bioRxiv Science⌕ Search

Biology subjects

Guerois, R.

Publications and source records attributed to Guerois, R..

At least 19 recordsLinked to original sources

RNA promotes MAPK signaling through direct interaction with BRAF

RNAs are increasingly recognized as direct regulators of protein function, yet whether RNA can control the activation of signaling kinases remains poorly understood. Here we identify RNA as a direct regulator of BRAF, a central kinase in the RAS-RAF-MEK-ERK (MAPK) pathway. Using complementary crosslinking and RNA-protein interaction approaches, we show that BRAF directly binds RNA in cells and that this interaction is enhanced upon MAPK pathway activation. Depletion of intracellular RNA reduces BRAF:CRAF complex assembly and impairs MEK1/2 phosphorylation, indicating that RNA contributes to RAF complex formation and MAPK signal transduction. We identify a positively charged surface within the BRAF kinase domain that mediates RNA binding. A separation-of-function RNA-binding mutant selectively attenuates BRAF-dependent MEK1/2 phosphorylation, demonstrating that RNA binding is required for efficient MAPK signaling. Moreover, artificial recruitment of the RNA-binding-deficient BRAF mutant to an RNA scaffold restores its signaling activity, supporting a direct role for RNA-mediated BRAF complex assembly. These data identify RNA as an active component of RAF kinase regulation and establish RNA-protein interactions as an additional determinant of MAPK pathway activation.

molecular biology↗

Coordinated Assembly of WAVE and WASH complexes

WAVE and WASH polymerize Arp2/3-mediated branched actin at the leading edge of migrating cells and on the surface of endosomes, respectively. These two proteins are regulated within similar multiprotein complexes, the assembly mechanism of which remains poorly understood. Here we found using mass spectrometry that the two smallest subunits, BRK1 of the WAVE complex and CCDC53 of the WASH complex, interact with each other and with the assembly factor HSBP1. HSBP1 promotes WAVE and WASH assemblies and their respective activities, migration persistence and endosomal branched actin. Reduced levels of WAVE and WASH complexes caused by HSBP1 depletion can be corrected by providing cells with an excess of BRK1 or CCDC53 to assemble their specific complex. Unexpectedly, endogenous levels of BRK1 cross-regulate WASH assembly, whereas those of CCDC53 cross-regulate WAVE assembly. We found that various oligomers comprising BRK1, CCDC53 and HSBP1, including a heterotrimer containing one molecule of each, are formed through their promiscuous coiled coils. These oligomers play a coordinating role in the assembly of WAVE and WASH complexes.

cell biology↗

Machine Learning-Guided Engineering of High-Affinity Cross-Reactive Antibodies with Minimal Mutations

Antibodies raised against human targets often fail to recognize their animal orthologs, limiting preclinical evaluation in relevant models. We developed a Deep Mutational Scanning (DMS)-coupled deep learning strategy to engineer potent cross-reactive antibodies with minimal sequence divergence. Starting from C4, a fully human anti-PD-L1 antibody with weak recognition of murine PD-L1, DMS identified substitutions that improved binding to both human and mouse antigens. Conventional recombination of beneficial mutations generated highly cross-reactive antibodies but required 13 to 15 substitutions. To reduce this mutational burden, a deep learning model trained on DMS-derived sequence-binding data was used to identify minimal mutation combinations predicted to retain high affinity. This approach yielded variants carrying only 4 to 5 substitutions, with in vitro and cellular binding properties comparable to highly mutated antibodies. Epitope mapping, structural modeling and in vivo assessment further confirmed that these engineered antibodies retained PD-1/PD-L1 blockade and demonstrated therapeutic activity in a mouse tumor model.

molecular biology↗

Mapping the genetic landscape of the DNA damage response with Cas12a-based combinatorial knockout screens

