Macrophage MERTK restrains GRAMD1A-driven mitochondrial oxysterol trafficking to limit colitis

Abstract

Impaired macrophage efferocytosis drives chronic inflammation, yet the underlying immunometabolic mechanisms remain poorly defined. In this study, we identify that impaired macrophage efferocytosis triggers excessive mitochondrial oxysterol accumulation, which drives mitochondrial ROS-associated inflammation in Crohn's disease (CD). We report an inverse correlation between CD activity and the abundance of MERTK⁺ macrophages. Mechanistically, MERTK loss enhances cholesterol biosynthesis in the endoplasmic reticulum (ER) and stabilizes the cholesterol transporter GRAMD1A. This promotes excessive cholesterol trafficking from the ER to mitochondria. Within mitochondria, this cholesterol is metabolized into specific oxysterols that, unlike other mevalonate pathway metabolites, directly impair the electron transport chain. This impairment triggers a burst of mitochondrial reactive oxygen species (ROS). Collectively, our findings define a pathogenic axis linking defective efferocytosis, dysregulated cholesterol metabolism, and oxysterol-induced mitochondrial damage.

Graphical abstract

Highlights

• •MERTK deficiency aggravates colitis via a mitochondrial-dependent inflammatory axis.
• •Overload of oxysterols, not cholesterol biosynthetic intermediates, trigger mtROS.
• •Oxysterols act as a critical driver for TBK1/NF-κB activation.
• •Targeting Gramd1a alleviates intestinal inflammation in vivo.

1. Introduction

Efficient efferocytosis by professional phagocytes is indispensable for physiological homeostasis [1]. In both resting and inflammatory states, most somatic cells die by apoptosis, and their prompt, silent removal prevents aberrant immune activation [2]. Several studies [3,4] have shown that enhancing intestinal efferocytosis mitigates tissue inflammation. Nevertheless, the impact of defective phagocytic signaling on the immunometabolic status and survival of the phagocyte itself remains largely unexplored.

The molecular recognition of apoptotic cells is orchestrated by a suite of receptors that recognize exposed phosphatidylserine, which is a canonical ‘eat-me’ signal [5]. Among these, MERTK is distinguished by its predominant expression on professional phagocytes, particularly macrophages, where its abundance correlates with tissue maturation and homeostatic clearance capacity. Crohn's disease (CD), a debilitating and etiologically enigmatic form of inflammatory bowel disease (IBD), is pathologically defined by recurrent, transmural granulomatous inflammation [6]. Our prior work [7] revealed a striking loss of MERTK expression in the inflamed intestinal mucosa of patients with highly active CD, posing an unresolved question regarding its precise mechanistic contribution to mucosal inflammation.

Converging lines of evidence implicate profound lipid metabolic dysregulation as a central pathogenic feature of IBD [[8], [9], [10]], with cholesterol dysmetabolism frequently observed in both the circulation and inflamed intestinal tissues. Indeed, clinical data suggests that exposure to statins, cholesterol-lowering drugs, is associated with a reduced risk of IBD [11]. Cellular cholesterol biosynthesis is primarily localized to the endoplasmic reticulum (ER). At specialized ER-mitochondria membrane contact sites, cholesterol is trafficked to mitochondria, where it is metabolized by enzymes like cytochrome P450s into a spectrum of oxysterols [12]. Accumulating evidence has established that mitochondrial cholesterol accumulation impairs mitochondrial function [13], and mitochondrial lipid peroxidation also compromises oxidative phosphorylation [14]. However, no previous studies have definitively elucidated the impact of oxysterols, derived from both cholesterol synthesis and catabolism, on mitochondrial injury.

In this study, we bridge these disparate observations to uncover a mechanistic axis linking defective efferocytosis to immunometabolic collapse in macrophages. We demonstrate that MERTK deficiency unleashes uncontrolled ER-localized cholesterol biosynthesis. Employing a combination of live-cell imaging and targeted metabolomics, we reveal that this results in aberrant cholesterol trafficking from the ER to mitochondria. This mitochondrial cholesterol overload precipitates profound bioenergetic failure, evidenced by impaired oxidative phosphorylation, and triggers the accumulation of pro-inflammatory oxysterols. This metabolic crisis culminates in the activation of TANK-binding kinase 1(TBK1), thereby perpetuating intestinal inflammation. Our findings offer a new conceptual framework for therapeutic intervention in CD.

2. CD intestinal inflammation is characterized by profound depletion of MERTK ⁺ macrophages

To investigate the relationship between MERTK expression and Crohn's disease (CD) activity, we analyzed inflamed mucosal tissues from patients stratified by CD activity index (<150 vs. ≥150). We found that MERTK protein levels were significantly reduced in ulcerative mucosa from patients with active CD compared to those in remission (Fig. 1A), a result confirmed by immunohistochemistry, flow cytometric analysis of intestinal macrophages and mRNA qualification (Fig. 1B, Fig. S1A–B). The metalloprotease ADAM17 [8], a known IBD susceptibility locus, mediates MERTK inactivation via ectodomain shedding. We therefore measured soluble MER (sMer) levels in sera from CD patients. Circulating sMer showed a significant positive correlation with CDAI, fecal calprotectin, serum IL-6 and serum TNF-α, but showed no significant correlation with serum hemoglobin, total cholesterol, albumin or ESR (Fig. 1C). Subgroup analysis revealed no significant correlation between peripheral blood sMer levels and prior exposure to biologics (within one month of admission), biologic types or enteral nutrition. According to the Montreal classification, sMer levels were not associated with age or inflammatory location, but were significantly correlated with inflammatory behavior and disease activity. Patients with active inflammation, stricture or fistula exhibited markedly higher peripheral blood sMer levels compared with healthy controls and patients with quiescent CD (Fig. 1D–H, Fig. S1D–E).

Fig. 1.

Downregulation of MERTK correlates with disease severity in Crohn's disease and exacerbates DSS-induced colitis.

(A) Representative Western blot analysis of MERTK protein expression in intestinal biopsies from the non-inflamed and inflamed mucosa of patients with Crohn's disease (CD) compared to healthy controls.

(B) Representative immunohistochemical (IHC) staining of MERTK in non-inflamed and inflamed intestinal tissues from CD patients. Scale bar, 50 μm (estimate).

(C) Correlation analysis of soluble MerTK (sMerTK) levels with the Crohn's Disease Activity Index (CDAI), pro-inflammatory cytokines (TNF-α, IL-6, IFN-γ) and anti-inflammatory cytokine (IL-10) in CD patients. Pearson correlation coefficient ($R^2$) and P-value are indicated.

(D-H) sMerTK levels in CD patients stratified by biologic use (no biologics, infliximab, adalimumab, ustekinumab), Montreal classification disease behavior (B1: non-stricturing/non-penetrating, B2: stricturing, B3: penetrating, B1+P: perianal lesions) with a longitudinal analysis of sMerTK levels over disease course (n = 82), age at diagnosis (A1: ≤16, A2: 17–40, A3: >40) and disease location (L1: ileal, L2: colonic, L3: ileocolonic, L1+upper: upper gastrointestinal).

(I) Gene Set Enrichment Analysis (GSEA) showing the top enriched biological processes in MERTK⁺ versus MERTK⁻ macrophages.

(J) Representative Western blot showing the temporal changes of MERTK expression in the colon of WT mice treated with 2.5% DSS on days 1, 4, and 8.

(K) Flow cytometry analysis of MERTK⁺ cells within the CD45⁺ population in the colonic lamina propria of control and DSS-treated mice.

(L) Schematic illustration of the experimental design for DSS-induced colitis in Wild-type (WT) and mertk^−/−mice.

(M − O) Assessment of colitis severity in WT and mertk^−/− mice following DSS treatment, evidenced by representative macroscopic images of colons (M), body weight changes (N), and colon length quantification (O).

Integrative reanalysis of public datasets (GSE112057, GSE143057, GSE159034, SCP1884) and our in-house scRNA-seq data (GSE236459) consistently revealed a marked depletion of MERTK transcripts alongside elevated ADAM17 expression in inflamed CD tissues (Figs. S1–2). Single-cell profiling localized MERTK primarily to macrophages, where its expression significantly diminished during active inflammation, and its ligand GAS6 to epithelial cells. These findings highlight a disrupted MERTK signaling axis in active CD and suggest that ectodomain shedding may yield a viable circulatory biomarker for disease activity.

Next, we quantified the mRNA expression of canonical inflammatory mediators in MERTK⁺ and MERTK⁻ macrophage subsets. With the exception of CCL5, all analyzed CD-associated inflammatory factors were significantly upregulated in MERTK⁻ macrophages (Fig. S1F), including CCL2, CXCL1, CXCL2, CXCR4, IL-1β, IL-6, IFN-γ and ICAM1.To further determine the functional consequences of altered MERTK signaling, we compared MERTK⁺ and MERTK⁻ macrophage subsets via differential pathway enrichment analysis. MERTK⁻ macrophages were enriched for pathways related to dysregulation of oxidative stress metabolism and mitochondrial dysfunction (Fig. 1I). In contrast, MERTK⁺ macrophages showed greater enrichment in pathways governing innate immune regulation.

3. MERTK deficiency promotes pro-inflammatory macrophage accumulation independent of direct polarization effects

We further investigated the molecular mechanisms by which MERTK regulates intestinal inflammation using a dextran sodium sulfate (DSS)-induced mouse model (Fig. 1J). MERTK protein expression in the intestinal tissue progressively decreased as the duration of DSS treatment and the severity of colitis increased (Fig. 1J). Flow cytometry analysis of isolated lamina propria macrophages revealed a significant reduction in the proportion of MERTK⁺ cells within the CD45 positive population following DSS stimulation (Fig. 1K).

