Recombinant strain pLA-ETEC K88/L.casei was constructed and stored in our laboratory (Wen et al., 2012). Briefly, The 851-bp DNA fragment encoding the fimbrial protein K88 (GenBank: M29375.1) was amplified from the ETEC strain by PCR, and then the PCR product was inserted into the vector pLA to construct the recombinant plasmid pLA-K88. Electroporation of Lactobacillus casei (ATCC-334) was carried out according to the transformation condition which was 2.0 kV/cm, 200 Ω, 25 μF, by using a Gene Pulser (Bio-Rad, Richmond, CA). It grew anaerobically at 37°C with 34 mg/ml of chloromycetin (Cm; Sigma) in MRS broth medium (Difco).

A total of 48 newborn piglets from 5 litters (Landrace, newborn average litter weight, 1.90 ± 0.05 kg) were obtained from the Sanhe farm (Qiqihar, Heilongjiang Province, China). All pigs in this study were chosen from one delivery room and had similar genetic backgrounds and husbandry practices. These piglets were allocated randomly to two groups (Ctrl: no feeding pLA-ETEC K88/L.casei, OA: feeding pLA-ETEC K88/L.casei on days 1–5) for the 28-day experiment. The piglets in the two groups were placed in three pens with a similar environment, the room temperature was maintained at 30°C, and the humidity was maintained constant at 65%–75%. The 24 piglets in the Ctrl group were fed the basic diet without any probiotics in the 28-day experiment. For the treatment group, 24 piglets in the OA group were treated with the protocol described by Li-Juan Wen (Wen et al., 2012). Briefly, pLA-ETEC K88/L.casei (5 × 10¹¹ CFU/ml) cells were orally administered daily on days 0–5 and treated according to animal protocols approved by the Institutional Animal Care and Use Committee (IACUC). All of the newborn piglets used in this study were weighed at birth, days 15 and 28 of age. The weights were calculated according to groups ctrl and OA no matter with or without diarrhea.

For the next-generation sequencing, five piglet samples were chosen randomly in each group. Piglets’ fecal samples (200 mg) were collected in a cryopreservation tube (Axygen) on days 15 and 28 for different groups. The Ctrl group samples and OA group samples were both collected on days 15 and 28 and named Ctrl15, Ctrl28, OA15, and OA28. After sample collection, the samples were quickly placed into the sterile tubes, then thrown into liquid nitrogen for half an hour and stored at -80°C until sequencing.

For special antibody detection, the piglets’ fecal samples (200 mg) were collected in the cryopreservation tube (Axygen) on days 0 (preimmune), 5, 10, 15, 20, and 28 for the Ctrl group and OA group. After collection, the samples were quickly placed into the sterile tubes, then thrown into liquid nitrogen for half an hour and stored at -80°C until antibody detection. The blood samples were collected from the front cavity vein of piglets on days 0 (preimmune), 5, 10, 15, 20, and 28 for the Ctrl group and OA group. Serum was isolated from the blood and stored at -20° until analysis.

The incidence of diarrhea was recorded twice a day (monitoring time: 9:00 a.m. and 4:00 p.m.) according to the method of Ou et al. (Ou et al., 2007). In detail, scores were assessed as 0 = normal, solid feces; 1 = slight diarrhea, soft and loose feces; 2 = moderate diarrhea, semi-liquid feces; or 3 = severe diarrhea, liquid and unformed feces. Diarrhea was defined as a score of 2 or 3 per day, and the incidence of diarrhea (%) was calculated as (number of piglets with diarrhea)/(total number of experiment piglets) × 100%. On days 0 (preimmune) and 28, the piglets were weighed.

According to the score of piglet diarrhea, we found that six piglets had obvious symptoms of diarrhea, and then piglet feces were collected.

In order to determine the diarrhea symptoms in the newborn piglets caused by ETEC K88, we conducted PCR-specific detection of feces collected from diarrheal piglets. We firstly extracted DNA from the feces samples, then used the designed K88 gene primers (5′-CGCGGATCCTTT GGTAATGTATTGAATG-3T, 5′-CGGGGTACCTTACTCTTTGAATCTGTC-3′) to amplify by PCR. The PCR products were detected by 1% agarose gel.

The specific antibodies IgG, IgA in serum, and sIgA in fecal samples were determined by enzyme-linked immunosorbent assay (ELISA), as described previously (Bai et al., 2020). The optical density was measured at 450 nm by using an ELISA auto-reader (Molecular Devices, San Jose, CA, USA) and visualized by R software (version 3.6.1) with ggplot2 package.

