“Candidatus Gortzia shahrazadis”, a Novel Endosymbiont of Paramecium multimicronucleatum and a Revision of the Biogeographical Distribution of Holospora-Like Bacteria

Holospora spp. and “Candidatus Gortzia infectiva”, known as Holospora-like bacteria (HLB), are commonly found as nuclear endosymbionts of ciliates, especially the Paramecium genus. HLB are related by phylogenetic relationships, morphological features, and life-cycles, which involve two alternating morphotypes: reproductive and infectious forms (RF, IF). In this paper we describe a novel species belonging to the “Ca. Gortzia” genus, detected in P. multimicronucleatum, a ciliate for which infection by an HLB has not been reported, discovered in India. This novel endosymbiont shows unusual and surprising features with respect to other HLB, such as large variations in IF morphology and the occasional ability to reproduce in the host cytoplasm. We propose the name of “Candidatus Gortzia shahrazadis” for this novel HLB. Moreover, we report two additional species of HLB from Indian Paramecium populations: “Ca. Gortzia infectiva” (from P. jenningsi), and H. obtusa (from P. caudatum); the latter is the first record of Holospora from a tropical country. Although tropical, we retrieved H. obtusa at an elevation of 706 m corresponding to a moderate climate not unlike conditions where Holospora are normally found, suggesting the genus Holospora does exist in tropical countries, but restricted to higher elevations.

Holospora-like bacteria (HLB) can be spread by both horizontal and vertical transmission (reviewed in Fokin and Görtz, 2009). The IFs invade a new host cell via a phagocytotic pathway and, after acidosome fusion, are able to escape from digestive vacuoles and reach the nucleus of the ciliate  via movement mediated by the host cytoskeleton (Fokin et al., 2003;Sabaneyeva et al., 2005Sabaneyeva et al., , 2009. Inside the nucleus, IFs differentiate into reproductive forms, which divide by binary fission. During host cell division, RFs are shared between two daughter nuclei. In most Holospora species (i.e., "H. acuminata", "H. curviuscula", H. elegans, H. obtusa, "H. recta", and H. undulata) the majority of infectious forms are concentrated in a central part of the dividing nucleus-called the "connecting piece"-that will be released into the environment . The remaining Holospora species (i.e., "H. bacillata", H. caryophila, "H. curvata") and "Ca. Gortzia infectiva" do not show this particular localization (Fokin et al., 1996;Boscaro et al., 2013a;Fokin, 2015), and instead are released from the host by an inverted path of infection (Fokin and Sabaneyeva, 1997;Fokin, 2015): IFs are released singularly or in small groups from the nucleus into the cytoplasm, and later into the environment.
Holospora species have been widely recorded by protistologists, allowing researchers to produce a biogeography of these endosymbionts , although the infection rate in Paramecium populations is not always permanent and follows cyclic fluctuations Duncan et al., 2015).
Although the genus Paramecium is nearly cosmopolitan (Fokin, 1997(Fokin, , 2010Przyboś and Fokin, 2000), Holospora have only been isolated in cold to temperate areas, in the north of America, Asia and Europe (Hori and Fujishima, 2003;Fokin, 2004b;Fokin et al., 2006). However, this pattern is most likely biased by the lack of sampling in tropical countries, where sampling efforts have been limited (Fokin et al., 2004;Görtz, 2008;Fokin and Sera, 2014). An exception is "Ca. Gortzia infectiva", which has only been found in a sample from Thailand, (Boscaro et al., 2013a), showing that HLB can exist in tropical climates.
In this paper, we report the retrieval of HLB from another tropical area, the south of India. We found and described a novel species of HLB, which is the first reported from P. multimicronucleatum. Moreover, we found "Ca. Gortzia infectiva" in the MA of an Indian population of P. jenningsi and, quite surprisingly, also H. obtusa infecting the MA of P. caudatum from India. Our records represent the first report of "Ca. Gortzia infectiva" from India and the first finding of the Holospora genus in a tropical country (although from a higher elevation with a moderate climate), at the lowest latitude ever reported. Thus, our study provides new and important information about HLB distribution, reshaping HLB biogeography. Samples were grown at room temperature and fed Cerophyll medium inoculated with Raoultella planticola (see Boscaro et al., 2013a for details). Unfortunately the SH2 and BJ1 populations survived only a few weeks in laboratory conditions, but for P. multimicronucleatum population TP2 we were able to create a monoclonal culture, named TP2-2.

