3.1 Effect of different fiber content on phenotype of two breeds of pigs
At the end of the experiment, we examined the apparent digestibility of dietary fiber and the feed/gain ratio of the four groups (Fig. 1A-B). We found that increasing the dietary fiber concentration significantly decreased the apparent digestibility of dietary fiber in DLY pigs (P < 0.001), while it had no effect on Ding’an pigs. Additionally, after increasing dietary fiber, the feed/gain ratio of Ding’an pigs significantly reduced (P < 0.05), but DLY pigs displayed an increasing tendency. After the animals were slaughtered, we calculated the dressing percentage of all individuals (Fig. 1C). We discovered that increasing dietary fiber had no effect on Ding’an pigs, while it significantly decreased the dressing percentage in DLY pigs (P < 0.001). These results suggest that, compared to DLY pigs, Ding’an pigs may digest more dietary fiber to provide energy for growth.
3.2 Influence of Dietary Fiber on Intestinal Morphology
The results of intestinal morphology of foregut, including duodenum, jejunum and ileum, are shown in Table 1 and Fig. 1D. The villus height and V/C in the duodenum and ileum of DLY pigs was significantly reduced after increasing the dietary fiber content (P < 0.001), while that of Ding’an pigs was not affected. In both DLY pigs and Ding’an pigs, increasing the dietary fiber content had no effect on the crypt depth of the three intestinal segments.
Table 1
Effect of dietary fiber on intestinal morphology of Ding’an pigs and DLY pigs
Item | Ding’an pigs | DLY pigs | SEM | P-value |
DAL | DAH | DLYL | DLYH | Breed | DF | Interaction |
Duodenum | | | | | | | | |
Villus height(µm) | 376.73b | 374.27b | 416.41a | 374.73b | 1.636 | 0.001 | 0.001 | 0.001 |
Crypt depth(µm) | 170.74b | 162.96b | 199.74a | 202.16a | 2.737 | 0.001 | 0.626 | 0.354 |
V/C | 2.25ab | 2.34a | 2.11b | 1.92c | 0.030 | 0.001 | 0.378 | 0.019 |
Jejunum | | | | | | | | |
Villus height(µm) | 411.68a | 397.09ab | 388.92b | 393.14ab | 2.815 | 0.020 | 0.359 | 0.098 |
Crypt depth(µm) | 185.05 | 179.49 | 174.51 | 171.88 | 2.874 | 0.114 | 0.474 | 0.798 |
V/C | 2.30 | 2.25 | 2.25 | 2.33 | 0.032 | 0.726 | 0.793 | 0.318 |
Ileum | | | | | | | | |
Villus height(µm) | 393.42a | 405.29a | 398.22a | 353.12b | 2.602 | 0.001 | 0.002 | 0.001 |
Crypt depth(µm) | 194.03 | 193.48 | 185.33 | 188.46 | 2.819 | 0.227 | 0.820 | 0.744 |
V/C | 2.06ab | 2.13a | 2.18a | 1.92b | 0.029 | 0.451 | 0.129 | 0.006 |
a,b,c Means with different superscripts in the same row differ significantly(n = 3, P < 0.05). V/C, the ratio of villus height to crypt depth. SEM, standard error of the mean.
3.3 Metagenomic Sequencing Date of Colonic Contents
Twelve samples from colonic contents of pigs were sequenced to explore the microbial composition, with an average of 85,222,539 raw reads per sample. The analysis showed that the rarefaction curves stabilized(Fig. S1A). After removing low-quality sequences and host sequences, there were an average of 83,484,708 high-quality clean reads per sample. A total of 5,171,138 contigs were obtained after assembly, with an average of 430,928 contigs per sample. Finally, we obtained a non-redundant gene catalogue with 2,964,637 genes for further research, with an average length of 752bp (Table S3).
