Effects of Short Chain Fatty Acids (SCFAs) Modulation on Potentially Diarrhea Causing Pathogens in Yaks Through Metagenomics Sequencing

Yaks are of great importance on the plateau; however, an emerging endemic diarrheal disease during the last few years is posing a great threat to the health of these animals. Yaks have special gut microbiotal community and short-chain fatty acids (SCFAs) which are not only the principle nutrient substrates of intestinal epithelial cells but can also regulate the epithelial barrier. Until now, metagenomics sequencing has not been reported in diarrheal yaks. A scarce information is available regarding the levels of fecal SCFAs and diarrhea in yaks. The purpose of our study was to identify the potential pathogens that cause the emerging diarrhea and also to explore the potential relationship of short-chain fatty acids in this issue. We estimated diarrhea rate in yaks after collecting the equal number of fecal samples from affected animals. Metagenomics sequencing and quantitative analysis of SCFAs were performed which revealed 15-25% and 5-10% prevalence in diarrheal yak’s calves and adults yaks respectively. Signicant difference was observed in GC contents (44.69%~46.08% vs 46.12%~46.38%) under two reference groups (p<0.05). Violin box plot also showed the higher degree of dispersion in gene abundance distribution of diarrhea yaks, while genes of normal yaks were relatively gathered. We found 366163 signicant differential abundance genes in diarrheal yaks, with 141305 up-regulated and 224858 down-regulated genes as compared with normal yaks via DESeq analysis. Metagenomic binning analysis indicated the higher signicant of bin 33 (Bacteroidales) (p<0.05) in diarrheal animals, while bin 10 (p<0.0001), bin 30 (Clostridiales) (p<0.05), bin 51 (Lactobacillales) (p<0.05), bin 8 (Lachnospiraceae) (p<0.05) and bin 47 (Bacteria) (p<0.05) were obviously higher in normal animals. At different levels, an obviously difference in Phylum, Class, Oder, Family, Genus and Species was noticed as 4, 8, 8, 16, 17 and 30 respectively. Compared with healthy yaks, Acetic acid (p<0.01), Propionic acid (p<0.01), Butyric acid (p<0.01), Isobutyric acid, Isovaleric acid (p<0.05) and Caproic acid (p<0.01) were all observed obviously at lower rate in diarrheal yaks. In conclusion, besides the increased pathogens level of Staphylococcus aureus, Babesia ovata, Anaplasma phagocytophilum, Bacteroides uxus, viruses, Klebsiella pneumonia, and inammation-related bacteria; the decreased of SCFAs caused by the imbalance of intestinal microbiota may potentially leads to emergence of diarrhea in yaks.


Introduction
The long-haired bovine specie i.e. yak is a well-known animal, which is an indispensable economic pillar on the Qing-Tibetan plateau [1]. There are approximately 15 million yaks in China accounting for over 90% of the world's yak population [2,3]. At the average of 3000-5000 m above the sea level, yak is depicted as a symbolic animal with the dependency of herdsmen's lives [1]. Yaks serve as transportations, especially in the zigzag mountainy areas [4]. Meat, butter and milk from yaks are also considered as an essential food items for local Tibetans [3]. Its hide is used for making boots, rafts, aprons, leather bags and leather harness [2]. While the long hairs and dungs are commonly used for livelihood purpose [4].
Hongyuan is located in the eastern of Qinghai-Tibetan plateau and northwestern of Sichuan province, with the northern latitude of 31°51′-33°33′ and eastern longitude of 101°51′-103°22′. In this continental plateau the temperature is ranged from − 22.8 o C to 24.6 o C with the average temperate of 2.9 o C and 860.8 mm precipitation annually. In this region animal husbandry is the primary industry with 75383 yak's population according to latest reports. About 7765 Kg of yak's meat has also been produced annually accounting for 96.23%. However, an emerging endemic diarrheal disease in yaks during the last few years (usually from May to August) not only has caused deaths to animals but also caused a huge constraint on the development of local economy.
Bovine diarrhea is a common disease throughout the farms with world-wide distribution. It has been causing a heavy economic loss in concern with fertility rate, milk production, and animals growth [5]. In yaks, diarrhea causing pathogens i.e. Cryptosporidium parvum, bovine viral diarrhea (BVD) virus, Escherichia coli have been reported continuously [6][7][8]. Although, many measures have employed to improve the hygiene and feeding management with the utilization of extensive drugs, but problem is still at peak [9].
