Divergence in Regulatory Regions and Gene Duplications May Underlie Chronobiological Adaptation in Desert Tortoises

Abstract

Many cellular processes and organismal behaviours are time-dependent, and asynchrony of these phenomena can facilitate speciation through reinforcement mechanisms. The Mojave and Sonoran desert tortoises (Gopherus agassizii and G. morafkai respectively) reside in adjoining deserts with distinct seasonal rainfall patterns and they exhibit asynchronous winter brumation and reproductive behaviours. We used whole genome sequencing of 21 individuals from the two tortoise species and an outgroup to understand genes potentially underlying these characteristics. Genes within the most diverged 1% of the genome (F_ST ≥ 0.63) with putatively functional variation showed extensive divergence in regulatory elements, particularly promoter regions. Such genes related to UV nucleotide excision repair, mitonuclear and homeostasis functions. Genes mediating chronobiological (cell cycle, circadian and circannual) processes were also among the most highly diverged regions (e.g., XPA and ZFHX3). Putative promoter variants had significant enrichment of genes related to regulatory machinery (ARC-Mediator complex), suggesting that transcriptional cascades driven by regulatory divergence may underlie the behavioural differences between these species, leading to asynchrony-based prezygotic isolation. Further investigation revealed extensive expansion of respiratory and intestinal mucins (MUC5B and MUC5AC) within Gopherus, particularly G. morafkai. This expansion could be a xeric-adaptation to water retention and/or contribute to differential Mycoplasma agassizii infection rates between the two species, as mucins help clear inhaled dust and bacterial. Overall, results highlight the diverse array of genetic changes underlying divergence, adaptation and reinforcement during speciation.

Keywords: circadian; circannual; mucin; promoter; reinforcement; speciation.

Figure 1.. Published studies that used whole genome sequencing to study speciation.

Estimated age of divergence and number of generations since divergence from other studied speciation events. Tortoises (this study) are thought to have been diverged for a greater number of generations than other well-studied speciation events. For a list of data sources see Methods.

Figure 2. Principal Components Analysis of whole-genome variation

for A) Mojave Desert tortoise (G. agassizii), Sonoran Desert tortoise (G. morafkai), and outgroup Texas tortoise (G. berlandieri) and B) solely within G. agassizii and G. morafkai.

Figure 3. Historical demography for the three Gopherus lineages

based on whole genome sequencing and single-genome analysis in PSMC (note differences in scale of y axis (between all panels and G. morafkai is shown twice (right) for clarity) and log x-axes). Analysis assumes a 25-year generation time and mutation rate of 2.7 x10⁻⁸. Pleistocene glacial cycles (peak glacial periods) are depicted in greyscale. Effective population size is much higher for G. morafkai (middle) and display unequal responses to glaciations. Gopherus agassizii appears to have undergone a population expansion during the LGM, while G. morafkai has expanded during interglacial periods. B) Historical demography of G. agassizii and G. morafkai based on 10 low coverage genomes per species in SMC++.

Figure 4. Distribution of how variants in the 1% most-diverged genomic regions map to genomic elements.

(A) Most variants map to flanking intergenic regions, defined as more than 1 kb upstream of the gene’s annotated start site; (B) Of variants that mapped to genic regions in (a), these are the total variants mapping to genic elements according to the annotation for G. agassizii. Variants mapped to promoter, UTRs, or exons are considered ‘functional’ in downstream analyses. Promoter is defined as 1kb immediately upstream of the gene.

Figure 5.. String interaction network of genes within highly diverged F_ST windows that interact based on available evidence.

Clusters with identifiable themes are colored-coded (see Discussion). Each node is a gene, edges are weighted to reflect the degree of confidence of node-node interactions. All genes shown were in the highly diverged regions with putatively functional variants; nodes are scaled by the number of variants in putatively functional regions (promoter, exons, or UTR). Non-interacting genes are not shown.

Figure 6.. Characterization and summary of diverged genes.

A) Manhattan plots showing F_ST of highly diverged genes discussed in the text (red): Mediation Complex Subunit 12 (MED12), DNA Damage Recognition and Repair Factor (XPA), and Mucin 5B (MUC5B). Outliers in black contained “non-functional” variants and are not discussed. B) Characterization of promoter INDEL variants of MED12 as well as the variant proportion in G. agassizii (left) and G. morafkai (right) relative to the G. agassizii reference allele. C) Phylogenetic reconstruction of the MUC2-5-6 gene array based on predicted proteins from mucin genes identified in this study, and D) synteny of the mucin gene array based showing the expansion of MUC5 genes in G. morafkai. Gene colors are based on original annotations and colored boxes are homology estimations based on the phylogeny in C. Genes with white centers are pseudogenized.

Figure 7.. Evolutionary divergence in the regulatory regions of chronobiological processes.

A) XPA plays a key role in nucleotide excision repair of UV DNA damage and exhibits a circadian rhythm regulated by the clock gene BMAL1. XPA and its regulator XAB2 showed extensive regulatory divergence which may relate to differences in circadian rhythm and DNA repair. Offsetting of internal clocks between the two species could results in hybrid incompatibilities. Circadian differences of tortoises shown here are only suggestions. B) MED12 and MED26 (part of the transcriptional Mediator Complex, grey) and ZFHX3 all have promoter changes; ZFHX3 is a transcription factor expressed in the Suprachiasmatic Nucleus of the Hypothalamus that is directly involved in the regulation of sleep-wake cycles and circadian rhythms. Regulatory changes can impact the timing, specificity, or magnitude of transcription that underlie known seasonal differences in the timing of mating, laying, and brumation behaviors of the tortoise species. Cell cycle, circadian, and circannual rhythms are proposed to be coupled (see Discussion).