Effects of Fumarate and Nitroglycerin on In Vitro Rumen Fermentation, Methane and Hydrogen Production, and on Microbiota

Simple Summary An important strategy to mitigate global warming is to reduce methane produced by ruminants. However, the inhibition of rumen methanogenesis generally results in hydrogen accumulation, which would affect the normal fermentation in the rumen. In this study, we used a combination of two chemicals (fumarate and nitroglycerin) to mitigate the rumen methane production. Nitroglycerin inhibits the activities of methanogens. Fumarate eliminates hydrogen accumulation. In vitro rumen fermentation was used to investigate the effects of this combination on rumen fermentation, methane and hydrogen production, and microbiota. The results showed that the addition of fumarate decreased the hydrogen accumulation and increased the concentration of propionate and microbial crude protein when methanogen activities were inhibited by nitroglycerin. The bacterial and archaeal communities were altered by the addition of the two chemicals, with several taxa changed in the relative abundance. Conclusively, the combination of fumarate and nitroglycerin inhibited methane production, reduced hydrogen accumulation, improved rumen fermentation and altered rumen microbiota. This study provides an alternative way of using these chemicals in order to mitigate methane emission in ruminants. Abstract This study aimed to investigate the effects of fumarate and nitroglycerin on rumen fermentation, methane and hydrogen production, and microbiota. In vitro rumen fermentation was used in this study with four treatment groups: control (CON), fumarate (FA), nitroglycerin (NG) and fumarate plus nitroglycerin (FN). Real-time PCR and 16S rRNA gene sequencing were used to analyze microbiota. The results showed that nitroglycerin completely inhibited methane production and that this resulted in hydrogen accumulation. Fumarate decreased the hydrogen accumulation and improved the rumen fermentation parameters. Fumarate increased the concentration of propionate and microbial crude protein, and decreased the ratio of acetate to propionate in FN. Fumarate, nitroglycerin and their combination did not affect the abundance of bacteria, protozoa and anaerobic fungi, but altered archaea. The PCoA showed that the bacterial (Anosim, R = 0.747, p = 0.001) and archaeal communities (Anosim, R = 0.410, p = 0.005) were different among the four treatments. Compared with CON, fumarate restored Bacteroidetes, Firmicutes, Spirochaetae, Actinobacteria, Unclassified Ruminococcaceae, Streptococcus, Treponema and Bifidobacterium in relative abundance in FN, but did not affect Succinivibrio, Ruminobacter and archaeal taxa. The results indicated that fumarate alleviated the depressed rumen fermentation caused by the inhibition of methanogenesis by nitroglycerin. This may potentially provide an alternative way to use these chemicals to mitigate methane emission in ruminants.


Introduction
Methane (CH 4 ) is a major contributor to global climate change [1]. About 40% of greenhouse gas emissions from livestock production can be attributed to ruminant CH 4 production, which accounts for approximately 6% of global anthropogenic greenhouse gas emissions [2]. Ruminal CH 4 production does not only concern greenhouse gas emissions, but also relates to energy loss for ruminants (up to 12% of the total energy intake) [3]. The rumen is rich in bacteria, protozoa, fungi and methanogens, which can ferment coarse feedstuffs to produce volatile fatty acids, carbon dioxide and CH 4 . Hydrogen (H 2 ) is an important intermediate in most of those biochemical processes [4]. CH 4 is generally produced through the utilization of CO 2 and H 2 by methanogenic archaea in the rumen [5]. The accumulation of H 2 could affect the normal rumen fermentation, therefore, CH 4 generation plays an important role in H 2 elimination in the rumen.
Strategies like the use of feed additives, nutrition management and animal genetic improvement have been proposed for use in reducing CH 4 emissions from ruminants [6][7][8]. Chemicals such as sulfate, nitrate and fumarate were studied for their potential to reduce rumen CH 4 emissions [9][10][11]. The CH 4 inhibitor 3-nitrooxypropanol [12,13] and the macroalga Asparagopsis taxiformis [14,15] were recently developed as promising rumen CH 4 -mitigating agents. However, the inhibition of methanogenic activity usually results in abnormal rumen fermentation caused by H 2 accumulation. Therefore, it is necessary to devise an alternative means of eliminating H 2 when inhibiting the activity of methanogens. Fumarate is a metabolic intermediate and can be reduced to succinate by H 2 in the rumen. Succinate is then decarboxylated into propionate, which is a major energy source for ruminants [16]. Fumarate is a promising H 2 -comsuming chemical in the rumen. Nitroglycerin, targeting methanogens, has the same functional group as the chemical 3-nitrooxypropanol, which was reported as being an effective means of reducing ruminal CH 4 emissions in in vitro and in vivo studies [17,18]. Its metabolic end products are propionate and ammonia in the rumen.
This study hypothesized that fumarate would alleviate the abnormal rumen fermentation when methanogenesis was inhibited by nitroglycerin. The objective of this study was to investigate the effects of fumarate and nitroglycerin on rumen fermentation, CH 4 and H 2 production and microbiota in an in vitro rumen trial. The results of this study could help to develop an alternative means of mitigating rumen CH 4 emissions.

