Neuroinflammation and microvascular dysfunction after experimental subarachnoid hemorrhage: Emerging components of early brain injury related to outcome

Abstract

Aneurysmal subarachnoid hemorrhage has a high mortality rate and, for those who survive this devastating injury, can lead to life-long impairment. Clinical trials have demonstrated that cerebral vasospasm of larger extraparenchymal vessels is not the sole contributor to neurologic outcome. Recently, the focus of intense investigation has turned to mechanisms of early brain injury that may play a larger role in outcome, including neuroinflammation and microvascular dysfunction. Extravasated blood after aneurysm rupture results in a robust inflammatory response characterized by activation of microglia, upregulation of cellular adhesion molecules, recruitment of peripheral immune cells, as well as impaired neurovascular coupling, disruption of the blood-brain barrier, and imbalances in endogenous vasodilators and vasoconstrictors. Each of these phenomena are either directly or indirectly associated with neuronal death and brain injury. Here, we review recent studies investigating these various mechanisms in experimental models of subarachnoid hemorrhage with special emphasis on neuroinflammation and its effect on microvascular dysfunction. We discuss the various therapeutic targets that have risen from these mechanistic studies and suggest the utility of a multi-targeted approach to preventing delayed injury and improving outcome after SAH.

Introduction

Non-traumatic subarachnoid hemorrhage (SAH) is a devastating neurological emergency resulting most commonly from the rupture of cerebral aneurysms. It occurs in 7.2–10.5 out of 100,000 people and accounts for approximately 5% of strokes annually.^{1, 2} While the incidence of aneurysmal SAH is lower compared to that of ischemic stroke, it occurs in younger patient populations, has a higher mortality, and confers a significant impairment on quality of life. Long-term disability and mortality from SAH are estimated to account for up to 27% of all stroke-related years of potential life lost before age 65.³ For those who survive, life expectancy is greatly reduced with reported excess mortality rates of approximately 17% at 20 years compared to the general population.⁴ Mortality after SAH can occur at various time points after aneurysmal rupture. Approximately 10–15% of patients die before receiving medical attention.^{1, 5} This is likely due to a sharp rise in intracranial pressure that reduces cerebral perfusion, inducing global cerebral ischemia and resulting in metabolic crisis. For those who make it to the hospital, another 25% may succumb within the first 24–72 hours.⁵ This is a critical window during which multiple pathogenic processes are occurring that are collectively referred to as early brain injury (EBI). Clinically, severity of EBI is determined by various admission factors which include hemorrhage volume, level of consciousness, and presence of neurological deficits. Various mechanisms have been investigated to target EBI after SAH, including a robust inflammatory response, cerebral edema, and microvascular dysfunction throughout the brain.⁶

Of those who survive the initial 24–72 hours, there is additional risk of delayed cerebral vasospasm and ischemia which occur in 70% and 30% of patients, respectively.^7–9 Vasospasm typically affects the medium- and large-sized intracranial arteries and occurs within days 3 and 14 after SAH. This luminal narrowing has been associated with delayed cerebral ischemia (DCI), cerebral infarction, and long-term neurocognitive impairment. Clinical deterioration caused by DCI is a distinct entity that presents with focal neurological deficits or a decrease in Glasgow Coma Scale of at least two points for at least one hour.¹⁰ Hypoperfusion from DCI may progress to infarction in some cases, detected via CT, MRI, or on autopsy.^{1, 10} For decades, cerebral vasospasm was the subject of intense investigation as it was attributed to be the principal contributor to poor outcome. However, recent clinical trials using the endothelin-1 (ET-1) receptor antagonist clazosentan demonstrated a reduction in the incidence of angiographic vasospasm, but no significant change in functional outcome or mortality.^11–13 The results of these trials served as a major turning point in the field and highlighted the need for further investigation into other pathogenic mechanisms of brain injury after SAH.

Several mechanisms during the acute phase of SAH contribute to DCI and poor outcome. These include neuroinflammation, microthrombosis, cortical spreading depolarizations, disrupted integrity of the blood-brain barrier (BBB), and microvascular dysfunction in addition to well-studied macrovascular cerebral vasospasm.⁶ There are additional systemic responses after SAH including stress hyperglycemia, fever, infection, and dysregulation of coagulation pathways which may also affect clinical outcome, although systemic complications are not the focus of our present review. Recently, our lab has focused on two of these phenomena and the complex interplay between them – neuroinflammation and microvascular dysfunction. Results from our work and that of others suggest that these pathophysiological processes play a highly influential role during the EBI phase and set the stage for long-term complications and outcome (Figure 1). Further, these events may be interrelated as inflammatory responses following SAH may result in microvascular dysfunction, which in turn could drive further inflammation.

Figure 1. Neuroinflammation and microvascular dysfunction after aneurysmal subarachnoid hemorrhage.

In addition to the well-studied phenomenon of cerebral vasospasm, release of blood products into the subarachnoid space can trigger a robust inflammatory response consisting of activated microglia, secretion of pro-inflammatory cytokines, increased expression of CAMs, peripheral leukocyte recruitment, and BBB disruption. This can further contribute to microvascular dysfunction including arteriolar vasoconstriction, microthrombosis, and imbalance of endogenous vasoconstrictors and vasodilators, which further compromises CBF and drives delayed ischemic damage.

