TAS3 miR390-dependent loci in non-vascular land plants: towards a comprehensive reconstruction of the gene evolutionary history

TAS3 loci in Bryophyta (classes Sphagnopsida and Takakiopsida)

It is commonly accepted that mosses of classes Sphagnopsida and Takakiopsida represent most basal lineages in Bryophyta (Shaw et al., 2010; Shaw, Szövényi & Shaw, 2011; Rosato et al., 2016). Previously, using primers, which have allowed us to detect pre-miR390 and TAS3 loci in Bryopsida and some other moss classes, we failed to identify pre-miR390 and TAS3 genes in genus Sphagnum (Krasnikova et al., 2013). However, a predicted sequence of pri-miR390 from Sphagnum fallax was recently reported (Xia, Xu & Meyers, 2017). This finding prompted us to re-evaluate the occurrence of TAS3-like loci in Sphagnopsida. To this end, we designed a new pair of degenerated PCR primers Spha-TASP and Spha-TASM, which differed from those used previously (Krasnikova et al., 2011; Krasnikova et al., 2013). As a positive control, we used plasmid DNA carrying cloned TAS3 gene of Andreaea rupestris, a representative of basal Bryophyta (Krasnikova et al., 2013). Like the positive control, two total DNA probes from Sphagnum angustifolium and S. girgensohnii gave a single main PCR product of the expected size (Fig. 1). Cloning and sequencing of these PCR fragments revealed two TAS3-like primary structures having 285 (S. angustifolium) and 292 (S. girgensohnii) bases in length and exhibiting 96% identity (e-value = 2e−131). We named these loci as Sphan-285 and Sphgi-292, (Fig. 2, Fig. S1 and Table 1). Despite that the degenerate miR390-mimicking primers were used for amplification of these loci, miR390 recognition sites in TAS3 species well corresponded to genomic loci of TAS3 extracted from genomic and transcriptomic data of other plants from genus Sphagnum (Fig. 2).

Peatmosses S. angustifolium and S. girgensohnii belong to subgenera Cuspidata and Acutifolia, respectively (Shaw et al., 2010; Shaw et al., 2016). To extend search for TAS3-like loci inside genus Sphagnum we performed bioinformatics analysis of the nucleotide sequences in databases available at NCBI (Sequence Read Archive) and Phytozome (version 12.1). Phytozome has recently released genome assembly of bog moss S. fallax (version 0.5). Bog moss belongs to subgenus Cuspidata and represents the most closely related moss to S. angustifolium (Shaw et al., 2016). BLASTN search at Phytozome allowed us to reveal a TAS3-like locus (supercontig super_37), which has 100% identity to the TAS3 locus of S. angustifolium sequenced in this study (Fig. S1 and Table 1). Unexpectedly, we found an additional TAS3-like locus in S. fallax (transcript Sphfalx0293s0011, supercontig super_293). This TAS3 locus in bog moss has 277 nucleotides in length and showed only a distant relation to the S. angustifolium TAS3 (Fig. 2, Fig. S1 and Table 1).

To further analyze Sphagnopsida TAS3-related loci, we used BLAST analysis of Sequence Read Archive (SRA), which is the NCBI database collecting sequence data obtained by the use of next generation sequence (NGS) technology. Assembly of sequence reads of S. recurvum (subgenus Cuspidata) retrieved by BLAST search using S.fallax sequences as queries revealed two TAS3 loci (Table 1). The first locus (Sphre-283) is 283 nucleotides in length and has 98% identity to Sphan-285. The second locus (Sphre-277) shows 98% identity to Sphfalx0293s0011 (Table 1, Fig. S1). These findings indicate that two distant TAS3 loci in species of a particular subgenus of genus Sphagnum are extremely similar.

We also analyzed the SRA database of subgenus Sphagnum (Shaw et al., 2010; Shaw et al., 2016). It was found that S. magellanicum belonging to this subgenus also encode two TAS3 loci called Sphma-285 (285 nt size) and Sphma-286 (286 nt size) (Fig. S1 and Table 1). Unlike S. fallax and S. recurvum, in S. magellanicum TAS3 loci are more similar, showing 86% identity (Fig. 2). Both Sphma-285 and Sphma-286 had 85% identity to Sphan-285 (Fig. 2). It was found that TAS3-like locus (Sphpa) from one more representative of subgenus Sphagnum (S. palustre) exhibited 98% identity to Sphma-285 (Fig. S1 and Table 1). The SRA database also contained sequence reads of two representatives from subgenus Subsecunda (Shaw et al., 2010; Shaw et al., 2016). Our BLAST analysis and subsequent assembly of retrieved reads revealed a single TAS3 locus in S. cribrosum (Sphcri, 291 nt size) showing 95% identity to Sphan-285 and 81% identity to Sphma-286 (Fig. 2, Fig. S1 and Table 1) and a partial TAS3-like sequence in S. lescurii (Fig. S1 and Table 1).

