- Research article
- Open Access
Molecular phylogeny of diplomonads and enteromonads based on SSU rRNA, alpha-tubulin and HSP90 genes: Implications for the evolutionary history of the double karyomastigont of diplomonads
BMC Evolutionary Biology volume 8, Article number: 205 (2008)
Fornicata is a relatively recently established group of protists that includes the diplokaryotic diplomonads (which have two similar nuclei per cell), and the monokaryotic enteromonads, retortamonads and Carpediemonas, with the more typical one nucleus per cell. The monophyly of the group was confirmed by molecular phylogenetic studies, but neither the internal phylogeny nor its position on the eukaryotic tree has been clearly resolved.
Here we have introduced data for three genes (SSU rRNA, α-tubulin and HSP90) with a wide taxonomic sampling of Fornicata, including ten isolates of enteromonads, representing the genera Trimitus and Enteromonas, and a new undescribed enteromonad genus. The diplomonad sequences formed two main clades in individual gene and combined gene analyses, with Giardia (and Octomitus) on one side of the basal divergence and Spironucleus, Hexamita and Trepomonas on the other. Contrary to earlier evolutionary scenarios, none of the studied enteromonads appeared basal to diplokaryotic diplomonads. Instead, the enteromonad isolates were all robustly situated within the second of the two diplomonad clades. Furthermore, our analyses suggested that enteromonads do not constitute a monophyletic group, and enteromonad monophyly was statistically rejected in 'approximately unbiased' tests of the combined gene data.
We suggest that all higher taxa intended to unite multiple enteromonad genera be abandoned, that Trimitus and Enteromonas be considered as part of Hexamitinae, and that the term 'enteromonads' be used in a strictly utilitarian sense. Our result suggests either that the diplokaryotic condition characteristic of diplomonads arose several times independently, or that the monokaryotic cell of enteromonads originated several times independently by secondary reduction from the diplokaryotic state. Both scenarios are evolutionarily complex. More comparative data on the similarity of the genomes of the two nuclei of diplomonads will be necessary to resolve which evolutionary scenario is more probable.
Diplomonads and their close relatives – enteromonads, retortamonads and Carpediemonas membranifera – are small flagellates that tend to be found in low-oxygen habitats. Recently they were classified within Fornicata (Metamonada, Excavata) . Very recently an additional member of Fornicata, Dysnectes brevis, was described . Most diplomonads, all described enteromonads, and all described retortamonads except one are endobionts or parasites of animals, with several causing serious and highly prevalent diseases in fish, domestic animals and man .
Diplomonads and their relatives have been interesting for students of the evolution of the eukaryotic cell for several reasons. Firstly, they lack classical mitochondria. Secondly, diplomonads and their relatives branch at the base of the eukaryotic tree in the majority of phylogenies in which eukaryotes are rooted using prokaryotic outgroups [4–6]. Thirdly, diplomonads, but not their relatives, possess a double karyomastigont, in other words, they have two similar or identical nuclei and two flagellar apparatuses per cell [7, 8]. In the late 1980s and early 1990s it was widely supposed that the last common ancestor of most or all living eukaryotes was a 'fornicate-like' amitochondriate organism [4, 9], and some models even proposed that almost all living eukaryotes were descended from ancestors with a double karyomastigont . The best-studied diplomonad – Giardia intestinalis (= G. lamblia) was looked to as a model for understanding early eukaryotic cells.
Recent studies, however, have shown that Fornicata are secondarily amitochondriate [11, 12], and that many, perhaps all, retain mitochondrion-related organelles. Small double membrane-bounded organelles called 'mitosomes' were identified in the diplomonad Giardia intestinalis using antibodies against IscS and IscU (nucleus encoded proteins targeted to the mitochondrion of eukaryotes) and the function of this organelle in iron-sulfur cluster synthesis was demonstrated in vitro . Using electron microscopy, a hydrogenosome-like double membrane-bounded organelle was observed superficially in Carpediemonas membranifera . Furthermore, the position of Fornicata at the base of rooted eukaryotic trees is now thought by many to be caused by long-branch attraction [15–17]. Some authors, however, have argued that this basal position might nonetheless still be correct .
Diplomonads and enteromonads deserve special attention within Fornicata. These two groups were considered as closely related on the basis of ultrastructural studies [3, 7, 8] and their close affinity was confirmed recently by molecular phylogenetic methods . The morphology of diplomonads is extremely similar to the morphology of enteromonads – the main character distinguishing these two groups is the doubled karyomastigont of diplomonads [3, 7]. In a very simplified way, the cell of diplomonads could be described as two enteromonad cells joined together (and conversely the enteromonad cell could be described as half of a diplomonad cell). The most straightforward scenario explaining the evolution of the doubled karyomastigont is that diplomonads arose from enteromonads in a single evolutionary event. Siddall, Hong and Desser  proposed a mechanism of double karyomastigont formation from a single karyomastigont ancestor by secondary karyokinesis (mitosis) and mastigont duplication after delay or arrest of cytokinesis (cell division), resulting in a cell with four karyomastigonts. This cell could then have divided into two cells, each with a doubled karyomastigont. However, some other authors consider as plausible the opposite scenario – secondary simplification from the double karyomastigont morphology of diplomonads to the single karyomastigont morphology of enteromonads [1, 20].
