This study was approved by the Experimental Animal Ethics Committee, Chongqing Academy of Animal Science, under reference number XKY-No. 20210606.

To investigate differences in the gut microbiota of SPF pigs and their interaction with the host, we selected Bama females provided by Chongqing Academy of Animal Science (Chongqing, China) that were disease-free, unrelated, in good physical condition, and normal in appearance. These females were pregnant and tested negative for the placental transmission diseases, African swine fever, porcine pseudorabies, porcine parvovirus disease, and porcine reproductive and respiratory syndrome, and for other serum-specific pathogens, such as African swine fever, foot-and-mouth disease, Japanese encephalitis, porcine circovirus, swine influenza, and porcine brucellosis.

Three days before the expected delivery date, piglets were delivered through aseptic cesarean section. Six female piglets were randomly selected as subjects for this study (Figure 1). These piglets were reared following the SPF pig model construction and rearing standards (Zhang et al., 2023) on the standardized aseptic pig breeding platform at the Institute of Biotechnology, Chongqing Academy of Animal Science. In summary, SPF piglets were fed with sterilized milk powder (0–21 days) and pelleted feed (21–42 days) diluted 1:4 with sterile water and sterilized using Co60 γ radiation. During the feeding process, milk powder, feed, and water were introduced into the isolator through a transfer tube and disinfected with 1% peracetic acid to prevent microbial contamination.

After feeding for 42 days, in order to comprehensively study the gut microbial characteristics of SPF pigs and their interactions with the host, we collected a total of 42 GIT content samples and 28 intestinal tissue samples from 6 SPF Bama females at the Chongqing Academy of Animal Sciences. Among them, the GIT content samples included samples from four different sites, including 6 stomach samples, 18 small intestine samples (duodenum, jejunum, and ileum), 12 large intestine samples (cecum and colon), and 6 rectum samples. The intestinal tissue samples consisted of 17 small intestine samples (duodenum, jejunum, and ileum) and 11 large intestine samples (cecum and colon) (Supplementary Tables S1, S2). GIT contents and intestinal tissue samples from the SPF pigs were collected and stored at −80°C prior to sequencing.

Total DNA was extracted from GIT content samples using a QIAamp DNA Stool Mini Kit (Qiagen, Germany), following the manufacturer’s instructions. Each sample yielded 0.2 μg of total DNA, which was used as input material for DNA library preparation. Briefly, genomic DNA samples were fragmented by sonication to an average size of 350 bp. The DNA fragments then underwent end-polishing, A-tailing, and ligation with full-length adapters for Illumina sequencing, followed by PCR amplification. The PCR products were purified using the AMPure XP system from Beverly. Next, we assessed the quality of the libraries using the Agilent 5400 system and quantified by QPCR (1.5 nM). Based on the effective library concentration and the required amount of data, qualified libraries were pooled and sent to Novogene Bioinformatics Technology Co., Ltd. (Beijing, China) for metagenomic sequencing using the Illumina NovaSeq 6000 platform with a paired-end-150 bp strategy, aiming to generate at least 10 G of raw data per sample. Forty-two samples were sequenced, resulting in approximately 448.08 G of raw data. Basic quality processing was performed on the raw data using Fastp (Chen et al., 2018) (v0.23.1) to obtain clean data. The data processing steps were as follows: if either of the paired reads contained adapter contamination, the reads were discarded; if either of the paired reads had more than 10% of bases with uncertainty, the reads were discarded; if the proportion of low-quality bases (Phred quality < 5) in either of the paired reads exceeded 50%, the reads were discarded. To remove reads that might originate from the host or food, we downloaded the reference genome sequence of Sus scrofa (Sus_scrofa.Sscrofa11.1.dna.toplevel.fa) from Ensembl and aligned the sequencing data to the host genome using BWA-MEM (Li, 2013) (v0.7.17). Subsequently, we used SAMtools (Li et al., 2009) (v1.17) to remove host reads. After this step, we obtained approximately 271.94 G of filtered data and 1.866 billion high-quality paired-end filtered reads for subsequent analysis.

The high-quality reads for each sample were assembled using MEGAHIT (Li et al., 2015) (v1.2.9) with the parameter “-min-contig-len 500,” resulting in 446,890 contigs. The assembly quality was assessed using QUAST (Gurevich et al., 2013) (v5.2.0). Next, open reading frames were predicted using metaProdigal (Hyatt et al., 2010) (v2.6.3) with the parameter “-p meta,” resulting in 1,213,609 open reading frames from which 620,325 complete genes were extracted (51.1%, partial = 00; Supplementary Figure S1). Subsequently, CD-HIT (Huang et al., 2010) (v4.8.1) was employed to remove redundancy among the predicted genes and proteins. Clustering was performed using 95% sequence identity and 90% sequence coverage, resulting in the formation of the most similar clusters. The nucleotide sequences were then translated into their corresponding protein sequences. The abundance of non-redundant genes was calculated as transcripts per million (TPM) fragments using Salmon (Patro et al., 2017) (v1.10.1).

