Phylogenetic relationships among asteroids remain to be extremely controversial in spite of many morphological and molecular studies have been applied to this issue. In the present study, especially focusing on resolving the relationship of Asterina and Solasteridae, we reconstructed the molecular phylogenetic tree of asteroids using nuclear 18S rDNA. A close relationship between Asterina and Solasteridae, which has been suggested from analyses of mitochondrial 12S rDNA and 16S rDNA, is supported here by the nuclear 18S rDNA dataset. The support is even stronger when the sequences of mitochondrial rDNAs and nuclear 18S rDNA are combined as a total dataset. The independent support from both nuclear 18S rDNA and mitochondrial rDNAs strongly argues for a close relationship between the Asterina and Solasteridae.
Asteroids (Echinodermata, Asteroidea) are familiar and diverse marine invertebrates. Seven orders of approximately 35 families, 300 genera and 1800 species are recognized (Clark and Downey, 1992; Hendler et al., 1995).
Since classification of asteroids have been reformulated in these twenty years (Blake, 1987; Gale, 1987; Clark and Downey, 1992), several molecular phylogenetic studies have been performed to evaluate the phylogenetic relationships (Lafay et al., 1995; Wada et al., 1996; Knott and Wray, 2000). Although these molecular analyses have not succeeded to provide convincing phylogenetic framework of asteroids so far, the affinity between Asterina and Solaster is supported with relatively high supporting values (Wada et al., 1996).
Traditionally, the Asterinidae and the Solasteridae had been classified into the order Spinulosida, which is characterized by adambulacral mouth frame, reduced marginals, lack of true paxillae and arrangement of abactinal ossicles (either reticulated or imbricated) (Spencer and Wright, 1966; Blake, 1981). However, the order Spinulosida had been recognized as a provisional group constituted by assemblage of asteroids which is not included in other orders (Fisher, 1911; Spencer and Wright, 1966), and the two families are classified into different suborders: Solasterids in suborder Eugnathina and Asterina in suborder Leptognathina (Spencer and Wright, 1966). Blake (1981) re-examined the classification of the Spinulosida, and transferred Asterinidae from the order Spinulosida to the order Valvatida, mainly based on ossicle morphology of ambulacral column, and he also pointed out that imbricate pattern of abactinal ossicles of Asterina has no analogy with the Solasteridae. Since then, nobody has suggested the affinity between Asterinidae and Solasteridae, and Clark and Downey (1992) classified them into different orders, Valvatida and Velatida, respectively.
Therefore the close affinity between Asterina and Solaster suggested by the molecular phylogenetic analyses using mitochondrial 12S and 16S rDNA (Wada et al., 1996) was hard to be accepted from the morphological standpoint. Smith (1997) mentioned that the close relationship between Solasteridae and Asterina has never been proposed, and thus even suspected laboratory contaminant in the analyses by Wada et al. (1996).
In the present study, especially focusing on the relationship of Asterina and Solasteridae suggested by mitochondrial rDNAs (Wada et al., 1996), we re-examined the phylo-genetic relationships among asteroid species using nuclear marker: 18S rDNA. In order to disprove the laboratory contaminant suspected by Smith (1997), we re-sampled all the specimens.
MATERIALS AND METHODS
Biological materials and Isolation of genomic DNA
Eighteen species from nine families of asteroids (with asterisk in Table 1) were collected off the coast of Japan and Mauritius (Table 1), and processed to purify genomic DNA. In order to check the laboratory contaminant suspected by Smith (1997), all of these specimens were re-sampled from the field. Genomic DNAs were extracted from gonads or tube feet with a DNeasy Tissue Kit (Qiagen).
List of specimens analyzed in this study
Table 1 Continued
Sequencing strategy for 18S rDNAs
Approximately 1.8 kb of 18S rDNA were amplified by the polymerase chain reaction (PCR). The primers used for the amplifications have been described in Wada and Satoh (1994a). Amplification reactions were carried out in 20 μl volumes of a reaction mix with KOD-Plus-DNA polymerase (Toyobo). The temperature regime was 1 min at 94°C, 2 min at 50–60°C, and 5 min at 72°C for 35 cycles.
As the quantity of PCR products of some taxa was insufficient for direct sequencing, the sequences were determined after sub-cloned into plasmid vector. The amplified DNA fragment was purified by electrophoresis in a 1% agarose gel, and inserted into the vector pBluescript II SK (Stratagene). Multiple copies for the 18S rDNA gene are known to exist in the genome, and most of them maintain identical sequence by means of gene conversion. In some case, however, minor copies (which may be pseudogenes) may be picked up when PCR products are subcloned into plasmid vector (Wada, 1998). In order to avoid using sequences from these minor copies as representative of the species, we partially sequenced three independent clones, and confirmed that they had identical sequences. Since the KOD-Plus-DNA polymerase is a proof-reading DNA polymerase, we ignored errors due to misamplification during PCR. Sequencing was performed on an ABI prism 310 automatic sequencer using an ABI Prism BigDye terminator cycle sequencing kit (Applied Biosystems).