Our understanding of the internal phylogeny of Fornicata is based to a large extent on molecular phylogenetic studies, especially with the relatively recent addition of several important taxa to the small subunit ribosomal RNA gene database, namely retortamonads , Carpediemonas membranifera , Dysnectes brevis one clade of enteromonads , the diplomonad Octomitus sp.  and several new species from the diplomonad genera Spironucleus and Hexamita . The monophyly of Fornicata is very strongly supported by molecular phylogenetic studies and there is strong support for a position of Carpediemonas membranifera and Dysnectes brevis at the base of the Fornicata clade [1, 2, 22, 25]. Meanwhile molecular phylogenies almost invariably divide diplomonads into two major clades, Hexamitinae and Giardiinae, that were already recognized on morphological grounds by Kulda and Nohýnková [3, 19, 21, 24], with the former also identified by the synapomorphy of a non-canonical genetic code . Recently the diplomonad Octomitus sp. was confirmed as a sister branch of Giardia within Giardiinae , and at least one enteromonad group was surprisingly shown to fall within Hexamitinae . Nonetheless, our understanding of the relationships amongst Fornicata is incomplete. For example, different analyses of SSU rRNA gene data place retortamonads either as a sister group of the diplomonad-enteromonad clade , as predicted by morphology [1, 3], or as a sister branch of the Giardia-Octomitus clade, thereby making diplomonads appear paraphyletic [19, 21, 22]. The main problem in resolving the phylogeny of Fornicata is the limited amount of data, both in terms of taxon sampling and the amount of sequence information per taxon. For example, to date only one enteromonad genus has been studied by molecular means, using only a single gene , whereas there are three genera of enteromonads already described, which are quite different in morphology, and enteromonads are generally not recovered as a clade in phylogenetic analyses of morphological data [1, 8].
In this study we aim to clarify relationships within fornicates, especially among enteromonads and Hexamitinae diplomonads, and to better understand the evolutionary history of single and double karyomastigonts. We introduce several important taxa into our molecular analyses, including ten new isolates of enteromonads that represent at least three genera. We also introduce two protein-coding genes into the analyses, α-tubulin and HSP90.
SSU rDNA phylogeny
Our analyses of SSU rRNA genes include the broadest taxonomic sampling of Fornicates examined so far. In addition to previously available data we have included five new isolates of the enteromonad genus Trimitus (the previously published strain KRPO3 was shown to be very closely related to other representatives of the genus Trimitus, and therefore we consider KRPO3 as a member of Trimitus), three new isolates of the genus Enteromonas, two isolates of a new enteromonad genus (isolates PSEUD and PYX, manuscript in preparation), an isolate of Trepomonas steini (see Additional files 1 and 2), a new isolate corresponding to the morphospecies Trepomonas agilis (PPS-6), a novel Spironucleus isolate (GEPA2H) and one uncultured eukaryote (CHESI2). We also extended the previously incomplete SSU rDNA sequence for Spironucleus muris. Trimitus sp. – IT1 and Trimitus sp. – KOMPKOJ represent the first observations and isolations of free-living enteromonads reported so far.
The SSU rDNA analyses were done using three different alignments, 'large' including a broad eukaryotic outgroup, 'main' including all Fornicata and a restricted outgroup, and 'small' including only Hexamitinae diplomonads and enteromonads. Figure 1 shows the ML tree based on SSU rRNA genes, with a restricted outgroup (i.e. the 'main' data set). The overall topology of the SSU rDNA tree is as follows: Fornicata forms a clade with high statistical support (bootstrap support – BS 100/99/88 and bayesian posterior probability – PP 1), enteromonads branch robustly within the Hexamitinae subtree (BS 100/99/100 and PP 1), and Giardia and Octomitus form the sister clade of the Hexamitinae-enteromonad subtree, thus rendering diplomonads plus enteromonads monophyletic to the exclusion of retortamonads, but usually with low or no statistical support (BS 43/40/85 and 0.43 PP). Very similar results were obtained when a broader outgroup sampling was employed ('large' dataset; statistical support for Fornicata monophyly: BS 98/97/82 and PP 1; for enteromonad-Hexamitinae monophyly: BS 99/98/99 and PP 1; for diplomonad/enteromonad monophyly: BS */*/64, and not recovered in Bayesian analyses).
Given the strong support for a clade consisting of all Hexamitinae and all enteromonads, we performed additional analyses with narrower taxon sampling to examine the internal relationships of this grouping (Figure 2). Within the Hexamitinae-enteromonad clade, members of the genus Spironucleus form three separated basal clades in all rooted analyses: i) Spironucleus vortens forms a clade with S. salmonis; ii) S. barkhanus branches with S. torosa and S. salmonicida; and iii) S. muris and S. meleagridis form a clade (see Figure 1). Our isolate of Spironucleus, GEPA2H, branches as a sister of S. torosa, suggesting that it could be an isolate of this species, although the genetic distance between GEPA2H and S. torosa is greater than that between S. barkhanus and S. salmonicida (Figure 2). Representatives of the genus Enteromonas constitute a weakly supported or unsupported clade in analyses of the main and small datasets, but are polyphyletic in the ML tree estimated for the large dataset, because the isolate Enteromonas hominis branches as a sister to Spironucleus muris and S. meleagridis. The members of the genus Hexamita and uncultured eukaryote CHESI2 form a clade with the enteromonads of the genus Trimitus, as well as with the new enteromonad genus. All Trimitus isolates constitute a highly supported monophyletic group. Hexamita, by contrast, does not constitute a monophyletic group within that clade. Genus Trepomonas constitutes an unsupported clade (17% ML bootstrap support) in analyses of the main and small datasets and forms a paraphyletic group in analyses of the large data set.
Concatenation of SSU rDNA, and protein gene data
We have obtained sequences of HSP90 and/or α-tubulin coding genes for all outstanding sampled genera of diplomonads, except Trepomonas. For three taxa we obtained only one of the protein coding genes (Table 1). Interestingly, the alternative genetic code usage described earlier in all Hexamitinae (TAA and TAG encode glutamine instead of stop codons; 26), was also identified in the protein coding gene sequences from two out of three enteromonad genera (Enteromonas and the undescribed enteromonad genus). Analyses of HSP90 and α-tubulin genes support the monophyly of the genus Enteromonas, as Enteromonas GECA2 and Enteromonas hominis ENTEROII constitute a highly supported clade in both cases (see Additional files 1, 3 and 4).
