Evolution in an unchanging environment

In Richard Lenski's long-term experiment, polymorphic communities formed from identical bacteria

Alexander Markov, "Elements"

A long-term evolutionary experiment on Escherichia coli bacteria, which has been going on for almost 30 years, has yielded new unexpected results. Sequencing of bacterial DNA from the frozen "fossil record" of the experiment showed that over 60,000 generations, the evolution of experimental populations has not stopped or even slowed down. Ecological diversification occurred in at least 9 out of 12 populations: the original monoculture was divided into clades connected by ecological interactions and not displacing each other. Within each clade, evolution continues in full swing, with further changes being guided by both the previous evolutionary history and the changing ecological situation. Thus, evolution "outsmarted" researchers who hoped to study the effect of mutations and selection during adaptation to stable conditions in an extremely simple artificial system.

Elements has repeatedly talked about a long-term evolutionary experiment started in 1988 by Richard Lenski. During the experiment, 12 initially identical populations of Escherichia coli Escherichia coli have been cultivated under the same conditions for more than 60,000 generations: in a liquid nutrient medium, where the only food source is glucose. However, one of the populations, Ara-3, has also learned to eat citrate, which is present in the medium as an auxiliary substance (see: In a long-term experiment, the gradual formation of an evolutionary innovation was recorded, "Elements", 25.09.2012). Once a day, a small part (about 10 ⁷ cells) is taken from each population and transplanted into a fresh nutrient medium, where the bacteria first multiply rapidly, and then, as glucose is exhausted, the number growth slows down. Every 500 generations, part of the bacteria from each population is frozen for further study. Microbes remain viable at the same time. Thus, the researchers have at their disposal a "living paleontological chronicle" of the experiment.

Although the conditions of the experiment look extremely simple, many of its results were unexpected. For example, it would be logical to assume that after a short period of rapid adaptation (growth of fitness) to new conditions, populations will reach an optimum (rise to the peak of the fitness landscape, see Fitness landscape), the stock of possible beneficial mutations will be exhausted, changes will slow down and an indefinitely long period of evolutionary stasis will occur. It was not there: even after 50,000 generations, useful mutations still appeared, and fitness continued to grow, albeit with a slowdown (see: New results of a long-term evolutionary experiment: the fitness of experimental bacteria continues to grow, "Elements", December 23, 2013).

The experiment was originally planned in such a way as to minimize all complicating circumstances, such as changes in environmental conditions, genetic exchange and ecological interactions between organisms. The researchers wanted to obtain in its purest form the simplest and most fundamental evolutionary process – adaptation to a stable environment based on mutations and selection. However, as Leslie Orgel aptly noted, "evolution is smarter than you" (see: Orgel's rules). A new article by Lenski and his colleagues, published on October 18 in the journal Nature, shows that it was not possible to avoid difficulties after all. As it turned out, ecological interactions based on the diversification and division of niches arise by themselves in experimental populations, which forces bacteria to adapt to the changing biotic environment.

The authors conducted a genetic analysis of the entire frozen "fossil record" of the experiment, accumulated over 60,000 bacterial generations and totaling about 1,440 samples (60,000 / 500 = 120 samples for each of the 12 populations). A metagenomic analysis with 50-fold coverage was performed for each sample (see: Coverage). This turned out to be enough to reliably identify all new mutations that occurred in experimental populations and reached a frequency of at least 10% (that is, they occurred in at least every tenth bacterium) in at least two samples. Mutations that were not so widespread were not taken into account because they are difficult to distinguish from random sequencing errors.

As a result, a detailed reconstruction of the evolutionary process in 12 populations was obtained (Fig. 1).

