Illumina Sequencing and Metabolomics Analysis Reveal Thiamine Modulation of Ruminal Microbiota and Metabolome Characteristics in Goats Fed a High-Concentrate Diet

All animal experiments were conducted in line with the Animal Protection Law based on the Guidelines for the Care and Use of Laboratory Animals approved by the Ethics Committee of Yangzhou University (SXXY 2015-0054).

Eighteen mid-lactating (148 ± 3 DIM) Saanen goats (body weight = 36.5 ± 1.99 kg; body condition score = 2.73 ± 0.16, where 0 = emaciated and 5 = obese; Russel et al., 1969) in parity 1 or 2 were used. All goats were randomly fed one of three diets: (1) control diet (CON; n = 6; concentrate:forage 30:70), (2) high-concentrate diet (HG; n = 6; concentrate:forage 70:30), (3) high-concentrate diet with 200 mg of thiamine/kg of DMI (HGT; n = 6; concentrate:forage 70:30). The supplementation dose of thiamine was based on previous studies on lambs (Neville et al., 2010), dairy goats (Zhang H.W. et al., 2010), and cows (Zhang et al., 2014; Pan et al., 2017b). Goats in each group were fed on the respective diets for 8 weeks. Dietary components of dairy goats were formulated according to NRC (2001). The dietary components are presented in Supplementary Table 1, and the data have been published in the Journal of Dairy Science (Zhang et al., 2019). The goats were fed at 7:00 and 18:00 and with one half of the allowed daily dose at each feeding. Thiamine was mixed with the concentrate to be licked by the goats. During the trial period, goats were allowed free access to freshwater.

On the last day of week 8, the goats were slaughtered 4 h after feeding, and a representative sample of ruminal digesta (at least 200 ml) was collected from the ruminal ventral sac. Then, 200 ml of rumen fluid was collected from each animal and transferred into a separate sterilized container, immediately frozen with liquid nitrogen, and stored at −80°C for further processing. In total, 18 rumen fluid samples were collected from the goats; however, two rumen fluid samples from the HGT group were damaged, and therefore, metabolomic analysis was only performed for the remaining 16 rumen fluid samples. A pH meter (Sartorius, Göttingen, Germany) was used to measure rumen pH using the collected samples. One hundred milliliters of the fluid samples was filtered through four layers of cheesecloth and centrifuged at 10,000 × g for 15 min at 4°C to obtain the supernatant, which was used to determine the concentration of volatile fatty acids (VFAs) and ammonia nitrogen (NH₃-N). The content of thiamine in the rumen fluid was measured by high-performance liquid chromatography (HPLC) according to the Analytical Methods Committee (2000). Another 100 ml of aliquot was stored at −80°C immediately for microbial DNA extraction and metabolomic analysis.

The collected rumen samples were thawed at room temperature, and a 1-ml aliquot was centrifuged at 10,000 × g for 1 min at 4°C, and then the supernatant was discarded. DNA was extracted with a bead-beating method using a mini-bead beater (BioSpec Products, Bartlesville, United States) as per manufacturer’s instructions, followed by phenol–chloroform extraction (Zoetendal et al., 1998). After the solution was precipitated with anhydrous ethanol, the pellets were suspended in the EDTA buffer. The DNA was determined spectrophotometrically using a Nanodrop spectrophotometer (Nyxor Biotech, Paris, France), following staining with QIAamp DNA Stool Mini Kit (Qiagen, Hilden, Germany). DNA was stored at −80°C until further processing.

Next-generation sequencing (NGS) library preparations and Illumina MiSeq sequencing were conducted at Novogene Inc. (Nanjing, China). The V3–V4 regions of the bacterial 16S rRNA gene were amplified by PCR using a forward primer (CCTACGGRRBGCASCAGKVRVGAAT) and reverse primer (GGACTACNVGGGTWTCTAATCC). These primers also contained adapter sequences that allow uniform amplification of highly complex libraries in preparation for downstream NGS on Illumina MiSeq. All PCRs were performed in a 50-μl thermal cycler (Bio-Rad, United States) containing 4 μl of fivefold FastPfu buffer, 10 ng of DNA, 0.8 μM of each primer, 0.5 U of Pfu polymerase, and 2 μl of 2.5 mM dNTPs. PCR conditions utilized for amplification were as follows: first step of 95°C for 3 min, followed by 27 cycles at 95°C for 30 s, 55°C for 30 s, 72°C for 45 s, and a final extension at 72°C for 10 min. The amplicons were purified using the AxyPrep DNA gel Extraction Kit (Axygen Biosciences, Union City, CA, United States). Thereafter, amplicon libraries were generated using a TruSeq^TM DNA Sample Prep Kit (TransGen Biotech, China) according to the manufacturer’s protocol. Paired-end (2 × 250 bp) sequencing was performed with the Illumina MiSeq platform based on the manufacturer’s guidelines (Caporaso et al., 2010).

