Growth stimulation of Bifidobacterium from human colon using daikenchuto in an in vitro model of human intestinal microbiota

Daikenchuto (DKT) is a Japanese traditional herbal (Kampo) medicine containing ginseng, processed ginger, and Japanese or Chinese pepper. We aimed to determine how DKT affects human colonic microbiota. An in vitro microbiota model was established using fecal inocula collected from nine healthy volunteers, and each model was found to retain operational taxonomic units similar to the ones in the original human fecal samples. DKT was added to the in vitro microbiota model culture at a concentration of 0.5% by weight. Next-generation sequencing of bacterial 16S rRNA gene revealed a significant increase in the relative abundance of bacteria related to the Bifidobacterium genus in the model after incubation with DKT. In pure cultures, DKT significantly promoted the growth of Bifidobacterium adolescentis, but not that of Fusobacterium nucleatum or Escherichia coli. Additionally, in pure cultures, B. adolescentis transformed ginsenoside Rc to Rd, which was then probably utilized for its growth. Our study reveals the in vitro bifidogenic effect of DKT that likely contributes to its beneficial effects on the human colon.


Results
Increase in the abundance of Bifidobacterium following administration of DKT to KUHIMM. An in vitro human colonic microbiota model, KUHIMM, was established using each of the nine human fecal samples as inoculum. We added 0.5% wt. DKT to KUHIMM to examine its effects on the microbiota. In all KUHIMM samples, eubacterial copy numbers were initially 2.41 ± 3.76 × 10 8 copies/mL and reached 3.45 ± 2.42 × 10 11 copies/mL at 48 h of cultivation.
Next-generation sequencing was used for bacterial 16S rRNA gene sequence analysis and yielded 4,776,828 quality reads for bacteria from the original fecal samples and KUHIMM cultures with and without DKT (Table 1 and Supplementary Figure S1). The corresponding KUHIMMs, with and without DKT, harbored similar numbers of operational taxonomic units (OTUs) as those in fecal inoculums (Mann-Whitney U-test, P = 0.19 and 0.93, respectively). Moreover, the Chao1 richness estimator indicated no significant differences between fecal inoculums and corresponding KUHIMMs with and without DKT (Mann-Whitney U-test, P = 0.29 and 0.72, respectively). Bacterial community diversity indices (Shannon and Inverse Simpson) were lower in the corresponding KUHIMMs without DKT than in fecal inoculums (Mann-Whitney U-test, P = 0.027 and 0.042, respectively). However, no significant differences were observed in the Shannon and Inverse Simpson indices of KUHIMMs with DKT and without DKT (Mann-Whitney U-test, P = 0.29 and 0.38, respectively).
Principal coordinate analysis revealed that DKT administration did not have such impact on total microbiota structure in the KUHIMM culture ( Fig. 1). Figure 2 shows the mean bacterial genus-level distribution in the original fecal inocula and the corresponding KUHIMMs, with and without DKT, for the nine healthy human subjects. Most of the bacterial genera found in the human fecal samples belonged to the phyla Actinobacteria, Bacteroidetes, Firmicutes, Fusobacteria, Proteobacteria, and Verrucomicrobia; these phyla were also found in the KUHIMMs ( Fig. 2A). Interestingly, the relative abundance of bacteria related to the Bifidobacterium genus was significantly higher in the KUHIMMs with DKT than in the KUHIMMs without DKT (Wilcoxon signed-rank test, P = 0.020, Fig. 2B).
Direct effects of DKT on gut microbiota. To directly investigate how 0.5% wt. DKT affects microbial growth, we cultured each bacterial species, Bifidobacterium adolescentis, Fusobacterium nucleatum, and Escherichia coli, with and without DKT. DKT significantly promoted the growth of B. adolescentis in pure cultures ( Fig. 3A, P = 0.024, two-way analysis of variance (ANOVA) with repeated measurements). However, DKT did not significantly promote the growth of F. nucleatum or E. coli (Fig. 3B,C, P = 0.62 and 0.052, respectively, twoway ANOVA with repeated measurements).
Next, we measured the concentrations of DKT constituents before and after cultivation with each bacterial species (Table 2). First, we measured the concentrations of 0.5% wt. DKT before cultivation and compared these concentrations to those after cultivation with B. adolescentis, F. nucleatum, or E. coli. Interestingly, although ginsenoside Rc concentration was lower after cultivation with B. adolescentis (P = 0.0005, Student's t-test), it did not change after cultivation with F. nucleatum or E. coli (P = 0.083 and 0.10, respectively, Student's t-test). Thus, B. adolescentis metabolized Rc to a greater extent than did F. nucleatum and E. coli. After cultivation with B. adolescentis, ginsenoside Rd concentration was higher than in the original DKT (P = 0.0002, Student's t-test), but lower after cultivation with F. nucleatum or E. coli (P = 0.0061 and 0.0026, respectively, Student's t-test). Thus, B. adolescentis produced more ginsenoside Rd than did F. nucleatum and E. coli.

