Full-length 16S rRNA gene amplicon analysis of gut microbiota in pigs fed with different diets in growing and finishing stages

Abstract

The present study utilized full-length 16S rRNA gene sequencing to investigate the impact of dietary protein content on the composition and function of gut microbiota, and to analyze the gut microbiota of pigs in the growing (30 kg) and finishing (120 kg) stages under different feeding conditions. The results indicated that the gut microbiota was significantly different between pigs fed high- and low-protein diets. Comparing fecal samples from pigs at 30 and 120 kg, pigs at 30 kg showed a significant increase in the relative abundance of Clostridium butyricum, whereas at 120 kg, the abundance of Lactobacillus reuteri and Lactobacillus johnsonii decreased. To access the functional profiles and metabolic pathways based on amplicon sequence variants (ASVs), the microbiome of the 120 kg exhibited significant enrichments in Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways related to metabolism-related category, including Alanine, aspartate and glutamate metabolism, Tyrosine and Thiamin metabolism, and Inositol phosphate metabolism. Meanwhile, analysis using the MetaCyc database showed that the metabolic pathways of the 30 kg group were significantly distinct when compared to the 120 kg of fecal samples. Overall, the findings indicated that the gut microbiota composition and function in the 30 and 120 kg fecal samples were markedly shaped by different dietary protein levels.

Introduction

Domesticated pig is an economically important food source. Global demand for pig products has increased in recent decades. According to USDA, Livestock and Poultry: World Markets and Trade, global pork production is expected to reach 114.1 million tons in 2023. Modern pig production utilizes sustainable, efficient, and cost-effective methods to raise pigs. To maximize economic return, various strategies were applied in pig growth stages.¹ One feeding strategy is to implement a low-protein diet during the finishing phases. Since dietary crude protein is the most expensive nutrient in pig diet, reducing its level becomes especially important as feed intake increases during the finishing phases.² Additionally, maintaining high feed efficiency is crucial for ensuring farm profitability, as feed costs represent more than 75% of overall costs in pork production.³ Feed efficiency in pigs is influenced by several factors, including pig characteristics (e.g., species, age, and health), nutrient composition, feeding phases, ambient temperature, and space allowance.^4,5 Previous studies showed that the gut microbiome of pigs with higher feed efficiency had a greater abundance of bacteria that produce short-chain fatty acids (SCFAs), indicating that the enzymatic activities of the microbiome significantly affect the regulation of feed efficiency.^6–9

Many studies demonstrated that the composition of the gut microbiota of pigs positively impacts pigs’ reproduction, immunity, metabolism, growth rate, and meat quality.¹⁰ In pigs, the diverse gut microbiota, particularly mutualistic microbes, is increasingly recognized for its many benefits to gut structure, health, and functionality. Specifically, the different metabolic activities and their metabolites significantly impact swine nutrition and performance.¹¹ A previous study indicates that variations in gut microbiota are significantly linked to growth performance in pigs, playing a crucial role in enhancing piglet health and reducing antibiotic use.¹² Patil et al.¹³ reviewed several studies on pig intestinal bacteria and indicated that the richness and diversity of these bacteria can be regarded as significant indicators of pig health.¹³ Moreover, the gut bacteria produce various metabolites, including SCFAs, bile acids, indoles, polyamines, phenols, and vitamins. These metabolites participate in important physiological processes, such as regulating intestinal health, immune function, and metabolism.¹⁴

