Regulation of the Blood–Brain Barrier in Health and Disease

Abstract

The neurovascular unit is a dynamic microenvironment with tightly controlled signaling and transport coordinated by the blood–brain barrier (BBB). A properly functioning BBB allows sufficient movement of ions and macromolecules to meet the high metabolic demand of the central nervous system (CNS), while protecting the brain from pathogenic and noxious insults. This review describes the main cell types comprising the BBB and unique molecular signatures of these cells. Additionally, major signaling pathways for BBB development and maintenance are highlighted. Finally, we describe the pathophysiology of BBB diseases, their relationship to barrier dysfunction, and identify avenues for therapeutic intervention.

The cerebrovasculature is a highly specialized vascular bed that tightly regulates the movement of nutrients, ions, and cells between the blood and the brain parenchyma. The stringent integrity of brain blood vessels is controlled by a series of properties termed the blood–brain barrier (BBB). The BBB selectively allows passage of ions and macromolecules to meet the high metabolic demands of the brain, while specifically preventing transmigration of toxins and pathogens to protect particularly sensitive neurons from injury. While the BBB protects the brain from pathogenic and noxious particles, it also prevents passage of cells, antibodies, and pharmaceuticals into the parenchyma, creating a unique problem for drug delivery to the brain. Here, we discuss cell types, molecular signatures, and major signaling pathways of the BBB and the pathophysiology of these core BBB components in diseases.

CELL TYPES

The BBB is comprised of a neurovascular unit (NVU) consisting of vasculature, neuronal cells, and extracellular matrix (ECM) (Fig. 1). Vascular cells include endothelial cells ([ECs] the innermost lining of blood vessels) and mural cells (pericytes and smooth muscle cells), shared with the peripheral vasculature, yet with NVU-specific properties. First, brain EC have increased intercellular tight junctions creating a much tighter barrier. Brain EC also have BBB-specific efflux and nutrient transporters to allow movement of essential ions and nutrients into the brain and to convey waste products from the brain. Brain ECs exhibit decreased expression of adhesion molecules such as ICAM1 and VCAM1 to prevent immune cell transmigration into the brain parenchyma (May et al. 1993; Male et al. 1994; Engelhardt and Ransohoff 2012). Instead, the brain possesses macrophage-like resident immune cells termed microglia, which are not derived from blood-associated myeloid cells. This renders microglia functionally and developmentally distinct from monocytes and provide a critical brain-immune niche (Li and Barres 2018).

Figure 1.

The neurovascular unit (NVU) in healthy brain, stroke brain, Alzheimer's disease brain, and multiple sclerosis brain. Dashed box shows blood–brain barrier (BBB)-specific molecular changes for each healthy or diseased NVU. (Figure created with BioRender.com.)

Pericytes cover the abluminal side of the endothelium and are connected to ECs by the adhesion molecule, N-cadherin (Gerhardt et al. 1999; Tillet et al. 2005). The pericyte coverage of endothelium is 30–100× denser in brain vasculature versus peripheral vasculature (Shepro and Morel 1993). Smooth muscle cells in the cerebrovasculature surround larger arteries and arterioles and regulate blood flow in response to neuronal activity. Pericytes also regulate vessel diameters (Hall et al. 2014), although to a much lesser extent than smooth muscle cells.

Astrocytes are a major glial cell that surrounds the vasculature. Traditionally, astrocytes were thought to provide structural support to the BBB but are now appreciated to control neuronal and vascular functions. Astrocytes mediate neurovascular coupling, linking blood flow and neuronal activity (Attwell et al. 2010; Choi et al. 2012; Liu et al. 2018). A single astrocyte can be in contact with both the vasculature and multiple neuronal synapses (Oberheim et al. 2009; Liu et al. 2018). Additionally, astrocytes ferry glucose and oxygen from blood to neurons (Choi et al. 2012).

Surrounding brain vascular cells is a basement membrane composed of ECM synthesized by pericytes and ECs and comprised of collagen IV, laminin, nidogen and perlecan, and other glycoproteins (Chen et al. 2019b). A second basement membrane, termed the glia limitans, forms the final separation from the brain parenchyma and is formed by astrocyte endfeet processes and astrocyte secretions (Horng et al. 2017). Whereas pericytes cover a significant portion of the EC, astrocytic endfeet completely enwrap the vasculature, thus comprising the final layer of the BBB.

MOLECULAR SIGNATURES OF THE BBB

The BBB has many unique molecular features that differentiate it from the peripheral vasculature. A well-established characteristic of brain EC is increased expression of tight junction proteins, which prevent passive diffusion, creating a high molecular weight barrier only allowing passage of nonpolar molecules <400 kDa in size. Tight junctions are comprised of occludins, claudins, and junctional adhesion molecules that interact to provide junctional stability through interactions with zona occludin proteins to connect to the actin cytoskeleton. Additionally, tight junctions function as rheostats and increase barrier properties, and thus are expressed most highly in the BBB, whereas the peripheral vasculature has more adherens junctions (Harris and Nelson 2010; Daneman and Prat 2015). The enhanced barrier function of tight junctions in the cerebral vasculature is demonstrated in mice lacking claudin-5, the highest expressed tight junction, which exhibit BBB defects and early postnatal death (Nitta et al. 2003).