Discussion
In this study, we incubated fecal samples with 0.5% wt. DKT in a model culture system. In this system comprising many human fecal microbiota samples, we observed a significant increase in the relative abundance of the Bifidobacterium genus. This observation was supported by our finding that DKT increased the growth of B. adolescentis in vitro using pure cultures. B. adolescentis strains represent key taxa of an adult-associated bifidobacteria because they appear to specifically colonize the guts of adult individuals 13 . Health-promoting probiotic properties have been associated with some bifidobacterial strains belonging to the species, B. adolescentis, B. animalis, B. bifidum, B. breve, and B. longum 14 . Thus, DKT can be considered a prebiotic that stimulates the growth of beneficial bacteria such as B. adolescentis. Oral supplementation with B. adolescentis protected against the development of nonalcoholic steatohepatitis in a mouse model 15 . Furthermore, anti-hepatitis B viral activity has been reported for B. adolescentis 16 . We found that B. adolescentis converted ginsenoside Rc into Rd. Biotransformation of ginsenoside Rc into Rd by α-L -arabinofuranosidase has been reported in Bifidobacterium longum H-1 17 and Leuconostoc sp. 22-3 18 . Further, Suzuki et al. 19 identified α-L -arabinofuranosidase enzyme activity in B. adolescentis. Substrates of the α-L -arabinofuranosidase enzyme include the terminal residues of 1,5-α-L -arabinan and 1,5-α-L -arabinooligosaccharide and the branched arabinofuranoside residues of arabinoxylan and arabinan 19 . Thus, α-Larabinofuranosidase in B. adolescentis in this study and the gut could catalyze the hydrolysis of the arabinofuranoside moiety attached to ginsenoside Rc 18 . Arabinofuranose is expected to be transported into the cell via transporters and utilized via arabinose utilization pathways 20 . B. longum subsp. longum utilizes arabinose as a carbon source 21 . Compound K, which is efficiently produced from ginsenoside Rd to a greater extent than from Rc, Rb1, and Rb2 and is absorbed by various target cells, has recently attracted a great deal of attention because of its anti-tumor, anti-inflammatory, anti-allergic, and hepatoprotective effects 1,22 . The above effects of compound K could be attributed to the stimulation of the growth of bifidobacteria.
DKT was added at a concentration of 0.5% wt. to the in vitro human colonic microbiota model, KUHIMM, which has a working volume of 100 mL. Colon content has been estimated to be approximately 400 mL 23 . Thus, we concluded that 2 g (= 0.5 × 4) of DKT is required to exert its bifidogenic effect in the colon. This amount of DKT is comparable with the amount previously used in human clinical studies (2.5-7.5 g-DKT/day) 24 . A

Methods
Fecal specimen collection. Fecal samples were obtained from nine healthy Japanese volunteers, who had not been treated with antibiotics for more than 6 months prior to the experiment. All participants were recruited according to the inclusion criteria: age of 30 to 60 years, being Japanese, non-smoking status, and good health and physical condition (Supplementary Table S1). All subjects provided written informed consent prior to specimen collection. Immediately following collection, each fecal sample was stored in an anaerobic culture swab (212550 BD BBL Culture Swab; Becton, Dickinson and Company, New Jersey, USA) and used within 24 h. The study was performed in accordance with the guidelines of Kobe University Hospital, and approved by the institutional ethics review board of Kobe University. All methods in this study were in accordance with the principles of the Declaration of Helsinki.

