Introduction

The 2015 UNFCCC Paris agreement aims to pursue efforts to limit the increase in global warming to 1.5 °C1 above temperatures prevailing during the pre-industrialisation era1. Globally, agriculture, forestry and other land use account for about 24% of annual greenhouse gas (GHG) emissions2. These include direct emissions from the cultivation of crops and livestock and also as a result of deforestation3. Enteric methane (CH4) emissions from the livestock sector is the single largest contributor to global CH4 emissions (40%)3.

Most of the variation in the composition of the rumen microbiota is a direct result of feed composition and quantity consumed4,5,6. A period of feed-restriction followed by ad libitum feeding is widely used in livestock management for the beef industry to reduce feed costs7. Following a period of dietary restriction cattle typically undergo accelerated growth referred to as compensatory growth8, upon re-alimentation. Feed restricted animals have been previously shown to have slower passage rate of feed through the rumen9,10 which has previously been correlated with increased CH4 emissions per unit of feed from ruminants9,11,12. We previously found that feed-restricted animals had increased relative abundance of 16S DNA from the Methanobrevibacter gottschalkii clade5. This observation was positively correlated with a dramatic decrease (from 30% to less than 1%) in the relative abundance of a single bacterial 16S DNA sequence which could only be identified confidently as an uncultured Proteobacteria species and appeared to be related to the family Succinivibrionaceae5. As some Succinivibrionaceae species utilise hydrogen13 we hypothesized that the aforementioned putative Succinivibrionaceae species was in competition for substrate, possibly hydrogen, with the Methanobrevibacter gottschalkii clade5 in feed restricted animals. This was consistent with a decrease in ruminal propionate and attendant increased acetate:propionate (A:P) ratio, an indication of greater methanogenesis in these animals5. Following two months of re-alimentation, the volatile fatty acid ratios of ruminal digesta from previously feed restricted animals mirrored that of their non-restricted contemporaries, while a reversal in the relative abundance of 16S DNA from the Methanobrevibacter gottschalkii clade and the putative Succinivibrionaceae 16S DNA5 was also observed.

It is still not clear if analysis of 16S DNA accurately reflects the metabolic activity of rumen bacteria and archaea as metabolically inactive and dead bacteria retain large amounts of DNA14. We were therefore interested in determining which ruminal methanogens showed the highest methanogenic activity in cattle undergoing feed restriction followed by re-alimentation and compensatory growth. To do this we employed high-throughput amplicon sequencing of mcrA gene cDNA. Methyl co-enzyme reductase (MCR) is a multi-subunit enzyme of methanogenic archaea which catalyses the final step in CH4 formation and the initial step in anaerobic CH4 oxidation15. It reduces methyl-coenzyme M (CH3-S-CoM) by 7-mercaptoheptanoylthreonine phosphate (H-S-HTP)16. The MCR subunits are coded for by the mcr gene operon which usually comprises the genes mcrA, B, C, D and G. The mcrA gene codes for subunit A of MCR and can be used both as a phylogenetic marker for the identification of archaea and, as mcrA transcript abundance is positively correlated with methane production in peat soil17, is a quantitative marker for methane production. The objective of this study was to determine which methanogens were contributing to methanogenesis in the bovine rumen during periods of dietary restriction. To our knowledge this is the first time that high-throughput sequencing of mcrA cDNA amplicon libraries has been used to explore the diversity of metabolically active methanogens present in the rumen. This may be useful as a convenient and inexpensive method for use in large numbers of animals to answer the, as yet unanswered question, ‘which methanogens are making the greatest relative contribution to ruminal methane production’?

