Exploring the link between ruminal methane production and physiological changes in Japanese Black cattle during fattening

Abstract

Gastroenteric release of methane from livestock accounts for a substantial portion of anthropogenic greenhouse gas emissions worldwide. In this study, we investigated the rumen microbiome and physiological characteristics associated with methane production in Japanese Black cattle (n = 21). Methane emissions were measured during three fattening phases—early (13 months), middle (20 months), and late (28 months). Using a mixed-effects model to consolidate period-specific measurements, animals were classified into high- and low-methane emitting groups based on residual methane emission (RME). Methane emissions in the high-methane (HME; n = 6) and low-methane (LME; n = 6) groups were 261.6 and 216.5 L/day at T1, 276.9 and 201.7 L/day at T2, and 282.9 and 221.1 L/day at T3, respectively. Overall estimated methane emissions were 273.8 L/day in the HME group and 214.4 L/day in the LME group. Hepatic transcriptome profiles, blood metabolites and hormones, rumen fermentation parameters, and rumen microbiota were subsequently analyzed to identify physiological and microbial features associated with methane emission potential. Hydrogen-sinking microbes such as Anaerovorax and Succinivibrio were present at low levels, whereas the prevalence of hydrogen-producing microbes including Christensenellaceae, Clostridium methylpentosum, and Mogibacterium was high in HME. Functional profiling of rumen microbiota revealed decreased coenzyme M biosynthesis and an increased hydrogen sink from L-glutamate biosynthesis in LME. In the liver, glutamate-derived ornithine and elevated ornithine transcarbamylase gene expression increased ammonia detoxification in LME, whereas the glutamate transporter-encoding gene SLC1A1 was upregulated in HME. These ruminal and physiological changes have potential as biomarkers for monitoring the methanogenic potential of Japanese Black cattle and highlight that the upstream oxoglutarate-to-glutamate biosynthesis pathway is associated with methane production by reducing hydrogen in the rumen.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-026-36644-6.

Introduction

Methane (CH₄), a greenhouse gas, is 28–34 times more potent than carbon dioxide (CO₂)¹. It constitutes 14% of the total² and approximately 5% of the anthropogenic greenhouse gas emissions worldwide, primarily owing to the ruminal digestion of livestock³. Moreover, without proper management, the negative effects of enteric methane emissions from the livestock industry will likely worsen owing to the increasing demand for milk and meat fuelled by population growth and urbanisation. Therefore, coordinated efforts are essential to ensure animal production system sustainability and health. Ruminants, such as cows and sheep, have a long digestive tract that harbours diverse microbiota in their rumen. These microbiotas consist of numerous microorganisms, including bacteria, archaea, and fungi, which enable ruminants to convert otherwise indigestible plant components into high-quality protein products such as milk and meat. During this process, ruminal microorganisms break down plant polymers and produce volatile fatty acids (VFAs), hydrogen (H₂), and carbon dioxide as the primary fermentation by-products. The VFAs are absorbed along the ruminal epithelium, providing energy for host animal growth and production. However, as continuous ruminal fermentation requires hydrogen removal, methanogens metabolize hydrogen to facilitate CO₂ reduction to CH₄, which is then emitted from the body. This process consumes approximately 2–15% of the gross energy intake of ruminants⁴, resulting in energy loss and reduced metabolic efficiency. Therefore, understanding the mechanisms underlying methane generation is essential for improving energy utilization and mitigating environmental impacts.

Various methane mitigation strategies influence ruminal methanogenesis, including dietary manipulation, rumen control, and selective breeding^5–7. However, despite extensive feeding additive research, few techniques for reducing enteric CH₄ emissions are cost-effective, non-residual, non-toxic, and acceptable to farmers⁸. Alternatively, a promising supplement-free strategy for regulating ruminal methane generation is to divert hydrogen from methanogenesis to more nutritionally advantageous pathways; e.g. by reducing acetate and butyrate while increasing propionate production. Better understanding the causes of methane production and identifying biomarkers such as rumen fermentation characteristics, blood metabolites, and associated ruminal microorganisms are key to developing effective methane reduction strategies.

Japanese Black cattle are raised under unique feeding systems during the fattening period. They are provided with large amounts of concentrates to promote intramuscular fat accumulation, resulting in the highly marbled beef. Therefore, their ruminal environment and physiological characteristics likely differ from those of other beef cattle breeds. We previously demonstrated that Japanese Black cattle exhibit unique physiological traits at different stages of fattening⁹. However, the relationship between ruminal microbiota and methane emissions during the fattening period and an in-depth understanding of the physiological changes related to methane production in this breed remain to be elucidated. Thus, in this study, we analysed the physiological traits of Japanese Black steers raised using conventional livestock practices in Japan, examining the influence of ruminal microbiota on the gut–liver axis and blood metabolites. By providing comprehensive data on physiological features including blood metabolites, hepatic gene expression, and ruminal fermentation characteristics along with ruminal microbiota compositional and functional profiling, this study offers new insights that into methane emissions in Japanese Black cattle.

