Machine learning revealed novel ferroptosis-related genes and construction ceRNA network in dermal lymphatic endothelial cells of diabetic foot ulcer

doi:10.21203/rs.3.rs-3095868/v1

Download PDF

Research Article

Machine learning revealed novel ferroptosis-related genes and construction ceRNA network in dermal lymphatic endothelial cells of diabetic foot ulcer

https://doi.org/10.21203/rs.3.rs-3095868/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this latest preprint version

Background

Diabetic foot ulcer (DFU) is a common chronic and serious complication that impairs patients' quality of life. The relationship between ferroptosis and complications of diabetes has attracted much attention in recent years. Here, this study aims to apply the gene expression profile of dermal lymphatic endothelial cells (DLECs) to build a ceRNA network and explore potential ferroptosis-related biomarkers and pathways related to the molecular mechanism of DFU.

Methods

The GSE38396 dataset from the Gene Expression Omnibus (GEO) collection was utilized to analyze differentially expressed genes (DEGs) in DLECs of DFU. The protein-protein interaction (PPI) network and enrichment analysis of DEGs were carried out. Subsequently, we performed a comprehensive analysis of hub genes identified from the PPI. To yield the key ferroptosis-related genes strongly associated with DLECs of DFU, we integrated multiple datasets and the least absolute shrinkage and selection operator (LASSO) model, which was validated with external datasets (GSE147890) via receiver operating characteristic (ROC) curves. Meanwhile, GSE147890 and GSE29221 were utilized as external datasets to validate the efficiency of key genes as potential biomarkers for DFU.

Results

The 149 DEGs in DLECs of DFU were obtained using the GSE38396 dataset. Gene Ontology (GO) analysis showed that the collagen-containing extracellular matrix was primarily enriched. The HIF-1 signaling pathway was considered the key pathway by the Kyoto Encyclopedia of Genes and Genomes (KEGG) analysis. We obtained 12 hub genes from the subnetwork in the PPI network and identified EGFR as a key ferroptosis-related gene by the LASSO model, which had a high AUC value (0.76). In addition, two external datasets validated EGFR with high efficiency (GSE147890: 0.67, GSE29221: 0.72). Ultimately, we constructed a ceRNA network consisting of 5 lncRNAs, 2 miRNAs, and 1 mRNA around EGFR.

Conclusions

As a key ferroptosis-related gene related to DLECs of DFU, EGFR may be regulated by upstream lncRNA, which in turn affects the activity of the HIF-1 pathway and affects the occurrence and development of DFU. Thus, the results of this study can provide a certain direction and basis for follow-up studies of DFU and provide new insights into the prevention and treatment of DFU.

diabetic foot ulcer

ferroptosis

lncRNA-miRNA-mRNA

biomarkers

LASSO

Diabetes mellitus (DM), as a prevalent chronic metabolic illness, affects roughly 8% of the world's population and is still rising (1). DFU is one of the most prevalent consequences of diabetes, and roughly one-fifth of patients require some kind of amputation, putting a significant financial strain on families and society (1). However, there is no widely available, effective biomarkers for DFU.

The pathophysiology of DFU is multifactorial, including peripheral neuropathy, peripheral vascular disease, repetitive trauma, and overlapping foot infections. According to existing studies, up to 85% of DFU can be attributed to potential peripheral neuropathy that changes foot pressure (2). Continuous hyperglycemia leads to the complex formation of advanced glycation end products and cytokines, which further induces oxidative stress in nerve cells and leads to motor nerve, autonomic nerve, and sensory nerve lesions, so it is also known as nerve foot ulcers (3). At the same time, hyperglycemia and its related changes in glucose metabolism can lead to endothelial injury, hyperlipidemia, increased platelet viscosity and activity, and develop into atherosclerosis over time (4). Foot perfusion below levels sufficient to maintain skin function can lead to ischemic ulcers or gangrene.

The skin lymphatic micro vessel density in patients with type 2 diabetes was increased, and there were DEGs between type 2 diabetes and non-diabetic LECs based on BMonika Haemmerle et al.'s transcriptome study of skin lymphatic vessels in type 2 diabetes (5).This suggests that the occurrence of DFU is possibly related to variations of DLECs in type 2 diabetic patients. Ferroptosis is a new iron-dependent form of regulated death that is morphologically, biochemically and genetically distinct from apoptosis, necrosis and autophagy(6). Recently, it was shown that the development and progression of diabetic complications are closely linked to ferroptosis and ferritinophagy due to disruptions in cellular metabolic pathways(7, 8). However, it is not known what the exact mechanism of action of ferroptosis is in DLECs of DFU.

