Abstract
Background
Primary graft dysfunction (PGD) usually occurs within 72 hours after lung transplantation and is primarily caused by ischemia-reperfusion injury (IRI). Patients who develop PGD after lung transplantation tend to have a poor prognosis. However, effective clinical strategies to reduce the incidence of primary graft dysfunction remain limited. Therefore, a comprehensive understanding of the mechanisms underlying lung ischemia-reperfusion injury is essential for improving outcomes in lung transplant recipients.
Methods
In this study, we explored the differential expression of metabolism-related genes in lung transplantation induced IRI and identify its potential molecular mechanisms by bioinformatics analysis. Next, we used two machine learning algorithms and further screened for key genes in them. The outside dataset GSE8021 was used to validated the accuracy of the model established by metabolism-related genes machine learning genes. In addition, we observed the distribution and localization of metabolism-related machine learning genes in the single-cell dataset GSE220797 and analyzed the correlation between metabolism-related machine learning genes and immune cells by the CIBERSORT immune infiltration algorithm. Finally, we validated the nine metabolism-related machine learning genes by rat orthotopic left lung transplantation model and Real-Time Quantitative Polymerase Chain Reaction (RT-qPCR), we found that seven of these metabolism-related machine learning genes were consistent with the results of the bioinformatics analysis.
Results
We identified multiple metabolism-related genes machine learning genes (PDE4B, CDA, HMOX1, EHHADH, AMD1, GUCY1A1, GUCY1B1, UGCG, and FPGT). Significant changes were observed in some of these genes following ischemia-reperfusion. They represent important biomarkers in ischemia-reperfusion injury induced by lung transplantation and hold promise as therapeutic targets for mitigating lung ischemia-reperfusion injury and reducing the incidence of primary graft dysfunction.
Keywords: Ischemia/Reperfusion injury, Metabolism, Lung transplantation, Primary graft dysfunction, Bioinformatics
Introduction
Lung transplantation serves as an effective life-saving intervention and significantly enhances the quality of life for patients with end-stage lung diseases (e.g., idiopathic pulmonary fibrosis, chronic obstructive pulmonary disease, pulmonary hypertension, and silicosis) (Adegunsoye et al., 2017). In the past few years, the number of lung transplantations has increased significantly. Advances in donor-recipient matching, organ preservation, surgical techniques, and perioperative care have improved posttransplant survival rates (Catelli et al., 2024). However, a substantial proportion of recipients still experience limited long-term survival (Singh et al., 2023). Primary graft dysfunction (PGD) represents not only the leading cause of early mortality following lung transplantation (Wong & Hsin, 2024) but also a significant risk factor for the development of bronchiolitis obliterans syndrome (BOS) in posttransplant patients (Daud et al., 2007). PGD usually manifests within the first 72 h after lung transplantation (Snell et al., 2017) and is characterized by progressive hypoxemia and the development of diffuse alveolar infiltrates on radiographic imaging (Porteous & Lee, 2017). Ischemia-reperfusion injury (IRI) is a major cause of PGD (Wong & Hsin, 2024). Therefore, it is important to understand the pathogenesis of lung ischemia-reperfusion injury (LIRI) after lung transplantation and utilize this information to attenuate the onset and progression of PGD.
IRI refers to the phenomenon of restoring blood perfusion to an organ on the basis of ischemia, thereby causing further aggravation of ischemic injury in that organ or even irreversible injury. During the course of IRI, a series of pathophysiological changes occur in lung tissues, such as increased intracellular calcium ion concentrations, overactivation of inflammatory responses, excessive production of reactive oxygen radicals, and various types of cell death (Capuzzimati, Hough & Liu, 2022). This subsequently leads to pulmonary vascular endothelial cell dysfunction and endothelial barrier breakage (Ve & Ak, 2016), ultimately leading to pulmonary oedema and severe respiratory failure.
Many studies in organ transplantation (such as heart, kidney) have shown that abnormal changes in metabolic function occur in cells and tissues, both in the ischemic state and during reperfusion (Feng et al., 2025; Huang et al., 2025). In recent years, several studies have identified metabolic changes occurring in LIRI induced by lung transplantation. First of all, Van Slambrouck et al. (2025) performed metabolomic sequencing on lung tissue before and after transplantation. They observed elevated levels of amino acids, hypoxanthine, uric acid, and glutathione disulfide after lung transplantation, alongside decreased levels of glucose and carnitine. Meanwhile, Liang et al. (2025) also reported changes in specific metabolite levels during ex vivo lung perfusion (EVLP) and transplantation in a Yorkshire pig model. These findings indicate metabolic reprogramming occurs during the ischemia-reperfusion injury induced by lung transplantation. Furthermore, Wang et al. (2025) observed changes in certain metabolic pathways in lungs stored at 10 °C compared to ice-preserved lungs, (e.g., increased activity in the Tricarboxylic Acid Cycle). This indicates that alterations in lung tissue metabolism are associated with improved donor lung quality and reduced incidence of PGD. Lung transplantation-induced ischemia-reperfusion injury involves not only changes in metabolites but also differences in metabolism-related genes. For instance, Wong et al. (2020)’s transcriptomic sequencing revealed that genes associated with fatty acid β-oxidation and protein synthesis were downregulated in LIRI induced by EVLP and lung transplantation. In summary, the levels of metabolic products and the expression of metabolism-related genes undergo significant alterations in pulmonary ischemia-reperfusion injury, which warrant our attention.
Given the critical roles of metabolic alterations in all aspects of IRI, elucidating the underlying molecular mechanisms is essential for identifying suitable biomarkers and therapeutic targets. With the advancement of scientific technology, transcriptomic sequencing is capable of generating hundreds to thousands of transcripts and isoforms (Krassowski et al., 2020); thus, we are able to gain a comprehensive understanding of gene regulation in a wide range of diseases. For example, transcriptomic analyses have revealed that inflammatory response activation is a key mechanism in LIRI (Jeon et al., 2024). In addition, single-cell transcriptomics allows for cell type-specific resolution of gene expression dynamics (Vandereyken et al., 2023), thus informing the rational selection of suitable cell populations for experimental verification of molecular mechanisms (Li et al., 2021). However, the metabolic alterations associated with LIRI have not been well studied. Few studies have analyzed the expression of metabolism-related genes in these tissues or cells via public databases. To address this gap, we screened metabolism-related genes that are significant in LIRI via a public database, followed by experimental validation in animal models. Finally, our study revealed several metabolism-related genes that are differentially expressed in rat IRI models.
Materials & Methods
Data source
All datasets were downloaded from the Gene Expression Omnibus (GEO) public database (http://www.ncbi.nlm.nih.gov/geo/). GSE127003 comprises 46 pairs of pre-transplantation and post-transplantation human lung tissue (Wong et al., 2020), GSE8021 consists of 16 lung tissue from patients developed PGD and 36 lung tissue from patients did not develop PGD, and GSE220797 consists of six pre-transplantation and six post-transplantation samples (Wong et al., 2024). Moreover, 948 metabolism-related genes (MRGs) were obtained from the Molecular Signatures Database (Msigdb, https://www.gsea-msigdb.org/gsea/msigdb) (Liberzon et al., 2015). Figure 1 presents a general flowchart of this study.
Figure 1. Flowchart of the steps of this bioinformatic analysis and experimental validation.
DEGs, differentially expressed genes; MRGs, metabolism-related genes; WGCNA, weighted correlation network analysis; GO, Gene Ontology; KEGG, Kyoto Encyclopedia of Genes and Genomes.
Analysis of differentially expressed genes
We employed the R package “limma” to identify differentially expressed genes (DEGs) between pre-transplantation and post-transplantation samples in the GSE127003 dataset. The statistical criteria for differential expression were a —log2-fold change (logFC)— > 0.5 and a P value < 0.05. Visualization of downregulated and upregulated DEGs was performed via the R package “ggvolcano” for volcano plots. Visualization of the expression levels of the 150 most differentially expressed genes in pre-transplantation and post-transplantation lung tissues was conducted via the R package “pheatmap” for generating heatmaps.
Weighted correlation network analysis and functional enrichment analysis
We used the R package weighted correlation network analysis (WGCNA) to identify module genes associated with IRI after lung transplantation (Langfelder & Horvath, 2008). The genes included in the DEG, WGCNA, and metabolism-related genes analyses were described via the R package “ggvenn”.
