Differential metabolites facilitate metabolic dysfunction-associated fatty liver disease malignancy via immune evasion and M2-polarized macrophages

Abstract

BACKGROUND

Differential metabolites (DMs) are associated with metabolic dysfunction-associated fatty liver disease (MAFLD) malignant transformation. However, their underlying mechanisms remain to be identified.

AIM

To investigate the dynamic alterations of DMs, carnitine palmitoyl transferase-II (CPT-II) and immune cells during MAFLD malignancy.

METHODS

A rat model was constructed with high fat diet plus 2-fluorenylacetamide to induce hepatocyte malignancy. Livers were divided into MAFLD, metabolic dysfunction-associated steatohepatitis, liver cirrhosis (LC) and hepatocellular carcinoma (HCC) groups based on hematoxylin and eosin staining, with normal rats as control. DMs were identified via RNA transcriptomics or metabolomics. Proteins were detected by western blotting, and immune cells were analyzed by single-cell sequencing.

RESULTS

Model livers with obvious lipid accumulation and cells were examined during MAFLD malignancy from inflammation or necrosis to LC or HCC and exhibited a pathological alteration of nuclear pleomorphism, disordered arrangement and a progressive decrease in CPT-II activity. The number of DMs was 131 in MAFLD, 134 in metabolic dysfunction-associated steatohepatitis, 27 in liver fibrosis/LC, and 130 in HCC, respectively. Cyclin B1 and cyclin-dependent kinase 1 were involved in the P53 pathway and cell cycle, and they held key positions in the protein interaction network, which involved metabolic regulation of cell response to stimuli. DMs, such as phosphatidylcholine or sphingomyelin in steroid biosynthesis, were significantly related to MAFLD malignancy. Mechanistically, liver immune cells undergo dynamic changes in a fat-rich microenvironment, with decreased T cell abundance and increased programmed death ligand 1 expression and M2-polarized macrophages.

CONCLUSION

Downregulated CPT-II aggravates the accumulation of metabolites associated with MAFLD malignancy via immune evasion and M2-polarized macrophages.

Key Words: Differential metabolites; Carnitine palmitoyl transferase-II; Immune escape; Differentially expressed genes; M2-polarized macrophages; Hepatocarcinogenesis

Core Tip: Dynamic alterations in metabolites and immune cell populations provide insight into metabolic dysfunction-associated fatty liver disease-related malignancy. These changes include loss of carnitine palmitoyltransferase II activity and significant alterations in phosphatidylcholines and sphingomyelins associated with steroid biosynthesis. Additionally, the fat-rich microenvironment is characterized by immune modulation, including reduced T-cell function, increased in programmed death ligand-1 expression that facilitates immune escape, and enrichment of M2-polarized macrophages, all of which contribute to the progression of metabolic dysfunction-associated fatty liver disease-related malignancy.

INTRODUCTION

Metabolic dysfunction-associated fatty liver disease (MAFLD), previously known as nonalcoholic fatty liver disease, represents a major health burden[1,2]. Emerging metabolomics-based studies have suggested links between lipid metabolism and MAFLD risk, which arises from excessive lipid accumulation in the liver due to an imbalance between lipid anabolism and catabolism[3]. MAFLD is the most common cause of chronic liver disease and is characterized by abnormal lipid metabolism. Excess lipid accumulation in the liver triggers inflammatory responses that can progress to metabolic dysfunction-associated steatohepatitis (MASH), fibrosis, or hepatocellular carcinoma (HCC)[4]. The MAFLD disease spectrum includes simple steatosis, MASH, metabolic dysfunction-associated liver cirrhosis (LC), and HCC. Although MAFLD is often relatively benign in its early stages, the lobular inflammation that characterizes MASH is considered a key driver of MAFLD-related malignancy[5,6]. Alterations in differentially expressed genes (DEGs)[7] and differential metabolites (DMs)[8], along with reduced activity of mitochondrial carnitine palmitoyl transferase-II (CPT-II) in fat-rich environments, represent important components of MAFLD pathogenesis[9]. These changes involve the activation of multiple signaling pathways and contribute to hepatic immune dysregulation. Both the innate and adaptive immune systems play crucial roles in disease progression, with complex crosstalk between hepatic and immune cells driving MAFLD development[10,11].

Hepatic mitochondrial dysfunction has been recognized as an important contributor to abnormal lipid accumulation in liver, including mitochondrial DNA damage and impaired β-oxidation damage in MASH[12,13]. Hepatocyte inflammation arises from complex interactions among cell death mediators, cytokines, and resident immune cells, collectively forming the hepatic microenvironment. Previous studies have shown that MAFLD pathogenesis involves the activation of multiple signaling pathways, immune dysregulation, endoplasmic reticulum stress, reactive oxygen species formation, and mitochondrial damage[14]. Both the innate and adaptive immune systems play crucial roles, with intricate crosstalk between hepatic and immune cells driving MAFLD progression[15,16]. The interplay among these inflammatory mediators, shifts in immune responses, and immune cell-mediated injury ultimately determines the progression of MAFLD pathology. However, the underlying mechanisms remain incompletely understood.

Therefore, the objective of this study was to investigate the roles of dynamic differential metabolites (DMs), CPT-II, and immune cell alterations in MAFLD malignancy, with the aim of providing new evidence for early prevention and targeted interventions.

MATERIALS AND METHODS

Dynamic model of MAFLD

The dynamic transformation model of hepatocytes subjected to lipid accumulation was established according to previously published protocols. All animal experiments were approved (No. P20230327-001) and conducted in accordance with the guidelines of the Animal Care and Use Committee of Nantong University, China.

Male Sprague-Dawley rats (SD, 4-week-old) were fed a high-fat diet (HFD; 43.7% fat, 36.6% carbohydrate, 19.7% protein, 0.2% cholesterol) supplemented with 0.05% 2-fluorenylacetamide (Sigma, MO, United States) for 16 weeks to induce MAFLD and hepatocarcinogenesis[1]. Age-matched SD rats fed a normal control diet (NC; 18% fat, 58% carbohydrate, 24% protein) served as controls[16].