Diet is a key factor affecting the composition of the gut microbiota. Various studies have shown that the gut microbiome can be shaped by different types of dietary fiber in pig diets.¹³ Typically, a diet high in dietary fiber is advantageous for the growth of bacteria with the ability to ferment fiber compared to other species.¹⁵ Moreover, growing pigs fed with rapeseed meal showed a greater abundance of Bacteroidota compared to those fed with soybean meal.¹⁶ The intestinal microbiota of pigs underwent extensive changes at different growth phases or following dietary changes. Frese et al.¹⁷ reported that the gut microbiome in young pigs was significantly affected by the composition of dietary glycans, resulting in varied functional abilities of the microbiome before and after weaning. In addition, dietary fatty acid composition has been observed to influence the microbial communities in pigs.^18,19 Studies have demonstrated that polyunsaturated fatty acids (PUFAs) can modulate the growth performance and pork quality of both Chinese local fatty breed pigs and lean breed pigs.^20,21 Dietary crude protein (CP) is also considered a major dietary factor regulating the gut microbiota composition. Previous studies have investigated the impact of dietary protein on microbial composition and health in pigs.^22,23 Protein is an essential nutrient for the health and growth of livestock throughout various stages of life processes, and it is one of the most expensive ingredients in livestock diets.²⁴ Therefore, reducing the protein content in pig diets has become a contemporary topic of interest, offering the potential to lower production costs and mitigate environmental impact. However, the interaction of this nutrient strategy on the gut microbial composition and intestinal health is still unclear. Several studies have indicated that even with the addition of essential amino acid supplementation, low-protein diets can have a negative impact on growth performance of pigs.^25–27 Therefore, the aims of this study were to examine the effects of dietary protein level on the gut microbiome and metabolome in growing and finishing stages using full-length 16S rRNA gene sequencing, to provide a better understanding of dietary interventions on the microbiota and contribute to the development of strategies to overcome the limitations associated with low-protein diets in commercial swine production.

Materials and methods

Animal experiment and sampling

Thirty-six crossbred pigs averaging 84 d of the age, including eighteen LYD (Landrace × Yorkshire × Duroc) and eighteen LD (Landrace × Duroc), with an initial body weight of 30.5 ± 2.2 kg were raised from Taitung Animal Propagation Station. Before the trial, the pigs were acclimated to the experimental environment with concrete floor and natural ventilation for seven days. During the trial, three pigs were randomly allocated to each pen (2.7 × 3.4 m) with crossbred scheme sex as block. Diets based on corn, soybean meal, and fish meal were prepared. Piglets from weaning to 30 kg were fed diet 1, and pigs from 30 to 120 kg were fed diet 2 (Table 1). Diet 1 was formulated without supplemental amino acid and the levels of crude protein, lysine, methionine, threonine, tryptophan, and metabolizable energy were 18.2, 1.03, 0.70, 0.35, 0.21%, and 3290 kcal/kg, respectively. In diet 2, crude protein level was reduced by partial substitution soybean meal and fish meal with wheat bran. Meanwhile, diet 2 was supplemented with L-lysine, L-threonine, DL-methionine, and L-tryptophan and the levels of crude protein, lysine, methionine, threonine, tryptophan, and metabolizable energy were 13.5, 0.97, 0.64, 0.28, 0.17%, and 3550 kcal/kg, respectively. The supplemental levels of amino acids were referred to NRC (v.2012) recommendations for all nutrient requirements. Besides, the energy level of diet 2 was referred to Myer and Bucklin²⁸ for improving the growth performance under high ambient temperature. Animals were fed ad libitum access to feed and water.

Table 1.

Composition of the experimental diets.

Ingredient (g/kg)	Stage
	30 kg (Diet 1)	120 kg (Diet 2)
Corn	677.5	676.5
Soybean meal	190.0	130.9
Wheat bran	0.0	50.0
Limestone	8.0	8.2
Dicalcium phosphate	16.0	11.3
Fish meal	50.0	20.0
Choline chloride	1.0	1.0
Molasses	0.0	15.0
Salt	5.0	5.0
Skim milk	20.0	0.0
Whey	20.0	0.0
Vitamin premixa	1.0	1.0
Mineral premixb	1.5	1.5
Blend oilc	10.0	74.0
L-lysine ‧ HCl (78%)	0.0	3.7
DL-methionine (99%)	0.0	0.7
L-threonine (98.5%)	0.0	1.0
Tryptophan (98%)	0.0	0.2
Total	1000.0	1000.0
Calculated composition
Metabolizable energy (kcal/kg)	3290	3550
Crude protein (%)	18.2	13.5
Crude fiber (%)	2.7	2.6
Lysine (%)	1.03	0.97
Threonine (%)	0.70	0.64
Methionine (%)	0.35	0.28
Tryptophan (%)	0.21	0.17

