Analysis of the complete mitochondrial DNA sequence of the brachiopod Terebratulina retusa places Brachiopoda within the protostomes

Analysis of the complete mitochondrial DNA sequence of the brachiopod Terebratulina retusa places Brachiopoda within the protostomes Alexandra Stechmann * and Martin Schlegel UniversitÌt Leipzig, Institut fïr Zoologie/Spezielle Zoologie,Talstr. 33, 04103 Leipzig, Germany Brachiopod phylogeny is still a controversial subject. Analyses using nuclear 18SrRNA and mitochondrial 12SrDNA sequences place them within the protostomes but some recent interpretations of morphological data support a relationship with deuterostomes. In order to investigate brachiopod a nities within the metazoa further,we compared the gene arrangement on the brachiopod mitochondrial genome with several metazoan taxa. The complete (15 451bp) mitochondrial DNA (mtdna) sequence of the articulate brachiopod Terebratulina retusa was determined from two overlapping long polymerase chain reaction products. All the genes are encoded on the same strand and gene order comparisons showed that only one major rearrangement is required to interconvert the T. retusa and Katharina tunicata (Mollusca: Polyplacophora) mitochondrial genomes. The partial mtdna sequence of the prosobranch mollusc Littorina saxatilis shows complete congruence with the T. retusa gene arrangement with regard to the ribosomal and protein coding genes. This high similarity in gene arrangement is the rst to be reported within the protostomes. Sequence analyses of mitochondrial protein coding genes also support a close relationship of the brachiopod with molluscs and annelids,thus supporting the clade Lophotrochozoa. Though being highly informative,sequence analyses of the mitochondrial protein coding genes failed to resolve the branching order within the lophotrochozoa. Keywords: brachiopod phylogeny; mitochondrial genome; gene order; sequence analysis; Lophotrochozoa 1. INTRODUCTION Brachiopods are a group of marine invertebrates comprising around 350 extant and 12 000 fossil species. Their fossil record dates back to the Lower Cambrian. Together with the Phoronida and Bryozoa they have been classi ed as Tentaculata (or Lophophorata) because they possess a similar suspension feeding apparatus,the lophophore,which is a horseshoe-shaped structure surrounding the mouth with ciliated tentacles arranged on it (Emig 1976). Because they show both protostome and deuterostome features,the Lophophorata have been seen as members of the protostomes,as deuterostomes or as intermediates. More recently the lophophorates have been classi ed as deuterostomes on the basis of morphological and larval features (Salvini-Plawen 1982; Emig 1984; Brusca & Brusca 1990; Eernisse et al. 1992; Backeljau et al. 1993; Nielsen 1995). Evidence from the fossil record gave rise to speculations about protostome a nities of the brachiopods. It has been suggested that the setae of brachiopods are derived from the sclerites of the fossil Halkieriids,which are equivalent to the setae of annelids (Conway-Morris 1995,1998; Conway-Morris & Peel 1995). This would mean that brachiopods and annelids * Author for correspondence (stech@rzuni-leipzig.de). share a common ancestry with those fossil Halkieriids, implying an ancestral metamerous body plan for both of them. Segmentation must then have been secondarily lost in brachiopods,but so far there is no clear evidence for any segmentation in brachiopods. With the rst partial brachiopod 18SrRNA (small subunit ribosomal RNA) sequence (Field et al. 1988) used in a molecular analysis of di erent metazoan taxa,new emphasis was put on the idea of a protostome a nity of the Lophophorata. Although the method and interpretation of this work was criticized,a reanalysis led to the same result (Lake 1990). Analyses of complete 18SrRNA sequences from di erent lophophorate taxa demonstrated that all three groups (Phoronida,Bryozoa and Brachiopoda) always cluster within the protostomes along with molluscs and annelids (Conway-Morris 1995; Halanych et al. 1995; Conway-Morris et al. 1996; Mackey et al. 1996; Cohen & Gawthrop 1997) in a clade named Lophotrochozoa (Halanych et al. 1995). Analyses using partial mitochondrial 12SrDNA (small subunit ribosomal RNA) sequences (Cohen et al. 1998b) con rmed the clustering of brachiopods within the protostomes. To address the question of brachiopod a nity within the metazoa further,we determined the complete mitochondrial DNA (mtdna) sequence of Terebratulina retusa. In the last few years mitochondrial gene order comparisons have 266,2043^2052 2043 & 1999 The Royal Society Received 28 June 1999 Accepted 23 July 1999

2044 A. Stechmann and M. Schlegel Mitochondrial genome of the brachiopod T. retusa become a promising new tool for investigating phylogenetic relationships within the major metazoan groups (see Boore & Brown 1998; Boore 1999). Several features of mitochondrial genomes support the idea of investigating deep level phylogenies in the metazoa. The gene content of metazoan mtdna is almost invariant,thus a comparable data set is available for all taxa (Moritz et al. 1987; Wolstenholme 1992). Stable structural gene rearrangements are rare events due to the fact that functional genomes have to be maintained. It is rather unlikely that two or more taxa will independently converge on the same gene order,thus an identical gene order strongly supports a common ancestry of those taxa. Furthermore,reversals of gene order are presumed to be unlikely. Thus,there is little danger of homoplastic events hiding the true phylogeny (Smith et al. 1993;Boore&Brown1994a,b; Booreet al. 1995; Lee & Kocher 1995). Studies on the complete mtdna of T. retusa have already been carried out byjacobs et al. (1988) using blot hybridizations to determine the relative positions of three genes on the mitochondrial genome. Cohen et al. (1991) investigated the population genetic structure between Scottish and Canadian Terebratulina populations using,among other methods,restriction fragment length polymorphism (RFLP) analysis of the mtdna. However,the complete mtdna sequence was not determined in any of these works. 