Fig. 1. The course of molecular evolution in 12 populations of E. coli. Changes in the frequency of occurrence of all registered new mutations are shown. On the horizontal axis – time in generations. A drawing from the discussed article in Nature

It should be remembered that almost all changes in allele frequencies, noticeable in Figure 1, reflect the work of natural selection, and not genetic drift (random fluctuations in allele frequencies). Drift works much slower. The effective number of experimental populations is approximately 107, and with such a number, in order for a new mutation to be fixed (reach the frequency 1) due to drift, it takes about 107 generations, that is, hundreds of times more than has passed since the beginning of the experiment. All mutations that reached a noticeable frequency during the experiment did so under the influence of selection. They were either useful themselves and supported by selection, or they were in the same genome with a useful mutation and spread due to "genetic hitchhiking" (see: Genetic hitchhiking), typical for asexual populations (see: Sexual reproduction helps selection to separate beneficial mutations from harmful ones, "Elements", 01.03.2016).

As already mentioned, the growth of overall fitness (which is estimated by the rate of reproduction compared to the original, ancestral strain) slowed down, but did not stop (Fig. 2, a; for more details, see in the news, New results of a long-term evolutionary experiment: the fitness of experimental bacteria continues to grow, "Elements", December 23, 2013). The rate of accumulation of new mutations remained high (Fig. 2, b). Mutator alleles (mutations that reduce the accuracy of replication or repair) were recorded in six of the twelve populations, which dramatically accelerated both the appearance and fixation of new mutations (graphs running up in Fig. 2, b). However, somewhat later in populations with mutators, alleles-"antimutators" began to spread, reducing the rate of mutagenesis. This can be seen by the slowdown in the growth of the number of mutations in some populations with mutators (inset in Fig. 2, b).

Fig. 2. Dynamics of fitness and accumulation of mutations in 12 populations. a – fitness, estimated by the rate of reproduction of bacteria in comparison with the original ancestral strain. b is the average number of accumulated mutations. The graphs that go up sharply correspond to six populations in which the rate of mutagenesis has increased due to the fixation of "mutator" alleles (mutations that reduce the accuracy of replication or repair). The inset shows the dynamics of mutation accumulation in six populations with mutators. The white line shows the average dynamics for populations in which mutators have not been recorded. A drawing from the discussed article in Nature

But the main discovery made by the authors is not this. The evolutionary dynamics shown in Figure 1 does not fit into the simplest model, according to which the adaptive evolution of a monoculture of asexual organisms in stable conditions is reduced to the sequential fixation by selection of newly emerging beneficial mutations.

Statistical processing of the data shown in Fig. 1 showed that this simplest model cannot explain the observed picture even taking into account such complicating circumstances as genetic hitchhiking and clonal interference (competition between bacterial clones with different beneficial mutations; see: Clonal interference). For example, many mutations, having reached a certain frequency, suddenly stop spreading, that is, move further towards fixation (this is the natural course of events if a clone with this mutation has a higher fitness than other bacteria in the population). But these clones do not die out, having lost the competition to clones with more successful mutations. Instead, the mutation frequency begins to fluctuate around some level. These fluctuations can last tens of thousands of generations, and the level around which the fluctuations occur may change over time.

The metagenomic data obtained for each of the 1,440 samples is a set of sequenced pieces of DNA belonging to different individuals. Therefore, it is impossible to immediately understand which mutations belong to the same clone, and which ones belong to different ones. However, the authors managed to figure this out by analyzing the consistency of mutation frequency changes over time (since the frequencies of mutations located in the same genome change synchronously). As a result, it turned out that at least nine of the twelve experimental populations had a stable coexistence of at least two clades (evolutionary lines) for a long time (over 10,000 generations). Within these clades there were their own evolutionary processes, that is, various mutations appeared and were fixed (Fig. 3).

Fig. 3. Long-term coexistence of clades in experimental populations. a – detailed population history Ara-6. In the upper figure, the mutations of the "basal clade" (basal clade) are shown in different colors, from which the "main" clade (major clade, the second figure from above) and the "minor" clade (minor clade, the third figure) subsequently originated. Which of the treasures was considered the main one, and which was secondary, depended on their number at the end of the observation period. The bottom figure shows the fate of extinct mutations – the history of all twelve populations, in nine of which two clades (purple and pink lines) have been established. Image from the discussed article in Nature

This means that ecological diversification has occurred in most of the experimental populations. Different clades somehow divided ecological niches among themselves and began to co-exist steadily, adapting now not to initially set stable environmental conditions, but to a specific and changeable biotic environment.

