Introduction

Ischemic stroke (IS) is usually caused by cerebral arterial occlusion and a consequent abrupt reduction or interruption of regional cerebral blood flow, and it accounts for a major proportion of stroke-related mortality and long-term disability worldwide [1]. Reduced cerebral perfusion restricts oxygen and glucose delivery to the ischemic core and penumbra to different extents, leading to inadequate adenosine triphosphate (ATP) production, disruption of ionic gradients, and glutamate accumulation. These changes promote cell injury and death through Ca²⁺ overload, mitochondrial dysfunction, and oxidative stress [2]. As neurons and other brain cells are damaged, damage-associated molecular patterns (DAMPs), cellular debris, and metabolites are released into the extracellular space. Concurrently, disruption of blood-brain barrier integrity facilitates the entry of peripheral immune cells into ischemic tissue, together creating an inflammatory and metabolic microenvironment that evolves over the course of disease [3]. During recovery, the sustained clearance of myelin and membrane-lipid debris further increases the lipid and cholesterol-handling burden on microglia [4]. Thus, the post-stroke extracellular milieu is not defined solely by hypoxia or inflammation; rather, it imposes multiple concurrent stresses, including energy-substrate imbalance, injury-associated signals, immune-cell infiltration, and lipid overload, which collectively influence microglial inflammation, phagocytic clearance, and tissue repair.

Microglia are the principal resident immune cells of the brain and among the earliest cell types to sense local injury after IS. Although the conventional M1/M2 dichotomy captures certain pro-inflammatory and repair-associated responses, it does not adequately represent the complex and dynamic functional states that emerge after stroke [5]. Single-cell RNA sequencing and spatial transcriptomic studies have further revealed multiple microglial transcriptional and functional states whose distribution varies with disease progression and the local ischemic milieu [6]. These findings suggest that the local environment may shape microglial metabolism; however, pathway-enrichment analyses primarily indicate transcriptional tendencies and do not directly demonstrate corresponding changes in metabolic flux. Microglial functional changes must therefore be interpreted in the context of local injury severity and mitochondrial status. Glycolytic adaptation can provide rapid energy for damage sensing, migration, and inflammatory-mediator production during the acute phase, whereas mitochondrial oxidative metabolism, lysosomal degradation, and lipid-handling capacity support the efficient clearance of apoptotic cells, neutrophils, and myelin debris [7, 8]. Accordingly, post-stroke microglial metabolic reprogramming is not merely a secondary consequence of energy deprivation but an important link between the ischemic microenvironment and shifts among inflammatory, clearance, and reparative functions. Current studies have implicated glycolysis, lactate transport, tricarboxylic acid (TCA) cycle intermediates, lipid handling, danger-signal sensing, and mitochondrial quality control in the regulation of microglial function [4, 9,10,11]; nevertheless, an integrated understanding of how the ischemic microenvironment and mitochondrial responses jointly shape microglial function remains lacking. This review therefore examines post-stroke microglial immunometabolic changes from the perspective of microenvironment-mitochondria interactions and discusses the scope and limitations of the available evidence.

Literature Search and Selection

This narrative review examined microglial immunometabolic remodeling and mitochondrial function after IS, with particular emphasis on their interactions with the ischemic microenvironment. PubMed and Web of Science were searched from database inception through August 2026. Medical Subject Headings and free-text terms were combined. The principal search terms included “microglia,” “ischemic stroke,” “cerebral ischemia,” “immunometabolism,” “metabolic reprogramming,” “glycolysis,” “pentose phosphate pathway,” “lactate,” “monocarboxylate transporter,” “TCA cycle,” “succinate,” “reverse electron transport,” “lipid metabolism,” “lipid droplet,” “cholesterol metabolism,” “TREM2,” “P2 × 7,” “mitochondrial DNA,” “mitochondria,” “oxidative phosphorylation,” and “mitophagy,” combined with the Boolean operators “AND” and “OR.” Searches addressing clinical translation additionally incorporated “patient,” “biomarker,” “prognosis,” “clinical outcome,” and “clinical trial.” Reference lists of eligible studies and relevant reviews were also screened to identify original and foundational studies not captured by the database searches.

Records were initially screened by title and abstract, followed by full-text assessment. Studies were eligible if they investigated IS, cerebral ischemia-reperfusion, or a relevant ischemic/hypoxic model; included microglia as the study population or a principal outcome; and addressed metabolic reprogramming, mitochondrial function, inflammation, phagocytic clearance, tissue repair, or clinical translation. Original studies directly evaluating microglial metabolism or mitochondrial alterations after stroke were prioritized. When direct stroke evidence was unavailable, non-stroke microglial studies or tissue-level investigations that could inform the relevant mechanisms were considered, with their evidentiary scope specified in the text. Duplicate publications, studies without a clear relationship to IS or microglia, studies that did not address metabolic or mitochondrial mechanisms, and reports with insufficient full-text information were excluded. For clinical evidence, priority was given to studies of patients with IS that evaluated metabolic, mitochondrial, oxidative-stress, or myeloid-immune indices and reported clearly defined clinical, imaging, or prognostic outcomes. Reviews and meta-analyses were used mainly to define concepts, compare evidence, and trace original studies, whereas mechanistic and interventional effects were evaluated preferentially from primary research. Because this was a narrative review, no quantitative synthesis or formal risk-of-bias assessment was performed.

Spatiotemporal Heterogeneity and Metabolic Requirements of Microglia after Stroke

Limitations of the M1/M2 Dichotomy and Microglial Heterogeneity After Stroke

Following cerebral ischemia, microglia rapidly sense local danger signals and participate in inflammatory-mediator release, debris clearance, tissue remodeling, and neural repair [12]. Conventionally, post-stroke microglia have been divided into M1-like and M2-like states. The former is typically characterized by increased expression of pro-inflammatory molecules, including inducible nitric oxide synthase (iNOS), CD86, tumor necrosis factor-α (TNF-α), and interleukin-1β (IL-1β), and is considered to contribute to neuroinflammatory amplification and secondary tissue injury. The latter is commonly represented by arginase 1 (Arg1), CD206, interleukin-10 (IL-10), and transforming growth factor-β (TGF-β), and is associated with inflammation resolution, phagocytic clearance, and tissue repair [13]. Although this framework is useful for summarizing certain inflammation-related phenotypes, it was derived largely from polarization paradigms established under in vitro culture conditions and cannot adequately capture the continuum and heterogeneity of microglial states shaped by time, spatial microenvironment, and metabolism after stroke [6, 14]. Recent single-cell RNA sequencing studies have identified multiple coexisting microglial states during acute ischemia and early reperfusion, including homeostatic-like, preactivated, inflammation-associated, and neuropeptide-associated states. Even within cell populations displaying M1-like transcriptional features, some cells with relatively weak inflammatory responses lack typical M2-like features, highlighting substantial state heterogeneity after stroke [14]. Subsequent studies have identified ischemia-associated MKI67⁺, CH25H⁺, and OASL⁺ microglial populations. CH25H⁺ cells exhibit pronounced phagocytic and neuroprotective features, whereas OASL⁺ cells are more closely associated with interferon responses and the progression of neuroinflammation [15]. Disease-associated microglia-like, proliferative, macrophage-like, and interferon-response states have likewise been observed in permanent cerebral ischemia [16]. Collectively, these findings indicate that post-stroke microglia can engage inflammatory, proliferative, phagocytic, lipid-handling, and tissue-reparative programs to varying degrees. The conventional M1/M2 classification should therefore be viewed as reflecting relative shifts in specific molecular and functional modules rather than two stable, mutually exclusive cell populations.

