Background and Objectives: Dietary components are differentially susceptible to digestion in the small intestine, contributing to variation in residual dietary composition along the gastrointestinal (GI) tract. Thus, quantifying diet as consumed orally does not capture which fraction of diet reaches and structures the gut microbiome, which is densest and most studied in the colon at the end of the GI tract. Here, we use DNA metabarcoding approaches to index dietary composition at it appears in the colon using fecal samples from humans and mice fed matched diets varying in their small intestinal digestibility due to cooking, compare colonic dietary composition to diet as consumed orally, and investigate relationships between dietary and microbial taxonomic composition in the colon.

Methods: In a cross-sectional study, mice were fed one of six plant substrates in either raw or cooked form for four days (n=48). In a counterbalanced crossover study, humans were fed mixed plant-based meals that were compositionally identical but served either raw or cooked for three days each (n=10). DNA was isolated from daily fecal samples collected during the dietary intervention, the 16S rRNA microbial marker gene was sequenced to quantify gut microbial composition, and the trnL chloroplast marker gene was assayed via sequencing and qPCR to index the relative taxonomic composition and absolute abundance of plant DNA, respectively.

Results: Mice fed cooked plants had fewer trnL genes per gram of feces than mice fed the same plants in raw form (all p<0.05). Human participants likewise had fewer trnL genes per gram of feces when consuming cooked diets than when consuming raw diets (F=4.717, p=0.037). In humans fed mixed plant-based diets, plant DNA in fecal samples was biased towards substrates with low digestibility, such as chia seeds. Despite being fed identical diets, both murine hosts and human hosts significantly varied in the fraction of plant DNA mapping to specific taxa, and co-analyzing plant relative abundance and microbial relative abundance in fecal samples revealed associations between nutrients and specific microbial taxa. For example, in mice we found inverse associations between log₁₀ relative abundances of DNA from soy and from the microbial genus Bacteroides, a known metabolizer of soy isoflavone (F= 5.78, p= 0.018), and between log₁₀ relative abundances of DNA from wheat and from the microbial family Muribaculaceace, which are complex-carbohydrate degraders  (F=6.46, p=0.013).

Conclusions: DNA metabarcoding can capture which dietary components reach and structure the gut microhome as well as differential digestibility across hosts, dietary information missed by traditional macronutrient-based nutritional assays. Our data confirm that the fecal density of plant DNA is higher on raw versus cooked diets and biased towards substrates with low expected small intestinal digestibility. Further, we find that gut microbiome composition mediates relative amounts of residual dietary components in hosts fed identical diets, providing insights into both host and microbial dietary digestion.