Animals were monitored at different time points throughout the experiment. One group of rats was sacrificed every 2 weeks, and liver tissues and blood samples were collected for subsequent analyses.

Pathological grouping and Oil Red O staining

Liver sections were stained with hematoxylin and eosin and divided into the MAFLD, MASH, LC, HCC and NC groups. Liver sections were generated and stained with 0.5% Oil Red O solution (Jiangcheng Bioeng. Ins., Nanjing, Jiangsu Province, China), then observed and photographed under an Olympus light microscope (IX71-A12FL/PH, Japan). The red lipid/total area ratio in each microscopic field was determined using Image-Pro Plus 6.0.

Biochemical determination

Livers and fasting blood samples were collected at baseline and every 2 weeks. The laboratory data included the levels of circulating aspartate transaminase, alanine transaminase, alpha-fetoprotein, and lipids, such as total cholesterol, low-density lipoprotein, and high-density lipoprotein.

RNA isolation

Liver tissues from model rats were placed in RNase-free tubes containing RNAlater (Ambion, TX, United States). Total RNA was extracted from liver tissues using TRIzol reagent (Gibco BRL, CA, United States) according to the manufacturer’s instructions. Total RNA (1 μg) was reverse-transcribed into complementary DNA (cDNA) using random primers and reverse transcriptase (Gibco BRL, CA, United States). The resulting cDNA was subsequently amplified by polymerase chain reaction (PCR) following the manufacturer’s protocol.

Metabolite measurements

Quantification of 630 metabolites including lipids was performed using a tandem mass spectrometry (MS/MS) MxP^® Quant 500 Kit (Biocrates Life Sciences, Innsbruck, Austria) according to the manufacturer’s protocol. Briefly, after preprocessing, samples were analyzed using flow-injection analysis-MS/MS on a SCIEX 5500 QTrapTM (SCIEX, Darmstadt, Germany) for lipids and liquid chromatography-MS/MS for small molecules utilizing an Agilent 1290 Infinity II liquid chromatography (Santa Clara, CA, United States) linked with a SCIEX 5500 QTrapTM employing multiple reaction monitoring to detect analytes. After data preprocessing and normalization, peak integration and metabolite concentration were determined using METIDQTM software (Biocrates, Innsbruck, Austria), quantifying 106 small molecules and free fatty acids in chromatography mode and 524 complex lipids in positive flow-injection mode (flow-injection analysis-MS/MS), exploring a broad range of metabolic pathways.

As a result, 436 metabolites qualified for further analysis. Additionally, short-chain fatty acids (acetic, propionic, and butyric acids) and metabolites involved in the tricarboxylic acid cycle (citrate, cis-aconitate, isocitrate, α-ketoglutarate, succinate, fumarate, malate, lactate, pyruvate, and oxaloacetate) were quantified with laboratory-developed tests using liquid chromatography-MS/MS. When the level of a metabolite in a subject was recorded as 0, it was imputed as half of the smallest nonzero value observed among all participants. Principal component analysis (PCA) of the DMs was performed using volcano maps, orthogonal partial least squares-discriminant analysis (OPLS-DA) and S-shaped OPLS-DA plot analysis (Splot-OPLS-DA)[17].

Differential gene expression analysis

Gene expression in liver samples from different groups was analyzed using the DESseq2 R package, and differentially expressed genes (DEGs) were identified using the following criteria: |log₂FC| > 1.5, P < 0.05. The |log₂FC| values of the top 10 DEGs are presented in a heatmap. DEG data from liver tissues at different disease stages were used for subsequent functional enrichment analysis[18].

Gene Ontology (GO) analysis was performed to annotate biological processes, molecular functions, and cellular components. In addition, GO and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis of DEGs were performed using the clusterProfiler R package[19].

Single-cell RNA sequencing

Single-cell suspensions were prepared from five liver tissues (n = 5) representing different disease stages, healthy liver, MAFLD, MASH, LC and HCC, as confirmed by hematoxylin and eosin (H&E) staining. Cell viability was assessed, and the suspensions were filtered through a cell strainer and centrifuged at 300 × g at 4 °C for 5 min. After removal of the supernatant, 1000 μL of cell protective solution was added, and the cells were resuspended for downstream analysis.

Single-cell transcriptome sequencing was subsequently performed using the 10X Genomics platform using the Gene Denovo under strict quality control. The number and proportion of hepatocyte populations at the single-cell level were analyzed using R (version 4.1.3), Cell Ranger (version 6.1.0), Seurat (version 4.1.0), and CellChat (version 2.1.0), based on established gene markers.

Protein-protein interactions

Protein-protein interaction (PPI) enrichment analysis of DEGs at different stages was performed via STRING (version 10.5) online software (https://stringdb.org/), resulting in a PPI network (score > 0.4) that was visualized using Cytoscape 3.7.1. The MCODE plug-in identified the most important module in the PPI network. Hub genes were screened via the Cytohubba plugin, and the top 10 genes were selected as hub genes by using the maximal clique centrality algorithm.

Statistical analysis

Livers were divided into HCC, LC, MASH, MAFLD and NC groups according to H&E staining. Data are expressed as the mean ± SD and analyzed via SPSS 20.0, t tests, one-way ANOVA, and linear correlation. Comparative analysis of DMs was performed via PCA, OPLS-DA, Splot-OPLS-DA and correlations with overall distribution, multivariate analysis and outliers and plotted with R, GraphPad Prism 5.0, Image Pro-Plus 6.0 software or hierarchical clustering. Heatmaps and TreeView were generated via Cluster 3.0 or STRING software. T-distributed stochastic neighbor embedding was used for visualizing tissue data. A P value less than 0.05 was considered statistically significant.

RESULTS

Dynamic pathology and DEG transcription during MAFLD malignancy

Dynamic histopathological changes and DEG transcription profiles during MAFLD malignant transformation are shown in Figure 1. Gross examination revealed that control livers were bright red, whereas livers from HFD-fed rats appeared yellow, with small nodules forming on the liver surface during hepatocarcinogenesis. Based on histological changes, liver samples were classified into five groups: NC, MAFLD, MASH, LC, and HCC (Figure 1A).

