Identification of IL-34 and Slc7al as potential key regulators in MASLD progression through epigenomic profiling

ABSTRACT

Objective

Epigenetic alterations are critical regulators in the progression of metabolic dysfunction-associated steatotic liver disease (MASLD); however, the dynamic epigenomic landscapes are not well defined. Our previous study found that H3K27ac and H3K9me3 play important roles in regulating lipid metabolic pathways in the early stages of MASLD. However, the epigenomic status in the inflammation stages still needs to be determined.

Method

C57BL/6 male mice were fed with the methionine- and choline-deficient (MCD) or normal diet, and their serum and liver samples were collected after 6 weeks. Serum alanine aminotransferase (ALT), aspartate amino transferase (AST), total cholesterol (TC), triglyceride (TG), high-density lipoprotein cholesterol (HDL-C), and low-density lipoprotein cholesterol (LDL-C) levels were measured. Chromatin immunoprecipitation sequencing (ChIP-Seq) for H3K27ac and H3K9me3 was performed together with RNA sequencing (RNA-seq) and key regulators were analyzed.

Results

The target genes of enhancers with increased H3K27ac and decreased H3K9me3 signals are enriched in lipid metabolism and immuno-inflammatory pathways. Il-34 and Slc7al are identified as potential regulators in MASLD.

Conclusion

Our study reveals that active enhancers and heterochromatin associated with metabolic and inflammatory genes are extensively reprogrammed in MCD-diet mice, and Il-34 and Slc7al are potentially key genes regulating the progression of MASLD.

1. Introduction

The global prevalence of metabolic dysfunction-associated steatotic liver disease (MASLD) had risen from 25.26% in 1990–2006 to 38.0% in 2016–2019 and had become one of the most common chronic diseases in the world [1]. Metabolic dysfunction-associated steatohepatitis (MASH), a more active form characterized by hepatic steatosis, inflammation, and hepatocellular swelling, is emerging as one of the leading causes of cirrhosis, cirrhotic complications, hepatocellular carcinoma, and liver-related deaths [2]. Although weight loss is the most effective treatment for MASLD, only 10% of the patients actually achieved this goal through lifestyle interventions [3]. In addition, with the long natural history and complex pathogenesis of MASLD, many questions and significant challenges remain in the treatment of MASLD.

Abnormal histone modifications are involved in the progression of insulin resistance and subsequent MASLD [4]. H3K27ac is the acetylation product of the 27th lysine residue at the N-terminal of the histone H3 protein, which determines the compactness and accessibility of chromatin and is an important epigenetic marker of active enhancers [5,6]. In a rat model of high-fat diet (HFD)-induced MASLD, 1831 different H3K27ac peaks were identified, which were significantly associated with transcription factors and target genes involved in lipid and energy balance, such as Cyp8b1, Pla2g12b, Slc27a5, Cyp7a1 and Apoc3 [7]. In another MASLD mice model induced by HFD, the H3K27ac-labeled active enhancers were enriched in the genes of lipid metabolism pathway, indicating that the transcriptional regulation of lipid metabolism genes had been reprogrammed. Further analysis of acquired enhancers showed that transcription factors such as Pparα, Cebp, Hnf4a and Fos were activated [8]. In addition, Ya-Ling Zhu et al. found that lipopolysaccharide-binding protein deletion caused an increase in H3K27ac through the transcription factor C/EBPβ, which activated the expression of downstream Scd genes and aggravated MASLD [9]. Similarly, sterol regulatory element-binding protein (SREBP) cleavage-activating protein (SCAP) N- glycosylation exacerbates inflammation and lipid accumulation in hepatocytes by enhancing Acss2-mediated H3K27ac [10]. Histone lysine methylation is one of the important epigenetic mechanisms for transcriptional regulation, and involved in the regulation of cell cycle, genomic stability and nuclear structure [11,12]. H3K9me3 controls the structure and functions of heterochromatin and inhibits gene transcription [12]. It has been reported that lipid accumulation in the liver led to abnormal H3K9me3 and H3K4me3 states of Pparα and lipid metabolic genes, which decreased their mRNA expression [13]. In addition, the average peak numbers and intensity of H3K9me3 inhibitory markers decreased in MASLD mice, and the enhancers located in the missing H3k9me3 loci were enriched in lipid metabolism and inflammatory pathways [8]. Kim JH et al. found that by reducing the enrichment of H3K9me3 near the ligand-activated liver X receptor (LXR) response element (LXRE) in the promoter region of LXRα target genes, LXRα activation and thus up-regulation of the expression of adipogenic genes led to hepatic lipid accumulation [14]. Conversely, increased H3K9me3 enrichment at Fasn, Srebp1 and Pparg genes to suppress gene expression and thereby reduce lipid accumulation [15]. Therefore, H3K27ac and H3K9me3 modification status and their alterations are of great importance for improving MASLD. However, the roles and mechanisms of the above histone modifications in MASLD still requires further exploration, and detailed illustrations of dynamic epigenomic changes in different diets will be helpful to fully understand the transcriptional regulation in MASLD.

The methionine- and choline-deficient (MCD) diet mice model is one of the widely used models to mimic MASLD progression [16,17]. MCD has been broadly used over 40 years. It produces the severe phenotype in a shortest timeframe, and induces hepatic steatosis in mice within 2–4 weeks which progresses to inflammation and fibrosis shortly thereafter [18]. The MCD model is often used to mimic hepatic inflammation and fibrosis in MASLD, but is not used in studying other phenotypes, particularly regarding insulin resistance and obesity [19–21]. The MCD diet significantly induces weight loss, which is uncommon in most human MASLD patients and may obscure the effects of certain metabolic factors. Additionally, liver damage caused by choline deficiency does not fully align with the mechanisms observed in human MASLD, limiting the broader applicability of this model. Furthermore, the MCD diet tends to induce more severe liver inflammation and fibrosis, leading to pathological differences compared to certain stages of human MASLD. The imbalance in dietary components may also introduce additional variables related to nutritional deficiencies, complicating the resulting interpretation. Therefore, considering the limitations of the MCD diet, we incorporated HFD and high fructose, high fat and high cholesterol (HFHC) diets for further analysis and validation. We have previously studied the epigenetic profiles in the mice livers fed with HFD [8]. To fully understand the epigenetic reprogramming during MASLD with different diets, we have constructed an MCD-diet induced MASLD mice model in the current study, performed the integrative analysis of RNA sequencing (RNA-seq), H3K27ac and H3K9me3 chromatin immunoprecipitation sequencing (ChIP-Seq) in the liver tissues, and screened for potential new target genes.

