“Ectrogella” Parasitoids of the Diatom Licmophora sp. are Polyphyletic

The diatom genera Licmophora and Fragilaria are frequent epiphytes on marine macroalgae and can be infected by intracellular parasitoids traditionally assigned to the oomycete genus Ectrogella. Much debate and uncertainty remains about the taxonomy of these oomycetes, not least due to their morphological and developmental plasticity. Here, we used single‐cell techniques to obtain partial sequences of the parasitoids 18S and cox2 genes. The former falls into two recently identified clades of Pseudo‐nitzschia parasites temporarily named OOM_1_2 and OOM_2, closely related to the genera of brown and red algal pathogens Anisolpidium and Olpidiopsis. A third group of sequences falls at the base of the red algal parasites assigned to Olpidiopsis. In one instance, two oomycete parasitoids seemed to co‐exist in a single diatom cell; this co‐occurrence of distinct parasitoid taxa not only within a population of diatom epiphytes, but also within the same host cell, possibly explains the ongoing confusion in the taxonomy of these parasitoids. We demonstrate the polyphyly of Licmophora parasitoids previously assigned to Ectrogella (sensu Sparrow, 1960) and show that parasites of red algae assigned to the genus Olpidiopsis are most likely not monophyletic. We conclude that combining single‐cell microscopy and molecular methods is necessary for their full characterisation.

PARASITES play a key role in phytoplankton ecological succession and in their host population dynamics (Chambouvet et al. 2008). Recent studies in marine and freshwater systems, especially when molecular barcoding has been used, have consistently highlighted the importance and continuing lack of knowledge of parasitic relationships in structuring pelagic microbial communities (Lep ere et al. 2008;Lima-Mendez et al. 2015;de Vargas et al. 2015). Planktonic diatoms are no exception to this pattern, and the most frequently described parasites of diatoms are Chytridiomycota (Fungi) and Oomycota (Stramenopila; Ibelings et al. 2004;Hanic et al. 2009;Guti errez et al. 2016;Taylor and Cunliffe 2016). Although less extensively studied, recent studies have shown that benthic and epiphytic diatoms are similarly subject to infection by parasites (Scholz et al. 2014(Scholz et al. , 2016). An abundant body of literature dating back from the 19 th century describes the presence of intracellular oomycetes infecting diatoms and mostly assigns them to the genus Ectrogella, which has traditionally been included in the Saprolegniales (Sparrow 1960). The original description of the genus corresponds to Ectrogella bacillariacearum infecting a freshwater Synedra; it stresses its spherical shape, lack of pigmentation, intracellular coenocytic habit and describes its zoosporogenesis. In this paper, a particular feature of Ectrogella is the presence of several discharge tubes in the bigger sporangia, and the uniflagellate, monoplanetic zoospores (Zopf 1884). This original description, however, conflicts with a second account given by Scherffel (1925), who stressed that zoospores emerging from sporangia were in fact biflagellate and diplanetic (see detailed review and illustration in Garvetto et al. 2018). Afterwards, several Ectrogella species were defined, but in most cases, important criteria, such as zoosporogenesis, were not observed; on the other hand, additional criteria absent from the original description were added while the presence of sexual reproduction or resting spores was reported for other species in the genus (Scherffel 1925). Thus, generally speaking, Ectrogella species are described as intracellular, unbranched, obligate biotrophic, holocarpic parasites. In the late stage of infection, the mature zoosporangium occupies the whole cell and disintegrated diatom plastids can be seen surrounding it. In their syntheses, Sparrow (1960) and Karling (1981) recognised eight species of Ectrogella, seven of them infecting diatoms, with only three in marine hosts: E. licmophorae (diplanetic with spores encysting at the mouth of the discharge tube, i.e. achlyoid), E. perforans (monoplanetic with spores swarming directly away from the sporangium, i.e. saprolegnoid) and E. eurychasmoides (diplanetic with primary spores encysting within the sporangium, i.e. eurychasmoid). In 2001, and without much more evidence at hand, Dick proposed to synonymise all the freshwater species of Ectrogella into E. bacillariacearum, to the exception of E. monostoma, which was synonymised with Aphanomycopsis bacillariacearum (Dick, 2001). Thus, he reduced the number of extant marine species to two, by including E. licmophorae into E. perforans, and keeping E. eurychasmoides as a "doubtfully distinct" species (Table 1), whereas Johnson (1966) argued to synonymise it with E. perforans and keep E. licmophorae a separate entity. Recently, molecular data have started to clarify the phylogenetic relationships between these parasitoids of diatoms and other intracellular parasites of brown and red algae. The genus of brown algal pathogens Anisolpidium (and the order Anisolpidiales) was transferred from the hyphochytrids into the oomycetes (Gachon et al. 2017). Several species of red algal pathogens assigned to the genus Olpidiopsis were sequenced (Badis et al. 2019;Klochkova et al. 2016;Kwak et al. 2017;Sekimoto et al. 2008Sekimoto et al. , 2009). However, the recent epitypification of the genus Olpidiopsis, due to the molecular description of Olpidiopsis saprolegniae, ascertained that these red algal pathogens do not belong to the same clade. The same authors also sequenced and defined an epitype for the red algal pathogen Pontisma lagenioides, and thus tentatively reassigned all other Olpidiopsis pathogens of red algae to the genus Pontisma despite the lack of strong phylogenetic support (Buaya et al. 2019a). To date, all red algal pathogens characterised molecularly cluster in three closely related clades; however, their monophyly and exact relatedness with Pontisma, Anisolpidium and the pathogen of freshwater diatoms Olpidiopsis gillii (Buaya et al., 2019b) remain to be established. Garvetto et al. (2018) recently identified novel pathogens of the marine diatoms Pseudo-nitzschia and Melosira; but in the light of the doubtful typification of Ectrogella and the unresolved relationships between different clades, they adopted a transient nomenclature OOM_1 (containing three subclades OOM_1_1, OOM_1_2 and OOM_1_3) and OOM_2. In parallel, Buaya et al. (2017) described the species Olpidiopsis drebesii, which falls in the OOM_2 clade, and defined a novel monogeneric family, the Miraculaceae. The latter contains the genus Miracula, with two species M. helgolandica (Buaya et al., 2017) and M. moenusica (Buaya and Thines, 2019), both belonging to the subclade OOM_1_1. Among the hosts of E. eurychasmoides, E. licmophorae, and E. perforans are the diatoms Licmophora sp. and Fragilaria sp. (Table 1). Licmophora is an almost ubiquitous marine, colonial, mostly epiphytic diatom with wedge-shaped cells in girdle view, while Fragilaria is a filament-forming diatom which can be planktonic, benthic, free living in colonies or epiphytic. Ectrogella perforans has been reported to cause epidemics outbreaks affecting 30% of Fragilaria and up to an astonishing 99% of Licmophora populations in the USA (Sparrow 1969). In the early 1980s, Raghu Kumar successfully cultivated a pathogen of Licmophora also identified as E. perforans (Raghu Kumar, 1978) and described its ultrastructure (Raghu Kumar 1980a, 1980b. Interestingly however, Johnson (1966) highlighted the variability of the morphological criteria used to delimit species among Ectrogella parasitoids of Licmophora (namely, L. abbreviata and L. gracilis). He mentioned the co-existence in field-collected materials of several modes of zoospore discharge and concluded that the morphological features used to define each species were unreliably polymorphic. Apparently, he did not envisage that several parasites species may co-exist in a same sample. With the above in mind, we developed single-cell approaches based on individual whole genome amplification and subsequent selection of suitable molecular markers, in order to investigate species boundaries within the genus Ectrogella and similar parasitoid genera.

