Authors Andrs Zeke et al MAPK-binding linear motifs Molecular Systems Biology distribution was highly divergent between related species. A further proof for the late evolution of MAPK partnerships is found when comparing paralogs. These latter are closely related copies of the same ancestral gene that often preserved linear motifs from before their split. Most vertebrate proteins come in groups of 2, 3, or 4 closely related paralogs due to twin genome duplications–and subsequent gene loss–at the dawn of vertebrate evolution. Interestingly, most of the better-known MAPK target proteins possess a D-motif in more than one vertebrate paralogs. However, the same is not true for the majority of novel partner proteins. Comparison of vertebrate proteins with those from earlier-branching genomes also helped us to determine whether a motif developed after the gene duplications or before. Our analysis suggests that the presence of a motif in more than one paralog is predictive for ancient motif emergence. In this case, motif loss appeared to be the dominant mechanism to create differences between vertebrate paralogs. Only a very few new motifs emerged in-between the two whole-genome duplication events, suggesting that this evolutionary stage was short-lived. On the other hand, where only a single paralog contained the motif, this motif was predominantly a new invention after the twin duplications–and not a result of an ancient motif being lost. This was the most common scenario for newly found D-motifs. Many of these novel MAPK-recruiting motifs are suspected to provide a paralog -specific regulation, thereby JW 55 offering unique roles to otherwise highly similar human proteins. Having obtained a sufficient number of experimentally verified examples, we could also test some theories on the evolutionary processes creating the linear motifs. The motifs we validated could be classified based on their predicted origins. Not PubMed ID:http://www.ncbi.nlm.nih.gov/pubmed/19844694 surprisingly, the most common way of motif emergence appeared to be random mutations in an already-existing disordered segment. This can be illustrated by a known interactor, the Smoothelin-like protein 2 . Here, gradual sequence changes in terrestrial vertebrates led to the creation of the motif, which is restricted to placental mammals . There were also several examples for creation from scratch. This could mean either translational start shift or splicing site shift. While the N-terminal expansion of the protein is seen in MCL1, another newly identified partner, KSR2, serves as an example for splicing site rearrangements. Here, the paralog KSR1 retains the ancestral intronexon boundaries, which appear to have shifted in KSR2. We could even find examples for proteins where this mechanism may still be active: The motif can be included or excluded due to alternative splicing or initiation. This is the case with the PDE4 genes, where most paralogs still retain an ancestral, alternative exon containing a JIP1-type motif. Interestingly, linear motifs can also transmute into each other: Some examples in the dataset show potential switching between different MAPK-docking motif classes. As a result, distant organisms may show different motif types at the same location: That is, the JIP1-type motif that we identified in MKP5 corresponds to an NFAT4-type motif in distant organisms. In contrast, CCSER1 
has an NFAT4-type motif in humans, but a JIP1type one in zebrafish. The motif in ELK1 is of the JIP1 type in humans, but Far1 type in protosto
