A Structure-Informed Atlas of Human-Virus Interactions

Abstract

While knowledge of protein-protein interactions (PPIs) is critical for understanding virus-host relationships, limitations on the scalability of high-throughput methods have hampered their identification beyond a number of well-studied viruses. Here, we implement an in silico computational framework (pathogen host interactome prediction using structure similarity [P-HIPSTer]) that employs structural information to predict ∼282,000 pan viral-human PPIs with an experimental validation rate of ∼76%. In addition to rediscovering known biology, P-HIPSTer has yielded a series of new findings: the discovery of shared and unique machinery employed across human-infecting viruses, a likely role for ZIKV-ESR1 interactions in modulating viral replication, the identification of PPIs that discriminate between human papilloma viruses (HPVs) with high and low oncogenic potential, and a structure-enabled history of evolutionary selective pressure imposed on the human proteome. Further, P-HIPSTer enables discovery of previously unappreciated cellular circuits that act on human-infecting viruses and provides insight into experimentally intractable viruses.

Keywords: evolution; immunology; protein structure; protein-protein interactions; systems biology; virology.

Figure 1.. P-HIPSTer enables human-virus interactome mapping and interrogation.

A) P-HIPSTer uses protein structure homology modeling to evaluate viral-human PPIs mediated by domain-domain or peptide domain contacts. B) empirical validation by co-IP of 65 predicted viral-human PPIs. Positive and negative interactions are shown in yellow and purple, respectively (TP: true positive rate; TN: true negative rate). C) Posterior analyses leveraging P-HIPSTer human-virus protein interactome.

Figure 2.. Discovery of novel ZIKV-human protein interactions.

A) Combining P-HIPSTer predictions with known human PPIs followed by topological analysis identifies connectivity-based modules that are subjected to functional interrogation. B) ZIKV-human PPI network with enriched biological pathways and phenotypes for each topological module. C) Interaction models and experimental validation for 7 predicted ZIKV-human PPIs (blue and red models correspond to human and viral proteins, respectively).

Figure 3.. Functional interrogation of ZIKV cellular partners and identification of ESR1 as an inhibitor of viral replication.

A) Effect of overexpressing cellular factors on ZIKV infection in 293T cells with or without IFNβ priming. Stars indicate significant difference in viral titer (determined by focus forming assay) or cellular response to IFNβ. B-C) Effect of ESR1 knockdown (siRNA) on ZIKV replication. B) Focus forming assay (red staining indicates ZIKV foci). C) qPCR of ZIKV mRNA, data representative of 3 independent experiments. One and two stars indicates p-values <0.05 and <0.0001 respectively. (EV; empty vector)

Figure 4.. P-HIPSTer derived Bayesian Network classifier discriminates high- and low-risk HPVs.

A) Machine learning on P-HIPSTer interactomes for HR and LR HPVs (highlighted in ‘B’ with red and blue rectangles, respectively) is used to identify features associated with viral oncogenic potential. B) Hierarchical clustering of alpha HPVs based on a constellation of 10 viral-host PPIs (5 associated with HR, 5 associated with LR; Group I and Group II, respectively) discriminates HR and LR HPVs. Group III describes 18 viral-host PPIs shared across alpha HPVs. Dark and open circles denote binding profiles (LR ≥ 100: dark circle; LR < 100 open circle). Human proteins with known roles during HPV infection are highlighted in red.

Figure 5.. P-HIPSTer reveals functional and evolutionary relationships across the human virome.

A) Strategy to cluster viruses, described by their set of enriched pathways. B) Dendrogram of 568 viruses clustered based on their enriched pathways. Inner color ring specifies Baltimore category; outer ring specifies viral family for each virus. Schematic representations of the Interleukin 10 (IL-10; enriched in Cluster 8), Rho GTPase signaling (enriched in cluster 7), and RNA metabolism (enriched in cluster 5) pathways. Pathway components predicted to be targeted by viruses within each cluster are highlighted in blue.

Figure 6.. P-HIPSTer reveals history of selection imposed by human viruses.

A) Evolutionary tree derived from 12 whole genome primate sequences. B) ΔN/ΔS values for each of 14,974 aligned genes were plotted across human-virus interacting and non-interacting proteins. Listed in the insert are virus-interacting proteins with ΔN/ΔS values >1. Shown are functional enrichments (p-value<0.05 and FDR<20%) for interacting (dark circle) and non-interacting (light circle) proteins with ΔN/ΔS values >1. Red dot denotes p-value that did not pass FDR correction. Stars indicate Wilcoxon p-value<0.0001, W>10⁶.