To exclude toxic molecules, the BBB expresses specific efflux transporters that transport substances back into the blood from the ECs. One example is P-glycoprotein (Pgp), which exports cell-permeable foreign substances such as drugs. Pgp is a multidrug resistance protein expressed in tumor cells and the BBB, providing a substantial challenge for CNS drug delivery (Aller et al. 2009; Aryal et al. 2017). To allow transport of larger or polar molecules into the brain parenchyma, highly specialized endothelial transporters mediate transcellular permeability. Multiple transporters facilitate the movement of molecules such as glucose and amino acids against their concentration gradients into the brain (nutrient transporters) as well as large molecules including transferrin, insulin, and leptin. The GLUT1 glucose transporter is highly expressed in brain endothelium; expression is often modulated in neurodegenerative disorders. Additionally, the transferrin receptor (TfR) has emerged as an important therapeutic target for movement of covalently modified antibodies into the brain and is discussed below (Jefferies et al. 1984; Niewoehner et al. 2014).

The major facilitator superfamily domain-containing protein 2 (MFSD2A) is a specific marker of BBB ECs (Ben-Zvi et al. 2014; Wood et al. 2021). MFSD2A is expressed only in BBB ECs, and not surrounding pericytes or the choroid plexus vasculature (CNS blood vessels lacking a BBB). MFSD2A specifically transports docosahexaenoic acid (DHA), an omega-3 fatty acid that is required for brain function and is not synthesized de novo in brain (Nguyen et al. 2014). Genetic ablation of Mfsd2a induces BBB disruption through increased vesicular transcytosis, independent of tight junctions (Ben-Zvi et al. 2014).

Interestingly, many BBB transporters, including MFSD2A and TfR, decrease during aging, potentially producing a “leakier” BBB. Additionally, the specific transport of younger brains decreases with age, whereas older brains have increased caveolin gene expression and nonspecific IgG influx (Yousef et al. 2019; Yang et al. 2020), indicating that the type rather than the amount of plasma protein leak is more indicative of aging.

Single-cell transcriptional and epigenetic analysis of BBB-specific endothelial beds and mural cells have emerged as powerful tools to elucidate cell-type molecular signatures (Sabbagh et al. 2018; Saunders et al. 2018; Vanlandewijck et al. 2018; Munji et al. 2019; Garcia et al. 2022). The cerebrovasculature exhibits higher expression of transcription factors in cerebral arteries, whereas transporters are more prevalent in veins and capillaries (Vanlandewijck et al. 2018; Yang et al. 2020). Cerebrovascular zonation exists in six categories including tip cells, mitotic, venous, capillary-venous, capillary-arterial, and arterial ECs, each with novel markers (Sabbagh et al. 2018). Single-cell sequencing has also identified unique aspects of the NVU such as novel perivascular Lama1⁺ fibroblast-like cells surrounding all cerebrovascular cells except capillaries (Vanlandewijck et al. 2018). Human-specific neurovascular transcriptome signatures have also been described (Garcia et al. 2022). Experimental validation is still required for many of these genes to define their underlying biology.

SIGNALING PATHWAYS ESSENTIAL FOR BBB DEVELOPMENT AND MAINTENANCE

NVU development and homeostasis are tightly regulated by intercellular signaling events culminating in a functional BBB. Spatiotemporal signaling of multiple, highly coordinated pathways leads to formation of CNS vasculature. Additionally, many of these same signaling mediators are required for BBB maintenance and modulation. Here, we highlight key signaling pathways regulating BBB generation and maintenance.

VEGF Signaling

Vascular endothelial growth factor (VEGF) is a canonical driver of capillary guidance during embryonic sprouting angiogenesis (Gerhardt et al. 2003; Raab et al. 2004). VEGF also potently induces BBB leakage (Ferrara et al. 2003; Jiang et al. 2014). VEGF-driven brain neovascularization and angiogenesis are tightly regulated at both ligand and receptor levels (Simons et al. 2016). The VEGF family includes VEGF-A, -B, -C, -D, and PlGF that bind VEGF receptors (VEGFRs) 1–3 and neuropilin coreceptors (NRP1 and NRP2) (Ferrara et al. 2003; Groblewska and Mroczko 2021). VEGFR2 and NRP1 primarily mediate brain angiogenesis, with VEGFR2 further required for vascular regeneration upon head injury in adult mice and humans (Gu et al. 2003; Koh et al. 2020). VEGF receptors exhibit spatially differentiated expression with VEGFR1 luminal and VEGFR2 abluminal, facing neuronal tissues (Hudson et al. 2014). VEGFR1, also known as FLT1, possesses a nonessential kinase domain (Rahimi 2006) and negatively regulates angiogenesis through a decoy-soluble (s)FLT1 receptor; stalk cells up-regulate VEGFR1 to decrease sensitivity to VEGF and inhibit angiogenesis (Kendall and Thomas 1993; Rahimi 2006). Hypoxia-induced pericytes to secrete VEGF, creating a gradient for endothelial tip cells to migrate to create new vascular beds (Fig. 2; Darland et al. 2003).

