STING controls glycolysis and histone lactylation to drive macrophage metabolic reprogramming in postoperative ileus

Abstract

Postoperative ileus (POI) is characterized by dysregulated inflammation within the intestinal muscular layer, which significantly disrupts gastrointestinal motility and presents a major challenge to postoperative recovery. Although macrophages are known to contribute to inflammation through glycolytic bursts that support rapid energy production, the role of the stimulator of interferon genes (STING) in orchestrating macrophage glycolysis and modulating phenotypic polarization remains poorly defined. To address this gap, we examined the regulatory relationship between STING and macrophage metabolism. Here, we demonstrate that lipopolysaccharide (LPS)-stimulated RAW 264.7 cells display a pronounced enhancement of glycolysis, an effect that was markedly attenuated in STING knockout (STING KO) cells. Further analysis revealed that STING deletion reduces histone lactylation, consequently restricting chromatin accessibility at the hexokinase 2 (HK2) gene loci. Through CUT&Tag sequencing, we identified IRF3 as a transcription factor that directly binds to the promoter regions of HK2 and enhances its expression. Our results delineate a STING-regulated glycolytic feedback loop in macrophages: STING stabilizes hypoxia-inducible factor 1-alpha (HIF1α), thereby amplifying glycolysis and promoting histone lactylation at HK2 loci. This epigenetic modification facilitates IRF3 binding to the HK2 promoter, further boosting HK2 expression and sustaining glycolytic flux. Together, these findings elucidate a molecular mechanism through which STING modulates macrophage polarization via metabolic reprogramming, highlighting the therapeutic potential of targeting STING to regulate macrophage metabolism, alleviate inflammation, and improve outcomes in POI.

This study shows STING drives macrophage metabolic reprogramming in postoperative ileus (POI). It stabilizes HIF1α to boost glycolysis, enhances H4K8la at HK2 loci, aids IRF3 binding to HK2 promoter, and promotes M1 polarization. Targeting STING may ease POI inflammation.

Introduction

Postoperative ileus (POI) is a common complication following abdominal surgery, characterized by delayed recovery of intestinal function, prolonged time to flatus and bowel movements, nausea, vomiting, inability to eat, and the need for parenteral nutrition¹. Despite advancements in surgical techniques to reduce the risk of POI, it remains prevalent, with an incidence of up to 17% following colorectal surgery. POI can extend hospitalization by ~29% and increase costs by 15%² The pathophysiology of POI is multifactorial, involving interactions between neurons and immune cells^3,4. Macrophages, present in the lamina propria and muscularis layers of the gastrointestinal tract, play essential roles in maintaining epithelial integrity and regulating gastrointestinal motility. Surgical manipulation induces inflammatory cell infiltration into the intestinal muscularis, and the degree of paralytic ileus correlates with the severity of muscularis inflammation^5,6 Surgical trauma promotes reactive oxygen species (ROS) release from the intestinal epithelium, increasing epithelial permeability and causing microbial translocation^7,8. While subepithelial macrophages exhibit weak responses to microbial translocation, resident muscularis macrophages (MMs) are rapidly activated, releasing pro-inflammatory mediators and recruiting circulating leukocytes⁹. Suppression of early MM activation and reduction of inflammatory cell infiltration into the muscularis can improve gastrointestinal motility¹⁰. Infiltrating immune cells, particularly macrophages, sustain a pro-inflammatory phenotype, amplifying inflammation. Understanding how infiltrating macrophages influence the muscularis inflammatory microenvironment is crucial for improving postoperative intestinal recovery.

Under inflammatory stress, immune cells exhibit a metabolic shift reminiscent of the “Warburg effect” in tumor cells, transitioning from oxidative phosphorylation (OXPHOS) to aerobic glycolysis to rapidly generate energy required for activation and proliferation¹¹. Macrophages undergoing cytokine or endotoxin stimulation display this metabolic switch, yielding two primary effects: (1) In certain conditions, the switch to glycolysis sustains macrophage survival. For instance, in acute respiratory distress syndrome (ARDS), hypoxia in alveolar macrophages suppresses OXPHOS, and enhanced glycolysis provides the energy necessary for cell survival¹². (2) Glycolysis supports macrophage polarization into a pro-inflammatory phenotype. For example, LPS-induced glycolytic bursts drive the pentose phosphate pathway to produce NADPH and ROS, such as H₂O₂, which act as intracellular signals to activate downstream pathways like NF-κB, promoting the expression of IL-1β and TNF-α¹¹ The transcription factor HIF1α, under hypoxic or inflammatory conditions, enhances the transcription of glycolytic genes. M1 macrophage polarization is glycolysis-dependent, and pharmacological inhibition of HIF1α suppresses glycolytic enzyme expression, mitigating M1 polarization¹³.

Stimulator of Interferon Genes (STING) is a pivotal molecule in innate immunity. Initially discovered for its role in inducing type II interferons in response to viral DNA, STING has also been implicated in immune responses to bacterial infections and oxidative stress. In macrophages, the STING-IRF3-NLRP3 pathway is well-studied. Damage-associated mitochondrial DNA activates STING, recruiting TBK1 to phosphorylate IRF3, which translocates to the nucleus to induce NLRP3 expression, exacerbating oxidative stress¹⁴. TBK1, a downstream effector of STING, drives macrophage inflammatory phenotypes by phosphorylating ACLY, an enzyme that generates acetyl-CoA from pyruvate. Elevated acetyl-CoA promotes histone acetylation, enhancing inflammatory gene expression¹⁵.

Recent studies have highlighted STING’s involvement in metabolic reprogramming—an adaptive process by which cells alter metabolic pathways to maintain functionality under physiological or pathological stress. Emerging evidence suggests STING contributes to metabolic reprogramming under specific conditions: (1) During bacterial infections, STING activation by cyclic dinucleotides enhances glycolysis and synergizes with STAT3 to accelerate antimicrobial peptide production¹⁶. (2) In tumor microenvironments, STING establishes a positive feedback loop with glycolysis in tumor-associated dendritic cells (DCs), supporting antitumor immunity by enhancing HIF1α-induced glycolytic gene expression. ATP produced from glycolysis sustains STING phosphorylation, maintaining downstream signaling¹⁷. (3) In renal fibrosis models, STING inhibition reduces tubular glycolysis, mitigating pro-fibrotic phenotypes¹⁸.

Although extensive research links STING to glycolysis-driven phenotype modulation, its precise mechanisms remain unclear. IRF3, a classical downstream effector of STING, might bridge STING and glycolysis. As a member of the IRF family, IRF3 is known for its antiviral roles, but emerging evidence highlights its involvement in metabolic regulation, including limiting thermogenesis in adipocytes and contributing to insulin resistance¹⁹. IRF3 promotes glycolysis by binding to the promoter region of PFKFB3, a key glycolytic enzyme¹⁸. However, the mechanisms by which STING regulates macrophage glycolysis remain poorly understood.

Given the strong links between the STING-IRF3 pathway, metabolic reprogramming, and glycolysis, this axis represents a promising therapeutic target for modulating inflammatory microenvironments. This study aims to elucidate the relationship between STING, macrophage glycolysis, and polarization, particularly in the intestinal muscularis inflammatory microenvironment. We propose that STING-mediated enhancement of glycolysis accelerates lactate production, thereby inducing histone lactylation at glycolytic gene promoter regions. This mechanism amplifies gene expression and establishes a positive feedback loop essential for macrophage polarization. To validate this hypothesis, we activated STING and glycolysis in macrophages using LPS, assessed glycolysis in STING knockout RAW macrophages, and utilized epigenetic techniques such as ATAC-seq and CUT&Tag-seq to explore the regulatory mechanisms of histone lactylation and identify new IRF3 binding sites.

