Alterations of the blood-brain barrier during aging

Abstract

The blood-brain barrier (BBB) is a complex and dynamic interface that regulates the exchange of molecules and cells between the blood and the central nervous system. It undergoes structural and functional changes during aging, which may compromise its integrity and contribute to the pathogenesis of neurodegenerative diseases. In recent years, advances in microscopy and high-throughput bioinformatics have allowed a more in-depth investigation of the aging mechanisms of BBB. This review summarizes age-related alterations of the BBB structure and function from six perspectives: endothelial cells, astrocytes, pericytes, basement membrane, microglia and perivascular macrophages, and fibroblasts, ranging from the molecular level to the human multi-system level. These basic components are essential for the proper functioning of the BBB. Recent imaging methods of BBB were also reviewed. Elucidation of age-associated BBB changes may offer insights into BBB homeostasis and may provide effective therapeutic strategies to protect it during aging.

Introduction

The blood-brain barrier (BBB) is the complex and dynamic structure of the central nervous system (CNS) that protects against and responds to peripheral environmental changes. The concept of the BBB was proposed in the 19th century and has been a subject of intense research ever since. ¹ The BBB consists of a variety of elements and cells, ² as depicted in Figure 1(a). These cells interact closely to each other and provide structural and functional support to the BBB and modulate its permeability and response to stimuli. The BBB creates a distinct environment for the CNS that differs from the rest of the body. ³

Figure 1.

The structure of the BBB and how it interacts with other body systems. (a) The normal structure of the BBB. The image on the right shows a cross-section of the CNS capillaries. From the outside to the center, the BBB is composed of endothelial cells, pericytes, and endfeet of astrocytes. Microglia and perivascular macrophages are also members of the BBB, with the former located in the intercellular matrix and the latter in the Virchow-Robin space. The fibroblasts also exist in the Virchow-Robin space. The BBB has two layers of basement membrane (BM): that of the endothelial cells and that of the astrocyte endfeet and (b) the interaction of the CNS with other systems. Endothelial cells, as the most peripheral component of the BBB, interact most with other body systems. It is important to emphasize that the circulatory system has a two-way effect on the CNS. Blood flow not only affects the inflammatory response and endothelial cell structure, but is also regulated by endothelial cells and pericytes. The image is created with BioRender.com.

As the basic structure of CNS homeostasis, changes in the BBB are associated with aging, Alzheimer’s disease and dementia, and many other physiological and pathological processes. ³ There have been many conclusive studies on the alterations of the BBB in CNS diseases. In contrast, the relationship between the BBB alterations and normal aging has received less attention. Aging is a physiological process that occurs throughout the life cycle, and it is often the basis of disease, especially neurodegenerative diseases, including hippocampal sclerosis and frontotemporal dementia.^4,5 Much research has been done on the molecular mechanisms of aging, and some findings have been widely accepted, such as telomeres. ⁵ Similarly, a clear and effective marker of brain aging is necessary for the CNS. And as the predominant interface between the brain and peripheral systems, markers of BBB senescence would be the most reliable markers of an aging brain.

Aging is a process that can affect all body systems. In the cardiovascular system, senescence leads to vascular aging, resulting in weak blood vessels and hypertension. ⁶ Under normal physiological conditions, the CNS blood vessels can automatically adjust in response to changes in blood pressure.^2,7 During aging, the basic structures of vessels degenerates. This weakened buffering function of the vascular system would increase the shear stress induced by changes in blood pressure and, in turn, cause further damage to the BBB. ⁸ In addition, aging affects CNS function through the immunological and metabolic processes. Peripheral immune cells can enter the CNS via the glymphatic system or by adhering to and passing through the CNS endothelial cells. ⁹ These recruited immune cells play an important role in neuroinflammation. In addition, blood glucose levels influence the expression of glucose transporters (GLUTs) on endothelial cells and the rate of brain glucose uptake into the brain. ¹⁰ The gut-brain axis has been found to regulate the activation of microglia and perivascular macrophages, which represent the link between the CNS and the digestive system. ¹¹ Taken together, aging may lead to altered interactions between the CNS and other body systems as shown in Figure 1(b). The impact of other body systems is discussed in detail in the following sections.

Previous researchers have done sufficient study on the structure and function of the BBB. Most of the studies have focused on the BBB alterations in specific pathological conditions. However, relatively few studies have investigated the physiological age-related changes, especially for these newly discovered BBB components that lack of specific markers like fibroblasts and pericytes. Since aging is the risk factor of many CNS diseases, age-related changes of the BBB can be considered as subclinical or preclinical changes of diseases, which deserve deeper studies. We intend to provide an overview of the changes and interactions of each cellular component of the BBB during aging, the communication between the BBB and different body systems, and the advances in BBB imaging. Table 1 lists the major findings documented. These discoveries are explained in detail in the following sections and the potential direction of future studies on each component is also discussed.

Table 1.

Important findings of each component of the BBB.

