Lactobacillus delbrueckii ssp. bulgaricus B-30892 (LDB B-30892) was obtained from LacPro Industries, LLC, Fort Wayne, Indiana and was routinely grown anaerobically in Difco Lactobacilli MRS broth (Becton, Dickinson and Company [BD], Sparks, MD) or on MRS agar (BD) at 37°C. Clostridium difficile strain 9689 (CD-9689), a cytotoxin-producing strain, was purchased from the American Type Culture Collection (ATCC) and routinely grown anaerobically in Difco Reinforced Clostridial Medium (RCM; BD) or on RCM agar at 37°C. Six other commercially available probiotic and conventional lactobacilli strains; L. delbrueckii ssp. bulgaricus (LB-1 and LB-6), L. acidophilus (LB-2 and LB-3), and L. casei (LB-4 and LB-5 were isolated and biochemically characterized by API 50 CH system (bioMérieux, Hazelwood, MO) (Table 1). These lactobacillus cultures (LB-1 through LB-6, Table 1) were grown in MRS medium under the same conditions as LDB B-30892. In the present paper, Lactobacillus delbrueckii ssp. bulgaricus B-30892 is mentioned as LDB or LDB B-30892, C. difficile strain 9689 is mentioned as CD-9689 or CD and all other lactobacilli strains used were designated as LB or LB-1 through LB-6.

The human intestinal epithelial cell line, Caco-2 (HTB-37), was purchased from the ATTC and cultured in Dulbecco's Modified Eagle medium (DMEM; Gibco, Invitrogen Corp.) with 20% fetal bovine serum (FBS). This cell line has been well established as a model to study the intestinal barrier function 30. For the cytotoxicity assays, Caco-2 cells were grown to confluency in 96-well cell culture plates (Falcon, BD) at 37°C in 7% CO₂.

The effect of different lactobacilli on C. difficile induced cytotoxicity on a Caco-2 cell monolayer was tested using two groups of bacterial cell free supernatant (CFS) or bacterial conditioned medium (cell free) as described below:

The goal of the adhesion assay was to investigate the ability of the cell free conditioned medium (CFSs) obtained from LDB B-30892 and other lactobacilli to modulate the adhesion of CD-9689 to the intestinal epithelial cell monolayer in vitro. The experimental set-up for this was similar to the detoxification experiment using Caco-2 cell as model. The adhesion assay was done as described previously 323334 with some modification. In short, Caco-2 cells (1.1 × 10⁴ cells/ml) at the post-confluence stage were seeded on Lab-Tek II Chamber Slide Systems (Nalge Nunc International, USA). After incubation for 14 days at 37°C in 7% CO₂, the slides were washed with sterile PBS buffer and the test bacterial suspension of CD-9689 (with multiplicity of infection of 100 C. difficile cells to one Caco-2 cell; the C. difficile suspension medium in this case was a 1:1 volume/volume RCM-lactobacilli CFS) was added to each chamber with a Caco-2 monolayer. A 1:1 volume/volume RCM-MRS (without any lactobacillus conditioning) served as the control. The chambered slides were incubated at 37°C in 7% CO₂ for 2 h. After incubation, Caco-2 cells were washed four times with PBS and then immersed in PBS for 30 s. Cells were immersed in 1.0 ml of 100% ethanol for 5 min, air dried, and immersed in 1.0 ml of Giemsa staining solution (2.5 ml of KaryoMAX Giemsa staining solution [Invitrogen Corp., Carlsbad, Calif.] and 48.5 ml of 10 mM potassium phosphate buffer [20 mM KH₂PO₄, 20 mM K₂HPO₄; pH 6.8]), washed twice with distilled water, air dried again, and examined under a Leica DAS Mikroskop at a magnification of ×1,000. Counts of bacterial adhesion were taken at four to five random locations for a total of at least 150 Caco-2 cells, averaged, and statistically analyzed by the Duncan test by using SAS software (SAS Institute, Cary, N.C.). The inhibition of adhesion was calculated by taking the adhesion of CD-9689 without any treatment as 100 percent by the formula described.

Co-cultures of C. difficile (CD-9689) and Lactobacillus (LDB B-30892) on solid medium as well as in broth were done. In one such study, CD-9689 was streaked on RCM agar and cross-streaked with LDB B-30892. In some other experiments, CD-9689 and LDB B-30892 were co-cultured in RCM broth followed by selective plating on RCM and MRS agar plates, respectively. To study if the LDB-CFS caused any growth inhibition on CD-9689 cells, different ratios of LDB-CFS were added to RCM containing CD-9689 cells. All plates and tubes were incubated anaerobically at 37°C.

