Optimized quantitative bacterial two-hybrid (qB2H) for protein-protein interaction assessment

Understanding protein quaternary assemblies is essential for both biology and clinical applications. Insight into these assemblies can be gained by assessing how mutations alter their function. Massively parallel analyses of protein-protein interaction (PPI) variants enable the identification of interaction interfaces, key contact residues, and the generation of datasets suitable for machine learning-based approaches. In cellulo strategies, such as two-hybrid systems, provide straightforward access to such information. However, the reliability of the resulting insights critically depends on the quality of the underlying quantitative data. Here, we evaluated the quantitative properties of previously described bacterial two-hybrid (B2H) systems and found limitations that constrain their ability to generate robust datasets. Building on this analysis, we engineered and benchmarked optimized quantitative B2H (qB2H) alternatives. These systems support strain-independent assays, improve data reproducibility, and enable the generation of high-quality, quantitative datasets suitable for downstream applications. We show the utility of qB2H in two contexts: interface mapping and binder optimization. For interface mapping, perturbation analysis of single-site variants accurately recovered known contact positions within ASF1 complexes, in agreement with crystallographic data. For binder optimization, we show that a peptide binder of ASF1 can be improved by approximately 70-fold. Together, these results establish qB2H as a convenient and robust system for quantitative PPI characterization, providing a reliable framework for protein engineering and for generating datasets that can fuel machine learning-driven discovery.