Ultrastructure and phylogenetic position of Chrysoculter rhomboideus gen. et sp. nov. (Prymnesiophyceae), a new flagellate haptophyte from Japanese coastal waters

T. Nakayama, M. Yoshida, M.-H. Noël, M. Kawachi and I. Inouye. 2005. Ultrastructure and phylogenetic position of Chrysoculter rhomboideus gen. et sp. nov. (Prymnesiophyceae), a new flagellate haptophyte from Japanese coastal waters. Phycologia 44: 369–383. We isolated a novel haptophyte alga from the coastal waters of the northern part of Japan. The cell is asymmetrical spindle-to knife-shaped and possesses two nearly equal flagella and a haptonema from the anterior tip. Two yellowish chloroplasts including immersed pyrenoids are situated asymmetrically. Two types of organic scales, small elliptical and large rhomboidal scales, cover the cell. The rhomboidal scale has two tubular projections at the longitudinal poles and closely touched to each other. The transition region of the flagellum includes only one transitional plate, which is probably homologous with the distal plate in other prymnesiophyceans. The flagellar apparatus of this alga has basic components found in other prymnesiophyceans. Developed root 1 (R1) is divided into two components that join again. The R1 extends toward the posterior tip of the cell with associated endoplasmic reticulum (ER) and forms a cytoplasmic tongue. A fibrous root, which has only been reported in some coccolithophorids, connects R1 and basal body 1. The proximal end of R1 is associated with the electron-dense plate and teeth-like structure. Four microtubules of the root 2 are arranged in an arc shape and have appendages. The free part of the haptonema includes five microtubules. In the traditional taxonomy, this alga is apparently a member of the genus Chrysochromulina because the cell possesses a long haptonema and no coccoliths. However, recent molecular phylogenetic studies have shown the polyphyly of the genus Chrysochromulina. Diagnostic characters of Chrysochromulina would be plesiomorphies of the Prymnesiophyceae. Phylogenetic analysis using 18S rDNA and rbcL sequences indicated that the alga reported here is not closely related to any other prymnesiophyceans, including Chrysochromulina, but suggested the basal position of the coccolithophorids. Based on these results, we propose Chrysoculter rhomboideus gen. et sp. nov. and Chrysoculteraceae fam. nov. for this unique haptophyte.


INTRODUCTION
The Haptophyta is an important algal group of photosynthetic primary producers especially in the open ocean. In addition to being photosynthetic, some haptophytes exhibit phagotrophy and are notable consumers in the ocean (e.g. Parke et al. 1956;Kawachi et al. 1991;Jones et al. 1994). Haptophyte algae are also known to play an important role in carbon and sulphur cycles. Some species of the Haptophyta (Isochrysidales and Coccolithales; called the coccolithophorids hereafter) produce body scales composed of calcium carbonate and are known as coccoliths. Although the haptophytes play important ecological roles in nature, their diversity and evolution including the origin of the coccolithophorids are not well understood. Recent molecular phylogenetic studies suggest that there are many undescribed haptophyte algae in the ocean (Edvardsen et al. 2000;Diéz et al. 2001;Moon-Van Der Staay et al. 2001).
2003). Species of Chrysochromulina are unicellular flagellates covered by organic scales but no coccoliths. Although members of Chrysochromulina have two flagella and a conspicuous haptonema as the common feature, they show notable diversity in cell shape, scale morphology, and ultrastructure (e.g. Manton & Leadbeater 1974;Green & Hori 1994;Birkhead & Pienaar 1995;Eikrem & Moestrup 1998;Puigserver et al. 2003). Furthermore, recent molecular phylogenetic studies have shown the polyphyletic nature of the genus Chrysochromulina (Edvardsen et al. 2000;Fujiwara et al. 2001). This evidence suggests that the genus Chrysochromulina is the ancestral stock of some haptophyte lineages, and so the phylogenetic study of Chrysochromulina is important to clarify the evolutional history of the Haptophyta.
