Main

Experimental studies in a well characterized perinatal stroke model, elicited by unilateral carotid artery ligation and subsequent timed exposure to moderate hypoxia in immature rats(1) have facilitated analysis of developmental stage-specific features of the pathogenesis of hypoxic-ischemic brain injury. Many interrelated mechanisms, including EAA receptor overactivation, intracellular Ca2+ accumulation, lipid peroxidation, and free radical generation, contribute to the evolution of hypoxic-ischemic injury(1, 2). Recent evidence implicates inflammatory cells and cytokines as mediators of hypoxic-ischemic injury in immature brain. Neutrophils(3) and activated macrophages and microglia(4, 5) accumulate, and cytokine (e.g. IL-1β and tumor necrosis factor-α) gene expression(6) or activity(7) are increased within the first 6 h after the hypoxic-ischemic insult. Neutrophil depletion(3) and treatment with a pharmacologic antagonist of IL-1(8) both confer neuroprotection in this model.

Recent data suggested that PAF(1-O-alkyl-2-acetyl-sn-glycero-3-phosphocholine), a potent phospholipid inflammatory mediator, could play a pathophysiologically important role in the progression of ischemic brain injury(9, 10). PAF has also been implicated as a mediator of tissue injury in a broad range of pathologic processes, including septic shock, and gastrointestinal, cardiac, and renal ischemia(11). Little is known about the role of PAF or the neuroprotective efficacy of PAF antagonists in the immature brain; delayed PAF metabolism and indirect evidence of increased brain PAF levels were reported in a rat model of intrauterine asphyxia(12).

PAF is synthesized(12–14) and PAF receptors are expressed in the brain(14–16). PAF is produced by neutrophils, monocyte-macrophages, eosinophils, platelets, and endothelia(17). It is uncertain which CNS cells synthesize PAF or express PAF receptors under normal or pathologic circumstances(15). In vivo, ischemia or convulsions increase CNS PAF levels(13, 18, 19).

In addition to its role in inflammation, PAF may also play a role in EAA synaptic activity(20, 21). PAF and PAF antagonists influence EAA-mediated events, but not by direct effects on postsynaptic EAA receptors(22). Presynaptic PAF receptors may augment EAA release; the PAF analog carbamyl-PAF augments glutamate-mediated evoked excitatory transmission in vitro(22), and the PAF receptor antagonist BN 52021 inhibits stimulated hippocampal EAA release under “ischemic” conditions in vitro(23). PAF antagonists could attenuate CNS injury by several possible mechanisms, including interruption of the inflammatory response, attenuation of EAA release, and/or prevention of cerebrovascular microthrombosis.

BN 52021 protects neurons from glutamate toxicity in vitro(24). In some mature animal models of cerebral ischemia PAF antagonists are neuroprotective(15); however, results have been inconsistent(25); factors that account for variability in outcome are uncertain. PAF may play a more prominent functional role in developing than in mature brain; in humans, expression of the gene for the γ catalytic subunit of the crucial PAF degradative enzyme cytosolic PAF-acetylhydrolase is developmentally regulated, with the greatest expression in fetal brain(26). Evidence of the critical role that PAF plays in brain development was provided by the intriguing observation that the α subunit of cytosolic PAF-acetylhydrolase is a product of the LIS-1 gene; mutation ofLIS-1 is causally linked to the developmental CNS malformation Miller-Dieker lissencephaly(27). Thus, immature brain injury models may provide the optimal setting in which to evaluate potential detrimental effects of PAF antagonist strategies.

To test the hypothesis that PAF antagonist treatment would attenuate perinatal hypoxic-ischemic brain injury, we evaluated the neuroprotective efficacy of the competitive cell membrane PAF receptor antagonist BN 52021(ginkgolide B)(28) in immature (P7) rats. The initial dose regimen was based on a regimen which attenuated ischemic injury in an adult rat model of transient global cerebral ischemia(24). To elicit focal hypoxic-ischemic brain injury, P7 rats underwent right carotid ligation, followed by timed (2.5-3.25 h) exposure to 8% O2; severity of injury in treated and control animals was compared 5 d later. We hypothesized that if PAF antagonists inhibited the postinjury inflammatory cascade, treatment initiated after the onset of injury could be effective; therefore we evaluated the efficacy of BN 52021 in both prophylactic and posthypoxia-ischemia regimens.

