Introduction

Orthonectida is probably one of the most enigmatic groups of so-called lower Metazoa, whose phylogenetic relationships remain uncertain (Slusarev 2018). Representatives of this group are parasites of a wide range of marine invertebrates. Their life cycle is rather simple and consists of two alternating generations: a free-living sexual stage and a parasitic stage commonly referred to as plasmodium. The phylum is not divided into classes or orders and consists of about 25 known species (Kozloff 1992; Slusarev 2018). The systematics of the orthonectids is still unresolved (Slusarev 2018). Recent molecular and morphological findings suggest that they belong to Lophotrochozoa and are close to the Annelida (Schiffer et al. 2018; Slusarev 2018). Only one mitochondrial genome from Orthonectida is available to date (Schiffer et al. 2018). There is a lack of information concerning the mt genomes for other species of Orthonectida which is needed to derive insights into their taxonomy, phylogeny, and possible molecular markers. Nothing is known about the rearrangement events within Orthonectida mt genomes and how and when these occurred. More detailed knowledge about these processes could help to improve the understanding of the mechanisms of Orthonectida species evolution.

Due to the abundance of mitochondrial genomes in tissues, their faster rate of evolution compared with the nuclear genome and the availability of a large number of informative data sets, they are widely used for phylogenetic analyses across broad taxonomic levels (Rota-Stabelli et al. 2010; Bernt et al. 2013). During evolution, mitochondrial genomes are modified by large-scale structural events, such as rearrangements, deletions or insertions of DNA blocks. Differences in gene arrangements and genetic codes may be used as valuable phylogenetic markers (Bernt et al. 2013). Also, gene loss and duplications, strand asymmetry in nucleotide composition, length, and structure of the control region, features of intergenic non-coding regions, codon usage, variation in gene length, variation in start and stop codons, gene diversity levels, mutation rates, may be of interest to comparative mitogenomics (Curole and Kocher 1999; Gissi et al. 2008; Plazzi et al. 2016).

Here we present the mitochondrial genomes of two orthonectid species Intoshia variabili and Rhopalura litoralis. These genomes were sequenced, assembled, and annotated as circular DNA molecules. We compared the I. variabili and R. litoralis mt genomes with the I. linei mt genome and reconstructed ancestral gene order for the better understanding of the evolution of these genomes. Additionally, we performed a phylogenomic investigation to assess relationships of Orthonectida with Annelida.

Materials and methods

Material collection and DNA extraction

The orthonectid I. variabili (Alexandrov and Sljusarev 1992) occurs in the turbellarian Macrorhynchus crocea Graff, 1882 (Platyhelminthes: Rhabditophora, suborder Calyptorhynchia). The orthonectid R. litoralis (Shtein 1953) is a parasite of the gastropod Onoba aculeus (Gastropoda: order Littorinimorpha, superfamily Rissooidea). The turbellarians and gastropods were collected in August 2017, in the Barents Sea at the marine biological station Dalnie Zelentsi (69°07′N, 36°05′E). The hosts were collected at the low tide and maintained in Petri dishes with filtered seawater. Free-living stages of I. variabili and R. litoralis were collected as described in detail elsewhere (Slyusarev 1994; Slyusarev and Ferraguti 2002). For genomic sequencing, the collected samples of I. variabili and R. litoralis were fixed using ethanol. Total genomic DNA was isolated using the PicoPure DNA Extraction Kit (Thermo Fisher) according to the recommendations of the manufacturer. DNA quantity assessments were performed with Qubit fluorometric quantification (Life Technologies).

Mitochondrial genome sequencing, assembly, and annotation

The genomic DNA libraries of I. variabili and R. litoralis were constructed using TruSeq library preparation protocol (Illumina) and sequenced on a HiSeq 2000 sequencing system. 62.8M 100 bp paired-end reads were obtained. Quality control check on raw sequence data was performed using FastQC (http://www.bioinformatics.babraham.ac.uk/projects/fastqc/, last accessed November 2017). De novo assemblies were implemented using SPAdes assembler (Bankevich et al. 2012). The quality of assemblies was evaluated using QUAST (Gurevich et al. 2013). The mitochondrial scaffold was identified in the assembly using BLAST (Altschul et al. 1997). Initial annotation was performed using the MITOS web server under the code for invertebrate mitochondria (Bernt et al. 2013). However, the MITOS server did not correctly identify the start and stop codons, so all protein-coding genes were further adjusted and corrected manually using mt genome of I. linei (Schiffer et al. 2018). Putative tRNA and rRNA genes were identified using the MITOS web server. To check prediction and analyze duplication events of all tRNAs, the program tRNAcan-SE Search Server v.1.21 (Lowe and Eddy 1997) was applied. The physical map was generated by our script written in Python (script available into GitHub, https://github.com/nilannik). The mitochondrial genomes have been deposited in GenBank under the accession numbers MG893580 (I. variabili) and MG917727 (R. litoralis).

