Increasing buffering capacity enhances rumen fermentation characteristics and alters rumen microbiota composition of high-concentrate fed Hanwoo steers

The buffering capacity of buffer agents and their effects on in vitro and in vivo rumen fermentation characteristics, and bacterial composition of a high-concentrate fed Hanwoo steers were investigated in this study. Treatments were comprised of CON (no buffer added), BC0.3% (low buffering capacity, 0.3% buffer), BC0.5% (medium buffering capacity, 0.5% buffer), and BC0.9% (high buffering capacity, 0.9% buffer). Four Hanwoo steers in a 4 × 4 Latin square design were used for the in vivo trial to assess the effect of treatments. Results on in vitro experiment showed that buffering capacity, pH, and ammonia–nitrogen concentration (NH3-N) were significantly higher in BC0.9% and BC0.5% than the other treatments after 24 h incubation. Individual and total volatile fatty acids (VFA) concentration of CON were lowest compared to treatment groups. Meanwhile, in vivo experiment revealed that Bacteroidetes were dominant for all treatments followed by Firmicutes and Proteobacteria. The abundances of Barnesiella intestinihominis, Treponema porcinum, and Vibrio marisflavi were relatively highest under BC0.9%, Ruminoccocus bromii and Succiniclasticum ruminis under BC0.5%, and Bacteroides massiliensis under BC0.3%. The normalized data of relative abundance of observed OTUs’ representative families have grouped the CON with BC0.3% in the same cluster, whereas BC0.5% and BC0.9% were clustered separately which indicates the effect of varying buffering capacity of buffer agents. Principal coordinate analysis (PCoA) on unweighted UniFrac distances revealed close similarity of bacterial community structures within and between treatments and control, in which BC0.9% and BC0.3% groups showed dispersed community distribution. Overall, increasing the buffering capacity by supplementation of BC0.5% and and BC0.9% buffer agents enhanced rumen fermentation characteristics and altered the rumen bacterial community, which could help prevent ruminal acidosis during a high-concentrate diet.


Results
Effect of different buffering capacities on in vitro rumen fermentation parameters. The buffering capacity of BC 0.9% and BC 0.5% were significantly greater (P < 0.05) after 24 h incubation compared to BC 0.3% and CON (Table 1). Both BC 0.9% and BC 0.5% exhibited significantly higher (P < 0.05) buffering capacity. BC 0.9% and BC 0.5% showed higher (P < 0.05) ruminal pH than other treatments throughout the incubation period. Total gas production increased significantly (P < 0.05) in BC 0.9% , BC 0.5% , and BC 0.3% after 12 h, maintaining the CON as the lowest gas produced (182.67 ml, 179.00 ml, 187.00 ml, and 169.00 ml, respectively). The NH 3 -N concentrations for BC 0.9% and BC 0.5% were significantly higher (P < 0.05) than BC 0.3% and CON after 3 h incubation. However, no significant differences were observed between treatments during the 6 and 12 h incubation periods. After 24 h incubation, production of NH 3 -N was observed to be higher under BC 0.9% , followed by BC 0.5%, BC 0.3% , and CON.
