Analysis of the genome of the New Zealand giant collembolan (Holacanthella duospinosa) sheds light on hexapod evolution

Abstract

Background: The New Zealand collembolan genus Holacanthella contains the largest species of springtails (Collembola) in the world. Using Illumina technology we have sequenced and assembled a draft genome and transcriptome from Holacanthella duospinosa (Salmon). We have used this annotated assembly to investigate the genetic basis of a range of traits critical to the evolution of the Hexapoda, the phylogenetic position of H. duospinosa and potential horizontal gene transfer events.

Results: Our genome assembly was ~375 Mbp in size with a scaffold N50 of ~230 Kbp and sequencing coverage of ~180×. DNA elements, LTRs and simple repeats and LINEs formed the largest components and SINEs were very rare. Phylogenomics (370,877 amino acids) placed H. duospinosa within the Neanuridae. We recovered orthologs of the conserved sex determination genes thought to play a role in sex determination. Analysis of CpG content suggested the absence of DNA methylation, and consistent with this we were unable to detect orthologs of the DNA methyltransferase enzymes. The small subunit rRNA gene contained a possible retrotransposon. The Hox gene complex was broken over two scaffolds. For chemosensory ability, at least 15 and 18 ionotropic glutamate and gustatory receptors were identified, respectively. However, we were unable to identify any odorant receptors or their obligate co-receptor Orco. Twenty-three chitinase-like genes were identified from the assembly. Members of this multigene family may play roles in the digestion of fungal cell walls, a common food source for these saproxylic organisms. We also detected 59 and 96 genes that blasted to bacteria and fungi, respectively, but were located on scaffolds that otherwise contained arthropod genes.

Conclusions: The genome of H. duospinosa contains some unusual features including a Hox complex broken over two scaffolds, in a different manner to other arthropod species, a lack of odorant receptor genes and an apparent lack of environmentally responsive DNA methylation, unlike many other arthropods. Our detection of candidate horizontal gene transfer candidates confirms that this phenomenon is occurring across Collembola. These findings allow us to narrow down the regions of the arthropod phylogeny where key innovations have occurred that have facilitated the evolutionary success of Hexapoda.

Keywords: Chemoreceptors; Developmental biology; Epigenetics; Genome assembly; Hexapoda; Horizontal gene transfer; Methylation; Neanuridae; Phylogenomics; RNA; Sex determination.

Fig. 1

Distribution of gene parameters for the genome assembly of Holacanthella duspinosa

Fig. 2

Kmer spectrum for the genome assembly of Holacanthella duspinosa

Fig. 3

Signatures of normalised CpG content (CpG[o/e]) reveal the presence and absence of historical DNA methylation in hexapods. Graphs are frequency histograms of CpG[o/e] with the y-axis depicting the number of genes with the specific CpG[o/e] values given on the x-axis. a Analysis of gene bodies in the honeybee (Apis mellifera), which has an intact DNA methylation system, reveals a bimodal distribution. b In contrast, the same analysis in Drosophila melanogaster, which does not have an intact DNA methylation system, reveals a unimodal distribution. c Analysis of Holacanthella duospinosa transcripts reveals a similar unimodal distribution consistent with the absence of an intact DNA methylation system in this species. The mean of this distribution is similar to the mean obtained for 1 kb fragments of the genome (d) and is consistent with a slightly lower than expected CpG content in the DNA sequence of H. duospinosa

Fig. 4

Arrangement of the Hox gene cluster in Holacanthella duospinosa relative to the collembolan Folsomia candida and a hypothetical ancestral insect

Fig. 5

Phylogenetic trees of doublesex (a) protein sequences and Sex-lethal (b) protein sequences. F and M denote female and male splice variants, respectively. Numbers above branches are bootstrap proportions where only values greater than 50% are given. The scale bars show the expected number of amino acid substitutions per site

Fig. 6

Maximum likelihood phylogenetic tree of the glycosyl hydrolase family 18 (GH18) domains from chitinase-like proteins of insects. Includes sequences from Aedes aegypti (Aaeg), Anopheles gambaie (Agam), Apis mellifera (Amel), Acyrthosiphon pisum (Apis), Bombyx mori (Bmor), Cerapachys biroi (Cbir), Drosophila melanogaster (Dmel), Daphnia pulex (Dpul), Helicoverpa armigera (Harm), Holacanthella duospinosa (Hduo), Nilaparvata lugens (Nlug), Ostrinia furnacalis (Ofur), Pediculus humanus corporis (Phum), Spodoptera litura (Slit), and Tribolium castaneum (Tcas). Values at the nodes are bootstrap support percentages over 50%. Chitinase-like proteins identified from H. duospinosa are indicated in bold. Classification of chitinase groups follows [75]

Fig. 7

Best phylogenetic tree inferred with a Maximum Likelihood approach by IQtree (see Methods). Non-parametric bootstrap support was derived from 300 bootstrap replicates. The tree was rooted with Diplura. For both datasets, all inferred relationships revealed maximal support except for the placement of Sminthurus viridis: in black: bootstrap support for the dataset with domain-based meta-partitions (in parentheses support from the SH-LRT test), in grey: bootstrap support for the dataset with gene-based meta-partitions (in parentheses support from the aLRT test). Nodal black dots indicate maximal bootstrap and single branch test support. Photograph: Mark I. Stevens