Weaning Stress Perturbs Gut Microbiome and Its Metabolic Profile in Piglets.

Weaned piglets are vulnerable to nutritional, physiological, and psychological stressors, leading to abrupt taxonomic and functional shifts in the intestinal microbiome. In this study, an integrated approach combination of 16S rDNA gene sequencing and the mass spectrometry-based metabolomics techniques was used to investigate the effects of weaning stress on intestinal microbial composition and its metabolic profiles of piglets. Three litters of suckling piglets with same parity were chosen. The samples of colonic contents were collected from each selected piglets (weaned day, 3 days after weaned) for microbial and metabolomics analysis. The results showed that Lachnospiraceae, Negativicutes, Selenomonadales, Campylobacterales and other 15 species increased after weaning, while Porphyromonadaceace, Alloprevotella, Barnesiella and Oscillibacter decreased. Based on the function profiles prediction and metabolomic analysis, five key metabolic pathways including Phenylalanine metabolism, Citrate cycle (TCA cycle), Glycolysis or Gluconeogenesis, Propanoate metabolism, Nicotinate and nicotinamide metabolism might be the relevant pathways involved in weaning stress-induced gut microbiota dysbiosis. Taken together, these results indicated that weaning stress not only changed microbial composition and function but altered the microbial metabolic profiles in the intestine, which might provide a new insight in alleviating weaning stress and facilitating disease prevention during the period of weaning in piglets.


Predicted Function of Gut Microbiota in Nursing and Weaned Piglets. PICRUSt analysis was
performed to predict the potential functions of gut microbiota. In comparison two groups, nursing piglets had higher enrichment of the pathways in Lipid biosynthesis proteins, Energy metabolism, Transcription machinery, Nicotinate and nicotinamide metabolism, Carbon fixation pathways in prokaryote, Propanoate metabolism, naphthalene degradation, Streptomycin biosynthesis and peroxisome. While other pathways such as Photosythesis proteins, Thiamine metabolism, Sulfur relay system were overrepresented in weaned piglets (Fig. 4).

Metabolic Differences of Gut Microbiota between Nursing and Weaned Piglets.
Untargeted metabolomics analysis was performed to evaluate the metabolic differences of gut microbiota in two groups. An OPLS-DA method was performed to better understand the different metabolic patterns. Figure 5A showed significantly separated clusters between two groups. The values for R 2 X and Q 2 and the results of permutation tests indicated that the samples were of reasonable quality (Fig. 5B). In total, relative levels of 433 biomarker metabolites differed significantly between nursing and weaned piglets. There was a clear distribution in metabolite between two groups ( Supplementary Fig. 1). Figure 6 showed the variety of metabolic profiles in gut microbiome, with 239 increased and 194 decreased molecular features based on a VIP > 1 in 95% jack-knifed confidence intervals, respectively. And a total of 83 biomarker metabolites were filtered according to the standard with similarity >600, VIP > 1 and p < 0.05. These metabolites, including lipids, amino acids, carbohydrates, benzene, organic acids, amines and others, involved in multiple biochemical processes in the colon ( Table 2).
Key metabolic pathways analysis. The commercial databases including KEGG (http://www.genome.jp/ kegg/) and MetaboAnalyst (http://www.metaboanalyst.ca/) was applied to search for the pathways of metabolites. The result showed that beta-Alanine metabolism, Arginine and proline metabolism, Alanine, aspartate and glutamate metabolism, Nitrogen metabolism and other pathways were altered (Table 3). Among them, five metabolic pathways including Phenylalanine metabolism, Citrate cycle (TCA cycle), Glycolysis or gluconeogenesis, Propanoate metabolism and Nicotinate and nicotinamide metabolism were also found in the functional prediction of gut microbiota as shown in Fig. 4.

