Introduction

Trichocysts are ejective organelles found in cryptomonads, dinoflagellates, and the prominent peniculine ciliates, which form needle-like, hair-like, or tubular structures when they are discharged from the cell. However, as the ultrastructure of trichocysts in peniculine ciliates and dinoflagellates greatly differs from that of cryptomonads, it has been thought that they have independently evolved, but it has been remained to compare by molecular and biochemical approaches.

Cryptomonads are unicellular biflagellate algae common in marine and freshwater habitats. Cells are typically ovoid to bean shaped, and have an asymmetric cell construction with a ventral furrow/gullet system. Fourteen genera of cryptomonads can be delineated based on both ultrastructural and molecular characters. As a matter of convenience, genera are informally grouped according to pigmentation: red colored genera (Pyrenomonas, Storeatula, Rhinomonas, Proteomonas, Geminigera, Teleaulax, Guillardia, Plagioselmis, Hemiselmis red species); brown colored genera (Cryptomonas); blue/green colored genera (Chroomonas, Komma, Falcomonas, Hemiselmis blue/green species); and colorless genera (Goniomonas). Throughout all these genera, the presence of trichocysts (= ejectisomes) is a characteristic feature. Cryptophyte trichocysts have been well documented as to general appearance and distribution in the cell (Dragesco 1951; Anderson 1962; Schuster 1968; Wehrmeyer 1970; Hausmann 1978). The trichocysts occur in two ranges of size according to their location in the cell. A group of large trichocysts of uniform size (400–600 nm in diameter) are found in several rows, outlining the gullet area (which is the oblique apical depression characteristically present in the majority of species). Smaller trichocysts, often as small as one half the diameter of the large ones, are located directly below the periplast and are distributed throughout the surface of the cell. The undischarged trichocysts contain two cylindrical coiled ribbons, each a tightly wound coil of tapering ribbon having the narrow end of the ribbon at the centre of the coil. The minor coil is trucked at an angle into the funnel-shaped depression at the distal end of the major coil and is attached to it by a special connecting piece. When the trichocyst discharges from the cell, a straight or slightly curved “telescopic” hollow tube is formed by the lateral inward rolling of the edges of the ribbon (Hovasse et al. 1967; Morrall and Greenwood 1980). The main shaft of the tube is formed from the major coiled ribbon, and the minor coil forms a similar shorter tube that remains attached to the broad end of the main shaft and is carried at a distinctive angle, giving to it a characteristic beak-like tip. Although the role of this exocytotic response remains unclear, it has been proposed that the discharge of the coiled ribbons from the trichocysts has a defensive role against predation by protozoa.

The katablepharids are a well-defined group of colorless heterotrophic flagellates; however, their taxonomic and phylogenetic position is inconclusive. Based on their cell shape, flagellar orientation, and the presence of trichocysts, they have been classified in Cryptophyceae (Cryptophyta) (Bourrelly 1970), but electron microscopy suggests that the katablepharids are distinct from cryptomonads. A possible affinity with the Alveolata is based on the resemblance of their feeding apparatus to the apical complex of the Apicomplexa (Clay and Kugrens 1999; Leander and Keeling 2003). Okamoto and Inouye (2005) suggested that the katablepharids are a distinct sister group of the Cryptophyta and proposed a new division/phylum based on molecular phylogenetic analyses of SSU rDNA and beta-tubulin. The katablepharids have trichocysts that contain cylindrical coiled ribbons similar to those of cryptomonads, but when discharged, these trichocysts display different ways of rolling into a tube (Kugrens et al. 1994). Molecular analyses of trichocysts would provide valuable insights into taxonomic and phylogenetic position of the katablepharids.

