Rare Freshwater Ciliate Paramecium chlorelligerum Kahl, 1935 and Its Macronuclear Symbiotic Bacterium “Candidatus Holospora parva”

Ciliated protists often form symbioses with many diverse microorganisms. In particular, symbiotic associations between ciliates and green algae, as well as between ciliates and intracellular bacteria, are rather wide-spread in nature. In this study, we describe the complex symbiotic system between a very rare ciliate, Paramecium chlorelligerum, unicellular algae inhabiting its cytoplasm, and novel bacteria colonizing the host macronucleus. Paramecium chlorelligerum, previously found only twice in Germany, was retrieved from a novel location in vicinity of St. Petersburg in Russia. Species identification was based on both classical morphological methods and analysis of the small subunit rDNA. Numerous algae occupying the cytoplasm of this ciliate were identified with ultrastructural and molecular methods as representatives of the Meyerella genus, which before was not considered among symbiotic algae. In the same locality at least fifteen other species of “green” ciliates were found, thus it is indeed a biodiversity hot-spot for such protists. A novel species of bacterial symbionts living in the macronucleus of Paramecium chlorelligerum cells was morphologically and ultrastructurally investigated in detail with the description of its life cycle and infection capabilities. The new endosymbiont was molecularly characterized following the full-cycle rRNA approach. Furthermore, phylogenetic analysis confirmed that the novel bacterium is a member of Holospora genus branching basally but sharing all characteristics of the genus except inducing connecting piece formation during the infected host nucleus division. We propose the name “Candidatus Holospora parva” for this newly described species. The described complex system raises new questions on how these microorganisms evolve and interact in symbiosis.


Introduction
One of the most studied ciliate genus is Paramecium (Ciliophora, Oligohymenophorea), which comprises nineteen valid morphospecies [1]. Some of these species seem to be cosmopolitan, while other Paramecium species are less widely distributed or might be even considered endemic [2,3]. After the initial work of Müller in 1786 [4], more than forty species descriptions of the Paramecium genus have been published [1,2,[5][6]. Some of these species have been retrieved only once and have not been found anymore after description. This set of species has been treated as uncertain, and some have even been abolished [2]. However, among these doubtful paramecia, some have been found and redescribed as true morphospecies during last two decades: P. duboscqui [7,8], P. nephridiatum [9, 10], P. chlorelligerum [11,12]. As matter of fact, the first two species are currently considered to be widely distributed, instead P. chlorelligerum can be estimated as a very rare ciliate.
This "forgotten" ciliate bears unicellular symbiotic algae in its cytoplasm, and, thus, it can be considered as a "green" Paramecium, like well-known P. bursaria. It has been recently found and redescribed in only a particular pond of Southern Germany, in Simmelried moorland [12]. In 2014 the same ciliate has been retrieved by us also in Peterhof (St Petersburg district, Russia) in a small permanent ditch, which, by some ecological parameters, appears to be quite similar to the German one in Simmelried [12]. The newly found freshwater locality manifests very unusual high diversity and number of green ciliates and can be considered a biodiversity hot-spot for these protists. Some representatives of the Russian population of P. chlorelligerum showed bacterial infection in the macronucleus. The morphology of these endosymbionts fits quite well to the known Holospora-like bacteria [13]. Indeed, representatives of this alphaproteobacterial genus, and some other closely related Rickettsiales, have been found in seven Paramecium morphospecies as well as in Frontonia salmastra and Frontonia leucas [13,14]. These intracellular bacteria are highly infectious nuclear symbionts with unique morphology and life cycle, and they have been considered to be a very promising model organism for investigations of symbiotic associations between eu-and prokaryotes [15]. Indeed, in the last years, several studies have been performed to facilitate their use as model organisms [16][17][18][19], and they have been extensively used in evolutionary ecology studies [20][21][22]. Up to now, the genus Holospora comprises ten species [13,23], which have been described morphologically, but only for several of them molecular characterization is available [23][24][25][26][27].
In the present study, we combined classical morphological approaches with molecular analysis to improve our knowledge on P. chlorelligerum, an extremely rare Paramecium species, examining its variability, physiology and symbiotic interactions both with unicellular algae and prokaryotic microorganisms. Infectious bacteria found in the host ciliate macronucleus were described as a new species of the genus Holospora.

Materials and Methods
Host isolation, cultivation and identification P. chlorelligerum has been found in the small but relatively deep permanent ditch in the corner of the so called English park (Peterhof, St Petersburg district, Russia (N 59˚52 0 47@, E 295 1 0 56@). After the first finding, dated September 2014, sampling was repeated throughout the year covering four different seasons (Fig 1). P. chlorelligerum was constantly present among a large number of ciliated protists, many of which were inhabited by green symbiotic algae.
We succeeded to establish clonal cultures of P. chlorelligerum using combination of classical feeding medium (wheat grass medium inoculated with Klebsiella aerogenes or lettuce medium with the same bacterium) and addition of beta-sitosterol (Merck, Darmstadt, Germany) work in Pisa 2015 mobility grant 1.42.1450.2015 of St Petersburg State University (spbu.ru) to AP; algae studies and infection experiments were supported by RFBR (grant 16-04-01195,rffi.ru) to AP; sequencing was done with a support from RSF (grant 16-14-10157, grant.rscf.ru) to AP. The publication costs were covered by European Commission FP7 post grant Open Access Pilot Open AIRE. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
(0.8 mg/l) and 1/3 volume of mineral medium [28]. Cultures were kept in plastic Petri dishes inside climatic chamber Sanyo at 18˚C with illumination regime of 10 hours darkness and 14 hours light by 2000 lx lamp.
P. chlorelligerum is characterized by unusually low division rate for Paramecium representatives (only 1-2 divisions per week). This could be related with the presence of very few food vacuoles in the cytoplasm of ciliates, indeed we could keep ciliate culture in a good condition for about one month even without feeding.
For the study, four clonal cultures established in laboratory (indexes HSG1 and HSG2 correspond to sampling of 2014 and 2015 respectively) were used: HSG1-6, HSG2-10 (without bacterial infection in the macronucleus), and infected HSG1-11 and HSG2-12. These clones are available from the Centre of Core Facilities "Culture Collections of Microorganisms", St Petersburg State University, Russia.
Species identification of obtained clones was performed according to the ciliate morphology [11,12], and then confirmed by molecular analysis.