To obtain the taxonomic composition of the gut microbiota for each sample, we used Kraken 2 (Lu et al., 2022) (v2.1.3) with the default parameters to align the translated protein sequences against its database (k2_pluspf_20231009). We calculated alpha diversity based on the Shannon index using the vegen package in R (v4.3.1), computed beta diversity distances using Bray–Curtis dissimilarity, and performed species rarefaction analysis using USEARCH (Edgar, 2010) (v10.0) (Figure 1). Differences in the GIT microbiota between SPF pigs were assessed using the Wilcoxon test. Spearman rank correlation was used to calculate the correlation coefficients of the 20 most abundant microorganisms at the species level in the GIT. Only correlations with Spearman correlation coefficient (r) ≥ 0.3 or ≤ −0.3 and p < 0.05 were retained and visualized using R (v4.3.1). Linear discriminant analysis implemented in the LEfSe (Segata et al., 2011) tool was used to identify different taxonomic groups between sample groups. We identified discriminative groups with a linear discriminant analysis score > 3 and assessed the statistical significance of abundance differences between GIT microbiota groups in SPF pigs using the Wilcoxon test.

We used GhostKOALA (Kanehisa et al., 2016) to map non-redundant protein sequences to the KEGG database for KEGG Orthology annotation. For protein Cluster of Orthologous Groups annotation, we aligned non-redundant protein sequences to the eggNOG database (v5.0.2) using the default parameters of eggNOG-mapper (Cantalapiedra et al., 2021) (v2.1.12). Additionally, we used the BLASTp option of DIAMOND (Buchfink et al., 2015) (v2.0.15) with the parameter “-e 1e-102” to align non-redundant protein sequences to the latest CAZy database (CAZyDB.07262023) for the detection of carbohydrate-active enzymes (CAZymes). To detect antimicrobial resistance genes in each microorganism, we used the RGI (Alcock et al., 2023) (v5.2.0) tool with the option “DIAMOND --include_loose” to align non-redundant protein sequences to the latest CARD database (v3.2.6) (Figure 1).

We extracted total RNA from intestinal samples using the Trizol method. After qualifying the RNA samples, eukaryotic mRNAs were enriched using Oligo(dT) magnetic beads. For prokaryotes, a mRNA purification kit to remove rRNA and enrich mRNA was used. Subsequently, we added a cleavage buffer to break the mRNA into short fragments, which was then used as a template to synthesize single-stranded cDNA using random hexamers. Next, a buffer, dNTPs, DNA polymerase I, and RNase H were added to synthesize double-stranded cDNA, which was then purified using AMPure XP beads. The cDNA then underwent end repair, A-tailing, and sequencing adapter ligation, followed by fragment size selection using AMPure XP beads. Finally, PCR amplification was performed, and the PCR products were purified using AMPure XP beads to obtain the final library. After construction, we quantified the library using Qubit 2.0 and diluted it accordingly. Subsequently, we used the Agilent 2100 system to assess the insert size of the library, ensuring that the insert fragments met the expected size. Next, we used q-PCR to accurately quantify the effective concentration of the library, ensuring library quality. After passing the quality check, the library was sequenced using DNBSEQ-T7.

The data obtained from sequencing is referred to as raw reads. Prior to reference genome alignment, we obtained clean reads by filtering out reads containing adapters (adapter contamination), reads containing unknown bases (N bases), and reads with low sequencing quality. Based on the pig reference genome and annotation files downloaded from Ensembl, we performed alignment of the clean reads using STAR (Dobin et al., 2012) (v2.7.10b), assembly using Cufflinks (Trapnell et al., 2010) (v2.2.1) and TACO (Niknafs et al., 2017) (v0.7.3), and calculation of the protein-coding potential of the assembled transcripts using CPC (Kang et al., 2017) (v2.0). Ultimately, we obtained reference genome files for mRNAs and lncRNAs, which were used for subsequent analysis. We quantified the data using Kallisto (Bray et al., 2016) (v0.49.0) to obtain the counts and TPM values for mRNAs and lncRNAs, which were used for downstream analysis. After calculating expression levels, we grouped the samples based on treatments and conditions. First, we selected expressed genes by choosing genes that were expressed (TPM > 0) in all samples. Subsequently, differential gene expression analysis was conducted using the edgeR package (Chen et al., 2016) in R (v4.3.1). We defined differentially expressed genes (DEGs) as genes with |Log₂(FoldChange)| > 1 and FDR.P (false discovery rate adjusted p-value) < 0.05. Kyoto Encyclopedia of Genes and Genomes (KEGG) (Kanehisa and Goto, 2000) and Gene Ontology (GO) (Gene Ontology Consortium, 2007) enrichment analyses were performed using the clusterProfiler package (Wu et al., 2021) (Figure 1). The gene sets used were primarily of mouse or human origin; therefore, we retained only the genes that exhibited one-to-one orthology with pigs.

