On Lactococcus lactis UL719 competitivity and nisin (Nisaplin®) capacity to inhibit Clostridium difficile in a model of human colon

Clostridium difficile is the most frequently identified enteric pathogen in patients with nosocomially acquired, antibiotic-associated diarrhea and pseudomembranous colitis. Although metronidazole and vancomycin were effective, an increasing number of treatment failures and recurrence of C. difficile infection are being reported. Use of probiotics, particularly metabolically active lactic acid bacteria, was recently proposed as an alternative for the medical community. The aim of this study was to assess a probiotic candidate, nisin Z-producer Lactococcus lactis UL719, competitivity and nisin (Nisaplin®) capacity to inhibit C. difficile in a model of human colon. Bacterial populations was enumerated by qPCR coupled to PMA treatment. L. lactis UL719 was able to survive and proliferate under simulated human colon, did not alter microbiota composition, but failed to inhibit C. difficile. While a single dose of 19 μmol/L (5× the MIC) was not sufficient to inhibit C. difficile, nisin at 76 μmol/L (20×the MIC) was effective at killing the pathogen. Nisin (at 76 μmol/L) caused some temporary changes in the microbiota with Gram-positive bacteria being the mostly affected. These results highlight the capacity of L. lactis UL719 to survive under simulated human colon and the efficacy of nisin as an alternative in the treatment of C. difficile infections.


Introduction
Clostridium difficile is a Gram-positive anaerobic sporulating pathogen causing intestinal infections following disturbance of the human and animal gut microbiota, usually subsequent to an antibiotic therapy. C. difficile is now thought to be responsible for a wide range of diseases including acute diarrhea and pseudomembranous colitis, and could lead to colonic perforation and death if untreated (Borriello et al., 1990). Although metronidazole and vancomycin are well-established treatments for C. difficile infections (CDI) (Surowiec et al., 2006;Kelly and LaMont, 2008), an increasing number of treatment failures with these antibiotics and recurrence of C. difficile infection are being reported, reviewed in Vardakas et al. (2012). Vancomycin is also losing its attractiveness for CDI treatment with emergence of vancomycin-resistant enterococci and dissemination of antibiotic-resistance determinants within the hospital environment (Lagrotteria et al., 2006). The emergence of C. difficile isolates with multiple-drug resistance is rarely explicitly mentioned (Peláez et al., 2002;Mutlu et al., 2007), but constitutes further a serious public health threat that urges the need of novel antimicrobial treatments.
Previously, a large number of clinical trials highlighted the positive role of probiotics in the treatment of diarrhea by either shortening its duration and/or preventing its complications in infants and young children, reviewed in Guandalini (2011). In instance, a yogurt containing a combination of Lactobacillus rhamnosus GG, L. acidophilus La-5, and Bifidobacterium lactis Bb12 was shown to be an effective method for reducing the incidence of antibiotic-associated diarrhea in children (Fox et al., 2015). Moreover, different probiotics (Saccharomyces boulardii, L. casei DN114001, a mixture of L. acidophilus and B. bifidum, and a mixture of L. acidophilus, L. casei and L. rhamnosus) significantly improved CDI prevention, reviewed in McFarland (2015). Although several meta-analyses pointed the positive effect of probiotics, their role in the prevention of CDI remains unclear. The health-promoting properties of probiotics are numerous and their effects on host include competition with pathogens for adhesion sites and nutrients, stimulation of immunity/immunomodulation, and production of inhibitory substances such as bacteriocins (Fliss et al., 2011). Bacteriocins have been suggested as promising alternative to conventional antibiotics (Rea et al., 2007(Rea et al., , 2010, and their production is being considered as a probiotic trait although not clearly demonstrated in vivo (Dobson et al., 2012). While several bacteriocins including nisin (Le Blay et al., 2007;Le Lay et al., under revision), Microbisporicin (Castiglione et al., 2008), Lacticin 3147 (Rea et al., 2007) and thuricin CD (Rea et al., 2010) were shown effective against C. difficile, to date only nisin is approved by the American Food and Drug Administration, the World Health Organization, and the European Union as natural food additive (Delves-Broughton, 1990). Nisin displays high antibacterial activity against multi-resistant Streptococcus pneumoniae, methicillinresistant Staphylococcus aureus (MRSA), vancomycin-resistant Enterococcus faecium, E. faecalis, and C. difficile (Severina et al., 1998;Le Blay et al., 2007).
Previously, we have observed that potential probiotic Lactococcus lactis UL719, a nisin Z producer, was able to survive through the gastrointestinal tract (unpublished data). The strain L. lactis UL719 was able to grow and inhibit Listeria in a medium simulating the nutrient composition of the human colon (Fernandez et al., 2013). The aim of this study was to evaluate the capacity of L. lactis UL719 and nisin (Nisaplin R ) to inhibit C. difficile in a model of the colon mimicking physiological and microbiological conditions of the large intestine. In addition, impact of both strain and its bacteriocin on the gut microbiota composition were also investigated.

Feces Collection and Immobilization in Gel Beads
A fresh fecal sample was obtained from one 27 years old healthy donor who had not taken antibiotics for the previous 3 months. The collected fecal sample was used for immobilization following procedure described by Le Blay et al. (2012). The entire process was completed aseptically under anaerobic conditions within 1 h after sample collection.

