Innate immune sensors and regulators at the blood brain barrier: focus on toll-like receptors and inflammasomes as mediators of neuro-immune crosstalk and inflammation

Abstract

Cerebral endothelial cells (CEC) that form the brain capillaries are the principal constituents of the blood brain barrier (BBB), the main active interface between the blood and the brain which plays a protective role by restricting the infiltration of pathogens, harmful substances and immune cells into the brain while allowing the entry of essential nutrients. Aberrant CEC function often leads to increased permeability of the BBB altering the bidirectional communication between the brain and the bloodstream and facilitating the extravasation of immune cells into the brain. In addition to their role as essential gatekeepers of the BBB, CEC exhibit immune cell properties as they can receive and transmit signals between the blood and the brain partly via release of inflammatory effectors in pathological conditions. Cerebral endothelial cells express innate immune receptors, including toll like receptors (TLRs) and inflammasomes which are the first sensors of exogenous or endogenous dangers and initiators of immune and inflammatory responses which drive neural dysfunction and degeneration. Accumulating evidence indicates that activation of TLRs and inflammasomes in CEC compromises BBB integrity, promotes aberrant neuroimmune interactions and modulates both systemic and neuroinflammation, common pathological features of neurodegenerative and psychiatric diseases and central nervous system (CNS) infections and injuries. The goal of the present review is to provide an overview of the pivotal roles played by TLRs and inflammasomes in CEC function and discuss the molecular and cellular mechanisms by which they contribute to BBB disruption and neuroinflammation especially in the context of traumatic and ischemic brain injuries and brain infections. We will especially focus on the most recent advances and literature reports in the field to highlight the knowledge gaps. We will discuss future research directions that can advance our understanding of the central contribution of innate immune receptors to CEC and BBB dysfunction and the potential of innate immune receptors at the BBB as promising therapeutic targets in a wide variety of pathological conditions of the brain.

Introduction

The brain has a complex vascular network which tightly regulates the exchange of substances between the blood and the brain. The blood brain barrier (BBB) is a dynamic metabolic and physical barrier that restricts the entry of circulating macromolecules, toxins, pathogens and immune cells into the brain parenchyma while allowing the passage of essential nutrients and energy substrates via passive permeability or active transport. Impairment of the BBB disrupts brain homeostasis, enables the extravasation of immune cells into the brain and causes inadequate regulation of molecular transport. The BBB is often compromised in normal aging, CNS infections, brain injuries as well as neurodegenerative and psychiatric diseases and disorders [1,2,3,4]. Highly specialized cerebral endothelial cells (CEC) that form the capillaries are the principal constituents of the BBB. Brain pericytes, embedded in the basement membrane and astrocytes which extend endfeet that tightly ensheathe blood vessels are considered essential components of the BBB because they modulate CEC function and BBB permeability (Fig. 1A) [1, 5,6,7]. Since astrocytes also contact synapses with their numerous processes, they mediate the crosstalk between neurons and BBB cells. In addition, the interactions between CEC and microglia can either facilitate the maintenance of the BBB integrity or impair BBB function, in a time- and context-dependent manner and through activation of distinct mechanisms (Fig. 1B) [8]. The cellular interplay between microglia and CEC in health and disease and the effects of these interactions on BBB integrity have been discussed in a recent comprehensive review [9].

Fig. 1

Cells of the blood brain barrier (BBB). (A) The principal cells of the BBB are cerebral endothelial cells (CEC) which form the capillaries of the central nervous system. Pericytes embedded in the basement membrane and astrocytes which send endfeet are additional components of the BBB. (B) Astrocytes also interact with neurons and microglia and mediate endothelial cell-neuron and endothelial cell-microglia communication. Created with BioRender.com

CEC are polarized cells and have a luminal (apical/blood-facing) and abluminal (basolateral/brain-facing) cell membrane that differ in receptor and transporter composition. The CEC mediate blood-to-brain and brain-to-blood signaling because they respond to both systemic and central cues. Consequently, CEC can modulate both systemic and central inflammation which are essential hallmarks of infections, neurodegenerative diseases, psychiatric disorders and CNS injuries, whereas low-grade, persistent neuroinflammation as a result of increased BBB permeability is observed in normal aging and has been implicated aging-related mild cognitive impairment [10, 11].

The CEC display unique properties and differ from endothelial cells in the vasculature of the periphery because they are interconnected through tight junctions formed between adjacent CEC, lack fenestrae, have low vesicular trafficking [12] and show suppressed transcytosis, a vesicular transcellular transport mechanism which facilitates the shuttling of internalized molecules across the CEC cytoplasm to cross the BBB [13]. The tight junctions are formed by the transmembrane proteins Occludin and Claudins and the cytoplasmic adaptor Zona Occludens (ZO) proteins (Fig. 1A). These proteins form complexes and associate with the cytoskeleton, sealing the paracellular space [14,15,16]. Changes in the levels of tight junction proteins and their cellular mislocalization as well as the disassembly of the protein complexes are main factors underlying the increased permeability of the BBB in neuropathology. Inflammatory signals, including cytokines and reactive oxygen species (ROS), are among triggers that regulate tight junction proteins and, in doing so, BBB permeability [17, 18]. ROS causes damage by oxidizing lipids and proteins and activating matrix metalloproteinases (MPP) which disrupt tight junction proteins [19] and the basement membrane [20]. CEC also express intercellular adhesion molecules that facilitate the migration of circulating immune cells into the brain. E- and P-selectins, intercellular adhesion molecule (ICAM) and vascular cell adhesion molecule (VCAM) are the principal cell adhesion molecules that are expressed in CEC. Inflammatory triggers also regulate these cell adhesion molecules [21,22,23].

