Enterohemorrhagic Escherichia coli O157:H7 Survival in an In Vitro Model of the Human Large Intestine and Interactions with Probiotic Yeasts and Resident Microbiota

Abstract

This is the first report on the fate of enterohemorrhagic Escherichia coli O157:H7 in simulated human colonic conditions. The pathogen was progressively eliminated from the bioreactor and did not modify the major populations of resident microbiota. The coadministration of the Saccharomyces cerevisiae CNCM I-3856 probiotic strain led to a significant increase in acetate production but did not reduce pathogen viability.

TEXT

Enterohemorrhagic Escherichia coli (EHEC) is a major food-borne zoonotic agent associated with outbreaks worldwide. Human contamination occurs mainly after consumption of raw or undercooked ground beef, vegetables, water, and dairy products contaminated by bovine feces (1). EHEC causes illnesses ranging from uncomplicated diarrhea to life-threatening complications, such as hemolytic-uremic syndrome. Most of the outbreaks and sporadic cases are caused by EHEC strains belonging to serotype O157:H7, but other serotypes can be involved. For instance, EHEC of the O104:H4 serotype was responsible for a large outbreak in June 2011 in Europe (2, 3). The virulence of EHEC strains is mainly associated with their ability to damage intestinal epithelial cells and produce Shiga toxins Stx1 and/or Stx2.

Although the terminal ileum and colon are assumed to be the main sites of EHEC colonization and pathology in humans (4, 5), data on bacterial survival and regulation of virulence genes in the human intestine are scant. Human studies are obviously unethical when pathogenic microorganisms are involved, and no animal model can reproduce the physiopathology of EHEC infection as a whole (6). Hence, for technical, economic, and ethical reasons, in vitro models offer a relevant alternative to in vivo studies (7, 8). EHEC survival was recently evaluated in a dynamic in vitro system reproducing the luminal conditions of the human stomach and small intestine (9), but no data are available on their viability in human large intestinal conditions. Recent studies have suggested that human intestinal microbiota may play a key role in the outcome of EHEC infection (10, 11). de Sablet et al. (10) showed, in vitro and ex vivo, using the cecal contents of human digestive microbiota-associated rats, that soluble factors released by the human microbiota repressed Stx synthesis in EHEC O157:H7. Conversely, DNase colicins produced by colicinogenic flora may enhance Stx production (11). Moreover, Kendall et al. (12) showed that ethanolamine, a major component of mammalian and bacterial membranes found in the large intestine, regulates O157:H7 virulence gene expression.

Since antibiotic therapy to treat EHEC infection is controversial (13, 14), probiotics are being investigated as an alternative strategy. Probiotics are defined as live microorganisms that resist digestion and reach the colon alive and that when administered in adequate amounts, confer a health benefit on the host (15). They can exert protective effects against enteric pathogens via direct antagonism, immunomodulation, and/or competitive exclusion (16). Probiotic yeasts have already shown beneficial effects in the control of EHEC infection. Saccharomyces cerevisiae var. boulardii (S. boulardii) reduces O157:H7 growth in ruminal fluids (17) and decreases proinflammatory pathways in EHEC-infected T84 human colonic cells (18, 19). Using an in vitro model of the human gastrointestinal tract, Etienne-Mesmin et al. (9) also showed that another probiotic yeast, S. cerevisiae CNCM I-3856, significantly decreases EHEC O157:H7 growth resumption in the distal parts of the small intestine, possibly through the combined action of ethanol and other inhibitory substances.

This study used the in vitro model of the human large intestine ARCOL (artificial colon) to (i) evaluate the survival of an EHEC O157:H7 strain in simulated human colonic conditions and (ii) investigate the effect of probiotic treatment with S. cerevisiae strains. In addition, the influence of both pathogen and probiotics on the main populations of resident microbiota was also assessed.

Fermentation in the ARCOL.

ARCOL is a one-stage fermentation system (Applikon, Schiedam, The Netherlands) that integrates the main parameters of the in vivo human colonic environment (20), including pH, temperature, supply of ileal effluents, retention time, anaerobiosis maintained by the sole activity of resident microbiota, and passive absorption of water and fermentation metabolites by dialysis fibers (Table 1). The bioreactor was inoculated with fresh feces from a healthy volunteer who had no history of antibiotic treatment 3 months before the study, and it was used under semicontinuous conditions.