Results

Animal Performance

Data on the performance of the animals from which ruminal digesta was harvested for this study, have been reported previously by Keogh et al.8. Abbreviations differ in previous publication due to different scientific objectives and for ease of comparison. A and AA refer to ADLIB from period one and two respectively, R and RA refer to RES period one and period two respectively8. Briefly, dry matter intake (DMI) (kg/d) for each animal group was A = 12.80 ± 0.43, R = 5.44 ± 0.17, AA = 12.22 ± 0.98 and RA = 11.8 ± 0.97. The average daily increase in body weight (kg) for each group was, A = 1.93 ± 0.15, R = 0.62 ± 0.07, AA = 1.4 ± 0.30 and RA = 2.5 ± 0.52. The average body weight of the animals at slaughter (kg) for each group was, A = 631.8 ± 41.6, R = 466.1 ± 44.6, AA = 667.1 ± 56.0, and RA = 575.5 ± 36.9. Body weight of R animals was less (P < 0.05) than that of A animals and similarly body weight of RA was less (P < 0.05) than that of AA animals.

Weight of both the full and empty reticulorumen complex was recorded at slaughter8. While the weight of the reticulorumen was lower for R animals in comparison to the other treatments, there was no difference in the weight of digesta between the treatment groups, potentially reflecting the slower ruminal passage of digesta in R animals.

Phylogenetic Analysis

A total of 19 from 25 MCRA clusters were differentially represented amongst the four treatment groups (Fig. 1). Methanobacteriaceae was the dominant family identified from the phylogenetic analysis. Twelve MCRA clusters were identified to be the most closely related to the genus Methanobrevibacter (MCRA cluster 1, 3, 10, 883, 7083, 7084, 7087, 7089, 7092, 7094, 7105 and 7159) and five MCRA clusters were grouped with another genus from the Methanobacteriaceae family; Methanosphaera (MCRA Cluster 4, 22291, 23687, 23941, 23942). The remaining eight MCRA clusters were identified to be related to the Methanomassiliicoccaceae family (MCRA cluster 0, 2, 6, 11, 13, 14, 32, and 144). Of these twelve MCRA clusters with Methanobrevibacter lineage, MCRA cluster: 1, 7083, 7084, 7092 and 7105 grouped with known methanogens from the Methanobrevibacter gottschalkii clade. Two MCRA clusters: 883 and 7094, assembled closely with members of the Methanobrevibacter ruminantium clade.

Figure 1
figure 1

The evolutionary history which approximately identifies the phylogeny of clustered sequences was inferred using the Maximum Parsimony method. The percentage of replicate trees in which the associated taxa clustered together in the bootstrap test (500 replicates) are shown next to the branches. Branches with less than 30% support were collapsed. The MP tree was obtained using the Subtree-Pruning-Regrafting (SPR) algorithm with search level 1 in which the initial trees were obtained by the random addition of sequences (10 replicates). All positions with less than 85% site coverage were eliminated. There were a total of 157 positions in the final dataset. The 4 numbers in red separated by a “;” after each cluster represent the average representation of this cluster in R, A, RA and AA groups respectively.

Full size image

Relative abundance of MCRA amino acid sequences (translated from mcrA cDNA) indicated that the family Methanobacteriaceae had the highest methanogenic activity in the rumen of all bulls sampled regardless of treatment (Table 1). There was no difference in Methanobacteriaceae (P > 0.05) between the R and A groups although relative abundance of translated mcrA transcripts from this family were increased (P < 0.001) in group R compared to groups RA and AA.

Table 1 Mean relative abundances for the four treatment groups of clusters identified from maximum parsimony tree at family and genus level.
Full size table

There was no difference in the relative abundance of MCRA clusters assigned to Methanosphaera (P > 0.05), a genus of the Methanobacteriaceae family, between any of the four treatment groups. The major changes in relative abundance of translated mcrA transcripts were attributed to the dominant genus Methanobrevibacter. Relative abundance of Methanobrevibacter was elevated in group R in comparison to groups A, AA and RA (P < 0.001). The abundance of mcrA transcripts assigned to the M. gottschalkii clade was increased in group R relative to all other treatments groups (P < 0.0001) whereas the mcrA transcripts assigned to the M. ruminantium clade were dramatically reduced in group R compared to groups A, AA and RA (P < 0.0001).