Cerebral endothelial cells have many of the characteristics of neuroimmune cells. They release cytokines and chemokines [24, 25] and respond to cytokines such as Interleukin 1β (IL-1β) and Tumor Necrosis Factor α (TNF-α) [26] as well as chemokines [27]. Chemokines released by CEC recruit immune cells circulating in the bloodstream and facilitate their infiltration into the brain [28]. The importance of CEC immune function is exemplified by a recent study showing that innate immune response pathways are highly upregulated in capillary CEC of old mice. Whereas circulatory cues in plasma of young mice reverse the upregulation of aging-related transcriptome in old mice, circulatory cues in plasma of old mice precipitate aging-related transcriptome changes in young mice [29]. Because genes implicated in innate immunity show robust differential regulation under the experimental paradigm described above, the findings suggest that innate immune responses of CEC are regulated by circulatory cues in an age-dependent manner.

Cerebral endothelial cells express innate immune receptors which are the first line of defense against pathogens and endogenous threats. Innate immune receptors sense danger during infections and following tissue damage and initiate a robust and acute inflammatory response that aims to eliminate the danger [30, 31]. However, if unresolved, persistent inflammation can lead to deleterious outcomes. Innate immune receptors comprise pattern recognition receptors (PRRs) which recognize highly conserved molecular motifs in pathogens, collectively called pathogen associated molecular patterns (PAMPs). Endogenous ligands released by damaged and stressed cells in CNS inflammation, injury and neurodegeneration as well as pathogenic molecules such as amyloid beta (Aβ) in Alzheimer’s disease and α-synuclein (α-syn) in synucleinopathies also bind and activate PRRs [31]. These ligands are referred to as damage associated molecular patterns (DAMPs). Upon ligand binding, PRR activate different intracellular signaling cascades that lead to the release of inflammatory effector molecules [32]. PRRs have been classified into several families including toll-like receptors (TLRs), nucleotide binding and oligomerization domain (NOD)-like receptors (NLR), retinoic acid-inducible gene (RIG)-like receptors (RLR), absent in melanoma 2 (AIM2)-like receptors (ALR), C-type lectin receptors (CLR) and triggering receptor expressed on myeloid cells (TREM) [31, 33]. NLR and ALR are PRR that form inflammasomes. Inflammasomes are cytosolic multimeric protein complexes which play key roles in the initiation of the innate immune response and inflammation.

The present review highlights the recent advances on the innate immune function of cerebral vascular cells and especially CEC. We focus on TLRs and inflammasomes as key players in CEC innate immunity, BBB integrity and the cross talk between the brain and the immune system in systemic and neuroinflammation associated with infections and brain injuries.

Toll-like receptors

Toll-like receptors are the first and best-characterized PRR family [34]. Ten TLRs (TLR1-10) have been identified in humans, whereas rodents additionally express TLR1-13 but do not express TLR10 [35, 36]. In rodents, TLR3, TLR7-9 and TLR13 are found in the endosomal compartments while other TLRs are localized on the cell surface [30,31,32]. Ligand binding initiates homodimer formation followed by activation of receptor signaling pathways (Fig. 2). TLR2 forms heterodimers with TLR1 or TLR6. TLRs form receptor signaling complexes with co-receptors including CD-14 and Myeloid differentiation factor 2 (MD2) [37]. TLRs initiate signaling by engaging the adaptor proteins, myeloid differentiation primary response 88 (MyD88), or TIR-domain-containing adaptor-inducing interferon-β (TRIF) through their cytoplasmic Toll/IL-1 receptor (TIR) domain. All TLRs utilize the MyD88-dependent signaling pathway except for TLR3 which signals primarily through the TRIF-dependent pathway. TLR4 uses both the MyD88-dependent or the TRIF-dependent signaling pathways. Activation of TLR signaling leads to the translocation of nuclear factor kappa light chain enhancer of activated B cells (NF-κB), interferon regulatory factors (IRFs), or transcription factor activator protein 1 (AP-1) to the nucleus resulting in the transcription of inflammatory cytokines or Type I Interferons (IFNs) [32, 36].

Fig. 2

Toll-like receptor signaling pathways. TLR2, TLR4, and TLR5 are cell surface receptors whereas TLR3 and TLR7-9 are intracellular receptors. Most TLRs utilize the MyD88-dependent signaling pathway, except for TLR3, which signals through TRIF. Ligand binding to TLRs triggers the formation of homodimers. TLR2 forms heterodimers with TLR1 or TLR6. TLR2 and TLR4 interact with MyD88 via the bridging adaptor TIRAP. TLR4 also signals via the TRIF pathway and utilizes the bridging adaptor TRAM. Activation of the MyD88 signaling pathway leads to the activation of MAPKs resulting in AP-1 dependent transcription. Activation of IKK complex causes the translocation of NF-κB to the nucleus and initiation of NF-κB-mediated transcription. Translocation of IRF5 to the nucleus promotes IRF5-dependent transcription. Activation of the TRIF pathway induces either IRF3- and IRF7-dependent transcription, or AP-1, NF-κB and IRF5-dependent transcription through TRAF6. This leads to pro-inflammatory cytokine and Type 1 IFNs expression and release. AP-1, activator protein 1; IFN, interferon; IRAK, interleukin-1 receptor-associated kinase; IKK, IκB kinase; IRF, interferon regulatory factor; MyD88, myeloid differentiation primary response 88; MAPK, mitogen-activated protein kinase; NF-κB, nuclear factor kappa B; TLR, toll-like receptor; TIRAP, toll/interleukin-1 (IL-1) receptor (TIR) domain containing adaptor protein; TRAM, toll/IL-1R domain-containing adaptor-inducing IFN-β-related adaptor molecule; TRIF, Toll/IL-1R domain-containing adaptor-inducing IFN-β. Created with BioRender.com

TLRs in CEC and pericytes

Endothelial cells of the peripheral vasculature express PRRs including TLRs [38]. Similarly, CEC express TLRs, NLR and RIG-1 [39,40,41]. Rodent and human CEC, and human CEC lines express TLR2, TLR3, TLR4, TLR6 and TLR9 [41,42,43]. Cytokines such as IL-1β and TNF-α and oxidative stress upregulate TLR expression in CEC. Moreover, the serine protease tryptase, which is released by mast cells, increases expression of TLR4 in CEC while downregulating tight junction protein expression [44].

