Production of maternal fermented diet
The fermenting substrate consisted of corn, soybean meal, and wine lees (2:2:1). Sterile water was added to achieve an optimal 40% moisture content. Bacillus subtilis CW4 (NCBI Accession No. MH885533, 1 × 108 CFU/g) and Enterococcus faecalis CWEF (NCBI Accession No. MN038173, 1 × 108 CFU/ g) were included in the diet to promote fermentation over the course of 3 days. The nutritional values of MFD are shown in Table S1.
Experimental design
Sixty Yorkshire × Landrace sows from the week before parturition to the day of weaning were randomly divided into three groups (Fig. 1a): (i) CON group (control diet, n = 20), (ii) MFD group (basal diet + 10% MFD, n = 20) and (iii) PROB group (basal diet + equal amount of B. subtilis and E. faecalis, n = 20). For the PROB group, 3.4 g kg−1 of B. subtilis powder and 1.3 g kg−1 of E. faecalis powder were supplied in water. The diets had same amounts of crude protein and digestive energy. The ingredients and nutritional values are shown in Table S2. The sow feces were collected on day 28 and the piglet feces were obtained on day 7, 14 and 28.
To examine the effects of MFD on the resistance of offspring to colonic inflammation, piglets were challenged with LPS (Sigma-Aldrich, St. Louis, MO, USA) (Fig. 1b). Healthy piglets with similar weights were randomly selected from CON, MFD, and PROB groups. Six piglets in each group were intraperitoneally injected with 10 mg/kg of LPS. The other six piglets in the CON group were intraperitoneally injected with an equivalent amount PBS. Piglets were divided into CON, LPS, MFD + LPS, and PROB + LPS groups, slaughtered on day 21, and sampled after 12 h.
To further investigate whether MFD could affect early-life gut health by regulating the maternal gut microbiome, we performed an FMT assay using mice (Fig. 1c). Twenty-four healthy pregnant female C57BL/6J mice were allotted into four groups with six mice per group. After 2 weeks of broad-spectrum antibiotic treatment, the dams were intragastrically administered a sow fecal microbiota suspension every other day. At the time of weaning on day 21, the dams and pups were intraperitoneally injected with 10 mg/kg LPS, and the feces and colon samples were collected 12 h later. Dams and pups were divided into CON-trans, CON-trans + LPS, MFD-trans + LPS and PROB-trans + LPS groups.
Mice were further used to verify the properties of maternal effectors in vivo (Fig. 1d). Thirty healthy 4-week-old C57BL/6J male mice were allotted into five groups with six mice per group. They were pre-fed for one week and treated with 1×108 CFU/d LR (Lactobacillus reuteri, ATCC 53608), 300 mg/kg GLN (Solarbio, Beijing, China), or a combination of both agents (MIX) for 21 days. Thereafter, the mice were intraperitoneally injected with 10 mg/kg LPS for 12 h and the feces and colon samples were collected. The groups were CON, LPS, LR + LPS, GLN + LPS, and MIX + LPS.
Porcine macrophage 3D4/2 and human Caco-2 cells were selected to investigate the effects of maternal effectors on LPS-induced inflammation in vitro (Fig. 1e). Cells were treated with 10 µg/mL of LPS or an equivalent amount of PBS for 12 h after treatment (1×108CFU/mL LR or 2.5 mM GLN) for 8 h. Then the total protein and RNA of the cells were collected.
16S sequencing and data analyses
Microbial DNA was obtained from fecal homogenates using the E.Z.N.A. Soil DNA Kit (Omega Bio-tek, Norcross, GA, USA). To ensure that no contamination had occurred, the concentration and purity of the DNA samples were measured with the NanoDrop 2000 UV-vis spectrophotometer (Thermo Scientific, Wilmington, MA, USA) and examined by electrophoresis using 1% agarose gels. The primer pairs 515F (5’-GTGCCAGCMGCCGCGG-3’) and 806R (5’-GGACTACHVGGGTWTCTAAT-3’) were selected to amplify the 16S rRNA gene using the Illumina MiSeq platform at Shanghai Majorbio Biopharm Technology Co., Ltd. (Shanghai, China).
The reads were preprocessed, denoised, quality controlled, and merged in dada2 [14]. UPARSE 7.1 and QIIME2’s q2-feature-classifier plugin were used to cluster the operational taxonomic units (OTUs) with 97% and 100% similarity cutoffs [15], respectively, and chimeric sequences were identified and removed. The taxonomies from phylum to genus levels were assigned by the Greengenes database [16], and the OTUs were analyzed against the Silva database. Sequences of interest were further confirmed through NCBI platforms, and highly similar sequences were selected based on the optimization of the BLAST algorithm. Diversity analyses were performed using Qiime1 and Qiime2 [17]. In addition, predicted microbial metabolic functional annotations were obtained using Kyoto Encyclopedia of Genes and Genomes (KEGG) under PICRUSt2 (https://github.com/picrust/picrust2) [18].
