Inhibition of Cronobacter sakazakii in an infant simulator of the human intestinal microbial ecosystem using a potential synbiotic

Abstract

Powdered infant formula (PIF) can be contaminated with Cronobacter sakazakii, which can cause severe illnesses in infants. Synbiotics, a combination of probiotics and prebiotics, could act as an alternative control measure for C. sakazakii contamination in PIF and within the infant gut, but synbiotics have not been well studied for their ability to inhibit C. sakazakii. Using a Simulator of the Human Intestinal Microbial Ecosystem (SHIME^®) inoculated with infant fecal matter, we demonstrated that a potential synbiotic, consisting of six lactic acid bacteria (LAB) strains and Vivinal GOS, can inhibit the growth of C. sakazakii in an infant possibly through either the production of antimicrobial metabolites like acetate, increasing species diversity within the SHIME compartments to compete for nutrients or a combination of mechanisms. Using a triple SHIME set-up, i.e., three identical SHIME compartments, the first SHIME (SHIME 1) was designated as the control SHIME in the absence of a treatment, whereas SHIME 2 and 3 were the treated SHIME over 2, 1-week treatment periods. The addition of the potential synbiotic (LAB + VGOS) resulted in a significant decrease in C. sakazakii levels within 1 week (p < 0.05), but in the absence of a treatment the significant decline took 2 weeks (p < 0.05), and the LAB treatment did not decrease C. sakazakii levels (p ≥ 0.05). The principal component analysis showed a distinction between metabolomic profiles for the control and LAB treatment, but similar profiles for the LAB + VGOS treatment. The addition of the potential synbiotic (LAB + VGOS) in the first treatment period slightly increased species diversity (p ≥ 0.05) compared to the control and LAB, which may have had an effect on the survival of C. sakazakii throughout the treatment period. Our results also revealed that the relative abundance of Bifidobacterium was negatively correlated with Cronobacter when no treatments were added (ρ = -0.96; p < 0.05). These findings suggest that C. sakazakii could be inhibited by the native gut microbiota, and inhibition can be accelerated by the potential synbiotic treatment.

Keywords: 16S sequencing; Cronobacter sakazakii; gut model; metabolomics; synbiotic.

FIGURE 1

Schematic diagram of the infant triple SHIME. Tubing connections are only shown for SHIME 1 as an example, but SHIME 2 and 3 are connected in the same manner. Each SHIME compartment had one stomach/small intestine (ST/SI), proximal and distal colon vessel. Feed and contents from the proximal colon were transferred into each of the ST/SI vessels three times a day. Contents from the ST/SI vessels are then transferred into the proximal colon while contents from the proximal colon were transferred into the distal colon. HCl and NaOH were added into the vessels as needed to maintain a uniform pH range. Contents from the distal colon were transferred into a waste bucket for disposal. All vessels were kept at 37°C and consistently flushed with nitrogen to maintain an anaerobic environment. PJ refers to pancreatic juice.

FIGURE 2

Survival of C. sakazakii in different SHIME compartments and under different treatment conditions (n = 2). At all time points, C. sakazakii was plated on Brilliance C. sakazakii agar. The gray dotted line indicates the start of the second treatment period. Values are the average of C. sakazakii plate counts from the proximal and distal colon vessels ± the standard deviation.

FIGURE 3

Principal Component Analysis (PCA) biplot of metabolite profiles from SHIME 1 (A), SHIME 2 (B), and SHIME 3 (C) over a 2-week treatment period. Each sample (bubble) represents a different day during the treatment period. The 2-week treatment period is separated with approximately 1 week for each treatment period and labeled accordingly based on color (i.e., blue for the first treatment period and yellow for the second). MRS represents de Man, Rogosa and Sharpe media, LAB represents lactic acid bacteria, VGOS represents Vivinal GOS, and SCFA represents short-chain fatty acids. Arrows represent the different metabolites profiled and colored based on the type of metabolite.

FIGURE 4

Principal Component Analysis (PCA) biplot of metabolite profiles from all SHIMEs across two treatment periods. The labels indicate the SHIME and treatment period. For example, S1T1 indicates that the light blue samples are for the first treatment period of SHIME 1. Each SHIME is represented by a different color and treatment periods are differentiated by the shade of the color, i.e., light or dark shade.

FIGURE 5

Regression analysis of C. sakazakii levels with concentrations (mM) of acetate (A), ethanol (B), acetone (C), and glycine (D). There is a significant correlation between the concentrations of these metabolites and C. sakazakii levels (p < 0.05). Acetate is negatively correlated with C. sakazakii, whereas the other metabolites are positively correlated.

FIGURE 6

Microbial community composition of SHIME 1 (A), SHIME 2 (B), and SHIME 3 (C) before and throughout the treatment period. Microbial composition shown as average relative abundance and colored by genera. Relative abundances were based on 16S rRNA gene sequencing of SHIME samples from the distal colon vessels. Days –10 and –3 indicate the time points before the start of the first treatment period, as shown by a dotted line on day 0. Day 9 indicates the start of the second treatment period (solid line).

FIGURE 7

Shannon (top) and Gini-Simpson (bottom) indices for measurement of alpha diversity between SHIMEs (A,D), treatment periods (B,E), and treatments (C,F). Significant differences in alpha diversity, as denoted by the asterisks in (D), were calculated based on the Wilcoxon Rank Sum test and adjusted using the Benjamini-Hochberg false discovery rate (p < 0.05).

FIGURE 8

Principal coordinate analyses (PCoA) plots of unweighted UniFrac (A) and Bray-Curtis dissimilarity (B) based on operational taxonomical units from SHIME samples as visualized by treatment periods and SHIME, with ellipses indicating 80% confidence interval. The labels indicate the SHIME and treatment period. For example, S1T1 indicates that the light blue samples are for the first treatment period of SHIME 1. Each SHIME is represented by a different color and treatment periods are differentiated by the shade of the color, i.e., light or dark shade. Significant differences in groups were calculated based on PERMANOVA and adjusted using the Benjamini-Hochberg false discovery rate (p < 0.05).

FIGURE 9

Heatmap of Spearman’s rank correlation coefficients of select metabolites and microbial genera between the control (A), LAB (B), and LAB + VGOS (C) treatments. Microbial genera are colored in green, and metabolites are colored in purple. A positive correlation indicates that when a microbial genera or metabolite becomes more or less abundant, so does the other feature. Conversely, a negative correlation indicates that when a genera or metabolite becomes less abundant, the other feature increases in abundance. Statistical significance is indicated by the asterisks after correction with Benjamini-Hochberg false discovery rate, *P < 0.05, **P < 0.01.