Single-Cell DNA Barcoding Using Sequences from the Small Subunit rRNA and Internal Transcribed Spacer Region Identifies New Species of Trichonympha and Trichomitopsis from the Hindgut of the Termite Zootermopsis angusticollis

To aid in their digestion of wood, lower termites are known to harbour a diverse community of prokaryotes as well as parabasalid and oxymonad protist symbionts. One of the best-studied lower termite gut communities is that of Zootermopsis angusticollis which has been known for almost 100 years to possess 3 species of Trichonympha (T. campanula, T. collaris, and T. sphaerica), 1 species of Trichomitopsis (T. termopsidis), as well as smaller flagellates. We have re-assessed this community by sequencing the small subunit (SSU) rRNA gene and the internal transcribed spacer (ITS) region from a large number of single Trichonympha and Trichomitopsis cells for which morphology was also documented. Based on phylogenetic clustering and sequence divergence, we identify 3 new species: Trichonympha postcylindrica, Trichomitopsis minor, and Trichomitopsis parvus spp. nov. Once identified by sequencing, the morphology of the isolated cells for all 3 new species was re-examined and found to be distinct from the previously described species: Trichonympha postcylindrica can be morphologically distinguished from the other Trichonympha species by an extension on its posterior end, whereas Trichomitopsis minor and T. parvus are smaller than T. termopsidis but similar in size to each other and cannot be distinguished based on morphology using light microscopy. Given that Z. angusticollis has one of the best characterized hindgut communities, the near doubling of the number of the largest and most easily identifiable symbiont species suggests that the diversity of hindgut symbionts is substantially underestimated in other termites as well. Accurate descriptions of the diversity of these microbial communities are essential for understanding hindgut ecology and disentangling the interactions among the symbionts, and molecular barcoding should be a priority for these systems.


Introduction
Termites harbour a diverse community of microbial symbionts in their hindguts, many of which aid the digestion of lignocellulose. The ''lower'' termites (Mastotermitidae, Kalotermitidae, Termopsidae, Hodotermitidae, Serritermitidae, and Rhinotermitidae) host both protist and prokaryotic symbionts, while the ''higher'' termites (Termitidae) host only prokaryotes [1]. The protists found in lower termites are primarily Parabasalia and Oxymonadida [2] and most of the known diversity of both lineages resides in the hindguts of termites. The composition of the hindgut community is generally species-specific: each termite species has its own set of symbionts, but related termites share a similar community of related symbionts, altogether indicating some degree of co-evolution between symbionts and termite hosts [3].
The identification and classification of termite hindgut symbionts has a history extending over more than 100 years and has largely been based on morphological criteria. Molecular characterization has been applied slowly, in part because almost none of the symbionts have been brought into culture. Molecular sequencing of the small subunit (SSU) rRNA gene from manually isolated cells has only recently begun to substantially supplement the morphological characterization, but the resulting molecular phylogenetic analyses have already contributed a great deal to clarifying taxonomy, in particular with respect to parabasalians [4][5][6][7][8][9][10][11]. Sequencing of SSU rRNA genes from the hindgut community followed by fluorescent in situ hybridization has also linked molecular data to the morphologically described species [12][13][14].
In general, molecular data have been used to test hypotheses concerning the evolution of established species [15][16][17][18][19] or to formally describe new symbiont species from termites that have not been investigated previously using classical criteria [20,21]. Another question of equal importance, however, is whether the classical morphology-based descriptions of termite hindgut communities can be validated using molecular markers. Despite the obvious utility of molecular data to test the defined compositions of these communities, they have seldom been tested specifically (but see Strassert et al. 2009). This is unfortunate, because an accurate description of a hindgut community is a basic first step to understanding its ecology, the interactions between the biota, the evolution of hindgut symbionts, and the factors influencing community composition.
Here we specifically test the seemingly well-known composition of the protist community in the hindgut of the Pacific Dampwood termite, Zootermopsis angusticollis. This community has been studied for nearly 100 years, and since the early 1900 s it has been documented to contain seven species of protist: the parabasalians Hexamastix termopsidis, Tricercomitus termopsidis, Trichomitopsis termopsidis, Trichonympha campanula, Trichonympha collaris, and Trichonympha sphaerica, and the oxymonad Streblomastix strix [22][23][24][25][26][27][28][29]. These same protist species are also found in the hindgut of Z. nevadensis, the closest relative of Z. angusticollis. This community has been revisited in a variety of studies, including some of the earliest molecular characterization studies [4,6,30] and metatranscriptomic analyses of protist hindgut symbionts [31,32], making it arguably the best-studied community of any lower termite.
By characterizing sequences from the SSU rRNA gene and internal transcribed spacer (ITS) region from over 50 manually isolated cells of the largest symbionts from the Z. angusticollis hindgut, we find that even the species-level diversity of this wellstudied community has been substantially underestimated. Rather than three species of Trichonympha we find there are four, and rather than one species of Trichomitopsis we find there are three, almost doubling the number of large and most easily identifiable species. Interestingly, all three new species found by molecular characterization also correlate with morphological variation. This expansion of characterized diversity of the largest symbionts within such a well-studied host termite suggests that parabasalian symbiont diversity as a whole may be even more significantly underestimated and a detailed characterization of symbiont diversity at the molecular level should be a first step in tackling hindgut ecology and symbiont interactions in any other model termite.

