Garlic Skin Induced Shifts in the Rumen Microbiome and Metabolome of Fattening Sheep

Background: Garlic and its constituents exhibit activities on modifying rumen fermentation and improving growth performance. As a by-product of garlic processing, garlic skin contains similar bioactive components as garlic bulb. However, studies in ruminants using garlic skin are scarce. This experiment was conducted to investigate the effects of garlic skin supplementation on rumen fermentation characterizes, growth performance, and involved mechanism in ruminants. Twelve Hu lambs were randomly assigned into one of two treatments: basal diet (CON) or basal diet supplemented with 80 g/kg DM of gallic skin (GAS). The experiment lasted for10 weeks, with the rst 2 weeks for adaptation. Results: The results revealed that the average daily gain and volatile fatty acid concentration were higher (P < 0.05) in lambs fed GAS than that in the control group. Garlic skin supplementation did not signicantly (P > 0.10) affect the α-diversity indices. Increased (P < 0.05) abundances of Prevotella, Bulleidia, Howardella, Methanosphaera but a decreased (P < 0.05) abundance of Fretibacterium were observed in GAS-fed lambs. Besides, the garlic skin supplementation favorably regulated (P < 0.05) pyrimidine metabolism, purine metabolism, vitamin B6 and B1 metabolism. Moreover, high correlations were observed between uctuant rumen microbiota and metabolites. Conclusions: Supplementation of garlic skin improved the growth performance of sheep by modifying rumen fermentation through inducing shifts in the rumen microbiome and metabolome.


Introduction
In recent years, plant-derived bioactive compounds have been concerned for its potential alternatives to growth promoting antibiotics in ruminant production [1][2]. Garlic (Allium sativum L.), which has been widely used as a foodstuff in the world, contains numerous active metabolites such as sulfur compounds (allicin, alliin, diallyl sul de, and diallyl trisul de) [3][4]. These components are known to possess antimicrobial, antibacterial, antioxidant, anti-in ammatory, and anticancer [5][6][7].Due to its antimicrobial properties, garlic power [8] and garlic oil [9] exhibit activities on modifying rumen fermentation parameters, improving nutrient digestibility, decreasing rumen protozoa numbers, and reducing methane emissions.
The annual production of garlic is approximately 20 million tons in the world, with China, India, and Korea being the main producers [10]. As a by-product of garlic processing, garlic skin contained similar bioactive components as garlic bulb [11]. Garlic bulbs yield approximately 760 g cloves, and 240 g garlic skin per kilogram [12]. Therefore, garlic skin could be of importance due to its abundance and possible utilization as ruminant feed stuff. However, to the best of our knowledge, no previous studies investigated the effects of garlic skin on the rumen fermentation characteristics, growth performance, and involved mechanism in ruminants.
Rumen harbors a wide variety of microorganisms, and rumen microbiome can directly and indirectly affect performance, health, and immune system of the host [13][14].During diet digestion, the rumen microbiota coproduces a large array of small molecules, which play critical roles in shuttling information between the microbial symbionts and their host's cells [15]. Therefore, a better understanding of the rumen microbiota composition and the metabolome is crucial to investigate the mechanism involved in feedstuff affecting rumen fermentation. We hypothesized that the rumen microbiota and rumen metabolites might be affected by garlic skin, which would demonstrate a regulatory role of garlic skin on rumen metabolism. The aim of this study was to investigate the effects of garlic skin on the dynamic changes in the ruminal microbiome and metabolome in growing sheep.

Material And Method
Animals and experimental design All the experimental protocols were approved by the Animal Care Committee of Anhui Agricultural University. Twelve Hu lambs (local breed, 23 ± 2.3 kg initial body weight) divided into two groups (n = 6), were fed basal diet (CON) or basal diet supplemented with 80 g/kg DM of gallic skin (GAS). The experimental period lasted for 10 weeks, with the rst 2 weeks for adaptation. The basal diets were formulated to meet the Feeding Standards of Meat-producing Sheep and Goats in order to achieve a daily gain of 0.250 kg (Ministry of Agriculture of P. R. China, 2004; Table 1). All the animals were housed in individual pen with free access to water and were fed twice daily at 07:00 and 19:00 h, with approximately 10% feed refusal.  [20]. Another two 5 mL subsamples of were infused into two10 mL spiral centrifuge tubes and immediately placed into a liquid nitrogen container and then transported to the laboratory, stored at -80 °C for further analysis of microbiota and metabolites.