Symbiotic algae isolation, cultivation and identification
Attempts of P. chlorelligerum symbiotic algae isolation and cultivation were performed. Several paramecia cells were thoroughly washed in sterile water, then separately squashed and grinded with a glass mini-spatula on agar plates containing Bold Bristol Medium (BBM) [29], glucose (1 g/l) and 10% amino-peptide, and ampicillin (100 μg/ml) in order to minimize bacterial growth. Plates were kept at +24˚C with constant daylight for four-five weeks. Several colonies of green algae were obtained directly on BBM agar plates, then grown in liquid BBM, and purified by repetitive passages on BBM plates. In the experiment, several dozens of paramecia were used, and only once cyst-like cells were observed after several weeks of slow growth. These cysts were transferred to liquid BBM, and after one week of growth a non-axenic algae culture was obtained. Identification of algae cultures was performed by phase contrast microscopy and by molecular methods (see DNA extraction and molecular characterization).

Morphological and Ultrastructural Characterization
Living cells were immobilized for observation with the help of a special device [30]. Impregnation was performed with the silver nitrate procedure made after Champy's fixation [31,32], and nuclear apparatus was also stained with Feulgen method. Living and fixed cells were examined and photographed by bright field and Nomarski interference (DIC) microscopy using an Orthoplan Leitz microscope (Leitz, Germany) at ×300-1.250 magnifications with a digital camera Canon S45 and True Chrome HDII Screen, as well as with Polyvar microscope (Reichert-Jung, Austria).
For the investigation of macronuclear bacteria and cytoplasmic algal symbionts, the material was fixed for electron microscope according to a protocol routinely used [33].

Experimental infections
Experimental bacterial infection was carried out using a homogenate prepared from infected cells according to Preer [34]. In particular, P. chlorelligerum cells were infected mixing equal volumes of a dense cell culture and of homogenate in a 3-mL depression slide, and maintained at 18˚C. To check the infection development, at least ten living cells were observed 2 hours, 24 hours, 48 hours, 4 days, 7 days, and 28 days after procedure. Then infection status was controlled every 10-15 days.

DNA extraction and molecular characterization
About 50 starved Paramecium cells were washed several times in sterile distilled water and fixed in 70% ethanol. Total genomic DNA was extracted using the NucleoSpin™ Plant II DNA extraction kit (Macherey-Nagel, Germany), following the protocol for mycelium.
All polymerase chain reactions (PCRs) were performed in a C1000TM Thermal Cycler (BioRad, Hercules, CA) with the TaKaRa ExTaq (TaKaRa Bio Inc., Otsu, Japan). Each PCR consisted of 35 cycles divided as follows: a preliminary denaturation step at 94˚C for 3 minutes, then denaturation at 94˚C for 30 seconds, annealing at variable temperature (between 50 and 55˚C) for 30 seconds and elongation at 72˚C for 90 seconds, and a final elongation step at 72˚C for 5 minutes. In every PCR experiment, a negative control without DNA was always included. After evaluation by electrophoresis on 1% agarose gel, PCR products were purified with EuroGold CyclePure kit (EuroClone1, Milan, Italy) and directly sequenced using appropriate primers.
The complete ITS1-5.8S-ITS2 sequence of the ciliate was obtained following Boscaro et al. [41] and the host cytochrome c oxidase subunit I (COI) gene was amplified and sequenced as described by Strüder-Kypke et al. [42].

Probe design and fluorescence in situ hybridization
The obtained 16S rRNA gene sequences were used to design a specific probe HoloParv_645 5'-CCGTACTCTAGTCTCCC-3' (Tm = 55.2˚C). The probe was synthesized and labeled with Cy3 by Eurofins GMBH (Ebersberg, Germany), and its specificity was tested in silico both on Ribosomal Database Project (RDP [45]) and on TestProbe tool 3.0 (SILVA rRNA database project [46]) allowing 0 and 1 mismatches ( Table 1). The probe sequence was deposited in Probebase [47].
Fluorescence in situ hybridization (FISH) experiments were carried out using different formamide concentrations in the hybridization buffer (0, 10, 20 and 30% v/v), in order to test the different stringency levels. The signal intensity was the best at 0% formamide. All FISH experiments were then performed following the protocol by Manz et al. [48] and negative controls, namely experiments without the use of probes, were always included. An additional probe, EUB338 5'-GCTGCCTCCCGTAGGAGT-3' [49], was used as a control to exclude the presence of other intracellular bacteria and to detect unambiguously the presence of the endosymbionts together with the specific probe inside the host cells.