Experimental Design
In vitro rumen fermentation was carried out with a completely randomized design (CRD) for 4 treatments: control (CON), fumarate at 12 mmol/L (FA), nitroglycerin at 99 µmol/L (NG) and fumarate at 12 mmol/L plus nitroglycerin at 99 µmol/L (FN). The dosages of FN and NG used in this study were determined according to previous studies [19,20] and the results of the pre-experiments. Three replicates were prepared for each Biology 2023, 12, 1011 3 of 13 treatment. Additionally, three independent incubation runs were performed at different times [21]. Each run consisted of 4 treatments with 3 replicates and 4 blanks containing only the inoculum. The experimental procedure was conducted according to the study of Martínez-Fernández et al. [22]. The rumen fluid was collected from 3 Hu sheep before the morning feeding was performed, and this fluid was then pooled and filtered through 4 layers of cheesecloth. The filtered rumen fluid and buffer were mixed thoroughly (1:3 [vol/vol]) in a water bath at 39 • C under anaerobic conditions. Additionally, each 100 mL of the mixture was dispensed into a 180 mL serum bottle containing 1.0 g substrate and chemicals (fumarate or nitroglycerin). All serum bottles were sealed and incubated at 39 • C for 24 h at 80 rpm. After 24 h of incubation, all the fermentation flasks were taken out and put into ice water to terminate the fermentation. Samples were collected and stored for subsequent analysis. The buffer was composed of 8.75 g NaHCO 3 , 1.00 g NH 4

Sample Collection and Chemical Analysis
At the end of the fermentation, the pH value was measured using a pH meter (Ecoscan pH 5, Thermo Fisher Scientific Inc., Singapore). Then, the bottles were immediately put into ice water to stop fermentation. The supernatant of the fermentation fluid was collected and stored at −20 • C for the determination of volatile fatty acids (VFAs), lactate, microbial crude protein (MCP) and ammonia nitrogen (NH 3 -N). The mixture of the substrate and fermentation fluid was collected and stored at −80 • C for the analysis of the microbiota.

Sample Collection and Chemical Analysis
Gas production was assessed using a pressure transducer [24]. Methane (CH 4 ) and hydrogen (H 2 ) production were measured following the gas measurement procedure using a GC-TCD instrument (Agilent 7890B, Agilent Technologies Inc., Santa Clara, CA, USA). Gases were separated on packed GC columns (Porapak Q packing & MolSieve 5A packing, Agilent Technologies Inc., CA, USA) at a column temperature of 80 • C, a 200 • C injection temperature and a 200 • C TCD detector temperature. N 2 was the carrier gas. The VFAs were determined according to Jin et al. [17]. Each 1.0 mL sample was mixed with 0.2 mL deproteinization-acidification solution [metaphosphoric acid (25% w/v) and crotonic acid (0.65% w/v)] before undergoing analysis via gas chromatography (Agilent 7890B instrument, Agilent Technologies Inc., CA, USA). The sample was separated using a fused silica capillary column (Supelco, Bellefonte, PA, USA) with a programmed heating process (110 • C for 3 min, 110-150 (40 • C/min)). The injection temperature was 200 • C. The flame ionization detector temperature was 220 • C. The carrier gas was nitrogen. Lactate was measured using an assay kit in accordance with the instructions of the manufacturer (Jiancheng Bioengineering Research Institute, Nanjing, China). Microbial crude protein was determined with a commercial reagent kit (BCA Protein Assay Kit, Tiandz Inc., Beijing, China) in accordance with manufacturer instructions. The concentration of NH 3 -N was analyzed using an indophenol method with an acidified procedure [25].

