The Antibiotic Erythromycin Induces Tolerance against Transient Global Cerebral Ischemia in Rats (Pharmacologic Preconditioning)

The experimental protocol was approved by the local Animal Care Committee (Bezirksregierung Rheinhessen-Pfalz, Neustadt an der Weinstrasse, Germany). Male Wistar rats (270–370 g; Charles River, Kisslegg, Germany) were injected with 25 mg/kg erythromycin lactobionate13(a soluble salt of the macrolide antibiotic erythromycin suitable for intravenous and intramuscular application, Erythrocin; Abbott GmbH & Co. KG, Wiesbaden, Germany; estimated half life in rats approximately 2 h),16,17or with vehicle (normal saline, 0.9%) intramuscularly. One cohort of animals served to test the hypothesis that erythromycin pretreatment is able to induce neuronal tolerance against the effects of global cerebral ischemia. A second cohort was studied in parallel to assess the effects of erythromycin on cerebral bcl-2 mRNA and protein expression.

Transient global cerebral ischemia was induced 6, 12, or 24 h after pretreatment with erythromycin or vehicle. After a power analysis, 56 animals were randomized to the respective groups, with n = 12 for the 6-h and n = 8 for the 12- and 24-h pretreatment paradigms (vehicle and erythromycin, respectively). Four animals (vehicle pretreated, n = 3; erythromycin pretreated, n = 1) had to be excluded during the recovery period, because of “no reflow” (n = 1), no electrophysiologic recovery (n = 1), or premature death during the first 24 h after the injury (n = 2). Animals recovered for 7 days after the injury, and neurologic deficit was assessed. On day 7, animals were killed and neuronal survival was assessed. Sham-operated (n = 6) and naive (n = 6) animals served as controls.

In parallel, 18 animals were treated for the bcl-2 mRNA expression study and were killed 6 h (vehicle, n = 4; erythromycin, n = 4) or 24 h (n = 5, respectively) after treatment. Bcl-2 protein expression was assessed in three groups of animals killed 6 h after erythromycin (n = 7) or vehicle (n = 7) or 24 h after erythromycin (n = 7).

All animals were randomized to treatment groups, and investigators were blinded to group assignment throughout the experiments and until full completion of the data acquisition.

Animals were fasted overnight before surgery. Fifteen minutes of transient global ischemia was achieved using bilateral carotid artery occlusion and hypobaric hypotension, as previously described in detail.10,18In brief, rats were anesthetized (chloral hydrate, 360 mg/kg intraperitoneally), intubated, and mechanically ventilated. Both carotid arteries were exposed, and the left side was catheterized for blood pressure monitoring (complete vessel occlusion). For the brain injury, the right carotid artery was occluded by a thread wrapped around it and pulled by a 5-g weight. Fifteen minutes of global cerebral ischemia was achieved by the simultaneous reduction of mean arterial blood pressure to 35 mmHg using the hypobaric hypotension technique.18,19Ischemia was verified by laser-Doppler monitoring as previously described (repetitive regional cerebral blood flow measurements at 30 different locations [scanning technique] over the right hemisphere [closed cranial window, 24 mm², 1.3–5.3 mm lateral to the right and 1.5–7.5 mm occipital of the bregma]; stationary laser-Doppler flow probe [local cerebral blood flow] over the left hemisphere [closed cranial window, 4 mm², 3.3–5.3 mm lateral to the left and 5.0–7.0 mm occipital of the bregma]10,20,21). After 15 min of ischemia, the nylon thread was removed and hypobaric hypotension was terminated to allow reperfusion of the brain. Rectal temperature was controlled to 37.5°± 0.1°C using a thermostatically regulated warming blanket. Peribrain temperature was recorded at two different sites (right temporal muscle, left auricular tube) and controlled to 37°C using a near-infrared heating lamp throughout the experiment, as previously described.22After 90 min of reperfusion, the carotid artery catheter was removed, incisions were closed, and the animals were extubated and returned to their cage. Animals randomized to sham-operation were treated in the same way, except ischemia was not induced.

