Dinoflagellates can serve as predators or parasites of tintinnid ciliates. Known predators feed on the ciliate from outside the tintinnid lorica, while parasites either grow in the host cytoplasm or feed inside in the lorica while attached to the outside of the host cell. Here we report mixotrophic species of Scrippsiella that enter the lorica to consume the ciliate zooid of Helicostomella subulata from Denmark and multiple tintinnid species from Korea. We contrast morphology and life-history stages of these mixotrophic predators with dinokaryote parasites of tintinnids and address phylogenetic relationships based on rDNA sequences. Mixotrophic Scrippsiella species sometimes attack tintinnids that are simultaneously infected by syndinean dinoflagellates, complicating study of life histories and potentially leading to confusion about trophic status and taxonomy.
Dinoflagellate predators of tintinnids include photosynthetic and heterotrophic species that either ingest the entire prey, or consume the ciliate zooid from outside the prey lorica. For example, Gymnodinium instriatum presses its sulcus against to oral opening of the prey lorica and directly engulfs the ciliate zooid (Uchida et al., 1997), while Oxyphysis oxytoxoides and Dinophysis rotundata ingest prey cytoplasm by inserting a feeding tube into the lorica (Inoue et al., 1993). By contrast, dinoflagellates characterized as parasites of tintinnids attack the ciliate from inside the lorica. Some syndinian dinoflagellates belonging to the genera Amoebophrya and Euduboscquella act as intracellular parasites of tintinnids, while dinokaryotes of the genera Duboscquella, Duboscquodinium, and Tintinnophagous are ectoparasites of tintinnids (Coats and Bachvaroff, 2013). Tintinnophagus acutus ingests host cytoplasm via a feeding tube, while the mechanism of feeding in species of Duboscquella and Duboscquodinium remains uncertain. At the end of the growth cycle, both syndinian and dinokaryote parasites of tintinnids undergo multiple divisions to produce dispersal, infective cells.
Here we report feeding and life-cycle process of a dinoflagellate from the Danish Kattegat, tentatively identified as Scrippsiella sp., that enters the lorica of Helicostomella subulata and consumes the ciliate zooid. We also consider the morphology and development of similar dinoflagellates that attack tintinnids in coastal waters of Korea and examine their phylogeny based on rDNA sequences.
Samples from the harbor adjacent to the Marine Biological Laboratory, Helsingør, Denmark were collected by towing of a 20-µm plankton net through the upper 3 to 4 m of water. Net tows obtained between July 30 and August 6, 2011 were maintained and processed at near ambient water temperature (∼20 °C). Helicostomella subulata loricae containing post-feeding cells of Scrippsiella sp. were isolated using a micropipette, washed through several changes of 0.45-µm filtered seawater, and transferred to ring slides, Sedgwick-Rafter chambers, or multi-well plates for subsequent observation. Two strategies were used to assess in vivo morphology, development, and behavior of Scrippsiella sp. First, post-feeding specimens were periodically examined, photographed, and video recorded using an Olympus CK2 inverted microscope with uEye digital video camera. Second, H. subulata loricae housing only the ciliate zooid (= naive H. subulata) were isolated as above, added to wells containing free-swimming daughter cells of Scrippsiella sp., and monitored over time. In combination, these approaches permitted documentation of prey capture and feeding, post-feeding morphology/behavior, cell division, daughter cell morphology/behavior, and cyst morphology. An Olympus BX50 microscope equipped with epifluorescence capability and digital camera was used to examine and photograph individual specimens, plastid autofluorescence, and thecal plates stained with calcofluor white M2R (Fritz and Triemer, 1985).
