Experimental evolution for cell biology

Abstract

Evolutionary cell biology explores the origins, principles, and core functions of cellular features and regulatory networks through the lens of evolution. This emerging field relies heavily on comparative experiments and genomic analyses that focus exclusively on extant diversity and historical events, providing limited opportunities for experimental validation. In this opinion article, we explore the potential for experimental laboratory evolution to augment the evolutionary cell biology toolbox, drawing inspiration from recent studies that combine laboratory evolution with cell biological assays. Primarily focusing on approaches for single cells, we provide a generalisable template for adapting experimental evolution protocols to provide fresh insight into long-standing questions in cell biology.

Evolutionary cell biology - success and opportunity in an emerging interdisciplinary field

There are few questions in biology that spark the imagination as much as those exploring the origins and diversity of life on Earth. How did organisms, and their constituent cells, end up being the way they are today? And once we know how, can we also understand why cells function the way they do? While humans have been asking such questions throughout recorded history, only recently have we developed the tools needed to answer such queries in a systematic way. Evolutionary cell biology employs and integrates the mature fields of evolutionary biology and cell biology to explore the origins and diversity of cellular complexity (Figure 1A). The aims of evolutionary cell biology are twofold: to develop a mechanistic understanding of evolutionary processes by exploring cell biological diversity across the tree of life, and to use an evolutionary perspective to elucidate how cellular processes work [1].

Figure 1 |. Evolutionary cell biology: integrating evolutionary biology and cell biology.

A. Evolutionary cell biology tries to integrate the fields of evolutionary biology and cell biology, each composed of their own sub-disciplines. B. Example highlighting the power of combining comparative phylogenetics with cell biological characterizations across different species. By mapping specific cellular features such as the presence or absence of centrioles and cilia on phylogenetic trees and correlating them back to the species’ genotypes, genotypes can be linked to phenotypes, even for complex cellular processes. This example highlights how the presence (black circles) and absence (white circles) of particular genes coding for proteins involved in centriolar structure correlates with the occurrence of centrioles and cilia/flagella. Gray circles denote instances where a similar, non-orthologous gene is present. Figure modified from [2].

The first steps in integrating modern cell and evolutionary biology have already proven successful. By combining comparative phylogenetic methods (see Glossary) with in-depth cell biological characterizations of selected cellular features across different species, researchers have retraced the evolutionary origins of proteins, protein complexes, and their corresponding cellular phenotypes. For example, the co-evolving ancestral control module for centrioles, cilia, and flagella was uncovered, together with a number of previously unknown components, using this approach (Figure 1B) [2,3]. More recently, an inventory of kinetochore diversity, inferring orthology from 90 diverse eukaryotes, was used to show that the last eukaryotic common ancestor (LECA) already possessed a complex kinetochore [4]. New genomes are being sequenced and made publicly available daily. Such new data, coupled with continued cell biological characterizations, will allow us to identify the minimal sets of proteins required to generate and regulate specific cellular features, and will reveal universal principles underlying certain cellular phenotypes.

Despite the successes highlighted above, far less progress has been made in combining knowledge from other sub-disciplines of evolutionary biology with cell biology (Figure 1A). As a result, questions relating to how evolutionary dynamics and phenomena like epistasis have shaped and contributed to the evolution of cellular features remain mostly unanswered. In addition, even though comparative phylogenetics is clearly a powerful tool in evolutionary cell biology, it is an inherently observational, retrospective approach, with limited potential for systematic experimental validation.

Here, we explore the potential for a prospective approach, experimental evolution, to bridge this gap. We give examples of some recent studies that combine laboratory evolution with cell biological assays, and discuss how classic set-ups can be incorporated into a new integrated design to specifically answer questions related to evolutionary cell biology.

Building on existing evolution experiments for evolutionary cell biology

Experimental evolution is a powerful tool for studying evolutionary processes in real time. The principle is simple: cells or organisms are propagated under a defined selective pressure and lineages carrying mutations that increase fitness outcompete others. Over the last several decades, hundreds of such experiments have provided insight into various aspects of the evolutionary process [5–7]. Using microbes, researchers have systematically explored how mutations spread through populations in relation to their effects on fitness [8], how different mutations interact (epistasis) and how this influences the evolutionary process [9,10]. Using multicellular model organisms such as Drosophila melanogaster and Caenorhabditis elegans, experimental evolution has helped us understand how sexual selection factors into the evolutionary process [11,12], and how different aspects of development can be linked to evolution [13]. At an interspecies level, evolutionary ecologists have studied how interactions between different species and their environments have been shaped through evolution, and how factors such as inter- or intra-species competition can drive evolutionary dynamics [14,15]. However, only a small but growing subset of recent studies explores why beneficial mutations are adaptive from a cell biological point of view.

