All animal procedures used in this study were approved by the Animal Care and Use Committee of the Institute of Animal Sciences of the Chinese Academy of Agricultural Sciences. Twenty-five piglets, born at full-term from three pure line large white sows were used in current study. All piglets were maintained in the same heat preserving pigsty at 28°C and fed solely on maternal milk (no, antibiotics, immunizations, or additives). Piglet plasma, cecal mucosa, and cecal content were collected from 5 piglets at each time point: d0 (at birth; n = 5), d1 (1 d post-birth; n = 5), d7 (7 d post-birth; n = 5), d14 (14 d post-birth; n = 5), and d21 (21 d post-birth; n = 5). At each collection point, 3 samples of cecal mucosal tissue were taken: 1) cecal tissue was cut and fixed in 10% formaldehyde for processing into histochemical blocks; 2) cecal mucosa was washed thrice in PBS buffer, gently scraped, and the removed tissue was used for the isolation of single cells for scRNA-seq analysis; 3) cecal mucosa was washed thrice in PBS buffer, gently scraped, and the removed tissue was quickly frozen in liquid nitrogen, and kept in −80°C refrigerator for proteomics and Western blotting analyses (Supplementary Figure S1A; Meng et al., 2021; Zhang et al., 2020; Zhao et al., 2020).

Cecal sample proteomics analysis was performed as reported in our earlier publications (Hou et al., 2020; Liu et al., 2020; Meng et al., 2021).

Piglet plasma was collected and kept at −80°C. The protein was removed from the plasma samples on ice before LC-MS/MS analysis using ACQUITY UPLC and AB Sciex Triple TOF 5600 (LC/MS) as described previously.

The condition for HPLC was: ACQUITY UPLC BEH C18 column (100 mm × 2.1 mm, 1.7 μm), solvent A [aqueous solution with 0.1% (v/v) formic acid], and solvent B [acetonitrile with 0.1% (v/v) formic acid] with a gradient program: 0–2 min, 5–20% B; 2–4 min, 20%–25% B; 4–9 min, 25–60% B; 9–17 min, 60–100% B; 17–19 min, 100% B; 19–19.1 min, 100–5% B; and 19.1–20.1 min, 5% B. The flow rate was set at 0.4 ml/min and 5 μl was injected. ESI was used in the mass spectrometry program. Progenesis QI v. 2.3 (Nonlinear Dynamics, Newcastle, United Kingdom) was applied to normalize the peaks. Data were characterized using the Human Metabolome Database (HMDB), Lipidmaps (v. 2.3), and METLIN software. In addition, the data were analyzed with SIMCA software (v. 14.0, Umetrics, Umea, Sweden) and KEGG database (http://www.genome.jp/KEGG/pathway.html) was applied for the pathway enrichment analysis (Zhang et al., 2020).

Cecal tissue segments were fixed in 10% neutral formalin, paraffin embedded, cut into 5 μm sections, and stained with hematoxylin and eosin (H&E) for histopathological analysis.

The protocol for immunofluorescence staining was reported in our recent publications (Zhang et al., 2020; Zhao et al., 2020; Meng et al., 2021). Supplementary Table S1 listed the primary antibodies that were used. In brief, riefly, 5 μm thick tissue sections were rehydrated gradient, subjected to antigen retrieval, and first blocked with normal goat serum in TBS, followed by incubation (1:150 in TBS-1% BSA) with primary antibodies at 4°C overnight. Sections were washed (TBS-1‰ Tween 20, 10 min X3) and then incubated with a Cy3/FITC labeled goat anti-rabbit or donkey anti-goat secondary Abs (1:150 in TBS-1% BSA; Beyotime Institute of Biotechnology, Shanghai, P.R. China) at 37°C for 30 min. After three times washes with TBST and then counterstained with Hoechst 33,342. Stained sections were examined under a Nikon Eclipse TE2000-U fluorescence microscope (Nikon, Inc., Melville, NY), and the resulting fluorescence images were analyzed with ImageJ software.

