The forage legume species biserrula (Biserrula pelecinus) is among a few forage species with low enteric methane emission when fermented by rumen microbes and it is unclear whether metabolites in biserrula play a role in this. The hypothesis of this study was that specific metabolites in biserrula are candidate biomarkers for and associated with low methanogenesis. We characterized the metabolomic profiles of 30 accessions of biserrula or its core collection using nuclear magnetic resonance and liquid chromatography mass spectrometry. These profiles were then analyzed for association with methanogenesis potential in the rumen. Metabolomic profile was predictive of methanogenesis. Forty-seven putative metabolites were identified by liquid chromatography mass spectrometry that were highly expressed (P < 0.05) in the associations with low methanogenic potential. Fragmentation analysis of a subset of these metabolites suggested saponins, with one feature tentatively identified as an ursolic or oleanolic terpene glucoside, consistent with the nuclear magnetic resonance data. Accessions with Eritrean geographic origin were metabolomic outliers. Here, we show for the first time that some metabolites contribute to the methane mitigation effects of biserrula. If in vivo results confirm this, the environmental impact of this study would be the availability of biomolecules for livestock vaccination to mitigate methane emission resulting in an economic impact of lower cost of production in countries with a price on environmental emissions. . . . . . Keywords Biserrula pelecinus (biserrula) Enteric methane mitigation Geographic origin Legume Metabolomics Saponin 1 Introduction Electronic supplementary material The online version of this article (https://doi.org/10.1007/s13593-018-0510-x) contains supplementary CH is a common byproduct of anaerobic microbial fermenta- material, which is available to authorized users. 4 tion of forageinthe rumen(Fig. 1a). Ruminants are vital to food production in most countries of the world, and in Australia and * Kioumars Ghamkhar Kioumars.firstname.lastname@example.org New Zealand, their products are important contributors to eco- nomic and social wellbeing. Enteric CH emission occurs as a result of microbial fermentation of feed components (Johnson Forage Improvement, Grasslands Research Centre, AgResearch, Palmerston North, New Zealand and Johnson 1995). CH produced by domesticated ruminants represents a loss of 2–15% of the gross energy intake (Holter and Centre for Plant Genetics and Breeding, The University of Western Australia, Perth, Western Australia, Australia Young 1992), thus being an important inefficiency in ruminant production systems (Moss et al. 2000). To abate CH emission GIn Silico Pty Ltd, Melbourne, Victoria, Australia and increase the efficiency of digestion in ruminants, the focus is Department of Economic Development, Jobs, Transport and to divert some of the energy lost as CH into animal products Resources, Melbourne, Victoria, Australia (Mitsumori and Sun 2008). Enteric fermentation can vary widely School of Applied Systems Biology, La Trobe University, depending on factors such as type of the animal, feed intake, the Melbourne, Victoria, Australia type and quality of feed, environment, adding dietary fat, feed School of Animal Biology, Faculty of Natural and Agricultural additives, rumen microbiota, and, indirectly, body weight of the Sciences, The University of Western Australia, Perth, Western Australia, Australia animal through feed intake (Moss et al. 2000). There are urgent needs to understand the various factors Department of Agriculture and Food Western Australia, Perth, Western Australia, Australia affecting variability in enteric CH production to decrease 4 30 Page 2 of 10 Agron. Sustain. Dev. (2018) 38:30 Fig. 1 a Schematic presentation of methanogenesis in rumen. The type of feed can affect the extent of methanogenesis. The type of rumen flora also affects the final outcome of methanogenesis/homoacetogenesis. b Spaced biserrula plants in the field the uncertainty in GHG emission inventories and to identify Western Australia), held by the Department of Agriculture viable GHG reduction strategies (Sejian et al. 2011). and Food Western Australia (DAFWA) (Ghamkhar et al. Improving pasture quality is viewed as a means of reducing 2012). Preliminary LCMS analysis of a subset (12) of the CH emissions because animal productivity may be enhanced biserrula accessions has previously been reported by (i.e., lowering CH emissions per unit of animal product) with Ghamkhar et al. (2016). Here, we report the use of metabolo- less dietary energy lost as CH due to reduced fiber content. In mics techniques to investigate the effect of all accessions of New Zealand, Friesian and Jersey dairy cows grazing sulla the core collection of biserrula on CH production when sup- (Hedysarum coronarium), a condensed tannins-containing le- plied to a rumen fluid suspension in vitro. In the past, some gume, emitted less CH per unit of dry matter than cows phytochemicals have been suggested to have a direct toxic grazing perennial ryegrass pasture (Woodward et al. 2002). effect on methanogens (e.g., condensed tannins) or protozoa Further, the antimicrobial properties of essential oils have (e.g., saponins) (Patra and Saxena 2009) and the