Figure 1 Dynamic pathology and differentially expressed genes in the liver during metabolic dysfunction-associated fatty liver disease malignancy. A: Livers of Sprague-Dawley rats fed a high-fat diet plus 2-fluorenylacetamide; B: Liver grouping according to hematoxylin and eosin staining; C: Lipid analysis via Oil Red O staining; D: Enrichment analysis of differentially expressed genes associated with metabolic dysfunction-associated fatty liver disease malignancy; E: Lipid metabolism differentially expressed genes network. NC: Normal control; MAFLD: Metabolic dysfunction-associated fatty liver disease; MASH: Metabolic dysfunction-associated steatohepatitis; LC: Liver cirrhosis; HCC: Hepatocellular carcinoma; H&E: Hematoxylin and eosin.

At the early stage of MAFLD, hepatocytes exhibited cytoplasmic granular degeneration and inflammatory infiltration. In the intermediate stage, hepatic plates became thickened, nuclear chromatin appeared coarse, and fibrosis and sclerosis were evident, with occasional cells displaying an increased nuclear-to-cytoplasmic ratio. In the late stage, cells were arranged in nests or thick cords, normal hepatic architecture was lost, nuclear size became variable, chromatin was coarse, and the nuclear-to-cytoplasmic ratio was further increased (Figure 1B).

HFD induced lipid accumulation in the hepatocyte microenvironment with the presence of fatty vacuoles (Figure 1C). Enrichment analysis revealed that the DEGs (Figure 1D) were primarily associated with steroid biosynthesis, terpenoid backbone biosynthesis, and fatty acid biosynthesis pathways, as well as immune-related pathways, including antigen presentation and graft-versus-host disease. DEGs in the MASH stage were primarily enriched in pathways related to steroid and fatty acid biosynthesis, peroxisome proliferator-activated receptors (PPAR) signaling, and amino acid metabolism, including glutathione, glycine, serine and threonine metabolism. In the LC stage, DEGs were enriched in steroid biosynthesis, PPAR signaling, arachidonic acid metabolism, amino acid metabolism, the cell cycle, and the p53 pathway associated with HCC. In the HCC stage, DEGs were enriched in steroid biosynthesis, PPAR signaling, focal adhesion, extracellular matrix-receptor interaction, glutathione or histidine metabolism, and the p53 and cell cycle pathways.

Overall, the number of enriched DEGs was 790 vs 103 in the biological process category, 119 vs 24 in the cell component category, 254 vs 116 in the molecular function category, and 26 vs 9 in the Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway category (P < 0.05). PPI network analysis was performed using the MCODE plug-in in Cytoscape, which identified hub genes in the comparison between HCC and MAFLD (Figure 1E). Additionally, serum levels of total cholesterol, triglycerides (TG), and low-density lipoprotein increased, whereas high-density lipoprotein level decreased (Table 1). Levels of alanine transaminase, aspartate transaminase, and alpha-fetoprotein also increased significantly during the progression of MAFLD. These data indicate that MAFLD malignancy is associated with hepatocyte injury and dynamic alterations in DEGs.

Table 1 Dynamic levels of lipids, enzymes and alpha-fetoprotein in the sera of the metabolic dysfunction-associated fatty liver disease rats.

Group	n	Tch (mmol/L)	TG (mmol/L)	HDL (mmol/L)	LDL (mmol/L)	ALT (IU/L)	AST (IU/L)	AFP (ng/mL)
NC	6	1.32 ± 0.49	0.61 ± 0.19	2.20 ± 0.42	0.38 ± 0.09	14.75 ± 3.02	15.88 ± 3.29	0.88 ± 0.09
MAFLD	12	2.72 ± 0.46a	0.99 ± 0.41a	0.67 ± 0.20a	1.19 ± 0.40a	44.61 ± 14.34a	44.86 ± 9.90a	0.96 ± 0.30
MASH	11	3.76 ± 0.54a	1.78 ± 0.25a	1.36 ± 0.17a	0.65 ± 0.09a	62.19 ± 29.79a	44.46 ± 9.25a	1.46 ± 0.41
LC	13	3.29 ± 0.24a	1.81 ± 0.27a	1.17 ± 0.13a	0.75 ± 0.09a	55.75 ± 26.89a	61.99 ± 18.21a	1.60 ± 0.42
HCC	6	3.50 ± 0.60a	1.22 ± 0.21a	1.47 ± 0.11a	0.63 ± 0.05a	46.08 ± 13.39a	47.91 ± 21.22a	1.83 ± 0.54a
F value		21.341	13.014	52.861	32.857	8.128	11.600	8.015
P value		< 0.001	< 0.001	< 0.001	< 0.001	< 0.001	< 0.001	< 0.001

^aP < 0.001 vs normal control group.

NC: Normal control; MAFLD: Metabolic dysfunction-associated fatty liver disease; MASH: Metabolic dysfunction-associated steatohepatitis; LC: Liver cirrhosis; HCC: Hepatocellular carcinoma; Tch: Total cholesterol; TG: Triglyceride; HDL: High-density lipoprotein; LDL: Low-density lipoprotein; ALT: Alanine aminotransferase; AST: Aspartate transaminase; AFP: Alpha-fetoprotein.

Promoting roles of DEGs during MAFLD malignancy

Dynamic up-/down-regulated DEG transcripts during MAFLD malignancy progression are shown in Table 2. The significantly upregulated DEGs were BAX, CCND1, CDK1, CPY11, GAL3ST1, HEXB-β, LDHB, PPARγ, and ST3GAL5 (P < 0.001), which are mainly involved in liver injury, apoptosis, the cell cycle and biosynthesis signaling pathways, whereas CPT-II, YP51, EGR1, HOXD3, MTHFR, SAT2 and TYMP, which are associated with mitochondrial damage and lipid metabolism disorders, were significantly downregulated (P < 0.001), especially in the progression of MASH to HCC. Comparative analysis of the top 10 DEGs at different stages of MAFLD malignancy is shown in Table 3. The genes involved in these stages are different, except that they share the common features of lipid accumulation, hepatocyte damage and inhibited steroid biosynthesis. Compared with those in the NC group, significant alterations were observed during the early stage in terpenoid backbone biosynthesis, immune-related networks, antigen presentation, unsaturated fatty acid synthesis, and diabetes-related genes; the main features during the middle stage were glutathione and glycine/serine/threonine metabolism, the p53 pathway, gap junctions, unsaturated fatty acid synthesis, the PPAR signaling pathway, retinoid metabolism, the cell cycle, and xenobiotic metabolism by cytochrome P450. However, significant alterations (P < 0.001) in the later stages were associated with adhesion molecules, extracellular-receptor interactions, the cell cycle, the PPAR pathway, glutathione, the p53 pathway, histidine, and xenobiotic metabolism by cytochrome P450, especially in HCC-related signaling pathways. The alterations in the DEGs indicated that many genes related to energy metabolism, molecular interactions, steroid biosynthesis, the cell cycle and p53 are involved in MAFLD-related hepatocarcinogenesis.

