Colonic luminal microbiota and bacterial metabolite composition in pregnant Huanjiang mini-pigs: effects of food composition at different times of pregnancy

The gut harbours diverse and complex microbiota, which influence body health including nutrient metabolism, immune development, and protection from pathogens. Pregnancy is associated with immune and metabolic changes that might be related to microbiota compositional dynamics. We therefore investigated the colonic luminal bacteria community in Huanjiang mini-pigs fed diets with different nutrient levels from the first to third trimester of pregnancy. The concentrations of intestinal metabolites including short-chain fat acids, NH3-N, indole, skatole, and bioamines were also determined. We found that the colonic bacteria species richness estimators (Chao1 and ACE) decreased with increased gestational age. The dominant phyla identified were Firmicutes and Bacteroidetes; the dominant genera were Lactobacillus, Treponema, Ruminococcus, Clostridium, and Prevotella. In addition, microbiota displayed spatial and temporal heterogeneity in composition, diversity, and species abundance in different colonic segments from the first to third trimester of pregnancy. Furthermore, the bacterial metabolites also changed according to the diet used and the pregnancy stage. These findings suggest that colonic bacteria richness decreased as gestational age increased, and that the higher nutrient level diet increased the production of metabolites related to nitrogen metabolism. However, although the higher nutrient diet was associated with pregnancy syndrome, causal links remain to be determined.

and Chao1 and ACE were significantly decreased (P < 0.05) with pregnancy stage. Bacterial diversity showed no difference between colonic content samples (P > 0.05).
Changes in the bacterial phylum, genera, and OTUs of colonic content microbiota. Principal coordinate analyses (PCoA) were conducted to determine the phylogenetic relationship between the tested microbial community samples ( Supplementary Fig. S1). Using the detected OTUs, we determined whether the data separated into distinct sample clusters. The CNF and NRC diet groups could almost be separated using the proximal colon contents according to pregnancy stage. Pregnancy day 45, 75, and 110 proximal colon content samples were clearly separated in the NRC but not in the CNF diet group. No separation was detected in clusters using the distal colon contents ( Supplementary Fig. S1). At the phylum level, the abundance of 10 phyla in any sample was ≥ 0.5%; i.e., Firmicutes, Bacteroidetes, Proteobacteria, Actinobacteria, Spirochaetes, Tenericutes, Verrucomicrobia, Fibrobacteres, Cyanobacteria, and Deferribacteres. In the colonic samples, Firmicutes and Bacteroidetes were the two dominant phyla. Firmicutes were decreased in the proximal colon content samples but Bacteroidetes increased from pregnancy days 45 to 110 of sows fed the NRC diet. In comparison, the relative abundance of Firmicutes was higher, but Bacteroidetes was lower in NRC diet-fed than in CNF diet-fed sows (Fig. 1a). In distal colon content samples, the Firmicutes abundance increased whereas Bacteroidetes from pregnancy days 45 to 110 with the NRC diet decreased. The Firmicutes abundance was almost stable from pregnancy days 45 to 110 (Fig. 1b). Overall, the Firmicutes abundance was greater, but Bacteroidetes abundance was lower in the distal compared to the proximal colon content samples.
The relative abundance of genera, among the 37 observed, was ≥ 0.5% in every sample (Fig. 2). In the proximal colon contents, within the Bacteroidetes phylum, the genus Paludibacter and Parabacteroides abundances were higher (P < 0.05) in the CNF vs. the NRC diet group. Within the Firmicutes phylum, the genera Anaerovibrio, Lactobacillus, Mitsuokella, and Roseburia abundances were lower (P < 0.05) in the CNF diet group. The unclassified genera changed significantly with diet (P < 0.05), with higher abundance in the CNF diet group. With increasing pregnancy time, Corynebacterium, Allobaculum, Mitsuokella, Roseburia, Solibacillus, Staphylococcus, Acinetobacter, and the unclassified genus changed significantly (P < 0.05). Within the Actinobacteria and Proteobacteria phyla, genera Corynebacterium and Acinetobacter abundance, respectively, decreased linearly with pregnancy stage (P < 0.05) as did genus Allobaculum abundance within the Firmicutes phylum. Unclassified genus and Firmicutes phylum Mitsuokella abundance peaked on pregnancy day 45 and exhibited its lowest level on pregnancy day 75 (P < 0.05), when Firmicutes Roseburia and Solibacillus abundances were highest (P < 0.05) but Staphylococcus was lowest (P < 0.05). Diet type interacted with the pregnancy stage for the genera of Bifidobacterium, Allobaculum, Megasphaera, Mitsuokella, Roseburia, Solibacillus, and Staphylococcus.
