The Zip4 protein directly couples meiotic crossover formation to synaptonemal complex assembly.

Meiotic recombination is triggered by programmed double-strand breaks (DSBs), a subset of these being repaired as crossovers, promoted by eight evolutionarily conserved proteins, named ZMM. Crossover formation is functionally linked to synaptonemal complex (SC) assembly between homologous chromosomes, but the underlying mechanism is unknown. Here we show that Ecm11, a SC central element protein, localizes on both DSB sites and sites that attach chromatin loops to the chromosome axis, which are the starting points of SC formation, in a way that strictly requires the ZMM protein Zip4. Furthermore, Zip4 directly interacts with Ecm11, and point mutants that specifically abolish this interaction lose Ecm11 binding to chromosomes and exhibit defective SC assembly. This can be partially rescued by artificially tethering interaction-defective Ecm11 to Zip4. Mechanistically, this direct connection ensuring SC assembly from CO sites could be a way for the meiotic cell to shut down further DSB formation once enough recombination sites have been selected for crossovers, thereby preventing excess crossovers. Finally, the mammalian ortholog of Zip4, TEX11, also interacts with the SC central element TEX12, suggesting a general mechanism.

A meiotic XPF-ERCC1-like complex recognizes joint molecule recombination intermediates to promote crossover formation.

Meiotic crossover formation requires the stabilization of early recombination intermediates by a set of proteins and occurs within the environment of the chromosome axis, a structure important for the regulation of meiotic recombination events. The molecular mechanisms underlying and connecting crossover recombination and axis localization are elusive. Here, we identified the ZZS (Zip2-Zip4-Spo16) complex, required for crossover formation, which carries two distinct activities: one provided by Zip4, which acts as hub through physical interactions with components of the chromosome axis and the crossover machinery, and the other carried by Zip2 and Spo16, which preferentially bind branched DNA molecules in vitro. We found that Zip2 and Spo16 share structural similarities to the structure-specific XPF-ERCC1 nuclease, although it lacks endonuclease activity. The XPF domain of Zip2 is required for crossover formation, suggesting that, together with Spo16, it has a noncatalytic DNA recognition function. Our results suggest that the ZZS complex shepherds recombination intermediates toward crossovers as a dynamic structural module that connects recombination events to the chromosome axis. The identification of the ZZS complex improves our understanding of the various activities required for crossover implementation and is likely applicable to other organisms, including mammals.

Genome-wide Analysis of RNA Polymerase II Termination at Protein-Coding Genes.

At the end of protein-coding genes, RNA polymerase (Pol) II undergoes a concerted transition that involves 3'-processing of the pre-mRNA and transcription termination. Here, we present a genome-wide analysis of the 3'-transition in budding yeast. We find that the 3'-transition globally requires the Pol II elongation factor Spt5 and factors involved in the recognition of the polyadenylation (pA) site and in endonucleolytic RNA cleavage. Pol II release from DNA occurs in a narrow termination window downstream of the pA site and requires the "torpedo" exonuclease Rat1 (XRN2 in human). The Rat1-interacting factor Rai1 contributes to RNA degradation downstream of the pA site. Defects in the 3'-transition can result in increased transcription at downstream genes.

Structural and functional basis of inositol hexaphosphate stimulation of NHEJ through stabilization of Ku-XLF interaction.

The classical Non-Homologous End Joining (c-NHEJ) pathway is the predominant process in mammals for repairing endogenous, accidental or programmed DNA Double-Strand Breaks. c-NHEJ is regulated by several accessory factors, post-translational modifications, endogenous chemical agents and metabolites. The metabolite inositol-hexaphosphate (IP6) stimulates c-NHEJ by interacting with the Ku70-Ku80 heterodimer (Ku). We report cryo-EM structures of apo- and DNA-bound Ku in complex with IP6, at 3.5 Å and 2.74 Å resolutions respectively, and an X-ray crystallography structure of a Ku in complex with DNA and IP6 at 3.7 Å. The Ku-IP6 interaction is mediated predominantly via salt bridges at the interface of the Ku70 and Ku80 subunits. This interaction is distant from the DNA, DNA-PKcs, APLF and PAXX binding sites and in close proximity to XLF binding site. Biophysical experiments show that IP6 binding increases the thermal stability of Ku by 2°C in a DNA-dependent manner, stabilizes Ku on DNA and enhances XLF affinity for Ku. In cells, selected mutagenesis of the IP6 binding pocket reduces both Ku accrual at damaged sites and XLF enrolment in the NHEJ complex, which translate into a lower end-joining efficiency. Thus, this study defines the molecular bases of the IP6 metabolite stimulatory effect on the c-NHEJ repair activity.

