nm xestospongin C, a highly specific inositol triphosphate (IP³) receptor antagonist. Anesthetic-induced IP³receptor–dependent release of Ca²⁺from the endoplasmic reticulum is consistent with previous reports.27 

Figure 5shows that isoflurane preconditioning is associated with significant increases in immunostaining of p42/44 in whole slice cultures and in the CA1 region as well. The labels next to the Western blots refer to the compounds present during preconditioning (isoflurane or U0126) and whether an injury of OGD followed 24 h later. Samples taken at t = 0 were obtained just before the OGD. In Western blots, the largest increases in immunostaining in preconditioned slices occurred in the p44 band (isoflurane and Iso/OGD groups). A mitogen-activated extracellular kinase (MEK) inhibitor greatly reduced the phosphorylation of p42/44 during preconditioning (those slices treated with U0126). In slices treated with the MEK inhibitor, p42/44 levels detected with Western blots returned toward baseline after 24 h, although in situ  immunostaining indicated a sustained increase in p42/44 staining 24 h after preconditioning, perhaps reflecting regional rather than global expression of this phosphoprotein in the slice.

Figure 6shows the impact of the calmodulin inhibitor calmidazolium and the MEK inhibitor U0126 on isoflurane preconditioning. Slices were preconditioned with 1.5% isoflurane either with or without the inhibitors present. Preconditioning was not seen in slices exposed to either of these agents. It is unlikely that toxicity of these inhibitors were responsible for these findings because they did not increase cell death in slices subjected to OGD or in slices exposed only to the inhibitors and not to OGD.

Preconditioning was associated with increased phosphorylation of the antiapoptotic protein Akt (fig. 7). This increase in immunostaining was not altered by the MEK inhibitor U0126. Akt phosphorylation did not persist 24 h after preconditioning. Similarly, increased phosphorylation of the Akt-activated protein GSK-3β was also seen after preconditioning, but this increase was not seen 24 h later. The role of Akt in preconditioning neuroprotection was evaluated in slice cultures treated with the phosphatidylinositol-3 kinase inhibitor LY294002, which is an effective means of decreasing Akt phosphorylation in slice cultures (Jonathan Gray, Department of Anesthesia, UCSF; unpublished observations). The presence of LY294002 during preconditioning prevented a reduction in cell death observed 48 h after the injury (fig. 8).

We found that both 0.5 and 1.5% isoflurane preconditions neurons in hippocampal slice cultures by mechanisms that seem to involve intracellular Ca²⁺and the MAP kinase–ERK pathway. Preconditioning was also associated with increased levels of the antiapoptotic protein kinase Akt (protein kinase B). Although APC was statistically significant in several cell regions and with both 0.5 and 1.5% isoflurane, the magnitude of the protection was modest, with reductions in cell death of approximately 15–40%. In similar studies with isoflurane present at the same time as OGD, we found reductions in cell death of 65–90%.23,28The greater potency of isoflurane in protecting neurons when it is present at the time of the simulated ischemia may reflect the direct effects of isoflurane on excitatory ion channels and on excitatory neurotransmitter transporters29,30as opposed to those involved in preconditioning. The reason for the differences in the capacity for isoflurane to precondition neurons in different regions of the hippocampus was not evident. We can speculate that although 0.5% isoflurane may be capable of preconditioning dentate neurons, a higher concentration (or a longer preconditioning duration) is required to precondition CA1 neurons. Regional differences in the expression of the signaling molecules involved in preconditioning, regional differences in calcium homeostasis, or regional differences in sensitivity to hypoxic injury are among the possible factors, but none of these were examined in this investigation.

