I. Background

The cerebral circulation can be divided into categories based on a number of criteria. For example, vessel size can be utilized to divide the brain circulation into the macro-circulation and the microcirculation. On the arterial side, the former begins in the neck and extends to pial arteries. The microcirculation consists of the pial and penetrating arterioles, the capillaries and the venules. Within the arterial circulation, the major function of the macro-circulation is conductance of blood whereas the principal activity of the arterial component of the microcirculation is regulation of flow. Although the capillaries and venules are a significant contributor to cerebrovascular resistance, their role in regulation of flow is minimal under physiologic conditions.

In addition to size, vessel location can be used to characterize the brain circulation. Thus, the brain vasculature can be divided in relation to the parenchyma into an extrinsic and an intrinsic component. The former encompasses the large conductance vessels (i.e., the macrocirculation) plus the pial circulation, whereas the latter consists of the intracerebral circulation composed of small arterioles, capillaries and venules. Penetrating arterioles represent the transition from the extrinsic to the intrinsic system. Clearly these two systems differ in regard to a number of features, such as proximity to the parenchyma, influence of neuronal and astrocytic activity, and presence and origin of innervation. However, despite these differences, both the extrinsic and intrinsic components of the cerebral circulation have an integrated response to cerebral challenges: co-dependency is the norm under physiologic and most pathologic conditions.

II. Microcirculation: Normal Anatomy & Special Features

III. Physiology of the Pial and Penetrating Arterial Circulation

The microcirculation of the brain shares many anatomic, physiologic and pharmacologic features with the microcirculation elsewhere in the body, but also has unique aspects (see below, III). The general principles of fluid dynamics apply: CBF (F) is directly related to pressure (P) and inversely affected by resistance (R):

Resistance is theoretically and experimentally inversely related to the 4^th power of the diameter (Kobari et al., 1984). Thus, small increases in the diameter of the arterioles result in significant decreases in resistance, and, assuming that there is sufficient perfusion pressure, flow will increase dramatically.

Flow is also affected by rheology and varies with hematocrit, erythrocyte flexibility, platelet aggregation and plasma viscosity (Kee and Wood, 1984). In vessels smaller than 200 to 300 μm diameter, red blood cells move into the central, rapid moving part of the flowstream, while white blood cells and other constituents of the plasma move more slowly next to the vessel wall (Chien et al., 1984). At asymmetrical branch points cells are distributed in an uneven fashion, the greater proportion goes to the branches deviating the least from the original direction and to those with higher flow. Tributaries close to a right angle may receive only plasma, platelets and a few white cells. Thus, hematocrit in the microvasculature is lower than in the average large blood vessel (Chien et al., 1984). White blood cells, on the contrary, are represented in a higher proportion in the microcirculation (Chien et al., 1984).

Furthermore, two distinct physiologic entities exert a dominant influence on CBF in the microvasculature: the Fahraeus effect (Fahraeus and Lindqvist, 1931) and the inversion phenomenon (Dintenfass, 1967). The Fahraeus effect describes an apparent decrease in the viscosity of blood flowing through tubes of progressive smaller diameter, whereas the inversion phenomenon explains that this reduction in blood viscosity ceases when the vessel diameter approaches 5–7 μm and reverses with further vessel diameter reductions.

Under physiologic conditions, a significant proportion of the arterial cerebrovascular resistance (CVR) is located in the pial and penetrating circulations (Heistad and Kontos, 1983). In order to increase flow to the intrinsic circulatory bed, the extrinsic pial vessels must dilate, otherwise a vascular steal would occur. Therefore, a coordinated response between extra- and intraparenchymal components of the vascular circulation is necessary to fulfill the flow demands of the cerebral tissue. The mechanisms responsible for coordinating the pial and penetrating arteriolar response are unclear and may be stimulus specific. Thus, with cortical activation and increased metabolic demand (and in the absence of hypoxia or hypotension), the initiation of pial arteriolar dilatation may involve a signal conducted from the activated parenchymal compartment to the pial vessels. The pathway of this conducted signal may involve astrocytes and their extensions into the glia limitans (Xu et al., 2008). Alternatively, the conducted signal for vasodilatation may travel in the arterial wall, as occurs in the non-cerebral circulation (Segal et al., 1989). Conduction in the arteriole wall by the endothelium and the vascular smooth muscle has been shown in penetrating arterioles (Horiuchi et al., 2002) and to be increased after ischemia (Ngai et al., 2007). In contrast to cortical activation, hypoxia and hypotension (and ischemia) may have direct effects on the wall of the extra-parenchymal arteries, obviating a need for a conducted signal from the cortex and the parenchymal circulation.