A Perspective on the Prospective Use of AI in Protein Structure Prediction.

AlphaFold2 (AF2) and RoseTTaFold (RF) have revolutionized structural biology, serving as highly reliable and effective methods for predicting protein structures. This article explores their impact and limitations, focusing on their integration into experimental pipelines and their application in diverse protein classes, including membrane proteins, intrinsically disordered proteins (IDPs), and oligomers. In experimental pipelines, AF2 models help X-ray crystallography in resolving the phase problem, while complementarity with mass spectrometry and NMR data enhances structure determination and protein flexibility prediction. Predicting the structure of membrane proteins remains challenging for both AF2 and RF due to difficulties in capturing conformational ensembles and interactions with the membrane. Improvements in incorporating membrane-specific features and predicting the structural effect of mutations are crucial. For intrinsically disordered proteins, AF2's confidence score (pLDDT) serves as a competitive disorder predictor, but integrative approaches including molecular dynamics (MD) simulations or hydrophobic cluster analyses are advocated for accurate dynamics representation. AF2 and RF show promising results for oligomeric models, outperforming traditional docking methods, with AlphaFold-Multimer showing improved performance. However, some caveats remain in particular for membrane proteins. Real-life examples demonstrate AF2's predictive capabilities in unknown protein structures, but models should be evaluated for their agreement with experimental data. Furthermore, AF2 models can be used complementarily with MD simulations. In this Perspective, we propose a "wish list" for improving deep-learning-based protein folding prediction models, including using experimental data as constraints and modifying models with binding partners or post-translational modifications. Additionally, a meta-tool for ranking and suggesting composite models is suggested, driving future advancements in this rapidly evolving field.

Molecular basis of the dual role of the Mlh1-Mlh3 endonuclease in MMR and in meiotic crossover formation.

In budding yeast, the MutL homolog heterodimer Mlh1-Mlh3 (MutLγ) plays a central role in the formation of meiotic crossovers. It is also involved in the repair of a subset of mismatches besides the main mismatch repair (MMR) endonuclease Mlh1-Pms1 (MutLα). The heterodimer interface and endonuclease sites of MutLγ and MutLα are located in their C-terminal domain (CTD). The molecular basis of MutLγ's dual roles in MMR and meiosis is not known. To better understand the specificity of MutLγ, we characterized the crystal structure of Saccharomyces cerevisiae MutLγ(CTD). Although MutLγ(CTD) presents overall similarities with MutLα(CTD), it harbors some rearrangement of the surface surrounding the active site, which indicates altered substrate preference. The last amino acids of Mlh1 participate in the Mlh3 endonuclease site as previously reported for Pms1. We characterized mlh1 alleles and showed a critical role of this Mlh1 extreme C terminus both in MMR and in meiotic recombination. We showed that the MutLγ(CTD) preferentially binds Holliday junctions, contrary to MutLα(CTD). We characterized Mlh3 positions on the N-terminal domain (NTD) and CTD that could contribute to the positioning of the NTD close to the CTD in the context of the full-length MutLγ. Finally, crystal packing revealed an assembly of MutLγ(CTD) molecules in filament structures. Mutation at the corresponding interfaces reduced crossover formation, suggesting that these superstructures may contribute to the oligomer formation proposed for MutLγ. This study defines clear divergent features between the MutL homologs and identifies, at the molecular level, their specialization toward MMR or meiotic recombination functions.

Discriminating physiological from non-physiological interfaces in structures of protein complexes: A community-wide study.

Reliably scoring and ranking candidate models of protein complexes and assigning their oligomeric state from the structure of the crystal lattice represent outstanding challenges. A community-wide effort was launched to tackle these challenges. The latest resources on protein complexes and interfaces were exploited to derive a benchmark dataset consisting of 1677 homodimer protein crystal structures, including a balanced mix of physiological and non-physiological complexes. The non-physiological complexes in the benchmark were selected to bury a similar or larger interface area than their physiological counterparts, making it more difficult for scoring functions to differentiate between them. Next, 252 functions for scoring protein-protein interfaces previously developed by 13 groups were collected and evaluated for their ability to discriminate between physiological and non-physiological complexes. A simple consensus score generated using the best performing score of each of the 13 groups, and a cross-validated Random Forest (RF) classifier were created. Both approaches showed excellent performance, with an area under the Receiver Operating Characteristic (ROC) curve of 0.93 and 0.94, respectively, outperforming individual scores developed by different groups. Additionally, AlphaFold2 engines recalled the physiological dimers with significantly higher accuracy than the non-physiological set, lending support to the reliability of our benchmark dataset annotations. Optimizing the combined power of interface scoring functions and evaluating it on challenging benchmark datasets appears to be a promising strategy.

