Low amounts of dietary fibre increase in vitro production of short-chain fatty acids without changing human colonic microbiota structure

This study investigated the effect of various prebiotics (indigestible dextrin, α-cyclodextrin, and dextran) on human colonic microbiota at a dosage corresponding to a daily intake of 6 g of prebiotics per person (0.2% of dietary intake). We used an in vitro human colonic microbiota model based on batch fermentation starting from a faecal inoculum. Bacterial 16S rRNA gene sequence analysis showed that addition of 0.2% prebiotics did not change the diversity and composition of colonic microbiota. This finding coincided with results from a clinical study showing that the microbiota composition of human faecal samples remained unchanged following administration of 6 g of prebiotics over seven days. However, compared to absence of prebiotics, their addition reduced the pH and increased the generation of acetate and propionate in the in vitro system. Thus, even at such relatively low amounts, prebiotics appear capable of activating the metabolism of colonic microbiota.

have shown that administration of 1-kDa dextrans (linear and α-1,2 branched) at a concentration of 1% could selectively increase the number of bacterial cells belonging to the genus Bifidobacterium 11,12 . However, given that the daily Japanese human diet is approximately 3 L (3,000 g) 13,14 , the amounts of prebiotics administered in previous studies (same or more than 1%, i.e., ≥30 g/day) were too high and should be decreased to reflect a more realistic dosage.
In vitro models do not pose the same ethical constraints as in vivo human trials and allow dynamic sampling to study microbial activity in vivo. Thus, in vitro batch fermentation systems inoculated with human faecal matter to mimic the human digestive tract environment have been applied to measure production of SCFAs 15 and test the modulatory effect of potential prebiotics on human gut microbiota [16][17][18] . By carefully constructing the necessary anaerobic conditions, we previously developed an in vitro batch fermentation system (hereafter referred to as Kobe University Human Intestinal Microbiota Model, KUHIMM) that is capable of hosting more than 500 microbial species found in a human faecal inoculum and can effectively mimic human colonic microbiota 13 . The number of microbial species in this system was similar to the that (400-1000) previously reported in human faecal samples 19,20 . Our KUHIMM was able to reproduce the bifidogenic effects of prebiotic materials (i.e., fructo-, galacto-, iso-malto-and xylo-oligosaccharides), in line with the results from human clinical trials 21 .
The aim of this study was to assess the effect of relatively low amounts of three different prebiotics, i.e., indigestible dextrin (DEX), α-cyclodextrin (αCD), and dextran (DXR), on human colonic microbiota using our KUHIMM system ( Table 1). The prebiotic concentration was set to 0.2%, corresponding to 6 g prebiotics per 3 L of daily dietary intake 13,14 , and was sufficiently low to avoid the occurrence of any intestinal discomfort. The bacterial composition in the KUHIMM following prebiotic treatment was evaluated by next-generation sequencing (NGS) and compared with that in the absence of prebiotics. Similarly, we compared also pH changes and production of SCFAs. In addition, a small human trial was performed with a daily intake of 6 g of DEX or αCD to compare the microbiota composition in actual human faeces with that in the KUHIMM.

Results
Prebiotics reduce the pH in the KUHIMM. The KUHIMM was operated by adding 0.2% DEX, αCD, or DXR, and each of eight human faecal samples (designated as F26, F40, F62, M27, M37, M38, M39, and M60) was used as the inoculum. A control that included no prebiotics was also prepared. During the fermentation, pH transition occurred due to production of mainly SCFAs and ammonia ( Supplementary Fig. S1). Continuous monitoring ( Fig. 1) revealed an overall reduction in pH in the later phase of fermentation in the presence of prebiotics compared to the control. The lower pH reflected changes in environmental conditions resulting from the interplay between prebiotics and human gut microbiota.
Addition of 0.2% prebiotics does not alter microbial diversity. DNA was extracted from human faeces and from KUHIMM samples collected at 30 h of fermentation. In all KUHIMM samples, eubacterial copy numbers, evaluated by quantitative PCR, reached 9.15 × 10 10 copies/mL (Supplementary Table S1) and were comparable to the reported cell densities in human faeces (approximately 10 11 /wet-g) 22 .
