Multi-gene-based phylogenetic analysis of oligotrich ciliates with emphasis on two dominant groups: cyrtostrombidiids

Phylogenetic analyses of ciliated protists are frequently based on single molecular markers, usually the small subunit ribosomal RNA gene (SSU rDNA), despite the well-known limitations of this approach. Here, 78 new sequences of three linked genes (SSU rDNA, ITS1-5.8S rDNA-ITS2, LSU rDNA) were characterized and applied to phylogenetic analyses of oligotrichs (s. str.). It was found that: (1) three taxa, that is tontoniids, pelagostrombidiids and cyrtostrombidiids should be split from the family Strombidiidae (s.l.), which supports Agatha’s classification based on morphological characters; (2) the families Tontoniidae and Cyrtostrombidiidae are both monophyletic whereas Strombidiidae is polyphyletic; (3) the positions of the families Cyrtostrombidiidae and Pelagostrombidiidae varied in different trees although with low support values; (4) the close relationship between Varistrombidium and Apostrombidium is confirmed, which updates the evolutionary hypothesis for oligotrichs based on ciliary patterns; and (5) two relatively stable clades were found in the family Strombidiidae.


Introduction
The oligotrichous ciliates (s.l.) are a group that are often present in great abundance in oceanic waters and play an important role in the microbial loop (Calbet and Saiz, 2005;Edwards and Burkill, 1995;Pierce and Turner, 1992).Nevertheless, the taxonomy and phylogeny of this lineage are insufficiently studied.It is conservatively estimated that there are more than 1000 oligotrich species worldwide, but only about 100 well described morphospecies, representing ca. 15 genera, have been documented (Liu et al., 2015a;Liu et al., 2015b;Song, 2005;Song and Bradbury, 1998;Song and Packroff, 1997;Song et al., 2000;Tsai et al., 2010;Xu and Song, 2006;Xu et al., 2006).Molecular phylogenetic analyses of oligotrichs are even more poorly documented, since only about 30 species from the subclass Oligotrichia have available SSU rDNA sequences.
For much of its history, the subclass Oligotrichia was not subdivided because somatic ciliary patterns did not seem to be restricted to distinct genera (Agatha, 2004b).Agatha divided Oligotrichia into four families (Tontoniidae, Pelagostrombidiidae, Cyrtostrombidiidae, and Strombidiidae) based on their morphological, ontogenetic and ultrastructural characters (Agatha, 2004b).However, in a later classification scheme proposed by Lynn (2008), species of the families Pelagostrombidiidae and Cyrtostrombidiidae were conservatively assigned into Strombidiidae.Until now, these two schemes have not been tested in terms of the placement of Pelagostrombidiidae, since only two very short SSU rDNA sequences (~170bp) are available for the family Pelagostrombidiidae.More sequences, especially from the family Pelagostrombidiidae, are urgently needed to reveal the evolutionary relationships among oligotrich taxa.Hitherto, most molecular phylogenetic studies focusing on oligotrichs have been based only on SSU rDNA data (Gao et al., 2009;Liu et al., 2015a;Liu et al., 2012;McManus et al., 2010;Song et al., 2013;Tsai et al., 2010;Zhang et al., 2010) , and phylogenetic assignments of some taxa are ambiguous due to conflicting results from SSU rDNA trees and morphological characters.For example, in a recent phylogenetic study based on SSU rDNA sequences (Tsai et al., 2015), Cyrtostrombidium is sister to the cluster consisting of the family Tontoniidae and the genus Apostrombidium.However, in cladistic analyses inferred from morphological data (Agatha, 2004a), Cyrtostrombidium is closer to some genera in Strombidiidae than to Tontoniidae.In recent years the ITS1-5.8SrDNA-ITS2 region and the large subunit ribosomal RNA gene (LSU rDNA), which are tandemly linked to the SSU rDNA gene, are increasingly used in phylogenetic analyses of ciliates (Gao et al., 2014;Hewitt et al., 2003;Huang et al., 2014;Liu, 2011;Marande et al., 2009;Yi et al., 2014;Zhao et al., 2015).Previously, phylogenetic trees inferred from just a few ITS1-5.8SrDNA-ITS2 region sequences have been applied in analyses of evolutionary relationships within the Oligotrichia (Li et al., 2013;Snoeyenbos-West et al., 2002).
Although the LSU rDNA has not been utilized to analyze Oligotrichia, its utility has been demonstrated in an investigation of a related group, the Tintinnida (Spirotrichea: Choreotrichia) (Santoferrara et al., 2013).
As mentioned above, the phylogeny of oligotrichs is still poorly understood, which is mainly due to: (i) low numbers of available sequences for morphologically described species and (ii) low numbers of gene markers (usually SSU rDNA) applied in phylogenetic analyses.Therefore, in the present investigation, 76 new sequences (17 SSU rDNA, 27 ITS1-5.8SrDNA-ITS2, 32 LSU rDNA) of oligotrichs were added and multi-gene-based analyses were applied in order to increase our understanding of the phylogeny of this group.