Methanomassiliicoccaceae was not statistically significantly different between treatment groups A and R. However, it did differ between treatment groups A and R and between treatment groups AA and RA (P < 0.001) (Table 1).

Methanobrevibacter ruminantium clade

MCRA clusters 7094 and 883 were the only clusters that were assigned to the M. ruminantium clade (Fig. 1) and both were (P < 0.0001) reduced in the restricted group R in comparison to groups A, AA and RA.

Methanobrevibacter gottschalkii clade

Five MCRA clusters (1, 7083, 7084, 7092 and 7105) (Table 2) were assigned to the M. gottschalkii clade. Four of these MCRA clusters (1, 7083, 7084 and 7092) were increased (P > 0.0001) in group R relative to groups A, AA and RA. MCRA cluster 1 is the largest cluster of the dataset as the majority of translated mcrA cDNA sequences were assigned to this cluster. The relative abundance of this MCRA cluster was increased by approximately 2-fold in the feed restricted animals.

Table 2 Mean relative abundances of the 25 MCRA clusters for the four treatment groups.
Full size table

Methanomassiliicoccaceae

There was no difference between treatments in the overall relative abundance of MCRA clusters assigned to the family Methanomassiliicoccaceae (Table 1). However, the relative abundance of four individual MCRA clusters (13, 14, 32 and 144) within this family was reduced (P < 0.001) in treatment group R compared to groups A, AA and RA (Table 2).

Ordination of Diversity of MCRA clusters

nMDS (Fig. 2) and PCoA (Fig. 3) plots and dendrogram analysis (Fig. 4) were used to assess the similarities and diversity of populations of MCRA clusters within the four treatment groups. These show a tight cluster of points representing the restricted animals which is separate from most of the points representing animals from the ad libitium (A, AA and RA) treatment groups. This indicates that methane was being generated by different methanogen species in the feed-restricted animals than in the ad libitum fed animals. These cluster analyses also suggests that the methanogenically active methanogens were more similar between feed-restricted animals than between ad libitum animals.

Figure 2
figure 2

nMDS plot indicating similarity of metabolically active methanogens between treatments. Distance between samples, based on similarity of cluster composition (similarity 97%) of each sample calculated using Bray-curtis similarity index and plotted using non-metric multidimensional scaling (nMDS). Each point represents a different sample plotted according to their cluster conformation and abundance (stress value = 0.1). A greater distance between two points infers a lower similarity between them, whereas samples composed of similar MCRA clusters will group closer together.

Full size image
Figure 3
figure 3

Principal coordinate analysis (PCoA) of the relative abundances of clusters. PCoA was used to compare variation of relative cluster abundances within the liquid rumen samples from treatment A, AA, R and RA.

Full size image
Figure 4
figure 4

Dendrogram of Bray-Curtis similarity of MCRA clusters in treatment groups; A, AA, R and RA.

Full size image

Diversity matrices also showed a reduction in diversity of MCRA clusters in treatment group R relative to the three ad libitum treatment groups (Table 3). This was apparent in the heatmap (Fig. 5). A distinct pattern of MCRA clusters representing active methanogens that are specific to the R bulls was observed. The heatmap highlights the relative increase in representation and increased homogeneity between samples of the MCRA clusters 1, 7083, 7084, 7092 of the M. gottschalkii clade in restricted animals in comparison to treatment groups A, AA and RA. The relative under representation and homogeneity between samples of MCRA cluster 883 and 7094 of the M. ruminantium clade and MCRA cluster 13, 14, 32 and 144 of the Methanomassiliicoccaceae family in treatment group R in comparison to treatment groups A, AA, RA was also evident from the heatmap. MCRA cluster 3 of Methanobrevibacter lineage was not significantly different between treatment groups, despite a numerical difference in R group bulls (Table 2). This cluster had the second greatest relative methanogenic activity regardless of treatment and this activity is consistent across treatment. This can be visualised from the heatmap (Fig. 5).

