Lactobacillus rhamnosus GG and Saccharomyces cerevisiae boulardii exert synergistic antipathogenic activity in vitro against enterotoxigenic Escherichia coli

Short-term colonic in vitro batch incubations were performed to elucidate the possible synergistic effects of Lactobacillus rhamnosus GG (CNCM-I-4798) and Saccharomyces cerevisiae boulardii (CNCM-I-1079) (associated in Smebiocta/Smectaflora Protect®) on the colonic microbial fermentation process, as well as their antipathogenic activity against enterotoxigenic Escherichia coli (LMG2092) (ETEC). These incubations adequately simulate the native microbiota and environmental conditions of the proximal colon of both adult and toddler donors, including the colonic mucosal layer. Results indicated that both strains were capable of growing together without showing antagonistic effects. Co-cultivation of both strains resulted in increased butyrate (stimulated by L. rhamnosus GG), propionate (stimulated by S. boulardii), and ethanol (produced by S. boulardii) production compared to the control incubations, revealing the additive effect of both strains. After inoculation of ETEC under simulated dysbiotic conditions, a 40 and 46% reduction in the concentration of ETEC was observed upon addition of both strains during the experiments with the adult and toddler donor, respectively. Furthermore, ETEC toxin levels decreased upon S. boulardii inoculation, probably due to proteolytic activity of this strain, with a synergistic effect being observed upon co-cultivation of L. rhamnosus GG and S. boulardii resulting in a reduction of 57 and 46% for the adult and toddler donor, respectively. Altogether, the results suggest that both probiotics together may help microbiota functionality, in both adults and toddlers and under healthy or impaired conditions, which could be of great interest when the colonic microbiota is dysbiotic and therefore sensitive to pathogenic invasion such as during antibiotic treatment.


Introduction
Probiotics are recognised for their role in improving human health, most likely by modifying the gut microbiota composition in a transient manner. Indeed, probiotics are defined as live microorganisms which when administered in adequate amounts confer a health benefit on the host (Hill et al., 2014). Among the most documented probiotic strains, Lactobacillus rhamnosus GG (Doron et al., 2005) has been associated with several health-promoting properties including prevention of antibiotic-induced diarrhoea (Vanderhoof et al., 1999), recovery from disease-related types of diarrhoea (Guandalini et al., 2000;Szajewska et al., 2011), prevention of atopic disease (Kalliomaki et al., 2001), improvement of immune function (Pena and Versalovic, 2003;Pessi et al., 2000) and antipathogenic activity (De Keersmaecker et al., 2006) due to the production of antimicrobial compounds. For example, the antimicrobial activity of L. rhamnosus GG against the intestinal pathogen Salmonella typhimurium has been linked to lactic acid secretion into the culture medium (De Keersmaecker et al., 2006). Another group investigating the antipathogenic effect of L. rhamnosus GG against the respiratory tract pathogen Moraxella catarrhalis obtained similar results (Van den Broek et al., 2018) Another extensively used healthpromoting microorganism is the non-pathogenic yeast Saccharomyces cerevisiae boulardii. Yeasts are resistant to antibiotics and therefore are frequently used as preventive Lactobacillus rhamnosus GG and Saccharomyces cerevisiae boulardii exert synergistic antipathogenic activity in vitro against enterotoxigenic Escherichia coli F. Moens 1 , C. Duysburgh 1 , P. van den Abbeele 1 , M. Morera 2 and M. Marzorati 1,3* 1 ProDigest bvba, Technologiepark 82, 9052 Ghent, Belgium;2 Ipsen Pharma SAS, 65 Quai George Gors, 92650 Boulogne Billancourt Cedex, France;3 Center of Microbial Ecology and Technology (CMET), Ghent University,Coupure Links 653,9000 Ghent,Belgium;massimo.marzorati@prodigest.eu and therapeutic agents for antibiotic-associated diarrhoea (Czerucka et al., 2007). Furthermore, S. boulardii has shown excellent survival through the upper gastro-intestinal tract, which is an important characteristic of probiotic products intended for oral consumption (Tompkins et al., 2011). Several mechanisms of action have been described for the health-promoting activity of S. boulardii, including regulation of intestinal microbial homeostasis, regulation of immune response, improvement of gut barrier function, antitoxin effects and antimicrobial activity (Kelesidis and Pothoulakis, 2012). S. boulardii has been shown to exert antipathogenic activity against several disease-causing Escherichia coli strains, such as enterohemorrhagic E. coli (EHEC) (Dahan et al., 2003;Dalmasso et al., 2006), enteropathogenic E. coli (EPEC) (Czerucka et al., 2000) and enterotoxigenic E. coli (ETEC) (Badia et al., 2012).