2. MATERIAL AND METHODS (a) DNA extraction, cloning, polymerase chain reaction and sequencing Single specimens of T. retusa (Linne 1758) were collected by divers in the Gullmarnfjord near Kristineberg,Sweden. Five specimens were shipped alive in cold seawater to Leipzig, Germany. Total genomic DNA was isolated from the gonads and muscles of a single specimen using standard phenol/chloroform protocols (Sambrook et al. 1989). The entire mtdna was ampli ed by two overlapping polymerase chain reactions (PCRs). The PCR strategy is described in the following. Two short fragments of the 16SrRNA and cytochrome c oxidase subunit I () genes were obtained using standard PCR conditions with slightly changed universal primers from Palumbi et al. (1991): 5'-CGCCTGTTTATCAAAAACAT-3' (16Sar-L),5'-CCGGTCTGAACTCAGATCAYGT-3' (16Sbr-H), 5'-CCTGCAGGAGGAGGAGAYCC-3' (f-l) and5'-agtat- AAGCATCTGGGTARTC-3' (a-h). Both gene fragments were cloned,sequenced and compared to the existing database (BLASTN; Altschul et al. 1995) to ensure that no contaminants were ampli ed. In order to amplify larger fragments of the mitochondrial genome the 16SrRNA and primer pairs were used in all four combinations. The combination of f-l and 16Sbr-H yielded a single product of ca. 4.7 kb length. This product was digested with EcoRI and XbaI,generating three and four fragments,respectively. A double digest with both restriction enzymes was also carried out. All restriction fragments were cloned into puc18 and puc19 (Boehringer, Mannheim,Germany) and sequenced using dye-labelled universal M13 forward and reverse sequencing primers. Sequencing reactions were performed using Thermo Sequenase (Amersham,Freiburg,Germany) and separated on an automated DNA sequencer (LI-COR). Using restriction fragments from single and double digests and vectors with complementary orientated polylinkers,both strands were sequenced in an overlapping mode. The 5'- and 3'-ends of this 4.7 kb fragment were identical to the and 16SrRNA genes of Terebratulina that had been obtained previously. From this 4.7 kb sequence the following two primers were designed to amplify the remaining mitochondrial genome: 5'-GTATCGTCGAGGTATCCCTCC- TAATCCTAG-3' (Tr/LCOX) and 5'-CGAATAAGGAGATTG- CGACCTCGATGTTGG-3' (Tr/L16S). Both primers were used in a long PCR with a combination of Taq- and Pwo-polymerases (Expand TM Long Template PCR System,Boehringer, Mannheim,Germany) and cycling conditions as recommended by the manufacturer. Electrophoretic analysis of the reaction showed a single product of ca. 11.5 kb length. This product was digested with the following restriction enzymes to generate overlapping fragments: ApaI, BamHI, EcoRI, HindIII, PstI, SacI, SmaI and XbaI. Double digests were carried out to create smaller fragments if necessary. All restriction fragments were cloned into the vectors puc18,puc19 (Boehringer,Mannheim, Germany) and pbluescript SK + (Stratagene,Amsterdam,The Netherlands). Sequencing was performed as already described for the 4.7 kb fragment. (b) Sequence analysis and phylogenetic reconstruction Gene sequences were identi ed as follows. Ribosomal genes were determined by their similarity to corresponding genes in alignments with other metazoan mtdnas. Open reading frames (ORFs) were determined with the program DNASIS for Windows,v. 2 using the genetic code for invertebrate mtdna. ORFs were identi ed according to the sequence similarity of their inferred amino-acid sequence in alignments using BLASTX (Altschul et al. 1995). In the case of ND4L and (NADH dehydrogenase subunits 4L and 6) and ATP8 (ATP synthetase subunit 8),which show only weak amino-acid sequence similarities,hydropathy pro les (Kyte & Doolittle 1982) were calculated and compared with Katharina tunicata (Boore & Brown 1994b), Drosophila yakuba (Clary & Wolstenholme 1985), Lumbricus terrestris (Boore & Brown 1995) and Balanoglossus carnosus (Castresana et al. 1998) (data not shown). Transfer RNA (trna) genes were identi ed manually by their potential to be folded into the stems and loops characteristic of mitochondrial trnas and speci ed by the triplet code in their putative anticodon loop. The gene arrangement of T. retusa was compared manually with di erent protostome and deuterostome mitochondrial genomes to assess brachiopod a nity within the metazoa. Using the criterion of parsimony the least number of steps necessary to interconvert two mitochondrial genomes was interpreted as more probable for a common ancestry. To elucidate brachiopod relationships within the metazoa further,phylogenetic trees were reconstructed from alignments of amino-acid sequences. Since the prosobranch Littorina saxatilis plays a major role in brachiopod a nities within the protostomes, only those amino-acid sequences which have already been determined for this mollusc were used. Furthermore,the alignments of the and Cytb (cytochrome b apoenzyme) amino-acid sequences were adjusted with respect to the sequenced portion of Littorina. Alignments were carried out with CLUSTAL W,v. 1.74 (Thompson et al. 1994) using default parameters. Amino-acid sequences from di erent metazoan taxa (see table 1) of the following protein coding genes were obtained from GenBank:, ATP8,,, Cytb, ND1 and. Alignment of the ATP8 amino-acid sequence produced only a poor t and was excluded from subsequent analyses. Regions with di cult