The trichocysts of cryptomonads are found across the taxon. Although the genus Pyramimonas belongs to the Prasinophyceae (Chlorophyta), a class of green flagellates possessing chlorophyll b and therefore related to the green algae and land plants, some Pyramimonas species also have trichocysts that contain coiled ribbon similar to cryptomonads. However, the manner of rolling into a tube differs from cryptomonads and katablepharids, that is, in Pyramimonas, the discharged trichocyst forms a tube by spiral rolling of the ribbon, whereas in cryptomonads and katablepharids, a tube is formed by the lateral inward rolling of the edges of the ribbon (Kugrens et al. 1994). On the other hand, the trichocysts found in cryptomonads have been suggested to show morphological similarity to R-bodies (Anderson et al. 1964; Hausmann 1978; Hovasse 1965), which are unusual inclusion bodies found in a few gram-negative bacterial species. They are highly insoluble protein ribbons, generally seen coiled into cylindrical structures within cells, as are the trichocysts of cryptomonads. It has been noted that R-bodies unroll in either of two ways, from the outside or from the inside, in a telescopic fashion, under certain conditions (Pond et al. 1989). R-bodies are found in all Caedibacter species and in some Pseudomonas species (Fusté et al. 1986; Lalucat et al. 1979; Quackenbush 1978, 1987; Wells and Horne 1983). Members of the genus Caedibacter are obligate endosymbionts of paramecia, commonly referred to as kappa particles, and easily distinguished from other bacterial endosymbionts of paramecia by their ability to produce R-bodies. In contrast, R-body-producing Pseudomonas are free living. The paramecia that contain Caedibacter cells producing the R-bodies kill the paramecia that do not contain endosymbionts or that contain endosymbionts that are not the same species of Caedibacter bacteria, and a Caedibacter mutant defective in R-body production does not exhibit killer traits, suggesting that R-bodies play an important role in the killing process of other paramecia (Dilts and Quackenbush 1986; Pond et al. 1989; Preer et al. 1974). A series of experiments with a recombinant R-body-encoding plasmid derived from pKAP47 showed that the rebA, rebB, rebC, and/or rebD genes of Caedibacter are involved in R-body synthesis and assembly, and they are present on plasmids (Jeblick and Kusch 2005; Heruth et al. 1994).

Thus, coiled ribbon structures have been found not only throughout the cryptomonads but also in some green algae and proteobacteria. Identification of trichocyst-related proteins would be a first step in elucidating evolutionary of trichocysts and the physical properties of trichocyst ribbons, in particular their ability to assume a hollow tubular form. Recently, Rhiel and Westermann (2011) reported that the trichocyst ribbon proteins from Chroomonas and Cryptomonas species were dominated by polypeptides of 40–44, 23–25, and 16–18 kDa. However, trichocyst-related proteins have not yet been identified. In this study, we aimed to identify the trichocyst component proteins in cryptomonads. It has been very difficult to perform biochemical analyses of cryptomonads due to their generally low growth rates. However, by using the red cryptomonad, Pyrenomonas helgolandii, which shows a relatively high growth rate (Yamagishi and Kawai 2011), we could identify four trichocyst-related proteins and here provide new knowledge on cryptomonads trichocysts.

Materials and Methods

Culture

Pyrenomonas helgolandii (SAG 28.87) was cultured in Provasoli’s enriched seawater (PES; Provasoli 1966) at 20 °C illuminated by white fluorescent lighting (approximately 40 μmol photons m−2 s−1) with a 12:12 light:dark (L:D) photoperiod.

Electron Microscopy

Pyrenomonas helgolandii cells were collected by centrifugation at 4 °C, 500×g for 10 min. Pellets of cells were fixed with 2 % glutaraldehyde (TAAB, Aldermaston, UK) in 0.2-M cacodylate buffer containing 0.4-M sucrose and 5-mM EGTA for 1 h at 4 °C. After three rinses with 0.1-M cacodylate buffer containing 2.5-mM EGTA, pellets were post-fixed with 1 % osmium tetroxide (OsO4) in 0.1-M cacodylate buffer containing 2.5-mM EGTA for 1 h at 4 °C, followed by three rinses with 0.2-M cacodylate buffer containing 5-mM EGTA. The material was then dehydrated in a graded ethanol series, then an acetone series, and embedded in Spurr’s resin (Spurr 1969). Sections cut using a diamond knife with a Reichert ULTRACUT microtome (Reichert-Jung, Vienna, Austria) were double stained in 2 % uranyl acetate for 15 min followed by lead citrate (Reynolds 1963) for 10 min, then observed with a JEM-1010D transmission electron microscope (JEOL, Tokyo, Japan). For whole-mount staining, isolated trichocyst ribbons were directly mounted onto formvar-coated grids and dried, followed by a rinse with distilled water. Dried materials were stained in 2 % uranyl acetate for 5 min, and after rinsing with distilled water, observed with a JEM-1010D transmission electron microscope.

Isolation of Trichocyst Ribbons

Pyrenomonas helgolandii cells were collected by centrifugation at 4 °C, 500×g for 10 min, followed by washing once with fresh PES medium by centrifugation at 4 °C, 500×g for 10 min. The cells were resuspended in 15 mL PES medium and incubated at 4 °C for 1 h, and then the flagella were detached by agitation of the centrifuge tube with a vibratory mixer (Taitec, Tokyo, Japan) for 2 min. Deflagellated cell bodies were collected by centrifugation at 4 °C, 500×g for 10 min and then resuspended in 15 mL PES medium with pH lowered to 3.5 by addition of HCl. Almost all trichocyst ribbons were discharged from the cells in approximately 10 s. The pH of the suspension was immediately raised to 7.0 by addition of 4-N NaOH and the cell bodies were removed from medium by centrifugation, five times at 4 °C, 500×g for 10 min. If cell bodies were still included in the supernatant, additional centrifugations were performed until they were completely removed. The supernatant containing the trichocyst ribbons was centrifuged at 4 °C, 13,000×g for 30 min to form a pellet of the trichocyst ribbons. The purified trichocyst ribbons were stored at −80 °C until use.