Data are expressed as mean ± Standard Error of Mean (SEM). Statistical analyses of data were performed using GraphPad Prism (version 3.02, GraphPad Software, San Diego, CA). Comparisons of cytotoxicity values between single control and CFS or CFT exposed sample means were made using two-tailed Student's t test or Tukey's pairwise comparison test. The limit for statistical significance was set at P < 0.05. Results of inhibition of adhesion assay were statistically analyzed by the Duncan test by using SAS software (SAS Institute, Cary, N.C.).

C. difficile-CFT causes a significant cytopathic effect on Caco-2 cells, presumably because of the production of extracellular cytotoxin(s) by this strain. Figure 1A shows the normal appearance of Caco-2 cells (untreated control). Caco-2 cells remain healthy following treatment with RCM (Figure 1B) and anti-toxin preparation from LDB B-30892 (LDB-CFS) (Figure 1C). While C. difficile cytotoxin after incubation with CFT from CD-9689-conditioned RCM for 24 h caused cytopathic effect (rounding and cell detachment) on Caco-2 cells (Figure 1D). The lack of a cytopathic effect of CFT from CD-9689-conditioned RCM after co-culturing CD-9689 and LDB B-30892 together for 24 h on Caco-2 cells was observed (Figure 1E). In contrast, Figure 1F shows the cytopathic effect when no LDB B-30892 were co-cultured with CD-9689 (same as depicted in Figure 1D). We observed the similar cytotopathic effects (as in Figures 1D and 1F) when heat killed LDB B- 30892 were incubated with CD-9689 (data not shown).

The lack of a cytotoxic effect shown in Figure 1E provided clues that CFT from CD-9689-conditioned RCM was no longer cytotoxic after co-culturing with LDB B-30892, possibly due to detoxification of the C. difficile cytotoxin(s) by LDB B-30892 or factor(s) produced by this organism. To verify this effect, we preformed a semi-quantitative assessment of C. difficile cytotoxin activity and its detoxification by LDB B-30892.

Table 2 shows representative results of the semi-quantitative assessment of C. difficile cytotoxin activity and its detoxification by LDB B-30892. The C. difficile-conditioned medium was cytotoxic up to a dilution of 1:8, while the highest concentrations of RCM, MRS, or LDB B-30892-conditioned medium (1:2) was non-cytotoxic and maintained normal Caco-2 cellular morphologies. Importantly, the LDB B-30892/C. difficile co-culture supernatant was also non-cytotoxic at its highest concentration (1:2). This detoxification by LDB B-30892 required living cells (which are metabolically active during the co-culturing with CD-9689 for 24 h) since adding heat-killed LDB B-30892 cells with C. difficile resulted in cytotoxicity equivalent to that of the C. difficile-conditioned medium alone.

Based on these initial in-vitro observations, we further investigated the following two issues: (i) Does LDB B-30892 release a bioactive component in the CFS, which results in the detoxification of C. difficile-mediated cytotoxicity; and (ii) Is the detoxification of C. difficile-mediated cytotoxicity unique to LDB B-30892?

To test the first question, LDB B-30892 and CD-9689 were grown separately in MRS and RCM broth, respectively, under the conditions mentioned earlier. CFS from the LDB B-30892 culture was added in different ratios to CFT from the CD-9689 culture (in a similar manner as in the semi-quantitative assay described above). We found that CFS from LDB B-30892 culture inhibited the cytotoxic effects of CFT from CD-9689 culture when added at a ratio of 1:1 through 1:8, and this inhibitory effect diminished in ratios lower than 1:16. These results indicate that LDB B-30892-conditioned CFS contains a bioactive component(s) or in other words, LDB B-30892 releases one or more extracellular components in the growth medium, which were responsible for inhibiting or deactivating the exotoxins released by C. difficile (in growth medium), thus protecting Caco-2 cells from C. difficile-mediated cytotoxicity.

So far, we observed that cell-free supernatant from co-cultured LDB/CD lacks cytotoxic property, also, when supernatants from individually grown culture of LDB B-30892 and CD-9689 were mixed together, the CFS and CFT mixture showed no cytotoxic effect. This observation indicates that CFS from LDB B-30892 resulted in the inhibition of cytotoxicity of CFT from CD-9689. To test whether the inhibitory property of the above-mentioned cytotoxicity of C. difficile is unique to LDB B-30892 or if it is typical to any lactobacillus, we tested six other commercially available probiotic or conventional lactobacillus strains (LB-1, LB-2, LB-3, LB-4, LB-5, LB-6; Table 1). CFS samples from the lactobacilli cultures (LB-1 through LB-6), either co-cultured with CD-9689 (LB/CD co-culture supernatant) or separately grown and then mixed together (LB/CD mixed CFS/CFT) were tested for cytotoxicity and compared with the results from LDB B-30892. We found no other lactobacillus tested was able to inhibit the cytopathic effect of CD-9689-conditioned CFT, in either co-culture or in CFS/CFT mixtures of LB/CD. We deduced the efficacy of LDB B-30892 to protect Caco-2 cells from the cytotoxic effect of CD-9689-conditioned CFT and compared that with other lactobacilli by a live-dead assay using fluorescence dyes propidium iodide (PI, red, dead cell indicator) and acridine orange (AO, green, live cell indicator) 31 (Figure 2 and Figure 3).