Evidence indicates that DAMPs and PAMPs circulating in the bloodstream are recognized by TLRs on the luminal side of CEC which is followed by release of proinflammatory mediators into the brain which affect brain function. Conversely, DAMPs or PAMPs present in brain infections, injury or neurodegenerative diseases activate TLRs in the abluminal side resulting in the release of inflammatory mediators into the bloodstream. This notion has been supported by in vitro and in vivo studies. CEC monolayer cultures grown in transwell inserts equipped with microporous membranes have been frequently used to create two compartments that mimic the luminal and abluminal side. In this culture model, CEC maintain the polarized expression of receptors and transporters [45]. When a TLR4 ligand, lipopolysaccharide (LPS) derived from gram negative bacteria, is applied either to the luminal or abluminal side, it induces release of cytokines in the luminal compartment. Interestingly, this response is polarized in a cytokine-dependent manner. Whereas Interleukin 10 (IL-10), Granulocyte Macrophage Colony Stimulating Factor (GM-CSF) and TNF-α are released in comparable amounts, whether LPS is applied to the luminal or abluminal compartments, IL-6 release in the luminal compartment is higher when LPS is applied to the abluminal rather than the luminal compartment [25].

There is in vivo evidence suggesting that systemic LPS, acting through TLR4 on the luminal membrane, affects cellular function in the central nervous system. A recent study reported that systemic LPS promotes microglia activation in the retina [46]. Since systemic LPS does not penetrate the BBB in any significant quantity [47], it is likely that it acts by initiating luminal TLR4 signaling which leads to release of inflammatory effectors. These effectors, in turn, activate microglia in the retina. Micoglial response to systemic LPS is abrogated in endothelial cell conditional Tlr4 knockout mice [46], indicating that the effects of LPS depend on endothelial TLR4 expression and are not mediated by microglial TLR4.

Systemic LPS treatment also affects CEC transporter systems. An important group of transporters at the BBB is the superfamily of organic anion-transporting polypeptides (OATP in humans and Oatps in rodents). Several members of the OATP/Oatp family including OATP/Oatp1a1, OATP/Oatp1a2, OATP/Oatp1a4 and OATP/Oatp1c1 are expressed in human or rodent endothelial cells at the BBB [48, 49]. Oatp1c1 (also known as Oatp14) is highly expressed in brain capillary endothelial cells of rodents [50]. It is one of the primary thyroid hormone transporters localized to both the luminal and abluminal membrane of CEC [51]. Oatp1c1 is significantly downregulated in cerebral blood vessels following systemic LPS treatment of rodents [52]. Although the downregulation of Oatp1c1 has been attributed to the effect of cytokines and other inflammatory signals released into the circulation in response to systemic administration of LPS, it is conceivable that LPS-induced activation of luminal TLR4 in CEC contributes to the regulation of Oatp1c1 in inflammatory conditions. This also applies to other endothelial transporters such as the prostaglandin E₂ (PGE₂) transporter oatp1a4 which is downregulated in response to systemic LPS treatment in mice [53].

In addition to LPS, human CEC respond to both TLR2 and TLR3 ligands. Polyinosinic: polycytidylic acid [Poly(I: C)], a TLR3 ligand induces IL-6, IL-8, chemokines, and soluble ICAM and soluble VCAM release by CEC, in vitro. Importantly, both Poly(I: C) and LPS acutely decrease CEC barrier strength [54]. The TLR2 agonist Pam3CSK4 differentially regulates matrix metalloproteinase (MMP)-9 and 2 release by mouse primary CEC and this is paralleled by a decrease in tight junction proteins occludin and ZO-1 and the basement membrane protein collagen IV [55].

It is worth noting that human brain pericytes also express TLR4 and respond to LPS, in vitro. Transcriptional profiling of human pericyte cultures showed upregulation of cytokines, chemokines, ICAM1 and VCAM1 in response to LPS. The increase in ICAM1 and VCAM1 was paralleled by more pronounced adhesion of peripheral blood leukocytes suggesting that pericytes participate in the trafficking of immune cells into the brain under inflammatory conditions. An endogenous TLR4 ligand, high mobility group box 1 (HMGB1), induced expression of IL-6 and IL-8 indicating that human pericytes can mount an immune response when TLR4 is stimulated either by a bacteria-derived or endogenous ligand [56]. In addition, TLR2, TLR5, TLR6 and TLR10 were constitutively expressed in human brain pericytes whereas TLR9 expression, which was not observed under basal conditions, was induced upon treatment with TNF-α, IL-1β and IFN-γ [57].

TLRs as regulators of CEC function in brain injuries

Stroke

Ischemic stroke

Toll like receptor 4 signaling in CEC has been implicated in BBB dysfunction in animal models of ischemic stroke and in vitro. Using middle cerebral artery occlusion (MCAO) as a rat model of ischemic stroke together with oxygen glucose deprivation (OGD) of CEC-astrocyte co-cultures as a cellular model of ischemia [58], a link between TLR4 signaling and the expression and activity of permeability glycoprotein (P-glycoprotein) in CEC has been described. P-glycoprotein, also known as the multidrug resistance protein-1, is an ATP-binding cassette (ABC) efflux transporter that pumps drugs out of cells and clears other substances including pathogenic molecules such as -β-Amyloid from the brain by transport across the BBB [59]. It is enriched in the luminal CEC membrane and limits the penetration of large drugs into the brain by back-transport into the blood. While this limits drug-induced neurotoxicity, it also reduces the efficacy of pharmacological treatments for CNS diseases by interfering with drug penetration into the brain [60]. Therefore, regulation of P-glycoprotein expression and activity under distinct circumstances impacts the effects of drugs differently. In fact, increased P-glycoprotein expression following MCAO has been associated with BBB dysfunction [61] whereas P-glycoprotein deficiency at the BBB has been implicated in the exacerbation of amyloid-β deposition in a mouse model of Alzheimer’s Disease [62]. MCAO-induced upregulation of P-glycoprotein expression and activity is paralleled by increased TLR4 expression and NF-κB activation in CEC. Pretreatment of rats undergoing MCAO with a TLR4, NF-κB or HMGB1 inhibitor, prevents the changes in P-glycoprotein expression and activity confirming the link between P-glycoprotein and TLR4 signaling. in vitro studies investigated the underlying mechanisms and demonstrated that OGD of astrocyte-CEC co-cultures results in release of HMGB1 by activated astrocytes. Exposure of the co-cultures to the HMGB1 inhibitor prevents the increase in P-glycoprotein expression and activity and abrogates TLR4 signaling and NF-κB activation in CEC. Taken together, these findings raise the possibility that following ischemia, HMGB1, potentially released by reactive astrocytes, induces TLR4/NF-κB signaling in CEC and this, in turn, affects CEC function by increasing P-glycoprotein expression and activity [58].

