Microglia replacement therapy for neurological diseases: a mechanistic perspective focused on the Golgi apparatus

Abstract

Microglia-mediated neuroinflammation is a pathological hallmark of various neurological diseases. Altered microglial states promote disease progression and induce an inflammatory microenvironment. Therapies aimed at modulating microglial signaling demonstrate certain therapeutic effects, yet their efficacy is limited in terms of durability and may sometimes have a double-edged impact. This is presumably because disease-associated microglial dysfunction stems from its multi-layered and self-reinforcing mechanisms. Microglia replacement therapy has emerged as an approach that first depletes resident microglia and then either directly transplants microglia or induces the repopulation/engraftment of microglia in the niche. This strategy shows therapeutic effects in pre-clinical models, but the intracellular mechanisms supporting its therapeutic durability remain unclear. In this review, we present a Golgi-centered, evidence-stratified perspective, proposing that Golgi homeostasis may represent a candidate organelle-level checkpoint for post-replacement microglial state stability. As a structural and signaling hub that integrates receptor maturation, innate immune signaling, lipid handling, and secretory competence, the Golgi apparatus provides a testable framework for understanding how intracellular organization may contribute to durable microglial recalibration after replacement. Direct evidence linking microglial replacement to Golgi homeostasis remains limited, and whether Golgi recovery contributes to therapeutic durability requires future validation.

Background

Microglia-mediated neuroinflammation is a pathological hallmark of major neurological diseases, including Alzheimer’s disease (AD), synucleinopathies such as Parkinson’s disease (PD) and multiple system atrophy (MSA), intracerebral hemorrhage (ICH), ageing-related dysfunction, and brain injury [1–8]. Disease-associated microglia are not simply activated or quiescent; rather, they undergo state transitions that reshape immune signaling, receptor processing, lipid metabolism, phagocytosis, and secretory profiles [9, 10]. These changes can amplify tissue injury when inflammatory, metabolic, and intracellular stress programs become self-reinforcing.

Microglia replacement therapy has emerged as a strategy to reset the diseased microglial compartment rather than transiently suppress isolated inflammatory mediators [11]. Replacement may involve endogenous repopulation after depletion, peripheral myeloid engraftment, microglia transplantation, iPSC-derived microglia, or allogeneic brain-restricted strategies [12–16]. However, durable benefit likely depends on more than cell number. Replacement-derived cells must acquire stable microglial identity, integrate into diseased CNS niches, and maintain functional programs over time [17, 18].

Disease-associated microglial dysfunction is unlikely to be maintained by a single pathway. Several microglial signaling pathways, including TREM2-APOE signaling [19], TLR-NF-κB signaling [20, 21], and type I interferon responses [22], have been explored as therapeutic entry points, but their effects are context-dependent and may not fully reset the diseased microglial compartment [23]. These observations support a broader view in which receptor signaling, lipid metabolism, organelle stress, and secretory regulation form interconnected networks that reinforce maladaptive microglial states. From this perspective, microglia replacement is motivated not simply by the need to suppress individual inflammatory mediators, but by the possibility of resetting the microglial compartment as a whole. Whether such resetting also involves durable intracellular reorganization remains a hypothesis requiring direct validation.

Newly engrafted or repopulated microglia may encounter CNS niches characterized by persistent lipid stress, increased receptor-processing demand, antiviral or sterile inflammatory signaling, and sustained secretory pressure. Durable therapeutic benefit may therefore depend, at least in part, on whether these cells can maintain coordinated intracellular trafficking and stress-adaptation programs over time [10, 22–25]. Within this framework, the Golgi apparatus represents a potential point of convergence for these processes rather than an established therapeutic target in microglia replacement models [26, 27].

The Golgi apparatus is a highly organized and polarized organelle consisting of stacked cisternae, including cis-, medial-, and trans-Golgi compartments [28]. This polarized organization enables sequential post-translational modification, sorting and trafficking of newly synthesized proteins and lipids. In addition to being a fundamental part of secretion, the Golgi has been recognized as a dynamic signaling platform especially in immune cells where rapid responses to environmental changes are needed [29, 30]. Golgi organization is closely linked to inflammatory activation and cytokine secretion in innate immune cells, including macrophages [30, 31]. Under inflammatory or cellular stress, Golgi remodeling and fragmentation can influence signal propagation, vesicular trafficking, and cytokine release [30–32].

In this Review, we frame the Golgi apparatus as a hypothesis-generating intracellular checkpoint for replacement durability, given that receptor maturation, immune signaling, secretion, glycosylation, ion homeostasis, and lipid trafficking all converge at this organelle [26, 27, 30, 33–35]. We do not propose that Golgi biology is an established explanation for replacement efficacy. Instead, we identify Golgi-associated structural and functional features that can be directly examined in replacement-derived microglia [26].

Because direct evidence linking Golgi recovery to microglial replacement durability remains limited [11, 12, 36–39], this Review adopts an evidence-stratified framework. We distinguish direct microglial evidence [40–43], microglia-relevant receptor or trafficking evidence [44–47], mechanisms inferred from non-microglial Golgi biology, and hypotheses that require validation in replacement-derived microglia. This distinction is important because STING/NLRP3 signaling and lipid-trafficking pathways involving CD36, LPL, ABCA1, SCAP-SREBP, and PLIN2 are supported by different levels of evidence and should not be treated as equivalent [2, 48–58]. Rather than applying general Golgi biology to replacement therapy, we use these pathways to define measurable features—structural integrity, immune signaling, secretion, ion homeostasis, and lipid trafficking—for future study in replacement-derived microglia.

Microglial state plasticity and the therapeutic rationale for replacement

Microglial functions in CNS homeostasis

Microglial homeostasis is best understood as an actively maintained homeostatic program rather than a quiescent baseline. Microglia are CNS-resident macrophages with distinct developmental origins and niche-dependent identity, and in the adult CNS they continuously calibrate surveillance, inflammatory tone, and interactions with surrounding neural and vascular compartments [59, 60]. In vivo imaging studies show that homeostatic microglia extend highly motile processes through the parenchyma and respond rapidly to local ATP-dependent injury signals [61, 62]. They also monitor synaptic elements in an activity-sensitive manner and can contribute to synaptic remodeling through interaction, pruning-related programs, and trogocytosis-like processes [63–65]. Together with trophic support and apoptotic cell or debris clearance, these functions define homeostasis as an active state of niche sensing and functional restraint [59, 60].