Cell type	Age-related change of specific cell type	Main molecules	Specific change	Related BBB function		Reference
Endothelial cell	Surface receptors	GLUT1	↓	Metabolism	glucose	12 –14
		IGFR-1	↓		insulin	15
		Nrf2, FPN1	↑		iron	16
		KIR2.1, KCa	↓	CBF regulation		17 –19
	Tight junction	Claudin 5, ZO-1, occludin	↓	Barrier function		20 , 21
		VE-cadherin	↓	Neuroinflammation		22
	Surface receptors	ICAM-1, VCAM-1	↑			22 , 23
Astrocyte	Secreta	SASP	↑			24
		NF-κb	↑			25
		TGFβ	↑			26
		MEGF10	↑	neuroplasticity		27
	Surface receptors	Noradrenaline	↓	Metabolism	glucose	28,29
		IGFR-1	↓		insulin	29
		MCT4/1, PPARγ, CNTF	↑		lipid	30 –32
Pericyte	Contractility	–	↓	CBF regulation		33 , 34
	Secreta	LPS, CypA, IL-6	↑	Neuroinflammation		35 , 36
		IL-33	↓
	Surfece receptors	ICAM-1, CCL-2	↑			22 , 36 , 37
Microglia&PVM	Surface receptors	S100A8/9, CD14, CD64	↑			38 , 39
		TOLLIP	↓			40
Basement membrane		Collagen IV, nidogen	↑	Barrier function		41 –43

PVM: perivascular macrophages; GLUT: glucose transporter; IGFR: Insulin-like growth factor; Nrf: NF-E2-related factor; FPN: ferroportin; K_IR: inward-rectifier K⁺ channel; K_Ca: voltage-insensitive K⁺ channel; ZO-1: zonula occludens-1; ICAM: Intercellular adhesion molecule; VCAM: vascular cell adhesion molecule; SASP: senescence-associated secretory phenotype; TGFβ: transforming growth factor β; MEGF 10: multiple epidermal growth factor-like domains protein 10; MCT: monocarboxylate transporter; PPARγ: peroxisome proliferator-activated receptor gamma; CNTF: ciliary neurotrophic factor; LPS: lipopolysaccharide; CypA: cyclophilin A; IL: Interleukin; CCL: chemokine CC chemokine ligand; TOLLIP: toll-interacting protein; CBF: cerebral blood flow.

Endothelial cells: the interface function failure

During vascular aging, the altered structure of the BBB cannot buffer shear stress and studies have focused on the influence on endothelial cells. ³ Endothelial cells are the major components of the BBB, which has both barrier and transcytosis functions. In young and normal brain, central endothelial cells (CECs) are connected by a variety of tight junctions, including the claudin family, the ZO family, and occludin, which is the most obvious difference between CECs and peripheral endothelial cells. ⁴⁴ These tight junctions play an important role in the physics of the BBB. In addition, endothelial cells have different modes of molecular transport. In this section, we briefly summarize the contribution of endothelial cells to senescence, including disruption of tight junctions, the impairment of transcytosis function, and other inflammatory or metabolic changes of endothelial cells in the aging brain.

The disruption of tight junctions

The loss of tight junctions (TJs) may increase paracellular transport and lead to uncontrolled molecular exchange between the CNS and the peripheral blood. This process is also known as “BBB permeability/leakage”. ³ Studies have confirmed the reduction of several types of TJs, mainly including claudin-5, occludin and ZO-1.^45,46 The BBB leakage as detected by dynamic contrast-enhanced MRI (DCEM) has been shown to be able to predict brain aging and cognitive decline. ⁴⁷

Many factors in the aging brain have been shown to be associated with TJ degradation. Sirtuin-1 (Sirt1), an NAD-dependent deacetylase belonging to the sirtuin family, has been found to be decreased in aging mice with reduced claudin-5 and ZO-1. It’s overexpression may delay aging.^20,21 MiR-195 has also been shown to improve TJ clearance through the autophagy-lysosome pathway. ⁴⁸ Angiotensin-II (Ang-II), a well-known risk factor for hypertension and CNS disease, may also influence ZO-1 and VE-cadherin expression and is downstream of PPAR alpha-related pathways. ⁴⁹ In addition, the inflammatory milieu, such as increased levels of matrix metalloproteinase (MMP)-2/9 and decreased levels of their inhibitors (tissue inhibitors of metalloproteinase or TIMPs), has been found to contribute to TJ degradation. ⁴⁷

Despite these findings, the exact process of TJ degradation during aging remains unclear. Due to the high predictive value of the BBB leakage in aging, it is important to elucidate the pathway of TJ degradation in order to identify novel targets for intervention in aging.

The damage to transcytosis function

As the most superficial structure of the BBB, the endothelial cell is the crucial regulator of molecular exchange between the brain parenchyma and the peripheral blood. Different molecules cross the BBB in different ways, some through receptors and others through endocytosis. If transcytosis function is impaired, this will affect the flow of molecules in and out, leading to unintended leakage of neurotoxic factors from the periphery and deposition in the CNS. ³ During aging, many receptors and channels have been found to change structurally or functionally, leading to various types of CEC dysfunction, including reduced nutrient uptake, loss of control over cerebral blood flow, and disruption of cell signaling.