In the distal colon contents, within the Actinobacteria phylum, the genus Bifidobacterium abundance was higher (P < 0.05) in CNF vs. NRC diet groups. Within the Firmicutes phylum, the genera Anaerovibrio, Dialister,  Roseburia, and Veillonella abundances were lower (P < 0.05) in the CNF diet groups. The effects of pregnancy stage on bacterial composition were as follows: within the Firmicutes phylum, the genera Allobaculum, Dialister, Lactobacillus, Megasphaera, Mitsuokella, Staphylococcus, and [Ruminococcus] abundances changed significantly with pregnancy stage, being lowest (P < 0.05) on pregnancy day 75 for Allobaculum, Megasphaera, and Mitsuokella and on pregnancy day 45 for Dialister (P < 0.05). Lactobacillus and [Ruminococcus] abundances increased whereas Staphylococcus abundance decreased linearly with pregnancy stage (P < 0.05). The unclassified genus abundance was lowest on pregnancy day 75 (P < 0.05) (Fig. 2).
Intestinal metabolites: SCFAs. Table 4 shows the SCFA concentrations in the colonic contents of pregnant Huanjiang mini-pigs fed two different nutrient level diets. In the proximal colon, isovalerate and total branched-chain fatty acids (BCFAs) decreased linearly from the first to the third trimester of pregnancy (P < 0.05). Acetate and total SCFAs trended toward decreasing with pregnancy trimester (0.05 ≤ P < 0.10). Acetate, propionate, and total SCFAs displayed significant differences according to diet × stage interaction  (P < 0.05), with isobutyrate trending toward change (0.05 ≤ P < 0.10). In the distal colon, isobutyrate displayed significant differences according to diet (P < 0.05). Propionate and total BCFAs trended toward decreasing according to pregnancy stage (0.05 ≤ P < 0.10). Propionate, isobutyrate, total SCFAs, and total BCFAs displayed significant differences (P < 0.05) according to diet × stage interaction, whereas acetate, isopentanoate, and total BCFAs only trended toward change (0.05 ≤ P < 0.10).

Discussion
Intestinal metabolites. Complex oligosaccharides and indigestible fibre matter not absorbed in the upper intestinal tract are fermented by the colonic anaerobic microbial community to produce SCFAs 18 ; SCFAs can also be produced from some luminal amino acids 19 . Acetate, propionate, and butyrate are the major SCFAs produced in the colon. Among these, butyrate is thought to exhibit health benefits by acting as a major energy source for colonic epithelial cells, regulating gene expression in colonocytes, and exhibiting immunomodulatory and anti-inflammatory properties 20 . BCFAs including isobutyrate and isovalerate are produced during protein fermentation by gut bacteria 21 . Isobutyrate and isovalerate are produced from L-leucine and L-valine, respectively 22 . The amount of BCFAs in the faeces and intestinal contents thus constitutes an indicator of protein catabolism in the large intestine 19 . Decreased BCFA amounts suggest that fewer dietary and/or endogenous proteins escape digestion in the small intestine and enter into the large intestine. In this study, isovalerate and total BCFAs decreased linearly from the first to the third trimester of pregnancy in the proximal colon, and acetate and total SCFAs trended towards decrease. Additionally, propionate in the distal colon decreased linearly with pregnancy progression. Several studies have shown that the SCFA levels in the large intestine influence body weight gain and adiposity 23,24 . Specifically, obese mice demonstrated significantly increased caecal propionate and butyrate contents vs. their lean counterparts 23 . The reduced acetate levels would likely result in decreased lipogenesis and the increased propionate would contribute to inhibiting acetate conversion into lipid in the liver and adipose tissue 25 . The colonic microbiota catabolises nitrogenous compounds to putrefactive catabolites such as ammonia, biogenic amines, indoles, and phenols 19 . NH 3 -N in the colonic lumen increases with an increase in dietary protein intake 26 and is produced by the microbiota via urea degradation and amino acid deamination 19 . In the proximal colon, NH 3 -N was higher after NRC diet consumption than in the CNF diet groups; notably, excessive NH 3 -N in the colonic contents has been shown to inhibit colonocyte oxygen consumption 27 . Leuconostoc, Bacillus, Atopostipes, Bacteroides, and Pseudomonas metabolise tryptophan to indole-3-lactate and then to indole, or further to skatole in animal faeces 28 . In the distal colon, indole was significantly higher in both the NRC diet groups and according to pregnancy stage, consistent with the findings of Keseler et al., wherein lesser amounts of indoles were produced by the lower vs. the higher crude protein group 28 . Indole may be beneficial for the colonic epithelium by increasing epithelial cell tight-junction resistance 29 , whereas skatole has been associated with poor organoleptic characteristics of pig meat 30 , although its effects on the colonic mucosa remain unknown.
