Letter to the Editor

Toward the Monophyly of Haeckel's : 18S rRNA Environmental Data Support the Sisterhood of Polycystinea and Puri®cacioÂn LoÂpez-GarcõÂa,*² Francisco RodrõÂguez-Valera,² and David Moreira* *Universite Pierre et Marie Curie, Paris and ²DivisioÂn de MicrobiologõÂa, Universidad Miguel HernaÂndez, San Juan de Alicante, Spain

subsequently constructed a cosmid genomic library of the 0.2- to 5-␮m biomass fraction from 500-m-deep wa- In 1887, Ernst Haeckel published a monumental ters at the same location (59Њ19ЈS, 55Њ45ЈW). The con- and amazingly illustrated monograph describing several struction of the environmental genomic libraries has a thousand different radiolarian species, which had been number of advantages over a direct PCR ampli®cation collected during the 4-year journey (1872±1876) of the of a target gene. First, the diversity retrieved is not af- British oceanographic corvette H.M.S. Challenger fected by well-known PCR-related biases. Second, these (Haeckel 1887). Radiolaria consist of diverse marine kinds of libraries offer the possibility of retrieving other planktonic protists, mostly unicellular, usually endowed gene sequences, in addition to the 18S rRNA, from the with complex, conspicuous mineral skeletons. Haeckel same genomic fragment. applied the term Radiolaria to three different groups: Our environmental genomic library was construct- Acantharea, Polycystinea (Spumellaria and Nassellaria), ed using 3.5 ␮g of DNA. DNA was mechanically bro- and . All of them are united by the posses- ken by intense shearing and fractionated in a 0.8% low sion of a central capsule de®ning an intracapsular and melting point agarose gel. The band corresponding to an extracapsular region in the cytoplasm, and some of 35±45 kb DNA fragments was cut off and digested with them, (all Acantharea and several Spumellaria), also by gelase. After concentration, the DNA was cloned in the the ability to secrete sulfate (SrSO ) (Ander- 4 8,179-bp cosmid vector of the pWEB Cloning System son 1983). This led some contemporary authors to clas- (Epicentre Technologies). A total of 6,107 cosmid sify, as Haeckel did one century ago, Acantharea, Po- clones were PCR-screened for the presence of eukary- lycystinea, and Phaeodarea within a common group otic 18S rRNA genes using the eukaryotic speci®c prim- (Cavalier-Smith 1987). Nevertheless, a recent study ers EK-82F (GAAACTGCGAATGGCTC) and EK- based on the phylogenetic analysis of small subunit ri- 1520R (CYGCAGGTTCACCTAC). Three cosmids bosomal RNA (18S rRNA) sequences challenged this were positive, and the respective 18S rRNA genes were view and supported the early emergence of completely sequenced with previously described internal before Acantharea (Zettler, Sogin, and Caron 1997). primers (LoÂpez-GarcõÂa et al. 2001). One of them cor- This would render Haeckel's Radiolaria polyphyletic responded to an alveolate sequence belonging to the re- and inappropriate as a taxonomic entity. We present here cently described marine Alveolate Group I (LoÂpez- a phylogenetic analysis, including new Polycystinea- GarcõÂa et al. 2001) (not shown). The other two were and Acantharea-related 18S rRNA sequences retrieved related to Radiolaria: exhaustive phylogenetic recon- from an oceanic environmental genomic library (500 m struction using all the available 18S rRNA sequences depth, Antarctic Polar Front). Our study strongly sup- showed that HA2 branched at the base of the Polycys- ports the monophyly of both groups, which constitutes tinea and KW16 at the base of the Acantharea, both with a ®rst piece of evidence for the legitimacy of Radiolaria strong statistical support (®g. 1). sensu Haeckel as a valid taxon. Besides its basal position, HA2 displays a branch During a recent study on the 18S rRNA gene±based signi®cantly shorter than that of its close polycystinean diversity of small protists (0.2±5 ␮m fraction) inhabiting relatives (®g. 1). The possibility that Polycystinea is a deep Antarctic waters (250±3,000 m depth), we re- trieved a sequence close to the Acantharea (DH147- fast-evolving lineage has already been advanced (Zettler, EKD17, see ®g. 1) from 2,000 m depth (LoÂpez-GarcõÂa Sogin, and Caron 1997; Cavalier-Smith 1999). In fact, et al. 2001). Three additional acantharean environmental Polycystinea 18S rRNA sequences display very long sequences were determined from surface waters (75 m branches and a long basal, unbroken, branch (Zettler, depth) in the equatorial Paci®c by other authors (Moon- Anderson, and Caron 1999). Therefore, HA2 sequence van der Staay, De Wachter, and Vaulot 2001). To com- was potentially very useful to break the long Polycys- plement our previous PCR-based study avoiding possi- tinea branch that, given its accelerated evolutionary rate, ble ampli®cation-induced biases (Wintzingerode, GoÈbel, could be affected by phylogenetic reconstruction prob- and Stackerbrandt 1997; Polz and Cavanaugh 1998), we lems, such as the well-known long-branch attraction (LBA) artifact (Felsenstein 1978). KW16 could repre- sent a similar situation within the Acantharea, although Key words: Radiolaria, Haeckel, molecular ecology, 18S rRNA, phylogeny. these display shorter branches than the Polycystinea. A close inspection of the 18S rRNA alignment already Address for correspondence and reprints: David Moreira, Univ- ersite Pierre et Marie Curie, UMR 7622, 9 quai St Bernard, 75005 showed that both sequences had signatures common to Paris, France. E-mail: [email protected]. Polycystinea and Acantharea, a trend most remarkable Mol. Biol. Evol. 19(1):118±121. 2002 for HA2 (not shown). Indeed, the addition of HA2 and ᭧ 2002 by the Society for Molecular Biology and Evolution. ISSN: 0737-4038 KW16 sequences to the eukaryotic 18S rRNA data set