The DNA damage response (DDR) is a complex network of cellular pathways that ensures the faithful maintenance of our genomes upon a wide array of genomic insults. To elucidate the functional architecture of this network, we conducted unbiased genetic interaction screens using the Cas12a genome editor to disrupt 233 DDR genes frequently mutated in cancer and other genetic diseases, either individually or in pairwise combinations. This approach enabled us to assess the phenotypic effects induced by the disruption of >27,000 DDR gene pair combinations under unperturbed cell growth conditions. From this analysis, we identified over 750 high-confidence positive (buffering) or negative (synthetic lethal/sick) gene-gene interactions, along with multiple connections between previously unlinked DDR pathways and modules, allowing us to define novel aspects of the cellular response to spontaneous, DNA replication-associated DNA damage. Among the identified genetic interactions, we uncovered profound synthetic lethal interactions between genes encoding 1) the translesion polymerase REV1-Pol {zeta} complex and the MCM8-MCM9-HROB DNA helicase complex; 2) Fanconi Anemia (FA) proteins and the mitotic DNA repair factors GEN1, CIP2A, and RHINO; and 3) the DNA translocase SMARCAL1 and components of the FANCM complex, suggesting novel opportunities for targeted therapies in tumors carrying mutations in these genes. Additionally, we identified robust suppressor interactions between the DCLRE1B gene encoding the nuclease APOLLO and the core non-homologous end joining (NHEJ) genes XRCC4, LIG4, and NHEJ1, suggesting that NHEJ impairs the fitness of APOLLO-deficient cells. This work provides a functional map of the DDR network and demonstrates the power of Cas12a-based screens for identifying synthetic lethal and buffering interactions with therapeutic potential.

genetics↗

Unveiling a missing component of the atypical type IV secretion system required for natural transformation of Helicobacter pylori

Exchange of genetic information by natural transformation shapes bacterial evolution. In Helicobacter pylori it is thought to drive its unusually high recombination rate, which has a crucial role in the evolution of virulence and the propagation of antibiotics resistance genes. While in most cases uptake of the incoming DNA into the periplasm is mediated by type IV pili, in H. pylori this initial step of natural transformation requires ComB, a unique competence-specific type IV secretion system (T4SS). The mechanisms by which ComB mediates DNA uptake are still poorly understood, since T4SS are usually involved in an opposite process of DNA export. Here, we identify a gene (hp1421) that is absolutely required for uptake of the transforming DNA into the periplasm, although distant from the comB operons. We show that hp1421 codes for a hexameric ATPase from the VirB11 family. HP1421 is present in the cytoplasm and interacts with ComB4, another ATPase of the T4SS inner membrane subcomplex. The structural modelling and functional analysis of HP1421 and its interaction with ComB4 indicate that HP1421 is a missing component of the ComB inner-membrane subcomplex that we propose to name ComB11. Phylogenetic analyses show that comB11 is a H. pylori core gene and suggest that the competence-dedicated ComB T4SS was a recent acquisition within Helicobacteraceae. Hence, co-option of the T4SS for DNA transformation requires nearly all the proteins that were previously essential for DNA conjugation. Author SummaryThe capacity of bacteria to exchange genetic information contributes in the case of pathogens to the spreading of antibiotic resistance and virulence factors. For Helicobacter pylori, a Gram-negative pathogen that colonises about half of the world population and is at the origin of diseases such as ulcers and gastric cancers, natural transformation is the major mechanism of horizontal gene transfer. However, H. pylori uses a very unusual system to capture and internalise the foreign DNA. Indeed, a Type 4 secretion system mediates this process. Here, we identify a so far missing and essential component of the T4SS, coded by a gene distant from the operon coding the other subunits. Through a combination of structural modelling, biochemical and microscopy approaches we show that this ATPase is an indispensable part of the ComB T4SS. Our study provides new insights into the mechanism by which the peculiar ComB T4SS works backwards to allow the passage of the tDNA from the bacterial environment into the periplasm.

microbiology↗

Human PMS1-dependent non-canonical mismatch repair converges with MBD4 to repair 5-methylcytosine deamination

CpG dinucleotides are hotspots for mutagenesis by spontaneous deamination of 5-methylcytosine (5mC) into thymine, resulting in T:G mismatches that can lead to C>T transitions. These mutations are a hallmark of aging and cancer and a major force shaping the evolution of vertebrate genomes. We have previously uncovered MBD4 as the primary base excision repair (BER) glycosylase responsible for 5mC deamination repair. In this study, we employ a cytosine base-editing system, comprising an APOBEC1 deaminase fused to a catalytically dead Cas9, to induce targeted cytosine deamination in native chromatin and track its repair. This approach reveals that MBD4 cooperates with a non-canonical branch of mismatch repair (MMR) to elicit 5mC deamination repair. We demonstrate that MBD4 activity depends on MMR complexes MutS (MSH2-MSH6) and MutL{beta} (MLH1-PMS1), but not on post-replicative MMR elicited by the MutL (MLH1-PMS2) complex. We find that PMS1 loss phenocopies the genome-wide CpG>TpG hypermutated profile associated with MBD4 deficiency, uncovering that 5mC deamination repair may represent one of the primary functions of the MutL{beta} complex. The mutational landscape of MMR-deficient tumors aligns with our experimental results, showing that replication-independent CpG>TpG mutagenesis partly contributes to the mutational burden of tumors inactivated for MLH1, MSH2 and MSH6. Finally, using structural predictions alongside biochemical validation, we show that MBD4 physically interacts with MutL{beta} in an MLH1-dependent manner, illuminating the structural basis for the convergence of the BER and MMR pathways. Altogether, we uncover a novel function of non-canonical MMR that underscores its interplay with BER in safeguarding genomic integrity against damage to methylated DNA.