Table 2 Dynamic transcription of metabolic dysfunction-associated fatty liver disease-related genes (mRNAs), mean ± SD.

Gene	n	NC	MAFLD	MASH	LC	HCC
BAX (× 104)	6	0.64 ± 0.91	0.66 ± 0.09	1.86 ± 0.39b	1.93 ± 0.73b	1.56 ± 0.16b
CCND1 (× 104)	6	1.64 ± 0.88	2.54 ± 1.35	6.44 ± 2.04c	4.61 ± 1.04b	4.88 ± 1.90b
CDK1 (× 103)	6	1.22 ± 0.59	1.46 ± 0.92	7.75 ± 2.48c	7.32 ± 1.38c	8.01 ± 2.48c
CPT-II (× 104)	6	5.41 ± 0.88	4.48 ± 0.31	4.02 ± 0.29a	3.85 ± 0.43a	3.66 ± 0.30a
CYP1A1 (× 104)	6	0.11 ± 0.05	0.08 ± 0.08	4.95 ± 3.79b	3.84 ± 2.18b	1.54 ± 0.32a
CYP51 (× 104)	6	4.90 ± 0.69	0.44 ± 0.13c	0.72 ± 0.30c	1.14 ± 0.38c	0.72 ± 0.18c
EGR1 (× 104)	6	1.64 ± 0.46	1.98 ± 0.40	1.74 ± 0.48	1.02 ± 0.10	0.94 ± 0.25
GAL3ST1 (× 103)	6	0.73 ± 0.23	0.86 ± 0.22	1.10 ± 0.20	0.89 ± 0.19	1.57 ± 0.50a
HEXB-β (× 104)	6	1.90 ± 0.54	2.61 ± 0.68	2.32 ± 0.40	2.88 ± 0.88a	3.27 ± 0.33b
HOXD3 (× 102)	6	8.21 ± 1.33	9.20 ± 2.17	8.16 ± 1.99	4.91 ± 0.62a	4.75 ± 1.22a
HSD17B7 (× 103)	6	5.18 ± 1.38	1.14 ± 0.18c	2.33 ± 0.47c	2.35 ± 0.46c	1.98 ± 0.77c
LDHB (× 103)	6	2.12 ± 1.16	2.01 ± 0.48	3.46 ± 1.54a	3.57 ± 1.27a	5.67 ± 3.31c
MTHFR (× 103)	6	2.20 ± 0.43	1.87 ± 0.50	1.45 ± 0.33	1.18 ± 0.34a	1.43 ± 0.32a
PPARγ (× 103)	6	0.19 ± 0.69	3.34 ± 0.67c	2.68 ± 0.74c	3.05 ± 0.72c	2.67 ± 0.97c
PTDSS1 (× 104)	6	2.26 ± 0.20	2.00 ± 0.17	1.86 ± 0.14a	1.88 ± 0.11a	1.90 ± 0.24a
SAT2 (× 104)	6	1.23 ± 0.22	1.48 ± 0.26	0.97 ± 0.22	0.69 ± 0.14a	0.48 ± 0.08a
SCD1 (× 104)	6	3.17 ± 1.98	1.33 ± 0.88a	0.26 ± 0.10c	0.47 ± 0.22c	0.87 ± 0.83c
ST3GAL5 (× 104)	6	0.65 ± 0.08	0.56 ± 0.06	0.73 ± 0.18	0.90 ± 0.22	1.31 ± 0.73a
TYMP (× 104)	6	3.59 ± 1.45	4.54 ± 1.18a	3.19 ± 0.64	1.90 ± 0.33c	1.63 ± 0.65c

^aP < 0.05 vs normal control group.

^bP < 0.01 vs normal control group.

^cP < 0.001 vs normal control group.

NC: Normal control; MAFLD: Metabolic dysfunction-associated fatty liver disease; MASH: Metabolic dysfunction-associated steatohepatitis; LC: Liver cirrhosis; HCC: Hepatocellular carcinoma.

Table 3 Enrichment analysis of the top 10 genes associated with metabolic dysfunction-associated fatty liver disease malignancy.