3.4 Taxonomic composition of colonic microbiota in the four groups
The microbial composition of the colonic contents was discussed in order to understand the effects of low and high dietary fiber on the intestinal microbiota of these two breeds of pigs. The principal co-ordinates analysis(PCoA) revealed that differences in the composition of colonic microbiota between Ding’an pigs and DLY pigs (Fig. S1B). At the phylum level, Firmicutes was most abundant in all four groups, followed by Bacteroidetes and Proteobacteria (Fig. 2A). However, Bacteroidetes and Proteobacteria showed different trends in abundance in Ding’an pigs and DLY pigs. The abundance of Bacteroidetes in Ding’an pigs was higher than that in DLY pigs in both the low and high dietary fiber groups (Fig. 2D, high dietary fiber : P < 0.01, low dietary fiber : P > 0.05). In addition, the abundance of Spirochaetes was higher in the DLY pigs than Ding’an pigs (high dietary fiber : P > 0.05, low dietary fiber : P < 0.05). At the genus level, the dominant genera of all groups are Prevotella, Streptococcus, Lactobacillus, Clostridium, Bacteroides, Roseburia, Ruminococcus, Eubacterium, Treponema and Faecalibacterium (Fig. 2B). The abundance of Prevotella in Ding’an pigs was higher than that in DLY pigs in the high dietary fiber groups(Fig. 2E).We also calculated the ratio of Prevotella to Bacteroides, although it was not statistically significant, the ratio was higher in Ding’an pigs (Fig. 2F). We used LEfSe to identify biomarkers with significant differences between the four groups. Seven genera significantly enriched in the DAH group, with Limosilactobacillus having the highest LDA score, which is a well-known probiotic. The largest number of biomarkers was found in the DLYH group, but most of these genera belong to pathogens (Fig. S2A). Meanwhile, we found that half of the top ten species in terms of abundance belong to Prevotella, such as Prevotella copri, Prevotella sp P2 − 180, Prevotella sp P5 − 92 (Fig. 2C). There were 20 biomarkers in the DAH group at species level. Notably, species that have been reported to be positively correlated with dietary fiber addition were significantly enriched, such as Eubacterium rectale, Blautia obeum, Bacteroides acidifaciens. Eighteen biomarkers were enriched in the DLYH group. In conformity with our expectations, most of these species were pathogenic microorganisms such as Clostridium baratii, Enterobacter hormaechei, Clostridium perfringens, Klebsiella variicola (Fig. S2B). In conclusion, there were significant differences in the colonic microbial composition among the four groups, especially between the two breeds of pigs.
3.5 Effect of feed factor on the colonic microbiota and metabolites
In order to understand the effect of dietary fiber on gut microbial composition, we conducted difference analysis on DAH vs. DAL group, DLYH vs. DLYL group, respectively. We found that there were differences in microbial composition between DAH and DAL group (Fig. S1B). The results showed that there were 67 differential species in DAH vs DAL group and 45 differential species in DLYH vs DLYL group (Fig. S3). They share four differential species, which are Bacteroides pectinophilus CAG437, Clostridium baratii, Clostridium sp CAG465 and Clostridium sp CAG678 (Fig. 3A). There were 63 species specifically altered in Ding’an pigs, and 31 of which showed increased abundance after increasing dietary fiber content, such as Limosilactobacillus reuteri, Firmicutes bacterium CAG882, Intestinibaculum porci and Prevotella pectinovora (Fig. 3B). We speculated that the up-regulation of the abundance of these species may allow Ding’an pigs to adapt higher dietary fiber content. Then we compared the effects of different dietary fiber content on intestinal microbial function. Seventeen differential pathways were enriched in Ding’an pigs, while twelve differential pathways were present in DLY pigs (Fig. 3C-D). These two breeds of pigs shared three differential pathways, pantothenate and CoA biosynthesis, histidine metabolism and vitamin B6 metabolism (Fig. 3E). Interestingly, these three pathways showed completely contrary trends in the two breeds, with abundance of these pathways up-regulated in Ding’an pigs and down-regulated in DLY pigs after high dietary fiber feeding. To further understand the effect of different fiber contents on the metabolites of Ding’an pigs and DLY pigs, we performed untargeted metabolomics analysis of the colonic contents. The OPLS-DA model showed a separation of metabolite profiles for the four groups. According to the results of the substitution test, the model's Q2 = 0.596 (P < 0.05) and R2Y = 0.991 (P < 0.05) values show that it has good explanatory and predictive power (Fig. S4). Compared with the DAL group, we found that 2-oxoadipic acid and numerous plant metabolites, including ginsenoside, genistein, coumestrol, 3-o-methylgallic acid, gentisic acid, catechin pentaacetate, tricin, pectolinarigenin, soyasaponin i were enriched in DAH group (Fig. 3F). Interestingly, compared with the DLYL group, we found a large number of dipeptides and tripeptides with increased abundance in DLYH group (Fig. 3G). Then we examined the mRNA expression of several inflammatory factors, including IFNG, IL-8, TNF-α, IL-18, IL-26 (Fig. S5). The mRNA expression of IL-8 was significantly decreased in Ding’an pigs feed on high dietary fiber, and the expression of other inflammatory factors also showed a decreasing trend.