Samples from Jangmok Bay, Korea (34°59’N, 128°40’E) were collected in summer of 2015 (20 July to 24 August) and 2016 (27-31 July) by towing a 20-µm plankton net through the upper 5 m. Samples from both years were immediately fixed with neutral Lugol’s iodine solution and stored at 4 °C. In addition, unfixed samples from 2016 were maintained at near ambient temperature (22-24 °C) for in vivo examination of specimens. Tintinnid loricae containing dinoflagellates were isolated from Lugol-fixed samples using micropipettes and a Zeiss Axiovert 10 equipped with a AxioCam ER c5 digital camera, washed through several changes of 0.22-µm filtered seawater, photographed using a Zeiss Imager A2 equipped with epifluorescence capability and a Zeiss AxioCam ICc1 rev. 4 digital camera, and processed for molecular and phylogenetic analysis as described below. Unfixed samples were scanned using the inverted microscope to locate and photograph tintinnid loricae containing a dinoflagellate. Specimens were isolated, washed with filtered sea water, transferred to multi-well plates, and monitored over time to assess developmental events. Daughter cells resulting from division of post-feeding cells were initially cultured in F/20-Si phytoplankton growth medium and subsequently transitioned to F/8 medium for stock culture at 20 °C, with irradiance of 120 µmol m−2 h−1 on a 12:12 light:dark cycle.
Cellular measurements were obtained using Zeiss Axiovision software, with means ± standard error of the mean provided in the text along with sample size (n).
Molecular analysis of specimens isolated from net-tow material followed the procedures of Jung et al. (2016), coupled with use of two PCR primer pairs, a EukA/EukB set (Medlin et al., 1988) and a EukA/LSU Rev4 set (Sonnenberg et al., 2007), to amplify rDNA. For molecular analysis of dinoflagellate cultures, a single cell was isolated by microcapillary using a Zeiss Stemi 508 microscope, washed at least three times in sterile-filtered seawater, and transferred to a microfuge tube for rDNA amplification as above.
Sequence fragments were assembled and aligned using Geneious R10.2.5 (Kearse et al., 2012) with internal plugin ClustalW 2.1 (Thompson et al., 1994). P-distances were calculated using MEGA v7.0.26 (Kumar et al., 2016). To determine an appropriate DNA substitution model for Bayesian inference (BI) and maximum likelihood (ML) analyses, we used the Akaike information criterion to identify the best-fit model using the jModelTest 2.1.10 (Darriba et al., 2012). The model selected was GTR + I (0.3160) + G (0.6370). Phylogenetic relationships were evaluated using BI (Bayesian inference) and ML (maximum likelihood) analyses. MrBayes 3.2.6 (Ronquist et al., 2012) was used for BI assessment and IQ-TREE (Nguyen et al., 2015) was used for ML analysis. For further information about sequence alignment, determination of p-distances, and assessment of phylogenetic relationships, see supplemental file: Molecular_details.pdf.
Helicostomella subulata loricae collected from the Danish Kattegat sometimes contained dinoflagellates tentatively identified as Scrippsiella sp. (Figure 1a). Loricae containing Scrippsiella sp. lacked a tintinnid zooid, but often contained debris presumably derived from the missing ciliate. Dinoflagellates appeared firmly wedged in position and were always oriented with the epicone directed toward the oral end of the lorica. The cytoplasm had a faint yellow-gold pigmentation and contained a large food vacuole filled with amorphous material and positioned anterior to the dinoflagellate nucleus (Figure 1a). Such specimens were considered post-feeding cells.
Post-feeding specimens measured 40 ± 3.0 µm by 18 ± 0.4 µm (n = 6; range 33-53 µm long by 17-20 µm wide) and typically remained inside the tintinnid lorica for a few hours, during which time they occasionally rotated slowly, with short anterior or posterior movement. Cells exited the lorica (Figure 1b) in 4-8 h, swimming slowly for short distances between periods of prolonged quiescence. That behavior persisted for several hours, as the food vacuole slowly decreased in size and the cytoplasm acquired more pronounced pigmentation. Specimens divided about 12 h after feeding to form two somewhat rounded, slow moving daughter cells. A second cell division usually occurred over the following 24 h. Daughter cells became more active over time, swimming rapidly with frequent and erratic changes in direction. Rapidly swimming cells (Figure 1c) had marked yellow-gold pigmentation, a sharply pointed epicone, and averaged 15 ± 0.6 µm by 13 ± 0.5 µm (n = 7; range: 13-18 µm long by 12-15 µm wide). Plastids showed strong Chl a autofluorescence (Figure 1d). Calcofluor staining revealed bipesoid epithecal plate tabulation consisting of Po, x, 4’, 3a, 7’’ (Figure 1e,f), with a symmetrical first apical plate and a six-sided, second anterior intercalary. There were six cingular plates with two sutures visible in mid-dorsal perspective. Sulcal and hypothecal plate tabulations were not resolved. Daughter cells held in filtered seawater for several days formed spherical cysts measuring 18-20 µm in diameter (n = 3). Cysts were covered with numerous short spines and contained 2 or 3 red globules (Figure 1g).