We subsequently generated mertk knockout (KO) mice using CRISPR Cas9 technology (Fig. 1L–S2E). To assess the severity of colitis, we compared multiple parameters between WT and mertk^−/− mice. We monitored body weight, colon length, intestinal barrier damage and the levels of colonic inflammatory cytokines (Fig. 1M − O, 2A-B). Results showed that mertk knockout increases susceptibility to DSS-induced colitis, manifesting as a more severe inflammatory phenotype. In comprehensive transcriptomic profiling of intestinal tissue (Fig. 2C–D, S2F-G), while baseline transcriptional differences between genotypes were minimal (Fig. S2F), mertk^−/− mice exhibited profound alterations under colitic conditions, characterized by aberrant antigen presentation, immune dysregulation, lipid metabolic perturbations, and impaired phagocytic capacity (Fig. 2C and D). The observed differences in intestinal barrier integrity and alcian blue staining from Fig. 2A suggested that mertk deficiency might be accompanied by aberrant microbial colonization. A 16S rRNA gene sequencing on fecal samples showed that mertk deficiency led to reduced microbial diversity and a decreased abundance of beneficial bacteria in the DSS-induced group, such as Lactobacillus and Muribaculaceae (Fig. 2E and F).

Fig. 2.

Mertk deficiency aggravates colonic inflammation, compromises barrier integrity, and alters gut microbiota composition.

(A) Representative histological assessment of colonic tissues involving Alcian Blue staining (mucus), immunofluorescence for tight junction proteins (ZO-1, Occludin) and macrophage markers (CD206, CD86), and H&E staining. Scale bars are indicated.

(B) Quantification of pro-inflammatory and anti-inflammatory cytokines in colonic homogenates.

(C, D) Transcriptomic profiling of colonic tissues. Heatmap of differentially expressed genes and KEGG pathway enrichment analysis (D) highlighting upregulated immune responses and downregulated lipid metabolic processes in mertk^−/− mice.

(E, F) 16S rRNA gene sequencing analysis of gut microbiota, showing Alpha diversity (Simpson index) (E) and relative abundance of bacterial taxa at the family level (F).

(G) Flow cytometric analysis of immune cell infiltration in the colonic lamina propria, quantifying leukocytes (CD45⁺), macrophages (CD11b⁺F4/80⁺), and macrophage polarization states (CD80 vs. CD206).

(H) In vitro validation of macrophage polarization defects in BMDMs following stimulation with IL-4 or LPS.

To further elucidate the relationship between MERTK deficiency, microbiota alterations and colitis, we performed antibiotic (ABX) cocktail gavage to establish gnotobiotic mice (Fig. S3A), and then recapitulated the colitis model in these mice to investigate whether MERTK knockout-induced colitis exacerbation persisted in the absence of the gut microbiota. Two weeks of ABX gavage effectively depleted the gut microbiota in both wild-type and mertk-knockout mice (Fig. S3B–C). After antibiotics treatment, mertk^−/− mice still exhibited increased susceptibility to colitis, as evidenced by significant changes in body weight, colon length, intestinal H&E pathology and inflammatory cytokine expression compared with WT mice (Fig. S3D–F). These results revealed that the observed microbiota alterations (Fig. S4A–C) are correlative factors rather than causal drivers of colitis in mertk-deficient mice.

The localization of MERTK to macrophages in our scRNA-seq data led us to speculate that the worsened inflammation may result from macrophage dysfunction. As expected, mertk deletion impaired efferocytosis both in vitro and in vivo (Fig. S4D–E). Immunophenotyping of the colonic lamina propria from colitic mice revealed exacerbated infiltration of CD45⁺ leukocytes, particularly CD11b⁺F4/80⁺CD80⁺ pro-inflammatory macrophages (Fig. 2G–S4F). To determine whether MERTK directly controls this phenotypic shift, we evaluated WT and mertk^−/− BMDMs in vitro. Strikingly, mertk deficiency did not alter intrinsic M1 (LPS) or M2 (IL-4) polarization capacities (Fig. 2H). These findings demonstrate that the pronounced pro-inflammatory macrophage accumulation in vivo is not due to a skewed canonical polarization program, but rather driven by an unrecognized, intrinsic mechanism linked to defective efferocytosis.

4. Cholesterol homeostasis dysregulation contributes to inflammatory progression following MERTK deficiency

Pathway enrichment analysis of the transcriptomic data identified cholesterol biosynthesis as the most significantly upregulated metabolic program in mertk-deficient tissue, a finding distinct from the canonical glycolytic reprogramming associated with macrophage activation (Fig. S4G–H). This metabolic signature was corroborated by 16S rRNA sequencing, which revealed a significant enrichment of the cholesterol-metabolizing commensal Oscillibacter [9] in mertk^−/− mice, an effect that was further exacerbated during active colitis (Fig. 3A).

Fig. 3.

MERTK deficiency triggers metabolic reprogramming and mitochondrial dysfunction, leading to TBK1 pathway activation.

(A) Relative abundance of specific bacterial genera in colonic contents, highlighting the enrichment of Oscillibacter in mertk-deficient groups.

(B) Western blot analysis of SREBP2 nuclear translocation (n-SREBP2) in cytoplasmic and nuclear fractions of macrophages treated with the MERTK inhibitor (UNC2025) or SREBP inhibitor (Fatostatin).

(C) Assessment of cholesterol metabolism evidenced by total cholesterol quantification in colonic tissues.

(D–F) Evaluation of mitochondrial integrity. (D) Mitochondrial ROS levels detected by MitoSOX Red. (E) Mitochondrial membrane potential assessed by JC-1 staining (aggregates vs. monomers). (F) Representative Transmission Electron Microscopy (TEM) images showing normal mitochondria (yellow arrowheads), damaged mitochondria (red arrowheads), and mitophagy structures (purple arrowheads). Scale bar, 1 μm.

(G–I) Western blot analysis of the STING-TBK1 signaling axis. (G) Time-course activation of STING/TBK1 in DMXAA-stimulated BMDMs. (H) Impact of NLRP3 and MERTK deficiency on TBK1 phosphorylation. (I) Pathway dependency analysis using WT, tbk1^−/−, and sting^−/− RAW264.7 cells.

(J-K) Confocal microscopy visualizing cytosolic accumulation of dsRNA (J2 antibody, green, left panels) and leakage of mitochondrial DNA (right panels) following MERTK inhibition.

(L, M) Schematic of the efferocytosis assay (L) and immunoblotting of mitochondrial fractions (M) assessing LC3 recruitment and cargo clearance. Scale bar, 50 μm 

(N) Representative confocal images of cells treated with DMSO (control) or UNC2025, stained for mitochondria (red/white) and nuclei (Hoechst, blue). Scale bar: 50 μm. Quantification of mitochondrial morphological parameters: average area, average perimeter, form factor, and aspect ratio, comparing DMSO (gray) and UNC2025 (red) treated cells. Data are presented as mean ± SEM; statistical significance was determined by t-test (p < 0.001, p < 0.0001).

Cholesterol biosynthesis is canonically driven by SREBP2 [13], which requires ER-to-Golgi transit and proteolytic cleavage to enter the nucleus and activate sterol-responsive genes. To determine if MERTK regulates this process, we treated immortalized BMDMs with the MERTK inhibitor UNC2025, with or without fatostatin (an inhibitor of SREBP2 trafficking). Subcellular immunoblotting revealed that MERTK inhibition alone was sufficient to drive the nuclear accumulation of cleaved SREBP2, an effect completely abrogated by fatostatin co-treatment (Fig. 3B). These data establish that MERTK kinase activity directly restrains SREBP2 activation. Furthermore, since efferocytosis potently induces MERTK expression [15], we re-analyzed published multi-omic datasets of macrophages engulfing apoptotic cells. Consistent with our in vitro findings, this MERTK-activating process resulted in a robust suppression of the cholesterol biosynthesis pathway (Fig. S4I).

As an essential structural component of cellular membranes, cholesterol is critical for maintaining organellar architecture and function. During efferocytosis, the engulfment of apoptotic cells provides an exogenous source of lipids, thereby obviating the need for de novo synthesis [16]. Consequently, MERTK deficiency imposes a dual metabolic burden: it both abrogates the acquisition of lipids via efferocytosis and triggers a compensatory upregulation of endogenous cholesterol synthesis. However, this state of unrestrained synthesis may predispose the cell to lipid peroxidation and subsequent inflammatory activation.

5. MERTK deficiency induces mitochondrial ROS and activates TBK1-Mediated innate immunity through mtDNA/RNA release

Transcriptomic analysis of human intestinal macrophages linked MERTK expression to mitochondrial metabolism, particularly oxidative phosphorylation (Fig. 1I). Combining this with our murine lipid data, we hypothesized that aberrant cholesterol biosynthesis drives mitochondrial dysfunction and subsequent inflammation. Supporting this, pharmacological MERTK blockade (UNC2025) in vitro significantly elevated mitochondrial ROS, as an effect partially rescued by fatostatin (Fig. 3D), and collapsed the mitochondrial membrane potential (JC-1 assay; Fig. 3E). Furthermore, transmission electron microscopy (TEM) revealed severe ultrastructural damage in MERTK-inhibited cells, characterized by mitochondrial swelling, cristae disorganization (red arrows), and active mitophagy (purple arrows; Fig. 3F).

Considering mitochondrial damage potently triggers innate immunity and releases mtDNA and RNA to activate the cGAS-STING [17] and RIG-I/MAVS [18] pathways, we examined their convergence point: the kinase TBK1 [19]. Indeed, mertk^−/− BMDMs exhibited hyper-responsiveness to the STING agonist DMXAA, indicating primed TBK1 signaling (Fig. 3G). Concurrently, our transcriptomic data revealed an enrichment of inflammasome pathways (Fig. S5A). This aligns with the known capacity of mitochondrial stress byproducts, such as lipid peroxides, to activate the NLRP3 inflammasome and drive gasdermin D-mediated pyroptosis [[20], [21], [22]].