Results

Effect of methane production on the ruminal microbial community

Figure 1a and b show the dominant members of the bacterial and archaeal communities within the high-methane (HME) and low-methane-emitting groups (LME) groups at the phylum and genus levels. At the phylum level, Bacillota and Bacteroidota dominated the ruminal microbial community with relative abundances of 56.62% and 36.0%, respectively. Patescibacteria was the next most abundant bacterial phylum (2.83%), followed by Actinomycetota (1.32%), Methanobacteriota (1.01%), Planctomycetota (0.65%), and Spirochaetota (0.63%). Prevotella was the most abundant bacterial genus (20.21% mean relative abundance across all samples), followed by Acetitomaculum (3.88%), Muribaculum (3.97%), and Ruminococcus (3.10%). Figure 1c shows differentially abundant microbiota between the HME and LME groups across the entire fattening period (T1–T3), identified using a mixed-effects model at the phylum, family, and genus levels. The HME exhibited enriched Bacteroidota and Methanobacteriota, Prevotellaceae and Christensenellaceae, and Prevotella, Christensenellaceae R-7 group, and Methanobrevibacter, whereas LME showed enriched Bacillota, Succinivibrionaceae, Succinivibrio, Clostridia UCG-014, Eubacterium nodatum group, Methanosphaera, and Anaerovorax (all P < 0.05). Figure 1d and 1e present Venn diagrams illustrating shared and exclusive amplicon sequence variants (ASVs) in the ruminal microbial community with 100% prevalence. The HME group contained higher numbers of exclusive ASVs during each of the three fattening periods. Among the periods, exclusively detected ASVs were more abundant during the middle fattening period (T2) in the HME and during the late fattening period (T3) in the LME groups. As revealed by PCoA and permutational multivariate analysis of variance (PERMANOVA) in Fig. 1f, the overall microbial community structure differed between the two groups at T2 (PERMANOVA; P < 0.05).

Regression tree for methane emission in Japanese Black cattle

Figure 1 presents the regression tree analysis with methane emissions as the dependent variable. Ruminal microbes that were differentially abundant between HME and LME were used as independent variables in Fig. 1a. All cattle were split into subgroups based on Christensenellaceae abundance (Node 1); Nodes 2 and 3 represent cattle with an abundance of ≤ 6.209 and > 6.209, respectively. At the second tree depth, Node 3 was partitioned based on Methanosphaera abundance; Nodes 4 and 5 reflect ≤ 3.765 and > 3.765 abundance, respectively. At the third tree depth, Node 5 was further split based on Prevotellaceae abundance, with respective Node 6 and 7 abundances of ≤ 25.214 and > 25.214.

Fig. 1.

Dynamic changes of ruminal microbiota according to methane emission. A total of 62 measurements from the early (n = 21), middle (n = 20), and late fattening periods (n = 21) were used in the analysis. In each period, the HME and LME groups consisted of six animals (n = 6 per group). (a, b) Relative microbial abundance at the (a) phylum and (b) genus level. (c) Relative fold change of differentially abundant microbiota between the HME (n = 18) and LME (n = 18) groups across the entire fattening period (T1–T3) based on a mixed-effects model at the phylum, family, and genus levels. (d, e) Shared core ASVs with 100% prevalence between the HME and LME groups, and among T1, T2 and T3. (f) PCoA plots based on Bray–Curtis dissimilarity for comparing the HME and LME groups. HME, group of high methane-emission cattle; LME, group of low methane-emission cattle; PCoA, principal coordinates analysis; T1, early fattening period; T2, middle fattening period; T3, late fattening period. ***P < 0.001, **P < 0.01, and *P < 0.05, respectively.

Ruminal microbes with 100% prevalence were used as independent variables in Fig. 1b. All cattle were divided into subgroups based on Clostridium methylpentosum group abundance (Node 1); Nodes 2 and 5 represent cattle with abundance ≤ 0.307 and > 0.307, respectively. At the second tree depth, Node 2 was partitioned based on Mogibacterium abundance; Nodes 3 and 4 represent abundances of ≤ 0.16 and > 0.16. Node 5 was partitioned based on the abundance of the Lachnospiraceae NK3A20 group; Nodes 6 and 7 are indicative of ≤ 2.211 and > 2.211 abundance. At the third tree depth, Node 7 was further split based on Desulfobacteroita abundance, with Node 8 and 9 abundances of ≤ 0.03 and > 0.03, respectively.

Functional profiles of ruminal microbiota related to methane emission

Functional genetic profiling revealed a total of 398 MetaCyc pathways among the 62 ruminal microbiome samples across the entire fattening period. Various analyses were conducted to identify the MetaCyc pathways related to methane emissions levels (Fig. 2). Spearman’s correlation analysis (correlation coefficient, |r|≥ 0.5; P < 0.05) revealed P261-PWY and PWY-7431 as positively and negatively correlated with methane emission levels, respectively (Fig. 2a). All significantly differential MetaCyc pathways of rumen microbiota between the HME and LME groups are presented in Supplementary Table S1. Among these, PWY-5505 and GLUTORN-PWY gene expression levels, involved in glutamate metabolism, were significantly higher (P < 0.01) in the LME group (Fig. 2b). In a regression tree with methane emissions as the dependent variable and MetaCyc pathways as independent variables, all cattle were significantly split into subgroups based on P261-PWY pathway abundance (Node 1), with Nodes 2 and 5 representing cattle with abundance ≤ 0.016 and > 0.016, respectively (Fig. 2c). At the second tree depth, Node 2 was significantly partitioned based on PWY-5100 pathway abundance, with Node 3 and 4 abundances of ≤ 0.894 and > 0.894. Node 5 was significantly partitioned based on PWY-5695 pathway abundance, creating Nodes 6 and 7 at ≤ 0.754 and > 0.754 abundance, respectively.

Fig. 2.

The linear mixed-effects model regression tree for methane emission. A total of 62 measurements for the early (n = 21), middle (n = 20), and late fattening (n = 21) periods were used in the analysis. Methane emission (L/d) was used as the dependent variable. Ruminal microbiota (a) selected using the mixed-effect model analysis and (b) those with 100% prevalence were used as independent variables.

Link between methane emissions, blood metabolites, hepatic transcriptome, ruminal microbiota, and fermentation

To identify physiological factors and hepatic transcriptome differences between the HME and LME groups, data from all fattening stages (T1–T3) were analyzed using a mixed-effects model (Supplementary Table S2 and S3). The regression tree analysis utilized rumen fermentation characteristics, blood metabolites, blood amino acids, and hepatic genes as independent variables and methane emissions as the dependent variable (Supplementary Figs. 1 and 2). Ruminal fermentation characteristics (e.g. propionate, butyrate, and ammonia concentrations), blood metabolites (e.g. BHBA and ornithine), and hepatic genes [(e.g. solute carrier family 1 member 1 (SLC1A1), Ras-related protein Rab-6A (RAB6A), and ornithine transcarbamylase (OTC)] significantly influenced the methane emission levels. Based on the regression tree and ANOVA results, the potential correlation between methane emission level and physiological parameters was examined using a Bayesian network (Fig. 3). The tabu search algorithm produced 8 directed connections in LME and 6 in HME from their respective latent variables. HME-group methane emissions were potentially moderated by butyrate (arc strength = 0.63), BHBA (arc strength = 0.68), and PWY-5505 (arc strength = 0.64), whereas PWY-5505 (arc strength = 0.69) potentially moderated LME-group emissions.

