Euplotes aediculatus specimens were collected from the anaerobic side tank (0 ppt salinity) of the San Colombano activated sludge plant placed in Lastra a Signa (Florence, Italy), in November 2015. A single cell was isolated from the original population and was washed several times in sterile San Benedetto water (San Benedetto S. p. A. Italy) in order to minimize the presence of contaminating microorganisms. The cell was then accommodated in San Benedetto water and daily fed for 5 days with a drop of cerophyll medium (CM) inoculated with Raoultella planticola DSM3069 (Gammaproteobacteria, Enterobacteriaceae) to induce its exponential growth. CM is a commercially available mixture of pulverized wheat grasses. One liter of CM is composed by Cerophyll buffer (79,75 g/l Na2HPO4 and 27,1 g/l NaH2PO4), balanced salt solution (BSS = 20.8 g/l NaCl, 8 g/l MgSO4 × 7 H2O, 17 g/l MgCl2 × 6 H2O, 2,7 g/l CaCl2 × 2 H2O, and 4,6 g/l KCl), and stigmasterol (5.0 mg/ml in 99.8% ethanol). Food bacteria were stored at -20°C in 1 ml aliquots with 50% glycerin (v/v). Each aliquot was prepared in order to be sufficient to guarantee a good density of bacteria in 1 L of CM after 24 h of incubation at a temperature of 37°C. For the inoculation of CM, R. planticola aliquots were thawed and centrifuged for 5 min at 5,500 ×g. After discharging the supernatant, the pellet was washed and re-suspended with 1 ml of CM. The bacteria were centrifuged again for 3 min at 5,500 ×g to form a pellet. Lastly, they were resuspended in 1 ml of CM and then inoculated in 1 L of CM. Once E. aediculatus monoclonal strain EASCc1 was established, the culture was maintained in incubator at a temperature of 19 ± 1°C, with a 12:12 h irradiance of 200 μmol photons m^-2s^-1 (see Boscaro et al., 2013b and Castelli et al., 2018b for details). The obtained monoclonal culture was fed once a week with 1/4 bacterized CM of the total culture volume. In this culture conditions, ciliate doubling time was about 36 h. However, all ciliates used in the experiments were kept in starving conditions for 4 days before the experiment to ensure food digestion. The culture was constantly checked to ensure cell healthy conditions during the whole time of the experiments. E. aediculatus cell density was calculated under a stereo-microscope (Leica Wild M3C) using three technical replicates of 100 μl each; the cell number was estimated by mean, and calculated for 1 ml of culture (±150 cells/ml).

Living cells were immobilized on a slide for observation with the help of a coverslip without deforming them (Skovorodkin, 1990). Living cells were examined under differential interference contrast (DIC) using an Orthoplan Leitz microscope (Leitz, Germany) at ×300–1.250 magnifications, and photographed with a digital camera Canon S45 and True Chrome HDII. Ciliate identification was based on observations of morphological key-characters (i.e., cell shape and sizes, number and positions of cirri, length of buccal aperture, number and positions of dorsal kineties, type of argyrome on the dorsal surface) on an appropriate number of living individuals, and on the comparison of collected data with previous literature (Curds, 1975).