The phylogenetic tree estimated for the concatenated SSU rDNA, HSP90 and α-tubulin data is shown in Figure 3. As in the SSU rDNA trees, Spironucleus vortens and Spironucleus salmonicida branch paraphyletically at the base of the Hexamitinae-enteromonad clade with high statistical support. Spironucleus muris, the genera Enteromonas, Hexamita and Trimitus, and the new enteromonad genus (represented by isolate PSEUD) constitute a clade with weakly resolved internal relationships. By contrast, Giardia intestinalis branches as a sister group of the Hexamitinae-enteromonad clade with high statistical support (BS 82), leaving retortamonads in a sister position to the whole 'diplomonads plus enteromonads' clade.
In order to explore further the evolutionary positions of enteromonads we performed AU tests of alternative topologies. In the case of the SSU rRNA gene data, we compared a set of 1000 reasonable trees, including both the ML tree and the topology of highest likelihood in which enteromonads were monophyletic. For the concatenated dataset, α-tubulin, and HSP90, all reasonable trees were examined (945, 945, and 15 trees respectively, see Methods). The monophyly of enteromonads was not rejected by analyses of any one single gene. In the analysis of the concatenated dataset, however, the monophyly of enteromonads was rejected at the level of 5% (p = 0.048).
Molecular phylogeny of Fornicata
Our analyses of SSU rRNA genes include the broadest taxonomic sampling of Fornicata published so far. Our results are generally consistent with those of other recent studies [19, 21–23]. The genus Spironucleus constitutes three separate branches close to the base of the Hexamitinae-enteromonad subtree. This topology is in good agreement with previous studies [21, 24, 27], however, all three clades of Spironucleus constitute long branches and their position at the base of the Hexamitinae-enteromonad subtree could be a long-branch attraction artifact . The presence of Spironucleus isolate GEPA2H within the clade of Spironucleus barkhanus, S. torosa, and S. salmonicida is noteworthy, since GEPA2H was isolated from the terrestrial tortoise Geochelone pardalis, while the latter three species infect marine teleosts. This argues against Jørgensen and Sterud's  proposal that there are more-or-less distinct marine, freshwater, and terrestrial clades within Spironucleus. The internal relationships within the remainder of the Hexamitinae-enteromonad subtree were weakly or not supported and vary widely with the method of tree reconstruction and alignment. Further data will be required for a complete picture of the relationships within Hexamitinae.
Analyses of concatenated genes quite strongly support the monophyly of diplomonads plus enteromonads to the exclusion of retortamonads and Carpediemonas, albeit within the context of a smaller taxon sampling than is presently available for SSU rDNA alone. This supports previous morphological studies and analyses [1, 7, 8] but contrasts with some previous studies of SSU rDNA data, in which retortamonads branch weakly within diplomonads, as the sister group to Giardiinae [19, 21, 22]. The tendency for retortamonads to branch within diplomonads in SSU rDNA analyses is most probably an analysis artifact, but the cause of this artifact is not clear. Our preliminary analyses do not support the notion that either base composition heterogeneity (Giardia SSU rDNA is notable for its high GC content, however neither LogDet distance correction nor RY recoding results in a different well resolved topology) or simple long-branch attraction is to blame (data not shown). It is also possible that SSU rDNA simply does not contain enough information to resolve the phylogenetic relationships among the ingroup taxa.
The positions of enteromonads
Enteromonads and Hexamitinae diplomonads form a monophyletic group to the exclusion of other Fornicata. This is consistent with previous molecular phylogenies that included a much smaller sampling of enteromonads  and one morphological study . The clade has very strong statistical support in our analyses, and is further supported by a molecular synapomorphy: the non-canonical genetic code common to all studied Hexamitinae  appears also to be present in at least two of the three enteromonads for which we obtained protein-coding gene sequence data.
The most striking result of our study is the non-monophyly of enteromonads. The possibility of enteromonads being polyphyletic was suggested speculatively by Simpson  on the basis of morphological data. The internal relationships of the Hexamitinae-enteromonad clade are weakly supported at present, and need further investigation. More extensive taxon sampling for studied protein-coding genes is necessary, as well as more protein-coding genes. Nonetheless, our results support the non-monophyletic status of enteromonads, as the monophyly of enteromonads was rejected by AU test with the concatenated data set. Interestingly, none of the single gene phylogenies themselves rejected enteromonad monophyly. The differing results from analysis of the concatenated dataset and the single-gene datasets could be caused either by an insufficient amount of data in single gene analyses or by conflicting signal between single gene phylogenies. However, none of the single gene ML trees was rejected in AU tests of the concatenated dataset, suggesting that there is no strong conflict between single gene phylogenies. It is reasonable to assume therefore that the rejection of enteromonad monophyly on the basis of concatenated data, but not in to single gene analyses, is due to insufficient signal rather than conflicting single-gene data.
Many authors have treated diplomonads and enteromonads as taxa of equivalent rank [3, 7, 28], implicitly or explicitly reflecting a widespread assumption that diplomonads and enteromonads are sister clades, or that enteromonads represent a paraphyletic group from which a monophyletic diplomonads group evolved [8, 28]. Our analysis demonstrates, however, that many, and probably all enteromonads fall within diplomonads and specifically within Hexamitinae. Moreover, enteromonads represent a non-monophyletic group within Hexamitinae. There is no possible interpretation of our results that would allow recognition of diplomonads and enteromonads as separate taxa, without at least one of them being polyphyletic. Therefore, we suggest that the taxon Diplomonadida and any of its effective synonyms be considered to include enteromonads, and that the taxa Enteromonadida and Enteromonadinae no longer be used. The term 'enteromonads' should be taken to have a purely descriptive, and not taxonomic, meaning.