Bioamines comprise organic bases with one or more amine (NH 2 ) group (s), and are formed mainly through the decarboxylation of free amino acids by bacterial decarboxylases 31 . Putrescine is synthesised from arginine or ornithine, which can occur simultaneously in many bacteria 32 . Putrescine appears to be strictly necessary for optimal undifferentiated colonic epithelial cell mitosis 33 . The precursor of cadaverine and tyramine is lysine 34 , and tyramine is synthesised by the decarboxylation of tyrosine 35 . Spermine is produced by spermidine, which in turn is produced from histidine, ornithine, arginine, and methionine 36 .
The bacteria community. The bacterial community of the colon was determined from the first (day 45), second (day 75), and third (day 110) stages of pregnancy. To accurately capture the majority of bacterial OTUs within the colonic contents of Huanjiang mini-pigs, the study was normalised to samples at a depth of 31,465 sequences/sample with good coverage (> 99.2%). We selected the V4 variable regions to interrogate the colon content bacterial communities. Alpha diversity analyses were performed using the proximal colonic contents samples across higher and lower nutrient diets and the three pregnancy trimesters. The OTUs decreased significantly according to pregnancy stage, and richness (Chao1 and ACE) changed significantly according to the pregnancy stage in both colonic segments analysed, suggesting that the microbiota diversity is lower at the end than at the beginning of pregnancy. This result is consistent with the results from Collado et al., who detected lower total faecal cell counts in the third than at the first pregnancy trimester 7 . In addition, Koren et al. also found that gut microbiota diversity and richness decreased with pregnancy 6 . Furthermore, Le Chatelier et al. found that non-obese and obese Danish individuals differed in bacterial richness with low richness associated with higher overall levels of body fat 37 . The sows in our present study also gained considerable weight from the period of remating to late pregnancy. The unweighted UniFrac PCoA reflected the similarities of the bacterial communities within the colon. In contrast, we found that the CNF and NRC diet groups could almost be separated in the proximal colonic samples. However, unlike the samples recovered from the pregnant sows on days 45, 75, and 110 in the NRC diet group, the obtained proximal colonic samples in the CNF diet group could not be separated. Furthermore, our analyses identified no cluster separation in the distal colonic samples. These results demonstrate that the structure or diversity of the bacterial community likely changes to a greater extent in proximal colonic contents and that NRC diet consumption is associated with higher level of changes in the colonic content than is consumption of a CNF diet.
Firmicutes was prominent in the digesta-associated colon microbiota in all samples, for the two diets as well as throughout the three pregnancy trimesters. Firmicutes was composed mainly of the genera Lactobacillus, Clostridium, Oscillospira, Ruminococcus, Turicibacter, Roseburia (≥ 1%), and the other genera (< 1%), which together comprised up to 45.43-74.57% of the total reads in some groups. The dominant taxa belonging to Firmicutes in the digesta-associated with the distal colon likely play an important role in starch and fibre degradation 38 . Lactobacillus abundance was higher in the proximal colon of pigs receiving the NRC diet vs. the CNF diet. OTU-31 (s: Lactobacillus reuteri) and OTU-1511 (s: L. reuteri) in the distal colon contents were decreased with pregnancy progression. L. reuteri exhibits strain-specific beneficial properties relevant to human health, with human associated lineages distinguished by genes related to bacteriophages, vitamin biosynthesis, antimicrobial production, and immunomodulation, although the strains show evidence of host adaptation 39 . In addition, Scientific RepoRts | 6:37224 | DOI: 10.1038/srep37224 Lactobacillus species have been associated with weight changes in humans and animals 40 . In comparison, the Roseburia component was larger in the whole colon of pigs receiving the NRC diet. Haro et al. reported that Roseburia was modulated by diet and that the degree of Roseburia increase was associated with a protective effect on the development of type 2 diabetes based on a human obese population with different diets 41 . In our study, the relative abundance of Staphylococcus, which accounted for 0.05-0.29% of the total sequences in the colon, linearly decreased from the first to the third trimester of pregnancy, which is not consistent with the results obtained in pregnant women 7 .