Significantly higher concentrations (P < 0.05) of acetate were observed in BC 0.9% at 12 h; however, BC 0.5% and BC 0.3% obtained higher value (P < 0.05) after 24 h (Table 2). Propionate concentration was significantly higher (P < 0.05) in BC 0.3% and BC 0.9% at 6 h. Subsequently, distinct effects of BC 0.3% , BC 0.5% , and BC 0.9% were observed at 24 h which had significantly higher (P < 0.05) propionate concentrations than CON. A similar pattern was noticeable with butyrate at 12 h such that BC 0.3% , BC 0.5% , and BC 0.9% obtained the higher concentration (P < 0.05) compared with CON. During this period, a similar effect can be seen between the 3 treatments; however, no significant effect was noticed on the 24 h observation. Total volatile fatty acid concentrations were higher (P < 0.05) in BC 0.3% , BC 0.5% and BC 0.9% at 12 to 24 h incubation periods compared to CON. At this time point, treatments BC 0.3% and BC 0.5% were higher (P < 0.05) compared to BC 0.9% and CON. Furthermore, there were no treatment effects on acetate to propionate ratio after 24 h incubation. Consequently, increasing the concentration of buffering capacity showed linear effects (P < 0.05) on pH, total gas production, NH 3 -N, and at some certain time point of individual VFA. Table 1. Effect of different buffering capacity concentrations on in vitro rumen fermentation parameters at 3, 6, 12 and 24 h. SEM, standard error of the mean. e CON (no buffer added); BC 0.3% (0.3% buffer); BC 0.5% (0.5% buffer); BC 0.9% (0.9% buffer). a-d Means with different superscripts in a row differ significantly (P < 0.05); x,y,z Means within a row indicate linear effect among CON, BC 0.3% , and BC 0.5% (P < 0.05). Bacterial diversity of the rumen contents of Hanwoo steers. The boxplot representation of alpha diversity indices is shown in Fig. 1. Alpha diversity indices are composite indices that reflect abundance and consistency. Chao1 which reflect the OTU abundance in the samples showed that BC 0.9% was the highest among treatments followed by BC 0.5% and the rest of the treatments (Fig. 1a). Shannon index which reflects the diversity of the OTU in samples presented BC 0.9% as the most diverse among treatments and BC 0.3% being the least (Fig. 1b). Moreover, Fig. 1c shows the boxplot of OTUs of observed species from the samples. The number of OTUs in BC 0.9% was higher followed by BC 0.5% and the rest of the treatments. Our results showed that the rumen bacterial composition of BC 0.5% and BC 0.9% had overall higher alpha diversity than other treatment groups, although no significant difference was observed after statistical analysis.

Effect of treatments on bacterial community composition of Hanwoo steers rumen contents.
Bacterial taxonomic compositions at the phylum, genera, and species level are shown in Fig. 2

. Results
at the phylum level revealed that 15 bacterial phyla were identified in the rumen digesta samples of Hanwoo steers (Fig. 2a) Fig. 2b. Prevotella was found to be predominant among the genera. The relative abundance of Ruminococcus and Succiniclasticum were higher in BC0.3% and BC0.5%. Supplementation with BC0.9% increased the abundance of Vibrio, Barnesiella, and Treponema, whereas it decreased Paludibacter and Bacteroides. Meanwhile, a noticeable increase in abundance was also observed in Barnesiella and Bacteroides with the supplementation of BC0.3%. At the species level, Prevotella ruminicola was the most abundant among the identified species for all treatments (Fig. 2c). The supplementation of BC 0.3% increased its relative abundance. The comparison of single species  24 and computed using QIIME 25 . CON (no buffer added); BC 0.3% (0.3% buffer); BC 0.5% (0.5% buffer); BC 0.9% (0.9% buffer). www.nature.com/scientificreports/ analyzed through statistical analysis showed a significant effect of the treatments only in the case of Prevotella brevis. This species was more abundant in the CON than those supplemented with BC 0.9% , BC 0.5% and BC 0.3% .
Owing to the BC 0.5% supplemented in the diet, a decreasing abundance of Paludibacter propionicigenes was observed; however, an increase in abundance was notable under CON. Incorporation of BC 0.5% increased the microbial population of Ruminococcus bromii and Succiniclasticum ruminis. Moreover, regardless of the low concentration of BC 0.3% , supplementation of this treatment increased the abundance of Bacteroides massiliensis. Supplementing buffering agents of different buffering capacity concentration may alter the rumen microbiota as what have presented in the result of the present study. The core, shared and unique bacterial community of observed species of the rumen microbiome after treatment of buffering agents with varying level of buffering capacity is presented in Fig. 3a Table 2.
