Super-enhancer-driven recruitment of C/EBPβ by SULT1B1 is implicated in metabolic dysfunction-associated steatotic liver disease progression

Abstract

Background

Metabolic dysfunction-associated steatotic liver disease (MASLD) disrupts core hepatic physiological functions, with super-enhancers (SEs) playing a pivotal role in orchestrating the expression of genes associated with disease pathogenesis. This study aimed to elucidate the regulatory mechanisms of SEs in MASLD pathogenesis.

Methods

We conducted genome-wide H3K27ac profiling and Rank Ordering of SEs (ROSE) in a high-fat diet (HFD)-induced rat model to characterize histone modification and SE reprograming. Integrative analysis of ChIP-Seq and RNA-Seq, combined with transcription factor (TF) binding analysis and siRNA knockdown, was performed to investigate the regulatory mechanisms of SEs, while single-cell RNA sequencing analysis revealed cell-type-specific expression patterns. In vitro experiments further evaluated JQ1-mediated SE inhibition using human and rat hepatocyte cell lines.

Results

We observed significant H3K27ac remodeling and transcriptional reprogramming in both MASLD patients and HFD-induced rat models, highlighting epigenetic dysregulation in MASLD progression. Nineteen differential active SEs were identified in the rat model, with SULT1B1 emerging as a core SE-associated gene. Mechanistic analyses revealed that the TF C/EBPβ promotes SULT1B1 transcription through H3K27ac modification. Single-cell analysis further localized this regulatory axis specifically to hepatocytes. Functionally, targeted inhibition of the SE suppressed SULT1B1 expression and significantly mitigated lipid accumulation in both human and rat hepatocytes, supporting its pathogenic role in MASLD.

Conclusions

Our study establishes a mechanistic link between epigenetic-driven SULT1B1 overexpression and MASLD pathogenesis, highlighting SE mediated H3K27ac/C/EBPβ/SULT1B1 axis emerges as a critical regulatory pathway in MASLD, which may offer new therapeutic targets and strategies for treating metabolic liver diseases.

Supplementary Information

The online version contains supplementary material available at 10.1186/s13148-026-02121-0.

Background

Metabolic dysfunction-associated steatotic liver disease (MASLD) is a chronic liver condition characterized by excessive lipid accumulation in hepatocytes, closely associated with a spectrum of metabolic disorders [1]. Epidemiological studies indicate that the global prevalence of MASLD has reached 25%, posing a significant public health challenge [2]. However, current MASLD treatments focus primarily on antioxidants or interventions that improve glucose and lipid metabolism, yet therapeutic outcomes remain suboptimal [3, 4]. An increasing number of studies have shown that MASLD not only induces insulin resistance, lipid degeneration, and inflammatory responses, but also causes changes in epigenetic remodeling [5, 6]. Therefore, there is an urgent need for a systematic and comprehensive analysis of the pathogenesis of MASLD, especially the role of epigenetic remodeling, which could provide more effective strategies for the prevention and treatment of MASLD.

Super-enhancers (SEs), defined as densely clustered enhancer arrays (typically spaced < 12.5 kb apart), are pivotal epigenetic regulators that orchestrate transcriptional dysregulation in metabolic diseases. SEs leverage H3K27ac-mediated chromatin opening to concentrate transcription factors (TFs), the mediator (MED) complex, and RNA polymerase II (Pol II), thereby exceeding the regulatory potency of typical enhancers [7, 8]. Numerous studies have documented the critical roles of SEs in various pathologies, including lung adenocarcinoma, neuroblastoma, and hepatocellular carcinoma (HCC) [9–11]. In HCC cells, SE-driven long noncoding RNAs (lncRNAs) contribute to epithelial-mesenchymal transition and cancer cell metastasis [9]. However, the precise molecular mechanisms underlying SEs’ contribution to MASLD progression remain unclear. In this study, we performed genome-wide H3K27ac profiling and transcriptome analysis to identify differential SEs and their regulated genes. We observed extensive SE remodeling in human MASLD and experimental models, establishing SEs as critical drivers of pathogenesis. Notably, SULT1B1 (Sulfotransferase family 1B member 1) emerged as a core SE-associated gene in MASLD, which is driven by TF C/EBPβ (CCAAT enhancer binding protein beta) through H3K27ac modification. Collectively, this study establishes a direct mechanistic link between SE-driven epigenetic changes and MASLD progression, specifically through the H3K27ac/C/EBPβ/SULT1B1 axis, thereby providing a novel mechanistic foundation for MASLD prevention and treatment.

Methods

Animals models and treatment

Twenty-four 6-week-old male Sprague-Dawley (SD) rats were housed under specific pathogen-free (SPF) conditions with a 12-hour light/dark cycle at 24 ± 2 °C and 50%±10% relative humidity. The rats were randomly divided into two groups: the HFD group received a high-fat diet (20.2% protein, 45.4% fat, 34.5% carbohydrates; XTHF45-1, Xietong Pharma, China) for 8 weeks, while the control group (CTR) was maintained on a normal diet (20.6% protein, 12% fat, 67.4% carbohydrates; SWS9102, Xietong Pharma, China). At the end of the experimental period, the liver tissues and plasma samples were harvested for further analyses. Phenotype indices including body weight, liver weight, alanine aminotransferase (ALT), aspartate aminotransferase (AST), total cholesterol (TC), triglyceride (TG), low-density lipoprotein cholesterol (LDL-C), and high-density lipoprotein cholesterol (HDL-C) were measured. Serum levels were analyzed with a HITACHI Automatic Analyzer (3100, Tokyo, Japan) according to the manufacturer’s protocols.

Cell culture and cell treatment

The rat hepatocyte cell line BRL-3 A was obtained from the Type Culture Collection of the Chinese Academy of Sciences (Beijing, China). The human hepatocyte cell line HepG2 was purchased from the American Type Culture Collection (ATCC, Rockville, MD). Cells were cultured in Dulbecco’s Modified Eagle Medium (DMEM) supplemented with 10% fetal bovine serum (FBS) in a humidified incubator with 5% CO₂ at 37 °C. To establish an in vitro model of lipid accumulation, 0.3 mmol/L oleic acid (OA) was added to the culture medium of BRL-3 A and HepG2 cells. In the OA+JQ1 group, 10 µmol/L of the H3K27ac inhibitor JQ1 (MedChemExpress, NJ, USA) was added to the culture. To evaluate intracellular lipid content, cells were stained with Oil Red O (ORO) as previously described [12].