Sequences comparison and phylogenetic analysis
The three molecular markers (18S rDNA, COI, ITS1-5.8S-ITS2) employed to characterize P. chlorelligerum were compared with sequences already present in database using BLASTN. In order to verify previous results and clarify uncertain sequences, cells from the German population (kindly provided by Dr. S. Krenek) were used to resequence all three molecular markers. All phylogenetic analyses were performed both with Maximum Likelihood (ML) and Bayesian Inference (BI) methods. In case of ML, trees were calculated with 1000 bootstrap pseudoreplicates using PHYML software 3.0 [50] from ARB package [51]. BI was carried out with MrBayes 3.2 [52] running three runs, with one cold and three heated Monte Carlo Markov chains each, for 1000000 generations with a burn-in of 25%. After verifying that the average standard deviation of the split frequencies had reached a value 0.01 or below, the runs were stopped.
The optimal model of nucleotide substitution was chosen according to the Akaike Information Criterion (AIC) calculated with jModelTest [53], which selected the GTR + I + G model (using a discrete, four-categories gamma function) in all.
The 18S rRNA gene sequences were firstly aligned against more than 3500 ciliates sequences from SILVA 119.1 database [46] with ARB software package, then the alignment was optimized manually according to 18S rDNA secondary structure. Phylogenetic analysis was inferred using 34 sequences of ciliates belonging to the order Peniculida and the final alignment consisted of 1646 nucleotides (S1 Alignment).
The ITS1-5.8S-ITS2 secondary structures were predicted using mFOLD version 3.2 [54], and the presence of compensating base changes (CBCs) was examined as shown in Coleman [55]. After all, in total 29 sequences were aligned with MUSCLE 3.8.31 [56], and phylogeny was inferred with a final alignment of 999 characters (S2 Alignment).
The obtained 16S rRNA gene sequences of the novel bacterial endosymbionts were firstly aligned against those present in the latest version of SILVA database (release 123) using the ARB software package. Afterwards, the alignment was manually edited in order to optimize base-pairing in the predicted rRNA stem regions. The aligned sequences were reduced at both ends to the length of the shortest one, and gaps were treated as a missing character.
Phylogenetic analysis was performed using 44 sequences from the order Rickettsiales and 6 alphaproteobacterial sequences as out-group. The final alignment contained 1356 nucleotide columns, which were used to infer phylogeny (S3 Alignment).

Diversity of ciliates in Peterhof ditch
Ciliates were repeatedly collected in a freshwater ditch (Peterhof, English park) during all seasons for one year. Throughout sampling, the temperature of water varied from +1˚C (January) to +23˚C (August), and pH ranged from 6.6 to 7.15 (measured in spring, summer, and autumn). In the same samples, taken from the water column and from the upper layer of bottom, numerous different ciliates with algal cytoplasmic symbionts were detected: Frontonia vernalis, Frontonia leucas, Stentor polymorphus, Climacostomum virens, Spirostomum semivirescens, Loxodes rostrum, Pelagotrix plancticola, Microthorax viridis, Spathidium chlorelligerum, Prorodon niveus, Euplotes daidaleos, Stichotricha secunda, Dileptus sp., and Paramecium bursaria. Four other morphospecies of Paramecium genus were found in the water body: P. caudatum, P. aurelia, P. multimicronucleatum, and P. putrinum. Bacteria belonging to Holospora genus, namely H. acuminata and H. obtusa, were retrieved in cells of P. bursaria and P. caudatum respectively.
The number of P. chlorelligerum cells present in different samples, which were collected in a distance of about 30-50 cm from each other, might vary significantly, but these ciliates were always present. P. chlorelligerum was also omnipresent through all the year seasons. In December-February all surface of the ditch was frozen, hence in January 2015 the surface was covered with ice 10 cm thick ( Fig 1B). Despite this, the general composition of ciliate community under the ice was relatively similar to the one observed during the other seasons, but with much less abundance.
In absence of a special cultivation conditions (see Materials and Methods) P. chlorelligerum usually disappeared in few days after collection from the original samples, instead, for instance, Spirostomum semivirescens or Frontonia vernalis could survive in samples at least several weeks, while Loxodes rostrum was present for more than a month.

