The Closest Unicellular Relatives of Animals

View metadata, citation and similar papers at core.ac.uk brought to you by CORE

provided by Elsevier - Publisher Connector Current Biology, Vol. 12, 1773–1778, October 15, 2002, 2002 Elsevier Science Ltd. All rights reserved. PII S0960-9822(02)01187-9 The Closest Unicellular Relatives of Animals

B.F. Lang,1,2 C. O’Kelly,1,3 T. Nerad,4 M.W. Gray,1,5 Results and Discussion and G. Burger1,2,6 1The Canadian Institute for Advanced Research The evolution of the Metazoa from single-celled protists Program in Evolutionary Biology is an issue that has intrigued biologists for more than a 2 De´ partement de Biochimie century. Early morphological and more recent ultra- Universite´ de Montre´ al structural and molecular studies have converged in sup- Succursale Centre-Ville porting the now widely accepted view that animals are Montre´ al, Que´ bec H3C 3J7 related to Fungi, choanoflagellates, and ichthyosporean Canada protists. However, controversy persists as to the spe- 3 Bigelow Laboratory for Ocean Sciences cific evolutionary relationships among these major P.O. Box 475 groups. This uncertainty is reflected in the plethora of 180 McKown Point Road published molecular phylogenies that propose virtually West Boothbay Harbor, Maine 04575 all of the possible alternative tree topologies involving 4 American Type Culture Collection Choanoflagellata, Fungi, Ichthyosporea, and Metazoa. 10801 University Boulevard For example, a monophyletic MetazoaϩChoanoflagel- Manassas, Virginia 20110 lata group has been suggested on the basis of small 5 Department of Biochemistry subunit (SSU) rDNA sequences [1, 6, 7]. Other studies and Molecular Biology using the same sequences have allied Choanoflagellata Dalhousie University with the Fungi [8], placed Choanoflagellata prior to the Halifax, Nova Scotia B3H 4H7 divergence of animals and Fungi [9], or even placed Canada them prior to the divergence of green algae and land plants [10]. Moreover, Ichthyosporea [11], a newly cre- ated taxon that was provisionally designated DRIPs (re- ferring to the four initial members, Dermocystidium, ro- Summary sette agent, Ichthyophonus, and Psorospermium [6]) and later Mesomycetozoa [12], has tentatively been Molecular phylogenies support a common ancestry placed somewhere near [12] (in one case, immediately between animals (Metazoa) and Fungi [1–3], but the before [6]) the animal-fungal divergence. Testifying to evolutionary descent of the Metazoa from single-celled the taxonomic uncertainty surrounding the Ichthyo- eukaryotes (protists) and the nature and taxonomic affili- sporea, one class of this phylum, i.e., the Amoebidiales, ation of these ancestral protists remain elusive. We ad- had traditionally been classified as trichomycete fungi. dressed this question by sequencing complete mito- Conflicting scenarios as to the relationship among chondrial genomes from taxonomically diverse protists Choanoflagellata, Ichthyosporea, and animals were to generate a large body of molecular data for phyloge- critically addressed in the course of a recent analysis netic analyses. Trees inferred from multiple concate- using complete SSU and large subunit (LSU) rRNA data, nated mitochondrial protein sequences demonstrate examined either individually or in combination [13]. This that animals are specifically affiliated with two mor- particular study demonstrates the weakness of tree as- phologically dissimilar unicellular protist taxa: Mono- sessments that are uniquely based on nonparametric siga brevicollis (Choanoflagellata), a flagellate, and bootstrap values and lends credence to the view that Amoebidium parasiticum (Ichthyosporea), a fungus- the precise interpretation of such values is not only diffi- like organism. Statistical evaluation of competing evo- cult [14] but also often leads to overconfidence in the lutionary hypotheses [4] confirms beyond a doubt that wrong tree [4]. For example, the monophyly of Choano- Choanoflagellata and multicellular animals share a flagellataϩIchthyosporea is supported by a high boot- close sister group relationship, originally proposed strap value (94%) based on the LSU data and by more than a century ago on morphological grounds a marginal bootstrap value (61%) in the combined [5]. For the first time, our trees convincingly resolve LSUϩSSU data set, whereas the Kishino-Hasegawa the currently controversial phylogenetic position of (KH) and Shimodaira-Hasegawa (SH) tests did not re- the Ichthyosporea, which the trees place basal to Cho- cover support for this topology with any combination of anoflagellata and Metazoa but after the divergence of data used [13]. Based upon the highly conflicting results, Fungi. Considering these results, we propose the new the authors come to the conclusion that the available taxonomic group Holozoa, comprising Ichthyosporea, data may be insufficient to resolve the question of Choanoflagellata, and Metazoa. Our findings provide whether Choanoflagellata, Ichthyosporea, or the two insight into the nature of the animal ancestor and have combined are the closest living relatives of Metazoa. broad implications for our understanding of the evolu- Basal animal phylogeny has also been investigated via tionary transition from unicellular protists to multicel- single nucleus-encoded proteins, but again, the support lular animals. for this deep divergence relies solely on the interpretation of often weak and variable bootstrap or quartet- puzzle support values. For instance, trees based on 6 Correspondence: [email protected] Hsp70 proteins [15] indicate a closer relationship of Current Biology 1774

