The complement system is a notable component of innate immunity (45). Microglia express complement receptors, including C3aR and C5aR, and produce complement components such as C1q and C3 (45). Microglia-derived complement contributes to neuroinflammation and neurodegeneration, with increased C1q and C3 expression associated with microglial synaptic phagocytosis in neuropathological conditions, including depression and neurodegenerative disease (46). Under physiological conditions, microglia produce complement components while expressing regulatory proteins such as CD55 and CD59 to constrain excessive complement activation (47). Following CNS injury or disease, damage-associated signals induce increased complement production by microglia, thereby promoting local complement activation (47).

Complement C3 is a key component of the complement cascade and a convergence point for the classical, alternative and lectin pathways in NMOSD, with key effects on microglial activation and function (Fig. 2) (48). Dysregulated C3 expression sustains microglial activation and contributes to synaptic dysfunction and cognitive impairment. Systemic inflammation, for example, promotes microglial activation through increased C3 expression, leading to synaptic loss and cognitive deficit (8). In NMOSD, excessive C3 activation is associated with optic nerve and spinal cord injury, whereas C3 inhibition decreases complement deposition and demyelination and improves motor function in experimental models (49,50).

Under physiological conditions, astrocytes express little or no C3. In pathological states, NMO-IgG induces robust C3 production by astrocytes (51), followed by its cleavage into C3a and C3b. C3a binds C3aR and promotes NLRP3 inflammasome assembly, facilitating the maturation and release of IL-1β and IL-18 and enhancing microglial inflammatory responses (Fig. 2) (8). CDC and ADCC are also established mechanisms of astrocyte injury in NMOSD (52,53). Complement activation promotes macrophage recruitment, while macrophage-derived cytokines and ROS injure neurons and oligodendrocytes, contributing to astrocyte damage and secondary demyelination (54).

C5a is an inflammatory effector molecule (55). Elevated C5a levels in NMOSD are associated with disease activity and increased concentrations of IL-6 and TNF-α (56,57). C5a/C5aR1 signaling promotes microglial chemotaxis, proliferation and cytokine production (58), whereas astrocytic C5a receptors may enhance reactive astrogliosis and local complement production (59). These effects reinforce reciprocal astrocyte-microglia activation. C5aR1 blockade is neuroprotective in experimental neuroinflammatory settings, although its microglia-specific contribution to NMOSD remains to be defined (60). Complement cleavage also generates C5b, which initiates assembly of the C5b-9 membrane attack complex (MAC) following AQP4-IgG binding (61). MAC disrupts astrocyte membranes and contributes to tissue injury (62), distinguishing terminal complement-mediated cytotoxicity from C5a-mediated inflammatory signaling.

AQP4-IgG promotes AQP4 internalization and astrocyte injury through CDC and ADCC (63). Activated microglia secrete C1q, contributing to local neuronal and oligodendrocyte injury (64). Astrocyte-derived ATP, HMGB1 and S100B activate microglial receptors, including P2Y12 and TLR4/receptor for advanced glycation end products (RAGE), whereas microglial IL-1α, TNF and ROS can further impair astrocyte function (65). Complement signaling also regulates microglial processing of myelin debris (66). Excessive complement activation may compromise effective debris clearance and sustain demyelination, potentially via NF-κB signaling (66). Thus, complement-dependent phagocytosis should be considered in the context of both tissue injury and repair.

Several key questions remain unresolved, including the interplay between C3a and C5a signaling, their relevance to AQP4-IgG-seronegative disease and their potential utility as biomarkers. Longitudinal studies are needed to define their association with relapse and treatment response, while pathway-selective interventions should distinguish detrimental inflammatory signaling from the protective functions of complement.

