Abstract
Objectives:
This study aimed to identify mitophagy-related biomarkers in severe acute pancreatitis through integrated bioinformatics and machine-learning approaches, and to elucidate their potential regulatory mechanisms and diagnostic value.
Materials and Methods:
Differentially expressed genes (DEGs) related to SAP were screened based on the GSE194331 dataset (human whole blood samples), and candidate genes were obtained by taking the intersection with the mitophagy gene set. Key genes were screened by combining the least absolute shrinkage and selection operator (LASSO) regression and Support Vector Machine-Recursive Feature Elimination (SVM-RFE) algorithms; the diagnostic efficacy was evaluated using the ROC curve, and a nomogram model was constructed. Furthermore, gene functions and regulatory mechanisms were revealed through Gene Set Enrichment Analysis (GSEA), Gene Set Variation Analysis (GSVA), immune infiltration analysis, and ceRNA network, and potential targeted drugs were predicted using molecular docking.
Results:
A total of 16 DEGs related to mitophagy were identified. Functional enrichment analysis showed that they were significantly associated with metabolic and immune regulation pathways. Machine learning and expression level validation jointly screened PGD and LMNB1 genes as two key genes. The key genes showed significant expression differences in the training set and external validation set (upregulated in the SAP group), and had excellent diagnostic efficacy (area under the curve (AUC) > 0.85). Immune infiltration analysis showed that the infiltration of 13 types of immune cells increased in the SAP group, and PGD was highly positively correlated with immune cells such as activated dendritic cells (r = 0.82). Molecular docking indicated that estradiol and progesterone might target and regulate PGD and LMNB1.
Conclusion:
PGD and LMNB1 are key genes related to mitophagy in SAP and have excellent diagnostic value. This study provides a theoretical basis for the analysis of the molecular mechanism of SAP and the development of precise diagnosis and treatment strategies.
Keywords: GSEA, Immune infiltration, Mitophagy, machine learning, Molecular docking, Severe acute pancreatitis
1. Background
Acute pancreatitis (AP) is a common digestive system disease. Its clinical manifestations include abdominal pain, vomiting, and fever. Laboratory tests may show elevated serum amylase levels and peripancreatic exudation. Severe acute pancreatitis (SAP) is associated with a mortality rate ranging from 15% to 30% ( 1 ), and frequently leads to multiple organ dysfunction. Previous studies suggest that pathways such as dysregulation of pancreatic enzyme activation, apoptosis and oxygen free radicals ( 2 ), inflammasomes and pyroptosis of macrophages ( 3 ) play a role in the occurrence and development of SAP. Our understanding of the genetic mechanisms underlying SAP exacerbation remains limited, and there is currently a lack of effective targeted therapies. This article aims to provide new ideas for the early genetic identification and treatment of SAP.
SAP is characterized by necrosis of pancreatic tissue cells and systemic inflammation, and its onset is closely related to mitochondrial dysfunction of pancreatic cells, mainly manifested as loss of mitochondrial membrane potential (MMP) and mitochondrial structural damage. In the early stage of SAP, mitochondrial calcium overload leads to sustained opening of the permeability transition pore (mPTP), which subsequently activates pyroptosis ( 4 ). In the later stage, mitochondrial electron transport chain dysfunction, excessive accumulation of ROS, resulting in oxidative stress imbalance and cell destruction ( 5 ). Mitochondrial autophagy plays a crucial role in maintaining cellular homeostasis by clearing damaged mitochondria. In SAP, mitochondrial autophagy is impaired and the capacity for mitochondrial clearance declines, thereby accelerating cell death. Studies have found that the stress protein Sestrin2 (Sesn2) enhances mitochondrial autophagy by regulating the PINK1-Parkin pathway, reduces mitochondrial superoxide levels and the expression of apoptotic proteins such as Bax and cleaved caspase-3, and promotes the polarization of macrophages to the M2 anti-inflammatory phenotype ( 6 ). There are also new immunomodulators, such as Tuftsin, that reduce SAP by regulating mitophagy ( 7 ). Although studies have shown that autophagy has a significant impact on the pathological damage of SAP, the related genes have not been fully identified, and the specific regulatory mechanism remains to be further clarified.
2. Objectives
In this study, we screened genes related to mitochondrial autophagy in SAP based on bioinformatics methods, identified SAP-related key genes through machine learning, and further explored the mechanism by which biomarkers affect SAP using methods such as GSEA, drug prediction, and molecular docking response.
3. Materials and Methods
3.1. Data Source
In this study, the SAP transcriptome data were retrieved from the GEO database. Specifically, the training dataset GSE194331 (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE194331) originated from human blood samples, comprising 32 normal whole blood samples (Control group) and 10 SAP whole blood samples (SAP group). The validation dataset GSE151927 (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE151927), on the other hand, was derived from mouse samples, consisting of 10 normal liver tissue samples (Control group) and 10 SAP liver tissue samples (SAP group). Mitophagy-related genes were collected from both the Genecards database and the Msigdb database. Following the consolidation and elimination of duplicate genes, a total of 1,765 mitophagy-related genes were identified.