There are excellent reviews describing the association of inflammation with hemostasis, dysregulation of large conduit arterial tone, and DCI.^14–16 This review summarizes the efforts to date investigating the role of neuroinflammation and microvascular dysfunction during EBI as well as mechanistic and therapeutic targets that may ameliorate EBI and improve outcome after SAH (Table 1). By highlighting various mechanisms contributing to EBI and long-term outcome, we suggest a multifactorial approach to the management of brain injury following SAH.

Table 1.

Major strategies in experimental subarachnoid hemorrhage aimed at targeting neuroinflammation and/or microvascular dysfunction.

Major Experimental Targets	Genetic or Pharmacologic Approach	Results	Translated to SAH Patients	Reference(s)
Pattern-Recognition Receptors
Toll-like Receptor 4 (TLR4)	TLR4−/− mice	Decreased vasospasm and neuronal apoptosis on days 7 and 15 after SAH	Not applicable	(27)
	TLR4 antagonists (IAXO-102, TAK-242)	Higher neurological scores and reduced brain water content at 24 hours compared to controls, reduced BBB disruption with decreased MMP-9 and preserved tight junctions	No	(45)
Resident Cells of the CNS
Microglia	Clodronate liposomes	Depletion of microglia results in significant ablation of vasospasm at day 7 and 15 in mice, reduced neuronal death at day 7 but not at day 15 compared to vehicle-treated controls	No	(27)
	CS11b-HSVTK+/− mice	Depletion of microglia results in reduced neuronal cell death	Not applicable	(31)
Pro-inflammatory Cytokines
Interleukin-1 (IL-1)	IL-1 receptor antagonist (IL-1Ra)	In rodents, treatment resulted in decreased BBB breakdown and subsequent brain injury	Yes. The SCILSAH study has shown safety, tolerability, and effective reduction in peripheral inflammatory markers, supporting a Phase III clinical trial	(42–44)
Cellular Adhesion Molecules
CD11/CD18 (includes LFA-1 and Mac-1, also known as CD11a/CD18 and CD11b/CD18, respectively)	Anti-LFA-1 antibody	Reduction in femoral artery spasm following blood exposure in rats	No	(63)
	Anti-CD11/CD18 antibody	Reduction in non-human primate cerebral vasospasm from baseline angiography compared to vehicle-treated animals; Reduction of rabbit basilar artery spasm and increased peripheral white blood cell count	No	(64–66)
Intercellular adhesion molecule-1 (ICAM-1)	Anti-ICAM-1 antibody	Reduction in rabbit basilar artery spasm, synergistic with effects of anti- CD18 Ab; reduction in femoral artery spasm following blood exposure in rats	No	(63,65)
Vascular adhesion protein-1 (VAP-1)	LJP-1586	Inhibition of VAP-1 results in reduced leukocyte adhesion and infiltration, enhanced microvascular reactivity, and improved shortterm neurologic outcome	No	(67,69)
Peripheral Immune Cells
Neutrophils	Anti-neutrophil serum	Reduction in leukocyte infiltration into CNS, preservation of pial arteriolar dilating function, and protection of neurobehavioral l function; reduction in vascular collagenase activity	No, may prolong bleeding time from ruptured artery based on preclinical data	(69,70)
	Anti-Ly6G/C antibody	Reduction in middle cerebral artery (MCA) vasospasm and improved behavioral testing via Ymaze and Barnes maze tests; reduced cerebral inflammation and decreased impairment in long-term potentiation (LTP) at day 6 after SAH in mice	No	(68,71)
Lymphocytes	Corticosteroids (Dexamethasone, Methylprednisolone, etc.)	Reduced alterations in contractile and cytoskeletal proteins of rabbit cerebral arteries; decreased CSF citrulline (contributor to NO production) and leukocytosis; mixed results related to effect on vasospasm	Yes, with conflicting results. Overall, no strong evidence of beneficial or adverse effect	(76–78,82)
	Cyclosporine	Reduction in canine basilar artery vasospasm with prophylactic treatment; reduction in neuronal apoptosis and BBB disruption in mice with improved neurological outcome	Yes, with conflicting results. Some have shown improved neurological outcome while others have shown no effect on vasospasm or DCI	(79–81,83,84)
	Fingolimod (FTY-720)	Reduced intravascular leukocyte adhesion to pial venules, preserved pial arteriolar reactivity, improved neurological outcome	No	(85)
Endogenous Vasoconstrictors
Endothelin-1 (ET-1)	Clazosentan (ET-1 receptor antagonist)	Prophylactic treatment in rats prevented continued reduction in cerebral blood flow after acute hypoperfusion; reduced largeartery vasospasm but did not prevent formation of microthrombi, neuronal cell death, or loss of LTP	Yes, reduction in angiographic vasospasm but no statistically significant effect on morbidity, mortality, or functional outcome	(11,12,97,98)
20-hydroxyeicosatetranoic acid (20-HETE)	TS-011, 17-octadecynoic acid, HET0016 (selective CYP450 inhibitors)	Pre-treatment resulted in faster recovery of cerebral blood flow in the acute setting following SAH; reversal of delayed vasospasm in vitro and in vivo	No	(163–165)
Endogenous Vasodilators
Nitric Oxide (NO)	NO donors (Larginine, Snitrosoglutathione, sodium nitroprusside, transdermal nitroglycerin, etc.)	Improved CBF recovery, reduction in cerebral vasospasm, decreased glutamate excitotoxicity, and transient decrease in systemic blood pressure	Yes, with conflicting results. Side effects including systemic hypotension, headache, and rebound hypertension possible, limiting clinical use.	(135–137,140–143)
	Inhaled NO	Reduction in microvascular constriction with limited effects on large artery spasms, decreased cerebral edema, hippocampal neuronal loss, and mortality; improved neurological outcome	No	(133)
	Phosphodiesterase (PDE)-V inhibitors (sildenafil)	Reduction of vasospasm and neuronal cell death with improved neurological outcome in mice	Yes, Phase I study demonstrate d safety and tolerability, with some data suggesting potential improvement of vasospasm	(144,145)
Vascular smooth muscle cell (vSMC) relaxation	PDE-III inhibitors (milrinone)	Prevented angiographic vasospasm in canine model; improved CBF and neurobehavioral l outcome, reduced DCI in mice	Yes, reduction in delayed cerebral vasospasm warranting further study	(146–151)
	Magnesium sulfate	Reduction of infarct size, reversal of vasospasm in vivo and in vitro, and improved cerebral blood flow recovery	Yes, MASH-II and IMAGES trials failed to show clinical benefit	(152–158)