DNA Extraction and Real-Time PCR
Genomic DNA was extracted from a 1.0 mL sample using a bead-beating and phenolchloroform-isopentanol extraction method [26]. Each DNA sample was divided into two parts to perform sequencing and real-time PCR.
Archaea, bacteria, anaerobic fungi and protozoa were quantified using an Applied Biosystems 7300 Real time PCR system (Applied Biosystems, Thermo Fisher Scientific Inc.,  Table A1. An SYBR ® Premix Ex Tag TM (TaKaRa, Dalian, China) was used to prepare the reaction mixture. The copy number of DNA in each sample was measured in triplicate, and the average value was calculated. The external standards were prepared with plasmid DNA of clones of each microbial population. The results are expressed as the number of copies of marker genes per milliliter of fermentation liquid.
Fastp (version 0.20.0) and FLASH (version 1.2.7) were used to filter and merge 16S rRNA sequences, and the chimeras were filtered to obtain effective reads [27]. UPARSE (version 7.1) was used to pick up the operational taxonomic unit (OTU) with a 97% similarity truncation value [28]. Taxonomic assignment was performed for bacteria using RDP classifier (version 2.11) based on the SILVA database (version 138), and via the RIM-DB database for methanogens [18]. QIIME 2 was used for alpha diversity analysis. The principal coordinate analysis (PCoA) was conducted based on Bray-Curtis distance [29]. The significance of the differences among groups was assessed with ANOSIM using the vegan package in R.

Statistical Analysis
The analyses of the in vitro fermentation parameters and the real-time PCR data were performed using the MIXED procedure of SAS 9.4 version (SAS Institute, Inc., Cary, NC, USA), and the data were tested to determine their normality using the Shapiro-Wilk test of SAS. The model used for data analysis was Yijk = µ + P i + S j + PS ij + e ij , where Y ijk is the observed value, µ is the overall mean, P i is the fixed effect of treatment with nitroglycerin, S j is the fixed effect of treatment with fumarate, PS ij is the interaction effect of nitroglycerin * fumarate and e ij is the random error. The variables that had non-normal distributions were analyzed using the Kruskal-Wallis test procedure. The Tukey test was used to identify differences (p < 0.05) between means.

Total Gas, Hydrogen and Methane Production
The total gas production in FA was the highest among the four groups (p < 0.05, Table 1), and it was higher in FN than CON and NG (p < 0.05). Hydrogen was accumulated in NG and FN. Additionally, NG had the highest hydrogen production (p < 0.05). Methane was only accumulated in CON and FA. There was no methane detected in NG and FN. Methane production was higher in FA than CON (p < 0.05).

Fermentation Characteristics
The in vitro fermentation characteristics are presented in Table 2. Total VFA was only higher in FA than NG (p < 0.05). Acetate was higher in CON and FA than NG and FN (p < 0.05). Propionate was higher in FA and FN than CON and NG (p < 0.05). The ratio of acetate/propionate in CON was the highest (p < 0.05), and it was higher in FA and NG than FN (p < 0.05). Isobutyrate in FA was the highest (p < 0.05), and it was higher in CON than NG (p < 0.05). Valerate was higher in NG and FN than CON and FA (p < 0.05). Isovalerate in FA was the highest (p < 0.05), and it was higher in CON than NG and FN (p < 0.05).
Ammonia nitrogen in FA was the highest (p < 0.05). Microbial crude protein was higher in CON and FA than NG and FN (p < 0.05), and it was higher in FN than NG (p < 0.05). There were no significant differences in pH, butyrate and lactate among the four groups (p > 0.05).