We isolated from Japanese coastal waters an enigmatic haptophyte alga with two flagella, a haptonema, and organic scales. In this paper, we report on morphological and ultrastructural characteristics of this alga. The phylogenetic position of the alga was also analysed, based on 18S rDNA and rbcL sequences. In a traditional taxonomical sense, the alga studied in this paper can be classified as a species of Chrysochromulina. However, we propose a new genus and species, Chrysoculter rhomboideus gen. et sp. nov., based on its ultrastructural and molecular characters.    (2004). For light microscopical observations, a Nikon Optiphoto microscope (Nikon, Tokyo, Japan) equipped with differential interference contrast (DIC) optics was used. Whole mount preparations for transmission electron microscopy (TEM) were prepared using the method of Marin & Melkonian (1994). Material for thin sections was fixed in equal volumes of fixative (5% glutaraldehyde, a few drops of 4% OsO 4 , 0.5 M sucrose, in 0.1 M cacodylate buffer) for 1 h. After rinsing with 0.05 M cacodylate buffer, cells were postfixed in 2% OsO 4 at 4ЊC for 1 h, then rinsed once with the same buffer. Cells were embedded in Spurr's resin (Spurr 1969) after dehydration in a graded ethanol series. Sections were cut with a diamond knife and double stained with 2% uranyl acetate and lead citrate (Reynolds 1963). Observations were carried out with a Jeol JEM 100CXII TEM (Jeol, Tokyo, Japan). Material for scanning electron microscopy (SEM) was fixed with 5% glutaraldehyde for 2 h and adhered to a SEM plate treated with 0.1% poly L-lysine. After rinsing thrice with 0.2 M cacodylate buffer, cells were postfixed in 1% OsO 4 for 30 min, then rinsed once with the same buffer. After dehydration in a graded ethanol series, absolute ethanol was replaced with tbutyl alcohol. The samples were dried in a VFD-21S (Shinku-Device, Ibaraki, Japan) and coated with platinum-palladium using an ion sputter E-102 (Hitachi, Tokyo, Japan). For SEM observations, a JEOL JSM-6330F (Jeol) was used.

Chrysoculter rhomboideus
Cells solitary, mostly asymmetric knife-shaped, 10-16 m long, 3-6 m wide. Two equal to subequal flagella, 14-26 m, inserted at the cell apex. Haptonema 4-10 m, noncoiling. Cells covered by two types of scales. Scales of the inner layer elliptical, 0.35 ϫ 0.18 m, with a narrow inflexed rim, and a radial pattern. Scales of the outer layer rhomboidal, 1.1 ϫ 0.6 m, with upright rim (50-80 nm), two projections, and a radial pattern. Rhomboidal scales do not overlap. Two pale yellow chloroplasts, lateral and parietal, are displaced horizontally, each with an immersed pyrenoid. The flagellar transitional region with a terminal transitional plate.
A unialgal culture used in this study is deposited at the National Institute of Environmental Studies, Japan as P1544.
ETYMOLOGY: The generic name refers to the colour and unique asymmetrical cell shape (chryso ϭ golden, culter ϭ knife), and the specific epithet refers to the shape of the large organic scales (rhomboideus ϭ rhomboidal).

Cell structure
Living cells of C. rhomboideus are usually slender spindle-to knife-shaped, 10-16 m long, and 3-6 m wide (Figs 1, 2, 5-7, 12). Cells are asymmetric in outline; one lateral side is somewhat swollen, and the opposite lateral side is flattened (Figs 2, 5, 12). Small cells with an asymmetrical ovoidal outline were sometimes observed (Fig. 3). The posterior end of the cell sometimes protruded as a short hyaline tail (Fig. 5). Two equal to subequal flagella (14-26 m long) and a haptonema (4-10 m long) emerged from the anterior tip of the cell (Figs 1-7, 12, 13). We did not observe a coiling haptonema. This alga was not an active swimmer, but it sometimes attached to the substratum by the anterior side of the cell or flagella. Two yellowish parietal chloroplasts were situated asymmetrically. One was situated at the anterior-lateral side of the cell, and the other was positioned at the posterior, op- posite lateral side (Figs 1, 2, 5, 12, 13). The anterior chloroplast was usually situated in the swelling (Figs 2, 5, 12). The chloroplast contained three thylakoid lamellae but no girdle lamella and it was covered by periplastidal endoplasmic reticulum merging with the nuclear envelope (Figs 12, 14). Each chloroplast included an immersed pyrenoid traversed by a pair of thylakoids (Fig. 14). The nucleus was situated at the centre of the cell (Figs 1, 5, 12). Sections showed mitochondrial profiles with tubular cristae that probably represented a large reticulate mitochondrion (Figs 12,13,15). A Golgi body with typical dilated cisternae was located between the nucleus and the basal bodies, and the maturing face opened into the area between the basal bodies and the posterior chloroplast (Figs 12,13,52). Large posterior vacuoles without conspicuous contents were located at the posterior part of the cell (Fig.  13). Smaller vacuoles containing some material were present in the posterior to central part of the cell, and this type of vacuole was usually associated with rough endoplasmic reticulum (Figs 13, 15). Peripheral endoplasmic reticulum (PER) was distributed beneath the cell membrane (Figs 15,38,58). C. rhomboideus reproduced by binary fission of the swimming cell (Fig. 4), and sexual reproduction or occurrence of different stages was not observed.

