Stroke and the neurovascular unit: glial cells, sex differences, and hypertension

Abstract

A functional neurovascular unit (NVU) is central to meeting the brain’s dynamic metabolic needs. Poststroke damage to the NVU within the ipsilateral hemisphere ranges from cell dysfunction to complete cell loss. Thus, understanding poststroke cell-cell communication within the NVU is of critical importance. Loss of coordinated NVU function exacerbates ischemic injury. However, particular cells of the NVU (e.g., astrocytes) and those with ancillary roles (e.g., microglia) also contribute to repair mechanisms. Epidemiological studies support the notion that infarct size and recovery outcomes are heterogeneous and greatly influenced by modifiable and nonmodifiable factors such as sex and the co-morbid condition common to stroke: hypertension. The mechanisms whereby sex and hypertension modulate NVU function are explored, to some extent, in preclinical laboratory studies. We present a review of the NVU in the context of ischemic stroke with a focus on glial contributions to NVU function and dysfunction. We explore the impact of sex and hypertension as modifiable and nonmodifiable risk factors and the underlying cellular mechanisms that may underlie heterogeneous stroke outcomes. Most of the preclinical investigative studies of poststroke NVU dysfunction are carried out primarily in male stroke models lacking underlying co-morbid conditions, which is very different from the human condition. As such, the evolution of translational medicine to target the NVU for improved stroke outcomes remains elusive; however, it is attainable with further research.

INTRODUCTION

Modified from the original phrase “time is money” (Benjamin Franklin in 1748), neurologists and stroke scientists coined the phrase “time is brain,” which speaks to the brain’s currency. Time without blood flow results in irreversible brain damage. It is estimated that an average stroke patient with a blocked large vessel will lose ∼7 miles of axonal fibers, 13.8 billion synapses, and 1.9 million neurons for each hour that ischemia persists (171). Unlike other organ systems, the brain has little capacity to store energy and, therefore, relies on a precisely regulated and steady blood supply for nutrient delivery and waste removal (39). An intricate process, known as neurovascular coupling, fulfills the “on-demand” and diverse cellular metabolic needs resulting from variable levels of neuronal activity simultaneously occurring throughout the brain (182). The neurovascular unit (NVU), having both physical and molecular signaling attributes, is key to carrying out neurovascular coupling (224). In the traditional sense, elements of the NVU are neurons, astrocytes, and vasculature. Brain cells once considered peripheral to the NVU, microglia and oligodendrocytes, may indirectly contribute to neurovascular coupling, as they contribute to NVU homeostasis in health and disease conditions. In contrast to neurovascular coupling, a finely tuned process (recently reviewed in Ref. 92), is autoregulation, a more coarse method of ensuring blood flow. Mechanisms of autoregulation ensure that blood flow is consistent when systemic blood pressure varies (221). These two distinct mechanisms, coarse and fine, highlight the physiological importance and dedication to balancing the brain’s energy supply and demand. Coordinated physiological processes at the NVU maintain brain homeostasis and prevent pathologies such as cognitive impairments. Ischemic stroke, however, severely interrupts both the physical and molecular signaling aspects of the NVU, so that it is either entirely lost in some regions or merely dysfunctional in other regions. In this review we highlight the current knowledge of the effects of ischemic stroke on the NVU, with particular focus on astrocytes and microglia intercellular communication in health and in response to, or mediation of, the deleterious poststroke parenchymal milieu. The heterogeneity of ischemic stroke in terms of risk and severity is due to both modifiable and nonmodifiable factors, such as sex and co-morbid conditions; whenever possible, a discussion of these factors is included.

ISCHEMIC STROKE

Stroke is the fifth-leading cause of death and disability in the United States (21). There are two major types of strokes: ischemic and hemorrhagic. Ischemic stroke is caused by a thrombus or emboli blocking blood flow in the vasculature or by decreased perfusion (132) and is, therefore, a clinical case of the ultimate imbalance between blood flow supply and demand in the brain. Ischemic strokes are by far the most prevalent (~80%); however, hemorrhagic strokes are considerably more deadly, because, with vessel rupture, blood accumulates within the skull and results in sudden and dramatic parenchymal compression and displacement and overwhelming rapid cell death (21). Cell death also occurs with ischemic stroke. Without a continuous source of energy [cerebral blood flow (CBF) ~40 ml·100 mg⁻¹·min⁻¹], the electrical function of neurons is lost (CBF less than ~12 ml·100 g⁻¹·min⁻¹) and cell membrane failure occurs (CBF 6–8 ml·100 g⁻¹·min⁻¹), resulting in cell death from the lack of energy reserves (132). Dying neurons in the ischemic core release cell contents (e.g., ATP, glutamate, and K⁺), which diffuse into the surrounding parenchyma and initiate a deleterious cascade of events. Peri-infarct regions near the ischemic core are viable, but at-risk, parenchyma (46, 81). Although not ischemic, the peri-infarct region is hypoperfused and subjected to deleterious signals originating from the ischemic core that can result in peri-infarct-spreading depressions, neuronal hyperexcitability, biochemical and molecular changes, and vascular impairments (86). These peri-infarct events exacerbate the supply-and-demand deficit that overwhelms mechanisms of functional hyperemia and severely disrupts neurovascular coupling (165). From a vascular standpoint, therapies to limit ischemic cell death are singular: reperfusion. Reperfusion is the restoration of blood flow to the ischemic brain, and although it may occur spontaneously, more often administration of an antifibrinolytic, such as recombinant tissue plasminogen activator, or endovascular retrieval is needed for best outcomes (188). With continued ischemia (i.e., no reperfusion therapy), the ischemic core will expand to envelop the peri-infarct region. A severe limitation has been that the treatment window is short and finite, however recently expanded (152), and thus not all stroke patients will benefit from this singular treatment option.