Neuroinflammation

Immediately after aneurysm rupture, blood rushes into the subarachnoid space under arterial pressure. This leads to a sharp rise in intracranial pressure and reduction in cerebral blood flow (CBF), which compromises tissue perfusion and causes diffuse brain injury and potential death. In addition, extravasated red blood cells (RBCs) in the subarachnoid space undergo degradation, releasing a host of bioactive and potentially toxic molecules including hemoglobin, methemoglobin, bilirubin, coagulation factors such as fibrinogen, and more (Figure 1).^16–22 Several of these molecules, including free hemoglobin and its subsequent byproducts, have long been associated with the development of cerebral vasospasm and outcome.^16–23 Experiments have suggested that mechanisms responsible for this may include production of free radicals and other metabolites with vasoconstrictive and pro-inflammatory properties. Bilirubin oxidation products (BOXes) formed from the breakdown of hemoglobin have peak concentrations correlating with the occurrence of vasospasm.^{21, 22} Several endogenous scavenging mechanisms in place, such as the CD163-haptoglobin-hemoglobin system, may act to may help clear toxic hemoglobin and its metabolites; however, studies suggest that these systems may quickly become saturated following SAH.²⁴ While direct interaction with the cerebral vasculature is one mechanism by which these molecules can produce long-term impairments, they may also interact with neurons, glia, and immune cells as blood products contact adjacent tissue or infiltrate brain parenchyma via paravascular spaces and disrupt normal flow between interstitial fluid and cerebrospinal fluid (CSF).^{25, 26} The interactions between blood products and cells throughout the central nervous system (CNS) results in a cascade of molecular events triggered initially during EBI that may persist and result in both acute and delayed brain injury.

Microglial Response

One of the first cells to respond to these extravasated RBC products is microglia, the resident immune cell of the CNS. Under normal physiological circumstances, microglia are the main immune cell actively surveying the CNS, which is otherwise somewhat restricted from peripheral immune cell trafficking. Blood products such as hemoglobin have been shown to bind to pattern recognition receptors (PRRs), such as toll-like receptor 4 (TLR4), on the surface of immune cells such as microglia (Figure 1).^{18, 27} Activation of TLR4 and other PRRs can lead to activation of downstream inflammatory signaling cascades including NF-kB, MyD88/TRIF, and MAPK pathways.^{16, 27–29} This results in activation of microglia, which take on a more amoeboid shape and release pro-inflammatory cytokines. In animal models of SAH, a robust increase in microglia and pro-inflammatory cytokine expression throughout the brain was associated with long-term sensorimotor deficits.³⁰ Depletion of microglia using CD11b HSVTK^+/− mice attenuated neuronal loss after experimental SAH.³¹ Further, increased microglial expression of heme oxygenase-1, responsible for the metabolism of free heme, has been shown to reduce neuronal cell death, vasospasm, and cognitive impairment in murine models.³² While most studies have investigated the deleterious effects of microglial activation and promotion of a pro-inflammatory environment, it is also well-known that microglia can be polarized to a more anti-inflammatory phenotype.³³ Potential therapeutic strategies early after SAH may thus seek to promote the activation of these microglia towards an anti-inflammatory phenotype which may confer neuroprotective benefits,³⁴ including but not limited to effects on neurogenesis and neurorepair which have been suggested in other forms of stroke.

Cytokines and Secreted Proteins

Several pro-inflammatory cytokines including interleukin-1 (IL-1), IL-6, tumor necrosis factor-a (TNF-a), and others have been demonstrated to be upregulated in CSF and serum after SAH in humans and animal models (Figure 1).^{35, 36} Pro-inflammatory cytokines can potentiate brain damage by triggering apoptotic pathways, interfering with the balance of endogenous vasodilators and vasoconstrictors, activating clotting factors leading to microthrombosis, and recruiting peripheral immune cells via upregulation of cellular adhesion molecules (CAMs). This initial release of cytokines and chemokines occurs from resident cells of the CNS such as microglia, but subsequent infiltration of peripheral immune cells further drives production of cytokines within the subarachnoid space and the brain parenchyma. IL-1, in particular, increases BBB permeability, enhances glial-mediated neurotoxicity, and promotes ischemic changes after SAH in preclinical models.^37–41 Based on these results in experimental SAH and other forms of stroke, the SCIL-SAH study targeted the pro-inflammatory cytokines IL-1 and downstream IL-6 using the naturally occurring IL-1 receptor antagonist anakinra in humans.^42–44 In addition to being safe and well-tolerated, results of the Phase II trial demonstrated reduction of IL-6, C reactive protein (CRP), and fibrinogen in the active arm, supporting a Phase III trial to investigate the effect on outcome.^{42, 43} Upstream targeting of TLR4 in experimental SAH through genetic knockouts and pharmacologic interventions has also been effective at reducing vasospasm and improving outcome highlighting the contribution of immune mechanisms to vascular tone regulation.^{16, 27, 45}