Evolution of single and double karyomastigont cell organization
In a previous study  it was shown that at least one enteromonad clade branches within diplomonads, and branches with or within the diplomonad subgroup Hexamitinae, suggesting either that enteromonads are secondarily simplified, possibly in one evolutionary event, or that the double karyomastigont morphology characteristic of diplomonads arose more than once during their evolution. Our present study indicates that enteromonads do not constitute a monophyletic group within Hexamitinae, and thus switches between single- and double-karyomastigont morphology must have occurred several times during evolution, irrespective of the direction in which these switches occurred. If we assume only one direction of change within the diplomonad-enteromonad clade and assuming, for argument's sake, that our topologies are correct, there are two basic possibilities: either the double-karyomastigont morphology arose many times in closely related groups (at least seven times according to the SSU rRNA gene tree, or at least five times, according to the multigene analyses) or enteromonads arose by secondary reduction from double-karyomastigont ancestors at least three times independently.
There are three potential scenarios describing the switch between single and double karyomastigont morphologies (Fig 4). For comparison, the standard cell cycle of an enteromonad is depicted in Figure 4A.
1. The switch from single karyomastigont morphology to double karyomastigont morphology was envisaged by Siddall, Hong and Desser  as a change in the relative timing between karyokinesis and cytokinesis (Figure 4B). Under this model a cell with a single karyomastigont goes through nuclear and mastigont division and is prepared for cell division. Cell division is arrested, however, and the cell goes through another nuclear and mastigont division, resulting in a cell with four karyomastigonts. The cell with four karyomastigonts then divides into two cells, each with a double karyomastigont.
2. The opposite switch (Figure 4C), from a double to single karyomastigont, could also be explained as a change in the timing of karyokinesis and cytokinesis. Cells with a double karyomastigont morphology could just go through cell division without nuclear and mastigont division. Another similar event would involve a cell with four karyomastigonts (i.e. a cell normally with a double karyomastigont prepared for cell division) dividing into four daughter cells instead of two.
3. The last scenario suggests fusion of the nuclei in a cell with the double-karyomastigont morphology. The resulting cell could then lose the second mastigont, resulting in a cell with a single-karyomastigont morphology.
These scenarios have different strengths and weaknesses in terms of plausibility. Scenario one seems implausible from a phylogenetic perspective, as it requires a very large number of parallel evolutions of the distinctive doubled cell morphology within the diplomonad-enteromonad clade. Scenario two invokes many fewer evolutionary events, but requires that the two nuclei of the parental diplomonad cell be nearly identical (or, at a minimum, that at least one of the nuclei retains all essential genes), otherwise the single karyomastigont progeny would not be viable. Scenario three requires the same number of evolutionary transitions as scenario two, and is compatible with a diplomonad parent that had non-identical and essential nuclei, but is otherwise a more complex mechanism. Therefore, the key to understanding the evolution of nucleus number in diplomonads lies in knowing whether the nuclei of diplomonads are identical. Naïvely, it might be expected that the nuclei are non-identical, since one copy of each nucleus is transmitted to each daughter cell during division , and, in principle at least, the two nuclei would represent separate lineages once the diplomonad state has been fixed. This would allow essential genes to be lost from one or the other nucleus, such that both nuclei would soon be required for the cell lineage to persist and any uninucleate progeny would be unviable. However, this could be avoided if there were mechanisms that frequently generated cells in which both nuclei were copies of a single parental nucleus (most likely a sexual process).
Empirical data on the nature of diplomonad nuclei are limited and conflicting. The human parasite Giardia intestinalis is the only diplomonad whose molecular and cellular biology has been studied in detail. Yu, Birky and Adam  used probes against selected genes to indicate that each nucleus of G. intestinalis contains a complete set of genetic information. Bernarder, Palm and Svard  have also deduced from results of FACS analysis that the two nuclei of G. intestinalis are diploid. The genome sequence of Giardia intestinalis (lamblia) strain WB clone C6 was published by Morrison et al. , and minimal heterozygosity was detected. Very recently, a process of physical transfer of DNA between nuclei was reported in Giardia cysts . However, Tůmová et al.  reported that the two nuclei of G. intestinalis possess different numbers of chromosomes. The focus on Giardia is unfortunate in some respects, as Giardia is a highly specialized parasite with an organization of cytoskeletal components (at least) that is substantially different from other diplomonads. It is possible that results obtained for Giardia may not be applicable to other diplomonads. Giardia also represents the diplomonad group that is most distantly related to the various enteromonad taxa. It would be of great interest to compare the nuclei in one or more Hexamitinae diplomonads (e.g.: Spironucleus, Hexamita or Trepomonas).
There are some observations that would support scenario 1, albeit indirectly. Firstly, the populations of enteromonads often contain some double individuals resembling Hexamita in their morphology, with two fully developed karyomastigonts and no apparent signs of cytokinesis [[7, 34], M. K. personal observation]. This may suggest that there is some general tendency for delayed or arrested cytokinesis. Secondly, a recent study of the flagellar cycle of Giardia intestinalis has shown that the two karyomastigonts are not independent, as basal bodies migrate between the two karyomastigonts . This means that the flagellar maturation cycle would be corrupted in cells that switched back to a single karyomastigont.
Our molecular phylogenies contradict any scenario invoking just one unique evolutionary transition between single and double karyomastigont morphologies within diplomonads. According to the inferred topology, the most parsimonious scenario would have one transition from single to double karyomastigont morphology at the base of diplomonads and several independent reversals to the single karyomastigont morphology. However, taking into account possible inaccuracies in our estimated tree, principally involving Spironucleus, and the limited data on diplomonad cell biology, several independent transitions from single to double morphology (and no reversals) cannot be excluded. More data on the molecular and cellular biology of diplomonads in addition to Giardia will be necessary for understanding the enigmatic evolution of the double karyomastigont of diplomonads.
Our analyses of SSU rRNA, HSP90 and α-tubulin genes strongly positioned all enteromonads within Hexamitinae diplomonads and showed that enteromonads do not constitute a monophyletic group. These results suggest that transformations between single- and double-karyomastigont morphologies have occurred several times during the evolution of diplomonads, however, it is not possible to confidently determine the direction of these switches without more information about the cellular and molecular biology of diplomonads and enteromonads. We suggest that the high level taxa Enteromonadida, Enteromonadidae and Enteromonadinae should be abandoned and the genera Enteromonas and Trimitus should be considered as members of Hexamitinae diplomonads. The term 'enteromonad' should have a purely utilitarian meaning – Diplomonadida with a single karyomastigont.
