Table1_Identification and analysis of hub genes of hypoxia-immunity in type 2 diabetes mellitus.DOCX

The chronic metabolic disease named type 2 diabetes (T2D) accounts for over 90% of diabetes mellitus. An increasing number of evidences have revealed that hypoxia has a significantly suppressive effect on cell-mediated immunity, as well as the utilization of glucose in diabetics. Therefore, we aimed to screen and identify hypoxia-immune-related hub genes in T2D through bioinformatic analysis. The Gene Expression Omnibus (GEO) database was used to get T2D gene expression profile data in the peripheral blood samples (GSE184050), and hypoxia-related genes were acquired from Molecular Signatures Database (MSigDB). Differentially expressed mRNAs (DEGs) and lncRNAs (DELs) between T2D and normal samples were identified by DeSeq2 package. The clusterProfiler package was used to perform enrichment analyses for the overlapped genes of DEGs and hypoxia-related genes. Further, Hypoxia-related hub genes were discovered using two machine learning algorithms. Next, the compositional patterns of immune and stromal cells in T2D and healthy groups were estimated by using xCell algorithm. Moreover, we used the weighted correlation network analysis (WGCNA) to examine the connection between genes and immune cells to screen immune-related genes. Gene Set Enrichment Analysis (GSEA) was used to investigate the functions of the hypoxia-immune-related hub genes. Finally, two peripheral blood cohorts of T2D (GSE184050 and GSE95849) as well as the quantitative real-time PCR (qRT-PCR) experiments for clicinal peripheral blood samples with T2D were used for verification analyses of hub genes. And meanwhile, a lncRNA-TF-mRNA network was constructed. Following the differentially expressed analysis, 38 out of 3822 DEGs were screened as hypoxia-related DEGs, and 493 DELs were found. These hypoxia-related DEGs were mainly enriched in the GO terms of pyruvate metabolic process, cytoplasmic vesicle lumen and monosaccharide binding, and the KEGG pathways of glycolysis/gluconeogenesis, pentose phosphate pathway and biosynthesis of nucleotide sugars. Moreover, 7 out of hypoxia-related DEGs were identified as hub genes. There were six differentially expressed immune cell types between T2D and healthy samples, which were further used as the clinical traits for WGCNA to identify AMPD3 and IER3 as the hypoxia-immune-related hub genes. The results of the KEGG pathways of genes in high-expression groups of AMPD3 and IER3 were mainly concentrated in glycosaminoglycan degradation and vasopressin-regulated water reabsorption, while the low-expression groups of AMPD3 and IER3 were mainly associated with RNA degradation and nucleotide excision repair. Finally, when compared to normal samples, both the AMPD3 and IER3 were highly expressed in the T2D groups in the GSE184050 and GSE95849 datasets. The result of lncRNA-TF-mRNA regulatory network showed that lncRNAs such as BACH1-IT1 and SNHG15 might induce the expression of the corresponding TFs such as TFAM and THAP12 and upregulate the expression of AMPD3. This study identified AMPD3 and IER3 as hypoxia-immune-related hub genes and potential regulatory mechanism for T2D, which provided a new perspective for elucidating the upstream molecular regulatory mechanism of diabetes mellitus.

2型糖尿病（type 2 diabetes, T2D）作为一种慢性代谢性疾病，其患者占糖尿病（diabetes mellitus）总病例的90%以上。越来越多的研究证据表明，缺氧（hypoxia）可显著抑制糖尿病患者的细胞免疫功能与葡萄糖利用过程。为此，本研究旨在通过生物信息学分析，筛选并鉴定2型糖尿病中与缺氧-免疫相关的核心基因（hub genes）。本研究从基因表达综合数据库（Gene Expression Omnibus, GEO）下载外周血样本来源的2型糖尿病基因表达谱数据（数据集编号GSE184050），并从分子特征数据库（Molecular Signatures Database, MSigDB）获取缺氧相关基因。使用DeSeq2 R包鉴定2型糖尿病患者与正常对照样本间的差异表达mRNA（differentially expressed mRNAs, DEGs）与差异表达长链非编码RNA（differentially expressed lncRNAs, DELs）。采用clusterProfiler R包对DEGs与缺氧相关基因的交集基因进行富集分析。进一步通过两种机器学习算法筛选得到缺氧相关核心基因。随后，利用xCell算法估算2型糖尿病组与健康对照组的免疫细胞与基质细胞组成模式。此外，通过加权基因共表达网络分析（weighted correlation network analysis, WGCNA）探究基因与免疫细胞间的关联，以筛选免疫相关基因。使用基因集富集分析（Gene Set Enrichment Analysis, GSEA）解析缺氧-免疫相关核心基因的功能。最后，采用两项外周血来源的2型糖尿病队列（数据集GSE184050与GSE95849），以及2型糖尿病患者临床外周血样本的定量实时荧光定量PCR（quantitative real-time PCR, qRT-PCR）实验，对核心基因进行验证分析。同时，构建了lncRNA-转录因子（transcription factor, TF）-mRNA调控网络。经差异表达分析后，本研究从3822个DEGs中筛选出38个缺氧相关DEGs，并鉴定得到493个DELs。这些缺氧相关DEGs主要富集于丙酮酸代谢过程、细胞质囊泡腔及单糖结合等基因本体（Gene Ontology, GO）条目，以及糖酵解/糖异生、磷酸戊糖途径与核苷酸糖生物合成等京都基因与基因组百科全书（Kyoto Encyclopedia of Genes and Genomes, KEGG）通路。此外，从缺氧相关DEGs中鉴定出7个核心基因。2型糖尿病组与健康对照样本间存在6种差异表达免疫细胞类型，将其作为临床表型纳入WGCNA分析，最终鉴定出AMPD3与IER3为缺氧-免疫相关核心基因。AMPD3与IER3高表达组的基因KEGG通路主要富集于糖胺聚糖降解及血管加压素调节的水重吸收过程，而二者低表达组的基因通路则主要与RNA降解及核苷酸切除修复相关。最后，在GSE184050与GSE95849数据集中，与正常对照样本相比，AMPD3与IER3在2型糖尿病组中均呈高表达。lncRNA-TF-mRNA调控网络分析结果显示，BACH1-IT1、SNHG15等长链非编码RNA可通过调控TFAM、THAP12等对应转录因子的表达，进而上调AMPD3的基因表达。本研究鉴定出AMPD3与IER3为2型糖尿病的缺氧-免疫相关核心基因及潜在调控机制，为阐明糖尿病的上游分子调控机制提供了新的研究视角。