heteroge- been shown through in vitro and in vivo studies to inhibit a neous group of compounds known as essential oils also show number of bacteria and yeasts and to control fermentation correlation with methanogenesis (Martin et al. 2010). Recent gases, livestock waste odors, and human pathogenic bacteria studies suggested plant secondary compounds as a cue in me- (Helander et al. 1998; Lee et al. 2002; Wallace et al. 2002). thanogenic potential of biserrula (Swinny et al. 2015;Banik et Experiments conducted in vitro (Ulyatt et al. 2002) and in vivo al. 2016). Identification of metabolites in this study will pro- (Bayaru et al. 2001) show that addition of organic acids to the vide the same, similar, or different candidate phytochemical diet reduces CH production. targets for development of novel classes of methanogen inhib- In a recent study, a broad range of dietary pasture and itors. These measurements and correlations should assist plant forage species and their effect on CH production in rumi- breeders to make better choices in plant selection, in particular nants was reported (Banik et al. 2013). The most promising targeting CH mitigation from livestock. among these species was biserrula (Biserrula pelecinus), an The aims of the current study were to: important annual legume for the wheat belt of southern Australia (Fig. 1b). 1. Compare the metabolomics profiles of the accessions of Recently, a core collection of 30 accessions of biserrula has the biserrula core collection and their association with been developed from the genetic resource of 279 accessions in CH emission using an in vitro fermentation technique the Australian Trifolium Genetic Resource Centre (Perth, (IVFT); and Agron. Sustain. Dev. (2018) 38:30 Page 3 of 10 30 2. Identify candidate metabolites that can reduce methane 10% CO :10% H to expel oxygen from the tubes. Before the 2 2 production in the rumen and suggest future environmen- commencement of fermentation, rumen fluid was collected tally and phytochemically targeted plant breeding options from two fistulated sheep fed a general maintenance diet in biserrula. consisting of lupins and oaten chaff 2 h after feeding. After straining, rumen fluid was pooled, transferred into the anaerobic chamber, and buffered to pH 7.2. Then 10 mL of this mix was dispensed into prepared tubes. Inside the cham- 2 Materials and methods ber, the tubes were sealed and incubated for 24 h at 39 °C, with constant shaking at 50 rpm. Tubes were then placed in a water 2.1 Plant material bath at 39 °C. After measuring gas pressure, 5 mL of head- space gas was used for subsequent methane analysis by gas The core collection of biserrula (Biserrula pelecinus L.), chromatography (GC). Methane concentration in the gas sam- made up of 30 accessions conserved in the ATGRC, and ple was determined using a Micro-GC with a PoraPLOT Q two controls, red clover (Trifolium pratense L. cv. 0.25 mm × 10 m column (Varian/Agilent Technologies, Redquin) and subterranean clover (Trifolium subterraneum Australia Pty Ltd., Mulgrave, Vic.), at 50 °C, a thermal con- L. cv. Woogenellup) (Table 1) were sampled and screened ductivity detector, and Galaxie software (Varian, Australia). for their in vitro CH output analysis as described by (Banik The methanogenic potential of the plant was expressed as total et al. 2013). Briefly, 47 plants of 30 accessions of the methane produced (mL/g DM incubated, (Patra et al. 2006)). biserrula core collection with half of them replicated, and two control species, subterranean and red clover, were 2.3 Mass spectrometry grown in pots in a glasshouse, and plant material was col- lected for in vitro analysis of methanogenic potential at LCMS was based on the methodology described by the au- 77 days post-sowing. The herbage consisting of leaf and thors (Ghamkhar et al. 2016), but in this experiment, a much stem 3 cm above ground level when elevated were harvested larger sample of 30 accessions was examined with the follow- from individual plants, and the material was freeze-dried in ing steps: a bench-top freeze dryer (BenchTop, VirTis/SP Scientific, Warminster,PA) andgroundinagrindingmill (Glen 2.3.1 Extraction Creston, Stanmore, UK) to pass through a 1.0-mm screen. Material was stored at room temperature in sealed con- In brief, ground, freeze-dried samples (100 mg) were extracted tainers until analysis by Banik et al. (2013). in 1 mL 80% methanol:water, by sonication for 10 min. After The same ground plant material was used for metabolo- centrifugation, 200 μL of supernatant from each sample was mics analysis except that accession numbers 11, 21, 29, placed into an HPLC vial for LCMS analysis. This LCMS and 37, the latter two being replications of the same acces- sample was stored at 4 °C overnight prior to analysis. The sion were missing and not available for this analysis extraction process was repeated on the pellet for a second time. (Table 1). However, 27 accessions of biserrula from differ- Combined extracts were then reduced in volume to less than ent geographic origins and a diverse range of CH output 300 μL by evaporation under N gas whilst heating at 30 °C. 4 2 as well as the two clover controls were still available for Samples were frozen at − 80 °C and then freeze-dried using a metabolomics analysis as shown in Table 1. Christ Freeze Drying System (ice condenser, − 55 °C; vacuum: 0.030 mbar) (Martin Christ Gefriertrocknungsanlagen GmbH, 2.2 In vitro fermentation technique Germany). Freeze-dried samples were stored at − 80 °C over desiccant. Duplicates were prepared for three samples to test Data from Banik et al. (2013) were sorted and processed for method reproducibility. use as the base information for the 27 accessions of biserrula and the control species (Table 1). Briefly, plants were harvest- 2.3.2 LCMS sample analysis ed at 10 weeks. The edible parts of the plants was freeze-dried in a bench-top freeze dryer (BenchTop, VirTis/SP Scientific, LCMS data was acquired in negative ionization mode at high Warminster, PA) and ground in a grinding mill (Glen Creston, resolution. Negative Fourier transform mode lenses were Stanmore, UK). Fermentability was examined in an in vitro tuned to m/z of 739.20 by infusion of sample number 3 when batch fermentation system commonly used to examine plant teed with mobile phase at initial chromatographic conditions. fermentability (Bodas et al. 2008). One day before the testing, HPLC separation was achieved using a 150 mm × 2.1 mm 0.1 g of plant material was weighed and transferred to an ThermoFisher 1.9 μm C18 HPLC column (Thermo Fisher anaerobic chamber (Coy Laboratory Products Inc., Grass Scientific, Australia) fitted to an Agilent series 1290 high- Lake, MI) maintained at 39 °C and supplied with 80% N : performance liquid chromatograph (binary pump, cooled 2 30 Page 4 of 10 Agron. Sustain. Dev. (2018) 38:30 Table 1 Core collection of Accession ID Accession Country Mean methane Methane biserrula and the two control −1 number of origin output (mL g DM) group species, their origin, and in vitro methane output. Methane output Red clover (cv. Redquin) 47 N/A 36.18 Very high was derived from IVFT over short incubations by Banik et al. Subterranean clover 46 N/A 28.38 Very high (2013). High-end methane groups (cv. Woogenellup) are high within biserrula 2004ERI1PEL 43 Eritrea 7.96 High germplasm not at interspecific 2004ERI56PEL 4 Eritrea 7.71 High level 2004ESP19PEL 7, 31 Canary 7.36 High Islands 2006MAR29PEL 1, 35 Morocco 6.31 High 2004ERI38PEL 30 Eritrea 5.75 High 139362 19, 40 Morocco 5.55 High 139363 20, 41 Morocco 5.54 High 143467 12, 38 Italy 5.24 High 139026 22 Greek Islands 4.99 Low GEH71PEL 25 Greek Islands 4.80 Low 143415 10 Spain 4.65 Low 139058 2 Greek Islands 4.60 Low 2004ESP64PEL 9, 44 Spain 4.50 Low 139049 18, 45 Greek Islands 4.13 Low 2006ISR20PEL 28, 42 Israel 4.00 Low 143464 11 Italy 3.93 Low 2004ERI37PEL 3 Eritrea 3.91 Low cv. MAURO 15, 39 Italy 3.44 Low 143469 13 Italy 2.57 Low cv. CASBAH 27, 33 Morocco 2.22 Low 143474 14, 32 Italy 2.16 Low 93ITA45PEL 24 Italy 2.06 Low 2005GRC77PEL 5 Greek Islands 1.95 Low GEH77PEL 26, 36 Greek Islands 1.64 Low 93FRA4PEL 23 France 1.49 Low 2004ESP39PEL 8, 34 Canary 1.07 Low Islands 143267a 16 Italy 0.91 Low 143267b 17 Italy 0.68 Low 2006MAR22PEL 29, 37 Morocco 0.66 Low 138972 21 Morocco 0.54 Low auto-sampler maintained at 4 °C, column heater maintained at MS2 and MS3 analyses were carried out on selected sam- 30 °C and photodiode array detector) utilizing a gradient elu- ples using a data-dependent approach with accurate mass in tion, (98 to 0%A over 20 min followed by a 5 minute wash MS1 and MS2 set at resolution of 15,000 for MS2 and MS3. with 100% B before returning to the initial gradient conditions Dynamic exclusion was engaged with a repeat duration of 30 s for 4 minutes; A: 0.1% formic acid in water; B: 0.1% formic and an exclusion duration of 180 s. −1 acid in acetonitrile, at 0.3 ml min ). The HPLC was connect- ed to a ThermoFisher LTQ Velos orbitrap mass spectrometer 2.3.3 LCMS data analysis (Thermo Fisher Scientific, USA), operating with a heated electrospray ionization probe operating in a negative FT mode Initial data processing was carried using Expressionist Refiner over 100–2000 m/z. The heated capillary was maintained at MS (Genedata, Basel, Switzerland) to align MS data, carry out 300 °C with a source heater temperature of 350 °C and the noise reduction, and for peak picking. Cluster analysis was sheath, auxiliary, and sweep gases at 37, 13, and 7 units re- undertaken to group isotope peaks and singletons removed. spectively. The source voltage was 3.3 kV. The cluster volume was analyzed statistically in Expressionist Agron. Sustain. Dev. (2018) 38:30 Page 5 of 10 30 Analyst (Genedata, Basel, Switzerland). Samples were classi- replicates and sample 2 were removed and the residual solvent fied into groups according to methanogenic potential and by (2.48–2.55 ppm) was deleted. This data was used in all subse- country of origin (Table 1). Principal components analysis quent PCA analyses. Samples were classified into groups ac- (PCA) and hierarchical cluster analysis (HCA)—Manhattan— cording to methanogenic potential (Table 1). were used to explore the relationship between the accessions. For both methods valid values were set to a minimum of 20%. To investigate specific features that were elevated in the low 3 Results and discussion methanogenic potential group compared to the high group, a t test was used with 30% valid values per group with Initially, the “high” and “low” categories in methanogenic bootstrapping (10 repeats). P values were adjusted using potential assigned in Table 1 involved the drawing