The human colon adenocarcinoma cell-lines Caco-2 (ECACC #86010202), SW620 (ECACC #87051203) and SW480 (ECACC #87092801) were obtained from the European Collection of Authenticated Cell Cultures (Public Health Laboratory Service; Wiltshire, United Kingdom). Intestine-407 cells (ATCC #CCL-6) were obtained from the American Type Culture Collection (LGC Standards: Teddington, United Kingdom). All cell-lines were initially grown in T-75 tissue culture flasks until 80% confluent. Cells were trypsinized, and re-seeded at 1 × 10⁵ cells/well to 24-well tissue culture plates (Corning Costar, High Wycombe, United Kingdom) for infection assays. For luminescence assays, cells were seeded at 9 × 10⁴ per well in 96-well white bottom culture plates (Costar). Parallel cultures in 96-well clear bottom plates were used to assess confluency by light microscopy. All cell-lines were cultured in complete Dulbecco’s Modified Eagle’s Medium (DMEM) supplemented with 10% v/v foetal bovine serum (Invitrogen; Paisley, United Kingdom), 100 U/ml penicillin, 100 μg/ml streptomycin and 8 mM glutamine (Sigma Aldrich; Poole, United Kingdom), and maintained at 37°C in a humidified atmosphere of 5% CO₂/95% air for 24–48 h.

Intestinal organoids were utilised to assess epithelial cytotoxicity in response to C. difficile purified toxins, as per (Jones et al., 2019). Mice (C57BL/6J strain, at 10-weeks old) purchased from Charles River Ltd. (Margate, United Kingdom) and were maintained for 2 weeks prior to tissue harvesting on a standard CRM pelleted rodent chow diet (Special Diet Services; Witham, United Kingdom) with water available ad libitum, in a specific-pathogen-free environment at the Biomedical Services Unit (University of Liverpool). Mice were sacrificed, under Schedule 1 of the Animals (Scientific Procedures) Act 1986, by rising CO₂ levels, followed with cervical dislocation. Small intestine was dissected and transported on ice in sterile phosphate-buffered saline (PBS) pH 7.3, for use in establishing organoid cultures over 6 days, as previously described (Jones et al., 2019). On day 7, cultures were passaged by mechanically disruption through a 27G needle, resuspended in fresh Matrigel and plated out again to 24-well plates. Experimental treatment of organoids was only performed following a minimum of 1 passage.

Plantain (Musa x paradisiaca L.) non-starch polysaccharide (NSP) preparations from excipient-free plantain flour (Trobana Green Plantain flour AAB group (Horn)); Confoco International Ltd.; Ripley, United Kingdom) were prepared by Provexis plc (Reading, United Kingdom) at the Teagasc Food Research Centre (Moorepark, Ireland) as previously described (Roberts et al., 2010; Roberts et al., 2013). A range of other NSP were also prepared, selected to include representatives of dicots, commelinoid monocots and non-commelinoid monocots. This selection ensured a wide variety of NSP composition. The plants selected included those from the families Rosaceae—apple, pear and strawberry (dicots); Ericaceae—blueberry (dicot); Musaceae—plantain (green) (commelinoid monocot); Solanaceae—tomato and bell pepper (dicots); Poaceae—oat (commelinoid monocot); Fabaceae—runner bean (dicot); Brassicaceae—broccoli (dicot); Apiaceae—celery (dicot); and Amaryllidaceae—leek (non-commelinoid monocot). All were prepared to the same specification (Provexis plc); see Supplementary Table S1. Soluble NSP concentrations tested were within the range of effective luminal concentrations in the human distal colon that would be readily achievable with dietary supplementation (Roberts et al., 2010). Neutral and acidic polysaccharide fractions of soluble NSP from plantain were prepared by Q-Sepharose anion-exchange chromatography, as per (Parsons et al., 2014).