Table 3 Effect of dietary restriction and ad libitum feeding, pre and post restriction, on alpha diversity, species richness and presence of MCRA clusters in treatment groups; A, AA, R and RA.
Full size table
Figure 5
figure 5

Square root transformed relative abundance heat map (white = absence, dark blue = high) of MCRA clusters in treatment groups; A, AA, R and RA.

Full size image

Discussion

We previously reported, using high-throughput 16S DNA amplicon sequencing, that a period of moderate feed restriction led to greater abundance of M. gottschalkii clade in the ruminal liquor of moderately feed restricted animals with an inverse relationship to a putative Succinivibrionaceae species5. Results from the study5 show that both ruminal fractions, i.e. solid and liquid, exhibited similar community diversity, however, differences linked to feeding regimes were amplified in the rumen liquid. This informed the decision to go forward with detailed examination of the microbial RNA of the rumen liquid fraction in order to address the main research objective to identify which methanogens dominate methanogenesis during periods of bovine dietary restriction.

Our approach utilised mcrA cDNA amplicon sequencing to determine the abundance of mcrA transcripts and their phylogenetic origin to identify the predominant methane generating species under different feed treatments. High-throughput mcrA amplicon sequencing, appears to have potential as a convenient and inexpensive tool for the identification of methane producing archaea and to assess their potential contribution to methane production in a large number of animals. There are two isoenzymes of MCR, denoted MCRI, which is encoded by operon mcrABCDG and MCR II, encoded by operon mrtABDG 16. The different isoenyzme systems allow methanogens to be supported by different growth environments18. The mcrA gene is contained in both enzyme operons hence making it a useful molecular marker for methanogen activity in different environments.

The RNA gene expression work described here partly supports the results of our 16S DNA amplicon sequencing. Methanobacteriaceae and Methanomassiliicoccaceae were found to be the dominant families in all treatment groups both at the RNA and DNA level. Methanobrevibacter was both the dominant methanogenic genus present in the 16S DNA study and produced the most mcrA transcripts relative to the other methanogen genera detected in the current study. The 16S amplicon sequencing analysis that we previously reported on these animals showed that relative abundance of the M. gottschalkii clade increased in bulls fed a restricted diet in comparison to the cattle fed an ad libitum diet5. The present study shows that this increase also occurred at the RNA level, indicating that feed restriction favours members of this clade. However, whereas our 16S DNA amplicon study on these animals showed that members of the M. ruminantium clade remained unchanged between treatments, our mcrA cDNA amplicon study shows a dramatic decrease in abundance in restricted animals. There are at least two possible explanations for this discrepancy. The first of these is that 16S DNA amplicon sequencing detects the presence or absence of bacteria and/or archaea, but does not give direct information on the activities and physiological states of microorganisms in samples14. The M. ruminantium clade detected in our 16S DNA amplicon study may have been present but not generating methane, while our mcrA cDNA amplicon data most likely provides a more accurate picture of which species are generating methane in the rumen. Secondly, our 16S DNA amplicon sequencing simultaneously targeted bacteria and archaea so it was possible to determine the abundance of archaea relative to bacteria. The mcrA primers, however, only show abundance of archaea relative to other archaea so the large increase in clusters in MCRA clusters in the M. gottschalkii clade would make it appear as though there was a decrease in other MCRA clusters. To resolve this, a further study is required in which an exogenous control RNA sequence is added at the start of RNA extraction to obtain an absolute rather than relative measure of mcrA transcript abundance. Alternatively, RNA-Seq analysis, for example, would permit the identification of mcrA transcripts relative to all other transcripts. Relative abundance of Methanomassiliicoccaceae did not significantly differ between treatment groups at either the DNA or RNA level.

