A Novel Probiotic Bacillus subtilis Strain Confers Cytoprotection to Host Pig Intestinal Epithelial Cells during Enterotoxic Escherichia coli Infection

To examine the impact of CP9 in attenuating ETEC-induced cytotoxicity in IPEC-J2 cells, we performed co- and preincubation assays. No morphological changes were observed when IPEC-J2 cells were incubated with CP9 or commercially available B. subtilis, CS. However, a clear cell damage was morphologically observed for the cells incubated with ETEC (Fig. 1A and C). Our results showed that ETEC-induced cell damage was clearly attenuated when IPEC-J2 cells were simultaneously or preincubated with CP9, but not with CS (Fig. 1A and C). Furthermore, the viability of the IPEC-J2 cells was not affected by the incubation with CP9 or CS; however, ETEC caused significant (P < 0.0001) IPEC-J2 cell death after 4 h of incubation (Fig. 1B and D). Consistently, the viability of IPEC-J2 cells infected with ETEC was significantly (P < 0.0001) maintained when co- or preincubated with CP9, but not with CS (Fig. 1B and D). Combined, these results suggest that CP9 protects IPEC-J2 cells from ETEC-induced toxicity and cell damage.

To investigate the inhibitory effect of CP9 on ETEC adhesion to the IPEC-J2 cells, we performed a coincubation assay, where equal numbers of CP9 and ETEC were incubated with IPEC-J2 cells for 3 h at 37°C. CP9 monoculture showed higher adhesion to the intestinal cells than ETEC monoculture, and CP9 significantly (P < 0.05) reduced the ETEC adhesion to IPEC-J2 cells when incubated together with ETEC (Fig. 2A). In addition, analysis of the viable ETEC in the spent media showed significantly (P < 0.05) lower ETEC counts compared to IPEC-J2 cells incubated with ETEC alone (Fig. 2B). These results suggest that CP9 may inhibit ETEC adhesion to IPEC-J2 cells by competitively acquiring the binding sites on IPEC-J2 cells and also reducing ETEC growth in a coculture.

To analyze the ability of the probiotic in modulating NO production in IPEC-J2 cells during ETEC infection, we performed a coculture assay. Our results showed that ETEC induced significantly higher (P < 0.01) levels of NO production in IPEC-J2 cells compared to the no-treatment control and CP9-treated cells (Fig. 3). In addition, NO production by CP9 was significantly higher (P < 0.05) than the no-treatment control cells. Interestingly, the results showed that when coincubated with ETEC, CP9 reduced the NO production in the IPEC-J2-infected cells, which was still significantly (P < 0.01) higher than for the no-treatment control cells. Combined, these results suggest that CP9 may exhibit its cytoprotective role in IPEC-J2 cells by moderating the NO production in the cells.

To understand the impact of CP9 on inflammatory responses induced by ETEC-infected IPEC-J2 cells, we first analyzed the relative expression of mRNAs for selected genes encoding the TLRs (Fig. 4). Compared to no-treatment control, incubation with CP9 alone significantly (P < 0.05) increased the relative expression of TLR2 and TLR9 mRNAs but not TLR4 mRNAs. As expected, the relative expression of TLR2, TLR4, and TLR9 mRNAs significantly (P < 0.05) increased in the ETEC-infected cells. The relative expression of TLR2, TLR4, and TLR9 mRNAs was significantly (P < 0.05) lower in cells coincubated with CP9 and ETEC. Interestingly, incubation with CP9 lowered the relative expression of the TLR2, TLR4, and TLR9 mRNAs to the negative-control levels in ETEC-infected cells, indicating a homeostatic expression of the TLRs.

To test whether CP9 can attenuate the inflammatory response induced by ETEC in IPEC-J2 cells, we measured the relative expression of tumor necrosis factor alpha (TNF-α), interleukin-6 (IL-6), IL-8, granulocyte-macrophage colony-stimulating factor (GM-CSF), and IL-10 mRNAs (Fig. 5). Compared to no-treatment control, incubation with CP9 alone significantly (P < 0.05) reduced the relative expression of IL-6 mRNA but not of TNF-α and IL-8 mRNA (Fig. 5A to C). As expected, infection with ETEC significantly (P < 0.05) increased the levels of TNF-α, IL-6, and IL-8 mRNAs compared to no-treatment control and CP9 alone. Interestingly, this upsurge was significantly (P < 0.05) attenuated when CP9 was coincubated with ETEC (Fig. 5A to C). Furthermore, incubation with CP9 alone significantly (P < 0.05) increased the relative expression of GM-CSF mRNA (Fig. 5D); however, no significant change was observed in ETEC-infected cells. Similarly, in the presence of CP9, ETEC-infected IPEC-J2 cells showed a higher relative expression of GM-CSF mRNA. In addition, there was no significant difference observed in the relative levels of IL-10 mRNA when IPEC-J2 cells were incubated with CP9 alone, ETEC alone, or both in coculture (Fig. 5E).

