Diversity and Universality of Endosymbiotic Rickettsia in the Fish Parasite Ichthyophthirius multifiliis

Although the presence of endosymbiotic rickettsial bacteria, specifically Candidatus Megaira, has been reported in diverse habitats and a wide range of eukaryotic hosts, it remains unclear how broadly Ca. Megaira are distributed in a single host species. In this study we seek to address whether Ca. Megaira are present in most, if not all isolates, of the parasitic ciliate Ichthyophthirius multifiliis. Conserved regions of bacterial 16S rRNA genes were either PCR amplified, or assembled from deep sequencing data, from 18 isolates/populations of I. multifiliis sampled worldwide (Brazil, Taiwan, and USA). We found that rickettsial rRNA sequences belonging to three out of four Ca. Megaira subclades could be consistently detected in all I. multifiliis samples. I. multifiliis collected from local fish farms tend to be inhabited by the same subclade of Ca. Megaira, whereas those derived from pet fish are often inhabited by more than one subclade of Ca. Megaira. Distributions of Ca. Megaira in I. multifiliis thus better reflect the travel history, but not the phylogeny, of I. multifiliis. In summary, our results suggest that I. multifiliis may be dependent on this endosymbiotic relationship, and the association between Ca. Megaira and I. multifiliis is more diverse than previously thought.


INTRODUCTION
Rickettsial bacteria (Order Rickettsiales), members of alphaproteobacteria, are well-known as the causative agents for insect-borne human diseases such as typhus, scrub typhus, and Rocky Mountain spotted fever (Walker and Ismail, 2008). These bacteria are gram-negative, obligate intracellular organisms, and their presence was once thought to be limited to animals, particularly insects and vertebrates (Raoult and Roux, 1997). Recently, surveys of environmental samples revealed that in addition to the pathogenic rickettsia, rickettsia-like bacteria could be found as endosymbionts in a variety of species and from different habitats. However, the functions of these rickettsial endosymbionts in their hosts remain unclear.
The other group of rickettsia-like endosymbiotic bacteria, Candidatus Megaira, forms a sister clade to the genus Rickettsia (family Rickettsiaceae) (Schrallhammer et al., 2013). Based on SSU rRNA sequences, Schrallhammer et al. (2013) further classified Ca. Megaira into three subclades. Members of the subclade Ca. Megaira polyxenophila were identified in both marine and freshwater ciliates (Vannini et al., 2005), in green algae (Kawafune et al., 2012), in lake water from the US (Percent et al., 2008) and China, in subsurface water from South Africa, and in aquaria in Greece (Vlahos et al., 2013). The other two subclades, Ca. Megaira B and C, contain species found in diverse hosts and habitats including: ciliate Ichthyophthirius multifiliis (Sun et al., 2009), cnidarians (Fraune and Bosch, 2007;Sunagawa et al., 2009), siphonous green algae (Hollants et al., 2013), lake water from the US (Percent et al., 2008), water from a lagoon in North Pacific (Galand et al., 2012), and a wastewater treatment plant in France (Chouari et al., 2010). There have been no reports that these bacteria are pathogenic, and the growth and reproduction of ciliate Diophrys were not affected when inhabited by Ca. Megaira (Vannini et al., 2003).
While it seems that Ca. Megaira are widely spread, it is not clear how ubiquitous they are. Furthermore, how universal these bacteria are in isolates/populations of particular host species is less well-studied. Research carried out by Kawafune et al. (2012) showed that Ca. Megaira were present only in 1 of 12 isolates of four unicellular green algal species (Cateria), and in one of nine isolates of multicellular green algae Volvox carteri (Kawafune et al., 2014), suggesting that Ca. Megaira might not be ubiquitously found in all isolates of one species. However, despite the works on non-phagotrophic green alga, to our knowledge there have been no other research systematically examining the distribution of Ca. Megaira in one single species, particularly in phagotrophic ones.
The parasitic ciliate I. multifiliis is the etiological agent for the 'white spot disease' in freshwater fish (Matthews, 2005;Dickerson, 2011). I. multifiliis contains an oral apparatus (Dickerson, 2006), and are apparently phagotrophic (Loboda-Dunha and Azevedo, 1993). Moreover, endosymbiotic Sphingobacteria and rickettsial alphaproteobacteria were detected in two I. multifiliis isolates isolated from the state of Georgia, USA (Sun et al., 2009;Coyne et al., 2011). The rickettsial alphaproteobacteria were later identified as members of the Ca. Megaira subclade C (Schrallhammer et al., 2013). We are therefore intrigued to determine if Ca. Megaira can be detected in most, if not all, isolates of the phagotrophic I. multifiliis. Furthermore, the phylogenetic relationships among different isolates of I. multifiliis can now be well-resolved by using mitochondrial sequences (MacColl et al., 2015). The phylogenies of Ca. Megaira, if they are present in most isolates of I. multifiliis, can then be compared to that of I. multifiliis to help deduce transmission routes of Ca. Megaira.
In this study we show that Ca. Megaira can be detected in 18 isolates of I. multifiliis, collected from Brazil, Taiwan, and the US. Ca. Megaira can now be classified into four subclades based on their rRNA sequences, and at least three of the subclades are capable of inhabiting I. multifiliis. The significance of the ubiquitous distribution of Ca. Megaira in I. multifiliis, and the transmission routes of Ca. Megaira, are discussed.