of a Bonferonni’s correction (P = 0.05 equivalent to P =7.7e-6). In somewhat arbitrary line through a continuous distribution. order to identify specific metabolomic features that could be However, analysis of the LCMS data by unsupervised contributing to the low methanogenic potential of the biserrula modeling techniques including PCA and HCA revealed accessions, the LCMS data was re-examined. The clover sam- clear trends in the data. In the PCA scores plot (data not ples were omitted from analysis so that only the high and low shown), each accession was classified as having low or high biserrula accessions were considered. These were then analyzed (within biserrula’s already low range of methanogenesis), by a t test. The features that were highly statistically significant and very high (in two clovers) methanogenic potential and that had an effect size of ten or greater (elevated in the low based on the in vitro CH output (Banik et al. 2013). In CH accessions) were selected as the most likely candidate general, the accessions cluster based on methanogenic po- features with CH lowering potential. tential with low and high groups separated on PC1 and the very high group separated on PC2. The clear outlier here 2.4 NMR sample preparation was sample 3, which did not appear to cluster according to CH group. Generally, the HCA showed that the clustering Freeze-dried sample extracts were re-dissolved for NMR anal- of the accessions by their methanogenic potential was in ysis in 600 uL of DMSO-d6 by vortexing and sonicating. line with their origin (Fig. 2). Samples were centrifuged for 10 min at 14,000 rpm and Initial analysis of NMR data with all accessions and control 20 °C. A total of 550 uL was transferred to NMR tubes for clover species using PCA revealed a similar pattern to the analysis. LCMS results with low and high accessions separating on PC1 and very high accessions separating on PC2 with the 2.4.1 NMR analysis expected graduation across the two groups (not shown here). Accession 3 still stayed with its contrasting CH output group. NMR data was acquired on a Bruker 700-MHz instrument Results obtained from both LCMS and NMR PCA suggest equipped with a cryoprobe and an autosampler. The that high and low CH output accessions are, to a large extent, noesypr1d pulse sequence was used with a spectral width of two chemically distinct groups within the core collection of 18.6 ppm, 256 scans (8 dummy scans), and a total acquisition biserrula. Further, hierarchical cluster analysis indicates that time of 1.26 s. The data were processed in Topspin 3.2 (Bruker within the high CH output group of accessions, the metabolic Biospin, Germany) with a line broadening of 0.3 Hz. Spectra profile of accessions from Eritrea is clearly distinct from acces- were referenced to DMSO (2.5 ppm). 2D NMR data was sions of the northern Mediterranean region. This supports the acquired for samples 4 (high) and 14 (low). The pulse se- previous suggestion (Ghamkhar et al. 2012) that the Eritrean quences used were socygpqf, hsqcedetgpsisp 2.2, and accessions are a differentiated group of accessions within B. hmbcgplpndqf. pelecinus germplasm and distinctly different from the other accessions of biserrula, in agro-morphological traits, eco- 2.4.2 NMR data analysis geographical distribution, and genomic profile. However, this study suggests that they still share some metabolites with some Proton spectra were imported into MatLab (R2015b, accessions from Morocco and Canary Islands, regions geo- Mathworks) using a modified Prometab script, as previously graphically isolated from the northern Mediterranean region. described (Rochfort et al. 2015). Spectra were baseline corrected It can be seen that the outlier in PCA (accession 3) clusters and normalized (total spectral area set to one) in PLSToolbox with its respective eco-geographical origins after the HCA (Ver 8.0.2, Eigenvector Research). The spectra were analyzed (Fig. 2), while the latter analysis identified 6484 features in by PCA with mean centering pre-processing. This analysis con- metabolomics data. It separated the control species more firmed that the replicate samples (for 17, 36, and 46) clustered clearly from the biserrula accessions (Fig. 2). It also separated closely to each other and revealed the spectrum of sample 2 to be the accessions with higher CH output from the ones with an outlier due to poor spectral quality. The spectra from lower CH output, although there were three exceptions. 4 30 Page 6 of 10 Agron. Sustain. Dev. (2018) 38:30 Methane output Geographic origin Subterranean clover Very Red clover high Eritrea High Low Morocco Canary Islands Morocco Canary Islands Italy - Sardinia Spain Italy - Sardinia France Italy - Sardinia Morocco Spain Italy - Sardinia Morocco Greece Israel Fig. 2 Hierarchical clustering of liquid chromatography mass coded methane output (arrow indicating accession 3). Countries of spectrometry (LCMS) data in the accessions of biserrula and the two origin of the accessions are projected on the clusters clover control species. Accessions are clustered based on their color- Accession number 20 (same accession as 41) with a high CH Italy), and the eastern Mediterranean region or cluster D output was clustered with the lower CH output cluster. Also, (Greece and Israel) (Fig. 2). as in the PCA, accession 