Phylogenetic placement of the enigmatic parasite, Polypodium hydriforme, within the Phylum Cnidaria
© Evans et al. 2008
Received: 04 December 2007
Accepted: 09 May 2008
Published: 09 May 2008
Skip to main content
© Evans et al. 2008
Received: 04 December 2007
Accepted: 09 May 2008
Published: 09 May 2008
Polypodium hydriforme is a parasite with an unusual life cycle and peculiar morphology, both of which have made its systematic position uncertain. Polypodium has traditionally been considered a cnidarian because it possesses nematocysts, the stinging structures characteristic of this phylum. However, recent molecular phylogenetic studies using 18S rDNA sequence data have challenged this interpretation, and have shown that Polypodium is a close relative to myxozoans and together they share a closer affinity to bilaterians than cnidarians. Due to the variable rates of 18S rDNA sequences, these results have been suggested to be an artifact of long-branch attraction (LBA). A recent study, using multiple protein coding markers, shows that the myxozoan Buddenbrockia, is nested within cnidarians. Polypodium was not included in this study. To further investigate the phylogenetic placement of Polypodium, we have performed phylogenetic analyses of metazoans with 18S and partial 28S rDNA sequences in a large dataset that includes Polypodium and a comprehensive sampling of cnidarian taxa.
Analyses of a combined dataset of 18S and partial 28S sequences, and partial 28S alone, support the placement of Polypodium within Cnidaria. Removal of the long-branched myxozoans from the 18S dataset also results in Polypodium being nested within Cnidaria. These results suggest that previous reports showing that Polypodium and Myxozoa form a sister group to Bilateria were an artifact of long-branch attraction.
By including 28S rDNA sequences and a comprehensive sampling of cnidarian taxa, we demonstrate that previously conflicting hypotheses concerning the phylogenetic placement of Polypodium can be reconciled. Specifically, the data presented provide evidence that Polypodium is indeed a cnidarian and is either the sister taxon to Hydrozoa, or part of the hydrozoan clade, Leptothecata. The former hypothesis is consistent with the traditional view that Polypodium should be placed in its own cnidarian class, Polypodiozoa.
Two conflicting hypotheses have been proposed regarding the phylogenetic placement of Polypodium. The first, more traditional hypothesis is that Polypodium is a cnidarian. Some have suggested it is nested within a derived group of hydrozoans, the Narcomedusae [10–13] or the cnidarian class Scyphozoa ; while others have suggested it belongs to a separate cnidarian class, Polypodiozoa [1, 15, 16]. The assignment of Polypodium to Cnidaria is based primarily on morphological evidence, most notably the fact that Polypodium possesses nematocysts [17, 18], the stinging structures characteristic of all cnidarians. In addition, the presence of tentacles and overall body-plan organization of Polypodium are reminiscent of cnidarians, although it is unclear if the adult free-living stage is homologous to a polyp or medusa stage. This hypothesis is supported by a cladistic analysis of small subunit nuclear ribosomal DNA (18S rDNA) sequences in conjunction with morphological characters (including nematocysts) . In this study, Polypodium falls within the medusozoan clade of cnidarians, although the non-cnidarian placozoan, Trichoplax [19, 20], also fell within this clade, rendering Cnidaria paraphyletic.
The second hypothesis is that Polypodium is the sister taxon to Myxozoa, a diverse group of parasites in aquatic animals, and that Polypodium + Myxozoa is the sister group to Bilateria [2–4]. This hypothesis is derived from cladistic analyses utilizing 18S rDNA sequences [2–4]. However, because Polypodium and myxozoans have unusually high divergence rates in their 18S rDNA sequences, these cladistic analyses have been criticized by a number of authors who suggest that the data might be unduly affected by long-branch attraction (LBA) [5, 21, 22]. Despite some attempts to overcome the effects of LBA through the use of a maximum likelihood (ML) approach [21–23] and pruning long branches [5, 22], these results have been largely silent on the placement of Polypodium. For instance, Kim et al.  applied a maximum likelihood approach to 18S rDNA sequence data and found that myxozoans and Polypodium did not group together. Instead, Polypodium was part of an unresolved polytomy that included several cnidarian lineages and Trichoplax, as well as myxozoans + Bilateria. Most recently, Jimenez-Guri et al.  utilized multiple protein-coding gene sequences in a ML analysis and found the myxozoan, Buddenbrockia plumatellae nested within cnidarians. Unfortunately, this study had relatively limited sampling of cnidarians and did not include Polypodium.