I. multifiliis and DNA Isolation
Ichthyophthirius multifiliis was collected from infected fish in the US, Taiwan, and Brazil, and each isolate likely derived from a distinct population. This collection represents more than 20 years of effort-on many researchers' part-in the collection and storage of samples from fish farms and pet stores across the world ( Table 1). Isolates were named with a letter(s) denoting the state or the country of its origin and a sequential number in the order they were discovered ( Table 1). Among the 18 I. multifiliis isolates 9 have been previously reported (Lin et al., 1996;MacColl et al., 2015), but only endosymbionts in the G5 isolate have been studied (Sun et al., 2009). Isolates collected in the US were at one point in time cultivated in the lab following previously established protocols (Noe and Dickerson, 1995), and except for G15 and NY3, all other US isolates were clonal lines. I. multifiliis trophont cells were collected from infected fish either by gently rubbing the skin of fish (Cassidy-Hanley et al., 2011), or by using saline shock (Schmahl et al., 1989). DNA was extracted either following protocols described elsewhere (Cassidy-Hanley et al., 2011;MacColl et al., 2015), or using the Qiagen DNeasy Blood & Tissue Kit (Redwood City, CA, USA) following manufacturer's protocols. The protocol of using fish (to pass I. multifiliis) was approved by the Institutional Animal Care and Use Committee of Cornell University (protocol number 1996-0083).

Phylogenetic Analyses
DNA sequences were first aligned using T-Coffee (Notredame et al., 2000), and alignments were further manually corrected in Jalview (Waterhouse et al., 2009) and/or BioEdit (Hall, 1999). For phylogenetic tree reconstructions protocols described elsewhere were followed (MacColl et al., 2015). Briefly, maximum likelihood (ML) trees were constructed with models predetermined by jModeltest (rRNA sequences: GTR+G; concatenated cox-1 and nad1_b: TIM1+I) (Darriba et al., 2012) and 1,000 bootstrapping replicates (Guindon and Gascuel, 2003). For Bayesian analyses (MB), MrBayes was used with the setting of: GTR+G model, two independent runs (each with three heated chains and one cold chain), 2,500,000 MCMC steps, and a sampling frequency of 1,000 (Ronquist et al., 2012). By the end of MCMC the standard deviation of split frequencies reached 0.0068. A burn-in of 25%, or 625, was used to generate both parameters and the consensus tree.

