Critical Nodes of Virus–Host Interaction Revealed Through an Integrated Network Analysis

Korbinian Bösl; Aleksandr Ianevski; Thoa T. Than; Petter I. Andersen; Suvi Kuivanen; Mona Teppor; Eva Zusinaite; Uga Dumpis; Astra Vitkauskiene; Rebecca J. Cox; Hannimari Kallio-Kokko; Anders Bergqvist; Tanel Tenson; Valentyn Oksenych; Magnar Bjørås; Marit W. Anthonsen; David Shum; Mari Kaarbø; Olli Vapalahti; Marc P. Windisch; Giulio Superti-Furga; Berend Snijder; Denis Kainov; Richard K. Kandasamy

To provide new and critical insights into viral evasion mechanisms we performed a comprehensive meta-analysis of the host-virus interaction landscape. We assembled the host-virus protein-protein interaction data ('hvPPI') from published studies (Fig. 1A ) (Pichlmair et al. 2012 , Rozenblatt-Rosen et al. 2012 , Jäger et al. 2011 , Shapira et al. 2009 , de Chassey et al. 2008 , Khadka et al. 2011 , Zhang et al. 2009 , Vidal et al. 2011 ). This dataset covered 17 different viruses including adeno-associated virus 5 (AAV5), dengue virus (DENV), Epstein-Barr virus (EBV), influenza A virus PR8 (IAV-PR8), influenza virus Udorn (IAV-Udorn), hepatitis C virus (HCV), human immunodeficiency virus 1 (HIV-1), human papilloma virus 5 (HPV5), human papilloma virus 6B (HPV6B), human papilloma virus 8 (HPV8), human papilloma virus 11 (HPV11), human papilloma virus 16 (HPV16), human papilloma virus 18 (HPV18), human papilloma virus 33 (HPV33), Merkel cell polyomavirus (MCPyV), Simian virus 40 (SV40) and Vaccinia virus (VACV). This dataset comprised of protein-protein interactions from two different types of experimental methods -affinity purification mass spectrometry (AP-MS) and yeast two-hybrid screens (Y2H). Altogether, this combined dataset includes 183 viral proteins, 2381 host proteins and 5781 protein-protein interactions ( Fig. 1B and Fig. S1 ). Many interactome networks including yeast and human are scale-free networks, where a large portion of the nodes (e.g. a protein in the network) have few interactions and only a few nodes have large number of interactions. The latter are often referred to as "hubs" which are crucial in keeping the network intact (Vidal et al. 2011) . We performed network topology analysis to infer the properties of the host proteins targeted by the viral proteins in the context of the human protein interactome. We considered two important parameters -relative betweenness centrality (which reflects the amount of information that passes through this protein in the human interactome) and degree (number of binding partners in the human interactome) of the host proteins targeted by each virus. The targets of all the viruses showed higher betweenness centrality and degree as compared to an average protein in the human interactome ( Fig. 1C and Fig. 1D ). This shows that viruses, by targeting "hubs" and proteins that serves as key communication nodes, have evolved the best way to disrupt the scale-free human interactome. This topological property thereby shows how viruses having small genomes achieve the maximal effect in rewiring the human interactome to benefit viral survival and replication. Our analysis is in agreement with several previous studies, which have highlighted this property (Pichlmair et al. 2012 , Rozenblatt-Rosen et al. 2012 , de Chassey et al. 2008 , Durmuş et al. 2015 , Franzosa and Xia 2011 . We propose that this could be a general principle for all viruses. preprint author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (which was not peer-reviewed) is the . https://doi.org/10.1101/548909 doi: bioRxiv preprint