Activation of the Reverse Cholesterol Transport Pathway: A Strategy to Ameliorate Lipid Metabolic Dyshomeostasis in the Microenvironment Following Spinal Cord Injury Authors: Xiaoxin Wang, Guina Tan, Ru Feng, Min Ni, Feng Gao, Jianjun Li Published: 2026 ABSTRACT AbstractSpinal cord injury (SCI) causes massive release of myelin‑derived lipids, driving macrophage‑to‑foam‑cell transformation that sustains inflammation and impedes repair. Although reverse cholesterol transport (RCT) mediates lipid efflux, its functional dynamics within the SCI microenvironment remain unclear. Using a mouse T10 contusion model, we examined key RCT components (ABCA1, ABCG1, and APOE) by qRT‑PCR, Western blot, and immunofluorescence. We also evaluated the LXR agonist AZ876 for its capacity to modulate RCT and promote functional recovery. We observed a pronounced imbalance in RCT after SCI. ABCA1 and APOE were persistently upregulated, but ABCA1 localized predominantly to astrocytes rather than macrophages. Notably, ABCG1 protein levels declined progressively despite increased mRNA expression, indicating a post‑transcriptional bottleneck that favors foam‑cell accumulation. AZ876 activated the LXR/RXR axis and induced long‑term transcriptomic remodeling of lipid metabolism. Early AZ876 treatment increased ABCA1 and partially restored ABCG1 protein expression, markedly reducing Oil Red O–positive foam‑cell accumulation at the lesion site. These microenvironmental improvements were associated with decreased TNF‑α levels and enhanced locomotor recovery, as evidenced by improved BMS scores and open‑field performance. The lipid-rich environment after spinal cord injury rapidly activates the initial steps of RCT, but a pronounced post-transcriptional suppression of ABCG1 impedes downstream RCT processes. Although the LXR agonist AZ876 partially overcomes this bottleneck and improves neurological outcomes, it fails to fully restore ABCG1 protein levels, indicating that transcriptional activation alone is insufficient. Future strategies should target the mechanisms of ABCG1 protein instability to fully correct lipid metabolic dysregulation.Graphical AbstractLXR Activation Mitigates Lipid-Driven Macrophage Dysfunction After Spinal Cord Injury. Spinal cord injury generates a lipid-rich microenvironment that drives macrophage foam cell formation, impairs phagocytic function, and sustains inflammation. Activation of LXR by AZ876 upregulates ABCA1 and ABCG1, enhances reverse cholesterol transport, reduces lipid accumulation, and promotes neurological recovery.Figure elements were partly created with Servier Medical Art (Servier); reproduced elements licensed under CC BY 4.0(https://creativecommons.org/licenses/by/4.0/) Supplementary InformationThe online version contains supplementary material available at https://doi.org/10.1007/s10571-026-01754-2.Keywords: Spinal cord injury, Foam cells, Reverse cholesterol transport, ABCG1, LXR agonist, Lipid metabolism Introduction Spinal cord injury (SCI) disrupts the structural and functional integrity of the spinal cord through primary and secondary mechanisms, resulting in severe loss of motor, sensory, and autonomic functions. Following the initial mechanical insult, a cascade of secondary events—including neuroinflammation, edema, vascular dysfunction, oxidative stress, and ionic imbalance—drives extensive demyelination, necrosis, and apoptosis. Consequently, the lesion site becomes densely populated with apoptotic cells and necrotic debris (Hu et al. 2023). Emerging evidence indicates that tissue fragments, particularly myelin-derived debris, act as potent inhibitors of axonal regeneration and neural stem cell differentiation (Fan et al. 2018). Therefore, timely clearance of this debris—primarily mediated by macrophages, including resident microglia and infiltrating monocyte-derived macrophages—is essential for neurological recovery (Van Broeckhoven et al. 2021). Myelin debris constitutes the majority of necrotic material at the lesion site and is highly enriched in lipids. Unlike typical biological membranes (approximately 25% cholesterol, 65% phospholipids, and 10% glycolipids), myelin displays a markedly different lipid composition, with an estimated cholesterol:phospholipid:glycolipid ratio of roughly 40%:40%:20% (O'Brien 1965). This cholesterol-rich environment can overwhelm macrophage lipid-handling capacity, leading to pathological intracellular lipid accumulation and conversion of macrophages into lipid-laden “foam cells,” analogous to those observed in atherosclerotic lesions (Ryan et al. 2022). Foam cell formation impairs phagocytic function and sustains chronic inflammation, thereby hindering neural repair (Kong et al. 2020). Pathologically, foam cell formation is associated with increased cholesterol esterification and defective cholesterol efflux mechanisms (Ouyang et al. 2023). These observations highlight cholesterol accumulation as a key driver of microenvironmental deterioration after SCI, with macrophage foaminess playing a central role. Accordingly, strategies that mitigate foam cell formation and accumulation represent promising therapeutic approaches to improve the injury microenvironment and promote tissue regeneration. In this study, we characterize the expression patterns of proteins involved in reverse cholesterol transport (RCT) following SCI. We found that ATP-binding cassette transporter A1 (ABCA1) and apolipoprotein E (APOE) are rapidly upregulated and remain elevated after injury, whereas the expression of the transporter ABCG1 is markedly insufficient. To address this metabolic bottleneck, we