Discussion
Weaning is a sudden, complex and highly stressful event in piglets' life, accumulating evidence indicated that piglets have an abrupt taxonomic and functional shift in the intestinal microbiota after weaned 8,12 . However, the effects of weaning stress on microbial metabolites in piglets are still an unsolved issue. In this study, piglets at weaned day (nursing group) and 3 days after weaned (weaned group) were chose to investigate the effects of weaning stress on the composition of gut microbiota and its metabolites in the colon. The results showed that the diversity of gut microbiota in weaned piglets was similar with nursing piglets, while the richness decreased. These results were analogous with previous study, which found a continuously decreased alpha diversity until 11 days after weaned 16 . However, other studies presented opposite phenomenon. Those showed a continuous increase in   alpha diversity of gut microbiota during weaning transition 17,18 . The reason might be that our studies focused on the gut microbiota during the early period after weaning, whereas their studies investigated the gut microbiota during the period from weaning to adulthood in pigs. It seems that alpha diversity of gut bacterial community undergoes decreases first during the early period after weaning, and then increases from weaning to adulthood in pigs on the whole.
In line with previous studies, Bacteroidetes and Firmicutes were the two most dominant phyla in the intestine of piglets, regardless of weaning 16,19 . Proteobacteria and Fusobacteria were the followed abundant phylum. There was no difference in bacterial dominant phyla in the colon between nursing and weaned piglets. And Candidatus Saccharibacteria, Planctomycetes, Verrucomicrobia and Lentisphaerae even cannot be detected in the gut of weaned piglets, suggesting that these bacterial species may not adapt to the intestinal tract environment after weaning. What's more, the results showed that several bacterial species had a remarkable difference between nursing and weaned piglets. For example, Alloprevotella and Oscillospira were decreased, while Campylobacterales, Campylobacteraceae and Campylobacter were increased in weaned piglets. Alloprevotella mainly produce succinate and acetate, which could improve the gut barrier and exhibit anti-inflammatory function 20 . Oscillospira species are butyrate producers and can use host glycans as a source of energy 21 . Study found that the presence of Oscillospira is reduced in diseases that involve inflammation 22 . Campylobacterales was one of opportunistic pathogens, causing a life-threatening gastrointestinal disease 23 . And Campylobacteraceae and Campylobacter were also increased after weaning. The increase of this bacterial species in weaned piglets may be one of the major reasons of post-weaning diarrhea. These results indicated that unhealthy alterations in the composition of gut microbiota triggered by weaning stress.
Gut microbiota participate in many functional metabolic pathways through which nutrient digestion and absorption, lipid metabolism, and hormone biosynthes are influenced 24 . To investigate how the functional capacity of the gut microbiota changed after weaning in piglets, Functional metagenomics prediction approach was used to analyze the KEGG pathways compositions in microbial populations. The results showed that many metabolic pathways such as lipid biosynthesis proteins, energy metabolism and phenylalanine metabolism were changed during the period of weaning. Gut microbiota could also directly alter its metabolic capacity through microbial products, affecting intestinal function. Butyrate, a kind of metabolite, produced by gut microbiome could regulate energy metabolism and autophagy in the mammalian colon 25 . So GC-TOF/MS method was applied in the current study to quantitatively measure small molecular metabolites in biological samples and identify relevant Pathways. Based on PICRUSt and metabolomics analysis, we found that five metabolic pathways including Phenylalanine metabolism, Citrate cycle (TCA cycle), Glycolysis or Gluconeogenesis, Propanoate metabolism and Nicotinate and nicotinamide metabolism were the key relevant metabolic pathways involved in weaning stress-induced gut microbiota dysbiosis.
