Both Akashiwo sanguinea and its corresponding strain of Amoebophrya sp. were isolated from Jiaozhou Bay (36°24′ N, 120°11′ E), China in June 2020 using single cell isolation methods, and established as continuous cultures as described in Kim et al. (2004). The identifications of A. sanguinea and Amoebophrya sp. were first made with microscopic observation (by visualization of the natural bright-green auto-fluorescence of the parasite, Coats and Bockstahler, 1994), and further confirmed with molecular characterization (18S rDNA sequencing, see details in Supplementary Text). In brief, A. sanguinea was maintained in f/2-Si medium (Guillard and Ryther, 1962) made with sterile natural seawater (salinity of 30 ± 0.1). The Amoebophrya sp. ex A. sanguinea cross was maintained in stock culture by sequentially transferring aliquots of infected host culture into uninfected host culture of exponentially growing A. sanguinea at approximately 2-day intervals. Stock cultures were incubated at 20 ± 1°C under a 14 h: 10 h light: dark cycle, with cool white fluorescent light providing 175 μE m^–2 s^–1.

The time-course infection of Amoebophrya sp. in A. sanguinea was established following the methods of Coats and Park (2002), which covered a complete life cycle of the parasite. Briefly, newly formed dinospores (≤6 h old) were separated from the parasite cultures by gravity filtration using 47-mm polycarbonate membranes (Whatman Nuclepore, 10-μm pore size). The subsamples (1 ml) of newly formed dinospores (≤6 h old) were immediately fixed with buffered formaldehyde (final conc. 1%), and enumerated with hemacytometers using an Olympus IX71 epifluorescence microscope (excitation 450–480 nm, emission 500 nm; Coats and Bockstahler, 1994).

Batch cultures of exponentially growing A. sanguinea (800 ml) were prepared in 1-L culture flasks, with an initial density of approximately 3 × 10³ cells ml^–1. To investigate the responses of A. sanguinea to Amoebophrya sp., A. sanguinea cultures were inoculated with the recently formed dinospores (≤6 h old) at a dinospore: host ratio of 10:1 to insure infection of most host cells (Chen et al., 2018). To examine the potential effects of the waterborne cues from the parasite, A. sanguinea cultures were treated with the addition of an equivalent volume of dinospore filtrate (Millipore PC filters, 1-μm pore size). In addition, an equivalent volume of sterile f/2-Si medium was added in the separate culture as experimental control. All treatments were conducted in triplicate and incubated under the same condition as mentioned above.

Host abundance and parasite prevalence (i.e., percent cells infected) were assessed and further quantified at time points of 0, 12, 30, 54, and 70 h post inoculation. To determine host abundance, subsamples (1 ml) were collected and fixed with buffered formaldehyde (final conc. 1%), then enumerated using triplicate plankton counting chambers under an epifluorescence microscope (IX71, Olympus, Japan). The parasitized A. sanguinea were identified based on the characteristic green autofluorescence of the parasite and red fluorescence of the host chloroplasts under the blue light excitation (excitation 450–480 nm, emission 500 nm; Coats and Bockstahler, 1994). The prevalence was calculated as the percentage of infected cells when at least 100 host cells per sample were counted. Morphological variation, photosynthesis activity, pigment composition, and transcription of photosynthesis-related genes of host cells were also studied at the five time points (0, 12, 30, 54, and 70 h) and as described below.

Epifluorescence microscopy, scanning electron microscopy (SEM), and transmission electron microscopy (TEM) were used to characterize the morphology and development of Amoebophrya sp. ex A. sanguinea. Live observation of the algal culture (1 ml) was made at 400 × magnification and recorded with an Olympus DP73 digital camera attached to an Olympus IX71 epifluorescence microscope.

Aliquots of 30 ml of newly formed dinospores (≤ 6 h old) of Amoebophrya sp. ex A. sanguinea, with a density of ∼10⁵ cells ml^–1 were fixed with OsO4 (final conc. 1%, Sigma-Aldrich, Seoul, Korea) for 40–50 min at room temperature, and rinsed with deionized water. The solution was gently filtered using gravity through a 5-μm pore sized Millipore nylon membrane (Cork, Ireland) to capture the fixed cells. Then the sample was dehydrated in a graded acetone series (10, 30, 50, 70, 90%, 3 × 100%) for 15 min at each step, critical point dried with liquid CO₂ (EM CPD300, Leica, Vienna, Austria), and sputter-coated with gold (EM ACE200, Leica, Vienna, Austria). Digital images were collected with an S-3400 N SEM (Hitachi, Hitachinaka, Japan) and processed using Adobe Photoshop (Adobe, San Jose, CA, United States). Cell dimensions of 50 cells were calculated from SEM micrographs.