Histological analysis and staining

All histological analyses were performed by Servicebio Technology (Wuhan, China). Briefly, Hematoxylin and Eosin (HE) staining was performed on paraffin-embedded sections, while ORO staining was applied to frozen sections to visualize lipid accumulation. Images of stained sections were captured by ZEISS AXIO fluorescence microscope.

ChIP-Seq and analysis of differentially H3K27-acetylated peak

ChIP-Seq was performed using the SimpleChIP^® Plus Enzymatic Chromatin IP Kit (Magnetic Beads) (CST, USA), following standard protocols. The ChIP and input library were constructed following the manufacturer’s protocol and sequenced using the HiSeq 2500 platform (Illumina) by Novogene (United States). Sequencing reads were aligned to the Rattus norvegicus reference genome (version 6.0) utilizing the Burrows-Wheeler Aligner (BWA) [13]. Subsequently, the H3K27ac peak regions were identified using Model-based Analysis for ChIP-Seq version 2.1.0 (MACS 2.1.0) with a q-value threshold of 1e-5 [14]. Bedtools (version 2.27.0) was utilized to merge the output peak files into a single group and skip short segments with small gaps, merging those with more than 1 kb distance (merged regions were considered putative active enhancers if their region boundaries were ≥ 1 kb from the transcription start site (TSS)), following previous study [15]. Acetylated peaks were quantified between CTR and HFD groups by sorting and analyzing all BAM files with the ‘bedcov’ utility in samtools (version 1.2) [16]. ChIP and input sample coverage were normalized by peak length, with the subtraction of input coverage from ChIP coverage to limit the impact of fragmentation bias. Finally, differential H3K27ac analysis was performed using the DESeq2 R package, with significance threshold at p-value < 0.05 and |log₂(Fold change)| ≥ 1. All analyses were conducted in R version 3.5.1. SEs were identified using the ROSE algorithm following previous reported methods [17]. defined as groups of putative enhancers in close genomic proximity (12.5 kb) with signal densities higher than the threshold (plot slope > 1).

Gene Ontology (GO) and pathway enrichment analysis

Gene Ontology (GO) [18] and Kyoto Encyclopedia of Genes and Genomes (KEGG) [19] analyses of DEGs were conducted using the clusterProfiler package (v4.11.1) in R, aiming to identify significantly enriched terms and biological pathways associated with DEGs and altered metabolites. P-values < 0.05 were considered significant.

RNA-Seq and analysis of differential expressed gene

Total RNA was extracted from liver tissues using Trizol reagent (Invitrogen, CA, USA) following the manufacturer’s protocol. RNA samples were processed for library preparation and sequenced on the Illumina HiSeq 4000 platform to generate 150 bp paired-end reads. High-quality reads were then aligned to the Rattus_norvegicus_6.0 reference genome using STAR aligner (v2.5.3a) [20]. Raw sequencing reads underwent quality assessment with FastQC (v0.11.9) [21], followed by adapter trimming and quality filtering using Trimmomatic (v0.39) with the following parameters: SLIDINGWINDOW:4:20, MINLEN:36. Transcript quantification was performed with Salmon (v1.6.0) [22] in selective alignment mode, and transcript-level abundances were summarized to gene-level counts using tximport (v1.18.0) [23] in the R/Bioconductor environment (v4.1.0). Differential expressed genes (DEGs) were identified using DESeq2, with the criteria of |log₂(Fold change)| ≥ 1 and p-value < 0.05.

Integrative analysis of H3K27ac ChIP-seq and RNA-seq

To identify putative H3K27ac peak target genes, integrative analyses of acetylome H3K27ac ChIP-Seq and transcriptome RNA-Seq were conducted by calculating the correlation between gene transcription level and H3K27ac peak [24]. The genome-wide “four-way” analysis was performed to determine whether differential peaks could regulate target genes in a positive or negative manner at a threshold of |Cor (peak-gene correlation)| ≥ 0.5 and p-value < 0.05. Among these, PP represented upregulated SE-genes positively regulated by H3K27ac (log₂(Fold change) (ChIP ≥ 1 & RNA ≥ 1)); PN was known as downregulated SE-genes positively regulated by H3K27ac (log₂(Fold change) (ChIP ≥ 1 & RNA ≤ −1)), NP as upregulated SE-genes negatively regulated by H3K27ac (log₂(Fold change) (ChIP ≤ −1 & RNA ≥ 1)), NN as downregulated SE-genes negatively regulated by H3K27ac (log₂(Fold change) (ChIP ≤ −1 & RNA ≤ −1)).

Determination of core TFs

To identify the core transcription factors (TFs) of SULT1B1, we predicted the TFs binding to the enhancer and promoter region of SULT1B1 based on UCSC [25] and PROMO [26] databases. Next, protein-protein interaction (PPI) network of top overlapping TFs related to promoter and enhancer of SULT1B1 was constructed via STRING [27] database and visualized through Cytoscape (v3.10.0). Homology analysis was employed using UniProt [28] to investigate the homologies of SULT1B1 and CEBPB among human, mouse and rat. Finally, further validation of candidate TFs was conducted in the Cistrome [29] database. Potential binding sites for the candidate TF were predicted using the JASPAR [30] database.

Molecular docking

The amino acid sequence of C/EBPβ was obtained from the UniProt [28] database, while the promoter sequence of SULT1B1 was retrieved from the UCSC [25]database. Protein-DNA docking was conducted using AlphaFold3 [31], and the docking results were visualized and analyzed with PyMOL software (v2.5.0).