Previous study indicated that alterations in the microbiota composition elicited by nutrition are associated with neurotransmitter concentrations 26 . Nutrition can influence hormonal, neurotransmitter, and signalling pathways, which in turn modulate several brain processes impacting both appetite and mood 27 . Phenylethylamine, as a neuromodulator or neurotransmitter, participates in the phenylalanine metabolic pathway and plays an important role in mood regulation 28 . In the present study, the phenylethylamine level decreased in weaned piglets, which indicated piglets had a bad mood caused by stress. And a low level of phenylethylamine could not modulate the mood well, which may cause a decreased feed intake and weight gain of piglets. In addition, changed neurotransmitter noradrenaline and 4-aminobutyric acid level were also found in the present work. Gut microbiota could influence the immune system via neurotransmitters, and proinflammatory cytokines released through microbiota-induced immune response can in turn have a direct influence on the secretion of neurotransmitters such as norepinephrine 29 . Compelling studies have linked weaning to an upregulation of expression of inflammatory cytokines in the intestine of piglets 10,30 . Thus the regulatory mechanism and potential role of "microbiota-brain-gut" axis that play during weaning period in piglets are warranted to further investigation.
During a short period after weaning, small intestine length, microvilli height, crypt depth, and barrier function of piglets were changed. What's more, digestive enzyme activity and absorptive capacity of small intestine were reduced, resulting in a low absorption capacity of amino acids and carbohydrates 31 . In the present study, decreased essential amino acids such as valine and some carbohydrate in weaned piglets supported this view. TCA cycle is responsible for the oxidative degradation of sugars, fats, and amino acids and could link different types of nutrients together. Pyruvic acid, as an important intermediate involved in the Citrate cycle (TCA cycle), was decreased in the weaned piglets. The reduction in energy metabolism may be part of the gut adaptation process. Microorganisms may engage energy for adaptation mechanisms to ensure necessary physiological functions 32 .
Notably, pathways related to short chain fatty acid (SCFA) metabolism such as propanoate metabolism were enrichment in piglets. Functional metagenomics prediction analysis using PICRUSt revealed that genes involved in Propanoate metabolism were more abundant in nursing piglets. And the bacteria Oscillospira, known to be as a butyrate producer, decreased in the gut microbiota of weaning piglets, which might indicate the decreased SCFA production after weaning. SCFAs are major anions in the gut and could be absorbed rapidly by colonic epithelial cells. Growing evidence proved that these small molecules could protect the host against colonic diseases, improve the gut barrier function and exhibit anti-inflammatory effects 33,34 . In addition, SCFA-mediated beneficial metabolic effects on health can be mediated by induction of intestinal gluconeogenesis 35 . A recent study showed that SCFAs could also improve the feed efficiency of swine 36 . Thus the reduction in Propanoate metabolism of weaned piglets in the current study might be related to a low growth performance and unhealthy gut environment after weaning. Administration of niacin (nicotinic acid and nicotinamide) has been shown to beneficially effect the host-microbiome interaction in a mouse model 37 . Nicotinamide adenine dinucleotide (NAD+) is the central cofactor of metabolism, mediating ATP generation, reactive oxygen species (ROS) detoxification, biosynthetic processes, DNA repair, and nutritionally sensitive gene regulation 38 . In the present study, decreased nicotinic acid was found in weaning piglets, which indicated an adverse influence on piglets trigged by stress.

