Pseudotrichonympha leei, Pseudotrichonympha lifesoni, and Pseudotrichonympha pearti, new species of parabasalian flagellates and the description of a rotating subcellular structure

Pseudotrichonympha is a large and structurally complex genus of parabasalian protists that play a key role in the digestion of lignocellulose in the termite hindgut. Like many termite symbionts, it has a conspicuous body plan that makes genus-level identification relatively easy, but species-level diversity of Pseudotrichonympha is understudied. Molecular surveys have suggested the diversity is much greater than the current number of described species, and that many “species” described in multiple hosts are in fact different, but gene sequences from formally described species remain a rarity. Here we describe three new species from Coptotermes and Prorhinotermes hosts, including small subunit ribosomal RNA (SSU rRNA) sequences from single cells. Based on host identification by morphology and DNA barcoding, as well as the morphology and phylogenetic position of each symbiont, all three represent new Pseudotrichonympha species: P. leei, P. lifesoni, and P. pearti. Pseudotrichonympha leei and P. lifesoni, both from Coptotermes, are closely related to other Coptotermes symbionts including the type species, P. hertwigi. Pseudotrichonympha pearti is the outlier of the trio, more distantly related to P. leei and P. lifesoni than they are to one another, and contains unique features, including an unusual rotating intracellular structure of unknown function.

or tens of thousands of flagella, typically organized in longitudinal rows that extend over most of the cell body and emerge from an anterior organizing centre beneath an apical cap. Other hypermastigotes can be found in the class Cristamonadea, which typically have multiple karyomastigonts rather than a single nucleus (with exceptions such as Deltotrichonympha and Kofoidia), and in the class Spirotrichonymphea, where flagellar bands are organized in a spiral array consisting of multiple right-handed helices.
The overall body plan of Pseudotrichonympha matches the main characteristics of Trichonymphida as a whole, and is specifically distinguished by a robust apical cap, a sub-apical rostrum with an anterior band of short flagella and a posterior band of distinctively long flagella, and a main body that is typically elongated and covered with shorter flagella organized in longitudinal rows (which may have a slight spiral on close inspection). Pseudotrichonympha occurs in many different termite host species, but its diversity and taxonomy are both understudied, and a long history of mistaken identities has led to too much contradictions and confusion in its classification. The genus was first described over a century ago, but was initially and erroneously interpreted as one of the 'sexes' of Trichonympha 5 . This was corrected by the creation of the new genus, with the type species Pseudotrichonympha hertwigi 6 . The type species has since been re-examined by light and electron microscopy and its small subunit rRNA gene sequences (SSU rRNA) characterized for phylogenetic analysis 7 . But most of the diversity of Pseudotrichonympha remains unexamined. They are known to be common, perhaps ubiquitous 8 , in the rhinotermitids, a widely-distributed and speciose family of termites, but only about a dozen species have been formally described 7 . Their diversity is also evident from molecular surveys of termite hindgut communities, which have yielded a diverse clade of environmental sequences that are inferred to be from Pseudotrichonympha, but currently sequences from organisms lacking formal taxonomic description outnumber sequences from described species by a ratio of five to one 7 . Here, we describe three new species of Pseudotrichonympha from barcoded Coptotermes and Prorhinotermes hosts from North America and Australia, using single cell isolation to provide molecular data from the SSU rRNA for phylogenetic analysis so as to further fill the spaces between identified species and environmental sequences in the Pseudotrichonympha tree.

