Polyphyletic origin, intracellular invasion, and meiotic genes in the putatively asexual agamococcidians (Apicomplexa incertae sedis)

Agamococcidians are enigmatic and poorly studied parasites of marine invertebrates with unexplored diversity and unclear relationships to other sporozoans such as the human pathogens Plasmodium and Toxoplasma. It is believed that agamococcidians are not capable of sexual reproduction, which is essential for life cycle completion in all well studied parasitic apicomplexans. Here, we describe three new species of agamococcidians belonging to the genus Rhytidocystis. We examined their cell morphology and ultrastructure, resolved their phylogenetic position by using near-complete rRNA operon sequences, and searched for genes associated with meiosis and oocyst wall formation in two rhytidocystid transcriptomes. Phylogenetic analyses consistently recovered rhytidocystids as basal coccidiomorphs and away from the corallicolids, demonstrating that the order Agamococcidiorida Levine, 1979 is polyphyletic. Light and transmission electron microscopy revealed that the development of rhytidocystids begins inside the gut epithelial cells, a characteristic which links them specifically with other coccidiomorphs to the exclusion of gregarines and suggests that intracellular invasion evolved early in the coccidiomorphs. We propose a new superorder Eococcidia for early coccidiomorphs. Transcriptomic analysis demonstrated that both the meiotic machinery and oocyst wall proteins are preserved in rhytidocystids. The conservation of meiotic genes and ultrastructural similarity of rhytidocystid trophozoites to macrogamonts of true coccidians point to an undescribed, cryptic sexual process in the group.


Results
occurrence and morphology of new rhytidocystid species. Rhytidocystis nekhoroshkovae sp. n.
Parasites were found in all 30 examined polychaetes Pectinaria (Cistenides) hyperborea collected in the vicinity of Educational and Research Station "Belomorskaya" of Saint Petersburg State University (ERS SPbU, see Methods for details). Infected midguts showed plenty of white dots on the outside surface, which corresponded to rhytidocystids, generally located at the basal part of the midgut epithelium. The polychaetes were usually heavily infected (hundreds of parasites per host).
Spindle-shaped zoites were observed inside the host enterocytes (Fig. 1A). As zoites grew, they lost their elongated shape and transformed into trophozoites (Fig. 1B). Early development took place intracellularly and young trophozoites were located inside parasitophorous vacuoles (Fig. 1C). Both young and adult trophozoites had near-round or irregular shape with a slightly uneven border and measured 13.0-68.0 µm in maximal dimension (av. 47.4 ± 2.54 µm, n = 23). The trophozoites' cytoplasm was filled with granules of storage carbohydrate (presumably, amylopectin), and smaller cells were more transparent than larger ones (Fig. 1D). Live parasites had a spherical nucleus located centrally and measured 6.6-19.2 µm in diameter (av. 14.64 ± 0.41 µm, n = 17). A single medium-sized spherical nucleolus was eccentric. Adult trophozoites were outside host cells close to the basal lamina of the midgut epithelium. No pathological changes were observed in infected tissue: the neighboring enterocytes had an appearance of active digestive cells with numerous phagosomes (Fig. 1E). Parasites isolated from the host gut were immotile. Their cell surface was rugose with longitudinal and transverse grooves, little creases, and small depressions (Fig. 1F). Numerous micropores on the parasite surface were arranged in curved rows, which merged with each other (Fig. 1G).
Rhytidocystis dobrovolskiji sp. n. Parasites were found in 34 out of 70 (48.6%) examined polychaetes Ophelia limacina collected in the vicinity of ERS SPbU (see Methods for details). Parasites were embedded in the host midgut epithelium and visible as white dots from the inside and outside the gut. The hosts contained between several and several dozens of parasites.