InterEvDock3: a combined template-based and free docking server with increased performance through explicit modeling of complex homologs and integration of covariation-based contact maps.

The InterEvDock3 protein docking server exploits the constraints of evolution by multiple means to generate structural models of protein assemblies. The server takes as input either several sequences or 3D structures of proteins known to interact. It returns a set of 10 consensus candidate complexes, together with interface predictions to guide further experimental validation interactively. Three key novelties were implemented in InterEvDock3 to help obtain more reliable models: users can (i) generate template-based structural models of assemblies using close and remote homologs of known 3D structure, detected through an automated search protocol, (ii) select the assembly models most consistent with contact maps from external methods that implement covariation-based contact prediction with or without deep learning and (iii) exploit a novel coevolution-based scoring scheme at atomic level, which leads to significantly higher free docking success rates. The performance of the server was validated on two large free docking benchmark databases, containing respectively 230 unbound targets (Weng dataset) and 812 models of unbound targets (PPI4DOCK dataset). Its effectiveness has also been proven on a number of challenging examples. The InterEvDock3 web interface is available at http://bioserv.rpbs.univ-paris-diderot.fr/services/InterEvDock3/.

Proteo3Dnet: a web server for the integration of structural information with interactomics data.

Proteo3Dnet is a web server dedicated to the analysis of mass spectrometry interactomics experiments. Given a flat list of proteins, its aim is to organize it in terms of structural interactions to provide a clearer overview of the data. This is achieved using three means: (i) the search for interologs with resolved structure available in the protein data bank, including cross-species remote homology search, (ii) the search for possibly weaker interactions mediated through Short Linear Motifs as predicted by ELM-a unique feature of Proteo3Dnet, (iii) the search for protein-protein interactions physically validated in the BioGRID database. The server then compiles this information and returns a graph of the identified interactions and details about the different searches. The graph can be interactively explored to understand the way the core complexes identified could interact. It can also suggest undetected partners to the experimentalists, or specific cases of conditionally exclusive binding. The interest of Proteo3Dnet, previously demonstrated for the difficult cases of the proteasome and pragmin complexes data is, here, illustrated in the context of yeast precursors to the small ribosomal subunits and the smaller interactome of 14-3-3zeta frequent interactors. The Proteo3Dnet web server is accessible at http://bioserv.rpbs.univ-paris-diderot.fr/services/Proteo3Dnet/.

Study of the DnaB:DciA interplay reveals insights into the primary mode of loading of the bacterial replicative helicase.

Replicative helicases are essential proteins that unwind DNA in front of replication forks. Their loading depends on accessory proteins and in bacteria, DnaC and DnaI are well characterized loaders. However, most bacteria do not express either of these two proteins. Instead, they are proposed to rely on DciA, an ancestral protein unrelated to DnaC/I. While the DciA structure from Vibrio cholerae shares no homology with DnaC, it reveals similarities with DnaA and DnaX, two proteins involved during replication initiation. As other bacterial replicative helicases, VcDnaB adopts a toroid-shaped homo-hexameric structure, but with a slightly open dynamic conformation in the free state. We show that VcDnaB can load itself on DNA in vitro and that VcDciA stimulates this function, resulting in an increased DNA unwinding. VcDciA interacts with VcDnaB with a 3/6 stoichiometry and we show that a determinant residue, which discriminates DciA- and DnaC/I-helicases, is critical in vivo. Our work is the first step toward the understanding of the ancestral mode of loading of bacterial replicative helicases on DNA. It sheds light on the strategy employed by phage helicase loaders to hijack bacterial replicative helicases and may explain the recurrent domestication of dnaC/I through evolution in bacteria.

Structural Insights of the DciA Helicase Loader in Its Relationship with DNA.

