Portal de Periódicos da CAPES

Animal Phylogeny: Fatal Attraction

2005; Elsevier BV; Volume: 15; Issue: 8 Linguagem: Inglês

10.1016/j.cub.2005.04.001

1879-0445

Maximilian J. Telford, Richard R. Copley,

Genetics, Aging, and Longevity in Model Organisms

MPhylogenetic analyses of hundreds of genes from model animals have placed flies closer to vertebrates than to nematodes; recent work suggests this may be due to an artefact known as long branch attraction. MPhylogenetic analyses of hundreds of genes from model animals have placed flies closer to vertebrates than to nematodes; recent work suggests this may be due to an artefact known as long branch attraction. The traditional view of animal evolution is one of gradually increasing complexity. The earliest-branching flatworms lack the body cavity known as a coelom, which is a characteristic feature of the two traditional groups of 'higher' animals: deuterostomes, including echinoderms and chordates, and protostomes, such as annelids, molluscs and arthropods. Between these two extremes, according to the traditional view, lie the pseudoceolomate worms such as the nematodes, the body cavities of which lack the refinements of a true coelom. This hierarchical view was shaken in the mid 1990s by a phylogenetic study of small subunit ribosomal (r)RNA genes [1Aguinaldo A.A. Turbeville J.M. Linford L.S. Rivera M.C. Garey J.R. Raff R.A. Lake J.A. Evidence for a clade of nematodes, arthropods and other moulting animals.Nature. 1997; 387: 489-493Crossref PubMed Scopus (1216) Google Scholar]. This work elevated the acoelomate flatworms to a close relationship with the coelomate annelids and molluscs, in a group called the Lophotrochozoa, and pseudocoelomate nematodes moved close to the coelomate arthropods, creating a group called the Ecdysozoa. Opposing the 'new animal phylogeny', as this new scheme has been called [2Adoutte A. Balavoine G. Lartillot N. Lespinet O. Prud'homme B. de Rosa R. The new animal phylogeny: reliability and implications.Proc. Natl. Acad. Sci. USA. 2000; 97: 4453-4456Crossref PubMed Scopus (411) Google Scholar], are several analyses [3Wolf Y.I. Rogozin I.B. Koonin E.V. Coelomata and not Ecdysozoa: Evidence from genome-wide phylogenetic analysis.Genome Res. 2004; 14: 29-36Crossref PubMed Scopus (172) Google Scholar, 4Mushegian A.R. Garey J.R. Martin J. Liu L.X. Large-scale taxonomic profiling of eukaryotic model organisms: a comparison of orthologous proteins encoded by the human, fly, nematode and yeast genomes.Genome Res. 1998; 8: 590-598Crossref PubMed Scopus (144) Google Scholar, 5Blair J.E. Ikeo K. Gojobori T. Hedges S.B. The evolutionary position of nematodes.BMC Evol. Biol. 2002; 2: 7Crossref PubMed Scopus (158) Google Scholar] of huge numbers of genes — close to 800 in the most recent [6Philip, G.K., Creevey, C.J., and McInerney, J.O. (2005). The Opisthokonta and the Ecdysozoa may not be clades: stronger support for the grouping of plant and animal than for animal and fungi and stronger support for the Coelomata than Ecdysozoa. Mol. Biol. Evol. Advance Access doi:10.1093/molbev/msi10.Google Scholar] — sampled from the few animals with completely sequenced genomes: fruitfly, nematode and various vertebrates. These multigene analyses are unanimous in grouping coelomate arthropods and vertebrates to the exclusion of the pseudocoelomate nematodes, so reverting to traditional views of their relationships. The overwhelming number of genes supporting the old scheme might suggest that the new animal phylogeny was finished — an artefact of a small data set. New work, however, suggests this conclusion is premature, and that the multigene result might itself be based on an artefact called long branch attraction [7Philippe, H., Lartillot, N., and Brinkmann, H. (2005). Multigene analyses of bilaterian animals corroborate the monophyly of Ecdysozoa, Lophotrochozoa and Protostomia. Mol. Biol. Evol. Advance Access doi:10.1093/molbev/msi11.Google Scholar]. The pernicious effects of long branch attraction occur when sequences from some species in a phylogenetic analysis have evolved much faster than others, making them 'long branch' species [8Felsenstein J. Cases