Table 1.

Parameters of in vitro fermentations in the ARCOL system

Parameters of in vitro fermentation	Exptl conditions
Temp	37°C
Continuous stirring	400 rpm
pH	6.3
Pressure	1.1 bar
Vol of fermentative medium	450 ml
Retention time	36 h
Redox potential	−400 mV

Two fermentations were carried out (Fig. 1). Both started after a 3-day stabilization phase (phase A) and were divided into three different 3-day phases (B, C, and D). In fermentation 1 (F1), E. coli O157:H7 (an isogenic mutant of the reference strain EDL 933 lacking the stx₁ gene but harboring stx₂) (21) was first administered alone (phase B), then coadministered with S. boulardii (Biocodex, Gentilly, France) (phase C), and finally coadministered with S. cerevisiae CNCM I-3856 (Lesaffre Human Care, Milwaukee, WI) (phase D). In fermentation 2 (F2), phases C and D were inverted to ensure that there was no carryover of treatment effects between phases. An aerobic culture (LB, 24 h, 37°C) of E. coli O157:H7 was introduced in the vessel (10⁵ CFU/ml of colonic medium) (9), whereas yeasts were supplied in active dried powder form (10⁷ CFU/ml of colonic medium) (22). Samples were regularly collected from the colonic medium and plated onto selective media (Table 2) to determine O157:H7 and probiotic yeast survival kinetics. Survival rates were assessed by comparing the profiles obtained for bacteria and yeasts with that of a theoretical marker which simulates the behavior of an inert (i.e., nondegraded and nonabsorbed) compound. Its removal from the bioreactor was described by C_t = C₀ × e^(−t/τ), where C_t is the concentration of the marker at time t, C₀ its initial concentration, t the time of fermentation, and τ the residence time (23). In parallel, the main bacterial populations of human intestinal microbiota were measured by plate counts (Table 2) and real-time quantitative PCR (qPCR) analysis. Total genomic DNA was extracted from 250 μl colonic medium by using the two first steps of Yu and Morrison's protocol (24) and the QIAamp DNA stool minikit (Qiagen, Courtaboeuf, France) according to the manufacturer's instructions. Specific primers used in this study are listed in Table 3. PCR amplification and detection were performed with a Stratagene Mx3005P QPCR system (Agilent Technologies, Massy, France). Ethanol concentrations in the colonic medium were determined by an enzymatic UV test (Biosentec, Toulouse, France). Dialysis outflow was daily sampled to determine short-chain fatty acid (SCFA) production by gas chromatography, as described by Gérard-Champod et al. (30). For all survival kinetics, significant differences between treatments were tested by repeated-measure two-way analysis of variance (ANOVA) followed by a Bonferroni correction for multiple comparisons. For data on intestinal bacterial populations and SCFA production, a Kruskal-Wallis test was performed followed, if the results were significant, by a Mann-Whitney test for pairwise comparisons (GraphPad Software, Inc., La Jolla, CA). Values of P < 0.05 were taken as indicating statistically significant differences.

Fig 1.

Schedule of in vitro fermentations in the ARCOL system.

Table 2.

Selective media used^a

Microorganism	Selective medium	Incubationb
E. coli EDL 933 Δstx1	Luria-Bertani + 100 μg/liter kanamycin	1 day, 37°C, A
Yeasts	Sabouraud dextrose agar + 50 mg/liter chloramphenicol + 10 mg/liter gentamicin	3 days, 30°C, A
Total anaerobes	Brain heart infusion	7 days, 37°C, AN
Total aerobes	Brain heart infusion	3 days, 37°C, A
Clostridia/Eubacteria	Reinforced clostridial medium	7 days, 37°C, AN
Bacteroides	Bacteroides mineral salt	7 days, 37°C, AN
Bifidobacteria	Beerens	4 days, 37°C, AN
Enterobacteriaceae	MacConkey	1 day, 37°C, A
Enterococci	d-Coccosel	2 days, 37°C, A

^aOverview of the selective media used to enumerate E. coli O157:H7, probiotic yeasts, and the main cultivable bacterial populations of the human intestinal microbiota.

^bCultures were incubated aerobically (A) or anaerobically (AN).