tested a small-molecule liver X receptor (LXR) agonist to upregulate ABCA1 and ABCG1, enhance cholesterol efflux, and reduce foam cell formation. Our results show that although upregulation of these transporters can partially reduce foam cell formation and accumulation, transcriptional activation alone is insufficient to fully prevent macrophage foaminess. Materials and Methods Animals Specific‑pathogen‑free (SPF) C57BL/6 mice were obtained from Beijing Speh‑Bio Biotechnology Co., Ltd. Mice were housed at 24–26 °C with 60–70% relative humidity and ambient noise below 60 dB, under a 12‑hour light/dark cycle, with ad libitum access to standard chow and water. Spinal Cord Injury Model Eight-week-old female C57BL/6 mice were anesthetized with isoflurane (2–3%). Following a midline incision to expose the thoracic vertebrae, a T9 laminectomy was performed to reveal the T10 spinal cord segment. A moderate contusion injury was induced using an IH-0400 impactor (Precision Systems and Instrumentation, LLC; Natick, MA, USA). A 1.3-mm diameter impactor tip was delivered at a force of 70 kdyn from a height of 6.25 mm onto the midline of the exposed spinal cord. Following the injury, the musculature and skin were sutured in layers. Post-operative care included manual bladder expression performed twice daily until the recovery of spontaneous voiding. Drug Administration The LXR agonist AZ876 (HY-18282, MCE) was first dissolved in dimethyl sulfoxide (DMSO) (HY-Y0320, MCE). This stock solution was then diluted in a 20% SBE-β-CD (HY-17031, MCE) saline solution at a ratio of 1:9 to prepare the working solution. Immediately following SCI, mice received intraperitoneal (i.p.) injections of AZ876 at a dose of 5 mg/kg (van der Hoorn et al. 2011). The injection volume was standardized to 100 μL per dose, administered once daily for a duration of one week. Behavioral Evaluations Locomotor recovery was assessed using the Basso Mouse Scale (BMS) for open-field locomotion. Evaluations were conducted immediately post-injury, at 3 days post-injury, and weekly thereafter for a period of 4 weeks. After a 2-min acclimatization period, mice were observed for 5 min in a circular arena (90 cm in diameter, enclosed by 10-cm high transparent walls) without external stimulation. Scoring was performed by two independent observers blinded to the experimental groups. To further analyze locomotor activity and anxiety- or depression-like behaviors, the Open Field Test (OFT) was performed at 4 and 8 weeks post-injury. The apparatus consisted of a polyvinyl chloride box (50 cm × 50 cm × 50 cm). A camera mounted vertically above the center of the arena monitored movement within the central and peripheral zones, with the central zone defined as 50% of the total area. Video data were processed using automated tracking software to analyze the following parameters: center-to-periphery ratio, total horizontal distance traveled, and mean horizontal velocity. BMS scoring and OFT analysis were performed using 6–8 biological replicates per group. No technical replicates apply to behavioral tests. Histology and Immunofluorescence(IF) Mice were deeply anesthetized by intraperitoneal injection of 1.25% tribromoethanol in saline (0.2 mL/10 g; HY‑B1372, MCE) and transcardially perfused with ice‑cold saline followed by 4% paraformaldehyde (PFA). Spinal cord segments containing the lesion site were harvested, post‑fixed in 4% PFA overnight at 4 °C, and cryoprotected in 15% then 30% sucrose until the tissue sank. Samples were embedded in Optimal Cutting Temperature (OCT) compound and flash‑frozen in liquid nitrogen. Longitudinal cryosections (15 μm) were cut and collected for staining. For each time point and each staining assay (IF or ORO), 3–4 biological replicates were analyzed per group. For IF, sections were washed in PBS and placed in EDTA antigen retrieval buffer (pH 8.0) for microwave-assisted antigen retrieval. After delineating the sections with a hydrophobic pen, non-specific binding was blocked with 3% bovine serum albumin (BSA) for 30 min. Sections were then incubated with primary antibodies, including: F4/80(Rat Monoclonal, MedChemExpress USA, HY-P80665, RRID:AB_3102652, 1:200), Anti-ABCG1 (Rabbit Monoclonal, Abcam, ab52617, RRID:AB_867471, 1:100), Anti-ABCA1 (Rabbit Polyclonal, Invitrogen, PA1-16,789, RRID: AB_2288917, 1:100), and Anti-GFAP (Mouse Monoclonal, Servicebio, GB15096, RRID:AB_3720197, 1:500). Subsequently, sections were incubated with the following secondary antibodies: Alexa Fluor™ 488-conjugated Goat anti-Rabbit IgG (Polyclonal, Invitrogen, A32731, RRID:AB_2633280, 1:500), Alexa Fluor™ Plus 555-conjugated Goat anti-Rat IgG (Polyclonal, Invitrogen, A48263, RRID:AB_2896332, 1:500), and Alexa Fluor™ Plus 647-conjugated Goat anti-Rabbit IgG (Polyclonal, Invitrogen, A32733, RRID:AB_2633282, 1:500). Nuclei were counterstained with DAPI, and tissue autofluorescence was quenched prior to mounting. Oil Red O (ORO) Staining Cryosections were equilibrated to room temperature and stained with freshly prepared Oil Red O (ORO) working solution (Servicebio, G1015) for 8–10 min in the dark. After staining, slides were briefly drained (~ 3 s) and differentiated by immersion in 60% isopropanol for 3–5 s, repeated twice. Sections were then rinsed in deionized water with two quick changes (≈10 s each). Stained slides were air‑dried and coverslipped using an aqueous mounting medium compatible with lipid stains. Digital Slide Scanning and Image Acquisition All sections were scanned at room temperature using Plan-Apochromat 20 × /0.50 NA and 40 × /0.75 NA objectives. The confocal