The role of inflammasomes in vascular cognitive impairment

Disruptions to CBF directly results in reduced glucose and oxygen supply, leading to immediate bioenergetic impairment and ionic imbalance, and development of excitotoxicity, oxidative stress and inflammation (Fig. 1).

The aforementioned molecular mechanisms initiated by CCH are critical drivers of subsequent pathogenic cellular mechanisms in VCI such as glial activation, blood-brain barrier dysfunction, cell death and demyelination (Fig. 1).

Several structural pathological features are observed in the brain during VCI. Advances in neuroimaging have allowed detection of white matter lesions, lacunes, microinfarcts, microbleeds and enlarged periventricular spaces, which are now considered standard diagnostic features of VCI [15, 127, 128]. In this section, several key pathogenic structural damages and cognitive impairment associated with CCH will be discussed (Fig. 1).

VCI encompasses a broad spectrum of cognitive dysfunctions ranging from subjective cognitive impairment, mild cognitive impairment to dementia (VaD) [8, 13]. The underlying pathophysiological mechanisms responsible for CCH-induced cognitive impairment may be extensive and severe. CCH initiates several cellular mechanisms - glial activation, BBB dysfunction, cell death and demyelination, via the activation of numerous pathogenic molecular mechanisms, including bioenergetic impairment and ionic imbalance, excitotoxicity, oxidative stress and neuroinflammation in different types of brain cells. These mechanisms all contribute to structural damage such as WMLs, microinfarcts and hippocampal atrophy. The influence of these structural damages on cognitive function was discussed previously.

Several guidelines have proposed that VCI is a syndrome with evidence of either clinical stroke or subclinical vascular brain injury, and cognitive impairment affecting at least one cognitive domain [8, 15, 167]. The cognitive domains involved in diagnosing VCI are executive/attention, memory, language, and visuospatial function [8]. Executive dysfunction is a well characterized neurological feature of VaD and can be present in VCI patients who are not demented [168, 169]. Impairments to executive functions are heavily associated with lesions and damage within the frontal lobes or downstream frontal-subcortical circuits of VCI/VaD patients [170, 171]. Memory impairment is a key feature of AD and is also important in VaD. Studies have shown that VaD patients experienced greater impairment in semantic memory than AD patients, possibly due to the inability to retrieve information from short and long-term memory [172‐174]. Impairments in language and visuospatial functions can also be observed in VCI. When presented with a picture description task, VaD patients exhibit lower semantic content production while maintaining a fluency comparable to healthy patients [175]. VaD patients show impairments in most of the tests associated with visuospatial tasks, indicating a deficit in constructional and visuoperceptual ability [174, 176].

Inflammasomes are macromolecular protein complexes that are essential signaling platforms capable of detecting pathogenic signals via pathogen-associated molecular patterns (PAMPs) and endogenous sterile stressors via damage-associated molecular patterns (DAMPs) [38]. Upon detection of PAMPs or DAMPs, the inflammasome complex is activated to trigger downstream inflammatory cascades such as the production of inflammatory cytokines. Given its sterile nature, neuroinflammation in VCI can be considered exclusively driven by DAMPs. The two main characterized inflammasome signaling pathways are the canonical and non-canonical pathways, each of which involves two steps to achieve inflammasome signaling: priming and activation [38, 195, 196].

