Identification and Validation of Biomarkers for Designing Enteric Methane Mitigation Strategies in Cattle

Introduction

The major challenge in livestock farming worldwide is to increase and improve production while limiting environmental impact and competition for resources for human consumption. Ruminants are key to food security as they convert forages, not directly usable by monogastric animals and humans, into animal products for human food. They supply 51% of all protein with respectively 67% and 33% from milk and meat (Gerber et al., 2013). However, ruminant livestock contributes substantially to greenhouse gas (GHG) emissions (14.5% of total anthropogenic emissions). Cattle (beef, milk) are responsible for about two-thirds of that total, largely due to methane (CH₄) emissions resulting from rumen fermentation, and with a global warming potential 28 times that of carbon dioxide (IPCC, 2014).

Developing operational strategies to reduce enteric CH₄emissions in cattle is important, while maintaining optimal production to preserve the competitiveness of farms, and meeting consumer demand for safe and quality food. Many strategies (nutrition, biotechnologies, management and genetics) have been evaluated individually worldwide, but few cost-effective solutions are currently available to producers (see review by Beauchemin et al., 2020). The future challenge in significantly and sustainably mitigating methanogenesis may be to design new strategies, such as combining feeding management and animal genotyping for low CH₄emissions.

Another important issue is the quantification of CH₄emissions to evaluate the effectiveness of different mitigating strategies. Under controlled conditions available in experimental research stations, direct measurement methods are commonly implemented (Guyader et al., 2015; Arbre et al., 2016; Doreau et al., 2018). These methods are accurate and repeatable, but are not adapted to a large number of animals and are difficult to implement on farms. To address these issues, alternative methods of estimating CH₄emissions using biomarkers (or proxies) are currently under evaluation (see review by Negussie et al., 2017).

The purpose of this paper is to provide an update of research on methanogenesis by reporting data from recent case studies on promising CH₄mitigation strategies and large-scale proxies to estimate CH₄emissions in cattle.

Combination of Strategies to Mitigate Enteric Methane Emissions

Numerous CH₄ mitigation strategies have been proposed in recent decades (Martin et al., 2010). Most concern manipulation of the rumen microbiome by nutrition (lipids, starch, natural and chemical additives, etc.) or by biotechnologies (vaccines, probiotics, early life programming, etc.). Strategies focusing on the animals (genetic, etc.) are more recent and have been reviewed by Beauchemin et al. (2020), who considered not only their potential of CH₄mitigation, but also their expected availability and feasibility of implementation on farms, as well as their limits. It is important not to focus on a single strategy of CH₄mitigation, but to investigate several at the same time for diverse production systems and environments. The future challenge in significantly and sustainably mitigating enteric CH₄emissions in ruminants may be to combine different strategies, such as feeding management and animal genotyping for low CH₄emissions (Beauchemin et al., 2020).

Additive Effect of Different Anti-methanogenic Dietary Strategies

Lipids have emerged as a persistent option for mitigating enteric CH₄emissions from ruminants. However, their potential of mitigation is moderate (~20%) if used at a suitable dose, while avoiding negative effects on animal performance (Martin et al., 2021). We tested whether it was possible to increase the CH₄mitigation potential of lipids (linseed oil) by combining them with another dietary strategy (nitrate) with a different mode of action on the metabolism of H2 in the rumen. In a meta-analysis, we reported that lipids may be relevant in reducing H₂ production (via reduction of protozoa), whereas nitrate may stimulate H₂ consumption (H₂ sink) by a pathway of competition with methanogenesis (Guyader et al., 2014). We assumed that simultaneous manipulation of H₂ production and HH₂ 2 utilization allows a greater reduction in CH₄emissions than when acting on a single pathway. To test this hypothesis, we investigated the effect of linseed oil and nitrate distributed alone or in combination on CH₄emissions and digestive processes in non-lactating cows fed a hay-based diet (Guyader et al., 2015). The persistency of the effect of linseed+nitrate on CH₄emissions and lactating performance was also assessed in dairy cows fed a maize silage-based diet (Guyader et al., 2016).