Western blotting analysis followed our previously reported protocols (Zhang et al., 2020; Zhao et al., 2020; Meng et al., 2021). Briefly, cecal mucosal tissue samples were lysed in RIPA buffer containing a protease inhibitor cocktail from Sangong Biotech, Ltd. (Shanghai, China). Protein concentration was determined using a BCA kit (Beyotime Institute of Biotechnology). Information regarding the primary antibodies used is given in Supplementary Table S1. Actin was used as the loading control. Secondary donkey anti-goat Abs (Cat no: A0181) was purchased from Beyotime Institute of Biotechnology, and goat anti-rabbit (Cat no: A24531) Abs were purchased from Novex® by Life Technologies (United States). Protein samples (50 μg/sample) were loaded onto 10% SDS polyacrylamide electrophoresis gels. The gels were transferred to a polyvinylidene fluoride (PVDF) membrane at 300 mA for 2 h at 4°C. Membranes were then blocked with 5% BSA for 1 h at RT, followed by 3 washes with 0.1‰ Tween-20 in TBS (TBST). Membranes were then incubated with primary Abs diluted to 1:500 in TBST with 1% BSA overnight at 4°C. Following a further 3 washes with TBST, the blots were, respectively, incubated with the HRP-labelled secondary goat anti-rabbit or donkey anti-goat Ab, for 1 h at RT. Secondary donkey anti-goat Ab (Cat no: A0181) was purchased from Beyotime Institute of Biotechnology, and goat anti-rabbit (Cat no: A24531) Abs were purchased from Novex® by Life Technologies. After three washes, the blots were imaged.

For cecal mucosa or content microbiota data analysis, data that were not normally distributed following log transformation or that had un-equal variances were subjected to nonparametric analysis using the Kruskal–Wallis test within the NPAR1WAY procedure of SAS.

The 10x sequencing raw data are deposited in NCBI’s Gene Expression Omnibus under accession number: GSE163272. Proteomics data are deposited at the Integrated Proteome resources (https://www.iprox.org/) with the ID: IPX0002622002. The microbiota raw sequencing data generated in this study has been uploaded to the NCBI SRA database with the accession number PRJNA688810.

In this study, we investigated the piglet cecal development during the neonatal window [from birth (d0) to 21 days of age (d21)] through scRNA-seq, proteomics, gut microbiota, and plasma metabolism (Supplementary Figure S1A; Study scheme). The piglet cecum developed gradually during its mucosal layer maturation (Figure 1A; Supplementary Figure S1B). At birth (d0), there were finger-like projections known as villi in the cecum similar as in the small intestine (Mowat and Agace, 2014). As the piglet grew, the projections became smaller and flatter to represent the cecal maturation lacking villi and a brush border showed by the vil1 staining (Figure 1A; Supplementary Figure S1C). A single-cell profiling was applied to create a map of cecal epithelia of the piglets at d0, d1, d7, d14, and d21 with 6,861, 6,451, 6,520, 6,415, and 6,439 cells (after quality control), respectively (Figures 1B–F; Supplementary Figure S1D–F; Supplementary Table S2). All the cells were combined together for further analysis and 6 clusters were partitioned by unsupervised graph clustering visualized by Uniform Manifold Approximation and Projection (UMAP; Figure 1B) as reported in early articles (Gao et al., 2018; Parikh et al., 2019; Zhao et al., 2020). This is the first study to cluster the cecum cells using scRNA-seq analysis using a method similar to Parikh et al. (2019).

The six clusters of cecal cells (Parikh et al., 2019) include undifferentiated cells [undifferentiated 1 cells (U1) and undifferentiated 2 cells (U2)], immune cells (Ims), cecal enterocyte (cecumocytes; CCs) (Parikh et al., 2019), goblet, Paneth like cells (PLCs), and enteroendocrine cells (EECs) with corresponding marker genes (Figure 1C; Supplementary Figure S1E, F; Supplementary Table S3). The timing of cell developmental potential was confirmed by RNA velocity analysis (Figure 1Cii) (La Manno et al., 2018; Joost et al., 2020).