Fig. 3.

MetaCyc pathways of rumen microbiota related to methane production. A total of 62 measurements for the early (n = 21), middle (n = 20), and late fattening (n = 21) periods were used in the analysis. (a) Metacyc pathways significantly correlated (|r| > 0.5, P < 0.05) with methane production based on Spearman’s rank correlation analysis. (b) Differentially abundant MetaCyc pathways between HME and LME groups based on a linear mixed-effect model. (c) Linear mixed-effects model regression tree for methane emission (L/d), using MetaCyc pathways as independent variables, incorporating 62 measurements across the early (n = 21), middle (n = 20), and late (n = 21) fattening periods. GLUTORN-PWY, L-ornithine biosynthesis; HME, group of high methane-emission cattle; LME, group of low methane-emission cattle; P261-PWY, coenzyme M biosynthesis; PWY-5100, pyruvate fermentation to acetate and lactate; PWY-5505, L-glutamate biosynthesis; PWY-5695, urate biosynthesis/inosine 5’-phosphate degradation; PWY-7431, aromatic biogenic amine degradation.

Discussion

Methane production in cattle is a thermodynamic requirement for the microbial conversion of feed into nutrients¹⁰. The associated removal of hydrogen is critical for the ruminal ecosystem and host because low hydrogen concentrations ensure high fermentation rates and efficient feed digestion¹¹. Therefore, methane production is a natural metabolic process. Enteric methane production in Japanese Black cattle is influenced by changes in the rumen microbiome and affects overall metabolism (Figs. 1 and 2), rendering it crucial to understand the mechanisms of methane production and its metabolic side effects for improving cattle growth and well-being. In particular, we hypothesized that the metabolic changes induced by methane production in cattle are reflected in the ruminal environment, blood metabolites, and liver metabolism.

Rumen microbial features related to methane production

A more diverse microbial community enables the rumen ecosystem to better adapt to dietary changes, which has been associated with improved ruminant growth performance¹², suggesting that the higher ruminal microbial diversity in the HME group facilitated better adaptation to dietary changes than that in the LME group. Additionally, diverse microbial communities can effectively degrade a wide range of plant materials, potentially enhancing the nutrient intake from feed¹³. The increased fiber degradation can lead to higher ruminal hydrogen production, potentially increasing methane production. Notably, ruminants with larger rumens emit increased methane levels, likely because of longer ruminal feed retention¹⁴. Therefore, the higher ruminal diversity in the HME group may be related to a larger rumen size than that in the LME group.

Regression tree and mixed-effect model analyses during the fattening period revealed several key microbiotas associated with methane emissions in Japanese Black cattle. Christensenellaceae, Clostridium methylpentosum, and Mogibacterium, which are related to hydrogen production, were more prevalent in HME cattle. Christensenellaceae, a crucial ruminal hydrogen-producing group¹⁵, effectively breakdown carbohydrates, amino acids, and carboxylic acids to produce acetate and butyrate¹⁶. This family is associated with methane emissions in Holstein cows¹⁷, sheep¹⁸, and beef cattle¹⁹. Clostridium methylpentosum specializes in decomposing specific plant materials in the rumen that other bacteria may not utilize efficiently, producing acetate, glycolaldehyde, carbon dioxide, and hydrogen during the fermentation of L-lyxose and B-arabinose²⁰. Mogibacterium, a hydrogen-producing fibrolytic bacterium, contributes to methane production by generating phenylacetate, which may facilitate cellulose degradation by R. albus strains¹⁸. Consistent with our findings, elevated Mogibacterium levels are also observed in cattle with high ruminal methane emissions^21,22. In contrast, Succinivibrionaceae, Succinivibrio, Anaerovorax, and Lachnospiraceae NK3A20 were more abundant in LME cattle. Succinivibrio, a member of the family Succinivibrionaceae, produces propionate as its primary fermentation product in the rumen²³. This genus helps mitigate methane levels through hydrogen consumption and negatively correlates with methane emissions in sheep²⁴ and dairy cattle²⁵. Although propionate production is the primary hydrogen sink, biohydrogenation also plays this role during ruminal conversion of unsaturated to saturated fatty acids^26,27. Identification of Anaerovorax as a potential biohydrogenating bacterium²⁸ also suggests a role in hydrogen sinking.

Eubacterium nodatum produces acetate from lysine²⁹. Although higher ruminal acetate concentrations were expected in the HME group owing to increased hydrogen production, the acetate generated from amino acids by this bacterium likely contributed to the comparable acetate levels between the HME and LME groups. Additionally, the PWY-5100 pathway, which involves pyruvate fermentation to acetate, was more active in LME cattles. The equivalent acetate concentrations among HME and LME cattle are consistent with our previous finding, which showed no significant difference between high- and low- RME Japanese Black cattle³⁰. Considering that the connection between ruminal fermentation traits and methane production is likely affected by factors such as breed and feeding management practices³¹, high concentrate feeding in Japanese Black cattle likely reduces their reliance on the pyruvate-to-acetate microbial pathway in the rumen, leading to comparable acetate levels regardless of methane emission status. Prevotella can reduce methane emissions by channelling hydrogen into propionic acid production, thereby lowering methanogenesis^32. Conversely, Prevotella levels were higher in the HME group. Similar observations were made in Holstein cattle, where Prevotella phylotypes were more abundant in animals with lower propionate levels²³. In sheep, Prevotella bryantii is an indicator of low-methane ruminotypes, whereas other Prevotella phylotypes are associated with high-methane ruminotypes³³. This discrepancy could be due to differences in propionate production at the phylotype level or variations in metabolic pathways. Metabolic profiling of Prevotella has revealed numerous pathways involving amino acid, carbohydrate, lipid, cofactor and vitamin, nucleotide, and energy (ATP) metabolism³². Therefore, further research is necessary to clarify the influence of Prevotella on methane production in Japanese Black cattle.