To analyze the biological functions and pathways associated with metabolism-related genes, the R package “clusterProfiler” was used to perform Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) analyses (Wu et al., 2021). In the GO analysis, the identified metabolism-related genes were categorized into three classes: biological process (BP), cellular component (CC), and molecular function (MF). KEGG enrichment analysis was subsequently conducted to predict signaling pathways. The criterion for significant enrichment was P < 0.05. The “ggplot2” package in R software was used to generate scatter plots.
Selection of core genes via machine learning
Further screening was conducted to identify genes associated with metabolism in Least Absolute Shrinkage and Selection Operator (LASSO) logistic regression. The R package “glmnet” was employed for LASSO logistic regression, where the minimum lambda was considered optimal. Through random forest (RF) with the R package “randomForest”, we identified key genes by pinpointing the point of lowest cross-validation error. Subsequently, receiver operating characteristic (ROC) curve and area under the curve (AUC) calculations were performed on the GSE127003 and GSE8021 datasets via the R package “pROC” to evaluate the sensitivity and specificity of key genes (Robin et al., 2011).
Immune cell infiltration analysis
The CIBERSORT algorithm was used to infer the infiltration status of immune cell types in the samples. “Heatmaps” and “box plots” in R were employed to visualize the differences in immune cell infiltration between pre-reperfusion and post-reperfusion samples, with P < 0.05 considered statistically significant. Additionally, Pearson correlation analysis was conducted to validate the associations between key genes and immune infiltration. Visualization of the results was performed via the R package “corrplot”.
Differential expression of core genes in the single-cell database
We downloaded the scRNA-seq data of IRI patients from the GSE220797 dataset, which is composed of six pre-transplantation samples and six post-transplantation samples. In R software, we used the “Seurat” pipeline to create a Seurat-format object (Hao et al., 2021) and calculated the proportions of mitochondrial genes and haemoglobin genes. Low-quality cells that had more than 15% mitochondrial genes and more than 5% haemoglobin genes were excluded. We also guaranteed that all the genes were expressed at levels between 300 and 4,000. A total of 91,340 cells were kept for further exploration. In accordance with other researchers’ experience, the remaining cells were further scaled and normalized via a linear regression model with the “log normalization” method, and the top 2,000 variable genes were detected via the “FindVariableFeatures” function (Zhou et al., 2025). Subsequently, the dimensionality of the scRNA-seq data was diminished through principal component analysis (PCA). Uniform manifold approximation and projection (UMAP) dimensional reduction, dataset integration, and cell types were annotated with the markers we searched from the website cellmarker2.0 (Hu et al., 2023). To remove the batch effects among the samples, we used the “Harmony” package (Korsunsky et al., 2019).
Animals
Specific pathogen free (SPF) grade male SD rats were purchased from Changzhou Cavans Animal Experiment Co., Ltd. (Changzhou, China). All animals were housed in a temperature-controlled (22 ± 2 °C) and humidity-controlled room (45–50%) with free access to both fresh water and standard laboratory food at Wuxi People’s Hospital Animal Experiment Center. The rats involved in the experiment were confirmed dead after their lungs were removed under anesthesia and the remaining ones were euthanized before the end of experiment program. The rats were euthanized by inhalation of carbon dioxide gas (flow rate: 50–60% of cage volume/min, no prefilled chambers), ensuring that no other animals were present at the execution site. The bodies were properly handled only after the rats were confirmed dead. All animal experiments were conducted in accordance with the Guide for the Care and Use of Laboratory Animals, and the animal study protocol was approved by the Institutional Ethics Committee of Nanjing Medical University Affiliated Wuxi People’s Hospital (DL2024023, 25 November 2024).
Rat orthotopic left lung transplantation model and hematoxylin & eosin staining
Nine- to ten-week-old Sprague Dawley rats weighing 250 to 300 g were used for left lung transplantation. The protocol of the surgical procedures are studied from our recent article (Yang et al., 2025). We deprived all rats of food twelve hours before harvesting and transplanting the lungs. The donor rats were anesthetized with 5% isoflurane and placed in a supine position. After hair removal and disinfection, a midline thoracoabdominal incision was performed. Heparin (500 IU/kg) was administered via the inferior vena cava, followed by pulmonary artery perfusion with low-potassium dextran (LPD) solution (4 °C, 20 mL). The heart-lung blocks were then excised, inflated with a 50% oxygen and 50% nitrogen gas mixture (2.5 mL), and the main bronchus was ligated. The heart-lung block was stored on ice in cold LPD solution at 4 °C for 12 h. An orthotopic left lung transplantation model was established using the cuff technique. After 4 h of reperfusion, the recipient rats were euthanized, and the transplanted lungs were collected for further analyses. Then, lung tissues were fixed in 4% paraformaldehyde and then embedded in paraffin. Tissue blocks were cut into 5-µm slices, stained with hematoxylin and eosin (H&E).
RNA extraction and reverse transcription
All lung tissue samples were stored at −80 °C and processed for RNA extraction within one week. Total RNA was extracted from rat left lung tissues via an RNAeasyTM kit (Beyotime, Shanghai, China, R0026). Throughout the extraction procedure, we exclusively used DNase/RNase-free ultrapure water to prevent nucleic acid degradation. RNA concentration was quantified using a NanoDrop 2000c spectrophotometer (Thermo Fisher Scientific, Waltham, MA, USA). Samples with 260/280 absorbance ratios between 1.8 and 2.2, indicating acceptable purity, were subsequently subjected to reverse transcription. After that, the total RNA was reverse transcribed into cDNA via HiScript III RT SuperMix for qPCR (Vazyme, Beijing, China, R223-01) with the following reaction system: one µg total RNA in a 20 µL volume containing four µL 5 × HiScript III qRT SuperMix, with RNase-free H2O added to adjust the final volume. The reaction mixture was briefly centrifuged and subjected to thermal cycling under the following conditions: 37 °C for 15 min, 85 °C for 5 s, followed by cooling to 4 °C. The synthesized cDNA was stored at −80 °C for subsequent use.
Real-time quantitative polymerase chain reaction (RT-qPCR)
Real-time quantitative polymerase chain reaction (RT-qPCR) was conducted via the use of ChamQ Universal SYBR qPCR Master Mix (Vazyme, Beijing, China, Q711-02) on an Applied Biosystems 7500 Real-time PCR Detection System (ABI, United States). All primers were commercially synthesized by Accurate Biology (Hunan, China) and subsequently diluted in nuclease-free ultrapure water to prevent degradation, the sequences of primers used are listed in Table S1. The following reaction was: two µL cDNA, five µL SYBR qPCR Master Mix, 0.5 µL the primers, with RNase-free H2O added to adjust the final volume. Thermal cycling conditions included an initial denaturation at 95 °C for 60 s, 40 cycles of denaturation at 95 °C for 10 s, annealing at 50 °C for 120 s, and extension at 60 °C for 30 s. The relative expression levels of the target genes were normalized to those of β-actin (ACTB) (N = 3), and the fold changes were calculated via the 2−ΔΔCt method, we selected the results within the Ct value range from 15 to 35. The results of fold changes were listed in Table S2. All experiments were repeated three times.
Statistical analysis
Our statistical analyses were performed via R software (version 4.3.3) and GraphPad Prism (version 9.0.0). Differences between data from two sample groups were calculated via an unpaired, two-tailed t test. Pearson correlation coefficients were used to access the correlation between different variables, and p < 0.05 was considered statistically significant. The sequences of primers used are listed below (Table S1).
Results
Differential expression analysis and weighted gene co-expression network analysis
Through analysis of the GSE127003 dataset, we identified 695 DEGs, consisting of 555 upregulated and 140 downregulated genes (Fig. 2B). Principal component analysis (PCA) revealed that the comparability between the two groups was obvious (Fig. 2A). The heatmap illustrates the expression of the 150 genes with the greatest differences in expression among the samples subjected to IRI induced by lung transplantation (Fig. 2C). Additionally, we identified a soft-thresholding power of 11 as optimal by WGCNA, and a total of 14 modules were identified through analysis (Fig. 2E). The module-trait heatmap depicts the association between gene modules and ischemia-reperfusion injury (Fig. 2D). We selected six key modules (green, yellow, black, red, brown, and turquoise) consisting of 4,478 genes that exhibited a strong correlation with lung transplant reperfusion injury (correlation coefficient > 0.4).