Aliquots of experimental cultures (30 ml) were fixed with 2.5% glutaraldehyde (final conc.) for 3 h at 4°C, and post-fixed with OsO₄ prepared with filtered seawater (final conc. 1%) overnight at 4°C. Cells were then centrifuged at 720 g for 5 min to form pellets and remove the supernatant. Pellets were rinsed three times (10 min each) in fresh f/2 medium. Dehydration was carried out in a graded acetone series (10, 30, 50, 7%, 90%) for 10 min each and then rinsed three times in absolute acetone for 15 min each. Finally, the pellets were infiltrated with 2:1, 1:1, and 1:2 mixtures of mixtures of acetone: Spurr’s resin (at least 8 h for each dilution) and embedded in 100% Spurr’s resin (Spurr Formula Kit I, SPI Supplies West Chester, Pennsylvania, United States; Spurr, 1969) overnight. The material was sectioned on an ultramicrotome (EM UC7, Leica, Vienna, Austria). Sections were mounted on Formvar-coated grids, stained with uranyl acetate (3%) and lead citrate (2%), and examined with an HT7700 TEM (Hitachi, Hitachinaka, Japan).

Aliquots of 2 ml algal culture were used to determine the chlorophyll fluorescence of A. sanguinea using a FEM-2 pulse-modulated fluorometer (Hansatech, Norfolk, United Kingdom). The maximal photosynthetic efficiency (F_v/F_m) was calculated as:

Where F_v represents the variable fluorescence, F_m represents the maximum fluorescence, and F_o represents the minimum fluorescence. Samples were dark adapted for 30 min (Schreiber et al., 1995) before transfer to a cuvette for the measurement of F_o, and after a saturating pulse (0.3 s; 2,600 μmol photonsm^–2 s^–1), the measurement of F_m. The actual photochemical efficiency of PSII in the light (Φ_PSII) was also measured at the same time.

Another 10 ml of algal culture were filtered through the Whatman GF/F filters (25-mm in diameter, Whatman Inc.) with a vacuum pump (<0.03 MPa). Filters were then folded and wrapped against light in aluminum foil, and immediately frozen at –80°C for further processing. Pigments were determined by an alliance High-Pressure Liquid Chromatograph (e2695, Waters, Milford, Massachusetts, United States) using a 100 μl sample injection according to Kong et al. (2012).

Aliquots of 30 ml algal cultures were pipetted and concentrated at 4°C via gentle centrifugation (500 g, 5 min) to form pellets and remove the supernatant, then flash-frozen in liquid nitrogen, and stored at –80°C until RNA extraction. Total RNA was extracted using TRIzol reagent (TakaRa, Japan), and further treated with RNase-free DNase I (TakaRa, Japan). The quality of RNA samples was ascertained by the spectrophotometric method with an A260 nm/A280 nm ratio from 1.8 to 2.0, and further assessed with 1% agarose gel electrophoresis based on the integrity of 18S and 28S rRNA bands. The cDNA was synthesized from 1 μg of total RNA using the PrimeScript^TM RT reagent Kit (TaKaRa, Japan) with addition of gDNA Eraser to eliminate the genomic DNA contamination according to manufacture protocols. The cDNA templates were diluted 10 times with nuclease-free water for subsequent analyses.

The transcriptions of photosynthesis-related genes, including psbA, psbD, psaA, psaB, petA, and petB, were detected with the quantitative Real Time PCR (qRT-PCR) assay using the specific primers reported in Liu et al. (2020). To normalize the mRNA expression of the target genes, 18S rRNA, TUA and GAPDH were selected as endogenous control and their expression stability in Amoebophrya infected cultures was confirmed prior to the experiments based on the assays described in Pfaffl et al. (2004) and Deng et al. (2016). The qRT-PCR assay was performed with a Rotor-Gene Q 2plex HRM thermocycler (QIAGEN, Germany). Reactions (triplicate for each sample) were carried out in a total reaction volume of 20 μl, containing 10 μl SYBR Premix Ex Taq II (2 ×) (TaKaRa, Japan), 0.5 μl for each primer (10 μM), 1 μl template cDNA, and DNAse/RNAse free water (8 μl, Sigma-Aldrich, MO, United States). Amplification conditions were 95°C for 30s, followed by 40 cycles of denaturation at 95°C for 5 s, and annealing at 60°C for 30 s. Melt curve analysis was added at the end of each PCR thermal profile to assess the specificity of amplification. Amplification efficiency (E) was calculated from the given slopes in Rotor-gene Real-Time PCR System Manager software by a 10-fold diluted cDNA sample series, with six dilution points measured in triplicate. The relative mRNA levels were calculated by the 2^{–Δ Δ Ct} method as described in Livak and Schmittgen (2001). The gene expression levels in control cultures were used as control.