Our results demonstrated that systemic administration of BN 52021 resulted in attenuation of hypoxic-ischemic brain injury in P7 rats, even when therapy was instituted after the hypoxic-ischemic insult and provide support for the hypothesis that PAF is a mediator of hypoxic-ischemic injury in the developing brain.

METHODS

Surgical methods. All surgical procedures were performed in P7 unsexed Sprague-Dawley rats (Charles River Laboratories), and used previously published methods(1). The protocols were approved by the University of Michigan Committee on Care and Use of Animals. All efforts were made to minimize animal suffering and minimize the number of animals used.

Induction of unilateral hypoxic-ischemic brain injury. Methoxyflurane-anesthetized P7 rats underwent right carotid artery ligation, and, after 1 h of recovery in a warm air incubator (ambient temperature: 35-36°C), were exposed to 8% O2, in glass chambers partially immersed in a 37 °C water bath. The duration of hypoxia ranged from 2.5 to 3.25 h among experiments; this was done both to compensate for to minor variations in Fio2 among commercially prepared gas mixtures and also to attempt to produce lesions of similar severity in all controls. In our experience with P7 rats, the duration of hypoxia necessary to induce severe ipsilateral forebrain infarction varies inversely with body weight (our unpublished observations); because mean animal weight/litter ranged from 11 to 16 g, the individuals performing the experiments (X.-H.L,B.-L.E.) determined the duration of hypoxic exposure for each litter based on animal size. In each experiment, all animals were exposed to the same duration of hypoxia.

To ensure that normothermia was maintained, for the first 4 h of the posthypoxia-ischemia recovery period, animals were housed in a warm air incubator (ambient temperature: 35-36 °C); then they were returned to the dam. Although fasting is mildly neuroprotective in this model(29), animals in all treatment groups were equally fasted. In one representative experiment, body temperature was monitored regularly in animals from each treatment group during this period (every 15 min for 1 h, then every 30 min), using a 0.6-mm diameter flexible probe, inserted orally to a depth of ≥3cm (YSI thermometer 43TA with probe 554, Yellow Spring Instruments, Yellow Springs, OH).

Animals were killed by decapitation on P12; brains were removed and either frozen under powdered dry ice or processed for hemisphere weights (see below).

Intracerebral injection. To determine whether BN 52021 neuroprotection could be mediated by antagonism of NMDA-sensitive EAA receptors, we tested whether BN 52021 treatment attenuated brain injury resulting from intracerebral NMDA injection. Methoxyflurane-anesthetized P7 rats were placed in a stereotaxic frame. Skull surface landmarks were exposed with a midline scalp incision, a 21 gauge needle was used to make a burr hole in the skull over the targeted injection site, and then a 25 gauge needle attached to a Hamilton syringe was inserted through the burr hole. The injection coordinates [anteroposterior: -2.0 mm, mediolateral: 2.5 mm, ventral: 4.0 mm, anteroposterior and mediolateral relative to bregma; depth relative to dura] were targeted to right dorsolateral hippocampus(30). NMDA 10 or 12.5 nmol/0.5 μL (dissolved in PBS, final pH 7.4) was injected over 2 min; the needle was left in place for another 2 min, to minimize leakage. After recovery from anesthesia, animals were returned to their dam. On P12, animals were killed by decapitation, and brains were removed and frozen under powdered dry ice.

