On Thursday, March 13, the Department of Biology hosted Greg Lang, associate professor of biological sciences at Lehigh University. Lang delivered a talk titled “Genome evolution in laboratory populations of yeast” as a part of the Department’s Seminar Series, and explored his lab’s work conducting experimental yeast evolution on the bench.
Yeast — particularly saccharomyces cerevisiae — have become ideal eukaryotic model organisms due to their extremely fast generation times, ease of genetic modification and well-annotated genomes.
“Yeast grows very quickly, so we get 10 generations of growth in a single day and over 3,000 generations of evolution in a year,” Lang said.
However, even with the yeast system’s inherent advantages, studying evolution is not easy. Experimental evolution requires thousands of initially equivalent populations all undergoing thousands of generations of growth, which necessitates constant and precise maintenance. However, the group has developed a mechanized solution to this problem.
“My lab uses robotics,” Lang explained, “It takes about half an hour to transfer 1000 [yeast] cultures. This combination of automation and miniaturization allows us to do really massive-scale, high-throughput experimental evolution in the lab.”
Lang’s system also utilizes the ability to capture and characterize particular samples at different evolutionary stages.
“We can freeze everything down, so we have our frozen ‘fossil’ record of all of our populations,” Lang said. “We can go back to any population at any time point, pull it out of the freezer, measure its fitness, sequence the genome, or even do evolutionary restarts — we can rewind the tape of life and let [evolution] play out again from just before or just after an interesting event.”
Since these experiments are performed in controlled laboratory conditions, the team can modify the yeast’s genome or external environment in specific ways to study the results of these modifications on evolution.
“There's a lot we can see about dynamics, types of mutations and how selection is activated. So this is a really personal system,” Lang explained. “We've done stuff with haploid, diploids, different strains and species. We can change the environment we grow them in.”
In the first project Lang described, he subjected haploid, asexually reproducing yeast to normal growth conditions — 30-degrees-Celsius temperatures and glucose-rich media — and let evolution play out for 1,000 generations. Three populations were isolated: one “ancestral” early clone from generation zero, one “intermediate” clone from the middle of the experiment and one “late” clone from the thousandth generation. To determine the relative fitness of these clones, Lang co-cultured the different yeast populations and measured their resulting growth rates. In nearly every population, the late clones exhibited higher fitness than the intermediate and early clones, indicating adaptive evolution. However, there was one outlier that did not exhibit this expected phenomenon.
“Simple math and transitivity predicts that we should get a 5% fitness advantage when we compete the late clone against the ancestor,” Lang argued. “However, to our surprise, the late clone actually lost in direct competition to the ancestor.”
Lang realized that this was a “nontransitive” evolutionary phenomenon, meaning that a series of adaptive evolutionary events had led to organisms that were more fit than a recent ancestor but unexpectedly less fit than a distant ancestor. At first, the group was perplexed by this result, as it was the only population in their experiment that exhibited nontransitivity. However, upon further analysis, the group found the unlikely cause of their observation.
“There are only a few phenomena that give rise to these nontransitive interactions and to positive frequency dependence, and these are usually toxin-antitoxin systems,“ he stated. “Yeast are known to contain a yeast killer toxin, which is a toxic-antitoxin system.”
According to Lang, a “killer” toxin virus is a double-stranded ribonucleic acid that encodes one open reading frame: the toxin. However, the toxin isn’t able to replicate independently, so it requires another “helper” virus to spread, which contains additional components that the killer virus uses to complete its viral life cycle. However, Lang and his lab had initially believed that his yeast strains did not contain this yeast toxin system, which made its discovery more surprising.
To test if the killer phenotypes changed during evolution, Lang’s team developed a high-throughput assay to determine yeast-killing ability and used it to test different generations of yeast.
“Half of our populations evolved changes to the killer phenotype — they either lost or had an attenuation of killing ability,” Lang explained. “We had an early clone that could both produce and resist this toxin. We then had evolution, and by the time we got to the intermediate clone, we fixed three nuclear mutations in the yeast genome and it lost the ability to produce toxin. Then the late clone fixed an additional ten mutations and then it lost the ability to be immune to toxins.”
Though this non-transitivity only occurred once in around 600 experimental evolution populations, it reflects a common theme in evolution.
“Sometimes evolutionary landscapes take a walk that looks more like the Penrose staircase, where it seems like [fitness] keeps going up at every individual step, but in reality, [the evolved organism] could be less fit than it was 1,000 generations ago,” Lang concluded.
After extensively using experimental evolution to study the yeast model system, Lang and his team began to recognize the potential of the yeast evolution system in studying human health and disease.
“We realized that experimental evolution is a really good tool to fish out [protein] interactions... We thought that we should try to put some rare human disease genes in our populations, let them grow...and ask if we can identify any kind of genetic interactions that may not have been previously known,” Lang said.
The first group of rare diseases Lang studied were congenital disorders of glycosylation (CDGs), which disrupt the proper glycosylation of proteins and lipids in the body. The dysregulation of this process leads to symptoms including poor growth, developmental delays, liver disease and heart problems. There are over 130 known CDGs affecting thousands of patients worldwide.
“Intriguingly, [through our yeast CDG evolution] we see an overrepresentation of genes that are also associated with variants of congenital disorders of glycosylation in humans,” Lang stated. “While some of these were known genetic interactors, a lot of them were previously unknown.”
Lang explained that although rare diseases are individually uncommon, the sheer number of them combine to create a tricky problem for researchers to study.
“Right now, it's estimated that there are about 7,000 rare human diseases affecting about 300 million people worldwide. By definition, no rare disease affects more than 40,000 people, so... even though they have a really large global footprint, they’re individually under-studied,” Lang explained. “Sometimes, the molecular mechanisms are completely unknown. We are pushing the idea of building an extensive yeast collection of rare human diseases.”
To conclude his talk, Lang summarized his insights and emphasized the strong potential of the yeast experimental evolution system in progressing translational medicine.
“We learned a lot about the dynamics of adaptation, we learned some fundamental principles and we have built a really nice tool for fishing out interactions,“ he said.