To investigate potential functional interactions between DEGs in the host gut and the gut microbiota (top 20 species), we performed Spearman rank correlation analysis. We defined DEGs as genes with |Log₂(FoldChange)| > 1 and FDR.P < 0.05. Specifically, we selected the 130 most up-and down-regulated DEGs based on |Log₂(FoldChange)| (after converting human genes to pig homologs). Spearman rank correlation, p-values and significant associations (FDR.P < 0.05) between genes were calculated using the corr.test() function from the psych package in R.

We used the species annotation software, Kraken2, to classify the non-redundant protein sequences from the GIT of SPF pigs. A total of 269,251 non-redundant proteins (43.4% of total proteins) were annotated into four domains (Bacteria 89.4%, Eukaryota 8.2%, Viruses 1.3%, and Archaea 0.1%; Figure 2A). Sixty-one phyla, 113 classes, 226 orders, 470 families, 1,646 genera, and 5,421 species were annotated (Supplementary Table S3). We then performed rarefaction analysis on the 5,421 annotated species. Moreover, with increasing sequencing depth, the species rarefaction curves obtained from different GIT sites reached a saturation stage (Supplementary Figure S2A).

We then performed statistical analysis on the dominant phyla and genera in the GIT. The dominant phyla were Firmicutes, Proteobacteria, and Bacteroidetes, while the dominant genera were Lactobacillus, Limosilactobacillus, and Romboutsia (Supplementary Figure S2B). Additionally, to demonstrate better, we constructed an evolutionary tree using the selected bacteria after filtering and annotated them to 15 phyla and 146 genera. The four most abundant bacterial phyla (dominant phyla) were Firmicutes, Proteobacteria, Actinobacteria, and Bacteroidetes (Figure 2B), which is consistent with studies conducted on humans and other animals (Xiao et al., 2017; Long et al., 2022; Maiuolo et al., 2022; Yu et al., 2023). To further investigate changes in the relative abundance of microbiota along the GIT, we analyzed the genus-level microbial composition, as shown in Figure 2E. We found that Lactobacillus, Limosilactobacillus, Romboutsia, and Bacteroides were the predominant bacteria in the GIT. These bacteria are considered probiotics and play a significant role in maintaining GIT health of animals (Abuqwider et al., 2022). Both Lactobacillus and Limosilactobacillus exhibited a gradual decrease in abundance moving anally along the GIT (Figure 2E).

Using 0 as the threshold for presence (a feature is considered present in a group only if the total abundance across all samples in that group is greater than the threshold), we found that 1,441, 1,545, 1,635, and 1,640 genera were present in samples from the four GIT sites (stomach, small intestine, large intestine, rectum) (Figure 2C). Additionally, we observed that 1,401 genera (85% of all annotated genera) were present in all four GIT sites, 138 genera (8%) were present in samples from three sites, and only a small fraction of genera showed site-specificity (two in the stomach; Figure 2C). According to these criteria, we did not find any site-specific genera in other sites, possibly because of the similar composition of microbial species along the SPF pig GIT. At the species level, seven showed gastric specificity. Among them, the most abundant were Capnocytophaga sp. H2931 [SUM Abundance (all samples) = 543.11] from the Capnocytophaga genus [a Gram-negative bacterial genus typically found in the oral-pharyngeal cavity of mammals and associated with the pathogenesis of periodontal disease and infection following animal bites (Jolivet-Gougeon et al., 2007)], and Neisseria weaveri [SUM Abundance (all samples) = 395.36] from the Neisseria genus [a Gram-negative bacterial genus colonizing the mucosal surfaces of various animals (Panagea et al., 2002)] (Figure 2C). The reason for their high abundance might result from their advantageous use of nutrients compared with other gastric-specific microorganisms. Alternatively, they may have a close interaction with receptors on the gastric mucosa that interferes with the colonization of other microbes.