Microglial heterogeneity extends beyond an increase in the number of transcriptionally defined subpopulations and is jointly shaped by the local spatial microenvironment and myeloid-cell origin. Integrated single-cell and spatial transcriptomic analyses have identified ischemic core-associated microglia (ICAM) and ischemic penumbra-associated microglia (IPAM). ICAM are enriched predominantly in the ischemic core, whereas IPAM are found mainly in the penumbra, and the two states differ in their metabolic requirements [6]. ICAM and IPAM are more appropriately interpreted as context-dependent reactive states rather than cell types with fixed identities [17], because the boundaries of the ischemic core and penumbra evolve with reperfusion and secondary injury, while microglia themselves migrate and transition between states. The post-stroke myeloid compartment also contains multiple monocyte-derived populations whose spatial distribution and response profiles are influenced by developmental origin, local tissue injury, and regional brain connectivity [18]. A recent study combining TMEM119-CreER lineage tracing with single-cell sequencing further demonstrated that resident microglia and infiltrating monocyte-derived macrophages exhibit distinct spatial enrichment, temporal dynamics, and transcriptional states in the post-stroke brain, although they share several myeloid markers within the injured milieu [19]. These findings support viewing post-stroke microglia as a dynamic functional continuum regulated by disease phase, spatial microenvironment, and cellular origin. Nevertheless, single-cell clustering remains sensitive to factors such as stroke model, animal age, and sampled brain region; transcriptionally defined clusters and pathway enrichment should not be equated directly with stable functional subtypes or actual metabolic activity.

Functional Transitions and Metabolic Requirements of Microglia Across Stroke Phases

Microglia exhibit substantial metabolic plasticity and can adjust their use of glycolysis, oxidative phosphorylation (OXPHOS), lipid metabolism, and amino acid metabolism in response to the local microenvironment and functional demand [20]. During the acute phase of ischemia, limited oxygen and glucose availability and mitochondrial dysfunction constrain OXPHOS, whereas danger-signal recognition, chemotactic migration, inflammatory-mediator synthesis, and phagocytic clearance increase energetic and biosynthetic demands. Microglia may consequently increase glucose utilization and glycolysis to meet the need for rapid energy production [21]. Following ischemia-reperfusion, increased levels of glycolysis-related metabolites, including glucose-6-phosphate, fructose-6-phosphate, pyruvate, and lactate, have been detected in brain tissue, indicating substantial remodeling of glucose metabolism early after stroke [22]. Consistent with this observation, chemokine-like factor 1 (CKLF1) suppresses OXPHOS and enhances glycolysis in microglia under cerebral ischemic conditions, further supporting post-stroke metabolic reprogramming in these cells [23]. Enhanced glycolysis, however, does not necessarily correspond to a single pro-inflammatory state. Microglia in the ischemic core are enriched for glycolytic, inflammatory, and chemotactic programs, whereas those in the penumbra exhibit stronger TCA-cycle, OXPHOS, and myelin-supporting signatures [6]. Thus, early post-stroke microglial metabolic reprogramming is spatially heterogeneous, and its functional consequences depend on local ischemic severity, duration, and mitochondrial status.

As inflammation progresses from acute injury toward resolution and repair, microglial functional demands gradually shift from danger sensing and inflammatory initiation to cellular clearance and tissue remodeling. Temporal analysis after transient middle cerebral artery occlusion showed that Itgb2⁺ microglia were enriched mainly for energy metabolism, cell-cycle, and inflammatory-response programs on day 1, angiogenesis-related programs on day 3, and myelination, axonal ensheathment, and tissue-repair programs on day 7 [24]. These findings suggest phase-dependent adjustment of microglial functional programs, although pathway enrichment alone does not demonstrate that the corresponding functions have been executed. Functional studies further showed that resolvin D1 promotes microglial engulfment of neutrophils and enhances OXPHOS and ATP production through AMP-activated protein kinase (AMPK)-dependent glutamine uptake and glutaminolysis [8]. These findings indicate that high phagocytic loads require support from oxidative metabolism and anaplerotic substrate supply.

During the recovery and chronic phases, lipid debris derived from myelin, cell membranes, and necrotic tissue imposes a sustained metabolic burden, extending microglial demands from debris uptake to lysosomal hydrolysis, lipid-droplet turnover, and cholesterol efflux. In the acute phase, lipid-droplet formation may buffer lipotoxicity by temporarily sequestering free fatty acids and peroxidized lipids; inhibition of lipid-droplet breakdown can reduce inflammatory-mediator release and attenuate acute ischemic brain injury [25]. As disease progresses, cerebral cholesteryl esters and triacylglycerols continue to undergo remodeling, while lipid-droplet-enriched microglia exhibit increased expression of perilipin 2 (PLIN2) and the inflammatory mediators TNF-α, interleukin-6 (IL-6), and IL-1β [26]. Myelin-derived lipids, lipid droplets, and foamy myeloid cells may persist in chronically infarcted tissue, accompanied by altered expression of lipid-handling genes, including Trem2, Apoe, Abca1, Npc1/2, and Lamp1 [27]. The functional outcome of microglia during recovery therefore depends on coordination among lipid uptake, lysosomal degradation, and cholesterol efflux. Short-term lipid-droplet formation may be adaptive, whereas persistent lipid influx that exceeds cellular handling capacity may cause post-phagocytic lipid-metabolic dysfunction and impede tissue repair Fig. 1.

Fig. 1
Fig. 1
Full size image

Spatial heterogeneity, functional transitions, and metabolic requirements of microglia after stroke

Spatially, ischemic core-associated microglia (ICAM) are predominantly enriched for inflammatory, chemotactic, and glycolytic programs and exhibit restricted OXPHOS, whereas ischemic penumbra-associated microglia (IPAM) display stronger OXPHOS, phagocytic-clearance, myelin-support, and tissue-repair features. ICAM and IPAM represent spatially biased reactive states rather than fixed, mutually exclusive cell populations. Microglial functions and metabolic requirements also change across disease phases. During the acute phase (0–24 h), increased glucose transporter 1 (GLUT1)-mediated glucose uptake and glycolysis provide rapid energy and support inflammatory-mediator release. During the subacute phase (1–7 days), the predominant demand shifts toward the phagocytic clearance of apoptotic cells, neutrophils, and tissue debris, supported by glutamine anaplerosis and OXPHOS. During the recovery phase (> 7 days), lipid uptake, lysosomal processing, and ATP-binding cassette transporter A1/G1 (ABCA1/ABCG1)-mediated cholesterol efflux help maintain lipid homeostasis and support myelin repair, whereas excessive lipid loading may promote persistent inflammation. These metabolic programs overlap across phases and change dynamically with local ischemic severity and reparative demand.

Abbreviations: ICAM, ischemic core-associated microglia; IPAM, ischemic penumbra-associated microglia; DAMPs, damage-associated molecular patterns; GLUT1, glucose transporter 1; OXPHOS, oxidative phosphorylation; ATP, adenosine triphosphate; TCA, tricarboxylic acid cycle; α-KG, α-ketoglutarate; CoA, coenzyme A; TREM2, triggering receptor expressed on myeloid cells 2; ApoE, apolipoprotein E; ABCA1, ATP-binding cassette transporter A1; ABCG1, ATP-binding cassette transporter G1; h, hour; d, day.

Metabolic Coupling between the Ischemic Microenvironment and Microglial Mitochondria

Glucose Metabolic Reprogramming and Lactate Signaling: Links between Inflammatory Activation and Mitochondrial Quality Control

Ischemia- and Hypoxia-Induced Glycolysis and Inflammatory Activation

Ischemia and hypoxia can shift overall brain metabolism away from mitochondrial OXPHOS toward increased glycolysis. Metabolomic studies have detected elevated levels of glycolysis-related metabolites, including glucose-6-phosphate, fructose-6-phosphate, lactate, and pyruvate, in brain tissue after stroke, consistent with a metabolic signature of increased glycolytic flux in the ischemic brain [20]. Whole-brain metabolic changes, however, cannot be equated directly with microglia-specific glycolytic reprogramming, because neurons, astrocytes, and infiltrating immune cells may all contribute to the observed metabolic signals. Under homeostatic conditions, microglial energy production relies relatively strongly on mitochondrial OXPHOS. Under ischemic, hypoxic, or inflammatory conditions, microglia can suppress OXPHOS and increase glycolysis to meet the rapid demands of damage sensing, chemotactic migration, phagocytic clearance, and inflammatory-cytokine release [28]. In BV2 cells subjected to oxygen-glucose deprivation/reoxygenation (OGD/R), lactate production and the extracellular acidification rate (ECAR) increased, together with upregulation of glycolytic enzymes. Triggering receptor expressed on myeloid cells 2 (TREM2) overexpression suppressed glycolysis through inhibition of the Janus kinase 2/signal transducer and activator of transcription 3 axis and enhanced the expression of repair-associated molecules [29]. Further work showed that Toll-like receptor 4 (TLR4) activation increased glycolysis in both human induced microglia-like cells and mouse microglia, although the regulation of metabolic enzymes such as PFKL, PFKP, 6-phosphofructo-2-kinase/fructose-2,6-bisphosphatase 3 (PFKFB3), ACO1, and MDH1 differed between species. Increased glycolysis in mouse microglia depended more strongly on hexokinase activity, whereas that in human induced microglia-like cells depended more strongly on phosphofructokinase activity [30]. Thus, enhanced glycolysis in post-stroke microglia should not be interpreted simply as either a pro-inflammatory or a protective response, but rather as an immunometabolic adaptation jointly regulated by stroke phase, cellular origin, inflammatory intensity, and the ischemic microenvironment.

