Cerebral Vascular Disease and Neurovascular Injury in Ischemic Stroke

Abstract

The consequences of cerebrovascular disease are among the leading health issues worldwide. Large and small cerebral vessel disease can trigger stroke and contribute to the vascular component of other forms of neurological dysfunction and degeneration. Both forms of vascular disease are driven by diverse risk factors, with hypertension as the leading contributor. Despite the importance of neurovascular disease and subsequent injury following ischemic events, fundamental knowledge in these areas lag behind our current understanding of neuroprotection and vascular biology in general. The goal of this review is to address select key structural and functional changes in the vasculature that promote hypoperfusion and ischemia, while also affecting the extent of injury and effectiveness of therapy. In addition, as damage to the blood-brain barrier (BBB) is one of the major consequences of ischemia, we discuss cellular and molecular mechanisms underlying ischemia-induced changes in BBB integrity and function, including alterations in endothelial cells and the contribution of pericytes, immune cells, and matrix metalloproteinases. Identification of cell types, pathways, and molecules that control vascular changes before and after ischemia may result in novel approaches to slow the progression of cerebrovascular disease and lessen both the frequency and impact of ischemic events.

The consequences of cerebrovascular disease are among the leading health issues worldwide.^{1, 2} Although ischemic and hemorrhagic stroke is perhaps the best known end-organ effects, diseases of the cerebral circulation are a major contributor to hemorrhagic stroke, dementias (Alzheimer’s disease and vascular dementia) and other forms of neurological dysfunction and degeneration.^3–5 Ischemic stroke is primarily a consequence of carotid and cerebrovascular disease, the latter of which includes both large and small vessel disease (SVD). In addition to larger ischemic strokes produced by thrombosis in carotid or cerebral arteries, hemorrhagic stroke can also occur. Microvascular changes during SVD result in small regions of ischemia and microbleeds (microhemorrhages).^{6, 7} Once initiated, both large and SVD typically progress slowly over a period of many years. This rate of progression as well as the frequency of large and small ischemic events and microbleeds is accelerated during aging and in the presence of vascular risk factors (Figure 1). In addition, disease modifiers like genetics or diet can reduce or increase the rate of progression of vascular disease and events in brain (Figure 1).

Figure 1. Risk factors and end-organ effects of vascular disease.

Schematic illustration of leading risk factors for large and small vessel disease and stroke as well as key changes in vascular function, the BBB, atherosclerosis, and vascular structure. The lower portion of the figure illustrates progression vascular disease over time with major end-organ effects.

This review is organized into two interrelated sections. First, we highlight functional and structural changes in the vasculature that affect cerebral blood flow (CBF) and adaptive vascular responses in ways that promote hypoperfusion, underlie ischemic events, and are determinants of the extent of ischemic-induced injury. Examples of common underlying mechanisms of vascular disease are presented. Second, changes in the blood-brain barrier (BBB) and components of the neurovascular unit (NVU) are integrated elements of this pathophysiology. Recent insight into cellular and molecular mechanisms that underlie vascular and non-vascular injury and repair after ischemia with or without reperfusion (I/R) are presented with an emphasis on BBB, pericytes, inflammatory cells, along with related molecules. The impact of other NVU components, such as astrocytes have been extensively reviewed elsewhere.^{8, 9}

Part I. Cerebrovascular Disease: The Prelude to Stroke

Overview of mechanisms that control CBF

Several major mechanisms contribute to regulation of CBF. Collectively, these mechanisms determine baseline CBF and underlie adjustments in perfusion that occur acutely (from seconds to minutes) and chronically (from months to years) in response to physiological and pathophysiological conditions. To begin, we provide an overview of such mechanisms, each of which is affected by vascular disease and ischemic events.

Vascular endothelium

The impact of endothelial cells in health and disease should not be underestimated. These cells are the site of the BBB and thus have a major role in controlling the movement of ions, molecules, and cells into and out of the brain.^{10, 11} Endothelium determines the thromboresistance property of vessels via anti-platelet, anti-thrombotic, and fibinolytic mechanisms.^{12, 13} Through various intrinsic mechanisms, endothelial cells are normally in a state which suppresses proinflammatory gene expression, recruitment of monocytes, and development of atherosclerosis.^14–16

Endothelium-dependent effects on underlying smooth muscle are a major regulator of vessel tone. Two primary mechanisms are involved – production and release of vasoactive molecules that diffuse to specific molecular targets as well as endothelium-dependent hyperpolarization of vascular muscle.^16–18 Through these mechanisms, endothelial cells affect resting CBF and mediate vasodilator responses to shear stress, neurotransmitters, metabolic factors, and therapeutic agents.^{16, 17} An even broader impact of endothelial cells becomes apparent when their influence on amyloid precursor protein (APP), amyloid β, and phosphorylation of tau (critical for the pathogenesis of Alzheimer’s disease) is taken into account.^{19, 20} In addition, endothelium affects the function of various non-vascular cells including neurons, microglia, and oligodendrocytes.^{17, 19, 21} Although not the only mechanism involved, most of these effects are mediated through nitric oxide (NO) produced by endothelial NO synthase (eNOS)(Figure 2).^{12, 13, 16, 17, 19, 20, 22, 23}

Figure 2. Mechanisms regulating function of eNOS and its impact.

Activity of eNOS is increased in response to receptor- and shear stress-mediated effects. Activation of transient receptor potential V4 channels (TRPV4) is involved for some stimuli. Enzyme activity is dependent on L-arginine, calcium (Ca²⁺), calmodulin (CaM), and tetrahydrobiopterin (BH₄) and is inhibited by caveolin-1. eNOS-mediated effects in healthy endothelium is show in the upper right. The bottom of the figure highlights major mechanisms that contribute to endothelial dysfunction, including the RAAS, oxidative stress, asymmetric dimethylarginine (ADMA), and Rho kinase. See text for additional details. Aldo, aldosterone; AT1R, AT1 receptor; Mito, mitochondria; ONOO⁻, peroxynitrite; O₂⁻, superoxide; GPx, glutathione peroxidase; DDAH, dimethylarginine dimethylaminohydrolase (DDAH); ET1R, endothelin-1 receptor.

Neurovascular coupling (NVC)

The process by which changes in neural activity elicit proportional and spatially controlled changes in CBF is often referred to as NVC (or functional hyperemia). Some common tools in neuroscience, such as functional MRI, are based on changes in hemodynamics during the NVC response.²⁴ This response involves a network of coordinated cells and signaling events that have neuronal activation at its core, with integrated contributions by astrocytes and endothelial cells.^{25, 26} Astrocytes make cellular contacts with vascular muscle in a heterogeneous pattern depending on brain region (relatively high in striatum, much lower in cerebral cortex).²⁷ Signaling between neurons, astrocytes, and vascular cells involves a family of molecules (NO, K⁺, adenosine, prostanoids) that may vary with cell type and brain region.^{24, 28} Pericytes may be involved as well, but their role in NVC is controversial.^{22, 29, 30} Through an extensive terminal network, capillaries provide the final local distribution of oxygen, glucose, and other nutrients. Some have suggested that capillaries dilate during NVC,²⁹ but other studies do not support this concept.^30–32

Autoregulation

The ability of the brain to maintain a relatively constant level of CBF over a range of perfusion pressures is well described and typically referred to as autoregulation.^{18, 33} Of the various mechanisms involved, myogenic function of resistance vessels is generally considered the most important. These mechanisms contribute to autoregulation in that vessels constrict when intravascular (transmural) pressure is increased and dilate when intravascular pressure is reduced.¹⁸ This basic feature is present in vessels from animals models and humans.^{18, 22, 34–36} When studied in vitro, myogenic tone (generated when a vessel is pressurized) and myogenic responses (changes in tone with changes in transmural pressure) are often differentiated.¹⁸ The myogenic response is an intrinsic property of vascular muscle,^{18, 36} but can be modulated by other cell types. Although widely studied in isolated arteries, our understanding of its mechanistic basis is still fragmented, particularly in relation to what actually senses changes in transmural pressure. Simply stated, increases in pressure produce depolarization of vascular muscle, likely via activation of transient receptor potential (TRP) channels, resulting in increases in intracellular calcium and contraction.¹⁸ As discussed below, myogenic mechanisms can change in disease.¹⁸ For example, the relationship of CBF to perfusion pressure, presented as the autoregulatory curve, normally shifts to the right (to higher levels of pressure) during chronic hypertension. With such a shift, the brain can be at risk for reductions in CBF and injury during decreases in perfusion pressure, such as that which occurs with systemic hypotension or occlusion of upstream vessels during ischemia.

Vascular risk factors

To a great extent, stroke is a disease of the aged. The rate of vascular events including stroke increases markedly with age.^37–39 The latest global analysis indicates that males have a higher incidence of ischemic and hemorrhagic stroke compared to age matched females.³⁹ Chronic hypertension has been and continues to be a leading risk factor for vascular disease, ischemic and hemorrhagic stroke.^{1, 38, 40, 41} For reasons which are not clear, hypertension is a greater risk factor for stroke than it is for myocardial infarction.⁴⁰ While some risk factors vary geographically in their impact, hypertension is the leading risk factor for stroke regardless of region.⁴¹ Other key risk factors for stroke are dietary (diets high in salt, low in fruits and whole grains), metabolic (elevated fasting glucose and body-mass index), behavioral (smoking, low physical activity) and environmental (air pollution)(Figure 1).⁴¹ Collectively these specific risk factors account for over 90% of all strokes.⁴¹

Large vessel versus small vessel disease

Cerebrovascular disease is often discussed in terms of large and SVD. In this context, small vessels refer to those on the brain surface and within the parenchyma, including smaller arteries, arterioles, capillaries, and venules.^{7, 42} Large and SVD share common features, but also have unique characteristics. In this section, we highlight some specific examples and concepts related to large and small blood vessels.