We used Cutadapt V1.9.1 software (Martin, 2011) to trim the low-quality part of sequence reads, while sample data were extracted from the reads obtained using barcodes. Next, we removed barcode and primer sequences to obtain raw reads. Then read sequences (Rognes et al., 2016) were compared against the species annotation database to detect the chimeric sequences, which were subsequently removed (Haas et al., 2011) to obtain the final clean reads. Using UPARSE V7.0 software, clean reads from all samples were clustered (Edgar, 2013), whereby 97% of identity sequences were clustered into operational taxonomic units (OTUs). Then, sequences with the highest occurrence frequency in OTUs were screened out as the representative sequences of OTUs following the principles of the algorithm. Additionally, community richness and diversity were measured using the Chao1 and Shannon indices, respectively. Species annotation analysis was conducted using the Mothur method and SILVA132 (Wang, 2007) SSU rRNA database (Quast and Pruesse, 2013) (the threshold value was set as 0.8–1) so as to obtain taxonomic information and make statistics of the community composition of each sample at various classification levels, including kingdom, phylum, class, order, family, genus, and species. The data of each sample were homogenized. We employed QIIME 1.9.1 and R 2.15.3 software to execute different analyses between alpha diversity index groups.

A total of 16 rumen samples were analyzed using the LC-MS platform as previously reported elsewhere (Chen et al., 2009; Zhu, 2013). Afterward, samples were resuspended with prechilled methanol and 0.1% formic acid by vortexing, followed by incubation on ice for 5 min and centrifugation at 15,000 rpm at 4°C for 5 min. After centrifugation, supernatants were diluted to a final concentration of 60% methanol using LC-MS-grade water. The samples were subsequently transferred to a fresh Eppendorf tube with a 0.22-μm filter and then centrifuged at 15,000 × g at 4°C for 10 min. Finally, the filtrate was injected into the LC-MS system for analysis. All LC-MS analyses were performed using a Vanquish UHPLC system (Thermo Fisher) coupled with an Orbitrap Q Exactive HF-X mass spectrometer (Thermo Fisher). Briefly, samples were injected onto a Hypersil Gold column (100 × 2.1 mm, 1.9 μm) using a 16-min linear gradient at a flow rate of 0.2 ml/min. The eluents for the positive polarity mode were eluent A (0.1% FA in water) and eluent B (methanol), while those of the negative polarity mode were eluent A (5 mM ammonium acetate, pH 9.0) and eluent B (methanol). The solvent gradient were set as follows: 2% B, 1.5 min; 2–100% B, 12.0 min; 100% B, 14.0 min; 100–2% B, 14.1 min; and 2% B, 16 min. Lastly, the Q Exactive HF-X mass spectrometer was operated in positive/negative polarity mode with a spray voltage at 3.2 kV, capillary temperature at 320°C, sheath gas flow rate at 35 arb, and aux gas flow rate at 10 arb.

To perform peak alignment, peak picking, and quantitation for each metabolite, we processed the raw data files generated by ultra-HPLC–mass spectrometry (UHPLC-MS/MS) using the Compound Discoverer 3.0 (CD 3.0, Thermo Fisher). For data collection, the main parameters were set as follows: retention time tolerance, 0.2 min; actual mass tolerance, 5 ppm; signal intensity tolerance, 30%; signal/noise ratio, 3; and minimum intensity, 100,000. Then, peak intensities were normalized to the total spectral intensity. The normalized data were used to predict the molecular formula based on additive ions, molecular ion peaks, and fragment ions. In order to obtain accurate qualitative and quantitative results, the peaks were matched against the mzCloud and ChemSpider databases.