118 Toward the Monophyly of Haeckel's Radiolaria 119

had very important effects on tree reconstruction. With- out them Polycystinea still branched before Acantharea, even with a larger taxonomic sampling for both groups than that used in previous studies (Zettler, Sogin, and Caron 1997) (not shown). When incorporated into the data set, Polycystinea and Acantharea emerged as sister groups (that is, the monophyly of two of the three groups of Radiolaria was retrieved) for all distance (neighbor-joining [NJ] with the Tamura-Nei [Tamura and Nei 1993] model of sequence evolution), maximum parsimony (MP, with heuristic search), and maximum likelihood (ML) phylogenetic reconstruction methods (see ®g. 2 for a ML ). The statistical support for this node, measured as bootstrap proportions FIG. 1.ÐMaximum likelihood (ML) phylogenetic tree of all avail- (BP), was medium to very high depending on the re- able Acantharea- and Polycystinea-related 18S rRNA sequences. Two new sequences, HA2 and KW16 (grey branches), were obtained from construction method employed. In fact, for the methods an environmental genomic DNA library of 500 m depth Antarctic more prone to LBA, NJ, and MP, the BP were 58% and . OLI sequences and DH147-EKD17 are PCR-ampli®ed ma- 79%, respectively, while for ML, generally the most ro- rine environmental sequences from 75 and 2,000 m depth, respectively bust method (Hasegawa and Fujiwara 1993), the BP rose (LoÂpez-GarcõÂa et al. 2001, Moon-van der Staay, De Wachter, and Vau- lot 2001). Eight sequences from heterokonts, alveolates, plants, and up to 92% (®g. 2). This strongly suggested that LBA fungi were used as the outgroup (grey triangle). The tree was con- was responsible for an artifactual early emergence of structed using a heuristic search (options Ϫq Ϫn 2500) with the MOL- Polycystinea in previous studies. Nevertheless, ML PHY 2.3 package (Adachi and Hasegawa 1996) and the branch lengths bootstrap values, calculated using the RELL method for the best tree were recalculated using a ⌫ law (eight rate classes ϩ invariant sites, ␣ϭ1.06) with the program PUZZLE (Strimmer and (Kishino, Miyata, and Hasegawa 1990), might be over- von Haeseler 1996). Numbers at the nodes are ML bootstrap values estimated. However, true bootstrap values calculated us- estimated with the RELL method (Kishino, Miyata, and Hasegawa ing ML with a more sophisticated and realistic model 1990) upon the 2,500 top-ranking trees using the MOLPHY 2.3 pack- of sequence evolution (accounting for among-site rate age. Bootstrap values concerning the position of the two new sequenc- es HA2 and KW16 are in bold. The scale bar corresponds to 10 sub- variation with a ⌫ law [Yang 1996]) still increased the stitutions per 100 positions for a unit branch length. support for the monophyly of Radiolaria, up to 96% (®g. 2). The monophyly of Polycystinea and Acantharea ap- pears stable as it was retrieved using different combi- nations of species. For instance, the inclusion of the