The well-known complex intestinal tract is colonized by a large and diverse type of microbial microbiota [10]. This community produces extensive amount of metabolic products in intestine, which interact intimately with host cells to maintain physiological processes and functions i.e. nutrition absorption, host metabolism, and immunity [10][11][12]. Mainly, microbiota bene ts the host through intestinal epithelium and by producing bene cial metabolites, that helps in food digestion and also against the pathogenic invasion. Gut microbiome has ability to convert fermentable dietary bers into short-chain fatty acids (SCFAs) that provides additional energy to the host [13]. These SCFAs are organic carboxylic acids with less than 6 carbon atoms of acetate; among them propionate and butyrate are the most abundant protraction in intestine [14]. A previous study reported that diets containing alfalfa meal and commodity concentrated ber could drop diarrhea rate via metabolic interactions between hindgut microbiota and SCFAs in piglets [15]. These SCFAs are also known for acting as ligands for G-protein coupled receptors by activating anti-in ammatory signaling [16]. Metagenomics sequencing is commonly utilized in microbial organisms, as it provides accurate classi cation of microbiota species and annotation to the bacteria at functional level rather than functional prediction [17][18][19].
It is generally accepted that microbiota composition and function contribute to the health status of the host [20]. Previously, dysfunctional gut microbiota was reported to be related with diseases like human in ammatory bowel disease, diabetes and cardiovascular disease [12,21]. The imbalance of such intestinal microbiota may cause diarrhea due to the growing conditional pathogens, mucosal barrier damage, immunity dropping and intestinal permeability [20]. However, it is still unclear how the changed microbiota can cause the emerging diarrheal disease in yaks. Hence, this study was carried out to explore such potential pathogens and short chain fatty acid changes that cause emerging diarrheal disease in yaks.
Sequencing data of yak microbiota samples and Gene abundance distribution Overall, 445199120 total reads and 445089080 clean reads were obtained from diarrheal yaks; while 285976660 and 285951940 total and clean reads were obtained respectively in normal yaks. Moreover, 66295299314 and 42671183625 clean bases were found in diarrheal and normal yak groups respectively. The Q20 and Q30 in both groups was more than 97% and 92%, which con rmed the reliable and accurate base recognitions [22]. Though no signi cant difference (p > 0.05) was observed in total reads, clean reads, Q20, and Q30. However, obvious difference was found in GC content between diarrheal (44.69%~46.08%) and normal yaks (46.12%~46.38%) (p < 0.05) (Fig. 3). According to violin box plot, the degree of dispersion in gene abundance distribution was higher in diarrheal than normal yaks (Fig. 4).

Species composition and abundance analysis
Annotated analysis was preformed via MetaPhlAn2 (http://huttenhower.sph.harvard.edu/metaphlan2, Version 2.0) by comparing with database. Results were showed through Krona [23] which revealed the abundance of Firmicutes and Proteobacteria in normal yaks. In comparison with normal, yaks with diarrhea were observed with a signi cant drop of Firmicutes (p < 0.05) (Fig. 5).
At Phylum level, Firmicutes and Bacteroidales were found primarily in both groups (Fig. 6a). Principal component analysis (PCA) found the left side location of D1, D2, D3, D4, D5, and D6 groups. While NA, NB, NC, and ND groups were found to be located on the right side in two-dimensional graphic representation. Samples in normal yaks were concentrically distributed as compared to the diarrheal yaks (Fig. 6b). Compared with normal animals, Bacteroidetes (p < 0.01) and Apicomplexa (p < 0.05) were signi cantly higher in diarrheal yaks, while Firmicutes (p < 0.05) and Euryarchaeota (p < 0.001) were obviously at lower levels ( Fig. 6c).