After Pareto scaling treatment, principal component analysis (PCA) was performed on the peaks extracted from all trial samples. The PLS-DA variable importance in the projection (VIP) values combined with P-values of the t-test were used to highlight significant differential expression metabolites. The screening criteria were as follows: threshold (VIP) > 1.0 and difference multiple (FC) > 2.0 or FC < 0.5 and P-value < 0.05. In addition, significantly differentially expressed metabolites were analyzed for expression pattern clustering using the gplots package in R (Warnes et al., 2016). Correlations between different metabolites and bacterial communities were explored with Spearman’s correlation analysis using the pheatmap package in R (Kolde, 2012). For all analyses, FDR-corrected P-values below 0.05 were considered statistically significant.

The fermentation and thiamine characteristics (pH, NH3-N, VFA, LPS, and PDH profile) are detailed in Supplementary Table 2. The data have been published in the Journal of Dairy Science (Zhang et al., 2019). In summary, thiamine altered rumen fermentation characteristics.

A total of 624,283 reads were detected from 12 samples via Illumina sequencing. Read shear filtration and quality control obtained an average of 58,551 effective data, while the quality control effective rate reached 93.88%. Sequences were clustered into OTUs at a 97% similarity level, generating a total of 1,257 OTUs. The Venn diagram shows that the intersection numbers of OTU for the CON, HG, and HGT groups were 764 (Figure 1A). Further, rarefaction curves revealed that the curve flattens out as the sequencing deepens, demonstrating that the OTU of the HGT group was higher than that of the HG group but lower than that of the CON group at the same sequencing depth (Figure 1B). Within the bacterial population, we uncovered three dominant phyla, namely, Bacteroidetes (53.41%), Firmicutes (36.32%), and Proteobacteria (3.36%) (Figure 2).

In terms of alpha bacterial diversity (Table 1), we observed no significant differences in OTU and Good’s coverage between the CON and HGT groups. Compared with the CON group, the feeding of high-grain diets (HG) markedly decreased Chao1 richness, abundance-based coverage estimator (ACE), and Shannon diversity index. Supplementation with thiamine significantly increased Shannon (P < 0.05), but the increases in ACE, Chao1 value, and Simpson index were insignificant (P > 0.05) in the HGT group compared with the HG group.

PCA revealed that goats fed with HG diet were distinctly separate from goats in the CON and HGT groups (Figure 3). Importantly, principal coordinates 1 and 2 accounted for 13.57 and 11.13% of the total variation, respectively.

At the phylum level (Table 2), the HG diet significantly decreased the relative abundance of Bacteroidetes (P < 0.01) and Firmicutes (P < 0.05), as well as increased the relative abundance of Firmicutes (P < 0.05) compared with the CON groups. Whereas supplementation with thiamine in the HGT group considerably increased the abundance of Bacteroidetes (P < 0.01), the increase of Fibrobacteres and the decrease of Firmicutes in relative abundance were insignificant (P > 0.05) compared with those in the CON group. We also identified no significant shifts in the relative abundance of Proteobacteria, Synergistetes, Euryarchaeota, Spirochaetes, Gracilibacteria, and Chloroflexi with the supplementation of HGT and thiamine compared with the CON group (P > 0.05).

At the genus level (Table 3), compared with the CON group, high-concentrate diet (HG) substantially decreased the relative abundance of Oribacterium, Acetitomaculum, Saccharofermentans, Bacteroides, uncultured Prevotellaceae, Anaerobiospirillum, Papillibacter, and Fibrobacter (P < 0.05) but also increased the abundance of Succiniclasticum, Prevotella_1 (P < 0.05), and Prevotella_7 (P < 0.01). While supplementation with thiamine largely increased the relative abundance of Oribacterium, uncultured Prevotellaceae (P < 0.05), Anaerobiospirillum, and Fibrobacter (P < 0.01), as well as decreased the relative abundance of Prevotella_1 (P < 0.05), but the effect on the relative abundance of Succiniclasticum, Acetitomaculum, Saccharofermentans, Bacteroides, and Prevotella_7 was insignificant compared with that of high-concentrate diet (P < 0.05). Notably, the results of correlation analysis revealed that phenyl propionic acid was positively associated (r > 0.38, P < 0.05) with 11 taxa and negatively associated (r < −0.38, P < 0.05) with four taxa and several ciliate and methanogens.