ETEC infection is recognised as one of the most common causes of traveller's diarrhoea and of childhood diarrhoea in developing countries (Anonymous, 2006). Furthermore, foodborne outbreaks of ETEC infection have emerged in developed countries, such as the USA (Dalton et al., 1999), Norway (MacDonald et al., 2015), Denmark (Pakalniskiene et al., 2009) and South Korea (Shin et al., 2016). One of the major virulence factors of ETEC infection includes adherence of the pathogenic strain to the epithelium of the small intestine, which is mediated by several colonisation factors including numerous fimbriate surface antigens (Gaastra and Svennerholm, 1996). Colonisation of the small intestine is followed by the release of either or both heat-stable and heat-labile enterotoxins, causing diarrhoea (Qadri et al., 2005). The resulting infection generally lasts for several days and ranges from illness with mild diarrhoea to severe cholera-like disease (Anonymous, 2006). Several studies have reported the potential role of probiotics in the prevention and/or treatment of ETEC infections (Roselli et al., 2006;Tsai et al., 2008). L. rhamnosus GG was shown to partially impair ETEC adhesion to intestinal epithelial cells in vitro, probably through competition for binding sites and/or secretion of a non-proteinaceous antimicrobial metabolite (Roselli et al., 2006). Additionally, L. rhamnosus GG exerted anti-inflammatory properties in response to ETEC-induced inflammation by regulation of chemokine and cytokine expression. Furthermore, S. boulardii was shown to prevent ETEC infection in vitro by preventing ETEC from adhering to intestinal epithelial cells and reducing pathogenic inflammation via secretion of anti-inflammatory factors (Badia et al., 2012). As such, co-administration of L. rhamnosus GG together with S. boulardii could therefore potentiate the beneficial effects of these probiotics on ETEC infection.
Further, since most studies regarding the interactions of probiotics with ETEC have been performed in vitro and focused on the small intestinal environment, only a limited amount of data exists on the interactions of ETEC with the colonic microbiota and probiotics (Roussel et al., 2017a,b). In humans, the colonic microbiota plays a predominant role on the possible invasion of ETEC and induction of disease (Pop et al., 2016). Indeed, microbiota compositional analysis revealed an association between the presence of 12 Operational Taxonomic Units (OTUs) in the microbiota and the absence of development of diarrheal disease upon ETEC challenge. Similar results were obtained during studies with piglets challenged with ETEC (Bin et al., 2018). Microbiome profiling has revealed that travellers who develop diarrhoea upon ETEC infection have a dysbiotic colon microbiota that is characterised by a high Firmicutes:Bacteroidetes ratio (Youmans et al., 2015). As such, modulation of the gut microbiota and its associated fermentation process through the ingestion of probiotics could be used as a strategy to protect against ETEC infection. Indeed, L. rhamnosus GG has been shown effective at preventing post-weaning diarrhoea in piglets, which was associated with modulation of the intestinal microbiota (Zhang et al., 2010). Modulation of the gut microbiota in newly weaned pigs through a select mixture of Bacillus species has been shown to improve ETECinduced enteritis symptoms (Zhang et al., 2017). Finally, pre-supplementation of the probiotic strain Lactobacillus reuteri HCM2 to mice prevents the dysbiosis of the colonic microbiota caused by ETEC through modulation of the gut microbial community (Wang et al., 2018).
The main objective of this study was to determine how L. rhamnosus GG and S. boulardii, when administered together, would affect the colonic microbial fermentation process and identify their respective and combined activity against ETEC. We hypothesised that these strains could possibly act synergistically when added together to a healthy and altered colonic microbiota environment in vitro.