Results and Discussion
To test whether our long-established understanding of the hindgut community composition of Z. angusticollis is correct, 77 single cells representing distinct morphotypes of Trichonympha and Trichomitopsis were manually isolated, photographed, and characterized by DNA sequencing. The termites from which these hindgut symbionts were isolated were confirmed to be Z. angusticollis as their mitochondrial cytochrome oxidase I (COI) gene sequences (GenBank accession KC136610 and KC136611) were identical to those previously sequenced from Z. angusticollis [33].
Diversity of Trichonympha in Z. angusticollis 52 single cells matching the overall description of Trichonympha were manually isolated from Z. angusticollis. After purification of their DNA and PCR amplification, nearly 1500 bp of the SSU rRNA gene from 42 single cells were successfully sequenced. In phylogenetic analyses including these new sequences together with existing Trichonympha homologues, the majority of the new sequences clustered with T. campanula (clones from ten cells), T. collaris (clones from eight cells), and T. sphaerica (clones from eight cells) sequences previously characterized from the closely related host, Z. nevadensis ( Figure 1). The morphology of the cells from which these clones were derived also matched the expected morphology of these species (Figure 1), and no cell yielded clones that fell into different clusters, as expected. However, sequences from 16 isolated cells formed a fourth lineage as distinct from other Zootermopsis Trichonympha species as they are from one another ( Figure 1). Within this cluster, the mean pairwise identity (6 standard deviation) was 99.760.2%, whereas the level of similarity between sequences from the new cluster and those of its nearest neighbor (T. sphaerica) fell to only 97.060.2% identity (Figure 2A). These values are similar to within-and between-species pairwise comparisons for the established species T. campanula and T. sphaerica. SSU sequences from T. collaris exhibited the greatest within-species diversity. Most of the sequences in the T. collaris cluster were nearly identical to the Z. nevadensis T. collaris, but four sequences formed long-branches that lowered the average withinspecies similarity to 98.261.4% (Figure 2A). Interestingly, one of these was a T. collaris sequence previously characterized from Z. angusticollis (AF023622) [4] which was not highly similar to any characterized here.
The coherence of this new cluster was also tested by analyzing the ITS region, which is typically a much more divergent marker. There is little ITS data available for parabasalians and these data are the first for the Trichonymphea. The ITS region was successfully amplified from 19 of the Z. angusticollis Trichonympha cells from which SSU rRNA was sequenced. Once again, the phylogenetic analyses of the ITS sequences showed four distinct clusters with the individual cells corresponding exactly to the SSU rRNA clusters ( Figure 2B). The mean within-cluster similarity was over 95.9% for all four clusters, whereas the mean between-cluster similarity was 88.0% or less (Figure 2A). T. collaris again shows a higher within-species diversity than the others due to a couple of divergent sequences (one from the same cell as a less divergent copy). It is common for protist cells to have divergent rRNA gene sequences as their genomes contain multiple copies of the rRNA operon, but the range of diversity within a species or in this case within a cell is an important consideration as molecular barcodes to identify protist species and other sequence-based diversity estimates are more widely applied [34,35].
The branching order of the four Trichonympha clusters was different in the SSU and ITS trees, but neither topology was rejected based on an approximately unbiased (AU) test of all possible topologies (Table S1). The order in which these species originated may be difficult to determine if the multiple species of Trichonympha radiated in a relatively short period of time.
As stated above, the morphology of the cells from which T. campanula, T. collaris, and T. sphaerica sequences were characterized corresponded to the known morphology of these symbiont species ( Figure 1, Figure 3D-F). We therefore examined the morphology of the 16 isolated cells from which all sequences falling into the fourth cluster were derived. These were similar in size and morphology to T. campanula averaging 180 mm667 mm with an average length to width ratio of 2.7. These cells also consistently presented an extension at the posterior end ( Figure 1, Figure 3A-C) that is not formally a defining character of T. campanula or any other Trichonympha species in Zootermopsis. Interestingly, in the formal description of T. campanula, Kofoid and Swezy remarked on the variation in morphology of the posterior end of Trichonympha cells [24], even describing cylindrical extensions on the posterior end that are identical to the ones we observed, but they did not consider this morphotype to be a distinct species. Based on this