Analysis of the SRA database of Takakia lepidozioides (class Takakiopsida) allowed us to reveal only one TAS3-like sequence (Takle-207) (Fig. S1 and Table 1). The same sequence was revealed in a longer assembly which was found recently upon search of 1KP database (Xia, Xu & Meyers, 2017).

Since Takakiopsida and Sphagnopsida are most basal sister lines to all other Bryophyta (Shaw et al., 2010; Shaw, Szövényi & Shaw, 2011; Rosato et al., 2016; Puttick et al., 2018), it was very interesting to compare the structural organization of Takakiopsida and Sphagnopsida TAS3 loci with other classes of Bryophyta. Our previous detailed analysis of approximately 40 TAS3 loci in Bryophyta (Krasnikova et al., 2011; Krasnikova et al., 2013) showed that the general structure of moss TAS3 is similar in all taxa and fits the structural organization of Physcomitrella patens genes, comprising dual miR390 target sites on the 5′ and 3′ borders and internal monomeric tasiAP2 sequence followed by tasiARF sequence positioned in 20–30 bases. We revealed that phylogenetic tree of TAS3-like loci in Bryophyta showed clear subdivision of their sequences into two main clades (see Fig 5 in Krasnikova et al., 2013). The first group was formed by a cluster of sequences close to P. patens TAS3 species PpTAS3a, PpTAS3d, and PpTAS3f, and the second one—by those close to PpTAS3b, PpTAS3c, and PpTAS3e. The recent paper on the structure of TAS3 loci in lower land plants (Xia, Xu & Meyers, 2017) has shown the structure-functional basis for this phylogenetic subdivision. TAS3 species of the first group (PpTAS3a/PpTAS3d/PpTAS3f cluster) were shown to form class III of TAS3-like loci and contain, in addition to the previously reported tasiAP2 and tasiARF-a2 sequences, newly discovered tasiARF-a3 sequence positioned 5′ according to tasiAP2 (Fig. 3). Among TAS3 species of basal Bryophyta, Andreaea rupestris locus 13-Aru (Krasnikova et al., 2013) belongs to class III (Fig. 3).

BLAST comparison of T. lepidozioides TAS3 with known Bryopsida loci showed that Takle-207 belongs to class II of TAS3 with typical positioning of tasiAP2 and tasiARF-a2 sequences (Fig. 2 and Fig. S1). On the other hand, none of Sphagnopsida TAS3-like sequences (Table 1) showed conventional internal structural organization of the most moss TAS3 species. The only recognizable conserved site, except miR390-targeting regions, was identified as tasiARF-a2 sequence, which was found to be conserved between two very distant TAS3 loci in S. fallax and S. recurvum (Fig. 2). The mentioned above tasiARF sequences, tasiARF-a2 and tasiARF-a3, showed no sequence similarity suggesting their independent origins. These tasiRNAs were found to be formed from different strands of the TAS3 dsRNA intermediate and target different regions of ARF genes (Xia, Xu & Meyers, 2017). Inhibition of production of both tasiARF RNAs in P. patens resulted in obvious developmental defects exhibited, in particular, as alterations in gametophore initiation, protonemal branch determinacy and caulonemal differentiation (Plavskin et al., 2016).

Comparison of nucleotide sequences between TAS3 species of several moss classes revealed in many plants obvious similarity of nucleotide sequence blocks including tasiAP2 site and immediate upstream 21 bp block occurring in the same 21-bp-phase (Fig. 4A). We hypothesized that this sequence block may correspond to novel previously unrecognized ta-siRNA in many moss species. This putative siRNA in its single-stranded form really presents in P. patens transcriptome as minus-sense 21-nucleotide ta-siRNA (see NCBI SRA accessions SRX903096–SRX903105) like tasiARF RNA (Arif et al., 2012). Thus, we speculated that novel hypothetical ta-siRNA might be produced from TAS3, and its minus-strand is complementary to uncharacterized well-conserved, protein-coding moss mRNAs which have homologs also in conifers and angiosperms (Fig. 4B; Fig. S2).