Biology of Paramecium chlorelligerum
Ciliate morphology and ultrastructure. The cell morphology of P. chlorelligerum from the newly discovered population matches quite well with the description presented in two previous publications [11,12]. Cell shape was mostly ellipsoidal, and the average size of living specimen was 113 x 45 μm; while impregnated ciliates were a bit smaller 106 x 44 μm (Figs 2A  and 3). The main morphological features which allowed to discriminate the species were: 1two contractile vacuoles (CV) of "vesicle" type (without collecting canals) terminated in cortex by single pore (CVP) each in 85% (n = 20) of population, while a small fraction of cells presented two pores in one of the two CV (Figs 2D, 2E, 3B and 3C); 2-caudal cilia (about 10), which were twice longer than the rest of somatic ciliature (about 10 μm (Fig 2G)); 3-relatively small (4.5-6.0 x 2.0-2.5μm) single spindle-shaped micronucleus of "compact" type with a hyaline "achromatic cap" (only once 2 micronuclei were detected in a cell) (Figs 2B, 2C, 3E, 3G and 3I). Just after cell division and in case of bimicronuclear cell, the form of the micronucleus was different from spindle-shape ( Fig 3H). In the cells of investigated clones, oral opening was located slightly in front of cell equator line; oral ciliature manifested sometimes irregularity (Fig 3A and 3E) of quadrulus (4 or 5 rows), but the classical composition (4 rows) was the most common.
The cytoplasm of P. chlorelligerum always contained a number (several hundreds) of green algae, which reminded at a first glance Chlorella sp. Slightly ellipsoidal algae cells were 5-7 x 3-6 μm in size. These endosymbionts were located in perialgal vacuoles, and were characterized by one cap-like chloroplast without distinctive pyrenoid inside. A single nucleus was generally situated close to chloroplast, but in some algae cells the nucleus was doubled. When the nucleus was stained by Feulgen, it looked as a vesicle structure (because of big nucleolus) localized in periphery of the cell (Figs 2A-2C, 2F, 3J and 4).
Endosymbiotic algae identification and cultivation attempts. Endosymbiotic algae were identified molecularly by PCR on fixed cells of P. chlorelligerum. A partial (1223 bp long, GenBank KX669637) 18S rRNA gene sequence was obtained, and its identity was 99.3% with Meyerella planctonica isolated from environmental samples [57,58] (Table 2). When this sequence was compared to the one of German P. chlorelligerum endosymbiotic algae (Gen-Bank JX010741), identity was 98.9%, thus confirming the presence of Meyerella sp. as endosymbiont of this green ciliate.
All algae cultures obtained directly from the initially grown on BBM plates were clearly not morphologically resembling the algae inhabiting the P. chlorelligerum cytoplasm, as they had distinct pyrenoid (data not shown). By partial 18S rDNA sequencing they were identified as Chlorella sorokiniana. Algae present in the single culture started from cyst-like cells (see Material and Methods) were morphologically similar to symbiotic algae observed in P. chlorelligerum cytoplasmic symbionts (Fig 5A). These opaque green algae lacked of pyrenoid. Several life stages were observed (Fig 5A): young vegetative cells, mature cells actively producing and secreting lipids (Fig 5E), a typical characteristic for Meyerella growing in restrictive conditions [59], and cyst-like cells. These algae multiplied either by dividing into four cells as tetrads or chains (Fig 5B and 5C), or producing autospores (Fig 5D), another distinctive feature of free-living M. planctonica [58]. Mature cells formed double-layered shells, thus, probably, preparing for encystment. Formation of cyst-like cells is typical for free-living algae which have to survive in unfavorable conditions, such as desiccation [60]. Such cyst-like cells are also known as hypnospores [61] and they are characterized by thick multilayered cell walls, and by accumulation of lipids, starch, and phenolic substances [60]. Unfortunately, the culture of these algae suddenly stopped growing in liquid BBM within a week, and all cells died before reaching sufficient amount for DNA extraction. Attempt to transfer them back from liquid medium to BBM agar plates also failed.
Molecular characterization, sequences comparison and phylogenetic analysis of Paramecium chlorelligerum. The almost complete 18S rRNA gene sequences (1716 bp) were obtained by direct sequencing after semi-nested amplification from the two strains (HSG1-11 GenBank KX669629, and HSG2-12 GenBank KX669630), and they were identical. These "Russian" P. chlorelligerum sequences were 99.8% identical to representatives of the "German" population [1,12]. In those sequences, respectively three (position 1029, 1298 and 1310) and two (positions 1298 and 1310) mismatches were present. The two 18S rDNA sequences produced from German population [1,12] were differing in a nucleotide, a W in [12]. To resolve this ambiguity, we resequenced a sample derived from the German population obtaining the same result as [1]. Thus, we can confidently assume that 18S rDNA from German population is differing from Russian one in two nucleotides. Intriguingly these two mutations are located on opposite site of the stem with the second mutation re-establishing the correct secondary structure (they represent a compensatory base changes (CBC)); the two mutations are consequently not evolutionary independent and possibly highlight a highly variable stem in P. chlorelligerum.
The two Russian strains had an identical sequence of COI gene (760 bp, HSG1-11 GenBank KX669632, HSG2-12 GenBank KX669633), which was 100% homologous to COI gene of the ciliates from German population [1]. On the contrary, the other COI sequence attributed in GenBank (KF110720) to P. chlorelligerum appeared to be completely different and probably is the result of organism misidentification. Indeed, in recent COI phylogenetic reconstruction [1], this sequence is present in the outgroup composed by different ciliate species. ITS1-5.8S-ITS2 sequences were 1080 bp long and 100% identical among all P. chlorelligerum characterized by us (HSG1-11 GenBank KX669629, HSG2-12 GenBank KX669630 and German population GenBank KX669631), and no CBC were found. Again, the GenBank sequence KF110708 attributed to P. chlorelligerum seems to be the result of a mixture of sequences belonging to Paramecium and some other ciliate, probably prostomatean (e.g. Prorodon sp.), thus showing a misidentification of ciliate cells or contamination during DNA sample preparation.
After the model GTR + I + G was chosen by jModelTest, phylogeny for 18S rDNA and ITS1-5.8S-ITS2 were inferred. As previously shown [1,12], the 18S rRNA gene phylogeny   Nuclear Symbiont Holospora Discovered in Rare Paramecium Species confirmed the position of P. chlorelligerum as clustering with the subgenera Cypriostomum and Paramecium (Fig 6). These results were confirmed also by ITS1-5.8S-ITS2 gene phylogeny, in which P. chlorelligerum was slightly associated as sister group of all species of the subgenus Cypriostomum as shown in Fig 7. The faint differences between German and Russian P. chlorelligerum 18S rDNA sequences are visible also on the tree (Fig 6), which is in agreement with the last 18S rDNA phylogenetic tree published on Paramecium genus [1]. On the contrary, our ITS1-5.8S-ITS2 phylogenetic tree (Fig 7) presents a different situation. First of all, we provided new and correct sequences of P. chlorelligerum and its position is completely different from previous studies. Indeed, this green ciliate clusters as sister group with Cypriostomum subgenus, which is in agreement with other phylogenetic reconstructions based on other molecular markers (e.g. 18S rDNA, COI) [1]. On the other hand, ITS1-5.8S-ITS2 phylogeny shows that P. polycarium does not cluster with the other members of Cypriostomum group, thus enlightening a necessity of revising this subgenus from a morphological and molecular point of view; and some positions within Paramecium aurelia complex are not solved due to a probably lack of sequences.

