Protein-Protein Interactions in Meiosis – PPIMei

The PPIMei project was built upon a fully integrated strategy, combining predictive bioinformatics approaches, innovative high-throughput experimental technologies, and functional analyses in cells. This combination made it possible to study the dynamics and specificity of protein-protein interactions involved in meiotic crossover formation with unprecedented resolution.

One of the project’s major innovations was the development and application of a Deep Mutational Scanning (DMS) technology within an optimized Bacterial Two-Hybrid (B2H) system. This method allows for the generation and testing of thousands of protein variants in parallel, enabling precise mapping of interaction interfaces. The approach was successfully applied to Zip4, a key protein in the ZMM complex, and its partners, leading to the identification of both loss-of-interaction and compensatory mutants—a technical achievement rarely reached in this context.

In parallel, the project anticipated and leveraged the AlphaFold2 revolution. The integration of 3D structure prediction via deep learning and coevolutionary analysis enabled the modeling of interfaces within disordered or poorly structured regions, which had long remained inaccessible to traditional methods. This strategy paved the way for a detailed analysis of interaction networks involving flexible domains, often underestimated in structural biology.

The project also demonstrated the generalizability of these tools by applying them to other critical meiotic systems. For instance, in the study of the Mre11-Rad50-Sae2 complex, compensatory mutations were used to validate a mechanism of endonuclease activation, illustrating the robustness of the tools developed within PPIMei.

Finally, the 2025 study explored structural interactions between the MLH1–MLH3, MSH4, and EXO1 complexes, which are essential for resolving Holliday junctions. The structural models guided the rational design of mutations targeting specific contact surfaces, which were subsequently validated through genetic and biochemical assays. This illustrates one of the project’s central goals: to build a bridge between structural prediction and functional validation.

The originality of the project lies in its ability to deploy a complete technological pipeline—from in silico prediction to in vivo functional validation, via high-throughput screening. These methods not only answered the project’s initial questions but also opened the way for broader applications, notably in the study of complex interactomes and human diseases linked to meiosis.

The PPIMei project was based on a fully integrated strategy, combining predictive bioinformatics approaches, innovative high-throughput experimental technologies, and in-cell functional analyses. This combination enabled the investigation of the dynamics and specificity of protein-protein interactions involved in meiotic crossover formation with unprecedented resolution. The approach was successfully applied to Zip4, a key component of the ZMM complex, and its partners [1], allowing the identification of loss-of-interaction mutants and validation of the proteins’ assembly mode.

In parallel, the project anticipated and took full advantage of the AlphaFold2 revolution. The integration of 3D structure prediction using deep learning, along with coevolutionary analyses, allowed the prediction of interfaces within disordered or poorly structured regions—long considered inaccessible to classical methods. This led to the development of the SCAN_IDR approach [2], which opened the way to detailed analyses of interaction networks involving flexible domains, often overlooked in structural studies. SCAN_IDR was, for example, successfully validated through the study of the Mre11-Rad50-Sae2 complex, where compensatory mutations supported a mechanism for endonuclease activation [3], illustrating the robustness of the tools developed within PPIMei.

Finally, [4] explored the structural interactions between the MLH1–MLH3, MSH4, and EXO1 complexes, which are essential for resolving Holliday junctions. Structural models guided the rational design of mutations targeting specific contact surfaces, which were subsequently validated through genetic and biochemical assays. These studies illustrate one of the project’s core objectives: to build a bridge between structural prediction and functional validation.

A major innovation of the project also involved the development and application of a Deep Mutational Scanning (DMS) technology within an optimized Bacterial Two-Hybrid (B2H) system. This method was optimized to test thousands of protein variants in parallel, providing high-resolution maps of interaction interfaces [5]. Although initially validated on model systems, the arrival of AlphaFold2 enabled faster progress in characterizing interfaces within the target proteins of the project, prompting a shift in focus. The DMS_B2H system was further developed in a 2D format, which has now been applied to map antibody-antigen complex surfaces in a separate study currently in preparation.

[1] de Muyt et al, Genes & Dev (2022), 10.1101/gad.348973.121

[2] Bret et al, Nat Comm (2024), 10.1038/s41467-023-44288-7

[3] Nicolas, Bret et al, Mol Cell (2024), 10.1016/j.molcel.2024.05.019

[4] Roy et al, Nat Comm (2025), 10.1038/s41467-025-59470-2

[5] Guyot et al, BioRxiv (2025), 10.1101/2025.10.29.680610

The PPIMei project laid the foundation for a robust technological pipeline that combines advanced structural modeling, high-throughput mutagenesis screening, and functional validation. Thanks to its efficiency and modularity, this approach is now being applied to the study of protein interaction networks beyond the context of yeast meiosis, including in complex systems related to DNA repair, cell signaling, and migration.

The bioinformatic tools developed—particularly those integrating deep learning, coevolutionary analysis, and experimental constraints from DMS—have been generalized for the analysis of large-scale interaction networks in various research projects. These methods are now being used to model protein complexes involved in both fundamental biological processes and disease mechanisms, with strong potential applications in structural biology, functional genetics, and translational medicine (as reflected in collaborative publications in journals such as Nature and Cell in 2024).

In addition, the high-throughput mutagenesis approach, coupled with quantitative functional readouts, has attracted interest from industrial partners. It has been deployed in collaborative projects with the pharmaceutical industry, particularly with Sanofi, as part of their iDEA-TECH Awards program, with the aim of exploring and reprogramming therapeutically relevant protein interfaces. These collaborations open promising perspectives for the rational design of modified proteins, target screening, and mechanistic insights into biomolecular complexes involved in human diseases.

Finally, the tools developed in PPIMei have been designed for wide dissemination. A set of resources—plasmids, scripts, protocols, mutagenesis libraries, and modeling servers—have been or will be made available to the scientific community to promote reuse, extension, and broader adoption of these methods. In the medium term, this technological foundation is expected to accelerate research on complex interactomes and support new initiatives in both biotechnology and fundamental biology.

Meiosis is an evolutionarily conserved cellular program orchestrated by several machineries that recombine parental DNA to form crossovers. The PPIMei project aims to dissect the molecular interactions involved in this process using a combination of proteomics, bioinformatics, high-throughput mutagenesis screening of complex interfaces and functional assays. Detailed analysis of this pathway in S. cerevisiae will have direct implications for our understanding of equivalent processes in humans. The original methodology proposed is based on biotechnological developments using directed evolution and NGS sequencing to generate a large number of interaction mutants. These mutants will provide invaluable tools for functional characterizations and will be integrated as constraints for the development of efficient structural modeling tools.