Short-term colonic microbiota batch incubations
To account for inter-individual differences, two different donors, an adult and a toddler, were used as a source of colonic background community. Under healthy conditions for both donors, the short-term colonic batch incubations were performed in 120 ml penicillin bottles and initiated https://www.wageningenacademic.com/doi/pdf/10.3920/BM2019.0064 -Friday, February 21, 2020 3:02:14 AM -Universiteit Gent IP Address:157.193.240.3 by the addition of 63 ml colonic background medium containing host-and diet-derived nutritional compounds. Four types of colonic incubations were performed. During the control (CTRL) incubations only the colonic background microbiota was added. During the 'LGG' experiments 10 9 cfu of L. rhamnosus GG were dosed to the incubations together with the colonic background microbiota. During the 'SB' experiments 50 mg of S. boulardii powder was added to the colonic incubations containing a metabolically active colonic microbiota. Both L. rhamnosus GG, S. boulardii and a colonic background microbiota were dosed to the 'LGG + SB' incubations. Background microbiota were obtained from both a healthy adult donor and from a healthy 3-year old toddler donor. Briefly, faecal samples were freshly collected in containers containing an AnaeroGen sachet to maintain anaerobic atmosphere. A faecal inoculum was prepared by making a 1:13 (w/v) mixture of the faecal sample with anaerobic phosphate buffer (K 2 HPO 4 8.8 g/l; KH 2 PO 4 6.8 g/l; sodium thioglycolate 0.1 g/l; sodium dithionite 0.015 g/l). After homogenisation (10 min, BagMixer® 400, Interscience, Louvain-La-Neuve, Belgium) and removal of big particles via centrifugation (2 min, 500×g), an inoculum corresponding to 10% (v/v) was added to the different incubations. Furthermore, during all incubations the mucosal layer of the colon was simulated by the addition of mucus beads (Van den Abbeele et al., 2012). Briefly, mucus beads were prepared by covering microcosms (Kaldness K1, AnoxKaldnes AB, Lund, Sweden) with a mucin agar solution and combining them in a polyethylene netting (Zakkencentrale, Rotterdam, the Netherlands) at a ratio of 1 microcosm per 9 ml of colonic medium. All incubations were performed under strict anaerobic conditions. Anaerobiosis was obtained by flushing with N 2 . All experiments were incubated at 37 °C for a period of 48 h under continuous mixing (90 rpm). All experiments were performed in biological triplicate.
Under dysbiotic conditions, colonic incubations were performed in a similar way, with the following modifications: during all incubations, only 0.02% (v/v) of background colonic microbial community was added in order to simulate dysbiosis. Furthermore, 2% (v/v) of an overnight ETEC culture was added to all incubations, resulting in approximately 7.5 log ETEC/ml at the start of the incubations. ETEC was grown overnight in nutrient broth at 37 °C under normal atmospheric conditions and continuous shaking (110 rpm). The overnight culture was subsequently centrifuged at 1,500×g for 10 min, the supernatant was removed and the pellet was resuspended in sterile phosphate-buffered saline (PBS).

Microbial growth
Quantification of the viable microbial cells of L. rhamnosus GG, S. boulardii, and ETEC during the colonic batch incubations was determined by generating cell pellets at different time points during the incubation (baseline, 6 h, 24 h, and 48 h) which were subjected to propidium monoazide (PMA, Biotium, Hayward, CA, USA) treatment. Briefly, 50 µl of sample was added to 450 µl of sterile anaerobic peptone water. PMA was added to reach a final concentration of 50 µM. Samples were shaken in the dark for 5 min at room temperature. Afterwards the PMAtreated cell suspensions were placed in a LED-Active Blue system (Ingenia Biosystems, Barcelona, Spain) for 15 min followed by centrifugation for 5 min at 7,690×g. Cell pellets were stored at -20 °C. Subsequently, DNA was extracted from these PMA-treated cells using the DNeasy UltraClean Microbial Kit (QIAGEN, Venlo, the Netherlands) according to manufacturer's instructions, followed by quantification of their concentration through qPCR with a QuantStudio™ 5 Real-Time PCR System (Applied Biosystems, Foster City, CA, USA) using primers specific for L. rhamnosus GG (Brandt and Alatossava, 2003), S. cerevisiae (Zott et al., 2010) and ETEC (Taniuchi et al., 2012). The qPCR protocols of L. rhamnosus GG and S. cerevisiae were optimised in order to allow quantification of both strains in the presence of a complex background microbiota. The qPCR assays resulted in an efficiency of 77 and 85% for L. rhamnosus GG and S. cerevisiae respectively, with good reproducibility between biological replicated. The mucin agar of the mucus beads was harvested after 48 h of incubation. DNA extraction was performed on 0.25 g of mucus followed by specific qPCR to study the adhesion of the strains to the mucosal layer.

ETEC toxin determination
The concentration of heat-labile toxin produced by ETEC was determined after 0, 6, 24, and 48 h of colonic incubation through GM1 ELISA as described previously by Salimian et al. (2010) with some minor modifications. Absorbance was measured using the Synergy HT Microplate reader (BioTek, Bad Friedrichshall, Germany). Both the concentration of toxin present in the fermentation liquid as well as the concentration of toxin that was bound to the cell wall were determined by separating the supernatant and cell pellet originating from 1 ml of sample. The cell pellet was resuspended in PBS and sonicated prior to toxin quantification.