At Species level, Firmicutes bacterium; CAG:110 and Clostridiales were found most abundantly in normal yaks, while Staphylococcus aureus was the main species in diarrheal yaks (Fig. 11a). PCA indicated that samples in diarrheal animals located separately as compared to normal animals (Fig. 11b). Compared with normal animals, Staphylococcus aureus (p < 0.05), Bacteroides coprophilus (p < 0.01), Bacteroides plebeius (p < 0.01), Butyricicoccus pullicaecorum (p < 0.01), Babesia ovata (p < 0.05), Fusobacterium mortiferum (p < 0.05),  (Fig. 11c). The statistics of compositional signi cant species in comparison with normal animals is shown in Fig. 12. Circus map indicated that the Phylum level of two groups is mainly consisted of Firmicutes, while obvious difference was found in case of Fusobacteria, Bacteroidetes and Proteobateria in both groups. While difference was found in Bacteroidia, Bacilli and Gammaproteo at the species level (Fig. 13).

Functional analysis of yak intestine microbiota
Functional pro le analysis was performed via annotating against the GO, KEGG, EggNOG and CAZy databases [24][25][26][27]. Annotated scores of one having HSP > 60 bits were selected for analyzing relative abundance at different functional levels [28][29][30][31]. In total, 354 990, 486 219, 778 943, 867 820, 366 984 and 188 719 nonredundant genes were found in GO, eggnog, KEGG, NR, swissport and CAZy, respectively. In KEGG, nonredundant genes were related to cellular community, energy metabolism and 40 more metabolic pathways in all yaks. Slightly lower of nervous system, development, and nucleotide metabolism was found in diarrheal yaks ( Fig. 14a). In eggNOG, about 24 cell metabolic pathways i.e. wall biogenesis, chromatin structure and dynamics were reported in all animals. In secondary metabolites biosynthesis, signal transduction mechanisms were a bit higher in diarrhea animals, while translation, ribosomal structure and biogenesis were slightly higher in healthy yaks (Fig. 14b). In CAZy, carbohydrate-binding modules, glycosyl transferases, glycoside hydrolases, polysaccharide lyases, auxiliary activities and carbohydrate esterase were seen in both of the two groups.
Moreover, glycoside hydrolases was found higher in normal yaks, while glycosyl transferases was higher in diarrhea yaks (Fig. 14c).
DESeq analysis was employed to uncover signi cant differential abundance genes between the two yak groups at Fold Change ≥ 2 and p-value < 0.01 [32]. Compared with normal yaks, there were 366163 obviously signi cant differential abundance genes in diarrheal yaks, with 141305 up-regulated and 224858 downregulated (Fig. 15a). Differential abundant genes were compared against cluster of orthologous proteins database. Most of the genes were found related with amino acid metabolism, replication, recombination, cell wall biogenesis, carbohydrate transportation & metabolism, translation, ribosomal structure and biogenesis ( Fig. 15b). Annotation of differential abundant genes in KEGG path showed the genes relationship with metabolism (Fig. 15c). Enrichment analysis of differential abundant genes in KEGG pathway revealed the top 24 obvious over-presentation genes. Enrichment Factor in the X-axis represented the signi cant enrichment level of differentially expressed abundant genes. Lipopolysaccharide biosynthesis gen was at the highest level of differentially expressed abundant genes (Fig. 15d). Most signi cantly different genes were related to ribosome (p < 0.001), peptidoglycan biosynthesis (p < 0.001), homologous recombination (p < 0.001).

Quantitative analysis of SCFAs in yaks
Sample quality control analysis showed that the TIC from different samples nearly overlapped completely, which indicated that the current data were repeatable and reliable (Fig. 17). The TIC from yak mixture samples showed several single waves without overlapping that reveals valid results of SCFAs (Fig. 18). In present study, all the correlation coe cient (R 2 ) of each equation of linear regressions were over 0.994 ≈ 1.000, which ensured the accuracy of SCFAs values (Table 4).
A signi cant difference of the seven SCFAs except Valeric acid was found between normal and diarrheal yaks. Acetic acid, Propionic acid, Butyric acid, Isobutyric acid and Caproic acid were found in normal yaks which were obviously higher than diarrheal yaks (p-value < 0.01). It was also admirable that Isovaleric acid in normal yaks was also a slightly higher than diarrhea yaks (p-value < 0.05) (Fig. 19).