The early trophozoite stages were crescent-shaped and measured 21.6-41.0 µm long (av. 32.9 ± 2.37 µm, n = 10) ( Fig. 2A). Occasionally, trophozoites were found tightly packed inside the host cells (Fig. 2B). Maturing trophozoites measured 36.3-67.0 µm in maximal dimension (av. 51.0 ± 1.13 µm, n = 33), were irregular or roundish (Fig. 2C) and became spherical in a short time after the release from the host tissue (Fig. 2D). All forms had a spherical nucleus with a relatively large eccentric or centric nucleolus, usually located in the central part of the cell and measured 3.6-10.0 µm in diameter (av. 6.8 ± 0.61 µm, n = 10) in crescent-shaped forms and 12.1-25.0 µm (av. 18.2 ± 0.83 µm, n = 18) in maturing ones. The cell surface of spherical trophozoites was smooth (Fig. 2E). Only once we observed young oocysts released from the host intestinal epithelium. They were spherical and covered by a thick transparent envelope (Fig. 2F). The nuclei were not clearly visible in all of them (Fig. 2G).
Rhytidocystis pertsovi sp. n. Parasites were found in 73 out of 106 (68.9%) examined polychaetes Ophelia limacina collected at the White Sea Biological Station of Lomonosov Moscow State University (WSBS MSU, see Methods for details). Similar to Rh. dobrovolskiji, parasites were located in the host midgut epithelium. The infected polychaetes contained from a few up to hundreds of parasites.
Intracellular crescent-shaped trophozoites measured 14.3-25.3 µm long (av. 19.86 ± 1.36 µm, n = 7) were found inside the host enterocytes (Fig. 3A). Crescent-shaped trophozoites were sometimes located in pairs inside a one host cell (Fig. 3B). We observed several crescent-shaped trophozoites slowly becoming bean-shaped after the releasing from the host midgut epithelium (Fig. 3C). A spherical nucleus measuring 4.0-6.6 µm in diameter (av. 5.1 ± 0.51 µm, n = 5) was usually located in the central part of the cell and had a spherical eccentric nucleolus. Larger trophozoites, 21.3-59.2 µm in maximal dimension (av. 49.15 ± 1.66 µm, n = 32), were irregular or roundish in shape (Fig. 3D). They had a spherical centric nucleus measuring 8.0-20.0 µm in diameter  (Fig. 2E).   Both Bayesian (BI) and Maximum Likelihood (ML) analyses produced almost identical tree topologies except for the position of blastogregarines, which were either the sister lineage to coccidiomorphs (ML; not shown) or placed among gregarines (BI; Fig. 5). The Bayesian tree inferred from the concatenated rDNA dataset of 99 taxa and 4,517 sites (Fig. 5) showed the monophyly of major alveolate lineages with high posterior probabilities (PP), but with moderate or low ML supports (bootstrap percentages, BP) in the apicomplexan part of the tree.
The sequences of the three new Rhytidocystis species had relatively short branches and grouped into a common rhytidocystid clade with the other representatives of the genus (Rh. cyamus and Rh. polygordiae), an undescribed parasite of the European oyster Ostrea edulis, and several environmental sequences from oceanic sediments with PP = 1 and BP = 88%. The rhytidocystids then formed a robust higher-order clade (PP = 1.0 and BP = 100%) with an unidentified parasite of Tridacna croecia, the coccidians Margolisiella islandica and Pseudoklossia pectinis, and an environmental sequence from a sulfidic karst spring in Slovenia (KT072247). The robust Rhytidocystis-Pseudoklossia-Margolisiella clade was the most early-branching lineage of coccidiomorphs, including coccidians and haematozoans (PP = 1, BP = 83%). We suggested the name "Eococcidia" for this clade (Fig. 5, also see "Discussion").
