Model-based selection of the robust JAK-STAT activation mechanism

Area of Science:

• Systems Biology
• Computational Biology
• Molecular Signaling

Background:

• The Janus kinase-signal transducer and activator of transcription (JAK-STAT) pathway is vital for eukaryotic cell functions, including growth and apoptosis.
• Dysregulation of JAK-STAT signaling is implicated in cancerogenesis, making its components attractive drug targets.
• Existing computational models face challenges in accurately representing receptor activation ambiguities within the JAK-STAT pathway.

Purpose of the Study:

• To compare four computational models of the JAK1/2-STAT1 signaling pathway with varying receptor activation mechanisms.
• To evaluate the effectiveness of model selection methods like Bayesian model selection (BMS) and global sensitivity analysis (GSA) for complex biological models.
• To identify key factors influencing pathway robustness and inform the selection of the most appropriate model.

Main Methods:

• Comparative analysis of four JAK1/2-STAT1 signaling pathway models using mass action kinetics.
• Application of Bayesian model selection (BMS) to assess model goodness of fit.
• Utilization of global sensitivity analysis (GSA) and identifiability analysis (IA) to investigate model robustness and parameter influence.

Main Results:

• Standard BMS failed to significantly differentiate between model variants due to marginal differences and limited data.
• Both BMS and GSA indicated a slight preference for the least complex model, focusing on the receptor activation component.
• Comprehensive GSA demonstrated that on-membrane dimer pre-assembly is more critical than the precise ligand-binding and dimerization reaction order for pathway robustness.

Conclusions:

• In scenarios with limited experimental data, GSA and IA can offer more insights than traditional BMS for selecting among similar complex biological models.
• Robustness analysis via GSA provides a valuable framework for understanding parameter importance and guiding expert-mediated model selection.
• The study highlights the importance of on-membrane receptor dimer pre-assembly in JAK-STAT signaling and offers a methodological approach for model evaluation in systems biology.