Single-cell RNA sequencing analysis

Single-cell transcriptomic data were obtained from the GEO dataset GSE174748 and GSE202379. For the GSE174748 dataset, cell quality control was implemented using the Seurat package (version 5.2.1), applying the following thresholds: genes detected per cell (nFeature_RNA; 300 < nFeature_RNA < 5,000), mitochondrial gene expression (percent.mt; percent.mt < 10%), and total RNA molecules per cell (nCount_RNA; nCount_RNA > 1,000 and below the 97th percentile). The harmony package (version 5.2.1) was used for batch-effect correction [32]. Subsequent cell clustering was performed using Seurat’s FindNeighbors and FindClusters functions. Cell type annotation was assigned based on established marker genes, referenced from both the original publication associated with GSE174748 and existing literature. For the GSE202379 dataset, the pre-processed RDS files containing the data and cell type annotations were directly obtained from GEO datasets.

Gene co-expression analysis

To assess the coordinated expression of SULT1B1 and CEBPB at single-cell resolution, we performed a gene co-expression analysis. Cells with detectable expression levels of both genes (expression value > 0) were classified as double-positive. This analysis was carried out separately in the Control and MASLD groups to compare co-expression patterns between conditions. The relationship between SULT1B1 and CEBPB expression in these double-positive cells was visualized using feature scatter plots. Furthermore, Spearman correlation analysis was applied specifically to the double-positive cell population to quantify the strength of association between the expression levels of the two genes.

Human transcriptomic data analysis

RNA-Seq datasets derived from human liver were obtained from the GEO under accession number GSE246221, GSE185051, GSE126848, GSE51143, GSE294293, and GSE158552. GSE246221 contains liver transcriptomes from 28 patients diagnosed with NAFLD and 4 healthy controls. In GSE246221, all NAFLD samples spanning fibrosis stages 1 through 4, were included to capture the overall disease-associated transcriptional profile. GSE185051 includes liver RNA-Seq profiles from 52 NAFLD patients and 5 healthy liver controls. GSE126848 comprises liver RNA-seq data from 15 NAFL patients and 14 healthy normal weight individuals. GSE51143 contains 2 JQ1-treated HepG2 cell samples and 3 control HepG2 cell samples. GSE294293 includes 3 JQ1-treated HepG2 cell samples and 3 control HepG2 cell samples. GSE158552 contains 3 JQ1-treated HepG2 cell samples and 3 control HepG2 cell samples. For all datasets, raw count matrices were normalized, and differentially expressed genes between MASLD and control groups were identified using the DESeq2 package with p < 0.05 as the significance threshold.

Transfection

To achieve gene knockdown, HepG2 and BRL-3A cells were transfected with 50 nM gene-specific siRNA targeting either SULT1B1 (si-SULT1B1; General Biol, Chuzhou, China) or CEBPB (si-CEBPB; General Biol, Chuzhou, China) using transfection reagent (lipofectamine 2000, Thermo Fisher Scientific, MA, China) according to the manufacturer’ s instructions. Non-targeting siRNA (si-NC) served as the negative control. After 24 h, 0.3 mmol/L oleic acid (OA) was added to the culture medium of BRL-3A and HepG2 cells to establish in vitro model of lipid accumulation. Twenty-four hours after OA-induced model establishment, the knockdown efficiency of target genes was evaluated by qRT-PCR and Western blot analysis.

Western blot analysis

Western blot analysis was performed as we previously described [33]. The primary antibodies used are listed in Supplementary Table 1.

RNA extraction and reverse transcription-quantitative real-time polymerase chain reaction (qRT-PCR)

To validate mRNA expression levels, total RNA was extracted from cellular samples utilizing the Trizol reagent (Invitrogen, USA) and reverse-transcribed into cDNA using a Reverse Transcription Kit (Vazyme Biotech, China). Subsequently, RT-PCR amplification was performed utilizing a SYBR Green Premix Kit (Vazyme Biotech, China), pre-denaturing cDNA at 95 °C for 10 min, followed by 40 cycles at 95 °C and 60 °C for 1 min. To accurately validate the transcriptional levels of mRNA, the expression levels were normalized and presented as fold changes relative to β-actin, calculated using the 2^△△Ct method. Relative primers used for RT-PCR analysis are listed in Supplementary Table 2.

Statistical analysis

Data were presented as mean ± standard deviation (SD) and analyzed using GraphPad Prism v.10.1 (GraphPad Software, USA). Group comparisons were conducted using t-tests or one-way analysis of variance (ANOVA), depending on the experimental design and data distribution. Pearson correlation analysis or Spearman rank correlation analysis was applied to assess associations between peak genes and samples. P-values were categorized as follows: * p < 0.05; ** p < 0.01; *** p < 0.001.

Results

The epigenetic marker H3K27ac plays a critical role in super-enhancers driven MASLD progression

To explore the epigenetic landscape associated with SEs formation and activity in MASLD, we adopted ChIP-Seq data of healthy and MASLD patients (GSE112221) [34] for three key histone marks-H3K27ac (active promoters and enhancers), H3K27me3 (suppressor), and H3K4me3 (active promoters), exhibiting distinct redistribution patterns during human MASLD progression (Fig. 1A). A total of 12,657 H3K27ac peaks, 7,646 H3K27me3 peaks, and 13,422 H3K4me3 peaks were identified with differential activity in MASLD compared to the normal group (Fig. 1B). Notably, the three modifications exhibited remarkably distinct patterns between normal individuals and patients (Fig. 1C, D), among which H3K27ac exhibited the most pronounced enrichment in pathways related to lipid metabolism and inflammation (Supplementary Fig. 1A), suggesting a pivotal role of H3K27ac in the regulation of MASLD progression.

Fig. 1.

Epigenetic landscape changes and SE identification in MASLD progression. (A) Three key histone marks in human MASLD progression. (B) Differential chromatin modification volcano plot comparing H3K27ac, H3K27me3, and H3K4me3 enrichment between normal liver and MASLD tissues (|log2(Fold change)| ≥ 5, p < 0.05) (GSE112221). Orange and green dots represent up-regulated peaks and down-regulated peaks respectively. (C) The peak counts of the three histone modifications and the difference between the number of up-regulated peaks (UR) and down-regulated peaks (DR). (D) Profiles and heatmaps showing the H3K27ac, H3K27me3, and H3K4me3 signals within 5 kb of transcription start site in the livers between normal and MASLD groups. (E) Pipeline for identifying super-enhancers (SEs). (F-K) Enhancer regions in control group (CTR) (n = 3). (F-H) and high fat diet group (HFD) (n = 3). (I-K) were plotted in ascending order based on their H3K27ac ChIP-Seq signal. Enhancers above the curve inflection point were classified as SEs.