For microglia replacement, these baseline functions become operational requirements for replacement-derived microglia rather than descriptive background [11, 12, 36, 37]. Therapeutic benefit is unlikely to depend only on cell entry, engraftment, or numerical repopulation; replacement-derived cells must also reacquire process motility, receptor responsiveness, phagocytic and lysosomal handling, lipid and metabolic balance, and secretory restraint within the diseased niche [59–67]. These capacities must remain coordinated over time so that new cells do not merely resemble microglia transcriptionally, but achieve functional integration. Disease studies linking receptor fitness and metabolic stress to microglial performance, including TREM2-related evidence, are therefore useful as a bridge to later sections, but the detailed TREM2, lipid, and Golgi mechanisms are left to Sects. "Current evidence linking Golgi dysfunction to disease-associated microglia" and "Evidence-stratified Golgi-related checkpoints for future microglial replacement studies" [44, 45, 48].

Depletion and repopulation studies provide the biological rationale for this replacement framework. Pharmacological CSF1R inhibition showed that adult microglia can be depleted and followed by robust repopulation, whereas genetic ablation revealed local self-renewing microglial clusters in the mammalian CNS [66, 67]. These studies establish compartment-level plasticity and niche-dependent reconstitution capacity, supporting the idea that diseased microglial compartments may be reset. However, repopulation alone does not guarantee durable homeostatic recalibration under chronic disease stress [11, 12, 36, 37, 66, 67]. The central question for replacement therapy is whether replacement-derived microglia can maintain homeostatic identity, functional integration, and stress resilience while facing inflammatory, proteostatic, lipid, and metabolic challenges. This logic leads directly to disease-associated state transitions in Sect. "Current evidence linking Golgi dysfunction to disease-associated microglia" and to the compartment-resetting strategy discussed in Sect. "Microglia replacement as a compartment-resetting strategy".

Current evidence linking Golgi dysfunction to disease-associated microglia

Direct evidence for structural or functional Golgi perturbation in disease-associated microglia remains limited but is not absent [40, 41, 43, 46]. Available microglial and microglia-relevant studies identify several Golgi-associated processes that are relevant to microglial dysfunction, including Golgi scaffold integrity, ER-Golgi receptor trafficking, Golgi ion homeostasis, antigen-presentation trafficking and secretory imbalance [31, 40, 41, 43–46, 68]. Together, these observations provide disease-linked perturbation evidence and define candidate Golgi-associated features for evaluation after microglial replacement. Table 1 summarizes these disease-linked observations according to disease context, Golgi-associated mechanism, evidence type, and interpretation for future replacement studies.

Table 1.

Disease-linked Golgi-associated perturbation and trafficking evidence relevant to microglial pathology

Disease/model	Golgi-associated mechanism	Evidence type	Microglial/pathological outcome	Interpretation for replacement studies
Ageing/LDAM [50]	Lipid-storage and secretory stress	Indirect microglial lipid-state evidence	Lipid-droplet-accumulating microglia show impaired phagocytosis, increased ROS, and pro-inflammatory output	Supports lipid-secretory stress as a candidate feature
AD/TREM2-related disease [44, 45, 69]	TREM2 ER-ERGIC-Golgi trafficking	Microglia-relevant receptor trafficking evidence	Impaired TREM2 maturation, ER/ERGIC cycling, and surface delivery may weaken phagocytic and metabolic fitness	Supports receptor-quality control as a candidate feature; not direct Golgi fragmentation evidence
Synucleinopathy/PD [4, 70, 71]	Golgi-associated protein machinery and HSPG-Golgi trafficking	Indirect disease-context evidence	Diseased neural tissue shows Golgi/TGN abnormalities while microglia undergo phagocytic and secretory remodeling	Plausible niche-level relevance, but not direct microglial Golgi evidence
ICH [43]	GM130/PD-L1	Direct microglial-related perturbation evidence	GM130 silencing aggravates Golgi injury, downregulates PD-L1 expression, shifts microglial polarization, and worsens BBB injury	Supports Golgi scaffold integrity as a microglia-relevant inflammatory checkpoint
HSV-1 neuroinflammation [40]	GM130/TLR3	Direct microglial evidence	HSV-1 downregulates GM130, promotes Golgi fragmentation, and suppresses TLR3-mediated innate immune responses	Supports Golgi integrity as a regulator of microglial antiviral signaling
Microglial inflammatory or metal stress [41, 72]	SPCA1/ATP2C1-dependent Golgi ion homeostasis	Direct microglial stress-model evidence	SPCA1 perturbation sensitizes microglia to Mn2+-induced Golgi fragmentation and inflammatory amplification	Supports Golgi ion homeostasis as a candidate stress-adaptation feature
EAE/inflammatory demyelination [46]	MHC-II/Ia trafficking through ER-Golgi	Microglia-like ultrastructural evidence	Ia/MHC-II reaction products are detected in ER, Golgi apparatus, and plasma membrane of reactive microglia-like cells	Supports Golgi-dependent antigen-presentation trafficking during CNS inflammation
Diabetes-associated cognitive impairment [51]	SCAP-SREBP/NLRP3	Microglial pathway-modulation evidence	Microglial SCAP deficiency reduces NLRP3-mediated neuroinflammation and improves cognitive outcomes	Supports a Golgi-linked lipid-inflammatory candidate axis requiring replacement validation
AD/ischemia/injury-associated lipid states [2, 48, 53–58, 73–78]	TREM2-cholesterol, LPL, ABCA1/APOE, lipid droplets	Indirect microglial lipid-handling evidence	Studies link microglial lipid droplets, cholesterol metabolism, ApoE lipidation, or ABCA1-dependent clearance to disease stress	Supports lipid handling and lysosomal stress as candidate features, not direct Golgi rescue evidence

Current evidence is based primarily on pathway modulation rather than direct restoration of Golgi structure

The strongest direct microglia-Golgi evidence comes from models in which Golgi-associated proteins or Golgi-localized trafficking events are linked to inflammatory outcomes [40, 41, 43, 46, 51]. In experimental intracerebral hemorrhage, GM130 silencing aggravates Golgi structural injury, downregulates PD-L1 expression, shifts microglial polarization, and worsens blood–brain barrier injury [43]. This provides disease-linked perturbation evidence linking Golgi scaffold integrity to microglia-associated inflammatory pathology and identifies GM130-associated organization as a candidate post-replacement readout. In HSV-1-infected microglia, viral downregulation of GM130 promotes Golgi fragmentation and suppresses TLR3-mediated innate immune responses [40], supporting a direct microglial link between Golgi scaffold disruption and antiviral signaling. In BV2 and primary microglia, inflammatory stimulation and Mn²⁺ stress implicate SPCA1-dependent Golgi ion handling in Golgi fragmentation and inflammatory amplification [41], making Golgi ion homeostasis a candidate stress-adaptation feature. Older EAE ultrastructural evidence further detected Ia/MHC-II reaction products in the ER, Golgi apparatus, and plasma membrane of reactive microglia-like cells, supporting Golgi-dependent antigen-presentation trafficking during CNS inflammation [46].