Microbial metabolic activity
Short-chain fatty acids (SCFA), lactate, ammonium and ethanol production were measured at the start of the incubation, and after 48 h. Lactate quantification was performed using a commercially available enzymatic assay kit (R-Biopharm, Darmstadt, Germany) according to the manufacturer's instructions. Ammonium analysis was performed using a KjelMaster K-375 device (Büchi, Hendrik-Ido-Ambacht, the Netherlands). Ammonium in the sample was liberated as ammonia by the addition of https://www.wageningenacademic.com/doi/pdf/10.3920/BM2019.0064 -Friday, February 21, 2020 3:02:14 AM -Universiteit Gent IP Address:157.193.240.3 32% NaOH. The released ammonia was distilled from the sample into a 2% boric acid solution. Finally, the ammonium in the distillate was determined by titration with a 0.02 M HCl solution. Short chain fatty acids (SCFA), including acetate, propionate, butyrate and branched SCFA (sum of isobutyrate, isovalerate and isocaproate), and ethanol were extracted by the addition of 0.1 g of NaCl and an acetonitrile mixture, containing 99.5% acetonitrile and 0.5% formic acid, in a 1:1 ratio. As an internal standard, 2-methyl hexanoic acid was added to all samples. Suspensions were centrifuged for 15 min at 15,080×g after which concentrations of SCFA and ethanol were determined using a GC-2014 gas chromatograph with flame ionisation (FID) (Shimadzu, 's-Hertogenbosch, the Netherlands). Nitrogen was used as a carrier gas at a constant flow rate of 2.19 ml/min and as a make-up gas. The injector and detector temperatures were set at 200 and 240 °C, respectively. The column temperature program was as follows: 0 min 40 °C; 18 min 220 °C and 23 min 220 °C. The injection volume was 1 µl and a split ratio of 40 was applied.

Agar well diffusion assay
First, second-generation cultures of ETEC, L. rhamnosus GG, and S. boulardii were prepared in Luria-Bertani (LB) medium, De Man-Rogosa-Sharpe (MRS) medium, and yeast extract peptone dextrose (YEPD) medium, respectively, by growing them during each generation for 24 h at 37 °C under normal atmospheric conditions. L. rhamnosus GG and S. boulardii cells were harvested and four aliquots were prepared: (1) 3 ml of the second-generation culture was centrifuged for 5 min at 1,500×g, the supernatant (pH of 4.4 for L. rhamnosus GG and 5.5 for S. boulardii) was collected and subsequently filtered (pore size of 0.2 µm) into a sterile recipient; (2) 3 ml of the secondgeneration culture was centrifuged for 5 min at 1,500×g, the supernatant was collected and pH adjusted to 6.5 using 1M NaOH or 1M HCl; (3) 3 ml of the second-generation culture was sonicated for 2 min; (4) 2 ml of the secondgeneration culture was centrifuged for 5 min at 5,000×g, and the pellet was resuspended in an equal amount of PBS. ETEC was prepared by centrifuging 2 ml of a secondgeneration culture for 5 min at 1,500×g. The resulting cell pellet was resuspended in 6 ml of sterile PBS to obtain an ETEC solution at 10 8 cfu/ml. The agar well diffusion assay was performed by first adding 0.5 ml of the resuspended ETEC culture to big petri dishes followed by the addition of 20 ml of YEPD agar medium. The plates were swung to mix ETEC equally over the agar plate. After drying the plates, were subsequently dried. Afterwards holes were punched in the solidified agar using a sterile Pasteur pipet to create wells. The four test aliquots of L. rhamnosus GG and S. boulardii were mixed with an equal amount of 2× concentrated molten MRS agar medium and 2× concentrated molten YEPD agar medium, respectively. The wells in the YEPD agar plates seeded with ETEC, were filled with the molten agar media containing the test solutions. The plates were allowed to dry and were subsequently incubated for 24 h at 37 °C under normal atmospheric conditions.