We analyzed microbial diversity further and found significant differences in alpha diversity among all GIT sites based on the Shannon index measurement (pairwise Wilcoxon rank-sum test, p < 0.05; Supplementary Figure S3A). The rectum exhibited the highest diversity, followed by the large intestine and small intestine, while the stomach showed the lowest diversity (Supplementary Figure S3A). However, considering different sources, the duodenum displayed the lowest diversity (Supplementary Figure S3B). To further determine the overall microbial characteristics of the four GIT sites, beta diversity comparisons were performed using paired non-parametric multivariate analysis of variance.

Based on principal coordinate analysis (PCoA) of Bray-Curtis distances between samples, the difference between the four GIT sites was extremely significant (R² = 0.4441, p = 0.001, paired non-parametric multivariate analysis of variance; Supplementary Figure S3C). Interestingly, samples from the cecum, colon, and rectum formed a highly clustered group (Supplementary Figure S3D), indicating strikingly similar microbial community structures in these three sources. This is consistent with their alpha diversity, as there was little difference in alpha diversity among these three sources (Supplementary Figure S3B).

To investigate the interactions among the GIT microbiota of SPF pigs, we selected core genera (average relative abundance >0.5%) and calculated the Spearman rank correlation coefficients, as shown in Figure 2D. Lactobacillus and Limosilactobacillus (the two most abundant genera) exhibited a strong positive correlation (Spearman’s rank correlation coefficient = 0.95, p = 0), indicating a potential mutualistic relationship between them. Similarly, there was a strong positive correlation among the three genera with the highest degree of centrality (Actinobacillus, Glaesserella, and Veillonella) (Figure 2D). In contrast, Moraxella showed a strong negative correlation with Clostridium, Romboutsia, and Mastadenovirus, indicating a potential competitive inhibition effect among them. Additionally, certain beneficial gut bacteria, such as Prevotella, Clostridioides, Faecalibacterium, Roseburia, Ruminococcus, and Ruthenibacterium, exhibited positive correlations with each other, forming a relatively independent and stable cluster (Figure 2D). Overall, the richness and diversity of the SPF pig GIT microbiota gradually increased from the stomach to the rectum. Moreover, these microorganisms formed a stable symbiotic network within the GIT.

In the ileum, we identified the presence of Porcine adenovirus A, which belongs to the phylum Preplasmiviricota (Supplementary Figure S3E). Infection with this virus is usually subclinical in pigs, and when clinical symptoms do occur, they manifest as mild and transient GIT symptoms, such as diarrhea, anorexia, and dehydration, and are most commonly observed in piglets (Liu et al., 2020; Gainor et al., 2022).

The overall microbial structures in the stomach, small intestine, large intestine, and rectum differ from one another. The phyla, Firmicutes and Bacteroidetes, were the two most abundant phyla, accounting for 95% (stomach), 86% (small intestine), 71% (large intestine), and 55% (rectum) of the relative bacterial abundance in the four GIT sites (Supplementary Table S4). Interestingly, the Actinobacteria and Bacteroidetes phyla showed a increasing trend in abundance moving anally along the GIT, while the Firmicutes and Proteobacteria phyla exhibited the opposite pattern (Figures 3A,C). Based on this pattern, the ratio of Firmicutes to Bacteroidetes (F/B) was lowest in the rectum and highest in the stomach. The F/B ratio is associated with energy harvesting (obesity) (Ley et al., 2006; Turnbaugh et al., 2006); therefore, its trend aligns with the physiological changes along the GIT, from food digestion (in the stomach) to energy harvesting (in the intestine).

The abundance of the phylum Bacteroidetes did not differ significantly between the stomach and small intestine (pairwise Wilcoxon rank-sum test, p > 0.05), but did show significant differences in abundance among other GIT sites (Figures 3A,C). This may be because of the presence of gastric acid, bile, and the lower pH in the stomach and small intestine affecting the survival and proliferation of the Bacteroidetes (Ding and Shah, 2007). Several functions have been reported for Bacteroidetes in the GIT. These microorganisms are involved in the fermentation of carbohydrates, converting complex carbohydrates into usable products (Hu et al., 2020; Fan and Pedersen, 2021). Additionally, they are capable of using nitrogenous substances and participating in the biotransformation of bile acids and other sterols in metabolic processes (Guzior and Quinn, 2021).