Phylogeny of Ca. Megaira
We identified more than 50 unique rickettsia-like 16S rRNA sequences from 17 isolates of I. multifiliis, and at least one sequence was detected in each of the isolates. The sequence of the rickettsial endosymbiotic bacterium found in the 18th isolate, G5, was published in an earlier report (Sun et al., 2009). If there were no deviations between two sequences beyond three or more positions, or 0.23% difference among 1,302 positions including gaps, only one was chosen for subsequent analyses. After such filtration 42 sequences were retained.
Both Bayesian (MB) and ML trees place all newly reported rickettsia-like 16S rRNA sequences and Ca. Megaira sequences in a monophyletic group next to the genus Rickettsia, a result consistent with a previous observation (Schrallhammer et al., 2013) (Figure 1). From this point we will follow the nomenclature system established by Schrallhammer et al. (2013) and collectively regard all endosymbiotic bacteria in this monophyletic group as Ca. Megaira.
Based on our phylogenetic analyses, Ca. Megaira can be further divided into four well-supported subclades, and at least three of the subclades are capable of inhabiting I. multifiliis (Figure 1). The early diverging position and the grouping of the subclade Ca. Megaira polyxenophilia are consistent with previous findings (Schrallhammer et al., 2013). However, Ca. Megaira polyxenophilia was not known to inhabit I. multifiliis and here we show ample examples (Figure 1).
The remaining three subclades might have descended from a common ancestor after the split from Ca. Megaira polyxenophilia (blue arrow in Figure 1), but the bootstrapping support of the branching point from ML analyses is only moderate. Among these three subclades, the grouping of the subclade Ca. Megaira B is consistent with findings reported in other studies (Schrallhammer et al., 2013), and this subclade has been the only one that has not been detected in I. multifiliis. Ca. Megaira subclade C includes endosymbionts found in hydra, in a US lake sample, and in I. multifiliis (Figure 1). It should be noted, however, that while the rickettsial endosymbionts in the G5 isolate of I. multifiliis were first classified as members in Ca. Megaira subclade C (Schrallhammer et al., 2013), our results show that these and a few other rickettsial endosymbionts detected in other I. multifiliis isolates should be classified as a new Ca. Megaira subclade D (Figure 1). Moreover, to date rickettsial endosymbionts in Ca. Megaira subclade D have only been detected in I. multifiliis. A nucleotide blast search using the Ca. Megaira 16S rRNA sequence found in G5 against the NCBI nr/nt database failed to identify sequences-including those derived from environmental samples-with higher similarities than those found in Ca. Megaira subclade C (data not shown).

Phylogeny of I. multifiliis
We then determined the phylogeny of the 18 host isolates of I. multifiliis using concatenated mitochondrial cox-1 and nad1_b sequences. In a previous study based on nine isolates I. multifiliis could be classified into two distinct groups with the possibility of a third group (MacColl et al., 2015). With more isolates included in this study it is clear that the 18 isolates of I. multifiliis belong to three distinct groups (Figure 2).
Group II contains seven out of eight isolates collected from Arkansas, demonstrating a pattern of local, repetitive infection of fish stocks. This pattern matches the isolate histories: most of these isolates were collected from local hatchery farms (Table 1). In contrast, for isolates obtained from pet stores ( Table 1) where the sources of I. multifiliis are expected to be variable, such a pattern does not exist. For instance, among the four isolates obtained in New York State, NY3, NY4, and NY7 were obtained from three different pet stores and belong to Groups I, II, and III, respectively (Figure 2). A similar pattern holds for isolates TW5 (Group II) and TW7 (Group I), which were imported from vendors in Thailand to aquarium shops in Taiwan (Figure 2).

Distributions of Ca. Megaira in I. multifiliis
We next mapped groupings of I. multifiliis to the phylogenetic tree of Ca. Megaira (Figure 1, short vertical color bars). While in most I. multifiliis isolates we could only detect Ca. Megaira endosymbionts from one subclade, a few isolates, particularly those derived from pet store aquaria-NY4, NY7, and TW5harbored endosymbionts from two subclades. BR1 was inhabited with endosymbionts from three Ca. Megaira subclades. The detailed history of this isolate is, however, not clear.