Analysis of meiosis-specific and oocyst wall protein transcripts. We examined the available transcriptomic data of rhytidocystids for transcripts of meiosis-specific genes to estimate whether meiotic recombination is possible in the reportedly asexual rhytidocystids. Homology searches identified seven meiosis-specific    (Fig. 6). Nearly all gene transcripts were partial or incompletely spliced in the assemblies, indicating low transcript presence in the sequencing libraries. Phylogenetic analyses grouped the rhytidocystid meiotic genes with other apicomplexans, ruling out that they could be contaminating sequences (Supplementary Figs. S1-S9). The rhytidocystid transcriptomes lack two genes of the core meiotic gene set-Msh4 and Msh5, but retain meiotic helicase Mer3 and a member of the Rad21/Rec8 cohesin family, which are both absent in sexual coccidians. The discovery of young oocysts in Rh. dobrovolskiji and the presence of putative wall-forming bodies in Rh. pertsovi prompted us to search the rhytidocystid transcriptomes for genes encoding oocyst wall proteins previously described in Toxoplasma and Cryptosporidium. We found over one hundred transcripts related to the apicomplexan oocyst wall family proteins COWP and TgOWP1-7 16,17 in the transcriptomes of the two Rhytidocystis species. Similarly to other apicomplexans, the rhytidocystid sequences are characterized by the N-terminal signal peptide followed by a series of cysteine-containing repeats. The tree reconstruction of apicomplexan COWP and TgOWP1-7 family proteins groups rhytidocystid sequences into 19 divergent clusters (Fig. 7A). The rhytidocystid OWP clusters are distributed evenly in the tree, pointing to their large diversity, with several clusters showing potential orthology to the characterized OWPs of Cryptosporidium and Toxoplasma: COWP4, COWP5, COWP9, Figure 6. Occurrences of meiosis-specific genes in alveolates. Rhytidocystis spp. represents combined transcriptomic data of Rh. pertsovi and Rhytidocystis sp. ex Travisia forbesii. Pie charts show completeness scores of underlying genomic or transcriptomic data as estimated by BUSCO; filled boxes correspond to complete or fragmented orthologs. The Spo11 box includes either Spo11-1, Spo11-2, or both orthologs; non-meiotic Spo11-3/Top6A orthologs were not considered. Homologs reported in the Rec8 category include any findings of the Rad21/Rec8 family, as specific orthology of the proteins is difficult to determine (see also Supplementary Figs. S1-S9). www.nature.com/scientificreports/ TgOWP3, and TgOWP4 sequences. Another abundant family of candidate OWPs in the two rhytidocystid species (over 200 of total identified sequences) is homologous to the more recently described TgOWP8-12 18 .
Similarly to the TgOWP1-7 family, the proteins contain periodical cysteine residues and a signal peptide. The TgOWP8-12 family was significantly expanded in rhytidocystids compared to the coccidians (Fig. 7B). However, that the majority of TgOWP1-7 and TgOWP8-12 homologs in the rhytidocystid transcriptomes are incomplete, which confounds their classification, especially considering the repetitive structure.

Discussion
In the present study, we described three new species of Rhytidocystis agamococcidians, defined their phylogenetic position and surveyed for molecular markers of their sexual reproduction and oocyst wall formation. The new data reveal several key points on rhytidocystid biology and evolution.
Rhytidocystids are most likely distributed worldwide and prefer to parasitize opheliid polychaetes. Previously Riser 21 had also observed putative rhytidocystids (described as "coccidians"), which were located within projected apical ends of host enterocytes, similar to Rh. pertsovi. Apparently, most of the real biodiversity of rhytidocystids remains undiscovered.
eococcidia: the sister group of coccidians and haematozoans. Previous phylogenetic studies of SSU rDNA sequences did not resolve the position of rhytidocystids 6,7 , making it unclear whether they could be related to Gemmocystis cylindrus, the other agammococcidian now represented by the "genotype-N" and coralicollid sequences. In some phylogenies, Rh. cyamus and Rh. polygordiae were the sister group of the coccidians Margolisiella islandica and Pseudoklossia pectinis 15,22,23 , but their common clade was never strongly affiliated with gregarines, cryptosporidians, or coccidiomorphs. Reasons for the lack of resolution might be two-fold: Rh. cyamus and Rh. polygordiae form long tree branches which may cause the long branch attraction artifact, and the earlier studies used SSU rDNA phylogenies only, which have inferior resolution to the whole rRNA operon [24][25][26][27] .
