An extended reconstruction of human gut microbiota metabolism of dietary compounds

Abstract

Understanding how diet and gut microbiota interact in the context of human health is a key question in personalized nutrition. Genome-scale metabolic networks and constraint-based modeling approaches are promising to systematically address this complex problem. However, when applied to nutritional questions, a major issue in existing reconstructions is the limited information about compounds in the diet that are metabolized by the gut microbiota. Here, we present AGREDA, an extended reconstruction of diet metabolism in the human gut microbiota. AGREDA adds the degradation pathways of 209 compounds present in the human diet, mainly phenolic compounds, a family of metabolites highly relevant for human health and nutrition. We show that AGREDA outperforms existing reconstructions in predicting diet-specific output metabolites from the gut microbiota. Using 16S rRNA gene sequencing data of faecal samples from Spanish children representing different clinical conditions, we illustrate the potential of AGREDA to establish relevant metabolic interactions between diet and gut microbiota.

Fig. 1. Summary of the reconstruction pipeline.

First, AGORA reconstructions (black) are merged into a single compartment. Duplicated reactions were deleted but the taxonomic assignment was kept. For example, the same reaction r_A in taxon 1 (T1) and taxon 3 (T3) in AGORA, r_A1 and r_A3,  were converted to only one reaction in AGREDA and its associated Boolean rule, T₁|T₃, which we term Taxonomy-Reaction (TR) rule. Next, the Model SEED reactions (green) are annotated to AGORA species through EC number information (see “Methods” section). Then, metabolites provided by i-Diet and manually curated literature knowledge are integrated with AGORA and the Model SEED (maroon). Finally, gap-filling techniques and single-species analysis, based on the Cobra Toolbox^,, are applied to derive AGREDA.

Fig. 2. Main features of AGREDA.

a Boxplots depict for each phylum the number of metabolites and reactions of its associated strains in AGORA (red) and AGREDA (blue). Bottom and top of the boxes denote the first and third quartiles, respectively, and whiskers represent the values within 1.5 interquartile range above and below the box. Center line represents the median value. The number of strains per phylum is shown in brackets. Blocked reactions and metabolites were excluded from metabolic models derived from AGORA and AGREDA; b Distribution of the 179 phenolic compounds added by AGREDA separated in 19 families. c Degradation capabilities for three families of phenolic compounds present in AGREDA. The total number of strains in each phylum is reported in brackets; d Other families of compounds in the diet included in AGREDA and AGORA.

Fig. 3. Nutritional composition of 20 representative recipes of the Mediterranean diet in AGREDA and AGORA.

a The number of input dietary nutrients that AGORA (red) and AGREDA (blue) capture for different recipes (R1, R2, …, R20). Note that all the metabolites present in AGORA are also included in AGREDA. b Differences between the nutritional content of the recipes captured by AGORA and AGREDA, respectively. The Jaccard’s distance between the compositions of the recipes is represented.

Fig. 4. In vitro experimental comparison of the predictions by AGREDA and AGORA.

a Comparison of AGREDA and AGORA for predicting the presence (positive) or absence (negative) of ten output microbial compounds derived from the fermentation of lentils with children feces and measured with a targeted metabolomics approach. b Confusion matrix and statistical details of the comparison shown in a. Sensitivity was determined as TP/(TP+FN), specificity as TN/(FP+TN), and accuracy as (TP+TN)/(TP+TN+FP+FN). The reported Fisher’s p value was two-sided. 34dhpgval 5-(3’,4’-Dihydroxyphenyl)-gamma-valerolactone, 3hpppn 3-(3-hydroxy-phenyl)propionate, 4hphac 4-hydroxyphenylacetate, 34dhpha (3,4-dihydroxyphenyl)acetate, CPDIM-6116 dihydrocaffeic acid; “AFF2,” “AFF4”, and “AFF7” denote samples 2, 4, and 7 from children allergic to cow’s milk, respectively; “CFF1,” “CFF4,” and “CFF7” denote samples 1, 4, and 7 from celiac children, respectively; “LFF1,” “LFF4,” and “LFF7” denote samples 1, 4, and 7 from lean children, respectively; “OFF1,” “OFF4,” and “OFF7” denote samples 1, 4, and 7 from obese children, respectively; TP true positives, TN true negatives, FP false positives, FN false negatives.

Fig. 5. AGREDA-predicted production of three output microbial metabolites across different recipes and clinical conditions.

Each entry in the heatmap represents the proportion of samples of a particular clinical condition where a 4-methoxyphenylacetic acid, b sesamolin, and c pyrogallol are produced. Seven samples were used for celiac, lean, and obese children, while 6 samples were used for children allergic to cow’s milk. The analysis was done for 20 different recipes (R1, R2, …, R20) and the minimal growth medium (denoted REF).