Transmissions of Ca. Megaira
If Ca. Megaira are transmitted solely vertically, i.e., to sister cells through asexual division, we should expect congruent evolution between Ca. Megaira and I. multifiliis. The distributions of Ca. Megaira in I. multifiliis isolates show that a congruent pattern does not exist (Figure 1). The three Ca. Megaira subclades that are found capable of inhabiting I. multifiliis can be detected in all three groups of I. multifiliis. This observation would argue strongly against solely vertical transmission.
Furthermore, we found that all isolated samples from Arkansas, which fell into two distinct groups, were inhabited with Ca. Megaira subclade C, suggesting a strong influence of the local environment on the acquisition of endosymbiont strains.

The Presence of Sphingobacteria in I. multifiliis
Since Sphingobacteria were also detected in two isolates of I. multifiliis collected from Georgia, USA (Sun et al., 2009), we set out to determine whether Sphingobacteria, like Ca. Megaira, were present in all 18 isolates of I. multifiliis. In three isolates, G15, Ark11, and Ark12, where genomic sequences were determined using next generation sequencing technologies with high coverages, assembled contigs with high similarity (>90%) to Sphingobacteria 16S rRNA sequence (GQ870456.1) were not detected (data not shown). We also failed to detect Sphingobacteria or Sphingobacteria-like 16S rRNA sequences in cloned PCR products. These results suggest that Sphingobacteria is not present in all I. multifiliis isolates.  Figure 2). Ark11 share identical Ca. Megaira 16S rRNA to that of Ark12_6 and is not shown. Scale bar represents 0.1 nucleotide substitutions per site.