2. Materials and methods

2.1. Mice model and animal housing

Nineteen C57BL/6 male mice (GemPharmatech Co., Ltd.) aged 5 weeks ~20 g were randomly divided into normal group (n = 9) and MCD group (n = 10). After 1 week of adaptive feeding, normal group was fed with a normal maintenance diet rich in methionine and choline (Wuhan Wanqian Jiaxing Biotechnology Co., Ltd.) and MCD group was fed with methionine and choline deficiency diet (Research Diets, A02082002B) for 6 weeks [22]. After 6 weeks of feeding, all mice were sacrificed by cervical dislocation after the whole blood was collected from the eyeballs. The mice were all housed in a controlled environment with a 12-hour light/dark cycle and provided with food and water ad libitum. The ambient temperature was maintained at 24 ± 2°C, with a relative humidity of 45%. All experimental procedures involving mice operation were conducted in accordance with the laboratory animal guidelines of Wuhan University and approved by the Animal Experimentations Ethics Committee (Protocol NO. 14110B).

2.2. Serologic testing

The serum total cholesterol (TC), triglyceride (TG), alanine aminotransferase (ALT), aspartate aminotransferase (AST), high-density lipoprotein cholesterol (HDL-C), and low-density lipoprotein cholesterol (LDL-C) were detected by Siemens automatic biochemical analyzer ADVIA 2400 (Erlangen, Germany) from Department of Laboratory Science, Renmin Hospital, Wuhan University.

2.3. Tissue section staining

H&E staining of paraffin-embedded tissue sections was carried out with a series of steps, including xylene and ethanol dehydration, hematoxylin staining, eosin staining, and dehydration with ethanol and xylene for preservation. For Masson staining, tissue slices were dewaxed with xylene and ethanol, stained with hematoxylin, rhododendron, and phosphomolybdic acid, re-stained with aniline blue, differentiated with glacial acetic acid, dehydrated, and sealed with neutral gum. For Oil Red O staining of frozen sections, frozen tissue sections were treated with fixed solution, oil red dye staining, isopropanol differentiation, hematoxylin staining, alcohol differentiation, and examined under a microscope.

2.4. MASLD scoring system

The MASLD scoring system evaluates liver damage based on three components [23]: steatosis (0–3 points: 0 = none, 1 = <33%, 2 = 33–66%, 3 = >66%), lobular inflammation (0–3 points: 0 = none, 1 = <2 foci, 2 = 2–4 foci, 3 = >4 foci), and hepatocyte ballooning (0–2 points: 0 = none, 1 = rare, 2 = frequent), with scores < 3 indicating mild MAFLD, 3–4 indicating moderate MAFLD, and ≥ 5 indicating MASH, calculated using the mode score for each characteristic.

2.5. ChIP assay and library construction

60 mg liver tissue was shredded to 1 mm in phosphate buffer saline (PBS) with ice bath, crosslinked for 10 min with 1% formaldehyde at room temperature and terminated with 0.125 M glycine at room temperature. The samples were fully ground in a tissue grinder and centrifuged at 4°C for 12,000 rpm for 10 min. The samples were washed with ice-cold PBS twice before adding a lysis buffer (50 mm pH 8.0 Tris-HCl, 0.1% SDS, 5 mm EDTA). After washing once with a digestion buffer (50 mm Tris-HCl, pH 8.0 mm CaCl2, 0.2% Triton X-100), digestion buffer and Micrococcal Nuclease (NEB, M0247S) were added to fragment the DNA into 150–300 bp. An ultrasonic crusher was used to break the cells. One-fourth of the fragmented DNA was used as input, and the other was co-immunoprecipitated with H3K27ac (Abcam, ab4729, RRID: AB_2118291), H3K9me3 (Abcam, ab176916, RRID: AB_2797591) and IgG antibodies with protein G (GE Healthcare 17,061,805), respectively. After incubation, the precipitates were then washed with Wash Buffer I (20 mm Tris-HCl, pH 8.0, 150 mm NaCl, 2 mm EDTA,1% Triton X-100, 0.1% SDS), Wash Buffer II (20 mm pH 8.0 Tris-HCl, 500 mm NaCl, 2 mm EDTA, 1% Triton X-100, 0.1 SDS), Wash Buffer III (10 mm pH 8.0 Tris-HCl, 0.25 m LiCl, 1 mm EDTA, 1% NP-40), and then incubated with elution buffer (0.1 M NaHCO₃, 1% SDS) and proteinase at 65°C. A universal DNA purification recovery kit (Tiangen Biotech Co., Ltd., DP214) was used for DNA recovery. The obtained DNA was used to construct the ChIP-seq library with VAHTS Universal DNA Library Prep Kit for Illumina V3 (Vazyme, ND607). Finally, the Illumina HiSeq × Ten platform was used for sequencing.

