1. Introduction
The Gastrotricha is an animal taxon with hitherto around 700 described species that are divided into the two main clades Macrodasyida and Chaetonotida [
1,
2,
3]. Gastrotrichs are microscopic animals that occur abundantly in marine and freshwater interstitial environments [
1,
4]. They show a direct development, and are bilaterally symmetric, with an acoelomate body, and worm-like body plan [
5]. Most gastrotrich species have a commissural brain [
6,
7,
8,
9] and within the clade there is a large variety of reproductive strategies [
5,
10]. Interestingly, most gastrotrichs lack an ectodermal hindgut, and a specialized respiratory system has not been described [
5]. Gastrotrichs share complex morphological characteristics, including a radial myoepithelial pharynx with terminal mouth, epidermal monociliation, and protonephridial structures, with several other lineages within Bilateria. This has made their phylogenetic position a matter of debate. [
11]. Initially, gastrotrichs were considered to be close to rotifers (“Trochelminthes”) [
12]. Later, classic systematists placed gastrotrichs either as the sister group of Introverta, together forming Cycloneuralia (i.e., Kinorhyncha + Loricifera + Priapulida + Nematoda + Nematomorpha) [
13,
14], or in the Neotrochozoa group that comprises Gnathostomulida + Gastrotricha [
15,
16]. When nucleotide sequence data became available, gastrotrichs were placed within Spiralia with uncertain relatedness, within the debated group of ‘Platyzoa’ comprised of Platyhelminthes, Gnathifera, and Gastrotricha [
17,
18,
19]. Recent phylogenomic studies suggest a sister group relationship between Gastrotricha and Platyhelminthes (Rouphozoa) [
20,
21]. However, this Gastrotricha-Platyhelminthes clade is still controversial because there is no morphological apomorphy supporting the monophyly of the clade [
22]. Even though gastrotrichs and most flatworms ingest food with a simple pharynx [
6,
23,
24], this common alimentary strategy is not necessarily an autapomorphy of Gastrotricha and Platyhelminthes. Thus, additional lines of evidence are required to further test the proposed close phylogenetic relationship of gastrotrichs and flatworms.
MicroRNAs (miRNAs) are a unique class of short non-coding RNAs with major roles as post-transcriptional gene regulators in animals and plants [
25]. In contrast to short interfering RNAs (siRNAs), responsible for RNA interference (RNAi), and PIWI-interacting RNAs (piRNAs), responsible for genome integrity [
25,
26], miRNAs are deeply conserved in metazoans. Their continuous addition to genomes during evolution has often been correlated with increasing morphological complexity and number of cell-types in a given organism [
27,
28,
29,
30]. Accordingly, absence/presence of miRNAs has also been successfully applied as phylogenetic and taxonomic arguments [
31,
32,
33,
34], and were also used to ascertain clade specificity of highly derived organisms such as Myzostomida [
35]. However, because miRNA annotation is a challenging task, many published miRNA complements have considerable issues with respect to quality and completeness [
36,
37]. For example, missing miRNA families have frequently been misinterpreted as secondary losses, and the conservation and the utility of miRNAs as phylogenetic markers has been questioned [
38]. However, losses of miRNA families have rarely been observed in well-annotated complements, and the few confirmed cases were correlated to high degrees of genomic and phenotypic reductions, for instance in parasitic species [
39,
40,
41]. Currently, information on the presence of the RNAi pathway and small RNAs in gastrotrichs is lacking, but several studies have reported on miRNAs, piRNA, and the RNAi protein machinery in flatworms [
40,
42,
43,
44,
45,
46]. Closer scrutiny of these studies revealed a dynamic evolution of RNAi proteins [
44,
45] and small RNAs, with the substantial loss of conserved and gain of novel miRNAs [
40], and the loss of piRNAs in obligate parasitic flatworms and nematodes [
43,
44,
47]. In free-living nematodes such as
Caenorhabditis elegans, however, piRNAs have been detected and described in detail [
48]. With respect to miRNAs there is substantial different between flatworms and nematodes. In addition to the miRNA families found in all protostome species, flatworms share two miRNA families MIR-1989 and MIR-1992 with spiralian taxa (we are following the nomenclature proposed by Fromm et al. [
36]). Nematodes are not part of Spiralia and, thus, do not share MIR-1989 and MIR-1992 but show instead additional four miRNA families specific for Ecdysozoa (MIR-305) and Nematoda (MIR-54, MIR-86, MIR-791), respectively. Thus at least six miRNA families differ between flatworms and nematodes. It is also noteworthy that, although free-living flatworms and nematodes have both piRNAs, the actual mature piRNAs and the protein machinery in nematodes are unique, and so far do not resemble what has been described for any other animal taxon. Thus, piRNAs of free-living flatworms and nematodes are clearly distinguishable. Using small-RNA sequencing and available transcriptome data, we here describe the first small RNA complement (miRNA and piRNA) as well as the corresponding protein machinery for a gastrotrich, namely the freshwater species
L. squamata. The analyses revealed that the genome of
L. squamata encodes an extensive small RNA repertoire of
bona fide miRNAs and piRNAs, as well as the full miRNA & piRNA biogenesis and RNAi protein machinery. The comparison of miRNAs, but also of piRNAs and RNAi protein machinery, to those of flatworms and nematodes together supports a close relationship of gastrotrichs to flatworms and not to nematodes.