The canonical inflammasome signaling pathway typically leads to the activation of caspase-1 and -8 (Fig. 2). The priming step (i.e. Signal 1) is initiated by the presence of DAMPs (e.g. HMGB1, IL-1α), which can activate various extracellular PRRs including toll-like receptors (TLR; TLR2, TLR4), receptor for advanced glycation end-products (RAGE) and interleukin-1 receptor 1 (IL-1R1) [195, 197], leading to downstream activation of NF-κB, MAPK, p53 and JAK-STAT pathways [38, 195, 198]. The main purpose of priming is to increase gene expression of key inflammasome components (i.e. receptor/sensor, adaptor and effector) and precursors of IL-1β and IL-18 in the cytosol [199, 200]. Priming can be independent of transcriptional expression whereby post-translational modifications (i.e. phosphorylation and de-ubiquitination) are essential to “license” inflammasome activation [201]. The activation step (i.e. Signal 2) involves stimulation of cytosolic inflammasome receptors/sensors that can be triggered by various DAMPs and/or disturbances in the cellular microenvironment, resulting in assembly and activation of the canonical inflammasome complex to facilitate activation of effectors caspase-1 and -8 [194, 202]. The NLRP3 complex is arguably the best characterized inflammasome, and along with the NLRP1, NLRC4 and AIM2 inflammasome complexes facilitates the self-cleavage of total caspase-1 and -8 into biologically active cleaved caspase-1 and -8, respectively [194, 195, 203, 204]. Both cleaved caspase-1 and -8 can cleave precursors of both IL-1β and IL-18 into mature proinflammatory cytokines that can then amplify downstream inflammation [205, 206]. Cleaved caspase-1 and -8 are also implicated in the activation of programmed cell death pathways such apoptosis and pyroptosis [207‐209], and will be discussed below.

The non-canonical inflammasome signaling pathway leads to the activation of the mouse homolog caspase-11, with the corresponding human homologs being caspase-4 and -5 (Fig. 2). It shares some similarities to the canonical activation of caspase-1, although the non-canonical pathway possesses some unique characteristics. Transcriptional priming is also present for caspase-11 in the non-canonical inflammasome signaling pathway, mediated by extracellular PRRs, including Toll-Like Receptors (TLRs) (i.e. TLR2, TLR4), RAGE and IL-1R1 [195, 197]. Total caspase-11 can detect cytosolic lipopolysaccharides (LPS) from gram-negative bacteria, and oligomerize to form a macromolecular complex of full length caspase-11 components. Subsequently, full length caspase-11 undergoes autoproteolytic cleavage via proximity-induced activation and conversion to its active cleaved form. Cleaved caspase-11 can induce pyroptosis and also indirect maturation of IL-1β and IL-18, in a caspase-1-dependent manner via induction of K⁺ efflux [210, 211]. Low intracellular K⁺ levels can activate NLRP1 and NLRP3 inflammasomes, thereby linking the non-canonical pathway to the canonical inflammasome pathway [212, 213]. Caspase-11 can also be activated by a class of DAMPs known as oxidized 1-palmitoyl-2-arachidonyl-sn-glycero-3-phosphorylcholine (oxPAPC), although oxPAPC upregulates the production and secretion of IL-1β, but does not induce pyroptosis [214].

A canonical inflammasome complex is typically comprised of three distinct components: a receptor/sensor, an adapter and effector (Fig. 3). The sensor is usually a cytosolic PRR that detects perturbations in the intracellular microenvironment that serve as the activation signal to initiate inflammasome assembly. The inflammasome complexes are named after their respective intracellular PRR, which are from two major families: the nucleotide-binding oligomerization domain-like receptor (NLR) family and the pyrin and hematopoietic expression, interferon-inducible, nuclear localization (HIN) domain-containing (PYHIN) family [38, 194]. Members of the NLR family share similar structural domains such as the NLR apoptosis inhibitory protein (NAIP), MHC class II transcription activator (CIITA), incompatibility locus protein from Podospora anserina (HET-E), and telomerase-associated protein (TP1), collectively known as NACHT, and a leucine-rich repeat (LRR) [215, 216]. The NACHT domain is responsible for NLR oligomerization, while LRR is the inhibitory unit folded onto the NACHT domain, keeping the receptor in its inactive state. The two other critical domains are the Pyrin (PYD) domain and the caspase activation and recruitment domain (CARD), which interact with other components with similar domains to form the inflammasome complex [38, 194, 217, 218]. Molecular structures of the AIM2, NLRP1, NLRP3, and NAIP-NLRC4 receptor complexes are discussed below (Fig. 3).