^aVitamin supplied the following per kilogram of diet: vitamin A, 6000 IU; vitamin D₃, 400 IU; vitamin E, 20 IU; vitamin K₃, 2 mg; vitamin B₁, 2.6 mg; vitamin B₂, 2 mg; Niacin, 30 mg; Pantothenic acid, 30 mg; Pyridoxine, 3 mg; vitamin B₁₂, 0.6 mg; Biotin, 0.2 mg.

^bMineral supplied the following per kilogram of diet: Fe (FeSO₄·7H₂O, 20.09% Fe), 80 mg; Cu (CuSO₄·5H₂O, 25.45% Cu), 5 mg; Mn (MnSO₄·H₂O, 32.49% Mn), 6 mg; Zn (ZnSO₄, 80.35% Zn), 45 mg; I (KI), 0.2 mg; Se (NaSeO₃, 45.56% Se), 0.1 mg; Co (CoSO₄·H₂O, 32% Co), 0.35 mg.

^cBlend oil consisted of one-third Soybean oil, one-third Tallow, and one-third Linseed oil.

Fecal sampling

Thirty-six fecal samples were collected from defecation immediately at 84 days (pigs reached to 30 kg) and 179 days (pigs reached to 120 kg) of experiment, respectively. The feces were collected into sanitized bags directly to avoid contamination and stored at −80 °C after sampling. The fecal sample for each sample was at least 100 g or more.

DNA extraction, amplicon generation, library construction, and sequencing

Genomic DNA of feces samples of each pig were extracted using Qiagen QIAamp^® PowerFecal^® kit (Qiagen Inc., CA, USA), according to the manufacturer instructions. The DNA concentration and quality were determined by NanoDrop spectrophotometer (NanoDrop Technologies). The DNA samples were submitted to Genomics Bioscience Technology Co., Ltd. (Taipei, Taiwan) for 16S full-length sequencing. For 16S rRNA gene sequencing, the primers 27 F (5′-AGRGTTYGATYMTGGCTCAG-3′) and 1492 R (5′-RGYTACCTTGTTACGACTT-3′) were used to amplify the V1 through V9 hypervariable regions of the 16S rRNA region. All PCR reactions were carried out in 25 μL reactions with KAPA HiFi HotStart 2× ReadyMix PCR Reagent (Kapa Biosystems, MA, USA), 0.375 μM of forward and reverse primers, and ∼1 ng DNA template. The amplification was performed with 27 cycles of denaturation at 95 °C for 30 s, annealing at 57 °C for 30 s, and elongation at 72 °C for 30 s, followed by 5 min extension at 72 °C. The PCR products were examined on 0.8% agarose gels and Qubit 2.0 Fluorometer (Thermo Scientific, MA, USA). The qualified PCR products were mixed in equal amount and purified via AMPure PB beads (PacBio, CA, USA). Following the manufacturer’s protocol (PacBio, CA, USA), the pooled 16S amplicons were used as template to construct the SMRTbell library using SMRTbell Express template prep kit 2.0. After the library construction, the amplicon libraries of each experimental group were sequenced on PacBio Sequel IIe sequencer by amplifying the V1-V9 domain of the 16S rRNA gene. The subreads were transferred into circular consensus sequencing (CCS) reads at least 0.99 accuracy and three passes. The CCS reads were demultiplexed by SMRTLink analysis pipeline version 9.0.

