Increasing Buffering Capacity Alters Rumen Microbiota Composition and Enhances Rumen Fermentation Characteristics of High-Concentrate Fed Hanwoo Steers

Background: Rumen bacterial community is mainly affected by the type of diet consumed by the host animals. High concentrate diet increases the abundance of lactic acid producers and utilizers due to high level of non-structural carbohydrates thus reducing the number of ber-degrading bacteria because of drastic decrease in pH. Dietary buffers are essential in regulating rumen pH through the compounds responsible in resisting drastic decrease in pH once cattle were fed with high-concentrate diet. However, no study has evaluated the effects of buffering capacity and eciency in alleviating chronic acidosis in rumen. Ruminal metataxonomic and fermentation characteristics analyses were conducted to evaluate the effect of different buffering capacities on in vitro and in vivo experiments in high-concentrate fed Hanwoo steers. Results: Results revealed that BC 0.9% and BC 0.5% had similar and signicant effect (P < 0.05) on in vitro ruminal fermentation at 3 to 24 h incubation. Both BC 0.9% and BC 0.5% had signicantly highest (P < 0.05) buffering capacity, pH, and ammonia-nitrogen (NH 3 -N) than BC 0.3% and CON at 24 h of incubation.

Results: Results revealed that BC 0.9% and BC 0.5% had similar and signi cant effect (P < 0.05) on in vitro ruminal fermentation at 3 to 24 h incubation. Both BC 0.9% and BC 0.5% had signi cantly highest (P < 0.05) buffering capacity, pH, and ammonia-nitrogen (NH 3 -N) than BC 0.3% and CON at 24 h of incubation.
Individual and total volatile fatty acids (VFA) were signi cantly lowest in CON. Increasing buffering capacity concentration showed linear effect on pH at 6 to 24 h while total gas and NH 3 -N at 3 and 12 h.
Phylum Bacteroidetes dominated all treatments but a higher abundance of Firmicutes in BC 0.5% than others. Ruminoccocus bromii and Succiniclasticum ruminis were dominant in BC 0.5% and Bacteroides massiliensis in BC 0.3% . The normalized data of relative abundance of observed OTUs' representative families have grouped the CON with BC 0.3% in the same cluster, whereas BC 0.5% and BC 0.9% 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 BC 0.9% and BC 0.3% groups showed dispersed community distribution.
Conclusion: Our ndings showed that increasing buffering capacity enhances rumen fermentation parameters and affects rumen microbiome by altering bacterial community through distinct structure between high and low buffering capacity, thus an important factor contributed to the prevention of ruminal acidosis during a high-concentrate diet.

