Amino Acid and Neurotransmitter Synthesis

Glucose, through pyruvate and other TCA cycle intermediates, is an important source of carbon backbones in the synthesis of a number of vital amino acids and neurotransmitters. Of these, glutamate assumes particular prominence as an amino acid in that it not only serves as the predominant excitatory neurotransmitter but is also a precursor of GABA, the major inhibitory neurotransmitter. One of the main biochemical routes to glutamate is linked to the oxidative metabolism of glucose by the TCA cycle intermediate α-ketoglutarate, which can accept amino groups from amino acids in a reductive transamination reaction that uses NADPH:

(6)

Glutamate can be converted back to α-ketoglutarate by oxidative deamination to undergo further oxidation in the TCA cycle for the purpose of energy generation³⁵ or redirected toward the synthesis of other amino acids, GABA, fatty acids, or glutathione.

Glycerol Synthesis

New evidence indicates that glucose undergoes conversion to free glycerol within the brain. In theory, this could occur through reduction of the glycolytic intermediate dihydroxyacetone phosphate to glycerol 3-phosphate by glycerol-3-phosphate dehydrogenase, followed by dephosphorylation to glycerol by a phosphatase. The natural operation of this pathway is suggested by the presence of the necessary enzymes in the brain. Glycerol 3-phosphate has been demonstrated in astrocytes and neurons,³⁶ as well as in the cytosol of oligodendrocytes.³⁷ More importantly, it has recently been reported that ¹³C-labeled glycerol accumulates in the media of cultured astrocytes and cerebellar granule cells and in the brains of rats within 5 minutes after treatment with [U-¹³C] glucose.³⁸

Pentose Phosphate Pathway

Up to 10% of glucose uptake goes toward the production of reducing power for a number of biosynthetic reactions in which the precursors are in a more oxidized state than the products. This reducing power takes the form of NADPH, which is produced when phosphorylated glucose (glucose 6-phosphate) from the first step of the glycolytic pathway is diverted to an alternative pathway called the pentose phosphate shunt:

(7)

NADPH is necessary for the synthesis of lipids and the excitatory neurotransmitter glutamate (which is coupled to synthesis of the neurotransmitters glutamine and GABA). It is also crucial for the scavenging of reactive oxygen species (ROSs) by maintenance of glutathione in a reduced state. The other major product of the pentose shunt is ribose 5-phosphate, a vital building block for essential biomolecules such as ATP, CoA, nucleotides, and nucleic acids.

Storage as Glycogen

The brain converts a limited amount of glucose into glycogen to form its principal energy reserve. Although there is typically 3 to 4 times more glycogen than free glucose in the brain, it still amounts to no more than approximately 4 µmol/g, and were it to serve as the sole fuel source, it would be consumed completely in no more than a few minutes. Instead, more contemporary evidence suggests that it exists as a metabolic buffer system and is metabolized slowly, so complete turnover of brain glycogen stores normally takes 3 to 5 days.³⁹

Glycogenesis requires the action of glycogen synthase on glucose subunits that have first undergone phosphorylation by ATP. Although both astrocytes and neurons possess the necessary enzymes, synthesis of glycogen is normally confined to astrocytes. Storage of glycogen is also almost exclusive to astrocytes and serves to augment neuronal energy requirements during periods of intense activity and pathologic shortages of glucose. Glycogen undergoes glycogenolysis by the enzyme glycogen phosphorylase to form lactate under stimulation by neurotransmitters such as norepinephrine, vasoactive intestinal polypeptide (VIP), histamine, serotonin, and certain metabolic by-products of neuronal activity such as K⁺ and adenosine.⁴⁰ It has been hypothesized that this lactate is transferred from astrocytes to neurons for use as an energy substrate.^41,42 In contrast to the metabolism of glucose, release of glucose equivalents from glycogen does not require prior “priming” with ATP.

Neurons

The contribution of neurons to cerebral metabolism is reflected by four features. First, they are quantitatively the main site of energy (ATP) generation, whether by oxidation of glucose, lactate, or, during hypoglycemia, ketone bodies. Second, as stated earlier, neurons are the greatest consumers of ATP in the brain, largely because of the unrelenting activity of the Na⁺,K⁺-ATPase pumps that maintain the transmembrane ion gradients. Third, neurons control the turnover of carbohydrates by astrocytes, including mobilization of the principal cerebral carbohydrate reserve, the glycogen store in astrocytes. Finally, neurons facilitate the coupling of blood flow to brain metabolism. Changes in local levels of metabolic by-products, including ions such as H⁺ and K⁺ and molecules such as adenosine, occur during neuronal activity, and the role of each of these in flow-metabolism coupling is elaborated on later.

Astrocytes

A growing body of research has resulted in a revision of the traditional view of astrocytes as a passive supporting act to neurons. The proliferation and development of astrocytes are indeed driven by trophic cues associated with neuronal activity, but the growth and survival of neurons are in turn dependent on astrocytes.⁴³ In terms of metabolism, astrocytes form a highly active compartment that is separate from that of neurons but with which it interacts in a dynamic and essential fashion. This interaction occurs via the narrow extracellular space (ECS) and involves the shuttling of ionic and molecular metabolites and neurotransmitters such as lactate and glutamate. Despite being essentially nonexcitable, astrocytes have also been implicated in the integration of neuronal input, modulation of synaptic activity, and even long-range signaling. Neurotransmitter-evoked elevations in astrocytic calcium can trigger the release of chemical transmitters that can cause sustained modulatory actions on neighboring neurons. Astrocytes are also involved in brain water homeostasis and induction of the BBB (see review by Bennaroch⁵).

Before considering the influence of astrocytes on cerebral metabolism, it is relevant to consider their structural relationship to each other and to other cells in the brain. In the mammalian brain, neurons make up no more than 50% of the cerebral cortical volume and in most regions are outnumbered 10 : 1 by astrocytes.⁴⁴ Each astrocyte typically defines a nonoverlapping three-dimensional domain and is polarized such that one or more astrocytic processes contact a capillary while many more are entwined within the neuropil and engage with hundreds to thousands of synapses. Where astrocytic processes abut capillaries, they are specialized into structures known as end-feet. These end-feet are so numerous that almost the entire surface of the capillaries is covered.⁴⁵ The potential significance of this interposition of astrocytes between neurons and the capillary (endothelium) has been recognized since the late 19th century.⁴⁴ It puts astrocytes in a special position to both sense neuronal signaling and capture glucose directly from the capillary, thereby permitting them to govern excitation-metabolism coupling. Another noteworthy feature of astrocytes is their ample connections with each other and the ECS through gap junctions and hemichannels composed mainly of connexin.^43,46,47

As alluded to before, astrocytes play a leading role in the flux of glucose into neurons for energy. The glucose taken up by astrocytes may have one of two primary fates: it may be converted to lactate via astrocytic glycolysis, or it may be converted via glycogenesis to the glucose storage polymer glycogen. By contrast, in adult neurons, aerobic glycolysis results in the formation of pyruvate, not lactate, and glycogen metabolism and storage normally do not occur. That glycogen, the storage form of glucose, is located almost entirely in astrocytes is one indication of the dominant position of astrocytes in the metabolic processing of glucose. However, it should be noted that glycogenolysis in astrocytes is dictated by specific neurotransmitters and neuromodulators, thus giving neurons tight rein over this energy store.⁴⁸ As elaborated on earlier, astrocytic glycolysis is also stimulated by neuronal activation, which leads to the production of lactate. It has been proposed that this lactate is secreted into the ECS and is taken up and oxidized by activated neurons as their main energy source. Notably, 1 molecule of lactate entering the TCA cycle to undergo oxidation can yield 17 ATP molecules under normoxic conditions, which is about half the energy production of aerobic glucose metabolism. Although controversial, the existence of astrocyte-neuron lactate shuttling is based on evidence. For instance, immunohistochemistry has identified a selective distribution of LDH isoforms, with neurons predominantly expressing LDH1, the form that is enriched in lactate-consuming tissues, and astrocytes expressing LDH5, the form that is enriched in lactate-producing tissues.⁶ Localization of the lactate monocarboxylate transporters MCT-1 and MCT-4 to astrocytes and MCT-2 to neurons demonstrates the capacity for lactate exchange between these cells.⁴⁹ When primary neuronal cultures are in the presence of lactate and glucose, they preferentially consume lactate as their oxidative substrate.