To identify age-related bacteria in the gut microbiome, a random forest model was used to train the data in the control group, and then the age of the microbial community was modeled for those same taxa in all groups. The maturation index of the intestinal microbiota was calculated.
Metabolomic profiling by LC-TOF/MS
Milk samples randomly selected from six sows in each group were used for extraction and sent for metabolomic analysis. The samples were extracted to obtain the supernatants. An equivalent volume was aliquoted from each sample, mixed to prepare the QC sample and dried in a vacuum concentrator. In addition, methoxymethyl amine salt was mixed with the dried sample, and bis(trimethylsilyl)trifluoroacetamide was added. After cooling to room temperature, fatty acid methyl ester was added to each sample and mixed. Ultra-performance liquid chromatography (1290 Infinity series UHPLC System, Agilent Technologies, Santa Clara, CA, USA) was applied for LC-TOF/MS analysis. A UPLC BEH amide column (internal diameter, 2.1 × 100 mm, 1.7 μm, Waters, Milford, MA, USA) was used for separation. The obtained data were applied for principal component analysis (PCA) and orthogonal partial least squares discriminant analysis (OPLS-DA). R2 and Q2 were used to evaluate the quality of PCA and OPLS-DA models. Differential metabolites were identified with variable importance projection (VIP) > 1.0 and P < 0.05. To further interpret the biological significance of the differential metabolites, metabolic pathway analyses were performed using an online analysis platform in MetaboAnalyst 4.0 (https://www.metaboanalyst.ca/). The different milk metabolites were present at Table S3
Intestinal morphology and histology
Approximately 2 cm of the proximal colon was fixed in 4% paraformaldehyde. Colonic tissues were stained with hematoxylin and eosin using the Leica DM3000 Microsystem, and Leica Application Suite 3.7.0 (Leica, Wetzlar, Germany) was applied to observe the colonic morphology. Histological scoring was performed as described by Shahanshah et al. [19].
A suitable size of colonic tissue was placed in 2.5% glutaraldehyde and fixed overnight. The fixed sample was washed with PBS, soaked in 1% osmic acid solution for 2 h and washed with PBS. Subsequently, the sample was dehydrated in different concentrations of alcohol and dehydrated twice with absolute ethanol. The sample was transferred to a 1:1 volume mixture of ethanol and isoamyl acetate for 30 min and then transferred to pure isoamyl acetate for 1 h. After the critical point was dried, the gold-platinum film was coated and observed with a field emission scanning electron microscope.
The colonic tissue fixed by the osmic acid solution was washed with PBS and dehydrated in different concentrations of alcohol and acetone. The sample was placed in a mixture of pure acetone and embedding agent for 1 h, transferred to a mixture of pure acetone and embedding agent for 3 h, and finally transferred to embedding agent for 12 h. The embedded sample was heated and then sectioned with an ultra-thin microtome. Finally, the sections were stained and treated with uranyl acetate and alkaline lead citrate. A suitable field of view was identified and imaged with a JEM-1011 transmission electron microscope (JEOL USA, Peabody, MA, USA).
Immunofluorescent staining and TUNEL
Colonic slides were incubated with antigen-recovering solution (Vector Laboratories, Inc. Burlingame, CA, USA) for 15 min, and the non-specific binding was blocked with 5% BSA. Slides were transferred to anti-ZO-1 and anti-β-catenin primary antibodies (ab96587 and ab32572, Abcam, Cambridge, MA, USA) at 4°C overnight. After washing with PBS, slides were incubated with fluorescent dye-conjugated secondary antibodies (Abcam) for 1 h and protected from light. Finally, 4, 6-diamidino-2-phenylindole (DAPI) was used to stain the nucleus. Sections were examined with a DM5000 B fluorescent microscope (Leica, Wetzlar, Germany).
After colonic sections were deparaffinized, DNase-free proteinase K (20 μg/mL) was applied and sections were incubated at 37°C for 20 min. After washing with PBS, the TUNEL detection solution was added dropwise, and sections were incubated in a 37°C incubator for 2 h. After adding the stop solution to complete the reaction, sections were washed with PBS. Thereafter, the streptavidin-HRP working solution was added dropwise, sections were incubated for 30 min. Sections were washed with PBS, followed by the addition of 0.2–0.5 mL of DAB and incubation at room temperature. The specific time was determined according to the degree of color development and images were acquired for analysis.
