机器学习与单细胞测序揭示胃癌进展中线粒体自噬的潜在调控因子。

Machine learning and single-cell sequencing reveal the potential regulatory factors of mitochondrial autophagy in the progression of gastric cancer.

作者信息

Wei Chen, Ma Yichao, Wang Fei, Chen Yuji, Liao Yiqun, Zhao Bin, Zhao Qi, Tang Dong

机构信息

Clinical Medical College, Yangzhou University, Yangzhou, Jiangsu Province, China.

Clinical Medical College, Dalian Medical University, Dalian, Liaoning Province, China.

出版信息

J Cancer Res Clin Oncol. 2023 Nov;149(17):15561-15572. doi: 10.1007/s00432-023-05287-9. Epub 2023 Aug 31.

DOI:10.1007/s00432-023-05287-9

PMID:37648811

Abstract

BACKGROUND

As an important regulatory mechanism to remove damaged mitochondria and maintain the balance between internal and external cells, mitochondrial autophagy plays a key role in the progression and treatment of cancer Onishi (EMBO J 40(3): e104705, 2021). The purpose of this study is to comprehensively analyze the role of mitochondrial autophagy-related genes in the progression of gastric cancer (GC) by RNA sequencing (RNA-seq) and single-cell RNA sequencing (scRNA-seq).

METHODS

GSE26942, GSE54129,GSE66229,GSE183904 and other data sets were obtained by GEO databases. Using support vector machine recursive feature elimination (SVM-RVF) algorithm and random forest algorithm, the mitochondrial autophagy-related genes related to gastric cancer were obtained, respectively. After that, the model was constructed and the inflammatory factors, immune score and immune cell infiltration were analyzed. Furthermore, according to the scRNA-seq data of 28,836 cells from 13 GC samples, 18 cell clusters and 7 cell types were identified by scRNA-seq analysis. The expression level and signal pathway of related genes were verified by cell communication analysis. Finally, the regulatory network of cells was analyzed by SCENIC.

RESULTS

MAP1LC3B, PGAW5, PINK1, TOMM40 and UBC are identified as key genes through machine learning algorithms. CXCL12-CXCR4, LGALS9-CD44, LGALS9-CD45 and MIF (CD74 + CD44) pathways may play an important role in endothelial cells with high score scores of T cells and monocytes in tumor environment. CEBPB, ETS1, GATA2, MATB, SPl1 and XBP1 were identified as candidate TF with specific regulatory expression in the GC cell cluster.

CONCLUSION

The results of this study will provide implications for the study of the mechanism, diagnosis and treatment of mitochondrial autophagy in GC.

摘要

背景

线粒体自噬作为清除受损线粒体并维持细胞内外平衡的重要调节机制，在癌症的进展和治疗中起着关键作用（大西，《欧洲分子生物学组织杂志》40(3): e104705，2021）。本研究旨在通过RNA测序（RNA-seq）和单细胞RNA测序（scRNA-seq）全面分析线粒体自噬相关基因在胃癌（GC）进展中的作用。

方法

通过GEO数据库获取GSE26942、GSE54129、GSE66229、GSE183904等数据集。分别使用支持向量机递归特征消除（SVM-RVF）算法和随机森林算法，获得与胃癌相关的线粒体自噬相关基因。之后，构建模型并分析炎症因子、免疫评分和免疫细胞浸润情况。此外，根据来自13个GC样本的28836个细胞的scRNA-seq数据，通过scRNA-seq分析鉴定出18个细胞簇和7种细胞类型。通过细胞通讯分析验证相关基因的表达水平和信号通路。最后，通过SCENIC分析细胞的调控网络。

结果

通过机器学习算法鉴定出MAP1LC3B、PGAW5、PINK1、TOMM40和UBC为关键基因。CXCL12-CXCR4、LGALS9-CD44、LGALS9-CD45和MIF（CD74 + CD44）通路可能在肿瘤环境中T细胞和单核细胞得分较高的内皮细胞中发挥重要作用。CEBPB、ETS1、GATA2、MATB、SPl1和XBP1被鉴定为在GC细胞簇中具有特定调控表达的候选转录因子。