The fidelity and safety of glutamate-mediated neurotransmission are dependent on very efficient uptake and modification of glutamate by astrocytes (Fig. 343-2). They keep the extracellular glutamate concentration very low by the rapid Na⁺-dependent GLT1 (predominant in the cortex and hippocampus) and GLAST (predominant in the cerebellum) cotransporter-mediated removal of the neurotransmitter from the synapse, thereby enabling quick termination of glutamate signaling before excitotoxic neuronal injury can occur.⁵⁰ Astrocytes can handle this influx of glutamate in several different ways. They possess the necessary aminotransferases to transfer the α-amino group of glutamate to oxaloacetate or pyruvate to yield α-ketoglutarate and either aspartate or alanine. The resultant α-ketoglutarate can be oxidized for energy metabolism in the TCA cycle. Indeed, cultured astrocytes have been reported to use glutamate as an energy substrate even in the presence of glucose.⁵¹ Glutamate can also be converted into α-ketoglutarate by glutamate dehydrogenase–catalyzed oxidative deamination. However, the most common means by which astrocytes inactivate glutamate is by conversion to glutamine through the attachment of ammonium ions (see Fig. 343-2). This last reaction is endergonic and requires glutamine synthase, an enzyme limited to astrocytes, and serves the additional important function of removing ammonia.⁵²

FIGURE 343-2 Functions of glutamate in the brain. Glutamate, synthesized from recycled astrocytic glutamine or neuronal α-ketoglutarate (α-KG), undergoes neuroglial uptake and participates in the glutamine/glutamate cycle between neurons and astrocytes. It is involved in normal excitatory neurotransmission (and therefore ionic homeostasis and the pathogenesis of excitotoxicity); ammonia metabolism; the biosynthesis of proteins, fatty acids, and glutathione (GSH, an antioxidant); astrocytic energy metabolism; and the synthesis and actions of neurotransmitters such as glutamine and γ-aminobutyric acid (GABA).

(8)

Being electrically “inert,” the glutamine can then be safely released to neurons for recycling into glutamate by neuronal mitochondrial glutaminase. This cycling between glutamine and glutamate is commonly referred to as the glutamine-glutamate shuttle (see Fig. 343-2). The sodium that is cotransported into astrocytes during the uptake of glutamate (glutamate/Na⁺ ratio, 1 : 3) stimulates the activity of Na⁺,K⁺-ATPase, which depletes intracellular ATP and in turn stimulates PFK, the principal rate-limiting enzyme of glycolysis.⁵³ The resulting net production by glycolysis of 2 molecules each of ATP and lactate per molecule of glucose would in theory supply enough energy for astrocytic uptake (1 ATP for the glutamate transporter) and conversion (1 ATP to drive amidation) of 1 molecule of glutamate to glutamine, thus leaving lactate in surplus of requirements.⁵⁴ Herein lies one possible explanation for the well-documented early rise in lactate in cerebral tissue during cortical activation, as occurs in seizures.^55,56 More importantly, this hypothesized stimulation of glycolysis by glutamate provides a mechanism for the observed coupling of excitation and metabolism in the brain. Cortical activation causes rapid glutamatergic signaling, which increases the glutamine/glutamate flux and therefore drives the increased astrocytic consumption of glucose by glycolysis. This logic is supported by evidence of coupling of the glutamate/glutamine shuttle to glucose energetics in the cerebral cortex in vivo.^57,58

Astrocytes are capable of long-range signaling and buffering functions through the use of their gap junctions to form a large intercellular network, or syncytium. For example, increases in the extracellular K⁺ concentration secondary to neuronal activity leads to the entry of K⁺ into astrocytes through strong inwardly rectifying K⁺ (K_ir) channels.^59,60 The resulting local depolarization is propagated electrotonically through the glial cell network via the gap junctions, which leads to the efflux of K⁺ at distant cell processes that are not experiencing the elevated K⁺ concentration. The high density of K⁺ channels on astrocytic end-feet allows K⁺ to be deposited in the perivascular space, where it can be recycled when neural activity ceases. This so-called spatial buffering of K⁺ is important because even modest efflux of K⁺ from neurons can lead to considerable changes in the concentration of K⁺ in the ECS, with potentially detrimental effects on maintenance of neuronal membrane potential, activation and inactivation of voltage-gated channels, synaptic transmission, and electrogenic transport of neurotransmitters. The gap junctions of astrocytes also contribute to the propagation of intercellular Ca²⁺ waves, probably by enhancing the release of ATP, as well as by providing an intercellular pathway. Astrocyte membrane depolarization by glutamate causes the mobilization of Ca²⁺ from the endoplasmic reticulum via the inositol triphosphate (IP₃) generated by the activation of metabotropic glutamate receptor 5.⁶¹ It has previously been thought that IP₃, Ca²⁺, or both are propagated across gap junctions to create a “Ca²⁺ wave.” More contemporary evidence suggests that the Ca²⁺ wave is propagated through the paracrine actions of ATP released from astrocytes on purinergic receptors.⁶² Release of glutamate from astrocytes has been associated with this calcium wave. Taken together, it is clear that an increase in neuronal depolarization is coupled to an increase in astrocytic depolarization, which in turn influences local metabolic and electrical activity via K⁺, H⁺, and Ca²⁺ ions.

Astrocytes are also involved in brain water homeostasis. They take up K⁺ through transporters, particularly Na⁺,K⁺-ATPase and Na⁺/K⁺/2Cl⁻ transporters. The resulting net ionic gain results in osmotic water uptake and slight swelling of the cell. Redistribution of this water is facilitated by the high density of aquaporin channels on perivascular end-feet. This excess metabolic water probably joins interstitial fluid in the perivascular spaces and is cleared in cerebrospinal fluid (CSF) or the lymphatics. There are separate mechanisms to clear metabolic water from neurotransmitter activation or ion uptake, some of which involve elevations in intracellular Ca²⁺ [Ca²⁺]_i.

Finally, yet another distinct role of astrocytes in cerebral metabolism is regulation of the development and function of the endothelial cells of the BBB. Astrocytes have important influences on the BBB, which plays a crucial role in cerebral metabolism (see later). First, contact of astrocytic foot processes with endothelial cells of the BBB upregulates the formation of tight junctions in the latter, mainly by inducing the production of occludin, an integral protein of tight junctions.^63,64 Second, astrocytes can also induce the expression in endothelial cells of membrane transporters and specialized enzymes such as γ-glutamyl transpeptidase.^65,66 Astrocytes submitted to hypoglycemic conditions may release factors that increase glucose uptake through the BBB.⁶⁷

Blood-Brain Barrier

The BBB is a protective structure formed by capillaries to restrict the exchange of solutes between the brain and blood and thus to ensure a chemically controlled intracerebral milieu for optimal cellular performance. Morphologically, the BBB is composed of a monolayer of specialized capillary endothelial cells surrounded by a thin basal lamina and closely invested by the foot processes of astrocytes and cells known as pericytes. Unlike endothelial cells elsewhere in the body, those of the BBB are characterized by specialized regions of circumferential intercellular contact known as tight junctions or zonula occludens, minimal pinocytotic activity, and the virtual absence of fenestrations, all of which combine to necessitate that molecules passing across the BBB take a transcellular rather than a paracellular route. Consequently, free diffusion through an intact BBB is limited to lipid-soluble substances such as CO₂, O₂, ethanol, and lipophilic drugs or to very small polar molecules with a radius of less than 0.8 nm.⁶⁸ Passage of even the smallest charged molecules (i.e., inorganic ions) is severely restricted, so transendothelial electrical resistance, which is typically 2 to 20 Ω•cm², can be increased to greater than 1000 Ω•cm².⁶⁹ Hence, ions are transported by channels or by active means, such as via Na⁺,K⁺-ATPase.⁷⁰ Transport of ions (particularly Na⁺) is coupled to the obligatory flow of water via osmotic forces, and both the luminal Na⁺ transporter and abluminal Na⁺,K⁺-ATPase are implicated in the secretion of extracellular fluid by brain capillaries.^68,70 There are specific transport systems for the transcellular traffic of small lipophilic nutrient and amino acid molecules across the BBB. For example, glucose and neutral amino acids are transferred by the GLUT1 and L-amino acid transporters, respectively. Large hydrophilic proteins such as insulin, transferrin, and insulin-like growth factor may be taken up by a saturable receptor-mediated transcytosis mechanism, whereas others, particularly polycationic proteins, may cross the BBB via a nonspecific, non–receptor-mediated process referred to as adsorptive transcytosis. The BBB also has active mechanisms for excluding potentially harmful substances from the brain. Physiologically expressed carriers such as P-glycoprotein actively transport lipophilic molecules across the BBB and out of the brain. In addition, a metabolic barrier is provided by a combination of intracellular and extracellular enzymes: ectoenzymes such as peptidases and nucleotidases for the degradation of peptides and ATP, respectively, and intracellular enzymes such as monoamine oxidase and cytochrome P-450 for the inactivation of neuroactive and toxic compounds.⁷¹ Finally many, if not most of the features of the BBB are dynamic. For instance, during hypoglycemia, upregulation of GLUT1 and MCT transporter expression allows transport of glucose and ketone bodies into the brain to be increased.⁷²