Bioinformatics and statistical analysis

DNA sequences from both ends of 27 F-1492R primers were trimmed by Cutadapt v3.3²⁹ with the default setting. DADA2 v1.12³⁰ was used to filter noisy sequence out, merge paired-end reads, and remove chimera to extract amplicon sequence variants. The QIIME2 pipeline³¹ was used to perform the analysis, and the quality filtering processes were run with the default setting. The alpha diversity was analyzed by a species richness estimator, including Chao1 and Fisher’s alpha.³² The species evenness estimator, including Shannon and Simpson,³³ were performed as well. The beta diversity was analyzed using UniFrac distance matrices with hierarchical clustering,^34–36 principal component analysis (PCA), and principal coordinate analysis (PCoA) with Bray-Curtis similarity matrix.³⁷ Non-metric Multi-dimensional Scaling (NMDS) ordination technique was employed in conjunction with correlation analysis to analyze the huge data sets for identifying the relationships between groups.³⁸

The process of identifying taxonomic biomarkers with discriminative properties involved the use of linear discriminant analysis (LDA) effect size (LEfSe), with a default cutoff LDA log score.³⁹ This was followed by the Kruskal-Wallis test, where a Wilcoxon test cutoff of p < 0.05 was applied. An online implementation of LEfSe, with a graphical interface was integrated in the Galaxy framework (http://huttenhower.sph.harvard.edu/galaxy). To access the functional pathways in KEGG Orthology (KOs) database and MetaCyc pathways database (https://metacyc.org/) using 16S rRNA gene sequencing data, we carried out PICRUSt2 (Phylogenetic Investigation of Communities by Reconstruction of Unobserved States) analysis, which was introduced by Langille et al.⁴⁰ We also assessed the alpha diversity and relative abundance comparison using a two-tailed Welch’s t-test. The STAMP (Statistical Analysis of Metagenomic Profiles) application (v. 2.1.3)⁴¹ was employed to analyze the statistical differences in microbiota taxa between two groups of samples. The hierarchical structures of taxonomic classifications were assessed by using R package Metacoder (https://github.com/grunwaldlab/metacoder). The Spearman’s rank correlation⁴² was performed by using R language (https://www.r-project.org/) to assess the structure of microbial population between 30 and 120 kg fecal samples.

Results

In this study, 16S rRNA gene sequences from 72 fecal samples of crossbred pig including LYD and LD were collected in two stages (pigs reached to 30 and 120 kg, respectively) (Fig. 1). The fecal samples were used to determine the influence of dietary protein (30 kg: 18.2% CP diet and 120 kg: 13.5% CP diet) on the microbiota composition. In total, 2,538,998 contiguous sequences were obtained from PacBio Sequel IIe sequencer. After quality control, a total of 1,131,957 high quality contiguous sequences were retained to detect 7925 Amplicon Sequences Variants (ASVs).

Figure 1.

Schematic depiction of the experiment and comparison of swine gut microbiome between 30 and 120 kg fecal samples.

Microbial diversity of 30 and 120 kg pig based on 16S rRNA gene data

When compared with the 30 kg stage, the diversity of microbial communities in the fecal samples increased at the 120 kg stage, as measured using Shannon, Simpson, and Chao1 diversity indices. The statistical significance was found between 30 and 120 kg groups (Table 1). The mean values of Shannon, Simpson, and Chao1 diversity indices identified in the 30 and 120 kg groups ranged from 5.95 to 6.59, 0.96 to 0.97, and 239.25 to 427.56, respectively (Table 2). Compositional differences between 30 and 120 kg groups were assessed to compare the gut microbial compositions. According to the Non-metric multidimensional scaling (NMDS), Principal Co-ordinates Analysis (PCoA), Analysis of similarities (ANOSIM), and Permutational multivariate analysis of variance (PERMANOVA) analyses, the dietary composition exhibited the most important effects on the microbiota composition. The NMDS results showed two clear clusters at ASVs level (Fig. 2), suggesting the dietary effect on the pig microbiota composition. According to PCoA, the PCoA1 and PCoA2 explained about 46% of the total variability of the relative abundance data. The significant differences in gut microbiome between 30 and 120 kg groups were also observed from the ANOSIM (R^2 = 0.67 and p-value = 1E-04) and PERMANOVA results (R^2 = 0.254 and p-value = 1E-04), indicating a significant effect of different stages (30 and 120 kg) at ASVs levels.

