Several validated dynamic in vitro models of the colon have been developed for humans, but there is no dynamic in vitro fermentation model for pigs. This study was conducted to modify the human, dynamic, computer-controlled TNO in vitro model of the colon (TIM-2) for pigs and investigate effects of different starch sources and polysaccharides on swine microbiota structure, ecological network, predictive functional profile, and short-chain fatty acids production. Our study showed that three different types of starch or two polysaccharides greatly impacted microbiota composition. Co-occurrence network analysis indicated that microbiota fed with different sources of starch changed the network topological properties. Functional profiles were predicted to vary significantly among the three starch treatments, and the original pig faecal inoculum was more similar to maize starch treatment. On the other hand, compared with maize starch and arabinoxylans (AX), the microbial composition of the original inoculum was more similar when AX-XG (arabinoxylans and xyloglucan) were added, and the functional profile of the original inoculum also clustered with AX-XG. The cumulative production of acetic, propionic, and butyric acid on maize starch were significantly higher than those on potato starch and wheat starch, while only the amount of acetic acid was significant higher on AX-XG than that on AX. In conclusion, supplementation of maize starch as the starch source together with AX and XG, leads to the bacteria being more stable in the in vitro model and closer to the original inoculum and microbial function compared to potato starch, wheat starch and AX. A maize basal diet may improve energy absorption in the large intestine in growing pigs.
Keywords: in vitro colon model; non-starch polysaccharide; pig; resistant starch; starch.