Cultural isolation of spore-forming bacteria in human feces using bile acids

Structurally-diversified bile acids (BAs) are involved in shaping of intestinal microbiota as well as absorption of dietary lipids. Taurocholic acid, a conjugated form of BA, has been reported to be a factor triggering germination of a wide range of spore-forming bacteria in intestine. To test a hypothesis that other BAs also promote germination of intestinal bacteria, we attempted culture of bacteria from ethanol-treated feces by using a series of BAs. It was found that conjugated-BAs, notably three glycine-conjugated BAs, glycodeoxycholic acid and glycochenodeoxycholic acid, significantly increased the number and the species variety of colonies formed on the agar plate. These colonized bacteria mostly belonged to class Clostridia, mainly consisting of families Lachnospiraceae, Clostridiaceae, and Peptostreptococcaceae. There were several types of bacteria associated with different sensitivity to each BA. Eventually, we isolated 72 bacterial species of which 61 are known and 11 novel. These results demonstrate that the culturable range of bacteria in intestine can be widened using the germination-inducing activity of BAs. This approach would advance the research on spore-forming Clostridia that contains important but difficult-to-cultured bacteria associate with host health and diseases.


Scientific RepoRtS
| (2020) 10:15041 | https://doi.org/10.1038/s41598-020-71883-1 www.nature.com/scientificreports/ With bactericidal activity, BAs provide selective pressure for the colonization of bile-sensitive bacteria 25 . In contrast, some spore-forming bacteria including pathogens and commensals, germinate in response to BAs in the intestinal tract 7 . The germination of Clostridioides difficile spores, the pathogen that causes diarrhea, is stimulated by a combination of nutrients and bile salts present in the gastrointestinal tract, specifically taurocholic acid (TCA) [26][27][28][29] . It was recently reported that TCA also induces germination of certain populations of spore-forming bacteria in the intestine and expands the culturability of gut bacteria 7 .
Here, we hypothesize that many spore-forming bacteria in intestinal tract control germination by using TCA as the environmental signal. To address this hypothesize, we demonstrated the germination-inducing activity of a series of human BAs against spore-formers in feces, which are expected to include a number of yet-to-be cultured bacterial species.

Results
Germination of spores in feces by using various BAs. To examine the activity of a series of BAs on spore germination, seven fresh fecal samples were treated with an equal volume of 70% (v/v) ethanol to kill vegetative cells. Surviving spores were treated with each BA then spread and culture on GAM agar medium. For a negative control, non-BA treated spores were also spread and culture on the plate in the same manner. The number of colonies formed from samples treated by each BA signatures was counted individually for the seven fecal samples and statistically compared with the different BA groups (Table S1). The colony forming units (cfu) of the ethanol untreated samples was about 10,000 times that of the ethanol treated sample. As shown in Fig. 1, the cfu. were significantly higher in the samples treated by glycodeoxycholic acid (GDCA), glycochenodeoxycholic acid (GCDCA), taurochenodeoxycholic acid (TCDCA), TCA and glycocholic acid (GCA) compared to the BA-untreated sample, suggesting that these BAs have germination inducing activity for spore-forming bacteria present in the intestine (p < 0.05, Wilcoxon signed-rank test). GCDCA and GDCA in particular, showed approximately two-fold higher activity compared to TCA, which is a known germination inducer, although not statistically significant (p > 0.375, Wilcoxon signed-rank test).

Spectrum of germination-inducing activity of the human BAs.