The NLR-pyrin domain containing 1 (NLRP1) belongs to the NLR family of PRRs, with only a single variant of NLRP1 found in humans [194, 219]. The NLRP1 receptor consists of an N-terminal PYD, NACHT, LRR, function-to-find domain (FIIND) and a C-terminal CARD domain. The FIIND domain is unique to NLRP1 in the NLR family, and may play an autoinhibitory role [219, 220] (Fig. 3). The NLR-pyrin domain containing 3 (NLRP3) belongs to the NLR family of PRRs and is perhaps the most studied of all inflammasomes. The NLRP3 receptor consists of an N-terminal PYD, NACHT and LRR domain [194, 221] (Fig. 3). The NLR apoptosis inhibitory protein (NAIP) and the NLR-CARD containing 4 (NLRC4) belong to the NLR family as they both contain NACHT and LRR domains. NAIP is distinct from other NLR proteins as they contain three baculovirus inhibitor-of-apoptosis repeats (BIR) at the N-terminal domain. Unlike the previously mentioned PRRs of the NLR family, NLRC4 contains an N-terminal CARD rather than a PYD domain, in addition to the NACHT and LRR domains [194, 222] (Fig. 3). The absent in melanoma 2 (AIM2) belongs to the PYHIN family of PRRs, and together with four other members forms the AIM2-like receptor (ALR) family of inflammasomes. AIM2 is a bipartite protein, consisting of an N-terminal PYD and a 200-amino-acid HIN domain with two oligonucleotide/ oligosaccharide-binding folds [218, 223]. The HIN domain can bind to double-stranded DNA (dsDNA) independent of its sequence, and during the absence of ligand binding can interact in an intramolecular manner with the PYD to result in autoinhibition [218] (Fig. 3).

The adapter, known as apoptosis-associated speck-like protein containing a CARD (ASC), is a protein responsible for the interactions between various inflammasome components, such as a scaffold protein that connects the PRR to the effector components. ASC contains both an N-terminal PYD domain and a C-terminal CARD domain [217, 224] (Fig. 3).

The effector protein components are a group of inflammatory caspases that catalyze a broad spectrum of substrates upon activation. Caspase-1, − 8 and − 11 play a significant role as the effectors in the inflammasome signaling pathway. Both caspase-1 and -11 contain an N-terminal CARD domain and C-terminal catalytic domain (composed of large p20 and small p10 subunits). Caspase-8 contains two death effector domains (DED) at its N-terminal and the catalytic domain at the C-terminal [225, 226]. Although these caspases are similar in structure, they differ in their domain linkers and residues in their catalytic pocket, contributing to their differential role in the inflammasome signaling pathway [227‐229] (Fig. 3).

During the pathogenesis of VCI, a multitude of stress signals and DAMPs are produced in the cytosol, released and detected by various PRRs, resulting in their eventual activation and formation of inflammasome complexes [194, 200]. This section discusses stimuli that potentially activate the NLRP1, NLRP3, NAIP-NLRC4 and AIM2 inflammasome receptors during CCH (Fig. 4).

The exact function of NLRP1 in innate immunity is not yet fully elucidated, but it is believed to be activated via autolytic proteolysis within the FIIND domain in response to severe bacterial infections [220]. NLRP1 levels are upregulated in ischemic conditions, and although the responsible activation signals are unclear, they are likely to be from aberrations in the cellular microenvironment such as depletion of intracellular ATP and reduction of intracellular K⁺ levels arising from K⁺ efflux during bioenergetic impairment [38, 212, 230]. The NLRP3 receptor can be activated by a multitude of intracellular signals including decreased K⁺, increased Ca²⁺ and oxidative stress during CCH [212, 231, 232]. Given that NLRP3 can respond to a diverse range of signals, it is likely to be responding to a common cellular event caused by these activators rather than directly to the activators themselves. Moreover, increased levels of hyaluronic acid was observed in the cerebrospinal fluid of VaD patients that appeared to serve as a potential NLRP3 stimuli [233].