Discussion
As a commonly reported disease, cattle diarrhea causes considerable economic losses for cattle producers world-widely [33]. In Norway, about 10 million US dollars loss was noted due to calves death effected by diarrhea in 2006 [34]. As an agricultural country, the development of animal husbandry is important especially in Hongyuan (China) like plateau areas. In our study, the prevalence of diarrhea in yaks was estimated about 15-25% and 5-10% in yak calves and adults yaks respectively (Table 1). Diarrhea in yaks was signi cantly higher in yak calves (Fig. 2), which was in line with the widely accepted knowledge that morbidity and mortality of diarrhea in calves is more serious [35]. Therefore, discovering the potential causes of this emerging diarrhea is urgent and meaningful, especially on the remote plateau.
The intestinal microbiota is also considered as an additional organ, which comprises of billions of microorganisms. Intestinal microbiota is not only important in the synthesis and metabolism of nutrients, hormones, vitamins, but also plays role in drugs utilization, pathogen's forti cation, and immune system maturation [13,36,37]. Therefore, the imbalance of intestinal microbiota may lead to serious disease.
Previously, we performed high-throughput sequencing of intestinal micro ora from diarrheal yaks, our study found 41 genera of bacteria in perinatal healthy yaks, while 145 genera of bacteria were only tested in healthy perinatal yaks [5]. Moreover, 212 genera of fungus were found in healthy yaks, 373 and 208 genera of fungus were found in calves and diarrheal yaks respectively [38]. However, 16 s RNA sequencing was limited to genus level. In the current study, metagenomics sequencing was employed to explore the potential pathogens of diarrhea in yaks. Obvious difference was found in GC content (44.69%~46.08% vs 46.12%~46.38%) of two yak groups (p < 0.05) (Fig. 3). Violin box plot also showed the higher gene abundance in diarrhea yaks in concern with the degree of dispersion than normal yaks (Fig. 4). Such results may predict the different composition of microbiota in diarrhea and normal animals. In different levels, we found more signi cant lower species composition in diarrhea yaks (Fig. 12). Overall, 366163 obvious signi cant differential abundance genes with 141305 up-regulated and 224858 down-regulated genes was found in diarrheal yaks as compared with normal yaks via DESeq analysis (Fig. 15a). Metagenomics binning analysis with bin 33 (Bacteroidales) (p < 0.05) was signi cantly higher in diarrheal animals, while bin 10 (p < 0.0001), bin 30 (Clostridiales) (p < 0.05), bin 51 (Lactobacillales) (p < 0.05), bin 8 (Lachnospiraceae) (p < 0.05) and bin 47 (Bacteria) (p < 0.05) was obviously higher in normal animals ( Fig. 16a & b).
Staphylococcus aureus is commonly known bacterium related to human and animal foodborne diseases [39]. This pathogen also causes orthopedic implant-associated infection, especially methicillin-resistant bacteria [40]. As infected animals are commonly treated with antimicrobial agents, thus the serious antimicrobial resistance is becoming a public concern world-widely [39]. Diseases such as gastroenteritis, nausea, vomiting, abdominal cramps and etc. are usually seen in infected individuals [41]. The increasing of Staphylococcus aureus in diarrheal yaks may indicate a potential threaten for local herdsmen. Bacteroides coprophilus was previously reported as pro-in ammatory in ankylosing spondylitis [42], which may infer with in ammatory status of diarrhea yaks. Bacteroides plebeius was previously found signi cantly higher in type 2 diabetes mellitus patients [43], which was also regarded as a biomarker of this disease. Thus the increase of Bacteroides plebeius in diarrhea animals means the abnormal glucose metabolism in yaks.