Both nonmodifiable (e.g., genetics, age, and sex) and modifiable (e.g., hypertension) risk factors have an impact on stroke incidence, severity, and associated neurological outcomes. Sex differences are well reported in both human and rodent studies of ischemic stroke. Epidemiological findings suggest less infarct risk and severity in premenopausal women than age-matched men (134). However, in women (48–55 yr of age), this protection is reversed following the postmenopausal decline in gonadal sex hormone levels; the average age for stroke is 75 yr in women and 71 yr for men (104). A history of clinical trials investigating the associations between long-term sex hormone treatment and stroke risk or severity conflicts with the abundant preclinical research (13) and laboratory evidence that suggest that estrogen therapy, supplied acutely after stroke, is protective. The stark contrast between clinical and preclinical evidence dampens the enthusiasm to entertain estrogen as a poststroke therapeutic. In 1985, a prospective study of the Framingham Heart Study cohort concluded that women with extended estrogen use for postmenopausal symptoms had an increased relative risk for stroke morbidity (211). In the early 1990s, the Women’s Health Initiative (WHI) began a large-scale investigation into the factors that affect the health of postmenopausal women. Two randomized and double-blind clinical trials emerged from the WHI specifically to investigate the relationship between sex hormone treatments and cardiovascular outcomes but were terminated early due to a harmful risk-to-benefit ratio most notable in ischemic stroke. In 2002, it was concluded that the combined use of estrogen and progesterone by healthy women significantly increased the risk of ischemic stroke, independent of hypertension, by >40% (209). Two years later, a second study that aimed to investigate estrogen alone was terminated due to similar outcomes (83). In both cases, gonadal hormones did not affect stroke severity (evidenced by the Glasgow Coma Scale) or stroke outcomes (83, 209), for better or worse. A more thorough analysis of these studies now indicates a complex role of estrogen replacement in cerebrovascular disease. An initial hypothesis, named the “timing” or “window” hypothesis, which emerged as an explanation of these initial findings, was that estrogen therapy had no effect or was harmful regarding risk following a period of estrogen deficiency. Rather, the benefits of estrogen therapy could be realized when treatment occurred during perimenopause or when initiated soon after menopause (17). However, because the WHI was not designed to test this hypothesis specifically, this interpretation is limited. Nonetheless, the age interval of hormone replacement therapy implementation is now a consideration for further investigation, and preclinical research provides some support of the timing hypothesis. In rodents, treatment with estrogen after 10 wk of estrogen deficiency, somewhat mimicking the WHI clinical trials, resulted in no neuroprotection and also suppressed estrogen’s anti-inflammatory actions, effects that were reversible when estrogen was administered immediately with ovariectomy (190). On the other hand, additional analyses of the WHI study, specifically designed to examine the timing hypothesis, were not supportive. In these studies, the effects of estrogen on WHI clinical outcomes, such as ischemic stroke, did not depend on the time between menopause and first use of the hormone therapy (14, 155).

Despite the findings of clinical investigations, a plethora of preclinical laboratory research has examined the fundamental phenomenon that stroke is less severe in premenopausal women than in men and postmenopausal women. A variety of ischemic stroke models have demonstrated that treatment with either estrogen or progesterone lessens infarct size after stroke when delivered to postmenopausal female or male rodent stroke models (4, 7, 34, 120, 183, 215). Gonadal hormones, in particular estrogen, are implicated in poststroke neuroprotection via multiple mechanisms (4, 105, 127). For example, estrogen promotes cell survival signaling and antioxidant pathways to lessen neuroinflammation (96, 162, 187) and edema (141), which are prominent sources of secondary injury in ischemic stroke. In addition, when administered during reperfusion to increase CBF and reduce infarct injury, estrogen has a vasodilatory effect (129). On the other hand, a treatment regimen of high doses of estrogen in older acyclic females increased infarct size, an outcome that was attributed to the decreased availability of insulin-like growth factor-1 (IGF-1) (174). It is likely that these preclinical findings do not coincide with clinical research for the following reasons. 1) Whereas preclinical studies investigate the effects of sex hormones on stroke severity and injury-related outcomes, they do not examine stroke risk, which was the primary finding of clinical research investigations. Although long-term treatment regimens of estrogen may have a poor risk-to-benefit ratio, the use of sex hormones as an acute treatment of ischemic stroke to lessen brain injury is a more viable option that, when mechanistically understood, should be considered. 2) Most preclinical studies used the surgical removal of ovaries (ovariectomy) as a model of menopause to study postmenopausal ischemic stroke (4); however, only 10% of women enter menopause surgically. Rather, the majority of women enter menopause gradually via the reduction of ovarian function and retention of ovarian tissue, a stark contrast to the ovariectomy model. This limitation can be addressed, while an age-matched experimental design is retained, by using an alternative model such as the 4-vinylcyclohexene diepoxide mouse model of menopause (27, 74, 200, 201).

ASTROCYTE CONTRIBUTIONS TO THE NVU

Over the past nearly five decades, our understanding of astrocyte function has transitioned from trophic, supportive, and metabolic-related to highly complex roles (for a detailed review see Ref. 203). Astrocytes account for 20–40% of all glial cells in the human brain and 10–20% in the rodent brain (189, 203) and are a highly heterogeneous cell population with brain region specificity (203). Diverse intercellular interactions with neurons, microglia, and vascular cells place astrocytes in a dominant position to influence numerous processes across the brain. In rodents, a single (protoplasmic) astrocyte is in contact with 20,000–120,000 synapses (30, 76, 203) and capillaries, allowing a bridge of information from neurons to blood vessels, and vice versa. In addition, extensive gap junction coupling, which forms the astrocytic syncytium, supports the coordination of neuronal networks and cellular responses over greater distances (33, 35). Significant astrocyte functions include synapse formation/elimination (37, 38); synaptic plasticity (148); release of neurotrophic factors; uptake and recycling of neurotransmitters; maintenance of the blood-brain barrier (BBB) (1); CBF regulation (11); and clearance and maintenance of ionic and extracellular neurotransmitters (94), glucose metabolism, and substrate delivery to neurons (125, 128). These functions are, in many instances, accomplished via the action of numerous specialized proteins, including glutamate transporters [e.g., glutamate transporter 1 and glutamate aspartate transporter (166)], K⁺ channels for the maintenance of extracellular K⁺ concentrations (e.g., via K_ir4.1), and aquaporin 4 (AQP4) for water homeostasis and glymphatic function (131), among others.