pinhole was set to 1.0 Airy unit. Excitation/emission settings were as follows: DAPI 405/450 nm, Alexa Fluor 488 488/525 nm, Alexa Fluor 555 561/590 nm, and Alexa Fluor 647 640/700 nm. Fluorescence signals were collected at room temperature using standard filter sets for the respective fluorophores. Images were acquired on an LG-FS80 digital slide scanner (Servicebio, Wuhan, China). ORO-stained sections were digitized in brightfield mode using a Pannoramic MIDI slide scanner (3 dHISTECH, Budapest, Hungary). Prior to acquisition, white-balance calibration and background correction were performed to preserve color fidelity. Digital slides and representative regions were visualized and exported using Saiviewer (Servicebio, China). All quantitative image analyses were performed with AIpathwell Histology/Fluorescence Analysis Software (v2.3.1, Servicebio, China). For IF, deep-learning algorithms were used to (1) segment nuclei by DAPI staining, (2) calculate the percentage of ABCG1 + and ABCA1 + cells within the lesion area, and (3) quantify colocalization of ABCG1/ABCA1 with F4/80 + macrophages and GFAP + astrocytes. For ORO Staining, regions of interest (ROIs) were defined at the lesion epicenter and the surrounding periphery. The lipid deposition was quantified as the positive area ratio, calculated by dividing the ORO-stained area by the total tissue area within the ROI. Western Blot Total protein was extracted from spinal cord tissue or cultured cells. For tissue samples, ~ 5-mm spinal cord segments centered on the lesion were homogenized in RIPA lysis buffer supplemented with protease and phosphatase inhibitor cocktails (RIPA: cocktail: PMSF: inhibitor A: inhibitor B = 100: 2: 1: 1: 1) using a tissue grinder with 3-mm beads. Lysates were cleared by centrifugation and protein concentrations were measured with a BCA assay kit (Servicebio, GB23301) according to the manufacturer’s instructions. Samples were mixed with 5 × reducing loading buffer at a 4:1 ratio and denatured at 95 °C for 10 min. Equal amounts of protein (30–50 μg per lane) were loaded and separated by SDS–PAGE using a standardized casting system. Electrophoresis was run at a constant voltage of 200 V for ~ 30 min, until the bromophenol blue dye front reached the bottom of the gel. Proteins were transferred to methanol-activated PVDF membranes (0.45 μm) using a wet transfer system at a constant current of 300 mA for 30 min. Membranes were blocked with 5% non-fat milk in TBST for 30 min at room temperature and incubated overnight at 4 °C with the following primary antibodies: Anti-ABCG1 (Rabbit Monoclonal, Abcam, ab52617, RRID:AB_867471, 1:1000), Anti-APOE (Rabbit Polyclonal, Servicebio, GB112288, RRID:AB_3644254, 1:500), Anti-ABCA1 (Mouse Monoclonal, Abcam, ab18180, RRID:AB_444302, 1:1000), and Anti-β-Actin (Rabbit Monoclonal, Servicebio, GB15003, RRID: AB_3083699, 1:5000) as the internal loading control. After three washes with TBST, membranes were incubated with HRP-conjugated Goat Anti-Rabbit (Servicebio, GB23303, RRID: AB_2811189, 1:5000) or HRP-conjugated Goat Anti-Mouse (Servicebio, GB23301, RRID: AB_2904020,1:3000) secondary antibodies for 30 min at room temperature. Protein bands were visualized using enhanced chemiluminescence (ECL) and imaged with a digital imaging system. Original TIFF images were analyzed with AIWBwell™ (Servicebio, Wuhan, China), which performs orientation correction and band localization using deep-learning algorithms. Integrated density values for target proteins (ABCA1, ABCG1, APOE) were normalized to the corresponding β-Actin band to yield relative expression levels (Ratio = Target density / β-Actin density). For each time point, n = 3 biological replicates per group were analyzed. For each time point, 3–4 biological replicates were included per group. RNA Isolation and Quantitative Real-Time PCR (qRT-PCR) Spinal cord segments (~ 5 mm) centered on the lesion were collected and homogenized in RNAiso Plus (9108; Takara) using a low‑temperature tissue grinder (JXFSTPRP‑CL; Shanghai Jingxin). Total RNA was extracted following the manufacturer’s protocol. RNA concentration and purity were assessed with a NanoDrop spectrophotometer (Thermo Fisher Scientific). First‑strand cDNA was synthesized from 500 ng–1 μg total RNA using PrimeScript RT Master Mix (RR036A; Takara) according to the supplier’s instructions. qRT‑PCR was performed on a QuantStudio 7 Flex Real‑Time PCR System (Thermo Fisher Scientific) using TB Green Premix Ex Taq II (RR820A; Takara). Primer sequences are listed in Supplementary file1: Table S1. Relative gene expression levels were calculated using the 2−ΔΔCt method, with β-actin serving as the internal housekeeping gene and the sham-operated group as the reference control. For each time point, n = 3 biological replicates per group were analyzed. Melt‑curve analysis was performed to confirm amplification specificity. RNA Sequencing and Bioinformatic Analysis Transcriptome sequencing (mRNA-seq) was performed by Servicebio Technology (Wuhan, China) using the DNBSEQ-T7RS platform. Total RNA was extracted from spinal cord tissue samples, and mRNA was purified using oligo (dT) magnetic beads. The purified mRNA was fragmented and used as a template for first-strand cDNA synthesis using random oligonucleotides and M-MuLV reverse transcriptase. After RNA strand degradation by RNase H, second-strand cDNA was synthesized using DNA polymerase I and dNTPs. The resulting double-stranded cDNA underwent end-repair, A-tailing, and ligation to sequencing adapters. Fragments of approximately 200 bp were selected using AMPure XP beads, followed