In one of these studies, budding yeast (Saccharomyces cerevisiae) cells with a defective beta-tubulin (TUB2) allele were evolved for 150 generations [16]. Cells with this defective allele are unable to properly polymerize microtubules and have difficulty assembling a functional spindle, resulting in delayed mitosis and elevated levels of chromosome missegregation [16]. After experimental evolution, cells with the defective allele had significantly restored microtubule functionality, primarily through missense mutations in tubulins themselves (TUB1, TUB2 and TUB4). By mapping the mutations onto protein structures, the authors showed that many recurrent amino acid changes are located in regions important for interactions with microtubule-binding proteins and in the proteins’ GTP-binding pockets. From a cell biological point of view, it would be interesting to further explore how these different tubulin mutations affect microtubule polymerization dynamics. Since tubulin proteins are highly conserved across eukaryotes, the mutations could also be contrasted with regions of sequence variation that are observed across different species, and instead of using the allele described above, TUB2 could be replaced by orthologs from different species or by a reconstructed ancestral version of the protein.

Instead of using defective alleles or orthologs, one can also look at paralogs to assess how duplicated proteins diverge within the same genome. Hsieh et al. replaced the mitotic kleisin Scc1 in S. cerevisiae with its meiotic paralog Rec8 [17]. Even though both proteins interact with cohesin proteins and help form the complex that holds sister chromosomes together, the mitotic kleisin cannot fully support meiotic chromosome segregation and vice versa [18]. Cells with only Rec8 display impaired chromosome linkage, start replicating their genome earlier in the cell cycle, and have a significant fitness defect [17]. After evolving the strains for ~1,750 generations, the authors observed mutations in the transcriptional mediator complex, cohesin-related genes, and regulators of the G1-S transition. They used fluorescence microscopy to show that evolved cells have improved chromosome cohesion, and measured replication dynamics to reveal that replication origins are fired at times more similar to the wild type. In budding yeast, cohesin is loaded onto the chromosomes before DNA replication, and functional linkages are established during S phase with the passage of the replication fork [19,20]. Based on these and their own observations, the authors proposed a more general cell biological hypothesis: that the dynamics of genome replication impact chromosome cohesion. This would imply that slowing genome replication would improve Rec8-dependent sister chromosome cohesion. The authors tested this idea, and indeed found that both delaying replication origin firing by removing two S phase cyclins, and reducing the speed of replication forks by removing a DNA helicase improves sister cohesion [17].

An example that highlights the added value of using not one, but several related defective alleles can be found in Escherichia coli. McCloskey et al. contrasted the evolution of strains lacking one of five different core metabolic genes and showed that each deletion strain followed a specific adaptive path [21]. By measuring gene expression levels, intracellular metabolite levels and metabolic fluxes, they were able to assign a cell biological explanation to several adaptive mutations. Cells lacking a functional succinate dehydrogenase, which converts succinate to fumarate in the TCA cycle, acquired mutations to reduce the flux through the TCA cycle [22]. By contrast, strains without phosphoglucose isomerase, inducing increased redox and sugar phosphate stress, evolved to reduce flux in those pathways [23]. An interesting next step to take from an evolutionary cell biological point of view would be to relate the original and adaptive mutations back to observed sequence variation in other natural strains or species, and try to link it to their respective metabolic profiles and ecological niches.