Sampling and single-cell isolation
On the 11 July 2017 and 27 July 2017, the macroalgae Laminaria digitata and Ulva sp. were sampled at Clachan Bridge (56°19 0 03.2″N 5°34 0 59.7″W; Isle of Seil, United Kingdom) and kept in seawater at 10°C under constant air supply, with a 12-h light-dark cycle. Algal fragments or their epiphytic community were screened for the presence of infected diatoms by bright field optical microscopy (Axioskop 2 plus, Zeiss, Oberkochen, Germany). Upon microscopic inspection of the epiphytic microbial community growing on Ulva sp. and Laminaria digitata, cells and chains of Licmophora sp. and Fragilaria sp. containing endobiotic thalli were recorded. The samples were further stained with 1% calcofluor white (CW) for 5 min in the dark, to facilitate the detection of oomycete infective structures. Pictures were taken with an AxioCam HRc coupled to the AxioVision software (Zeiss, version 4.7.1). The most developed thalli were stained with CW, and some exhibited discharge tubes, strongly suggestive of an infection by Ectrogella-like parasites. Samples containing infected diatoms were then transferred into sterile Petri dishes with sterile filtered seawater, and infected cells were isolated through mouth pipetting under an inverse microscope. Single cells (for Licmophora sp.) or single colonial chains (for Fragilaria sp.) were washed four times by sequential transfer in sterile seawater drops, transferred into sterile Eppendorf tubes, and stored at À20°C until further handling. This procedure resulted in fourteen single-cell isolates. Isolate names provide information on the infected diatom host (Inf = infected; Lic = Licmophora and Fr = Fragilaria), sampling date (the code S10 refers to samples collected on the 27 of July, while the absence of it refers to samples from the 11 of July) and ordinal number of the single-cell isolate (SC = single cell/single chain).