Hemorrhagic stroke

Hemorrhagic stroke can at times be due to disruption of cerebral cavernous malformations (CCM), a collection of enlarged blood capillaries with thinner walls which are prone to leak. CEC dysfunction has been implicated in the formation of CCM and some of the pathological mechanisms have been elucidated [63]. Systemic administration of Gram-negative bacteria or LPS accelerates the formation of CCM whereas loss of a single or both tlr4 alleles in CEC attenuates CCM formation. These findings suggest that endothelial TLR4 expressed on the luminal side plays a key role in this process. In humans, a single nucleotide polymorphism in TLR4 and its co-receptor CD14 were associated with increased lesion number. Interestingly, the authors reported that Gram-negative bacteria in gut microbiome and the CEC response to gram negative bacteria underlie CCM formation [64].

Traumatic brain injury

Studies have shown that LPS-preconditioning confers neuroprotection in TBI [65]. This is indicated by investigations demonstrating that prophylactic pre-conditioning of mice by repetitive systemic administration of low dose LPS protects the brain against TBI without altering BBB permeability, tight junction integrity and claudin-5 and occludin levels. LPS preconditioning activates CEC and induces differential gene expression. Upregulated transcripts include cytokines and chemokines, especially C-X-C motif chemokine ligand 10 (CXCL10). In addition, microglial activation is observed. Selective deletion of MyD88 in CEC prevents LPS-induced increase in CXCL10 expression and microglial activation whereas selective genetic deletion of cxcl10 in CEC abrogates the neuroprotective effects of LPS following brain injury. These findings support the notion that systemic LPS-preconditioning confers neuroprotection by activating TLR4 expressed on the luminal surface of CEC in a MyD88-dependent manner. CEC-derived CXCL10, in turn, modulates microglia activation and confers neuroprotection [66].

TLRs as regulators of CEC function in infections

Brain pathology induced by blood-borne pathogens is associated with impairment of the BBB. The innate immune response mounted by CEC following sensing of blood-borne pathogen-derived ligands by PRR and the dysregulation of CEC function by inflammatory effectors in the circulation are believed to play an essential role in increased BBB permeability and the subsequent brain pathology. Early studies have shown that TLR3 signaling in CEC triggers an IFN-dependent innate immune response to human immunodeficiency virus (HIV) infection. Double-stranded RNA derived from viruses is a TLR3 ligand and Poly(I: C), a synthetic double-stranded RNA, has been used to activate TLR3 under experimental conditions. In human CEC cultures, Poly(I: C) activates interferon regulatory transcription factor 3 (IRF3) and IRF7 which translocate to the nucleus leading to increased IFN-β and IFN-γ expression. IFN-β and IFN-γ in conditioned medium of Poly(I: C)-treated CEC cultures inhibit HIV replication in macrophages. Thus, the TLR3-mediated anti-viral response mounted by CEC can have beneficial effects in HIV infection by limiting viral infection, replication and potentially dissemination by macrophages [67]. However, a study on mouse brain CEC cultures indicated that direct exposure of cells to LPS, added to the luminal compartment, increases transcytosis of HIV-1 by engaging the p38 mitogen-activated protein kinase (p38MAPK) pathway and is potentially mediated by GM-CSF and IL-6 [68]. Thus, activation of different TLRs in CEC can lead to distinct and opposite outcomes in HIV infection.

Recent investigations highlight the important contribution of TLR3 to the anti-viral response mounted by CEC. In human CEC cultures, Poly(I: C) treatment increases the RNA and protein expression of Zinc Finger anti-viral protein S (ZAPS), an IFN-inducible gene product found in the cytoplasm which binds and degrades viral RNA. Since knockdown of ZAP enhances Japanese Encephalitis virus propagation, TLR3-initiated and ZAPS-mediated anti-viral response of CEC is essential to limit viral dissemination [69].

The impact of viral and bacterial infection on drug efflux transporters in CEC have been investigated [70]. In this study, Poly(I: C) and single stranded RNA (ssRNA) were used as viral mimetic ligands to activate TLR3 and TLR7/8, respectively, in human CEC cultures. P-glycoprotein and breast cancer resistance protein (BCRP), two ABC efflux transporters, were differentially regulated, depending on the TLR activated. While Poly(I: C) increased P-glycoprotein activity, it decreased BCRP activity. In contrast, ssRNA downregulated P-glycoprotein activity and upregulated BCRP activity. Thus, activation of TLR3 and TLR7/8 has opposite effects on efflux proteins in CEC and consequently, it could differentially affect the biodistribution of drugs in the brain during infections.