To explore the functional correlation between the rumen microbiome changes and metabolite perturbations, a Pearson correlation matrix was generated by calculating the Pearson correlation coefficient in high-concentrate diet with thiamine supplement (Figure 5). The results showed a clear correlation between microbial structure and metabolites (r > 0.38 or < 0.38, P < 0.05). More interestingly, correlation analysis revealed that acetate was positively associated with eight genera (r > 0.49, P < 0.05) and negatively associated with one genus (r < −0.62, P < 0.05). Propionate was positively associated with six genera (r > 0.41, P < 0.05) and negatively associated with four genera (r < −0.55, P < 0.05). Pyruvate was positively associated with 12 genera (r > 0.52, P < 0.05), while lactate was positively associated with three genera (r > 0.54, P < 0.01) and negatively associated with six genera (r < −0.45, P < 0.05). Succinates were positively associated with nine genera (r > 0.55, P < 0.05). Likewise, oxaloacetate was positively associated with nine genera (r > 0.47, P < 0.05). Leucine was positively associated with four genera (r > 0.48, P < 0.05) and negatively associated with five genera (r < −0.59, P < 0.05), whereas valine was positively correlated with five genera (r > 0.48, P < 0.05) and negatively associated with two genera (r < −0.62, P < 0.01). Putrescine was positively associated with two genera (r > 0.60, P < 0.05) and negatively associated with seven genera (r < −0.69, P < 0.01). Tyramine was positively associated with two genera (r > 0.76, P < 0.05) and negatively associated with seven genera (r < −0.66, P < 0.05). Histamine was positively associated with one genus (r > 0.78, P < 0.05) and negatively associated with eight genera (r < −0.65, P < 0.05). Linoleic acid was positively associated with three genera (r > 0.62, P < 0.05) and negatively associated with three genera (r < −0.52, P < 0.01). Docosahexaenoic acid was positively associated with four genera (r > 0.62, P < 0.05) and negatively associated with two genera (r < −0.57, P < 0.05). Finally, 4-phenylbutyric acid (4-PBA) was positively associated with one genus (r > 0.86, P < 0.01) and negatively associated with seven genera (r < −0.64, P < 0.05).

In the present manuscript, we have demonstrated the alteration of rumen fermentation characteristics by thiamine (Supplementary Table 2), and the data have been published in the Journal of Dairy Science (Zhang et al., 2019). To the best of our knowledge, the detailed mechanisms by which thiamine supplementation relieves SARA in goats and dairy cows remain largely unknown. Therefore, to determine the potential mechanism of thiamine on rumen fermentation and bacterial community, we analyzed a high-concentrate diet using 16S rRNA gene sequencing technology.

Our PCA results showed differences in the bacterial community among CON, HG, and HGT groups, indicating the impact of thiamine supplementation. Recent studies have shown that a long-term supply of HG diet results in an alteration of rumen microbial populations (Mao et al., 2015; Hua et al., 2017), while thiamine supplementation in our previous study could increase the abundance of rumen bacterial community (Pan et al., 2017a).

At the phyla level, we observed that high-concentrate dietary supplementation with thiamine might significantly improve the abundance of Bacteroidetes. We also noted that the abundance of Firmicutes decreased in the HGT group compared with the HG group, although the change was statistically insignificant. Of note, the phyla Firmicutes and Bacteroidetes are known for polysaccharide fermentation (Wang et al., 2019). Additionally, Bacteroidetes, one of the most abundant gram-negative bacteria in the lumen, are classified into three categories, namely, Bacteroidetes, Flavobacteria, and Sphingobacteria. Zened et al. (2013) reported that an increase in fermentable substrates led to a decrease in pH in the rumen during high-concentrate diet supplementation, which can subsequently lead to death and lysis of gram-negative bacteria (Cole et al., 2005). Firmicutes are predominately composed of gram-positive bacteria (Fernando et al., 2010), which are also insensitive to low pH. Similarly, these bacteria can still degrade easily fermentable carbohydrates under low pH (Wetzels et al., 2015). However, supplementation with thiamine partially reversed this effect.

At the genus level, a high abundance of Succiniclasticum was identified in the high-concentrate diet group than in the control group, whereas thiamine supplementation reversed this change, although this result was insignificant. A similar trend was also reported for Succiniclasticum with a high-concentrate diet elsewhere (Petri et al., 2013). Importantly, Succiniclasticum is an organism that specializes in fermenting succinate by converting it quantitatively to propionate, the most important precursor of glucose in ruminants (Van Gylswyk, 1995). Unfortunately, the function of Succiniclasticum in the rumen remains enigmatic. As shown in Supplementary Table 2, the changing trend of propionate content in the rumen was similar to that of Succiniclasticum. The low energy yield when anaerobes convert succinate to propionate should make this bacterium highly efficient in this conversion (Thauer et al., 1977). Thiamine is a coenzyme of pyruvate dehydrogenase (PDH) (Chen et al., 2016); the increasing thiamine concentration could facilitate the transformation of pyruvate to acetyl CoA, which ultimately might be hydrolyzed to acetate. The change reduced the production of succinate and subsequently decreased the proportion of Succiniclasticum in rumen concentration. Oribacterium has been identified as one of the predominant bacteria in the rumen of cows fed forage-based diets (Kong et al., 2010; Kenters et al., 2011), but its role in the rumen remains unknown. Noticeably, thiamine supplementation increased significantly the abundance of Oribacterium. However, the reasons for the change in abundance of this genus due to thiamine supplementation need further clarification.

