Galactooligosaccharides and Xylo-oligosaccharides Altering the Caecal Microbiome, Metabolome, and Transcriptome of Chickens Revealed by a Multi-Omics Analysis

Backgroud Studies have shown that prebiotics could affect meat quality, but the underlying mechanism are poorly understood. This study aimed to investigate whether prebiotics affect chicken’s meat quality through gut microbiome and metabolome. Methods The gut content were collected from chickens fed with or without prebiotics (galactooligosaccharides or xylo-oligosaccharides) and subjected to microbiome and metabolome analyses, and chicken breast was performed transcriptome sequencing. Results The prebiotics altered proportions of microbiota in gut contents at different levels, especially microbiota in the phylum of Bacteroidetes and Firmicutes, such as genus of Alistipes, Bacteroides, and Faecalibacterium. The prebiotics also altered contents of caecal metabolites such as lysophosphatidylcholine (lysoPC), intramuscular fat and avor compound (Benzaldehyde and myristic acid). Differentially expressed genes (DEGs) induced by prebiotics were signicantly involved in regulation of lipolysis inadipocytes and adipocytokine signaling pathway. Changes in gut microbiota and metabolites were remarkably correlated such as Bacteroidetes and Firmicutes was respectively positively and negatively correlated with lysoPC. DEGs were also interacted with caecal metabolites. Conclusion These ndings integrated and incorporated link among gut microbiota, metabolites and transcriptome, which proposed prebiotics may affect meat quality and avor of chickens. acid pathway. The perturbations of D-Glutamine and D-glutamate metabolism, Glutathione metabolism and Glycerophospholipid metabolism (the enriched pathway of LysoPC -like metabolites) were discovered in response to GOS treatment. These results suggested that prebiotics feeding may induce changes in intestinal microbial LysoPC and shikimic acid metabolism in broilers chicken.


Growth monitoring and Sample collection
The mortality and elimination rate of chickens were recorded daily, and the body weight for each group were measured once a week. The chickens were randomly selected for euthanasia. Growth performance (e.g. the breast muscle percentage, thigh muscle percentage, abdominal fat percentage and subcutaneous fat thickness) were detected and compared. The chicken breast and contents of large intestine, cecum, colon and rectum in selected broiler chickens from each group were collected, and frozen immediately at 80°C until used. Sequences were processed using the software package of the QIIME toolkit. Subsequently, operational taxonomic units (OTU) were picked at 97% similarity using Vsearch software. Alpha diversity index was calculated using QIIME. Beta diversity was performed on QIIME (v. 1.8.0) for Principal co-ordinates analysis (PCoA) on Bray-Curtis distance matrices. We performed linear discriminant analysis (LDA) coupled with effect size measurement (LefSe) analysis online (http://huttenhower.sph. harvard.edu/galaxy/). Statistical analyses of the microbial relative abundances were conducted using the Kruskal-Wallis H test.

Metabolic pro ling of caecal contents
Caecal contents from 30 chickens (n=10 per group) was utilized to extract metabolites for LC-MS analysis. Brie y, 50 mg of the sample was transferred into 2 ml centrifuge tube followed by the addition of a grinding ball 6 mm in diameter. A total of 400 µl methanol (methanol: water = 4:1 (v: v)) containing 0.02 mg/mL L-2-chlorophenylalanine as internal standards was added into the sample. The mixture was ground for 6 min under a condition of -10°C and 50 Hz, and sonicated for 30 min at -5°C and 40 KHz, and kept at -20°C for 30 min, and then centrifuged at 13000g at 4°C for 15 min. The supernatant was transferred into a liquid chromatography/ mass spectrometry (LC/MS) vial, and additional 20 µl was used as the quality control samples for analysis. Chromatographic separation was performed in an ultra-performance liquid chromatography tandem fourier transform mass spectrometry uHPLC-

