Both host and parasitic strains were isolated from the Penzé Estuary (North-West of France, English Channel; 48°37′N, 3°56′W). The culture of Amoebophrya sp. strain A120 was established by manually isolating a swimming infected dinoflagellate host (recognized thanks to the natural bright-green auto-fluorescence of the parasite) collected in the field using an epifluorescence microscope Olympus BX51 equipped with the U-MWB2 filter (excitation 450–480 nm, emission 500 nm; Coats and Bockstahler, 1994). The infected host was then transferred to and maintained in exponentially growing cultures of the non-toxic dinoflagellate Scrippsiella acuminata (previously known as S. trochoidea, Kretschmann et al., 2015) that was established from the germination of a single cyst collected from sediments in 2005 (strain ST147). Both strains are accessible at the Roscoff Culture Collection website¹ with the following accession numbers: ST147 = RCC1627 for S. acuminata and A120 = RCC4398 for the Amoebophrya-like parasite.

We used continuous light in order to minimize circadian variations of the host metabolism, such as pigment cell content and composition. We have found that continuous light does not seem to stress phytoplanktonic organisms, as illustrated by significant increases in growth rates under this condition (data not shown). Consequently, S. acuminata cultures were acclimated to continuous white light in exponential growth phase under 85 μmol photons m^–2 s^–1 for over 3 weeks prior to the experiments. We used fluorescent tubes (Sylvania Luxline plus) and adjusted the irradiance using a quantamer (Biospherical Instrument Inc.) and neutral density filters (LEE filters). The cultures were grown at 22°C in vented flasks (Culture One) filled with natural seawater from the Penzé Estuary complemented with F/2 medium (Sigma Aldrich) and 5% soil extract (Chambouvet et al., 2008).

We performed two independent experiments of infection of S. acuminata ST147 by Amoebophrya sp. A120, mainly differing by the initial parasite:host ratio and sampling times. Despites the differences in ratios, we observed massive infections for both experiments corresponding to 100% prevalence (see below). In both experiments, the free-living infective stage of the parasite (the dinospores) was freshly prepared from several infected S. acuminata cultures and harvested by gentle separation from the remaining host cells using gravity filtration through 5 μm polycarbonate filters (Whatman).

The first experiment (hereafter called “experiment 1”) was used to investigate the influence of light on the development of Amoebophrya sp. A120, as well as the photosynthetic activity and the pigment composition of the infected S. acuminata ST147 host. We also estimated the effect of light on parasite production, which was assessed based on the conversion of host to parasite biomass (i.e., carbon content). For that, hosts and parasites were mixed in 50 mL culture flasks with an initial dinospore:host ratio of 50:1 and placed either under light or dark conditions, each in four replicates (host and dinospore countings are detailed below). Flasks containing uninfected hosts from the initial cultures were similarly sampled during the experiment and used as control. In order to estimate host and dinospore cell density, we performed 11 samplings of both infected and control cultures over a 94 h period, and for 6 of those we estimated the pigment composition. In addition, parasite prevalences were estimated after 40 h of incubation to check for the effect of light conditions on parasite infectivity.

The second experiment (hereafter “experiment 2”) was used for transcriptomics analyses and has been previously described by Farhat et al. (2018). Briefly, 21 culture flasks (600 mL) were inoculated by mixing 150 mL of dinospores of Amoebophrya sp. A120 and 300 mL of exponentially growing S. acuminata. ST147 (dinospore:host ratio of 30:1) cultures. Every 6 h, three flasks were completely harvested (7 times from 0 to 36 h) for mRNA extraction and sequencing, as detailed in Farhat et al. (2018). Three flasks containing 450 mL of uninfected hosts were additionally prepared to be used as control and completely harvested at 0 h (T0). Before each harvest, infected cultures were sampled in order to estimate the prevalence of the parasite.

