Introduction

Healthy livestock farming technologies and management practices are a central concern in animal farming industry, which must provide high-quality meat products to safely meet human consumption requirements. In particular, pigs are one of the most important economic livestock species in numerous countries. Pigs represent the largest livestock product in China, with great commercial and economic values worldwide (Bai et al. 2014; Sun and Jia 2015; Yu and Abler 2014). However, intensive or large-scale livestock farming causes various problems such as serious feed resource shortages and grain competition between humans and livestock (Ilea 2009). In addition, the utilization of antibiotics in livestock husbandry potentially poses a threat to livestock health, and it could lead to environmental pollution as well as a decline in animal immunity, gut microbiota disorders, and an increased risk of the spread of antibiotic-resistant bacteria genes in the environment (Baquero et al. 2008; Wu et al. 2010; Zhao et al. 2018). Residues of antibiotics in livestock products are likely harmful to human health (Marshall and Levy 2011). However, the restricted use of antibiotics in feed used in livestock farming may increase disease rates in animals and decrease amounts of global products. Therefore, it is imperative to develop safe and effective feed additives that could replace antibiotics, thus improving livestock health and products. Recently, studies focused on replacing antibiotics in livestock husbandry have examined the effects of probiotics, prebiotics, synbiotics, and other dietary additives on livestock (Abudabos et al. 2017; Markowiak and Slizewska 2018; Marshall and Levy 2011). Moreover, these studies have aroused the interests of scientists globally, since these newly developed feed supplements were found to improve livestock health and growth (Markowiak and Slizewska 2018).

Gut microbiota has several beneficial functions in hosts, including food digestion, energy harvesting, immune regulation, and resistance against pathogens (Rooks and Garrett 2016; Stanley et al. 2016; Tremaroli and Bäckhed 2012). Evidence suggests that probiotics and prebiotics could beneficially modify gut microbiota composition and activity in humans and animals (Barba-Vidal et al. 2018; Moura et al. 2007), thereby improving host metabolism and health, notably at the intestinal level. For example, the probiotics Lactobacillus (Lactobacillus reuteri avibro) and Bacillus (Bacillus subtilis and Bacillus licheniformis) exhibited the ability to increase nutrient digestibility and animal performance, and they reduced the abundance of pathogens (Salmonella and Escherichia coli) (Ahmed et al. 2014). In pigs, some prebiotic products, such as xylo-oligosaccharides (XOS), β-mannanase, mannan-oligosaccharides, and yeast cultures have been used to modify gut microbiota composition, improve host immune response, stimulate the growth of more beneficial bacteria, and inhibit the colonization or abundance of pathogens by producing antimicrobial substances in humans and livestock (Barros et al. 2015; De Maesschalck et al. 2015; Liu et al. 2018; Rastall and Gibson 2015). For instance, in broiler chicken, XOS feed additives modulate the gut microbiota and increase the abundance of the potential beneficial bacteria Lactobacillus and Bifidobacterium (Pourabedin et al. 2015). Moreover, feed containing yeast cultures increased the abundance of Clostridium and Methylobacterium, which are important cellulose-degrading bacteria that may help herbivorous grass carp (Ctenopharyngodon idellus) to degrade ingested plant-based food (Liu et al. 2018).

Recently, liquor lees were been used as promising feed additive candidates, and they were found to be able to shape animal gut microbiota. For example, the content of dried distiller grains with solubles (DDGS) in feed is positively correlated with gut microbial diversity in birds (Abudabos et al. 2017), and increased microbial diversity is associated with metabolic activity and health in human and animal hosts (Li et al. 2017b, 2018; Tap et al. 2015). However, studies of the effects of liquor lees on livestock gut microbiota are still in their infancy and need further development.

Guizhou Maotai liquor is a fragrant and tasty wine (Wu et al. 2013), and it is renowned as the best Chinese sauce fragrance liquor. The production of Maotai wine was about 38,700 t, and the solid by-products (including lees) were about 110,000 t (Li et al. 2015). DDGS is a suitable substrate for solid-state fermentation (SSF), and only moderate changes were found in its nutritional profile after SSF (Yang et al. 2012). Fermented Maotai lees (FML) also have a high nutritional value (e.g., high protein, cellulose, amino acids, and organic acids), so they may be of interest for livestock farming. Here, we used MiSeq sequencing of 16S rRNA genes to evaluate the effects of FML on the gut microbiota diversity and also measured the gut metabolic products (short-chain fatty acids (SCFAs) and bioamines) in pigs. We solved three questions: (1) Do FML influence the SCFA and bioamine pofiles in the pig gut? (2) Whether FML modify gut microbiota composition and function? (3) Do FML increase the abundance of potential beneficial bacteria and decrease the abundance of specific pathogens?