DciA is the ancestral bacterial replicative helicase loader, punctually replaced during evolution by the DnaC/I loaders of phage origin. DnaC helps the helicase to load onto DNA by cracking open the hexameric ring, but the mechanism of loading by DciA remains unknown. We demonstrate by electron microscopy, nuclear magnetic resonance (NMR) spectroscopy, and biochemistry experiments that DciA, which folds into a KH-like domain, interacts with not only single-stranded but also double-stranded DNA, in an atypical mode. Some point mutations of the long α-helix 1 demonstrate its importance in the interaction of DciA for various DNA substrates mimicking single-stranded, double-stranded, and forked DNA. Some of these mutations also affect the loading of the helicase by DciA. We come to the hypothesis that DciA could be a DNA chaperone by intercalating itself between the two DNA strands to stabilize it. This work allows us to propose that the direct interaction of DciA with DNA could play a role in the loading mechanism of the helicase.

Atomic-level evolutionary information improves protein-protein interface scoring.

MOTIVATION: The crucial role of protein interactions and the difficulty in characterizing them experimentally strongly motivates the development of computational approaches for structural prediction. Even when protein-protein docking samples correct models, current scoring functions struggle to discriminate them from incorrect decoys. The previous incorporation of conservation and coevolution information has shown promise for improving protein-protein scoring. Here, we present a novel strategy to integrate atomic-level evolutionary information into different types of scoring functions to improve their docking discrimination. RESULTS: We applied this general strategy to our residue-level statistical potential from InterEvScore and to two atomic-level scores, SOAP-PP and Rosetta interface score (ISC). Including evolutionary information from as few as 10 homologous sequences improves the top 10 success rates of individual atomic-level scores SOAP-PP and Rosetta ISC by 6 and 13.5 percentage points, respectively, on a large benchmark of 752 docking cases. The best individual homology-enriched score reaches a top 10 success rate of 34.4%. A consensus approach based on the complementarity between different homology-enriched scores further increases the top 10 success rate to 40%. AVAILABILITY AND IMPLEMENTATION: All data used for benchmarking and scoring results, as well as a Singularity container of the pipeline, are available at http://biodev.cea.fr/interevol/interevdata/. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Genetic evidence for the involvement of mismatch repair proteins, PMS2 and MLH3, in a late step of homologous recombination.

Homologous recombination (HR) repairs DNA double-strand breaks using intact homologous sequences as template DNA. Broken DNA and intact homologous sequences form joint molecules (JMs), including Holliday junctions (HJs), as HR intermediates. HJs are resolved to form crossover and noncrossover products. A mismatch repair factor, MLH3 endonuclease, produces the majority of crossovers during meiotic HR, but it remains elusive whether mismatch repair factors promote HR in nonmeiotic cells. We disrupted genes encoding the MLH3 and PMS2 endonucleases in the human B cell line, TK6, generating null MLH3-/- and PMS2-/- mutant cells. We also inserted point mutations into the endonuclease motif of MLH3 and PMS2 genes, generating endonuclease death MLH3DN/DN and PMS2EK/EK cells. MLH3-/- and MLH3DN/DN cells showed a very similar phenotype, a 2.5-fold decrease in the frequency of heteroallelic HR-dependent repair of restriction enzyme-induced double-strand breaks. PMS2-/- and PMS2EK/EK cells showed a phenotype very similar to that of the MLH3 mutants. These data indicate that MLH3 and PMS2 promote HR as an endonuclease. The MLH3DN/DN and PMS2EK/EK mutations had an additive effect on the heteroallelic HR. MLH3DN/DN/PMS2EK/EK cells showed normal kinetics of γ-irradiation-induced Rad51 foci but a significant delay in the resolution of Rad51 foci and a 3-fold decrease in the number of cisplatin-induced sister chromatid exchanges. The ectopic expression of the Gen1 HJ re-solvase partially reversed the defective heteroallelic HR of MLH3DN/DN/PMS2EK/EK cells. Taken together, we propose that MLH3 and PMS2 promote HR as endonucleases, most likely by processing JMs in mammalian somatic cells.

Probing Protein Interaction Networks by Combining MS-Based Proteomics and Structural Data Integration.

Protein-protein interactions play a major role in the molecular machinery of life, and various techniques such as AP-MS are dedicated to their identification. However, those techniques return lists of proteins devoid of organizational structure, not detailing which proteins interact with which others. Proposing a hierarchical view of the interactions between the members of the flat list becomes highly tedious for large data sets when done by hand. To help hierarchize this data, we introduce a new bioinformatics protocol that integrates information of the multimeric protein 3D structures available in the Protein Data Bank using remote homology detection, as well as information related to Short Linear Motifs and interaction data from the BioGRID. We illustrate on two unrelated use-cases of different complexity how our approach can be useful to decipher the network of interactions hidden in the list of input proteins, and how it provides added value compared to state-of-the-art resources such as Interactome3D or STRING. Particularly, we show the added value of using homology detection to distinguish between orthologs and paralogs, and to distinguish between core obligate and more facultative interactions. We also demonstrate the potential of considering interactions occurring through Short Linear Motifs.