ILs are multifunctional cytokines that regulate immune responses and contribute to microglial activation and neuroinflammation (Fig. 3). IL-6 is involved in the initiation and regulation of NMOSD and can modulate microglial activation (67). IL-17A activates microglia and astrocytes, inducing the production of inflammatory mediators and exacerbating neuronal injury and dysfunction (68). The IL-20 family regulates BBB integrity, oxidative stress and autophagy partly through effects on microglial activation, linking these processes with neuroinflammatory and neurodegenerative pathology (69). IL-2 exhibits region- and cell type-specific changes across neurological diseases, including multiple sclerosis (MS), Alzheimer's disease (AD), ischemic stroke and encephalitis (70). By contrast, IL-10 is a key anti-inflammatory regulator that suppresses glial activation and protects against inflammation-associated neuronal degeneration (71). Collectively, these findings underscore the complex and context-dependent functions of ILs in NMOSD.

IL-6 is produced by microglia, lymphocytes, macrophages, fibroblasts and endothelial cells in response to infection, tissue damage and inflammatory stimuli (72) (Fig. 4). Its effects on immune cell survival, inflammatory signaling and barrier function make it particularly relevant to NMOSD.

IL-6 is produced by microglia, lymphocytes, macrophages, fibroblasts and endothelial cells in response to infection, tissue injury and inflammatory stimuli (72) (Fig. 4). Its effects on plasmablast survival, stimulates AQP4 antibody production and perpetuates an autoimmune cycle make it particularly relevant to NMOSD (73). IL-6 is a regulator of neuroinflammation, with key effects on microglial activation and function (74). In microglia, IL-6 signals via the JAK/STAT pathway to regulate proliferation and inflammatory responses (74). IL-6 receptor engagement activates downstream signaling pathways that enhance microglial migration and phagocytic activity (75) (Fig. 4).

Elevated IL-6 levels during NMOSD relapses suggest its potential use as a biomarker of disease activity (76). IL-6 may promote disease progression by modulating other cytokines and immune cell populations (76). For example, IL-6 and IL-17A impair BBB integrity, thereby exacerbating CNS inflammation (77). Increasing evidence implicates dysregulated microglial activation and IL-6 production in NMOSD pathogenesis (78-80). Elevated serum and cerebrospinal fluid (CSF) IL-6 levels during relapse are associated with disease severity and brain atrophy, suggesting that IL-6 may serve not only as a marker of disease activity but also as a mediator of pathological progression (81). IL-6 promotes plasmablast survival and AQP4 antibody production, thereby sustaining the autoimmune response in NMOSD (82). It can also work with TNF-α to increase the expression of endothelial adhesion molecules, including intercellular adhesion molecule (ICAM)-1, facilitating neutrophil infiltration across the BBB (83). This compromises BBB integrity (84), while neutrophil-derived MMP-9 disrupts myelin structure (85). In addition, IL-6 inhibits nuclear translocation of oligodendrocyte transcription factor 2 in oligodendrocyte precursor cells (OPCs), impairing their differentiation into mature oligodendrocytes and decreasing myelin basic protein (MBP) synthesis (86,87). IL-6 also activates suppressor of cytokine signaling 3 in OPCs (88), interfering with insulin-like growth factor-1 (IGF-1)-mediated repair signaling (89) and restricting OPC differentiation and remyelination. Because OPC proliferation and differentiation are key for myelin regeneration, impaired microglia-OPC interactions may further compromise repair in the inflamed CNS (90). IL-6 is directly involved in inflammatory signaling (91). Binding of IL-6 to IL-6R promotes gp130 homodimerization and formation of the functional receptor complex, leading to activation of JAK1 and JAK2 and subsequent phosphorylation and activation of STAT3. Activated STAT3 translocates to the nucleus and regulates the expression of inflammation-associated genes (Fig. 4) (92). In activated microglia, IL-6 signaling via the IL-6R/gp130 complex induces downstream pro-inflammatory mediators, including TNF-α and IL-1β, establishing a positive feedback loop that sustains inflammation and neuronal injury (93). This pathway increases microglial expression of inducible nitric oxide synthase and cyclooxygenase-2, promoting the production of ROS and prostaglandin E2 and thereby aggravating oxidative stress (93). IL-6 may cooperate with TGF-β to promote differentiation of CD4⁺ T cells toward the T helper (Th)17 lineage, enhancing autoimmune responses, increasing BBB permeability and intensifying inflammation in NMOSD (94).