Genetic code and nucleotide composition

AT and GC skew were calculated using the formulae: AT skew = [A − T]/[A + T] and GC skew = [G − C]/[G + C], for the strand encoding the majority of the protein-coding genes (Perna and Kocher 1995). Genetic code and codon usage were analyzed by GenDecoder v1.6 web tool (Abascal et al. 2006).

Ancestral genome reconstruction and gene order analyses

To show similarities in gene clusters alignment of Orthonectida mt genomes was performed with the Mauve (Darling et al. 2004). This program allows visualizing major rearrangements and inverted regions. For ancestral mt genome reconstruction, GRIMM program was used (Tesler 2002). For pairwise comparisons of the mitochondrial gene order and determining the most parsimonious genome rearrangement scenario CREx program was used (Bernt et al. 2007). The analysis was performed by applying the common intervals parameter for distance measurement.

Phylogenetic analysis

The protein-coding gene sequences of I. variabili and R. litoralis were translated using the “Invertebrate mitochondrial” genetic code per each taxon using the software Translate tool (https://web.expasy.org/translate). Intoshia linei and 28 Annelida species were selected from the NCBI’s organelles genome database and included in the phylogenetic analysis. The amino acid sequences were concatenated to the alignment using our own script written in Python. Multiple sequence alignment was implemented using muscle program integrated to SeaView (version 4; Gouy et al. 2010) with the default parameters for the protein-coding genes. Variable regions were recognized and excluded using the program Gblocks (Castresana 2000). Near 25% of the most varying sites were excluded from alignment. Concatenation of partitions for the combined data set was conducted with SeaView. Phylogenetic inference was performed by PhyloBayes (Lartillot and Philippe 2004). The PhyloBayes with two independent Monte Carlo Markov chains running for 20,000 cycles under the MtZoa model (Rota-Stabelli et al. 2009). The analysis was considered to have reached stationarity when the average standard deviation of split frequencies decreased to 0.01. Stationarity for each run was assessed by importing the parameter files into Tracer v. 1.5 (Rambaut and Drummond 2009). 36,002 trees were summed after removing first 1999 as burn-in. The resulting Bayesian tree was visualized in FigTree 1.4.3 (http://tree.bio.ed.ac.uk/software/figtree/, last accessed March 2018).

Results and discussion

Genome organization and nucleotide composition

BLAST searches identified the mt genomes of I. variabili and R. litoralis as single contigs. The complete mt genome of I. variabili is 13.989 bp (Fig. 1) and the one of R. litoralis is 14.299 (Fig. 2).

Fig. 1
figure 1

Intoshia variabili mitochondrial genome map. The tRNA genes are labeled based on the IUPACIUB single letter amino acid codes

Full size image
Fig. 2
figure 2

Rhopalura litoralis mitochondrial genome map. The tRNA genes are labeled based on the IUPACIUB single letter amino acid codes

Full size image

These two genomes are smaller than I. linei mt genome which has 15.217 bp in length (Schiffer et al. 2018). The difference in length is explained by the gene content and the size of the non-coding regions (Fig. 3; Tables 1, 2).

Fig. 3
figure 3

Comparison map of the three mitochondrial genomes of Orthonectida

Full size image
Table 1 Intoshia variabili genome organization
Full size table
Table 2 Rhopalura litoralis genome organization
Full size table

Compared with some other metazoan groups, the size variation in Orthonectida mt genomes is relatively small (approx. 14 kb ± 1 kb) (Gissi et al. 2008). The size of the two orthonectid mt genomes seems to be slightly smaller than typical metazoan mt genomes which vary in size from 15 to 20 kbp in length (Bernt et al. 2013; Boore 1999).