Relative abundance was computed using QIIME 25 . CON (no buffer added); BC 0.3% (0.3% buffer); BC 0.5% (0.5% buffer); BC 0.9% (0.9% buffer); asterisk (*): represents significant differences (P < 0.05).  Fig. 3b. The PCoA plots showed close similarity within and between treatments and control, whereas those under BC 0.9% and BC 0.3% groups showed distinct and spatial separation of bacterial communities. The normalized data presented in Fig. 4 shows the clustering based on the similarity of relative abundance between representative families of OTUs (row), and treatments (column). The analysis divided the representative families into two major clusters distinguishing families which represents low relative abundance on all treatments (upper cluster in red), and families that have varying relative abundance between treatments (lower cluster, colored from peach to blue). On the cluster presenting varying abundance between treatments, two subclusters were also distinguishable; (1) families which represent variation from very low (red) to average (peach) abundance, and (2) cluster representing families which have average to high (blue) abundance. The cluster in the bottom of the heatmap (labelled) contains the families that represent the above average relative abundance. On this cluster, family Prevotellaceae had branched out because it presents the highest abundance with very small variations between treatments (p = 0.092). Family Ruminicoccaceae were also found in all treatments, but varying relative abundance was observed, with BC 0.5% presenting the highest. Families Acidaminococcaceae and Lachnospiraceae were significantly higher (P < 0.05) in BC 0.3% and BC 0.5% , respectively. Also, the Unclassified Clostridiales had significantly highest (P < 0.05) relative abundance in BC 0.5% . A certain unclassified family under order Bacteroidales also showed major abundance especially in BC 0.3% , while families Vibrionaceae and Spirochaetaceae were highest in BC 0.9% . Meanwhile, the normalized data of relative abundance of representative families of observed OTUs have grouped the control sample together with BC 0.3% in a single cluster, while BC 0.5% and BC 0.9% are on their own cluster.

Monitoring of acidosis. The changes in the 24 h mean ruminal pH monitored for 30 d is presented in
supplementary Table 3. During this period of the feeding challenges, mean pH values were > 5.8. Minimum pH was lowest in CON, whereas it was highest in BC 0.5% . Additionally, BC 0.9% had a low minimum pH value second to that of CON. It was noticeable that BC 0.3% and BC 0.5% had higher minimum and mean pH values compared to BC 0.9% and CON. Obtained results indicated that the duration of time where pH was < 5.8 and 5.8 < 6.0 was longer in CON followed by BC 0.9% and BC 0.3% . Meanwhile, BC 0.3% also exhibited good results in the duration of time where pH was approximately 6.0 and above; however, BC 0.5% had even better effects in the rumen. Based on the data gathered, BC 0.5% stabilized the pH of rumen preventing it from becoming acidotic. www.nature.com/scientificreports/

Discussion
One of the major health issues in dairy farming is the sudden decline of ruminal pH that causes reduction of feed intake, problems with digestion, and production losses 10 . Its prevalence adversely affects cattle health resulting to high management cost. Methods of decreasing the rate and incidence of acidosis have already been studied, such as addition of buffering agents and supplementation of commercially available treatments in high-grain diets that presumably reduce acidosis by facilitating intake 29 . Various studies have reported that adding buffer solution, such as sodium bicarbonate and/with magnesium oxide increased dry matter intake when corn silage was the sole or major source of forage in the diet 19 . Moreover, several researches have conducted to control rumen pH through supplementing rations with buffers or neutralizing agents such as sodium bicarbonate or magnesium oxide 30 . Sodium bicarbonate is a naturally secreted buffering agent in the saliva of the cow however when too little saliva is produced, such as during feeding of high-grain rations, supplemental sodium bicarbonate can be of benefit. In addition, this buffer agent is widely used in preventing ruminal acidosis since it provides a natural buffer however, its high solubility limits the buffering activity against acidic condition 20 . Despite its buffering ability, it only functions for a short period of time and because of the high solubility, it is rapidly used by the ruminants. Most studies have suggested that magnesium oxide act either as a neutralizer or buffer in rumen or intestine 31 . It also increases starch digestion in the intestine of animals fed with a high-concentrate diet. This may result in an increase of pH in the small intestines allowing starch-digesting enzymes to become more active 22 . Mao et al. 32 reported that supplementation of the bicarbonate group had higher pH, total gas production, and total VFA concentration although ammonia-nitrogen concentrations remained unaltered. Addition of combined buffers in high concentrate rations altered rumen pH, liquid turnover, and patterns of rumen fermentation 33 . Consequently, commercial buffer agent (CBA) is developed as a buffer premix and considered as more powerful alternative to sodium bicarbonate. This premix is a mixture of various raw materials, differing in acid-binding capacity and solubility that contained live yeast, which promoted the conversion of lactate to propionate; thus, improving rumen conditions. Research data have shown its efficiency in maintaining the stability of ruminal pH, thus preventing