Diversity of Trichomitopsis in Z. angusticollis
The second large and distinctive genus in Zootermopsis is the trichomonad Trichomitopsis. Kofoid and Swezy [23] noted size variation in Trichomitopsis termopsidis ranging from 16 to over 200 mm in length, but did not consider that multiple species comprise this morphotype. It is thought that the very large cells were about to go through multiple fission events resulting in many smaller cells [23]. We did not observe the extremes of this size range, but did observe two distinct sizes of Trichomitopsis in the hindguts of Z. angusticollis ( Figure 4). The larger cells were approximately 50 mm in diameter, whereas the smaller cells were approximately 25 mm in diameter. Five larger and 20 smaller cells were collected individually and the SSU rRNA was sequenced successfully from 3 larger and 10 smaller single cells. Of these, ITS sequence data was also obtained from 3 and 7 cells, respectively.
Phylogenetic analyses of the Trichomitopsis SSU rRNA gene sequences resulted not in a single lineage, as expected, but rather three distinct lineages ( Figure 5). All sequences from the larger morphotype shared at least 99.5% identity with the existing T. termopsidis sequence, which was characterized from isolated cells at the larger end of the spectrum reported [6]. Sequences from the smaller morphotype resulted in two distinct lineages, each with greater than 99% within-cluster mean pairwise similarity, and sharing less than 98.5% identity with other clusters ( Figure 6A). Once again, this result was confirmed by analyses of ITS sequences, where the same three clusters were found and the same cells shown to correlate with each cluster ( Figure 6B). In this case, the within-cluster sequences share on average greater than 98% similarity, whereas the mean between-cluster similarities were less than 93% ( Figure 6A). Based on their morphological differences from T. termopsidis and the two distinct clusters that consistently form in analyses of molecular diversity from single cells, we have named these two species Trichomitopsis parvus sp. nov. and Trichomitopsis minor sp. nov. (see Taxonomic Synopsis below).