Cecal development during the neonatal window was reflected by the proportion of different clusters of cells at different times (Figures 1D,E). Figure 1D showed the UMAP maps for the samples in different time points with each cell type. Figure 1E presented the percentage of each type of cells at each time point. The change of the number of the cells during the time was presented in Supplementary Figure S1G. At d0, CCs were most abundant, followed by goblet, EECs, Ims, PLCs, and U1/U2. However, with advancing age, CCs decreased dramatically to a minimum at d7, then increased at d14 and d21 (Figures 1D,E). On the other hand, Ims increased dramatically from d0 to d7, then continued to rise in proportion at d14 and d21 (Figures 1D,E). Meanwhile, U1/U2 gradually increased throughout the period (Figures 1D,E). At the same time, goblet cells sharply reduced from d0 to d14, and EECs gradually reduced from d0 to d21; while PLCs increased from d0 to d7, then gradually decreased till d21. At d21, Ims were most common (61.08%), followed by CCs (17.11%), U1/U2 (8.50%), goblet (6.97%), PLCs (6.27%), and EECs (0.06%; Figures 1D,E). All cell types approached mature cecum development at d21 (Mowat and Agace, 2014; Peterson and Artis, 2014). U1/U2, CCs, goblet, PLCs, and EECs were isolated and a bifurcating trajectory (pseudotime analysis) was found for these cell clusters, arising from U1/U2, then separating to secretory and absorptive lineages (Figure 1F) (Parikh et al., 2019).

Proteomics of the cecum was determined with the cecal mucosal samples of 5 piglets at each time point (Figure 1G). In total, 6,847 cecal proteins were detected (Supplementary Table S4) with 1,453 differentially expressed proteins (Supplementary Table S5). In total, 916 out of the 1,453 differentially expressed protein genes were also found in the scRNA-seq data set (Supplementary Table S6). The 916 proteins were clustered into 3 groups (Figure 1G) and their functions were enriched by Metascape online. Group 1 included 303 proteins that were higher at d0 and d1, then decreased from d7 during cecal development (Figure 1G; Supplementary Table S7). These proteins were mainly associated with catabolic and metabolic process, correlating the functions of CCs which matched the scRNA-seq data (Figure 1G). Protein levels in group 2 gradually decreased from d1 to d21 with functions related to nuclei and organic compound metabolism, also correlating to CCs (Figure 1G). Group 3 protein levels were lower from d0 to d1 while higher from d7 to d21 and functions were related to defense and immune function, correlating the functions of Ims (Figure 1G). Overall, there was a good match between proteomic and scRNA-seq data.

During neonatal development, Ims cells increased significantly (Figures 2A,B; Supplementary Table S8) from 5.12 to 61.08% during d0–d21 (Figure 1E; Figure 2B). Figure 2A showed the UMAP map for Ims for different cell types in whole while Figure 2B presented the UMAP maps for Ims at each time point with different cell types. There were four subclusters of Ims in the neonatal cecum: T cells, B cells, mast cells and innate lymphoid cells (ILC; Figures 2A,B; Supplementary Figure S2A–C). The developmental trajectory of these Ims started from ILC to T cells and B cells; however, mast cells showed a different developmental trajectory (Figure 2A; Supplementary Figure S2C; RNA velocity). T cells, and B cells continued increasing from d0 to d21; while mast cells increased from d0 to d14 then decreased; however, ILC increased from d0 to d7, then fell quickly in proportion at d14 and d21 (Figure 2B). To search the correlation of gene expression patterns and cell populations, the expression levels of the top 50 specifically expressed genes from these Ims were determined. The expression of most of these 50 genes gradually increased from d0 to d21 (Figure 2C; Supplementary Table S9), which matched the increase in proportion of the Ims. Moreover, the protein levels showed a similar trend to that of gene expression (Figure 2D; Supplementary Table S6) in the proteomics data. According to IHF analysis, protein levels of CCL5 and IL6 increased from d0 and d1 to d21, which confirmed the proteomics data and scRNA-seq data (Figures 2C,E; Supplementary Figure S2E,F). Figure 2C was the relative expression pattern of the top 50 specifically expressed genes in Ims. Figure 2E showed the IHF staining for CCL5 and CD3. Concurrently, the proportion of CD3 (part of T-cell receptor/CD3 complex; involved in T-cell development and signal transduction) positive cells increased from d0 to d 21, which further confirmed the proteomics data and scRNA-seq data (Figures 2C,E; Supplementary Figure S2G).

