Many bacteria, including the model bacterium Escherichia coli can survive for years within spent media, following resource exhaustion. We carried out evolutionary experiments, followed by whole genome sequencing of hundreds of evolved clones to study the dynamics by which E. coli adapts during the first 4 months of survival under resource exhaustion. Our results reveal that bacteria evolving under resource exhaustion are subject to intense selection, manifesting in rapid mutation accumulation, enrichment in functional mutation categories and extremely convergent adaptation. In the most striking example of convergent adaptation, we found that across five independent populations adaptation to conditions of resource exhaustion occurs through mutations to the three same specific positions of the RNA polymerase core enzyme. Mutations to these three sites are strongly antagonistically pleiotropic, in that they sharply reduce exponential growth rates in fresh media. Such antagonistically pleiotropic mutations, combined with the accumulation of additional mutations, severely reduce the ability of bacteria surviving under resource exhaustion to grow exponentially in fresh media. We further demonstrate that the three positions at which these resource exhaustion mutations occur are conserved for the ancestral E. coli allele, across bacterial phyla, with the exception of nonculturable bacteria that carry the resource exhaustion allele at one of these positions, at very high frequencies. Finally, our results demonstrate that adaptation to resource exhaustion is not limited by mutational input and that bacteria are able to rapidly adapt under resource exhaustion in a temporally precise manner through allele frequency fluctuations.

Many bacterial species that cannot sporulate, such as the model bacterium Escherichia coli, can nevertheless survive for years, following exhaustion of external resources, in a state termed long-term stationary phase (LTSP). Here we describe the dynamics of E. coli adaptation during the first three years spent under LTSP. We show that during this time E. coli continuously adapts genetically, through the accumulation of mutations. For non-mutator clones, the majority of mutations accumulated appear to be adaptive under LTSP, reflected in an extremely convergent pattern of mutation accumulation. Despite the rapid and convergent manner in which populations adapt under LTSP, they continue to harbor extensive genetic variation. The dynamics of evolution of mutation rates under LTSP are particularly interesting. The emergence of mutators affects overall mutation accumulation rates as well as the mutational spectra and the ultimate spectrum of adaptive alleles acquired under LTSP. With time, mutators can evolve even higher mutation rates, through the acquisition of additional mutation-rate enhancing mutations. Different mutator and non-mutator clones within a single population and time point can display extreme variation in their mutation rates, resulting in differences in both the dynamics of adaptation and their associated deleterious burdens. Despite these differences, clones that vary greatly in their mutation rates tend to co-exist within their populations for many years, under LTSP.

The frequency of mutants resistant to the antibiotic rifampicin has been shown to increase in aging (starved), compared to young colonies of Eschierchia coli. These increases in resistance frequency occur in the absence of any antibiotic exposure, and similar increases have also been observed in response to additional growth limiting conditions. Understanding the causes of such increases in the frequency of resistance is important for understanding the dynamics of antibiotic resistance emergence and spread. Increased frequency of rifampicin resistant mutants in aging colonies is cited widely as evidence of stress-induced mutagenesis (SIM), a mechanism thought to allow bacteria to increase mutation rates upon exposure to growth-limiting stresses. At the same time it has been demonstrated that some rifampicin resistant mutants are relatively fitter in aging compared to young colonies, indicating that natural selection may also contribute to increased frequency of rifampicin resistance in aging colonies. Here, we demonstrate that the frequency of mutants resistant to both rifampicin and an additional antibiotic (nalidixic-acid) significantly increases in aging compared to young colonies of a lab strain of Escherichia coli. We then use whole genome sequencing to demonstrate conclusively that SIM cannot explain the observed magnitude of increased frequency of resistance to these two antibiotics. We further demonstrate that, as was previously shown for rifampicin resistance mutations, mutations conferring nalidixic acid resistance can also increase fitness in aging compared to young colonies. Our results show that increases in the frequency of antibiotic resistant mutants in aging colonies cannot be seen as evidence of SIM. Furthermore, they demonstrate that natural selection likely contributes to increases in the frequency of certain antibiotic resistance mutations, even when no selection is exerted due to the presence of antibiotics.

Codon usage bias affects the genomes of organisms from all kingdoms of life and results from both background substitution biases and natural selection. Natural selection on codon usage to increase translation accuracy and efficiency has long been known to affect gene sequences. Such selection is stronger on highly, compared with lowly expressed genes, resulting in higher levels of codon bias within genes with higher expression levels. Additionally, selection on translation accuracy affects more strongly codons encoding conserved amino acids, since these will more often affect protein folding and/or function. By applying tests of selection on the gene sequences of the bacterium Escherichia coli, we demonstrate that both highly and lowly expressed genes display signals of selection on codon usage. Such signals are found for both conserved and less conserved amino acid positions, even within the 10% of E. coli genes expressed at the lowest levels. We further demonstrate experimentally that single synonymous codon replacements within a lowly expressed, essential gene can carry substantial effects on bacterial fitness. Combined, our results demonstrate that even within genes expressed at relatively low levels there is substantial selection on codon usage and that single synonymous codon replacements within such genes can have a marked effect on bacterial fitness.

Many nonsporulating bacterial species can survive for years within exhausted growth media in a state termed long-term stationary phase (LTSP). We have been carrying out evolutionary experiments aimed at elucidating the dynamics of genetic adaptation under LTSP. We showed that Escherichia coli adapts to prolonged resource exhaustion through the highly convergent acquisition of mutations. In the most striking example of such convergent adaptation, we observed that across all independently evolving LTSP populations, over 90% of E. coli cells carry mutations to one of three specific sites of the RNA polymerase core enzyme (RNAPC). These LTSP adaptations reduce the ability of the cells carrying them to grow once fresh resources are again provided. Here, we examine how LTSP populations recover from costs associated with their adaptation once resources are again provided to them. We demonstrate that due to the ability of LTSP populations to maintain high levels of standing genetic variation during adaptation, costly adaptations are very rapidly purged from the population once they are provided with fresh resources. We further demonstrate that recovery from costs acquired during adaptation under LTSP occurs more rapidly than would be possible if LTSP adaptations had fixed during the time populations spent under resource exhaustion. Finally, we previously reported that under LTSP, some clones develop a mutator phenotype, greatly increasing their mutation accumulation rates. Here, we show that the mechanisms by which populations recover from costs associated with fixed adaptations may depend on mutator status. IMPORTANCE Many bacterial species can survive for decades under starvation, following the exhaustion of external growth resources. We have previously shown that bacteria genetically adapt under these conditions in a manner that reduces their ability to grow once resources again become available. Here, we study how populations that have been subject to very prolonged resource exhaustion recover from costs associated with their adaptation. We demonstrate that rapid adaptations acquired under prolonged starvation tend to be highly transient, rapidly reducing in frequency once bacteria are no longer starved. Our results shed light on the longer-term consequences of bacterial survival under prolonged starvation. More generally, these results may also be applicable to understanding longer-term consequences of rapid adaptation to additional conditions as well.