Bioinformatics, one of the most powerful tools in the field of biomedical research, contributes to the researcher's better understanding of the molecular mechanisms of DLECs in DFU (9). In the current study, we identified DEGs in DLECs of DFU via the GEO database using R software. Subsequently, utilizing machine learning methods, a robust prediction model for DFU was constructed and validated by employing external dataset. Finally, we determined the key ferroptosis-related gene verified by external dataset, and established ceRNA network of lncRNA, which is a collection of regulatory factors to activate or inhibit gene expression, as well as engage in biological processes such as cell proliferation, apoptosis, and organ development in human disorders (10). A flow diagram of the study is shown in Fig. 1.

Microarray dataset information and processing

One dataset in DLECs of DFU and three datasets of DFU were downloaded from the GEO (https://www.ncbi.nlm.nih.gov/geo/) (11), which were all from Homo sapiens (Table 1). The raw expression data was preprocessed using the Normalize Between Arrays function of limma package. The 259 genes associated with ferroptosis were available based on the FerrDb database (Table S5), which is the world's first database of ferroptosis regulators and markers and ferroptosis-disease associations.

Table 1

Details of four datasets.
Dataset	Platform	Count	Diabetic	Control
GSE38396	GPL570 [HG-U133_Plus_2] Affymetrix Human Genome U133 Plus 2.0 Array	8	4	4
GSE80178	GPL16686 [HuGene-2_0-st] Affymetrix Human Gene 2.0 ST Array	12	6	Diabetic Foot Skin: 3
GSE80178	GPL16686 [HuGene-2_0-st] Affymetrix Human Gene 2.0 ST Array	12	6	non-Diabetic Foot Skin: 3
GSE29221	GPL6947 Illumina HumanHT-12 V3.0 expression beadchip	24	12	12
GSE147890	GPL571[HG-U133A_2] Affymetrix Human Genome U133A 2.0 Array	26	Diabetic_0h: 6	Control_0h: 7
GSE147890	GPL571[HG-U133A_2] Affymetrix Human Genome U133A 2.0 Array	26	Diabetic_24h: 6	Control_24h: 7

Identification and anlysis of DEGs in GSE38396

The R package "Limma" was applied to identify DEGs between DLECs from type 2 diabetes patients and non-diabetic patients, with the selection criteria of the |log2 (fold-change) | > 1 and P-value < 0.05 (12). The volcano and heatmap plots were generated by ggrepel and heatmap packages in R software. The intra-group data repeatability of the dataset was tested by principal component analysis (PCA) based on DEGs.

Enrichment analysis of DEGs

Enrichment analysis of gene ontology, including biological processes (BP), cellular components (CC), and molecular functions (MF), and Kyoto Encyclopedia of Genes and Genomes pathways enrichment analysis were executed using Metascape (https://metascape.org/gp/index.html), which is a public database of gene annotation and analysis resources. In order to make the results more accurate, we performed GSEA enrichment analysis of DEGs using the 'clusterProfiler' package.

Protein-protein interaction (PPI) network analysis and hub genes identification

Loading the DEGs into the Search Tool for the Retrieval of Interacting Genes (STRING 11.5 https://string-db.org/), an online database of known and predicted PPI network, to construct a PPI network. The criterion for significant protein pairings was set at a combined score of > 0.4. The average node degree was 2.08, the average local clustering coefficient was 0.368, and the PPI enrichment p-value was 4.12e-13, according to database results. Cytoscape software was used to visualize and construct the network of proteins by nodes and edges. The high degree value was more likely to be hub genes, which were predicted from numerous candidates by Cytohubba. The plug-in Molecular Complex Detection (MCODE) of Cytoscape was used to extract the most highly connected cluster (degree cutoff = 2, node score cutoff = 0.2, k-core = 2, and max. depth = 100), which aimed to investigate more specific regulatory relationships of proteins (13).

Hub genes analysis

We confirmed the reliability of the hub genes from another perspective by verifying whether they were effective in distinguishing different samples through PCA. Enrichment analysis of hub genes were performed to gain a better understanding of molecular function. The enrichment analysis was carried out using clusterProfiler package, and the top ten results were visualized (14). Subsequently, we analyzed the correlation of hub genes with each other.