The arterial vasculature has characteristics intrinsic to the vessels (i.e., present in the absence of brain parenchyma). For example, isolated penetrating arterioles are capable of reacting to changes in pH and to a number of chemicals, compounds and peptides [e.g., NO (Furchgott and Zawadzki, 1980), adenosine (Meno et al., 1993), ATP (Dietrich et al., 1996; Janigro et al., 1993), K⁺ (Horiuchi et al., 2002; Kuffler and Potter, 1964), VIP (Heistad et al., 1980)]. Some of these substances exert their effects selectively with luminal or abluminal application. For example, ATP evokes a more significant vasodilation when applied intraluminally (Janigro et al., 1997), whereas H⁺ and adenosine are more potent applied abluminally (Ngai and Winn, 1993). Moreover, the response to some of these agents is size dependent (Kontos et al., 1978; Morii et al., 1986).

Another inherent property of the arterial vasculature is the ability to react to changes in intraluminal flow and pressure. Bayliss in 1902 (Bayliss, 1902) reported direct contraction and relaxation of canine hind limb arteries in response to an increase or decrease of intravascular pressure respectively, a reaction attributed to intrinsic properties of VSM. As demonstrated in Fig 1, isolated rat and mouse cerebral penetrating arterioles have the ability to constrict with increasing pressure, and to dilate with decreasing pressure (Ngai and Winn, 1995), (Coyne et al., 2002). The latter response is also observed with increasing flow and this flow-mediated vasodilatation can be altered by blocking NO production (see Fig. 2). Harder et al. suggest that VSM metabolizes arachidonic acid via a P4504A dependent pathway to 20-hydroxyeicosatetraenoic acid (20-HETE) (Harder et al., 1997). This compound is known as a potent vasoconstrictor and acts in part by blocking the open-state probability of Ca²⁺ activated K⁺ channels, thereby depolarizing VSM and causing vasoconstriction (Rubanyi et al., 1990). With a decrease in intraluminal pressure and possibly flow, the endothelium presumably acts as a mechanoreceptor, transducing variations in transmural pressure and shear stress into altered vascular tone.

Extrapolation of conclusions based on isolated vessel preparations may be questionable since these arterial segments are largely devoid of astrocytic endfeet, either because of their in vivo anatomy or because of loss of endfeet during harvesting. In vivo, astrocytes may directly influence or add a permissive element to the function of wholly intracerebral arterioles which not to be observed or attenuated in the in vitro setting. On the other hand, the penetrating arterioles, and to a significant extent, the pial arterioles and arteries, have variable contact with the astrocytic endfeet.

During hypotension in vivo, pial arteries and arterioles vasodilate, thereby decreasing cerebrovascular resistance (Kontos et al., 1978). This vasodilatation is prompt (5–20 s), with the smaller pial arteries and arterioles responding to a greater degree than the macrocirculation (Fig. 2). In a similar fashion, increases in MABP result in vasoconstriction (Kontos et al., 1978). Beyond either extreme, CBF becomes passive and varies directly with hypotension and hypertension.

IV. Changes in the Cerebral Vascular Response and Microcirculation during Ischemia

Hypotension and ischemia are in a continuum. In hypotension, the changes in the brain and the cerebral vasculature are, in general, physiologic and recoverable. The dominant response may be in the vascular compartment as demonstrated in isolated arterioles subjected to decreased intraluminal pressure and perfusion. In contrast, ischemia elicits pathophysiologic responses with the parenchyma being more sensitive to ischemia than the vasculature. Within the parenchyma, astrocytes are more resistant to ischemia than neurons, relating to differences in their intrinsic metabolism and signaling characteristics (Rossi et al., 2007; see Hertz, this issue).

Ischemia itself can be further divided by the degree (complete vs. incomplete), the volume (focal vs. global), the duration (transient or permanent) and the location within the brain, with certain areas being more vulnerable to ischemia.

The vascular response in ischemia will depend on the interplay of the factors noted above, as will the success of reperfusion. With ischemia, alteration in the endothelial cell wall results in leukocyte adherence and platelet rupture (del Zoppo and Hallenbeck, 2000) leading to potentially non-reversible intravascular injury within hours (Tagaya et al., 2001; Wagner et al., 1997). The breakdown of the blood brain barrier, occurring first in the venules, allows entry of multiple restricted substances into the brain. Larger arterial vessels are also affected by ischemia as demonstrated by the alteration of the biomechanics of the middle cerebral artery which occurs within 30 minutes of ischemic onset (Cipolla et al., 2001). Reperfusion after ischemia can trigger a series of events that are highly destructive and can add further injury to both the vasculature and the brain parenchyma. During minimal perfusion or reperfusion, activated leukocytes interact with the damaged abluminal endothelial wall, leading to plugging of the microvasculature and the development of a “no-reflow” phenomenon (Ames et al., 1968), although the existence of a leukocyte related no-flow phenomenon has been questioned (Aspey et al., 1989; Dirnagl et al., 1994). Reperfusion also has an independent effect on the parenchyma with the development of destructive processes, such as the release of inflammatory factors, cytokines and free radicals (Chan, 2001; Kuroda and Siesjo, 1997).

