The experiments were conducted at the Beef Cattle Research Station of China Agricultural University in Beijing. Animal management and research procedures were approved by the Animal Welfare and Ethical Committee of China Agricultural University (Permit No. DK3178). Experiments were performed as per the Regulations of the Administration of Affairs Concerning Experimental Animals (The State Science and Technology Commission of P. R. China, 1988).
In vitro incubation procedure
Nine, sulfur addition levels, including 0.4% (basal sulfur content of the substrate dry matter), 0.5% (using sodium sulfate as exogenous sulfur source), 0.6%, 0.7%, 0.8%, 0.9%, 1.0%, 1.1% and 1.2% were evaluated in an in vitro study with a complete random design. Fresh rumen fluid was collected from 3 Angus steers (350 kg±43 kg) having permanent rumen fistula, before their morning feed. They were fed a total mixed ration twice a day at 0800 and 1600 h, the fluid was then squeezed and filtered through four layers of cheesecloth into a vacuum bottle. The samples were transported immediately to the laboratory of Beef Cattle Research Center of China Agricultural University. The in vitro incubation was carried out according to the procedures of Menke et al. (1979). The rumen fluid was mixed with artificial saliva in a 1:2 (v/v) proportion, under a continuous flux of CO2 to prepare in vitro fermentation inoculum.
220 mg of feed samples, same as the ration fed to fistula cattle, with the right sulfur levels were transferred to 100 mL glass syringes, pre-incubated at 39 °C, with each treatment containing nine syringes. 30 mL of inoculum was injected into each syringe with a Varispenser (Eppendorf, Germany) and then incubated on an automatic shaker, in a water bath at 39 °C for 72 h. During the incubation, the volume of cumulative gas production (GP) was recorded manually at the time points of 0, 2, 4, 6, 8, 10, 12, 18, 24, 30, 36, 42, 48, 60, and 72 h.
At the end of 24 h incubation, triplicate gas samples were collected from each treatment after terminating the fermentation process in ice water, to determine the CH4 and H2S concentration. CH4 was determined by Gas Chromatography (TP-2060F, Beijing Beifen Tianpu Analytical Instrument Co., Ltd.), and H2S was determined using GASTEC (Japan) fast gas detection tube. The fermentation mixture was sampled at 24 h and 48 h and then centrifuged at 8,000×g for 15 min at 4 °C. Thus obtained supernatant was used to determine the volatile fatty acids (VFA) and ammonia-nitrogen (NH3-N) using methods described by Wu et al. (2015).
Animals, diets, and experimental design
Eight, 24-month-old Angus steers (350 kg±43 kg) having permanent rumen fistula were selected as the experimental animals in a repeated 4×4 Latin square design. Each of the four sulfur addition levels (0.4%, 0.6%, 0.8%, and 1.0%) was fed to two animals in a period of 20 days, where 18 days were meant for adaptation and 2 days for sample collection. The basic diet contained 0.4% sulfur (DM basis) and was used as a control (CON), while sodium sulfate (Na2SO4) was added as an exogenous sulfur source to provide the right amount of additional sulfur, which were set as low additional sulfur (LAS), moderate additional sulfur (MAS), and high additional sulfur (HAS) treatments (Table 1).
Sample collection and measurement
Collection of rumen fluid On the first day of each sampling period, rumen content was collected through fistula before the morning feed at the following four sites: rumen sac, dorsal cecum, abdominal sac, and abdominal cecum. It was then mixed and filtered using four layers of cheesecloth, and transferred to 20 mL centrifuge tubes to further concentrate VFA and measure the NH3-N content. Additionally, 2 mL of rumen contents were cryopreserved in triplicate and immediately stored in liquid nitrogen for further DNA extraction and microbial analysis.
Collection of rumen epithelial tissue samples After the first day of sampling, the experimental animals were allowed to fast, with no solid and liquid intake for the next 24 h, and the rumen epithelial tissue was collected from the abdominal sac portion of the rumen. The sac was pulled toward the fistula by hand after emptying the remaining fistula contents. The chyme residue on the surface of the rumen epithelium was washed with saline, and then the tissue was cut thoroughly with a surgical scissor soaked in DEPC water. The tissue samples were cut into 1×1 cm and 0.5×0.5 cm pieces and placed in 4% polyformaldehyde (Sigma, USA) and 2.5% glutaraldehyde solution for histomorphological analysis.