BN 52021 drug treatment. Outcome was compared in litter mates that underwent hypoxic-ischemic lesioning concurrently. BN 52021 was administered by intraperitoneal injection (volume: 0.1 mL); in all experiments, controls received injections of equal volumes of the vehicle. Neuroprotective efficacy versus hypoxic-ischemic injury and the optimal timing and dosage were evaluated in three groups of experiments. In the first group, animals were lesioned (n = 25), and then received BN 52021 (25 mg/kg/dose) (n = 12, two died during hypoxia) or vehicle (n = 13; four died during hypoxia), immediately after the end of hypoxia, and again 2 h later. In the second group, lesioned animals(n = 24) received the first dose of BN 52021 (25 mg/kg/dose)(n = 13; one died during hypoxia) or vehicle (n = 11) immediately before the onset of exposure to 8% O2, and the second dose at 1 h after the end of the hypoxia. In the third group animals underwent right carotid artery ligation, followed by 8% O2 exposure (n= 62), and then received treatment with BN 52021, 12.5, 25, or 50 mg/kg/dose(n = 11, 11 and 12, respectively), or vehicle (n = 28) immediately and again 2 h after the end of hypoxia.

Two groups of experiments were performed to determine whether BN 52021 treatment could attenuate NMDA-induced hippocampal injury. In the first group, animals were treated with BN 52021 (25 mg/kg/dose) (n = 13, two died on P10) or vehicle (n = 13, one died 2 h post-NMDA) 2 h before and again 2 h after intrahippocampal NMDA (10 nmol) injection. In the second group, animals were treated with BN 52021 (25 mg/kg/dose) (n = 11) or vehicle (n = 9, 1 died between P8 and P12) at 15 min and again at 2 h after intrahippocampal NMDA (12.5 nmol) injection.

Evaluation of injury. In P7 rats neither carotid ligation or hypoxia alone cause brain injury; hypoxia after ligation induces ipsilateral striatal, hippocampal, and cortical infarction(1). Severity of ipsilateral injury depends on several factors, including hypoxia duration, ambient temperature, and animal age. Typical findings on cresyl violet-stained sections 5 d after injury induction (i.e. on P12) include ipsilateral striatal, cortical, and hippocampal infarction, with pallor indicating neuronal loss; at this stage of rapid brain growth, atrophy is a prominent feature of cerebral infarction. Comparison of bilateral regional cross-sectional areas or of bilateral cerebral hemisphere weights on P12 provides a reliable measurement of the severity of hypoxic-ischemic injury and the efficacy of neuroprotective interventions(31). Intrahippocampal injection of NMDA (1-50 nmol) in P7 rats results in dose-dependent focal neuronal loss; at higher doses (≥12.5 nmol) substantial (>50%) atrophy of the ipsilateral hippocampus is evident by P12, and injury extends to adjacent striatum, thalamus and cortex(31).

Histopathology and image analysis. Twenty micrometer coronal frozen brain sections, postfixed over paraformaldehyde vapors, were stained with cresyl violet. Presence of infarction was determined by light microscopic inspection of at least 20 cresyl violet-stained sections/brain, by an observer unaware of animal treatment group. Criteria for cerebral infarction included pallor (i.e. loss of staining), atrophy, or frank tissue loss of cortex, striatum, hippocampus, or thalamus. To quantitate the severity of hypoxic-ischemic injury, bilateral cross-sectional areas of hemisphere, cortex, striatum, and hippocampus were measured in six standardized coronal sections/brain (see Fig. 1), using a microcomputer image analysis system (the public domain program NIH, Image, running on a Macintosh Quadra 650 with a Scion LG-3 capture board and a Sony XC-77 camera with macro lens, mounted above a stable illumination source). Cortex area measurements included only intact cortex, as judged by intensity and uniformity of cresyl violet staining. Striatal and hippocampal area measurements included the entire residual tissue, because it was not possible to consistently distinguish intact and injured tissue on the basis of distinct and well demarcated differences in staining characteristics.

Figure 1
figure 1

Standardized sections for image analysis. These six cresyl violet stained coronal sections from P12 rat illustrate the six standardized sections used for area measurements to evaluate the severity of brain injury. Areas measured bilaterally in each section are outlined in black. Cortex area measurements included only intact cortex, as judged by intensity and uniformity of cresyl violet staining. Striatal and hippocampal area measurements included the entire residual tissue, because it was not possible to consistently distinguish intact and injured tissue on the basis of distinct and well demarcated differences in staining characteristics.HIP, hippocampus; STR, striatum; CORT, cortex.

