ABSTRACT: Background: There are no obvious diagnostic markers in the early stages of ccRCC and current targeted therapeutic regimens are susceptible to drug resistance. Our goal is to explore the molecular mechanisms associated with autophagy during the development of ccRCC and to identify molecular markers of prognostic value using scRNA-seq data in combination with stRNA-seq data. Methods: We performed single-cell sequencing of 6 ccRCC samples (3 tumours and 3 controls), stRNA-seq data from 12 samples (GSE175540), and bulk RNA-seq data from 614 patients (TCGA-KIRC). The cell subpopulations were artificial annotated, and the cell pathway activity were assessed. Autophagy-related genes(ARGs) obtained by intersecting autophagy databases with single-cell data and key clusters were identified using AUCell, Subsequent analyses included ligand-receptor interactions and SCENIC,. Combining single-cell genomics and spatial genomics, we used RCTD to visualise cell spatial distribution, analysed spatial cell interactions, studied cell trajectories using stLearn. Finally, identified prognostic-related genes and constructed a nomogram model using LASSO regression and AUCs, based on the risk score, divided into high- and low-risk groups and underwent immune infiltration analysis.. Additionally, we analysed the spatial distribution of prognostic genes. Results Through single-cell sequencing data, we manually annotated 14 cell subpopulations, with TAE being the key cluster. By cross-referencing with the autophagy database, we obtained 216 ARGs. Further CellChat analysis revealed complex interactions. The SCENIC analysis identified the top 5 regulons with the highest RSS scores (EGR1, DDIT3, IRF1, XBP1, and KLF4) and associated 204 transcription factors (TFs). Through spatial transcriptomics analysis, we further determined the cell distribution from a three-dimensional perspective and visualised the spatial dynamics during cell development, the MISTy indicated strong interactions between TAE, fibroblasts, and T cells. By combining ligand-receptor interaction analysis in single cells, we identified the MIF-(CD74+CD44) pathway as having the highest signal intensity, with colocalization indicating a strong correlation between the two. Finally, by analysing the TCGA-KIRC data, we developed a prognostic model based on nine genes (KLF6, UBE2S, PMAIP1, NFKBIZ, CES2, RCN1, C1S, SRSF2, and PKHD1). The high-risk group exhibited significantly poorer survival outcomes, and the spatial distribution of prognostic genes was also visualized. Conclusion: Our study offers new insights into the pathogenesis of autophagy-related clear cell renal cell carcinoma (ccRCC) and potential therapeutic targets.