Table 2.

Diversity index of the pig gut microbiota in 30 and 120 kg groups.

Diversity index	30 kg (mean ± std) (n = 36)	120 kg (mean ± std) (n = 36)	p-Value
Shannon	5.95 ± 0.72	6.59 ± 0.74	2.91E-04**
Simpson	0.96 ± 0.03	0.97 ± 0.03	5.39E-02*
Chao1	239.25 ± 108.61	427.56 ± 125.92	2.06E-07**

Two-sided Welch’s t-test was used to verify the difference in the relative abundance of microbial taxa in 30 and 120 kg groups.

**p-Value < 0.01; *p-value < 0.1.

Figure 2.

Nonmetric multidimensional scaling (NMDS) analysis based on two different stages (30 vs. 120 kg).

Comparison of taxonomic classification of 30 and 120 kg fecal samples

Comparisons of the relative abundances of the gut microbiota compositions between 30 and 120 kg fecal samples at the phylum, order, genus and species levels are shown in Fig. 2. At the genus level, the bacterial sequences from the 30 kg samples were composed of the genus Lactobacillus (13.68%), Clostridium sensu stricto (9.46%), Terrisporobacter (7.86%), Streptococcus (6.50%), Prevotella (5.91%), and ∼40% are unclassified (division Bacteria) (Fig. 3). On the other hand, 120 kg fecal samples consisted of genus C. sensu stricto 1 (16.04%), Streptococcus (12.94%), Terrisporobacter (4.53%), Lactobacillus (2.75%), Muribaculaceae (2.15%), and around 45% of the total sequences analyzed in the 120 kg samples was unclassified (division Bacteria). In comparison with 30 and 120 kg fecal samples, the microorganism populations of the genus C. sensu stricto 1 increased by 70%, with an average of 9.46% in 30 kg samples to 16.04% in 120 kg samples. The population of genus Streptococcus also significantly increased by two times (p-value < 0.05) when shifting from 30 to 120 kg stage. The changes resulted in a significant average decrease of 86% in the populations of genus Lactobacillus, including Lactobacillus amylovorus, Lactobacillus reuteri, and Lactobacillus johnsonii, in 120 kg fecal samples compared to the populations observed in 30 kg fecal samples. At the family level in 30 kg samples, family Lactobacillaceae, Clostridiaceae, Prevotellaceae, Peptostreptococcaceae, and Streptococcaceae were consisted of 46.57% of ASVs, while 43% were still unclassified (belonging to division Bacteria). For 120 kg samples, the order of most abundant family was slightly different, with family Clostridiaceae (16.12%), Streptococcaceae (12.94%), Peptostreptococcaceae (5.81%), Prevotellaceae (3.75%), and Lactobacillaceae (2.76%), accounting for 41.37% of the total population. Notably, the family Lactobacillaceae (13.68%) was the most abundant population in 30 kg samples, while the family Clostridiaceae (16.12%) was the most abundant population in 120 kg samples.

Figure 3.

Taxonomic classification of the 16S rRNA gene sequence at the (A) phylum, (B) order, (C) genus, and (D) species levels for the 30 and 120 kg fecal samples.