Screening and validation of key genes

The R packages “glmnet” was used to build a machine learning model of the least absolute shrinkage and selection operator (LASSO) analysis, which was performed on the entire dataset to screen for key genes related with DLECs of DFU from hub genes. In addition, the GSE147890 dataset was used as an external dataset to validate the prediction model, and the prediction efficiency of the model was evaluated by ROC curves. By indexing GeneCards database (15) and using ‘[expression] (foot) AND [genomics] (lymphatic AND lymphatic AND cells)’ as key words, 232 lymphatic lymphatic cells -related genes expressed in feet were obtained. Finally, key ferroptosis-related gene was obtained by integrating the GSE80178 dataset and ferroptosis-related genes from the FerrDb database, GSE147890 and GSE29221 were utilized as external data for validation of key ferroptosis-related gene.

Construction of ceRNA network

The miRDB database (http://mirdb.org/miRDB/) was selected to predict the targeted miRNAs of hub genes. In addition, miRNet database was used to predict the upstream lncRNAs of miRNAs, which is a tool for miRNA-associated studies (16).

Identification and analysis of DEGs

R software was used to detect the normalization of the expression matrix, as shown in Fig. 2A and Fig. 2B. After normalizing the data, 149 genes were identified to be DEGs, with 56 being up-regulated and 93 being down-regulated genes. The details of the DEGs were shown in the Table S1. Based on the analysis of gene expression results, the volcano plot was drawn as shown in Fig. 2C, and partial genes were labeled with the selection criteria of the |log2 (fold-change) | >= 1 and P-value < 0.05. The heatmap plot (Fig. 2D) was constructed to visually show all DEGS. PCA indicated that DEGs allowed us to distinguish DFU samples from control samples (Fig. 2E).

Functional annotation and pathway enrichment analysis of the DEGs

GO enrichment analysis results of DEGs mainly enriched in positive regulation of protein phosphorylation, collagen-containing extracellular matrix and skeletal system development base on Metascape database (Fig. 3A, B). Additionally, GSEA enrichment analysis results were shown in Fig. 3C, which included basal plasma membrane and developmental process. KEGG pathway enrichment analysis uncovered 12 significantly enriched pathways (Fig. 3D). The detailed results of enrichment analysis were shown in Table S2. According to Log (p-value), HIF-1 signaling pathway, Pathways in cancer and African trypanosomiasis were considered as the more critical pathways.

PPI network construction of DEGs and further excavation of hub genes

The PPI network with 145 nodes and 151 edges was obtained from the STRING database, and the top 50 nodes of the PPI were visualized using Cytoscape according to degree score (Fig. 4A). Node color represented the difference in degree, and node size represented the difference in combined-score. The Cytohubba plug-in was used to determine the hub genes from the top 50 PPI network based on the degree (Fig. 4B). Excluding three isolated nodes, we identified a total of 12 hub genes. The details of the hub genes were shown in the Table S3. We extracted the two cluster from the PPI network using MCODE analysis (Fig. 4C and Fig. 4D).

Comprehensive analysis of hub genes

As shown in Fig. 5A, PCA revealed that hub genes enabled us to effectively distinguish DFU samples from control samples. In addition, a significantly distinct correlation among key genes is illustrated in Fig. 5B. To further explore the specific functions and signaling pathways involved in the 12 hub genes, enrichment analysis of hub genes was conducted. The GO enrichment analysis results revealed that 12 hub genes mainly enriched in apical part of cell, T cell activation and basolateral plasma membrane. The enrichment results for BP, CC and MF based on the top 10 of the p-value were presented in Fig. 5C, Fig. 5D and Fig. 5E. KEGG enrichment analysis of hub genes was shown in Fig. 5F. The detailed results of enrichment analysis were shown in Table S4. Combined with the p-value and enrichment results of DEGs, the HIF-1 pathway was considered as a signaling pathway with a key role in DLECs of DFU (Fig. 5G).

