AUTHOR=Niu Xiaoguang , Xu Man , Zhu Jian , Zhang Shaowei , Yang Yanning TITLE=Identification of the immune-associated characteristics and predictive biomarkers of keratoconus based on single-cell RNA-sequencing and bulk RNA-sequencing JOURNAL=Frontiers in Immunology VOLUME=14 YEAR=2023 URL=https://www.frontiersin.org/journals/immunology/articles/10.3389/fimmu.2023.1220646 DOI=10.3389/fimmu.2023.1220646 ISSN=1664-3224 ABSTRACT=Background

Whether keratoconus (KC) is an inflammatory disease is currently debated. Hence, we aimed to investigate the immune-related features of KC based on single-cell RNA sequencing (scRNA-seq) and bulk RNA sequencing (bulk RNA-seq) data.

Methods

scRNA-seq data were obtained from the Genome Sequence Archive (GSA), bulk RNA-seq data were obtained from the Gene Expression Omnibus (GEO), and immune-associated genes(IAGs) were obtained from the ImmPort database. Cell clusters of KC were annotated, and different cell clusters were then selected. The IAG score of each cell was calculated using the AUCell package. Three bulk RNA-seq datasets were merged and used to identify the differentially expressed genes (DEGs), biological functions, and immune characteristics. Weighted gene coexpression network analysis (WGCNA) was used to select the IAG score-related hub genes. Based on scRNA-seq and bulk RNA-seq analyses, three machine learning algorithms, including random forest (RF), support vector machine (SVM), and least absolute shrinkage and selection operator (LASSO) regression analysis, were used to identify potential prognostic markers for KC. A predictive nomogram was developed based on prognostic markers.

Results

Six cell clusters were identified in KC, and decreased corneal stromal cell-5 (CSC-5) and increased CSC-6 were found in KC. CSC and immune cell clusters had the highest IAG scores. The bulk RNA-seq analysis identified 1362 DEGs (553 upregulated and 809 downregulated) in KC. We found different immune cell populations and differentially expressed cytokines in KC. More than three key IAG score-related modules and 367 genes were identified. By integrating the scRNA-seq and bulk RNA-seq analyses, 250 IAGs were selected and then incorporated into three machine learning models, and 10 IAGs (CEP112, FYN, IFITM1, IGFBP5, LPIN2, MAP1B, RNASE1, RUNX3, SMIM10, and SRGN) were identified as potential prognostic genes that were significantly associated with cytokine and matrix metalloproteinase(MMP)1-14 expression. Finally, a predictive nomogram was constructed and validated.

Conclusion

Taken together, our results identified CSCs and immune cell clusters that may play a key role during KC progression by regulating immunological features and maintaining cell stability.