To determine whether CP9 can modulate expression of host defense peptides (HDPs), we analyzed the relative expression of BD3, MUC1, and PG-1 in the IPEC-J2 cells infected with ETEC (Fig. 6). Although an increasing trend in the relative expression of BD3 was observed in cells incubated with CP9 alone and in coculture with ETEC, the difference was not significant (Fig. 6A). Similarly, no difference in expression of BD3 was observed in the ETEC-alone cells compared to no-treatment control cells. Incubation with CP9 alone significantly (P < 0.05) increased the relative expression of MUC1 mRNA (Fig. 6B). Incubation with ETEC alone significantly (P < 0.05) decreased the relative expression of MUC1 mRNA, which was significantly (P < 0.05) attenuated when cells were cocultured with CP9. Lastly, incubation with CP9 alone significantly (P < 0.05) increased the relative expression of PG-1 mRNA (Fig. 6C); however, no significant difference was observed in PG-1 mRNA expression with other treatments. Collectively, these data suggest that CP9 has the ability to modulate the expression of immune markers by either decreasing or maintaining the homeostatic expression of inflammatory and HDP genes.

To test whether CP9 can attenuate the reduction in intestinal barrier function induced by ETEC in IPEC-J2 cells, we measured the relative expression of claudin-1, occludin, and tight-junction protein-1 (zona occludens-1) mRNAs (Fig. 7). Compared to control cells, incubation with CP9 alone did not demonstrate any significant change in any of the three intestinal barrier function genes tested (Fig. 7). As expected, infection with ETEC significantly (P < 0.05) reduced the relative expression of claudin-1, occludin, and zona occludens-1 mRNAs (Fig. 7). Interestingly, this reduction was significantly (P < 0.05) reversed in the coculture samples, where IPEC-J2 cells were coincubated with CP9 and ETEC. Combined, these results suggest ETEC reduced the mRNA expression of intestinal barrier function genes, and CP9 addition to the ETEC-infected IPEC-J2 cells may strengthen the barrier function by improving the levels of claudin-1, occludin, and zona occludens-1 mRNAs.

To test whether CP9 can attenuate the ETEC-induced apoptosis in IPEC-J2 cells, we measured the caspase-3 activity in IPEC-J2 cells infected with ETEC (Fig. 8). Although CP9 alone did not show any changes in caspase-3 activity compared to no-treatment control cells, the caspase-3 positive control and ETEC increased caspase-3 activity after 4 h of incubation in IPEC-J2 cells. Caspase-3 activity diminished with the addition of caspase-3 inhibitor in all treatment groups. Interestingly, caspase-3 activity was significantly (P < 0.05) reduced to no-treatment control samples in the CP9- and ETEC-coincubated IPEC-J2 cells. These data suggest that CP9 may have the ability to reduce apoptosis by reducing the caspase-3 activity either directly or by reducing ETEC-induced cytotoxicity in IPEC-J2 cells.

Enterotoxic E. coli induces cell cycle arrest and has been shown to negatively affect intestinal cell proliferation (39–41). Since apoptosis and cell proliferation are very critical steps in epithelial cell turn over in intestinal cells (42), we next assessed whether CP9 could attenuate the impact of ETEC on IPEC-J2 cell proliferation. As expected, incubation of IPEC-J2 cells with ETEC for 2 h caused a significant decrease in IPEC-J2 cell proliferation as assessed by CCK-8 assay after 8 h (Fig. 9). The results showed that metabolic activity of cells in proliferation was significantly lower in cells incubated with ETEC than in negative-control cells. In addition, when incubated with CP9 for 8 h, ETEC-incubated IPEC-J2 cells showed a metabolic activity similar to that of negative-control cells. This suggests that CP9 was able to reinstate cellular proliferation and attenuate the negative impact of ETEC on IPEC-J2 cell proliferation.