TAS3 loci in Anthocerotophyta

Taking into account the finding of TAS3-like loci in classes Sphagnopsida and Takakiopsida and previously published data (Krasnikova et al., 2013; Xia, Xu & Meyers, 2017), one can conclude that the only remaining blind-spot in land plants with respect to TAS3 is represented by phylum Anthocerotophyta. Relationships between liverworts, mosses and hornworts are still obscure. Moreover, the question remains which bryophyte phylum is a sister line to all other land plants (Qiu, 2008; Shaw, Szövényi & Shaw, 2011; Harrison, 2017; Puttick et al., 2018). Recent molecular phylogenetic analysis, in which three bryophyte lineages were resolved, revealed that a clade with mosses and liverworts could form a sister group to the tracheophytes, whereas the hornworts is sister line to all other land plants (Wickett et al., 2014). However, analyses of the plastid genome sequences suggested another branching order of the phylogenetic tree, with hornworts rather than moss/liverwort clade being a sister group to tracheophytes (Lewis, Mishler & Vilgalys, 1997; Samigullin et al., 2002; Ruhfel et al., 2014; Lemieux, Otis & Turmel, 2016). Moreover, some very recent nuclear gene comparisons also suggested that hornworts could be a sister clade to tracheophytes, and liverworts plus mosses might be closer to a common ancestor of land plants (Rosato et al., 2016; Bowman et al., 2017). However, this ancestor could have more tracheophyte-like characteristics than Setaphyta (liverworts/mosses) due to secondary simplification in liverworts (Puttick et al., 2018).

Analysis of the SRA database of Anthocerotophyta revealed a putative TAS3-like sequence in Folioceros fuciformis (family Anthocerotaceae). Unexpectedly, the discovered TAS3-like sequence (Folfu) was found to be 244 nucleotides in length and obviously similar to Bryophyta class III TAS3 species (Fig. 3, Fig. S3 and Table 2). The identity of Folfu to some moss TAS3 sequences exceeds 80% being therefore even higher than between some related Bryopsida species (Fig. 3). Thus these data clearly indicate a close relation of TAS3 in Anthocerotophyta to Bryophyta TAS3 (excepting Sphagnopsida).

TAS3 loci in Marchantiophyta

Some of the recent molecular phylogenetic reconstructions suggested that Marchantiophyta species could represent a sister clade to all other land plants (see above). Therefore, finding and comparative analyses of TAS3 loci in this taxon represented a significant interest for understanding early events in TAS3 evolution. In contrast to class Marchantiopsida, where putative TAS3 and pre-miR390 loci were previously identified (Krasnikova et al., 2013; Lin et al., 2016; Tsuzuki et al., 2016), for class Jungermanniopsida only potential pre-miR390 loci were found in Pellia endiviifolia and Harpanthus flotovianus (Krasnikova et al., 2013; Alaba et al., 2015). Assuming that miR390 was found to be among eight most conserved miRNA species in land plants (Xia et al., 2013; You et al., 2017; Liu et al., 2018), Jungermanniopsida could be expected to encode TAS3 loci.

To detect new potential TAS3 loci, we performed BLAST analysis of the SRA database for species of class Jungermanniopsida using Marchantia polymorpha TAS3 sequence (1-Mpo) as a query. Using this approach we revealed a set of reads and assembled a single TAS3-like locus (Pelen-192) for Pellia endiviifolia (192 nt size). In addition, TAS3 locus of 226 nucleotides in length was found in Metzgeria crassipilis (Metcr-226) (Fig. 5, Table 2, Fig. S3). The latter locus was also recently revealed in a search of 1KP database (Xia, Xu & Meyers, 2017).

TAS3 1-Mpo sequence was further used for BLAST analysis of other Marchantiopsida sequences available at the NCBI SRA database. As a result, we retrieved sequence reads and assembled five full-length TAS3-like sequences in Plagiochasma appendiculatum (Plaap-247), Dumortiera hirsuta (Dumhi-243), Marchantia emarginata (Marem-262), Ricciocarpos natans (Ricna-235) and Conocephalum japonicum (Conja-252) (Fig. 5, Table 2, Fig. S3). Recent bioinformatics analysis of 1KP database revealed three additional full-length TAS3-like sequences in Conocephalum conicum, Lunularia cruciata and Marchantia paleaceae (Xia, Xu & Meyers, 2017) (Table 2). Thus, totally 11 TAS3-like loci have been found in Marchantiophyta.

TAS6 loci in Bryophyta

Previous studies of P. patens revealed three novel non-coding PHAS loci (TAS6) which were located in rather close genomic proximity to PpTAS3 loci (PpTAS3a, PpTAS3d, and PpTAS3f) and expressed as common RNA precursors with these TAS3 species (Cho, Coruh & Axtell, 2012; Arif et al., 2012; Arif, Frank & Khraiwesh, 2013). Moreover, miR529 and miR156 were suggested to influence accumulation of ta-siRNAs specific not only for TAS6, but also for PpTAS3a (Cho, Coruh & Axtell, 2012). We have found that localization of TAS6 loci close to TAS3 genes in common transcripts was not unique for P. patens (subclass Funariidae), since these loci were also found to be encoded by three other mosses of subclasses Bryidae and Dicranidae (Krasnikova et al., 2013).