Table 3.

Primer and probe sequences used in real-time qPCR assays

Name	Sequence 5′–3′	Target	Annealing temp (°C)	Reference
SYBR green
BAC338F	ACTCCTACGGGAGGCAG	Total bacteria	58	25
BAC516F	GTATTACCGCGGCTGCTG
789cfbF	CRAACAGGATTAGATACCCT	Bacteroidetes	61	26
cfb967R	GGTAAGGTTCCTCGCGTAT
Act920F3	TACGGCCGCAAGGCTA	Actinobacteria	61	26
Act1200R	TCRTCCCCACCTTCCTCCG
928F-Firm	TGAAACTYAAAGGAATTGACG	Firmicutes	61	26
1040FirmR	ACCATGCACCACCTGTC
Eco1457F	CATTGACGTTACCCGCAGAAGAAGC	Enterobacteriaceae	63	27
Eco1652R	CTCTACGAGACTCAAGCTTGC
F_Lacto05	AGCAGTAGGGAATCTTCCA	Lactobacillus/Pediococcus/Leuconostoc	60	28
R_Lacto04	CGCCACTGGTGTCTYTCCATATA
TaqMan
F_Bact 1369	CGGTGAATACGTTCCCGG
P_TM1389F	FAM-CTTGTACACACCGCCCGTC-TAMRA	Total bacteria	60	28
R_Prok1492R	TACGGCTACCTTGTTACGACTT
E. coli-F	CATGCCGCGTGTATGAAGAA
E. coli-P	FAM-TATTAACTTTACTCCCTTCCTCCCCGCTGAA-TAMRA	Escherichia coli	60	29
E. coli-R	CGGGTAACGTCAATGAGCAAA
F_Bifid 09c	CGGGTGAGTAATGCGTGACC
P_Bifid	FAM-CTCCTGGAAACGGGTG-TAMRA	Bifidobacteria	60	28
R_Bifid 06	TGATAGGACGCGACCCCA
F_Bacter 11	CCTWCGATGGATAGGGGTT
P_Bac303	YY-AAGGTCCCCCACATTG-TAMRA	Bacteroides/Prevotella	60	28
R_Bacter 08	CACGCTACTTGGCTGGTTCAG

Survival of E. coli O157:H7 in simulated human colonic conditions.

This study is the first report of EHEC O157:H7 survival kinetics in a one-stage fermentation system reproducing human large intestinal conditions (Fig. 2). The pathogen was progressively eliminated from the bioreactor, more rapidly than the theoretical transit marker. After 36 h fermentation, 4.2 ± 0.03 log₁₀ CFU/ml of viable bacteria (n = 2) was recovered in the bioreactor, i.e., less than 5% of the initial intake versus 35% for the marker. These results indicate that the pathogen did not persist in this environment at a level similar to that of the marker, probably due to the barrier effect of intestinal microbiota. Resident microbiota can prevent the establishment of intestinal pathogen by producing antimicrobial substances, such as bacteriocins, or through competition for nutrients or ecological niches (31). In particular, E. coli strains producing colicins or microcins have been reported to be effective in vitro in inhibiting E. coli O157:H7 (32, 33). EHEC is considered to be a colonic pathogen. However, in the present study, E. coli O157:H7 does not persist in the colonic medium. Previous findings (4, 9, 34, 35) suggest that the distal small intestine would play a major role in EHEC infection, whence bacteria can spread to the colon as colonization becomes established.

Fig 2.

Survival of E. coli O157:H7 during in vitro fermentations in the ARCOL system and influence of probiotic treatment. The curves obtained for bacteria alone (□) or coadministered with S. boulardii (×) or S. cerevisiae CNCM I-3856 (▽) were compared with that of the theoretical transit marker (●). Results are expressed as means (log₁₀ CFU/ml) ± standard deviations (n = 2 for bacteria, n = 6 for theoretical transit marker).

Yeast viability in the ARCOL system.