Experimental evolution can also be used to study tumorigenesis, and the increasing ease of genome editing (e.g. using CRISPR/Cas9-based techniques) has enabled researchers to design experiments similar to the examples described above in mammalian cells. Karlsson et al. evolved TP53-deficient human gastric organoids for over two years and were able to repeatedly recapitulate malignant transformation [24]. They showed that loss of TP53 induced aneuploidy, copy number alterations and complex structural variation in every population, suggesting significant genome instability. Single-cell RNA sequencing at different timepoints throughout the experiment indicated that cells evolved towards malignant transcriptional states. While the expression of normal gastric cell type markers was consistent in wild-type organoids, TP53-deficient cultures showed a much more heterogeneous marker expression profile. Additionally, several pathways that are commonly implicated in malignancy (e.g. NF-κβ) were altered across multiple evolved samples. It will be especially interesting to explore whether these transcriptional differences are reflected in the cells’ cellular and subcellular morphology, and ultimately, to use systems like this to compile a cell-biological overview of the transition to malignancy.

These are examples of evolutionary repair experiments (Figure 2), a classic category of evolution experiments in which the ancestor contains at least one genetic perturbation. Most often, the perturbation is a single gene deletion or defective allele [16,24–30], but protein sequences can also be replaced by orthologs [31,32], paralogs [17] or reconstructed ancestral protein sequences [35] to help determine how protein sequence variability is shaped by, and contributes to the evolutionary process. Depending on the function of the gene that was originally perturbed, an evolving organism will have access to different mutational trajectories. By perturbing genes across the genetic interaction network and by comparing how they behave in evolutionary repair experiments, one can systematically explore how different components of cellular processes interact in an evolutionary setting [21,34,35]. Together, these different experimental set-ups allow researchers to explore a wide variety of evolutionary questions.

Figure 2 |. Different types of evolutionary repair experiments, using the kinetochore as an example of a cellular feature that can be perturbed.

A. Classic evolutionary repair experiments: one specific perturbation is introduced, after which the strain is evolved. Evolved strains are then characterised to evaluate how cells compensated for the perturbation. B. Homolog swap evolutionary repair experiments. Instead of introducing a defect (e.g. by removing the gene), proteins are replaced by orthologs, paralogs, or sometimes even reconstructed ancestral versions of the protein. C. Systems-level evolutionary repair experiments. This type of experiment follows the same basic scheme, but uses a more systematic approach by perturbing different components of the complex or process.

Using experimental evolution to fill in the gaps in evolutionary cell biology

Despite the examples shown above, few laboratory evolution experiments are designed to specifically answer questions related to evolutionary cell biology [36]. To remedy this, we propose an experimental design specific for evolutionary cell biology below. We also discuss one area of evolutionary biology that remains mostly unexplored within evolutionary cell biology: the influence of evolutionary dynamics on the evolution of cellular features. We finish with a discussion on the evolution of biological innovation, and consider if and how this phenomenon can be explored using experimental evolution.

Experimental design and generating hypotheses

Figure 3 gives an example of an integrated experimental design that is specifically tailored to answer cell biological questions using experimental evolution. Phylogenetics, in combination with cell biological assays, are an ideal starting point for further experimentation (Figure 3A). This combination helps to narrow down sets of genes that are associated with particular cellular features, and informs us how their sequences vary in nature and throughout evolution. Generating high-quality gene trees also provides the opportunity to infer ancestral genotypic and phenotypic states. The inferred links between genotype and phenotype can be tested in the lab by introducing the observed genetic variation into different experimentally-tractable organisms (ETOs) and using cell biological assays to evaluate the resulting phenotypes (Figure 3B). Next, experimental evolution can be used to explore the evolutionary response to these perturbations. Characterising adaptation on both a genetic (Figure 3C) and phenotypic (Figure 3D) level not only tells us more about how cellular features of interest evolve in a controlled setting, but also about the role of existing natural genetic variation in the evolutionary process. Finally, by further characterising a broader set of cellular features of the evolved progeny (Figure 3E), we can learn if and how different cellular features co-evolve and how this translates into evolutionary trade-offs. At different points in this experimental workflow, including knowledge from complementary fields such as systems biology, evolutionary theory, biophysics, and biochemistry may prove useful. For example, if introducing natural diversity from phylogenetic analyses in ETOs (Figure 3B) does not lead to sufficient variation in the cellular features of interest, knowledge from genome-wide screens in model organisms can supplement the genetic variation that will be included at the start of the evolution experiment.

Figure 3 |. Example of integrated experimental design for evolutionary cell biology.