All data were transformed [log (x + 1)] before analysis using the basic package “stats” in the R software (R Core Team)¹². For experiment 1, we used Mann-Whitney tests to compare dinospore production and host photosynthetic activities at each sampling time (N = 4) between light and dark conditions in both infected and uninfected treatments. The same test was used to compare the conversion of host to dinospore biomass under light and dark conditions (N = 4). To compare temporal patterns of gene expression in experiment 2, we used polynomial regressions to fit no-linear relationships between time (hours) and TPM (N = 21). In all polynomial regression analyses, we used 3-degrees polynomial features given they resulted in the best significant fit.

Host cell densities were nearly identical between healthy (control) and infected cultures under both light and dark conditions (Mann-Whitney test, p > 0.01) for the first 40 h of experiment 1, followed by a growth spurt in the uninfected treatments at 74 h under light conditions (net growth rate of 0.3 d^–1) and decrease in infected treatments (Figures 1A,B). Infection prevalence (79 and 75% after 40 h under light and dark conditions, respectively) and duration of the parasite internal development (46 h followed by bursts of dinospore) were similar in infected cells independent of the light condition (Mann-Whitney test, p = 0.23). Interestingly, we observed that the dinospore density was 5-fold lower at the end of the experiment in dark compared to light conditions (Mann-Whitney test, p = 0.03). Moreover, net conversions of the available host biomass (maximum of density estimated at 24 h) into dinospore biomass (maximum of density estimated at 88 h) were 75.62 and 14.77% (Figures 1C,D) in light and dark conditions, respectively (Mann-Whitney test, p = 0.002).

At the beginning of experiment 1, the photosystem II of S. acuminata chloroplasts displayed near-optimal quantum yield values (F_V/F_M = 0.6_; Mann-Whitney test, p < 0.01) in all cultures (Figures 1E,F). In light conditions, the F_V/F_M parameter of the infected culture were similar (Mann-Whitney test, p < 0.01) to those of the uninfected control during the first 46 h (Figure 1E), corresponding to the intracellular stages of the parasite life cycle. The photosystem II quantum yield remained constant in the controls and dropped in the infected cultures only after 46 h. By comparison, this parameter remained high until 24 h and then progressively decreased in both control and infected cultures under constant darkness (Figure 1F).

We measured the natural chl a cell fluorescence of S. acuminata cells by flow cytometry in the light condition experiments (Supplementary Figure S2A). The infected cells show similar values of chl a fluorescence when compared to the control during the first 46 h of infection (i.e., until the first release of dinospores in the medium) followed by a sharp drop in chl a fluorescence consecutive to the collapsing of the host cell density.

In order to identify possible modifications of light utilization capacities in the host chloroplasts during the infection, we monitored the variations of the thylakoidal accessory pigments relative to chl a under constant light (Supplementary Figures S2B–F). S. acuminata ST147 displayed a pigmentation profile typical of peridinin containing dinoflagellates (Roy et al., 2011). The relative content in chl c₂ (Supplementary Figure S2B) and the light-harvesting xanthophyll peridinin (Supplementary Figure S2C) remained stable throughout the infection cycle. We recorded similar relative contents of the photoprotective pigments β-carotene, diadinoxantin, and diatoxanthin between the infected and the healthy cultures throughout the infection (Mann-Whitney test, p < 0.01; Supplementary Figures S2D–F).

In this study, we monitored the physiological and genetic responses of the dinoflagellate Scrippsiella acuminata ST147 to infection by the Syndiniales Amoebophrya sp. A120, which primarily targets the host nucleus after entry into the cell. Our result show that the absence of light during the infection had no effect on the duration of the internal development nor the prevalence of the parasite. However, the lack of light negatively impacted the number of dinospores produced per infected host cell and the estimated biomass transfer yield from the host to the parasite (Figures 1A–D; Mann-Whitney test, p = 0.002).

