Gene expression noise can promote the fixation of beneficial mutations in fluctuating environments

Abstract

Nongenetic phenotypic variation can either speed up or slow down adaptive evolution. We show that it can speed up evolution in environments where available carbon and energy sources change over time. To this end, we use an experimentally validated model of Escherichia coli growth on two alternative carbon sources, glucose and acetate. On the superior carbon source (glucose), all cells achieve high growth rates, while on the inferior carbon source (acetate) only a small fraction of the population manages to initiate growth. Consequently, populations experience a bottleneck when the environment changes from the superior to the inferior carbon source. Growth on the inferior carbon source depends on a circuit under the control of a transcription factor that is repressed in the presence of the superior carbon source. We show that noise in the expression of this transcription factor can increase the probability that cells start growing on the inferior carbon source. In doing so, it can decrease the severity of the bottleneck and increase mean population fitness whenever this fitness is low. A modest amount of noise can also enhance the fitness effects of a beneficial allele that increases the fraction of a population initiating growth on acetate. Additionally, noise can protect this allele from extinction, accelerate its spread, and increase its likelihood of going to fixation. Central to the adaptation-enhancing principle we identify is the ability of noise to mitigate population bottlenecks, particularly in environments that fluctuate periodically. Because such bottlenecks are frequent in fluctuating environments, and because periodically fluctuating environments themselves are common, this principle may apply to a broad range of environments and organisms.

Bridging Tumorigenesis and Therapy Resistance With a Non-Darwinian and Non-Lamarckian Mechanism of Adaptive Evolution

“…The foregoing propositions are compatible with many observations. For example, they are consistent with empirical findings showing that environment-induced plasticity can promote adaptive evolution (90,91), that expression variability among environments can affect gene evolution (92), and that a trait's variance may be controlled by genes that are not directly involved in the trait being considered [the Omnigenic Model (93)]. They align with the increased rate of tandem duplications frequently associated with up-regulated stress-responsive genes in several organisms (94)(95)(96) and with the predictable and frequent formation of de novo copy number variation in independent experimental evolution lines of yeast (97)(98)(99).…”

Transcriptional control of the lacZ promoter is under directional and diversifying selection

“…However, this suboptimal regulation may be mitigated by this isolate having evolved increased noise at the lac locus. Theoretical models have predicted that high noise can be beneficial when the precise expression control and plasticity is not optimal (Wolf et al 2015; Schmiedel et al 2019; Schmutzer and Wagner 2020). This has also been experimentally tested in S. cerevisiae, in which high-noise variants of the TDH3 promoter resulted in on average higher fitness as compared to low noise variants with suboptimal expression levels (Duveau et al 2018).…”

Two levels of selection of rhythmicity in gene expression: energy saving for rhythmic proteins and noise optimization for rhythmic transcripts

“…Recent studies show that in a population of unicellular organisms, variation in gene expression among genetically identical cells produces heterogeneous phenotypes conferring selective advantages in stressful, changing, or fluctuating environments (as shown in S. cerevisiae and E. coli) (Acar et al 2008, Wolf et al 2015, Thattai & van Oudenaarden 2004, Schmutzer & Wagner 2020, Duveau et al 2018, Liu et al 2015. The heterogeneity in genetically identical cells ensures that some cells are always prepared for changes of the environment, called "blind anticipation" by Acar et al (2008).…”