Figure 1. Maximum Likelihood Tree of Con- catenated Proteins Encoded by mtDNA The sequences of 11 well-conserved proteins (Cox1,2,3, Cob, Atp6,9, and Nad1,3,4,4L,5) were concatenated. A ⌫ distribution model of site variation was used (see Experimental Procedures). Percent bootstrap support for PROML (300 samples) is shown above each branch, and that for PUZZLEBOOT/BIONJ (1000 samples) is shown below each branch. The scale bar denotes genetic distance. Taxon designations are as follows (GenBank Accession numbers within parentheses): Magnetospirillum magnetotacticum (NC_002725); Rickettsia prowazekii (NC_000963); Chrysodidymus synuroideus (NC_002174); Phytophthora infestans (NC_002387); Hyaloraphidium curvatum (NC_003048); Spizellomyces punctatus (NC_003052); Schizophyllum commune (NC_003049); Podospora anserina (NC_001329); Rhizopus stolonifer (unpublished; see Supplementary Material); Allo- myces macrogynus (NC_001715); Sarcophy- ton glaucum (AF064823, AF063191); Metridium senile (NC_000933); Homo sapiens (J01415); Monosiga brevicollis (this publication); Prototheca wickerhamii (NC_001613); Marchantia polymorpha (NC_001660); Amoe- bidium parasiticum (this publication); Por- phyra purpurea (NC_002007); and Chondrus crispus (NC_001677).

Choanoflagellata (represented by the taxon Monosiga For phylogenetic analyses, we used our own gener- ovata) to animals than to Fungi. However, the bootstrap ated mitochondrial protein sequences, including data value for this relationship is low (58%) and, as the au- reported here from M. brevicollis and A. parasiticum,as thors state [15], “rigorous statistical tests such as the well as sequences determined by others and retrieved Kishino-Hasegawa test were also carried out but were from public data repositories. The analyses included unable to provide statistical support for any of the alter- about 3000 aligned amino acid positions from 11 well- native trees.” Another recent phylogenetic study used conserved proteins whose sequences were concate- elongation factor 2 (EF-2), ␣- and ␤-tubulin, and actin nated (see Experimental Procedures section for details). proteins [16]. In all four trees, bootstrap and quartet- Figure 1 depicts the tree obtained with the maximum puzzle indices are only shown at selected branches and likelihood (ML) method implemented in PROML [18], are weak in support of either the fungal-animal diver- with site heterogeneity modeled by the discrete ⌫ distri- gence or the branching order within the animals, thus bution. Because the ML method is computationally calling into question the overall topology of these trees. highly demanding, the number of taxa was limited to 20. Finally, in this study among-site rate heterogeneity was Taxon selection was based on two criteria; we excluded not taken into consideration, and no statistical tree se- taxa that either completely lack mitochondrial nad genes lection tests were performed. or that display highly accelerated rates of evolution of We posit that one important reason for the contro- mitochondrial proteins (see Experimental Procedures versy about early animal evolution is that the available for details). Notably, within the Metazoa, essentially all sequence data have been insufficient to yield unambigu- taxa evolve quickly, with the exception of the sea anem- ous resolution of the taxa in question. To provide a one Metridium [19] and the leather coral Sarcophyton suitable data set, we sequenced complete mitochon- [20]. Therefore, only one metazoan (i.e., human) se- drial genomes from diverse protist phyla [17], including quence having a relatively long branch length has been the choanoflagellate Monosiga brevicollis and the ich- included in the analyses. The same data set was also thyosporean Amoebidium parasiticum, and thereby analyzed with an ML distance method (TREE-PUZZLE generated the first mitochondrial gene sequences for and BIONJ [21, 22]), PUZZLEBOOT [23] for bootstrap- Choanoflagellata and Ichthyosporea. With this new data ping, and the same model of site heterogeneity, which set, we addressed two specific questions: which of yielded a tree topology identical to that of the ML recon- these present-day protist groups is on the lineage lead- struction. ing specifically to Metazoa, and which one is the closest Figure 2 shows a tree inferred only with the distance living relative of animals? Phylogenetic analyses of method. Here, many more species could be included these new data are reported here, with the gene content because distance methods are computationally less de- and genome architecture of A. parasiticum and M. brevi- manding than ML analyses. Inclusion of additional ani- collis mtDNAs being presented elsewhere. mals, Fungi, and plants, and the addition of available The Closest Unicellular Relatives of Animals 1775