The Quantification of Protozoa, Bacteria, Anaerobic Fungi and Archaea
There was no significant difference in the abundance of bacteria, protozoa and anaerobic fungi (p > 0.530) among the four groups ( Table 3). The abundance of archaea was higher in FA than CON and NG (p < 0.05), and it was higher in FN than NG (p < 0.05).

Bacterial Community
A total of 390,011 bacterial sequences remained after filtering for quality. The average length was 418 bp. A total of 2722 OTUs were identified. There was no difference in the  Table 4), except for the Shannon index (p = 0.043). There was a clear separation of clusters on the 3D-PCoA plot of bacterial populations among the four groups (Anosim, R = 0.747, p = 0.001, Figure 1A). PC1, PC2 and PC3 accounted for 53.08%, 16.04% and 10.51% of the total variance, respectively. At the phylum level, 20 phyla were identified across all samples. The eight predominant phyla (the average relative abundances of phyla >1% in at least one group) were Bacteroidetes, Firmicutes, Proteobacteria, Spirochaetes, Actinobacteria, Candidate_division_SR1, Candidate_division_TM7 and Cyanobacteria ( Table 5). The relative abundance of Bacteroidetes was higher in FN than NG (p < 0.05), but it did not show significant differences between other groups (p > 0.05). Firmicutes and Actinobacteria were the highest in NG (p < 0.05). Proteobacteria in CON and FA were higher than NG and FN (p < 0.05). Spirochaetae were higher in FA and FN than NG (p < 0.05). Candidate_division_TM7 was the highest in CON (p < 0.05). The relative abundance of Cyanobacteria showed no significant difference among groups (p > 0.05). a-c Means within a row with different superscripts differ (p < 0.05). 1 CON, control group; FA, mmol/L fumarate was added; NG, 99 μmol/L nitroglycerin was added; FN, fumarate and nitrogl erin were added in combination (FA + NG, 12 mmol/L + 99 μmol/L). * Represents an interact effect.

Bacterial Community
A total of 390,011 bacterial sequences remained after filtering for quality. The avera length was 418 bp. A total of 2722 OTUs were identified. There was no difference in t alpha diversity indexes (p > 0.05, Table 4), except for the Shannon index (p = 0.043). The was a clear separation of clusters on the 3D-PCoA plot of bacterial populations among t four groups (Anosim, R = 0.747, p = 0.001, Figure 1A). PC1, PC2 and PC3 accounted 53.08%, 16.04% and 10.51% of the total variance, respectively.   A total of 307 bacterial genera were identified from all samples. The seven predominant genera (the average relative abundances of genera >2% in at least one group) were Streptococcus, Succinivibrio, Ruminobacter, Treponema, Bifidobacterium, Unclassified BS11_gut_group and Unclassified Ruminococcaceae ( Table 6). The relative abundance of Streptococcus and Bifidobacterium was the highest in NG (p < 0.05). Succinivibrio was higher in CON and FA than FN (p < 0.05). Ruminobacter was higher in CON than NG and FN (p < 0.05). Treponema was higher in FN than NG (p < 0.05). Unclassified BS11_gut_group was higher in CON and FA than the other two groups (p < 0.05). Unclassified Ruminococcaceae was lower in NG than CON (p < 0.05).