The phylogenetic assembly and relative abundance of mcrA cDNA found in this study are in agreement with the current literature on rumen composition derived from DNA based investigations11,19,20. This includes an international study19, which found that there is consistency in ruminant bacterial and archaeal communities regardless of species, diet and geographical location. Members of the M. gottschalkii and M. ruminantium clades were the two largest methanogenic groups in the rumen, accounting for approximately 74% of the total archaeal population19. Our mcrA cDNA results were consistent with this regardless of treatment. The two other dominant methanogen groups were Methanosphaera and Methanomassiliicoccaceae. In total, these four groups account for 89.2% of the total rumen archaeal population.

Significantly, we have identified, using alpha diversity analysis, that feed restriction has a dramatic effect on species which generate methane in the rumen. Dietary restriction appears to provide a niche environment for specific methanogenic groups and removes the necessary environment required to sustain a more diverse archaeal community. High feed intake was previously associated with increased passage rate9 and constant substrate supply to rumen microbes, ensuring diverse substrate to the rumen methanogen population. In our study, the microbiota of the restricted bulls probably resides in a different ecosystem to the ad libitum groups due to the smaller rumen size and reduced feed availability8. MCRA cluster 3, identified to be of the Methanobrevibacter genus, showed the second greatest methanogenic activity in all treatment groups. This indicates that dietary management inferred no bias toward the methanogenic activity of this Methanobrevibacter cluster. Feed-restriction did however, affect other clusters within the three major methanogenic groups active in the rumen. The relative abundance of MCRA clusters that were most closely related to the M. ruminantium clade and Methanomassiliicoccaceae were decreased in the rumen of feed restricted bulls, while the relative abundance of MRCA clusters of the Mbr. gottschalkii clade was increased.

M. ruminantium M1 possesses mcrI  21 but does not possess the mcrII gene system. However, another known member of the M. ruminantium clade, M. olleyae possess mcrII so it is unclear if the lack of mcrI is typical of members of the M. ruminantium clade22. If it is assumed that the clusters 7094 and 883 of the M. ruminantium clade, identified in this current study, possess a similar mcrI system to M. ruminantium M1, then this could explain the reduction in this clade in the feed restricted animals. Expression of mcrI and mcrII are controlled preferentially on the levels of substrate available to methanogens present18. Low and high substrate concentrations for methanogens upregulate the expression of mcrI and mcrII respectively18. Reduced feed intake is correlated with a reduction in apparent digestibility despite an increase in particle retention time10. There was increased bacterial alpha diversity in the restricted animals and increased time for microbial fermentation, along with assumed decreased digestibility. Therefore, it was possible that there was a steady state of substrate available to methanogens present in R animals in comparison to animals fed ad libitum 9,10. In animals fed ad libitum feed, there is typically an increase in digestibility leading to rapid substrate (H2) availability to methanogens, initially9. This has a limiting effect on H2 production from ruminal bacteria, as at a high H2 concentration, H2 producing pathways are less thermodynamically favourable and thus pathways that utilise H2 are favoured, such as propionate production9. We hypothesise that due to the reduction in feed intake in restricted animals and the reduced degradability, rumen H2 concentrations never reach the threshold concentration that was required for the negative feedback to be initiated. Steady substrate availability may have shifted expression from the mcrI isoenzyme in favour of mcrII, therefore decreasing the activity of members of the M. ruminantium clade. M. ruminantium clade members only possess mcrI therefore this increased the activity of members of the gottschalkii clade, where all members contain both versions of the isoenzyme. The methanogens active in the restricted group also were contained in a rumen with reduced volume in comparison to methanogens active in the ad libitum treatment groups8. This may contribute to the steady supply of H2 present in the rumen and why it is sufficient to switch activity predominately to mcrII. In the rumen of animals fed ad libitum, the Mbr. ruminantium clade was active. This may be due to variation in the amount of H2 present in the rumen at different times of the day, as members of this clade are active when supply of H2 was low.