The photosystem II (PSII) complex carries out the initial charge separations of the light phase of photosynthesis and is very sensitive to perturbation or stress, which trigger the photoinactivation process (see e.g., Takahashi and Murata, 2008). The quantum yield of PSII thus constitutes a useful parameter for monitoring the integrity of the chloroplastidial ETC under stress. Our results show that PSII complexes remain efficient in S. acuminata ST147 infected cells (Figure 1E), suggesting that the thylakoidal light phase of photosynthesis is quite functional during the parasite intracellular stages. These findings are corroborated by the resilience of cpDNA-encoded mRNAs when compared to their nuclear counterparts (Figures 3D,E). In addition, it is well-known that a number of β-carotene molecules are bound to the reaction center of PSII and act as sink for singlet oxygen (Telfer, 2005; Ishikita et al., 2007). The absence of significant β-carotene degradation and phaeopigment induction as a result of oxidative stress, along with the stability of both light-harvesting and light-dissipating pigments of the photosynthetic antennae together reflect consistent light utilization by the chloroplast during the infection. These observations strongly suggest that infection of S. acuminata ST147 by Amoebophrya sp. A120 does not induce any significant photophysiological stress in the host chloroplast. Instead, the chloroplast of infected S. acuminata ST147 likely maintains steady ATP and NADPH production in the plastidial stroma during the whole duration of the internal development of the parasite.

While the dinoflagellate host appears to be “alive,” i.e., continued swimming, intracellular infection will rapidly alter most of its physiological functions, following a modality that seems to differ depending whether the parasite first targets the host cytoplasmic or nucleus content. Previous studies reported that the dinoflagellate Levanderina fissa (formerly Gyrodinium instriatum) infected by a cytoplasmic strain of Amoebophrya sp. maintains plastidial carbon fixing activity throughout infection, suggesting that the Calvin-Benson cycle remained functional (Park et al., 2002b). By contrast, in the dinoflagellate Akashiwo sanguinea infected by an intranuclear strain, carbon fixation was impaired from the early stages of the infection, with pronounced degradation of chl a. The authors linked the chloroplast shutdown in A. sanguinea to the digestion of the nucleus by the parasite, whereas the intact nucleus of L. fissa would sustain chloroplast functions through the uninterrupted synthesis of nuclear-encoded photosynthetic proteins (Park et al., 2002b). In our study, infection by the intranuclear Amoebophrya sp. A120 did hinder photosynthetic functions in S. acuminata ST147 (Figure 1E). Considering the short timespan of the parasite intracellular development period (36 h under the experimental conditions of this study), it is likely that the chloroplast of S. acuminata was quite capable of relying upon the existing pool of nuclear-encoded photosynthetic proteins in order to maintain plastidial activity despite the early digestion of its nucleus. This suggests the existence of a variety of physiological infective strategies among Amoebophrya species.

Considering that during the first 46 h of the first experiment (1) the host density remained constant under dark condition; (2) no drastic difference in host density was observed between the light and the dark conditions; (3) the host chloroplasts remained functional under dark condition, we therefore conclude that non-infected hosts survived the first 46 h of darkness, likely in lower physiological conditions compared to the cells under light. While we cannot clearly disentangle the various effects of the dark conditions on the host physiology, our results strongly suggest that light, while not essential, is an important factor for the fitness of Amoebophrya sp. A120. This is in line with previous observations suggesting that environmental factors such as water quality and nutrient levels impact spore productivity and infectivity (Yih and Coats, 2000), whereas the duration of the internal development of the parasite is dependent upon the host-parasite system (Park et al., 2013). A major limitation of our study concerns its reliance on indirect evidence of kleptoplastidy by the parasite. Such shortcoming can be addressed by the use of stable isotopes of inorganic carbon (i.e., HCO3-), only accessible to the photosynthetic host but not to a heterotrophic parasite, on infected cultures and monitor whether ¹³C has been incorporated by the dinospores.