The LME and HME groups had higher relative abundances of Methanosphaera and Methanobrevibacter, respectively. Similar patterns have been reported in several studies; for example, sheep with lower methane yields contained more Methanosphaera and fewer Methanobrevibacter³⁴, and dairy cows with lower methane yields produced 26% less methane than their high methane-yield counterparts, exhibiting higher Methanosphaera and lower Methanobrevibacter abundance³⁵. Methanosphaera is a methylotrophic methanogen that depends entirely on hydrogen and utilizes alcohols but not carbon dioxide, formate, or methylamines^36,37. Alternatively, Methanobrevibacter is a hydrogenotrophic methanogen that uses hydrogen along with carbon dioxide or formate to produce methane. The relative abundances of these methanogens are negatively correlated with each other in ruminants^38,39. Furthermore, Methanosphaera, with its low hydrogen threshold, can outcompete Methanobrevibacter at low hydrogen partial pressures⁴⁰. Further research is needed to understand the competitive dynamics of hydrogen and factors influencing the selection of methanogenic lineages. Examining the relationships among methane, hydrogen, and specific methanogenic lineages could provide valuable insights regarding methanogenesis and help develop strategies to reduce enteric methane emissions in the rumen.

KEGG pathways related to methane emission

Several MetaCyc pathways of rumen microbiota associated with methane production were identified using correlation, mixed-effects modelling, and regression tree analyses. In particular, methyl-coenzyme M reductase is targeted by nitro compounds which inhibit methanogenesis showing the key role this enzyme has in methane biosynthesis. Methyl-coenzyme M reductase acts on methyl-coenzyme M (CH₃-S-CoM), which is an immediate precursor to methane. Coenzyme M is thus essential for methanogenesis in rumen microorganisms during the final step of methane production⁴¹. Similarly, we identified coenzyme M biosynthesis (P261-PWY) as a critical factor influencing methane emissions during the fattening period in Japanese Black cattle.

The PWY-5505 pathway involves a transamination process in which oxoglutarate and ammonium are converted into glutamate by glutamate dehydrogenase. This process is a major source of ammonia fixation and is important for converting non-protein nitrogen into proteins in ruminants⁴². In our analysis, the predicted activity of the PWY-5505 pathway was higher in the LME group during the fattening period, which may be consistent with the lower ruminal ammonia concentration observed in this group compared with HME animals. Additionally, oxoglutarate fermentation to glutamate has been described as an important hydrogen disposal pathway, wherein oxoglutarate combines with ammonium, NADH, and hydrogen to form glutamate, nicotinamide adenine dinucleotide (NAD+), and H₂O, respectively. Conversely, hydrogen production during NAD+ conversion to NADH, which occurs during glutamate deamination to oxoglutarate, may contribute to methane synthesis in the rumen⁴³. Furthermore, transamination reactions of ruminal bacteria involving glutamate dehydrogenase proceed at a significantly faster rate than those involving other amino acids, underscoring the elevated dehydrogenation capability of ruminal glutamate biosynthesis⁴³. In addition, formate, a substrate for methanogenesis, can be produced from oxoglutarate by ruminal microbes via a pathway other than the pyruvate formate-lyase reaction⁴⁴. Therefore, the conversion of oxoglutarate to glutamate may also reduce formate production from oxoglutarate and potentially influence methane-related hydrogen and formate fluxes. However, these interpretations should be viewed as preliminary because PICRUSt2 provides predicted functional potential rather than measured pathway expression or enzymatic activity. Thus, the potential association between PWY-5505 and methane metabolism requires further confirmation using direct measurements of pathway activity.

Induced physiological, ruminal, and hepatic gene changes according to methane production

The ruminal ornithine biosynthetic pathway begins with oxoglutarate and proceeds in two stages: glutamate synthesis and ornithine synthesis⁴⁵. The increased ruminal PWY-5505 and GLUTORN-PWY pathway activities in the LME group promotes oxoglutarate conversion first to glutamate then to ornithine, respectively. Increased GLUTORN-PWY pathway activity may be associated with elevated blood ornithine levels in the LME group. Regression tree analysis of blood amino acids and hepatic gene expression revealed that increased blood ornithine levels and OTC activity were characteristic of low-methane emission animals. OTC plays a crucial role in ammonia detoxification and nitrogen waste removal by catalysing the reaction between carbamoyl phosphate and ornithine to form citrulline in the second step of the urea cycle⁴⁶. The ornithine–urea cycle is the primary pathway for ammonia detoxification and urea synthesis in the livers of dairy cattle⁴⁷. The enhanced conversion of ammonia to urea via ornithine and OTC may help maintain body health by detoxifying ammonia. In the present study, the HME group showed higher ruminal ammonia concentrations than the LME group. Elevated ammonia levels can negatively influence rumen fermentation by reducing the efficiency of VFA production and other fermentation end products, ultimately affecting nutrient utilization and energy supply to the host⁴⁸. Relatively low ammonia detoxification through the ornithine–urea cycle and reduced transamination capacity (PWY-5505) may have contributed to the increased ruminal ammonia levels observed in HME cattle.