Changes of gut microbiota in 30 and 120 kg fecal samples

LEfSe was used to determine whether specific bacterial taxa were differentially enriched between the 30 and 120 kg fecal samples. A phylogenetic cladogram was generated. The cladogram presenting the taxonomic hierarchical structure of the fecal microbiota from class to family indicated significant differences in phylogenetic distributions between the microbiota of 30 and 120 kg samples (Fig. 4A). Using a logarithmic LDA score cutoff with default setting, this study identified genus Lactobacillus, Megaspharea, Alloprevotella, Solobacterium, Holdemanella, Catenispharea, and Prevotella were significantly over-represented in the feces of 30 kg samples, whereas genus Clostridium, C. sensu stricto, Romboutsia, Bifidobacterium, Oscillospira, Streptococcus, and Turicibacter were enriched in 120 kg samples (Fig. 4B), showing significant differences in fecal microbiota between 30 and 120 kg samples. The hierarchical structure of taxonomic classifications were assessed based on 30 and 120 kg gut microbiota samples (Figs. 5 and 6). Results also indicated that the gut microbiota between 30 and 120 kg were significantly different. The FDR corrections were used to verify the results of LEfSe for statistical analysis of metagenomic profiles (STAMP). The LEfSe results from the linear discriminant analysis (LDA) effect size revealed the microbiota was distinct between 30 and 120 kg samples, with the most differentially abundant taxa being identified (Fig. 7). Thirty kilogram samples had more abundance of Lactobacillus, Prevotella, Terrisporobacter, and Megasphaera compared with 120 kg samples. Alternatively, 120 kg samples had more abundance of Romboutsia, C. sensu stricto, Streptococcus, and Muribaculaceae. At species level, a relatively high richness of Lactobacillus was observed, including L. reuteri and L. johnsonii, while 120 kg samples contained higher abundant levels of Clostridium butyricum and species-unknown Clostridium. On the other hand, the spearman’s rank correlation was performed to evaluate the microbial community structure between 30 and 120 kg samples. The top thirty abundance genera between 30 and 120 kg samples were involved in the analysis, showing that the significant correlation was observed between Muribaculaceae, Romboutsia, Turicibacter, and C. sensu stricto, uncultured Eubacterium WCHB1-41, Parabacteroides, and Prevotellaceae UGC-001 (Fig. 8).

Figure 4.

The microbial communities of 30 and 120 kg samples were compared. (A) a phylogenetic cladogram generated from LEfSe analysis showed the taxonomic association between the microbiome communities of 30 and 120 kg samples, with each node representing a specific taxonomic type. (B) The LEfSe method was used to rank significantly different genera based on the log LDA scores of the two groups, using the non-parametric factorial Kruskal-Wallis sum-rank test. Default setting of featured LDA scores was employed.

Figure 5.

The heat trees of 30 kg samples. Each node represents a taxon used to classify OTUs/ASVs, while the connection determines the position of this node in the entire taxonomic hierarchy. The diameter of the node is proportional to the number of OTUs/ASVs classified into that category, while the width of the connection is proportional to the number of sequences.

Figure 6.

The heat trees of 120 kg samples.

Figure 7.

Microbial taxa were identified via LEfSe analysis based on 30 and 120 kg samples. Bar chart showing the taxonomic difference (corrected q-value < 0.05) at (A) family, (B) genus, and (C) species level between 30 and 120 kg samples. The data were analyzed by two-sided Welch’s t-test and subsequently filtered for false discoveries utilizing the Benjamini-Hochberg method. Only objects with q-values below 0.05 were displayed in the figure.

Figure 8.

The correlation between the top thirty abundant genera of microbiota population in 30 and 120 kg samples.

Metabolism-related KEGG and MetaCyc pathways were identified in pig microbiome

The PICRUSt2 analysis was performed to predict the functional profiles and metabolic pathways in two stages of pig gut microbiota using 16S rRNA gene sequencing. The top fifteen KEGG pathways and MetaCyc pathways are shown in Figs. 9 and 10, respectively. As shown in Supplementary Table 1, 189 KEGG pathways were estimated to have been significantly affected by two different stages (30 and 120 kg samples) (p-value < 0.05). The microbiome of 30 and 120 kg samples exhibited significant enrichment in pathways related to the Atrazine degradation, Inositol phosphate metabolism, Ubiquinone and other terpenoid-quinone biosynthesis, Pores ion channels, Thiamin metabolism, and Tyrosine metabolism, etc. On the other hand, MetaCyc pathway database was assessed, which was used to estimate the primary and secondary metabolism associated with different stages of microbiome (Supplementary Table 2). Results showed that 214 MetaCyc pathways were significantly affected by two different stages (30 and 120 kg samples) (p-value < 0.05). The microbiome exhibited enrichment in pathways related to the superpathway of (R,R)-butanediol biosynthesis, superpathway of UDP-N-acetylglucosamine-derived O-antigen building blocks biosynthesis, aromatic biogenic amine degradation (bacteria), allantoin degradation to glyoxylate III, isopropanol biosynthesis, superpathway of fucose and rhamnose degradation, and fucose degradation, etc. Significant differences were observed in metabolism-related pathways, suggesting that the different protein dietary potentially affects the metabolic pathways of pigs by changing the pig gut microbiome.