A second microglia-relevant layer concerns receptor trafficking rather than direct Golgi fragmentation. AD-related TREM2 studies provide disease-linked evidence that impaired ER-Golgi receptor processing can compromise microglial receptor fitness and downstream function [44, 45]. This evidence supports receptor trafficking as a candidate Golgi-related feature and motivates analysis of TREM2 maturation and surface delivery in replacement-derived microglia. The relevant maturation steps and post-replacement readouts are discussed in Sect. "Direct and microglia-relevant Golgi evidence". Microglial retromer function provides another disease-relevant trafficking layer: microglial VPS35 deficiency in 5XFAD mice impairs Aβ uptake and DAM development, reduces plaque-associated microglial responses, and exacerbates Aβ pathology and cognitive decline [79]. In PD-related synucleinopathy, microglia show phagocytic and secretory remodeling in response to α-synuclein pathology [70]. Recent evidence linking HSPG biosynthesis and Golgi-associated trafficking to α-synuclein fibril uptake in human iPSC-derived microglia further supports trafficking-related features as disease-contextual readouts for future replacement studies [71].

A third disease-oriented layer links lipid-state and metabolic stress to Golgi-relevant features. Ageing-related lipid-droplet-accumulating microglia show impaired phagocytosis, increased ROS production, and pro-inflammatory output [50], while AD-, ischemia-, spinal cord injury-, or diabetes-associated lipid-stress studies, together with TREM2-regulated cholesterol metabolism, LPL-dependent lipid uptake, ApoE secretion or lipidation, and ABCA1-mediated lipid handling, support lipid handling, secretory stress, and lysosomal adaptation as microglia-relevant features [48, 53–58, 73–76]. ICH-associated FABP4/S100A9-driven lipid droplet accumulation adds disease-linked lipid-state evidence [2], whereas microglial SCAP deficiency in diabetes-associated cognitive impairment supports SCAP-SREBP-NLRP3 signaling as a Golgi-linked lipid-inflammatory candidate axis [51]. The strength and proximity of the Golgi-related evidence vary across lipid-droplet, LPL, PLIN2, and ABCA1/APOE pathways. Collectively, these observations define testable trafficking, receptor-maturation, immune, ionic, and lipid-homeostasis features for evaluation in replacement-derived microglia.

Microglia replacement as a compartment-resetting strategy

Microglia replacement refers to strategies that remove, dilute, or functionally bypass dysfunctional resident microglia and reconstitute the CNS microglial compartment through endogenous repopulation, peripheral myeloid engraftment, or direct transplantation [11, 66, 67]. Current approaches include bone marrow transplantation-based microglia replacement (Mr BMT) [15], peripheral blood-based replacement (Mr PB) [14], microglia transplantation (Mr MT) [13], nonconditioning or reduced-conditioning approaches [12], iPSC-derived microglia [16], and allogeneic brain-restricted replacement [37].

These strategies are not interchangeable. Conditioning intensity, donor ontogeny, route of CNS access, genetic design, host niche inflammation, and immune compatibility may shape engraftment efficiency, lineage identity, functional integration, and long-term persistence. Accordingly, these design and host variables should be explicitly considered when comparing therapeutic efficacy and durability across replacement platforms [37, 80, 81] (Fig. 1).

Fig. 1.

Conceptual framework linking disease-associated microglial states, microglial replacement strategies, and candidate Golgi-associated features after replacement. Disease-associated microglial states in neurological diseases, including AD, PD, MSA, and ICH, may involve lipid dysregulation, impaired receptor maturation, secretory imbalance, organelle stress, and candidate Golgi stress, fragmentation, or trafficking dysfunction. Microglial replacement strategies may include microglial depletion or niche opening, such as CSF1R inhibition or conditioning-based depletion, followed by repopulation or donor-cell replacement. Donor-cell replacement may involve bone marrow transplantation-based microglia replacement (Mr BMT), peripheral blood-based replacement (Mr PB), microglia transplantation (Mr MT), iPSC-derived microglia, or allogeneic approaches. Because these approaches differ in the extent and mechanism of resident microglial depletion and niche opening, they may differentially shape Golgi-associated stress, trafficking demands, and post-replacement stabilization readouts. After replacement, Golgi-associated features, including protein processing and trafficking, receptor maturation, innate immune signaling, lipid regulation, and secretory balance, are proposed as candidate readouts requiring direct validation in replacement-derived microglia

Microglia replacement in neurological diseases and the Golgi-centered durability hypothesis

Determinants of post-replacement microglial identity

Replacement-derived microglia must acquire stable identity within diseased CNS environments [12, 16, 36, 37, 80, 82]. After engraftment, inflammatory, metabolic, and proteostatic cues within the diseased niche may further shape the transcriptional state, immune tone, and functional integration of replacement-derived microglia [11–16, 37, 39, 80]. Disease models suggest that replacement can reduce pathology or restore specific microglial functions [12, 16, 36, 37, 82], but durability remains an unresolved question because acute engraftment does not guarantee long-term state stability (Table 2).

Table 2.

