Bioinformatics analysis of ferroptosis-related genes and immune cell infiltration in non-alcoholic fatty liver disease

Non-alcoholic fatty liver disease (NAFLD) is an important liver disease that affects approximately 24% of the general population [1]. In the coming decades, NAFLD might become the leading cause of end-stage liver disease [1]. NAFLD encompasses a range of diseases from non-alcoholic fatty liver (NAFL) to non-alcoholic steatohepatitis (NASH), fibrosis, and cirrhosis [2]. An underlying progressive liver disease is typically observed in a subset of patients with NAFLD..

NAFLD is usually diagnosed by an invasive liver biopsy. Presently, there are no reliable biomarkers for accurately diagnosing and staging NAFLD, which makes it challenging to screen NAFLD cases worldwide [3]. Moreover, according to the current hypothesis, NAFLD is the hepatic manifestation of metabolic syndrome because of its bidirectional association with the components of metabolic syndrome [3, 4]. NAFLD patients show a high incidence rate of metabolic complications; hence, NAFLD is considered a growing burden on the healthcare system [5]. Therefore, it is crucial to identify new and efficient NAFLD biomarkers for the prompt diagnosis and treatment of this disease.

Ferroptosis is an iron-dependent form of programmed cell death. It is characterized by the cellular accumulation of lipid hydroperoxides to lethal levels [6]. The morphological effects of ferroptosis include reduced mitochondrial size, disappearance of mitochondrial cristae, and mitochondrial membrane rupture [7]. The primary changes in biochemical characteristics associated with ferroptosis are iron overload and decreased glutathione peroxidase 4 (GPX4) activity; these changes promote the production of reactive oxygen species (ROS), accelerate lipid peroxidation, and eventually lead to cell death [8]. Ferroptosis is associated with the onset and progression of many liver diseases such as NAFLD, alcohol-associated liver disease (ALD), hepatocellular carcinoma (HCC), and hepatitis C virus (HCV) infection [9‐13]. Alterations in several metabolic pathways, including decreased GPX4 activity, iron overload, acyl-CoA synthetase long-chain family member 4 (ACSL4) induction, and nuclear factor erythroid-2-related factor 2 (Nrf2) activation, have been implicated in ferroptosis [14, 15]. Therefore, ferroptosis inhibition could serve as a new treatment approach for NAFLD. It, however, remains unclear how ferroptosis regulates NAFLD.

In the study, we analyzed two NAFLD liver tissue-derived microarray datasets from the Gene Expression Omnibus (GEO) database and obtained differentially expressed genes (DEGs). From these DEGs, we selected ferroptosis-related genes (FRGs). The expression of these FRGs was further validated in another microarray dataset. Finally, five genes, namely SCP2, MUC1, DPP4, SLC1A4, and TF, were screened as target genes. Competing endogenous RNA (ceRNA) networks were constructed to determine the specific regulatory effects of noncoding RNAs on the FRGs in NAFLD. The ratios of immune cell infiltration in NAFLD and normal tissues were calculated using the CIBERSORT package. We also evaluated the correlations between the expression of FRGs and infiltration ratios of various immune cells. This in-depth research investigated the mechanism of NAFLD development at the transcriptome level and identified potential biomarkers for NAFLD diagnosis.

NAFLD is the most common chronic liver disease worldwide. Ferroptosis plays a critical role in the occurrence and progression of NAFLD through the regulation of iron homeostasis and lipid metabolism in the liver. Hence, the identification of new effective NAFLD biomarkers could enable prompt diagnosis timely and treatment of this disease.

In the present study, we attempted to identify ferroptosis-related biomarkers in NAFLD and examined the role of immune cell infiltration in NAFLD pathogenesis. Six FRGs were identified using two liver tissue microarray datasets (GSE72756 and GSE24807) from the GEO database and the FerrDb database. Of these, five FRGs (SCP2, MUC1, DPP4, SLC1A4, and TF) were finally identified after confirmation using the GSE89632 dataset. The pathway enrichment analysis revealed that these genes were primarily involved in metabolic processes. The most enriched GO categories were regulation of intracellular cholesterol and lipid transport, regulation of intracellular cholesterol and lipid transport, organic acid biosynthetic and catabolic process, fatty acid beta-oxidation using acyl-CoA oxidase, cholesterol-binding, fatty acid-binding, and acidic amino acid transmembrane transporter activity. The KEGG pathway enrichment analyses showed that the main functions of these genes were the biosynthesis and metabolism of fatty acids, primary bile acid biosynthesis, protein digestion and absorption, peroxisome, and regulation of the HIF-1 signaling pathway and the PPAR signaling pathway.