Sampling and identification
Sampling information for the 36 newly sequenced taxa (Fig. 1) are listed in Table 1.Species identifications were made based on microscopical observation of specimens both in vivo and following silver staining (Song et al., 2013).The genomic DNA is used in some taxa which were investigated in previous studies (Gao et al., 2009;Li et al., 2013;Liu et al., 2012;Liu et al., 2011;Zhang et al., 2010).

DNA extraction, PCR amplification and sequencing
One or more individuals of each species were isolated for DNA extraction.Total genomic DNA was extracted using the REDExtract-N-Amp Tissue PCR Kit (Sigma, St. Louis, USA) or the DNeasy Blood & Tissue Kit (Qiagen, CA), following the manufacturer's instructions.
The PCR amplification of the SSU rDNA was performed using the eukaryotic universal forward 18SF (5'-AAC CTG GTT GAT CCT GCC AGT-3') and the reverse 18SR (5'-TGA TCC TTC TGC AGG TTC ACC TAC-3') primers (Medlin et al., 1988).A fragment of about 500bp covering the ITS1, 5.8S ribosomal gene and ITS2 was amplified using primers ITS-F and ITS-R as described in Gao et al. (2012).Primers 28S-1F and 28S-3R were used to amplify part (about 1800 bp) of the LSU rDNA gene (Moreira et al., 2007).In some cases, the ITS1-5.8SrDNA-ITS2 region and the part of the LSU rDNA gene were amplified together using primers ITS-F and 28S-3R.PCR conditions were as follows: 5 min initial denaturation at 94°C; 35-40 cycles of 15-30s at 94°C, 60-75s at 56-60°C and 60-90s at 72°C; with a final extension of 7-10 min at 72°C.The PCR product was purified and inserted into the pMD™ 19-T vector (Takara Biotechnology, Dalian Co., Ltd.).Subsequently, the vector was transferred into competent E. coli DH5α cells.
Commercial sequencing was carried out on an ABI-PRISM 3730 automatic sequencer (Applied Biosystems, USA).Wherever possible, we selected the same DNA source for the amplification of SSU rDNA, ITS1-5.8SrDNA-ITS2, and LSU rDNA.