It is worth nothing that there are also reports showing that activation of TLR3 signaling in CEC can have deleterious effects in infections. West Nile virus (WNV) is a mosquito-transmitted RNA virus that can infect the brain in some susceptible individuals such as the elderly and immunocompromised leading to fatal encephalitis. Early studies on Tlr3-deficient mice and wild type controls have shown that systemic WNV increases TNF-α and IL-6 in wild type mice while these cytokines are lower in whole body Tlr3-deficient mice compared to wild type controls. Accordingly, viral load in the periphery of whole body Tlr3-deficient mice is higher than wild type controls. In contrast, viral load in the brain of Tlr3-deficient mice is lower compared to wild type mice and Tlr3-deficient mice are more resistant to lethal infection and neuropathology. These findings suggest that TLR3 plays a role in the entry of WNV into the brain potentially through transendothelial trafficking or via infiltration of infected myeloid cells into the brain following BBB impairment. In fact, systemic administration of WNV or Poly(I: C) disrupts BBB integrity in wild type but not Tlr3-deficient mice and this effect is dependent on TNF receptor-1 signaling. The authors suggest that following systemic WNV infection, the innate immune response initiated by TLR3 activation leads to TNF-α  secretion which, in turn, increases the permeability of the BBB. Subsequently, neuroinvasion by WNV increases lethality. Consistent with this idea, intracerebroventricular administration of WNV to Tlr3-deficient and wild type mice causes equal vulnerability to death. Taken together, the findings highlight the deleterious role of TLR3 in WNV neuroinvasion and lethal encephalitis [71].

Other studies indicate that activation of PRR on CEC by virus-derived PAMPs is sufficient to modulate BBB function in WNV infection [72]. BBB permeability changes during the course of WNV infection. An initial increase in BBB permeability is followed by a transient decrease and thereafter, a rebound. These alterations in BBB integrity correspond to different phases of peripheral and brain infection. The authors hypothesized that the changes observed in BBB permeability depend on competing cytokine signaling induced by activation of TLR3 and TLR7 in CEC. Using the transwell CEC culture model, the study first establishes that TNF-α- and IL-1β increase endothelial permeability and disrupt tight junction assembly whereas IFN-β enhances barrier properties and tight junction assembly and counteracts the effects of TNF-α and IL-1β. Subsequently, the investigations demonstrate that agonist-induced activation of TLR3 increases endothelial permeability and agonist-mediated activation of TLR7 enhances barrier properties. Using wild type and ifnar1 (type I IFN receptor) knockout CEC, together with TNF-α and IL-1β blocking antibodies, the authors show that the balance between TNF-α, IL-1β and type I IFNs determines endothelial permeability and tight junction integrity in response to activation of TLR3 and TLR7 by agonists or WNV-derived PAMPs. Thus, the changes observed in BBB integrity during the different phases of WNV infection could be due to the interplay between TNF-α, IL-1β, and type I IFNs at various phases after infection. Additionally, the authors report that transendothelial trafficking of WNV is differentially regulated by TNF-α, IL-1β and type I IFNs through engagement of Rac1 and RhoA [72], small GTPases which are involved in vesicular trafficking and endocytosis, respectively [73]. These findings are consistent with earlier investigations indicating that TLR3 promotes invasion of the CNS by WNV and fatal encephalitis [71] whereas TLR7 mitigates WNV-induced fatal encephalitis [74].

Blood brain barrier disruption [75,76,77], vascular damage [78], and CEC dysfunction and death [79,80,81] have also been shown in severe acute respiratory syndrome (SARS) coronavirus-2 (SARS-CoV-2) infection which caused the coronavirus disease 2019 (COVID-19) pandemic. In addition to respiratory problems, neurological and psychiatric symptoms are observed in some patients affected by COVID-19 [82,83,84,85,86,87,88], and have been linked to cerebral microvascular damage and neuroinflammation [81, 89,90,91,92,93,94,95,96,97]. The effects of SARS-CoV-2 on the brain and the mechanisms leading to neuroinflammation and neural damage have been discussed in recent comprehensive reviews [98, 99]. Evidence indicates that exposure of human CEC to SARS-CoV-2 elicits an inflammatory response [100]. However, the infection of CEC and pericytes by SARS-CoV-2 remains controversial. Studies on human post-mortem tissue and CEC cultures show that the cells can be infected by the virus [101,102,103,104] which disrupts the NF-kB signaling pathway leading to apoptosis [104]. In contrast, other reports indicate that human CEC are refractive to productive SARS-CoV-2 infection [105,106,107,108]. However, CEC respond to SARS-CoV-2 proteins without the virus entering the cells. Exposure of CEC to virus-derived S1 protein reduces tight junction expression [109] and induces the release of IL-6 and other mediators which, in turn, activate inflammatory pathways in a human microglial cell line [110]. The S1 protein is a subunit of the spike protein S. It recognizes and binds angiotensin-converting enzyme 2 (ACE2), the host receptor, thus mediating viral entry into the cell [111]. SARS-CoV-2-derived proteins act as PAMPs and activate TLR signaling in innate immune cells [99], a mechanism implicated in SARS-CoV-2-induced cytokine storm [112]. In fact, S1 protein activates TLR2 signaling in macrophages causing cytokine release [113]. Moreover, both TLR2 and TLR4 have been implicated in increased cytokine expression in the brain, neuroinflammation, and behavioral deficits following intracisternal injection of the envelope protein E or S1 protein [114, 115]. SARS-CoV-2-derived ssRNA can also activate TLR7/8-MyD88 signaling in dendritic cells [116] and human macrophages and microglia [117] leading to production of IFNs and cytokines. These studies, taken together, indicate that TLRs in both innate immune and CNS cells sense and respond to SARS-CoV-2 or PAMPs derived from SARS-CoV-2. Despite these findings, the specific contribution of TLRs expressed in CEC and pericytes to neuroinflammation and neural damage following SARS-CoV-2 infection has been understudied. It is plausible that TLRs in CEC are activated by SARS-CoV-2-derived PAMPs in the circulation since SARS-CoV-2 RNA has been detected in the blood [118]. It is also possible that PAMPs and DAMPs in the brain of individuals affected by COVID-19 activate TLRs in CEC. Such interactions could exacerbate both neuroinflammation and systemic inflammation in COVID-19. However, as mentioned above, studies on TLRs expressed in CEC in SARS-CoV-2 infection are scarce. This highlights a major gap in the field and warrants further investigations.