in which parsimony or compatibility methods will be positively misleading.Syst. Zool. 1978; 27: 401-410Crossref Google Scholar]. The result of this relatively common phenomenon is a tendency for all methods of tree reconstruction to group the long-branch species together regardless of their true relationship. Because the species used as an outgroup to root the tree inevitably has a relatively long branch, it can incorrectly 'attract' long branch species towards the base of the tree (Figure 1). The whole genome datasets are of such an overwhelming size that multigene analyses have to be taken seriously as a challenge to the new animal phylogeny. Even so, the suspicion has been that the multigene result could be due to long branch attraction dragging the long-branched nematode Caenorhabditis elegans towards the base of the tree, away from the rightful place of the nematodes adjacent to the arthropods [9Telford M.J. Animal phylogeny: Back to the Coelomata?.Curr. Biol. 2004; 14: R247-R276Abstract Full Text Full Text PDF PubMed Scopus (23) Google Scholar]. If such an artefact is affecting the majority of genes, the addition of more and more data would actually lead to stronger and stronger support for the wrong tree topology, a situation called inconsistency. Initial support for the Ecdysozoa came from work aiming to avoid long branch attraction by using the intensely sampled rRNA gene sequences to select nematode species with shorter branches. Whereas long branched species of nematodes branch at the root of the tree, discarding these and using the shorter branched species resulted in nematodes moving up the tree, adjacent to the arthropods [1Aguinaldo A.A. Turbeville J.M. Linford L.S. Rivera M.C. Garey J.R. Raff R.A. Lake J.A. Evidence for a clade of nematodes, arthropods and other moulting animals.Nature. 1997; 387: 489-493Crossref PubMed Scopus (1216) Google Scholar]. Lacking sequences from additional taxa, the multigene workers have had to use alternative approaches to challenge the possibility that their finding of a basally positioned nematode is due to long branch attraction. They have shown that discarding those genes that ought to be most prone to long branch attraction — those that evolve faster or with more uneven rates — does not remove the overall support for Coelomata [5Blair J.E. Ikeo K. Gojobori T. Hedges S.B. The evolutionary position of nematodes.BMC Evol. Biol. 2002; 2: 7Crossref PubMed Scopus (158) Google Scholar]. Furthermore, simulation studies, based on the null hypothesis that Ecdysozoa is a real grouping, suggest that the observed support for Coelomata cannot credibly be explained by long branch attraction [3Wolf Y.I. Rogozin I.B. Koonin E.V. Coelomata and not Ecdysozoa: Evidence from genome-wide phylogenetic analysis.Genome Res. 2004; 14: 29-36Crossref PubMed Scopus (172) Google Scholar]. The suspicion remains, however, that long branch attraction has not been conclusively ruled out, and that a study that combines the benefits of both the many-genes and many-species approaches is required to settle the dispute [9Telford M.J. Animal phylogeny: Back to the Coelomata?.Curr. Biol. 2004; 14: R247-R276Abstract Full Text Full Text PDF PubMed Scopus (23) Google Scholar]. The work of Philippe et al. [7Philippe, H., Lartillot, N., and Brinkmann, H. (2005). Multigene analyses of bilaterian animals corroborate the monophyly of Ecdysozoa, Lophotrochozoa and Protostomia. Mol. Biol. Evol. Advance Access doi:10.1093/molbev/msi11.Google Scholar] is the first attempt at using a dataset containing many genes and many species: 146 genes from 49 species, including 35 animals. To increase representation of more obscure phyla, many of the sequences used come from expressed-sequence tags (ESTs). The large number of genes should protect the authors from accusations of artefacts due to insufficient data, while the many taxa have allowed them to address the question of long branch attraction in a number of ways. In our view, the most important advance in this work is the use of an animal, rather than a fungus, to root the tree (Figure 2). Previous analyses used the closest completely sequenced, non-animal genome: that of