Statistics
All experiments were performed in biological triplicate. Statistical analysis was performed using the SPSS Statistics software, version 25 (SPSS, Chicago, IL, USA). Normality of data and equality of the variances were determined with a Shapiro-Wilk test and a Levene's test, respectively. For normally distributed data with equal variances a one-way ANOVA with a Bonferroni post-hoc test was used. For normally distributed data with unequal variances, a Welch test with a Games-Howell post-hoc test was conducted. For non-normally distributed data, a Kruskal-Wallis one-way ANOVA test with multiple post-hoc pairwise comparisons were performed. Bonferroni corrections were implemented on the significance levels of the multiple pairwise comparisons. In terms of statistics, the differences for all data discussed and indicated by 'P<0.05' were significant with a confidence interval of 95%.

Growth of Lactobacillus rhamnosus GG and Saccharomyces c. boulardii
Growth of L. rhamnosus GG and S. boulardii was measured in the presence of a healthy colonic background microbiota of both an adult ( Figure 1) and a toddler (Supplementary Figure S1) donor. Similar effects were observed for both donors. During control incubations, as well as during the incubations to which S. boulardii was added, low background levels of L. rhamnosus GG were detected throughout the incubation period. Addition of L. rhamnosus GG to the incubations (alone or in combination with S. boulardii) resulted in initial concentrations of 9.1 log cfu/ ml, followed by moderate increases up to a maximum concentration of 9.7 log cfu/ml throughout the incubation period. On the contrary, inoculating S. boulardii to the incubations resulted in initial concentrations of 7.7 log cfu/ ml when S. boulardii was inoculated alone, and 7.4 log cfu/ ml when S. boulardii was supplemented in combination with L. rhamnosus GG, followed by a reduction towards the end of the incubation period. In the adult model, significantly higher concentrations of S. boulardii were present after 24 and 48 h of incubation compared to the incubations that were performed with S. boulardii alone.
The possible adhesion of L. rhamnosus GG and S. boulardii to the mucosal layer was studied by the addition of mucincoated beads to the colon medium. Supplementation of L. rhamnosus GG resulted in the presence of 7.2 log cfu/ml for the adult donor and 6.8 log cfu/ml for the toddler donor after 48 h of incubation. Mucosal binding of S. boulardii was lower compared to L. rhamnosus GG, reaching concentrations of 3.5 log cfu/ml and 3.8 log cfu/ml for the adult and toddler donor, respectively. When L. rhamnosus GG and S. boulardii were supplemented together, a similar adhesion of the mucosal layer was observed.

Microbial activity of Lactobacillus rhamnosus GG and Saccharomyces c. boulardii under healthy and dysbiotic conditions
During the control incubations with the adult donor, an overall SCFA production of 86.5 mM, mainly attributed to acetate, propionate and butyrate, was observed under healthy conditions (Table 1), which indicated the high metabolic activity of the donor microbiota. When L. rhamnosus GG was added to the incubations, acetate levels significantly decreased compared to the control incubations. This coincided with statistically significant higher levels of butyrate during these incubations, i.e. 25.0 mM compared to 17.0 mM for the control incubations. During the incubations with S. boulardii significantly more propionate was produced compared to the control experiments. Administration of both L. rhamnosus GG and S. boulardii to the colonic microbiota resulted in higher concentrations of both butyrate and propionate compared to the control experiments, indicating the additive activity of both strains on the colonic microbiota.
Under dysbiotic conditions (Table 2), SCFA production was significantly lower during the control incubations compared to the healthy conditions (P=0.008 for acetate, P=0.009 for propionate and P<0.001 for butyrate), with the strongest reduction being observed for butyrate production, i.e. a reduction of 13.5 mM. Addition of L. rhamnosus GG, S. boulardii, or both had no major effect on SCFA production under dysbiotic conditions. Overall, similar trends in SCFA production were observed during the experiments performed with the toddler donor (Supplementary Table S1 for healthy conditions and Table  S2 for dysbiotic conditions), albeit not reaching statistical  Ammonium and branched chain fatty acids (BCFA) were produced during all experiments of the present study, indicating the occurrence of proteolytic fermentation. When comparing healthy and dysbiotic conditions for the control incubations, it was observed that ammonium and BCFA levels were significantly higher under healthy conditions in both donors tested, except for BCFA levels in the adult model where similar levels were observed between both conditions. Addition of L. rhamnosus GG to the colonic incubations did not result in increased production of ammonium and BCFA compared to the control incubations under both healthy (Table 1 for adult and Supplementary  Table S1 for toddler) and dysbiotic (Table 2 for adult and  Supplementary Table S2 for toddler) conditions. Under healthy conditions, supplementation of S. boulardii (alone or in combination with L. rhamnosus GG) resulted in a statistically significant increase in ammonium and BCFA production compared to the control experiments, for both donors tested. Under dysbiotic conditions, supplementation of L. rhamnosus GG together with S. boulardii resulted in statistically significantly higher ammonium production compared to the control experiments when using an adult (P<0.004) and a toddler (P=0.050) colonic background microbiota, whereas no statistically significant differences were observed for BCFA production.