The proposal that increases in [Ca²⁺]ⁱare a necessary component of APC in the brain is consistent both with what is known of other forms of ischemic preconditioning and tolerance and also with previously proposed mechanisms for APC in the brain and heart. An essential role for Ca²⁺in preconditioning is supported by our observations that APC is associated with moderate and reversible increases in [Ca²⁺]ⁱin CA1 neurons and by the observation that calmidazolium, an inhibitor of calmodulin, prevented preconditioning. Calmodulin is a key Ca²⁺-binding/sensing protein that is involved in a myriad of Ca²⁺-dependent processes. It is specifically tied to both the activation of the MAP kinase–ERK pathway in preconditioning in the brain10,17and to activation of Ca²⁺-calmodulin dependent protein kinase II activation in heart preconditioning.31Calmodulin also activates nitric oxide synthase. The nitric oxide system seems to be important in preconditioning both with N -methyl-d-aspartate receptor activation15and in APC with isoflurane and sevoflurane.4,5Finally, moderate increases in [Ca²⁺]ⁱ, such as those observed in the presence of isoflurane, could plausibly be linked to increased phosphorylation of the antiapoptotic protein Akt, because nitric oxide is an activator of PI³kinase, a major regulator of Akt. Akt seems to play a role in isoflurane preconditioning because LY294002, a compound that prevents the phosphorylation of Akt by PI³kinase, eliminated isoflurane preconditioning neuroprotection in CA1 neurons (fig. 8).

This study examined only several possible signaling mechanisms during preconditioning that are related to moderate increases in [Ca²⁺]ⁱ. Among the interesting possibilities that remain to be investigated are that changes in [Ca²⁺]ⁱare linked via  a calmodulin-dependent protein kinase (e.g. , Ca²⁺-calmodulin dependent protein kinase II) to the regulation of other protein kinases and phosphatases that modulate inhibitory and excitatory ion channels and neurotransmitter receptors.32In the context of brain ischemia, such actions may be important in limiting neurotoxicity from glutamate and other excitotoxins. It is also plausible that this type of signaling process is involved in the mechanism of anesthetic action of isoflurane.

Although APC was associated with a substantial immediate increase in MAP kinase p42/44 immunostaining, some increased staining was still evident 24 h later. A similar but more dramatic phenomenon was observed by Zheng and Zuo6with respect to APC-associated upregulation of MAP kinase p38 in intact rodents. The phosphorylated form of that protein was increased for at least 14 days after preconditioning with 2% isoflurane. In the study of Zheng and Zuo, increased expression of p42/44 (ERK1/2) was not observed, in contrast to our finding here and in studies of acute isoflurane neuroprotection in hippocampal slice cultures.20Sustained activation of MAP kinase signaling pathways is also interesting in view of the fact that sustained activation of MAP kinase pathways is associated with neuronal injury, not protection.33Transient up-regulation of MAP kinases, such as that observed after preconditioning (fig. 5), is more clearly linked to neuroprotective signal transduction.34As stressed by Dirnagl et al. ,2most, if not all, preconditioning stimuli are toxic in high doses or when they are sustained for long periods.

The release of Ca²⁺from intracellular stores (i.e. , the endoplasmic reticulum) is likely the source of Ca²⁺involved in APC. Isoflurane increases [Ca²⁺]ⁱin cortical brain slices, hippocampal brain slices, and isolated cortical and hippocampal neurons.12This process does not require extracellular Ca²⁺but is inhibited by dantrolene and azumolene, both ryanodine receptor antagonists.12The IP³receptor antagonist xestospongin C also antagonizes this release (fig. 4). The ryanodine and IP³receptor complex is the major avenue for intracellular signal–mediated Ca²⁺release in neurons.35It is not known whether anesthetics such as isoflurane act directly on this complex to release Ca²⁺or if other messengers such as IP³or reduced nicotinamide dinucleotide are required.

The slice culture model used in this study has significant advantages over other in vitro  models, including preserved synaptic structure, neuron–glia relations, and the possibility of measuring survival and [Ca²⁺]ⁱin different cell types. Like all in vitro  models, it has limitations that may restrict extrapolation to intact animals. One significant limitation is that in vitro  models, even slice cultures,23do not replicate the temporal loss of anesthetic neuroprotection observed in intact rodents.36It is not yet clear whether APC provides durable neuroprotection in either in vivo  or in vitro  models. Another possible limitation of APC is that, at least in our in vitro  model, it seemed weakly neuroprotective compared with conditions when isoflurane is present during the simulated ischemia. Further studies in both in vitro  and in vivo  models are needed to clarify the durability and power of APC.

The authors thank Jen Schuyler, B.S. (Technician, Department of Anesthesia, University of California at San Francisco, San Francisco, California), for technical assistance.