2.6. Data analysis for ChIP-Seq

Raw data quality control was performed by FastQC (version 0.11.9, https://www.bioinformatics.babraham.ac.uk/projects/fastqc/) and MultiQC (version 1.4, https://multiqc.info/). Subsequently, low-quality bases and adapters in the sequencing data were removed by Fastp (version 0.23.4, https://github.com/OpenGene/fastp, parameters were “-f 10 -t 10 -F 10 -T 10 -l 30”). The cleaned data were aligned to the mice UCSC reference genome mm10 with BOWTIE2 (version 2.3.5.1, https://bowtie-bio.sourceforge.net/bowtie2/index.shtml) with default parameters. Samtools (version 1.16.1, https://www.htslib.org/) was used for conversion between SAM and BAM files and PCR duplicate reads removal. MACS2 (version 2.1.1, https://github.com/taoliu/MACS) was used to call histone modification enrichment regions (peaks): H3K27ac parameters using -f BAMPE -g mm – nomodel – keep-dup all -B – extsize 147 -p 1E–6 –broad – broad-cutoff 1E–6, H3K9me3 parameters using -f BAMPE -g mm – nomodel – keep-dup all -B – extsize 147 -p 1E–4 –broad – broad-cutoff 1E–4. Bedtools intersect (version 2.30.0, https://bedtools.readthedocs.io/en/latest/index.html) was used to calculate the number of reads aligned to the mice mm10 genomic regions in ChIP-Seq data. HOMER (version 4.11, http://homer.ucsd.edu/homer/) was used for peak annotation on the mm10 genome and for finding significantly enriched motifs. BigWig files were generated by Deeptools bamCoverage (version 3.0.2, https://deeptools.readthedocs.io/en/latest/, parameters were “–bs 10 –normalizeUsing RPKM”). GO analysis and KEGG analysis of H3K9me3 modification were performed by DAVID (https://david.ncifcrf.gov/). The functional enrichment analysis of enhancer neighboring genes was performed by GREAT website (version 4.0.4, http://great.stanford.edu/public/html/). The H3K27ac and H3K9me3 bigwig files were visualized using the WashU Epigenome Browser (https://epigenomegateway.wustl.edu/browser/).

2.7. Identification of active enhancer and super enhancer

Since H3K27ac is generally present in the promoter and enhancer loci, we removed the peak located in the promoter loci (2 kb above and below the TSS) using bedtools intersect -v (version 2.30.0, https://bedtools.readthedocs.io/en/latest/index.html), and defined the remaining H3K27ac peak as the enhancer loci. The super enhancers (SEs) were recognized by ROSE (https://hpc.nih.gov/apps/ROSE.html, parameters were “-g MM10 -s 12,500 -t 12,500”). Bedtools intersect was used to calculate the number of reads aligned to the super enhancer regions in ChIP-Seq data.

2.8. Data normalization

RNA-Seq data were normalized to fragments per kilobase of bin per million mapped reads (FPKM), which was normalized by DESEQ2 when performing gene differential expression analysis. ChIP-Seq data were normalized to reads per kilobase per million mapped reads (RPKM) using deeptools (version 3.0.2, https://deeptools.readthedocs.io/en/latest/, parameters were “–bs 10 –normalizeUsing RPKM”).

2.9. Correlation and principal component analysis

The sample correlation was performed using the Spearman correlation coefficient. Principal Components Analysis (PCA) was performed with the R package FactoMineR (version 2.9), based on the results normalized by DESEQ2.

2.10. Identification of variant enhancer loci (VELs) and variant H3K9me3 loci

We merged the enhancer loci of all samples and the H3K9me3 peak of all samples into one single coordinate file, and then calculated the number of counts for all samples on the merged enhancer loci and the merged H3K9me3 peak loci. We performed a differential analysis with R package DESEQ2 (version 1.40.1, https://bioconductor.org/pack-ages/release/bioc/html/DESeq2.html) to look for VELs and variant H3K9me3 loci in the MCD and normal group, with a threshold of p-value <0.05 & fold change ≥ 2.

2.11. Identification of variant super enhancer loci (VSELs)

The recognition of VSELs was performed by DESEQ2 (version 1.40.1, https://bioconductor.org/pack-ages/release/bioc/html/DESeq2.html) with a threshold of p-value <0.05 & fold change ≥1.5.

2.12. Reverse transcription and quantitative PCR

Liver tissue samples were processed by adding TRIzon reagent and non-ribozyme magnetic beads for complete grinding, followed by vigorous shaking with chloroform and centrifugation for 10 minutes. The upper layer containing water was collected and mixed with isopropanol in equal volume for RNA extraction. The extracted RNA was purified using the Ultrapure RNA Kit (Cwbiotech, CW0581M) and its concentration was determined using NanoDrop. Reverse transcription was performed using HiScript III All-in-one RT SuperMix Perfect for qPCR (Vazyme, R333–01). Real-time fluorescence quantitative PCR was carried out using ChamQ Universal SYBR qPCR Master Mix (Vazyme, Q711), and CT values were calculated using the 2^^(-△△Ct) method with β-actin as the reference gene. The sequences of primers are in Table S1, Supporting Information.

2.13. Western blot analysis

Liver tissue samples were homogenized twice in PBS, followed by thorough homogenization with radio immunoprecipitation assay protein lysis buffer. The homogenate was centrifuged at 4°C for 10 minutes, and the supernatant was collected. An equal volume of SDS loading buffer was added, and the mixture was heated in a 98°C water bath for 15 minutes before storage at −80°C. Protein samples were separated by 10% SDS-PAGE and transferred to a nitrocellulose membrane (Millipore). The membrane was blocked with 5% skim milk, then incubated overnight at 4°C with primary antibodies IL-34 (Abcam, ab101443, RRID: AB_10711208) and GAPDH (ABclonal, AC033, RRID: AB_2769570). Afterward, the membrane was washed three times with TBS-T and incubated with secondary antibodies at room temperature for 1 hour. Protein expression was detected using the Clarity Western ECL substrate (BIO-RAD). Finally, Image J software version 1.54 (National Institutes of Health, Bethesda, MD, USA) was used to quantitatively measure the gray level of protein bands.

2.14. Statistics and reproducibility

For experiments other than next-generation sequencing (NGS), a minimum of three biological replicates were conducted for each experiment. The data is represented as mean values with standard error of the mean (SEM). Statistical analysis was carried out using a two-sided Student t test. The p value was either indicated on the respective items or included in the legends. GraphPad Prism version 9.2.0 was used for statistical analysis.

