As of today, it is 8 years ago that I presented a paper for a course, and I recently got reminded of this work. The paper, written by Espinosa-Soto et al. (2011)1 was about phenotypic plasticity (the same genotype giving rise to alternative phenotypes) and how it can help or hinder evolution. Since then, a lot of single-cell measurements have been done on microorganisms, revealing just how dominant phenotypic heterogeneity is in bacterial populations. For example, take a look at the awesome video by Alma Dal Co, revealing how microscale gradients in clonal populations give rise to cross-feeding and antibiotic tolerance:
Today, having gone back to the paper I had read all these years ago, I feel it is still very interesting, and very clearly written. I wanted to write down the gist of it. So. Here it goes.
Firstly, let’s consider the fact that evolution by natural selection is not selecting for genotypes, but for particular phenotypes. This means that, while maintaining a particular phenotype, evolution is “free” to drift to any other genotype, given that they 1) produce the same phenotype, and 2) are mutationally connected (see top panel of illustration below). All positions on such a genotype network are typically assumed to be equally fit, because they give rise to the same phenotype. In other words, neutrally drifting over this genotype space does not directly influence the phenotypes which are produced. As an aside, these positions on the genotype network do have access to different mutants, meaning that they may still give rise to a different shapes and forms through mutations. However, we will now discuss how different phenotypes can be explored through phenotypic plasticity.
Let us now include the fact that the translation from a particular genotype into a phenotype isn’t fully predictable. For example, many biologically relevant structures such as RNA molecules, proteins or gene regulatory networks, often have alternative outcomes. RNA molecules and proteins have alternative foldings, and gene regulatory networks can reach alternative expression patterns. This can happen through environmental noise, or noise intrinsic to the system. This means that one genotype does in fact not always lead to one resulting phenotype, but that these structures may have a spectrum of alternative phenotypes. Moreover, each of the genotypes on the mutationally connected network, may have their own alternative phenotypes (depicted in the bottom panel of the illustration). Now we can ask ourselves the question: does this phenomenon of phenotypic plasticity help, or hinder evolution towards new phenotypes?
Espinosa-Soto et al. discuss the conditions which have to hold for phenotypic plasticity to help evolution find new “better” phenotypes (see image below). They do so by postulating the following scenario. Consider a population which is adapted to a certain phenotype (P) in search of a new phenotype which is fitter (P*). While drifting along the genotype network , genotypes may arise that, sometimes, give rise to P* as an alternative phenotype. Over time, these genotypes will accumulate as they sporadically produce P* instead of P (i.e. those genotypes are slightly more fit than the other genotypes in the neutral network). These newly accumulated genotypes may in turn give rise to genotypes where P* occurs more often. Eventually, the population is located in a part of genotype space that produces P* much more than it produces P (potentially even avoiding P, as it was less fit).
In the above scenario, phenotypic plasticity helps because the genotypes that produce P* “by accident” somehow have better access to those genotypes that produce P* more frequently, and eventually even produce it natively. This is a nice story, but what needs to be true in order for this to actually speed up evolution from P towards P*?
Espinosa-Soto et al. formulate 4 requirements for phenotypic plasticity to help speed up evolution:
- Finding genotypes which produce P* through plasticity must be easier than finding the genotype network of P* (otherwise, it makes no difference)
- Genotypes near the genotype network of P* must have the tendency to produce P* by plasticity
- Genotypes that produce P* through plasticity must be mutationally connected on the genotype network of P, or the evolutionary search would not be able to progress towards the genotype network of P*
- Genotypes close to the genotype network of P* must more often produce P* through plasticity than genotype far from P*
Condition 1-3 are all very similar: there must be a mutationally connected genotype network P that produces P* as its alternative phenotype. The 4th condition is however more interesting, as it would cause selection to promote the stabilisation of P* through gradual increases in the P*‘s penetrance: how often it occurs “by accident”.
The authors next go on to show how these conditions do indeed hold for transcriptional regulation circuits. In order not to bore you, I think I will just leave you with the summarising figure from the article.
To wrap up, the paper is called “Phenotypic plasticity can facilitate adaptive evolution in gene regulatory circuits” (Espinosa-Soto et al., 2011). Give it a read if you found this blog interesting. I for one think it is really well written, and it was fun to rediscover after all these years. Perhaps, evolution in nature is also very much influenced by phenotypic plasticity!