Figure 2. DEG analysis and WGCNA of the GSE127003 dataset.
(A) PCA map of two groups of lung samples from the GSE127003 dataset, including pre-transplantation and post-transplantation samples. (B–C) Volcano plot and heatmap of differentially expressed genes from the GSE127003 dataset. (D) Module trait heatmap of the WGCNA results from the GSE127003 dataset. (E) Selecting soft thresholding power.
Identification of metabolism-related hub genes and functional enrichment analysis
We identified 948 metabolism-related genes from Msigdb, and the Venn diagram depicted 34 shared hub genes between metabolism-related genes, DEGs, and the key modules of WGCNA (Fig. 3A). The location of copy number variations on chromosomes is shown in Fig. 3B. Simultaneously, the heatmap illustrates the differential expression of these 34 hub genes across different groups of lung tissue (Fig. 3C). To investigate the biological functions of the 34 hub genes, we conducted GO and KEGG functional enrichment analyses. According to the GO enrichment analysis, the top ten enriched cellular components (CC) included the vesicle lumen, secretory granule lumen, and outer membrane. With respect to biological process (BP) and molecular function (MF), the hub genes were predominantly associated with lyase activity and purine nucleotide metabolic processes. Furthermore, these hub genes were enriched in KEGG pathways such as the sphingolipid metabolism pathway and retinol metabolism pathway (Fig. 3D).
Figure 3. Shared genes between metabolism-related genes, DEGs, and the key modules of WGCNA.
(A) Venn diagram of shared genes; (B) location of CNV alterations of the 34 hub genes on chromosomes; (C) heatmap of the differential expression of 34 hub genes; (D) GO enrichment and KEGG enrichment of 34 hub genes.
Metabolic-related machine learning genes identified via LASSO logistic regression and the RF algorithm
Furthermore, the 34 hub genes were further analyzed via LASSO logistic regression (Figs. 4A and 4B) and the RF algorithm (Figs. 4C and 4D) to identify metabolic-related machine learning genes (MLGs). LASSO revealed 14 identified genes (PDE4B, ARG1, MGAM, CDA, HMOX1, GFPT2, EHHADH, FMO2, AMD1, GUCY1A1, GUCY1B1, UGCG, GPAT3, and FPGT), whereas we selected the 20 most important genes in RF (PDE4B, CDA, HMOX1, SDS, NAMPT, GUCY1B1, ADH1B, SGMS2, GPAM, PFKFB3, PNP, EHHADH, SPHK1, FPGT, NUDT12, GUCY1A1, PTGS2, UPP1, AMD1, and UGCG). Finally, we screened nine overlapping metabolic-related machine learning genes via LASSO regression and the RF algorithm (Fig. 4E). Moreover, to evaluate the diagnostic value of these metabolic-related machine learning genes, we constructed a nomogram and evaluated its diagnostic value via receiver operating characteristic (ROC) curves. Finally, we found that the AUC value of these metabolic-related machine learning genes was greater than 0.8, indicating that these metabolic-related machine learning genes are important biomarkers of IRI (Fig. 4G); at the same time, we displayed the expression of the metabolic-related machine learning genes (Fig. 4F), and the boxplot indicated that PDE4B, CDA, HMOX1, AMD1, and UGCG were highly expressed in IRI lung samples, whereas EHHADH, GUCY1A1, GUCY1B1, and FPGT were significantly downregulated in IRI lung samples.
Figure 4. A predictive model and nomogram based on metabolism-related hub genes for LIRI.
(A–D) Identification of metabolism-related machine learning genes via LASSO regression and the RF algorithm. (E) Overlapping metabolism-related machine learning genes via LASSO regression and the RF algorithm are shown in a Venn diagram. (F–G) Nomogram and ROC curve of LIRI risk on the basis of the nine metabolism-related machine learning genes. (H) Differential expression of metabolism-related machine learning genes in GSE127003 (*** p ≤ 0.001 vs. CIT).
Expression validation and ROC curve of the metabolic-related machine learning genes in the validation dataset
To assess the clinical relevance of these metabolism-related machine learning genes, we selected the PGD-associated dataset GSE8021 for rigorous external validation. We constructed a logistic regression model based on these nine metabolism-related machine learning genes. The model achieved an overall AUC value of 0.939 (Fig. 5A), significantly higher than the AUC values of the individual genes (PDE4B, CDA, HMOX1, EHHADH, AMD1, GUCY1A1, GUCY1B1, UGCG, and FPGT, which were 0.436, 0.526, 0.608, 0.588, 0.597, 0.765, 0.619, 0.559, and 0.610, respectively (Fig. 5E), indicating the model’s strong predictive capability for PGD occurrence. Additionally, we constructed a risk map depicting PGD risk based on nine metabolism-related machine learning genes. Each gene projected upward as a point on the map, with the total score of the nine genes translating to an individual’s disease risk—higher scores corresponding to greater PGD occurrence risk (Fig. 5B). As shown in Fig. 5C, the calibration curve revealed no significant bias between observed and predicted values. We validated the clinical utility of this predictive map through decision curve analysis (DCA). The final DCA results demonstrated that the model exhibits higher overall net benefit within the 0–1 threshold probability range (Fig. 5D).
Figure 5. Validation results of metabolism-related machine learning genes in GSE8021.
(A) ROC curve for logistic regression models. (B) Nomogram of the logistic regression model. (C–D) Calibration curve and decision curve analysis of the metabolism-related machine learning genes. (E) ROC curve of MLGs in GSE8021.
Differential expression of metabolic-related machine learning genes in single-cell RNA sequences
Cells were classified into different types, including epithelial cells, endothelial cells, macrophages, fibroblasts, and T lymphocytes (Fig. 6A). Moreover, we determined the proportions of different cell types in different samples of lung tissue, and by combining these data, we detected a more pronounced change in the proportions of alveolar epithelial type II cells and endothelial cells (Fig. 6B). Next, we observed the distribution of metabolic-related machine learning genes in different cell types. The UMAP data revealed that in reperfusion-induced lung samples, HMOX1 mRNA expression was increased in alveolar epithelial type II cells, AMD1 mRNA expression was increased in alveolar epithelial type II cells and macrophages, and UGCG mRNA expression was increased in alveolar epithelial type II cells and endothelial cells (Figs. 6C–6K).
Figure 6. Metabolism-related machine learning genes in the single-cell RNA sequence dataset GSE220797.
(A) UMAP visualization results after clustering; (B) cell ratio and its variations in pre-transplantation and post-transplantation lung samples; (C–K) distribution of metabolism-related machine learning genes in different groups and in different cell types of lung samples.
Immune cell infiltration
In addition, we analysed immune cell infiltration in the GSE127003 dataset. First, we displayed the proportions of various types of immune cells in different lung samples (Fig. 7A). We subsequently calculated the relative expression levels of different immune cells between the pre-transplantation and post-transplantation lung samples. According to the boxplot, we found significant changes in some types of immune cells, including activated natural killer cells, M1 macrophages, M2 macrophages, and activated mast cells (Fig. 7B). Macrophage polarization plays an important role in inflammatory diseases; M1 polarization often promotes inflammation, whereas M2 polarization has anti-inflammatory and protumour genic biological effects (Luo et al., 2024; Locati, Curtale & Mantovani, 2020). Additionally, researchers have reported that in lung ischemia-reperfusion injury, natural killer cells migrate to the lung in large numbers and manifest cytotoxic effects on epithelial and endothelial cells, thus exacerbating PGD (Calabrese et al., 2020). In conclusion, the immune infiltration results we observed can be confirmed by several previous studies. Next, we investigated the relationships between these cells in the samples and found that some immune cells, such as M1 macrophages and activated memory CD4 T cells, were significantly positively correlated (Fig. 7C). Finally, we examined the correlation between the nine metabolic-related machine learning genes and these immune cells. We found that CDA is associated primarily with M2 macrophages, natural killer cells, eosinophils, and neutrophils. Similarly, PDE4B was positively related to memory B cells.
Figure 7. Immune cell infiltration results from the GSE127003 database.