3. Results

3.1. Construction and characterization of mcd-diet induced mice model

To established the MCD-diet mice model, five-week-old male C57/BL6 mice about 20 g were randomly divided into a normal diet group (n = 4) and MCD group (n = 5) after adaptive feeding for 1 week. Compared with the normal group, the body weight and liver/body weight ratio of MCD group decreased significantly (Figure 1(a), Supplementary Fig. S1A). Pathological staining showed obvious steatosis, inflammatory infiltration and slight fibrosis in MCD group (Figure 1(b)). Further analysis showed that NAS score, liver injury indexes ALT and AST increased significantly (Supplementary Fig. S1B-D), and serum TC, TG, HDL-C and LDL-C all significantly decreased (Supplementary Fig. S1E-H). The above results were consistent with the characteristics of the MCD model [24].

Figure 1.

Chromatin immunoprecipitation sequencing of H3K27ac and H3K9me3 in methionine and choline deficiency diet and normal liver tissues. (a) Body weight of MCD and normal group. (b) H&E, Masson, and oil-red of liver tissues of MCD and normal group. (c) Bar plot showed the number of significant peaks from H3K9me3 and H3K27ac ChIP-seq data in both MCD and normal diet mice. (d) the distribution of H3K27ac peaks of all samples in different genome elements, calculated by HOMER module annotatePeaks.Pl. (e) the distribution of H3K9me3 peaks of all samples in different genome. (f) Heat map representing correlations based on H3K9me3 and H3K27ac occupancy on mice genome. Correlations were calculated by the Spearman correlation coefficient. elements, calculated by HOMER module annotatePeaks.Pl. (g) PCA plot of ChIP-seq data. PCA analyses to classify H3K9me3 and H3K27ac modification between MCD and normal liver tissues. (h) Metagene plot of mean ChIP-seq signal of H3K9me3 and H3K27ac across the individual peaks of each mark. Metagene analysis was centered on the middle of peaks and 10 kb around peak centers are displayed (5 kb upstream and 5 kb downstream). The ChIP-seq data of each histone marks were merged from three individual replicates (MCD group: n = 3; normal group: n = 3).

MCD: Methionine and choline deficiency diet; ChIP-Seq: Chromatin immunoprecipitation sequencing. PCA: Principal components analysis.

3.2. Characterization of H3K27ac and H3K9me3 in MCD and normal liver tissue

To explore the epigenetic dynamics in the mice hepatitis tissues, we performed ChIP-Seq of H3K27ac and H3K9me3 modifications with MCD (n = 3) and normal (n = 3) mice livers. The peak numbers and the proportion of genome distribution of the two modifications were not significant different between groups (Figure 1(c-e)). The H3K27ac peaks were mainly distributed on promoter-TSS, intergenic, introns, and some exons (Figure 1(d)). The H3K9me3 peaks were mainly distributed in the intronic and interactive loci (Figure 1(e)). Sample correlation and principal component analysis showed that the two modifications were clearly distinguishable as expected, indicating the reliability of our data quality (Figure 1(f,g)). To explore whether the two histone modifications changed between MCD and normal groups, we calculated the signal intensity of on their peak loci. Their signal intensities were both elevated in the livers of MCD mice (Figure 1(h)), and the expression levels of most of these enzyme genes of H3K27 acetyltransferase and H3K9 methyltransferase were significantly up-regulated (Supplementary Fig S2A&B), indicating epigenetic reprogramming of H3K27ac and H3K9me3 occurred in the MCD-diet induced MASLD model.

3.3. Reprogrammed enhancers in MCD liver tissues are associated with metabolism and inflammation

Enhancers play important roles in the development of diseases. We identified active enhancer loci (ELs) and VELs according to the protocol described in the method section. We totally identified 26,663 active ELs, 1,638 gain VELs and 1,854 lost VELs in MCD liver tissues (Figure 2(a)). The H3K27ac signal on gain/lost VELs has been significantly altered (Figure 2(b-d)). GO analysis showed that the adjacent genes of both gain and lost VELs were enriched in lipid metabolism and other metabolic pathways, suggesting a switch of metabolic programs occurred in the liver of MCD (Figure 2(d)). In addition, the proximal genes of gain VELs, but not the lost VELs, in the MCD livers were enriched in inflammation and immune-related processes, suggesting that the enhancer activation was tightly associated with inflammation in the MCD livers (Figure 2(d)). Motif analysis of the gain VELs predicted activated transcription factors, including Fra2, Fosl2, Jun-Ap1, Fos, Hnf4a, Pparα, Foxo1, etc (Figure 2(e)), which are all well-known in the regulation of inflammation and metabolism. To further verify the changes in enhancers during MASLD development, we used public datasets (GSE242881) from the early stages of MASLD mice models of HFD [8]. We overlapped the proximal genes of the gain VELs of MCD and HFD, and performed gene enrichment analysis on the three parts of the genes (Supplementary Fig. S3A-D). We found that the overlapped VELs of MCD and HFD models were associated with metabolic processes and gene expression regulation (Supplementary Fig. S3B). The MCD gain VELs were associated with metabolic processes and immune process regulation, while HFD gain VELs were only associated with metabolic processes (Supplementary Fig. S3C&D), supporting that the epigenetic reprogramming of inflammation genes is critical for MASLD development.

Figure 2.

Identification of variant enhancer loci in methionine and choline deficiency diet liver tissues. (a) Enhancers and VELs statistics. (b) the heat map showed the H3K27ac signal for VELs of MCD group and normal group. (c) the average H3K27ac signal (RPKM) on gain/lost VELs. (d) Bar plots showed the enrichment analysis of biological processes of proximal genes for MCD gain/lost VELs. (e the top 20 transcription factors enriched in the MCD gain enhancer loci. (f) Identification of SEs in MCD group. (g) Identification of SEs in normal group. (h) the volcano diagram showed the number of VSELs. The threshold is fold change ≥ 1.5 and p value < 0.05. (i) Bar plots showed the enrichment analysis of biological processes of proximal genes for MCD gain VSELs. (j) WashU epigenome browser view to show the ChIP-seq density of H3K27ac on the SEs in Cd36 gene.