Enhanced glycolysis in post-stroke microglia is also strongly time dependent. A moderate early increase in glycolysis can support rapid microglial responses, including damage sensing, migration, and phagocytic clearance. For example, hexokinase 2 (HK2) deficiency reduces glycolytic flux in microglia, impairs damage sensing and chemotactic responses, and is accompanied by mitochondrial dysfunction, increased reactive oxygen species (ROS), and worse neurological impairment [7]. Under sustained hypoxia or intense inflammatory stimulation, however, enhanced glycolysis may shift from a compensatory energy response to a pro-inflammatory metabolic program. HK2 upregulation may also promote acetyl-coenzyme A accumulation and histone acetylation, thereby activating the expression of inflammatory mediators such as IL-1β [31]. Similarly, increased phosphoglycerate kinase 1 (PGK1) expression after middle cerebral artery occlusion (MCAO) has been associated with pro-inflammatory microglial activation, increased ECAR, and upregulation of lactate dehydrogenase (LDH), pyruvate kinase M2 (PKM2), and lactate dehydrogenase A (LDHA). PGK1 silencing reduces glycolysis and inflammatory-cytokine release, potentially through p300-mediated regulation of H3K27 acetylation [32]. PFKFB3 can likewise modulate glucose transporter 1 (GLUT1) expression, lactate production, and inflammatory-mediator release by increasing glycolytic flux. Inhibition of the PFKFB3/GLUT1 axis suppresses glycolysis, promotes a repair-associated microglial phenotype, and reduces pro-inflammatory-mediator release [33]. Sustained or excessive glycolytic activation may therefore convert an initially compensatory metabolic adaptation into a driver of secondary injury through epigenetic regulation, amplification of inflammatory transcription, ROS accumulation, and mitochondrial dysfunction.

The Pentose Phosphate Pathway and NADPH-Mediated Redox Regulation

The pentose phosphate pathway (PPP) originates from glucose-6-phosphate. In its oxidative branch, glucose-6-phosphate is first metabolized by glucose-6-phosphate dehydrogenase and subsequently by 6-phosphogluconate dehydrogenase, generating reduced nicotinamide adenine dinucleotide phosphate (NADPH) and ribulose-5-phosphate; the latter can be converted to ribose-5-phosphate for nucleotide synthesis. Allocation of glucose-derived carbon between glycolysis and the PPP may determine the balance between rapid energy production and antioxidant defense in ischemic cells. In MCAO mice and OGD-treated neurons, Ren et al. found that ischemic injury reduced PPP-related metabolites, including 6-phosphogluconate, xylulose-5-phosphate, ribose-5-phosphate, and erythrose-4-phosphate, together with the NADPH/NADP⁺ ratio and the reduced glutathione (GSH)-to-oxidized glutathione (GSSG) ratio. PKM2 inhibition redirected part of the glucose-derived carbon flux from glycolysis toward the PPP, enhanced PPP-dependent antioxidant capacity, and attenuated ischemic injury [34]. These findings suggest that glycolytic predominance under severe ischemia may restrict PPP substrate supply and cellular reducing capacity. Because the experiments were conducted mainly in neurons, however, they do not directly establish an equivalent change in PPP flux in post-stroke microglia.

NADPH has divergent functional destinations in microglia. On the one hand, it supplies reducing equivalents to glutathione reductase and thioredoxin reductase, promotes the conversion of GSSG to GSH, and maintains thioredoxin-system activity, thereby limiting ROS accumulation. Studies using gene-edited human microglia deficient in glucose-6-phosphate dehydrogenase (G6PD) showed that G6PD loss reduced NADPH levels, increased basal ROS, and impaired lysosomal acidification and phagocytic clearance. Supplementation with citrate and malate to support NADPH generation associated with isocitrate dehydrogenase 1 and malic enzyme 1 partially restored redox balance and lysosomal function [35, 36]. On the other hand, NADPH is also the electron donor used by NADPH oxidases, particularly NADPH oxidase 2 (NOX2), to generate superoxide. When NOX2 remains persistently active, NADPH supplied by an enhanced PPP may be preferentially consumed for ROS production [37]. In non-stroke microglial models, inflammatory stimulation increased G6PD expression and PPP activity, providing substrate for excessive NOX2 activation and promoting ROS generation and nuclear factor-κB (NF-κB)-associated inflammatory responses; G6PD inhibition or knockdown attenuated these effects [37]. NADPH is therefore not intrinsically pro-oxidant. Its net effect depends on how reducing equivalents are partitioned between the glutathione/thioredoxin antioxidant systems and NOX-mediated oxidation. Microglia-specific evidence in stroke remains limited, and an increase in NADPH should not be interpreted directly as evidence of aggravated oxidative stress.

Lactate Accumulation and Monocarboxylate Transporter-Mediated Metabolic Adaptation in Microglia

Lactate is the principal metabolite generated from pyruvate when glycolysis is increased [38]. In IS, lactate accumulation not only reflects compensatory glycolysis under hypoxia but may also regulate microglial phenotypes as a metabolic substrate and signaling molecule [39]. Intracerebroventricular lactate administration reduces the expression of pro-inflammatory molecules induced by lipopolysaccharide (LPS) and increases Arg1 and CD206 expression, suggesting that elevated lactate may suppress classical pro-inflammatory polarization and favor a protective phenotype [40]. In OGD- or OGD/R-based in vitro models, exogenous lactate improves microglial viability, attenuates iNOS, IL-1β, TNF-α, or NF-κB-associated inflammatory signaling, and increases the expression of protective molecules such as Arg1, IL-10, and TGF-β, thereby improving neuronal survival [41, 42]. Lactate can also contribute to transcriptional regulation after cerebral ischemia through protein lactylation. For example, methyl-CpG-binding protein 2 lactylation suppresses apoptosis-related gene expression, reduces neuronal apoptosis, and attenuates ischemic brain injury [43]. These findings indicate that lactate may serve simultaneously as an energy substrate, an anti-inflammatory signal, and an epigenetic regulator in the acute ischemic milieu. Its effects, however, are not uniformly protective. Hypoxia-inducible factor-1α (HIF-1α) is an important transcriptional regulator linking hypoxia sensing to increased glycolysis and lactate production. Ischemia and hypoxia stabilize HIF-1α and promote its nuclear translocation, thereby increasing glucose uptake and glycolytic-gene expression, directing more pyruvate toward lactate production, and limiting its entry into mitochondrial oxidative metabolism. Early after ischemia, this response may rapidly supply ATP to support microglial damage sensing, migration, and phagocytosis [44, 45]. Persistent HIF-1α activation, however, may sustain glycolysis-dependent inflammatory transcription, lactate accumulation, and inflammatory-mediator release. Increased mechanistic target of rapamycin/HIF-1α signaling after ischemia has been associated with upregulation of glycolytic proteins, including HK2, PFKM, and GLUT1, and with amplified inflammation [46]. HIF-1α-mediated lactate metabolism is therefore strongly dependent on time and cellular state, and its direction of effect may vary with ischemic severity and duration, lactate source and concentration, monocarboxylate transporter-mediated transmembrane transport, and mitochondrial status.