3.2. Obtaining DEGs in SAP
In the training set, differential gene expression analysis was performed by comparing SAP samples with Control samples using the “DESeq2” package in R. The goal of this analysis was to detect DEGs linked to SAP. Genes were selected based on the criteria of an absolute log2 fold change (|log2FC|) exceeding 2 and an adjusted p-value below 0.05. Following the analysis, the findings were illustrated through visualizations created with the “ggplot2” and “pheatmap” packages.
3.3. Identification and Functional Enrichment of Mitophagy-Related Degs
The overlap between mitophagy-related genes and DEGs was analyzed to identify mitophagy-associated DEGs. Subsequently, gene ontology (GO) functional annotation and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analyses were conducted on these mitophagy-associated DEGs to comprehensively explore the functional cooperation mechanisms of the candidate genes. Additionally, the STRING database was utilized to predict the protein-protein interaction networks of the mitophagy-related DEGs.
3.4. Machine Learning for Screening Candidate Key Genes
Next, we utilized two machine learning techniques to refine the gene selection process and identify potential key genes. Initially, we performed a LASSO regression analysis using the R package “glmnet.” To determine the optimal regularization parameter (lambda.min), we applied cross-validation with 10 folds. Following this, we executed SVM-RFE using the R package “e1071,” retaining the gene set that exhibited the lowest error rate and minimal size. Ultimately, we identified the candidate key genes by intersecting the gene lists derived from both machine learning approaches.
3.5. Validation of the Expression Levels of Candidate Key Genes and Drawing of the ROC Curve
In this study, the R package “ggplot2” was used to analyze the expression of four candidate key genes in the training set and the external validation set. The candidate key genes with consistent expression trends in the training set and the validation set were named key genes. At the same time, the R package “pROC” was used to draw the ROC curve of the key genes. The judgment criterion for the ROC curve is that the AUC is greater than 0.7.
3.6. Construction of the Nomogram
To more intuitively understand the relationship between key genes and SAP, we incorporated the standardized expression levels of key genes into the multivariate logistic regression analysis to construct a Nomogram. The accuracy of the nomogram prediction was verified using the Receiver Operating Characteristic Curve (ROC).
3.7. GSEA of Key Genes
To gain deeper insights into the signaling pathways associated with key genes and their underlying biological mechanisms, we initially assessed the correlation between the key genes and the expression levels of all genes within the training dataset. Following this, we ranked all genes in descending order based on their calculated correlation values and conducted GSEA enrichment analysis. The top five enriched pathways were highlighted in the results. The criteria for significance were set as |NES| > 1 (NES, Normalized Enrichment Score), P < 0.05, and q < 0.25.
3.8. Analysis of Pathway Differences between the SAP Group and the Control Group
GSVA can effectively capture the heterogeneity of gene expression profiles in different biological processes or signaling pathways. In this study, the “GSVA” and “limma” packages in R language, combined with the gene sets related to KEGG pathways included in the MSigDB database, were used to analyze the gene set variation between the SAP group and the Control group.
3.9. Immune Infiltration Analysis
In this study, the expression data of genes in the training set were used to infer the composition proportions of 28 types of immune cells in each sample through the ssGSEA algorithm. First, an immune infiltration characteristic analysis was performed on the samples, and then the Mann-Whitney U test was used to evaluate the differences in the immune cell infiltration levels between different groups. On this basis, the correlation between key genes and these differentially infiltrated immune cells was further explored.
3.10. Construction of the Cerna Network of Key Genes
To deeply analyze the upstream regulatory mechanism of key genes and construct a competitive endogenous RNA (ceRNA) regulatory network mediated by lncRNA and miRNA. Specifically, first, the miRDB database (https://mirdb.org/) was used to predict miRNAs related to key genes, and miRNAs with a Target Score higher than 60 were selected as candidate objects. Subsequently, these miRNAs were imported into the Starbase database to further predict the lncRNAs that might be regulated by them (clipExpNum ≥ 20). Finally, based on the above analysis results, a ternary regulatory network of “lncRNA-miRNA-key gene”, namely the ceRNA network, was successfully constructed.
3.11. Prediction of Small Molecule Drugs for Key Genes and Molecular Docking
Predict small molecule drugs related to key genes in the CTD database. The selection criteria are: Reference Count ≥ 2. Molecular docking is divided into three steps: Step 1: Download the tertiary structure of key genes from the PDB database and the tertiary structure of small molecule drugs from the pubchem database. Step 2: Use the software Autodock for molecular docking. Step 3: Use pymol for docking visualization (select the structures with lower binding free energy and hydrogen bonds in the output results (highest binding affinity)).
3.12. Statistical Analysis
Use R 4.0.3 for statistical analysis. The Shapiro-Wilk test verified that the data did not conform to a normal distribution, so the Wilcoxon rank sum test was used for inter-group difference analysis. A P value less than 0.05 is considered statistically significant.