Compared with the control diet, linseed oil and nitrate decreased CH₄emissions [g/kg dry matter intake (DMI)] by -17 and -22% when fed alone, respectively, and by -32% when combined, without altering diet digestibility. The daily kinetics of CH₄emission measurements in respiration chambers clearly showed the mode of action of each dietary treatment. Linseed oil supplementation reduced CH₄emissions throughout the day compared to the control diet, while nitrate had a transient but marked action for 3 h post-feeding. Combination of the strategies cumulated the two modes of action (Figure 1).

Figure 1. Daily methane production pattern of non-lactating cows fed 4 different diets containing linseed oil and calcium nitrate alone or in combination (n = 4). Treatments consisted of control diet (CON), CON plus 3% calcium nitrate (NIT), CON plus 4% linseed oil (LIN) and CON plus 4% linseed oil and 3% calcium nitrate (LIN+NIT). The arrows indicate time of feeding. Error bars indicate SD. Symbols indicate hourly statistical comparison between dietary treatments (†P ⩽0.10; *P ⩽0.05; **P ⩽0.01; ***P ⩽0.001). Adapted from Guyader et al. (2015).

In addition, we showed that linseed oil + nitrate fed to lactating cows for 2 months induced a repeatable and persistent reduction of CH₄ emissions [−47%, g/d; −30%, g/kg DMI; −33%, g/kg fat- and protein- corrected milk (FPCM)], on average; Figure 2), without any effect on digestibility of nutrients, nitrogen balance or animal health. Intake and milk yield tended to be lower for dairy cows fed the linseed oil + nitrate diet, but feed efficiency (kg FPCM/kg DMI) was unaffected. Nitrate or nitrite residuals were not detected in milk and associated products (yoghurts, whey, curd and 6-week ripened Saint-Nectaire cheese) suggesting that milk from cows fed nitrate is safe for human consumption. This persistent effect showed the absence of adaptation of the rumen microbiota. However, the energy benefits from the decreased CH₄ emissions with the linseed oil + nitrate diet did not appear beneficial for dairy performances.

Figure 2. Methane emissions from lactating cows fed a control diet based on maize silage (CON; n = 8) or CON supplemented with 10% extruded linseed plus 1.8% nitrate on a DM basis (LIN +NIT; n = 8) after 5 weeks (Wk 5) and 16 weeks (Wk 16). Error bars indicate SD. Letters indicate a significant effect (P = 0.002) between diets. Adapted from Guyader et al. (2016).

This work confirmed our initial hypothesis that combining dietary strategies with different mechanisms of action to reduce H₂ availability in the rumen reduces methanogenesis more markedly than when lipids and nitrate are fed individually. This proof of concept opens up a range of possibilities for designing new strategies to increase CH4 abatement.

Animal Phenotyping for Low Methane Emissions across Diets

Animal phenotyping as a strategy to mitigate enteric CH₄ emissions is more recent and less known than strategies related to nutrition (review of Løvendahl et al., 2018). Interest in selecting ruminants for their CH₄ emissions has increased since it was demonstrated that heritability of this trait is moderate (h2 ≈ 0.29 and 0.40 for CH₄ in g/d in sheep and cattle, respectively) (Pickering et al., 2015; Lassen and Løvendahl, 2016). Together with the heritability of traits, classification of individuals by their CH₄ emissions and stability of ranking over time for different diets are prerequisites for the purposes of animal breeding.

The repeatability of CH₄ emissions in a large number of animals in contrasting feeding conditions is a key point to consider. Coppa et al. (2020) investigated under farm conditions the repeatability [i.e. between animals/(between animals + unexplained) variability] and stability of dairy cow ranking in the long term according to their CH₄ emissions. Forty-five dairy cows fed three contrasted diets formulated to be more or less methanogenic were phenotyped over a 4- month period after the peak of lactation regarding their CH₄ emissions quantified using GreenFeed systems. The repeatability of CH₄ emissions was calculated from data averaged over 1, 2, 4, and 8 weeks for each animal. It increased with the duration of the measurement period and peaked (0.78 for CH₄ in g/d) with 8-week averaged periods. This high repeatability confirmed a high variability among cows, which is relevant for phenotyping regarding CH₄ emissions. However, despite this high individual variability, the dairy cow ranking on CH₄ emissions (g/d) was not stable over time between all individuals or within any of the three diets (Figure 3). Similar results were found by Rischewski et al. (2017) who showed an unstable ranking in a few dairy cows (n = 8) fed a similar diet between three periods (early, middle and late) in the second phase of lactation. These authors stated that only the 2 extreme cows (the highest and the lowest CH₄ emitters) stayed approximately (but not absolutely) near the extreme, whatever the lactation phase of measurement. Coppa et al. (2020) found that, in a large number of animals, even extreme cows changed ranking. Denninger et al. (2019) showed that differences in daily CH₄ emissions (g/d) between 2 groups of 10 high- or low-emitter dairy cows were maintained across 5 subsequent months of lactation and 2 different diets, but no information on individual ranking independently of group was available.