For further search of the combined TAS6/TAS3 loci, we performed bioinformatics analysis of 1KP database. Although nucleotide sequences of miR156 and related miR529, as well as their recognition sites in RNA transcripts, are highly conserved among land plants (Morea et al., 2016; Axtell & Meyers, 2018), the internal sequences between dual miR156/miR529 recognition sites show little or no similarity even between different TAS6 loci of P. patens (Arif et al., 2012). So we used, as queries for BLAST search, the individual full-length TAS6/TAS3 loci including most characterized locus encoding PpTAS3a (Fig. 6), as well as those for PpTAS3d and PpTAS3f. First, it was found that in addition to four previously found Bryopsida species, encoding TAS6/TAS3 loci, these loci could be revealed in basal subclasses Timmiidae (Timmia austriaca) and Diphysciidae (Diphyscium foliosum) (Shaw, Szövényi & Shaw, 2011; Table 3, Fig. S4). List of TAS6/TAS3 loci in other moss subclasses was also significantly extended: we found 18 new loci in Bryidae, seven loci in Dicranidae and four loci in Funariidae (Table 3, Fig. S4). These novel loci showed recognizable but varying sequence similarities to the PpTAS3a-containing locus (Fig. 6). Second, most importantly, putative TAS6/TAS3 loci were revealed in four basal classes of Bryophyta, namely, Tetraphidiopsida, Polytrichopsida, Andreaeopsida and Takakiopsida (Table 3, Fig. S4). These novel loci had a similar organization to Bryopsida TAS6/TAS3 species (Fig. 6). However, no TAS6-specific sequence signatures were found in the vicinity of genomic S. fallax and M. polymorpha TAS3 loci upon analysis of the corresponding Phytozome genome contigs.

Phylogeny of SGS3 as a characteristic molecular component of TAS3 pathway

It was shown that some species of green algae could encode ancient types of dicer-like proteins, RDRs, and AGOs. On the other hand, no encoded SGS3 proteins were revealed for these algae (Zheng et al., 2015). Since SGS3 was found to be essential for production of tasiARF RNAs in moss P. patens (Plavskin et al., 2016), we performed sequence analyses to identify possible SGS3 genes in charophytes. For identification of SGS3 protein orthologs among land nonvascular plants and charophytes, we used as a query the most conserved region of P. patens SGS3 including short zinc binding zf-XS domain and RNA recognition XS domain (Bateman, 2002; Zhang & Trudeau, 2008). Importantly, the short N-terminal zf-XS domain is characteristic for functional SGS3 proteins, since the XS domain-containing protein of Selaginella moellendofii lacking TAS-generating machinery (Banks et al., 2011) possesses no zf-XS domain upstream of XS domain and instead contains the C-terminal RING zf region (see NCBI accession XP_002979112). However, it should be noted that the lack of TAS3 pathway and SGS3 is not universal for lycophytes (Xia, Xu & Meyers, 2017).

In addition to class Bryopsida, SGS3 protein sequences were revealed for members of classes Marchantiopsida, Jungermanniopsida, Anthocerotopsida, Takakiopsida and Sphagnopsida (Fig. 7 and Fig. S5). Most importantly, search for the SGS3 coding sequences in transcriptomes of four charophyte classes (Klebsormidiophyceae) also revealed the SGS3-like proteins in representatives of all these taxa (Fig. 7, Figs. S5, S6). This observation was in agreement with the fact that SGS3-like coding sequence was found in the fully sequenced and annotated genome of Klebsormidium nitens (NCBI accession GAQ92898) (Hori et al., 2014). Moreover, the characteristic motifs of land plant SGS3 proteins (Bateman, 2002) were revealed in the protein sequences from charophyte algae (Figs. S5, S6).

Importantly, in the dendrogram based on comparisons of 27 aligned SGS3 protein sequences, the position of charophytes (Fig. 7) nearly corresponded to the commonly accepted Viridiplantae phylogenetic tree (Shaw, Szövényi & Shaw, 2011; Delwiche & Cooper, 2015; Harrison, 2017), where class Zygnemophyceae (Spirogyra pratensis) was a sister group for all land plants. It has become clear that evolving the SGS3-like genes was not directly connected to the appearance of TAS loci in Viridiplantae, since Chlorophyta species, lacking SGS3, encode not only critical enzyme machinery including DCLs, RDRs, and AGOs (You et al., 2017), but also PHAS loci (Zheng et al., 2015). Despite our extensive searches, no SGS3 genes could be identified also in brown and red algae, and this is in agreement with previously published data on green algae (Zheng et al., 2015).