To assess the population and the phylogenetic variety of intestinal bacteria responsive to the 15 BAs, the bacterial cells of colonies formed on each BA plate were collected and subjected to 16S rRNA gene amplicon sequencing. For the control, colonies from fecal samples without ethanol treatments, which represented cultivable vegetative cells of non-spore-forming and spore-forming bacteria, were analyzed in the same manner. The relative abundance of each bacterial family was determined for each plate. The data obtained from fecal samples of 7 donors were statistically compared among each BA-treated groups and the non-ethanol treated control group (Fig. 2). Neither Bacteroidaceae and Bifidobacteriaceae were observed in the ethanol-BA treated samples while their colonization was confirmed on the plate of non-ethanol treated control, confirming that the ethanol treatment killed these non-spore-forming bacteria of these families. On the other hand, families of Clostridiaceae, Lachnospiraceae and Peptostreptococ-   Fig. 3 shows the distribution of OTUs observed from each BA-treated sample and their annotation to bacterial families and species. Colored cells indicate the presence of OTUs among the colonies cultured from each BA-treated sample. These OTUs were classified into 12 families, constituting mainly of Lachnospiraceae, Clostridiaceae, Peptostreptococcaceae, and Oscillospiraceae. Further, these OTUs were clustered into five types based on the responsiveness to the different BAs. Type-1 OTUs were detected from all BA-treated samples. Type-2 OTUs were commonly detected from the samples treated by conjugated primary BAs. Type 3 and Type 4 OTUs were detected from the samples treated by glycine-conjugated BAs or taurine-conjugated BAs respectively. Type 5 OTUs were detected from the samples specifically treated by TCA or GDCA. A number of OTUs, particularly from the samples treated by glycine-conjugated BAs, were not annotated with any known species (similarity < 97%), suggesting the effective enrichment of as-yet-uncultured species whose germination is triggered by glycine conjugated BA 30 .
Taxonomy of cultured isolates. We cultured 438 bacterial colonies isolated from the ethanol-BA treated fecal samples of seven donors, as well as 67 colonies from untreated feces of the same donors (Table S2). Theses isolates were subjected to taxonomy analysis using full-length 16S rRNA gene sequences. The derived 16S rRNA gene sequences were dereplicated into 94 OTUs with more than 98.7% sequence identity. The number of the MiSeq sequence reads mapped to these 94 OTUs accounted for 49.9% of total amplicon reads from fecal microbiotas of the seven subjects, suggesting that approximately half portion of human gastro-intestinal microbiota was successfully cultured. Relative abundance (%) Relative abundance (%)  For the control to capture culturable vegetative cells of non-spore-forming and spore-forming bacteria, fecal sample was directly cultured on the medium without ethanol and BA treatments. Colonies on the plate were pooled and subjected to the 16S rRNA gene amplicon sequencing. Also, to capture the microbiome of original fecal samples, total bacterial DNA was extracted from fecal samples and subjected to the 16S rRNA gene amplicon sequencing. OTUs were generated from the obtained 16S rRNA gene V3-V4 region sequences. Box plots show the distribution of the relative abundance of bacterial family among the data from seven donors. Different letters (a-c) indicate significant differences between treated BA signatures (p < 0.05, pair-wise Mann Whitney U test). n.s. non-significantly. www.nature.com/scientificreports/ Each OTU was taxonomically annotated by comparison against the EzBioCloud 16S rRNA gene database. Consequently, 94 OTUs were classified into four main phyla of human intestinal bacteria, namely Firmicutes, Bacteroidetes, Actinobacteria, and Proteobacteria (Fig. 4). Seventy-two OTUs observed from the ethanol-BA treated fecal samples belonged primarily to families Lachnospiraceae, Clostridiaceae and Oscillospiraceae, whereas 30 OTUs from the untreated samples mainly belonged to the Bacteroidaceae and Bifidobacteriaceae families. Out of 94 OTUs, 7 were observed from both treated and untreated. The taxonomy of each OTU was identified at the species level with more than 98.7% identity and at genus level with more than 94.5% identity to the 16S rRNA gene sequence of the closest species in the database 31 Figure 3. Heat map of detected OTUs in each BA-treated sample. OTUs whose abundances were higher than 0.1% on each plate were selected and then clustered according to their abundance in each BA-treated sample. Each cell shows by color, the relative abundance to total population (black, < 0.1%; green, 0.1% >). The taxonomic classification at family level was displayed in the "Family" column with colors representing the bacterial taxonomies at the bottom panel. The OTUs were also annotated at species level, while novel candidate species were listed with 16S rRNA gene sequences showing less than 97% identity with any described species (black in the "Species" column).   Table S1. Specificity in the responsiveness of spore-forming bacteria to BAs. Figure 5 shows the number of shared and unshared OTUs among the different BA-treated samples. Glycine-conjugated BAs offered somewhat higher number of unique OTUs compared to taurine-conjugated BAs. Notably, 4 and 3 OTUs belonging to genera Roseburia and Blautia, respectively, were obtained only from the samples treated by glycine-conjugated BAs (Table S2). Furthermore, from glycine-conjugated BAs, 10 novel candidate species were obtained, while only 2 was found from taurine-conjugated BAs. These results suggest the specificity in the responsiveness of sporeforming bacteria to BAs as well as validity of glycine-conjugated BAs to culture uncultured species of human gastro-intestinal tract.

