In the arterioles, SMCs continuously enwrap the abluminal side of endothelial cell layer and make myoendothelial gap junction (Aydin et al., 1991). The SMCs in the arterioles have an inconspicuous soma and extend broad processes (Attwell et al., 2016) that strongly express αSMA transgelin, desmin, CD146, and CNN1 as well as shared mural cell markers including NG2, PDGFRβ, CD13, vimentin, and RGS5. Outside of the SMCs is perivascular space with fibroblast-like cells, surrounded by collagen layer and endfeet of astrocytes (Mastorakos and McGavern, 2019).

In the capillaries, the vessel size further decreases, and the endothelial layer is intermittently surrounded by pericytes. Compared to the peripheral vascular beds, CNS capillaries have higher pericyte-to-endothelial cell ratios (1:1 to 1:3) and around 70–80% of the capillary surface area is covered with pericyte cell processes (Bell et al., 2010; Winkler et al., 2010, 2014). Capillary pericytes have a conspicuous protruding ovoid cell body with long thin processes that course along the capillary for longer distances and are embedded within the basement membrane.

The role of capillary pericytes in CBF control has long been debated. Using mouse models, some researchers have reported that arteriolar SMCs but not capillary pericytes regulate CBF in response to neuronal activities or ischemic stress (Fernández-Klett et al., 2010; Hill et al., 2015). Others, however, have shown that capillary pericytes also change the vessel diameter and CBF by stimuli, neuronal activation, or ischemia (Peppiatt et al., 2006; Yemisci et al., 2009; Hall et al., 2014; Pieper et al., 2014; Kisler et al., 2017b; Rungta et al., 2018), suggesting capillary pericytes also contribute to CBF regulation.

Because pericytes are morphologically and functionally heterogeneous, pericytes are sometimes subclassified according to their topology, morphology, and the protein expression levels.

In the post-capillary venules, different shaped mesh pericytes, also called stellate/stellate-shaped pericytes, surround the endothelial layer. These cells have many slender and shorter branching processes than capillary pericytes (Hashitani and Lang, 2016; Yang et al., 2017; Arango-Lievano et al., 2018). They express αSMA, ABCC9, cysteine sulfinic acid decarboxylase (P-selectin), and endomucin (Marín-Padilla, 2012). The expression level of αSMA is lower than the mural cells in the arterioles and proximal capillaries (Grant et al., 2019). The pericytes in the post-capillary venules are thought to regulate immune cell entry to the brain parenchyma like those in other tissues (Proebstl et al., 2012; Stark et al., 2013; Attwell et al., 2016; Dalkara, 2019; Rudziak et al., 2019). Outside of the pericytes are astroglial end-feet forming glia limitans. Between the endothelial basement membrane and astrocytic basement membrane is perivascular space, where antigen presenting cells reside (Engelhardt et al., 2017; Mastorakos and McGavern, 2019). Fibroblast-like cells are within the astrocytic basement membrane (Mastorakos and McGavern, 2019).

Post-capillary venules are collected to form ascending venules. In the venules, the endothelial cell layer is surrounded by stellate-shaped SMCs with broad leaf-like processes (Ushiwata and Ushiki, 1990; Armulik et al., 2011). Venous SMCs express NG2, CD13, PDGFRβ, αSMA, transgelin, ABCC9, but not CNN1 (Vanlandewijck et al., 2018). And the expression level of αSMA and transgelin in the venous SMCs is lower than the mural cells in the arterioles (Vanlandewijck et al., 2018; Grant et al., 2019). SMCs in brain venules express NG2, which is different from the venules of peripheral tissues (Murfee et al., 2005; Stark et al., 2013). Outside of SMCs are perivascular space and fibroblast-like cells (Mastorakos and McGavern, 2019).

The CNS vascular system possess a highly selective semipermeable border formed by the BBB wherein tight junctions and adherens junctions created by endothelial cells strictly regulate the movement of ions, molecules, and circulating cells between the blood and the brain (Luissint et al., 2012; Daneman and Prat, 2015). The tight and adherens junctions are controlled by various types of cells surrounding the endothelium, such as pericytes, astrocytes, perivascular oligodendrocyte precursor cells (OPCs), interneurons, perivascular macrophages, microglia, and other immune cells (Abbott et al., 2010; Seo et al., 2014; Faraco et al., 2017; Stebbins et al., 2019) (Figure 2).

Capillary pericytes play especially crucial roles in the function of the BBB. Pericyte ablation leads to breakdown of the BBB in the mouse brain (Nikolakopoulou et al., 2019). Pericytes control protein expression in the tight junctions, their alignment with endothelial cells, and the bulk-flow transcytosis of fluid-filled vesicles across the BBB (Armulik et al., 2010; Bell et al., 2010; Daneman et al., 2010; Quaegebeur et al., 2010).

Pericytes also play a key role in the generation of new blood vessels. During angiogenesis, a complex web of bidirectional signaling pathways between endothelial cells and pericytes is essential for forming and stabilizing new blood vessels (Gaengel et al., 2009; Stapor et al., 2014). The signaling molecules involved in these processes include platelet-derived growth factor B (PDGFB)/PDGF receptor beta (PDGFRβ), transforming growth factor beta (TGFβ), Notch, vascular endothelial growth factor (VEGF), sphingosine-1 phosphate (S1P)/S1P receptor 1 (S1PR1 or EDG), and angiopoietin 1 and 2 (ANGPT1, ANGPT2)/TEK receptor tyrosine kinase (TEK, or TIE2) all of which differentially contribute to these signaling activities (Liu et al., 2000; Winkler et al., 2011b; Zechariah et al., 2013; Eilken et al., 2017; Teichert et al., 2017; Cheng et al., 2018). During angiogenesis, pericytes are reported to be recruited from the bone marrow as well as brain parenchyma in response to the PDGF-BB secreted from endothelial cells (Rajantie et al., 2004; Song et al., 2005; Kokovay et al., 2006; Gaengel et al., 2009). On the other hand, pericytes induce endothelial cell sprouting and stabilization via secreting TGFβ, VEGF, and ANGPT1 (Paik et al., 2004; Durham et al., 2014; Teichert et al., 2017; Blocki et al., 2018). Lack of pericytes leads to endothelial hyperplasia and abnormal vascular morphogenesis including microaneurysm (Lindahl et al., 1997; Hellström et al., 2001). When a single brain pericyte is ablated, the processes from neighboring pericytes are extended to contact uncovered regions of the endothelial cells and maintain the vessel diameter and vessel stability (Berthiaume et al., 2018).

