Эпигенетическая ось Лактата-H4K12la-HDAC1 регулирует переход к микроглиальному состоянию во время экспериментального аутоиммунного энцефаломиелита у женских мышей.

Introduction

Multiple sclerosis (MS) is a chronic autoimmune inflammatory disease affecting the brain and spinal cord 1 , 2 . Emerging evidence suggests that dysfunctional innate immunity, particularly involving microglia plays a critical pathogenic role in MS 3 – 5 . Microglia exhibit remarkable functional plasticity and are capable of adopting both pro-inflammatory and neuroprotective phenotypes depending on the microenvironmental context 6 , 7 . The mechanisms governing this functional switching are complex, but emerging evidence highlights metabolic and epigenetic reprogramming as a central regulatory hub 7 . However, the precise molecular mechanisms that govern these state transitions, especially those that sustain long-term functional changes during chronic inflammation, remain incompletely understood.

Recent studies suggest that metabolic adaptation is closely intertwined with microglial activation and function. Beyond serving as an energy source, metabolic intermediates can influence transcriptional programs and cellular identity. Among these, lactate has emerged as a metabolite of particular interest 8 . Once regarded primarily as a byproduct of glycolysis, lactate is now recognized as a signaling molecule capable of modulating immune cell behavior in a context-dependent manner 1 , 9 – 11 . In chronic inflammatory and autoimmune conditions, lactate accumulation has been linked to altered immune response, although its precise role in the inflamed central nervous system (CNS) remains unclear. Intriguingly, despite this association between endogenous lactate accumulation and severe disease states, prophylactic administration of exogenous lactate has been shown to markedly attenuate disease severity in the experimental autoimmune encephalomyelitis (EAE) model 10 . This apparent paradox suggests that while endogenous lactate accumulation may reflect metabolic stress or an insufficient adaptive response to injury, pharmacologically boosting this pathway could drive a latent protective mechanism. However, how sustained lactate availability specifically rewires microglial function during autoimmune neuroinflammation remains to be elucidated.

Epigenetic reprogramming has been proposed as the molecular mechanism responsible for long-term functional modification of the innate immune cells 12 . Histone lactylation, a post-translational modification, has emerged as a critical link between cellular metabolism and gene transcriptional regulation 13 – 15 . This modification involves the enzymatic transfer of a lactyl group from lactate-derived lactyl-CoA to histone lysine residues, directly coupling metabolic flux to chromatin remodeling and gene expression 16 . Recent studies have revealed that histone lactylation serves as a key mechanism of metabolic-epigenetic coupling across diverse biological contexts 17 . Notably, different histone lactylation sites appear to regulate distinct gene programs in a context-dependent manner. For example, H3K9la has been associated with pro-inflammatory gene expression in Parkinson’s disease 18 , while H3K18la enhances neuroprotective responses in cerebral ischemia 19 . In spinal cord injury models, exogenous lactate supplementation enhances histone lactylation in microglia 20 , increasing the expression of genes involved in debris clearance, scar stabilization, and axonal regeneration 21 . These findings suggest that histone lactylation may play a dual role in both inflammatory regulation and tissue repair. However, despite these advances, critical mechanistic questions remain unanswered. It is still unclear whether lactate-mediated histone lactylation in microglia operates in a disease-specific manner and whether it differentially influences homeostatic or reparative phenotypes in chronic neuroinflammatory diseases such as MS. Furthermore, the temporal dynamics, cell-type specificity, and functional reversibility of histone lactylation in the inflamed CNS remain largely undefined. Addressing these gaps is essential for determining whether targeting lactate-driven epigenetic remodeling represents a viable therapeutic strategy for autoimmune neuroinflammation.

In this study, we sought to characterize the epigenomic landscape associated with histone lactylation in microglia during autoimmune neuroinflammation. By integrating patient-derived data with in vivo EAE models, pharmacological perturbations, and single-cell resolution analyses, we identified a specific lactate-driven epigenetic axis that shapes microglial functional programs. We demonstrate that lactate promotes a shift toward a restorative phenotype through specific histone modifications, providing mechanistic insight into the metabolic control of microglial states and identifying potential therapeutic targets for modulating chronic neuroinflammation.

Results

Elevated CSF lactate correlates with disease severity and drives microglial lactylation in neuroinflammation

To investigate whether lactate acts as a functional metabolic signal in neuroinflammation, we measured lactate levels in cerebrospinal fluid (CSF) from 20 patients with MS and 20 non-neurological disease controls. Clinical characteristics and demographic information for MS patients and control subjects are summarized in Supplementary Tables S1 – S3 . CSF lactate concentrations were significantly elevated in MS patients compared with controls (Fig. 1a ). To model these metabolic changes experimentally, we next examined lactate dynamics in the EAE mouse model. Lactate production was significantly increased in the spinal cord tissues of EAE mice compared to controls (Fig. 1b ). Correlation analysis (Fig. 1c ) revealed a moderate positive association between CSF lactate levels and Expanded Disability Status Scale (EDSS) scores (r = 0.45), whereas no significant correlation was observed with peripheral acute inflammatory markers, including IL-6 or CRP (Supplementary Fig. 1a ). Fig. 1 Elevated lactate levels and cell-type specific protein lactylation in microglia during neuroinflammation. a Quantification of lactate concentrations in the CSF of MS patients ( n = 20 individuals) compared to non-neurological disease controls ( n = 20 individuals). Each dot represents one individual. b L-Lactate concentrations in spinal cord tissues from acute EAE mice and control mice ( n = 5 mice/group). Each dot represents one mouse. c Heatmap of Spearman’s rank correlation matrix between CSF lactate levels and clinical parameters in MS patients ( n = 20 individuals). Color intensity represents the Spearman correlation coefficient (ρ). No adjustment for multiple comparisons was made. Annotated P values: CSF Lactate vs EDSS ρ = 0.447, P = 0.083; vs Age ρ = 0.384, P = 0.095; EDSS vs Duration ρ = 0.489, P = 0.055; EDSS vs Age ρ = 0.612, P = 0.012; CSF WBC vs Age ρ = −0.559, P = 0.010. Representative immunofluorescence images ( d ) and quantification ( e ) of Pan-Kla (red) and Iba1 (green) in spinal cord ( n = 8 mice/group). Representative immunofluorescence images ( f ) and quantification ( g ) of Pan-Kla (red) and GFAP (green) in spinal cord ( n = 7 mice/group). Representative immunofluorescence images ( h ) and quantification ( i ) of Pan-Kla (red) and NeuN (green) in spinal cord ( n = 8 mice/group). Data are presented as mean ± SEM. Statistical significance was determined by an unpaired, two-tailed Student’s t test, except for ( a ) where a two-tailed Mann–Whitney U test was applied due to non-normal data distribution. Detailed statistical parameters are provided in the “Methods” section. Source data are provided as a Source Data file. Scale bars, 50 μm (applies to all image panels). Source data are provided as a Source Data file.

Stratified analysis based on clinical disease stage revealed a significant positive correlation between CSF lactate and EDSS scores in patients with chronic MS (r = 0.70, P = 0.011), whereas no such association was detected in patients classified as acute MS (Supplementary Fig. 1b ). Notably, this positive correlation does not imply that lactate drives pathology; rather, it suggests that endogenous lactate accumulation serves as a biomarker of unresolving metabolic distress and tissue hypoxia in chronic MS. This indicates an insufficient endogenous compensatory response, highlighting the necessity for robust pharmacological intervention to harness lactate’s latent neuroprotective potential.

Given that lactate has been implicated as a substrate for lysine lactylation, we assessed global lysine lactylation (Pan Kla) levels to determine whether elevated lactate is accompanied by corresponding changes in protein modification. Western blot analysis, utilizing Total Protein Normalization (TPN) via Coomassie Brilliant Blue (CBB) staining as a rigorous loading control, revealed a marked upregulation of Pan Kla in the spinal cords of EAE mice relative to controls (Supplementary Fig. 1c ). Densitometric quantification of the entire Pan Kla lanes, normalized to total protein density, confirmed a statistically significant increase in global Pan Kla levels in EAE mice compared to controls (Supplementary Fig. 1c ; P < 0.05, two-tailed Mann–Whitney U test).

To determine the cellular distribution of this increase, we performed immunofluorescence co-staining for Pan-Kla together with markers for microglia (Iba1), astrocytes (GFAP), and neurons (NeuN) in spinal cord lesions. Quantitative analysis demonstrated a significant increase in the density of Pan Kla+Iba1+ microglia in EAE mice compared to controls (Fig. 1d, e ; P < 0.001). In striking contrast, the densities of Pan Kla+ astrocytes (Fig. 1f, g ) and Pan Kla+ neurons (Fig. 1h, i ) were not significantly altered between groups. This robust Pan-Kla signal in microglia was observed in both the cytoplasm and the nucleus, which is consistent with the broad nature of the Pan-Kla antibody that targets lactylated residues on both metabolic enzymes and histones. These findings indicate that, during neuroinflammation, increased protein lactylation is predominantly observed in microglia, suggesting a cell-type-specific metabolic responsiveness to elevated lactate within the inflamed CNS.

Glycolytic activation selectively couples to H4K12la in CNS myeloid cells

Building on our observation that lactate accumulation and protein lactylation are preferentially enriched in microglia during neuroinflammation, we next sought to identify the upstream metabolic alterations that may contribute to this lactate-rich microenvironment. To this end, CD11b + cells were isolated from the spinal cords of EAE mice and controls using magnetic bead-based sorting (Fig. 2a ). As CD11b captures the broader CNS myeloid compartment, including resident microglia and infiltrating myeloid cells, this strategy was employed as an initial, unbiased screening approach to survey inflammation-associated metabolic remodeling. Fig. 2 Glycolytic activation and specific H4K12la are upregulated in CNS myeloid cells during EAE. a Schematic overview of the isolation of CD11b⁺ cells (representing the broader CNS myeloid compartment) from spinal cords for proteomic and RNA-seq analyses. b Proteomic analysis of the isolated CD11b⁺ fraction showing enriched KEGG pathways in EAE mice. Dot size represents protein number. Color indicates P value. c Heatmap displaying relative expression levels of key glycolytic enzymes identified by proteomic analysis in CD11b⁺ cells. Representative Western blots ( d ) and quantification ( e ) of HK2 and LDHA in the CD11b⁺ fraction during the acute phase. Relative expression was normalized to tubulin. Each dot represents one mouse ( n = 6 mice/group). Representative Western blots ( f ) and quantification ( g ) of H4K12la, H3K9la, H3K14la, and H3K18la in spinal cord tissues across disease stages. Relative expression was normalized to total Histone H4. Each dot represents one mouse ( n = 5 mice/group). Data are presented as mean ± SEM. Representative immunofluorescence images of Iba1 (green) co-stained with H4K12la ( h ), H3K14la ( i ), H3K18la ( j ), and H3K9la ( k ) in red. Scale bars, 50 µm (applies to all image panels). Data are presented as mean ± SEM. Statistical significance was determined by an unpaired, two-tailed Student’s t test for two-group comparisons ( e ) and a one-way ANOVA followed by Tukey’s multiple comparisons test for multiple groups ( g ). Exact P values: e HK2 P < 0.0001, LDHA P < 0.0001; g H4K12la ANOVA P = 0.0001, Control vs. Acute EAE adjusted P = 0.0003, Control vs. Chronic EAE adjusted P = 0.0008; H3K9la ANOVA P = 0.0437, Control vs. Chronic EAE adjusted P = 0.0343. Detailed statistical parameters are provided in the “Methods” section. Source data are provided as a Source Data file.

To identify initial metabolic remodeling at the peak of neuroinflammation, cells were harvested from EAE mice during the acute phase (Day 21 post-immunization). Quantitative proteomic analysis revealed a significant enrichment of glycolysis-related proteins in the EAE group relative to controls (Fig. 2b ). Notably, several rate-limiting glycolytic enzymes, including lactate dehydrogenase A (LDHA), hexokinase 2 (HK2), and pyruvate kinase M (PKM), were significantly upregulated (Fig. 2c ). Immunoblotting of the CD11b⁺fraction further confirmed elevated protein levels of HK2 and LDHA in EAE mice (Fig. 2d, e , P < 0.001), consistent with enhanced glycolytic activation and an increased capacity for lactate generation within the inflamed CNS myeloid compartment.

We next investigated whether this metabolic shift was accompanied by corresponding changes in lactate-derived epigenetic modifications. We systematically screened for dynamic changes in several histone lactylation marks (H4K12la, H3K18la, H3K9la, and H3K14la) in spinal cord tissues across disease stages (Fig. 2f ). Among these sites, H4K12la exhibited the most robust and persistent upregulation, being significantly elevated during the acute phase and sustained into the chronic phase (Fig. 2g , P < 0.001). In contrast, H3K9la showed a downward trend, while H3K18la and H3K14la remained largely unchanged. Immunofluorescence co-staining further validated this specificity. While H3K18la, H3K9la, and H3K14la showed minimal or inconsistent changes across cell types, H4K12la signals were prominently induced and strictly co-localized within Iba1+ cells within EAE lesions (Fig. 2h–k ). Confirmed by high-resolution confocal microscopy, H4K12la exhibited a selective nuclear-localized pattern overlapping with DAPI, contrasting with the broader nucleocytoplasmic distribution of global Pan-Kla (Supplementary Fig. 1d, e ). This localization aligns with its role as a specific histone modification. These findings identify H4K12la as the dominant lactate-associated histone modification and suggest that glycolytic reprogramming is selectively coupled to a defined epigenetic output in microglia.

