Hypothermic Preconditioning Increases Survival of Purkinje Neurons in Rat Cerebellar Slices after an In Vitro  Simulated Ischemia

The animal protocol was approved by the Institutional Animal Care and Use Committee at the University of Virginia. All animal experiments were carried out in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals  (NIH Publication No. 80–23) revised in 1996. All efforts were made to minimize the number of animals used and their suffering.

Unless specified below, all reagents were obtained from Sigma (St. Louis, MO). Glibenclamide, a selective K^ATPchannel blocker, and 8-cyclopentyl-1,3-dipropylxanthine, a selective adenosine A¹receptor antagonist, were first dissolved in ethanol and then diluted into the required concentrations with artificial cerebrospinal fluid (aCSF). The final ethanol concentrations in the aCSF used for incubation of cerebellar slices were less than 0.05%. A pilot study showed that 0.05% ethanol did not affect Purkinje cell survival compared with control slices in aCSF only under current experimental protocols. Farnesyl protein transferase inhibitor III (a potent and selective inhibitor of farnesyl transferase, which thus inhibits Ras farnesylation and processing) was aCSF soluble.

Similar to the methods reported previously, 31–33cerebellar brain slices were prepared from 2- to 3-month-old, 200- to 250-g, male Sprague-Dawley rats (Hilltop, Scotdale, PA). Rats were anesthetized with halothane and then decapitated. The cerebellum was removed rapidly and placed in ice-cold aCSF bubbled with 5% CO²and 95% O². The aCSF contained 116 mm NaCl, 26.2 mm NaHCO³, 5.4 mm KCl, 1.8 mm CaCl², 0.9 mm MgCl², 0.9 mm NaH²PO⁴, and 5.6 mm glucose, pH 7.4. The cerebellum was immediately hemisected and sectioned with a tissue slicer into 400-μm transverse slices in ice-cold cutting solution (260 mm sucrose, 26.2 mm NaHCO³, 3 mm KCl, 1.2 mm NaH²PO⁴, 5 mm MgCl², and 9 mm glucose, pH 7.4) bubbled with 5% CO²and 95% O². After sectioning, slices were wrapped in filter paper and placed into a tissue holder (made of plastic, with small holes in it to allow free diffusion of gases and water; this holder also helps to avoid direct gas bubbling on slices). These slices were then immersed in circulating aCSF continuously bubbled with 5% CO²and 95% O²(oxygenated aCSF) at 37°C for at least 1 h for recovery of the synaptic function. 31–33 

Ischemia was simulated in vitro  by oxygen-glucose deprivation (OGD). As described previously, 31,33cerebellar slices were transferred into a glass beaker containing glucose-free aCSF (also containing 1 mm dithionite, an oxygen absorbent) bubbled with 95% N²and 5% CO²(30-min bubbling before the placement of brain slices was allowed so as to reduce the oxygen content in the solution). Under these conditions, the Po²in the aCSF was lower than 0.1 mmHg as measured by a Clark oxygen electrode (Cameron Instrument Co., Port Aransas, TX). The beaker containing the slices was immersed in a water bath to keep the temperature of glucose-free aCSF in the beaker at 37°± 0.2°C as monitored by a thermometer. After 20 min of OGD, slices were recovered in circulating oxygenated aCSF at 37°C for 4 h to allow cell injury and death that may not be evident immediately after the OGD episode to become apparent.

Hypothermic preconditioning was performed by transferring cerebellar slices into a glass beaker containing oxygenated aCSF at 33°C. After 20 min of hypothermia, slices were placed in circulating oxygenated aCSF at 37°C for the indicated duration before they were subjected to the OGD insult.

In the first set of experiments as illustrated in figure 1, hypothermic preconditioning (33°C for 20 min) was performed immediately (0HI), 1 h (1HI), 2 h (2HI), or 3 h (3HI) before the OGD at 37°C for 20 min. Control experiments with the hypothermia applied at the corresponding time points but without the OGD were performed to evaluate the effects of hypothermia on cell survival (0HC, 1HC, 2HC, and 3HC groups). A time control experiment (C group) was also performed by keeping the slices in circulating oxygenated aCSF at 37°C for the whole experimental duration. Finally, the OGD-alone group (I group) was tested by subjecting the slices to the OGD at the time point corresponding to that in the other groups with the OGD insult.