Cerebral blood flow is dynamically altered in response to changes of transient neuronal activity, which is referred to as NVC (Abbott et al., 2010; Attwell et al., 2010; Kisler et al., 2017a). It is controlled by the cells within the neurovascular unit (NVU) including endothelial cells, pericytes, SMCs, astrocytes, OPCs, and neurons (Peppiatt et al., 2006; Attwell et al., 2010; Hall et al., 2014; Mishra et al., 2016; Kisler et al., 2017a, b; Rungta et al., 2018). In response to the different neurotransmitters, pericytes dilate capillaries and increase local CBF (Hamilton et al., 2010). In pathological conditions such as ischemic stroke (Hall et al., 2014; Yemisci et al., 2009) and AD (Nortley et al., 2019), brain capillaries are constricted by pericytes. In ischemic stroke mouse brains, damaged and dead pericytes squeeze the capillaries and sustain the reduction of CBF even after recanalization of the larger vessels, causing the no-reflow phenomenon (Yemisci et al., 2009; Kloner et al., 2018). The burden of amyloid beta (Aβ) oligomer causes pericyte contraction and capillary stenosis, which decreases CBF in the AD brains (Nortley et al., 2019).

The role of pericytes in regulation of vessel diameter has been heatedly debated. When Rouget first describe the branched cells on the capillary wall, which is nowadays called as pericytes, he regarded them as contractile cells (Rouget, 1873). Thereafter, the studies which supported or objected to the pericyte contractility were successively reported (Krueger and Bechmann, 2010). In terms of CBF regulation, SMCs located at arterioles were traditionally thought to control CBF (Iadecola, 2004). This view of CBF dynamics was revolutionized by the findings that capillary diameter also changes with neural activity (Peppiatt et al., 2006; Hall et al., 2014; Kisler et al., 2017b; Khennouf et al., 2018; Rungta et al., 2018) and ischemia (Yemisci et al., 2009; Hall et al., 2014). Furthermore, the loss of pericytes has been reported to lead to diminishing CBF in response to functional hyperemia in pericytes-deficient mice (Bell et al., 2010; Kisler et al., 2017b, 2020). However, Hill et al. (2015) refuted that pericytes are involved in the regulation of CBF, and put forward the view that arteriolar SMCs may be the key players regulating CBF. Fernández-Klett et al. (2010) also showed pre-capillary and penetrating arterioles, but not pericyte in capillaries are responsible for the CBF increase induced by neural activity. These controversial reports most likely stem from the different definitions of pericytes in the proximal capillaries. Some of these reports concur that the mural cells in the proximal branches coming off penetrating arterioles respond to the stimulations outlined above and change vessel diameters accordingly (Fernández-Klett et al., 2010; Hall et al., 2014; Hill et al., 2015; Cai et al., 2018; Khennouf et al., 2018). Pericytes residing at the proximal capillaries possess both characteristics of pericytes and SMCs (Hartmann et al., 2015), which may lead to discrepant interpretations by different investigators. Some groups showed stimulation-evoked increases in synaptic activity and capillary dilation starting mostly at the first- or second-order capillary then propagating along arterioles and downstream capillaries (Cai et al., 2018; Khennouf et al., 2018), which may position the pericytes in the proximal capillaries as the major regulators of CBF. In addition, Grubb et al. (2020) reported that a pre-capillary sphincter, at the junction between the penetrating arteriole and first order branch, modulated capillary flow while protecting the downstream capillary bed from adverse pressure fluctuations. Taken together, the proximal branches coming off arterioles seem to be the gatekeeper that controls CBF in the capillary beds.

Pericytes internalize small molecules and neurotoxic blood-derived products which enter the breached BBB (i.e., immunoglobulins, fibrin and albumin) through receptor-mediated endocytosis or non-specific pinocytosis (Armulik et al., 2010; Bell et al., 2010; Schultz et al., 2017). Pericytes also internalize large solid substance through phagocytosis. Engulfed molecules are transported to lysosomes for enzymatic degradation (Diaz-Flores et al., 2009) or possibly transported to the blood circulation (Zhao et al., 2015). While tumor necrosis factor alpha (TNFα) and interferon-γ (IFNγ) enhance phagocytic uptake, TGFB1 attenuates phagocytic uptake in pericytes.

Pericytes may clear substances derived from the brain parenchyma as well as around vessels. Pericyte loss aggravates Aβ deposition in transgenic mice (Sagare et al., 2013). Aβ clearance by pericytes is mainly performed through receptor-mediated endocytic pathways, especially low-density lipoprotein receptor-related protein 1 (LRP-1) (Shibata et al., 2000; Zlokovic et al., 2010; Ma et al., 2018).