Figure 2. Distance Method Tree of Concatenated Proteins Encoded by mtDNA The sequences of 11 well-conserved proteins (Cox1,2,3, Cob, Atp6,9, and Nad1,3,4,4L,5) were concatenated. A ⌫ distribution model of site variation was used (see Experimental Procedures). The number at each branch represents bootstrap support (percent) for PUZZLEBOOT (1000 samples). Taxon designations, in addition to those specified in Figure 1, are as follows: Sinorhizobium meliloti (NC_003047); Lumbricus terrestris (NC_001673); Drosophila yakuba (X03240); Branchiostoma lanceolatum (MTY16474); Mus musculus (J01420); Rhizophydium sp. (NC_003053); Aspergillus nidulans (L19866, X00790, V00650; X15441, X15011, X06960, X06961, AH001255, M35967, J01387, J01388, and J01389); Pichia canadensis (NC_001762); Yarrowia lipolytica (NC_002659); Acanthamoeba castellanii (NC_001637); Dictyostelium discoideum (NC_000895); Reclinomonas americana (AF007261); Malawimonas jakobiformis (NC_002553); Mesostigma viride (AF353999); Arabidopsis thaliana (NC_001284); and Nephroselmis olivata (AF110138). data from jakobids, did not change the relevant part of ware programs CONSEL and PAML [24, 25], which pro- the tree topology or the support in the Choanoflagel- vide the least biased and most rigorous tests available lataϩIchthyosporeaϩMetazoa clade, compared to Fig- to date [4, 14]. We tested the significance of each set ure 1. We observed, however, that support values for of competing tree topologies that included Choanofla- the monophyly of red algae and green algaeϩland plants gellata, Ichthyosporea, Metazoa, and Fungi. The results decreased as a consequence of the inclusion of jako- of the standard AU test, the weighted KH test (WKH), bids, an issue that will be addressed in more detail weighted SH test (WSH) [4], and bootstrap probability elsewhere. It should be noted that trees were also con- [18] derived from the data set used in Figure 1 are com- structed with individual, well-conserved proteins (Cob, piled in Table 1. The standard AU and WKH tests confirm Cox1) and with three combinations of several proteins the topology shown in Figure 1, i.e., they reject all alter- (Cob,Atp6,9; Cox1,2,3; Nad1,3,4,4L,5). The phylogenetic native scenarios (at a significance level of 0.05), namely position of A. parasiticum and M. brevicollis was identi- the ChoanoflagellataϩIchthyosporea, Ichthyosporeaϩ cal to the one shown in Figures 1 and 2, with four of Metazoa, IchthyosporeaϩFungi, and Choanoflagellataϩ these five data sets. However, the resulting trees did Fungi sister relationships. It should be noted that for the not yield significant support in favor of or against the tree including H. sapiens, the WSH test does not reject topology that was obtained with the full set of concate- (at the given significance level) the hypothesis that Ich- nated proteins. thyosporea are basal to Fungi and Metazoa (Table 1, When we used the concatenated data sets, both ML topology #3). However, this topology is unambiguously and distance approaches yielded identical tree topolog- rejected when H. sapiens is excluded from the data set ies with branches supported by robust (Ͼ90%) boot- (see Supplementary Material). We attribute this differ- strap indices. To assess the level of confidence in tree ence to the fact that H. sapiens displays the longest selection, we performed statistical tests with the soft- branch, and long branches are notoriously difficult to Current Biology 1776