NGS was used for bacterial 16S rRNA gene sequence analysis of faecal samples and corresponding cultures with and without prebiotics. In total, 5,922,873 quality reads with an average of 148,072 reads were obtained ( Table 2). The numbers of operational taxonomic units (OTUs) and α-diversity values (Chao1, Shannon index, and Simpson index) were not lower in cultures treated with or without prebiotics than in faecal samples (Kruskal-Wallis test). Thus, as described previously, bacterial numbers in human faeces were maintained in the corresponding cultures of our KUHIMM 13 . There were fewer OTUs in cultures treated with DEX and αCD than in control cultures (P < 0.05, Kruskal-Wallis test); however, OTU numbers were higher in the presence of DRX than in the absence of prebiotics (P < 0.05, Kruskal-Wallis test). Chao1 values, which indicate species richness, were lower in cultures with DEX and αCD than in controls (P < 0.05, Kruskal-Wallis test), but were not significantly different between DRX-treated cultures and controls (P > 0.05, Kruskal-Wallis test). Moreover, the Shannon and Simpson indexes were not significantly different between cultures containing prebiotics and controls (P > 0.05, Kruskal-Wallis test). Therefore, species diversity in our KUHIMM did not change following addition of 0.2% prebiotics.
Addition of 0.2% prebiotics does not change microbial composition. Principal coordinate analysis (PCoA) of unweighted UniFrac distances revealed that microbiota in faecal samples shifted in the same direction as in the corresponding cultures without prebiotics (P = 0.454, multivariate analysis of variance (MANOVA)) ( Fig. 2). No significant changes were detected between the microbiota of cultures with and without prebiotics (DEX: P = 0.571; αCD: P = 0.986; DXR: P = 0.996, MANOVA).

The effect on faecal microbiota composition is confirmed by a human intervention study.
To compare the effect of a daily intake of dietary fibre on healthy human microbiota with data from the KUHIMM, we performed a preliminary, small scale, open-label clinical study. To this end, we investigated each faecal sample obtained from four human volunteers (F26, M37, M38, and M39), who ate a daily amount of 6 g of DEX or αCD for seven days. Before eating prebiotics, the composition of human gut microbiota was calibrated by drinking 200 mL of water for 14 days (washout-1) ( Supplementary Fig. S2). Faecal samples were collected four times during the final days of eating prebiotics and drinking water. NGS analysis showed that species (OTU) numbers and microbial diversity (α-diversity) did not differ significantly between faeces collected after eating prebiotics or after  drinking water (P > 0.05, Kruskal-Wallis test) ( Table 3). Similar to the results in the KUHIMM, PCoA analysis revealed that microbiota compositions clustered equally in the four faecal samples (two with prebiotics and two without) from the same individual (Fig. 5). Thus, eating 6 g of DEX or αCD for seven days did not significantly affect the microbiota composition in the human colon. The results obtained in the human intervention study coincided with the findings in the KUHIMM.

Discussion
The aim of the present study was to investigate the impact of a low dosage (0.2%) of prebiotics (DEX, αCD, or DXR) on human colonic microbial ecology and metabolic end-products using an in vitro batch fermentation system, KUHIMM. We mimicked a daily intake of 6 g of prebiotics, assuming that the total Japanese human daily intake of foods and beverages is 3000 g 14 . Our in vitro and in vivo experiments suggest that a low dosage of prebiotics did not significantly affect the colonic microbiota composition. This contrasts with previous reports where a higher intake of prebiotics such as wheat dextrin (10-20 g/day in humans) 24,25 , αCD (1.5% in mice) 9 , and DXR (1-10% in an in vitro system) 26,27 substantially changed the colonic microbiota composition. However, addition   of prebiotics reduced the pH and promoted the generation of acetate and propionate even at low concentrations, as revealed by the in vitro experiments. This increase in acetate and propionate is in line with previous findings on the administration of wheat dextrin 28 , αCD 29 , and DXR 26 . Bacteria belonging to the phylum Bacteroidetes, which make up a large proportion of the intestinal microbiota, are known to produce mainly propionate along with acetate [30][31][32] . The phylum Bacteroidetes harbours a very broad saccharolytic potential 33 and plays an important role in fermenting indigestible carbohydrates in the proximal colon 30 . Here, the relative abundance of this phylum correlated positively with the prevalence of propionate among total SCFAs ( Supplementary Fig. S3), as observed previously in human faeces 34,35 . In addition, bacteria belonging to the genus Bifidobacterium, which were abundant in the phylum Actinobacteria in our system, are known to produce acetate 36,37 . Not surprisingly, the relative abundance of Actinobacteria showed a positive correlation with the proportion of acetate among total SCFAs ( Supplementary  Fig. S3). However, the cell abundance of phyla Bacteroidetes and Actinobacteria did not show a significant difference between control cultures and cultures with prebiotics, considering that the total number of eubacteria and relative abundance of these phyla were similar in the cultures, although cell abundance reportedly correlates with metabolic activity, i.e., rates of fermentation 34 . Therefore, it was expected that prebiotics such as DEX, αCD, and DXR could potentially increase the metabolic activity of bacteria belonging to the phyla Bacteroidetes and Actinobacteria, irrespective of the lack of increase in growth at low dosage amounts. In contrast, bacteria belonging to the phylum Firmicutes are known to produce butyrate as the primary metabolic end-product 30 . Thus, the relative abundance of the phylum Firmicutes correlated positively with the proportion of butyrate among total SCFAs (Supplementary Fig. S3). Low amounts of the tested prebiotics had no significant effect on the metabolic activity of bacteria belonging to the phylum Firmicutes.