molecular biology↗

It takes two (Ku) to repair: Mechanistic insights into the minimal NHEJ system of Mycobacterium Tuberculosis

In this study, we used a combination of structural simulations, ensemble assays and single molecule analysis to shed light on Non-Homologous End-Joining (NHEJ) in Mycobacterium Tuberculosis (Mtb), executed by the homodimeric Ku and Ligase D (LigD). We used a monomeric form of the Ku protein to confirm the necessity of homodimerization of Ku to bind DNA. We demonstrated that Mtb Ku and shows limited translocation on DNA and primarily instead stays bound at the DNA ends. The reconstitution of an active Mtb NHEJ machinery, consisting of Ku and LigD, allowed us to characterize the dynamics of each step of the assembly of the Mtb NHEJ machinery at the single molecule level, revealing competition between Ku and LigD for DNA binding. In silico investigations highlighted key residues in Mtb LigD - Ku complex formation. We conducted a mutational analysis of these residues in the polymerase and the ligase domains, respectively. This investigation demonstrated that, although formation of the NHEJ complex is preserved in LigD mutants, mutations in the polymerase domain result in reduced enzymatic activity, whereas mutations in the ligase domain led to its enhancement. Finally, we present a model for NHEJ in M. tuberculosis, where Ku must stay tightly at the DNA ends to properly recruit and regulate LigD enzymatic activity. GRAPHICAL ABSTRACT O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=80 SRC="FIGDIR/small/694593v1_ufig1.gif" ALT="Figure 1"> View larger version (53K): org.highwire.dtl.DTLVardef@9724ccorg.highwire.dtl.DTLVardef@f0dcc4org.highwire.dtl.DTLVardef@12df36dorg.highwire.dtl.DTLVardef@1c881ea_HPS_FORMAT_FIGEXP M_FIG C_FIG Mtb Ku and LigD compete for DNA binding, while interactions between Ku and the LigD polymerase (POL) and ligase (LIG) domains differentially modulate enzymatic activity. This image was generated with the assistance of ChatGPT and Gemini and subsequently edited by the authors.

biochemistry↗

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

Characterizing mutation effects on protein-protein interactions (PPIs) is crucial for elucidating protein structure and function. Massively parallel PPI variant analyses such as deep mutational scanning (DMS) enable interface identification and generate datasets for machine learning. In cellulo strategies such as two-hybrid systems provide straightforward access to such data, but reliability depends on quantitative properties. Here, we show that existing bacterial two-hybrid (B2H) systems have limitations constraining accurate dataset generation. We engineered and benchmarked optimized quantitative B2H (qB2H) alternatives, enabling strain-independent assays, improved metrics, and generation of high-quality datasets. We demonstrate qB2H utility through interface mapping and binder optimization. Perturbation analysis of single-site variants accurately recovered known ASF1 complex contact positions, matching crystallographic data. Integration of generative AI-based design yielded an ASF1-binding peptide with a 70-fold increase in affinity. qB2H offers to R&D scientists a robust, reusable platform for quantitative PPI analysis, enabling both rational protein engineering and data-driven discovery.

synthetic biology↗

USP7 deubiquitinase stabilizes FAN1 to support DNA crosslink repair and suppress CAG repeat expansion

Human FAN1 is a structure-specific endonuclease critical for the repair of DNA interstrand crosslinks (ICLs) and the excision of extrahelical CAG repeats-whose pathological expansion underlies Huntingtons disease (HD), a progressive and currently incurable neurodegenerative disorder. However, mechanisms of post-translational regulation of FAN1 are still largely unknown. Here, we identify the ubiquitin-specific protease 7 (USP7) as new interactor of FAN1. USP7 stabilizes FAN1 protein levels in a deubiquitination-dependent manner, preventing FAN1 from proteasomal degradation. Consequently, we demonstrate that USP7 depletion leads to reduced chromatin association of FAN1 and increased cellular hypersensitivity following ICL damage. Moreover, we find that loss of USP7 accelerates CAG repeat expansion in an HD cellular model. Collectively, our findings establish USP7 as a critical regulator of FAN1 activity in the maintenance of genome stability, highlighting potential therapeutic opportunities for cancer and HD.