The brain consumes 20–25% of the total body’s energy and relies primarily on glucose taken up by glucose transporters (GLUTs) across the BBB. ⁵⁰ There are 14 types of GLUTs in the human body, and GLUT1, encoded by SLC2a1, is the dominant one expressed by CEC.^50,51 Several studies have found that GLUT1 deficiency is associated with cognitive impairment in both humans and mice.^12,13 There was a clear association between GLUT1 reduction and Alzheimer’s disease as demonstrated by FDG-PET in individuals who already had or were at high risk of developing Alzheimer’s disease. ¹⁴ Although GLUT1 reduction has been shown to be associated with various pathological conditions, direct evidence of GLUT1 reduction within a physiologically aging brain is lacking. Whether GLUT1 is reduced in aging CECs, and whether it can be an effective target for predicting aging remains to be further investigated.

In recent years, CECs have been implicated in the control of cerebral blood flow through endothelium-derived hyperpolarization (EDH), which is controlled by Gq protein-coupled receptors (GqPCRs), small- and intermediate-Ca2⁺-activated K⁺ (SK_Ca/IK_Ca) channels and several types of K⁺ channels. ⁵² Longden’s study found that the endothelial cell inward-rectifier K⁺ (K_IR2.1) can respond to neurogenic extracellular K⁺ by causing capillary and arteriolar dilation and increasing blood flow, while K_IR2.1 are maintained by the plasma membrane phospholipid (PIP2). ⁵³ Many investigators have demonstrated age-related changes in EDH, focusing on different changes in different channels. In aged mice, some have observed K_Ca activation with increased current loss from CECs, ¹⁷ while others have found decreased KIR function. ¹⁸ The complexity of ion channels involved in EDH makes it difficult to fully identify the exact relationship between aging and CEC-driven changes in blood flow. ¹⁹ Whether new CEC channels are involved in EDH, and how they change specifically during aging, remains to be further investigated.

In addition to the receptors mentioned above, many different types of transcytosis processes have been found to be altered during aging. Acid sphingomyelinase (ASM) has been shown to increase caveolae-mediated transcytosis and cause damage to the caveolar cytoskeleton in endothelial cells. ⁵⁴ Circulating levels of insulin like growth factor receptor-1 (IGFR-1) have been found to decrease with age. Endothelial-specific knockout of IGFR-1 mimics the aging phenotype and inhibits NO-dependent telangiectasis. ¹⁵ In addition, the knockout of the transcription factor NF-E2-related factor 2 (Nrf2) has a protective effect on motor coordination in aged mice by reducing ferroportin 1 (FPN1) levels in CECs. ¹⁶

The transcytosis function of CECs makes them the gatekeepers of the BBB. Figure 2 shows the major changes in transcytosis function in aging CECs. Similarly, a precise understanding of how endothelial receptors change during aging can further our understanding of other age-related process changes in endothelial cells, such as carbohydrate metabolism, iron homeostasis, inflammation and oxidative stress. Besides, these significantly altered receptors may be potential targets for slowing aging.

Figure 2.

Changes in the transcytosis function of CECs in an aging blood-brain barrier. Changes in glucose uptake lead to neurological starvation and the production of more ROS. K⁺ channels may further influence vasoconstriction. The stability of the vesicle structure and the regulation of molecular receptors are closely related to CEC-induced neuroinflammation. “↑” indicates up-regulated, “↓” indicates down-regulated. The image is created with BioRender.com.

The endothelium-related vascular inflammation

Recently, vascular inflammation has been implicated in age-related cognitive impairment. Although CNS neuroinflammation is more closely associated with astrocytes and microglia, CECs play an important role in the recruitment of peripheral leukocytes. Peripheral leukocytes require several steps to adhere to and cross the BBB, each of which relies on different cell adhesion molecules and tight junctions. ²² During aging, many adhesion molecules are up-regulated. Intercellular adhesion molecule-1 (ICAM-1), a key component of T-cell migration across CECs, is up-regulated in aging mice, leading to increased T-cell invasion.^22,23 In addition, vascular cell adhesion molecule 1 (VCAM1), a protein that promotes vascular-immune cell interaction, is locally up-regulated in CECs during aging. ⁵⁵ In addition to adhesion molecules, CEC adherent junctions, which control the transendothelial migration of peripheral leukocytes, also show changes in expression during aging. The degradation of VE-cadherin is a major cause of uncontrolled leukocyte invasion in an aging brain. ²² Many factors can influence the amount, stability and phosphorylation of VE-cadherin. MMP-9 and the phosphorylation of p120 at S879 have been found to stimulate VE-cadherin degradation, ⁵⁶ whereas VE-PTP and integrin activation may protect VE-cadherin from destruction. ⁵⁷ These discoveries may become potential intervention target of VE-cadherin related inflammatory cells invasion, and the same idea can be used to avoid degradation of other adhesion molecules of the endothelial cells to ensure the first line of defense against peripheral leukocytes.