Sequence alignment and phylogenetic analyses
The newly obtained sequences were deposited in the NCBI database (for accession numbers see Table 1 in bold).Other sequences used in the present analyses were downloaded from the NCBI database.Three hypotrich and two halteriid taxa were selected as the outgroup.Sequences of SSU rDNA, ITS1-5.8SrDNA-ITS2, and LSU rDNA were aligned using the online server GUIDANCE (http://guidance.tau.ac.il/) with the alignment algorithm MAFFT (Penn et al., 2010).Ambiguous columns in the alignment were removed based on confidence scores calculated by GUIDANCE.The resulting alignment was further manually checked in SeaView v. 4 (Gouy et al., 2010).The final alignments included 1635 sites of SSU rDNA (75 taxa), 400 sites of ITS1-5.8SrDNA-ITS2 (52 taxa), and 1719 sites of LSU rDNA (49 taxa).The individual gene alignments were then concatenated to build a three-gene, 75-taxon matrix with 3754 characters using the concatenate option in SeaView v. 4..The taxa whose whole sequence data of some genes are unavailable were treated as missing data.
Phylogenetic trees were constructed as described in Gao et al. (2016).Maximum likelihood (ML) analyses with different gene selections (SSU rDNA, ITS1-5.8SrDNA-ITS2, LSU rDNA, and all three concatenated) were performed on the CIPRES Science Gateway (URL: http://www.phylo.org/sub_sections/portal)(Miller et al., 2010), with RAxML-HPC2 on XSEDE using the GTR + I + G model as selected by Modeltest v. 3.4 (Posada and Crandall, 1998).Searches for the best tree were conducted starting from 100 random trees, and 1000 nonparametric bootstrap replicates were done to assess the reliability of the internal branches.Bayesian inference (BI) analyses were also performed on the CIPRES Science Gateway using MrBayes v. 3.1.2on XSEDE with the GTR + I + G model selected by MrModeltest v. 2.2 (Nylander, 2004).Markov chain Monte Carlo (MCMC) simulations were run with two sets of four chains for 4,000,000 generations.Sampling frequency is every 100 generations.The first 25% of sampled trees were discarded as burn-in prior to consensus tree construction.Trees were viewed with MEGA v5 (Chen et al., 2015).

Topology testing
To test the monophyly of each oligotrich taxon and the robustness of phylogenetic associations of particular interest, the Approximately Unbiased (AU) tests was used (Shimodaira, 2002).Ten constraint ML topologies were generated base on the concatenated data and then compared with the unconstrained ML topology.For all constraints, internal relationships within the constrained groups were unspecified, and relationships among the remaining taxa were likewise unspecified.The site-wise likelihoods for the resulting constrained topologies and the non-constrained ML topology were calculated using PAUP (Swofford, 2002) and then analyzed in CONSEL (Shimodaira and Hasegawa, 2001) with standard parameters to obtain p-values.

Topology based on SSU rDNA
The topologies generated by BI and ML are generally consistent, therefore only the BI phylogenetic tree is presented with posterior probability values from BI and bootstrap values from ML analysis (Fig. 2).In the SSU rDNA tree, the traditional oligotrichs form two main clades.One clade is composed of the subclass Choreotrichia, the other clade comprises all the members of Oligotrichia (0.99 BI, 67% ML).well-supported group (0.99 BI, 70% ML).Outside the previous two clades, species generally form polytomies except for some small groups comprising closely related species, e.g.Strombidium basimorphum and S. cf.basimorphum clustering in a well-supported branch (1.00 BI, 86% ML), two populations of Strombidium paracalkinsi branching with each other (1.00 BI, 99% ML), Varistrombidium kielum pop2 and pop3 clustering together (1.00 BI, 71% ML), and Novistrombidium apsheronicum and N. testaceum grouping together with high support (1.00 BI, 83% ML).

Topology based on ITS1-5.8S rDNA-ITS2 region
For the ITS1-5.8SrDNA-ITS2 dataset with 38 oligotrichs and 7 choreotrichs, the BI tree is presented with support values from both BI and ML algorithms (Fig. 3).The topology of the ITS1-5.8SrDNA-ITS2 region tree has some similarities to that of the SSU rDNA tree, for example: (i) Lynnella branches with choreotrichs; (ii) the monophyly of Tontoniidae and the non-monophyly of both Strombidiidae and Strombidium; (iii) Novistrombidium species scattered into three parts, with N. apsheronicum and N. testaceum forming a fully supported clade; and (iv) Parallelostrombidium forms a highly supported clade with P. conicum excluded (1.00 BI, 97% ML).There are, however, also some differences in the topology of the ITS1-5.8SrDNA-ITS2 tree compared to that based on SSU rDNA, for example: (i) Tontoniidae and Cyrtostrombidiidae each clusters with Strombidium species respectively, rather than grouping together; and (ii) Omegastrombidium forms sister groups with Williophrya and Strombidium cf.capitatum, rather than grouping with Antestrombidium wilberti and Novistrombidium orientale.