Astrocytes are nonexcitable cells. Their functional state (although not exclusive) has been studied by monitoring of intracellular Ca²⁺ dynamics as well as their effect on neuronal function. The use of genetically encoded Ca²⁺ indicators [myosin light chain kinase (GCaMP3/6) (177, 197)] has unveiled compartmentalized Ca²⁺ events in soma, primary processes, and fine processes defined as microdomains (for suggested nomenclature see Ref. 177), but the Ca²⁺ source and signaling mechanisms underlying activity in subcompartments are poorly understood. Agarwal et al. showed that Ca²⁺ events within astrocyte microdomains have distinct kinetics and respond to neuromodulators (i.e., norepinephrine) differently from those in soma or larger processes (3). In their study, microdomain Ca²⁺ events originated from the opening of the mitochondria permeability transition (MPT) pore. Importantly, mitochondrial Ca²⁺ was associated with signaling pathways involved in astrocyte dysfunction and increased reactive oxygen species (ROS) (3). Agarwal et al. proposed aberrant astrocyte microdomain Ca²⁺ activity as a potential link to neurodegenerative conditions (3, 111), raising the question of whether these subcompartments constitute an important site for cerebral bioenergetic and redox dysregulation. Astrocyte metabolic activity is critical for neuronal survival; thus, mitochondrial function is of high importance during pathology such as ischemic stroke. Also, astrocyte mitochondrial function is sexually dimorphic, both in the healthy brain and after stroke (65). In the healthy rodent brain, pyruvate dehydrogenase complex activity is higher in females than males (66). In humans, positron emission tomography shows higher glucose metabolism in females than males (217). Ovariectomy and the decline of gonadal hormones diminish mitochondrial function (reviewed in Ref. 66).

ASTROCYTE RESPONSES TO ISCHEMIC INJURY AND CONTRIBUTIONS TO RECOVERY

After injury, astrocyte function is dependent on numerous factors, including the type of injury, age, sex, and co-morbid disorders. In stroke, astrocytes have both beneficial and deleterious roles (147). While the brain has little capacity for energy storage, astrocytes store glycogen, giving them an advantage over neurons to become more resistant to ischemic injury (73) and essential players in poststroke recovery. On the other hand, severe ischemia results in mitochondrial dysfunction, energy depletion, and, consequently, dysregulation of astrocyte-mediated brain homeostatic processes, resulting in ionic disequilibrium, membrane depolarization, increased glutamate and Ca²⁺ release, excitotoxicity (164), oxidative stress, breakdown of the BBB, water dysregulation, and inflammation (49). An interesting study by Hayakawa et al. showed that astrocytes may be able to release, via a cluster of differentiation 38-mediated mechanism, functional extracellular mitochondria particles, which in turn enter neurons and aid in their viability and recovery after stroke (79), increasing the relevance for astrocyte contributions to health and disease. Aberrant astrocyte Ca²⁺ regulation also follows ischemic injury, contributing to injury exacerbation. Sources of intracellular Ca²⁺ overload after stroke include N-methyl d-aspartate receptors, voltage-dependent Ca²⁺ channels, metabotropic receptors, and increased combined activity of the Na⁺/H⁺ and Na⁺/Ca²⁺ exchangers (reviewed in Ref. 109). Excitotoxic Ca²⁺ concentrations result in opening of the MPT pore, mitochondrial damage, and release of proapoptotic factors that exacerbate ischemic brain injury (29). Burstein et al. demonstrated mitochondria sex differences in the brain (29). Female mitochondria exhibited reduced Ca²⁺ capacity and increased propensity for Ca²⁺-dependent opening of the MPT pore compared with male mitochondria. Both inhibition and genetic deletion of cyclophilin D, a regulator of MPT, eliminated sex differences. In addition, estrogen receptor (ER)-β blockade increased mitochondria Ca²⁺ capacity, suggesting a functional link between ERβ and cyclophilin D (29). These studies support ERβ signaling as a potential therapeutic target against irreversible mitochondria damage. Using in vivo multiphoton imaging microscopy in a mouse model of ischemia, Rakers et al. demonstrated that peri-infarct depolarizations exacerbate tissue damage via an increase in Ca²⁺ in both neurons and astrocytes (157). Responses were ameliorated in an inositol trisphosphate type 2 receptor-deficient mouse model, suggesting that astrocyte Ca²⁺ is a primary contributor to excitotoxicity. In a later study, the same group showed a transient receptor potential vanilloid 4 (TRPV4) channel contribution to astrocytic and neuronal Ca²⁺ increases along with extracellular glutamate accumulation (156, 157).

