Genomics and metagenomics of trimethylamine-utilizing Archaea in the human gut microbiome

Abstract

The biological significance of Archaea in the human gut microbiota is largely unclear. We recently reported genomic and biochemical analyses of the Methanomassiliicoccales, a novel order of methanogenic Archaea dwelling in soil and the animal digestive tract. We now show that these Methanomassiliicoccales are present in published microbiome data sets from eight countries. They are represented by five Operational Taxonomic Units present in at least four cohorts and phylogenetically distributed into two clades. Genes for utilizing trimethylamine (TMA), a bacterial precursor to an atherosclerogenic human metabolite, were present in four of the six novel Methanomassiliicoccales genomes assembled from ELDERMET metagenomes. In addition to increased microbiota TMA production capacity in long-term residential care subjects, abundance of TMA-utilizing Methanomassiliicoccales correlated positively with bacterial gene count for TMA production and negatively with fecal TMA concentrations. The two large Methanomassiliicoccales clades have opposite correlations with host health status in the ELDERMET cohort and putative distinct genomic signatures for gut adaptation.

Figure 1

Maximum likelihood phylogenetic tree of the Methanomassiliicoccales showing the diversity of human-associated Methanomassiliicoccales. Sequences in red are from human-associated Methanomassiliicoccales and sequences in bold are obtained in this study; large arrows indicate draft genomes obtained in this study and small arrows indicate sequences corresponding to complete or draft genomes obtained in previous studies. Underlined sequences correspond to cultured representatives. Sequences of human-associated Methanobacteriales were used to root the tree. Mx-07 to Mx-09 sequences are not included due to too short sequences available for correct assignment but are given in Supplementary Information. Bootstrapping using 200 iterations was performed, black-filled dots indicate nodes with bootstrap supports >90%, gray-filled dot >70% and white-filled dot >50%.

Figure 2

Pan-genome of human-associated Methanomassiliicoccales. (a) Core, variable and unique gene families in the pan-genome of the human-associated Methanomassiliicoccales. Between brackets, the number of unique gene families and the percentage they represent within each genome. The number of unique genes and its percentage are also given for the pan-genome of ‘Ca. Mmc. intestinalis’ as the three genomes of this species are very close. Mint, ‘Ca. Mmc. intestinalis’ Malv, ‘Ca. Mmp. alvus’ Mlum, Mmc. luminyensis. (b) Rarefaction analysis of gene family discovery in the pan-genome of human-associated Methanomassiliicoccales, determined on the basis of identity percentage (dark green) and KO annotation (light green).

Figure 3

Microbial metabolisms modulating TMA concentrations in the human gut microbiome. (a) Schematic representation of the pathways and associated marker genes modulating TMA concentrations in the gut microbiome and health output related to high TMA productions. (b) Composition and carriage (%) of the potential TMA producers in the ELDERMET cohort. ‘Total’ corresponds to the percentage of carriage of each marker gene among the subjects, regardless of the affiliation of the marker. (c) Percentage of carriage of Methanomassiliicoccales representatives in subgroups of subjects having no TMA production pathway or one to four different TMA production pathways detected in their microbiota. (d) Concentrations of fecal TMA (logged) among subjects without or with Methanomassiliicoccales detected at different concentrations. No Mmass, no Methanomassiliicoccales detected; Mmass without mttB, Methanomassiliicoccales potentially lacking the capacity to deplete TMA; mttB Mmass <10⁸ and mttB Mmass >10⁸, Methanomassiliicoccales potentially depleting TMA having a concentration lower and higher than 10⁸ cell per g stool, respectively. Significant differences between groups are determined with Mann–Whitney test, *P<0.05, **P<0.001.

Figure 4

Prevalence and relative abundance of methanogens among elderly subjects living in community and according to the time spent in residential care. Prevalence and relative abundance are presented for Methanobrevibacter smithii (a and b), Methanosphaera stadtmanae (c and d), ‘Free-living clade’ Methanomassiliicoccales (e and f) and the ‘Host-associated clade’ Methanomassiliicoccales (g and h). In panels (a), (c), (e) and (g), the dark blue bars represent prevalence among subjects at T0, while the clear blue bars take into account subjects positive for at least one time point. For prevalence, Community (Com), n=125; Residential care 0–2 months (mth), n=61; 2–12 months, n=47; >1 year, n=51. Statistical analysis and representation of the relative abundance of methanogens were carried out on subjects with a positive detection of methanogens. *P<0.05, **P<0.01, ***P<0.001. Day hospital category is not presented for clarity but is never different from Community.

Figure 5

Occurrence of genes encoding for Flg_new and Sel1_repeat domains in Methanomassiliicoccales and other genomes from the three domains of life. Flg_new domains (a) and Sel1 repeat domains (b). Genomes are ordered from the highest occurrence to the lowest occurrence of the genes coding those domains.