Culturomics revealed the bacterial constituents of the microbiota of a 10-year-old laboratory culture of planarian species S. mediterranea

The planarian species Schmidtea mediterranea is a flatworm living in freshwater that is used in the research laboratory as a model to study developmental and regeneration mechanisms, as well as antibacterial mechanisms. However, the cultivable microbial repertoire of the microbes comprising its microbiota remains unknown. Here, we characterized the bacterial constituents of a 10-year-old laboratory culture of planarian species S. mediterranea via culturomics analysis. We isolated 40 cultivable bacterial species, including 1 unidentifiable species. The predominant phylum is Proteobacteria, and the most common genus is Pseudomonas. We discovered that parts of the bacterial flora of the planarian S. mediterranea can be classified as fish pathogens and opportunistic human pathogens.

). (B) The nature of the bacterial membrane was determined by gram staining colouration. (C) The representativeness of each bacterial species is illustrated. Five experiments were each performed on ten individual worms. www.nature.com/scientificreports/  Taken together, these data show that the predominant phylum in the S. mediterranea laboratory planarian flora is Proteobacteria, that the most common genus is Pseudomonas, and that planarians share microorganisms with both invertebrates and vertebrates.

Discussion
Analysis of the microbiome of the laboratory strain planarian species S. mediterranea using culturomics methods allowed us to identify 40 cultivable bacterial species forming the microbiota repertoire of S. mediterranea, most of which are found in the planarian gut. Among the 40 isolated bacterial species, we identified one non classified bacterial species (Pseudomonas sp.) that needs to be taxonomically and biochemically characterized in further work, four bacterial species (Pedobacter schmidteae, Pedobacter ghigonii, Chryseobacterium schmidteae, Metabacillus schmidteae) that to date have been identified only in planarians [27][28][29][30] , and one bacterial species (Comamonas aquatilis) that has been identified in both pond water and S. mediterranea 46,49 . In the calf liver used to feed S. mediterranea, we detected several bacterial species, including Brochothrix thermosphacta, Lactococcus piscium, Pseudomonas frederiksbergensis, Pseudomonas gessardii, Serratia proteamaculans, Staphylococcus hominis, and  www.nature.com/scientificreports/ Pseudomonas azotoformans (Table S5). None of these were found in the planarian S. mediterranea, suggesting that they were thus eliminated by planarians. The division of bacterial species between the planarian gut and epidermal mucus remains difficult to interpret. We cannot be sure of 100% bacterial species repartition because bacteria can be regurgitated by planarians through the pharynx. This phenomenon most likely occurs for the following bacteria, which are primarily from the phylum Proteobacteria: Aeromonas veronii, Chryseobacterium scophthalmum, and Pseudomonas brennerii, along with Pseudomonas anguilliseptica bacteroidetes, because this bacterial strain are detected in water containing planarians (Table S6). Notably, no bacteria were detected in the water control, which does not contain planarians. We also observed a variation in the microbiota composition as a function of starvation time. Indeed, whereas only 14 bacteria were detected after 1 week of starvation, 40 bacteria were detected after 2 and 4 weeks of Table 2. Bacterial composition of the microbiota of the laboratory strain S. mediterranea starved for different duration. This table lists the bacteria identified in the planarian S. mediterranea microbiota starved for 1 week, 2 weeks, and 4 weeks as follows: species, phyla, and nature of the membrane analysed by gram staining. The distribution remained unchanged after 2 weeks of starvation. Five experiments were each performed on ten individual worms (see also Table S3). Identical results were obtained for each experiment and each worm tested. www.nature.com/scientificreports/ starvation. Interestingly, the bacterial composition remained the same between 2 and 4 weeks of starvation, and the bacteria detected after one week of starvation were also found in the microbiota after 2 and 4 weeks of starvation. The bacterial distribution remained unchanged after 2 weeks of starvation. This microbiota evolution can be explained in several ways. First, after 1 week of starvation, the number of bacteria present in the worms was too low to allow for the detection of all the bacterial strains. Second, the bacteria present in the liver, although eliminated by planarians, can shape the composition of the microbiota after feeding by promoting the growth of different bacterial strains. Third, the 14 bacteria detected in planarians after 1 week of starvation promoted the growth of the other strains. Fourth, feeding induces the growth of planarians, thus modulating tissue homeostasis and regeneration; it cannot be ignored that feeding allows the development of several bacterial strains from planarian microbiota.
Firmicutes and Actinobacteria represent only 10% of the bacterial population found in the planarian S. mediterranea; Proteobacteria (60%) and Bacteroidetes (20%) are the predominant phyla in the microbiota. Proteobacteria is one of the largest bacterial phyla, with six classes and more than 116 families having been recognized (http:// www. bacte rio. net/). Proteobacteria are gram-negative bacteria and play numerous roles in diverse microbial ecosystems (aquatic, soil, plant, animal). Proteobacteria are involved in maintaining homeostasis of the gastrointestinal tract anaerobic environment and, thus, in the stability of the strictly anaerobic microbiota. Members of the phylum Bacteroidetes are known to be involved in the synthesis of short-chain fatty acids such as butyrate, propionate, and acetate, which are rich sources of energy for the host 50,51 . They also participate in carbohydrate metabolism by expressing enzymes such as glycosyl transferases, glycoside hydrolases, and polysaccharide lyases. Interestingly, Bacteroidetes have been shown to synthesize conjugated linoleic acid, which is reported to have  www.nature.com/scientificreports/ immunomodulatory properties [52][53][54] . In addition, the bacteria present in the planarian microbiome are mostly gram-negative, and gram-negative organisms produce antimicrobial peptides 55 . Thus, it can be hypothesized that Bacteroidetes and other gram-negative bacteria, such as Proteobacteria bacteria, participate strongly in the control of the antimicrobial properties of planarians via antimicrobial peptides and immune regulation.
It is difficult to consider and discuss each bacterial strain found in S. mediterranea and to compare them with the microbiota of other simple animal models or humans. However, it has been reported that the Proteobacteria is the dominant microbial phylum found in Danio rerio, Apis melifera, and Caenorhabditis elegans 1,56 . The predominance of Proteobacteria has also been reported in the microbiota of another planarian species called Dugesia japonica 57 , as well as in the fruit fly, Drosophila. melanogaster 58 , and Hydra oligactis 59 . Notably, Firmicutes are also the predominant phyla for C. elegans. The other phyla, Bacteroidetes and Actinobacteria, were not found in the simple animal models cited above. Stenostomum leucops, belonging to Catenulida within the phylum Platyhelminthes, are tiny planarians of the phylum Platyhelminthes that reproduce asexually and have a lifestyle close to S. mediterranea. They also share several microorganisms with S. mediterranea, including S. epidermidis, A. tumefaciens, E. adhaerens, P. fluorescens, and V. paradoxas 60 . Table 3. The bacterial species identified in the laboratory strain S. mediterranea starved for 2 weeks are shared with environment, vertebrates and invertebrates. Bibliographic analysis allowed us to illustrate the potential sharing of the bacterial strains forming S. mediterranea microbiota.