The genomes of I. variabili and R. litoralis have low GC content (17.38% and 21.28%, respectively) (Table 2). A is the most common base (41.76% in I. variabili and 39.72% R. litoralis), while G is the least common (8.85% and 10.67%, respectively). Both mt genomes have a positive AT and GC skew (Table 2) which differs from I. linei characterized by a negative AT and positive GC skew (Schiffer et al. 2018). These results indicated a low degree of strand asymmetry of the base composition in the I. variabili and R. litoralis mt genomes. Positive AT skew can be explained by approximately equal frequencies of A and T bases within each strand of I. variabili and R. litoralis mt DNA (a slight prevalence of A over T is characteristic of both species) (Table 3).

Table 3 Nucleotide composition characteristics of I. variabili and R. litoralis mitochondrial genomes
Full size table

Some genes partially overlapped in both species. Overlaps were detected in the genes located in major and minor strands. I. variabili mt genome has ten gene overlaps. The largest overlap, 28 bp long, is located between trnL2 and nad6. In R. litoralis mt genome, genes have 13 overlaps and the largest overlap of 77 bp is also located between nad6 and trnL2 genes. Coding gene overlapping seems to be a common feature in Orthonectida since the I. linei showed this feature in its genome too (Schiffer et al. 2018). Several non-coding regions are dispersed throughout the whole mt genome in both species. The non-coding regions in the mt genomes of I. variabili and R. litoralis are 1645 bp and 486 bp in total, respectively. R. litoralis has two major non-coding regions of 227 and 239 bp located between trnN and nad1 and between trnK and nad5, respectively. I. variabili mt genome, like that of I. linei, has the largest non-coding region with a size of 1662 bp located between the nad5 and the trnG genes.

Protein-coding genes

Typical metazoan mt genomes are considered to be relatively constant in gene content and order and to consist of a single circular DNA which codes 13 protein-coding genes (PCGs), the 2 subunits of the rRNA, and 22 tRNAs with 2 copies of serine and leucine (Bernt et al. 2013; Boore 1999). The two mt genomes contain the same gene sets as found in most metazoan mt genomes except some variations. They include two rRNA genes and 21 tRNA genes in both species, ten PCGs (atp8, nad2 and cox3 are missing) in I. variabili (Fig. 1; Table 1), and 12 PCGs (atp8 is missing) in R. litoralis (Fig. 2; Table 2). Loss of atp8 is also reported from several other, phylogenetically distant taxa, such as Platyhelminthes, Chaetognatha, and Nematoda (Gissi et al. 2008). In both genomes, some of the genes are transcribed from the major strand (17 genes for I. variabili; 22 genes for R. litoralis), others from the minor strand (16 genes for I. variabili; 13 genes for R. litoralis). By comparison, I. linei has only eight genes located in the minor strand (Schiffer et al. 2018). The cumulative length of PCGs, excluding all termination codons, is 8841 bp encoding 2947 amino acid residues for ten genes of I. variabili and 10.578 bp encoding 3526 amino acid residues for 12 genes of R. litoralis.

Codon usage

Both Othonectida species use standard invertebrate mitochondrial code (GenBank translation Table 5) and do not use TGA as the stop codon (TGA specifies tryptophan). The majority of I. variabili and R. litoralis PCGs have ATA as the start codon (Table 1, 2). Only nad1 and nad4l genes in I. variabili mt genome have ATG and ATT as start codons, respectively. As for R. litoralis, nad1 also has ATG as a start codon, nad6 and nad4 genes begin with the start codon ATT. I. variabili and R. litoralis PCGs have two stop codons TAA and TAG (Table 1, 2), and TAA is preferred over TAG. Diverse start and stop codons have been also found in other Orthonectida (Schiffer et al. 2018).

The mitochondrial PCGs of I. variabili lack six of the 64 possible codons: GCC for alanine, GCG, and CGG for arginine, CTC for leucine, AGC for serine, and TAG stop codon (Table 4). The PCGs of R. litoralis lack only two of the 64 possible codons: GCG for arginine and CCG for proline (Table 5). AT richness of both mt genomes affects the amino acid composition bias in PCGs towards the amino acids coded by the AT-rich codons (Tables 4, 5). There is also a codon usage bias in the two genomes (Table 4, 5).

In general, NNA and NNT codons are the most common codon types, while NNG and NNC codons are the least used. Thus, the eight most frequently used codons were TTT, TTA, ATT, ATA, TCT, TAT, AAT, and AAA in I. variabili (Table 4) and R. litoralis mt genomes (Table 5). These AT-rich codons account for 55.2% and 53.2% in I. variabili and R. litoralis, respectively. This is consistent with the high percentage of A + T content in the nucleotide composition of PCGs as in another mt genome for Orthonectida.

