Reperfusion Injury as the Mechanism of Brain Damage after Perinatal Asphyxia

Abstract

Upon reperfusion of ischemic tissues, reactive oxygen metabolites are generated and are responsible for much of the organ damage. Experimental studies have revealed two main sources of these metabolites: 1) the oxidation of hypoxanthine to xanthine and on to uric acid by the oxidase form of xanthine oxidoreductase and 2) neutrophils accumulating in ischemic and reperfused tissue. Blocking either source will reduce reperfusion damage in a number of experimental situations. Although xanthine oxidoreductase activity may be unmeasurably low in organs other than liver and intestine, it may be involved in reperfusion injury elsewhere because of its localization in capillary endothelial cells. Time course considerations suggest that substrate accumulation and NADH inhibition of dehydrogenase activity may be more important in the pathogenesis than conversion of xanthine dehydrogenase into the oxidase form. Neutrophil accumulation may be partly due to oxidants in the first place, suggesting a link between the two sources of reactive oxygen metabolites. In the clinical context, many of the sequelae of perinatal asphyxia may be accounted for by reperfusion damage to organs such as brain, kidney, heart, liver, and lungs. During asphyxia, substrates of xanthine oxidase accumulate, upon resuscitation the cosubstrate oxygen is introduced, and evidence for oxidant production and effects has been obtained. In the pathogenesis of brain damage after asphyxia, both microvascular injury and parenchymal cell damage are important. Oxygen metabolites are involved in the former, but in the latter process their role is less clear because ischemia-reperfusion triggers not only oxidant production but many other phenomena, including gene activation, ATP depletion, glutamate accumulation, and increase of intracellular calcium. A severe insult results in cell necrosis, but more moderate asphyxia may cause delayed neuronal death through apoptosis. The time course of the changes in high energy phosphates as well as of selective neuronal death suggest that in the first hours of life there is a“therapeutic window,” with future possibilities for prevention of permanent damage.

Main

Severe and prolonged ischemia or hypoxia of any organ will result in cell death and tissue damage. Thus, restoration of blood flow or oxygen delivery is necessary for organ survival, but it is, on the other hand, critical because damage may be amplified during this period. For example, if the superior mesenteric artery of a cat is clamped for 3 h and then reperfused for 1 h, mucosal damage is more severe than after 4 h of continuous ischemia⁽¹⁾. Reperfusion with deoxygenated blood causes no further damage beyond that observed after 3 h of ischemia⁽¹⁾, suggesting that oxygen and its reactive metabolites are implicated in the pathogenesis. A number of other studies have provided evidence for the basic concept and for the role of oxygen metabolites^{(2, 3)}.

Perinatal asphyxia, with its attendant bradycardia and hypotension, is an ischemic condition, which is followed by reperfusion upon resuscitation. After asphyxia, dysfunction of several organs is common, and permanent damage, particularly of the brain, is possible. In this review, we will consider the potential sources of reactive oxygen metabolites in reperfusion injury and their relevance to postasphyxial organ damage, with emphasis on the brain. We will also review the evidence on circulatory changes associated with experimental and clinical asphyxia and the mechanisms of hypoxic-ischemic cell damage, especially the role of oxidants.

GENERATION OF REACTIVE OXYGEN METABOLITES UPON REPERFUSION

Of the main reactive metabolites of oxygen relevant to human pathology, O₂[horizontal line over dot] and H₂O₂ can be formed through several metabolic pathways [reviewed in Kinnula⁽⁴⁾]. They in turn can produce the most reactive of the oxygen metabolites, OH^˙, through the Haber-Weiss or Fenton reactions catalyzed by Fe²⁺⁽⁵⁾. In the context of reperfusion injury, the two important sources of reactive oxygen metabolites are XOR^{(3, 6)} and inflammatory cells⁽⁷⁾, but other pathways may be significant, depending on the species and cell type^(8–11).