In addition to leukocyte recruitment, CECs also participate in vascular inflammation through interactions with other inflammatory cells and pathways, particularly the complement pathway. CECs have been shown to produce a variety of complement proteins, including C1, C4 and C3. ⁵⁸ Propson et al. have found that the C3a receptor on CECs is part of the complement pathway that inhibits VCAM-1 expression, increases microglial reactivity, and accelerates neurodegeneration during senescence. ⁵⁹ Elahi et al. have found an increase in multiple complement components in endothelial-derived exosomes (EDE) in the elderly. ⁶⁰ Miguel et al. intravenously injected clusterin, a complement cascade inhibitor, into the mouse model of acute brain inflammation and found that it bound to CECs and reduced neuroinflammatory gene expression. ⁶¹

As the first barrier of the BBB, CECs play an important role in neurovascular decoupling during aging. In addition to its own contribution, the CEC is extensively intertwined with other components of the BBB. Therefore, studying how and to what extent CECs interact with astrocytes, microglia and pericytes is also important for a comprehensive understanding of age-related changes in the BBB.

Astrocytes: the impairment of neurovascular unit homeostasis

Located between neurons and blood vessels, astrocytes cover nearly 100% of the microvascular wall and play an essential role in maintaining neuronal homeostasis.^2,62 Although astrocytes cannot generate action potentials like neurons, they can respond to environmental changes through Ca²⁺ signaling and mediate communication between neurons and capillaries. ⁶³ Astrocytes also control the flow of cerebrospinal fluid, which is referred to as “glymphatic function”. ⁶⁴ The hippocampus is a relatively active area of the brain that requires an abundant supply of blood and nutrients, as well as a rigorous immune system. Being involved in all of these critical functions, the astrocyte is a major factor in both physiological and pathological processes. Here, we will briefly review the changes in astrocytes during aging.

The effect of aging on the astrocyte itself

With aging, astrocytes undergo two major changes: (a) cellular senescence: astrocytes in an aging brain exhibit the phenotype of aging cells. (b) A1-like astrocyte activity: increases in two astrocyte-specific proteins that represent astrocyte activation, GFAP and vimentin, converting more astrocytes into neuroinflammatory A1-like astrocytes. Different from the neuronal loss, the number of astrocytes remains nearly constant in the aging brain. ²⁴ The major effect of aging on astrocytes occurs in cell function.

Aging astrocytes undergo changes in both cell morphology and gene expression patterns. Astrocytes perform their interface functions primarily through their endfeet, which are swollen structures at the end of the cytoplasm that coats blood vessels. As astrocytes age, both their area and the number of endfeet decrease, resulting in a reduction of the peri-endfoot area. ²⁴ This astrocyte “shrinkage” consequently reduces the astrocyte coverage of blood vessels as well as AQP-4 expression on the cell surface, which to some extent disrupts the neurovascular unit (NVU) barrier. ⁶⁵ In addition to morphological changes, aging astrocytes exhibit replicative senescence, including increased expression of growth-arresting factors (p16 and p53),^24,66 decreased expression of lamin-B1, ⁶⁷ and overproduction of ROS. ⁶⁸ Another important feature of senescent astrocytes is the senescence-associated secretory phenotype (SASP), which represents the secretion of many types of inflammatory factors, such as interleukin 6 (IL-6), CCL2, MMP-3 and MMP-9. Based on the importance of inflammation in aging, SASP is a potential marker of astrocyte senescence. ²⁴

Neuroinflammatory effect of astrocytes in aging brains

As mentioned above, astrocytes in aging brains may undergo reactive hyperplasia and transform into A1-like reactive astrocytes characterized by increased GFAP expression.^24,69 Such astrocytes are highly neurotoxic, making them a major killer of neurons during aging. Three major pathways contribute to the neuroinflammatory process of A1-like reactive astrocytes: the NF-κb pathway, the TGFβ pathway, and the complement pathway. The NF-κb pathway is widely expressed in different cells and is extensively involved in the stress response. In A1-like astrocytes, it stimulates the secretion of IL-6, MMP-3 and other inflammatory factors, which have a negative effect not only on neurons but also on pericytes and endothelial cells. Besides, NF-κb regulates SASP in normally aging astrocytes. ²⁵ When astrocytes interface with the BBB destruction, TGFβ has been found to be over-activated accompanied by up-regulation of inflammatory factors, resulting in impaired synaptic connection within both surrounding cells and astrocytes. ²⁶ There is evidence that TGFβ is also involved in the up-regulation of astrocyte inflammatory cytokines through microglial activation. ⁷⁰ In A1-like astrocytes, several complements, such as C3 and C4B, are over-activated and play an important role in neurotoxicity and microglial stimulation. ²⁵

The exact processes of astrocyte alteration during aging are not yet clear. Much work has been done to inhibit one of these key cytokines and to reverse A1-like astrocytes, but few have effectively stopped or slowed aging.^71,72 This may be because there are many alternatives to the silenced pathway. Given that astrocyte-driven neuroinflammatory pathways are indeed one of the mechanisms of aging, future research could focus more on how these pathways interact and compensate with each other.

The metabolic plasticity of astrocytes in aging brains

With multiple receptors on the cell surface, astrocytes also control nutrient uptake and multiple metabolic pathways of the NVU. The metabolic function of astrocytes is affected by many factors including aging. Nutrient supply and metabolic excitability of the brain decrease with age, in large part due to a decrease in astrocyte surface receptors. Here, we review the metabolic plasticity of astrocytes from three main perspectives.

