The Evolution of Cellular Error Rates
The total number of mutations arising in functional DNA per generation scales with the -0.75 power of the genetic effective population size.
Although mutations provide the ultimate material upon which natural selection depends, most mutations are deleterious, and in certain settings can lead to a substantial fitness load. We are attempting to understand the nearly 1000-fold range of variation in the mutation rate that exists across the Tree of Life, through the study of a diversity of invertebrates and unicellular eukaryotes and prokaryotes. This work exploits a mutation-accumulation strategy in which lines are propagated as single individuals to minimize the ability of natural selection to influence the fate of newly arisen mutations, often for hundreds to thousands of generations. Complete-genome sequencing of the derived lines yields unbiased estimates of the rate and spectrum of mutations at the DNA level, revealing a dramatic negative scaling of the mutation rate with population size, an apparently universal mutation pressure towards AT composition, and many other previously unknown mutational features.
The overall results of this mutation work support the drift-barrier hypothesis, which postulates that natural selection persistently operates to decrease the mutation rate until a point is reached at which the advantages of further improvement are overcome by the power of random genetic drift. We are now attempting to expand this work to other types of intracellular errors.
We have developed a novel method that provides estimates of transcription error rates (i.e., rates of misincorporation of ribonucleotides into mRNAs), and are now evaluating the degree to which these vary among eukaryotic lineages. Error rates at this level are typically >1000x those at the level of genome replication, but also quite constant across phylogenetic lineages, implying that >1% of transcripts typically contain an erroneous base.
The next step is to determine translation-error rates, and to this end, we are applying advanced techniques from mass spectrometry to the proteomes of multiple organisms. Although much work remains to be done, preliminary results indicate that error rates at the level of translation are on the order of 0.001 to 0.01 per codon.