Morey and Van Dolah (2013) reported that mRNA half-lives were substantially longer for genes involved in housekeeping processes, including energy metabolism and transport, whereas genes involved in organellar post-transcriptional regulation belonged to the pool with the shortest half-life. Amoebophrya sp. A120 primarily digests the host nucleus, thereby knocking out the main gene expression machinery of the host. Consequently, the apparent upregulation of genes related to information processing and cell cycle processes (Supplementary Figure S7) in our DE analyses could be explained by the longer half-lives rather than by the de novo synthesis of these mRNAs. However, our results feature stark differences in the rarefication rate of mRNAs for the same pathways (mitochondrial and plastidial ETCs) based on the location of those genes. For instance, while the expression of nuclear-encoded ETC genes steadily decreased throughout the infection, all mitochondrial (mtDNA) and most (9 out of 11) chloroplastidial (cpDNA) encoded protein genes displayed an apparent peak of expression at 18–24 h and 24–30 h, respectively (Figures 3D,E; Supplementary Figures S5E–I and Supplementary Figures S6H–J). One explanation for the apparent increase in mt and cp-mRNAs can involve the presence of a residual pool of RNA polymerases. Moreover, the degradation of organellar mRNAs is delayed (longer effective half-lives) compared to those from the nucleus as it relies solely on ribonucleases (RNase) present in the organelles while being protected from RNases from the parasite by intact mitochondrion and chloroplast membranes. Altogether, our results strongly suggest that the chloroplast and the mitochondrion remain active and might fuel the active swimming of the denucleated host during the internal stages of the infection (Supplementary Video S1), a behavior reminiscent of an undead “zombie” dinoflagellate. Active swimming has been previously observed in other dinoflagellates infected by Amoebophrya spp. (Park et al., 2002a) and might eventually increase the dispersion capacity of the dinospores and their probability to reach new hosts.

Amoebophyra sp. A120 is an aerobic phagotrophic endoparasite (Cachon, 1964) that uses oxygen throughout its life cycle (Farhat et al., 2018) and requires light in order to produce high numbers of infective dinospores (Figures 1A,B). Moreover, our results suggest that the parasite has access to oxygen, ATP, NADPH and other metabolites such as carbohydrates, from the host mitochondrion and chloroplast which remain functional during most of its intracellular stages of development. This suggest that the physiological stress of infection induced by Amoebophrya sp. A120 does not impact the functioning of the host organelles, a feature also reported in symbiotic dinoflagellates of collodarian where the chloroplast lasts under thermal stress until the very last stages of the cell lysis (Villar et al., 2018).

Frequently referred to as kleptochloroplastidy, plastid retention involves a modified grazing behavior where the predator sequestrates functional chloroplasts from its prey for a variable period of time (Johnson, 2011). Kleptochloroplastidy has been previously described in at least thirteen dinoflagellates genera (Stoecker et al., 2009). The stability of transient plastids may vary considerably: kleptochloroplastids of Nusuttodinium spp. and Pfisteria piscicida remain photosynthetically active for only a few days, whereas those of the genus Dinophysis are active over multiple cell divisions (Stoecker et al., 2009; Park et al., 2014). The organellar retention strategy observed in Amoebophrya lasts throughout the whole infection, a process similar to that used by the marine cyanophages infecting picocyanobacteria of the genera Prochlorococcus and Synechococcus. These viruses take control of the photosynthetic apparatus of their host during the infection (Dammeyer et al., 2008) by hijacking the transcriptional machinery in order to overexpress some photosynthetic proteins encoded in the viral genome, such as encoding the PSII core subunit PsbA (D1) (Mann et al., 2003; Lindell et al., 2005; Sharon et al., 2009). This strategy is so efficient that 60% of the psbA genes identifiable by barcoding at the ocean surfaces are thought to be of phage origin (Sharon et al., 2007). Amoebophrya strains lack the vestigial plastid-like structure found in other non-photosynthetic apicomplexans and the genomes of the sequenced strains are devoid of plastidial related genes (John et al., 2019). These suggest that Amoebophrya parasites are unable to take control of the host photosynthesis apparatus like cyanophages, but rather benefits passively from it. Given that organellar retention does not necessarily imply that plastids are actively maintained by the host, our results suggest that Amoebophrya sp. A120 performs a new form of “extra-cellular” organellar retention (klepto-organelly) involving both the chloroplast (kleptochloroplast) and potentially the mitochondrion (kleptomitochondrion) of its host. This is likely done by specifically avoiding the digestion of the host organelles during the parasite intracellular development, a process analogous to what happens during kleptoplastidy described in other protists/animals (Johnson, 2011). Comparison with other Amoebophrya strains digesting their host chloroplasts earlier during infection (Park et al., 2002b) may help understanding some of the mechanisms underlying organellar retention by the A120 strain.