3 was grouped with the high emitters Among the low methanogenic accessions, Greek acces- but it was clustered with its geographical group (Fig. 2). In sions have a similar metabolic profile with the Israeli acces- general, this analysis, although on metabolites, indicates a sions as they form a cluster together. Using agro- pattern from high CH output to low CH output accessions. morphological characteristics and molecular diversity 4 4 When the geographical site of the accessions is added to the profiles, Ghamkhar et al. (2012) concluded that accessions equation, Eritrea is identified as the hotspot of accessions with from the western and eastern Mediterranean regions can be high CH output. A change in CH output, from high to low, distinguished as two separate categories. In this study, the two 4 4 was observed in the collection sites from Africa to Europe via regions are distinct in their metabolomics profile as well, al- Morocco and the Canary Islands. The dendrogram was sub though Eritrean accessions are, again, an exception. However, divided into four separate clusters (Fig. 2). Clovers were clear- within cluster B, the East-West separation is also true. ly separate from all biserrula accessions (cluster “A” in Fig. 2). The discovery of 47 candidate metabolites (Table 2)that All accessions from Eritrea clustered, firstly, together and may affect CH output in the rumen is a starting point for then, distinctly with some Moroccan and Canary Islands ac- further investigation into these candidate metabolites as cessions. The rest of the accessions could be geographically markers for this trait in other pasture species. It also provides categorized to two regions: the western Mediterranean region an opportunity to discover candidate genes for controlling this or cluster C (Morocco, Canary Islands, Spain, France, and trait in a reverse genetics approach, from metabolites to the Agron. Sustain. Dev. (2018) 38:30 Page 7 of 10 30 Table 2 Clusters elevated in Cluster Effect size Adjusted p Mass m/z [M-H] RT (min) Possible molecular formula accessions of the biserrula core collection with low compared to 1 10.0 0.00051 858.4961 428.2408 8.35 C H O high methanogenic potential 22 36 8 2 19.9 0.00003 650.4024 649.3951 8.41 C H O 36 57 10 3 19.4 0.00005 652.4175 651.4102 8.42 4 15.1 0.00279 680.4483 679.4410 9.17 C H O 38 63 10 5 20.0 0.00006 684.4436 683.4363 8.72 6 14.7 0.00059 690.4115 689.4042 7.17 7 47.5 0.00249 696.4440 695.4368 7.44 C H O 38 63 11 8 12.9 0.00001 696.4441 695.4368 7.55 C H O 38 63 11 9 402.7 0.00020 696.4443 695.4370 7.73 C H O 38 63 11 10 151.8 0.00007 698.4234 697.4161 7.22 C H O 37 61 12 11 43.0 0.00026 700.4383 699.4310 7.17 C H O 37 63 12 12 34.4 0.00031 700.4393 699.4320 7.40 C H O 37 63 12 13 14.1 0.00169 706.4286 705.4213 12.33 C H O 39 61 11 14 10.5 0.00054 722.4604 721.4531 9.29 C H O 40 65 11 15 13.3 0.00153 724.4376 723.4303 9.17 C H O 39 63 12 16 8.8 0.00252 728.3089 727.3017 3.84 C H O 31 51 19 17 138.6 0.00026 740.4328 739.4255 7.44 C H O 39 63 13 18 17.2 0.00003 740.4332 739.4259 7.55 C H O 39 63 13 19 106.6 0.00005 740.4332 739.4260 7.73 C H O 39 63 13 20 12.7 0.00015 752.4703 751.4630 8.46 C H O 41 67 12 21 32.7 0.00298 768.4028 767.3955 7.17 C H O /C H O 50 55 7 32 63 20 22 11.0 0.00014 782.4810 781.4737 7.98 C H O 42 69 13 23 11.0 0.00008 782.4811 781.4738 8.63 C H O 42 69 13 24 25.3 0.00212 794.4828 793.4755 10.33 C H O 43 69 13 25 16.5 0.00004 796.4541 795.4468 10.28 C H O 49 63 9 26 19.6 0.00054 798.4752 797.4679 8.25 C H O 42 69 14 27 21.3 0.00115 816.4862 815.4790 6.27 C H O 42 71 15 28 33.0 0.00155 830.4803 829.4731 8.03 29 11.3 0.00020 830.5010 829.4937 8.11 30 11.2 0.00114 838.5149 837.5076 9.41 31 10.0 0.00015 846.4966 845.4893 7.81 C H O 43 73 16 32 29.3 0.00169 862.4916 861.4843 6.27 C H O 43 73 17 33 29.4 0.00003 862.4913 861.4840 6.44 34 28.2 0.00001 884.4711 883.4638 8.08 C H O 52 67 12 35 72.0 0.00022 902.4854 901.4781 6.71 C H O 45 73 18 36 342.2 0.00032 902.4862 901.4790 6.51 37 12.0 0.00065 922.4543 921.4471 10.50 C H O 47 69 18 38 9.2 0.00064 926.5208 925.5135 9.96 C H O /C H O 55 73 12 48 77 17 39 11.7 0.00023 926.4872 925.4799 9.42 40 13.7 0.00028 972.4871 971.4798 7.96 C H O 55 71 15 41 98.2 0.00069 1014.6110 1013.6040 7.54 C H O 53 89 18 42 18.3 0.00000 1046.3781 1045.3710 5.40 C H O 54 61 21 43 15.9 0.00019 1096.6862 1095.6790 7.78 C H O 66 95 13 44 53.3 0.00004 1098.7047 1097.6970 7.84 C H O 66 97 13 45 52.1 0.00148 1098.7025 1097.6953 8.03 C67 H99 O14 46 43.1 0.00016 1128.7149 1127.7080 7.82 C66 H97 O13 47 71.1 0.00046 1308.8670 1307.8600 7.16 C72 H123 O20 30 Page 8 of 10 Agron. Sustain. Dev. (2018) 38:30 transcriptome and eventually the genome by investigating en- the positive loadings. The aromatic methine resonances are zymes that control the biochemical pathways where these elevated in the high accessions although, again, this is a markers are most active/present. Ursolic acid, a candidate crowded spectral region and there are some resonances con- marker of interest in this study, is a pentacyclic triterpene tributing to the positive loadings as well. saponin and a phytosterol. It is a biochemical component with Oleanolic acid and ursolic acid and their derivatives have cytotoxic activity and also used as an emulsifier in pharma- anti-inflammatory and antioxidant effects as well (Pollier and ceuticals, cosmetics, and foods, similar to pomolic acid found Goossens 2012). In plants, specifically, they exist in the epi- in the wax coats of apples (He and Liu 2007). Both ursolic cuticular waxes as a first defense barrier against pathogens acid and its isomer, oleanolic terpene, identified as the two (Heinzen et al. 1996) and may function as defense compounds main candidate markers for CH output mitigation in rumen, against pathogens or as allelopathic agents (Szakiel et al. originate from the oleanyl cation in biosynthesis. 