by PCR amplification and purification to generate the final sequencing library. Library concentration was precisely quantified by Kapa qPCR, and fragment size was verified using an Agilent 4200 TapeStation. Qualified libraries were sequenced on the Illumina PE150 platform. Raw data were processed using the open-source software fastp (0.23.0) (Chen et al. 2018; Chen 2023) to remove adapter sequences and filter out low-quality reads (reads with > 5 bp of N bases, or those where > 40% of bases had a quality score(Q ≤ 20). Sliding window quality trimming was performed (window size 4 bp, mean quality < 20). Clean reads were aligned to the reference genome using HISAT2 (2.1.0) (Kim et al. 2015). Differentially expressed genes (DEGs) were identified using the DESeq2 package, with significance thresholds set at |log2 (FoldChange)|> 1 and a Benjamini–Hochberg adjusted P-value (P.adjust) < 0.05. Functional enrichment analyses, including Gene Ontology (GO), Kyoto Encyclopedia of Genes and Genomes (KEGG), and Reactome pathways, were performed on the DEGs using the clusterProfiler package. Pathways with a P.adjust < 0.05 were considered statistically significant (Wu et al. 2021). RNA-seq was performed using 3 biological replicates per group. All raw sequencing data have been deposited in the NCBI Gene Expression Omnibus (GEO) under accession number GSE328020. Statistics All statistical analyses were performed using SPSS version 27, and all graphs were generated using GraphPad Prism. Data distribution was assessed using the Shapiro–Wilk test for normality, and homogeneity of variance was evaluated using Levene’s test. For normally distributed data with equal variances, two‑group comparisons were conducted using independent‑samples t‑tests, and multiple‑group comparisons were analyzed using one‑way ANOVA. When ANOVA indicated significant differences, Bonferroni‑corrected post‑hoc tests were applied for multiple comparisons. For datasets that did not meet normality or variance homogeneity assumptions, non‑parametric tests were used (Mann–Whitney U test for two groups; Kruskal–Wallis test followed by Dunn’s post‑hoc test for multiple groups). For experiments with n = 3, only descriptive statistics (Mean ± SD) are reported without inferential statistical testing. Animals were randomly assigned to experimental groups using a computer‑generated randomization list. All behavioral scoring, histological quantification, and Western blot densitometry were performed by investigators blinded to group allocation. Statistical analyses were conducted after data collection was completed, and the investigators performing the analyses were also blinded to group identity. Detailed results of all statistical tests are provided in Supplementary file2: Results of Statistical Tests. Result Spatial and Temporal Patterns of Foam Cell Accumulation Post-SCI To define the regional and temporal progression of foam-cell formation after spinal cord injury, we examined lesion-centered spinal cord segments at acute and chronic stages (3 days, 1 week, 2 weeks, and 4 weeks post-injury) using ORO staining to visualize neutral lipid deposition. Foam cells were detectable within the lesion as early as 3 days post-injury (Fig. 1A). By 1 week, abundant foam-cell accumulation was evident mainly at the lesion border, whereas the lesion core displayed a relatively cavity-like appearance with few foam cells. Lipid deposition intensified over time in both the lesion epicenter and surrounding borders, peaking at 2 weeks and remaining elevated through 4 weeks (Fig. 1A–B). These results indicate a dynamic, region-specific pattern of lipid accumulation and foam-cell persistence following SCI. Spatiotemporal patterns of lipid deposition after SCI and the effects of AZ876. A Representative ORO staining images of the lesion core and lesion border at 3 d, 7 d, 2 w, and 4 w post‑SCI. (B1–B2) Quantification of ORO‑positive area (%) in the lesion core (B1) and lesion border (B2) in the CON group. (C1–C2) Comparison of ORO‑positive area between the AZ876 and CON groups in the lesion core (C1) and lesion border (C2). Sample size: CON: n = 4 per time point; AZ876: n = 4 per time point (biological replicates). Data are presented as Mean ± SD. Statistical significance: ns, not significant; *P < 0.05; **P < 0.01; ***P < 0.001. Scale bars: 50 μm (high‑magnification insets) Spatiotemporal Expression Patterns of Cholesterol Transporters Following SCI RCT represents a critical pathway for the clearance of excessive intracellular cholesterol. Within this process, ABCA1 facilitates the initial efflux of cholesterol to lipid-poor apolipoproteins to generate nascent (pre-β) HDL, serving as the rate-limiting step. Subsequently, ABCG1 primarily promotes cholesterol efflux to lipidated or mature HDL particles (Wang and Westerterp 2020). In the central nervous system (CNS), ApoE typically serves as the primary protein carrier for HDL-like particles (Cantuti-Castelvetri et al. 2018). We characterized the expression profile of ApoE following SCI using qRT-PCR and Western Blot. The results revealed that both the transcriptional levels and protein expression of ApoE were obviously upregulated at 1 week post-injury compared to pre-injury levels. This elevated expression was sustained for at least 4 weeks (Fig. 2 A, B1, C1). Expression patterns of APOE in the CON and AZ876 groups after SCI. A Western blot analysis of APOE protein expression at indicated time points. (B1–B2) Quantification of APOE protein levels relative to the Sham group (B1) and comparison between AZ876 and CON groups (B2). (C1–C2) qRT-PCR analysis of APOE