Figure 2.

Signaling schematic for WNT7, Norrin, and vascular endothelial growth factor (VEGF) in the neurovascular unit (NVU). Ligands are released from NVU cells to bind receptor or receptor complexes on the brain endothelium and elicit specific responses. (Figure created with BioRender.com.)

VEGF is also a potent driver of neurogenesis (Eichmann and Thomas 2013). Multiple NVU cell types either secrete or receive VEGF signals. Hypoxic astrocytes secrete VEGF to affect tip cell migration through VEGFR2 during brain angiogenesis (Gerhardt et al. 2003), or induce BBB permeability in fully developed brains (Franco et al. 2011; Argaw et al. 2012). VEGF secreted from hypoxic pericytes can induce angiogenesis and can also affect BBB permeability through claudin-5 disruption (Bai et al. 2015). VEGF also plays a major role in BBB pathophysiology as discussed below.

GPR124/RECK/WNT7 Signaling

Canonical WNT/β-catenin signaling is activated by binding of WNT ligands to Frizzled (FZD1-10) receptors in complex with coreceptor lipoprotein receptor–related protein 5/6 (LRP5/6). Stabilized β-catenin then translocates from the cytoplasm to the nucleus, binds with the TCF/LEF transcription factors, and activates transcription of WNT target genes (Nusse and Clevers 2017). Canonical WNT signaling is specifically activated in CNS ECs during CNS development (Liebner et al. 2008; Daneman et al. 2009). Canonical WNT ligands WNT7A and WNT7B are expressed by the neural progenitors in the ventral region of the forebrain and ventrolateral spinal cord during CNS development (Daneman et al. 2009). Wnt7a/Wnt7b double-knockout (KO) mice as well as endothelial-specific β-catenin KO mice exhibit vascular defects in CNS with the decreased expression of GLUT1 accompanied by severe hemorrhaging (Stenman et al. 2008; Daneman et al. 2009). Together, Wnt7a/Wnt7b are necessary for developmental CNS angiogenesis and BBB formation.

The GPR124 and RECK receptors are essential coactivators that selectively regulate WNT7 signaling in CNS endothelium (Fig. 2). The orphan adhesion G protein–coupled receptor, GPR124 (TEM5/ADGRA2), is predominantly expressed in CNS EC and pericytes (Kuhnert et al. 2010; Anderson et al. 2011; Cullen et al. 2011). Reversion-inducing cysteine-rich protein with Kazal motifs (RECK) is widely expressed in mesenchymal tissues including blood vessels in endothelial and mural cells (Oh et al. 2001; de Almeida et al. 2015). Genetic studies using transgenic mice and zebrafish and in vitro cell-based WNT reporter assays reveal that GPR124/RECK are associated with WNT7A/B signaling. Gpr124 KO (Kuhnert et al. 2010; Anderson et al. 2011; Cullen et al. 2011; Zhou and Nathans 2014; Posokhova et al. 2015) and Reck KO (Oh et al. 2001; de Almeida et al. 2015; Cho et al. 2017b) mice exhibit CNS angiogenesis defects in the forebrain and the ventral spinal cord. These closely phenocopy Wnt7a^−/−; Wnt7b^−/− embryos, which manifest additional angiogenic defects in midbrain, hindbrain, and dorsal spinal cord. Notably, Gpr124, Reck, and Wnt7a/Wnt7b KO mice are further unified by (1) embryonic lethality, severe CNS hemorrhage, (2) markedly reduced angiogenic sprouting with near avascularity, (3) CNS vascular malformations, and (4) loss of EC Glut-1 expression. Importantly, phenotypes of both Gpr124 KO and Reck KO embryos are rescued by endothelial-specific constitutive activation of canonical WNT/β-catenin signaling (Zhou and Nathans 2014; Cho et al. 2017b). Zebrafish Gpr124 and Reck mutants demonstrate that both GPR124 and RECK are required for endothelial-specific WNT signaling during brain vessel development and for tip cell–specific angiogenic sprouting in the brain (Vanhollebeke et al. 2015). These data implicate both GPR124 and RECK in CNS angiogenesis and BBB formation through WNT signaling.