A neurologic evaluation was performed once daily (day −1, i.e. , the day before the injury [baseline] through day 7) at the same time (around 5:00 pm) by the same investigator, who was unaware of the group assignment. Consciousness, breathing, smell, vision and hearing, reflexes, motor function, overall activity, orientation, and presence of seizures were scored, according to a neurologic deficit score (0–100 scale; 0 = no deficit, 100 = most severe deficit; adapted from Katz et al. ,23Neumar et al. ,24and Nakase et al.  25; table 1). Comparable scoring systems using the same or very similar items have previously been used in our laboratory,10as well as by others.26–29They allow the identification of neurobehavioral deficits after ischemic brain injury and were found to correlate well with other outcome end points, e.g. , electrophysiologic recovery or histopathologic damage.10,26–29All randomized animals were evaluated at baseline (day −1) to ensure a normal neurologic score.

Neuronal cell density (number of viable neurons/mm²) within parietal neocortex and hippocampus (CA1, CA2, CA3, CA4) was assessed in hematoxylin and eosin–stained coronal brain slices using a video microscope in conjunction with a computer system as previously described18and averaged for both hemispheres. The investigator was unaware of treatment assignment, and all brain slices evaluated were coded to ensure an unbiased cell counting and data evaluation.

Anesthetized rats (chloral hydrate, 360 mg/kg intraperitoneally) were decapitated; hippocampus and neocortex were dissected and snap-frozen. Tissue samples were homogenized for each animal, and total RNA was extracted.30First-strand complementary DNA was synthesized after DNase I digestion (Invitrogen, Carlsbad, CA). Real-time quantitative polymerase chain reaction was performed for bcl-2 and glyceraldehyde-3-phosphate-dehydrogenase on a LightCycler® (Roche Molecular Biochemicals, Mannheim, Germany) thermocycler, using FastStart SYBR Green® PCR Reagents (Roche). Bcl-2 expression was normalized to glyceraldehyde-3-phosphate-dehydrogenase and analyzed with RelQuant® software (Roche). Polymerase chain reaction experiments on duplicate samples were repeated twice.

Anesthetized animals (chloral hydrate, 360 mg/kg intraperitoneally) were perfused with freshly prepared ice-cold buffered paraformaldehyde. Brains were processed for immunohistochemical analysis as previously described.10Briefly, coronal brain slices (3 μm) were incubated with rabbit polyclonal anti–bcl-2 antibody (Santa Cruz Biotechnology, Santa Cruz, CA) and visualized with diaminobenzidine (DAKO, Hamburg, Germany). Antibody specificity was confirmed omitting the primary antibody.

The intensity of bcl-2 immunoreactivity was evaluated using a semiquantitative grading procedure.10,31Neuronal staining in parietal neocortex, hippocampus (CA1, CA3, CA4), and dentate gyrus was graded on a scale ranging from 0 for no staining to 4 for strong immunoreactivity.

Parametric data, i.e. , physiologic variables, neuronal cell densities, and bcl-2 mRNA expression are shown as mean ± SD and were found to be normally distributed. For these data sets, one-way analysis of variance and post hoc  Student-Newman-Keuls test was used to identify differences between experimental groups. Nonparametric data (i.e. , neurologic deficit score) were expressed as median ± 25th/75th percentiles, and groups were compared using the rank sum test. Significance was set at P < 0.05. Because of the semiquantitative character of the grading procedure for Bcl-2 protein expression, immunohistochemistry data were not statistically evaluated for differences between groups.

At baseline, vital signs of all animals from the cohort were within normal limits, independent of treatment assignment. Upon initiation of global cerebral ischemia, physiologic variables changed as expected for all animals randomized to the injury, but normalized during early reperfusion. Cerebral blood flow ceased immediately after induction of bilateral carotid artery occlusion and hypobaric hypotension, and laser-Doppler flow measurements showed values reflective of biologic zero (5 ± 1 laser-Doppler units during the first minute and 2 ± 1 during the remaining 14 min). No apparent differences were found between treatment groups besides a lower partial pressure of carbon dioxide and corresponding pH during reperfusion in the erythromycin group (see table 2for details).

All animals exhibited a moderate neurologic deficit on the first days after the ischemic injury. Deficit improved in all animals over time, although baseline (no neuronal deficit on the day before the brain injury; day −1) was not reached again during the postischemic observation period. Deficits were consistently less severe in animals that had been pretreated with erythromycin (fig. 1; P < 0.05 on days 4–7).