Table 4 The codon usage in Intoshia variabili
Full size table
Table 5 The codon usage in Rhopalura litoralis
Full size table

Transfer and ribosomal RNA genes

The majority of metazoans mt genomes typically encode 22 tRNAs with a single tRNA for each of 18 amino acids and 2 tRNA genes for serine and leucine. A total of 21 tRNA genes were found interspersed in the mt genomes of I. variabili and R. (Tables 1, 2). Besides two tRNASer and tRNALeu genes, the mt genome of I. variabili encodes two aspartic acid tRNA genes (tRNAAsp) (Table 1; Fig. 1). The tRNAAsp1 gene is located after tRNALeu1, and the tRNAAsp2 gene before cox1 gene. A small number of tRNAs of both mt genomes possess the common cloverleaf structure. Eight tRNAs of I. variabili and 11 tRNAs of R. litoralis have the TΨC arm simplified to a loop. Another three tRNAs of I. variabili and R. litoralis have the dihydrouridine arm simplified to a loop.

The length of 16S rRNA gene of both I. variabili and R. litoralis mt genomes is in the range of other metazoan mt LSU rDNA (from 1 to 1.5 kb), though 12S rRNA is slightly smaller (660 and 690 bp, while it is between 700 and 1.5 kb for other metazoans) (Wey-Fabrizius et al. 2013).

Ancestral genome reconstruction and gene order analysis

So far, the order of genes and their rearrangements within Orthonectida remained unknown because only one mt genome belonging to I. variabili is available (Schiffer et al. 2018). To show similarities in gene clusters and initial view to rearrangements the three Orthonectida mt genomes were compared using Mauve. The mt genome of I. linei was chosen as a reference. Significant homology was observed for the six regions represented by colored blocks connected between genomes by lines (Fig. 4).

Fig. 4
figure 4

A Mauve alignment of Intoshia linei, Intoshia variabili, and Rhopalura litoralis mitochondrial genomes. The colored blocks represent regions of homology between mt genomes as determined by Mauve alignment on default settings. Inside each block, Mauve draws a similarity profile of the genome sequence. Lines indicate which regions in each genome are homologous

Full size image

As shown in Fig. 4, several blocks are shifted downward relative to the reference genome, which indicates that such blocks are in the reverse complement orientation. Regions containing cox2 and nad3 gene, are not defined by Mauve program probably because the mt genome of I. variabili lacks such genes. Also, there are regions located outside the blocks, which are too divergent in genomes and contain lineage-specific sequences (Fig. 4). As shown in Fig. 4, the mt genome of I. linei can be transformed into that of I. variabili by a reverse transposition of regions number 2, 3 and 6 (genes located in regions are represented in Table 6).

Table 6 Genes located in the regions, determined by Mauve program
Full size table

As for R. litoralis, mt genome of I. linei can be transformed into that by a reverse transposition of regions number 1, 2 and 6. As seen from the analysis, the rearrangement landscape of Orthonectida mt genomes is rather complicated and it is unclear which of the gene order is more ancestral. For the better understanding of Orthonectida mt genome evolution, ancestral gene order was reconstructed (Fig. 5).

Fig. 5
figure 5

Comparison of ancestral gene order with Orthonectida species. Protein-coding and ribosomal genes are denoted by their names and one capital letter indicates the amino acid for the tRNAs

Full size image

Ancestral gene order is more similar to I. linei mt genome (Fig. 5). To estimate the number of rearrangements between the ancestral and Orthonectida mt genomes the CREx analysis was implemented (Table 4). According to the analysis, I. linei mt genome had only two reversal events compared with the ancestral genome. (Table 7; Fig. 6a). In the course of evolution I. variabili lost nad2, atp8, cox3, trnR, and trnF genes, duplicated trnD gene, and had two reversals of two gene blocks (Fig. 6b). As for R. litoralis, this mt genome lost atp8 and trnQ genes and had three reverse transposition and one reversal events (Fig. 6c).