Microglial Ldha deletion impairs H4K12la and exacerbates EAE pathogenesis

To elucidate the role of microglial LDHA in the pathogenesis of EAE, we generated mice with an inducible conditional deletion of Ldha in Cx3cr1 -positive microglia ( Ldha flox/flox ; Cx3cr1 CreERT2/+ , Ldha icKO) using a specific breeding strategy (Supplementary Fig. 2a ). The genotype was confirmed by PCR analysis (Supplementary Fig. 2b ). For inducible Ldha ablation in microglia, adult Ldha icKO and control mice received daily intraperitoneal injections of tamoxifen (TMX) for 5 consecutive days, with the final injection 14 days before EAE induction. Subsequently, EAE was induced in TMX-treated icKO and Ldha flox/flox control mice ( n = 8 per group) and disease progression was monitored daily (Fig. 3a, b ). Fig. 3 Microglia-specific Ldha deletion exacerbates EAE pathogenesis and reduces H4K12 lactylation. Breeding strategy ( a ) and experimental timeline ( b ) for inducible microglial Ldha knockout and EAE induction. Clinical EAE scores ( c ) and AUC quantification ( d ) in tamoxifen-treated Ldha icKO and control mice ( n = 8 mice/group). Representative H&E staining ( e ) and quantification of inflammatory cell infiltration density ( f ) in spinal cords ( n = 8 mice/group). Representative Nissl staining ( g ) and quantification of Nissl-positive neuron density ( h ) in spinal cords ( n = 8 mice/group). Representative LFB staining ( i ) and quantification of LFB scores ( j ) in spinal cords ( n = 8 mice/group). Representative immunofluorescence images ( k ) and quantification of relative H4K12la density ( l ) and the number of H4K12la⁺ Iba1⁺ cells ( m ) in the spinal cord acute phase ( n = 8 mice/group). Clinical EAE scores ( n ) and AUC quantification ( o ) in EAE mice treated with MS-275 or PBS ( n = 6 mice/group). Representative immunofluorescence images ( p ) and quantification ( q ) of H4K12la⁺ Iba1⁺ cell density in spinal cord lesions following MS-275 treatment ( n = 6 mice/group). Scale bars, 50 μm (applies to all image panels). Data are presented as mean ± SEM. Statistical significance was determined by an unpaired, two-tailed Student’s t test. Exact P values: d P = 0.0060; f P = 0.0290; h P = 0.0334; j P = 0.0486; l , m P < 0.0001; o P = 0.0002; q P = 0.0234. Source data are provided as a Source Data file.

To confirm the efficiency and specificity of the gene deletion, we first performed validation experiments on naive mice. Western blot analysis showed a significant reduction in LDHA protein levels in the cortex, hippocampus, cerebellum, and spinal cord of Ldha icKO mice compared to controls (Supplementary Fig. 2c ). Furthermore, immunofluorescence co-staining confirmed that LDHA expression (red) was specifically depleted in Iba1-positive microglia (green) in Ldha icKO mice (Supplementary Fig. 2d ). We next assessed whether microglial-specific Ldha deletion induced pathological changes under normal physiological conditions. Histological analysis of naive mice revealed no significant differences in inflammatory infiltration (H&E staining, Supplementary Fig. 2e, f ), neuronal damage (Nissl staining, Supplementary Fig. 2g, h ), or myelin damage (LFB staining, Supplementary Fig. 2i, j ) in the spinal cords of Ldha icKO mice compared to controls (all P > 0.05). However, immunofluorescence analysis of these naive Ldha icKO mice revealed reduced H4K12la modification in microglia compared to control mice (Supplementary Fig. 2k, l ), confirming that the gene deletion effectively modulated microglial lactate metabolism at baseline.

Following EAE induction, Ldha icKO mice exhibited significantly exacerbated disease severity compared to Ldha flox/flox control EAE mice ( P < 0.05) (Fig. 3c, d ). Importantly, EAE outcomes in tamoxifen-treated Cre-only controls ( Cx3cr1 CreERT2/+ ) were indistinguishable from those of wild-type controls, confirming that the phenotype was specifically attributable to microglial Ldha deletion (Supplementary Fig. 2m, n ). Consistent with this aggravated clinical phenotype, analysis of spinal cords from EAE-induced mice during the acute phase revealed that Ldha icKO mice had a marked increase in neuroinflammatory cell infiltration ( P < 0.05) (Fig. 3e, f ), accompanied by exacerbated neuronal damage ( P < 0.01) (Fig. 3g, h ) and demyelination ( P < 0.05) (Fig. 3i, j ) relative to EAE control mice. Importantly, immunofluorescence analysis of the spinal cord showed that microglia in EAE-induced Ldha icKO mice displayed persistently reduced H4K12la modification ( P < 0.01) (Fig. 3k, l, m ).

Critically, to validate the microglia-specificity of our conditional knockout model at the time of EAE induction, we assessed whether peripheral myeloid cells retained the Ldha deletion phenotype after the tamoxifen washout period. We isolated peripheral CD11b+ myeloid cells from the spleens of Cx3cr1 -CreERT2; Ldha flox/flox mice using magnetic bead sorting at 2 weeks post-tamoxifen injection, the exact time point used for EAE induction in our study. Intracellular lactate production in these splenic CD11b+ cells was comparable between tamoxifen-treated knockout mice and controls (Supplementary Fig. 3a ), confirming that the peripheral myeloid cell pool had been sufficiently renewed from bone marrow precursors and no longer exhibited the Ldha -deficient metabolic phenotype at this time point. These data functionally validate that the phenotypic effects observed in our EAE model are attributable to microglia-specific Ldha loss rather than peripheral myeloid cell involvement.

We hypothesized that preserving the lactate-driven H4K12la modification could be protective. Recent studies have identified histone deacetylase 1 (HDAC1) as a primary enzymatic “eraser” responsible for removing histone lactylation marks 22 , 23 . Accordingly, we first confirmed that microglial HDAC1 protein levels were significantly upregulated during the course of EAE in the absence of treatment (Supplementary Fig. 4a, b ). To test this hypothesis, we treated EAE mice with MS-275, a class I HDAC inhibitor, to prevent the removal of H4K12la by HDAC1. Strikingly, MS-275 treatment significantly ameliorated clinical symptoms and reduced disease burden compared to vehicle-treated controls (Fig. 3n, o ). Immunofluorescence analysis of spinal cord lesions confirmed that MS-275 treatment effectively restored H4K12la levels in microglia (Fig. 3p, q ). Collectively, these loss- and gain-of-function data demonstrate that the intracellular lactate–H4K12la axis is a critical determinant of disease outcome in EAE, with impaired microglial lactylation fueling neuroinflammation and exacerbating disease progression. Furthermore, restoring this metabolic–epigenetic axis through pharmacological intervention provides a potential therapeutic strategy to mitigate EAE severity by reprogramming microglial function towards a more neuroprotective state.

Exogenous lactate ameliorates neuroinflammation via microglial reprogramming, independent of restricting infiltration

Building on LDHA’s central role in endogenous lactate generation, we propose that microglial Ldha deletion, which not only impairs lactate production but also worsens neuroinflammatory injury in EAE, can be counteracted by exogenous lactate. Specifically, we hypothesize that supplemental lactate will boost microglial H4K12la, reshape the immune microenvironment toward a less inflammatory state, and thereby attenuate disease severity. To test this, wild-type mice were subjected to EAE induction, and daily intraperitoneal administration of lactate (500 mg kg⁻¹) or PBS as a control was initiated at day 10 post-induction (preclinical phase) and continued throughout the experiment (Fig. 4a ). Importantly, to confirm that systemically administered lactate effectively reaches the CNS, we quantified lactate levels in the spinal cord tissues and isolated microglial cells. As expected, i.p. administration of lactate significantly elevated the overall lactate concentrations in the inflamed spinal cords and the intracellular pool of CD11b⁺ cells of EAE mice compared to vehicle-treated controls (Supplementary Fig. 3b, c ). Fig. 4 Exogenous lactate supplementation ameliorates EAE severity and enhances CNS myeloid cell H4K12 lactylation. a Experimental design for EAE induction and daily intraperitoneal (i.p.) lactate administration (500 mg kg⁻¹/day) or PBS initiated at day 10 post-immunization (dpi). Clinical EAE scores of mice treated with PBS, lactate, or naive controls in the acute phase ( b , n = 12 mice/group) and extending to the chronic phase ( c , n = 6 mice/group). AUC quantification for EAE scores during the acute ( d ) and chronic ( e ) phases. Representative LFB ( f ) and H&E ( g ) staining in spinal cord sections at acute and chronic stages. Scale bars, 50 µm (applies to all image panels). Quantification of LFB ( h ) and H&E ( i ) scores ( n = 6 mice/group). Representative immunofluorescence of H4K12la (red) and Iba1 (green) in spinal cords from acute ( j ) and chronic ( k ) EAE mice. DAPI (blue) stains nuclei. Scale bars, 50 µm (applies to all image panels). l Quantification of H4K12la⁺ Iba1⁺ cell density from j and k ( n = 5 mice/group). Data are presented as mean ± SEM. Statistical significance was determined by an unpaired, two-tailed Student’s t test for two-group comparisons ( d , e ) and a one-way ANOVA followed by Šídák’s multiple comparisons test for multiple groups ( h , i , l ). Exact P values: d P < 0.0001; e P = 0.0021; h Acute P < 0.0001, Chronic P < 0.0001; i acute P = 0.0008, Chronic P = 0.0131; l acute P < 0.0001, chronic P = 0.0028. Source data are provided as a Source Data file.

Exogenous lactate treatment significantly alleviated EAE disease severity, as evidenced by reduced clinical scores during both the acute (Fig. 4b ) and chronic phases (Fig. 4c ) of the disease. Correspondingly, the area under the curve (AUC) for EAE scores was significantly lower ( P < 0.01) in lactate-treated mice during both the acute (Fig. 4d ) and chronic phases (Fig. 4e ) compared to PBS-treated EAE controls. Histopathological examination of spinal cord tissues further supported these clinical findings. Specifically, LFB staining demonstrated a notable amelioration of myelin damage (Fig. 4f, h ; P < 0.05), and H&E staining revealed a significant reduction in neuroinflammatory cell infiltration during both the acute and chronic stages (Fig. 4g, i ; P < 0.05). Crucially, immunofluorescence analysis demonstrated that while EAE itself was associated with an increase in H4K12la-positive Iba1⁺ microglia, exogenous lactate administration led to a further significant increase in the density of these cells in the spinal cords during both acute and chronic phases ( P < 0.01) (Fig. 4j–l ). We used flow cytometry to characterize the immune cell landscape across the disease course. Early lactate treatment significantly reduced the percentage of infiltrating monocytes (Ly6C⁺CD45hi) in the spinal cord during both acute and chronic EAE (Supplementary Fig. 5a, b ). This modulation was localized to the CNS. Analysis of peripheral immune organs showed no significant differences in CD8⁺ T cell populations in the spleen (Supplementary Fig. 5c, d ) or draining lymph nodes (Supplementary Fig. 5e, f ). These findings suggest that early lactate intervention suppresses local myeloid accumulation without inducing systemic immune suppression. However, as early treatment reduced total cell infiltration, it remains to be determined whether lactate acts by preventing recruitment or by functionally reprogramming cells within the established inflammatory environment.

To rigorously distinguish whether the therapeutic benefit of lactate relies on blocking peripheral immune cell recruitment or modulating local cellular function, we initiated lactate treatment at the peak of disease (day 14 post-immunization), a stage when immune infiltration into the CNS is already established (Fig. 5a ). To ensure rigorous experimental design, animals were ranked by clinical score at 14 dpi and allocated to treatment versus vehicle groups using a stratified randomization approach, ensuring comparable baseline disease severity between groups at the time of treatment initiation (Fig. 5b ). Strikingly, even with this delayed intervention, lactate treatment significantly attenuated disease progression. As shown in the clinical trajectories, while the disease course of the PBS and 14 dpi lactate-treated groups perfectly overlapped prior to treatment, they distinctly diverged shortly after lactate administration (Fig. 5b ). Evaluation of clinical scores at specific critical time points confirmed that late-stage lactate intervention progressively and significantly reduced disease severity compared to the PBS-treated vehicle group (Fig. 5c ; P < 0.01 at 17, 20, and 21 dpi). Histological analysis confirmed that late-stage lactate treatment effectively preserved myelin integrity (Fig. 5d, e ; LFB score, P < 0.01). Unlike the preclinical treatment, lactate administration starting at the peak did not significantly alter the density of inflammatory infiltrates (H&E staining, Fig. 5f, g ) or the proportions of Ly6C⁺ infiltrating myeloid cells (Fig. 5h, i ), CD4⁺ T cells, and CD8⁺ T cells (Fig. 5j, k ) in the spinal cord. These data provide compelling evidence that the therapeutic efficacy of lactate at this stage is not driven by restricting leukocyte entry into the CNS. Instead, immunofluorescence analysis revealed a significant upregulation of H4K12la specifically within Iba1⁺ microglia/macrophages in the lesions of lactate-treated mice (Fig. 5l, m ; P < 0.01). Collectively, these findings demonstrate that lactate can ameliorate ongoing neuroinflammation by functionally reprogramming CNS myeloid cells via the H4K12la epigenetic axis, independent of blocking peripheral immune infiltration. Fig. 5 Lactate treatment initiated at peak disease ameliorates EAE severity via CNS myeloid cell reprogramming without altering immune infiltration. a Schematic of the therapeutic regimen: lactate (500 mg kg⁻¹/day, i.p.) or PBS was administered starting from the peak of disease (day 14 post-immunization). Clinical EAE scores ( b ) and statistical comparisons at specific critical time points (17, 20, and 21 dpi) ( c ) for mice treated with PBS, lactate starting at day 10, or lactate starting at day 14 ( n = 8 mice/group). For the box plot in ( c ), the center line represents the median, the bounds of the box represent the 25th and 75th percentiles, and the whiskers extend to the minima and maxima. Representative LFB staining ( d ) and quantification of LFB scores ( e ) in spinal cord sections from PBS- and lactate-treated (day 14) mice ( n = 6 mice/group). Representative H&E staining ( f ) and quantification of inflammatory corpuscle density ( g ) ( n = 6 mice/group). Representative flow cytometry plots ( h ) and quantification ( i ) of infiltrating myeloid cells (Ly6C⁺CD45hi) in the spinal cord ( n = 5 mice/group). Representative flow cytometry plots ( j ) and quantification ( k ) of CD4⁺ and CD8⁺ T cells in the spinal cord ( n = 5 mice/group). Representative immunofluorescence images ( l ) and quantification ( m ) of H4K12la (red) in Iba1⁺ cells (green) ( n = 5 mice/group). Scale bars, 50 μm (applies to all image panels). Data are presented as mean ± SEM. Statistical significance was determined by two-way repeated measures ANOVA ( b ), one-way ANOVA followed by Šídák’s multiple comparisons test ( c ), and an unpaired, two-tailed Student’s t test for two-group comparisons ( e , g , i , k , m ). Exact P values are indicated in the figure. Source data are provided as a Source Data file.

