Prebiotic effects of yeast mannan, which selectively promotes Bacteroides thetaiotaomicron and Bacteroides ovatus in a human colonic microbiota model

Yeast mannan (YM) is an indigestible water-soluble polysaccharide of the yeast cell wall, with a notable prebiotic effect on the intestinal microbiota. We previously reported that YM increased Bacteroides thetaiotaomicron abundance in in vitro rat faeces fermentation, concluding that its effects on human colonic microbiota should be investigated. In this study, we show the effects of YM on human colonic microbiota and its metabolites using an in vitro human faeces fermentation system. Bacterial 16S rRNA gene sequence analysis showed that YM administration did not change the microbial diversity or composition. Quantitative real-time PCR analysis revealed that YM administration significantly increased the relative abundance of Bacteroides ovatus and B. thetaiotaomicron. Moreover, a positive correlation was observed between the relative ratio (with or without YM administration) of B. thetaiotaomicron and B. ovatus (r = 0.92), suggesting that these bacteria utilise YM in a coordinated manner. In addition, YM administration increased the production of acetate, propionate, and total short-chain fatty acids. These results demonstrate the potential of YM as a novel prebiotic that selectively increases B. thetaiotaomicron and B. ovatus and improves the intestinal environment. The findings also provide insights that might be useful for the development of novel functional foods.


Results
YM was utilised in the KUHIMM. YM was prepared from yeast cell wall slurry as described previously 32 , with a final mannan concentration of 50.5%. The KUHIMM was set up by adding a 0.4% YM preparation (0.2% mannan) (referred to as YM) under anaerobic conditions, and each of the eight human faecal samples (HS1, HS2, HS3, HS4, HS5, HS6, HS7, and HS8) (referred to as FEC) was used as the inoculum. A control culture without YM was also prepared (referred to as CUL). We investigated whether mannan was consumed by the human colonic microbiota in the KUHIMM after 30 h of fermentation. Mannan consumption was confirmed in all samples ( Supplementary Fig. S1).
YM administration did not alter bacterial genus-level composition and selectively stimulated the growth of Bacteroides thetaiotaomicron and Bacteroides ovatus. The effects of YM on human colonic microbiota were investigated using next-generation sequencing (NGS) and quantitative real-time PCR (qPCR). DNA was extracted from KUHIMM samples with and without YM collected after 30 h of fermentation. The eubacterial copy numbers, evaluated by qPCR, reached 2.81-4.90 × 10 11 copies/mL (Supplementary www.nature.com/scientificreports/ Table S1), which were comparable to the reported cell densities in the human colon (approximately 10 11 cells/ mL) 35 . NGS was used for the V3-V4 region of bacterial 16S rRNA for gene sequence analysis of faecal samples and the corresponding cultures with and without YM using the Illumina MiSeq system. In total, 4,407,318 quality reads were obtained from the eight faecal samples and the corresponding KUHIMMs with and without YM ( Table 1). The numbers of operational taxonomic units (OTUs) and the Chao1 values for species richness were lower in the CUL and YM groups than in the FEC group (Kruskal-Wallis test, p < 0.05); however, there was no significant difference between the CUL and YM groups (Kruskal-Wallis test, p > 0.05). The Shannon index for species diversity was lower in the CUL group than in the FEC group (Kruskal-Wallis test, p < 0.05); however, there was no significant difference between the CUL and YM groups (Kruskal-Wallis test, p > 0.05). The Simpson index for species diversity was not significantly different among the FEC, CUL, and YM groups (Kruskal-Wallis test, p > 0.05). Thus, the microbial diversity in the KUHIMMs did not change with the addition of the 0.4% YM preparation.