Another clearance system which might be related to pericytes is the CNS lymphatic drainage system. In the CNS, there are two major extracellular fluids, namely, (1) cerebrospinal fluid (CSF) and (2) interstitial fluid (ISF). CSF drains to cervical lymph nodes via the cribriform plate and nasal lymphatics (Kida et al., 1993; Spector et al., 2015), as well as via dural lymphatics (Aspelund et al., 2015; Louveau et al., 2015; Absinta et al., 2017; Ahn et al., 2019) and along cranial nerves (Hatterer et al., 2006; Aspelund et al., 2015). ISF containing metabolic products of the brain as well as Aβ and tau drains to lymph nodes by the shared or distinct pathways from CSF (Engelhardt et al., 2017; Cheng and Wang, 2020). Two different pathways for draining ISF to the periphery — the perivascular and paravascular pathways — are controversially proposed (Engelhardt et al., 2016). In the perivascular pathway, ISF and solutes from CNS parenchyma enter the basement membranes of capillaries, where pericytes are embedded, and drain directly via tunica media of arterioles and arteries out of the brain to cervical lymph nodes (Carare et al., 2008, 2014; Engelhardt et al., 2016). Paravascular pathway, also known as glymphatic system, denotes the moving of CSF into the brain along arterial perivascular spaces and successively into the interstitium to mix with ISF, which then guides flow toward the venous perivascular spaces, removing metabolic waste of ISF to the CSF via convective bulk flow (Iliff et al., 2012). In the glymphatic system, astrocytes are important for ion buffering and fluid exchange between the CSF and ISF (Jessen et al., 2015). The exchange through the glymphatic system is suggested to be dependent on the water channel aquaporin-4 (AQP4) located in astrocytic endfeet (Iliff et al., 2012). However, it remains enigmatic whether or not AQP4 is solely responsible for this fluid transport or not (Lendahl et al., 2019). Recently, insufficient PDGFB signaling in the Pdgfb^ret/ret mice has shown decreased pericyte coverage of the vessels with decreased AQP4 polarization to astrocyte endfeet, which impairs maturation of the glymphatic function (Munk et al., 2019). The focal absence of pericytes correlates with relocation of AQP4 from astrocytic endfeet to the soma of astrocytes (Armulik et al., 2010). Pericytes express laminin-α2 (LAMA2), laminin-β1, and laminin-γ1, which encode the subunits of laminin 211 (Vanlandewijck et al., 2018). Laminin 211 deposits in the vascular basement membrane and interacts with dystrophin in astrocytes, which acts as a molecular bridge to AQP4 to keep it in the astrocyte endfeet (Guadagno and Moukhles, 2004). Indeed, Lama2 knockout in mice results in BBB abnormalities in association with loss of AQP4 polarization to astrocyte endfeet (Menezes et al., 2014). The above referenced reports suggest that pericytes might influence the development of the glymphatic system through deposition of laminin 211 in the vascular basement membrane, which maintains the polarization of AQP4 at astrocytic endfeet. However, there are critical assessments of the proposed glymphatic system (Hladky and Barrand, 2014, 2019; Abbott et al., 2018). Several observations or simulations do not support the glymphatic mechanism (Jin et al., 2016; Smith et al., 2017) nor convective fluid flow of CSF (Asgari et al., 2016; Holter et al., 2017). Hence, the existence of the paravascular pathway as a CNS drainage system is still under debate.

Brain pericytes have many properties of immune regulating cells such as (1) responding to and expressing pro-inflammatory and anti-inflammatory molecules, (2) regulating leukocyte extravasation and trafficking, and (3) controlling immune cell activation including T cells, macrophages, and microglia (Rustenhoven et al., 2017; Thomas et al., 2017; Duan et al., 2018; Smyth L.C.D. et al., 2018). In the mouse brain, pericytes function as the initial sensor of systemic inflammation and relay the infection signal to neurons by secreting chemokine CC chemokine ligand 2 (CCL2, also known as monocyte chemotactic protein-1, MCP1) (Duan et al., 2018).

Pericytes express and release several mediator molecules that enhance leukocyte extravasation. Although the endothelial cells are well known to induce leukocyte crawling and extravasation (Muller, 2002), pericytes also contribute to leukocyte transmigration (Proebstl et al., 2012). In vivo observation of mouse skin vessels have demonstrated that leukocyte extravasation occur only post-capillary venular pericytes (Stark et al., 2013). After inflammation stimuli, neutrophils exhibited transendothelial migration (TEM) and sub-endothelial cell crawling along pericyte processes, which was supported by pericyte-derived intercellular adhesion molecule-1 (ICAM-1) and its leukocyte integrin ligands, macrophage-1 antigen (Mac-1) and lymphocyte function–associated antigen-1 (LFA-1). Then, the leukocytes transmigrated to the interstitium through the gaps between adjacent pericytes (Proebstl et al., 2012). After extravasation, the leukocytes interact with capillary pericytes as well. Pericyte-monocyte interaction is mediated mainly by macrophage migration-inhibitory factor (MIF) and CCL2, whereas neutrophil migration involves MIF and C-X3-C motif chemokine ligand 1 (CXCL8, also known as interleukin 8, IL8) (Stark et al., 2013).

Exposure of pericytes to cytokines such as interleukin 1 beta (IL1β) and TNFα triggers the release of inflammatory molecules and matrix metalloprotease 9 (MMP9), leading to BBB breakdown in vitro (Herland et al., 2016). The immunomodulatory factors secreted by pericytes including IL1β, TNFα, IFNγ, and interleukin 6 (IL6) induce a proinflammatory state in astrocytes, microglia, and endothelial cells, and cause apoptotic neuronal death (Kovac et al., 2011; Matsumoto et al., 2018).

Conversely, pericytes can also secrete several anti-inflammatory substances such as interleukin 33 (IL33) and C-X3-C motif chemokine ligand 1 (CX3CL1) (Rustenhoven et al., 2016, 2017; Yang et al., 2016), both of which are shown to promote anti-inflammatory microglial phenotype in mouse models (Cardona et al., 2006; Fu et al., 2016). Furthermore, depletion of pericytes induced inflammatory responses in endothelial cells and perivascular infiltration of macrophages in mouse retinal vessels, suggesting pericytes exerts an anti-inflammatory effect on endothelial cells under normal conditions (Ogura et al., 2017).