Result

Impaired intestinal motility and inflammatory cell infiltration following surgical manipulation

To investigate the inflammatory changes in the intestinal muscularis layer induced by surgical manipulation, we established an animal model of postoperative ileus (POI). Intestinal obstruction and edema were observed in mice following intestinal manipulation (Fig. 1A). Gastrointestinal transit (GIT) was measured using fluorescein isothiocyanate (FITC)-dextran distribution in the small and large intestines (Fig. 1B). Compared with the sham-operated group, the intestinal manipulation (IM) group exhibited a shorter GIT distance, indicating reduced gastrointestinal motility caused by IM.

Fig. 1. IM induces a substantial infiltration of inflammatory cells in the intestinal muscle layer.

A Gross pictures of the representational photo in the Sham and IM groups. The area in the white rectangles displays distinct edema and food obstruction in the small intestine. B FITC-dextran distribution in the intestinal tract of mice 24 h after IM and the geometric center (GC) of FITC-dextran distribution in each experiment group. n = 4 in the Sham group and n = 4 in the IM group. C The protein expression in the intestinal muscle layer of each group of mice was quantified by WB. D Representative images of HE staining in the intestinal muscle layer of POI mice. E UMAP embedding plot for cells from CD45+ immune cells in Sham and IM intestine muscularis derived from in GSE167465 dataset. F The bar chart compares the percentages of neutrophils, infiltrating macrophages, and resident macrophages in the intestinal muscle layer before and after intestinal manipulation. G The violin plot illustrates the expression of classical markers for each cell type in the intestinal muscle layer. H Differential gene analysis was performed on infiltrating macrophages. In the volcano plot, the red and blue dots represent upregulated and downregulated genes in the IM group, respectively. I Enrichment pathway analysis was performed on the upregulated differentially expressed genes (DEGs) in the volcano plot (IM vs. Sham), revealing several characteristic pathways. All data are represented as means ± SEM. All experiments were repeated at a minimum of three times. Statistical significance was determined by unpaired two-tailed Student’s t test: *P ≤ 0.05 and ****P ≤ 0.0001.

We isolated the intestinal muscularis layer and measured protein expression levels. Western blot (WB) analysis revealed a significant upregulation of inflammatory markers in the small intestinal muscularis following surgical manipulation (Fig. 1C). Hematoxylin and eosin (HE) staining of the small intestine showed congestion, edema, and inflammatory cell infiltration in the muscularis layer after intestinal manipulation (Fig. 1D). Immunohistochemistry (IHC) for MPO (Supplementary Fig. 1A) and CD86 (Supplementary Fig. 1B) further demonstrated extensive infiltration of neutrophils and macrophages in the muscularis layer post-manipulation.

Single-cell sequencing data revealed extensive macrophage infiltration characterized by a pro-inflammatory phenotype

To further clarify the type and phenotype of inflammatory infiltration induced by intestinal manipulation, we analyzed single-cell sequencing data of CD45+ cells from the intestinal muscularis layer in postoperative ileus models available in public databases. The immune cell types in the muscularis were predominantly composed of resident macrophages, infiltrating macrophages, dendritic cells (DCs), neutrophils, and lymphocytes, with cell type identification determined by differential expression of established markers (Fig. 1E, G). In terms of cell proportion changes, intestinal manipulation resulted in a significant increase in infiltrating macrophages and neutrophils (Fig. 1F), consistent with the immunohistochemical findings mentioned earlier. To further investigate the phenotypic changes in infiltrating macrophages after intestinal manipulation, we performed differential gene expression analysis of these cells before and after manipulation (Fig. 1H). Gene Ontology Biological Process (GOBP) enrichment analysis of the upregulated genes revealed that infiltrating macrophages exhibited an inflammatory and chemotactic phenotype post-manipulation, suggesting their role in exacerbating muscularis inflammation (Fig. 1I).

Transcriptomic and single-cell sequencing analyses identified macrophages with a highly glycolytic phenotype in the intestinal muscularis layer

Considering recent studies highlighting the close relationship between metabolic reprogramming and immune cell phenotypes, particularly the role of glycolysis in driving macrophage polarization, and given that macrophages are the predominant infiltrating cells in the intestinal muscularis following intestinal manipulation, we analyzed glycolytic phenotypes in the muscularis using transcriptomic and single-cell sequencing data. Bulk RNA sequencing revealed that intestinal manipulation significantly upregulated the expression of glycolytic enzymes, including Slc2a1, Hk2, Pkm, and Ldha (Fig. 2A). Differential pathway analysis showed significant enrichment of glycolysis and inflammatory activation pathways in IM group (Fig. 2B–E).

Fig. 2. IM induces enhanced glycolysis in muscle layer macrophages and upregulation of STING expression.

A Heatmap of mRNA levels of glycolysis genes (Slc2a1, Hk2, Pkm, and Ldha) in the intestinal muscle layer following intestinal manipulation. B–E Gene set enrichment analysis (GSEA) was performed to analyze the transcriptional features of the intestinal muscle layer before and after surgical manipulation. F Ucell cell scoring based on the glycolysis gene set was performed on immune cells in the intestinal muscle layer before and after IM using single-cell sequencing data. G Infiltrated macrophages were subjected to dimensionality reduction and clustering based on metabolic gene sets, and were classified into two metabolic states: State1 and State2. H The bar chart compares the percentages of infiltrating macrophages in the intestinal muscularis before and after intestinal manipulation in two metabolic states, State1 and State2. I Ucell cell scores were calculated for the infiltrating macrophages of the two metabolic states based on the glycolysis gene set and the TCA cycle gene set. J The dot plot displays the expression of Sting and Irf3 in infiltrating macrophages of State1 and State2. K The bar chart illustrates the top ten GOBP pathways enriched in infiltrating macrophages of State2. The size of the dot corresponds to the proportion of cells expressing each transcript within the group, while the color of the dot corresponds to the level of expression. L Expression levels of key genes in Response to Lipopolysaccharide pathway in infiltrating macrophages of State1 and State2. M The level of LPS in the intestinal muscularis before and after intestinal ileus (IM) as determined by ELISA. N The dot plot displays the expression of Sting, Irf3, Slc2a1, Eno1, Pkm, and Ldha in infiltrating macrophages of State1 and State2. O The expression level of STING in the muscular layer of the intestine before and after intestinal manipulation was determined by IHC. All data are represented as means ± SEM. All experiments were repeated at a minimum of three times. Statistical significance was determined by unpaired two-tailed Student’s t test: **P ≤ 0.01, ***P ≤ 0.001,****P ≤ 0.0001. ns not significant (P > 0.05).

To further determine cell-type-specific differences in enhanced glycolysis within the muscularis, UCell scoring was applied to single-cell sequencing data using a glycolysis gene set. Results indicated that resident and infiltrating macrophages exhibited the highest glycolysis scores among muscularis immune cells, with infiltrating macrophages in the intestinal manipulation group showing the most pronounced increase in glycolysis scores (Fig. 2F).