The brain is unique in relation to the distribution of vascular resistance. When small arterioles in the pial circulation are taken as a reference point, roughly half of vascular resistance resides in arterioles and arteries upstream, the other half is downstream within the circulation of the parenchyma.^{22, 33} Although the BBB is present throughout the cerebral circulation, it has heterogeneous features.^43–45 Capillaries often receive much of the focus, but many of the interactions between the BBB and immune cells,⁴⁴ along with dynamic changes in the integrity of tight junctions (TJ) in response to ischemia occur at the level of small venules.⁴⁶

While both arteries and arterioles display myogenic function, the degree of myogenic tone developed is related to vessel size. For example, isolated parenchymal arterioles develop greater myogenic tone than the middle cerebral artery.^{34, 47–50} Small pial arterioles have substantial tone in vivo,^51–56 greater than that present in larger vessels within the same network.^{57, 58} Through its molecular targets, eNOS influences tone in large arteries to the smallest parenchymal arterioles.^{17, 22, 34} In contrast, endothelium-dependent hyperpolarization that involves small and intermediate conductance K⁺ channels and myoendothelial gap junctions have more prominent effects in small arterioles.^{17, 22, 34} As discussed below changes in vascular mechanics and structure during disease can also differ in large versus small vessels.

Vascular regulation requires integration between large and small vessels. For example, both pial and parenchymal arterioles dilate during NVC.^{24, 25, 31, 59} Because resistance of large arteries is relatively high in brain,^{22, 33} an absence of upstream vasodilation would result in reductions in local microvascular (perfusion) pressure during NVC or with other stimuli that reduce small vessel resistance.^{22, 33, 60} Although both conducted and flow-mediated responses may be involved in vivo, mechanisms that control this integration is one of the least understood areas of vascular control. Endothelial cells sense shear stress^{16, 17} and are responsible for propagation of vasodilation during NVC, through mechanisms that have yet to be defined.²⁵

Endothelial dysfunction: Putting the brain at risk

Considering the broad impact of this cell, it is no surprise that loss of endothelial health represents a cornerstone event in the pathogenesis of vascular disease, and as a consequence, brain health (Figures 1 and 2).^{17, 19, 22} Endothelial-based abnormalities reside center stage in relation to cerebrovascular disease, stroke onset, neurovascular injury and repair. These changes in endothelium promote oxidative stress, low-grade inflammation, increased vascular tone, loss of BBB integrity, atherosclerosis, and thrombosis. Collectively, such changes are often referred to as endothelial dysfunction.^{17, 23, 61} Because a portion of the response is endothelium-dependent, endothelial dysfunction contributes to loss of NVC as well. In relation to long-term effects on brain health, reductions in NVC are thought to contribute to cellular injury and degeneration over time. However, the relative importance of baseline hypoperfusion versus reductions in NVC per se is not known.⁶² Disruption in the endothelial control of vascular tone is predictive of cardiovascular events including stroke in animals and humans.^{16, 17, 23}

How does endothelial dysfunction, often first detected based on changes in vascular tone, evolve and translate to more severe disease and ultimately ischemic events? Multiple mechanisms are now known to contribute. One key effect of eNOS-derived NO is suppression of atherosclerosis.¹⁴ This phenotype results from inhibitory effects of NO on platelets, expression of adhesion molecules with recruitment of monocytes, formation of oxidized low-density lipoproteins (ox-LDL) and isoprostanes, and activation of NF-κB (Figure 2).^{15, 61} Beyond their role in hemostasis, platelets and platelet-derived microparticles may contribute to the progression of atherosclerosis.⁶³ Platelet aggregation may be an early event in the pathogenesis of Alzheimer’s disease,^{64, 65} as platelets promote aggregation of β-amyloid through integrin and chaperone-dependent mechanisms.⁶⁵

Regulation of eNOS and its signaling is complex, occurring at the level of transcription, post-translational modification, enzyme activity, NO bioavailability, as well as molecular targets of NO (Figure 2).^{14, 16, 17, 66} Thus, abnormalities in this signaling network can occur at various levels. For example, genetic loss of eNOS augments atherogenesis in hyperlipidemic mice,¹⁴ while partial genetic deficiency, when combined with age, results in formation of microthrombi in brain, BBB abnormalities, localized loss of perfusion, and cognitive deficits.⁶⁷

There are many examples of oxidant-dependent mechanisms, reactive oxygen species (ROS)-mediated loss of NO signaling, and other aspects of vascular disease. Some of these mechanisms are common to both endothelial dysfunction and dysregulation of NVC (Figure 2).^{3, 17, 22, 62} For example, superoxide-mediated oxidative stress impairs regulation of vascular tone and CBF in models of aging and major risk factors for stroke.²² Oxidative stress can result from both increased production of superoxide or loss of antioxidant molecules that limit increases in superoxide (e.g., superoxide dismutases) (Figure 2). Major sources of superoxide in vascular cells include NADPH oxidases (Nox) and mitochondria (Figure 2).⁶⁸ Because superoxide reacts so efficiently with NO, the local concentration of superoxide is a determinant of the bioavailability of NO. In addition to NO, superoxide reacts with arachidonic acid forming isoprostanes.⁶⁹ Activation of thromboxane receptors by isoprostanes may contribute to atherosclerosis.^{69, 70} Other downstream effects of superoxide that contribute to vascular dysfunction include formation of peroxynitrite (the reaction product of NO and superoxide)(Figure 2) along with increased activity of ADP ribose polymerase and selective TRP channels.^{22, 62, 71}

Both local (organ-based) and circulating renin–angiotensin–aldosterone systems (RAAS)(Figure 2) contribute to vascular disease including atherosclerosis.^{17, 72–78} Angiotensin II (Ang II), a major effector peptide of the RAAS, produces oxidative stress, activation of NF-κB and low grade inflammation in both large and small vessels.^{17, 22, 71, 79–81} Many of the diverse effects of the RAAS are mediated by Ang II acting on AT1 receptors (AT1R)(Figure 2).^{17, 76} While less in known relative to Ang II, effects of aldosterone signaling via mineralocorticoid (MR) receptors on endothelium and smooth muscle have begun to emerge.^{82, 83} Circulating aldosterone produces Nox-2-dependent endothelial dysfunction without increasing arterial pressure (Figure 2).⁸⁴ In addition to detrimental effects, there is a protective arm of the RAAS. This includes the AT2 receptor, the angiotensin-converting enzyme 2 (ACE2) pathway, and angiotensin 1-7 (Ang 1-7) acting on mas receptors.^{76, 85} For example, activation of mas receptors produces NO-mediated endothelium-dependent vasodilation in cerebral arteries.⁸⁶

One of the mechanisms by which activity of eNOS and production of NO can be reduced is by formation and accumulation of methylated analogues of L-arginine including asymmetric dimethylarginine (ADMA), an inhibitor of NOS activity.^{66, 87, 88} Cellular concentrations of ADMA and effects on vessels are dependent in part on its hydrolysis by dimethylarginine dimethylaminohydrolases (DDAH)(Figure 2).⁸⁷ Interactions between ADMA, NADPH oxidase, and RAAS have been described.^{89, 90} In patients, circulating levels of ADMA positively correlate with the presence of intracranial atherosclerosis.⁹¹

Through inhibitory effects on expression and activity of eNOS (and other mechanisms), Rho kinase (ROCK) promotes endothelial dysfunction, vasoconstriction, and progression of atherosclerosis.⁹² Activation of RhoA and its target ROCK are required for Ang II- and endothelin-1 induced endothelial dysfunction and reduced NVC (Figure 2).^{50, 93} Subsequent experiments revealed that the ROCK2 isoform of ROCK was essential for Ang II-induced endothelial dysfunction.⁵⁰

Atherosclerosis and hypercholesterolemia

Atherosclerosis

In relation to ischemic events and associated cognitive deficits, atherosclerosis is one of the most important forms of vascular disease.^94–96 Atherosclerosis begins in regions where endothelial dysfunction and local hemodynamics support a process that includes retention and modification of lipids, recruitment of monocytes and other inflammatory cells, along with phenotypic switching of vascular muscle.¹⁵ Atherosclerosis is primarily a disease of large arteries,⁹⁷ but can extend into smaller vessels with the addition of hypertension,⁹⁸ and perhaps other risk factors.