molecular biology↗

CARM1-mediated methylation controls interactions of ALIX with key partners important for cytokinesis

CARM1 is an arginine methyltransferase with a well-established role in regulating gene expression, but its cytoplasmic functions remain largely uncharacterized. Here, we identified ALIX, a protein acting with the ESCRT-III machinery in numerous membrane remodeling events, as a main cytoplasmic partner and a relevant substrate of CARM1. We demonstrate that CARM1 methylates arginine residues within the proline-rich motif of ALIX. At the molecular level, ALIX methylation impairs SH3-dependent association with CD2AP, CIN85, and endophilin-A2 that are required for cytokinesis. Using a mutant of ALIX that is unable to bind these proteins, we further show that these interactions are essential for ALIX functions during cytokinesis. Altogether, this work highlights the significance of arginine methylation as a regulatory post-translational modification important for the final step of cell division, by modulating interactions between proline-rich domains and their SH3-containing partners.

cell biology↗

A tripartite protein complex promotes DNA transport during natural transformation in firmicutes.

Natural genetic transformation is a conserved mechanism of bacterial horizontal gene transfer, which is directed entirely by the recipient cell and facilitates the acquisition of new genetic traits such as antibiotic resistance. Transformation proceeds via the capture of exogenous DNA, its internalisation in single strand form (ssDNA) and its integration into the recipient chromosome by homologous recombination. While the proteins involved in these steps have mainly been identified, the specific mechanisms at play remain poorly characterised. This study takes advantage of recent advances in structural modelling to explore the uptake of ssDNA during transformation. Using the monoderm human pathogen Streptococcus pneumoniae, we model a tripartite protein complex composed of the transmembrane channel ComEC, and two cytoplasmic ssDNA-binding proteins ComFA and ComFC. Using targeted mutation and transformation assays, we propose that pneumococcal ComEC features a narrow channel for ssDNA passage, and we show this channel is conserved in the diderm Helicobacter pylori. We identify key residues involved in protein-protein and protein-ssDNA interactions in the pneumococcal tripartite complex model and we show them to be crucial for transformation efficiency. Structural modelling reveals that this tripartite protein complex and its interaction with ssDNA are conserved in firmicutes. Overall, this study validates a tripartite complex required for the internalisation of ssDNA during transformation in firmicutes, providing new insights into the molecular mechanisms involved in this horizontal gene transfer mechanism central to bacterial adaptation. It also demonstrates the power of recent structural modelling techniques such as AlphaFold3 as hypothesis generators and guides for designing experiments. Significance statementNatural genetic transformation is a key mechanism of horizontal gene transfer, conserved in bacteria. The investigation of transmembrane channel ComEC and its interaction partners was previously hindered by the difficulty of manipulating these proteins experimentally. Thanks to state-of-the-art structural modelling with AlphaFold3 and subsequent thorough experimental validation of this model using targeted mutation and transformation assays, we demonstrate the importance of the ComEC/ComFA/ComFC complex for single strand DNA (ssDNA) uptake during natural transformation in the human pathogen Streptococcus pneumoniae. Similar models in several other species suggest a widely conserved organization of this complex in firmicutes. In addition, we demonstrate that the ComEC transmembrane channel is also crucial for ssDNA uptake during natural transformation in Helicobacter pylori.

microbiology↗

Mechanistic Insights into Proteomic Mutation-Phenotype Linkages from Tiling Mutagenesis Screens