Characterization of Endosymbionts
Examination of living cells show that SH2, BJ1, and TP2/TP2-2 paramecia exhibit MA bacterial infections. Transmission electron microscopy (TEM) analysis was performed on the TP2-2 monoclonal strain following the protocol in Boscaro et al. (2013a).
We carried out cross-infection experiments using infected TP2-2 cells as donor, and the endosymbiont-free P. multimicronucleatum strain PC6 as recipient. Experimental infection was carried out by preparing a homogenate of infected cells according to Preer (1969). Recipient cells were infected by mixing equal volumes of cell culture with the donor cell homogenate in a 3 mL depression slide, and incubating at room temperature. In order to check the infection status a set of living cells (n = 10) was observed by DIC after 2, 24 and 48 h after mixture with the homogenate. We could test the trans-infection with only one species of Paramecium as receiver, due to the small number of cells in the TP2 culture.
Endosymbiont 16S sequences were obtained for all the populations using the Alphaproteobacteria universal primers 16S Alfa F19b, 16S R1522b (Table 1); sequencing was initially carried out using bacterial universal primers (16S F343 ND, 16S F785 ND, 16S R515 ND); subsequently we designed specific sequencing primers for each species (for details see Table 1). For the TP2 endosymbiont, two additional PCR reactions were performed with two different sets of primers: 16S Alfa F19b together with 16S Gortz R659 (annealing temperature was changed to 51 • C) and 16S F114HoloCaedi together with 16S R1488 Holo. The first PCR product was sequenced using R534 Gortzia, the second with F1008 Gortzia and 16S R1328HoloCaedi, and the sequencing results assembled into a single sequence. The 16S rRNA gene sequences were aligned using the ARB software package (Ludwig et al., 2004) and manually checked against more than 600,000 bacterial sequences  (Quast et al., 2013). For phylogenetic analyses 58 sequences were employed: together with the 3 new sequences, 48 selected sequences belonging to order Rickettsiales and 7 sequences belonging to class Alphaproteobacteria as outgroup (sequences not shown in the tree are listed in Table S1). The alignment was reduced in length, producing a 1632 character matrix. Maximum likelihood (ML) analyses (PHYML 5.3.2) (Guindon and Gascuel, 2003) and Bayesian inference (BI) analyses (MrBayes 3.2) (Ronquist et al., 2012) were performed, with the GTR +I +G substitution model, as indicated by AIC (Akaike's information criterion), calculated by jModelTest 2.2 (Darriba et al., 2012). ML analysis was applied with 1000 pseudoreplicates, while for BI analysis, three different Markov Chain Monte Carlo runs were employed, with one cold chain and three heated chains each, running for 500,000 generations. Fluorescence In Situ Hybridization (FISH) analyses was performed as described by Boscaro et al. (2013a). A first set of experiments was carried out with P. jenningsi BJ1 and P. multimicronucleatum TP2 using a probe designed for "Ca. Gortzia infectiva", GortProb659 (5 ′ -TTCCGTTTTCCTCTACCA-3 ′ ), (Boscaro et al., 2013a), labeled with a cyanine 3 (Cy3) fluorophore at the 5 ′ end, together with a Bacterial universal probe EUB338 (5 ′ -GCT GCCTCCCGTAGGAGT-3 ′ ) (Amann et al., 1990), labeled with fluorescein isothiocyanate (FITC) at the 5 ′ end. We then designed two new species-specific probes able to distinguish between the two endosymbionts from P. jenningsi and P. multimicronucleatum, since GortProb659 was found to label both. We designed GortzInf_1268 (5 ′ -TCCTGATTCGCTCA AGGTC-3 ′ ; FITC fluorophore in 5 ′ end), specific for "Ca. G. infectiva", and GortzSha_1266 (5 ′ -TTTTGATTTGCTCAAGGT CGC-3 ′ ; Cy3 fluorophore in 5 ′ end), specific for the new HLB from P. multimicronucleatum. Both probes were tested in silico on the RDP (ribosomal database project) (Cole et al., 2009) and SILVA database using TestProbe 3.0 (Quast et al., 2013), allowing 0 mismatches. It was not possible to experimentally test GortzInf_1268 because P. jenningsi BJ1 infected with "Ca. Gortzia infectiva" was lost. We tested GortzSha_1266 (Cy3) in competition with GortzInf_1268 (FITC) or EUB338 (FITC) on the P. multimicronucleatum endosymbiont.