Background
Energy and essential nutrients are obtained by ruminants through a complicated symbiotic relationship with the rumen microbiome [1] and bacterial community alterations can affect the productivity and health of the host animal [2]. A high forage diet is usually switched to a high concentrate diet to improve the productivity of the ruminants; however, it alters the rumen ecosystem due to high non-structural carbohydrates level [1]. Meanwhile, the core ruminal bacterial community is dominated by the phyla Bacteroidetes, Firmicutes, and Proteobacteria regardless of diet composition [3]. However, a highconcentrate diet induces death and cell lysis [4], thus decrease in abundance of Firmicutes in the rumen [3]. It also enhances the growth of lactic acid utilizers like Megasphaera elsdenii, Selenomonas ruminantium, and Veillonella parvula resulting to a drastic reduction of ber-degrading bacteria such as Fibrobacter succinogenes and Ruminococcus spp [1]. Feeding of highly fermentable diets is the current practices in high producing beef to increase growth rates, but it causes microbial disturbances resulting to digestive disorders such as ruminal acidosis [5]. The rapid fermentation of non-structural carbohydrates resulted in the accumulation of volatile fatty acid and lactic acid in the rumen causing a drastic decrease in pH [6]. Hence, the use of buffer could be useful to resist changes in rumen pH whenever cattle are being fed with high concentrate, low forage, fermented and ne-chopped forage [7].
Compounds that increase the buffering capacity of ruminal uid help maintain a more stable ruminal pH and direct neutralization of VFA especially during a diet or fermentation-related acid challenge [5,6]. Rumen buffering could avert the sudden decrease in pH, thus could enhance rumen microbial growth, activity and diversity, microbial protein synthesis, and fermentation end product [10]. Buffering capacity (BC) is then referred to as the number of moles of H + that should be added to a 1L solution to decrease pH by 1 unit [11]. Weak acids and bases are known to provide better buffering in comparison to strong acids and bases because of the equilibrium establishment between the acid and the conjugate base [12].
Various studies have reported that adding a buffer solution, such as sodium bicarbonate (NaHCO 3 ) with magnesium oxide (MgO) increased dry matter intake when corn silage was the sole or major source of forage in the diet [13]. NaHCO 3 is commonly used in preventing ruminal acidosis because it provides a natural buffer; however, its high solubility limits the buffering activity against acidic conditions [14]. Le Ruyet and Tucker [15] proved that NaHCO 3 had high BC in an in vitro study. It contained 26% more actively buffering the CO 3 portion of the molecule that is important to neutralize the acid. MgO, on the other hand, appears to work e ciently in combination with NaHCO 3 [14]. Shaver et al. [16] stated that supplementing NaHCO 3 and MgO in a 3:1 ratio is the recommended level of dietary buffer for the best response. The e ciency and mechanisms of buffer responsible for alleviating chronic acidosis are variable and often inconsistent [9]. Research on different level of buffering capacity in enhancing rumen fermentation parameters and microbiome during high concentrate diet has not yet been investigated. In the present work, ruminal metataxonomic and fermentation characteristics analyses were conducted using rumen uid samples to evaluate the effect of different buffering capacities on in vitro and in vivo trials in high-concentrate fed Hanwoo steers.

Effect of different buffering capacities on in vitro rumen fermentation parameters
The buffering capacity of BC 0.9% and BC 0.5% were signi cantly greatest (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 signi cantly highest (P < 0.05) buffering capacity value of 106.00 meq/L, hence had a similar effect on in vitro after 24 h. The ruminal pH obtained from BC 0.9% and BC 0.5% showed similar effects and were consistently higher (P < 0.05) than the other treatments throughout the incubation period. In gas production, BC 0.9% , BC 0.5% , and BC 0.3% had signi cantly higher (P < 0.05) gas produced than CON and showed similar effects at 3 and 12 h incubation. Ruminal NH 3 -N concentration was signi cantly higher (P < 0.05) in BC 0.9% and BC 0.5% , thus, it also had a similar effect on this parameter. However, at 6 and 12 h, no effect observed on treatments except that BC 0.5% tended to increase (P = 0.073) NH 3 -N concentration followed by BC 0.9% and the rest treatments.
Signi cantly 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 the highest value (P < 0.05) after 24 h (Table 2). Propionate and butyrate concentrations were both highest (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 signi cantly 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 highest concentration (P < 0.05) compared with CON. During this period, a similar effect can be seen between the 3 treatments; however, no signi cant effect was observed after 24 h. Total volatile fatty acid contents were greater (P < 0.05) in BC 0.3% , BC 0.5% and BC 0.9% at 12 h but had a slight change after 24 h. At this time point, treatments BC 0.3% and BC 0.5% were highest (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. x,y Means within a row indicate linear effect among CON, BC 0.3% , and BC 0.5% (P < 0.05)

