Investigating the loci that contribute to convergent craniofacial evolution between the thylacine and canids

Abstract

One of the most fundamental questions in evolutionary developmental biology is how phenotypic adaptations are controlled at the molecular level. One way we can address this question is by looking at examples of convergent evolution between distantly related species. Here we can ask the question; are similarities in morphology reflected by similarities in the genome? One of the most striking cases of convergent evolution in mammals is seen between the marsupial thylacine (or Tasmanian tiger) and placental canids (wolves, dingos and foxes) particularly in their cranial morphology. However, the extent of their morphological convergence has never been directly quantified. In my thesis I use a combination of morphological and molecular data to investigate candidate loci that may contribute to convergent craniofacial evolution between the thylacine and the canids.

Using a geometric morphometric comparison of cranial shape between extinct and extant marsupial and placental mammals, I showed that the adult thylacine and canids represent a remarkable case of craniofacial convergence. By additionally CT scanning and landmarking all known thylacine pouch young specimens, I was able to demonstrate that the marsupial thylacine overcame its conserved neonatal constraints towards the end of its developmental period in the pouch. The strong similarities between the thylacine and canids are likely driven by underlying changes in cranial neural crest cells (NCCs), which are directly responsible for patterning the facial skeleton.

I next investigated candidate loci that might be underpinning this extraordinary phenotypic convergence. RUNX2 is expressed in NCCs and is strongly implicated in driving facial length evolution in placental mammals. I hypothesized that similarities in the RUNX2 gene might partially explain similarities in facial shape between the thylacine and canids. However, unexpectedly, we found that the marsupials possess an invariant RUNX2 which cannot explain the diversity of facial shapes seen within marsupials nor craniofacial convergence. Instead, changes in facial length might be mediated through regulatory changes to RUNX2 expression.

Using a genome-wide approach, we investigated homoplasy in protein coding genes. While overall homoplasy was extremely rare, we identified multiple thylacine/canid homoplasious amino acid substitutions in the osteogenic chromatin remodeller, CHD9, a known upstream regulator of RUNX2. We found that the amino acid substitution in the DNA binding domain resulted in differential expression and activation of RUNX2 in vitro and may act as a contributor to RUNX2-mediated craniofacial convergence.

While I found evidence for changes in protein coding genes potentially contributing to convergence, the pleiotropic consequences of mutations in key developmental genes are thought to limit their evolvability. As such, we also used a genome wide approach to investigate accelerated evolution and convergence in the non-coding portion of the genome. We identified multiple putative cis-regulatory elements (CREs), including an enhancer upstream of the craniofacial TGF-β signalling receptor ACVR2A, also critical in NCCs. We found that the thylacine enhancer was able to drive craniofacial expression in the mouse and is a potential candidate mediating convergent craniofacial evolution. This finding suggests CREs may also play important roles in adaptive evolution and convergence.

In this thesis I find support for protein coding and CRE evolution driving convergent craniofacial similarities. This supports my hypothesis that convergence targets genes and CREs in the NCCs directing craniofacial convergence between the thylacine and canids.