Isolates used in this study are summarized in Table 2. All enteromonad isolates were obtained from animal guts or feces, except isolates KOMPKOJ and IT1, which were free living. Trepomonas steini and Trepomonas sp.-PPS6 were isolated from anoxic fresh water sediments by the authors. A culture of Spironucleus vortens was obtained from the American Type Culture Collection (ATCC #50386). DNA from Spironucleus muris was isolated from purified cysts obtained from the intestine of a SCID laboratory mouse. Xenic cultures of enteromonads were grown in Dobell-Leidlaw biphasic medium  and in TYSGM medium  without tween and mucin at 21°C, 27°C and 37°C. Spironucleus vortens was grown axenically in TYI-S33 medium as modified for Giardia at 27°C . Trepomonas sp.-PPS6 and Trepomonas steini were grown in cerophyll medium (ATCC #802) at 21°C (Table 2). DNA from Spironucleus sp. GEPA2H and uncultured eukaryote CHESI2 was isolated from crude cultures.
Gene amplification and sequencing
Genomic DNA was isolated using a High Pure PCR template kit (Roche Applied Science, UK) or using CTAB and organic extractions . SSU rDNA sequences were amplified by PCR using primers 'EntUnvF' and 'EntUnvR'  or universal eukaryotic primers . In the case of the 'new enteromonad genus' isolate 'PSEUD', the culture also contained a retortamonad species. The SSU rDNA segments from both eukaryotes were amplified and partially sequenced. Specific primers for the 'new enteromonad genus', DimA (5'-AGTCAAAGATTAAAACATGCATAT-3') and DimB (5'-TCCTCTAAGCCTTCTAGTTCGTGCAAA-3') were then designed and used for amplification of the SSU rDNA from isolate 'PYX', which is an enteromonad closely related to isolate 'PSEUD'. A specific forward primer (SSUSmur20F 5'-AACTGCGGACGGCTCATT-3') was designed for S. muris and used with the universal eukaryotic reverse primer.
Alpha-tubulin genes were amplified using primers AtubA and AtubB  and then by nested PCR with primers α-tubF1 and α-tubR1 . HSP90 genes were amplified using primers H90100X  and H90910XR . The annealing temperatures used were 45–53°C, 45–50°C and 48–53°C for SSU rDNA, α-tubulin and HSP90, respectively. SSU rDNA amplicons were sequenced directly where possible. Otherwise, major PCR fragments of the expected sizes were subcloned (TOPO TA cloning kit for sequencing, pCR4-TOPO vector, Invitrogen, USA; or pGEM-T Easy vector cloning kit, Promega, USA) and several clones (2–6) were partially sequenced. Obtained sequences were then subjected to BLAST searches  to confirm their identity. At least one of the positive clones was fully sequenced bidirectionally by primer walking. All sequences obtained during this study are deposited in GenBank [GenBank: EF551168 – EF551186, EU043230, AY921407 and AY921408].
All alignments used in this study were constructed using the program ClustalX 1.83  followed by manual editing in the program BioEdit 220.127.116.11  and are available upon request (see Additional files 1 and 5 for details about the sequences used).
SSU rRNA genes
Two data sets were constructed including all near-full-length Fornicata sequences, except some redundant close relatives within Giardia and retortamonads, plus an outgroup consisting of either i) a broad diversity of eukaryotes (large dataset), or ii) a few supposed close relatives of Fornicata – the excavate groups Parabasalia, Trimastix, Oxymonadida, Malawimonas and Andalucia (main dataset). These datasets included 887 and 1041 well-aligned sites, respectively. An additional dataset was generated that included only Hexamitinae and enteromonads, and also included 1041 sites (small dataset).
Each dataset was analyzed using several likelihood-based phylogenetic methods. The model of sequence evolution was selected by the Akaike information criterion, as implemented in the program Modeltest 3.7 . The general time reversible model of nucleotide substitution was used, with among-site rate variation modeled by a gamma distribution and a proportion of invariable sites (GTR + Γ + I model), with the gamma distribution approximated by 4 equiprobable discrete categories. Maximum likelihood (ML) analyses were performed using the program PAUP*4B10 , with 10 random taxon additions followed by tree bisection and reconnection branch rearrangements, while ML bootstrap support (200 replicates) was estimated using PAUP*4B10 (10 random taxa additions followed by TBR; for the large dataset only the ML bootstrap analysis was instead performed using the program IQPNNI 3.0.1. ), and LRSH-RELL bootstrapping (1000 replicates) was performed using the program Treefinder (version: February 2007) . The model of sequence evolution used was the same for all ML analyses. Least squares distance trees were estimated from ML distances using PAUP*4B10 and bootstrapped with 1000 replicates (each searched using 10 replicates of random taxon addition with TBR branch swapping). The Bayesian analysis was performed using the program MrBayes 3.1.2 , using the GTR + Γ + I model with two runs, each with four independent chains running for 3 × 106 generations (a burn-in of 5 × 105 generations was used), with default heating parameter and sampling frequency (which was also used in all subsequent Bayesian analyses).
Analyses of protein coding genes
The HSP90 and α-tubulin amino acid datasets included all available Fornicata sequences and an extensive eukaryotic outgroup consisting of representatives from major eukaryotic groups. The trimmed alignments included 370 sites for α-tubulin and 493 sites for HSP90. Both datasets were analyzed using the WAG + Γ + I model . The WAG matrix was selected over other substitution matrices by the Akaike information criterion, as implemented in the program ProtTest 1.4 . For each, the ML tree was estimated and bootstrap support (500 replicates) was estimated using IQPNNI 3.0.1, while LRSH-RELL bootstrap support (1000 replicates) was determined using Treefinder (version: February 2007). In addition, a Bayesian analysis (WAG + Γ + I model) was performed using MrBayes 3.1.2, with four independent chains running for 2 × 106 generations, and with a conservative burn-in of 5 × 105 generations.