the yeast Saccharomyces cerevisiae, which because of its considerable evolutionary distance from animals is found at the end of a rather long branch. Reasoning that long branch attraction might act to group long branched C. elegans with this outgroup at the root of the tree, Philippe et al. [7Philippe, H., Lartillot, N., and Brinkmann, H. (2005). Multigene analyses of bilaterian animals corroborate the monophyly of Ecdysozoa, Lophotrochozoa and Protostomia. Mol. Biol. Evol. Advance Access doi:10.1093/molbev/msi11.Google Scholar] compared results using the yeast with those obtained using the early branching animal Hydra magnipapillata as the outgroup. For these analyses they used just the three main animal models — fly, worm and vertebrate — with the addition of sequences from flatworms. In common with previous studies Philippe et al. [7Philippe, H., Lartillot, N., and Brinkmann, H. (2005). Multigene analyses of bilaterian animals corroborate the monophyly of Ecdysozoa, Lophotrochozoa and Protostomia. Mol. Biol. Evol. Advance Access doi:10.1093/molbev/msi11.Google Scholar] found that, using yeast as an outgroup, nematodes are located at the base of the tree with high statistical support. The flatworms are long branched too, and they are also found at the base of the tree. The change when short-branched Hydra is used instead of yeast is dramatic: both nematodes and flatworms jump up from the root of the tree to a position adjacent to the arthropods, strongly suggesting it was long branch attraction that placed them at the base. But this result is troubling, as there is now an unexpected close association between nematodes — which, as presumed ecdysozoans are appropriately close to the arthropods — and flatworms which, according to the new animal phylogeny ought to be grouped with annelids and molluscs in the Lophotrochozoa. Philippe et al. [7Philippe, H., Lartillot, N., and Brinkmann, H. (2005). Multigene analyses of bilaterian animals corroborate the monophyly of Ecdysozoa, Lophotrochozoa and Protostomia. Mol. Biol. Evol. Advance Access doi:10.1093/molbev/msi11.Google Scholar] explain this nematode–flatworm association as another manifestation of long branch attraction. As mentioned, both are long branched species and, while no longer attracted to the base of the tree thanks to the short branched outgroup, might still be attracted to each other. Philippe et al. [7Philippe, H., Lartillot, N., and Brinkmann, H. (2005). Multigene analyses of bilaterian animals corroborate the monophyly of Ecdysozoa, Lophotrochozoa and Protostomia. Mol. Biol. Evol. Advance Access doi:10.1093/molbev/msi11.Google Scholar] approached this problem in several ways. First, they tested for the suspected mutual attraction between nematode and flatworm by seeing what happens when they remove each in turn. They found that, when flatworms are excluded from analyses, the nematodes remain where they are, adjacent to the arthropods. When the nematodes are removed, however, the flatworms jump across the tree to sit next to the short-branched annelids and molluscs, as predicted by the new animal phylogeny. Their second approach was to select from among their multiple flatworm and nematode sequences those that are slowest evolving, mimicking the approach of the original rRNA gene paper [1Aguinaldo A.A. Turbeville J.M. Linford L.S. Rivera M.C. Garey J.R. Raff R.A. Lake J.A. Evidence for a clade of nematodes, arthropods and other moulting animals.Nature. 1997; 387: 489-493Crossref PubMed Scopus (1216) Google Scholar]. This, albeit with low statistical support, had the same effect; nematode and flatworms are no longer attracted, the nematodes grouping with the arthropods and the flatworms with the annelids and molluscs. Their final experiment was to rank their 146 genes according to the evenness of evolutionary rate — genes with more even rates of evolution can a priori be counted as least susceptible to long branch attraction. To do this they calculated the evolutionary distance between flatworms or nematodes and the outgroup. These distances were compared with the same measure for the vertebrates, echinoderms, annelids and molluscs. The more equal these