Mitochondrial genome of the brachiopod T. retusa A. Stechmann and M. Schlegel 2045 Table 1. Specimens and mtdna sequences used for phylogenetic analysis (Sequences of the following specimens representing di erent phyla were taken from GenBank and EMBL for alignments of mitochondrial amino-acid sequences to be used in phylogenetic analysis.) phylum species accession Chordata Xenopus laevis M10217 Petromyzon marinus U11880 Branchiostoma lanceolatum Y16474 Hemichordata Balanoglossus carnosus AF051097 Echinodermata Paracentrotus lividus J04815 Strongylocentrotus purpuratus X12631 Asterina pectinifera D16387 Florometra serratissima AF049132 Arthropoda Artemia franciscana X69067 Locusta migratoria X80245 Drosophila yakuba X03240 Mollusca Katharina tunicata U09810 Littorina saxatilis AJ132137 Annelida Lumbricus terrestris U24570 Nematoda Caenorhabditis elegans X54252 Ascaris suum X54253 Cnidaria Metridium senile AF000023 alignments were detected manually and removed,as well as gaps. All aligned amino-acid sequences were then concatenated to be used in phylogenetic analyses (alignment 1). Phylogenetic trees were obtained using di erent programs. Neighbour-joining trees were inferred with the program TREECON for Windows,v. 1.3b (Van de Peer & De Wachter 1994). Distance estimations were calculated using the Kimura (1983; Hasegawa & Fujiwara 1993) and the Tajima & Nei (1984) models. Bootstrap analysis was done with 500 replicates to test the robustness of the phylogenetic trees. Maximum-likelihood analyses were done with PUZZLE,v. 4.0.2 (Strimmer & Von Haeseler 1996) using the mtrev24 model (Adachi & Hasegawa 1996) of amino-acid substitution. Aminoacid frequencies were calculated by the program from the respective data sets. The robustness of the inferred phylogenetic trees was tested with likelihood mapping (Strimmer & Von Haeseler 1997) as implemented in PUZZLE,v. 4.0.2 and by subjecting alternative tree topologies to the Kishino & Hasegawa (1989) test. Maximum-likelihood analyses were done with 10 000 puzzling steps to test the reliability of internal branches. The variability of sites was estimated with PUZZLE,v. 4.0.2 which introduced eight gamma-distributed rates,where each position in the alignment is given one of eight rate categories according to Felsenstein & Churchill (1996). In three successive analyses those positions that were most variable (category 8),positions with the two most variable (category 7 + 8),and three most variable (category 6^8) sites were removed. The resulting data sets (alignments 2^4) were used for phylogenetic analyses as described above. The mtdna sequence of T. retusa is deposited in EMBL (accession number: AJ245743). Sequence alignments are available on request from the authors or from http://www.unileipzig.art/bde/fak/biowiss.htm. 3. RESULTS AND DISCUSSION (a) Gene content The complete mtdna sequence of T. retusa is 15 451bp long which is slightly less than the values given in Cohen et al. (1991). The complete sequence has an AT content of 57.2% and consists of all 37 genes usually found in metazoan mitochondrial genomes (see also gure 1): 13 protein coding genes (^I, Cytb,, ATP8, ND1^ and ND4L),two genes for subunits of the ribosomal RNA,a small subunit (12SrRNA) and a large subunit (16SrRNA),and 22 trna genes. The extreme economic arrangement of genes typical of metazoan mitochondrial genomes also applies in T. retusa. No introns are found and the longest region between two adjacent genes spans 12 bp (I^tRNA Thr ). Most of the genes abut directly or overlap by one to seven nucleotides. All genes are transcribed from the same strand. As already shown by Jacobs et al. (1988) the and genes are adjacent to each other,but the relative location of the 16SrRNA gene is somewhat di erent (cf. gure 10.3 in Jacobs et al. (1988)). These inconsistencies are not unexpected,considering the methods used: Southern blot hybridization versus complete sequence analysis (B. L. Cohen,personal communication). The Terebratulina mitochondrial genome contains one unassigned region of 794 bp length between the ND1 and genes with a slightly higher ATcontent (61.7%) as compared to the rest of the genome. The 5'- part of this segment consists of six copies of a 68 bp tandem repeat followed by a seventh incomplete copy which stops after 32 bp. A repeat region of 13 guanines is located 56 bp upstream from the inferred translation initiation of the gene. The exact number of nucleotides in this homopolymer run still needs to be determined. From reports of other metazoan mtdnas this could possibly represent the origin of replication (Clary & Wolstenholme 1985; Boore & Brown 1995) but,as long as this has not been shown experimentally,the assumption remains speculative. We generated a computer-based restriction fragment pattern using the same enzymes as in the study of Cohen et al. (1991) and compared this pattern to the di erent mitotypes given there. We could determine the equivalent mitomorph for six enzymes; although the fragments sometimes di ered in size,in general they resembled the Scottish samples. Whether these size di erences originate from a varying copy number of the tandem repeats in the long non-coding region remains unknown. Only complete sequences of otherterebratulina mtdnas will help clarify this question. A more detailed description of the mtdna of T. retusa will be given elsewhere. (b) Comparison of gene arrangements with mitochondrial genomes of di erent metazoan phyla Once the gene order of T. retusa was established the most striking feature was its similarity to the gene arrangement of the polyplacophoran mollusc K. tunicata (Boore & Brown 1994b). Only one major rearrangement is required to interconvert the brachiopod and polyplacophoran gene order (see gure 1). This involves an inversion of a fragment ranging from the 12SrRNA gene to the ND5 gene with the inclusion of some of the adjacent trna genes (trna Phe, trna Tyr, trna Cys and trna Met ). Of the 22 trna genes,ten keep their absolute and four their relative positions when comparing both genomes, whereas the positions of eight trna genes have changed. However,when we compared the brachiopod gene order with the partial mtdna of the prosobranch gastropod