Cell covering
One of the most distinctive features of C. rhomboideus was two layers of organic body scales. The inner layer consisted of scattered small elliptical scales (0.35 ϫ 0.18 m) with narrow rims 69). Some images suggested that the small elliptical scales were distributed beneath the border of the outer scales (Figs 9, 10). The large rhomboidal scales (1.1 ϫ 0.6 m) of the outer layer had upright rims (50-80 nm) (Figs 9-11, 69). The rim of the rhomboidal scale has two tubular projections (0.3 m long) at the longitudinal poles of the scale (Figs 9-11, 69). The large rhomboidal scales over-lapped each other in a regular arrangement (Figs 6-8). Scales seemed to form a scale casing around the cell (Fig. 8), but they probably were not tightly attached to each other because shedded single scales were frequently observed. Both types of scales had faint radial but no concentric fibrils (Figs 9-11, 69). The scaly covering sometimes extended to the haptonema and the posterior extension (Figs 3, 7). Fibrous material covered the proximal portion of the flagella (Figs 16, 17). This material may be used to attach the cell to the substratum.
No close relationship between R1 and mitochondrial profiles was observed around the flagellar apparatus. The root 2 (R2) comprised four MTs arranged in an arc-shaped structure at the proximal end (Figs 39,40,(43)(44)(45)64,65). Each microtubule of R2 possessed a short appendage extending to the cavity of the R2 arc (Fig. 39). The R2 originated from the area between the two basal bodies and was closely related (but never connected directly) to the electron-dense material, which is probably an extension of the distal fibre (Figs 39, 43, 64). The arc shape of R2 collapses very soon and the root terminates (Figs 31,41,42). Root 3 (R3) originates from the side of BB2 away from the haptonema and at first comprises two MTs attached to BB2 by electron-dense material (Figs 63, 64, 66, 71). The R3 extended along the inner surface of the anterior chloroplast, and an extra microtubule was soon added (Figs 62, 67,  70). Near the proximal end of R3, an electron-dense lump was situated on BB2 (Figs 27,33,35,36,39,43,59,60,64,70,71). Part of this lump extended under R2, but never attached directly (Figs 39, 43, 64). One section showed an opaque connection between the electron-dense lump and BB1 (Fig. 60). Root 4 (R4) was single-stranded and emanated from the area between hf1 and hf3 (Figs 62, 63, 68). It did not join up with R3 (Figs 59-61). No crystalline root or its homologue associated with the microtubular roots was observed.

