Capillary-Associated Microglia in Neurovascular Coupling: Localization, Mechanisms, and Disease Implications

Abstract

Capillary-Associated Microglia (CAMs) are increasingly recognized as integral components of the neurovascular unit (NVU) that interface with intracerebral capillaries and are implicated in microcirculatory regulation, neurovascular coupling (NVC), and blood–brain barrier (BBB) maintenance. Rather than a rigid lineage-defined subset, CAMs are often best conceptualized as an anatomically anchored and context-responsive microglial state in which somata and/or primary processes closely appose the capillary wall, enabling continuous sensing of vascular-derived signals and local modulation of vascular function. Advances in in vivo two-photon imaging, single-cell and spatial transcriptomics, and genetic perturbation approaches have begun to delineate their positioning dynamics and candidate molecular programs. Mechanistically, CAMs can engage purinergic signaling—most prominently via the PANX1–P2Y12R axis—to detect extracellular nucleotides and support process responses and capillary association, while additional pathways involving COX-1–dependent prostanoid production and CD39-mediated nucleotide hydrolysis may contribute to basal tone regulation and activity-dependent vascular reactivity, in part through local adenosine availability. Through coordinated interactions with endothelial cells, pericytes, astrocytes, and neurons, CAMs occupy a strategic immunovascular interface, where they may couple immune surveillance to BBB integrity and metabolic homeostasis. In disease, alterations in the positioning and signaling of CAMs have been linked to a range of neurological and neurovascular disorders, including Alzheimer’s and Parkinson’s diseases, cerebral small vessel disease, diabetic encephalopathy, and ischemic stroke, with emerging evidence suggesting that chronic inflammatory or metabolic stress can reshape these programs and compromise NVC. Key challenges remain in establishing operational definitions and cross-species comparability, achieving long-term in vivo tracking with causal resolution, and improving the physiological fidelity of in vitro models. Future work will benefit from standardized criteria for CAMs identification, longitudinal multimodal imaging integrated with spatial omics, refined microphysiological platforms, and targeted modulation or delivery strategies, with the aim of translating mechanistic insights into reliable biomarkers and therapeutic opportunities for neuroinflammatory and neurovascular pathologies.

Introduction

The neurovascular unit (NVU) constitutes a multicellular functional complex comprising neurons, astrocytes, microglia, pericytes, vascular endothelial cells, and the extracellular matrix. These cellular components operate synergistically through intricate signaling networks to maintain cerebral blood flow homeostasis, blood-brain barrier (BBB) integrity, and neural microenvironmental stability [1–3]. This process ensures continuous delivery of oxygen, glucose, and other essential nutrients to meet the metabolic demands of the highly energy-dependent brain, thereby underpinning critical higher-order functions such as synaptic transmission, learning, and memory [1, 4–8].

Conventional perspectives attributed NVC regulation primarily to neuron-derived vasoactive substances (e.g., nitric oxide) and direct astrocytic endfoot interactions with vascular smooth muscle cells [9, 10]. For instance, glutamate released during neuronal activation engages metabotropic glutamate receptors on astrocytes, triggering intracellular calcium signaling and subsequent release of vasodilatory mediators such as prostaglandin E2, leading to localized vasodilation [11]. Pericytes, as principal perivascular cells enveloping capillaries, have also been implicated in fine-tuning NVC by modulating capillary diameter; their dysfunction may result in neurovascular uncoupling and cerebral hypoperfusion [12–14]. Nevertheless, the role of microglia in NVC has long been overlooked [8], with most studies relegating them to passive “bystanders” activated only under pathological conditions to participate in inflammatory responses. This conceptual bias stemmed from their canonical immune effector functions and technological limitations in capturing their dynamic physiological behaviors.

Recent advances in functional imaging have progressively unveiled the multifaceted roles of microglia in NVU physiology. Beyond their established functions in debris clearance and synaptic pruning during brain development and homeostasis, microglia exhibit extensive physical and functional interactions with the vascular system. During development, they promote angiogenesis through secreted factors like vascular endothelial growth factor (VEGF) [15–17]. In the adult brain, microglial processes dynamically surveil the perivascular microenvironment and rapidly migrate toward injured vessels under pathological conditions such as cerebral ischemia, contributing to BBB repair [18–20]. Recent studies demonstrate that microglia sense neuronal activity and vascular status via purinergic signaling, directly influencing capillary diameter and CBF [21–24]. These findings challenge the traditional view of microglia as solely immune cells and support an integrated NVU model, in which Capillary-Associated Microglia (CAMs; singular CAM) intersect rather than replace canonical neuron-astrocyte-pericyte control of NVC. Against this backdrop, this review focuses on the mechanisms by which CAMs regulate capillaries and their potential roles in vascular-related central nervous system disorders, aiming to provide novel perspectives and evidence to guide future studies on CAMs.

CAMs reside in brain parenchyma and exhibit tight associations with capillaries

Definition, markers, and morphological characteristics of CAMs

The identification of CAMs stems from their specific anatomical relationship with intracerebral capillaries and their consequent unique functional attributes. These cells detect and respond to changes in the vascular microenvironment, thereby participating in the precise regulation of capillary diameter and cerebral blood flow.

Bisht et al. first systematically characterized this population [21], demonstrating via immunoelectron microscopy and confocal imaging that the somata of CAMs reside in the parenchyma outside the vascular basement membrane, while their processes directly contact capillary walls, often at sites with limited astrocytic endfoot. This localization distinguishes CAMs from perivascular macrophages (PVMs), which occupy the perivascular space within the basement membrane and are enriched around larger vessels and meninges, rarely adhering to capillaries (Fig. 1). Through in vivo two-photon imaging in CX3CR1^CreERT2:R26-Tomato mice, Bisht et al. visualized stable perivascular attachment and dynamic process motility, coining the term “capillary-associated microglia“ [21]. Multiple lines of evidence support that CAMs are bona fide microglia [21–24]: they highly express microglia-specific markers (e.g., Iba1, CX3CR1, P2Y12R) while largely lacking classic PVMs markers such as CD206; they exhibit characteristic ramified morphology, with slightly fewer branches and larger somata than PVMs under resting conditions, though overall area remains comparable; functionally, they retain environmental surveillance capacity and rapid response to injury, extending processes toward laser-induced microlesions.

Fig. 1.

Spatial segregation of capillary-associated microglia and perivascular macrophages at the microvascular interface. Schematic showing the anatomical boundary between the brain parenchyma and the Virchow–Robin space (perivascular space, VRS). TMEM119⁺ capillary-associated microglia (CAMs) reside in the parenchyma and can position their soma in close pericapillary proximity while extending processes that appose the capillary wall. In contrast, CD206⁺ perivascular macrophages (PVMs) are localized within the VRS as PVMs, occupying the perivascular compartment adjacent to the vessel wall. Red blood cells illustrate the capillary lumen

In the normal mouse brain, CAMs account for about one-third of microglial cells, which is significantly higher than the approximately 6% expected from random contact based on vascular density. This suggests that their adhesion is not coincidental but is jointly maintained by intrinsic mechanisms and the local microenvironment [21]. CAM–capillary interactions are highly dynamic: short-term observations reveal continuous extension and retraction of processes, while long-term tracking shows most CAMs maintaining positional stability, with occasional “detachment” or new “attachment” events, yet overall proportions remain constant [21]. Thus, CAMs likely represent a reversible spatial state regulated by microenvironmental signals rather than a fixed, independent lineage. Current evidence indicates that CAMs are ubiquitously distributed across cortical, hippocampal, and thalamic regions with similar densities. Bisht et al. reported that CAMs “could be detected in the neonatal (P5) brain,” and that the density of CAMs was maintained from P15 to 12 months [21]. However, quantitative staging across late-life remains limited. Thus, CAMs likely share the canonical developmental origin of parenchymal microglia. Through direct soma-capillary contact, they constitute an important component of the neurovascular unit.