As survival in the human gastrointestinal tract is a key feature of probiotic strains, yeast viability was followed during large intestine in vitro fermentations. Both strains were rapidly eliminated from the bioreactor (Fig. 3). However, S. cerevisiae CNCM I-3856 was significantly (P < 0.05) more affected by colonic conditions than S. boulardii, no viable yeast being recovered after 36 h and 60 h, respectively. S. boulardii is a variant of S. cerevisiae with specific genomic and phenotypic characteristics, including optimal growth at body temperature. Consistent with our results in colonic conditions, S. boulardii seems more resistant than S. cerevisiae in a simulated human gastric environment (36, 37). The extensive elimination of yeasts observed in vitro has already been described in humans (38, 39), supporting the usefulness of the ARCOL model. To maximize the number of viable yeasts in the human intestine and so maybe potentiate their probiotic effect, repeated administration of yeasts could be considered (20).

Fig 3.

Survival kinetics of S. boulardii (□) and S. cerevisiae CNCM I-3856 (▽) during in vitro fermentations in the ARCOL system. The results obtained for yeasts were compared with that of the theoretical transit marker (●). Results are expressed as means (log₁₀ CFU/ml) ± standard deviations (n = 2 for yeasts, n = 4 for theoretical transit marker).

Influence of probiotic treatment on EHEC survival.

When yeasts were coadministered with the pathogen, the survival rate of E. coli O157:H7 in the ARCOL (Fig. 2) was not modified (P > 0.05). Accordingly, the antagonistic effect observed in vitro with S. cerevisiae CNCM I-3856 in the distal small intestine (9) was not found here in human colonic conditions. This difference may be explained by (i) the low survival of yeasts in colonic medium, (ii) the lack of resident microbiota in the gastric and small intestinal model used by Etienne-Mesmin et al. (9), and (iii) higher concentrations in ethanol, which may act as an inhibitory agent (40), in the small intestinal compartments of this model (up to 0.6 g/liter) compared with those measured in colonic medium in the present study (0.05 g/liter).

Influence of E. coli O157:H7 and probiotic treatment on the main populations of human intestinal microbiota.

The major phyla of gut microbiota and their main members (41, 42) were counted throughout fermentations in the ARCOL (Table 4). During the stabilization phase (phase A), their concentrations were similar to those described in humans (43, 44) or in other in vitro models of the human gut (45, 46).

Table 4.

Influence of E. coli O157:H7 and probiotic treatment on human intestinal microbiota

Assay and target bacteria or population	Analysisa
	Feces	Phase A	Phase B	Phase C	Phase D
SYBR assay
Total bacteria	10.1 ± 0.2	10.7 ± 0.2a	10.8 ± 0.1a	10.3 ± 0.2b	10.4 ± 0.3ab
Bacteroidetes	9.0 ± 0.1	9.7 ± 0.1a	9.4 ± 0.4a	7.6 ± 0.5b	8.1 ± 0.8b
Actinobacteria	7.2 ± 0.1	7.0 ± 0.2a	6.9 ± 0.1a	6.5 ± 0.2b	6.2 ± 0.3b
Firmicutes	10.0 ± 0.1	10.2 ± 0.1ac	10.4 ± 0.2a	10.1 ± 0.1b	10.1 ± 0.2bc
Enterobacteriaceae	6.2 ± 0.4	6.3 ± 0.2a	5.7 ± 0.3b	5.7 ± 0.3b	5.7 ± 0.1b
Lactobacillus/Pediococcus/Leuconostoc	5.0 ± 0.4	5.4 ± 0.1a	5.5 ± 0.2a	5.3 ± 0.2a	5.3 ± 0.1a
TaqMan assay
Total bacteria	9.9 ± 0.1	10.4 ± 0.1a	10.4 ± 0.3ac	9.9 ± 0.1b	10.0 ± 0.3bc
Bifidobacteria	7.8 ± 0.2	7.6 ± 0.3a	5.1 ± 2.4b	3.0 ± 1.6b	4.7 ± 1.4b
Bacteroides/Prevotella	9.3 ± 0.1	10.1 ± 0.1a	9.8 ± 0.3a	8.6 ± 0.4b	8.9 ± 0.8b
Escherichia coli	6.2 ± 0.4	6.3 ± 0.1a	4.8 ± 1.4b	5.3 ± 0.5b	4.2 ± 1.8b
Bacterial populations
Total anaerobes	9.5 ± 0.0	9.9 ± 0.4a	9.9 ± 0.1a	9.8 ± 0.1a	10.0 ± 0.1a
Clostridia/Eubacteria	9.4 ± 0.0	10.0 ± 0.6a	9.9 ± 0.2a	9.8 ± 0.1a	10.0 ± 0.1a
Bacteroides	8.1 ± 0.0	8.2 ± 0.1a	7.8 ± 0.5a	8.2 ± 0.0a	8.5 ± 1.0a
Bifidobacteria	8.2 ± 0.0	8.0 ± 0.3a	7.0 ± 0.5b	6.0 ± 0.2c	6.1 ± 0.2c
Total aerobes	5.6 ± 0.1	5.8 ± 0.5a	5.3 ± 0.4a	5.8 ± 0.2a	5.9 ± 0.7a
Enterococcus	4.5 ± 0.3	4.6 ± 0.3a	4.2 ± 0.4a	4.9 ± 0.9a	4.1 ± 0.3a
Enterobacteriaceae	5.5 ± 0.1	5.8 ± 0.4a	4.8 ± 0.3b	4.5 ± 0.4b	4.7 ± 0.3b