Bioinformatics
The raw reads containing ambiguous bases and any longer than 480 base pairs (bp) were dislodged and those with a maximum homopolymer length of 6 bp were allowed, and sequence short than 200 bp were removed [21].Paired-end raw reads were merged as raw tags using FLSAH (v 1.2.3) with a minimum overlap of 10 bp [22].Noisy sequences of raw tags were ltered by the Pre.cluster tool to obtain clean tags. Chimeras were detected by using Chimera UCHIME. The effective reads were clustered into operational taxonomic units (OTUs) of ≥ 97% similarity using the UPARSE pipeline [23], and the representative sequence of each OUT was classi ed as organisms (phylum, class, order, family, and genus) by Naïve Bayesian assignment using an RDP classi er (version 2.12) [24].Species richness and diversity statistics including coverage, rarefaction, Chao1, ACE, Simpson, and Shannon were calculated using Mothur (version 1.30.1). A weighted and unweighted UniFrac distance matrix was also generated by Mothur (version 1.30.1), and a principal coordinate analysis (PCoA) was conducted based on the unweighted Unifrac distance method [25].

Metabolomics Data Analysis
The injection volume was 1 µL (positive) or 1 µL (negative), respectively. During the operation, the acquisition software (Analyst TF 1.7, AB Sciex) continuously evaluates the full scan survey MS data as it collects and triggers the acquisition of MS/MS spectra depending on preselected criteria. In each cycle, the most intensive 12 precursor ions with intensity above 100 were chosen for MS/MS at collision energy (CE) of 30 eV. The cycle time was 0.56 s. MS raw data were converted to the mz XML format by Proteo Wizard and processed by R package XCMS (version 3.2). The process includes peak deconvolution, alignment and integration. Minfrac and cut off are set as 0.5 and 0.3 respectively. In-house MS2 database was applied for metabolites identi cation. A total of 1812 resulting peaks with the peak numbers, sample names, and normalized peak areas were import to the SIMCA software package (V15.0.2, Sartorius Stedim Data Analytics AB, Umea, Sweden) for principal component analysis (PCA) and orthogonal projections to latent structures-discriminant analysis (OPLS-DA). Additionally, 7-fold permutation tests in the OPLS-DA model were used to verify model validity and robustness. Encyclopedia of Genes and Genomes (KEGG, http://www.genome.jp/kegg/) was utilized for metabolite identi cation and con rmation, and MetaboAnalyst 4.0 was used for the pathway topology analysis.

Statistical analysis
The matter intake (DMI), growth performance, and rumen fermentation characteristics were analyzed as a randomized block design using PROC MIXED of SAS software (version 9.3, SAS Institute Inc., Cary, NC, United States). Dietary treatment was included as a xed effect, and lamb was included as a random effect. Means were separated using the PDIFF option in the LSMEANS statement.
Statistical calculations rumen bacterial community, metabolomic, and their correlation data were carried out by conducting tests using the SPSS software package (SPSS version 23.0; SPSS Inc., Chicago, IL, United States). The Kruskal-Wallis sum-rank test was used to select and demonstrate differentially abundant taxa between the groups. For metabolite identi cation, the rst principal component with a variable importance in the projection (VIP) value > 1.0 and a P value < 0.05 in Student's t test were considered signi cantly different.
The correlations between different rumen microbial genera (P < 0.05 and relative abundance > 0.05% in at least one of the samples) and varied altered rumen metabolites (VIP > 1.0 and P < 0.05) were assessed by Spearman's correlation test. Signi cance was declared at P < 0.05, and a tendency was declared at 0.05 ≤ P < 0.10.

Results
Dry matter intake and growth performance As shown in Table 2, dry matter intake was not signi cantly affected by the dietary treatments (P > 0.05). Lambs consuming the GAS diet had signi cantly higher ADG than lambs in the control group (P < 0.05). Table 2 Dry matter intake (DMI) and growth performance for the dietary treatments (n = 6 lambs per treatment

Rumen Fermentation Parameters
Rumen pH was similar (P > 0.05) between GAS and the control group (Table 3). The ammonia-nitrogen decreased signi cantly in GAS compared with CON (P < 0.05). Total concentrations of VFA and concentrations of acetate demonstrated signi cant (P < 0.05) or trend (P < 0.10) responded to GAS compared with CON fermenters. Concentrations of propionate, butyrate, valerate, and acetate-topropionate ratio did not differ (P > 0.05) between the two dietary groups.

