**4. Discussion**

The green algal endosymbionts are often designated as *Chlorella*-like organisms or as simple zoochlorellae because of the di fficulties in identifying them based solely on the morphology (see Figure 1). Especially, identification at the species level is almost impossible using light microscopic observations. The isolation of clonal cultures from the ciliate hosts is quite di fficult and time-consuming, but the described method above resulted in some axenic strains of green algal endosymbionts. The main problem for isolating endosymbionts from their hosts is the slow growth and the requirement of additional nutrients, which is included in supplements, such as soil and beef extract. Especially, *Chlorella variabilis* requires organic compounds and vitamins [33,34]. Several attempts have been undertaken to isolate green algal endosymbionts from their hosts. Loefer [14] was the first to obtain the endosymbiont in culture. He isolated this alga by taking green algae from the sediment of an axenic *P. bursaria* culture and spreading it on agar plates containing tap water with unknown organic compounds. Since then, several methods for the isolation of endosymbionts have been described [35–38]. Similar to our approach, washing of the ciliate and the usage of antibiotics and transfer onto agar plates were used in di fferent variants. However, the crucial points of our approach are the starvation of the ciliate for 24 h before rupture on agar plates and the microscopical check during all steps, providing some security that the isolated algae are the endosymbionts of *Paramecium bursaria*. Especially, the last point is of grea<sup>t</sup> importance. For example, Hoshina and Imamura [20] described that the strain CCAP 1660/13 of *P. bursaria* had an additional endosymbiont (*Coccomyxa* sp.); however, Pröschold et al. [1] revealed that this alga was not an endosymbiont and represented only a free-living alga co-occurring in the culture of this *P. bursaria* strain.

Another critical point is the choice of culture media for the endosymbionts. As highlighted, most of them need organic compounds for growth. Therefore, it is mostly likely that the three protocols provided in Achilles-Day and Day [39] resulted in the cultivation of free-living green algae, which are co-cultivated with the hosts. As described in Achilles-Day and Day [39], all green algal endosymbionts grew on media without organic nutrients. In contrast, the three steps of our method (Figure 2) rely on the microscopical control at each step as well as the elimination of contaminants and free-living algae growing outside in the medium. This method can result in axenic clonal cultures of green algal endosymbionts when they are needed for further investigations.

If it is only required to know which endosymbiont species is in a strain of *Paramecium bursaria*, the presented diagnostic PCR approach revealed an easy and fast method for species identification. Diagnostic PCR approaches have been successfully established in several approaches, such as for the identification of harmful algae ([40] and references therein). In *P. bursaria*, Tanaka et al. [41] used a PCR-based approach to demonstrate the success of the elimination of green algal endosymbionts

from their hosts. This was based on the small subunit of the rubisco gene (*rbc*S) gene and did not focus on the identification of the endosymbionts. Here, we used the ITS-2 for our diagnostic PCR approach because this gene has been used for species delimitation as described above, and for all described species belonging to the Chlorellaceae, ITS rDNA sequences are available in GenBank, which is the only reliable dataset for species identification within this group until now ([22] and references therein). With species-specific primers (Figure 3) in PCR amplifications, *Chlorella variabilis* and *Micractinium conductrix* could be exactly identified, which were confirmed by ITS-2 sequencing (Figures 4 and 5). Both endosymbionts were di fferently distributed among the five syngens of *P. bursaria*. Whereas the syngens R3–R5 exclusively had *C. variabilis* as an endosymbiont, both endosymbionts could be discovered in strains belonging to syngens R1 and R2, which originated exclusively from Europe. Gaponova et al. [15] also found *M. conductrix* in *P. bursaria* isolates collected in north Karelia (Russia). It appears that this green algal endosymbiont occurred only in Europe, whereas *C. variabilis* was distributed worldwide. Consequently, the subdivision into the "American" and "European" endosymbionts groups, i.e., *C. variabilis* and *M. conductrix*, respectively, as proposed by Hoshina and Imamura [20] and the references therein needs to be revised as *C. variabilis* was also found in *P. bursaria* isolates originating from Europe: Ciliate strains PB-19 (Poland, syngen R1), CIL-46 (Germany, syngen R2), CIL-19, and CIL-20 (both from Austria, syngen R3). In contrast to our findings here, Summerer et al. [42,43] described that ciliate isolates from the two locations about 50 km apart (called PbPIB and PbW) bore *Chlorella* sp. as endosymbionts based on ITS-1 rDNA sequences. Unfortunately, neither the cultures nor the DNA of these endosymbionts are available anymore for comparative investigations. However, in our investigations, both Austrian *P. bursaria* strains and both algal strains revealed *C. variabilis*, which are the first European isolates. Considering the findings of Jeanniard et al. [4], who demonstrated that specific chloroviruses infected these algal species were widely distributed, indicating that hosts containing *C. variabilis* and *M. conductrix* also occurred in these freshwater habitats. As demonstrated above, our diagnostic PCR approach provided a quick and precise identification of the endosymbiotic green algae occurring in *P. bursaria*. This promising method will discover new endosymbionts not only in the model ciliate *P. bursaria* but also in other ciliate and invertebrate hosts and finally elucidate the biogeographic patterns of endosymbiotic green algal species.

The characteristics of both endosymbionts found in *Paramecium bursaria* are summarized in Pröschold et al. [1]. Both species di ffered in the ITS-2 secondary structures and their ITS-2 barcode (Figure 6). Despite the variations in the helices I–IV, both species were di fferentiated by one compensatory base change (CBC) and one hemi-CBC (one-sided base change) in the conserved region (ITS-2 barcode). The genomes of both species were sequenced [44,45]. The genome of the strain CCAP 211/84 (NC64A) *Chlorella variabilis* (ITS-2 barcode: CVAR in Figure 6) has a size of 46.2 Mb and contains 12 chromosomes [44]. *Micractinium conductrix*, strain SAG 241.80 (ITS-2 barcode: MCON in Figure 6), has a larger genome (60.8 Mb with more than 13 chromosomes; [45]). The chloroplast and mitochondrial genomes of both species are similar in size (125 vs. 129 Kb and 78 vs. 75 Kb; respectively; [45]). Fan et al. [46] questioned the separation of both species into two di fferent genera based on a comparison of the chloroplast and mitochondrial genomes. However, only a few taxa of *Chlorella* and *Micractinium* were included in this study. Before generic revision can be taken into account, more species of both genera need to be investigated using an integrative approach.

**Figure 6.** ITS-2 secondary structure of *Chlorella variabilis* (blue) and *Micractinium conductrix* (yellow). The ITS-2 barcodes of both species (CVAR and MCON) is given as number codes.