Unlike the developmental potential of neonatal enterocytes in the ileum (Meng et al., 2021), neonatal ceca followed a unique pattern of development (Figures 3A,B; Supplementary Figures S3A–C; Supplementary Table S10). During the neonatal window, CCs grew and differentiated very quickly (Figures 1D,E; Figures 3A,B; Supplementary Figure S3A). Figure 3A showed the UMAP map for CCs for different cell types in whole while Figure 3B presented the UMAP maps for CCs at each time point with different cell types. There were 5 subclusters of CCs: CC1, CC2, CC3, CC4, and CC5 (Figures 3A,B; Supplementary Figure S3A). Overall, the proportion of CCs quickly decreased from d1 to d7, then increased at d14 and d21 (Figure 1E; Figure 3B). The majority of CCs were grouped to the CC1 subcluster (Figures 3A,B). The proportion of CC1 continued decreasing from d0 to d21, while the proportion of CC2, CC3, CC4, and CC5 increased from d0 to d1 and then decreased from d1 to d7; however, there was an increase from d7 to d14 followed by a decrease to d21, which presented a complicated development trend (Figure 3B). Gene enrichment analysis showed that the main functions of the marker genes in CC1 and CC2 were closely associated with enterocytes; while the main functions of the marker genes in CC3 and CC4 were closely associated with blood vessel formation; meanwhile, the main functions of the marker genes in CC5 were related to neuron development. Owing to the wide function of these CCs, RNA velocity and unsupervised pseudotime analyses showed similar trends (Figure 3A; Supplementary Figure S3C).

To correlate gene expression patterns and cell population changes, the expression levels of the top 50 specifically expressed genes from CCs were analyzed. There were two expression peaks at d1 and d14 during neonatal development (Figure 3C). Furthermore, the protein levels of some of the 50 genes from the proteomics analysis showed a decreasing trend or very little change (Figure 3D). The main function of the large intestine (cecum and colon) is to reabsorb water, other small molecules, and fermentation to produce organic acids and other compounds (Dabareiner and White, 1997; Peterson and Artis, 2014). Gene expressions of the solute carrier family (Figure 3E), potassium channel subfamily (Figure 3F), proton channel family (Figure 3G), cation channel family (Figure 3H), and fatty acid binding protein family showed unique trends (Figure 3I) that correlated to CCs functions. The protein levels of ferritin (FTH1), the major intracellular iron storage protein, increased with neonatal cecal development (Supplementary Figures S3D,E). Another enterocyte protein, intestinal FABP, was expressed in piglet ceca at low levels (Supplementary Figures S3D, F) (Peterson and Artis, 2014). Levels of the absorption protein APOA1 increased along with cecal maturation (Figure 3J; Supplementary Figure S3G). At the same time the cell junction protein catenin was more condensed at d7–d21 than that at d0–d3, which indicated that the cecum maturated with time (Figure 3K; Supplementary Figure S3H).

