Cell-type specific single-cell signatures reveal nephrotoxic drug effects

Area of Science:

• Nephrology
• Genomics
• Computational Biology

Background:

• Drug-induced acute kidney injury (AKI) is a significant cause of morbidity and mortality in hospitalized patients.
• Current biomarkers for AKI, such as creatinine levels, detect injury only after it has occurred.
• Understanding the cellular mechanisms of drug-induced AKI is limited, hindering prediction and prevention.

Purpose of the Study:

• To investigate the role of specific kidney cell subtypes in drug-induced nephrotoxicity.
• To develop a predictive model for AKI using single-cell transcriptomic data.
• To identify novel molecular mechanisms underlying AKI.

Main Methods:

• Utilized single-cell RNA sequencing (scRNAseq) data from the Human Cell Atlas for 32 kidney cell types.
• Calculated cellular response scores to estimate drug effects on different kidney cell populations.
• Developed an XGBoost model to predict AKI based on transcriptomic signatures.

Main Results:

• Identified significant gene expression differences in 6 kidney cell types, including Indistinct intercalated cells and Epithelial Progenitor cells.
• The XGBoost model achieved an Area Under the Receiver Operating Characteristic curve (AUROC) of 0.6 on an external test set.
• Single-cell transcriptomic signatures provided insights into previously unexplained molecular mechanisms of nephrotoxicity.

Conclusions:

• Specific kidney cell subtypes play a crucial role in drug-induced nephrotoxicity.
• scRNAseq-based models offer improved prediction of AKI compared to traditional methods.
• The study provides a computational framework and dataset for future AKI research.