Hellbender genome sequences shed light on genomic expansion at the base of crown salamanders

Abstract

Among animals, genome sizes range from 20 Mb to 130 Gb, with 380-fold variation across vertebrates. Most of the largest vertebrate genomes are found in salamanders, an amphibian clade of 660 species. Thus, salamanders are an important system for studying causes and consequences of genomic gigantism. Previously, we showed that plethodontid salamander genomes accumulate higher levels of long terminal repeat (LTR) retrotransposons than do other vertebrates, although the evolutionary origins of such sequences remained unexplored. We also showed that some salamanders in the family Plethodontidae have relatively slow rates of DNA loss through small insertions and deletions. Here, we present new data from Cryptobranchus alleganiensis, the hellbender. Cryptobranchus and Plethodontidae span the basal phylogenetic split within salamanders; thus, analyses incorporating these taxa can shed light on the genome of the ancestral crown salamander lineage, which underwent expansion. We show that high levels of LTR retrotransposons likely characterize all crown salamanders, suggesting that disproportionate expansion of this transposable element (TE) class contributed to genomic expansion. Phylogenetic and age distribution analyses of salamander LTR retrotransposons indicate that salamanders' high TE levels reflect persistence and diversification of ancestral TEs rather than horizontal transfer events. Finally, we show that relatively slow DNA loss rates through small indels likely characterize all crown salamanders, suggesting that a decreased DNA loss rate contributed to genomic expansion at the clade's base. Our identification of shared genomic features across phylogenetically distant salamanders is a first step toward identifying the evolutionary processes underlying accumulation and persistence of high levels of repetitive sequence in salamander genomes.

Fig. 1.—

(A) Pie-chart summarizing the proportions of the Cryptobranchus genome identified by our repeat-mining pipeline as TEs, simple/tandem repeats, unknown repeats, and nonrepetitive sequences. The majority of the genome is repetitive. (B) Percentage of the shotgun data (bp) identified as TEs from different superfamilies.

Fig. 2.—

(A) Levels of LTR retrotransposons in the Cryptobranchus genome, as well as the genomes of six plethodontid salamanders and five vertebrates with typical genome sizes, shown as the percentage of the host genome (bp). (B) Levels of LTR retrotransposons in the Cryptobranchus genome, as well as the genomes of six plethodontid salamanders and five other vertebrates, shown as Gb. Cryptobranchus, as well as the plethodontid salamanders, has high levels of LTR retrotransposons relative to other vertebrates. Analyses of human genome 454 sequence data sets comparable to our salamander sequencing coverage (i.e., ∼1%) suggest that 1% shotgun data produce underestimates of LTR levels; thus, the differences we report here are likely conservative.

Fig. 3.—

(A) Levels of non-LTR retrotransposons in the Cryptobranchus genome, as well as the genomes of six plethodontid salamanders. Cryptobranchus has higher levels of non-LTR retrotransposons than do the other salamander species. (B) Levels of six abundant non-LTR retrotransposon superfamilies in the Cryptobranchus genome, as well as the genomes of six plethodontid salamanders. L1 and Penelope are much more abundant in the Cryptobranchus genome.

Fig. 4.—

Maximum likelihood tree estimated from aligned amino acid sequences of the RNase H domain from Gypsy/Ty3 retrotransposons in salamanders and other taxa. Tip names include genus name followed by retrotransposon family/subfamily name. Bootstraps above 50% are shown. Salamander sequences, and the ancestral lineages of the clades that contain salamander sequences, are indicated by color. The tree is unrooted. The majority of the salamander sequences are most closely related to other vertebrate sequences.

Fig. 5.—

Sequence divergence distributions for Gypsy/Ty3 elements in the Cryptobranchus alleganiensis genome, as well as four species of plethodontid salamanders. Distributions for Aneides flavipunctatus, Eurycea tynerensis, and Batrachoseps nigriventris suggest ongoing proliferation. Distributions for Desmognathus ochrophaeus and C. alleganiensis suggest a peak of proliferation in the past and decreased activity toward the present. The presence of a single peak in all species is inconsistent with multiple horizontal transfer events into salamander genomes.

Fig. 6.—

DNA loss rate (bp deleted − bp inserted/substitutions) from Cryptobranchus, four species of plethodontid salamanders, and five nonsalamander vertebrates with typical genome sizes. Cryptobranchus, as well as the plethodontid salamanders, has lower rates of DNA loss than the nonsalamander vertebrates. Analyses of human genome 454 sequence data sets comparable to our salamander sequencing coverage (i.e., ∼1%) suggest that 1% shotgun data produce less accurate estimates of DNA loss rate than whole-genome analyses; thus, the extent of the difference between salamanders and other vertebrates that we report should be interpreted with caution.