Transcriptome sequencing analysis
Total RNA from chicken breast (n=4 per group) was isolated using TRIzol reagent (Invitrogen, 15596-018). After removing genomic DNA contaminations and assessing RNA integrity, the quality and quantity of RNA were evaluated using an Agilent 2100 bioanalyzer with the RNA 6000 Nano Chip (Agilent Technologies) and Agilent 2100 Bioanalyzer (Agilent Technologies, CA, USA). The sequencing library was generated from the NEB Next UltraTM RNA Library prep Kit for Illumina (NEB, USA) according to the instruction manual. The twelve cDNA libraries that were obtained were sequenced on Hi-Seq platform (Illumina, USA) and generated paired-end reads.
The Fast-QC (http://www.bioinformatics.babraham.ac.uk/projects/fastqc/) was used to analyze the raw data to lter away the low-quality sequences and adaptor sequences. Then, we matched the high quality clean reads obtained to the Ensembl Gallus_gallus-5.0 (Gallus_gallus. Gallus_gallus-5.0.dna.toplevel.fa) database. The level of gene expression in chicken breast was calculated by FPKM, the differentially expressed genes (DEGs) with a P value < 0.05 and | log2 (fold change) |>1 was determined by the DE-Seq2 algorithm. Function analysis of DEGs were performed using Gene ontology (GO, http:\\www.geneontology.org\), and pathway analysis were assessed by kyoto encyclopedia of genes and genomes (KEGG, http:\\www.genome.jp\kegg\ analyses).

Statistical analyses
All data were analyzed with Graph Pad Prism 8 software. Growth performance parameter were presented as mean±standard deviation and analyzed through t test. All results were considered statistically signi cant at P <0.05.

Prebiotic supplementation slightly improves growth performance of broiler chickens
As depicted in Table 2, the GOS had a growth promoting effect on the average body weight of chickens, while there were no signi cant differences between XOS treatment group and control group (P>0.05). No statistical differences in performance metric of feed conversion ratio. Additionally, the breast muscle percentage and thigh muscle percentage were slightly increased in prebiotics group relative to the control group (P>0.05). The abdominal fat percentage and subcutaneous fat thickness (cm) showed slight decrease in prebiotic treatment group (P >0.05). Feed conversion ratio of the prebiotic treatment group was lower that of control group (P>0.05). Thus, considering the growth performance, the prebiotic group was slightly superior to the control group. Notes: There are no signi cant difference between the prebiotics group and the control group.

Prebiotic intervention alters microbiome composition
In our present microbiome investigation, the caecal microbial communities were compared between prebiotics (GOS and XOS) treatment group and control group. A total number of 1513565 valid reads and 795 OTUs at 97% sequence similarity were obtained from all samples, respectively. The rank abundance analysis on the OUT level revealed that the richness and evenness of the microbiota composition was the similar among three groups ( Figure 1A). The extent of the similarity of gut microbial communities between the three groups was measured PCoA at the OTU level, the results showed that the gut microbial communities was signi cantly separated between GOS, XOS and control group ( Figure 1B). Moreover, we found that the diversity of the microbial community signi cantly decreased in the prebiotics treatment group, especially in XOS group, as showed by abundance-based coverage estimators (ACE) and Chao1 indices ( Figure 1C and 1D).
Then, we displayed the composition of caecal microbiota at family and genus levels. Microbiota of the 30 samples from the prebiotics treatment group and control group were altered at family level. As shown in Figure 1E, the abundance of Ruminococcaceae, Barnesiellaceae, and Acidaminococcaceae were increased in the prebiotics treatment group. Unlike these bacteria, the prebiotics treatment group had a lower average relative abundance of Bacteroidaceae and Lactobacillaceae compared with control ( Figure  1E). At the Top10 genus level, the increased proportions of Alistipes, Faecalibacterium, unclassi ed_f_Barnesiellaceae, and Phascolarctobacterium, decreased proportions of Bacteroides and Lactobacillus were observed in prebiotics treatment group in comparison to control ( Figure 1F).
Consistent with above results, the circos plot at genus level revealed that Alistipes was the dominant ora in GOS group, accounting for 19.86% (Figure 2A). Faecalibacterium and Bacteroides were the dominant ora in XOS group, accounting for 18.15% and 17.73%, respectively ( Figure 2A). Bacteroides was the only dominant ora in control group, accounting for about 20.82% ( Figure 2A). Importantly, Alistipes and Bacteroides are a genus in the phylum Bacteroidetes, and Faecalibacterium is a genus in the phylum Firmicutes, indicating that prebiotics treatment may affect the quality of chicken mainly by changing the composition of Bacteroidetes and Firmicutes. Subsequently, LEfSe analysis revealed speci c bacteria that were associated with prebiotics ( Figure 2B). Several microbiota including f_ Porphyromonadaceae_g_Barnesiella_OTU436, f_Bacteroidaceae_g_Bacteroides_ OTU370, and s_Bacteroides_ caecigallinarum were all signi cantly over-represented (all LDA scores (log10) > 3) in the feces of chickens fed with XOS. Interestingly, the Porphyromonadaceae, Bacteroidaceae, and Bacteroides are belong to the phylum Bacteroidetes. The c_Clostridia_o_Oscillospirales_OTU445, belong to Firmicutes phylum, was identi ed as speci c taxa in chickens treated with GOS (all LDA scores (log10) > 2). Therefore, these results again suggested that prebiotics treatment primarily altered the abundance of Bacteroidetes and Firmicutes.

