Novel Polyfermentor intestinal model (PolyFermS) for controlled ecological studies: validation and effect of pH

Abstract

In vitro gut fermentation modeling offers a useful platform for ecological studies of the intestinal microbiota. In this study we describe a novel Polyfermentor Intestinal Model (PolyFermS) designed to compare the effects of different treatments on the same complex gut microbiota. The model operated in conditions of the proximal colon is composed of a first reactor containing fecal microbiota immobilized in gel beads, and used to continuously inoculate a set of parallel second-stage reactors. The PolyFermS model was validated with three independent intestinal fermentations conducted for 38 days with immobilized human fecal microbiota obtained from three child donors. The microbial diversity of reactor effluents was compared to donor feces using the HITChip, a high-density phylogenetic microarray targeting small subunit rRNA sequences of over 1100 phylotypes of the human gastrointestinal tract. Furthermore, the metabolic response to a decrease of pH from 5.7 to 5.5, applied to balance the high fermentative activity in inoculum reactors, was studied. We observed a reproducible development of stable intestinal communities representing major taxonomic bacterial groups at ratios similar to these in feces of healthy donors, a high similarity of microbiota composition produced in second-stage reactors within a model, and a high time stability of microbiota composition and metabolic activity over 38 day culture. For all tested models, the pH-drop of 0.2 units in inoculum reactors enhanced butyrate production at the expense of acetate, but was accompanied by a donor-specific reorganization of the reactor community, suggesting a concerted metabolic adaptation and trigger of community-specific lactate or acetate cross-feeding pathways in response to varying pH. Our data showed that the PolyFermS model allows the stable cultivation of complex intestinal microbiota akin to the fecal donor and can be developed for the direct comparison of different experimental conditions in parallel reactors continuously inoculated with the exact same microbiota.

Conflict of interest statement

Competing Interests: The authors have declared that no competing interests exist. The name of the PolyFermS has been registered at INPI (Institut National de la propriété industrielle, France) for identification of the system. ETH Zurich is the deponent. There are no patents or commercial interests.

Figures

Figure 1. Setup and design of the Polyfermentor Intestinal Model (PolyFermS).

Control (CR) and test reactors (TR) were continuously inoculated with effluents from inoculum reactors (IR) containing 30% (v/v) of fecal beads from donors A (model A), B (model B) and C (model C), respectively. Metabolites were quantified daily by HPLC analysis. Community structure was studied by HITChip analyses at selected time points t1, t2, t3 and t4. RT, mean retention time.

Figure 2. Composition of intestinal microbiota produced in effluents of PolyFermS models.

A. Similarity to the fecal donor. Open circles and straight lines correspond to single and mean similarity indices, respectively, based on Pearson product-moment correlation coefficients for HITChip fingerprints generated from fecal donor samples and PolyFermS model reactor effluents obtained from model A, B and C at time points t1 and t2 from all reactors and at t3 and t4 from IR and CR. B. Temporal diversity development. Open circles and lines correspond to single and mean Simpson’s reciprocal indices of diversity, respectively, calculated for effluent samples obtained from model A, B and C at time points t1 and t2 from all reactors and at t3 and t4 from IR and CR. Dashed lines indicate the Simpson’s reciprocal index of the corresponding fecal donor sample. A higher Simpson’s reciprocal index reflects a more diverse community, e.g. in terms of species richness and evenness.

Figure 3. Correlation between butyrate production and the fraction of predominant butyrate-producing bacteria in the total microbiota produced in effluents of PolyFermS models.

Open circles visualize the mean relative abundance (% of total flora) of predominant butyrate-producing bacteria (Faecalibacterium prausnitzii, Eubacterium rectale, Roseburia spp.) in IR, CR and TR of model A, B and C and colored bars indicate mean model-specific molar metabolite ratios (%) of acetate (in red), propionate (in green) and butyrate (in blue), with IR pH at 5.7 (t1) and 5.5 (t2).

Figure 4. Intra-model reproducibility of metabolic balance measured in effluent samples of PolyFermS models.

Molar metabolite ratios (%) measured daily in control reactors (CRA, CRB and CRB; x-axis) and test reactors (TRA, TRB and TRC; y-axis) from day 6 to 22. Mean daily ratios were calculated for TR1A, TR2A, TR3A (TRA). Numbers in color indicate model-specific Pearson correlation coefficients calculated for acetate (in red), propionate (in green) and butyrate (in blue).

Figure 5. Time stability of intestinal microbiota produced in effluents of PolyFermS models.

Single (open circles) and mean (line) similarity indices are based on Pearson product-moment correlation coefficients for HITChip fingerprints generated from PolyFermS reactor effluents obtained from model A, B and C at time points t1 and t2 from all reactors and at t3 and t4 from IR and CR compared to corresponding previous time points.