Additional secreted immune molecules further promote inflammation and worsen outcome after SAH. Previous studies have demonstrated an association between elevated CSF and plasma levels of complement proteins C3a and C4a, and outcome.^{46, 47} In addition, alterations in members of the mannose-binding lectin pathway of complement activation, such as ficolin-1, have been described.^{48, 49} Changes in complement proteins are an attractive target for further investigation given their proposed role in microglial-mediated synaptic alterations during development and aberrant reactivation in neurologic disease.^50–53 In contrast, anti-inflammatory mediators, such as IL-10 and various fatty acid-derived lipid mediators, may promote resolution of inflammation.^{54, 55} The results of these clinical and preclinical studies suggest a highly significant contribution of microglia, pro-inflammatory cytokines, and other secreted factors to poor outcome after SAH (Table 1).

Cellular Adhesion Molecules

The development of a pro-inflammatory state induced by secretion of cytokines and chemokines following SAH is also associated with increased expression of CAMs on the surface of endothelium, platelets, and leukocytes. CAMs such as E-selectin, vascular adhesion protein-1 (VAP-1), vascular cell adhesion molecule-1 (VCAM-1), intercellular adhesion molecule-1 (ICAM-1), macrophage-1 antigen (Mac-1), and lymphocyte function-associated antigen-1 (LFA-1) promote leukocyte adhesion of immune cells to luminal endothelia.^56–58 This robust inflammatory response leads to increased BBB permeability which facilitates the infiltration of peripheral leukocytes into the brain parenchyma (Figure 1). CAMs including E-selectin, VCAM-1, and ICAM-1 are elevated in the CSF of SAH patients, and these elevations correlate with the occurrence of vasospasm and DCI.^59–62 In animal models, similar elevations in CAM expression have been detected, and treatment with anti-CAM antibodies such as anti-ICAM1, anti-LFA-1, and anti-Mac-1 resulted in reduced leukocyte infiltration and arterial narrowing (Table 1).^{56, 58, 63–66} In our own lab, we have targeted endothelial VAP-1 using the VAP-1 inhibitor LJP-1586 in rats during the period of EBI, and demonstrated reduced leukocyte adhesion, enhanced microvascular reactivity, and improved short-term neurologic outcome.⁶⁷ Most strategies targeting CAMs are non-selective inhibitors of leukocyte adhesion, and thus the neurological benefits derived from CAM inhibition may be from neutrophils, monocytes, lymphocytes, or a combination. Future studies may seek to identify and target CAM pathways which selectively block infiltration of specific peripheral immune cell populations.

Peripheral Immune Cells

Recruitment of peripheral immune cells into the brain after SAH is a well-documented phenomenon and occurs early in the course of disease in response to increased expression of chemokines and CAMs. Infiltrating leukocytes attempt to phagocytose RBCs and debris induced by aneurysm rupture. Leukocyte adhesion to pial venules occurs within 48 hours after SAH, corresponding to the period of EBI.^67–69 The earliest peripheral immune cell to infiltrate the CNS after SAH is the neutrophil, believed to enter the CNS within 24–48 hours after injury.^{63, 64, 67, 69} Induction of neutropenia using anti-rat neutrophil serum reduced leukocyte adhesion to pial vessels and improved neurologic outcome suggesting that neutrophils play a predominant role in poor outcome following experimental SAH.⁶⁹ These observations were also described in other studies which either depleted neutrophils or limited their function.^68–70 Further, Provencio et al. (2016) showed that neutrophil depletion using an anti-Ly6G/C antibody after SAH results in improved spatial memory six days after SAH, and that this is mediated largely by attenuating dysfunction in long-term potentiation via NMDA receptors.⁷¹ In SAH patients, CSF neutrophil content has been shown to be an independent predictor of other delayed events including vasospasm.⁷² The extent of neutrophil infiltration into the subarachnoid space and CNS parenchyma helps determine the extent of the acute inflammatory response, which in turn may influence delayed events such as vasospasm and neurologic outcome.

Some studies have shown a significant role of other infiltrating leukocyte subtypes. Infiltrating monocytes enter the subarachnoid space and CNS parenchyma as active macrophages around 2–5 days post-SAH, and like neutrophils engage in phagocytosis of RBCs, clots, and debris.⁷³ Increased migration of monocytes across the cerebral microvasculature has been shown in vivo as well as in vitro using monocyte migration assays.⁷⁴ Further, due to the number of overlapping markers between monocyte-derived macrophages and resident microglia, few studies have successfully distinguished between the contribution of these two cell populations in SAH. Recent studies using RNA sequencing technology have successfully identified unique markers of resident microglia in the CNS (e.g. Tmem119).⁷⁵ Future studies in SAH would benefit from using more specific microglial and infiltrating monocyte markers to differentiate the unique role of these cells.