The butyrate-producing bacteria Butyricicoccus pullicaecorum is commonly linked with in ammatory conditions of intestinal ecosystem [44], which may cause inferred in ammatory response during diarrhea in yaks. Though Babesia ovata is a low pathogenic species, but its infection may lead to severe damages in cattle when co-infected with Theileria orientalis [45]. Previous study reported that the prevalence of T. orientalis in yaks was 9.7% on the plateau [46]. The infection of T. orientalis may be the main reason for bloody diarrhea in yaks (Fig. 1). Fusobacterium mortiferum was usually isolated from Crohn's and Behcet's disease patients [47], also Ruminococcus gnavus is a Crohn's disease-associated pathobiont [48], which was in line with the diarrhea symptom in yaks. Anaplasma phagocytophilum is a commonly reported emerging tick-borne zoonotic pathogen causing anaplasmosis [49]. This bacterium primarily infects host neutrophils, which break the rstline immune defensive barrier in mammalians [50]. The infected animals show typically anemia [51], which reveal that A. phagocytophilum may contribute to diarrhea in yaks. Bacteroides uxus is a pathogenic species of Bacteroides that displays numerous and high rate of antibiotic resistance. Higher abundance of Bacteroides uxus means, this bacterium acting as potential role in diarrhea. Firmicutes bacterium was reported to be associated with lipogenesis metabolism in animals with nonalcoholic fatty liver disease [52]. The increased Firmicutes bacterium (CAG:424) in diarrhea yaks may cause dyslipidemia. Bovine viral diarrhea and Rotavirus were also reported in yaks [53,54], which could be inferred that the increased abundance of these virus may cause diarrhea in yaks.
Butyrate producing Clostridiales bacterium is associated in protection of host from colorectal cancer, immune, and metabolic disorders [60]. It means dropped Clostridiales bacterium in yaks made contribute to diarrhea. Ruminococcus avefaciens works with noncellulolytic Treponema or Butyrivibrio species that can accelerates the digestion of cellulose [61]. The lower Ruminococcus avefaciens in diarrheal yaks may decrease the cellulose e ciency. Previously lower Ruminococcaceae bacterium was found in hospitalized patients, Cirrhosis [52], and diarrhea foals [62]. The deceased of this bacterium may insight that Ruminococcaceae bacterium has relationship with diarrhea in yaks. CAG:413, CAG:448 and Clostridia bacterium are belongs to Clostridium genus, which were recognized as bene cial bacteria to the host [63]. The lower of these three Clostridium spp. may promote diarrhea in yaks. Methanobrevibacter olleyae and Methanobrevibacter ruminantium composed the M. ruminantium clade, which belongs to ruminant Methanobrevibacter genus [64]. These two bacteria with other Methanobrevibacter spp. compose the rumen methanogenic community [65], which indicates that diarrheal yaks also have decreased production of methane. Bacteroidales bacterium WCE2008 is a Bacteroidales specie, which was accepted as"bene cial" microbes [66]. Previously dropped abundance of Bacteroidales was found in pediatric patients with CD [66], which may infer that the imbalance of this bacterium is related to diarrhea in yaks. Higher abundance of Bacteroidete bacterium is related to healthy lean of host, as it can generate three main SCFAs, butyrate, acetate and propionate [67]. The decreased Bacteroidete bacterium in animals contribute to diarrhea. Anaerotruncus can utilize cheese whey to produce acetic and butyric acids [68]. The decreased Anaerotruncus sp. Cag:390 in diarrhea may effect fatty acid metabolism in ruminants.
Micro ora is a key regulator of digestion, extraction, synthesis, and absorption of many nutrients and metabolites i.e. bile acids, lipids, amino acids, vitamins, and short-chain fatty acids (SCFAs) [37]. SCFAs are not only principle nutrient substrates of intestinal epithelial cells, but also can regulate the epithelial barrier [69].
Previously, concentrations of SCFAs was related with diarrhea-predominant irritable bowel syndrome patients [69]. SCFAs could mitigate adenine-induced chronic kidney disease [70], Preoperative fecal levels of SCFAs had an important impact on the occurrence of postoperative infectious complications in patients with esophageal cancer [71]. SCFAs in fecal samples was commonly used as an approximation of gut levels, which can infer the relationship between intestinal SCFAs production and fecal levels [72]. In the current study, 6 out of 7 SCFAs were uncovered signi cantly lower in diarrhea yaks from 100 mixed fecal samples by employing GC-MS/MS (p < 0.05) (Fig. 5). It reveals that the imbalance of gut microbiota dropped the levels of SCFAs in diarrhea due to the extensive immunological and regulatory functions of SCFAs in the host-microbe interactions [73], activating anti-in ammatory signaling via acting as ligands of G-protein coupled receptors e.g. GPR109A, GPR41, and GPR43 [16]. The current results are in line with diarrhea-dominant IBS with lower levels of SCFAs [74]. Among the common SCFAs, acetate (C2), propionate (C3) and butyrate (C4) are the most in number, produced by anaerobic fermentation of dietary bers in intestine [16]. Those three SCFAs are accounted for 90% of SCFAs produced by gut microbiota, which depicts the bene cial effects on intestinal epithelial cells and immune cells in the intestinal mucosa [75,76].