Effect of different buffering capacities on rumen fermentation characteristics in Hanwoo steers
The effect of different buffering capacity concentrations on rumen fermentation characteristics of Hanwoo steers in four treatments are presented in Table 3. Average pH had no signi cant effects among CON and treatments. However, BC 0.3% , BC 0.5% , and BC 0.9% had signi cantly higher (P < 0.05) buffering capacity value than CON, and showed linearly signi cant effect (P < 0.05). Ammonia-nitrogen, acetate to propionate ratio, individual and total VFA concentrations of rumen uid from steers under all treatments were not signi cant and showed similar effects after the in vivo experiment. x,y 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 Figure 1. Alpha diversity indices are composite indices that re ect abundance and consistency. Chao1 which re ect 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 ( Figure 1a). Shannon index which re ects the diversity of the OTU in samples presented BC 0.9% as the most diverse among treatments and BC 0.3% being the least ( Figure 1b). Moreover, Figure 1c showed 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. The diversity index is used to analyze the temporal and spatial changes in species composition which re ects whether bacterial communities between groups have differences. 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 signi cant 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 Figure 2. Results at the phylum level revealed that 15 bacterial phyla were identi ed in the rumen digesta samples of Hanwoo steers (Figure 2a). The majority of the sequences obtained from all treatments belonged to is incorporated into the diet, which had a reverse effect as did CON. Species-level analyses revealed that Prevotella ruminicola predominated the treatments CON, BC 0.3% , BC 0.5% and BC 0.9% with the relative abundance of 24.85%, 32.16%, 26.73%, and 23.17%, respectively ( Figure 2c). The comparison of single species analyzed through statistical analysis showed a signi cant effect of the treatments only in the case of Prevotella brevis. This species was more abundant (P = 0.015) in the CON and as steers received a diet supplemented with BC 0.9% , BC 0.5% and BC 0.3% its abundance decreased. Owing to the BC 0.5% supplemented in the diet, a decreasing abundance of Paludibacter propionicigenes was observed; however, it increased in CON. Incorporation of BC 0.5% increased the microbial population of Ruminococcus bromii and Succiniclasticum ruminis. Moreover, the smaller percentage of BC 0.3% resulted in a higher abundance of Bacteroides massiliensis which led to a sudden decrease in its population as the concentration of treatments increases. Supplementing buffers of different buffering capacity concentration may affect the rumen microbiota through the relative abundance of bacterial species.
The core, shared and unique bacterial community of observed species of the rumen microbiome after treatment of buffer agents with varying level of buffering capacity is presented in Figure 3 as Venn diagram. A total of 211 (59.6%) observed species can be found across all the samples (core), 79 (22.32%) for shared by 2 or 3 samples, and 64 (18.08%) are speci c and are distributed to the four samples.
The normalized data presented in Figure 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 sub-clusters were also distinguishable; (1) families which represent variation from very low (red) to average (peach) abundance, and (2)  The comparison of the bacterial communities by principal coordinate analysis (PCoA) is presented in Figure 5. The PCoA plots showed close similarity within and between treatments and control, whereas those under BC 0.9% and BC 0.3% groups showed dispersed distribution of bacterial communities. The PCoA plot showed dissimilarity of bacterial community and revealed a distinct structure between high buffering capacity and low buffering capacity.

Monitoring of acidosis
The changes in the 24 h mean ruminal pH monitored for 30 d is presented in Table 4. 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 and did not show any signs of acidosis in the rumen. Based on the data gathered, BC 0.5% stabilized the pH of rumen preventing it from becoming acidotic.