High-throughput mutagenesis screens are powerful tools for mapping mutations to phenotypes. However, deciphering the molecular mechanisms that link mutations to phenotypic outcomes remains a significant challenge. Here, we present ProTiler-Mut, a versatile computational framework that harnesses tiling mutagenesis screens, which introduce variants across entire protein sequences, to facilitate investigation of mutation-to-phenotype associations at multiple levels, including individual residues, protein substructures, and protein-protein interactions (PPIs). As demonstrated through our analyses of base editing (BE) screens targeting DNA Damage Response (DDR) proteins and T cell regulators, ProTiler-Mut provides novel insights into the mutation-phenotype linkages, including: i) refined classification of mutation that reveals separation-of-function (SOF) category beyond the conventional binary classification of loss-of-function (LOF) and gain-of-function (GOF); ii) definition of phenotype-associated hotspot substructures that enable the inference of the function of unscreened pathogenic mutations; and iii) identification of phenotype-associated PPIs disrupted by functional mutations. Through ProTiler-Mut analyses, we identified a substructure harboring pathogenic GOF mutations that disrupt interactions between the kinases MAPK1 and RSK1, leading to MAPK1 activation and elevated expression of the immune checkpoint receptor PD-1. Furthermore, we demonstrate the applicability of ProTiler-Mut to various mutagenesis screening platforms, highlighting its broad utility and generalizability.

bioinformatics↗

Rad52 Acts as an Assembly Chaperone to Form and Stabilize Rad51 Filaments Through a Large C-Terminus 85-Residue Segment

Homologous recombination (HR) is essential for the repair of DNA double-strand breaks and the restart of stalled replication forks. A critical step in HR is the formation of Rad51 nucleofilaments, which perform homology search and strand invasion of a homologous DNA sequence required for repair synthesis. In the yeast Saccharomyces cerevisiae, Rad52 facilitates Rad51 nucleofilament formation by mediating Rad51 loading onto ssDNA and counteracting the dissociation of Rad51 filaments by the DNA translocase Srs2. The molecular basis of these two Rad52 functions remains unclear. Our integrative structural analyses of the Rad51-Rad52 interaction, combining NMR, SAXS, and modeling, reveal that an 85-residue segment of Rad52, conserved in fungi, folds upon binding to a broad surface of a Rad51 monomer. Notably, it includes an FxxA motif conserved in the BRC repeats of BRCA2 and at the Rad51-Rad51 interface. This binding mode was validated through an extensive set of mutations. Using in vivo assays and a functional fluorescent GFP-Rad51 fusion protein, we demonstrated that this entire segment is critical for Rad51 filament formation. These findings highlight how Rad52 functions as an assembly chaperone, preventing Rad51 oligomerization, promoting nucleation of Rad51 nucleofilaments on ssDNA, and protecting Rad51 filaments from destabilization by Srs2.

molecular biology↗

DprA recruits ComM to facilitate recombination during natural transformation in Gram-negative bacteria

Natural transformation (NT) represents one of the major modes of horizontal gene transfer in bacterial species. During NT, cells can take up free DNA from the environment and integrate it into their genome by homologous recombination. While NT has been studied for >90 years, the molecular details underlying this recombination remain poorly understood. Recent work has demonstrated that ComM is an NT-specific hexameric helicase that promotes recombinational branch migration in Gram-negative bacteria. How ComM is loaded onto the post-synaptic recombination intermediate during NT, however, remains unclear. Another NT-specific recombination mediator protein that is ubiquitously conserved in both Gram-positive and Gram-negative bacteria is DprA. Here, we uncover that DprA homologs in Gram-negative species contain a C-terminal winged helix domain that is predicted to interact with ComM by AlphaFold. Using Helicobacter pylori and Vibrio cholerae as model systems, we demonstrate that ComM directly interacts with the DprA winged-helix domain, and that this interaction is critical for DprA to recruit ComM to the recombination site to promote branch migration during NT. These results advance our molecular understanding of recombination during this conserved mode of horizontal gene transfer. Furthermore, they demonstrate how structural modeling can help uncover unexpected interactions between well-studied proteins to provide deep mechanistic insight into the molecular coordination required for their activity. SIGNIFICANCE STATEMENTBacteria can acquire novel traits like antibiotic resistance and virulence through horizontal gene transfer by natural transformation. During this process, cells take up free DNA from the environment and integrate it into their genome by homologous recombination. Many of the molecular details underlying this process, however, remain incompletely understood. In this study, we identify a new protein-protein interaction between ComM and DprA, two factors that promote homologous recombination during natural transformation in Gram-negative species. Through a combination of bioinformatics, structural modeling, cell biological assays, and complementary genetic approaches, we demonstrate that this interaction is required for DprA to recruit ComM to the site of homologous recombination.

microbiology↗

Molecular insights into the activation of Mre11-Rad50 endonuclease activity by Sae2/CtIP