From a preliminary FISH screening, E. aediculatus strain EASCc1 was known to host in its cytoplasm two bacterial symbionts: P. necessarius (Betaproteobacteria), primary endosymbiont, and an unidentified Gammaproteobacterium, secondary endosymbiont, so for the molecular characterization of the latter, about 50 starved E. aediculatus strain EASCc1 cells were washed several times in sterile distilled water and fixed in 70% ethanol. Total genomic DNA extraction was performed using the NucleoSpin^TM Plant II DNA extraction kit (Macherey-Nagel, Germany), following the protocol for mycelium suggested by the company. The 16S rRNA gene of the bacterial symbiont was amplified in a C1000TM Thermal Cycler (BioRad, Hercules, CA, United States) with the TaKaRa ExTaq (TaKaRa Bio Inc., Otsu, Japan). The16S rRNA gene of the Gammaproteobacteria endosymbiont was amplified using the primers alfa F19a (5′-CCTGGCTCAGAACGAACG-3′ Vannini et al., 2004) and R1492 (5′-GGNWACCTTGTTACGACTT-3′, modified from Lane, 1991). Each amplification consisted of 35 cycles of: a preliminary denaturation step at 94°C for 3 min, denaturation at 94°C for 30 s, annealing at 50°C for 30 s and elongation at 72°C for 2 min, and a final elongation step at 72°C for 6 min. In each PCR experiment, a control without DNA was always included. After evaluation by electrophoresis on 1% agarose gel, PCR products were purified with EuroGold Cycle-Pure kit (EuroClone1, Milan, Italy) and directly sequenced using prokaryotic universal primers, 16S R515ND (5′-ACCGCGGCTGCTGGCAC-3′), 16S F785ND (5′-GGATTAGATACCCTGGTA-3′), and 16S F343 (5′-TACGGGAGGCAGCAG-3′) (Vannini et al., 2004).

Genomic DNA of Rheinheimera sp. strain EpRS3 was obtained through thermal lysis. Briefly, an isolated colony (about 10⁸ bacterial cells after 12 h growth, Lodish et al., 2000) was picked up from an overnight culture in solid medium (in this case, TSA) and dissolved in 50 μl of saline solution (0.9% NaCl); bacterial cells were then submitted to thermal lysis in a thermoblock at 95°C for 10 min, followed by cooling on ice for 5 min. The supernatant obtained after 3-min centrifugation at maximum speed was then placed in a sterile tube and used as template DNA for amplification. The amplification of the 16S rRNA gene was performed using the same amplification protocol described for the ciliate symbiont, with universal prokaryotic primers F7 (5′-AGRGTTYGATYMTGGCTCAG-3′Vannini et al., 2004) and R1492 (5′-GGNWACCTTGTTACGACTT-3′, modified from Lane, 1991). After evaluation by electrophoresis on 1% agarose gel, PCR products were purified with EuroGold Cycle-Pure kit (EuroClone, Milan, Italy) and directly sequenced using prokaryotic universal primers, 16S R515ND (5′-ACCGCGGCTGCTGGCAC-3′), 16S F785ND (5′-GGATTAGATACCCTGGTA-3′), and 16S F343 (5′-TACGGG AGGCAGCAG-3′) (Vannini et al., 2004). The full-length sequence of the 16S rRNA gene of Rheinheimera sp. strain EpRS3 was necessary to design the specific FISH probe as described below.

A specific probe was designed in silico to detect Rheinheimera sp. EpRS3 using the almost complete characterized 16S rRNA gene sequence (Accession Number MH540131, this work). The specificity of probe Rhein443 (5′-TACCAACCCTTCCTCCTC-3′, Tm = 56°C) was tested in silico both on Ribosomal Database Project (Cole et al., 2013) and TestProbe tool 3.0 (SILVA rRNA database project, Quast et al., 2013). The probe was synthesized and labeled with Cy3 by Eurofins GMBH (Ebersberg, Germany), and the probe sequence was deposited in ProbeBase (Greuter et al., 2015). FISH experiments were carried out according to the protocol by Manz et al. (1992) and different formamide concentrations (0, 10, 20, and 30% v/v) were tested to find the optimal stringency level for the newly designed probe.

The experimental procedure has been resumed in Figure 1. Rheinheimera sp. EpRS3 was grown overnight on tryptic soy broth medium (TSB) at 30°C in agitation. After growing, the liquid culture was immediately processed; five different treatments were prepared to treat five separate 1-ml ciliate sub-cultures of E. aediculatus strain EASCc1 consisting of 20–30 cells each (Supplementary Table S1). Previously, the ciliates were maintained in starving conditions for 4 days to let them complete digestion of their regular bacterial food (Raoultella planticola) and to minimize the presence of phagosomes with Gammaproteobacteria inside. All the sub-cultures received their specific treatment at the same time. These were the prepared treatments: (A) plain TSB medium (positive control); (B) absence of any treatment (negative control); (C) cell-free filtered supernatant from EpRS3 bacterial culture (“Supernatant” treatment); (D) liquid culture of EpRS3 bacterial cells with their growth TSB medium (“Tq” treatment); (E) EpRS3 cells without TSB growing medium, and resuspended in saline solution (0.9% NaCl in dH₂O) (“Pellet” treatment).