The astrocyte-neuronal L-lactate shuttle (ANLS) refers to neuronal ATP production with astrocyte-derived L-lactate, which is known to be significantly related to cognitive function. ⁷³ Due to the limitations of neuronal glycolysis, astrocytes help produce ATP and L-lactate, and then deliver them to neurons for energy. ²⁸ This process can be stimulated by noradrenaline (NA) released by neurons,^28,29 and several factors have been discovered to up-regulate ANLS, such as the vasoactive intestinal peptide (VIP) pathway and the amyloid beta (Aβ) 42-α7 nicotinic acetylcholine receptor interaction.^74,75 In the aging brain, NA levels decline, inhibiting astrocyte glycolysis and subsequently reducing ATP supply to neurons.^28,29

In addition to glycolysis, the nutrient supply of neurons also depends on glycogen storage in astrocytes. In astrocytes, some of the glycogen taken up from the blood by GLUT1 and GLUT4 is sent for gluconeogenesis, which is known as the “glycogen shunt pathway”. ⁷⁴ This process can be regulated by insulin, which has been shown to be able to cross the BBB via IGFR-1. ²⁹ With aging, the amount of IGFR-1 on the surface of astrocytes decreases, leading to the inhibition of insulin uptake and a lower rate of astrocyte gluconeogenesis. Studies have shown that decreased IGFR-1 is associated with increased ROS, decreased astrocyte autophagy, decreased blood perfusion, and decreased cognitive function. ⁷⁵ Whether insulin influences glucose uptake remains unclear. Some studies have used particular approaches and have shown that insulin increases glycogen levels locally within astrocytes without changing integrative glucose flux. ⁷⁶

Astrocytes are also involved in the lipid metabolism of the NVU. Although lipids are not the major source of energy in the brain, they affect not only nutrient supply but also inflammatory pathways, particularly LPS. ²⁹ A number of factors have been shown to be associated with changes in lipid uptake and metabolism in astrocytes during aging, including monocarboxylate transporter 4 (MCT4), MCT1, peroxisome proliferator-activated receptor gamma (PPARγ), and ciliary neurotrophic factor (CNTF).^{30 –32} The relationship between astrocytes and metabolism may extend to more complex disease states, such as diabetes and obesity. Some studies have focused on the relationship between aerobic exercise and aging, ⁷⁷ while others have focused on that between different diets and aging, such as high-carbohydrate diets and caloric restriction. ⁷⁸

Most of these studies show specific associations between metabolism and senescence. Since metabolic states are relatively easy to manipulate, more work is needed to elucidate the molecular and cellular mechanisms of these associations in order to find more effective approaches to delaying senescence.

The control of synaptic plasticity

In the adult brain, neuronal networks are constantly remodeling and new synapses are constantly being produced to meet the needs of continuous learning and memory, which is referred to as “neuroplasticity”. ⁷⁹ Glial cells are responsible for pruning newly formed synapses during the processes of synaptic production, maintenance and degradation. Astrocytes cover more than 50% of CNS synapses, making them the BBB component that is most closely related to synaptic plasticity.^2,79

Synaptic dysfunction is thought to be an important process leading to cognitive decline during aging. In terms of synapse formation, astrocytes secrete thrombospondin, glypican4/6 and hevin to interact with their receptors and stimulate synapse formation.^80,81 Ephrin-B is also an important astrocytic factor that controls synaptic remodeling in the hippocampus. ⁸² In terms of synaptic degradation, astrocytes rely primarily on the MEGF10/MERTK pathway to destroy mistakenly-formed synapses. ²⁷ With these changes, the number of correctly connected synapses decreases, leading to cognitive impairment in the elderly. Compared to microglia, the contribution of astrocytes to neuroplasticity is relatively understudied, but considering of the direct and persistent connection between astrocyte and synapse, the astrocyte may become a potential intervention target to increase correct synapses in an aging brain.

Pericytes: the abnormality in cerebral blood flow and BBB integrity

Pericytes are mural cells located between CECs and the basement membrane, which cover 30% of brain microvessels.^2,83 With the discovery of unique pericyte markers, more functions of pericytes in the NVU have been found. The main role of pericytes is to control the microvascular diameter and cerebral blood flow in response to environmental changes. On the other hand, pericytes secrete many kinds of inflammatory cytokines and assist in the trafficking of inflammatory cells, and they also help maintain the stability of the BBB structure. Here, we briefly summarize the role of pericytes in the aging process.

The pericyte-related reduction of cerebral blood flow

Adequate blood perfusion is essential for neurons to receive sufficient oxygen and nutrients. With a deeper insight into pericytes, more and more studies have demonstrated that the reduction of microcirculation in aging brains is closely associated with cognitive impairment, and that pericytes contribute to this reduction. Using optogenetic stimulation, pericytes were found to exert a substantial but slow influence on blood flow, and the slow vasoconstriction was inhibited by the clinically-used vasodilator fasudil, a Rho-kinase inhibitor that blocks the contractile machinery. ⁸⁴ In ischemic stroke, pericytes cause the no-reflow phenomenon, which refers to the reactive contraction of capillaries even after the blood flow of larger vessels has already been restored. ⁸⁵ Similarly, during physiological aging, pericytes cause alterations in capillary diameter and decrease CBF, leading to chronic neuronal ischemia and hypoxia. ³³