Share with References
Acidovorax wautersii Environment and human 36 Acinetobacter guillouiae Environment and human 80 Aeromonas hydrophila Environment and human 81 Comamonas testosteroni Environment and human 82,83 Delftia acidovorans Environment and human 84,85 Micrococcus luteus Environment and human 86,87 Sphingomonas paucimobilis Environment and human 88,89 Staphylococcus haemolyticus Environment and human 90  www.nature.com/scientificreports/ In contrast to that of planarian species S. mediterranea, the normal human gut microbiota is predominantly composed of two major phyla, Bacteroidetes and Firmicutes, followed by Actinobacteria and Verrucomicrobia. Verrucomicrobia were not detected in S. mediterranea. In humans, the gut microbiota contains only a minor proportion of the phylum Proteobacteria. It has been shown that an increase in the Proteobacteria phylum is a potential signature of dysbiosis and indicates a higher risk of disease 61 . In humans, the presence of Proteobacteria within the microbiota is associated with an adaptation of the gut microbial community to the host's diet, which could improve the ability of the host to harvest energy from indigestible polysaccharides 61 . Accordingly, we can hypothesize that the presence of large amounts of Proteobacteria in S. mediterranea might be associated with the capacity of planarians to adapt their body size to food availability. Indeed, in the absence of food, planarian size decreases, whereas in the presence of sufficient amounts of food, their body size increases greatly 62,63 .