Table 1. Likelihood Tests of Alternative Tree Topologies Tree Topology Ta AUb BPc WKHd WSHe #1 Best tree (Figure 1) Ϫ30.6 0.997 0.980 0.980 1.000 #2 M.b. basal to Metazoa/Fungi 30.6 0.003 0.003 0.006 0.020 #3 A.p. prior to Metazoa/Fungi 32.8 0.017 0.013 0.020 0.083f #4 M.b. and A.p. member of Fungi 33.0 0.004 0.002 0.005 0.017 #5 A.p. and M.b. sister taxa 37.2 0.001 0.000 0.001 0.005 #6 A.p. together with Fungi 40.4 0.007 0.002 0.004 0.018 #7 M.b. member of Fungi 57.2 0.001 0.000 0.001 0.002 #8 A.p. and M.b. basal to Metridium/Sarcophyton 140.2 0.001 0.000 0.000 0.000 #9 M.b. basal to Metridium/Sarcophyton 144.3 0.000 0.000 0.000 0.000 #10 A.p. sister to Metazoa; M.b. basal to A.p. 151.5 0.000 0.005 0.000 0.000

The program CONSEL [24] was employed, and the same factor for the discrete ⌫ distribution was used as in Figures 1 and 2. a Log likelihood difference. b Standard approximately unbiased test. c Bootstrap probability. d Weighted Kishino-Hasegawa test. e Weighted Shimodaira-Hasegawa test. f This value decreases to 0.032 when the long-branching taxon H. sapiens is removed from the data set.

place with confidence in phylogenetic analyses [26]. We gae), and the Holozoa defined here. Hence, multicellular also emphasize that the developers of the WSH test groups have a phylogenetically dispersed distribution. regard this test as biased and excessively critical in This fact, together with the existence of disparate tissue assessing the significance of likelihood differences types, developmental strategies, and cell-cell communi- among competing tree topologies, and they recommend cation mechanisms provides increasing evidence that the AU test for general tree testing [4]. multicellularity is a trait that has emerged independently, The results shown in Figures 1 and 2 clearly identify on several occasions, during eukaryotic evolution. M. brevicollis, and by implication the Choanoflagellata, Our results testify to the considerable potential of as a sister taxon to the Metazoa. This result confirms mitochondrial genomics as applied to protistan eukary- the hypothesis that sponges and all other animals otes. This approach not only reveals novel types of mito- evolved from a choanoflagellate-like ancestor, a pro- chondrial genome structure and gene expression (for a posal made as early as 1866 on the basis of the remark- review, see [29]), but it also generates large data sets able morphological similarities between feeding cells that are particularly well suited to resolving the phyloge- (choanocytes) of sponges and choanoflagellate protists netic relationships of deeply diverging eukaryotic lin- [5]. The second important conclusion of our results is eages; such relationships cannot be discerned by sin- that A. parasiticum, and by implication the Ichthyo- gle-gene analyses. sporea [27, 28], emerged prior to animals and choanoflagellates and clearly after the divergence of the fungi. Experimental Procedures Thus, Ichthyosporea, Choanoflagellata, and Metazoa to- Strains and Cultivation gether form a higher-order taxon that we term the Ho- M. brevicollis (ATCC 50154) was obtained from the American Type lozoa. Culture Collection. The organism was grown in batch cultures at approximately 25ЊC on sterile natural seawater and fed with live bacteria (Enterobacter aerogenes ATCC 13048). A. parasiticum JAP- Evolutionary Implications 7-2 was obtained from R.W. Lichtwardt (Department of Botany, With the data and analyses presented here, we are now University of Kansas, Lawrence, KS) and cultured in liquid medium able to provide unambiguous and compelling evidence (1% yeast extract, 3% glycerol) with shaking. that Choanoflagellata, Ichthyosporea, and Metazoa constitute a monophyletic assemblage, the Holozoa, to DNA Extraction, Cloning, and Sequencing the exclusion of Fungi and other eukaryotic groups. Cells of M. brevicollis and A. parasiticum were suspended in sorbitol buffer (0.6 M sorbitol, 5 mM EDTA, 50 mM Tris [pH 7.4]), broken Within the Holozoa, Ichthyosporea diverge basally, mechanically by being shaken with glass beads, and subsequently whereas Choanoflagellata represent the sister taxon to lysed in the presence of 1% SDS and 100 ␮g/ml proteinase K. animals. Because Choanoflagellata and Ichthyosporea SDS was eliminated by NaCl precipitation. Total nucleic acids were are both unicellular protist groups, the specific ances- fractionated by CsCl/Hoechst 33258 dye isopycnic centrifugation, ϩ tors of animals were most likely unicellular organisms whereby mitochondrial DNA forms the uppermost (A T-rich) band, as verified by hybridization of all fractions with a probe including as well. These postulated single-celled ancestors of the cox1 gene. The upper band was recentrifuged to achieve further Metazoa must have given rise to multicellular proto- purification. Random clone libraries were constructed by nebuliza- animals, from which the major extant metazoan lineages tion of the purified mtDNA (into fragment sizes of 1–3 kbp) and (Bilateria, Cnidaria, Ctenophora) subsequently emerged. cloning into pBluescript (Stratagene). The corresponding protocol Including the findings presented here, we now recognize is available [17]. Clones were sequenced by a combination of auto- a total of five major eukaryotic clades that encompass mated Li-Cor and manual methods. both unicellular and multicellular members. These are Sequence Analysis the Fungi, Streptophyta (charophyte algae ϩ land Sequence readings were assembled and proofread with the GAP plants), Rhodophyta (red algae), Phaeophyta (brown al- software suite [30]. The FASTA program [31] was employed for The Closest Unicellular Relatives of Animals 1777