Regression tree analysis of hepatic genes indicated that SLC1A1 expression was higher in HME cattle. In the liver, SLC1A1 is overexpressed compared to other organs⁴⁹. Its protein product contributes to the biosynthesis of glutathione, an abundant natural antioxidant in the liver, by facilitating transport of the glutathione precursors L-glutamate and L-cysteine, thereby protecting the liver cells from oxidative stress^49,50. Butyrate produced during rumen fermentation is absorbed across the ruminal epithelium and converted into BHBA, which is then transported through the bloodstream and used as an energy source in various tissues. Owing to the high-energy diets fed to Japanese Black cattle during the fattening period, BHBA may not be fully utilized as an energy source and may remain in the bloodstream and tissues, potentially inducing inflammatory injury and oxidative stress in cattle hepatocytes through the NF-κB signalling pathway⁵¹. The elevated hepatic SLC1A1 expression may help mitigate oxidative stress induced by residual BHBA in the HME group. Moreover, Bayesian network analysis revealed a strong association between BHBA and SLC1A1 (arc strength = 0.987), further supporting the role of SLC1A1 in mitigating oxidative stress. Notably, although the differences in blood urea cycle components, ornithine concentration, and OTC and SLC1A1 expression levels do not directly explain methane production, they may indicate downstream metabolic effects derived from the ruminal oxoglutarate-to-glutamate biosynthesis (PWY-5505) pathway. However, the significant differences in these downstream metabolic processes according to methane production status provide evidence that glutamate metabolism could be an important pathway for decreasing methane production by reducing ruminal hydrogen in fattening cattle fed a high concentrate diet.

This study has some methodological limitations that should be considered when interpreting its findings. The methane measurement technique used in this study, the “sniffer method,” is cheaper and easier to implement under on-farm conditions than respiration chambers, as it allows convenient sampling of exhaled air during normal routines by installing a fixed tube in a feed bin within an automatic milking system, where a portion of the animal’s breath during milking or feeding is collected⁵². The resulting spot measurements have been reported to correlate with methane emissions determined using respiration chambers⁵³. However, the method has certain constraints. Methane concentration was measured only during feeding, and because methane emissions fluctuate with time after feeding⁵⁴, such short-term spot measurements may not perfectly represent 24-h methane production. This limitation may affect the accuracy of absolute methane emission values; however, it does not compromise relative comparisons, as all animals were measured under the same conditions⁵⁵. Thus, while absolute values may differ from true daily emissions, the method remains suitable for the relative evaluation and ranking of animals. Another limitation of this study is that ruminal hydrogen concentration was not directly measured. However, in this study, the differences in methane emissions observed between the HME and LME groups were interpreted as being driven by alterations in hydrogen metabolism resulting from shifts in microbial community composition. Previous studies have demonstrated that variations in hydrogen partial pressure dynamically regulate methanogen activity and fermentation balance, thereby affecting methane yield^56,57. Future studies incorporating direct measurements of ruminal hydrogen concentration will be essential to validate these mechanisms and clarify the physiological basis underlying the differences between HME and LME groups.

Conclusion

This study provides crucial insights regarding the ruminal microbial community of Japanese Black cattle, highlighting its association with blood metabolites, hepatic gene expression, and methane emissions (Fig. 4). The more diverse ruminal microbial communities in the HME group may be linked to higher hydrogen production. Christensenellaceae, Clostridium methylpentosum, and Mogibacterium, which are related to hydrogen production, were more prevalent in HME animals. In contrast, Succinivibrionaceae, Succinivibrio, and Anaerovorax, which are associated with hydrogen sinks, were more abundant in LME cattle. Methanobrevibacter may have outcompeted Methanosphaera for hydrogen because of the presumed higher hydrogen concentrations in the HME group. In addition to these microbial shifts in hydrogen flow, the oxoglutarate-to-glutamate biosynthesis pathway may also contribute to hydrogen sinking, suggesting a potential metabolic mechanism underlying the differences in methane emissions. These metabolic differences were accompanied by distinct hepatic gene expression patterns between the HME and LME groups. In LME cattle, improved ammonia conversion to urea via ornithine and OTC detoxified ammonia, thereby promoting body health. In HME animals, higher hepatic SLC1A1 expression may help mitigate oxidative stress by promoting glutathione synthesis. Differences in these metabolic processes suggest that the oxoglutarate-to-glutamate biosynthesis pathway may contribute to hydrogen sinking, thereby serving as a crucial differentiating factor for methane emissions. These findings provide a mechanistic basis for identifying metabolic and microbial biomarkers associated with methane emissions, offering potential targets for mitigating enteric methane production in Japanese Black cattle.

Fig. 4.

Bayesian networks considering both methane emission and selected phenotypic factors. A total of 60 measurements for the early (n = 21), middle (n = 18), and late fattening (n = 21) periods were used in the analysis. The estimated phenotypic network structure of methane emission, PWY-5505, and the downstream metabolism is shown for (a) low and (b) high methane emission conditions. BHBA, beta-hydroxybutyric acid; C4, butyrate; CH4, methane emission; PWY-5505, L-glutamate biosynthesis; GLUTORN-PWY, L-ornithine biosynthesis; NH3, ammonia; OTC, ornithine transcarbamylase; SLC1A1, solute carrier family 1 member 1. The thickness of each arrow represents the edge strength.