Major Mediators

Cerebral Autoregulation

The brain maintains its regional and total blood flow (and therefore perfusion) at a fairly constant rate over a wide range of systemic blood pressure by a vasomotor phenomenon known as cerebral autoregulation. Autoregulation involves the rapid and sustained constriction or dilation of cerebral resistance vessels, principally the precapillary arterioles, purely in response to changes in transmural pressure caused by variations in CPP. Recall that in normal individuals, CPP varies directly with MAP because ICP is constant. In effect, the brain is protected from fluctuations in perfusion over a MAP range of 60 to 150 mm Hg (Fig. 343-3). Beyond these limits, CPP passively follows MAP. It is important to note that the limits of autoregulation are by no means invariable. For example, they may be right-shifted in chronic hypertension (associated with increased sympathetic tone) and in states associated with increased renin release; conversely, a left shift may be observed in sleep, “physiologic hypotension” in athletes, “pathologic hypotension” associated with hemorrhage, and the presence of angiotensin-converting enzyme inhibitors, prolonged hypoxemia, or hypercapnia.^192,193 Furthermore, autoregulation of CBF may be disturbed in acute, severe hypotension or hypertension and profoundly impaired or even abolished in severe cerebral ischemia, arteriovenous malformation beds, or brain injury and after aneurysmal SAH.⁷³

FIGURE 343-3 Classic cerebral autoregulation curve. Cerebral autoregulation maintains a relatively constant rate of cerebral blood flow (CBF) across a wide range of cerebral perfusion pressure (CPP) as shown (plateau). The normal pressure limits of autoregulation (vertical broken lines) may be shifted to the left or right in certain conditions (arrows) and impaired or abolished in others (see text for details). Note the respective relatively steep rise and fall in CBF as CPP changes above and below the normal limits of autoregulation.

Regulation of Cerebral Blood Flow by Blood Gases

Carbon Dioxide

CO₂ is considered to be the most important physiologic variable in chemoregulation. The arterial partial pressure of CO₂ (PaCO₂) exerts profound effects on CBF, particularly across the physiologic range of 30 to 50 mm Hg (Fig. 343-4). In mammals of all ages, hypercapnia (increased PaCO₂) is found to cause cerebral vasodilation, whereas hypocapnia causes the reverse. In fact, inhalation of 5% to 7% CO₂ is associated with an almost exponential increase in CBF of 50% to 100%, thus rendering CO₂ one of the most potent vasodilatory influences on the cerebral circulation.²²⁷ When alterations in PCO₂ have been sustained for 3 to 5 hours, there is an adaptive return of CBF toward baseline levels.

FIGURE 343-4 Effect of arterial partial pressure of CO₂ (PaCO₂) on cerebral blood flow (CBF). Note that there is a linear relationship between CBF and PaCO₂ across the physiologic range of PaCO₂ as marked by the vertical broken lines.

It is most likely that extracellular acidosis rather than intracellular acidosis or molecular CO₂ itself is the major determinant of hypercapnic hyperemia.^187,228,229 A direct effect of CO₂ on brain vessels has been disproved by experiments showing that manipulations of PCO₂ are not vasoactive when the pH of CSF is kept constant. The “pH hypothesis” of hypercapnic vasodilation states that the actions of CO₂ on CBF are mediated by the effects of H⁺ on VSM. When PaCO₂ increases, molecular CO₂ but not HCO₃⁻ or H⁺ diffuses across the BBB. Perivascular PCO₂ rises, which leads astrocytes to convert the CO₂ to HCO₃⁻ and H⁺ via carbonic anhydrase, thereby increasing the extracellular or perivascular concentration of H⁺ (i.e., there is a fall in local pH).²²⁷ This is thought to cause cerebral vasodilation via a sequence of effects on VSM cells consisting of increased potassium conductance, increased K⁺ efflux, hyperpolarization, closure of VGCCs, and reduced cytosolic Ca²⁺, a process culminating in relaxation.²³⁰ The exact mechanism of increased potassium conductance has not been resolved and may involve the opening of K_ATP or K_Ca channels^231,232 or increased H⁺/K⁺ exchange in VSM cells.^{179,233–235} The adaptation of cerebrovascular tone to hypercapnia is associated with a gradual return of CSF pH and HCO₃⁻ ion concentration toward normal values.^236,237

There is no consensus on the mechanism or mechanisms linking changes in extracellular pH to cerebral VSM tone, and some research has even suggested that there may be significant differences between adults and neonates. Prostanoids, NO, and cyclic nucleotides have all been implicated in mediation of the aforementioned changes in plasma membrane calcium and potassium conductance, but the literature is confusing. Inhibition of prostanoid synthesis by indomethacin has been found to abolish the vasodilatory response to hypercapnia in newborn pigs but not in adult humans.²³⁸ Some studies have reported that NOS inhibition significantly attenuates the vascular reactivity to hypercapnia in adult animals.^{158,239–241} Yet this effect of NOS inhibition has been disputed in newborn animals and adult humans.^96,242 Data comparing newborn and adult animals are insufficient to make any firm conclusions about possible differences in the mechanism of vascular responsiveness. Suggested explanations for the inconsistent data in adults are that NO may have a permissive role in facilitating the action of other mediators^243,244 or that multiple redundant pathways may interact in CO₂-induced vasodilation.^245–247 Alternatively, it could well reflect the variable specificity and other pitfalls of pharmacologic NOS blockade.¹¹⁶

Oxygen

Arterial oxygen is another important determinant of CVR and hence CBF. Elevated inspired oxygen concentrations elicit CVR and decrease CBF. Conversely, a fall in PaO₂ results in vasodilation. It is generally held that the response of CBF to PaO₂ is a threshold phenomenon that is evident only when PaO₂ falls below the normal physiologic range (Fig. 343-5), although some research suggests otherwise.²⁴⁸

FIGURE 343-5 Effect of the arterial partial pressure of O₂ (PaO₂) on cerebral blood flow (CBF). Note that there is a hyperbolic relationship between CBF and PaO₂. However, in the physiologic range of PaO₂, marked by the vertical broken lines, CBF remains relatively constant despite marked changes in PaO₂. Only when PaO₂ falls to levels indicative of marked hypoxia is there a compensatory (and relatively dramatic) increase in CBF.

Hypoxia is thought to elicit vasodilation by both direct effects on cerebral resistance vessels and indirect effects on nonvascular components. It has been proposed that hypoxia elicits VSM relaxation by inhibiting sarcoplasmic Ca²⁺ uptake and stimulating the production of EDRF over EDCF. Hypoxia also causes alterations in cellular metabolism that lead to increased generation and release of the vasoactive tissue factors K⁺, H⁺, and adenosine. The influence of K⁺ on CBF is discussed later in the context of flow-metabolism coupling. Cortical CBF has been shown to increase with increasing H⁺ concentration or acidity over a PaO₂ range of 40 to 300 torr (1 torr ≅ 1 mm Hg).²⁴⁸ Likewise, increasing blood flow has been seen after the infusion of increasing concentrations of adenosine or adenosine analogues into the brain.²⁴⁹ The preponderance of evidence suggests that arterial chemoreceptors do not contribute to hypoxic vasodilation because it persists after the deafferentation of cranial nerves IX and X. However, it has been proposed that hypoxia-induced stimulation of oxygen-sensing neurons in the rostral ventrolateral medulla may increase CBF via neurogenic vasodilation.^193,250 There is an overall consensus that NO signaling is probably not involved in mild to moderate hypoxia-induced vasodilation inasmuch as NOS inhibitors do not attenuate vasodilation in this setting.^251,252 However, NO signaling may be involved in severe hypoxia (PaO₂ <35 mm Hg) because the NOS inhibitor L-NAME has been found to attenuate vasodilation in these circumstances.¹⁹¹

Mechanisms

Despite much research, exactly how increased neuronal activity triggers enhanced capillary blood flow remains conjectural. Any proposed mechanism or mechanisms must be capable of inducing rapid vasodilation along the lines of what is observed in vivo and have a clearly defined site of action within the vasculature that can account for the effects of neuronal activity on the cerebral circulation. Originally, Roy and Sherrington proposed that blood flow through the microvasculature increases under the local influence of a buildup of metabolites from a relative energy deficit. Such metabolites, primarily K⁺, H⁺ (largely related to lactate and CO₂ levels), and adenosine, can affect VSM tone directly or indirectly through altered neurovascular transmission (see later). However, this theory is clearly inadequate in that it fails to anticipate any effect on the upstream resistance vessels or arterioles, an omission that goes against basic hemodynamic principles and laboratory observations in vascular beds throughout the body.²⁵⁵ The ability of functional activity to elicit coordinated changes in the precapillary resistance arterioles, as well as the microvessels in a simple model of the cerebral circulation, suggested that the vascular response must somehow be retrogradely propagated. This prompted speculation that the metabolic and electrical signals for vasodilation may be conducted upstream through gap junctions between endothelial cells or between endothelial and VSM cells or through flow-mediated vasodilation.^256–258 More recent investigations have suggested several potential mechanisms whereby synaptic activity may be incorporated as a trigger for neurovascular coupling.