Firmicutes was the most abundant phylum in the stomach because of the high abundance of the genera Lactobacillus and Limosilactobacillus (Figure 3B). Microorganisms belonging to these genera are able to survive and establish themselves in the highly acidic environment of the stomach (Su et al., 2011). Lactobacillus is a common beneficial bacterium and is often used as a probiotic in yogurt and other fermented dairy products (Oleksy and Klewicka, 2018). These microorganisms can produce acetate (a beneficial short-chain fatty acid), lactate, and antimicrobial substances, which help to inhibit pathogens (Tomova et al., 2019). A higher ratio of Firmicutes to Bacteroidetes in the gut of obese subjects has been observed, which may be associated with more efficient absorption of calories from food by Firmicutes, leading to obesity (Moreira Júnior et al., 2021) One reason for the increase in abundance of the phylum Bacteroidetes in the GIT is the higher abundance of the genus Prevotella. Prevotella is a major genus within the Bacteroidetes, with a high abundance in the rectum and significantly lower abundance in other sites (Figures 3B,D). Prevotella is associated with the degradation of non-fibrous plant tissue and can produce short-chain fatty acids through the fermentation of cellulose and other non-digestible carbohydrates (Rubino et al., 2017). Additionally, Prevotella plays an important role in influencing host immune regulation (Iljazovic et al., 2021).

There was a significant difference in the abundance of the genus Clostridium within the phylum Firmicutes between the stomach and the large intestine/rectum (pairwise Wilcoxon rank-sum test, p < 0.05; Figure 3B). This may be because of notable differences in environmental conditions between the stomach and the large intestine/rectum (Sengupta et al., 2014; Chu and Traverso, 2022) The stomach is an acidic environment with a high concentration of gastric acid, which has inhibitory effects on many bacteria. Certain strains within the genus Clostridium may have weaker resistance to the acidic environment, resulting in lower numbers in the stomach. In contrast, the environment of the large intestine and rectum is neutral to slightly alkaline, providing more suitable growth conditions for Clostridium strains, allowing them to thrive and reproduce. However, there was no significant difference in the abundance of the genus Romboutsia across different sites of the GIT (pairwise Wilcoxon rank-sum test, p > 0.05; Figure 3B). This may be because strains within the genus Romboutsia have relatively stronger adaptability to gastric acid and intestinal environments, allowing them to survive and reproduce in different locations. The varying abundance of these bacterial genera reflect the interactive trends between the microbiota and the different GIT sites that collectively promote digestion and absorption functions along the GIT.

Additionally, we employed a linear discriminant analysis effect size (LEfSe) and linear discriminant analysis scores >3 to identify 52 taxonomic groups with significant differences in abundance at the genus level (Supplementary Figure S4). Among these taxonomic groups, the genera Lactobacillus, Romboutsia, Roseburia, and Prevotella were significantly more abundant in the stomach, small intestine, large intestine, and rectum compared with other taxonomic groups (Supplementary Figure S4). Other taxonomic groups with significant differences in abundance may also contribute to the physiological functions of the corresponding GIT sites. Therefore, further research is needed to explore their specific roles.

As shown in Figure 4A, nearly half of the proteins (274,379, 44.3% of the total) were annotated to KEGG Orthology pathways. KEGG pathway mapping indicated the most abundant pathways in the GIT of SPF pigs to be metabolic pathways (25.3%). Additionally, a significant proportion of the proteins (34.9%) could be assigned to BRITE hierarchical classifications (Figure 4A). According to KEGG functional analysis (Supplementary Figure S5), the microbial community in the small intestine exhibited richness in nucleotide metabolism, xenobiotic biodegradation, and metabolism, as well as digestive system. The microbial community in the large intestine was enriched in functions related to amino acid metabolism, biosynthesis of other secondary metabolites, and energy metabolism. Interestingly, similar results were obtained based on eggNOG analysis (Supplementary Figures S6A,E). These findings are consistent with the large intestine being a site of abundant microbial fermentation and the production of various metabolites, such as amino acids and short-chain fatty acids (Williams et al., 2001; Dai et al., 2011; Vital et al., 2014; Oliphant and Allen-Vercoe, 2019), which are crucial for host health. Therefore, we believe that the differences in composition and functionality of the microbial communities in the small and large intestines are related to their positions and nutritional conditions in the digestive system.

According to the CAZy database, only a small fraction of proteins (19,857, 3.2% of the total) were annotated as CAZymes (Figure 4B). Similar to studies in ruminant animals (Tong et al., 2022; Cao et al., 2023), the majority of these enzymes were annotated as glycoside hydrolases (9,664, 48.7% of 19,857), followed by glycosyltransferases (6,788, 34.2%) and carbohydrate-binding modules (2,052, 10.3%). The other three classes of enzymes, including carbohydrate esterases (1,048, 5.3%), polysaccharide lyases (221, 1.1%), and auxiliary activity (74, 0.4%), were represented by much smaller numbers of proteins.