(A) Proportion of various types of immune cells in GSE127003; (B) boxplot of the relative expression of each immune cell subtype between the pre-transplantation and post-transplantation lung samples, ns: p ≥ 0.05, *: p < 0.05, **: p < 0.01, ***: p < 0.001, ****: p < 0.0001, *****: p < 0.00001; (C) correlated heatmap of the immune cells; (D) heatmap of the correlation between metabolism-related machine learning genes and different immune cells, ns: p ≥ 0.05, *: p < 0.05, **: p < 0.01, ***: p < 0.001.
Validation of metabolic-related machine learning genes in a rat orthotopic left lung transplantation model
To further validate the results of the analysis of public databases and better recapitulate the pathological progression of PGD in lung transplant patients, we successfully established a rat orthotopic left lung transplantation model. As shown in Fig. 8A, the transplanted left lung tissue displayed severe pulmonary oedema and congestion. Figure 8B shows the lung injury scores for different groups. Furthermore, according to the RT–qPCR results, seven metabolic-related machine learning genes were significantly differentially expressed compared with the control groups (Fig. 8C), which is consistent with our previous bioinformatics findings.
Figure 8. The levels of metabolism-related machine learning genes in lung transplantation-induced ischemia-reperfusion injury.
(A) A representative photograph and HE staining of the lungs. (B) Lung injury scores for different groups of lung tissue (n = 3, ## p < 0.05 vs. NC). (C) qPCR results of metabolism-related machine learning genes in the rat orthotopic left lung transplantation model (n = 3, ## p < 0.05 vs. NC). NC, natural control.
Discussion
Our research revealed nine key genes (PDE4B, CDA, HMOX1, EHHADH, AMD1, GUCY1A1, GUCY1B1, UGCG, and FPGT) involved in IRI, seven of which were validated in our animal experiments. This was the first bioinformatic study to characterize the relationships between metabolism-related genes and lung transplantation-induced ischemia-reperfusion injury in human samples.
HMOX1, encoding haem oxygenase-1, mainly participates in the catabolism of haem (Dunn et al., 2014). Hypothesized by researchers to protect the body from damage caused by oxidative stress (Chudy et al., 2024), HMOX1 is regulated by Kelch-like ECH-associated protein 1 (KEAP1) and nuclear factor erythroid 2 related factor 2 (NRF2), and under normal conditions, KEAP1 promotes ubiquitinated degradation of NRF2, whereas massive production of reactive oxygen species modifies KEAP1 and releases and stabilizes NRF2, which facilitates the translocation of NRF2 to the nucleus and thus promotes HMOX1 expression (O’Rourke, Shanley & Dunne, 2024). Haem can regulate the occurrence of inflammation, endothelial cell permeability, and various forms of cell death (Ryter, 2021). For instance, haem can activate microglia through the TLR4/MyD88/P65 signaling pathway, promoting inflammation and apoptosis. Meanwhile, haem has been found to increase inflammatory levels in macrophages, causing them to shift toward a pro-inflammatory phenotype (Pradhan et al., 2024). Multiple studies suggest that HMOX1 is a target for natural small molecule compounds for the treatment of IRI. For example, researchers have reported that corynoline can enhance NRF2/HMOX1 signalling, thus alleviating hepatic ischemia-reperfusion injury (Ge et al., 2024). Yao et al. (2024) reported that isoliquiritigenin improved myocardial ischemia-reperfusion injury by inhibiting ferroptosis through upregulation of the NRF2/HMOX1 pathway. Finally, HMOX1 deficiency in myeloid cells promotes macrophage M1 polarization, which exacerbates hepatic ischemia-reperfusion injury (Zhang et al., 2018).
EHHADH encodes a peroxisomal L-bifunctional enzyme, a type of peroxisomal enzyme that is expressed mainly in the liver and kidney (Pilz et al., 2024). This protein participates in the beta-oxidation of fatty acids (Klootwijk et al., 2014), and Sander et al. reported that this protein helps to maintain medium-chain dicarboxylic acid levels in mouse plasma (Houten et al., 2012). Research indicates that excessive fat accumulation stimulates mitochondrial β-oxidation of free fatty acids, thereby increasing ROS production and promoting inflammation and tissue damage (Jiang, Liu & Li, 2021). In recent years, researchers have proposed that EHHADH is a regulator of autophagy and that knockout of this gene can aggravate tubulointerstitial injury in diabetic mice by increasing the levels of reactive oxygen species and improving pexophagy (Kan et al., 2024). Since no research has explored the role that EHHADH plays in IRI, combined with our bioinformatic analysis and experimental validation, this molecule may become an important marker as well as an important protective factor in lung IRI.
AMD1 encodes the S-adenosylmethionine decarboxylase proenzyme, which takes part in the synthesis of spermine and spermidine (Bian et al., 2021). Many studies have shown that AMD1 is closely correlated with the survival and prognosis of many malignant tumours. For example, AMD1 facilitates prostate cancer progression through the modulation of polyamine metabolism (Zabala-Letona et al., 2017). In addition, Zhu et al. (2021) reported that PM2.5 exposure decreased AMD1 expression and spermidine synthesis, which aggravated neuronal apoptosis. According to previous studies, apoptosis plays an important role in IRI (Zhao et al., 2022), so AMD1 could be an important molecular marker for the prediction of IRI. Furthermore, spermine and spermidine are able to play important roles in certain diseases. First of all, Xu et al. (2024) found that in systemic lupus erythematosus and psoriasis mouse models, spermine binds to JAK1 and inhibits its phosphorylation, thereby reducing the production of inflammatory cytokines and the activation of immune cells (Xu et al., 2024). Secondly, in a T-cell transfer murine model of colitis, researchers observed that spermidine promotes the differentiation of T helper cells 17 into FOXP3+ regulatory T cells, thereby alleviating intestinal lesions (Carriche et al., 2021). Last but not least, spermidine activates autophagy in adipose tissue of obese mice and promotes the upregulation of genes associated with glucose metabolism and lipid metabolism (Ni et al., 2024). In summary, AMD1 and its metabolites play a crucial role in various diseases. In the future, we can explore the effects of AMD1 regulation on polyamine metabolism in relation to LIRI and PGD.
GUCY1A1 encodes the alpha subunit, whereas GUCY1B1 encodes the beta subunit of soluble guanylate cyclase (SGC) (Erdmann et al., 2013); these two genes are also known as GUCY1A3 and GUCY1B3. In mammals, SGC is activated by nitric oxide to produce cyclic guanosine monophosphate, which regulates a range of cellular signalling pathways (Poulos, 2006). Several studies have shown that the activation of SGC can exert a protective effect on the cardiovascular system, and this process is being used as a target for the development of SGC activators (Sandner et al., 2021). For example, a recent Mendelian randomization study revealed a negative association between GUCY1A3 and the risk of coronary artery disease (Le et al., 2025). Additionally, according to a clinical trial, MK-5475, an SGC activator, was able to improve pulmonary hypertension by selectively dilating the pulmonary vasculature (Humbert et al., 2024). SGC has also been studied in IRI, and Wang et al. (2019) reported that GUCY1B1 ameliorated myocardial ischemia-reperfusion injury by activating the PKC/AKT pathway. Overall, according to our bioinformatic analysis and experimental validation, GUCY1A1 and GUCY1B1 expression is reduced in IRI lung tissues, which demonstrates their potential as therapeutic targets.
UGCG, encoding UDP-glucose ceramide glucosyltransferase, is an enzyme that participates in the synthesis of glycosphingolipids (Fan et al., 2025). It has also been shown that UGCG is involved in the regulation of glutamine oxidation and glycolysis in tumour cells (Schömel et al., 2019; Schömel et al., 2020). Previous studies have demonstrated that sphingolipid metabolism is closely linked to inflammation and immunity. For example, interleukin-10 improves inflammatory responses by regulating sphingolipid metabolism (York et al., 2024). Simultaneously, researchers discovered that the absence of the endoplasmic reticulum membrane protein NOGO-B increases sphingosine-1-phosphate levels, thereby helping to improve endothelial cell function and prevent atherosclerosis (Manzo et al., 2024). Moreover, studies in a murine ulcerative colitis model revealed that intestinal epithelial cells secrete sphingosine-1-phosphate to promote the release of serum amyloid A protein 1/3 from macrophages, thereby inducing Th17 cell differentiation (Ma et al., 2024). In addition to sphingolipid metabolism, UGCG itself participates in the regulation of certain diseases. For instance, Additionally, UGCG can promote heart hypertrophy via mitochondrial oxidative stress (Cui et al., 2023). Furthermore, UGCG promotes the activation of autophagy (Jain et al., 2023; Fan et al., 2025). Given that autophagy plays a dual role in IRI, where moderate levels confer protection while excessive activation exacerbates cellular damage (Mokhtari & Badalzadeh, 2022), consistent with these findings, our bioinformatic analysis and experimental validation using a rat lung transplantation model demonstrated that UGCG mRNA levels are significantly elevated in lung tissue following IRI. Consequently, UGCG represents a promising biomarker and therapeutic target for LIRI.