the stimulation of subacute ruminal acidosis 34 . Meanwhile, the results of the present study are in accordance with their experimental output. The result of the present study showed that BC 0.9% , as well as the BC 0.5% , had similar effects on rumen content. These two treatments had significant effects on pH, buffering capacity, and ammonia-nitrogen concentration which only shows that supplementation of buffer agents can enhance ruminal fermentation parameters. An increase in ruminal pH upon supplementation of sodium bicarbonate is a result of dissociation of sodium (Na + ) and bicarbonate (HCO 3 − ) 17 . Dijkstra et al. 35 stated that the interactions between rumen and host and other complex factors that determine ruminal pH might result in wide range differences among animals in terms of rumen pH response to dietary changes. Meanwhile, the results on gas production were supported by the claims of Rauch et al. 36 and Kang and Wanapat 16 , who stated that supplementation with sodium bicarbonate enhanced gas production. The increase in gas production might be caused by the dissociation of sodium bicarbonate resulting to increase gas volume because of CO 2 liberation 36 . Also, it might be due to the conversion of some bicarbonate to carbonic acid which soon released as carbon dioxide 17 . NH 3 -N is the vital source of nitrogen for microbial protein synthesis in the rumen 67   . Heatmap presentation of relative abundance of representative families of observed OTUs. Treatments (columns) and families (rows) are clustered using Bray-Curtis dissimilarity test and Ward linkage. Normalized relative abundance are plotted from low (red), mid (peach), and high (blue). Heatmap clustering was generated in MetaCOMET 24 utilizing the InCHlib application 28 . CON (no buffer added); BC 0.3% (0.3% buffer); BC 0.5% (0.5% buffer); BC 0.9% (0.9% buffer); asterisk (*): represents significant differences (P < 0.05). www.nature.com/scientificreports/ ion concentration. Through this bicarbonate dependent mechanism, the rumen epithelium supplies adequate amount of bicarbonate to the rumen in comparison to the bicarbonate produced during salivation 35 . Thus, this mechanism influences rumen buffering capacity 35,37 . Shaver et al. 22 also stated that magnesium oxide and sodium bicarbonate were the best rumen buffers, which increased the acetate: propionate ratio and prevented declines in pH. The effect of buffers on VFA in this study was the same as the data obtained by Kang and Wanapat 16 wherein supplementation with buffering agents increased the total VFA. High ruminal VFA concentration is caused by increased carbohydrate fermentation in the rumen 38 . Although the present study did not show a significant effect on molar concentration of VFA, the noticeable increasing numerical values were observed in buffer-supplemented treatments. Subsequently, the 16S rRNA gene-based metataxonomic survey of bacterial community composition was identified in the rumen digesta samples of Hanwoo steers. Obtained results at the phylum level were in accordance with the data gathered by Nagata et al. 39 wherein the relative abundance of Bacteroidetes was higher during the high-concentrate period of the experimental animals. Additionally, Zhao et al. 40 stated that the microbial community of beef cattle was dominated by Bacteroidetes and Firmicutes at the phylum level regardless of group. An increase in the phylum Bacteroidetes resulted in increased Prevotella and repressed Firmicutes, which was attributed to decreasing Ruminococcaceae. Dodd et al. 41 and Naas et al. 42 indicated that the Bacteroidetes in the rumen represented another numerically dominating phylum that was not associated with cellulose degradation, rather its saccharolytic status is based on limited case studies of noncellulolytic Prevotella rumen isolates. Because of the ability of Prevotella to use a variety of substrates, it tends to dominate in the rumen under a range of diets 43 . In the present study, Prevotella ruminicola appeared to be the predominant species among all treatments. This species constitutes one of the most numerous groups recovered from the rumen and plays important roles in the utilization of polysaccharides of plant origin [44][45][46] and the metabolism of peptides and proteins [47][48][49][50][51] . Moreover, the low-relative abundance of Ruminococcus (8.93%) in this study was in contrast with the findings obtained by Klieve et al. 11 , who also used a high-grain diet (75% barley) for the animals, although, this genus was identified and largely comprised the cellulolytic bacteria. Meanwhile, Klieve et al. 11 and Kim et al. 12 have found that several species of Ruminococcus such as R. bromii are capable of fermenting starch. This might also explain the abundance of R. bromii in the study of Ramos et al. 10 who also used high-concentrate diet for the animals. High propionate concentration of BC 0.5% might be caused by the high relative abundance of Succiniclasticum ruminis. This result is in accordance with the previous studies which stated that this species specializes in fermenting and converting succinate to propionate 10,52-54 , which is an important precursor of glucose in ruminants 52 . Ueki et al. 55 described Bacteroides massiliensis as a producer of acetate, propionate, and succinate. We observed a higher abundance of this bacterium among treated groups in a modest level, which could be one of the major contributors on the increase in molar concentrations of VFA's on in vivo study. The abundance of Paludibacter propionicigenes might be due to its description as a sugars utilizer and a producer of acetate and propionate, an end product of fermentation 56 .