Comparison of major microglial replacement and repopulation strategies

Method	Name	Typical workflow (key steps)	Conditioning or niche-opening strategy	Post-procedure monitoring	Main limitations
Systemic hematopoietic transplantation-driven replacement	Systemic HCT/BMT-driven microglia replacement [15, 36, 38, 39, 82]	①Harvest donor BM/HSPCs → ②Myeloablation/conditioning of recipient → ③IV transplantation → ④Long-term engraftment and CNS entry of donor myeloid cells → ⑤Assess brain chimerism	Irradiation and/or busulfan; sometimes head shielding to reduce CNS irradiation	Supportive care (antibiotics, fluids), monitor blood counts/weight; assess peripheral chimerism; potential long-term monitoring for GVHD/infection	High systemic toxicity; infection risk; GVHD (allogeneic); confounding peripheral immune reconstitution; slow CNS replacement kinetics; not purely microglia-specific
Pharmacological depletion-repopulation	CSF1R inhibitor depletion → repopulation [66, 67, 82]	①Administer CSF1R inhibitor (e.g., PLX compounds) to deplete microglia → ②Drug withdrawal → ③Endogenous repopulation from residual/niche cells → ④Phenotype/function assessment	Beyond CSF1R inhibitor exposure (dose/duration critical)	Monitor depletion/repopulation kinetics; manage off-target systemic effects; repeat cycles possible	Repopulation often derives from residual microglia (not true “replacement”); incomplete depletion; variable rebound; potential peripheral macrophage effects; disease milieu may re-imprint maladaptive states
Brain-directed transplantation after pharmacologic or brain-restricted niche opening	CNS-directed donor-cell delivery [12, 37]	① Establish a donor-cell niche through repeated CSF1R-inhibitor-mediated depletion or brain-restricted preconditioning → ② Prepare primary microglia or committed myeloid progenitors → ③ Perform brain-directed donor-cell delivery → ④ Assess engraftment, maturation, and functional integration	Repeated CSF1R-inhibitor-mediated depletion or brain-restricted preconditioning, depending on the protocol	Postoperative analgesia; monitor neurological status; sometimes transient immunomodulation; verify engraftment (reporter, qPCR, flow cytometry)	Requires neurosurgical expertise; variable engraftment without optimized niche; potential procedure-related mortality; still may need some systemic conditioning; immune rejection possible
Peripheral-source replacement with competitive advantage	Peripheral delivery with competitive advantage (conditioning-sparing) [18]	①Generate/select donor cells with survival advantage under CSF1R inhibition (e.g., resistant variants) → ②Administer CSF1R inhibitor to open niche/competitive landscape → ③Peripheral infusion (IV) → ④CNS entry/selection → ⑤Validate identity/function	Typically relies on CSF1R inhibitor “selection pressure” ± minimal conditioning	Ongoing/periodic CSF1R inhibitor; immune monitoring; verify persistence/identity and off-target peripheral effects	CNS entry efficiency can be inconsistent; identity fidelity concerns (microglia-like vs true microglia); selection pressure may have systemic effects; regulatory complexity if genetically modified
Human iPSC-derived microglia transplantation and engineering	iPSC-microglia cell therapy/programmable microglia platform [16]	①iPSC expansion → ②Directed differentiation into microglia-like cells → ③Optional genetic engineering (e.g., payload, safety switch, reporter) → ④CNS delivery (ICV/intraparenchymal) → ⑤Long-term monitoring of function/payload	Immunosuppression or immunodeficient hosts; sometimes microglia niche opening (CSF1R inhibitor)	Long-term monitoring; immunosuppression management; tumorigenicity/safety QC; tracking integration/function (imaging, single-cell, flow cytometry)	Manufacturing/QC burden; batch variability; immune compatibility; safety concerns (tumorigenicity, ectopic differentiation); regulatory barriers; high cost and expertise requirement

Beyond differences in replacement modality, the key determinant of therapeutic durability is whether replacement-derived microglia can maintain stable identity and functional integration within a chronically diseased CNS niche. This requires sustained receptor fitness, immune responsiveness, secretory restraint, metabolic balance, and homeostatic surveillance rather than transient engraftment alone. This framing connects replacement biology to organelle-level quality control.

Translational feasibility and barriers in humans

For human translation, microglial replacement may be most immediately feasible in selected monogenic microgliopathies, including CSF1R-related disease, where restoration of a defective microglial compartment is conceptually aligned with disease mechanism [17, 36, 82]. Broader application to AD, PD, MSA, ICH, or ageing-related disorders will require careful consideration of safety, timing, conditioning, and disease-stage heterogeneity [2, 50, 83].

Major barriers include safe conditioning, immunosuppression, resident microglia depletion, blood–brain barrier access, HLA matching, graft rejection, graft-versus-host disease risk, lineage stability, tumorigenicity, and functional integration within diseased human CNS niches [11–16, 37, 39, 80]. These issues are not peripheral to the Golgi hypothesis: if intracellular features are proposed as markers of replacement quality, they must be interpretable across clinically realistic donor sources, conditioning regimens, and host environments.

A Golgi-centered intracellular hypothesis for replacement durability

Current replacement studies have not directly shown Golgi recovery after microglial replacement [11, 12, 36, 37]. Golgi-related changes are therefore proposed here as candidate mechanisms or candidate features, not as established replacement-specific drivers. The main question is not whether Golgi biology explains all aspects of replacement benefit, but which Golgi-dependent structural, immune, secretory, and lipid-handling features should be tested in post-replacement microglia.

This distinction is important for interpreting the evidence base. Direct microglia and microglia-relevant studies support selected links between Golgi organization and microglial cytoskeleton, innate immune responses, receptor trafficking, ion homeostasis, lipid-inflammatory signaling, and α-synuclein uptake-related trafficking [40–47, 51, 71]. In contrast, many STING, NLRP3, SREBP-processing, PLIN2, and post-Golgi trafficking mechanisms are derived from general innate immune or non-microglial systems and remain hypothesis-generating for replacement durability [30, 49, 52, 84–95]. Thus, the purpose of the following section is to identify a limited set of Golgi-related checkpoints that are either directly supported by microglial evidence or sufficiently plausible to serve as candidate features in future replacement studies.

Evidence-stratified Golgi-related checkpoints for future microglial replacement studies

Direct and microglia-relevant Golgi evidence

Golgi architecture is considered here only insofar as it informs microglial trafficking, polarity, and state stability, particularly because local ER-Golgi systems and Golgi outposts are increasingly recognized as organizers of trafficking and cytoskeletal architecture in ramified neural and glial processes [42, 96]. The strongest structural evidence comes from studies linking Golgi organization to microglial cytoskeletal remodeling. GM130, also known as GOLGA2, is a cis-Golgi matrix protein that contributes to Golgi ribbon organization, vesicle tethering, and Golgi-associated cytoskeletal organization [28, 40, 42, 43]. In this review, GM130 is discussed not only as a structural Golgi marker but also as a microglia-relevant scaffold whose perturbation has been linked to altered TLR3-dependent innate signaling and PD-L1-associated inflammatory regulation. Homeostatic microglia contain GM130-positive Golgi outposts that support acentrosomal microtubule organization and ramified morphology, whereas inflammatory reactivity is accompanied by remodeling toward centrosomal radial microtubule arrays that enable efficient cytokine release [42]. Golgi reorientation has also been implicated in microglial migration toward Aβ deposits through the chemerin/CMKLR1-p38 MAPK pathway [97]. In chronic Parkinsonian macaques, persistently phagocytic microglia show increased Golgi apparatus protein machinery, suggesting that Golgi-related organization may accompany long-lasting phagocytic microglial states in neurodegenerative tissue [70]. These observations identify Golgi organization and Golgi outposts as structural features relevant to post-replacement microglial stability.