Tricine–Sodium Dodecyl Sulfate-Polyacrylamide Gel Electrophoresis

Preparations of trichocyst ribbons were suspended in an SDS buffer (0.05 M Tris–HCl at pH 6.8, 2 % w/v SDS, 1 % w/v dithiothreitol, 10 % v/v glycerol) and incubated at 95 °C for 5 min, followed by centrifugation at 4 °C, 13,000×g for 30 min. Tricine–sodium dodecyl sulfate-polyacrylamide gel electrophoresis was performed on the supernatant to separate small proteins less than 10 kDa according to the method of Schägger and Jagow (1987). Resolved protein bands were stained with 0.25 % Coomassie brilliant blue R-250 in 30 % methanol and 10 % acetic acid and destained in 30 % methanol and 10 % acetic acid.

Two-Dimensional Gel Electrophoresis

Trichocyst ribbons were solubilized with alkaline urea buffer following Horst et al. (1980). An Immobiline DryStrip (7 cm, pH 3–10) (GE Healthcare UK, Buckinghamshire, UK) was rehydrated overnight in 125 μL of alkaline urea buffer containing trichocyst proteins. Isoelectric focusing (IEF) electrophoresis was performed with an Ettan IPGphor III system (GE Healthcare UK) according to the manufacturer’s instruction. The voltage was increased in a stepwise manner: 300 V for 40 min; 1,000 V for 300 Vh; 5,000 V for 4,000 Vh; 5,000 V for 12 s. After IEF electrophoresis, the gel was incubated with shaking for 15 min at room temperature in equilibration buffer (1.5 M Tris–HCl, pH 8.8, 6.0 M urea, 30 % glycerol, 2.0 % SDS, and 0.002 % bromophenol blue) containing 1.0 % dithiothreitol. The proteins in the equilibrated gel were separated using tricine–sodium dodecyl sulfate-polyacrylamide gel electrophoresis as previously described. Resolved protein spots were stained with Coomassie brilliant blue R-250 and visualized as previously described.

N-Terminal Amino Acid Sequencing

The proteins separated by 2D gel electrophoresis were electrophoretically transferred to polyvinyliden difluoride (PVDF) membranes (Clear Blot Membrane-p, Atto, Tokyo, Japan). The membrane was stained with 0.25 % Coomassie brilliant blue R-250 in 5 % methanol and 7.5 % acetic acid, then destained with methanol. The spots were excised and subjected to N-terminal amino acid sequencing (Hokkaido System Science, Hokkaido, Japan).

Isolation of Pyrenomonas helgolandii Trichocyst cDNAs

Total RNA was extracted from P. helgolandii with an RNeasy Plant Mini Kit (Qiagen, Hilden, Germany), according to the manufacturer’s instructions. The cDNA was derived from total RNA of P. helgolandii using reverse transcriptase (RT), and 3′ rapid amplification of the cDNA end (3′ RACE) was performed with a 3′ RACE System for Rapid Amplification of cDNA Ends (Invitrogen, Carlsbad, CA, USA), according to the manufacturer’s instructions. The degenerate primer sets for 3′ RACE were designed from N-terminal amino acid sequences of Tri1, Tri2, Tri3-1, and Tri3-2 as follows: 5′-CCNAAYGGNGGNMGNAAYGAYCC-3′ for the first round PCR and 5′-ATGTAYGAYACNAARATGGG-3′ for the nested PCR in the case of Tri1, and 5′-GGNAARAARGCNGCNAAYYTNTAYAC-3′ for the first round PCR and 5′-GCNTTYGTNCARAAYACNCARGC-3′ for the nested PCR in the case of Tri2, and 5′-TAYGGNCARACNGCNGCNGC-3′ for the first round PCR and 5′-TAYGCNAARGCNGCNGGNCARAA-3′ for the nested PCR in the case of Tri3-1, and 5′-TTYGAYYTNGARGAYTAYGCNGG-3′ for the first round PCR and 5′-ACNTAYMGNGARACNMGNAAYGC-3′ for the nested PCR in the case of Tri3-2. The PCR products were cloned into the plasmid vector pGEM-T Easy (Promega, Madison, WI, USA). Plasmid DNAs were prepared from positive clones with a Mini Prep Kit (Promega) and sequenced with a CEQ8000 sequencer (Beckman Coulter, Fullerton, CA, USA). To generate 5′ cDNA sequences, 5′ rapid amplification of the cDNA end (5′ RACE) was performed with a 5′ RACE System for Rapid Amplification of cDNA Ends (Invitrogen), according to the manufacturer’s instructions. The primer sets for 5′ RACE were designed from determined 3′ sequences of Tri1, Tri2, Tri3-1, and Tri3-2 as follows: 5′-CCATCATGTATTGCTTGC-3′ for reverse transcription, 5′-GATATTGCCATGACCACAACCC-3′ for the first round PCR and 5′-GTTTATGTGCTGCACGATGTAC-3′ for the nested PCR in the case of Tri1, and 5′-TAGACCAACAGCCGAAG-3′ for reverse transcription, 5′-CTAACCCCTTGTGTCAAAACAC-3′ for the first round PCR and 5′-TTCGTTCGACTGGTCCCTCGGG-3′ for the nested PCR in the case of Tri2, and 5′-TCTCAGAACAGCACGGT-3′ for reverse transcription, 5′-CGACCAAACTACGCCTTGAGCG-3′ for the first round PCR and 5′-CGCCGGCAACCAGCTCTAAACC-3′ for the nested PCR in the case of Tri3-1, and 5′-AGATACATCAATTTATG-3′ for the reverse transcription, 5′-TAGATGATGAATCGGATGAACG-3′ for the first PCR and 5′-TTGTTGCCAAGCGCCGACCAAAC-3′ for the nested PCR in the case of Tri3-2. The PCR products were cloned and sequenced as previously described.

