Protocol used in biology labs around the world affects mutations in unexpected ways

Mutations in laboratory populations of simple organisms such as bacteria are of central interest in the context of experimental evolution and a major concern for bioengineers whose goal it is to equip them with evolutionarily stable added functionality. This study shows that the changes in population size as they are imposed by common experimental protocols can have a strong impact on the likeliness for mutations to spread through the population and makes quantitative predictions how the spectrum of successful mutations is shaped by these population dynamics.

HFSP Cross-Disciplinary Fellow Philip Bittihn and colleagues
authored on Thu, 09 March 2017

For long term studies, liquid bacterial cultures in biology labs all around the world undergo a procedure known as “serial passage”: A small fraction of a saturated culture is transferred to fresh culture medium, where cells can multiply until they have used up the nutrients and a new round of dilution is necessary (see figure). At the same time, random genetic mutations occur constantly due to a variety of mechanisms, including environmental perturbations and intrinsic errors upon genome duplication. Once a mutation occurs in a single cell, there are two possibilities:the mutation can either be lost from the population because the mutants happen to be the cells that are removed, or the mutants take over the entire population, a process known as “fixation”. Beneficial mutations, i.e. those conferring a fitness advantage, are the driving force of evolutionary adaptation, and so their fixation is of central interest to evolutionary biologists and the subject of many laboratory-scale experimental evolution studies employing the serial passage protocol. To other scientists, beneficial mutations that get fixed in the population pose a problem: For example, in synthetic biology, bacteria are genetically modified to carry out additional tasks. In many cases, it is beneficial for the host cell to deactivate or at least somehow impair such synthetic functionality, because this would lessen the metabolic burden therefore lead to an increased growth rate. In this case, beneficial mutations are a threat to the stability of synthetic gene circuits, and higher fixation probabilities correspond to a shorter average “time to failure”.

Figure: In the serial passage protocol, a small fraction of dense bacterial culture is transferred to fresh growth media, leading to phases of exponential population growth alternating with instantaneous reductions in population size.

Working in a synthetic biology lab, HFSP fellow Philip Bittihn and colleagues asked themselves whether the repeated reduction in population size during serial passage could have an effect on the fixation probability of beneficial mutations and could therefore influence the outcome of experimental evolution studies and the stability of synthetic circuits. At first, it seems that the dilution of a liquid culture is a fair sampling process that does not impose any additional selective pressure, and so one might be led to believe that fixation probability should not be affected. Prior theoretical studies on this subject indicated that it could have an effect, but did not cover many of the relevant scenarios, and so Bittihn and colleagues set out to start their own theoretical investigation. Reducing the problem to a simple and well-established model of cells dividing stochastically by binary fission, they compared how the fixation probability of a mutation with a certain selective advantage differs between the serial passage protocol and the baseline scenario of a population whose size is kept constant by removing random individuals as cells divide. While the baseline case is one of the standard problems of population genetics that has long been solved, the serial passage scenario yielded surprising results.

The study shows that, compared to a constant population size, beneficial mutations are generally less likely to get fixed under the serial passage protocol. Of course, depending on how experiments are carried out, mutations might also appear in the population at a different rate if the time-averaged population size is different between the two scenarios, but this would be in addition to the discovered effect which was measured for equal average population sizes. It is purely a result of the different statistics of bacterial growth and the removal of cells in the two scenarios after the appearance of a single mutant. The fraction of beneficial mutations eliminated by serial passage depends on the dilution factor upon passage—stronger dilutions eliminate more mutations. Strikingly, however, the extent to which beneficial mutations are suppressed also depends non-trivially on the selective advantage of the mutation and the population size. In particular, mutations with certain intermediate fitness gains are maximally suppressed, while weakly and strongly beneficial mutations are less impacted by serial passage. This has the intriguing consequence that the spectrum of beneficial mutations which eventually become fixed is shaped by the experimental protocol itself, because the imposed population dynamics act as a biased filter. The bias poses a problem: for example, the rate of beneficial mutations inferred from evolution experiments would be distorted if the simple but well-established formulas for the fixation probability in constant populations were used.

To improve this situation, quantitative analytical predictions for the fixation probability under serial passage are derived in the study which, taken together, cover arbitrary population sizes, dilution factors and selective advantages. In numerical simulations, the results are also shown to hold when a more realistic model for cell division is used, which suggests that the discovered effects are quite robust and should persist even as the model is extended, thereby serving as a baseline for further research. The results are a first step in helping scientists to account for the effect of serial passage, e.g., to compare evolution experiments under different conditions or to correctly reverse calculate the rate with which beneficial mutations first occur. To date, mostly neutral mutations have been used to connect the underlying mutation rate to the evolutionary dynamics observed at the population level in order to circumvent these problems. At the same time, the results may provide a strategy to limit the impact of beneficial mutations and thus increase the stability of gene circuits in biotechnological applications by shaping the imposed population dynamics as much as practically possible.

Reference

Suppression of Beneficial Mutations in Dynamic Microbial Populations, P. Bittihn, J. Hasty, and L. S. Tsimring, Phys. Rev. Lett. 118, 028102 (2017).

Pubmed link

Link to article