The succinate producing bacterial family Succinivibrionaceae was significantly increased in animals fed an ad libitum diet and virtually absent in animals fed a restricted diet5. This bacterial family utilises H2 to produce succinate which is rapidly converted to propionate in the rumen23. The ad libitum groups all had increased propionate in their rumens in comparison to the restricted group5. This provides further evidence that in ad libitum animals alternative pathways for H2 utilisation are favoured9, while H2 was below the threshold level for alternative pathways to be prioritised in restricted animals and therefore utilised for CH4 production24 by a less diverse group of methanogen with the majority made up of the M. gottschalkii clade.

Four MCRA clusters identified to be from the Methanomassiliicoccaceae family were significantly less methanogenically active in the restricted treatment group in comparison to the ad libitum treatment groups. The Methanomassiliicoccaceae family are obligate H2-dependent methylotrophs, utilising methyl groups from methanol and methylamines (mono-, di-, and tri-methylamine) and methyl thiols for the production of methyl coenzyme25. It has previously been hypothesised that there is a positive association between Methanomassiliicoccaceae and Succinivibrionaceae19. Succinivibrionaceae degrade pectin26 to produce methanol, which is a substrate required for the growth of Methanomassiliicoccaceae27. Therefore, dietary restriction may also have an influence on methylotrophic methanogens which utilise by-products of the microbial fermentation other than hydrogen, such as methylamines. This reduces the diversity of the active methanogen population.

In conclusion we sequenced mcrA cDNA amplicon libraries to assess the abundance and diversity mcrA cDNA sequences present in the bovine rumen of animals undergoing dietary restriction and subsequent compensatory growth. This validated previous 16S DNA based data from our group and showed the most abundant methanogens were indeed the most methanogenically active. The study identifies a clear cohort of active methanogens in an environment where feed intake was restricted and methanogen substrate supply was limited but stable, when compared with when substrate was unrestricted. The majority of methanogens were hydrogentrophic and we hypothesise that H2 concentration in the rumen was the main driver of methanogenic activity in the rumen. We found that under the condition of dietary restriction, where CH4 production was predicted to be increased, relative abundance of the M. gottschalkii clade increased, while members of the M. ruminantium clade showed a dramatic decrease. Relative abundance of members of the Methanomassiliicoccaceae family, which are H2 dependant methyltrophs, were also reduced in diet restricted bulls, as H2 concentrations are regulated by passage rate. During the ad libitum feeding period we predicted that ruminal H2 concentration was increased, directing H2 utilisation away from CH4 production in favour of other H2 utilising pathways, such as succinate and propionate production pathways. In these pathways H2 was used endogenously in the microbial cell rather than accumulating in the rumen. This contributes to a diverse methanogenic population that can survive at a variety of H2 concentrations and also utilises by-products of the alternative H2 utilising pathways such as methylamines. Restricted animals had a slower passage rate and a lower rate of digestibility within the rumen, providing a stable H2 supply that is directed towards CH4 production, giving rise to a low diversity population of active methanogens. This study therefore implies that a common livestock management system used to reduce feed costs may be contributing to increased ruminant CH4 production. Research may be able to identify targeted CH4 mitigation approaches such as anti-methanogen vaccines, dietary strategies and probiotics and therefore warrant further exploration.

Methods

Animal model

All procedures involving animals were approved by the University College Dublin, Animal Research Ethics Committee and licensed by the Irish Department of Health and Children in accordance with the European Community Directive 86/609/EC.

Animal model and rumen digesta sample collection have previously been published by our group5,8 respectively.

