Evolution has conflicting opinions about spontaneous mutations. Spontaneous mutations produce the genetic variation that drives evolution in all organisms, but at the same time, most mutations that affect fitness are harmful for the organism. Despite being a pivotal component of evolution, our understanding of mutations is limited.
To understand the role of mutations in evolution, the following basic questions are essential: What types of spontaneous mutations occur? How frequently do they occur?
Such simple questions are surprisingly difficult to answer, but a recent study by Yuan Zhu (Zhu et al. 2014) has harnessed the power of next generation sequencing to get a better answer to these questions for the budding yeast S. cerevisiae.
A tricky measurement
Obtaining an unbiased measurement of spontaneous mutations is challenging because the nucleotide changes we observe can be biased by selection.
Imagine we sequence the genome of a yeast cell. Let’s then wait several generations and sequence its progeny. Sure, you may find a number of mutations – but is that really an unbiased measurement of spontaneous mutations? No! Progeny with deleterious or harmful mutations will be out-competed by others, so they will not be sequenced. In other words, this approach misses the deleterious mutations.
So, what’s the proper way to measure spontaneous mutations? A few methods have been used:
One method measures the mutations that occur in a location with no fitness effect, such as a pseudogene. Because mutations can occur within this location without affecting the fitness, this provides an unbiased picture of mutation rate. The downside is that the frequency and type of spontaneous mutations differ depending on the location of the genome, and this method is restricted to studying only one particular location.
To get a genome-wide approach, a pedigree-based method can be employed. This method looks at the differences between the parents and offspring, which provide an unbiased view of the mutational landscape (except for the most deleterious mutations). This method, however, is unfeasible for measuring many mutations because of the paucity of mutations occurring within one generation.
But not all hope is lost because there is a method that addresses both of the weaknesses presented above. This approach uses mutation accumulation (MA) lines – passaging an organism from generation to generation at very small population sizes. Maintaining small population sizes eliminates the effect of selection because the progeny are not allowed to compete with each other. The passaging can be performed for several generations resulting in a large number of spontaneous mutations (except again for the strongest deleterious mutations). MA lines have been studied for over 50 years in yeast, fruit flies, and nematodes, and have both reinforced and altered our understanding of spontaneous mutations and the distribution of fitness effects (Halligan and Keightley 2009).
300,000 generations of yeast
Although MA lines have been studied for so long, constraints in sequencing ability have prevented a large-scale analysis of the mutations. Combining the sequencing prowess of the Petrov Lab at Stanford and the MA lines from the Hall Lab at the University of Texas, Yuan and colleagues sequenced the genomes of 145 budding yeast diploid MA lines that were passaged for ~2000 generations each. This amounts to roughly 300,000 generations of mutation accumulation! Using innovative analyses to identify mutations with high confidence, they found numerous single nucleotide mutations, in addition to indels, CNVs, and whole-chromosome copy number changes.
Using this data, they were able to calculate the mutation rate in yeast. Their refinement of the mutation rate is valuable for molecular evolution models in yeast- but the authors also uncovered findings that are unique to this mutation accumulation study because of the whole-genome sequencing approach and the scale of the study. Here, two of these findings will be discussed: the prevalence of aneuploidies and the genomic-context of single nucleotide mutations.
Aneuploidy (almost) everywhere!
Just like in humans, aneuploidy can occur in yeast – but how often does it occur spontaneously? With their next-generation sequencing data, Zhu et al. were uniquely poised to answer this question. By analyzing the read-depth across chromosomes for each strain, they found many differences in whole-chromosome copy number. Roughly 20% of their strains (31/145) exhibited aneuploidy. And out of the 16 chromosomes in yeast, all but two had a duplication event (Figure 2).
Only a small fraction (2/31) of the aneuploidy events were chromosome deletions, while the rest were chromosome duplications. The relative lack of chromosome loss is likely because of its strongly deleterious effect. But the high prevalence of chromosome duplications raises a number of questions. For example, how common are aneuploidies in yeast strains used in other studies? Analyses on phenotypes or gene expression of yeast are often performed assuming that the strains of yeast are without aneuploidies, but an additional chromosome could affect such analyses significantly. To further investigate the role of these aneuplodies, a useful next step would be to study the distribution of fitness effects of these events.
Patterns of single mutations & methylation in yeast?
In addition to aneuplodies, Zhu et al. also identified 867 single-nucleotide mutations. The patterns of these mutations can be used as a baseline for the mutational landscape without selection.
For example, Zhu et al. examined the frequencies of specific nucleotide changes – such as the frequency of an A to T mutation or an A to G mutation. There are 6 detectable nucleotide changes (because the strand of mutation origin is unknown) that Zhu et al. found are not evenly distributed in their MA lines. C to T mutations were particularly frequent at 35%, twice as high as the 17% null expectation. These biases in mutation rate can be used to refine nucleotide evolution models such as the Jukes Cantor, which will improve future phylogenetic analyses and tests for selection in yeast.
In addition to looking at the mutations themselves, Zhu et al. also looked at the frequencies of the neighboring bases – resulting in a peculiar finding. They found that GC mutations were twice as likely to occur in CCG and TCG sites (Figure 3). This type of elevation would not surprising in mammals because of its association with methylation, but methylation is considered rare (and possibly absent) in budding yeast. The authors carefully suggest that methylation in yeast could be a “parsimonious explanation” for this difference, but further studies will be necessary to confirm this.
Continuing the saga
Three hundred thousand generations of mutation accumulation is a lot. With this long list of confirmations, refinements, and surprises uncovered by Zhu et al, it may seem that we could now stop this yeast MA experiment, as it may not yield many more results. But, as studies of similar style have shown (such as Richard Lenski‘s experiments), long-term passaging experiments continue to bear fruit even after many years. Time will tell whether these MA lines will continue to bring novel insight into the context and patterns of spontaneous mutations. One area the authors suggest could benefit from further data is the effect of genomic context on mutation rate, such as the difference of mutations occurring in noncoding versus coding regions.
Another area ripe for further experiments is the distribution of fitness effects of these mutations. It remains to be seen whether the various mutations seen in these lines are mostly neutral, deleterious, or advantageous. Such studies have both theoretical impacts, such as understanding the molecular clock, and practical implications, such as the relation of spontaneous mutations to disease (Eyre-Walker and Keightley 2007). However, calculating the distribution of fitness effects will be a difficult endeavor. Each strain has experienced several mutations, and thus determining the individual effect of mutations will likely be a challenging effort. Nevertheless, Yuan and colleagues have provided a great stepping-stone for studying the spontaneous mutation landscape at an unprecedented level of detail and scale.
Zhu Y.O., Siegal M.L., Hall D.W. & Petrov D.A. (2014). Precise estimates of mutation rate and spectrum in yeast., Proc. Nat. Academy of Sciences, DOI: 10.1073/pnas.1323011111
Halligan D.L. & Keightly P.D. (2009). Spontaneous mutation accumulation studies in evolutionary genetics, Annual Rev. Ecol. Evol. Syst., 40 151-172. DOI: 10.1146/annurev.ecolsys.39.110707.173437
Eyre-Walker A. & Keightley P.D. (2007). The distribution of fitness effects of new mutations., Nature Reviews Genetics, 8 610-618. DOI: 10.1038/nrg2146