Discussion
In this study, we focused on the germination-inducing activity of a series of human BAs against spore-forming bacteria in human gut microbiota and found that spores in human feces are largely responsive to BAs. The induction of germination depends on the BA signatures, conjugated BAs in particular. Two types of conjugated BAs were tested in this study, namely taurine-and glycine-conjugated BAs. These showed rather strong activity, in accordance with a previous study which showed the efficacy of TCA 6 notably for uncultured intestinal bacteria, as well as C. difficile whose spore formation is problematic to the antibiotic-associated colitis patients [26][27][28][29] . The present study showed that glycine conjugated BAs showed a novel germination-inducing capability in terms of effective range of target species including 9 yet-to-be cultured species belonging to class Clostridia. Our results showed that human intestinal clostridia tend to be more sensitive to glycine-conjugated BAs than TCA. The difference in the sensitivity spectra may be related to different intestinal microbial niche, such as taurine-conjugate 32 rich mouse intestine, glycine-conjugate rich adult human intestine 33 , and human infant's intestine lacking in glycine-conjugates 24 .
It is known that tere are complex machineries equipped in spores, allowing their germination in response to multiple environmental signals including not only bile acids but also nutrient signals such as amino acids, nucleic acids, and sugars. The systems are studied well in clostridial pathogens such as C. difficile, C. perfringens, and C. botulinum, in which they share common cascade signaling system with three Csp proteins 34 . These Csp proteases are also conserved in many Clostridia organisms, such as Clostridiaceae, Lachnospiraceae and Peptostreptococcaceae family 35 . However, germinant receptors are diversified such as Ger protein family to sense amino acids in C. perfringens and C. botulinum and CspC to sense taurocholate in C. difficile 34 . It is also known that C. difficile requires an amino acid signal, particularly glycine, as co-germinant of TCA, which may function through unknown receptor 36,37 . Glycine-conjugated BAs may interact with these cognate receptors as well as amino acid molecules and may efficiently trigger the germination of the wide range of gastro-intestinal tract bacteria . It warrants further study on the molecular mechanism of the signaling of glycine-conjugated BAs.
Previous large-scale isolation studies using TCA successfully isolated 105 novel species 7,8 . In this study, however, although on a smaller scale, most of the novel candidate species were isolated from the plate supplemented with glycine-conjugated BAs and only one was isolated from the TCA plate. This warrants a large-scale isolation study with varieties of BA molecules including novel conjugated BAs, such as phenylalanocholic acid, tyrosocholic acid and leucocholic acid, which were recently identified in human feces 38 .
In conclusion, our study showed the efficacy of a series of BAs for the cultivation and isolation of gastrointestinal tract spore-forming bacteria including uncultured species. The efficacy depends on the BA signatures www.nature.com/scientificreports/ and the target bacterial species. A further study to address its molecular mechanism is warranted, in order to understand the BA-mediated gut microbial ecosystem as well as to capture unculturable microbiota.

Materials and methods
Sample collection. To test spore germination activity of BAs for gastro-intestinal tract bacteria, we collected fresh fecal samples from seven healthy adults aged from 22 to 32 years. Fresh feces were collected in two sterile tube (size 76 × 20 mm), each with 2 ml RNA later for 16S rRNA gene sequencing analysis or phosphate buffered saline (PBS) for bacterial culture. Collected samples in RNAlater were stored at 4 °C until further processing for DNA extraction. Collected samples in PBS were immediately processed within 1 h to preserve the viability of anaerobic bacteria. . Stock solution of BAs was prepared by dissolving each BA in dimethyl sulfoxide to be 10% (w/v) except for TLCA, GLCA, TDCA, and GDCA to be 1% (w/v) followed by filter sterilization. Immediately before use, the BA solutions were diluted in reduced PBS to be 0.1% (w/v) except 0.01% (w/v) for the four BAs.