In the current study, we enlarged a broadly and evenly sampled alveolate dataset with environmental sequences and short-branching sequences of our new species. We also sequenced the first rhytidocystid LSU rDNA and analyzed their concatenated rDNA phylogeny in both Maximum likelihood and Bayesian frameworks. The resulting phylogenies have a resolution superior to earlier studies and resolve rhytidocystid and sporozoan relationships in two important ways. Firstly, rhytidocystids are unrelated to corallicolids which include Gemmocystis cylindrus (Fig. 5), demonstrating that the order Agamococcidiorida is polyphyletic. Secondly, the analyses unambiguously combined rhytidocystids with coccidiomorphs (coccidians and haematozoans). This finding matches the more sparsely-sampled but multiprotein phylogeny of apicomplexans based on 296 concatenated markers 14 , which recovered rhytidocystids as basal coccidiomorphs. Margolisiella islandica infecting the Iceland scallop Chlamys islandica is the closest described relative of rhytidocystids with complete Leukart's triad in monoxenous life cycle ( Fig. 5) 15,22,23 . Pseudoklossia pectinis is a putatively heteroxenous parasite of Pecten maximus: gamogony and sporogony occur in the great scallop, and a merogonic phase is supposed to be in some other host 28 . Other members of the Rhytidocystis-Pseudoklossia-Margolisiella clade are poorly-studied (unnamed parasites of Tridacna croecia and Ostrea edulis), and the whole clade lacks obvious shared morphological characteristics (synapomorphies). Nevertheless, since this robust clade has been recovered in several analyses and discovered as basal group of all coccidiomorphs, including coccidians and haematozoans, we suggest establishing the new taxon Eococcidia for basal coccidiomorphs mainly parasitizing marine invertebrates. The term "Eococcidia" refers to Eos-the goddess of the dawn in Greek mythology, who rose in the morning from the Oceanus; the prefix "eo-" is used in geology and biology for the designation of something to be early (Eococcidia-early coccidians). Coccidian genera Merocystis and Aggregata are also candidate members to Eococcidia since they were recovered as close relatives of Rhytidocystis, Pseudoklossia, and Margolisiella 23,29 . intracellular parasitism links rhytidocystids with other coccidiomorphs. The phylum Apicomplexa contains both extracellular and intracellular parasites. Development of many gregarines runs extracellularly, whereas coccidians and haematozoans invade host cells 30,31 . Four out of five previously described rhytidocystids have trophozoites in the host intestinal epithelium 3,4,6,7 , but their intracellular stages were never detected. Here we provide evidence of intracellular localization of three new rhytidocystid species. The findings of putative sporozoites inside host cells and young trophozoites within parasitophorous vacuoles in Rh. nekhoroshkovae, and trophozoites tightly packed inside host cells in Rh. dobrovolskiji suggest that the development of both species begins intracellularly. The trophozoites of Rh. pertsovi on squash preparations of the host enterocytes and in TEM sections were undoubtedly located under the plasma membrane of the host cell. Perhaps, the development of other rhytidocystid may also start with intracellular forms. Rh. polygordiae was the only species previously described with TEM, and its trophozoites were observed in the interstitial space between adjacent host enterocytes 6 . However, the structure purported to be "the plasma membrane of the adjacent epithelial cell" Scientific RepoRtS | (2020) 10:15847 | https://doi.org/10.1038/s41598-020-72287-x www.nature.com/scientificreports/ ( Fig. 19 in 6 ) has the thickness (a little more than 20 nm) and appearance of two membranes with the intercellular space between them, and most likely represents two plasma membranes of two adjoining enterocytes. Furthermore, "the interstitial space" (Fig. 19 in 6 ) is filled with structures resembling endoplasmic reticulum and apparently is the host cell cytoplasm. We therefore suppose that at least some trophozoites of Rh. polygordiae were located intracellularly. Overall, evidence for intracellular development strongly links rhytidocystids with other coccidiomorphs such as coccidians and haematozoans to the exclusion of gregarines and cryptosporidians, which are chiefly extracellular or epicellular. Congruent with their phylogeny (Fig. 5) 14 , and intracellular stages in Margolisiella 15 , this distribution suggests that intracellular invasion evolved in the common coccidiomorph ancestor. Unlike other coccidiomorphs, however, rhytidocystid trophozoites grow up to a relatively large size, destroy infected cells, and end up lying extracellularly within the host tissue. Intracellular apicomplexans may be either embedded in a parasitophorous vacuole (PV) made of components of host origin or host and parasite origin, or be in direct contact with the host cell cytoplasm [32][33][34] . In the first case a zoite penetrates the host cell membrane, induces the PV formation and becomes surrounded by PV since the beginning of its intracellular development 35,36 . Unexpectedly, younger forms of Rh. pertsovi were not within a parasitophorous vacuole in our TEM sections (Fig. 4A,B), whereas several larger ones were (Fig. 4C,D). This result does not correspond to typical PV development, so a thorough TEM investigation of rhytidocystid intracellular development is needed.
