Sequence co-evolution gives 3D contacts and structures of protein complexes

Abstract

Protein–protein interactions are fundamental to many biological processes. Experimental screens have identified tens of thousands of interactions, and structural biology has provided detailed functional insight for select 3D protein complexes. An alternative rich source of information about protein interactions is the evolutionary sequence record. Building on earlier work, we show that analysis of correlated evolutionary sequence changes across proteins identifies residues that are close in space with sufficient accuracy to determine the three-dimensional structure of the protein complexes. We evaluate prediction performance in blinded tests on 76 complexes of known 3D structure, predict protein–protein contacts in 32 complexes of unknown structure, and demonstrate how evolutionary couplings can be used to distinguish between interacting and non-interacting protein pairs in a large complex. With the current growth of sequences, we expect that the method can be generalized to genome-wide elucidation of protein–protein interaction networks and used for interaction predictions at residue resolution.DOI: http://dx.doi.org/10.7554/eLife.03430.001

Structure of the peptidoglycan polymerase RodA resolved by evolutionary coupling analysis

“…An MSA was generated for 9 different bitscores, a sequence inclusion threshold normalized to length and expressed as the number of bits per residue, with values ranging from the most-inclusive 0.1 to the least-inclusive 0.9. The alignment depth was chosen to optimize the number of non-redundant sequences with the fewest gaps in the alignment as described elsewhere 9,10 , although that the alignment choice was robust with respect to consistency of predicted ECs over a wide range of alignment depths. Blindly optimizing the an alignment choice resulted in two alignments with the smaller one at 31,505 non-redundant sequences of length 346 residues (10 – 355) with an “effective number” of 8,729 sequences after down-weighting sequences with more than 80% identity and no more than 30% gaps in any columns used for the EC model computation.…”

“…EVComplex 10 was used to predict inter-protein contacts between T. thermophilus full-length RodA and full-length PBP2 (Uniprot ID Q5SJ23). We constructed alignments for RodA (33670 sequences) and PBP2 (40764 sequences) as described above for RodA alone, using the April 2017 Uniprot release 42 for clarity on species identifiers.…”

“…We constructed alignments for RodA (33670 sequences) and PBP2 (40764 sequences) as described above for RodA alone, using the April 2017 Uniprot release 42 for clarity on species identifiers. We concatenated RodA and PBP2A sequences from each species when they were within 10,000 nucleotides of each other based on European Nucleotide Archive (ENA) data downloaded in February 2017 43 , and evolutionary couplings were computed on the complex alignment as previously described 44 . A mixture model approach was used to identify top-scoring contacts, both within individual monomers and between RodA and PBP2, and to assign a probability to each contact of being within the tail of the distribution of coupling scores as described previously 29 .…”

“…A mixture model approach was used to identify top-scoring contacts, both within individual monomers and between RodA and PBP2, and to assign a probability to each contact of being within the tail of the distribution of coupling scores as described previously 29 . This scoring method is accurate for predicting whether or not proteins interact as well as which residues are in contact 10 . The distribution of ECs is approximated by a Gaussian-lognormal mixture model, and we defined the tail of the distribution as those scores that have >95% probability of belonging to the lognormal component.…”

“…Cytological and protein-protein interaction studies indicate that SEDS proteins and bPBPs likely form a complex in cells 1,2 , and evolutionary coupling analysis shows strong co-variation between bPBP and SEDS protein sequences. This is sufficient to map the binding site between bPBPs and RodA to TM8 and TM9 (Figure 4a, b; Extended Data Figure 8), corresponding to the proposed interaction site between the divisome SEDS protein FtsW and its corresponding bPBP, FtsI 10,21,22 . RodA mutants in this interface exhibit a dominant negative effect in E. coli (Figure 4c).…”