TRPV4 channels are Ca²⁺-permeable nonselective cation channels (150) expressed in multiple cell types and, thus, are involved in a variety of functions, including mechanosensation, systemic tonicity, pain, inflammation, and vascular tone (57, 85, 98, 150, 168, 191, 204). Relevant to the NVU, TRPV4 channel expression has been shown in endothelium, vascular smooth muscle cells, astrocytes, neurons, and microglia cells (20, 88). However, the relative contribution of these channels to neurovascular-related physiological vs. pathological processes, such as ischemia, is poorly understood. In the vasculature, specifically parenchymal arterioles, TRPV4 channels detect and transduce arteriole myogenic responses (e.g., mechanical stress) (48, 102). In cerebral capillary endothelial cells, TRPV4 channels are regulated by the plasma membrane phosphatidylinositol 4,5-biphosphate (PIP₂). Interestingly, the presence of PIP₂ tonically inhibits TRPV4 channels, resulting in reduced baseline activity. Harraz et al. showed that G_q protein-coupled receptor activation, a step leading to PIP₂ depletion, activated TRPV4 channels (77). PIP₂-mediated channel regulation is important, because Harraz et al. showed that its presence has opposite effects on the activity of inwardly rectifying K⁺ (K_ir2.1) channels (77, 78, 124). While a physiological role for endothelial K_ir channels was proposed to be the spread of a hyperpolarizing current (from capillary to arteriole), resulting in upstream dilation and increases in CBF (124), the impact of PIP₂ depletion and, consequently, TRPV4 channel activation on neurovascular-related process remains unclear. To this end, Harraz et al. proposed that TRPV4 channel activation might act as a stop signal to K⁺ channel-mediated endothelial hyperpolarization (77, 78, 124). These studies highlight the importance of TRPV4-K_ir channel regulation in shaping the neurovascular coupling response and its implication in disease conditions.

TRPV4 channels are also expressed in the plasma membrane of astrocytes (19, 20, 119). We demonstrated that increases in intravascular pressure augmented astrocyte Ca²⁺, partly via a TRPV4-mediated activation (48). Interestingly, not all astrocytes express TRPV4 channels; however, activation of TRPV4-positive cells can trigger intracellular Ca²⁺ increases in TRPV4-negative astrocytes, a mechanism likely involving ATP/purinergic signaling (176) and amplified by ischemic injury. Under pathological conditions, both increases and decreases in TRPV4 expression/function have been reported. Increased TRPV4 expression and activity were shown in rat hippocampal CA1 region astrocytes following ischemic injury and astrogliosis (31, 178). These studies support the notion that while TRPV4 channels are involved in physiological processes, channel dysregulation via proinflammatory mechanisms may contribute to neurovascular unit dysfunction. To this end, Križaj et al. proposed that Ca²⁺ overload in retinal ganglion cells results from the enhanced activity of TRPV4, P2X purinoceptor 7, and pannexin channels (110). TRPV4-mediated Ca²⁺ entry can trigger ATP release (102, 176) and, ultimately, release of proinflammatory signals (e.g., TNF-α and IL-1β) (110). Of more recent interest is the potential of TRPV4 channel antagonists to lessen brain injury by limiting cation influx (to include Ca²⁺), cell swelling, and deleterious downstream signaling, such as matrix metalloproteinases (MMPs) (88). Future studies addressing the cross talk between vascular and glial cells in the context of hypertension, ischemic injury, and neurovascular unit homeostasis are needed.

Astrocytes near the ischemic infarct undergo functional and structural transformations, referred as reactive gliosis (26, 146, 147). Astrocytes increase expression of the intermediate filament protein glial fibrillary acidic protein (GFAP), an event that is dependent on their relative distance from the insult, used as a biomarker of ischemic damage (159). Astrocytes near the lesion proliferate and form the glial scar, composed of intertwined processes and extracellular matrix (16, 185, 208, 216), requiring ≥2 wk (measured in rodent models) to fully form around the infarct necrotic core. The glial scar demarcates the zone between dead and surviving parenchyma, and while it is thought to play a role in defending against infiltrating peripheral cells (16), recent evidence suggests that the glial scar is not a strong barrier against extracellular fluid entry into the brain (220). Moreover, blood vessels within the infarct core are leaky; therefore, the extracellular fluid leaking into the surrounding tissue likely contains plasma, a highly neurotoxic element (220). Although B lymphocytes and plasma cells remain largely sequestered from healthy tissue (54), products of these cells (e.g., antibodies), as well as MMP, cytokines, and chemokines, leak across the glial scar into the surrounding parenchyma (220). While the barrier properties (to infiltrating extracellular fluid) of the glial scar are unclear, the necrotic and neurotoxic ischemic core will eventually resolve: a stage termed cystic encephalomalacia (139, 220). Farther from the lesion, astrocytes remain in their territory but also undergo morphological changes, including increased gene expression of the intermediate filaments GFAP, vimentin, and nestin (185), changes that are likely to negatively influence NVU cell-cell communication. These findings are significant, considering drug delivery during the stroke recovery period. First, considering the leaky BBB in the weeks and months after stroke, therapeutics that may hasten a return to homeostasis have potential to reach the peri-infarct region. On the other hand, drugs delivered systemically as part of any therapeutic regimen may also leak past the glial scar for unintended effects. However, pharmacokinetic studies to determine the penetration of drugs into the parenchyma are necessary to fully appreciate the benefits or harmful outcomes of a leaky glial scar. In addition to a leaky glial scar, ischemic stroke alters the functional integrity of the BBB, with additional important implications for drug delivery. For example, BBB permeability increases during hypoxia via altered tight junctions, an action mediated by protein kinase C (PKC) isoforms along with zonula occludens-1, occludin, and claudin-5 (123). The redistribution of these proteins allows for increased movement of molecules across the BBB in a size-dependent manner. Importantly, BBB breakdown can result in cerebral edema, worsening stroke outcome. Both astrocytes and cerebral endothelial cells undergo functional changes, which facilitate ionic disturbances and water movement into the brain (207). During ischemia, edema formation is exacerbated by the activation of ion channels and transporters (e.g., K_ir4.1, AQP4, Na⁺-K⁺-Cl⁻ cotransporter, and Na⁺/H⁺ exchanger), impairing osmotic homeostasis (87, 167, 207, 218). Ischemia-induced increases in astrocyte Ca²⁺ concentrations raise cAMP, which in turn activates AQP4 channels, increasing water permeability and swelling (207). These events are further worsened by ischemia-induced loss of AQP4 polarization (45a, 144). The effects of sex differences on BBB permeability are poorly understood, but evidence supports the idea that estradiol treatment reduces astrocyte swelling and ischemia-induced overexpression of AQP4 (167). Understanding key components of the BBB in health and disease will allow for development of specific strategies that modulate the BBB during the acute phases of stroke, attenuate edema (a deleterious outcome of uncontrolled increased BBB permeability), and improve drug delivery during central nervous system (CNS) injury.