In an attempt to resolve this controversy, we sequenced an additional marker in Polypodium, a partial gene sequence of the large nuclear ribosomal unit (28S rDNA), and greatly expanded the taxonomic sampling of cnidarian sequences. Using this approach, we provide evidence that Polypodium is nested within Cnidaria and does not group with myxozoans.
Taxon and sequence list
Tenebrio sp. /Tenebrio molitor
Oncorhynchus sp. /O. kisutch
Pycnophyes sp. Tjarno
Polypodium (Host: Acipenser ruthenus)
Polypodium (Host: Polyodon spathula)
Polypodium (Host:Scaphirhynchus platorynchus)
Darwin carybdeid sp.
Garveia annulata/Garveia sp.
All Polypodium sequences were newly generated for this study. We did not include the previously published 18S Polypodium sequence (GenBank accession number U37526) because of concern over the quality of the sequence which included a number of ambiguities. Furthermore, while the two new Polypodium 18S sequences (from hosts Acipenser ruthensus and Polyodon spathula) differed from each other by a total of 8 sites they differed from #U37526 by 77 and 83 sites respectively. These differences included a large number of insertions and deletions. The two new 28S sequences (from hosts Acipenser ruthensus and Scaphirhynchus platorynchus) only differed from each other by 2 sites.
Myxozoans and Polypodium have unusually high rates of evolution in their 18S and 28S rDNA sequences relative to the other sampled taxa. To investigate the influence of myxozoans on the placement of Polypodium, we removed the myxozoans from our three datasets and re-ran each analysis. Under the ML analysis of 18S rDNA, the removal of myxozoans results in the placement of Polypodium nested within Cnidaria (Figure 3B and Additional file 3B). This result suggests that the placement of Polypodium at the base of bilaterians in the 18S analysis (Figure 3A) was indeed an artifact of LBA. The placement of Polypodium within Cnidaria was not effected by the removal of myxozoans in the 28S (Additional file 6) and combined datasets (Additional file 7).
To investigate the possible role of LBA on myxozoan placement, we removed Polypodium from the combined ML analyses and found that it did not affect the position of Myxozoa at the base of the Bilateria (not shown). Given that bilaterians also form long branches, we tried removing all bilaterian sequences in the combined ML analysis. This resulted in a Myxozoa + Polypodium clade nested within Cnidaria (not shown). However, when Polypodium and bilaterians were removed, myxozoans fell outside the cnidarians (not shown). Similar effects of myxozoan placement to long-branches were also found in parsimony analyses of the combined dataset (not shown).
Our metazoan dataset of 18S and partial 28S rDNA sequences, with a large taxonomic sample of cnidarians, places Polypodium within a monophyletic Cnidaria. This accords with the fact that Polypodium possesses nematocysts [17, 18] and a cnidarian-like body plan [7–9, 12]. The precise placement of Polypodium within Cnidaria is less certain. The ML combined analysis places Polypodium as sister to Hydrozoa (Figure 2A), a hypothesis consistent with the suggestion that Polypodium be considered a separate class of cnidarians, Polypodiozoa . By contrast, the combined parsimony analysis (Figure 2B) and the ML analyses of 28S alone (Additional file 2 and 6) place Polypodium within the hydrozoan clade Leptothecata. Given that leptothecates have relatively high rates of evolution within hydrozoans, one possible explanation for the conflicting hypotheses is that the placement of Polypodium within leptothecates is an artifact of LBA and that the combined data, in conjunction with the ML approach (Figure 2A), overcame this localized LBA artifact.