mRNA expression relative to the Sham group (C1) and comparison between groups (C2) Sample size: n = 3–4 biological replicates per group per time point. Data are presented as Mean ± SD. Statistical significance: ns, not significant; *P < 0.05; **P < 0.01; ***P < 0.001 ABCA1 and ABCG1 are key transporters for cellular cholesterol efflux. We profiled their expression at 3, 7, 14, and 28 days post‑injury by IF, qRT‑PCR, and Western blot. ABCA1 was persistently upregulated throughout the observation period after T10 contusion (Fig. 3). Robust ABCA1 protein expression was evident in both the lesion core and border as early as 3 days post‑injury and remained elevated through 4 weeks. Intriguingly, co‑localization analysis revealed that ABCA1 was not predominantly expressed by macrophages in the early phase. At 1 week post‑injury—when macrophages accumulated in the ring‑like spared tissue at the lesion border—the F4/80/ABCA1 co‑localization ratio was low (23.23 ± 7.55%). Even at 4 weeks, roughly 40% of macrophages remained ABCA1‑negative (Fig. 3D). In contrast, ABCA1 expression in GFAP + astrocytes increased from 1 week onward in adjacent uninjured tissue, reaching approximately 80% co‑localization by 4 weeks (Fig. 3E). Spatiotemporal expression patterns of ABCA1 following spinal cord injury. A–C Representative confocal immunofluorescence images of the CON group at 3 d (A), 1 w (B), and 4 w (C) post‑injury, showing co‑staining of ABCA1 (yellow), GFAP (green), F4/80 (red), and DAPI (blue) within the lesion core, lesion border, and spared tissue. D Quantification of the proportion of F4/80⁺ABCA1⁺ cells among total F4/80⁺ macrophages in the lesion border at 1 w and 4 w post‑injury (1 w: n = 4; 4 w: n = 3 biological replicates). E Quantification of the proportion of GFAP⁺ABCA1⁺ cells among total GFAP⁺ astrocytes in the spared tissue at 1 w and 4 w post‑injury (1 w: n = 4; 4 w: n = 3 biological replicates). Scale bars: 200 μm (panoramic views) and 20 μm (high‑magnification insets) Consistent with these histological findings, qRT‑PCR and Western blot confirmed that overall ABCA1 levels in injured spinal cord tissue were higher than in sham controls at all time points examined (Fig. 4). Together, these data indicate that astrocytes are major contributors to ABCA1 upregulation after SCI and that the initial, ABCA1‑dependent step of RCT remains active post‑injury. Expression patterns and differential levels of ABCA1 in the AZ876 and CON groups after spinal cord injury. A Western blot analysis of ABCA1 protein expression at 3, 7, 14, and 28 days post-injury in the CON and AZ876 groups, with the Sham group included as baseline control. B Quantification of ABCA1 protein levels relative to the Sham group at each time point. (C1–C2) qRT-PCR analysis of ABCA1 mRNA expression at the indicated time points, showing relative expression compared with the Sham group (C1) and comparison between the CON and AZ876 groups (C2) Sample size: Sham n = 6; CON n = 3; AZ876 n = 3 per time point. Data are presented as Mean ± SD. Statistical significance: ns, not significant In contrast to ABCA1, ABCG1 protein showed a marked and sustained decline following T10 contusion. Western blotting revealed a observably reduction in ABCG1 protein as early as 3 days post‑injury relative to sham, with suppression persisting through 4 weeks (Fig. 5A, B1). Immunofluorescence corroborated these results: although ABCG1 was detectable within the lesion at 3 days, its signal fell sharply from 1 week onward and remained at a low steady state thereafter (Fig. 5A, B1). Co‑localization resluts indicated that the vast majority of both F4/80 + macrophages and GFAP + astrocytes lacked detectable ABCG1 protein across the post‑injury time course (Figs.6,7). The divergent patterns of F4/80 co‑localization with ABCA1 versus ABCG1 point to the presence of macrophage subpopulations with distinct lipid‑processing capacities or functional states within the lesion microenvironment. Expression patterns and differential levels of ABCG1 in the AZ876 and CON groups after spinal cord injury. A Western blot analysis of ABCG1 protein expression at the indicated time points (Sham, 3 d, 7 d, 14 d, and 28 d) in the CON and AZ876 groups. (B1–B2) Quantification of ABCG1 protein levels. B1 Relative ABCG1 protein expression in the CON group compared with the Sham group across time points. B2 Comparison of ABCG1 protein levels between the AZ876 and CON groups at each time point. (C1–C2) qRT-PCR analysis of ABCG1 mRNA expression. C1 Relative ABCG1 mRNA levels in the CON group compared with the Sham group. C2 Comparison of ABCG1 mRNA expression between the AZ876 and CON groups across time points. Sample size: n = 3–4 biological replicates per group per time point. Data are presented as Mean ± SD. Statistical significance: ns, not significant; *P < 0.05; **P < 0.01; ***P < 0.001 Representative immunofluorescence co-staining of ABCG1 at 3 d and 1 w post-SCI in the AZ876 and CON groups. Representative confocal images showing the co-localization of ABCG1 (purple), GFAP (green), F4/80 (red), and DAPI (blue) within the lesion core, lesion border, and spared tissue at 3 d and 1 w post-injury. Sample size: n = 4 biological replicates per group at 3 d; n = 4 biological replicates per group at 1 w. Scale bars: 200 μm (panoramic views) and 20 μm (high-magnification insets) Immunofluorescence co-staining and quantitative analysis of ABCG1 at 4 weeks post-SCI in the AZ876 and CON groups. A Representative confocal images showing ABCG1 (purple), GFAP (green), F4/80 (red), and DAPI (blue) in the lesion core, lesion border, and spared