Confirmation of the key gene

We extracted 3 genes (EGFR, TRFC, CCL5) mostly associated with DLEGs of DFU from 12 hub genes by the LASSO prediction model (Fig. 6A and Fig. 6B). The result of external dataset validation suggested that it had a better predictive value (Fig. 6C). Two clusters of hub genes related of DFU were obtained by the GSE80178 dataset (Table 2), which was processed with the same method as GSE38396 data. Ultimately, the predictive model, two clusters of hub genes, genes (EGFR, TFRC, ENO2, SLC2A1, IL6R) contained in the HIF-1 signaling pathway and 232 lymphatic lymphatic cells -related genes expressed in feet from the GeneCards database (Table S5) were synthesized to obtain key gene (EGFR) strongly associated with DLECs of DFU (Fig. 6D).

Table 2

Two clusters of hub genes of GSE80178 dataset.
Cluster	Hub genes
GSE80178A: Diabetic Foot Ulcer VS. Diabetic Foot Skin	EGFR, SPRR1B, CXCL8, LOR, IVL, LCE2B, SPRR2B, LCE6A
GSE80178B: Diabetic Foot Ulcer VS. non-Diabetic Foot Skin	EGFR, SPRR1B, CXCL8, LOR, IVL, S100A7, FLG, SPRR1A, PTGS2, SPRR3

Verifying the key gene and constructing ceRNA network

As illustrated in Fig. 7A, EGFR was lowly expressed in DLECs of DFU samples compared to the control samples. To further determine that EGFR can be used as a key biomarker for the prediction and diagnosis of DLECs of DFU, the ROC curve based on the two external datasets was performed to analyze the sensitivity and specificity of EGFR as a predictor for DFU (Fig. 7B). Focusing on three hub genes (EGFR, TFRC, CCL5) obtained based on LASSO, upstream miRNAs were predicted by miRDB database and miRNAs common to the three genes were selected to predict lncRNAs based on miRNet database. Finally, we constructed a ceRNA network containing 2 mRNAs, 2 miRNAs and 5 lncRNAs (Fig. 7C, Table S6).

DFU is one of the most serious complications of diabetes. Up to a quarter of diabetic patients will develop DFU, and approximately 25% of the hospitalization time of diabetic patients is due to infection or ischemic DFU (17). At present, it is known that DFU is closely related to microvascular lesions, and type 2 diabetes can induce microvascular disease, resulting in endothelial cells that exhibit an imbalance between vasoconstrictor and dilator (18). However, there are few studies on the involvement of lymphatic vessels in diabetic skin changes. We compared the gene expression profiles of DLECs of type 2 diabetes patients to non-diabetic patients in this study, in order to reveal the gene expression differences of DLECs in diabetic skin. We considered EGFR as a potential biomarker which plays a key role in the development of DFU based on multiple database data and LASSO prediction model, and constructed a ceRNA network around EGFR. Furthermore, enrichment analysis implies that the HIF-1 signaling pathway probably plays a role in the pathogenesis of DFU.

Hypoxia is defined as the presence of a small amount of oxygen at the tissue level. Hypoxia is a characteristic condition during wound healing that is established immediately after skin injury due to vascular injury or collapse (19). Hypoxia-inducible factor-1 (HIF-1) is a protein that contains the HIF-1 and HIF-1 subunits and has been shown to have a function in controlling angiogenesis as well as wound healing after vascular damage (20). HIF-1 is a hypoxia-resistant transcription factor. HIF-1 is stable and accumulates in the nucleus under hypoxia conditions, where it binds to hypoxia response elements to trigger the activation of target genes (21). However, studies have shown that in the diabetic wound environment, the stability of HIF-1α under hypoxia will be seriously damaged, and the instability of HIF-1α is considered to be the main factor of diabetic wound nonunion (22). In this study, KEGG enrichment analysis showed that HIF-1 signaling pathway enrichment was the most significant. However, due to the low expression of genes involved in the regulatory pathway, including EGFR, ENO2, SLC2A1, IL6R and TFRC, in diabetic DLECs, the function of the HIF-1 signaling pathway was inhibited, resulting in delayed wound healing, which is consistent with the appeal conclusion. As a result, there are strong reasons to expect that understanding the molecular mechanism of HIF-1 signal down-regulation and the related low-expression genes would help to clarify novel predication and therapy options for DFU wound healing.