DNA extraction, high-throughput sequencing, and data processing
DNA was extracted from 0.5 g of each of the homogenized ruminal semi-fluid samples using the repeated bead-beating plus column purification method  and an oscillator (Precellys 24, Bertin Technologies, France). The rotation speed of the oscillator was 5,500 rpm with two circulations (30 s per circulation). The DNA quality was assessed by agarose gel (1%) electrophoresis, and metagenomic DNA concentrations were determined using a NanoDrop 2000 Spectrophotometer (Thermo Scientific, USA). Further, DNA was diluted to a concentration of 1 ng/uL using sterile water.
Sequencing was performed on an Illumina MiSeq PE300 platform. DNA was amplified using the universal eubacterial primer set (338F: 5’-ACTCCTACGGGAGGCAGCAG-3’ and 806R: 5’-GGACTACHVGGGTWTCTAAT-3’), which targets the hypervariable V3-V4 region of the 16S rRNA gene. The reverse primer contained a 6-bp error-correcting barcode, unique to each sample  and the 5`-end of the reverse primer was fused to an Ion A adaptor plus a key sequence and a sample barcode sequence; while the forward primers were fused to a truncated Ion P1 adapter sequence. PCR conditions and the amplification reaction protocol have been previously described by Zhou et al. (2017). Amplicons were examined on a 2% E-Gel Size SelectTM Agarose Gel and purified with Agencourt AMPure XP Reagent. Library size and molar concentration were determined by Agilent 2100 BioanalyzerTM using Agilent High Sensitivity DNA Kit (Agilent Technologies, Inc., Santa Clara, CA). Emulsion PCR was performed using the Ion OneTouchTM 200 Template Kit v2 DL (Life Technologies, Inc.) according to the manufacturer’s instructions. The sequencing of the amplicon libraries was performed on a 318 chip by the Ion Torrent Personal Genome Machine (PGM) system using the Ion PGMTM Sequencing 300 kit (Life Technologies, Inc.).
The Illumina MiSeq sequencing data was analyzed using QIIME software (version 1.7.0) . Filters were applied to the sequences before phylogenetic analysis. Depending on the appropriate fragment size for V3-V4 PCR (150–200 bp), bases after the 200th position were trimmed and the reads shorter than 150 bp were removed. Reads with a quality score of <25 were removed using the NGS QC Toolkit, and only sequences without any ambiguous characters were included in the analysis. FLASH 1.2.7v was used to merge the paired-end reads from raw sequencing data . Chimeric sequences were removed using USEARCH based on the UCHIME algorithm . In order to calculate the downstream diversity (alpha and beta diversity), all the samples were subsampled in equal sizes of 100,000 reads before comparing the bacterial communities. The sequences were clustered into operational taxonomic units (OTUs) at 97% sequence identity level, and the most abundant sequence from each OTU was chosen as a representative. Based on the OTUs, rarefaction curve and alpha diversity indices (i.e., abundance-based coverage estimator [ACE], and the Chao 1, Shannon, and Simpson estimators) were developed. The jackknifed beta diversity was visualized by Principal component analysis (PCA) and performed using the UnscramblerX program (CAMO Software Inc., Woodbridge, NJ) to identify the shifts in the microbial population structure.
Pretreatment of the Rumen Epithelial Tissue Section for Staining and Electron Microscopy
Ten papillae per animal were prepared for light microscopy histomorphometric analysis using methods previously described by Odongo et al. (2006). PFA-fixed, paraffin-embedded papillae were sectioned at 6 µm thickness, stained with hematoxylin and eosin, and mounted for analysis. The microscopist was blinded to the treatment conditions during the histomorphometric analysis. Measurement of each stratum was made using the 40× objective lens, and four images were captured per papillae for a total of 40 replicates per measurement per animal. Image-Pro Plus software (Media Cybernetics, Bethesda, MD, USA) was used to measure the predefined criteria previously described by Steel et al. (2011).