We calculated the abundance of all proteins in each sample and compared their distribution. As shown in Figure 4C, the abundance of six CAZy families was relatively similar in the large intestine and rectum, but was significantly different from the other two sites (pairwise Wilcoxon rank-sum test, p < 0.05). Interestingly, the abundance of all CAZy protein families showed a trend of initially decreasing and then increasing moving anally along the GIT, with the lowest abundance in the small intestine and the highest in the rectum (Figure 4C). This might be because the small intestine is primarily responsible for nutrient absorption (Chin et al., 2017), not carbohydrate breakdown, resulting in a lower demand for CAZymes. In contrast, the microbial diversity was significantly increased in the rectum (Supplementary Figure S3A), where abundant CAZymes are involved in the degradation and fermentation of complex carbohydrates. Therefore, this trend reflects the characteristics and changing demands of different GIT sites in carbohydrate digestion and metabolism.

There were no significant differences in abundance of polysaccharide lyases and auxiliary activity enzymes between the stomach and small intestine, while differences in carbohydrate-binding modules, carbohydrate esterases, glycoside hydrolases, and glycosyltransferases were highly significant (pairwise Wilcoxon rank-sum test, p < 0.05; Figure 4C). This may be because polysaccharide lyases and auxiliary activity enzymes are primarily involved in the degradation of polysaccharides, while carbohydrate-binding modules, carbohydrate esterases, glycoside hydrolases, and glycosyltransferase enzymes have different functions, such as polysaccharide structure recognition, binding, and modification (Ndeh et al., 2017; Ye et al., 2020). Therefore, the degradation and digestion processes of polysaccharides were relatively similar in the stomach and small intestine. Interestingly, unlike other families, the abundance differences of glycosyltransferases in the stomach, large intestine, and rectum were not significant (Figure 4C; pairwise Wilcoxon rank-sum test, p > 0.05). This can be attributed to similar expression patterns and functions of glycosyltransferases in these tissues and indicates that they might be expressed at similar levels in the stomach, large intestine, and rectum, resulting in non-significant differences in their abundances. The duodenum had the lowest abundance among the six families (Figure 4D); therefore, we identified it as an outlier, possibly because of its lowest alpha diversity (Supplementary Figure S3B).

We investigated the distribution of CAZyme subfamilies in four GIT sites. The most abundant subfamily in the GIT was carbohydrate-binding module (CBM)48 (Supplementary Figure S6D), which possesses the ability to recognize and bind polysaccharide substrates, particularly xylan substances (Zhang et al., 2022). Interestingly, CBM91 exhibited lower abundance in the small intestine and stomach, but higher abundance in the large intestine and rectum (Supplementary Figure S6D). CBM91 is capable of binding xylan substances (such as birch and oat bran) (Ito et al., 2022), and the large intestine and rectum are the sites of fiber digestion (Moran and Bedford, 2022). The abundance of CBM50 gradually decreased along the GIT (Supplementary Figure S6D); CBM50 is involved in the degradation of chitin or peptidoglycans (Gruber et al., 2011). CBM6 exhibited higher abundance in the large intestine and rectum. CBM34, which binds granular starch (Kuchtová and Janeček, 2016), showed higher abundance in the stomach and small intestine, but the lowest abundance in the rectum (Supplementary Figure S6D). These differences may be closely associated with the host’s diet.

We also performed protein sequence alignment with the Comprehensive Antibiotic Resistance Database (CARD) and found a total of 43,892 protein sequences that matched antibiotic resistance genes (accounting for 7.1% of the total; Supplementary Figure S6B). The most abundant protein in the GIT was macB (Supplementary Figure S6C), which forms an antibiotic efflux complex with macA and TolC. The resistance mechanism of this complex is achieved through antibiotic efflux, with the best antibody match being against Neisseria gonorrhoeae (Xu et al., 2009). It is worth noting that the abundance of vanU gene in vanG cluster and evgS were higher in the large intestine and rectum (Supplementary Figure S6C) and lower in the stomach and small intestine. Their resistance mechanisms involve antibiotic target alteration and antibiotic efflux, with the best matches to antibodies against Streptococcus agalactiae and Escherichia coli (Nishino and Yamaguchi, 2002; Boyd et al., 2006). The highest abundance of RanA was found in the stomach and small intestine, while it was lower in the large intestine and rectum (Supplementary Figure S6C). RanA, together with RanB, confers resistance to aminoglycoside antibiotics through antibiotic efflux, with the best to the antibody match being against Riemerella anatipestifer (Li et al., 2020). Therefore, we believe there are different antimicrobial resistance genes present in different sites of the SPF pig GIT that have varying resistance mechanisms and best antibody matches. The identification of abundance differences of these antimicrobial resistance genes in the GIT provides background information to inform antibiotic use in pig farming. In summary, these findings reveal the microbial composition and functional differences within the four GIT sites of SPF pigs, along with characteristics associated with digestion and metabolism.