Highly active daughter cells occasionally attacked naive H. subulata (see supplemental video file: Scrippsiella-H_subulata feeding.MP4). Dinoflagellates were not observed when entering tintinnid loricae, but feeding cells were found attached to ciliate zooids contracted within their loricae (Figure 1h-j). Specimens remained pressed against the prey, as cytoplasm was “sucked” into the developing food vacuole. Early feeding cells contained a small spherical to ovoid food vacuole and were slightly larger (18-21 µm by 14-17 µm; n = 2) than rapidly swimming cells. Feeding lasted 3-5 h, during which time the prey was almost entirely ingested and predator size increased dramatically. On two occasions, specimens were found attached to the outside of a prey lorica adjacent to the oral opening. One of those specimens soon abandoned the prey lorica. The other, however, remained attached as the tintinnid swam unimpeded until suddenly contracting into its lorica. Immediately upon contraction, a feeding tube was evident stretching from the dinoflagellate to its prey (Figure 1k). The feeding cell ingested most of the ciliate (Figure 1l) before detaching from the lorica and swimming away.
Ribosomal DNA sequences were not obtained for Danish specimens.
Dinoflagellates resembling Scrippsiella sp. from H. subulata were encountered in 52 loricae (44 from Lugol-preserved samples; 8 from unfixed material) representing seven tintinnid species from the southern coast of Korea: Helicostomella longa (Figure 2a-b), Rhizodomus tagatzi (Figure 2k-o), Tintinnopsis cf. beroidea (Figure 2c-f), Tintinnopsis cylindrica (Figure 2i), Tintinnopsis radix (Figure 2j), Tintinnopsis tocantinensis (Figure 2g), and Tintinnopsis sp. (Figure 2h). Thirty-three specimens were post-feeding cells with large food vacuoles (Figure 2a,e,g), 12 were specimens fixed while feeding (Figure 2c,k,o), and seven were dividing or post-division cells (Figure 2j,m,n).
Lugol-preserved post-feeding cells averaged 36 ± 1.1 µm by 22 ± 0.8 µm (n = 27; range: 26-46 long by 17-28 wide), while early feeding cells were 18-26 µm long by 13-19 µm wide (n = 4). The number of presumptive daughter cells contained in loricae ranged from two in T. cf. beroidea and T. radix to 8 in R. tagatzi. Scrippsielloid dinoflagellates sometimes attacked prey infected by syndinean dinoflagellates. For example, one early feeding cell was attached to R. tagatzi harboring a trophont of Euduboscquella sp. (Figure 2o), while another was located in the lorica of T. cf. beroidea containing sporocytes closely resembling those of Euduboscquella sp. (Figure 2f).
All specimens tracked over time in vivo (three each in loricae of H. longa and R. tagatzi; two in loricae of T. cf. beroidea) had pale yellow-gold pigmentation. Specimens from H. longa were very active, with two leaving the lorica immediately upon isolation. Only one of the cells from R. tagatzi and T cf. beroidea exited the lorica immediately, while the others were relatively quiescent, moving only slightly from time to time. All specimens, however, swam out of the lorica prior to division. Five cells, three from H. longa and two from R. tagatzi, divided over the following day, each producing two daughter cells. Daughter cells from H. longa loricae were easily grown as phototrophs. Cultured cells showed strong Chl a autofluorescence and were similar in size (mean 20 ± 0.4 by 14 ± 0.3 µm; range: 17-22 µm long, 12-16 µm wide; n = 12) to early feeding cells.