Principal coordinate analysis (PCoA) revealed that the microbiota in each KUHIMM was shifted in the same direction from the original faeces, and individual faecal samples and corresponding KUHIMMs with and without YM were assigned to the same cluster (Fig. 2). Microbiota β-diversity based on unweighted UniFrac distances was not significantly different between CUL and YM (permutational multivariate analysis of variance, PERMANOVA, p = 0.98). Bacterial genus-level compositional analyses of microbiota in the FEC, CUL, and YM are shown in Fig. 3. Almost all bacterial genera in the original faeces were also detected in the KUHIMMs. Comparing the relative abundance of 26 representative bacterial genera between CUL and YM, no significant Table 1. Summary of 16S rRNA gene sequencing data and α-diversity values (Chao1 estimator, Shannon index, and Simpson index). Eight human faecal samples, the corresponding culture without yeast mannan (CUL), and the corresponding culture with the 0.4% yeast mannan preparation (YM) were sampled 30 h after the initiation of fermentation. Values are the mean ± standard deviation. Asterisks (*) represent significant differences (*p < 0.05) (n = 8) between microbiota in the original faeces and the microbiota in corresponding cultures without or with yeast mannan using the Kruskal-Wallis test.  37 . After 30 h of fermentation, the numbers of six Bacteroides species were estimated by qPCR analysis (Fig. 4). As expected, the relative abundance of B. thetaiotaomicron was significantly increased in the YM group compared to that in the CUL group (Wilcoxon signed-rank test, p = 0.036). Remarkably, the relative abundance of B. ovatus was also significantly increased in the YM group compared to that in the CUL group (Wilcoxon signed-rank test, p = 0.036). Conversely, the relative abundance of the other Bacteroides species, B. caccae, B. uniformis, B. fragilis, and B. vulgatus, was not significantly different (Wilcoxon signed-rank test, p = 0.48, 0.61, 0.69, and 0.35, respectively) between CUL and YM. Thus, B. thetaiotaomicron and B. ovatus in the KUHIMMs were selectively increased by the addition of the 0.4% YM preparation.
YM administration reduced the pH and enhanced acetate and propionate production. The pH reflects the intestinal environmental condition, and low pH inhibits the growth of pathogenic bacteria, resulting in the reduction of putrefactive compounds 38 . Supplementary Figure S2 shows the results of continuous monitoring of pH during culture. After 30 h of fermentation, the pH was significantly reduced in the presence of YM compared to that of the CUL group (Wilcoxon signed-rank test, p = 0.025, Fig. 5a).
SCFAs are metabolic products of human gut microbiota, which act as signalling molecules and provide beneficial effects for host health 39 . Acetate, propionate, and butyrate are the most abundant (≥ 95%) SCFAs in the human colon 40 . The impact of YM administration on the production of SCFAs was examined after 30 h of fermentation (Fig. 5b). The concentrations of acetate, propionate, and total SCFAs were significantly higher in the YM group than in the CUL group (Wilcoxon signed-rank test, p = 0.036, 0.017, and 0.025, respectively). In contrast, the concentration of butyrate was not significantly different between YM and CUL (Wilcoxon signedrank test, p = 0.67).

Discussion
The most recent definition of prebiotics is 'a substrate that is selectively utilised by host microorganisms, conferring a health benefit' 41 , and numerous studies on prebiotics have found health benefits not only for the gut but also for the host in general 31,42 . Most traditional prebiotics increase the number of specific bacteria, such as Lactobacillus and Bifidobacterium 31 . In addition, they selectively increase the abundance at the bacterial genus www.nature.com/scientificreports/ level; few studies have reported on prebiotics that selectively increase abundance at the bacterial species level. It has been reported that bacteria of the genus Bacteroides have various beneficial effects 36 ; among these bacteria, B. thetaiotaomicron and B. ovatus are expected to be utilised as potential novel probiotics 43,44 . Conversely, several species are pathogens and associated with harmful effects on host health, e.g. B. fragilis with the induction of abscess formation 45 and B. vulgatus with the development of ulcerative colitis 46 . Therefore, a product that selectively increases beneficial bacteria of the genus Bacteroides could be a functional food ingredient as a novel prebiotic candidate.
In this study, we investigated the effect of YM on human colonic microbiota and metabolic end products using an in vitro human faeces fermentation system, the KUHIMM. Bacterial 16S rRNA gene sequence analysis showed that YM administration did not change microbial α-diversity, β-diversity, or the relative abundance of representative bacterial genera. Analysis of the growth profiles of six Bacteroides species in the KUHIMM revealed that YM administration stimulated the growth of only two species, B. thetaiotaomicron and B. ovatus, through the consumption of mannan. These results indicate that YM selectively increases the abundance of B. thetaiotaomicron and B. ovatus. To the best of our knowledge, there are few prebiotics that increase microbes in a species-specific manner, and YM is the first material that selectively increases B. thetaiotaomicron and B. ovatus in the complex of human colonic microbiota. B. ovatus is reported to exhibit immunogenic and immunomodulatory functions, such as expression of the tumour-specific Thomsen-Friedenreich antigen as a target for a cancer vaccine 47 and alleviation of lipopolysaccharide-induced inflammation 48 . In addition to B. thetaiotaomicron, several strains of B. ovatus, B. vulgatus, and B. caccae metabolised mannan in monoculture 7 , although among these species, only B. ovatus was increased in human colonic microbiota.