We determined transcriptome differences between the small intestine (n = 17) and large intestine (n = 11) by sequencing the transcriptomes of 28 tissue samples. In conjunction with the metagenomic data described above, we explored the interaction between SPF pig intestinal microbiota and the host.

By comparing gene expression profiles from different sites of the host GIT, we identified 1,618 DEGs between the small and large intestines (|Log₂(FoldChange)| > 1 and FDR.P < 0.05). Among these genes, 912 were up-regulated and 706 were down-regulated in the small intestine (Figures 5A,B).

We observed significant enrichment of up-regulated genes in multiple KEGG pathways in the small intestine (Figure 5C). Specifically, these up-regulated genes were significantly enriched in digestion and absorption-related pathways, including vitamin digestion and absorption, fat digestion and absorption, mineral absorption, and protein digestion and absorption (adjusted p < 0.05). Furthermore, immune-related pathways, such as primary immunodeficiency, complement and coagulation cascades, hematopoietic cell lineage, and viral protein interaction with cytokine and cytokine receptor, also showed significant enrichment (adjusted p < 0.05). We also observed significant enrichment in metabolism-related pathways, including glycerolipid metabolism, glycine, serine, and threonine metabolism, retinol metabolism, and arginine and proline metabolism (adjusted p < 0.05). Therefore, we conclude that the main functions of the SPF pig small intestine involve digestion and absorption, metabolism, and immune responses.

For the down-regulated genes in the small intestine (Figure 5D), we found significant enrichment in signaling transduction-related pathways, including focal adhesion, ECM-receptor interaction, and the PPAR signaling pathway (adjusted p < 0.05). Additionally, significant enrichment was observed in pathways related to nutrient metabolism, such as regulation of lipolysis in adipocytes, protein digestion and absorption, glycosaminoglycan biosynthesis—keratan sulfate, and nitrogen metabolism (adjusted p < 0.05). Therefore, we conclude that the main functions of the SPF pig large intestine involve nutrient metabolism and disease-related processes.

Further analysis of the GO enrichment results for Biological Process (BP) terms (Supplementary Figure S7A) revealed that the up-regulated genes in the small intestine were primarily enriched in immune system-related terms, such as mononuclear cell proliferation, leukocyte proliferation, lymphocyte proliferation, and lymphocyte differentiation (adjusted p < 0.05). Additionally, significant enrichment was observed in terms related to metabolic processes, such as alcohol metabolic process, and organic anion transport (adjusted p < 0.05). This indicates that the small intestine plays an important role in substance transport, homeostatic regulation, immune modulation, and cell activation.

The enriched down-regulated genes in the small intestine (Supplementary Figure S7B), were mainly enriched in terms related to extracellular matrix organization and development, such as extracellular matrix organization, extracellular structural organization, external encapsulating structure organization, wound healing, and cell-substrate adhesion (adjusted p < 0.05). Terms related to metabolic regulation, such as fatty acid metabolic process, and response to steroid hormone (adjusted p < 0.05), were also enriched. This indicates that the large intestine primarily functions in extracellular matrix organization and development, as well as nutrient metabolism.

In summary, there are differences in KEGG and GO enrichment between the small and large intestines, reflecting their specific biological functions and physiological processes. The small intestine is mainly involved in processes related to digestion and absorption, metabolism, and immune processes, while the large intestine is more involved in processes related to nutrient metabolism. These differences are consistent with our previous findings on the functional differences between the microbiota of the small and large intestines (Supplementary Figure S5). Overall, these findings indicate a co-evolutionary trend between the gut microbiota of SPF pigs and their hosts. These results provide important clues and foundations for further research on the characteristics and functions of SPF pig intestinal tissues, and contribute to a better understanding of the biological functions of the intestine and its role in host health.