The laboratory C. difficile reference strain 1470 (ATCC #43598) was obtained from the American Tissue Type Collection (LGC Standard; Teddington, United Kingdom). C. difficile clinical laboratory isolate 080042 (PCR ribotype 027) used in initial intestinal epithelial cell interaction assays was obtained from Dr Godfrey Smith (Department of Medical Microbiology, Royal Liverpool and Broadgreen University Hospitals NHS Trust, United Kingdom). C. difficile 1342, a pathogenicity locus (PaLoc) negative strain, was provided by Drs Gillian Douce and Anthony Buckley (Institute of Infection, Immunity and Inflammation, University of Glasgow, United Kingdom), a strain having been originally obtained from Dr Kate Dingle (Oxford University, Oxford, United Kingdom) (Stoesser et al., 2011). Eleven additional C. difficile clinical isolates of differing toxin status and PCR ribotype, and from various geographical locations, were also obtained from the Royal Liverpool and Broadgreen University Hospitals NHS Trust; see Table 1. All isolates were initially grown in the laboratory under anaerobic conditions, at 37°C for 48 h, on Brain Heart Infusion (BHIS) agar supplemented with 0.1% w/v L-cysteine and 5 mg/ml yeast extract. Prior to infection of cultured intestinal epithelial cells, C. difficile strains were grown on Fastidious anaerobic agar (Oxoid), washed three times in sterile PBS pH 7.3, and resuspended to an OD_550nm, equivalent to 1 × 10⁹ CFU/ml.

Spores from five selected C. difficile clinical isolates (98011, 108536, 108906, 98220, and 1342) were prepared according to methods previously described (Permpoonpattana et al., 2011), with some modifications. Briefly, for each isolate, a single bacterial colony was selected from BHIS agar anaerobic culture (at 37°C for 48 h) and used to inoculate 10 ml TGY broth (3% w/v tryptic soy broth, 2% w/v glucose, 1% w/v yeast extract, 0.1% w/v L-cysteine) After overnight incubation in static conditions, TGY culture medium was sub-cultured (1:100) into Sorbitol MacConkey (SMC) broth (9% w/v/peptone, 0.05% w/v proteose-peptone, 0.1% w/v (NH₄)₂SO₄, 0.15% w/v/Tris, 0.1% w/v L-cysteine) and incubated at 37°C to an OD_600nm = 0.5. Aliquots of culture broth were used to inoculate a dozen SMC agar plates per strain, which were incubated under anaerobic conditions, at 37°C for 10 days. After day 10, spores were harvested and washed twice with sterile water, then suspended in 2 ml sterile PBS pH 7.3 containing 125 mM Tris, 200 mM EDTA, 0.3 mg/ml proteinase K and 1% v/v sarcosyl, and incubated at 37°C for 2 h with gentle shaking. Spores were then centrifuged (4,500 x g for 20 min) and the pellet resuspended in cold sterile distilled water. This was repeated a further 10 times. After the final wash, spores were heat-treated (60°C, for 20 min) to kill any residual vegetative cells, and then frozen at −20°C until use. To calculate spore numbers, aliquots were serially diluted in sterile PBS and plated onto BHIS agar supplemented with 0.1% w/v sodium taurocholate for spore germination (Sorg and Sonenshein, 2008; Burns and Minton, 2011). After 48 h, spore titres were enumerated by enumerating vegetative bacteria colony forming units (CFU). To confirm purity of each spore preparation, a Schaeffer and Fulton endospore stain was routinely performed (Schaeffer and Fulton, 1933). Results indicated that the purification process effectively and reproducibly yielded spore titres of a high purity, effectively free of vegetative cells (see Supplementary Figure S1; Supplementary Table S2).

Confluent Caco-2 cell monolayers in 24-well plates were pre-treated for 30 min with or without soluble NSP from plantain (0–20 mg/ml in antibiotic-free DMEM). Cell monolayers were then infected with vegetative C. difficile clinical isolates or purified C. difficile spore preparations (1 × 10⁷ CFU/ml), at a multiplicity of infection (MOI) of 100, for 2 h. Each monolayer was then washed three times with sterile PBS to remove non-adherent bacteria or spores, and monolayers were lysed with sterile 1% v/v Triton X-100. Epithelial adherent vegetative C. difficile and C. difficile spores were both enumerated by performing serial dilutions of cell lysates, plating of 20 µl aliquots of lysate to BHIS agar (inclusive of 0.1% w/v sodium taurocholate for spores) in triplicate. Plates were then incubated for 48 h at 37°C in anaerobic conditions, and CFU quantified.