Lymphocytes as a prominent cell type of adaptive immunity may also play a role in post-SAH pathophysiology; however, studies are limited and inconsistent. Therapeutic strategies targeting T lymphocytes such as corticosteroids or cyclosporine have shown efficacy in some studies;^76–81 however, evidence supporting clinical use is lacking due to conflicting results or increased risk of adverse events.^{56, 82–84} Studies in our lab have employed the immunomodulatory agent fingolimod (FTY720) in a preclinical rat model of SAH.^{85, 86} Fingolimod is a well-tolerated, orally bioavailable, FDA-approved drug currently used for multiple sclerosis that acts as a sphingosine-1-phosphate (S1P) analog.⁸⁷ Its mechanism of action involves reversible phosphorylation and activation by sphingosine-kinase 2, which allows it to recognize and downregulate G-coupled S1P receptor (S1PR) type 1 expressed on peripheral lymphocytes. The immunomodulatory effect of fingolimod results from the sequestration of circulating mature lymphocytes in peripheral lymphoid tissues resulting in lymphopenia.⁸⁸ In addition, fingolimod crosses the BBB and binds to S1PRs expressed on CNS cells including neurons, oligodendrocytes, astrocytes, microglia, and brain endothelia, resulting in direct effects on these cells.^{86, 89, 90} Studies in ischemic stroke models, for example, have shown that fingolimod is neuroprotective, can enhance remyelination, restore BBB integrity, and reduce microglial activation and astrogliosis.⁸⁶ Treatment with fingolimod reduces intravascular leukocyte adhesion to pial venules, preserves pial arteriolar reactivity, and improves neurologic function at 48 hours after SAH in a rat model.⁸⁵ While the benefit of immunomodulators such as fingolimod has been demonstrated in experimental SAH EBI, other readily available biomarkers can be used clinically to evaluate immune responses following cerebral hemorrhage. One such marker is the serum neutrophil-to-lymphocyte ratio (NLR), which has not been thoroughly studied in EBI but has been associated with DCI.^91–93 Taken together, alterations in leukocyte trafficking after SAH seem to play an important role in driving outcome; however, the role of resident versus infiltrating immune cells in SAH-associated EBI remains a key area of investigation.

Microvascular Dysfunction

SAH-associated vascular dysfunction of large intracranial arteries has been the principal focus of preclinical and clinical studies over the last several decades. However, while large-vessel angiographic vasospasm is observed in up to 70% of patients, DCI is only observed in up to half of them.^{94, 95} Also, neurologic deterioration and radiologic evidence of cerebral ischemia can occur in the absence of vasospasm, and the reversal of vasospasm using ET-1 inhibitors does not demonstrably influence outcome despite some evidence from preclinical studies.^95–98 Moreover, the only medication that has been shown to be beneficial in SAH, the calcium-channel blocker nimodipine, does not seem to have a significant effect on vasospasm.⁹⁹ However, nimodipine does in fact appear to inhibit vasoconstriction at the level of small-diameter arterioles,¹⁰⁰ suggesting that targeting microvascular dysfunction could improve outcomes after SAH. The phase 3, multicenter, randomized NEWTON 2 trial was designed to compare the effect of an extended-release microparticle nimodipine preparation delivered directly to CSF to oral nimodipine.¹⁰¹ However, skepticism exists in relation to the efficacy of this approach as the Data Monitoring Committee has recommended discontinuation of the study due to low probability of meeting its primary endpoint for favorable outcome.¹⁰² This highlights the need for further mechanistic understanding of microvascular changes after SAH and identification of therapeutic targets.

Blood Vessel Reactivity

It has been estimated that at least 50% of the cerebrovascular resistance lies in arterioles and precapillary segments. Despite their central role in hemodynamic control, the contribution of microvessels to SAH outcome has received little attention. Increasing evidence suggests that microvascular dysfunction is associated with both EBI and DCI.^{103, 104} Various cell types within the cerebral microvasculature including endothelia, pericytes, and vascular smooth muscle cells (vSMCs) engage in constant communication with surrounding neurons and glia, collectively forming a functional neurovascular unit. The intense cross-talk between these various cells under normal conditions results in changes in microvascular tone and tissue perfusion in response to neuronal energetic needs.^{103, 105, 106} This process, called neurovascular coupling, is coordinated by neurons and astrocytes, which typically respond to increased extracellular glutamate and transmit signals to vSMCs in arterioles to promote vasodilation and enhanced blood flow in response to neuronal activity and increasing metabolic demands.¹⁰⁷ Following SAH, there appears to be inversion of neurovascular coupling starting 24 to 96 hours after injury, whereby neuronal activity instead promotes a vasoconstrictive response in arterioles.^{108, 109} This aberrant response to neuronal activity creates a mismatch between neuronal energetic needs and blood flow that can further potentiate brain injury.