To precisely dissect this signaling hierarchy, we employed pharmacological MERTK inhibition in macrophages genetically deficient for key signaling nodes (nlrp3^−/−, tbk1^−/−, and sting^−/−) (Fig. 3H and I). UNC2025-induced TBK1 phosphorylation was completely dependent on MERTK activity but was unaffected by the loss of nlrp3, ruling out a primary role for the inflammasome in TBK1 activation (Fig. 3H and I). Crucially, downstream analysis revealed a bifurcation of the TBK1-dependent signal. NF-κB activation was completely dependent on TBK1 but mostly independently of STING. In contrast, IRF3 phosphorylation and mitochondria-associated apoptosis (cleaved caspase-3) were partially dependent on both TBK1 and STING (Fig. 3I–S5B).

TBK1 activation is mediated by two distinct mechanisms: dsDNA-cGAS/STING and dsRNA-RIG-I/MAVS. To further quantify the relative contributions of these pathways, we performed in vitro functional experiments using pathway-selective inhibitors, including RIG-012 (RIG-I/MAVS inhibitor), H151 (cGAS/STING inhibitor), BAY 11-7082 (NF-κB inhibitor) and MRT67307 (TBK1 inhibitor). We investigated which pathway inhibition could reverse the inflammatory effects of MERTK signaling deficiency, and obtained the following key results (Fig. S5C). Inhibition of RIG-I/MAVS, cGAS/STING and TBK1 all attenuated TBK1 activation to varying degrees. Upstream of NF-κB/IRF3, RIG-012 and MRT67307 exhibited more potent suppressive effects than H151, indicating that the dsRNA-RIG-I/MAVS-TBK1 axis is the predominant pathway mediating downstream inflammation in the setting of MERTK signaling deficiency.

To directly visualize mitochondrial DNA/RNA release, we employed confocal microscopy using J2 antibody for dsRNA detection and live-cell mtDNA probes (Fig. 3J and K). UNC2025 treatment induced striking punctate accumulation of both dsRNA and mtDNA signals, a canonical signature of enhanced nucleic acid release from damaged mitochondria.

While efferocytosis is known to induce LC3-associated phagocytosis (LAP) [21], we sought to distinguish this from mertk deficiency-induced autophagy. Comparative analysis of whole-cell lysates, isolated mitochondria, and cytoplasmic fractions revealed that while both efferocytosis and MERTK inhibition induced LC3-II accumulation in whole-cell lysates, UNC2025 treatment specifically enriched LC3-II on mitochondrial membranes, indicating selective mitophagy rather than generalized autophagy (Fig. 3L and M). This was further validated using GFP-LC3-expressing RAW264.7 cells, where UNC2025 treatment induced striking colocalization of autophagosomes with mitochondria (Fig. S5D). MERTK signaling deficiency induced significant mitochondrial fragmentation via analyzed with super-resolution confocal microscopy (Fig. 3N). Functionally, these mitochondrial perturbations and inflammatory activation translated to pathological consequences. In macrophage-epithelial co-culture systems, UNC2025 significantly reduced transepithelial electrical resistance, providing evidence for inflammation-mediated barrier dysfunction (Fig. S5E). Level of inflammatory activation was evaluated by detecting the expression of key downstream inflammatory cytokines via qPCR and assessing the reversal of epithelial barrier dysfunction using transwell co-culture assays (Fig. S5F–H). The results were consistent with Western blot data. Inhibition dsRNA-RIG-I/MAVS signaling or direct TBK1 inhibition exerted superior rescue effects on downstream inflammation and epithelial barrier damage compared with STING inhibition.

6. Mitochondrial cholesterol overload impairs oxidative phosphorylation and drives ROS production

To directly link aberrant cholesterol biosynthesis to mitochondrial dysfunction, we dynamically tracked subcellular cholesterol distribution using live-cell fluorescent imaging. While exogenous cholesterol loading (cholesterol-MCD) induced only modest mitochondrial accumulation, MERTK inhibition (UNC2025) dramatically enhanced cholesterol-mitochondria colocalization (Fig. S6A), visually confirming that MERTK blockade drives pathological mitochondrial cholesterol sequestration.

Once transported to the inner mitochondrial membrane (IMM), cholesterol is converted by resident cytochrome P450 enzymes (e.g., CYP27A1) into diverse oxysterols, such as 27-hydroxycholesterol [[22], [23], [24], [25]]. To pinpoint this sub-organelle localization, we employed super-resolution confocal microscopy utilizing an IMM-targeted probe (Fig. S6B) and CYP27A1 immunostaining (Fig. 4A). Strikingly, mertk^−/− cells exhibited strong colocalization of cholesterol specifically with CYP27A1, as a pathogenic feature entirely absents in cells merely loaded with exogenous cholesterol.

Fig. 4.

Mitochondrial accumulation of specific oxysterols disrupts respiratory supercomplex assembly and induces oxidative stress.

(A) Confocal microscopy analysis showing the subcellular distribution of cholesterol. Representative images in macrophages treated with DMSO, cholesterol-cyclodextrin (CHO-MCD), or UNC2025. Intensity spatial profiles (line scans) illustrate the overlap between cholesterol (green) and mitochondrial (red) signals. Scale bar, 20 μm 

(B) Blue-Native PAGE (top) and SDS-PAGE (bottom) analysis of mitochondrial respiratory chain supercomplexes and individual subunits in BMDMs (left) and colonic lamina propria mononuclear cells (LPMCs, right) from WT and mertk^−/− mice.

(C) Schematic diagram of the cholesterol metabolic pathway, highlighting the specific oxysterols identified in the screen.

(D, E) Targeted lipidomics profiling of sterols and oxysterols. Heatmap (D) and violin plots (E) showing differentially abundant metabolites in WT and mertk^−/− macrophages.

(F) Effects of MerTK pathway inhibitors, TBK1 pathway inhibitors, cholesterol modulators on 27-OHC levels in whole cell lysates (WCL) and mitochondrial fractions (Mito).

(G) Blue-Native PAGE assessing the impact of specific oxysterols (25-HC, 24-HC, etc.) on the stability of mitochondrial respiratory supercomplexes.

(H–K) Assessment of mitochondrial ROS induction by different cholesterol metabolites. Flow cytometry histograms (H, J) and quantification (I, K) of MitoSOX Red intensity in cells treated with various oxysterols or mevalonate pathway intermediates (FPP, GGPP), in the presence or absence of the uncoupler CCCP. Note the potent pro-oxidant effect of di- and tri-hydroxy oxysterols.

While physiological oxysterols are essential metabolic precursors, their excessive accumulation in mitochondria disrupts the electron transport chain, compromising oxidative phosphorylation (OXPHOS) and triggering ROS [14]. Evaluating these bioenergetic consequences, we found that although baseline WT and mertk^−/− BMDMs were comparable, inflammatory stimulation unmasked profound OXPHOS defects in mertk^−/− LPMCs. This was evidenced by impaired respiratory complex assembly (blue native PAGE) and altered subunit expression (SDS-PAGE) (Fig. 4B). Notably, these massive macrophage-specific bioenergetic failures were completely masked in whole-tissue metabolic profiles (Fig. S2G), underscoring that MERTK orchestrates a highly cell-type-specific metabolic defense within the inflamed intestinal microenvironment.

We performed targeted metabolomic analysis of BMDMs from WT and mertk^−/− mice to comprehensively profile oxysterol dysregulation (Fig. S6C). Heat map visualization revealed striking alterations in the oxysterol landscape following mertk deficiency, with 27-hydroxycholesterol (27-OHC), 7β-hydroxycholesterol, 6α-hydroxy-5α-cholestane, 24S,25-epoxycholesterol, and cholestan-3β,5α,6β-triol emerging as the most significantly elevated species (Fig. 4D and E). CYP27A1 is the major enzyme converting mitochondrial cholesterol to 27-OHC (Fig. 4C). To clarify their roles in MERTK-mediated mitochondrial dysfunction, we quantified 27-OHC levels (Fig. 4F) and CYP27A1 protein (Fig. S6D–E) in whole cells and purified mitochondria after MERTK inhibition. UNC2025 significantly upregulated CYP27A1 expression in whole cells and mitochondria, an effect reversed by the ER-mitochondria cholesterol trafficking inhibitor autogramin-2 (AG-2). However, inhibition of TBK1 pathway activation following mitochondrial injury at multiple targets failed to alter the mitochondrial CYP27A1 changes induced by UNC2025 (Fig. S6D). These results establish a direct link between MERTK deficiency, mitochondrial CYP27A1 upregulation, 27-OHC accumulation, and mitochondrial dysfunction.

Based on these data, subcellular fractionation and quantitative cholesterol measurement (Fig. S6F–I) were established subsequently to complement our imaging data and provide direct evidence for mitochondrial cholesterol enrichment. We isolated mitochondria and ER from RAW264.7 cells and LPMCs via enzymatic digestion combined with differential centrifugation, and further fractionated mitochondria into IMM and outer mitochondrial membrane (OMM) fractions. MERTK deficiency, DSS-induced colitis or in vitro MERTK inhibition with UNC2025 all significantly increased cholesterol levels in total mitochondria, ER, IMM and OMM fractions. This effect was not observed in nlrp3, sting or tbk1 knockout cells, confirming the specificity of the MERTK signaling pathway (Fig. S6F–I).

We then additionally examined whether exogenous supplementation of sterol metabolites could rescue or exacerbate the effects under abnormal sterol metabolism. We treated cells with a panel of oxysterols (cholesterol catabolic products) versus canonical mevalonate metabolites (cholesterol synthesis intermediates), and subsequently evaluated changes in 27-OHC level (Fig. 4F), oxidative phosphorylation capacity (Fig. 4G) and mtROS production (Fig. 4H–K). Notably, neither cholesterol nor mevalonate intermediates directly triggered an upregulation of mtROS. Conversely, only oxysterols, especially 7β-hydroxycholesterol and cholestan-3β,5α,6β-triol, demonstrated superior potency in mtROS induction and disrupting respiratory complex assembly (Fig. 4G–K), establishing a hierarchy of oxysterol-mediated mitochondrial toxicity. However, mevalonate pathway intermediates reversed the abnormal elevation of 27-OHC and mtROS induced by MERTK deficiency.