Key questions remain regarding the interaction between IL-6 signaling and AQP4-IgG-mediated injury, its relevance to seronegative NMOSD and the determinants of treatment response. Longitudinal comparative studies and biomarker-guided patient stratification are needed to refine IL-6-targeted therapy and assess combination approaches for refractory disease.

IL-1β primarily promotes inflammation by activating immune cells and inducing the release of additional inflammatory mediators (95). IL-1β stimulates astrocytes and microglia to produce chemokines, thereby recruiting peripheral immune cells to sites of inflammation and amplifying neuronal injury (Fig. 3) (95). IL-1β levels are notably elevated during the acute phase of inflammatory demyelinating disease, including myelin oligodendrocyte glycoprotein antibody-associated disease (MOGAD) and AQP4-IgG-positive NMOSD (96,97). This increase is particularly pronounced during acute attacks, suggesting a role for IL-1β in early disease pathology. CSF IL-1β levels are elevated in NMOSD and associated with disease severity, supporting its involvement in NMOSD pathogenesis (98). IL-1β can act in concert with other inflammatory mediators to reinforce microglial activation. It promotes the production of pro-inflammatory cytokines, including TNF-α and IL-6, and chemokines such as CCL2 and CXCL10, thereby intensifying local inflammation (99). IL-1β may promote steroid resistance through activation of T helper 17 cells, further sustaining inflammatory responses (100). In astrocytes, IL-1β binding to IL-1 receptor activates NF-κB and MAPK signaling, inducing the production of CXCL10 and CCL2. These chemokines facilitate the recruitment of peripheral neutrophils and monocytes to the CNS, thereby amplifying inflammation and contributing to IL-1β-associated neuronal and oligodendrocyte injury (101). This cytokine-driven cascade may promote NMOSD pathology. Beyond its inflammatory effects, IL-1β influences tissue repair and cellular differentiation (102). Increased BBB permeability facilitates peripheral immune cell infiltration, while experimental evidence linking IL-1β with caspase-1 activation, pyroptosis and synaptic dysfunction suggests additional mechanisms of neuronal injury (103). The relative contribution of these pathways to human NMOSD remains to be established. IL-1β may contribute to NMOSD by regulating immune cell activation and trafficking. It enhances T and other immune cell activation (104) and increases the expression of endothelial adhesion molecules, including vascular CAM-1 and ICAM-1, facilitating transendothelial immune cell migration and compromising BBB integrity (105). IL-1β may promote the production of AQP4-targeting autoantibodies, potentially reinforcing the autoimmune response characteristic of NMOSD. IL-1β can inhibit OPC proliferation and differentiation through the microRNA-202-3p/β-catenin/glioma-associated oncogene 1 pathway, decreasing the expression of myelin proteins such as MBP and impairing remyelination (106). In neurons, IL-1β induces calcium overload and mitochondrial ROS production, contributing to axonal degeneration (107). It may also act with TNF-α to increase neutrophil-derived MMP-9, disrupting myelin integrity and promoting neuronal injury (108). Evidence supporting a role for IL-1β in NMOSD is predominantly preclinical or based on cross-sectional observations (96). Its association with AQP4-IgG-mediated astrocyte injury and seronegative disease requires further clarification. Longitudinal biomarker studies and controlled evaluations of pathway-specific inhibition are needed to determine whether IL-1β serves as a reliable marker of disease activity or a viable therapeutic target.

IFNs regulate antimicrobial defense, antigen presentation and immune cell activation (109,110). Their effects in NMOSD vary according to IFN subtype, cell context and experimental model, with potential roles in either inflammatory amplification or maintenance of immune tolerance (111) (Fig. 5).