Trichonympha and Trichomitopsis speciation
The new species of Trichonympha, T. postcylindrica, and the 3 previously described species form a monophyletic cluster exclusive of other Trichonympha symbionts from non-Zootermopsis termites. Therefore, the Trichonympha species in Zootermopsis likely diversified from a common ancestor within the hindgut after the divergence of the Zootermopsis lineage. Archotermopsis, the closest relative of Zootermopsis, does not host Trichonympha species in their hindguts, but A. wroughtoni does harbour a related symbiont -Protrichonympha pristina [2,36]. Based on the available SSU data, the closest relatives of the Trichonympha species from Zootermopsis are Trichonympha from Hodotermopsis and Reticulitermes (Figure 1).
Similarly, the multiple Trichomitopsis species form a monophyletic group and also likely originated within the Zootermopsis lineage. Currently, a single species, T. termopsidis, is thought to commonly occur in the hindguts of all 3 species of Zootermopsis. However, a closer examination of the hindguts of Z. nevadensis and Z. laticeps may also reveal multiple species comprising the Trichomitopsis morphotype. T. termitis from the hindgut of A. wroughtoni [36,37], T. barbouri in Glyptotermes angustus [38], and T. cartagoensis in G. contracticornis [38] are the closest known relatives of the Trichomitopsis from Zootermopsis, but molecular data are not available for these species so their phylogenetic relationships are not known. The closest relative to Trichomitopsis for which there are molecular data is Pseudotrypanosoma giganteum from the hindgut of Porotermes adamsoni, a relative of Zootermopsis, and this species groups outside of the Trichomitopsis cluster ( Figure 5).
In addition to examining species-level diversity, ITS data should provide a more variable marker to assess population level differences between Trichonympha and Trichomitopsis symbionts in Z. angusticollis and Z. nevadensis. In SSU trees, the Z. nevadensis symbiont sequences cluster with the corresponding Z. angusticollis symbiont species suggesting they are the same or closely related cospeciated symbionts. With more extensive sampling of Zootermopsis populations, ITS data may reveal genetic distinctions between the symbiont populations of Z. angusticollis and Z. nevadensis, and even between the subspecies Z. nevadensis subsp. nuttingi and Z. nevadensis subsp. nevadensis which have distinct geographic distributions. Given the close similarity between Z. angusticollis and Z. nevadensis, the correct identification of the host is important for symbiont identification and it will be desirable to verify the identity of the host using DNA barcodes. Ultimately, the symbionts of Z. angusticollis and Z. nevadensis provide an opportunity to examine the rate of symbiont diversification relative to that of their hosts and possible mechanisms of speciation.
Undoubtedly, better molecular sampling of Trichonympha and Trichomitopsis species and their relatives would clarify the evolutionary relationships of these symbionts with each other and with their hosts. Nevertheless, the monophyly of the Trichonympha and Trichomitopsis species from Zootermopsis hosts indicates that these symbionts have almost certainly diversified into multiple species within the environment of the Zootermopsis hindgut. This diversification is likely an example of sympatric speciation, but it is not known what factors have lead to the diversification of Trichonympha or Trichomitopsis. These species appear to be similarly distributed throughout the hindgut and comparative physiological studies do not exist. The hindgut environment, however, is not uniform and adaptation to microniches due to differences in oxygen or hydrogen concentrations is a possibility [39]. For Trichonympha, another possibility is their association with ecto-and endobacterial symbionts that may facilitate the separation of ecological niches within the hindgut by performing distinctive biochemical functions such as providing amino acids or other nitrogenous compounds [17,40,41].

Conclusions
Accurate descriptions of diversity are essential for understanding the ecology and evolution of any biological community. Especially for microbial communities, these descriptions are lacking because most microbes cannot be easily cultivated and descriptions based solely on morphology greatly underestimate the genetic diversity. For nearly 100 years, the hindgut of Z. angusticollis was known to harbour 3 species of Trichonympha and 1 species of Trichomitopsis, but even for these largest symbionts, we show that species diversity has been underestimated. We used single cell isolations and DNA sequencing of molecular markers alongside morphological obser-the next most closely related SSU rRNA sequences available: T. agilis from Reticulitermes speratus = AB003920, T. sp. from Hodotermopsis sjoestedti = AB326373, T. sp. from Reticulitermes santonensis = AB434787, T. tabogae from Incisitermes tabogae = AB434793. The best ML tree is shown. Numbers at nodes indicate ML bootstrap support and Bayesian posterior probability values. Statistical support is shown only for nodes with .70% bootstrap support and .0.90 posterior probability. doi:10.1371/journal.pone.0058728.g001 Figure 2. Comparing SSU and ITS barcodes from manually isolated Trichonympha cells from Zootermopsis angusticollis. A) Pairwise similarity matrix for DNA sequences from Trichonympha species. The upper and lower values are for SSU and ITS sequences, respectively. The mean pairwise similarities (6 standard deviation) for comparisons of sequences within and between species are reported. B) Phylogenetic tree of single-cell ITS barcodes from Trichonympha isolated from the hindgut of Zootermopsis angusticollis. Sequences from Hexamastix mitis and Monocercomonas colubrorum were included to root the tree. The best ML tree is shown. Numbers at nodes indicate ML bootstrap support and Bayesian posterior probability values. Statistical support is shown only for nodes with .70% bootstrap support and .0.90 posterior probability. Statistical support for the T. campanula node (in italics) is also shown which does not satisfy the above criteria. doi:10.1371/journal.pone.0058728.g002 vations and discovered a new Trichonympha species, T. postcylindrica sp. nov. and two species of Trichomitopsis, T. minor sp. nov. and T. parvus sp. nov. in the hindgut of Z. angusticollis. This expansion of the known diversity in the hindgut of Z. angusticollis provides necessary knowledge to better understand the ecological interactions, diversification, evolution, and host co-evolution of these symbiotic microorganisms.