All data were presented as mean ± standard error of the mean (SE) in the text, and subjected to a t-test to determine the difference in the three treatments using the SPSS 18.0 for windows (SPSS, Chicago, IL, United States), with the significance level of P < 0.05.

In the filtrate-added and control groups, Akashiwo sanguinea showed a steady growth during the experiment, reaching 10.05 ± 0.73 × 10³ and 10.35 ± 0.18 × 10³ cells ml^–1 by 70 h, respectively (Figure 1A). In the infected groups, the growth of A. sanguinea was significantly affected by introduction of Amoebophrya dinospores. The cell density of A. sanguinea decreased dramatically after 30 h post inoculation (p.i.), and the final density was 1.71 ± 0.74 × 10³ cells ml^–1, even though there was no significant difference in the cell densities among the three groups within the first 30 h (P > 0.05).

Using fluorescence microscopy, we were able to detect Amoebophrya infection in A. sanguinea based on the presence of green autofluorescence signals inside host nuclei (Figure 2). In the early infection stage (12 h p.i.), the green autofluorescing parasite (approximately 18.20 μm length) was surrounded by red-fluorescing chloroplasts of host dinoflagellates (Figure 2C). By this time, the abundance of infected cells was 1.16 ± 0.24 × 10³ cells ml^–1 and the prevalence in the infected cultures was 31.25 ± 1.73% (Figure 1B). In the middle stage of infection (30 h p.i.), the presence of the beehive-like structure (approximately 40.01 μm length) could be easily discriminated in the Ameobophrya-infected host cells (Figure 2D), even though the green autofluorescence was mostly concealed by the red autofluorescence of host chloroplasts. The abundance of infected host cells increased to 2.50 ± 0.24 × 10³ cells ml^–1 and the prevalence of Amoebophrya infection was 54.80 ± 2.84% at the time point (Figure 1B). In the late stage of infection (54 h p.i.), the characteristic “beehive” structure of Amoebophrya multinucleate trophonts (approximately 65.80 μm length) was easily observed in the infected A. sanguinea cells (Figure 2E) where they appeared bright green with grooved and doted structure when examined with an epifluorescence microscope. The abundance of infected host cells was 1.95 ± 0.37 × 10³ cells ml^–1 and the prevalence increased to 85.83 ± 1.40% at 54 h p.i. By 70 h p.i., only 0.17 ± 0.03 × 10³ cells ml^–1 were infected and the prevalence of Amoebophrya infection dropped to 10.05 ± 0.73% (Figure 1B). Most of the infected host cells broke up at the time point, and the parasite emerged as a conspicuously motile vermiform (approximately 133.33 μm length) (Figure 2F). The vermiform persisted for approximately 3–10 min before suddenly breaking apart to release numerous dinospores, and the abundance of motile dinospores was 9.25 ± 0.41 × 10⁵ cells ml^–1 in the infected groups. In the case of A. sanguinea as host, each infected cell produced 370 ± 51 Amoebophrya dinospores. No infected cells or dinospores were detected in the filtrate-added and control groups.

The free-living stage (dinospore) of Amoebophrya sp. was a flagellated zooid with a large bulbous episome (4.01 ± 0.37 μm in diameter, n = 50, using SEM) and a narrow hyposome, which was characterized by the presence of two flagella (Figure 3A). The transverse flagellum exhibited flagellar hairs over its full length and was approximately 11.21 ± 1.05 μm long (n = 50, using SEM), whereas the longitudinal flagellum was bare (4.05 ± 0.53 μm long, n = 50, using SEM) and progressively thinner toward the end. As indicated by the TEM ultrastructure, there was an elongated, typically compact nucleus located in the cell center of uninfected A. sanguinea, with a distinct nucleus membrane condensed chromosomes (Figure 3B). Chloroplasts were ellipsoidal or elongated, rod-like or irregularly shaped, and located almost throughout the healthy host cells.