There are 4 major types of secretory cells in the small intestine mucosal epithelium: goblet, Paneth, tuft, and EECs. It has previously been reported that there are no Paneth cells in the large intestine (Parikh et al., 2019); however, there are Paneth like cells (PLCs) in the colon (Parikh et al., 2019). Similarly, we found PLCs, goblet, and EECs (no tuft cells) in the cecum of piglets in the current study (Figures 1B–E); these cells showed a different developmental trend (Figure 1E).

In the small intestine, goblet cells synthesize and secrete mucus (Birchenough et al., 2015), which assists with the elimination of gut content and also immune defense (Kim and Ho, 2010). In the current study, cecal goblet cells were the second most prolific at d0 (Figure 1E; Supplementary Figure S1B). Specifically, there were 6 subclusters of goblet cells (Goblet 1, Goblet 2, Goblet 3, Goblet 4, Goblet 5, and Goblet 6; Figures 4A,B; Supplementary Figure S4A–C; Supplementary Table S11). Figure 4A showed the UMAP map for goblet cells for different cell types in whole while Figure 4B presented the UMAP maps for goblet cells at each time point with different cell types. Goblet 1 had the most cells while Goblet 4, Goblet 5, and Goblet 6 contained only a small proportion of cells. Goblet 2 acted as progenitor cells as demonstrated by the RNA velocity and pseudotime analysis (Figure 4A; Supplementary Figure S4C). Although the total proportion of goblet cells continuously declined (Figure 1E; Figure 4B), those of Goblet 2 cells initially decreased and then increased at d21 (Figure 4B). The top 50 specifically expressed goblet cell genes followed the same trend as the proportion of goblet cells, decreasing during d0–d21; this was especially relevant to COX2, which was most highly expressed in goblet cells (Figure 4C). Furthermore, the protein levels of some of these 50 genes followed the same trend as gene expression in goblet cells (Figure 4D) in the proteomics analysis. At the same time, the goblet markers MUC13, and TFF3 were also present in goblet cells (Figure 4E; Supplementary Figures S4D–F).

PLCs, are highly specialized cells with intensive secretory activity that are located at the base of crypts within the small intestine; their ability to produce copious secretions is owing to their extensive endoplasmic reticulum and Golgi structures (Bevins and Salzman, 2011; Clevers and Bevins, 2013). Also, within the small intestine, Paneth cells perform important antimicrobial functions as their large granules have the ability to release antimicrobial molecules including peptides (Bevins and Salzman, 2011; Clevers and Bevins, 2013). In the current study, there were 3 subclusters of PLCs with different marker genes (PLC1, PLC2, and PLC3; Figures 5A–E; Supplementary Table S12). Figure 5A showed the UMAP map for PLCs for different cell types in whole while Figure 5D presented the UMAP maps for PLCs at each time point with different cell types. The majority of PLC were in subcluster PLC1. Overall, the proportion of PLCs gradually increased from d0 to d7, then reduced at d4 and d21 (Figure 5D). The expression levels of the top 50 specifically expressed genes in PLCs followed cell proportion trends, especially for Reg4, HSPA6 and CD74 (Figure 5F). Protein levels of the PLCs marker LYZ (Sasaki et al., 2016; Haber et al., 2017) followed the trend of its gene expression pattern, which confirmed the scRNA-seq data (Figure 5G; Supplementary Figure S5A).

Intestinal EECs are known to be key sensory cells (Furness et al., 2013; Gribble and Reimann, 2016) that secrete various hormones and play important roles in nutrient and microbial metabolism (Haber et al., 2017; Worthington et al., 2018). In the small intestine, EECs are members of many overlapping sub-clusters (Habib et al., 2013; Gribble and Reimann, 2016). Cells expressing Sct, Cck, Gcg, or GIP are commonly, and respectively, called S, I, L, and K cells (Haber et al., 2017). In the current investigation, EEC in the neonatal piglet cecal epithelium were classed into 3 subclusters (Figures 6A–E; Supplementary Table S13). Figure 6A showed the UMAP map for EECs for different cell types in whole while Figure 6D presented the UMAP maps for EECs at each time point with different cell types. The proportion of EECs continued to fall from d0 to d21 (Figure 6D) and the expression levels of the top 50 specifically expressed genes in EEC followed the same trend (Figure 6F). Protein levels of CHGA (the marker gene of EECs) followed its gene expression pattern, which confirmed the scRNA-seq data (Figure 6G; Supplementary Figure S5B).