4. Results
4.1. Identification of DEGs Related To Mitophagy
There were a total of 277 DEGs between the SAP group and the Control group in the GSE194331dataset, including 229 upregulated genes and 48 downregulated genes (Fig. 1A-1B). Moreover, the heatmap clearly presents the expression difference pattern of DEGs in the two groups of samples. By taking the intersection of mitophagy-related genes and DEGs through a Venn diagram, a total of 16 DEGs related to mitophagy were obtained (Fig. 1C).
Figure 1.
Identification of DEGs related to mitophagy. Volcano plot (A) and heatmap (B) of DEGs between the SAP group and the Control group in the GSE194331 dataset. C) Venn diagram of mitophagy-related genes and DEGs.
4.2. Functional Enrichment and Construction of PPI Network of DEGs Related To Mitophagy
To analyze the biological characteristics of DEGs related to mitophagy, we performed functional enrichment analysis. The results of GO enrichment analysis showed that these genes were significantly associated with immunosuppression and metabolism-related processes in biological processes; at the cellular component level, they were mainly distributed in structures such as granule membrane structures and membrane microdomains; and their molecular functions were focused on the regulation of enzyme activity and substance transport (Fig. 2A). Further KEGG enrichment analysis indicated that these genes were mainly enriched in metabolism-related pathways, such as propanoate metabolism and pentose phosphate pathway; meanwhile, they were closely associated with infection and immune response pathways (Fig. 2B). To explore the characteristics of the protein-protein interaction network, we conducted PPI analysis on these genes. The results showed that except for the SNX3 gene, which had no interaction with other proteins, 55 interaction edges were formed among the remaining 15 proteins, suggesting that the core protein group had the potential for coordinated regulation in function (Fig. 2C).
Figure 2.
Construction of functional enrichment and PPI network. A) GO enrichment analysis of DEGs related to mitophagy. B) KEGG pathway enrichment map of DEGs related to mitophagy. C) PPI interaction network.
4.3. Machine Learning for Screening Candidate Key Genes
In order to further screen candidate key genes that can serve as diagnostic markers for SAP, we performed machine learning on 16 DEGs related to mitophagy. The LASSO regression algorithm screened out 4 characteristic genes from the DEGs related to mitophagy (Fig. 3A). The SVM-RFE algorithm identified 9 genes (Fig. 3B). Subsequently, the genes obtained by these two algorithms were overlapped using a Venn diagram, and finally four genes, namely PGD, LMNB1, TXN, and PYGL, were obtained (Fig. 3C).
Figure 3.
Screening of candidate key genes by machine learning. A) LASSO regression analysis. B) SVM-RFE algorithm. C) Venn diagram of genes obtained by the two algorithms, and finally four candidate key genes were obtained.
4.4. Expression Level Verification of Candidate Key Genes and ROC Curve
To further clarify the expression differences of candidate key genes in the disease group and their potential as diagnostic markers, this study analyzed the expression of four candidate key genes in the training set and external validation set. The results showed that in the SAP group and Control group of the training set, there were significant differences in the expression levels of PGD, LMNB1, TXN, and PYGL genes, and their expression levels were significantly increased in the SAP group (Fig. 4A). In the SAP group and Control group of the validation set, there were also significant differences in the expression levels of PGD, LMNB1, TXN, and PYGL genes. Among them, the expression levels of PGD and LMNB1 were significantly increased in the SAP group, while the expression levels of TXN and PYGL genes were significantly decreased in the SAP group (Fig. 4B). We named PGD and LMNB1, the genes with consistent trends of gene expression levels in the training set and validation set, as key genes.
Figure 4.
Expression level validation, ROC curves, Nomogram and key gene validation. A) Differential expression analysis of the training set. B) Differential expression analysis of the validation set. C) ROC curve of the training set. D) ROC curve of the validation set. E) Nomogram and key gene validation by construction of a nomogram based on PGD and LMNB1 genes. F) ROC curve of the nomogram.
In order to investigate whether PGD and LMNB1 have good diagnostic efficacy, this study plotted ROC curves in both the training set and the external validation set. The results showed that in the training set, the AUC value of PGD was 0.966, and the AUC value of LMNB1 was 0.909 (Fig. 4C). In the validation set (GSE151927, the mouse liver tissue validation set), the AUC value of PGD was 0.891, and the AUC value of LMNB1 was 0.969 (Fig. 4D). The AUC values of both genes in the training set and the validation set were greater than 0.85, indicating that the two genes have a good ability to distinguish between diseased and normal samples, and their diagnostic efficacy is good.
4.5. Evaluation of the Clinical Predictive Efficacy of Key Genes Based on the Nomogram Model
To clarify the clinical predictive ability of these two key genes, we constructed a Nomogram (Fig. 4E) based on the standardized expression levels of the key genes. The AUC of the ROC curve was 0.953, which is greater than 0.9, demonstrating that the Nomogram model has excellent predictive efficiency (Fig. 4F).