We further classified infiltrating macrophages into distinct metabolic states (State1 and State2) using metabolic gene sets from the GSEA database (Fig. 2G). Stacked bar plot analysis revealed that the proportion of State2 macrophages significantly increased after intestinal manipulation (Fig. 2H). UCell scoring showed that State2 macrophages had significantly higher glycolysis scores compared to State1 macrophages, with no significant differences in TCA cycle activity (Fig. 2I). Compared with State1 macrophages, State2 macrophages exhibit high expression of glycolytic genes, including Slc2a1, Eno1, Pkm, and Ldha (Fig. 2J). Differential gene expression analysis between State1 and State2 macrophages, followed by GOBP enrichment analysis, identified significant enrichment of pathways related to leukocyte chemotaxis (Fig. 2K) and response to lipopolysaccharide in State2 macrophages (Fig. 2L). These findings suggest that State2 macrophages, characterized by heightened glycolytic activity, play a role in exacerbating muscularis inflammation. Consistent with the single-cell sequencing data, the quantitative level of LPS in the intestinal muscularis was significantly higher in the IM group than in the Sham group, suggesting a substantial upregulation of LPS in the intestinal muscularis during ileus (Fig. 2M). Based on these findings, we used LPS to stimulate macrophages in subsequent cellular experiments to mimic the polarization phenotype of macrophages in the intestinal muscularis of mice with intestinal ileus.

Increased STING expression in macrophages of the intestinal muscularis layer

To elucidate the role of the STING-IRF3 signaling pathway in surgery-induced POI, we examined STING expression differences in single-cell sequencing data. The IM group showed a significant increase in State2 macrophages, which exhibited markedly higher STING expression compared to State1 macrophages (Fig. 2N). Immunohistochemistry staining further confirmed that the elevated STING expression in the intestinal muscularis layer(Fig. 2O).

Knocking out STING in vitro reduces the expression of glycolytic enzymes and ROS generation in LPS-stimulated RAW cells

Upon LPS stimulation, the expression of glycolytic enzymes GLUT1, HK2, PKM2, and LDHA was significantly increased in RAW264.7 cells. To investigate whether reduced STING expression affects the expression of these glycolytic enzymes, we generated STING knockout (STING KO) macrophages using CRISPR-Cas9 technology. The absence of STING led to a significant reduction in LPS-induced IRF3 phosphorylation (Fig. 3A) and expression of glycolytic enzymes (Fig. 3B). Given potential differences in transcriptional and metabolic profiles between BMDMs and RAW cells, we also examined STING-mediated regulation of glycolytic enzymes in BMDMs, and found that glycolytic enzyme levels were significantly lower in STING-knockout BMDMs than in wild-type controls (Supplementary Fig. 2A). We further analyzed energy metabolism-related metabolites via targeted metabolomics and found that glycolytic intermediates were generally lower in LPS-stimulated macrophages with STING knockout than in wild-type controls, among which the levels of glucose-6-phosphate(G6P), fructose-1,6-bisphosphate (FBP), phosphoenolpyruvate (PA), and lactate (LA) were significantly decreased (Fig. 3C–G). Similarly, extracellular acidification rate (ECAR) measurements revealed that silencing STING in vitro markedly reduced glycolysis levels (Fig. 3H, I). These findings indicate that STING knockout suppresses LPS-induced expression of glycolytic enzymes (GLUT1, HK2, PKM2, and LDHA) and blocks the glycolytic burst. This suggests that STING may serve as a potential target for modulating macrophage-mediated inflammatory activation in the intestinal muscularis. However, the mechanisms by which STING influences glycolysis warrant further investigation.

Fig. 3. Knockout of STING reduces the expression of glycolytic enzymes and the rate of glycolysis in LPS-induced RAW cells.

A Western blot (WB) analysis was performed for STING-IRF3 pathway proteins (pIRF3, IRF3, pTBK1,TBK1, STING) in RAW macrophages before and after LPS stimulation following the knockout of the STING gene. B Western blot (WB) analysis was performed for glycolytic enzymes (HIF1α, GLUT1, HK2, PKM2, and LDHA) in RAW macrophages before and after LPS stimulation following the knockout of the STING gene. C Targeted metabolomic analysis of energy metabolites in WT and STING knockout (KO) macrophages. D–G Quantification of key glycolytic metabolites: G1P + G6P + F6P, FBP, PA, and LA in macrophages. H, I Following LPS stimulation and STING knockout, the extracellular acidification rate (ECAR) was measured using the XF96 Seahorse analyzer, and glycolytic parameters were quantified. J Representative images of the culture medium of LPS-stimulated RAW264.7 cells after knockout of the STING gene. K Lactate levels in the supernatant of RAW264.7 macrophage cultures after LPS stimulation (n = 3 per group). All data are represented as means ± SEM. All experiments were repeated at a minimum of three times. Statistical significance was determined by unpaired two-tailed Student’s t test in (D) and ANOVA test in (A, B, I, K): *P ≤ 0.05, **P ≤ 0.01,***P ≤ 0.001,****P ≤ 0.0001. ns not significant (P > 0.05).

To better understand how STING regulates glycolysis in LPS-stimulated RAW macrophages, we assessed ROS production in STING KO macrophages using flow cytometry. As expected, LPS stimulation increased ROS generation in WT macrophages, while ROS levels remained significantly lower in STING KO cells (Supplementary Fig 3A, B). Given recent evidence that ROS can modulate HIF1α activity, and that HIF1α drives the expression of glycolytic enzymes, we hypothesize that STING enhances glycolysis by promoting ROS production, which in turn stabilizes HIF1α, in LPS-stimulated macrophages.

STING-driven lactate accumulation promotes H4K8 lactylation at the Hk2 gene promoter

During glycolysis rate measurements, we observed a rapid yellowing of the culture medium in LPS-stimulated macrophages, potentially due to significant lactate production (Fig. 3J). To confirm this, we measured lactate concentrations in the medium and found that LPS stimulation markedly increased lactate production in WT RAW cells, whereas this effect was attenuated in STING KO macrophages (Fig. 3K). Given the regulatory role of histone lactylation on gene expression, we quantified histone lactylation modifications in LPS-stimulated macrophages. The results showed a significant increase in pan-lysine lactylation (Pan Kla) in LPS-stimulated RAW cells. In contrast, Pan Kla was significantly reduced in STING KO RAW cells (Fig. 4A). LPS induces increased lactylation at multiple histone loci, including H3K9la, H3K18la, H4K8la, and H4K16la; notably, the H4K8la locus exhibits the most significant reduction upon STING knockout (Fig. 4B). In POI mice, we observed that the level of histone lactylation in intestinal muscularis macrophages was significantly lower in STING KO mice compared to WT mice (Supplementary Fig. 5).To investigate the downstream genes regulated by STING-driven histone lactylation, we performed CUT&Tag-seq using an H4K8la antibody in LPS-stimulated macrophages. Representative genome browser tracks from CUT&Tag-seq revealed significant enrichment of H4K8la at the Hk2 gene locus in LPS-stimulated RAW264.7 cells compared to the control group. This enrichment was notably diminished in STING KO cells (Fig. 4C). To validate the CUT&Tag-seq data, we designed three primer sets targeting the binding sites at the Hk2 promoter and conducted CUT-PCR experiments on RAW cell lysates. The results confirmed a significant increase in H4K8la abundance at the Hk2 promoter (Fig. 4D), further supporting the interaction between H4K8la and the Hk2 promoter region.

Fig. 4. Knockout of STING reduces histone lactylation levels and chromatin accessibility at Hk2 promotor region.

A Representative WB images showing the protein levels of Pan-Kla on WT and STING KO macrophages stimulated with LPS. B Representative WB images showing the protein levels of, H3K9la、H3K18la、H4K8la and H4K16la on WT and STING KO macrophages stimulated with LPS. C CUT&Tag tracks of H4K8la and ATAC-seq tracks at the HK2 loci in LPS-stimulated WT RAW and STING KO RAW macrophages. D CUT-PCR validated the results of CUT&Tag-seq. E The heatmap illustrates the normalized ATAC-seq signal intensity, representing chromatin accessibility from the transcription start site (TSS) to ±5 kb in macrophages. F The heatmap represents the normalized CUT&Tag intensity of H4K8la binding to DNA from the transcription start site (TSS) to the ±5 kb in macrophages. G, H The genome-wide distribution of differentiated H4K8la-binding peaks was analyzed between WT and STING KO macrophages. All data are represented as means ± SEM. All experiments were repeated at a minimum of three times. Statistical significance was determined by an unpaired two-tailed Student’s t test in (D) and ANOVA test in (A, B): *P ≤ 0.05, **P ≤ 0.01,***P ≤ 0.001,****P ≤ 0.0001. ns not significant (P > 0.05).