Intestinal cell renewal is orchestrated by intestinal stem cells (SCs) through their production of highly proliferative progenitor cells (Figures 7A–F; Supplementary Table S14); these cells form an undifferentiated cell pool with the potential to develop into all types of mature cecal cells: CCs, PLCs, goblet, Ims, and EECs (Williams et al., 2001; Labib et al., 2004; Peterson and Artis, 2014). Figure 7A showed the UMAP map for undifferentiated cells (U1/U2) for different cell types in whole while Figure 7D presented the UMAP maps for U1/U2 at each time point with different cell types. During neonatal development, these cecal undifferentiated cells underwent specific increases from 0.69 to 11.79% during d0–d21 (Figure 1E; Figure 7D). There were 3 subclusters of undifferentiated cells: stem cells (SCs) expressed stem cell markers; immune progenitor cells (IPCs) expressed proliferation markers and immune markers; and cecum enterocyte progenitor cells (CPCs) expressed proliferation markers and enterocyte markers (Figure 7B).

Expression of genes involved in cell cycle regulation was higher in undifferentiated cells in comparison with CCs or secretory cells (Figure 7G), and was unchanged from d0 to d21 (Figure 7H). To determine correlation of gene expression pattern and cell population, the expression levels of the top 50 differentially expressed genes from these undifferentiated cells were determined. The expression of most of these 50 genes gradually increased from d0 to d14 (Figure 7I), which matched the increase in proportion of the undifferentiated cells. Moreover, the protein levels from proteomics data showed a similar trend to that of gene expression (Figure 7J). The proportion of stem cell marker SOX9 (Verdile et al., 2019) increased at d7-d21 (Figure 7K; Supplementary Figure S5C). Similarly, the proportion of cell proliferation marker Ki67 positive cells increased at d7-d21 (Figure 7K; Supplementary Figure S5D). Concurrently, PCNA protein levels increased from d1 to d14 using WB analysis, which confirmed the proteomics data and scRNA-seq data (Figure 7L; Supplementary Figure S5E).

Gene regulatory network [GRNs; transcriptional factors (TFs)] (Aibar et al., 2017) analysis of the various types of cecal epithelia exposed several master regulators within each cell population (Figure 8A; Supplementary Table S15). Notably, the binary regulon activity heatmap indicated that CC, and Im had a predominantly high expression of regulons while the secretory cell clusters had relatively low regulon expression (Figure 8A). There was also some overlapping of regulon activity for different cluster of cells. TF protein levels confirmed their gene expression (Figure 8B; Supplementary Figure S5F–H; Supplementary Figure S6), including ZBTB1, RAB18, E2F8, Pou2AF1, and FOX O 3a.

Cell-cell communication takes place through ligand-receptor complexes; such coordination is important for multiple biological activities, including development, and differentiation (Vento-Tormo et al., 2018; Efremova et al., 2020). CellPhoneDB analysis (www.CellPhoneDB.org), based on ligand-receptor interacting pairs, was used in the exploration of cellular interaction at the cecal cell interface (Figure 8C; Supplementary Table S16). Overall, ligand-receptor interaction was higher in U1/U2, Im, and PLC, both with each other and with other cells (CCs, goblet, and EECs) compared with the interaction of CCs, goblet, and EECs with each other, or with U1/U2, Im, and PLCs (Figure 8C). The more significant pairs were CD74-APP, CD74-COPA, and CD74-MIF, followed by TNFSF10-RIPK1, TNFSF1B-GNR, and LAMP1-FAM3C. Some of these pairs have been reported to have broad functions; for example, CD74-MIF is involved in several biological processes associated with the modulation of inflammation, cardiac function, and tumor formation (Soppert et al., 2018).