IFN-β, a type I IFN, is key for antiviral defense and is produced by various resident CNS cells, including microglia, astrocytes, neurons and endothelial cells (28). Microglia are responsive to type I IFN signaling in the CNS (112). Type I IFNs have been investigated in immune-mediated demyelinating disease, particularly MS (113), in which IFN-β exerts therapeutic effects by modulating immune cell functions (108). However, excessive or context-dependent type I IFN signaling may produce adverse effects. In a mouse model of NMOSD lesions, IFN-β exacerbates NMO-like pathological changes, including increased AQP4 and glial fibrillary acidic protein expression and expansion of the CD11c⁺ microglial subset (114). Transcriptomic analysis have identified notable type I IFN signatures in affected tissue and microglial populations (115). Microglial depletion attenuates these pathological changes without substantially altering IFN-β expression, supporting a key role for microglia as downstream responders to type I IFN signaling (116). Type I IFN-responsive microglia may contribute to the inflammatory milieu and aggravate CNS injury (116). In NMOSD, type I IFN may exacerbate disease by suppressing Th1 cell function (117).

Experimental findings, however, are not consistent across models (118,119). Type I IFN receptor deficiency improves outcomes in an induced lesion model, whereas IFN-β has limited effects in organotypic slice cultures and improved disease in an active-immunization rat model (120). These discrepancies may reflect differences between T cell- and antibody/complement-mediated mechanisms of injury. Both CD11c⁺ and CD11c⁻ microglia exhibit IFN response signatures, whereas IFN-β-associated expansion of CD11c⁺ microglia is associated with greater pathology in induced lesions (121). Clinically, IFN-β treatment is associated with NMOSD exacerbations, including severe optic neuritis and myelitis (108,122-124). Proposed mechanisms include AQP4-IgG- and astrocyte injury-dependent microglial programming (125), altered STAT1/STAT2 signaling (Fig. 5) (126) and cooperation between astrocyte-derived DAMPs and microglial TLR4/NLRP3 activation (127). Th17 responses and AQP4-specific T cells may reinforce these pathways (108). IFN-β-responsive microglia produce IL-6, TNF-α and IL-1β (128), while IL-6 promotes Th17 differentiation and supports AQP4-IgG-producing plasmablasts (129), potentially accounting for effects opposite to those observed in MS (108,130). Other proposed mechanisms include increased C1q/C3 and Fcγ receptor expression (131), MMP-9/VEGF-associated BBB disruption (132) and CXCL9, CXCL10, and CCL2-mediated leukocyte recruitment (133). Microglial production of B cell activating factor (BAFF) and a proliferation-inducing ligand (APRIL) may support autoreactive B cell survival, linking IFN signaling with persistent antibody production and astrocyte injury (134). The relative contribution of these pathways to human NMOSD lesions is uncertain.

The aforementioned findings support further investigation of the disease-specific IFN-microglia axis. Single-cell and spatial transcriptomic approaches may identify IFN-responsive microglial populations, whereas epigenetic analyses may clarify persistent inflammatory programming. CNS-penetrant JAK inhibitors and IL-6 antagonists are potential mechanistic interventions, although proposed CSF biomarkers, including IL-6 and BAFF, require independent validation.

IFN-β engages microglia through the cyclic GMP-AMP synthase/stimulator of interferon genes/type I interferon (IFN-I) signaling pathway and plays a paradoxical role in NMOSD (135), in contrast to its therapeutic effect on multiple sclerosis. Activated microglia-derived IFN-β promotes expansion and infiltration of AQP4-specific autoreactive T cells, exacerbates astrocyte injury and inflammatory pathology in NMOSD (133). Nevertheless, the functional heterogeneity of microglial subsets, species differences between rodent and human systems and stage-dependent effects of IFN signaling must be resolved before pathway modulation can be translated into NMOSD therapy.