Further evidence from our lab and others has suggested that significant microvascular dysfunction after SAH occurs at the level of arterioles. Several studies have attempted to investigate microvascular reactivity after experimental SAH via direct visualization of vessels in vivo.^{67, 69, 85} Under normal conditions, cortical activation (achieved via sciatic nerve stimulation) or topical treatment of pial vessels with vasoactive agents, including adenosine, acetylcholine, nitric oxide (NO) donors, or carbon dioxide, result in pial arteriolar dilation. However, in the rodent model of SAH, impaired microvascular reactivity in response to these interventions was observed.^{17, 67, 103} These changes peak at 48 hours and slowly resolve within the subsequent 5–7 days post injury. In addition, Friedrich et al. (2012) showed that greater than 70% of arterioles were constricted in diameter up to 72 hours after SAH, with smaller arterioles having more constriction.¹⁰³ Similar findings have been reported in other studies, and have described arteriolar constrictions in a “pearl string” pattern.^{57, 103, 110} Interestingly, microthombi which have commonly been observed throughout the brain following SAH, are commonly found in areas of arteriolar constriction.^{6, 103, 105} Microthrombosis further compromises cerebral perfusion and can lead to ischemia and neuronal cell death (Figure 1). Activation of the coagulation cascade, formation of microthrombi, and neuroinflammation are closely linked to one another through a process known as thromboinflammation.^{6, 14, 111, 112} Microthrombosis has been reviewed elsewhere.^{103, 113}

Several other structural and cellular changes within the cerebral microvasculature can be observed after SAH. Microvilli have been shown to develop and extrude from the vessel wall, forming blebs that can detach from the basal lamina and obstruct the lumen.^{104, 106} In addition to affecting blood flow directly, these changes can lead to exposure of the basal lamina, triggering both platelet and leukocyte adhesion, promoting microthrombosis and neuroinflammation, respectively. The role of pericytes in various cerebrovascular conditions including SAH has received increasing attention for their contribution to vessel tone and alterations in CBF.^114–116 Li et. al. (2016) showed that penetration of hemoglobin into the brain parenchyma following SAH in a rat model resulted in phenotypic transformation of pericytes to a hypercontractile form that resulted in reduction in microvessel diameter.¹¹⁶ This transformation was further shown to be dependent on reduction in NO/cyclic guanosine monophosphate (cGMP) signaling, a well-documented phenomenon after SAH described below. In addition to pericyte-mediated vasoconstriction, swelling of astrocytic end feet can further compromise blood flow.^{57, 104, 106} Astrocytes also appear to serve as a source of the endogenous vasoconstrictor ET-1 and undergo proliferation after SAH in the cortex and hippocampus.^{117, 118} Further, astrocytes following exposure to CSF containing blood appear to enter a metabolic crisis related to release of intracellular pools of calcium that also underlie alterations in neurovascular coupling.^{108, 119} Taken together, it appears that several cell types within the neurovascular unit collectively drive microvascular dysfunction after SAH.

Blood-Brain Barrier

SAH also disrupts the integrity of the BBB, which can further compromise cerebral perfusion and facilitate neuroinflammation. Leakage of endogenous proteins and injected dyes normally restricted from the CNS have been observed after experimental SAH.^{57, 110} This increased permeability of the BBB drives cerebral edema and intracranial hypertension which further compromise cerebral perfusion. Mechanistically, loss of BBB integrity has been associated with the upregulation of matrix metalloproteinases (MMPs) and other proteases which degrade tight junctions and the basal lamina.^{16, 120} MMP-9 in particular has emerged as a key player in post-SAH pathophysiology based on several studies in patients and animal models, contributing to global cerebral edema following degradation of extracellular matrix proteins and disruption of tight junctions.¹²¹ Increased expression and subsequent activation of MMP-9 can occur in response to reactive oxygen species and pro-inflammatory cytokines such as TNF-a and IL-17, all of which are increased after SAH.¹²² The source of MMP-9 in SAH is not well described, but evidence obtained in ischemia-reperfusion models indicates that neutrophils may constitute the major source of MMP-9 acting on the BBB.¹²¹ MMP-9 can also drive neuroinflammation via activation of pro-inflammatory signals and clotting factors, triggering a positive feedback loop promoting thromboinflammation and neurotoxicity.¹²² Indeed, increased MMP-9 in both plasma and CSF of SAH patients have been observed and some studies have shown a correlation with the extent of EBI, vasospasm, and DCI.^123–125 The correlation of MMP-9 with vasospasm in human cohorts; however, remains controversial.¹²⁵ Beyond MMP-9, recent studies have also shown upregulation of sulfonylurea receptor 1-transient receptor potential melastatin 4 (Sur1-Trpm4) after SAH in rats, and that this upregulation is associated with BBB dysfunction, neuroinflammation, and deficits in spatial learning and memory.¹²⁶ Importantly, blockade of this channel using antisense oligonucleotides or the Sur1 inhibitor glibenclamide reduced these deficits.¹²⁶ While a detailed review of BBB changes after SAH is outside the scope of this review, it is clear that disruption of the BBB after SAH is closely linked to neuroinflammation and contributes to poor outcome.

Vasoconstrictors and Vasodilators

As mentioned above, SAH also results in an imbalance between endogenous vasoconstrictors (i.e., ET-1) and vasodilators (i.e., NO). Changes in these vasoactive substances may also be triggered by the initial inflammatory response following SAH (Figure 1).^{127, 128} Studies of SAH patients have shown that activated mononuclear leukocytes in CSF synthesize and release ET-1, and that this occurs in parallel with the release of pro-inflammatory cytokines such as IL-1b.¹²⁸ ET-1 was the target of previous clinical trials targeting large conduit arteries using the ET-1 receptor antagonist clazosentan; however, despite a reduction in large-vessel vasospasm, there was no improvement in long-term functional outcome.^11–13 Clazosentan is currently being re-examined in a more focused manner in the REACT trial (clinical trial registration number , clinicaltrials.gov). Although results have not yet been published, the aim of this trial is to identify subgroups of patients which may derive benefit from targeting ET-1.