To faithfully study the critical role of H3K27ac-associated SEs remodeling in MASLD, we first established HFD-induced MASLD model in rats (Supplementary Fig. 2). The livers of HFD-induced MASLD rats appeared light yellow color with thicker edges and greasier surface compared to the CTR group. Histological analysis through H&E staining and Oil Red O staining also revealed severe hepatic steatosis and pronounced lipid accumulation in the livers of HFD-induced rats (Supplementary Fig. 2A, B). In addition, the high-fat diet resulted in noticeable increase in both body weight and liver index (p < 0.05) (Supplementary Fig. 2C, D), as well as the significant elevations in serum levels of ALT (p < 0.001), AST (p < 0.05), TC (p < 0.01), TG (p < 0.05), LDL-C (p < 0.01), and the decrease in serum HDL-C (p < 0.05) levels of HFD-induced rats (Supplementary Fig. 2E-J), confirming the successful establishment of the MASLD model in this study.

Following the analytical pipeline illustrated in Fig. 1E, we employed H3K27ac marked ChIP-Seq to identify SEs. The number of SEs identified in HFD-induced MASLD samples (1246, 892, and 454) showed variation compared to control samples (456, 904, and 409) (Fig. 1F-K). These results highlight a differential H3K27ac-associated SE landscape between MASLD and control samples, suggesting that SE reprogramming plays a role in the epigenetic regulation of MASLD progression.

SE landscape reveals significant remodeling in MASLD

To further investigate the reprogramming of SEs and their potential impact on MASLD, we performed a comparative analysis between the CTR and HFD groups. Principal component analysis demonstrated clear segregation of SE landscapes between the two groups (Fig. 2A), while comparative analysis identified 19 differentially regulated SEs (|log₂(Fold change)| ≥ 1 and p < 0.05), consisting of 8 with decreased activity and 11 with increased activity in MASLD (Fig. 2B, Supplementary Table 3). Genomic distribution analysis revealed that H3K27ac peaks were predominantly enriched in distal intergenic regions (located > 3 kb away from TSS) [35](Fig. 2C), consistent with the localization pattern of SEs. Notably, significant enrichment was observed in pathways promoting lipid metabolism, such as the “Fatty acid metabolic process” and “Fatty acid biosynthetic process” (derived from GO analysis, Fig. 2D), as well as in pathways associated with lipid oxidation and insulin resistance, including the “Fatty acid biosynthesis” and “Insulin resistance” (derived from KEGG analysis, Fig. 2E). These findings collectively demonstrate that deregulation in SE landscape contributes to lipid related biological processes and pathways, which drives MASLD pathogenesis and progression.

Fig. 2.

Differential SEs profiles between CTR and HFD groups. (A) Principal Component Analysis (PCA) of ChIP-Seq data revealing distinct clustering between CTR and HFD groups. (B) Volcano plot of differential H3K27ac peaks between CTR and HFD groups (|log₂(Fold change)| ≥ 1, p < 0.05). Red and blue dots represent up-regulated peaks (n = 11) and down-regulated peaks (n = 8), respectively. (C) Distribution of H3K27ac peaks according to genomic region in livers from CTR and HFD groups. (D) Gene Ontology (GO) enrichment analysis of the top 10 enriched biological processes in HFD compared to CTR group. The rich factor represents the ratio of the proportion of differential expression genes annotated to a given pathway to the proportion of all background genes annotated to the same pathway (Rich Factor = GeneRatio/BgRatio). A higher Rich Factor indicates a greater degree of enrichment, while the -log₁₀(p-value) indicates the statistical significance. (E) Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis of the top 10 enriched pathways in HFD compared to CTR group.

SULT1B1 is identified as a core SE-driven gene by integrative analysis of ChIP-Seq and RNA-Seq

To identify the key genes regulated by SE in MASLD, we firstly carried out RNA-Seq analysis in CTR and HFD groups. PCA and cluster analysis of transcriptional profiles revealed clear segregation between the two groups, as visualized in Fig. 3A and B. A total of 1,041 significantly upregulated and 2,715 downregulated genes (|log₂(Fold change)| ≥ 1 and p < 0.05) were identified (Fig. 3C), with significant enrichment in biological processes related to lipid metabolism, including “Fatty acid metabolic process”, “Lipid transport”, and “Fatty acid biosynthetic process” (derived from GO analysis, Fig. 3D). And KEGG pathway analysis further indicated that upregulated genes were involved in “Non-alcoholic fatty liver disease” pathway (derived from KEGG analysis, Fig. 3E). These results suggest that transcriptional dysregulation is central to MASLD by regulating lipid metabolic pathways.

Fig. 3.

SULT1B1 is an epigenetically pivotal gene driven by SEs in MASLD. (A) PCA of RNA-Seq profiles revealing distinct clustering between CTR and HFD groups. (B) Dendrogram of RNA-Seq profiles. (C) Volcano plot of differentially expressed genes between CTR and HFD (|log₂(Fold change)| ≥ 1, p < 0.05). Red and blue dots represent up-regulated genes (n = 1,041) and down-regulated genes (n = 2,715) respectively. (D-E) Functional enrichment analysis of upregulated gene in HFD compared to CTR. The rich factor reflects the significance of each process, while the -log₁₀(p-value) indicates the statistical significance. (F) Genome-wide “four way” plot showing the genes with the thresholds of |Cor (peak-gene correlation)| ≥ 0.5 and p-value < 0.05, generated by integrative analysis of ChIP-Seq and RNA-Seq between CTR and HFD groups. PP SE-genes with upregulation and positive correlation are colored red, NN SE-genes with downregulation and negative correlation are colored blue, PN SE-genes with downregulation but positive correlation are colored green, and NP SE-genes with upregulation but negative correlation are colored orange. (G) Heatmap displaying the ChIP-Seq and RNA-Seq data of PP SE-genes. (H) Venn plot of PP SE-genes and genes up-regulated in MASLD patients. (I-J) The protein and relative mRNA levels of Sult1b1 expression in CTR and HFD groups. (K-M) Elevated expression level of SULT1B1 in MASLD versus CTR samples based on three GEO datasets (GSE185051, GSE246221 and GSE126848). (N) Positive correlation between expression level of Sult1b1 and H3K27ac peak density at the SE region (chr14:22375925-22457825).