Acetate was reported to mediate joint in ammation in a murine gout model via in ammasome assembly and IL-1β [77]. Propionic Acid was found to be increased in gut-associated Treg cells (relates to systemic immune reaction and disease amelioration) [78]. Butyrate is not only a primary energy source for colonocytes, but also can maintain intestinal homeostasis through anti-in ammatory actions via inhibiting nuclear factor kappa β, and histone deacetylation by promoting epithelial barrier function [16,79]. Previously, lower abundance of butyrate-producing bacteria and fecal butyrate were found in stroke patients as higher risk factors [80]. Bacteroidetes from Firmicutes mainly produce acetate and propionate, while Butyrate is mainly produced by phylum Firmicutes i.e. Faecalibacterium prausnitzii, Clostridium leptum, Eubacterium rectale and Roseburia spp. [81]. In a previous study, Firmicutes phylum was found clearly lower in diarrheal yaks (p < 0.05) [5]. Also genus of Clostridium_IV (p < 0.01) and Clostridium_XI (p < 0.05) were found obviously lower in diarrheal yaks except Clostridium XVIII (p < 0.01). Genera of Bacteroides (p < 0.05) and Faecalibacterium (p < 0.05) were found signi cantly higher in diarrhea yaks, while no signi cant difference was found in genera of Eubacterium, Eubacterium and Roseburia [5]. However, among all those genera, Clostridium_IV and Clostridium_XI were the dominant [5], which may uncover that the decreasing of clostridium may cause the drop of SCFAs (C2-C4). In current study, Isobutyric acid, Isovaleric acid and Caproic acid were found signi cantly lower in diarrheal animals (p < 0.05), which was in line with patients suffering from cirrhosis and neuromyelitis optica spectrum disorders [82,83]. As SCFAs plays a critical role in mucosal integrity and immune response [84]. So, the dropping of SCFAs (C4-C6) may mean the damage of mucosal and in ammation response. In a sentence we can say, SCFAs generation bacteria of Anaerotruncus sp. Cag:390, Clostridiales bacterium and Butyricicoccus pullicaecorum are lower in number in diarrheal yaks. Although Fusobacterium mortiferum producing butyric and acetic acids increase obviously [85]. But, it could not affect the dropping trend of SCFAs in diarrheal animals. Statistical analysis of SCFAs; relevant to dominant KEGG signal pathways related to SCFA Acetic acid (53.85%) (Fig. 19); which was the primary level of acetate (50-70%) in the intestine [86].
In conclusion, we estimated the prevalence of emerging diarrhea disease in yak calves (15-25%) and adults (5-10%). Besides the high prevalence of Staphylococcus aureus, Babesia ovata, Anaplasma phagocytophilum, Bacteroides uxus, viruses, Klebsiella pneumonia, and in ammation-related bacteria, the decreased of SCFAs may potentially lead to emerging diarrhea in yaks. Our results will make insights to the prevention and treatment of emerging diarrhea disease in yaks on the cold plateau.

Ethics statement
Fecal samples were collected under the permission of the relevant institutions. All procedures were performed under the instructions and approval of Laboratory Animals Research Centre of Hubei province and Sichuan province, and also under the ethics committee of Huazhong Agricultural University in P. R. China.

Sample collection
We visited 10 family yak farms with diarrhea out-break during June and July, 2019 (Table 1) in Hongyuan, Sichuan, China. The prevalence of diarrhea at farms was estimated by consulting animal owners as these bovines were free-ranged having grasslands without concentrated feed on the plateau. A total of 120 fresh fecal samples were collected from diarrheal (n = 60) and healthy (n = 60) yak calves. All the fecal samples were frozen immediately in liquid nitrogen and then transported to the laboratory of Huazhong Agricultural University. Samples were kept at -80 °C for further appraisement.

Library construction and sequencing
Metagenome shotgun sequencing libraries (400 bp) were constructed by using Illumina TruSeq Nano DNA LT Library Preparation Kit. Each library was sequenced by employing Illumina HiSeq X-ten platform (Illumina, USA) with PE150 strategy (Shanghai, China).

