Environmental change influences the pathway to the evolution of antibiotic resistance

Observed changes in (a) global average surface temperature; (b) global average sea level from tide gauge (blue) and satellite (red) data; and (c) Northern Hemisphere snow cover for March-April. All differences are relative to corresponding averages for the period 1961-1990. (IPCC 2007)

Observed environmental changes. All differences are relative to corresponding averages for the period 1961-1990. (IPCC 2007)

How will populations respond rapidly changing environmental conditions? We’ve all seen the imagery of the polar bears surrounded by thawing ice sheets, but this isn’t just a problem of the environmentally concerned. The rate of environmental change may be dramatic and making economically relevant impacts on our everyday lives. It seems obvious to scientists that global change is occurring (IPCC 2007). How do organisms respond, not just on an ecological basis, but also in an evolutionary sense? Microbe based experiments can help us understand the evolutionary processes that come into play in rapidly changing environments.

A recent paper (Lindsey et al., 2013) does just this…

The authors use replicated microbial populations and expose them to a changing environment. In this case, the changing environment is increasing concentrations of a deathly antibiotic. The authors then ask, by sequencing a gene responsible for antibiotic resistance, if the adaptations or mutations that occur are the same across all populations. In essence, the researchers are trying to understand how the bacteria solve the challenge of the harmful environment. In the case of a rapidly changing environment, there are only a handful of solutions and most of the test populations go extinct before the mutations occur. For populations that experience a slow increase in the deathly poison, there appear to be many more ways to evolve resistance. What is especially fascinating about this research is that it appears that these pathways to resistance are only available when the environment changes slowly. Here those intermediate steps are only revealed at intermediate concentrations of the antibiotic.

Beyond Open Access: In a bold move of scientific transparency, you can go beyond the Nature paper and the supplemental information and discover the details for this project on its very own RifRamp Project website. There you’ll find well-organized and annotated data, molecular protocols, as well as computer code for supplementary simulations and generating their amazing visualizations. I commend the authors for taking this extra step in data sharing.

So let’s get to the details of how they figure this all out.

The experiment: Experimental evolution using populations of E. coli in various concentrations of the antibiotic rifampicin. That means hundreds of individual wells in a plate filled with independent populations. All of the populations were initially grown without antibiotic and were later exposed to increasing concentrations of a deathly antibiotic. In order to understand the process of adaptation to environmental change, the experimenters used three different treatments: 1) a gradual increase in concentration of the abiotic at each transfer where the maximum concentration (190 ug/ml) is reached during the last transfer; 2) a moderate increase in concentration (twice the rate as the gradual), so the maximum concentration is reached half way through the experiment; and 3) a sudden shift to the maximum concentration of antibiotic during the first transfer.

While the end concentration of antibiotic was the same across treatments, the time spent in lower, sub-inhibitory, concentrations differed across treatments. The authors asked

…does evolutionary rescue occur by the same set of mutations in all treatments?

What they found: The highest rates of survival occurred when the populations were gradually exposed to the antibiotic. Nearly 90% of the gradual treatment populations survived the antibiotics and almost 50% of the moderate populations survived. In contrast, less than 2% of the populations with the sudden shift survived the course of antibiotics. Measurements of growth rates among the end populations revealed significant differences among treatments suggesting that different mutations had become fixed in the different treatments even though they were adapting to the same antibiotic.

Using previously published research on the evolution of resistance to this particular antibiotic, the authors sequenced a single gene (rpoB) from samples of each of the different treatments. There was variation in the mutations within each treatment. For instance, four different mutations occurred within the Sudden treatment. However, no population in the Sudden treatment had more than one mutation. In stark contrast to this, the other treatments where the antibiotic concentration had been more slowly increased, the majority of populations had fixed multiple sequential mutations. The authors went back to the freezer, and through careful sequencing of a population through time, were able to determine the order of mutations that fixed within a particular population.

The significance of multiple mutations

For those populations that contained multiple sequential mutations, the authors created mutants with the just the first mutation and measured how they grew in different concentrations of antibiotic. They found that the these early mutations yielded high growth rates at low to moderate concentrations of antibiotic; they grew faster than the ancestral populations without the mutations. However, when the concentrations got very high, the early single mutations failed to provide resistance. The authors posit that

… subsequent mutations are required to salvage the lineage at the highest concentration. Such secondary mutations might only be selectively accessible after the first mutation, which might only be selectively accessible under sub-maximal antibiotic concentrations.