To analyze whether CP9 could cause metabolic changes in the IPEC-J2 cells, we performed extracellular untargeted metabolomic analyses on the IPEC-J2 cells incubated with CP9, using liquid chromatography coupled to a mass spectrometer (LC-MS). A total of 113 metabolites were successfully identified (see Table S2 in the supplemental material) which were then statistically analyzed through Metaboanalyst (version 5.0) online analysis software. Incubation with CP9 substantially altered the metabolome of IPEC-J2 cells (see Fig. S1 in the supplemental material). To detect the modulation in the metabolome resulting from the CP9 incubation, we performed principal-component analysis (PCA) on the metabolic profile of CP9-incubated IPEC-J2 cells and no-treatment IPEC-J2 cells. A distinct separation was observed between the metabolomic profiles of no-treatment cells and CP9-incubated cells (Fig. 10A). Component 1 explained 60.7% of the variance, and component 2 explained 26.8% of the variance between the two groups. Analysis of statistically significant (P < 0.05) metabolites identified by Welch’s two-sample t test revealed that more metabolites had increased in abundance than had decreased in abundance in the CP9-incubated cells (see Table S3 in the supplemental material). Of the 49 significant metabolites, 37 increased in abundance, and 12 decreased in abundance in response to CP9 administration (see Table S3). To further analyze which metabolites caused maximum variance between the two groups, we performed supervised partial least-squares discriminant analysis (PLS-DA; Fig. 10B). The predictive performance of the model by cross-validation coefficient (Q²) was greater than 0.8 (see Fig. S2), suggesting that the models used were of reasonable and acceptable quality. Consistently, a clear separation was observed between the metabolomic profiles of no-treatment cells and CP9-incubated cells via PLS-DA, indicating considerable variation in the IPEC-J2 cells’ metabolite profiles upon incubation with CP9. Furthermore, the variable importance in projection (VIP) score plot in PLS-DA model showed 34 metabolites (VIP score >1.2) responsible for causing maximum separation between the two groups (Fig. 10C). Statistically different metabolites between the groups were then subjected to the pathway impact analysis nodule (in Metaboanalyst version 5) to unravel the metabolic pathways that were being modulated by CP9 in IPEC-J2 cells. The results revealed that CP9 administration caused major changes in 17 metabolic pathways metabolism; out of which, the greatest impact was seen in six major pathways: (i) alanine, aspartate, and glutamate metabolism; (ii) pyrimidine metabolism; (iii) nicotinate and nicotinamide metabolism; (iv) glutathione metabolism; (v) the citrate cycle (the tricarboxylic acid [TCA] cycle); and (vi) arginine and proline metabolism (Fig. 11A). Hierarchical clustering further revealed that metabolites involved in the alanine, aspartate and glutamate metabolism (l-aspartate, dl-glutamine, and succinate) were seen in higher abundance in the CP9-incubated IPEC-J2 cells (Fig. 11B), which may reflect a higher energy demand in the cells (43, 44). Metabolites involved in pyrimidine metabolism (dl-glutamine, cytidine, and thymine) were seen in higher abundance in the CP9-incubated IPEC-J2 cells, which may reflect the higher demand of nucleic acid production owing to cell growth and glucose transport (45, 46). Metabolites involved in the nicotinate and nicotinamide metabolism (dl-tryptophan, l-aspartate, and niacin) were seen in greater abundance in the CP9-incubated IPEC-J2 cells, which may reflect increased energy derivation in the cells for enhanced physiological functions (47, 48). Metabolites involved in glutathione metabolism (cysteine-glutathione disulfide, reduced glutathione, and l-gamma-glutamyl-l-leucine) were seen in greater abundance in the CP9-incubated IPEC-J2 cells, which may reflect a higher antioxidant status and nutrient metabolism in the cells (49), possibly induced by CP9. Finally, metabolites involved in the TCA cycle (succinic acid) and in arginine and proline metabolism (l-proline and dl-arginine) were seen in greater abundance in CP9-incubated IPEC-J2 cells, which may further reflect an increased production of amino acids, cell growth, and energy derivation (50–52) in the CP9-incubated cells.

After hierarchical cluster analysis, we performed metabolite set enrichment analysis by using Metaboanalyst (version 5) to identify detailed pathways that were modulated by the significantly observed metabolites. The results showed that 24 pathways were impacted with CP9 administration (Fig. 11C). Consistently, alanine, aspartate, and glutamate metabolism, pyrimidine metabolism, nicotinate and nicotinamide metabolism, glutathione metabolism, and arginine and proline metabolism were among the top 10 pathways substantially modulated between the two groups. Collectively, these findings indicate that upon the administration of probiotic CP9, the host IPEC-J2 cells undergo notable metabolic changes that may reflect the cytoprotective ability of CP9.