Most of the studies in this area are cause-and-effect studies without a precise molecular mechanism. To prove the contribution of pericytes to the control of CBF, researchers specifically knocked out pericytes and observed the change in CBF in old mice. ³⁴ This revealed more information about pericytes. Compared to young mice, the remodeling and self-repairing function of pericytes in old mice is impaired, resulting in prolonged flow heterogeneity in older brains. ³⁴

Despite these positive findings, whether pericytes are contractile remains controversial. Some studies suggest that smooth muscle cells (SMCs) rather than pericytes play a key role in CBF regulation. ⁸⁶ We believe that the reason for this debate is mainly due to the difficulty in precisely defining the cell type, since the pericytes located on arterioles and venules exhibit both SMC and capillary pericyte phenotype. ⁸⁷ In the future, more work is needed on both of these aspects. First, more effective markers that separate SMCs from pericytes remain to be discovered, which may help us find the more crucial controller of the cerebral microcirculation. Second, it is important to figure out the molecular mechanisms of how pericytes control CBF and how these mechanisms change during aging.

The initiator of neuroinflammation

Pericytes can secrete several types of pro- and anti-inflammatory cytokines, allowing them to first sense peripheral inflammatory changes and then send signals to other cells of the BBB, including neurons. ⁸³ Pericytes also play an important role in age-related neuroinflammation. Pericytes sense systemic inflammatory factors, including lipopolysaccharide (LPS), cyclophilin A (CypA), and TNF-α, and secrete various cytokines through specific pathways. Among these, matrix metalloprotease 9, chemokine CC chemokine ligand 2 (CCL2), and IL-6 are pro-inflammatory factors, whereas C-X3-C motif chemokine ligand 1 (CX3CL1) and IL-33 are anti-inflammatory factors.^35,36 During aging, more pro-inflammatory factors are secreted, leading to the breakdown of the basement membrane and endothelial tight junctions and the activation of astrocytes and microglia. ⁸³

In addition to transmitting inflammatory factors, pericytes also participate in peripheral leukocyte invasion. Pericytes express ICAM-1 and cooperate with endothelial cells in leukocyte migration. ³⁷ Other studies have shown that pericytes interact with monocytes via CCL2 and with neutrophils via CXCL8.^22,36 However, the contribution of pericytes to the BBB barrier function is not as strong as that of CECs due to the gap between pericytes.

As the initial sensor and promoter of CNS neuroinflammation, pericytes may be a potential target for inflammation research. In addition, if pericytes are considered as a breakthrough, it is important to find out how pericytes crosstalk with other components of the NVU.

The participant of the maintenance of the BBB integrity

Besides regulating CBF and inflammation, pericytes also actively regulate the BBB integrity. Early studies have proved that the absolute pericyte coverage can determine relative vascular permeability, even if the pericytes cover only 30% of the capillary.^83,88 During rat CNS development, the pericytes are recruited to and interact with the endothelial cells earlier than the astrocytes and oligodendrocytes, and contribute to the formation of endothelial tight junctions and vesicle trafficking. ⁸⁹ During natural aging, the pericyte loss and the BBB disruption have been observed cocurrently, but there has been little direct evidence indicating the causal relationship between these two observations, so it remains intriguing to what extent do pericytes are involved in age-dependent BBB leakage.

Recent studies have reported the heterogeneity of the pericyte contribution to BBB integrity both in different brain regions and in the response of adjacent cells. Some have found that in the midbrain, interbrain, basal forebrain, and pons, the pericyte counts are higher and the BBB permeability is lower compared to that in the cortex, hippocampus and striatum within the same PDGF-B knockout mouse.^90,91 A single-cell analysis focused on the endothelial cell response to pericyte loss and found that most endothelial cells exhibited a venous-shifted molecular pattern, while those at the hotspot leakage sites exhibited a arteriolar-shifted molecular pattern. ⁹² This newly discovered complexity may possibly be due to the fact that pericytes do not influence the BBB integrity directly, but through other BBB components especially endothelial cells, inspiring researchers to focus more on the communication and interaction between different BBB components.

Basement membrane: the internal interface of the endothelial cells

Lying between the endothelial cells and the endfeet of astrocytes, the basement membrane (BM) is one of the major components that contributes to structural support, cell anchoring and signal transduction. ⁹³ The BM is composed of multiple highly organized extracellular matrix proteins including collagen IV, laminins, nidogens, heparin sulfate proteoglycan, and fibronectin. ⁹⁴ Many researchers have examined each of the four components of the BM using loss-of-function studies. Most global knockout mice exhibited embryonic lethality, and some conditional knockout mice (including Tie2, PDGFRβ and SM22α, etc.) showed BBB breakdown and cerebral hemorrhage.^41,93