The intersection of GPR124, RECK, and WNT7A/B with β-catenin signaling has been investigated mechanistically (Posokhova et al. 2015; Cho et al. 2017b; Eubelen et al. 2018; Vallon et al. 2018). In vitro studies of GPR124 and RECK using the TOP-Flash WNT reporter assay demonstrate that GPR124 and RECK together dramatically enhance signaling by WNT7A/B but not by the other 17 WNT family members; this is further augmented by cotransfection of numerous FZD species (Vanhollebeke et al. 2015; Vallon et al. 2018). RECK and GPR124 ectodomains physically interact (Vanhollebeke et al. 2015; Cho et al. 2017b; Vallon et al. 2018). In unbiased coimmunoprecipitation studies from rat brain blood vessels, RECK was also identified as a predominant binding partner of GPR124 (Vallon et al. 2018). Notably, RECK is a WNT7 coreceptor capable of binding full-length WNT7 (Vallon et al. 2018) or WNT7 peptides (Eubelen et al. 2018). WNT7 rapidly aggregates into an inactive hydrophilic form upon secretion from cells, but RECK stabilizes newly produced WNT7 into an active monomeric species and thus dramatically enhances WNT7 binding to FZD8 (Vallon et al. 2018).

GPR124/RECK interactions with WNT/FZD are also being explored. It has been proposed that the intracellular FZD effector, Dishevelled (Dvl), is recruited to the GPR124 intracellular domain (ICD). Upon FZD and GPR124 bridging, Dvl polymerizes and enhances WNT signaling. Further, co-IP and colocalization studies reveal constitutive binding of Dvl to the GPR124 ICD. The GPR124 ICD is essential for enhanced RECK/WNT7 signaling as zebrafish ICD deletion mutants are inactive in promoting RECK/WNT7-dependent WNT reporter activity in vitro and zebrafish CNS angiogenesis in vivo (Eubelen et al. 2018). Conversely, it has been demonstrated that GPR124 ICD deletion mutants still mediate WNT7 signaling, albeit at partially suppressed levels, and that the GPR124-recombinant ECD is sufficient to augment WNT7 signaling (Vallon et al. 2018). Initial WNT7 binding to RECK increases subsequent binding of WNT7 to FZD8 by RECK, suggesting that initial WNT7 stabilization by GPR124/RECK may promote subsequent WNT7 transfer to FZD/LRP (Vallon et al. 2018). Further studies will be needed to further define these mechanisms.

NORRIN/Fzd4 SIGNALING

Norrin, encoded by the Norrie disease protein (NDP/Ndp) gene, is a secreted cysteine knot-like growth factor (Berger et al. 1992; Chen et al. 1992) that activates the canonical WNT/β-catenin pathway by interaction with WNT receptor FZD4, lipoprotein receptor-related protein 5 (LRP5), and the auxiliary four-pass transmembrane protein tetraspanin-12 (Tspan12) (Fig. 2; Junge et al. 2009; Ke et al. 2013). Of the 10 FZD family members, Norrin specially binds to only FZD4 (Smallwood et al. 2007). X-ray crystallography reveals that two FZD4 molecules interact with Norrin dimers, which in turn bind LRP5 (Chang et al. 2015).

Norrin is essential for brain, spinal cord, and retina development (Xu et al. 2004; Liebner et al. 2008; Stenman et al. 2008; Daneman et al. 2009; Wang et al. 2012; Zhou et al. 2014). Mutation of the Ndp, or Norrin receptors (Fzd4, Lrp5, Tspan12) causes human vitreoretinal diseases including Norrie disease, familial exudative vitreoretinopathy, and Coats’ disease (Ye et al. 2010; Ohlmann and Tamm 2012). Norrin signaling is required to maintain BBB and blood–retina barriers (BRBs). Ndp mutation leads to impaired barrier competence, while restoration of Norrin expression or rescue of Norrin by downstream WNT signaling activation is sufficient to rescue the Ndp mutant phenotype (Wang et al. 2012, 2018; Zhou et al. 2014; Díaz-Coránguez et al. 2020). Norrin also plays an essential role in CNS angiogenesis, with mutations of Ndp, Fzd4, Lrp5, and Tspan12, resulting in incomplete brain and retina vascularization (Xu et al. 2004; Junge et al. 2009; Wang et al. 2012, 2018). Norrin also effects other parts of the NVU and has neuroprotective effects on retinal neurons (Ohlmann et al. 2010; Seitz et al. 2010).