Group	Code	Biological function	P value1	Gene symbol
MAFLD	rno00100	Steroid biosynthesis	< 0.001	Sc5d, Ebp, Msmo1, Cyp51, Sqle, Tm7sf2
	rno00900	Terpenoid backbone biosynthesis	< 0.001	Hmgcr, Hmgcs1, Pmvk, Mvd, Mvk
	rno05310	Asthma	0.002	RT1-Da, RT1-Db1, RT1-Bb, RT1-B1
	rno04672	IgA intestinal immune network	0.012	RT1-Da, RT1-Db1, RT1-Bb, RT1-B1
	rno04612	Antigen processing and presentation	0.014	Cd74, RT1-Da, RT1-Db1, RT1-Ba, RT1-Bb
	rno05332	Graft-vs-host disease	0.016	RT1-Da, RT1-Db1, RT1-Ba, RT1-Bb
	rno05330	Allograft rejection	0.018	RT1-Da, RT1-Db1, RT1-Ba, RT1-Bb
	rno04940	Type I diabetes mellitus	0.026	RT1-Da, RT1-Db1, RT1-Ba, RT1-Bb
	rno01040	Biosynthesis of unsaturated FA	0.026	Acaa1b, Fads2, Fads1
	rno05320	Autoimmune thyroid disease	0.026	RT1-Da, RT1-Db1, RT1-Ba, RT1-Bb
MASH	rno00100	Steroid biosynthesis	< 0.001	Ebp, Msmo1, Cyp51, Sqle, Tm7sf2
	rno00480	Glutathione metabolism	0.002	Ggt1, Gsr, Gclc, Gpx6, Gclm, Gsta3
	rno00260	Gly, Ser and Thr metabolism	0.003	Dao, Cbs, Agxt, Sds, Alas2, Psat1, Phgdh
	rno04115	p53 signaling pathway	0.003	Cdkn1a, Ccnb1, Ccng1, Gadd45b, Gtse1
	rno04540	Gap junction	0.004	Drd2, Egfr, Gja1, Gucy1b2, Pdgfa, Gnai1
	rno01040	Biosynthesis of unsaturated FA	0.005	Acot2, Acaa1b, Acot1, Acot4, Fads2, Fads1
	rno03320	PPAR signaling pathway	0.006	Acsl3, Lpl, Cyp7a1, Fabp2, Cyp4a8, Aqp7
	rno00830	Retinol metabolism	0.006	Aldh1a2, Cyp3a9, Aldh1a1, Cyp1a1
	rno04110	Cell cycle	0.006	Cdkn1a, Bub1b, Myc, Cdkn1c, Ccnb1
	rno00980	Xenobiotics by cytochrome P450	0.006	Cyp3a9, Cyp1a1, Cyp2c12, Aldh3a1
HCC	rno00100	Steroid biosynthesis	< 0.001	Sc5d, Ebp, Msmo1, Cyp51, Sqle
	rno04510	Focal adhesion	< 0.001	Ppp1r12a, Lama5, Lamc2, Egfr, Hgf
	rno04512	ECM-receptor interaction	< 0.001	Lama5, Lamc2, Ibsp, Itga1, Spp1, Hmmr
	rno04110	Cell cycle	< 0.001	Cdkn1a, Bub1b, Myc, Cdkn1c, Cdkn2a
	rno03320	PPAR signaling pathway	< 0.001	Acsl3, Acsl6, Olr1, Lpl, Me1, Cyp7a1
	rno00480	Glutathione metabolism	< 0.001	Ggt1, Gsr, G6pd, Gclc, Gpx2, Gclm
	rno04115	p53 signaling pathway	< 0.001	Cdkn1a, Fas, Serpine1, Cdkn2a, Ccnb1
	rno00340	Histidine metabolism	0.002	Ddc, Aldh3a1, Aldh1a3, Hal, Amdhd1
	rno05410	Hypertrophic cardiomyopathy	0.002	Tnnt2, Itga1, Itgb4, Atp2a2, Sgcg
	rno00980	Xenobiotics by cytochrome P450	0.008	Cyp3a9, Cyp1a1, Cyp2c12, Aldh3a1

¹Compared with the control group.

MAFLD: Metabolic dysfunction-associated fatty liver disease; MASH: Metabolic dysfunction-associated steatohepatitis; HCC: Hepatocellular carcinoma; FA: Fatty acids; PPAR: Peroxisome proliferator-activated receptors; Gly: Glycine; Ser: Serine; Thr: Threonine; ECM: Extracellular matrix; IgA: Immunoglobulin A.

Dynamic DM alterations during MAFLD progression

The alterations in hepatic DMs during MAFLD malignancy are shown in Figure 2. Compared with those in the NC group, there were 132 DMs in MAFLD, 135 in MASH, 127 in LC, and 130 in the HCC group. Compared with the MAFLD group, there were 99 in MASH, 111 in LC, and 99 in HCC. Compared with the MASH group, there were 41 in LC and 56 in HCC; and there were 42 between the LC and HCC groups. We used volcano maps to screen for DMs between MAFLD and NC (Figure 2A) and between HCC and MASH (Figure 2B). We used DMs based on unsupervised PCA to observe the overall sample distribution and analyzed them via OPLS-DA in MAFLD vs NC (Figure 2C) and HCC vs MASH (Figure 2D). Splot-OPLS-DA was used to show the eigenvalues of the metabolites vs the vertical correlations between the sample scores and the metabolites. Significant differences were found between the MAFLD and NC groups (Figure 2E) and between the HCC and MASH groups (Figure 2F). The upregulated DMs were mainly sphingomyelins (SMs: D14:1/28:1, d22:0/19:1 and d24:0/18:1), whereas the downregulated DMs were phosphatidylcholines (PCs: 16:0/20:4, 10:0/241, 17:0/17:1, and 16:0/18:2). All the points in the visualized graph were distributed in the 1^st and 3^rd quadrants. The DMs were closer to the upper-right and lower-left corners (Figure 2F). DMs in the biological process category were confirmed between MAFLD and HCC, suggesting that abnormal DMs may promote MAFLD malignant progression.

Figure 2 Dynamic alterations in hepatic differential metabolites during metabolic dysfunction-associated fatty liver disease malignancy. A and B: Differential metabolite (DM) volcano maps (P value vs fold change). Metabolic dysfunction-associated fatty liver disease (MAFLD) vs normal control (NC) (A). Hepatocellular carcinoma (HCC) vs metabolic dysfunction-associated steatohepatitis (MASH) (B); C and D: Principal least squares-discriminant analysis of DM. MAFLD vs NC (C). HCC vs MASH (D); E and F: S-shaped orthogonal principal least squares-discriminant analysis plot analysis of DM. MAFLD vs NC (E) and HCC vs MASH (F). NC: Normal control; MAFLD: Metabolic dysfunction-associated fatty liver disease; FC: Fold change; PLS-DA: Principal least squares-discriminant analysis; Splot-OPLS-DA: S-shaped orthogonal principal least squares-discriminant analysis plot analysis.