Monocarboxylate transporter (MCT) family members are key mediators of lactate, pyruvate, and other monocarboxylate substrate exchange [47]. In animal models of ischemia, MCT1 and MCT2 expression is increased in peri-infarct regions and is partially localized to activated microglia [48], suggesting that MCT-mediated lactate transport contributes to post-stroke microglial metabolic adaptation. Functionally, MCT-mediated transport may help maintain glycolytic flux and monocarboxylate utilization. Lactate can be oxidized to pyruvate and subsequently converted to acetyl-coenzyme A, thereby providing substrate for the TCA cycle, mitochondrial OXPHOS, and ATP production. Conversely, MCT1-mediated substrate supply may also support PFKFB3 activity and pro-inflammatory-mediator expression, thereby sustaining pro-inflammatory microglial activation [40]. MCT dysfunction, particularly microglial MCT4 deficiency, impairs lactate-induced lysosomal acidification and activation of the autophagy-lysosomal pathway, interferes with the removal of damaged mitochondria, and consequently promotes defective mitochondrial quality control, ROS accumulation, and inflammation [49]. MCT-mediated lactate transport thus connects increased glycolysis with substrate reutilization and also contributes to microglial functional transitions by modulating autophagy-lysosomal function, mitochondrial quality control, and inflammatory signaling. The phase- and cell type-specific functions of individual MCT isoforms remain central to explaining the bidirectional effects of lactate.

Effects of Altered Glucose Metabolism on Mitochondrial Fission and Mitophagy

A glycolytic shift and mitochondrial quality control influence one another. Under ischemic and hypoxic conditions, restricted mitochondrial OXPHOS, ROS accumulation, and loss of membrane potential may further increase microglial dependence on glycolysis; conversely, a persistent glycolytic shift may exacerbate mitochondrial dysfunction [50]. Mitochondrial fission and mitophagy are important regulatory processes within this interaction. LPS stimulation alters mitochondrial morphology in microglia, decreases the oxygen consumption rate (OCR), and increases ECAR. Inhibition of dynamin-related protein 1 (Drp1)-associated mitochondrial fission reduces succinate and ROS levels, improves mitochondrial membrane potential, and partially reverses the glycolytic shift [51]. Mitochondrial dynamics and metabolic reprogramming, however, are not linked by a simple linear causal relationship. Although the mitochondrial-division inhibitor 1 (Mdivi-1) reduces pro-inflammatory marker expression induced by LPS/interferon-γ, it does not fully restore OXPHOS or induce microglial repolarization [52]. Drp1-dependent fission may therefore contribute to inflammatory activation rather than serve as the sole determinant of glycolytic reprogramming. Moreover, Mdivi-1 may exert pharmacological effects independent of Drp1, and findings obtained with this inhibitor alone are insufficient to establish a causal relationship between Drp1-dependent fission and microglial metabolic reprogramming.

Mitophagy removes damaged mitochondria, limits ROS accumulation, and maintains microglial metabolic homeostasis, thereby influencing the inflammatory consequences of a glycolytic shift. After cerebral ischemia, PTEN-induced kinase 1 (PINK1)-associated mitophagy is not invariably protective. In a photothrombotic stroke model, delivery of PINK1 small interfering RNA attenuated ischemic injury [53], suggesting that the effects of this pathway may depend on the degree of activation, injury phase, and autophagic flux. CKLF1 can promote microglial activation by inducing defective mitophagy and mitochondrial dysfunction, as reflected by reduced basal, maximal, ATP-linked, and spare respiratory capacities, together with increased ECAR, decreased OCR, and enhanced mitochondrial fission [11]. The functional consequences of mitophagy therefore cannot be inferred solely from changes in the expression of mitophagy-related molecules; autophagic flux and mitochondrial functional status must also be considered. Collectively, increased glycolysis and lactate/MCT transport can influence microglial inflammatory and metabolic states through effects on mitochondrial fission and autophagic flux. When mitochondrial quality control cannot fully buffer ischemia-associated metabolic stress, restricted TCA-cycle flux and redistribution of intermediates such as succinate may become key links between mitochondrial metabolic dysfunction and ROS-mediated inflammatory amplification.

TCA-Cycle Disruption and Succinate Signaling in Mitochondrial Metabolic Dysfunction and Oxidative-Stress Amplification

Ischemia-Induced TCA-Cycle Interruption and Inflammatory Metabolic Transitions in Microglia

Ischemia and hypoxia can disrupt the continuity of mitochondrial oxidative metabolism in microglia, restricting pyruvate oxidation, acetyl-coenzyme A generation, and TCA-cycle flux and redistributing intermediates such as citrate, cis-aconitate, α-ketoglutarate, succinate, and malate [54]. These changes not only reflect impaired mitochondrial energy production but may also contribute to inflammatory metabolic transitions in microglia by modifying redox status, histone modifications, and inflammatory-signaling thresholds. The long non-coding RNA TUG1 has been reported to alter citrate, cis-aconitate, and other intermediates by modulating TCA-cycle activity and to be associated with transitions in microglial inflammatory phenotype. TUG1 downregulation improves the interrupted TCA cycle and mitochondrial metabolism and is accompanied by reduced pro-inflammatory responses [55]. The mitochondrial citrate carrier SLC25A1 mediates citrate/isocitrate-malate exchange between the mitochondrial matrix and cytosol. Its dysregulation may alter the export of TCA intermediates and the generation of cytosolic acetyl-coenzyme A, thereby linking mitochondrial metabolism to inflammatory transcription. Modulation of SLC25A1 can improve the interrupted TCA cycle and mitochondrial OXPHOS and attenuate inflammation [56]. In addition, reduced Suclg1 expression has been detected in brain tissue after MCAO/reperfusion, suggesting that the conversion of succinyl-coenzyme A to succinate and the associated TCA flux may be impaired. Restoration of the relevant metabolic node reduced microglial accumulation and suppressed NOD-, LRR- and pyrin domain-containing protein 3 (NLRP3) inflammasome signaling [57]. Thus, post-ischemic TCA-cycle interruption represents more than inadequate energy supply; by altering intermediate flux, mitochondrial redox status, and the sensitivity of inflammatory signaling, it may contribute to inflammatory metabolic transitions in microglia.

Succinate Oxidation and Reverse Electron Transport in Oxidative-Stress and Microglial Inflammatory Amplification

Among TCA-cycle intermediates, succinate is an important ischemia-associated metabolite linking mitochondrial metabolic dysfunction, oxidative stress, and immune regulation [58]. During cerebral ischemia-reperfusion, succinate accumulation and altered electron transfer during its oxidation by succinate dehydrogenase (SDH) are considered important mechanisms linking TCA-cycle disruption to mitochondrial ROS production. Succinate has been reported to increase in both the ischemic core and peri-infarct tissue and to be associated with mitochondrial superoxide production during reperfusion [59]. Classical metabolic studies of ischemia-reperfusion have also shown that selective succinate accumulation is a prominent metabolic feature of ischemic tissue. Upon reperfusion, rapid SDH-mediated oxidation of accumulated succinate promotes reverse electron transport (RET) at complex I and drives ROS production, whereas limiting succinate accumulation attenuates ischemia-reperfusion injury in mouse stroke models [60]. Restricting SDH-mediated succinate oxidation is therefore considered a potential strategy for modulating RET and downstream oxidative stress. Altered electron transfer during SDH-mediated succinate oxidation may represent an important metabolic link in mitochondrial ROS generation and oxidative-stress amplification after cerebral ischemia-reperfusion.

Available evidence indicates that succinate signaling does not exert uniform effects in microglia, potentially because extracellular receptor-mediated signaling differs from intracellular metabolic actions. Extracellular succinate can act as a signaling molecule and regulate immunometabolism through succinate receptor 1 (SUCNR1) [61]. In models of central nervous system inflammation, transplanted neural stem cells increase SLC13A3 and SLC13A5 expression through a SUCNR1-related mechanism, promote extracellular succinate uptake, reduce local inflammatory succinate levels, and induce an anti-inflammatory microglial phenotype [62]. By contrast, an increase in intracellular succinate does not necessarily amplify inflammation. Moderate succinate concentrations suppress LPS-induced pro-inflammatory transitions in primary microglia and reduce mitochondrial fission and intracellular ROS production. This protective effect is independent of SUCNR1 and is more likely related to mitochondrial redox status and intracellular metabolic regulation [58]. Succinate signaling in post-stroke microglia should therefore be interpreted by distinguishing extracellular receptor-mediated signaling from intracellular SDH-associated metabolic effects. Its net effect may depend jointly on the source and local concentration of succinate, SDH activity, SUCNR1 involvement, ischemic phase, and mitochondrial status.