Genomic Sequencing

Total genomic DNA was prepared from Pyrenomonas helgolandii with a DNeasy Plant Mini Kit (Qiagen), according to the manufacturer’s instructions. To amplify genomic sequences of Tri1, Tri2, Tri3-1, and Tri3-2, four primer sets were designed as follows: 5′-ACCAGCTTCGCCATGCTCTTTC-3′ (forward) and 5′-CCATAGTTTATGTGCTGCACGATG-3′ (reverse) for Tri1, and 5′-CAAACTTCAGTATGCGCACTGTTC-3′ (forward) and 5′-GTACAACAGCTTAATGCCTGTTG-3′ (reverse) for Tri2, and 5′-CGAAAGATGTTCCAGACTCTTTAC-3′ (forward) and 5′-CTCTCAGAACAGCACGGTCTGCG-3′ (reverse) for Tri3-1, and 5′-CACAAGATGATCGCCAAGCTT-3′ (forward) and 5′-CAGATACATCAATTTATGAGTA-3′ (reverse) for Tri3-2. The PCR products were cloned and sequenced as previously described.

Nucleotide Sequence Accession Number

The accession numbers for Tri1, Tri2, Tri3-1, and Tri3-2 are JF826281, JF826282, JF710318, and JF710319, respectively, which will appear in the DNA Data Bank of Japan (DDBJ), GenBank, and European Molecular Biology Laboratory (EMBL) nucleotide data bases.

Sequence Alignment and Bioinformatic Analysis

Multiple amino acid alignments were performed by using ClustalW (1.83, DDBJ). Hydropathy profiles were analyzed with ProtScale (Gasteiger et al. 2005), using the Kyte-Doolittle algorithm and a window of nine amino acid residues. Secondary structure predictions were made using Chou-Fasman algorithm (Chou and Fasman 1974a, b).

Results

Trichocysts lie outlining the gullet in Pyrenomonas helgolandii, and small trichocysts at the cell surface can scarcely be recognized with the optical microscope (Fig. 1a, b). Trichocysts fixed by the standard glutaraldehyde and osmium tetraoxide (OsO4) method were observed as a coiled ribbon of concentric layers, which appeared densely osmiophilic, however, their substructures could not be recognized (Fig. 1c). The ribbon-shaped structure contained in the trichocyst was easily discharged by acid stimulation without cellular damage; therefore, the isolated trichocyst preparations were nearly free of other cellular material (Fig. 2a–c). The discharged trichocyst content was observed as an unrolled ribbon and measured ca. 80 nm in diameter (Fig. 2c).