Acidosis was defined as impaired ruminal health accompanied by a reversible ruminal pH depression 11,57-60 . Ruminal microbes convert carbohydrates to short-chain fatty acids at a rate that exceeds the rumen's absorptive, buffering, and outflow capacity causing a rapid decrease in ruminal pH 61 . Meanwhile, Zamarreño et al. 23 stated that the use of sodium bicarbonate and magnesium oxide or even mixed antacids were recommended for satisfactory results. They concluded that the increase in buffering capacity and increase in acid consuming capacity contributed to the correction of animal acidosis.
In conclusion, supplementation of BC 0.5% and BC 0.9% buffer agents increased the buffering capacity, enhanced the rumen fermentation, and alters rumen microbiome, which is an important factor contributing positively to the correction of animal acidosis during a high-concentrate diet. Furthermore, this study also considers other mechanisms such as salivation of the animals and VFA absorption through epithelium, which might play an equal relevant roles in altering rumen microbiota.

Animals, rumen fluid collection and in vitro rumen fermentation. Three ruminally cannulated
Hanwoo steers (500 ± 47 kg body weight; 20 mos. of age) were used to provide ruminal fluid for in vitro rumen fermentation. The animals were fed twice daily with concentrate feed and kleingrass (Panicum coloratum L.). Ruminal contents were collected before morning feeding. Samples were squeezed and strained through four layers of surgical gauze and pooled in an amber bottle with an oxygen-free headspace, which was subsequently capped after collection. Collected samples were immediately transported to the laboratory while being maintaining at a temperature of 39 °C 62 .
Seventy milliliters of rumen fluid were dispensed into serum bottles containing each treatment and 2.5 g dry matter of ground corn grain served as substrate, mixed, and flushed with CO 2 63 . Samples were in triplicate and incubated at 39 °C for 3, 6, 12, and 24 h while shaking horizontally at 100 rpm, as described by Hattori and Matsui 64 . The buffering agents used in treatments are composed of calcium carbonate, magnesium oxide, sodium carbonate, and calcified seaweed (Rupromin Balance™, Rotterdam, Netherlands). Treatments consisted of CON (negative control, no buffer added), BC 0.3% (low buffering capacity, 0.3% buffering agent), BC 0.5% (medium buffering capacity, 0.5% buffering agent), and BC 0.9% (high buffering capacity, 0.9% buffering agent www.nature.com/scientificreports/ agents and the concentrate given to experimental animals were supplied by Purina® Cargill, Korea. The ingredients and chemical composition of the experimental concentrate offered are presented in supplementary Table 4. Treatments were initially tested for determining their neutralizing (NC) and buffering capacity (BC) through titration using 2 N acetic acid from its initial pH to 6.50, and 5.50, respectively (Supplementary Table 5). The buffering agents used in every treatment are in powdered form.

Analyses of in vitro rumen fermentation parameters and buffering capacity.