The thiamine treatment group exhibited a higher proportion of the Anaerobiospirillum, which can ferment glucose or starch into succinate (Nghiem et al., 1997; Lee et al., 2000). Interestingly, thiamine is associated with the production of acetyl coenzyme A, while acetyl coenzyme A is also related to the production of succinate (Nyström, 2015). The supply of thiamine increased the utilization of the substrate by Anaerobiospirillum, hence promoting its proliferation. Besides, Asma et al. (2013) reported the differences due to dietary starch levels in the relative abundances of Fibrobacteres, whose relative abundance decreased by more than two times when the dietary starch level increased. This changing trend is consistent with our results. At present, two species belonging to the genus Fibrobacter are known, including Fibrobacter succinogenes and Fibrobacter intestinalis (Amann et al., 1992). In particular, F. succinogenes has been found to utilize its incomplete citric acid cycle to produce succinate, the bacterium’s major fermentative end product (Miller, 1978). Additionally, a recent study has demonstrated that F. succinogenes contains both a phosphoenolpyruvate carboxykinase and a pyruvate carboxylase that can reversibly convert pyruvate, respectively, to oxaloacetate, which is sequentially converted to malate, fumarate, and succinate. This degradation process requires thiamine pyrophosphate as a coenzyme factor, which is possibly one of the reasons for the increased content of succinate in the rumen. Moreover, Prevotella_1, a polysaccharide-degrading bacteria belonging to the Prevotella ruminicola family, in the HGT group tended to be lessened by thiamine supplementation that can promote the transformation of some substances in this degradation pathway (Purushe et al., 2010).

Thiamine contributes significantly to carbohydrate metabolism and is essential for normal cellular function and growth. For the effects of thiamine supplementation on carbohydrate metabolism, our metabolome data indicated that thiamine supplementation alters fermentation type with a high-concentrate diet in the rumen, implying that the ruminal metabolism pattern might be associated with structural changes in bacterial communities. Previous studies have elucidated that supplementation of nonstructural carbohydrates can affect metabolism (Afzal, 1999). As a result, change in carbohydrate structure affects the bacterial communities in the rumen (Mcallan and Smith, 1974). Carbohydrate metabolism plays a vital role in regulating animal health. For example, Sutton (1985) found that the end products of carbohydrates act as energy substrates, inflammation modulators, and signaling molecules in the rumen. As a metabolite of carbohydrates, VFA can regulate the differentiation of the rumen epithelium, cell apoptosis, and morphology of the rumen papilla (Mentschel et al., 2001; Inagaki and Sakata, 2005; Shen et al., 2005). Our findings unearthed that thiamine supplementation in a high-concentrate diet markedly increased the content of acetate and decreased the content of propionate compared with those in the HG group diet in ruminal fluid. The ratio of acetate to propionate is positively correlated with the content of the concentrate (Lechartier and Peyraud, 2010). Studies have emphasized that the succinate pathway is the principal pathway of propionate production, which is used by Bacteroidetes to generate propionate from a substrate (Louis et al., 2014). More importantly, this may be a plausible explanation for the decrease in the ratio of acetate to propionate with thiamine supplementation relative to the high-concentrate diet group. Besides, as a coenzyme factor of PDH (Bubber et al., 2004), deficiency of thiamine leads to the accumulation of pyruvate, which is then converted to lactate under the catalytic action of lactate dehydrogenase (LDH) (Kumar et al., 2015). The supplement of thiamine can promote the conversion of pyruvate into acetyl-CoA, reducing the accumulation of lactate and pyruvate. Moreover, thiamine is a coenzyme required for the activity of α-ketoglutarate dehydrogenase (α-KGDHC) that is involved in the Krebs cycle (Hanson and Cox, 1967). The increase in oxaloacetate content can be interpreted as the enhancement of the tricarboxylic acid cycle, which promotes the transformation of oxaloacetate.