Subsequently, an integrative analysis of H3K27ac ChIP-Seq and RNA-Seq data was performed to identify differential SEs and their target genes in the pathological process of MASLD. We conducted Genome-wide “four-way” analysis and identified differential SE-gene correlations of both positive and negative with a threshold of (log₂(Fold change) (|ChIP| > 1 & |RNA| > 1)), including 29 PP SE-genes (log₂(Fold change) (ChIP > 1 & RNA > 1)), 31 NN SE-genes (log₂(Fold change) (ChIP < −1 & RNA < −1)), 26 PN SE-genes (log₂(Fold change) (ChIP > 1 & RNA < −1)) and 31 NP SE-genes (log₂(Fold change) (ChIP < −1 & RNA > 1)) (Fig. 3F). And the PP SE-genes were visualized in the heatmap (Fig. 3G). To further confirm the core SE-genes influencing the progression of MASLD, we carried out an overlap analysis by combining the PP SE-genes and MASLD-related human datasets (GSE185051, GSE246221, GSE126848) [36–38] (Fig. 3H). Interestingly, SULT1B1 gene was uniquely overlapped and significantly increased in all four groups (Figure I-M). Moreover, Sult1b1 exhibited a strong positive correlation between expression level (RNA-Seq) and H3K27ac density at the SE region (ChIP-Seq) (Fig. 3N). Notably, we further validated a higher density of H3K27 acetylation at the region of chr14:22375925-22457825 in MASLD group, which was defined as putative SE region of Sult1b1 (Supplementary Table 4, Correlation = 0.9613). Together, SULT1B1, which was characterized as a core SE-driven gene in MASLD upon the epigenetic regulation of H3K27ac, potentially manifested a pivotal role in liver pathology.

Transcription factor C/EBPβ promotes transcriptional activation of SULT1B1 in MASLD

Given that SEs leverage H3K27ac-mediated chromatin opening to concentrate TFs and coactivators to amplify gene expression [39, 40], we sought to explore the upstream regulatory mechanisms by identifying TFs that interact with SE regions and activate SULT1B1 expression in MASLD. We initially characterized the top three peaks with high H3K27ac density in the SULT1B1_SE locus, and these peaks showed conserved enrichment patterns across MASLD rats and clinical patient samples (GSE267119) [41] (Fig. 4A, B, Supplementary Fig. 4A-D). In total, nine overlapped TFs were identified to potentially target both promoter and SE regions of Sult1b1 through PROMO [26]database (Fig. 4C), and C/EBPβ was pointed out as the most relevant TF due to its central role of widespread interaction with other TFs by PPI network analysis using the STRING [27]database (Fig. 4D). Moreover, regulatory potential scores were obtained from the Cistrome database to evaluate the relative transcriptional activity of each TF in liver tissue. Among them, C/EBPβ exhibited the highest regulatory potential, supporting its role as a key transcriptional regulator of SULT1B1 (Fig. 4E, Supplementary Fig. 3, Supplementary Table 5). Notably, both the mRNA and protein expression levels of Cebpb were markedly increased in MASLD rats (Fig. 4F, G). Furthermore, homology analysis of CEBPB and SULT1B1 via the UniProt database [28]demonstrated high sequence conservation across human, mouse, and rat (Fig. 4H, I). Pearson correlation analysis of the clinical transcriptomic dataset GSE126848 [38] validated a significant positive correlation between CEBPB and SULT1B1 expression levels (Fig. 4J). The expression level of both Cebpb and Sult1b1 exhibited positive correlations with serum ALT, AST, TG, TC, and LDL-C levels, but negative correlation with HDL-C (Fig. 4K), indicating their involvement in MASLD pathogenesis. Molecular docking analysis using AlphaFold3 [31] demonstrated binding of C/EBPβ to SE and promoter regions of Sult1b1 (Fig. 4L, Supplementary Fig. 4E-G). Moreover, we predicted C/EBPβ binding sites within the SULT1B1 promoter (1-2000 bp), identifying a key binding region (1862-1994 bp) (Fig. 4L). This region overlapped with the “ACTTGCCTCAT” motif (1912-1922 bp) predicted by JASPAR [30] analysis (Supplementary Fig. 4H, I, Supplementary Table 5), strongly suggesting physical interaction between C/EBPβ and the SULT1B1 promoter. To experimentally validate the predicted regulatory relationship, we transfected siRNAs targeting CEBPB in BRL-3 A and HepG2 cells (Fig. 5A, B). Following CEBPB silencing, the mRNA and protein levels of SULT1B1 was significantly reduced (Fig. 5C-H), confirming that C/EBPB positively regulates SULT1B1 expression at both the transcriptional and translational levels. Collectively, these findings provide mechanistic evidence supporting C/EBPβ as an upstream transcriptional regulator of SULT1B1, and their coordinated dysregulation may drive MASLD progression through lipid metabolism alterations.

Fig. 4.

C/EBPβ is a key transcription factor of SULT1B1 in MASLD. (A-B) ChIP-Seq profiles for H3K27ac at the SULT1B1 loci in MASLD rats and patients. E1, E2, and E3 indicate the top three enhancers with most prominent peaks within the SE region. (C) Venn diagram of putative TFs predicted to bind the SE and promoter regions of Sult1b1 in rat. (D) PPI network highlighting C/EBPβ as a central regulator. (E) Cistrome database validation of C/EBPβ binding to SULT1B1 gene in human liver tissue. (F) Relative mRNA levels of Cebpb in MASLD rat liver tissues. (G) Western blot analysis of C/EBPβ protein expression in CTR and HFD groups. (H-I) Sequence conservation analysis of CEBPB (H) and SULT1B1 (I) among human, mouse and rat. (J) Positive correlation between CEBPB and SULT1B1 mRNA levels (GSE126848). (K) Correlation heatmap illustrating the associations between Cebpb and Sult1b1 expression levels and serum lipid profiles. (L) AlphaFold3-predicted C/EBPβ binding region (1862-1994 bp) on SULT1B1 promoter.

Fig. 5.