Materials and methods

Experimental design and ethical standards

A total of 24 Duroc × Large White × Landrace hybrid barrows (male), with an initial body weight (BW) of 42.28 ± 1.23 kg (mean ± SE), were fed a corn and soybean meal-based diet (Additional file 1: Table S1), which met the National Research Council (NRC 2012) requirements for growing-finishing pigs. After these pigs were fed the same diet for 1 week, the animals were randomly arranged to one of the four treatments. Each treatment group had 6 replicates. Pigs in the control group were fed a basic diet, whereas those in the experimental groups were fed the basic diet containing 5%, 10%, or 15% fermented Maotai lees (FML) [0% as Control, 5% (treat 1), 10% (treat 2) and 15% (treat 3)] during the experimental period. Notably, the FML used in the present study were provided by the Road Biological Environmental Co., Ltd., Sichuan, China. The FML products were fermented by saccharomyces yeasts with acid-resisting, overproducing enzyme, and high vigor. After fermentation, the mixture was dried at 60 °C using laboratory incubator, and then smashed. The FML were then stored in − 20 °C for experimental use. The determined nutrient levels (%) of the FML based on dry matter content (92.97%) were as follows: ash, 9.28; gross energy (GE), 18.29; crude protein (CP), 23.96; ether extract (EE), 5.39; crude fiber (CF), 17.67; acid detergent fiber (ADF), 38.06; neutral detergent fiber, 47.28; Ca, 0.53; P, 0.55.

Experimental pigs were housed in cages (3.5 m × 5.0 m) equipped with feed intake recording equipment (Bei**g Hamoer Automation Equipment Co., Ltd., Bei**g, China). The space provided by the equipment allows one pig at a time to have ad libitum access to the diet. Animals had 24 h access to feed and water during the whole course of the experiment. Pigs were labeled with an individual electronic ear marker. The feeding trial lasted for 90 days. The final body weight of each pig was also measured to evaluate the growth status.

The experimental design and procedures in this study were reviewed and approved by the Animal Care and Use Committee of the Institute of Subtropical Agriculture, Chinese Academy of Science. Processing of animal experiments and sample collection strictly followed the relevant guidelines.

Sample collection

At the end of the feeding trial, the diet was removed 12 h before slaughter. All pigs were transported from the farm to the processing facility (approximately 40 km) at 7:00 h, and sacrificed at 19:00 h under commercial conditions using electrical stunning (120 V, 200 Hz). Colonic contents were collected into 50 ml sterile tubes. After complete mixing, all the gut contents were stored at − 20 °C for the following microbiota and metabolic analysis.

Measurements of short-chain fatty acids (SCFAs) and bioamines

Gut SCFAs (mg/g), including straight-chain fatty acids (acetate, propionate, butyrate, and valerate) and BCFAs (branched-chain fatty acid, including isobutyrate and isovalerate) were analyzed using gas chromatography as previously described (Ji et al. 2018). Bioamines (μg/g), including putrescine, cadaverine, tyramine, spermidine, and spermine were measured using high-performance liquid chromatography as previously described (Ji et al. 2018).

DNA extraction and MiSeq sequencing of microbial 16S rRNA gene

Genomic DNA of colonic contents was extracted using a QIAamp DNA Stool Mini Kit (Qiagen, Hilden, Germany) according to the manufacturer’s instructions. The DNA concentration of each sample was measured with a NanoDrop® ND-1000 instrument (NanoDrop Technologies Inc., Dover, USA). The extracted DNA was diluted to 10 ng/µl for the following polymerase chain reaction (PCR) amplification. The protocols of PCR amplification, gel extraction, and sequencing library construction were described previously (Li et al. 2011). Due to possible contamination of chloroplast sequences in PCR amplification, the Metaxa2 software tool was used to remove chloroplast sequences from our large sequencing datasets (Bengtsson-Palme et al. 2015). After filtering chimeras and chloroplasts, the remaining sequences were clustered into operational taxonomic units (OTUs) at a 97% identity threshold with an open-reference OTU picking method using the Uclust algorithm (Edgar 2010). Those sequences not classifying to bacteria (Eukaryota and Archaea lineages) were filtered out. Singleton OTUs were also removed. The most abundant sequences within each OTU were defined as “representative sequences”. Taxonomic classification of representative sequences was implemented using the Ribosomal Database Project classifier in the QIIME platform (Wang et al. 2007).

To minimize the impact of uneven sequencing depth, each sample was rarefied to 32,805 sequences. To evaluate alpha diversity indices, Goods coverage, Chao1, observed OTUs, Shannon diversity and evenness were calculated. The rarefaction curves were generated from the observed OTUs at the OTU level. To assess beta diversity, Jaccard and Bray–Curtis distance metrics were produced through the QIIME pipeline. Jaccard distance is based on the presence/absence of OTUs/species (Jaccard 1912), whereas Bray–Curtis distance is based on the both OTU abundance and presence/absence (Bray and Curtis 1957). Differences in overall bacterial community structure among groups were visualized using the non-metric multidimensional scaling (NMDS) plots of the two dissimilarity matrices.

