Effects of Different Carbohydrates on the Simulation of Human Intestinal Bacterial Flora with in Vitro Culture

Objective: To investigate the optimal growth condition of human fecal bacterial flora in vitro by comparing the effect of different carbohydrates as cultural media. Methods: Three fecal samples (1, 2, 3) were collected and inoculated into a single-stage chemostat system, in which starch medium (VI) and starch polysaccharide medium(XP) were used. Samples were collected for bacterial genomic DNA extraction and polymerase chain reaction-denaturing gradient gel electrophoresis (PCR-DGGE) analysis. Bacterial composition and short chain fatty acid (SCFA) were then analyzed. Results: The single stage chemostat system reached steady after operating 8 days, when evaluated by the PCR-DGGE. Bacterial 16s rRNA high-throughput sequencing showed that the intestinal bacteria of these three volunteers was mainly composed of four bacterial phyla, namely, Bacteroidetes, Firmicutes, Proteobacteria and Actinobacteria. When the influence of bacterial abundance was considered, the similarity of bacterial composition between the original fecal samples to the harvested flora after culture was 0.847, 0.825, 0.968 in VI medium and 0.927, 0.926, 0.836 in XP medium, respectively. The similarity was decreased to 0.553, 0.580, 0.623 with VI medium and 0.617, 0.520, 0.574 with XP medium, when the number of bacterial species was considered. The variation of host individual also influenced the simulation. VI medium favored fecal sample 3, while XP medium more benefited sample 1 and 2. Bacteroides and Lachnospiraceae_incertae_sedis grew in both VI and XP medium. However, some species were only detected in VI medium and some were specifically found in the XP medium. The SCFA concentration in fermenters was 15-35 mmol/L, mainly propionate and butyrate. Conclusions: The chemostat system works for stimulating human gut bacterial flora in vitro. The bacterial composition is affected by different carbohydrate in the culture medium yet with close simulation higher than 80%