In brief, this experiment was conducted as part of a larger study designed to examine the physiological and molecular control of compensatory growth in growing beef cattle8. Animals were managed on the same farm from two weeks of age prior to being transferred to Teagasc Grange Beef Research Centre, Dunsany, Co. Meath, Ireland. In order to acclimatise the animals to their environment and reduce any latent influence of previous environments, all animals were subjected to a 3 month common feeding period of ad libitum grass silage plus 2 kg of concentrate per head per day prior to commencing the experiment. Sixty purebred Holstein Friesian bulls (mean live weight 370 ± 35 kg; mean age 479 ± 15 d) were blocked on the basis of live weight, age and sire into four groups with fifteen animals in each group. Two of these groups received a restricted dietary allowance for 125 days (groups R and RA), with the other two groups (groups A and AA) receiving feed ad libitum during the same time. This 125 day period of differential feeding was followed by a subsequent ad libitum re-alimentation period for RA and AA groups which lasted 55 days. Throughout the 180 day trial all animals received the same total mixed ration diet consisting of 70:30 concentrate:forage (grass silage). Further details of the diet employed are provided by Keogh et al.8. The 55 d ad libitum period was inclusive of a 15 d transition period in order to build up R group animals to ad libitum feed intake. This transition period was implemented to allow animals to acclimatise to a higher plane of nutrition while preventing the development of intestinal disorders, such as acidosis. R animals were managed to grow at a rate of 0.6 kg /d, with A animals expected to gain in excess of 1.5 kg/d during the first 125 d period.

Groups R and A were slaughtered after 125 days of feeding, whilst groups RA and AA were slaughtered after the subsequent 55 day ad libitum re-alimentation period (following 180 days of feeding). All animals were slaughtered in an EU licensed abattoir (Euro Farm Foods Ltd, Cooksgrove, Duleek, Co. Meath, Ireland). Slaughter order was randomized to account for potential confounding effects on treatment outcomes. Two A group animals, two AA group and one R group were removed from this experiment due to illness. Two samples, one from the A group and one from the R group, were mislaid. This left 12 A, 14AA, 13 R and 15RA samples.

Sample Collection at Slaughter

Digesta contents were sampled from five different locations, inclusive of the dorsal and ventral sacs, within the rumen of each bull immediately after slaughter. Liquid and solid fractions were separated by squeezing rumen contents through 4 layers of cheese cloth and immediately frozen in liquid nitrogen, transported on dry ice, and then stored at −80 °C.

RNA Extraction and cDNA Synthesis

Approximately 20 g of frozen rumen liquid sample from each of the animals was considered as representative. Each sample was homogenised to a fine frozen powder under liquid nitrogen using a pestle and mortar and stored at −80 °C.

Approximately 60 mg of the homogenized frozen powder was used for RNA extraction of liquid digesta sample. RNA was extracted using the Qiagen RNeasy plus kit (Qiagen, Manchester, UK). RNA extract yield and purity were assessed with two consecutive readings on the Nanodrop 1000 spectrophotometer. RNA integrity was estimated with an Agilent 2100 Bioanalyzer (Agilent Technologies, Santa Clara, CA). Total RNA (6 µg) was DNase treated using a TURBO DNA-free™ Kit (Thermo Fisher Scientific, MA, USA) to remove genomic DNA. DNase treated RNA was then purified with Zymo RNA clean and concentrate kit (Zymo Research Corp, Irvine, CA, USA). The RNA concentration in each sample was normalised to 50 ng/µL and then reverse transcribed into cDNA using random hexamers and a High Capacity cDNA Reverse Transcription Kit (Thermo Fisher Scientific, MA, USA).

mcrA Amplicon Library Preparation

mcrA amplicon libraries (n = 54) were generated by PCR amplification using 2.5 ng cDNA as a template and Mlas forward/ mcrA primers reverse28 fused with Nextera XT overhang adapters (Illumina, San Diego, CA, USA). PCR was performed using 1 × KAPA HiFi HotStart ReadyMix DNA polymerase (Roche Diagnositics, West Sussex, UK) and cycle conditions were initial denaturation at 95 °C for 3 minutes, then 35 amplification cycles (95 °C for 30 seconds, 55 °C for 30 seconds, 72 °C for 30 seconds) then a final extension at 72 °C for 5 minutes.