2005). This suggests a potentially crucial role for these com- The loadings plot for PC1 (Fig. 3a) reveal some clear met- ponents as key biomolecular components to be added to future abolic signatures that are responsible for the discrimination on vaccines against high methanogenesis in the rumen, possibly PC1 and so between the low and high groups. The resonances through their antibacterial (Pollier and Goossens 2012)or between 0 and 2.5 ppm are typical of terpenes with methyl anti-proliferative (Neto 2007) activity. However, Banik et al. singlets resonating between 0.7 and 1.2 ppm. Methylene and (2016) suggested that rumen microflora can change or adapt methine resonances between 0.2 and 2.4 are separated and over time in response to biserrula substrate; therefore, the well resolved, typical of protons on more complex cyclic short-term bioassay data has to be used with some caution. structures such as diterpenes and triterpenes. These terpene Ultimately, in vivo and in-field testing is required to confirm resonances are elevated in the spectra from the low CH ac- these properties. cessions. The majority of the carbohydrate residues are reso- nating between 2.8 and 4.2 ppm with anomeric methines res- 3.1 Data mining for specific metabolites for CH4 onating between 3.8 and 5.5 ppm. This is a crowded region of output the spectrum and there are resonances contributing to both the positive and negative loadings. The olefinic methines between A total of 47 features were identified (Table 2) that were sig- 5.2 and 5.5 ppm seem to be contributing to the separation on nificantly elevated in the low group (P < 0.05 and 10 fold or OH OH OH HO OH O O O O HO OH OH OH PC 1 (26.01%) - - Chemical Formula: C H O 36 57 10 Chemical Formula: C H O 30 47 5 a Exact Mass: 649.3957 Exact Mass: 487.3429 Observed Mass: 649.3947 Observed Mass: 487.3436 OH OH HO OH O HO OH b O O O O d OH OH OH OH - - Chemical Formula: C H O Chemical Formula: C H O 36 55 9 35 57 8 Exact Mass: 631.3852 Exact Mass: 605.4059 Observed Mass: 631.3890 Observed Mass: 605.4063 O O MS3 HO OH OH - - Chemical Formula: C H O Chemical Formula: C H O 29 45 2 29 47 3 Exact Mass: 443.3531 Exact Mass: 425.3425 Observed Mass: 443.3534 Observed Mass: 425.3424 ppm Fig. 3 a Principal component analysis loadings plot of NMR data for features can be identified that may play a role in diversity of the biserrula. The resonances on the positive side of the Y axis are methanogenic potential of the accessions: a–e. b MSn analysis of elevated in the accessions with low methanogenic potential (upward m/z 649.3957 where the parent ion (m/z 649.3957) generates 5 arrow) and on the negative side of the Y axis in accessions with high fragments in MS2 and the MS2 base peak (m/z 443.3534) generates methanogenic potential (downward arrow). Several structural m/z 425.3424 in MS3 Agron. Sustain. Dev. (2018) 38:30 Page 9 of 10 30 greater change in cluster volume). These 47 features corre- and isolation of metabolites for NMR structure determination spond to metabolites of relatively high molecular weight, be- to confirm this. Several of the proposed molecular formulae tween 650 and 1310 amu. Accurate mass measurement was do not have a match in databases (Scifinder, FoodDB, etc.) used to predict the most probable molecular formulae. suggesting that it is very likely that some of these metabolites However, even with a mass error of 3–5 ppm, allowed there will be novel biomarkers for methane mitigation. are over 30 possible formulae for many of the ions. For this reason, and considering both the NMR and MS2 data, formulae were restricted to CHO composition. This allowed 4 Conclusion putative molecular formulae to be proposed. Joblin (1999) stated that the management of H production In conclusion, metabolomic and methanogenic profiles of in the rumen is the most important factor to be considered the biserrula core collection are very congruent. We suggest when developing strategies to control ruminant CH emis- that metabolites discovered in this study and most likely the sions. It should therefore be possible to reduce CH produc- ursolic or oleanolic terpene glycosides affect the methano- tion by inhibiting H -liberating reactions or by promoting al- genic potential in the rumen and the novel ones can be po- ternative H -using reactions or routes for disposing of H dur- 2 2 tentially used for vaccine manufacturing in the future. ing fermentation. An achievement in reducing CH produc- Further mass spectrometry analysis of samples with CH tion in livestock farms would be the development of a vaccine output measurements in the field will be needed to confirm containing an antigen derived from methanogenic bacteria the relevance and true effects of the discovered metabolic (Gworgwor et al. 2006) and an immunogenic preparation markers. The