CAMs sense capillary-derived cues and exhibit directed chemotaxis

Purinergic signaling serves as the principal mechanism for CAMs to detect capillaries and guide their chemotactic recruitment [21]. Capillary endothelial cells, pericytes, and erythrocytes express Pannexin-1 (PANX1) channels, enabling release of ATP and other purinergic signals [25–27]. Microglia highly express P2Y12 receptors (P2Y12R), whose activation by ATP/ADP elevates intracellular calcium levels, driving process extension toward metabolically active regions [28–30]. Experimental evidence demonstrates that genetic ablation of microglial P2Y12R or vascular PANX1 significantly reduces the density of CAMs [21]. Thus, the coupling of ATP/ADP released by PANX1 with the P2Y12 receptor on microglia is a necessary signaling mechanism for eading to a failure in adapting perfusion to metabolic represents to capillaries.

Critically, the PANX1-P2Y12R axis mediates a “chemotaxis-contraction coupling” effect on capillary diameter and hemodynamics. Under steady-state conditions, CAMs exert a basal contractile tone on capillaries through close physical interactions, preventing excessive dilation. Depletion of microglia or pharmacological blockade of P2Y12R increases capillary diameter by approximately 15% and elevates baseline CBF by ~ 20% [21], while simultaneously impairing hypercapnia (CO₂ challenge)-induced vasodilation. Similarly, P2Y12R^−/− or PANX1^−/− mice exhibit enlarged capillary diameters, increased resting CBF, and compromised vascular responses to elevated CO₂ [21]. These findings indicate that microglia, via PANX1-P2Y12R-dependent coupling to capillaries, continuously fine-tune contractile tone to optimize luminal diameter and perfusion under resting conditions. This contractile regulation primarily influences capillary basal tone and hemodynamic responsiveness without altering structural integrity. Neither microglial ablation nor P2Y12R deficiency affects capillary density, pericyte coverage, or astrocytic endfoot ensheathment [21], indicating that microglial regulation of capillary tone is functionally mediated rather than structurally transformative. However, it is pivotal to note that P2Y12R is downregulated during microglial activation [30, 31]. Thus, under neuroinflammatory conditions, the loss of this ‘’sensory’’ mechanism may uncouple CAMs from capillary walls, leading to a failure to adapt perfusion to metabolic demands—a phenomenon termed “inflammatory neurovascular uncoupling”.

In addition to the purinergic axis, the CX3CL1-CX3CR1 signaling pathway may influence the distribution and activity of microglia at vascular sites [18, 32]. During development, microglia frequently associate with nascent vessels and contribute to vascular growth and remodeling; analogous enrichment of microglia within angiogenic niches is also evident in retinal vascular development and neovascular disease contexts [33, 34]. Experimental studies further link CX3CR1 signaling to microglial recruitment kinetics and process dynamics in the retina; CX3CR1 deficiency reduces process velocity under basal conditions and after focal laser injury and significantly slows microglial soma migration [35]. Conversely, exogenous CX3CL1 can evoke microglia-dependent vasoregulatory responses and modulate retinal capillary tone and vascular reactivity in vivo [36]. In the mature CNS, constitutive neuronal CX3CL1 is widely regarded as a homeostatic cue that helps maintain microglia in a surveillant, restrained state [37], whereas disrupted CX3CR1 signaling has been associated with heightened inflammatory reactivity and altered microglial behavior that could secondarily weaken the stability of vascular apposition [38]. Thus, the CX3CL1-CX3CR1 axis may be regarded as a “contextual signal” modulating microglial affinity and homeostatic function: adequate fractalkine signaling promotes stable vascular association and supportive roles, whereas its attenuation may bias microglia toward less stable capillary engagement and aberrant reactivity.

CAMs enhance capillary reactivity and provide basal tonic influence

CAMs potentiate capillary reactivity via CD39

CD39 is another purinergic signaling enzyme highly expressed on microglia. When microglia detect high levels of ATP signals (such as during neuronal excitation or hypercapnia), CD39/TNAP sequentially hydrolyzes ATP/ADP into AMP, which is further converted to adenosine [39]. Adenosine acts as a potent vasodilator in the brain, inducing vascular smooth muscle relaxation and vasodilation by acting on A2A/A2B adenosine receptors located on vascular smooth muscle and endothelial cells [40].

Hypercapnia serves as a classic stimulus for inducing cerebral vasodilation. Studies demonstrate that under elevated CO₂ conditions, microglia rapidly release substantial amounts of adenosine. Research by Császár et al. revealed that elevated CO₂ triggers a rapid intracellular Ca² surge in microglia, prompting the extension of vascular-adhering filopodia while concurrently stimulating robust adenosine production to facilitate vasodilation [24]. In the presence of functional microglia, mechanical whisker stimulation significantly elevates cortical adenosine concentrations, promoting capillary dilation and consequently enhancing local blood flow. Conversely, pharmacological depletion of microglia substantially reduces cerebral adenosine levels, concomitantly impairing cerebrovascular dilation responses. This impairment can be rescued by exogenous adenosine administration. These findings collectively establish microglia as pivotal contributors to cerebral adenosine production, with microglia-derived adenosine mediating protective vasodilation and negative feedback regulation. Thus, CAMs function as signal amplifiers through the CD39-adenosine pathway: during localized neuronal excitation, they rapidly generate adenosine to amplify vascular reactivity, thereby preventing cerebral hypoperfusion due to inadequate capillary dilation. Beyond converting ATP to adenosine to promote vasodilation, CD39 activity in CAMs also shapes the local immunological milieu, as adenosine signaling via A₂A and A₂B receptors can dampen microglial cytokine production and restrain excessive neuroinflammation [41].

Furthermore, the classical NO-sGC-cGMP vasodilation axis is also modulated by microglial activity. Under conditions such as hemodynamic changes, hypercapnia, or inflammation, endothelial and surrounding cells enhance NO production, thereby triggering vasodilation [42–44]. However, studies reveal that pharmacological blockade of microglial P2Y12R prevents the significant elevation of cGMP levels typically induced by hypercapnia. This observation does not imply that P2Y12R acts as a direct upstream regulator of the NO-sGC-cGMP pathway, as cGMP concentrations recover when NO is directly supplied via sodium nitroprusside. Administration of the NOS inhibitor L-NAME exacerbates the impaired vasodilatory response caused by P2Y12R blockade. These findings suggest that the CD39-adenosine axis operates in parallel with the NO-sGC-cGMP pathway, while the suppressed cGMP elevation following P2Y12R inhibition may stem from disrupted amplification of “vasodilation-demand” signals normally mediated by microglia [24].