By integrating computational and experimental approaches, we can systematically probe how specific cellular features evolve. A. A combination of comparative phylogenetics and cell biological characterizations in different species gives us an idea of how variation in particular gene products translates into specific phenotypes (here: cellular features). ETO: experimentally-tractable organism. Red outline: ancestral reconstructed states. Figure is simplified and only shows presence/absence of genes, but in reality we can look at variation in protein sequences. B. By introducing the observed genetic variation in ETOs, we can evaluate the validity of the inferred genotype-phenotype links. ETOs with different perturbations are evolved, after which their C. mutations and D. cellular features are characterised, allowing us to test the role of natural variation in the evolution of cellular features in real time. E. To learn more about the relationships between different cellular features and trade-offs within an evolutionary context, a broader set of cellular features can be assayed.

Another important aspect of experimental design for evolutionary cell biology is the inclusion of multiple species across the phylogenetic tree. The vast majority of classic evolution experiments use a single focal species, making it difficult to draw generalizable conclusions. We know that even within a species, genetic background is an important determinant of evolutionary outcome [37]. Hence, the field would benefit from expanding the repertoire of experimentally-tractable organisms, and including sets of representative species in each evolution experiment. In the previous section, we highlighted examples of evolution experiments that include a cell-biological point of view with S. cerevisiae, E. coli and human cells, but many more organisms across the tree of life have proven to be amenable to experimental evolution. To date, researchers have evolved various animals [11–13], fungi [16,17], plants [38], (micro)algae [39], ciliates [40,41], amoeba [42], bacteria [21,43], and viruses [44]. When designing new evolution experiments, such as the ones depicted in Figure 3, with species other than those that are traditionally used, two factors have to be taken into account. First, one has to consider the ease of experimentation, both in terms of genome manipulation and microscopic characterization. For some of the species listed above, both are readily available. For example, the C. elegans genome can now be edited using CRISPR-Cas9 and microscopic techniques are well-established for this species. The worm’s physical transparency in particular makes it an attractive option for evolution experiments that include microscopic phenotyping. Experimentation with non-model species is typically more difficult, although that does not necessarily preclude every setup. In the absence of genome editing, a combination of fluorescent dyes and immunological staining can give an accurate representation of intracellular organisation, and adding targeted inhibitors to the selective regime can replicate some aspects of traditional gene knockouts. A second important factor to take into account when choosing new focal species for evolution experiments is the organism’s sexual cycle. Perhaps unsurprisingly, a large proportion of the literature has used organisms that reproduce asexually, which makes it easier to identify beneficial mutations and determine their effect on fitness. Sexual reproduction, on the other hand, can speed up adaptation since beneficial mutations can be combined between different lineages, and recombination can help prevent deleterious mutations from reaching fixation [45]. Accordingly, if one is also interested in characterizing evolutionary dynamics during the experiment (see further below), one has to appreciate that the organism’s sexual cycle plays a substantial role in this.

Finally, one other limiting aspect of classic evolution experiments is that selection is almost always inherently linked to an organism’s growth characteristics. Without a clear link to the cellular phenotype of interest, an evolution experiment may have outcomes that are orthogonal to the questions one would like to answer. Techniques that can couple selection based on cellular phenotypes, e.g. high-speed image-enabled cell sorting [46], to lab evolution, would therefore be ideal for expanding the scope of lab evolution protocols. Cells within evolving populations could be imaged, and those images used to select cells with desired cellular phenotypes to found the next generation. High-speed image-enabled cell sorting is able to distinguish between a vast range of cellular compartments in mammalian cells, including organelles such as the nucleus, mitochondria, and centrosomes, and can sort up to 15,000 events per second [46]. Before being able to use it with other, potentially smaller species, it will be necessary to first determine which cellular features can be resolved. In cases where a direct measurement is not possible (because of reasons related to either resolution or biology), one could still resort to using indirect reporter systems (e.g. relocalization of a reporter protein to the nucleus, or by using dyes compatible with live-cell imaging). Image-enabled cell sorting could also be used to circumvent some of the limitations of using less-established ETOs. If genome engineering proves difficult in such species, populations could be mutagenized and cells with desired phenotypes could be sorted into new populations. Such cells could then serve as ancestral populations for evolutionary repair experiments.