This phenomenon was previously detected in one of the twelve populations (Ara-2). Two clades coexisting in this population have different metabolism and use to their advantage the waste products of the other clade. Stable coexistence is ensured by frequency-dependent balancing selection. This means that the relative fitness of a clade is higher the lower its number (J. Plucain et al., 2014. Epistasis and allele specificity in the emergence of a stable polymorphism in Escherichia coli). New data have shown that a similar situation has developed in at least nine out of twelve populations. Thus, ecological diversification is not a random episode, but a general pattern.

Analysis of the history of individual clades has shown that adaptive evolution within the clade continues to be in full swing: new useful (for this clade) mutations appear, their frequencies grow under the influence of selection, other (not so useful) mutations spread along with them "hitchhiking"; many genetic variants, having reached a noticeable frequency, subsequently die out, displaced more successful competitors. And all this is no longer happening on the scale of the entire experimental population, but separately in each of the treasures. Therefore, the assessment of the fitness of bacteria by their growth rate in comparison with the ancestral strain is partly meaningless (Fig. 2, a): after all, now their real fitness also depends on how successfully they interact with representatives of coexisting clades.

Statistical analysis of the distribution of mutations over time showed that in some genes mutations were mainly fixed at the beginning of the experiment (at the early stages of adaptation), while in other genes mutations began to be fixed only at late stages. This is explained by three reasons, and all three, according to the authors, actually work during the experiment (it cannot be strictly proved yet, but there are indirect statistical arguments in favor of the reality of all three reasons).

Firstly, mutations in some genes are the most beneficial (give the greatest increase in fitness) for the original genotype, and therefore such mutations are fixed first, displacing other, less useful mutations in other genes during clonal interference. These latter begin to spread later, when the first portion of the most "obvious" beneficial mutations has already been fixed.

Secondly, mutations recorded earlier affect the usefulness or harmfulness of mutations that appear later (the influence of some genetic variants on phenotypic manifestations, including usefulness, others are called epistasis, see: Epistasis). Therefore, some mutations become useful and get a chance to fix themselves only after a suitable genetic context is formed due to other mutations (see: The evolution of proteins is constrained by the low patency of the fitness landscape, "Elements", 09.02.2015). This was the case with the mutation, thanks to which bacteria from the Ara-3 population were able to feed on citrate (see: In a long-term experiment, the gradual formation of an evolutionary innovation was recorded, "Elements", 25.09.2012).

Thirdly, the emerging ecological interactions between the clades fundamentally change the "rules of the game", forcing bacteria to adapt not to a stable and extremely simple environment (as was conceived by the researchers), but to a dynamic biotic environment. This means that the direction of selection changes all the time, and therefore mutations that are useful for a given clade at the moment will not necessarily be useful at another time or for other clades.

Thus, a long-term evolutionary experiment has refuted overly simplistic ideas about how an asexual population should adapt to stable environmental conditions. There is nothing like slowing down and stopping adaptive evolution as it approaches the optimum (peak on the fitness landscape), the stock of potentially useful mutations is not exhausted, and even the rate of their accumulation practically does not decrease. Instead, we see a spontaneous complication of an evolving community, which turns from a monoculture into an ecosystem with subdivided niches and co-evolving clades and is clearly not going to go into a state of evolutionary stasis in the foreseeable future. So Leslie Orgel was, of course, right about who is smarter – evolution or theorists who think everyone knows about it.

A source: Good et al., The dynamics of molecular evolution over 60,000 generations // Nature. Published online 18 October 2017.

Portal "Eternal youth" http://vechnayamolodost.ru