Fig. 1
figure 1

Trichocysts in Pyrenomonas helgolandii. Arrowheads indicate trichocysts. a DIC image of entire cell. b Magnified DIC image of trichocysts. c TEM image of a trichocyst in longitudinal section. Bar = 5 μm (a); 2.5 μm (b); 0.2 μm (c)

Full size image
Fig. 2
figure 2

Trichocyst ribbons discharged from Pyrenomonas helgolandii cells. a DIC image of trichocyst ribbons discharged from P. helgolandii. b, c Electron micrograph of negatively stained trichocyst ribbons. Bar = 5 μm (a); 0.5 μm (b, c)

Full size image

Isolated trichocyst ribbons were analyzed by tricine–sodium dodecyl sulfate-polyacrylamide gel electrophoresis to separate small proteins less than 10 kDa. Five major bands could be observed, one of which corresponded to 12 kDa in molecular weight, and the other four were less than 10 kDa (Fig. 3a). The trichocyst ribbons were insoluble without SDS, and therefore it was difficult to resolve the trichocyst-related proteins with 2D electrophoresis. Therefore, we used the alkaline urea solubilization method to dissolve the trichocyst ribbons, and 2D analysis with this method showed that they included six polypeptides (Tri1, Tri2, Tri3, Tri4, Tri5, and Tri6) from 12 to 2 kDa in molecular weight (Fig. 3b). To identify these polypeptides, N-terminal amino acid sequencing was performed. We obtained adequate amino acid sequences for Tri1, Tri2, and Tri3 (which contained two polypeptides, Tri3-1 and Tri3-2). For Tri1, we could determine 20 amino acid residues; YASPNGGRNDPYSMYDTKMG. For Tri2, we could determine 20 amino acid residues; YGKKAANLYTQAFVQNTQAR. For Tri3-1, we could determine 18 amino acid residues; YGQTAAALYAKAAGQNAQa, in which “a” (Ala) is an estimated residue (Fig. 3b). For Tri3-2, 19 amino acid residues were determined; FDLEDYAGTYRETRNAYLK (Fig. 3b). We could not find any proteins homologous to these N-terminal amino acid sequences in public databases.

Fig. 3
figure 3

Electrophoretic analysis of trichocyst ribbons of Pyrenomonas helgolandii. a Tricine–sodium dodecyl sulfate-polyacrylamide gel electrophoresis of trichocyst ribbons isolated from P. helgolandii. The left lane is molecular marker, and molecular sizes (Da) are indicated on the left side. Major five bands were detected. b 2D electrophoresis of trichocyst ribbons. Trichocyst ribbons were separated in the first dimension (left to right) by IEF-PAGE, and in the second dimension (top to bottom) by tricine–sodium dodecyl sulfate-polyacrylamide gel electrophoresis. Molecular size (Da) and isoelectoric points (pI) are indicated on the left and top, respectively. Major six spots (Tri1, Tri2, Tri3, Tri4, Tri5, and Tri6) were found. N-terminal amino acid sequences were determined in Tri1 (YASPNGGRNDPYSMYDTKMG), Tri2 (YGKKAANLYTQAFVQNTQAR), and Tri3. Tri3 contained two polypeptides, Tri3-1 (YGQTAAALYAKAAGQNAQAa) and Tri3-2 (FDLEDYAGTYRETRNAYLKR)