Ruminal fermentation parameters were monitored at the end of each incubation time period. Total gas production was measured from each serum bottle after the incubation time using a pressure meter (Laurel Electronics, Inc., Costa Mesa, Calif., USA). Consequently, a needle channel connected to the machine was extended into the sealed fermentation bottle for measuring positive pressure created by the gas build up inside the bottle. A gas flow regulator was then opened to allow gas flow inside a syringe barrel and the plunger was subsequently pulled gradually until the pressure reading on the machine display was zero. The volume of gas trapped inside the barrel was recorded as the total gas produced 62,65 .
The pH value was determined using a pH meter (Metler Toledo, Germany) after uncapping each serum bottle. Samples of fermenta were also collected into two 1.5 ml microcentrifuge tubes and stored at − 80 °C prior to ammonia-nitrogen and VFA analyses. Frozen samples were thawed at room temperature; after which, they were centrifuged for 10 min at 13,000 rpm at 4 °C using a Micro 17TR centrifuge (Hanil Science Industrial, Korea). The resulting supernatant was used for ammonia-nitrogen and VFA concentration analyses. Ammonia-nitrogen concentration was measured according to the colorimetric method developed by Chaney and Marbach 66 using a Libra S22 spectrophotometer (Biochrom Ltd., CB40FJ, England) at an absorbance of 630 nm. Analysis of volatile fatty acid concentration was done using high-performance liquid chromatography (Agilent Technologies 1200 series, Tokyo, Japan) with a UV detector set at 210 nm and 220 nm. Samples were isocratically eluted with 0.0085N H 2 SO 4 at a flow rate of 0.6 ml/min and a column temperature of 35 °C.
Ruminal fluid pH was recorded following 1 min of equilibration. Buffering capacity, defined as the resistance to change in pH from pH 7 to 5, was determined by titrating a 30 ml aliquot of ruminal fluid with continuous stirring from its initial pH to pH 5 with 1 N HCl and titrating an additional 30 ml aliquot from its initial pH to a pH of 7 with 1 N NaOH. If the initial pH was higher than 7, only the volume of acid required to reduce the pH from 7 to 5 was recorded. Buffering capacity was converted to milliequivalents per liter as follows: BC = [(milliliters of 1 N HCl) + (milliliters of 1 N NaOH)] × 10 3 /30 21 .

Analysis of rumen fermentation characteristics in Hanwoo steers.
In vivo experiment was conducted using four non-cannulated Hanwoo steers (765 ± 60 kg body weight; 24 mos. of age) in a 4 × 4 Latin square design to assess the effects of treatments on rumen fermentation characteristics and ruminal bacterial composition and diversity of the experimental animals for four months. The feeding trial was conducted with 4 treatments comprised of CON which served as the non-buffer supplemented group, BC 0.3% , BC 0.5% , and BC 0.9% . Each experimental period lasted for a month with 10 days washing (animals were fed with control), 20 days adaptation period (animals were fed with different treatments or control), and rumen sampling collection at the end of the period.
Experimental design and animal management. The Hanwoo steers were fed daily of 2:8 forage (2 kg) and concentrate (2 kg) ratio in 2 equal portions at 9:00 AM and 4:00 PM. Animals in all treatments received the same vaccinations, medications, and were under the same management programs unless otherwise stated. Steers were confined in free-stall barns and had free access to water and exercise lots.
