Towards an engineering theory of evolution

Abstract

Biological technologies are fundamentally unlike any other because biology evolves. Bioengineering therefore requires novel design methodologies with evolution at their core. Knowledge about evolution is currently applied to the design of biosystems ad hoc. Unless we have an engineering theory of evolution, we will neither be able to meet evolution's potential as an engineering tool, nor understand or limit its unintended consequences for our biological designs. Here, we propose the evotype as a helpful concept for engineering the evolutionary potential of biosystems, or other self-adaptive technologies, potentially beyond the realm of biology.

Fig. 1. The evotype and its key properties.

a The evotype visualised as a landscape surrounding the design type (red square), where fitneity (the combined function of fitness and utility) is plotted as a vertical axis against a 2D plane of sequence space with the probability of evolution exploring regions of sequence space overlaid in grey. The properties of this landscape are determined by the interaction of three components: variation, function and selection. b A variation probability distribution can be projected onto sequence space, which represents the likelihood of exploring a given sequence through genetic variation. Darker regions represent regions of higher probability. This is the sum of the distributions of the individual variation operators present in the system (variation operator set). For example, point mutation (bottom layer in set), recombination of homologous regions (middle layer in set), and slip-strand mutation (top layer in set). Red arrows in the middle and bottom layers represent algorithmic and point mutations, respectively. c How phenotypic functions are distributed in sequence space surrounding the design type is critical. Function space may be considered as discrete (top), where the space may have high genotypic robustness (left grid) or high variability (right grid). A continuous utility space (bottom) plotted against a 1D projection of sequence space. The colour under the curve represents the discrete function associated with that region of sequence space and the utility that each has as a continuous value. For example, if the goal is to produce blue-like functions, dark blue may have the highest utility, followed by lighter variants in the spectrum. The bioengineer must define a minimal threshold (dashed line), below which the design is deemed to be a failure (e.g., non-desired function is exhibited). d Sequences differ in their reproductive fitness. This is the driver of natural selection and can be plotted across sequence space as an adaptive landscape (red dotted line). Utility (blue dashed line) may or may not correlate with reproductive fitness across sequence space. The fitneity (grey solid line) is the combination of the fitness and utility. Bioengineers must optimise fitneity both for the design type and throughout the landscape.

Fig. 2. Engineering evotypes by sculpting their landscapes.

Different biosystem designs may share the same phenotype but have very different evotypes (top row). Rational engineering approaches could be used to transform a naive design (middle column), where evotype has not been considered, into either evolutionarily stable (left column) or specifically evolvable (right column) evotypes, which are characterised by their fitneity landscapes. Bioengineers can sculpt the evotype by modifying three major factors: genetic variation, production of function and selection. Genetic variation (green row): in a naive design, a mixture of variation operators may be in play. This might create a system that can reach many different regions of sequence space. It could be made more stable by reducing global mutation rates (e.g., host strain engineering) or by removing homologous regions to reduce the chance of recombination. Conversely, a naive design might be made more evolvable by increasing mutation rates in focused areas of sequence space (e.g., via methylation) and incorporating site-specific recombination or gene shuffling (e.g., the SCRaMbLE system). Function (blue row): a naive design may have high utility; however, if its function changes rapidly and chaotically across sequence space, it may be inherently unstable. A robust evotype has large neutral regions in function space. Conversely, a design can be made more evolvable if it can access a large range of new phenotypes, of a specific class (e.g., produce a colour) and the landscape may be smoothed (e.g., through removing crosstalk between features) and thus made amenable to evolutionary search. Production of function may be engineered by using prevalent phenotypes, designing in redundancy, modularity, regularity and hierarchy, increasing environmental robustness or by designing a system’s parameter space. Selection (orange row): if, as in the naive design, reproductive fitness (red dotted line) and utility (blue dashed line) are highly uncorrelated, then the design type may have a strong selection pressure acting against it and regions where both fitness and utility are maximised may be rare or non-existent; thus, high fitneity (grey solid line) may not be achievable. For a stable design, one might act to reduce the effects of natural selection through global increases in fitness (e.g., through reducing metabolic burden of a genetic circuit), by reducing toxicity of gene products or by reducing the fitness of neighbouring sequences (e.g., using minimised chassis organisms). A naive design can be made more evolvable by closely correlating fitness and utility (e.g., through coupling function to reproduction). This means natural selection will act to drive up the utility of the design: the precise goal of a directed evolution experiment.