The ability of astrocytes to participate in repair processes is lessened by aging, which increases the state of inflammation and results in more pronounced stroke (36). Proliferation of reactive astrocytes is higher in the aged brain (151), and the neuroprotective function of astrocytes is decreased (36). With aging, the trophic factor IGF-1 declines, an event that is also linked to sex differences (174). Okoreeh et al. determined the importance of astrocyte IGF-1 in middle-aged (10- to 12-mo-old) acyclic female rats (142). They showed that injection of recombinant adeno-associated virus serotype 5 containing the GFAP promoter into the striatum and cortex improved ischemia-induced deleterious outcomes. IGF-1 gene transfer to astrocytes also reduced serum GFAP levels and improved blood pressure, neurological score, and sensory-motor performance (142).

Astrocytes are a target for gonadal hormones (10). Both ERα and ERβ are expressed in astrocytes (5, 67). Studies in primary hypothalamic astrocytes from female rats confirmed the presence of transmembrane G protein-coupled ER (GPER), which is involved in nongenomic signaling (113) and contributes to the neuroprotective effects of 17β-estradiol (192). Estrogen’s neuroprotective actions are well established and extend to the regulation of spine density (72), synaptic number, and synthesis of neurotrophic factors (2, 99, 186). Thus, ischemia-induced increased ERα expression in astrocytes is generally considered neuroprotective (47).

Astrocytes are active participants in inflammatory processes. Astrocytes secrete proinflammatory (IL-6 and IL-1) and anti-inflammatory (IL-10) signals, chemokines [C-C motif ligand 2 (CCL2); C-X-C motif chemokine ligand (CXCL) 1 (CXCL1); CXCL10, also known as interferon-γ-induced protein 10 (IP-10); and CXCL12], and other small-molecule messengers [S100 Ca²⁺-binding protein B (S100B) and nitric oxide (NO)] (51), which signal to immune cells in physiological and pathological conditions. Within the peri-infarct region, astrocytes are critically important for recovery processes after stroke; however, evidence also suggests that astrocytes release proinflammatory signals that exacerbate tissue damage. Using an in vitro and an in vivo approach, including proteomic profile array analysis (24 h after stroke), Li et al. showed that astrocyte-derived IL-15, which in the periphery plays a role in the maintenance and function of immune cells, including natural killer cells and T lymphocyte (108, 206), induced activation of natural killer cells and CD8⁺ T lymphocytes (115). IL-15 overexpression in astrocytes increased infiltration of lymphocytes into the brain, exacerbating tissue damage and neurological dysfunction. On the other hand, IL-15 knockdown (using a lentiviral approach) reduced damage and improved neurological outcomes (115). Paralleling the macrophage literature, astrocyte phenotypes, induced by ischemic injury or lipopolysaccharide, are broadly categorized as A1 (neurotoxic) or A2 (neuroprotective) reactive astrocytes (117, 219). Subsequent studies show that A1 astrocytes are highly toxic to a broad array of cell types, including retinal ganglion cells, cortical neurons, embryonic spinal motor neurons, and differentiated oligodendrocytes (117). However, other cell types (preganglionic and γ-mother neurons) were not vulnerable to A1 astrocytes. In addition, A1 astrocytes have a reduced capacity to form robust excitatory synapses due to decreased glypican secretion. In contrast to A1 astrocytes, A2 astrocytes express high levels of neurotrophic factors and cytokines that are useful in the poststroke repair process (219). Using two-photon imaging of male mice subjected to middle cerebral artery occlusion (MCAO model), Fordsmann et al. demonstrated reduced spontaneous Ca²⁺ activity in the ischemic penumbra in neurons and astrocytes from young (3- to 4-mo-old) mice vs. electrically silent spontaneous Ca²⁺ activity in neurons and astrocytes from older (18- to 24-mo-old) mice (63). Spontaneous Ca²⁺ activity was driven by presynaptic and purinergic signaling mechanisms. Importantly, the ischemic penumbra of aged brains was more vulnerable than that of younger animals, in part because of the harmful actions of aged astrocytes under ischemic conditions (63). Delineating between the A1 and A2 phenotypes in a spatiotemporal manner to the ischemic injury and how gonadal hormones may influence these phenotypes is a necessary next step to fully appreciate the complexity of astrocyte-mediated injury and repair processes after stroke.

MICROGLIA: AN ASSET TO THE NVU

Microglia are the immune cells and resident phagocytes in the brain; their primary essential functions are continuous parenchymal surveillance, maintenance of neural network homeostasis, and response to disease or injury. Microglia are highly ramified (branched) cells with dynamically moving processes, which in turn facilitate their essential functions and cell-cell interactions (194, 212). This ramified morphology is associated with dynamic activities involving process extension and retraction necessary to continuously monitor their specific parenchymal domains. In fact, microglia process extension and retraction rates in the healthy brain are estimated to be 1.47 μm/min (140), a rate necessary for complete brain assessment every few hours. Therefore, the historical functional categorization of ramified microglia as “resting” marginalizes their essential contributions to brain maintenance and homeostasis. Microglia are phagocytes and have an important role in maintaining brain health (i.e., synaptic pruning) (184) and restoration of neuronal function during injury and disease (61, 199). Microglia have a significant role in neurological (43, 184) and behavioral (114) development; they also maintain neuronal networks in the adult brain (95, 205) and work collaboratively with other glia (50). While microglial surveillance functions are vital to maintenance of the complex neural network of glia, neurons, and vasculature, they are rarely considered a key member of the NVU.