3. Discussion
Gastrotrichs are an abundant group of organisms in marine habitats with important roles in aquatic ecosystems [
1,
4]. However, only little is known about their molecular biology [
5]. In order to obtain a better molecular understanding of this, and to contribute to the discussion on their phylogenetic relatedness to other protostome species, we have studied the miRNA and piRNA gene-regulatory pathways in the chaetonotid gastrotrich
Lepidodermella squamata using a ‘multi-omics’ approach.
The miRNA complement of
L. squamata consisted of highly conserved and some novel miRNAs. There was an almost complete set of the expected miRNA families and the spiralian-specific MIR-1992, but no Ecdysozoa- or Nematoda-specific miRNAs were detected. Some of the few losses observed in
L. squamata were shared with all other flatworms studied so far [
40,
49,
61,
62]. Interestingly, some of the gastrotrich miRNAs showed extended precursor lengths of up to 386 nucleotides, which, to this date, has only been described in flatworms [
40,
51,
63]. Taken together, we found a near complete set of miRNAs that by the presence and absence of particular miRNA families, their high sequence similarities and the length of some precursors support a close relationship of gastrotrichs and flatworms.
The conservation of individual piRNA sequences is extremely low [
64,
65,
66] and thus their potential for evolutionary or phylogenetic analyses is thought to be limited. However, we showed that similarities in piRNA biogenesis patterns can be used to infer relatedness of animal groups such as the gastrotrichs to flatworms or nematodes. While both free-living flatworms and free-living nematodes possess piRNAs, many parasitic lineages have lost them altogether [
48] or, in the case of nematodes, have retained a highly derived pathway, producing piRNAs that are clearly distinct from most other animals studied thus far [
52]. For instance, piRNAs in
C. elegans are not produced from long primary transcripts derived from genomic regions known as piRNA clusters and have a size of 21 nucleotides instead of the conventional 24–32 nt lengths typical of piRNAs in other animals [
67].
Although, the four identified piRNA clusters do not constitute the full piRNA complement of
L. squamata, they already allowed for some insights into the piRNA processing machinery in Gastrotricha. Three piRNA clusters with piRNAs mapping to only one DNA strand and a fourth cluster including piRNAs from both DNA strands were identified. The piRNAs of the latter cluster showed processing patterns consistent with the so-called ping-pong amplification mechanism [
52], and the majority of the piRNAs were derived from the minus strand and showed 1T bias. In contrast, plus strand-derived piRNAs, which are antisense to the putative precursor transcript, showed strong bias towards adenine at the 10th position and overlapped minus strand-derived piRNAs by exactly ten nucleotides. It is known that the 10A bias is caused by the intrinsic affinity of PIWI-clade proteins for targets bearing adenine at position target-1 or t1 (i.e., in front of the piRNA 5′ end). When PIWI cleaves their targets in a piRNA-directed manner, it does so at position 10 of the target counting from the piRNA 5′ end. As a consequence, adenine at position t1 becomes adenine at position guide-10 or g10 [
68]. piRNAs generated in this way were initially termed as secondary piRNAs, although the term ‘responder piRNA’ has been recommended [
52]. In
Drosophila melanogaster (an ecdysozoan), the ping-pong cycle involves two of the three PIWI-clade proteins present in the genome. Aubergine (
Aub) binds piRNAs produced from piRNA precursors while Argonaute 3 (
AGO3) associate with complementary transcripts, usually corresponding to TEs. Aub-bound sequences tend to start with uridine, while AGO3-associated piRNAs show 10A, but no 5′ bias [
69]. A similar heterotypic ping-pong amplification cycle was also recently described in mollusks, where two different PIWI proteins bind sense and antisense transcripts, respectively [
70]. These two proteins bind piRNAs of 29–30 nt (with 1U bias) and 24–25 nt (with 10A bias), respectively. Interestingly, homotypic ping-pong amplification was also described in the oyster
Crassostrea gigas, involving a single PIWI-clade protein acting on both sense and antisense transcripts. As a consequence of concomitant homotypic and heterotypic ping-pong amplification in the gonads of
C. gigas, 1U-biased piRNAs showed a size distribution of 29–30 nt, while partially complementary, 10A-biased piRNAs were either 24–25 (heterotypic) or 29–30 (homotypic) nt long.