Pseudomonas brenneri
In the planarian S. mediterranea, the major genus found is Pseudomonas, unlike in Drosophila, where the most common genus is Klebsiella 64 . Several bacterial species found in planarians are common to the Drosophila microbiota, such as Micrococcus luteus, Micrococcus yuennansis, and Microbacterium oxydans. In S. mediterranea, Aeromonas veronii was also detected. Although its function remains unknown, it has been shown that Aeromonas veronii plays a crucial role in the immune response of D. rerio for upregulating neutrophil abundance, which leads to a downmodulation of inflammation 1 .
In 2016, Arnold et al. 65 reported that planarian S. mediterranea microbiota analysed by metagenomics contain more than 300 bacterial strains. Here, in our study, we reported 40 bacterial strains, among which we identified and cultivated 14 bacterial strains described by Arnold et al., including M. luteus, M. yunnanensis, S. capitis, S. epidermitis, A. guillouiae, A. tumefaciens, C. testosteroni, D. acidovorans, P. anguilliseptica, P. brenneri, P. fluorescens, P. gessardii, S. ginsenosidimutans, and V. paradoxus. Such a disparity can be explained by the methodology used; metagenomics highlights any nucleotide sequence related to a bacterial strain, but the cultivability of the strain remains unknown. In addition, as shown by Arnold et al., the methods of culturing S. mediterranea can easily change the microbiota composition 65 . The role of the feeding (beef liver vs. calf liver) cannot be excluded. Thus, the composition of microbiota from the same species of planarians, here S. mediterranea CI4W, but kept in another laboratory will be affected by the methodology of bacterial detection and the culture conditions. A similar situation has been described for the microbiota of D. melanogaster 66 .
We have also observed that the bacteria comprising the microbiota of S. mediterranea are shared with the environment (soil, water, brackish water, sewer, and plants) as well as with fish, which is in accordance with the lifestyle of planarians. Indeed, planarians are zoophages and live in water. We also found bacteria that are shared with humans. Indeed, 16 bacterial strains have been described as opportunistic human pathogens, which represent 40% of the S. mediterranea flora and are responsible for diarrhoea, bacteremia, and endocarditis. Some of them have been reported to cause nosocomial infections, such as Acinetobacter guillouiae, or infection in immunocompromised people, such as Micrococcus luteus, which is involved in bacteraemia associated with intravascular catheters and endocarditis, peritonitis, ocular infections, and urinary tract infections. We also identified Aeromonas veronii, which is commonly hosted by leeches and is known to be responsible for gastroenteritis in humans 67 . The virulence and opportunistic capacity remain unclear for some of the bacterial species identified, such as Pseudomonas fluorescens and Comamonas testosterone, which might be a cause of bacteraemia or gastroenteritis 68,69 . As in D. melanogaster, we detected bacteria that have been described to be opportunistic human pathogens, such as Micrococcus luteus. This is also an important pathogen for aquatic animals 70 , but it is also a probiotic which promotes the growth of Nile tilapia Oreochromis niloticus" 71 . We also found several Staphylococcus species responsible for bacteraemia, sepsis, and nosocomial infection. The planarian S. mediterranea microbiota flora also includes Aeromonas veronii, which is commonly hosted by leeches. Cases of infection, such as gastroenteritis, have been reported in people using leech therapy procedures 35 or eating contaminated fishes 72 . Similarly, Mycobacterium marinum infections have been reported to be associated with the exposure of damaged skin to polluted water from fish pools or objects contaminated with infected fish 73 . Planarians are known to be fish tank invaders. Thus, although the planarian S. mediterranea is a free-living flatworm, it cannot be ignored that planarians are a reservoir or host of several human microbial pathogens that might be transmitted to predators of planarians or released in fish tanks contaminated with planarians.
Several studies suggest the role of probiotics in tissue homeostasis, as well as in tissue regeneration 74 , and that manipulation of the microbiome could be a way to resolve some tissue homeostasis deficiencies 75 . It has been shown that antibiotic treatment affects the planarian microbiota, which leads to an alteration of the regeneration process of S. mediterranea 65 . In Dugesia japonica, metabolites such as indole produced by the bacteria Aquitalea sp. delay the regeneration of the tissue after amputation 57 . Although there are contradictory and controversial findings, it appears that commensals, symbionts, and pathogens from the human cutaneous microbiome can play an important role in the resolution of nonhealing wounds 75 .
The role of the microbiota in the antimicrobial capacity of S. mediterranea and in their ability to have trained immunity remain to be elucidated. For this purpose, it is important to have information concerning the composition of the microbiota of the laboratory strain used for the experiments and to consider that a divergence in results can be caused by the mode of culture.