In the second set of experiments, cerebellar slices were preconditioned with hypothermia (33°C for 20 min) 1 h before the OGD at 37°C for 20 min. In some experiments, 8-cyclopentyl-1,3-dipropylxanthine (100 nm), glibenclamide (0.3 μm), or farnesyl protein transferase inhibitor III (0.25 mm) was present during the period of hypothermia. Control experiments were performed by incubating the slices with these three drugs at the corresponding time point, and these slices were not exposed to the hypothermia or to both the hypothermia and the OGD. Similarly, time control and OGD-alone experiments were performed as described in the first set of experiments. Brain slices from the first and second sets of experiments were processed for morphologic evaluation of Purkinje cell survival.

In the third set of experiments, cerebellar slices were subjected to the hypothermia alone, OGD alone, hypothermia plus OGD, and time control with the same experimental protocols as described in the second set of experiments. These slices were used for Western blotting to detect the expression of HMG I(Y) proteins.

After incubation and recovery, cerebellar slices were fixed in 4% paraformaldehyde in phosphate-buffered saline (pH 7.4) overnight at 4°C. The slices then were paraffin embedded and sectioned into 5-μm thick sections. The sections were cut from an interior region of the brain slices (∼ 120 μm from the edge) to avoid areas that may have slicing trauma during slice preparation. Morphologic examination was performed under light microscopy after sections were stained with hematoxylin and eosin. The sections were examined by an observer blinded to the group assignment to determine the percentage of morphologically normal Purkinje cells (intact rate) and remaining (morphologically normal and injured) Purkinje cells (survival rate) in total counted (morphologically normal, injured, and dead) Purkinje cells. Purkinje cells were recognized based on the morphology and location (between molecular and granular layers, fig. 2). Injured neurons were identified if the cells had one of the following presentations: cell swelling, vacuolization, or presence of shrunken and darkened nuclei. 34Dead neurons were recognized by cell fragmentations. All Purkinje neurons present in one section were counted, and at least two sections were read for each experimental condition (total of 200–400 Purkinje cells) to compute the percentages.

Cerebellar slices were homogenized in 25 mm Tris-HCl, pH 7.4, containing 1 mm EDTA, 1 mm EGTA, 0.1% (vol/vol) 2-mercaptoethanol, 1 mm phenylmethylsulfonyl fluoride, 2 μm leupeptin, and 1 μm pepstatin A. The crude homogenates were centrifuged at 1,000 g  for 10 min at 4°C. Protein content in the supernatants was quantitated by the Lowry assay using a protein assay kit from Sigma. About 100 μg per lane of proteins was used for Western blotting as described previously. 35Goat anti-HMG I(Y) polyclonal antibody was raised against a 19–amino acid peptide at the N-terminus of human HMG I(Y). This antibody cross-reacts with rat HMG I(Y) and is from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA). Rabbit anti-β-actin polyclonal antibody was an affinity-purified antibody raised against the C-terminal actin fragment (11 amino acids) attached to multiple antigen peptide backbone and was acquired from Sigma. The volume of the protein bands was quantitated by an ImageQuant 5.0 Molecular Dynamics Densitometer (Molecular Dynamics, Sunnyvale, CA). The HMG I(Y) results were then normalized to those of β-actin to control for errors in protein sample loading and transferring during the Western blotting.

Each experimental condition was repeated eight times with cerebellar slices from different rats. Statistical analysis for the cell survival/intact rates was performed by Kruskal–Wallis ANOVA for unpaired nonparametric data, followed by the Mann–Whitney rank sum test. Statistical analysis for HMG I(Y) expression data was performed by unpaired t  test. A P < 0.05 was accepted as significant.

In vitro  simulated ischemia (OGD) for 20 min followed by a 4-h recovery period reduced the percentages of remaining and morphologically normal Purkinje neurons in the cerebellar slices (figs. 2 and 3). This OGD-induced neuronal injury/death was abolished by a 20-min hypothermia (33°C) applied before the OGD (figures 2 and 3). The hypothermic preconditioning was similarly effective to prevent OGD-induced Purkinje cell injury/death when the hypothermia was applied from 3 h to immediately before the OGD (figure 3). Hypothermia alone did not alter neuronal injury/death rates under control conditions (fig. 3). These findings indicated that hypothermic preconditioning induced an acute phase of neuroprotection that developed immediately and persisted for at least 3 h.

In the mechanistic studies, we chose to apply the hypothermic stimulus 1 h before the OGD. Consistent with the results presented previously, this preconditioning protocol also abolished OGD-induced Purkinje cell injury/death in this set of experiments (fig. 4). However, this hypothermic preconditioning–induced neuroprotection was abolished by 8-cyclopentyl-1,3-dipropylxanthine, glibenclamide, or farnesyl protein transferase inhibitor III, whereas these three drugs did not affect the neuronal injury/death rates under control and OGD-only conditions (fig. 4). These results suggest that adenosine A¹receptors, K^ATPchannels, and Ras play a role in hypothermic preconditioning–induced neuroprotection.