results obtained from all datasets suggest that which reduces the activity of rumen protozoa (Baker et al. the Eritrean accessions likely belong to a different taxon 2004). Such vaccines have the potential to provide a cost- within the species B. pelecinus with potentially different effective treatment to reduce CH emission and enhance ani- agronomic traits. Finally, crossing the low and high methan- mal production. The candidate metabolites are also potential ogenic accessions of biserrula investigated in this study will precursors for future vaccine development if their inhibiting lead to materials and populations for gene discovery of CH properties on methanogenic bacteria are proved. output through transcriptomics and reverse genomics tools MS data was acquired in a data independent mode in sep- and techniques in the near future. arate analyses. Useful MS data was acquired on only a few of the 47 potential biomarkers. The best data was obtained for m/ Acknowledgements The authors would like to thank Vilnis Ezernieks of z 649.3957. The parent ion (m/z 649.3957) generates five the Department of Economic Development, Victoria, for MS lab experiments. main fragments in MS (m/z 631.3890, 605.4063, 487.3436, 443.3534, and 425.3424). In addition, the MS base peak (m/z Compliance with ethical standards 443.3534) generates m/z 425.3424 in MS . Accurate mass measurements on the fragment ions greatly facilitated the as- Conflict of interest The authors declare that they have no conflict of signment of the molecular formulae of this molecule and interest. allowed determination of the molecular formulae of the frag- ments. The mass and fragmentation pattern are consistent with an ursolic or oleanolic terpene glycoside. The fragmentation Open Access This article is distributed under the terms of the Creative of this molecule follows the same general trend reported by Commons Attribution 4.0 International License (http:// Xia et al. (2015). In particular, the observation of the loss of creativecommons.org/licenses/by/4.0/), which permits unrestricted use, CO to form the ion at m/z 605.4063 confirmed that the gly- distribution, and reproduction in any medium, provided you give cosylation cannot occur at this position and also confirms the appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. presence of the carboxylic acid moiety in the molecule. The regiochemistry of the molecule cannot be defined by the MS data, hence the structure may be an ursolic or oleanolic ter- pene. The NMR data is consistent with this interpretation in References that there are resonances in the regions appropriate for both the carboxylic acids and terpene glycosides (Fig. 3b). Indeed, Baker SK, Gnanasampanthan G, Purser DB, Hoskinson RM (2004) the NMR data suggests there are many such molecules in the Immunogenic preparation and method for improving the productiv- ity of ruminant animals. Commonwealth Scientific and Industrial accessions. However, the NMR data is complex and no single Research Organization (CSIRO), U.S. Patent 6,689,354 molecule could be positively identified. Banik BK, Durmic Z, Erskine W, Nichols P, Ghamkhar K, Vercoe P (2013) Although the remaining biomarkers remain unidentified, it Variability of in vitro ruminal fermentation and methanogenic poten- seems very likely that they are also terpene glycosides. Further tial in the pasture legume biserrula (Biserrula pelecinus L.). Crop work will need to be done with both targeted MS approaches Pasture Sci 64(4):409–416. https://doi.org/10.1071/CP13073 30 Page 10 of 10 Agron. Sustain. Dev. (2018) 38:30 Banik BK, Durmic Z, Erskine W, Revell CK, Vadhanabhuti J, Martin C, Morgavi D, Doreau M (2010) Methane mitigation in rumi- McSweeney CS, Padmanabha J, Flematti GR, Algreiby AA, nants: from microbe to the farm scale. Animal 4(03):351–365. Vercoe PE (2016) Bioactive fractions from the pasture legume https://doi.org/10.1017/S1751731109990620 Biserrula pelecinus L. have an anti-methanogenic effect against Mitsumori M, Sun W (2008) Control of rumen microbial fermentation for key rumen methanogens. Anaerobe 39:173–182. https://doi.org/10. mitigating methane emissions from the rumen. Asian Australas J 1016/j.anaerobe.2016.04.004 Anim Sci 21(1):144–154 Bayaru E, Kanda S, Kamada T, Itabashi H, Andoh S, Nishida T, Ishida M, Moss AR, Jouany JP, Newbold J (2000) Methane production by rumi- Itoh T, Nagara K, Isobe Y (2001) Effect of fumaric acid on methane nants: its contribution to global warming. Ann Zootech 49(3):231– production, rumen fermentation and digestibility of cattle fed rough- 253. https://doi.org/10.1051/animres:2000119 age alone. Anim Sci J 72(2):139–146. https://doi.org/10.2508/ Neto CC (2007) Cranberry and its phytochemicals: a review of in vitro chikusan.72.139 anticancer studies. J Nutr 137(1):186S–193S Bodas R, López S, Fernandez M, García-González R, Rodríguez A, Patra AK, Saxena J (2009) Dietary phytochemicals as rumen mod- Wallace R, González J (2008) In vitro screening of the potential of ifiers: a review of the effects on microbial populations. Antonie numerous plant species as antimethanogenic feed additives for ru- Van Leeuwenhoek 96(4):363–375. https://doi.org/10.1007/ minants. Anim Feed Sci Technol 145(1):245–258. https://doi.org/ s10482-009-9364-1 10.1016/j.anifeedsci.2007.04.015 Patra A, Kamra D, Agarwal N (2006) Effect of plant extracts on in vitro Ghamkhar K, Revell C, Erskine W (2012) Biserrula pelecinus L.