Analyses of concatenated SSU rDNA and protein sequences
The concatenated alignment of SSU rDNA, α-tubulin and HSP90 genes was analyzed using the program MrBayes 3.1.2 (two runs each with four independent chains running for 5 × 106 generations with a burn-in of 1.5 × 106 generations), with among-site rate variation for each gene modeled by a discrete approximation of a gamma distribution, proportion of invariable sites and a covarion model. The GTR substitution model was used for the SSU rDNA partition and the WAG substitution matrix  was used for the protein coding genes. The branch lengths, α parameter, proportion of invariable sites and parameter for switching rates in the covarion model were estimated separately for each gene (both runs converged to the same level). The branch lengths in the depicted tree are those estimated for the HSP90 partition of the data (The alternative of displaying the average of the estimated branch lengths over all three genes was not followed on the grounds that this average does not reflect any actual parameter examined under the model of evolution we used). In addition to examining posterior probabilities we performed a full bootstrap analysis with 100 replicate samples. Each gene was re-sampled independently (using the program Seqboot from the Phylip package ) and then each bootstrap sample was created by concatenating one replicate from each gene. Each bootstrap replicate was analyzed under the same conditions as the starting dataset but using only 2.5 × 105 generations (a burn-in of 50000 generations was used). The consensus tree was made for each bootstrap replicate in MrBayes 3.1.2. The bootstrap consensus tree was then estimated from the 100 resulting trees using the program Consense from the package Phylip 3.67 .
Testing of topologies
The topologies were compared using 'Approximately Unbiased' (AU) tests implemented in the program Consel 1.19 . We performed separate AU tests on four datasets – 1. SSU rRNA genes, 2. α-tubulin, 3. HSP90, and 4. the concatenated dataset (with the missing genes treated as missing data). For the SSU rRNA test, the small dataset, which includes only Hexamitinae and enteromonads was used. For the α-tubulin, HSP90 and combined datasets, the alignments used for estimating the ML tree were used, except that taxa other than Hexamitinae and enteromonads were excluded. For AU tests using the SSU rRNA gene data we generated a set of ,reasonable trees'. This set was generated by saving the 999 trees with the highest likelihood found during ML analyses in PAUP*4B10 (10 random sequence additions plus TBR). The tree representing monophyletic enteromonads was generated using a constraint search in PAUP*4B10 (10 random addition replicates plus TBR). In the case of α-tubulin, HSP90, and concatenated analyses, we included all possible trees that were consistent with a constraint where nodes corresponding to those that had received 100% bootstrap support in the concatenated genes analysis were fixed. Site likelihoods were calculated using PAUP*4B10 for the SSU rRNA gene data, and using the PAML package  for protein data. For the concatenated gene analysis site likelihoods were generated separately for all three genes and then concatenated prior to analysis in Consel 1.19.
Simpson AGB: Cytoskeletal organization, phylogenetic affinities and systematics in the contentious taxon Excavata (Eukaryota). Int J Syst Evol Microbiol. 2003, 53: 1759-1777. 10.1099/ijs.0.02578-0.
Yubuki N, Inagaki Y, Nakayama T, Inouye I: Ultrastructure and ribosomal RNA phylogeny of the free-living heterotrophic flagellate Dysnectes brevis n. gen., n sp., a new member of the Fornicata. J Eukaryot Microbiol. 2007, 54: 191-200. 10.1111/j.1550-7408.2007.00252.x.
Kulda J, Nohýnková E: Flagellates of the human intestine and of intestines of other species. Parasitic protozoa: Intestinal flagellates, Histomonads, Trichomonads, Amoeba, Opalinids, and Ciliates. Edited by: Kreier JP. 1978, San Diego, Academic Press, II: 1-128.
Sogin ML, Gunderson JH, Elwood HJ, Alonso RA, Peattie DA: Phylogenetic meaning of the kingdom concept: an unusual ribosomal RNA from Giardia lamblia. Science. 1989, 243: 75-77. 10.1126/science.2911720.
Bapteste E, Brinkman H, Lee JA, Moore DV, Sensen CW, Gordon P, Duruflé L, Gaasterland T, Lopez P, Müller M, Philippe H: The analysis of 100 genes supports the grouping of three highly divergent amoebae: Dictyostelium, Entamoeba, and Mastigamoeba. Proc Natl Acad Sci USA. 2002, 99: 1414-1419. 10.1073/pnas.032662799.
Ciccarelli DF, Doerks T, von Mering C, Creevey CJ, Snel B, Bork P: Toward automatic reconstruction of a highly resolved tree of life. Science. 2006, 311: 1283-1287. 10.1126/science.1123061.
Brugerolle G: Contribution a l'étude cytologique et phylétique des diplozoaires (Zoomastigophorea, Diplozoa, Dangeard 1910). VI. Caractères généraux des diplozoaires. Protistologica. 1975, 11: 111-118.
Siddall ME, Hong H, Desser SS: Phylogenetic analysis of the Diplomonadida (Wenyon, 1926) Brugerolle, 1975: evidence for heterochrony in protozoa and against Giardia lamblia as a "missing link". J Protozool. 1992, 39: 361-367.
Cavalier-Smith T: A 6-kingdom classification and a unified phylogeny. Endocytobiology II. Edited by: Schwemmler W, Schenk HEA. 1983, Berlin, de Gruyter, 1027-1034.
Cavalier-Smith T: Cell cycles, diplokaryosis and the archezoan origin of sex. Arch Protistenkd. 1995, 145: 189-207.
Roger AJ, Svard SG, Tovar J, Clark CG, Smith MW, Gillin FD, Sogin ML: A mitochondrial-like chaperonin 60 gene in Giardia lamblia: Evidence that diplomonads once harbored an endosymbiont related to the progenitor of mitochondria. Proc Natl Acad Sci USA. 1998, 95: 229-234. 10.1073/pnas.95.1.229.