XOR. XOR catalyzes the last two steps of purine catabolism in man, oxidation of hypoxanthine to xanthine and on to uric acid^{(12, 13)}. Under physiologic conditions, the enzyme exists mainly as a dehydrogenase (XOR_D), uses NAD⁺ as the electron acceptor, and generates no harmful metabolites. In ischemic tissue XOR_D can be converted into an oxidase (XOR_O), which utilizes molecular oxygen as the electron acceptor and produces O₂[horizontal line over dot] and H₂O₂⁽¹⁴⁾. Also during ischemia, ATP is catabolized and hypoxanthine accumulates^{(3, 15–17)}. During reperfusion the cosubstrate O₂ is reintroduced, and the stage is set for production of reactive oxygen metabolites (Fig. 1). The other product of XOR, uric acid, is a free radical scavenger and may function as an extracellular antioxidant in man⁽¹⁸⁾. Whether it has any significance in reperfusion injury is unclear.

Figure 1

Production of reactive oxygen metabolites by XOR during reperfusion. Dark boxes represent compounds accumulating during ischemia. Broken lines indicate inhibition of XOR_D by NADH.

Evidence for the role of XOR in reperfusion injury is mainly indirect, based on prevention of organ damage in animal models with XOR inhibitors(allopurinol or oxipurinol). Hundreds of such studies have been reported, relevant to reperfusion damage of the heart [reviewed in Godin and Ko⁽¹⁹⁾], intestine [reviewed in Granger⁽⁶⁾], kidney⁽²⁰⁾, liver [reviewed in Lemasters and Thurman⁽²¹⁾], lung⁽²²⁾, skeletal muscle⁽²³⁾, and brain [reviewed in Palmer⁽²⁴⁾]. However, amelioration of reperfusion injury by allopurinol has not been consistent, and mechanisms unrelated to XOR inhibition may be at play, e.g. increased salvage of purine bases to maintain the ATP pool⁽²⁵⁾ or direct scavenging of free radicals by allopurinol⁽²⁶⁾. Therefore, the argument for the role of XOR is strengthened by another approach, namely pretreatment of animals with tungsten, which inactivates XOR by replacing molybdenum in the active site of the enzyme⁽²⁷⁾. Such animals are protected against reperfusion damage to the heart⁽²⁸⁾, intestine⁽²⁹⁾, kidney⁽³⁰⁾, and lung⁽²²⁾.

Using both direct and spin-trapping techniques of electron paramagnetic resonance spectroscopy, a burst of O₂[horizontal line over dot] generation has been shown in the first minutes after reperfusion of ischemic rat heart⁽³¹⁾ and reoxygenation of anoxic bovine⁽³²⁾ and human⁽³³⁾ endothelial cells. In each case, allopurinol or oxipurinol prevented the production of superoxide at concentrations that are too low to scavenge free radicals.

In normal adults, XOR activity is found mainly in liver and intestine, with lower levels in kidney and undetectable activities in other organs^{(12, 34, 35)}. Our findings of immunoreactive XOR in fetal and neonatal organs⁽³⁶⁾ are in accordance with the adult data. XOR activity decreases in human intestine but increases in liver during fetal development⁽³⁷⁾. Immunolocalization studies have shown that in bovine organs XOR is mainly present in capillary endothelial cells⁽³⁸⁾, and similar findings have been reported for human tissues^{(39, 40)}. Activity measurements in cells isolated from rat⁽⁴¹⁾ and bovine⁽⁴²⁾ brain capillaries as well as from cultured microvascular endothelium⁽⁴³⁾ have supported this localization. Furthermore, endothelial XOR activity can be induced by γ-interferon⁽⁴⁴⁾ and by hypoxia^(45–47). Thus, even if in whole tissue homogenates enzyme activity is low, XOR may be relevant, because capillary damage and increased permeability are important in the pathogenesis of reperfusion injury.

The conversion of XOR_D to XOR_O is initially reversible and occurs through sulfhydryl group oxidation, but then becomes irreversible, presumably through proteolysis^{(13, 14, 48)}. The hypothesis that XOR is the prime source of reactive oxygen metabolites after reperfusion was originally based on the notion that in ischemic organs the conversion of XOR_D to XOR_O occurs within minutes⁽²⁾. In subsequent studies, the time scale has turned out to be hours rather than minutes both in ischemic rat liver and intestine^{(49, 50)}. In the liver it has also been shown that the rates of conversion depend on the nutritional status⁽⁵¹⁾ and differ greatly between hepatocytes, Kupffer cells, and endothelial cells⁽⁵²⁾. In any case, the rate of conversion appears too slow to account for the burst of radical production and the rapid onset of postischemic damage. However, a minimum of 10-15% of total XOR activity is always in the O-form even in well oxygenated tissues. This is sufficient for rapid generation of oxidants, if the accumulated hypoxanthine and xanthine are metabolized through the oxidase and not the dehydrogenase reaction. This is very likely to happen in ischemic cells, which have elevated levels of NADH, a strong inhibitor of XOR_D but not of XOR_O⁽⁵³⁾.