Characterization of New HLB from P. multimicronucleatum TP2 Cells of P. multimicronucleatum TP2 exhibited 100% MA infection by rod shaped bacteria (Figure 1). A few weeks after isolation Paramecium cells manifested hyperinfected MAs, with dense clusters of bacteria (Figures 1A,B). The symbionts showed the typical morphology of HLB, with two different forms: the small RF (2.5-3.3 µm) and the larger, rod shaped IF, (6.9-10.7 µm) ( Figure 1E). The IF were characterized by slightly tapered ends and by darker/reflecting parts in their bodies. In addition, some transient forms (TF)-large but not containing reflecting material-were present ( Figure 1E). One year after isolation and subcloning, we repeated the analysis to check the infection status of TP2-2 and noticed that the distribution of the bacteria in the MA had changed: IF were clustered together in several small groups inside the nucleus (Figures 1F,G). IF and TF increased their length in comparison to previous observations, reaching 8.2-14.7 µm and 6.7-15.5 µm, respectively. In addition, the number of TF increased dramatically, whereas classical RF were fewer and/or less visible, compared to the first observations. We also observed IF dividing by binary fission (Figure 1K). Moreover, many IF showed an abnormal distribution of periplasmic parts inside the cell, forming a striated/dotted pattern (Figures 1J,K).
Quite surprisingly, we found these HLB not only in the MA but also evenly distributed in the cytoplasm, inside and outside food vacuoles (Figures 1H,I, 4G, 5D,F,G); using TEM and FISH techniques we detected large numbers of IFs and TFs, but also some RFs (Figures 4G, 5G). This phenomenon was observed during a random check 4 months after sampling and occurred in 10-20% of analyzed cells (n = 40). Due to the low number of cells in our culture, it was not possible to periodically check the status of infection and its progress, however, 1 year after sampling this unusual cytoplasmic distribution was present in 50% of the analyzed cells (n = 42).
During host cell division, we never observed the "connecting piece" (Figures 1C,D, 5A,B). Endosymbionts were spread among daughter cells within MA pieces which, in many cases, were not equally shared in terms of size, suggesting that the bacteria interfere in host cell division.
During our cross-infection assay, we observed the presence of HLB inside the MA of P. multimicronucleatum PC6 48 h after the beginning of the experiment, showing the initial RF morphology  observed in TP2 cells after a few weeks in cultivation ( Figure 1L). From that moment, HLB cells in PC6 strain started to grow and differentiate, but after some days host cells started to decline and die.