While no study to date has directly investigated the effect of CCH to cause bioenergetic deficits, blood flow reduction during ischemia is known to induce K⁺ efflux [46, 234]. This is a possible mechanism by which CCH activates the NLRP inflammasome receptors. Lower cerebral blood flow reduces ATP production and impairs ATP-dependent transporters such as the Na⁺/K⁺-ATPase pump leading to K⁺ accumulation in the extracellular space during ischemia [46, 234]. Alternatively, ATP released by damaged cells can bind to the P2X purinoceptor 4 (P2X4) and P2X purinoceptor 7 (P2X7) on neighboring cells, leading to opening of the ligand-gated ion channel and K⁺ efflux. In addition, K⁺ can be passively released into the extracellular environment due to increased permeability of the plasma membrane from damaged cells [38, 235]. Elevated levels of extracellular K⁺ can activate Pannexin-1 channels on the plasma membrane through a mechanism independent of the membrane potential, to further promote release of ATP, creating a positive feedback loop for K⁺ efflux [38, 236, 237]. K⁺ efflux has been identified as a key step for NLRP1 and NLRP3 inflammasome activation, although there is no clear understanding of the precise mechanism linking K⁺ efflux and NLRP receptor activation [232, 238]. However, a recent study showed that low intracellular K⁺ concentrations can open the inactive structure of NLRP3 by altering a domain in between the PYD and NACHT domains, resulting in a stable structure that promotes the functional oligomerization of NLRP3 into active oligomers [239] (Fig. 4).

Other than K⁺ efflux, a reduced intracellular cyclic adenosine monophosphate (cAMP) concentration during CCH may activate the NLRP3 inflammasome [235, 240]. Binding of cAMP to the NLRP3 receptor inhibits its ability for inflammasome assembly. When cAMP is reduced upon CCH, NLRP3 activation is promoted [231, 235, 240]. While the underlying cause of cAMP reduction is unknown, evidence from mechanistic studies from ischemic models suggest an involvement of Ca²⁺. When Ca²⁺ is released by damaged cells, it activates G-protein coupled calcium-sensing receptors (CaSRs) on the plasma membrane, allowing it to interact with Gαi and inhibit adenylate cyclase, reducing the conversion of ATP to cAMP in neighboring cells [38, 231, 235]. The involvement of Ca²⁺ in mediating NLRP3 inflammasome activation is highly plausible as the Ca²⁺-permeable channel, transient receptor potential melastatin 2 (TRPM2), plays a significant role in regulating the production of IL-1β upon CCH [70]. One possible explanation for this is that it allows influx of Ca²⁺, which activates protein kinase R (PKR) in the cytoplasm. Upon activation, PKR can phosphorylate NLRP1 or NLRP3 receptors for inflammasome activation, leading to production of IL-1β [38, 241, 242] (Fig. 4).

Numerous lines of evidence have shown that NLRP3 inflammasome activation is closely linked to increased levels of ROS in neurological diseases [243, 244]. Several studies have already demonstrated a close association between ROS, thioredoxin-interacting protein (TXNIP) and NLRP3 [243‐245]. The generation of ROS facilitates the uncoupling of TXNIP from thioredoxin (TRX), allowing TXNIP to bind to the NLRP3 receptor for inflammasome activation [244‐246]. Due to increased oxidative stress observed in CCH mouse models of VCI [39, 55, 247], one study demonstrated that decreased TXNIP-associated oxidative stress was associated with reduced NLRP3 and IL-1β expression, in conjunction with better cognitive performance [248]. Another major source of ROS is from the mitochondria as increased hydrogen peroxide production induced oxidative stress that were observed in rodent models of CCH [39] (Fig. 4).

NAIP-NLRC4 is generally activated by pathogenic bacteria, with NAIP as a direct receptor to these bacterial signals [249] However, activation of NAIP-NLRC4 has been shown to be associated with lysophosphatidylcholine (LPC), which are lipids arising from the hydrolytic activity of phospholipase A2 (PLA₂) under stressed conditions [250, 251]. Although LPC has been found to activate NLRC4 and NLRP3 inflammasomes in neuroinflammatory disease mouse models [250, 252], it is unlikely to be involved in VCI as the levels of LPC do not differ significantly upon CCH [253]. Nucleotide-derived metabolites, including adenine and N4-acetylcytidine (N4A), can both prime and activate the NAIP-NLRC4 inflammasome [254], and could be relevant given that oxidative DNA damage has been described in VCI patients [60]. However, no evidence to date have investigated the effect of CCH on nucleotide-derived metabolites.