Bacterial morphology and its life cycle
In the sampled P. chlorelligerum populations, 5-7% of cells manifested macronuclear infection by bacteria morphologically resembling Holospora sp. After subcloning from two samples in 2014 and 2015, two infected strains were established (HSG1-11, HSG2-12), and cells of these were investigated from morphological and molecular points of view. Two types of straight non-motile bacteria with different size and structure were observed in the infected macronucleus. There were two morphological forms, which looked like infectious (IF) and reproductive (RF) forms of classical Holospora-like bacteria [13]. RFs were short, slightly ovoid, and their length was 1-3 μm; IFs were spindle-shaped and 3-5 μm long; diameter of both forms of the bacterium was estimated as 0.6-0.7 μm (Figs 8-10). The RF manifested typical homogenous prokaryotic cytoplasm (Figs 9, 10A, 10B and 10D). The bacterium IF with distinctive tapered ends displayed differentiated cytoplasmic and periplasmic parts and a recognition tip-like structure (Fig 10A and 10B). The recognition tip contained less osmiophilic material and reminded those of other Holospora species, but periplasmic region always showed two different parts, a larger and denser one, and another one more transparent. The latter was located close to recognition tip structure and looked as a minor part of periplasm (Figs 9, 10A and 10B). Sometimes the same cell could manifest two strips of dense periplasm, but the transparent part of it was always single and situated close to the recognition tip (Fig 10A and 10B). Another difference between IF and RF was the composition of its surface membrane. Indeed, in majority of IFs their surface was decorated with fine fibrous material (Fig 10B-10D).
The microorganism had a typical Holospora-like life cycle with alternation of IFs and RFs [13], and it could be completed (during experimental infection) only in 6 weeks, while the life cycle of Holospora sp. takes normally just 7-12 days [13]. In fact, the first IFs were detected in experimentally infected culture after 40 days. Before the nucleoplasm was populated exclusively by RFs. Sometimes, infected macronucleus might be overpopulated by symbionts with several hundred of IFs (Fig 8A-8C), becoming distinctively roundish and larger in size than uninfected ovoid nucleus (Figs 2B, 2C and 3F). During cultivation, growth of infected cells was as slow as a growth rate of uninfected ones, only 1-2 divisions per week. The fate of infected cell lines was different. HSG1-11 manifested 100% infection, but after 6 months died. Instead, cells of HSG2-12 showed 100% of infection and then almost lost infection after 2 months of cultivation, but after 6 other months the strain recovered bacterium presence up to 10-90% (subpopulations of the same stock in different Petri dishes). It was possible to check only few dividers, but the connecting piece was never observed during division of infected macronucleus. The connecting piece, an equatorial part of the dividing host nucleus where the majority of IFs are gathered, is a feature manifested by all "classical" holosporas [62]. The daughter cells of infected P. chlorelligerum always inherited both bacterial forms in their macronuclei, and sometimes IFs could even produce a kind of clusters in the karyoplasm. The process of IFs release from the macronucleus into the ciliate cytoplasm was not investigated, but some bacteria were detected in cytoplasm during electron microscopy investigation.

Molecular characterization and phylogenetic analysis of the bacterial endosymbiont
Almost complete 16S rRNA gene sequence (1404 bp) was obtained by direct sequencing from the two strains (HSG1-11 GenBank KX669635, and HSG2-12 GenBank KX669636) harboring endosymbionts. These were identical between each other and presented an identity value ranging from 96.5% to 98.1 with other Holospora species (Table 3).
A species-specific probe was designed and FISH experiments were carried out in order to detect the presence of the bacterial endosymbiont (Fig 11), as expected by "full-cycle rRNA approach" [49]. According to the 16S rRNA gene secondary structure and diversity of the Holospora genus sequences, regions suitable for species-specific probe design were limited. Indeed, the HoloParv_645 probe was designed in the only region, which resulted both accessible and also sufficiently variable. According to RDP results, HoloParv_645 probe may recognize, in addition to "Ca. Holospora parva", a handful of sequences of uncultivable organisms. When a single mismatch was allowed, the number of hits significantly increased including also H. obtusa. In order to verify species-specificity of newly designed probe in condition of one mismatch, hybridization experiments with probe HoloParv_645 and Paramecium caudatum bearing H. obtusa were performed in 0, 15, 30% formamide conditions. The probe was never binding to H. obtusa at any formamide concentration (data not shown). All cells screened showed the presence of bacteria in the host macronucleus (Fig 11). Moreover, signals from both probes EUB338 and HoloParv_645 overlapped, thus excluding the presence of other bacterial endosymbionts.
ML and BI trees were inferred (Fig 12) after the GTR+I+G substitution model was chosen by jModelTest. The ML and BI trees showed good statistical values of bootstrap and posterior probability in the majority of nodes. As shown in the Fig 12, the monophyly of the families within the order Rickettsiales was confirmed, and their evolutionary relations were in agreement with recent literature [63][64][65][66][67], namely Rickettsiaceae as sister group of Anaplasmataceae-"Candidatus Midichloriaceae" clade (99% bootstrap value for ML and 1.00 posterior probability for BI). All members of the Holospora genus cluster together forming a monophyletic