Western blotting
Total proteins were obtained using the Protein Extraction Kit (KeyGen BioTECH, Nanjing, China). SDS-PAGE separates proteins of different sizes and electroporates them onto PVDF membranes (Millipore, Bedford, MA, USA). Membranes were blocked with 5% skimmed milk and incubated with anti-ZO-1, anti-Occludin, anti-Claudin-1, anti-Cleaved capase3, anti-Bax, anti-Bcl-2, anti-iNOS, anti-Arg1, anti-p-p38, anti-p38, anti-p-JNK, anti-JNK, anti-p-ERK1/2, anti-ERK1/2, and anti-β-actin antibodies at 4°C overnight. After washing with TBST, membranes were transferred to secondary antibodies for 1 h. Protein bands were visualized with an ECL Assay Kit (Biosharp, Hangzhou, China) and measured with Image J software (NIH, Bethesda, MD, USA).
ELISA assay
Approximately 0.1 g of colonic tissue was homogenized after milling at 12000 × g for 10 min at 4°C. Cytokine concentrations in supernatants were measured with ELISA Kits (Jiangsu Meibiao Biological Technology Co., Ltd., Jiangsu, China).
In vitro fermentation
Approximately 1 g of pre-digested MFD was suspended in 25 mL of sterile fermentation medium and boiled for 10 min. The boiled MFD solution was introduced into the anaerobic chamber, cooled down to room temperature, and reduced for 2 h. Thereafter, 5 mL of sterile fermentation medium was combined with 0.1 g of faecal sample and homogenized, followed by the mixing of 2.5 mL of the homogenized faecal suspension and 2.5 mL of the MFD solution (final MFD concentration, 1%; final fiber concentration, 2%) and incubation under anaerobic conditions at 37°C with 125 rpm shaking for 14 h. All fermentation steps were conducted in an anaerobic chamber (Bactron, Sheldon Manufacturing, Inc., Cornelius, OR, USA) in an anaerobic environment containing 5% CO2, 5% H2, and 90% N2.
Quantitative real time PCR and absolute quantification of L. reuteri
Primers were used to measure the copy numbers of the 16S rRNA gene of specific bacteria (V3-V4). The Lreu gene was selected to analyze the abundance of the L. reuter strain. Forward and reverse primers were used as follows: 5′-CAGACAATCTTTGATTGTTTAG-3′ and 5′-GCTTGTTGGTTTGGGCTCTTC-3′ [20]. To obtain a standard curve, plasmids containing target DNA PCR fragments of 16S rRNA and Lreu genes were diluted 10-fold. qPCR reactions and analyses were conducted as previously described [20]. The ΔCt method was used to calculate the copy numbers of Lreu genes from the standard curve of 16S rRNA.
Cell culture
Caco-2 and 3D4/2 cells were obtained from ATCC Cell Bank (Shanghai, China). DMEM-F12 medium supplemented with 10% fetal bovine serum and double antibiotics at 37°C was used. Cells were incubated in an incubator containing 5% CO2.
In vitro gut paracellular permeability
Approximately 100 µL of 4-kDa FD4 (1 mg/mL) was added to the apical chamber after the last TEER measurement. The plates were placed at humidified incubator for 30 min. The prepared FD4 solution was serially diluted (5, 10, 20, 40, 80, 160, 320, 640, 1,280 ng/mL). Subsequently, the level of FD4 in the basolateral chamber was measured (excitation wavelength, 492nm; emission wavelength, 520nm).
Transepithelial electrical resistance (TEER)
Transwell filters (Corning, NY, USA) were used to culture Caco-2 cells. When the TEER stabilized, cells were used for subsequent experiments. The TEER was determined using the Millicell ERS voltohmmeter (Millipore, Burlington, MA, USA) at 0, 2, 4, 6, 8, 10, and 12 h after LPS stimulation. Changes were analyzed as a fold of TEER at 0 h.
Statistical analysis
All data were presented as means ± SEM as analyzed by SPSS 26.0 (SPSS Inc. Chicago, IL, USA). The “corrplot” package in R (R Core Team, 2014) was used to obtain the Pearson correlation coefficient and significance. Differential gut microbes were verified by the ANOVA test and Benjamini–Hochberg FDR adjusting of STAMP (version 2.1.3). The contribution of maternal factors on gut microbiota in offspring was calculated by variance partitioning analysis (VPA) using the “varpart” package of R.