AIM2 is activated by the binding of dsDNA to the HIN domain, thereby removing the autoinhibitory effect of the intramolecular interaction between the PYD-HIN domains. This ligand binding is achieved via electrostatic attractions between the positively-charged HIN domain and negatively-charged dsDNA [218, 223, 255]. The dsDNA is required to be at least 80 base pairs in length and is conventionally from viral or bacterial origins [256]. Given the sterile nature of VCI, it is likely that AIM2 is activated by host dsDNA instead because ischemic conditions produce anoxic depolarization and release of mitochondrial DNA into the cytosol due to ATP-induced mitochondrial apoptosis. Coupled with the extracellular release of nuclear and mitochondrial DNA by injured neurons and glial cells, this provides the appropriate activation signals in the form of dsDNA to AIM2 [257, 258]. During CCH, DNA fragmentation has been observed to occur in astrocytes and oligodendrocytes [130, 259]. Furthermore, it was recently shown that CCH increases plasma levels of double-stranded DNA, and induces AIM2 inflammasome-mediated neuropathology and cognitive impairment in a mouse model of VaD [71]. Moreover, CCH-induced receptor for advanced glycation end-products (RAGE) upregulation can also promote DNA uptake into the cell via the action of endosomes, hence providing a mechanism through which extracellular DNA can interact with cytosolic AIM2 [260, 261] (Fig. 4).

In general, stimulation of the inflammasome receptor causes the LRR inhibitory unit to unfold from the NACHT domain [38, 216]. Consequently, the inflammasome receptor reconfigures into an “open” structure to allow homo-oligomerization with neighbouring inflammasome receptors. As more inflammasome receptors converge, they form a macro-molecular platform with their N-terminal PYD domains pointing towards each other [224, 262]. A PYD domain attracts other PYD domains via homotypic interactions. One of them is the adaptor protein, ASC, which facilitates inflammasome complex formation by binding to the PYD domain of the inflammasome receptor platform using its N-terminal PYD domain through homo-oligomerization [38, 224]. Many ASC proteins will come together during this process, forming a filamentous structure from the inflammasome receptor platform. These filamentous macromolecular aggregates are known as ASC specks [263]. As such, the CARD domain on the C-terminal of ASC is made available to bind with full-length caspases. These caspases bind with their N-terminal CARD domain to the ASC specks via CARD-CARD domain interactions [217, 263], and therefore directly interact with CARD domains on inflammasome receptors in the absence of the ASC adaptor. NLRP1 and NLRC4 receptors can mediate ASC-independent inflammasome activation [249, 264]. The structural formation of each major inflammasome complex is explained below (Fig. 5).

Following stimulation of the NLRP1 inflammasome receptor, the FIIND domain undergoes autolytic proteolysis to facilitate activation and oligomerization to form the inflammasome core [220]. NLRP1 oligomers appear to interact with ASC via homotypic PYD-PYD interactions [220]. After ASC speck formation, the ASC speck can recruit multiple effector inflammatory caspases via homotypic CARD-CARD interactions [219]. Despite being able to mediate ASC-independent NLRP1 inflammasome activation, NLRP1 activity level is higher in the presence of an ASC speck [220] (Fig. 5). Prior to NLRP3 activation, priming must occur in the form of increased expression and de-ubiquitination of NLRP3 [199, 201]. Assembly of the NLRP3 inflammasome also requires ASC to be linearly ubiquitinated [265]. The NLRP3 receptor binds to ASC via homotypic PYD-PYD interactions, and through the resulting ASC speck recruits effector inflammatory caspases via homotypic CARD-CARD interactions [221, 224] (Fig. 5). Upon binding of NAIP to the ligand, a single NAIP oligomerizes with multiple NLRC4 receptors. Phosphorylation of a single evolutionarily conserved serine 533 residue in NLRC4 is necessary for the assembly of the NAIP-NLRC4 inflammasome [266]. NAIP-NLRC4 then interacts with ASC likely through homotypic CARD-CARD interactions, given the absence of a PYD domain in the NAIP-NLRC4 receptor. ASC speck formation ensues and is accompanied by recruitment of multiple effector inflammatory caspases through homotypic CARD-CARD interactions. NLRC4, however, possesses an innate CARD domain, and so can recruit the effector inflammatory caspases independent of ASC, although the presence of ASC can enhance the assembly of the NAIP-NLRC4 inflammasome complex [222, 249, 263] (Fig. 5). The PYD domain of AIM2 possesses an intrinsic tendency to self-aggregate, resulting in homo-oligomerization of several AIM2 receptors after activation via dsDNA binding to the HIN domain. The AIM2 receptor complex then recruits ASC concomitant with ASC speck formation via homotypic PYD-PYD interactions. AIM2 recruits multiple effector inflammatory caspases via homotypic CARD-CARD interactions in an ASC-dependent manner, due to the absence of CARD in the AIM2 receptor [255, 267] (Fig. 5).