4.6. Functional Synergy and Pathway Specificity between PGD and LMNB1 Genes
The functional pathways of PGD and LMNB1 genes were analyzed through GSEA. The results showed that both the PGD gene and the LMNB1 gene were significantly enriched in signaling pathways such as cell phagocytosis, chemokine signaling, and ribosome. This indicates that these two genes have a coordinated regulatory role in the synergistic enhancement of immune defense, the optimization of cell signal transduction efficiency, and the maintenance of protein synthesis homeostasis. The PGD gene was separately enriched in the actin cytoskeleton regulation pathway, suggesting that it may be involved in processes related to cell shape maintenance or movement, which is involved in the maintenance of pancreatic cell morphology, leukocyte migration, and may affect the process of pancreatic tissue damage and repair in SAP. The LMNB1 gene was separately enriched in leukocyte transendothelial migration, indicating that it may be associated with the regulatory mechanism of immune cell migration (Fig. 5A-5B).
Figure 5.
Pathway enrichment. A) GSEA-enriched pathways of the PGD gene. B) GSEA-enriched pathways of the LMNB1 gene. C) Diagram of GSVA pathway differences between groups.
4.7. Analysis of Pathway Differences between Groups
GSVA can identify pathways with significant differences among samples. Through GSVA enrichment analysis between the disease group and the normal group, the results show that pathways such as biological metabolism, glycolysis, and lipogenesis are significantly upregulated, while pathways such as WNT-β-catenin signaling and unfolded protein response were significantly downregulated (Fig. 5C). This indicates that there are differences in the activities of metabolism, signal transduction, and stress response-related pathways between the two groups.
4.8. Differential Analysis of the Immune Microenvironment and Correlation Analysis of Immune Cells
We further analyzed the differences in the immune microenvironment between the two groups through immune infiltration analysis and evaluated the correlation between key genes and immune cells. The results showed that among 28 types of immune cells, 15 immune cells had significant differences between the SAP group and the Control group. Among them, 13 immune cells showed a significantly increased immune cell score in the SAP group, and two immune cells showed a significantly decreased score, namely Activated B cell and Activated CD8 T cell (Fig. 6A). Most of the differentially expressed immune cells were strongly positively correlated with the PGD gene, such as Activated dendritic cell, Macrophage, Mast cell, Monocyte, Neutrophil, Gamma delta T cell, Immature dendritic cell, MDSC, Plasmacytoid dendritic cell, and Regulatory T cell (r > 0.6). Among them, activated dendritic cell had the strongest correlation, with an r value of 0.82 (Fig. 6B).
Figure 6.
Immune infiltration analysis and ceRNA network. A) Box plot of the differences in immune cell infiltration scores between groups. B) Correlation between key genes and differentially expressed immune cells. C) ceRNA network related to PGD and LMNB1 genes (lncRNA-miRNA-mRNA triple interaction). The red ones are key genes; the orange ones are miRNAs; the blue ones are lncRNAs.
4.9. Construction of the ceRNA Network of Key Genes
We attempted to predict and screen the regulatory networks of PGD and LMNB1 genes in SAP. Two miRNAs targeting PGD and six miRNAs targeting LMNB1 were predicted respectively. Meanwhile, a total of 28 lncRNAs related to 8 miRNAs were predicted. The final network topological structure with PGD and LMNB1 as the core presented 38 nodes and 68 edges (Fig. 6C).
4.10. Prediction of Small Molecule Drugs for Key Genes and Molecular Docking
To screen for potential therapeutic molecules targeting the key genes (PGD and LMNB1), we predicted the relevant small molecule drugs for the key genes through the CTD database. The small molecules predicted for PGD were Dietary Fats, Estradiol, and Zinc. Among them, neither Dietary Fats nor Zinc had a 3D structure. Therefore, we performed docking between PGD and Estradiol, and the docking affinity was -7.56 kcal.moL-1 (Fig. 7A). The small molecule drugs docked with LMNB1 were Cyclosporine, Tretinoin, Benzo (a)pyrene, Estradiol, Progesterone, and Valproic Acid. Among them, Cyclosporine and Tretinoin had no 3D structure. Therefore, we performed molecular docking between LMNB1 and Benzo (a)pyrene, Estradiol, Progesterone, and Valproic Acid respectively. The docking affinities were -5.79 kcal.moL-1 (Fig. 7B), -5.78 kcal.moL-1 (Fig. 7C), -6.88 kcal.moL-1 (Fig. 7D), and -2.67 kcal.moL-1 (Fig. 7E). With a binding free energy ≤ -5 kcal.moL-1 defined as the strong binding standard.
Figure 7.
Prediction of small molecule compounds and molecular docking. A) Docking conformation of PGD with Estradiol. B) Docking conformation of LMNB1 with Benzo (a)pyrene. C) Docking conformation of LMNB1 with Estradiol. D) Docking conformation of LMNB1 with Progesterone. E) Docking conformation of LMNB1 with Valproic Acid.