Taxonomic Synopsis
Trichonympha postcylindrica Tai and Keeling, sp. nov.   Diagnosis. Symbiont from the hindgut of Zootermopsis angusticollis ranging from 12 to 37 mm in diameter (averaging 25 mm) with a recurrent flagellum forming an undulating membrane and a protruding axostyle. Distinct from Trichomitopsis termopsidis based on its smaller size. Distinct from other Trichomitopsis based on SSU rRNA gene and ITS region sequences.
Hapantotype. Mounted slide deposited at the Beaty Biodiversity Museum, University of British Columbia, Vancouver, Canada under the accession number MI-PR202.
Etymology. Small in size.
Type host. Zootermopsis angusticollis. Diagnosis. Symbiont from the hindgut of Zootermopsis angusticollis ranging from 13 to 30 mm (averaging 25 mm) in diameter with a recurrent flagellum forming an undulating membrane and a protruding axostyle. Distinct from Trichomitopsis termopsidis based on its smaller size. Distinct from other Trichomitopsis based on SSUrRNA and ITS sequences.
Hapantotype. Mounted slide deposited at the Beaty Biodiversity Museum, University of British Columbia, Vancouver, Canada under the accession number MI-PR202.
Etymology. Small in size.

Single cell isolation
Late instar nymphs from Z. angusticollis colonies were collected from decaying logs in Pacific Spirit Park, adjacent to the University of British Columbia campus, Vancouver, Canada. The samples were collected under the Metro Vancouver Regional Parks Research Permit No. VTPAC2011. The identity of the termites was confirmed by obtaining DNA sequences from their mitochondrial COI gene as described in Booth et al. 2012 [33]. The hindgut was removed and the contents were resuspended in Trager's Medium U [42]. Using an inverted microscope and micromanipulation, single cells that were morphologically consistent with Trichonympha and Trichomitopsis species were transferred to fresh buffer 3 times, placed in a microcentrifuge tube, and stored at 220uC. Photographs were taken of all individually isolated cells.
DNA extraction, PCR amplification, and sequencing DNA from single cells was extracted using the MasterPure Complete DNA and RNA Purification kit (Epicentre) following the manufacturer's instructions except the extracted DNA was resuspended in 4.5 mL TE buffer. For all PCR reactions, a 25 mL reaction mix consisted of 5 pmoles each of the forward and reverse primers, 2 mL of DNA template, and 1X EconoTaq PLUS GREEN (Lucigen).
Nearly the entire SSU (18S) rRNA gene was amplified by PCR using the PF1 and FAD4 primers [21]. If this PCR did not amplify the SSU rRNA gene fragment sufficiently, a nested PCR was performed using 1 mL of the primary PCR as template and the primers GGF and GGR [21]. The primary and nested PCRs were incubated using the following thermal profile: 94uC for 2 min, 35 cycles of 94uC for 30 s, 50uC for 1 min, and 72uC for 2 min, and a final extension at 72uC for 10 min.
The ITS region was amplified using the forward primer ITSFpara (59-GTC CCT GCC CTT TGT ACA CAC C-39) modified from [43], and the reverse primer NC2 (59-TTA GTT TCT TTT CCT CCG CT-39) [44]. These primers anneal to the 39 end of the SSU rRNA gene and the 59 end of the large subunit (LSU, 28S) rRNA gene, respectively. The PCR conditions were: 94uC for 2 min, 35 cycles of 94uC for 30 s, 54uC for 30 s, and 72uC for 1 min, and a final extension at 72uC for 5 min.
For Trichonympha cells, a larger portion of the SSU rRNA gene was amplified with the ITS region using TrichoSSUmidF (59-CGA GAC TAC CGC CAA ATA-39) and NC2. TrichoSSUmidF is a Trichonympha-specific primer that anneals in the middle of the SSU rRNA gene approximately 400 bp away from the 39 end. This larger DNA fragment was amplified in order to have sufficient SSU rRNA sequence data to distinguish the Trichonympha species. The PCRs were incubated at 94uC for 2 min, followed by 35 cycles of 94uC for 30 s, 52uC for 1 min, and 72uC for 1 min, and a final extension at 72uC for 10 min.
All PCR products were ligated into plasmid vectors and cloned using the Strataclone PCR cloning kit (Agilent Technologies) following the manufacturer's protocol. Plasmid DNA was extracted from positive clones using the FastPlasmid Mini Kit (5 Prime). For each single Trichonympha or Trichomitopsis cell, two clones of the SSU rRNA gene fragment and two clones of the ITS region were Sanger sequenced (NAPS facility, UBC) from both strands with the BigDye Terminator kit v. 3.1 (Applied BioSystems). In a few rare cases, only a single clone was sequenced. All of the sequences have been deposited in GenBank under accession numbers KC136612-KC136766.