3.6 Effect of breed factor on the colonic microbiota and metabolites
In order to investigate the differences in the composition and function of intestinal microorganisms among different breeds of pigs, we conducted difference analysis on DAH vs. DLYH group, DAL vs. DLYL group, respectively. The microbial composition of Ding’an pigs and DLY pigs was different in both the low-fiber and high-fiber diets (Fig. S1B). There were 156 differential species in the DAH vs. DLYH group, and 146 differential species in the DAL vs. DLYL group (Fig. S6). We found that fourteen species, both high and low dietary fiber, were more abundant in Ding’an pigs than in DLY pigs (Fig. 4A-B). We thought that these species are responsible for the higher apparent digestibility of dietary fiber in Ding’an pigs. We assessed the function of microorganisms by KEGG annotation. There were 41 and 31 differential functions in the high and low dietary fiber groups, respectively (Fig. 4C-D). Notably, some amino acid metabolism pathways, such as histidine metabolism, arginine biosynthesis, phenylalanine, tyrosine and tryptophan biosynthesis, were significantly enriched in DLY pigs when feed on low dietary fiber. However, amino acid metabolism pathways such as glycine, serine and threonine metabolism, lysine biosynthesis were represented in Ding’an pigs after increasing the dietary fiber content. Veen diagram showed the common enrichment pathways in Ding’an pigs in both comparisons (Fig. 4E). Among them, thiamine metabolism, riboflavin metabolism, arginine and proline metabolism and other metabolism pathways were significantly enriched. We displayed some metabolites that were specifically altered in the high dietary fiber group by heatmap (i.e., not changed in low fiber feeding, Fig. 4F). We found that several amino acid metabolites, including N-acetylserine, N-acetylhistamine, 1-methylhistamine, histamine, hydroxyisocaproic acid, 2-Isopropylmalic acid and guanidoacetic acid were enriched in the DAH group. In addition, two cholic acids and four l-carnitines were also found to be enriched in DAH group. Eight fatty acids (docosahexaenoic acid, traumatic acid, undecanedioic acid, 3,3-dimethylglutaric acid, oleic acid, 3-hydroxyvalproic acid, docosapentaenoic acid, butanoic acid) were enriched in DLYH group. We also identified two tryptophan metabolites (kynurenic acid and serotonin) were enriched in DLYH group. For the differential metabolites, we further conducted pathway enrichment analysis, and we discovered that several metabolic pathways, including cortisol synthesis and secretion, starch and sucrose metabolism, nucleotide metabolism, and histidine metabolism, were enriched (Fig. 4G). Notably, the metabolites enriched in histidine metabolism were all upregulated in the DAH group, suggesting that the pathway was activated in DAH group.
3.7 Cooccurrence networks of intestinal microbiota
The interaction between species in the colon was analyzed by co-occurrence network (Fig. 5A). The figure only show species that have interactions with more than two other species. We discovered that many of the differential bacteria identified with LEfSe were at the core of the co-occurrence network and interact with many other bacteria, such as I. porci, P. sp.CAG:520, S. ruminantium, E. rectale, etc. Among the species screened for feed factor and breed factor, the P. sp. CAG:520 was co-enriched. We found that P. sp. CAG:520 was positively correlated with many fiber-degrading bacteria, such as P. copri, E. rectale and S. ruminantium (Fig. 5B).
3.8 Correlation analysis of phenotype, microbiota and metabolites
We then analyzed correlations between important differential species, metabolites and host phenotypes. We found a strong correlation between apparent digestibility of dietary fiber and differential species or metabolites(Fig. 6A-B). In agreement with our expectation, many fiber-degrading species such as R. torques, E. rectale and Prevotella spp. were positively correlated with apparent digestibility of dietary fiber. Strong correlation was also observed between these fiber-degrading species and metabolites(Fig. 6C). Plant metabolites, including ginsenoside f1, soyasaponin i, ginsenoside rg6, etc., showed positive correlations with almost all of these species. Furthermore, amino acid metabolites, including N-acetylserine, N-acetylhistamine, 2-isopropylmalic acid, guanidinoacetic acid, etc., are also positively correlated with these species. Glucose 1-phosphate, oleic acid and docosapentaenoic acid were negatively correlated with these species.
3.9 Increasing fiber content can improve the absorption of SCFAs in Ding’an pigs
We detected the relative concentrations of two SCFAs, propionic acid and butanoic acid, through the untargeted metabolomics analysis (Fig. 7A-B). After increasing dietary fiber, we found that the concentration of butanoic acid significantly decreased in Ding’an pigs (P < 0.05), and the concentration of propionic acid also showed a decreasing trend. Although the difference in DLY pigs was not statistically significant, both of these SCFAs showed an upward trend after increasing dietary fiber. However, we found that some well-known propionate and butyrate-producing bacteria were most abundant in the DAH group (Fig. S7). Then we examined the mRNA expression of genes related to the absorption and sensing of SCFAs, including SLC16A1, FFAR2, FFAR3, PYY, GCG (Fig. 7C-D). The mRNA expression of SLC16A1, PYY and GCG was significantly increased in Ding’an pigs feed on high dietary fiber (P < 0.01). However, in DLY pigs, the dietary fiber content had no impact on the expression of these five genes.