scRNA-seq reveals lactate-driven suppression of inflammatory microglial states

Having established that exogenous lactate exerts robust neuroprotection, we utilized the standard preclinical treatment regimen (initiation at dpi 10) as a foundation for all subsequent mechanistic investigations. To rigorously dissect the cellular mechanisms underlying lactate-mediated neuroprotection, we performed single-cell RNA sequencing (scRNA-seq) on spinal cord tissues from EAE mice treated with PBS (Control) or lactate (from dpi 10) during the acute phase (Day 21). To ensure robust cell-type resolution and minimize technical artifacts, we implemented a stringent quality-control pipeline prior to downstream analyses. Specifically, we utilized SoupX to estimate and subtract ambient RNA contamination—such as hemoglobin and myelin transcripts—that inevitably arises from tissue dissociation (Supplementary Fig. 6a–b ). Furthermore, rigorous doublet removal was performed using DoubletFinder (Supplementary Fig. 6c–e ). This approach yielded 51,902 high-quality cells, allowing us to unambiguously distinguish resident microglia ( Tmem119 + , P2ry12 + , Cx3cr1 + ) from infiltrating macrophages ( Ccr2 + , Ly6c2 + , CD45-high ) and other CNS lineages (Fig. 6a ; Supplementary Fig. 7a–b ), thereby definitively confirming that the lactate-driven transcriptomic and metabolic reprogramming occurs intrinsically within resident microglia, rather than merely reflecting a shift in myeloid cell composition. Fig. 6 Lactate intervention reprograms microglial state and metabolism, modulating neuroinflammation in EAE progression. a Schematic illustration of the experimental design for single-cell RNA sequencing (scRNA-seq). EAE mice were treated with PBS or lactate (500 mg kg⁻¹, i.p.) daily from day 10 to day 21 (acute phase). Spinal cord tissues were collected at peak disease (Day 21) for single-cell dissociation and sequencing. The dataset comprises two independent biological replicates per condition, each consisting of a pooled sample from two individual mice. b UMAP visualization of microglial subclustering. The analysis identifies eight transcriptional states including Homeostatic, IFN-responsive (1 and 2), Inflammatory, DAM, DAM-like, Lipid-Regulating, and Cycling microglia. c Relative proportions of microglial subpopulations in Control and Lactate groups. Lactate treatment reduces the fraction of Inflammatory and DAM clusters. d Violin plots displaying the expression levels of representative genes related to inflammation ( Ccl5, Tlr2, C3, Nfkbia ), homeostasis ( Tgfbr1, Il10ra ), and metabolic pathways ( Pgk1, Ldha, Pkm, Atp5md, Cox5a, Ndufa1 ) across the microglial subpopulations. ( n = 10,749 cells for Control; n = 14,769 cells for Lactate, with each replicate pooled from 2 mice, totaling 4 mice per group). For the internal box plots, the center line represents the median, the bounds of the box represent the 25th and 75th percentiles, and the whiskers extend to 1.5 times the interquartile range. e Metabolic pathway activity in microglia. High-flux pathways including Glycolysis, Oxidative phosphorylation, and the TCA cycle are downregulated under lactate treatment. f Intercellular communication networks determined by CellChat. Dot plots show ligand–receptor pairs with increased signaling (left) and decreased signaling (right) in the Lactate group. Communication probabilities were inferred based on ligand and receptor expression in sender and receiver cell populations. Statistical significance was evaluated using the one-sided permutation test implemented in CellChat. Dot size indicates nominal P value, and color represents the inferred communication probability. No additional multiple-testing correction was applied for this visualization.

Given the central role of microglia in our hypothesis, we performed focused subclustering of the microglial compartment. This analysis resolved microglia into eight transcriptionally distinct states, defined based on differential gene expression patterns: Homeostatic, IFN-responsive (1 and 2), Inflammatory, Disease-associated Microglia (DAM), DAM-like, Lipid-regulating, and Cycling microglia (Fig. 6b ; Supplementary Fig. 7c ). Notably, DAM-associated and DAM-like states were operationally defined based on the enrichment of previously reported DAM signature genes, without implying full equivalence to the canonical DAM programs originally described in neurodegenerative disease models. Lactate treatment was associated with a marked shift in the microglial landscape. While the PBS-treated EAE group was dominated by Inflammatory, IFN-responsive, and DAM-associated transcriptional states, lactate-treated mice exhibited a relative enrichment of Homeostatic microglia, accompanied by a significant reduction in the representation of Inflammatory, IFN-responsive, and DAM-associated states These changes suggest that lactate supplementation is linked to an attenuation of inflammatory microglial programs rather than a global suppression of microglial abundance (Fig. 6c ).

We next examined the transcriptional programs underlying these state-level changes. Differential expression analysis revealed that lactate treatment was associated with the downregulation of genes involved in pro-inflammatory signaling and innate immune activation, including Ccl5 , Tlr2 , C3 , and Nfkbia (Fig. 6d ). Conversely, genes critical for microglial homeostasis and anti-inflammatory regulation, including Tgfbr1 and Il10ra , were maintained or upregulated (Fig. 6d ). To independently validate these transcriptomic shifts with adequate biological replication, we isolated pure microglial populations by FACS sorting (gated as CD45int CD11b + Ly6C-) from new, independent biological replicates ( n = 5 per group). RT-qPCR analysis confirmed that lactate treatment significantly downregulated the expression of the pro-inflammatory/DAM marker C3 and upregulated the reparative marker Arg1 in purified microglia compared to vehicle-treated controls (Supplementary Fig. 8 ; P < 0.05, unpaired two-tailed Student’s t-test). Furthermore, the expression of the inflammatory receptor Tlr2 and the neurotrophic factor Gdnf exhibited consistent trends of downregulation and upregulation, respectively, perfectly mirroring the scRNA-seq profiles (Supplementary Fig. 8 ). These transcriptional changes are consistent with a shift toward a tissue-supportive and immunoregulatory microglial state, rather than a simple binary transition between activated and resting phenotypes. To investigate how exogenous lactate influences intracellular metabolic programs, we mapped metabolic gene expression across the single-cell landscape. Microglia from lactate-treated mice displayed a coordinated downregulation of genes involved in glycolysis ( Pgk1, Pkm, Ldha ) and mitochondrial respiration ( Cox5a, Ndufa1, Atp5md ) compared with PBS-treated controls (Fig. 6d ; Supplementary Fig. 9a ). Pathway enrichment analysis further confirmed a global reduction in high-flux metabolic pathways, including glycolysis, pyruvate entry into the TCA cycle, and oxidative phosphorylation (Fig. 6e ; Supplementary Fig. 9a–c ). This transcriptional profile suggests a relative dampening of energy-intensive metabolic programs, consistent with reduced metabolic strain in inflammatory microglia under lactate supplementation. Box plots show median (center line), interquartile range (IQR, box bounds), and whiskers extending to 1.5× IQR.

Finally, to determine whether these intracellular changes were accompanied by altered intercellular communication within the CNS microenvironment, we utilized CellChat analysis. In the PBS-treated EAE group, microglia exhibited strong outgoing signaling through pro-inflammatory ligand-receptor pathways, including the Spp1-Cd44 axis and Ccl chemokine signaling (Fig. 6f , decreased signaling). In contrast, lactate-treated mice showed a pronounced reduction in the strength and number of these inflammatory interactions, alongside a relative enhancement of immunoregulatory signaling pathways (Fig. 6f , increased signaling; Supplementary Fig. 9d–f ). Collectively, these data demonstrate that lactate supplementation is associated with coordinated transcriptional, metabolic, and communicative remodeling of microglia, favoring a CNS environment permissive to neuroprotection during EAE.

Lactate drives a stage-dependent transition toward a neurotrophic microglial phenotype

To delineate the stage-dependent molecular impact of lactate on microglial states during neuroinflammation, we performed bulk RNA sequencing on isolated CD11b⁺ cells from EAE mice at defined acute and chronic disease stages (as determined by clinical course and post-immunization time; see “Methods”), following the experimental timeline illustrated in Fig. 7a . Differential expression analysis revealed a substantial lactate-associated transcriptomic remodeling in microglia during the acute phase (Acute EAE+ La vs. Acute EAE). Using a stringent threshold (Q < 0.05 and |log2FC | > 1), we identified 33 significantly differentially expressed genes (DEGs), all of which were downregulated, reflecting a precise inhibition of pro-inflammatory programs (Supplementary Fig. 10a ). KEGG pathway analysis highlighted significant modulation of key immune and metabolic programs, including the “PPAR signaling pathway and C-type lectin receptor signaling pathway” (Fig. 7b, c ). Gene Set Enrichment Analysis (GSEA) further clarified that the “C-type lectin receptor signaling pathway” was significantly downregulated in the lactate-treated group (NES = −1.91, P < 0.001; Fig. 7d ), indicating an attenuation of innate immune sensing programs typically associated with acute inflammatory activation. Consistently, heatmap analysis revealed downregulation of multiple lectin receptors and associated signaling molecules (Supplementary Fig. 11a ). To validate these functional shifts at the protein level, we performed immunofluorescence staining. Consistent with a shift away from a pro-inflammatory state, the density of Iba1+ microglia expressing the inflammatory marker NOS2 was markedly reduced following lactate treatment ( P < 0.05) (Fig. 7l, p ), whereas the density of Arginase-1 (Arg1)-expressing Iba1+ microglia was substantially increased in lactate-treated mice ( P < 0.05) (Fig. 7k, o ). These findings suggest that lactate biases microglia away from acute inflammatory activation toward a more reparative functional state during early EAE. Fig. 7 Lactate reprograms CNS myeloid cell transcriptional and functional landscapes from acute to chronic EAE. a Schematic of the bulk RNA-seq and validation workflow in isolated CD11b⁺ cells. Acute phase comparison between Acute EAE+La and Acute EAE groups showing a heatmap of DEGs ( b ), KEGG enrichment ( c ), and GSEA plot for the C-type lectin receptor signaling pathway ( d ). Chronic phase comparison between Chronic EAE+La and Chronic EAE groups showing a heatmap of DEGs ( e ), Reactome enrichment ( f ), and GSEA plot for chromatin modifying enzymes ( g ). Temporal comparison between Acute EAE+La and Chronic EAE+La groups showing a heatmap of DEGs ( h ), WikiPathways enrichment ( i ), and GSEA plot for burn wound healing ( j ). Representative immunofluorescence of ARG1 ( k ), NOS2 ( l ), HDAC1 ( m ), and GDNF ( n ) in spinal cord lesions. Scale bars, 50 μm (applies to all image panels). Quantification of ARG1⁺ Iba1⁺ cell density ( o ), NOS2⁺ Iba1⁺ cell density ( p ), relative HDAC1 intensity ( q ), and relative GDNF intensity ( r ) in microglia ( n = 6 mice/group). Data are presented as mean ± SEM. Statistical significance was determined by an unpaired, two-tailed Student’s t est. Exact P values: o P = 0.0087; p P = 0.0140; q P = 0.0004; r P = 0.0046. Source data are provided as a Source Data file.

In the chronic phase (Chronic EAE+ La vs. Chronic EAE), transcriptomic differences persisted, identifying 355 DEGs (Supplementary Fig. 10b ) and suggesting a long-term reprogramming (Fig. 7e ). Pathway analysis revealed significant modulation of pathways related to epigenetic regulation, including “HATs acetylate histones” and “Epigenetic regulation of gene expression” (Fig. 7f ). GSEA revealed a significant positive enrichment of the “Chromatin modifying enzymes” gene set (NES = 1.35, P = 0.032; Fig. 7g ), with a heatmap showing dynamic changes in multiple histone deacetylases and acetyltransferases (Supplementary Fig. 11b ).

Finally, direct comparison of microglia from lactate-treated mice across disease phases (Chronic EAE+ La vs. Acute EAE+ La) revealed broader temporal evolution with 5043 DEGs (Supplementary Fig. 10c ; Fig. 7h ). Enriched pathways included immunomodulatory and tissue-reparative signaling cascades such as “IL-6 signaling” and “TGF-β signaling” (Fig. 7i ). GSEA confirmed a robust positive enrichment of a gene ontology module annotated as “Burn wound healing” (NES = 1.53, P = 0.039; Fig. 7j ), reflecting the activation of conserved reparative transcriptional programs rather than literal wound repair. This enrichment was driven by increased expression of reparative and neurotrophic genes, including Tgfb1 and Vegfa, as shown in the gene expression heatmap (Supplementary Fig. 11c ). Consistent with these transcriptomic findings, immunofluorescence analysis showed a marked increase in the protein expression of the neurotrophic factor glial cell line-derived neurotrophic factor (GDNF) within Iba1+ microglia in the chronic lactate-treated group relative to the acute lactate-treated group ( P < 0.01) (Fig. 7n, r ), supporting a sustained neuroprotective microglial phenotype.

Based on prior evidence implicating HDAC1 as a principal eraser of histone lactylation marks in inflammatory contexts, we next focused on HDAC1 dynamics during EAE progression. Immunofluorescence analysis revealed that HDAC1 protein levels in spinal cord microglia were significantly upregulated during the acute phase of EAE and remained persistently elevated into the chronic phase compared to naive controls (Supplementary Fig. 4a, b ), identifying sustained HDAC1 induction as a pathological feature of chronic neuroinflammation. Notably, lactate treatment significantly reduced HDAC1 expression in microglia during the chronic phase ( P < 0.01) (Fig. 7 , m and q), providing a mechanistic basis for the long-term maintenance of the beneficial H4K12la epigenetic signature under prolonged inflammatory conditions.