Genomic DNA was extracted from each group at each time point. The 16S rRNA genes were amplified via a pair of the universal bacterial 16S primers 338F (5′-ACTCCTACGGGAGGCAG CAG-3′) and 806R (5′-GGACTACHVGGGTWTC TAAT-3′), covering the V3–V4 regions of the 16S rRNA gene. The sequencing library was quantified by Qubit and qPCR, and the barcode V3 and V4 PCR amplicons were sequenced using the Illumina MiSeq platform (Shanghai Personal Biotechnology Co., Ltd., Shanghai, China). All the raw sequencing data were submitted to the NCBI Sequence Read Archive (SRA) database under accession nos. SRP282258 and SRP344853.

The Student’s t-test was used to evaluate differences in fecal microbiome between the control and experimental groups, growth details, differences in the relative abundances of the genes involved in the KEGG pathway, and correlation analysis. Differences were considered significant when the p value was <0.05.

The 48 newborn piglets were grouped as Ctrl group (24 piglets) and OA group (24 piglets). The piglets in the Ctrl group were fed the normal diet without any probiotics and those in the OA group fed pLA-ETEC K88/L.casei at days 0–5. On day 0, there was no significant difference in average body weight of piglets between the Ctrl group and OA group (p > 0.05) ( Figure 1A ). However, there was a significant difference in average body weight between the Ctrl group and OA group on day 28 (p < 0.05) ( Figure 1B ). These results indicated that the recombinant L.casei can improve and promote the growth and development of newborn piglets.

The specific anti-K88 IgG antibody in serum samples from immunized piglets was determined ( Figure 1C ). After the oral immunization, the anti-K88-specific IgG level was higher in the OA group than that in the Ctrl group and the highest value was found on day 15.

In order to determine the mucosal immune response, the specific sIgA level in fecal samples and the IgA level in serum samples were measured by ELISA ( Figures 1D, E ). The results showed that the levels of the specific sIgA and IgA were all higher than the control, suggesting that the systemic and mucosal immune responses were induced by oral immunization with pLA-ETEC K88/L.casei.

During the 28 days of the experiment, in the Ctrl group, there were 6 piglets that developed diarrhea (levels 2–3) caused by ETEC K88, which was confirmed by PCR detection ( Figure 2B ), and the diarrhea incidence was 25% ( Figure 2A ). However, in the OA group there was none. Thus, pLA-ETEC K88/L.casei improved the clinical performance of piglets by reducing incidence of diarrhea and morbidity.

The intestinal flora in the three groups of piglets was analyzed by sequencing the bacterial 16S rRNA gene amplicons (V3+V4 region). After removing the low-quality sequences, 1,050,897 clean tags were identified as a total of 73,146 OTUs presenting in these samples. In α-diversity, the Chao1 and ACE indexes revealed the minor change but not statistical significance in the OA15 group when compared with the Ctrl15 group. Most interestingly, the Chao1 and ACE values of the α-diversity index were statistically significant in the OA28 group in comparison with the Ctrl28 group (p < 0.05). The Shannon and Simpson indexes of the α-diversity index were statistically significant in the OA28 group in comparison with the Ctrl28 group (p<0.05) ( Figure 3 ). These data suggest that after the recombinant L.casei treatment, there was no clear difference in inter-group for the richness and variation of the intestinal flora on day 15. However, there was a significant difference in inter-group for the richness and variation of the intestinal flora on day 28.

In analyzing the degree of similarity between each group, principal component analysis (PCA) based on OTU relative abundances was taken for this experiment. In the coordinate system, the closer the two points, the higher the similarity was. The analysis results showed that the relative dispersion between different groups was relatively concentrated in the same group. According to the results of PCA analysis, there were significant differences between the Ctrl15 group and treatment group (OA group) ( Figure 4A ). Weighted UniFrac distances were used to estimate β–diversity and to compare among the three groups. The PCoA plot of the weighted UniFrac distances showed that the OA groups tended to separate from the Ctrl group ( Figure 4B ). Similar results were also observed in the analysis via non-metric multidimensional scaling (NMDS) ( Figure 4C ).

To determine the degree of similarity among the samples, a clustering tree of the samples was constructed ( Figure 5 ). We found out that the treatment component was divided into two clusters, and the Ctrl15 component was divided into one cluster. However, three samples of the OA15 group were in the OA28 cluster. It was supposed that the intestinal microflora of piglets in the OA28 group was similar to that in the OA15 group.

We measured the relative abundance of intestinal flora in levels of phylum ( Figure 6A ), class ( Figure 6B , family ( Figure 6C ), and genus ( Figure 6D ). The abundance of the two major phyla Firmicutes and Proteobacteria in the intestinal flora was 80.06% and 4.72% of the total abundance in the Ctlr15 group, respectively. However, after the recombinant L.casei treatment, the phyla Firmicutes was decreased in the OA15 group which constituted 72.26% and Proteobacteria was increased to 9.43%. Notably, in the OA28 group, the phyla Proteobacteria was decreased compared to the OA15 group, which constituted 8.88% of the total abundance. The abundance of the phyla Firmicutes was increased on day 28 ( Figure 6A ).