2046 A. Stechmann and M. Schlegel Mitochondrial genome of the brachiopod T. retusa ATP8 S(UCN) L(CUN) MYQE KRI S(AGN) ND5 ND4 ND4L Cytb ND1 16S 12S I ND3 ND2 Katharina D F H T P L(UUR) V CWG AN ATP8 C V A P N W E R S(AGN) 12S 16S ND1 Cytb ND4L ND4 ND5 I ND3 ND2 Terebratulina D YM L(CUN) L(UUR) KS(UCN) QH FG T I ATP8 MCQE L(UUR) P 12S 16S ND1 Cytb Littorina D YWG V L(CUN) Figure 1. Gene organization of the T. retusa (Brachiopoda) mitochondrial genome compared to K. tunicata and L. saxatilis (Mollusca). Protein coding genes and ribosomal genes are designated by their abbreviations as given in the text (see } 3(a)). trna genes are indicated by the one letter code of the amino acid they specify. Genes coding on the opposite strand are underlined. Unassigned regions are marked by a shaded box. Double arrows indicate the rearrangements required to interconvert the genomes of Terebratulina and Katharina with respect to protein coding and ribosomal genes. Rearrangements of trna genes are not shown. The region spanning the major rearrangement between the Terebratulina and Katharina mtdna sequence is underlined. Note that the mitochondrial genome of L. saxatilis is only partial. Littorina saxatilis (Wilding et al. 1999) we found that both genomes show complete congruence in gene order with respect to the ribosomal and protein coding genes. Out of the 12 trna genes determined so far in the Littorina mtdna,two retain their absolute and ve their relative positions,whereas ve trna genes have changed positions compared to Terebratulina ( gure 1). The partial mtdna sequence of four genes of another prosobranch mollusc, Plicopurpura columellaris (Boore et al. 1995),shows complete congruence with the Terebratulina mtdna gene order. The recently published partial mtdna sequence of a cephalopod (Loligo bleekeri; Sasuga et al. 1999) requires only one rearrangement event concerning the protein coding genes when compared to the Terebratulina mitochondrial gene arrangement. This involves the translocation and inversion of the ND5^ND4^ND4L protein gene block. Out of the 14 trna genes determined so far,two keep their positions,whereas 12 have changed their positions when compared to Terebratulina (not shown). Other complete molluscan mtdna sequences from Mytilus edulis (Ho mann et al. 1992) and pulmonate gastropods (Hatzoglou et al. 1995; Terrett et al. 1996; Yamazaki et al. 1997) show a rather di erent arrangement to the Terebratulina, Katharina and Littorina gene order (data not shown). Comparisons of the Terebratulina mitochondrial gene order with di erent arthropods revealed that three rearrangement events of protein coding and ribosomal genes are necessary to interconvert these genomes (e.g. Limulus polyphemus; Staton et al. 1997). The annelid L. terrestris (Boore & Brown 1995) requires at least four rearrangement events to produce an identical gene order to that of the brachiopod (data not shown). Comparisons of protein coding and ribosomal genes on the brachiopod mtdna sequence with di erent deuterostome sequences revealed the following results (see gure 2). In the echinoderms three di erent arrangements of the mtdna sequence have so far been found,which can be related to di erent lineages in this group (Smith et al. 1993; Asakawa et al. 1995; A. Scouras and M. J. Smith, unpublished data). The brachiopod gene order shows only little similarity to any of the echinoderm arrangements, requiring a minimum of six rearrangement events. At least four rearrangement events are needed to convert the Terebratulina gene order into the hemichordate (B. carnosus; Castresana et al. 1998) or cephalochordate (B. lanceolatum; Spruyt et al. 1998) gene orders. These involve translocations of the I/ND3 and 12SrRNA/ 16SrRNA/ND1 blocks,translocations of the and Cytb genes and an inversion of the gene ( gure 2).

Mitochondrial genome of the brachiopod T. retusa A. Stechmann and M. Schlegel 2047 D ATP8 G R H EP V L(CUN) QM NAY I ND3 ND4L ND4 ND5 Cytb 12S 16S ND1 ND2 Balanoglossus S(UCN) K S(AGN) T F L(UUR) I WC ATP8 C V A N W E R S(AGN) 12S 16S ND1 Cytb ND4L ND4 ND5 I ND3 ND2 Terebratulina D YM L(CUN) L(UUR) P K S(UCN) QH FG TI ATP8 H PNACMT F G Y I ND4L I ND3 ND4 ND5 Cytb 12S 16S ND2 ND1 Florometra R K S(UCN) S(AGN) QLWVDE L(UUR) Figure 2. Gene organization of the T. retusa (Brachiopoda) mitochondrial genome compared to B. carnosus (Hemichordata) and F. serratissima (Echinodermata). Abbreviations are as in gure 1. Double arrows indicate the rearrangements required to convert the brachiopod mtdna sequence into the hemichordate and echinoderm sequence. Regions of rearrangements are underlined. Rearrangements of trna genes are not shown. With respect to the trna genes numerous rearrangements are required: in the Balanoglossus and Branchiostoma mitochondrial genomes 17 and 18 trna genes,respectively, have changed their positions compared to the brachiopod gene arrangement. The relatively high number of rearrangement events concerning the trna genes conforms with the view that these genes are more mobile on the mitochondrial genome (Cantatore et al. 1987; Moritz et al. 1987). (c) Phylogenetic implications from gene order comparisons Analysis of mitochondrial gene order can be done using the overall similarity in gene order,i.e. similar or identical gene arrangements imply common ancestry. Because rearrangements have to be considered as rare events,the most parsimonious explanation is favoured. Those genomes requiring the least number of rearrangements when compared are likely to indicate a close relationship. Since trna genes seem to rearrange at a much higher rate than ribosomal and protein coding genes, they are probably more informative for within-phyla relationships (see Boore et al. 1995) and,therefore,not very useful for deep-level phylogenetic relationships. For this reason,the phylogenetic implications from comparisons of the gene order on the mitochondrial genome will be restricted to the ribosomal and protein coding genes. The several independent rearrangement events required to interconvert the Terebratulina and any of the deuterostome gene arrangements imply that a close brachiopod relationship with any of the deuterostome phyla seems highly unlikely. In contrast,only one major rearrangement event separates the brachiopod from the K. tunicata (Mollusca: Polyplacophora) gene order. In addition,the gene order of L. bleekeri (Mollusca: Cephalopoda),as determined so far,requires one rearrangement event when compared to Terebratulina. However,with the exclusion of trna genes,the gene order of the partial mtdna sequence of L. saxatilis (Mollusca: Prosobranchia) is completely congruent with the brachiopod. This high degree of similarity in gene order is the rst to be reported between di erent protostome phyla. In contrast,the gene arrangements found in the bivalve M. edulis and the pulmonate mtdnas are rather di erent from Terebratulina. These di erences are interpreted to represent highly derived states within those lineages (Sasuga et al. 1999; Wilding et al. 1999,see also Boore 1999). With the mtdna data available so far,we cannot completely rule out an independent evolution of the lineages leading to brachiopods and molluscs. However, the nearly identical gene order in T. retusa and the three molluscs Katharina, Loligo and Littorina suggests a close a nity and perhaps even common ancestry of both