Similar to SARS-CoV-2, SARS-CoV-1 infection, which caused the SARS outbreak, and influenza A virus (IAV) infections are paralleled by neurological deficits, neuroinflammation and neural damage [119,120,121]. Investigations on the role of TLRs in SARS-CoV-1 and IVA infections focus mostly on innate immune cells [122,123,124,125,126,127,128]. IAV nuclear protein interacts with TLR2 and TLR4 leading to cytokine release by immune cells [129]. IAV also exerts effects on human immortalized endothelial cells in vitro and downregulates tight junction proteins [130] whereas inoculation of CEC with IAV elicits innate immune and inflammatory responses and causes cell damage by disrupting the cytoskeleton [131]. Despite reports showing the effects of IAV on CEC, the specific contribution of TLRs expressed in CEC to IAV-associated encephalopathy and neuroinflammation has not been addressed and necessitates future investigations.

Inflammasomes

Inflammasomes are categorized into canonical and non-canonical inflammasomes. Canonical inflammasomes comprise a sensor (NLR), an adaptor protein and pro-caspase 1. Most, but not all, canonical inflammasomes utilize the adaptor protein apoptosis-associated speck-like protein containing a caspase recruitment domain/CARD (ASC) which mediates the assembly of the inflammasome complex by interacting with both NLR and pro-caspase-1, the inactive precursors of the aspartate-specific cysteine protease caspase-1. One of the best characterized and broadly studied inflammasomes is the NACHT, leucine-rich repeat (LRR), and N-terminal pyrine domains-containing protein 3 (NLRP3) inflammasome (Fig. 3). The sensor of this inflammasome is NLRP3 which contains a C-terminal LRR domain, a central NACHT domain, and N-terminal PYD. The components of the NLRP3 inflammasomes are not assembled under homeostatic conditions. Sensing of danger signals initiates complex assembly. NLRP3 forms oligomers which recruit the adaptor protein ASC via PYD-PYD interactions. ASC, in turn, recruits pro-caspase-1 through interactions between CARD-CARD domains. Following complex assembly, autoproteolysis of pro-caspase 1 leads to caspase-1 activation. (Fig. 3).

Fig. 3

Assembly of the NLRP3 inflammasome. Components of the NLRP3 inflammasome (upper panel). The components, pro-caspase-1, ASC and NLRP3 are not assembled in homeostatic conditions. NLRP3 contains a C-terminal LRR domain, a central NACHT domain, and N-terminal PYD. Upon danger sensing by NLRP3, complex assembly is initiated; NLRP3 oligomers are formed and recruit the adaptor protein ASC via PYD-PYD interactions whereas ASC interacts with pro-caspase-1 through CARD-CARD (lower panel). Following assembly, autoproteolysis of pro-caspase 1 leads to caspase-1 activation. ASC, apoptosis-associated speck-like protein containing a CARD (ASC); CARD, caspase activation and recruitment domain; DAMP, danger associated molecular pattern; LRR, leucine rich repeat; NLR, nucleotide binding and oligomerization domain (NOD)-like receptors. Created with BioRender.com

Unlike other inflammasomes, NLRP3 inflammasome activation requires two-steps: priming and activation. A first signal, such as a TLR ligand, primes cells. This increases the expression of NLRP3 and the other components of the inflammasome complex. Activation of TLR signaling also leads to production of the inactive precursor molecules pro-IL-1β and pro-IL-18. A second signal, including PAMPs of bacterial, fungal, and viral origin and DAMPs such as Aβ, extracellular ATP, uric acid, silica crystals, lysosomal cathepsin and reactive oxygen species (ROS), triggers the assembly of the NLRP3 inflammasome complex followed by autoproteolytic cleavage of pro-caspase 1 and activation of caspase-1 (Fig. 4). Subsequently, active caspase 1 cleaves pro-IL-1β and pro-IL-18, leading to release of biologically active IL-1β and IL-18 which are potent inducers of inflammation. Activation of caspase-1 can also cause pyroptosis, a form of programmed inflammatory cell death. Gasdermin D (GSDMD), a pyroptosis effector protein, contains an amino-terminal pore-forming domain and a carboxy-terminal repression domain (RD). GSDMD is inserted into the cell membrane via its N-terminal. Cleavage by caspase-1 separates the N-terminal from the C-terminal. The N-terminal oligomerizes and forms large pores in the cell membrane leading to rapid membrane rupture and release of the cellular content including DAMPs and cytokines which initiate inflammation or pyroptotic death (Fig. 3). Reviews of additional inflammasomes has been provided before [132,133,134].

Fig. 4

Activation of NLRP3 inflammasome. Activation of NLRP3 inflammasome involves two steps. The first step, priming (left panel) is triggered by cytokines or PAMPs/DAMPs acting through TLRs and leads to pro-IL-1β and pro-IL-18 production. The second step (right panel) is NLRP3 inflammasome assembly and activation, which is triggered by several signals, including K⁺ efflux, Ca²⁺ influx, extracellular ATP, lysosomal cathepsin and ROS. Increased intracellular Ca²⁺ due to ER stress causes mitochondrial damage and ROS release whereas ruptured lysosomes release cathepsin. This triggers the assembly and activation of NLRP3 inflammasome. Activation of caspase-1 results in the proteolytic cleavage of pro-IL-1β and pro-IL-18 and secretion of active IL-1β and IL-18. In addition, NLRP3 inflammasome activation results in cleavage of GSDMD. The N-terminal fragment (N-GSDMD) generated is incorporated into the cell membrane and forms pores that mediate pyroptosis. ASC, apoptosis-associated speck-like protein containing a CARD; CARD9, caspase recruitment domain-containing protein 9; DAMP, danger associated molecular pattern; GSDMD, Gasdermin D; IL-1β, interleukin 1β; IL-1R, interleukin receptor 1; NF-κB, nuclear factor kappa B; NLRP3, NACHT, LRR- and pyrin domains-containing protein 3; P2X, purinoceptor 7; PAMP, pathogen associated molecular pattern; ROS, reactive oxygen species; TLR, toll-like receptor; TNF, tumor necrosis factor; TNFR, TNF receptor. Created with BioRender.com

In contrast to canonical inflammasomes which have distinct sensors and effectors, non-canonical inflammasomes utilize the same protein as both sensor and effector. The pro-inflammatory caspase-11 in mice and the human orthologue, caspase-4/5, directly recognize LPS. The sensing of LPS initiates the oligomerization of the caspase-11-LPS complex and the autoproteolytic activation of caspase-11 [135] which cleaves GSDMD and initiates pyroptosis [136]. Activation of non-canonical inflammasome does not directly mediate IL-1β and IL-18 release. Instead, the cooperation between caspase-11 and the NLRP3 inflammasome indirectly triggers the release of IL-1β and IL-18 through the GSDMD pores (Fig. 5) [137]. Thus, LPS promotes inflammation and pyroptotic death in a TLR4-independent manner by activating non-canonical and canonical inflammasome signaling [136, 138, 139].