Bacillus subtilis CP9 was isolated previously from sub-Saharan camel feces and characterized in our lab at the University of Guelph, Guelph, Ontario, Canada (37). Fresh stocks of the strain were stored and preserved in a −80°C Ultra-low-temperature Freezer (Thomas Scientific). CP9 was grown in Luria-Bertani (LB) liquid broth medium at 37°C, in shaking incubator under aerophilic conditions.

The E. coli strain used in our study was F4-expressing enterotoxigenic E. coli (ETEC), serotype K88. The strain was acquired from the Animal Health Lab, University of Guelph, and was grown in LB medium at 37°C in a shaking incubator under aerophilic conditions. As an additional control, a commercially available B. subtilis strain (CS) was used for comparative analysis. The strain was grown in LB liquid broth medium at 37°C in a shaking incubator under aerophilic conditions.

The swine intestinal epithelial cell line IPEC-J2, originally derived from jejuna of neonatal piglets was acquired from the American Type Culture Collection (Cedarlane, Burlington, Ontario, Canada). IPEC-J2 cells were cultured in a 1:1 mixture of Dulbecco modified Eagle medium/Ham’s nutrient medium F-12 (DMEM/F-12) supplemented with 10% fetal bovine serum (FBS), 1% penicillin-streptomycin solution (Gibco/Fisher Scientific, Mississauga, Ontario, Canada) under 5% CO₂ in a 95% air atmosphere with 90% humidity at 37°C. The culture medium was replaced every 3 days, and the cells were split after reaching 80% confluence with 0.25% trypsin-EDTA (Gibco/Fisher Scientific).

For bacterial coincubation experiments, supplemented DMEM/F-12 nutrient medium was replaced after 24 h with CO₂-independent medium (Gibco/Fisher Scientific) supplemented with l-glutamine (GlutaMAX supplement; Gibco/Fisher Scientific) and 10% FBS.

ETEC infection model was created as previously described (121) with minor modifications. Briefly, IPEC-J2 cells were seeded at a density of 4 × 10⁵ cells per well in six-well tissue culture plates (Corning; Fisher Scientific). After 24 h of incubation, the medium was replaced with incomplete CO₂-independent medium supplemented with l-glutamine and FBS. For the preincubation assay, IPEC-J2 cells were pretreated with phosphate-buffered saline (PBS)-washed 10⁸ CFU of CP9 or CS/well for 3 h and then stimulated by the addition of 10⁸ CFU of ETEC/well. The cells were incubated in a 37°C incubator for 4 h. Similarly, for the coincubation assay, IPEC-J2 cells were incubated with 10⁸ CFU of CP9 or CS/well, along with 10⁸ CFU of ETEC/well, followed by incubation in a 37°C incubator for 4 h. The morphology of the cells was assessed by using Cytation 5 Cell Imaging Multi-Mode Reader (BioTek Instruments, Winooski, VT) at ×20 magnification, using a 100-μm scale in bright-field mode. After removal of the medium, cells were washed twice with PBS and collected using 0.25% trypsin-EDTA. The numbers of viable cells were counted by using a standard trypan blue assay with a TC20 Automated Cell Counter (Bio-Rad Laboratories, Ltd., Mississauga, Ontario, Canada).

The cell adhesion assay was performed as previously described (122) with minor modifications. Briefly, IPEC-J2 cells were seeded at a density of 2 × 10⁵ cells per well in six-well tissue culture plates (Corning) and grown under 5% CO₂ in a 95% air atmosphere with 90% humidity at 37°C. After 24 h of incubation, the medium was replaced with incomplete CO₂-independent medium supplemented with l-glutamine and FBS, followed by incubation in a 37°C incubator for 2 h. After an initial adjustment for 2 h, PBS-washed (twice) 10⁸ CFU of CP9, ETEC, and CP9-ETEC/mL together were added to IPEC-J2 cells in respective wells. The cells were cultured for an additional 3 h. The culture medium was removed, and the cells were washed twice with PBS to remove unbound bacterial cells. IPEC-J2 cells were collected using 0.25% trypsin-EDTA incubation for 2 min. Adherent bacteria were serial diluted by 10-fold in LB medium and plated on LB agar plates. The numbers of CP9 and ETEC colonies were counted after aerobic incubation for 24 h. CP9 and ETEC were enumerated in LB and MacConkey agar plates, respectively.