Bacteroidetes comprised the second dominant phylum in digesta-associated colonic microbiota. This phylum was more abundant in the proximal colonic content in CNF diet-fed animals, and more abundant in the proximal than the distal colon content samples. This phylum was prominent and mainly composed of the genera Prevotella, YRC22, and Parabacteroides. Prevotella comprised 1.83-7.28% of the total sequences in some groups. OTU-62, OTU-89, and OTU-1967 (all g: Prevotella) were linearly decreased with pregnancy. Prevotella has been proposed (together with other genera such as Xylanibacter, Butyrivibrio, and Treponema) to enhance caloric extraction from resistant starches and oligosaccharides, as well as from indigested carbohydrates 42 . Another study showed that long-term consumption of diets rich in carbohydrates can increase the abundance of Prevotella 43 . In addition, a study showed that Prevotella decreased in rats fed a high fat diet 44 .
Firmicutes and Bacteroidetes comprised the dominant colonic phyla, as previously suggested 45 . Earlier studies showed that obesity and glucose intolerance are associated with an increase in Firmicutes 37 . Obesity is also associated with a higher Firmicutes/Bacteroidetes ratio 46 . A high-fat diet can increase the Firmicutes/Bacteroidetes ratio within a small range 47 . Conversely, a higher saccharolytic diet may preferentially increase Bacteroidetes as compared to Firmicutes 48 . In our study, Firmicutes abundance was higher in distal colonic contents, whereas Bacteroidetes showed the opposite. Firmicutes was affected by diet and stage, with the NRC diet increasing its abundance in the proximal colon. In addition, Firmicutes was increased at the end of the pregnancy period in the distal colon. This increase was considered to alter the Firmicutes metabolic potential and enhance the body's capacity to harvest energy from the diet 49 , a potential advantage during pregnancy to support growth of the foetus and prepare the body for the energetic demands of lactation. Jost et al. revealed that Firmicutes exhibited no detectable change over the perinatal period 8 . However, Khan et al. found that Firmicutes abundance was profoundly impacted during pregnancy 50 . Bacteroidetes was affected by diet only in the proximal colon, and the NRC diet decreased its abundance. The ratio of Firmicutes/Bacteroidetes was larger in groups fed the NRC vs. the CNF diet, suggesting that the gut microbiota of sows fed the NRC diet was more similar to obesity models than those fed the CNF diet.
Large abundances of other phyla were identified for Proteobacteria, Actinobacteria, Spirochaetes, and Tenericutes, which are present in the majority of gut-associated phylotypes in mammals 51 . These phyla are ubiquitous within the mammalian gut suggesting their critical role in the microbial ecology of the mammalian intestine 52 . Notably, Enterobacteriaceae and E. coli are higher in overweight than in normal-weight women and in women with excessive weight gain during pregnancy 53 . In the current study, Escherichia (phylum Proteobacteria) were more abundant in the proximal colon of pigs fed with HN NRC diet. As the NRC diet contains higher nutrient level than the CNF diet, the sows fed NRC demonstrated higher body weights. E. coli is sufficient to induce inflammation and glucose and insulin tolerance in germ free mice 54 , as well as acute diarrhoea in pregnant women 55 .
Several studies suggested that the gut microbiota change during pregnancy may also induce pregnancy complications such as excessive maternal weight gain 56 . Notably, our previous study showed that the average back-fat thickness of pregnant Huanjiang mini-pigs increased from days 45 (27.20 mm and 26.90 mm) to 75 (36.60 mm and 28.10 mm) post-service in both NRC and CNF diet groups, respectively 57 ; the live body weight of these sows also increased from days 45 (73.82 kg and 67.52 kg) to 75 (86.14 kg and 75.28 kg) in both groups (Kong et al., unpublished data). These findings suggested that the pregnant sows became obese during pregnancy, especially in the NRC group, and that this was associated with more abundant Escherichia. Actinobacteria was more abundant in the distal colonic contents recovered from animals fed the CNF diet and was in addition more abundant in the distal than in the proximal colon. Among bacteria, Bifidobacterium and Corynebacterium were more abundant in the distal colon from the CNF vs. the NRC groups, consistent with a previous study showing that high-fat feeding is associated with lower Bifidobacteria concentrations 9 . Notably, the Bifidobacterium group is present in higher numbers in normal-weight than in overweight women 7 , and in women with lower weight gain over pregnancy 7 . The present study indicates that consumption of the diet with higher nutrient level coincides with the presence of a colonic microbiota in pregnant sows that resembles the gut bacteria composition observed in obese animals.