Sequencing strategy for mitochondrial rDNAs
Approximately 550 bp for mitochondrial (mt.) 16S rDNA, and 400 bp for 12S rDNA were amplified using primers reported in Wada et al. (1996). The temperature regime was 1 min at 94°C, 1 min at 50–60°C, and 1min at 72°C for 35 cycles. Amplified DNA fragments were purified by electrophoresis in a 1% agarose gel, and processed for direct sequencing.
Sequence Alignment and Phylogenetic Analysis
Sequences were aligned using the SeqApp 1.9 manual aligner for Macintosh (Gilbert, 1993). For 18S rDNA analysis, the nucleotide sequences were aligned with outgroup sequences from two brittle stars, Ophioplocus japonicus and Ophiocanops fugiens, and one sea urchin, Strongyrocentrotus purpuratus (Table 1). Confidently aligned 1669 nucleotide sites were used for analyses. In mitochondrial 12S and 16S rDNA analysis, two sea urchin, Strongylocentrotus purpuratus and Paracentrotus lividus (Table 1) were used as outgroup. Confidently aligned 559 nucleptide sites were processed for phylogenetic analyses. In total dataset analysis combining 18S rDNA and mt. rDNAs, sequences were aligned with one sea urchin species, Strongylocentrotus purpuratus and one hemi-chordate, Balanoglossus carnosus.
Phylogenetic trees were constructed using the neighbor-joining (NJ; Saitou and Nei, 1987), maximum likelihood (ML; Felsenstein, 1981), and maximum parsimony (MP) methods. Clustal X (Thompson et al., 1997) was used for the NJ method. Evolutionary distance was calculated according to Kimura's two-paramerter method (Kimura, 1980), and gaps and insertions were excluded from the analyses. ML analyses were performed with Tree-Puzzle 5.0 (Schmidt et al., 2002). Substitution rate heterogeneity was corrected by introducing Gamma-distributed rates for variable sites. We followed the HKY85 model (Hasegawa et al., 1985) for substitution process. PAUP 4.0b10 (Swafford, 2002) was used for the maximum parsimony analyses. MP analyses were constructed with transversion weighted equal to two transitions. The confidence for each node was estimated by bootstrap resampling (Felsenstein, 1985) for NJ and MP, and by quartet-puzzling for ML.
18S rDNA sequences
In this study, we determined the nucleotide sequences of 18S rDNA for 17 species included 9 families which cover five of the seven orders recognized in the classification by Clark and Downey (1992; Table 1).
Although 18S rDNA analyses have made a significant contribution for resolving phylogenetic relations (e.g. Wainright et al., 1993; Aguinaldo et al., 1997), some problems have been suggested to use rDNAs as a molecular marker. First, because rRNA functions by forming the secondary structure, the mutation rate is not uniform among sites (Wheeler and Honeycutt, 1988; Hills and Dixon, 1991). In order to overcome this problem, we performed ML analyses by correcting substitution rate heterogeneity introducing Gamma-distributed rates for variable sites. Second, heterogeneity of GC contents may also lead to a failure in recovering correct tree (Hasegawa and Hashimoto, 1993). We confirmed that the GC contents of the sequence used in the present analyses are between 51% and 56%.
Fig. 1 shows phylogenetic trees constructed by using the NJ, ML and MP methods. In these trees, the close affinities of species belonging to the same family, such as those in Luidiidae, Astropectinidae and Archasteridae, were well supported. In contrast, the confidence values for the branches linking different families were lower. Thus, 18S rDNA barely resolved the relationships among higher taxa.
A striking exception is a monophyly of the group that includes Asterina (Asterinidae) and two species of Solasteridae (Solaster and Crossaster). The monophyly of these groups is consistently supported by NJ and MP methods with relatively high bootstrap values. Although it is not supported by ML method, a close relationship between two Asterinidae species and two Solasteridae species is supported.
Mitochondrial rDNA sequences
Since affinity between Asterinidae and Solasteridae is suggested by Wada et al. (1996) based on mt. rDNA sequences, we analyzed mt. 12S and 16S rDNA sequence dataset with higher taxonomic density. About 550 bp of mt. 16S rDNA and 400 bp of 12S rDNA were sequenced for nine species in the present study (Table 1). The results are similar to those of 18S rDNA, namely, relationships between higher taxa were not resolved except for close relationship between Asterina and Solasteridae (Fig. 2). Although the early divergence of Luidiidae is supported by the previous analyses (Wada et al., 1996), this conclusion is not supported in the present analyses of higher taxonomic density. The close relationship between Asterina and Solasteridae is recovered from all methods, although bootstrap support from MP analysys is lower than 50%.