The oxysterol-induced mitochondrial damage, also known as oxiapoptophagy [23], manifested through three interconnected pathways: mitochondrial ROS burst (previously demonstrated), enhanced mitophagy, and activation of the intrinsic apoptotic cascade via caspase-3 [23,24]. These effects exhibited dose-dependent responses to UNC2025 (Fig. S7A–B), and were partially rescued by the mitochondria-targeted ROS scavenger Mito-TEMPO (Fig. S7B), confirming that oxidative stress serves as a central mediator of MERTK deficiency-induced mitochondrial pathology.

7. Loss of MERTK unleashes pathological cholesterol trafficking to mitochondria

Mitochondria orchestrate complex intracellular cholesterol trafficking from multiple pathways. While cytosolic free cholesterol transport via TSPO/StARD1 is well-established, de novo cholesterol synthesized in the ER is also dynamically trafficked through membrane contact sites or cytosolic shuttles (Fig. 5A). Having demonstrated that MERTK deficiency triggers a compensatory, SREBP2-dependent upregulation of cholesterol biosynthesis due to impaired efferocytosis (Fig. 3B and C), we hypothesized that the pathological consequences extend far beyond mere overproduction. Instead, we posited that MERTK deficiency profoundly dysregulates the intracellular trafficking network, actively funneling this newly synthesized cholesterol into mitochondria for aberrant accumulation.

Fig. 5.

Gramd1a mediates mitochondrial dysfunction and inflammatory signaling downstream of MERTK deficiency.

(A) Schematic illustration of cholesterol transport proteins (StARD1, TSPO, Gramd1a) localized at ER-Mitochondria contact sites (MAMs).

(B-D) Western blot analysis in iBMDMs treated with Bafilomycin A1 (Baf A1, 100 nM), UNC2025 (1 μM), and/or AG2 (50 nM) as indicated.

(E, F) qRT-PCR analysis of Gramd1a, inflammatory cytokines (E), and cholesterol biosynthesis genes (F).

(G, H) Seahorse XF analysis of mitochondrial respiration in BMDMs. (G) Real-time Oxygen Consumption Rate (OCR) profiles following sequential injection of Oligomycin, FCCP, and Rotenone/Antimycin A (ROT/AA). (H) Quantification of basal respiration, maximal respiration, ATP production, and spare respiratory capacity.

The flux of cholesterol between the ER and mitochondria is principally governed by the GRAM Domain Containing 1 (GRAMD1) protein [22]. GRAMD1A prominently localized to the inter-organellar contact sites that facilitate this [26]. Autogramin-2 [27] (AG-2) is a natural product derived inhibitor that selectively targets GRAMD1 [27]. We sought to delineate the contributions of distinct cholesterol trafficking routes by using AG-2 to block the ER-Mito axis and siRNA (Fig. S7C) to silence StARD1 dependent transport, assessing the capacity of each intervention to reverse phenotypes induced by MERTK inhibition.

Treatment with AG-2 dose-dependently reversed the accumulation of oxysterols, TBK1 phosphorylation, and the induction of autophagy caused by UNC2025 (Figs. 4F and 5B-D, S5C-D, S6D, S7D-F). This effect was specific to the ER-Mito pathway, as silencing StARD1 did not suppress TBK1 activation (Fig. 5B). As a control, inhibition of downstream autophagic flux with bafilomycin A1 also had no effect, positioning the action of AG-2 upstream of lysosomal fusion (Fig. 5C). These data implicate GRAMD1A-mediated transport as the principal pathway for MERTK-related mitochondrial oxysterols accumulation. Upregulation of GRAMD1A protein following UNC2025 treatment occurred with only a slight increase in its transcript, indicating that this regulation is primarily post-transcriptional, while GRAMD1B and GRAMD1C remained unchanged ((Fig. S7E, 5D-E).

We next sought to determine if the inflammatory and bioenergetic consequences of MERTK inhibition were functionally dependent on this GRAMD1A-mediated cholesterol transport. Results showed that AG-2 effectively suppressed the transcriptional upregulation of inflammatory factors and improved epithelial barrier injury induced by UNC2025 (Fig. 5E-F, S5F-H). Additionally, seahorse analysis also revealed that MERTK inhibition severely impaired macrophage ATP production and cellular energy reserves. These deficits were substantially rescued by AG-2 treatment (Fig. 5G and H).

This specific metabolic impairment is in agreement with our transcriptomic analysis of macrophages isolated from CD patients with (Fig. 1I). MERTK-negative macrophages from inflamed intestinal tissue exhibited signatures of severe mitochondrial dysfunction. These included profound respiratory chain uncoupling, diminished ATP metabolism, and impaired utilization of precursor energy substrates. This powerful concordance suggests that the function of MERTK in the context of CD pathology is critically dependent on its phosphorylation status.

8. MERTK-ITIM motif recruits SHP1 to orchestrate RNF123-Mediated K48-Ubiquitination of GRAMD1A

Having identified GRAMD1A as the driver of mitochondrial cholesterol overload, we hypothesized that this dysregulation arises from impaired signaling through MERTK's intracellular tyrosine kinase domain. This region contains a conserved immunoreceptor tyrosine-based inhibitory motif (ITIM) centered at Y749 [28], which serves as a docking site for SHP1/SHP2 phosphatases upon phosphorylation [29]. To identify MERTK's specific signaling partners, we performed endogenous immunoprecipitation followed by mass spectrometry (IP-MS) in RAW264.7 cells. This unbiased approach identified SHP1 (ptpn6) as a MERTK-associated protein, while SHP2 was not detected (Fig. 6A). Validation experiments showed that MERTK inhibition (UNC2025) abolished SHP1 phosphorylation, as an effect partially rescued by the mitochondrial permeability transition pore (mPTP) inhibitor cyclosporin A (Fig. 6B). These results suggest that MERTK maintains mitochondrial cholesterol homeostasis by recruiting and activating SHP1 via its cytoplasmic kinase domain.

Fig. 6.

MERTK recruits SHP1 to the endoplasmic reticulum via Y749 phosphorylation to suppress mitochondrial oxidative stress.

(A) Proteomic identification of MERTK downstream effectors. Venn diagram and table highlight PTPN6 (SHP1) as a specific target enriched following GAS6 stimulation.

(B) Western blot analysis in THP-1 cells treated with increasing concentrations of UNC2025, showing dose-dependent suppression of SHP1 phosphorylation (p-SHP1) and concurrent activation of TBK1.

(C) Schematic representation of Wild-type (WT) MERTK and its mutant forms: the Y749F point mutant (ITIM domain disrupted) and the kinase-dead mutant (δ659-943).

(D) Western blot analysis in 293T cells transfected with the indicated MERTK constructs. Note that the Y749F mutation abolishes MERTK-mediated SHP1 phosphorylation, mirroring the effect of the kinase-dead mutant or UNC2025 treatment.

(E, F) Confocal microscopy assessing the subcellular localization of SHP1. Representative images (E) and co-localization analysis (F) (line scans and Pearson's correlation coefficient) demonstrate that MERTK deficiency impairs the recruitment of SHP1 (red) to the Endoplasmic Reticulum (ER, green). (F) Fluorescent imaging of total ROS (green) in iBMDMs. Scale bar, 300 μm 

(G, H) Functional validation of the SHP1-ROS axis. Fluorescence imaging (G) and flow cytometric quantification (H) of mitochondrial ROS (mtSOX) in cells transfected with si-SHP1.

To pinpoint the precise mechanism of SHP1 recruitment, we engineered three MERTK plasmids: wild-type (WT), a Y749F point mutant disrupting the ITIM motif, and a truncation mutant lacking the entire kinase domain (Δ659-943) (Fig. 6C). Transfection studies in 293T cells revealed that while WT-MERTK underwent robust autophosphorylation and induced concomitant SHP1 activation, the Y749F mutant retained autophosphorylation capacity but failed to recruit SHP1. The Δ659-943 truncation abolished both MERTK autophosphorylation and SHP1 recruitment (Fig. 6D), establishing the ITIM motif as essential for SHP1 activation despite being dispensable for MERTK catalytic activity.

Given that GRAMD1A localizes to the ER, we next examined SHP1 subcellular distribution using confocal microscopy. MERTK activation induced pronounced SHP1-ER colocalization in BMDMs (Fig. 6E and F). Functional validation showed that Ptpn6 knockdown exacerbated UNC2025-induced ROS production, while knockdown of Slc25a5, a mitochondrial nucleic acid release channel [30], served as a control for ROS attenuation (Fig. 6G and H). Besides, Ptpn6 knockdown also increased cholesterol levels in total mitochondria, ER and IMM (but not OMM), consistent with its role as a downstream effector of MERTK (Fig. 4F–S6F-I).

Our observation that GRAMD1A regulation occurs post-transcriptionally (Fig. 5D and E), combined with the inability of bafilomycin A1 to affect GRAMD1A levels (Fig. 5D), suggested proteasomal rather than autophagic degradation. Since K48-linked ubiquitination mediates more than 90% of proteasomal degradation, we transfected 293T with wild-type, K48-only, or K63-only ubiquitin plasmids. As shown in Fig. 7A, inhibition of MERTK phosphorylation suppressed the K48-linked, but not K63-linked, ubiquitination of GRAMD1A.

Fig. 7.

MERTK Regulates GRAMD1A Stability via a SHP1-RNF123 Ubiquitination Axis.