The rich innervation of cortical microvessels by several neurotransmitter systems suggests that perivascularly released neurotransmitters are important in adaptations of flow to neuronal activity. Somatosensory cortical hemodynamic responses have been shown to correlate with local field potentials, thus implying that vascular changes reflect the incoming neuronal input and local processing in a given area.²⁵⁹ GABA interneurons are candidates for such a neural processing role because they provide a particularly rich innervation to microvessels. In this context, it has been of interest that selected activation of single cortical GABA interneurons can not only increase but also decrease the diameter of neighboring microvessels through the expression of VIP, NOS, and somatostatin.^260,261 This provides a plausible mechanism for the redistribution of perfusion to activated neuronal networks to occur over a backdrop of stable perfusion to the entire brain.

Through painstaking research, a picture has gradually emerged of astrocytic end-feet serving as individual vasoregulatory units through neurotransmitter-evoked Ca²⁺-dependent signaling events. It has been demonstrated that arteriolar vasodilation occurs in a similar time frame to the rise in astrocytic [Ca²⁺]_i induced by glutamatergic neuron activity or the application of metabotropic glutamate receptor (mGluR) agonists.¹³⁵ Depending on the resting state of the arteriole, similarly evoked astrocytic increases in [Ca²⁺]_i have also been found to be temporally related to vasoconstriction. However, arteriolar tone is affected by increases in astrocytic [Ca²⁺]_i only when it is large and propagates into end-feet.⁸⁵ Currently, it is thought that binding of mGluR by glutamate activates PLC and generates IP₃, which not only promotes cytosolic release of Ca²⁺ from the endoplasmic reticulum but also causes a separate, spatially restricted rise in [Ca²⁺]_i through Ca²⁺ release channels associated with end-feet IP₃ receptors. Hence, the spatially restricted liberation of IP₃ within an astrocyte end-foot has been linked to the production of a localized large-amplitude [Ca²⁺]_i signal that immediately precedes vasodilation of an approximately 30-µm segment of arteriole within a cortical slice.²⁶² Although most experimental work has been done with glutamate, it is probable that other neurotransmitters such as GABA, somatostatin, and VIP may likewise produce vasoactive effects through the generation of astrocytic Ca²⁺ signals, as well as direct innervation of vessels.^260,262 In light of the evidence for a specialized function of the end-foot in neurovascular coupling, it is of interest that the local interneurons that so richly innervate cerebral arterioles make extensive contacts with end-feet.²¹⁶

It is probable that astrocytic Ca²⁺ signals cause vasodilation through mediators that bring about alterations in Ca²⁺ signaling processes in arteriolar VSM cells. Studies on brain slices have found that induction of synaptic activity leads to an increase in astrocyte [Ca²⁺]_i, followed closely (<1 second) by the inhibition of spontaneous VSM cell [Ca²⁺]_i oscillation and arteriolar vasodilation.²⁶³ The leading candidates in mediation of this process are K⁺ and arachidonic acid metabolites. Elevation of astrocytic [Ca²⁺]_i can lead to Ca²⁺-dependent activation of PLA and liberation of arachidonic acid, which is subject to metabolism by a host of enzymes in astrocytes and endothelial cells to form vasoactive substances.²⁶⁴ Alternatively, a rise in astrocytic [Ca²⁺]_i might activate end-foot big conductance Ca²⁺-sensitive K⁺ (BK_Ca) channels and lead to the release of K⁺ into the ECS and VSM cell depolarization as a result of the activation of K_ir channels.²⁶⁵

One aspect that has been neglected in the account of the astrocytic vasoregulatory unit just presented is the responsiveness of cerebrovascular tone to the dynamic metabolic state of the local brain environment. Therefore, it is of interest that a mechanism whereby metabolites can modulate the vasomotor influence of astrocytes has been newly revealed in the rat brain.²⁶⁶ When oxygen availability is low and astrocyte [Ca²⁺]_i is elevated, astrocyte glycolysis is maximized, thereby leading to extracellular accumulation of lactate and adenosine. This in turn can attenuate transporter-mediated uptake of the vasorelaxant PGE₂ from the ECS and block astrocyte-mediated constriction, both of which facilitate vasodilation.

Inert Nondiffusible Tracer Techniques

Two primary CT-based imaging techniques are currently available as commercial products for the quantitative evaluation of CBF: Xe-CT and dynamic CTP imaging.

Single-Photon Emission Computed Tomography

SPECT determines the three-dimensional distribution of a single gamma ray–emitting radiotracer in the body. Hence, SPECT machines are designed to detect single gamma rays (photons) and determine their point of origin from their trajectory. Most modern SPECT studies use the radioisotope technetium 99m attached to a carrier that easily crosses the BBB, is distributed in the brain in proportion to regional CBF, and is temporarily retained there, thus allowing detection of the radioisotope anytime within the subsequent few hours. A variety of such carriers are in clinical use, each with its own advantages and disadvantages. The most commonly used SPECT brain blood flow tracer is technetium 99m-hexamethyl-propyleneamine oxime (^99mTc-HMPAO [Ceretec]), a lipid-soluble macrocyclic amine. It is extracted on first pass by the brain, where it becomes fixed for several hours by conversion to a hydrophilic compound in the presence of intercellular glutathione. With ^99mTc-HMPAO, CBF quantification is not straightforward, and a semiquantitative approach is generally used in which counts in a region of interest are determined and compared with those in an analogous region in the opposite, presumably normal hemisphere.²⁸⁵ By contrast, ¹³³Xe-SPECT is ideally suited for quantitative CBF determination with the inert gas clearance technique. The rapid clearance of ¹³³Xe from the brain has the advantage of allowing repeated studies within a short interval but unfortunately requires dynamic instrumentation, which has the significant disadvantage, when combined with the low energy of the emitted photons, of rendering poor spatial resolution of the resulting images. Also to be considered are the same concerns about the adverse effects of xenon inhalation that were mentioned for Xe-CT.

The main advantage of SPECT is that the necessary equipment and radiotracers are readily available in the average nuclear medicine department. It is also relatively simple to use, and even with high-resolution systems, it is considerably less expensive than PET. The disadvantage is that when compared with equivalent CT or MRI studies, the spatial resolution is inferior, although this can be partly addressed by digital superimposition on the patient’s CT and MRI data set.

Positron Emission Tomography

PET involves the production of two- or three-dimensional images of the distribution of inhaled or intravenously administered tracer compounds that have been labeled with radionuclides containing an excess of positrons. Because these tracers are taken up by and accumulate in the brain, the radionuclides spontaneously decay by positron emission. The positrons travel up to a few millimeters through tissue before eventually becoming annihilated by collision with an electron, in the process simultaneously emitting two 511-keV photons (gamma rays) traveling at 180 degrees to each other that are registered by a ring array of external detectors. The detector electronics are designed in a manner to recognize the close temporal coincidence of two detection events as the result of one annihilation event. The trajectories of each pair of photon emissions can then be used to calculate the point of origin of each annihilation event. The coincidence events are stored in arrays corresponding to projections through the patient and reconstructed with standard tomographic techniques to generate a map of radioactivity as a function of location. The more intense the radioactivity, the greater the concentration of radiotracer in an area of interest.

The type of information that can be obtained with PET relies on the radiotracer that is administered. PET can be used to study regional cerebral glucose utilization through the use of intravenous [¹⁸F]fluorodeoxyglucose (FDG). Active brain regions take up FDG as though it were glucose. Once inside the cell, it is phosphorylated by hexokinase into FDG 6-phosphate, which is not a substrate for glucose transport and cannot be metabolized by phosphohexose isomerase, the enzyme catalyzing the next step in glucose metabolism. Thus, labeled FDG 6-glucose becomes trapped within the cell and contributes to a pattern of radioactivity that reflects local glucose uptake and metabolism. ¹⁵O can be inhaled for assessment of oxygen metabolism. The most commonly used tracer for PET CBF imaging is radiolabeled water (H₂¹⁵O) because it distributes freely in blood. This tracer has a very short half-life (≈2 minutes), so a bolus injection provides a snapshot that can be repeated, if desired, every 12 to 15 minutes.²⁸⁶

The principal disadvantage of PET is the need for an expensive on-site cyclotron. Positron-emitting radionuclides are produced in these cyclotrons, and because those in common use have a half-life ranging from mere minutes to just under 2 hours, they must be produced in close proximity to the PET scanner.