Single-cell whole genome amplification, PCR and sequencing
Multiple displacement amplification (MDA) was carried out using the REPLI-g â Single Cell Kit (Qiagen, Hilden, Germany). Briefly, 4 ll PBS buffer and 3 ll lysis buffer were added to single cells. After 10 min of incubation at 65°C, the lysis was stopped with 3 ll of Stop Solution. 40 ll of master mix (containing φ-29 polymerase) was added to each single cell, and samples were incubated for 8 h at  Table S1, including specific primers designed to discriminate two oomycetes co-infecting a single Licmophora sp. cell. In order to rule out cross-contamination in the coinfected sample, both 18S rDNA and cox2 sequences were repeatedly retrieved from different dilution of the MDA product and after all reagents and laboratory consumables had been replaced. PCR products were checked by gel electrophoresis. When present, bands of different length were excised from the gel and purified with the Xtracta Gel Extraction Tool (Sigma-Aldrich, St. Louis, MO). If a single amplification product was detected, the purification was performed using the Thermo Scientific GeneJET Gel Extraction and DNA Cleanup Micro Kit, and PCR products were Sanger-sequenced (GATC Biotech AG, K€ oln, Germany). Overall, marker genes were successfully amplified for seven of the single-cell isolates.

Phylogenetic reconstruction
Sequences were quality-controlled, trimmed and assembled in contigs using Geneious 6.1.8 (Kearse et al. 2012). The dataset of 18S rDNA sequences of oomycetes used in this study is amended from Garvetto et al. (2018) in order to include novel sequences of parasitoids of diatoms. Alignments were carried out using the MAFFT (Katoh et al. 2002) algorithm implemented in Geneious 6.1.8, manually checked and curated with Gblocks 0.91b (Castresana 2000). Substitution models were assessed through ModelFinder (Kalyaanamoorthy et al. 2017) in IQ-TREE 1.5.5 (Nguyen et al. 2015), resulting in TN + R4. IQ-TREE 1.5.5 was also used for the phylogenetic reconstruction, which was computed with a Maximum-likelihood method using the ultrafast bootstrap approximation (UFBoot, 1,000 replicates) test of phylogeny (Minh et al. 2013). Aiming at backing up the results from the 18S rDNA, a second phylogenetic reconstruction was computed (as described above) based on the amino acid sequence of the mitochondrial marker cox2, under the substitution model mtZOA + F + G4 (Rota-Stabelli et al.

Observed morphological features
Different morphological features were observed in the oomycetes present in the environmental samples (Fig. 1A, B) containing the isolated single cells (Fig. 1C-I, Table S2), which are in line with descriptions of Ectrogella perforans infecting Licmophora (Sparrow 1960) and Fragilaria (Sparrow 1969). In Fragilaria sp. parasitic thalli undertook an ellipsoidal shape in the centre of the cell; generally between two collapsed plastids (Fig. 1C, 2B). In one instance, a big central vacuole was observed in one thallus, whereas multiple thalli in the same diatom cell were often observed (Fig. 2B). CW staining highlighted the presence of broadly conical discharge tubes (Fig. 2C, arrowhead) as well as more elongated ones (Fig. 2D, arrowheads). Hollow spore cysts have been observed within empty sporangia (Fig. 2C, arrows), although clusters of spore cysts at the mouth of the discharge tube were also observed (Fig. 2E). Both these zoospore behaviours have been reported for the Fragilaria islandica infected by E. perforans in Sparrow material (1969). In parasites of Licmophora sp., thalli were mainly spherical in shape and closely surrounded by red-brown degraded plastids when developing (Fig. 1G, H, 2G). At maturity, the sporangial cell wall was weakly stained by CW (Fig. 2G, inset). Empty sporangia observed in our sample were less regular in shape and bore up to four broadly conical discharge tubes (Fig. 2H, I; arrowheads). These sporangial features are consistent with the descriptions of lenticular/spherical or irregular/saccate sporangia reported for Licmophora infected by E. perforans in Johnson (1966). In few cases, spores could be observed within the sporangium, but in no instance, we were able to observe their flagellation. Spores are spherical and seem to surround a central vacuole when within the sporangium (Fig. 2J). Empty spore cysts could be observed both within and outside the sporangium (Fig. 2H, I; arrows). Despite the lack of first-hand experience of spore discharge behaviour, our observations of empty spore cysts are consistent with observations made by Raghu Kumar (1980b) on E. perforans in culture, where zoospores were observed to either escape before encysting or to encyst within the sporangium or at the mouth of the discharge tube. Discharge tubes with a circular or broadly ellipsoidal outline were observed to coexist in the same sporangium, incorporated in a thickened area of the sporangial wall (Fig. 2K, L).