IFN-γ is the sole member of the type II IFN family and is a notable regulator of inflammatory responses. Dysregulated IFN-γ signaling is implicated in several autoinflammatory and autoimmune diseases such as MS and NMOSD, reflecting its dual immunostimulatory and immunoregulatory functions (136). Primarily produced by Th1, CD8⁺ T and natural killer cells, IFN-γ promotes the polarization of microglia toward an inflammatory state (137). In NMOSD, IFN-γ is typically considered pathogenic due to its role in enhancing inflammatory responses and activating immune pathways targeting the CNS (Fig. 5) (108). Th1-derived IFN-γ promotes inflammatory microglial polarization via JAK1/STAT1 signaling, suppresses reparative markers such as arginase-1 and CD206 and increases the expression of pro-inflammatory cytokines and chemokines, thereby aggravating NMOSD-associated pathology (138). IFN-γ can activate the NLRP3 inflammasome through STAT1, intensifying microglial inflammatory responses (139). In addition, IFN-γ promotes CD8⁺ T cell and monocyte recruitment through the CXCL10/CXCR3 pathway, facilitating inflammatory cell accumulation within lesions and potentially exacerbating optic nerve and spinal cord injury (140). IFN-γ may induce Fas expression on oligodendrocytes, enhancing FasL-mediated apoptotic signaling (141), while activating NADPH oxidase 2 in microglia and increasing superoxide production, which contribute to myelin injury (142). IFN-γ-associated BBB disruption may facilitate the entry of autoantibodies and peripheral immune cells into the CNS, thereby promoting demyelination and neuronal injury (143).

Recent findings, however, indicate that type I and II IFNs may exert opposing effects in AQP4^201-220-induced disease (138). In this model, type II IFN decreases disease incidence and severity, challenging the predominant view of IFN-γ as pathogenic (144). Further studies suggest IFN-γ contributes to immune tolerance toward AQP4, whereas its deficiency promotes autoimmune inflammation (108,145). Decreased IFN-γ signaling may increase IL-6 expression and enhance Th17 responses, thereby facilitating autoimmune activation (136). The effects of IFN-γ are associated with peripheral immune-tolerance mechanisms that regulate autoreactive B and T cell activation (146). These observations suggest IFN-γ may exert protective effects under specific immunological conditions and highlight the context dependence of IFN signaling in NMOSD.

The divergent effects of IFN-γ underscore the limitations of extrapolating findings across experimental models and disease stages. Regional microglial heterogeneity, species differences, interactions with other cytokines and limited access to human lesion tissue complicate interpretation. Spatial profiling and functional studies are needed to distinguish inflammatory effects within CNS lesions from peripheral mechanisms by which IFN-γ contributes to AQP4-specific immune tolerance.

Disrupted lipid metabolism promotes oxidative stress and neuroinflammation (147,148). Acyl-coA synthetase long chain family member 4/vestigial like family member 4(VGLL4)-associated signaling is one example of lipid-dependent regulation of microglial function (149). In NMOSD, altered lipid handling may influence microglial response and myelin integrity (150), although much of the mechanistic evidence derives from other types of neurological disorders, particularly AD and MS (10). Lipid metabolic pathways should therefore be considered potential contributors to NMOSD pathology rather than established disease-specific mechanisms.

TREM family regulates inflammation, immune response and tissue homeostasis. TREM1 and TREM2 are of particular interest because they link myeloid cell activation with inflammatory signaling, phagocytosis and metabolic adaptation (151).

TREM1 signals through the adaptor protein tyrosine kinase-binding protein 12 (DAP12) to activate spleen tyrosine kinase (SYK) and NF-κB, promoting the release of IL-1β and TNF-α (152) (Fig. 6). TREM1 contributes to inflammatory signaling and macrophage polarization in numerous types of inflammatory disorder such as NMOSD, MS, and ankylosing spondylitis (153). By contrast, TREM2 recognizes lipid-associated signals and damaged cell material and supports microglial survival and phagocytosis (154). Through DAP12 and PI3K/Akt signaling, TREM2 enhances phagocytic activity and favors reparative microglial responses (Fig. 6) (155). Microglia in these states produce IL-10 and TGF-β, which restrain inflammation and support neuronal survival (156). Recognition of myelin debris activates SYK/PI3K-dependent lysosomal programs, facilitating debris processing in demyelinated regions and limiting the accumulation of potentially toxic lipid material (157). TREM2 may suppress TLR4/NF-κB signaling and IL-1β production, linking debris clearance with control of excessive inflammation (158). These functions are relevant to tissue repair because effective phagocytosis requires not only recognition and uptake of damaged material but also the metabolic capacity to process its lipid content (159). However, TREM2 expression alone does not define a uniform reparative state and its functional effects depend on lesion stage and the surrounding inflammatory milieu.