Glutamine Anaplerosis and Arginine Metabolism in the Regulation of Inflammatory Metabolism

In the post-stroke metabolically stressed milieu, glutamine and glutamate metabolism provide complementary links among the TCA cycle, microglial functional changes, and neurotoxicity. Glutamine can be converted through glutamate to α-ketoglutarate, which enters the TCA cycle as an anaplerotic substrate. In ischemic stroke models, resolvin D1 promotes microglial glutamine uptake and glutaminolysis and supports OXPHOS and ATP production through an AMPK-dependent mechanism, thereby enhancing the clearance of neutrophils by microglia [8]. These findings identify glutamine metabolism as an important metabolic link between TCA-cycle anaplerosis and microglial phagocytic clearance. Unlike the anaplerotic role of glutamine, dysregulated glutamate release or clearance is primarily associated with excitotoxicity and inflammatory amplification after stroke. Insufficient ATP production, disrupted ionic homeostasis, and impaired glutamate transport after ischemia can cause extracellular glutamate accumulation. In experimental stroke, miR-29a deficiency increased the expression of pro-inflammatory molecules in microglia and glutamate release, thereby aggravating neural injury [64]. Glutamine-derived α-ketoglutarate may therefore support TCA-cycle activity and phagocytic clearance, whereas dysregulated microglia-associated glutamate release may contribute to inflammatory amplification and secondary neural injury.

Arginine metabolism influences microglial inflammatory states primarily through the iNOS/nitric oxide and Arg1 branches. Under inflammatory stimulation, iNOS uses arginine to generate nitric oxide (NO) [63]. Physiological amounts of NO contribute to vasodilation and immune regulation, whereas sustained or excessive NO and its reactive nitrogen derivatives can inhibit mitochondrial respiratory complexes, exacerbate oxidative and nitrosative stress, and act together with ROS in secondary inflammatory injury after ischemia [64]. By contrast, Arg1-mediated arginine metabolism directs substrates toward ornithine, polyamines, proline, and related products and is generally associated with inflammation resolution, cellular repair, and extracellular-matrix remodeling [65]. Nevertheless, post-stroke Arg1-associated effects may depend on cellular origin and disease phase and should not be equated automatically with protection. Experimental stroke studies have shown that depletion of Arg1⁺ microglia/macrophages aggravates ischemic injury, whereas Arg1 expressed by infiltrating macrophages may, in some contexts, alter the inflammatory microenvironment and influence functional recovery. Arginine has also been reported to attenuate inflammation and exert neuroprotection after cerebral ischemia-reperfusion in rats by suppressing LDHA signaling in microglia [66]. Although arginine is not a major direct carbon source for the TCA cycle, it may indirectly participate in microglial inflammatory metabolic remodeling during TCA-cycle dysfunction by modulating iNOS/NO production, mitochondrial respiratory stress, and associated inflammatory transcriptional programs.

Persistent Microglial Inflammation Driven by Defective Lipid Handling: From Lipid-Droplet Accumulation to Impaired Cholesterol Efflux

Increased Lipid Burden and Lipid-Droplet Accumulation after Ischemia

Following IS, necrotic cellular debris, myelin damage, and membrane-lipid release markedly increase the lipid-handling burden on microglia within and around the ischemic region [25, 27]. In acute MCAO, microglia within the ischemic lesion accumulate abundant lipid droplets and exhibit inflammation- and phagocytosis-related transcriptional features. Pharmacological promotion of lipid-droplet formation reduces pro-inflammatory cytokine production, infarct volume, and neurological deficits, suggesting that lipid-droplet formation may represent a compensatory mechanism by which microglia buffer acute lipid loading and constrain inflammation [25]. Spatiotemporal changes in the cerebral lipidome under ischemic conditions can also increase microglial PLIN2 expression, lipid-droplet accumulation, and the expression of TNF-α, IL-6, and IL-1β, indicating that lipid-droplet accumulation is not uniformly protective in relation to inflammatory phenotypes [26]. Non-stroke studies likewise show that blocking adipose triglyceride lipase-mediated lipid-droplet breakdown in microglia reduces LPS-induced expression and secretion of C-C motif chemokine ligand 2, IL-1β, and IL-6, further supporting a role for abnormal lipid-droplet mobilization in amplifying microglial inflammation [67]. From a whole-organism lipid-metabolism perspective, edaravone studies have shown that serum triglycerides, total cholesterol, and low-density lipoprotein increase after MCAO/reperfusion, whereas edaravone lowers these indices, reduces inflammatory-cell accumulation and neuronal injury, and implicates an FDFT1-related lipid-metabolic pathway in ischemic brain injury [68]. In the chronic phase, abnormal lipid metabolism may progress to defective cholesterol handling. Repeated administration of hydroxypropyl-β-cyclodextrin attenuates chronic inflammation and secondary neurodegeneration after experimental stroke, suggesting that inadequate cholesterol and lipid clearance contributes to persistent inflammation [27]. Recent single-cell transcriptomic studies have further identified stroke-associated foamy microglia during the chronic phase. These cells are characterized by reprogrammed cholesterol metabolism, cholesterol accumulation, and persistent inflammatory activation; enhancing CYP46A1-mediated cholesterol metabolism alleviates microglial cholesterol overload and promotes white-matter repair and functional recovery [4]. Post-stroke lipid-droplet accumulation in microglia should therefore be interpreted as a phase-dependent metabolic adaptation: acute lipid-droplet formation may buffer lipids and constrain inflammation, whereas persistent defects in cholesterol handling and TREM2-related lipid metabolism may drive chronic inflammation and restrict repair.

Microglial Lipid Clearance and Regulation of Cholesterol Homeostasis

Under sustained lipid loading, microglia must coordinate lipid sensing, phagocytosis of myelin debris, lysosomal degradation, and cholesterol efflux to maintain lipid homeostasis. During recovery, CD11c⁺ microglia progressively increase from days 7 to 30 after stroke and exhibit higher expression of genes associated with phagocytosis, myelin support, and lipid metabolism, including Axl, Cd68, Igf1, Spp1, Csf1, Abca1, Abcg1, Apoe, Apoc1, and Lpl. These changes indicate persistent activation of microglial lipid handling and cholesterol transport during white-matter repair [69]. TREM2 is an important link among lipid sensing, phagocytic clearance, and metabolic adaptation to lipid loading. TREM2 knockdown impairs microglial phagocytosis, promotes cholesteryl-ester accumulation and lipid-droplet formation, increases PLIN2 and SOAT1 expression, and decreases LIPA, ApoE, ABCA1, NCEH1, and NPC2 expression. These changes are associated with weakened TGF-β1/Smad2/3 signaling, a stronger pro-inflammatory phenotype, and aggravated post-ischemic brain injury [9]. TREM2 should therefore be understood not simply as a promoter of phagocytosis but as an important metabolic regulatory node linking myelin-debris clearance, cholesterol metabolism, and inflammation resolution.

Cholesterol efflux mediated by ATP-binding cassette transporters A1 and G1 (ABCA1 and ABCG1) is another key mechanism by which microglia respond to post-stroke lipid loading. ABCA1 and ABCG1 promote the efflux of excess intracellular cholesterol and, together with the lipidation of apolipoproteins such as apolipoprotein E (ApoE), maintain cholesterol homeostasis and thereby limit lipid deposition, lipid-droplet accumulation, and cellular stress. In stroke-related studies, the experimental liver X receptor-α (LXRα) agonist CKN activates the LXRα/ABCA1 pathway, increases ABCA1 expression in penumbral microglia, reduces lipid deposition and lipid-droplet formation, and suppresses TLR4-mediated inflammatory signaling, thereby attenuating cerebral ischemia-reperfusion injury [70]. FTY720 has also been shown to promote TREM2-dependent redistribution of ABCA1 from lysosomes to the plasma membrane, increase cholesterol efflux from microglia, reduce intracellular cholesterol and lipid-droplet accumulation, and facilitate cholesterol transfer from microglia to oligodendrocytes to support myelination [71]. Thus, TREM2 primarily coordinates lipid sensing, myelin-debris clearance, and cholesterol recycling, whereas ABCA1/ABCG1 primarily mediates cholesterol efflux and lipid-homeostasis maintenance. Together, these pathways form an important metabolic network linking microglial lipid handling to white-matter repair after stroke Fig. 2.