During normal human aging, most observations noted BM thickening. The chemical characteristic change of the BM remained intriguing and varied for different components and models. ⁴¹ For example, aged mice showed decreased collagen IV, nidogens and fibronectin, whereas aged humans showed increased collagen IV. Besides, recent studies showed BM thickening in old mice due to the deposition of lipid droplets and aggregates, revealing the possible function of the BM to store metabolites, which may also provide inspiration to Alzheimer’s Disease. ⁴² Recently, some researchers started to explore the molecular mechanism of how the changes of the BM cause BBB disruption. Some have found that fibrinogen in mouse cerebral microvessels induces BBB dysregulation with aging via a dynamin-related protein 1 (DRP1)-dependent pathway. ⁴³ Others have found that some molecule produced by other components of the BBB can influence the integrity of the BM: the endothelial Atg7 downregulates the expression of fibronectin, ⁹⁴ while ApoE4 impairs pericyte-mediated basement membrane formation. ⁹⁵

In conclusion, the BM is necessary to the angiogenesis during the embryonic period, and it exhibits age-dependent changes that can influence the stability of the BBB. The molecular mechanism of this influence is still remains unclear and requires further study.

Microglia and perivascular macrophages: the overactivation of the CNS immune system

Microglia and perivascular macrophages (PVMs) play an important role in the immune system of the CNS. In a normal brain, these cells move around the neuronal extracellular matrix or the Virchow-Robin space, clearing up dead cells and metabolic waste. ² In an aging brain, however, their function is overactivated, leading to the activation of neuroinflammatory pathways. Many studies have proved the upregulation of several types of inflammatory factors, inflammasomes and inflammatory cell receptors during aging.

The alteration of inflammatory-related receptors

Surface receptors are important for immune cells to sense and respond to environmental changes. During aging, both microglia and PVM appear to undergo an increase in different kinds of surface receptors, which is known to be associated with the chronic activation of these immune cells, leading to the secretion of neuroinflammatory factors, including IL-6, IL-1β and tumor necrosis factor (TNF-α). ⁹⁶

One of the most significant increases is in the toll-like receptors (TLRs). TLR is a pattern-recognition receptor located on M1 microglia that acts as a receptor for LPS and an initiator of NF-κB inflammatory pathways in PVMs. ⁹⁷ In aging brains, upregulation of TLR expression parallels changes in the expression of many other factors. Myeloid-related protein (MRP) 8 and 14, also known as S100A8 and S100A9, are members of TLR ligands that have been extensively studied in neurodegenerative disorders or cancer. ³⁸ In aging brains, these inflammatory factors are observed to be upregulated and are strongly associated with increases in TRL. ³⁸ Besides, CD14, which is expressed by monocytes, has also been found to be a co-activator of TRLs. ³⁹ In parallel with these up-regulated factors, the toll-interacting protein (TOLLIP), one of the inhibitors of TLRs, was significantly downregulated at multiple sites in the brain. ⁴⁰

The expression of other inflammatory receptors has been found to be regulated in activated microglia and PVMs. In microglia, the expression of CD64, CD32, CD16b, CD11b and FCER1a is upregulated. ⁹⁸ In PVM, fractalkine, the ligand of CX3CR1 is downregulated, while major histocompatibility complex II (MHC II) and Trem-2 receptors which are responsible for antigen presentation and monocyte recruitment are upregulated. ⁴⁰ If the related pathways of these differently expressed receptors can be further elucidated in future studies, these receptors may become potential targets for delaying and interfering with senescence.

The influence of intestinal flora on microglia and PVMs

Many studies have shown that changes in the gut flora influence the activation and infiltration of both microglia and PVMs to some extent. The bidirectional communication between the CNS and the gut is called the gut-brain axis. ¹¹ Relying on the well-controlled exchange of substances between the CNS and the peripheral blood provided by the BBB, metabolites released by the gut flora can cross through the barrier and act on the NVU. The microglial transcriptome of germ-free mice shows a lower expression of inflammatory factors than normal mice. ⁹⁹

A variety of gut flora metabolites, including short-chain fatty acids (SCFAs), choline-derived trimethylamine N-oxide (TMAO), and bile acids, have been shown to interact with microglia and PVM. SCFAs may inhibit the expression of inflammatory factors and protect aging neurons. TMAO, which is secreted at higher levels in aging individuals, has controversial effects on neurons and microglia. ¹⁰⁰ Some studies have observed increased neuroinflammation under TMAO treatment, ¹⁰¹ while other studies have confirmed a protective effect of chronic TMAO exposure. ¹⁰² In recent years, more research has been conducted on the TMAO precursor δ-valerobetaine, which may help to further elucidate the relationship between TMAO and aging. ¹⁰⁰

In summary, changes in the gut flora during aging may have an impact on neuroinflammation and BBB integrity through the gut-brain axis. Since the gut responds directly to diet, if the exact pathways of the gut-brain axis are elucidated, it may become possible to intervene in aging through diet, which may be one of the easiest ways to delay aging, possibly with the fewest side effects.

Fibroblasts: Scar formation

With the recent use of precise single-cell and single-nucleus sequencing, CNS fibroblasts have gained attention as a new component of the BBB, which is mainly located in the meninges and perivascular spaces. ¹⁰³ Recent studies have identified several specific markers for CNS fibroblasts, including PDGFRα, COL1A1 and ERTR7.^87,103,104 In a healthy brain, some studies consider CNS fibroblasts to be part of the support of the BBB structure, while others suggest that CNS fibroblasts are involved in the glymphatic system.^103,105 Although few studies have mentioned age-related changes in CNS fibroblasts, their main function in several pathological conditions (multiple sclerosis, stroke, and etc.) is in common: scar formation after CNS injury or neuroinflammation.