Prebiotic changes global metabolome of caecum
Since signature of microbiota related with prebiotics in the chickens was demonstrated, we speculated that alterations in metabolic pathways may be at least partially in uenced by prebiotic-driven gut microbiota in chickens. Thus, we subsequently analyzed intestinal metabolites in caecal feces of chickens, using a nontargeted LC-MS technology. As shown in Fig. S1A-1D, the score plots of PLS-DA con rmed that the metabolic pro les had signi cant differences between prebiotics treatment group and control. Permutation test for the OPLS-DA model to further visualized group separation that XOS group and control group generated intercepts of R2 = 0.9848 and Q2 = -0.1713 (Fig. S1E), and that of GOS group and E group produced intercepts of R2 = 0.9851 and Q2 = -0.1205 (Fig. S1F), which revealed OPLS-DA had well tted-effect.
Subsequently, we summarized the distribution of differential metabolites separating the two groups. Overall, 135 differential metabolites were identi ed in comparison of prebiotics treatment group relative to control (VIP > 1, P < 0.05). There were 79 and 92 differential metabolites respectively found in XOS vs control and GOS vs control, respectively (Table S1 and Table S2). A heat map was utilized to visualize the abundance of the differential metabolites ( Figure 3A and 3B). In total, two clusters were respectively generated in XOS vs control and GOS vs control. Among these, 36 metabolites were shared (Table S3) Moreover, we identi ed several altering metabolic pathways that speci c related with prebiotics treatment. As depicted in Figure 3C and 3D, the Phenylalanine, tyrosine and tryptophan biosynthesis, Purine metabolism, and Glutathione metabolism were observed to be strikingly disturbed in response to XOS. Interestingly, the biosynthesis of these amino acids depend on shikimic acid pathway. The perturbations of D-Glutamine and D-glutamate metabolism, Glutathione metabolism and Glycerophospholipid metabolism (the enriched pathway of LysoPC -like metabolites) were discovered in response to GOS treatment. These results suggested that prebiotics feeding may induce changes in intestinal microbial LysoPC and shikimic acid metabolism in broilers chicken.

Alteration of meat metabolites
Subsequently, we further detected the concentration of common nutrients, including the proteins, amino acids, fatty acids in the chicken breast. As presented in Table S4, overall, prebiotics treatment signi cantly decreased the content of histidine compared to control samples but had no signi cant effect on other amino acids in chicken breast. Although administration of prebiotics signi cantly decreased the proportion of some fatty acids, such as myristic acid (C14) and palmitoleic acid (C16), the proportion of IMF remarkably increased in prebiotics group. Notably, although concentration of these metabolites such as oleic acid and polyunsaturated fatty acids (linoleic acid and arachidonic acid) had no signi cant alteration, that of them had a slight increase in chicken breast and viscera samples from XOS or GOS treatment group relative to control group. Accumulating studies have con rmed that juiciness, avor, and tenderness correlate positively with the muscle total fat content [22,23]. Polyunsaturated fatty acids are heated and oxidized to generate volatile components, such as 2, 4-decaldehyde, which improve the avor of meat [24]. Together, the addition of prebiotics had an effect on meat quality and avor of chicken.
3.5. Pro ling of genes expressed in the chicken breast of the prebiotics treatment group and control group We further investigated the metabolite gene response. In total, 599 genes that were signi cantly altered by transcriptome sequencing identi ed. There were 381 and 298 DEGs genes respectively obtained in XOS vs control and GOS vs control (Table S5). Among these, 80 DEGs were shared, including 36 upregulated DEGs and 42 down-regulated DEGs (data not shown). We further analyzed the expression tendency of these common genes, the trend for 10 genes of subcluster_5, such as LRRTM3, LRRC10B, and ENSGALG00000046353 was similar; these genes all highly expressed in prebiotics group relative to control group. The trend for 4 genes of subcluster_9, such as CFAP44 and CCL21 presented same result as that for subcluster_5 ( Figure 4A). Subsequently, corresponding functions of the 80 common genes were determined ( Figure 4B), which were mainly found to function as biological process in the "regulation of lipid kinase activity", "regulation of lipid metabolic process", "regulation of phospholipid metabolic process", and "regulation of hormone biosynthetic process. KEGG annotation analysis showed these genes were mainly involved in organismal systems of "endocrine system", "immune system", "immune system", "digestive system", and various metabolism (e.g. nucleotide, carbohydrate, biosynthesis of other secondary, amino acid metabolism) ( Figure 5A). Subsequently, KEGG enrichment analysis delineated 80 genes primarily implicated in metabolism pathway of "regulationof lipolysis inadipocytes", "adipocytokine signaling pathway", "protein digestion and absorption", "cytokine-cytokine receptor interaction" and "Jak-STAT signaling pathway" ( Figure 5B). These pathways were related with lipid and protein metabolism and may affect meat quality and avor.