Fig. 2
Fig. 2
Full size image

Defective lipid handling and TREM2-mediated regulation of cholesterol homeostasis in microglia after ischemic stroke

Following ischemic stroke, necrotic cellular debris, myelin debris, and membrane-lipid release increase the lipid-handling burden on microglia.Moderate lipid-droplet formation may temporarily sequester excess lipids and reduce lipotoxicity.When TREM2 function is intact or activated, TGF-β1/Smad2/3 and LXRα–ABCA1/ABCG1 signaling regulate the microglial inflammatory state and cholesterol transport, respectively, while cooperating with ApoE to promote cholesterol efflux.These processes reduce lipid deposition, suppress TLR4-mediated inflammation, favor a repair-associated microglial state, and support white-matter repair.Conversely, TREM2 deficiency or dysfunction increases PLIN2 and SOAT1 expression while decreasing LIPA, ApoE, ABCA1,ABCG1,NCEH1,and NPC2,leading to impaired lipid hydrolysis, cholesteryl-ester and lipid-droplet accumulation, and defective cholesterol efflux.Sustained lipid and cholesterol overload may further activate TLR4-associated inflammatory signaling, drive microglia toward pro-inflammatory and foamy states, maintain inflammation, and restrict tissue repair.Green arrows indicate lipid-clearance and repair-associated regulation; red arrows indicate defective lipid metabolism and inflammatory injury; and green dashed arrows indicate cooperation between ApoE and cholesterol transporters.

Abbreviations: TREM2,triggering receptor expressed on myeloid cells 2;TGF-β1,transforming growth factor-β1;Smad2/3,SMAD family member 2/3;LXRα,liver X receptor-α;ABCA1,ATP-binding cassette transporter A1;ABCG1,ATP-binding cassette transporter G1;ApoE, apolipoprotein E; TLR4,Toll-like receptor 4;PLIN2,perilipin 2;SOAT1,sterol O-acyltransferase 1;LIPA, lysosomal acid lipase; NCEH1,neutral cholesterol ester hydrolase 1;NPC2,Niemann–Pick disease type C2 protein.

Danger Signals in the Ischemic Microenvironment and Amplification of Microglial Inflammation

Complement Activation in Microglial Phagocytosis and Inflammation

Following cerebral ischemia-reperfusion, the complement cascade contributes to neuroinflammation as a key mechanism of damage recognition and innate immune amplification. Under physiological conditions, complement participates in synaptic pruning and homeostatic maintenance; however, abnormal or persistent activation in the ischemic penumbra or during recovery may disrupt glial responses, promote synaptic loss, and cause secondary neural injury. Taohong Siwu decoction has been reported to reduce C1qc, C3, and C5aR expression through regulation of the circDnajc1/miR-27a-5p/C1qc axis in an MCAO/reperfusion model and to attenuate microglial activation, inflammatory-cytokine release, and neuronal apoptosis, implicating C1q-related complement signaling in post-ischemic inflammatory amplification [72]. In addition, bone marrow stromal antigen 2 expressed by boundary astrocytes promotes microglial recruitment to the lesion border through C3/C3aR signaling, indicating that complement contributes to spatial interactions between astrocytes and microglia after stroke [73]. By contrast, endothelial progenitor-cell transplantation enhances the engulfment of apoptotic cells by microglia expressing repair-associated markers in the ischemic brain, increases complement receptor 3 (CR3) expression, and reduces synaptic loss, suggesting that C3/CR3-associated phagocytosis may facilitate damage clearance and tissue repair at specific stages [74]. Complement-mediated microglial responses are therefore strongly dependent on time and target: appropriate complement tagging may facilitate the clearance of apoptotic cells and tissue debris, whereas persistent activation or mistargeting may promote synaptic engulfment, inflammatory amplification, and impaired repair.

In addition to C1q/C3-mediated opsonophagocytosis, the complement cleavage products C3a and C5a regulate post-stroke neuroinflammation through C3aR and C5aR1. Binding of C3a/C5a to receptors on microglia or astrocytes can activate inflammatory signaling, promote cytokine release, and amplify inflammatory crosstalk between glial cells [75]. Sustained or excessive C3aR/C5aR1 activation may also increase intracellular Ca²⁺ loading and ROS production, thereby converting extracellular damage recognition into intracellular oxidative and metabolic stress in microglia [76]. Complement cleavage products are not uniformly detrimental, however. Some stroke studies have shown that C3a treatment promotes functional recovery by modulating astrocytic responses and cortical connectivity [77]. Complement signaling in post-stroke microglial inflammation is therefore phase- and cell type-dependent: it participates in damage recognition and phagocytic clearance but may also amplify inflammation when persistently activated.

The Extracellular ATP–P2 × 7–Mitochondrial DNA Axis in Microglial Inflammatory Amplification

After IS, cellular necrosis, loss of membrane integrity, and ionic disequilibrium increase extracellular ATP, converting ATP from an intracellular energy currency into a prototypical extracellular danger signal. P2X purinoceptor 7 (P2 × 7) is an ATP-gated ion channel that is highly expressed in microglia and is particularly sensitive to high extracellular ATP concentrations. Its activation induces Ca²⁺ influx and K⁺ efflux, promotes membrane-pore formation and inflammasome activation, and facilitates the maturation and release of inflammatory cytokines such as IL-1β [10]. In rats subjected to MCAO/reperfusion, increased P2 × 7 expression is associated with pro-inflammatory microglial polarization, NLRP3 activation, ROS accumulation, and post-stroke memory impairment. P2 × 7 inhibition reduces inflammation and oxidative stress and shifts microglia toward an Arg1⁺ phenotype [78]. The extracellular ATP–P2 × 7 pathway can therefore be considered an important route by which soluble danger signals in the ischemic microenvironment activate microglial inflammation.

Ca²⁺ influx, K⁺ efflux, and ROS accumulation triggered by P2 × 7 activation may further aggravate mitochondrial stress and provide upstream conditions for the release of mitochondria-derived DAMPs [10]. Stroke studies have shown that ischemia-reperfusion induces mitochondrial DNA (mtDNA) leakage from microglia and activates stimulator of interferon genes (STING) signaling, promoting interferon regulatory factor 3/NF-κB-associated inflammatory transcription. STING inhibition reduces cerebral infarction, edema, neuronal injury, and neurological deficits [79]. STING can also interact with NLRP3 to promote microglial pyroptosis and neuroinflammation after MCAO [79]. The extracellular ATP–P2 × 7 pathway and mtDNA-mediated inflammation may therefore constitute interconnected processes in post-ischemic microglial inflammatory amplification: the former converts extracellular ATP signaling into ionic disequilibrium and mitochondrial stress, whereas the latter translates mitochondrial injury into DAMP-mediated inflammatory amplification. Notably, relatively direct stroke evidence supports P2 × 7-mediated pro-inflammatory activation of microglia, whereas direct evidence that P2 × 7-induced mitochondrial injury causes mtDNA release in post-stroke microglia remains limited.

Collectively, post-stroke metabolic changes in microglia do not represent a set of independent pathways. During the acute phase, increased glycolysis and lactate transport can supply substrates for rapid responses; when mitochondrial quality control is inadequate, however, restricted TCA-cycle activity and succinate-associated oxidative stress may further exacerbate mitochondrial dysfunction. Sustained oxidative stress can in turn impair lysosomal function and lipid handling, promoting lipid-droplet and cholesterol accumulation. Concurrently, extracellular ATP, complement, and other danger signals may further amplify P2 × 7- and mtDNA-related inflammatory responses. These interacting processes may shift microglia from an initially compensatory response toward persistent inflammation, defective phagocytosis, and impaired repair Fig 3.

Fig. 3
Fig. 3
Full size image

Microglial immunometabolic remodeling and functional outcomes after ischemic stroke

The ischemic extracellular microenvironment formed after ischemia/reperfusion is characterized by hypoxia and glucose deprivation, lactate and succinate accumulation, increased lipid and myelin-debris burden, and the release of danger signals such as extracellular ATP. Together, these stimuli reshape microglial function through four interconnected metabolic and mitochondrial axes: (A) the lactate–MCT–mitophagy axis, involving a glycolytic shift, MCT-dependent lactate transport, lysosomal acidification, and mitochondrial quality control; (B) the succinate–SDH–RET axis, involving TCA-cycle disruption, SDH-dependent succinate oxidation, reverse electron transport at mitochondrial complex I, and an mtROS burst; (C) the lipid droplet–TREM2–cholesterol efflux axis, involving lipid uptake, lipid-droplet formation, TREM2 signaling, and ABCA1/ABCG1-mediated cholesterol efflux; and (D) the putative ATP–P2 × 7–mtDNA inflammatory axis, involving extracellular ATP-induced P2 × 7 activation, Ca²⁺ influx and K⁺ efflux, mtDNA release, and activation of cGAS–STING/NLRP3 signaling.