5. Discussion
SAP represents a significant global health challenge. Approximately 20% of patients with acute pancreatitis progress to SAP, with a mortality rate reaching up to 30%.The current treatment for SAP is mainly supportive care that inhibits pancreatic enzyme activity and organ function damage, while immunotherapy and gene-targeted therapy are still in the research stage. Mitochondrial damage and abnormal activation of inflammatory factors are direct causes of the occurrence and development of SAP. Screening for targeting genes involved in mitochondrial autophagy can facilitate the early detection of SAP. Therefore, through bioinformatics analysis, we identified PGD and LMNB1 as key genes related to mitochondrial autophagy in SAP, among which PGD was highly positively correlated with immune cells such as activated dendritic cells, providing a theoretical basis for analyzing the molecular mechanism of SAP and formulating precise diagnosis and treatment strategies.
In this study, we identified PGD and LMNB1 as key genes for mitochondrial autophagy in SAP. We found upregulation trend of PGD and LMNB1 in the SAP group. PGD is considered an antioxidant enzyme that mainly participates in the synthesis and metabolism of NADPH and glutathione, thereby reducing ROS and achieving an antioxidant effect, and plays a significant role in cellular oxidative stress response, cell growth and proliferation. Recent studies have found that PGD can inhibit glucose metabolism and migration of tumors ( 8 ). And mitochondrial autophagy is closely associated with cellular ferroptosis. In sepsis-induced myocardial injury, PGD downregulation leads to impaired mitochondrial autophagy, resulting in cellular ferroptosis and organ damage ( 9 ). LMNB1 is predominantly localized to the inner nuclear membrane, where it interacts with transcription factors to modulate gene transcription and DNA replication, contributing to the maintenance of nuclear structural integrity. In some studies, it has been found that LMNB1 is overexpressed in tumors, thereby promoting tumor cell growth rate and colony-forming ability, and can serve as an early biomarker for prostate cancer and hepatocellular carcinoma ( 10 ). In sepsis-induced acute kidney injury, LMNB1 mediates mitochondrial damage and ferroptosis in renal tubular epithelium through the MDM2-P53 pathway ( 11 ). Another study found that LMNB1 expression was significantly increased in sepsis and could induce mitochondrial damage in cells through activating the NF-κB and JAK2/STAT3 pathways ( 11 ). Both PGD and LMNB1 are associated with mitochondrial functional status, but the relationship with SAP has not been reported so far.
To further explore some of the related signaling pathways and potential biological mechanisms present in the key genes, this study conducted GSEA enrichment analyses of the PGD and LMNB1 genes and identified several important pathways. Enrichment analysis revealed that the important pathways affected by PGD were Fc-γ receptor-mediated phagocytosis, chemokine signaling, regulation of actin cytoskeleton, and ribosome. The important pathways affected by the LMNB1 gene are: Fc-γ receptor-mediated phagocytosis, chemokine signaling, and leukocyte transendothelial migration. The Fc-γ receptor in the Fcγ receptor-mediated phagocytosis signaling pathway is primarily expressed on the surface of macrophages, dendritic cells, and NK cells. It specifically binds to the Fc region of immunoglobulin G (IgG). When IgG binds to targets such as pathogen/tumor cells, FcγR recognizes IgG-Fc segments and aggregates, FcγR intracellular segments contain immune receptor tyrosine activation motifs (ITAM), which are phosphorylated by Src family kinases and recruit Syk kinases. Syk activates PI3K, Rac/Rho GTP enzymes, drives actin remodeling, and forms the phagocytic cup. After maturation of the phagosome, it fuses with the lysosome, where the engulfed target is degraded. The FcγR phagocytic pathway is the core clearance mechanism of innate immunity, and its abnormal activation or inhibition is closely related to cancer and inflammatory diseases. Targeted regulation of this pathway has become an important direction for novel immunotherapy strategies ( 12 ). At present, research on this pathway is mainly focused on rheumatic immune diseases such as viral infections and systemic sclerosis ( 13 ), where abnormal enhancement leads to abnormal activation of inflammatory factors such as IL-6, reactive oxygen species and phagocytosis. In recent years, bionic drugs have been developed to treat atopic dermatitis by inhibiting Fcγ R-mediated phagocytic signaling and reducing the continuous accumulation of ROS ( 14 ). But the role in SAP has not been studied, and it may become a new research direction in the future. Chemokine signaling pathway, which describes the activation of intracellular G proteins or kinases by chemokines binding to cell surface receptors (such as CCR, CXCR), ultimately regulates processes such as cell migration and immune response by activating downstream pathways including JAK/STAT, Ras, ERK, Akt. Mitochondria release of immune-stimulating molecules by regulating chemokine secretion leads to inflammatory signaling, and the interaction between mitochondria and the signaling pathways that regulate inflammation affects the progression of age-related diseases, including neurodegenerative diseases (Alzheimer’s disease and Parkinson’s disease, as well as amyotrophic lateral sclerosis), osteoarthritis, and sarcopenia. Chemokine pathways are significantly linked to the occurrence and development of acute pancreatitis. Analysis of mRNA expression profiles in patients with acute pancreatitis and normal individuals revealed that three chemokine-related genes (DECRG) were associated with the severity of acute pancreatitis by regulating the Th1 and Th2 cell differentiation signaling pathways ( 15 ). Chemokines are closely associated with inflammatory responses induced by neutrophils and macrophages. Fibroblasts secrete CXCL1, which functions to chemotactically recruit neutrophils and promote excessive neutrophil infiltration during acute pancreatitis ( 16 ). This mechanism was further supported by studies in CXCL2 gene knockout mice, which exhibited a protective effect against acute pancreatitis ( 17 ). In vitro cell experiments also revealed that CCL3 activates the JNK/p38MAPK pathway in pancreatic cells and promotes the secretion of TNF-α and IL6, leading to aggravated pancreatitis ( 18 ). Chemokines are also involved in the migration and activation of macrophages. HMGB1 secreted by macrophages can induce CXCL12 / CXCR4, and targeting inhibition of the CXCL12 / CXCR4 pathway can alleviate the inflammatory response in pancreatitis ( 19 ). CCL2 also induces macrophages to migrate from the bone marrow to the periphery during pancreatitis, triggering pancreatitis ( 20 ). In addition to its effects on neutrophils and macrophages, chemokines such as MCP-1 can also cause acute pancreatitis by increasing calcium ion levels in pancreatic acinar cells ( 21 ).