DISCUSSION
In this study we surveyed the presence and distribution of rickettsial endosymbionts Ca. Megaira in 18 isolates of the parasitic ciliate I. multifiliis. In contrast to findings reported in non-phagotrophic green alga, where only a few isolates harbored this endosymbiont (Kawafune et al., 2012(Kawafune et al., , 2014, Ca. Megaira could be detected in all 18 isolates of I. multifiliis, collected from North and South America, and Southeast Asia. Like many other ciliates, I. multifiliis has an oral apparatus and is presumed to be phagocytic when feeding on fish (Ewing et al., 1985;Dickerson, 2006). It is therefore possible that I. multifiliis acquires Ca. Megaira through phagocytosis and Ca. Megaira subsequently escapes from phagolysosomes using a mechanism similar to that used by pathogenic rickettsiae (Teysseire et al., 1995;Whitworth et al., 2005). On the other hand, ciliates are also equipped with a clathrin-mediated endocytosis pathway (Ramoino et al., 2001;Elde et al., 2005). Because rickettsiae invade mammalian cells in a clathrin-dependent manner (Chan et al., 2009), it is possible that Ca. Megaira enters I. multifiliis through clathrin-mediated endocytosis. Further experiments are needed to elucidate which mechanism(s) Ca. Megaira use to enter their hosts.
The ubiquitous presence of Ca. Megaira in I. multifiliis prompts us to consider whether the bacteria and the ciliate host have formed a dependent relationship, which may well be an example of hyperparasitism between bacteria (hyperparasite) and protozoan (pathogen) (Parratt and Laine, 2016). I. multifiliis shows a significant reduction of its somatic genome size (∼50 Mb) compared to two other free-living ciliate species in the same class (Oligohymenophorea) -Paramecium tetraurelia (72 Mb) (Aury et al., 2006) and Tetrahymena thermophila (104 Mb) (Eisen et al., 2006). Although parasitic protozoans tend to have smaller genomes (Ersfeld, 2003;Hertz-Fowler et al., 2005), I. multifiliis might have undergone further genome reduction due to the formation of a mutualistic symbiotic relationships with Ca. Megaira.
Prokaryotic endosymbionts have been identified in more than 200 ciliate species (Fokin, 2004;Schweikert et al., 2013), and these endosymbionts have been shown to provide hosts with nutritional support (Vannini et al., 2003), defense (Beale et al., 1969;Preer et al., 1972;Quackenbush and Burbach, 1983), and/or access to better environments (Finlay and Fenchel, 1989;Fenchel and Finlay, 1991a,b). It is unclear what roles Ca. Megaira may play in I. multifiliis biology. Lobo-da-Cunha and Azevedo showed that endosymbiotic bacteria in I. multifiliis, likely Ca. Megaira, were surrounded by glycogen in the cytoplasm (Loboda-Cunha and Azevedo, 1988). While this raises the interesting possibility that Ca. Megaira utilizes glycogen/glucoses derived from I. multifiliis, it does not immediately suggest that Ca. Megaira provides anything to its I. multifiliis host. Further details on metabolic dependencies between Ca. Megaira and I. multifiliis may be revealed when genomic sequences of Ca. Megaira become available.
With the addition of Ca. Megaira 16S rRNA sequences derived from I. multifiliis it is now clear that Ca. Megaira can be further divided into at least four subclades. Three of these subclades (polyxenophila, C and D) are capable of inhabiting I. multifiliis, while subclade B, which is found primarily in seawater samples, is not (Figure 1). The geographic isolation and/or changes in host tropism may contribute to this phenomenon (I. multifiliis only infect freshwater fish). Along the same lines, the fact that Ca.
Megaira subclade D appears to be present only in I. multifiliis suggests a specific tropism for these bacteria, although undersampling may explain this as well.
By comparing the phylogenies of I. multifiliis and Ca. Megaira, we concluded that transmission of Ca. Megaira is not solely vertical. Isolates of I. multifiliis in the same phylogroup may be inhabited with different subclades of Ca. Megaira. Moreover, local Ca. Megaira strains seem to play a more important role in determining which bacteria are present in I. multifiliis. The fact that all Arkansas isolates of I. multifiliis, regardless their phylogroups, were inhabited with Ca. Megaira subclade C, and not with other subclades, supports this idea. In this case, local bacteria may transmit horizontally to I. multifiliis and outcompete Ca. Megaira already inhabiting I. multifiliis. The presence of two subclades of Ca. Megaira in some clonally derived parasite lines (for example, NY4 and NY7), may reflect complex travel histories (exposures to different Ca. Megaira in different locations) and balanced competition between endosymbiont groups. The host/bacterial nature histories may always be more complicated than simple scenario we propose. Multiple gains/losses, in combination with horizontal/vertical transmissions, could result in what we observe today. More molecular sequences from Ca. Megaira will allow us to make better phylogenetic inferences not only between Ca. Megaira subclades, but also within subclades, with which we will be better able to determine the transmission routes of Ca.

Megaira.
Finally, host range and tropisms of Ca. Megaira remain to be further investigated. Ca. Megaira polyxenophilia is capable of inhabiting at least six ciliate species encompassing three distinct classes: Hypotrichea (Diophrys oligothrix, Diophrys appendiculata, and Euplotes octocarinatus), Heterotrichea (Spirostomum sp.), and Oligohymenophorea (Paramecium caudatum and I. multifiliis). Do these observations imply that Ca. Megaira (polyxenophilia) may be able to inhabit most, if not all, species in these three classes? Moreover, could Ca. Megaira subclade C and subclade D inhabit ciliate species other than I. multifiliis? In pathogenic rickettsiae it has been shown that several surface proteins, e.g., rOmpB (Sca5), Sca1, and Sca2, play significant roles in host cell adherence and invasion, and may be involved in determining host tropisms Chan et al., 2009;Riley et al., 2010;Uchiyama, 2012). An extensive survey of variations of these protein sequences from different Ca. Megaira subclades may help us gain insight on the host specificities of Ca. Megaira.

ETHICS STATEMENT
The protocol of using fish (to pass I. multifiliis) was approved by the Institutional Animal Care and Use Committee of Cornell University (protocol number 1996-0083).