Effect of Lactobacillus rhamnosus GG and Saccharomyces c. boulardii against ETEC
Determination of the growth of ETEC during the control incubations demonstrated that this pathogenic strain was capable to outgrow in the presence of the dysbiotic colonic background community of both the adult ( Figure  2) and toddler (Supplementary Figure S2)   ETEC was able to adhere to the mucus beads for both the adult ( Figure 3) and the toddler (Supplementary Figure  S3) donor, resulting in an average adhesion of 6.6 and 7.0 log cfu/g, respectively. L. rhamnosus GG and S. boulardii as well as their combination did not reduce ETEC binding for both donors.
The results obtained during the agar well diffusion experiments (Table 3) indicated that the culture supernatant of L. rhamnosus GG had an antimicrobial effect on ETEC. This was mainly due to the low pH of the supernatant (pH=4.4) since no inhibition zones were obtained when the culture supernatant was adjusted to pH 6.5. Large inhibition zones were obtained when the sonicated cell suspensions of L. rhamnosus GG were administered to the plates. Finally, metabolically active cells of L. rhamnosus GG also resulted in the occurrence of inhibition zones. The culture supernatant of S. boulardii had no antimicrobial effect against ETEC. On the contrary, the sonicated fraction of the S. boulardii culture resulted in inhibition zones through its antimicrobial activity against ETEC. Finally, also the metabolically active S. boulardii cells resulted in inhibition zones.
Determination of the concentration of toxin produced in the supernatant revealed that the growth of ETEC resulted in the production of its toxin over the course of the control incubations and this for both donors (Figure 4 for adult and Supplementary Figure S5 for toddler). Administration of L. rhamnosus GG to the incubations had no effect on the production of the toxin by ETEC in the supernatant. On the LGG

CTRL
LGG SB LGG+SB 0 2 4 6 8 10 12 log (counts/g)  Figure S4. rhamnosus GG together with S. boulardii had no effect on the concentration of toxin that was bound to the cell wall. The sum of the concentrations of the toxin in the supernatant and bound to the cell wall showed that L. rhamnosus GG and S. boulardii had a synergistic effect against toxin production by ETEC and this both for the adult donor (P=0.018) and for the toddler (P=0.055) donor, resulting in a reduction of 57.4 and 45.5%, respectively.

Discussion
To our knowledge, the well-documented strains L. rhamnosus GG and S. boulardii have never been studied together before. As both strains confer several health benefits, we aimed to study the interaction between L. rhamnosus GG and S. boulardii at concentrations which are in the range of usual therapeutic doses in humans, their modulatory effect on the colonic microbial functionality and the possible synergistic effects when both strains are added together into a healthy colonic environment using an in vitro model. Two different donors were used as a source of a colonic background community, and this to account for inter-individual differences. As the microbiome in toddlers is similar to that in adults (Voreades et al., 2014), both types of donors were selected for the current study.
Addition of L. rhamnosus GG to the colonic incubations resulted in a high concentration of this strain at the start of the experiments, followed by minor growth throughout the incubation period. Furthermore, determination of L. rhamnosus GG concentrations present on the mucin-coated beads revealed that this strain was capable to adhere to the mucosal layer of the simulated colon. Indeed, Kankainen et al. (2009) revealed that the surface of L. rhamnosus GG contains mucus-binding pili, which shows its ability to bind to the mucosal layer in the colon. Co-cultivation of L. rhamnosus GG together with S. boulardii resulted in comparable concentrations of L. rhamnosus GG in both the luminal and mucosal compartments of the incubations, which indicated that the activity of S. boulardii had no negative impact on L. rhamnosus GG in the lumen and that S. boulardii did not compete with L. rhamnosus GG for adhesion sites in the mucosa. Dosing of S. boulardii to the incubations resulted in high concentrations of this strain during the first 6 h of the experiments. Afterwards, levels of S. boulardii decreased and this especially after 48 h. This was probably due to competition for substrates with the colonic background microbiota and not due to competition with L. rhamnosus GG, as this occurred when S. boulardii was grown alone and similar effects were observed when both strains were added to the incubations. In the adult model, even significantly higher concentrations of S. boulardii were observed after 24 and 48h of incubation compared to the incubations that were performed with S. boulardii alone. This could indicate a protective effect of L. rhamnosus GG towards S. boulardii. Furthermore, S. boulardii was also capable of binding the mucin-coated beads present in the incubations and its adhesion was not affected by the presence of L. rhamnosus GG. Adhesion was lower compared with L. rhamnosus GG, which could be explained by the absence of mucus-binding pili for S. boulardii (Moré and Swidsinski, 2015).