Our findings demonstrate that in LPS-stimulated RAW cells, STING-mediated H4K8la epigenetically regulates Hk2 transcription, establishing a direct link between STING-driven histone lactylation and glycolytic gene activation. The ATAC-seq data analysis revealed a widespread and significant decrease in chromatin accessibility at promoter regions following STING KO (Fig. 4E). Specifically, the binding peaks at the promoter region of the HK2 gene were markedly reduced (Fig. 4C), which may be associated with the H4K8la level at the HK2 gene promoter. Furthermore, the H4K8la binding peaks in the STING KO group were significantly diminished (Fig. 4F), with nearly 20% of these binding peaks located at promoter regions (Fig. 4G, H). These findings suggest that the STING signaling pathway may regulate the expression of Hk2 by modulating chromatin accessibility and histone lactylation modifications, thereby playing a crucial role in Glycolysis.

Enhanced glycolysis and histone lactylation regulate HK2 expression via IRF3

Considering that phosphorylated IRF3 translocates to the nucleus to regulate gene transcription, and that LPS stimulation promotes IRF3 phosphorylation via STING and TRIF3, we hypothesized that enhanced glycolysis and histone lactylation might further regulate glycolytic gene expression at the transcriptional level through phosphorylated IRF3 (pIRF3). To investigate whether IRF3 phosphorylation contributes to LPS-induced glycolytic enhancement, we utilized lentivirus-mediated knockdown of IRF3 in RAW264.7 cells. Western blot (WB) analysis showed reduced expression of HK2 in IRF3-knockdown cells (LV-IRF3) (Fig. 5A). These findings were further confirmed by qRT-PCR (Fig. 5B), indicating that IRF3 activation is likely associated with LPS-induced glycolytic enhancement. This conclusion was supported by Seahorse assays (Supplementary Fig. 5), which demonstrated a significant reduction in glycolytic activity in IRF3-knockdown cells.

Fig. 5. IRF3 binds to the promoter regions of HK2, promoting gene expression.

A Protein levels of HK2 in shIRF3 RAW and WT RAW cells after LPS stimulation. B mRNA expression levels of HK2 in shIRF3 RAW and WT RAW cells after LPS stimulation. C Representative CUT&Tag-seq genome browser tracks showing IRF3-regulated gene loci, including Hk2. D CUT-PCR validated the results of CUT&Tag-seq. E The heatmap illustrates the normalized CUT&Tag intensity of IRF3 binding to DNA from the transcription start site (TSS) to ±5 kb in macrophages. F The genome-wide distribution of differentiated IRF3-binding peaks was analyzed between WT and STING KO macrophages. G A bar plot of KEGG pathway enrichment based on the elevated IRF3-binding peaks versus the IgG group.All data are represented as means ± SEM. All experiments were repeated at a minimum of three times. Statistical significance was determined by an unpaired two-tailed Student’s t test in (D) and ANOVA test in (A, B): *P ≤ 0.05, **P ≤ 0.01, ****P ≤ 0.0001. ns not significant (P > 0.05).

To test this hypothesis, we first analyzed publicly available ChIP-seq data using an anti-pIRF3 antibody. The results indicated that pIRF3 interacts with DNA fragments in the Hk2 promoter region (Fig. 5C). To further confirm this interaction, we performed CUT&Tag-seq on LPS-stimulated RAW macrophages. Representative genome browser tracks from CUT&Tag-seq revealed significant enrichment of pIRF3 binding at the Hk2 locus (Fig. 5C), with signal intensities markedly higher than those in the control group (IgG). To validate the CUT&Tag-seq data, we designed three primer sets targeting the binding sequences and performed CUT-PCR experiments using RAW cell lysates. The results confirmed a strong interaction between pIRF3 and the Hk2 promoter region (Fig. 5D).

The CUT&Tag analysis revealed a significant increase in IRF3 binding peaks compared to the IgG control group, particularly in the vicinity of the transcription start site (TSS) (Fig. 5E). Approximately 30% of the IRF3 binding regions were localized to promoter regions (Fig. 5F). Among the target genes exhibiting differential IRF3 binding at their promoters, metabolic pathways were significantly upregulated (Fig. 5G). These findings demonstrate that pIRF3 directly regulates Hk2 transcription by binding to its promoter region, linking enhanced glycolysis and histone lactylation to transcriptional regulation mediated by IRF3 phosphorylation.

STING knockout restricts LPS-induced M1 polarization in RAW264.7 cells

As previously mentioned, STING knockout restricts glycolysis, and the rapid glycolytic burst provides the energy required for macrophage polarization. To elucidate the role of glycolysis in macrophage polarization, we found that treatment with the glucose analog 2-deoxy-D-glucose (2DG) and the HIF1α inhibitor PX478 attenuated LPS-induced M1 polarization of macrophages. This effect was confirmed by flow cytometry, which showed reduced expression of iNOS (Supplementary Fig 6). To investigate the downstream effects of STING on glycolysis in RAW264.7 cells, we employed RNA-seq, flow cytometry and ELISA to analyze LPS-induced STING KO macrophages (Fig. 6A, B). Flow cytometry revealed that LPS stimulation significantly upregulated the expression of iNOS (M1 marker) and CD206 (M2 marker) in WT RAW264.7 cells (Fig. 6C–E). Additionally, ELISA demonstrated that LPS-induced WT RAW264.7 cells secreted significantly higher levels of inflammatory cytokines, including IL-1β, IL-6, and TNF-α, into the supernatant. In contrast, silencing STING reduced both the expression of macrophage polarization markers and the secretion of inflammatory cytokines (Fig. 6F–H). These findings suggest that STING plays a critical role in regulating macrophage polarization and the production of inflammatory mediators through its effects on glycolysis.

Fig. 6. Knockout of STING suppresses LPS-induced M1 polarization of RAW cells.

A Schematic diagram of samples and grouping for RNA sequencing. B Heatmap of mRNA levels of M1 marker genes (Il1b, Il6, Tnf, Cd86, iNOS), M2 marker genes (Mrc1) and Sting in RAW264.7 cells r following LPS-stimulation. C Following the knockout of the STING gene, flow cytometry was used to analyze and quantify the M1 (iNOS) (D)and M2 (CD206) (E) polarization phenotypes in LPS-stimulated macrophages. F–H ELISA analysis of IL-1β, IL-6, and TNF-α in the supernatant of culture medium from LPS-stimulated macrophages after STING gene knockout. All data are represented as means ± SEM. All experiments were repeated at a minimum of three times. Statistical significance was determined by an unpaired two-tailed Student’s t test and ANOVA test: *P ≤ 0.05, **P ≤ 0.01,***P ≤ 0.001,****P ≤ 0.0001. ns not significant (P > 0.05).