Co-existence of two taxonomically distinct oomycetes in a single host cell
Single-cell (or chain) whole genome amplification followed by PCR allowed to gather genetic markers used to phylogenetically place the Ectrogella-like organisms detected and described above. Two distinct oomycete 18S rDNA and cox2 sequences were amplified from the cell InfLicSC2 (noted InfLicSC2-a and InfLicSC2-b in Fig. 3, 4; Table S2). The 18S rDNA InfLicSC2-b was very closely related to Olpidiopsis drebesii within the OOM_2 clade whereas InfLicSC2-a was identical with the 18S rDNA sequences retrieved from InfLicSC1, InfLicSC3, InfLicSC4 (Fig. 3). cox2 data were in full agreement with results from the 18S rDNA, though less easily resolved due to the paucity of reference sequences for the OOM_2 clade (Fig. 4). Bearing in mind that some oomycetes affiliated to the genus Olpidiopsis are themselves intracellular parasites of oomycetes (e.g. the type of the genus, Olpidiopsis saprolegniae) and that hyperparasites (i.e. parasitic organisms whose host is itself a parasite) have been reported in endocellular holocarpic oomycetes (e.g. Pythiella besseyi in Olpidiopsis schenkiana Sparrow and Ellison 1949;Sparrow 1960), we next investigated the possibility of a cryptic hyperparasitism of the Ectrogella parasite: we reasoned that if a hyperparasite was present, its sequence should always be associated to its oomycete host. Therefore, two specific reverse primers were designed to target either of the two rDNA sequences (18S_SC1spe_R1 and 18S_SC1spe_R2 for the 18S rDNA of InfLicSC1 and 18S_SC2spe_R3 and 18S_SC2spe_R4 for the 18S rDNA of InfLicSC2-b, Table S1). Those were used in direct and nested PCRs on the MDA-amplified InfLicSC1, InfLicSC2, InfLicSC3, and InfLicSC4. Overall, only one oomycete sequence was retrieved from each isolated cell, except in InfLicSC2. On the other hand, the InfLicSC2-b 18S and cox2 sequences have never been obtained in our laboratory apart from this sample, limiting the risk of an external contamination of the InfLicSC2 isolate by foreign DNA. Finally, a close re-inspection of our photographs highlighted the possible presence of two thalli inside the host diatom cell (Fig. 1F). Therefore, we conclude that InfLicSC2 was concurrently infected by two taxonomically distinct oomycete parasitoids; our data do not give us any reason to suspect a nonrandom association between them, such as a hyperparasitic relationship.

Molecular phylogeny of the parasitoids infecting
Licmophora sp. and Fragilaria sp Overall, three distinct 18S rDNA sequences were retrieved. The 18S rDNA of InfLicSC1 (identical to InfLicSC2-a, InfLicSC3, InfLicSC4) diverges early and without bootstrap support from the group of Olpidiopsis parasites of red seaweeds, together with Olpidiopsis pyropiae, a pathogen of Pyropia sp. recently discovered in Korean laver farms (Klochkova et al. 2016) and Pontisma lagenidioides, recently isolated from Ceramium rubrum (Buaya et al. 2019a). The 18S rDNA InfLicSC2-b is closely related to Olpidiopsis drebesii within the OOM_2 clade. Finally, the 18S rDNA S10InfFrLic and S10InfLicSC (identical to S10InfLicSC6) fall within the OOM_1_2 subclade, the only morphologically known member of which is a parasitoid of Pseudo-nitzschia fraudulenta (Garvetto et al. 2018; Fig. 3). Likewise, three distinct cox2 sequences were retrieved. Their position in the phylogenetic tree is consistent with the 18S rDNA data, despite a generally lower resolution of the tree (Fig. 4). The latter is due to the limited availability of reference sequences in the alignment, and their shorter length, a well-known limitation of cox2-based oomycete phylogenies (Badis et al. 2019).