searches in local databases, and the BLAST network service [32] References was employed for similarity searches in GenBank at the National Center for Biotechnology Information. Custom-made batch utilities 1. Wainright, P.O., Hinkle, G., Sogin, M.L., and Stickel, S.K. (1993). used for submitting queries and browsing the results are avail- Monophyletic origins of the metazoa: an evolutionary link with able [17]. fungi. Science 260, 340–342. 2. Baldauf, S.L., and Palmer, J.D. (1993). Animals and fungi are Phylogenetic Analyses each other’s closest relatives: congruent evidence from multiple For phylogenetic analyses, we used a total of 2969 amino acid proteins. Proc. Natl. Acad. Sci. USA 90, 11558–11562. positions from 11 concatenated, well-conserved protein sequences 3. Baldauf, S.L., Roger, A.J., Wenk-Siefert, I., and Doolittle, W.F. (Cox1,2,3, Cob, Atp6,9, and Nad1,3,4,4L,5) that are encoded in (2000). A kingdom-level phylogeny of eukaryotes based on com- mtDNAs of most eukaryotes. Exceptions are Atp9, which is nucleus- bined protein data. Science 290, 972–977. encoded in Metazoa and in Podospora anserina, and Nad1, 2, and 4. Shimodaira, H. (2002). An approximately unbiased test of phylo- 4, for which complete genes in A. parasiticum have not been identi- genetic tree selection. Syst. Biol. 51, 492–508. fied so far. Multiple sequence alignment was performed with CLUS- 5. James-Clark, H. (1866). Note on the infusoria flagellata and the TAL W [33], and only amino acid positions that could be aligned spongiae ciliatae. Am. J. Sci. 1, 113–114. without ambiguity were used in the analysis (the data used for the 6. Ragan, M.A., Goggin, C.L., Cawthorn, R.J., Cerenius, L., phylogenetic analysis will be made available, on request). Phyloge- Jamieson, A.V.C., Plourde, S.M., Rand, T.G., So¨ derha¨ ll, K., and netic inferences employed either the ML method as implemented Gutell, R.R. (1996). A novel clade of protistan parasites near in PROML [18] or a distance approach. For the calculation, we the animal-fungal divergence. Proc. Natl. Acad. Sci. USA 93, used the distance table TREE-PUZZLE [21], which allows a Jin/ 11907–11912. Nei correction for unequal rates of change at different amino acid 7. Kumar, S., and Rzhetsky, A. (1996). Evolutionary relationships positions, and we used BIONJ [22] for tree inference. Taxon sam- of eukaryotic kingdoms. J. Mol. Evol. 42, 183–193. pling was given particular consideration. In both distance and ML 8. Cavalier-Smith, T. (1987). Evolutionary Biology of the Fungi. analyses, we excluded taxa that either completely lack mitochon- Symposium of the British Mycological Society. A.D.M. Rayner, drial nad genes (e.g., Saccharomyces cerevisiae, Schizosaccharo- C.M. Brasier, and D. Moore, eds. (Cambridge, UK: Cambridge myces pombe, Plasmodium falciparum) or have highly accelerated University Press), pp. 339–353. rates of mitochondrial evolution (e.g., ciliates, trypanosomatids, and 9. Van de Peer, Y., and De Wachter, R. (1997). Evolutionary rela- green algae of the Chlamydomonas group as well as Pedinomonas tionships among the eukaryotic crown taxa taking into account minor). In ML analyses, the number of taxa had to be restricted to site-to-site rate variation in 18S rRNA. J. Mol. Evol. 45, 619–630. 20 per analysis; this was necessary because this method is compu- 10. Smothers, J.F., von Dohlen, C.D., Smith, L.H., Jr., and Spall, tationally very demanding, especially when one uses evolutionary R.D. (1994). Molecular evidence that the myxozoan protists are models that permit rate change at all amino acid positions of the metazoans. Science 265, 1719–1721. protein sequences and applies subsequent bootstrap analyses. The 11. Cavalier-Smith, T. (1998). Neomonada and the origin of animals 20 species used to generate the tree shown in Figure 1 were chosen and fungi. In Evolutionary Relationships Among Protozoa, G.H. to include (i) all fungal and animal species with moderate branch Coombs, K. Vickerman, M.A. Sleigh, and A. Warren, eds (Kluwer, length; (ii) one representative of higher animals