In separate experiments, a variety of soluble dietary plant fibres, including NSP isolated from dicots such as apple, runner bean, blueberry, broccoli, celery, pear, bell pepper, strawberry and tomato, were compared to NSP from the non-commelinoid monocot leek, and the commelinoid monocots oat and plantain. Each NSP was screened at 5 mg/ml for ability to block C. difficile interactions with Caco-2 cell monolayers.

Additional experiments were also performed to determine whether action of plantain NSP to block C. difficile adhesion was via an interaction with the intestinal epithelial monolayer or through direct interaction with bacteria. To test the former, plantain NSP was added to Caco-2 cells 30 min prior to infection as described above, but then removed by three washes with sterile PBS (1 min each; at 37°C). Monolayers were then provided with fresh antibiotic-free DMEM, infected and levels of C. difficile adherent to cells assessed. To test for direct interaction with bacteria, plantain NSP was incubated with C. difficile for 30 min, followed by centrifugation, resuspension of bacteria in antibiotic-free media and inoculation of epithelial monolayers. To assess whether the inhibitory activity of soluble NSP from plantain to block C. difficile-host intestinal epithelium interactions lay within an acidic (pectin-rich) or neutral polysaccharide component, Caco-2 monolayers were pre-treated for 30 min with 5 mg/ml purified, desalted Q-Sepharose anion-exchange NSP fractions, as previously described (Parsons et al., 2014).

Alternative approaches were also used to assess the impact of plantain NSP on C. difficile adhesion to Caco-2 cells, including Giemsa staining coupled with light microscopy (as per (Roberts et al., 2013)) and flow cytometry (based on the method of Drudy and colleagues (Drudy et al., 2001). Briefly, C. difficile 080042 was cultured anaerobically in 50 ml Fastidious Anaerobe Broth (Lab M; Bury, United Kingdom) without shaking for 36 h, at 37°C. Bacteria were washed and resuspended in sterile PBS, pH 7.3. For fluorochrome labelling, 1 × 10⁸ bacteria/ml were incubated with 1 μM of 2′,7′-bis-(2-carboxyethyl)-5-(and-6)-carboxyfluorescein, acetoxymethyl ester (BCECF/AM) in the dark, at 37°C for 1 h. Excess fluorochrome was removed by 5 sequential washes in sterile PBS followed by centrifugation, each for 5 min. Bacteria were re-suspended in PBS containing plantain NSP, or PBS alone and incubated for 30 min in the dark, then added to 1 × 10⁶ Caco-2 cells in single cell suspension, at 37°C for 1 h. After incubation, cells were washed three times in PBS (300 x g, for 20 min) to remove any non-adherent bacteria. After the final wash, pellets were resuspended in 1 ml PBS and subjected to flow cytometry using a FACScan flow cytometer (Becton, Dickinson Ltd.; Wokingham, United Kingdom). A total of 10,000 events were acquired, with data analysed using Cell Quest software (Becton Dickinson). Mean fluorescence intensity (MFI) using the geometric mean of each sample was used to assess bacterial adherence to Caco-2 cells.

Cellular chloride secretion was assessed using a colorimetric assay that measures cellular iodide efflux as a surrogate for monitoring chloride ion channel activity, as per Tang and Wildey (2004). Caco-2 cells were seeded in 96-well plates at 2.5 × 10⁴ cells/well and incubated for 24 h at 37°C. Cells were then loaded with 100 µl pre-warmed iodine loading buffer and incubated for 4 h at 37°C. Following three washes with sterile PBS, and incubated for 30 min with plantain NSP (2.5–10 mg/ml) or chloride ion channel agonists, forskolin (200 μM; Sigma) or RP107 (100 μM; Sigma) as a positive controls (Noel et al., 2006). Following incubation, cells were lysed with 1% (v/v) Triton X-100 in deionised water. Intracellular iodine concentration was determined using the modified Sandell-Kolthoff reaction, and absorbances measured at OD_410nm (Tang and Wildey, 2004). Chloride channel blockers, cystic fibrosis transmembrane conductance regulator (CFTR) inhibitor Inh-172 (12.5–200 μM; Sigma) (Linsdell, 2014), or 5-nitro-2-(3-phenylpropylamino)benzoic acid (NPPB, 50–800 μM; Sigma) (Keeling et al., 1991), were evaluated for their ability to block cellular chloride channel activation, and were present during the 4 h iodine-loading step prior to incubation with soluble NSP from plantain, RP107 or forskolin.

