Comprehensive Transcriptomic Analysis Integrating Bulk and Single-Cell RNA-Seq with Machine Learning to Identify and Validate Mitochondrial Unfolded Protein Response Biomarkers in Patients with Ischemic Stroke

Lu Zhang ^1* Lei Yue ² Ziqi Cheng ^3* Jiwen Liu ^4*

• ¹ Institute of Science and Technology for Brain-Inspired Intelligence, Fudan University, Shanghai, Shanghai Municipality, China
• ² Department of Neurology, Shangrao Municipal Hospital Shangrao 334000 Jiangxi Province China, Shangrao, China
• ³ Department of Neurosurgery, Renmin Hospital of Wuhan University, Wuhan, Hubei Province, China
• ⁴ Department of Emergency Medicine, Shanghai Pudong New Area Gongli Hospital, Miaopu Road ，Shanghai 200135, Shanghai，China, Shanghai, China

Background: Ischaemic stroke (IS) represents a significant contributor to morbidity and mortality globally. The relationship between IS and mitochondrial unfolded protein response UPRmt was presently uncertain. This study endeavors to explore the fundamental mechanism of UPRmt in IS by utilizing bioinformatics methods.Methods: In GSE58294, differentially expressed genes (DEGs) were obtained, which were overlapped with key module genes of UPRmt-related gene (UPRmt-RGs) for producing candidate genes. The biomarkers were identified from the candidate genes through machine learning, expression validation, and receiver operating characteristic (ROC) curves. In order to verify the biomarkers, reverse transcription-quantitative PCR (RT-qPCR) experiments were performed on human peripheral blood. Subsequently, a predictive nomogram was created to estimate the likelihood of developing IS. Next, the mechanisms and functions related to the biomarkers were explored by enrichment analysis and immune infiltration. In addition, cells enriched with biomarkers were identified, and the biological processes involved in these cells were analyzed through intercellular communication analysis and virtual knockout experiments.Results:MCEMP1, CACNA1E, and CLEC4D were identified as biomarkers and subsequently validated by RT-qPCR. RT-qPCR revealed that CLEC4D is the most sensitive biomarker. The nomogram analysis revealed that these biomarkers possess strong diagnostic value. Immune infiltration analysis indicated that all three biomarkers are strongly correlated with neutrophils. Additionally, in the single-cell transcriptome data, these biomarkers were predominantly enriched in neutrophils. Compared to the sham group, the middle cerebral artery occlusion (MCAO) group exhibited enhanced immune-inflammatory responses. Virtual knockout experiments provide preliminary evidence that CLEC4D functions as a regulatory molecule in neutrophil-mediated inflammation, rather than serving merely as a passive marker.Conclusion:CLEC4D was identified as the most sensitive biomarker for IS related to UPRmt-RGs, offering a new reference for IS diagnosis and treatment.