NO is an alternative target gaining attraction due to its ability to induce vascular dilation via cGMP-dependent relaxation of vSMCs and its involvement in the inflammatory response.^129–132 A constant supply of NO is important under normal homeostatic conditions in the maintenance of arteriolar diameter, in addition to preventing the activation of platelets and leukocytes.¹⁰⁶ This constitutive production of NO is mostly provided by neuronal nitric oxide synthase (nNOS) and endothelial NOS (eNOS), whereas NO involved in inflammatory processes is mainly produced by inducible NOS (iNOS).¹²⁷ Immediately following SAH, different mechanisms result in reduced bioavailability of NO, including decreased synthesis, uncoupling of eNOS, endothelial damage, upregulation of endogenous NOS inhibitors (such as asymmetric dimethyl arginine), and sequestration of NO by various byproducts including hemoglobin and reactive oxygen species through a sink effect.^{106, 116, 131} Free hemoglobin released from dying RBCs can undergo oxidation and serve as a strong NO scavenger in addition to suppression of NO signaling.¹¹⁶ NOS uncoupling refers to a pathological condition by which the NO synthesized by this enzyme reacts with superoxide anion (•O²⁻) and forms the reactive nitrogen species peroxynitrite (ONOO⁻) which has toxic effects on lipids, genetic material, and proteins, and contributes to endothelial dysfunction, vasoconstriction, and thrombosis.^{104, 106} Different NOS isoforms also undergo different changes following SAH – nNOS is primarily downregulated, iNOS is primarily upregulated, and eNOS undergoes complex changes characterized by decreased endothelial expression and increased parenchymal expression.¹²⁷ Upregulation of iNOS by cells such as microglia or astrocytes can generate large amounts of NO that leads to downstream inflammation and cytotoxicity through uncoupling.¹²⁷ This suggests that rather than a simplistic model of decreased NO following SAH, the specific enzymes producing NO at a particular time and place may in fact regulate both neuroinflammation and microvascular function.

Therapeutic targeting of NO may thus serve two purposes – reduction of pro-inflammatory mediators and attenuation of microvascular dysfunction. After SAH, infiltrating immune cells such as neutrophils or macrophages may increase production of inflammatory reactive nitrogen species through upregulation of iNOS while impairments in constitutive NO signaling can interfere with microvascular function.¹²⁷ Attempts to restore the balance of constitutive NO production have been effective in experimental models, including the use of genetic elimination of eNOS and more clinically-relevant NO supplementation using pharmacological NO donors and inhaled NO.^{56, 133–135} Drugs such as L-arginine and S-nitrosoglutathione showed efficacy in improving outcome after SAH in animal models,^{136, 137} but were associated with drops in systemic blood pressure.^{133, 138} However, inhaled NO has limited effects on systemic blood pressure and was shown in rodents to reduce the number and severity of microvascular constrictions with subsequent reduction in mortality and improvement in outcome.¹³³ In SAH patients, NO donors including sodium nitroprusside and transdermal nitroglycerin have been used.¹³⁹ Some of these studies showed promise; however, they were underpowered and side effects of systemic hypotension, headache, and rebound hypertension limit routine use.^139–143

Beyond targeting NO directly, many therapeutic strategies have sought to modulate NO production by interfering with vSMC relaxation in other ways. Phosphodiesterase V (PDE-V) is a key regulator of the eNOS-NO-cGMP pathway that hydrolyzes cGMP and prevents vSMC relaxation and subsequent vasodilation. Inhibition of PDE-V using sildenafil showed promising results in experimental SAH and was recently tested in a phase I safety and proof-of-concept trial.^{144, 145} Other PDE inhibitors were tested which have more direct effects on vSMCs themselves, including the PDE-III inhibitor milrinone which showed some efficacy at reducing vasospasm and improving outcome (Table 1).^146–151 Besides PDE inhibitors, magnesium sulfate showed promise in experimental SAH with reduction in cerebral infarct size, reversal of vasospasm, and improved cerebral perfusion based on its ability to promote vSMC relaxation.^152–155 However, two large phase III clinical trials failed to demonstrate clinical benefit.^156–158 Some recent studies have suggested that the use of higher dose magnesium sulfate may have some benefit, although this deserves further study.¹⁵⁹ Thus, although targeting of vasoconstrictive mediators such as ET-1 did not appear to improve long-term outcome after SAH, perhaps targeting dysfunction of vasodilatory molecules such as NO may prove efficacious.

