Surface frustration re-patterning underlies the structural landscape and evolvability of fungal orphan candidate effectors

Abstract

Pathogens secrete effector proteins to subvert host physiology and cause disease. Effectors are engaged in a molecular arms race with the host resulting in conflicting evolutionary constraints to manipulate host cells without triggering immune responses. The molecular mechanisms allowing effectors to be at the same time robust and evolvable remain largely enigmatic. Here, we show that 62 conserved structure-related families encompass the majority of fungal orphan effector candidates in the Pezizomycotina subphylum. These effectors diversified through changes in patterns of thermodynamic frustration at surface residues. The underlying mutations tended to increase the robustness of the overall effector protein structure while switching potential binding interfaces. This mechanism could explain how conserved effector families maintained biological activity over long evolutionary timespans in different host environments and provides a model for the emergence of sequence-unrelated effector families with conserved structures.

Fig. 1. Structural relationship between orphan candidate effectors in 20 fungal genomes.

A Bioinformatics pipeline for the systematic analysis of structures of orphan candidate effectors (OCEs) in fungal genomes. Filtering and analysis steps are shown as gray boxes with the number of retained proteins indicated between brackets. Analytical tools are indicated in blue. Out of 227,823 predicted proteins, 20,451 had a secretion signal, 4987 we selected for structure prediction, 3925 had a reliable structure predicted. B Fungal species selected to cover the phylogenetic and lifestyle diversity of the Pezizomycotina. For each proteome, the distribution of secreted proteins with different characteristics based on the pipeline is indicated as a histogram. C OCE structural similarity network with vertices colored according to detected communities forming structure-related families. The 12 major families including ≥54 proteins are numbered by decreasing protein count. D Representative predicted structures corresponding to the 12 major families. Numbered and colored stickers refer to communities shown in (C). Residues are colored according to their pLDDT score. The number of OCEs and species in which they occur (sp.) is indicated. E Contribution of fungal species classified by lineage, host type and lifestyle to each of the 62 major families. Each circle represents one OCE fold, sized according to the number of proteins it contains. The 12 major families are colored as in (D). Source data are provided as a Source data file.

Fig. 2. The extant sequence and structural diversity in major OCE families reveal stable cores with flexible surface-exposed regions.

A Structural similarity (Dali Z-score) according to evolutionary distance (Jukes-Cantor corrected distance calculated on curated sequence alignments) for histone 2A and the OCE families Alt-A1 and KP4 toxins. The green dotted lines are the local average. Pairwise structure comparisons falling outside of the 97.5% prediction interval (PI, gray ribbons) are shown as red dots. Numbers in brackets indicate the number of protein structures compared. The percent of proteins with Z values above the predicted interval is indicated for each family. B Structure similarity neighbour-joining trees for members of the Alt-A1 and BoNT families, with representative structures shown as red circles. C Structural alignment of the corresponding representative proteins colored according to the residue relative exposure. D Relative residue exposure according to local structural alignment accuracy (RMSD, in Angstrom). Gray ribbons show LOESS regression-centered 50% prediction intervals. Dots are colored according to residue conservation. E The overall diversity of the Alt-A1 family obtained through sensitive HMM comparisons of homologs from the NCBI nr database. Each circle corresponds to a cluster of 1 to 169 sequences, with homology relationships shows as edges. The 13 superclusters are numbered and shown in different colors. Superclusters 1 to 7 are connected by homology to a single member. Source data are provided as a Source data file.

Fig. 3. Mapping of robustness-increasing mutations based on natural diversity and in silico mutation scans in clades of KP6 and Alt-A1 OCEs.

A Reconstruction of OCE ancestral structure in KP6 clade 43 and Alt-A1 clade 480. The phylogenetic trees used for ancestral sequence reconstruction show modern OCEs as black circles and ancestral nodes used in structure prediction as red circles. The Alphafold models of the common ancestors are shown in rainbow colors. B Sequence conservation (%) in modern OCES mapped onto the protein structure of their common ancestor. C Mapping of clusters of residues significantly co-mutated onto ancestral protein structures. D Structural divergence (SD) caused by extant natural variants (‘Modern’) and artificial mutations in ancestral OCEs. Structural divergence between variant OCEs is expressed relative to ancestral OCE self-alignment score. Boxplots show median values (thick line), first and third quartile values (box) and 1.5 times the interquartile range (whiskers). E Mapping of residues net SD effect onto the structure of the ancestral OCE. Net SD corresponds to the difference between SD obtained after the mutation/deletion of individual residues and SD obtained after multiple simultaneous mutations. F Relationship between residues SD effect (Y-axis) and their probability of co-selection (X-axis, as log10 of p-value). Residues are colored as in (C) according to the cluster they belong to. The dotted line shows local average, the gray area corresponds to the 95% confidence interval. Source data are provided as a Source data file.

Fig. 4. Re-patterning of surface residue frustration buffers structural variation in OCEs.

A Frustration index negatively correlates with relative surface exposure in Alt-A1, KP6, and GNK2/KP4 OCE families. Frustration index decreases with the degree of residue frustration. The red line shows loess regression, gray area is the 90% confidence interval, dotted lines delimit minimally (0.78) and highly (−1.0) frustrated thresholds. Data points are colored according to Cα structural deviation from predicted ancestral structure (RMSD, root mean square deviation, in Angstroms). B Comparison of relative surface exposure for residue classified as neutrally/minimally frustrated and highly frustrated in Alt-A1, KP6, and GNK2/KP4 OCE families. Boxplots show median values (thick line), first and third quartile values (box) and 1.5 times the interquartile range (whiskers). C Frustration index variance in clades from four OCE families mapped on the structure of the reconstructed clade ancestor. D Frustration variation (red, left side) and structure divergence (SD, blue, right side) rates during the evolution of OCEs from an Alt-A1 and a KP6 clade mapped on time-calibrated phylogenies. N is the number of modern OCEs per group, r is Pearson’s product-moment correlation between frustration variation and structural divergence across all branches of the tree. E Frustration variation (red) and structural divergence (blue) over time across every possible evolutionary path in trees shown in (D). Lines show mean value from a loess regression, shaded areas are the 95% confidence interval. F Reconstructed evolution of structure and residue frustration in two lineages of members of the Alt-A1 family. Each protein structure is shown as molecular surface (top) and ribbon diagram (bottom), with the reconstructed ancestral protein at the top of the panel and two of its modern descendants at the bottom. Z, structural similarity score; ΣFI, sum of frustration index for all residues in the protein. Black dotted lines delimit patches of increased frustration, black stars indicate the position of highly variable loops. G Frustration variance for residues showing positive (green) and negative (red) SD effect in mutational scans of two OCE groups. Mya, Million years ago. Boxplots show median values (thick line), first and third quartile values (box) and 1.5 times the interquartile range (whiskers). Source data are provided as a Source data file.