FPGT encodes GDP-L-fucose pyrophosphorylase, a type of bifunctional enzyme. This enzyme catalyses the biosynthesis of GDP-L-fucose from GTP and L-fucose-1-phosphate and participates in a salvage pathway to reutilize L-fucose arising from the turnover of glycoproteins and glycolipids (Pastuszak et al., 1998). Although the role of FPGT in inflammation and oxidative stress has been infrequently reported, our integrated bioinformatic analysis and experimental validation demonstrated that alterations in FPGT mRNA expression levels may serve as potential biomarkers for IRI.
A key strength of this study lies in the identification of critical metabolism-related genes that are consistently differentially expressed in lung transplantation-induced IRI, which was achieved through comprehensive bioinformatics analyses. These findings were further validated in animal models, highlighting their potential relevance to metabolic dysregulation in IRI. As a result, the identified metabolic-related machine learning genes may serve as promising targets for future investigations into the metabolic alterations associated with this disease. However, several limitations of this study need to be addressed. First, the construction and validation of the prognostic model were based on retrospective data from the GEO database, and the sample size of the cohort was relatively small. Thus, a prospective study with a large sample size is needed to identify its clinical application. Second, clinical information, such as patients’ lung function data and the partial pressure of oxygenation in the arterial blood, was not complete in the datasets; thus, the significance of the prognostic model was restricted. Finally, the mechanisms of the metabolic-related machine learning genes need to be further explored in basic medical experiments. In summary, metabolic abnormalities play an important role in ischemia-reperfusion injury, and we can use these abnormalities to explore useful biomarkers or therapeutic targets.
Conclusions
This study screened and identified nine metabolism-related genes through analysis of public databases. A subset of these genes was validated in a rat lung transplantation model, thereby identifying potential biomarkers and therapeutic targets for the occurrence of PGD in lung transplant patients. Our study developed a novel risk score prognostic model based on nine metabolism-related genes, providing a new approach for predicting the prognosis of lung transplant patients and potential therapeutic targets for PGD. Although this study represents preliminary work, it establishes a rigorous methodological framework for future research in this critical field, with the potential to inform clinical decision-making and enhance patient outcomes.
Supplemental Information
Funding Statement
This research was funded by the Noncommunicable Chronic Diseases-National Science and Technology Major Project (2023ZD0505900), the National Key Research and Development Program of China (2023YFC2507100). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Contributor Information
Dong Wei, Email: weidongmd@126.com.
Jingyu Chen, Email: chenjy@wuxiph.com, 2023122356@stu.njmu.edu.cn.
Additional Information and Declarations
Competing Interests
The authors declare there are no competing interests.
Author Contributions
Longfei Zhu conceived and designed the experiments, performed the experiments, analyzed the data, prepared figures and/or tables, and approved the final draft.
Jiaqi Ding conceived and designed the experiments, performed the experiments, analyzed the data, prepared figures and/or tables, and approved the final draft.
Dong Wei analyzed the data, authored or reviewed drafts of the article, and approved the final draft.
Jingyu Chen analyzed the data, authored or reviewed drafts of the article, and approved the final draft.
Animal Ethics
The following information was supplied relating to ethical approvals (i.e., approving body and any reference numbers):
The study was conducted in accordance with the Declaration of Helsinki, and the animal study protocol was approved by the Institutional Ethics Committee of Nanjing Medical University Affricated Wuxi People’s Hospital (DL2024023, 25 November 2024).
Data Availability
The following information was supplied regarding data availability:
The raw measurements are available in the Supplementary Files and at GEO: GSE127003, GSE145989, GSE220797.
References
- Adegunsoye et al. (2017).Adegunsoye A, Strek ME, Garrity E, Guzy R, Bag R. Comprehensive care of the lung transplant patient. Chest. 2017;152:150–164. doi: 10.1016/j.chest.2016.10.001. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Bian et al. (2021).Bian X, Shi D, Xing K, Zhou H, Lu L, Yu D, Wu W. AMD1 upregulates hepatocellular carcinoma cells stemness by FTO mediated mRNA demethylation. Clinical and Translational Medicine. 2021;11:e352. doi: 10.1002/ctm2.352. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Calabrese et al. (2020).Calabrese DR, Aminian E, Mallavia B, Liu F, Cleary SJ, Aguilar OA, Wang P, Singer JP, Hays SR, Golden JA, Kukreja J, Dugger D, Nakamura M, Lanier LL, Looney MR, Greenland JR. Natural killer cells activated through NKG2D mediate lung ischemia-reperfusion injury. The Journal of Clinical Investigation. 2020;131:e137047. doi: 10.1172/JCI137047. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Capuzzimati, Hough & Liu (2022).Capuzzimati M, Hough O, Liu M. Cell death and ischemia-reperfusion injury in lung transplantation. The Journal of Heart and Lung Transplantation. 2022;41:1003–1013. doi: 10.1016/j.healun.2022.05.013. [DOI] [PubMed] [Google Scholar]
- Carriche et al. (2021).Carriche GM, Almeida L, Stüve P, Velasquez L, Dhillon-LaBrooy A, Roy U, Lindenberg M, Strowig T, Plaza-Sirvent C, Schmitz I, Lochner M, Simon AK, Sparwasser T. Regulating T-cell differentiation through the polyamine spermidine. The Journal of Allergy and Clinical Immunology. 2021;147:335–348.e11. doi: 10.1016/j.jaci.2020.04.037. [DOI] [PubMed] [Google Scholar]
- Catelli et al. (2024).Catelli C, Faccioli E, Silvestrin S, Lorenzoni G, Luzzi L, Bennett D, Schiavon M, Campisi A, Bargagli E, Dell’Amore A, Rea F. Lung transplantation in patients with previous or unknown oncological disease: evaluation of short- and long-term outcomes. Cancer. 2024;16:538. doi: 10.3390/cancers16030538. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Chudy et al. (2024).Chudy P, Kochan J, Wawro M, Nguyen P, Gorczyca M, Varanko A, Retka A, Ghadei SS, Napieralska E, Grochot-Przęczek A, Szade K, Berendes L-S, Park J, Sokołowski G, Yu Q, Józkowicz A, Nowak WN, Krzeptowski W. Heme oxygenase-1 protects cells from replication stress. Redox Biology. 2024;75:103247. doi: 10.1016/j.redox.2024.103247. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Cui et al. (2023).Cui S, Zhang X, Li Y, Hu S, Wu B, Fang Z, Gao J, Li M, Wu H, Tao B, Xia H, Xu L. UGCG modulates heart hypertrophy through B4GalT5-mediated mitochondrial oxidative stress and the ERK signaling pathway. Cellular & Molecular Biology Letters. 2023;28:71. doi: 10.1186/s11658-023-00484-3. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Daud et al. (2007).Daud SA, Yusen RD, Meyers BF, Chakinala MM, Walter MJ, Aloush AA, Patterson GA, Trulock EP, Hachem RR. Impact of immediate primary lung allograft dysfunction on bronchiolitis obliterans syndrome. American Journal of Respiratory and Critical Care Medicine. 2007;175:507–513. doi: 10.1164/rccm.200608-1079OC. [DOI] [PubMed] [Google Scholar]