The induction of NO by IPEC-J2 cells upon treatment with ETEC and CP9 was assessed as previously described (123), with minor modifications. Briefly, IPEC-J2 cells were seeded in a 96-well tissue culture plate until 80% confluence. The cells were then stimulated with 10⁸ CFU/mL ETEC or CP9 monocultures or 10⁸ CFU/mL ETEC-CP9 cocultures, followed by incubation at 37°C for 3 h. Nitric oxide production induced by different treatments was measured by using a Griess reagent system (Promega, Madison, WI), which measures nitrite as a primary breakdown product of NO. Manufacturer’s instructions were followed to perform this assay. Briefly, 100 μL of the cell culture mixture was transferred to a new 96-well tissue culture plate, and 100 μL of Griess reagent was added to the each well; samples were incubated for 15 min and protected from light. NO was then detected by measuring the absorbance at 540 nm by using a spectrophotometer with a Cytation 5 Cell Imaging Multi-Mode Reader (BioTek Instruments). The results were expressed as nitrite concentrations (μM) produced, by plotting a standard curve.

The impact of CP9 on the proliferation of IPEC-J2 cells during ETEC infection was assessed as previously described (124), with minor modifications. Briefly, IPEC-J2 cells were seeded in a 96-well plate at a density of 3 × 10⁴ cells/well, followed by incubation at 37°C for 24 h. Spent medium was then replaced with CO₂-independent medium without antibiotics. IPEC-J2 cells were stimulated with 10⁸ CFU of ETEC/mL for 2 h at 37°C. The spent medium was then removed, and cells were incubated with fresh medium containing 10⁸ CFU of CP9/mL at 37°C for an additional 6 h. The final volume in each well was 100 μL. Cells with or without FBS were used as controls. Cell proliferation was measured by using a CCK-8 kit (Bimake, Houston, TX) according to the manufacturer’s instructions. Briefly, 10 μL of the CCK-8 reagent was added to each well, followed by incubation at 37°C for 2 h. Metabolically active cells in proliferation were counted by measuring the absorbance at 450 nm using a Cytation 5 Cell Imaging Multi-Mode Reader (BioTek Instruments).

The activity of caspase-3 was determined as previously described (125) by using a caspase-3 activity kit (Sigma-Aldrich, Oakville, Ontario, Canada), which is based on the ability of caspase-3 to hydrolyze the acetyl-Asp-Glu-Val-Asp p-nitroanilide (Ac-DEVD-pNA) substrate into a yellow formazan p-nitroaniline (pNA) moiety. Briefly, a coincubation assay was performed as described above. After 4 h of incubation, cell lysates were prepared by centrifuging cells at 12,000 × g for 15 min at 4°C. Cell lysates (10 μL) were added to each well in a flat-bottom, 96-well culture plate, followed by incubation with 10 μL of Ac-DEVD-pNA overnight at 37°C. Controls used in the assay included the following: no treatment, a caspase-3 positive control (provided with kit), and a caspase-3 inhibitor (provided with the kit). The caspase-3 activity was determined by measuring the absorbance value of released pNA at 405 nm by using a Cytation 5 Cell Imaging Multi-Mode Reader.

A coincubation assay with ETEC infection was performed as described above. Briefly, IPEC-J2 cells were seeded at a density of 4 × 10⁵ cells per well in six-well tissue culture plates. After 24 h of incubation, the medium was replaced with incomplete CO₂-independent medium supplemented with l-glutamine and FBS. IPEC-J2 cells were then infected with 10⁸ CFU of ETEC/well, along with 10⁸ CFU of CP9/well, and incubated in a 37°C incubator for 4 h. The cells were collected, and the total RNA was isolated from the cells using a Norgen total RNA purification kit (Norgen BioTek, Thorold, Ontario, Canada). The RNA yield and quality were assessed spectrophotometrically with a NanoDrop 8000 spectrophotometer (Thermo Fisher Scientific). An extracted RNA sample was reverse transcribed to complementary DNA using an iScript Reverse Transcription Supermix kit (Bio-Rad Laboratories, Ltd., Berkeley, CA). Quantitative real-time-PCR (qPCR) was conducted using a Bio-Rad CFX Connect real-time system. Primers were designed using the Primer-BLAST tool (National Center for Biotechnology Information) and synthesized by Integrated DNA Technologies, Guelph, Ontario, Canada. Primer information is listed in Table S1 in the supplemental material. The efficiencies of the primers were calculated by using CFX Manager Software (Bio-Rad Laboratories, Ltd.). Real-time quantitative PCR was performed at 95°C, activating the enzyme for 2 min; at 95°C, with denaturing for 5 s; and then at 60°C, with annealing for 20 s. Reactions were repeated for 40 cycles. The efficiency of each primer set was calculated by performing qPCR, and only primer sets with an efficiency between 90 and 110% were selected for the downstream gene expression analysis. The product size of each primer set was verified by agarose gel electrophoresis. Samples were assessed in triplicate. The relative mRNA expression was analyzed by normalizing the cycle threshold (C_T) values of the target genes to the C_T values of the housekeeping gene encoding GAPDH. The results are presented as the fold change calculated using the 2^−ΔΔCT method.