The piRNAs of cluster #4 piRNAs in
L. squamata (
Figure 2a–e) meet this pattern, with the only difference that the 1U-biased piRNAs derived from the putative piRNA precursor transcript were slightly longer in Gastrotricha (30–32 nt vs 29–30 nt,
Figure 2a,b). Mammalian PIWI proteins are also associated with piRNAs of slightly different lengths. For instance, mouse piRNAs bound to MILI or MIWI2 show mean lengths of 26 and 28 nt, respectively [
71]. 1U-biased piRNAs in
L. squamata tend to be slightly longer than in the majority of animals analyzed to date [
70,
71,
72]. They are very similar to what was previously reported in the planarian
S. mediterranea [
42], where bands of 31–32 nt were identified by both sequencing and Northern blot analyses. However, an exhaustive analysis of piRNA lengths across metazoan species is still lacking.
Although we cannot exclude the existence of 21U piRNAs (as not all 21 nt reads were assigned to other classes), the presence of canonical piRNAs and their genomic distribution in piRNA clusters suggest that piRNA processing patterns in
L. squamata is different than in nematodes, but, instead, very similar to what was recently reported for mollusks [
70] (a spiralian) and comparable in size to the piRNA complement in flatworms [
42,
45]. Accordingly, the piRNA data indicate that Gastrotricha (
L. squamata) are more closely related to Platyhelminthes within Spiralia rather than to Nematoda within Ecdysozoa.
A strict homology approach to the protein repertoires of the RNAi and PIWI protein machinery across protostomes (i.e., spiralians and ecdysozoans) corroborated that nematodes (including
C. elegans) have a more derived RNAi pathway [
52] compared with other spiralians. Among spiralians, we found that RNAi proteins from gastrotrichs (including
L. squamata) and flatworms are very similar in terms of RNAi protein phylogenetic distribution and domain architecture, supporting their closer relationship and the Rouphozoa phylogenetic clade [
20,
21]. On the other hand, our analyses on PIWI protein repertoires in
L. squamata revealed all necessary protein machinery for piRNA biogenesis [
25]. In addition, the presence of several PIWI-like proteins, some of them with similarity to
D. melanogaster Aub and
AGO3, suggest that piRNAs might be controlled by a system similar what has been observed in Arthropoda. Although further experimental evidence is needed to support this hypothesis, we demonstrated that
L. squamata PIWI repertoire is not similar to nematodes studied so far. Taken together, all data inferences covering reciprocal BLAST, functional domain annotation, and phylogenetic analysis, strongly support the close phylogenetic relationship between Gastrotricha and Platyhelminthes, which is congruent with miRNA and piRNA results.
In conclusion, we show that the freshwater gastrotrich Lepidodermella squamata (Chaetonotida) expresses a comprehensive set of bona fide miRNAs and piRNAs and the corresponding biogenesis and RNAi machinery. Furthermore, we demonstrate that the occurrence, structure and sequence of miRNA genes and families is useful to assess the phylogenetic position of gastrotrichs and strongly supports close relatedness to flatworms and not to nematodes. Congruently, by comparing the length and structure of mature and precursor piRNAs, we show that the presence of bona fide piRNAs and ‘responder piRNA’ in particular, clearly supports a closer relationship to flatworms, as well. This is consistently followed by the presence/absence distribution and domain architecture of RNAi proteins supporting this Rouphozoa clade, too.
With miRNA and piRNA complements for only one species we certainly cannot make statements about the monophyly of gastrotrichs [
20]. Nevertheless, these first small non-coding RNA complements of
L. squamata add important information for an understanding of the evolution of non-coding RNAs across the metazoan tree of life. We showed that not only the highly conserved miRNAs, but also the piRNAs and the corresponding protein machinery can add valuable phylogenetic information in support of a close relationship of
L. squamata to flatworms.
Our knowledge on the evolution of ncRNAs is broad but not very detailed because the annotation of bona fide ncRNA complements, especially for microscopic organisms, remains a challenge as it requires material from often very sparse samples, expert knowledge, and manual curation. Surprisingly, even major endeavors such as the International Helminth Genomes Consortium have neglected ncRNA analyses despite the obvious availability of high quality genome assemblies [
73]. In the absence of such assemblies, our pipeline miRCandRef can be used for miRNA and piRNA annotations that can be complemented by information of key RNAi and biogenesis proteins through readily available transcriptome assemblies for all major metazoan groups toward a better understanding on the evolution of coding and non-coding elements.