- Dunn et al. (2014).Dunn LL, Midwinter RG, Ni J, Hamid HA, Parish CR, Stocker R. New insights into intracellular locations and functions of heme oxygenase-1. Antioxidants & Redox Signaling. 2014;20:1723–1742. doi: 10.1089/ars.2013.5675. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Erdmann et al. (2013).Erdmann J, Stark K, Esslinger UB, Rumpf PM, Koesling D, De Wit C, Kaiser FJ, Braunholz D, Medack A, Fischer M, Zimmermann ME, Tennstedt S, Graf E, Eck S, Aherrahrou Z, Nahrstaedt J, Willenborg C, Bruse P, Brænne I, Nöthen MM, Hofmann P, Braund PS, Mergia E, Reinhard W, Burgdorf C, Schreiber S, Balmforth AJ, Hall AS, Bertram L, Steinhagen-Thiessen E, Li S-C, März W, Reilly M, Kathiresan S, McPherson R, Walter U, CARDIoGRAM. Ott J, Samani NJ, Strom TM, Meitinger T, Hengstenberg C, Schunkert H. Dysfunctional nitric oxide signalling increases risk of myocardial infarction. Nature. 2013;504:432–436. doi: 10.1038/nature12722. [DOI] [PubMed] [Google Scholar]
- Fan et al. (2025).Fan W, Yao C, Ma Y, Wang H, Liu P, Zhang Z, Chu B, Yang G, Wang M. Inhibiting UGCG prevents PRV infection by decreasing lysosome-associated autophage. International Journal of Biological Macromolecules. 2025;285:138303. doi: 10.1016/j.ijbiomac.2024.138303. [DOI] [PubMed] [Google Scholar]
- Feng et al. (2025).Feng Y, Peng Y, Hou J, Wang Z, Lai J, Xiong T, Shi J, Wang Y, Yim WY, Chen Y, Dong N. FABP3 and FABP4 promote lipid peroxidation injury during static cold storage of donor heart: insights from multi-omics and therapeutic targeting. Journal of Molecular and Cellular Cardiology Plus. 2025;14:100497. doi: 10.1016/j.jmccpl.2025.100497. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Ge et al. (2024).Ge X, Gu Y, Wang W, Guo W, Wang P, Du P. Corynoline alleviates hepatic ischemia-reperfusion injury by inhibiting NLRP3 inflammasome activation through enhancing Nrf2/HO-1 signaling. Inflammation Research. 2024;73:2069–2085. doi: 10.1007/s00011-024-01949-7. [DOI] [PubMed] [Google Scholar]
- Hao et al. (2021).Hao Y, Hao S, Andersen-Nissen E, Mauck WM, Zheng S, Butler A, Lee MJ, Wilk AJ, Darby C, Zager M, Hoffman P, Stoeckius M, Papalexi E, Mimitou EP, Jain J, Srivastava A, Stuart T, Fleming LM, Yeung B, Rogers AJ, McElrath JM, Blish CA, Gottardo R, Smibert P, Satija R. Integrated analysis of multimodal single-cell data. Cell. 2021;184:3573–3587. doi: 10.1016/j.cell.2021.04.048. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Houten et al. (2012).Houten SM, Denis S, Argmann CA, Jia Y, Ferdinandusse S, Reddy JK, Wanders RJA. Peroxisomal L-bifunctional enzyme (Ehhadh) is essential for the production of medium-chain dicarboxylic acids. Journal of Lipid Research. 2012;53:1296–1303. doi: 10.1194/jlr.M024463. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Hu et al. (2023).Hu C, Li T, Xu Y, Zhang X, Li F, Bai J, Chen J, Jiang W, Yang K, Ou Q, Li X, Wang P, Zhang Y. CellMarker 2.0: an updated database of manually curated cell markers in human/mouse and web tools based on scRNA-seq data. Nucleic Acids Research. 2023;51:D870–D876. doi: 10.1093/nar/gkac947. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Huang et al. (2025).Huang X, Chen L, Liu L, Zhou Y, Zhou H, Zhang Y. Five-aminolevulinic acid as a potential biomarker for renal insufficiency after heart transplantation. Clinical Transplantation. 2025;39:e70284. doi: 10.1111/ctr.70284. [DOI] [PubMed] [Google Scholar]
- Humbert et al. (2024).Humbert M, Hassoun PM, Chin KM, Bortman G, Patel MJ, La Rosa C, Fu W, Loureiro MJ, Hoeper MM. MK-5475, an inhaled soluble guanylate cyclase stimulator, for treatment of pulmonary arterial hypertension: the INSIGNIA-PAH study. The European Respiratory Journal. 2024;64:2401110. doi: 10.1183/13993003.01110-2024. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Jain et al. (2023).Jain V, Harper SL, Versace AM, Fingerman D, Brown GS, Bhardwaj M, Crissey MAS, Goldman AR, Ruthel G, Liu Q, Zivkovic A, Stark H, Herlyn M, Gimotty PA, Speicher DW, Amaravadi RK. Targeting UGCG overcomes resistance to lysosomal autophagy inhibition. Cancer Discovery. 2023;13:454–473. doi: 10.1158/2159-8290.CD-22-0535. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Jeon et al. (2024).Jeon JE, Rajapaksa Y, Keshavjee S, Liu M. Applications of transcriptomics in ischemia reperfusion research in lung transplantation. The Journal of Heart and Lung Transplantation. 2024;43:1501–1513. doi: 10.1016/j.healun.2024.03.006. [DOI] [PubMed] [Google Scholar]
- Jiang, Liu & Li (2021).Jiang S, Liu H, Li C. Dietary regulation of oxidative stress in chronic metabolic diseases. Foods. 2021;10:1854. doi: 10.3390/foods10081854. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Kan et al. (2024).Kan S, Hou Q, Shi J, Zhang M, Xu F, Liu Z, Jiang S. EHHADH deficiency regulates pexophagy and accelerates tubulointerstitial injury in diabetic kidney disease. Cell Death Discovery. 2024;10:289. doi: 10.1038/s41420-024-02066-4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Klootwijk et al. (2014).Klootwijk ED, Reichold M, Helip-Wooley A, Tolaymat A, Broeker C, Robinette SL, Reinders J, Peindl D, Renner K, Eberhart K, Assmann N, Oefner PJ, Dettmer K, Sterner C, Schroeder J, Zorger N, Witzgall R, Reinhold SW, Stanescu HC, Bockenhauer D, Jaureguiberry G, Courtneidge H, Hall AM, Wijeyesekera AD, Holmes E, Nicholson JK, O’Brien K, Bernardini I, Krasnewich DM, Arcos-Burgos M, Izumi Y, Nonoguchi H, Jia Y, Reddy JK, Ilyas M, Unwin RJ, Gahl WA, Warth R, Kleta R. Mistargeting of peroxisomal EHHADH and inherited renal Fanconi’s syndrome. The New England Journal of Medicine. 2014;370:129–138. doi: 10.1056/NEJMoa1307581. [DOI] [PubMed] [Google Scholar]
- Korsunsky et al. (2019).Korsunsky I, Millard N, Fan J, Slowikowski K, Zhang F, Wei K, Baglaenko Y, Brenner M, Loh P, Raychaudhuri S. Fast, sensitive and accurate integration of single-cell data with harmony. Nature Methods. 2019;16:1289–1296. doi: 10.1038/s41592-019-0619-0. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Krassowski et al. (2020).Krassowski M, Das V, Sahu SK, Misra BB. State of the field in multi-omics research: from computational needs to data mining and sharing. Frontiers in Genetics. 2020;11:610798. doi: 10.3389/fgene.2020.610798. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Langfelder & Horvath (2008).Langfelder P, Horvath S. WGCNA: an R package for weighted correlation network analysis. BMC Bioinformatics. 2008;9:559. doi: 10.1186/1471-2105-9-559. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Le et al. (2025).Le NN, Tran TQB, McClure J, Gill D, Padmanabhan S. Emerging antihypertensive therapies and cardiovascular, kidney, and metabolic outcomes: a Mendelian randomization study. European Heart Journal. Cardiovascular Pharmacotherapy. 2025;11:264–274. doi: 10.1093/ehjcvp/pvaf015. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Li et al. (2021).Li L, Xiong F, Wang Y, Zhang S, Gong Z, Li X, He Y, Shi L, Wang F, Liao Q, Xiang B, Zhou M, Li X, Li Y, Li G, Zeng Z, Xiong W, Guo C. What are the applications of single-cell RNA sequencing in cancer research: a systematic review. Journal of Experimental & Clinical Cancer Research. 2021;40:163. doi: 10.1186/s13046-021-01955-1. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Liang et al. (2025).Liang L, Hsin MK, Zhao Y, Wang A, Machuca T, Yeung J, Cypel M, Keshavjee S, Liu M. Metabolic changes during cold ischemic preservation and reperfusion in porcine lung transplants. American Journal of Transplantation. 2025;25:2090–2103. doi: 10.1016/j.ajt.2025.05.021. [DOI] [PubMed] [Google Scholar]