(A) Ubiquitination assay analyzing the linkage specificity of Gramd1a ubiquitination. 293T cells were transfected with Myc-Gramd1a and different HA-Ubiquitin mutants (WT, K48-only, K63-only) in the presence or absence of UNC2025.

(B) Schematic of the SHP1 interactome screen. SHP1-interacting proteins were identified by IP-MS from GAS6-stimulated THP-1 cells.

(C) Western blot validation of the functional hierarchy. Knockdown of either RNF123 or PTPN6 (SHP1) abolishes the suppressive effect of GAS6/MERTK signaling on TBK1 activation, mirroring the phenotype of UNC2025 treatment.

(D) Model of the MERTK-SHP1-RNF123 signaling axis.

(E-G) Co-IP mapping of the SHP1-RNF123 interaction in 293T cells.

(H) Co-IP for GRAMD1A ubiquitinaiton in 293T cells after RNF123 knockdown (siRNA) and MG132 treatment (10 μM, 6h).

Considering that ubiquitination requires an E3 ligase, we next tried to identify the ligase recruited by SHP1. We immunoprecipitated SHP1 and analyzed associated proteins by 4D-LC/MS, focusing on those involved in the ubiquitin process (Fig. 7B). Among these, RNF123 was the most abundant E3 ubiquitin ligase. Functional validation confirmed that siRNA-mediated knockdown of either SHP1 or RNF123 enhanced UNC2025-induced TBK1 activation (Fig. 7C), establishing the SHP1/RNF123 axis as critical for restricting GRAMD1A-mediated cholesterol transfer and maintaining mitochondrial homeostasis (Fig. 7D).

To confirm the recruitment of RNF123 by SHP1, we mapped their interaction using a series of truncation constructs (Fig. 7E–G). Co-immunoprecipitation assays revealed a direct binding between the PTP domain of SHP1 (residues 244-595) and the SPRY domain of RNF123 (residues 74-254). Crucially, depletion of RNF123 significantly reduced GRAMD1A ubiquitination (Fig. 7H).

These findings establish a novel regulatory circuit wherein MERTK's ITIM motif recruits and activates SHP1, which subsequently translocate to the ER and engages the RING-type E3 ligase RNF123. This complex promotes K48-linked ubiquitination of GRAMD1A, thereby constraining ER-mitochondria cholesterol transfer.

We turn to determine the in vivo relevance of the GRAMD1A pathway by employing an AAV9-shRNA system for myeloid-specific Gramd1a knockdown in a model of colitis (Fig. 8A–C). Myeloid-specific ablation of Gramd1a conferred remarkable protection against DSS-induced acute colitis, a benefit substantially reflected in the amelioration of body weight loss, colon shortening, inflammatory cell infiltration, and cytokine production. (Fig. 8D–I). Notably, Gramd1a-KD mice exhibited a reduction in diarrhea, maintaining formed stools even after a 7-day DSS challenge. This protective effect was further substantiated by preserved tissue architecture in H&E-stained colon sections, where EGFP-positive transduced cells were readily detected (Fig. 8B).

Fig. 8.

Myeloid-specific knockdown of Gramd1a alleviates DSS-induced colitis severity.

(A) Schematic representation of the experimental strategy for myeloid-specific Gramd1a knockdown. Lyz2-Cre mice were injected with AAV9 vectors carrying a Cre-dependent shRNA cassette (FLEX-shRNA-Gramd1a) prior to DSS induction.

(B) Representative fluorescence images of colonic sections showing successful transduction of the AAV vector (GFP, green) in the colonic mucosa.

(C) Western blot confirmation of Gramd1a knockdown efficiency in colonic lamina propria mononuclear cells (LPMCs) isolated from Lyz2-Cre mice at 4 and 5 weeks post-injection.

(D) Daily body weight change in DSS-treated WT, Mertk^−/−, and Lyz2-Cre; shGramd1a-AAV mice. Data are mean ± SEM (n = 5).

(E-G) Body weight, colon length (cm), and gross colon morphology on day 7.

(H) H&E staining of distal colon sections. Scale bar, 100 μm.

(I) qRT-PCR analysis of pro-inflammatory cytokines (Tnf, Il1b, Il6) and tight junction proteins (Occludin, Claudin1) in colonic tissues.

Given GRAMD1A's established roles in sterol homeostasis and autophagy, we performed long-term follow-up of myeloid-specific Gramd1a-knockdown mice until advanced age (>18 months). Additionally, we established a chronic colitis model (64 days with 3 cycles of 1.5% DSS administration) in these knockdown mice and age-matched control mice to investigate the effects of long-term GRAMD1A inhibition on both non-inflammatory and chronic intestinal inflammatory states (Fig. S8A). At 25 weeks of age, long-term GRAMD1A signaling inhibition had no adverse effects on mouse body weight or normal hair development (Fig. S8B), indicating no overt developmental abnormalities under basal non-inflammatory conditions. In the chronic colitis model, GRAMD1A signaling inhibition remained a protective factor. Multi-dimensional analyses (including collagen deposition, inflammatory cytokine expression, H&E staining, colon length, intestinal wall thickness and body weight) demonstrated that GRAMD1A suppression attenuated chronic colonic inflammation and colonic fibrosis without inducing spontaneous inflammation or impairing normal growth and development (Fig. S8B–F). Together, our findings establish myeloid GRAMD1A as a critical regulator of intestinal inflammation and validate it as a compelling therapeutic target for colitis.

9. Discussion

The nexus between sterol metabolism and autoimmune pathogenesis has emerged as a cornerstone of immunometabolism. However, whether the spatiotemporal dysregulation of sterol trafficking at the subcellular level drives the immunopathology of Crohn's disease (CD) remains a critical frontier. By integrating high-resolution single-cell and spatial transcriptomics with targeted lipidomics and proteomics, our study unmasks a previously unrecognized metabolic-immune checkpoint governed by MERTK. We demonstrate that MERTK deficiency does not merely elevate bulk cholesterol, but precipitates a catastrophic mitochondrial redox collapse. This pathogenic cascade is orchestrated by the SREBP2-driven biosynthesis and GRAMD1A-mediated funneling of cholesterol into mitochondria, where its conversion into pro-oxidant oxysterols directly impairs the electron transport chain. Ultimately, this bioenergetic failure culminates in the release of mitochondrial nucleic acids and the aberrant activation of the TBK1/NF-κB/IRF3 innate immune axis, providing a novel mechanistic rationale for macrophage-driven mucosal inflammation in CD.

The pathogenic significance of cholesterol oxidation products is increasingly recognized, particularly given that genome-wide association studies (GWAS) have linked the GPR183-oxysterol axis to IBD susceptibility [[31], [32], [33]]. However, existing literature frequently conflates the immunological effects of cholesterol biosynthesis precursors with those of its catabolic breakdown products. Our study provides a critical conceptual refinement by dissecting their differential impacts on mitochondrial integrity. Strikingly, we demonstrated that mevalonate pathway intermediates (e.g., FPP and GGPP) remain immunologically inert regarding mitochondrial oxidative damage at physiological concentrations. In stark contrast, their terminal conversion into mitochondrial oxysterols provokes massive structural damage and mtROS bursts. Crucially, we identified a structure-function relationship wherein the pro-oxidant potency is strictly dictated by the degree of hydroxylation: di- and tri-hydroxylated oxysterols uniquely drive mitochondrial collapse, whereas mono-hydroxylated forms are largely benign. This distinct capacity of multi-hydroxylated oxysterols to shatter mitochondrial bioenergetics constitutes a major mechanistic novelty of our work.

Among TAM receptor tyrosine kinases, MERTK is unique for its pronounced macrophage-restricted expression [34]. While traditionally viewed as a mere marker of macrophage maturation [35], our patient-derived scRNA-seq and knockout mouse data fundamentally challenge this paradigm. MERTK deficiency impairs neither macrophage development nor the maintenance of the intestinal pool; rather, it functions as a critical metabolic rheostat. It calibrates the macrophage's threshold for PAMP/DAMP responsiveness by maintaining subcellular lipid homeostasis. This concept perfectly contextualizes our previous finding that MERTK hyperactivation paradoxically drives intestinal fibrosis. Together, these bifurcating roles, where deficiency unleashes mucosal inflammation and hyperactivation fuels fibrogenesis, position MERTK as a master molecular switch governing CD phenotypic transition. This is further corroborated by recent evidence linking MERTK to ferroptosis resistance via SLC7A11 [36], independently validating our conclusion that MERTK is indispensable for shielding cells against catastrophic lipid peroxidation.

Mechanistically, our delineation of the SHP1-RNF123-GRAMD1A axis provides a spatial explanation for this metabolic control. RNF123, a RING-type E3 ligase [37], requires spatial proximity to its substrates. We discovered that SHP1 bridges this gap by binding the substrate-recognition SPRY domain of RNF123, thereby tethering the E2 ubiquitin machinery to the ER-resident GRAMD1A. Importantly, GRAMD1A exhibits striking functional plasticity: under cholesterol-replete conditions, it localizes to ER-plasma membrane junctions [38], whereas nutrient deprivation drives it to the ER-mitochondria interface [39]. By abrogating efferocytosis, MERTK deficiency essentially imposes a state of “pseudo-starvation” on macrophages. This perceived nutrient deficit hijacks GRAMD1A, redirecting it to funnel unesterified sterols directly into the mitochondria, thereby setting the stage for oxysterol-induced toxicity.

Once mitochondrial integrity is breached, the consequent release of mitochondrial DAMPs (mtDNA and dsRNA) must be translated into an inflammatory response. Our data position the kinase TBK1 as the master signal integrator in this cascade. By sensing these ectopic nucleic acids via the converged cGAS-STING and RIG-I/MAVS axes [18], TBK1 is massively phosphorylated in mertk^−/− macrophages. This early and robust TBK1 activation perfectly explains the subsequent, sustained hyperactivation of the NF-κB and IRF3 transcriptional programs driving colitis.