Inflammatory cells. Ischemia and reperfusion induce an inflammatory response in many tissues, including the brain⁽⁵⁴⁾. The expression of a number of cytokines is elevated, including IL-1, IL-6, transforming growth factor-β, and fibroblast growth factor. Some of these changes are rapid, e.g. the mRNA for IL-1β is increased within 15 min of forebrain ischemia in the rat⁽⁵⁵⁾. The primary cellular origin of the cytokine response appears to be activated microglia^(56–58). The pathogenic role of IL-1 is suggested by the beneficial effects of IL-1 antibody⁽⁵⁹⁾, as well as IL-1 receptor antagonist administration⁽⁶⁰⁾ or gene transfer⁽⁶¹⁾ on postischemic brain damage. On the other hand, TNF appears to be neuroprotective, because focal ischemic damage was exacerbated in mice lacking TNF receptors⁽⁶²⁾.

Cytokines presumably mediate the influx of inflammatory cells into reperfused tissue (Fig. 2). Through increased expression of P- and E-selectins and ICAM-1 on endothelial cells^{(63, 64)}, and of integrins on leukocytes⁽⁶⁵⁾, neutrophils adhere to the endothelium, extravasate through the vessel wall, and accumulate in the interstitium^{(6, 24, 66, 67)}. The neutrophils are activated by the inflammatory mediators (cytokines, platelet-activating factor) released during ischemia, and they produce primarily O₂[horizontal line over dot], secondarily other oxygen metabolites, and hypochlorous acid. That neutrophils are involved in reperfusion damage is suggested by indirect evidence, based on attenuation by granulocyte depletion of cardiac⁽⁶⁸⁾ and intestinal⁽⁶⁹⁾ injury. Similar protection is obtained by blocking neutrophil adhesion to endothelium with antibodies to either neutrophil^(69–71) or endothelial cell⁽⁷²⁾ adhesion molecules. Transgenic mice deficient in ICAM-1 have smaller infarct volumes as well as better survival and neurologic function than do wild type animals after transient cerebral ischemia and reperfusion⁽⁷³⁾.

Figure 2

Reperfusion injury to the brain. Ischemia and reperfusion induce both rapid and delayed changes in gene expression. Reactive oxygen metabolites are generated in endothelial and parenchymal cells. They damage cells and promote the expression of adhesion molecules on endothelial cells. Granulocytes accumulate, adhere to and injure endothelial cells, and invade into the parenchyma. Free radicals together with glutamate-induced excitotocity and cytokines from activated microglia cause neuronal damage (see text). Abbreviations: NGF, nerve growth factor; BDNF, brain-derived neurotrophic factor; TGF, transforming growth factor;PAF, platelet-activating factor; ONOO^-, peroxynitrite.

Neutrophil accumulation in ischemic tissues can be prevented by allopurinol⁽⁷⁴⁾. Thus it has been hypothesized that XOR_O-derived reactive oxygen metabolites are important for the recruitment of neutrophils⁽⁶⁾. A potential mechanism for this is up-regulation of adhesion molecules on endothelial cells^{(75, 76)}.

CEREBRAL ISCHEMIA-REPERFUSION IN PERINATAL ASPHYXIA

Animal studies have characterized the hemodynamic changes triggered by asphyxia, i.e. bradycardia, hypotension, and redistribution of blood flow, and documented the changes in cerebral circulation after resuscitation. Also the mechanisms of brain damage after asphyxia, i.e. hypoxic-ischemic encephalopathy, have been studied in experimental animals as well as cultured cells of neural origin. Many of the animal models are unsatisfactory, either because they employ adult rather than immature animals⁽⁷⁷⁾, or because they involve ligation of vessels, which results in pathophysiology equivalent to stroke and different from that after global ischemia⁽⁷⁸⁾.