^aMajor phyla of gut microbiota and their main members were measured by plate counts (results shown in 1og₁₀ CFU/ml) and/or real-time qPCR analysis (results shown in log₁₀ mean copy number/ml). Data are means ± standard deviations for the 3 days of each phase during F1 and F2 assays (n = 6 for phases A, B, C, and D). Samples with different letters in a row are significantly different according to Kruskal-Wallis test followed, if the results were significant, by Mann-Whitney test; P < 0.05. Phase A, stabilization; phase B, E. coli O157:H7; phase C, E. coli O157:H7 plus S. boulardii; phase D, E. coli O157:H7 plus S. cerevisiae CNCM I-3856.

No study had hitherto evaluated the influence of EHEC O157:H7 on human intestinal microbiota. In phase B of both assays (F1 and F2), no change was observed by cultural and molecular approaches after addition of EHEC O157:H7, except for bifidobacteria, Enterobacteriaceae, and E. coli (P < 0.05). Each administration of E. coli O157:H7 was followed by a decrease in concentrations of Enterobacteriaceae (Fig. 4). Our results suggest that EHEC colonization of the large intestine is not associated with any major modifications in the profile of dominant bacterial populations. The decrease in Enterobacteriaceae may be explained by Shiga toxin-encoding phage transfer from E. coli O157:H7 to commensal E. coli resulting in bacterial lysis. According to Gamage et al. (47), about 10% of the normal E. coli in the human intestine is sensitive to infection by Shiga-toxin-encoding phage.

Fig 4.

Survival kinetics of E. coli O157:H7 and Enterobacteriaceae during in vitro fermentations in the ARCOL system. E. coli O157:H7 and Enterobacteriaceae were regularly counted (log₁₀ CFU/ml) during phases A and B of F1 (closed symbol) and F2 (open symbol) assays. *, addition of E. coli O157:H7 (10⁵ CFU/ml of colonic medium); phase A, stabilization; phase B, E. coli O157:H7.

When probiotic yeasts were coadministered with the pathogen (phases C and D), no profound change was observed in gut microbiota compared to phase B. The only modifications identified by qPCR were slight decreases in the levels of Bacteroidetes, Actinobacteria, and Firmicutes (P < 0.05). Our results are consistent with those obtained in healthy human volunteers, showing that S. boulardii and S. cerevisiae do not alter radically the dominant bacterial groups of fecal microbiota (39, 48).

Influence of E. coli O157:H7 and probiotic treatment on the metabolic activity of human intestinal microbiota.