Once each ciliate sub-culture was provided with its treatment, eukaryotic cells were counted at defined time points under the stereomicroscope (×10). The first count was performed immediately after the beginning of each treatment. Then cell counts were performed every 60 min, until 24 h. When E. aediculatus cells had showed evident signs of suffering, such as a clear swim deceleration, a prolonged absence of cell movement with ciliature still beating and/or cell number decrease, they were considered close to death and were collected and fixed for TEM and FISH analyses.

In order to study the possible cell damage that ciliates were facing, cell were fixed immediately before their death. A comparison with negative and positive control specimens was performed to highlight Euplotes cell structures affected by different EpRS3 treatments. Any potentially observed treatment effect on the ciliate obligate endosymbiont P. necessarius and on the secondary endosymbiont “Ca. Nebulobacter yamunensis” was recorded as well.

A total of ten replicate experiments were conducted in different days in order to confirm the harmful effect of Rheinheimera sp. EpRS3 vs. the ciliate E. aediculatus EASCc1. Once the effect was confirmed, an experiment was conducted with the aim to obtain fixed cells (treated and control ciliates) for FISH and TEM analyses: three different culture replicates, monitored up to 24 h, were performed to test the effect of the five different treatments reported in Figure 1 on the growth and survival of the E. aediculatus EASCc1. Variation in ciliate cell number was calculated in each case and expressed as % of T0 cell number (100%).

Fluorescence in situ hybridization analysis was used both to assess the presence of the two different endosymbionts in E. aediculatus confirming preliminary observations using specific probes already available, and to verify the potential presence of Rheineimera sp. EpRS3 inside E. aediculatus EASCc1 cells which underwent culture experiments using the specific probe Rhein_443 (this work).

About 15 ciliate cells were randomly collected from both experimental and control cultures and were washed three times in sterile San Benedetto water and separately fixed on slides for FISH experiments as described by Szokoli et al. (2016).

Several different specific probes were used for FISH analysis (Table 1): Poly_862 (Vannini et al., 2005, labeled either with fluorescein or Cy3, green signal and red signal, respectively) to detect P. necessarius, NebProbe_203 (Boscaro et al., 2012, labeled with Cy3, red signal) to observe “Ca. Nebulobacter yamunensis,” and the newly designed specific probe Rhein443 (labeled with Cy3, red signal) to screen for Rheinheimera sp. EpRS3 presence. Two universal probes were employed as control to exclude the presence of other intracellular bacteria, namely Gamma_42 (Manz et al., 1992, labeled with Cy3, red signal) to detect Gammaproteobacteria, and Eub_338 (labeled with fluorescein or Cy3, green signal and red signal, respectively) targeting almost total Eubacterial diversity (Amann et al., 1990; Table 1).The presence of DAPI in the SlowFade^® Gold Antifade Reagent (Invitrogen) was used as additional control.

The pellet of Rheinheimera sp. EpRS3 from 2 ml, overnight liquid TSB culture (about 10¹⁰ cells/ml) was fixed in duplicate for FISH experiments using 2% PFA. 50 μl of the fixed bacterial pellet were dropped on the slide and hybridized according to Szokoli et al. (2016). FISH was performed using probe Rhein443 (labeled with Cy3, red signal) and Eub_338 (labeled with fluorescein, green signal) (Table 1) in order to verify the efficiency of the newly designed probe Rhein443 on Rheinheimera sp. EpRS3.

Images were obtained with a Leica DMR microscope, equipped with an HBO 50W/AC-L2 fluorescent lamp, a Leica DFC490 video camera, and a dedicated Software Leica IM1000 (v.1.0).