Discussion
The water body in Peterhof, Russia, as a diversity hot-spot for green ciliated protists The newly found freshwater locality manifested very unusual high biodiversity of green ciliates, at least fifteen species. In spite of 40 years practices in ciliate sampling in Russia, as well as in different parts of the World [68], we never found so high diversity of such ciliates in the same place. Some of mentioned species (e.g. Loxodes rostrum, Pelagotrix plancticola, Microthorax viridis, Spathidium chlorelligerum, Prorodon niveus, Frontonia vernalis and Stichotricha  Nuclear Symbiont Holospora Discovered in Rare Paramecium Species secunda), were never previously recorded in Russia. In literature, the case of six green ciliate species present in the same pond was treated as "spectacular abundance" [69] and reported number of ciliate species associated with endosymbiotic algae could be from one to eight in the same water body [70][71][72]. Indeed, the finding of so many green ciliate species demonstrates that Peterhof ditch can be considered as a hot-spot for ciliates inhabited by green algae. To our best knowledge, the diversity of ciliates with cytoplasmic green algae symbionts was higher only in Simmelried moorland ponds, nearby the town of Konstanz, where 30 species were found [73]. These two localities hosting P. chlorelligerum shared some general ecological features. On one hand, ground flora is rather different between Simmelried and Peterhof biotopes. Moss (Sphagnum fallax) which is a dominant plant in Simmelried, does not grow in proximity to the Peterhof ditch. Also edificatory trees in Simmelried were Pinus sylvestris, Pinus mungo, Picea abies, and Betula sp. with, probably, Alnus sp. (according to pictures from [73]), while in Peterhof the most common trees close to the ditch were Betula pendula, Alnus glutinosa, Salix sp., and Quercus robur (minority). However, at the same time both areas are humid and a bit marshy, and the waterbodies possess similar set of water vegetation: Lemna sp., Carex sp., Phragmites australis, Typha sp. and Caricetum sp. (minority in Peterhof). Some water plants like Cladium marisci absent in Peterhof just indicate more south localization of Simmelried. This similarity between Simmelried and Peterhof water bodies gives ground to doubt another finding of P. chlorelligerum in Romania, Herăstrău Lake, Bucharest [74] as that lake is big, rather clean and does not remind at all, according to the water vegetation set, the first two water bodies. On the contrary, the first description made by A. Kahl [11], even if short, indicates mire environment: "I know it only from the mud of a mire puddle" (page 830). So, we can assume that rather particular environmental conditions are needed for P. chlorelligerum presence. Probably specific quantities of microelements and substances present in water and soil of these marshy biotopes produce favorable conditions for specific water vegetation and stimulate the growth of unicellular algae prone for symbiosis with protists [75]. Such mire environments are not yet so much studied; if our considerations are correct, we can predict to find P. chlorelligerum and diverse green ciliates in some similar biotopes.