Through analyses such as ceRNA network construction and drug prediction, our molecular docking indicates that estradiol and progesterone target the regulation of PGD and LMNB1. Previous clinical observations have found that elevated serum estradiol levels at admission in patients with severe pancreatitis are associated with high mortality ( 22 ); Postmenopausal estrogen and progesterone replacement therapy increases the risk of acute pancreatitis ( 23 ), and some studies suggest that hormone replacement therapy may cause abnormally elevated triglycerides, but the specific mechanism is unknown ( 24 ). However, the effect of estradiol is multifactorial, and the findings in the literature are contradictory. Other animal experiments have shown that in pancreatitis rats, the application of estradiol and estradiol agonists can provide protection by reducing oxidative stress damage in pancreatic and lung tissues ( 25 ). Recent COVID-19 studies have also found that estradiol can be involved in anti-inflammatory and immunomodulatory properties in the context of cytokine storm immune dysregulation ( 26 ). After induced pancreatitis in ovariectomized rats, it was found that estradiol treatment may reduce autophagic flux, leading to more severe pancreatitis ( 27 ). Based on our target analysis, it is speculated that estrogen may cause mitochondrial autophagy disorder through the regulation of PGD or LMNB1 genes, leading to the occurrence and development of pancreatitis, but the specific mechanism of action needs to be further confirmed through experiments.
At present, research on diagnostic markers for mitochondrial autophagy-related genes in SAP is still relatively scarce. This study screened out potential key mitochondrial autophagy-related genes in SAP through bioinformatics analysis of transcriptome data: PGD and LMNB1. The analysis suggests that both may be involved in the disease progression of SAP by regulating Fc-γ receptor-mediated phagocytosis and related cytokine signaling pathways. However, this study has certain limitations. The specific molecular mechanisms through which the PGD and LMNB1 genes regulate their downstream signaling pathways remain unclear and warrant further investigation in future studies.
6. Conclusion
In this study, we screened genes related to mitochondrial autophagy in SAP based on bioinformatics methods. We found that PGD and LMNB1 are key genes related to mitophagy in SAP and have excellent diagnostic value. The analysis suggests that both LMNB1 and PGD may be involved in the disease progression of SAP by regulating Fc-γ receptor-mediated phagocytosis and related cytokine signaling pathways. We also predicted the relevant small molecule drugs for the key genes, such as Estradiol and Progesterone. Previous clinical observations have found that elevated serum estradiol levels in patients with severe pancreatitis are associated with high mortality, but the mechanism is unknown. Based on our target analysis, it is speculated that estrogen may cause mitochondrial autophagy disorder through the regulation of PGD or LMNB1 genes, leading to the occurrence and development of pancreatitis. Overall, PGD and LMNB1 are key genes related to mitophagy in SAP and have excellent diagnostic value. This study provides a theoretical basis for the analysis of the molecular mechanism of SAP and the development of precise diagnosis and treatment strategies.
Acknowledgements
We thank all authors for their support.
List of abbreviations
SAP: Severe acute pancreatitis; DEGs :Differentially expressed genes; LASSO: least absolute shrinkage and selection operator; SVM-RFE: Support Vector Machine-Recursive Feature Elimination; ROC: receiver operator characteristic; GSEA: Gene Set Enrichment Analysis; GSVA: Gene Set Variation Analysis; ceRNA: competitive endogenous RNA;PGD: phosphogluconate dehydrogenase; AUC: area under the curve; MMP :mitochondrial membrane potential; mPTP :permeability transition pore; ROS: reactive oxygen species; GO: gene ontology; KEGG :Kyoto Encyclopedia of Genes and Genomes; DECRG: chemokine-related genes.
Ethics approval and consent to participate
Not applicable
Consent for publication
Not applicable
Availability of data and materials
The datasets used and/or analyzed during the current study are available from GEO. Mitophagy-related genes were collected from both the Genecards database and the Msigdb database.
