Diverse hydrogen production and consumption pathways influence methane production in ruminants

Abstract

Farmed ruminants are the largest source of anthropogenic methane emissions globally. The methanogenic archaea responsible for these emissions use molecular hydrogen (H₂), produced during bacterial and eukaryotic carbohydrate fermentation, as their primary energy source. In this work, we used comparative genomic, metatranscriptomic and co-culture-based approaches to gain a system-wide understanding of the organisms and pathways responsible for ruminal H₂ metabolism. Two-thirds of sequenced rumen bacterial and archaeal genomes encode enzymes that catalyse H₂ production or consumption, including 26 distinct hydrogenase subgroups. Metatranscriptomic analysis confirmed that these hydrogenases are differentially expressed in sheep rumen. Electron-bifurcating [FeFe]-hydrogenases from carbohydrate-fermenting Clostridia (e.g., Ruminococcus) accounted for half of all hydrogenase transcripts. Various H₂ uptake pathways were also expressed, including methanogenesis (Methanobrevibacter), fumarate and nitrite reduction (Selenomonas), and acetogenesis (Blautia). Whereas methanogenesis-related transcripts predominated in high methane yield sheep, alternative uptake pathways were significantly upregulated in low methane yield sheep. Complementing these findings, we observed significant differential expression and activity of the hydrogenases of the hydrogenogenic cellulose fermenter Ruminococcus albus and the hydrogenotrophic fumarate reducer Wolinella succinogenes in co-culture compared with pure culture. We conclude that H₂ metabolism is a more complex and widespread trait among rumen microorganisms than previously recognised. There is evidence that alternative hydrogenotrophs, including acetogenic and respiratory bacteria, can prosper in the rumen and effectively compete with methanogens for H₂. These findings may help to inform ongoing strategies to mitigate methane emissions by increasing flux through alternative H₂ uptake pathways, including through animal selection, dietary supplementation and methanogenesis inhibitors.

Fig. 1

Heatmap showing distribution of enzymes mediating H₂ production and H₂ consumption in orders of rumen microorganisms. Results are shown based on screens of the 501 genomes of cultured rumen bacteria and archaea (410 from the Hungate collection plus 91 other genomes). Partial hydrogenase sequences were also retrieved and classified from four rumen ciliates and two rumen fungi. The left-hand side of the heatmap shows the distribution of the catalytic subunits of enzymes that catalyse H₂ oxidation and production. These are divided into fermentative hydrogenases (H₂-producing; group A1, A2, B FeFe-hydrogenases), bifurcating hydrogenases (bidirectional; group A3, A4 FeFe-hydrogenases), respiratory hydrogenases (H₂-uptake; group 1b, 1c, 1d, 1f, 1i, 2d NiFe-hydrogenases), methanogenic hydrogenases (H₂-uptake; group 1k, 3a, 3c, 4h, 4i NiFe-hydrogenases, Fe-hydrogenases), energy-converting hydrogenases (bidirectional; group 4a, 4c, 4e, 4f, 4g NiFe-hydrogenases), sensory hydrogenases (group C FeFe-hydrogenases) and nitrogenases (H₂-producing; NifH). The right-hand side shows the distribution of the catalytic subunits of key reductases in H₂ consumption pathways. They are genes for methanogenesis (McrA, methyl-CoM reductase), acetogenesis (AcsB, acetyl-CoA synthase), sulfate reduction (DsrA, dissimilatory sulfite reductase; AprA, adenylylsulfate reductase; AsrA, alternative sulfite reductase), fumarate reduction (FrdA, fumarate reductase), nitrate ammonification (NarG, dissimilatory nitrate reductase; NapA, periplasmic nitrate reductase; NrfA, ammonia-forming nitrite reductase), dimethyl sulfoxide and trimethylamine N-oxide reduction (DmsA, DMSO and TMAO reductase) and aerobic respiration (CydA, cytochrome bd oxidase). Only hydrogenase-encoding orders are shown. Table S2 shows the distribution of these enzymes by genome, Figure S1 depicts hydrogenase subgroup distribution by class, and Table S1 lists the FASTA sequences of the retrieved reads