Excluding reasons related to the methodological analysis, the most probable interpretation of the ranking variation over time is a change in CH₄ emissions within individuals. Dietary factors (DM and GE intakes) or milk performances (milk yield and FPCM) are known to explain variability of absolute CH₄ emissions over time in dairy cattle (Beauchemin et al., 2020). However, in our trial, ranking stability was not improved when expressing CH₄ emissions per unit of intake or unit of milk, irrespective of the diets (Figure 3). Other factors like changes in animal physiological status over time (i.e. pregnancy, heat events, etc.) or external factors like climatic changes (i.e. occurrence of heat stress period) or barn management activities (i.e. insemination practices, etc.) could perturb feeding behavior over time (i.e. variations in meal frequency and consistency) and thus CH₄ emissions (Garnsworthy et al., 2012; Hammond et al., 2016). Our data highlight the importance of phenotyping animals across environments in which they will be expected to perform.

Figure 3. Individual ranking of CH₄ emissions from 2 subsequent periods of 8 weeks (P1: week 11–18 and P2: week 19–26) for diet CH₄+ (grey triangles; n = 15), CH₄ int (white dots; n = 15), and CH₄- (black squares; n = 15) diets; full lines represent the y = x equations; (Rs = Spearmann correlation coefficients between the ranking of the same cow in the two subsequent measurement periods); *, P < 0.05; ns, not significant. CH₄+, diet formulated to produce high methane emissions; CH₄ int, diet formulated to produce intermediate methane emissions; CH₄-, diet formulated to produce low methane emissions. Adapted from Coppa et al. (2020).

Figure 3. Individual ranking of CH4 emissions from 2 subsequent periods of 8 weeks (P1: week 11–18 and P2: week 19–26) for diet CH4+ (grey triangles; n = 15), CH4 int (white dots; n = 15), and CH4- (black squares; n = 15) diets; full lines represent the y = x equations; (Rs = Spearmann correlation coefficients between the ranking of the same cow in the two subsequent measurement periods); *, P < 0.05; ns, not significant. CH4+, diet formulated to produce high methane emissions; CH4 int, diet formulated to produce intermediate methane emissions; CH4-, diet formulated to produce low methane emissions. Adapted from Coppa et al. (2020).

The profitability of ruminant livestock is assumed to increase if their environmental impact is reduced through the selection of low CH₄ emitters. Indeed, the feed efficiency of animals would increase by saving energy lost in the form of CH₄, and CH₄ emissions expressed per kg of product would decrease (Capper et al., 2009; Legesse et al., 2016). The relationship between the two phenotypes (CH₄ emissions and feed efficiency) is largely unknown (Basarab et al., 2013) due to the absence of simultaneous measurements on these two large-scale criteria over sufficiently long and representative breeding periods, and is also difficult to investigate because of the complexity of their mode of expression.

Bes et al. (2021) recently investigated the relationship between the individual variability of enteric CH₄ emissions to that of feed efficiency in growing steers fed two contrasted diets. One hundred young Charolais bulls fattened for 6 months received ad libitum a total mixed ration based either on maize silage (rich in starch; MS-S) or on grass silage (rich in fiber; GS-F). Methane emissions (g/d) were measured individually using GreenFeed systems and individual feed efficiency was calculated as residual feed intake (RFI = difference between the real and theoretical amount of DM intake in kg/d) during the same periods.