CAMs confer basal capillary tone via COX-1

Beyond the aforementioned purinergic signaling-mediated regulation of vasoconstriction and vasodilation, CAMs also provide sustained vasodilatory tone to capillaries through the arachidonic acid metabolic pathway [23]. Specifically, microglia demonstrate high expression of cyclooxygenase-1 (COX-1), which catalyzes the production of vasoactive substances such as prostaglandin E₂ (PGE₂) [45]. Unlike the inducible COX-2 predominantly expressed in neurons and activated astrocytes, COX-1 in normal brain tissue is almost exclusively expressed by microglia transcriptomic data reveal that microglial COX-1 expression levels exceed those in astrocytes by more than tenfold, ranking highest among all cerebral cell types [23]. Consequently, microglia can be regarded as localized “production centers” for prostaglandins in the perivascular microenvironment, continuously releasing appropriate amounts of vasodilatory mediators to maintain capillary vasodilation.

Recent investigations have provided direct evidence demonstrating the crucial role of microglial COX-1 activity in modulating capillary diameter. Initial studies by Bisht et al. and Császár et al. revealed that microglial depletion via CSF1R inhibitors altered baseline capillary diameter and hemodynamic responses, though the specific mechanisms remained ambiguous due to concurrent elimination of perivascular macrophages and other cell populations. To delineate microglia-specific functions, researchers employed conditional knockout models targeting COX-1 exclusively in microglia. The results demonstrated that selective ablation of microglial COX-1 induced significant capillary constriction, reduced luminal diameter, and substantially diminished red blood cell perfusion in capillaries [23]. This phenotype closely mirrored the effects observed with complete microglial elimination or localized ablation of CAMs, effectively constituting a “phenocopy” of these interventions. These findings indicate that under physiological conditions, COX-1-derived metabolites from microglia maintain constitutive vasodilatory tone, preserving capillary patency. Loss of microglial COX-1 function consequently removes this vasodilatory support, leading to excessive capillary constriction and impaired local perfusion.

The COX-1-mediated vasodilation mechanism further elucidates why microglia are indispensable for neurovascular coupling. Prostaglandins generated by COX-1, particularly PGE₂, function as classical vasodilators by activating EP2/EP4 receptors on vascular mural cells, thereby elevating intracellular cAMP levels to induce vasodilation [46]. Additional evidence indicates that COX-1, rather than COX-2, predominates in hypercapnia-induced cerebrovascular responses [47]. This suggests that during elevated CO₂ conditions, microglia not only engage vascular structures through P2Y12R-mediated adhesion but also potentially release PGE₂ via COX-1 activation, facilitating rapid capillary dilation to enhance blood flow and clear excess CO₂. Thus, microglial COX-1 establishes sustained vasodilatory tone that maintains baseline capillary patency during resting conditions while enabling sufficient dilation during increased metabolic demands. These findings substantially expand our understanding of microglial regulation of cerebral microcirculation: beyond their established purinergic signaling mechanisms, microglia additionally modulate capillary diameter through COX-1-mediated pathways, collectively fine-tuning vascular tone to maintain cerebral hemodynamic homeostasis. However, this mechanism represents a fragile equilibrium vulnerable to inflammatory remodeling. Under pathological conditions, the metabolic balance may shift from homeostatic COX-1 toward inducible COX-2 [48], skewing the lipid profile from vasodilatory PGE₂ to vasoconstrictive prostanoids (e.g., Thromboxane A2), which drives capillary stalling [49]. Furthermore, PGE₂ exerts a dual mandate: beyond regulating vascular tone, it acts via EP receptors to modulate the inflammatory polarization of CAMs themselves, thereby inextricably coupling microcirculatory control with local immune surveillance [50–53].

It is crucial to emphasize that CAM-related signaling pathways are not antagonistic but operate synergistically within a unified purinergic framework. These three components form temporally coordinated and functionally complementary mechanisms, enabling capillaries to maintain stable resting diameters while retaining sufficient dynamic range for adaptive responses (Fig. 2). The strength of causal evidence supporting each pathway is summarized in Table 1.

Fig. 2.

Physiological recruitment and regulatory roles of capillary-associated microglia in neurovascular coupling. Capillary-associated microglia (purple) sense extracellular ATP signals in the microenvironment (particularly derived from the vasculature (red) and neurons (pink)) via P2Y12R, extend their processes to tightly appose the capillary surface at sites with relatively sparse astrocytic(green) coverage, and regulate capillary caliber and local blood flow through the COX-1–mediated production of prostanoids (e.g., PGE2) and a CD39/adenosine axis driven by an extracellular nucleotide–metabolic cascade initiated by CD39. CAMs may also indirectly adjust vessel caliber by interacting with EP1 and EP4 receptors on pericytes(yellow). Those processes may be coordinated by CX3CL1–CX3CR1 and Piezo1 signaling. Solid arrows indicate the predominant direction of signaling or mediator flow, and dashed arrows indicate intracellular coupling within CAMs. ATP, adenosine triphosphate; CAMs, capillary-associated microglia; COX-1, cyclooxygenase-1; PGE2, prostaglandin E2; EP1/EP4, prostaglandin E receptor 1/4; CX3CL1, C-X3-C motif chemokine ligand 1; CX3CR1, CX3C chemokine receptor 1; PANX1, pannexin-1; P2Y12R, purinergic receptor P2Y12

Table 1.

Evidence map for CAM-associated mechanisms regulating capillary tone and reactivity

Mechanism	Reference	Intervention type	Evidence tier*
PANX1-P2Y12R purinergic coupling	Ref [21]	Germline knockout; pharmacological depletion and repopulation	B
Microglial CD39 purinergic metabolism	Ref [22]	Microglia-enriched conditional knockout; genetic microglia ablation; pharmacological inhibition; chemogenetic manipulation	A
Microglial COX-1 prostanoid signaling	Ref [23]	Microglia-enriched conditional knockout; focal ablation of CAMs; pharmacological depletion	A
P2Y12R-dependent purinergic microglial actions	Ref [24]	Pharmacological depletion; germline knockout; pharmacological receptor antagonism; chemogenetic activation	B
CX3CL1-CX3CR1 signaling	Ref [38]	Ligand stimulation; germline receptor disruption; pharmacological receptor antagonism; disease induction	B
Microglial Piezo1 mechanosensation	Ref [54]	Microglia-enriched conditional knockout; pharmacological activation/inhibition; disease model	C

Evidence strength scale: A = in vivo causal evidence with microglia-enriched genetic manipulation and vascular readouts; B = in vivo or ex vivo functional evidence with depletion, repopulation, germline genetics, or pharmacology; C = mechanistic evidence in disease context without direct causal readouts linking CAMs to vascular function

CAMs perceive hemodynamic changes via Piezo1

Microglia possess the capacity to detect mechanical stimuli. During hemodynamic or tissue mechanical alterations, microglia may respond through mechanosensitive ion channels such as Piezo1 [55]. Emerging evidence indicates that microglia can “sense” changes in environmental stiffness or pressure. Capillary pulsation, shear stress, or variations in brain tissue rigidity are likely detected by microglia via mechanosensitive channels (e.g., Piezo1), thereby influencing calcium influx and functional states [54, 56–59]. For instance, capillary flow pulsations and distension may induce subtle morphological adaptations in adherent microglia, initiating intracellular signaling cascades. This mechanotransduction potentially prompts microglia to modulate their adhesion dynamics or release vasoactive factors (dilators/constrictors) to adapt to fluctuating hemodynamic conditions [60, 61]. Although research on mechanosensory regulation of microglia remains nascent, existing studies demonstrate that microglia alter their phenotype and metabolic activity under fluid shear stress in vitro [58, 62]. Thus, it is plausible that CAMs utilize mechanosensitive mechanisms to continuously monitor vascular physical states and initiate feedback responses when necessary, ensuring capillary adaptation to dynamic blood flow variations.