- Liberzon et al. (2015).Liberzon A, Birger C, Thorvaldsdóttir H, Ghandi M, Mesirov JP, Tamayo P. The molecular signatures database (MSigDB) hallmark gene set collection. Cell Systems. 2015;1:417–425. doi: 10.1016/j.cels.2015.12.004. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Locati, Curtale & Mantovani (2020).Locati M, Curtale G, Mantovani A. Diversity, mechanisms, and significance of macrophage plasticity. Annual Review of Pathology. 2020;15:123–147. doi: 10.1146/annurev-pathmechdis-012418-012718. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Luo et al. (2024).Luo M, Zhao F, Cheng H, Su M, Wang Y. Macrophage polarization: an important role in inflammatory diseases. Frontiers in Immunology. 2024;15:1352946. doi: 10.3389/fimmu.2024.1352946. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Ma et al. (2024).Ma Y, Zhang X, Xuan B, Li D, Yin N, Ning L, Zhou Y-L, Yan Y, Tong T, Zhu X, Huang X, Hu M, Wang Z, Cui Z, Li H, Wang J, Fang J-Y, Liu R, Chen H, Hong J. Disruption of CerS6-mediated sphingolipid metabolism by FTO deficiency aggravates ulcerative colitis. Gut. 2024;73:268–281. doi: 10.1136/gutjnl-2023-330009. [DOI] [PubMed] [Google Scholar]
- Manzo et al. (2024).Manzo OL, Nour J, Sasset L, Marino A, Rubinelli L, Palikhe S, Smimmo M, Hu Y, Bucci MR, Borczuk A, Elemento O, Freed JK, Norata GD, Di Lorenzo A. Rewiring Endothelial Sphingolipid Metabolism to Favor S1P Over Ceramide Protects From Coronary Atherosclerosis. Circulation Research. 2024;134:990–1005. doi: 10.1161/CIRCRESAHA.123.323826. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Mokhtari & Badalzadeh (2022).Mokhtari B, Badalzadeh R. Protective and deleterious effects of autophagy in the setting of myocardial ischemia/reperfusion injury: an overview. Molecular Biology Reports. 2022;49:11081–11099. doi: 10.1007/s11033-022-07837-9. [DOI] [PubMed] [Google Scholar]
- Ni et al. (2024).Ni Y, Zheng L, Zhang L, Li J, Pan Y, Du H, Wang Z, Fu Z. Spermidine activates adipose tissue thermogenesis through autophagy and fibroblast growth factor 21. The Journal of Nutritional Biochemistry. 2024;125:109569. doi: 10.1016/j.jnutbio.2024.109569. [DOI] [PubMed] [Google Scholar]
- O’Rourke, Shanley & Dunne (2024).O’Rourke SA, Shanley LC, Dunne A. The Nrf2-HO-1 system and inflammaging. Frontiers in Immunology. 2024;15:1457010. doi: 10.3389/fimmu.2024.1457010. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Pastuszak et al. (1998).Pastuszak I, Ketchum C, Hermanson G, Sjoberg EJ, Drake R, Elbein AD. GDP-L-fucose pyrophosphorylase, purification, cDNA cloning, and properties of the enzyme. The Journal of Biological Chemistry. 1998;273:30165–30174. doi: 10.1074/jbc.273.46.30165. [DOI] [PubMed] [Google Scholar]
- Pilz et al. (2024).Pilz JF, Klein M, Neumann-Haefelin E, Ganner A. VHL-dependence of EHHADH expression in a human renal cell carcinoma cell line. Journal of Kidney Cancer and VHL. 2024;11:12–18. doi: 10.15586/jkcvhl.v11i1.322. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Porteous & Lee (2017).Porteous MK, Lee JC. Primary graft dysfunction after lung transplantation. Clinics in Chest Medicine. 2017;38:641–654. doi: 10.1016/j.ccm.2017.07.005. [DOI] [PubMed] [Google Scholar]
- Poulos (2006).Poulos TL. Soluble guanylate cyclase. Current Opinion in Structural Biology. 2006;16:736–743. doi: 10.1016/j.sbi.2006.09.006. [DOI] [PubMed] [Google Scholar]
- Pradhan et al. (2024).Pradhan P, Vijayan V, Liu B, Martinez-Delgado B, Matamala N, Nikolin C, Greite R, De Luca DS, Janciauskiene S, Motterlini R, Foresti R, Immenschuh S. Distinct metabolic responses to heme in inflammatory human and mouse macrophages—role of nitric oxide. Redox Biology. 2024;73:103191. doi: 10.1016/j.redox.2024.103191. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Robin et al. (2011).Robin X, Turck N, Hainard A, Tiberti N, Lisacek F, Sanchez J-C, Müller M. pROC: an open-source package for R and S+ to analyze and compare ROC curves. BMC Bioinformatics. 2011;12:77. doi: 10.1186/1471-2105-12-77. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Ryter (2021).Ryter SW. Significance of heme and heme degradation in the pathogenesis of acute lung and inflammatory disorders. International Journal of Molecular Sciences. 2021;22:5509. doi: 10.3390/ijms22115509. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Sandner et al. (2021).Sandner P, Zimmer DP, Milne GT, Follmann M, Hobbs A, Stasch J-P. Soluble guanylate cyclase stimulators and activators. HandBook of Experimental Pharmacology. 2021;264:355–394. doi: 10.1007/164_2018_197. [DOI] [PubMed] [Google Scholar]
- Schömel et al. (2020).Schömel N, Gruber L, Alexopoulos SJ, Trautmann S, Olzomer EM, Byrne FL, Hoehn KL, Gurke R, Thomas D, Ferreirós N, Geisslinger G, Wegner M-S. UGCG overexpression leads to increased glycolysis and increased oxidative phosphorylation of breast cancer cells. Scientific Reports. 2020;10:8182. doi: 10.1038/s41598-020-65182-y. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Schömel et al. (2019).Schömel N, Hancock SE, Gruber L, Olzomer EM, Byrne FL, Shah D, Hoehn KL, Turner N, Grösch S, Geisslinger G, Wegner M-S. UGCG influences glutamine metabolism of breast cancer cells. Scientific Reports. 2019;9:15665. doi: 10.1038/s41598-019-52169-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Singh et al. (2023).Singh TP, Cherikh WS, Hsich E, Lewis A, Perch M, Kian S, Hayes D, Potena L, Stehlik J, Zuckermann A, Cogswell R, International S for H and LT Graft survival in primary thoracic organ transplant recipients: a special report from the International Thoracic Organ Transplant Registry of the International Society for Heart and Lung Transplantation. The Journal of Heart and Lung Transplantation. 2023;42:1321–1333. doi: 10.1016/j.healun.2023.07.017. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Snell et al. (2017).Snell GI, Yusen RD, Weill D, Strueber M, Garrity E, Reed A, Pelaez A, Whelan TP, Perch M, Bag R, Budev M, Corris PA, Crespo MM, Witt C, Cantu E, Christie JD. Report of the ISHLT working group on primary lung graft dysfunction, part I: definition and grading-A 2016 consensus group statement of the international society for heart and lung transplantation. The Journal of Heart and Lung Transplantation. 2017;36:1097–1103. doi: 10.1016/j.healun.2017.07.021. [DOI] [PubMed] [Google Scholar]