Ultrastructural analysis of TP2 P. multimicronucleatum after 1 year of cultivation (Figure 4) supported live investigations using DIC microscope during the same period (Figures 1F-K). We were able to recognize different ultrastructural features characterizing the life-cycle stages described above: RF, IF, and TF. RF appeared short and roundish with the typical, homogeneous and relatively electron-transparent prokaryotic cytoplasm. The IF, longer and rod-shaped with slightly tapered ends, displayed differentiated cytoplasmic and periplasmic parts and a recognition tip-like structure in the apical part of the body (Figures 4A-C). In some cases, recognition tip-like structures were present at both ends of the cell (Figure 4A). The periplasmic regions of IF exhibited a very dense, osmiophilic pattern, while the recognition tip contained less osmiophilic material. The periplasm constituted a rather thin layer, often unevenly distributed beneath the outer cellular membrane of IF, sometimes forming outgrowths or sublayer invaginations in the cytoplasm (Figure 4F). We could discriminate at least two different periplasmic regions on the bases of their density: a darker, electron-dense part and a lighter one (Figures 4B,C). The IF cytoplasm was more heterogeneous and denser than the RF cytoplasm. Extensions of cytoplasm could be seen protruding into the periplasm in some cases (Figures 4B,C,F). Sometimes the same cell would manifest two or more stripes of dense  periplasm, interposed with other cytoplasmic regions (probably forming the striped/dotted pattern previously observed via DIC microscopy). Another difference between IF and RF was their membrane composition: in the majority of IF the surface membrane was surrounded by fine fibrous material and, in some cases, manifests an additional membrane covering this fibrous-like layer (Figures 4C,F). TF appeared rod-shaped, with a size comparable to IF. The TF ultrastructure showed the presence of prokaryotic cytoplasm, without a distinctive periplasmic part (Figure 4E). The density of the TF cytoplasm resembles that observed in IF. Some TF manifest recognition tips at one or both ends of the body, as we observed in IF ( Figure 4A). All forms showed a constant diameter of 0.7-0.8 µm.
A 1398 bp long 16S rDNA sequence was obtained from this P. multimicronucleatum HLB and is available from ENA database under the accession number LT549002. From similarity matrix calculation we observed an identity value of 98.93% (15 mismatches) with "Ca. Gortzia infectiva" HE797907 and 90-91% with Holospora sequences ( Table 2). In phylogenetic trees, this HLB species appeared strongly supported by statistical values as a member of the "Ca. Gortzia" clade (100/1.00), being a sister species of "Ca. Gortzia infectiva" (Figure 6).
Preliminary FISH analyses suggested the presence of members of the "Ca. Gortzia" genus in P. multimicronucleatum TP2 (Figures 5A,B). In a double hybridization experiment with the newly designed species-specific probes GortzSha_1266 and GortzInf_1268, HLB from P. multimicronucleatum showed a positive signal only to GortzSha_1266 and not to the specific probe for "Ca. Gortzia infectiva" (Figures 5C-E). Moreover, FISH results were positive to GortzSha_1266 and EUB338 probes, with completely overlapping signals (Figures 5F,G), suggesting that the new HLB was the only bacterial species harbored by the TP2 cells. As detected from live observations, in many cases we found bacteria positive to the GortzSha_1266 probe not only in the MA but also in the cytoplasm (Figure 5G), with a homogeneous distribution. No broken MA were detected, which could have explained this unusual distribution.
These results confirmed both the presence of HLB corresponding to our 16S sequences inside the P. multimicronucleatum host and that it is possible to discriminate this bacterium from the already characterized "Ca. Gortzia infectiva", using species-specific probes. The sequence of probe GortzSha_1266 matches no sequences in the RDP database, while GortzInf_1268 matches 4 sequences belonging to "Ca. Gortzia infectiva" clones. Sequences of these two new probes were deposited into probeBase database (Greuter et al., 2016). "Ca. Gortzia infectiva" from India, Endosymbiont of P. jenningsi BJ1 Paramecium jenningsi BJ1 showed a "Ca. Gortzia infectiva"related infection in the MA, from images taken of live specimens and from FISH results (Figure 2). The 16S rDNA sequence from this endosymbiont, 1432 bp long, showed 99.93% of identity (1 mismatch) with "Ca. Gortzia infectiva" HE797907 from Thailand, with which it forms a monophyletic specieslevel group, supported by strong statistical values (100/0.97)  ( Figure 6). The identity values with the other two newly characterized endosymbiont sequences, TP2 and SH2, were 99.00 and 90.76%, respectively (14 mismatches and 115 mismatches plus 18 indels, respectively) ( Table 2). Our sequence is available from ENA database under the following accession number: LT549000, "Ca. Gortzia infectiva" BJ1.