However, this mechanosensation may also serve as a double-edged sword. In pathological contexts characterized by vascular stiffening (e.g., arteriosclerosis or amyloid angiopathy), Piezo1 may become hyperactivated. Recent studies reveal that excessive Piezo1 signaling can trigger calcium-dependent potassium efflux (via KCNN4), which acts as a potent activator of the NLRP3 inflammasome [63, 64]. Accordingly, CAMs may couple abnormal vascular stiffness to inflammatory signaling, rather than simply adapting to mechanical forces. This raises the possibility of a “mechano-inflammatory” route by which chronic vascular stiffening could contribute to neuroinflammation, although its in vivo relevance and sufficiency remain to be established.

Intercellular interactions between CAMs and other NVU components

Beyond their direct regulatory effects on vascular endothelial cells, CAMs engage in dynamic crosstalk with NVU components including astrocytes and pericytes [19, 65]. Pericytes, as primary regulators of capillary tone, interact intricately with CAMs to modulate capillary diameter and cerebral perfusion [12–14]. Microglia regulate pericyte contractility through soluble mediators and extracellular vesicles (EVs) [19, 66]. For instance, activated microglia-derived PGE₂ downregulates N-cadherin and connexin 43 expression in pericytes via EP1/EP4 receptors, disrupting pericyte-endothelial junctions and diminishing contractile capacity to induce capillary dilation [67]. Conversely, under homeostatic conditions, COX-1-derived prostaglandins from microglia may exert tonic influences on pericytes to maintain baseline capillary tension [23]. In pathological contexts such as Alzheimer’s disease, Aβ deposition activates microglia to release inflammatory cytokines, disrupting pericyte calcium homeostasis and inducing apoptosis, ultimately compromising capillary integrity and hemodynamic regulation [68].

Astrocytes represent classical regulators of NVC [11], while their interplay with CAMs enables coordinated cerebral blood flow modulation. CAMs frequently localize to capillary segments lacking astrocytic endfoot coverage [21]. In these regions of diminished astrocytic influence, microglia may assume a more direct regulatory role. For example, when neuronal activity increases, astrocytes typically release substances such as PGE₂ and ATP through Ca²⁺ fluctuations, leading to the dilation of adjacent blood vessels [69]; however, capillary segments devoid of astrocytic ensheathment may rely on microglial sensing of neuronal activity to modulate vascular tone. Furthermore, CAM-astrocyte interactions extend to regulating perivascular space (PVS) clearance function [19]. While astrocytic endfeet delineate PVS boundaries, microglial processes infiltrate these spaces, phagocytosing metabolic waste (e.g., Aβ) [65, 70, 71]. Their synergistic activities maintain PVS hydrodynamic properties, ensuring efficient waste clearance. During sleep or anesthesia, enhanced microglial phagocytic activity coincides with polarized AQP4 redistribution in astrocytic endfeet, collectively accelerating metabolite removal [70, 72, 73]. This spatial division of labor enables the NVU to maintain precise hemodynamic control across diverse structural contexts.

Conceptually, this architecture is consistent with a tiered model of NVC. Neuronal and astrocytic pathways provide fast initiating signals, pericytes implement capillary diameter changes as the primary contractile effectors, and CAMs primarily tune the set-point and gain of capillary responsiveness through P2Y12R-dependent nucleotide sensing, CD39-driven extracellular nucleotide metabolism coupled to adenosine-mediated dilation, and COX-1-derived prostanoid tone that stabilizes basal capillary patency [21–24]. In this framework, CAMs act mainly as modulators during routine physiological coupling by shaping baseline tone and sensitivity rather than initiating responses. They may shift toward a more driver-like role when canonical astrocytic endfoot control is locally attenuated or during global metabolic challenges such as hypercapnia, when purinergic, adenosinergic, and prostanoid programs become disproportionately important for sustaining microcirculatory perfusion [21–24]. Accordingly, CAMs constitute an immunovascular interface through which systemic inflammatory states can reset local NVC gain and microcirculatory resilience.

CAMs as potential regulators in vascular-associated pathologies

CAMs as therapeutic targets for neurodegenerative diseases

Neurodegenerative diseases, such as Alzheimer’s disease (AD) and Parkinson’s disease (PD), are principally characterized by progressive neuronal loss, aberrant activation of glial cells, and dysfunction of the neurovascular unit [74–76], often accompanied by persistent neuroinflammation at the neurovascular interface. These pathologies are frequently concomitant with cerebral hypoperfusion, capillary degeneration, and compromise of the BBB, thereby establishing microcirculatory dysfunction as a pivotal driver of disease pathogenesis and progression and a feed-forward amplifier of inflammatory stress. Functioning as ‘’sentinels’’ ensheathing the capillaries, CAMs assume a critical role. Aberrations in their structure and function may precede the manifestation of clinical symptoms and contribute to multiple facets of the disease, including the amplification of inflammatory responses, BBB disruption, impairment of cerebral blood flow regulation, and deficits in the clearance of pathological proteins, thereby reinforcing neurodegenerative cascades (Fig. 3).

Fig. 3.

Capillary-associated microglia as potential regulators across vascular-related brain diseases. This conceptual schematic summarizes how capillary-associated microglia (CAMs) may engage distinct signaling modules at the capillary interface across four disease contexts. In neurodegenerative diseases, pathological protein accumulation (Aβ/α-syn) is associated with reduced P2Y12R signaling and impaired clearance of Aβ/α-syn, together with increased inflammasome activation (NLRP3↑), reduced vasoactive mediators (CO/adenosine↓), impaired neurovascular coupling (NVC↓), blood–brain barrier leakage (BBB leak↑), and decreased cerebral blood flow (CBF↓). In cerebral hypoperfusion-related disorders, purinergic cues (ATP) and vasoactive signaling (adenosine/CO) are linked to dilation responses, whereas vascular damage and inflammatory mediators (IL-1β/TNF-α) coincide with immune-cell recruitment, lipid droplet accumulation, BBB dysfunction, and reduced microvessel density (↓). In infectious neurovascular diseases, pathogen-associated molecular patterns (PAMPs) and chemokines (e.g., CCL5) activate innate immune pathways (TLRs↑; cGAS–STING↑), elevating inflammatory effectors (IL-1β, TNF-α, ROS) that contribute to BBB leakage and impaired autoregulation. In metabolic neurovascular diseases, hyperglycemia/hyperlipidemia and AngII–AT1 signaling (↑), together with CX3CL1-related cues, converge on inflammatory outputs (IL-1β/TNF-α/ROS) and are associated with BBB leakage, capillary constriction, synapse loss, and white-matter injury. Solid arrows indicate the proposed direction of influence/association among stimuli, CAM-linked pathways, and vascular outcomes; upward and downward arrows denote relative increases or decreases as depicted. Aβ, amyloid-β; α-syn, alpha-synuclein; AngII, angiotensin II; AT1, angiotensin II type 1 receptor; ATP, adenosine triphosphate; BBB, blood–brain barrier; CBF, cerebral blood flow; CCL5, C-C motif chemokine ligand 5; cGAS–STING, cyclic GMP–AMP synthase–stimulator of interferon genes pathway; CO, carbon monoxide; CX3CL1, C-X3-C motif chemokine ligand 1; IL-1β, interleukin-1 beta; NLRP3, NOD-like receptor family pyrin domain-containing 3; NVC, neurovascular coupling; P2Y12R, purinergic receptor P2Y12; PAMPs, pathogen-associated molecular patterns; ROS, reactive oxygen species; TLRs, Toll-like receptors; TNF-α, tumor necrosis factor alpha