Pericytes display some similarities to mesenchymal stem cells (Wong et al., 2015). Responding to the stimuli and environmental changes, pericytes may transform into multipotent stem cells and differentiate into various cells including neural, vascular, and glial cells (Dore-Duffy et al., 2006; Nakagomi et al., 2015b; Pombero et al., 2016). Pericytes extracted from ischemic mouse brain and human brain pericytes under oxygen-glucose deprivation states develop stem properties in vitro (Nakagomi et al., 2015a). Pericytes under ischemic condition in vivo and in vitro are also reported to acquire a microglial phenotype corresponding with increased phagocytic property (Özen et al., 2014; Sakuma et al., 2016).

These phenotype changes of pericytes under stimulation can be beneficial for the compensatory remodeling after brain injury and ischemia, rapid response to infection and inflammation, and clearing compromised cells or neurotoxic substances breaching an impaired BBB. However, no multipotency of pericytes in aging and injury in vivo has been reported, challenging the current view of pericytes as tissue-resident multipotent progenitors (Guimaraes-Camboa et al., 2017).

Central nervous system injury evokes the recruitment of astroglia and scar formation. Pericytes and OPCs as well as astrocytes are observed within glial scars. After spinal injury or ischemic stroke, pericytes proliferate and migrate to the injured region and form a glial scar (Göritz et al., 2011; Makihara et al., 2015; Dias et al., 2018; Hesp et al., 2018). Extracellular matrix proteins, such as periostin have shown to be expressed in the extracellular space of the injury region, which induces pericyte proliferation and leads to scar formation (Yokota et al., 2017). The glial scar around the injury site forms a barrier between the injured and the non-injured tissue to prevent further neuronal loss, which eventually hinders the axonal regeneration in the scarred area (Zhu et al., 2015; Anderson et al., 2016; Cheng et al., 2018; Dias et al., 2018). Recent evidence has demonstrated that the glial scar can also promote CNS regeneration after injury (Anderson et al., 2016), suggesting a dual function. The complexity and heterogeneity of the glial scar derived from different cell types (i.e., astrocytes, pericytes, and OPCs) at various phases in CNS diseases remains to be elucidated.

Pericytes and endothelial cells are connected to a shared basement membrane by several types of integrin molecules. In areas lacking the basement membrane, interdigitations of pericytes and endothelial cell membranes, called peg and socket contacts, form direct connections by N-cadherin and connexin 43 (Armulik et al., 2005; Winkler et al., 2011b). The crosstalk between pericytes and endothelial cells is indispensable for angiogenesis, vascular stability, and BBB formation. For CNS pericytes and endothelial cells, PDGFB/PDGFRβ, TGFβ, Notch, VEGF, and S1P/S1PR1 signaling events are well investigated (Darland et al., 2003; Paik et al., 2004; Gaengel et al., 2009; Walshe et al., 2009; Liu et al., 2010; Li et al., 2011). ANGPT signaling is investigated in the retinal endothelial cells and pericytes (Winkler et al., 2011b).

In the angiogenesis of the mouse brain, PDGF-BB secreted by endothelial cells recruits PDGFRβ-positive pericytes and progenitor cells (Tallquist et al., 2003; Gaengel et al., 2009). PDGFB signaling also stimulates pericyte proliferation (Geranmayeh et al., 2019), and sustained PDGF-BB–PDGFRβ signaling in the adult CNS is required for pericyte cell survival (Geraldes et al., 2009; Bell et al., 2010).

TGFβ signaling is vital for microvessel stability affecting both endothelial cells and pericytes. Endothelially secreted TGFβ regulates differentiation of pericyte progenitors (Ribatti et al., 2011) and induces pericyte contractile protein expression and extracellular matrix production and facilitates proper pericyte attachment in coordination with Notch signaling (Li et al., 2011; Winkler et al., 2011a). Pericyte-derived TGFβ contributes to endothelial maturation through SMAD signaling (Winkler et al., 2011b).

Vascular endothelial growth factor produced by pericytes and endothelial cells also shows reciprocal interaction (Sweeney et al., 2016). Pericyte-derived VEGF in the mouse brain promotes endothelial sprouting and cell survival (Franco et al., 2011; Eilken et al., 2017). VEGF treatment enhances pericyte coverage of brain endothelial cells with increased N-cadherin production (Zechariah et al., 2013). VEGF induces proliferation and migration of pericytes as well as endothelial cell stabilization (Darland et al., 2003).

S1P is originally described as secreted by endothelial cells. Its receptor, S1PR1 is expressed in mural cells including pericytes. S1P secreted by endothelial cells is essential for pericytes coverage in the mouse brain (Allende et al., 2003) and stabilizes endothelial/pericyte cell adhesion through N-cadherin (Paik et al., 2004; Gaengel et al., 2009) and maintains the BBB (Yanagida et al., 2017). Human pericytes secrete S1P, which induces the expression of adhesion proteins in human retinal endothelial cells in vitro (McGuire et al., 2011).

ANGPT1 and ANGPT2 differently contribute to angiogenesis. The ANGPT1 is mainly expressed in pericytes and ANGPT2 is mainly expressed in endothelial cells. The ligand of ANGPT, TEK is mainly expressed in endothelial cells (Sundberg et al., 2002). Pericyte-derived ANGPT1 activates endothelial TEK and promotes endothelial survival (Geevarghese and Herman, 2014). TEK is also expressed at lower levels by pericytes and its downstream signaling in pericytes is essential for angiogenesis (Teichert et al., 2017). In angiogenesis, ANGPT2 was thought to antagonize ANGPT1, but later was found to act as both agonist/antagonist of TEK signaling in the endothelium (Yuan et al., 2009; Akwii et al., 2019). ANGPT2 expressed by mouse endothelial cells leads to the dissociation of TEK expressing pericytes from vessels, which initiates endothelial cell sprouting (Armulik et al., 2005).