The epidermal growth factor receptor (EGFR) is a transmembrane glycoprotein that belongs to the ErbB tyrosine kinase receptor family, which has four members (23). EGFR is made up of three domains: (1) extracellular domain, which binds to the ligand that activates the receptor; (2) hydrophobic transmembrane region, which is involved in receptor dimerization; and (3) intracellular tyrosine kinase domain, which binds to a tyrosine residue on the phosphorylated substrate protein (24). EGFR plays a key role in cell growth, cell differentiation, tissue development, and function. It is an integration of extracellular growth and survival signals that converge and transform into intracellular output (25). Its function is related to a series of neurological metabolic disorders, including diabetes, Alzheimer 's disease and aging (25). The binding of EGFR to polypeptide growth factors can induce the interaction of homologous and heterologous dimers between family members (26). This initiates a series of downstream signaling pathways leading to cell growth, differentiation, adhesion, migration, cell survival (anti-apoptosis), drug and radiation sensitivity (26). In addition, enrichment analysis showed that EGFR has the function of binding to exogenous proteins. It is well known that proteins show excellent biological activity in promoting diabetic wound healing and skin regeneration (27). However, the results of this study showed low expression of EGFR in the DLECs of DFU. We considered that EGFR function cannot be fully manifested in patients with diabetes, which leads to the difficulty of healing ulcer wounds. In addition, we constructed a ceRNA network of lncRNAs. The results revealed that OLMALINC, KCNQ1OT1, HELLPAR, OIP5-AS1, and SNHG16 are likely to act as EGFR upstream regulators affecting the development of DFU. As a driver of ferroptosis, EGFR has been employed as a target for treatment in a variety of diseases(28, 29). The present study demonstrated that EGFR expression was lower in DFU patients than in healthy subjects, but this does not indicate that ferroptosis is inhibited in DFU, as disease development is caused by multifactorial and multifactorial regulation. Therefore, the specific mechanism of action played by EGFR as a target of ferroptosis in DFU needs to be further analyzed and explored.

However, there are limitations to this study. Firstly, the machine learning prediction model of LASSO in the external validation cohort affects the accuracy due to the small sample size, leading to misdiagnosis and missed diagnoses. Thus, the larger DFU sample size can improve the prediction accuracy. Secondly, EGFR as a ferroptosis-related biomarker for DLECs of DFU identified in this study that require further literature support and basic experimental validation.

In conclusion, low expression of EGFR was identified as a key ferroptosis-related biomarker of DLEGs of DFU and the HIF-1 signaling pathway is likely to be involved in the process of DFU development. Moreover, we hypothesized that upstream lncRNAs (OLMALINC, KCNQ1OT1, HELLPAR, OIP5-AS1, and SNHG16) regulate EGFR expression and thus modulate the HIF-1 signaling pathway, ultimately affecting DLECs of DFU progression.

DFU: Dabetic foot ulcer; DLECs: Dermal lymphatic endothelial cells; GEO: Gene expression omnibus; DEGs: Differentially expressed genes; PPI: Protein-protein interaction; LASSO: Least absolute shrinkage and selection operator; ROC: Receiver operating characteristic; GO: Gene Ontology; KEGG: Kyoto Encyclopedia of Genes and Genomes; DM: Diabetes mellitus; BP: Biological processes; MF: Molecular functions; CC: Cellular components; MCODE: Molecular Complex Detection; HIF-1: Hypoxia-inducible factor-1; EGFR: Epidermal growth factor receptor.

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Availability of data and materials

Publicly available datasets were analyzed in this study. The datasets (GSE147890, GSE38396, GSE80178, GSE29221) analyzed during the current study are available in Gene Expression Omnibus (https://www.ncbi.nlm.nih.gov/geo/).

Competing interests

The authors declare that they have no competing interests.

Funding

This study was supported by the National Natural Science Foundation of China (82074426, 82104864, 82204822), Natural Science Foundation of Liaoning Province (2021-BS-215, 2022-MS-25, 2023-MS-13), Liaoning Revitalization Talents Program (XLYC1802014), Liaoning Key Research and Development Planning Project (2017226015), Basic Research Projects of Liaoning Provincial Department of Education (LJKMZ20221286), Naural Science Foundation of Tibet Autonomous Region and Regional Science(XZ202301ZR0030G) and Technology Project of Naqu City.

Author’s contributions

XW, LZ and LM collected the original data, finished the analysis and drafted the initial version. MS, XS, ZJ, LZ (Lin Zhao), XH, HW and PW helped revise the manuscript. JZ and ML offered constructive comments on experimental studies. SW, JZ and ML provided the funding. SW designed and conceptualized the study, put forward many constructive comments for the final version and supervised the study. The final version of the manuscript was approved by all authors.