This structural evidence is useful because it links Golgi organization to microglial polarity, migration, phagocytic orientation, and secretory competence rather than to generic organelle morphology [42, 70, 97]. In the context of replacement, these findings should be interpreted narrowly: Golgi outposts, Golgi-associated microtubule organization, and centrosomal remodeling provide measurable structural features for assessing whether replacement-derived microglia re-establish microglia-like polarity under disease stress.

Golgi scaffold proteins provide direct microglia-specific support for immune regulation. In HSV-1-infected microglia, viral downregulation of GM130 promotes Golgi fragmentation and suppresses TLR3-mediated innate immune responses, facilitating immune escape [40]. In experimental intracerebral hemorrhage, GM130 silencing aggravates Golgi structural injury, downregulates PD-L1 expression, shifts microglial polarization, and worsens blood–brain barrier injury [43]. These findings support GM130 integrity, TLR3 signaling, and PD-L1 expression as immune-regulatory features that should be tested after replacement. The relevance of GM130 is therefore not simply that it marks Golgi integrity, but that its perturbation has been linked to defined immune outputs, including TLR3-dependent antiviral signaling and PD-L1-associated inflammatory regulation [40, 43]. This evidence positions GM130-associated Golgi scaffolding as a microglia-relevant immune checkpoint and a practical readout for testing Golgi-immune organization in replacement-derived microglia (Fig. 2).

Fig. 2.

Golgi-associated immune checkpoints in microglia. A The somatic Golgi/TGN is shown as a potential spatial platform for immune signaling checkpoints. Under regulated conditions, STING trafficking from the ER to the Golgi supports canonical cGAS-STING-TBK1-IRF3-IFN-I signaling, whereas NLRP3 inflammasome assembly at PI4P-enriched TGN membranes is depicted as a candidate innate immune checkpoint. GM130-associated Golgi integrity is also depicted as a microglia-relevant checkpoint supporting TLR3-linked innate signaling. B Under disease-associated stress, Golgi/TGN disruption is hypothesized to be associated with sustained STING/IFN-I signaling, candidate facilitation of NLRP3 inflammasome assembly, and GM130 perturbation with altered immune signaling. Solid arrows indicate established canonical pathway direction, whereas dashed arrows indicate microglia-relevant or candidate regulatory links requiring direct validation. The somatic Golgi/TGN is shown here; process-associated Golgi outposts are discussed separately

TREM2 ER-ERGIC-Golgi trafficking provides a microglia-relevant receptor-fitness checkpoint. The TREM2/DAP12 complex requires proper folding, glycosylation, ER-Golgi trafficking, surface localization, and downstream signaling competence [44, 69]. Neurodegeneration-associated mutant TREM2 proteins can undergo abortive cycling between the ER and ER-Golgi intermediate compartment, reducing mature receptor delivery and weakening phagocytic and metabolic programs [45]. Replacement studies should therefore examine whether donor-derived microglia maintain TREM2 maturation, surface localization, soluble or surface TREM2 balance, DAP12-associated signaling, and downstream phagocytic-metabolic responses under disease stress (Fig. 3).

Fig. 3.

Golgi-associated secretory and trafficking features in microglia. A Candidate homeostatic trafficking features include TREM2/DAP12 maturation and surface delivery, TGN/recycling-endosome trafficking involving synaptotagmin-11 (Syt11), HSPG/heparan sulfate biosynthesis, endolysosomal routing and membrane supply, and TGN-derived secretory vesicles. These pathways may support receptor maturation, phagosome maturation, cargo recycling, regulated secretion, aggregate handling, and membrane homeostasis. HSPG/heparan sulfate-related trafficking is also depicted in relation to α-synuclein preformed fibril (α-syn PFF) uptake. B Under disease-associated stress, impaired TREM2/DAP12 maturation and delivery, altered TGN/recycling-endosome trafficking, altered HSPG/heparan sulfate biosynthesis, disturbed endolysosomal routing or retromer stress, and altered secretory output may contribute to impaired phagocytic, metabolic, and inflammatory outputs. Solid arrows indicate relatively established trafficking directions, whereas dashed arrows indicate candidate, microglia-relevant, or stress-associated regulatory links. These mechanisms are presented as Golgi-associated trafficking features relevant to microglial function, with several components remaining candidate readouts requiring further validation in replacement-derived microglia

Golgi ion homeostasis is retained as a focused microglia-relevant checkpoint through SPCA1/ATP2C1. In BV2 and primary microglia, inflammatory stimulation increases SPCA1 expression, and SPCA1 perturbation sensitizes microglia to Mn²⁺⁻induced Golgi fragmentation with enhanced pro-inflammatory output [41]. These findings position SPCA1-dependent Golgi ion handling as a focused experimental endpoint for assessing whether replacement-derived microglia maintain ion homeostasis, Golgi morphology, and inflammatory restraint under disease stress. Earlier non-microglial work supports the role of SPCA1 in Golgi Ca²⁺ handling and should be treated as background support rather than direct microglial replacement evidence [72] (Fig. 4).

Fig. 4.

SPCA1/ATP2C1-dependent Golgi ion homeostasis as a microglia-relevant stress checkpoint. Under homeostatic conditions, SPCA1/ATP2C1-dependent Ca²⁺/Mn²⁺ transport supports Golgi luminal ion homeostasis, protein glycosylation, complex glycan processing, secretory vesicle formation, and microglial functional homeostasis. Under inflammatory or Mn²⁺ stress, inflammatory mediators and disrupted SPCA1 transport may disturb Golgi ion handling and contribute to elevated cytosolic Ca²⁺ or Mn²⁺, Golgi fragmentation or stress, perturbed SPCA1 expression or function, and inflammatory signaling amplification. Candidate consequences include altered cytokine and chemokine output, altered innate immune responses, and reduced microglial functional stability. Replacement-associated stabilization of SPCA1 expression or function, glycosylation or secretory stress, Golgi morphology, inflammatory amplification, and Mn²⁺ sensitivity is proposed as a candidate beneficial outcome that requires direct validation in replacement-derived microglia. This schematic integrates direct microglia-relevant evidence with candidate downstream interpretations

Additional trafficking and glycosylation-linked mechanisms are microglia-relevant but should be interpreted within clear evidence boundaries. Ia/MHC-II reaction products have been detected in the ER, Golgi apparatus, and plasma membrane of reactive microglia-like cells during neuroinflammation [46]. Microglia-intrinsic polysialylated proteins have been identified in the Golgi compartment and can be released after inflammatory stimulation, providing negative feedback through Siglec-E to restrain excessive microglial activation [47]. Sialylation-SIGLEC signaling more broadly represents an inhibitory glycan-recognition system relevant to microglial immune homeostasis [47, 98]. In addition, synaptotagmin-11 localizes to the trans-Golgi network and recycling endosomes in microglia and suppresses cytokine secretion and phagocytosis, including α-synuclein fibril phagocytosis [99, 100]. Emerging evidence also suggests that Golgi-dependent heparan sulfate proteoglycan homeostasis can influence α-synuclein fibril uptake in iPSC-derived microglia [71]. These pathways therefore define candidate secretory, glycosylation-linked, and trafficking-related endpoints for evaluating receptor maturation, regulated secretion, aggregate handling, and immune restraint in replacement-derived microglia (Fig. 3).