2048 A. Stechmann and M. Schlegel Mitochondrial genome of the brachiopod T. retusa Table 2. Phylogenetic analysis of six amino-acid sequences encoded on mtdna type of alignment a number of sites %of constant sites lophotrochozoa ln L b ((prot,brach), deut) c ln L s.e. (prot,(brach, deut)) ln L s.e. (chor,((echi, hemi),brach)) ln L s.e. (chor,((hemi, brach),echi)) ln L s.e. 1 (full) 1143 28.3 21 125.48 * 33.25 14.09 79.55 20.19 154.64 26.26 189.37 28.01 2( category 8) 1005 32.1 15 480.30 * 34.62 11.02 75.52 16.73 135.20 22.26 175.64 23.98 3( category 8 + 7) 854 37.8 10 610.60 * 43.17 15.20 91.38 20.65 127.57 23.68 153.48 24.53 4( category 8 + 7 + 6) 691 46.7 6554.41 * 43.78 12.83 80.08 17.34 100.37 19.78 117.92 20.74 a Di erent alignments containing all eight gamma-distributed rate categories (alignment 1,full),exclusion of the most variable category, (alignment 2, category 8),exclusion of the two most variable categories,(alignment 3, category 8 + 7),exclusion of the three most variable categories,(alignment 4, category 8 + 7 + 6). b Log-likelihood values of the preferred topologies (ln L); di erences in the log likelihood (ln L) and standard error of this value in relation to the preferred topolgy. An asterisk indicates the preferred maximum-likelihood topology. c Five di erent topologies relating the brachiopod (brach) with protostomes (prot),deuterostomes (deut),echinoderms (echi) and hemichordates (hemi); lophotrochozoa: brachiopods + molluscs + annelids. phyla. This is further supported by 18SrRNA sequence analyses (Field et al. 1988; Lake 1990; Halanych et al. 1995; Mackey et al. 1996; Cohen & Gawthrop 1997; Cohen et al. 1998a) and fossil evidence (e.g. Conway- Morris 1998). If it holds true that the Cambrian explosion represents the simultaneous diversi cation of three already long-existing stem lineages,i.e. Lophotrochozoa, Ecdysozoa and Deuterostomia (Balavoine & Adoutte 1998),with brachiopods having a close phylogenetic relationship to molluscs,we can infer the following hypothesis concerning mtdna gene arrangements in these lineages. We interpret the gene order of the prosobranch L. saxatilis to represent the ancestral molluscan state. In each of the lineages leading to the polyplacophorans and cephalopods a rearrangement must have occurred resulting in the apomorphic gene order of K. tunicata and L. bleekeri, respectively. The brachiopod Terebratulina has kept the ancestral state and,therefore,both the brachiopod and Littorina represent the plesiomorphic state. Following the criteria of phylogenetic systematics (Hennig 1966) plesiomorphic features do not count in phylogenetic reconstructions. Thus,we are not able to make any inferences about brachiopod and mollusc phylogenetic relationships since they both share primitive characters. More data are needed,such as,for example,from inarticulate brachiopods and more basal branching articulate brachiopods,to see whether they share the same gene order with Terebratulina or whether the Terebratulina gene order is derived within the brachiopod lineage. In addition,data from more basal branching molluscs could help to assess the ancestral gene order of all molluscs reliably. To test whether the brachiopods share a close relationship with annelids,as suggested by 18SrRNA sequence comparisons, additional data from mitochondrial genomes of oligochaetes and polychaetes would be helpful. (d) Phylogenetic trees inferred from amino-acid sequence alignments The original alignment of the six concatenated aminoacid sequences with 1143 sites (alignment 1) was tested for rate heterogeneity introducing eight categories of variable sites with PUZZLE,v. 4.0.2. In subsequent analyses three di erent alignments (alignments 2^4) were produced by stepwise omission of variable sites (see } 2). In all,four di erent alignments were used for the phylogenetic 31.5% 0.3% 33.9% 0.1% 0.3% 33.7% Figure 3. Likelihood mapping analysis of the original alignment (alignment 1) represented as a triangle. Values at the corners show the percentages of well-resolved phylogenies for all possible quartets,values at the lateral region show the percentages of unresolved quartets and the value in the centre shows the percentage of completely unresolved quartets. The high number of all corner values (99.1%) indicates that the data set is phylogenetically informative. analyses (see table 2). The phylogenetic content of these four alignments was tested using the likelihood mapping analysis as implemented in PUZZLE,v. 4.0.2 (see gure 3). The percentage of well-resolved phylogenies for all possible quartets decreased only slightly with the reduced alignments,from 99.1% in the original alignment (alignment 1) to 94.2% in the shortest alignment (alignment 4) (cf. table 2). Correspondingly,unresolved and starlike phylogenies increased slightly from 0.9 and 0.1% to 3.1 and 2.6%,respectively. Still,those values indicate a high phylogenetic content in all four data sets. The neighbour-joining analyses inferred from the four di erent alignments showed that the brachiopod T. retusa always emerged within the protostomes along with the two molluscs K. tunicata and L. saxatilis and the annelid L. terrestris with bootstrap support of 597%. The maximum-likelihood analyses also showed the brachiopod T. retusa within the Lophotrochozoan clade with quartet puzzling support of 589% ( gure 4). Di erences emerged in resolving the phylogeny within the Lophotrochozoa. In all neighbour-joining analyses the two molluscs clearly branch o together,but the branching of the brachiopod and the annelid could not be resolved and is 0.3%

