Impact of Different Fecal Processing Methods on Assessments of Bacterial Diversity in the Human Intestine

Yu-Hsin Hsieh, Courtney M. Peterson, Anne Raggio, Michael J. Keenan, Roy J. Martin, Eric Ravussin, Maria L. Marco

The intestinal microbiota are integral to understanding the relationships between nutrition and health. Therefore, fecal sampling and processing protocols for metagenomic surveys should be sufficiently robust, accurate, and reliable to identify the microorganisms present. We investigated the use of different fecal preparation methods on the bacterial community structures identified in human stools. Complete stools were collected from six healthy individuals and processed according to the following methods: (i) randomly sampled fresh stool, (ii) fresh stool homogenized in a blender for 2 min, (iii) randomly sampled frozen stool, and (iv) frozen stool homogenized in a blender for 2 min, or (v) homogenized in a pneumatic mixer for either 10, 20, or 30 min. High-throughput DNA sequencing of the 16S rRNA V4 regions of bacterial community DNA extracted from the stools showed that the fecal microbiota remained distinct between individuals, independent of processing method. Moreover, the different stool preparation approaches did not alter intra-individual bacterial diversity. Distinctions were found at the level of individual taxa, however. Stools that were frozen and then homogenized tended to have higher proportions of Faecalibacterium, Streptococcus, and Bifidobacterium and decreased quantities of Oscillospira, Bacteroides, and Parabacteroides compared to stools that were collected in small quantities and not mixed prior to DNA extraction. These findings indicate that certain taxa are at particular risk for under or over sampling due to protocol differences. Importantly, homogenization by any method significantly reduced the intra-individual variation in bacteria detected per stool. Our results confirm the robustness of fecal homogenization for microbial analyses and underscore the value of collecting and mixing large stool sample quantities in human nutrition intervention studies.