Deletion of CEBPB diminishes SULT1B1 expression in BRL-3 A and HepG2 hepatic steatosis models. (A-D) The mRNA expression levels of CEBPB (A-B) and SULT1B1 (C-D) were quantified by RT-qPCR in BRL-3 A and HepG2 cells transfected with siRNAs against CEBPB (si-CEBPB_1, si-CEBPB_2, si-CEBPB_3) or negative control siRNA (si-NC). (E-H) The protein levels of C/EBPβ (E-F) and ST1B1 (G-H) in BRL-3 A and HepG2 cells transfected with siRNAs against CEBPB or negative control siRNA were detected by Western blot. *: p < 0.05, **: p < 0.01, ***: p < 0.001

Single-cell analysis reveals enrichment of SULT1B1 and CEBPB in hepatocytes in MASLD

To delineate the cell-type-specific expression profiles of CEBPB and SULT1B1 in MASLD pathogenesis, we performed single-cell RNA sequencing (scRNA-seq) analysis on healthy and MASLD livers obtained from the GEO dataset (GSE174748) [42]. Following quality control (QC) of the cells (Supplementary Fig. 5), we conducted dimensionality reduction using Uniform Manifold Approximation and Projection (UMAP) and clustered the cells into 10 subgroups (Fig. 6A). These clusters were then annotated into 6 major cell types, including “hepatocyte”, “liver sinusoidal endothelial cells” (LSECs), “T cells”, “macrophage”, “fibroblasts”, and “cholangiocytes” (Fig. 6B). The marker gene expression profiles for each cell type are illustrated in Fig. 6C. Furthermore, distinct cell distributions were also observed between control and MASLD groups (Fig. 6D). Notably, both SULT1B1 and CEBPB were significantly overexpressed in hepatocytes (Figs. 6E, F), with higher expression levels observed in the MASLD groups (Figs. 6G–J). Simultaneously, cells co-expressing SULT1B1 and CEBPB were also significantly enriched in hepatocytes (Figs. 6K, L). Moreover, the positive correlation between SULT1B1 and CEBPB was revealed in both control (Spearman’s R = 0.5019, p < 0.001) and MASLD (Spearman’s R = 0.6023, p < 0.001) (Supplementary Figs. 6A). To further examine the relationship between SULT1B1 and CEBPB across MASLD stages, we performed an analysis of a human scRNA dataset (GSE202379) [43], which shows expression patterns from early MASLD to advanced carcinogenesis (Supplementary Figs. 6B-F). Our analysis of this dataset revealed that SULT1B1 and CEBPB have a consistent expression trend across multiple stages in hepatocyte cells (Supplementary Figs. 6G-H), providing compelling evidence for their close association throughout disease progression. In conclusion, these results indicate that SULT1B1 and CEBPB are specifically enriched in MASLD hepatocytes, implicating their potential association in the progression of MASLD.

Fig. 6.

scRNA-seq reveals hepatocyte-specific expression of SULT1B1 and CEBPB in MASLD. (A) UMAP visualization of scRNA-seq data (GSE174748). (B) Annotation of clusters into six major liver cell types: hepatocytes, liver sinusoidal endothelial cells (LSECs), T cells, macrophages, fibroblasts, and cholangiocytes. (C) Marker gene expression profiles defining each annotated cell population. (D) Comparative cellular distribution between control and MASLD groups. (E-F) Cell type-specific expression patterns of SULT1B1 (E) and CEBPB (F). (G-J) Elevated expression of SULT1B1 (G-H) and CEBPB (I-J) in MASLD hepatocytes compared to controls. (K-L) Co-expression analysis revealing significant overlap of SULT1B1 and CEBPB in hepatocytes (K: UMAP projection, L: pie chart of cell type distribution).

SE inhibition with JQ1 reverses SULT1B1 expression and hepatic lipid accumulation

To systematically validate the functional interplay between SEs and SULT1B1 in hepatic lipid accumulation, we employed pharmacological inhibition of SEs using JQ1 in vitro MASLD models. JQ1, a commonly used inhibitor of BRD4, hinders the binding of BRD4 to H3K27ac, thereby reducing BRD4 occupancy in the SE region and disrupting SEs activity. We found that JQ1 treatment effectively decreased lipid droplets in BRL-3A and HepG2 cells (Fig. 7A-F), and inhibited the expression of SULT1B1 at mRNA levels (Fig. 7G-H). Moreover, western blot also confirmed the decreased protein levels of ST1B1 and H3K27ac, supporting the regulatory role of SEs in controlling SULT1B1 expression (Fig. 7I-L).

Fig. 7.

JQ1 treatment reduces lipid accumulation and downregulates SULT1B1 and H3K27ac expression. (A-B) Representative micrographs of Oil Red O staining and Bodipy 493/503 fluorescence staining in BRL-3A and HepG2 cells across experimental groups. (C-F) Quantitative analyses of lipid accumulation presented as percentage of total cellular area for both Oil Red O- and Bodipy-stained regions. (G-H) RT-qPCR quantification of SULT1B1 mRNA expression levels in respective treatment groups. (I-L) Western blot analysis of ST1B1 (I, K) and H3K27ac (J, L) protein expression in hepatic specimens from indicated cohorts. (M) Quantification of expression of ST1B1 protein levels in BRL-3A and HepG2 cells. (N) Quantification of expression of H3K27ac protein levels in BRL-3A and HepG2 cells. *: p < 0.05, **: p < 0.01, ***: p < 0.001. All data are shown as mean ± SD. BSA: Bovine serum albumin, OA: oleic acid, DMSO: Dimethyl sulfoxide.

It is important to note that JQ1 is a pan-BET inhibitor and its effect on lipid accumulation is likely the composite result of suppressing multiple SE-driven genes involved in lipid metabolism. However, given the specific identification of SULT1B1 as a SE-associated gene in our prior analyses, and its significant downregulation in JQ1-treated HepG2 datasets (GSE51143, GSE294293, GSE158552, Supplementary Figs. 7) [44–46], we identified SULT1B1 as a key mediator. To further determine the role of SE-mediated SULT1B1, we specifically knocked down SULT1B1 (Fig. 8A-D), and the result shown that SULT1B1 knockdown was sufficient to exert the anti-steatotic effect in MASLD (Fig. 8E-J), underscoring SULT1B1 is a pivotal SE-mediated gene in driving lipid accumulation.