In addition to ET-1 and NO, additional potent vasomodulators have been described which may serve as therapeutic targets. Such targets include arachidonic acid and its metabolites.^{160, 161} One of the most well-studied of this family is 20-hydroxyeicosatetraenoic acid (20-HETE), shown to be elevated following SAH in patients and animal models.^160–163 Produced by cytochrome P450 enzymes in vSMCs, neurons, and glia, 20-HETE can induce vasoconstriction.¹⁶⁰ Mechanistically, 20-HETE levels are increased following loss or scavenging of NO.^{160, 161} Selective inhibition of 20-HETE synthesis using pharmacological inhibitors reverses delayed vasospasm and improves acute CBF recovery (Table 1).^163–165 20-HETE levels are elevated in the CSF of SAH patients, and this elevation is associated with acute and long-term outcomes.^{166, 167} Another arachidonic acid metabolite, 14,15-epoxyeicosatrienoic acid (14,15-EET), may be protective against the actions of 20-HETE.¹⁶⁸ The cumulative data suggests that arachidonic acid metabolites play an active role in SAH pathophysiology, and may offer novel therapeutic targets.

Discussion

Through extensive has become investigation of neuroinflammation and microvascular dysfunction after SAH, it clear that they play an important role in EBI and contribute to poor outcome. These two mechanisms are also tightly linked, as pro-inflammatory signaling can promote disruption of the microvasculature and vice versa (Figure 1). The release of RBC components such as hemoglobin into the subarachnoid space following aneurysmal rupture likely triggers an initial inflammatory reaction by microglia, which secrete numerous pro-inflammatory chemokines and cytokines. These signals increase expression of CAMs on endothelia, drive peripheral leukocyte transmigration, and may also promote microvascular dysfunction. Meanwhile, changes in NO bioavailability coupled with damage to BBB and neurovascular unit dysfunction likely compromise vascular tone regulation and lead to the formation of microthrombi. One exciting area of investigation is the role of cortical spreading depolarizations after SAH, which may be related to both neuroinflammation and microvascular dysfunction.^169–171 Taking into account the various mechanistic changes occurring in the brain after SAH, management must be comprehensive and play close attention to acute and delayed brain injury as well as systemic complications. Systemic complications of SAH include hyperglycemia, fever, infection, and dysregulation of coagulation cascades, all of which can influence clinical outcome. Glycosylated hemoglobin, monomeric CRP, and other biochemical mediators have been associated with outcome in other stroke subtypes.^{172, 173} However, their value in SAH has not been conclusively demonstrated. While large vessel cerebral vasospasm likely contributes to poor outcome, it is no longer believed to be the determining factor and additional studies of neuroinflammation and microvascular dysfunction will likely provide both mechanistic information and therapeutic targets (Table 1).

Despite the evidence of a clear role of neuroinflammation and microvascular dysfunction in poor outcomes after SAH, there are some limitations to current studies. One such limitation is the lack of a standardized animal model of SAH. Studies use a variety of different animal models ranging from autologous blood injection to endovascular perforation models, contributing to variability both within species and between species.^{30, 174, 175} Within these models, those such as the endovascular perforation model have a relatively high mortality rate, and thus studies may only be conducted on those animals which survive and thus may have limited EBI. The endovascular perforation model provides translational relevance to human SAH by recapitulating a hemorrhagic lesion under arterial pressure but may also have high variability in the location and severity of hemorrhage. The more commonly used blood injection models are easier to control and have lower mortality rates; however, a limitation to these models is that they do not reproduce the complex hemodynamic changes seen in SAH and therefore may have limited EBI.¹⁷⁶ One additional shortcoming of using rodent models to investigate the role of inflammation in human SAH is related to differences in immune responses across species, including the major immune cells involved in the response as well as the timeline and major signaling pathways.^{177, 178} At the level of clinical studies, neuroinflammation and microvascular dysfunction are much more difficult to assess compared to large vessel vasospasm, although studies for various biomarkers in both the peripheral blood and CSF are underway.

Future experimental studies in preclinical models should focus on a multi-pronged effect of targeting both neuroinflammation and microvascular dysfunction to improve outcomes after SAH. While single molecular targets have shown promise in experimental SAH, they have not easily translated to success in clinical trials. Targeting neuroinflammation may alleviate the microvascular dysfunction observed after SAH, as the initial inflammatory response may serve as the primordial factor that contributes to abnormal vascular reactivity and imbalance of endogenous vasodilators and vasoconstrictors. Further, several drugs currently under investigation show promise in targeting both neuroinflammation and microvascular dysfunction, and may be effective in improving outcomes after SAH. Statins were considered an attractive target due to their pleiotropic effects including anti-inflammation, neuroprotection, and increase in eNOS.^179–181 However, clinical trials have had less success demonstrating a significant effect on DCI, infarction, or mortality.^182–184 More promising results have been obtained with low-dose, unfractionated heparin, which has several biologic effects independent of its anticoagulant properties. By complexing with oxyhemoglobin, heparin can block the formation of free radicals and act as an antagonist to ET-1-mediated vasoconstriction and cytokine-mediated neuroinflammation.^185–188 Recently, clinical trials of low-dose unfractionated heparin in SAH patients have shown a favorable safety profile, reduction in DCI without a change in angiographic vasospasm, and improved cognitive outcomes.^{187, 189} These data have supported the initiation of an ongoing, large-scale, randomized control trial, the Aneurysmal Subarachnoid Hemorrhage Trial Randomizing Heparin (ASTROH, clinical trial registration no. , clinicaltrials.gov).

In summary, a body of evidence supports the notion that the pathophysiology of brain injury in SAH is multifactorial and targeting only one process will likely be insufficient to derive clinical benefit. The complex interplay of microvascular dysfunction and neuroinflammation point to new and exciting areas of current investigation that may result in the development of new therapeutics that reduce long-term impairments after SAH.