GPCRs, TGF-β signaling, and BMP signaling greatly affect intestinal development (Haber et al., 2017; Gao et al., 2018). In total, 35 GPCRs were expressed in the cecal epithelium (Supplementary Figures S7A, B). Some of these receptors such as signal sequence receptor subunit 4 (SSR4) and CCR7 were specifically expressed in some cell types or at some time points (Supplementary Figures S7A,B). The TGF-β signaling pathway members TGFBR2, TGFBR1, SMAD7, SMAD4, SMAD2, and BMP signaling pathway members BMP2K and BMP4 were also specifically expressed in some types of cecal cells or at some time points (Supplementary Figures S7C, D). Our results suggest that, during the neonatal window, these factors may affect cecal cell development.

At the point of birth, a neonate moves from a sterile uterine environment to an external microbe-rich environment (Palmer et al., 2007; Costello et al., 2009; Dominguez-Bello et al., 2010; Koenig et al., 2011; Renz et al., 2011). Shortly after birth, almost no microbiota was found in piglet cecal mucosa or content, while with development, the diversity of cecal microbiota (both mucosa and content) increased to initiate their influence on intestinal cell development (Figures 9A–C; Supplementary Figures S8A–E) (Palmer et al., 2007; Costello et al., 2009; Dominguez-Bello et al., 2010; Koenig et al., 2011; Renz et al., 2011). Figure 9A showed the PCA analysis of the microflora in cecal mucosal samples from different time points. Figure 9B present the differences of bacterial abundance at the genus level. The relative proportion of the 10 major cecal mucosal microbes (at the genus level) changed during the neonatal window (Figure 9C; Supplementary Table S17), as did the relative proportion of the 10 major cecal content microbes (Supplementary Figure S8E; Supplementary Table S18). In both cecal content and mucosa, one of the major microbe populations was lactobacillus (Figure 9C; Supplementary Figure S8E). Moreover, there was a good correlation between cecal content microbes and mucosal microbes (Figure 9D; Supplementary Table S19). The cecal mucosa microbes and the cecal cell population (sRNA-seq) were well correlated (Figure 9E; Supplementary Table S20). The cecal content microbes and the cecal cell population (sRNA-seq) were also well correlated (Supplementary Figure S8F; Supplementary Table S21).

The “beneficial” microbiota lactobacillus (Zhang et al., 2020) was first observed in swine cecal content at d1 and from there increased in proportion to become the major microbiota at d21. Meanwhile the proportion of other major microbiota either decreased or fluctuated over the same period (Supplementary Figure S8E). Lactobacillus started to appear in the swine cecal mucosa at d7 and increased to become the major microbiota at d21, while other major microbiota either decreased or fluctuated over the same period of development (Figure 9C). This may be because the experimental piglets were raised solely on maternal milk, without interventions such as antibiotics, additives, or immunization.

Furthermore, piglet plasma metabolism was determined by LC/MS, and 5,388 metabolites were found in the plasma samples (Supplementary Table S22). Sixty-two metabolites continued increasing from d1 to d21 (Figure 10A; Supplementary Table S23). Melibiitol, belonging to galactose metabolism, quickly increased from d0 to d1, then decreased to d21 (Figure 10B). The two bile acids metabolites taurochenodeoxycholic acid and taurochenodesoxycholic acid decreased from d0 to d1, then increased to d21, which indicated the developmental potential of the intestine (Figure 10C). The plasma metabolites and cecal mucosa microbiota showed a more profound correlation (Figure 10D; Supplementary Table S24) than that of plasma metabolites and cecal content microbiota (Figure 10E; Supplementary Table S25). The data indicated that cecal cells and microbiota were involved in the changes taking place during plasma metabolism.