Although the Norrin and WNT7A/B pathways both activate β-catenin signaling in vascular ECs to control embryonic BBB and BRB development and maintenance, there are several differences. At the level of receptor complexes, WNT7A/B signaling requires GRP124, RECK, FZD, and LRP5/6 to functionally interact to enhance β-catenin signaling (Zhou and Nathans 2014; Posokhova et al. 2015; Cho et al. 2017a; Eubelen et al. 2018). However, Norrin only binds FZD4 together with LRP5 and TSPAN12 to activate β-catenin signaling (Junge et al. 2009; Ke et al. 2013). Additional differences in tissue and cell-type specificity exist in the cerebral cortex; WNT7A/B are expressed by astroglia, oligodendrocytes, and neurons, whereas Norrin is expressed by astroglia and oligodendrocytes (Fig. 2; Zhang et al. 2014). In the retina, Norrin is expressed by Müller glia cells (Ye et al. 2009) and embryonic angiogenesis is controlled by the Norrin system but not WNT7A/B (Xu et al. 2004; Junge et al. 2009; Ye et al. 2009; Wang et al. 2012). In contrast, embryonic angiogenesis in cerebral cortex and medial ganglionic eminences are predominantly controlled by WNT7A/B and not Norrin (Stenman et al. 2008; Daneman et al. 2009; Kuhnert et al. 2010; Cullen et al. 2011). Hindbrain angiogenesis, however, is controlled by both Norrin and WNT7A/B systems, with developmental defects only observed upon simultaneous mutation of both pathways (Zhou and Nathans 2014; Cho et al. 2017a; Wang et al. 2018). Redundancy between these systems can occur, since in the postnatal brain, combined loss of Norrin and either GPR124 or RECK results in more severe BBB defects in cortex, thalamus, and brainstem than deficiency in either pathway alone (Zhou and Nathans 2014; Cho et al. 2017a).

BBB DYSREGULATION IN PATHOLOGICAL CONDITIONS

Brain Tumor Angiogenesis

Gliomas are the most common primary intracranial tumors, accounting for 70%–80% of all brain neoplasms (Ohgaki 2009; Ostrom et al. 2017). Among subtypes of glioma, glioblastoma (GBM) is the most aggressive and invasive, with survival time of 12–15 mo after diagnosis and <5% patients surviving past 5 yr (Louis et al. 2007; Dolecek et al. 2012). GBM is highly vascular and exhibits aggressive local dissemination (Louis et al. 2007). GBM angiogenesis is influenced by VEGF, transforming growth factor-β, fibroblast growth factors (FGFs), epidermal growth factor, and angiopoietin-1/2, which are released from GBM cells in the hypoxic tumor microenvironment (Stacker et al. 2001; Bhattacharya et al. 2015; Shim and Madsen 2018; Diniz et al. 2019; Huang et al. 2019; Rattner et al. 2019; Ardizzone et al. 2020).

VEGF is highly expressed by brain tumor cells and mediates BBB dysfunction during tumor growth (Ahir et al. 2020). VEGF synergizes with other factors, including FGFs and PlGF, to induce angiogenesis (Carmeliet et al. 2001; Sun et al. 2004). VEGF is detected in low-grade gliomas and higher levels are observed in necrotic regions of high-grade gliomas/GBM (Phillips et al. 1993; van Tellingen et al. 2015; Di Tacchio et al. 2019; Nicolas et al. 2019). VEGF and its receptors are prognostic biomarkers for glioma patients (Yao et al. 2001; Hervey-Jumper et al. 2014; Jiang et al. 2018) and VEGF inhibition is used therapeutically for GBM patients (Kim et al. 2018; Di Tacchio et al. 2019); the anti-VEGF monoclonal antibody bevacizumab was the first antiangiogenesis agent approved for clinical use for brain cancer with antiglioma activity in recurrent GBM patients (Kreisl et al. 2009). A major mechanism of the antiglioma action of VEGF inhibition may reside in the inhibition of VEGF vascular permeability factor activity, which improves BBB function and decreases edema (Gerstner et al. 2009). Moreover, addition of bevacizumab to other therapeutics improves progression-free and overall survival (Taal et al. 2014; Jakobsen et al. 2018).

Additional cerebrovascular signaling pathways have emerged as novel therapeutics and diagnostic strategies for brain tumors. Hypoxia-induced gene 2 (HIG2), a marker of hypoxia, has been reported to be a diagnostic biomarker for several cancers including GBM (Kim et al. 2013). Loss of the PTEN tumor suppressor increases VEGFR2 expression in GBM tumor cells, possibly facilitating resistance against antiangiogenic treatments (Paul-Samojedny et al. 2015; Bidinotto et al. 2016); inhibition of PTEN affects BBB permeability (Ding et al. 2013). Novel small molecule inhibitors of the hypoxia-inducible factor-2α transcription factor (HIF-2α) (Wallace et al. 2016; Choueiri et al. 2021) could be used to indirectly antagonize the downstream target VEGF and its effects on BBB permeability.

Cerebral Cavernous and Arteriovenous Malformations

Cerebral cavernous malformations (CCMs) and arteriovenous malformations (AVMs) are comprised of irregular and enlarged collections of brain blood vessels, and represent low- and high-flow lesions, respectively. Brain AVMs are either sporadic (80%) or genetically linked (20%) and highly aggressive manifestations can cause stroke or seizures in young adults (Fischer et al. 2013).

CCMs are classically derived from a loss-of-function (LOF) mutation in one of the three CCM complex genes, KRIT1, CCM2, or PDCD10 in ECs (Fischer et al. 2013). CCM gene LOF leads to increase in MEKK3 and KLF2/4 signaling, which are important signaling pathways in endothelial shear stress and in mediation of barrier properties (Zhou et al. 2016; Ren et al. 2021). However, a new paradigm suggests a model akin to cancer in which co-occurrence of (1) LOF of a CCM gene that constrains vessel growth, and (2) gain-of-function of PIK3CA, together cause aggressive CCM growth (Ren et al. 2021). PIK3CA is important signaling molecule for angiogenesis and survival, creating excessive, leaky cerebrovascular without a proper BBB (Zhao et al. 2017).