Amplicons were subsequently purified using a QIAquick PCR Purification Kit (Qiagen, Manchester, UK). A second PCR step attached dual indices and Illumina sequencing adapters using the Nextera XT index kit (Illumina, San Diego, CA, USA). Cycle conditions were an initial denaturation at 95 °C for 3 minutes, then 8 amplification cycles (95 °C for 30 seconds, 55 °C for 30 seconds, 72 °C for 30 seconds) followed by a final extension at 72 °C for 5 minutes. Amplicons were analysed on a 2% (w/v) agarose gel to check they were the correct size then pooled in equal concentrations. Pooled amplicons were gel purified to remove unwanted products using the Qiagen Gel Extraction Kit (Qiagen, Manchester, UK). An extra purification with the QIAquick purification kit was used to remove residual agarose. The pooled purified libraries were measured for purity and quantity on a Nanodrop 1000 and further quantified using the KAPA SYBR FAST universal kit with Illumina Primer Premix (Roche Diagnositics, West Sussex, UK). The library pool was then diluted and denatured as recommended by the Illumina MiSeq library preparation guide (Illumina, San Diego, CA, USA). Sequencing was conducted using 500 cycle MiSeq reagent kits (Illumina, San Diego, CA, USA).

Sequence analysis

Read 1 and Read 2 of mcrA amplicons were quality checked using FASTQC29 (http://www.bioinformatics.babraham.ac.uk/projects/fastqc/) and overlapping reads were merged using BBMerge from the BBMap package (https://sourceforge.net/projects/bbmap/). Sequences were then size selected based on predicted amplicon fragment size (472 bp + /−1). Chimeras were removed using both a de novo method and against the mcrA database from the Functional Gene pipeline & repository (http://fungene.cme.msu.edu/). This resulted in 3,526,425 total reads from samples.

Each read was translated in all six reading frames in an effort to identify intact open reading frames. For each sequence read, a single reading frame was obtained with an expected size of ~155–160 amino acids in length and contained the conserved translation of the forward and reverse sequencing primers. This resulted in 3,526,425 MCRA peptide sequences.

Peptide sequences were clustered at 97% identity. MCRA clusters were required to account for at least 0.1% of the total peptide sequences and clusters below this threshold were discarded. This resulted in 25 MCRA clusters accounting for 96.2% of the total peptide sequences.

Evolutionary Analysis

A representative amino acid sequence from each cluster was obtained. MCRA protein sequences from sequenced methanogen genomes were downloaded from the NCBI database (https://www.ncbi.nlm.nih.gov/protein). Genome sequences with known taxonomy were also downloaded from the mcrA reference database30 and translated into MCRA protein sequences. A database of 67 MCRA protein sequences was compiled consisting of the downloaded and representative cluster sequences. Evolutionary history was inferred using the maximum parsimony method using MEGA7.

Statistical Analysis

Relative abundance of MCRA clusters was computed using raw sequence counts per cluster. MCRA clusters were analysed using a Kruskal-Wallis test with correction for multiple testing using the Benjamini Hochberg (BH) method to identify MCRA clusters differently represented across treatment groups. A transformation based principal coordinate analysis (PCoA) and non-metric multidimensional scaling (nMDS) was performed by square root transformation of the relative abundance data and subjecting Bray Curtis similarity analysis and then plotted using PRIMER v7 software. A heatmap was generated using PRIMER v7 software and the square root transformed relative abundance data for all four treatments. Alpha diversity (observed cluster, richness, Shannon and Simpson diversity indices) was calculated using PRIMER v7 software. Significance of alpha diversity was tested with a one-way ANOVA in R, multiple comparison analysis was assessed between treatments using the Tukey Honest Significance Difference (HSD) test in R.