Protein–protein interactions (PPIs) occur when two or more proteins physically associate to perform a biological function. In doing so, they regulate most major cellular functions, including metabolism, DNA replication, cell signaling, and the maintenance of membrane and cellular architecture. Disruption of these interactions, for example when one of the proteins is mutated, can lead to severe diseases.
HIGH-THROUGHPUT ANALYSIS OF PROTEIN–PROTEIN INTERACTIONS
Analysis of protein complexes based on the effects of mutations provides valuable information for precisely locating interface positions, quantifying the contribution of individual residues to binding strength (structural constraints), and identifying critical binding sites. When generated on a very large scale through deep mutational scanning and integrated into machine-learning models, these datasets on the effects of mutations can guide molecular optimization in research, diagnostics, and therapeutic development. In cellulo strategies, such as two-hybrid systems, provide easy access to this type of data, but their reliability depends on their quantitative properties.
SUPERIORITY OF THE QUANTITATIVE BACTERIAL TWO-HYBRID SYSTEM qB2HSBE
Aware of the limitations of existing bacterial two-hybrid systems (B2H), which hinder the generation of accurate datasets, the authors present in this study a low-stochasticity system developed in their laboratory. Evaluated, optimized, and validated using complexes between the histone chaperone protein Asf1 and synthetic peptides displaying varying affinities for Asf1, qB2HSBE outperforms earlier-generation B2H systems. It enables strain-independent analyses, improved measurements, and the generation of high-quality protein-interaction datasets ideally suited for training AI models. Analysis of perturbations induced by single-site variants accurately identified the known contact positions between Asf1 and the peptides, in agreement with crystallographic data. Integrating a generative AI-based design strategy led to the development of a peptide exhibiting a 70-fold increase in affinity for Asf1.
The qB2HSBE system described in this study provides R&D researchers with a robust and reusable platform for the quantitative analysis of protein–protein interactions, enabling both rational protein engineering and data-driven discovery. The code, data, and materials are freely available to the scientific community.
Joliot/I2BC institute contacts : Oscar P. Ramos (oscar.pereira-ramos@cea.fr) ; Françoise Ochsenbein (francoise.ochsenbein@cea.fr ; francoise.ochsenbein@i2bc.paris-saclay.fr)
Legend : Fluorescence of bacterial colonies expressing the GFP protein, indicating an interaction between Asf1 and two peptides with different affinities: stronger fluorescence (right) is observed when the peptide has higher affinity. © Guyot et al., Comput Struct Biotechnol J. 2026
- The two-hybrid system is a genetic system used to detect a physical interaction between two proteins within a living cell. Here, the amount of GFP (Green Fluorescent Protein) produced by a bacterium is proportional to the strength of the interaction between Asf1 and a given peptide. A second reporter of interaction strength was implemented, making it possible to correlate interaction strength with cell proliferation and thereby perform high-throughput quantification.
- Françoise Ochsenbein (I2BC) and her team have been studying the histone chaperone protein Asf1 (Anti-Silencing Function 1) for several years as a potential therapeutic target for cancer. The team designs anticancer peptides that competitively inhibit the interaction between Asf1 and histones, an interaction that contributes to uncontrolled cell proliferation. These peptides exhibit very high affinity for Asf1 and are capable of disrupting its physical interaction with histones.