Although this H. obtusa from India showed a slightly higher divergence in comparison with its previously known conspecifics, from the phylogenetic analysis these four sequences clearly form a monophyletic clade (Figure 6), supported by high values of bootstrap and posterior probability (100/1.00). Holospora obtusa was the sister group of the H. undulata (NZ_ARPM03000111) and H. elegans (BAUP01000039) clade. Together with "H. curviuscula" (JF713683) and "H. acuminata" (KC164379) and one uncultured bacterium (JF681416) they formed a monophyletic lineage, sister clade of the "Ca. Gortzia" genus. The sequences of H. undulata and H. elegans showed 99.93% identity, differing by only 1 bp.

DISCUSSION
Novel HLB from P. multimicronucleatum TP2 In the present study, we describe a novel HLB, belonging to the "Ca. Gortzia" genus. It occurred in P. multimicronucletum, a Paramecium species in which HLB have never been documented ). This endosymbiont shares many traits with the Holospora and "Ca. Gortzia" genera, such as the peculiar life-cycle involving two different stages (IF, RF), and ultrastructural internal compartmentalization of IF Wiemann and Görtz, 1989). Furthermore, it can be included in the group of "Ca. Gortzia infectiva", H. caryophila, "H. bacillata", and "H. curvata", due to the absence of a "connecting piece" and the presence of several membrane layers surrounding IF. This particular feature, never observed in those Holospora species able to induce the "connecting piece", is probably related to host-invasion or a releasing mechanism (Fokin et al., 1996;Fokin and Sabaneyeva, 1997;Fokin, 2015). Moreover, as observed in "H. bacillata" (Fokin, 1989), IF were able to divide by binary fission. On the other hand, we detected ultrastructural differences with other HLB Görtz et al., 1990;Wiemann and Görtz, 1991): IF of the new bacterium showed a darker, less homogeneous cytoplasm, an unevenly distributed periplasm with two distinguishable regions, and a recognition tip that did not show the additional subdivisions detected in "Ca. Gortzia infectiva" (Boscaro et al., 2013a). The presence of two recognition tip-like structures in IF and TF has been interpreted as an early stage of division for IF (since IF can divide) or differentiation and division, in the case of TF.
Our analyses emphasizes additional and surprising features, such as the simultaneous and extensive presence of bacteria both in the host cytoplasm and MA: this pattern has never been reported for any species of Holospora nor for "Ca. Gortzia infectiva" Boscaro et al., 2013a). The presence of IF in the cytoplasm has previously been reported and interpreted as a secondary infection of the host or as IF trafficking through the cytosol to exit the cell (Fokin and Sabaneyeva, 1997;Fokin, 2015). The presence of dividing RF in the cytoplasm (Figure 5G) suggest that in this case, unique in HLB, the symbiont is able to complete its life-cycle in the cytoplasm. Further investigation is needed to resolve this issue.
The only organism with a similar behavior among phylogenetically relatives is "Ca. Paraholospora nucleivisitans", a cytoplasmatic endosymbiont of P. sexaurelia, but observed to enter the MA of its host (Eschbach et al., 2009); but it was rarely present in both cell compartments simultaneous. Furthermore, it showed different morphological features and low 16S rDNA identity (83.95%) with all HLB, indicating that this feature is not a shared derived character of the two organisms.
The other new characteristic of this novel HLB, as mentioned above, is the highly variable size and shape of IF and TF, observed after several months of culturing. IF, indeed, showed an abnormal pattern of growth (it almost doubled its length in some cases) and bacterial periplasmic distribution, when compared to the classical HLB morphology. We do not yet know the cause for these unusual variations in IF/TF shape. The overall supposition is that "aberrant" IFs faced some physiological problem in developing the classical morphology. Indeed, the increased number of TFs in the MA could be caused by an inability to exit this stage, with a consequent increase in size. When they were able to develop cellular differentiation, the periplasmic part has been produced in an unevenly pattern. In other words, these traits could be due to physiological stress, disease, or mutations that accumulate in vertically inherited bacteria, with a probable reduction in the efficiency of horizontal transfer, or due to changes in host/symbiont interactions occurred under laboratory growth conditions. The fact that the classical HLB morphology is observed in the experimentally infected PC6 strain could be due either to a positive selection for non-altered bacteria, still able to perform horizontal infection correctly, or to the capability of bacteria to restore the initial cell morphology in a different, unaltered, host environment. Although these are just hypothesis, the observed features can be used for taxonomic purposes. From a molecular point of view, this new HLB is phylogenetically close to "Ca. Gortzia infectiva" from P. jenningsi. Our results suggest that "Ca. Gortzia infectiva" and the new HLB coexist in the same geographical region and may have host specificity, a feature common in the sister genus Holospora.