Microglia are intricately linked to their cellular environment, and, rather than being innately driven, much of their activity is heavily influenced by neuronal and glial signaling mediated by the release of small-molecule messengers and soluble factors (212), as summarized in Fig. 1. Hence, the influence of the neighboring environment is an important rationale for studying microglia in vivo or ex vivo, rather than in vitro (71). Microglia are highly responsive to neuronal activity via purines (45, 106, 140), neurotransmitters (62), and chemokines (170), which vary according to the functions of brain regions, sex, and age. Estrogens rapidly modulate neuronal activity. Thus, gonadal hormones may have a significant direct or indirect influence on microglial activity. In the cortex, estrogen potentiates excitatory synapses via increased sensitivity to glutamate through multiple mechanisms, which include altered cellular responsiveness to ionotropic glutamate receptor activation and opioid, dopamine, and GABA receptors (reviewed in Refs. 100, 126, 173), as well as more long-term effects by regulating the brain transcriptome (56).

Fig. 1.

Brain cell-microglia interactions: a nonexhaustive listing of oligodendrocytes (50, 145, 158) and neuronal (198, 212) and astrocytic (116) communication modalities with microglia mediated by small-molecule messengers, soluble factors, cytokines, chemokines (from cells), and associated microglial receptors (color-coded to match ligand and receptor). Such ligand-receptor interactions promote microglial responses, such as changes in morphology (e.g., polarization), increased or decreased process activity, process-directed phagocytosis, synaptic remodeling, and cytokine and chemokine production, which, in turn, elicit a response from responding brain cells. AdR, adenosine receptor; CD200L, cluster of differentiation (CD) 200 ligand; CD200R, CD200 receptor; CSF1, colony-stimulating factor type 1; CSF1R, CSF1 receptor; CX₃CL1, C-X₃-C motif ligand 1; GABAR, GABA receptor; GLUT, glucose transporter; HSP, heat shock protein; HSPR, HSP receptor; NE, norepinephrine; NMDAR, N-methyl-d-aspartate receptor; P2Y6/7R, pyrimidinergic receptor 6/7; P2Y12R, pyrimidinergic receptor 12; RAGE, receptor for advanced glycation end products; S100B, S100 Ca²⁺-binding protein B; TGFβ, transforming growth factor-β; TGFβR, TGFβ receptor. [Oligodendrocyte artwork is from Domingues et al. (50), with permission from Frontiers;astrocyte artwork is from Wilhelmsson et al. (210), with permission; copyright (2006) National Academy of Sciences.]

Aging superimposes a progressive and intrinsic change in brain anatomy and physiology (135). As average brain weight decreases, average neuronal and glial numbers decline. In particular, microglial distribution and morphology change (213). Decreased cell numbers and ramification may impair surveillance functions or merely parallel a change in the parenchymal environment, including reduced neuronal activity. With age, microglia expression of immune response signaling receptors, such as major histocompatibility complex II and complement receptor (CR) 3, increases (149). However, it is unclear if these changes, observed in young mice, contribute to, or are a result of, disease pathology.

MICROGLIAL RESPONSES TO ISCHEMIC INJURY AND CONTRIBUTIONS TO RECOVERY

Microglia have an immediate and dynamic response to injury that is dependent on the modality and severity of the injury, such that the response to an acute and robust insult, such as ischemic stroke, is different from the response to chronic neurodegenerative diseases. Such plasticity allows for a more specific and appropriate response from the innate immune system to a variety of infections, diseases, and injury. Similar to the macrophage literature, microglial responses to injury are now grossly classified into two types: M1 and M2. The M1 response is elicited by molecules such as bacterial lipopolysaccharide or the specific T-helper cell 1 cytokines interleukin (IL) 2 and interferon-γ to elicit high-level production of deleterious proinflammatory chemokines (e.g., CXCL8 and CXCL10) and cytokines (e.g., IL-1β, IL-12, and IL-6) and ROS. Such actions aid in amplifying the immune response to injury and are necessary for mounting a defense evolving from a host response to bacteria. On the other hand, the M2 response is elicited by T-helper cell cytokines such as IL-4 and IL-10 and corresponds to tissue recovery, repair, and growth. Over a period that is much debated, microglia transition from the M1 to the M2 response, and IL-4 is reported to play a major role in this transition (223). While the macrophage literature also includes subclassifications of M2, such as M2b (associated with B cell responses) and M2c (associated with macrophage deactivation), these subclassifications have yet to be fully investigated in the context of microglia phenotypes. Studies of the period during which microglia transition between polarities (M1 vs. M2) after stroke and a consensus regarding corresponding functions associated with beneficial or harmful processes to improve stroke outcomes are needed. This critical knowledge will aid in development of targeted therapeutics that optimize neuroinflammatory responses after stroke.

Both age and sex affect microglia M1 vs. M2 polarization. In aged (18-mo-old) mice, microglia appeared less ramified than in young (2-mo-old) mice and lipofuscin deposits within the cells and surrounding tissue were increased (181). Moreover, mRNA expression of the proinflammatory cytokines IL-1β, IL-6, and TNF-α, as well as transforming growth factor-β1 (considered anti-inflammatory), but not IL-10 or IL-12/p40, was increased in microglia from aged mice relative to young mice. Estrogen also modulates microglia-mediated neuroinflammatory proteins via ERs. ER subtypes are classified as intranuclear and membrane-associated. Estrogen binding to intranuclear ERs (e.g., ERα and ERβ) results in protein modulation at the transcriptional level, while binding to membrane-associated ERs, such as GPER-1, will immediately modulate cellular functions and signaling to include an attenuation of neuroinflammatory responses in models of CNS diseases and injury (130, 153, 160, 202). Although all ER subtypes are expressed in the brain, the level of expression of each receptor differs according to brain region (25, 80, 112). Primary microglia, in vitro, express both ERα and ERβ (169), and application of anti-ERα, -ERβ, and -GPER-1 agonists elicits potent anti-inflammatory and neuroprotective responses, whereas antagonists block these responses, in a wide array of brain injury models, including ischemic stroke (121, 169, 222). Progesterone receptor expression on microglia and the role of progesterone-mediated neuroinflammatory responses have been much less investigated.