H4K12la directly regulates neurotrophic gene expression via an HDAC1-sensitive mechanism

To elucidate the molecular mechanism linking metabolic flux to microglial neuroprotective programs, we employed the human microglial cell line HMC3 as a reductionist system to dissect epigenetic causality under controlled inflammatory conditions. To map the epigenetic landscape modulated by inflammation, we performed H4K12la CUT&Tag analysis following stimulation with LPS and IFN-γ. Genome-wide analysis revealed that inflammatory stimulation induced a global increase in H4K12la peaks, which were predominantly localized to promoter regions and 5’ untranslated regions (UTRs), consistent with a transcriptional permissive chromatin state (Fig. 8a-e ). Notably, H4K12la was preferentially enriched at the promoter regions of neurotrophic factors Gdnf and Fgf1 (Fig. 8f, g ), implicating lactylation in the regulation of microglial reparative programs. Fig. 8 Lactate-induced H4K12la regulates neurotrophic factor expression via an HDAC1-sensitive mechanism. H4K12la CUT&Tag profiling in HMC3 cells, including signal density at transcription start sites ( a ), genomic distribution ( b ), peak overlap ( c ), and average profile across gene bodies ( d ). e Volcano plot displaying differentially expressed genes in LPS + IFN-γ-stimulated microglia. Genome browser tracks ( f ) and qPCR validation ( g ) of H4K12la enrichment at Gdnf and Fgf1 promoters ( n = 3 independent cell cultures). KEGG enrichment of H4K12la-associated genes ( h ) and schematic illustration of the lactate-driven mechanism ( i ). Western blot ( j ) and immunofluorescence ( k ) showing lactate-mediated rescue of GDNF and H4K12la levels following stimulation. Western blot analysis of target proteins under HDAC1-GFP overexpression or MS-275 treatment ( l ), and 2-DG-mediated inhibition with subsequent lactate rescue ( m ). Representative immunofluorescence images ( n ) and quantification ( o ) of H4K12la and GDNF in cells overexpressing HDAC1-GFP ( n = 4 independent cell cultures). White boxes indicate co-localization; white arrowheads mark non-transfected cells. Scale bars, 20 μm (applies to all image panels). For representative Western blots and micrographs ( j – m ), experiments were independently repeated three times with similar results. Data are presented as mean ± SEM. Statistical significance was determined by an unpaired, two-tailed Student’s t test. Exact P values: g Gdnf P = 0.5331, Fgf1 P = 0.2814; o H4K12la P = 0.0001, GDNF P = 0.0014. Source data are provided as a Source Data file.

To determine whether lactate availability functionally constrains this epigenetic–transcriptional axis beyond the baseline inflammatory response, we supplemented LPS + IFN-γ-stimulated HMC3 cells with exogenous lactate (20 mM). Western blot and immunofluorescence analyses demonstrated that lactate supplementation significantly potentiated the inflammation-induced accumulation of H4K12la and further augmented the protein expression of GDNF and FGF1 compared to the vehicle-treated inflammatory group (Fig. 8j, k ). This confirms that expanding the intracellular lactate pool beyond the endogenous level generated by inflammation is sufficient to reinforce the reparative phenotype.

To definitively link glycolytic flux to this epigenetic modification, we utilized 2-Deoxy-D-glucose (2-DG) to inhibit glycolysis. 2-DG treatment significantly abrogated the inflammation-induced upregulation of H4K12la and GDNF. Importantly, this suppression was effectively rescued by exogenous lactate supplementation (Fig. 8m ), proving that lactate is the essential substrate coupling glycolytic activity to epigenetic regulation.

Given that HDAC1 exhibits deacetylase and delactylase activities, we sought to decouple the contributions of histone acetylation and lactylation. To selectively disrupt acetyl-CoA–dependent acetylation, we treated cells with the ACLY inhibitor BMS-303141. Quantitative immunofluorescence analysis confirmed that while BMS treatment significantly reduced global H4K12ac levels (Supplementary Fig. 12b, e ), it preserved H4K12la and downstream FGF1 expression (Supplementary Fig. 12a, c, d ). Strikingly, when HDAC1-GFP was overexpressed in these BMS-treated cells, we observed a spatially restricted depletion of H4K12la specifically within HDAC1-positive nuclei, accompanied by a concomitant loss of FGF1 expression (Supplementary Fig. 12a, c, f, g ), establishing HDAC1 as a potent enzymatic “eraser” and negative regulator of lactylation-dependent transcription. Consistently, pharmacological inhibition of HDAC1 using MS-275 effectively prevented H4K12la erasure and significantly enhanced neurotrophic factor expression, as demonstrated by Western blot analysis (Fig. 8l ). Furthermore, immunofluorescence analysis confirmed that genetic overexpression of HDAC1 specifically erased H4K12la marks in situ and suppressed GDNF expression (Fig. 8n-o ). Collectively, these findings establish an HDAC1-sensitive epigenetic mechanism in which lactate-dependent H4K12la is functionally coupled to neurotrophic gene expression in microglia under inflammatory conditions, indicating a direct and cell-autonomous role for HDAC1 in lactylation erasure.

Discussion

Microglial metabolic states shape CNS autoimmune inflammation in MS 24 . We identify a lactate-driven metabolic–epigenetic axis that reprograms microglia through H4K12la, linking elevated glycolysis to the expression of neuroprotective gene 10 , 25 . H4K12la is preferentially enriched in microglia, promoting a sustained, pro-repair phenotype conducive to CNS recovery in MS. Central to our findings is a lactate–H4K12la–HDAC1 regulatory axis that orchestrates this microglial reparative program. Accordingly, microglia-specific deletion of Ldha markedly worsens EAE, highlighting the protective function of this pathway. The axis also exhibits distinct spatial and temporal features in which lactate functions as a spatiotemporal cue that preconditions microglia during the preclinical phase of EAE, promoting an anti-inflammatory and reparative state, while H4K12la becomes preferentially enriched in spinal cord microglia. These findings establish microglia as context-sensitive “lactate readers” and highlight the lactate–H4K12la axis as a metabolically tunable target for MS therapy.

Our study revisits the function of lactate in the context of neuroinflammation, proposing that lactate, acting not merely as a metabolic byproduct of the Warburg effect, but as an intrinsic “brake” on the inflammatory response 26 . In the inflamed CNS, the metabolic shift toward aerobic glycolysis naturally leads to lactate accumulation, which has traditionally been viewed as a hallmark of metabolic stress. Our data suggest that microglia may utilize elevated intracellular lactate as a substrate for histone lactylation (H4K12la), leading to a transcriptional shift associated with a negative feedback loop involving genes such as Gdnf and Fgf1. This mechanism suggests that lactate may function as a key “immunometabolic switch” poised to initiate tissue repair. This pattern is recapitulated in EAE models, where elevated spinal cord lactate is accompanied by increased microglial lactylation. Together, these findings support the translational relevance of the lactate–lactylation axis in MS pathophysiology. The elevated lactate in patient CSF reflects heightened glycolytic activity in immune cells, acting not only as a dynamic biomarker but also as a mediator that drives the observed epigenetic modifications 17 , 27 , 28 . Our findings expand on previous observations by demonstrating that early lactate exposure not only delays disease onset10, but actively reprograms microglia through H4K12la, inducing a persistent anti-inflammatory phenotype. This preemptive effect likely arises from early epigenetic priming of microglia, where lactate-induced elevation of H4K12la levels prior to immune challenge induces a more anti-inflammatory chromatin configuration at key cytokine and chemokine loci 20 . As a result, when autoreactive T cells infiltrate the CNS, microglia are predisposed to a regulatory phenotype, secreting higher levels of IL-10 and TGF-β and exhibiting reduced MHC-II expression. Early lactate exposure may also condition myeloid precursors in the bone marrow to adopt a less inflammatory phenotype upon CNS entry 13 , 29 . Lactate therefore plays a critical pre-conditioning role reprograming both CNS and peripheral innate immune compartments, offering a window of metabolic intervention to prevent or mitigate neuroinflammation. Spatially, our immunofluorescence data demonstrate that protein lactylation in EAE is most pronounced within spinal cord microglia, with only modest changes observed in neurons and astrocytes. This cell-type specificity reflects microglial distinct capacities for glycolytic flux, lactate export, and histone-modifying enzyme expression30. Essentially, microglia serve as privileged “lactate sensors” and “lactate readers,” integrating metabolic signals into tailored epigenetic outputs 5 , 30 , 31 .

Our findings on lactate-driven epigenetic reprogramming suggest a potentially protective role for this pathway in neuroinflammation and contribute to a growing body of evidence highlighting the context-dependent function of lactate. In certain pathological settings, such as acute CNS infections, elevated lactate in the CSF is a recognized biomarker of poor prognosis 32 , often reflecting severe tissue hypoxia and metabolic crisis, rather than an adaptive signaling response. However, this contrasts with its role in chronic autoimmune settings. A recent review has underscored that lactate is an emerging immunoregulatory molecule whose effects are contingent on its concentration, duration of exposure, and the specific pathological environment. In the context of CNS autoimmunity, it has been demonstrated that lactate can limit disease progression in EAE models by stabilizing HIF-1α in dendritic cells, thereby revealing a protective immunometabolic mechanism distinct from, but conceptually aligned with, the pathway described in the present study 11 . Building upon these observations, our study further refines this framework by demonstrating that lactate, acting as a signaling molecule, induces a sustained epigenetic reprogramming state with features reminiscent of long-term functional reprogramming in microglia through histone lactylation. Therefore, these findings underscore the functional plasticity of lactate in the CNS, while it may serve as a marker of metabolic distress in acute infections, it can also act as a context-dependent immunoregulatory signal with potential therapeutic implications in chronic autoimmune diseases such as MS.

In recent years, the concept of long-term functional reprogramming has redefined innate immune function, demonstrating that monocytes, macrophages, and NK cells can develop memory-like traits through metabolic rewiring and histone modifications following primary stimulation 33 – 36 . Emerging evidence indicates that microglia, the CNS-resident innate immune cells, also exhibit such durable reprogramming 37 . Our data demonstrate that lactate, acting as a signaling molecule, induces transcriptional activation during acute inflammation via histone H4K12la and sustains this chromatin state into the chronic phase, representing a form of long-term functional reprogramming in the CNS. Mechanistically, H4K12la locks microglia into a repair-favoring transcriptional state, extending prior findings in macrophages where lactate-driven histone lactylation activates pro-resolving genes and promotes a remyelination-permissive environment 13 , 38 – 40 . It is now appreciated that the simple M1/M2 classification is an oversimplification of the complex spectrum of microglial states 38 .

Microglial lactate metabolism, specifically via LDHA, mediates epigenetic reprogramming critical for resolving neuroinflammation 41 . Microglial-specific Ldha is indispensable for generating endogenous lactate that fuels H4K12la and imprints a reparative immune phenotype during EAE. Deletion of Ldha disrupts this metabolic–epigenetic axis, abolishes H4K12la, locks microglia into a pro-inflammatory transcriptional program, aggravating disease pathology. Restoration of lactate levels through exogenous supplementation reinstates H4K12la, reactivates anti-inflammatory and pro-regenerative gene expression, and ameliorates neuroinflammatory damage. This mechanism closely resembles long-term functional reprogramming, glycolysis-derived lactate promotes histone lactylation-dependent memory in models of Alzheimer’s disease, Parkinson’s disease, and spinal cord injury 18 , 20 , 21 , 30 , 42 . Inhibition of glycolysis, in contrast, erases this immune memory. Together, these findings delineate a microglia-specific lactate–H4K12la–HDAC1 axis that is indispensable for constraining EAE progression in models of autoimmune demyelination. Metabolic reprogramming of microglia toward lactate-supported chromatin states offers a therapeutic entry point for shifting the inflammatory–repair balance in favor of tissue regeneration 38 , 43 . Importantly, our findings necessitate a clear distinction between lactate’s traditional role as an energy substrate and its emerging function as an epigenetic signaling molecule. While the utilization of lactate as a metabolic fuel could partially explain the observed shifts in energy metabolism, our Ldha icKO model demonstrates that metabolic compensation alone is insufficient to recapitulate the neuroprotective phenotype. The exacerbation of EAE in Ldha -deficient mice, despite the presumed availability of alternative energy substrates, strongly indicates that lactate-driven epigenetic signaling via H4K12la is indispensable for the protective response. Thus, while metabolic remodeling and epigenetic programming are intricately linked, they represent mechanistically distinct layers of regulation, with H4K12la serving as the critical epigenetic switch governing microglial functional reprogramming.

The dynamic regulation of H4K12 lactylation involves active enzymatic deacetylation, primarily mediated by HDAC1 22 , 23 . Overexpression of HDAC1 eliminates H4K12la marks and concomitantly suppresses the expression of Gdnf and Fgf1 , two key neurotrophic genes. Co-localization assays reveal targeted recruitment of HDAC1–GFP to H4K12la-enriched chromatin, where it erases lactylation accumulated under pro-inflammatory conditions. By antagonizing lactate-mediated epigenetic activation, HDAC1 dampens the neurotrophic gene program in microglia. In contrast, either pharmacological inhibition of HDAC1 or intracellular lactate elevation stabilizes H4K12la at regulatory loci, reinforcing a pro-restorative, tissue-restorative transcriptional landscape. These findings are consistent with emerging evidence that Class I HDACs act as histone delactylases, positioning H4K12la as a metabolically regulated and druggable epigenetic node in neuroinflammation 22 , 23 . Importantly, our study establishes a well-supported functional framework linking metabolism to epigenetic regulation through complementary loss- and gain-of-function approaches. By genetically ablating the lactate-generating enzyme in microglia, we demonstrated that endogenous lactate production is necessary for maintaining the H4K12la mark and its protective effects. Conversely, inhibiting the “eraser” enzyme HDAC1 preserved H4K12la and enhanced neurotrophic factor expression. Together, this bidirectional modulation supports a model in which the Lactate-H4K12la-HDAC1 axis functions as a tunable metabolic switch, acting largely independently of overt systemic metabolic alterations.

In summary, our study reveals that lactate serves as a metabolite-encoded epigenetic signal that reprograms microglia from a pro-inflammatory to a pro-regenerative state, thereby promoting CNS repair in models of autoimmune demyelination. Central to this mechanism is histone H4K12 lactylation, which imprints a stable transcriptional program that sustains neurotrophic and reparative gene expression. The establishment and maintenance of this chromatin state are intrinsically dependent on glycolytic flux through microglial LDHA, thereby linking inflammatory metabolism to regenerative capacity. Targeting the lactate–H4K12la axis offers a mechanism-based strategy to activate endogenous repair programs in the inflamed CNS, thereby complementing and potentially surpassing traditional immunosuppressive therapies. Realizing this therapeutic potential will require temporal precision, dose optimization, and cell-specific delivery. Nonetheless, leveraging a simple metabolite to reprogram the innate immune landscape of the brain establishes a conceptual and translational framework for metabolic immunotherapy in MS (Fig. 9 ). Fig. 9 Mechanism of lactate-mediated microglial repair via H4K12la and HDAC1. a Clinical and experimental evidence of sustained lactate elevation in MS patients and EAE models. Experimental interventions including genetic ablation ( Ldha icKO), HDAC1 inhibition (MS275), and lactate supplementation ( b ) validated through multi-dimensional omics and histological analyses ( c ). d Proposed molecular mechanism of microglial reprogramming. In EAE, pathological HDAC1 levels antagonize endogenous H4K12la, suppressing the reparative program. Exogenous lactate treatment or HDAC1 inhibition restores H4K12la levels and stabilizes the expression of neurotrophic factors Gdnf and Fgf1 . This epigenetic axis drives the transition from pro-inflammatory activation to a reparative microglial phenotype, promoting metabolic quiescence and neuroprotection during both acute and chronic phases.