With vehicle pretreatment (n = 25, no difference between treatment intervals; pooled), neuronal cell density was reduced by 75% seven days after transient global cerebral ischemia, as compared with sham-operated controls (n = 6). More neurons survived in animals treated with erythromycin before the injury. Erythromycin pretreatment 12 h before ischemia increased the number of surviving neurons in CA1 by threefold, as compared with vehicle pretreatment (P < 0.05; fig. 2).

With vehicle pretreatment, approximately 20% of viable neurons in this brain region were lost 7 days after transient global cerebral ischemia (not significant, P = 0.052). In animals pretreated with erythromycin 12 h before the injury, 95% of neurons survived (P = 0.066 vs . vehicle; fig. 2).

Transient global cerebral ischemia with vehicle pretreatment caused a significant reduction of neuronal cell counts in CA2 and CA3 (P < 0.05, as compared with sham operation) but not CA4 subregions. Animals pretreated with erythromycin 12 h before ischemia had cell counts similar to sham-operated controls (table 3; P < 0.05 vs.  vehicle in CA3).

Erythromycin injection was followed by an increased expression of bcl-2 mRNA in hippocampus (1.6-fold compared with vehicle; P < 0.05) and neocortex (1.9-fold; not significant) at 6 h but not at 24 h (fig. 3). We observed a spatial difference in bcl-2 mRNA expression between neocortex and hippocampus at 6 h but not at 24 h, independent of treatment assignment.

Bcl-2 protein staining was strictly localized to the cytoplasm of the neurons evaluated (fig. 4). At baseline (6 h after vehicle treatment, n = 7), staining was virtually absent (grade 0) in dentate gyrus, faint (1) in CA1, mild (2) in CA4, and mild to moderate (2–3) in CA3 and the parietal neocortex. However, Bcl-2 immunoreactivity did not change at either 6 or 24 h after erythromycin treatment compared with vehicle (n = 7 each; fig. 5).

The presented data show for the first time that a single dose of the antibiotic erythromycin, applied 6–24 h before transient global cerebral ischemia, induces tolerance and improves postischemic outcome, as measured by neurologic deficit and neuronal survival, in a well-controlled chronic rat model (pharmacologic preconditioning).

The applied injury resulted in the significant histopathologic damage in ischemia susceptible brain regions, i.e. , hippocampal CA1, CA2, and CA3, 7 days after global ischemia, that is expected in this model.19The CA1 and CA3 regions were well protected by erythromycin pretreatment 12 h before the injury. In contrast, the same treatment did not render a statistical significant improvement in hippocampal CA2 and parietal neocortex. This may be due to the relatively smaller damage in these regions, compared with CA1. However, differences were close to statistical significance (P = 0.052–0.086), so small sample sizes may have prevented us from detecting protection in these regions. Alternatively, these findings could suggest region-specific differences in the neuroprotective potential of erythromycin preconditioning. Neuronal damage in hippocampal CA4 usually is small with the applied injury, limiting the potential to detect a significant neuroprotection in this region.

Although neuronal density is a readily available tool to assess ischemic damage, neurologic deficit and function are even more relevant for clinical applications. The erythromycin-pretreated animals fared better throughout the entire 7-day observation period. This was significant in the second half of the follow-up period, at a time when animals have recovered from immediate surgical stress and residual anesthetic effects, and cerebral edema has vanished.

Our in vivo  findings confirm previous ex vivo  work on hippocampal slices, demonstrating tolerance against hypoxia after erythromycin pretreatment.13The extent of hippocampal CA1 protection from ischemic damage that could be achieved by erythromycin pretreatment in our model (i.e. , tripling of the number of surviving neurons) was similar to, or even more pronounced than, the effects we observed earlier after a chemical preconditioning regimen using a single dose of 3-nitropropionic acid10(50% increase in surviving CA1 neurons). In animals pretreated with erythromycin 12 h before ischemia, 75% of CA1 neurons survived 7 days after ischemia, a survival rate that is comparable to the 70–80% surviving CA1 neurons found after classic ischemic preconditioning in rat models of global cerebral ischemia using four-vessel occlusion32or bilateral carotid artery occlusion with hemorrhagic hypotension.33,34Tolerance induction by the clinically used drug erythromycin may therefore be similarly effective as these earlier, more invasive regimens, but contrary to its predecessors, this new strategy lacks significant side effects and could therefore potentially be applied to human patients.