Purified C. difficile TcdA and TcdB, isolated from the highly toxic strain reference strain VPI10463, were provided by Dr Clifford Shone (Public Health England; Porton Down, United Kingdom) (Maynard-Smith et al., 2014). Binary toxin (CDT) was generated from combination (1:2) of purified component CDTa and activated component CDTb (precursor CDTb digested with trypsin), produced and purified from a ribotype 027 strain, by Drs Panagiotis Papatheodorou and Carsten Schwan (Institute for Experimental and Clinical Pharmacology and Toxicology, Albert-Ludwigs University Freiburg, Germany) (Schwan et al., 2009; Aktories et al., 2018).

Serial dilution of each toxin was made in sterile culture media to achieve final test concentrations between 1–1,000 ng/ml. Caco-2 cells were seeded at 2 × 10⁵ cells/well in 6-well tissue culture plates and incubated for 24 h. Following three washes with sterile PBS, monolayers were pre-treatment for 30 min with or without 10 mg/ml soluble NSP from plantain. Cells were then treated with toxins for up to 24 h.

Intestinal organoids, on day 4 of culture, were pre-treated with 10 mg/ml plantain NSP or saline vehicle, followed by treatment with C. difficile toxins (1–100 ng/ml). Images were taken using an Axiovert 25 microscope (Zeiss) and a 20X objective lens with a Hitachi HV-C20A camera before treatment and 24 h post-toxin treatment. Organoid circularity (as a measure of cytotoxicity) was scored as previously described (Jones et al., 2019).

C. difficile toxin mediated effects were assessed by immunoblot using two monoclonal antibodies raised against Ras-related C3 botulinum toxin substrate (Rac1); clone 102 (BD Transduction Lab; Oxford, United Kingdom) previously shown to lose affinity for Rac1 following C. difficile toxin-mediated mono-glucosylation (Gerhard et al., 2008) and clone 238a (Abcam; Cambridge, United Kingdom) which recognises both non-glucosylated and mono-glucosylated Rac1 isoforms (Gerhard et al., 2008). Cells were lysed in ice cold 20 mM Tris-HCl buffer, pH 7.4. Whole-cell protein lysates (25 µg) were separated using 4–20% acrylamide gradient gels on SDS-PAGE at 150 V for 2 h, followed by electro-transfer to nitrocellulose membrane (100 V, for 1 h). Nitrocellulose membranes were blocked using 1% (v/v) bovine serum albumin (BSA) in PBS containing 0.1% (v/v) Tween-20 overnight at 4°C, then incubated with primary antibody (1:1,000) for 2 h at room temperature, followed by three 5 min washes with PBS-0.1% (v/v) Tween 20. Blots were incubated with horse radish peroxidase (HRP)-conjugated rabbit anti-mouse Immunoglobulin antibody (1:2,000) for 1 h at room temperature. Membranes were again washed three times with PBS-0.1% (v/v) Tween 20 and protein bands visualised with SuperSignal West Dura Extended Duration Substrate (Thermo-Fisher Scientific; Loughborough, United Kingdom) and quantified using Image Lab Software v3.0.1 (Biorad; Hemel Hempstead, United Kingdom).

