Carole Knibbe

New insights into bacterial adaptation through in vivo and in silico experimental evolution

Microbiology research has recently undergone major developments that have led to great progress towards obtaining an integrated view of microbial cell function. Microbial genetics, high-throughput technologies and systems biology have all provided an improved understanding of the structure and function of bacterial genomes and cellular networks. However, integrated evolutionary perspectives are needed to relate the dynamics of adaptive changes to the phenotypic and genotypic landscapes of living organisms. Here, we review evolution experiments, carried out both in vivo with microorganisms and in silico with artificial organisms, that have provided insights into bacterial adaptation and emphasize the potential of bacterial regulatory networks to evolve.

Reductive genome evolution at both ends of the bacterial population size spectrum

Bacterial genomes show substantial variations in size. The smallest bacterial genomes are those of endocellular symbionts of eukaryotic hosts, which have undergone massive genome reduction and show patterns that are consistent with the degenerative processes that are predicted to occur in species with small effective population sizes. However, similar genome reduction is found in some free-living marine cyanobacteria that are characterized by extremely large populations. In this Opinion article, we discuss the different hypotheses that have been proposed to account for this reductive genome evolution at both ends of the bacterial population size spectrum.

In silico experimental evolution: a tool to test evolutionary scenarios

Comparative genomics has revealed that some species have exceptional genomes, compared to their closest relatives. For instance, some species have undergone a strong reduction of their genome with a drastic reduction of their genic repertoire. Deciphering the causes of these atypical trajectories can be very difficult because of the many phenomena that are intertwined during their evolution (e.g. changes of population size, environment structure and dynamics, selection strength, mutation rates...). Here we propose a methodology based on synthetic experiments to test the individual effect of these phenomena on a population of simulated organisms. We developed an evolutionary model - aevol - in which evolutionary conditions can be changed one at a time to test their effects on genome size and organization (e.g. coding ratio). To illustrate the proposed approach, we used aevol to test the effects of a strong reduction in the selection strength on a population of (simulated) bacteria. Our results show that this reduction of selection strength leads to a genome reduction of ~35% with a slight loss of coding sequences (~15% of the genes are lost - mainly those for which the contribution to fitness is the lowest). More surprisingly, under a low selection strength, genomes undergo a strong reduction of the noncoding compartment (~55% of the noncoding sequences being lost). These results are consistent with what is observed in reduced Prochlorococcus strains (marine cyanobacteria) when compared to close relatives.

Scaling Laws in Bacterial Genomes: A Side-Effect of Selection of Mutational Robustness

In the past few years, numerous research projects have focused on identifying and understanding scaling properties in the gene content of prokaryote genomes and the intricacy of their regulation networks. Yet, and despite the increasing amount of data available, the origins of these scalings remain an open question. The RAevol model, a digital genetics model, provides us with an insight into the mechanisms involved in an evolutionary process. The results we present here show that (i) our model reproduces qualitatively these scaling laws and that (ii) these laws are not due to differences in lifestyles but to differences in the spontaneous rates of mutations and rearrangements. We argue that this is due to an indirect selective pressure for robustness that constrains the genome size.

Breaking Good: Accounting for Fragility of Genomic Regions in Rearrangement Distance Estimation

Models of evolution by genome rearrangements are prone to two types of flaws: One is to ignore the diversity of susceptibility to breakage across genomic regions, and the other is to suppose that susceptibility values are given. Without necessarily supposing their precise localization, we call “solid” the regions that are improbably broken by rearrangements and “fragile” the regions outside solid ones. We propose a model of evolution by inversions where breakage probabilities vary across fragile regions and over time. It contains as a particular case the uniform breakage model on the nucleotidic sequence, where breakage probabilities are proportional to fragile region lengths. This is very different from the frequently used pseudouniform model where all fragile regions have the same probability to break. Estimations of rearrangement distances based on the pseudouniform model completely fail on simulations with the truly uniform model. On pairs of amniote genomes, we show that identifying coding genes with solid regions yields incoherent distance estimations, especially with the pseudouniform model, and to a lesser extent with the truly uniform model. This incoherence is solved when we coestimate the number of fragile regions with the rearrangement distance. The estimated number of fragile regions is surprisingly small, suggesting that a minority of regions are recurrently used by rearrangements. Estimations for several pairs of genomes at different divergence times are in agreement with a slowly evolvable colocalization of active genomic regions in the cell.