Caecal microbiota correlated to growth performance of broiler chickens
To investigate the correlation between caecal microbiota alteration and growth performance, we analyzed the correlation between species abundance (Bacteroidetes and Firmicutes, and other taxa at family level mentioned above) and growth performance by Pearson correlation coe cient analysis. At the phylum level, only Bacteroidetes showed signi cantly positively correlation with average body weight (Figure 6). At the family level, caecal microbiota of Barnesiellaceae and Ruminococcaceae were signi cantly positively correlation with breast muscle percentage and average body weight, respectively ( Figure 6). No signi cant correlation was detected between other caecal microbiota and growth performance.

Correlations between the prebiotics-induced gut microbiome and metabolome
A pearman correlation analysis of differential metabolites and top40 OTUs in abundance (microbes) was performed. As depicted in Figure 7A, based on differential metabolites in XOS group, the f_Acidaminococcaceae_g_Phascolarctobacterium_OTU754 was signi cantly positively related to retapamulin, N-Methyl-14-O-demethylepiporphyroxine, mandelonitrile rutinoside, N-Arachidonoyl tyrosine, benzaldehyde and L-Tryptophan.

Metabolite-gene network analysis
To extract the interactions among the all differential metabolites and gene expression in prebioticssupplemented chickens, a network diagram was constructed. Among the differential metabolites, a total of twelve metabolites interacted with differentially expressed genes ( Figure 7B). Particularly, the AMD1(Sadenosylmethionine decarboxylase)-cadaverine or omithine pairs, and PLA2G1B (phospholipase A2 group 1B) -indoleacetic acid pairs were discovered in the network. These genes were likely hubs of the prebiotics treatment on meat quality of chickens.