Tachezy J, Sánchez LB, Müller M: Mitochondrial type iron-sulfur cluster assembly in the amitochondriate eukaryotes Trichomonas vaginalis and Giardia intestinalis, as indicated by the phylogeny of IscS. Mol Biol Evol. 2001, 18: 1919-1928.
Tovar J, Leon-Avila G, Sánchez LB, Šut'ák R, Tachezy J, Giezen van der M, Hernandez M, Müller M, Lucocq JM: Mitochondrial remnant organelles of Giardia function in iron-sulfur protein maturation. Nature. 2003, 426: 172-176. 10.1038/nature01945.
Simpson AGB, Patterson DJ: The ultrastructure of Carpediemonas membranifera (Eukaryota) with reference to the excavate hypothesis. Europ J Protistol. 1999, 35: 353-370.
Felsenstein J: Inferring phylogenies. 2004, Sunderland, Massachusetts, Sinauer Associates, Inc
Brinkmann H, Giezen van der M, Zhou Y, De Raucourt GP, Philippe H: An empirical assessment of long-branch attraction artefacts in deep eukaryotic phylogenomics. Syst Biol. 2005, 54: 743-757. 10.1080/10635150500234609.
Philippe H, Delsuc F, Brinkmann H, Lartillot N: Phylogenomics. Annu Rev Ecol Evol Syst. 2005, 36: 541-62. 10.1146/annurev.ecolsys.35.112202.130205.
Best AA, Morrison HG, McArthur AG, Sogin ML, Olsen GJ: Evolution of eukaryotic transcription: insights from the genome of Giardia lamblia. Genome Res. 2004, 14: 1537-1547. 10.1101/gr.2256604.
Kolisko M, Cepicka I, Hampl V, Kulda J, Flegr J: The phylogenetic position of enteromonads: a challenge for the present models of diplomonad evolution. Int J Syst Evol Microbiol. 2005, 55: 1729-1733. 10.1099/ijs.0.63542-0.
Brugerolle G, Taylor FJR: Taxonomy, cytology and evolution of the Mastigophora. Proceedings of the Fifth International Congress of Protozoology New York. Edited by: Hutner SH. 1977, New York USA, Pace University, 14-28.
Silberman JD, Simpson AGB, Kulda J, Cepicka I, Hampl V, Johnson PJ, Roger AJ: Retortamonad flagellates are closely related to diplomonads – implications for the history of mitochondrial function in eukaryote evolution. Mol Biol Evol. 2002, 19: 777-786.
Simpson AGB, Roger AJ, Silberman JD, Leipe DD, Edgcomb VP, Jermiin LS, Patterson DJ, Sogin ML: Evolutionary history of "early-diverging" eukaryotes: the excavate taxon Carpediemonas is a close relative of Giardia. Mol Biol Evol. 2002, 19: 1782-1791.
Keeling PJ, Brugerolle G: Evidence from SSU rRNA phylogeny that Octomitus is a sister lineage of Giardia. Protist. 2006, 157: 205-212. 10.1016/j.protis.2006.01.003.
Jørgensen A, Sterud E: Phylogeny of Spironucleus [Eopharyngia: Diplomonadida: Hexamitinae]. Protist. 2007, 158: 247-254. 10.1016/j.protis.2006.12.003.
Simpson AGB, Inagaki Y, Roger AJ: Comprehensive multigene phylogenies of excavate protists reveal the evolutionary positions of "primitive" eukaryotes. Mol Biol Evol. 2006, 23: 615-625. 10.1093/molbev/msj068.
Keeling PJ, Doolittle WF: A non-canonical genetic code in an early diverging eukaryotic lineage. EMBO J. 1996, 15 (9): 2285-2290.
Cavalier-Smith T, Chao EE: Molecular phylogeny of the free-living archezoan Trepomonas agilis and the nature of the first eukaryote. J Mol Evol. 1996, 43: 551-562. 10.1007/BF02202103.
Cavalier-Smith T: The excavate protozoan phyla Metamonada Grasse emend. (Anaeromonadea, Parabasalia, Carpediemonas, Eopharyngia) and Loukozoa emend. (Jakobea, Malawimonas): their evolutionary affinities and new higher taxa. Int J Syst Evol Microbiol. 2003, 53: 1741-1758. 10.1099/ijs.0.02548-0.
Tůmová P, Hofštetrová K, Nohýnková E, Hovorka O, Král J: Cytogenetic evidence for diversity of two nuclei within a single diplomonad cell of Giardia. Chromosoma. 2006, 116: 65-78. 10.1007/s00412-006-0082-4.
Yu LZ, Birky CW, Adam RD: The two nuclei of Giardia each have complete copies of the genome and are partitioned equationally at cytokinesis. Eukaryot Cell. 2002, 1: 191-199. 10.1128/EC.1.2.191-199.2002.
Bernander R, Palm JE, Svard SG: Genome ploidy in different stages of the Giardia lamblia life cycle. Cell Microbiol. 2001, 3: 55-62. 10.1046/j.1462-5822.2001.00094.x.
Morrison HG, McArthur AG, Gillin FD, Aley SB, Adam RD, Olsen GJ, Best AA, Cande WZ, Chen F, Cipriano MJ, Davids BJ, Dawson SC, Elmendorf HG, Hehl AB, Holder ME, Huse SM, Kim UU, Lasek-Nesselquist E, Manning G, Nigam A, Nixon JEJ, Palm D, Passamaneck NE, Prabhu A, Reich CI, Reiner DS, Samuelson G, Svard SG, Sogin ML: Genomic minimalism in the early diverging intestinal parasite Giardia lamblia. Science. 2007, 317: 1921-1926. 10.1126/science.1143837.
Poxleitner MK, Carpenter ML, Mancuso JJ, Wang C-JR, Dawson SC, Cande WZ: Evidence for karyogamy and exchange of genetic material in the binucleate intestinal parasite Giardia intestinalis. Science. 2008, 319: 1530-1533. 10.1126/science.1153752.
