Phylogenomics of nonavian reptiles and the structure of the ancestral amniote genome

Abstract

We report results of a megabase-scale phylogenomic analysis of the Reptilia, the sister group of mammals. Large-scale end-sequence scanning of genomic clones of a turtle, alligator, and lizard reveals diverse, mammal-like landscapes of retroelements and simple sequence repeats (SSRs) not found in the chicken. Several global genomic traits, including distinctive phylogenetic lineages of CR1-like long interspersed elements (LINEs) and a paucity of A-T rich SSRs, characterize turtles and archosaur genomes, whereas higher frequencies of tandem repeats and a lower global GC content reveal mammal-like features in Anolis. Nonavian reptile genomes also possess a high frequency of diverse and novel 50-bp unit tandem duplications not found in chicken or mammals. The frequency distributions of approximately 65,000 8-mer oligonucleotides suggest that rates of DNA-word frequency change are an order of magnitude slower in reptiles than in mammals. These results suggest a diverse array of interspersed and SSRs in the common ancestor of amniotes and a genomic conservatism and gradual loss of retroelements in reptiles that culminated in the minimalist chicken genome. The sequences reported in this paper have been deposited in the GenBank database (accession nos. CZ 250707-CZ 257443 and DX 390731-DX 389174).

Fig. 1.

Summary of interspersed and tandem repeats in nonavian reptiles. (a) Estimated copy number per genome of repetitive elements for four reptilian species. Estimates with error bars are based on RepeatMasker (37) queries against the chicken and primate database, summarized in SI Table 1. Error bars are 95% confidence intervals for the entire genome. Copy numbers for chicken are taken from published whole-genome assembly results (5) and targeted hybridization studies of avian microsatellites (23). DNA TE, DNA transposable element. (b) Histogram of frequencies of total tandem repeat array lengths, measured in base pairs, for the same sequence data examined in a. Details of repeat detection and analysis are presented in Materials and Methods and SI Text.

Fig. 2.

Phylogenetic analysis of CR1 elements. (a) Diagram of the ≈4.5-kb full-length CR1-like LINE element structure. The 3′ terminal region analyzed is boxed, including the untranslated region (UTR) and conserved ORF (ORF-2) reverse-transcriptase domains. (b) Neighbor-joining tree of genetic distances among 308 3′ CR1 termini (alignment length = 168–976 bp) for four reptilian species with bootstrap support and host-species indicated by color. Outgroup is arbitrary and is not meant to indicate ancestral lineages. Bayesian analysis yielded similar results (see text). (c) Relative frequency of species representation in nonoverlapping CR1 clades with >80% bootstrap support annotated in b. T, turtle; A, alligator; C, chicken; An, Anolis. Details of sequence alignment and phylogenetic analysis are listed in Materials and Methods.

Fig. 3.

History of amniote genomes and genomic signatures. Neighbor-joining tree of relationships based on Euclidean distances between signatures is shown. All nodes are resolved by >70% bootstrap support except at node b (10³ replications). Genomic signatures are presented for eight vertebrates (zebrafish = outgroup) based on the frequency of all possible 8-nt DNA words contained in sequences analyzed. A key illustrates dark-colored (more frequent words) and light-colored (less frequent words) pixels used to construct signatures. Approximate amount of DNA sequence in megabases (and genomic source) used to construct genomic signatures are as follows: alligator, 2.4; turtle, 2.4; chicken, 6.1 (multiple chromosomes); Anolis (1.3); mouse, 23.7 (chromosome 17); human, 32.7 (chromosome 22); Xenopus, 2.6 (multiple chromosomes); Zebrafish, 14.9 (multiple chromosomes). Trends in amniote genome evolution are annotated with specific nodes and tips labeled a–k. Estimated amounts of evolutionary change indicated for CR1 LINE copies and average GC content are based on optimization of these traits across the tree using a phylogenetic generalized least squares analysis implemented in COMPARE v. 4.6 (ref. and SI Table 4). Details of genome signature construction and phylogenetic analysis are presented in the text and Materials and Methods.

Fig. 4.

Rapid evolution of genomic word-frequency change in mammals. Estimates of amounts of lineage-specific change are based on a phylogenetic generalized least-squares analysis implemented in COMPARE v. 4.6 (35). Rates and standard errors for a subset of the most rapidly evolving words analyzed are listed in SI Table 3. Word rank order plotted for each lineage is determined by rank order amount of change within each lineage and similar but not identical between lineages. Divergence times used for rate estimations are listed in SI Text.