Acknowledgments

We thank all the participants who supported our study. In particular, thanks to the GEO database and R software for the analytical data.

Chen R, Liang X, Huang T, Zhong W, Luo X. Effects of type 1 diabetes mellitus on lumbar disc degeneration: a retrospective study of 118 patients. J Orthop Surg Res. 2020;15(1):280.
Moon K-C, Kim K-B, Han S-K, Jeong S-H, Dhong E-S. Risk Factors for Major Amputation on Hindfoot Ulcers in Hospitalized Diabetic Patients. Adv Wound Care (New Rochelle). 2019;8(5):177-85.
Liu Z, Zaid M, Hisamatsu T, Tanaka S, Fujiyoshi A, Miyagawa N, et al. Elevated Fasting Blood Glucose Levels Are Associated With Lower Cognitive Function, With a Threshold in Non-Diabetic Individuals: A Population-Based Study. J Epidemiol. 2020;30(3):121-7.
Gawryś J, Wiśniewski J, Szahidewicz-Krupska E, Gajecki D, Leśniewska J, Majda F, et al. Increased Intraplatelet ADMA Level May Promote Platelet Activation in Diabetes Mellitus. Oxid Med Cell Longev. 2020;2020:6938629.
Haemmerle M, Keller T, Egger G, Schachner H, Steiner CW, Stokic D, et al. Enhanced lymph vessel density, remodeling, and inflammation are reflected by gene expression signatures in dermal lymphatic endothelial cells in type 2 diabetes. Diabetes. 2013;62(7):2509-29.
Dixon SJ, Lemberg KM, Lamprecht MR, Skouta R, Zaitsev EM, Gleason CE, et al. Ferroptosis: an iron-dependent form of nonapoptotic cell death. Cell. 2012;149(5):1060-72.
Venkatesan P, Varghese J, Arthi TS, James JV, Anura A, Prasad J, et al. Evidence of dysregulated iron homeostasis in newly diagnosed diabetics, but not in pre-diabetics. J Diabetes Complications. 2021;35(9):107977.
Feng X, Wang S, Sun Z, Dong H, Yu H, Huang M, et al. Ferroptosis Enhanced Diabetic Renal Tubular Injury via HIF-1α/HO-1 Pathway in db/db Mice. Front Endocrinol (Lausanne). 2021;12:626390.
Zheng L, Dou X, Song H, Wang P, Qu W, Zheng X. Bioinformatics analysis of key genes and pathways in Hashimoto thyroiditis tissues. Biosci Rep. 2020;40(7).
Li J, Wang M, Chen X. Long non-coding RNA UCA1 modulates cell proliferation and apoptosis by regulating miR-296-3p/Myc axis in acute myeloid leukemia. Cell Cycle. 2020;19(12):1454-65.
Zhao X, Zhang L, Wang J, Zhang M, Song Z, Ni B, et al. Identification of key biomarkers and immune infiltration in systemic lupus erythematosus by integrated bioinformatics analysis. J Transl Med. 2021;19(1):35.
Cai W, Li H, Zhang Y, Han G. Identification of key biomarkers and immune infiltration in the synovial tissue of osteoarthritis by bioinformatics analysis. PeerJ. 2020;8:e8390.
Yi X-H, Zhang B, Fu Y-R, Yi Z-J. STAT1 and its related molecules as potential biomarkers in Mycobacterium tuberculosis infection. J Cell Mol Med. 2020;24(5):2866-78.
Guo T, Ma H, Zhou Y. Bioinformatics analysis of microarray data to identify the candidate biomarkers of lung adenocarcinoma. PeerJ. 2019;7:e7313.
Safran M, Rosen N, Twik M, BarShir R, Stein TI, Dahary D, et al. The GeneCards Suite. In: Abugessaisa I, Kasukawa T, editors. Practical Guide to Life Science Databases. Singapore: Springer Nature Singapore; 2021. p. 27-56.
Fan Y, Xia J. miRNet-Functional Analysis and Visual Exploration of miRNA-Target Interactions in a Network Context. Methods Mol Biol. 2018;1819:215-33.
Kalan LR, Meisel JS, Loesche MA, Horwinski J, Soaita I, Chen X, et al. Strain- and Species-Level Variation in the Microbiome of Diabetic Wounds Is Associated with Clinical Outcomes and Therapeutic Efficacy. Cell Host Microbe. 2019;25(5).