culturing. Bacteria culturing was performed under anoxic conditions in an anaerobic chamber (gas condition was 10% carbon dioxide, 10% hydrogen and 80% nitrogen). The fecal samples were homogenized in PBS (0.1 g stool per ml PBS), added by an equal volume of 70% (v/v) ethanol, and then incubated for 4 h at room temperature under ambient aerobic condition in order to kill vegetative cells 7 . Thereafter, the material was washed three times with PBS. Then, the ethanol-treated feces were resuspended with BA solutions (0.1 or 0.01 mg BAs per ml PBS) in order to stimulate spore germination and a spread on agar plate with Gifu-anaerobic medium (GAM, Nissui Pharmaceutical, Tokyo, Japan). For ethanol-untreated control, the fecal suspension was serial dilution and spread on the agar plates with GAM. The inoculated agar plates were incubated for 72 h at 37 °C in the anaerobic chamber. The colonies formed on the agar medium were picked and inoculated in 3 ml GAM broth, and thereafter incubated overnight for the liquid culture. The cultured bacteria were reserved in glycerol stock at − 80 °C while analyzing the sequence of full-length 16S rRNA gene.
16S rRNA gene amplicon sequencing and analysis. All colonies on the plates were picked and suspended in 1 ml PBS by using inoculating loop. The genomic DNA was extracted from the pool by the beadsphenol method as previously described 39 . Bacterial DNA was also extracted from fecal samples using the same DNA extraction process. The extracted DNA was stored at − 30 °C until sequencing was performed on the Illumina MiSeq.
The variable region, V3-V4, of the 16S rRNA gene, was amplified from the extracted DNA as a template using the universal primers 341f. (5′-CGC TCT TCC GAT CTC TGC CTA CGG GNGGC WGC AG -3′) and 805r (5′-TGC TCT TCC GAT CTG ACG ACTACHVGGG TAT CTA ATC C -3′), and TaKaRa Ex Taq HS (Takara Bio). The amplified products were then used as templates in a second PCR for further amplification using barcode-tagged primers. Both PCRs were performed as described previously 39 . PCR products were pooled and combined in equimolar amounts for sequencing using the Illumina MiSeq platform, generating 300 bp paired-end reads.
The sequences obtained from feces and cultured bacteria were processed using the Uparse pipeline in Usearch version 9.2 and 10.0. (https ://drive 5.com/useac h/manua l/upp_ill_pe.html). 40 Firstly, the pair of sequence reads were merged using the fastq_mergepairs script with mismatched windows up to 20 bases. After quality filtering, dereplication, the discard of singletons and length trimming, the merged sequences were clustered into 286 OTUs, each representing greater than 97% identity, using the UPARSE-OTU algorithm. The taxonomy of each OTU was assigned at phylum to species level using the EzBioCloud 16S rRNA gene database and the VSEARCH program 41 and a cut-off of 0.97.
Taxonomic classification of isolates. The taxonomy of the cultured isolates was analyzed using fulllength 16S rRNA gene sequences. The full-length 16S rRNA gene fragment was amplified by PCR using the 27F (5′-AGA GTT TGA TCC TGG CTC AG -3′) forward and 1497R (5′-GGT TAC CTT GTT ACG ACT T -3′) reverse primers. Purified PCR products were subjected to Sanger sequencing by GENEWIZ (New jersey, USA). The resulting sequences were aligned using the Uparse/Usearch pipeline and were then grouped into OTUs, each representing greater than 98.7% sequence identity as per the UPARSE-OTU algorithm. The full-length 16S rRNA gene sequence of each species-level OTU was searched in the EzBioCloud 16S rRNA gene database (last updated Nov. 12th, 2019) to assign taxonomic designations to the species level and either known or novel candidate species status 31 . For phylogenetic analysis, all OTU sequences were aligned using ClustalW 2.1 (https ://clust ralw.ddbj.nig/) and then analyzed using the neighbor joining method 42 . For the neighbor-joining tree, evolutional distances were calculated based on a two-parameter model 43 . The rarefaction analysis was calculated by using the Usearch alpha_rare command.