distances, the more even the rate of evolution. They then repeatedly constructed phylogenies using concatenations of their genes, each time removing a few genes in order of inequality of rates, starting with the most unequal. They found that, as the uneven genes are gradually discarded, the support steadily increases for the groupings of Ecdysozoa (nematodes plus arthropods) and Lophotrochozoa (flatworms plus annelids/molluscs). Each of the experiments described is designed to counteract long branch attraction, and in each case the tree converges on the new animal phylogeny. This result demonstrates the great strength of broadly sampled data sets, and implies that multigene results have indeed been compromised by the affects of long branch attraction. However, although their method seems to improve matters, some aspects of their tree remain puzzling from a biological point of view: the placement of sea squirts rather than amphioxus next to vertebrates is certainly unorthodox, and the sistergroup relationship between tardigrades and nematodes smacks of long branch attraction. One other possible approach for avoiding long branch attraction that has provided additional recent support for the grouping of arthropods with nematodes rather than with deuterostomes is the use of so-called 'rare genomic changes' whose characteristics can be guessed from their name. Any heritable character that has a very low rate of change is relatively immune to long branch attraction effects. This immunity stems from the fact that, because any given character change is very rare, the likelihood of the same change happening independently in unrelated long branches — the ultimate cause of long branch attraction — is even rarer. We have used the gain and loss of gene orthologs as such putatively slowly evolving characters and found some evidence supporting Ecdysozoa over Coelomata [10Copley R.R. Aloy P. Russell R.B. Telford M.J. Systematic searches for molecular synapomorphies in model metazoan genomes give some support for Ecdysozoa after accounting for idiosyncratic nematode.Evol. Dev. 2004; 6: 164-169Crossref PubMed Scopus (58) Google Scholar]. More unequivocal is a recent analysis [11Roy S.W. Gilbert W. Resolution of a deep animal divergence by the pattern of intron conservation.Proc. Natl. Acad. Sci. USA. 2005; 102: 4403-4408Crossref PubMed Scopus (49) Google Scholar] of the gain and loss of introns in arthropods, nematodes and deuterostomes that, like the study of Philippe et al. [7Philippe, H., Lartillot, N., and Brinkmann, H. (2005). Multigene analyses of bilaterian animals corroborate the monophyly of Ecdysozoa, Lophotrochozoa and Protostomia. Mol. Biol. Evol. Advance Access doi:10.1093/molbev/msi11.Google Scholar], gives strong support for the Ecdysozoa and hence for the new animal phylogeny. The biggest problem that some may have with the work of Philippe et al. [7Philippe, H., Lartillot, N., and Brinkmann, H. (2005). Multigene analyses of bilaterian animals corroborate the monophyly of Ecdysozoa, Lophotrochozoa and Protostomia. Mol. Biol. Evol. Advance Access doi:10.1093/molbev/msi11.Google Scholar] will be the worrying levels of missing data. Of their 49 species, only three have sequences from all sampled genes, and on average 35% of the 146 genes are missing. In their defence, the same group [12Philippe H. Snell E.A. Bapteste E. Lopez P. Holland P.W.H. Casane D. Phylogenomics of eukaryotes: impact of missing data on large alignments.Mol. Biol. Evol. 2004; 21: 1740-1752Crossref PubMed Scopus (316) Google Scholar] showed previously that such analyses are unaffected by this level of missing data, but confirmation of their results awaits a few judiciously chosen new genome sequences. As far as it goes, this work makes a significant dent in the question of the relationships of the metazoan phyla, but although eight animal phyla have been related more or less convincingly, roughly 20 phyla remain unplaced. As phylogenetic considerations rarely seem to drive the choice of genomes for sequencing, this EST approach, whatever its limitations, may be the best route to a completely resolved tree of all animal phyla sooner, rather than later.