Materials and methods
Culture of the planarian species Schmidtea mediterranea. The S. mediterranea asexual clonal line ClW4 24 was maintained at 18 °C in water. The water was first filtered through charcoal and ceramics with pores of 0.2 µm (manufactured by Fairey Industrial Ceramics Limited) and through a membrane of 0.2 µm (Thermo Scientific Nalgene Filtration Products) for 10 years. Microbiological analysis of the filtered water was performed by inoculation of 5% sheep blood-enriched Columbia agar plates (bioMérieux, Marcy l' étoile, France) with 25, Culturomics. Two-week starved planarians 0.4-0.6 mm in length were used for the experiments. Selected worms were placed on agar plates (13%) and pressed slightly to collect their epidermal mucus (also denoted as mucus in the manuscript). The recovered epidermal mucus (one planarian per sample) was mixed with sterile phosphate-buffered saline (PBS), and then 100 µL of sample was inoculated on 5% sheep blood-enriched Columbia agar (bioMérieux, Marcy l' étoile, France), buffered charcoal yeast extract (BCYE) (Oxoid Deutschland GmbH, Wesel, Germany), and lysogeny broth (LB) under anaerobic and aerobic conditions and incubated at 19, 28, or 37 °C for 1, 2, 3, and 4 days 76 . The whole microbiota (gut and mucus) was then characterized by grinding one two-week-starved animal in PBS (one worm per sample). Homogenates were inoculated on 5% sheep blood-enriched Columbia agar (bioMérieux, Marcy l' étoile, France), BCYE (Oxoid Deutschland GmbH, Wesel, Germany), and lysogeny broth (LB) under anaerobic and aerobic conditions and then incubated at 19, 28 or 37 °C for 1, 2, 3, and 4 days.

MALDI-TOF MS and bacterial identification.
Individual bacterial colonies were collected every day for 4 days, and then each colony was identified by matrix-assisted laser desorption-ionization time-of-flight mass spectrometry (MALDI-TOF MS) (Microflex Spectrometer; Bruker Daltonics, Bremen, Germany) as previously described 77 . The obtained MALDI-TOF MS spectra were imported into MALDI Biotyper 3.0 software (Bruker Daltonics) and analysed against the reference bacterial spectral database. The MALDI Biotyper RTC software interprets the results according to predefined values, i.e., values between 2.00 ≤ species identified ≤ 3.00; of 1.70 ≤ probably identified ≤ 1.99 and 0.00 ≤ no identification ≤ 1.69. The unidentified colonies (with values from 0.00 to 1.99) were sequenced using the complete 16S rRNA gene.

Sequencing of the 16S rRNA gene and bacterial identification. The unidentified bacterial colonies
were cultured under the appropriate conditions, and the genomic DNA of each bacterium was extracted using an EZ1 automate (BioRobot) and the EZ1 DNA tissue kit (Cat No./ID: 953034, Qiagen, Hilden, Germany) according to the manufacturing protocol. The genomic materials were quantified using a Qubit assay (Life Technologies, Carlsbad, CA, USA) and then amplified by standard PCR. The standard PCR protocol was performed in a Thermal Cycler Peltier PTC200 cycler thermal model (MJ Research Inc., Watertown, MA, USA). Each reaction was conducted in a final volume of 50 μL, containing 5 μL of DNA from each sample, 25 HotstarTaq-Ampli-Taq Gold (Life Technologies, Carlsbad, CA, USA), 1.5 μL of primers (Fd1-AGA GTT TGA TCC TGG CTC AG; Rp2-ACG GCT ACC TTG TTA CGA CTT (Eurogentec, Angers, France)) 78 and 17 μL DNAse/RNAse-free water. The amplification was performed as follows: an initial denaturation step at 95 °C for 15 min, 40 cycles of denaturation at 95 °C for 30 s, step hybridization at a temperature of 52 °C for 30 s, and elongation at 72 °C for 60 s. All PCR products were resolved in 0.5× Tris Borate EDTA buffer (Ref. ET020-A, EUROMEDEX, Souffelweyersheim, France) and 1.5% agarose (Ref. LE-8200-B, EUROMEDEX, Souffelweyersheim, France), purified using NucleoFast 96 PCR plates (Macherey-Nagel EURL, Hoerdt, France), and sequenced using the Big Dye Terminator Cycle sequencing kit (Perkin Elmer Applied Biosystems, Foster City, CA, USA) with an ABI Prism 3130xl Genetic Analyser capillary sequencer (Applied Biosystems, Bedford, MA, USA). The following primers were used for the sequencing of the complete 16S rRNA: Fd1-AGA GTT TGA TCC TGG CTC AG; Rp2-ACG GCT ACC TTG TTA CGA CTT; F536-CAG CAG CCG CGG TAA TAC; R536-GTA TTA CCG CGG CTG CTG; F800-ATT AGA TAC CCT GGTAG; R800-CTA CCA GGG TAT CTAAT; F1050-TGT CGT CAG CTC GTG; and R1050-CAC GAG CTG ACG ACA (Eurogentec, Angers, France). CodonCode Aligner software was used for alignment and assembly and to correct the sequence (https:// www. codon code. com/). A consensus sequence was generated after analysis. BLASTn searches were performed against the nr database to check the similarity of the sequence (https:// blast. ncbi. nlm. nih. gov/ Blast. cgi). A sequence similarity threshold of 98.65% by comparison with the phylogenetically closest species with standing in the literature was used to delineate species 79  www.nature.com/scientificreports/