Mitochondrial genome of the brachiopod T. retusa A. Stechmann and M. Schlegel 2049 89/94 97/98 80/100 97/99 Katharina tunicata Littorina saxatilis Terebratulina retusa 61/94 83/99 99/100 93/96 79/100 89/99 82/100 99/100 98/100 100/100 94/100 98/100 88/99 84/100 99/100 99/100 83/99 83/99 93/100 99/100 97/100 75/97 Lumbricus terrestris Ascaris suum Caenorhabditis elegans Drosophila yakuba Locusta migratoria Artemia franciscana Paracentrotus lividus Strongylocentrotus purpuratus Asterina pectinifera Florometra serratissima Balanoglossus carnosus Xenopus laevis Petromyzon marinus Branchiostoma lanceolatum PROTOSTOMIA DEUTEROSTOMIA Metridium senile Figure 4. Phylogenetic tree of protostome and deuterostome taxa inferred from alignments 1^4 (for more information see } 2). Phylogenetic inference was performed with maximum likelihood and neighbour joining. Upper numbers indicate quartet puzzling support values from 10 000 puzzling steps in the maximum-likelihood analyses of all four alignments (lowest and highest value). Lower numbers represent bootstrap values calculated with 500 replicates in the neighbour-joining analyses. Di erences emerged in the branching order of Lumbricus and Terebratulina,depending on the data set used for phylogenetic analysis (tree drawn as a multifurcation). In the maximum-likelihood and neighbour-joining analysis of alignment 4 the two taxa branch o as sister groups with support values of 83 and 45%. In the neighbour-joining analyses the branching order of Lumbricus and Terebratulina varied and the bootstrap support was always below 50%. Since the branching of the two nematodes within the protostome clade could not be resolved in the di erent analyses it is also drawn as a multifurcation. therefore drawn as a multifurcation. Maximum-likelihood analysis of the original alignment (alignment 1) shows Lumbricus to emerge as the rst taxon in the Lophotrochozoa followed by Terebratulina with quartet puzzling support of 78% ( gure 5). In the following maximum-likelihood analysis,reducing the alignment by the most variable site ( category 8; alignment 2) produced a lower quartet support value for the branching of Lumbricus and Terebratulina within the Lophotrochozoa,leaving their branching order unresolved. Further reduction of the alignment ( category 7 + 8; alignment 3) shows the annelid branching o with 83% followed by Terebratulina. But in the shortest alignment consisting of the most conserved sites ( category 6^8; alignment 4),the annelid and the brachiopod branch o as sister-clades with 83% support. Obviously,the data set does not contain enough information to resolve the ingroup relationship of the Lophotrochozoa,although likelihood mapping analyses indicated that the data set is phylogenetically informative. Another problem concerns the placement of the two nematodes C. elegans and A. suum. Recent analyses of 18SrRNA sequences showed that the nematodes branch o together

2050 A. Stechmann and M. Schlegel Mitochondrial genome of the brachiopod T. retusa 0.1 100 78 94 100 52 94 88 99 100 100 94 93 97 Katharina tunicata Littorina saxatilis Terebratulina retusa Lumbricus terrestris Ascaris suum 99 Caenorhabditis elegans Drosophila yakuba Locusta migratoria Artemia franciscana Paracentrotus lividus 100 Strongylocentrotus purpuratus Asterina pectinifera Florometra serratissima Balanoglossus carnosus Xenopus laevis Petromyzon marinus LOPHOTROCHOZOA ECDYSOZOA DEUTEROSTOMIA Branchiostoma lanceolatum Metridium senile Figure 5. Maximum-likelihood tree calculated from alignment 1 (1143 sites) with 10 000 puzzling steps showing maximum-likelihood branch lengths. The branch leading to the nematodes A. suum and C. elegans is around 4.3 times longer as shown. All bifurcations are supported with values above 50%,which correspond to the bootstrap values. In this analysis three groups emerge in the bilaterian clade: Lophotrochozoa,Ecdysozoa and Deuterostomia. The scale bar represents a distance of 0.1 substitutions per site. with the arthropods forming a clade named Ecdysozoa (Aguinaldo et al. 1997; Ruiz-Trillo et al. 1999). This could only be con rmed in one analysis with a su ciently high quartet support value (see gure 5). All other analyses failed to resolve the branching of the nematodes with su ciently high support (see also gure 4). However,they branch within the protostomes in all analyses and do not emerge at the base of bilaterian animals. To test for various hypotheses concerning brachiopod relationships within the metazoa we compared maximum-likelihood tree topologies with alternative topologies using the Kishino & Hasegawa (1989) test. The alternative tree topologies included a brachiopod sister-group relationship with protostomes or deuterostomes,a sister-group relationship of brachiopods with a clade including hemichordates and echinoderms and a sister-group relationship of brachiopods and hemichordates. The latter has been discussed since pterobranchs and brachiopods both possess lophophores (Nielsen 1987; Halanych 1993,1995). Tests were carried out from all alignments and the results are shown in table 2. The di erences in log likelihood had to exceed the standard error 1.96 times before being signi cant ( p50.05) (Strimmer & Von Haeseler 1996). In all tests the signi cantly best tree was the one showing the brachiopod branching o with the annelid and the molluscs,