Magnetic Resonance Imaging and Spectroscopy

Clinical MRI uses the magnetic properties of hydrogen and its interaction with a large external magnetic field and radio waves to produce highly detailed images of the human brain. Subatomic particles such as protons behave like a spinning charge and induce microscopic loops of electric current. The nucleus of hydrogen contains a single proton and is described as having a half-integer spin. As a result, it has a very small magnetic field or magnetic moment that aligns itself either parallel or antiparallel to the direction of a strong external magnetic field with a circular oscillation of a certain frequency. When electromagnetic energy in the form of a radiofrequency pulse is delivered at this frequency, the hydrogen atoms are excited by absorption of this energy, which causes their magnetic moments to line up in the same direction. When the external energy is turned off, the absorbed energy is released, and the magnetic moments return to their previous orientations. The energy released is the basis of the MRI signal and is unique to the varying molecular structures and amount of hydrogen in various brain regions.

Two approaches can be used for the measurement of CBF with MRI. Akin to first-pass CTP imaging, the more established approach involves administration of the paramagnetic agent gadolinium and is variously referred to as dynamic susceptibility contrast, first-pass, or bolus perfusion MRI. Of note, the contrast agent remains wholly intravascular and only indirectly alters MRI signal intensity, thus hindering absolute quantification of hemodynamic parameters. The other MRI approach for imaging CBF is called arterial spin labeling and uses radiofrequency pulses to “tag” arterial blood water molecules, thereby obviating the need for exogenous contrast agent. Changes in the amplitude of the MRI signal can be used to construct quantitative images of cerebral perfusion, although there are still outstanding issues with absolute quantification.²⁸⁷

Clinical magnetic resonance spectroscopy (MRS) is usually based on the principle that protons resonate at different frequencies in the same static magnetic field when located in different chemical microenvironments. It can be performed with a variety of pulse sequences, the simplest being the application of a 90-degree radiofrequency pulse called free induction decay, which is then converted to a spectrum by a Fourier transformation. The spectral chemical shift is measured in parts per million and is a characteristic of the variation in resonance frequency. Its specific dependency on the chemical microenvironment of a particular nucleus makes it resemble a “fingerprint” of the analyzed substance. Both ¹H-MRS and ³¹P-MRS are in clinical use, although ¹H-MRS is more widely used. ¹H-MRS allows the measurement of several brain metabolites, including lactate, N-acetylaspartate, total creatine, glutamine/glutamate, and choline, whereas ³¹P-MRS can measure intracellular pH, as well as signals from ATP and PCr.

MRI has the principal advantages of avoiding ionizing radiation exposure and providing a high degree of temporal and spatial resolution. The main limit on the wealth of diagnostic information that can be obtained is the duration of the examination. However, it is susceptible to movement artifact and is contraindicated in patients with ferromagnetic implants.

Regulation of Cellular Calcium

[Ca²⁺]_i is tightly regulated within narrow limits by multiple mechanisms.³³⁴ Under normal conditions, entry of Ca²⁺ into the cell from the ECS is favored by a 10,000-fold higher extracellular calcium concentration ([Ca²⁺]_e) than [Ca²⁺]_i and a negative resting cell membrane potential. One determinant of [Ca²⁺]_i is the balance between passive influx and active extrusion across the plasma membrane. Ca²⁺ enters cells by voltage-sensitive and agonist-operated calcium channels. At the level of the plasma membrane, the extrusion mechanisms consist of ATPase-driven export of Ca²⁺ and a 3Na⁺-Ca²⁺ (NCX) exchanger that indirectly derives its energy from the Na⁺ gradient. Because the NCX is electrogenic and carries 3 Na⁺ (in) for 1 Ca²⁺ (out), it is driven not only by the Na⁺ gradient but also by the membrane potential. In some circumstances, such as collapse of the Na⁺ gradient and membrane depolarization, Ca²⁺ translocation by NCX can be reversed and cause cell calcium loading.

Other regulatory mechanisms control movement of Ca²⁺ between the endoplasmic reticulum and the cytosol and between mitochondria and the cytosol.^335,336 The endoplasmic reticulum is the main compartment for Ca²⁺ storage in the cell. Sequestration of free cytosolic Ca²⁺ within the endoplasmic reticulum depends on sarcoplasmic-endoplasmic reticulum Ca²⁺-ATPase (SERCA), whereas release back to the cytoplasm depends on signaling mediated by ryanodine and IP₃ receptors.^337,338 The calcium-binding proteins calreticulin and calnexin enhance this buffering capacity of the endoplasmic reticulum.³³⁹ Mitochondrial influx of calcium is driven principally by a uniporter (i.e., by the mitochondrial membrane potential Δψ_m), which being negative inside, exerts electrophoretic drag on the positively charged calcium ion.^{335,336,340–342} The rate of influx through the uniporter is proportional to [Ca²⁺]_i, which has a large capacity (Vmax). Conversely, the normal efflux pathway for mitochondrial calcium involves Na⁺-dependent and Na⁺-independent secondary active transporters, which are substantially slower than the uniporter.³⁴³ Hence, the intramitochondrial calcium concentration ([Ca²⁺]_m) can acutely reach excessive levels if [Ca²⁺]_i is elevated.

Depletion of Adenosine Triphosphate and Terminal Depolarization

The primary insult of cerebral ischemia is loss of ATP as a result of a reduction in oxidative phosphorylation. The degree and duration of the ischemia determine the severity of the energy impairment. In complete and prolonged ischemia, energy metabolism ceases altogether. Whether an interruption in glucose supply from blood depletes cellular ATP depends on the availability of alternative glycolytic and oxidative substrates and on the cell’s rate of consumption of ATP. More commonly, ischemia is incomplete, and when the oxygen supply is critical, greater reliance is placed on anaerobic metabolism. Recall that astrocytes possess their own glycogen stores and, by converting these stores to glucose, can, at least in theory, temporarily continue to supply a limited quantity of ATP by anaerobic glycolysis.³⁴⁴ Meanwhile, interruption of oxidative phosphorylation causes ATP synthase to run in reverse, and this can accelerate the depletion of residual ATP.³⁴⁵

In the core of a focal ischemic lesion, severe or complete reduction of ATP levels to 25% or less of basal levels initiates an early disturbance in ion gradients.³²⁹ Of foremost importance is a reduction in the activity of plasma membrane Na⁺,K⁺-ATPase. This affects neurons before astrocytes because the former have higher energy demands.³⁴⁶ Because Na⁺,K⁺-ATPase counters the inherent “leakiness” of the cell membrane, the membrane potential gradually rises, which leads to the opening of voltage-gated ion channels. This permits the influx of Na⁺, Ca²⁺, and Cl⁻ and the efflux of K⁺ along concentration gradients previously established by primary and secondary active transport. The influx of Na⁺ and Cl⁻ is accompanied by osmotically obligated water, thereby resulting in cell swelling or osmolytic damage. This cell swelling is not considered to be a principal cause of cell death in vivo because of the limited Na⁺, Cl⁻, and water content of extracellular fluid (about 20% of tissue volume). More importantly, profound dissipation of ion gradients ensues, and with insufficient Na⁺,K⁺-ATPase activity to reestablish them, neurons become terminally depolarized and electrically silent.³⁴⁷ This sequence of events probably underlies cell death in the ischemic core.

In the penumbra, the catastrophic events of the core do not occur because tissue energy status is nearly normal or relatively well preserved. The suppression of electroencephalographic activity along with the supply of sufficient substrates for ongoing anaerobic synthesis of ATP avoids the rapid onset of extreme ionic disturbances seen in the core. Nevertheless, cell death eventually ensues if blood flow is persistently depressed. There is experimental evidence from rodents that the occurrence of repetitive peri-infarct depolarization may ultimately lead to terminal depolarization.³⁴⁸ These depolarizations occur very abruptly, last between 3 and 5 minutes, have the characteristic form of spreading depression, and are associated with K⁺ efflux and Ca²⁺ influx. Their frequencies are halved by NMDA or AMPA antagonists, and hence they are believed to originate from glutamate release in the ischemic core.³⁴⁹ They probably aggravate the energy demands on penumbral brain because they necessitate the consumption of ATP to restore normal transmembrane gradients. Certainly, biochemical measurements confirm the episodic depletion of ATP, and factors that reduce the frequency of these depolarizations also reduce the final infarct size.^348–351 Nonetheless, it remains to be established that such depolarization waves occur in primates or in humans. Instead, cell death in the penumbra has been attributed to a host of secondary phenomena triggered by acidosis, excess glutamate signaling, and cytosolic Ca²⁺ overloading.