Microglia phagocytosis, a specific aspect of neuroinflammation, has a strong role in both M1 and M2 phenotypes. However, distinguishing between beneficial and deleterious phagocytosis in vivo is challenging from a methodological standpoint. Phagocytosis is a temporal and step-wise signaling cascade (179) carefully orchestrated to result in the efficient and purposeful internalization and degradation of specific targets (15, 75, 137), which is an asset to the NVU. 1) A signal for phagocytosis is released by affected cells in health or disease situations. The “find-me” (179) signals, such as UTP, ATP, and fractalkine (58, 106, 107, 179), induce a strong and immediate morphological and functional response from adjacent microglia (45, 140). Aberrant find-me signaling results in excessive microglial responses and increased, rather than decreased, injury (122). In a culture model of cell injury, increased signaling via UTP/P2RY6 ligand-receptor interactions resulted in exuberant phagocytosis and ingestion of viable neurons (138). 2) Microglia upregulation of “eat-me” signals (i.e., CR3) mediates interactions with affected cells via microglia surface receptors (101, 179). 3) Microglia engulf debris, evidenced by ball-and-chain morphology and phagosome formation. 4) The internalized debris is degraded, and immune signaling, via cytokine and chemokine release, occurs to amplify phagocytosis. Microglia produce ROS (196) and MMPs during this process as a mechanism to degrade engulfed debris. Any imbalance (deficit or augmentation) of this process, such as aberrant find-me signaling, reduced eat-me signals, or poor recruitment signaling, reduces phagocytosis efficiency. Inefficient, dysfunctional, or excessive phagocytosis can result in engulfment of healthy cells (28), failure to clear or digest debris, and continued ROS and MMP production, for an overall increased neurotoxic milieu (9, 122), and increased injury. Such a scenario eliminates the benefits of phagocytosis (28, 122, 179). Few have addressed whether poststroke microglia phagocytosis differs with sex or sex hormone treatments. In 2016, we showed that the constitutive expression of the phagocytic receptor CR3 differs between males and females and was a first indication that phagocytosis signaling is different and, possibly, a more efficient process in females than males (133). While there is ample literature to describe microglia-related neuroinflammatory responses after stroke (10, 13, 55), considerably more research is necessary to investigate the specifics of microglia phagocytosis in health and disease (42, 44, 225) and the related signaling mechanisms (75, 138) that may differ according to sex and, in females, menopausal status.

ASTROCYTE AND MICROGLIA CROSS TALK

Ischemic injury brings about a profound inflammatory response, and we have described how each of these cells contributes independently to the poststroke inflammatory milieu. Given their robust responses to ischemic injury and close proximities to each other throughout the brain, modalities of astrocyte and microglia cross talk are an important consideration in stroke research and treatment development. For example, therapeutics to modulate microglial responses may affect astrocyte functions or responses. In the face of injury, glia communicate with each other via release of cytokines and small-molecule messengers to promote or propagate microglial or astrocytic actions. As demonstrated in an elegant series of experiments, S100β, a small-molecule messenger released by astrocytes in large quantities after stroke, binds to the receptor for advanced glycated end products and increases concentrations of IL-1β and TNF-α via both NF-κB and activator protein 1 modalities, increasing cyclooxygenase (COX)-2 expression by microglia (23, 24). In neurodegenerative diseases, astrocytes have a role in regulating phagocytosis by releasing factors such as complement 3 (116) and plasminogen activator type 1 (93), as well as monocyte chemoattractant protein-1 and IP-10 (193). In the other direction, from microglia to astrocytes, the combined release of TNF, IL-1α, and complement protein C1q by microglia strongly induces the astrocyte A1 phenotype described above (see astrocyte responses to ischemic injury and contributions to recovery) (117). However, these communication modalities need to be confirmed in the context of ischemic stroke.

HYPERTENSION: NEUROVASCULAR UNIT DYSFUNCTION AND RISK FACTOR FOR ISCHEMIC STROKE

Hypertension is the leading modifiable risk factor for ischemic stroke. The American Heart Association defines hypertension as a resting blood pressure that is consistently >120/80 mmHg. By most current estimates, hypertension occurs in ~45% of the population, or ~60% when considering the additional 16% that are undiagnosed and unaware of their hypertensive status (21). The brain’s energy metabolism and CBF are altered by chronic hypertension (64), and this is particularly the case if cerebrovascular perfusion is compromised. Unresolved hypertension results in extensive remodeling of peripheral vessels, along with increased cerebrovascular resistance, resulting in cerebral hypoperfusion (8, 89) and impaired neurovascular coupling (69). The interactive mechanisms driving hypertension-evoked structural and functional changes in the cerebral circulation are complex, poorly understood, and specific to vessel size, while also dependent on the stage and severity of hypertension. The temporal aspect of these injury mechanisms is also poorly understood, and it is possible that injury cascades are set in motion long before functional and structural changes can be observed. Therefore, hypertension can have profound effects on the functional and structural state of the NVU that are “silent” until pathology has progressed extensively. Considering the prevalence of hypertension, when an ischemic stroke happens, it likely does not occur in a “healthy” brain but, rather, in a brain that has been subjected to years of chronic hypertension and the pathobiology summarized here. Therefore, hypertension not only predisposes one to ischemic stroke, but when stroke does occur in a hypertensive individual, this event should be considered a second hit.