Our antibiotic depletion experiments (ABX) revealed that oxysterol accumulation, mitochondrial dysfunction, and colitis persist even in a pseudo-germ-free state, identifying the MERTK-GRAMD1A axis as a primary, macrophage-intrinsic pathogenic driver. While we acknowledge the correlative nature of 16S-based analyses, these findings suggest that the host-intrinsic metabolic defect precedes microbial shifts, although the latter may further amplify the inflammatory loop, as a feedback mechanism to be definitively resolved in future FMT-based studies. In conclusion, our work uncovers a previously unrecognized subcellular checkpoint where MERTK orchestrates mitochondrial sterol homeostasis to prevent innate immune activation. By linking defective efferocytosis to a metabolic-mitochondrial catastrophe, we identify the MERTK-Oxysterol-TBK1 axis as a pivotal rheostat of mucosal immunity and a promising therapeutic target for restoring intestinal homeostasis in Crohn's disease.

10. Conclusion

The self-regulatory MERTK mechanism maintains mitochondrial cholesterol homeostasis, preventing the pathological accumulation of cholesterol and its conversion to toxic oxysterols that would otherwise compromise oxidative phosphorylation, trigger mitochondrial nucleic acid release and ROS production, and ultimately activate TBK1-mediated inflammatory signaling in Crohn's disease.

11. Materials and methods

11.1. Mouse

Mice with a C57BL/6 background were used in this study. Mertk^−/− and Nlrp3^−/− mice were generated by C. Wu from Nanjing University. Lyz2-Cre mice were purchased from Cyagen Biosciences (Suzhou, China). All animal experiments were approved by the Institutional Animal Care and Use Committee (IACUC) of Jinling hospital (DZYJSKTB20240927004; DZYISKT202508080002). C57BL/6 mice (6-8 weeks old, male) were housed under specific-pathogen-free (SPF) conditions. Acute colitis was induced by administering 2.5% (w/v) dextran sulfate sodium (MP Biomedicals) in the drinking water ad libitum for 7 consecutive days. Control mice received regular drinking water. Mice were monitored daily for body weight, stool consistency, and rectal bleeding. On day 8, mice were euthanized, and the entire colon was excised. Colon length was measured as a primary indicator of inflammation. Distal colon segments were collected for histological analysis, protein extraction, or LPL isolation.

11.2. Antibiotic (ABX) cocktail-induced gnotobiotic mouse model establishment

Gnotobiotic mice were established by oral gavage of an antibiotic (ABX) cocktail to deplete the gut microbiota. WT or mertk^−/− mice (8–10 weeks old) were randomly divided into control and ABX-treated groups. The ABX cocktail consisted of ampicillin (1 g/L), neomycin (1 g/L), metronidazole (1 g/L), and vancomycin (0.5 g/L), dissolved in sterile drinking water. Mice in the ABX group received the ABX daily gavage for 2 weeks, while the control group received water without antibiotics. Successful gut microbiota depletion was verified by bacterial culture plating and fecal DNA detection.

11.3. Cells

RAW264.7, RAW-Lucia™ ISG-KO-STING, RAW-Lucia™ ISG-KO-TBK1 and RAW mLC3™ cells were purchased from Invivogen. THP-1, NCM460, HEK293T and iBMDMs cell lines were purchased from KeyGEN BioTECH. All cells were cultivated in a 37 °C incubator with 5% CO2. RAW, 293T or iBMDM cells were cultured in Dulbecco's modified Eagle's medium (THP-1 and NCM460 cells were cultured in1640 medium) supplemented with 10% fetal bovine serum (FBS) and 1% penicillin and streptomycin at 37 °C and 5% CO2. Cell culture dishes, round coverslips, and centrifuge tubes were obtained from NEST biotechnology. The pipette tips were obtained from KeyGEN.

12. Method details

12.1. Intestinal sample collection

The collection of human specimens that were taken during bowel resection surgery from CD patients, and the collection of non-inflamed, non-stenotic controls were approved by the Institutional Review Board at Jinling Hospital (Ethics Approval No.:2018GKJDWLS-03-032; 2026DZKY-013-01). Each participant provided written informed consent.

12.2. Western blotting

Total protein was extracted from cells or tissues using RIPA lysis buffer supplemented with protease and phosphatase inhibitor cocktails. Protein concentration was determined using the bicinchoninic acid (BCA) assay. Equal amounts of protein (20-30 μg) per sample were resolved by SDS-PAGE and transferred onto polyvinylidene difluoride membranes. Membranes were blocked with 5% non-fat milk or bovine serum albumin (BSA) in Tris-buffered saline with 0.1% Tween 20 (TBST) for 1 h at room temperature. Membranes were then incubated overnight at 4 °C with primary antibodies. Following washing with TBST, membranes were incubated with HRP-conjugated secondary antibodies for 1 h at room temperature. Protein bands were visualized using an enhanced chemiluminescence (ECL) detection kit and. Band intensities were quantified using ImageJ software (NIH) and normalized to the corresponding loading control. Detailed information of the reagents involved is provided in Supplementary Table.

12.3. Quantitative real-time PCR (RT-qPCR)

Total RNA was isolated from samples using TRIzol reagent or an RNA isolation kit according to the manufacturer's protocol. RNA concentration and purity were assessed using a NanoDrop spectrophotometer. First-strand cDNA was synthesized from 1 μg of total RNA. RT-qPCR was performed on a Applied Biosystems 7500 using SYBR. The thermal cycling conditions were: 5 °C for 10 min, followed by 40 cycles of 95 °C for 15 s and 60 °C for 1 min. The relative expression of target genes was calculated using the 2⁻ΔΔCt method and normalized to the expression of the housekeeping gene. All primer sequences are listed in Supplementary Table.

12.4. Enzyme-linked immunosorbent assay (ELISA)

The levels of IL-10,TNF-α,IFN-γ,IL-1β, IL-5,CCL2 and CCL5 in colon samples were measured by multiplex secretome analysis (XMPlex02240714 XMplex Human 12-Plex Custom panel) according to manufacturer's instructions with the assistance of SXM Biotechnology Co., Ltd. (WuHan, China). All ELISA assays for tissues were performed after the total protein concentration of the homogenate was quantified to 2 mg/ml. Briefly, samples and standards were added to pre-coated 96-well plates and incubated for 2 h at room temperature. Following washing, a biotinylated detection antibody was added, followed by incubation with streptavidin-HRP. The reaction was developed using a TMB substrate solution and stopped with a stop solution. The optical density was measured at 450 nm using a microplate reader. Analyte concentrations were calculated based on the standard curve.

12.5. Isolation of mitochondrial and endoplasmic reticulum (ER) fractions

Mitochondrial and ER fractions were isolated from cells using commercial kits, with minor modifications to the manufacturer's protocols (Beyotime Biotechnology, China; Invent Biotechnologies, USA). Briefly, harvested cells were washed with pre-cooled PBS, resuspended in kit-provided isolation buffer, and homogenized on ice. After centrifugation at 600g for 10 min (4 °C) to remove nuclei, the supernatant was centrifuged at 11,000g for 15 min (4 °C) to collect mitochondrial pellets. Pellets were washed, resuspended in lysis buffer, and stored at −80 °C. ER fractions were isolated using the ER Isolation Kit. Cells were processed as above, resuspended in ER isolation buffer with protease inhibitors, and homogenized. After centrifugation at 800g (10 min, 4 °C) and 12,000g (20 min, 4 °C) to remove debris and mitochondria, the supernatant was centrifuged at 100,000g for 60 min (4 °C) to pellet ER.

Protein concentrations of all fractions were measured by BCA assay (Beyotime Biotechnology, China) for equal loading in subsequent experiments.

12.6. Histology and immunohistochemistry

Formalin-fixed, paraffin-embedded (FFPE) mouse colon or human intestinal mucosal biopsies were sectioned at 4-5 μm thickness and mounted on charged glass slides. All staining procedures were preceded by deparaffinization in xylene and rehydration through a graded ethanol series to water.

12.7. Hematoxylin and eosin (H&E) and alcian blue staining

For morphological assessment, sections were stained with Harris's hematoxylin followed by eosin Y (H&E). To visualize acidic mucins, sections were stained with Alcian Blue solution (pH 2.5) for 30 min, followed by a counterstain with Nuclear Fast Red. After staining, all slides were dehydrated through graded ethanol, cleared in xylene, and mounted with a permanent mounting medium.

12.8. Immunofluorescence (IF) staining

Following rehydration, heat-induced epitope retrieval (HIER) was performed by incubating slides in a citrate buffer (10 mM sodium citrate, pH 6.0) at 95 °C for 20 min. Sections were then cooled, washed in PBS, and permeabilized with 0.2% Triton X-100 in PBS for 10 min. Non-specific binding was blocked using 5% normal goat serum in PBS for 1 h at room temperature. Slides were incubated overnight at 4 °C with a cocktail of primary antibodies: rat anti-mouse CD80, rabbit anti-mouse CD206, and goat anti-mouse MUC2. After washing, sections were incubated for 1 h at room temperature with species-specific, Alexa Fluor-conjugated secondary antibodies. Nuclei were counterstained with DAPI (4’,6-diamidino-2-phenylindole). Slides were mounted with ProLong Gold Antifade Mountant (Invitrogen). Detailed information of the reagents involved is provided in Supplementary Table.

12.9. Immunohistochemistry (IHC)

For MERTK detection in human tissue, FFPE sections underwent HIER as described above. Endogenous peroxidase activity was quenched by incubating slides in 3% hydrogen peroxide for 15 min. Following a blocking step with 5% normal goat serum, sections were incubated overnight at 4 °C with a primary antibody against human MERTK. Antibody binding was detected using a high-sensitivity HRP-polymer-conjugated secondary antibody detection system. The signal was visualized with 3,3′-Diaminobenzidine (DAB) substrate, resulting in a brown precipitate. Sections were counterstained with hematoxylin, dehydrated, and mounted.