A protein band with a mobility consistent with a molecular weight of ∼ 25 kDa on sodium dodecyl sulfate–polyacrylamide gel electrophoresis was detected by Western blot analysis with an anti-HMG I(Y) antibody. Brain slices that had a 20-min OGD expressed a significantly higher level of HMG I(Y) proteins than controls. However, brain slices with hypothermic preconditioning and the OGD had a similar level of HMG I(Y) proteins to that in controls (fig. 5).

The delayed phase of hypothermic preconditioning–induced neuroprotection has been investigated in two recent studies from the same group of investigators. 18,36They gave adult rats a 20-min hypothermia (28.5°C or 31.5°C; the actual hypothermic period is about 80 min, because 30 min each was needed to cool and warm the animals to/from the target temperatures under halothane anesthesia) before brain ischemia. The ischemia was induced by clamping one middle cerebral artery and both carotid arteries for 1 h. The infarct size was measured 24 h after the ischemia. Animals with hypothermic preconditioning at 6, 24, and 48 h but not 7 days before the ischemia had significantly smaller cerebral infarct sizes. 18In our study, we used a brief (20 min) and milder hypothermia (33°C) than that used in the previous studies and investigated the acute phase (0–3 h) of hypothermic preconditioning–induced neuroprotection. These features in our study were designed to provide more relevant information for situations during surgery, because a short episode of hypothermia at 33°C is usually well tolerated by most patients and easily applied under anesthesia with minimal untoward effects on many organ systems, including the cardiovascular system. The time interval (0–3 h) between the preconditioning stimulus and an ischemic insult is also suitable for application in many clinical situations such as during carotid endarterectomy. In addition, the acute phase of ischemic preconditioning–induced cardioprotection has been shown to last for only 2 to 3 h. 3,5,37Our results demonstrated that a 20-min hypothermia applied from 3 h to immediately before the OGD was equally effective in protecting Purkinje neurons from OGD-induced injury/death. These data suggest that inducing hypothermia even just before an anticipated ischemia will precondition neurons against ischemia.

Our findings may have important implications. Anticipated episodes of brain ischemia during surgery, especially cerebrovascular surgery, are common. Effective, practical, and safe methods to induce ischemic tolerance are needed to minimize ischemic brain injury in those patients. Our data suggest that hypothermic preconditioning may be used for this purpose. Hypothermia is usually noninvasive and has a substantial safety margin with respect to neuronal or cardiac injury. This is especially true when only mild hypothermia is needed for inducing neuroprotection as demonstrated in our study. Mild hypothermia has fewer and less severe side effects than the moderate to severe hypothermia used in the previous studies by Nishio et al.  18and Yunoki et al.  36and is easy to achieve during surgery. Taken together, using hypothermia as a preconditioning stimulus for inducing neuroprotection is worth further evaluation for potential clinical use.

Our study has limitations. We used brain tissues from rats. It is not appropriate to extrapolate these data to human patients. It is also not appropriate to extrapolate our in vitro  results to in vivo  conditions. In addition, we evaluated Purkinje cell injury/death 4 h after the OGD, because neuronal death in brain slices is manifest within 4 to 5 h after OGD, hypoxia, or overstimulation of glutamate receptors. 31–33,38However, delayed cell death that can occur even 14 days after brain ischemia has been reported in rats. 39–41Thus, our study may have underestimated the neuronal death induced by OGD, and it is not known whether hypothermic preconditioning–induced neuroprotection observed at 4 h after OGD will persist at later time points. Interestingly, a recent study has shown that the effects of the hypothermic preconditioning–induced delayed phase of neuroprotection were increased with the increase in the duration (from 20 to 180 min tested in the study) of the hypothermia (33°C) or in the depth (from 31.5° to 25.5°C tested in the study) of hypothermia. 36Further studies are needed to determine whether this association exists in the acute phase of hypothermic preconditioning–induced neuroprotection. Lastly, all slices (including control slices) in our study had a brief episode of hypothermia during the preparation of slices. Thus, our results suggest that preconditioning with two episodes of hypothermia may make Purkinje cells more resistant to OGD than that with only one episode of hypothermia. Nevertheless, our study still revealed an interesting phenomenon whose mechanistic studies may suggest new interventions for ischemic brain injury. In this regard, our in vitro  model does provide some advantages over in vivo  models. For example, it is easier to apply pharmacologic manipulations for mechanistic studies in in vitro  models than in in vivo  models. Temperature changes of the brain slices can be quickly achieved by transferring them to a beaker with the desired temperature. Moreover, our in vitro  model also eliminates the interference of systemic variables such as blood pressure and cerebral blood flow existing in an in vivo  model. Consequently, our results also suggest that hypothermic preconditioning–induced neuroprotection was caused, at least in part, by cellular elements intrinsic to the neuropil.