Figure 2A depicts control Caco-2 cells without any treatment; most cells are viable as they incorporated AO but not PI. Caco-2 cells treated with CFT from CD-9689 are shown in Figure 2B, where most of the cells were dead as indicated by the incorporation of PI by the vast majority of the cells. Figures 2C and 2D represent the protective effect of LDB B-30892-conditioned CFS on Caco-2 cells when co-cultured (Fig 2C) with CD-9689 or when separately cultured LDB-CFS was added to CFT from CD-9689 (Fig 2D). In both Figures 2C and 2D, the vast majority of the cells excluded PI (red fluorescence) indicating a higher live cell population. However, no other lactobacilli tested in the present study were able to detoxify CD-9689-conditioned CFT as similarly demonstrated by LDB B-30892 CFS. Figures 2E and 2F show the inability of LB-4, a L. casei isolated from a commercial probiotic yogurt and LB-1, a L. delbrueckii ssp. bulgaricus isolated from a probiotic supplement, to inhibit the cytotoxicity induced by CFT from CD-9689, as depicted by higher PI counts indicating that the majority of cells are dead.

Figure 3 shows the counts of live and dead Caco-2 cells after exposure to CFT from CD-9689, with or without treatments with different lactobacilli CFS. A mixture of cell-free extracts from CD-9689/LDB B-30892 at a ratio of 1:1 resulted in less than 4% cell death, while the mixture at a ratio of 4:1 caused only 12% cell death. No other probiotic organisms tested could result in such protection of Caco-2 cell death when the probiotic CFS aliquots were mixed to CD-9689-CFT.

The lack of a cytotoxic effect of CD-9689 in the presence of LDB B-30892-conditioned CFS on Caco-2 cells is thought to be due to a possible proteolytic activity of the antitoxic component released by LDB B-30892. Indeed, a similar detoxification of the C. difficile toxin A and B by a protease produced by Saccharomyces boulardii has been reported 3536.

The ability of different lactobacilli strains to inhibit the adhesion of enteropathogens varies significantly. For example, Ingrassia et al. 37 reported the ability of L. casei DN-114 001 to inhibit the adhesion of adherent-invasive Escherichia coli isolated from Crohn's disease patients. On the contrary, Gueimonde et al. (2006) reported that L. casei TMC 0409 actually increased the adhesion of C. diffcile ATCC 9689 (CD-9689) in a Caco-2 model. These apparent contradictory results can be found regularly in probiotic or gut microbiology literature. It is widely accepted that commensal or probiotic organisms prevent colonization of enteropathogens on the gut epithelial surface by competitive exclusion. Sharing of common carbohydrate-binding sites in probiotic organisms allows the blocking of adhesin receptors, thus promoting the inhibition of pathogen adhesion by steric hindrance 3238. Apart from whole cell bindings, another prominent mechanism which is believed to play a crucial role are the soluble factors (such as loosely adhered surface proteins of certain lactobacilli) released in the gut lumen which may cause the inhibition of adhesion or colonization of enteropathogens 3839. We were interested to investigate whether CFSs from different lactobacilli including LDB B-30892 contain such soluble factor(s) that may reduce the adhesion of CD-9689. It is evident from Figure 4, that culturing C. difficile (CD-9689) with CFS from LacPro's probiotic L. delbrueckii spp.bulgaricus (LDB B-30892) significantly reduced the adhesion of this pathogen to the Caco-2 cell monolayer. No other lactobacilli CFS tested showed an inhibition of adhesion of CD-9689 as efficient as LDB B-30892.

To investigate if the whole cell or CFS from LacPro's probiotic L. bulgaricus (LDB B-30892) were able to inhibit the growth of pathogenic C. difficile (CD-9689), we did a co-culture experiment on solid medium (trypticase soy agar, TSA) as well as in broth (RCM). Our result of growth inhibition study reveals that neither LDB B-30892 nor CFS from LDB B-30892 inhibited the growth or viability of CD-9689 (Figure 5). Since the CD-9689 tested grew well in presence of LDB B-30892, it appeared that no soluble and diffusible growth-inhibiting substances were released from this bacterium, as has been reported for another probiotic Lactobacillus 40. Although, recent reports have shown that L. casei inhibits the infection of Caco-2 cells by S. Typhimurium 41 and L. rhamnosus blocks epithelial barrier disruption by E. coli O157:H7 42 without producing any growth-inhibiting substances.