STING knockout alleviates intestinal muscularis inflammation and motility impairment in POI mice

To investigate the relationship between STING, glycolysis, and macrophage polarization in vivo, we utilized a STING knockout (STING KO) mouse model. We isolated the intestinal muscularis from POI mice and performed Western blot (WB) analysis. We found that the macrophage M1 polarization marker (iNOS) was significantly upregulated in wild-type (WT) mice, but markedly decreased in STING knockout (KO) mice (Supplementary Fig 7A). Bulk transcriptomic analysis of the intestinal muscularis revealed that STING knockout significantly reduced the expression of glycolytic enzymes, including Glut1(Slc2a1), Hk2, Pkm, and Ldha (Fig. 7A, B). Gene Set Enrichment Analysis revealed that both glycolysis-related and inflammation-associated pathways were significantly suppressed in the STING KO IM group (Fig. 7C–F). Flow cytometry demonstrated that STING knockout decreased macrophage infiltration in the intestinal muscularis (Fig. 7G) while reducing the proportion of M1-polarized macrophages and increasing the proportion of M2-polarized macrophages (Fig. 7H, I, Supplementary Fig7D).Meanwhile, STING KO reduced neutrophil infiltration in the intestinal muscularis (Supplementary Fig 7B, C), indicating that STING KO alleviates post-operative inflammation in the intestinal muscularis. Furthermore, the distribution of FITC-dextran in the gastrointestinal tract indicated that STING KO ameliorated intestinal motility impairment caused by intestinal manipulation (Supplementary Fig. 7E, F). To validate the potential of STING as a clinical therapeutic target for intestinal ileus, we intraperitoneally injected the STING inhibitor C176 into WT POI mice. Results showed that C176 reduced M1 polarization of intestinal muscularis macrophages and alleviated intestinal motility impairment (Supplementary Fig. 7E, F). To clarify the critical role of STING-mediated HK2 regulation in POI mice, we overexpressed HK2 in intestinal muscularis macrophages of STING knockout (KO) mice via tail vein injection of AAV-HK2. We found that reversing STING KO-induced HK2 downregulation partially restored M1 polarization in the intestinal muscularis (Fig. 7H–J), and exacerbated intestinal motility impairment (Supplementary Fig. 7E, F)

Fig. 7. Knockout of the STING gene alleviates intestinal muscularis inflammation and motility impairment in POI mice.

A Schematic diagram of samples and grouping for RNA sequencing. B Heatmap of glycolytic gene mRNA levels (Slc2a1, Hk2, Pkm, and Ldha) in the small intestinal muscular layer of WT and STING KO mice after intestinal manipulation. C–F Gene set enrichment analysis (GSEA) was performed to analyze the transcriptional characteristics of the intestinal muscular layer in WT and STING KO mice before and after IM. G Flow cytometry was used to analyze and quantify the infiltration of macrophages in the intestinal muscular layer of WT and STING KO mice following IM. H Flow cytometry was used to analyze the M1 and M2 polarization phenotypes of macrophages in the intestinal muscular layer of WT IM, STING KO IM, WT IM + C176, and STING KO IM + AAV-HK2. I Quantification of flow cytometry data. J Representative WB images showing the expression of HK2, CD86, and IL-1β in STING KO mice subjected to IM following tail vein injection of AAV-HK2. All data are represented as means ± SEM. All experiments were repeated at a minimum of three times. Statistical significance was determined by an unpaired two-tailed Student’s t test in (G) and ANOVA test in (I, J): *P ≤ 0.05, **P ≤ 0.01,***P ≤ 0.001,****P ≤ 0.0001. ns not significant (P > 0.05).

Discussion

POI progresses in two phases. The first phase begins during surgery and concludes shortly after. This phase is neurogenically mediated, involving alterations in parasympathetic and sympathetic nervous system activity. The second phase begins 3–4 h post-surgery and is driven by inflammation. The late inflammatory phase of POI lasts longer than the early neurogenic phase, offering a potential window for intervention. Previous studies on the inflammatory microenvironment of the intestinal muscularis have primarily focused on inflammatory response regulation, neglecting metabolism as a critical driving factor. With advances in immunometabolic research, the interplay between metabolism and immunity has become increasingly evident. Targeting metabolic pathways to regulate inflammation presents a promising therapeutic strategy. For example, controlling glucose levels to limit macrophage M1 polarization has been shown to accelerate wound healing in diabetic ulcers^20,21. Supplementation with tryptophan has been reported to reverse cGAS-STING activation²². Thus, we conducted the first investigation into the immunometabolism of macrophages within the intestinal muscularis inflammatory microenvironment.

The intestinal wall comprises the mucosa, lamina propria, and muscularis layers. While the mucosa contains numerous immune cells, our understanding of muscularis macrophages remains limited compared to those in the mucosa and submucosa. Notably, the pathological site of POI is primarily within the intestinal muscularis. Isolating cells from the muscularis layer presents significant challenges. Building upon the method described by Vilz et al.²³, we developed an improved protocol to isolate the muscularis layer effectively. This approach enables subsequent analyses, such as protein quantification and transcriptome sequencing, while minimizing mucosal layer contamination. The resulting data offer a more accurate representation of the inflammatory microenvironment in the muscularis.

Using single-cell sequencing data, we analyzed the immune cell composition in the muscularis layer in the muscularis layer of a POI model. The analysis revealed significant infiltration of macrophages and neutrophils, with macrophages being the predominant immune cell population. Gene enrichment analyses and UCell-based gene set scoring indicated that macrophages in the POI group exhibited significantly elevated glycolysis scores. This change is closely associated with the enhanced inflammatory response and the M1 polarization phenotype of macrophages, which may exacerbate the inflammatory microenvironment in the muscularis.

Recognizing the metabolic plasticity of macrophages, we aimed to comprehensively characterize their metabolic features. Following the methodology of Xiaoxiang Dong et al.²⁴, we extracted 1486 genes associated with metabolic pathways from the MsigDB database. These genes were utilized for principal component analysis (PCA) to perform dimensionality reduction and clustering. This analysis stratified macrophages into two distinct metabolic states, designated as State1 and State2. Macrophages in State2 exhibited markedly higher glycolysis scores but relatively low TCA cycle scores. Pathway enrichment analysis further revealed that State2 macrophages were characterized by both enhanced glycolysis and a pronounced pro-inflammatory phenotype, highlighting their potential role in exacerbating the inflammatory microenvironment.

Gene expression regulation depends on the chromatin accessibility of corresponding promoter regions. Modulating the chromatin accessibility of target genes is a key therapeutic strategy for certain diseases. For example, nanoparticles delivering the histone demethylase JMJD3 have been shown to inhibit histone methylation at the STING promoter region, thereby reducing chromatin accessibility and mitigating persistent inflammation in diabetic wounds²⁵. Protein lactylation is a novel post-translational modification (PTM) identified in 2019, where lactyl groups covalently attach to proteins. This modification represents a chemical interaction between lactate and proteins. As an epigenetic modification, histone lactylation directly regulates gene expression²⁶.

In this study, ATAC-Seq data revealed increased chromatin accessibility in LPS-stimulated macrophages. However, following STING KO, ATAC-Seq analysis indicated reduced chromatin accessibility at glycolysis-related gene HK2. During cell culture, we observed significantly less yellowing in the culture supernatant of LPS-stimulated STING KO cells compared to WT cells. This finding suggests that STING may regulate macrophage metabolism during glycolysis through lactate production. Lactate concentration measurements confirmed that lactate levels in the KO group were markedly lower than those in the WT group. Moreover, STING KO inhibited the LPS-induced increases in histone lactylation. These findings suggest that STING may intervene in histone lactylation through a certain pathway.

Chromatin accessibility is regulated by histone lactylation, and recent studies have utilized CUT&Tag-Seq to identify genes modulated by lactylation modifications^27,28. Using CUT&Tag-Seq, we identified HK2 as target genes regulated by H4K8la. Genome fragment sequencing results revealed significant enrichment of H4K8la-bound fragments at the promoter regions of HK2. The increased chromatin accessibility facilitates the binding of LPS-induced transcription factor IRF3 to the HK2 promoter, thereby promoting its expression.