—ge- methanogenesis, enzyme activities and fermentation of feed in ru- netic diversity in a promising pasture legume for the future. Crop men liquor of buffalo. Anim Feed Sci Technol 128(3):276–291. Pasture Sci 63(9):833–839. https://doi.org/10.1071/CP12126 https://doi.org/10.1016/j.anifeedsci.2005.11.001 Ghamkhar K, Rochfort S, Banik B, Revell C (2016) Metabolomics for Pollier J, Goossens A (2012) Oleanolic acid. Phytochemistry 77:10–15. methane mitigation: a case study in Biserrula (Biserrula pelecinus https://doi.org/10.1016/j.phytochem.2011.12.022 L.). In: Roldán-Ruiz I, Baert J, Reheul D (eds) Breeding in a world Rochfort S, Ezernieks V, Mele P, Kitching M (2015) NMR metabolomics of scarcity: proceedings of the 2015 Meeting of the Section “Forage for soil analysis provide complementary, orthogonal data to MIR Crops and Amenity Grasses” of Eucarpia. Springer International and traditional soil chemistry approaches—a land use study. Magn Publishing, Cham, pp 271–275 Reson Chem 53(9):719–725. https://doi.org/10.1002/mrc.4187 Gworgwor Z, Mbahi T, Yakubu B (2006) Environmental implications of Sejian V, Lal R, Lakritz J, Ezeji T (2011) Measurement and prediction of methane production by ruminants: a review. J Sustain Dev Agric enteric methane emission. Int J Biometeorol 55(1):1–16. https://doi. Environ 2:1 org/10.1007/s00484-010-0356-7 He X, Liu RH (2007) Triterpenoids isolated from apple peels have potent Swinny E, Revell C, Campbell N, Spadek E, Russo C (2015) In search of antiproliferative activity and may be partially responsible for apple’s photosensitising compounds in the annual forage legume Biserrula anticancer activity. J Agric Food Chem 55(11):4366–4370. https:// pelecinus L. Crop Pasture Sci 66(11):1161–1166. https://doi.org/10. doi.org/10.1021/jf063563o 1071/CP14342 Heinzen H, de Vries JX, Moyna P, Remberg G, Martinez R, Tietze LF (1996) Mass spectrometry of labelled triterpenoids: thermospray Szakiel A, Ruszkowski D, Janiszowska W (2005) Saponins in Calendula and electron impact ionization analysis. Phytochem Anal 7(5): officinalis L.—structure, biosynthesis, transport and biological ac- 237–244. https://doi.org/10.1002/(SICI)1099-1565(199609)7: tivity. Phytochem Rev 4(2–3):151–158. https://doi.org/10.1007/ 5<237::AID-PCA310>3.0.CO;2-M s11101-005-4053-9 Helander IM, Alakomi H-L, Latva-Kala K, Mattila-Sandholm T, Pol I, Ulyatt M, Lassey K, Shelton I, Walker C (2002) Methane emission from Smid EJ, Gorris LG, von Wright A (1998) Characterization of the dairy cows and wether sheep fed subtropical grass-dominant pas- action of selected essential oil components on Gram-negative bac- tures in midsummer in New Zealand. N Z J Agric Res 45(4):227– teria. J Agric Food Chem 46(9):3590–3595. https://doi.org/10.1021/ 234. https://doi.org/10.1080/00288233.2002.9513513 jf980154m Wallace RJ, McEwan NR, McIntosh FM, Teferedegne B, Newbold CJ Holter J, Young A (1992) Methane prediction in dry and lactating (2002) Natural products as manipulators of rumen fermentation. Holstein cows. J Dairy Sci 75(8):2165–2175. https://doi.org/10. Asian Australas J Anim Sci 15(10):1458–1468 3168/jds.S0022-0302(92)77976-4 Woodward SL, Waghorn GC, Lassey KR, Laboyrie PG (2002) Does feed- Joblin K (1999) Ruminal acetogens and their potential to lower ruminant ing sulla (Hedysarum coronarium) reduce methane emissions from methane emissions. Crop Pasture Sci 50(8):1307–1314. https://doi. dairy cows? Proceedings-New Zealand Society of Animal Production org/10.1071/AR99004 62:227–230 Johnson KA, Johnson DE (1995) Methane emissions from cattle. J Anim Xia Y-G, Liang J, Li G-Y, Yang B-Y, Kuang H-X (2015) Analysis of Sci 73(8):2483–2492. https://doi.org/10.2527/1995.7382483x oligosaccharide sequences of trace Caulophyllum robustum sapo- Lee SS, Hsu J-T, Mantovani HC, Russell JB (2002) The effect of bovicin nins by direct infusion multiple-stage tandem mass spectrometry. J HC5, a bacteriocin from Streptococcus bovis HC5, on ruminal Pharm Biomed Anal 112:106–115. https://doi.org/10.1016/j.jpba. methane production in vitro. FEMS Microbiol Lett 217(1):51–55. 2015.04.024 https://doi.org/10.1111/j.1574-6968.2002.tb11455.x
Agronomy for Sustainable Development – Springer Journals
Published: May 30, 2018
It’s your single place to instantly
discover and read the research
that matters to you.
Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.
All for just $49/month
Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly
Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.
Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.
Read from thousands of the leading scholarly journals from SpringerNature, Wiley-Blackwell, Oxford University Press and more.
All the latest content is available, no embargo periods.
“Hi guys, I cannot tell you how much I love this resource. Incredible. I really believe you've hit the nail on the head with this site in regards to solving the research-purchase issue.”Daniel C.
“Whoa! It’s like Spotify but for academic articles.”@Phil_Robichaud
“I must say, @deepdyve is a fabulous solution to the independent researcher's problem of #access to #information.”@deepthiw
“My last article couldn't be possible without the platform @deepdyve that makes journal papers cheaper.”@JoseServera