A total of nine novel rDNA sequences were obtained and deposited into GenBank under the following accession numbers: MN319487-MN319488, two culture strains each derived from specimens associated with different Helicostomella longa loricae (Figure 2a; specimen for MN319488 not shown); MN314591, a post-feeding cell from H. longa (Figure 2b); MN319489-MN319490, post-feeding cells from each of two Tintinnopsis cf. beroidea loricae (Figure 2d,e, respectively); MN319491, a pair of daughter cells in a Tintinnopsis radix lorica (Figure 2j); MN319492-MN319493 and MN314592, Rhizodomus tagatzi loricae containing an early feeding cell, post-feeding cell, and eight daughter cells (Figure 2k,l,n, respectively).
Sequences for the two culture strains were 2992 and 2969 bp long (MN319487 and MN319488, respectively) and encompassed the SSU (small subunit), ITS1 (internal transcribed spacer 1), 5.8S, ITS2 and LSU D1-D2 (large subunit D1-D2 region) of the rRNA gene (supplemental Table 1). The genetic dissimilarity between the two sequences was 0.11% in the SSU (differing by only two nucleotides) and 0% in the ITS1-LSU regions (supplemental Table 2). The top hit BLAST sequences in GenBank were attributed to Scrippsiella trochoidea (HM483396; 96.38% and 96.43% similarity, with 108 and 106 bp differences including 17 gaps) and Duboscquodinium collini (HM483398; 96.33% and 96.37% similarity, with 110 and 108 bp differences including 11 gaps). Only the SSU was amplified for the Lugol-fixed, post-feeding cell from H. longa. The sequence was 1484 bp long and encompassed a deletion of 271 bp at the subterminal end of the SSU that was identified through gene alignment with other Scrippsiella sequences. Excluding the 271 bp gap, the SSU of the post-feeding cell had a genetic dissimilarity of 0.13% and 0% relative to the SSU of the two cultures (MN319487 and MN319488, respectively).
SSU-LSU sequences were obtained for the two post-feeding cells associated with T. cf. beroidea, the pair of daughter cells from T. radix, and two of the three specimens from R. tagatzi (MN319492 and MN319493), with sequence length ranging from 2889-2984 bp (supplemental Table 1). Only the SSU was sequenced for the eight daughter cells located in the lorica of R. tagatzi (MN314592, 1773 bp). SSU sequences for these six specimens were identical. The five ITS1–LSU sequences were either identical or very highly similar (genetic dissimilarity 0–0.18%), the latter differing at two loci in the ITS1 and one locus in LSU D1–D2 region (supplemental Table 2). The most similar sequences in GenBank based on BLAST searches were attributed to Scrippsiella trochoidea (HM483396; 98.07–98.31% identity with 49–56 bp differences including 11 gaps) and Duboscquodinium collini (HM483398; 97.72–97.96% identity with 59–68 bp differences including 3 gaps).
SSU rRNA gene tree topology revealed by Bayesian inference (BI) and maximum likelihood (ML) analyses (Figure 3) nested the nine novel sequences within a well-supported scrippsielloid group (BI/ML, 0.97/100%). The novel sequences sorted into two clades representing separate species of Scrippsiella (Species 1 and Species 2). The clade for Species 1 consisted of the three sequences derived from dinoflagellates associated with H. longa loricae and had high support values (BI/ML, 1.00/99%). The clade for Species 2 was composed of the six sequences obtained for specimens from loricae of T. cf. beroidea, T. radix, and R. tagatzi and also had high support (BI/ML, 0.92/99%). Genetic dissimilarities between Species 1 and 2 were 0.28–0.34% in the SSU and 7.15–7.33% in ITS1–LSU regions (supplemental Table 2).
The symmetrical first apical plate, bipesoid epithecal tabulation, and six-sided, second anterior intercalary place dinoflagellates that attacked Helicostomella subulata within the order Peridiniales, subfamily Calciodinelloideae (Fensome et al., 1993), the latter now included in the family Thoracosphaeraceae (Elbrächter et al., 2008). The presence of six cingular plates, with two cingular sutures visible in mid-doral view, link specimens from H. subulata with the genus Scrippsiella (Zinssmeister et al., 2011). Thus, our tentative designation of specimens from H. subulata as Scrippsiella sp.