Sign Environmental Epistasis

Modified from figure 3 of Lindsey et al 2013. Fitness landscape of four genotypes (ab, Ab, aB, and AB) in two different environments (x and y).

Sign environmental epistasis, what the heck does that mean? The authors provide an abstract example to help illustrate their point. First, imagine a simple fitness landscape (for more on that check out this book) of just four genotypes (two genes with two alternative versions possible at each gene). The above figure shows the same genotypes in two different environments (x and y). In environment x, the ab genotype has the highest fitness. A mutation to either gene (ab -> aB or Ab) results in a decrease in fitness (highlighted by the yellow arrow). However, if the environment changes to y, which could represent a different concentration of antibiotic, the ab genotype is no longer the fittest of the bunch. Any single mutation results in an increase in fitness (highlighted by the other yellow arrow). This is an example of sign environmental epistasis, a type of GxE (genotype-by-environment) interaction. Now imagine that there are many different types of environments and the fitness landscape continues to change, the pathway of mutations that evolve within a population may be determined by the changing fitness landscape as influenced by the changing environmental landscape.  As the authors say

Each step in a selectively accessible path involves an increase in genotype fitness. However, because the fitness of genotypes can change with the environment, the historical sequence of environments can qualitatively affect which paths are selectively accessible. Consider a path from one genotype to another that is selectively accessible under a sequence of distinct environments.

This pathway then becomes historically contingent. The sequence of mutations is determined partially by the sequence of environmental conditions.

To return to the experiment at hand, the authors recreated these same kinds of fitness landscapes for all of the mutations found in particular populations at many different antibiotic concentrations including the initial antibiotic free state and the final maximum concentration. Using these measures of fitness, they were then able to determine the likely pathways available to the microbes. In some examples, the set of mutations that fixed would only be possible when the intermediate concentrations of antibiotic were present and changed the shape of the fitness landscape. That is, the pathway to antibiotic resistance was historically contingent on the changing environment. Importantly, a rapidly changing environment would not have made these pathways available. The slowly changing environment generated sign environmental epistasis that allowed a population to cross fitness valleys.

Now what about the polar bear now surrounded by a sea instead of ice? This study found that not only was the particular sequence of mutations that evolved dependent on the changing environmental conditions, but also the probability of finding a solution decreased as the rate of change increased. “In our system, we find that rapid environmental change closes off paths that are accessible under gradual change.” As anthropogenic climate change increases, the ability of organisms to respond to this change may decrease. This is a positive relationship we want to avoid. On the other end of human induced catastrophes, a word about antibiotic resistance from Lindsey et al: “… low drug concentrations can evolutionarily ‘prime’ bacterial populations by bringing a population (mutationally) closer to genotypes that would enable drug resistance at high concentrations.”

A few further thoughts:

On the process of doing research: Often, when I read a paper, I think, “wow, that was a fortuitous choice of a 15 molar concentration” or something like that. However, it is rarely a happy accident. My current postdoctoral advisor, Michael Wade, was talking about this very subject the other day when describing the particular population size that he used in a group selection experiment. He explained that choosing 16 beetles per container was based on 10s of years of research from his advisor in the same system. Knowing quite a bit about the natural history of an organism can help produce better-designed experiments. One can see this same kind of logic explained in the “Sampling rationale” section in the methods in this highlighted paper. I find these types of statements from the authors helpful as someone without much knowledge in this experimental system. It provides me a little bit more insight into the process of designing the experiment.

Scholarship beyond a well-designed experiment: As an addendum to the excellent research, I’d like to point out that this paper was simply a pleasure to read. As we are being slowly overwhelmed with new research coming out, it is nice to sit down with a manuscript that not only provides some excellent research, but also coveys the story of the science.

Reference

  • Lindsey HA, Gallie J, Taylor S, Kerr B (2013) Evolutionary Rescue from Extinction Is Contingent on a Lower Rate of Environmental Change. Nature 494: 463-467. DOI: 10.1038/nature11879