Hu M, Cai X, Yang W, Zhang S, Nie L, Ji L. Effect of Hemoglobin A1c Reduction or Weight Reduction on Blood Pressure in Glucagon-Like Peptide-1 Receptor Agonist and Sodium-Glucose Cotransporter-2 Inhibitor Treatment in Type 2 Diabetes Mellitus: A Meta-Analysis. J Am Heart Assoc. 2020;9(7):e015323.
Sun Y, Xiong X, Wang X. HIF1α/miR-199a/ADM feedback loop modulates the proliferation of human dermal microvascular endothelial cells (HDMECs) under hypoxic condition. Cell Cycle. 2019;18(21):2998-3009.
Ling K, Xu A, Chen Y, Chen X, Li Y, Wang W. Protective effect of a hydrogen sulfide donor on balloon injury-induced restenosis via the Nrf2/HIF-1α signaling pathway. Int J Mol Med. 2019;43(3):1299-310.
Ivashkiv LB. The hypoxia-lactate axis tempers inflammation. Nat Rev Immunol. 2020;20(2):85-6.
Yang P, Wang D, Shi Y, Li M, Gao M, Yu T, et al. Insulin-Containing Wound Dressing Promotes Diabetic Wound Healing Through Stabilizing HIF-1α. Front Bioeng Biotechnol. 2020;8:592833.
Jutten B, Keulers TG, Peeters HJM, Schaaf MBE, Savelkouls KGM, Compter I, et al. EGFRvIII expression triggers a metabolic dependency and therapeutic vulnerability sensitive to autophagy inhibition. Autophagy. 2018;14(2):283-95.
Al Qaraghuli MM. Biotherapeutic Antibodies for the Treatment of Head and Neck Cancer: Current Approaches and Future Considerations of Photothermal Therapies. Front Oncol. 2020;10:559596.
Bruban J, Voloudakis G, Huang Q, Kajiwara Y, Al Rahim M, Yoon Y, et al. Presenilin 1 is necessary for neuronal, but not glial, EGFR expression and neuroprotection via γ-secretase-independent transcriptional mechanisms. FASEB J. 2015;29(9):3702-12.
Cai W-Q, Zeng L-S, Wang L-F, Wang Y-Y, Cheng J-T, Zhang Y, et al. The Latest Battles Between EGFR Monoclonal Antibodies and Resistant Tumor Cells. Front Oncol. 2020;10:1249.
Li Y, Xu T, Tu Z, Dai W, Xue Y, Tang C, et al. Bioactive antibacterial silica-based nanocomposites hydrogel scaffolds with high angiogenesis for promoting diabetic wound healing and skin repair. Theranostics. 2020;10(11):4929-43.
Fang Y, Fang D. Comprehensive analysis of placental gene-expression profiles and identification of EGFR-mediated autophagy and ferroptosis suppression in intrahepatic cholestasis of pregnancy. Gene. 2022;834:146594.
Sun J, Zhou C, Zhao Y, Zhang X, Chen W, Zhou Q, et al. Quiescin sulfhydryl oxidase 1 promotes sorafenib-induced ferroptosis in hepatocellular carcinoma by driving EGFR endosomal trafficking and inhibiting NRF2 activation. Redox Biol. 2021;41:101942.

No competing interests reported.

Download PDF

Version 1

posted

You are reading this latest preprint version

Machine learning revealed novel ferroptosis-related genes and construction ceRNA network in dermal lymphatic endothelial cells of diabetic foot ulcer

Status:

Version 1

Abstract

Figures

Introduction

Methods

Microarray dataset information and processing

Identification and anlysis of DEGs in GSE38396

Enrichment analysis of DEGs

Protein-protein interaction (PPI) network analysis and hub genes identification

Hub genes analysis

Screening and validation of key genes

Construction of ceRNA network

Results

Identification and analysis of DEGs

Functional annotation and pathway enrichment analysis of the DEGs

PPI network construction of DEGs and further excavation of hub genes

Comprehensive analysis of hub genes

Confirmation of the key gene

Verifying the key gene and constructing ceRNA network

Discussion

Conclusions

Abbreviations

Declarations

References

Additional Declarations

Supplementary Files

Status:

Version 1