12.10. Isolation of lamina propria mononuclear cells

The colon was excised, flushed with ice-cold PBS to remove luminal contents, and Peyer's patches were excised. The tissue was opened longitudinally, cut into 1-2 cm pieces, and incubated twice in HBSS (Ca²⁺/Mg²⁺-free) containing 5 mM EDTA and 1 mM DTT for 20 min at 37 °C with shaking to remove the epithelial layer. The remaining tissue fragments were minced and digested in RPMI 1640 medium supplemented with 1 mg/mL Collagenase D (Roche) and 20 μg/mL DNase I (Sigma-Aldrich) for 45 min at 37 °C with continuous agitation. The resulting cell suspension was passed through a 70 μm cell strainer. Leukocytes were enriched by density gradient centrifugation over a 40%/80% Percoll gradient at 800×g for 20 min at room temperature without brake. The leukocyte-rich interface was collected, washed, and resuspended in FACS buffer (PBS containing 2% FBS and 1 mM EDTA) for subsequent analysis. Detailed information of the reagents involved is provided in Supplementary Table.

12.11. Flow cytometry analysis of macrophage subsets

Approximately 1-2 × 10⁶ isolated LPLs per sample were incubated with an Fc receptor blocking antibody (anti-CD16, BioLegend) for 15 min on ice. Cells were then stained with a pre-titered cocktail of fluorochrome-conjugated antibodies for 30 min at 4 °C in the dark. A representative antibody panel for macrophage analysis included: CD45, CD11b, F4/80, Ly6C, CD64, MerTK, CD86, and CD206. Following surface staining, cells were washed and stained with a viability dye to exclude dead cells from analysis.

Data were acquired on a BD™ cytometer using BD FACSDiva™ software. Compensation was established using single-stained antibody-capture beads. Data were analyzed using FlowJo™ software (v10, BD). The gating strategy involved identifying live, single, CD45⁺ leukocytes, followed by gating on CD11b⁺ F4/80⁺ macrophages. Polarization and functional subsets within this population were then defined based on the differential expression of markers such as CD86 (M1) and CD206 (M2). Detailed information of the reagents involved is provided in Supplementary Table.

12.12. Plasmid and siRNA transfection

For lipid-based transfection, 293T were seeded at a density of 2 × 10⁵ cells/well in 24-well plates. Transfection was performed using Lipofectamine 3000 (Invitrogen) according to the manufacturer's protocol. Briefly, 25 pmol of siRNA or 500 ng of plasmid DNA was complexed with Lipofectamine 3000 and P3000 reagent in Opti-MEM medium. The complexes were added to cells, which were then cultured for 24-48 h before subsequent analysis.

For electroporation of primary cells, 1 × 10⁶ cells were resuspended in 100 μL of SF Cell Nucleofector™ Solution (Lonza, Basel, Switzerland) mixed with 50 pmol siRNA or 1 μg plasmid DNA. The mixture was transferred to a certified cuvette and electroporated using a pre-optimized program on the Lonza 4D-Nucleofector™ System. Cells were immediately transferred to pre-warmed culture medium and analyzed after 24-48 h. Knockdown or overexpression efficiency was confirmed by RT-qPCR or Western blotting. Detailed information of the reagents involved is provided in Supplementary Table.

12.13. Measurement of transepithelial electrical resistance (TEER)

Human intestinal epithelial cells were seeded at a density of 1 × 10⁵ cells/cm² onto 0.4 μm pore size, 12-well Transwell® polyester membrane inserts (Corning, NY, USA). TEER was measured using Millicell® ERS-2 Voltohmmeter (Merck KGaA, Darmstadt, Germany). The resistance of a blank insert was subtracted from all readings, and the resulting value was multiplied by the surface area of the membrane to yield the final TEER in Ω·cm².

12.14. Live-cell confocal microscopy

Cells were seeded on 35 mm glass-bottom dishes. For organelle visualization, cells were incubated with specific live-cell probes for 30 min at 37 °C. Imaging was performed on a Olympus IXplore SpinSR confocal microscope equipped with a 63x/1.4 NA oil-immersion objective and a stage-top environmental chamber maintaining conditions at 37 °C and 5% CO₂. Time-lapse images were acquired at specified intervals. Image processing and analysis, including colocalization or object tracking, were performed using ImageJ.

12.15. Ultrastructural analysis by transmission electron microscopy (TEM)

For TEM analysis, cells were fixed in 2.5% glutaraldehyde in 0.1 M sodium cacodylate buffer (pH 7.4) for 1 h at room temperature. After washing with cacodylate buffer, samples were post-fixed in 1% osmium tetroxide (OsO₄) with 1.5% potassium ferrocyanide in the same buffer for 1 h on ice to enhance membrane contrast. Samples were then dehydrated through a graded ethanol series and embedded in Eponate 12™ resin. The resin was polymerized at 60 °C for 48 h. Ultrathin sections (70-80 nm) were cut on a Leica UC7 ultramicrotome, collected on 200-mesh copper grids, and post-stained with uranyl acetate and lead citrate. The grids were examined and imaged on transmission electron microscope operated at 80-120 kV. Images were captured with a Gatan digital camera.

12.16. Blue native-polyacrylamide gel electrophoresis (BN-PAGE)

To preserve native protein complexes, 50 μg of protein was solubilized in NativePAGE™ Sample Buffer (Beyotime) containing 4% digitonin (Beyotime) on ice for 5 min. Following centrifugation at 20,000×g for 30 min at 4 °C, the supernatant was collected, and G-250 Sample Additive was added. The protein complexes were separated on a NativePAGE™ 4–16% Bis-Tris Gel (Beyotime) according to the manufacturer's instructions. For Western blot analysis, proteins were transferred to a PVDF membrane using a semi-dry transfer system. After transfer, the membrane was briefly destained with methanol to visualize the protein lanes, blocked with 5% non-fat milk in TBST, and incubated overnight at 4 °C with primary antibodies targeting subunits of the OXPHOS complexes. A representative antibody cocktail included: NDUFB8 (Complex I), SDHB (Complex II), UQCRC2 (Complex III), MTCO1 (Complex IV), and ATP5A (Complex V). Signal was detected using HRP-conjugated secondary antibodies and an ECL substrate. Detailed information of the reagents involved is provided in Supplementary Table.

12.17. Mitochondrial respiration assay

Real-time oxygen consumption rates (OCR) were measured using a Seahorse XFe96 Analyzer (Agilent Technologies). THP1 were seeded at an optimized density of 5 × 10⁴ cells/well onto a Seahorse XF96 cell culture microplate and allowed to adhere overnight. One day prior to the assay, the sensor cartridge was hydrated in sterile, deionized water and incubated at 37 °C in a non-CO₂ incubator.

On the day of the assay, the cell culture medium was replaced with XF Assay Medium (Agilent Seahorse XF Base Medium supplemented with 10 mM glucose, 1 mM sodium pyruvate, and 2 mM l-glutamine), adjusted to pH 7.4. The cells were then incubated at 37 °C in a non-CO₂ incubator for 1 h to allow for temperature and pH equilibration. During this time, mitochondrial respiratory modulators were loaded into the appropriate ports of the hydrated sensor cartridge. The assay protocol consisted of baseline OCR measurements followed by the sequential injection of the following compounds to achieve final working concentrations of: Oligomycin (1.5 μM), an ATP synthase inhibitor. Carbonyl cyanide-4-(trifluoromethoxy) phenylhydrazone (1.0 μM), a protonophore that uncouples the mitochondrial inner membrane.

Rotenone and Antimycin A (0.5 μM each), inhibitors of Complex I and Complex III, respectively. After the assay, OCR values were normalized to the total protein content in each well, which was determined using a bicinchoninic acid assay.

12.18. Targeted metabolomics of oxidized lipids

Lipids were extracted from cells using a modified version of the Bligh and Dyer's protocol. Lipid extract was resuspended in 500 μL of ethanol containing 5 μg of butylated hydroxytoluene (BHT). An internal standard cocktail (50 μL) comprising d7-24-hydroxcholesterol, d7-7β-hydroxycholesterol, d6-25-hydroxycholesterol, d6-27-hydroxycholesterol, d7-7-keto-cholesterol, d7-7α-hydroxy-cholestenone, d6-TMAS, d7-4β-hydroxycholesterol, d6-24,25-epoxycholesterol, d7-desmosterol, d3-3β-7α-dihydroxycholest-5-enoic acid (Avanti Polar Lipids) was added to the samples. The samples were incubated at 1200 rpm for 15 min at 4 °C. At the end of incubation, 250 μL of MilliQ water and 1 ml of n-hexane was added. The samples were mixed thoroughly by vortexing, and centrifuged at 12, 000 rpm for 5 min 4 °C. Clear upper phase containing oxysterols and sterols in hexane was transferred to new tube. The extraction was repeated once with another 1 ml of n-hexane. The pooled extract was dried in a SpeedVac under organic mode. Oxysterols were derivatised to obtain their picolinic acid esters prior to LC/MS analysis on a Shimadzu 40X3B-UPLC coupled to Sciex QTRAP 6500 Plus and quantitated.

12.19. Detection of 27-hydroxycholesterol (27-OHC) levels

The levels of 27-OHC were detected using the 27-OHC (27-Hydroxycholesterol) ELISA Kit (FineTest, EU2632) following the manufacturer's instructions. Briefly, samples and standards were added to the pre-coated microplate wells and incubated at 37 °C for 30 min. After washing, the detection reagent was added and incubated, followed by another washing step. The substrate solution was added, and the microplate was incubated in the dark at 37 °C for 15 min. The reaction was terminated with stop solution, and the absorbance was measured at 450 nm using a microplate reader. The concentration of 27-OHC in samples was calculated according to the standard curve.