Conclusion
In conclusion, our current work demonstrated that weaning stress not only disturbed the gut microbiota but also altered the metabolome profiles in the colon. After weaning, some benefit bacterial species such as Alloprevotella and Oscillibacter decreased in weaned piglets, while some opportunistic pathogens such as Campylobacterales, Campylobacteraceae and Campylobacter increased. In addition, based on PICRUSt and metabolomics analysis, five key metabolic pathways including Phenylalanine metabolism, Citrate cycle (TCA cycle), Glycolysis or Gluconeogenesis, Propanoate metabolism, Nicotinate and nicotinamide metabolism were identified as the relevant pathways associate with weaning stress-induced gut microbiota dysbiosis. These results may provide a new insight in alleviating weaning stress of piglets and facilitating disease prevention during weaning by manipulating host-microbiome metabolites interactions. And future study might focus on the potential functions of neurotransmitters in gut microbiome, neurogenesis, behavior and immune of piglets suffering weaning stress.

Materials and Methods
Animals and Sample Collection. All      Sequencing data analysis. The paired-end reads were obtained 2 × 250 bp using MiSeq platform. In order to obtain high quality sequences, tags with length of <220 nt, average quality score of <20, and tags containing >3 ambiguous bases were removed by PANDAseq. Data was de-chimeric and clustered using Usearch software, a standard clustering of 97% similarity was applied to obtain OTU (Operational Taxonomic Units). And the taxonomy-based analysis to the OTUs was performed by RDP algorithm using GreenGene database (http:// greengenes.lbl.gov). QIIME was used for analyzing alpha diversity, which included calculation of the observed species, Chao 1, Shannon, and Simpson indices. And principal coordinate analysis (PCoA), Anosim Analysis, and unweighted pair group method with arithmetic mean (UPGMA) were performed using QIIME for beta-diversity analysis. Linear discriminant analysis coupled with effect size (LEfSe) was applied to identify the bacterial taxa differentially represented between groups at genus or higher taxonomy levels 40 .
Metagenome prediction based on 16S rDNA gene data. Phylogenetic Investigation of Communities by Reconstruction of Unobserved States (PICRUSt), which makes up for the shortage of 16S rDNA gene studies, was used for predicting the gene family abundances of bacterial populations based on the precalculated GreenGenes (v13.5) database. In brief, the analysis process is divided into two steps: gene content inference and metagenome inference. The software STAMP was conducted to detect the differentially abundant Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways in two groups with false discovery rate correction. The rank sum test was used to identify species with significant differences between groups. Two-sided Welch's t-test and Benjamini-Hochberg FDR correction were performed in two-group functional prediction analysis. P < 0.05 was considered statistically significant. And the prediction accuracy of PICRUSt was evaluated by the Nearest Sequenced Taxon Index (NSTI), with lower value indicating a higher accuracy of prediction.  GC-TOF/MS analysis. The derivatized samples were analyzed using an Agilent 7890 GC system equipped with a Pegasus 4D TOFMS (LECO, St. Joseph, MI) with a DB-5MS capillary column (30 m × 250 μm inner diameter, 0.25 μm film thickness coated with 95% dimethylpolysiloxane cross-linked with 5% diphenyl under the following conditions: initial temperature was kept at 80 °C for 0.2 min, increased to 180 °C at a rate of 10 °C/min, to 240 °C at a rate of 5 °C/min, and further to 290 °C at a rate of 20 °C/min; the column was then maintained for 11 min. One μL of sample solution was injected with helium as the carrier gas at a flow rate of 1 mL/minute. The temperatures of transfer line, and ion source were 245 °C, and 220 °C, respectively. The MS data were acquired with a mass-to-charge ratio (m/z) range of 20−600 in a full-scan mode.
GC-TOF/MS data acquisition and processing. The raw peaks extraction, data baselines filtering and calibration, peak alignment, deconvolution analysis, peak identification, and integration of the peak area were operated using the Chroma TOF4.3X software (LECO) and LECO-Fiehn Rtx5 database. The peak identification was tested using the RI (retention time index) method. The missing values of the original data were filled by half of the minimum value via a numerical simulation method. Noise removal was performed based on an interquartile range to filter data, then, data were normalized by area normalization methods. The SIMCA14.1 software package (Umetrics, Umea, Sweden) was used for multivariate variable pattern recognition analysis: principal component analysis (PCA), partial least squares discriminant analysis (PLS-DA), and orthogonal partial least-squares discriminant analysis (OPLS-DA). PCA was used to show the internal structure of the data and display the similarity and difference. OPLS-DA was applied to obtain a higher level of group separation and better explain the variables. To evaluate the predictive ability and fitting level of the model, the parameters R 2 Y and Q 2 were applied. The metabolites responsible for differentiating two groups were filtered with the following requirements: variable importance in the projection (VIP) >1 and P-values of 0.05 (threshold) with 95% Hotelling's T-squared ellipse. Data was analyzed by SPSS 20.0 using Student's t-test to profile metabolite differences between two groups. The commercial databases including KEGG (http://www.genome.jp/kegg/) and MetaboAnalyst (http://www. metaboanalyst.ca/) was utilized to search for the pathways of metabolites. We can further screen pathways and find the key pathways with the highest correlation with metabolites through comprehensive analysis of the pathways where the differential metabolites are located (including enrichment analysis and topological analysis).

Data Availability
All data generated or analysed during this study are included in this article and its Supplementary Information files.