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
LJJ and CJX designed the study. CJX performed the bioinformatics analysis and interpretation of the data. LJJ wrote the manuscript. WZP modified the manuscript. MYM and QWJ revised the manuscript and gave final approval of the version to be published. All authors read and approved the final manuscript. All authors have accepted responsibility for the entire content of this manuscript and approved its submission.
References
- 1.Iannuzzi JP, King JA, Leong JH, Quan J, Windsor JW, Tanyingoh D, et al. Global Incidence of Acute Pancreatitis Is Increasing Over Time: A Systematic Review and Meta-Analysis. Gastroenterology. 2022;162(1):122–134. doi: 10.1053/j.gastro.2021.09.043. [DOI] [PubMed] [Google Scholar]
- 2.Ye W, Zheng C, Yu D, Zhang F, Pan R, Ni X, et al. Lipoxin A4 Ameliorates Acute Pancreatitis-Associated Acute Lung Injury through the Antioxidative and Anti-Inflammatory Effects of the Nrf2 Pathway. Oxid Med Cell Longev. 2019;2019:2197017. doi: 10.1155/2019/2197017. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Wu X, Sun H, Luo Z, Cheng L, Duan X, Ren J. Plasma-derived exosomes contribute to pancreatitis-associated lung injury by triggering NLRP3-dependent pyroptosis in alveolar macrophages. Bba-Mol Basis Dis. 2020;1866(5):165685. doi: 10.1016/j.bbadis.2020.165685. [DOI] [PubMed] [Google Scholar]
- 4.Yan J, Liu H, Xu Y, Sun X, Meng X, Wei S, et al. Tailored “Three-stage booster” nano-extinguisher for synergistic treatment of severe acute pancreatitis by rectifying mitochondrial dysfunction and inhibiting pancreatic autodigestion. Acta Biomater. 2025;200:569–590. doi: 10.1016/j.actbio.2025.05.049. [DOI] [PubMed] [Google Scholar]
- 5.Kang H, Hu Q, Yang Y, Huang G, Li J, Zhao X, et al. Urolithin A’s Role in Alleviating Severe Acute Pancreatitis via Endoplasmic Reticulum-Mitochondrial Calcium Channel Modulation. Acs Nano. 2024;18(21):13885–13898. doi: 10.1021/acsnano.4c03044. [DOI] [PubMed] [Google Scholar]
- 6.Yang Y, Peng Y, Li Y, Shi T, Xu N, Luan Y, et al. Sestrin2 balances mitophagy and apoptosis through the PINK1-Parkin pathway to attenuate severe acute pancreatitis. Cell Signal. 2025;126:111518. doi: 10.1016/j.cellsig.2024.111518. [DOI] [PubMed] [Google Scholar]
- 7.Wen E, Tian Y, Fang M, Zhang Y, Zhao H, Wang Z, et al. The P2X7-Mediated Mitochondrial ROS as an Emerging Core Target of Tuftsin Nanoparticles in Severe Acute Pancreatitis Therapy via Regulating Mitophagy. Acs Appl Mater Inter. 2025;17(5):7521–7538. doi: 10.1021/acsami.4c21010. [DOI] [PubMed] [Google Scholar]
- 8.Chen J, Chen W, Wang Z, Zhou L, Lin Q, Huang Q, et al. PGD: Shared gene linking polycystic ovary syndrome and endometrial cancer, influencing proliferation and migration through glycometabolism. Cancer Sci. 2024;115(9):2908–2922. doi: 10.1111/cas.16212. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Liu R, Li F, Hao S, Hou D, Zeng X, Huang H, et al. Low-dose Olaparib improves septic cardiac function by reducing ferroptosis via accelerated mitophagy flux. Pharmacol Res. 2024;200:107056. doi: 10.1016/j.phrs.2023.107056. [DOI] [PubMed] [Google Scholar]
- 10.Hong J, Liang S, Wang AS, Yeh C, Huang H, Sun C, et al. LMNB1, a potential marker for early prostate cancer progression. Am J Cancer Res. 2022;12(7):3390–3404. [PMC free article] [PubMed] [Google Scholar]
- 11.Hu C, Zhang B, Zhao S. METTL3-mediated N6-methyl-adenosine modification stimulates mitochondrial damage and ferroptosis of kidney tubular epithelial cells following acute kidney injury by modulating the stabilization of MDM2-p53-LMNB1 axis. Eur J Med Chem. 2023;259:115677. doi: 10.1016/j.ejmech.2023.115677. [DOI] [PubMed] [Google Scholar]
- 12.Luo J, Chen Y, Xu Y, Tang M, Zhang X. Morphine contributed to the deterioration of cancer via miR-543/MARCKS/FcgammaR-mediated phagocytosis pathway. J Pharm Pharmacol. 2019;71(10):1584–1598. doi: 10.1111/jphp.13146. [DOI] [PubMed] [Google Scholar]