Previous work suggests that preconditioning has two windows of opportunity, one early (at approximately 3 h) and one late (24 h and beyond).1,10,25,35–37The effect of erythromycin preconditioning in our hands was maximal after a pretreatment interval of 12 h before the injury, somewhat earlier than the classic delayed preconditioning window. This is in accordance with ex vivo  work that found brief pretreatment intervals to be most effective for tolerance induction by erythromycin.13 

A small and transient increase of bcl-2 mRNA was noted early after erythromycin treatment. Expression of bcl-2 mRNA was increased 6 h but not 24 h after erythromycin in both neocortical and hippocampal specimens, but Bcl-2 protein did not increase in the same regions at either time point.

Baseline expression of both bcl-2 mRNA and Bcl-2 protein was higher in neocortical than in hippocampal (CA1) neurons. This correlates well with previous reports10,31and may in part explain the relatively higher intrinsic tolerance of cortical neurons against the applied ischemic injury in this model.

Erythromycin is known to inhibit mitochondrial protein synthesis38–40and allegedly reduces cell respiration by reducing the production of enzymes necessary for optimal function of the respiratory chain.40,41A functional impairment of mitochondrial oxidative phosphorylation after the application of a single dose of erythromycin might therefore cause a mild burst of reactive oxygen species in neuronal cells, which subsequently may induce gene expression of, for example, apoptosis-inhibiting bcl-2, similar to the mechanism that has been described for 3-nitropropionic acid.10,42 

However, differential bcl-2 expression after erythromycin injection was transient and less pronounced than what we saw after 3-nitropropionic acid. It may therefore only partially explain the observed neuroprotection provided by the antibiotic.

On the other hand, erythromycin may possibly suppress the immunologic response that mediates damage after cerebral ischemia,43–45because this macrolide has been reported to influence several inflammatory mechanisms in other organ systems.46,47This concept is supported by recent reports on the reduction of postischemic cytokine expression and inflammation after cerebral preconditioning48,49and the beneficial effects of antibiotic treatment after experimental ischemia,50and certainly warrants further experimental work.

Sample size is a potential limitation of our study. We had performed a power analysis to determine sample size before we started our study. However, besides profound protection by erythromycin in hippocampal CA1 and CA3, histopathologic analysis found differences between treatment groups in parietal neocortex and hippocampal CA2 that were very close to but did not reach statistical significance. Interpretation of these data are difficult because, by convention, there are no differences, but one is tempted to state a trend. Future studies in other models may help to decide whether tolerance induction by erythromycin is indeed region specific, and less pronounced in CA2 compared with CA1.

We assessed bcl-2 expression after a 6-h and a 24-h but not after a 12-h pretreatment interval. However, our previous work using a model of chemical preconditioning showed an early and sustained increase of bcl-2 expression starting at 3 h, most pronounced at 24 h after pretreatment.10We found only small and transient changes of bcl-2 mRNA, but not protein, expression early after erythromycin treatment. Therefore, we do not expect results to be different after 12 h. We therefore conclude that bcl-2 is not as relevant for tolerance induction by erythromycin as it is for other preconditioning regimens.

Erythromycin preconditioning improves neurologic function and neuronal survival after ischemia and is associated with a transiently increased cerebral expression of the antiapoptotic gene bcl-2 in rats. This neuroprotective effect of erythromycin suggests a potential clinical strategy for preischemic conditioning, which could be beneficial for patients scheduled to undergo surgical procedures associated with an increased risk of perioperative brain ischemia, e.g. , in cardiovascular surgery or neurosurgery. Additional research is needed to determine the clinical role for this new method of preemptive neuroprotection and to further clarify the molecular mechanisms involved.

The authors thank Annett Ehlert (Technician, Institute for Neurosurgical Pathophysiology, Johannes Gutenberg-University, Mainz, Germany), Mitch Malzahn (Technician, Institute for Neurosurgical Pathophysiology, Johannes Gutenberg-University), and Dana Pieter (Technician, Department of Anesthesiology, Johannes Gutenberg-University) for excellent technical assistance.