Statistical analysis

Permutational multivariate analysis of variance (PERMANOVA) was used to reveal whether the structures of gut microbiota were significantly different among groups based on the Jaccard and Bray–Curtis distance matrices using “adonis” in the R ‘vegan’ package (Li et al.

Table 1 The comparison of body weight, fecal short chain fatty acids (SCFAs), and bioamine concentrations in pigs with fermented Mao-tai lees (FML) contents
Full size table
Fig. 1
figure 1

NMDS plots of dissimilarity metrics comparing the profiles of gut microbiota, SCFAs and bioamines among groups. a SCFA profile based on Bray–Curtis dissimilarities. b Bioamine profile based on Bray–Curtis dissimilarities. c Gut microbiota structure based on Jaccard dissimilarities. d Gut microbiota structure based on Bray–Curtis dissimilarities

Full size image

In addition, the concentrations of most bioamines showed no differences between groups, whereas the spermidine concentration of 10% FML group was significantly higher than that of control group (Table 1). Over bioamine profile structure also showed no significant separation between control and treat groups (Fig. 1b; Additional file 1: Table S2; PERMANOVA, all P > 0.05). Final body weight also had no significant impacts in sha** the bioamine profile (PERMANOVA, P > 0.05).

The influences of FML on alpha and beta diversity of pig gut microbiota

A total of 2,394,088 sequences were generated from the 24 pig samples. After filtering out low-quality sequences, chimeras, chloroplasts, singletons, and those sequences not classifying to bacteria, we obtained 922,668 valid sequences. To compare samples with different sequencing depth, each sample was rarefied to 32,805 sequences. At a threshold of 97% sequence similarity, 21,920 unique OTUs were identified using Uclust clustering. The rarefaction curve of observed OTUs across all samples reach stable values (Additional file 1: Fig. S1), indicating that our sequencing had captured most species although additional sequencing may detect some rare OTUs. In addition, the Goods coverage (mean ± SE) of gut microbiota at OTU level across all samples was 95.01% ± 0.26%, confirming that our sequencing depth is enough to detect a majority of gut bacterial species. The alpha diversity values (including Chao1, observed OTUs, Shannon diversity, and evenness) had no significant differences between the control and treat groups (Mann–Whitney U tests, all P > 0.05; Table 2).

Table 2 Alpha diversity values of the gut microbiota were evaluated using OTUs defined at 97% similarity threshold
Full size table

However, the beta diversity values of gut microbiota had significant differences across groups based on Jaccard (Fig. 1c; Table 3; PERMANOVA, R2 = 0.057, P = 0.003) and Bray–Curtis distance matrices (Fig. 1d; Table 3; PERMANOVA, R2 = 0.212, P = 0.008). When we compared the pairwise dissimilarity matrices between any two groups, our results showed that only the community structure of 15% FML group had significant differences with that of the control group (Table 3) based on Jaccard (PERMANOVA, R2 = 0.11, P = 0.015) or Bray–Curtis distance (PERMANOVA, R2 = 0.171, P = 0.017), indicating that only 15% FML influences the community structure of pig gut microbiota. In addition, the group 15% FML showed different community structures with 5% FML based on Jaccard (PERMANOVA, R2 = 0.114, P = 0.023) or Bray–Curtis distance (PERMANOVA, R2 = 0.228, P = 0.008). However, final body weight had no significant impacts in sha** the beta diversity values of pig gut microbiota (Jaccard, PERMANOVA, P > 0.05; Bray–Curtis, P > 0.05).

Table 3 PERMANOVA showing different community compositions and structures among different groups
Full size table

FML treatments affect the taxonomic composition of gut microbiota

Across all samples, approximately 99% of the total sequences were assigned into 23 phyla. Among these phyla, Firmicutes (62.1%), Bacteroidetes (27.0%), Spirochaetes (3.2%), Proteobacteria (3.0%), Tenericutes (2.2%), and Cyanobacteria (1.2%) were the six dominant bacterial taxa (mean relative abundance > 1%) in the pig gut. Some rare phyla included Actinobacteria, TM7, and WPS-2. Those phyla with mean relative abundance > 0.1% are shown in Fig. 2a. At genus level, S24-7, Prevotella, Lachnospiraceae, Ruminococcaceae, Lactobacillus, SMB53, Clostridiales, Oscillospira, Bacteroidales, Clostridiaceae, Treponema, Escherichia, Clostridium, Ruminococcus, RF39, Roseburia, p-75-a5, Lachnospira, Streptococcus, YS2, and Coprococcus were the dominant bacterial genera (Fig. 2b).