These metabolic programs may lead to divergent functional outcomes. Adaptive remodeling supports damage sensing, debris clearance, phagocytosis, resolution of inflammation, white-matter repair, and remyelination, whereas persistent or maladaptive remodeling promotes inflammatory amplification, ROS accumulation, impaired mitophagy, foamy microglia, chronic neuroinflammation, and limited repair. These processes evolve dynamically from the acute phase through the subacute phase to the chronic phase. The resulting functional outcomes may in turn reshape the ischemic extracellular microenvironment, forming either adaptive feedback that promotes resolution or vicious cycles that perpetuate injury. Green arrows indicate adaptive or repair-associated responses, red arrows indicate pathological or injury-amplifying responses, and purple arrows indicate feedback from functional outcomes to the ischemic microenvironment.

Abbreviations: ABCA1, ATP-binding cassette transporter A1; ABCG1, ATP-binding cassette transporter G1; ATP, adenosine triphosphate; C3b, complement component 3b; cGAS, cyclic GMP–AMP synthase; CR3, complement receptor 3; MCT1/MCT4, monocarboxylate transporter 1/4; mtDNA, mitochondrial DNA; mtROS, mitochondrial reactive oxygen species; NLRP3, NOD-, LRR- and pyrin domain-containing protein 3; O₂, oxygen; P2 × 7, P2X purinoceptor 7; RET, reverse electron transport; ROS, reactive oxygen species; SDH, succinate dehydrogenase; STING, stimulator of interferon genes; TCA, tricarboxylic acid cycle; TREM2, triggering receptor expressed on myeloid cells 2.

Clinical Translation and Challenges

Although extensive preclinical evidence implicates microglial metabolic reprogramming, mitochondrial dysfunction, and changes in the extracellular milieu in IS, clinical translation remains at an early stage. The clinically relevant evidence currently available derives mainly from peripheral-blood metabolomic and lipidomic studies and observational analyses of prognosis. These indices may reflect systemic disturbances in energy metabolism and immunometabolism after stroke to some extent, but peripheral metabolites cannot be equated directly with microglia-specific metabolic changes in the brain. This section therefore focuses on the translational relevance of acylcarnitines, glycolysis-related metabolites, succinate, and soluble TREM2, while emphasizing the limits imposed by uncertain cellular origin, mechanistic specificity, and clinical interpretation.

Translational Value of Metabolism-Related Biomarkers

Peripheral Metabolites and Mitochondrial Metabolic Stress

Lactate, pyruvate, and LDH-related indices may reflect hypoxic metabolic stress and systemic energy-metabolic remodeling after stroke. Metabolomic analyses of patient samples have identified changes in lactate, pyruvate, and related energy metabolites after acute IS, consistent with a glycolytic shift and disturbed energy metabolism [80]. Further clinical studies have associated dynamic changes in serum lactate, the lactate-to-albumin ratio, LDH, and the LDH-to-albumin ratio with short-term mortality, poor functional outcomes, or recurrent stroke [81,82,83]. Succinate, a TCA-cycle intermediate, links TCA-cycle restriction, SDH-associated oxidation, and mitochondrial ROS (mtROS) generation during reperfusion [60]. Clinical evidence also associates plasma succinate with cardiovascular events and recurrent stroke after IS [84]. Lactate-related indices and succinate may therefore provide complementary information on glycolytic displacement and mitochondrial oxidative stress, respectively, for metabolic risk stratification after stroke. These circulating metabolites are nevertheless influenced by hypoperfusion, infection, hepatic and renal function, nutritional status, cardiometabolic background, reperfusion therapy, and systemic inflammation and thus cannot be interpreted as direct measures of glycolysis or TCA-cycle activity in brain microglia.

Among lipid-metabolism-related indices, acylcarnitine profiles show particular translational potential. Clinical metabolomic studies have identified changes in multiple plasma acylcarnitines in patients with acute IS, suggesting abnormalities in fatty-acid oxidation and mitochondrial energy metabolism during the acute phase. Systematic reviews likewise suggest that lipids, carnitine/acylcarnitines, and other circulating metabolites may aid diagnosis, risk assessment, and prognostic evaluation, although studies differ substantially in sample source, analytical platform, and outcome definitions [85, 86]. At the patient level, long-chain acylcarnitines are elevated on admission for acute IS and are associated with poor outcomes, supporting their potential as blood biomarkers of acute mitochondrial lipid-metabolic stress [87]. Metabolomic analyses from acute-stroke registries have also associated medium-chain acylcarnitines with cardioembolic stroke and recurrent-stroke risk, suggesting that acylcarnitine profiles may capture differences in stroke etiology and recurrence risk [88]. In addition, clinical cohort studies have associated plasma L-carnitine with cardiovascular events and recurrent stroke after IS, further supporting the potential value of the carnitine-metabolism axis in risk stratification for secondary prevention [89]. Clinically, acylcarnitine profiles are therefore better regarded as candidate biomarkers of peripheral mitochondrial fatty-acid oxidation, cardiometabolic background, and recurrence risk. Because they are affected by hepatic and skeletal-muscle metabolism, cardiometabolic status, nutrition, and medication use, they cannot be used to infer TREM2/ABCA1-mediated lipid handling in brain microglia directly.

Soluble TREM2 and Myeloid Immune Activation

Soluble TREM2 (sTREM2) may serve as a candidate marker of myeloid-cell activation and tissue-injury severity but should not be interpreted directly as a marker of microglial metabolism [90]. TREM2 is an important immune receptor in microglia and other myeloid cells and participates in lipid recognition, debris clearance, phagocytosis, cell survival, and inflammatory regulation. Mechanistic studies link the TREM2 axis to microglial metabolism. Single-nucleus RNA sequencing analyses have identified a microglial subpopulation with high Trem2 and Igf1 expression in ischemic brain tissue; this population exhibits a neuroprotective transcriptional signature and increased OXPHOS, while the Trem2-Igf1 axis remodels the microglial immunometabolic profile and promotes a neuroprotective phenotype [91]. Plasma sTREM2, however, cannot be equated directly with membrane-bound TREM2 signaling in brain microglia. In patients with acute IS, elevated plasma sTREM2 is associated with higher risks of death and cardiovascular events, and the risk of composite outcomes is greatest when both sTREM2 and galectin-3 are elevated [92]. Plasma sTREM2 has also been associated with the risk of post-stroke depression [93]. These findings have several limitations. First, the source of circulating sTREM2 is uncertain and may be influenced by central microglia, peripheral monocytes/macrophages, altered blood-brain barrier permeability, and systemic inflammation. Second, elevated sTREM2 may reflect the extent of tissue injury, phagocytic activity, or myeloid-cell activation rather than a specific change in glycolysis, fatty-acid oxidation, or mitochondrial function. Current evidence therefore supports sTREM2 as a candidate biomarker of myeloid immune activation and poor prognosis but does not establish that it specifically reflects microglial metabolic status.

Emerging Metabolic and Mitochondria-Targeted Interventions

After IS, a glycolytic shift, lactate accumulation, restricted mitochondrial oxidative metabolism, and increased mtROS provide the principal pathophysiological rationale for metabolic intervention. Clinical studies offer preliminary support for the feasibility of targeting metabolic and mitochondrial processes, but no clinical evidence yet supports a microglia-specific therapy. Dichloroacetate (DCA), an inhibitor of pyruvate dehydrogenase kinase, promotes the entry of pyruvate into mitochondrial oxidative metabolism and is therefore mechanistically relevant to post-ischemic lactate accumulation and glycolytic displacement. Direct clinical evidence in IS derives mainly from a double-blind, placebo-controlled proton magnetic resonance spectroscopy study of sodium dichloroacetate initiated 1–5 days after stroke. The lesion lactate-to-N-acetyl compound ratio tended to decrease in patients who received a higher dose or were treated earlier, but no improvement in neurological outcomes was demonstrated. DCA has therefore been investigated in stroke mainly through metabolic imaging endpoints, and evidence regarding its clinical efficacy and optimal treatment window remains insufficient.