Among the five selected FRGs, SCP2 and DPP4 are ferroptosis driver genes, MUC1 is a ferroptosis suppressor gene, SLC1A4 is a ferroptosis marker gene, and TF is a ferroptosis driver and marker gene [26]. We focused on two genes: SCP2 related to lipid metabolism and MUC1 with the highest diagnostic specificity. The lipid transport protein SCP2 can combine with fatty acids and fatty acyl-COA to regulate fatty acid metabolism in the liver [27, 28]. It is also a key regulator of cholesterol metabolism in the liver and plays a beneficial role in NAFLD. In contrast to the protective effect of SCP2 on NAFLD, a previous study showed that SCP2 can promote the accumulation of low-density lipoprotein cholesterol (LDL-C), thereby promoting the development of atherosclerosis and hyperlipidemia [29]. Previous studies have indicated that SCP2 can suppress ferroptosis inhibitors (GPX4 and cav1) and activate ferroptosis promoters (PRKAA1 and PRKAA2) [30]. MUC1 is a large O-type glycoprotein essential for maintaining the function of the epithelial cell surface [31]. It is composed of two subunits: the MUC1 N-terminal subunit (MUC1-n) and the carcinogenic MUC1 C-terminal subunit (MUC1-c) that form a heterogeneous complex on the cell membrane [32]. Many MUC1-c subunits can be detected in the mitochondria and nuclei of cancer cells. Several studies have shown that MUC1 plays a key regulatory role in tumor invasion, metastasis, angiogenesis, and inflammation [33‐36]. MUC1 can also induce apoptosis and necrosis by inhibiting ROS accumulation [37]. Hasegawa et al. demonstrated that targeting MUC1-c with ferroptosis inhibitors induces ROS-mediated death [38].

Yangchunxie et al. confirmed that DPP4 (also known as CD26) plays a role in ferroptosis regulation and found that the loss of TP53 prevented the nuclear accumulation of DPP4 in colorectal cancer cells, thereby facilitating the plasma membrane-related DPP4-dependent lipid peroxidation and ultimately leading to ferroptosis [39]. These data support DPP4 as the driving factor of ferroptosis. TF is an extremely important factor in regulating iron trafficking and metabolism. The increased expression of TF is suggested to induce ferroptosis [40]. SLC1A4 is one of the members of solute carrier family 1, and it can promote ferroptosis [41]. It should be noted that most of these aforementioned genes have been identified in tumors; however, there is a general lack of evidence regarding their role in NAFLD.

In the present study, two ceRNA networks were constructed to determine the regulatory mechanisms of these five FRGs by predicting their miRNA targets as well as the lncRNAs and circRNAs targeted by these miRNAs. According to the ceRNA hypothesis, we searched literature related to NAFLD in the PubMed database and selected 3 reported miRNAs, 2 lncRNAs, and 1 circRNA for further investigation. Based on our findings, we suggest that MALAT1-miR-485-5p-MUC1, NEAT1-miR-1224-5p-SCP2, and NEAT1-miR-485-5p-MUC1 might be the regulatory pathways for the pathogenesis and progression of NAFLD. MALAT1 and NEAT1 are important lncRNAs, and recent studies have reported that their expression is upregulated in the liver tissues of NAFLD patients. MALAT1 knockdown reversed free fatty acid -induced lipid accumulation in hepatocytes; moreover, MALAT1 promoted the progression of liver fibrosis [42]. NEAT1 was previously identified as an oncogene that promotes tumor cell proliferation [43]. Several recent studies have shown that NEAT1 participates in NAFLD progression by promoting lipid deposition in the liver [44]. The regulatory relationship between NEAT1 and ferroptosis has been reported in recent literature. Zhang et al. discovered that NEAT1 overexpression enhances both extracellular and intracellular ferroptosis, thereby increasing the anti-tumor activity of erastin [45]. miR-1224-5p promotes hepatic lipogenesis by inhibiting AMPKα1 expression [46]. miR-1224-5p inhibitors deserve further investigation as a potential therapeutic tool for treating NAFLD. miR-485-5p is associated with inflammation and immune responses, and it upregulates MUC1 to promote liver cancer progression [47]. MALAT1 and NEAT1 target miRNAs in NAFLD to regulate ferroptosis; this aspect requires further investigation. Regarding circRNAs, although circ_CNOT6 has not been reported in NAFLD, it is likely to play a critical role in other metabolic diseases such as diabetes [48]. Therefore, we hypothesized that circ_CNOT6-miR-145-5p-SLC1A4 might be involved in NAFLD development. Further prospective cohort studies are required to confirm our hypothesis.