The RFI values ranged between -0.98 and 1.08 (SD = 0.59) for the MS-S diet and between -0.93 and 0.82 (SD = 0.42) for the GS-F diet. Methane emissions differed between the two diets [between 201 and 371 g/d (SD = 31) and between 154 and 329 g/d (SD = 29) for MS-S and GSF diets, respectively; P < 0.01). Regression analysis revealed that daily CH₄ emissions (g/d) were positively linked with RFI (r = 0.50, P ≤ 0.01) with a slope averaging 26 g/kg DM intake for both diets (Figure 4). Animals that ingested food in excess of their maintenance and growth requirements emitted more CH₄ per day with both diets. These data are in agreement with those reported recently in dairy heifers (Flay et al., 2019) and steers when CH₄ emissions were expressed in g/kg DMI (Renand et al., 2019). In contrast, two older studies showed that CH₄ emissions (g/d) were lower for the more efficient steers (Hegarty et al., 2007; Nkrumah et al., 2006).

Figure 4. Relationship between individual CH₄ emissions and RFI in growing bulls (n = 100) fed a diet based on maize silage (rich in starch; MS-S) or based on grass silage (rich in fiber; GSF). Adapted from Bes et al. (2021).

Figure 4. Relationship between individual CH4 emissions and RFI in growing bulls (n = 100) fed a diet based on maize silage (rich in starch; MS-S) or based on grass silage (rich in fiber; GSF). Adapted from Bes et al. (2021).

All these results confirm the complexity of studying the relationship between these 2 criteria and the need for further research in order to understand the underlying mechanisms. Incorporating CH₄ production in a genetic selection program needs to consider the potential risks of counter-selecting other phenotypes of interest (Løvendahl et al., 2018). In their review, these authors report that the genetic selection of animals that emit low CH₄ would reduce the digestive efficiency of fiber, a phenotype which is particularly interesting to consider in the context of ruminant breeding that does not compete with human food production.

Biomarkers (or proxies) to Estimate Enteric Methane Emissions

In the last few years, much progress has been made in terms of the precision and accuracy of direct methods of measuring daily enteric CH₄ emissions (g/d). However, most methods used to quantify these emissions on an individual scale in ruminants (respiration chambers, SF₆ tracer technique, GreenFeed®) remain expensive, labor-intensive, technically challenging and thus limited to research (Hammond et al., 2016). There is a need to develop easy methods of estimating CH₄ emissions on a large scale and in practical conditions, so as to evaluate the effect of different mitigating strategies (diets, animals, management systems) compared to what is currently done in controlled experimental conditions.

Negussie et al. (2017) reviewed a large range of available CH₄ proxies that have been recently explored in dairy cattle, including parameters related to (1) intake, (2) digestive function, (3) milk performance, (4) the whole animal. All proxies were summarized in terms of their characteristics, including simplicity, cost, accuracy, invasiveness, and throughput.

In this section, biomarkers (or proxies) are defined as ‘indicators’ measurable in body matrices easy to access, and that can be used to predict CH₄ emissions. Methane proxies might be less accurate than direct methods, but can be measured frequently to reduce random noise. We focused on proxies from two body matrices: milk routinely obtained on dairy farms, and plasma for its potential availability in all cattle including beef and growing replacement dairy heifers.

Proxies from milk

Proxies related to milk composition, like milk fatty acid (MFA) and mid-infrared (MIR) profiles, seems to be a promising way to predict CH₄ emissions because precursors for methanogenesis and de novo synthesis of MFA both arise in the rumen (Negussie et al, 2017). Milk FA composition was first used as CH₄ emission proxies in dairy cows fed corn silage-based diets containing linseed (Chilliard et al., 2009). Other prediction equations based on milk FA concentrations have been reviewed by van Gastelen and Dijkstra (2016), but all were developed in a narrow range of diets and had limited data numbers, thus restricting their range of application.

In this context, Bougouin et al. (2019) used a meta-analytical approach to construct a set of CH₄ prediction equations from MFA in lactating dairy cows fed a wide range of diets. A large international data set¹ (n = 825) was created including individual CH₄ emissions (mostly measured using a respiration chamber and SF₆ tracer techniques), individual MFA profile (exclusively determined by gas chromatography), and data on daily DMI, dietary composition, milk performances (yield and composition), and animal characteristics (days in milk, body weight).