The molecular assembly of different inflammasome complexes can vary between inflammasome receptors in the canonical pathway. The ultimate purpose of inflammasome assembly is to bring together the effector caspases in close proximity. Increasing the local concentration of effector caspase-1 around the complex facilitates dimerization of caspase-1 monomers, and enables autoproteolytic activation [255, 267]. Subsequently, a transient activity hub of cleaved caspase-1 p33/p10 is produced for substrate catalysis. Different cell types and inflammasome sizes influence the kinetics of p33/p10 processing. Therefore, the final product of cleaved caspase-1 p20/p10 serves as a classic hallmark of caspase-1 activation as it indicates termination of protease activity [226, 268] (Figs. 2, 3 & 5). Unlike caspase-1, which binds to the inflammasome complex via CARD-CARD interactions, caspase-8 relies on its DED domain to carry out heterotypic interactions with the adaptor ASC. This is possibly due to the similarities with its self-association between DEDs and PYDs domains. As such, caspase-8 is similarly able to undergo proximity-induced activation, producing cleaved caspase-8 for substrate processing [224, 269, 270]. Nonetheless, caspase-8 is capable of cleaving inflammasome substrates (IL-1β) directly via the complex formation of caspase recruitment domain-containing protein-9 (CARD9), B-cell lymphoma/leukemia-10 (BCL10), Mucosa-associated lymphoid tissue lymphoma translocation protein-1 (MALT1), ASC and caspase-8. Caspase-8 can also interact with receptor-interacting serine/threonine kinase 1 (RIPK1) directly cleaving caspase-1 leading to caspase-1 activation [205, 206].

Organization of the canonical inflammasome complex usually involves a receptor, adaptor and effector caspases-1 and -8. However, in the non-canonical inflammasome pathway, caspase-11 can undergo homo-oligomerization in the absence of a receptor and adaptor protein component. Upon binding of a stimuli such as oxPAPC or LPS to the N-terminal CARD domain on caspase-11, the CARD domain interacts with the same domain on another caspase-11 to form a dimer [271, 272]. The dimer then interacts with others to form a homo-tetramer, which then oligomerizes to induces auto-proteolysis. As caspase-11 undergoes self-cleavage, it releases the pro-domain from the catalytic domain of caspase-11, producing biologically active cleaved caspase-11 [271, 272] (Figs. 2, 3 & 5).

Upon the activation of the inflammasome signaling pathway, the mature cytokines IL-1β and IL-18, are produced leading to the downstream inflammatory response. Simultaneously, these effector proteins also initiate a wide variety of cell death pathways such as apoptosis, pyroptosis and secondary necrosis [38] (Fig. 2).