with a long branch London), pp. 375–407. length (human); (iii) two to three representatives each of plants and 12. Herr, R.A., Ajello, L., Taylor, J.W., Arseculeratne, S.N., and Men- green algae, red algae, and stramenopiles; and (iv) a bacterial out- doza, L. (1999). Phylogenetic analysis of Rhinosporidium see- group with the two ␣-proteobacteria that are most closely related beri’s 18S small-subunit ribosomal DNA groups this pathogen to mitochondria (Rickettsia and Magnetospirillum). In the distance- among members of the protoctistan Mesomycetozoa clade. J. based phylogenetic analysis shown in Figure 2, we included addi- Clin. Microbiol. 37, 2750–2754. tional data from ␣-proteobacteria and all mitochondrial data that 13. Medina, M., Collins, A.G., Silberman, J.D., and Sogin, M.L. satisfy the requirements specified above, except that the number (2001). Evaluating hypotheses of basal animal phylogeny using of metazoan taxa, which exceeds by far that from all other phyla, complete sequences of large and small subunit rRNA. Proc. was reduced to five. Bootstrap analysis was applied to both PROML Natl. Acad. Sci. USA 98, 9707–9712. and TREE-PUZZLE/BIONJ trees. For likelihood tests, P values were 14. Whelan, S., Lio, P., and Goldman, N. (2001). Molecular phyloge- calculated by the CONSEL software [24], and log-likelihood values netics: state of the art methods for looking into the past. Trends were calculated by the PAML package [25]. A ⌫ distribution model Genet. 17, 262–272. of site variation with an ␣ factor of 0.7, eight categories, and the 15. Snell, E.A., Furlong, R.F., and Holland, P.W.H. (2001). Hsp70 JTT matrix were used [34]. sequences indicate that choanoflagellates are closely related to animals. Curr. Biol. 11, 967–970. Supplementary Material 16. King, N., and Carroll, S.B. (2001). A receptor tyrosine kinase The ML tree excluding humans, a table showing the corresponding from choanoflagellates: molecular insights into early animal results of likelihood tests, and the unpublished protein sequences evolution. Proc. Natl. Acad. Sci. USA 98, 15032–15037. of Rhizopus stolonifer used in these analyses are available at http:// 17. The Organelle Genome Megasequencing Program (OGMP). images.cellpress.com/supmat/supmatin.htm http://megasun.bch.umontreal.ca/ogmp/ 18. Felsenstein, J. (2001). PHYLIP version 3.6a2. (Seattle, WA: De- Acknowledgments partment of Genetics, University of Washington, Seattle). 19. Beagley, C.T., Okimoto, R., and Wolstenholme, D.R. (1998). The We thank R.W. Lichtwardt (Department of Botany, University of mitochondrial genome of the sea anemone Metridium senile Kansas, Lawrence, KS) for supplying an axenic culture of A. parasit- (Cnidaria): introns, a paucity of tRNA genes, and a near-standard icum, L. Forget and I. Plante for clone library construction, and Z. genetic code. Genetics 148, 1091–1108. Wang, Y. Zhu, and S. Cagna for DNA sequencing. We also thank the 20. Beaton, M.J., Roger, A.J., and Cavalier-Smith, T. (1998). Se- anonymous referees of the manuscript for constructive suggestions. quence analysis of the mitochondrial genome of Sarcophyton This project was supported by a grant from the Canadian Institutes glaucum: conserved gene order among octocorals. J. Mol. Evol. for Health Research and equipment grants from Sun Microsystems 47, 697–708. (Palo Alto, CA) and Li-Cor (Lincoln, NE). Salary and interaction sup- 21. Strimmer, K., and von Haeseler, A. (1996). Quartet puzzling: port from the Canadian Institute for Advanced Research to G.B., a quartet maximum-likelihood method for reconstructing tree M.W.G., C.J.O., and B.F.L. is gratefully acknowledged. topologies. Mol. Biol. Evol. 13, 964–969. 22. Gascuel, O. (1997). BIONJ: an improved version of the NJ algo- Received: July 1, 2002 rithm based on a simple model of sequence data. Mol. Biol. Revised: August 19, 2002 Evol. 14, 685–695. Accepted: August 19, 2002 23. Holder, M., and Roger, A.J. PUZZLEBOOT version 1.03. http:// Published: October 15, 2002 hades.biochem.dal.ca/Rogerlab/Software/software.html. Current Biology 1778