Traumatic Brain Injury (TBI)

TBI results from external forces on the head and impairs cognitive, physical, and psychosocial functions (Maas et al. 2008). TBI is classified as mild (80% of cases), moderate, and severe based on severity. Even mild TBI has serious and long-lasting effects and is a long-term risk factor for neurodegenerative diseases including Alzheimer's disease (AD) (Smith et al. 2013). BBB disruption is observed in both mild TBI patients and animal models (Wu et al. 2020). TBI induces acute microvascular injury and BBB disruption accompanied by EC death, decreased expression of tight junction proteins, and up-regulation of matrix metalloproteinases (MMPs), increased BBB permeability, and inflammation (Shlosberg et al. 2010; Jullienne et al. 2016; Wu et al. 2020). Recruitment of immune cells such as neutrophils, activation of microglia, up-regulation of cytokines, chemokines, and other proinflammatory mediators increase BBB permeability acutely (Shlosberg et al. 2010; Wu et al. 2020) and long-term (Jullienne et al. 2016; Sandsmark et al. 2019). BBB dysfunction is linked to chronic neuroinflammation, neuronal damage, and accumulation of pathologic products such as amyloid β, that can promote neurodegeneration (Jullienne et al. 2016).

Stroke

After ischemic stroke, BBB breakdown results in brain edema, hemorrhagic transformation, and neuroinflammation, leading to further brain injury, neurological impairment, and poor clinical prognosis (Prakash and Carmichael 2015). Thrombolytic therapy such as tissue plasminogen activator (tPA) is the only pharmacologic treatment for ischemic stroke, but it has a restricted therapeutic time window and is associated with increased risk of intracerebral hemorrhagic transformation (Sussman and Connolly 2013).

Mechanisms underlying increased BBB permeability after cerebral ischemia injury are poorly understood. BBB disruption begins by tight junction disassembly, deregulation of transporter properties, and ECM degradation. Animal stroke models such as transient middle cerebral ischemia show ischemia reperfusion injury induces early structural EC changes with subtle BBB leakage. These initial changes in the BBB facilitate the infiltration of immune cells, leading to junctional and ECM degradation and a secondary increase in BBB permeability allowing larger macromolecule leakage into the brain parenchyma (Fig. 2; Knowland et al. 2014; Shi et al. 2016). Transcriptome analysis of EC in mouse stroke and TBI models reveal similar endothelial gene expression changes during BBB disruption, shifting the CNS EC phenotype to one akin to that of peripheral endothelium (Munji et al. 2019).

BBB impairment can be initiated by cytoskeletal rearrangement in brain EC and endothelial death via oxidative stress from free radical generation after ischemia. Stressed ECs and perivascular cells promote expression of chemokines, cytokines, and adhesion molecules. The inflammatory response recruits peripheral immune cells to sites of injury, causing neuronal impairment (Shi et al. 2019). Perivascular activated microglia also contribute to post-stroke neuronal injury and BBB breakdown (Jolivel et al. 2015; Su et al. 2017; Chen et al. 2019a). Additionally, activated MMPs degrade endothelial tight junction and ECM (Yang et al. 2007; Liu et al. 2012), exacerbating BBB injury. MMP-9 and plasma albumin are used as biomarkers for BBB injury in experimental stroke and clinical studies (Jickling and Sharp 2011; Li et al. 2018).

As described, canonical WNT signaling induces BBB formation and maturation. Active β-catenin levels are induced predominantly in brain EC early after ischemic stroke in mouse models (Jean LeBlanc et al. 2019). Yet, low levels of β-catenin are observed near bleeding sites in the brain of hemorrhagic stroke patients (Tran et al. 2016) and WNT target genes are decreased in mouse stroke vasculature (Chang et al. 2017). Endothelial-specific Gpr124 KO and pharmacological inhibition of WNT/β-catenin signaling by the XAV939 compound enhance BBB breakdown and increase hemorrhaging in the mouse brain after ischemic stroke (Chang et al. 2017; Jean LeBlanc et al. 2019). Conversely, genetic activation of WNT/β-catenin signaling in Gpr124 KO mice and pharmacological WNT pathway activation by GSK3β inhibitors in stroke models rescue BBB disruption and hemorrhagic defects (Chang et al. 2017; Jean LeBlanc et al. 2019), implicating WNT/β-catenin signaling in postischemic maintenance of BBB integrity. Targeting the NVU by protecting BBB integrity and function could elicit long-term functional improvement after stroke and extend the therapeutic time window of interventions such as tPA or stenting.