Twenty equations for prediction of enteric CH₄ have been published. Equations including MFA alone confirmed common rumen metabolic pathways between methanogenesis and lipid metabolism in dairy cows. Figure 5 presents the most correlated individual MFA among the 5 main families (SFA, odd- and branched-chain FA, cis MUFA, trans MUFA, and PUFA) with 3 CH₄ emission metrics (g/d, g/kg DMI, g/kg milk). Prediction equations based on MFA alone had a root mean squared error of 65.1 g/d, 2.8 g/kg of DMI, and 2.9 g/kg of milk, respectively.

Complex equations that additionally used other variables [DMI, dietary chemical composition (neutral detergent fiber, ether extract contents), animal characteristics (days in milk, body weight)] had a lower root mean squared error of 46.6 g/d, 2.6 g/kg of DMI, and 2.7 g/kg of milk, respectively. Few MFA (cis-11 C18:1) and (trans-10 C18:1) were commonly found among prediction equations in this study and in the literature (Dijkstra et al., 2011; Mohammed et al., 2011; Rico et al., 2016; van Gastelen et al., 2018). Performance of the prediction equations was not consistent, meaning that MFA used alone had a limited potential to predict CH₄ emissions. Our results confirmed that equations predicting CH₄ from MFA depend on diet (Mohammed et al., 2011; Dijkstra et al., 2016; Rico et al., 2016) and the lactation stage (Vanrobays et al., 2016). In addition, increasing the complexity of prediction equations, by combining proxies, increased their robustness (Negussie et al, 2017), probably because more complex equations explain an additional proportion of the variability not taken into account in simple equations with MFA alone.

A minimum difference of 16% (simple equations) and 11% (complex equations) in CH₄ emissions (g/d) between mitigating strategies can be evidenced with the best prediction equations (Bougouin et al., 2019). These authors highlight the low potential of applicability of their equation on farms, because of the limitations (time-consuming, expensive) associated with routine use of gas chromatography, which is considered the gold standard in determining MFA.

Figure 5. Significant correlation between CH₄ emissions (g/d, g/kg DMI, g/kg milk) and individual MFA in dairy cows. BCFA = odd- and branched-chain FA; SFA = saturated FA; RBH = rumen biohydrogenation intermediates with MUFA = mono-unsaturated FA and PUFA = polyunsaturated FA. (+) and (-) indicate positive and negative correlation, respectively. Adapted from Bougouin et al. (2019)

Figure 5. Significant correlation between CH4 emissions (g/d, g/kg DMI, g/kg milk) and individual MFA in dairy cows. BCFA = odd- and branched-chain FA; SFA = saturated FA; RBH = rumen biohydrogenation intermediates with MUFA = mono-unsaturated FA and PUFA = polyunsaturated FA. (+) and (-) indicate positive and negative correlation, respectively. Adapted from Bougouin et al. (2019)

Milk composition (fat, protein, lactose and urea contents), including some FA (Soyeurt et al., 2011), can be determined routinely and at low cost by MIR spectroscopy in milk recording laboratories. Some authors report a good potential of milk MIR spectra as a proxy for prediction of CH₄ emissions in dairy cattle, especially when combined with animal characteristics such as lactation stage (van Gastelen and Dijkstra, 2016), milk yield, parity and breed (Vanlierde et al., 2020). This high throughput approach allows CH₄ production to be incorporated in dairy cow breeding programs.

Proxies from plasma

A plasma metabolomics approach was successfully used to assess performance efficiency in growing steers (Artegoitia et al., 2017) and heat stress in dairy cows (Tian et al., 2015). Information is scarce on the metabolic consequences in the host ruminant when rumen methanogenesis is depressed. Understanding these possible changes is important for acceptance by producers of mitigation strategies, and for exploring new metabolites that could be used as proxies of enteric CH₄ emissions. In addition, plasma is a biological fluid that has the advantage of being easily accessible from all types of ruminants.

Yanibada et al. (2020) hypothesized that plasma metabolites that originate from rumen microbes, from the host or both (co-metabolites), are impacted by a reduction of ruminal methanogenesis. Twenty-five Holstein primiparous cows were fed the same diet with (n=12) or without (n=13) a specific inhibitor of methanogenesis for 6 weeks to obtain two groups of cows classified as lowor high- CH₄ emitters according to their different (23%) CH4 emissions (g/d). Milk production, food consumption, body weight and indicators of the health status of animals were comparable between all animals. The plasma metabolome was explored using untargeted [nuclear magentic resonance (NMR) and liquid chromatography-mass spectrometry (LC-MS)] and targeted (LCMS/MS) approaches.