Full size image

We determined sequences of the full-length cDNA corresponding to Tri1, Tri2, Tri3-1, and Tri3-2 by using degenerate primers designed from the N-terminal amino acid sequences of Tri1, Tri2, Tri3-1, and Tri3-2. The size of the full-length Tri1 cDNA was 431 bp, with a 5′ UTR of 14 bp and a 3′ UTR of 135 bp. Tri1 genomic sequences contained a single intron of 327 bp between 88 and 89 bp from the 5′ end of the cDNA. The size of the full-length Tri2 cDNA was 323 bp, with a 5′ UTR of 12 and a 3′ UTR of 125. Tri3-2 genomic sequences did not contain any introns. The size of the full-length Tri3-1 cDNA was 287 bp, with a 5′ UTR of 11 bp and a 3′ UTR of 90 bp. In addition, Tri3-1 genomic sequences did not contain any introns. The size of the full-length Tri3-2 cDNA was 272 bp, with a 5′ UTR of 7 bp and a 3′ UTR of 94 bp. Tri3-2 genomic sequences did not contain any introns. The predicted amino acid sequences of Tri1, Tri2, Tri3-1, and Tri3-2 were estimated to have average molecular masses of 10.4, 7.6, 7.0, and 6.2 kDa, and theoretical pI of 4.6, 8.0, 8.0, and 8.9, respectively. A Blast search did not find any proteins homologous to predicted amino acid sequences of Tri1, Tri2, Tri3-1, and Tri3-2 in public databases. However, when these sequences were compared with the reb family proteins (rebA, B, C, and/or D) that are involved in R-body synthesis and assembly in Caedibacter taeniospiralis, several interesting similarities to rebB were found. Although rebB proteins display little amino acid sequence similarity among proteobacteria, they share a region that displays relatively highly conserved amino acids (pfam11747: rebB superfamily domain) (Fig. 4a and Supplementary file 1). When Tri1 was multiply aligned with rebBs from proteobacteria, it contained a domain similar to the rebB superfamily domain (Fig. 4a and Suppementary file 1). In addition, Tri1 and rebBs were similar in size and did not contain any cysteine residues (Fig. 4a and Supplementary file 1). Tri1 showed the highest similarity (identity/positive identity = 19.7 %/40.6 %, alignment score = 20) to rebB in the gammaproteobacterium, C. taeniospiralis (Table 1). Furthermore, to evaluate whether this similarity is reliable, the hydropathicity and the secondary structures were analyzed between the rebB superfamily domain of C. taeniospiralis and the rebB superfamily-like domain of Tri1. The rebB superfamily domain of C. taeniospiralis and the rebB superfamily-like domain of Tri1 showed hydrophilic characteristics with averages of −0.36 and −1.0, respectively. Hydropathy plots of rebB superfamily domain in C. taeniospiralis showed similarity to that of the rebB superfamily-like domain in Tri1 (Fig. 5a). However, the secondary structure of the rebB superfamily-like domain in P. helgolandii was different from that of C. taeniospiralis, although they were similar at a point that alternates α-helix and β-strand regions (Fig. 6a). The predicted amino acid sequences of Tri2, Tri3-1, and Tri3-2 showed statistically significant similarity to one another, suggesting that these proteins are encoded by a single gene family. Tri2, Tri3-1, and Tri3-2 could be aligned only with C. taeniospiralis among proteobacteria, and showed regular similarity (identity/positive identity = 21.4 %/33.9 %, 28.5 %/41.0 %, and 8.9 %/17.8 %, respectively) to the latter half region of rebB of C. taeniospiralis (Fig. 4a, b and Supplementary file 1). Hydropathy plots of Tri3-2 were similar to that of C. taeniospiralis (Fig. 5b). Although hydropathy plots of Tri3-1 and Tri2 were similar to each other, with strong hydrophilic characteristics, they were very different from that of C. taeniospiralis (Fig. 5b). The secondary structure of Tri3-1 was similar to that of C. taeniospiralis (Fig. 6b). However, the secondary structures of Tri2 and Tri3-2 were very different from that of C. taeniospiralis (Fig. 6b). A domain search by SMART revealed that N-terminal regions of Tri1, Tri2, Tri3-1, and Tri3-2 do not have the characteristics of a signal sequence and cleavage site, suggesting that these proteins are unlikely to be secretory proteins (Fig. 4a, b). However, the 2D analysis showed that N-terminal regions of Tri1, Tri2, Tri3-1, and Tri3-2 were cleaved (Fig. 4a, b). The N-terminal-cleaved sequences of Tri1, Tri2, Tri3-1, and Tri3-2 were estimated to have average molecular masses of 9.3, 6.8, 6.3, and 5.7 kDa, and theoretical pI of 4.6, 8.2, 8.2, and 8.1, respectively, which agreed well with the values observed by 2D electorophoresis.

Fig. 4
figure 4

Alignment of the predicted amino acid sequences of Tri1, Tri2, Tri3-1, and Tri3-2 and their structures. a Amino acid sequence alignment among Tri1, Tri2, Tri3-1, and Tri3-2 in Pyrenomonas helgolandii, and rebBs in proteobacteria: areas shaded in black, identical and similar amino acids; box of broken line, conserved region among rebB of Caedibacter taeniospiralis, Tri3-1, Tri2, and Tri3-2; underbars in N-terminal region of Tri1, Tri2, Tri3-1, and Tri3-2, cleaved amino acid regions. b Schematic representation of comparison among rebB protein of C. taeniospiralis and Tri1, Tri2, Tri3-1, and Tri3-2 proteins of P. helgolandii. Predicted Tri2, Tri3-1, and Tri3-2 showed homology to the latter half region of rebB of C. taeniospiralis. Area shown in black in the N-terminal region of Tri1, Tri2, Tri3-1, and Tri3-2 shows unknown signal sequences. The abbreviations, reference numbers, and taxonomic groups for the amino acid sequences obtained from the National Center for Biotechnology Information Web site are as follows: P. helgolandii Tri1 (PheTri1, JF826281, Cryptophyceae), A. caulinodans rebB (AcaRebB, YP_001526698, alphaproteobacteria), B. oklahomensis rebB (BokRebB, ZP_02361292, betaproteobacteria), B. ambifaria rebB (BamRebB, YP_773108, betaproteobacteria), D. alkenivorans rebB (DalRebB, YP_002433825), P. pacifica rebB (PpaRebB, ZP_01912580, deltaproteobacteria), S. maltophilia rebB (SmaRebB1 and 2, YP_002028487 and YP_002028486, gammaproteobacteria), Pseudovibrio sp. rebB (PviRebB, ZP_05083799, alphaproteobacteria), X. axonopodis rebB (XaxRebB, NP_643396, gammaproteobacteria), C. taeniospiralis rebB (CtaRebB, YP_025468, gammaproteobacteria), and P. helgolandii Tri2, Tri3-1, and Tri3-2 (PheTri2, PheTri3-1, and PheTri3-2, JF826282, JF10318, and JF10319, Cryptophyceae)