Harvested media from human cell-lines was assayed for the presence of pro-inflammatory IL-8 using a solid-phase sandwich enzyme-linked immunosorbent assay (ELISA) (Human Elipair Kit; Abcam, Cambridge, United Kingdom). For murine intestinal organoid cultures, media was assayed for the CXCL1/KC chemokine using a Duoset mouse CXCL1/KC ELISA kit (R & D systems, Abingdon, United Kingdom). For both assays, capture antibody in 0.5 M carbonate-bicarbonate buffer (pH 9.6) was coated onto a 96-well high-binding γ-irradiated ELISA plates (Costar) overnight at 4°C. Following three washes with PBS-0.1% (v/v) Tween 20, plates were blocked with 1% (w/v) bovine serum albumin for 2 h at room temperature. Triplicate treatment samples (100 µl) of cell-free medium were then incubated overnight at 4°C, plates washed three times with 12.0 µl wash buffer, and the respective biotinylated detection antibody added for 2 h at room temperature. Subsequently, ExtrAvidin peroxidase (Sigma; Poole, United Kingdom) was added to each well, incubated for 20 min at room temperature. Plates were again washed three times with PBS-Tween, developed with SigmaFast o-phenylenediamine substrate (Sigma) for 10 min and colour development stopped with 100 µl 4 M H₂SO₄. ELISAs were measured on a Tecan Sunrise plate reader (Tecan; Reading, United Kingdom) at OD_492nm, with a reference filter set at OD_750nm. Treatment samples were quantified against calibration curves of recombinant human IL-8 (0–2000 pg/ml) and mouse KC (0–1,000 pg/ml) standards in culture medium.

Cells were seeded at 1 × 10⁴ cells/ml in clear bottomed white wall 96-well tissue culture plates (Corning/Costar) and cultured overnight at 37°C. Following treatment with purified toxins and/or C. difficile clinical isolates, cell monolayers were assayed using a commercial Caspase-Glo 3/7 assay (Promega; Southampton, United Kingdom), as per manufacturer instructions. Luminescence was detected using the Infinite F200 plate reader (Tecan). Experiments were performed in triplicate, with background luminescence determined from wells containing culture medium alone.

Epithelial cell viability during NSP treatments and in response to C. difficile infection or toxin treatment was monitored by measurement of adenylate kinase (normally found within intact cells) released to the culture medium from Caco-2 cells using a ToxiLight bioassay (Lonza; Slough, United Kingdom). Intestinal organoid cytotoxicity was measured as previously described (Jones et al., 2019).

Data are expressed as mean ± SD, unless otherwise stated. N numbers indicate the total number of independent experiments performed, where each experiment was performed with at least n = 3 replicates for any individual treatment group, unless otherwise indicated. Independent sample groups were first assessed for normality (Shapiro-Wilks test) and equality of variances (Levene’s test). Two-sample comparisons were performed using unpaired t-test or non-parametric Mann-Whitney U test as appropriate. Multiple treatment groups were analysed using either parametric one-way analysis of variance (ANOVA) or non-parametric Kruskal-Wallis test (Stats-Direct v3.0.171; Sale, United Kingdom). Differences were considered significant when p < 0.05.

Soluble NSP from plantain was observed to dose-dependently block adhesion of clinical C. difficile isolates to Caco-2 intestinal epithelial cell monolayers. Peak inhibition of C. difficile adhesion to Caco-2 cells was seen following pre-treatment with plantain NSP at 10 mg/ml for ribotype 027, toxin positive isolate C. difficile 80042 (64.2 ± 4.4% inhibition) was seen compared to untreated controls (N = 3, n = 3, p < 0.001, Kruskal-Wallis test), see Figure 1A. Plantain NSP also dose dependently reduced C. difficile adhesion of other C. difficile isolates, including the ATCC reference strain C. difficile 1470 to Caco-2 cells (peak inhibition seen at 10 mg/ml, 81.2 ± 2.4% in comparison to the untreated control (N = 3, n = 3, p < 0.001); Figure 1A. Using the optimum inhibitory dose of 10 mg/ml plantain NSP, similar levels of blockade of C. difficile 80042 interaction to epithelial cells was observed using two other colonic cell-lines SW620 (69.2 ± 3.0% inhibition) and SW480 (70.6 ± 1.4%) as was seen using Caco-2 cells (75.6 ± 5.1%), compared to untreated controls (N = 3; p < 0.001, Mann-Whitney U test), Figure 1B. Alternative approaches to measure bacterial adhesion, by flow cytometry (Figures 1C,D) and Giemsa staining coupled with light microscopy (Figure 1E), also revealed marked reduction in adhesion of C. difficile to Caco-2 cells in the presence of plantain NSP.

Plantain NSP (at 10 mg/ml) also significantly reduced adhesion of a panel of C. difficile clinical isolates differing in their geographical origin, toxin expression and PCR ribotype (Table 1) to Caco-2 cells, with an overall median reduction for all isolates tested of 73.8% (range 52.2–98.6%) in comparison to the untreated inoculated monolayers (Figures 2A–D, for all; p < 0.001; Kruskal-Wallis test). These results demonstrate that the inhibitory effect of plantain NSP is independent of C. difficile toxin expression and ribotype status.