- Van Slambrouck et al. (2025).Van Slambrouck J, Loopmans S, Prisciandaro E, Barbarossa A, Kortleven P, Feys S, Vandervelde CM, Jin X, Cenik I, Moermans K, Fieuws S, Provoost A-L, Willems A, De Leyn P, Van Veer H, Depypere L, Jansen Y, Pirenne J, Neyrinck A, Weynand B, Vanaudenaerde B, Carmeliet G, Vos R, Van Raemdonck D, Ghesquière B, Van Weyenbergh J, Ceulemans LJ. The effect of rewarming ischemia on tissue transcriptome and metabolome signatures: a clinical observational study in lung transplantation. The Journal of Heart and Lung Transplantation. 2025;44:437–447. doi: 10.1016/j.healun.2024.10.020. [DOI] [PubMed] [Google Scholar]
- Vandereyken et al. (2023).Vandereyken K, Sifrim A, Thienpont B, Voet T. Methods and applications for single-cell and spatial multi-omics. Nature Reviews Genetics. 2023;24:494–515. doi: 10.1038/s41576-023-00580-2. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Ve & Ak (2016).Ve L, Ak S. Mechanisms of lung ischemia-reperfusion injury. Current Opinion in Organ Transplantation. 2016;21(3):246–252. doi: 10.1097/MOT.0000000000000304. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Wang et al. (2025).Wang A, Ali A, Baciu C, Bellissimo C, Siebiger G, Yamanashi K, Montagne J, Garza G, Goligher E, Keshavjee S, Liu M, Cypel M. Metabolomic studies reveal an organ-protective hibernation state in donor lungs preserved at 10 °C. The Journal of Thoracic and Cardiovascular Surgery. 2025;169:796–810. doi: 10.1016/j.jtcvs.2024.08.015. [DOI] [PubMed] [Google Scholar]
- Wang et al. (2019).Wang X, Du W, Li M, Zhang Y, Li H, Sun K, Liu J, Dong P, Meng X, Yi W, Yang L, Zhao R, Hu J. The β subunit of soluble guanylyl cyclase GUCY1B3 exerts cardioprotective effects against ischemic injury via the PKCɛ/Akt pathway. Journal of Cellular Biochemistry. 2019;120:3071–3081. doi: 10.1002/jcb.27479. [DOI] [PubMed] [Google Scholar]
- Wong et al. (2024).Wong A, Duong A, Wilson G, Yeung J, MacParland S, Han H, Cypel M, Keshavjee S, Liu M. Ischemia-reperfusion responses in human lung transplants at the single-cell resolution. American Journal of Transplantation. 2024;24:2199–2211. doi: 10.1016/j.ajt.2024.08.019. [DOI] [PubMed] [Google Scholar]
- Wong & Hsin (2024).Wong KHM, Hsin KYM. Primary graft dysfunction in lung transplantation: still a thorn in the side of lung transplant. Journal of Thoracic Disease. 2024;16:1–5. doi: 10.21037/jtd-23-1618. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Wong et al. (2020).Wong A, Zamel R, Yeung J, Bader GD, Dos Santos CC, Bai X, Wang Y, Keshavjee S, Liu M. Potential therapeutic targets for lung repair during human ex vivo lung perfusion. The European Respiratory Journal. 2020;55:1902222. doi: 10.1183/13993003.02222-2019. [DOI] [PubMed] [Google Scholar]
- Wu et al. (2021).Wu T, Hu E, Xu S, Chen M, Guo P, Dai Z, Feng T, Zhou L, Tang W, Zhan L, Fu X, Liu S, Bo X, Yu G. clusterProfiler 4.0: a universal enrichment tool for interpreting omics data. Innovation. 2021;2:100141. doi: 10.1016/j.xinn.2021.100141. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Xu et al. (2024).Xu H, Zhang X, Wang X, Li B, Yu H, Quan Y, Jiang Y, You Y, Wang Y, Wen M, Liu J, Wang M, Zhang B, Li Y, Zhang X, Lu Q, Yu C-Y, Cao X. Cellular spermine targets JAK signaling to restrain cytokine-mediated autoimmunity. Immunity. 2024;57:1796–1811.e8. doi: 10.1016/j.immuni.2024.05.025. [DOI] [PubMed] [Google Scholar]
- Yang et al. (2025).Yang X, Hong S, Yan T, Liu M, Liu M, Zhao J, Yue B, Wu D, Shao J, Huang M, Chen J. MiR-146a engineered extracellular vesicles derived from mesenchymal stromal cells more potently attenuate ischaemia-reperfusion injury in lung transplantation. Clinical and Translational Medicine. 2025;15:e70298. doi: 10.1002/ctm2.70298. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Yao et al. (2024).Yao D, Bao L, Wang S, Tan M, Xu Y, Wu T, Zhang Z, Gong K. Isoliquiritigenin alleviates myocardial ischemia-reperfusion injury by regulating the Nrf2/HO-1/SLC7a11/GPX4 axis in mice. Free Radical Biology and Medicine. 2024;221:1–12. doi: 10.1016/j.freeradbiomed.2024.05.012. [DOI] [PubMed] [Google Scholar]
- York et al. (2024).York AG, Skadow MH, Oh J, Qu R, Zhou QD, Hsieh W-Y, Mowel WK, Brewer JR, Kaffe E, Williams KJ, Kluger Y, Smale ST, Crawford JM, Bensinger SJ, Flavell RA. IL-10 constrains sphingolipid metabolism to limit inflammation. Nature. 2024;627:628–635. doi: 10.1038/s41586-024-07098-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Zabala-Letona et al. (2017).Zabala-Letona A, Arruabarrena-Aristorena A, Martín-Martín N, Fernandez-Ruiz S, Sutherland JD, Clasquin M, Tomas-Cortazar J, Jimenez J, Torres I, Quang P, Ximenez-Embun P, Bago R, Ugalde-Olano A, Loizaga-Iriarte A, Lacasa-Viscasillas I, Unda M, Torrano V, Cabrera D, Van Liempd SM, Cendon Y, Castro E, Murray S, Revandkar A, Alimonti A, Zhang Y, Barnett A, Lein G, Pirman D, Cortazar AR, Arreal L, Prudkin L, Astobiza I, Valcarcel-Jimenez L, Zuñiga García P, Fernandez-Dominguez I, Piva M, Caro-Maldonado A, Sánchez-Mosquera P, Castillo-Martín M, Serra V, Beraza N, Gentilella A, Thomas G, Azkargorta M, Elortza F, Farràs R, Olmos D, Efeyan A, Anguita J, Muñoz J, Falcón-Pérez JM, Barrio R, Macarulla T, Mato JM, Martinez-Chantar ML, Cordon-Cardo C, Aransay AM, Marks K, Baselga J, Tabernero J, Nuciforo P, Manning BD, Marjon K, Carracedo A. mTORC1-dependent AMD1 regulation sustains polyamine metabolism in prostate cancer. Nature. 2017;547:109–113. doi: 10.1038/nature22964. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Zhang et al. (2018).Zhang M, Nakamura K, Kageyama S, Lawal AO, Gong KW, Bhetraratana M, Fujii T, Sulaiman D, Hirao H, Bolisetty S, Kupiec-Weglinski JW, Araujo JA. Myeloid HO-1 modulates macrophage polarization and protects against ischemia-reperfusion injury. JCI Insight. 2018;3:e120596. doi: 10.1172/jci.insight.120596. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Zhao et al. (2022).Zhao J, Wang G, Han K, Wang Y, Wang L, Gao J, Zhao S, Wang G, Chen S, Luo A, Wu J, Wang G. Mitochondrial PKM2 deacetylation by procyanidin B2-induced SIRT3 upregulation alleviates lung ischemia/reperfusion injury. Cell Death & Disease. 2022;13:594. doi: 10.1038/s41419-022-05051-w. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Zhou et al. (2025).Zhou Y, Tong Z, Zhu X, Wu C, Zhou Y, Dong Z. Deciphering the cellular and molecular landscape of pulmonary fibrosis through single-cell sequencing and machine learning. Journal of Translational Medicine. 2025;23:3. doi: 10.1186/s12967-024-06031-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Zhu et al. (2021).Zhu X, Shou Y, Ji X, Hu Y, Wang H. S-adenosylmethionine decarboxylase 1 and its related spermidine synthesis mediate PM2.5 exposure-induced neuronal apoptosis. Ecotoxicology and Environmental Safety. 2021;224:112678. doi: 10.1016/j.ecoenv.2021.112678. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
The following information was supplied regarding data availability:
The raw measurements are available in the Supplementary Files and at GEO: GSE127003, GSE145989, GSE220797.