Intracranial atherosclerosis is a common cause of ischemic stroke, but a form of atherosclerosis with unique features.^94–96 In both animal models and humans, intracranial atherosclerosis develops at a slower rate than extracranial atherosclerosis.^94–96 While fatty streaks (precursors of plaques) are present early in life, the number and size of these lesions is less in cerebral than in extracranial arteries.⁹⁹ Intracranial atherosclerosis progresses with age and is very common in the elderly.^{95, 96, 100, 101} Significant racial differences exist with cerebral atherosclerosis being more common in African Americans, Hispanics, and Asians.^95–97 Its rate of progression is accelerated when combined with select risk factors which include hypertension and diabetes.^{95, 96} A similar accelerating effect is seen in animals models.⁹⁸ The role of lipids and effects of tobacco smoking appear to be less for cerebral atherosclerosis compared to arteries outside the brain.^{95, 96} There are also suggestions that sex differences may be less prominent with intracranial atherosclerosis.^{95, 96}

Advanced atherosclerosis can become stenotic, physically encroaching on the vessel lumen,⁹⁸ thus affecting downstream or collateral perfusion (Figure 3). These lesions can also become unstable and rupture, with resulting vasoconstriction, platelet activation, and thrombosis. In addition to potential effects on hemodynamics and thrombosis, intracranial atherosclerosis has been linked to Alzheimer’s disease.^{100, 102} For example, the more severe the extent of atherosclerosis, the lesser the performance over a range of cognitive domains and the greater risk for dementia.¹⁰⁰ Intracranial atherosclerosis may impact Alzheimer’s disease by contributing to hypoperfusion but also other mechanisms including increased processing of APP and reduced clearance of β-amyloid.¹⁰³ Because it has pro-oxidant and pro-inflammatory effects, β-amyloid may also promote atherogenesis.¹⁰³

Figure 3. Changes in vascular structure and mechanics.

Major structural and mechanical changes in the vasculature (shown in cross-section) that collectively reduce the vascular lumen, affect vasodilator responses, and limit vasodilator reserve (increase minimal vascular resistance). Physiological consequences of these changes are listed on the bottom.

Why atherosclerosis develops more slowly in brain is not clear, but several possibilities might be considered. Mechanistic studies that focus on intracranial atherosclerosis are relatively rare, but it seems very likely that fundamental differences in endothelial cells, recruitment of immune cells, and the balance between pro- and anti-atherogenic mechanisms must exist. Pivotal transcription factors that determine the rate of atherosclerosis progression outside of brain include endothelial NF-κB and Kruppel-like factor 2.¹⁵ Thus, differences between hemodynamics, key molecular integrators as well as the impact of endothelium-derived NO may play major roles. Intracranial arteries may be less susceptible to hypercholesterolemia and ox-LDL in part because of differences in antioxidants,^{99, 101} known to affect progression of experimental atherosclerosis.¹⁰⁴ In addition, the gut microbiota can affect the rate of progression of experimental atherosclerosis.^{105, 106} Whether this or other microbiota affects intracranial atherosclerosis is unknown. Gut microbes can affect platelet responsiveness and clot formation,¹⁰⁷ important factors in hemostasis and thrombosis.

In relation to specific mechanisms, Ang II promotes atherosclerosis via the AT1R with downstream effects on plaque composition and stability. Atherosclerotic lesions express major components of the RAAS.⁷⁵ Genetic deficiency in AT1R results in reductions in lipid content, superoxide, immune cells, pro-inflammatory cytokines, and matrix metalloproteinase (MMP) activation within the vessel wall - effects that collectively result in smaller and more stable atherosclerotic plaques.^{75, 108} Of note, the local RAAS contributes to the progression of atherosclerosis even in the absence of increases in plasma Ang II.⁷⁵ In addition to effects on atherosclerosis, AT1R deficiency decreases the rate of amyloid deposition and β-amyloid production in an Alzheimer’s disease model via effects on the γ–secretase complex.¹⁰⁹

Hypercholesterolemia

Dietary and genetic models of hypercholesterolemia continue to be used to study effects of hyperlipidemia on the cerebral circulation. Several concepts have emerged from these efforts. Some studies found that endothelial function in cerebral arteries was normal in animals that exhibit atherosclerotic lesions and endothelial dysfunction in the carotid artery and aorta.^110–112 Acute treatment with ox-LDL impairs endothelium-dependent relaxation in the carotid but not the basilar artery.¹¹³ Thus, intracranial arteries can exhibit some resistance to ox-LDL and chronic hypercholesterolemia, experimental findings consistent with the delayed development of intracranial atherosclerosis. In other studies, hypercholesterolemia was sufficient to produce vascular dysfunction in animal models including primates. For example, in models where cerebral arteries had minimal or no detectable atherosclerotic lesions, hypercholesterolemia reduced resting CBF, the influence of basally-produce NO on vascular tone,^114–116 as well as agonist- and platelet-induced endothelium-dependent relaxation.^117–120 Vasoconstrictor responses to endothelin-1 were substantially increased in hyperlipidemic mice.¹¹⁷ There is limited data in this area in human vessels, but contraction of cerebral arteries in response to platelets and leukocytes are enhanced in vessels with atherosclerotic lesions.¹²¹ In addition, other key vasodilator responses are reduced during hypercholesterolemia including NVC, effects of hypercapnia, and autoregulation of CBF.^{116, 122} In the microcirculation, there are increased interactions between both leukocytes and platelets with endothelial cells.¹²³ Lastly, hypercholesterolemia can affect pericytes and their interactions with endothelial cells¹²⁴ as well as producing age-dependent increases in BBB permeability.¹²⁵

Mechanisms that underlie cerebrovascular dysfunction with hypercholesterolemia are not well defined but include effects on NO signaling (Figure 2). There is evidence for decreased production of NO, with reduced phosphorylation of eNOS at Ser^1179.117 ROS and the Nox2-containing NADPH oxidase has been implicated for several endpoints.^{115, 118, 123, 125} ROCK can promote progression of atherosclerosis via several mechanisms including effects on NADPH oxidases and eNOS (Figure 2).⁹² Expression of both ROCK1 and ROCK2 are increased in brain microvessels in hyperlipidemic mice.¹²⁶ A substantial reduction in expression of the nuclear receptor peroxisome proliferator-activated receptor-γ (PPARγ) occurs during hypercholesterolemia.¹²⁶ PPARγ normally exerts protective effects in large and small blood vessels including promoting NO-dependent signaling while suppressing oxidative stress, superoxide-mediated endothelial dysfunction, inward vascular remodeling, and ROCK-mediated increases in vascular tone.^{49, 54, 86, 127, 128} Consistent with a protective role, reduced levels of PPARγ in macrophages and vascular muscle are associated with more advanced atherosclerosis.¹²⁹ Genetic interference with PPARγ in endothelium or vascular muscle enhances atherosclerosis.¹³⁰

Despite these advances, there is a notable lack of studies using preclinical models that develop intracranial atherosclerosis. As a consequence, we have limited insight into mechanisms that control its pathogenesis. To better understand this form of atherosclerosis, new models may be needed. Current models of hyperlipidemia may still be useful if animals are allowed to age sufficiently to develop cerebral atherosclerosis, thus mimicking the natural progression of the disease. Incorporation of major risk factors for intracranial atherosclerosis like hypertension can have marked effects on disease severity.⁹⁸ Importantly, we also have limited insight into the functional impact of regression of intracranial atherosclerosis along with underlying mechanisms. The findings that the lipid content within the vessel wall and endothelial function in the internal carotid artery rapidly improve with dietary lipid lowering in a primate model of atherosclerosis suggest that regression may have substantial vascular effects.¹³¹

Vascular structure and mechanics

In addition to function, changes in the structural or mechanical state of the vasculature has a major influence on vascular resistance and local hemodynamics, the extent of injury once ischemia is initiated, and the effectiveness of collateral-dependent CBF. During hypertension for example, vascular resistance increases due to both narrowing of the vascular lumen and loss of vessels. Such changes are among those commonly seen in both large and SVD. With time and the natural progression of disease, degeneration of cerebral arterioles occurs in humans and preclinical models.^132–136 Degeneration of the vessel wall and loss of BBB integrity promote passage of molecules out of the circulation along with microbleeds and reductions in microvascular density.^{137, 138}

Changes in vascular structure or dispensability can have important effects on resting CBF (hypoperfusion), vasodilator responses, and vasodilator capacity (Figure 3).^{3, 135, 136} Because this impairment can affect both submaximal and maximal vasodilation,^{22, 55} key adaptive responses such as NVC, conducted or flow-dependent vasodilation, and autoregulation can each be impacted. In this section, we summarize select examples and key concepts related to changes in vascular structure and mechanics.^{22, 35, 139}

Vascular hypertrophy

Increases in the cross-sectional area of the vessel wall (referred to as hypertrophy here) represent an adaptive response to reduce wall stress during hypertension. Vascular hypertrophy is common, being present in genetic, renal and pharmacological models of hypertension.^{35, 52, 139–144} Wall thickening is often described in cerebral arteries in humans with hypertension, particularly in the early phase of the disease.¹³⁸ In relation to its impact on vascular resistance, hypertrophy of the vessel wall during hypertension contributes to the reduction in lumen diameter that is present in maximally dilated vessels (Figure 3).^{55, 142} Vascular hypertrophy is not unique to hypertension however, occurring in models of hyperhomocysteinemia, oxidative stress, diabetes, and genetic interference with eNOS or PPARγ.^{51, 53, 127, 145–147} In contrast, atrophy or thinning of the arteriolar wall occurs with aging,¹⁴⁸ a process that may contribute to the increased frequency of microbleeds with age.^{6, 149}