To screen for inhibitory activity, Caco-2 cell monolayers underwent pre-treatment using soluble NSP from a range of dietary plants prior to infection with C. difficile 80042 (Figure 3A). Pre-incubation of intestinal cells with 5 mg/ml NSP from broccoli, leek and plantain resulted in the highest levels of inhibition of C. difficile 80042 adhesion to Caco-2 cells (>43% inhibition; p ≤ 0.01 (N = 3, n = 3–4 for broccoli NSP and plantain NSP; N = 1, n = 3 for leek NSP). NSP from runner bean, bell pepper or tomato was observed to be less effective (>22% inhibition; p < 0.05) whilst others such as apple, blueberry, celery, oat, pear and strawberry, had little or no inhibitory effect (N = 3, n = 3–4); see Figure 3A. Further study of leek NSP and broccoli NSP indicated a dose-dependent inhibitory effect on C. difficile-epithelial cell adhesion, with peak inhibition seen at 20 mg/ml (leek NSP, 57.5 ± 6.5%; and broccoli NSP, 56.9 ± 7.2% compared to untreated controls (100%); both p < 0.001). Both these dietary plant soluble fibres had a lower efficacy than that seen with plantain NSP; used as a positive control at 10 mg/ml, where the epithelial adhesion of C. difficile was reduced by 81.8 ± 4.6% (p < 0.001; N = 3, n = 3); see Figures 3B,C.

Plantain NSP appears to exert its inhibitory effect on C. difficile adhesion via action on the epithelium. When plantain NSP (10 mg/ml) was added to Caco-2 cell monolayers for 30 min, then removed by three washes with sterile PBS, levels of adherent C. difficile were observed to be significantly reduced (59.2 ± 5.0%) compared to the untreated inoculated controls (p < 0.01; Kruskal-Wallis test), albeit lower than that seen in experiments where plantain NSP was added to cells for 30 min without removal before infection (92.9 ± 2.0%, compared to the untreated control; p < 0.001; Kruskal-Wallis test). In contrast, pre-incubation of plantain NSP directly with bacteria for 30 min, followed by removal of soluble dietary fibre from bacteria by centrifugation, resulted in no significant inhibition of C. difficile adhesion compared with untreated controls (see Supplementary Figure S2).

In addition, treatment of Caco-2 cells with 2.5–20 mg/ml plantain NSP also resulted in a dose-dependent increase in cellular chloride ion channel activity, with peak activity at 20 mg/ml plantain NSP (5.4 ± 0.4-fold increase in comparison to the untreated control; p < 0.001 Kruskal-Wallis test, N = 3, n ≥ 3). This level of activation was similar to that seen with chloride ion channel agonists forskolin (200 μM) and RP107 (100 μM), which increased chloride ion channel activity by 5.4 ± 0.5 and 8.2 ± 1.0-fold respectively, in comparison to the untreated vehicle controls (p < 0.001). Other dietary plant NSP’s (at 5 mg/ml), previously shown to have little or no inhibitory action on C. difficile adhesion to Caco-2 cells, exhibited less efficacy with oat and runner bean NSP increasing chloride ion channel activity at a much lower levels (2.2 ± 0.6 and 2.0 ± 0.3 fold, respectively; both p < 0.001), with tomato, pear, bell pepper and apple NSP actually reducing chloride ion channel activity in comparison to the untreated controls (see Supplementary Figure S3). Chloride ion channel activity in Caco-2 cells induced by 10 mg/ml plantain NSP, was significantly inhibited by inhibitors NPPB and Inh-172 in dose-dependent manner; with peak levels of inhibition seen using 800 µM NPPB and 200 µM Inh-172, reducing chloride efflux by 94.7 ± 10.0% and 86.9 ± 8.5% respectively, in comparison to the vehicle-treated controls (both p < 0.001). However, bacterial interaction assays indicated that in the presence of NPPB or Inh-172, pre-treatment of Caco-2 cells with 10 mg/ml plantain NSP still possessed significant ability to reduce C. difficile 80042 adhesion. C. difficile adhesion to Caco-2 cells was significantly decreased when treated with chloride ion channel agonist.

