The interplay between genome and network evolution in eukaryotes

Abstract

We assume that the functional relations that a protein engages in, influences it’s evolutionary dynamics. An extreme case is a functional module in which all components are strongly dependent on each other, for example a protein complex or metabolic pathway. If a protein’s function is only relevant on the level of the module, selection will act on module level as well. This leads to distinct phylogenetic patterns: either the module is completely present or completely absent. In other words, the functional module is also an evolutionary module. Surprisingly, we find that most functional modules do not behave like evolutionary modules. In order to remove any biases resulting from erroneous module definitions, we filter our functional modules using high-throughput protein interaction data as well as by cross-comparing different module datasets. We observe only a minor increase in evolutionary modularity. Without strong module-level selection, how is the modular structure of the cellular organization maintained? To answer this question, we represent the molecular machinery as a network of interacting proteins and infer how this network may have evolved. We use an existing model in which networks grow through gene duplication followed by subfunctionalization, to simulate network evolution. Networks that are produced by this model, have a modular structure that is similar to that of real protein interaction networks. We increase the biological relevance of this model by incorporating gene loss. We find that in this extended model, modularity is more easily established and maintained. Importantly, we find that if duplication of existing interactions is the only source of new connections, networks tend to break up. This strongly suggests that gain of completely novel interactions is pivotal in network evolution. In addition to comparing purely topological characteristics of model networks with those of real large-scale protein interaction networks, we study how proteins of different ages are connected in different networks. Previous studies report that protein tend to interact with priteins of a similar age and we demonstrate that this can be partially explained by the process of gene duplcation followed by subfunctionalization. In conclusion, we find that the organization of proteins in specific modules is not conserved in evolution: functional modules do not necessarily behave as evolutionary modules. Moreover, we find that functional modules can spontaneously arise from basic, local evolutionary processes, such as gene duplication.