At the Class level ( Figure 6B ), the intestinal flora was dominated by class Clostridia which constituted 29.23% of the total abundance in the Ctlr15 group. After the recombinant L.casei treatment, the class Bacilli was significantly increased in the OA15 group which constituted 37.57% of their total abundance, and Gammaproteobacteria was dramatically increased to 10.29% in their total abundance.

At the Family level ( Figure 6C ), we took the top 10 taxons of total abundance. The intestinal flora was dominated by the families Ruminococcaceae and Lactobacillaceae, which constituted 65.75% and 7.56% of the total abundance in the Ctlr15 group. After feeding the recombinant L.casei, the family Lactobacillaceae was increased in the OA15 group which constituted 37.02% of their total abundance and the family Ruminococcaceae was dramatically decreased to 9.62% in their total abundance. However, surprisingly, Lactobacillaceae decreased to 20.27% in total abundance in the OA28 group.

At the Genus level ( Figure 6D ), we took the top 10 taxons of total abundance. After the recombinant L.casei treatment, Lactobacillus was markedly elevated in the OA15 group which achieved 37.02% of the total abundance. As unexpected, Lactobacillus decreased to 20.27% in total abundance in the OA28 group. Altogether, these results indicate that the recombinant L.casei intervention altered the microbiota composition in the piglet gut.

Basing on the OTU relative abundance of each sample, we analyzed the differences of bacterial abundances in genus level among groups Ctrl15, Ctrl28, OA15, and OA28 by using STAMP software with two-sided Student’s t-test ( Figures 7A, B ). The genus was ordered by effect sizes. In the Ctrl15 vs. OA15 group, the mean percentage of the genus Lactobacillus in the OA15 group was significantly higher than that in the Ctrl15 group (p < 0.05) ( Figure 7A ). In the Ctrl28 vs. OA28 group, the mean proportions of the genera Lactobacillus and Treponema in the OA28 group were upregulated in comparison to the Ctrl28 group (p < 0.05) ( Figure 7B ). However, it is not what we expected; the mean percentage values of the genus Treponema, Oscillospira, Streptococcus, and Catenibacterium in the OA28 group were upregulated when compared to those in the OA15 group ( Figure 7C ). This phenomenon also appeared to be consistent with the analysis results of the species classification tree ( Figure 7D ) and heat maps ( Figures 7E, F ) basing on OTU relative abundance between each sample.

The relative abundances of bacterial taxa and random forest regression analysis were carried out in each group to screen the key species affecting the change of intestinal flora. If a genus was removed, the model could predict the magnitude of the increase in error rate to determine its importance value which was ranked according to its importance to the model. The higher the value is, the greater the contribution of the genus to the prediction accuracy of the model. We took the top 20 taxons of the genus to predict via the model and visualize by R software ( Figure 8 ). The genus Lactobacillus was significantly higher than other genera, indicating that the intervention of L.casei played an important contribution in the change of intestinal flora.

16s rRNA marker gene sequences were used to predict the functional profiling of microbial communities by using Kyoto Encyclopedia of Genes and Genomes (KEGG), which were obtained through a phylogenetic prediction investigation of communities via reconstruction of unobserved states (PICRUSt). All the analysis results are shown in Figure 9A . We discovered that after the L.casei treatment, the relative abundances of the genes involved in membrane transport, replication, and repair were significantly increased, but in lipid metabolism, metabolism of cofactors and vitamins, metabolism of terpenoids and polyketides, and metabolism of other amino acids were significantly declined (p < 0.05) ( Figure 9B ). In the OA28 group, the relative abundances of the genes involved in membrane transport, replication, and repair were significantly increased (p < 0.05) ( Figure 9C ). In the OA15 vs. OA28 group, the relative abundances of the genes involved in cell motility and environmental adaptation were significantly decreased ( Figure 9D ).

The Spearman rank correlation analysis was used to study the potential relationship between intestinal microbial composition and body function in piglets ( Figure 10 ). The genus Lactobacillus abundance enhancement was positively correlated with the increased levels of translation, nucleotide metabolism, replication and repair, serum IgG value, serum IgA value, and fecal sIgA value (p < 0.05) but was negatively correlated with the fecal score (p < 0.05). The genus Collinsella abundance enhancement was also positively related to the increased levels of translation, nucleotide metabolism, replication and repair, and serum IgA value (p < 0.05). However, the genus Treponema abundance enhancement and the increased levels of carbohydrate metabolism were negatively correlated (p < 0.05). Bacteroides abundance enhancement was negatively correlated with the lipid metabolism increased level (p < 0.05).