Fig. 8.

SULT1B1 is a critical SE-regulated gene that drives lipid accumulation in MASLD. (A-B) The mRNA expression levels of SULT1B1 in BRL-3A and HepG2 cells transfected with siRNAs against SULT1B1 (si-SULT1B1_1, si-SULT1B1_2, si-SULT1B1_3) were quantified by RT-PCR to assess knockdown efficiency. (C-D) The protein levels of ST1B1 in the BRL-3A cells (C) and HepG2 cells (D) transfected with siRNAs against SULT1B1 or negative control siRNA (si-NC) were determined by Western blot. (E-F) Representative micrographs of Oil Red O staining and Bodipy 493/503 fluorescence staining in BRL-3A and HepG2 cells transfected with siRNAs against SULT1B1 or negative control siRNA. (G-J) Quantitative analyses of lipid accumulation presented as percentage of total cellular area for both Oil Red O- and Bodipy-stained regions. *: p < 0.05, **: p < 0.01, ***: p < 0.001

Take together, these findings highlight the functional importance of SE-mediated transcriptional control in MASLD and suggest that targeting SE components such as BRD4 may be a promising strategy to modulate pathogenic gene expression.

Discussion

As the most prevalent liver disease globally, MASLD poses a growing health threat through its strong associations with metabolic disorders, while recent evidence underscores the critical role of epigenetic dysregulation in driving its pathological gene expression profiles [47, 48]. In this study, we integrated ChIP-Seq, RNA-Seq, and single-cell transcriptomics data to provide a holistic view of the super-enhancer mediated epigenetic and transcriptional changes in MASLD, highlighting the key gene SULT1B1 as a core SE-driven gene in disease pathogenesis. Mechanistically, we found that SULT1B1 transcription was activated by C/EBPβ dependent H3K27ac modification. Functionally, pharmacological inhibition of SE using JQ1 effectively reduced lipid accumulation in both BRL-3 A and HepG2 hepatocyte models, and also decreased protein levels of ST1B1 and H3K27ac, further supporting the regulatory role of SEs in mediating SULT1B1 expression. These findings established a novel regulatory axis of SE by H3K27ac/C/EBPβ/SULT1B1 in MASLD, offering potential therapeutic targets and strategies for treating metabolic liver diseases (Fig. 9).

Fig. 9.

A proposed model of SE-mediated SULT1B1 activation in MASLD and its pharmacological inhibition. H3K27ac was identified as a key histone modification profiled in human MASLD, where its aberrant enrichment disrupted the regulatory functions of SEs. In HFD-induced rat model, integrative multi-omics analysis revealed pivotal SE regions, target gene, and TF involved in MASLD progression. C/EBPβ was found to bind to both the promoter and SE regions of SULT1B1, driving its transcriptional activation in MASLD. This effect can be reversed by JQ1, a small-molecule inhibitor targeting BRD4, thereby suppressing SULT1B1 expression and offering potential therapeutic insight

SEs are clusters of enhancers spanning large genomic regions, typically reaching ten times the size of regular enhancers [17]. These SE regions are characterized by high densities of epigenetic marks, such as H3K27ac, which maintains a highly activated chromatin state [49, 50]. This activated state facilitates the robust recruitment of transcriptional co-activators, including BRD4 and MED [51]. Notably, research by Sabari et al. demonstrated that the intrinsically disordered regions (IDRs) of BRD4 and Med1 undergo phase separation at SE sites, forming liquid-like condensates [52]. This condensation critically compartmentalizes the transcription process by concentrating transcriptional apparatus in proximity to SEs, thereby significantly enhancing transcriptional efficiency. In this study, we also demonstrated substantial SE reorganization, which impacts lipid related biological processes and pathways to drive MASLD and progression. Furthermore, we treated MASLD cellular models with JQ1, a specific inhibitor targeting BRD4 [53]. JQ1 treatment not only effectively reversed lipid accumulation and H3K27ac enrichment in the MASLD models but also significantly downregulated SULT1B1 expression at both transcriptional and protein levels. Although JQ1 is an inhibitor broadly suppressing transcription, our integrated genomics approach pinpointed SULT1B1 as a prominent SE-associated candidate. The confirmed sensitivity of SULT1B1 to JQ1 in human hepatoma cells further supports its status as a BET-sensitive gene. The subsequent validation using siRNA-mediated knockdown, provides compelling evidence that SULT1B1 is a functionally important target within the broader transcriptional program regulated by BET proteins in hepatic steatosis. Future studies utilizing more tools, such as BRD4 degraders or CRISPRi targeting the SEs of SULT1B1 and other lipid-related genes, could help delineate the precise contribution of SULT1B1 within the BRD4-regulated transcriptional network.

During the progression of various diseases, aberrant acquisition of SEs at disease-associated genes is a widely observed phenomenon. In small cell lung cancer cells, SEs have been found to be associated with a number of oncogenic genes, such as MYC, SOX2, and NFIB [54]. Liver cancer cells also acquire SEs at the loci of critical oncogenic genes, such as SPHK1, MYC, MYCN, SHH, and YAP1, to drive their substantial overexpression [55]. In this study, we integrated H3K27ac ChIP-Seq and RNA-Seq data, revealing SULT1B1 as a core SE-driven gene in MASLD. Intriguingly, ST1B1 is a member of the cytosolic sulfotransferases (SULTs) superfamily, which can catalyze the transfer of a sulfonate group and have a major effect on the chemical and functional homeostasis of substrate chemicals [56, 57]. Previous studies have established SULT1B1’s role in detoxification and lipid metabolism, implying its potential involvement in MASLD [58, 59]. This study further elucidates the specific role of SE mediated SULTB1 in lipid accumulation in BRL-3A cells and HepG2 hepatocyte models, providing mechanistic insights that reinforce its potential as a therapeutic target for MASLD.