Transmission electron microscopy analysis was used both to assess the morphology of the two different endosymbionts in E. aediculatus, and to verify the potential presence of Rheineimera sp. EpRS3 inside E. aediculatus EASCc1 cells which underwent culture experiments. For TEM analysis, eukaryotic cells sampled from each of the protist sub-cultures of culture experiments were fixed in a 1:1 mixture of 2% OsO₄ in sterile distilled water and 2.5% glutaraldehyde in 0.2 M cacodylate buffer (pH 7.4); then cells were ethanol-dehydrated, transferred to 100% acetone, and embedded in an Epon araldite mixture (Modeo et al., 2013a).

After overnight growth in liquid TSB medium the pellet of Rheinheimera sp. EpRS3 (about 10¹⁰ bacterial cells/ml) was fixed for TEM analysis according to Nitla et al. (2018). Prokaryotic cells were fixed in 1.5-ml Eppendorf tubes and all the solutions (fixatives, ethanol, acetone, and Epon araldite embedding mixture) were directly added. At each step, the Eppendorf tube with prokaryotic cells was vortexed and centrifuged for 5 min at 5,500 × g; then, after discharging the supernatant, the next solution was added, and the pellet was re-suspended.

Ultrathin sections of both eukaryotic and prokaryotic cell TEM preparations were placed on copper grids and stained with uranyl acetate and lead citrate prior to observation with a JEOL 100S TEM.

The production of hydrogen peroxide by Rheinheimera sp. EpRS3 was assessed with a colorimetric assay on agar medium based on the Prussian blue-forming reaction (Saito et al., 2007), that was used as described in Chen et al. (2010). The agar medium allows the identification of hydrogen peroxide-producing bacterial strains by the appearance of dark blue halos around the colonies, due to the formation of a precipitate. The Prussian blue agar is made as follows (1 L): 1 g of FeCl₃⋅6H₂O dissolved in 50 ml of distilled water (solution A); 1 g of potassium hexacyanoferrate(III) K₃[Fe^III(CN)₆] dissolved in 50 ml of distilled water (solution B); preparation of 900 ml TSA medium; mixing solutions A and B, and pouring the mixture into TSA medium; the obtained medium, must be sterilized by autoclaving at 121°C for 20 min. Bacterial sample were streaked on Prussian blue TSA (PB-TSA), and incubated for 5 days at 30°C, until satisfactory growth; in parallel, different amounts of hydrogen peroxide were dropped on the same medium, and put under the same temperature conditions, in order to observe the blue precipitate formation.

Ciliate strain EASCc1 was confirmed in morphological inspections as Euplotes aediculatus as perfectly matching the description of the species (Curds, 1975); additionally, molecular analysis based on 18S rRNA gene sequencing confirmed the species assignation by morphological identification (C. Sigona, pers. commun.).

The almost complete 16S rRNA gene sequence of the Gammaproteobacteria endosymbiont was achieved by direct sequencing after PCR on E. aediculatus cells. The sequence obtained was 1409 bp long (Acc. Number MH608339, this work) and was 100% identical with “Ca. Nebulobacter yamunensis” (GenBank HE794998) already present in NCBI database.

The Rheinheimera sp. EpRS3 strain 16S rRNA gene sequence was obtained by direct sequencing from the bacterial strain DNA. The sequence was 1455 bp long (Acc. Number MH540131, this work) and 98.27% identical with Rheinheimera pacifica (GenBank NR114230).

The experimental procedure has been resumed in Figure 1. In all the replicate experiments, C+ and C- treatments (i.e., the treatment with plain bacteria TSB medium and the absence of any treatment, respectively) did not affect the overall health status of ciliates (Figure 2 and Supplementary Figure S2). However, several of the other treatments did affect ciliate survival (Figure 2). Supernatant treatment showed conflicting results depending on the different replicate: in two out of ten replicates, cell number dramatically decreased between 6 and 24 h from T0 (beginning of the treatment); moreover, cells stopped moving 2–4 h from T0. This behavior, with the same timing, was also recorded in Tq-treated ciliates. In a third replicate cell number remained constant over time, suggesting that there was not any effect on cell survival. Tq treatment impacted ciliates in all the ten replicates. After 2–4 h from T0, all ciliates stopped movement and showed cell suffering; in a single replicate ciliate number started a dramatic decreasing already after 1 h from T0. In all the replicates, the almost complete disappearance of treated ciliate sub-culture was recorded between 6 and 24 h.