Topology based on LSU rDNA
Hitherto, only four LSU rDNA sequences of oligotrichs were available prior to this study.In this study, 30 newly characterized oligotrichid LSU rDNA sequences are supplied, and a BI phylogenetic tree based on these data is presented with support values from both BI and ML analyses (Fig. 4).The topology of the LSU rDNA gene tree differs from that of SSU rDNA mainly by: (i) Lynnella grouping with oligotrichs (1.00 BI, 78% ML) rather than with choreotrichs;

Topology based on concatenated data
The concatenated data of SSU rDNA, ITS1-5.8SrDNA-ITS2 and LSU rDNA sequences comes from the same 60 oligotrichs and the same 8 choreotrichs that were used for the SSU rDNA analyses.The topology of the concatenated gene tree (Fig. 5) is similar to that of the SSU rDNA tree, the main differences being: (i) Lynnella grouping with the oligotrichs in the concatenated gene tree rather than with the choreotrichs; (ii) Tontoniidae clustering with the clade of Strombidium cf.basimorphum and S. basimorphum (0.97 BI, 67% ML) rather than grouping with Cyrtostrombidiidae; (iii) Novistrombidium sinicum forming a polytomy with other oligotrichs rather than clustering with the Parallelostrombidium clade.

Classification of families in the subclass Oligotrichia
The classification of the subclass Oligotrichia has been disputed for a long time.Suggestions for the subdivision at family level within this group have included: no subdivision at family level (Modeo et al., 2003); the recognition of two families, i.e., Strombidiidae and Tontoniidae (Lynn, 2008); and the recognition of four families, i.e., Tontoniidae, Cyrtostrombidiidae, Pelagostrombidiidae and Strombidiidae (Agatha, 2004b).Only one gene sequence is available for the Pelagostrombidiidae, so it is impossible for us to infer whether or not this family is monophyletic.The monophyly of Cyrtostrombidiidae is indicated in trees based both on SSU rDNA and on concatenated data, and the monophyly of Tontoniidae is supported in all trees except that based on LSU rDNA sequences (Fig. 4).
The family Strombidiidae is not monophyletic in any of our trees, which is consistent with previous investigations (Li et al., 2013;Song et al., 2015).These findings partly support Agatha's classification based on morphological characters, i.e., that the families Tontoniidae, Pelagostrombidiidae, and Cyrtostrombidiidae should be split from the family Strombidiidae (Agatha, 2004b).
The family Pelagostrombidiidae was established by Agatha (2004b) based on their freshwater habitat and the presence of the neoformation organelle, a permanent tube in which the oral primordium develops (Krainer, 1991;Krainer, 1995).Two genera, Pelagostrombidium and Limnostrombidium, were included in this family.According to cladistic analyses based on morphological data (Agatha and Strüder-Kypke, 2014), Pelagostrombidiidae is closely related to Cyrtostrombidiidae and certain genera in the family Strombidiidae (e.g.Strombidium, Foissneridium, and Williophrya).Limnostrombidium, the only representative of Pelagostrombidiidae in our phylogenetic analyses, grouped either with Strombidium species in the SSU rDNA and concatenated trees or with Laboea and Spirotontonia (Tontoniidae) in the LSU rDNA trees, which is partly consistent with the morphological data.
Unfortunately, molecular data for the type genus Pelagostrombidium is still lacking, therefore the systematic position of Pelagostrombidiidae will remain uncertain pending a re-evaluation following the acquisition of additional data, especially from the type genus and species.
In recent cladistic analyses inferred from morphological data, Cyrtostrombidium is more closely related to certain genera in Strombidiidae than to Tontoniidae (Agatha and Strüder-Kypke, 2014).By contrast, phylogenetic analysis based on SSU rDNA sequences indicated that Cyrtostrombidium is sister to the group consisting of the family Tontoniidae and the genus Apostrombidium (Tsai et al., 2015).In the present multi-gene phylogenetic study, the position of Cyrtostrombidiidae was found to be variable in different trees, either clustering with Tontoniidae or grouping with some species of Strombidium, albeit with low support (Figs. 2-5).The grouping of Cyrtostrombidium and Tontoniidae was not rejected by the AU test while the grouping of Cyrtostrombidium and Strombidiidae was rejected, which may be due to the non-monophyly of Strombidiidae.Therefore, these two possibilities should be further tested when more information becomes available.