24. Shimodaira, H., and Hasegawa, M. (2001). CONSEL: for assessing the confidence of phylogenetic tree selection. Bioinfor- matics 17, 1246–1247. 25. Yang, Z. (1997). PAML: a program package for phylogenetic analysis by maximum likelihood. Comput. Appl. Biosci. 13, 555–556. 26. Felsenstein, J. (1978). Cases in which parsimony or compatibil- ity methods will be positively misleading. Syst. Zool. 20, 401–410. 27. Ustinova, I., Krienitz, L., and Huss, V.A. (2000). Hyaloraphidium curvatum is not a green alga, but a lower fungus: Amoebidium parasiticum is not a fungus, but a member of the DRIPs. Protist 151, 253–262. 28. Benny, G.L., and O’Donnell, K. (2000). Amoebidium parasiticum is a protozoan, not a Trichomycete. Mycologia 92, 1133–1137. 29. Lang, B.F., Gray, M.W., and Burger, G. (1999). Mitochondrial genome evolution and the origin of eukaryotes. Annu. Rev. Genet. 33, 351–397. 30. Staden, R. (1996). The Staden sequence analysis package. Mol. Biotechnol. 5, 233–241. 31. Pearson, W.R. (1990). Rapid and sensitive sequence compari- son with FASTP and FASTA. Methods Enzymol. 183, 63–98. 32. Altschul, S.F., Gish, W., Miller, W., Myers, E.W., and Lipman, D.J. (1990). Basic local alignment search tool. J. Mol. Biol. 215, 403–410. 33. Thompson, J.D., Higgins, D.G., and Gibson, T.J. (1994). CLUS- TAL W: improving the sensitivity of progressive multiple sequence alignment through sequence weighting, position-specific gap penalties and weight matrix choice. Nucleic Acids Res. 22, 4673–4680. 34. Jones, D.T., Taylor, W.R., and Thornton, J.M. (1992). The rapid generation of mutation data matrices from protein sequences. Comput. Appl. Biosci. 8, 275–282.

Accession Numbers

The GenBank accession numbers for the sequences reported in this paper are AF538042-AF538052 (A. parasiticum mtDNA) and AF538053 (M. brevicollis mtDNA).