Figure 9.

Top 15 KEGG pathways comparison based on 30 and 120 kg samples.

Figure 10.

Top 15 MetaCyc pathways comparison based on 30 and 120 kg samples.

Discussion

Growing pigs have a significant demand for essential nutrients. To optimize their nutrient intake and promote rapid growth, they are provided with feed that is rich in protein and other nutrients. Previous studies experimented with various additives and ingredients in the diets of weaned and growing pigs to evaluate their effects on growth performance and changes in the gut microbiome.^1,43–45 For finishing pigs, it is common to employ a low-protein diet supplemented with synthetic amino acids to reduce feed costs and environmental impact.^1,46–48 Most studies of low-protein diets for finishing pigs have focused on production performance, slaughter performance, and meat quality. However, only a limited number of studies evaluated the effects of low-protein diets on the fecal microbiota of finishing pigs. The objective of this study was to evaluate the influence of dietary protein level on the gut microbiome and metabolome of pigs in growing and finishing stages. In the present study, pigs were subjected to different diets during the growing and finishing stages. After weaning, the pigs were given an 18.2% CP diet consisting of various protein sources, such as soybean meal, skim milk, whey, and fish meal, until they reached a weight of 30 kg. Subsequently, the pigs were transitioned to a 13.5% CP diet until they reached a weight of 120 kg. In this study, the impacts of dietary protein levels on the microbial composition in the 30 and 120 kg groups showed remarkable differences, as suggested by NMDS and LEfSe analysis. Dietary protein is vital for lean gain in pigs, as it fulfills the amino acid requirement for optimal growth performance. Previous researches indicated that dietary protein content impacted the gut microbiota, showing a positive correlation with microbial diversity.^44,49,50 This correlation is directly influenced by both the quantity and quality of the protein.⁴⁹ Additionally, Li et al.⁵¹ demonstrated that reducing the CP level from 18.3 to 15.1% during the growing phase led to a decrease in average daily gain and gain to feed, even though the calculated ratio of all essential amino acids to lysine was consistent with the NRC.⁵² Additionally, reducing the CP level from 16.3 to 13.2% in finishing phase resulted in a reduction in average daily gain and gain to feed, while improving meat quality.⁵³ Furthermore, there was a significant distinction observed in taxonomic composition among the different groups. The gut microbiota of pigs is primarily composed of Firmicutes and Bacteroidetes before and after weaning.⁵⁴ At the phylum level, Firmicutes were the most abundant phylum in both the 30 and 120 kg groups, representing 42.04 and 41.05%, respectively. Compared to the 120 kg group, pigs fed an 18.2% CP diet (30 kg) exhibited high abundances of the phyla Bacteroidota and Actinobacteriota, and low abundances of the phylum Planctomycetota. Research conducted by Guo et al.⁵⁵ indicated that the intestines of obese pigs showed a high abundance of Firmicutes and a low abundance of Bacteroidetes. It is a common strategy in the pork industry to increase the fat content in the finishing diet, as it is a cost-effective energy source. In the current study, both diets had a similar metabolizable energy; however, due to a lower protein content, the diet with 13.5% CP had a higher fat content. Previous studies indicated that the gut microbial composition is altered by a high-fat diet, leading to disruption in the balance between energy intake and expenditure.⁵⁶ The gut microbial composition in mammals is significantly influenced by various types of dietary fat.⁵⁷ For instance, medium-chain fatty acids have antibacterial properties,⁵⁸ and lipolytic bacteria utilize hydrolyzates, such as glycerol, fatty acids, or triglycerides to generate short-chain fatty acids, which serve as an energy source for both bacteria and intestinal epithelial cells.⁵⁹ At the genus level, we found that the proportions of the genera Lactobacillus and Prevotella were significantly higher in the 18.2% CP diet (30 kg), showing nearly 5 times and 4.5 times more abundance, respectively, than in the 13.5% CP diet. Lactobacillus and Prevotella are recognized as important constituents of a healthy microbiota. Lactobacillus is a well-known genus associated with pig health.⁶⁰ The enterotype driven by Prevotella has demonstrated positive correlations with animal traits, such as weight gain,⁶¹ feed efficiency,⁶² feed intake,⁶³ and incidence of diarrhea.⁶⁴ These findings suggested that Prevotella plays a significant role in influencing growth performance and disease resistance in pigs. In contrast, the high abundance of genus C. sensu stricto 1 and genus Streptococcus were observed in the 13.5% CP diet. Clostridium sensu stricto 1 is an opportunistic pathogen^65,66 that can induce intestinal inflammation and reduce the levels of short-chain fatty acids (SCFAs).⁶⁷ Genus Streptococcus comprises multiple species, some of which have been reported to have pathogenic potential including Streptococcus suis, Streptococcus dysgalactiae, Streptococcus porcinus, and Streptococcus pyogenes.⁶⁸ Moreover, in this study, comparisons of the top 15 KEGG and the top 15 MetaCyc pathways between diets were conducted. We observed that the low-protein diet group exhibited higher abundance in all 15 KEGG pathways, whereas 3 MetaCyc pathways showed higher abundance in the high-protein diet group. The KEGG analysis showed that the pathways related to amino acid metabolism were enriched in low protein diet group, as well as the pathway related to zeatin and ubiquinone, and other terpenoid-quinone biosynthesis. Comparison of MetaCyc pathways between diets, the pathways including Calvin-Benson-Bassham cycle, L-1,2-propanediol degradation, and hexitol fermentation to lactate, formate, ethanol, and acetate were enriched in high protein diet group. The pathways defined in the MetaCyc and KEGG databases revealed that the gut microbiome of the high-protein and low-protein diet groups showed different abundances.

