As John Maynard Smith and Eros Szathmary in their highly influential book realised, the rise of biological complexity has been marked by a small number of events in which self-replicating entities, by various means, align reproductive fates and come to replicate as a single collective. Such transitions include the evolution of networks of self-replicating chemistries (from single autocatalytic reactions), the evolution of chromosomes (from once separate genes), evolution of the eukaryotic cell (from ancestral archaebacterial and eubacterial cells), the evolution of sex (from asexual types), the evolution of multicellularity (from single cells) and evolution of eusociality (from asocial types).
The following is adapted from Rainey et al (2017): A central issue for each transition concerns the emergence of Darwinian properties of variation, reproduction and heredity without which the process of evolution by natural selection cannot occur. With exception of the transition from non-life to life, these properties manifest at the lower level, but their existence at, for example, the level of the individual cell, does not mean that these same properties automatically appear at the level of collectives of cells. Consider a nascent multicellular collective that arises from a mutation that causes cell adhesion. The collective that results grows by virtue of the reproductive capacity of individual cells, but in the absence of a means of collective level reproduction, the collective – like soma – is an evolutionary dead end. This leads to a very particular dilemma, namely, the need to explain the evolution of Darwinian properties, in non-Darwinian entities (e.g., the first collective), by non-Darwinian means (it is not possible to invoke the thing one wants to explain (the evolution or reproduction) as the cause of its own evolution).
One potential solution comes from recognising the role that the environment can play in exogenously imposing Darwinian properties on otherwise “unwitting” particles. The idea, in general form, we refer to as “ecological scaffolding”. It has been further developed in collaboration with Andrew Black and Pierrick Bourrat. A preprint is available here. A principle motivator behind the concept came from a sense that there was something general about the ecological context of our experimental work on the evolution of multicellularity.
Imposing Darwinian properties on otherwise “asocial” particles requires nothing more than patchily distributed resources and a means of migration between patches. For Pseudomonas growing in unshaken liquid, the limiting resource is oxygen but this can be obtained at the air-liquid interface. This requires types that can colonise this environment (it is largely inaccessible to the ancestral type), which in turn requires mutations that cause overproduction of adhesive polymers leading to the formation of WS mats. However mats are not buoyant and cannot remain at the air-liquid interface without a physical structure to which attachment is possible.
Consider a pond with randomly placed reeds, each of which constitutes a scaffold around which a mat can form. Mats eventually collapse and go extinct due to their increasing mass, but provided a mat can re-establish at the original reed, or around one or more new reeds, then a process akin to collective-level reproduction occurs. With this comes the possibility of a Darwinian process at the level of mat collectives. Importantly when this occurs, selection works with potency over a timescale that is longer than the doubling time of cells. Just how the reproductive event occurs is not trivial, but possibilities range from simple fragmentation (although this is unlikely to be successful over the long term), to the possibility that cheating types arising within mats act as dispersing agents — rather like a germ line — with capacity to regenerate the mat-forming type. See the cartoon in the section on the evolution of multicellularity.
Under a scenario like this, ecology is everything: the structure of the environment permits realisation of Darwinian properties at the collective level even in the absence of these properties being endogenously determined. As Hammerschmidt et al (2014) showed, there exists ample possibility for Darwinian properties to become endogenous. A first step observed through experimentation was evolution of a simple but effective genetic switch that allowed reliable transition between soma- and germ-like phases of a two-stage life cycle that marks a first step in the evolution of development.
When Darwinian properties manifest over time scales that exceed that of the doubling time of individual cells a process of special evolutionary significance begins to unfold. Required are conditions that ensure that particle-level variation is partitioned into discrete collective-level packages and that these packages manifest heritable variance in fitness. Once achieved, then selection working over the longer timescale, defined by the birth (and associated death) of collectives, stands to trump selection working on the short timescale (for example, the doubling time of particles). This causes particle fitness to increasingly align with the fitness of collectives. Because particle growth rate is no longer the focus of selection, drift becomes an important factor allowing individual particles (and thus collectives) to explore a much greater range of phenotypic possibility. For example, the process allows evolution of phenotypes that are altogether new and that emerge from the action of collective-level selection. The possibilities arising from implementation of a second collective-level timescale extend beyond research in evolutionary transitions and opens new ways of thinking about how experimental evolution might be used to engineer communities to generate new functions, chemistries and processes with application in biotechnology, medicine and agriculture.
At the core of these possibilities are emulsion-based technologies known as micro- and milli-fluidics. By dividing a community of cells into discrete parcels by means of enclosure within a droplet surrounded by oil, variation at a level above that of the individual cell is achieved. Droplets, harbouring up to 106 – 107 cells, can be manipulated in a myriad of ways: the activity of contents can be monitored by fluorescent-based assay (in real time), contents replenished, sampled, and substrates added; droplets can be diluted, sorted, merged, and they can be split. The latter being akin to a droplet-level reproduction event and immediately affords the possibility of implementing a birth / death process based on some community level readout of performance, over a longer time scale than that of the doubling time of individual cells, with potentially dramatic effect.
We anticipate the combination of technology and its uses to explore evolutionary process over two or more timescales to provoke a wave of new theory and experimental possibility. Such approaches have already been used to investigate the importance of spatial structure and migration on the evolutionary dynamics of bacterial populations. But there exist — through ability to miniaturise and parallelise — numerous ordinary uses beyond the imposition of Darwinian properties on populations and collectives, including for example, ability to re-run Lenski’s experiment with thousands of replicate populations and with automated serial transfer occurring the moment cells reach late exponential phase thus significantly decreasing the time required to pass through thousands of generations. Using standard serial transfer approaches, a train of droplets can be constructed so that in a single experiment the evolutionary response to multiple different environments can be studied. Additional possibilities exist for analysis of evolution in small populations, for testing ideas concerning the evolution of new genes, for systematic analysis of the effects of drift, migration and environmental fluctuation and, if coupled to genomic and phenotypic assays studies that connect genotype to phenotype can move to a new high throughput era. But emulsion technologies are not the only possibility for manipulation of experimental populations. Interesting opportunities are set to emerge from electrowetting that allows the same set of possibilities afforded by emulsion-based technologies, but without the need to package droplets within a matrix of oil.
Building upon this is some super theoretical work by Guilhem Doulcier (and here) at ESPCI and in collaboration with Silvia De Monte and Amaury Lambert that looks at the evolutionary emergence of community-level heredity. This work lays the foundations for planned experiments with a new microfluidic device developed by Jerome Bibette’s (LCMB) lab in which it is possible to implement selection on droplets using a device akin to a Turing Machine.