FIG. 2.ÐML phylogenetic tree of eukaryotic 18S rRNA sequences showing the monophyly of Radiolaria. The tree was obtained by comparing 1,250 unambiguously aligned positions, applying the method described in ®gure 1 with an ␣ parameter of 0.39. Bootstrap values are shown only for the radiolarian nodes. For the internal nodes, only ML bootstrap values are indicated. For the node concerning the monophyly of Acantharea and Polycystinea, bootstrap values correspond from top to bottom to: ML, ML applying a ⌫ law, maximum parsimony (MP), and Neighbor-Joining (NJ). Bootstraps were calculated as described in ®gure 1 legend for ML values and, for the rest, from 100 replicates using the programs PUZZLEBOOT (Holder and Roger 1999) and PUZZLE (Strimmer and von Haeseler 1996) (⌫ law-ML), and from 1,000 replicates using PAUP 3.1 (Swofford 1993) (MP), and the MUST package (Philippe 1993) (NJ). Triangles correspond to groups of two sequences. The scale bar corresponds to 5 substitutions per 100 positions for a unit branch length. 120 LoÂpez-GarcõÂa et al.

Mastigamoeba balamuthi sequence in our analyses did should be reinterpreted in the light of this newly vali- not affect the tree topology or the statistical support (®g. dated phylogenetic framework. 2) in contrast to previous analyses (Zettler, Sogin, and Caron 1997). In addition, when HA2 was included as Acknowledgments the only representative for the Polycystinea group, all This work was supported by the European MIDAS reconstruction methods retrieved its sisterhood to the project. The HeÂsperides campaign He052 was ®nanced Acantharea with a BP of 100% (not shown). by the Spanish Research Council (CSIC). We thank Tom As our sequences were obtained directly from en- Cavalier-Smith for critical reading of the manuscript. vironmental samples, without classical identi®cation and Sequences have been deposited in GenBank under ac- characterization of the corresponding microorganisms, it cession numbers AF382824 and AF382825. could be argued that they do not belong to the true Ra- diolaria, so our results are not an evidence for the mono- LITERATURE CITED phyly of this group. Nevertheless, given the topology of the phylogenetic tree obtained, a simple application of ADACHI, J., and M. HASEGAWA. 1996. MOLPHY Version 2.3: programs for molecular phylogenetics based on maximum basic cladistic rules strongly suggests that these new or- likelihood. Comput. Sci. Monogr. 28:1±150. ganisms should display several radiolarian features, in AGUINALDO, A. M., J. M. TURBEVILLE,L.S.LINFORD,M.C. particular the possession of central capsule and axopo- RIVERA,J.R.GAREY,R.A.RAFF, and J. A. LAKE. 1997. dial extracapsular network, and perhaps even the ability Evidence for a clade of nematodes, arthropods and other to secrete SrSO4. We cannot exclude the possibility of moulting animals. Nature 387:489±493. HA2 being a member of the Phaeodarea, given that se- ANDERSON, R. O. 1983. Radiolaria. Springer, New York. quences from known species of this group are not avail- BALDAUF, S. L., A. J. ROGER,I.WENK-SIEFERT, and W. F. able. However, this does not invalidate our conclusion, DOOLITTLE. 2000. A -level phylogeny of eukary- otes based on combined protein data. Science 290:972±977. as Polycystinea and Acantharea would still be members CAVALIER-SMITH, T. 1987. The origin of eukaryotic and ar- of a same monophyletic group. Furthermore, if HA2 chaebacterial cells. Ann. N. Y. Acad. Sci. 503:17±54. turns out to be a phaeodarean, this would imply the ÐÐÐ. 1999. Principles of protein and lipid targeting in sec- monophyly of Haeckel's Radiolaria sensu stricto. ondary symbiogenesis: euglenoid, dino¯agellate, and spo- In any case, HA2 and KW16 have a strong phy- rozoan plastids origin and the family tree. J. Eu- logenetic impact, bringing together Polycystinea and karyot. Microbiol. 46:347±366. Acantharea with strong statistical support. Actually, the FELSENSTEIN, J. 1978. Cases in which parsimony or compati- use of environmental sequences has already proved use- bility methods will be positively misleading. Syst. Zool. 27: ful to stabilize other problematic branches of the eu- 401±410. HAECKEL, E. 1887. Report on radiolaria collected by H. M. S. karyotic tree (Moreira, LoÂpez-GarcõÂa, and RodrõÂguez- Challenger during the years 1873±1876. Rep. Sci. Res. Voy- Valera 2001). At present, there are three major strategies age H.M.S. Challenger 1873±1876 18:1±1803. to solve these unstable regions of the eukaryotic tree, HASEGAWA, M., and M. FUJIWARA. 1993. Relative ef®ciencies namely, the retention of good supported relationships of the maximum likelihood, maximum parsimony, and retrieved from the analyses of different markers, the use neighbor-joining methods for estimating protein phylogeny. of large multigene fusions (Baldauf et al. 2000; Moreira, Mol. Phylogenet. Evol. 2:1±5. Le Guyader, and Philippe 2000), and the use of repre- HOLDER, A., and M. ROGER. 1999. PUZZLEBOOT. Marine sentative slow-evolving sequences (Aguinaldo et al. Biological Laboratory, Woods Hole, Mass. KISHINO, H., T. MIYATA, and M. HASEGAWA. 1990. Maximum 1997). Whereas massive genome sequencing will pro- likelihood inference of protein phylogeny, and the origin of vide the necessary data for comparative phylogeny and chloroplasts. J. Mol. Evol. 31:151±160. gene-fusion analyses, molecular ecology can be an in- KNOLL, A. H. 1992. The early evolution of : a geo- valuable source of slow-evolving genes. logical perspective. Science 256:622±627. The recovered monophyly of Polycystinea and LOÂ PEZ-GARCÂõA, P., F. RODRÂõGUEZ-VALERA,C.PEDROÂ S-ALIOÂ , Acantharea has important consequences. From an evo- and D. MOREIRA. 2001. Unexpected diversity of small eu- lutionary point of view, Polycystinea do not appear to karyotes in deep-sea Antarctic plankton. Nature 409:603± be an ancient lineage as previously inferred from certain 607. MOON-VAN DER STAAY, S. Y., R. DE WACHTER, and D. VAU- phylogenetic analyses (Zettler, Sogin, and Caron 1997) LOT. 2001. Oceanic 18S rDNA sequences from picoplank- but, instead, a fast-evolving branch emerging together ton reveal unsuspected eukaryotic diversity. Nature 409: with Acantharea in the apical region of the eukaryotic 607±610. tree (the so-called crown [Knoll 1992]). Polycystinea MOREIRA, D., H. LE GUYADER, and H. PHILIPPE. 2000. The have also, within Radiolaria, left a vast fossil record origin of red algae and the evolution of chloroplasts. Nature (Anderson 1983). Polycystinean siliceous skeletons are 405:69±72. MOREIRA, D., P. LOÂ PEZ-GARCIÂA, and F. RODRIÂGUEZ-VALERA. easily fossilizable, in contrast to the very soluble SrSO4 skeletons of acanthareans, and have recorded the history 2001. New insights on the phylogenetic position of diplo- of some environments on Earth (in particular the deep nemids: environmental sequences, GC content bias and dif- ferences of evolutionary rate. Int. J. Syst. Evol. Microbiol. sea, as calcareous fossils are rare at very high depth) 51:2211±2219. and paleoclimate (as climatic oscillations induce alter- PHILIPPE, H. 1993. MUST, a computer package of management ations in the morphology of the skeletons) (Steineck and utilities for sequences and trees. Nucleic Acids Res. 21: Casey 1990). Accordingly, the Radiolaria fossil record 5264±5272. Toward the Monophyly of Haeckel's Radiolaria 121