To conclude, for this HLB we suggest the status of a novel species for the following reasons: 1) the different morphology and the striking phenotypic plasticity, never observed before in HLB; 2) the ability to infect both the MA and the cytoplasm; 3) the new host species in which it has been found, P. multimicronucleatum, given the fact that species-specificity between host and endosymbiont is well-known for most HLB; 4) the 16S rDNA distance from closely related species, which is compatible with threshold values proposed for new species description (Rosselló-Móra and Amann, 2015). For this new HLB species, belonging to the "Ca. Gortzia" genus, we propose the name "Candidatus Gortzia shahrazadis", accordingly with Murray and colleagues (Murray and Schleifer, 1994;Murray and Stackebrandt, 1995). This specific name was chosen due to the Asian origin of this species and because one of the authors is known to say "I will finish this paper after one thousand and one nights." A description of this new species is present at the end of the discussion.

"Ca. Gortzia infectiva" and Holospora obtusa from India
We detected HLB in an Indian Paramecium population. A first analysis of phenotypic characters and host specificity led to the species attributions "Ca. Gortzia infectiva" and H. obtusa, respectively, from P. jenningsi and P. caudatum. In both cases, the typical IF and RF were observed inside the host MA, with typical size and features described in the literature Boscaro et al., 2013a). Molecular analyses was performed on fixed material and, in both cases, results were congruent with live observations, confirming the species assignment, despite both Paramecium caudatum 18S rDNA sequence and its H. obusa 16S rDNA sequences showed some peculiarities when compared to sequences of conspecific organisms, suggesting a level of divergence. Phylogenetic analysis showed clear association of the Indian H. obtusa with its conspecifics sampled in other countries. Nevertheless, according to branch lengths, Indian H. obtusa SH2 showed a clear divergence from northern strains.
In our trees all HLB cluster together: Holospora clade appeared as sister group of the "Ca. Gortzia" genus. In general, our phylogeny of the order Rickettsiales is in agreement with previously published analyses (Boscaro et al., 2013a;Hess et al., 2016). Additionally, we note that the sequences belonging to H. undulata and H. elegans are very similar. While past analyses included a short partial sequence (479 bp) from H. elegans (AB297813) (e.g., Boscaro et al., 2013a), to our knowledge this is the first study reporting the full-length gene sequence of this species, derived from its genomic assembly (Dohra et al., 2014). On the basis of the present data and given the fact that both H. undulata and H. elegans have been found in MI of P. caudatum, H. undulata-which has an unusual undulated shape in comparison to other Holospora-could be a morphotype of the species H. elegans. Nevertheless, further investigation is needed, including analysis of a greater number of strains and comparison at the whole-genome level. To our knowledge, the several strains of H. undulata molecularly characterized by our and other laboratories are all identical to the presently published 16S rDNA sequences (data not shown), while for H. elegans only a single complete sequence is available (Dohra et al., 2014). In any case, since these two species were described in 1890 by Hafkine (Hafkine, 1890), it seems particularly important to clarify this issue.

Reconsideration about HLBs Distribution
Our data confirm the presence of HLB endosymbionts in Paramecium species inhabiting different and distant locations in India. P. multimicronucleatum (TP2) was collected from freshwater Kolleru lake in Andhra Pradesh; P. jenningsi (BJ1) was sampled in Chilka Lake, a brackish lagoon in Odisha; and P. caudatum (SH2) was found in a freshwater stream at the top of Araku Hills (Andhra Pradesh). All these findings strongly suggest a relatively common presence of HLB in Indian ciliate populations, adapted to different types of tropical habitats.