As mentioned, microglia and astrocytes are key players in mediating neuroinflammation and tissue damage. Microglial cells are equipped with different PRRs that screen the microenvironment in the resting state [286, 287]. In an RNA-sequencing transcriptome study using mouse brain, gene expression of NLRP1, NLRP3, NLRC4 and AIM2 in the microglia was highest among all brain cell types. Moreover, microglial expression of caspase-1 and IL-1β was around four and thirty fold higher than neurons, respectively [288]. This primes microglia for rapid activation when CCH induces ionic imbalance, oxidative stress and inflammation. Canonical classification of activated microglia is broadly described as either proinflammatory M1-like or anti-inflammatory M2-like phenotypes. M1-polarised microglia produce proinflammatory cytokines such as IL-1β and TNF-α that serve to activate more microglia cells in a paracrine manner [289]. However, emerging transcriptomic studies revealed more disease-associated subtypes of microglia such as Keratan sulfate proteoglycan (KSPG)-microglia (associated with amyotrophic lateral sclerosis), highly active “dark microglia” which interact with blood vessels and synapses (associated with Alzheimer’s disease) and CD11c-microglia which interact with oligodendrocytes (associated with demyelination) [290]. Based on current evidence, microglia are likely to promote inflammatory and non-inflammatory responses via an involvement of NLRP3 and AIM2 inflammasomes during VCI [71, 291]. In fact, a study showed reduced proinflammatory microglia signatures in the hippocampus of mice with AIM2 deficiency upon CCH [71]. Attenuating NLRP3 inflammasome activity via pharmacological inhibitors during CCH also reduces microglial overactivation, possibly due to lower production of ROS during drug treatment [291].

Astrocytes are another immune effector cell type in the brain. Similar to microglial polarization states, activated astrocytes exist in two phenotypes based on traditional understanding: neurotoxic A1 and neuroprotective A2 astrocytes [292, 293]. Beyond this binary classification, reactive astrocytes were found to be more heterogeneous in terms of their morphology, locality, cellular interaction, and molecular expression. Therefore, the status of reactive astrocytes varies in a context-, time- and stimulus-specific manner [294]. In the context of neurodegenerative diseases, reactive astrocytes are consistently being identified with hypertrophied morphology and reduced expression of essential ion and neurotransmitter channels and receptors such as ATP-sensitive inward rectifier potassium channel 10 (Kir4.1), glial glutamate transporter 1 (GLT1), and increased expression of the glial fibrillary acidic protein (GFAP). This state of reactive astrocytes is commonly observed in A1 astrocytes that release proinflammatory and toxic mediators [294‐297]. Both A1 and A2 phenotypes were present in conjunction with inflammasome activation following CCH.

Astrocytes have been shown to express NLRP1, NLRP3, AIM2 and NLRC4 receptors under different conditions. However, genetic deletion of the AIM2 receptor did not affect the expression of GFAP in either the cortex or hippocampus of mice during CCH, suggesting potential involvement of other inflammasome receptors [71, 257, 298, 299]. We recently collected unpublished data that the inflammasome signaling pathway is activated in fibrous and protoplasmic astrocytes under in vitro ischemic conditions, resulting in mature cytokine production, and apoptotic and pyroptotic cell death. Hence, we postulate that inflammasome activation may serve as a signal for astrocyte polarization.

BBB disruption can arise from neuroinflammation, primarily through degradation of various tight junction proteins (TJPs). Matrix metalloproteinases (MMPs) released by glial cells can break down the extracellular matrix [300] and have been implicated in the disruption of the BBB under ischemic conditions [104]. MMP-2 and MMP-9 are known to be involved in BBB disruption and primarily localized to ischemic regions of astrocytic foot processes, during which TJPs undergo degradation over time [301]. Inflammasome signaling contributes to BBB dysfunction via the action of IL-1β, which can upregulate the expression and release of MMPs from glial cells [302, 303]. Application of exogenous IL-1β elicits BBB dysfunction in rat and human brain microvascular endothelium [94, 304]. In addition, both IL-1β and IL-18 can upregulate expression of various chemokines in the extracellular space and cell adhesion molecules on the endothelium [276, 305]. Cell adhesion molecules, such as ICAM-1 and VCAM, attract and facilitate immune cell infiltration into the brain during neuroinflammation. These peripheral cells can also release active MMP-9, further exacerbating BBB dysfunction [306, 307]. A recent study showed that an interleukin-1 receptor antagonist preserved BBB integrity, attenuated changes in expression and localization of TJPs and MMPs in a rat model of ischemia-reperfusion [308]. Inhibition of caspase-1 enzymatic activity in a similar model further revealed caspase-1 to induce BBB dysfunction through the activation of pyroptosis [309]. Therefore, a number of studies support a critical role of inflammasomes in BBB disruption, especially following disruption to CBF.