What are the mechanisms for hypothermic preconditioning–induced neuroprotection? Nishio et al.  18suggested that protein synthesis is required for the development of the hypothermic preconditioning–induced delayed phase of neuroprotection. Similar mechanisms were proposed for the development of the delayed phase of ischemic preconditioning–induced neuroprotection and cardioprotection. 6,7,19,37Multiple mediators, including adenosine receptors and protein kinase C in the heart and Ras in the brain, have been proposed to be involved in inducing protein synthesis after ischemic preconditioning. 19,21,42–45However, the acute phase of ischemic preconditioning–induced neuroprotection and cardioprotection may not require protein synthesis because of the rapid onset of protection (within minutes after the preconditioning stimuli). It is generally accepted that activation of cell signaling molecules such as adenosine receptors, protein kinase C, and K^ATPchannels is important for the development of the acute phase of ischemic or anesthetic preconditioning–induced cardioprotection. 19,20,22–24Similarly, our results suggest that adenosine A¹receptors and K^ATPchannels play an important role in the acute phase of hypothermic preconditioning–induced neuroprotection. In addition, our results suggest that Ras activation may be necessary in this neuroprotection (the implication of adenosine A¹receptors, K^ATPchannels, and Ras in the hypothermic preconditioning–induced neuroprotection was based on the results using one inhibitor for each of these signaling molecules, and the inhibitory effects of these inhibitors on their respective target molecules were not confirmed biochemically or electrophysiologically in our study). A previous study showed that N -methyl-d-aspartate receptor–initiated and Ras-dependent activation of mitogen-activated protein kinases was critical in the development of the delayed phase of OGD preconditioning–induced neuroprotection. 21In a recent study from our laboratory, we showed that Ras was also important for the acute phase of OGD preconditioning–induced neuroprotection. 31However, Ras and K^ATPchannel activation was not demonstrated to be involved in isoflurane preconditioning–induced neuroprotection. 31Thus, our results presented here suggest that hypothermic preconditioning–induced neuroprotection involves signaling molecules that are important for ischemic preconditioning–induced neuroprotection and are different from those for isoflurane preconditioning–induced neuroprotection. The exact relation between adenosine receptors, K^ATPchannels, and Ras in cellular signaling pathways to mediate hypothermic preconditioning–induced neuroprotection is not known from our study. However, a potential link between adenosine receptors, K^ATPchannels, and Ras may be proposed, because it has been reported that activation of adenosine receptors can lead to opening of K^ATPchannels, which can induce production of oxygen free radicals to activate Ras. 20,23,24,46,47 

The development of brain injury after ischemia involves the synthesis of multiple putatively damaging proteins such as inducible nitric oxide synthase, cyclooxygenase-2, and cytokines. 48Thus, inhibition of the synthesis of these proteins may reduce ischemic brain injury. HMG I(Y), an important nuclear transcription protein, has been demonstrated to enhance the expression of inducible nitric oxide synthase, cyclooxygenase-2, and cytokines. 25–28Therefore, the expression of HMG I(Y) proteins may be increased after ischemia. This phenomenon was demonstrated in a previous study. #Our current study confirmed this finding and further showed that hypothermic preconditioning attenuated the OGD-induced HMG I(Y) increase. Thus, inhibition of HMG I(Y) expression and, therefore, a decrease in the expression of putatively damaging proteins after ischemia is an attractive mechanism for hypothermic preconditioning–induced neuroprotection. However, it is not known from our current study whether hypothermic preconditioning–induced inhibition of HMG I(Y) protein expression plays a role in the development of the acute phase or delayed phase or both phases of hypothermic preconditioning–induced neuroprotection.

In conclusion, we showed that hypothermic preconditioning induced an acute phase of neuroprotection in rat cerebellar slices. Activation of adenosine A¹receptors, K^ATPchannels, and Ras may be involved in this neuroprotection. In addition, our data suggest that hypothermic preconditioning attenuates OGD-induced increase in HMG I(Y) expression, a potential mechanism for hypothermic preconditioning–induced neuroprotection.