Biology of Paramecium chlorelligerum
General morphology. Morphological characteristics of P. chlorelligerum from Russia match quite well with original description and redescription [11,12]. Some small variations occurred in cell size, position of oral aperture, composition of the quadrulus and the number of contractile vacuole pores. Small morphometric deviations could be explained, probably, by measurement technique; on the contrary, quadrulus structure and variable number of pores could be explained by different geographical origin of the two investigated populations. General structure of contractile vacuole and single spindle-shaped micronucleus of "compact" type with a hyaline "achromatic cap" are stable for representatives of P. chlorelligerum and allow to identify it unequivocally, and, in particular, to discriminate between P. chlorelligerum and another green Paramecium, P. bursaria. However, it is worth mentioning after the German colleagues that P. chlorelligerum might have been overlooked many times in natural samples, as at a glance check it is very easy to confuse with P. bursaria or even with green prostomateans [12]. Thus, the frequency of P. chlorelligerum in nature might be significantly underestimated.
Sequence comparison, molecular variability and phylogeny of Paramecium chlorelligerum. The retrieval of another P. chlorelligerum population in a new location gave the possibility to investigate in more detail this rare green ciliate both from molecular and phylogenetic points of view. For this purpose, three molecular markers (18S rDNA, COI, ITS1-5.8S-ITS2) with different rate of evolution were employed (Figs 6 and 7).
The novel Russian locality hosted a very stable and homogeneous population from genetic point of view, and, thus, probably, isolated population as no difference was observed in the diverse molecular markers used to characterize strains isolated in different years. Our recharacterization of the German population confirmed previous data by Krenek et al. [1] and provided a novel and correct ITS1-5.8S-ITS2 sequence for this ciliate. Our phylogenetic reconstructions supported the previous ones inferred on 18S rDNA and COI [1,12] and provided a novel phylogenetic tree based on ITS1-5.8S-ITS2, which confirmed the position of this rare ciliate species as a sister group of Cypriostomum subgenus, although with a low support. Unfortunately, it was not possible to infer phylogeny on the complete rDNA locus, as for most of Paramecium species present in databases some genes are missing. The comparison of P. chlorelligerum sequences revealed that this species displays a very low genetic variability, even in geographically separated populations. Indeed, only one out of the three molecular markers showed differences between the German and the Russian P. chlorelligerum populations. COI and ITS1-5.8S-ITS2, molecular markers usually employed to investigate intraspecific diversity, did not show any difference in the populations as they were 100% identical. Surprisingly, evolutionary more conserved 18S rDNA presented a couple of mismatches (CBC), which could be suitable to design a population-specific FISH probe, as an alternative and faster method to discriminate strains within the same morphospecies [76,77]. The same phenomenon, but less emphasized, was found also in other members of the genus Paramecium [46,78], where intraspecific variability was higher in 18S rDNA sequence than in the usually less conserved markers.
Nevertheless, the increase of sampling and studies on marshy environments could lead to the retrieval of other P. chlorelligerum populations, thus allowing a deeper understanding of the real genetic diversity of this rare ciliate. Furthermore, as several sequences in GenBank have been incorrectly assigned to P. chlorelligerum (KF110720, KF110708), now it becomes more and more important to combine molecular tools with morphological analysis of the studied microorganism to avoid further mistakes.
Characterization and ecology of endosymbiotic algae. The characterization of Russian P. chlorelligerum algal endosymbionts was performed both morphologically and molecularly, and they were identified as members of the Meyerella genus. This algal genus splits off the Chlorella group basally [57]. The main discriminating morphological feature of this genus is absence of pyrenoid in chloroplasts. This feature characterizes only three groups of green microalgae, namely Kalinella, Leptochlorella and Meyerella [79,80]. The presence of a single chloroplast was revealed in symbiotic algae of Russian P. chlorelligerum (Fig 4), thus resembling M. planktonica description [58]. Still, ultrastucture of M. planktonica is not identical with symbionts from P. chlorelligerum. Some morphological differences were revealed also between Meyerella sp. from Russian paramecia and other members of the Meyerella genus. At the moment two free-living species of Meyerella have been described, namely M. planktonica from the Itaska lake in Minnesota [58], and Meyerella sp. from the soil crust in Utah desert area [57]; also microalgae inhabiting German P. chlorelligerum have been identified as Meyerella sp. [12]. In particular, two chromatophores, leaving blank one polar region, were detected in symbiotic algae of German P. chlorelligerum [12], while Meyerella sp. isolated from the soil crusts was characterized by lobed or even fragmented chloroplasts [57]. Significant diversity within the Meyerella genus was observed also from molecular point of view. Indeed, two free-living Meyerella sp. were molecularly characterized [57,58] and they displayed 98.3% similarity of 18S rDNA. At the same time, two 18S rDNA sequences of symbiotic Meyerella lineages were 98.9% identical. Comparison between free-living and symbiotic Meyerella sequences showed differences from 97.6 to 99.3% (Table 2). Thus, symbiotic Meyerella from Russian P. chlorelligerum was even more similar to M. planctonica, than two free-living species of Meyerella to each other.
Meyerella has never been reported as a symbiont of any other ciliate, except P. chlorelligerum. Until recently [79], all known symbioses between freshwater ciliates and algae generally have been presumed to involve representatives of Chlorella genus (Chlorophyta). Systematic revision of zoochlorellae [79], redescription of P. chlorelligerum [12], new data on green symbionts of Loxodes rostrum [71] and on Tetrahymena utriculariae [81], changed this view on algae inhabiting cytoplasm of ciliates. Comparative analysis of different Chlorella-like symbionts [79] revealed five phylogenetically distinct symbionts, namely Chlorella, Choricystis, Coccomyxa, Scenedesmus and Micractinium, which could be found in several ciliates. Recently representatives of a new green algae group called "Chlorb" (probably forming a new genus) were retrieved from four green ciliates isolated from Lake Biwa in Japan [82]. Now also the genus Meyerella should be added to the list of symbiosis-forming algae after its retrieval in the cytoplasm of both German [12] and Russian P. chlorelligerum. Two independent findings of these algae in geographically diverse populations of P. chlorelligerum allow to suggest that this symbiotic association is not occasional. The failure in stable cultivation of ex-symbiotic Meyerella on BBM, considered as a quality standard for maintenance of unicellular algae including exsymbiotic Chlorella and Micractinium [83], may be explained by the lack of some special conditions which these algae meet inside the ciliate cell. For example, these endosymbiotic algae may get a certain advantage living in a host with slow growth rate, as propagation enhanced by artificially favorable conditions could be too fast. Other possible explanations of unsuccessful cultivation attempts of symbiotic Meyerella may include low frequency of cystlike cells formation readily in Paramecium, occasional overgrowth of it by Chlorella if the latter are present in medium or engulfed but not digested by the ciliate, and even presence of a latent virus which activates and lyses the host cells when they start rapid growing outside of Paramecium [84,85].
All these data indicate that there is a need of revision of Meyerella species, as this genus includes both free-living and diverse symbiotic algae which possess also morphological differences. Furthermore, our findings raise a question on the origin and evolution of symbiosis between these algae and ciliates and suggest that P. chlorelligerum from the two populations independently acquired Meyerella from the environment.