This study has several limitations. This study expands the paradigm of lactylation by revealing a direct transcriptional regulation of neurotrophic pathways in inflammatory microglia via H4K12la 5 , 44 , establishing a modifiable epigenetic mechanism in MS. However, several limitations warrant discussion. First, while our Ldha icKO and HDAC1 inhibition models underscore the functional necessity of the lactate–H4K12la axis in microglial reprogramming, we acknowledge the pleiotropic nature of lactate. Its therapeutic effects on the CNS may not be exclusively mediated by epigenetic mechanisms, but could also involve additional pathways such as pH modulation 45 , 46 , or receptor-dependent signaling. Although our pharmacokinetic data confirm that exogenous lactate effectively reaches microglia, and our Ldha icKO experiments underscore the centrality of the lactate-lactylation pathway, we recognize that other lactate-mediated effects may contribute to the overall phenotype. Regarding the biochemical conversion of lactate into histone modifications, our study builds upon the foundational work by Zhang et al. (2019), which definitively established lactate as the primary carbon source for histone lactylation using mass spectrometry. Accordingly, rather than extending atomic-level tracing, we strategically prioritized defining the functional necessity of this metabolic-epigenetic axis in a disease-relevant context. By selectively manipulating the “writer” machinery (via Ldha icKO to restrict substrate availability) and the “eraser” machinery (via HDAC1 inhibition), we demonstrate that the Lactate-H4K12la axis is functionally required for microglial reprogramming. While lactate may exert additional pleiotropic effects through mechanisms such as GPR81 signaling or extracellular pH modulation, the profound loss of neuroprotection observed in our microglia-specific Ldha knockout mice supports intracellular lactylation as a major mechanistic contributor to the reparative phenotype. Furthermore, as our work primarily focused on microglia, future studies should investigate lactate’s impact on other CNS cell types, including astrocytes and neurons, to determine whether their responses synergize with or counteract microglial epigenetic reprogramming. Second, our primary experimental design relied heavily on prophylactic intervention. Although our supplementary data demonstrate that therapeutic administration during the peak phase of EAE also ameliorates disease severity—suggesting that lactate promotes a pro-repair microglial phenotype even amid ongoing inflammation—the optimal therapeutic window, dosing kinetics, and long-term safety of exogenous lactate administration remain unresolved. Future studies must rigorously define these pharmacological parameters for clinical viability. Finally, while we observed a correlation between lactate levels and neuroinflammation in both patient CSF and mouse CNS tissues, translational relevance is constrained by interspecies differences. The EAE model does not fully capture the chronic, compartmentalized neurodegeneration seen in human MS. Validating lactate-induced immunometabolic remodeling in human MS brain tissues or patient-derived in vitro models is necessary for future clinical translation. Together, these limitations define clear next steps and provide a tractable roadmap for extending the mechanistic and translational potential of this lactate–epigenetic regulatory axis.

Methods

Animal

C57BL/6 J female mice (8 weeks old) were purchased from Beijing Huafukang Bioscience Co., Ltd. Inducible conditional knockout mice with microglia-specific deletion of Ldha ( Ldha flox/flox Cx3cr1 CreERT2/+ , Ldha icKO) were generated by crossing Ldha flox/flox mice with Cx3cr1 CreERT2/+ mice to generate inducible microglial-specific Ldha -deficient mice (Shanghai Model Organisms Center, Inc). Both Ldha flox/flox and Ldha icKO mice were on a C57BL/6 J genetic background. For inducible Ldha ablation in microglia, adult Ldha icKO and control ( Ldha flox/flox ) mice received daily intraperitoneal (i.p.) injections of tamoxifen (20 mg kg⁻¹) for 5 consecutive days, with the final injection 14 days before EAE induction. For exogenous lactate supplementation, wild-type (WT) EAE mice received daily i.p. administration of lactate (500 mg kg⁻¹) or PBS as a control. Treatment was initiated either at day 10 post-EAE induction (early acute phase) or at day 14 post-EAE induction (therapeutic/peak stage). For pharmacological inhibition of HDAC1 in vivo, EAE mice received daily i.p. injections of MS-275 (5 mg kg⁻¹) or vehicle starting from the acute phase (approximately day 10 post-induction). All treatments continued until the termination of the experiment. Mice were housed in a specific-pathogen-free (SPF) facility under a 12-h light/dark cycle at an ambient temperature of 22 ± 2 °C and 50 ± 10% humidity, with free access to standard chow and autoclaved water. Mice were euthanized by deep isoflurane anesthesia followed by cervical dislocation prior to tissue collection. All experimental animal procedures were approved by the institutional animal care and use committee of Xuanwu Hospital (protocol number: XW-20211216-2).

HMC3 cells culture

HMC3 microglial cells (catalog no. CL-0620) were sourced from Procell and grown in MEM (Procell, catalog no. PM150410) with 10% FBS and 1% penicillin–streptomycin. Cells were maintained in an incubator at 37 °C supplemented with 5% CO2. For experimental treatments, HMC3 cells were stimulated with lipopolysaccharide (LPS, 1 μg mL⁻¹) and interferon-gamma (IFN-γ, 100 ng mL⁻¹) to induce an inflammatory state. To pharmacologically modulate the lactate–HDAC1 axis, cells were treated with the ACLY inhibitor BMS-303141 (10 μM), the class I HDAC inhibitor MS-275 (1 μM), or the glycolysis inhibitor 2-deoxy-D-glucose (2-DG, 5 mM) as indicated. For rescue experiments, exogenous L-lactate (20 mM) was supplemented to the culture medium. Inhibitors were typically added 2 h prior to or simultaneously with the inflammatory stimulation.

Human participants and clinical sample collection

This observational study was conducted at Xuanwu Hospital. The study protocol was reviewed and approved by the Ethics Committee of Xuanwu Hospital, Capital Medical University (Approval Number: 临研审[2025]064 号−001), and all research procedures were conducted in strict accordance with the principles of the Declaration of Helsinki. Written informed consent was obtained from all participants prior to their inclusion in the study and cerebrospinal fluid (CSF) sample collection. Inclusion criteria for the MS group were: (1) diagnosis of MS according to the 2017 McDonald criteria; (2) age 18–65 years; (3) ability to provide written informed consent. Inclusion criteria for the control group were: patients undergoing lumbar puncture for non-neurological indications (e.g., orthopedic surgery requiring spinal anesthesia) with normal CSF findings. Exclusion criteria for both groups were: (1) concurrent central nervous system infection or other autoimmune diseases; (2) use of immunosuppressive or immunomodulatory therapy within 3 months prior to CSF collection; (3) pregnancy or lactation. The study cohorts consisted of 20 patients diagnosed with Multiple Sclerosis (MS) and 20 non-neurological control subjects. The MS group included 5 males and 15 females with a mean age of 41.4 ± 10.6 years and a median disease duration of 84 months (range: 0–360 months); 12 patients were in the chronic phase and 8 in the acute phase. The control group included 12 males and 8 females with a mean age of 35.8 ± 12.5 years. Individual-level demographic data are provided in Supplementary Table 1 . CSF samples were collected via standard lumbar puncture during routine clinical diagnostic procedures. Samples were promptly centrifuged, aliquoted, and stored at −80 °C until further analysis of lactate concentrations. In accordance with the SAGER guidelines, we considered sex as a biological variable. While the sex of all participants was recorded and reported, it was not designated as a primary stratification variable in the initial study design due to the observational nature of the cohort and sample size limitations. Furthermore, as a single-center observational study, potential selection bias may exist, and the relatively small sample size may limit the broader generalizability of the findings. However, strict inclusion criteria were applied to minimize confounding factors.

Induction and clinical evaluation of EAE

Experimental autoimmune encephalomyelitis (EAE) was initiated in female C57BL/6 J mice at eight weeks of age. On the initial day (Day 0), a stable water-in-oil emulsion was formulated. This involved vigorously combining equal parts of a Myelin Oligodendrocyte Glycoprotein (MOG) 35-55 peptide solution (2 mg mL⁻¹ in 20 nM glycine-HCl, pH 3.0) with Complete Freund’s Adjuvant (CFA) supplemented with Mycobacterium tuberculosis H37RA at a concentration of 5 mg mL⁻¹. Using a three-way stopcock, the mixture was emulsified on ice for a duration of 30–120 min. Stability was confirmed when a small droplet of the emulsion did not disperse upon contact with a water surface. Every animal was administered a 200 µL total volume of the preparation via two 100 µL subcutaneous injections at dorsal locations, specifically the nape and the base of the tail. Directly after the emulsion administration, an intraperitoneal injection of 200 ng Pertussis Toxin (PTX) in 0.2 mL of phosphate-buffered saline (PBS) was delivered. A subsequent dose of PTX was given via the same route 48 h post-induction (Day 2). The animals were subsequently observed each day to track the emergence and development of EAE clinical indicators. Post-induction, disease severity for each mouse was graded daily according to a scoring system where 0 indicated no observable disease symptoms; 1 signified a limp or partially paralyzed tail; 2 represented hind limb weakness manifesting as an unstable gait or tremors; 2.5 denoted paralysis of a single hind limb; 3 corresponded to total paralysis affecting both hind limbs; 4 was assigned for paralysis of all four limbs (quadriplegia); and 5 indicated a moribund condition or fatality. Peak clinical severity, defining the acute disease stage, typically occurred around Day 17 after immunization. Assessment of the chronic stage, which could include periods of remission and possible recurrence, was conducted at later intervals, usually starting from Day 53.

Tissue collection and processing

Mice were anesthetized by intraperitoneal injection of 1.25% Avertin (2,2,2-tribromoethanol, 250 mg kg⁻¹) and transcardially perfused with ice-cold phosphate-buffered saline to clear blood from the tissues. For histological examination, the brain and spinal cord were meticulously excised. These specimens were subsequently immersed in a 4% paraformaldehyde (PFA) solution for a period of 6–8 h at 4 °C for fixation. Cryoprotection was then achieved through successive submersion in 15% and later 30% sucrose until the tissues were fully saturated and submerged. The specimens were then embedded within Tissue-Tek® O.C.T. compound (Sakura® Finetek Inc., USA), flash-frozen utilizing liquid nitrogen, and subsequently held at −80 °C pending immunofluorescence staining. In preparation for protein extraction and Western blotting, samples from the spinal cord or isolated microglia/monocytes were gathered, immediately flash-frozen with liquid nitrogen, and kept at −80 °C. For procedures involving RNA extraction for quantitative real-time PCR (qPCR) or RNA sequencing (RNA-seq), designated tissue areas were similarly flash-frozen in liquid nitrogen, homogenized in TRIzol Reagent (Invitrogen, USA), and maintained at −80 °C until required for analysis.

Immunofluorescence staining and confocal microscopy

Coronal sections of mouse spinal cords (20 µm thickness) were cut using a cryostat (CM3050S, Leica), immediately mounted onto glass slides, and stored at −80 °C until use. For staining, sections were thoroughly washed three times with PBS. For antigen retrieval, sections were incubated in 10 mM citrate buffer (pH 6.0) and autoclaved at 120 °C for 5 min. After naturally cooling to room temperature, sections were washed three times with PBS. Following these steps, sections were incubated with blocking buffer (0.3% Triton X-100, 5% goat serum, and 5% BSA in PBS) for 1 h at room temperature. This was followed by overnight incubation at 4 °C with primary antibodies. The primary antibodies used in this study include: rabbit polyclonal anti-Pan Kla (PTM-1401, PTM BIO, 1:100), rabbit polyclonal anti-H4K12la (PTM-1411, PTM BIO, 1:100), rabbit polyclonal anti-Iba1 (ab178847, Abcam, 1:100), guinea pig anti-Iba1 (OB-PGP049, Oasisbiofarm, 1:100), mouse polyclonal anti-NeuN (ab104224, Abcam, 1:100), rat polyclonal anti-GFAP (13-0300, Invitrogen, 1:100), mouse polyclonal anti-Arg1 (sc-271430, Santa Cruz Biotechnology, 1:50), mouse polyclonal anti-NOS2 (sc-7271, Santa Cruz Biotechnology, 1:50), rabbit polyclonal anti-LDHA (ER00702, Huabio, 1:100), rabbit polyclonal anti-GDNF (ET1704-46, Huabio, 1:100), rabbit polyclonal anti-FGF1 (HA722695, Huabio, 1:100), rabbit polyclonal anti-HDAC1 (ET1605-35, Huabio, 1:100), and mouse polyclonal anti-HDAC1 (66085-1-Ig, Proteintech, 1:100). After the primary antibody step and several rinses, detection was performed using fluorescently conjugated secondary antibodies, including Alexa Fluor 488 donkey anti-rabbit IgG (A-21206, Thermo Fisher Scientific, 1:100), Alexa Fluor 546 donkey anti-mouse IgG (A10036, Thermo Fisher Scientific, 1:100), and Alexa Fluor 647 donkey anti-rat IgG (A78947, Thermo Fisher Scientific, 1:100). Concluding the protocol with final washes, the stained sections were imaged on a Zeiss confocal microscopy platform (Germany).

Protein extraction and Western blot analysis

Spinal cord tissues or isolated microglial cells were immersed in ice-cold radioimmunoprecipitation assay (RIPA) lysis buffer containing a protease and phosphatase inhibitor cocktail (Roche, Germany), and homogenized by sonication. Protein concentrations of extracted total proteins were quantified using a bicinchoninic acid protein (BCA) assay kit (P0010, Beyotime). An equal amount of protein (30 µg) from each sample was electrophoretically separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and then transferred to polyvinylidene fluoride (PVDF) membranes (Merck).