Alzheimer’s disease

AD is not merely a quintessential neurodegenerative disease-marked by the deposition of amyloid-β (Aβ), neurofibrillary tangles, and synaptic and neuronal loss-but is increasingly recognized as a convergent “neurovascular” pathology [74, 77–79], in which microcirculatory failure and chronic neuroinflammatory stress frequently co-evolve. Prior to the onset of clinical symptoms, a substantial proportion of AD patients exhibit diminished CBF and blunted or lost NVC responses, indicating that vascular and microcirculatory impairment constitutes a seminal event in early AD pathogenesis [2, 7, 74]. Numerous neuroimaging and hemodynamic studies have documented reduced perfusion or attenuated hemodynamic responses in typically vulnerable brain regions-such as the frontal lobe, cingulate cortex, and precuneus-during the stage of mild cognitive impairment (MCI) or early AD [75]. Cerebral amyloid angiopathy (CAA), a common cerebrovascular alteration in AD, is characterized by the deposition of Aβ within capillary and arteriolar walls, culminating in thickened, fragile vasculature and compromised perfusion, thereby creating a permissive substrate for sustained perivascular immune activation [80–83].

As resident immune sentinels of the brain, microglia under physiological conditions rapidly respond to neuronal activity and metabolic shifts, engaging in phagocytic clearance of Aβ oligomers and deposits, eliminating cellular debris, and supporting homeostatic functions of the neurovascular unit [21]. Ablation of microglia exacerbates cerebrovascular pathology in AD models, leading to increased vascular hemorrhage and augmented Aβ deposition within vessel walls [84]. However, during AD progression, microglia manifest significant abnormalities, including numerical depletion, functional impairment, and phenotypic dysregulation. On one hand, a marked reduction in microglial density is observed in AD brains, particularly in regions vulnerable to ischemic-hypoxic stress such as the hippocampus [65]. On the other hand, microglia associated with Aβ plaques in human AD brains frequently exhibit a loss of P2Y12R expression, adopting an activated phenotype [85]. Normally, microglia highly express the P2Y12R, enabling them to sense neuronally-released ATP/ADP and thereby modulate capillary diameter and local CBF [22]. However, plaque-associated CAMs exhibit a marked loss of P2Y12R, a phenotype consistent with chronic inflammatory signaling and disease-associated microglial reprogramming, potentially involving inflammasome-linked pathways [86–88]. This loss may render CAMs less responsive to purinergic cues linked to vascular dynamics, functionally weakening the PANX1-P2Y12R sensing module. Consequently, their capacity to maintain blood flow and clear pathological aggregates may be undermined. Furthermore, the enzymatic machinery of CAMs undergoes pathological remodeling. While COX-1 inhibition has shown therapeutic potential [89], evidence suggests that the inflammatory milieu may promote a “metabolic switch” from homeostatic COX-1 activity toward inducible COX-2 and a more constrictive/pro-inflammatory prostanoid spectrum, including Thromboxane A2 production [90, 91]. This shift not only withdraws basal vasodilatory support but may actively drive capillary constriction and stalling, thereby mechanistically coupling lipid remodeling to microvascular inflammation and flow failure. Thus, the alteration of COX pathways in AD represents a dual failure of vasoregulation and immune resolution.

These alterations not only impair microcirculatory perfusion but also disrupt the coupling between cerebral metabolism and blood flow, leading to insufficient energy supply to neurons and thereby accelerating the neurodegenerative cascade [92], while simultaneously reinforcing perivascular inflammatory stress within the NVU. Consequently, CAMs represent a potential target for intervention. However, effective therapeutic strategies must account for the temporal evolution of this dysfunction. We propose that CAMs participate in a stage-dependent “neurovascular vicious cycle”: in the early phase, amyloid-associated vascular stress and chronic inflammatory cues may promote the loss of P2Y12R expression in CAMs, establishing a reciprocal loop in which impaired purinergic sensing worsens local hypoperfusion and Aβ clearance, thereby amplifying vascular stress and amyloid burden [74, 93, 94]. As the disease advances, the aforementioned COX-pathway remodeling may superimpose active vasoconstrictive/pro-inflammatory prostanoid outputs (e.g., COX-2/TXA2-skewed signaling) upon this sensory failure, further driving capillary constriction and stalling. Thus, rather than broadly suppressing microglia, future interventions could be staged-aiming to restore “sensory” (P2Y12R) homeostasis early to decouple reciprocal injury, while targeting maladaptive prostanoid remodeling in later stages.

To make such stage-tailored interventions actionable, translational studies will need biomarkers that reflect microcirculatory perfusion and the evolving myeloid state in the living brain. Perfusion MRI, including arterial spin labeling, can quantify cerebral blood flow and flow heterogeneity [95, 96], and contrast-based permeability imaging such as dynamic contrast-enhanced MRI can inform blood-brain barrier disruption as a vascular readout aligned with neurovascular unit stress [97–99]. Microglia-oriented PET can provide pharmacodynamic evidence of target engagement, with TSPO as an established starting point and an expanding toolkit of non-TSPO immune targets [100–102]. Representative candidates include cerebrospinal fluid soluble TREM2 as a readout of microglial activation, cerebrospinal fluid soluble PDGFRβ as an index of pericyte injury and barrier breakdown, and blood-based GFAP and neurofilament light as complementary markers of astroglial and axonal injury [103–106]. Systemic inflammatory markers and routine coagulation panels can be co-collected to flag peripheral confounding and safety liabilities.

Parkinson’s disease

The core pathological hallmarks of PD are the progressive loss of dopaminergic neurons in the substantia nigra pars compacta and the formation of Lewy bodies, primarily composed of α-synuclein [107, 108]. While traditionally conceptualized as a predominantly neurodegenerative disorder, accumulating evidence underscores cerebral small vessel disease and microcirculatory dysfunction as integral components of its pathological progression [109, 110], often alongside sustained glial activation and neuroinflammatory stress at the neurovascular interface [111, 112]. Alterations in microvessel density, capillary wall thickening, and increased BBB permeability have been observed in both PD patients and animal models [113], with these vascular changes being closely correlated with neurological deterioration, suggesting that microvascular injury can amplify neuronal vulnerability rather than serving as a passive epiphenomenon.