In animal models, resuscitation after mild to moderate ischemia is accompanied by an initial period of hyperemia, but a severe insult is followed by hypoperfusion (no reflow phenomenon)^{(24, 79, 80)}. The latter has been ascribed to endothelial injury and swelling^{(81, 82)}, to granulocyte plugging of microvessels^{(73, 83, 84)}, or to intravascular clotting⁽⁸⁵⁾. The role of granulocytes is supported by the finding of markedly higher cerebral blood flow after reperfusion in transgenic mice deficient in ICAM-1⁽⁸⁶⁾. The microvascular injury is associated with disruption of the blood-brain barrier, secondary brain edema, and possibly secondary neuronal loss⁽²⁴⁾.

When cerebral blood flow is compromised, brain cells are deprived of oxygen and glucose, resulting in a decrease in high energy phosphate (ATP) levels. Ionic gradients across the cell membrane cannot be maintained, partly due to ATP depletion and loss of Na⁺/K⁺-ATPase activity. Influx of Na⁺ and Ca²⁺ into the cell is accompanied by water, resulting in cytotoxic edema^{(78, 87)}. These changes may become irreversible and lead to cell necrosis and brain infarction as a result of ischemia alone. Reperfusion may temporarily correct the energy failure, but it may trigger a chain of events leading to delayed neuronal death^{(78, 88)} or secondary damage due to brain swelling.

In a newborn piglet model of hypoxic-ischemic encephalopathy, temporary occlusion of the carotid arteries and hypoxic ventilation resulted in“primary” energy failure, documented in MRS as a rapid decrease in the ratios of PCr to P_i and of nucleoside triphosphates to the exchangeable phosphate pool. These changes were equally rapidly normalized upon resuscitation⁽⁸⁹⁾. Within several hours, the piglets then developed “secondary” energy failure, as shown by a progressive decrease in high energy phosphate levels. The outcome, in terms of brain damage or death, was correlated with the severity of the primary energy failure. The secondary failure was not associated with a decrease in intracellular pH nor recurring hypoxia or ischemia. Its delayed onset and gradual progression suggest a process triggered by the primary insult, possibly related to excitotoxicity, reactive oxygen metabolites, or cytokine release⁽⁸⁹⁾.

The role of reperfusion in the alterations of energy metabolism in the brain after asphyxia is not clear. Both spin-trapping^{(90, 91)} and direct⁽⁹²⁾ electron paramagnetic resonance measurements have indicated that reactive oxygen metabolites are formed in the first minutes of reperfusion after cerebral ischemia. Some of these metabolites may arise in brain parenchyma, because increased amounts are recovered by microdialysis from the striatum⁽⁹⁰⁾. However, the main site of their generation appears to be the endothelial cell (Fig. 2). Reduction of their formation by oxipurinol implicates XOR_O as one source⁽⁹¹⁾, and a similar effect of indomethacin suggests prostanoid synthesis as another⁽⁹⁾. Free radicals are probably the major cause of endothelial damage, increased permeability, and brain edema after asphyxia. This conclusion is supported by the finding of reduced infarct volume and neurologic deficits after reperfusion in transgenic mice overexpressing human SOD⁽⁹³⁾.

In human infants, the transition from fetal to newborn life is normally associated with an immediate decrease in cerebral blood flow velocity, followed by an increase above the fetal level by the age of 24 h, as indicated by the Doppler flow resistance index⁽⁹⁴⁾. In asphyxiated infants, the changes in cerebral blood flow have not been clearly documented, and it is unknown whether a “no reflow phenomenon” exists. This is because of methodologic and ethical constraints in the acute stage immediately after birth, and because global cerebral blood flow measurements do not reveal possible flow changes in critical regions.