Acidosis

As a consequence of increased anaerobic glycolysis, the proton and lactate levels of the brain rise during ischemia, with a corresponding reduction in intracellular and extracellular pH.^346,352 The pH of ischemic brain may fall below 6.2 in normoglycemia and below 5.6 in the presence of preischemic hyperglycemia.³⁵³ Evidence has been presented that acidosis inhibits glycolysis, thus compromising the brain’s attempt to maintain ATP levels.³⁵⁴ Acidotic conditions during ischemia also reduce the activity of endogenous antioxidants and favor the liberation of ferrous iron from binding sites on ferritin and other protein complexes, which catalyzes hydroxyl radical (·OH⁻) formation from hydrogen peroxide (H₂O₂) (the so-called Fenton or Haber-Weiss reaction).³⁵⁵

Glutamate-Dependent and Glutamate-Independent Cellular Calcium Overload

The depolarization of neurons by ischemia triggers the release of glutamate and other excitatory neurotransmitters. Glutamate can activate two types of ionotropic glutamate receptors on neurons—those preferentially activated by AMPA or NMDA. Glutamate acts on the AMPA receptor to open a channel that allows the passage of monovalent cations (Na⁺, K⁺, and H⁺). When Na⁺ enters this channel down its electrochemical gradient, it depolarizes the postsynaptic membrane and allows the influx of Ca²⁺ through L- and T-type VGCCs on the postsynaptic membranes of the dendrites and cell body. In addition, it relieves the magnesium (Mg²⁺) block on postsynaptic NMDA-gated channels, thereby allowing Ca²⁺ to pass through these particularly high-conductance channels. Although AMPA-gated channels are generally considered to conduct monovalent cations only, it has been established that they may permit an inward Ca²⁺ current with certain combinations of receptor subunits (GluR1 to GluR4) and the degree of RNA editing.^356,357 According to the GluR2 hypothesis, the assembly of AMPA receptors containing the GluR2 subunit, which are impermeable to Ca²⁺, is preferentially suppressed relative to that of other AMPA receptors after cerebral ischemia.³⁵⁸ Additionally, activation of metabotropic glutamate receptors results in the mobilization of intracellular endoplasmic reticulum calcium stores via activation of PLC and the production of second messengers such as IP₃.

The glutamate-stimulated increase in free [Ca²⁺]_i is normally promptly terminated by reuptake of glutamate into astrocytes by Na⁺-glutamate cotransporters. Recall that glutamate uptake by GLT1 and GLAST is driven by the electrochemical gradient of Na⁺, which is maintained by Na⁺,K⁺-ATPase (see earlier). Because of the failure of Na⁺,K⁺-ATPase, uptake of glutamate by these transporters is inhibited. It is also possible that glutamate uptake by glial GLT1 and the excitatory amino acid transporter EAAT2 can be downregulated by free radical–mediated oxidation of a redox site on these transporters.³⁵⁹ Furthermore, because these transporters are electrogenic (i.e., normally transferring a positive charge inward), membrane depolarization can lead to a reversal of the direction of Na⁺ transport and actively contribute to excess of glutamate in the ECS. Thus, impaired glutamate uptake and enhanced glutamate release combine to cause increasing concentrations of extracellular glutamate, abnormal prolongation of glutamate signaling, further cellular Ca²⁺ influx, and ever increasing elevations in free [Ca²⁺]_i. Initiation of this positive feedback loop is crucial to the theory of glutamate-mediated excitotoxicity.

Researchers have also identified an important role for ischemic cation influx secondary to glutamate-independent mechanisms,³⁶⁰ including TRPM7, a nonselectively permeable member of the melastatin subfamily of TRP channels that is activated by oxidative stress; acid-sensitive ion channel 1a (ASIC1a), which carries an inward Na⁺ and Ca²⁺ current in response to extracellular acidification; Na⁺/K⁺/2Cl⁻ cotransporter isoform 1 (NKCC-1); and NCX, which can work as an antiporter in either the Ca²⁺ efflux/Na⁺ influx (forward) or Na⁺ efflux/Ca²⁺ influx (reverse) modes. The first two have been found to contribute to intracellular Na⁺ and Ca²⁺ overload in cerebral ischemia.^361–363 With regard to NCX, investigators have previously found evidence of both detrimental and beneficial functions in the setting of ischemia.³⁶⁴ NCX has three isoforms that are encoded by three different genes. Significantly, although NCX1 and NCX2 require ATP to function, NCX3 does not. Intracellular sodium overload through channels such as TRP, ASIC1a, and NKCC-1 has the potential to cause NCX to function in reverse mode, thereby exacerbating the cytosolic Ca²⁺ load.³⁶⁵ Alternatively, there is evidence that cleavage of NCX proteins by the Ca²⁺-activated protease calpain promotes the excitotoxic death of neurons,^366,367 which suggests that the forward function of NCX may indeed be operating to reduce the cytosolic Ca²⁺ level. In this regard, it is of interest that ischemic brain damage is much worsened in NCX3-null mice.³⁶⁸

There is convincing evidence that mitochondria actively accumulate most of the Ca²⁺ that enters neurons during pathologic glutamate signaling in ischemic regions. In this situation, many of the mechanisms for the reduction of cytosolic Ca²⁺ are disabled because of ATP depletion. In fact, under extreme ischemic stress, endoplasmic reticulum Ca²⁺ stores are discharged into the cytoplasm, thereby aggravating the cytosolic Ca²⁺ overload.^369,370 Hence, there is an increase in [Ca²⁺]_i and a proportional rise in activity of the uniporter responsible for the mitochondrial Ca²⁺ influx. This usually serves the physiologic purpose of stimulating the activity of matrix Ca²⁺-sensitive dehydrogenase, thus increasing the supply of reducing equivalents to the respiratory chain and enhancing ATP synthesis. However, because the mitochondrial Ca²⁺ efflux mechanisms are substantially slower than the uniporter, they are readily overwhelmed by excessive Ca²⁺ influx through the uniporter, with significant adverse consequences on mitochondrial function (see later).³³⁴

Activation of Inflammation

A number of proinflammatory genes, including those for tumor necrosis factor-α (TNF-α) and interleukin-1β (IL-1β) and the transcription factors HIF-1, nuclear factor κB (NF-κB), and interferon-regulatory factor-1, are activated in response to the hypoxia, superoxide radical (·O₂⁻) formation, and Ca²⁺ influx in the acute phase of cerebral ischemia.^371,372 These products influence the expression of adhesion proteins such as intercellular adhesion molecule-1 (ICAM-1), P-selectins, and E-selectins, which permit neutrophils to accumulate in the vasculature.³⁷³ A major role in inflammation is played by resident microglia, whose proliferation and activation are a striking feature of the ischemic penumbra.³⁷⁴ In addition, chemokines such as monocyte chemoattractant protein-1 (MCP-1) induce the transmigration of leukocytes and monocytes into ischemic brain.³⁷⁵ About a third of these monocytes differentiate into cells that are indistinguishable from locally present microglia (Fig. 343-8).

FIGURE 343-8 Schematic illustration of major leukocyte influx into brain tissue subjected to ischemia. Apart from the upregulation of vascular adhesion molecules, several other factors, such as a chemokine gradient over the vascular endothelium, govern the final flux of leukocytes into the ischemic tissue. Although not fully clarified, current data point to neutrophils as the major causative agent in secondary inflammatory tissue injury after ischemia. Monocytes may transduce ameliorating signals such as anti-inflammatory cytokines and growth-promoting factors.

(Reproduced from P Siesjo, BK Siesjo. Cerebral metabolism and the pathophysiology of ischemic brain damage. In: Winn HR, ed. Youmans Neurological Surgery, 5th ed. Philadelphia: WB Saunders, 2004.)