NEURODEGENERATIVE DISEASES

Neurodegenerative diseases are prevalent in the aging population. While minor BBB breakdown occurs during aging, BBB degeneration is significantly enhanced in many neurodegenerative diseases (Montagne et al. 2015). BBB breakdown permits invasion of xenobiotics and non-CNS resident cells into the brain parenchyma, eliciting inflammatory and immune responses. CNS inflammation is thought to induce multiple pathways leading to neurodegeneration. Indeed, neurodegenerative BBB breakdown is associated with microbleeds, impaired glucose transport, deregulation of tight junctions, altered Pgp function, and erythrocyte and leukocyte infiltration (Sweeney et al. 2018).

Alzheimer's Disease

In the most prevalent neurodegeneration, AD, cerebrovascular damage precedes neuronal injury and amyloid β (Aβ) accumulation (Ujiie et al. 2003; Ramos-Cejudo et al. 2018). Pericyte loss drives many BBB defects characteristic of AD as pericytes can both clear Aβ and promote endothelial angiogenesis (Sagare et al. 2013; Sengillo et al. 2013). BBB disruption induces microbleeds and extravasation of erythrocyte, neutrophils, and nonresident macrophages, which illicit inflammatory responses in the AD brain (Fig. 1; Fiala et al. 2002; Zenaro et al. 2015). Additionally, pericytes remaining after AD progression can cause further capillary blood flow restriction through reactive oxygen species (ROS) released by Aβ accumulation (Nortley et al. 2019). ROS production then induces endothelin-1 signaling in pericytes to restrict capillary blood flow and further preventing oxygen and nutrients from reaching ailing neurons (Nortley et al. 2019).

Besides the mechanical breakdown of BBB in AD, many molecular alterations are present. Pericytes, together with endothelium, translocate Aβ from the brain through the BBB via the low-density LRP1 transporter. However, during AD, LRP1 is down-regulated in EC and pericytes allowing accumulation of Aβ plaques (Shibata et al. 2000; Deane et al. 2004; Shinohara et al. 2017). Aβ toxicity to LRP1-expressing pericytes may facilitate this down-regulation, forming a negative feedback loop inducing further Aβ accumulation. Additionally, levels of receptors for advanced glycosylation end products (RAGE), which transport Aβ in the opposite direction (into the brain from the blood) from LRP1 are increased in AD, further exacerbating Aβ accumulation (Yan et al. 1996; Deane et al. 2012). Another molecule that can expel Aβ from the brain is the xenobiotic transporter, Pgp, which is also down-regulated in AD (Cirrito et al. 2005). These molecular adaptations in the BBB during AD provide a challenge for therapeutic strategies as it becomes all but impossible to clear Aβ. However, inhibition of proinflammatory prostaglandins in myeloid and microglial cells can restore youthful, healthy immune functions and rescue cognition in aged mice, suggesting a potential therapeutic target for AD (Minhas et al. 2021).

Parkinson's Disease

BBB disruption is also present in Parkinson's disease (PD) (Gray and Woulfe 2015; Gerwien et al. 2016; Sweeney et al. 2018), the second-most prevalent neurodegenerative disorder. While AD is associated with accumulation of toxic Aβ and tau proteins to destroy neuronal synapses, PD pathophysiology centers around the degradation of dopaminergic neurons with the accumulation of α-synuclein. Via endocytosis from neurons, astrocytes accumulate extracellular α-synuclein aggregates, which cannot be cleared, causing cellular stress and death (Lee et al. 2010); astrocytic death leads to BBB defects including microbleeds with erythrocyte extravasation (Gray and Woulfe 2015).

Multiple Sclerosis

Other neurodegenerative diseases such as multiple sclerosis (MS), amyotrophic lateral sclerosis (Eubelen et al. 2018), and HIV-1-associated dementia pathophysiology are largely driven by leukocyte infiltration into the CNS through a disrupted BBB (Persidsky et al. 2006; Winkler et al. 2013; Gerwien et al. 2016). MS is an autoimmune disease where proinflammatory signaling compromises the BBB. Subsequent leukocyte transmigration, primarily in the white matter, causes demyelination and axonal loss (Ortiz et al. 2014). Once in the brain parenchyma, the T cells become activated from myelin sheath protein fragments and release proinflammatory and toxic mediators such as cytokines and free radicals (Hernández-Pedro et al. 2013). CD8⁺ T cells trafficking to the brain are dependent on Pgp; silencing of Pgp, significantly alters T-cell ability to move into the brain by decreasing the binding of T cells to ICAM (Kooij et al. 2014). Many current MS therapeutics dampen immune activation by either blocking release of proinflammatory factors or inhibiting leukocyte adherence to EC (Haarmann et al. 2015; Lamb 2020). Ozanimod, a sphingosine 1-phosphate receptor (S1PR) antagonist, was recently approved for treatment of MS; it induces S1PR internalization and prevents immune egress from lymph nodes (Subei and Cohen 2015; Lamb 2020).

