We have constructed a computational model of the open-ended evolution of a transcriptional regulatory network. The model uses parameters taken from real data from Saccharomyces cerevisiae, and captures intrinsic stochasticity and delays. It captures two key features of mutation bias: ease of evolving weak-affinity transcription factor binding sites that can subsequently evolve greater affinity, and the network patterns introduced through the process of gene duplication.
We have used the model to study adaptive and non-adaptive hypotheses for the evolution of network motifs such as the two over-represented types of feed forward loops (C1-FFLs and I1-FFLs). We reproduced adaptive evolution of C1-FFLs for their purported function of filtering out short spurious signals, but our results emphasized that dynamics are more important than topology in achieving this function (Xiong et al. 2019). C1-FFLs use an AND-gate to integrate information from slow and fast signalling pathways, in order to filter out short spurious signals. However, C1-FFLs are not the only motif to evolve - anything that combines a slow and fast pathway will work. In the real world, because transcriptional regulation is intrinsically slow, this dynamic module is more likely to arise from combining transcriptional with non-transcriptional (e.g. post-translational) regulation.
Both I1-FFLs and negative feedback loops (NFBLs) can generate short, sharp pulses of expression in response to a signal. We found that both motifs are equally capable of solving the problem, and that which motif evolves is determined by their evolutionary accessibility (Xiong et al. 2021). When I1-FFLs are initially more accessible, NFBLs generally evolve via a network that combines both motifs at once (Xiong et al. 2021).
More broadly, biological robustness is often distributed across a complex network, rather than being a simple consequence of redundant parts. We think networks are constrained not only by what they need to do, but also by the threat from pervasive non-specific interactions. Complex interacting networks can act as evolutionary capacitors by concealing and revealing variation, and in the process can turn “wasteful mess” into a creative force in evolution (Masel 2004, Masel & Siegal 2009).