Sequence Analysis
Further analysis to achieve quality-ltered reads, the sequencing adapters were removed from raw sequencing reads by using Cutadapt (v1.2.1) [87]. Low quality reads were trimmed by performing a sliding-window algorithm. Reads were aligned to the host genome via BWA (http://bio-bwa.sourceforge.net/) to remove host gene contamination [88]. Then quality-ltered reads were de novo assembled to construct the metagenome for each mixed sample by IDBA-UD (Iterative De Bruijn graph Assembler for sequencing data with highly Uneven Depth) [89]. All coding regions (CDS) of metagenomic scaffolds longer than 300 bp were predicted by MetaGeneMark (http://exon.gatech.edu/GeneMark/metagenome) [90]. CDS sequences of current samples were clustered by CD-HIT at 90% protein sequence identity, to obtain non-redundant gene catalog [91]. Gene abundance in each sample was estimated (http://soap.genomics.org.cn/) on the base of aligned reads number.
The lowest common ancestor taxonomy of the non-redundant genes was obtained by aligning them against the NCBI-NT database by BLASTN (e value < 0.001). Similarly, the functional pro les of the non-redundant genes were obtained by annotated against the GO, KEGG, EggNOG and CAZy databases by utilizing DIAMOND (Buch nk) alignment algorithm [92].
Comparing of the difference of intestine microbiota between normal and diarrheal yaks Based on the taxonomic and functional pro les of non-redundant genes, LEfSe (Linear discriminant analysis effect size) was performed to detect differentially abundant taxa and functions across the groups by using default parameters [93]. Beta diversity analysis was performed to investigate the compositional and functional variation of microbial communities across diarrheal and healthy yak samples through Bray-Curtis distance metrics [94]. Visualization was done via principal coordinate analysis (PCoA), nonmetric multidimensional scaling (NMDS) and unweighted pair-group method with arithmetic means (UPGMA) hierarchical clustering [95]. Differential gene (Up-regulated gene and Down-regulated gene) abundance analysis was preformed via DESeq at Fold Change ≥ 2 and p-value < 0.01 [28]. Metagenomics binning analysis was carried out by using Maxbin2 and Maxbat2 [96] at > 50% genomic integrity and < 10% contamination rate.
Extraction of fatty acids from fecal samples Firstly 20 mg fecal from each sample was taken out to mix with 1 mL Phosphoric acid (0.5% v/v) in a sterile 2 mL EP tube, and then mixed up thoroughly via vortex and ultra-sonication for 10 and 5 minutes respectively. Secondly, 0.1 mL sample was taken out and then putted into a sterile 1.5 mL EP tube with the edition of 0.5 mL MTBE (CAS NO. 1634-04-4). Final product was mixed up thoroughly via vortex and ultra-sonication for 3 and 5 minutes respectively. Thirdly, at 12000 rpm sample was centrifuged at 4 o C for 10 min. After taking out 0.2 mL from supernatant. Mixed 10 extracted samples from the same group and vortex for 1 minute. At the end, 0.2 mL mixture sample was taken out into sample detection vial for further analysis through GC-MS/MS (Agilent).

Qualitative and quantitative analysis of SCFAs
Total ions current (TIC) and standard quality of all mixture samples were detected through GC-MS/MS (Agilent) by using the procedures and parameters showed in Table 2. Sample's quality control analysis was carried out to ensure the validity of the method. Standard of quality samples was checked thrice in concern with instrument stability. This standard quality sample was tested in every ten samples to monitor the repeatability of the analysis process. Qualitative and quantitative analyses of SCFAs were performed by Agilent MassHunter.
Standard curves for all SCFAs were generated by detecting standard quality control samples which were

Statistical Analysis
The prevalence of diarrhea at different farms was analyzed through IBM SPSS Statistics (SPSS 22.0) by using chi-square (results were followed as upper and lower limits prevalence). Quantitative analyses of SCFAs were expressed as means ± standard deviation (SD). While the difference of SCFAs among the groups were analyzed via Wilcoxon test and Fold changes through piloting SPSS (IBM, 22.0). Signi cance level was kept as p < 0.05. T-test was also performed to compare intestinal microbiota difference by using IBM SPSS Statistics. Declarations