Epithelial cells and immune cells are missing in in vitro models 15 . Addition of enterocytes and/or immune cells to an in vitro model will broaden the current understanding of the effect of different microbial metabolite patterns on host cells. In summary, the KUHIMM was able to capture an increase in SCFA production due to increased metabolic activity of human colonic microbiota, with no accompanying alteration in microbial composition, in response to a low-dosage supply of prebiotics. These results suggest that the positive effect of prebiotics seen in vitro using our human colonic microbiota model can be achieved also in vivo in human subjects upon administration of 6 g of prebiotics per day, although increased SCFA production is difficult to detect in the human intestine. This amount represents a realistic intake and does not pose a burden on a person's eating habits.

Materials and Methods
Faecal sample collection from human volunteers. Faecal samples were obtained from eight Japanese healthy human volunteers, who had not been treated with antibiotics for more than two months prior to the experiment. All participants were recruited according to the inclusion criteria, which comprised an age of 20 to Four human volunteers (designated as F26, M37, M38, and M39) participated in this study. In one subject, two faecal samples (designated as washout-1 (plot 1) and washout-2 (plot 3)) were collected after the subjects drank water for 14 days and another two faecal samples (designated as αCD (plot 2) and DEX (plot 4)) were collected after the subjects consumed α-cyclodextrin or indigestible dextrin, respectively, for seven days. In addition, PCoA plots, which were obtained from the KUHIMM without prebiotics (CUL (plot b)) and with indigestible dextrin (DEX (plot c)) or α-cyclodextrin (αCD (plot d)) (already shown in Fig. 2), are also presented. Faecal samples from the subjects participating in the human intervention study were used to inoculate the cultures in the KUHIMM.
Scientific REPORTS | (2018) 8:435 | DOI:10.1038/s41598-017-18877-8 65 years, being Japanese, non-smoking status, good health and physical condition. The exclusion criteria included significant clinical deviation from normal as determined by investigators; history or suspicion of diabetes, liver disease, kidney disease, or having a food allergy; or taking supplemental dietary fibre or lipid-lowering medications. Volunteers were designated as F26 (female, age 26), F40 (female, age 40), F63 (female, age 63), M27 (male, age 27), M37 (male, age 37), M38 (male, age 38), M39 (male, age 39), and M60 (male, age 60). All subjects provided written informed consent prior to specimen collection. 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 Declaration of Helsinki. The authors have no financial or personal relationships that could inappropriately influence this research.
Faecal samples were immediately collected in an anaerobic culture swab (212550 BD BBL Culture Swab; Becton, Dickinson and Company, Franklin Lakes, NJ, USA) and used within 24 h.
Operation of the KUHIMM with and without prebiotics. The KUHIMM was operated using a multi-channel fermenter (Bio Jr.8; ABLE, Tokyo, Japan), as described previously 13,21 with some modifications. The KUHIMM consisted of eight parallel and independent vessels. Each vessel contained 100 mL of Gifu Anaerobic Medium (GAM; Nissui Pharmaceutical Co., Ltd., Tokyo, Japan). The medium was autoclaved at 115 °C for 15 min and the initial pH was adjusted to 6.5. Anaerobic conditions in the vessel were achieved by purging with a mixture of N 2 and CO 2 (80:20; 15 mL/min) that was filter-sterilised through a 0.2-µm PTFE membrane (Pall Corporation, Port Washington, NY, USA) at 37 °C for 1 h prior to cultivation. To prepare the inoculum, the faecal sample in the swab was suspended in 2.0 mL of 0.1 M phosphate buffer (pH 6.5, consisting of 0.2 M NaH 2 PO 4 and 0.1 M Na 2 HPO 4 ) supplemented with 1.0% L-ascorbic acid (Wako Pure Chemical Industries, Osaka, Japan).