After blocking with 5% non-fat milk in Tris-buffered saline with Tween 20 (TBST, TBS + 0.1% Tween 20) for 1 h at room temperature, membranes were incubated with primary antibodies overnight at 4 °C. The primary antibodies used in this study include: rabbit polyclonal anti-Pan Kla (PTM-1401, PTM BIO, 1:1000), rabbit polyclonal anti-H4K12la (PTM-1411, PTM BIO, 1:1000), rabbit polyclonal anti-LDHA (ER00702, Huabio, 1:1000), rabbit polyclonal anti-HK2 (ER1803-23, Huabio, 1:1000), rabbit polyclonal anti-HDAC1 (ET1605-35, Huabio, 1:1000), rabbit polyclonal anti-GDNF (ET1704-46, Huabio, 1:1000), rabbit polyclonal anti-Histone H3 (PTM-201, PTM BIO, 1:1000), mouse anti-β-actin mAb (TA-09, ZSGB-Bio, 1:5000), and mouse anti-β-tubulin mAb (TA-10, ZSGB-Bio, 1:5000).

Following primary antibody incubation, membranes were washed three times with TBST and then incubated with species-specific horseradish peroxidase (HRP)-conjugated secondary antibodies, including HRP-labeled goat anti-rabbit IgG (ZB-2301, ZSGB-Bio, 1:5000) and HRP-labeled goat anti-mouse IgG (ZB-2305, ZSGB-Bio, 1:5000), for 1 h at room temperature. Protein bands were visualized using an enhanced chemiluminescence (ECL) substrate (Millipore) on a Bio-Rad Gel Doc Imager (Bio-Rad, USA). The intensity of each band was measured using densitometry (ImageJ software). Relative protein expression levels were normalized against β-actin, β-tubulin, or Histone H3 as loading controls.

For validation of microglia-specific knockout, spleens were harvested from Cx3cr1 -CreERT2; Ldha flox/flox mice and control mice at 2 weeks post-tamoxifen injection. Peripheral CD11b+ myeloid cells were isolated using CD11b magnetic microbeads (Miltenyi Biotec) according to the manufacturer’s instructions. Intracellular lactate levels were measured using a lactate assay kit as described above (Supplementary Fig. 3a ).

Quantitative real-time PCR (qPCR)

Total RNA was extracted from fresh spinal cord and brain tissues or isolated microglial cells using TRIzol Reagent (Invitrogen, USA) as per the manufacturer’s protocol. The concentration and purity of total RNA for each sample were determined using a NanoDrop 2000 spectrophotometer (Thermo Scientific, USA). Total RNA (2 µg) was reverse-transcribed into cDNA in a 20 µL reaction volume using FastKing RT Enzyme Mix (TIANGEN, China). Subsequently, cDNA was amplified by SYBR Green (TIANGEN, China) on a Bio-Rad CFX96 Detection System (Bio-Rad, USA). All qPCR reactions were performed in triplicate. The relative expression levels of genes of interest were normalized to that of β -actin using the 2^−ΔΔCt comparative method. Sequences of primers used in qPCR are displayed in Supplementary Table 4 .

Flow cytometry analysis

For flow cytometric analysis of CNS-infiltrating and peripheral immune cells, spinal cords, spleens, and draining lymph nodes were harvested following transcardial perfusion with ice-cold PBS. Spinal cord tissues were mechanically dissociated and enzymatically digested with collagenase D (1 mg mL⁻¹, Roche) and DNase I (0.1 mg mL⁻¹, Sigma-Aldrich) at 37 °C for 30 min. Single-cell suspensions were prepared by passing through a 100 μm cell strainer and subjected to a 30% Percoll gradient centrifugation to enrich mononuclear cells. Spleen and lymph node tissues were mechanically homogenized, passed through cell strainers, and subjected to red blood cell lysis (for spleens) to obtain single-cell suspensions.

To quantify absolute cell numbers in the chronic-phase experiments, Precision Count Beads™ (BioLegend, Cat# 424902) were added (50 μL per sample in a final volume of 350 μL) prior to acquisition. Absolute cell counts were calculated using the formula: absolute count = (number of cell events / number of bead events) × assigned bead count per 50 μL (as provided in the lot-specific certificate of analysis). For surface marker staining, cell suspensions were first incubated with Zombie Green™ Fixable Viability Kit (423112, BioLegend, 1:1000) for 15 min at room temperature in the dark to exclude dead cells. After washing, cells were incubated with anti-CD16/32 antibody (Fc block) to minimize non-specific binding, followed by staining with the following fluorophore-conjugated antibodies for 30 min on ice: PerCP/Cyanine5.5 anti-mouse CD45 (103132, BioLegend, 1:100), APC/Cyanine7 anti-mouse/human CD11b (101226, BioLegend, 1:100), Brilliant Violet 711™ anti-mouse Ly-6C (128037, BioLegend, 1:100), Alexa Fluor® 700 anti-mouse CD3 (100216, BioLegend, 1:100), PE/Cyanine7 anti-mouse CD4 (100421, BioLegend, 1:100), and FITC anti-mouse CD8a (100705, BioLegend, 1:100). Immune cell subsets were identified after sequential gating on singlets (FSC-A vs. FSC-H) and viable cells (Zombie-negative): resident microglia were defined as CD45 int CD11b⁺ cells, infiltrating monocytes/macrophages as CD45 hi CD11b⁺Ly6C⁺ cells, and T lymphocyte subsets as CD3⁺CD4⁺ or CD3⁺CD8a⁺ cells (The detailed gating strategy is provided in Supplementary Fig. 13 ). Samples were acquired on a BD LSRFortessa flow cytometer (BD Biosciences) and analyzed using FlowJo software (v10.8, Tree Star).

Isolation of CNS myeloid cells

To isolate CNS myeloid cells for downstream bulk applications (proteomics and RNA-seq), single-cell suspensions were prepared from mouse spinal cord tissues. To capture the broader CNS myeloid compartment (including both resident microglia and infiltrating myeloid cells), cells were enriched using CD11b magnetic microbeads (Miltenyi Biotec) following the manufacturer’s instructions. In brief, single-cell suspensions (typically 10 6 cells) were resuspended in MACS buffer and incubated with CD11b microbeads. Labeled cells were then isolated using MACS columns. These CD11b-enriched populations were subsequently utilized for bulk proteomic and transcriptomic analyses. For high-resolution validation of microglial-specific effects, scRNA-seq was performed on the total CNS cell population, allowing for computational separation of resident microglia from infiltrating myeloid cells based on established lineage-specific markers (e.g., Tmem119 , P2ry12 , Ccr2 , Ly6c2 ).

RNA isolation and library preparation

Total RNA was extracted using the TRIzol reagent (Invitrogen, CA, USA) according to the manufacturer’s protocol. RNA purity and quantification were evaluated using the NanoDrop 2000 spectrophotometer (Thermo Scientific, USA). RNA integrity was assessed using the Agilent 2100 Bioanalyzer (Agilent Technologies, Santa Clara, CA, USA). Then the libraries were constructed using VAHTS Universal V6 RNA-seq Library Prep Kit according to the manufacturer’s instructions. The transcriptome sequencing and analysis were conducted by OE Biotech Co., Ltd. (Shanghai, China).

RNA sequencing and differentially expressed genes analysis

The libraries were sequenced on an Illumina NovaSeq 6000 platform and 150 bp paired-end reads were generated. Raw reads of fastq format were firstly processed using fastp 47 and the low quality reads were removed to obtain the clean reads. Then about 20,000 clean reads for each sample were retained for subsequent analyses. The clean reads were mapped to the reference genome using HISAT2 48 . FPKM3 of each gene was calculated and the read counts of each gene were obtained by HTSeq-count 49 . PCA analysis were performed using R (v 3.2.0) to evaluate the biological duplication of samples. Differential expression analysis was performed using the DESeq2 50 . Q value < 0.05 and foldchange >2 or foldchange <0.5 was set as the threshold for significantly differential expression gene (DEGs). Hierarchical cluster analysis of DEGs was performed using R (v 3.2.0) to demonstrate the expression pattern of genes in different groups and samples. The radar map of top 30 genes was drew to show the expression of up-regulated or down-regulated DEGs using R packet ggradar. Based on the hypergeometric distribution, GO, KEGG pathway, Reactome and WikiPathways enrichment analysis of DEGs were performed to screen the significant enriched term using R (v 3.2.0), respectively. R (v 3.2.0) was used to draw the column diagram, the chord diagram and bubble diagram of the significant enrichment term.

Quantification of lactate levels

Lactate levels in spinal cord tissue and magnetically sorted cells were measured using the CheKine™ Micro Lactate Assay Kit (Abbkine, Cat# KTB1100) following the manufacturer’s protocol with modifications for low cell numbers. For spinal cord tissue homogenates, tissues were weighed and homogenized on ice in Lactate Assay Buffer at a ratio of 1 mL per 0.1 g of tissue. Homogenates were centrifuged at 12,000 × g for 5 min at 4 °C, and supernatants were collected for measurement. Lactate concentrations in tissue samples were expressed as μmol mL⁻¹. For magnetically sorted cells (e.g., CD11b⁺ microglia/macrophages isolated by MACS), cell pellets were washed twice with cold PBS to remove residual culture medium, and the exact cell number was recorded. Cells were resuspended in Lactate Assay Buffer (100–200 μL, adjusted according to cell yield) and lysed by sonication on ice (200 W, 3 s on/7 s off, 30 cycles). Lysates were centrifuged at 12,000

Plasmid transfection

HMC3 human microglial cells were seeded in 6-well plates and cultured to approximately 60–70% confluence. For overexpression experiments, cells were transfected with the pcDNA3.1-HDAC1-EGFP (Cat#HG11486-ACG, Sino Biological Inc.) plasmid using Lipofectamine™ 2000 (Invitrogen, 11668019) according to the manufacturer’s instructions. Briefly, plasmids were diluted in Opti-MEM™ I Reduced Serum Medium (Gibco) and incubated with Lipofectamine™ 2000. The DNA-lipid complexes were then added dropwise to the cells. After 6–8 h of incubation, the transfection medium was replaced with fresh complete culture medium. Transfected cells were incubated for 48 h before being harvested for subsequent experiments, including Western blot analysis, immunofluorescence staining, and functional assays, to confirm HDAC1 overexpression.

scRNA-seq workflow and bioinformatics

Experimental autoimmune encephalomyelitis (EAE) was induced in both normal-diet and lactate-supplemented mice. On day 10 post-induction, spinal cord tissues were harvested from each group (two mice per group (constituting two independent biological replicates per condition, each pooled from two individual animals)), and single-cell suspensions were prepared. Cell viability was assessed using a Countstar instrument; only samples with viability exceeding 80% were selected for single-cell RNA sequencing (scRNA-seq). scRNA-seq libraries were generated using the 10x Genomics platform, following the manufacturer’s protocol, and sequenced on an Illumina NovaSeq 6000. All downstream analyses were conducted using the Seurat R package (v4.0) 51 . Raw data were first processed with CellRanger (v6.0) for alignment to the mm10 mouse genome and generation of count matrices. Ambient RNA contamination was corrected for each sample using the SoupX R package (default parameters, with contamination fraction estimated via autoEstCont). Corrected matrices were then loaded into Seurat for quality control: cells with fewer than 200 or more than 5000 unique feature counts, or mitochondrial gene percentage >10%, were excluded. Doublets were detected and removed using DoubletFinder, resulting in 51,902 high-quality cells. Data were normalized with NormalizeData. Batch effects were corrected using Harmony 52 . Highly variable features were identified with FindVariableFeatures. Dimensionality reduction was performed via PCA, with significant dimensions selected using JackStrawPlot (dims = 1:30) and ElbowPlot (ndims = 30), leading to the use of the first 20 dimensions where variance explained plateaued. Clustering was executed with FindNeighbors (reduction = “harmony”, dims = 1:20) and FindClusters (resolution = 0.8), yielding 28 clusters. UMAP visualization was generated with RunUMAP (reduction = “harmony”, dims = 1:20). Cell types were annotated using lineage-defining genes and published immunological signatures. Metabolic pathway analysis in microglial clusters was performed with scMetabolism 53 , after converting mouse genes to human orthologs via biomaRt (Ensembl database); pathways with >40% missing genes were excluded. Cell-cell communication was inferred using CellChat 54 based on normalized gene expression data.

CUT&Tag

CUT&Tag 55 was performed with Hyperactive Universal CUT&Tag Assay Kit for Illumina Pro (TD904, Vazyme Biotech) according to the manufacturer’s instructions. Briefly, HMC3 cells, treated with either PBS (control) or LPS plus INF-γ, were bound to ConA magnetic beads, permeabilized with Digitonin, and sequentially incubated with a primary antibody (H4K12la), a secondary antibody, and the Hyperactive pA-Tn5 Transposase. The pA-Tn5 transposase cuts target-bound DNA while simultaneously ligating P5/P7 sequencing adaptors. The samples were then split for two separate analyses. For sequencing, one portion of the sample was PCR-amplified, purified, and quality-controlled with an Agilent 2100 Bioanalyzer before being sequenced on an Illumina NovaSeq 6000 platform. For qPCR validation, the other portion was treated with Stop Buffer at 95 °C for 5 min to release the DNA fragments. The released DNA, which is unsuitable for library construction, was analyzed by SYBR Green qPCR. Enrichment was calculated using the 2-ΔΔCt method, with normalization to a DNA Spike-in and an IgG control.

The raw sequence data were firstly quality trimmed by fastp software to obtain the clean reads. Then the clean reads were aligned to the human genome (GRCh38.p12) using Bowtie2 and subsequently analyzed by the SEACR software based the ‘stringent’parameter to detect genomic regions enriched for multiple overlapping DNA fragments (peaks) that we considered to be putative binding sites. Visualization of peak distribution along genomic regions of interested genes was performed with IGV. Peaks were then annotated using CHIPseeker software to obtain the genes and gene annotations about peaks.