Determination of the metabolites produced during these experiments revealed the additive effect of L. rhamnosus GG and S. boulardii when inoculated together to the colon under healthy conditions. Dosing of L. rhamnosus GG to the colonic microbiota resulted in the production of increased levels of butyrate compared to the control. Butyrate is a major energy source for colonocytes (Basson et al., 1996) and has been associated with several healthpromoting properties, such as reduction of oxidative stress, improvement of gut barrier function and enhancement of anti-inflammatory response (Hamer et al., 2008). As one of the main end metabolites of L. rhamnosus GG fermentation is lactate (De Keersmaecker et al., 2006), the observed increase in butyrate levels was probably attributed to stimulation of specific lactate-utilising, butyrate-producing bacterial species such as Anaerostipes caccae and Eubacterium hallii (Duncan et al., 2004). The utilisation of lactate by these microorganisms probably resulted in an absence of increase on lactate levels during the incubation period due to production and subsequent consumption of this transient metabolite (Moens et al., 2017). Furthermore, during conversion of lactate into butyrate, acetate is an essential co-substrate that needs to be consumed to complete the final reaction step in butyrate synthesis (Duncan et al., 2002), which was indeed observed by a reduction of acetate levels upon dosing L. rhamnosus GG to the experiments. Dosing of S. boulardii gave rise to the production of high concentrations of ethanol. Ethanol is known for its antimicrobial properties but is in general not a major end-metabolite produced during colonic fermentation processes. Therefore, ethanol production can mainly be attributed to the fermentative activity of S. boulardii during the aforementioned incubations. Furthermore, addition of S. boulardii to the colonic microbiota resulted in increased production of propionate throughout the course of the incubation, which could be linked to the decreased concentrations of luminal S. boulardii after 48 h of incubation. Indeed, the decreased S. boulardii concentrations suggested the occurrence of cell lysis resulting in the release of major cell wall components in the medium, which could be used as substrates by propionate-producing bacteria. Cuskin et al. (2015) reported that the propionate-producing Bacteroidetes species can utilise mannan-oligosaccharides, which are present in the outer cell wall of yeast. Hughes et al. (2008) reported that beta-glucans, another major cell wall component of yeasts (Manners et al., 1973), specifically stimulate propionate during colonic fermentation. Cocultivation of both L. rhamnosus GG and S. boulardii resulted in increased butyrate, propionate, and ethanol production compared to the control incubations indicating the cooperative effect of both strains.
Moreover, the antipathogenic effect of L. rhamnosus GG and S. boulardii against ETEC was investigated in a colonic dysbiotic environment. Dysbiosis was confirmed by the significantly decreased levels of SCFA and ammonium compared to healthy conditions. ETEC was capable to outgrow in the dysbiotic microbiota of both donors and to adhere to the mucosal surface. Addition of L. rhamnosus GG to the incubations resulted in lower concentrations of luminal ETEC at each sampling point, with the lowest concentration of ETEC observed when L. rhamnosus GG and S. boulardii were dosed together. The antimicrobial activity of L. rhamnosus GG was mainly attributed to the active production of toxic metabolites as large inhibition zones were obtained during the agar well diffusion assay in the presence of the culture supernatant at low environmental pH. The antimicrobial activity of L. rhamnosus GG was probably attributed to the production of lactate, as was already reported by De Keersmaecker et al. (2006). Lactate decreases the pH of the environment (Macfarlane and Englyst, 1986). Especially at low pH values, lactate can exert strong antimicrobial effects against pathogens, as protonated lactic acid can penetrate the microbial cell after which it dissociates and releases protons within the cell, resulting in acidification and microbial cell death (Alakomi et al., 2000;Raybaudi-Massilia et al., 2009;Stratford and Eklund, 2003). Therefore, the antimicrobial activity of L. rhamnosus GG could be attributed to lactate production under acidic environmental conditions. Administration of S. boulardii also resulted in antimicrobial activity against ETEC in the luminal environment. The sonicated fraction of the S. boulardii culture resulted in the largest inhibition zones, indicating that cell components of S. boulardii were toxic for ETEC. Additionally, dosing of S. boulardii to the experiments resulted in lower toxin concentrations after 48 h of incubation compared to the control experiments, with an increased effect being observed when L. rhamnosus GG and S. boulardii were co-cultivated. The effect on toxin production by S. boulardii could be explained by the secretion of proteolytic enzymes. Indeed, the