- 13.Hukara A, Bonazza GA, Tabib T, Micheroli R, Jordan S, Burki K, et al. Elevated Fcy receptor expression augments pro-inflammatory macrophage phagocytosis in systemic sclerosis and associated rheumatic diseases. Rheumatology. 2025;64(6):3975–3988. doi: 10.1093/rheumatology/keae688. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Huang F, Lu X, Kuai L, Ru Y, Jiang J, Song J, et al. Dual-Site Biomimetic Cu/Zn-MOF for Atopic Dermatitis Catalytic Therapy via Suppressing FcgammaR-Mediated Phagocytosis. J Am Chem Soc. 2024;146(5):3186–3199. doi: 10.1021/jacs.3c11059. [DOI] [PubMed] [Google Scholar]
- 15.Mo S, Wu W, Luo K, Huang C, Wang Y, Qin H, et al. Identification and analysis of chemokine-related and NETosis-related genes in acute pancreatitis to develop a predictive model. Front Genet. 2024;15:1389936. doi: 10.3389/fgene.2024.1389936. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Wang Q, Zhang X, Han C, Lv Z, Zheng Y, Liu X, et al. Immunodynamic axis of fibroblast-driven neutrophil infiltration in acute pancreatitis: NF-kappaB-HIF-1alpha-CXCL1. Cell Mol Biol Lett. 2025;30(1):57. doi: 10.1186/s11658-025-00734-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Steele CW, Karim SA, Foth M, Rishi L, Leach JDG, Porter RJ, et al. CXCR2 inhibition suppresses acute and chronic pancreatic inflammation. J Pathol. 2015;237(1):85–97. doi: 10.1002/path.4555. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Wu X, Ji K, Wang H, Zhao Y, Jia J, Gao X, et al. MIP-1alpha induces inflammatory responses by upregulating chemokine receptor 1/chemokine receptor 5 and activating c-Jun N-terminal kinase and mitogen-activated protein kinase signaling pathways in acute pancreatitis. J Cell Biochem. 2019;120(3):2994–3000. doi: 10.1002/jcb.27049. [DOI] [PubMed] [Google Scholar]
- 19.Irie Y, Tsubota M, Ishikura H, Sekiguchi F, Terada Y, Tsujiuchi T, et al. Macrophage-derived HMGB1 as a Pain Mediator in the Early Stage of Acute Pancreatitis in Mice: Targeting RAGE and CXCL12/CXCR4 Axis. J Neuroimmune Pharm. 2017;12(4):693–707. doi: 10.1007/s11481-017-9757-2. [DOI] [PubMed] [Google Scholar]
- 20.Saeki K, Kanai T, Nakano M, Nakamura Y, Miyata N, Sujino T, et al. CCL2-induced migration and SOCS3-mediated activation of macrophages are involved in cerulein-induced pancreatitis in mice. Gastroenterology. 2012;142(4):1010–1020. doi: 10.1053/j.gastro.2011.12.054. [DOI] [PubMed] [Google Scholar]
- 21.Ramnath RD, Sun J, Bhatia M. Role of calcium in substance P-induced chemokine synthesis in mouse pancreatic acinar cells. Brit J Pharmacol. 2008;154(6):1339–1348. doi: 10.1038/bjp.2008.188. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Lu C, Liu L, Hsieh Y, Yang L, Chen R, Hsieh C. Increased admission serum estradiol level is correlated with high mortality in patients with severe acute pancreatitis. J Gastroenterol. 2013;48(3):374–381. doi: 10.1007/s00535-012-0636-6. [DOI] [PubMed] [Google Scholar]
- 23.Oskarsson V, Orsini N, Sadr-Azodi O, Wolk A. Postmenopausal hormone replacement therapy and risk of acute pancreatitis: a prospective cohort study. Can Med Assoc J. 2014;186(5):338–344. doi: 10.1503/cmaj.131064. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Asai Y, Zhang Q, Zheng L, Li J, Zhang Z, Gao HL. Acute pancreatitis caused by oral contraceptives. Intern Emerg Med. 2025;20(2):529–533. doi: 10.1007/s11739-025-03897-4. [DOI] [PubMed] [Google Scholar]
- 25.Guleken Z, Ozbeyli D, Acikel-Elmas M, Oktay S, Alev B, Sirvanci S, et al. The effect of estrogen receptor agonists on pancreaticobiliary duct ligation induced experimental acute pancreatitis. J Physiol Pharmacol. 2017;68(6):847–858. [PubMed] [Google Scholar]
- 26.Mauvais-Jarvis F, Klein SL, Levin ER. Estradiol, Progesterone, Immunomodulation, and COVID-19 Outcomes. Endocrinology. 2020;161(9) doi: 10.1210/endocr/bqaa127. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Wang M, Gorelick F. Ovariectomy Affects Acute Pancreatitis in Mice. Digest Dis Sci. 2022;67(7):2971–2980. doi: 10.1007/s10620-021-07116-w. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Data Availability Statement
The datasets used and/or analyzed during the current study are available from GEO. Mitophagy-related genes were collected from both the Genecards database and the Msigdb database.