Discussion
Currently, one of the major health issues in dairy farming is the sudden decline of ruminal pH which causes a reduction of feed intake, problems with digestion, and production losses. Cattle health mainly suffers and additional costs in management increase due to its prevalence. Sodium bicarbonate is widely used for the prevention of rumen acidosis because it serves as a natural buffer in the rumen. 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 [22]. 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 starchdigesting enzymes to become more active [16]. Mao et al. [23] reported that supplementation of the bicarbonate group had higher pH, total gas production, and total VFA concentration although ammonianitrogen concentrations remained unaltered. Addition of combined buffers in high concentrate rations altered rumen pH, liquid turnover, and patterns of rumen fermentation [24]. 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 e ciency in maintaining the stability of ruminal pH, thus preventing the stimulation of subacute ruminal acidosis (Provimi™, Rotterdam, Netherlands). 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. Both treatments had signi cant effects on pH, buffering capacity, and ammonia-nitrogen concentration relative to that of the negative control. An increase in ruminal pH upon supplementation of sodium bicarbonate is a result of dissociation of sodium (Na + ) and bicarbonate (HCO 3 − ) [11]. Meanwhile, the results on gas production were supported by the claims of Rauch et al. [25] and Kang and Wanapat [10], 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 [25]. Also, it might be due to the conversion of some bicarbonate to carbonic acid which soon released as carbon dioxide [11]. Moreover, obtained data from the present study is in accordance with the results of Le Ruyet and Tucker [15] on the temporal effects of ruminal buffers in terms of buffering capacity and pH of ruminal uid from cows fed a high concentration diet.
Buffering compounds increased the ruminal uid buffering value index and were bene cial in preventing postprandial increases in ruminal uid hydrogen ion concentration. Shaver et al. [16] 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 [10] wherein supplementation with buffering agents increased the total VFA. High ruminal VFA concentration is caused by increased carbohydrate fermentation in the rumen [26]. Although the present study did not show a signi cant effect on molar concentration of VFA, the noticeable increasing numerical values were observed in buffer-supplemented treatments.
Subsequently, the metagenomic survey of bacterial community composition was identi ed 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. [27] wherein the relative abundance of Bacteroidetes was higher during the high-concentrate period of the experimental animals. Additionally, Zhao et al. [28] 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. [29] and Naas et al. [30] 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 [31]. 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 [32][33][34] and the metabolism of peptides and proteins [35][36][37][38][39]. Moreover, the low-relative abundance of Ruminococcus (8.93%) in this study was in contrast with the ndings obtained by Klieve et al. [40], who used a high-grain diet (75% barley) for the animals, although this genus was identi ed and largely comprised the cellulolytic bacteria. 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 study of Van Gylswyk [41], who stated that this species specializes in fermenting and converting succinate to propionate, which is an important precursor of glucose in ruminants. Ueki et al. [42] described Bacteroides massiliensis as a producer of acetate, propionate, and succinate which can explain the increase in molar concentrations of VFA 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 [43].
Acidosis was de ned as impaired ruminal health accompanied by a reversible ruminal pH depression [40,[44][45][46][47]. Ruminal microbes convert carbohydrates to short-chain fatty acids at a rate that exceeds the rumen's absorptive, buffering, and out ow capacity causing a rapid decrease in ruminal pH [48]. Data gathered in this experiment agreed with the results obtained by Tucker et al. [8] that the addition of a buffer, especially sodium bicarbonate, was effective in reducing ruminal uid acidity and retards the drop in pH that normally occurs from 6 to 12 h post-feeding. Also, Zamarreño et al. [9] 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.

Animals, rumen uid collection and in vitro rumen fermentation
Three ruminally cannulated Hanwoo steers (500 ± 47 kg body weight; 20 mos. of age) were used to provide rumina uid for in vitro rumen fermentation. The animals were fed twice daily with concentrate feed and kleingrass. 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 [49].
Seventy milliliters of rumen uid were dispensed into serum bottles containing each treatment and 2.5 g dry matter of ground corn grain served as substrate, mixed, and ushed with CO 2 [50]. 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 [51]. The buffer used in treatments is composed of calcium carbonate, magnesium oxide, sodium carbonate, and calci ed seaweed (Rupromin Balance™, Rotterdam, Netherlands).
Treatments consisted of CON (negative control, no buffer added), BC 0.3% (low buffering capacity, 0.3% buffer), BC 0.5% (medium buffering capacity, 0.5% buffer), and BC 0.9% (high buffering capacity, 0.9% buffer). The buffer 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 Table 5. Treatments were initially tested for determining their neutralizing (NC) and buffering capacity (BC) through titration using 2N acetic acid from its initial pH to 6.50, and 5.50, respectively ( Table 6). 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 ow regulator was then opened to allow gas ow 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 [49,52].
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 [53] using a Libra S22 spectrophotometer (Biochrom Ltd., CB40FJ, England) at an absorbance of 630 nm. NH 3 -N is the vital source of nitrogen for microbial protein synthesis in the rumen [54]. Analysis of volatile fatty acid concentration was done using highperformance 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 ow rate of 0.6 mL/min and a column temperature of 35 °C.
Ruminal uid pH was recorded following 1 min of equilibration. Buffering capacity, de ned as the resistance to change in pH from pH 7 to 5, was determined by titrating a 30 ml aliquot of ruminal uid with continuous stirring from its initial pH to pH 5 with 1N HCl and titrating an additional 30 ml aliquot from its initial pH to a pH of 7 with 1N 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 1N HCl) + (milliliters of 1N NaOH)] × 10 3 /30 [15].