Crosstalk between pericytes and endothelial cells is also mediated by circular RNA. Diabetes-related stress up-regulates a circular RNA, cPWWP2A (PWWP domain containing 2A) expression in pericytes, which inhibit microRNA-579 and regulate vascular integrity (Liu C. et al., 2019).

The crosstalk between pericytes and astrocytes contributes to BBB maintenance, NVC, and white matter attenuation under chronic hypoperfusion (Bonkowski et al., 2011).

Pericytes facilitate the attachment of astrocyte endfeet to the BBB (Ihara and Yamamoto, 2016; Geranmayeh et al., 2019) and pericyte-deficient mice lose AQP4 in the endfeet of astrocytes (Armulik et al., 2010). On the other hand, astrocytes control pericyte migration, differentiation, and the juxtaposition of pericytes to endothelial cells (Nakagawa et al., 2009; Yao et al., 2014). Astrocyte-derived apolipoproteins differently regulate cyclophilin A (CypA) signaling in pericytes, which controls BBB integrity (Bell et al., 2012). In NVC, astrocytic calcium signaling mediates capillary dilation via pericytes (Mishra et al., 2016; Kisler et al., 2017a).

Oligodendrocyte precursor cells (OPCs) have recently emerged as one of the contributors to BBB. According to their regional differences, OPCs can be divided into two subtypes, namely, perivascular OPCs and parenchymal OPCs (Maki, 2017; Kishida et al., 2019). In the human and mouse brain, perivascular OPCs are attached to cerebral endothelial cells and pericytes through basal lamina, and thereby are thought to become novel components of the BBB (Seo et al., 2013; Maki et al., 2015). OPC-specific TGFβ1 depleted mice exhibited cerebral hemorrhage and loss of BBB function, showing the role of OPCs for BBB maintenance through TGFβ1 signaling (Seo et al., 2014). In vitro experiments have shown that OPC-derived factors increase pericyte proliferation whereas pericyte-derived factors support OPC self-renewal and differentiation (Maki et al., 2015, 2018). In the developing mouse forebrain, pericyte-derived TGFβ family proteins contribute to the migration and distribution of OPCs in brain parenchyma (Choe et al., 2014), while perivascular OPC migration to the vessels in the developing CNS requires interaction with endothelium but not pericytes (Tsai et al., 2016).

In the adult mouse brain, pericytes respond to toxin-induced demyelination in the brain and stimulate OPC differentiation during remyelination through Lama2 (De La Fuente et al., 2017). Pericyte-derived Lama2 also instructs neuronal stem cells to an oligodendrocyte fate (Silva et al., 2019).

A recent report has shown that odor triggers rapid Ca²⁺ elevations in OPC processes before pericytes and SMCs dilate the vessels responding to synaptic activation, suggesting possible relationship between OPCs and pericytes in the NVC (Rungta et al., 2018).

Microglia have been regarded as the main executor of inflammation after acute and chronic CNS disorders. The interaction between microglia and vascular cells – including pericytes – has important roles for vascular inflammation, angiogenesis, and BBB integrity (Ding et al., 2018; Thurgur and Pinteaux, 2019). Although endothelial cells are thought to be the main source of cytokines and chemokines which trigger microglial activation upon vascular inflammation, pericytes are also known to be key mediators in this process. In response to TNFα, rat brain pericytes in vitro produce IL6 and macrophage inflammatory protein 1 (MIP1), which trigger microglial activation (Matsumoto et al., 2014). Activated microglia disrupt the BBB, which triggers angiogenesis (Dudvarski Stankovic et al., 2016; Shigemoto-Mogami et al., 2018). In the mouse brain, pericytes initially respond to the systemic inflammation within 2 h and secrete CCL2 before the response of astrocytes or microglia (Duan et al., 2018). Given that CCL2 is also known to activate microglia (He M. et al., 2016; Zhang et al., 2017) and microglial process motility dynamics are altered 48 h after systemic infection (Gyoneva et al., 2014), pericytes might modulate microglial process motility and physical dynamics around the vessels in response to infection. Furthermore, pericytes themselves acquire a microglial phenotype after ischemic stroke as mentioned above (Özen et al., 2014; Sakuma et al., 2016). Conversely, pericytes also secrete several anti-inflammatory substances such as IL33 and CX3CL1 (Rustenhoven et al., 2016, 2017; Yang et al., 2016), both of which has shown promote anti-inflammatory microglial phenotype in mouse models (Cardona et al., 2006; Fu et al., 2016).

In the human and mouse brain, perivascular macrophages lie under the basement membrane alongside pericytes (Fabriek et al., 2005; Goldmann et al., 2016). Perivascular macrophages maintain tight junctions between endothelial cells and limit vessel permeability, phagocytose potential pathogens before they enter tissues from the blood and restrict inappropriate inflammation (Zenker et al., 2003). Although pericytes and perivascular macrophages are localized close to each other and possess shared functions including regulation of vascular permeability and phagocytosis, little is known about how pericytes interact with perivascular macrophages in the vascular niche (Lapenna et al., 2018).

SVD is characterized by pathological changes in the small vessels with a diameter < 100 μm, with concentric smooth muscle thickening in arterioles, as well as pericyte degeneration, basal membrane thickening, endothelial, and astrocyte end-feet swelling in capillaries (Craggs et al., 2014; Bosetti et al., 2016; Østergaard et al., 2016). The slowly progressive worsening of microcirculatory structure and function results in white matter changes, which can be detected by magnetic resonance imaging (MRI).