Although the fresh water habitat of Polypodium is unusual for cnidarians, it is not unheard of, especially within hydrozoans. For instance, the model organism Hydra and the jellyfish Craspedacusta are both exclusively fresh-water hydrozoans. Hydra and Craspedacusta are distantly related  and our analyses do not indicate a close phylogenetic affinity of Polypodium to either of the clades containing these taxa. Thus, it appears that in the evolution of cnidarians, invasion to fresh-water habitats has happened at least three separate times.
Although Polypodium is the only known intracellular cnidarian parasite, other cnidarians have adopted parasitic life-styles [11, 26–29]. For example, parasites belonging to the Narcomedusae (Hydrozoa) have been reported to live in the stomach cavities of other narcomedusae [11, 27] and anthomedusae . In addition, the sea anemone Edwardsiella lineata parasitizes the stomach cavity of the ctenophore Mnemiopsis leidyi  and the anemone Peachia quinquecapitata is reported to parasitize the stomachs of hydromedusa .
The well-documented effects of long-branch attraction artifacts (reviewed in Bergsten ) are particularly concerning when investigating relationships amongst early-diverging metazoans, where rates between lineages vary greatly . Suggestions for avoiding LBA artifacts include choice of appropriate markers [31, 32], increased taxonomic sampling to effectively break up long branches [33, 34] and utilization of best-fit models that incorporate rate variation [21–23]. Previous conflicting reports that show Polypodium and myxozoans form a sister taxon to Bilateria [2–4] can be explained by limited taxon sampling and an inadequate number of informative characters in their analyses, both of which confound long-branch problems. In this study, the increased taxonomic sampling of cnidarians and the addition of 28S rDNA sequence data proved critical to placing the highly divergent Polypodium taxon within Cnidaria. The choice of optimality criteria (ML vs. parsimony) both supported Polypodium as a cnidarian but did affect the placement within Cnidaria.
Our analyses are inconclusive in the placement of Myxozoa within metazoans. We found that myxozoans consistently grouped with long-branched taxa and that removal of long-branches resulted in myxozoans being placed to the next longest branch. For example myxozoans group with Polypodium in the absence of Bilateria and group with Bilateria in the absence of Polypodium (not shown).
Jimenez-Guri et al.  sampled the myxozoan, Buddenbrockia, and found it to fall within Cnidaria, as the sister group to two hydrozoan representatives and a single scyphozoan. Previous studies have suggested a sister group relationship between cnidarians and myxozoans [2–4], and some morphological evidence has been used to support this view . Although our present study does not support this relationship, further investigation is merited. Myxozoans are a highly diverse group (reviewed in Kent et al. ) that comprise two clades, the Myxosporea and the Malacosporea . We were only able to include 28S rDNA sequences from myxosporeans, although the malacosporean Buddenbrockia was included in our 18S analysis and found to group with other myxozoans and outside of Cnidaria. Future studies with a comprehensive sampling of myxozoans together with Polypodium, in a dataset that includes a large taxonomic sampling of cnidarians, should shed further light on the relationships between myxozoans and Polypodium.
Although previous molecular phylogenetic hypotheses conflicted with the traditional interpretation of cnidarian affinity for Polypodium, the molecular evidence we present, using an augmented dataset, ultimately confirms and reconciles this traditional hypothesis and suggests that Polypodium is indeed a cnidarian. This study also reaffirms the importance to large taxonomic sampling and inclusion of additional informative characters for avoiding long-branch attraction artifacts.
Genomic DNA was extracted using Qiagen DNeasy kits according to manufacturer's protocol (QIAGEN Inc., Mississauga, ON) or a standard phenol/chloroform protocol. The latter method involved tissue digestion with proteinase K (20 mg/ml) in a lysis buffer (20 mM Tris-CL pH 8.0, 5 mM EDTA pH 8.0, 400 mM NaCl, 2%SDS), extraction with phenol/chloroform (1:1), precipitation with 2.5 vol. 95% EtOH, and elution in TE or H2O.