Non-microglial Golgi mechanisms as candidate inflammatory features

STING–Golgi trafficking and NLRP3–TGN scaffolding provide measurable inflammatory checkpoints derived primarily from general innate immune systems. STING activation requires ER-to-Golgi trafficking, where Golgi-associated organization supports TBK1 recruitment and type I interferon output [30, 52, 92]. Related work linking STING trafficking to LC3 lipidation, STIM1, or autophagy machinery is derived mainly from general innate immune systems and should be treated as non-microglial support [93–95] (Fig. 2).

NLRP3 can be recruited to dispersed PI4P-enriched trans-Golgi network membranes after stimulation, supporting inflammasome assembly and ASC recruitment [49]. This mechanism is also mainly established in general inflammasome biology. For future replacement studies, STING Golgi localization, TBK1 recruitment, IFN-I output, NLRP3-TGN localization, PI4P-enriched TGN dispersion, ASC speck formation, caspase-1 activation, and IL-1β release provide measurable endpoints for testing Golgi-associated inflammatory organization in replacement-derived microglia (Fig. 2).

The boundary between lipid handling and innate immune signaling may be more porous than a pathway-by-pathway framework implies. In Niemann-Pick disease type C, NPC1 loss links defective lysosomal lipid handling to STING trafficking and SREBP2-associated signaling, resulting in cGAS-independent STING activation, microglial activation, and neurodegeneration [92]. This suggests that Golgi- and lysosome-associated trafficking can couple lipid-sensing pathways to innate immune activation. In microglia, related lipid-immune interfaces may include SCAP-SREBP-dependent lipogenic activation [51], ACAT1/SOAT1-regulated control of TLR4 fate [101], ABCA1/APOE-mediated lipid efflux and lipidation [55–57, 76, 77], and disease-associated lipid-droplet programs [2, 48, 50, 53, 58, 73, 74]. Together, these observations define an integrated set of lipid-flux, receptor-trafficking, and inflammatory-signaling features for direct evaluation in replacement-derived microglia (Fig. 5).

Fig. 5.

Golgi-associated lipid-flux features in microglia under homeostatic and disease-stress conditions. A Under homeostatic conditions, microglia maintain lipid-flux balance through coordinated lipid uptake and processing, ABCA1/APOE-associated cholesterol efflux, SCAP-SREBP retention/retrieval and controlled ER-to-Golgi trafficking, and lipid-droplet turnover with lysosomal degradation. In this context, the Golgi apparatus is depicted as a lipid regulatory platform that may support receptor or lipid-protein sorting, cholesterol efflux capacity, and controlled lipogenic signaling. B Under disease-associated stress, microglia may develop candidate lipid-flux imbalance characterized by altered lipid uptake and sorting, impaired ABCA1/APOE efflux routing, SCAP-SREBP activation under metabolic stress, and lipid-droplet accumulation or persistence. Yellow spherical structures indicate lipid droplets. Golgi-lipid droplet contacts, PLIN2-associated droplet persistence, and ABCA1 post-Golgi routing are shown as candidate mechanisms, several of which are supported mainly by indirect or non-microglial evidence. Arrows indicate proposed directions of lipid flux, trafficking, or regulatory influence rather than confirmed causal pathways in replacement-derived microglia. These lipid-flux features should therefore be interpreted as candidate readouts requiring direct validation in replacement-derived microglia

Golgi-associated lipid handling as candidate metabolic features

Lipid-droplet-accumulating microglia are potentially relevant to replacement durability because they are associated with ageing, impaired phagocytosis, increased reactive oxygen species, and pro-inflammatory output [50]. This phenotype links chronic disease stress to lipid uptake, lipid synthesis, droplet persistence, efflux failure, and lysosomal or autophagy-linked degradation [50, 53, 54, 58, 73–75]. For this reason, it is treated as a microglial lipid-state feature that motivates direct testing of Golgi-linked lipid flux in replacement-derived microglia.

Recent studies further suggest that microglial lipid stress should be interpreted as a lipid-flux problem rather than simply as lipid-droplet accumulation. In AD, APOE4/4 is associated with ACSL1-positive lipid-droplet microglia, and fibrillar Aβ can induce APOE-dependent triglyceride synthesis, lipid-droplet formation, and neurotoxic microglial outputs in human iPSC-derived microglia [76]. Conversely, reducing lipid load in brain macrophages, including microglia and border-associated macrophages, enhances Aβ phagocytosis and reduces Aβ deposition in an AD mouse model [78]. These observations support the view that replacement-derived microglia should be assessed for balanced lipid uptake, utilization, storage, mobilization, and ABCA1/APOE-mediated efflux rather than for lipid-droplet burden alone [55–57, 76, 77].

Among lipid-related pathways, SCAP-SREBP provides the most direct Golgi-linked entry point because its activation is anatomically dependent on ER-to-Golgi trafficking. Under sterol-depleted or metabolic stress conditions, the SCAP-SREBP complex moves from the ER to the Golgi, where SREBP is cleaved by S1P/MBTPS1 and S2P/MBTPS2 to generate transcriptionally active nuclear fragments [84, 85]. This trafficking-dependent cleavage links Golgi access to lipogenic transcriptional programs. In microglia, abnormal SCAP translocation to the Golgi has been associated with NLRP3 inflammasome activation and cognitive impairment, making this pathway a Golgi-linked lipid-inflammatory axis that warrants testing in future replacement studies [51]. Regulators such as PAQR3, GP73/GOLM1, and C12orf49 further support Golgi control of SREBP processing in non-microglial systems, but these mechanisms should be treated as supporting biology rather than replacement-specific evidence [86–88].