TREM2 is a transmembrane receptor predominantly expressed by microglia and serves as an important regulator of cell metabolism and inflammatory responses through extracellular lipid sensing (160). In AD, TREM2 signaling enhances the metabolic adaptability and survival of microglia, thereby shaping the response to amyloid-β plaques (161). TREM2 is implicated in other neurodegenerative disease; in AD and MS, it contributes to the balance between protective and detrimental inflammatory responses (162). Its effects extend to lipid metabolism, through which TREM2 maintains cell energy homeostasis and functional adaptation (163). TREM2 deficiency causes metabolic abnormality characterized by autophagosome accumulation and impaired mitochondrial function (20). TREM2 regulates mTOR signaling, thereby influencing microglial metabolism and autophagy, processes that are essential for cellular energy production and biosynthetic activity (164). In NMOSD, TREM2 deficiency impairs microglial response and exacerbates demyelinating injury (165). Acute TREM2 inhibition decreases the population of v-maf avian musculoaponeurotic fibrosarcoma oncogene homolog B-expressing microglia and impairs remyelination (166). Loss of this microglial population is associated with decreased oligodendrocyte generation, an essential process for myelin restoration (167). During NMOSD pathology, TREM2 expression is associated with microglial aggregation, proliferation and phagocytic activity (157). TREM2 deficiency aggravates tissue injury and neurological dysfunction while decreasing microglial accumulation and impairing myelin debris clearance (157), supporting a neuroprotective role for TREM2 during demyelination. TREM2-activated microglia promote OPC differentiation into mature oligodendrocytes through the release of platelet-derived growth factor BB, while enhancing the synthesis and secretion of myelin proteins such as proteolipid protein and MBP, thereby facilitating remyelination (168). TREM2-dependent debris clearance and secretion of NFs, including BDNF and glial cell line-derived neurotrophic factor (GDNF), may also contribute to extracellular matrix remodeling, reduce glial scar formation and provide a permissive environment for OPC migration and remyelination (169). TREM2 is associated with microenvironmental changes in NMOSD. Elevated CSF levels of soluble TREM2 (sTREM2) are associated with neuronal injury and other neuroinflammatory markers (neurofilament light chain, intercellular adhesion molecule 1 and IFN-γ) in patients with NMOSD (170). sTREM2 may influence microglial dysfunction through NF-κB signaling, potentially contributing to disease progression (31,171). In animal models of NMOSD, TREM2 deficiency decreases microglial phagocytic activity and myelin debris degradation, supporting its role in maintaining microglial function during demyelination (165,166,172). The interaction between TREM2 and apolipoprotein (Apo)E is key for microglial activation and lipid handling. A biphasic TREM2 response is associated with microglial state transitions, raising the possibility that TREM2 may support inflammatory responses during early inflammation while promoting repair and inflammation resolution at later stages (173). TREM1 may similarly regulate microglial activity through interconnected inflammatory pathways, but its specific contribution to NMOSD is poorly defined. TREM2 signaling in AQP4-IgG-mediated astrocyte injury and seronegative NMOSD is incompletely characterized. Larger longitudinal cohorts are needed to determine associations between TREM2 expression, sTREM2 levels, genetic variation, disease activity and prognosis. Whether TREM2 activation or inhibition is therapeutically beneficial may depend on disease stage and the specific microglial function being targeted.

ApoE is a major regulator of lipid and cholesterol transport in the CNS. Its three notable alleles, ε2, ε3 and ε4, differ in their effects on lipid handling and disease susceptibility, with ε4 associated with late-onset AD (174,175). These established functions in lipid metabolism and neuroinflammation provide a rationale for investigating ApoE in NMOSD, although genetic associations established in AD cannot be directly extrapolated to optic-spinal autoimmunity.