Mitochondrial genome of the brachiopod T. retusa A. Stechmann and M. Schlegel 2051 supporting the Lophotrochozoa concept. Di erent placements of the brachiopod were signi cantly worse and therefore rejected by the test. 4. CONCLUSIONS Both gene order comparisons and amino-acid sequence comparisons strongly support a protostome relationship of T. retusa. It becomes more evident that the long-held view of a deuterostome relationship of brachiopods could be wrong,at least this is evident from the molecular viewpoint. Furthermore,the new concept of the clade Lophotrochozoa within the protostomes becomes more and more acceptable with the accumulating evidence from independent markers. This work now provides new evidence from mtdna sequences. Gene order comparisons suggest a protostome a nity of brachiopods. Further,sequence comparisons of six protein-coding genes support a closer relationship of brachiopods,annelids and molluscs. The close a nity of T. retusa and the polyplacophoran K. tunicata is consistent with the results of Cohen & Gawthrop (1997) and Cohen et al. (1998a),who found the mollusc to be the most proximal outgroup to brachiopods using 18SrRNA sequences. Still,we are not able to resolve the detailed relationships within the clade Lophotrochozoa. Most studies concerning complete mitochondrial genomes have been on chordates and arthropods while the mitochondrial genomes of many invertebrate phyla are not yet known. We assume that the problems arising in resolving the detailed relationships between lophotrochozoan taxa are due to the shortage of sequence data available,particularly in groups such as the annelids and brachiopods. Both phyla are so far represented by one genome,which is not su cient to draw any major conclusions. Thus,we propose that our data should be treated as rst evidence and that we must await more mitochondrial genomes of those taxa before coming to any nal conclusion. Furthermore,it should be noted that T. retusa does not represent a basal branching brachiopod,but rather belongs to the more derived group of articulate brachiopods (Cohen & Gawthrop 1997; Cohen et al. 1998a). Additional data from inarticulate brachiopods are needed to clarify questions such as what does the basic arrangement of the mitochondrial genome (`Grundplan') from brachiopods look like? We are very grateful to Dr C. LÏter and some unknown divers for providing us with living specimens of T. retusa. Many thanks are due to Dr J. Castresana for helpful instructions about the long PCR method. A.S. would like to thank Dr B. L. Cohen for all his help and encouragement during this work and for the unpublished material he provided. For critical reading of this manuscript we thank S. Theophilou and Dr B. L. Cohen. Many thanks are due to H. H. Fo«rster for his last-minute help on the graphics. REFERENCES Adachi,J. & Hasegawa,M. 1996 Model of amino acid substitution in proteins encoded by mitochondrial DNA. J. Mol. Evol. 42,459^468. Aguinaldo,A. M. A.,Turbeville,J. M.,Linford,L. S.,Rivera, M. C.,Garey,J. R.,Ra,R. A. & Lake,J. A. 1997 Evidence for a clade of nematodes,arthropods and other moulting animals. Nature 387,489^493. Altschul,S. F.,Gish,W.,Miller,W.,Myers,E. W. & Lipman, D. L. 1995 Basic local alignment search tool. J. Mol. Biol. 215, 403^410. Asakawa,S.,Himeno,H.,Miura,K. & Watanabe,K. 1995 Nucleotide sequence and gene organization of the star sh Asterina pectinifera mitochondrial genome. Genetics 140,1047^1060. Backeljau,T.,Winnepenninckx, B. & De Bruyn, L. 1993 Cladistic analysis of metazoan relationships: a reappraisal. Cladistics 9,167^181. Balavoine,G. & Adoutte,A. 1998 One or three Cambrian radiations? Science 280,397^398. Boore,J. L. 1999 Animal mitochondrial genomes. Nucl. Acids Res. 27,1767^1780. Boore,J. L. & Brown,W. M. 1994a Mitochondrial genomes and the phylogeny of mollusks. Nautilus 2(Suppl.),61^78. Boore,J. L. & Brown,W. M. 1994b Complete DNA sequence of the mitochondrial genome of the black chiton Katharina tunicata. Genetics 138,423^443. Boore,J. L. & Brown,W. M. 1995 Complete sequence of the mitochondrial DNA of the annelid worm Lumbricus terrestris. Genetics 141,305^319. Boore,J. L. & Brown,W. M. 1998 Big tress from little genomes: mitochondrial gene order as a phylogenetic tool. Curr. Opin. Genet. Dev. 8,668^674. Boore,J. L.,Collins,T. M.,Stanton,D.,Daehler,L. L. & Brown,W. M. 1995 Deducing the pattern of arthropod phylogeny from mitochondrial DNA rearrangements. Nature 376, 163^165. Boore,J. L.,Lavrov,D. V. & Brown,W. M. 1998 Gene translocation links insects and crustaceans. Science 392,667^668. Brusca,R. C. & Brusca,G. J. 1990 Invertebrates. Sunderland, MA: Sinauer Associates. Cantatore,P.,Gadaleta,M. N.,Roberti,M.,Saccone,C. & Wilson A. C. 1987 Duplication and remoulding of trna genes during the evolutionary rearrangement of mitochondrial genomes. Nature 329,853^855. Castresana,J.,Feldmaier-Fuchs,G.,Yokobori,S.,Satoh,N. & PÌÌbo,S. 1998 The mitochondrial genome of the hemichordate Balanoglossus carnosus and the evolution of deuterostome mitochondria. Genetics 150,1115^1123. Clary,D. O. & Wolstenholme,D. R. 1985 The mitochondrial DNA molecule of Drosophila yakuba: nucleotide sequence,gene organization and genetic code. J. Mol. Evol. 22,252^271. Cohen,B. L. & Gawthrop,A. B. 1997 The brachiopod genome. In Treatise on invertebrate paleontology. Brachiopoda,revised (ed. A. Williams),pp. 189^211. Lawrence,KS: Geological Society of America and University of Kansas Press. Cohen,B. L.,Balfe,P. & Cohen,M. 1991 Molecular evolution and morphological speciation in North Atlantic brachiopods (Terebratulina ssp.). Can. J. Zool. 69,2903^2911. Cohen,B. L.,Gawthrop,A. & Cavalier-Smith,T. 1998a Molecular phylogeny of brachiopods and phoronids based on nuclear-encoded small subunit ribosomal RNA gene sequences. Proc. R. Soc. Lond. B 353,2039^2061. Cohen,B. L.,Stark,S.,Gawthrop,A. B.,Burke,M. E. & Thayer,C. W. 1998b Comparison of articulate brachiopod nuclear and mitochondrial gene trees leads to a clade-based rede nition of protostomes (Protostomozoa) and deuterostomes (Deuterostomozoa). Proc. R. Soc. Lond. B 265, 475^482. Conway-Morris,S. 1995 Nailing the Lophophorates. Science 375, 365^366. Conway-Morris,S. 1998 Metazoan phylogenies: falling into place or falling to pieces? A palaeontological perspective. Curr. Opin. Genet. Dev. 8,662^667. Conway-Morris,S. & Peel,S. 1995 Articulated halkieriids from the Lower Cambrian of North Greenland and their role in early protostome evolution. Phil. Trans. R. Soc. Lond. B 347, 305^358.