PROSPECTS FOR CLINICAL THERAPEUTICS

Despite a current lack of universal BBB therapies, significant opportunities exist for pharmacologic modulation of the BBB. Approaches could either open the BBB to enable drug delivery, or enhance BBB function for reduction of inflammation, edema, or hemorrhage. One approach exploits receptor-mediated transcytosis (RMT) via covalent modification of an endogenous trafficking pathway, allowing receptor-mediated internalization of a drug-receptor complex into the brain parenchyma (Jones and Shusta 2007). Specifically, the TfR, which is highly expressed by brain ECs, has been targeted by anti-TfR antibodies possessing a carboxyl-terminus peptide linker to an enzyme or drug of interest. This allows internalization of a therapeutic via the lysosomal sorting pathway (Vanlandewijck et al. 2018; Kariolis et al. 2020) and has been used to treat Hunter syndrome, a brain-specific lysosomal storage disorder phenotype in mice (Ullman et al. 2020). The use of TfR-mediated RMT has potential as a therapeutic for other CNS diseases such as AD that do not exhibit decreased TfR expression during disease progression (Bourassa et al. 2019). For AD treatment, slight alterations of TfR binding motifs allow transport of covalently modified antibodies to the brain parenchyma instead of the lysosome (Niewoehner et al. 2014), allowing antibodies to cross the BBB into the brain to target CNS proteins.

Other approaches to selectively and transiently open the BBB include focused ultrasound with magnetic resonance monitoring. This method does not disrupt neuronal function and has been safely demonstrated in patients (Hynynen et al. 2001; Abrahao et al. 2019). Additionally, intracarotid injection of the hyperosmolar agent mannitol opens the BBB via osmotic effects (Wang et al. 2007). This method can again be monitored by magnetic resonance to fine tune and predict specific BBB openings and allow drug delivery to the brain (Chu et al. 2018). Systemic RNAi delivery targeting claudin-5 has also been reported to induce BBB opening (Campbell et al. 2008). While these novel techniques can increase drug delivery through the BBB, additional validation and targeting specificity is necessary.

Last, pharmacologic targeting of developmental pathways that regulate the BBB represents a viable strategy. Such proof-of-principle is represented by decreased cerebral edema upon VEGF inhibition in GBM clinical trials (Gerstner et al. 2009). Therapeutic manipulation of the WNT7AB/GPR124/RECK and NORRIN/TSPAN12/FZD4 receptor complexes, converging on β-catenin signaling, could allow either BBB or BRB opening or closure through a variety of strategies ranging from nonselective (lithium, GSK3β inhibitors) to global modulators of β-catenin signaling (tankyrase or porcupine antagonists) (Habib et al. 2020). WNT7A/WNT7B are poorly suited for direct use as therapeutics because of obligate palmitoylation, hydrophobicity, and rapid aggregation (Janda et al. 2012; Vallon et al. 2018). Similarly, NORRIN is poorly secreted as recombinant protein and is highly associated with ECM, thus limiting its action to short-range effects (Perez-Vilar and Hill 1997; Niehrs 2004). Newly described artificial bioengineered WNT surrogates that completely lack palmitoylation can cross-link FZD and LRP5/6 to initiate β-catenin signaling in vivo upon systemic infusion (Janda et al. 2017; Yan et al. 2017; Miao et al. 2020) and analogous FZD4-selective surrogates exhibit activity in BRB stabilization (Chidiac et al. 2021). Further, WNT7 mutants exhibit specificity for GPR124/RECK as opposed to generalized FZD activation and stabilizes the BBB in stroke and GBM models (Martin et al. 2022). However, this WNT7 fragment likely retains palmitoylation, which could compromise systemic bioavailability and indeed only local BBB delivery via adeno-associated virus has been demonstrated (Martin et al. 2022). Certainly, such paradigms could be extended to therapeutic manipulation of other developmental pathways including CCM loci.

CODA

Overall, a wealth of investigations has generated tremendous insight into the cellular composition of and molecular components underlying the formation and maintenance of the BBB. Clearly, the coordinated action of multiple cell types within the NVU cooperates with junctional and transporter elements, all governed by the activity of BBB-specific receptor complexes and their cognate ligands. The recent advent of single-cell technologies has further illuminated the cellular heterogeneity of the BBB, particularly within ECs. These studies, in aggregate, now provide substantial opportunities for mechanism-based intervention in the BBB.

However, in general, these mechanistic insights have not been translated to patient care. An urgent need yet remains for targeted agents that either promote or inhibit BBB function during pathophysiologic states. Indeed, many of the BBB diseases highlighted in this review are characterized by sequelae of prolonged or inappropriate BBB dysregulation. Conceivably, molecular pathways that have been functionally implicated in BBB regulation by in vivo genetic or in vitro cell culture studies offer routes to mechanism-based, selective promotion of BBB activity, or reciprocal options to promote barrier permeability. Such therapeutic potential highlights both the need for further mechanistic explorations of BBB regulation and for clinical translation into effective therapy.