ApoE modulates CNS inflammatory responses in experimental MS (176) and may influence NMOSD through lipid metabolic pathways (Fig. 7) (177). Studies of astrocyte-derived extracellular vesicles and mouse models implicate the APOE/low density lipoprotein receptor-related protein 1 pathway in limiting microglial reactivity, complement activation and tissue injury (178,179). ApoE serves as a ligand for TREM2 and supports microglial survival, lesion-associated clustering and transition toward DAM states, although these responses may have inflammatory consequences (31). ApoE interactions with C1q and C3b may influence opsonization (180), whereas ATP-binding cassette transporter A1 (ABCA1)/ABCG1-dependent cholesterol efflux facilitates the removal of damaged membrane lipids (181). ApoE regulates T cell proliferation and Th1/Th17 response (182), potentially influencing the inflammatory milieu in NMOSD (183). Following astrocyte injury, increased microglial ApoE expression may partially compensate for impaired astrocytic lipid support and facilitate tissue repair (184). Associations between ApoE4 and ischemic optic nerve injury provide an additional potential mechanism but do not establish an equivalent effect in NMOSD (185). The identification of APOE/TREM2/lipoprotein lipase-expressing microglial subsets (186), alterations in CSF or serum ApoE levels (187) and early ApoE/TREM2 upregulation in experimental NMOSD (188) connect this pathway with microglial phagocytosis and lipid metabolism. ApoE/TREM2 functional and clinical value, however, requires validation.

The timing and direction of APOE/TREM2 signaling in NMOSD are uncertain. Independent human cohorts should examine associations between ApoE genetic variation, pathway activity, relapse risk and disease severity, including in seronegative NMOSD. Stage-specific therapeutic strategies require evidence that modifying lipid handling enhances debris clearance and repair without sustaining detrimental inflammatory responses. TREM2/DAP12/Syk signaling connects recognition of damaged material with phagocytosis, metabolic adaptation and regulation of complement and inflammatory pathways (Fig. 8) (189). Its protective capacity may be insufficient under intense AQP4-IgG-driven inflammation (189). AQP4-IgG-triggered complement cascade and rapid infiltration of neutrophils/eosinophils generate tissue destruction far exceeding the pathway's physiological clearance and anti-inflammatory threshold; the highly oxidative NMOSD lesion microenvironment induces excessive TREM2 ectodomain shedding, depleting functional membrane TREM2 and abrogating downstream Syk signaling; and the pathway's myeloid-restricted regulatory scope cannot directly counteract the core pathological drivers of AQP4-IgG-mediated astrocyte cytotoxicity and persistent autoantibody production. Defining the conditions under which TREM2 signaling supports recovery and determining whether genetic or metabolic factors modify these effects in NMOSD is key for evaluating its therapeutic potential.

