**4. Discussion**

Here, we report a new *Holospora*-like bacterium from the macronucleus of *P. putrinum*. All features of this bacterium, such as morphology, intracellular localization, complex life cycle, host and nuclear specificity, and infectivity, indicate a close relation of this endosymbiont to other HLB. The phylogenetic analysis based on the 16S rRNA gene sequence also shows that the endosymbiont is close to other HLB and belongs to *Holosporaceae* family. Recently described macronuclear symbionts "*Ca.* G. shahrazadis" and "*Ca.* G. infectiva" are the closest relatives of the new endosymbiont, and together they form a well-supported clade sister to *Holospora* genus. Consequently, we assign this symbiotic bacterium to the genus "*Ca.* Gortzia" and name it "*Ca.* Gortzia yakutica" sp. nov.

As with "*Ca.* G. infectiva" and "*Ca.* G. shahrazadis", the new endosymbiont does not induce formation of the connecting piece during the host division, and can escape into host's cytoplasm; these two features together can be considered to be distinctive for the genus. Another feature to discuss is the manner of transforming IFs to RFs. It is described that IFs of *Holospora* species do not undergo a binary division, but constrict at several points forming a specific chain of cells and divide into several cells simultaneously [6,37]. On the contrary Serra et al. describe binary fission of IFs for the "*Ca*. Gortzia shahrazadis" [12]. We observed a classical IFs chain formation as well as the behavior similar to what is described for "*Ca*. Gortzia" (Figure 2, shown with the white arrowhead). Our knowledge about the transformation of IFs into RFs for the new HLB is limited and based on observations of live *Paramecium* cultures with DIC microscopy, and henceforth is far from comprehensive. At the same time, this phenomenon certainly deserves to be carefully investigated in further studies.

The ability to form the connecting piece also places the new symbiont close to *"H. bacillata"*, *"H. curvata"* and "*H. sp.* from the macronucleus of *P. putrinum*" [1,8]. It is possible that this species was described previously by Fokin et al. as "*Holospora* sp. from macronucleus of *P. putrinum*" from Germany [1,38]. The original description has the information about localization, the shape and sizes of IFs and RFs; the ability to induce the formation of the connecting piece. The new HLB reported here and the endosymbiont from *P. putrinum* reported by Fokin et al. have a similar phenotype (same host and localization, same shape of the cell, both do not form the connecting piece), but the described sizes are different, with the new HLB being notably smaller (e.g., the length of IFs is 12 μm vs. 17 μm). As the culture of "*Holospora* sp. from macronucleus of *P. putrinum*" had been lost precluding a more detailed characterization, it is impossible to establish whether these two endosymbionts belong to the same species.

All HLB have very distinctive morphological and physiological features and form a monophyletic clade within *Holosporaceae* family [10–12,14]. *Holosporaceae* includes four HLB genera—*Holospora*, "*Ca.* Gortzia," "*Ca.* Hafkinia", *Preeria*, and several other genera, which do not share HLB phenotype. Takeshita et al. recently described an endosymbiont from an anaerobic Scuticociliate—"*Ca.* Hydrogenosomobacter endosymbioticus" [39]. This endosymbiont has an uncertain position on the phylogenetic trees based on 16S rRNA genes: according to Takeshita et al. it forms a sister taxon to HLB, but with the low branch support (less than 70%) [39]. In our analysis this species appears within the HLB clade (Figure 6), but the branch support is quite low as well. "*Ca*. H. endosymbioticus" does not have HLB characteristic features discussed above, and its phylogenetic placement would have to be revised when some additional molecular data become available. Another issue arises with "*Ca.* Hafkinia", which was described as a separate genus within HLB, based on the 93.9–94.5% similarity with *Holospora* species [14], whereas our analysis shows 96–96.5% similarity, which places it within the genus *Holospora*.

While phylogenetic analysis based on 16S rRNA gene sequences is undoubtedly useful and widely used to make decisions on bacterial taxonomy, the examples given above show the limitations of such approach. Different thresholds proposed by various authors [15,40], different approaches to multiple alignments and substitution models can affect similarity values and topology of phylogenetic trees. It has been recently demonstrated that complete genome sequences could be used to better define bacterial species [41,42]. Thus, we can conclude that we would be in a much better position to infer the phylogenetic relationships of the HLB clade when complete genomes of *Gortzia* spp. become available.

#### **5. Description of "***Candidatus* **Gortzia yakutica" sp. nov.**

Gortzia yakutica (Gor'tzi.a ya.ku'ti.ca; N.L. fem. n. Gortzia, in honour of Professor emeritus Hans-Dieter Görtz; N.L. fem. adj. yakutica, of or belonging to Republic of Yakutia, the name of the region where the bacterium was first collected).

Obligate macronuclear endosymbionts of the free-living ciliate *P. putrinum*, occasionally can be found in the cytoplasm. Sampled from the freshwater pond in Republic of Yakutia, Russia. Has two life stages: small reproductive forms (1–2 by 2–4 μm) and long infectious forms (1–2 by 7–12 μm, rod-shaped with tapered ends, sometimes slightly curved). No formation of the connecting piece was observed. Basis of assignment: SSU rRNA gene sequence (GenBank accession numbers: MT421875.1–MT421895.1) and positive match with the species-specific FISH oligonucleotide probe Gyak567 (5'-AGGTAGCCACCTACACA-3').

Type strain is YA111-52 carried by *Paramecium putrinum* YA111-52 (Culture Collection of Ciliates and their Symbionts, CCCS 1024, St. Petersburg State University). Unculturable outside of host cells so far.