Results and Discussion
Host collection and identification. Prorhinotermes simplex and Coptotermes gestroi were collected in Fort Lauderdale, Florida, USA, and their morphological identification was confirmed by DNA barcoding (accessions JX975355 and MF373426, respectively). A second species of Coptotermes was collected in Mount Glorious, Queensland, Australia. It was identified to the genus level by morphology in the field. Ethanol-preserved samples were returned to the lab for DNA barcoding using the mitochondrial large subunit rRNA gene (accession KJ438378), which allowed for comparisons with two of the three species of Coptotermes known from the area. Coptotermes frenchi barcodes shared on average 95.5% identity with the Australian barcode and showed no close affinity in the phylogenetic tree (Fig. 1, Suppl. Figure 1). The identity shared with Coptotermes lacteus barcodes was on average 96.7%, and in phylogenetic trees the Australian barcode branched closer to C. lacteus, but did not Coptotermes. All available Coptotermes barcodes are included. The new specimen from Florida is confirmed to correspond to C. gestroi, while the new specimen from Australia is a species for which no barcode is currently available. Based on the available barcodes to which it does not correspond and the diversity of Coptotermes from this region of Queensland, we infer this is most likely a specimen of C. acinaciformis, and therefore refer to this as Coptotermes cf. acinaciformis. Conspecific clades are collapsed and represented by triangles. Numbers at nodes correspond to ML bootstrap support over 50% (values for nodes with lower support are not shown for clarity), and the scale bar represents a distance of 0.06 substitutions per site. The complete tree with outgroups can be seen in Suppl. Figure 1. branch within the clade of C. lacteus barcode isolates (which share 99-100% identity). Based on this, we conclude that the collection corresponds to neither C. lacteus nor C. frenchi. The only other Coptotermes known in this region is Coptotermes acinaciformis, but unfortunately there is no barcoded vouchered specimen of this species (Coptotermes are notorious difficult to identify). Therefore, we conclude that the probable identity of our isolate is C. acinaciformis, but it could also be a presently unknown species. We will henceforth refer to this specimen as Coptotermes cf. acinaciformis.

Morphology of new Pseudotrichonympha species.
In all three termite species we observed and photographed cells matching the description of Pseudotrichonympha (Fig. 2). Specifically, all three symbionts were characterized by an apical cap with an adjacent apical band of short rostral flagella, a distinctive posterior band of very long rostral flagella, and intermediate-length flagella covering the remainder of the post-rostral body. Cells contained a single nucleus central in the sub-rostral area (e.g. see Fig. 2E). The Pseudotrichonympha from Coptotermes cf. acinaciformis (Fig. 2F) was very large (length average: 392 µm, length range: 332-448 µm, width average: 100 µm,width range: 94-108, n = 5) and had a wide ovoid shape and a sharply tapering posterior tip. The Pseudotrichonympha from C. gestroi ( Fig. 2A-C) had a more slender body (length average: 303 µm, length range: 280-330 µm, width average: 62 µm,width range: 56-71 µm, n = 5), and was distinguished by the tendency to possess a robust sub-apical shoulder region that frequently had the greatest diameter of the body (Fig. 2B). In stressed cells (e.g. during isolation), this region expanded greatly and produced a distinctive cup-shaped cell with the apex at the centre of the cup (Fig. 2C). The Pseudotrichonympha from P. simplex (Fig. 2D,E) tended to have a wide body around the midline (length average: 262 µm, length range: 200-316 µm, width average: 122 µm, width range: 89-180 µm, n = 12), and was most obviously defined by the presence of a strange intracellular body we dubbed the "rotatosome".
The rotatosome (see Fig. 3, and Supplemental video files) is a spherical body on average 13 µm in diameter (n = 3) that was only ever observed in a single copy per cell. It could only be observed in some focal planes, so whether it is in all cells or not is not known. It continuously rotated, causing visible turbulence in the cytoplasm around it. The lumen of the rotatosome was not uniform. At certain observational planes the rotatosome could be seen to include a rod 6-7 µm in length that emerged from one side of the circle (which side depended on the angle from which it was viewed), was stationary with respect to the cytoplasm, and in which the material appears to stream towards the outermost layer of the sphere, where it can then be seen to turn and turn again in the direction of the rotation. We examined the possibility that the rotatosome is an endosymbiont living within Pseudotrichonympha by Hoechst staining. In fixed and stained cells, the nucleus consistently showed a strong signal for the presence of DNA (in the pattern expected for a parabasalian, which have condensed interphase chromosomes), but the rotatosome retained no detectible stain (Fig. 3C,D). Similarly, we manually broke three Pseudotrichonympha cells open under the microscope to see if a rotatosome could be observed swimming freely, but the body consistently disappeared when cells were disrupted (not shown). A spherical structure was previously observed, in 1923, in the cytoplasm of a species identified as P. sphaerophora from a termite identified as Rhinotermes nasutus in British Guiana 9 . In this case, the sphere was reported to be larger (25 µm), consisting of a non-staining centre covered by a lighter layer in a concentric organization. The structured was reported from a fixed and stained sample, so no rotation was observed, and it was hypothesized at the time to function as a stercoma for storing excretory material. Based on the difference in size, lack of detail in the 1923 description, and the fact this is apparently not the same species of host or symbiont, or the same collection location, it would be premature to conclude these are the same structure. Overall, in either case, many questions remain about the function and the form of the rotatosome that will hopefully be resolved in the fullness of time, but it appears to be a sub-cellular structure rather than an endosymbiont.