Morphology of trophozoite and oocyst stages. The trophozoite cell shape of earlier described Rhytidocystis species varies from oblong Rh. polygordiae and bean-shaped Rh. cyamus to flat oval cells of Rh. opheliae and Rh. henneguyi 3,4,6,7 . In the case of Rh. pertsovi and Rh. dobrovolskiji, it seems to be that crescent-shaped, bean-shaped, irregular, roundish and spherical forms represent successive stages of development. A zoite invades the host cell and transforms into a trophozoite. Presumably, during the growth inside the limited space of the host cell, the young trophozoite bends and becomes crescent-shaped, then it loses peaked cell poles and becomes bean-shaped; over time, it undergoes marked growth and becomes tightly packed under the host cell membrane. Released from the host cell to the interstitial space of the tissue, a trophozoite unbends, becomes irregular, then roundish and spherical eventually. Spherical trophozoites, apparently, represent the transitional form to the oocyst stage. Young oocysts Rh. dobrovolskiji look like spherical trophozoites covered by a thick transparent envelope and resemble young oocysts of Rh. sthenelais 37 .
The cytoplasm of intracellular trophozoites Rh. pertsovi looks typical of sporozoans and possesses all general structures. Rh. pertsovi retains an active apicoplast 14 but none was observed in this TEM study. Instead, we found numerous dense bodies of oval or round shape, which could be homologous to wall-forming bodies (WFBs) in coccidian and cryptosporidian macrogamonts [38][39][40][41] . The WFBs mediate the formation of the oocyst wall, which more than 90% is made up of proteins 42 . The presence of oocysts and putative WFBs has prompted us to search in rhytidocystid transcriptomes for homologs of oocyst wall proteins (OWPs), some of which are proven to be located in WFBs 40,43 . The analysis revealed an astounding diversity of transcripts related to COWP, TgOWP1-7 and TgOWP8-12 family proteins, supporting the presence of WFBs in rhytidocystids and a common mechanism for their oocyst wall formation with coccidia and cryptosporidia. Discovery of meiotic genes in "asexual" rhytidocystids. Since rhytidocystids are closely related to M. islandica, which is an eucoccidian-like protist, whose life cycle includes all three types of sporozoan reproduction: gametogony, merogony, and sporogony (Leuckart's triad), and which, hence, produces sexual gamonts 15 , the supposed absence of sexual life stages in rhytidocystids raises the possibility that they recently lost sexual reproduction. Generally conserved across the eukaryotes, the core meiosis machinery includes nine proteins (Spo11, Hop1, Hop2, Mnd1, Dmc1, Mer3, Msh4, Msh5, Rec8), with functions spanning sister chromatid cohesion, induction of double-strand breaks, heteroduplex DNA and synaptonemal complex formation, and Holliday junction resolution 44,45 . In rhytidocystids, the core meiotic gene set is short of two genes for Msh4 and Msh5, unlike the chrompodellid Vitrella brassicaformis, which retains a full set of core meiosis-specific genes and where sexual process has been proposed 46 . The absence of the heterodimer-forming Msh4 and Msh5 in rhytidocystids is consistent with their absence in other coccidiomorphs' genomes ( Fig. 6): they are involved in the stabilization of Holliday junctions and meiotic crossover interference in model organisms 47 but apparently dispensable in apicomplexans 48 . Notably, the closely related Msh2 family, which is involved in DNA repair, has expanded in Rhytidocystis sp. ex Travisia forbesii. Unlike many sporozoans, rhytidocystids retain the meiotic helicase Mer3 and a member of the Rad21/Rec8 cohesin family. Thus, the inventory of meiosis-specific genes in rhytidocystids does not display evidence of reduction in relation to the same gene set of sexual coccidians. The presence of these genes alone, however, does not constitute conclusive evidence of sexual reproduction in the family. Meiosis-specific genes, contrary to their designation, were reported to have functions outside of meiosis, specifically in homologous recombination and DNA repair 49 . The preservation of these genes weighs in favor of meiotic recombination in rhytidocystids, but more direct evidence would be necessary to verify the existence of sexual process.