MCD: Methionine and choline deficiency diet; VELs: Variant enhancers loci; VSELs: Variant super-enhancer loci; ChIP-Seq: Chromatin immunoprecipitation sequencing. SEs: super enhancers.

3.4. Gain VSELs in the liver of MCD are associated with inflammation

SEs are a cluster of enhancers containing much stronger regulatory effects on gene expression and cell functions compared with typical enhancers. SEs are often considered as hallmarks for cell identification [25,26]. To investigate the dynamics of SEs in our MASLD mice model, we analyzed SEs and identified 926 SEs in the MCD group and 951 SEs in the normal diet group (Figure 2(f,g)), including 81 gain VSELs and 57 lost VSELs (Figure 2(h)) in the liver tissues of the MCD group. The proximal genes of gain VSELs were enriched in inflammation and immune-related pathways (Figure 2(i)). Cd36 is a scavenger receptor and plays important roles in lipid metabolism and inflammatory responses [27]. Cd36 were selected as a sample and the WashU epigenome browser view showed the H3K27ac upregulation on SEs (Figure 2(j)). Our previous study showed that only very few SEs were changed in the liver tissues of a HFD-diet fat liver mice model [8]. The comparison of the two studies imply that change of SE activity is more critical for the MCD-diet liver and contributes to the expression of inflammatory genes.

3.5. Reprogramming of H3K9me3 in MCD diet liver tissues

H3K9me3 is one of the important inhibitory histone modifications which is distributed mostly in heterochromatin regions. After analyzing the distribution of H3K9me3 loci, we identified 1130 h3K9me3 acquisition sites and 1663 h3K9me3 loss sites in MCD liver (Figure 3(a)). The proximal genes of lost H3K9me3 loci were mainly enriched in immune and inflammation-related processes, which indicated that changes of certain immune pathways in the liver of MCD were associated with H3K9me3 reprogramming (Figure 3(b)). The proximal genes of gain H3K9me3 loci are mainly enriched in metabolically related processes, which indicated that the changes of metabolic processes in the MCD liver are associated with H3K9me3 reprogramming (Figure 3(c)).

Figure 3.

Identification of variant H3K9me3 loci in methionine and choline deficiency diet liver tissues and heterochromatin loci enhancer analysis. (a) H3K9me3 loci and variant H3K9me3 loci statistics. (b) Bar plots showed the enrichment analysis of biological processes of proximal genes for MCD lost H3K9me3 loci. (c) Bar plots showed the enrichment analysis of biological processes of proximal genes for MCD gain H3K9me3 loci. (d) the upper plot showed the average H3K9me3 ChIP-seq density across enhancers in lost H3K9me3 loci between MCD and normal liver tissues. The lower plot showed the average H3K27ac ChIP-seq density across enhancers in lost H3K9me3 loci between MCD and normal liver tissues. (e) Bar plots showed biological process enrichment analysis of proximal genes for enhancers in lost H3K9me3 loci. Items were ordered by the P-value.

MCD: Methionine and choline deficiency diet; ChIP-Seq: Chromatin immunoprecipitation sequencing.

It has been reported that H3K9me3 reprogramming is often associated with alteration of enhancer activity [12,28]. We combined all the enhancer loci of the MCD and normal group, and then overlapped the lost H3K9me3 loci with them. Subsequently, we calculated the signals of H3K27ac and H3K9me3 on these enhancers. The H3K9me3 signals on the enhancers in the lost H3K9me3 loci were indeed reduced, while their H3K27ac signals were enhanced (Figure 3(d)). Their proximal genes were enriched in immune and inflammation-related pathways (Figure 3(e)), suggesting that the enhancers in the lost H3K9me3 loci are associated with the inflammatory processes in the MCD livers. Then, we overlapped the proximal genes of the lost H3K9me3 loci of MCD and HFD, and performed gene enrichment analysis on the three groups of genes (Supplementary Fig. S4A-C). We found that the lost H3K9me3 loci in MCD tissues were associated with inflammatory responses and regulation of immune processes, while lost H3K9me3 loci in HFD tissues were associated with cell adhesion and gene expression processes (Supplementary Fig. S4B&C).

3.6. Identification of inflammation-related genes in MCD livers

Based on the above analyses, we focused on the proximal genes of gain VELs and lost H3K9me3 loci in MCD mice liver. By intersecting different gene sets, we identified three genes, Il34, Irf8 and Slc7a1 as candidate genes (Figure 4(a,b)). The immune-related genes were extracted from the mice GO-BP dataset (https://www.gsea-msigdb.org/gsea/msigdb/mice/genesets.jsp?collection=GO:BP). We acquired RNA-Seq data of MCD-diet induced C57BL/6 mice models from the GEO database (GSE205974). The expression of these three genes in the liver of MCD mice was higher than that in the control diet group (Figure 4(b,c,e,g)). It has been shown that Il34 gene could be used as a marker for liver fibrosis [29,30], but little is known about its functions in the development of MASLD. Although the H3K9me3 signal of Il34 gene did not change significantly in MCD liver, the H3K27ac signal on its promoter and enhancer significantly increased; while in the HFD diet mice, H3K9me3 signal close to its enhancer decreased significantly (Figure 4(d)). In the liver of MCD mice, the H3K27ac signals on Irf8 and Slc7a1 genes were increased, and their H3K9me3 signals were decreased (Figure 4(f,h)). In conclusion, our integrative analysis of multi-omic data revealed that Il34, Irf8 and Scl7a1 may be involved in the development of MASLD.

Figure 4.

mRNA levels and histone modification levels of theIl-34Irf8andSlc7a1. (a) the Venn diagram showed the overlap of immune genes set, MCD up regulated genes, proximal genes for gain VELs and lost H3K9me3 loci in MCD group. (b) the heat map showed the expression (FPKM) for Il-34, Irf8 and Slc7a1 in MCD and normal diet group. (c) the normalized expression of Il-34 according to the published datasets in MASLD mode. (d) WashU epigenome browser view to show the ChIP-seq density of H3K27ac and H3K9me3 on the Il-34 gene loci. (e) the normalized expression of Irf8 according to the published datasets in MASLD mode. (f) WashU epigenome browser view to show the ChIP-seq density of H3K27ac and H3K9me3 on the Irf8 gene loci. (g) the normalized expression of Slc7a1 according to the published datasets in MASLD mode. (h) WashU epigenome browser view to show the ChIP-seq density of H3K27ac and H3K9me3 on the Slc7a1 gene loci.