Molecular phylogeny.
To determine how these Pseudotrichonympha species relate to other members of the genus, we sequenced the SSU rRNA gene from single isolated cells and small pools of cells. In all cases, sequences from the symbionts of a given host were 98.5-99% identical and one representative sequence per taxon was used for phylogenetic analyses and submitted to GenBank under accessions MF373810 -MF373812. In phylogenetic trees, the Pseudotrichonympha symbionts from Coptotermes hosts branch within a single clade, albeit with no support (Fig. 4). However, within this clade are two well-supported subgroups, one of these consists of three sequences from Southeast Asian specimens of Coptotermes (and includes one known species, P. grassii from C. formosanus), and the other consists of P. hertwigi from its type host, C. testaceus, and the two Coptotermes symbionts described here. The P. simplex symbiont is excluded from this clade, but its position is otherwise unresolved and the real relation to the underlying tree of currently sampled Pseudotrichonympha sequences unclear (Fig. 4).
Taxonomic considerations. All three specimens of Pseudotrichonympha characterized here represent new species. In the case of C. gestroi, flagellates have been reported 10 but none, including members of the genus Pseudotrichonympha, have been formally described. In the case of P. simplex, several spirotrichonymphids and the smaller flagellate, Cthulhu, have been described 11 , but although Pseudotrichonympha has been reported to be present, it was never formally described 10-12 . In the case of the Australian Coptotermes cf. acinaciformis, we can conclude the symbiont is undescribed without knowing the host species with absolute certainty. Of the three Coptotermes known to exist in the region, one (C. frenchi) has no described Pseudotrichonympha, and the other two (C. lacteus and C. acinaciformis) are both documented as containing "Pseudotrichonympha hertwigi" 10,13 . However, this host barcode excludes both C. frenchi and C. lacteus, and moreover the confusing history of P. hertwigi is inconsistent with the presence of this species in any of these hosts (reviewed in detail in ref. 7 ). Since its original description and subsequent re-classificiation to a new genus 5,6,14 , several authors have assumed that Pseudotrichonympha observed in other Coptotermes hosts correspond to the same species, even in geographically distant Australian and Asian termites 10,12,13 . But this was invariably based only on light microscopy and limited numbers of distinguishable characters. The identity of the host species is also convoluted. The Coptotermes host of P. hertwigi was named Coptotermes hartmanni 15 , but it lacked formal description and is now invalid 16 . Subsequent surveys of neotropical Coptotermes have shown all Brazilian specimens correspond to Coptotermes testaceus: this host has now been barcoded and the P. hertwigi SSU rRNA from this type host has been characterized 7 . This species does not exist in Australia, and neither the Pseudotrichonympha from the Australian Coptotermes nor the host barcodes match their Brazilian counterparts, leading to the conclusion that neither Australian C. lacteus nor C. acinaciformis (nor any other Australian termite) contain P. hertwigi, but rather contain undescribed species that have been misidentified. The same is also likely true for Asian termites concluded by light microscopy alone to harbour P. hertwigi. Based on this and the phylogenetic and morphological data, we propose three new species of Pseudotrichonympha as detailed below. Prorhinotermes simplex and Coptotermes gestroi were maintained in conical tubes with wood from their habitats at room temperature in the laboratory. Coptotermes cf. acinaciformis was processed directly in the field. Termite identities were determined morphologically and by barcoding using the mitochondrial 16 S (LSU) ribosomal RNA gene amplified and sequenced using the primers LR-N-13398 5′-CGC CTG TTT ATC AAA AAC AT-3′ 17 and LR-J-13017 5′-TTA CGC TGT TAT CCC TAA-3′ 18 . These were aligned with Coptotermes LSU termite barcodes from Genbank using using MAFFT v7.310 19 (setting: -auto). Poorly aligned regions were automatically removed with trimAl v1.4 using a gap threshold of 0.3 20 (settings: -gt 0.3 -st 0.001); columns with missing data at both ends were removed. For the termite phylogeny, maximum likelihood (ML) inference was carried out using RAxML v8.2.9 assuming the GTR + Γ substitution model with 1,000 random starting trees and statistical support obtained from 1,000 bootstrap replicates 21  Microscopy. Termites were dissected and hindgut contents were suspended in Trager's medium U 23 .