Fig. 5

Non-canonical inflammasome activation. Sensing of intracellular LPS by pro-caspase-11 (or pro-caspase-4/5) leads to oligomerization and activation of the non-canonical inflammasome resulting in activation of caspase 11. Activated caspase-11 cleaves pore forming GSDMD, leading to pyroptosis. Caspase 11 activation also results in NLRP3 inflammasome activation, which, in turn, leads to IL-1β production and release. GSDMD, Gasdermin D; IL-1β, Interleukin β; LPS, Lipopolysaccharide; Created with BioRender.com

Inflammasomes in CEC

Cerebral endothelial cells express NLRs including NLRP1, NLRP3, NLRP5, NLRC4 and NLRC5. Agents that induce oxidative stress, IFN-γ, IL-1β and TNF-α differentially regulate the expression of NLRs in CEC. Exposure of CEC cultures to LPS (priming step) followed by activation of NLRP3 by bacterial muramyl dipeptide (MDP) upregulates the mRNA and protein expression of canonical inflammasome components including NLRP3 and caspase 1. Moreover, this treatment significantly increases IL-1β in cell lysates and culture medium in a caspase-dependent manner. These findings indicate that the NLRP3 inflammasome plays a key role in CEC-mediated inflammation by facilitating IL-1β release [40]. Inflammasome activation in CEC cultures is induced whether stimuli are applied in the luminal or abluminal compartment. Moreover, inflammasome activators in the luminal side induce IL-1β release in both the luminal and abluminal side indicating that sensing of circulating inflammatory cues by CEC results in release of cytokines into the brain [140].

Non-canonical inflammasome pathways are also activated in CEC, in vitro. When LPS is incorporated into the CEC, it is sensed by caspase 11, activating the non-canonical inflammasome leading to IL-1β release. Non-canonical inflammasome activation results in more robust IL-1β release than that observed following canonical inflammasome activation. Disruption of CEC tight junctions is associated with activation of the non-canonical inflammasome [140].

A recent report further supports the notion that during sepsis a non-canonical inflammasome-dependent mechanism in CEC underlies BBB impairment. Circulating LPS, internalized following LPS-binding protein (LBP)-mediated transfer to CD14, binds and activates Caspase 11, which cleaves GSDMD leading to pore formation, CEC membrane permeabilization and pyroptosis in vivo and in vitro. BBB disruption requires Lbp, Cd14, Casp11 and Gsdmd as demonstrated by studies on Lbp^−/−, Cd14^−/−, Casp 11^−/− and Gsdmd^−/− mice which are resistant to BBB impairment following a LPS challenge. Expression of the pore forming N-terminal domain of GSDMD in CEC is sufficient to disrupt BBB integrity even in the absence of LPS whereas neutralization of the N-terminal GSDMD pore forming activity by use of a nanobody prevents BBB breakdown in response to LPS [141].

Inflammasomes in pericytes and CEC-pericyte communication

Human pericytes express several inflammasome components including NLRP1-3, NLRP5, NLRP9, NLRP10, NLRC5 as well as caspase-1, 4 and 5 and ASC. Treatments with agents that induce oxidative stress and cytokines such as TNF-α and IL-1β as well as IFN-γ, differentially regulate the expression of inflammasome components. Although stimulation of pericytes by cytokines and other inflammatory signals increases the expression of pro-IL-1β, it does not induce release of IL-1β suggesting that canonical inflammasome pathways are not activated. In contrast, when LPS is delivered into pericytes, stimulation with a combination of IFN-γ and TNF-α increases both pro-IL-1β expression and IL-1β release. Thus, pericytes contribute to inflammation by activation of non-canonical rather than canonical inflammasomes [57].

Inflammasome-induced signals mediate CEC-pericyte communication [140]. Cultured CEC release IL-1β when exposed to conditioned medium of pericytes which were treated with a combination of IFN-γ and TNF-α. These findings suggest that under the influence of inflammatory cytokines, pericytes release effectors that act on CEC and activate canonical inflammasomes leading to IL-1β release. However, conditioned medium of pericytes that have been transfected with LPS to activate the non-canonical inflammasome and treated with IFN-γ and TNF-α induces more robust IL-1β release when added to CEC cultures. Thus, in an inflammatory milieu, activation of the non- canonical inflammasome in pericytes has a pronounced effect on CEC. The pericyte conditioned medium decreases tight junction protein expression suggesting that pericyte-derived signals during inflammation cause BBB impairment [140].

Role of CEC inflammasomes in brain pathology

Stroke

NLRP3 inflammasome activation has been implicated in stroke-induced impairment of CEC function. Oxygen and glucose deprivation of a mouse immortalized brain endothelial cell line causes pyroptotic cell death. Treatment with an NLRP3 inflammasome inhibitor (MCC950) rescues cells from pyroptotic death. This is paralleled by a decrease in cleaved GSDMD and an increase in uncleaved GSDMD. The NLRP3 inhibitor attenuates the post-ischemic release of inflammatory chemokines and cytokines as well as matrix metalloproteinases by the brain endothelial cell line. In vivo investigations using the mouse MCAO model showed that treatment with the NLRP3 inflammasome inhibitor reduces the lesion volume and permeability of the BBB while also preventing the death of CD-31 positive CEC, which is accompanied by an increase in uncleaved GSDMD, indicating reduced pyroptosis. Collectively, these findings indicate that NLRP3 inflammasome is a key player in the inflammatory response and pyroptotic death of CEC under ischemic conditions and inhibition of NLRP3 inflammasome might have protective effects in ischemic stroke. However, NLRP3 is also expressed in microglia and astrocytes as well as immune cells, and it is possible that the protective effects of the inhibitor are mediated not only through CEC but also glia and infiltrating immune cells [142]. Further investigations on a murine immortalized endothelial cell line (bEnd.3) showed that inhibition of NLRP3 inflammasome attenuates tissue plasminogen activator-induced endothelial cell toxicity under ischemic conditions [143].