At 12 h p.i., the condensed chromosomes of host cells were overtly disrupted and trophonts of Amoebophya sp. appeared as spherical cells showing electron-dense indentation at the surface (Figure 3C). By 30 h p.i., the parasite grew in size as they consumed the nucleonic contents of the host. The host nucleus was significantly enlarged and mostly occupied by the developing trophont. The parasite started to take on a “beehive” appearance (Figure 3D), resulting from helical extension of the parasitic cingulum and sinking of the parasitic apex during its growth. Subsequently, the development of the parasite largely consumed the nucleonic contents of the host and took up a large proportion of the host dinoflagellate. Further invagination of the developing parasite created a cavity structure (lobed cavity and mastigocoel) within the trophont cytoplasm, appearing as a narrow space associated with basal bodies. By 54 h p.i., cytoplasm was relegated to the margin of the host cells by the well-developed trophont of the parasite (Figure 3E), the chloroplasts of the host cell remained intact at the late stage of the parasitic infection (Figure 3G). Inside the giant trophont, multiple nuclei (will develop into the nuclei of dinospores in next few hours) lined up in a continuous fashion along the folds of the beehive, each section lobe contained one nucleus, with numerous flagella extended into the cavity (amphiesmalined mastigocoel) (Figure 3F). Ultimately, the multiflagellated and multinucleated trophont ruptured the host cell and emerged as a conspicuously motile vermiform, which split into fragments of variable sizes before suddenly break, releasing numerous individual dinospores (Supplementary Videos 1–3).

Photosynthesis is an important determinant of the ability of marine microalgae to survive biotic stresses (Baker, 2008), and the F_v/F_m is widely used to indicate photosynthetic activity of microalgae (Maxwell and Johnson, 2000). The photosynthetic activities of A. sanguinea were overtly impacted by Amoebophrya sp. infection (Figure 4). The F_v/F_m of A. sanguinea remained steady in the filtrate-added and control cultures, with near-optimal quantum yield values (F_v/F_m = 0.6) over the experimental period. In the cultures inoculated with Amoebophrya dinospores, the F_v/F_m decreased continuously to 0.45 ± 0.03 at 54 h p.i., then recovered to 0.51 ± 0.02 by 70 h p.i., which was significantly different from that of control and filtrate-added groups (P < 0.05) (Figure 4A). The values of photochemical efficiency (Φ_PSII) in the infected cultures ranged from 0.47 ± 0.03 to 0.54 ± 0.01, significantly lower than that of the control and filtrate-added cultures (0.47 ± 0.03 to 0.57 ± 0.04) (P < 0.05), but exhibited a similar pattern of variation (Figure 4B).

Photosynthesis of marine dinoflagellates largely depends on the cellular photosynthetic pigments (Blankenship, 2014), thus variations in the contents of major pigments were further quantified to assess the effects of Amoebophrya infection on A. sanguinea. For the filtrate-added and control cultures, the cellular photosynthetic pigments showed a similar pattern of variation, and no significant difference was found between the two groups (P > 0.05). In the infected cultures, chlorophyll a (Chl a) exhibited a significant difference from the filtrate-added and control groups, which decreased sharply by 30 h p.i. (36.41 ± 2.66 pg cell^–1) (P < 0.05), then recovered gradually to 48.86 ± 2.83 pg cell^–1 at 70 h p.i. (Figure 5A). The light-harvesting pigment peridinin of the infected cultures ranged from 40.01 ± 0.03 to 48.22 ± 2.35 pg cell^–1, Figure 5B). The photoprotective pigments diatoxanthin, β-carotene, and diadinoxanthin increased gradually in the infected cultures, with fluctuated variation during the time course (Figures 5C–E). It was noteworthy that the β-carotene of the infected cultures decreased overtly by 30 h p.i., then gradually increased by 70 h p.i. (Figure 5D).

The two photosystem I-related genes—psaA and psaB—were down-regulated (< 1-fold) in the A. sanguinea cultures with Amoebophrya infection, but showed different patterns. The psaA transcripts were significantly down-regulated at 30 h p.i. and 70 h p.i. (P < 0.05), with the lowest level of transcription (0.06 ± 0.01-fold) at 30 h p.i. (Figure 6A). The psaB transcripts decreased significantly during the infection cycle (P < 0.05), with the lowest transcription (0.18 ± 0.05-fold) at 30 h p.i. (Figure 6B). The photosystem II-related genes—psbA and psbD—were significantly suppressed during the time course of infection (P < 0.05), with the lowest level of transcription at 70 and 30 h p.i., respectively (Figures 6C,D). In contrast, the psbA and psbD were almost up-regulated (> 1-fold) in the control cultures, with the highest level of transcription at 70 and 54 h p.i., respectively. The two cytochrome b6f complex-related genes—petA and petB showed various patterns in the infected cultures. The petA gene was down-regulated (< one fold) during the time course, with significant suppression at 12 and 30 h p.i. (P < 0.05) (Figure 6E). The petB was down-regulated significantly (P < 0.05), with the lowest transcription (0.13 ± 0.07-fold) at 30 h p.i. (Figure 6F). The petB transcripts in the control cultures increased during the time course, with the highest transcription (2.00 ± 0.04-fold) at 54 h p.i (P < 0.05). Compared to control cultures, all genes except psbA were down-regulated in the filtrate-added cultures (< one fold), and the psbD was significantly suppressed throughout the infection cycle (P < 0.05).