Inflammation has many potentially deleterious effects in the ischemic brain. The accumulation of neutrophils, platelets, and red blood cells can be so intense that it leads to capillary plugging and results in focal areas of downstream microvascular perfusion defects (the so-called no-reflow phenomenon).³⁷⁶ Soluble factors released from neutrophils have been shown to activate endothelial matrix metalloproteinases (MMPs).³⁷⁷ Another source of MMPs is activated microglial cells.³⁷⁸ MMPs cleave protein components of the extracellular matrix, as well as process some cell surface receptors and soluble proteins.³⁷⁵ Studies suggest that MMP-2 and MMP-9 have in vivo activity in the conversion of pro-IL-1β into the mature proinflammatory cytokine IL-1β, which is known to be a crucial mediator of neurodegeneration in experimental models of cerebral ischemia.³⁷⁹ Both MCP-1 and MMPs have been found to cause BBB breakdown with disorganization of tight junctions as reflected by the redistribution and loss of key proteins such as occludin and zonula occludens-1.^380–382 The permeability of the BBB is further exacerbated by hypoxia and neutrophil-induced activation of the contractile apparatus of the endothelial cell, which causes the formation of paracellular gaps.^380,383 Destruction of the BBB permits the transmigration of leukocytes, promotes the extravasation of albumin and other osmotic compounds leading to vasogenic edema, and increases the risk of hemorrhagic conversion of an infarct.³⁸⁴ In addition, the elaboration of oxygen free radicals by inflammatory cells contributes to oxidative stress and vascular injury.³⁸⁵

Enhanced Activity of Free Radical Species

Ischemia causes an increase in the generation of both ROSs and reactive nitrogen species (RNSs) to levels that cannot be managed by endogenous antioxidant enzymes and factors. In this scenario, mitochondria are strongly implicated in the production of excessive superoxide production.³⁸⁶ An initial burst is triggered by the interruption of oxidative phosphorylation at complex IV in the mitochondrial respiratory chain by the lack of oxygen.³⁴⁵ Later mitochondrial free radical generation is under the influence of high Ca²⁺, Na⁺ and ADP in ischemic cells.³⁸⁷ Invading inflammatory cells also make an important contribution to ROS and RNS formation in several in vivo animal models of cerebral ischemia.^388,389 Besides their direct cellular toxicity, these ROSs and RNSs can trigger the release of proinflammatory cytokines such as TNF-α, IL-1β, and IL-6, with subsequent upregulation of ICAM-1 and vascular cell adhesion molecule (VCAM), thereby leading to positive feedback.³⁹⁰

Ca²⁺ provides a major stimulus for the production of ROSs and RNSs in several ways. During the ischemic period, excessive consumption of ATP leads to accumulation of the purine metabolites xanthine and hypoxanthine in the brain, which on subsequent reperfusion are metabolized by xanthine oxidase to yield massive amounts of H₂O₂ and ·O₂⁻. Ca²⁺ influences this process by inducing the activation of a cytosolic serine protease that converts xanthine dehydrogenase (which does not produce ROSs) to xanthine oxidase by limited proteolysis. In other cases, Ca²⁺ may trigger a reaction sequence that secondarily gives rise to ROS production. This occurs when Ca²⁺ activates PLA₂, which causes hydrolysis of phospholipids and accumulation of arachidonic acid. It is during the metabolism of arachidonic acid by COX and lipoxygenase that ROSs, notably ·O₂⁻, are formed.^391,392 Elevations in [Ca²⁺]_i can also produce several free radical species through stimulating the synthesis of NO by calcium-dependent nNOS. NO synthesis is particularly linked to Ca²⁺ influx via the NMDA receptor–operated channel through coupling of nNOS to the NMDA receptor by postsynaptic density 95 protein (PSD-95).³⁹³ NO itself has limited radical reactivity, and its half-life is brief because of its rapid reaction with ·O₂⁻ and possibly H₂O₂ to produce neurotoxic peroxynitrite (ONOO⁻), which can in turn decompose to another toxic species, ·OH⁻, the hydroxyl radical.³⁹⁴ This important reaction sequence is described by the following three equations:

(17)

(18)

(19)

Of all the NOS isoforms, nNOS also has a propensity for generating ·O₂⁻ through the oxidation of NADPH at the expense of O₂. Such uncoupling of O₂ reduction from NO formation is more likely under the acidic pH of the ischemic brain.³⁹⁵

It has been demonstrated in numerous studies that free radicals are directly involved in the damage to cellular macromolecules, especially lipids, proteins, and nucleic acids,³⁹⁶ and influence redox signal transduction pathways. With regard to the latter, free radicals can trigger the translocation of cytochrome c from the mitochondria to the cytoplasm and give rise to apoptosis.³⁹⁷ Through reduction of the redox state of the cell, free radicals can also rapidly reduce levels of the constitutively expressed DNA repair enzyme apurinic/apyrimidic endonuclease-1/redox factor-1 (APE/Ref-1) and activate transcription factors involved in the upregulation of proinflammatory genes.³⁹⁸ Experiments have implicated this reduction of APE/Ref-1 in the fragmentation of cellular DNA in focal cerebral ischemia, which may then precipitate apoptotic pathways.³⁹⁸ In addition, ROSs such as superoxide have the capacity to alter the vascular response to CO₂, and endothelium-dependent vasodilators such as acetylcholine increase platelet aggregability, increase endothelial and BBB permeability, and disrupt endothelial membranes.^385,399

DNA Damage

DNA damage in ischemia is primarily mediated by free radicals. Like other free radicals, ONOO⁻ exerts toxic effects via a multiplicity of mechanisms.^400,401 However, ONOO⁻ is unique among the free radical species in having a long enough half-life to travel within and between cells and eventually entering the cell nucleus to trigger single-strand breaks in DNA.⁴⁰² In this respect it is regarded as the key pathophysiologic species in the oxidation of and damage to DNA, with subsequent activation of poly(ADP-ribose) polymerase-1 (PARP-1), which as elaborated later, is believed to be a key mediator of cell death.^403,404 In addition, intraneuronal Ca²⁺ overload or acidification can activate Ca²⁺/Mg²⁺-dependent endonucleases, or DNase II, and lead to internucleosomal cleavage of DNA.⁴⁰⁵

Generation of Lipid Mediators

Lipid catabolism is enhanced in ischemia, both because of insufficient ATP and cytidine triphosphate to catalyze the resynthesis of phospholipids in the course of their normal turnover and because calcium activates PLA₂ and PLC.^406,407 Hence, ischemia leads to a marked increase in free fatty acid levels, particularly arachidonic acid. Once reperfusion is initiated, oxidative metabolism of the accumulated arachidonic acid leads to the formation of COX and lipoxygenase products, some of which are active in promoting inflammatory responses, including the chemotactic accumulation of leukocytes in ischemic or postischemic tissue, whereas others promote cerebral edema, vasoconstriction, and platelet aggregation.⁴⁰⁸ PLA₂ activation also gives rise to the formation of platelet-activating factor (PAF) from glycerophosphocholine (a minor phospholipid component of the membranes). The PLA₂-activated reaction produces lyso-PAF, but this is converted to PAF by an endogenous acetyltransferase.⁴⁰⁹ PAF itself is a potent vasoconstrictor that may worsen flow reduction in the ischemic brain.⁴¹⁰ Furthermore, PAF can promote transcriptional activation of primary response genes, including c-jun, c-fos, and the gene for the inducible form of prostaglandin synthase (PGS-2), which leads to the elaboration of a spectrum of vasoactive eicosanoids.⁴⁰⁶

Proteolysis by Calpains and Cathepsins

The elevations in [Ca²⁺]_i resulting from ischemia can cause toxicity through the mediation of calcium-dependent proteases known as calpains. Calpains cleave α-spectrin, actin, and fodrin-α, thereby causing collapse of the cytoskeleton.^411,412 In itself, this can cause serious problems for intracellular signaling, which depends on the integrity of the cytoskeleton. In addition, it has been proposed that damage to cytoskeletal proteins can contribute to subsequent damage to mitochondria. To be specific, many proteins in mitochondria, including mitochondrial RNA polymerase and most of the 13 subunits that make up cytochrome oxidase (complex IV) for the respiratory chain, are encoded only by nuclear DNA. Hence, neuronal mitochondria need to be periodically shuttled to and from the perikaryon by the cytoskeletal proteins cytoplasmic dynein and kinesin to obtain their full complement of proteins. Because this shuttle system is essential to mitochondrial function, intra-ischemic and postischemic damage to the motor proteins (dynein and kinesin) may be a cause of delayed neuronal damage, simply because mitochondria in regions far from the perikaryon run out of proteins to sustain oxidative phosphorylation.⁴¹³

Calpains have also been implicated in cell death through the liberation of destructive proteases from the lysosome, the “suicide bags” of the cell.⁴¹⁴ The lysosomal proteases cathepsin B and L have been found in the cytoplasm of ischemic cells, presumably after injury to the lysosomal membrane by free radicals or extralysosomal hydrolytic enzymes.³¹⁵ With regard to the latter, involvement of calpains has been hypothesized because they have been found to rapidly localize to lysosomal membranes in primate models of ischemia.^415,416 Among various targets, cathepsin B can activate the proinflammatory caspase-1 (also known as interleukin-1β–converting enzyme [ICE]) and caspase-11, which play important roles in brain ischemia by triggering inflammation and apoptosis.^370,417 In addition, the proteolytic effects of cathepsin B include the activation of Bax and Bak, two proapoptotic members of the B-cell lymphoma (Bcl-2) family, and the subsequent release of proapoptotic cytochrome c through permeabilization of the mitochondrial membrane.