Conclusion
Consumption of a higher nutrient level diet increases the colonic luminal content of bacterial metabolites related to nitrogen metabolism, suggesting increased dietary and endogenous protein transfer from the small to the large intestine 58 and/or modified composition/metabolic capacities of the intestinal microbiota toward luminal protein/amino acids. The dominate phyla in the colonic luminal contents were, to different extents, Firmicutes, Bacteroidetes, Proteobacteria, Actinobacteria, Spirochaetes, and Tenericutes. The dominate genera in the colonic content of pregnant sows were Lactobacillus, Treponema, Ruminococcus, Clostridium, and Prevotella. The richness of bacteria in the colon decreased from the first to the third trimester of pregnancy. Lastly, consumption of an NRC diet resulted in a colonic microbiota profile of sows resembling the gut bacteria composition observed in human obesity. This might be associated with an increased risk of pregnancy syndrome, although no causal link between these parameters was established in our study, calling for further experimental works.

Materials and Methods
Animals, housing, and treatment. The present study was carried out in accordance with the Chinese guidelines for animal welfare and experimental protocols and approved by the Animal Care and Use Committee of the Institute of Subtropical Agriculture, Chinese Academy of Sciences.
Forty-eight primiparous Huanjiang mini-pigs with a mean body weight (BW) of 43.02 ± 8.87 kg were obtained from a mini-pig farm located in Jixiang town (108°27′ 40.8″ E and 25°9′ 50″ N, altitude 578 m), Huanjiang county (Guangxi province, China). These sows were randomly assigned to one of the two dietary groups post service (24 sows per each dietary group and three sows per pen). One group of sows received diet with a higher nutrient level and the other received diet with a lower nutrient level. The higher nutrient level diet was formulated to meet the nutrient requirements recommended by the National Research Council (1998) (NRC diet) and contained 14.50 MJ/kg digestible energy, 13.10% crude protein, and 4.56% crude fibre. The lower nutrient level diet was formulated to meet recommendations of the Chinese National Feeding Standard for Swine (CNF diet), and thus contained 12.20 MJ/kg of digestible energy, 11.00% of crude protein, and 6.86% of crude fibre (Supplementary Table S3). These diets are widely used in commercial crossbred pig farms and commercial Huanjiang mini-pig farms, respectively.
All animals were housed in 2 m × 3 m pens with cement-sclerified flooring. Each pen was equipped with a feeder and a nipple drinker. The room temperature was maintained at 22-28 °C. All pigs had ad libitum access to drinking water and were fed twice daily (08: 30 and 16:30) with their diets (about 2.5% of body weight) after service. The feed intake, excreta, and mental state per sow were monitored twice each day throughout the entire experimental period.
Sample collection and preparation. According to the report of Johnston and Trottier (1999) 59 , the early, middle, and later stages of pregnancy in pigs are from days 1 to 30, from days 30 to 75, and from days 75 to delivery, respectively. Considering that the Huanjiang mini-pigs represent miniature animals and it is difficult to collect the conceptus samples to determine foetus development for subsequent study, we chose 45 days post-service to sample as the early pregnancy stage.
This study was conducted using samples collected from 31 sows offered two experimental diets. On day 45, 75 or 110 of service, the pregnant sows from each treatment group were sacrificed for sample collection at 12 h after the last feeding 60 . Briefly, the pigs were administered general anaesthesia by intravenous injection of 4% sodium pentobarbital solution (40 mg/kg BW), and sacrificed by exsanguination via the carotid artery 61 . After colon recovery, the luminal contents were collected from a region 10 cm posterior to the ileocecal valve and a region 10 cm at the end of the colon, respectively, and then stored at − 80 °C. Subsequently, the contents of ammonia (considered as the sum of NH 3 and NH 4 + ), SCFAs, indole, skatole (3-methylindole), and bioamines were analysed, and the composition of the gut microbiota was determined.