Evidence suggests that the combined effects of increases in shear stress, endothelial dysfunction (59, 82), and circulating factors, such as mineralocorticoids (52), angiotensin II (ANG II) (18, 52), and inflammatory signals, play a role in the vascular dysfunction and remodeling observed in hypertension. Chronic increases in intraluminal pressure give rise to a progressive increase in vessel wall thickening (wall hypertrophy) as a way to buffer pressure fluctuations and normalize vascular wall tension (136, 154). While, during physiological laminar flow conditions, vasculature exposure to mechanical forces (e.g., flow, shear stress, and pressure) is stable and protected by signals such as NO, increases in wall stiffening can result in turbulent flow, damage to the microcirculation, and increases in the incidence of stroke and cognitive impairment (12). To this end, hypertension has been shown to increase the production of flow-induced vasoconstricting signals such as 20-hydroxyeicosatetraenoic acid, reduce NO availability, and increase inflammatory signaling via NF-κB (6, 12, 195). Thus, changes in the vessel architecture may be considered a protective mechanism against hyperperfusion, with remodeling occurring in larger vessels first to protect downstream arterioles. However, extensive remodeling of the large intracranial vessels and smaller parenchymal arterioles will eventually limit CBF, for which autoregulation and neurovascular coupling are unable to compensate. Such delineations are important, considering the diversity of NVU structure and function according to vessel size (reviewed in Ref. 92). In addition to biomechanical changes to the NVU, ANG II, a key player in human hypertension, has been shown to cause deleterious changes in multiple cell types of the NVU. In hypothalamic 4B cells, ANG II increased TRPV4 channel trafficking to the membrane and intracellular Ca²⁺ (172). In parenchymal arteriole myocytes, circulating ANG II modulated TRPV4 channel activity via downstream activation of PKCα bound to A-kinase anchoring protein 150 (191). Activity of the TRPV4 channel was dependent on its distance from A-kinase anchoring protein 150-associated PKCα, an interaction that showed sex differences and, thus, may be hormonally regulated (191). In astrocytes, we showed that chronic ANG II infusion increased TRPV4 channel expression and enhanced TRPV4-mediated currents and astrocyte Ca²⁺ events (48). Moreover, we also demonstrated that chronic ANG II infusion increased astrogliosis and hippocampus microglia activation (48). These studies support the notion that, in hypertension, cells of the NVU may be subjected to alterations in intracellular Ca²⁺, likely increasing their vulnerability to ischemia and reducing their recovery capacity after stroke.

Ultimately, the structural and functional changes induced by chronic hypertension predispose an individual to ischemic stroke. Moreover, the hypertension-evoked remodeling of cerebral vessels leads to hypoperfusion and can exacerbate ischemia-driven pathways that disrupt BBB integrity and allow infiltration of inflammatory cells and cytokines into the brain (60, 163, 175). ANG II, a pressor peptide that is dysregulated and overabundant in several forms of hypertension, drives changes in cerebral vascular structure and function (18), including, among others, increased mitochondrial contributions to oxidative stress (53, 103). Evidence suggests that Toll-like receptor (TLR)-4 is a link between ANG II and vascular inflammation. In the aorta, TLR4 expression and cytokine and chemokine production were abolished with anti-TLR4 treatment in ANG II-treated animals (84). The possibility that TLR4 may be a direct link between ANG II and vascular and neural inflammation has been recently reviewed (22).

Hypertension prevalence, pathophysiology, and vascular outcomes are not the same between males and females. Similar to ischemic stroke, premenopausal women stave off the development of hypertension until menopause, at which point hypertension prevalence in women surpasses that in men (21, 118). Multiple mechanisms likely account for these observed sex differences. Much research supports the notion that estrogen plays a large role in modulating the effects of ANG II to increase blood pressure (70, 214) and regulate cerebral circulation (70), as well as to mitigate ANG II-mediated cerebral vascular dysfunction and vascular oxidative stress (32). ANG II stimulates intracellular Ca²⁺ stores to result in myocyte contractility, contributing to vascular tone. Ca²⁺ dysregulation occurs in conjunction with ANG II dysregulation, contributing to vascular dysfunction and remodeling (40). Sex differences in cytosolic Ca²⁺ concentrations, Ca²⁺ signaling, and Ca²⁺ mobilization from intracellular stores in rodent hypertension models (68) suggest an additional layer of protection in vascular responses to hypertension in females that is not present in males.

The balance of signals that promote vessel relaxation (e.g., NO) vs. constriction (e.g., COX-derived factors) also favors the female. Estrogen preserves NO production, limits shear stress, and maintains vasodilatory effects in female hypertensive rats; it also has restorative effects in male hypertensive rats (90, 91). By limiting circulating gonadal hormones via orchiectomy, mean arterial blood pressure and impaired NO-mediated endothelial function were reduced in Sprague-Dawley rats fed a high-salt diet, whereas testosterone replacement restored these dysfunctions (143). Vasoconstriction is increased, via the release of COX-derived elements, in hypertensive male vs. female rats (41, 97). In combination, these data suggest that the balance between vasodilation and constriction is favored in females over males. Such positive effects on the vasculature in the female may have a pivotal role in delaying the onset of endothelial dysfunction and vascular remodeling observed in female vs. male models of hypertension.

CONCLUSION

While the incidence of stroke has declined, the prevalence of cognitive impairments is on the rise, increasing the demand for knowledge of the events that lead to dysfunction of the NVU. While much knowledge has been gained about the cellular mechanism underlying NVU function before and after stroke, much more remains to be discovered. This is particularly the case in terms of an understanding of how modifiable and nonmodifiable stroke risk factors influence the risk for stroke, as well as the repair processes following an ischemic insult. The heterogeneous nature of ischemic stroke in humans signals that personalized treatments are necessary for improved outcomes. Without the underlying knowledge to guide personalization, such therapies will remain elusive. Expanding investigations beyond the male healthy rodent stroke model to understand NVU function and dysfunction will broaden our understanding of stroke pathophysiology and mechanisms of recovery that may be exploited for therapy development. Although challenging, such efforts will aid in addressing the gap in translating stroke bench science to bedside treatments.

GRANTS

This work was supported by National Institutes of Health Grants R01 HL-089067 and 1R01 NS-082521-01 (to J. A. Filosa).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

H.W.M. prepared figure; H.W.M. and J.A.F. drafted manuscript; H.W.M. and J.A.F. edited and revised manuscript; H.W.M. and J.A.F. approved final version of manuscript.