Conclusion

In this study, we characterized the variances in fecal microbiota composition between two diets with distinct protein contents, examined at stages corresponding to 30 and 120 kg in pigs. Significant differences in gut microbial composition were observed between the two stages, suggesting that the microbial community was significantly shaped by dietary protein. Decreasing dietary protein levels with the supplement of fat resulted in fewer probiotics, such as Lactobacillus, suggesting that a low protein-high fat diet may affect gut health. Additionally, the associated gut microbiota and metabolism pathways were predicted. The present study uncovered the influence of prevalent feeding practice in growing and finishing pigs on the gut microbiome and metabolome, with potential implications for intestinal health. Overall, our findings offer valuable insights for the pork industry and may assist in adjusting feeding strategies or optimize feed formulas.

Funding Statement

This study was supported by Taiwan Livestock Research Institute, Ministry of Agriculture, Taiwan.

Ethical approval

The animal research protocol was approved and instructed by Institutional Animal Care and Use Committee of Taitung Animal Propagation Station, Livestock Research Institute, Council of Agriculture, Taiwan (IACUC 111-09). All experimental treatments and animal husbandry were followed the Animal Protection Act issued by Council of Agriculture, Taiwan and fourth edition of the Guide for the Care and Use of Agricultural Animals in Research and Teaching.

Consent for publication

Not applicable.

Author contributions

HSW, SYS, YLH, and HYT conducted the experiments. SYS, HSW, and HYT prepared the manuscript, and HSW, CCC, and HYT conceived the investigation. All authors have read and agreed to the manuscript.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

All data generated or analyzed during this study are included in this published article and supplementary files. Further inquiries can be directed to the corresponding author.