Vascular remodeling

Remodeling is a term used by some to describe any structural change in the vasculature. Experts in the area argue that the term should be used specifically to describe changes in lumen diameter in fully dilated vessels that are not due to changes in wall stiffness or dispensability.¹⁵⁰ Thus, inward remodeling represents a rearrangement of the vessel wall such that lumen diameter is reduced even when the vessel is maximally dilated (Figure 3). Both vascular hypertrophy and inward remodeling occur during hypertension, the latter making the greatest contribution to the reduction in lumen diameter and therefore vascular resistance.^{55, 142} Structural reductions in lumen diameter can have predictive value in relation to cardiovascular events.¹⁵¹ Although detrimental hemodynamic effects are clear, the increase in vessel resistance that results from inward remodeling also protects distal vessels and the BBB from increases in upstream pressure.¹⁵²

Inward remodeling is seen most commonly in resistance vessels. It is present in both pial and parenchymal arterioles in some,^{52, 140–144, 148, 153–156} but not all models of hypertension.^{52, 140} Collectively, these findings and others⁵⁵ support the concept that Ang II is an important determinant of inward vascular remodeling. Further support for this concept comes from the observation that chronic infusion of a non-pressor dose of Ang II is sufficient to produce inward remodeling in cerebral arterioles.¹⁴³ In genetically hypertensive rats, inhibition of ACE is more effective than other antihypertensives in lessening inward remodeling and impairment of vasodilation during reductions in arterial pressure.⁵⁵ In humans with essential hypertension, there is evidence that inward remodeling occurs in small cerebral arteries.^157–159

In some disease models, vascular hypertrophy is present, but not inward remodeling.^{51, 53, 56, 145} An exception are models with genetic interference of PPARγ where both inward modeling and vascular hypertrophy occur.^{127, 146} It is noteworthy that one of the pleiotropic effects of PPARγ is inhibition of expression and function of AT1R.^{17, 128, 160} Thus, PPARγ interference mimics effects of Ang II. These studies highlight that loss of endogenous protective mechanisms is an additional cause of changes in vascular structure.

Some mechanistic features that underlie these changes have emerged. Oxidative stress or loss of eNOS-derived NO, in the absence of hypertension or Ang II, fail to produce inward vascular remodeling.^{51, 53} The pattern differs when Ang II is involved. Inward arteriolar remodeling in response to non-pressor and pressor doses of Ang II requires normal expression of Nox2.¹⁴³ In addition to activation of AT1R, Ang II transactivates epidermal growth factor receptor (EGFR) via a disintegrin and metalloprotease 17 intermediate.¹⁶¹ Ang II produces phosphorylation of EGFR in cerebral arterioles while pharmacological or genetic inhibition of EGFR prevents Ang II-induced hypertrophy, but not inward remodeling.¹⁵³ Many of the signaling molecules associated with AT1R and EGFR are localized in caveolae.¹⁶¹ Ang II increases vascular expression of caveolin-1, which plays an essential role in both inward remodeling and hypertrophy.¹⁵⁴ In contrast, Ang II increases expression of MMP9, but MMP9 is only involved in inward remodeling.¹⁵⁴ Other molecules that have been implicated in these changes are endothelin-1, aldosterone, MR receptors, and chloride channels (TMEM16A).^162–164

Vascular distensibility

Changes in vascular stiffness or distensibility commonly occur during hypertension and aging,¹⁶⁵ but can be seen with other vascular risk factors as well. In hypertension, pattern of changes differ between large and small vessels. Large arteries show collagen deposition with reduced distensibility,^{139, 166} while small vessels exhibit inward remodeling but no stiffening.¹³⁹ Importantly, both changes can contribute to reduced vasodilator capacity and thus affect CBF (Figure 3). With aging, loss of elastin and vascular muscle occur with a resulting reduction in arteriolar distensibility, changes that also contribute to impaired vasodilation.¹⁴⁸ A similar loss of elastin and vascular muscle has been described in cerebral arteries in subjects with Alzheimer’s disease.¹³⁶ In the absence of hypertension, mutations in Notch3 that cause cerebral autosomal dominant arteriopathy with subcortical infarcts and leukoencephalopathy (CADASIL), the most common known genetic cause of SVD, also produces reductions in arterial distensibility.¹⁶⁷

Vascular rarefaction

Rarefaction represents the loss of vessels, commonly capillaries and arterioles, within a vascular bed (Figure 3). A reduction in vascular density in these segments can occur with disease as well as in response to genetic factors or stimuli that impact vessel destruction or angiogenesis. In models of hypertension for example, there is loss of brain arterioles as well as capillaries.¹⁶⁸ Similar reductions in microvascular density are seen in humans with essential hypertension.^{157, 159} The vascular degeneration that has been described in small arteries and arterioles would contribute to this process as well.¹³⁷ Preclinical models of CADASIL exhibit age-dependent reductions in capillary density and CBF.^{42, 169}

Collateral vessels

Collateral vessels are connections between arteries or between arterioles in the microcirculation. In patients and preclinical models, the structural and functional status of collateral vessels determines the extent of cellular injury and death after occlusion of supply arteries upstream.^{170, 171} This segment of the vasculature contributes to maintenance of CBF during ischemic challenges, reductions in perfusion pressure, or loss of autoregulation. The status of collaterals may also impact the outcome of vascular-based stroke therapies. Their location, number, and diameter determine the effectiveness of collaterals. The collateral network within the pial circulation is extensive.¹⁷² In contrast, penetrating arterioles within the parenchyma often travel long distances with few branches.^{7, 135, 173} Faber and others have provided recent insight into collateral size and numbers as well as the impact of genetics, vascular risk factors, and aging on the pial collateral network.^{170, 172} Hypertension produces a progressive loss of collaterals (both vessel size and diameter) in the pial circulation. These changes occur much earlier than the loss of collaterals that occurs normally with aging.^{170, 172} In addition, models of metabolic syndrome and leptin deficiency (used as a model of obesity) exhibit similar changes.¹⁷² Vascular changes of this magnitude substantially increase resistance of collateral vessels and are associated with increased infarct size in models of ischemia.^{170, 172} One of the normal functions of eNOS-derived NO is maintenance of vascular collaterals.^{170, 172} Loss of this protective molecule appears to account for rarefaction of collaterals in the pial circulation in the face of vascular risk factors.^{170, 172}

In contrast to efforts toward defining structural changes, very little has been done related to understanding mechanisms that control vascular tone in cerebral collaterals. A recent study found that myogenic tone and reactivity are quite different in collateral versus non-collateral arterioles.¹⁷⁴ During hypertension for example, collaterals have increased tone and impaired vasodilation which may result in lower levels of collateral-dependent blood flow, contributing to greater infarct size in response to ischemia.¹⁷⁴

Part II. Neurovascular Injury and Repair after Ischemic Stroke

Our current understanding of the mechanisms of brain injury following ischemia and I/R and repair include multicellular interactions, involving the BBB, activation of glia, immune cell infiltration, and neuronal death. In contrast to the more traditional neurocentric view, recent studies emphasize the contribution of non-neuronal components of the NVU to brain injury and repair after stroke. This section highlights recent progress in our understanding of the cellular and molecular mechanisms underlying vascular injury and repair after ischemic stroke.

Effects of ischemia on the regulation of CBF

Ischemic-induced injury affects regulation of CBF as well as integrity and function of the BBB. Insight into these effects and the importance of preserving vascular function following stroke has begun to emerge. Here we summarize select effects of ischemia (and I/R) on the vasculature.

Endothelial function

Early work in this area revealed that global I/R impairs endothelium-dependent regulation of vascular tone,^175–177 with some evidence implicating ROS as mediators of dysfunction.¹⁷⁷ Other preclinical studies found that focal I/R impairs both basal and receptor-mediated endothelium-dependent vasodilation.^{34, 178–181} Impaired NO-mediated responses occurred despite an increase in eNOS expression after ischemia.¹⁸² Mechanistically, these effects were due to scavenging of NO by superoxide as treatment with superoxide dismutase (SOD) or related mimetics normalize vascular responses after ischemia.¹⁸⁰ The finding that the influence of NO on vascular tone was not impaired after ischemia in Nox2-deficient mice indicates NADPH oxidase plays a key role.¹⁸⁰ Interest in this area has now extended to parenchymal arterioles. Similar to findings in large arteries, the influence of endothelium-derived NO on vascular tone is impaired in parenchymal arterioles following ischemia.^{34, 183}

NVC

Both global and focal ischemia impairs NVC in preclinical models.^{181, 184, 185} For example, glucose utilization and the increase in CBF with somatosensory activation is decreased following global I/R.¹⁸⁶ As neural activity recovers after reperfusion, these findings suggest that communication between neurons and the vasculature is impaired after ischemic injury. In contrast, there may be a generalized and more sustained suppression of brain function in addition to impaired NVC following permanent focal ischemia in regions remote from the infarct.¹⁸⁷ There is currently limited insight into mechanisms that contribute to these changes. Vascular responses to direct activation of vascular muscle were not affected by I/R, suggesting communication between cells is impaired, not dysfunction of smooth muscle per se.¹⁸¹ Interestingly, impaired NVC following focal I/R is prevented by lipopolysaccharide (LPS)-induced ischemic tolerance.¹⁸¹ This protection is dependent on iNOS- and Nox2-derived peroxynitrite.¹⁸¹ Another study suggested that decreased function of inward rectifier K⁺ channels mediate impaired NVC following global I/R.¹⁸⁸

Autoregulation

Ischemic injury has heterogeneous effects on the ability of the vasculature to respond the changes in perfusion pressure. Both myogenic tone and myogenic reactivity are reduced in cerebral arteries after I/R,^{34, 188–191} but are not sex-dependent.¹⁷⁸ Mechanisms that contribute to impaired myogenic tone include oxidative and nitrosative stress. The content of F-actin is reduced in cerebral arteries after ischemia, an effect that was dependent on formation of peroxynitrite.^{189, 190, 192}

In contrast to arteries upstream, basal tone and myogenic reactivity are intact^{34, 183, 193} or elevated⁴⁷ following I/R in parenchymal arterioles. This effect may be due to changes in calcium sensitivity and activation of ROCK.⁴⁷ The preservation or increase in myogenic tone in parenchymal arterioles after ischemia would be predicted to offset potential increases in CBF that would occur as a result of decreased myogenic tone of larger arteries upstream. Compromised CBF may lead to an expansion of brain injury and cell death.