Brugerolle G: Séparation des genres Trimitus (Diplomonadida) and Tricercomitus (Trichomonadida) d'après leur ultrastructure. Protistologica. 1986, 22: 31-37.
Nohýnková E, Tůmová P, Kulda J: Cell division of Giardia intestinalis: flagellar developmental cycle involves transformation and exchange of flagella between mastigonts of a diplomonad cell. Eukaryot Cell. 2006, 5: 753-761. 10.1128/EC.5.4.753-761.2006.
Dobell C, Leidlaw PP: On the cultivation of Entamoeba histolytica and some other parasitic amoebae. Parasitology. 1926, 18: 283-318.
Clark CG, Diamond LS: Methods for cultivation of luminal parasitic protists of clinical importance. Clin Microbiol Rev. 2002, 15: 329-341. 10.1128/CMR.15.3.329-341.2002.
Keister DB: Axenic culture of Giardia lamblia in TYI-S-33 medium supplemented with bile. Trans R Soc Trop Med Hyg. 1983, 77: 487-488. 10.1016/0035-9203(83)90120-7.
Clark CG: DNA purification from polysaccharide-rich cells. Protocols in protozoology. Edited by: Lee JJ, Soldo AT. 1992, Kansas, Lawrence, Allen Press, 1: D-3.1-D-3.2.
Medlin L, Elwood HJ, Stickles S, Sogin ML: The characterization of enzymatically amplified eukaryotic 16S-like rRNA-coding regions. Gene. 1988, 71: 491-499. 10.1016/0378-1119(88)90066-2.
Edgcomb VP, Roger AJ, Simpson AGB, Kysela DT, Sogin ML: Evolutionary relationships among "jakobid" flagellates as indicated by alpha- and beta-tubulin phylogenies. Mol Biol Evol. 2001, 18: 514-522.
Moriya S, Tanaka M, Ohkuma M, Sugano S, Kudo T: Diversification of the microtubule system in the early stage of eukaryotic evolution: elongation factor-1α and α-tubulin protein phylogeny of termite symbiotic oxymonad and hypermastigote protists. J Mol Evol. 2001, 52: 6-16.
Simpson AGB, Lukeš J, Roger AJ: The evolutionary history of kinetoplastids and their kinetoplasts. Mol Biol Evol. 2002, 19: 2071-2083.
Altschul SF, Gish W, Miller W, Myers EW, Lipman DJ: Basic local alignment search tool. J Mol Biol. 1990, 215: 403-410.
Thompson JD, Gibson TJ, Plewniak F, Jeanmougin F, Higgins DG: The ClustalX windows interface: flexible strategies for multiple sequence alignment aided by quality analysis tools. Nucleic Acids Res. 1997, 24: 4876-4882. 10.1093/nar/25.24.4876.
Hall TA: BioEdit: a user-friendly biological sequence alignment editor and analysis program for Windows 95/98/NT. Nucl Acids Symp Ser. 1999, 41: 95-98.
Posada D, Crandall CA: MODELTEST: testing the model of DNA substitution. Bioinformatics. 1998, 14: 817-818. 10.1093/bioinformatics/14.9.817.
Swofford DL: PAUP*. Phylogenetic analysis using parsimony (* and other methods). Version 4*B10. 2002, Sunderland, Massachusetts: Sinauer Associates
Vinh LS, von Haeseler A: IQPNNI: moving fast through tree space and stopping in time. Mol Biol Evol. 2004, 8: 1565-1571. 10.1093/molbev/msh176.
Jobb G, von Haeseler A, Strimmer K: TREEFINDER: a powerful graphical analysis environment for molecular phylogenetics. BMC Evol Biol. 2004, 4: 18-10.1186/1471-2148-4-18.
Huelsenbeck JP: MrBayes: Bayesian inference of phylogeny. Distributed by the author. 2000, Department of Biology, University of Rochester
Whelan S, Goldman N: A general empirical model of protein evolution derived from multiple protein families using a maximum likelihood approach. Mol Biol Evol. 2001, 18: 691-699.
Abascal F, Zardoya R, Posada D: Selection of best-fit models of protein evolution. Bioinformatics. 2005, 21: 2104-2105. 10.1093/bioinformatics/bti263.
Felsenstein J: PHYLIP (Phylogeny inference package) version 3.6. Distributed by the author. 2004, Department of Genome Sciences, University of Washington, Seattle
Shimodaira H, Hasegawa M: CONSEL: for assessing the confidence of phylogenetic tree selection. Bioinformatics. 2001, 17: 1246-1247. 10.1093/bioinformatics/17.12.1246.
Yang Z: PAML: a program package for phylogenetic analysis by maximum likelihood. Comput Appl Biosci. 1997, 13: 555-556.
We would like to thank Magdalena Uzlikova for providing the sequence from isolate Spironucleus sp. GEPA2H, and Jeffrey Silberman and two anonymous reviewers for helpful comments. The work was supported by the Czech Ministry of Education (project 0021620828), the Czech Science Foundation (project 206/05/0371) and NSERC Discovery grants 298366-04 and 227085-05 awarded to AGBS and AJR respectively. AJR and AGBS are supported as a Fellow and Scholar respectively of the Canadian Institute for Advanced Research (CIFAR), program in Integrated Microbial Biodiversity. VH is supported by a postdoctoral fellowship from CIFAR.
MK contributed by gene amplification and sequencing (SSU and HSP90), data analyses and by drafting the manuscript. IC isolated most of the new strains examined in this study, and obtained α-tubulin sequences. VH contributed to data interpretation and by critical reading of the manuscript. JL contributed to data analyses and by critical reading. AJR contributed by critical reading and by supervision of JL and MK. JK contributed by critical reading and isolation of isolate ENTEROII. AGBS contributed to data interpretation, writing the manuscript and by supervising MK. JF contributed to data interpretation, writing the manuscript and by supervising MK at Charles University, Prague.