Microglia are posited to play a pivotal role. Beyond the conventional paradigm wherein activated microglia release inflammatory cytokines [114], directly damaging vascular endothelial tight junctions and disrupting the BBB [115], microglia positioned at capillaries may act as a gatekeeper of “immunovascular” homeostasis, and their vasoregulatory functions appear to be compromised in PD. Studies utilizing PD animal models compounded by metabolic stress have revealed a significant reduction in perivascular microglia in the striatum, concomitant with decreased pericyte density and impaired angiogenic responses, suggesting deficient microvascular remodeling [113]. This attenuation of microglia-vessel interaction not only diminishes local microcirculatory adaptability but also exacerbates ischemic and hypoxic conditions in the substant nigra [116, 117], thereby propelling the degeneration of dopaminergic neurons, and may further facilitate a feed-forward loop in which hypoxia and BBB leakage perpetuate microglial inflammatory priming. Restoring or enhancing the surveillance and supportive functions of microglia, specifically CAMs, toward capillaries may offer a therapeutic avenue to improve regional cerebral blood flow, alleviate ischemic stress, and thus decelerate dopaminergic neuron loss and disease progression, potentially without indiscriminately suppressing microglial protective programs.

CAMs as therapeutic targets in cerebral hypoperfusion pathologies

Cerebral hypoperfusion denotes a state in which CBF falls below the level required to sustain normal neurological function [118]. In both acute cerebral perfusion deficits, such as ischemia/reperfusion injury, and chronic microcirculatory insufficiency, including cerebral small vessel disease and vascular dementia, the dynamic interplay between microglia and the vasculature becomes particularly prominent [29, 119, 120], because perfusion failure and sterile inflammation tend to co-escalate within the NVU [121, 122]. Their functional impairment is not only a consequence of diminished perfusion but may also constitute a significant factor exacerbating cerebral injury, by destabilizing capillary tone control and amplifying inflammatory remodeling of the microvascular niche [123–125]. Consequently, CAMs represent a promising potential target for strategies aimed at improving cerebral perfusion and treating associated disorders, particularly those in which neuroinflammation and microcirculatory dysfunction are mechanistically intertwined.

Acute cerebral perfusion disturbances

Acute ischemia/reperfusion (I/R) injury, as exemplified in acute ischemic stroke, triggers a rapid response from cerebral immune cells. Among these, microglia represent the earliest activated resident immune cells within the central nervous system [126], positioning them as a first-line “immunovascular” sensor at the microcirculatory interface. Within hours of stroke onset, they undergo a morphological transformation from a resting ramified state to an amoeboid or bushy appearance, swiftly secreting pro-inflammatory cytokines such as IL-1β and TNF-α, releasing reactive oxygen species, and initiating phagocytosis of necrotic cells and blood components [127–129], including erythrocyte-derived danger signals that are highly relevant to capillary-adjacent microglial surveillance [130–133]. Significant elevations of IL-1α/β and TNF-α within the ischemic core are detectable as early as 3 h post-ictus, whereas infiltration of peripheral neutrophils and monocytes typically manifests after 24 h, indicating the primacy of microglia in driving the initial inflammatory cascade [134, 135], which can secondarily remodel endothelial activation status and microvascular patency [136, 137]. Although these responses facilitate the clearance of necrotic debris, they concurrently exacerbate blood-brain barrier disruption; furthermore, released iron and heme products can even induce cellular apoptosis via the TLR4 pathway, amplifying tissue damage [138], thereby providing a mechanistic bridge between early innate immune signaling and downstream microcirculatory failure.

The rapid microglial response is intimately linked to their P2Y12 receptors [139]. During the early ischemic phase, extracellular ATP efflux is sensitively detected by microglia via P2Y12R, inducing rapid process extension and convergence towards the ischemic area [29, 140]. This process serves a dual function: it prompts microglia to release neurotrophic factors like BDNF and NGF, exerting neuroprotective effects [141, 142], while also enabling the release of vasoactive molecules such as prostaglandin E2 and adenosine, which modulate capillary tone and contribute to neurovascular coupling, thereby partially facilitating reperfusion, suggesting that early microglial activation can transiently align vasoregulation with tissue rescue. However, this protective window is likely transient. Under severe or prolonged ischemia, the role of CAMs appears to shift from compensatory support to maladaptive injury. They polarize towards a pro-inflammatory phenotype, generating copious amounts of cytokines, chemokines, and ROS/RNS, ultimately aggravating injury [119, 142, 143], and potentially undermining vasoregulatory programs by favoring inflammatory remodeling over capillary support.

The “no-reflow” phenomenon, primarily caused by capillary occlusion, constitutes a critical mechanism in acute perfusion failure during reperfusion [144]. On one hand, ischemic endothelial cells and vascular smooth muscle cells release endothelin-1 (ET-1), inducing sustained pericyte contraction which impedes the restoration of capillary luminal patency [145, 146]; On the other hand, post-reperfusion inflammation and thrombotic activation promote neutrophil adherence within the microvasculature, with two-photon imaging revealing that approximately 20–30% of capillaries in the ischemic region become obstructed by this mechanism [147]. The aggregation of microglia around capillaries is implicated in this process through several mechanisms. Firstly, activated microglia during I/R release a plethora of inflammatory mediators including IL-1β and TNF-α, which upregulate endothelial adhesion molecules and recruit neutrophils that subsequently embolize the microvasculature, exacerbating the formation of non-perfused areas [148]. Secondly, reactive oxygen species and proteolytic enzymes (e.g., MMP-9) produced by microglia degrade the microvascular basal lamina and tight junctions, leading to vascular wall edema, increased leakage, and elevated microcirculatory resistance [18], thereby converting inflammatory signaling into a biophysical barrier to reperfusion. Thirdly, hyperactivated microglia may physically impede capillary blood flow either by closely apposing capillary walls or through excessive interactions with pericytes [65], highlighting a potential “immuno-mechanical” component of capillary stalling during acute neuroinflammation.

This dual-edged nature of CAMs positions them as a potential therapeutic target, particularly in settings where acute neuroinflammation and microcirculatory failure co-emerge. On one hand, appropriately modulating their acute inflammatory response may mitigate I/R injury. For instance, P2RY12 antagonists (e.g., clopidogrel) or genetic deletion can reduce excessive microglial aggregation and neuronal “collateral damage“ [140]; inhibiting the TLR4-NFκB signaling pathway alleviates vasospasm and secondary injury following subarachnoid hemorrhage [149]; minocycline reduces edema and functional deficits in intracerebral hemorrhage models by suppressing microglial activation [150, 151]. On the other hand, preserving and harnessing the beneficial functions of CAMs is equally crucial. Adenosine released by microglia subsequently promotes prostaglandin-mediated vasodilation [22, 23]; HO-1-mediated heme catabolism and its metabolic byproduct carbon monoxide (CO) facilitate vasodilation and attenuate vasospasm. Microglial deficiency in HO-1 impairs hematoma resolution and exacerbates delayed pathological outcomes [152]. However, translating systemic P2Y12 inhibition is constrained by hemostatic off-target effects because platelet P2Y12 is essential for aggregation, creating a clinically meaningful bleeding liability and complicating attribution of platelet- versus CAM-driven effects in vivo [153]. Moreover, platelet P2Y12 signaling participates in thrombo-inflammatory crosstalk, such that peripheral engagement may exert broader immunomodulatory effects beyond the CNS [154]. Defining the precise temporal switch from the protective P2Y12R-mediated phase to the deleterious inflammatory phase will be crucial for determining the therapeutic window.