Some Tq-treated cells were isolated soon after stopping movement and transferred to pure freshwater in order to roughly test whether the treatment effect was immediate and/or permanent. They apparently recovered well after some minutes, but it is not possible to assert if they were able to survive for a longer time and also to divide in the light of the TEM observation which reported on heavy Euplotes/endosymbiont cell damages (see later).

Pellet treatment did not affect protozoan survival in any of the ten replicate experiments, and the ciliate sub-culture survived over time: cells showed a healthy aspect, and, in some cases, they duplicated with an increment of number.

In the present study the probe Rhein_443 (5′-TACCAACCCTTCCTCCTC-3′, Tm = 56°C) was tested for its optimum formamide concentration by applying formamide from 0 to 30%, and the highest signal intensity was recorded at 0%. Its specificity was verified in silico both on Ribosomal Database Project (Cole et al., 2013) and on TestProbe tool 3.0 (SILVA rRNA database project Quast et al., 2013). The newly designed probe recognized in silico four non-target sequences, but its specificity was always verified during FISH experiments using the universal probe Gamma_42 targeting Gammaproteobacteria (Table 1).

In the cytoplasm of the ciliate, the endosymbiont P. necessarius was well visible in culture experiments under the light microscope due to its large size and high abundance (Supplementary Figures S1A,B). Preliminary FISH experiments on the ciliates with Poly_862, Gamma_42, and NebProb_203 probes highlighted the presence in the cytoplasm of two distinct endosymbiotic bacteria: the betaproteobacterium P. necessarius (Supplementary Figures S1C,D) and the gammaproteobacterium “Ca. Nebulobacter yamunensis” (Supplementary Figures S1E,F). FISH experiments on Rheinheimera sp. EpRS3 bacterial cells produced positive signals with both EUB_338_Fluo (Supplementary Figure S3A) and Rhein443_cy3 (Supplementary Figure S3B) probes. FISH experiments on C- and C+ ciliate subcultures hybridized with Poly_862, Gamma_42, and NebProb_203 probes, confirmed the presence of the two bacterial endosymbionts in ciliate cytoplasm. On the contrary, FISH experiments on the same two ciliate subcultures with Eub_338 and Rhein443 gave positive signals only with the first probe, revealing the absence of Rheinheimera sp. EpRS3 cells (C-, Figure 3A,B; C+, data not shown).

In Supernatant-treated sub-culture ciliates Rhein443 probe gave negative signal indicating the absence of Rheinheimera sp. EpRS3 (Figure 3C,D). In Pellet-treated ciliates, Rhein443 probe highlighted the presence of Rheinheimera sp. EpRS3 inside large food vacuoles (Figure 3E,F).

On ciliates of the Tq-treated subculture, Rhein443 probe revealed positive signals in the cytoplasm, indicating the presence of Rheinheimera sp. EpRS3 (Figure 3G,H). It is worth noting that these positive signals indicated the presence of Rheinheimera sp. EpRS3 both in small phagosomes and free in the cytoplasm, i.e., not included in phagosomes. Additionally, some Rheinheimera sp. EpRS3 bacteria were targeted outside the ciliate (Supplementary Figure S4), divergently from what observed in all other FISH experiments.

A preliminary TEM experiment was carried out on Rheinheimera sp. EpRS3 in order to obtain the first morphological-ultrastructural data on these bacterial cells (Figure 4). Cells generally appeared rod-shaped and encircled by a double membrane. They showed a linear outline and a highly electron-dense peripheral region; cell electron-density decreased in the inner structure, leading to a clear central core. Cells were fixed during their stationary growth phase and did not show the presence of different bacterial morphologies, which might occur with diverse life cycle stage. No bacterial flagellum was observed; however, the lack of flagella could also be due to TEM procedure. Bacterial cell dimensions ranged from ∼ 1.5 μm × 0.8 μm up to ∼ 4 μm × 0.8 μm.