Archaeal Community
A total of 568, 155 archaeal sequences were obtained after quality filtering. The average length was 355 bp. No difference in the alpha diversity index was found (p > 0.05, Table 4). These sequences were clustered into 198 OTUs. There was a clear separation of clusters on the 3D-PCoA plot of archaeal populations among the four groups (Anosim, R = 0.410, p = 0.005, Figure 1B). PC1, PC2 and PC3 accounted for 73.62%, 8.94% and 5.96% of the total variance, respectively.
A total of two archaeal orders were identified from all samples. The two predominant orders (the average relative abundance of orders > 1% in at least one group) were Methanobacteriales and Methanomassiliicoccales ( Table 7). The relative abundance of Methanobacteriales was the lowest in FN (p < 0.05), and it was lower in FA than CON or NG (p < 0.05). Additionally, there was no significant difference between CON and NG (p > 0.05). The relative abundance of Methanomassiliicoccales was the highest in FN (p < 0.05), and it was higher in FA than CON and NG (p < 0.05).
A total of 47 archaeal species were identified. Six predominant species (the average relative abundance of species > 1% in at least one group) are shown in Table 8. The relative abundance of Methanobrevibacter gottschalkii clade was lower in FN than CON (p < 0.05). Group12 sp. ISO4-H5 was the highest in FN (p < 0.05). Group9 sp. ISO4-G1 was higher in FN than the other groups (p < 0.05), and it was higher in FA than CON and NG (p < 0.05). Table 7. The relative abundances of archaea at the order level (the average relative abundances of orders > 1% in at least one group are presented). a-c Means within a row with different superscripts differ (p < 0.05). 1 CON, control group; FA, 12 mmol/L fumarate was added; NG, 99 µmol/L nitroglycerin was added; FN, fumarate and nitroglycerin were added in combination (FA + NG, 12 mmol/L + 99 µmol/L).