In intestinal epithelial cells, 10-carboxymethyl-9-acridanone (CMA) activates the STING pathway, enhancing the antimicrobial function of epithelial cells while concurrently increasing the expression of HK2 and LDHA. However, the study did not elucidate the regulatory mechanism linking STING to the upregulation of HK2 and LDHA¹⁶. This gap in understanding STING’s role in metabolic regulation is particularly significant, as metabolic reprogramming plays a central role in immune responses. Metabolic reprogramming in macrophages stimulated by LPS and IFNγ is highly dynamic. During the early stages, significant accumulation of succinate and itaconate occurs, which strongly promotes the stabilization of HIF1α²⁹. Numerous studies have emphasized the critical role of HIF1α in initiating downstream inflammatory programs through the regulation of glycolysis^13,30,31. Additionally, research has demonstrated the involvement of HIF1α in STING-mediated glycolysis regulation. cGAMP activation of STING in dendritic cells is accompanied by enhanced glycolysis. It was observed that HIF1α is regulated by both STING and PKM. Despite these findings, the precise mechanism by which STING regulates HIF1α remains unclear¹⁷. In the present study, we determined that STING stabilizes HIF1α by maintaining intracellular ROS levels.

A study has provided an epigenetic perspective on how STING regulates glycolysis through its downstream transcription factor IRF3. In a renal fibrosis model, activation of the STING-IRF3 pathway was shown to enhance glycolysis. The authors utilized ChIP-PCR to demonstrate that IRF3 binds to the promoter region of phosphofructokinase-2/fructose-2,6-bisphosphatase 3 (PFKFB3), promoting its expression and subsequently driving glycolysis. ChIP-seq and CUT&Tag-seq are powerful tools for identifying transcription factor binding sites and recognizing motif regions, offering valuable insights into transcriptional regulation^28,32. In this study, we integrated ChIP-seq data with CUT&Tag-seq to reveal that IRF3 binds to the promoter regions of the HK2 gene in macrophages. To validate these findings, we constructed IRF3 knockdown RAW264.7 cells using lentivirus-mediated gene silencing^18,32. In this study, LPS was used to stimulate macrophages to induce M1 polarization. Building on the stabilization of HIF1α by STING, LPS stimulation further enhances glycolysis in macrophages, resulting in substantial lactate production. The accumulation of lactate increases histone lactylation at the promoter region of the HK2 gene. Concurrently, LPS-induced phosphorylated IRF3 binds to DNA fragments at the HK2 promoter region, thereby promoting HK2 expression and driving glycolysis.

Macrophage STING has emerged as a potential therapeutic target for disease intervention, with several innovative strategies validated in preclinical research. For instance, a macrophage-targeted cell therapy involves the subcutaneous injection of STING KO bone marrow-derived macrophages (BMDMs) near wound sites. This approach induces macrophage M2 polarization, promoting angiogenesis and matrix deposition, thereby accelerating diabetic wound (DW) healing²⁰. In another approach, liposomes carrying small interfering RNA (siRNA) have been utilized to suppress the expression of upstream cGAS at the transcriptional level, thereby reducing macrophage pyroptosis³³.

STING serves as both a regulator of metabolism and a target of metabolic intermediates, presenting promising opportunities for modulating the STING pathway through metabolic interventions. For instance, 4-octyl itaconate (4-OI), an analog of the TCA cycle-derived metabolite itaconate, has been shown to attenuate excessive STING activation. Mechanistically, 4-OI enhances Nrf2 expression, thereby influencing the stability of STING mRNA³⁴. On the other hand, 4-OI limits STING activation by promoting its alkylation³⁵. Lipid metabolic pathways can also regulate STING activity. For instance, oxidized low-density lipoprotein (OxLDL) activates the STING pathway in macrophages, driving sustained inflammation³⁶. Conversely, the LXR-induced protein SMPDL3A degrades 2′3′-cGAMP, thereby limiting the activation of the cGAS-STING signaling pathway³⁷. Based on these findings, future strategies may leverage endogenous and exogenous metabolites to modulate excessive STING activation in pathological conditions.

This study has several limitations. Although this study validated the role of STING in POI using a STING knockout mouse model, we believe that future research should focus on developing more precise macrophage-specific STING-targeted therapeutic strategies, such as gene editing-based cell therapies or siRNA delivery systems, to enhance therapeutic efficacy and minimize side effects.

Conclusion

Our study demonstrates that STING plays a critical role in the pathogenesis of POI by regulating glycolysis in macrophages and promoting M1 polarization. From the perspective of downstream mechanisms, STING stabilizes HIF1α under inflammatory conditions by maintaining intracellular ROS levels, thereby sustaining LPS-induced glycolytic enhancement and providing a rapid energy supply for macrophage polarization.

From the perspective of histone modifications, STING-driven glycolysis generates increased lactate, which enhances histone lactylation at the promoter regions of HK2, thereby increasing chromatin accessibility. Additionally, from an epigenetic standpoint, we identified HK2 as a novel regulatory target of IRF3. Under STING activation, IRF3 binds to the promoter regions of these genes, further promoting their expression.

In summary, our study highlights the mechanisms by which STING regulates macrophage polarization through the glycolytic pathway and suggests that targeting macrophage STING to disrupt this positive feedback loop, thereby inhibiting the rapid glycolytic burst and cutting off the energy supply, represents a promising therapeutic strategy for inflammation.

Material and methods

Animal study

All experimental animals were fasted for 24 h prior to modeling. The surgical procedure was performed as follows^23,38: after anesthesia, the abdominal skin was sterilized with alcohol, and a midline incision was made. The entire small intestine was gently manipulated using saline-moistened cotton swabs to push the intestinal contents from the small intestine to the cecum. This procedure was repeated three times within 15 min, after which the small intestine was returned to the abdominal cavity, and the incision was sutured. One day post-surgery, the small intestine was harvested, and the muscularis layer was isolated for subsequent experiments. In the Sham group, only the abdominal incision was made and sutured without intestinal manipulation.The animals were divided into four groups: WT Sham, WT IM, STING^−/− Sham, and STING^−/− IM. The STING^−/− mice were on a C57BL/6 background, and both STING knockout and wild-type C57BL/6 mice were purchased from Jiangsu Gempharmatech Biotechnology Co., Ltd. The mice were male, aged 6–8 weeks, and weighed 20 ± 2 g. All mice were housed in the specific pathogen-free (SPF) Animal Center at Nanjing Drum Tower Hospital, the Affiliated Hospital of Nanjing University Medical School. The homozygosity of STING−/− mice was confirmed by PCR prior to their use in experiments. We have complied with all relevant ethical regulations for animal use. All experiments were conducted in accordance with the guidelines approved by the Animal Ethics Committee of Nanjing Drum Tower Hospital (Approval Number: 202312209).

Cell culture

RAW264.7 WT (rawl, Invivogen, USA) and RAW264.7 STING KO cells (rawl-kostg, Invivogen, USA) were cultured in DMEM supplemented with 10% fetal bovine serum (FBS) and 1% penicillin-streptomycin in a humidified incubator at 37 °C with 5% CO₂. M1 polarization was induced by treating RAW264.7 cells with LPS (1 μg/mL, Escherichia coli O111:B4, Sigma-Aldrich, USA). The cell culture supernatant was collected for the measurement of inflammatory cytokines and lactate secretion. The detection of inflammatory cytokines in cell supernatants was performed using the XMplex Custom Panel kit and the XMplex-100 multi-parameter flow cytometer (XM BIOTECH). Lactate levels were measured using the Lactic Acid (LA) Content Assay Kit (BC2235, Solarbio), following the manufacturer’s instructions.