Table 7 Pairwise distance matrix of the Orthonectida mt genomes
Full size table
Fig. 6
figure 6

A minimum number of events that rearranges the mitochondrial gene order of ancestral genome (top) via the intermediate gene order (middle) into the gene order of Orthonectida (bottom)

Full size image

The arrangement observed for I. linei can be obtained with a minimum number of changes only when starting from that of the ancestral genome (Table 7). I. linei mt genome has minimum rearrangement events and poses the closest gene order to the ancestral genome (Table 7). Mt genome of I. variabili has the largest number of rearrangement events when starting from that of the ancestral genome. Rearrangements are represented by two reversals of significant gene blocks, deletions of five genes (nad2, atp8, cox3, trnR, and trnF), and one duplication event (Fig. 6). R. litoralis, compared with I. variabili, has lower number of rearrangements: three reverse transpositions of different gene blocks, one reversal event and two gene deletions (atp8 and trnQ) (Fig. 6).

Starting from any Orthonectida mt genome arrangement, the number of changes is approximately the same (Table 7). This is explained by the fact that it is not possible to calculate correctly all rearrangements, because programs do not consider duplications and loss of genes. In these cases, all rearrangements for all genomes are represented only by reversals or reverse translocations. All known species of Orthonectida have a unique gene order which was not defined previously (Fig. 7). If we compare Orthonectida to the putative ground pattern of Lophotrochozoa we observe three common gene blocks (nad4 and nad5, rrnL and rrnS, nad1, nad6 and cob) along with complicated rearrangement landscape within Orthonectida (Fig. 7a, c). If we compare Orthonectida with Pleistoannelida we can see only two common gene blocks (rrnL and rrnS, nad6 and cob) (Fig. 7a, c). The exception is I. variabili, which has three common gene blocks with Lophotrochozoa and Pleistoannelida (Fig. 7b). Interestingly, ancestral genome has four common gene blocks with Lophotrochozoa and three with Pleistoannelida (Fig. 7d).

Fig. 7
figure 7

Comparison of Orthonectida and ancestral mitochondrial genes with the putative ground pattern of Pleistoannelida and Lophotrochozoa. Only PCGs are shown. The direction of the transcription is not considered

Full size image

Phylogenetic analysis

According to previous studies, position of the Orthonectida on the phylogenetic tree has been changed (Mikhailov et al. 2016). Based on genomic data of I. linei, Mikhailov et al. identified orthonectids as highly simplified spiralians located close to Annelida in the phylogenetic tree. The most recent studies based on single mt genome of I. linei defined orthonectids as a highly degenerate annelid worms but could not precisely place them within Annelida (Schiffer et al. 2018). Based on only one Orthonectida species the authors found a short block of genes (nad1, nad6, cob) which are found in the same order as in the lophotrochozoan ancestor but did not find a similar block in the Pleistoannelida (Schiffer et al. 2018).

Our gene order analysis and ancestral gene order reconstruction showed that orthonectids and their ancestral genome have up to three common gene blocks with Pleistoannelida (Fig. 7) while in basal branch of Annelida we observe a variety of gene orders and, therefore, it is more difficult to find common gene blocks between all basal branch species and orthonectids (Weigert et al. 2016). According to this, we proposed that Orthonectida are closer to Pleistoannelida, than to the Basal Branch Annelida. To check our hypothesis, we inferred phylogenetic relationships between Orthonectida and Annelida species where most of them belong to Pleistoannelida and several of them to basal branch as an outgroup.

Our results suggest that all orthonectid species are grouped together within Annelida. Moreover, they are forming a long branch, which is close to Clitellata (Fig. 8). As seen from the analysis, Orthonectida might have separated from Clitellata a long time ago, and their mt genomes evolved faster than Annelida mt genomes, hence orthonectids are located on a longer branch. Noteworthy, a clitellum made of secretory cells in the midbody has been registered in some orthonectids (Metschnikoff 1881; Shtein 1953).

Fig. 8
figure 8

Phylogenetic relationships of Orthonectida based on mitochondrial genome data (31 taxa, 2759 amino acid positions). Bootstrap support values (BS) are shown for each node. Node numbers show poster probability values. Scale bar represents substitutions/site

Full size image

To summarize, mt genomes of orthonectids have undergone reduction and gene loss, they have complicated organization, which is manifested in strand asymmetry in nucleotide composition, as well as in some features of intergenic non-coding regions, tRNA duplication and folding. Moreover, all species of Orthonectida have a unique gene order with complicated rearrangement landscape. Significant differences in mitochondrial genomes in the three orthonectid species could be explained by the fact that their host species belong to different taxa (flat worms, nemertines and gastropods). Among analyzed mt genomes of Orthonectida, I. linei possesses the closest gene order to the ancestral genome. All Orthonectida species are monophyletic, and in the phylogenetic tree are close to Pleistoannelida, and specifically, to Clitellata.