Methods

Animals and sample collection

The study involved animal experiments conducted at the Hyogo Prefectural Technology Center of Agriculture, Forestry, and Fisheries in Japan. The experiments followed the guidelines provided by the Institute of Livestock and Grassland Science⁵⁸ and the ethical guidance of the Hyogo Prefectural Institute of Agriculture and Forestry and Fisheries Animal Care and Use Committee. The protocol for the experiments was evaluated and approved by Hyogo Prefectural Institute of Agriculture of Forestry and Fisheries Animal Care and Use Committee (approval number: H2018-01), and all methods were performed in accordance with the ARRIVE guidelines (https://arriveguidelines.org). The study used 21 Japanese Black steers that were raised from 12 months of age (initial body weight, 335.6 ± 19.8 kg) until they reached 30 months of age (final body weight, 742.1 ± 49.9 kg). The experimental period was divided into three phases: early fattening (12 ~ 14 months; T1), middle fattening (15 ~ 22 months; T2), and late fattening (until 23 ~ 30 months; T3). All animals were accustomed for 7 days to eating from the sheet-covered feed bin. To promote voluntary entry and maintain a stable posture during sampling, roughage was provided approximately 30 min before gas collection, and concentrate was offered immediately prior to measurement. Roughage was fed twice daily at 09:00 and 15:00, with concentrate supplied 30 min after each roughage feeding. The growth performance, feed intake and nutritional composition of the formula diet are shown in Table 1. Experimental samples, including blood, liver tissue, and rumen fluid, were collected from 21 Japanese Black cattle during the early (13 months of age), middle (20 months of age), and late fattening phases (28 months of age). Blood samples were collected from the jugular vein at 13:00, three hours after the morning feeding, using heparin–sodium tubes (Venoject II VP-H100K, Terumo, Tokyo, Japan). Samples were transported to the laboratory on ice, centrifuged at 3,000 × g for 15 min, and the resulting plasma was stored at −30 °C until metabolic profiling. Rumen fluid was collected via a catheter at 14:00, five hours after the morning feeding. Approximately 200 mL was obtained into a sterilized flask. The sample was immediately filtered through four layers of cheesecloth to remove feed particles, and the filtrates were stored at −80 °C until component analysis and DNA extraction. Liver tissue biopsies were performed as described previously⁵⁹. Samples were taken from the intersection of the right 11th–12th intercostal space and a line extending from the acromion to the tuber coxa. After local anesthesia with xylazine, a 0.5 cm incision was made, and tissue was collected using an automatic biopsy gun equipped with a 150 mm needle (ACECUT ACE-141502, DELTA Surgical, Staffordshire, UK). The biopsy site was disinfected with 10% isodine gel, and all tissues were immediately frozen in liquid nitrogen and stored at −80 °C until RNA extraction.

Table 1.

Comparison of growth performance, feed intake, and nutritional composition of Japanese Black cattle between HME and LME group during the early, middle, and late fattening periods.

Variables	All perioda		T1		T2		T3		SEM				P-value
	HME	LME	HME	LME	HME	LME	HME	LME	All period	T1	T2	T3	All period	T1	T2	T3
Growth performance
Average body weight (kg)	573.2	558.1	398.3	387.5	590.8	562.5	730.5	724.5	23.6	3.62	8.82	10.68	0.26	0.15	0.11	0.79
Average daily gain (kg/day)	0.76	0.79	0.78	0.89	0.88	0.83	0.61	0.66	0.03	0.05	0.03	0.02	0.11	0.29	0.44	0.30
Feed intake (kg/day)
Concentrate	7.16	7.32	5.99	6.10	8.11	8.24	7.38	7.61	0.20	0.03	0.31	0.28	0.96	0.15	0.84	0.70
Rice straw	1.01	1.12	1.50	1.63	0.83	0.94	0.72	0.80	0.08	0.12	0.06	0.05	0.87	0.59	0.38	0.43
Kraft pulp feed	0.38	0.25	0.57	0.38	0.30	0.20	0.28	0.18	0.06	0.13	0.08	0.06	0.51	0.52	0.50	0.45
Dry matter intake	5.82	5.89	5.23	5.23	6.40	6.47	5.84	5.97	0.12	0.06	0.21	0.19	0.88	0.97	0.89	0.76
Total digestible nutrients	7.40	7.53	6.94	7.01	8.01	8.14	7.26	7.45	0.14	0.03	0.26	0.24	0.90	0.25	0.82	0.71

^aCalculated across T1, T2, and T3 using the mixed-effects model. T1: Early fattening period (13 months of age). Concentrates in T1: steam-flaked corn 42%, wheat bran 27%, corn gluten meal 10%, soybean meal 12%, soybean hull 6%, salt 1% (total digestible nutrients (TDN) 71.2%, crude protein (CP) 15.9%); T2: Middle fattening period (20 months of age) Concentrates in T2: steam-flaked corn 42%, barley 14%, wheat bran 21%, corn gluten meal 5%, soybean meal 10%, soybean hull 6%, salt 1% (TDN 72.5%, CP 14.4%); T3: Late fattening period (28 months) Concentrates in T3: steam-flaked corn 44%, barley 25%, wheat bran 14%, soybean meal 5%, soybean hull 10%, salt 1% (TDN 72.8%, CP 12.0%). Rice straw: dry matter (DM) 87.8%, TDN 37.7%, CP 4.7%; Kraft pulp feed: DM 76.9%, TDN 51.1%, CP 0.3%; HME: High methane-emission cattle (n = 6); LME: Low methane-emission cattle (n = 6). Data sourced from our previous study (Kim et al., 2022).

Methane emissions measurement

Methane emissions were measured in 21 Japanese Black steers at different stages of fattening: early (T1, 13 months old), middle (T2, 20 months old), and late (T3, 28 months old). Methane measurements were obtained from all animals in all three periods, resulting in a total of 63 observations (21 animals × 3 periods). Following our previous protocol^7,55, individual breath gas was collected by covering the feed bin with a sheet fitted with sampling inlets immediately after concentrate feeding. CH₄ and CO₂ were monitored while the animals consumed concentrate following roughage feeding, with two measurements per day for three consecutive days. Air inside the feed bin was continuously drawn using a pump at a flow rate of 6 L/min and directed to a gas analyzer (GLA131; LGR, San Jose, CA, USA) through tubing equipped with a particle filter. CH₄ and CO₂ concentrations were recorded at 1-s intervals for 360 s per measurement. Background gas concentrations were measured before and after each pen-level sampling session, and mean values were used to correct the sampled gas concentrations. Background-corrected concentrations were calculated by subtracting background values from the corresponding sampled gas concentration. To avoid distortion from periods with minimal breath contribution, all 1-s CO₂ values below 500 ppm were removed. Measurements with < 240 s of valid post-filtering data were discarded. For each measurement, the CH₄/CO₂ ratio was calculated as the mean corrected CH₄ concentration divided by the mean corrected CO₂ concentration. Outliers exceeding the mean ± 3 × SD within each fattening stage were excluded.