12.20. 4D proteomics analysis method summary

Protein samples were subjected to reduction and alkylation, followed by enzymatic digestion with trypsin at a 1:50 enzyme-to-protein mass ratio. The digestion was carried out at 37 °C for approximately 20 h. The resulting peptides were desalted, dried under vacuum, and reconstituted in a solution of 0.1% formic acid (FA) for subsequent analysis.

Peptides were separated using a nanoElute HPLC system. The separation was performed on a Thermo Scientific EASY-Spray™ C18 column (25 cm × 75 μm ID, 1.9 μm) with a 30-min gradient at a flow rate of 300 nL/min. The mobile phases consisted of A (0.1% FA in water) and B (0.1% FA in 99.9% acetonitrile). The gradient was as follows: 5-35% B over 18 min, 35-80% B over 2 min, and holding at 80% B for 10 min.

The eluted peptides were analyzed on a timsTOF Pro mass spectrometer operating in positive ion mode. Data were acquired using the Parallel Accumulation-Serial Fragmentation (PASEF) method over a mass range of 100-1700 m/z. The ion source voltage was set to 1.5 kV. Each duty cycle (0.95 s) consisted of one full MS1 scan followed by 8 PASEF MS/MS scans of precursor ions with charge states from 0 to 5. Dynamic exclusion was enabled for 24 s to prevent re-sequencing of abundant precursors.

Raw data files were processed using MaxQuant software (version 1.6.14). Database searching was performed against the UniProt Homo sapiens database.

12.21. 16S rDNA amplification and sequencing

The luminal contents and colon samples were collected for 16S rDNA amplification and sequencing. The V3–V4 region was selected for 16S rDNA amplification, and the universal primers used were 341F (5′-CCTACGGGRSGCAGCAG-3′) and 806R (5′-GGACTACVVGGGTATCTAATC-3′). The index sequence and connector sequence suitable for Illumina Novaseq PE250 sequencing were added to the 5 ‘end of the universal primer to complete the design of the specific primer. PCR amplification using the KAPA HiFi Hotstart ReadyMix PCR kit high-fidelity enzyme ensured the accuracy of the amplification and high efficiency. PCR products were detected by 2% agarose gel electrophoresis, and the PCR products were recovered by gel-cutting with an AxyPrep DNA gel recovery kit (Axygen). Then the Illumina Novaseq PE250 was used for sequencing.

12.22. Bulk RNA-seq library preparation and analysis

Colon segments from wild-type (WT) and mertk^−/− mice were isolated. Total RNA was extracted and purified using TRIzol reagent (Invitrogen) according to the manufacturer's protocol. RNA quantity and purity were assessed using NanoDrop ND-1000 (NanoDrop), and RNA integrity was confirmed by Bioanalyzer 2100 (Agilent) with RNA Integrity Number (RIN) > 7.0, further validated by denaturing agarose gel electrophoresis. Poly(A) RNA was purified from 1 μg total RNA using Dynabeads Oligo(dT) (Thermo Fisher) through two rounds of purification, followed by fragmentation with the Magnesium RNA Fragmentation Module (NEB) at 94 °C for 5-7 min. The fragmented RNA was reverse-transcribed into cDNA using SuperScript™ II Reverse Transcriptase (Invitrogen) and subsequently used to synthesize U-labeled second-strand DNA with E. coli DNA polymerase I, RNase H (NEB), and dUTP solution (Thermo Fisher). After PCR amplification, the final cDNA library had an average insert size of 300 ± 50 bp. Sequencing was performed using 2 × 150 bp paired-end sequencing (PE150) on an Illumina Novaseq™ 6000, following the manufacturer's protocol. Library preparation was assisted by APTBIO Co., Ltd., Hangzhou. Quality control was performed using FastQC with default parameters. Adapters were trimmed using Cutadapt, retaining read pairs >30 bp. Reads were aligned to the mouse reference genome GRCm38.102 using STAR. Differentially expressed genes were identified using DESeq2. Gene Set Enrichment Analysis (GSEA) was conducted with ClusterProfiler following the default pipeline. Visualizations, including volcano plots, were generated using ggplot2. Associated statistical analyses were carried out using R (R software foundation).

12.23. Plasmid construction

Recombinant plasmids targeting mouse Gramd1a (Gene ID: 52857) were constructed using the GV718 vector (CMV-bGlobin-FLEX-EGFP-MIR155(mcs)-WPRE-hGH polyA). The Gramd1a shRNA, designed as 3 tandem copies (each: 5′-GCTGAAGGCTGTATGCTGAGGT-GAGGTGGTG-ACCTCAGCATACAGCCTT CAGC-3’; sense: GCTGAAGGCTGTATGCTGAGGT, loop: GAGGTGGTG, antisense: ACCTCAGCATACAGCCTTCAGC), was cloned into the MIR155(mcs) via NheI/MfeI sites. PCR amplification of the 459 bp shRNA-containing fragment used primers K25J0608-P1 (NheI site) and K25J0608-P2 (MfeI site). After ligation and transformation into E. coli DH5α, positive clones were screened by colony PCR (Globin-F/EGFP-P2: 828.0 bp for insert, 409.0 bp for empty vector) and verified by Sanger sequencing (confirming shRNA tandem structure and no mutations). A non-targeting control (CON545: TTCTCCGAACGTGTCACGT) was constructed identically.

12.24. AAV production & in vivo administration

Recombinant AAV9 was produced by co-transfecting 293T cells with the GV718-Gramd1a-shRNA (or CON545) plasmid, pHelper, and pRepCap (Lipofectamine 2000). Viruses were harvested at 72 h, purified via iodixanol gradient ultracentrifugation (150,000×g, 4 °C, 2 h), and concentrated (Amicon Ultra, 100 kDa). Titers (qPCR, WPRE region) were adjusted to 1 × 10¹² vg/mL. Lyz2-Cre mice received 1 × 10¹² vg/mouse via tail vein injection for myeloid-specific Gramd1a knockdown.

12.25. Statistical analysis

No animals or data points were excluded from any analyses in this study. For data analysis, unless otherwise stated below, statistical analyses were carried out using R (R software foundation) or GraphPad Prism (version 9.00; GraphPad software). For flow cytometry analyses, data processing and analysis were performed with FlowJo v10.10.0 software (BD Life Sciences). For comparing the mean of two groups, an unpaired student's t-test or a Wilcoxon test was used where appropriate. Data distribution was assumed to be normal and not formally tested, or known to deviate from a normal distribution. Differences between multiple groups were tested using ANOVA or Kruskal–Wallis tests. For testing differences in frequency, Fisher's exact test was used. Unless otherwise specified, all data are presented as the mean ± s.d. P values < 0.05 are considered statistically significant (∗P < 0.05, ∗∗P < 0.01, ∗∗∗P < 0.001, ∗∗∗∗P < 0.0001).

Data and materials availability

All unique reagents generated in this study are available from the lead contact. The raw single-cell RNA-sequence, bulk RNA sequence and 16S rRNA gene amplicon sequence data are available at the Gene Expression Omnibus under accession codes (GSE236459, GSE309427, PRJNA1359577). Raw LC/MS data for IP has been deposited to the ProteomeXchange Consortium via the iProX partner repository (PXD052743; PXD069433). Raw LC/MS data for cholesterol-targeted metabolomics are not publicly available due to limitations of LipidALL Technologies (http://www.lipidall.com/). Their rationale is that mass spectrometry files contain content that could harm their commercial interests. If necessary, additional information on raw mass spectrometry in this study (Project: 2025-064-C-01) can be requested from LipidALL by email (qmchu@lipidall.com, jywang@lipidall.com). The authors declare that the remaining data generated or analyzed during this study are available within the article, Supplementary Information, or Source Data file. Source data are provided with this paper.

Declaration of generative AI and AI-assisted technologies in the writing process

None.

Funding

Supported by Basic Research Program of Jiangsu BK20250108 and Jiangsu Provincial Medical Innovation Center CXZX202217.

CRediT authorship contribution statement

Juanhan Liu: Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Visualization, Writing – original draft, Writing – review & editing. Peizhao Liu: Formal analysis, Methodology, Software, Visualization, Writing – review & editing. Yangguang Li: Conceptualization, Formal analysis, Software, Visualization, Writing – review & editing. Tao Zheng: Data curation, Methodology, Project administration, Resources, Validation, Writing – review & editing. Cunxia Wu: Data curation, Methodology, Validation, Writing – review & editing. Wenbin Gong: Conceptualization, Formal analysis, Project administration, Validation. Ze Li: Conceptualization, Validation, Visualization. Guiwen Qu: Data curation, Visualization, Writing – review & editing. Xuanheng Li: Data curation, Formal analysis, Writing – review & editing. Yiyu Yang: Formal analysis, Visualization, Writing – review & editing. Qianwen Li: Data curation, Formal analysis, Writing – review & editing. Fansen Lin: Conceptualization, Data curation, Writing – review & editing. Jinyao Liu: Formal analysis, Visualization. Zexing Lin: Data curation, Formal analysis. Yun Zhao: Conceptualization, Data curation, Supervision, Writing – original draft, Writing – review & editing. Jianan Ren: Conceptualization, Data curation, Formal analysis, Funding acquisition, Methodology, Project administration, Software, Supervision, Validation, Writing – original draft, Writing – review & editing. Xiuwen Wu: Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Project administration, Resources, Software, Supervision, Validation, Visualization, Writing – original draft, Writing – review & editing.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix A. Supplementary data

The following are the Supplementary data to this article:

figs1.

figs2.

figs3.

figs4.

figs5.

figs6.

figs7.

figs8.

Data availability

The data involved in this study have been uploaded to public platforms, and the specific information has been provided in the manuscript.