The primary cause of NAFLD is metabolic dysfunction. Immune cell-mediated inflammatory processes also contribute to NAFLD. The liver immune cell landscape directly affects the severity of NAFLD. A study conducted by the German Cancer Research Center showed the accumulation of a large number of CD8/PD-1 double-positive abnormal T cells in the liver of NASH patients. PD-1/L1 inhibitors can activate these T cells; however, treatment with PD-1/L1 inhibitors not only kill the tumor cells but also aggravates liver tissue damage [49]. Previous studies have confirmed that immunotherapy has no survival benefit for liver cancer patients with NAFLD.

To better understand immune cell infiltration, we used the CIBERSORT algorithm to evaluate immune cell infiltration in NAFLD tissues. We found that increased infiltration of M1 macrophages and neutrophil was associated with NAFLD occurrence and development. We also found that the M1 macrophage activation was negatively correlated with MUC1 and SLC1A4 and neutrophil activation was negatively correlated with SLC1A4.

According to previous studies, the beneficial effects of neutrophils during infections are opposite to those in noninfectious diseases. Neutrophils usually produce neutrophil extracellular traps, proteases, cytokines, and ROS to induce adverse effects on the infectious agent [50, 51]. Several studies have reported significantly increased neutrophil infiltration in the liver of patients with NASH [52, 53]. Neutrophils are involved in the early stages of NASH development. However, their role in the advanced stage of NASH remains unclear [54]. In an in vivo study, Zhao et al. confirmed that methionine-choline-deficient and high-fat (MCDHF) diet-induced liver injury was significantly reduced by an intraperitoneal injection of deoxyribonuclease I [53]. Blood monocytes are recruited to hepatic sinusoids and differentiate into macrophages, thereby increasing the macrophage pool of the liver [55]. Recent studies have shown that monocyte-derived macrophages exhibit more apparent inflammatory characteristics in NASH and can promote injury by limiting liver lipid storage in the liver [56]. Monocyte-derived macrophages in mouse livers are located in the tissue fibrosis area near desmin-positive hepatic stellate cells, thus indicating their contribution to liver fibrosis [57]. These studies and our present analysis support the concept that immune cell infiltration is an important factor in NAFLD pathogenesis. Future studies should focus on the correlation between FGRs and M1 macrophages and neutrophils.

In the present study, we identified DEGs associated with ferroptosis in NAFLD. Our findings also suggest a certain correlation between FRGs and immune cell infiltration in NAFLD. Furthermore, we identified NAFLD-related miRNAs, lncRNAs, and circRNAs. However, because this was a strictly bioinformatics analysis, in future studies, we will focus on the expression patterns and functions of these genes to understand the precise molecular mechanism of ferroptosis in NAFLD development. First, we will quantify the expression of these genes at the transcriptional and translational levels and confirm their interactions through immunohistochemical and immunofluorescence assays. Second, we will determine the specific mechanisms of ferroptosis in lipid accumulation, hepatocyte injury, and immune responses by using cellular models. Third, we will collect more liver tissue samples from NAFLD patients for conducting large-scale research. Our research will focus on identifying more effective ferroptosis-specific biomarkers and developing ferroptosis modulators with improved properties for alleviating NAFLD.

The present study has several limitations. First, we did not perform an additional in vivo experiment to validate whether the selected FRGs regulate ferroptosis in NAFLD. Second, we are aware of FRGs only from the FerrDb database, and only a few studies have examined the role of ferroptosis in NAFLD. Third, we did not evaluate the different stages of NAFLD. To overcome these limitations, prospective clinical trials should be designed to elucidate the mechanisms of action of the five FRGs in different stages of NAFLD.

In summary, based on our bioinformatics analysis, we identified five FRGs (SCP2, MUC1, DPP4, SLC1A4, and TF) that could predict NAFLD development and explored the potential pathway of liver tissue damage in NAFLD. We also provided new insights into the molecular mechanisms of NAFLD pathogenesis. Further research is required to confirm our preliminary evidence and to validate these FRGs as proposed biomarkers for NAFLD diagnosis in clinical practice.