production of a specific protease by S. boulardii has already been shown to inhibit the effects of Clostridium difficile toxins A and B in human colonic mucosa (Castagliuolo et al., 1999). Also, Buts et al. (2006) reported that S. boulardii produces a protein phosphatase that inhibited E. coli endotoxins in the small intestine of rats. Furthermore, Roussel et al. (2018)  ammonium and branched SCFA concentrations observed during the incubations with S. boulardii confirms the stimulation of proteolytic activity. Hence, addition of L. rhamnosus GG together with S. boulardii resulted in a stronger effect which was probably due to the fact that the growth of L. rhamnosus GG resulted in a faster and more pronounced depletion of carbohydrates thereby increasing the proteolytic activity of S. boulardii towards the toxin. This also explains why the toxin concentration initially increased during the experiments with S. boulardii to comparable concentrations as the control, as during the initial 24 h of the incubation carbohydrates were still present in the medium which stimulated saccharolytic fermentation by S. boulardii thereby minimising its proteolytic activity. While all antimicrobial activity against ETEC was observed in the luminal environment, both L. rhamnosus GG and S. boulardii were not capable to compete for binding sites with ETEC on the mucosa. However, the mucus layer generally creates a protective barrier (Cornick et al., 2015) and therefore one single dose of the test strains in the current experimental setup was probably not effective to observe antipathogenic effects in the mucosal environment. Whereas, L. rhamnosus GG and S. boulardii had a detrimental effect on ETEC in the lumen, they did not seem to exert these effects on the residing microbial community of both donors, as was seen by the stable SCFA levels compared to the control under these dysbiotic conditions. The activity of the strains even resulted in the production of additional butyrate (by L. rhamnosus GG) and ethanol (by S. boulardii) during the colonic incubations. Altogether, these results demonstrated the complementarity of the antipathogenic effects of these strains on ETEC, resulting in a significantly lower amount of ETEC toxin after 48 h.
Some potential shortcomings of the study design include the absence of analysis of the microbial community composition.
Indeed, in the current study, impact of L. rhamnosus GG and S. boulardii on the residing microbiota under both healthy and dysbiotic conditions was mainly investigated through assessment of microbial functionality. Analysis of microbial community composition might have provided further insight into the effect of both probiotic strains on the residing microbial community. Indeed, presence of specific microbial groups in the microbial community of the respective donors might have contributed to the observed synergistic effects and therefore investigation of the potential correlations between the observed effects and the initial microbial community composition might provide insight into the responsiveness of subjects to the probiotic formulation. Furthermore, the effect of both strains was examined upon administration of one single dose, while in vivo probiotics are generally consumed over several days or weeks. Inclusion of the long-term evaluation of repeated probiotic intake might have provided further insight on the effect of L. rhamnosus GG, S. boulardii and their combination against ETEC. Finally, in the current experimental setting, a dysbiotic microbial community was simulated by the administration of very low percentage of the background microbial community of the donors under investigation. The strength of this approach lies in the fact that the impact of inter-individual variation was minimised allowing for physiologically relevant comparisons between both healthy and impaired conditions. However, while dilution of the faecal inoculum will have resulted in a dysbiotic microbial community in a quantitative way, the qualitative microbiota composition might not be as accurate as would be observed in vivo.
In conclusion, the results indicated that L. rhamnosus GG and S. boulardii were capable to grow together and no antagonistic effects were observed between both strains. Co-cultivation showed or resulted in the complementary of both strains at the metabolic level, which potentiated the antipathogenic and antitoxin activity against ETEC. Altogether, the results reinforce the hypothesis that both probiotics together may help microbiota functionality, either in adults or toddlers and under healthy or impaired conditions. This effect could be of great interest under dysbiotic conditions such as during antibiotic therapy, when the colonic microbiota is more sensitive to invasion of pathogens. Future research is warranted to further explore this hypothesis.

Supplementary material
Supplementary material can be found online at https://doi. org/10.3920/BM2019.0064. Figure S1. Growth of Lactobacillus rhamnosus GG and Saccharomyces boulardii in microbiota colonic incubations from a toddler donor.