CH₄ emissions (L/d) were calculated from estimated oxygen (O₂) consumption (L/d), respiratory quotient (RQ; CO₂ emission/O₂ consumption), and the CH₄/CO₂ ratio as follows:

O2 consumption and RQ were estimated from related variables, including body weight (BW, kg), dry matter intake (DMI, kg/d), total digestible nutrients (TDN, %), and ratio of roughage intake to total DMI (Rrate, %). O2 consumption was estimated from heat production (HP), representing the metabolizable energy (ME) required for maintenance and growth, using the equation:

where metabolic BW (MBW) was calculated as BW^0.75. ME intake (MEI, Mcal/day) was estimated as 3.62 × TDN intake (TDNI, kg/day)⁶⁰, with TDNI obtained as DMI × TDN. O2 consumption was estimated from HP by dividing by 4.89⁶¹. Parameter a is the ME requirement for maintenance per unit MBW (0.1124), and b is the efficiency of ME utilization for growth (0.52)⁶².

The RQ was estimated from MEI per MBW and Rrate following Jakobsen & Thorbekt (1993)⁶³:

Individual absolute methane emissions (CH₄, L/d) measured at the early (T1), middle (T2), and late (T3) fattening stages are summarized in Supplementary Table S4.

To consolidate the three measurements from the T1, T2, and T3 phases, the estimated values of the methane emission (L/d) and DMI were calibrated using a mixed model analysis of variance as follows:

where , which represents methane emissions and DMI, is a dependent variable, µ is the overall mean, is the fixed effect of fattening period (three classes: T1, T2, and T3), is the random effect of animal i, and is the random error.

Accordingly, the predicted values of methane emissions (L/d) were obtained through a linear regression model (CH₄(L/d) = 165.8 + 10.07 DMI) of dry matter intake to evaluate the levels of methane emission (L/d). Residual methane emission (RME) was calculated as the difference between the estimated values and the predicted value (n = 21). Three individuals were excluded from group classification because their data points fell below the threshold of one standard deviation below the mean (µ1σ). The upper group (n = 6, mean RME: 23.35 ± 4.93) was classified as the high- (HME) methane-emitting group, and the lower group (n = 6, mean RME: −19.33 ± 4.03) as the low- (LME) methane-emitting group. The estimated CH₄, predicted CH₄ values, and RME of individual cattles are provided in Supplementary Table S4.

Blood metabolites and rumen fermentation characteristics

Blood metabolites were analyzed using an Automatic Biochemical Analyzer (HITACHI 7070, Hitachi, Ltd., Tokyo, Japan), measuring total protein, albumin, blood urea nitrogen (BUN), creatinine, total cholesterol, triglycerides, non-esterified fatty acid (NEFA), glucose, alkaline phosphatase (ALP), aspartate aminotransferase (AST), alanine aminotransferase (ALT), lactate dehydrogenase (LD), gamma (γ)-glutamyl transferase (γ-GTP), creatine kinase, acetate, BHBA, and total ketone bodies. Plasma levels of insulin, insulin-like growth factor 1 (IGF-1), and cortisol were determined by enzyme immunoassay using specific kits: the multi-species insulin ELISA kit (Mercodia bovine insulin ELISA, Mercodia AB, Uppsala, Sweden), the human IGF-1 ELISA kit (Human IGF-I, R&D Systems, Minneapolis, USA), and the cortisol ELISA kit (Cortisol ELISA kit, Enzo Life Sciences Inc., Budapest, Hungary), following the manufacturers’ protocols. For blood amino acid concentrations, trichloroacetic acid was added to the plasma, and proteins were filtered through a membrane filter before analysis with a high-speed amino acid spectrometer (L-8900, Hitachi High Tech, Tokyo, Japan). Total VFA in the rumen fluid and individual components, such as acetic acid, propionic acid, butyric acid, and valeric acid were quantified by gas chromatography (GC2014, Shimadzu, Kyoto, Japan) using a Thermon-3000 [3%] packed glass column on a Shimalite TPA 60–80 support (Shinwa Chemical Industries Ltd., Kyoto, Japan). The gas chromatography was operated under the following conditions: nitrogen as the carrier gas at a flow rate of 30 mL/min, with the column injection and FID detection temperatures set at 220 °C, and the column oven at 140 °C. Ammonium nitrogen concentration in the rumen fluid was measured using the steam distillation method with an automatic nitrogen analyzer (Kjeltec Auto 1035, Tecator, Sweden).

Transcriptomics analyses of hepatocytes

Liver tissue samples from 20 cattle at early fattening (T1), 20 at middle fattening (T2), and 19 at late fattening (T3) were used for RNA-Seq. The tissues were homogenized in 200 µL RNAiso Plus (TAKARA Bio Inc., Shiga, Japan) with a Multibeads shocker (YASUIKIKAI Inc., Osaka, Japan) following the manufacturer’s instructions. Homogenization was performed twice at 2000 rpm for 10–15 s, followed by the addition of 800 µL RNAiso at 25 °C. The homogenate was collected in 1.5 mL tubes, mixed with 200 µL chloroform, and centrifuged at 12,000×g for 15 min at 4 °C. The supernatant was mixed with 500 µL isopropanol and centrifuged again to precipitate the RNA. The RNA pellet was washed twice with 75% cold ethanol, dissolved in RNase-free water, and quantified using a Nano Drop ND-1000 Spectrophotometer (Thermo Fisher Scientific, Waltham, MA, USA). RNA purity was confirmed with an A260/A280 ratio > 1.8 and verified via 1.0% agarose gel electrophoresis. RNA integrity was confirmed using a Tape Station 4200 (Agilent Technologies, Santa Clara, CA, USA) with an average RNA integrity number (RIN) of 7.1. RNA-seq libraries were prepared using the TruSeq stranded mRNA Kit (Illumina, San Diego, CA, USA) and sequenced on a NovaSeq 6000 platform at Macrogen Japan Corporation. Sequencing read quality was evaluated with FastQC software (version 0.11.8), trimmed with Trim Galore (version 0.5.0), and mapped to the ARS-UCD1.2 bovine reference genome using HISAT2 ⁶⁸. Read counts were calculated using StringTie⁶⁴ with the GTF Bovine gene annotation file. Normalization of counts and PCA were performed using the DESeq2 package⁶⁵ in R statistical software and two T2 samples identified as outliers were removed.