Full size image

To evaluate the severity of NMDA-induced hippocampal injury, bilateral hippocampal cross-sectional areas of every sixth section were measured; values were summated and multiplied by the distance between sections (120 μm)(32). Dorsal hippocampal volumes were estimated, based on measurements beginning at the rostral tip and continuing posteriorly to the level of the caudal limit of the corpus callosum.

Hemisphere weight measurements. Presence or absence of ipsilateral cerebral infarction (liquefaction, pallor, atrophy) was first assessed by an observer unaware of treatment identity. Cerebellum and brainstem were removed and hemispheres separated using a razor blade. Bilateral wet hemisphere weights were measured on an electronic balance.

Data analysis. Microcomputer-based statistical programs[Statview II (ABACUS, Berkeley, CA) and Systat (Systat, Inc., Evanston, IL)] were used. In animals that underwent hypoxic-ischemic lesioning, incidence of cerebral infarction in pre- and posttreatment experiments was compared between groups using Fisher's exact test. Right-sided cortical, striatal, and hippocampal areas were compared between animals treated with BN 52021 after the end of hypoxia and concurrent vehicle injected controls by two-way ANOVA, factoring treatment group and section level. Percent damage {i.e. mean% left-right (L-R) difference in areas [100·(L-R)/L]} was first compared independently for each of the four regions analyzed, among pre- and posttreated groups and controls by two-way ANOVA, factoring treatment group, and standardized section level. This initial analysis indicated no effect of section level on percent damage. To evaluate regional differences in neuroprotective efficacy of BN 52021, percent damage was also compared among treatment groups by two-way ANOVA factoring treatment group and region.Post hoc testing (Fisher's least significant difference test) was used to evaluate the significance of subgroup differences. For all percent damage comparisons, values from controls injected either pre- or posthypoxia were pooled for analysis, because they did not differ in any region (data not shown). Percent Protection by BN 52021 was based on comparison of the severity of injury in drug and vehicle treated animals, using the formula{100·[1-(% damage drug-treated/% damage vehicle-treated)]}. As a measure of the range of neuroprotection values observed, the SEM for percent protection was calculated as {100·(SEM, drug treated/% damage vehicle treated)}, as previously described(31).

Mean bilateral hemisphere weights and interhemispheric percent differences in weight [100·(L-R)/L] were compared among groups by ANOVA.Post hoc testing (Fisher's least significant difference test) was used to evaluate subgroup differences. Percent protection was calculated as above, using interhemispheric percent differences in hemisphere weights.

In animals that received intrahippocampal NMDA injection, bilateral hippocampal volumes were compared between BN 52021- and vehicle-injected animals by a two-tailed t test.

Materials. BN 52021 used in initial experiments, in which regional specificity of neuroprotection was assessed, was a gift from Dr. N. G. Bazan (Louisiana State University, New Orleans, LA). BN 52021 used to evaluate dose dependence of neuroprotection was obtained from Biomol Inc.(Plymouth Meeting, PA). BN 52021 was suspended in 5% gum arabic (Sigma Chemical Co., St. Louis MO) in water, with vigorous agitation before each administration. NMDA (Sigma Chemical Co.) was dissolved in PBS, and the pH was adjusted to 7.4 with NaOH.

RESULTS

Right carotid ligation followed by timed hypoxic exposure (2.5-3.25 h, Fio2 = 0.08) on P7 resulted in cerebral infarction, detectable by histopathologic examination on P12, in 90% (18/20) of vehicle-treated controls. In experiments in which hemisphere wet weights were compared, in vehicle-injected controls the mean weight of the right hemisphere was reduced by 31%, in comparison with the weight of the contralateral hemisphere. When infarction was detected by gross inspection of the brain, typical histopathologic correlates included cortical and striatal infarction, with pallor, and cortical, striatal, hippocampal, and thalamic atrophy; in the most severely lesioned brains, the majority of the right hemisphere was liquefied, and few anatomic landmarks were recognizable.