Correlation between DMs and MAFLD malignancy

A summary of each of the top five up-/down-regulated DMs at different stages of MAFLD is shown in Table 4. Compared with those in the NC group, the significantly increased DMs in the MASH stage were PCs (16:022:4, 18:0/20:4), platelet-activating factor (PAF, 36:4, 8:6), and lysophosphatidyl ethanolamine (20:4). The significantly decreased DMs were PCs (18:0/18:3, 36:0/20:3), PAF (34:3), dimethylphosphatidylethanolamine (18:0/18:2), and SM (d16:0/26:2). In the HCC stage, PCs (16:0/22:4, 18:0/20:4), PAFs (36:4, 38:6), and lysophosphatidylethanolamine (20:4) were significantly increased. In contrast, PCs (16:0/18:1, 18:0/24:1) and phosphatidylethanolamine (32:1/34:1) were significantly decreased. During the MASH to HCC stage, dimethylphosphatidylethanolamine (16:2/18:2), SMs (d14:1/28:1, d22:0/19:1, d2:0/18:1), and PC (18:0/18:3) increased, whereas PCs (16:0/0:4, 10:0/24:1, 17:0/17:1, 16:0/8:2) and PAF (34:4) significantly decreased. As the model progressed, TG and acylglycerophosphocholine were consistently upregulated, PC was downregulated, and SM increased. DMs included in the panel primarily focused on the phosphorylation pathway.

Table 4 Top five up/downregulated differential metabolites during metabolic dysfunction-associated fatty liver disease progression.

Groups	Top 5 upregulated	Top 5 downregulated
MAFLD vs NC	PC (16:0/22:4)	PC (18:0/18:3)
	PAF (36:4)	PC (36:0/20:3)
	PC (18:0/20:4)	dMePE (18:0/18:2)
	LPE (20:4)	SM (d16:0/26:2)
	PAF (38:6)	PAF (34:3)
HCC vs NC	PC (16:0/22:4)	PC (16:0/18:1)
	PAF (36:4)	PC (10:0/24:1)
	PC (18:0/20:4)	PAE (32:1)
	LPE (20:4)	PAF (34:1)
	PAF (38:6)	PC (18:0/18:1)
HCC vs MASH	dMePE (16:2/18:2)	PC (16:0/20:4)
	SM (d14:1/28:1)	PC (10:0/24:1)
	SM (d22:0/19:1)	PC (17:0/17:1)
	SM (d24:0/18:1)	PC (16:0/18:2)
	PC (18:0/18:3)	PAF (34:4)

NC: Normal control; MAFLD: Metabolic dysfunction-associated fatty liver disease; MASH: Metabolic dysfunction-associated steatohepatitis; HCC: Hepatocellular carcinoma; PC: Phosphatidylcholine; PAF: Platelet-activating factor; LPE: Lysophosphatidylethanolamine; dMePE: Dimethylphosphatidylethanolamine; SM: Sphingomyelin.

A comparative analysis of DMs during the malignant progression of MAFLD from MASH to HCC is shown in Figure 3. TG, acylglycerolphosphorylcholine, and SMs were upregulated. The DMs of lipid metabolism significantly differed during MAFLD progression according to hierarchical clustering (Figure 3A) and the Pearson correlation coefficient of DMs (Figure 3B). Interestingly, choline and glycerophospholipid metabolism significantly decreased by nearly 50% from MASH to HCC, along with comprehensive downregulation of steroid biosynthesis and upregulation of the cell cycle pathway.

Figure 3 Correlation analysis of differential metabolites during metabolic dysfunction-associated fatty liver disease malignancy. A: Hierarchical clustering of differential metabolites significantly different between the hepatocellular carcinoma (n = 10) and metabolic dysfunction-associated steatohepatitis (n = 10) groups. Horizontal, samples; vertical, differential metabolite (DM). Blue, low-abundance DM; red, high-abundance DM; B: Differential metabolites (hepatocellular carcinoma vs metabolic dysfunction-associated steatohepatitis) were analyzed with Pearson’s correlation coefficient. Blue, negative correlation; red, positive correlation. rHCC: Rat hepatocellular carcinoma; H-deg: Hepatocyte degeneration.

Dynamic immune cells during MAFLD progression

All cells isolated from fresh liver tissues during MAFLD malignancy progression were subjected to single-cell sequencing, and the resulting cell subpopulations are shown in Figure 4. Hepatic cells were divided into 34 subtypes in healthy liver, including 24 identified cell populations and 10 unidentified populations (Figure 4A). The most abundant cell subpopulations in the healthy liver were hepatocytes, B cells, endothelial cells, natural killer (NK) cells, T/monocyte/macrophage populations, and dendritic cells (DCs) (Figure 4B). However, several cell subpopulations, particularly immune cells, were significantly and dynamically altered during MAFLD malignancy (Figure 4C). Among the 16 most abundant subpopulations (Figure 4D), the percentage of immune/inflammatory cells during MAFLD progression significantly changed, with increased numbers of B/T lymphocytes, monocytes macrophages, NKs, Kupffer cells, DCs, and neutrophils and decreased numbers of hepatocytes and endothelial cells (Figure 4E), indicating that MAFLD malignant transformation is closely associated with the dynamic accumulation of immune/inflammatory cells, especially T cells and macrophages, in the liver.

Figure 4 Hepatic cell clustering under lipid aggregation. A: Cell clustering of hepatic cells in healthy liver; B: Number of hepatic cell in each subpopulation; C: Dynamic liver cell subpopulations during metabolic dysfunction-associated fatty liver disease malignancy. Identifiable cells: Hepatocytes (0, 20, 23, 24, 27); B cells (2, 25, 28); endothelial cells (3, 17); natural killer cells (6, 10, 13); T cells/monocytes/macrophages (7, 9, 2, 15); dendritic cells (8); neutrophils (11); bile duct cells (19); red blood cells (21); plasma cells (26); hepatic stellate cells (29); dendritic cells (33). Unidentified cells (1, 4, 5, 14, 16, 18, 22, 30, 31, 32); D: The 16 most abundant subpopulations of immune cells in the healthy liver: Hepatocytes (0); macrophages (1, 4, 5, 14, 16); B cells (2); endothelial cells (3); natural killer cells (6, 10, 13); T cells/monocytes (7, 9, 12, 15); dendritic cells (8); and neutrophils (11); E: The percentage of immune/inflammatory cells during metabolic dysfunction-associated fatty liver disease malignancy progression. NC: Normal control; MAFLD: Metabolic dysfunction-associated fatty liver disease; MASH: Metabolic dysfunction-associated steatohepatitis; LC: Liver cirrhosis; HCC: Hepatocellular carcinoma; NK: Natural killer; DC: Dendritic cell; HSC: Hepatic stellate cell.