CEC express G-protein coupled receptor 124 (GPR124) which plays a key role in brain-specific angiogenesis and BBB integrity. Whereas GPR124 deficiency causes BBB disruption in ischemic mice, increased GPR124 induces inflammation. OGD of the bEnd.3 cell line causes an initial increase in GPR124 followed by a decrease. Overexpression of GPR124 in CEC upregulates NLRP3, cleaved caspase-1, N-terminal GSDMD and IL-1β and leads to pyroptosis. These detrimental effects of GPR124 are prevented by use of a specific inhibitor which reduces pyroptosis and infarct volume in rats subjected to MCAO. Thus, NLRP3 inflammasome-induced pyroptosis of CEC cells is a mechanism by which increased GPR124 can impair BBB function [144].

Traumatic brain injury

Increasing evidence indicates that bidirectional communication between the brain and immune system cells occurs not only through secreted factors but also via extracellular vesicles (EVs) which can cross the BBB in both directions via transcytosis [145, 146]. EVs are membranous structures which are secreted by a vast variety of cells. They carry cargo derived from the cells that they originate and deliver it to recipient cells altering their function. The cargo consists of a vast variety of molecules including lipids, proteins, RNA and DNA [147, 148]. Extracellular vesicles (TEV) isolated from the plasma of individuals who sustained a traumatic brain injury (TBI) exacerbated cerebral edema and vascular leakage when intravenously infused to mice with TBI. Additionally, the high adherence of TEV to the endothelium of TEV-infused injured mice was attributed to increased ICAM1 expression in CEC. Cleaved caspase-1, NLRP3, ASC, and GSDMD-N Terminal were upregulated in CEC following treatment with TEV, suggesting that CEC undergo pyroptotic death. Moreover, tight junction proteins were downregulated indicating disruption of the BBB. Even in uninjured mice, infusion with TEV induced NLRP3/caspase-1/GSDMD-dependent CEC pyroptosis. HMGB1, a DAMP that is increased in the blood and cerebrospinal fluid of individuals who sustained a TBI was implicated in the deleterious effects of TEV as an HMGB1 inhibitor, administered together with TEV, alleviated CEC pathology. A subpopulation of TEV containing HMGB1 (HMGB1⁺TEV) was identified and used to treat CEC isolated from injured mice. This treatment upregulated NLRP3 and GSDMD, which was prevented by the HMGB1 inhibitor [149]. These results, taken together, indicate that EVs isolated from individuals who sustained a TBI carry cargo such as HMGB1 which is delivered to CEC inducing deleterious mechanisms that lead to NLRP3/caspase-1/GSDMD-mediated pyroptotic death. While the stimuli that induce CEC pyroptosis in various pathological conditions might be different, activation of the NLRP3-GSDMD pathway could be the common mechanism underlying CEC injury and BBB dysfunction as observed both in stoke and TBI.

Role of CEC inflammasomes in brain infection

Cerebral malaria (CM) is a fatal disease that primarily affects children, especially those under five years of age. Children who survive CM develop cognitive deficits. Plasmodium falciparum (P. Falciparum) is the pathogen that leads to CM upon infection of the host. A hallmark of P. Falciparum infection is the release of histidine-rich protein II (HRPII), a parasite-derived protein, into the bloodstream. High plasma levels of HRPII have been associated with CM and HRPII deposition vasculature. In the blood of individuals affected by CM and in parasite cultures, HRPII is found as heme-bound nanoparticle consisting of 14 polypeptides loaded with heme. Addition of heme loaded HRPII nanoparticles to human CEC cultures results in binding and internalization of the nanoparticles. Subsequently, intracellular release of heme followed by generation of ferric iron and ROS results in IL-1β release leading to endothelial leakage. Knockout as well as inhibitors of NLRP3 inflammasome and caspase-1 abrogate these effects. However, because endothelial cell activation does not require a priming step and the release of IL-1β is independent of ASC, the authors suggested that IL-1β release might involve non-canonical inflammasome activation [150].

Conclusions

Currently, the immune function of CEC in addition to their role as gatekeepers of the BBB is well recognized. There is vast evidence demonstrating that CEC release inflammatory effectors that contribute to both systemic and neuroinflammation. The effects of inflammatory cues that modulate CEC function, especially the tight junctions and BBB permeability have also been addressed in numerous studies. However, the contribution of innate immune receptors to CEC function requires further investigations. Because TLR4 and NLRP3 inflammasomes are the best characterized and most frequently studied innate immune receptors in the context of neuroinflammation, they have also been investigated in CEC more widely than other innate immune receptors. However, a more comprehensive insight into the innate immune function of CEC necessitates broader investigations on other TLRs, inflammasomes and additional innate immune receptors. New mechanistic insights into the interplay between different innate immune receptors in CEC is needed to advance the understanding of inflammatory BBB breakdown. While CEC have been the primary focus of studies on TLRs and inflammasomes at the BBB, the immune function of pericytes and the contribution of innate immune receptors to inflammatory responses mounted by pericytes are promising research areas given the key role played by pericytes in the maintenance of BBB integrity and the neurovascular unit. Unraveling the role of innate immune receptors in the crosstalk between cells of the neurovascular unit is essential to advance our understanding of BBB impairment in brain pathology. This knowledge is critical to identify potential therapeutic targets and develop new therapies that prevent BBB disruption, facilitate repair and protect the brain from the deleterious consequences of BBB dysfunction in various neurological and psychiatric diseases and disorders and CNS injuries and infections.