Metataxonomics analyses of rumen microbiota

Genomic DNA was extracted from rumen fluid using the Fast DNA kit (MP Biomedicals, CA, USA) and quantified with the NanoDrop ND-1000 Spectrophotometer (Thermo Fisher Scientific, France), with integrity verified by agarose gel electrophoresis. One sample in T2 was excluded due to poor gDNA quality. Approximately 10 µg of rumen gDNA was used for PCR amplification to generate libraries of the bacterial 16S rRNA V3-V4 region using 341 F-806R primers (forward: CCTACGGGNGGCWGCAG, reverse: GACTACHVGGGTATCTAATCC) and the archaeal 16S rRNA V4 region using Arch 349F-806R primers (forward: GYGCASCAGKCGMGAAW, reverse: GGACTACVSGGGTATCTAAT) following the 16S metagenomics sequencing library preparation protocol⁶⁶. Sequencing was performed on Illumina’s MiSeq platform (San Diego, CA, United States) using the Paired-End method (2 × 300 bp). Metataxonomic analysis of the bacterial and archaeal rumen microbiome was conducted using QIIME2 (version 2023.2)⁶⁷, with adapter and primer sequences trimmed using Cutadapt⁶⁸ and further processing with DADA2⁶⁹ for chimera removal, denoising, and quality filtering. Sequences were clustered at 99% similarity using the Naive Bayes classifier⁷⁰ with the SILVA v13.8 database^71. Taxonomic assignment was refined by excluding unassigned ASVs, chloroplast, mitochondria, and non-bacteria taxa. Beta diversity of overall rumen microbiotas between HME and LME across fattening stages were analyzed using MicrobiomeAnalyst⁷², with Principal Coordinates Analysis (PCoA) based on Bray-Curtis dissimilarity⁷³. A Venn diagram illustrated ASVs with 100% prevalence present in both groups and unique to each group using InteractiVenn software⁷⁴. Functional genetic profiles were predicted using PICRUSt2 (Phylogenetic Investigation of Communities by Reconstruction of Unobserved States 2) v2.5.2⁷⁵, with differences in Kyoto Encyclopedia for Genes and Genomes (KEGG) profiles between HME and LME groups analyzed.

Statistical analysis

The comparison of feed intake, growth performance, blood metabolites, hormones, amino acids, hepatic genes, rumen fermentation, rumen microbiota, functional profiles, and methane emission levels between the HME (n = 6) and LME (n = 6) groups was conducted by applying the linear mixed model as follows:

where is the observation of animal i for fattening period j; is the total mean; is the fixed effects of methane emission group (two classes: HME and LME groups); is the fixed effects of fattening period j; is the interaction between methane emission group and fattening period; is the random effect of animal i; is the residual random effect. The ‘lme4’ R package⁷⁶ was used for fitting the above linear mixed model, and an analysis of variance (ANOVA) was then performed the fixed effects using ‘car’ R package⁷⁷. The entire dataset of 21 animals measured across three periods, excluding outliers and low-quality samples, was subjected to multivariate analyses, including correlation analysis, regression tree analysis, and Bayesian network analysis. Spearman’s correlation coefficients were analyzed using the ‘hmisc’ R package⁷⁸, with statistical significance determined at |r|≥ 0.5 and P ≤ 0.05 to identify pairwise associations between MetaCyc pathways and methane emission. Regression tree analysis was conducted using the ‘lmertree’ R package⁷⁹. Methane emissions were treated as the dependent variable, while phenotypic variables—including rumen microbial taxa, MetaCyc pathway abundances, and hepatic transcriptomic profiles—as independent variables. Bayesian networks were constructed to infer significant interactions between methane emission and host phenotypic factors. The structure of the Bayesian network was developed using a tabu search algorithm in the ‘bnlearn’ R package⁸⁰, and the uncertainty of the edge strength and direction within the network was estimated using 2,000 bootstrap replicates.

Fig. 5.

Proposed model for blood metabolites, hepatic transcriptomes, and rumen fermentation associated with high and low methane emission in Japanese Black cattle. BHBA, beta-hydroxybutyric acid; PWY-5505, L-glutamate biosynthesis; GLUTORN-PWY, L-ornithine biosynthesis; OTC, ornithine transcarbamylase; SLC1A1, solute carrier family 1 member 1.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

H.L., M.K. contributed to sample collection, sample analysis, data analysis, and writing; T.M., K.I., and E.I. contributed to experimental design, sample collection, and data analyses; O.K., K.O., I.N., A.A., U.Y. contributed to data analysis, data interpretation, and manuscript revision; S.H., F.T. and S.R. contributed to experimental design, data interpretation, writing, and revision. All authors read and approved the final manuscript.

Funding

This work was supported by the MAFF Commissioned project study on “Development of Technologies to Reduce Greenhouse Gas Emissions in the Livestock Sector” (Grant Number JPJ011299). This work was partly supported by JSPS KAKENHI (grant number: 23K18075). Huseong Lee was granted by JST (grant number: JPMJSP2114).

Data availability

Raw sequence reads data obtained from the present study have been deposited in the NCBI BioProject database, assigned the project number PRJNA1031765. The data will be available with the following link: [https://www.ncbi.nlm.nih.gov/bioproject/PRJNA1031765](https:/www.ncbi.nlm.nih.gov/bioproject/PRJNA1031765) .

Declarations

Competing interests

The authors declare no competing interests.