In vivo studies in this area are generally consistent with studies of isolated vessels. When hypotension was induced during or after global ischemia, CBF was significantly reduced, an effect that was not observed during non-ischemic conditions.¹⁹⁴ Focal ischemia impairs autoregulation to both hypotension and hypertension with CBF passively following changes in arterial pressure.¹⁹⁵ Autoregulation was also impaired in areas of milder ischemia; with persistent effects 24 hours post injury.^{195, 196}

Structural injury and repair of microvascular components within the NVU

At the level of the NVU, endothelial cells and pericytes form the basic structure of the BBB. BBB damage is one of the most disabling consequences of stroke. Loss of BBB integrity after ischemia permits penetration of intravascular proteins, fluid, and immune cells into the extracellular space, resulting in vasoactive edema and expansion of tissue damage.¹⁹⁷ Severity of BBB damage predicts neurological outcome after stroke.¹⁹⁸ While it is well accepted that BBB disruption occurs early after ischemia, the exact temporal profile remains vague. Some studies reported a monophasic BBB leakage, starting as early as 25 min after post-ischemic reperfusion and lasting for 5 weeks.^{199, 200} In other studies, a biphasic increase of BBB permeability was observed, with the first peak occurring within hours and the second peak 2 to 3 days after stroke.^{201, 202} These discrepancies may be due to differences in stroke models or experimental conditions. Nevertheless, structural alterations in endothelial cells or pericytes are consistently regarded as the mechanisms underlying early BBB leakage.

Endothelium is at the center for neurovascular injury and repair after ischemia

Cerebral endothelial cells are the foundation of the BBB. They are specialized non-fenestrated cells sealed by highly restrictive TJs. Intact endothelium plays a critical role in controlling the exchange of ions, molecules and cells between the CNS and the periphery. Structural changes in endothelial cells or related TJs are the first steps to open the BBB after ischemic injury.

Cytoskeletal rearrangement in endothelial cells: Opening of paracellular pathways early after stroke

Cerebral I/R rapidly initiates vascular changes including increases in paracellular permeability. Such early stage hyperpermeability is usually not accompanied by overt injury or detachment of endothelial cells from the vessel wall. Instead, it reflects more subtle changes, loosening of the endothelial paracellular junctions.²⁰³ Structurally, the interface of adjacent endothelial cells are fused together by intercellular junctions, including TJ and adherens junctions (AJ). Both junctions are highly specialized protein complexes. The main components of the TJ complex (occludin and claudin) and the junctional adhesion molecules, are anchored to the actin cytoskeleton through TJ accessory or anchoring proteins [zonula occludens (ZO)-1, ZO-2, ZO-3)].²⁰⁴ Cadherin, the main component of AJ, is also stabilized by connections to actin filaments.²⁰⁵ The dynamic interaction between actin cytoskeleton and junctional proteins is critical for regulation of junctional integrity and endothelial permeability.

In normal conditions, actin filaments are distributed throughout endothelial cells as short filaments of F-actin at the cell marginal band (cortical actin).^{206, 207} F-actin functions to sustain the shape of endothelial cells as well as maintain the integrity of TJ.²⁰⁸ In response to certain stressors, the actin filaments are polymerized into linear stress fibers. This polymerization is often accompanied by actomyosin contraction and increased cytoskeletal tension, which result in a contracted cell morphology and impaired junctional sealing efficiency.^209–212 Such cytoskeletal rearrangement results in increased permeability and BBB leakage.

Several signaling events are important in the regulation of actin dynamics (Figure 4). Among them, phosphorylation of myosin light chains (MLC) is critical for actin-myosin contraction and disruption of endothelial cell-cell junctions. MLC is phosphorylated by MLC kinase (MLCK) in a Ca²⁺/calmodulin-dependent manner. RhoA, a small GTPase, and its downstream effector ROCK also potentiate MLC phosphorylation by either direct action or inhibition of MLC phosphatase (MLCP) activity.²¹³ In addition, RhoA is involved in the signaling pathway that controls actin stress fiber formation.²¹⁴ A variety of I/R related factors, including hypoxia, ROS, and cytokines, contribute to the disarrangement of cytoskeletal and junctional proteins in brain endothelial cells by activating MLC signaling.^{209, 210, 212, 215} It has been recently reported that early BBB disruption after ischemia was caused by the activation of ROCK/MLC signaling, which was accompanied by persistent actin polymerization and disassembly of junctional proteins within endothelial cells.²¹⁶ This study suggests that cytoskeleton rearrangement and structural alterations in endothelial cells are a novel mechanism for early BBB disruption after stroke, which in turn contributes to late stage MMP-dependent BBB opening. More importantly, early in the disease process, BBB dysfunction may be a cause rather than a consequence of parenchymal cell injury. Therefore, stabilizing endothelial cell structure may represent an overlooked therapeutic opportunity to target the early disturbance in BBB integrity and to prevent subsequent adverse events after stroke.

Figure 4. The structural alterations in endothelial cells are critical for BBB opening early after ischemic stroke.

A. The opening of paracellular pathways: cytoskeletal rearrangements and related signaling in endothelial cells. B. The transcellular pathway of BBB leakage. ADF: actin-depolymerizing factor; CaM: calmodulin; LIMK: LIM kinase; MLC: myosin light chain; MLCK: MLC kinase; MLCP: MLC phosphatase; MMP: matrix metallopeptidase; ROCK: Rho kinase TESK1: testicular protein kinase 1.

Recent work has highlighted the importance of cofilin/actin-depolymerizing factor (ADF) proteins in stabilizing endothelial structure by regulating actin dynamics (Figure 4). The binding of activated ADF or cofilin proteins to actin filaments results in actin depolymerization, reduced stress fiber formation, stabilized endothelial structure, and maintained BBB integrity.^{210, 217} Both cofilin and ADF are activated when dephosphorylated. LIM kinase and testicular protein kinase 1 (TESK1) are two molecules that phosphorylate cofilin/ADF at Ser-3 in a ROCK-dependent or -independent manner, respectively, which leads to cofilin/ADF inactivation, microtubule destabilization and actin polymerization.^218–220 However, controversial evidence suggested hyperactive cofilin may produce loss of BBB integrity by disassembling F-actin, decreasing the expression of TJ proteins^{221, 222} or redistributing actin and TJ proteins.^{218, 223} Such an apparent discrepancy may be attributed to the differences in the sources of endothelial cells and the dynamic changes of endothelial structure under different challenges. In the case of brain ischemia, overexpression of constitutively active ADF in endothelial cells reduces actin polymerization and junctional protein disassembly, attenuates early BBB leakage, and improves long-term histological and neurological outcomes. By contrast, ADF inactivation led to sustained actin polymerization and junctional disruption in endothelial cells. Therefore, maintaining BBB integrity might be achieved by targeting ADF activity early after stroke.²¹⁶

Transcytosis: Another mechanism for BBB leakage after ischemia

During transcytosis, vesicles containing various cargo molecules are trafficked across the interior of a cell.²²⁴ A recent study reported that the early increase of BBB permeability after stroke is due to increased transcytosis in endothelial cells. Interestingly, in contrast to the concept of cytoskeletal rearrangement-mediated early BBB damage, upregulation of transcellular pathways preceded the opening of paracellular pathways (Figure 4).⁴⁶ Increased transcytotic vesicles were observed in endothelial cells as early as 6 hours post stroke while the paracellular pathways of the BBB were not impaired until 48 hours after stroke. This observation is consistent with previous reports,²²⁵ including the demonstration of increased vesicles in the endothelial cells in the peri-infarct cortex at 3 hours post ischemia in both aged and young mice.²²⁶ A study in diabetic mice also suggested that loss of BBB integrity after stroke is primarily attributed to increased transcytosis.²²⁷

The debate about the importance of transcellular and paracellular pathway for BBB leakage is centered at the function of caveolin-1. Caveolae, membrane microdomains formed within the plasma membrane, are the main organelles in transcytosis. Caveolin is a group of proteins that play key roles in plasma membrane invagination and caveolae formation.²²⁴ Among the family of caveolins, caveolin-1 is mainly distributed in endothelial cells and is known to be important in the regulation of BBB permeability.²²⁴ The expression of caveolin-1 and transendothelial vesicles increases prior to the disruption of TJ and BBB integrity.^{228, 229} Further studies showed that transcellular but not paracellular permeability of cortical blood vessels was ameliorated early after stroke in caveolin-1 deficient mice, suggesting that caveolin-1 is involved in transcellular endothelial leakage in response to stroke.⁴⁶ However, both in vivo and in vitro data obtained by another group suggested that caveolin-1-mediated transcellular mechanisms do not play a dominant role in BBB disruption up to 24 h after ischemia, disagreeing with the contribution of transcytosis in early BBB damage after stroke.²¹⁶ Such discrepancy may be explained by differences in the severity of ischemia or possibly genetic background. Interestingly, caveolin-1 also interacts with TJ proteins to influence intercellular transport under stroke and inflammation conditions. ^{230, 231} It is therefore possible that the two biological mechanisms regulating transcellular and paracellular barrier properties are interconnected. Further studies are required to define the relative importance of transcellular and paracellular pathways and their temporal sequence in early BBB dysfunction after stroke.