The plasma metabolome differed between high- and low-CH₄ emitter dairy cows, although the differences were moderate. A wide range of discriminating plasma metabolites (n=48) were identified and 7 metabolomic pathways were associated in the low-CH₄ emitters (Figure 6). Some metabolites were of microbial origin, such as dimethyl sulfone, formic acid and metabolites containing methyl groups such as stachydrine. They are known to be involved in methanogenesis or the use of H₂ in the rumen and can potentially be used as proxies of methanogenesis. Other discriminating plasma metabolites are produced by the host or are of mixed microbial-host origin. The latter metabolites, which increased in low-CH₄ emitters, belong to general energy and amino acid metabolic pathways, suggesting that reduction of methanogenesis occurs without negative effects on dairy cows.

Figure 6. Analysis of the metabolic pathways modified in low-CH₄ emitters. Discriminating metabolites identified showed 7 impacted metabolic pathways: 1) Valine, leucine and isoleucine biosynthesis 2) Taurine and hypotaurine metabolism 3) Glycine, serine and threonine metabolism 4) Phenylalanine, tyrosine and tryptophan biosynthesis 5) Methane metabolism 6) Glyoxylate and dicarboxylate metabolism 7) Arginine and proline metabolism. The plot was built based on the pathway enrichment analysis (node colors) and on the pathway impact values resulting from the pathway topology analysis (node size). Adapted from Yanibada et al. (2020)

Figure 6. Analysis of the metabolic pathways modified in low-CH4 emitters. Discriminating metabolites identified showed 7 impacted metabolic pathways: 1) Valine, leucine and isoleucine biosynthesis 2) Taurine and hypotaurine metabolism 3) Glycine, serine and threonine metabolism 4) Phenylalanine, tyrosine and tryptophan biosynthesis 5) Methane metabolism 6) Glyoxylate and dicarboxylate metabolism 7) Arginine and proline metabolism. The plot was built based on the pathway enrichment analysis (node colors) and on the pathway impact values resulting from the pathway topology analysis (node size). Adapted from Yanibada et al. (2020)

Yanibada et al. (2020) demonstrated the proof of principle that plasma metabolome reflects changes in enteric CH₄ emissions in dairy cattle. Plasma metabolites identified were useful in improving understanding of the physiological effects on the ruminant host induced by a reduction in methanogenesis. Plasma is a biofluid of interest for developing new proxies of methanogenesis. Validation of metabolomic approach and discriminant metabolites is required in studies using anti-methanogenic additives with different modes of action, in a large number of animals and in different production conditions.

In conclusion, large-scale studies covering a wide range of experimental conditions is necessary to increase the accuracy of the existing prediction equations and to validate new proxies. The integration of databases from different matrices and reconciliation of all the data would offer the possibility of proposing combinations of even more robust and discriminating models for prediction of individual emissions of enteric CH₄ in ruminants. The development of proxies is at a relatively early stage and should be a priority in future research for designing new enteric methane mitigation strategies in cattle.

Acknowledgements

The author thanks the organizing committee of the 5th annual Animal Nutrition Conference of Canada (ANCC) for their invitation and encouragement to prepare this article. The author thanks the technical and scientific staff, permanent and non-permanent, of the UMR 1213 Herbivores for their input in this collaborative research as well as institutes and private partners for their financial support [Adisseo France SAS (Antony, France), Agrial Caen, France), APIS-GENE (Paris, France), Deltavit (Janze, France), DSM Nutritional Products AG (Kaiseraugst, Switzerland), Institut de l’Elevage (Paris, France), Lallemand (Blagnac, France), Moy Park Beef Orleans (Fleury-les-Aubrais, France), ADM nutrition animale - Neovia (Saint-Nolff, France), Techna France Nutrition (Coueron, France), and Valorex (Combourtille, France)].

C. Martin (https://orcid.org/0000-0002-2265-2048).

Presented at the 2021 Animal Nutrition Conference of Canada. For information on the next edition, click here.

Identification and Validation of Biomarkers for Designing Enteric Methane Mitigation Strategies in Cattle

Predict and act: Heat stress management made better

Diets of Productive Sheep & Goats: Performances & Health - Dr. Cannas, University of Sassari