Treatment with BN 52021 (25 mg/kg/dose in two serial doses), initiated either before or after hypoxia, substantially reduced the incidence of histopathologically detectable cerebral infarction (see Table 1); comparison of the frequency of infarction in vehicle and BN 52021-treated groups (by Fisher's exact test) demonstrated significant intergroup differences (pretreatment regimen:p < 0.01; posttreatment regimen p < 0.02).

Table 1 Reduced incidence of neuropathologically detectable cerebral infarction after BN 52021 administration*
Full size table

Both pre- and post treatment with BN 52021 reduced the severity of tissue damage in cortex, hippocampus, and striatum (compared with vehicle-injected controls, p < 0.0001, ANOVA) (see Tables 2 and 3 and Figures 2 and3). There were differences in the efficacy of neuroprotection between pre- and posttreatment protocols. Pretreatment was more effective than treatment initiated posthypoxia in hippocampus and cortex (Fisher's least-significant-difference test,p ≤ 0.05); our method of image analysis may have underestimated additional benefit from pretreatment in the striatum, because we measured the entire residual striatum, rather than attempting to distinguish regions with intact Nissl staining (p = 0.10, Fisher's least significant difference test). For both treatment protocols, there were differences in the efficacy of neuroprotection among regions. When compared with vehicle-treated controls, pretreatment with BN 52021 eliminated from 74% (in striatum) to 90%(in hippocampus) of hypoxic-ischemic injury (see Table 3); posttreatment eliminated from 46% (in striatum) to 72% (in cortex) of hypoxic-ischemic injury.

Table 2 Comparison of regional area measurements in BN 52021 and vehicle treated animals*
Full size table
Table 3 Protection from posthypoxic-ischemic tissue loss by BN 52021 treatment*
Full size table
Figure 2
figure 2

Posthypoxic BN 52021 treatment attenuates perinatal cerebral infarction. This montage compares representative 20-μm coronal cresyl violetstained frozen sections from two P12 animals that underwent right carotid ligation and 2.5-h 8% O2 exposure on P7, followed by administration of BN 52021 25 mg/kg (A and B) or an equal volume of vehicle (C and D), intraperitoneally immediately and again 2 h after the end of hypoxia. In C and D, note right cortical infarction (star). striatal atrophy and infarction(asterisk), and hippocampal atrophy and pyramidal cell loss(arrows). In A and B, note subtle reduction in right hemisphere area, without evidence of focal infarction (scale bar = 1 mm).

Full size image
Figure 3
figure 3

Pre- or posthypoxic BN 52021 treatment attenuates regional ischemia-induced atrophy.% Damage indicates the mean percent reduction in right sided areas, measured on P12, ipsilateral to carotid ligation, in comparison to intact left sided regions. Area measurements used were the means of six levels for cortex, four levels for striatum, and three levels for hippocampus. All animals underwent right carotid ligation and 2.5-3.25-h 8% O2 exposure on P7. BN 52021 (25 mg/kg/dose) was administered in 2 serial doses, either immediately before and 1 h after the end of hypoxia (BN pre), or immediately after and again 2 h after the end of hypoxia (BN post). Because mean area measurements in vehicle controls did not differ between experiments, all controls were pooled for this statistical analysis. Both pre- and post treatment with BN 52021 conferred neuroprotection (*p < 0.0001, ANOVA). Posthoc statistical analysis indicated greater attenuation of injury by BN 52021 pretreatment compared with posttreatment in cortex and hippocampus(**p ≤ 0.05, Fisher least significant difference test).