tissue at 4 w post‑injury. (B1–B2) Quantification of ABCG1⁺ cells in the lesion core (B1) and lesion border (B2) at 3 d, 1 w, and 4 w post‑injury. Sample size: CON: n = 4 (3 d), n = 4 (1 w), n = 3 (4 w) biological replicates; AZ876: n = 4 biological replicates per time point. Data are presented as Mean ± SD. Statistical significance: ns, not significant or no statistical analysis was performed; *P < 0.05; **P < 0.01; ***P < 0.001. Scale bars: 200 μm (panoramic views) and 20 μm (high‑magnification insets) Strikingly, ABCG1 mRNA was not reduced; qRT‑PCR showed upregulation at 1 and 2 weeks post‑injury compared with sham (Fig. 5C1). The discordance between elevated transcript levels and diminished protein abundance strongly implicates post‑transcriptional regulation—for example, translational inhibition, accelerated proteolytic degradation, or other mechanisms affecting protein stability—as the primary driver of ABCG1 loss after SCI. LXR Agonist AZ876 Induces Long-Term Lipid Metabolic Remodeling in the Injured Spinal Cord The LXR is a pivotal nuclear receptor that plays a central role in the regulation of lipid metabolism (Wang and Tontonoz 2018). Upon activation, LXR forms an obligate heterodimer with the Retinoid X Receptor (RXR). The resulting LXR/RXR complex subsequently binds to the promoter regions of the Abca1 and Abcg1 genes to induce their transcription (Kennedy et al. 2005; Tang et al. 2004). Furthermore, LXR has been shown to upregulate the expression of ApoE (Mak et al. 2002). Based on this mechanism, mice received daily intraperitoneal injections of the lipophilic LXR agonist AZ876 for 7 days immediately after injury to limit foam‑cell formation (van der Hoorn et al. 2011). Transcriptomic profiling at 6 weeks post-injury revealed that AZ876 treatment significantly modulated pathways related to lipid metabolism and immune response (Fig. 8). KEGG enrichment analysis identified significant upregulation of the Lipid and atherosclerosis pathway (FDR‑adjusted P = 0.036), encompassing lipid metabolism regulation, scavenger receptor–mediated uptake, inflammatory signaling, and endothelial activation. Reactome analysis revealed significant enrichment of Binding and Uptake of Ligands by Scavenger Receptors (FDR‑adjusted P = 0.029), indicating enhanced ligand recognition and internalization capacity. Together, these data indicate that AZ876 activates the LXR/RXR axis in the injured spinal cord, remodeling transcriptional programs that promote macrophage-mediated clearance and lipid processing and thereby shifting the lesion microenvironment toward improved lipid homeostasis and immune modulation. Differential expression and functional enrichment of bulk transcriptomic data from the AZ876 and CON groups at 6 weeks post-SCI. A Volcano plot of differentially expressed genes (DEGs), illustrating the distribution of upregulated and downregulated genes in the AZ876 group compared to the CON group at 6 weeks post-injury. B Reactome pathway enrichment analysis (upregulated genes), showing significant enrichment in pathways related to scavenger receptor ligand binding and uptake. C Gene Ontology (GO) functional enrichment analysis (upregulated genes), categorized by Biological Process (BP), Cellular Component (CC), and Molecular Function (MF). D KEGG pathway enrichment analysis (upregulated genes), highlighting significant enrichment in pathways associated with lipid metabolism and atherosclerosis. Sample size: n = 3 biological replicates per group. E qRT PCR analysis of To further validate the transcriptomic signatures, we examined the expression of two representative differentially expressed genes (DEGs) associated with lipid transport and neuroinflammation. Specifically, we focused on Apol11b and Apol7c, members of the Apolipoprotein L (APOL) family. Functional annotations and public database records indicate that Apol11b and Apol7c possess features consistent with lipid‑binding/transport activity and show associations with immune‑related processes. Bulk RNA-sequencing at 6 weeks post-injury revealed significant upregulation of Apol11b (FDR-adjusted P = 0.0012) and Apol7c (FDR-adjusted P = 0.0042) following AZ876 treatment. Independent qRT-PCR validation at 7 days post-injury confirmed that the mRNA levels of these target genes were elevated in the AZ876 group compared with the control group (Fig. 8E, F). The concordance between the early-stage qRT-PCR validation and chronic-phase RNA-seq data suggests that AZ876 intervention within the first week post-injury initiates a sustained remodeling of the lipid metabolic network, thereby enhancing the capacity of LXR agonists to drive the long-term stabilization of the lesion microenvironment. AZ876 effects on ABCA1 and ABCG1 expression ORO staining showed that, beginning at 2 weeks post‑injury, AZ876 treatment significantly reduced foam‑cell accumulation in both the lesion core and spared tissue relative to controls (Fig. 1A, C1–C2). At the level of reverse cholesterol transport, AZ876 induced an early increase in ABCA1: ABCA1 protein within the lesion core was elevated by day 3, and ABCA1 mRNA was increased at 1 week, with a general trend toward higher expression across core, border, and spared tissue during the chronic phase (Fig. 9). Spatiotemporal expression patterns of ABCA1 following spinal cord injury in the AZ876 group. A–C Representative confocal immunofluorescence images of ABCA1 (yellow), GFAP (green), F4/80 (red), and DAPI (blue) in the lesion core, lesion border, and spared tissue at 3 d (A), 1 w (C), and 4 w (B) post‑injury. (D1–D3) Quantification of ABCA1⁺ cells relative to