From a hemodynamic perspective and as noted previously, the capillaries and venules are a significant contributor to cerebrovascular resistance, but their role in regulation of flow is minimal under physiologic conditions. With ischemia, the resistance arterioles and arteries maximally dilate and become less responsive. The capillaries and the venules thus account for a greater component of the CVR, but are then directly exposed to unbuffered arterial blood pressure. As a consequence and in concert with cellular and molecular events related to alteration in the blood brain barrier, cytotoxic edema results with increase in tissue pressure and closure of capillaries (Hossmann, 2006). This phenomenon replicates the “choked flow” or “hemodynamic waterfall” predicted from the theoretical collapsible tube further complicating tissue perfusion (Hoffman and Spaan, 1990).

The alterations in the brain parenchyma during ischemia are multiple and complex whether the insult be local or global, transient or permanent, neuronal or astrocytic. The changes can be divided into early or late phases. The former is measured in seconds, whereas the latter extends from minutes to hours. Since this review focuses on factors affecting and potentially regulating the arterial micro-vasculature during ischemia, we restrict our assessment to the acute period. The major event during the initial period of complete ischemia is energy failure due to the lack of glucose and oxygen. With these deficits, synthesis of ATP attenuates or ceases, ATP concentrations fall, Na+-K+ ATPase activity decreases and trans-membrane ion gradients are disrupted (Rossi et al., 2007). Within 5 s of the onset of complete ischemia (Fig. 3), a time span during which pial arterioles dilate (Kontos et al., 1978), the concentration of phosphocreatine (PCr) and adenosine (ADO) are significantly altered (PCr down, ADO up), as demonstrated in freeze blown rat tissue (Winn et al., 1979). Adenosine is a potent vasodilator. Consequently, the rapid increase in adenosine concentrations may account for the changes in cerebral arteriolar diameter and the decrease in CVR during the acute period of ischemia. By 60 s, ATP begins to fall and other adenine nucleotides are altered (Winn et al., 1979). K+ concentration then rises in two stages: an initial (60–120 s) moderate increase followed by a more profound elevation, and is associated with a decrease in Na and pH (Rossi et al., 2007). Dependent on the species, spreading depression and anoxic depolarization occurs (Nedergaard and Hansen, 1993) with subsequent release of extracellular glutamate (Obrenovitch and Urenjak, 1997), stimulation of the N-methyl-D-aspartate receptors (Choi and Rothman, 1990) and increase in intracellular calcium (Choi and Rothman, 1990), leading to eventual cell death by either necrotic or apoptotic mechanisms (Zhao et al., 2007; Doyle et al., this issue). As previously noted, astrocytes are less sensitive to oxygen and glucose deprivation, therefore it would be reasonable to conclude that metabolic changes during the acute period of ischmia outlined above are reflective of neuronal rather than astrocytic alterations.

Almost all ischemic insults to the brain can be divided into a central core of dead or destined to die tissue and a surrounding region, the penumbra, which has marginal flow and cells with the potential to survive (Hossmann, 2006). All the changes reviewed above also apply to a variable extent to these two zones, but with some unique events in the penumbra such as reduction in protein synthesis (Castellanos et al., 2006; Rebel et al., 2005). A complex interplay of these factors with reperfusion determines the ultimate outcome of the parenchyma in the penumbra zone (Hossmann, 2006).

Additional influences on the extent and severity of parenchymal injury caused by ischemia are the temperature of the brain and the level of the blood glucose. Thus, even mild increases or decreases in brain temperature can significantly alter the metabolic rate and the lethality of the same degree of ischemia (Busto et al., 1987). Zhao, Steinberg and Sapolsky, in a recent review, comprehensively summarize the suggested mechanism involved in hypothermia in ischemic brain, but question its effectiveness in attenuating ischemic insults (Zhao et al., 2007).

High blood glucose levels result in worse outcomes in experimental settings and in humans following stroke and head injury (Lam et al., 1991; Pulsinelli et al., 1982). Hyperglycemia has been demonstrated by Cipolla et al. to dilate cerebral arteries and to diminish myogenic tone (Cipolla et al., 1997). This direct effect of hyperglycemia on the arteries could be abolished with the removal of the endothelium (Cipolla et al., 1997). The mechanisms responsible for the more extensive and pronounced parenchymal injury seen with hyperglycemia are unclear, but studies suggest a variety of possible causes, such as an increase in lactic acid and decrease in pH (Folbergrova et al., 1997), although MR spectroscopy failed to show a difference in intracellular pH with hyperglycemic or normoglycemic ischemia (Hsu et al., 1994)), a release of glucocorticoids (Schurr, 2002), or a decrease in cerebral adenosine production (Higashi et al., 2002; Hsu et al., 1991; Meno et al., 2003; Zhou et al., 1994).

V. Conclusion

Ischemia has multi-factorial and complex effects on the cerebral vasculature and the brain parenchyma. During ischemia, the pathophysiologic responses in the vasculature and the brain compartments are interconnected. Further understanding of the physiology and pathophysiology of ischemia will allow the rational development of pharmacological therapies for human stroke and brain injury.

Acknowledgments

Bibliography