Molecular phylogenetic analyses
Phylogenetic analyses based on the 18S rDNA sequences resulted in a tree similar to that of Edvardsen et al.    (Figs 60, 61). Note an opaque connection between the electron-dense lump and BB1 (arrowhead in Fig. 60). for these relationships were relatively low (Ͼ 77%). Phylogenetic analysis including partial 18S rDNA sequences reported by Diéz et al. (2001) also showed no close relatives of Chrysoculter (not shown). The MP analysis based on the partial sequences of rbcL resulted in a single most parsimonious tree (length ϭ 1686, consistency index ϭ 04075, retention index ϭ 0.4165) (not shown). The hierarchical likelihood ratio test selected the general time reversible model (Rodríguez et al. 1990) with an estimated proportion of invariant sites (0.5788) and rate heterogeneity among variable sites approximated as a discrete gamma distribution (2.7228) (GTR ϩ I ϩ G). Base frequencies were A ϭ 0.2706, C ϭ 0.1871, G ϭ 0.2031, T ϭ 0.3391; rate matrix was A-C ϭ 0.3242, A-G ϭ 6.4582, A-T ϭ 8.8998, C-G ϭ 0.8879, C-T ϭ 7.2952, G-T ϭ 1.0000. The ML tree is shown Fig. 73. All analyses generated similar trees, in which Chrysoculter was distantly related to Chrysochromulina species and formed a clade with the Coccolithales and Isochrysidales (bootstrap values less than 50%). In the rbcL trees, the monophyly of Chrysochromulina sensu stricto (C. alifera ϩ C. parva), position of Imantonia, and relationship within the clade coccolithophorids ϩ Chrysoculter were not settled.

DISCUSSION
Chrysoculter rhomboideus is apparently a member of the Haptophyta as it possesses (1) a haptonema, (2) two yellowish chloroplasts with periplastidal endoplasmic reticulum and no girdle lamella, and (3) a system of PER. Two nearly equal flagella with no appendages and organic scales with radial ribs covering the cell indicate that it is a member of the Prymnesiophyceae (sensu Edvardsen et al. 2000). Some ultrastructural features, such as splitting R1 Birkhead & Pienaar 1995;Eikrem & Moestrup 1998), the arcshaped R2 (Eikrem & Moestrup 1998), and the additional microtubule on R3 (Green & Hori 1994) also support inclusion of Chrysoculter in the Prymnesiophyceae. Flagellate prymnesiophycean algae possessing a conspicuous haptonema and various types of organic scales but no coccoliths have been classified in the genus Chrysochromulina. Chrysoculter rhom- boideus would be classified in the genus Chrysochromulina in this sense. However, the genus Chrysochromulina is very diverse morphologically (e.g. Green & Hori 1994;Birkhead & Pienaar 1995;Eikrem & Moestrup 1998;Eikrem & Edvardsen 1999). Recent molecular phylogenetic studies support the polyphyletic nature of the genus Chrysochromulina (Edvardsen et al. 2000;Fujiwara et al. 2001) and indicate that the genus should be recircumscribed and divided into several genera (Edvardsen et al. 2000). Because the type species, C. parva Lackey, has a saddle-shaped cell with a long haptonema, Chrysochromulina sensu stricto should be limited to the saddle-shaped species such as C. acantha Leadbeater & Manton, C. campanulifera Manton & Leadbeater, C. scutellum Eikrem & Moestrup, and C. throndsenii Eikrem. In the 18S rDNA tree, these species (C. parva has not been studied yet) form a robust clade distantly related to C. rhomboideus. Furthermore, we found no ultrastructural apomorphic feature common to the saddle-shaped Chrysochromulina and Chrysoculter. The alga therefore cannot be included in the genus Chrysochromulina. In the molecular phylogenetic trees, most of other 'Chrysochromulina' species (e.g. C. hirta Manton, C.  Fujiwara et al. 2001) with Prymnesium Massart. Although the apomorphic character of this clade is uncertain, there is no significant ultrastructural similarity between Chrysoculter and the algae in clade B1. The 18S rDNA analysis also places C. rhomboideus distantly from the clade B1 and from any other prymnesiophyceans analysed. Based on this evidence, we propose the new genus, Chrysoculter. Most species of the genus Chrysochromulina are characterized by their scale morphology, and some species were described based on only the scales. The large rhomboidal scale of C. rhomboideus is a very distinctive character, and there is no previous report of such a unique scale. The asymmetrical knife-shaped cell is also a characteristic feature that has not been reported previously.