MAFLD progression associated with immune evasion

A comparative analysis of gene transcription among immune/inflammatory cells during MAFLD malignancy is shown in Figure 5. Strongly expressed gene transcripts associated with MAFLD were identified in immune cells (Figure 5A), such as hepatocytes (Apoa2, Apoc2, Alb), HSCs (Dcn, Col3a1), T cells (Cd3d, Cd3e, Cd3 g), NK cells (Nkg7, Klrd1), B cells (Cd19, Ms4a1, Cd79a), macrophages (Cd68), neutrophils (S100a8), pDCs (Naaa, Ppt1, Xpt1), cDCs (Jchain, Siglech, Ccr9, Irt8), and mDCs (Ccr7, Fscn1, Tmem123). The genes whose expression was upregulated or downregulated from NC to MAFLD progression (Figure 5B) displayed significant inflammatory-related responses, such as those in macrophages and T and cDC cells. The transcript changes from MAFLD to HCC were similar to those from NC to MAFLD, with the exception of those in inhibited T cells (Figure 5C). The interaction analysis of immune cells revealed that macrophages play an important role in MAFLD progression (Figure 5D), especially T, NK, and B cells, which are more important than epithelial, cDC and neutrophil populations. An interesting finding was the activation of programmed death ligand 1 (PD-L1), which mediates immune escape during MAFLD malignancy (Figure 5E). These results indicate that T-cell inhibition and PD-L1 signaling activation are related to the creation of an immunosuppressive microenvironment to help hepatocyte malignant transformation evade immune surveillance.

Figure 5 Gene transcripts of immune cells during metabolic dysfunction-associated fatty liver disease malignancy progression. A: Gene transcripts of hepatic immune cells; B: Transcript numbers of genes whose expression was upregulated or downregulated from healthy liver to metabolic dysfunction-associated fatty liver disease; C: Transcript numbers of genes whose expression increased or decreased from metabolic dysfunction-associated fatty liver disease to hepatocellular carcinoma; D: Interaction analysis of hepatic immune cells; E: Macrophages with immune cells in the programmed death ligand 1-activating network. MAFLD: Metabolic dysfunction-associated fatty liver disease; MASH: Metabolic dysfunction-associated steatohepatitis; HCC: Hepatocellular carcinoma.

M2-polarized macrophages promote MAFLD malignancy

The dynamics of macrophages and related signaling pathways during MAFLD malignancy are shown in Figure 6. Although macrophages are among the most important immune cells involved in MAFLD progression, the genes involved in their regulation vary at different stages (Figure 6A). Compared with M1 macrophages (CD86- and NOS2-positive), activated M2 macrophages (CD68, CD163 and MIC1, Figure 6B) were significantly increased and exhibited upregulated gene expression (Figure 6C). This change was accompanied by inhibition of the steroid biosynthesis pathway (Figure 6D) and abnormal activation of the cell cycle signaling pathway (Figure 6E). These data indicated that in a lipid-rich environment, DMs and M2-polarized macrophages could be associated with MAFLD malignancy (Figure 6F) through signaling pathways that inhibit steroid biosynthesis or activate the cell cycle.

Figure 6 M2 macrophages and signaling pathways during metabolic dysfunction-associated fatty liver disease malignancy. A: Dynamic alterations in the expression of the main genes in hepatic macrophages during metabolic dysfunction-associated fatty liver disease malignancy; B: M2-polarized macrophages (CD68+ or CD163+); C: Gene signal intensity of M1- or M2- macrophages; D: Inhibition of the steroid biosynthesis pathway; E: Activation of the cell cycle signaling pathway; F: Schematic diagram for interactions among differential metabolites, inhibited T cells and M2 macrophages promoting metabolic dysfunction-associated fatty liver disease malignancy. PD-L1: Programmed death ligand 1; HFD: High-fat diet; CPT-II: Carnitine palmitoyl transferase II; DMs: Differential metabolites; NC: Normal control; MAFLD: Metabolic dysfunction-associated fatty liver disease; MASH: Metabolic dysfunction-associated steatohepatitis; LC: Liver cirrhosis; HCC: Hepatocellular carcinoma.

DISCUSSION

MAFLD malignancy progression is a complex process involving genes and gene regulation, with overall dynamic changes in multiple genes during its stages[20]. MAFLD includes a range of liver manifestations, starting with steatosis and potentially evolving toward MASH, cirrhosis or even HCC[21]. Currently, the most widely accepted theory of MAFLD pathogenesis is a synergistic result of mitochondrial dysfunction, lipotoxic endoplasmic reticulum stress and several immune cell-mediated inflammatory processes caused by steatosis and lipid oxidation, particularly at the MASH stage, which indicates that inflammation is integral to disease progression[22,23]. Previous studies reported that CPT-II is located on the inner membrane of mitochondria in MAFLD models and that the level of CPT-II in patients gradually decreases with disease severity[24,25]. However, the exact mechanism underlying MAFLD malignancy remains underexplored. In this study, we used dynamic MAFLD models to investigate alterations in CPT-II, DMs and immune cells during MAFLD malignant progression to provide new evidence to support the development of strategies for early prevention or specific targeted intervention.

Genomic analysis revealed gene signatures and robust DEGs during MAFLD malignancy[6,26,27]. Many DEGs involved in complex regulatory networks have been confirmed by models of MAFLD progression, with significantly upregulated expression of BAX, CCND1, CDK1, Y1A1, GAL3ST1, HEXB-β, LDHB, PPARγ and ST3GAL5, involving pathways such as cell damage, apoptosis, cell cycle and biogenesis, whereas CPT-II, CYP51, EGR1, HOD3, MTHFR, SAT2 and TYMP were significantly downregulated, involving mitochondrial damage and mitochondrial enzyme activity[28,29]. Genes associated with steroid synthesis (Cyp51) and Tm7sf2, P53, and cell cycle-related CCNB1 and CDK1 pathways were enriched. DEGs enriched in the steroid synthesis pathway, p53 signaling pathway, and cell cycle signaling pathway were enriched, and Cyp51 and Tm7sf2 were significantly downregulated. Additionally, they play key roles in the protein interaction network. These data indicate that mitochondrial damage can significantly affect MAFLD progression and that CPT-II may be a new target for anti-inflammatory or anticancer therapy[3,13].