Fig. 2

Hydrogenase content in the metagenomes and metatranscriptomes of the microbial communities within rumen contents of high and low methane yield sheep. Hydrogenase content is shown based on hydrogenase subgroup a, b and predicted taxonomic affiliation c, d for metagenome data sets a, c and metatranscriptome data sets b, d. Hydrogenase-encoding sequences were retrieved from 20 paired shotgun metagenomes and metatranscriptomes randomly subsampled at five million reads. Reads were classified into hydrogenase subgroups and taxonomically assigned at the order level based on their closest match to the hydrogenases within the genomes screened (Fig. 1). L01 to L10 are data sets for sheep that were low methane yield at time of sampling, H01 to H10 are data sets from sheep that were high methane yield at time of sampling (see Table S3 for full details)

Fig. 3

Comparison of expression levels of H₂ production and H₂ uptake pathways in low and high methane yield sheep. Results are shown for 10 metatranscriptome data sets each from low methane yield sheep (orange) and high methane yield sheep (blue) that were randomly subsampled at five million reads. a Normalised count of hydrogenase transcript reads based on hydrogenase subgroup. b Normalised count of hydrogenase transcript reads based on predicted taxonomic affiliation. c Normalised count of transcript reads of key enzymes involved in H₂ production and H₂ consumption, namely the catalytic subunits of [NiFe]-hydrogenases (NiFe), [FeFe]-hydrogenases (FeFe), [Fe]-hydrogenases (Fe), hydrogenase-associated diaphorases (HydB), nitrogenases (NifH), methyl-CoM reductases (McrA), acetyl-CoA synthases (AcsB), adenylylsulfate reductases (AprA), dissimilatory sulphite reductases (DsrA), alternative sulfite reductases (AsrA), fumarate reductases (FrdA), dissimilatory nitrate reductases (NarG), periplasmic nitrate reductases (NapA), ammonia-forming nitrite reductases (NrfA), DMSO/TMAO reductases (DmsA) and cytochrome bd oxidases (CydA) are provided. For FrdA, NrfA and CydA, the numerous reads from non-hydrogenotrophic organisms (e.g., Bacteroidetes) were excluded. Each boxplot shows the 10 datapoints and their range, mean and quartiles. Significance was tested using independent two-group Wilcoxon rank-sum tests (*p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001; full p values in Table S9, S10 and S11). Note the metagenome abundance and RNA/DNA ratio of these genes is shown in Figure S2 (hydrogenase subgroup), Figure S3 (hydrogenase taxonomic affiliation) and Figure S4 (H₂ uptake pathways). A full list of metagenome and metatranscriptome hits is provided for hydrogenases in Table S4 and H₂ uptake pathways in Table S5

Fig. 4

Comparison of whole genome expression levels of Ruminococcus albus and Wolinella succinogenes in pure culture and co-culture. Pure cultures and co-cultures of Ruminococcus albus 7 a–c and Wolinella succinogenes DSM 1740 d–f were harvested in duplicate during mid-exponential phase and subject to RNA sequencing. a, d Volcano plots of the ratio of normalised average transcript abundance for co-cultures over pure cultures. Each gene is represented by a grey dot and key metabolic genes, including hydrogenases, are highlighted as per the legend. The horizontal dotted lines indicate q values of 0.05 and the vertical dotted lines indicate twofold changes. b, d Predicted operon structure of the three hydrogenases of R. albus and two hydrogenases of W. succinogenes. e Comparison of dominant fermentation pathways of R. albus in pure culture (left) and co-culture (right) based on transcriptome reads and metabolite profiling. Three enzymes with decreased expression in co-culture are in red font. f Respiratory chain composition of W. succinogenes in pure culture and co-culture based on transcriptome reads. Metabolite profiling indicated that the respiratory hydrogenase and fumarate reductases were active in both conditions. Proton translocation is thought to occur primarily through redox-loop mechanisms. A full list of read counts and expression ratios for each gene is provided in Table 1