Operation of the model culture system with and without DKT.
We used a small-scale multi-channel fermentor (Bio Jr.8, ABLE, Tokyo, Japan) composed of eight parallel and independent anaerobic culturing vessels, as described by Sasaki et al. 12  and Na 2 HPO 4 ) supplemented with L-ascorbic acid (1.0% w/v; Wako Pure Chemical Industries, Osaka, Japan). Cultivations were initiated by inoculating one fecal suspension (100 μL) into each vessel to construct the in vitro human colonic microbiota model, KUHIMM. During fermentation at 37 °C, the culture broth was stirred at 300 rpm with a magnetic stirrer and continuously purged with a filter-sterilized gas mixture. Aliquots of culture broth were collected from the vessel at 48 h after the initiation of cultivation. Feces and culture broth samples were stored at -20 °C until use.
DKT used in this study was supplied by Matsuura Yakugyo Co., Ltd. (Aichi, Japan). It was a brown, soft extract, 1.25 g of which was produced from 1 g of Zanthoxylum peel, 2 g of processed ginger, and 1 g of ginseng.  www.nature.com/scientificreports/ Unlike the Kampo drug generally known as Daikenchuto, this DKT extract did not include starch syrup, which mainly consists of maltose and other sugars, to avoid the starch syrup effect on the growth of gut microbiota. To determine the effect of DKT, it was added into one of the vessels at a final concentration of 5.0 g/L (0.5% wt. per 100-mL vessel) prior to cultivation. Additionally, a control vessel was prepared without DKT.
16S rRNA bacterial profiling. Microbial genomic DNA was extracted from suspended feces and culture broth obtained from KUHIMM. Purified DNA was eluted into TE buffer (10 mM Tris-HCl, 1.0 mM EDTA) and stored at -20 °C until use. Bacterial 16S rRNA genes (V3-V4 region) were amplified using genomic DNA as the template and primers S-D-Bact-0341-b-S-17 (5′-CCT ACG GGNGGC WGC AG-3′) and S-D-Bact-0785-a-A-21 (5′-GAC TAC HVGGG TAT CTA ATC C-3′) 25 , as previously described 12 . The libraries were generated using a Nextera kit (Nextera XT Index Kit; Illumina Inc., San Diego, CA, USA). Paired-end sequencing reactions were performed on a MiSeq platform (Illumina). Paired-end reads with quality scores ≥ 20 were joined using Quantitative Insights Into Microbial Ecology (QIIME) version 1.9.1 software 26 . The UCLUST algorithm was used to cluster filtered sequences into OTUs based on a 97% similarity threshold 27 . Chimeric sequences were identified and excluded from the library using USEARCH 28 . Paired-end reads were taxonomically classified via the Greengenes taxonomic database using the Ribosomal Database Project Classifier 29 .
Real-time PCR analysis. Real-time PCR was performed using LightCycler 96 (Roche Diagnostics GmBH, Germany). The FastStart SYBR Green master mix kit (Roche) was used for reactions with a primer set targeting all eubacteria 30 . The PCR reaction and amplification were performed as described previously 30 . Analysis of DKT constituent. GAM (1 mL) containing DKT (5.0 g/L) and cultures of the above three strains were lyophilized using an FZ-2.5 Labconco vacuum freeze-drier (Asahi Life Science Co. Ltd., Saitama, Japan) after centrifugation at 6000 rpm for 5 min to remove microorganisms. All standard ginsenosides (except ginsenoside Rc), [6]-gingerol and hydroxysanshool were isolated and identified by comparing their spectral data with previously reported data [31][32][33] . Lyophilized samples were individually dissolved in methanol (1 mL). Each solution was filtered through a 0.45 μm Millipore filter unit (Advantec, Tokyo, Japan), and the filtrate samples (1 μL) were injected into the LC-MS system for analysis. The LC-MS analyses were performed using a Shimadzu LC-IT-TOF mass spectrometer (Shimadzu, Kyoto, Japan) equipped with an ESI interface. The ESI parameters were as follows: source voltage, + 4.5 kV (positive ion mode) and -3. Bioinformatics and statistical analysis. The Chao 1, Shannon, and Inverse Simpson indices were calculated using the QIIME software package (Caporaso et al. 2010). Data were compared between the groups using the Mann-Whitney U test, Wilcoxon signed-rank test, or a two-way ANOVA with repeated measurements in the JMP version 12. P < 0.05 was considered statistically significant.

Data availability
All sequences from the original fecal samples and corresponding KUHIMMs were deposited in MG-RAST as "Model Culture System of Human Colonic Microbiota_Daikenchuto" under the accession numbers mgm4891148.3-mgm4891174.3.