As an essential component of the BBB, the endothelial cell expresses a wide range of inflammasome receptors such as NLRP1, NLRP3 and NLRC4 [310]. Studies showed endothelial NLRP3 in modulating the BBB during different disease conditions, but evidence for CCH is still lacking [311‐313]. NLRP3 may mediate BBB dysfunction for CCH based on in vitro and in vivo evidence from studies focusing on ischemic brain injury. Pharmacological inhibition of NLRP3 inflammasome activity has been shown to attenuate cerebral ischemia-induced BBB dysfunction by reducing its permeability and upregulating TJPs [313]. In addition, the study confirmed that reduced NLRP3 activity in endothelial cells increased expression of TJPs, cell viability and reduced barrier leakage [313]. This mechanism observed may explain the BBB dysfunction that is observed during CCH.

Pericytes also express the NLRP3 receptor along with other NLRP and NLRC inflammasome receptors [314]. Upon stimulation of pericytes by oxidative stress and proinflammatory mediators, in vitro cerebral pericytes demonstrated upregulation of NLRP3 and NLRC4 mRNA expression [314]. Despite an increase in inflammasome receptor mRNA expression, activation of the inflammasome complex was not detected in cerebral pericytes following exposure with a wide range of DAMPs [314]. However, as pericytes are increasingly implicated in BBB dysfunction during CCH [91, 100, 101], additional studies are needed to establish the role of inflammasomes in pericytes during VCI.

Inflammation is closely linked to cell death at the molecular level. The cell death pathway can be triggered by various proinflammatory mediators [315]. As DAMPs initiate the inflammasome signaling pathway via the NLR family and interferon-inducible protein, it activates key effector proteins such as caspase-1 and caspase-8 to cleave caspase-3, resulting in apoptosis and secondary necrosis [206, 283]. Together with the non-canonical caspase-11, these effector proteins catalyze pyroptosis via cleavage of gasdermin-D (GSDMD) [203, 208, 210]. Inhibition of caspase-1 reduces inflammasome activation and cell death in primary cortical neurons and murine microglial cells subjected to ischemia-like conditions [200, 203]. The NLRP3 inhibitor, MCC950, decreases apoptotic cell death and brain infarct size in a mouse model of ischemic stroke [316]. In a mouse model of VCI, AIM2 knockout mice expressed reduced inflammasome activity as well as apoptotic and pyroptotic cell death in neurons and microglia in the cortex and hippocampus. Similarly, a higher neuronal count was observed in the CA2 and CA3 area of the hippocampus in AIM2 knockout mice during CCH [71]. Therefore, the AIM2 inflammasome is involved in CCH-induced neuroinflammation by mediating cell death and neuronal loss during VCI disease progression.

Numerous studies have indicated a close association of inflammasome activity with the formation of WMLs that often occur together with activated microglia [77, 203, 317]. As mentioned, activated microglia release numerous inflammatory mediators that contribute to demyelination, including IL-1β and IL-18 that are produced during inflammasome activation. IL-1β was shown to impede oligodendrocyte migration and white matter repair in mouse models of VCI [69]. Conversely, reduced secretion of IL-1β preserved myelin integrity and attenuated the formation of WMLs under CCH [71]. Anti-inflammatory pharmaceutical interventions and transgenic animal models have attenuated WML formation via suppressing microglial activation, and preventing caspase-1 and IL-1β production [70, 318, 319]. Other than in microglia, inflammasomes are also activated during CCH in oligodendrocytes. Our recent study found increased levels of cleaved caspase-1 in oligodendrocytes of mice after 30 days of BCAS [71]. Similar observations of inflammasome-mediated apoptotic and pyroptotic cell death markers were also found within oligodendrocytes. The evidence suggests that the inflammasome signaling pathway likely plays a causative role upstream of CCH-induced WML formation.