Small vessel disease is commonly known to be co-morbid brain pathology in a wide range of neurodegenerative diseases including CTE, MS, LBD, and diverse tauopathies including AD (Erdener and Dalkara, 2019). In sporadic SVD, pericytes play pivotal roles as they reside in the small vessels and contribute to maintenance of the BBB, vascular integrity, inflammation, and angiogenesis. Chronic hypoperfusion in the rodent brain results in degeneration of pericytes and decreased pericyte coverage in brain blood vessels, and increased BBB permeability followed by white matter attenuation (Ueno et al., 2002; Liu Q. et al., 2019). Pericyte-deficient mice also cause circulatory failure in the brain which can trigger white matter functional deficits and neuronal loss (Bell et al., 2010; Montagne et al., 2018; Nikolakopoulou et al., 2019).

A leaky BBB allows for the extravasation of toxic-blood derived products such as fibrinogen, which accumulates around the vasculature as insoluble fibrin (Bell et al., 2010; Montagne et al., 2018; Nikolakopoulou et al., 2019). Fibrinogen/fibrin infiltration results in clustering and activation of macrophages and microglia as well as chemokine- and antigen presentation-mediated recruitment and activation of T cells, causing axonal degeneration (Davalos et al., 2012; Ryu et al., 2015). Pericytes produce numerous pro-inflammatory mediators including reactive oxygen/nitrogen species (ROS/RNS), which induces neurons to undergo stress-induced apoptosis (Rustenhoven et al., 2017). This pro-inflammatory status in the vessels induces leukocyte adhesion and microglial activation (Matsumoto et al., 2018; Erdener and Dalkara, 2019). Under chronic hypoperfusion, bone morphogenetic protein 4 (BMP4) expression is increased by pericytes, which induces astrogliogenesis and aggravates white matter attenuation (Uemura et al., 2018).

Chronic hypoperfusion induces compensatory angiogenesis by increasing the expression of angiogenetic factors such as VEGF, ANGPT1/2, and MMP9 (Jian et al., 2003; Ohtaki et al., 2006; Min-Soo et al., 2018). VEGF and ANGPT1 promote sprouting and proliferation of endothelial cells, and recruitment of pericytes (Shane and Didier, 2011). MMP9 regulates the detachment of pericytes from vessels thereby triggering angiogenesis (Joyce, 2005).

The importance of pericytes in SVD may be emphasized by the fact that one of the most common inherited cerebral SVD, cerebral autosomal dominant arteriopathy with subcortical infarcts and leukoencephalopathy (CADASIL), shows aggregation of mutant Notch3 protein around capillary pericytes as well as arteriolar SMCs (Ihara and Yamamoto, 2016). Pericytes express Notch3 and are first affected by Notch3 aggregation in Notch3^R169C mice, suggesting pericytes might be a main contributor in the pathogenesis of CADASIL (Ghosh et al., 2015). A recent study, however, has shown no change in pericyte coverage in the white matter lesion of CADASIL patients nor Notch3^R169C mice, arguing against the prevailing hypothesis that pericyte loss is the primary driver of white matter lesions (Rajani et al., 2019). The jury is still out on the contribution of pericytes to the white matter damage.

Residing in the microvessels, pericytes have a great influence on the condition of the brain following acute ischemic stroke caused by thrombosis or embolism affecting larger vessels. The biological roles of pericytes, such as regulation of CBF, BBB maintenance, inflammation and immunological properties, angiogenesis, and scar formation are all involved in the status of the ischemic brain (Gautam and Yao, 2018).

During arterial obstruction, pericytes positioned on the proximal capillaries constrict the vessels and impede capillary blood flow, which lasts even after arterial recanalization, developing a no-reflow phenomenon (Dalkara and Arsava, 2012; Gursoy-Ozdemir et al., 2012; Hall et al., 2014, Hill et al., 2015). The debate about whether these cells are pericytes or SMCs was discussed above.

During stroke, BBB permeability is increased, and sustained ischemia leads to increased BBB disruption (Simpkins et al., 2016). Severe BBB disruption during stroke increases the risk of hemorrhage in patients treated with intravenous tissue-type plasminogen activator (Deguchi et al., 2014). During ischemia, as in SVD, ROS production and MMP9 up-regulation in pericytes contribute to BBB breakdown. An enzymatic source of ROS production, nicotinamide adenine dinucleotide phosphate oxidase 4 (NOX4), is highly up-regulated by pericytes in the peri-infarct region of the mouse brain subjected to middle cerebral artery occlusion (MCAO), and overexpression of NOX4 in pericytes induces BBB breakdown by up-regulating MMP9 (Nishimura et al., 2016). Pericytes also directly release MMP9 during ischemia, which interrupts the tight junctions between endothelial cells and the binding of astrocyte endfeet to the vascular wall (Underly et al., 2017). VEGF up-regulation by pericytes under ischemic conditions has also been reported to disrupt the BBB in vitro and in vivo (Zheng Gang et al., 2000; Bai et al., 2015), and this triggers further angiogenesis. However, a report has shown that prolonged exposure to VEGF enhances post-ischemic BBB integrity and reduces infarct volume in mice subjected to transient MCAO (Zechariah et al., 2013). Thus, VEGF might have a pluripotent role in BBB integrity according to the dosage and timing of its release.

Pericytes might also play a role in regulating ischemia-induced leukocyte infiltration as pericytes express cell surface adhesion molecules and induce leukocyte transmigration in response to inflammatory mediators (Balabanov et al., 1999; Pieper et al., 2013; Stark et al., 2013). Pericytes express ICAM-1, which guide leukocyte migration through gaps between pericytes by interacting with the integrin ligands on leukocytes (Proebstl et al., 2012).

