Discrepancies in microbiota composition along the pig gastrointestinal tract between in vivo observations and an in vitro batch fermentation model.

ABSTRACT

In vitro fermentation models are increasingly used to assess prebiotic potential of novel indigestible carbohydrates (CHO). A trial was performed to assess the validity of such approaches by comparing the influence of fermentation of inulin and cellulose on microbiota in vivo and in vitro. Two semipurified diets based on 5% inulin or 5% cellulose were fed to 2 groups of four 25-kg pigs. After 3 wk, the pigs were slaughtered and digesta was sampled from jejunum, ileum, cecum, and 3 parts of the colon to measure pH and microbiota population. An in vitro gas fermentation test was also performed on inulin and cellulose using fresh feces of the experimental pigs as bacterial inoculum. The gas production kinetics were modeled and fermentation broth sampled after 5, 8, 12, 24, and 72 h. Bacterial DNA was extracted and quantitative PCR was performed to quantify total bacteria, lactobacilli, bifidobacteria, Bacteroides, Clostridium cluster I, and Escherichia coli. Total bacteria quantification was similar between both systems. In vivo, total bacteria increased (P < 0.001) along the gut until the second part of the colon (from 10.5(7) to 10(10) cfu/mg) and then decreased (P < 0.05) to 10(9) cfu whereas in vitro, it increased (P < 0.05) until 12 to 24 h of fermentation (from 10(9) to 10.5(9) cfu/mL) and then decreased (P < 0.05) to initial level (10(9) cfu/mL). This evolution was consistent with fermentation kinetics. In both models, inulin increased (P < 0.05) the ratio of bifidobacteria and E. coli populations in the total microflora compared to cellulose. However, in vivo this was observed only in the first parts of the gut whereas in vitro the effect lasted for 72 h. Inulin also increased (P < 0.001) Bacteroides genus in vitro but not in vivo where the evolutions of Bacteroides were similar (P > 0.05) for both CHO. Evolutions of lactobacilli and Clostridium populations in both systems were also not consistent. This can be ascribed to specific bacterial properties as, for example, adhesive properties or sensitivity to sulfur reducing agent used in the in vitro model. As is, the in vitro model does not reflect properly changes in microbiota along the digestive tract induced by specific feed ingredients compared to in vivo observations.