Centre for Systems Biology, School of Biosciences, University of Birmingham, Edgbaston, Birmingham B15 2TT, UK.
J Mol Evol. 2010 Feb;70(2):215-31. doi: 10.1007/s00239-010-9323-5. Epub 2010 Feb 12.
Gene regulation is one important mechanism in producing observed phenotypes and heterogeneity. Consequently, the study of gene regulatory network (GRN) architecture, function and evolution now forms a major part of modern biology. However, it is impossible to experimentally observe the evolution of GRNs on the timescales on which living species evolve. In silico evolution provides an approach to studying the long-term evolution of GRNs, but many models have either considered network architecture from non-adaptive evolution, or evolution to non-biological objectives. Here, we address a number of important modelling and biological questions about the evolution of GRNs to the realistic goal of biomass production. Can different commonly used simulation paradigms, in particular deterministic and stochastic Boolean networks, with and without basal gene expression, be used to compare adaptive with non-adaptive evolution of GRNs? Are these paradigms together with this goal sufficient to generate a range of solutions? Will the interaction between a biological goal and evolutionary dynamics produce trade-offs between growth and mutational robustness? We show that stochastic basal gene expression forces shrinkage of genomes due to energetic constraints and is a prerequisite for some solutions. In systems that are able to evolve rates of basal expression, two optima, one with and one without basal expression, are observed. Simulation paradigms without basal expression generate bloated networks with non-functional elements. Further, a range of functional solutions was observed under identical conditions only in stochastic networks. Moreover, there are trade-offs between efficiency and yield, indicating an inherent intertwining of fitness and evolutionary dynamics.
基因调控是产生观察到的表型和异质性的重要机制之一。因此，研究基因调控网络（GRN）的结构、功能和进化现在构成了现代生物学的主要部分。然而，在生物进化的时间尺度上，不可能对 GRN 的进化进行实验观察。计算机进化提供了一种研究 GRN 长期进化的方法，但许多模型要么考虑了非适应性进化的网络结构，要么考虑了非生物目标的进化。在这里，我们针对 GRN 向生物量生产这一现实目标的进化提出了一些关于模型和生物学的重要问题。不同的常用模拟范例（特别是确定性和随机布尔网络），是否有基础基因表达，能否用于比较 GRN 的适应性进化和非适应性进化？这些范例以及这个目标是否足以产生一系列解决方案？生物目标和进化动态之间的相互作用是否会在生长和突变稳健性之间产生权衡？我们表明，随机基础基因表达由于能量限制迫使基因组收缩，并且是某些解决方案的先决条件。在能够进化基础表达率的系统中，观察到一个有基础表达和一个没有基础表达的两个最优值。没有基础表达的模拟范例会产生具有非功能元素的肿胀网络。此外，只有在随机网络中，才能在相同条件下观察到一系列功能解决方案。此外，效率和产量之间存在权衡，表明适应性和进化动态之间存在内在的交织。