Metabolic reprogramming is a key determinant of microglial functional state and has emerged as an important focus in neuroscience (Fig. 9) (190,191). Pro-inflammatory microglia typically exhibit increased glycolytic metabolism, whereas reparative microglia rely more heavily on oxidative phosphorylation (OXPHOS) to meet their energetic demands (192). These metabolic states are coupled to microglial function: Glycolysis provides rapid ATP production and biosynthetic intermediates to support inflammatory activation, whereas OXPHOS provides sustained energy for phagocytosis, tissue repair and homeostatic functions. The mTOR signaling pathway is a key metabolic regulator that connects cellular energy status with microglial functional phenotype (193). Modulation of mTOR-associated metabolic pathways promotes the transition from inflammatory toward reparative microglial states and may alleviate neuroinflammation and neurodegenerative pathology (193). Lipid metabolism is key for CNS homeostasis and myelin maintenance; its disruption can increase myelin vulnerability and contribute to neurological dysfunction (194). In NMOSD, impaired lipid handling may further compromise myelin integrity and nerve conduction. Microglial metabolic reprogramming may contribute to remyelination failure in NMOSD through interconnected alterations in energy metabolism, inflammatory state persistence and intercellular signaling (195). In acute NMOSD lesions, AQP4-IgG-mediated astrocyte injury activates microglial purinergic receptor P2X 7 (P2X7) and RAGE) receptors, promoting glycolytic metabolism via NF-κB and hypoxia-inducible factor (HIF)-1α signaling (196,197). Increased glycolytic activity and lactate accumulation may impair OPC differentiation and enhance inflammatory gene expression through lactylation, thereby reinforcing local inflammation (198). At later stages, mitochondrial dysfunction and impaired OXPHOS may decrease ATP availability and compromise the efficient clearance of myelin debris, limiting OPC maturation (199). Persistent metabolic imbalance may stabilize a glycolysis-dependent inflammatory state through epigenetic mechanisms, including histone lactylation and DNA methylation, restricting the transition toward an OXPHOS-supported reparative phenotype (200). Metabolic reprogramming may alter microglia-oligodendrocyte and microglia-astrocyte communication. Lactate released by glycolytic microglia activates G protein-coupled receptor 81 on OPCs and inhibits differentiation through ERK and PI3K/Akt signaling, whereas CXCL10/CXCR3 signaling between microglia and reactive astrocytes may promote glial scar formation and restrict OPC migration (201,202). Together, these metabolic and intercellular alterations may establish a persistent inflammatory microenvironment that limits remyelination in NMOSD. Direct characterization of microglial metabolism in NMOSD remains limited, with evidence derived from MS, AD and spinal cord injury models (191,203). Current experimental systems incompletely reproduce AQP4-IgG-mediated astrocytopathy (30,204), while lesion-stage heterogeneity (202,206), static metabolic measurement (207), difficulty in distinguishing resident microglia from infiltrating macrophages (208) and limited access to human lesion tissue (209) constrain mechanistic interpretation. Future studies should stratify NMOSD by disease phenotype and lesion stage and distinguish NMOSD from MOGAD rather than assuming shared metabolic mechanisms (210). Single-cell and spatial profiling, metabolic-flux analysis, induced pluripotent stem cell (iPSC)-derived models, and neurovascular co-culture systems may provide more direct evidence of cell-specific metabolic changes. Candidate interventions include itaconate derivatives, HIF-1α modulation, AMPK activation, PPARγ agonism and TREM2 modulation (211). Their therapeutic potential depends on disease stage, cell specificity and compatibility with existing immunotherapies, which require investigation in NMOSD.

Despite increasing interest in microglial function and immune-regulatory mechanisms in NMOSD (Table VI) (79,258-260), several limitations remain in disease modeling, cell resolution, functional assessment and clinical translation. Rodent microglia differ notably from their human counterparts, while passive transfer and active immunization models do not fully reproduce the chronic, relapsing features of NMOSD (Table VII) (12,99,188,261-266). The conventional M1/M2 framework oversimplifies microglial heterogeneity and fails to capture dynamic interactions with astrocytes, neurons and peripheral immune cells. MRI and PET provide valuable information at the tissue level but lack sufficient spatial resolution to resolve individual cell responses, whereas single-cell sequencing may under-represent rare populations or low-abundance transcripts. Shared molecular pathways between resident microglia and infiltrating myeloid cells complicate cell attribution and target specificity. Reliable biomarkers that reflect microglial function and treatment response are limited, and available human tissue samples typically represent advanced or end-stage lesions. These constraints restrict causal interpretation and indicate that decreased inflammatory markers alone cannot establish neuroprotection or tissue repair.

Future studies should integrate human iPSC-derived microglia, neurovascular co-culture systems and improved chronic or relapsing models with single-cell and spatial multi-omics. Such approaches may resolve microglial subsets, characterize their interactions with neighboring cells and define the temporal transition from inflammatory activation to tissue repair. Longitudinal imaging and functional assays should be combined with patient-derived samples to validate candidate biomarkers rather than infer microglial activity from non-specific inflammatory signals. Translational studies should prioritize selective intervention with adequate CNS exposure and assess outcomes that capture both relapse control and tissue recovery. Human tissue biobanks, standardized experimental protocols and collaborative data sharing may improve reproducibility and enable validation across independent cohorts. Metabolic interventions are promising, but direct evidence is needed to establish whether altering microglial energy metabolism can restore remyelination in NMOSD.