Software development for improved sensitivity of massspectrometry-based thermal shift assays (MS-TSA) for target engagement and drug discovery

Importance of the Topic

Mass spectrometry–based thermal shift assays (MS-TSA) have emerged as a powerful approach to probe protein stability changes upon ligand binding, enabling high-throughput target engagement analyses in drug discovery. Enhancing the sensitivity and robustness of MS-TSA data processing is critical to detect subtle thermal shifts, reduce false negatives, and accelerate the identification of true binders in complex proteomes.

Goals and Overview of the Study

The study introduces a comprehensive software pipeline designed to improve the detection of thermal stability changes in MS-TSA experiments. Key objectives are to implement advanced peptide filtering strategies, integrate rigorous quality-control metrics, and apply statistical hypothesis testing on fitted melting curves to reliably distinguish true thermal shifts from background variation.

Methodology and Instrumentation

The workflow comprises cultured Jurkat cells subjected to lysis, reduction, alkylation, enzymatic digestion, and labeling for isobaric quantitation. Peptide mixtures are analyzed on high-resolution Thermo Fisher Scientific mass spectrometers. Data processing employs R and Shiny for user interaction, Excel for input formatting, and STRINGdb for functional annotation. Search parameters include 1% false discovery rate, up to three missed cleavages, and ±15 ppm mass tolerance. Thermal profiles are modeled by spline curve fitting at protein and peptide levels.

Key Results and Discussion

• Peptide Filtering: Implementing shared versus unique peptide filters raised the number of proteins with evaluable melt curves by up to 27%.
• Statistical Analysis: A residual-sum-of-squares (RSS) F-test approach robustly discriminates shifted versus non-shifted proteins, with RSS1 ≪ RSS0 indicating significant thermal stabilization or destabilization.
• Quality Control: Missing value patterns correlate with low baseline reporter-ion intensities; higher charge states exhibit poorer signal-to-noise ratios; variability increases at elevated temperatures, highlighting the need for rigorous QC thresholds.
• Multi-Experiment Handling: The pipeline can process between one and eight or more replicate experiments, flagging outliers and overlaps in melting behavior.

Benefits and Practical Applications

• Enhanced Sensitivity: Automated filtering and statistical testing reduce false negatives and improve detection of weak protein–ligand interactions.
• Scalability: Modular design allows users to scale from single assays to large screening campaigns.
• Community Adoption: Implementation in R/Shiny promotes transparency, reproducibility, and ease of customization for diverse proteomic laboratories.

Future Trends and Applications

• External Data Validation: Applying the pipeline to public thermal proteome profiling datasets to benchmark performance improvements.
• Receiver Operating Characteristic (ROC) Analysis: Developing ROC curves to quantify assay sensitivity and specificity across diverse target classes.
• Software Distribution: Packaging as an open-source R library for the broader scientific community.

Conclusion

This work delivers a robust, flexible software solution for MS-TSA data analysis, combining advanced peptide filtering, stringent quality control, and statistical rigor to enhance the reliability of thermal shift detection. By increasing the number of proteins with valid melt curves and enabling multi-experiment integration, the pipeline stands to accelerate target identification and compound prioritization in drug discovery efforts.

Reference

• Franken H., Mathieson T., Childs D., Sweetman G. M., Werner T., Tögel I., … Savitski M. M. Thermal proteome profiling for unbiased identification of direct and indirect drug targets using multiplexed quantitative mass spectrometry. Nature Protocols. 2015;10(10):1567–1593.
• Childs D., Bach K., Franken H., Anders S., Kurzawa N., Bantscheff M., Savitski M. M., Huber W. Non-parametric analysis of thermal proteome profiles reveals novel drug-binding proteins. Molecular & Cellular Proteomics. 2019; mcp.TIR119.001481.