The scar that commonly formes after CNS injury, consists of two major components: the astroglial scar and the fibrotic extracellular matrix (ECM). ¹⁰⁶ The former is formed by resident astrocytes, while the latter is formed by inflammatory cells in the lesion core and the perivascular fibroblasts together.^103,106 When activated by inflammatory factors, fibroblasts produce large amounts of fibronectin, collagen and laminin, forming a dense barrier around the lesion. ¹⁰⁷ This stimulated barrier is a double-edged sword. On the one hand, it prevents the spread of inflammation, thereby reducing the extent of the damage. On the other hand, it is a major obstacle to neural and vascular regeneration and blood perfusion after injury.^106,107 Most studies generally agree that scar-forming glial cells and perivascular fibroblasts are potential targets for injury repair and disease intervention. ¹⁰³

The relationship between CNS fibroblast function and aging has received little attention. In an AD brain, some components of the ECM, such as proteoglycans, have been found to aggregate around cells along with the Aβ protein. ¹⁰⁸ In an aging brain, CNS fibroblasts may contribute to the aging process by altering the composition of the ECM. In addition, there is some evidence that CNS fibroblasts can crosstalk with other cells, particularly CNS immune cells. ¹⁰³ As a very newly discovered element of the BBB, whether CNS fibroblasts can be used as targets for senescence intervention still requires further investigation.

Advances in imaging of BBB

With the advancement of the progression in the medical imaging technology, the BBB has gradually changed from a microcosmic structure of basic research to a visible structure that can be applied to clinical practice. Although the specific structure of the BBB cannot be seen, the widely used imaging method, including MRI and CT, can reflect many kinds of function of the BBB. The dynamic contrast-enhanced (DCE) or perfusion imaging protocols on MRI or CT scanners can quantify the BBB permeability, by injecting contrast agents such as gadolinium chelates and iodinated contrast media before imaging.^109,110 More recently, the development of arterial spin labeling techniques used in MRI has provided the ability to non-invasively measure the BBB permeability to water and has been used in several CNS diseases including CADASIL and neurodegenerative diseases.^111,112 Delayed CSF enhancement on post-contrast FLAIR images can also quantify BBB leakage. ¹¹³ Neuroinflammation can also be visualized by tracing various surface receptors using labeled ligands, including tracing translocator protein (TSPO) based on PET-CT or tracing ultrasmall particles of iron oxide (USPIOs) based on MRI to see the activation of microglia, macrophages and some inflammatory factors such as MMP, with the latter being more accurate but more unstable,^114,115 and tracing activated endothelial cells by conjugating micron-sized particles of iron oxide (MPIOs) with specific monoclonal antibodies. ¹¹⁶ Some researchers have used the injection of contrast agent into the subarachnoid space of mice to see the CSF circulation function of astrocytes, but it has not yet been practiced on patients. ¹¹⁷ In summary, compared with the BBB permeability imaging, which is more mature and widely used, the clinical imaging of neuroinflammation still needs more experiments and data to achieve higher sensitivity, specificity and safety to better help diagnosis of various types of CNS diseases.

Conclusion

In summary, this review focused on the impact of aging on the key components of the BBB, which regulates the exchange of substances between blood and CNS. The review discussed how aging affects the BBB structure and function, as well as the NVU interactions, in relation to synaptic plasticity, neurotransmission, cerebral metabolism and perfusion, and neuroinflammation. Figure 3 shows the main molecular changes of the BBB during aging.

Figure 3.

Neuroinflammatory processes in the aging brain. Neuroinflammation is an important process that occurs during aging. Members of the neurovascular unit interact and cooperate with each other, ultimately leading to neuronal destruction and the aging phenotype of the brain. The image is created with BioRender.com.

Previous researchers have done sufficient studies on the structure and function of the BBB. More and more studies started to focus on the communication between different components within the BBB and between different body systems, and more advanced imaging methods and biological information analysis methods have provided more comprehensive information about those communications. However, most of the studies have focused on the BBB alterations in specific pathological conditions. Especially for some newly discovered components that lack of specific markers (such as fibroblasts and pericytes), relatively fewer studies have focused on the physiological age-related changes. Since aging is the risk factor for many CNS diseases, some age-related changes of the BBB can be considered as subclinical or preclinical changes of diseases, which deserve deeper studies. In our perspective, given the complexity of the BBB, further research on the BBB should address: (1) The heterogeneity and delicate function of different BBB components by defining them with more specific markers and involving pathways. (2) The crosstalk between components and how signals are transmitted between the peripheral system and the BBB. (3) Realistic in vitro models that well simulate the structure and cellular environment of the BBB, which can facilitate accurate investigation of the BBB changes. (4) High-precision, safe and real-time in vivo imaging methods to observe the dynamic changes of the BBB. Undoubtedly, rapid progress in molecular, bioinformatic and imaging approaches in future would facilitate the discovery of dominant and comprehensive changes of the BBB in the future, thus providing effective therapeutic strategies to protect it during aging.