In S. cerevisiae, the Mre11-Rad50-Xrs2 (MRX)-Sae2 nuclease activity is required for the resection of DNA breaks with secondary structures or protein blocks, while in humans, the MRE11-RAD50-NBS1 (MRN) homologue with CtIP is needed to initiate DNA end resection of all breaks. Phosphorylated Sae2/CtIP stimulates the endonuclease activity of MRX/N. Structural insights into the activation of the Mre11 nuclease are available only for organisms lacking Sae2/CtIP, so little is known how Sae2/CtIP activates the nuclease ensemble. Here, we uncover the mechanism of Mre11 activation by Sae2 using a combination of AlphaFold2 structural modeling, biochemical and genetic assays. We show that Sae2 stabilizes the Mre11 nuclease in a conformation poised to cleave substrate DNA. Several designs of compensatory mutations establish how Sae2 activates MRX in vitro and in vivo, validating the structural model. Last, our study uncovers how human CtIP, despite considerable sequence divergence, employs a similar mechanism to activate MRN.

biochemistry↗

Mechanism of DNA unwinding by hexameric MCM8-9 in complex with HROB

The human MCM8-9 helicase functions in concert with HROB in the context of homologous recombination, but its precise function is unknown. To gain insights into how HROB regulates MCM8-9, we first used molecular modeling and biochemistry to define their interaction interface. We show that HROB makes important contacts with both MCM8 and MCM9 subunits, which directly promotes its DNA-dependent ATPase and helicase activities. MCM8-9-HROB preferentially binds and unwinds branched DNA structures, and single-molecule experiments reveal a low DNA unwinding processivity. MCM8-9 unwinds DNA as a hexameric complex that assembles from dimers on DNA in the presence of ATP, which is prerequisite for its helicase function. The hexamer formation thus involves two repeating protein-protein interfaces forming between the alternating MCM8 and MCM9 subunits. One of these interfaces is rather stable and forms an obligate heterodimer, while the other interface is labile and mediates the assembly of the hexamer on DNA, independently of HROB. The ATPase site composed of the subunits forming the labile interface disproportionally contributes to DNA unwinding. HROB does not affect the MCM8-9 ring formation, but promotes DNA unwinding downstream by possibly coordinating ATP hydrolysis with structural transitions accompanying translocation of MCM8-9 on DNA.

biochemistry↗

Disordered regions and folded modules in CAF-1 promote histone deposition in S. pombe

Genome and epigenome integrity in eukaryotes depends on the proper coupling of histone deposition with DNA synthesis. This process relies on the evolutionary conserved histone chaperone CAF-1 for which the links between structure and functions are still a puzzle. While studies of the S. cerevisiae CAF-1 complex enabled to propose a model for the histone deposition mechanism, we still lack a framework to demonstrate its generality and in particular, how its interaction with the polymerase accessory factor PCNA is operating. Here, we reconstituted a complete SpCAF-1 from fission yeast. We characterized its dynamic structure using NMR, SAXS and molecular modeling together with in vitro and in vivo functional studies on rationally designed interaction mutants. Importantly, we identify the unfolded nature of the acidic domain which folds up when binding to histones. We also show how the long KER helix mediates DNA binding and stimulates SpCAF-1 association with PCNA. Our study highlights how the organization of CAF-1 comprising both disordered regions and folded modules enables the dynamics of multiple interactions to promote synthesis-coupled histone deposition essential for its DNA replication, heterochromatin maintenance, and genome stability functions.

biochemistry↗

From interaction networks to interfaces: Scanning intrinsically disordered regions using AlphaFold2

The revolution brought about by AlphaFold2 and the performance of AlphaFold2-Multimer open promising perspectives to unravel the complexity of protein-protein interaction networks. Nevertheless, the analysis of interaction networks obtained from proteomics experiments does not systematically provide the delimitations of the interaction regions. This is of particular concern in the case of interactions mediated by intrinsically disordered regions, in which the interaction site is generally small. Using a dataset of protein-peptide complexes involving intrinsically disordered protein regions that are non-redundant with the structures used in AlphaFold2 training, we show that when using the full sequences of the proteins involved in the interaction networks, AlphaFold2-Multimer only achieves 40% success rate in identifying the correct site and structure of the interface. By delineating the interaction region into fragments of decreasing size and combining different strategies for integrating evolutionary information, we managed to raise this success rate up to 90%. Beyond the correct identification of the interaction site, our study also explores specificity issues. We show the advantages and limitations of using the AlphaFold2 confidence score to discriminate between alternative binding partners, a task that can be particularly challenging in the case of small interaction motifs.

bioinformatics↗