Parasitoids of Licmophora are polyphyletic
Here, we show that parasitoids of Licmophora sp. fall into three distinct taxonomic entities: all of them belong to the "early diverging oomycetes" (sensu Beakes et al. 2012), a poorly known group that contains the Haliphthorales, Anisolpidiales, Pontismatales, Olpidiopsidales (sensu Badis et al. 2019), Eurychasmales and Haptoglossales (Garvetto et al. 2018;Buaya et al. 2019a; Fig. 3). However, these three parasitoid taxa of Licmophora are polyphyletic and may co-occur in nature. The implications of our findings are several-fold. Firstly, we demonstrate that the genus chains, the infected chain shows developing parasitic thalli and empty sporangia (asterisks). (C) DIC and CW epifluorescence overlaid picture featuring a growing thallus (left) and three empty sporangia. Note the thickened discharge tubes (arrowheads) and empty spore cysts (arrows). CW epifluorescence (D) and DIC (E) pictures of an infected Fragilaria sp. chain in apical view. Note the thick-walled discharge tubes (arrowheads) and a cluster of released spores at the mouth of the discharge tube at the top of the picture. (F) Healthy uninfected Licmophora sp. cell. (G) Spherical mature sporangium surrounded by collapsed red-brown chloroplasts. Note the sporangia wall reacted positively to CW (inset). (H and I) The same empty sporangium featured in two different focal planes showing four discharge tubes (arrowheads) and empty spore cysts both within and outside the sporangium (arrows). (J) Spherical spores as observed within the sporangium. Note a big vacuole seems to occupy the centre of the sporangium. (K) A mature sporangium in Licmophora sp. is filled with zoospores and shows an opened discharge tube (arrowhead). (L) The same sporangium marked with CW and observed in epifluorescent microscopy highlights a second discharge tube (arrowheads) in a thickened area of the sporangial wall.

Figure 3
Maximum-likelihood reconstruction (1,000 ultrafast bootstrap) of oomycetes (grey area) phylogeny based on the 18S rDNA gene sequence. Isolate names in bold indicate parasites sequenced in this study and in smaller font share a > 99% sequence identity with the represented sequence and have been therefore omitted from the alignment. Well-supported (> 95% UFBoot support) nodes highlighted in red may correspond to Ectrogella, in the light of our findings.
Ectrogella is polyphyletic. Unfortunately, evidence is lacking to link either of our sequences to the type of the genus, the freshwater E. bacilliacearum (Zopf, 1884). In the light of our results, Ectrogella may thus correspond either to the subclade OOM_1_2, the clade OOM_1, the clade OOM_2, encompass the branch retrieved with the parasitoid InfLicSC1 or fall in a clade yet to be sequenced (Fig. 3). Therefore, we suggest to retain the temporary nomenclature OOM_1 and OOM_2 (Garvetto et al. 2018), compliant with the EukRef guidelines (Berney et al. 2017) until the typification of Ectrogella is clarified and enough information is available to formally describe each taxon. Secondly, we demonstrate that the Olpidiopsis parasites of red algae are unlikely to be monophyletic and that our diatom parasitoids are distinct from the freshwater parasitoid of Pleurosigma sp., Olpidiopsis gilli (Buaya et al., 2019b). Support is lacking to link either of the "pyropiae," "porphyrae" or "bostrychiae" lineages (sensu Badis et al. 2019) to the genus type, Pontisma lagenidioides; therefore, the re-assignation of all red algal Olpidiopsis pathogens to the genus Pontisma Figure 4 Maximum-likelihood phylogenetic tree reconstruction (1,000 ultrafast bootstrap) using amino acid sequences for the mitochondrial genetic marker cox2 (196 positions) of the parasites of Licmophora sp. and Fragilaria sp. (isolate names in bold) within the oomycetes. Isolate names in brackets and in smaller font share a > 99% sequence identity with the represented sequence and have been therefore omitted from the alignment.
proposed by Buaya et al. (2019a) remains unconfirmed. Overall, until type identities and phylogenetic placement are ascertained, we feel that caution should be applied before updating the nomenclature of these groups. Finally, the natural co-existence of three phylogenetically distinct parasitoids infecting Licmophora in the same sample means that all morphological observations conducted thus far on diatom parasitoids, to the exception perhaps of the ultrastructural work by Raghu Kumar (1980a,b), may have been performed on a mix of several species. This co-existence may explain to some extent the ongoing disagreement between respected taxonomists about species delimitation in Ectrogella; it also means that a very careful appraisal of the literature is required to link existing taxonomic descriptions with novel molecular data. Although useful, the limited morphological evidence that we have gathered on a limited number of cells is insufficient to tackle this question. In the future, only single-cell approaches, hopefully combined with the laboratory cultivation of clonal isolates, will enable to resolve the taxonomy of these parasitoids.

ACKNOWLEDGMENTS
This project has received funding from the European Union's Horizon 2020 research and innovation programme under the Marie Skłodowska-Curie grant agreement No. 642575 and the grant agreement No. 727892.