In infants developing brain injury after severe asphyxia, normal resistance indices in major arteries were generally found in Doppler studies during the first hours of life⁽⁹⁵⁾. Later on, usually on the 2nd d of life, a decrease in the resistance index, i.e. an increase in diastolic flow, was observed⁽⁹⁶⁾. Increased global cerebral blood flow, in combination with decreased cross-brain oxygen extraction, was shown on the 1st d after severe asphyxia using the Kety-Schmidt method⁽⁹⁷⁾. This is in agreement with the finding of increased cerebral blood volume at an equivalent age using near-infrared spectroscopy⁽⁹⁸⁾. A 4-h recording in asphyxiated infants using the latter method showed a decrease in cerebral blood volume between 2 and 12 h of age, which was interpreted as hypoperfusion⁽⁹⁹⁾. However, this decrement could represent normalization of hyperperfusion, because in another study, using a xenon-133 technique, hyperemia with abolished autoregulation was observed at the same ages⁽¹⁰⁰⁾. Nevertheless, toward the end of the 1st wk of life, positron emission tomography has revealed hypoperfusion in the parasagittal areas⁽¹⁰¹⁾, and preliminary studies using single-photon emission tomography suggest similar findings⁽¹⁰²⁾. Microvascular abnormalities have not been documented in human infants, but cerebral edema is a well known complication of asphyxia.

In newborn infants studied using MRS after resuscitation from asphyxia at the age of 2-4 h⁽¹⁰³⁾, the cerebral energy status was usually normal, as judged by the PCr/P_i ratio or the ratio of ATP to total phosphate. In babies with a good outcome, the PCr/P_i remained in the normal range, whereas in babies subsequently developing brain damage the ratio began to decrease after a delay of 12-24 h⁽¹⁰⁴⁾. A subnormal ATP/P_i ratio predicted fatal outcome⁽¹⁰⁴⁾. The similarity of these findings with those obtained with the piglet model⁽⁸⁹⁾ suggests that a primary energy failure occurs in the acute stage also in the human infant, followed by rapid normalization upon resuscitation.

HYPOXIC-ISCHEMIC CELL DAMAGE

Severe asphyxia results in typical neuropathologic changes, which correlate with neurologic abnormalities upon follow-up⁽¹⁰⁵⁾. The localization and extent of the pathology depend not only on the extent and duration of the insult but also on the maturity of the brain. In term infants the parasagittal parietal cortex and hippocampus are the most vulnerable, in preterm infants the periventricular white matter. Early MRI studies of term asphyxiated infants have shown specific patterns of brain damage localized to areas undergoing active myelinization^{(106, 107)}. Cell damage may occur both during the ischemic phase and upon reperfusion, probably by different mechanisms.

One mechanism of neuronal death after hypoxia-ischemia appears to be excitotoxicity, triggered mainly by elevation of extracellular glutamate concentrations. This is followed by influx of Ca²⁺ and Na⁺ through channels associated with NMDA- and other glutamate-activated receptors^(108–110). Subsequent intracellular events are poorly characterized but may include activation of lipases, proteases, endonucleases, and nitric oxide synthase⁽¹⁰⁸⁾. Reactive oxygen metabolites may be directly involved in glutamate release⁽¹¹¹⁾ but, more importantly, they may participate in the excitotoxic process itself. O₂[horizontal line over dot] production has been demonstrated after exposure of cultured cerebellar granule cells to NMDA⁽¹¹²⁾. SOD was neuroprotective both in an in vivo model of transient ischemic injury⁽¹¹³⁾ and in cultured neural cells subjected to hypoxia and reoxygenation, under depolarizing conditions to allow enzyme entry⁽¹¹⁴⁾. Overexpression of human SOD ameliorated the neurologic sequelae of transient focal ischemia in transgenic mice⁽⁹³⁾, but this may partly be accounted for by microvascular effects. However, inasmuch as ischemia alone seems sufficient to induce glutamate neurotoxicity, the role of reperfusion in this context is unclear.

Increase of intracellular Ca²⁺ levels activates nitric oxide synthase, and inhibitors of this enzyme protect cultured cortical neurons against NMDA-induced damage⁽¹¹⁵⁾. The neurotoxicity of NO may partly be ascribed to peroxynitrite⁽¹¹⁶⁾, which is formed in a reaction of NO with O₂[horizontal line over dot]⁽¹¹⁷⁾.