In terms of appearance, rhytidocystid trophozoites are similar to macrogamonts of their closest relativessexual coccidians: they develop intracellularly, amass a supply of nutrients, produce wall-forming bodies and eventually become the oocysts. The presence of meiosis-specific transcripts challenges the long held belief that rhytidocystids lack gamogony and inspires search for their cryptic sexual process, which has remained hidden for over a hundred years. Future research on early rhytidocystid development and genetics are awaited to contribute to this matter.

Scientific RepoRtS
| (2020) 10:15847 | https://doi.org/10.1038/s41598-020-72287-x www.nature.com/scientificreports/ taxonomic summary. Here we use the most recent Adl et al. 13 system for higher-ranks as phylum and class, despite the inconsistency of this system to the actual phylogeny of Apicomplexa (Fig. 5) 14,26,50 , and due to the absence of a correct system. The eococcidians are as close to coccidians as to haematozoans (Aconoidasida) in our phylogeny. However, we classify Eococcidia into Conoidasida and Coccidia because of the findings of the apical complex in their zoites 6,37,51 . We suggest the order Pseudoklossiida for Pseudoklossia and Margolisiella (former Eimeriorina) as P. pectinis and M. islandica were recovered strongly within the eococcidians, but not within the eimeriids (Fig. 5) 15,23 . We keep the rhytidocystids into the order Agamococcidiida as no microgamonts and microgametes were found, only trophozoites that potentially may be macrogamonts. We do not include Gemmocystis to the Agamococcidiida as we consider it belonging to the corallicolids. The mature sporulated oocysts of the newly described rhytidocystids were not observed, but we use the characteristics of the previously described rhytidocystid oocysts for the taxon diagnosis 3,37 . We still lack distinct morphological synapomorphies both for Eococcidia and for Pseudoklossiida; therefore, the establishment of these taxa is mainly based on molecular data.
Coccidia. Homoxenous and heteroxenous parasites of marine invertebrates, predominantely polychaetes and mollusks. Molecular data: earlier robust sister clade to coccidians and haematozoans in rDNA and multiprotein phylogenies. Etymology: from "eo-", a prefix meaning the earliest appearance, and "coccidia".
Agamococcidiida. Early development intracellular; adult trophozoites extracellular in the host intestinal epithelium or coelom; large oocysts with many tens of sporocysts in the host intestinal epithelium or coelom; in annelids. For TEM study dehydrated samples were transferred to an ethanol/acetone mixture 1:1 (v/v), rinsed twice in pure acetone, and embedded in Epon resin using a standard procedure. Ultrathin sections obtained using LKB-III (LKB-produkter, Sweden) or Leica EM UC6 (Leica Microsystems, Germany) ultramicrotomes were contrasted with uranyl acetate and lead citrate 53 and examined under a JEM 1011 electron microscope (JEOL, Tokyo, Japan). DNA extraction, PCR amplification, and sequencing of rDNA. Isolated trophozoites (up to 50 cells from each location) were washed three times in filtered sea water and deposited into 1.5 ml microcentrifuge tubes. Samples of Rh. nekhoroshkovae and Rh. pertsovi were fixed with 96% ethanol, the sample of Rh. dobrovolskiji was fixed with RNA-later (Life Technologies, USA). Extraction of DNA from fixed cells was performed using the NucleoSpin Tissue kit (Macherey-Nagel GmbH & Co. KG, Germany). Whole Genome Amplification (WGA) was performed for Rh. nekhoroshkovae and Rh. dobrovolskiji samples using REPLI-g Midi Kit (Qiagen, UK). The contiguous nucleotide sequences (SSU, ITS1, 5.8S, ITS2 and LSU rDNAs) were assembled from a series of overlapping fragments obtained by PCR with different pairs of primers, followed by Sanger sequencing (see 26,27,50 for the general approach). The rDNA fragments were amplified with Encyclo PCR kit (Evrogen, Russia) in a total volume of 20 µl using a DNA Engine Dyad thermocycler (Bio-Rad) and a T100 Thermal Cycler (Bio-Rad). General scheme of PCR protocol as follows: lid temperature 100 °C; initial denaturation at 95 °C for 2.5 min; 40 cycles of 95 °C for 30 s (denaturation); 48-55 °C for 30 s (annealing); 72 °C for 1.5 (elongation) and a final extension at 72 °C for 10 min. Table 1 shows the lengths of amplified and overlapping fragments, sequences of used oligonucleotides and exact annealing temperatures.