Chronic microcirculatory insufficiency

Sustained cerebral hypoperfusion can precipitate impaired neuronal energy metabolism, heightened oxidative stress, activation of neuroinflammation, and disruption of NVU structure and function, ultimately culminating in cognitive decline and dementia [120, 155, 156], with chronic inflammatory remodeling progressively constraining capillary reserve and perfusion adaptability [53]. Consequently, CAMs represent a promising potential target for strategies aimed at improving cerebral perfusion and treating associated disorders, especially those characterized by sustained neuroinflammation at the NVU.

In chronic perfusion deficits such as cerebral small vessel disease (CSVD), CAMs also exhibit alterations in their spatial distribution and phenotypic profile, potentially influencing disease progression, thereby providing a cellular substrate through which chronic microvascular stress may be translated into persistent immune remodeling at capillaries [24, 148]. In hypertension-related arteriolosclerotic CSVD, microvessels are subjected to prolonged high perfusion pressure and low-grade inflammatory stimulation, gradually developing wall thickening, luminal stenosis, and BBB disruption [148]. Postmortem studies comparing brain tissue from CSVD patients and age-matched controls have revealed significant morphological alterations in cortical microglia of CSVD patients: enlarged somata, thickened and shortened processes, and an overall reduction in branching complexity [148]. Particularly noteworthy is the marked increase in the density of microglia directly apposed to capillaries per unit area in the cortex of CSVD patients [148, 157], suggesting that under chronic ischemic and hypoxic conditions, microglia exhibit a heightened propensity to perivascular accumulation, likely in response to stimuli such as hypoperfusion and BBB leakage. Concurrently, dynamic shifts in microglial surface receptor expression were observed: in early hypertension, certain microglial subpopulations paradoxically upregulated markers associated with a mildly activated/surveillant state, such as the P2Y12 receptor and CD200R [148], potentially reflecting an attempt to preserve purinergic sensing and immune restraint at stressed capillary segments. However, in the advanced chronic hypertension stage, microglia manifested a more classic inflammatory phenotype [158]. This progression strongly suggests a “compensation-to-decompensation” trajectory: CAMs initially increase coverage to stabilize perfusion but eventually succumb to chronic stress, transitioning towards a decompensated, inflammatory state that exacerbates ischemic injury.

Microglia-mediated chronic inflammation is recognized as a significant contributor to brain injury and cognitive impairment in CSVD. As previously discussed, the release of pro-inflammatory factors by microglia directly damages vascular endothelium, maintaining BBB in a state of chronic hyperpermeability. This leads to white matter edema and inflammatory cell infiltration. Furthermore, through the secretion of chemokines, microglia recruit peripheral monocytes that cross the compromised BBB into the brain parenchyma. These monocytes subsequently differentiate into macrophages, releasing additional inflammatory mediators and establishing a self-perpetuating cycle of chronic inflammation [18]. This cascade ultimately results in neuronal loss and impairs synaptic plasticity and learning-memory functions [116, 159]. Concurrently, microglia engage in phagocytosis of myelin debris, leading to substantial intracellular lipid accumulation. An increase in “lipid-laden microglia” has been observed in the white matter lesions of CSVD, hypothesized to be one of the drivers underlying white matter pathology [160]. These lipid-burdened microglia become functionally impaired, further compromising their capacity to clear debris and provide axonal support. This dysfunction contributes to the progression of white matter hyperintensities and cognitive decline [148, 160].

In vascular cognitive impairment (VCI), cerebral hypoperfusion serves as the primary pathological substrate. Therefore, aberrant activation of CAMs may represent a critical link connecting cerebral ischemia with cognitive dysfunction [141, 161, 162], by coupling impaired capillary perfusion adaptability to a persistent, chronic inflammatory microenvironment.

In light of this, anti-inflammatory strategies and metabolic modulation have emerged as important interventional approaches targeting CAMs in CSVD. Preclinical studies demonstrate that anti-inflammatory treatments in animal models can shift microglial phenotype, reduce capillary damage, and slow the progression of hypertensive CSVD-related brain injury [163]. Clinically, there is preliminary evidence supporting interventions aimed at the chronic inflammatory state associated with CSVD. For instance, one study found that the level of neuroinflammation in CSVD patients negatively correlates with their cognitive function and lesion burden [164]. Aggressive management of cardiovascular risk factors, such as hypertension and hyperglycemia, and the use of statins, which possess ancillary anti-inflammatory properties, may indirectly mitigate excessive microglial activation, thereby potentially slowing CSVD progression [148], and may be particularly relevant for limiting inflammatory remodeling at the capillary niche.

CAMs as therapeutic targets in metabolic neurovascular disorders

Metabolic diseases are frequently accompanied by CNS complications, among which neurovascular pathologies represent a significant factor contributing to disease progression and poor prognosis [165–167]. Metabolic disorders such as diabetes can induce structural and functional abnormalities in the cerebral vasculature through chronic hyperglycemia, insulin resistance, and associated metabolic dysregulation [168]. These systemic metabolic insults are typically coupled to persistent low-grade inflammation and oxidative stress, which may activate microglia at the microvascular interface and thereby lower the threshold for NVU dysfunction.

Diabetic encephalopathy, a common neurological complication of diabetes, is characterized by reduced cerebral blood flow, BBB disruption, neuroinflammation, and cognitive decline [169, 170]. These changes are associated with downregulation of BBB tight junction proteins, increased permeability, regional perfusion deficits, and ultimately, cognitive impairment [18, 171–173]. Concurrently, elevated circulating angiotensin II (AngII) levels act upon AT1 receptors expressed by cerebral microglia, promoting their perivascular accumulation and activation, a phenomenon characterized by an increase in so-called vasculature-associated microglia (VAM) [169]. Acute AngII administration replicated this phenomenon, whereas AT1a receptor blockade reversed the abnormal VAM clustering and inflammatory activation in diabetic rats, concomitantly improving cognitive function [169]. This AngII-AT1R axis may therefore represent a mechanistic bridge through which systemic metabolic inflammation primes CAMs and increases NVU vulnerability.

Recent research has identified a critical role for CAM-like retinal microglia in the early pathophysiology of DR. Under physiological conditions, these CAM-like retinal microglia help maintain normal retinal microvascular structure and function via CX3CR1 receptor binding to CX3CL1 secreted by neurons and vascular endothelial cells [36, 174]. Studies indicate that CAM-like retinal microglia can induce local capillary constriction through the fractalkine-CX3CR1 signaling axis, a process dependent on activation of the RAS. AngII then acts on AT1 receptors on vascular endothelial cells and pericytes, inducing vasoconstriction [36, 174]. In the diabetic state, hyperglycemia-induced oxidative stress in retinal neurons and endothelial cells upregulates fractalkine expression, leading to excessive activation of microglial CX3CR1 receptors [36, 175]. Simultaneously, hyperglycemia directly activates the microglia-intrinsic RAS program, increasing AngII production and upregulating AT1 receptor expression. This hyperactivated microglia-RAS axis causes sustained retinal capillary constriction, reduced lumen diameter, and decreased retinal blood flow, ultimately resulting in retinal ischemia and hypoxia [36, 174]. In streptozotocin-induced diabetic rat models, early stages are marked by observable perivascular clustering and activation of capillary-apposed retinal microglia, the extent of which correlates significantly with diminished capillary diameter and reduced retinal blood flow [36, 176]. Administration of AT1 receptor antagonists effectively suppresses abnormal microglia activation and vasoconstriction in diabetic rat retinas, partially restoring retinal blood flow; however, overall hemodynamic improvement may be limited due to potential maldistribution of flow consequent to larger vessel dilation [36]. These findings provide evidence for “microglia-mediated vascular regulation” and offer a comparable mechanistic blueprint for extrapolating the role of microglial inflammatory signaling in DR.