To further investigate the effects of both pathogen and probiotics on intestinal microbiota, its metabolic activity was followed by assessing the production of major and minor SCFAs in the dialysis outflow of the ARCOL (Table 5). Whichever the treatment, similar trends were observed for SCFA production and acetate was the predominant SCFA, followed by propionate and butyrate. The molar ratios of acetate-propionate-butyrate obtained in vitro were consistent with that measured in humans (49, 50). The coadministration of S. boulardii (phase C) led to no significant change in SCFA production. In contrast, when EHEC O157:H7 and S. cerevisiae CNCM I-3856 were coadministered (phase D), the production of acetate was significantly increased (P < 0.01) and that of butyrate significantly decreased (P < 0.05) compared with phases A and B. Hitherto, S. boulardii was known to have no effect on SCFA profiles in healthy humans but was found to increase butyrate and acetate production in patients with diarrhea who were on long-term total enteral nutrition (51). Although the increase in acetate production with S. cerevisiae CNCM I-3856 was not linked to any decrease in E. coli O157:H7 viability, appropriate manipulation of the SCFA levels through probiotic treatment may be a potentially useful approach in the fight against EHEC infection. Indeed, in mice, the anti-infectious activity of bifidobacteria against E. coli O157:H7 has been thought to be related to an increase in acetate production, leading to either (i) a lower gut pH (52) or (ii) inhibition of Stx production and translocation (53). Interestingly, S. cerevisiae CNCM I-3856, when coadministered with E. coli O157:H7, led to a decrease in butyrate concentrations, while this SCFA has been shown to enhance the expression of virulence-associated genes in EHEC (35). The ARCOL system would potentially enable us to determine what experimental conditions could lead to high acetate and low butyrate production when probiotic yeasts are added, with the aim of inhibiting EHEC O157:H7.

Table 5.

Influence of E. coli O157:H7 and probiotic treatment on SCFA production^a

SCFA	Amt during phaseb:
	A		B		C		D
	mmol/h	%	mmol/h	%	mmol/h	%	mmol/h	%
Totalc	9.4 ± 0.7a		8.3 ± 1.0a		8.2 ± 1.4a		9.4 ± 0.4a
Acetate	5.8 ± 0.3a	62.2α	5.3 ± 0.8a	63.9α	5.4 ± 1.1a	65.2α	6.6 ± 0.3b	70.2β
Propionate	1.9 ± 0.3a	19.7α	1.5 ± 0.3a	17.9α	1.4 ± 0.2a	17.5α	1.5 ± 0.2a	16.1α
Butyrate	1.7 ± 0.3a	18.1α	1.5 ± 0.2a	18.2α	1.4 ± 0.3ab	17.3α	1.3 ± 0.1b	13.7β
iso-Butyrate	0.2 ± 0.02a		0.2 ± 0.01a		0.2 ± 0.03a		0.2 ± 0.01a
iso-Valerate	0.3 ± 0.04a		0.3 ± 0.01a		0.3 ± 0.1a		0.3 ± 0.01a
Valerate	0.4 ± 0.05a		0.4 ± 0.03a		0.3 ± 0.05a		0.4 ± 0.01a
Hexanoic acid	0.3 ± 0.05a		0.3 ± 0.1a		0.4 ± 0.1a		0.4 ± 0.1a
Heptanoic acid	0.1 ± 0.03a		0.1 ± 0.02a		0.2 ± 0.04a		0.2 ± 0.03a

^aSCFA in the dialysis outflow were measured by gas chromatography. Data are means ± standard deviations for the 3 days of each phase during F1 and F2 assays (n = 6 for phases A, B, C, and D). Samples with different letters in a row (a, b, and c for values in mmol/h; α and β for percentages) are significantly different according to Kruskal-Wallis test followed, if the results were significant, by Mann-Whitney test; P < 0.05.

^bPhase A, stabilization; phase B, E. coli O157:H7; phase C, E. coli O157:H7 plus S. boulardii; phase D, E. coli O157:H7 plus S. cerevisiae CNCM I-3856.

^cDefined as the sum of acetate, propionate, and butyrate.

In conclusion, this study is the first report on the fate of EHEC O157:H7 in simulated human colonic conditions. Our experiments provide new information on EHEC survival in the human gastrointestinal tract and its interactions with resident microbiota, which is essential for a full understanding of EHEC pathogenesis. There is growing interest in developing new strategies, such as the use of probiotics, in the fight against this pathogen. In our experimental conditions, S. cerevisiae strains did not exert any clear-cut antagonistic effect against E. coli O157:H7, but the present results open new fields in research on probiotics. The ARCOL system could be advantageously used to investigate the effect of resident microbiota on EHEC virulence or to screen probiotic strains for their ability to modulate pathogen infectivity in the human gastrointestinal tract. In a more holistic view of EHEC behavior in the human digestive environment, the ARCOL model should be used in combination with gastric and small intestinal systems (20, 54), if possible including a resident microbiota (55).