Aside from these detrimental roles, pericytes play a beneficial role in ischemic stroke via promoting angiogenesis and scar formation. During MCAO-inducing ischemia in the mouse brains, pericytes are recruited from the periphery as well as parenchyma, and are involved in angiogenesis and blood vessel stabilization (Renner et al., 2003; Kokovay et al., 2006). Pericyte migration to the infarcted area forming the core of the scar, which is distinct from the astroglial scar surrounding the core (Fernández-Klett et al., 2013). Consistent with that, Pdgfrb^{+ ⁣/−} mice demonstrated decreased fibrosis in the ischemic area and enlarged infarct volume (Makihara et al., 2015). Although glial scar formation is beneficial to prevent toxic substances from spreading, it should be noted that excessive or long-lasting glial scar formation inhibits axonal regeneration and stalls the recovery process (Dias et al., 2018). Furthermore, detachment of pericytes from capillaries allows them to migrate toward ischemic region thereby causing further leakage of the BBB. The same is true of angiogenesis. While angiogenesis increases the blood supply to the peri-infarct area, insufficient angiogenesis results in leaky blood vessels leading to brain hemorrhage (Kuhnert et al., 2010; Cullen et al., 2011).

AD is the most prevailing dementia among the elderly and is defined pathologically by the presence of Aβ accumulation in brain parenchyma as Aβ plaque and aggregation of hyperphosphorylated tau as neurofibrillary tangles as well as neuritic plaques and neuropil threads. Aβ also accumulate in the vessels as cerebral amyloid angiopathy (CAA). Recently, it has been increasingly recognized that the decreased CBF and white matter attenuation associated with BBB breakdown correlates with the accumulation of AD pathology, and contributes to the onset and progression of dementia (Iturria-Medina et al., 2016; Leijenaar et al., 2017; Park et al., 2019). CBF reduction, BBB breakdown in the hippocampus, and an increase in PDGFRβ level in the CSF occur even in the very early stages of cognitive impairment (Iris et al., 2007; Montagne et al., 2015; Iturria-Medina et al., 2016; Nation et al., 2019) as well as later stages of AD (Miners et al., 2019). In AD patient brains, microvessels are frequently narrowed and irregular in diameter especially in the vicinity of the senile plaques, which is accompanied by decreased capillary bed densities (Kitaguchi et al., 2007). Some vessels in these area are collapsed with lacking endothelial cells, and do not carry blood flow, called string vessels (Hunter et al., 2012). In the mouse brain, infusion of Aβ caused endothelin-1 (ET1) upregulation in cerebral vasculature through receptor for advanced glycation end products (RAGE), which contributes to Aβ-induced CBF reduction (Deane et al., 2003). CBF reduction accompanied by increased vascular RAGE and ET1 is also observed in Tg2576 mice, which is ameliorated by blocking Aβ and RAGE binding (Deane et al., 2003). A recent study has shown that capillaries in the AD brains are constricted by pericytes, which causes a decrease in CBF (Nortley et al., 2019). In rat brains, Aβ oligomer-induced ROS triggers the release of ET1 to stimulate pericytes contraction and CBF reduction (Nortley et al., 2019). AD patients also show a decrease in pericyte coverage with an increase in extravascular immunoglobulin G and fibrin deposition (Sengillo et al., 2013). The apolipoprotein E4 genotype, which is a major genetic risk factor for late-onset AD, leads to pericyte loss and enhances CypA-MMP9 pathway of BBB degradation (Halliday et al., 2016). Pericytes express LRP1 and other Aβ-binding receptors such as the low density lipoprotein receptor (LDLR), RAGE, and CD36 in brains with AD pathology including CAA (Zenaro et al., 2017). Aβ accumulation in pericytes is observed in human AD brains and in the brains of APP^sw/0 mice (Ma et al., 2018). At the ultrastructural level of AD brains, pericytes are disorganized and exhibit mitochondrial abnormalities, pinocytotic vesicles, and accumulation of osmophilic material (Farkas and Luiten, 2001; Baloyannis and Baloyannis, 2012).

Loss of BBB integrity caused by pericyte deterioration may induce an influx of immune cells into the brain, driving inflammation, and CBF stagnation and thereby impairing Aβ clearance, all of which aggravate AD pathology (Mazza et al., 2011; Kinney et al., 2018). Indeed, both the depletion of pericytes in the APP^sw/0 mice (APP^sw/0; Pdfgfb^{+ ⁣/−}) (Sagare et al., 2013) and chronic cerebral hypoperfusion in the APP^SwInd Tg mice (Kitaguchi et al., 2009; Yamada et al., 2011) aggravate AD pathology such as increasing Aβ deposition and tau phosphorylation followed by neuronal loss. Notably, a high-fat diet, which leads to vascular related diseases, exacerbates AD pathology accompanied by pericyte dysfunction in the APP^sw/PS1 mice (Theriault et al., 2016).

Although Aβ is toxic to the pericytes, pericytes basically take an active part in the clearance of Aβ by phagocytosis and translocation through BBB (Winkler et al., 2014; Alla et al., 2019). Pericytes clear Aβ aggregates via an LRP1/ApoE isoform-specific mechanisms, suggesting a potential therapeutic target for controlling Aβ clearance in AD (Winkler et al., 2014; Ma et al., 2018).

Loss of pericytes and BBB dysfunction are common in a variety of neurological disorders including cerebrovascular diseases and AD. Thus, promoting the interaction of pericytes and endothelial cells by regulating PDGFB/PDGFRβ, TGFβ, Notch, ANGPT/TEK, and VEGF signaling could be therapeutic by preventing BBB dysfunction and facilitating proper angiogenesis and vascular stability. For instance, increasing PDGF-BB in the endothelial cells and/or PDGFRβ in the pericytes could boost pericyte proliferation and migration to microvessels while increasing TGFβ signaling could promote pericyte proliferation and attachment to the vessels (Winkler et al., 2011b). Indeed, administration of TGFβ showed increased BBB formation under ischemic condition in vitro and in vivo (Shen et al., 2019). VEGF treatment was also shown to enhance post-ischemic BBB integrity and reduce infarct volume in rodent models of MCAO (Zheng Gang et al., 2000; Zechariah et al., 2013). Recently, regulating RNA for maintaining the BBB has been investigated and overexpression of cPWWP2A or silencing microRNA-579 expression promoted pericyte-endothelial cell crosstalk and microvascular stability (Liu C. et al., 2019). Administration of microRNA-149-5p attenuated BBB permeability and improved the outcomes of rat subjected with transient MCAO (Wan et al., 2018).