Pericytes: Versatile players in ischemia and reperfusion

Pericytes are components of the NVU located on the abluminal aspect of endothelial cells, sandwiched between astrocytes and endothelium and almost entirely embedded within the basal lamina.²³² As constituents of the BBB, pericytes not only exert barrier function themselves, but also influence physiological functions of endothelial cells, basal lamina and astrocytes.²³³ Pericytes also possess immunological functions, contributing to immune responses.²³⁴ In addition, pericytes may differentiate into other cell types and thus may be important for CNS renewal.²³⁵

Pericytes are early responders to brain hypoxia. After ischemic stroke, pericytes change from a quiescent flat shape into an ameboid morphology, while expressing specific proteins such as RGS5. Pericytes detach from the basal lamina as early as 1 hour after ischemic stroke,²³⁶ followed by the migration toward the hypoperfusion lesion.²³⁷ Detachment and migration of pericytes are associated with their secretion of MMPs. Pericytes have been shown to be an important source of MMP9 in basal lamina after ischemic stroke.²³⁸ The chemotactic factors controlling pericyte migration are not well characterized. Hypoperfusion in the peri-infarct area can induce pericyte migration. Hypoxia-induced vascular endothelial growth factor (VEGF) expression can also stimulate pericyte migration in a concentration-dependent manner.²³⁹ After arriving at the site of injury, pericytes display both beneficial and detrimental functions during I/R and contribute significantly to BBB damage and repair (Figure 5).

Figure 5. Pericytes play multifaceted roles in ischemia and reperfusion.

Pericytes display both beneficial and detrimental functions during ischemia and reperfusion phases, and contribute significantly to the BBB damage and repair. 1) Pericyte contraction and dilation regulate cerebral blood flow in the ischemic and peri-lesion areas. 2) Pericyte protects other NVU components through releasing protective/trophic factors such as nerve growth factor (NGF), neurotrophin-3 (NT-3), vascular endothelial growth factor (VEGF), angiopoietin (Ang-1) and glial cell line-derived neurotrophic factor (GDNF). 3) Phagocytotic pericytes help to eliminate dead or injured tissue in the ischemic core, which in turn mitigates local inflammation and reduce secondary tissue damage. 4) Pericyte-endothelial cell interaction promotes angiogenesis after stroke. 5) Pericytes have the potential to serve as an origin of NVU components during tissue repair after ischemic stroke.

Impact of contractile pericytes on capillary blood flow

Recent studies have highlighted possible roles for pericytes in regulation of capillary perfusion. However, some concepts that have arisen are controversial.^{22, 29, 30, 240} After ischemic stroke, hypoperfusion resulting from vascular occlusion leads to tissue damage. Re-flow of occluded vessels is crucial for tissue preservation and restoration. However, absence of re-flow in the microvasculature is common, impeding reperfusion of ischemic tissue. Contraction and relaxation of pericytes may contribute to control of capillary blood flow.^{29, 241} The lack of oxygen after ischemia promptly induces relaxation of pericytes which may help dilate blocked vessels and restore CBF. Multiple mediators including platelet-derived growth factor (PDGF)-β, adenosine and NO have been identified as potential regulators of pericyte contractility after stroke.^{29, 242} For example, PDGF-β affects contractility of pericytes depending on metabolic status.²⁴² In particular, PDGF-β contracts pericytes when the circulation is normal and relaxes pericytes during hypoperfusion. Higher levels of PDGF-β are evident after stroke. Increased PDGF-β after ischemia relaxes pericytes, increases microvessel diameter, and elevates blood supply in the microcirculation. Other studies suggest pericytes contract following ischemia due to a reduced energy supply and may subsequently die in a contracted state.^{29, 241} Thus, even with recanalization after ischemia, hypoperfusion may persist if red blood cells are unable to pass through capillaries due to constriction of pericytes and/or plugging by microthrombi (or immune cells).

Mechanistically, roles for excitotoxicity or oxidative or nitrosative stress in post-ischemic pericyte-induced contraction have been proposed.^{29, 241} In support of a ROS-dependent mechanism, hydrogen peroxide and radiation-induced production of ROS cause contraction of cultured pericytes.²⁴³ The precise mechanism(s) underlying oxidative stress-induced pericyte contraction is unclear. Possible mechanisms involve excessive calcium influx during reperfusion and inactivation of potassium channels or sodium/hydrogen exchangers.²⁴¹ In short, pericyte contractility may vary at different stages of I/R, potentially impacting capillary perfusion and stroke outcome.

In contrast to the aforementioned studies, Hill et al. provided evidence against a role for pericytes in regulating capillary perfusion in ischemic brain.³⁰ Based on several lines of evidence, they concluded that parenchymal arterioles are the final point in the vasculature where CBF can be modulated and it is constriction of vascular muscle in arterioles that causes hypoperfusion during ischemia.³⁰ This concept is consistent with the findings of maintained or increased myogenic tone in isolated parenchymal arterioles after ischemia.^{34, 47, 183, 193} Despite uncertainty regarding the role of pericytes in regulating capillary perfusion, this remains a potentially important area of research. Direct evidence regarding the influence of pericytes on capillary perfusion in vivo is still very limited. Pericytes are difficult to study in vivo due to heterogeneity along with a lack of selective molecular markers^{240, 244} or cell-specific promoters.

Pericytes protect adjacent cells against ischemic injury

Pericytes may express a variety of factors in response to ischemia that protect adjacent cells, including neurons and other NVU components. For example, ischemia induces PDGF-β expression by endothelial cells and PDGFR-β expression by pericytes.²⁴² Increased PDGF-β may protect pericytes from apoptosis and promote pericyte proliferation. PDGF-β also enhances the expression of neuroprotective factors including nerve growth factor (NGF) and neurotrophin-3 (NT-3) in pericytes. Moreover, increased pericyte production of NT3 activates astrocytes and raises astrocytic secretion of NGF.²⁴⁵

In addition to neuronal protection, pericytes may protect the BBB at several levels including TJ integrity, protecting endothelial cells from necrosis, and promoting angiogenesis in the ischemic brain. Pericyte-produced angiopoietin (Ang-1) fortifies TJ connections by enhancing the expression of TJ proteins including ZO-1.²⁴⁶ Glial cell line-derived neurotrophic factor (GDNF) produced by pericytes up-regulates the expression of claudin-5 in endothelium, promoting integrity of the BBB.²⁴⁷ In addition, pericyte-derived VEGF can enhance survival of endothelial cells and preserve their function in ischemic stroke.²⁴⁸ Pericytes can also secrete transforming growth factor (TGF)-β1, which enhances VEGF receptor 1 (VEGFR1) expression on endothelial cells and may improve endothelial survival in ischemia. Overall, the interactions of pericytes with endothelial cells, neurons and astrocytes support the preservation and reconstruction of the NVU after cerebral ischemia.^{249, 250}

Phagocytotic pericytes clear cell debris after stroke

Electron microscopy suggests there are two types of brain pericytes, granular and filamentous.²⁵¹ These subtypes can be distinguished based on the presence of cytoplasmic lysosome-like granules.²⁵² Pericytes containing lysosome-like granules can serve as scavenger cells in injured brain. In an animal model of ischemia where human atheroma samples were injected intravascularly, an increase in granular pericytes was detected in the injured brain area as early as 2 hours after injection. Moreover, these granular pericytes accumulated lipid components of the injected atheroma,²⁵³ suggesting a phagocytotic property of granular pericytes. With the capacity of multi-potent differentiation, pericytes can also acquire microglia phenotype after ischemia.^{254, 255} These phagocytotic pericytes and microglia derived from multi-potent pericytes may help to eliminate dead and injured tissue in the ischemic core, which in turn mitigates local inflammation and reduces secondary tissue damage.