Quantification and statistical analysis

All statistical analyses and visualizations were performed using GraphPad Prism software (version 9.0). Prior to applying statistical tests, data distribution was assessed for normality using the Shapiro-Wilk test. Outlier identification was performed using the ROUT method (Q = 1%) where appropriate to eliminate technical artifacts. Based on the normality assessment and sample size, a tailored statistical approach was employed. For normally distributed datasets with sufficient sample sizes ( n ≥ 5), parametric tests were maintained to ensure optimal statistical power. Specifically, an unpaired, two-tailed Student’s t-test was used for two-group comparisons, while Welch’s t-test was applied for groups with unequal variances. A one-way analysis of variance (ANOVA) followed by Tukey’s or Šídák’s post-hoc tests was utilized for multiple comparisons. For longitudinal clinical EAE score trajectories, a two-way repeated measures ANOVA was applied. For datasets that violated normality assumptions (e.g., human clinical CSF parameters), non-parametric tests, specifically the Mann–Whitney U test, were applied. Furthermore, for in vitro assays limited to small sample sizes ( n = 3 or 4 independent biological replicates), parametric tests were strictly retained, as non-parametric alternatives lack sufficient mathematical statistical power to detect significance at such sample sizes. Correlation analyses were evaluated using Spearman’s rank correlation coefficient. All quantitative data are presented as mean ± standard error of the mean (SEM) unless otherwise stated. For box plots, the center line represents the median, the bounds of the box represent the 25th and 75th percentiles, and the whiskers extend to the minima and maxima. Statistical significance was defined as P < 0.05. To ensure full transparency, exact P values, test statistics ( t , F , or U ), degrees of freedom ( df ), and 95% confidence intervals (CI) are explicitly detailed in the respective figure legends and the Source Data file. All experiments showing representative micrographs or immunoblots were independently repeated at least three times with similar results.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Supplementary information

Supplementary Information Reporting Summary Transparent Peer Review file

Source data

Soure Data

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

These authors contributed equally: Jiahong Li, Su Meng, Jiajian Li.

Supplementary information

The online version contains supplementary material available at https://doi.org/10.1038/s41467-026-76302-z .

Acknowledgements

We thank the staff at the animal facility and the NIC Neuroimmunology Laboratory of Xuanwu Hospital, Capital Medical University, for their invaluable technical support and assistance.

Author contributions

Jiahong L., S.M. and Jiajian L. designed the experiments. Jiahong L. and S.M. performed the most of the experiments. Jiajian L. and Q.W. performed bioinformatic analysis. X.H. helped perform the immunofluorescence staining of mouse samples. J.F., W.H., H.W. and Y.Z. wrote the manuscript. All authors discussed and commented on the manuscript.

Peer review

Peer review information

Nature Communications thanks the anonymous reviewer(s) for their contribution to the peer review of this work. A peer review file is available.

Funding

This study was supported by the Beijing Natural Science Foundation (Grant No. L2510031), the National Natural Science Foundation of China (Grant Nos. 82571539 and 82501612), the Beijing Research Ward Excellence Program (Grant No. BRWEP2024W022010104), the grant from the Chinese Institutes for Medical Research, Beijing (Grant No. CX23YZ15), the Beijing Scholars Program (Grant No. 106), the Project for Innovation and Development of Beijing Municipal Geriatric Medical Research Center (Grant No. 11000023T000002041657), the Dengfeng Talent Program (Grant No. DFL20220701), the Xuanwu Hospital Talent Convergence Program – Leading Talents (Grant No. HZ2021ZCLJ008), the Beijing Health System Leading Talent Grant (Grant No. 2022-02-10), the Beijing Hospitals Authority’s Ascent Plan (Grant No. DFL20240801), and the Beijing “Huizhi” Talent Program – Cultivation Program for Leading Talents (Grant No. HZ2025PYDTR003).

Data availability

The mass spectrometry proteomics data generated in this study have been deposited in the PRIDE database under accession code PXD067298 . The bulk RNA-seq data generated in this study have been deposited in the Genome Sequence Archive (GSA) database under accession code CRA028941 . The single-cell RNA sequencing data generated in this study have been deposited in the GSA database under accession code CRA028954 . The processed CUT&Tag data generated in this study are available in the OMIX database under accession code OMIX011457 . The raw CUT&Tag sequencing data generated in this study have been deposited in the Genome Sequence Archive for Human (GSA-Human) database under accession code HRA012789 . All other relevant data generated in this study are provided in the Supplementary Information and Source Data files. Source data are provided with this paper.

Competing interests

The authors declare no conflicts of interest.

References

1. Thompson AJ Baranzini SE Geurts J Hemmer B Ciccarelli O Multiple sclerosis Lancet 2018 391 1622 1636 10.1016/S0140-6736(18)30481-1 29576504 Thompson, A. J., Baranzini, S. E., Geurts, J., Hemmer, B. & Ciccarelli, O. Multiple sclerosis. Lancet 391 , 1622–1636 (2018). 29576504 10.1016/S0140-6736(18)30481-1

2. Distéfano-Gagné F Bitarafan S Lacroix S Gosselin D Roles and regulation of microglia activity in multiple sclerosis: Insights from animal models Nat. Rev. Neurosci. 2023 24 397 415 10.1038/s41583-023-00709-6 37268822 Distéfano-Gagné, F., Bitarafan, S., Lacroix, S. & Gosselin, D. Roles and regulation of microglia activity in multiple sclerosis: Insights from animal models. Nat. Rev. Neurosci. 24 , 397–415 (2023). 37268822 10.1038/s41583-023-00709-6

3. Hernández-Pedro NY Espinosa-Ramirez G de la Cruz VP Pineda B Sotelo J Initial immunopathogenesis of multiple sclerosis: Innate immune response J. Immunol. Res. 2013 2013 413465 10.1155/2013/413465 PMC3794540 24174969 Hernández-Pedro, N. Y., Espinosa-Ramirez, G., de la Cruz, V. P., Pineda, B. & Sotelo, J. Initial immunopathogenesis of multiple sclerosis: Innate immune response. J. Immunol. Res. 2013 , 413465 (2013). 10.1155/2013/413465 PMC3794540 24174969

4. Hemmer B Kerschensteiner M Korn T Role of the innate and adaptive immune responses in the course of multiple sclerosis Lancet Neurol. 2015 14 406 419 10.1016/S1474-4422(14)70305-9 25792099 Hemmer, B., Kerschensteiner, M. & Korn, T. Role of the innate and adaptive immune responses in the course of multiple sclerosis. Lancet Neurol. 14 , 406–419 (2015). 25792099 10.1016/S1474-4422(14)70305-9

5. Bao C Histone lactylation in macrophage biology and disease: from plasticity regulation to therapeutic implications EBioMedicine 2025 111 105502 10.1016/j.ebiom.2024.105502 39662177 PMC11697715 Bao, C. et al. Histone lactylation in macrophage biology and disease: from plasticity regulation to therapeutic implications. EBioMedicine 111 , 105502 (2025). 39662177 10.1016/j.ebiom.2024.105502 PMC11697715

6. Hickman S Izzy S Sen P Morsett L El Khoury J Microglia in neurodegeneration Nat. Neurosci. 2018 21 1359 1369 10.1038/s41593-018-0242-x 30258234 PMC6817969 Hickman, S., Izzy, S., Sen, P., Morsett, L. & El Khoury, J. Microglia in neurodegeneration. Nat. Neurosci. 21 , 1359–1369 (2018). 30258234 10.1038/s41593-018-0242-x PMC6817969

7. Baik SH A breakdown in metabolic reprogramming causes microglia dysfunction in Alzheimer’s disease Cell Metab. 2019 30 493 507.e6 10.1016/j.cmet.2019.06.005 31257151 Baik, S. H. et al. A breakdown in metabolic reprogramming causes microglia dysfunction in Alzheimer’s disease. Cell Metab. 30 , 493–507.e6 (2019). 31257151 10.1016/j.cmet.2019.06.005

8. Monsorno K Buckinx A Paolicelli RC Microglial metabolic flexibility: emerging roles for lactate Trends Endocrinol. Metab. 2022 33 186 195 10.1016/j.tem.2021.12.001 34996673 Monsorno, K., Buckinx, A. & Paolicelli, R. C. Microglial metabolic flexibility: emerging roles for lactate. Trends Endocrinol. Metab. 33 , 186–195 (2022). 34996673 10.1016/j.tem.2021.12.001

9. Li X Lactate metabolism in human health and disease Signal Transduct. Target. Ther. 2022 7 305 10.1038/s41392-022-01151-3 36050306 PMC9434547 Li, X. et al. Lactate metabolism in human health and disease. Signal Transduct. Target. Ther. 7 , 305 (2022). 36050306 10.1038/s41392-022-01151-3 PMC9434547

10. Sanmarco LM Lactate limits CNS autoimmunity by stabilizing HIF-1α in dendritic cells Nature 2023 620 881 889 10.1038/s41586-023-06409-6 37558878 PMC10725186 Sanmarco, L. M. et al. Lactate limits CNS autoimmunity by stabilizing HIF-1α in dendritic cells. Nature 620 , 881–889 (2023). 37558878 10.1038/s41586-023-06409-6 PMC10725186

11. Certo M Tsai C-H Pucino V Ho P-C Mauro C Lactate modulation of immune responses in inflammatory versus tumour microenvironments Nat. Rev. Immunol. 2021 21 151 161 10.1038/s41577-020-0406-2 32839570 Certo, M., Tsai, C.-H., Pucino, V., Ho, P.-C. & Mauro, C. Lactate modulation of immune responses in inflammatory versus tumour microenvironments. Nat. Rev. Immunol. 21 , 151–161 (2021). 32839570 10.1038/s41577-020-0406-2

12. Netea MG Defining trained immunity and its role in health and disease Nat. Rev. Immunol. 2020 20 375 388 10.1038/s41577-020-0285-6 32132681 PMC7186935 Netea, M. G. et al. Defining trained immunity and its role in health and disease. Nat. Rev. Immunol. 20 , 375–388 (2020). 32132681 10.1038/s41577-020-0285-6 PMC7186935

13. Zhang D Metabolic regulation of gene expression by histone lactylation Nature 2019 574 575 580 10.1038/s41586-019-1678-1 31645732 PMC6818755 Zhang, D. et al. Metabolic regulation of gene expression by histone lactylation. Nature 574 , 575–580 (2019). 31645732 10.1038/s41586-019-1678-1 PMC6818755

14. Gong H Zhong H Cheng L Li L-P Zhang D-K Post-translational protein lactylation modification in health and diseases: A double-edged sword J. Transl. Med. 2024 22 41 10.1186/s12967-023-04842-9 38200523 PMC10777551 Gong, H., Zhong, H., Cheng, L., Li, L.-P. & Zhang, D.-K. Post-translational protein lactylation modification in health and diseases: A double-edged sword. J. Transl. Med. 22 , 41 (2024). 38200523 10.1186/s12967-023-04842-9 PMC10777551

15. Liu J Zhao F Qu Y Lactylation: A novel post-translational modification with clinical implications in CNS diseases Biomolecules 2024 14 1175 10.3390/biom14091175 39334941 PMC11430557 Liu, J., Zhao, F. & Qu, Y. Lactylation: A novel post-translational modification with clinical implications in CNS diseases. Biomolecules 14 , 1175 (2024). 39334941 10.3390/biom14091175 PMC11430557

16. Hu Y Lactylation: the histone modification influence on gene expression, protein function, and disease Clin. Epigen. 2024 16 72 10.1186/s13148-024-01682-2 PMC11138093 38812044 Hu, Y. et al. Lactylation: the histone modification influence on gene expression, protein function, and disease. Clin. Epigen. 16 , 72 (2024). 10.1186/s13148-024-01682-2 PMC11138093 38812044

17. Wang Y Lactate metabolism and histone lactylation in the central nervous system disorders: impacts and molecular mechanisms J. Neuroinflamm. 2024 21 308 10.1186/s12974-024-03303-4 PMC11605911 39609834 Wang, Y. et al. Lactate metabolism and histone lactylation in the central nervous system disorders: impacts and molecular mechanisms. J. Neuroinflamm. 21 , 308 (2024). 10.1186/s12974-024-03303-4 PMC11605911 39609834

18. Qin Q Enhanced glycolysis-derived lactate promotes microglial activation in Parkinson’s disease via histone lactylation NPJ Park. Dis. 2025 11 3 10.1038/s41531-024-00858-0 PMC11698869 39753581 Qin, Q. et al. Enhanced glycolysis-derived lactate promotes microglial activation in Parkinson’s disease via histone lactylation. NPJ Park. Dis. 11 , 3 (2025). 10.1038/s41531-024-00858-0 PMC11698869 39753581

19. Li P Histone H3 lysine 18 lactylation attenuates neuroinflammation and neurological damage by regulating microglial plxnb2 after ischemic stroke Neurobiol. Dis. 2025 215 107061 10.1016/j.nbd.2025.107061 40835170 Li, P. et al. Histone H3 lysine 18 lactylation attenuates neuroinflammation and neurological damage by regulating microglial plxnb2 after ischemic stroke. Neurobiol. Dis. 215 , 107061 (2025). 40835170 10.1016/j.nbd.2025.107061

20. Hu X Lactate promotes microglial scar formation and facilitates locomotor function recovery by enhancing histone H4 lysine 12 lactylation after spinal cord injury J. Neuroinflamm. 2024 21 193 10.1186/s12974-024-03186-5 PMC11297795 39095832 Hu, X. et al. Lactate promotes microglial scar formation and facilitates locomotor function recovery by enhancing histone H4 lysine 12 lactylation after spinal cord injury. J. Neuroinflamm. 21 , 193 (2024). 10.1186/s12974-024-03186-5 PMC11297795 39095832

21. Zhang B Single-cell RNA sequencing integrated with bulk RNA sequencing analysis reveals the protective effects of lactate-mediated lactylation of microglia-related proteins on spinal cord injury CNS Neurosci. Ther. 2024 30 e70028 10.1111/cns.70028 39218784 PMC11366449 Zhang, B. et al. Single-cell RNA sequencing integrated with bulk RNA sequencing analysis reveals the protective effects of lactate-mediated lactylation of microglia-related proteins on spinal cord injury. CNS Neurosci. Ther. 30 , e70028 (2024). 39218784 10.1111/cns.70028 PMC11366449

22. Sheng X Lin H Cole PA Zhao Y Biochemistry and regulation of histone lysine l-lactylation Nat. Rev. Mol. Cell Biol. 2026 27 95 109 10.1038/s41580-025-00876-7 40830268 PMC12920031 Sheng, X., Lin, H., Cole, P. A. & Zhao, Y. Biochemistry and regulation of histone lysine l-lactylation. Nat. Rev. Mol. Cell Biol. 27 , 95–109 (2026). 40830268 10.1038/s41580-025-00876-7 PMC12920031