DNA Extraction and PCR Amplification. Total genomic DNA was extracted from the colon content samples using a QIAamp DNA Stool Mini Kit (Qiagen, Hilden, Germany), according to the manufacturer's instructions. The quality and quantity of DNA were measured using a NanoDrop ND-1000 spectrophotometer (NanoDrop Technologies Inc., Wilmington, DE, USA). An absorption ratio (260/280 nm) within 1.8-2.0 was deemed to be of sufficient purity to be used for subsequent analyses. The total DNA of the colon contents was then diluted to 50 ng/μ l and used in the preparation of amplicons for high-throughput sequencing of microbial 16S rRNA. Conventional PCR was performed to amplify the V4 regions of the 16S rRNA gene using universal primers 515F (5′ -GTG CCA GCM GCC GCG GTA A-3′ ) and 806R (5′ -GGA CTA CHV GGG TWT CTA AT-3′ ). The reverse primer contained a 6-bp error-correcting barcode unique to each sample. DNA was amplified using the protocol described previously 62 . DNA samples were sent to a commercial service provider (Novogene, Beijing, China) for amplicon pyrosequencing on an Illumina MiSeq platform according to the manufacturer's instruction 63 . Microbial community analysis. Raw data were screened and assembled by QIIME 64 and FLASH 63 software packages. The reads flagged as chimeras were removed to form the "effective sequences" collection for each sample. QIIME software package (http://qiime.org/) and UPARSE pipeline (http://drive5.com/uparse) were used for analysis of the sequences and determination of the Operational Taxonomic Units (OTUs) with an identity threshold of 97%. Meanwhile, we picked a representative sequence for each OTU and used the RDP classifier 65 to assign taxonomic data to each representative sequence. Taxonomy classifications were assigned against RDP Classifier (Version 2.2, http://sourceforge.net/projects/rdpclassifier/) 65 and the Greengenes database (http:// greengenes.lbl.gov/cgi-bin/nph-index.cgi) 66 . Alpha diversity values for colon bacterial communities of pregnant Huanjiang mini-pigs were obtained using abundance-based coverage estimator (ACE), bias-corrected Chao richness estimator, Shannon index, Simpson index, and Good's coverage, at a sequence depth of 31,460 for all samples. Principal coordinate analysis (PCoA) of microbial communities (jackknifed beta diversity from resampling 100 times at depth of 31,460 sequences) was performed using Bray-Curtis distance.
Metabolites in colon contents. SCFAs including acetate, propionate, butyrate, isobutyrate, valerate, and isovalerate were analysed as described previously 67 . To ensure the homogenicity of the intestine content sample, the freeze-dried samples were prepared using a Vacuum freeze-dryer (Hrist ALPHA 2-4/LSC, Germany) at − 80 °C. Briefly, freeze-dried samples (0.5-0.6 g) were weighed into 10 ml centrifuge tubes and mixed with 8 ml ddH 2 O, homogenised, and centrifuged in sealed tube at 7,000 g and 4 °C for 10 min. A mixture of the supernatant fluid and 25% metaphosphoric acid solution (0.9 and 0.1 ml, respectively) was centrifuged at 20,000 g and 4 °C for 10 min after standing in a 2 ml sealed tube at 4 °C for over 2 h. The supernatant portion was then filtered through a 0.45-μ m polysulfone filter and analysed using Agilent 6890 gas chromatography (Agilent Technologies, Inc, Scientific RepoRts | 6:37224 | DOI: 10.1038/srep37224 Palo Alto, CA, USA) with a flame ionisation detector and a 1.82 m × 0.2 mm I.D. glass column that was packed with 10% SP-1200/1% H 3 PO 4 on the 80/100 Chromosorb W AW (HP, Inc., Boise, ID, USA). The concentration of NH 3 -N in the supernatant fluid was measured at 550 nm using a UV-2450 spectrophotometer (Shimadzu, Kyoto, Japan) 68 . The bioamines including 1,7-heptyl diamine, cadaverine, phenylethylamine, putrescine, trytamine, tyramine, spermidine, and spermine, as well as the indoles and skatoles, were analysed as described previously 69 . Statistical analysis. The alpha diversity indices of bacterial communities, the apparent relative abundances at phylum and genus levels, as well as colon metabolites (SCFA, NH 3 -N, indole, skatole, and bioamines) were analysed in a completely randomised design with the GLM procedure of SAS (SAS Institute, Inc., Cary, NC). The PCoA of overall diversity of microbial communities based on an un-weighted Unifrac metric was performed using Bray-Curtis distance to compare all samples. A study of the relationship among parameters was also carried out using Pearson's linear correlation coefficient. Phylum and genus at < 0.5% relative abundance for both diets were excluded from all analyses. Differences with P values < 0.05 were considered as statistically significant while a tendency was considered to exist at 0.05 ≤ P < 0.10.