To investigate correlations between host genes and the gut microbiota and potential roles that any associations may have in intestinal function, we conducted a correlation analysis between 260 DEGs and the 20 most abundant microbial taxa in the SPF pig gut. We identified 1729 specific and strong microbiota-host associations in the small intestine and large intestine (Spearman rank correlation ≥ 0.5 or ≤ −0.5, FDR.P ≤ 0.01). Interestingly, the microbial taxa associated with up-regulated and down-regulated DEGs in the small intestine were mostly distinct (Figure 6A). Specifically, we identified 11 microbial taxa (Limosilactobacillus reuteri, Lactobacillus amylovorus, Bacteroides fragilis, Lactobacillus mucosae, Roseburia hominis, Prevotella ruminicola, Prevotella dentalis, Clostridioides difficile, Ruthenibacterium lactatiformans, Ruminococcus bicirculans, and Desulfovibrio piger) that showed positive correlations with most down-regulated genes but negative correlations with up-regulated genes (Figure 6A). In contrast, we found that nine microbial taxa (Glaesserella parasuis, Actinobacillus indolicus, Romboutsia ilealis, Turicibacter sanguinis, Lactobacillus crispatus, Porcine mastadenovirus A, Escherichia coli, Lactobacillus johnsonii, and Ligilactobacillus salivarius) exhibited positive correlations with the majority of up-regulated genes but negative correlations with down-regulated genes (Figure 6A).

We then selected eight microbial taxa of interest to describe the pairwise correlations between host gene expression and microbial taxonomic groups (Figure 6B). The TCN1 gene encodes vitamin B12-binding protein, which binds to and protects vitamin B12 from the acidic environment of the stomach (Johnston et al., 1992). There was a strong negative correlation between Prevotella dentalis and TCN1 gene expression (R = −0.79, p = 2.5e-06; Figure 6B). This negative correlation may be attributed to Prevotella dentalis being a Gram-negative bacteria with metabolic pathways related to vitamin B12 (Franco-Lopez et al., 2020). The transport protein encoded by the TCN1 gene is responsible for transporting vitamin B12 into cells and participating in the metabolic processes in which it is involved. If Prevotella dentalis is involved in the intracellular metabolism of vitamin B12, a negative correlation with TCN1 expression may exist. Similarly, the ADA gene encodes adenosine deaminase, an enzyme involved in adenosine metabolism. Adenosine is an important cell signaling molecule that plays a significant role in regulating intestinal immune function and inflammatory responses. The function of adenosine deaminase is to convert adenosine into inosine, thereby regulating adenosine levels (Eltzschig et al., 2006). Although the exact function and impact of Turicibacter sanguinis in the intestines are still under investigation, some studies suggest its association with gut health and metabolism (Hoffman and Margolis, 2020). The involvement of the ADA enzyme in regulating adenosine levels and the important role of adenosine in immune regulation and inflammatory responses mean that Turicibacter sanguinis may indirectly affect ADA gene expression or enzyme activity by influencing immune cells or inflammatory signaling pathways. This could result in the negative correlation between ADA gene expression and Turicibacter sanguinis (R = −0.79, p = 5.2e-07; Figure 6B).

In addition, the 3-hydroxybutyrate dehydrogenase encoded by BDH1 participates in reactions of the fatty acid metabolism pathway. It is responsible for converting 3-hydroxybutyrate into acetoacetate, which is an important step in fatty acid oxidation metabolism. This process contributes to the generation of energy supply and the maintenance of normal intestinal cell function (Marks et al., 1992; Otsuka et al., 2020). Meanwhile, Roseburia hominis, a beneficial bacteria in the gut microbiota, is primarily involved in the degradation of cellulose and other complex carbohydrates. Bacteria belonging to the Roseburia genus play important roles in the human intestinal tract, particularly in the production of short-chain fatty acids, such as propionate, butyrate, and valerate (Tamanai-Shacoori et al., 2017). Therefore, there was a positive correlation between the BDH1 gene and Roseburia hominis (R = 0.85, p = 1.3e-06; Figure 6B). Finally, the DRD1 gene encodes the dopamine D1 receptor, which plays a crucial role in the nervous system. The dopamine D1 receptor is a key component in the transmission of dopamine neurotransmitter signals and is involved in regulating numerous functions, such as motor control, emotion, cognition, and reward (Dearry et al., 1990). Clostridioides difficile is a pathogenic bacteria that can cause intestinal infections and inflammation. It is typically transmitted through the ingestion of spores or bacteria, and produces toxins that result in symptoms such as diarrhea, abdominal pain, and fever (Sandhu and Mcbride, 2018). There was a positive correlation between Clostridioides difficile infection and the DRD1 gene (R = 0.92, p = 5.1e-07; Figure 6B). A possible reason for this is that Clostridioides difficile infection alters the gut microbiota environment, including the composition and functionality of the microbial community. Such environmental changes may impact the activity of the dopamine signaling pathway and expression of the DRD1 gene.