TEM observation on E. aediculatus cells confirmed the results of FISH experiments. As expected, in C- cells (no treatment) the presence of two endosymbionts was observed, i.e., P. necessarius (maximum size: ∼ 6.5 μm × 0.4 μm; very abundant) (Figure 5A) and “Ca. Nebulobacter yamunensis” (size: ∼ 1.5 μm × 0.6 μm; rare) (Figure 5B,C). Both the endosymbionts showed a double membrane as typical of Gram-negative bacteria. P. necessarius cells were characterized by a slender aspect, an undulating external membrane, a non-homogeneous, generally electron dense cytoplasm, with some inner clearer regions where the presence of nucleoids sometimes is visible. “Ca. Nebulobacter yamunensis” usually appeared squatter and showed a homogeneous, clear inner cytoplasm, which becomes more electron dense in its periphery. A clear halo was sometimes visible around P. necessarius cells.

Positive control cells (C+, treatment with sterile TSB medium) showed no ultrastructural differences with C- cells (data not shown), as bacteria growth medium itself did not affect ciliate cells in any possible way. By TEM observation, E. aediculatus cells treated with Supernatant treatment (Rheinheimera sp. EpRS3 cell-free supernatant), showed a slightly vacuolized cytoplasm (Figure 6A,B), while mitochondria and plasma membrane appeared not affected (Figure 6C). Ciliates still hosted in the cytoplasm many P. necessarius bacterial endosymbionts, but several of them showed clear suffering signs, such as high cytoplasm vacuolation and degradation, being altered at different grades (Figure 6A,C,D). Representatives of “Ca. Nebulobacter yamunensis” generally showed a rather regular low abundance and appearance (Figure 6B).

Several very electrodense bodies enclosed in vacuoles were observed in the cytoplasm of both Supernatant (Figure 6D) and Tq (see later) treated cells; these are obliquely- or cross-sectioned, highly degraded P. necessarius cells, as their abundance and general appearance fit with those of longitudinally sectioned, degraded P. necessarius endosymbionts observed in the same TEM sections.

In Tq-treated ciliates, plasma membrane was still well conserved, while majority of P. necessarius endosymbionts were vacuolized and degraded as much as in Supernatant-treated cells. (Figure 7A). Autophagosomes were also present in the cytoplasm, in some cases including P. necessarius cells (Figure 7B). The endosymbiont “Ca. Nebulobacter yamunensis” was apparently not affected by the treatment (Figure 7C). Unlike what observed in Supernatant-treated cell, in Tq-treated ciliates mitochondria were very degraded as well; many of them were broken and, in some cases, they appeared to have undergone a kind of membrane disassembly (Figure 7A,B,D).

In Tq treated-cells, EpRS3 bacterial cells were observed both enclosed (Figure 7F) and not enclosed in phagosomes (i.e., free in the cytoplasm) (Figure 7G–I). They could be identified based on shape, dimensions, and appearance that distinguished them both from P. necessarius and “Ca. Nebulobacter yamunensis” endosymbionts. These results confirmed FISH experiment findings.

Ciliate cells treated with Pellet treatment showed cytoplasm and endosymbiont appearance similar to C- cells (Figure 8A,B). In Pellet-treated ciliates many Rheinheimera sp. EpRS3 cells (R) were observed inside phagosomes, in different digestion stages (Figure 8B).

Prussian blue agar medium assay revealed that Rheinheimera EpRS3 was able to grow on the modified PB-TSA medium. The growth after 5 days incubation did not reveal any hydrogen peroxide production by Rheinheimera EpRS3; this was evidenced by the absence of the blue halo around bacterial cells (Supplementary Figure S5, right side). This was also confirmed through comparison with the blue halos appeared around hydrogen peroxide dropped at different concentration on the same modified medium (Supplementary Figure S5, left side).