Reduction of ROS and MMP9 should also prevent pericyte-mediated BBB breakdown. A recent study observed that MMP9 inhibitors reduced pericyte-associated BBB leakage (Underly et al., 2017). A free radical scavenger edaravone has been sown to ameliorate brain damage after ischemia via pericyte-mediated angiogenesis and vessel stability (Deguchi et al., 2014). Further, cilostazol – a phosphodiesterase 3 inhibitor – promoted angiogenesis through pericyte proliferation with inhibition of the MMP9, which maintained vascular integrity in spontaneously hypertensive stroke prone (SHR-SP) rat (Omote et al., 2014) while cilostazol also ameliorated cerebral hemorrhage in mice by protecting the BBB (Takagi et al., 2017), suggesting cilostazol has additional effects of vascular stability aside from antithrombosis.

Since the no-reflow phenomenon hampers the tissue recovery after recanalization in the arteries, researchers have tried to find clues to prevent capillary constriction by pericytes after ischemia. These ischemia-induced pericyte contraction have shown to be relieved by suppressing ROS/RNS (Yemisci et al., 2009; Deguchi et al., 2014; Hall et al., 2014), removal of external Ca²⁺ (Hall et al., 2014), or administration of adenosine and sodium nitroprusside (Neuhaus et al., 2017; O’Farrell et al., 2017). Capillary constriction by pericytes and CBF reduction are also observed in AD brains, which may aggravate cognitive decline. Aβ infusion into mouse brain causes RAGE-ET1 mediated CBF reduction, similar to the CBF reduction observed in aged Tg2576 mice (Deane et al., 2003). Furthermore, a RAGE-specific inhibitor recovered CBF and lowered the Aβ burden in APP^sw/0 mice (Deane et al., 2012). Aβ oligomers induce pericyte constriction by ROS mediated ET1 release (Nortley et al., 2019). This Aβ-evoked constriction was reversed by applying the vasodilator C-type natriuretic peptide and could be halted by blocking NOX4 or ET1 receptors, suggesting potential therapeutic target for CBF reduction in AD.

Because pericytes take up Aβ, then degrade or excrete it into the circulation, boosting pericyte function could be therapeutic target for Aβ clearance in AD. Consistent with that, pericyte loss in the APP^sw/0 mice showed increased Aβ accumulation and tau phosphorylation (Sagare et al., 2013). Pericytes internalize and clear aggregated Aβ by LRP1-dependent ApoE isoform-specific mechanism (Ma et al., 2018), highlighting up-regulation of LRP1 as a therapeutic target for Aβ clearance.

Neuroinflammation is present in almost all neurological diseases. Pericytes release both pro-inflammatory and anti-inflammatory mediators and regulate recruitment of immune cells from the blood to the brain parenchyma. Although inflammation may have some positive aspects such as immunoprotection against pathogens, clearance of toxic substances, and support of angiogenesis, excessive inflammation causes BBB leakage, tissue damages and neuronal loss.

Targeting ROS-mediated inflammation is a feasible therapeutic target in terms of suppressing release of pro-inflammatory cytokines by pericytes with preventing pericyte loss and BBB dysfunction. A free radical scavenger edaravone has been sown to ameliorate brain damage after ischemia (Deguchi et al., 2014). Targeting receptor or downstream signaling stimulated by pericyte-secreted cytokines may also be potential therapeutic target. Blocking pericyte-derived BMP4 by its receptor antagonist noggin treatment was shown to suppress astrogliogenesis and alleviate white matter damage resulting from chronic cerebral hypoperfusion (Uemura et al., 2018). Targeting transcription factors that regulate immune functions following inflammatory insults might be another option to suppress the detrimental effects of inflammation.

Implantation of mesenchymal stem cell-derived pericytes in mice that model AD plaque pathology reduces the Aβ burden, demonstrating the possibility of cell-based therapy for AD treatment (Tachibana et al., 2018). Easy accessibility of pericytes for autologous transplantation highlights their capabilities for future therapeutic studies (Geranmayeh et al., 2019). Since pericyte-like cells derived from induced pluripotent stem cells (iPSC) acquire BBB properties and are incorporated with iPSC-derived endothelial cells, astrocytes and neurons (Faal et al., 2019; Stebbins et al., 2019), iPSC-derived pericytes might also be promising for implantation therapy of AD and other neurological disorders. Pericytes themselves have been shown to acquire stem cell-like and microglial properties after ischemia (Özen et al., 2014; Nakagomi et al., 2015a), which offers another potential therapeutic target for recovery form CNS diseases.

As discussed above, scar formation in the ischemic brain and brain/spinal cord injury has pluripotent effects on diverse CNS conditions. While the scar formation by pericytes and glia play fundamental roles in promoting angiogenesis and tissue remodeling (Hesp et al., 2018), reducing pericyte-derived scar formation has been reported to promote axonal regeneration and recovery from spinal cord injury (Dias et al., 2018). Administration of periostin-neutralizing antibody, which suppresses pericyte-induced scar formation, ameliorates functional recovery after spinal cord injury (Yokota et al., 2017), but further studies are needed to demonstrate the therapeutic potential of these strategies including the appropriate timing and degree of intervention.