Full size image

Attenuation of ipsilateral forebrain injury by posthypoxic administration of BN 52021 was confirmed in additional experiments; measurement of bilateral hemisphere weights on P12 was used to quantitate severity of injury in animals that were treated with two serial doses of BN 52021 (12.5, 25, or 50 mg/kg/dose) immediately and 2 h posthypoxia. In animals treated with 25 or 50 mg/kg/dose, mean percent interhemispheric weight differences (a measure of tissue damage) were significantly less than in vehicle-injected controls (seeFig 4). Percent protection (based on inter hemispheric weight differences, mean ± SEM) did not differ significantly among the three BN 52021 doses [18.9% (±12.3) for 12.5 mg/kg/dose; 44.8%(±9.6) for 25 mg/kg/dose; 32.5% (±6.1) for 50 mg/kg/dose]; however, it is possible that some dose dependence was not detected due to a type II error.

Figure 4
figure 4

Efficacy of posthypoxic BN 52021 treatment: dose effect. The vertical axis indicates percent interhemispheric weight difference(mean, ±SEM) in P12 rats that underwent right carotid ligation and 2.5-3.25-h 8% O2 exposure on P7, followed by administration of BN 52021 12.5 (BN 12.5), 25 (BN 25), or 50 (BN 50) mg/kg/dose, or vehicle (VEH), in two serial doses immediately and again 2 h after the end of hypoxia. Right hemisphere tissue loss was attenuated by BN 52021 administration (p < 0.005, ANOVA).Post hoc testing indicated that inter hemispheric weight differences were significantly reduced (indicating neuroprotection) in animals treated with 25 or 50 mg/kg/dose, compared with controls (*p < 0.05, Fisher least significant difference test); however, there was no significant difference in the efficacy of neuroprotection among the three drug dose groups.

Full size image

In contrast, BN 52021 did not attenuate NMDA-induced hippocampal injury. Treatment initiated 2 h before intrahippocampal injection of 10 nmol of NMDA(25 mg/kg/dose, 2 h pre- and 2 h post-NMDA), did not reduce injury [mean right hippocampal volume (mm3, ±SEM): NMDA + BN 52021 6.0 ± 1.1versus NMDA + vehicle 6.3 ± 0.9, p = NS]. The outcome was similar when BN 52021 treatment was instituted after intrahippocampal injection of 12.5 nmol of NMDA (25 mg/kg/dose, 15 min and 2 h post-NMDA), [mean right hippocampal volume (mm3, ±SEM): NMDA + BN 52021 4.7 ± 1.2 versus NMDA + vehicle 5.1 ± 1.0,p = NS].

There were no apparent behavioral effects and no increase in mortality associated with drug administration to hypoxic-ischemic lesioned animals. One BN 52021-pretreated animal died during hypoxia, all other deaths in animals assigned to BN 52021 treatment groups occurred during the period of exposure to hypoxia, before drug treatment was administered. There were no late deaths(i.e. between P8 and P12) in control or BN 52021-treated animals. There was no effect of BN 52021 posttreatment on body temperature.

DISCUSSION

These results demonstrate the neuroprotective efficacy of the PAF receptor antagonist BN 52021 in a perinatal rat model of cerebral hypoxic-ischemic injury, and support the hypothesis that PAF is a mediator of hypoxic-ischemic injury in the immature brain. Of potential therapeutic importance is our finding that treatment with BN 52021 after the hypoxic-ischemic episode resulted in substantial neuroprotection. In addition, this finding suggests that a major mode of action of this PAF antagonist involves suppression of the postinsult inflammatory cascade.