Symbionts from C. gestroi and P. simplex were observed on an Axioplan 2 compound microscope (Zeiss, Oberkochen, Germany) using differential interference contrast and documented with a 3CCD HD video camera XL H1S (Canon, Tokyo, Japan). Multiple cells of each new species were observed, filmed, and measured to estimate ranges of cell size and shape, but in all cases the size range was variable and many cells were observed to deviate from the norm, as these species are known to display significant plasticity. For this reason, we also carried out molecular characterization using single cell isolations of these symbionts, using an Axiovert 200 (Zeiss, Oberkochen, Germany) inverted microscope. Isolated cells were photographed with a MicroImager II (QImaging, Surrey, BC, Canada). For Coptotermes cf. acinaciformis, microscopy was carried out at the sampling location using a Swift Field Master phase contrast field microscope (Swift Optical Instruments, Schertz, TX) and imaging using a Canon G9 camera (Canon, Tokyo, Japan). Whether the "rotatosome" structure was an endosymbiotic cell was tested by puncturing individual cells with a small-diameter micro-pipette. We observed that with a certain amount of force the cell membrane could be disrupted without damaging comparably sized organelles, such as the nucleus.
For fluorescence microscopy, cells from P. simplex were fixed and stained in 2% paraformaldehyde in PBS buffer (pH 7.4) with 10 μg/mL Bisbenzimide H 33258 (Hoescht stain: Sigma, St Louis, MO) for 10 min. The cells were observed under an Axioplan2 fluorescent microscope and filmed with the Canon XL H1S. Bisbenzimide fluorescence were detected with the UV filter (excitation BP 365/12 nm, emission LP 397 nm).
Single cell isolation, sequencing, and phylogenetic analysis. Individual Pseudotrichonympha cells were isolated by glass micropipette. Symbionts from P. simplex and C. gestroi were isolated in the lab using a Zeiss Axiovert, while individual Pseudotrichonympha cells from Coptotermes cf. acinaciformis were isolated directly in the field using the Swift Field Master. Representatives of this species were fixed in 95% ethanol with a view towards its preservation and shipped to the University of British Columbia for all subsequent molecular analyses. For all three species, single cells or small pools of cells were manually picked and rinsed three times under constant observation. In cases where the cell was so close to the edge of the cavity slide as to obscure the view, they were washed an additional time and inspected to confirm its state of integrity was not compromised. DNA was extracted from isolated cells using the Masterpure Complete DNA and RNA Purification Kit (Epicentre, Madison, WI, USA). SSU rRNA genes were amplified from purified DNA using the eukaryote specific primers PFI 5′-TGC GCT ACC TGG TTG ATC CTG CC-3′ and FAD4 5′-TGA TCC TTC TGC AGG TTC ACC TAC-3′. PCR conditions included a 3-minute denaturation at 95 °C followed by 30 cycles of 95 °C for 30 seconds, 55 °C for 30 seconds, and 72 °C for 1 minute 30 seconds, then an additional 7 minutes at 72 °C. Products were purified, cloned into the pCR2.1 vector using the TOPO-TA cloning kit (Invitrogen, Carlsbad, CA, USA), and sequenced on both strands with BigDye Terminator v 3.1 (Applied Biosystems, Carlsbad, CA, USA). Multiple clones were sequenced from each isolation. Sequences were submitted to GenBank under accessions MF373810 -MF373812.
New Pseudotrichonympha sequences were aligned with previously published sequences spanning the phylogenetic diversity of parabasalians using MAFFT with default settings 19 . Highly variable regions were removed using trimAl 20 (settings: -gt 0.3 -st 0.001). Preliminary analyses confirmed with complete support that all new sequences corresponded to Pseudotrichonympha, and a more detailed analysis focused exclusively on sequences from this genus and outgroups, resulting in a final alignment of 28 taxa and 1,388 positions. ML analyses were performed with RAxML v8.2.9 21 , using the GTR + Γmodel and 1,000 random starting trees. For the ML analysis, support was assessed from 1,000 bootstrap replicates.