SEs are marked by high levels of H3K27ac and serve as focal points for TF binding and coactivators recruitment [17]. Specifically, TFs play a key role in recognizing modifications and binding to gene regulatory elements [60–62]. By analyzing the core SE and promoter sequences of SULT1B1, we identified C/EBPβ as a critical transcription factor of this gene. Recent studies have revealed that upregulation of C/EBPβ promotes genome-wide H3K27ac deposition [63]. Furthermore, as an essential TF governing adipocyte differentiation and lipid droplet biogenesis, C/EBPβ plays critical role in adipogenesis, a process intimately linked to metabolic pathologies including MASLD [12, 64]. Collectively, these previous studies demonstrate functional parallels between C/EBPβ and SULT1B1. Our study further validates a significant molecular and cellular connection between C/EBPβ and SULT1B1. Molecular docking simulations revealed specific C/EBPβ binding motifs within the SULT1B1 promoter, indicating structural compatibility for binding. Targeted silencing of CEBPB resulted in significant downregulation of SULT1B1 mRNA and protein levels. Single-cell RNA sequencing analysis confirmed robust co-expression of SULT1B1 and CEBPB in hepatocytes, with both genes showing significantly elevated expression in MASLD samples compared to healthy controls. Collectively, these findings establish a mechanistic connection between SE-driven epigenetic remodeling and MASLD progression, specifically through the H3K27ac/C/EBPβ/SULT1B1 regulatory axis.

Our study advances MASLD research through multi-omics integration of H3K27ac ChIP-Seq, RNA-Seq, and single-cell transcriptomics, identifying SE-driven SULT1B1 and C/EBPβ activation as a critical epigenetic axis in disease progression. However, several limitations warrant consideration. First, despite the clear separation of groups observed in PCA and the statistical significance (p < 0.05) of our key results, the relatively modest sample size of both human and animal datasets may limit the statistical power and generalizability of our findings. Future studies involving larger independent cohorts are warranted to validate and extend our results. Second, the regulatory mechanisms by which SEs control SULT1B1 in different species warrant further discussion. On one hand, comparative studies indicate that the overall SE landscape often exhibits species divergence, while core SE modules can be conserved [65]. On the other hand, although differences in catalytic efficiency exist between species, our analysis further demonstrates that ST1B1 is highly conserved among human, mouse, and rat (Fig. 4I), which is consistent with previous work by Fujita et al. [66]. Future investigations should prioritize primary human hepatocytes or patient derived samples to better delineate the species-specific functions of SEs and SULT1B1. Third, although our data suggest that the SE recruits C/EBPβ to activate SULT1B1, the detailed regulatory mechanism requires further experimental validation. Finally, while SULT1B1 appears to promote hepatic steatosis, its precise role and downstream effectors in the context of MASLD progression remain to be elucidated. Despite these limitations, our work establishes a foundational framework for exploring SE-mediated epigenetic dysregulation in MASLD, advancing our understanding of transcriptional complexity in metabolic diseases and paving the way for therapeutic strategies.

Conclusion

In summary, our study establishes a direct mechanistic link between SE-driven epigenetic remodeling and MASLD progression, specifically mediated by the H3K27ac/C/EBPβ/SULT1B1 axis, which not only provides new insights into the epigenetic regulation in liver physiology, but also highlights the therapeutic potential of targeting SE-associated epigenetic modifiers for metabolic liver diseases.

Abbreviations

ALT

Alanine aminotransferase

AST

Aspartate aminotransferase

ChIP-Seq

Chromatin immunoprecipitation sequencing

CTR

Control group

C/EBPβ

CCAAT enhancer binding protein beta

DEGs

Differential expressed genes

GO

Gene Ontology

HCC

Hepatocellular carcinoma

HDL-C

High density lipoprotein holesterol

H&E

Hematoxylin and Eosin

HFD

High fat diet

H3K27ac

Acetylation of the lysine residue at Nterminal position 27 of the histone H3

KEGG

Kyoto Encyclopedia of Genes and Genomes

LDL-C

Low density lipoprotein cholesterol

lncRNAs

Long noncoding RNAs

MASLD

Metabolic dysfunction-associated steatotic liver disease

OA

Oleic acid

PCA

Principal Component Analysis

Pol II

RNA polymerase II

qRT-PCR

Quantitative reverse transcription polymerase chain reaction

RNA-Seq

RNA sequencing

ROSE

Rank ordering of SEs

scRNA-Seq

Single cell RNA sequencing

SE

Super-enhancer

SULT1B1

Sulfotransferase family 1B member 1

TC

Total cholesterol

TF

Transcription factor

TG

Triglyceride

TSS

Transcription start site

UMAP

Uniform Manifold Approximation and Projection

Author contributions

X. L.: conceptualization, visualization, formal analysis, writing - original draft preparation, writing - review &editing. Y. Y.: visualization, formal analysis, writing - review & editing. J. L.: validation, writing - review & editing. X. P.: visualization, formal analysis. M. Z.: visualization, writing - review &editing. X. Z.: investigation, writing - review &editing. F. Z.: validation. Y. T.: validation, resources, writing - review & editing. Y. Z.: conceptualization, data curation, project administration, supervision, validation, funding acquisition.

Funding

This work was funded by the National Natural Science Foundation of China, grant number 82300661; the Natural Science Foundation of Anhui province, grant number 2308085QH246; Basic and Clinical Cooperative Research Promotion Program of Anhui Medical University, grant number 2022xkjT013. Scientific Research Foundation of Anhui Medical University, grant number 2023xkj002.

Data availability

The dataset supporting the conclusions of this article is available in the Genome Sequence Archive repository, [CRA008234 to datasets in [https://ngdc.cncb.ac.cn/gsa/](https:/ngdc.cncb.ac.cn/gsa)]. The GEO([https://www.ncbi.nlm.nih.gov/geo/](https:/www.ncbi.nlm.nih.gov/geo)) dataset used for this study can be found below: ChIP-Seq data of multi histone modifications from GSE112221, GSE267119; RNA-Seq data from GSE246221, GSE185051, GSE126848, GSE51143, GSE294293 and GSE158552; Single-cell data from GSE174748, GSE202379.

Declarations

Ethics approval and consent to participate

This study has been reviewed and approved by the Institutional Review Board (IRB) at Anhui medical university (#LLSC20230636; 83230226). All research procedures will be conducted in accordance with the ethical standards outlined in the Declaration of Helsinki, and applicable local and international guidelines for human subjects research.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.