Other pathways, including CD36/LPL-associated lipid uptake, PLIN2-associated droplet persistence, Golgi-lipid droplet contacts, and ABCA1/APOE-mediated efflux, are biologically plausible but supported by different levels of evidence [2, 48, 54–57, 75–78]. LPL, TREM2-cholesterol metabolism, FABP4/S100A9-driven lipid-droplet accumulation after intracerebral hemorrhage, APOE4/ACSL1-associated lipid-droplet microglia, and lipid-load reduction studies provide microglia-relevant disease or functional evidence [2, 48, 54, 75, 76, 78]. By contrast, PLIN2-Cideb-Golgi mechanisms are derived mainly from hepatocyte studies [90], and ABCA transporter post-Golgi routing is supported by non-microglial trafficking evidence [91].

These lipid-state and trafficking observations raise a focused experimental question for replacement studies: whether Golgi-linked lipid sensing, processing, and trafficking enable replacement-derived microglia to preserve balanced lipid flux under disease stress through coordinated lipid uptake, mobilization, lysosomal degradation, and ABCA1/APOE-mediated efflux.

Summary of evidence boundaries

Taken together, the strongest microglia-relevant support comes from Golgi outposts and microtubule remodeling, GM130-associated immune regulation, TREM2 ER-Golgi trafficking, SPCA1-dependent Golgi integrity, and SCAP-associated inflammatory lipid signaling. STING-Golgi trafficking, NLRP3-TGN scaffolding, LPL sorting, PLIN2-associated Golgi-lipid-droplet interactions, and ABCA1 post-Golgi routing are retained as candidate mechanisms that define testable post-replacement readouts and priorities for direct validation. This distinction defines the experimental agenda for the field.

Experimental framework and future directions

Future studies should test Golgi-related experimental endpoints alongside conventional inflammatory, behavioral, histological, transcriptomic, and lineage endpoints used in microglial replacement studies [11, 12, 16, 36, 37]. The central experimental question is not simply whether the Golgi apparatus appears more intact after microglial replacement, but whether changes are associated with durable functional recalibration of replacement-derived microglia under disease stress. Recent discussions of microglial microtubule remodeling further support the value of assessing Golgi-associated cytoskeletal organization as part of future neuroinflammation studies [102]. Therefore, Golgi experimental endpoints should be assessed in parallel with microglial identity, inflammatory responsiveness, lipid-handling capacity, phagocytic function, and long-term persistence (Table 3).

Table 3.

Experimental parameters for testing Golgi-related mechanisms in microglial replacement studies

Parameter class and key references	Representative parameters	Experimental purpose	Interpretation boundary
Structural Golgi parameters [28, 42, 70, 96, 97, 102]	GM130/GOLGA2; TGN46/TGOLN2; Golgi compactness, fragmentation, dispersion, and perinuclear organization; process-associated Golgi outposts; Golgi orientation toward phagocytic or migratory fronts; centrosomal versus acentrosomal microtubule organization	Test whether replacement-derived microglia re-establish microglia-like Golgi architecture, polarity, trafficking orientation, and cytoskeletal organization under baseline and disease-stress conditions	Structural improvement should be interpreted as a measurable feature rather than proof of therapeutic causality unless linked to lineage stability and functional recovery
Immune/secretory parameters [30, 31, 40, 43–47, 49, 52, 69, 92–95, 98–100]	GM130-related TLR3 signaling; GM130/PD-L1-associated inflammatory regulation; TREM2 maturation and surface delivery; MHC-II/Ia trafficking; polysialic acid-Siglec-E signaling; Syt11-associated TGN/recycling-endosome trafficking; STING/TBK1 localization; NLRP3/ASC assembly; caspase-1 activation; IL-1β, TNF, and IFN-I output	Test whether replacement-derived microglia maintain receptor maturation, antigen-presentation trafficking, innate immune responsiveness, cytokine secretion, and inflammatory signal resolution	GM130, TREM2, Syt11, and MHC-II/Ia are microglia-relevant features; STING/NLRP3-TGN mechanisms remain candidate checkpoints unless directly validated in replacement-derived microglia
Lipid/metabolic parameters [2, 48, 50, 51, 53–58, 73–78, 84–91]	Lipid-droplet (LD) burden; CD36/LPL-associated lipid uptake; TREM2-regulated cholesterol metabolism; SCAP-SREBP activation; S1P/S2P processing; PLIN2-associated droplet persistence; ABCA1/APOE-mediated efflux; APOE lipidation; lysosomal degradation; lipid-load reduction; ROS production; phagocytosis	Test whether replacement-derived microglia preserve lipid flux, cholesterol handling, lipid-droplet mobilization, lysosomal clearance, metabolic adaptation, and phagocytic competence under disease stress	Lipid-droplet accumulation alone does not prove Golgi dysfunction. PLIN2-Golgi contacts and ABCA1 post-Golgi routing include indirect or non-microglial evidence and should be treated as candidate readouts
Durability and lineage parameters [11, 12, 16, 36–39, 80–82]	Donor-cell tracing; persistence over time; single-cell transcriptomic stability; chromatin-state profiling; stress-challenge response; homeostatic signature recovery; disease-associated state suppression; relapse of maladaptive microglial states; long-term engraftment and functional integration	Test whether Golgi-related structural, immune, secretory, and lipid-handling changes persist as stable recalibration rather than transient adaptation after replacement	These parameters are needed to distinguish driver, marker, and consequence relationships. Durable Golgi-associated changes should not be interpreted as causal unless experimentally linked to replacement benefit

A practical validation framework should include comparisons among replacement-derived microglia, diseased resident microglia, and, where possible, homeostatic control microglia. Experimental endpoints should be measured under baseline conditions and after disease-relevant stress challenges, including inflammatory stimulation, lipid overload, protein aggregates, hemoglobin-derived products, metal stress, or metabolic stress. In this framework, structural Golgi organization, immune and secretory recalibration, lipid-metabolic adaptation, and lineage stability are interpreted together rather than as isolated endpoints.

Operationally, structural imaging should include cis-Golgi scaffold GM130/GOLGA2, the trans-Golgi network marker TGN46/TGOLN2, Golgi compactness, fragmentation, and process-associated Golgi outposts, while functional assays should pair these markers with TREM2 maturation and surface delivery, GM130-TLR3/PD-L1 signaling, SPCA1-linked ion stress, SCAP-SREBP activation, lipid droplet burden, ABCA1/APOE efflux, and STING/NLRP3 localization, as summarized in Sects. "Direct and microglia-relevant Golgi evidence"-"Golgi-associated lipid handling as candidate metabolic features". Longitudinal lineage tracing, single-cell transcriptomics, chromatin-state profiling, and repeated stress-challenge experiments are needed to determine whether Golgi-related changes persist as stable recalibration or merely reflect transient adaptation. This design is intended to distinguish driver, marker, and consequence relationships rather than assume that Golgi recovery explains durable therapeutic benefit.