For amino acid metabolism, amino acids play a pivotal role in the rumen, whereby rumen microbes are associated with amino acid metabolism (Chalupa, 1975). Our data demonstrate that the content of leucine and valine in the rumen was considerably increased by thiamine supplementation in the high-concentrate diet. Amino acids are the degradation products of microbial or dietary proteins and precursors of protein synthesis in the rumen, which can regulate some metabolic pathways (Mariz et al., 2018). Previous studies have shown that L-leucine tended to decrease the level of acetate in the rumen (Menahan and Schultz, 1964). This result concurs with our findings. The relationship between leucine and thiamine has been rarely reported in the rumen. Thiamine supplementation has been noted to promote the transformation of acetyl-CoA, which is key to amino acid metabolism (Schwartau et al., 1984). Furthermore, TPP, as a coenzyme factor of hydroxyl acetate synthase, catalyzes the first step reaction to the biosynthesis of valine (Byrne, 1998), while thiamine supplementation can enhance this metabolic pathway.

For biogenic amine metabolism, high concentrations of bioamines such as histamine, putrescine, and tyramine may induce a series of inflammatory responses in the rumen epithelium (Nocek, 1997; Fusi et al., 2004), which emerges in the rumen because of decarboxylation of precursor amino acids under the degradation of ruminal microbes (Wang et al., 2013). Studies have suggested that the absorption of histamine in the rumen increases the content of histamine in the whole organism accompanied by the damage of the rumen epithelial cells caused by lactate poisoning, which is the main factor that aggravates the condition of ruminal acidosis (Andersen and Jarlv, 1990). In this work, we found that thiamine supplementation with a high-concentrate diet significantly decreased the content of histamine, putrescine, and tyramine. It has been postulated that gram-negative bacteria die and lyse when carbohydrates ferment quickly and pH drops sharply, leading to the release and absorption of LPS and histamine (Duffield et al., 2004; Khafipour et al., 2009). The stimulation of toll-like receptor 4 (TLR4) by LPS induces the release of critical proinflammatory cytokines (Lu et al., 2008), while thiamine might regulate the pathways (Pan et al., 2017b). Besides, competitive inhibition of histamine and thiamine have been reported in the study of bacterial metabolism (Terzl and Keek, 1949), which is likely one of the reasons that thiamine supplementation reduces biogenic amines.

For fatty acid (FA) metabolism, diet composition can modify the content and composition of ruminal bacterial FAs (Bas and Archimède, 2003). Rumen microbes are assumed to play an essential role in the metabolism of trans-FA and conjugated linoleic acid (Kalscheur et al., 1997). Also, a shift in the rumen bacteria community has been reported as dietary NDF content (Dehority and Orpin, 1997). This could have influenced FA composition due to the large differences in FA composition in different community bacterial structures (Minato et al., 1988). In this study, the contents of linoleic acid and docosahexaenoic acid in the rumen were decreased significantly with thiamine supplementation in a high-concentrate diet. A reasonable explanation for FA alteration may be that thiamine supplement perhaps changed the bacterial community structure (Pan et al., 2017a). Moreover, thiamine is a coenzyme factor in the synthesis of acetyl-CoA, which is a precursor for the synthesis of FA; thus, thiamine supplementation can facilitate this pathway.

For organic acid metabolism, the 4-PBA plays an integral role in maintaining proteostasis and apoptosis of cells (Sun et al., 2012; Kolb et al., 2015). Studies have revealed that the 4-PBA reduced cellular damage as well as cell death (Sawada et al., 2008) and attenuated inflammation (Qi et al., 2011). This may be an indication that thiamine supplementation increased the content of 4-PBA, thus alleviating rumen epithelial inflammation. Currently, there are few reports on the association between 4-PBA and bacterial community structure in animals. It is also important to note that the underlying mechanism by which thiamine supplementation affects the metabolism of 4-PBA requires further investigation.

The relationship between bacterial community and metabolism with thiamine supplementation has been discussed above. In general, there was a utilization or productive association between bacteria and metabolomes in the rumen throughout the study. Mao et al. (2015) concluded that the type of feeding affects the concentration of metabolites by influencing bacterial community structure, hence affecting the rumen function. Taken together, these changes and correlations provide new insights that can reveal the mechanism of action of thiamine on animal health.

The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found below: PRJNA700131↗.