total cells in the lesion core (D1), lesion border (D2), and spared tissue (D3), comparing CON and AZ876 groups. Sample size: CON: n = 4 (3 d), n = 4 (1 w), n = 3 (4 w) biological replicates. AZ876: n = 4 biological replicates per time point. Data are presented as Mean ± SD. Statistical significance: ns, not significant or no statistical analysis was performed; *P < 0.05; **P < 0.01; ***P < 0.001. Scale bars: 200 μm (panoramic views) and 20 μm (high‑magnification insets) ABCG1 exhibited a transient response to AZ876: both ABCG1 mRNA and protein were increased at 1 week post‑injury (Fig. 5B2, C2), but this effect was not maintained long term, and by 4 weeks ABCG1 protein levels in the AZ876 group were no longer significantly different from controls (Fig. 7A, B1–B2). Assessing the effects of qRT-PCR and Western blot, AZ876 treatment did not affect APOE expression. (Fig. 2B2, C2). These findings indicate that AZ876 activates LXR‑dependent lipid‑handling pathways and promotes early cholesterol efflux and foam‑cell clearance, whereas the failure to sustain ABCG1 protein expression may limit prolonged enhancement of reverse cholesterol transport. Impact of AZ876 on Inflammation and Neurological Recovery After SCI AZ876 treatment for 1 week attenuated post‑injury inflammation and improved functional recovery. At 7 days post‑injury, mice receiving AZ876 exhibited reduced inflammatory markers compared with controls (Fig. 10C). By 4 weeks post‑injury, AZ876‑treated animals showed significantly higher BMS scores than controls, indicating enhanced hindlimb locomotor recovery (Fig. 10B2). In the OFT at 4 weeks, AZ876‑treated mice demonstrated increased center‑to‑periphery ratio, mean velocity, and total distance traveled relative to controls (Fig. 10A), consistent with reduced anxiety‑like behavior and improved overall locomotor capacity. These results suggest that AZ876 treatment effectively reduces anxiety-like behavior and enhances overall locomotor capacity in mice following spinal cord injury. AZ876 improves locomotor function and reduces early inflammatory responses following SCI. A OFT performance at 4 w and 8w post‑injury, including percentage of time spent in the center, total distance traveled, and mean velocity. B BMS scores showing hindlimb locomotor recovery from 1d to 4 w post‑injury. C qRT‑PCR analysis of TNF‑α mRNA expression in injured spinal cord tissue. Sample size: OFT: Sham n = 8; CON n = 8 (4 w), n = 6 (8w); AZ876 n = 8. BMS: Sham n = 8; CON n = 9; AZ876 n = 9; qRT‑PCR: n = 3 biological replicates per group per time point. Data are presented as Mean ± SD. Statistical significance: ns, not significant; *P < 0.05; **P < 0.01; ***P < 0.001 Discussion Following SCI, the lesion microenvironment undergoes profound changes characterized by extensive release of myelin debris and lipids. Foam‑cell formation and persistent lipid sequestration are major contributors to chronic inflammation and impaired tissue repair (Peng et al. 2025). In our model, foam cells were already present at 1 week post‑injury and were concentrated primarily at the perilesional border, whereas the lesion epicenter exhibited a relatively “hollowed‑out” appearance. This spatial heterogeneity likely reflects severe necrosis and extracellular matrix disruption at the epicenter, which restricts early macrophage infiltration and redirects phagocytic activity to the more accessible border regions. Over time, lipid deposition intensified in both core and border regions, indicating that foam‑cell formation is a progressive process driven by the extent of tissue damage and the temporal sequence of cellular infiltration. RCT is the canonical pathway for intracellular lipid efflux. Consistent with prior reports (Zheng et al. 2023; Yao et al. 2022), we observed sustained upregulation of ApoE and ABCA1 at both the mRNA and protein levels after SCI, indicating activation of the initial steps of RCT following trauma. Notably, immunohistochemical co‑localization revealed that ABCA1 expression during the early post‑injury phase was predominantly astrocytic rather than macrophage‑associated. Even at 1 week, only a minority of macrophages expressed ABCA1, and approximately half of the macrophage population remained ABCA1‑negative by 4 weeks. These data identify astrocytes as principal mediators of the early ABCA1‑dependent efflux response and reveal substantial metabolic heterogeneity among macrophage subsets within the lesion. In stark contrast to ABCA1, ABCG1 protein levels declined progressively after injury despite significant transcriptional upregulation at 1–2 weeks. The discordance between ABCG1 mRNA and protein strongly implicates post-transcriptional regulation—such as translational repression or accelerated proteasomal degradation—as the cause of ABCG1 deficiency. Because ABCG1 primarily mediates cholesterol transfer to mature HDL-like particles, its sustained loss would be expected to impair intracellular lipid trafficking and favor foam-cell transformation. Thus, the RCT pathway after SCI appears functionally imbalanced: the initial ABCA1-dependent efflux step is activated, whereas the downstream ABCG1-dependent step is compromised, providing a plausible molecular basis for long-term lipid sequestration. We tested whether pharmacologic activation of the LXR/RXR axis could correct this imbalance. A one‑week course of the lipophilic LXR agonist AZ876 produced durable transcriptomic remodeling at 6 weeks post‑injury, with enrichment of lipid‑metabolism and scavenger‑receptor pathways. At the protein level, AZ876 further increased ABCA1 expression and transiently