Computational Structural Genomics Unravels Common Folds and Novel Families in the Secretome of Fungal PhytopathogenMagnaporthe oryzae

Abstract

Structural biology has the potential to illuminate the evolution of pathogen effectors and their commonalities that cannot be readily detected at the primary sequence level. Recent breakthroughs in protein structure modeling have demonstrated the feasibility to predict the protein folds without depending on homologous templates. These advances enabled a genome-wide computational structural biology approach to help understand proteins based on their predicted folds. In this study, we employed structure prediction methods on the secretome of the destructive fungal pathogen Magnaporthe oryzae. Out of 1,854 secreted proteins, we predicted the folds of 1,295 proteins (70%). We showed that template-free modeling by TrRosetta captured 514 folds missed by homology modeling, including many known effectors and virulence factors, and that TrRosetta generally produced higher quality models for secreted proteins. Along with sensitive homology search, we employed structure-based clustering, defining not only homologous groups with divergent members but also sequence-unrelated structurally analogous groups. We demonstrate that this approach can reveal new putative members of structurally similar MAX effectors and novel analogous effector families present in M. oryzae and possibly in other phytopathogens. We also investigated the evolution of expanded putative ADP-ribose transferases with predicted structures. We suggest that the loss of catalytic activities of the enzymes might have led them to new evolutionary trajectories to be specialized as protein binders. Collectively, we propose that computational structural genomics approaches can be an integral part of studying effector biology and provide valuable resources that were inaccessible before the advent of machine learning-based structure prediction. [Formula: see text]Copyright © 2021 The Author(s). This is an open access article distributed under the CC BY-NC-ND 4.0 International license .

TheVenturia inaequaliseffector repertoire is expressed in waves and is dominated by expanded families with predicted structural similarity to avirulence proteins from other plant-pathogenic fungi

“…Taken together, our study on V. inaequalis , along with previous studies on effector proteins from M. oryzae , F. oxysporum and other fungi [43, 53, 62, 117, 120], reinforce the idea that fungal effectors are often sequence-diverse, but share a limited number of structural folds. The presence of common structural folds in effectors without obvious sequence similarity could be the result of diversifying selection, where the effectors have evolved rapidly to a point where almost all amino acid sequence similarity, with the exception of residues involved in the maintenance of the overall structural fold, is lost [53].…”

“…Similarly, different ECs from C. fulvum and a necrosis-inducing effector from Cercospora beticola are predicted to adopt this fold [41, 116]. Intriguingly, through structural modelling, the ferredoxin-like fold was also found to be the most abundant fold in the M. oryzae secretome, and was predicted for the BAS4 effector protein from this fungus [117]. Altogether, this suggests that this ferredoxin/KP6-like fold may be a widely conserved structural family in different phytopathogens.…”

Comparative computational structural genomics highlights divergent evolution of fungal effectors

“…Structural similarity was considered significant only if the pair of structures were predicted with pTM scores > 0.5, and their structural similarity were measured with TM-score > 0.5 normalized for both structures. The parameters were set more stringent than the criteria used in our previous work to reduce false clustering (Seong and Krasileva, 2021). To identify the RNAse supercluster, the previously used parameters were adopted for the B. graminis secretome: E-value < 10E-4 and bidirectional coverage > 50% for sequence similarity searches, and TM scores were > 0.5 for both structures or > 0.6 and > 0.4 for each structure for structural similarity searches.…”

“…That is, a group of sequence-unrelated structurally similar effectors originated from a common ancestor but have lost detectable sequence similarity through rapid diversification. We proposed computational structural genomics as a framework to reveal such evolutionary connections obscured by sequence dissimilarity with predicted structures (Seong and Krasileva, 2021). The success of the methodology was exemplified by the identification of the MAX effector cluster, which could not be revealed by remote homology searches alone (Jones et al, 2021).…”

“…We proposed that computational structural genomics at a comparative scale would reveal novelty and commonality of effector and better elucidate effector evolution across diverse species in the fungal kingdom (Seong and Krasileva, 2021). To elucidate effector evolution at the structural level, we predicted with AlphaFold the folds of 26,653 secreted proteins from 14 agriculturally important fungal phytopathogens (Dean et al, 2012), 6 non-pathogenic fungi, and oomycete Phytophthora infestans as an outgroup.…”