Chronic high-fat diets and elevated cholesterol levels not only promote atherosclerosis but also contribute to cerebral capillary degeneration and cognitive impairment [177–179]. Under hyperlipidemic conditions, microglia readily internalize excess lipids, transforming into “foam cells“ [180]. These lipid-laden cells exhibit impaired functionality in clearing myelin debris and supporting neuronal metabolism, while simultaneously sustaining the release of inflammatory mediators and oxidative products, potentially driving white matter pathology and cognitive deficits [160]. In rats fed a high-fat diet, researchers observed significant microglial activation in cerebral white matter regions, accompanied by intensified perivascular inflammation [164]. Furthermore, hyperlipidemia often coexists with insulin resistance and elevated inflammatory cytokines, all of which impact microglia through the blood-brain barrier, maintaining them in a state of chronic stress [181–183]. High-fat diets markedly exacerbate microvascular pathology in Parkinson’s disease models, manifesting as reduced microglia-capillary interactions, pericyte loss, and impaired vascular remodeling [113]. This suggests that hyperlipidemia synergistically worsens neurovascular pathology in other diseases by amplifying microglia-mediated inflammation. This locks CAMs into a persistent vasoconstrictive and pro-inflammatory state, thereby converting a potentially adaptive response into a driver of microvascular rarefaction. This mechanistic insight positions CAMs as a convergent therapeutic node for strategies designed to restore microcirculatory stability while tempering neuroinflammation.

CAMs as therapeutic targets in infectious neurovascular diseases

Infectious diseases are similarly associated with CNS complications, which can compromise NVU integrity through direct invasion or by inducing systemic and local inflammatory responses [184]. In this setting, CAMs represent a strategically positioned innate immune sensor that can translate circulating inflammatory cues into microvascular consequences.

Infectious conditions such as viral encephalitis, sepsis-associated encephalopathy, and bacterial meningitis frequently induce NVU dysfunction by disrupting BBB and interfering with cerebral blood flow regulation [185, 186]. Viral pathogens can directly traverse the BBB through degradation of tight junction proteins or via “Trojan horse” mechanisms, leading to plasma component leakage, vascular endothelial damage, and cerebral edema [185]. Concurrently, systemic infections like sepsis trigger endothelial dysfunction and dysregulation of cerebral circulation dynamics, with nearly half of sepsis patients exhibiting impaired cerebral autoregulation [186]. These structural and functional alterations ultimately result in inadequate oxygen delivery, neuronal injury, and cognitive deficits [187, 188].

As resident immune sentinels of the CNS, microglia possess sophisticated pathogen recognition capabilities [189, 190]. During early infection, microglia recognize pathogen-associated molecular patterns (PAMPs) through various pattern recognition receptors expressed on their surface and intracellularly, including TLRs and cGAS [191, 192]. For instance, DNA viruses such as herpes simplex virus (HSV) activate the cGAS-STING pathway within microglia, inducing type I interferon production and initiating antiviral immune responses [192]. Microglia respond to CCL5 by migrating and adhering tightly to the cerebrovascular endothelium, expressing significant amounts of claudin-5 to maintain BBB integrity [186]. STING activation in microglia not only mediates antiviral immunity but also plays a significant role in maintaining vascular homeostasis: experimental studies demonstrate that STING knockout leads to more severe cerebral edema, microhemorrhages, and neurological deficits [192, 193]. However, because Pattern Recognition Receptor (PRR) signaling can also propagate NF-κB-driven cytokine programs [194], clarifying when CAMs’ responses transition from vasculoprotective immune surveillance to maladaptive endothelial activation remains an important neuroinflammation-focused question.

Animal studies provide further evidence supporting microglial influence on cerebrovascular damage. Depletion of microglia using CSF-1R inhibitors or genetic knockout approaches significantly exacerbates cerebrovascular injury in various viral encephalitis models and septic encephalopathy [189]. These investigations reveal that microglial ablation results in increased viral load and mortality, frequently accompanied by exacerbated BBB disruption, microhemorrhages, and neuroinflammation [189, 195]. This underscores the protective function of microglia toward cerebral vasculature in infectious brain disorders.

It should be noted that direct evidence demonstrating microglial capacity to ameliorate vascular damage through modifying vascular tone or repairing endothelial function remains lacking. Although studies report microglial accumulation at vascular walls during infection and inflammation [190], their precise role in direct vasomotor regulation remains unclear. A further gap is whether vasoactive programs in CAMs are engaged or instead overridden during infection-driven cytokine storms and endothelial dysfunction. As the first immune defense line in the brain, CAMs exhibit potential vasculoprotective effects, yet their overall impact is likely biphasic. In the initial phase of invasion, PAMP-mediated recruitment of CAMs may act as a rapid ‘’immunological seal’’ to preserve BBB integrity and limit pathogen entry. However, a critical tipping point may exist: if the infection triggers an overwhelming cytokine storm (e.g., in severe sepsis), this protective surveillance may collapse. At this stage, hyperactivated CAMs may exacerbate endothelial injury and barrier leakage, shifting from ‘’vascular guardians’’ to effectors of collateral damage. Future research elucidating the mechanisms governing this protective-to-deleterious switch may yield novel therapeutic targets for intervening in infection-related neurovascular pathologies.

Conclusion

CAMs, functioning as ‘’sentinels’’ of the neurovascular unit, are increasingly recognized for their critical importance in maintaining brain homeostasis and contributing to disease pathogenesis. Future research should place greater emphasis on integrated analyses of CAMs within the broader NVU context and on translating fundamental discoveries into clinical applications, ultimately achieving the goal of preserving brain health and mitigating or curing neurovascular diseases through targeted modulation of CAMs’ function.

Authors’ contributions

H.L., Y.Z., and C.C. contributed equally to this work. H.L. and Y.Z. performed the literature search, drafted the original manuscript, and prepared the figures. C.C. performed additional literature searches and evidence synthesis, contributed to drafting and revising key sections of the manuscript, and contributed to the preparation and revision of the figures, and provided critical feedback. H.X. assisted in the literature review and provided feedback on the manuscript. Y.L. and S.Q. conceived the review, supervised the project, and revised the manuscript critically for important intellectual content. All authors reviewed and approved the final version of the manuscript.

Funding

This work was supported by the Zhong Nanshan Youth Science and Technology Innovation Award Fund of the China Youth Entrepreneurship and Employment Foundation and the Huzhou Municipal Public Welfare Application Research Project (Grant No. 2022GZ63) and by the Zhong Nanshan Youth Science and Technology Innovation Award Fund of the China Youth Entrepreneurship and Employment Foundation (no grant number).

Data availability

Not applicable.

Declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.