Researchers report on a genome-wide reconstruction of human microbes to develop personalized therapies

By Pooja Toshniwal PahariaReviewed by Aimee MolineuxJan 26 2023

In a recent study published in Nature Biotechnology, researchers reported on AGORA2, an expansion of their previously published resource, AGORA, both resources of genomewide metabolic reconstructions of human intestinal microbes.

Human intestinal microbes synthesize biological metabolites, including neurotransmitters, hormones, and short-chain fatty acids, and metabolize several medications, resulting in the activation or inactivation, and detoxification of the drugs. Precision therapies accounting for genetics, microbes, and diet need an in-depth understanding of drug metabolism by intestinal microbes, gut microbiota taxa, and the stoichiometry of drug transformations. The AGORA resource, previously reported by the authors, comprises 773 genomewide reconstructions of 65 human intestinal microbes of 14 phyla.

About the study

In the present study, researchers presented AGORA2, comprising metabolically reconstructed genomes, for the precise capture of microbe-specific potentials.

Newly reconstructed microbes were selected, and their full-genomic sequences of 4,185 human intestine microbial strains were retrieved from the PubSEED database. In addition, 1,324 microbial strains across 127.0 genomes of mice-related strains identified from a literature search were included. Moreover, 26.0 genomes of the strains of Eggerthella lenta, retrieved from the national center for biotechnology information (NCBI) database, and 761.0 human microbiota genomes available in HBC (Human Gastrointestinal Bacteria Culture Collection) were included.

Comparative genomics analysis was performed to refine metabolic pathways and genetic annotations manually. Following this, 34 metabolic subsystems were curated, and sequence-based gap-filling was performed using the bidirectional best-hit (BBH) approach. Drug metabolic genes were annotated, the PubMed database was searched, and data were obtained from 732 papers and greater than 8,000.0 pages of relevant textbooks. The strain taxonomy was classified according to the NCBI database taxonomy. Data on habitat, morphology, gram status, size, metabolism, oxygenation status, genomic size, and motility were retrieved manually from the IMG/M (integrated microbial genomes and microbiomes) database.

Subsequently, draft genomic reconstructions were prepared using the KBase database. The DEMETER (data-driven metabolic network refinement) semiautomated, data-driven refinement pipeline was revised, quality control reports were generated, and 15 microbial enzymes (retrieved from a literature search), encoded by 25 genes were analyzed. The structures of 287 drug metabolites and drug degradation products were retrieved from 73 peer-reviewed papers, the human metabolome (HMDB) database, DrugBank, and the Transformer database. For a total of 98 medications, 353 metabolites, 381 enzyme-involving reactions, 373 interchange-type reactions and 710 transportation-type reactions were formulated.

Atom-atom mapping analysis was performed for 5,583 (65.0%) reactions captured, and molecular simulations were performed. Seventy-two fully manually designed reconstructions were downloaded from the BiGG database, and reconstructions were generated. Flux and stoichiometrically consistent reactions were assessed, and the results were validated using datasets such as NJC19, the BacDive database, and the dataset of the study by Madin et al. Drug yields were assessed, and shadow price analysis was performed. In addition, a dataset of 365 Japanese CRC (colorectal cancer) patients and 251 healthy individuals were included, and community modelling was performed for estimating drug metabolizing potential. Further, a sensitivity analysis was performed by recomputing the drug-metabolizing capacities based on European diets rather than Japanese diets.  

Results

AGORA 2 comprised microbial reconstructions for 7,302 intestinal microbial strains, 1,738 species and 25 phyla, showing a quality control score of 73% and high accuracy against the experimental independent datasets. The findings indicated that AGORA2 reconstructions can be used independently or together for investigating microbial metabolism and host–microbiota co-metabolism in silico. The models derived from AGORA2 capture taxon-specific metabolic traits of the reconstructed microorganisms.

The AGORA2 resource reconstructions captured established features of microbes well, out-competing other semi-automated reconstructions of microbial strains, and were comparable to manually curated reconstructions. AGORA2 performed especially well in terms of estimating the intake and secretion of drug metabolites. All 5,438 analyzed microbial strains carried ≥1.0 drug-metabolizing enzyme. The estimated drug-metabolizing potentials estimated were validated against the experimental datasets for 253 drug–microorganism pairs. The findings indicated that the transporters and enzymes involved in drug metabolism are distributed widely, with critical strain- and phyla-specific differences.

Notably, 97.0% of the microbes were mapped using AGORA2 (in comparison with 72.0% using the AGORA resource). For enzymes such as dopamine dehydroxylase and dihydropyrimidine dehydrogenase, the drug transformation ability correlated less with the profuseness of drug-metabolism reactions, indicating metabolic roadblocks limited by flux. The in silico analysis showed that the majority of the medications were metabolized qualitatively by ≥95.0% of the microbiota; however, only 53.0% metabolized digoxin, and 86.0% and 46.0% metabolized levodopa to generate dopamine, and tyramine, respectively, and required Eggerthella lenta for the reactions.

All except three microorganisms showed activated balazide against inflammatory bowel disorder via azoreductase activities. The example of balsalazide showcases the use of AGORA2 for informing research and facilitating tailored therapy. The sensitivity analysis findings indicated that the capabilities of drug metabolism were unaltered for the drugs, and therefore, robust for dietary limitations. However, considerable individual differences in drug-metabolizing capacity, irrespective of illness status, existed due to distinctive microbial compositions. AGORA2-based community modeling could estimate the direction of species metabolite associations for several metabolites.

Overall, the study findings highlighted AGORA2 as a resource for estimating the drug conversion capability of intestinal microbes, and developing personalized, intestinal microbial drug metabolism therapies for colorectal cancer patients and control individuals.