Both B. thetaiotaomicron and B. ovatus can degrade various indigestible polysaccharides utilising Sus-like systems 22 . These bacteria break down polysaccharides extracellularly to liberate polysaccharide breakdown products (PBPs). Some of them produce PBPs exclusively for their own use, while others produce PBPs that they do not necessarily require but can be used for growth by other Bacteroides spp. having limited or no ability to use the polysaccharides 23,49,50 . In addition, there are potential effects outside the genus Bacteroides; B. ovatus liberates PBPs during growth on xylan, which can support the growth of Bifidobacterium adolescentis that are normally unable to utilise it 51 . One study showed that B. thetaiotaomicron uses YM exclusively through a selfish  Table S2). A model has been proposed wherein YM is degraded extracellularly by at least two GH76s (endo-α-1, 6-mannanases BT2623 and BT3792) and a GH99 (endo-α-1, 2-mannosidase and endo-α-1, 2-mannanase BT3862) within these PULs to liberate PBPs, which are then transported into the periplasm, where they are depolymerised to mannose 7 . Of these three proteins, B. ovatus possesses only one GH76 (BO3915), with a relatively lower degree of homology with the other GHs in MAN-PUL2, suggesting that the extracellular degradation of YM is incomplete. Thus, in the in vitro human colonic microbiota model, B. ovatus appears to have utilised PBPs generated by B. thetaiotaomicron from the YM by incorporating them into the periplasm, revealing a novel cooperative relationship between Bacteroides species. This interesting phenomenon might have evolved cooperatively between B. thetaiotaomicron and B. ovatus in complex gut microbial ecosystems where various microbes compete for limited nutrients.
An increase in the production of acetate, propionate, and total SCFAs was observed in the culture with YM. The phylum Bacteroidetes is known to primarily produce acetate and propionate as metabolic end products 40 . Therefore, it was suggested that YM administration stimulated the growth of B. thetaiotaomicron and B. ovatus, increased the relative abundance of the phylum Bacteroidetes, and resulted in an increase in acetate and propionate. These SCFAs are thought to have reduced the pH. Acetate and propionate are the most potent activators of GPR43, a receptor on the cell surfaces of adipose tissue 52 . Because one SCFA is utilised by intestinal bacteria to www.nature.com/scientificreports/ produce another SCFA, and changes in the intestinal microbiota compositions are associated with the production of SCFAs 39 , an increase in one SCFA ideally should not reduce the levels of another beneficial SCFA. Therefore, YM might be a useful prebiotic because it increased the production of acetate, propionate, and total SCFAs and did not decrease that of butyrate. Because there are various factors in the complex intestinal microbial ecosystem, including competition for prebiotics and cross-feeding interactions among microorganisms, even if a material is utilised by certain intestinal bacteria in monoculture, it may not necessarily increase the bacteria in the intestinal microbiota. Furthermore, the intake of prebiotics by humans may cause a considerable increase in the abundance of bifidobacteria, even if they do not affect the growth of bifidobacteria in in vitro monoculture 53 . Therefore, it was important to confirm that YM selectively increased B. thetaiotaomicron and B. ovatus abundance in the in vitro human colonic microbiota fermentation system, which reproduces the in vivo microbiota changes induced by prebiotics. The prebiotic effects of YM were confirmed at doses as low as 0.4% (0.2% mannan). Previous studies using this system have confirmed the bifidogenic effects of prebiotic oligosaccharides at a concentration of 0.5% 34 , while 0.2% did not change the colonic microbiota composition as reported in human and animal studies 33 . For this reason, compared to conventional prebiotics, YM may also exert prebiotic effects at lower doses in in vivo human clinical studies. However, when YM is ingested by humans, it may be affected by variation in diets; therefore, it is not  www.nature.com/scientificreports/ clear whether YM exhibits the same prebiotic effect. To develop YM as a microbiota-directed food ingredient for human consumption that selectively increases the abundance of B. thetaiotaomicron and B. ovatus, clinical studies are required to verify its prebiotic effect, the resulting health benefits, and the doses at which these effects are produced.

Conclusion
YM selectively increased the relative abundance of B. thetaiotaomicron and B. ovatus in the human colonic microbiota model. In addition, YM increased the production of acetate, propionate, and total SCFAs. These results show the potential of YM as a novel prebiotic that selectively increases B. thetaiotaomicron and B. ovatus and improves the intestinal environment.

Methods
Preparation of YM. YM was produced from yeast cell wall slurry provided by Asahi Group Foods, Ltd.