Full size image
Table 1 Similarity among PheTri1 and rebBs from representative proteobacteria
Full size table
Fig. 5
figure 5

Comparisons of hydropathy plots among Caedibacter taeniospiralis, Tri1, Tri2, Tri3-1, and Tri3-2. a Hydropathy plots of rebB superfamily domain in C. taeniospiralis and rebB superfamily-like domain in Tri1. b Hydropathy plots of conserved region among C. taeniospiralis, Tri3-1, Tri3-2, and Tri2

Full size image
Fig. 6
figure 6

Comparisons of secondary structures among rebB of Caedibacter taeniospiralis, Tri1, Tri2, Tri3-1, and Tri3-2. a Secondary structures of rebB superfamily domain in C. taeniospiralis and rebB superfamily-like domain in Tri1. b Secondary structures of conserved region among C. taeniospiralis, Tri3-1, Tri3-2, and Tri2. Letters “e” and “h” show β-strand and α-helix, respectively. Box shows similar region of secondary structure between rebB of C. taeniospiralis and Tri3-1

Full size image

Discussion

In this study, we isolated the discharged trichocyst ribbons and sequenced four putative trichocyst-related genes, Tri1, Tri2, Tri3-1, and Tri3-2 from the red cryptomonad, Pyrenomonas helgolandii. These genes were coded on nuclear genome because orthologs of these Tri genes were found in nuclear genome of the red cryptomonad, Guillardia theta (JGI: http://genome.jgi.doe.gov/Guith1.home.html). Furthermore, Tri2, Tri3-1, and Tri3-2 are grouped in the same Tri family of low molecular weight proteins based on statistically significant similarity to one another. With regard to Tri1, although the Blast search did not find any proteins homologous to Tri1, we consider that the amino acids encoded by Tri1 is homologous to the R-body-constituted protein rebB of proteobacteria for the following reasons: (1) the similarity between rebB of Caedibacter taeniospilaris and Tri1 (alignment score = 20) is high enough to be defined as homologous, because an alignment score greater than 20, which is equivalent to an adjusted 20 % identity, almost always indicates a genuine relationship (Doolittle 1981). For example, the cytochrome f from the cyanobacterium Arthrospira maxima (accession: 1KIB_F) and the cytochrome c from guinea pig Tentative sequence (accession: C04604), which share a common ancestry, show an alignment score of 20 (Doolittle 1981); (2) Tri1 and the rebB proteins of proteobacteria were close in size; and (3) Tri1 possessed rebB superfamily-like domain, which showed a hydrophilic characteristic similar to that of rebB. With regard to Tri2, Tri3-1, and Tri3-2, although we could not find any proteins homologous to these proteins in public databases, they showed regular similarity to the latter half of rebB of Caedibacter taeniospiralis. Similarities of Tri2, Tri3-1, and Tri3-2 to rebB of C. taeniospiralis (alignment score = 43, 30, and 13, respectively) were high enough to warrant serious scrutiny, except for Tri3-2. However, the sizes of Tri2, Tri3-1, and Tri3-2 were only half that of rebB of C. taeniospiralis and were not long enough to be defined as homologous. On the other hand, similarities in the hydropathy patterns and the secondary structure to rebB were found in Tri3-2 and Tri3-1. Although these results support the notion that these proteins are homologs of rebB of C. taeniospiralis, these similarities were relatively weak and were not consistently found throughout three proteins. Therefore, the conclusion should be led with analyses of functions and three-dimensional structures of proteins.

R-bodies were not dissociated to polypeptides by the following treatment: 8 M urea; boiling in dilute HCl (pH 1.8) for 5 min; incubation for 2 h in 8.6 M guanidine buffer; and incubation in 6-M guanidine thiocyanate (Kanabrocki et al. 1986). These treatments also had no noticeable effect on dissociation of the trichocyst ribbons in cryptomonads, suggesting their biochemical similarity to the R-body. We could identify four polypeptides constituting the trichocyst ribbons in this study, and it is still unclear whether other trichocyst-related proteins show homology to reb proteins. It is necessary to identify other proteins to understand their structures and relationship with R-bodies in more detail.

