CoMiniGut-a small volume in vitro colon model for the screening of gut microbial fermentation processes

Abstract

Driven by the growing recognition of the influence of the gut microbiota (GM) on human health and disease, there is a rapidly increasing interest in understanding how dietary components, pharmaceuticals and pre- and probiotics influence GM. In vitro colon models represent an attractive tool for this purpose. With the dual objective of facilitating the investigation of rare and expensive compounds, as well as an increased throughput, we have developed a prototype in vitro parallel gut microbial fermentation screening tool with a working volume of only 5 ml consisting of five parallel reactor units that can be expanded with multiples of five to increase throughput. This allows e.g., the investigation of interpersonal variations in gut microbial dynamics and the acquisition of larger data sets with enhanced statistical inference. The functionality of the in vitro colon model, Copenhagen MiniGut (CoMiniGut) was first demonstrated in experiments with two common prebiotics using the oligosaccharide inulin and the disaccharide lactulose at 1% (w/v). We then investigated fermentation of the scarce and expensive human milk oligosaccharides (HMOs) 3-Fucosyllactose, 3-Sialyllactose, 6-Sialyllactose and the more common Fructooligosaccharide in fermentations with infant gut microbial communities. Investigations of microbial community composition dynamics in the CoMiniGut reactors by MiSeq-based 16S rRNA gene amplicon high throughput sequencing showed excellent experimental reproducibility and allowed us to extract significant differences in gut microbial composition after 24 h of fermentation for all investigated substrates and fecal donors. Furthermore, short chain fatty acids (SCFAs) were quantified for all treatments and donors. Fermentations with inulin and lactulose showed that inulin leads to a microbiota dominated by obligate anaerobes, with high relative abundance of Bacteroidetes, while the more easily fermented lactulose leads to higher relative abundance of Proteobacteria. The subsequent study on the influence of HMOs on two infant GM communities, revealed the strongest bifidogenic effect for 3'SL for both infants. Inter-individual differences of infant GM, especially with regards to the occurrence of Bacteroidetes and differences in bifidobacterial species composition, correlated with varying degrees of HMO utilization foremost of 6'SL and 3'FL, indicating species and strain related differences in HMO utilization which was also reflected in SCFAs concentrations, with 3'SL and 6'SL resulting in significantly higher butyrate production compared to 3'FL. In conclusion, the increased throughput of CoMiniGut strengthens experimental conclusions through elimination of statistical interferences originating from low number of repetitions. Its small working volume moreover allows the investigation of rare and expensive bioactives.

Keywords: 16S rRNA gene; Bifidogenic effect; Gut microbiome; Human milk oligosaccharide; In vitro colon model; Short chain fatty acids.

Figure 1. CoMiniGut model overview.

1. Climate Box, 2. Water bath, 3. Heating plate, 4. Temperature probe, 5. Temperature data logger, 6. Ventilators, 7. Lab-jack table, 8. Magnetic stirrer, 9. PMMA compartment, 10. Reaction vial, 11. Gas inlet, 12. Gas outlet, 13. Flow meters, 14. Nitrogen/gas tap, 15. pH probe and needle inlets, 16. Lid with septa, 17. Multichannel syringe pump, 18. pH meter, 19. Computer.

Figure 2. pH profiles of inulin and lactulose fermentations.

Fermentation pH profiles of controlled and uncontrolled inulin (A) and lactulose (B) (1% w/v) fermentations using fecal inocula from adult donors F1 and F2 (n = 4, line is mean value, shadow area is convex hull).

Figure 3. PCoA plot of 16S rRNA gene sequences.

PCoA score plot of 16S rRNA gene tag-encoded sequence reads based on weighted UniFrac distance metrics (n = 4, for all six experimental conditions).

Figure 4. Phyla (A) and SCFA (B) composition of the inulin and lactulose fermentations.

(A) Phyla composition (as determined by 16S rRNA gene amplicon sequencing) of pH controlled fermentations (24 h) of inulin and lactulose (1% w/v) using fecal inocula from adult donors F1 and F2 in CoMiniGut. ANOVA analysis of F1 and F2 inulin fermentations revealed significant differences (phylum level) in the relative abundances of Bacteroidetes, Proteobacteria and Firmicutes (p < 0.01), while no significant differences were detected between donors for Fusobacteria (p = 0.1) and Actinobacteria (p = 0.5). Variation between technical replicates was not significant (n = 4, F1 p = 0.4, F2 p = 0.4). For lactulose fermentations all phyla differed significantly between the two donors (p ≤ 0.05), while variation between technical replicates was not significant for both donors, (n = 4, F1 p = 0.6; F2 p = 0.8). (B) Relative concentrations of short chain fatty acid profiles of pH controlled CoMiniGut fermentations with inulin and or lactulose (1% w/v) using fecal inocula from adults donors F1 and F2 (n = 4).

Figure 5. Average Phyla composition of HMO fermentations.

Average Phyla composition, (relative % fraction of total) all fermentations for baby 1 and baby 2 with 3’SL, 6’FL, 6’SL and FOS as substrates as determined by 16S rRNA gene amplicon sequencing. 3’FL fermentations differed significantly in the relative abundance of assigned phyla between donors p ≤ 0.02, but no significant differences were found between technical replicates (n = 4, B1 p = 0.7, B2 p = 0.6). For 3’SL fermentations significant differences between donors were detected for Proteobacteria and Firmicutes p < 0.01, whereas no significant variation was detected between technical replicates (n = 4, B1 p = 0.8, B2 p = 0.6). For 6’SL fermentations, relative abundance were significantly different for all phyla p ≤ 0.05, whereas no significant variation was detected for technical replicates (n = 4, B1 p = 0.8, B2 p = 0.5). For FOS significant differences were detected for Bacteroidetes, Firmicutes and Proteobacteria p ≤ 0.05, whereas no significant differences were detected between technical replicates (n = 4, B1 p = 0.8, B2 p = 0.5). For the negative control no significant differences between phyla composition were detected between donors ≥0.3 and technical replicates (n = 4, B1 p = 0.6, B2 p = 0.3).

Figure 6. SCFA content of HMO fermentates.

Quantified SCFAs (µM) produced after 24 h of fermentation of 3’SL, 6’SL, 3’FL and FOS by GM of baby 1 and 2. Amounts displayed represent values after subtraction of SCFAs produced in the control fermentations.