This week’s BEACON Researchers at Work blog post is by University of Texas at Austin faculty member Jeffrey Barrick.
For a long time, I thought that I’d become a synthetic organic chemist. Synthesizing intricate molecules would be a natural next step from a childhood (ok: permanent) obsession with building blocks: first Duplo, then Lego, and finally Atoms. This train of thought was derailed when, as a college freshman, I learned about methods that biochemists had devised to recapitulate the process of Darwinian evolution (select, replicate, repeat) on RNA and protein molecules in a test tube.
I was hooked. Why build one specific molecule or protein sequence when you could kick things up a level? You didn’t necessarily need to design a protein that would fold into a precise three-dimensional structure and speed up a specific chemical reaction (which is still an extremely difficult scientific challenge). Instead, you could design a clever test, crank the Darwinian cycle, and be surprised by what crawled out. Evolution would teach you the relevant biochemistry.
I did eventually find some “winning” proteins that solved a binding challenge in an unexpected way, but my research experiences also brought me face to face with the limitations of evolution. Sometimes you just didn’t get any solutions if you pushed too hard. There might not be any molecules in your test tube that could pass the test. Nature has eons for evolution to generate and test so many sequences that it can eventually get around many of these walls, but we can’t always wait for those rare innovations in our lifetimes. So, like many in the BEACON community, I started to think about how can we force, coax, and jumpstart evolutionary processes so that they will more quickly discover better solutions to more challenging problems.
My focus shifted up a level of biological organization to bacteria in a postdoc with Richard Lenski at Michigan State University. We began sequencing genomes from one of twelve flasks of the lab’s 20-year evolution experiment with Escherichia coli (the subject of recent posts by Mike Wiser and Caroline Turner). One striking observation from our work was that the rate of sequence evolution was really quite slow. The genomes of these bacteria had only experienced one new mutation every 500 generations (roughly 75 days). Furthermore, even though these bacteria have lived in the laboratory under conditions where a majority of their genes are dispensable, only about 1% of their chromosome has been deleted after 20 years.
So, bacteria have evolved to be conservative in an evolutionary sense. This makes sense: mutations in genes are more likely to be disruptive than beneficial. It’s easier to “break” a protein and unbalance regulatory networks, metabolic pathways, and cell physiology with a random change than to improve this well-tuned engine. Mutations that remove the coding regions for entire genes are even worse. They can cause a bacterium and all of its descendants to completely “forget” how to utilize some nutrient. This loss might not have any immediate ramifications, but those genes might prove crucial for survival later if the environment changed.
It turns out that 6 of the other 11 flasks are now dominated by bacteria that are risk takers. The winning bacteria in these flasks are “hypermutators.” They accumulate mutations at 10–100 times the rate of their ancestors because they have evolved defects in DNA repair and proofreading. Why? Having an elevated mutation rate is not immediately beneficial—in fact, it can be deleterious for the reasons discussed above. However, hypermutators also have more chances to throw the mutational “dice” and generate descendants with highly beneficial but very rare mutations or to “yahtzee” together multiple beneficial mutations more rapidly. So, hypermutators can adapt more quickly than their competitors and win when opportunities outweigh risks. This result can be described as “second-order selection” for greater evolutionary potential.
We recently studied how differences in evolutionary potential of another kind were important in the history of another flask. This E. coli population had diverged from its ancestor into two contending groups characterized by different beneficial mutations by 500 generations. Individuals in one group had evolved a significant fitness advantage over the other, but it turned out that descendants of the less-fit group later prevailed and drove the ones that had been ahead extinct by 2,000 generations. That could have meant that the ones that were behind were just lucky and happened to “draw” an extremely beneficial mutation that caused them to leapfrog their more-fit competitors. However, by replaying evolution many times from these different starting points, Robert Woods was able to show in his thesis research that neither of these explanations was the case. The eventual winners were actually able to out-adapt the group that had been ahead earlier, on average.
To understand why this was the case, we sequenced the genomes of bacteria from these experiments. None had become hypermutators. Instead, we discovered that each group of bacteria had a different, beneficial mutation in a protein that regulates how tightly the DNA of the bacterial chromosome is wound (i.e., supercoiling). Supercoiling affects how accessible the genome is to polymerases that must unwind the chromosome to transcribe RNA. Therefore, this protein is a global regulator, and mutations in it can alter the expression levels of hundreds of cellular proteins. Interestingly, the bacteria that were initially less fit, but later won, often had subsequent, highly beneficial, mutations in a second global regulatory protein that turns many genes on and off in response to starvation, but mutations in this protein were never found in descendants of bacteria from the group that lost.
Tim Cooper’s lab at the University of Houston mixed and matched these key regulatory mutations together in different combinations to definitively show that the group that had been winning early in the experiment had, in a way, painted itself into a corner. Despite having the most beneficial set of mutations early, further mutations — including key ones in the starvation response protein — were no longer as beneficial in these bacteria. Furthermore, although each group had mutations in the supercoiling regulator early on, only the mutation from the eventual losers restricted further evolution. The slow-but-steady lineage that won in the end had maintained its capacity to benefit more from mutations in the starvation response gene (and probably elsewhere). This situation can be described as second-order selection to maintain a more evolvable “genetic architecture”.
In January, I moved to the Department of Chemistry and Biochemistry at the University of Texas at Austin. My new lab is using systems biology tools to better understand how the interactions between global regulatory proteins, such as these, can affect evolutionary potential, and we’re also looking for other eventual winner and eventual loser stories in hopes of discovering some commonalities. The eventual goal is to use synthetic biology and computational approaches to test ideas for how we can isolate, engineer, and evolve bacteria that are more capable of rapidly evolving in useful and creative ways.
As you can see, even though bacterial evolution appears to be fairly conservative most of the time in nature, it’s clear that both risk-takers and prudently mutating lineages can evolve as a second-order effect of natural selection. This gives us a chance to learn something from an evolution experiment about what it means on a biochemical level to unleash evolutionary potential.
You can learn more about the Barrick lab’s research on their website.