To determine the metabolic impact of CP9 on IPEC-J2 cells, a coculture experiment was performed wherein CP9 was incubated with IPEC-J2 cells for 5 h. After the end of the incubation period, spent supernatants from negative-control cells and CP9-incubated cells were collected and filter sterilized using Fisher 0.22-μm-pore-size nylon filters. Nutrient medium (CO₂-independent medium) was used as a blank. The samples were immediately frozen in liquid nitrogen and stored in a –80°C Ultra-Freezer. Samples were packed in dry ice and shipped to the BioZone Mass Spectrometry Facility in the Chemical Engineering Department, University of Toronto, for metabolite extraction and LC-MS analysis (courtesy of metabolomics specialist Robert Flick). Analysis was performed as previously described (38). Briefly, metabolites were vacuum dried after protein precipitation and resuspended in appropriate starting solvent for each chromatography analysis. A Thermo Scientific Ultimate 3000 UHPLC (Thermo Fisher Scientific, Waltham, MA) equipped with Hypersil Gold C₁₈ column (50 mm × 2.1 mm, 1.9 μm) or a Phenomenex Luna NH₂ column (150 mm × 2mm, 3 μm) were then used to analyze the samples. Temperature at 40°C and a flow rate of 300 μL min⁻¹ were set for the column. Water and acetonitrile containing 0.1% formic acid were used as eluents. Two separate gradients were performed for each column. A gradient at 5% B (1 min), a linear gradient at 98% B (6 min), maintained at 98% B (3 min), returned to 5% B (0.5 min), and finally a reequilibration at 5% B (4.5 min) (total runtime, 15 min) were performed for the C₁₈ column. A gradient at 90% B (1 min), a linear gradient at 5% B (4 min), maintained at 5% B (8 min), returned to 90% B (1 min), and finally a reequilibration at 90% B (6 min) (total runtime, 20 min) were performed for the Luna NH₂ column. Liquid samples (10 μL) were loaded onto the autosampler. The autosampler temperature was kept at 10°C. A Q-Exactive Orbitrap mass spectrometer (Thermo Fisher Scientific) equipped with a heated electrospray ionization (HESI II) probe was used for compound detection. The system was operated in negative and positive ionization modes to generate spectra. After generating the raw peaks, we processed the untargeted metabolomic data (raw signal exacting, data baseline filtering, peak identification, and integration) and metabolite detection (KEGG and BioCyc database) using a differential analysis software package (Compound Discoverer 2.1; Thermo Scientific).

All experiments were performed in three biological replicates, and data are presented as means ± the standard errors of the mean (SEM). For gene expression analysis, experiments were performed in triplicate (n = 3), and data are presented as means ± the SEM. Data were analyzed using Prism v7.0 (GraphPad Software, Inc., San Diego, CA) using one-way or two-way analysis of variance (ANOVA) with the Tukey post hoc test. A P value of <0.05 was considered significant for all statistical tests.

Metaboanalyst (version 5.0) online analysis software (https://www.metaboanalyst.ca; accessed 20 May 2021) was used to analyze metabolomic data. Briefly, samples were first normalized to the internal control and blank control. Processed data were then submitted to Metabolanalyst software for performing univariate and multivariate analyses. PCA and PLS-DA, combined with t tests (two-sample t tests and Wilcoxon rank sum tests) analyses, were used to screen the significantly differential metabolites. A P value of <0.05 was considered significant for all statistical tests. The model was evaluated by the cross-validation (leave-one-out cross-validation) method, using Q² as a performance measure. Clustering and pathway analysis was performed by generating a heat map using Euclidean distances and complete linkages with t test results.

The data presented in this study are available on request from the corresponding author.