Bacterial Analyses Of Rumen Samples
In total, 1,233,294 raw reads were obtained for the bacterial 16S rRNA genes in the two groups. After screening, 1,217,625 effective tags were obtained, accounting for 98.7% of the raw reads. The Good's coverage values for all samples were greater than 99.5%. The attened rarefaction demonstrated that most of the diversity was captured (Fig. S1). The α-diversity indices showed no signi cantly difference (P > 0.05) between GAS and CON. Based on the unweighted UniFrac distances, the PCoA revealed that the microbiota structure of the GAS group was separated from that of the control group (Fig. 1).
The rumen-microbiota taxonomic distributions at the phylum, family, and genus levels are shown in Fig. 2. In total, 20 bacterial phyla were identi ed in the rumen samples. The relative abundances revealed that Bacteroidetes (58.7 ± 7.72;mean ± standard deviation) was the most dominant phylum, followed by Firmicutes (31.3 ± 8.00) ( Fig. 2A, B). The relative abundances of Synergisteteswere signi cantly decreased (P < 0.05) in GSA compared with CON. However, the relative abundance of Firmicutes tend to increase (0.05 < P < 0.10) in GSA compared with CON.
There were 260 bacterial taxa identi ed at the genus level (Fig. 2C, D) Table 5 compares the ruminal bacteria relative abundance between the diets. Among these, 5 genera bacteria were identi ed in comparisons between GAS and CON. The percentages of Prevotella (P < 0.05), Bulleidia (P < 0.05), Howardella (P < 0.01), and Methanosphaera (P < 0.01) were signi cantly increased in GAS compared with CON. In contrast, the proportions of Fretibacterium were signi cantly decreased (P < 0.05) in GAS compared with CON.

Metabolomic Analyses Of Rumen Samples
Our untargeted LC-MS approach led to the dei cation of 1004 metabolites out of all 12 sheep samples.
The PCA plot was conducted to visualize the trends ad outliers, showing noticeable separations between GAS and CON (Fig. 3A, B). In order to further examine the metabolic changes, OPLS-DA analysis was performed to compare the two groups' rumen uid samples (Fig. 3C, D). R2Y (cum) and Q2 (cum) were respectively of 0.956 and 0.608 in negative ion mode and 0.959 and 0.678 in positive ion mode, which indicated a stable and accurate prediction of the both models. In addition, a random-permutations test was carried out to prevent over t of the models (Fig. 3E, F). The intercept of Q2 on Y axis was − 0.63 in negative ion mode and − 0.80 in positive ion mode (less than 0.05), showing that the models had good predictability and did not over t. All the samples in the score plots of the rumen uid inside the 95% Hotelling's T-squared ellipse. Clear separation and discrimination were found between GAS and CON both in the PCA and OPLS-DA plots.
The potential biomarkers were screened according to the VIP value (> 1.0) from OPLS-DA modeling and statistical tests (P < 0.05) in Student's t test. Compared with those in the control group, a total of 139 metabolites (40 in the negative mode and 99 in the positive mode) changed signi cantly in the rumen uid of the GAS sheep. Among these metabolites, 93 metabolites in GAS were up-regulated, and 46 metabolites were down-regulated (Table S1). These metabolites involved in protein digestion and absorption (including amino acids, peptides, and phenols), carbohydrate metabolism (organic acids and derivatives), lipids metabolism (VFA, linoleic acid, and α-linolenic acid), vitamin metabolism (B1, B2 and B6), purine metabolism, and pyrimidine metabolism. Generally, most of the carbohydrates, fatty acids, amino acids, dipeptide, pyrimidine, and purine were higher for the GAS group than for the control group.
Different metabolites were used in hierarchical clustering with Cluster 3.0 software. Rumen metabolites from the GAS group clustered separately from the CON group (Fig. 4A). As shown in Table S2, 22 metabolic pathways were generated in the GAS group compared with in the CON group. The signi cantly different metabolites between the two groups were mainly involved in pyrimidine metabolism, purine metabolism, vitamin B6 metabolism, thiamine metabolism, ribo avin metabolism, and alpha-linolenic metabolism (Fig. 4B).