Docking proteins and peptides under evolutionary constraints in Critical Assessment of PRediction of Interactions rounds 38 to 45.

Computational structural prediction of macromolecular interactions is a fundamental tool toward the global understanding of cellular processes. The Critical Assessment of PRediction of Interactions (CAPRI) community-wide experiment provides excellent opportunities for blind testing computational docking methods and includes original targets, thus widening the range of docking applications. Our participation in CAPRI rounds 38 to 45 enabled us to expand the way we include evolutionary information in structural predictions beyond our standard free docking InterEvDock pipeline. InterEvDock integrates a coarse-grained potential that accounts for interface coevolution based on joint multiple sequence alignments of two protein partners (co-alignments). However, even though such co-alignments could be built for none of the CAPRI targets in rounds 38 to 45, including host-pathogen and protein-oligosaccharide complexes and a redesigned interface, we identified multiple strategies that can be used to incorporate evolutionary constraints, which helped us to identify the most likely macromolecular binding modes. These strategies include template-based modeling where only local adjustments should be applied when query-template sequence identity is above 30% and larger perturbations are needed below this threshold; covariation-based structure prediction for individual protein partners; and the identification of evolutionarily conserved and structurally recurrent anchoring interface motifs. Overall, we submitted correct predictions among the top 5 models for 12 out of 19 interface challenges, including four High- and five Medium-quality predictions. Our top 20 models included correct predictions for three out of the five targets we missed in the top 5, including two targets for which misleading biological data led us to downgrade correct free docking models.

InterEvDock2: an expanded server for protein docking using evolutionary and biological information from homology models and multimeric inputs.

Computational protein docking is a powerful strategy to predict structures of protein-protein interactions and provides crucial insights for the functional characterization of macromolecular cross-talks. We previously developed InterEvDock, a server for ab initio protein docking based on rigid-body sampling followed by consensus scoring using physics-based and statistical potentials, including the InterEvScore function specifically developed to incorporate co-evolutionary information in docking. InterEvDock2 is a major evolution of InterEvDock which allows users to submit input sequences - not only structures - and multimeric inputs and to specify constraints for the pairwise docking process based on previous knowledge about the interaction. For this purpose, we added modules in InterEvDock2 for automatic template search and comparative modeling of the input proteins. The InterEvDock2 pipeline was benchmarked on 812 complexes for which unbound homology models of the two partners and co-evolutionary information are available in the PPI4DOCK database. InterEvDock2 identified a correct model among the top 10 consensus in 29% of these cases (compared to 15-24% for individual scoring functions) and at least one correct interface residue among 10 predicted in 91% of these cases. InterEvDock2 is thus a unique protein docking server, designed to be useful for the experimental biology community. The InterEvDock2 web interface is available at http://bioserv.rpbs.univ-paris-diderot.fr/services/InterEvDock2/.

Characterization of the binding mode of JNK-interacting protein 1 (JIP1) to kinesin-light chain 1 (KLC1).

JIP1 was first identified as scaffold protein for the MAP kinase JNK and is a cargo protein for the kinesin1 molecular motor. JIP1 plays significant and broad roles in neurons, mainly as a regulator of kinesin1-dependent transport, and is associated with human pathologies such as cancer and Alzheimer disease. JIP1 is specifically recruited by the kinesin-light chain 1 (KLC1) of kinesin1, but the details of this interaction are not yet fully elucidated. Here, using calorimetry, we extensively biochemically characterized the interaction between KLC1 and JIP1. Using various truncated fragments of the tetratricopeptide repeat (TPR) domain of KLC1, we narrowed down its JIP1-binding region and identified seven KLC1 residues critical for JIP1 binding. These isothermal titration calorimetry (ITC)-based binding data enabled us to footprint the JIP1-binding site on KLC1-TPR. This footprint was used to uncover the structural basis for the marginal inhibition of JIP1 binding by the autoinhibitory LFP-acidic motif of KLC1, as well as for the competition between JIP1 and another cargo protein of kinesin1, the W-acidic motif-containing alcadein-α. Also, we examined the role of each of these critical residues of KLC1 for JIP1 binding in light of the previously reported crystal structure of the KLC1-TPR:JIP1 complex. Finally, sequence search in eukaryotic genomes identified several proteins, among which is SH2D6, that exhibit a motif similar to the KLC1-binding motif of JIP1. Overall, our extensive biochemical characterization of the KLC:JIP1 interaction, as well as identification of potential KLC1-binding partners, improves the understanding of how this growing family of cargos is recruited to kinesin1 by KLC1.