The total dataset of 18S rDNA and mt. 16S and 12S rDNA sequences
Finally, we analyzed twenty species for which both nuclear 18S rDNA and mt. 12S and 16S rDNA sequences were determined (Table 1). Even in these analyses of the total dataset, the phylogenetic relationships among higher taxa were not resolved with satisfying confidence values (Fig. 3). However, the monophyly of the Asterina and Solasteridae is supported consistently in all trees by all three methods with higher confidence values than in the analyses described above (Fig. 3).
In this study, we tried to resolve the phylogenetic relationship of higher taxa in the Asteroidea by analyzing 18S rDNA sequences. Although even in the analyses where 18S rDNA sequences were combined with those of mt. rDNAs, most of the interfamilial relationships were not resolved, the close relationship between Asterina and Solasteridae is consistently supported with relatively higher confidence values. The independent results based on nucleotide sequences of nuclear 18S rDNA and mitochondrial rDNAs support the close relationship between the five Asterina species and two species of Solasteridae (one Solaster and one Crossaster) (Fig. 1, 2). This is even strongly supported when the two datasets were combined (Fig. 3). While nine genera are recognized in the family Asterinidae (Clark and Downey, 1992), we investigated only the genus Asterina from Asterinidae. Thus, we cannot be certain whether the family Asterinidae is a monophyletic group. Since, in the analyses by Knott and Wray (2000), monophyly of the Asterinidae is not supported, we should be careful to regard Asterina as a representative of the family Asterinidae.
Lafay et al. (1995) analyzed the first 400bp of 28S rDNA in nine taxa. NJ and MP analyses supported the idea that Solasteridae species, represented by Crossaster, are more closely related to Henricia and Echinaster than to Asterina. However, the branches leading to Henricia and Echinaster are quite long, and this may disturb the position of these branches. In the ML tree, which is less sensitive to substitution rate heterogeneity, the sister group of Asterina and Crossaster was significantly supported. Thus, 28S rDNA analyses and the present results do not contradict each other. In analyses of mt. tRNAs and COI by Knott and Wray (2000), the close relationship between Asterinidae and Solasteridae is not recovered. However, in ML trees of their analysis, a sister group relationship between Patiriella (probably synonym of Asterina; Hart et al., 1997) and Solasteridae is suggested, although with bootstrap values of less than 50%. In fact, the phylogenetic status of either Asterinidae or Solasteridae is barely resolved in Knott and Wray (2000).
From the aspect of morphology, there is little support for the affinity between Asterinidae and Solasteridae. Regarding the Solasteridae, Blake (1981, 1987) and Gale (1987) agreed on its close relationship with Pterasteridae. However, no affinity is suggested between Solasteridae and Pterasteridae in the present analyses, neither in Knott and Wray (2000). Rather, NJ and ML results of total dataset in the present study suggest that Pteraster branched off from the rest of the species in the early stage of the asteroid evolution. Regarding the phylogenetic status of the Asterinidae, Blake (1987) classified it into the superfamily Ganeriacea with the other two families (Ganeriidae and Poraniidae). Although Gareriidae is not included in the present study, we could not detect any signals that suggest the phylogenetic affinity between Asterina and Porania.
Blake (1981) re-examined the phylogeny of Spinulosida, and found significant differences of three families including Asterinidae to the rest of the families of the Spinulosida, especially in ossicle morphology of ambulacral column, and thus classified them into different order Valvatida. However, it should be noted that in the same article he also pointed that the presence of metapaxilae on abactical ossicles favors the affinity of Asterinidae to Solasteridae and Echinasteridae (Blake, 1981). Since the phylogenetic status of the Echinasteridae was not resolved in the present study, we hesitate to stress the metapaxillae as a phylogenetically useful character. Careful observation may be necessary to evaluate the present results from a morphological point of view, and on the other hand, more robust phylogenetic framework should be recovered from molecular phylogenetic studies in order to examine phylohgenetic usefulness of each morphological character.
We thank Daisuke Fujita, Kazumasa Hashimoto, Shinji Kirihara, Atsushi Yoshinaka, Hiroshi Misaki, Yasuo Kano, Hidetomo Tanase, Taiji Yamamoto, Mitsuru Ohta and the members of Kushimoto Marine Park and Seto Marine Biological Laboratory, Kyoto University for providing specimens. This study was supported by a Toray Science and Technology Grant to H.W.