Intracellular symbionts of Paramecium chlorelligerum are a novel Holospora species
The characterized bacterium shares many morphological and life cycle similarities with members of the Holospora genus. The pattern of features usually employed to identify bacteria of this genus was previously described for several species [86][87][88][89][90][91][92][93][94]. Indeed, alternation of IF and RF in their life cycle, high infectivity of the IF, nuclear specificity, and high selectivity for the host Paramecium species are considered to be distinctive features of Holospora species. The novel bacterium found in P. chlorelligerum macronucleus fits to all these criteria, and, thus, can be considered as a representative of the Holospora genus. This conclusion is also supported by ultrastructural data on morphology of IFs. The ultrastructure of IF slightly varies in the species of the Holospora genus. Among them, P. chlorelligerum symbiont is the smallest in size and according to shape and length reminds H. acuminata [87]. Finally, 16S rDNA sequence similarity with several Holospora species of circa 97% and phylogenetic analysis has shown that novel bacterial symbiont of P. chlorelligerum clusters with H. obtusa, H. unduluta, and H. elegans, thus unequivocally including it in the Holospora genus (Fig 12).
This new Holospora species proved to be infectious, as experiments in stock HSG2-12 showed that reinfection is a possible event in the P. chlorelligerum population. However, outcome of experimental infection was relatively low, and development of infection took much longer than in all other Holospora species. Low rate of this process is probably connected with physiology of the host ciliate, which is characterized by unusually slow reproduction and, consequently, may put certain limitations on the readiness of experimental (or native) infection. It is also possible that laboratory conditions are insufficient to allow faster growth of the host ciliate and, consequently, the symbiotic bacteria slow down their reproduction.
All members of the Holospora genus recover maximal support and form a monophyletic group (Fig 12). Up to now, the validity of the Holosporaceae family within the order Rickettsiales is under discussion. As shown in some recent literature based on 16S rDNA and genome phylogenies [23,63,64], the family Holosporaceae is placed as basal to the rest of Rickettsiales and within this order. Other studies utilizing different and less widespread molecular markers, such as 23S rDNA and protein-coding genes, placed Holosporaceae outside of the order Rickettsiales and suggested to rank it as a separate order [95][96][97]; a recent study formally elevated Holosporaceae to the order Holosporales providing an overall revision of families belonging to this new order [98].
However, our phylogenetic reconstruction of Holosporaceae and Holosporales is in accordance with previous studies [23,63,64,98] and its inner relations are confirmed. Indeed, all H. obtusa sequences cluster together with H. undulata as sister group, and the earlier divergence of Holospora species associated with P. bursaria is confirmed and they cluster together [23]. Within the group of P. caudatum "classical" holosporas, the novel species from P. chlorelligerum occupies basal position, thus forming a separate lineage. "Candidatus Gortzia infectiva", an Holospora-like bacterium, is confirmed as the closest relative of Holospora genus [23,99].
These phylogenetic findings make doubtful the importance of inducing the connecting piece as another feature considered to be distinctive between two groups of holosporas. Connecting piece is the result of IF concentration in a particular median body during division process of the infected host nucleus facilitating further release of symbionts to environment [100]. It was shown that the presence of the connecting piece is characteristic for classical Holospora species (i.e. those infecting P. caudatum and P. bursaria), while this structure is not induced by other Holospora-like bacteria [13,23,62]. If present, this event could be considered as a particular adaptation of bacteria evolved to accomplish more efficiently the complex life-cycle, exploiting host cell divison machinery [100,101]. This feature has not been found for two Holospora species infecting macronuclei of representatives of the Frontonia genus, namely F. salmastra and F. leucas [13,68]. Moreover, the lack of inducing the connecting piece was observed also in "Candidatus Gortzia infectiva" from P. jenningsi [23]. Interestingly, the novel macronuclear endosymbiont of P. chlorelligerum also does not induce the connecting piece reminding "Ca. Gortzia infectiva" rather than "classical" holosporas, though phylogenetically it is closer to the latter. Considering both morphological and molecular data, we can not claim anymore the presence of the connecting piece as an apomorphic feature for all Holospora species.
According to the current microbiological rules [102] for nomenclature, we can coin only the status of "Candidatus" for the novel macronuclear bacterium of P. chlorelligerum as a representative of non-cultivable microorganisms.
Holospora parva (N.L. adj. parva, a little) Rod-shaped gram-negative bacteria, with differentiated reproductive (RF) and infectious (IF) forms. RF 1.0-3.0×0.6-0.7 μm with homogeneous cytoplasm without visible inclusions. IF 3.0-5.0×0.6-0.7 μm straight rods, with distinctive tapered ends, extensive periplasmic space divided into two parts with different osmiophilic density, and a recognition tip. Does not induce the formation of a distinctive connecting piece during host cell division. Macronuclear endosymbiont of the free-living ciliate Paramecium chlorelligerum, identified in samples taken from permanent ditch in English park of Peterhof (St. Petersburg district, Russia). Capable of horizontal and vertical transmission in the host species. Basis of assignment: 16S rRNA gene sequence (GenBank KX669636) and positive matching with the 16S rRNA-targeting oligonucleotide probe HoloParv_645; morphological characters pattern as above. Uncultured thus far.