Dyslipidemia and DM production are important steps in early MAFLD malignancy[5,30] and might directly cause chronic hepatocyte inflammation via harmful metabolites[31,32]. On the basis of the dynamic model of MAFLD malignancy, the early stage upregulated DMs are directly related to lipid, fatty acid and retinol metabolism during adipocyte differentiation. Conversely, PC and phosphatidylethanolamine (phospholipids and brain phospholipids) are downregulated, and the structure of the liver cell membrane changes. In addition, diglycerides, TG, and acylglycerophosphocholine are upregulated. Downregulated Creb3 L1 leads to lower cAMP in HCC cells, which is a precursor to MAFLD malignancy[33-35]. In this study, more than 130 DMs were identified according to vertical correlation and distribution trends at different stages. The main upregulated DMs were SMs, and the main downregulated DMs were PCs. SMs and PCs are a series of lipid molecules[36,37] that exist in the cell membrane, where they are involved in energy supply, signal trans-duction, and the regulation of cell growth, differentiation, migration and apoptosis[38,39].

Moreover, dyslipidemia inevitably affects T cells in vivo[40,41]. Elov1.6, which plays a key role in fatty acid chain elongation, was upregulated, as was Cryl1, which catalyzes the formation of NADH from NAD, because HCC cells are energy-consuming. Cryl1, which plays an important role in the elongation of fatty acid chains, was upregulated, and correspondingly, Cryl1, which catalyzes the formation of NADH from NAD+, was upregulated, leading to energy consumption by HCC cells[35,42]. Consistent with energy storage, TG and acylglycerophosphocholine were upregulated. Significantly downregulated PCs and upregulated SMs may be directly related to MAFLD malignancy. The complexity of the DEGs during MAFLD malignancy progression (Cyp51, Tm7sf2, CCNB1, and CDK1, etc.) and the discovery of sphingolipid enrichment via lipidomics suggest that its intervention might constitute an innovative strategy to increase the efficacy of immunotherapy. However, whether an increase in membrane sphingolipids can predict HCC progression, responsiveness to immunotherapy or the effect of 2-fluorenylacetamide on hepatocytes are unknown. Sphingolipids cannot be ignored in the process of cancer immune escape. Therefore, among the DMs that significantly increased or decreased, T-cell depletion was closely related to MAFLD malignancy[43,44].

The accumulation of hepatic DMs leads to immune suppression in MAFLD in the liver microenvironment[45,46]. In the MAFLD stage, the upregulated DMs from MASH to HCC were SMs with fatty acids, retinol, diacylglycerol, TG, acylglycerophosphocholine and adipocytes[47]. The downregulated DMs were PC and phosphatidylethanolamine (lecithin and cephalin), which are associated with the membrane structure of hepatocytes[44,48]. Changes in MAFLD-related choline, retrograde endocannabinoid, linoleic, α-linolenic and arachidonic acid metabolism, choline and glycerophospholipid metabolism are signs of MAFLD malignancy. Specific toxic lipids rather than total lipid burden within the liver are the driving factors for MASH deterioration. Sphingolipid metabolism, especially the core metabolite ceramide, has been demonstrated to involve key molecules in lipotoxicity. Therefore, targeting sphingolipids could be a potential intervention for MASH treatment[48,49].

Previous studies have shown that tumor necrosis factor alpha and interleukin-10 are secreted by activated monocytes/macrophages to induce PD-L1 expression in an autocrine manner[50-52]. M2 macrophages mainly exhibit anti-inflammatory, proangiogenic, and protumor progression effects[53,54]. Therefore, studying immune functions in the microenvironment is crucial for understanding their role and potential mechanisms in tumor immunity[55]. Complex interrelationships exist among the major liver cell subpopulations, with key interactions occurring between macrophages and NK cells, T cells, endothelial cells, and B cells. In addition, gene transcription in inflammatory and immune cells, including T cells, NK and macrophages, as well as adaptive immune cells such as T and B cells, is altered in lipid-accumulating hepatocytes[46,56]. The main pathways involved in MAFLD progression are involved in steroid biosynthesis, cholesterol metabolism, redox, p53 and the cell cycle. Choline is a key DM for the synthesis of phospholipids and neurotransmitters, and while it can be transported into the mitochondria, it normally localizes to the cytoplasm where it acts as a methyl donor for methionine synthesis. CCNB1 and CDK1 are involved not only in the p53 pathway, but also in the cell cycle and have key positions in the PPI network. Both are enriched in M phase, cell cycle and cell division, sterol synthesis, and the response to endogenous stimuli, steroid hormones, hormonal stimuli, and biosynthetic processes[57,58].

CONCLUSION

In conclusion, the dynamic alterations of DMs and immune cells are very useful for understanding MAFLD transformation via hepatocarcinogenesis with decreasing mitochondrial CPT-II activity. Significant DEGs, such as Cyp51 and Tm7sf2, and DMs (PCs and SMs), which affect the biosynthesis of steroids and sphingolipids, PC and phosphatidylethanolamine, are associated with MAFLD progression. CCNB1 and CDK1 are related to MAFLD malignancy via the steroid biosynthesis, cell cycle and lipid metabolism signaling pathways in the PPI network. Interestingly, hepatic immune cells undergo dynamic alterations in a fat-rich microenvironment characterized by a decrease in T-cell dysfunction, an increase in PD-L1 expression for immune escape and M2-polarized macrophages to promote MAFLD malignancy, providing novel evidence to assist in developing strategies for early prevention or specific targeted intervention.

ACKNOWLEDGEMENTS

The authors thank Professor Li Wang for providing an illustration of the schematic model of metabolic dysfunction-associated fatty liver disease.