In Engineering in life sciences
Recent studies have shown perturbed gut microbiota associated with gouty arthritis, a metabolic disease characterized by an imbalance between uric acid production and excretion. To mechanistically investigate altered microbiota metabolism associated with gout disease, 16S rRNA gene amplicon sequence data from stool samples of gout patients and healthy controls were computationally analyzed through bacterial community metabolic models. Patient-specific community models constructed with the metagenomics modeling pipeline, mgPipe, were used to perform k-means clustering of samples according to their metabolic capabilities. The clustering analysis generated statistically significant partitioning of samples into a Bacteroides-dominated, high gout cluster and a Faecalibacterium-elevated, low gout cluster. The high gout cluster was predicted to allow elevated synthesis of the amino acids D-alanine and L-alanine and byproducts of branched-chain amino acid catabolism, while the low gout cluster allowed higher production of butyrate, the sulfur-containing amino acids L-cysteine and L-methionine, and the L-cysteine catabolic product H2S. By expanding the capabilities of mgPipe to provide taxa-level resolution of metabolite exchange rates, acetate, D-lactate and succinate exchanged from Bacteroides to Faecalibacterium were predicted to enhance butyrate production in the low gout cluster. Model predictions suggested that sulfur-containing amino acid metabolism generally and H2S more specifically could be novel gout disease markers.
Henson Michael A
bacterial communities, gout, gut microbiota, machine learning, metabolic modeling