POLZ, M. F., and C. M. CAVANAUGH. 1998. Bias in template- WINTZINGERODE, F., U. B. GOÈ BEL, and E. STACKEBRANDT. to-product ratios in multitemplate PCR. Appl. Environ. Mi- 1997. Determination if microbial diversity in environmental crobiol. 64:3724±3730. samples: pitfalls of PCR-based analysis. FEMS Microbiol. STEINECK, P. L., and R. E. CASEY. 1990. Ecology and paleo- Rev. 21:213±229. biology of and Radiolaria. Pp. 89±138 in G. YANG, Z. 1996. Among-site rate variation and its impact on M. Capriulo, ed. Ecology of marine . Oxford Uni- phylogenetic analyses. Trends Ecol. Evol. 11:367±370. versity Press, Oxford. ZETTLER, L. A., O. R. ANDERSON, and D. A. CARON. 1999. STRIMMER, K., and A. VON HAESELER. 1996. Quartet puzzling: Towards a molecular phylogeny of colonial spumellarian a quartet maximum likelihood method for reconstructing radiolaria. Mar. Micropaleontol. 36:67±79. tree topologies. Mol. Biol. Evol. 13:964±969. ZETTLER, L. A., M. L. SOGIN, and D. A. CARON. 1997. Phy- SWOFFORD, D. L. 1993. PAUP: phylogenetic analysis using logenetic relationships between the Acantharea and the Po- parsimony. Version 3.1.1. Illinois Natural History Survey, lycystinea: a molecular perspective on Haeckel's Radiolaria. Champaign, Illinois. Proc. Natl. Acad. Sci. USA 94:11411±11416. TAMURA, K., and M. NEI. 1993. Estimation of the number of nucleotide substitutions in the control region of mitochon- GEOFFREY MCFADDEN, reviewing editor drial DNA in humans and chimpanzees. Mol. Biol. Evol. 10:512±526. Accepted August 27, 2001