In addition to the characters mentioned above (scale and cell morphology, gene sequences), certain ultrastructural characters also support the isolated phylogenetic position of Chrysoculter in the Prymnesiophyceae. A single transitional plate in the transition region of the flagellum is one of the most distinctive characters of Chrysoculter. Most members of the Prymnesiophyceae have two, a distal and proximal, transitional plates (Moestrup 1982;Preisig 1989;Green & Hori 1994). Coccolithophorid species have been reported to have a single transitional plate with or without the helical bands Hori & Green 1991;Kawachi & Inouye 1994;Sym & Kawachi 2000). Because of its position in the flagellum and its morphology (i.e. a perforated septum with an axosome), the single transitional plate of the coccolithophorids is probably homologous with the proximal transitional plate of other prymnesiophyceans. On the other hand, the single transitional plate of Chrysoculter is apparently homologous with the distal transitional plate of other prymnesiophyceans. The loss of the proximal transitional plate is unique among members of the Prymnesiophyceae, and it is an autapomorphic characteristic of Chrysoculter. Although the fragile cylindrical structure found in the transition region of Chrysoculter has not been reported previously, a similar struc-  Scherffel (figs 26, 27 in Parke et al. 1971). Interestingly, the position of this structure corresponds to the place where flagellar shedding takes place in the Prymnesiophyceae (Eikrem & Moestrup 1998) and to the place where the stellate structure is situated in the Viridiplantae (Sanders & Salisbury 1989). This structure may be more common, although it is difficult to detect because of its tenuous nature. The teeth-like structure on the electron-dense plate overlying the R1 microtubular sheet is another distinctive feature of Chrysoculter. It has not been reported in other haptophytes. However, the thinness of the structure (it appears in only one or two sections) could cause it to be overlooked in other species. Billard (1994) proposed that the haptophyte life cycle in-cludes a haploid cell with dimorphic scale faces and a diploid cell with monomorphic scale faces. Recent findings on Chrysochromulina polylepis and Prymnesium parvum N. Carter reinforce this idea (Edvardsen & Paasche 1992;Edvardsen et al. 1996;Edvardsen & Medlin 1998;Larsen 1999), and morphologically distinct haptophytes sometimes represent different generations of the same species (Parke & Adams 1960;Gayral & Fresnel 1983a;Green et al. 1996;Noël et al. 2004). Thus, C. rhomboideus may represent a generation of a known species. There is superficial similarity between the organic scales of C. rhomboideus and the coccoliths of some heterococcolithophorids such as Anoplosolenia Deflandre and Calciosolenia Gran. However, the monomorphic scale face of C. rhomboideus indicates that this is a diploid generation. This is the same ploidy stage as a heterococcolithophorid, and it is therefore unlikely that Chrysoculter is a stage in the life cycle of a heterococcolithophorid. Neither the production of the coc- coliths nor morphologically different phases were observed in this study. We therefore consider C. rhomboideus to be a distinct, new species. Interestingly, some ultrastructural characters suggest phylogenetic affinity between Chrysoculter and coccolithophorids. The fibrous root connecting BB1 and the posterior component of R1 is a notable feature of Chrysoculter. This structure, called F1 by Roberts & Mills (1992), was described also from the Coccolithales (Inouye & Chihara 1983;Inouye & Pienaar 1985;Roberts & Mills 1992;Kawachi & Inouye 1994) and the Isochrysidales (Hori & Green 1991). Although Green & Hori (1990 noted the presence of F1 in P. parvum (as P. patellifera Green, Hibberd & Pienaar), the published micrograph ( fig. 6h in Green & Hori 1990) shows that this fibrous structure originates from the side of BB1 facing the microtubular sheet of R1. In the coccolithophorids, F1 emerges from the side of BB1 away from the haptonema. So, the true F1 is probably a synapomorphic character of the coccolithophorids (Coccolithales and Isochrysidales). The occurrence of F1 in Chrysoculter, although it is less developed, suggests a close relationship to the coccolithophorids. Although the bootstrap values were low, this relationship was also shown in the molecular analyses. The evidence suggests that Chrysoculter is derived from the common ancestor of the coccolithophorids before the ability to form coccoliths was developed. Other ultrastructural characters also