Clarifying the molecular basis of adaptive phenotypic evolution through comparative genomics in marsupial and eutherian mammals

© 2017 Dr Charles FeiginAdaptive phenotypic evolution is influenced by many factors, occurring at multiple levels of biological organization. The interplay of natural selection and genetic variation in populations is the ultimate cause of adaptation. As new variants with phenotypic consequences arise, they are filtered by selection to those which increase fitness. However, the genetic context in which mutations occur and the effects that they have on development profoundly influence their contributions to adaptation. Therefore, a modern, predictive evolutionary theory must also account for these proximate aspects of evolution. Comparative genomics has emerged as a powerful approach for identifying and characterising adaptively relevant loci. In this thesis, I performed several comparative studies between marsupials and eutherians, two sister taxa that comprise the vast majority of extant mammalian species. I focused chiefly on the extinct marsupial Tasmanian tiger or thylacine (Thylacinus cynocephalus) and the eutherian canids (family Canidae). Together, these species are perhaps the most widely recognised example of convergent evolution. I sequenced the genome of the thylacine from an ethanol- preserved pouch young specimen. I then analysed protein-coding genes and cis- regulatory elements to uncover loci that contribute to their exquisite convergence. I first identified homoplasious amino acid substitutions between the thylacine and canid protein-coding orthologs. I found that the number of homoplasies between these species was similar to that in non-convergent species and did not exceed that expected under a model of neutral evolution. Surprisingly, I found that despite their extensive phenotypic convergence, positive selection has targeted different genes and pathways in these species. Furthermore, no homoplasious genes were under positive selection in both. These results suggest that adaptive amino acid homoplasy is not a primary driver of phenotypic convergence in the thylacine and canids. I then used computational methods to identify and characterize non-coding, cis- regulatory elements in the thylacine and wolf genomes. I first predicted craniofacial enhancers and tested these elements for accelerated evolutionary rates in the thylacine iand wolf. In stark contrast to my findings in protein-coding genes, I identified hundreds of convergently accelerated, non-coding elements in both of these species. Many elements lay close to key craniofacial, osteogenic and neurological genes and showed convergent enrichment and depletion of binding motifs corresponding to key developmental transcription factors. There was also considerable overlap between thylacine and wolf accelerated regions and known mouse and human enhancers. This provided an external validation of my computational predictions and yielded insights into the expression patterns of several predicted enhancers. These results show rich potential for evolutionarily relevant changes in cis-regulatory elements that may contribute to convergent adaptations in the thylacine and wolf. Taking a candidate gene approach, I then examined an alternative route for protein-coding evolution through coding repeat mutations. In certain evolutionary contexts, coding repeats may provide an important source of rapid, quantitative phenotypic variation. RUNX2 QA repeat variation has been associated with craniofacial evolution in several groups of eutherian mammals but is largely absent in marsupials, suggesting evolutionary constraint. Analyses of RUNX2 QA repeat variation among mammal lineages revealed that marsupials and precocial eutherians, which have intensive locomotor requirements at birth, show less RUNX2 QA repeat variation than altricial eutherians. Strict requirements on the timing of perinatal ossification in these lineages may limit phenotypic evolution through RUNX2 QA repeats. My research has important implications for our understanding of how selection targets the genome to drive phenotypic adaptation. My findings are consistent with long- standing hypotheses that protein-coding genes are evolutionarily constrained by their pleiotropic functions. In distantly related species, such changes may not be major contributors to adaptive phenotypic evolution. In contrast my findings show rich potential for cis-regulatory elements to encode evolutionarily relevant variation. Furthermore, my research has yielded novel insights into the biology and evolution of an iconic, extinct marsupial predator