MCD: Methionine and choline deficiency diet; VELs: Variant enhancers loci; MASLD: Metabolic dysfunction-associated steatotic liver disease; FPKM: Frag-ments per kilobase of bin per million mapped reads; ChIP-Seq: Chromatin immunoprecipitation sequencing.

3.7. Validation of inflammation-related genes in MCD liver tissue

To further verify the changes of Il-34, Irf8 and Scl7a1 in MASLD, we used multiple public datasets to verify their gene expression level under different dietary patterns (MCD: GSE162863; HFD: GSE242881) and MASLD patients (GSE213621). We found that Il-34, Irf8, and Slc7a1 mRNA were significantly increased in the MCD mice model (Figure 5(a-c)). In the HFD diet model, only Il-34 mRNA increased significantly (Figure 5(d)), while no significant difference for Irf8 and Slc7a1 mRNA (Figure 5(e,f)). In MASLD patients, along with the development of liver fibrosis, IL-34 and SLC7A1 mRNA significantly increased (Figure 5(g,i)), while IRF8 mRNA was not (Figure 5(h)). Finally, to further confirm the results, we re-fed a batch of MCD-diet induced MASLD mice (MCD group: n = 5, normal group: n = 5) and found that Il-34 and Slc7a1 mRNA were significantly increased (Figure 5(j,k)), and H3K27ac modification on the enhancer was also significantly increased. However, no significant difference was observed for Irf8 (Figure 5(i)). Moreover, we found that IL-34 protein levels significantly decreased in the 6-week MCD and 28-week HFHC diet groups (Figure 5(m,n)), suggesting that IL34 may act as a protective factor in the development and progression of MASLD. Summary, Il-34 and Slc7a1 may be important regulatory genes in regulating the progression of MASLD, especially in promoting inflammation.

Figure 5.

Hepatic Il-34, Irf8 and Slc7a1 mRNA expression and H3K27ac modification levels in methionine and choline deficiency diet and normal diet mice. (a-c) Il-34, Irf8 and Slc7a1 mRNA in MCD mice model (GSE162863); (d-f) Il-34, Irf8 and Slc7a1 mRNA in HFD mice model (GSE242881); (g-h) Il-34, Irf8 and Slc7a1 mRNA in human MASLD samples (GSE213621); (j-l) Il-34, Slc7a1 and Irf8 mRNA and H3K27ac modification levels in 6-week MCD mice. (m) IL-34 protein level in 6-week MCD (n) IL-34 protein level in 28-week HFHC mice. MCD group: n = 10; normal group: n = 9. HFHC group: n = 7; normal group: n = 3. All experiments were performed in at least three technical replicates.

MCD: Methionine and choline deficiency diet; HFD: High-fat diet; MASLD: Metabolic dysfunction-associated steatotic liver disease. HFHC: High fat and high cholesterol.

3.8. Distribution of inflammation-related genes in mice livers

To investigate the expression of Il34, Slc7a1, and Irf8 in different liver cell types, we retrieved single cell RNA sequencing (scRNA-Seq) data from multiple public databases (MCD: GSE225868; HFHC: GSE210501) and conducted analysis. According to the provided data which contain information for various cell types except hepatocytes (Supplementary Fig. S5A&B), we found that Il34 was expressed in a small amount in fibroblasts and macrophages (Supplementary Fig. S5C&D). Furthermore, we identified various cell types such as hepatocytes, Kupffer cells, and endothelial cells in the liver of MASLD mice induced by HFHC diet and normal diet mice (Supplementary Fig. S6A&B). The Il34 gene was expressed in small amounts in Kupffer cells and endothelial cells (Supplementary Fig. S6C&D). The Slc7a1 and Irf8 genes were expressed in Kupffer cells, endothelial cells, and hepatocytes (Supplementary Fig. S6C&E&F). In summary, Il34 and Slc7a1 may be almost unexpressed in normal liver, but weakly expressed in Kupffer cells and endothelial cells during MASLD development.

4. Discussion

MASLD patients undergo several distinct stages before progressing to irreversible cirrhosis and hepatocellular carcinoma. Steatohepatitis, characterized by an extensive inflammatory infiltrate in the liver, was the central stage of MASLD. However, the epigenomic alterations caused by fat accumulation and inflammatory infiltration are currently unknown. Our previous findings identified significant modification alterations of H3K27ac and H3K9me3 in the early stages of MASLD, especially H3K9me3 may play important roles in the pathogenesis of MASLD by regulating enhancer accessibility, but SEs change very little [8]. In this study, using the MCD-diet induced MASLD mice model, we found that typical enhancers were mainly enriched in lipid metabolism and immune inflammation. SEs were also significantly altered in liver tissues of MASLD mice, mainly enriched in TNF, immune-inflammatory and lipid metabolic pathways, suggesting the cells of MCD-diet liver tissues experienced more dramatic changes compared with those in HFD-diet livers. In addition, we found that Il-34 and Slc7a1 may be important regulators in modulating the development of inflammation.