These experimental endpoints should be interpreted within explicit evidence boundaries. Future studies should integrate structural Golgi recovery, STING/NLRP3 activity, lipid-trafficking changes, lineage stability, and disease-outcome measurements to determine whether Golgi-related features act as drivers, markers, or consequences of durable microglial recalibration.

Conclusions and future prospects

By using an evidence-stratified framework, this Review proposes that selected Golgi-related structural, immune, secretory, ionic, and lipid-handling measurable features can be tested as markers or candidate mechanisms of replacement durability. Current evidence does not yet prove Golgi recovery after microglial replacement, but it identifies several microglia-relevant entry points, including Golgi outposts, GM130-associated immune regulation, TREM2 trafficking, SPCA1-dependent Golgi integrity, and SCAP-associated inflammatory lipid signaling. STING, NLRP3, PLIN2-Golgi-lipid droplet contacts, and ABCA1 routing remain largely hypothesis-generating and should be evaluated in future replacement studies.

Future work should directly determine whether replacement-derived microglia exhibit improved Golgi organization, receptor trafficking, inflammatory signal resolution, lipid handling, and lineage stability in diseased CNS niches. Most importantly, these studies should clarify how Golgi-related measurable features relate to durable microglial recalibration and replacement benefit in disease-relevant CNS niches. This evidence boundary keeps the Golgi-centered model focused, experimentally tractable, and aligned with the central therapeutic question of microglia replacement durability.

Abbreviations

ABCA1

ATP-binding cassette transporter A1

AD

Alzheimer’s disease

APOE

Apolipoprotein E

ASC

Apoptosis-associated speck-like protein containing a CARD

ATP2C1

ATPase secretory pathway Ca²⁺/Mn²⁺ transporting 1

Aβ

Amyloid-β

α-syn

α-Synuclein

BM

Bone marrow

BMT

Bone marrow transplantation

Ca²⁺

Calcium ion

cGAS

Cyclic GMP-AMP synthase

CNS

Central nervous system

COPI

Coat protein complex I

CSF1R

Colony-stimulating factor 1 receptor

DAP12

DNAX-activating protein of 12 kDa

ER

Endoplasmic reticulum

ERGIC

ER-Golgi intermediate compartment

GM130

Golgi matrix protein 130

GOLGA2

Golgin subfamily A member 2

GP73

Golgi phosphoprotein 2

GVHD

Graft-versus-host disease

HCT

Hematopoietic cell transplantation

HDL

High-density lipoprotein

HLA

Human leukocyte antigen

HSPCs

Hematopoietic stem and progenitor cells

HSPG

Heparan sulfate proteoglycan

HSV-1

Herpes simplex virus type 1

ICH

Intracerebral hemorrhage

ICV

Intracerebroventricular

IFN

Interferon

IFN-I

Type I interferon

IL-1β

Interleukin-1β

IL-21

Interleukin-21

iPSCs

Induced pluripotent stem cells

IV

Intravenous

LC3

Microtubule-associated protein 1 light chain 3

LD

Lipid droplet

LDAM

Lipid-droplet-accumulating microglia

LPL

Lipoprotein lipase

LPS

Lipopolysaccharide

MBTPS1

Membrane-bound transcription factor peptidase, site 1

MHC-II

Major histocompatibility complex class II

Mn²⁺

Manganese ion

Mr BMT

Bone marrow transplantation-based microglia replacement

Mr MT

Microglia transplantation

Mr PB

Peripheral blood-based microglia replacement

MSA

Multiple system atrophy

NF-κB

Nuclear factor kappa B

NLRP3

NOD-like receptor family pyrin domain containing 3

PAQR3

Progestin and adipoQ receptor family member 3

PD

Parkinson’s disease

PD-L1

Programmed death-ligand 1

PFF 

Preformed fibril

PI4P

Phosphatidylinositol-4-phosphate

PLIN2

Perilipin 2

PPARα

Peroxisome proliferator-activated receptor alpha

qPCR

Quantitative polymerase chain reaction

QC

Quality control

Rab6

Ras-related protein Rab6

ROS

Reactive oxygen species

S1P

Site-1 protease

S2P

Site-2 protease

SCAP

SREBP cleavage-activating protein

SPCA1

Secretory pathway Ca²⁺/Mn²⁺-ATPase 1

SREBP

Sterol regulatory element-binding protein

STIM1

Stromal interaction molecule 1

STING

Stimulator of interferon genes

Syt11

Synaptotagmin-11

TBK1

TANK-binding kinase 1

TFEB

Transcription factor EB

TGN

Trans-Golgi network

TLR

Toll-like receptor

TNF

Tumor necrosis factor

TREM2

Triggering receptor expressed on myeloid cells 2

VLDL

Very-low-density lipoprotein

VPS35

Vacuolar protein sorting-associated protein 35

WT

Wild type

Authors’ contributions

Shuwen Deng drafted the manuscript. Hong Jiang conceived the review and supervised the work. Rong Qiu and Jinze Wu supervised the figures and tables. All authors read and approved the final manuscript.

Funding

This study was funded by the National Key R&D Program of China (No. 2021YFA0805200 to H Jiang), the National Natural Science Foundation of China (U25A2058 and 82571433 to H Jiang; 82301518 to Shuwen Deng), the Scientific Research Program of FuRong Laboratory (Nos. 2025PT5014 and 2025PT5001 to H Jiang), the Natural Science Foundation of Hunan Province (No. 2024JJ3050 to H Jiang), the Science and Technology Innovation Program of Hunan Province (No. 2026RC9017 to H Jiang), the Major Scientific Research Project for High-level Health Talent in Hunan Province (No. R2023047 to H Jiang), the MIIT Project for Open Competition and Leadership in AI Medical Device Innovation (No. 2025-AISJK-01009 to H Jiang), the Central South University Research Programme of Advanced Interdisciplinary Study (No. 2023QYJC010 to H Jiang), and the Central South University Clinical Research Zhang Xiao-qian Program (No. ZXQ2026A08 to H Jiang).

Data availability

No datasets were generated or analysed during the current study.

Declarations

Ethics approval and consent to participate

Not applicable.

Competing interests

The authors declare no competing interests.