Fig. 2
figure 2

Taxonomic compositions of gut bacterial communities with different dietary treatments with fermented Mao-tai lees in pigs. a The relative abundance of bacterial phyla. b Their relative abundance of top 21 bacterial taxa within a group at genus level

Full size image

We further compared the significant difference of bacterial genera among groups. A total of 14 genera showed significant difference among groups (Table 4). Among these genera, the relative abundance of Roseburia, Desulfovibrio, Phascolarctobacterium, Oxalobacter, Defluviitalea, and Prevotella in the group 15% FML was significantly higher than that of the control group (one-way ANOVA, all P < 0.05). A total of 8 genera, such as Butyricicoccus, Dorea, Clostridium, SMB53, and Peptococcus, showed a lower abundance than that of the control group (all P < 0.05).

Table 4 Comparsion of mean relative abundance of bacterial genera among groups
Full size table

To further compare the taxonomic difference between the control and treat groups, we used LEfSe analysis to differential abundance of bacterial taxa at OTU level among different treatments (Fig. 3). The results showed that a total of 25 bacterial biomarkers/OTUs were differentially abundant among the four groups. OTU54 (belonging to Lactobacillus reuteri), OTU57 (Turicibacter), OTU73 (Clostridiaceae), OTU72 (Clostridiaceae), OTU74 (Clostridium butyricum), OTU75 (Clostridium celatum), and OTU76 (SMB53) were the dominant microbes in the 5% FML group, while OTU 29 (Prevotella) and OTU115 (Oscillospira) were mostly in the control group. A total of 13 OTUs, including OTU117 (Ruminococcus bromii), OTU121 (Phascolarctobacterium), OTU132 (Treponema), OTU 25, 26, and 31 (Prevotella), OTU106 (Faecalibacterium), OTU94 (Roseburia), and OTU36 (Prevotella copri), were significantly enriched in the 15% FML group.

Fig. 3
figure 3

Differences in the pig gut microbiota among the different fermented Mao-tai lees (FML) treatments. Linear discriminant analysis (LDA) effect size (LEfSe) results show that bacterial OTUs/markers were significantly different in abundance between control and FML-treated groups

Full size image

Comparison of the potential beneficial bacteria and pathogens abundance

In order to detect the effects of FML on the potential probiotics (Bifidobacterium, Bacillus, Lactobacillus, Akkermansia, and Faecalibacterium) and pathogens (Escherichia) in pigs (Barba-Vidal et al. 2018; Rastall and Gibson 2015), we compared the abundance of these genera among treatments (Fig. 4). We found that the relative abundance of Bacillus, Lactobacillus, Akkermansia, and Faecalibacterium in 15% FML group is higher than that of the control group. However, the 15% FML showed a lower abundance of Escherichia than that of the control group.

Fig. 4
figure 4

The distribution of potential probiotics and pathogens of pigs among treatments. The mean relative abundance of these genera was normalized using Z-score transformation

Full size image

The differences of predicted gene functions across groups

Based on the functional prediction of 16S rRNA gene sequences, the difference of gene functions among groups were visualized based on NMDS plot of Bray–Curtis dissimilarity at level 3. Overall functional profiles showed no significant differences among groups (PERMANOVA, R2 = 0.197, P > 0.05, Additional file 1: Fig. S2). When we compared the differences between the 15% FML and control groups, a total of 29 gene functions at level 3 showed significant differences between the two groups. Compared with the control group, the 15% FML was more abundant in 15 gene functions associated with metabolism, such as flavone and flavonol biosynthesis, phenylpropanoid biosynthesis, starch and sucrose metabolism, carbon fixation in photosynthetic organisms, sulfur metabolism, nicotinate and nicotinamide metabolism, retinol metabolism, and cyanoamino acid metabolism (Additional file 1: Table S3).

Correlations between bacterial biomarkers and gut metabolites

To understand the relationship between bacterial biomarkers and gut metabolites, we calculated the spearman correlations between those bacterial OTUs (based on LefSe analysis) and SCFAs or bioamines (Fig. 5). Our results showed that acetate was negatively correlated with OTU57, OTU72, OTU75, or OTU76, whereas was positively associated with OTU17, OTU25, OTU26, OTU36, OTU47, OTU82, OTU94, OTU106, OTU118, and OTU121. Propionate only showed positive correlations with OTU17, OTU47, OTU118, and OTU121. Butyrate was related to OTU17, OTU36, OTU47, OTU106, OTU118, and OTU121. These results indicated that most OTUs enriched in the 15% FML group showed correlations with SCFAs. In addition, OTU22 was positively correlated with Putrescine. Tyramine showed positive associations with OTU40, OTU49, and OTU106.

Fig. 5
figure 5

The heatmap plot of spearman correlations between bacterial markers and SCFAs or bioamines. Only those correlations with P < 0.05 are shown

Full size image