A cross-talk between cellular and viral proteins: functional inference by a comparative proteome approach

Poster number: 19

Iris Bahir, Michal Linial

  1. Dept of Biological Chemistry, Life Science Institute, The Hebrew University, Jerusalem, Israel

A useful antiviral agent must fulfill the following criteria: (i) it must inhibit virus specific process; (ii) it must not interfere with the host metabolism thus reducing the risk for toxicity and (iii) it typically has a restricted, well defined spectrum of activities. As of today, an integrative knowledge on any potential antiviral agent is limited to molecular biology considerations with limited contribution of system biology perspective. The information on viral sequences continues to grow with already ~1500 complete viral genomes that comprise ~16% the protein sequences archived in UniProtKB. Existing viral oriented databases are mainly focused on comparing sequences of different strains, genomic variations and mutations. Other databases focus on structural and functional information gathered for a particular viral protein. Our approach aims at developing a comprehensive integrated source of information that will eventually guide the overlap in function and interaction or viral and cellular proteins. In the long run, this will lead to the development of a specific antiviral agent by a systematic approach.


While the ultimate goal is towards human health, we initiate an unbiased approach that is applicable for all viruses and hosts (bacterial viruses are not included). We first scan for potential similarity and sequence overlap between the virus and potential hosts. Each viral protein is mapped to its correlating proteins in terms of sequence similarity. Practically, a correlating protein is one that results in a BLAST e-score<10 considering covering the query protein by sequential partition to 60 amino acid (aa) segments (with 30 aa overlap). The mapped proteins are listed separately for viral and non-viral correlating proteins. The origin of the corresponding protein is also recorded.


We demonstrate the potential contribution of our approach by refereeing to an experimentally verified analysis of human papilloma (HPV). E7 proteins were found to bind to cellular TATA box binding protein (TBP). It was also shown that the HPV-18 E6 and the HPV-16 E2 proteins bind TBP as well as P53 and pRB proteins. Sequence and structure similarity of E7 to adenovirus E1a and E2 suggest additional transcription regulators as partners including the co-activator p300. Interestingly, the binding to all of the above transcription regulators occurs by the same short C- terminal residues in the viral proteins. Mapping the correlation between E7, E2, E6, Adenovirus E1a and E2 to TBP and a p53 is expected to construct a functional network. Of course, our approach is not restricted to known viral proteins but is also valid for viral ORFs. From the protein correlated list we will focus on potential conserved residues shared by variety of related viral proteins. Incorporation of surface data to include the protein-protein interactions, structure, tissues and organelle specificity will be essential to enrich our functional attributes. The results will be a systematic database with an online interactive tool available to the community.