Secondary Energy Failure

Some evidence suggests that free radical overproduction causes ischemic death by compromising mitochondrial energy metabolism on a number of levels. The NO produced by nNOS, as well as by mtNOS, reversibly binds to the Cu²⁺ center of cytochrome oxidase and consequently inhibits electron transfer to O₂ and mitochondrial O₂ uptake.⁴¹⁸ The free radical species ONOO⁻, ·OH, and ·O₂⁻ have been shown to cause direct inactivation of essential enzymes of the TCA cycle and metalloprotein complexes of the electron transport chain. For example, the Fe-S centers in electron transfer complexes are exquisitely sensitive to radical-mediated inactivation, as are pyruvate and succinate dehydrogenases, succinate oxidase, complex I, and F₀F₁-ATPase.^419,420 ONOO⁻ can also trigger the high conductance state of the mitochondrial permeability transition pore (MPTP), thus dissipating the mitochondrial membrane potential and resulting in cessation of electron transport and ATP synthesis.⁴⁰¹ It is possible, too, that alterations in the fluidity of membranes subsequent to the peroxidation of their lipid constituents by free radicals could undermine electron flow by affecting lateral diffusion within the inner mitochondrial membrane.⁴²¹

Yet another important contribution to secondary energy failure occurs through the activation of PARP-1 by the damaging actions of ONOO⁻ on DNA. Evidence of PARP-1 activation is prominent in human cerebral infarcts.⁴²² PARP-1 is an abundant nuclear enzyme whose activation after DNA damage is an early event elicited by very small amounts of DNA damage. The activated PARP-1 binds to DNA and hydrolyzes NAD⁺ to form polymers of ADP-ribose on acceptor proteins. Activation of PARP-1 is energetically expensive. For every mole of ADP-ribose transferred from NAD⁺, one mole of NAD⁺ is consumed and four free energy equivalents of ATP are required to regenerate NAD⁺ to normal cellular levels. This process can potentially result in a dramatic decline in cellular NAD⁺, particularly when NAD⁺ biosynthesis by the ATP-dependent NAD synthetase is inhibited. Once the NAD⁺ concentration falls below the approximately 1-mM level necessary to sustain the glycolytic glyceraldehyde-3-phosphate dehydrogenase reaction or the approximately 0.1-mM level necessary for intramitochondrial dehydrogenase reactions, the rate of ATP production is impaired, and a vicious cycle results that if not reversed will lead to permanent metabolic failure and necrotic death.⁴²³

In addition, PARP-1 can catalyze the activation of proinflammatory pathways via enhancement of NF-κB–mediated transcription, which plays a central role in the expression of inflammatory cytokines, chemokines, adhesion molecules, and inflammatory mediators.⁴²⁴ Hence, activation of PARP-1 can result in the migration of macrophages and microglial cells, with the potential for production of large quantities of free radicals and toxic cytokines. Alternatively, PARP-1 activation can also act as a signal for the initiation of a caspase-independent apoptotic program through the nuclear translocation of apoptosis-inducing factor (AIF).

Mechanisms of Apoptosis

Intrinsic Pathway: Mitochondrial Release of Proapoptotic Molecules

Current understanding of mitochondrial involvement in apoptosis is that mitochondria have a key role in initiation of the process through the release of a number of apoptogenic proteins from the intermembrane space. These proteins include cytochrome c, the second mitochondrial activator of apoptosis/direct inhibitor of apoptosis (IAP) binding protein with low pI (smac/DIABLO), and AIF. Experimental evidence suggests that release of these mitochondrial proteins can occur through two main mechanisms: nonspecific rupture of the outer mitochondrial membrane (OMM) after the induction of an MPTP and selective OMM permeabilization by Bcl-2 proteins (Fig. 343-9).

FIGURE 343-9 The extrinsic versus intrinsic pathways of apoptosis. The intrinsic pathway is mediated by insults that trigger the exposed cell’s mitochondria and their permeability transition pores (MPTPs). The pathway is modulated by Bcl-2 and cytochrome c. The extrinsic pathway is mediated by death ligands acting on cell surface (e.g., Fas) receptors. Caspases are involved in both pathways, but earlier in the extrinsic pathway. Inhibitor of apoptosis (IAP) proteins inhibit apoptosis by binding to activated caspases. They inhibit signals generated through both these two major pathways of apoptosis. APAF, apoptotic peptidase-activating factor; UV, ultraviolet.

(Downloaded from http://www.imgenex.com/emarketing/081606_LivinorSurvivin/LivinorSurvivin_forweb.htm on 3/3/09.)

The MPTP is a multiprotein complex situated in the inner mitochondrial membrane; it was originally discovered as a mechanism by which mitochondrial calcium overload is relieved. The MPTP has been shown to be a key player in the cell death caused by Ca²⁺ overload, as well as by oxidative stress, high P_i levels, and decreased ATP, conditions that are created by ischemia.^425,426 There is ongoing debate over the composition of the MPTP, but one crucial component is known to be cyclophilin D, which is so named because it is bound and inhibited by the drug cyclosporine. Cyclophilin D is thought to facilitate a calcium-triggered conformational change in the component that forms the transmembrane channel by converting it to an “open” pore.^427,428 This causes indiscriminate discharge of Ca²⁺ and other molecules with a molecular mass of up to 1500 kD, dissipation of the mitochondrial electrochemical hydrogen ion gradient (Δψ_m), loss of matrix pyridine nucleotides, and rupture of the OMM, thus having an adverse impact on both the function and structural integrity of the organelle.

The MPTP was originally proposed to be the universal mechanism for mitochondrial involvement in apoptosis, but more recent studies have supported the provision by Bcl-2 molecules of a more important MPTP-independent mechanism, as well as regulatory mechanisms for the MPTP.^428,429 Proteins in the Bcl-2 family share up to four Bcl-2 homology domains (BH₁ to BH₄) and are divided between proapoptotic and antiapoptotic members.⁴³⁰ It has been proposed that BH₃-only proteins induce oligomerization of the multidomain proapoptotic Bax and Bak, which then insert themselves into the OMM and form large pores through which proteins may be able to leave the intermembrane space.⁴³¹ Unlike the MPTP, this Bax- and Bak-mediated mitochondrial membrane permeabilization appears to be facilitated by Mg²⁺ and insensitive to cyclosporine.⁴³² In contrast, Bcl-2 and Bcl-X are well-characterized inhibitors of apoptosis, and evidence suggests that they can inhibit opening of the MPTP.^433,434

Release of cytochrome c from the confines of mitochondria is the commitment step in both the intrinsic and extrinsic pathways of apoptosis.^324,435 In the cytosol, cytochrome c binds to deoxy-ATP and apoptotic peptidase-activating factor (APAF-1), which then oligomerizes into a septameric apoptosome. This process exposes APAF-1’s caspase activation and recruitment domain (CARD), thereby leading to the formation of a holoenzyme complex with caspase-9 and the subsequent activation of caspase-3 and other caspases such as types 2, 6, 8, and 10.⁴³⁶ Caspase-3 is known as an executioner of apoptotic cell death because it initiates the final common pathway in apoptosis that ultimately causes the morphologic and biochemical changes seen in apoptotic cells. Caspase-3 cleaves a range of critical cellular substrates and destroys their functions in the process, including the DNA repair enzyme PARP, the plasma membrane Ca²⁺-2H⁺ pump, the cytoskeletal components gelsolin and fodrin, and nuclear lamins.^324,437,438 Caspase-3 can also activate proteins such as caspase-activated DNAse (CAD), which is responsible for the distinctive electrophoretic DNA ladder of apoptosis. Apoptosis can be blocked by binding of the X-linked inhibitor of apoptosis (XIAP) to APAF-1/caspase-9. However, XIAP is negatively regulated by smac/DIABLO, which is released from injured mitochondria together with cytochrome c.

By contrast, excitotoxic stimulation of NMDA receptors leads to the generation of free radicals and induction of apoptosis through PARP-1 activation. After PARP-1 activation, the flavoprotein AIF translocates from the mitochondria to the nucleus, preceding the release of cytochrome c and activation of caspase-3, the major executioner caspase. At the nucleus, AIF initiates nuclear condensation and large-scale chromatin fragmentation, thus dooming the cell to death. After AIF translocation, collapse of the mitochondrial membrane potential and initiation of the release of cytochrome c take place.⁴³⁹ However, the apoptogenic actions of AIF appear to be independent of caspase.⁴³⁹ Exactly how PARP-1 activation leads to the release of AIF is unclear, but the depletion of NAD⁺ after PARP-1 activation or some product of PARP-1 could serve as the signal.