In both homeostasis and disease states, hepatic cis-regulatory networks were established through the coordinated action of liver-enriched transcription factors (TFs). These TFs defined enhancer landscapes that activated a wide range of gene programs with spatiotemporal resolution. In addition, genomic studies of MASLD patients and MASLD models have demonstrated that a generalized regulatory remodeling occurs in the liver of MASLD patients, which is reflected in aberrant gene expression profiles [31]. In this study, we found that enhancers of inflammation and immune related genes were activated in MASLD and motif analysis showed that transcription factors such as Fra2, Fosl2, Jun-AP1, Fos, Hnf4α, Pparα, Foxo1 were associated with MASLD. Fos-like antigen 2 (Fra2/Fosl2) belongs to the AP-1 family of transcription factors, which includes various isoforms of Fos and Jun [32]. Fra2/Fosl2 is involved in the regulation of cellular responses to a wide range of extracellular stimuli, stressors, and intracellular changes, and its aberrant expression or regulation leads to severe growth defects or a wide range of pathologies, including pulmonary fibrosis, chronic obstructive pulmonary disease, colon cancer, and hepatocellular carcinoma [33–36]. It has been shown that in MASLD, Fosl2 expression was significantly increased and promoted Ly6d transcription for inflammatory progression through binding to the Ly6d promoter, whereas knockdown of Fosl2 significantly inhibited the MASLD-associated hepatocyte apoptosis [37]. In obese MASLD patients, significant activation of the pro-inflammatory transcription factor AP-1 was significantly associated with oxidative stress and insulin resistance and was involved in the progression of inflammation in MASLD as well as the formation of hepatic fibrosis in concert with the transcription factor Jun [38,39]. Hnf4α regulates regulatory elements in promoters and enhancers of genes related to cholesterol, fatty acid and glucose metabolism, and activates hepatic gluconeogenesis and regulates the expression of several genes including lipoproteins [40–42]. Hnf4α expression was significantly upregulated in MASLD samples compared with healthy livers, and the expression of this TF increased with increasing MASLD score in MASLD samples [43]. These findings suggest that Hnf4α may be a core gene in the pathogenesis of MASLD. Pparα is a ligand-activated transcription factor that is abundantly expressed in the liver and efficiently induces the expression of genes related to a variety of lipid metabolism pathways including microsomal, peroxisomal, and mitochondrial fatty acid oxidation, synthesis and catabolism of TG and lipid droplets, lipoprotein metabolism, gluconeogenesis, bile acid metabolism, and various other metabolic pathways and genes [44]. Foxo1 was a conserved transcription factor involved in energy metabolism that mediated the transcription of a variety of genes downstream of metabolic regulation, including diabetes, obesity, nonalcoholic fatty liver disease, and atherosclerosis [45]. Foxo1 promoted gluconeogenesis, lipoprotein and TG secretion, but inhibited the expression of genes related to glucose utilization and lipogenesis in the liver [45]. In addition, Foxo1 expression was upregulated in hepatic macrophages and was associated with hepatic inflammation, steatosis and fibrosis in mice and MASLD patients [46]. The above suggests that reprogramming of lipid metabolism and inflammation occurs in the MASLD model, which plays an important role in promoting MASLD progression.

In addition to the above findings that enhancers are enriched for transcription factors associated with metabolic and inflammatory reprogramming, we further analyzed the roles of SEs in the MASLD model. SEs are clusters of enhancers densely distributed in non-coding positions that drive the expression of genes controlling cellular properties and diseases at higher levels [25,26]. Our previous study found that SEs in hepatocytes in the early stages of MASLD did not change much, suggesting that cellular properties may not change [8]. However, in the MASLD model, we found many changes in SEs, with 57 up-regulated and 81 down-regulated SEs, mainly enriched in the inflammatory and lipid metabolism signaling pathways. Cd36 was a membrane glycoprotein on the cell surface binding fatty acids to facilitate lipid transport. Its expression was significantly increased in response to lipid overload or activation of nuclear receptors in livers [47]. The activation of various signaling pathways, including Ppars, Ampk, and miRNAs pathways, has been linked to Cd36 expression, which provides a way to molecularly manipulate Cd36-lipid metabolism [47–50]. The identification of Cd36 on SEs in our study implies a potential novel mechanism for the regulation of Cd36 in livers, similar to a recent study [51].

In addition, we focused on the changes in inflammation-related genes that were epigenetically regulated, looking for key genes that may regulate the development of MASLD. We found that Il-34 and Slc7a1 may be important key genes that regulate the development of inflammation in MASLD, which have not been previously reported. IL-34 is a homodimeric protein widely expressed in brain, heart, liver, spleen, kidney, and colon tissues, which binds to colony-stimulating factor-1 receptor (CSF-1 R), receptor-type protein tyrosine phosphatase zeta (RPTP-ζ), and the transmembrane heparan sulfate proteoglycan syndecan-1, and modulates macrophage polarization and influenced the immune-inflammatory response [52,53]. A study showed that the expression of Il-34 in adipose tissues was significantly increased in obese populations, and that TNF-α and IL-1β stimulated adipocytes to further produce IL-34, which promoted adipose accumulation and inhibited the stimulatory effect of insulin on glucose transport [54]. Another study showed that IL-34 could be produced by activated fibroblasts in MASLD and that serum IL-34 level was strongly associated with liver fibrosis, which could be used as a diagnostic biomarker for MASLD/MASH patients [29,30]. Thus, IL-34 may play important roles as a cytokine in the regulation of lipid metabolism and immune inflammation. We found that Il-34 mRNA expression was significantly increased in the livers of MASLD mice, and its H3K27ac modification was also increased, suggesting that Il-34 may be involved in the development of inflammation in MASLD, but the specific mechanism needs to be further investigated. Surprisingly, we also found that IL-34 protein levels were significantly reduced in 6-week MCD and 28-week HFHC mice models, a phenomenon not previously reported. Studies have shown that the liver microenvironment dynamically regulates macrophage polarization into pro-inflammatory M1 and anti-inflammatory M2 subtypes [55]. IL-34 suppresses inflammation by driving M2 macrophage polarization, protecting the liver from autoimmune hepatitis [56]. Il-34 knockout increases pro-inflammatory cytokines, decreases anti-inflammatory cytokines and M2 markers, and exacerbates inflammation and bile duct injury in primary biliary cholangitis mice models [57], suggesting a critical role of IL-34 in maintaining liver immune homeostasis. Additionally, IL-34 expression was reduced in patients with chronic hepatitis B virus (HBV) infection. In vitro experiments had demonstrated that supplementation with IL-34 could modulate the host’s innate or adaptive immune response, thereby inhibiting HBV replication [58]. Single-cell sequencing analysis revealed that Il-34 was primarily expressed by fibroblasts, macrophages, Kupffer cells, and endothelial cells, but at minimal levels. Additionally, studies had shown that serum IL-34 levels increase with the progression of liver fibrosis in MASLD patients [29,30]. We speculated that hepatic IL-34 may be secreted into the serum or originate from other tissues. Despite epigenetic regulation promoting Il-34 mRNA expression, the inability of endogenous or exogenous IL-34 protein to reach the liver resulted in significantly reduced hepatic IL-34 levels. This reduction limited macrophage polarization, thereby exacerbating MASLD. Supplementing IL-34 cytokines might help alleviate MASLD, but further studies are needed to clarify the sources and regulatory mechanisms of IL-34. SLC7A1 was expressed in most human tissues and its main function is to transport ornithine, lysine and arginine [59]. Current studies have shown that Slc7a1 is associated with hepatocellular carcinoma, colorectal carcinoma, esophageal carcinoma, renal carcinoma, lung carcinoma, and hypertension, and is involved in T cell growth and proliferation [60–62]. Our study also found that Slc7a1 expression was significantly increased in different dietary patterns and patients, and histone modifications were also significantly altered; thus, we hypothesized that Slc7a1, as an arginine transporter protein, may have an important role in the development of MASLD. However, Slc7a1 has not yet been reported in MASLD. Thus, Il-34 and Slc7a1 are expected to be new key genes for early intervention and blocking the progression of MASLD and deserve further investigation.