The eyespot clade
In a recent study of oligotrich systematics based on SSU rDNA sequence data, Liu et al. (2016) recovered a highly supported clade comprising Strombidium apolatum, S. rassoulzadegani, S. oculatum, S. purpureum and Williophrya maedai, suggesting that the presence of an eyespot might be an important synapomorphy of this group.In the present study, this hypothesis is tested for the first time using multi-gene data.
Among Strombidiidae, species with an eyespot (Strombidium apolatum, S. rassoulzadegani, S. cf.parastylifer, S. guangdongensei, S. oculatum, S. purpureum and Williophrya maedai) grouped together in the SSU rDNA, LSU rDNA and concatenated gene trees, all with high support values.Of these species, only S. purpureum lacks detailed in vivo information, while the rest are all known to possess an eyespot (Liu, 2011;Song et al., 2009;Song et al., 2015).Moreover, the eyespot is the only unique morphological character shared among all members of this clade, which can also explain why Williophrya clustered within it.It is noteworthy that tontoniid species possessing an eyespot are excluded from this clade suggesting that the eye-spot evolved more than once within the oligotrichs.

The close relationship between Apostrombidium and Varistrombidium: updating the evolutionary hypothesis based on oligotrich ciliary patterns
In this study, three isolates of Varistrombidium kielum were included: pop 1 (sampled from Qingdao, northern of China) (Xu et al., 2011), pop 2 (sampled from Zhanjiang, southern of China; Table 1), and pop 3 (sampled from Qingdao, present study).Although they share the same morphological features, the SSU rDNA sequence similarity between pop2 and pop3 is 98.4% while pop1 is even more divergent with a similarity as low as 96.7%.The sequence identities of ITS-5.8S and LSU rDNA cannot currently be calculated due to lack of data.The three populations of V. kielum clustered with different lineages in the phylogenetic tree based on SSU rDNA (Fig. 2), pop 1 forming a polytomy with other oligotrichs, and pop 2 and pop 3 clustering with Apostrombidium parakielum (0.95 BI, 42% ML).The possibility that the three populations cluster together was also rejected by the AU test.This phenomenon may reveal the existence of cryptic species and high gene diversity within the oligotrichs (Snoeyenbos-West et al., 2002).
In the present study, multi-gene information for Varistrombidium and Apostrombidium is analyzed for the first time making it possible to conduct a further analysis of the evolutionary relationships among these three genera.In the trees based on SSU rDNA, LSU rDNA and concatenated genes, A. parakielum always groups with V. kielum pop 2 and pop 3 with high support values.Additionally, the ITS1-5.8SrDNA-ITS2 sequence similarities between Omegastrombidium and Apostrombidium are 83.4%-84.6%;and those between Varistrombidium and Omegastrombidium are 84.8%-85.7%.These values are sigificanlty lower than the ITS1-5.8SrDNA-ITS2 sequence similarity between Apostrombidium and Varistrombidium (92.0%).Furthermore, the groupings of Varistrombidium with Omegastrombidium and Apostrombidium with Omegastrombidium are both also rejected by the AU test.

Fig. 4 .
Fig. 4. BI tree of Oligotrichia based on LSU rDNA sequences of 1719 sites from 49 taxa.Numbers at

Table 1
Sampling information for newly sequenced oligotrichs in this study (accession numbers in bold are newly sequenced gene fragments in the present study).

Table 2
Approximately Unbiased test results based on the concatenated data.Rejected monophyly (p < 0.05) is highlighted in gray.