RP107, with an observed reduction in adhesion of 23.7 ± 9.0% (at 100 μM), from vehicle treated controls (set at 100%) (p < 0.01; Kruskal-Wallis test), with this action reversed in the presence of chloride ion channel antagonists NPPB or Inh-172, increasing bacterial adhesion by 7.9 ± 8.1% and 4.5 ± 4.1% respectively above the level seen in controls (100%); see Supplementary Figure S3).

At 5 mg/ml, the acidic (pectin-rich) polysaccharide fraction of plantain NSP, significantly blocked adhesion of C. difficile isolate 080042 to intestinal Caco-2 cell-line by 52.3 ± 7.2% (p < 0.01), and was at least as effective as the unfractionated plantain soluble fibre, which reduced adhesion by 62.3 ± 17.1% (p < 0.01). Conversely, pre-treatment of Caco-2 cells with 5 mg/ml neutral polysaccharide fraction (unbound to Q-Sepharose) of plantain NSP showed significantly reduced ability to inhibit C. difficile epithelial adhesion (Figure 4).

Soluble NSP from plantain, at 10 mg/ml, significantly reduced the adhesion of all five C. difficile isolate spore preparations to Caco-2 cells, with inhibition ranging from 27.9 ± 5.4% to 33.8 ± 6.9% (all p ≤ 0.05, Kruskal-Wallis test); see Table 2. Blockade of spores was less than that observed with the corresponding vegetative C. difficile bacterium, from which the spores were purified, at the same MOI 100, where pre-incubation with 10 mg/ml plantain NSP resulted in inhibition ranging from 78.9 ± 3.1% to 97.9 ± 1.5% (for all, p < 0.001). It is worth noting that C. difficile spores (from all isolates) showed markedly greater adhesion to Caco-2 cells (85 ± 7.3% versus 4.0 ± 0.6% adhesion of the original inoculum of spores and vegetative cells (1 × 10⁷ CFU/ml), respectively; N = 5 isolates).

Treatment of SW480 colonocytes with C. difficile binary toxin (CDT) dose-dependently increased cytotoxicity, as measured by enhanced caspase 3/7 activity, with peak response seen at 10 ng/ml CDT (283,890 ± 125,878 RFU) compared to untreated controls, 109,333 ± 35,941 (p < 0.05 Kruskal-Wallis test; N = 2, n = 3). Pre-treatment with plantain NSP inhibited the dose-dependent cytotoxicity induced by CDT, with peak reduction in toxin-treated SW480 colonocytes occurring at 100 ng/ml CDT (259,914 ± 67,103 RFU) which was reduced by >70% to 73,060 ± 53,767 (N = 2, n = 3; p < 0.001); see Figure 8A. Similarly, treatment with CDT also effected a dose-dependent increase in release of pro-inflammatory IL-8 from SW480 cells, with peak response seen at 100 ng/ml CDT (1041.92 ± 84.11 pg/ml/24 h) compared to untreated controls, 27.33 ± 10.81 pg/ml/24 h (p < 0.001). Again, it was seen that pre-treatment with plantain NSP inhibited the dose-dependent actions of CDT, with peak IL-8 release in toxin-treated SW480 colonocytes occurring at 100 ng/ml CDT reduced by approximately 50%, to 540.67 ± 41.21 pg/ml (p < 0.001); see Figure 8B.

CDT was also observed to induce release of inflammatory IL-8 from Intestine-407 cells, at test concentrations of 10 and 100 ng/ml over 24 h (both p < 0.001 one-way ANOVA; N = 2, n = 3). Pre-incubation of Intestine-407 cells with 10 mg/ml plantain NSP, reduced the amount of IL-8 released to the medium by both 10 and 100 ng/ml CDT (both p < 0.01; N = 2, n = 3 Kruskal-Wallis test). Plantain NSP was also effective in reducing Intestine-407 cytotoxicity induced by CDT (p < 0.05). No such inhibitory activity was seen with pre-treatment using 10 mg/ml oat NSP; see Supplementary Figure S6).