Discussion
Recent ndings showed that the garlic (for example garlic powder), garlic extracts (for example garlic oil) and garlic by-products (for example garlic husk and garlic leaf) had potential to modify rumen fermentation, improve animal performance, and might be an alternative for growth promoting feed antibiotics [27][28][29].In the present study, an increase of ADG was observed in the GAS group compared with that in the control group, which is consistent with previous studies which reported that ADG was increased by garlic powder and garlic leaf supplementing in growing lambs [29]. Ma et al. (2016) reported that supplementation with garlic extraction improved nutrient utilization by stimulating cellulolytic bacterial activity in the rumen [30]. Bioactive components present in garlic skin are polyphenolic in nature [31] and induce a positive effect on energy metabolism [32]. Therefore, improved growth rates in the GAS group could be attributed to lower energy loss, and higher nutrient utilization.
The average pH value was not affected by garlic skin in the present study. The results are consistent with other studies where rumen pH values did not differ on addition of garlic components [30], garlic oil [33],or garlic leaf [34].Supplementary garlic skin increased the ruminal concentration of total VFA, but decreased that of ammonia, which is similar to the results reported by Busquet et al. (2005) [35]and Klevenhusen et al (2011) [36]who supplemented various garlic components in vitro and in growing sheep diet. The increased total VFA concentrations with the garlic skin supplementation indicated stimulated rumen microbial fermentation activities and help support the improved ADG as was observed in the present study. The increased population of R. avefaciens with garlic extracts supplementation was observed by Ma et al. (2016) [30]. Increased concentrations of VFA and acetate production in our study could be attributed to an increased ber-digesting bacteria population. The ratio of acetate to propionate was not affected by garlic skin suggested that rumen fermentation pattern was unchanged. The increased of branched-chain volatile fatty acids, for example 2,2-dimethyl succinic acid, 3,3-dimethylglutaric acid, and 2-methylglutaric acid might be due to the fact that more branched-chain amino acids were degraded as less energy (metabolizable energy: 9.74 MJ/Kg vs 9.83 MJ/Kg) and carbon skeleton (crude protein:14.49% vs 15.10%) were available in the GAS diets than that in the CON.
The taxonomic assignment of ruminal bacteria in our study revealed that the most abundant phyla were Bacteroidetes and Firmicutes, which was in accordance with the results of previous studies [37,38].The Firmicutes phylum are the main bacteria that degrade bers, including a large number of bacteria which can promote decomposition of cellulose and fermentation of polysaccharides [39]. It was reported that Bulleidia has the capability to utilize saccharides and starch as energy source [40]. It has been reported that Prevotella accounting for 60-70% of rumen microorganisms [41], and several Prevotella were known to degrade oligosaccharides and hemicellulose [42]. It was also reported that Prevotella has the ability to utilize starch, xylan and pectin [43].Thus, the greater relative abundance of the phylum Firmicutes, the genus Bulleidia and Prevotella in the GAS group indicated that garlic skin supplementation effectively improved carbohydrate metabolism, especially the cellulose digestion. In addition, allicin is one of the active components of garlic [44]. Ma et al. (2016) found that supplementing a basal diet with allicin effectively increased the apparent digestibility of neutral detergent ber and acid detergent ber in growing lambs [30].
Methanosphaera, a methanogen capable to utilize methylated substrates like the Methanomassiliicoccaceae, increased in the GAS group. This agrees with Martinez-Fernandez et al. (2015) who observed that the abundances of Methanosphaera were increased in anti-methanogenic organosulphur supplementation lambs [45]. Saro et al. (2018) reported that greater abundance of Methanosphaera and lower abundance of Methanomassiliicoccaceae in the rumen of low methaneemitting lambs treated with a combination of garlic essential oil and linseed oil, which partially agrees with our results [46]. It has also been reported that the genus Methanosphaera was H2 utilizing species within the rumen [47]. The genus Fretibacterium was reported to participate in hydrogenation of longchain fatty acid [38]. Thus, the decreased relative abundance of Fretibacterium in the GAS group could be partially attributed to the increased abundance of Methanosphaera.