Our electrophoretic analysis results were significantly different from those of Rhiel and Westermann (2011). They reported that the trichocyst ribbons from Chroomonas and Cryptomonas species were dominated by polypeptides of 40–44, 23–25, and 16–18 kDa, whereas in P. helgolandii, they were constituted of at least six small polypeptides from 2 to 12 kDa in molecular weight. The difference may be explained by the difference in the analytical methods between the two experiments, although there remains some possibility that the two taxa have different types of polypeptides. Rhiel and Westermann (2011) also found low molecular weight peptides migrating in the front during electrophoresis, which were more intensively stained by Coomassie blue than polypeptides of 40–44, 23–25, and 16–18 kDa. However, these low molecular weight peptides were not noticed in the trichocyst-constituted proteins because they were not resolved as clear bands, presumably because Tris–SDS-PAGE, which cannot resolve small polypeptides less than 10 kDa in molecular weight, was used for the eletrophoretic analysis. Probably the low molecular weight peptides found in Cryptomonas and Chroomonas may correspond to the polypeptides that we found in P. helgolandii.

In Cyathomonas truncata (= Goniomonas truncata), it has been reported that the trichocysts disappeared in axenic culture, but re-appeared on addition of some bacteria to the culture (Schuster 1968). In addition, 3H-thimidine uptake in Cyathomonas occurred when the trichocysts were reappearing in the cells, suggesting that the trichocysts contain nucleic acids (Schuster 1968). These findings suggested the hypothesis that the trichocysts in cryptomonads are a type of bacterial endosymbionts as in Paramecium. In contrast, Anderson (1962) and Mignot (1965) reported the development of trichocysts from vesicles in the Golgi region in the cryptomonads, Chilomonas and Cyathomonas, respectively. Wehrmeyer (1970) also reported that in Cryptomonas and Hemiselmis (Cryptophyceae), the trichocysts lie exclusively in close proximity to a single dictyosome, and the trichocysts originated from it. The trichocysts in cryptomonads are, therefore, generally regarded as cytoplasmic organelles, and not bacterial endosymbionts. Our results support the notion that trichocysts are not independent bacteria, because the cDNA of the trichocyst-related genes Tri1, Tri2, Tri3-1, and Tri3-2 had a long poly(A) tail that is polyadenylated at the RNA′s 3′ end in eukaryotes. Furthermore, orthologs of these four genes with high homology were found in the genome of the red cryptomonad, Guillardia theta. On the other hand, prospective amino acid sequences of Tri1, Tri2, Tri3-1, and Tri3-2 did not have a cleavable signal sequence in the N-terminal region, suggesting that Tri1, Tri2, Tri3-1, and Tri3-2 are unlikely to be secretory proteins produced via the endoplasmic reticulum and Golgi apparatus. However, N-terminal amino acid sequences from the discharged trichocyst ribbon proteins showed that Tri1, Tri2, Tri3-1, and Tri3-2 were clearly cleaved in the N-terminal region, suggesting that those may have unknown signal peptides. Further studies of the function or target of N-terminal sequences in Tri1, Tri2, Tri3-1, and Tri3-2 are needed to clarify the pathway of the trichocyst formation.

As to the biochemical makeup of the trichocyst constituents of cryptomonads and ciliates, the proteins composing the crystalline trichocyst matrix of Paramecium (trichocyst matrix proteins, TMPs) are the products of a large family of 30–100 polypeptides from 15 to 20 kDa in molecular size (Gautier et al. 1994, 1996; Tindall et al. 1989). On the other hand, 2D analysis showed that the trichocyst ribbons in P. helgolandii are constituted of only six polypeptides, four of which, Tri1, Tri2, Tri3-1, and Tri3-2, did not show homology to TMPs. Therefore, we consider that the biochemical makeups of the trichocysts are entirely different between cryptomonads and peniculine ciliates, and they have independently evolved in the two lineages.

Most of the species possessing reb-homologous genes belong to the phylum proteobacteria (Akiba et al. 2010). The evolution of R-bodies in bacteria has long been discussed, and one hypothesis is that the genes encoding R-bodies were passed on by horizontal gene transfer by phages or plasmids, because the phylogenetic relationship of proteins encoded by the reb-homologous genes of some selected species is quite different from that of 16S rRNA genes (Akiba et al. 2010; Beier et al. 2002; Pond et al. 1989; Quackenbush and Burbach 1983). We could not find any eukaryote proteins homologous to rebB proteins in public databases, and the proteins constituting the cryptomonad trichocyst were unique among the eukaryotes. On the other hand, the genes homologous to rebA, rebB, and/or rebD are distributed in a variety of gram-negative bacteria, although no genes homologous to rebC have been identified. Therefore, we consider that the trichocyst-related gene, Tri1, might have been acquired by horizontal gene transfer from some proteobacteria that have the ability to produce the R-bodies. Current taxon sampling of eukaryotic sequences, however, is poor. For further evaluation of this scenario, more sequences throughout phylogenetically informative taxa are necessary.