A Model for Genome Size Evolution

We present a model for genome size evolution that takes into account both local mutations such as small insertions and small deletions, and large chromosomal rearrangements such as duplications and large deletions. We introduce the possibility of undergoing several mutations within one generation. The model, albeit minimalist, reveals a non-trivial spontaneous dynamics of genome size: in the absence of selection, an arbitrary large part of genomes remains beneath a finite size, even for a duplication rate 2.6-fold higher than the rate of large deletions, and even if there is also a systematic bias toward small insertions compared to small deletions. Specifically, we show that the condition of existence of an asymptotic stationary distribution for genome size non-trivially depends on the rates and mean sizes of the different mutation types. We also give upper bounds for the median and other quantiles of the genome size distribution, and argue that these bounds cannot be overcome by selection. Taken together, our results show that the spontaneous dynamics of genome size naturally prevents it from growing infinitely, even in cases where intuition would suggest an infinite growth. Using quantitative numerical examples, we show that, in practice, a shrinkage bias appears very quickly in genomes undergoing mutation accumulation, even though DNA gains and losses appear to be perfectly symmetrical at first sight. We discuss this spontaneous dynamics in the light of the other evolutionary forces proposed in the literature and argue that it provides them a stability-related size limit below which they can act.

Comparative Genomics on Artificial Life

Molecular evolutionary methods and tools are difficult to validate as we have almost no direct access to ancient molecules. Inference methods may be tested with simulated data, producing full scenarios they can be compared with. But often simulations design is concomitant with the design of a particular method, developed by a same team, based on the same assumptions, when both should be blind to each other. In silico experimental evolution consists in evolving digital organisms with the aim of testing or discovering complex evolutionary processes. Models were not designed with a particular inference method in mind, only with basic biological principles. As such they provide a unique opportunity to blind test the behavior of inference methods. We give a proof of this concept on a comparative genomics problem: inferring the number of inversions separating two genomes. We use Aevol, an in silico experimental evolution platform, to produce benchmarks, and show that most combinatorial or statistical estimators of the number of inversions fail on this dataset while they were behaving perfectly on ad-hoc simulations. We argue that biological data is probably closer to the difficult situation.

Beware batch culture: Seasonality and niche construction predicted to favor bacterial adaptive diversification

Metabolic cross-feeding interactions between microbial strains are common in nature, and emerge during evolution experiments in the laboratory, even in homogeneous environments providing a single carbon source. In sympatry, when the environment is well-mixed, the reasons why emerging cross-feeding interactions may sometimes become stable and lead to monophyletic genotypic clusters occupying specific niches, named ecotypes, remain unclear. As an alternative to evolution experiments in the laboratory, we developed EvoSim, a multi-scale model of in silico experimental evolution, equipped with the whole tool case of experimental setups, competition assays, phylogenetic analysis, and, most importantly, allowing for evolvable ecological interactions. Digital organisms with an evolvable genome structure encoding an evolvable metabolic network evolved for tens of thousands of generations in environments mimicking the dynamics of real controlled environments, including chemostat or batch culture providing a single limiting resource. We show here that the evolution of stable cross-feeding interactions requires seasonal batch conditions. In this case, adaptive diversification events result in two stably co-existing ecotypes, with one feeding on the primary resource and the other on by-products. We show that the regularity of serial transfers is essential for the maintenance of the polymorphism, as it allows for at least two stable seasons and thus two temporal niches. A first season is externally generated by the transfer into fresh medium, while a second one is internally generated by niche construction as the provided nutrient is replaced by secreted by-products derived from bacterial growth. In chemostat conditions, even if cross-feeding interactions emerge, they are not stable on the long-term because fitter mutants eventually invade the whole population. We also show that the long-term evolution of the two stable ecotypes leads to character displacement, at the level of the metabolic network but also of the genome structure. This difference of genome structure between both ecotypes impacts the stability of the cross-feeding interaction, when the population is propagated in chemostat conditions. This study shows the crucial role played by seasonality in temporal PLOS Computational Biology | https://doi.org/10.1371/journal.pcbi.1005459 March 30, 2017 1 / 32 a1111111111 a1111111111 a1111111111 a1111111111 a1111111111

Self-adaptation of genome size in artificial organisms

In this paper we investigate the evolutionary pressures influencing genome size in artificial organisms. These were designed with three organisation levels (genome, proteome, phenotype) and are submitted to local mutations as well as rearrangements of the genomic structure. Experiments with various per-locus mutation rates show that the genome size always stabilises, although the fitness computation does not penalise genome length. The equilibrium value is closely dependent on the mutational pressure, resulting in a constant genome-wide mutation rate and a constant average impact of rearrangements. Genome size therefore self-adapts to the variation intensity, reflecting a balance between at least two pressures: evolving more and more complex functions with more and more genes, and preserving genome robustness by keeping it small.

Souvenirs from ECAL 2017: create, play, experiment, discover – revealing the experimental power of virtual worlds

This report presents some highlights from ECAL 2017, the Fourteenth European Conference on Artificial Life, which was held on 4–8 September 2017 in Lyon, France.