Cultivations were initiated by inoculating one faecal suspension (100 µL) into each vessel. During fermentation at 37 °C, the culture broth was stirred at 300 rpm with a magnetic stirrer and continuously purged with a filter-sterilised mixture of gas to maintain anaerobic conditions. Aliquots (1 mL) of culture broth were sampled from the vessel at 30 h after the initiation of cultivation. Faeces and culture broth samples were stored at −20 °C until use.
To evaluate the effect of prebiotics (Table 1), one type of prebiotics was added into one of the vessels at a final concentration of 2.0 g/L (0.2% per 100-mL vessel) prior to fermentation. The following prebiotics were used: DEX (Fibersol-2; Matsutani Chemical Industry Co., Ltd., Hyogo, Japan), αCD (Cyclochem Co., Ltd., Hyogo, Japan), and DXR (Dextran 40,000; Wako Pure Chemical Industries). A control vessel without prebiotics was prepared.
Human intervention study. All participants were recruited according to the inclusion criteria, which comprised an age of 20 to 50 years, being Japanese, non-smoking status, good health and physical condition, and a fasting triglyceride level of 120 mg/dL to 199 mg/dL. The exclusion criteria included significant clinical deviation from normal as determined by investigators; history or suspicion of diabetes, liver disease, kidney disease, or having a food allergy; or taking supplemental dietary fibre or lipid-lowering medications. A human intervention study was conducted on four volunteers (F26, M37, M38, and M39) ( Supplementary Fig. S2). At first, the volunteer drank water (200 mL) twice a day for 14 days (days 1-14, washout-1), and faeces were sampled using the swab on day 14. Then, the volunteer ate 3.0 g of αCD dissolved in water (200 mL) twice a day (6.0 g/day) for seven days (days [15][16][17][18][19][20][21], and faeces were sampled on day 21. Next, the volunteer drank water (200 mL) twice a day for 14 days (days 22-35, washout-2) again, and faeces were sampled on day 35. Thereafter, the volunteer ate 3.0 g of DEX dissolved in water (200 mL) twice a day (6.0 g/day) for seven days (days [36][37][38][39][40][41][42], and faeces were sampled on day 42. During the study, the four volunteers avoided drinking alcohol or consuming other dietary fibres. 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 Declaration of Helsinki and the approved guidelines by the Medical Ethics Committee at Kobe University (research code; 290001, approved date; 21 Mar 2017). The authors have no financial or personal relationships that could inappropriately influence this research.
Extraction of microbial genomic DNA. Microbial genomic DNA was extracted from suspended faeces and culture broth from the KUHIMM at 30 h, as described previously 21 . Purified DNA was eluted into TE buffer (10 mM TrisHCl, 1.0 mM EDTA) and stored at −20 °C until use.
Illumina library generation. Bacterial 16S rRNA genes (V3-V4 region) were amplified using genomic DNA as template and primers S-D-Bact-0341-b-S-17 (5′-CCTACGGGNGGCWGCAG-3′) and S-D-Bact-0785-a-A-21 (5′-GACTACHVGGGTATCTAATCC-3′) 38 , as described previously 13 . Index primers (Nextera XT Index Kit; Illumina Inc., San Diego, CA, USA) overhanging the amplified sequences were added to the gene-specific sequences. The PCR reaction was performed according to the manufacturer's instructions. Amplicons were purified with AMPure XP DNA purification beads (Beckman Coulter, Brea, CA, USA) and eluted in 25 µL of 10 mM Tris (pH 8.5). Purified amplicons were quantified using an Agilent Bioanalyzer 2100 with DNA 1000 chips (Agilent Technology, Santa Clara, CA, USA) and Qubit 2.0 (Thermo Fisher Inc., Waltham, MA, USA), and pooled in equimolar concentration (5 nM). The 16S rRNA genes along with an internal control (PhiX control v3; Illumina) were subjected to paired-end sequencing using a Miseq sequencer (Illumina) with a Miseq reagent kit v3 (600 cycles; Illumina). The PhiX sequences were removed, and paired-end reads with Q scores ≥ 20 were joined using the software package MacQIIME version 1.9.1 39 . The UCLUST algorithm 40 was used to cluster filtered sequences into OTUs based on a ≥97% similarity threshold. Chimeric sequences were checked and removed from the library using ChimeraSlayer 41 . Representative sequences from each OTU were taxonomically classified via the GreenGenes taxonomic database using the Ribosomal Database Project Classifier 42 . OTUs were