support this hypothesis. Thus both Chrysoculter and some members of the Coccolithales and Isochrysidales have an electron-dense plate located next to the R1 microtubular sheet on the side facing BB1 (Inouye & Pienaar 1985, 1988Hori & Green 1991). Other prymnesiophyceans have no such a structure. However, it should be noted that some prymnesialean algae possess a flange-like structure on the R1 microtubule, and longitudinal sections of R1 sometimes show structure of similar appearance to the electron-dense plate in the coccolithophorids (Green & Hori 1990;Birkhead & Pienaar 1994a, 1995Eikrem & Moestrup 1998). The number of microtubules in R2 is usually one to three in prymnesiophyceans except in the Coccolithales (Green & Hori 1994;Eikrem & Moestrup 1998). In most coccolithophorids studied R2 comprises four microtubules as in Chrysoculter (Inouye & Pienaar 1984, 1988Kawachi & Inouye 1994;Sym & Kawachi 2000). Green & Hori (1990) reported appendages on the R2 microtubules of P. parvum. However, the appendages of P. parvum and Chrysoculter are situated on opposite sides. Interestingly, published micrographs of some coccolithophorids suggest that appendages are present on R2 as in Chrysoculter ( fig. 21 in Inouye & Chihara 1983, fig. 23 in Inouye & Pienaar 1984, fig. 18 in Inouye & Pienaar 1985. The numbers of microtubules in the free part of the haptonema vary in the Haptophyta. Most prymnesialean, phaeocystidalean, and pavlovophycean species investigated have six or seven microtubules in the haptonema (reviewed by Edvardsen et al. 2000). No more than five microtubules have been reported in some coccolithophorids (Inouye & Chihara 1983;Hori & Green 1991;Roberts & Mills 1992;Kawachi & Inouye 1994;Sym & Kawachi 2000), and in this respect Chrysoculter resembles the coccolithophorids. However, some species of the Coccolithales also have haptonema with six or seven microtubules (Inouye & Pienaar 1985, 1988. In conclusion, the fibrous root (F1), the electron-dense plate on the R1 microtubular sheet, R2 of four microtubules with appendages, and the low number of haptonema microtubules suggest a phylogenetic relationship between Chrysoculter and coccolithophorids. Some (or all) of these characters are synapomorphies of the clade that includes Chrysoculter and coccolithophorids.
A compound structure composed of R1 microtubules and PER, the cytoplasmic tongue, has been considered a characteristic feature of the coccolithophorids (Gayral & Fresnel 1983a;. This structure is also present in Chrysoculter. However, the cytoplasmic tongue has now been reported from some members of the B1 clade, such as P. nemamethecum (Birkhead & Pienaar 1994a;Pienaar & Birkhead 1994) and 'eyelash' Chrysochromulina (Birkhead & Pienaar 1995). This structure may be a synapomorphy of the lineage including coccolithophorids, Chrysoculter, and the B1 clade, although some species have subsequently lost it again. Another character shared with the coccolithophorids and B1 clade is the compound R1. An R1 with accessory bundles of microtubules has been reported from some species of Prymnesium (Birkhead & Pienaar 1994a), Platychrysis (Gayral & Fresnel 1983b), Chrysochromulina (except Chrysochromulina sensu stricto; Birkhead & Pienaar 1995), and many coccolithophorids (e.g. Gayral & Fresnel 1983a;Inouye & Chihara 1983;Inouye & Pienaar 1984Roberts & Mills 1992;Kawachi & Inouye 1994), and it may be another apomorphy shared between the coccolithophorids and the B1 clade. Chrysoculter lacks a compound R1, but some coccolithophorids and the B1 clade also have very reduced or no compound root (e.g. Green & Hori 1986Inouye & Pienaar 1988;Sym & Kawachi 2000). The frequent loss of the compound root has occurred also in the Prymnesiophyceae. A close relationship between the cell cycle and the development of the compound R1 also makes it difficult to judge the phylogenetic relevance of this characteristic Birkhead & Pienaar 1995).
The phylogenetic considerations based on ultrastructural and molecular evidence have induced us to propose a new monotypic family, Chrysoculteraceae, for C. rhomboideus. A new order may also have to be created for this alga, but we prefer to postpone the erection of a new order until the relationships and taxonomic relevance of the prymnesiophycean orders have been clarified better. Swimming cell with two equal flagella and a haptonema. Chloroplasts two, parietal, with pyrenoid traversed by a pair of thylakoids. Cell body covered with scales of two types. R1 with fibrous root (F1), thin plate, and teeth-like structure. Transitional region includes a terminal transitional plate.