(Tokyo, Japan) as described previously 32 . The mannan concentration was measured by Japan Food Research Laboratories (Tokyo, Japan). The mannan concentration was 50.5%, calculated based on the mannose concentration after hydrolysis, which was quantified by high-performance liquid chromatography (HPLC).
Human faecal sample collection from volunteers. Faecal samples were obtained from eight healthy subjects in their thirties to forties who had not taken antibiotics for at least 2 months prior to sampling, as described previously 33  Operation of the KUHIMM with and without YM. The model culture system was operated using a multi-channel fermenter (Bio Jr. 8; ABLE, Tokyo, Japan) to construct the KUHIMM as described previously 33,34 .
Briefly, each vessel in the system contained autoclaved Gifu anaerobic medium broth (100 mL; Nissui Pharmaceutical Co.), with the initial pH adjusted to 6.5. Next-generation sequencing and data processing. NGS analysis was performed by Macrogen Japan Corp. (Kyoto, Japan). Samples for sequencing were prepared according to the Illumina 16S Metagenomic Sequencing Library Preparation protocols to amplify the V3 and V4 regions of the 16S rRNA genes. Bacterial 16S rRNA genes (V3-V4 region) were amplified using genomic DNA as the template. The following primers were used: 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′) 54 . PCR was performed according to the manufacturer's instructions. Amplicons were purified using AMPure XP beads (Beckman Coulter, Inc., CA, USA). Paired-end sequencing was performed on the Illumina MiSeq platform. Overlapping reads were merged using fast length adjustment of short reads 55 . Pre-processing and clustering of sequences to identify OTUs was performed using the CD-HIT-OTU software 56 . After short reads were filtered out and extra-long tails were trimmed, chimeric reads were identified and discarded. The remaining representative reads were clustered into OTUs based on a ≥ 97% similarity threshold. Taxonomic composition for each sample from phylum to species levels was generated using QIIME-UCLUST 57 against the RDP-16S rRNA gene database 58  www.nature.com/scientificreports/ each method. Standard curves were prepared by diluting reference fragments (10 1 -10 8 copies). To confirm the specificity of the amplification using SYBR Green, a melting-point-determination analysis was performed.
Measurement of mannan concentration in the model culture system. The remaining mannan in the culture medium was analysed at 0 h and 30 h after the initiation of fermentation and was calculated based on the mannose concentration after hydrolysis, which allows the measurement of mannan utilisation. Samples were prepared according to the method described by Goubet et al. 62 with minor modification. Briefly, each culture broth was centrifuged at 10,000 × g for 5 min, and 100 μL of the supernatant was recovered, which was then hydrolysed using 1 mL of 2 M trifluoroacetic acid for 4.5 h at 100 °C. The samples were combined with 1 mL of 99.5% ethanol and dried using a centrifugal evaporator (Genevac Ltd., Ipswich, UK). The dry residue was resuspended in water, and the low-molecular-weight (< 10 kDa) fraction was recovered using Vivaspin 500 MWCO 10,000 PES (Sartorius Stedim Biotech, Göttingen, Germany). The mannose concentration was determined by high-performance anion-exchange chromatography with a pulsed amperometric detector (HPAEC-PAD). HPAEC-PAD analysis was performed using a Dionex ICS-5000 (Thermo Fisher Scientific, CA, USA). The system was equipped with a CarboPac PA1 column (2 × 250 mm) in combination with a CarboPac PA1 guard column (2 × 50 mm) (Thermo Fisher Scientific). The mobile phases consisted of 10 mM NaOH (A) and 500 mM NaOH (B). Samples (10 μL) were applied to the column and eluted at a flow rate of 0.25 mL/min using the following linear gradient: 0 min-0% B; 20 min-0% B; 20.01 min-60% B; 35 min-60% B; 35.01 min-0% B; 50 min-0% B.
Measurement of SCFA concentrations. The concentrations of lactate, succinate, acetate, propionate, and butyrate were measured using an HPLC system (Shimadzu, Kyoto, Japan) equipped with an Aminex HPX-87H column (Bio-Rad Laboratories, Hercules, CA, USA) and an RID-10A refractive index detector (Shimadzu), as described previously 33 . Bioinformatics and statistical analyses. PULs similar to MAN-PUL1, MAN-PUL2, and MAN-PUL3 were searched using the PUL prediction tool described in PULDB 63 . The Kruskal-Wallis test and Wilcoxon signed-rank test were performed using SPSS software ver. 23 (IBM Japan, Ltd., Tokyo, Japan). PERMANOVA was performed using the R ver. 3.6.0 Vegan package. A p-value < 0.05 was considered statistically significant.