23. Moreno-Yruela C Class I histone deacetylases (HDAC1-3) are histone lysine delactylases Sci. Adv. 2022 8 eabi6696 10.1126/sciadv.abi6696 35044827 PMC8769552 Moreno-Yruela, C. et al. Class I histone deacetylases (HDAC1-3) are histone lysine delactylases. Sci. Adv. 8 , eabi6696 (2022). 35044827 10.1126/sciadv.abi6696 PMC8769552

24. Dendrou CA Fugger L Friese MA Immunopathology of multiple sclerosis Nat. Rev. Immunol. 2015 15 545 558 10.1038/nri3871 26250739 Dendrou, C. A., Fugger, L. & Friese, M. A. Immunopathology of multiple sclerosis. Nat. Rev. Immunol. 15 , 545–558 (2015). 26250739 10.1038/nri3871

25. Wang N Histone Lactylation Boosts Reparative Gene Activation Post–Myocardial Infarction Circ. Res. 2022 131 893 908 10.1161/CIRCRESAHA.122.320488 36268709 Wang, N. et al. Histone Lactylation Boosts Reparative Gene Activation Post–Myocardial Infarction. Circ. Res. 131 , 893–908 (2022). 36268709 10.1161/CIRCRESAHA.122.320488

26. Kelly B O’Neill LA Metabolic reprogramming in macrophages and dendritic cells in innate immunity Cell Res 2015 25 771 784 10.1038/cr.2015.68 26045163 PMC4493277 Kelly, B. & O’Neill, L. A. Metabolic reprogramming in macrophages and dendritic cells in innate immunity. Cell Res . 25 , 771–784 (2015). 26045163 10.1038/cr.2015.68 PMC4493277

27. Magistretti PJ Allaman I Lactate in the brain: from metabolic end-product to signalling molecule Nat. Rev. Neurosci. 2018 19 235 249 10.1038/nrn.2018.19 29515192 Magistretti, P. J. & Allaman, I. Lactate in the brain: from metabolic end-product to signalling molecule. Nat. Rev. Neurosci. 19 , 235–249 (2018). 29515192 10.1038/nrn.2018.19

28. Sun M Zhang X Feng X Liang L PKM2 accelerated the progression of chronic fatigue syndrome via promoting the H4K12la/ NF-κB induced neuroinflammation and mitochondrial damage Sci. Rep. 2025 15 10772 10.1038/s41598-025-93313-w 40155479 PMC11953386 Sun, M., Zhang, X., Feng, X. & Liang, L. PKM2 accelerated the progression of chronic fatigue syndrome via promoting the H4K12la/ NF-κB induced neuroinflammation and mitochondrial damage. Sci. Rep. 15 , 10772 (2025). 40155479 10.1038/s41598-025-93313-w PMC11953386

29. Colegio OR Functional polarization of tumour-associated macrophages by tumour-derived lactic acid Nature 2014 513 559 563 10.1038/nature13490 25043024 PMC4301845 Colegio, O. R. et al. Functional polarization of tumour-associated macrophages by tumour-derived lactic acid. Nature 513 , 559–563 (2014). 25043024 10.1038/nature13490 PMC4301845

30. Pan R-Y Positive feedback regulation of microglial glucose metabolism by histone H4 lysine 12 lactylation in Alzheimer’s disease Cell Metab. 2022 34 634 648.e6 10.1016/j.cmet.2022.02.013 35303422 Pan, R.-Y. et al. Positive feedback regulation of microglial glucose metabolism by histone H4 lysine 12 lactylation in Alzheimer’s disease. Cell Metab. 34 , 634–648.e6 (2022). 35303422 10.1016/j.cmet.2022.02.013

31. Wei L H3K18 lactylation of senescent microglia potentiates brain aging and Alzheimer’s disease through the NFκB signaling pathway J. Neuroinflamm. 2023 20 208 10.1186/s12974-023-02879-7 PMC10494370 37697347 Wei, L. et al. H3K18 lactylation of senescent microglia potentiates brain aging and Alzheimer’s disease through the NFκB signaling pathway. J. Neuroinflamm. 20 , 208 (2023). 10.1186/s12974-023-02879-7 PMC10494370 37697347

32. Sakushima K Hayashino Y Kawaguchi T Jackson JL Fukuhara S Diagnostic accuracy of cerebrospinal fluid lactate for differentiating bacterial meningitis from aseptic meningitis: A meta-analysis J. Infect. 2011 62 255 262 10.1016/j.jinf.2011.02.010 21382412 Sakushima, K., Hayashino, Y., Kawaguchi, T., Jackson, J. L. & Fukuhara, S. Diagnostic accuracy of cerebrospinal fluid lactate for differentiating bacterial meningitis from aseptic meningitis: A meta-analysis. J. Infect. 62 , 255–262 (2011). 21382412 10.1016/j.jinf.2011.02.010

33. Netea MG Trained immunity: A program of innate immune memory in health and disease Science 2016 352 aaf1098 10.1126/science.aaf1098 27102489 PMC5087274 Netea, M. G. et al. Trained immunity: A program of innate immune memory in health and disease. Science 352 , aaf1098 (2016). 27102489 10.1126/science.aaf1098 PMC5087274

34. Kleinnijenhuis J Bacille Calmette-Guérin induces NOD2-dependent nonspecific protection from reinfection via epigenetic reprogramming of monocytes Proc. Natl. Acad. Sci. 2012 109 17537 17542 10.1073/pnas.1202870109 22988082 PMC3491454 Kleinnijenhuis, J. et al. Bacille Calmette-Guérin induces NOD2-dependent nonspecific protection from reinfection via epigenetic reprogramming of monocytes. Proc. Natl. Acad. Sci. 109 , 17537–17542 (2012). 22988082 10.1073/pnas.1202870109 PMC3491454

35. Ziogas A Long-term histone lactylation connects metabolic and epigenetic rewiring in innate immune memory Cell 2025 188 2992 3012.e16 10.1016/j.cell.2025.03.048 40318634 Ziogas, A. et al. Long-term histone lactylation connects metabolic and epigenetic rewiring in innate immune memory. Cell 188 , 2992–3012.e16 (2025). 40318634 10.1016/j.cell.2025.03.048

36. Cai H Lactate activates trained immunity by fueling the tricarboxylic acid cycle and regulating histone lactylation Nat. Commun. 2025 16 3230 10.1038/s41467-025-58563-2 40185732 PMC11971257 Cai, H. et al. Lactate activates trained immunity by fueling the tricarboxylic acid cycle and regulating histone lactylation. Nat. Commun. 16 , 3230 (2025). 40185732 10.1038/s41467-025-58563-2 PMC11971257

37. Ransohoff RM Perry VH Microglial physiology: Unique stimuli, specialized responses Annu. Rev. Immunol. 2009 27 119 145 10.1146/annurev.immunol.021908.132528 19302036 Ransohoff, R. M. & Perry, V. H. Microglial physiology: Unique stimuli, specialized responses. Annu. Rev. Immunol. 27 , 119–145 (2009). 19302036 10.1146/annurev.immunol.021908.132528

38. Ransohoff RM A polarizing question: Do M1 and M2 microglia exist? Nat. Neurosci. 2016 19 987 991 10.1038/nn.4338 27459405 Ransohoff, R. M. A polarizing question: Do M1 and M2 microglia exist? Nat. Neurosci. 19 , 987–991 (2016). 27459405 10.1038/nn.4338

39. Guo, S., Wang, H. & Yin, Y. Microglia polarization from M1 to M2 in neurodegenerative diseases. Front. Aging Neurosci . 14 , 815347 (2022). 10.3389/fnagi.2022.815347 PMC8888930 35250543

40. Miron VE M2 microglia/macrophages drive oligodendrocyte differentiation during CNS remyelination Nat. Neurosci. 2013 16 1211 1218 10.1038/nn.3469 23872599 PMC3977045 Miron, V. E. et al. M2 microglia/macrophages drive oligodendrocyte differentiation during CNS remyelination. Nat. Neurosci. 16 , 1211–1218 (2013). 23872599 10.1038/nn.3469 PMC3977045

41. Lloyd AF Miron VE The pro-remyelination properties of microglia in the central nervous system Nat. Rev. Neurol. 2019 15 447 458 10.1038/s41582-019-0184-2 31256193 Lloyd, A. F. & Miron, V. E. The pro-remyelination properties of microglia in the central nervous system. Nat. Rev. Neurol. 15 , 447–458 (2019). 31256193 10.1038/s41582-019-0184-2

42. Han H Exercise improves cognitive dysfunction and neuroinflammation in mice through Histone H3 lactylation in microglia Immun. Ageing A 2023 20 63 10.1186/s12979-023-00390-4 PMC10655345 37978517 Han, H. et al. Exercise improves cognitive dysfunction and neuroinflammation in mice through Histone H3 lactylation in microglia. Immun. Ageing A . 20 , 63 (2023). 10.1186/s12979-023-00390-4 PMC10655345 37978517

43. Hu X Microglial and macrophage polarization-new prospects for brain repair Nat. Rev. Neurol. 2015 11 56 64 10.1038/nrneurol.2014.207 25385337 PMC4395497 Hu, X. et al. Microglial and macrophage polarization-new prospects for brain repair. Nat. Rev. Neurol. 11 , 56–64 (2015). 25385337 10.1038/nrneurol.2014.207 PMC4395497

44. Holtman IR Skola D Glass CK Transcriptional control of microglia phenotypes in health and disease J. Clin. Invest. 2017 127 3220 3229 10.1172/JCI90604 28758903 PMC5669536 Holtman, I. R., Skola, D. & Glass, C. K. Transcriptional control of microglia phenotypes in health and disease. J. Clin. Invest. 127 , 3220–3229 (2017). 28758903 10.1172/JCI90604 PMC5669536

45. Nohesara S Abdolmaleky HM Thiagalingam S Potential for new therapeutic approaches by targeting lactate and pH mediated epigenetic dysregulation in major mental diseases Biomedicines 2024 12 457 10.3390/biomedicines12020457 38398057 PMC10887322 Nohesara, S., Abdolmaleky, H. M. & Thiagalingam, S. Potential for new therapeutic approaches by targeting lactate and pH mediated epigenetic dysregulation in major mental diseases. Biomedicines 12 , 457 (2024). 38398057 10.3390/biomedicines12020457 PMC10887322

46. Wang Z-H Peng W-B Zhang P Yang X-P Zhou Q Lactate in the tumour microenvironment: From immune modulation to therapy EBioMedicine 2021 73 103627 10.1016/j.ebiom.2021.103627 34656878 PMC8524104 Wang, Z.-H., Peng, W.-B., Zhang, P., Yang, X.-P. & Zhou, Q. Lactate in the tumour microenvironment: From immune modulation to therapy. EBioMedicine 73 , 103627 (2021). 34656878 10.1016/j.ebiom.2021.103627 PMC8524104

47. Chen S Zhou Y Chen Y Gu J fastp: An ultra-fast all-in-one FASTQ preprocessor Bioinformatics 2018 34 i884 i890 10.1093/bioinformatics/bty560 30423086 PMC6129281 Chen, S., Zhou, Y., Chen, Y. & Gu, J. fastp: An ultra-fast all-in-one FASTQ preprocessor. Bioinformatics 34 , i884–i890 (2018). 30423086 10.1093/bioinformatics/bty560 PMC6129281

48. Kim D Langmead B Salzberg SL HISAT: A fast spliced aligner with low memory requirements Nat. Methods 2015 12 357 360 10.1038/nmeth.3317 25751142 PMC4655817 Kim, D., Langmead, B. & Salzberg, S. L. HISAT: A fast spliced aligner with low memory requirements. Nat. Methods 12 , 357–360 (2015). 25751142 10.1038/nmeth.3317 PMC4655817

49. Anders S Pyl PT Huber W HTSeq—a Python framework to work with high-throughput sequencing data Bioinformatics 2015 31 166 169 10.1093/bioinformatics/btu638 25260700 PMC4287950 Anders, S., Pyl, P. T. & Huber, W. HTSeq—a Python framework to work with high-throughput sequencing data. Bioinformatics 31 , 166–169 (2015). 25260700 10.1093/bioinformatics/btu638 PMC4287950

50. Love MI Huber W Anders S Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2 Genome Biol. 2014 15 550 10.1186/s13059-014-0550-8 25516281 PMC4302049 Love, M. I., Huber, W. & Anders, S. Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biol. 15 , 550 (2014). 25516281 10.1186/s13059-014-0550-8 PMC4302049

51. Hao Y Integrated analysis of multimodal single-cell data Cell 2021 184 3573 3587.e29 10.1016/j.cell.2021.04.048 34062119 PMC8238499 Hao, Y. et al. Integrated analysis of multimodal single-cell data. Cell 184 , 3573–3587.e29 (2021). 34062119 10.1016/j.cell.2021.04.048 PMC8238499

52. Korsunsky I Fast, sensitive, and accurate integration of single cell data with Harmony Nat. Methods 2019 16 1289 1296 10.1038/s41592-019-0619-0 31740819 PMC6884693 Korsunsky, I. et al. Fast, sensitive, and accurate integration of single cell data with Harmony. Nat. Methods 16 , 1289–1296 (2019). 31740819 10.1038/s41592-019-0619-0 PMC6884693

53. Wu Y Spatiotemporal immune landscape of colorectal cancer liver metastasis at single-cell level Cancer Discov. 2022 12 134 153 10.1158/2159-8290.CD-21-0316 34417225 Wu, Y. et al. Spatiotemporal immune landscape of colorectal cancer liver metastasis at single-cell level. Cancer Discov. 12 , 134–153 (2022). 34417225 10.1158/2159-8290.CD-21-0316

54. Jin S Inference and analysis of cell-cell communication using CellChat Nat. Commun. 2021 12 1088 10.1038/s41467-021-21246-9 33597522 PMC7889871 Jin, S. et al. Inference and analysis of cell-cell communication using CellChat. Nat. Commun. 12 , 1088 (2021). 33597522 10.1038/s41467-021-21246-9 PMC7889871

55. Kaya-Okur HS CUT&Tag for efficient epigenomic profiling of small samples and single cells Nat. Commun. 2019 10 1930 10.1038/s41467-019-09982-5 31036827 PMC6488672 Kaya-Okur, H. S. et al. CUT&Tag for efficient epigenomic profiling of small samples and single cells. Nat. Commun. 10 , 1930 (2019). 31036827 10.1038/s41467-019-09982-5 PMC6488672