In the current study, we found that H3K27ac and H3K9me3 of Il-34 and Slc7a1 are reprogrammed in MCD-diet mice liver, which are correlated with their mRNA expression level. H3K27ac on enhancers is catalyzed by histone acetylase p300 and CBP [63]; H3K9me3 by histone methylases such as SETDB1/2 and SUV39H1/2 [64]. The altered epigenetic alterations on the identified genome elements may be helpful in disease diagnosis. It is also possible to develop novel strategies for disease treatment, through inhibiting the enzyme activities using small molecular chemicals, or by loci-specific genome editing. Moreover, IL-34 is a cytokine and it will be interesting to test its functions by directly injecting it into animals or overexpressing it in livers. These might provide novel insights for the development of disease treatment.

5. Conclusion

In summary, both H3K27ac and H3K9me3 are reprogrammed in the live tissues of MCD-diet mice; and the VELs and VSELs were enriched in lipid metabolism and immune inflammation-related pathways. In addition, our newly identified Il-34 and Slc7a1 may be the key inflammation-related genes regulating the progression of MASLD, and the functions and molecular mechanisms of these genes require further studies in the future.

Funding Statement

This manuscript was funded by National Key Research and Development Program of China to Min Wu [2023YFA0913400] and to Mingkai Chen [2023YFC2507405], the Fundamental Research Funds for the Central Universities [2042022dx0003], National Natural Science Foundation of China to Lian-Yun Li [32170718] and Min Wu [32470620], The Interdisciplinary Innovative Talents Foundation from Renmin Hospital of Wuhan University to Mingkai Chen [JCRCFZ-2022-017]. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Article highlights

• H3K27ac and H3K9me3 are reprogrammed in the live tissues of MCD-diet mice.

• The target genes of the enhancers with increased H3K27ac and decreased H3K9me3 signals are enriched in lipid metabolism and immuno-inflammatory pathways.

• The presence of 81 gain VSELs and 57 lost VSELs in the liver tissues of the MCD group compared to the normal group. The proximal genes of gain VSELs were enriched in inflammatory and immune-related pathways.

• Cd36 was identified as a gain VSELs, which is a scavenger receptor that plays an important role in lipid metabolism and inflammatory responses.

• Il-34 and Slc7al are identified as potential regulators in modulating inflammation in MASLD.

Author contributions

Chuanfei Zeng and Huan Li performed most of the experiments; Mingliang Wei did data analysis; Linxin Yu, Chuang Wang, Ziqi Mu, Ziyin Huang and Yujia Ke helped in animal experiments; Yong Xiao, Lian-Yun Li, Min Wu and Ming-Kai Chen discussed the project and provided suggestions; Ming-Kai Chen and Min Wu directed the study; Chuanfei Zeng, Huan Li and Mingliang Wei wrote the manuscript; Yong Xiao, Lian-Yun Li, Min Wu and Ming-Kai Chen edited the manuscript.

Disclosure statement

The authors have no relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript. This includes employment, consultancies, honoraria, stock ownership or options, expert testimony, grants or patents received or pending, or royalties.

No writing assistance was utilized in the production of this manuscript.

Ethical declaration

All experimental procedures involving mice operation were conducted in accordance with the laboratory animal guidelines of Wuhan University and approved by the Animal Experimentations Ethics Committee (Protocol NO. 14110B).

Data availability statement

Availability of NGS data

The raw ChIP-seq data used in this study are available in the Gene Expression Omnibus (GEO, https://www.ncbi.nlm.nih.gov/geo/) database under accession number (GSE266077). We acquired MASLD-related RNA-seq and ChIP-seq data from the GEO database to aid our research. RNA-seq data of MCD-diet induced C57BL/6 mice models were obtained from the GEO database (GSE205974, GSE162863). RNA-seq data of HFD-induced C57BL/6 mice models were obtained from the GEO database (GSE242881). RNA-seq data from human clinical samples of MASLD liver tissue were obtained from the GEO database (GSE213621). ScRNA-Seq data of MCD-diet induced mice liver tissue were obtained from the GEO database (GSE225786). ScRNA-Seq data of GAN-induced mice liver tissue were obtained from the GEO database (GSE210501). H3K27ac and H3K9me3 ChIP-Seq of HFD-induced C57BL/6 mice models were obtained from the GEO database (GSE226171). The expression level of genes related to immunity and inflammation were calculated by the DESEQ2 (version 1.40.1, https://bioconductor.org/pack-ages/release/bioc/html/DESeq2.html) and normalized by FPKM.

Supplementary material

Supplemental data for this article can be accessed online at https://doi.org/10.1080/17501911.2025.2467028