Production of ROS by pericytes after stroke

In addition to the aforementioned beneficial effects, pericytes may release detrimental factors in response to ischemia. In particular, hypoxic stress induces ROS production by pericytes. NADPH oxidase (Nox4) is a major source of ROS in human brain pericytes.²⁵⁶ Nox4 is upregulated in pericytes after acute ischemia in peri-infarct areas and enhances BBB leakage by activating NF-κB and MMP9 production.²⁵⁷ The upregulation of Nox4 is greater in permanent compared to transient focal ischemia, suggesting that ischemia is a strong inducer for Nox4 and subsequent ROS production in pericytes. The increased level of ROS causes detrimental effects in other already compromised cells within the NVU. Pericytes themselves are sensitive to low concentrations of ROS and may be more fragile than endothelial cells. Apoptosis of pericytes was evident in ischemic areas after reperfusion, when production of ROS was elevated. Anti-oxidative treatment inhibited pericyte death after I/R,²⁴³ suggesting a key role for ROS in the demise of pericytes.²⁹ Overall, pericytes are much more than supportive cells to endothelium. They have vital functions within the BBB and NVU. Pericytes display morphological and functional alterations during ischemia and reperfusion, and may be involved in multiple processes after stroke.

Pericyte-endothelial interactions in angiogenesis and neovascularization

A neurovascular regenerative program is activated after stroke as a mechanism of brain repair. The capacity of such neurovascular repair is critical for long-term recovery after stroke. Angiogenesis, or the growth of new blood vessels, is an important component of BBB remodeling. In addition to its potential to re-establish the blood supply to hypoperfused areas, early upregulation of angiogenic factors including VEGF and Ang-1 can promote cell survival and enhance the removal of cell debris.²⁵⁸ Interestingly, one study suggested that angiogenic responses were transiently activated after stroke to enhance macrophage infiltration and debris clearance, rather than increasing microvessel density.²⁵⁸ Such a “clean up alone” hypothesis is challenged however by a large number of animal and human studies showing positive effects of angiogenesis on post-stroke functional recovery.^{259, 260} Further investigations are warranted to confirm the causal relationship between angiogenesis and functional improvement after stroke.

The processes of angiogenesis include proliferation of vessel cells, recruitment of pericytes, coverage of endothelial tubes by pericytes, and maturation of newly formed vessels. Endothelial cells, pericytes, and the communication between these cells are essential for the regulation of angiogenesis.²⁶¹ Endothelial cells start to proliferate and grow vessel sprouts within 1 day after brain ischemia, leading to formation of new vessels in the peri-infarct region several days after ischemic injury.^{262, 263} Meanwhile, pericytes with upregulated PDGFR-β start to proliferate and migrate from the microvessel wall to the new vessel sprouts to foster their maturation.²⁶¹ Administration of a phosphodiesterase-3 inhibitor that promotes pericyte proliferation decreased the final infarct size by enhancing new vessel formation after stroke in hypertensive rats.²⁶⁴ In contrast, post-stroke angiogenesis was impaired in hyperlipidemic or diabetic mice with pericyte dysfunction and weakened pericyte-endothelial cell communication.^{249, 259} A recent study found that injection of blood-derived pericyte-like cells could rescue affected tissue by accelerating angiogenesis in a model of hind limb ischemia.²⁶⁵ This result indicates that transplantation of pericyte progenitor cells may be a promising therapy for ischemia and could be possibly applied in ischemic stroke. More intriguingly, pericytes isolated from the ischemic regions of mouse or human brains reveal mesenchymal multi-lineage developmental properties when cultured under oxygen/glucose depriving environment, and could differentiate into both neural and vascular lineage cells.²⁶⁶ Therefore, pericytes may have the potential to serve as an origin of NVU components during tissue repair after ischemic stroke.

Despite the potential beneficial effects of angiogenesis in post-stroke recovery, the elevation of angiogenic factors such as VEGF may increase vascular permeability and in turn contribute to edema or hemorrhage.²⁶⁷ A thorough understanding of the dynamics and mechanisms of angiogenesis is essential for developing an effective therapy for stroke.

Inflammation in BBB disruption after ischemia and reperfusion

Neurovascular inflammation involves a complex interaction between endothelial cells, resident microglia, and invading leukocytes, and plays a critical role in BBB disruption following I/R. Cerebral I/R triggers expression and release of inflammatory mediators and proteases from endothelium and immune cells and causes BBB disruption.²⁶⁸ These destructive factors include: 1) MMPs that degrade the extracellular matrix; 2) chemokines, such as monocyte chemoattractant-1 (MCP-1) that attract peripheral leukocytes; 3) adhesive molecules, such as selectins and ICAM that enhance leukocyte-endothelial cell interaction; and 4) inflammatory cytokines that exacerbate inflammatory injury to components of the NVU. Meanwhile, local microglia are recruited to ischemic sites, where they secrete immune mediators and further activate endothelial cells, leading to BBB breakdown. In addition, the primed BBB permits the extravasation of neutrophils, monocytes, and other peripheral immune cells, which carry with them even more deleterious inflammatory mediators. Microvascular accumulation of these immune cells, pro-inflammatory mediators, and proteases not only potentiate junctional disassembly and endothelial malfunction but also degrade the extracellular matrix, resulting in irreversible BBB disruption. In this section, we focus on current concepts regarding the function of MMPs and effects of several immune cell populations in NVU injury and repair after ischemia.

MMPs in BBB disruption after ischemia and reperfusion

MMPs are a family of proteins specialized in the degradation of extracellular matrix and basement membrane. MMP2 and MMP9 can be released from a variety of cells including endothelial cells, glial cells, and recruited immune cells.²⁶⁹ Both these MMPs are strongly implicated in the disruption of the BBB following ischemic injury in both rodent models^{230, 270} and stroke patients.²⁷¹ Genetic ablation or pharmacological inhibition of MMP2/9 provides protection on BBB integrity and reduces brain damage after I/R.^{230, 270}

Interestingly, recent studies demonstrated that the increase of brain MMP2/9 seems to be relatively delayed compare to the early loss of BBB integrity.²¹⁶ The BBB becomes permeable to smaller molecules (≤10 kDa) very early (30 min) after cerebral I/R. In contrast, active MMP9 and MMP2 were observed in the ischemic brain beginning at 3 hours after focal ischemia, progressively increasing over the course of 24 hours.²⁷² In vitro studies suggest that the inhibition of MMP2/9 activity greatly reduced changes in endothelial permeability at 4–6 hours after oxygen-glucose deprivation (OGD), but failed to protect the early breach of endothelial cell integrity at 1–3 hours after OGD. In vivo animal studies showed that MMP2 or MMP9 knockout or inhibition can prevent BBB leakage of larger molecules (>40 kDa) at 3 or 24 hours after focal ischemia but failed to reduce the leakage of smaller molecules, suggesting that MMP2/9 contributes to severe BBB disruption following I/R in a relatively delayed manner. Notably, MMP2/9 ablation transiently reduced infarct volume at 24 hours after focal ischemia but failed to provide long-term protection.²¹⁶ These data suggest that MMPs may represent a target for BBB protection after I/R; however, inhibiting MMPs alone is not sufficient to provide early and prolonged protection to the BBB or the ischemic brain.

It should be noted that MMPs are not always detrimental in ischemia as they serve critical functions related to stroke recovery and NVU remodeling.²⁷³ Treatment with MMP inhibitors 7 days after stroke increases ischemic brain injury and impairs functional recovery at 14 days. Bioprocessing of VEGF by MMP-9 was proposed as one of the underlying mechanisms of neurovascular remodeling after cerebral ischemia.²⁷⁴ Considering the dual role of MMPs at different phases after cerebral ischemia, timing is an important factor to be considered for therapeutic strategies targeting MMPs.

Neutrophils: A key player in early BBB disruption

Neutrophils migrate toward the injured brain within a few hours after stroke onset in response to the upregulation of adhesion molecules, such as ICAM-1 and P-selectin, on injured endothelium.^{275, 276} The importance of neutrophils in post-ischemic BBB damage is confirmed by the fact that inhibition or depletion of neutrophils mitigated BBB leakage after stroke and reduced the risk of hemorrhagic transformation (HT) after thrombolysis.^{277, 278} In contrast, triggering neutrophil activation with LPS greatly enhances BBB disruption in a rodent model of stroke.²⁷⁹

Neutrophils greatly contribute to BBB disruption through the release of a variety of proteases, including MMPs, elastase, cathepsin G, and proteinase 3.^{278, 280–282} Neutrophils are major sources of MMP9, so their infiltration into the ischemic brain enhances central MMP9 levels by releasing the MMP9 proform.^{280, 283} Reperfusion after tPA treatment promotes the degranulation of human neutrophils and release of MMP9.²⁸⁴ Clinical data revealed that MMP9-positive neutrophil infiltration is associated with BBB breakdown, basal lamina type IV collagen degradation and HT.²⁷¹ In addition to MMP9, neutrophil elastase is involved in BBB breakdown by degrading basal lamina and extracellular matrix. Accordingly, pharmacological inhibition or genetic ablation of neutrophil elastase reduced BBB damage and decreased cerebral edema.^{285, 286} Furthermore, inhibition of elastase in MMP9 deficient mice further decreased infarct volume and BBB disruption, suggesting that the contribution of neutrophil elastase and MMP9 to BBB damage is independent of each other.²⁸⁶ A therapeutic strategy with the capacity to inhibit both MMP9 and elastase may be more effective than targeting either of them individually.