An approximately 1.8 kb portion of the gene coding for 18S was amplified and sequenced with universal eukaryotic primers as described by Medlin et al. , with the annealing temperature modified to 57°C. With the exception of Polypodium samples, a nearly complete, roughly 3 kb portion of the gene coding for 28S was amplified and sequenced with an approach modified from that reported in Collins et al. . 28S was directly amplified in two fragments with combinations of primers F63mod+R2077sq and F1379+R3264 from Medina et al.  or newly developed medusozoan specific primers F97+R2084 and F1383+R3238 (F97: CCYYAGTAACGGCGAGT, R2084: AGAGCCAATCCTTTTCC, F1383: GGACGGTGGCCATGGAAGT, and R3238: SWACAGATGGTAGCTTCG). Amplifications of 28S were conducted with the following thermal profile: 4 minutes at 94°C; 30 cycles of 30 seconds at 94°C, 1 minute at 45°C, and 3 minutes at 72°C; and 10 minutes at 72°C. For Polypodium, a portion of the 5' end of 28S (approx. 0.8–1.0 kbps) was amplified using two universal metazoan primers (fw1and rev2) as reported by Sonnenberg et al. . Sequencing was carried out using amplification primers and F635sq and R635sq from Medina et al. .
All gene fragments were purified and sequenced by Cogenics, Inc. (Houston, TX) and assembled and edited using Sequencher v4.5 (Gene Code Co., 2005). Sequences for each marker were aligned using the program MUSCLE . The 28S sequence alignment was then trimmed to reflect only that region which included sequence data for Polypodium. This trimmed 28S dataset was analyzed separately and used in conjunction with the complete 18S sequences to create the combined dataset.
Phylogenetic analyses were performed using both maximum likelihood (ML) and parsimony criteria. ML searches were performed using GARLI v0.951.OsX-GUI  under an assumed GTR model with rates estimated from the data. The assumed model of nucleotide substitution was selected by using the Akaike Information Criterion (AIC) as implemented in ModelTest . Each run was repeated 10 times from random starting trees using default termination conditions. Each run gave identical topologies and similar likelihood scores. 100 bootstrap replications were run in GARLI v0.951.0sX-GUI  under the same parameters.
To assess the effect that omitting length-variable regions has on topology, we removed these regions from the combined dataset, using the less stringent settings of Gblocks . This dataset contained 126 metazoan taxa, 2415 characters, 1391 of which are parsimony informative. We found that removal of length-variable regions had no effect on the placement of Polypodium and minimal effect on overall topology in our combined ML analyses (Additional file 8). Therefore we performed all other analyses with the complete datasets, including the more variable regions.
Parsimony analyses were performed using TNTv.1.1 . Separate tree searches were performed with gaps coded as missing and gaps coded as a fifth state. However, with one exception (see results for myxozoan placement with 18S data) there was no significant difference in topology. Numerous search methods available in TNT were utilized to search the tree space but the following approach was found to consistently recover trees with minimum lengths from our datasets. The implemented search was a driven new technology search with a random seed of 0 (where 0 = time). Default settings for sectorial searches (RSS and CSS) and tree fusing were used , with 5 replicates per repetition, and a requirement that the global optimum be found 20 times. TBR branch swapping was performed on the resulting trees and a strict consensus was calculated. TNT was used to calculate standard bootstrap values (1000 replicates). Alignments and trees for 18S, 28S and combined datasets have been submitted to TreeBASE http://www.treebase.org/treebase/index.html.
We thank S. Ash Bullard, Jan Dean, Bobby Reed and Ron Bruch for contributions of Polypodium specimens. We also thank Peter Schuchert, Casey Dunn and Michael Dawson for contributions of other cnidarian specimens. NME acknowledges the instructors at the 2007 Ohio State Cladistics workshop, especially P. Goloboff, for help with TNT. We thank M. Holder for discussions, B. Bentlage, B. Lieberman, A. Nawrocki, and five anonymous reviewers for helpful comments on earlier versions of this manuscript. This work was supported by grants from NSF AToL EF-0531779 (to PC and AGC) and support for AL from NSF (PEET DEB-9978086) and FAPESP (06/02960-8/05821-9/60327-0).
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.