Rumen is an important organ where microbes ferment the plant biomass into proteins, VFAs, and other nutrients that can be used by the host. In total, 139 metabolites were changed as a result of garlic skin supplementation, 93 out of 139 signi cantly different metabolites had higher concentration including amino acids, dipeptides, fatty acids, and carbohydrates in GAS-fed lambs compared with CON-fed lambs. These amino acids and dipeptides can be used as precursors for the synthesis of microbial protein, while carbohydrates and fatty acids can provide energy for the synthesis of microbial protein [48].Additionally, our study also demonstrated pyrimidine and purine metabolism pathways were changed in GAS-fed lambs. Purine and pyrimidine are basic components of DNA and RNA. The levels of uridine, deoxycytidine, deoxyuridine, thymidine, and uracil in the pyrimidine metabolism as well as xanthine, deoxyadenosine, deoxyinosine, deoxyguanosine, and guanosine in the pyrimidine metabolism were all signi cantly higher for the GAS group than for the control group. In the rumen, xanthine and hypoxanthine are the nucleic acid degradation products of rumen microbial nucleic acids; thus, xanthine and hypoxanthine have been used as biomarkers of microbial protein synthesis [49]. Our ndings suggest that lambs may obtain more nutrients from GAS than from CON to produce more microbial protein.
Pyridoxal is the main active form of vitamin B6, and 4-pyridoxic acid is the major urinary catabolite of vitamin B6 [50].In our study, garlic skin supplementation increased the levels of pyridoxal, 4-pyridoxic acid, and thiamine, which indicated the upregulation of thiamine (vitamin B1) and vitamin B6 metabolisms. By serving as coenzymes, including pyruvate dehydrogenase and α-ketoneglutaric acid dehydrogenase, thiamine plays a critical role in carbohydrate metabolism [51].Besides, thiamine could participate in the production of acetyl-CoA via pyruvate-ferredoxin oxidoreductase in Megasphaera [52]. A positive correlation between thiamine Megasphaera was observed in our study, and the higher concentration of ruminal acetate in GAS-fed lambs might contribute from the higher thiamine and Megasphaera. Vitamin B6 were reported to involve in amino acid, glucose and lipid metabolism [53].It is well known that B vitamins can be synthesized by the ruminal microbiota, which is the main source for ruminants [48].Thus, in our study, the up-regulated levels of B vitamins and their metabolites indicated that the utilization of carbohydrates and proteins might be enhanced.
Poly-unsaturated fatty acids are thought to be more toxic than saturated fatty acid, therefore, biohydrogenation of poly-unsaturated fatty acid into fatty acid is one of the important microbial processes in the rumen [54].Biohydrogenation process provide a variety of biohydrogenation intermediates including vaccenic acid, conjugated linoleic acid, and conjugated linolenic acid [55].In our study, garlic skin supplementation decreased the level of alpha-linolenic acid and trans-vaccenic acid, which indicated the downregulation of alpha-linolenic acid metabolism. Spearman's correlation analysis revealed that Howardella was negatively correlated with alpha-linolenic acid, indicating that garlic skin probably affected alpha-linolenic acid metabolism through enriching the abundance of Howardella, which was consistent with previous study reported that higher abundances of Howardella in the gut inversely contributed to the enrichment of alpha-linolenic acid metabolism [56].

Conclusion
In conclusion, garlic skin supplementation improved the rumen fermentation, increased the average daily gain. We further investigated the effects of garlic skin supplementation on the rumen microbiome structure by 16S compositional sequencing and on the ruminal metabolome using untargeted metabolomics approach. Garlic skin supplementation did not signi cantly affect the α-diversity indices. Increased abundances of Prevotella, Bulleidia, Howardella, Methanosphaera but a decreased abundance of Fretibacterium were observed in GAS-fed sheep. Besides, the garlic skin supplementation favorably regulated pyrimidine metabolism, purine metabolism, vitamin B6 and B1 metabolism. These ndings suggested that garlic skin supplementation not only changed the community of the rumen microbiome but also altered ruminal uid metabolism in fattening sheep.

Declarations
Ethics approval and consent to participate The procedures of this study were approved by the Animal Care and Use Committee of Anhui Agricultural University (Hefei, China) and were in accordance with the university's guidelines for animal research (SYXK(Wan)2016-007).

Consent for publication
Not applicable.

Availability of data and materials
Raw unprocessed sequence data in this study are available from the National Centre for Biotechnology Information Sequence Read Archive (SRA) database under accession number: PRJNA648813.

Competing interests
The authors declare no con ict of interest.