Use of host-like peptide motifs in viral proteins is a prevalent strategy in host-virus interactions

Abstract

Viruses interact extensively with host proteins, but the mechanisms controlling these interactions are not well understood. We present a comprehensive analysis of eukaryotic linear motifs (ELMs) in 2,208 viral genomes and reveal that viruses exploit molecular mimicry of host-like ELMs to possibly assist in host-virus interactions. Using a statistical genomics approach, we identify a large number of potentially functional ELMs and observe that the occurrence of ELMs is often evolutionarily conserved but not uniform across virus families. Some viral proteins contain multiple types of ELMs, in striking similarity to complex regulatory modules in host proteins, suggesting that ELMs may act combinatorially to assist viral replication. Furthermore, a simple evolutionary model suggests that the inherent structural simplicity of ELMs often enables them to tolerate mutations and evolve quickly. Our findings suggest that ELMs may allow fast rewiring of host-virus interactions, which likely assists rapid viral evolution and adaptation to diverse environments.

Figure 1. ELMs and viral proteins

(A) An example of a viral motif-host domain interaction: The PPxY motif of the Epstein–Barr virus LMP2 protein (in magenta) interacts with the host E3 ligase WW domain (in purple) to promote degradation of Tyr-kinases (PDB: 2JO9). (B) A non-redundant set of 2,208 viruses, in which 1,672 Eukaryotic viruses were compared to 536 prokaryotic viruses. (C) A correlation between ELM complexity (according to their information content) and their observed occurrence in total in the entire viral proteome (y = 4*10⁶x + 225; r²=0.96). P/VxxP (an SH3-domain binding motif) is an example of a simple ELM and RGD (an integrin binding motif) is an example of a complex ELM. (D) Correlations between disorder content (the total number of disordered residues in a virus) and the total number of ELM occurrences in that virus in eukaryotic (blue; y=2*10⁶x+78; r²=0.94) and prokaryotic viruses (red; y=2*10⁶x+147; r²=0.93).

Figure 2. Occurrence of ELMs that are rare in shuffled sequences

(A) The percentage of ELMs that occur in less than 0.1% of the 100,000 shuffled sequences in prokaryotic (red), eukaryotic (dark blue), animal (cyan) and plant (green) viruses. Eukaryotic viruses have significantly higher fractions of ELMs that are hard to achieve by random shuffling. (B) The distribution of ELMs in eukaryotic viruses (as a function of complexity): the entire set of ELM-matching patterns (before shuffling, in purple) and the subset of ELMs that occur in less than 0.1% of the 100,000 shuffled sequences (in blue). (C) The percentage of ELMs that occur in less than 0.1% of the shuffled sequences in two viral families and two species (3 strains of EBV and 6 strains of HCV).

Figure 3. The effects of single non-synonymous mutations on the occurrence of ELMs in HIV-1 genes

Top - 47.7% of the mutants remain with the same distribution of ELMs as occurs in the wildtype (in orange) 59% of them occur outside of ELMs regions, while 41% occur within ELMs but still conform to the wildtype ELM(cyan)). Of the remaining 52.3% mutants (left circle in gray) – which differ in their ELMs – 33.3% have a reduced number of ELMs (red, top right circle), 27.7% have an increased number of ELMs (purple, middle circle), and 32.9% have evolved a new type of ELM (violet, bottom circle), with respect to the wildtype.

Figure 4. The evolutionary origins of viral mimics

(A) Structured domain mimics can be acquired from the host through HGT (such as in the case of the cytomegalovirus MHC-I mimic m144 (PDB: 1U58, purple) that highly resembles in sequence and structure the murine homolog (PDB: 1VAC, grey)); or evolve in a convergent manner (such as the pathogen RhoGAP mimic (PDB: 1HE1, purple) that has similar activity to that of the host (PDB: 1TX4, grey) (in red – the two Arg fingers which are important for the GTPase reaction and are similarly positioned) despite of no sequence similarity. (B) Motif mimics can less frequently be acquired by HGT, such as the WH2 motif occurrence in baculoviruses that is similar in sequence and in location of other regions to host WASP (several regions and motifs are shown, based on a previous annotation(Machesky et al., 2001) Many motifs emerge in pathogens in a convergent manner, such as the RGD motif which is found on the capsid surface of various unrelated picornaviruses to support their cell entry (a schematic clade with several picornaviruses is shown; RGD-containing species appear in purple). (C) The median of similarities of human-mouse and human-virus homolog pairs in ordered and disordered regions.

Figure 5. Occurrence of multiple types of ELMs

(A) Occurrence of 17 ELM switches in various viral families (In the Y-axis, families are colored with various shades according to their replication types: blue – dsDNA, red – ssDNA, green – RT, yellow - +ssRNA, pink - -ssRNA, dark brown – dsRNA). Each of these 17 pairs (the names of the two ELMs that compose the switch are marked in the X-axis in blue and red in the bottom) occurs in several viral families as well as in their host – suggesting convergent evolution in using ELM switches by unrelated viruses and to similarity to their host. (B) EBNA-2 ELMs and their known or suggested interactions with host proteins (solid or dashed arrow respectively) (structures of the host’s interacting domains appear in matching colors). We associated identified ELMs in EBNA-2 with domains of host proteins that are known to interact with this viral protein according to a two-hybrid screening(Calderwood et al., 2007). In each case, the link was made based on the ELM type and the occurrence of a relevant domain in the host interacting proteins (e.g – an SH3-binding motif in EBNA-2 was linked to the host Endophillin-B1 which contains an SH3 domain).