elevated ABCG1 mRNA and protein at 1 week, although ABCG1 protein remained below baseline and the effect was not sustained into the chronic phase. Importantly, AZ876 treatment reduced foam‑cell accumulation from 2 weeks onward and improved behavioral outcomes (BMS and OFT) at 4 weeks. These findings indicate that early LXR activation can enhance lipid handling and macrophage‑mediated clearance sufficiently to lower the local lipid burden and promote functional recovery, even if ABCG1 protein restoration is incomplete. These findings indicate that early LXR activation can enhance lipid handling and macrophage‑mediated clearance sufficiently to lower the local lipid burden and promote functional recovery, even if ABCG1 protein restoration is incomplete. Several mechanistic and translational implications arise from these results. First, astrocyte‑driven ABCA1 upregulation may represent an endogenous compensatory mechanism for early cholesterol efflux that could be therapeutically amplified. Second, the post‑transcriptional loss of ABCG1 identifies a critical bottleneck in RCT that must be addressed to achieve sustained restoration of intracellular lipid trafficking. Third, short‑term LXR activation can reprogram lesion metabolism and immune responses in a manner that reduces foam‑cell formation and supports recovery, but strategies that maintain ABCG1 protein stability or provide prolonged, targeted delivery may be required for durable correction of RCT. This study has several limitations. The mechanisms underlying the mRNA–protein dissociation for ABCG1 were not defined; future work should examine translational control, ubiquitin‑proteasome activity, and microRNA‑mediated silencing. The transient nature of ABCG1 induction by AZ876 suggests that alternative dosing regimens, targeted delivery systems, or combination therapies may be necessary to sustain protein expression without systemic adverse effects. Furthermore, while our confocal imaging and bioinformatics analysis point to astrocytes as a key source of ABCA1, we recognize the inherent resolution limits of immunofluorescence in distinguishing tightly associated cell populations or detecting low-abundance transcripts. Specifically, the cell‑type–specific contributions of astrocytic ABCA1 and the metabolic heterogeneity of macrophage subsets should be further validated using complementary high-resolution approaches, such as Fluorescence-Activated Cell Sorting, RNAscope, or conditional knockouts and lineage tracing. Consequently, our current conclusions regarding the cell-specific metabolic roles should be interpreted as identifying the predominant cell populations involved within the observed spatiotemporal context of spinal cord injury. Conclusion In the injured spinal cord, key RCT proteins ABCA1 and APOE remain persistently upregulated, whereas ABCG1 is subject to post‑transcriptional suppression, resulting in reduced protein levels. Our data indicate that astrocytes play a major role in the initial steps of RCT, and that distinct macrophage subtypes exhibit marked differences in their lipid‑metabolic responses to overload. Although the LXR agonist AZ876 partially relieves the ABCG1 bottleneck and improves neurological outcomes, it does not fully restore ABCG1 protein expression, indicating that transcriptional activation alone is insufficient to correct lipid dyshomeostasis. Future studies should focus on elucidating the molecular mechanisms underlying ABCG1 protein instability and on characterizing the functional heterogeneity of macrophage subtypes under lipid overload, with the aim of fully restoring lipid homeostasis at the lesion site. Supplementary Information Below is the link to the electronic supplementary material. Supplementary file1 (DOCX 16 KB) Supplementary file2 (XLSX 13 KB) Supplementary file3 (ZIP 101878 KB) Author Contributions All authors contributed to the study conception and design. Material preparation and data collection were performed by *Xiaoxin Wang, Guina Tan, Feng Ru,* and *Min Ni.* Data analysis and interpretation were conducted by *Xiaoxin Wang and Guina Tan.* The first draft of the manuscript was written by *Xiaoxin Wang. Jianjun Li* and *Feng Gao* supervised the study and critically revised the manuscript for important intellectual content. All authors commented on previous versions of the manuscript, and all authors read and approved the final manuscript. Funding This work was funded by The Fundamental Research Funds for Central Public Welfare Research Institutes (CRSI2024CZ-7) and Researchers at the China Rehabilitation Research Center initiated a scientific research project (2025YZDW-04). Data Availability No datasets were generated or analysed during the current study. Declarations Conflict of interest The authors declare no competing interests. Ethical Approval All animal procedures were performed in strict accordance with the protocols approved by the Institutional Animal Care and Use Committee (IACUC) of Capital Medical University, Beijing, China (Ethical Approval No.: AEEI-2026–183). Generative AI and AI-Assisted Technologies in the Manuscript Preparation Process During the preparation of this work the author(s) used Gemini 3.1 in order to assist with English expression, translation, and language refinement. After using this tool, the author(s) reviewed and edited the content as needed and take full responsibility for the content of the published article. Footnotes Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. References Associated Data Data Availability Statement No datasets were generated or analysed during the current study.