Evolutionary dynamics - an under-explored facet of evolutionary cell biology

So far, we have described how evolutionary data - either from phylogenetic analyses or experimental evolution - can be used to learn more about the evolution of cellular features. Doing the opposite, however, has been explored even less. Cell biology is only rarely used to gain insight into the dynamics of evolution. Experimental evolution has been crucial in elucidating different aspects of evolutionary dynamics: how mutations spread through populations, how different lineages compete with each other over time, and how epistasis determines the shape of the fitness landscape and evolutionary trajectories. Regardless, we often do not know why certain adaptive mutations are more beneficial than others and how their fitness effects are brought about on a cellular level. Which components of genetic and especially phenotypic variation are actually adaptive? What is the influence of evolutionary dynamics and population genetics on the evolution of cellular features? For example, how much do typical parameters from population genetics, such as population size and mutation rate, influence this process?

Lineage tracking using DNA barcodes is a powerful technique for studying evolutionary dynamics [47]. By introducing short, but highly variable barcodes into the population, one can track thousands of lineages simultaneously and detect the establishment of beneficial mutations at high temporal resolution. The technique was pioneered in S. cerevisiae but is now also available in E. coli [48] and human organoids [24]. We propose an experimental setup that combines this technology with fluorescence-activated cell sorting (FACS) based on a fluorescent readout that is coupled to the cellular phenotype of interest (Figure 4). At any point during the evolution experiment, phenotypic variation can be examined and coupled to lineage and mutational trajectories by FACS and genome sequencing. This approach not only allows us to explore how much variation at a cellular level contributes to adaptation, it also allows us to link these observations to mutations that arise during the same evolution experiment. One important constraint is that this technique would currently only be feasible in species with efficient transformation protocols. Additionally, the method loses power over longer timescales, when adaptive lineages overtake the population and barcode diversity is lost. Renewable barcoding systems, in which new barcodes are periodically added to evolving populations do exist and have been shown to significantly increase the time over which evolutionary dynamics can be quantified [49]. However, such protocols also impose additional bottlenecks on the population, as the transformation efficiency of evolving lineages might differ depending on the mutations they acquired.

Figure 4 |. Adding evolutionary dynamics to evolutionary cell biology.

Combining a fluorescent readout of the cellular phenotype of interest (here represented by shades of gray and green) and a conventional lineage tracking technique using DNA barcodes would be a powerful way of coupling evolutionary dynamics to evolutionary cell biology. FACS can be used to sort the population based on their cellular phenotypes, allowing us to link variation in specific cellular features to lineage trajectories and the mutations that arise throughout adaptation. This technique also allows us to add an additional dimension to conventional Muller plots (here represented by shades of gray and green).

Nonetheless, this modular setup is potentially very powerful and can be expanded as needed. One can combine compatible fluorophores to simultaneously follow different cellular processes or different parts of the same cellular pathway. Depending on the resolution of the cell sorting system, the readout can be intensity-based, based on relocalization, or on morphological variation. One can start the experiment with different genotypes and/or species to reflect observed phylogenetic trends. Sorted populations can also be subjected to further cell biological assays to obtain a more comprehensive overview of how cellular features changed throughout the evolution experiment. Collectively, this setup would give us the ability to disentangle the effects of genetic and phenotypic diversity on the evolution of cellular features.

Evolution of innovation - horizontal gene transfer and endosymbionts

Most cellular processes change gradually over the course of evolution, making them ideal for analyses such as comparative phylogenetics. However, at different points in time, organisms evolve features that were previously absent in their ancestors. The paradox of such ‘novelty’ is that these new features arose despite the fact that natural selection presumably only had ‘old’ DNA to work with [50]. Regardless, there are several examples of evolution experiments in which innovation has been observed, e.g. Pseudomona aeruginosa cells acquiring the ability to grow on new nutrients [51], single-celled yeast developing multicellularity [52], and viruses acquiring new modes of transmission [53].

However, in addition to their ‘old’ DNA, organisms can acquire new genetic material by non-vertical transmission, for example by horizontal gene transfer and endosymbiosis, which is an important mechanism of genome evolution not only in prokaryotes, but also in eukaryotes. Such genetic changes can lie at the basis of dramatic cellular innovations. Indeed, mitochondria and plastids evolved from free-living prokaryotes that were acquired through endosymbiosis more than a billion years ago [54], and in polar climates, ice-binding proteins from prokaryotic donors enable the survival of protists such as sea ice diatoms [55] and Chlamydomonas species [56].