Discussion
The inhibition of methanogenic activities usually results in H 2 accumulation and causes depressed rumen fermentation [30]. This can affect the animal production performance. The results of this study showed that the addition of fumarate alleviated H 2 accumulation and improved the depressed rumen fermentation parameters when methanogenesis was inhibited by nitroglycerin. Nitroglycerin was demonstrated to be effective at reducing rumen methane production in several in vitro and in vivo studies [17,18,20]. It was able to completely inhibit methane production and caused an accumulation of hydrogen in in vitro rumen fermentation [17,20]. Moreover, the final metabolites of nitroglycerin were propionate and ammonia in the rumen, which have no negative effect on rumen fermentation. Therefore, nitroglycerin was used to successfully establish a model of methane depression and hydrogen accumulation in this study. Nitroglycerin caused about 4.8% hydrogen of accumulation (% total gas production). The current experiment also observed that nitroglycerin inhibited methane production, but did not affect the abundance of archaea compared with the control. This result is consistent with that of a previous study [20]. In another study, an opposite result was observed, whereby the abundance of archaea declined when the methane production was depressed by nitroglycerin [17], which was consistent with the research on the other methanogenic inhibitors [31,32]. However, the mechanisms of the different results for archaeal abundance in different studies are unclear and further work is needed to elucidate this point.
Fumarate is an intermediate in the rumen metabolism and is finally reduced to propionate [33]. The reduction of fumarate has a lower H 2 -consuming threshold (0.02 ppm) and produces more Gibbs free energy than the methanogenesis of H 2 and CO 2 . The fumarate reduction should be more effective than methanogenesis in the rumen [34][35][36]. Therefore, fumarate was used as a rumen CH 4 -mitigation agent in many previous studies [37][38][39]. However, the effects of fumarate on rumen CH 4 mitigation were found to be inconsistent. Bayaru et al. [37] observed that CH 4 production in steers was reduced by 23% when fumarate was added to the complete diet at 20 g/kg dry matter. In contrast, no effect was observed in steers fed barley silage and concentrate with fumarate (12 g/kg dry matter) [38] and in lambs fed dried alfalfa with fumarate (100 g/kg dry matter) [39]. Fumarate increased CH 4 production in sheep fed a mixed diet [19], which is consistent with the current study. Fumarate increased the abundance of archaea, methane production and acetate with the absence of nitroglycerin. Fumarate was expected to consume H 2 and reduce methane production; the increase in methane production was not expected. Fumarate can be metabolized into acetate via the malate-pyruvate pathway in the rumen [19]. In this process, there is net [H] produced (C 4 H 4 O 4 + 2H 2 O→C 2 H 4 O 2 + 2CO 2 + 4H), which could account for the increase in the abundance of archaea and methane production. The increased concentration of acetate supported this speculation. Moreover, Gibbs free energy calculation shows that the production of acetate from fumarate under rumen conditions is thermodynamically feasible even at very low fumarate concentrations [35]. Fumarate was metabolized into acetate instead of propionate, which could have occurred as the microbial populations that reduce fumarate to succinate/propionate had not yet been completely established. It may also give an explanation for the inconsistent results obtained in different studies on fumarate.
Fumarate and nitroglycerin altered the relative abundance of Bacteroides and Firmicutes as well as Streptococcus and several unclassified genera to the two phyla. The relative abundance of Streptococcus was increased by nitroglycerin, but restored to the level of that in CON by the addition of fumarate. The underlying mechanism of the changes in Streptococcus is unclear. However, Streptococcus had been reported to produce bacteriocin, which could inhibit methane production [40]. Succinivibrio, belonging to Proteobaceria, produces succinate [41]. The relative abundance of Succinivibrio was decreased by the combination of fumarate and nitroglycerin. Mao et al. [42] reported that the relative abundance of Succinivibrio dextrinisolvens was increased in the rumen of goats fed disodium fumarate. It seems that the combination of the two chemicals had an opposite impact on Succinivibrio. Treponema, belonging to Spirochaetae, produces succinate, formate and acetate [43]. The relative abundance of Treponema was restored by the addition of fumarate. Jin et al. [44] observed an increase in Treponema due to disodium fumarate in an in vitro rumen fermentation. Therefore, Treponema might play a role in the restoration of the propionate concentration in FN. Ruminobacter, belonging to Proteobacteria, is associated with ruminal fiber degradation [45]. Nitroglycerin decreased the proportion of Ruminobacter, suggesting the inhibition of fiber degradation. It might be partly related to the decrease in acetate concentration caused by nitroglycerin. Bifidobacteria, belonging to the phylum of Actinobacteria, were known fermenters of starch and simple sugars [46]. The addition of nitroglycerin increased the relative abundance of Bifidobacteria, but fumarate restored it. The mechanism underlying the changes in Bifidobacteria is unclear.
The most dominant methanogens belonged to Methanobacteriales and Methanomassiliicoccales, which was consistent with the results of previous studies [17,18]. Fumarate decreased the relative abundance of Methanobacteriales and increased that of Methanomassiliicoccales. Members of the Methanobacteriales primarily use H 2 and CO 2 to produce CH 4 , which generates lower amounts of Gibbs free energy and has a higher H 2 -utilizing threshold than fumarate reduction [35]. The fumarate reduction might have decreased the H 2 concentration, which depressed the growth of Methanobacteriales. Members of the Methanomassiliicoccales are H 2 -dependent methyltrophic methanogens which produce more Gibbs free energy and have a lower H 2 -utilizing threshold than Methanobacteriales. Moreover, the repair system of Methanomassiliicoccales seems to be more resilient than that of Methanobacteriales in the presence of nitroglycerin [47]. This might explain the changes in the relative abundance in the two methanogenic orders.

Conclusions
Fumarate decreased hydrogen accumulation and increased the concentration of propionate and MCP in the presence of nitroglycerin. Treatments did not affect the abundance of bacteria, protozoa and anaerobic fungi, but altered the abundance of archaea. Fumarate restored the relative abundance of several bacterial taxa to the levels in CON in the presence of nitroglycerin, but did not affect Succinivibrio, Ruminobacter and archaeal taxa. Collectively, fumarate alleviated the depressed rumen fermentation caused by the inhibition of methanogenesis by nitroglycerin. This might provide an alternative way to use those chemicals to mitigate methane emissions in ruminants. However, further studies are needed in order to evaluate the effects of nitroglycerin combined with fumarate on animal health, production performance, rumen fermentation and the microbial community in vivo.

Data Availability Statement:
The data presented in this study are available from the corresponding author on reasonable request.