2052 A. Stechmann and M. Schlegel Mitochondrial genome of the brachiopod T. retusa Conway-Morris,S.,Cohen,B. L.,Gawthrop,A. B.,Cavalier- Smith,T. & Winnepenninckx, B. 1996 Lophophorate phylogeny. Science 272,282. Eernisse,D. J.,Albert,J. S. & Anderson,F. E. 1992 Annelida and Arthropoda are not sister-taxa: a phylogenetic analysis of spiralian metazoan morphology. Syst. Biol. 41,305^330. Emig,C. 1976 Le Lophophoreöstructure signi cative des Lophophorates (Brachiopoda,Bryozoa,Phoronida). Zool. Scr. 5,133^137. Emig,C. 1984 On the origin of the Lophophorata. Z. Zool. Syst. Evolut.-Forsch. 22,91^94. Felsenstein,J. & Churchill,G. A. 1996 A hidden Markov model approach to variation among sites in rate of evolution. Mol. Biol. Evol. 13,93^104. Field,K. G.,Olsen,G. J.,Lane,D. J.,Giovannoni,S. J., Ghiselin,M. T.,Ra,E. C.,Pace,N. R. & Ra,R. A. 1988 Molecular phylogeny of the animal kingdom. Science 239, 748^753. Halanych,K. M. 1993 Suspension feeding by the lophophorelike apparatus of the pterobranch hemichordate Rhabdopleura normani. Biol. Bull. 185,417^427. Halanych,K. M. 1995 The phylogenetic position of the pterobranch hemichordates based on 18SrDNA sequence data. Mol. Phyl. Evol. 4,72^76. Halanych,K. M.,Bacheller,J. D.,Aguinaldo,A. M. A.,Liva, S. M.,Hillis,D. M. & Lake,J. A. 1995 Evidence from 18S ribosomal DNA that the Lophophorates are protostome animals. Science 267,1641^1643. Hasegawa,M. & Fujiwara,M. 1993 Relative e ciencies of the maximum likelihood,maximum parsimony,and neighborjoining methods for estimating protein phylogeny. Mol. Phyl. Evol. 2,1^5. Hatzoglou,E.,Rodakis,G. C. & Lecanidou,R. 1995 Complete sequence and gene organization of the mitochondrial genome of the land snail Albinaria coerulea. Genetics 140, 1353^1366. Hennig,W. 1966 Phylogenetic systematics. Urbana,IL: University of Illinois Press. Ho mann,r. J.,Boore,J. L. & Brown,W. M. 1992 A novel mitochondrial gene organization for the blue mussel, Mytilus edulis. Genetics 131,397^412. Jacobs,H. T.,Balfe,P.,Cohen,B. L.,Farquharson,A. & Comito, L. 1988 Phylogenetic implications of genome rearrangement and sequence evolution in echinoderm mitochondrial DNA. In Echinoderm phylogeny and evolutionary biology (ed. C. R. C. Paul & A. B. Smith),pp.121^137. Oxford,UK: Clarendon Press. Kimura,M. 1983 The neutral theoryof molecular evolution. Cambridge University Press. Kishino,H. & Hasegawa,M. 1989 Evaluation of the maximum likelihood estimate of the evolutionary tree topologies from DNA sequence data,and the branching order in Hominoidea. J. Mol. Evol. 29,170^179. Kyte,J. & Doolittle,R. F. 1982 A simple method for displaying the hydropathic character of a protein. J. Mol. Biol. 157, 105^132. Lake,J. A. 1990 Origin of the metazoa. Proc. Natl Acad. Sci. USA 87,763^766. Lee,W. J. & Kocher,T. D. 1995 Complete sequence of a sea lamprey (Petromyzon marinus) mitochondrial genome: early establishment of the vertebrate genome organization. Genetics 139,873^887. Mackey,L. Y.,Winnepenninckx,B.,De Wachter,R.,Backeljau, T.,Emschermann,P. & Garey,J. R. 1996 18SrRNA suggests that Entoprocta are protostomes,unrelated to Ectoprocta. J. Mol. Evol. 42,552^559. Moritz,C.,Dowling,T. E. & Brown,W. M. 1987 Evolution of animal mitochondrial DNA: relevance for population biology and systematics. A. Rev. Ecol. Syst. 18,269^292. Nielsen,C. 1987 Structure and function of metazoan ciliary bands and their phylogenetic signi cance. Acta Zool. 68,205^262. Nielsen,C. 1995 Animal evolution. Interrelationships of the living phyla. Oxford University Press. Palumbi,S.,Martin,A.,Romano,W. O.,Stice,L. & Grabowski,G. 1991 The simple fools guide to PCR, version 2.0. Honolulu: University of Hawaii. Ruiz-Trillo,I.,Riutort,M.,Littlewood,D. T. J.,Herniou,E. A. & Bagu a,j. 1999 Acoel atworms: earliest extant bilaterian metazoans,not members of platyhelminthes. Science 283, 1919^1923. Salvini-Plawen,L. V. 1982 A paedomorphic origin of the oligomerous animals? Zool. Scr. 11,77^81. Sambrook,J.,Fritsch,E. T. & Maniatis,T. 1989 Molecular cloning. A laboratorymanual,2nd edn. New York: Cold Spring Harbor Laboratory Press. Sasuga,J.,Yokobori,S.,Kaifu,M.,Ueda,T.,Nishikawa,K. & Watanabe,K. 1999 Gene contents and organization of a mitochondrial DNA segment of the squid Loligo bleekeri. J. Mol. Evol. 48,692^702. Smith,M. J.,Arndt,A.,Gorski,S. & Fajber,E. 1993 The phylogeny of echinoderm classes based on mitochondrial gene arrangements. J. Mol. Evol. 36,545^554. Spruyt,N.,Delarbre,C.,Gachelin,G. & Laudet,V. 1998 Complete sequence of the amphioxus (Branchiostoma lanceolatum) mitochondrial genome: relations to vertebrates. Nucl. Acids Res. 26,3279^3285. Staton,J. L.,Daehler,L. L. & Brown,W. M. 1997 Mitochondrial gene arrangement of the horseshoe crab Limulus polyphemus L.: conservation of major features among arthropod classes. Mol. Biol. Evol. 14,867^874. Strimmer,K. & Von Haeseler,A. 1996 Quartet puzzling: a quartet maximum likelihood method for reconstructing tree topologies. Mol. Biol. Evol. 13,240^247. Strimmer,K. & Von Haeseler,A. 1997 Likelihood-mapping: a simple method to visualize phylogenetic content of a sequence alignment. Proc. Natl Acad. Sci. USA 94,6815^6819. Tajima,F. & Nei,M. 1984 Estimation of evolutionary distance between nucleotide sequences. Mol. Biol. Evol. 1,269^285. Terrett,J. A.,Miles,S. & Thomas,R. H. 1996 Complete DNA sequence of the mitochondrial genome of Cepaea nemoralis (Gastropoda: Pulmonata). J. Mol. Evol. 42,160^168. Thompson,J. D.,Higgins,D. G. & Gibson,T. J. 1994 CLUSTAL W: improving the sensitivity of progressive multiple sequence alignment through sequence weighting, position-speci c gap penalties and weight matrix choice. Nucl. Acids Res. 22,4673^4680. Van de Peer,Y. & De Wachter, R. 1994 TREECON for Windows: a software package for the construction and drawing of evolutionary trees for the Microsoft Windows environment. Comput. Appl. Biosci. 10,569^570. Wilding,C. S.,Mill,P. J. & Grahame,J. 1999 Partial sequence of the mitochondrial genome of Littorina saxatilis: relevance to gastropod phylogenetics. J. Mol. Evol. 48,348^359. Wolstenholme,D. R. 1992 Animal mitochondrial DNA: structure and evolution. Int. Rev. Cytol. 141,173^216. Yamazaki,N. (and 11 others) 1997 Evolution of pulmonate gastropod mitochondrial genomes: comparisons of gene organization of Euhadra, Cepaea and Albinaria and implications of unusual trna secondary structures. Genetics 145,749^758. As this paper exceeds the maximum length normally permitted, the authors have agreed to contribute to production costs.