Ribosomal RNA gene sequences available for dinoflagellates attacking four species of Korean tintinnids (Helicostomella longa, Tintinnopsis cf. beroidea, Tintinnopsis radix, and Rhizodomus tagatzi) nested among Scrippsiella clades, indicating that they are also Scrippsiella spp. The novel sequences sorted into two highly supported clades, suggesting that Korean tintinnids serve as prey for at least two species of Scrippsiella. Sequences for one of the species (Species 1 in Figure 3) were only associated with specimens from H. longa, indicating possible prey specificity. Species 2 sequences, however, represented specimens from T. cf. beroidea, T. radix, and R. tagatzi, indicating that the second Scrippsiella species lacks prey specificity. Our phylogenetic tree also indicates that Scrippsiella species feeding on Korean tintinnids were distinct from Duboscquodinium collini and Tintinnophagus acutus (both parasites of tintinnids), as well as Scrippsiella trochoidea (the top hit in GenBank BLAST searches). Tintinnophagus acutus ex. Tintinnopsis cylindrica (HM483397) was located outside the scrippsielloid lineage and basal to Cryptoperidinopsis-Pfiesteria-Paulsenella clades. Duboscquodinium collini ex. Eutintinnus fraknoii (HM483398, HM483399) and Scrippsiella trochoidea (AJ415515, EF492513, HM483396) were located within the scrippsielloid lineage, but sorted into highly supported clades (BI/ML 1.00 and >90%) separate from our novel sequences.
Scrippsiella species feeding on H. subulata and H. longa are clearly mixotrophic organisms, as they possessed plastids, with the latter easily grown in culture as a phototroph. Scrippsiella sp. from Tintinnopsis cf. beroidea and Rhizodomus tagatzi are likely mixotrophs, given their yellow-gold pigmentation in vivo, however, the presence of plastids was not confirmed. Specimens from T. radix are also likely mixotrophs, given molecular analyses indicating that they are conspecific with Scrippsiella sp. from T. cf. beroidea, and R. tagatzi. We lack morphological or molecular data to infer that dinoflagellates associated with T. cylindrica, T. tocantinensis, and Tintinnopsis sp. are mixotrophs.
Four species of dinokaryotes, Duboscquella tintinnicola, Duboscquodinium collini and kofoidi, and Tintinnophagus acutus, are considered parasites of tintinnids (Coats and Bachvaroff, 2013). Duboscquella tintinnicola and Duboscquodinium kofoidi are enigmatic species that have not been reported since their original descriptions, making comparison with Scrippsiella spp. feeding on tintinnids tenuous at best. Our mixotrophic, scrippsielloid predators show little resemblance to Tintinnophagus acutus, as the latter lacks sulcus, girdle, and flagella when attached to its host, only forms a small food vacuole during prolonged feeding lasting days, and undergoes sporogenesis within an outer membrane or cyst wall. Scrippsiella sp. from H. subulata and specimens found in the loricae of Korean tintinnids superficially resemble Duboscquodinium collini. Like D. collini, (Coats et al., 2010, Coats and Bachvaroff, 2013), our specimens appeared lodged inside tintinnid loricae as post-feeding cells, had a more or less ovoid shape with an acute protuberance (the apex of the epicone) directed toward the oral end of the lorica, and appeared posteriorly attached to the ciliate zooid when feeding. They differed, however, from the original description of D. collini Grassé 1952, published in Chatton (1952), by the lack of: (1) a conspicuous, outer double-membrane envelope; (2) a differential first division producing a trophocyte and gonocyte; and (3) palintomic sporogenesis. Also, our specimens from H. subulata clearly had chloroplasts, whereas Coats et al. (2010) failed to detect plastids in post-feeding D. collini, an observation that needs corroboration given the phylogenetic relationship of D. collini with photosynthetic species of Scrippsiella (Zinssmeister et al., 2011, Gottschling et al., 2012).
