Errors, stop codons, and the evolution of genomes
: (Alternative Format Thesis)

  • Alex Ho

Student thesis: Doctoral ThesisPhD


The canonical view of protein evolution is one where selection acts upon the final product of gene expression. As gene expression is extremely prone to error, and these errors are deleterious, it is however becoming increasingly apparent that selection acts far earlier to oppose erroneous protein synthesis. Genomes may evolve to become more error-proof by preventing their occurrence or by mitigating their impacts. In this thesis I seek to resolve several questions regarding error control using translational read-through (TR) as an exemplar. TR occurs when the stop codon of an mRNA transcript is missed by the termination machinery during translation, leading to the continuation of translation into the 3’ UTR and potential generation of C-terminally extended proteins. TR prevention and mitigation can be achieved by selection for stop codons. TR rate can be reduced by using the least error-prone stop codon, TAA, to terminate translation while TR may theoretically be mitigated by 3’ in-frame additional stop codons (ASCs) which could act as a fail-safe mechanism. In Chapter 2, I ask how often we see evidence for error mitigation strategies in response to TR. I present evidence for ASC enrichment in some, not all, unicellular eukaryotes but no such evidence in multicellular species or bacteria. I note that the strength of selection for TR mitigation should depend on the TR rate, thus in Chapter 3 I investigate how ASCs and TAA stop codons co-evolve, asking whether there is a preferred evolutionary route for error handling. I observe that TAA enrichment significantly correlates with effective population size (Ne), while ASC enrichment does not. As nearly neutral theory predicts that selection is most efficient in species with high Ne, from these results I infer that error prevention might be optimal. If TAA is positively selected to prevent TR, how then might we explain variation in the usage of the nonoptimal stops, TGA and TAG? In Chapter 4, I re-examine the long-standing hypothesis that bacterial stop codon usage adapts to the cellular abundance of RF1 and RF2 release factors, finding evidence to the contrary. I note also that TGA is enigmatically highly abundant and highly conserved in mammals despite its high intrinsic TR rate. This, however, I find in Chapter 5 to be better explained by the action of GC-biased gene conversion than selection for TGA stop codons or mutation bias. All the above I frame within the wider literature in a review article presented as “discussion part 1” in Chapter 6. During the pandemic I also contributed to two papers concerning the evolution of SARS-CoV-2, which are presented as appendices.
Date of Award22 Feb 2023
Original languageEnglish
Awarding Institution
  • University of Bath
SupervisorLaurence Hurst (Supervisor), Araxi Urrutia (Supervisor) & Laurence Hurst (Supervisor)

Cite this