In this animal model, ipsilateral cerebral blood flow is reduced during hypoxia(33), but is restored to normal during the first 3 d after hypoxia exposure(34); it is likely that a combination of hypoxia-ischemia and reperfusion injury contribute to tissue damage. Because cerebral blood flow is restored to normal after hypoxia(34), it is unlikely that effects of posthypoxic BN 52021 administration on cerebral blood flow contribute substantially to neuroprotection; however, we did not directly evaluate the effect of this drug on cerebral blood flow. In other tissues, PAF, a by-product of lipid peroxidation, may be one of the earliest inflammatory mediators released in response to ischemia-reperfusion injury(36). Increased tissue PAF production during ischemia or reperfusion may rapidly induce expression of leukocyte adhesion molecules in endothelium and result in local accumulation and activation of inflammatory cells, e.g. neutrophils and monocyte-macroph-ages; this would, in turn, induce local production of proinflammatory cytokines (such as IL-1β, tumor necrosis factor-α, and IL-6) as well as further increase PAF production(36). PAF receptor antagonist therapy may interrupt both early activation and later persistence of the inflammatory response to cerebral ischemia-reperfusion injury at multiple steps, including selectin-or integrin-mediated neutrophil or monocyte-macrophage accumulation/activation and increased cytokine production. The neuroprotective mechanisms of the PAF receptor antagonist BN 52021 are incompletely understood. A chemically distinct PAF antagonist, WEB 2170, decreases early cerebral edema, but not later neuropathologic changes, in this injury model(35); we did not evaluate the effect of BN 52021 treatment on early cerebral edema.

Studies using PAF antagonists in mature animal models of cerebral ischemia have yielded inconsistent results. Both post-ischemic PAF antagonist administration(37, 38) and pretreatment(24) were reported to protect the mature CNS from ischemic injury. However, other reports have failed to substantiate neuroprotective effects of PAF antagonists(25). Possible explanations for these inconsistencies include differences in animal species, disease models, or outcomes measured. Recent data suggest that previously reported elevations in CNS concentrations of PAF in mature animal cerebral ischemia models may have been overestimated(39, 40); thus PAF may not contribute significantly to neuronal death in some adult cerebral ischemia models. The remarkable neuroprotective effect of BN 52021 in our studies raises the possibility that PAF antagonists may be more efficacious neuroprotective agents in immature than adult brain. Indeed, there is some evidence that PAF is functionally more important in immature than mature brain(26, 27).

Intracerebral NMDA injection in P7 rats replicates many pathophysiologic features of the perinatal stroke model(41), and both intracerebral NMDA injection and hypoxic-ischemic lesioning stimulate cytokine gene expression(6) and microglial activation(42). Yet, our finding that BN 52021 did not attenuate NMDA-induced hippocampal injury suggests that inflammatory responses to hypoxic-ischemic and excitotoxic injury differ in some critical respects. This discrepant outcome may ultimately facilitate identification of the primary mode of action of BN 52021 as a neuroprotective agent.

In this experimental model, hypoxic-ischemic brain damage is attenuated by posthypoxic-ischemic administration of several agents with diverse modes of action, including: the NMDA receptor antagonist MK-801 (43-76% protection)(43, 44); the AMPA[α-amino-3-hydroxy-5-methyl-4-isoxazole propionate] receptor antagonist NBQX (28% protection)(44); and recombinant IL-1 receptor antagonist (75-100% protection)(8). Thus it appears that drugs acting upon a wide range of cellular and molecular targets can suppress progression of hypoxic-ischemic brain injury. Our results demonstrate that BN 52021 may be one of the most potent postischemic neuroprotective agents for perinatal cerebral hypoxia-ischemia. Whether this reflects actions at multiple target sites is uncertain. It will be of interest to evaluate whether low doses of PAF antagonists have additive or synergistic effects with other neuroprotective drugs. Such interactions could facilitate delineation of the mode of action of these drugs.

Some neuroprotective drugs have significant detrimental side effects in immature animals, e.g. sedation and respiratory depression with MK-801 and NBQX, and nephrotoxicity with NBQX. Potential adverse effects of PAF antagonists include altered excitatory synaptic transmission acutely(20–22) or disruption of neuronal connectivity(45). We did not detect any adverse effects associated with BN 52021 administration; there was no CNS depression or readily evident disruption of CNS development. However, more detailed long-term studies will be required to identify subtle adverse effects in CNS maturation.

In conclusion, our results further support the role of the acute inflammatory response as a mediator of perinatal hypoxic-ischemic brain injury. PAF antagonists, or other strategies to decrease PAF-mediated effects, may offer a novel pharmacologic approach to decrease the incidence and severity of perinatal hypoxic-ischemic brain injury.