Phylogeny
Separate alignments were used to calculate the phylogeny for the SSU rRNA sequences from Trichonympha and Trichomitopsis. Related parabasalid SSU rRNA sequences were obtained from GenBank and aligned to the Trichonympha or Trichomitopsis sequences using MAFFT [45] from an online server (http:// www.ebi.ac.uk/Tools/msa/mafft/) with the default settings ( = L-INS-i). The ends of the alignments were trimmed manually. Gblocks was used to remove highly variable and ambiguously aligned sites, but allowing smaller final blocks, gap positions, and less strict flanking positions (http://molevol.cmima.csic.es/ castresana/Gblocks_server.html) [46]. Phylogenetic trees were calculated from maximum likelihood (ML) analysis using RAxML 7.0.4 [47] and Bayesian analysis using MrBayes 3.2 [48]. The ML analyses implemented a general time reversible (GTR) model of nucleotide substitution with the gamma model of rate heterogeneity. Statistical support for the consensus tree was assessed from 1000 bootstrap replicates. The Bayesian analyses also used a GTR + gamma model. For the Trichonympha alignment, 4 chains were sampled every 100 generations from 2 runs for 1 500 000 generations. Diagnostics were run every 1000 generations with a relative burnin of 25% of the tree samples. After 1 500 000 generations, the average standard deviation of the split frequencies from the 2 runs was less than 0.01. The Trichomitopsis Bayesian analysis was run for 1 000 000 generations.
The ITS sequences from Trichonympha, Trichomitopsis, and representative parabasalid ITS sequences from GenBank were used to calculate phylogenetic trees using ML and Bayesian analyses as described above. The Bayesian analyses were run for 1 000 000 generations for both Trichonympha and Trichomitopsis ITS alignments.

Topology test
The approximately unbiased (AU) test was used to assess the confidence of all possible branching topologies in the phylogeny of the 4 distinct clusters ofTrichonympha from Z. angusticollis. Given 4 clusters of Trichonympha, there are 15 possible topologies for these clusters. For both SSU and ITS data, each of these topologies was generated by editing the branching order from the best ML tree in TreeView [49]. RAxML 7.0.4 [47] was used to generate per-site log likelihoods from the SSU and ITS alignments for each topology and CONSEL [50] was used to conduct the AU test.

Nomenclatural Acts
The electronic edition of this article conforms to the requirements of the amended International Code of Zoological Nomen-clature, and hence the new names contained herein are available under that Code from the electronic edition of this article. This published work and the nomenclatural acts it contains have been registered in ZooBank, the online registration system for the ICZN. The ZooBank LSIDs (Life Science Identifiers) can be resolved and the associated information viewed through any standard web browser by appending the LSID to the prefix ''http://zoobank.org/''. The LSID for this publication is: urn:lsid:zoobank.org:pub:EF0EDBC0-9C8F-4B2F-8B62-9C748691B5D3. The electronic edition of this work was published in a journal with an ISSN, and has been archived and is available from the following digital repositories: PubMed Central, LOCKSS.