Rumen fluid samples were collected before morning feeding using an oral stomach tube on the 30th day right before transitioning to the next feeding trial for the analysis of ruminal fermentation parameters. These parameters which include rumen pH, total gas, NH 3 -N, and VFAs were all evaluated using the same protocol as used in the in vitro experiment. However, rumen pH change in every experimental period of about 30 days was monitored using eCow (hathor.ecow.co.uk). The eCow system monitors the rumen pH every minute and averaged every 15 min. It was done basically to monitor the occurrence of acidosis through a pH value of < 5.8. This threshold value was used because it is harmful to ruminal cellulolytic bacteria 75 . 16S ribosomal RNA gene amplicon sequencing and metataxonomic analyses. Samples obtained from each treatment were sent to Macrogen, Korea for DNA extraction, 16S rRNA gene sequencing and microbiome analysis. Ruminal fluid samples were transported with dry ice to protect the samples from degradation due to extended shipping time or elevated temperatures. In brief, DNA was extracted using DNeasy Power Soil Kit (Qiagen, Hilden, Germany) according to the manufacturer's instructions. The extracted DNA was quantified using Quant-IT PicoGreen (Invitrogen). The sequencing libraries were prepared according to the Illumina 16S Metagenomic Sequencing Library protocols to amplify the V3 and V4 region. The input gDNA was PCR amplified with 1 × reaction buffer, 1 nM of dNTP mix, 500 nM each of the universal F/R PCR primer, and 2.5 U of Herculase II fusion DNA polymerase (Agilent Technologies, Santa Clara, CA). The cycle condition for 1st PCR was 3 min at 95 °C for heat activation, and 25 cycles of 30 s at 95 °C, 30 s at 55 °C and 30 s at 72 °C, followed by a 5-min final extension at 72 °C. The universal primer pair with Illumina adapter overhang sequences used for the first amplification was V3-F (5′-TCG TCG GCA GCG TCA GAT GTG TAT AAG AGA CAG CCT ACG GGN GGC WGC AG-3′) and V4-R (5′-GTC TCG TGG GCT CGG AGA TGT GTA TAA GAG ACA GGA CTA CHV GGG TAT CTA ATC C-3′). The 1st PCR product was purified with AMPure beads (Agencourt Bioscience, Beverly, MA). Following purification, the 2 μL of 1st PCR product was PCR amplified for final library construction containing the index using NexteraXT Indexed Primer. The cycle condition for the 2nd PCR was www.nature.com/scientificreports/ the same as the 1st PCR condition except for 10 additional cycles. The PCR product was purified with AMPure beads. The final purified product is then quantified using qPCR according to the qPCR Quantification Protocol Guide (KAPA Library Quantification kits for Illumina Sequencing platforms) and qualified using the TapeStation D1000 ScreenTape (Agilent Technologies, Waldbronn, Germany). Sequencing was done using the Illumina Miseq (Illumina Inc., San Diego, CA, USA) platform. The raw data files (fastq) containing the sequenced paired-end (PE) reads were obtained using the bcls2fastq package (Illumina Inc., San Diego, CA, USA) from the base call binary data produced by real-time analysis. The PE raw reads were filtered from adapter sequences using Scythe (v0.994) 68 and Sickle 69 programs then assembled using Fast Length Adjustment of Short Reads (FLASH 1.2.11) 70 . Assembled reads were quality filtered and trimmed for short and extra-long reads, and duplicate reads were removed, then clustered at 100% identity using CD-HIT-OTU 71 . Chimeric reads were identified and the initial clusters were recruited to primary clusters. Then, noise filtering was done and the remaining non-chimeric clusters were binned to operational taxonomic units (OTU) following a greedy algorithm with a cut-off value of 97% species level identity using CD-HIT-OTU 71 . Representative sequences from the clustered OTU were taxonomically assigned using Quantitative Insights Into Microbial Ecology (QIIME Version 1) 25 from the NCBI 16S rRNA gene database, and the taxonomy composition was generated using QIIME-UCLUST 72 . The produced bacterial taxonomy and composition data were used to generate a biological information matrix (BIOM) 73 in Mothur 74 . The generated BIOM file were used to visualize the alpha and beta diversity indices, and the bacterial composition using programs utilized by Metagenomics Core Microbiome Exploration Tool (MetaCOMET) 24 .
Statistical analysis. Data analysis was performed using Statistical Analysis Systems (SAS) version 9.1 (SAS Institute Inc., Cary, NC). The data of rumen fermentation, alpha diversity indices and relative abundance of individual taxa of rumen microbiota were statistically evaluated using Proc general linear model (GLM) for a completely randomized design. All treatments in the in vitro experiment were conducted in triplicate and Duncan's Multiple Range Test (DMRT) was used to identify differences between specific treatments. The linear effects of different buffering capacity concentrations were analyzed using orthogonal polynomial coefficients to describe the functional relationships among the control and treatment groups. A P < 0.05 was considered indicative of significant differences.