Neutrophils are also important sources of ROS following I/R.^{287, 288} ROS disrupts the NVU through damage to multiple components including endothelium, pericytes, smooth muscle, neurons and astrocytes. These effects result in increased BBB permeability and HT. Superoxide is one of the most important mediators of BBB damage during reperfusion.²⁸⁹ Neutrophils express high levels of NADPH oxidase, a major source of superoxide. Inhibition or depletion of Nox reduces the degree of neutrophil infiltration, which is accompanied by ameliorated BBB disruption and reduced infarct volume.^{290, 291}

Lastly, neutrophils in the microvasculature may physically obstruct capillaries contributing to no-reflow during reperfusion.²⁹² Therefore, inhibition of neutrophil adherence to endothelial cells may promote the microvascular patency by reducing the no-reflow phenomenon in regions affected by I/R.²⁹³

Microglia/macrophage polarization influences BBB injury and repair

Recent studies highlighted the importance of microglia/macrophages phenotypes in brain injury and repair. Upon activation, microglia/macrophages can develop into a spectrum of different but overlapping functional phenotypes. The “classically activated” M1 phenotype and the “alternatively activated” M2 phenotype represent two ends of this spectrum, with a variety of phenotypes in between. M1-like microglia/macrophages are generally characterized by the release of destructive pro-inflammatory mediators. In contrast, M2-like microglia/macrophages typically release protective/trophic factors to preserve brain tissue or promote brain repair.²⁹⁴ Accumulating studies have documented the distinct functions of M1-like and M2-like microglia/macrophage in the BBB injury and repair, suggesting a possible therapeutic strategy to promote NVU integrity by modulating the phenotypic balance within these cell types.

Microglia/macrophage polarity in BBB injury after ischemic stroke

Activated M1 phenotype microglia may play a detrimental role in BBB integrity by eliciting the expression of pro-inflammatory cytokines such as interleukin (IL)-1β, IL-6, and tumor necrosis factor-α (TNF-α). Several M1 cytokines contribute to BBB damage after stroke. Interactions of TNFα with endothelial cells increase the paracellular permeability of the BBB by altering cytoskeletal organization and TJ expression and promoting production of MMPs.²⁹⁵ The upregulation of IL-1β after ischemia could induce endothelial expression of adhesion molecules, including ICAM-1 and vascular cell adhesion molecule (VCAM)-1, which in turn promotes neutrophil and other immune cell adhesion and infiltration into sites of injury.²⁹⁶ The M2 phenotype microglia, in contrast, enhance BBB integrity by increasing TJ expression after stroke.²⁹⁷

Effect of microglia/macrophage polarity in post-stroke angiogenesis and brain repair

Studies of vascular repair during wound healing have suggested phenotype-specific roles for macrophage in angiogenesis.^{298, 299} M2 macrophages are pro-angiogenic by releasing VEGF,³⁰⁰ IL-8,²⁹⁸ and a unique tissue inhibitor of metalloproteinases-1 (TIMP-1) free form of pro-MMP9.²⁹⁹ The influence of microglial phenotype on angiogenesis, however, has not been well-addressed. Nevertheless, the activation state of microglia has been shown to regulate brain endothelial cell proliferation.²⁹⁷ Specifically, the M1-type cytokine TNF-α and the M2-type cytokine TGF-β exhibit distinct effects on the proliferation of endothelial cells. Furthermore, minocycline and some other experimental treatments potentiate microglia/macrophage M2 polarization while promoting angiogenesis or BBB remodeling after stroke.^301–303 These studies suggest that the polarization state of microglia/macrophage might be a key regulator of angiogenesis and BBB repair after stroke.

Interestingly, the phenotypic diversity described above is not restricted to microglia/macrophages. Neutrophils also exhibit N1 and N2 phenotypes in ischemic brain, with N2 polarization being associated with resolution of inflammation.³⁰⁴ Such differential effects of distinct immune cell phenotypes on NVU injury and repair need to be further investigated.

Risk factors for NVU integrity after ischemic stroke

The effects of aging on vascular components of the NVU and their response to ischemia are evident. First, aging reduces endothelium-dependent regulation of vascular tone,^{17, 22, 77} the regenerative capacity of endothelial cells,³⁰⁵ and the number of circulating endothelial progenitor cells.³⁰⁶ Second, associations between pericytes and capillaries are reduced with aging.^307,308 Third, age-related alterations in microglial function are prominent making them less mobile and less efficient in CNS surveillance.³⁰⁹ Aged microglia tend to polarize into an M1 phenotype,^{310, 311} which may impair their function in NVU repair. Lastly, the function of other NVU components, including astrocytes, oligodendrocytes, and neurons are each compromised with aging.³¹² These structural and functional changes likely increase vulnerability to stroke and enhance ischemic brain injury in the elderly.^{313, 314}

Other vascular risk factors, including hyperlipidemia, diabetes and hypertension enhance BBB damage after ischemia. ^{315, 316} Increased lipid peroxidation, elevated protease activity, downregulation of TJ expression, and increased inflammation are potential mechanisms for dyslipidemia-enhanced BBB damage.^{315, 317} Hypertension is associated with disorganization of TJs, BBB dysfunction, and cerebral edema.³¹⁸ Diabetes also exacerbates BBB damage in different neurological disorders. Specifically, fluctuations in plasma glucose levels are associated with altered BBB transport, impaired TJ integrity, and elevated oxidative stress.³¹⁹ Hyperglycemia exacerbates inflammatory responses, such as cytokine expression and neutrophil infiltration,³²⁰ in the NVU. Consequently, vascular injury in response to ischemia is increased in diabetes.³²¹ In addition to endothelium and pericytes, other components of the NVU can be affected by vascular risk factors and ischemia. For example, hypertension produces focal swelling and other changes in astrocyte end feet around capillaries,^{251, 318} changes that likely contribute to loss of BBB integrity and may impact local regulation of CBF. Overall, treatment or prevention of these and other comorbidities may lessen NVU damage after ischemia and improve disease outcome.

Conclusions

In this review, we have outlined major mechanisms that regulate CBF along with features of large and small vessel disease that underlie ischemic events and impact their severity. Loss of endothelial health is a central player in the onset and progression of cerebrovascular disease. We highlighted common underlying mechanisms that contribute to structural and functional changes, particularly endothelial-based abnormalities. Damage to the BBB is one of the consequences of stroke with the greatest impact. How changes in endothelial TJs, the cytoskeleton, and the rate of transcytosis impact BBB integrity over time is gradually becoming clear. Protective and detrimental effects of pericytes on the BBB and other components of the NVU have begun to emerge. Lastly, the molecular basis of contributions by immune cells and related factors such as MMPs has been actively investigated. A thorough mechanistic understanding of CBF regulation, vascular, and NVU injury during the formation and development of ischemic stroke should provide therapeutic strategies to preserve post-stroke vascular integrity. Investigations into the impact of leading risk factors for cerebrovascular disease, stroke and BBB injury will provide further opportunities to prevent the occurrence of ischemic events and improve disease outcome.

Non-standard Abbreviations and Acronyms

ADF

actin-depolymerizing factor

AJ

adherens junction

APP

amyloid precursor protein

Ang-1

angiopoietin

Ang 1-7

angiotensin 1-7

Ang II

angiotensin II

ACE

angiotensin-converting enzyme

ADMA

asymmetric dimethylarginine

AT1R

AT1 receptor

BBB

blood-brain barrier

CaM

calmodulin

CADASIL

cerebral autosomal dominant arteriopathy with subcortical infarcts and leukoencephalopathy

CBF

cerebral blood flow

DDAH

dimethylarginine dimethylaminohydrolase

eNOS

endothelial NO synthase

EGFR

epidermal growth factor receptor

GDNF

glial cell line-derived neurotrophic factor

HT

hemorrhagic transformation

HUVECs

human umbilical vein endothelial cells

I/R

ischemia with or without reperfusion

LPS

lipopolysaccharide

MMP

matrix metalloproteinase

MR

mineralocorticoid receptor

MLCK

MLC kinase

MLCP

MLC phosphatase

MCP-1

monocyte chemoattractant-1

MLC

myosin light chain

Nox

NADPH oxidases

NGF

nerve growth factor

NT-3

neurotrophin-3

NVU

neurovascular unit

NVC

neurovascular coupling

NO

nitric oxide

ox-LDL

oxidized low-density lipoproteins

OGD

oxygen-glucose deprivation

PPARγ

peroxisome proliferator-activated receptor-γ

PDGF

platelet-derived growth factor

ROS

reactive oxygen species

RAAS

renin–angiotensin–aldosterone system

ROCK

rho kinase

SVD

small vessel disease

SOD

superoxide dismutase

TESK1

testicular protein kinase 1

BH₄

tetrahydrobiopterin

TJ

tight junction

TIMP-1

tissue inhibitor of metalloproteinases-1

TGF

transforming growth factor

TRP

transient receptor potential channels

TRPV4

transient receptor potential V4 channels

TNF-α

tumor necrosis factor-α

VCAM

vascular cell adhesion molecule

VEGF

vascular endothelial growth factor

VEGFR1

VEGF receptor 1

ZO

zonula occluden