Detection of intracellular ROS

The intracellular reactive oxygen species (ROS) concentration was measured using the ROS Detection Kit (Green Fluorescence) (abs580232, Absin). Treated cells were incubated with an appropriate volume of diluted DCFH-DA working solution, ensuring the cells were fully covered. The cells were washed 1–2 times with serum-free culture medium to remove any unincorporated DCFH-DA. ROS levels were determined by flow cytometry using the FL1 channel, with excitation at 488 nm and emission measurement at 530 nm.

Lentivirus-mediated knockdown of IRF3 gene expression in RAW264.7 cells

RAW264.7 cells were transduced with lentivirus-mediated Irf3 knockdown (lenti-Irf3-shRNA) purchased from Shanghai GeneChem Co., Ltd. The cells were infected for 48 h. The shRNA targeting Irf3 consisted of the following sequences: sense strand, 5′-GATCCGAGTTAGTTTGACAGCTAATTCAAGAGATTAGCTGT CAAACTAACTCTTTTTTG-3′, and antisense strand, 5′-AATTCAAAAAAGAGTT AGTTTGACAGCTAATCTCTTGAATTAGCTGTCAAACTAACTCG-3′. Stable cell lines were selected and maintained using puromycin (5 μg/mL). The resulting stable RAW264.7 shIRF3 cell line was used for subsequent experiments.

Bulk-RNA sequencing and analysis

Total RNA from RAW264.7 cells and intestinal muscularis was extracted using the TRIzol method, with or without LPS treatment. Library preparation was performed using the KAPA Stranded RNA-Seq Library Preparation Kit (Illumina, San Diego, CA, USA). Sequencing was conducted on the Illumina HiSeq 4000 platform, and RNA library quality was assessed using the Agilent 2100 Bioanalyzer (Agilent Technologies, Santa Clara, CA, USA). Differentially expressed genes (DEGs) were identified based on |log2FC| ≥ 0.5 and p-values and q-values ≤ 0.05. Following differential expression analysis, all genes were ranked based on log2FC values and subsequently subjected to Gene Set Enrichment Analysis (GSEA) to identify significantly enriched gene sets.

Bulk RNA seq analysis of publicly available datasets

We downloaded the expression matrix of bulk RNA sequencing data from the mouse intestinal muscularis (GSE134942) post-intestinal manipulation from the NCBI Gene Expression Omnibus (GEO) database³⁹. Differential analysis and enrichment analysis were performed according to the methods described in the previous section³⁹.

Single-cell sequencing data analysis

Single-cell sequencing data from a postoperative ileus mouse model were obtained from the GEO database (Accession ID: GSE167465)³⁸. Cell types were identified using classical cell markers. Gene set scoring for each cell type was performed using the R package UCell. For metabolic state analysis of single cells²⁴, 1486 metabolism-related genes were extracted from the Molecular Signatures Database (http://www.broad.mit.edu/gsea/msigdb/msigdb_index.html). These genes were used for dimensionality reduction and clustering analysis of infiltrating macrophages. Principal component analysis (PCA) was conducted using the metabolic genes, and the top five principal components were applied in UMAP dimensionality reduction and Seurat-based clustering using the k-nearest neighbor method. The parameter k was set to 100.

Western Blot analysis

Small intestinal muscularis tissues and RAW264.7 cells were collected, and total protein was extracted using RIPA lysis buffer containing protease and phosphatase inhibitors. Proteins were separated by sodium dodecyl sulfate (SDS)-polyacrylamide gel electrophoresis (PAGE) and then transferred onto polyvinylidene fluoride (PVDF) membranes. The membranes were blocked at room temperature with 5% BSA (BIO Froxx) for one hour and subsequently incubated overnight at 4 °C with primary antibodies. The primary antibodies used in this study targeted the following proteins: CD68(ab213363, 1:1000), iNOS (ab178945, 1:1000), IL1β (ab234437, 1:1000), and HIF1α(ab179483, 1:1000) from Abcam; COX2(12282, 1:1000), STING(13647, 1:1000), pSTING(72971T, 1:1000), TBK1(3504, 1:1000), pTBK1(5483T, 1:1000), IRF3(4302S, 1:1000) and pIRF3(29047T, 1:1000) from CST; ACTB(66009-1-Ig, 1:2000), LDHA(19987-1-AP, 1:1000), PKM2(15822-1-AP, 1:1000), HK2(22029-1-AP, 1:1000), GLUT1(21829-1-AP, 1:1000) from PTG; H3K9la (PTM-1419RM, 1:1000), H3K18la (PTM-1406RM, 1:1000), H4K8la (PTM-1415RM, 1:1000), H4K16la (PTM-1417RM, 1:1000) and panKla(PTM-1401RM, 1:1000) from PTM BIO; Histone H3 (A17562, 1:2000) from Abclone. Finally, the membranes were incubated with HRP-conjugated secondary antibodies, and chemiluminescence was detected using an ECL detection system.

Immunofluorescence (IF) staining and Immunohistochemistry (IHC)

Fresh small intestines were fixed in 4% paraformaldehyde, embedded in paraffin, and sectioned using a microtome. Sections were deparaffinized, rehydrated, and washed with distilled water. Antigen retrieval was performed under high-temperature and high-pressure conditions using EDTA buffer (pH 9.0). The sections were blocked with 10% serum at 37 °C for 30 min. After removing the serum, the sections were incubated overnight with primary antibodies. The sections were incubated with secondary antibody working solution diluted in TBST at 37 °C for 45 min, followed by three washes with TBST. For tyramide signal amplification, a tyramide working solution was applied to each section, incubated at room temperature for 10 min, and washed. The process was repeated for staining the second marker.

Finally, nuclei were counterstained with DAPI (1:500, Solarbio, C0060) for 5 min in the dark. The sections were mounted with fluorescent mounting medium and scanned using a fluorescence scanner (3DHISTECH, Pannoramic MIDI, Hungary). Images were captured under identical parameters.

Metabolic assays using the seahorse xFe96 analysis

RAW264.7 cells were seeded into an XFe96 microplate at a density of 1 ×10⁵ cells per well and incubated overnight. After incubation, the cells were treated with or without 100 ng/mL lipopolysaccharide (LPS) for 24 h. The culture medium was then replaced with XF DMEM base medium supplemented with glucose (10 mM), sodium pyruvate (1 mM), and glutamine (2 mM). The cells were incubated in a CO2-free incubator at 37 °C for 1 h. Using the XFe96 analyzer, the extracellular acidification rate (ECAR) was measured following the sequential injection of compounds from the XF Glycolysis Stress Test Kit, including rotenone and antimycin A (0.5 μM) and 2-deoxyglucose (50 mM). ECAR values were automatically calculated using Seahorse XFe96 software.

Flow cytometry

The isolated intestinal muscularis was digested at 37 °C in a shaking water bath for 40 min using the following enzyme mixture: RPMI 1640 + 3% FBS + 0.2 mg/mL collagenase IV + 0.1 mg/mL DNase I + 1 mg/mL neutral protease. The resulting cell suspension was filtered through a 70 μm cell strainer and stained at 4 °C for 20–30 min with fluorophore-conjugated monoclonal antibodies diluted 1:200. Flow cytometric analysis was performed on an Arial II (BD, USA). All flow cytometry data were analyzed using FlowJo software (BD, version 10.8.1). The staining procedure for RAW264.7 cells was consistent with the aforementioned protocol. The antibodies included CD45-FITC(553079), iNOS-PE(696806) from Biolegend; CD86-BV605(563055), CD206-AF647(565250), CD11b-PerCP-Cy5.5(550993) and F4/80-BV421(565411) from BD Pharmingen. Dead cells were excluded using FVS510(564406) from BD Pharmingen.