Genus
Assembling of rRnAs sequences from transcriptomic data. The rDNAs of Pterospora schizosoma, Monocystis agilis, Lecudina tuzetae, and Heliospora caprellae were assembled from the transcriptome sequencing data generated by Mathur et al. 56  www.nature.com/scientificreports/ SRR8980200-SRR8980205, SRR8980208-SRR8980213, SRR1300212), and the assemblies were performed with SPAdes 58 utilizing k-mer size 127 and Trinity 59 programs. Assembled rDNA contigs were aligned using MAFFT 60 and inspected by eye for assembly errors and chimeric sequences.  63 under the GTR + Г model and CAT approximation (25 rate categories per site). The procedure included 100 alternative runs of the ML analysis and 1,000 replicates of multiparametric bootstrap. Bootstrap percentages were merged on the user trees (both ML and BI) with the same program. Bayesian inference (BI) analyses were done in MrBayes 3.2.6 64 under GTR + Г + I model with 12 discrete categories of gamma distribution. The following parameters were used: nst = 6, Table 1. Main characteristics of the rhytidocystid sequences obtained in this study. a The primer sequence was based on Medlin et al. 54 . b The primer sequences were based on Van der Auwera et al. 55 . b The primer sequence was tailored specially for Rh. dobrovolskiji. www.nature.com/scientificreports/ ngammacat = 8, rates = invgamma; parameters of Metropolis Coupling Markov Chains Monte Carlo (mcmc): nruns = 2, nchains = 4, temp = 0.2, ngen = 10,000,000, samplefreq = 1,000, burninfrac = 0.5. The average standard deviation of split frequencies at the end of computations was 0.001441.

Analysis of meiosis-specific and oocyst wall protein transcripts. Searches for meiosis-specific and
oocyst wall protein families in the transcriptomes of Rhytidocystis species were carried out with HMMER 65 using profiles constructed from protein family alignments. The corresponding protein families were identified with OrthoFinder 66 clustering with predicted protein sequences in a set of 70 eukaryotic genomes; the protein family alignments were generated using MAFFT 60 . We utilized the following sources for genomic data: Genome database of NCBI (https ://www.ncbi.nlm.nih.gov/genom e), Genome Portal of DOE JGI (https ://genom e.jgi.doe. gov/porta l/), Ensembl Protists resources (https ://proti sts.ensem bl.org/), genome projects of Marine Genomics Unit (https ://marin egeno mics.oist.jp/), and genomic resources of multicellgenome lab (https ://multi cellg enome .com/). The transcriptomic data for Rhytidocystis pertsovi and Rhytidocystis sp. ex Travisia forbesii were obtained from GenBank transcriptome shotgun assembly projects GHVQ00000000.1 and GHVS00000000.1. The transcriptome assemblies of Rhytidocystis species were processed with TransDecoder 67 utilizing BLAST 68 and HMMER searches against the UniProtKB/Swiss-Prot 69 and Pfam 70 databases for ORF prediction. Orthology of proteins discovered by HMMER profile searches was verified by reciprocal BLAST searches against OrthoFinder orthogroups: proteins with best hit outside of the queried protein family were excluded from the set of findings. The findings satisfying reciprocal BLAST search criterion were added to the protein family alignments using MAFFT 60 , and the family membership was further inspected by reconstructing phylogenies. The trees for meiosis-specific protein families were reconstructed by IQ-TREE 71 using the LG + C10 + F + G4 profile mixture model, and ultrafast bootstrap approximation 72 with 1,000 replicates for estimation of branch support; IQ-TREE reconstructions for oocyst wall protein families utilized ModelFinder 73 to automatically select the best-fit model. The trees were visualized using MEGA 74 and iTOL 75 . The completeness estimates for genomic and transcriptomic data were performed with BUSCO 76 using the eukaryota_odb9 dataset.