Apart from some recent studies in bacterial species [57–59], the effects of horizontal gene transfer on cellular evolution have not been examined by experimental evolution. It will therefore be especially interesting to explore whether supplying ETOs with exogenous DNA can lead to innovation (e.g. the ability to grow in new niches), and how this is brought about within the cell. Recapitulating the early stages of endosymbiont evolution and organelle formation, unsurprisingly, is even more experimentally challenging. Nevertheless, Mehta et al. engineered a system in which they were able to introduce an ATP-producing E. coli strain as an endosymbiont of a respiration-deficient S. cerevisiae strain [60]. Such groundbreaking approaches in combination with experimental evolution will be crucial for developing a better understanding of the most dramatic innovations within the tree of life.

Concluding remarks

By integrating evolutionary biology and cell biology, with a combination of cell biological assays and comparative phylogenetics in particular, evolutionary cell biology has begun to significantly improve our understanding of the origins and diversity of cellular complexity. However, many questions within evolutionary cell biology remain unanswered, partly because of the limitations of approaches which often focus on what has happened in the remote past. Here, we discussed how experimental evolution in the lab holds the potential to bridge some of the knowledge gaps present in the field by asking what can happen, by either replaying segments of evolutionary history in real time, or by employing evolutionary repair to understand how cellular systems can circumvent breakage of specific components. We laid out some examples of integrated experimental design for evolution experiments, so that they can be specifically tailored to answer evolutionary cell biological questions, including some of the most important Outstanding Questions in the field. Consequently, we anticipate that such frameworks will enable us to answer some of the most fundamental questions pertaining to the origin and evolution of cellular life.

Outstanding questions.

• How did evolutionary processes shape cells, cellular processes, organelles, and proteins?

• How much cellular diversity is the result of adaptation? How have mutations, recombination, and genetic drift shaped cellular complexity?

• Does the evolutionary history of an organism (i.e. the mutations and cellular features it acquired along the way) contribute to future evolutionary trajectories?

• Do universal principles or regulatory networks underlie common processes?

• To what extent do shared cellular features reflect common ancestry and orthology?

• How do cellular innovations arise, and where do they occur across the tree of life?

• How do we experimentally validate comparative phylogenetic studies?

Highlights.

• Evolutionary cell biology is an emerging field that explores how cellular processes have been shaped by evolution, and conversely, how cell biological variation has influenced the evolutionary process.

• Complementing insights from comparative phylogenetics, experimental evolution is now increasingly used in combination with cell biological assays to replay segments of evolutionary history in real time.

• Formal adoption of experimental evolution into the established evolutionary cell biology toolbox requires an integrated experimental framework with tailored experimental design.

Glossary

Comparative phylogenetics

Comparative phylogenetics uses a combination of two types of data: genomic data to create phylogenetic trees and different types of species characterizations (organismal, cellular, or subcellular) that are mapped onto the trees. Differences and similarities are then used to test evolutionary hypotheses.

Epistasis

Epistasis is the phenomenon whereby the effect of one mutation depends on the presence or absence of other mutations.

Evolutionary dynamics

Evolutionary dynamics is the study of how allele frequencies in populations change throughout evolution in function of different evolutionary parameters such as selective pressure, fitness effects of mutations, and epistasis.

Experimentally-tractable organisms (ETOs)

Experimentally-tractable organisms are organisms for which we have sufficient tools to render them amenable to experimentation. At the very least, this means that the organisms can be stably cultured in a laboratory setting, but other features that contribute to an organism’s tractability are ease of imaging, ease of genetic manipulation, and the availability of omics data.

Fitness landscape

Fitness landscapes are a useful concept to get an intuitive understanding of epistasis and its role in evolution. In an adaptive fitness landscape, each xy coordinate represents a point in the genotypic space, and the value on the z-axis indicates the corresponding fitness of the genotype. In this way, evolution can be visualized as an exploration of the landscape, where selection favors genotypes that climb a fitness peak.

Genetic interaction network

A genetic interaction network consists of a network of genes or gene products, linked together by pairwise genetic interactions. Two genes are said to have a genetic interaction when mutations in both genes result in a phenotype that cannot be explained by adding the individual effect of each mutation (i.e. there is epistasis).