Metformin

Metformin is among the candidate metabolic interventions with the largest body of clinical translational evidence in stroke. Systematic reviews and meta-analyses have associated prestroke metformin use with better clinical outcomes in patients with diabetes and acute IS. In a meta-analysis of seven studies involving 11,589 patients with diabetes and acute IS, Pakkam et al. found that prestroke metformin use was associated with a higher likelihood of functional independence at discharge and lower 90-day mortality, without a significant increase in symptomatic intracranial hemorrhage [94]. Another systematic review and meta-analysis published in BMJ Open included 11 studies and 18,525 patients and likewise associated prestroke metformin use with lower risks of an adverse post-stroke course and mortality. Prior metformin use has also been associated with favorable outcomes in specific clinical subgroups [95]. Among patients with diabetes and acute IS who underwent endovascular treatment, prestroke metformin use was associated with less stroke progression, a lower risk of symptomatic hemorrhagic transformation, and better functional outcomes [96]. Among patients with type 2 diabetes and IS who did not undergo mechanical thrombectomy, prestroke metformin use was also associated with less severe neurological deficits at admission and better functional status at discharge, with a stronger association in the small-vessel disease subtype [97]. These findings suggest translational potential for metformin; however, existing clinical evidence derives mainly from observational studies of prestroke exposure and remains susceptible to confounding by indication, diabetes duration and control, concomitant medication, and healthy-user bias [98]. Although small studies have explored the feasibility of initiating metformin after stroke, large multicenter randomized controlled trials with clearly defined mechanistic stratification remain lacking [99].

Alternative Energy Substrates: Ketogenic Diet and β-Hydroxybutyrate

The ketogenic diet and β-hydroxybutyrate (BHB) have attracted interest as alternative energy-substrate strategies for stroke. The rationale is that increasing ketone availability may provide the post-ischemic brain with a glucose-independent energy source while modulating oxidative stress and inflammation. Stroke-specific clinical evidence, however, remains very limited. In a multicenter study, You et al. associated urinary ketone positivity on admission with poor functional status at discharge among patients with acute IS [100]. Wang et al. further showed that urinary ketone positivity predicted all-cause mortality and poor functional outcomes in patients with acute IS or transient ischemic attack, whereas Dai et al. associated urinary ketone positivity with higher risks of stroke recurrence at 3 months and 1 year [101, 102]. Metabolomic studies have also detected elevated serum ketones and energy-metabolism-related compounds during acute IS, with levels tending to decline toward those of controls in the chronic phase, suggesting that ketone elevation is part of the systemic metabolic response to acute stroke [103]. In contrast to these risk-marker studies, Lin et al. associated BHB with favorable outcomes in patients with stroke and showed in experimental models that BHB promoted functional recovery through γ-aminobutyric acid transporter 1-dependent cortical network remodeling, suggesting possible neuroreparative effects during recovery [104]. The ketogenic diet and BHB are therefore more appropriately considered exploratory candidates for alternative-energy intervention [105]. Although exogenous BHB or ketone-ester supplementation may increase circulating ketones more rapidly, their efficacy, safety, optimal ketone range, and treatment window in patients with stroke have not been established in high-quality clinical trials.

Mitochondria-Related Redox Modulation

After ischemia-reperfusion, ROS generation, lipid peroxidation, loss of mitochondrial membrane stability, and amplified inflammation jointly contribute to injury of the neurovascular unit, making redox modulation a longstanding strategy for neuroprotection in stroke. MitoQ, SS-31/elamipretide, SkQ1, and other mitochondria-targeted antioxidants lack established clinical-efficacy evidence in patients with acute IS. Stroke-specific clinical studies have instead focused largely on antioxidant cytoprotective agents that are not mitochondria specific. Edaravone dexborneol is supported by a comparatively larger evidence base. The TASTE trial enrolled 1,165 patients with acute IS within 48 h of onset and showed that edaravone dexborneol increased the proportion of patients achieving a modified Rankin Scale (mRS) score ≤ 1 at 90 days compared with edaravone alone, suggesting potential clinical benefit from multitarget antioxidant and anti-inflammatory treatment [106]. The subsequent TASTE-SL trial showed that sublingual edaravone dexborneol increased the proportion of patients achieving an mRS score ≤ 1 at 90 days compared with placebo and had a generally acceptable safety profile [107]. In patients with large-vessel occlusion and successful endovascular reperfusion, however, intravenous edaravone dexborneol did not significantly increase the proportion achieving an mRS score of 0–2 at 90 days, suggesting that the effects of redox-targeted neuroprotection may depend on reperfusion status, lesion burden, route of administration, and patient selection [108]. Autologous mitochondrial transplantation has entered early clinical investigation as a more direct mitochondrial intervention for acute cerebral ischemia. A phase I, open-label, single-arm study reported that intravascular autologous mitochondrial transplantation during mechanical thrombectomy provided preliminary evidence of feasibility, with safety outcomes comparable to those of matched controls, but the small sample precluded conclusions regarding efficacy [109]. Thus, non-mitochondria-specific cytoprotective agents such as edaravone dexborneol have shown some clinical-efficacy signals, but a causal relationship with mitochondrial oxidative stress in microglia has not been established clinically. Mitochondrial transplantation and genuinely mitochondria-targeted antioxidants more closely align with mechanistic targeting, but their safety, therapeutic window, blood-brain barrier penetration, and effects on the National Institutes of Health Stroke Scale, infarct volume, and long-term mRS outcomes require further evaluation in acute IS.

Limitations

Microglial metabolic remodeling after IS is not a change in a single pathway but a dynamic process jointly driven by extracellular metabolic stress, mitochondrial stress, and intercellular communication. The available evidence nevertheless has substantial limitations in its applicability across models. Most mechanistic studies reviewed here were conducted in rodent stroke models, mouse primary microglia, BV2 cells, or in vitro OGD/R systems. Although these models help clarify mechanistic links among glycolysis, OXPHOS, lipid handling, mitochondrial stress, and inflammation, their cellular origins, experimental conditions, and injury contexts differ from those of human stroke. Single-cell studies have identified some conserved homeostatic gene-expression features and disease-associated response programs in mouse and human microglia, but regional heterogeneity, disease-associated subpopulation composition, and activation trajectories are not identical between species [110, 111]. In immunometabolism specifically, inflammatory stimulation increases glycolysis in both human induced microglia-like cells and mouse microglia, but the regulation of the relevant metabolic enzymes and mitochondrial oxidative responses differs [30]. Moreover, as an immortalized mouse microglial cell line, BV2 cells differ from primary microglia in baseline transcriptional profiles, inflammatory-response thresholds, proliferative state, and energy metabolism [112]. OGD/R or LPS stimulation of BV2 cells also cannot fully reproduce the complex post-stroke brain milieu of hypoxia, glucose deprivation, mitochondrial injury, and lipid-debris loading. Results from animal and in vitro models should therefore be regarded as mechanistic leads rather than direct representations of the metabolic state of microglia in the human post-stroke brain.

Conclusion and Perspectives

Overall, current evidence remains largely associative and is insufficient to identify which metabolic alterations drive microglial functional transitions after stroke. Future studies should therefore move from descriptive characterization toward causal testing. A central question is whether restoring mitochondrial metabolic adaptability and quality control in microglia can improve their clearance and reparative functions. This question can be addressed at three levels. First, stroke models should incorporate microglia-specific genetic manipulation or targeted delivery to selectively modulate OXPHOS, mitochondrial biogenesis, or mitophagy, together with measurements of metabolic flux, mtROS, phagocytosis, infarct volume, and long-term behavioral outcomes, to establish necessity and sufficiency. Second, future studies should distinguish disease phase, ischemic region, and cellular origin while integrating transcriptional state with actual metabolic activity and functional outcomes, thereby defining the roles of distinct microglial states in post-stroke injury and repair. Third, human microglia, neurovascular-unit co-culture systems, patient-derived samples, and longitudinal clinical cohorts should be used to determine whether relevant metabolic features are associated with stroke severity, inflammatory state, reperfusion status, and functional recovery. Only after cell specificity, therapeutic windows, and causal effects have been established can targeting microglial metabolism progress from a mechanistic hypothesis toward a clinically translatable stroke intervention.