Analysis of rumen fermentation characteristics in Hanwoo steers
In vivo experiment was conducted using four 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 negative control, BC 0.3% , BC 0.5% , and BC 0.9% .
The Hanwoo steers were fed daily of 2:8 forage and concentrate ratio in 2 equal portions at 0900 and 1600 h. Animals in all treatments received the same vaccinations, medications, and were under the same management programs unless otherwise stated. Steers were con ned in free-stall barns and had free access to water and exercise lots.
Rumen uid samples were collected before morning feeding using an oral stomach tube on the 30 th day right before transitioning to the next feeding trial for the analysis of ruminal fermentation parameters.
These parameters 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). It was done basically to monitor the occurrence of acidosis through a pH value of <5.8 for several hours a day.
16S rRNA amplicon sequencing and metataxonomic analyses Samples obtained from each treatment were sent to Macrogen, Korea for DNA extraction, 16S rRNA sequencing and microbiome analysis. In brief, DNA was extracted using DNeasy Power Soil Kit (Qiagen, Hilden, Germany) according to the manufacturer's instructions. The extracted DNA was quanti ed 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 ampli ed 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 sec at 95 °C, 30 sec at 55 °C and 30 sec at 72 °C, followed by a 5-min nal extension at 72 °C. The universal primer pair with Illumina adapter overhang sequences used for the rst ampli cation 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 puri ed with AMPure beads (Agencourt Bioscience, Beverly, MA). Following puri cation, the 2 uL of 1st PCR product was PCR ampli ed for nal library construction containing the index using NexteraXT Indexed Primer. The cycle condition for 2nd PCR was the same as the 1st PCR condition except for 10 cycles. The PCR product was puri ed with AMPure beads. The nal puri ed product is then quanti ed using qPCR according to the qPCR Quanti cation Protocol Guide (KAPA Library Quanti cation kits for Illumina Sequencing platforms) and quali ed 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 les (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 ltered from adapter sequences using Scythe (v0.994) [55] and Sickle [56] programs then assembled using Fast Length Adjustment of Short Reads (FLASH 1.2.11) [57].
Assembled reads were quality ltered and trimmed for short and extra-long reads, and duplicate reads were removed, then clustered at 100% identity using CD-HIT-OTU [58]. Chimeric reads were identi ed and the initial clusters were recruited to primary clusters. Then, noise ltering 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 [58]. Representative sequences from the clustered OTU were taxonomically assigned using Quantitative Insights Into Microbial Ecology (QIIME Version 1) [18] from the NCBI 16S rRNA database, and the taxonomy composition was generated using QIIME-UCLUST [59]. The produced bacterial taxonomy and composition data were used to generate a biological information matrix (BIOM) [60] in Mothur [61]. The generated BIOM le were used to visualize the alpha and beta diversity indices, and the bacterial composition using programs utilized by Metagenomics Core Microbiome Exploration Tool (MetaCOMET) [17].