Conservation and evolution of a tethering function across a viral phylogeny through the prediction of multivalent binding affinity

JULIANA GLAVINA; NICOLAS SEBASTIAN GONZALEZ FOUTEL; MATIAS SAFRANCHIK; NICOLAS GARRONE; CLARA BLANES-MIRA; GREGORIO FERNANDEZ BALLESTER; IGNACIO E SANXHEZ; ROHIT PAPPU; ALEX HOLEHOUSE; GARY DAUGHDRILL; LUCIA CHEMES

Diablerets

Conferencia; Intrinsically Disordered Proteins, GRC Conference. ?The Functional Role of Disorder in Biological Systems?; 2022

GRC

Linear motifs (LMs) are ubiquitous short sequence elements that mediate protein-protein interactions. DNA tumor viruses mimic LMs to hijack the cell cycle inducing proliferation and creating an environment that facilitates viral replication. In some cases, viral LMs hijack cell regulation by evolving higher affinities that outcompete host interactions. Viral LMs can also cooperate with other LMs or domains within the same protein through tethering, forming multivalent arrangements that effectively displace cellular complexes. A predictor that represents LMs tethered by disordered linkers would facilitate the understanding of these mechanisms. The Mastadenovirus E1A protein uses two LMs, the E2F and LxCxE motifs, tethered by a disordered linker to bind to the Retinoblastoma (Rb) protein and hijack the eukaryotic cell cycle. We used E1A as a model system to study a tethering function. In order to predict tethering across the entire Mastadenovirus phylogeny we scored the presence and absence of the E2F and LxCxE motifs in 116 E1A proteins and predicted binding affinity of individual LMs by using position specific scoring matrices obtained from the FoldX force field. We used a worm-like chain model (WLC) that describes the linker as an entropic chain to predict the effective concentration (Ceff) of linkers of different stiffness and length. Finally, we calculated the global binding affinity using the individual LM affinities and the Ceff, and mapped all features onto a phylogenetic tree representing the Mastadenovirus phylogeny.Tethering was predicted to be highly conserved across E1A proteins with linkers of 41 to 75 residues. Predictions were validated using grafting experiments with endogenous linkers and the LMs of the prototype sequence (HAdV5-E1A) and experiments with endogenous E1A proteins. Grafting experiments for five E1A proteins from Mastadenovirus infecting primates yielded a conserved Rb binding affinity, similar to that of HAdV5-E1A. The predicted binding affinity for canine and bat E1As was higher than for HAdV5-E1A but the grafing experiment, which tested functionality of the linker, showed a reduced binding affinity, suggesting that the bat linker was suboptimal. However, an E1A construct containing the endogenous motifs and linker from bat restored the high affinity, indicating a compensation between linker mutations that weaken the linker functionality and LM mutations that increase affinity for Rb. In contrast, bioinformatic analysis predicted that binding to Rb was impaired or lost due to the presence of short linkers coupled to low affinity or missing motifs in a divergent branch of adenoviruses infecting cows, pigs, rodents or treeshrews. This was validated for the shorter bovine linkers with grafting experiments. Probably the bovine linker is too short to accommodate the LMs for Rb-interaction. Altogether these results show that tethering is the main mechanism that allows optimal binding of E1A to Rb and cell cycle hijack. The results also suggest that the LMs and the linker are under co-evolutionary selection, such that either the LMs and linker are jointly optimized, or selection pressure is lost on both elements. These results agree with previous studies of our group showing that E1A-LMs evolve in a coordinated manner and concurrently with host switch evolutionary events, which suggests that coevolution of LMs and linkers underlies viral adaptation in order to maximize the competition with host interactions. Our approach provides new insights into the evolutionary patterns of IDPs, and reveals that conserved functions can be encoded within a variable disordered region by co-evolution between distant functional elements of the protein.