Discussion
Growing evidence shows that novel additives such as prebiotics or probiotics can regulate composition and/or activity of the gut microbiota, thus attributing a bene cial physiological effect on the host [25,26]. Bene cial effects of prebiotics (XOS and GOS) on performance and gut microbiota have already been demonstrated in chickens [10,27,28]. However, whether the gut microbial community, metabolites, and transcriptomic changes of chicken breast were correlated in chicken fed with prebiotics is not unknown. In the study, we proved the clue that prebiotics improved meat quality and avor in broiler through modulating gut microbiota, metabolites and chicken breast transcriptome.
Selective fermentation of some prebiotics has been demonstrated to cause alteration in the composition and/or activity of the gastrointestinal microbiota, contributing to the host health [29]. A study has reported dietary xylooligosaccharides prebiotics can improve growth performance, enhance endocrine metabolism and immune function of broilers [30]. Additionally, changes in enteric bacteria in the cecum [31] and improved intestinal morphology have been found in chickens given with dietary mannanoligosaccharide [32]. In the present study, supplementation with either GOS or XOS improved the growth performance of chickens, and increased the relative abundance of dominant and bene cial bacteria (Alistipes, Faecalibacterium, Phascolarctobacterium), while decreased that of potential bacterial pathogens (Rikenellaceae_RC9_gut_group). It can be also assumed that the improvement in growth performance is associated with the bene cial effect of prebiotics on the intestinal microbiota.
LysoPC, a derivative of PC hydrolyzed by phospholipase A2, is a highly abundant bioactive lipid mediator that exist in circulation [33]. Different contens of PC determine whether the vesicles release cholesterol, so the changes in LysoPC are a dynamic re ection of cholesterol [33]. In this study, prebiotic treatment resulted in signi cant changes in the content of 7 types of PC, suggesting that prebiotic feeding disrupted the original cholesterol metabolism of broiler chicken. Furthermore, we found that caecal micro ora Bacteroidetes and Firmicutes are related to the metabolism of LysoPC. Consistent with our results, Bai et al. reported that the in ammation-related metabolites of LysoPC(16:0) was signi cantly correlated with genera that belonged to phyla Firmicutes [34]. The results of Gao et al. reveal that LysoPC(20:0) has a similar correlation trend with us, LysoPC(20:0) was respectively negatively correlated with Firmicutes and positively correlated with Bacteroidetes [35]. Therefore, we speculated that prebiotic feeding stimulated the changes of Firmicutes and Bacteroidetes, which further changed the content of LysoPC and affected lipid accumulation in broiler chicken.
The saturated fatty acids and omega (ω)-3 polyunsaturated fatty acids (PUFAs) like linoleic acid and αlinolenic acid, and IMF were also associated with the umami avor [36]. Similarly, the concentration of avoring fatty acids like myristic acid and IMF were signi cantly higher in chickens fed with prebiotics.
The concentration of PUFAs has a slight increase in prebiotics group. The fatty acids play an important role in determining the meat quality and avor of chickens. For example, the myristic acid (C14:0) contribute to avor and is used as a avoring agent in food items [37]. The benzaldehyde is known to arise from linolenic acid (C18:3n-6), and considered as the characteristic avor components [38]. Here, the higher concentration of benzaldehyde was found in prebiotics group, and it had signi cantly positive and negative correlation with g_Phascolarctobacterium_OTU754 and g_Lactobacillus_OTU4, respectively.
Consistently, the abundance of Phascolarctobacterium and Lactobacillus respectively increased and decreased in prebiotics group in comparison to control. Other non-volatile avor precursor components including amino acids also affect the meat quality of meat [39]. The monosodium glutamate-like amino acids, including glutamic and aspartic acid were ingredients for umami taste [40]. Here, the L-Glutamate was found to be dramatically positively related with f_Barnesiellaceae_g_unclassi ed_f_OTU716; the proportion of Barnesiellaceae increased in prebiotics group. Additionally, other differential metabolites (e.g. retapamulin, N-Methyl-14-O-demethylepiporphyroxine, LysoPC, et al) had also correlation with microbial taxa (e.g. g_Alistipes_OTU714 and OTU326, g_CHKCI001_OTU416, g_Barnesiella_OTU436, et al). Therefore, the microbiota -metabolites axis was regulated by prebiotics may affect meat quality and avor of chickens.
The fatty acids are oxidized and used as energy source, or stored and deposited in adipose tissues [41]. The dietary fatty acids exert regulator role in gene expression and nally control enzyme activity [42].
Therefore, the lipid uptake, transport, storage and biosynthesis are complex steps to regulate balance of lipid metabolism. We found the perturbations of lipid metabolism related pathway such as D-Glutamine and D-glutamate metabolism, Glutathione metabolism and Glycerophospholipid metabolism were discovered in response to prebiotics treatment. To further understand the potential mechanism of lipid metabolism, the GO was analyzed. Many DEGs in chicken breast from prebiotics group and control group were mainly involved in biological process of the "regulation of lipid kinase activity" and "regulation of lipid. To further study the links between gut metabolites and gene expression, a network diagram was constructed. The twelve metabolites interacted with many genes (e.g. AMD1 and PLA2G1B).
Accumulating studies have showed AMD1 is associated with sustaining polyamine metabolism in prostate cancer [43,44]. PLA2G1B, as a secreted phospholipase, is reported to mediate lipid absorption [45]. However, the links of these genes with metabolism were further con rmed.

Conclusion
The bene cial effect of prebiotics on gut microbiome was determined. The prebiotics affected

Consent for publication
Not applicable Availability of data and materials All data generated or analysed during this study are included in this published article and its supplementary information les.

Competing interests
The authors declare no con ict of interest. Relative gut microbiota abundance at the family and genus level in fecal samples from the control and prebiotics groups (GOS group and XOS group). GOS group indicated chickens was fed galactooligosaccharides (GOS); XOS group presented chickens was fed xylo-oligosaccharides (XOS).

Figure 2
Page 22/27 The relationship and abundance of ora in three groups. (A) The relationship between ora and samples and visualized as a circos plot. (B) The differences in abundance among the three groups analyzed by LEfSe.    Caecal microbiota correlated to growth performance of broiler chickens.