Additional papillae were prepared for electron microscopy using methodology reported by Graham and Simmons (2005). Washed papillae were immediately fixed in 2.5% glutaraldehyde for 24 h, then fixed in 1% osmium for 1 h, and dehydrated in a series of graded ethanol solutions. For scanning electron microscopy (SEM), the papillae were subjected to critical-point drying using liquid CO2 as a medium, then mounted, and coated with gold. The samples were then examined using SEM (Hitachi Model S-3000N, Hitachi Technologies, Tokyo, Japan). For transmission electron microscopy (TEM), dehydrated samples were placed in a mixture of Spurr resin and acetone (1:1) for 30 min, followed by 10 h in 100% resin. These samples were placed in molds and polymerized at 40 °C or 60 °C for 48 h. Semithin (0.25–0.5 µm) sections were cut with glass knives and stained with 1% toluidine blue-O in 1% sodium borate. Ultrathin (70–90 nm) sections were cut using a diamond knife, stained with methanolic uranyl acetate followed by lead citrate, and examined using TEM (Hitachi H-7650, Hitachi Technologies, Tokyo, Japan) .
Measurement of relative expression of RNA
RNA isolation and cDNA synthesis Total RNA was extracted from the papillae samples using Trizol as described by Chomczynski and Sacchi (1987). The RNA concentration was quantified using a Nanodrop spectrophotometer ND-1000UV-Vis (Thermo Fisher Scientific, Madison, Wisconsin, USA). The absorption ratio (260/280 nm) of all the samples was between 1.96 and 2.09, indicating a high RNA purity. Aliquots of RNA samples were subjected to electrophoresis using 1.4% agarose-formaldehyde gel to verify its integrity. The concentration of RNA was adjusted to 1 µg/µL based on the optical density and then stored at –80 °C. Total RNA (1 µg) was reverse-transcribed using a PrimeScript® RT reagent Kit with gDNA Eraser according to the manufacturer’s instructions.
Primer design and qRT-PCR Primer sets were designed to recognize and amplify the conserved nucleotide sequences encoding bovine TJ proteins and cytokines. cDNA sequences were identified using BLAST (Basic Local Alignment Search Tool) (National Center for Biotechnology Information, Bethesda, MD, USA) and primers were designed using the Primer 5.0 (Whitehead Institute, Cambridge, MA, USA). All the primers were synthesized by Genenode Biotechnologies (Beijing, China). Real-time quantitative PCR of target genes and β-actin were performed using the ABI 7500 real-time PCR system (Applied Biosystems, Foster, California, USA) by detecting the fluorescence of SYBR green dye. Amplification conditions were as follows: 94 °C for 2 min, followed by 94 °C for 15 s, 60 °C for 15 s and 72 °C for 30 s for 30 cycles. Each sample contained 2 µL of cDNA in 5 µL 2×SYBRGreen PCR Master Mix with 10µM of each primer in a final volume of 10 µL. All the measurements were performed in triplicate. A reverse-transcription-negative blank for each sample and a no-template blank were used as negative controls. The relative amount of each studied mRNA was normalized to the mRNA level of the housekeeping gene, β-actin, and the data were analyzed according to the 2–△△CT method. The primers and amplicon sizes of all the genes are presented in Table 2.
In order to estimate the kinetic parameters of Cumulative Gas production (GP), all the results of GP were fitted using the NLIN Procedure of the SAS 9.0 statistical software, according to France et al. (2000). The formula used is as follows: a =b× (1−e−ct), where: a = volume (mL) of gas production per 0.2 g DM substrate at time t; b = asymptotic gas production (mL) of 0.2 g DM substrate; c = rate of gas production per hour. The test data were preliminarily sorted by Excel, and the one-way ANOVA was performed by SAS 9.0 statistical software. Duncan’s multiple comparison test was used to calculate the SEM and P values. Additionally, the linear and quadratic curve trend analysis (Orthogonal polynomial contrast) was carried out. P<0.05 (significant) and P<0.01 (extremely significant) were used as the criteria to judge the significance of differences.