Not only the extent but also the mechanism of postischemic neuronal damage may depend on the intensity of the insult. Although exposure of cultured cortical cells to high concentrations of NMDA or peroxynitrite results in necrosis, less intense exposure triggers delayed neuronal death through apoptosis^{(118, 119)}. This is characterized by cell shrinkage, membrane blebbing, chromatin condensation, and DNA fragmentation due to endonuclease activation, presumably by elevated Ca^2+(120). The typical “ladder” in DNA electrophoresis has been demonstrated in models of focal cerebral ischemia^(120–123). In rats, after transient occlusion of the middle cerebral artery followed by reperfusion, the number of apoptotic cells in striatum and cortex was directly proportional to the duration of ischemia⁽¹²⁴⁾. In newborn piglets, the severity of the initial ischemic insult, as judged by high energy phosphate depletion in MRS, was correlated with the proportion of apoptotic cells in the cingulate sulcus studied at autopsy after 48 h⁽¹²⁵⁾. Glutamate-induced death of cultured cortical neurons was largely prevented by cycloheximide, indicating the dependence of programmed cell death on protein synthesis in this model⁽¹²⁶⁾. Partial prevention of apoptosis by inhibition of RNA synthesis with actinomycin D, as well as by inhibition of endonuclease with aurintricarboxylic acid, was shown in cultured sympathetic neurons exposed to hypoxia for 24 h and then reoxygenated for 2 h⁽¹²⁷⁾. Increased expression of bcl-2, a known inhibitor of apoptosis and oxidant stress⁽¹²⁸⁾, also partially prevented glutamate-induced death of cultured neurons^{(129, 130)} and ameliorated tissue damage after focal cerebral ischemia in rats⁽¹³¹⁾.

Ischemia and reperfusion induce both rapid and delayed changes in gene expression (Fig. 2). The transcription of several members of the c-fos and c-jun proto-oncogene families is increased in postischemic brain⁽¹³¹⁾, heart⁽¹³²⁾, and kidney⁽¹³³⁾. The earliest event may be phosphorylation of the transcription factors c-jun and ATF-2 by stress-activated protein kinases⁽¹³⁴⁾, detectable after 5 min of reperfusion of ischemic kidney. This is followed by altered binding of these and other transcription factors, including AP-1, to specific promoters. A similar chain of events occurs in cultured renal epithelial cells, when ATP levels are repleted after chemical anoxia^{(134, 135)}. The significance of the resulting pattern of gene activation is poorly understood and does not necessarily result in expression of the corresponding protein. However, one end result appears to be programmed cell death, for which induction of c-jun has been considered necessary⁽¹³⁶⁾. Other induced proteins such as vimentin⁽¹³³⁾ and nerve growth factors⁽¹³⁷⁾ may play a role in tissue regeneration and repair after ischemia.

CONCLUDING REMARKS

Out of the complex phenomena that are triggered on the cellular level by perinatal asphyxia, some general principles concerning reperfusion injury are beginning to emerge. Early on, a burst of reactive oxygen metabolites is generated, mainly by the activity of xanthine oxidase in endothelial and, depending on the tissue, in parenchymal cells. These metabolites may directly damage cells, but they also promote the expression of adhesion molecules on endothelial cells. This leads to accumulation of granulocytes in the reperfused tissue, further circulatory disturbances, and amplified cell damage by proteases and oxidants. In the brain, and probably in other tissues as well, reperfusion after severe asphyxia may result in early necrotic cell death. Even if the energy status is initially corrected, a progressive secondary energy failure may develop, leading to neuronal loss through apoptosis.

Animal studies have indicated that after short-term intestinal or renal ischemia, reperfusion is the main mechanism of injury, but direct ischemic organ damage becomes more important as a function of ischemia time⁽¹³⁸⁾. Before the dominance of the latter process, there is a “therapeutic window” for potential intervention⁽¹³⁸⁾. A similar window of opportunity is suggested by the time course of changes in high energy phosphate levels in piglet⁽⁸⁹⁾ and human infant⁽¹⁰⁴⁾ brain after an asphyxial insult. Several potential approaches to intervention can be envisaged based on increasing information on the pathogenic mechanisms of reperfusion injury. Prevention of reactive oxygen metabolite formation, strengthening of cellular antioxidant defenses, amelioration of microvascular injury, blocking of Ca²⁺ entry into cells, and arrest of the process of apoptosis have all been tested in vitro and in animal models, but still await confirmation and clinical trials in asphyxiated human infants.