Gaines and Elbrächter (1987) considered parasitic dinoflagellates “as having morphologically different feeding and reproductive stages and producing numerous progeny after only one feeding act.” Scrippsiella sp. that attacked H. subulata maintained typical dinoflagellate structures throughout the feeding process, although the girdle and sulcus were sometimes difficult to detect due to swelling of the cell during feeding. Subtle movements of the cell during and after feeding, and the ability to swim out of the lorica when disturbed, suggest that the flagella were never lost. In addition, post-feeding cells from H. subulata, H. longa, Tintinnopsis cf. beroidea, and Rhizodomus tagatzi exhibited no significant change in morphology prior to division, typically exited the lorica before division, and did not undergo sequential divisions closely set in time. Repeated division during or after feeding appear possible, however, as loricae of R. tagatzi sometimes containing four or eight daughter cells. Thus, Scrippsiella species that attacked tintinnids in our Danish and Korean samples are not parasitic dinoflagellates as defined by Gaines and Elbrächter (1987). Instead, we consider them to be mixotrophic predators.
We were unable to determine how feeding cells get into the lorica of their prey. For prey having loricae with a large diameter (e.g. Tintinnopsis radix and Rhizodomus tagatzi), pre-feeding cells might swim directly into the lorica and rotate 180° before or after attaching to the ciliate zooid, thereby achieving the orientation observed in feeding and post-feeding specimens. That possibility seems less likely, however, for prey having a lorica diameter only slightly larger than pre-feeding cells (e.g. Helicostomella subulata, H. longa, and Tintinnopsis cf. beroidea). Also unlikely is the possibility that dinoflagellates might swim backward (i.e. antapical end first) into the prey lorica. Since pre-feeding cells are able to attach with their antapical end against the prey lorica, they might also do the same at the oral region of the tintinnid zooid. If so, then contraction of the ciliate as the dinoflagellate begins to feed would draw the attached cell into the lorica, posterior end first.
Occasionally, feeding cells of Scrippsiella species attached to ciliate zooids that were infected by Euduboscquella sp., or were located in loricae containing sporocytes of Euduboscquella sp. Those observations raise concerns about using preserved plankton samples to assess life-history stages of predatory or parasitic dinoflagellates developing inside tintinnid loricae. For example, a post-feeding cell of Scrippsiella sp. occurring along with sporocytes of Euduboscquella sp. might be mistaken for a trophocyte and dividing gonocyte of D. colloni. We would have similar concerns about life-history stages of Scrippsiella sp. from our preserved Korean samples had molecular analyses not shown identical, or nearly identical, rDNA sequences for early feeding, post-feeding, and daughter cells from loricae of Tintinnopsis cf. beroidea, T. radix and Rhizodomus tagatzi. Thus, descriptive studies of intra-lorica parasites and predators of tintinnids might be best resolved using correlative data for morphology/cytology, development in vivo, and rDNA sequences.
Tintinnid populations of Danish and Korean waters are subject to predation by mixotrophic species of Scrippsiella. These mixotrophs enter the prey lorica, increase dramatically in size while feeding, and become difficult to distinguish as free-living dinoflagellates. Reproduction occurs outside or inside the prey lorica, with the latter resulting in multiple daughter cells inside a single lorica and their possible misidentification as parasites. Tintinnids attacked by Scrippsiella spp. can be simultaneously infected by intracellular parasites, potentially leading to further confusion when examining life-histories based on preserved samples.
Research in Denmark supported in part by a 2011 international Ph.D. course sponsored by the Faculty of Science, University of Copenhagen Helsingør, Denmark. Travel costs for Y. Lu provided by the PACES research program, Alfred-Wegener-Institute Helmholtz-Zentrum für Polar- und Meeresforschung. Korea work supported by the Basic Core Technology Development Program for the Oceans and the Polar Regions of the National Research Foundation (NRF), with funding from the Ministry of Science, ICT and Future Planning, Republic of Korea (grant number NRF2016M1A5A1027456).
Supplemental data for this article can be accessed on the publisher’s website.