ELISA

The levels of IL-1β、IL6、TNF-α and LPS were measured by multiplex secretome analysis (XMPlex01240648, XMplex Mouse 4-Plex Custom Panel and SXME04Q004, LPS ELISA Kit) according to manufacturer’s instructions with the assistance of SXM Biotechnology Co., Ltd. (WuHan, China).

ATAC-seq

According to the manufacturer’s instructions, the ATAC experiment was performed using the Hyperactive ATAC-Seq Library Prep Kit for Illumina (TD711, Vazyme). Briefly, cells were collected at room temperature, lysed using Lysis Buffer, followed by low-speed centrifugation at 2300 rpm for 10 min at 4 °C to collect nuclei. The nuclei were fragmented using Fragmentation Mix, and DNA was extracted from the fragmented products using ATAC DNA Extract Beads. After transposase activation and tagging, the DNA was extracted, amplified, and purified for library construction. Once qualified libraries were prepared, sequencing was performed on the Illumina Novaseq platform at Novogene Technology Co., Ltd. (Beijing, China), generating 150 bp paired-end reads.

Chip-seq data analysis

We downloaded the Chip-seq bedgraph file of the transcription factor IRF3, treated with pIC for 90 min in DC cells (GSE125340)³², from the NCBI Gene Expression Omnibus (GEO) database and visualized it using the IGV viewer.

Quantitative PCR analysis

Total RNA from cells was extracted using TRIzol® Reagent (Invitrogen). The RNA samples were then reverse-transcribed to complementary DNA (cDNA) using HiScript III RT SuperMix for qPCR (Vazyme Biotech Co., Ltd., Nanjing, China). The synthesized cDNAs were amplified using a ViiATM 7 Real-Time PCR system (Applied Biosystems, United States). Melting curve analysis was performed to confirm specific PCR products. Relative mRNA expression was calculated by the 2^-∆∆CT method. Suitable primers were designed in our laboratory and were synthesized by TSINGKE Biological Technology Inc. (Nanjing, China). The primers are listed in Supplementary Table S1.

CUT&Tag assay and CUT-PCR

According to the manufacturer’s instructions, the Hyperactive Universal CUT&Tag Assay Kit for Illumina Pro (TD904, Vazyme) was used to perform the CUT&Tag experiment. Briefly, RAW macrophages were collected and their nuclei were extracted. The extracted nuclei were then collected and bound to Concanavalin A-coated magnetic beads. The Concanavalin A-bound nuclei were resuspended in antibody buffer and incubated with primary antibodies targeting H4K8la and p-IRF3, followed by secondary antibody incubation. pA-Tn5 transposase was added to the samples. After transposase activation and labeling, DNA was extracted, amplified, and purified for library construction. The qualified libraries were then sequenced on the Illumina Novaseq platform at Novogene Technology Co., Ltd. (Beijing, China), generating 150 bp paired-end reads.

First, we performed basic quality statistics of the raw read sequences using FastQC. Then, we preprocessed the Illumina-generated FASTQ format read sequences using Trimmomatic software. The remaining reads that passed the filtering steps were considered as “clean reads,” and all subsequent analyses were based on these clean reads. Finally, we conducted another round of basic quality statistics on the clean reads using FastQC. The reference genome and gene model annotation files were directly downloaded from the genome database. We used Bowtie2 to build an index for the reference genome and aligned the clean read sequences to it. Peak calling was performed using MACS2. The peak summit positions relative to the transcription start site (TSS) were used to predict protein-gene interaction sites. Peak Annotator was employed to identify the nearest TSS for each peak and visualize the distribution of distances between the peaks and TSS.

The CUT-PCR experiments were conducted using Hyperactive Universal CUT&Tag Assay Kit from Vazyme (TD904), and experiments were conducted according to the instructions. The qPCR primers are listed in Supplementary Table S2.

Targeted metabolomic profiling

Macrophages stimulated with LPS were collected and rapidly frozen in liquid nitrogen. The samples were resuspended in methanol, and metabolites were extracted via freeze-thaw cycles. Subsequently, analysis was performed using a liquid chromatography-electrospray ionization tandem mass spectrometry (LC-ESI MS/MS) system, consisting of an EClassical 3200, a Waters ACQUITY H-Class Liquid Chromatograph, and a QTRAP 6500+ Mass Spectrometer. This targeted metabolomics analysis was conducted by PANOMIX Co., Ltd. (Suzhou, China).

Plasmid construction and adeno-associated virus production

Macrophage-specific HK2 overexpression was achieved using recombinant adeno-associated virus serotype 9 (AAV9). The macrophage-specific AAV9 vector (pAAV-F4/80p-HK2-EGFP-3Flag-SV40 PolyA) was constructed as follows: Mouse Hk2 gene (NM_013820) was obtained from the cDNA library of Genechem (Shanghai, China) using the following primers: forward 5’-GTCTCGAGGGATCCGCTAGCCGCCACC-3’ and reverse 5’-CAGCTATGGTGGCGGGGCCC-3’. The AAV vector plasmid GV576 (F4/80p-MCS-SV40 PolyA, Shanghai Genechem Co., Ltd.) and Hk2 gene sequence were digested with AgeI and NheI restriction enzymes, then cloned via In-Fusion recombination. The recombinant vector was verified by DNA sequencing.

The viral vector was cotransfected into 293T cells using Lipofectamine 2000 (Invitrogen; Thermo Fisher Scientific, Inc.) together with the pHelper and pRepCap plasmids. Adeno-associated viruses were harvested 72 h post-transfection. AAV9 was purified via iodixanol gradient ultracentrifugation followed by concentration. Purified AAV viruses were titered using a quantitative PCR-based method. All AAVs used in this study were prepared in a 0.001% Pluronic F-68 solution (Poloxamer 188 Solution, PFL01-100ML, Caisson Laboratories, Smithfield, UT, USA).

8-week-old male STING^−/− mice were injected via the tail vein with 100 μL of AAV9-HK2 (2 × 10¹¹ vg total) or control AAV9-EGFP (empty vector). Experiments were performed 3 weeks after AAV injection to ensure sufficient HK2 overexpression.

Statistics and reproducibility

GraphPad Prism 9.5.0 software (GraphPad, USA) was used for all statistical analyses. Results are expressed as mean ± standard error (SEM). For comparisons involving more than two groups, ANOVA was performed, while t tests were used for comparisons between two groups. A p-value < 0.05 was considered statistically significant

Reporting summary

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

Author contributions

All authors contributed to the study conception and design. C.K. conceived and designed the experiments. C.K., L.G. and W.X. performed the experiments. C.K., C.Y., Z.X. analyzed the data. C.K. wrote the article. H.Q. and L.S. revised the manuscript. G.W., H.Q. and L.S. conceived and supervised the study. All authors read and revised the manuscript. The authors agreed on the authorship and publication of this article.

Peer review

Peer review information

Communications Biology thanks the anonymous reviewers for their contribution to the peer review of this work. Primary Handling Editors: Toshiro Moroishi and Mengtan Xing.

Data availability

The accession number for the bulk RNA-seq, ATAC-seq and CUTTAG-seq data is CRA034657 and CRA036362. The data supporting the findings of this study are available from the corresponding author upon reasonable request. The source data behind the graphs in the paper is available in Supplementary Data 1.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s42003-026-09602-1.