InterEvDock: a docking server to predict the structure of protein-protein interactions using evolutionary information.

The structural modeling of protein-protein interactions is key in understanding how cell machineries cross-talk with each other. Molecular docking simulations provide efficient means to explore how two unbound protein structures interact. InterEvDock is a server for protein docking based on a free rigid-body docking strategy. A systematic rigid-body docking search is performed using the FRODOCK program and the resulting models are re-scored with InterEvScore and SOAP-PP statistical potentials. The InterEvScore potential was specifically designed to integrate co-evolutionary information in the docking process. InterEvDock server is thus particularly well suited in case homologous sequences are available for both binding partners. The server returns 10 structures of the most likely consensus models together with 10 predicted residues most likely involved in the interface. In 91% of all complexes tested in the benchmark, at least one residue out of the 10 predicted is involved in the interface, providing useful guidelines for mutagenesis. InterEvDock is able to identify a correct model among the top10 models for 49% of the rigid-body cases with evolutionary information, making it a unique and efficient tool to explore structural interactomes under an evolutionary perspective. The InterEvDock web interface is available at http://bioserv.rpbs.univ-paris-diderot.fr/services/InterEvDock/.

Lessons from (co-)evolution in the docking of proteins and peptides for CAPRI Rounds 28-35.

Computational protein-protein docking is of great importance for understanding protein interactions at the structural level. Critical assessment of prediction of interactions (CAPRI) experiments provide the protein docking community with a unique opportunity to blindly test methods based on real-life cases and help accelerate methodology development. For CAPRI Rounds 28-35, we used an automatic docking pipeline integrating the coarse-grained co-evolution-based potential InterEvScore. This score was developed to exploit the information contained in the multiple sequence alignments of binding partners and selectively recognize co-evolved interfaces. Together with Zdock/Frodock for rigid-body docking, SOAP-PP for atomic potential and Rosetta applications for structural refinement, this pipeline reached high performance on a majority of targets. For protein-peptide docking and interfacial water position predictions, we also explored different means of taking evolutionary information into account. Overall, our group ranked 1st by correctly predicting 10 targets, composed of 1 High, 7 Medium and 2 Acceptable predictions. Excellent and Outstanding levels of accuracy were reached for each of the two water prediction targets, respectively. Altogether, in 15 out of 18 targets in total, evolutionary information, either through co-evolution or conservation analyses, could provide key constraints to guide modeling towards the most likely assemblies. These results open promising perspectives regarding the way evolutionary information can be valuable to improve docking prediction accuracy. Proteins 2017; 85:378-390. © 2016 Wiley Periodicals, Inc.

bbcontacts: prediction of ß-strand pairing from direct coupling patterns.

MOTIVATION: It has recently become possible to build reliable de novo models of proteins if a multiple sequence alignment (MSA) of at least 1000 homologous sequences can be built. Methods of global statistical network analysis can explain the observed correlations between columns in the MSA by a small set of directly coupled pairs of columns. Strong couplings are indicative of residue-residue contacts, and from the predicted contacts a structure can be computed. Here, we exploit the structural regularity of paired ß-strands that leads to characteristic patterns in the noisy matrices of couplings. The ß-ß contacts should be detected more reliably than single contacts, reducing the required number of sequences in the MSAs. RESULTS: bbcontacts predicts ß-ß contacts by detecting these characteristic patterns in the 2D map of coupling scores using two hidden Markov models (HMMs), one for parallel and one for antiparallel contacts. ß-bulges are modelled as indel states. In contrast to existing methods, bbcontacts uses predicted instead of true secondary structure. On a standard set of 916 test proteins, 34% of which have MSAs with < 1000 sequences, bbcontacts achieves 50% precision for contacting ß-ß residue pairs at 50% recall using predicted secondary structure and 64% precision at 64% recall using true secondary structure, while existing tools achieve around 45% precision at 45% recall using true secondary structure. AVAILABILITY AND IMPLEMENTATION: bbcontacts is open source software (GNU Affero GPL v3) available at https://bitbucket.org/soedinglab/bbcontacts .