To date Holospora has been considered an endosymbiont adapted to cold and temperate countries. It has been demonstrated that transmission efficiency of H. undulata, for example, is strongly affected by temperature, being more effective at 10 • C than at 23 and 30 • C (Fels and Kaltz, 2006). Moreover, Holospora was never found in tropical habitats. For all these reasons, our data introduce new and interesting information about HLB's distribution, being the lowest latitude record for the Holospora genus reported (Figure 7; Table 3).
Nevertheless, the stream from which we sampled H. obtusa was located on the top of Araku hills, 700 m above the sea FIGURE 7 | Holospora-like bacteria distribution. Yellow, orange, and red indicate countries in which Holospora-like bacteria have so far been detected, whereas a gray indicates those areas not yet investigated for their presence. Asterisk indicates sampling area of Holospora obtusa and "Ca. Gortzia" species characterized in the present work. laboratory experiments (Hori and Fujishima, 2003). Thus, it is possible that HLB confer a degree of benefit to their host if exposed to temperature stress, although this hypothesis is controversial (Duncan et al., 2010).
Obviously, another necessary consideration is that the biogeographic distribution of HLB is highly dependent on their hosts' distribution. The Paramecium genus is cosmopolitan, although some species seem to be adapted to certain climatic conditions. P. jenningsi, for example, has been found mostly in tropical countries, whereas P. caudatum is reported mainly from northern areas with temperate climates (Wichterman, 1986;Fokin, 2010/11). However, fewer sampling efforts in tropical countries strongly affects the present understanding of ciliate distribution, and thus the distribution of their endosymbionts (Görtz, 2008;Fokin and Sera, 2014).
To conclude, our record of H. obtusa from India reshapes the known Holospora distribution and underlines the need for further sampling and research efforts, to better clarify the ecological significance of HLB.

Description of "Candidatus Gortzia shahrazadis"
Gortzia shahrazadis (Gor'tzi.a shah.ra.za'dis; N.L. fem. n. Gortzia, in honor of Professor Emeritus Hans-Dieter Görtz; N.L. adj. shahrazadis, of Shahrazad, main character and charming story teller in the Arabian collection of tales, "The One Thousand and One Nights"). Mainly macronuclear endosymbiont which can sometimes be observed in the cytoplasm of P. multimicronucleatum, sampled from Kolleru, a freshwater lake in India. It has two life-cycle stages: a small, reproductive form (RF: 2.5-3.3 µm) and a rod-shaped infectious form (IF). The latter shows cellular compartmentalization: cytoplasm, periplasm (with two areas distinguishable by different density), and recognition tip. Two morphotypes of IF are detected during different periods of laboratory culturing: a shorter IF at the beginning (6.9-10.7 µm), and a longer one (8.2-14.7 µm) with irregular distribution of periplasm after 1 year of cultivation. The latter are found together with a high number of transient forms (TF), which are long (6.7-15.5 µm), rod-shaped and not yet differentiated into IF at an ultrastructural level. All forms showed a constant diameter of 0.7-0.8 µm. No production of a "connecting piece" during host cell division was observed. Several membranaceous layers surrounded fully differentiated IF. Basis of assignment: 16S rRNA gene sequence (ENA database, accession number LT549002) and positive matching with the 16S rRNA-targeting oligonucleotide probe GortzSha_1266 (5 ′ -TTTTGATTTGCTCAAGGTCGC-3 ′ ).

AUTHOR CONTRIBUTIONS
VS, SF, CB, and VN actively performed sampling in India. CB, VN, BS, CK organized and logistically supported sampling activity in India, VS carried out most of FISH experiments, molecular and phylogenetic analyses, probes and primers design. SF performed preliminary microscopic observation finding all the endosymbionts, morphologically identified the hosts, and performed TEM analysis. MC helped to realize the phylogenetic study. CB and VN performed part of the molecular characterization. SF, FV, BS, CK, and GP carefully supervised all the experiments and gave suggestions for data interpretation. VS wrote the manuscript. All the authors were involved in manuscript revision. GP coordinated the whole research activity.