Nutritive Medium
The culture medium used for colonic fermentation was the same as described by Macfarlane et al. (1998) with some modifications. Briefly, 0.5 mL of a vitamin solution (mg/L: pyridoxine-HCl 20; p-aminobenzoic acid 10; nicotinic acid 10; biotin 4; folic acid 4; vitamin B12 1; thiamine 8; riboflavin 10; menadione 2; vitamin K1 0.005; pantothenate 20) described by Gibson and Wang (1994) was added to each liter of the culture medium. The vitamin solution was sterilized by filtration (0.2 μm, VWR) and added to the autoclaved medium (15 min, 121 • C) after cooling at room temperature.

Experimental Setup and Sampling
The colonic fermentation was based on the model described by Cinquin et al. (2004). A single-stage reactor (Bioflo III, New Brunswick Scientific Inc., Edison, NJ, USA) with 1 L working volume containing 30% (v/v) of freshly prepared beads was used to mimic the microbial ecosystem of adult distal colon. The colonization of beads with fecal microbiota was carried out during 2 days, and the nutritive medium was aseptically replaced by fresh culture medium every 12 h. pH (6.2) and anaerobic and temperature (37 • C) conditions were maintained during the whole fermentation by addition of 5 M NaOH and a continuous flow of pure CO 2 in the headspace. The continuous fermentation was carried out in the same reactor connected to a stirred feedstock vessel containing the sterile culture medium at 4 • C under a CO 2 atmosphere and to an effluent-receiving vessel. Feed flow rate was adjusted to 83.3 mL/h to mimic a mean retention time of 12 h encountered in adult distal colon.
The fermentation process was carried out for a total of 82 days and microbiota was stabilized 2 weeks before challenging tests. First, a cell suspension of L. lactis UL719 (at final concentration 10 9 CFU/mL in the reactor) was added twice to the reactor 3 | Impact of L. lactis UL719 (10 9 CFU/mL) and/or C. difficile ATCC43255 (5 × 10 6 CFU/mL) addition on the microbiota. (day 17 and 22) (Figure 1).

Analyses of Metabolites
Short-chain fatty acids (SCFA: acetate, propionate, butyrate, and valerate) and isoacids (isobutyrate and isovalerate) were determined by high-performance liquid chromatography (HPLC) analysis (Waters, Milford, MA, USA) equipped with an Ion 300 column (Transgenomic, San Jose, CA, USA), a differential refractometer (Model R410, Waters) as previously described by Cinquin et al. (2004). The analysis was performed at a flow rate of 0.4 mL/min at 37 • C, with an injection volume of 100 μL. Each analysis was done in duplicate. The mean metabolite concentrations were expressed in mmol/L.

Statistical Analysis
Data are presented as means ± SD. Cell counts values were log 10 -transformed and analyzed for repeated measures using the PROC MIXED procedure of SAS v9.2 statistical package (SAS Institute Inc., Cary, NC, USA). The statistical differences in metabolites contents between treatments were evaluated using a one-way ANOVA t-test. The level of significance was P ≤ 0.05.

Microbiota Composition during Stabilization Period
Bacterial populations enumerated by qPCR coupled to PMA treatment in the fecal inoculum and effluent samples at the end of stabilization period are summarized in Table 2.
The fecal inoculum presented a total bacterial cell counts of 11.84 ± 0.04 log 10 CFU/g, which was dominated by Bacteroidetes (10.85 ± 0.02 log 10 CFU/g), clostridia (10.55 ± 0.02 log 10 CFU/g), and bifidobacteria (10.16 ± 0.15 log 10 CFU/g). At the end of the stabilization period (16 days) under simulated colon conditions, the microbiota population reached a pseudo steady state in which a slight change was observed in the microbial balance, compared to the initial fecal inoculum. The microbiota decreased by −0.7 log 10 CFU/mL at this stage and was dominated by Bacteroidetes group with 10.52 ± 0.08 log 10 CFU/mL. While Enterobacteriaceae group increased by 1.49 log 10 and reached 8.73 ± 0.01 log 10 CFU/mL, bifidobacteria, and Lactobacillaceae/Leuconostocaceae group populations dropped to 6.14 ± 0.08 and 3.82 ± 0.14 log 10 CFU/mL, respectively. Nevertheless, the obtained results are in accordance to those previously reported for colonic fermentation models (Brück et al., 2002;Probert and Gibson, 2004;Cleusix et al., 2008;Le Blay et al., 2012).
Frontiers in Microbiology | www.frontiersin.org After the stabilization period, L. lactis UL719, C. difficile ATCC43255, and their combination were successively added to the bioreactor and the microbiota populations were monitored by qPCR (Table 3). Interestingly, the addition of L. lactis UL719 at 1 × 10 9 CFU/mL to the bioreactor, did not induce any significant change neither in the intestinal microbiota composition nor in metabolites production ( Table 4). Since the last addition of L. lactis UL719 to the reactor, the strain was detected at about 0.1 − 1 × 10 9 CFU/mL during the remaining 20 days of fermentation (Figure 2). While the infection of the bioreactor with 5 × 10 6 CFU/mL of C. difficile did not affect the microbiota composition, a slight but significant decrease (p < 0.05) of acetate and butyrate was detected (from 76.24 to 72.59 mmol/L and from 32.13 to 29.54 mmol/L, respectively) ( Table 4).
Simultaneous addition of C. difficile and L. lactis UL719 had no impact on the microbiota cell counts but a significant decrease (p < 0.05) of butyrate (from 32.13 to 28.40 mmol/L). Under these conditions, L. lactis has no inhibitory effect on C. difficile (Figure 3).

A Nisin Concentration of 20× the MIC is Required to Effective Inhibition of C. difficile ATCC43255 in a Model of Human Colon
The microbiota was challenged by 5× and 20× the MIC vs. C. difficile ATCC43255. Nisin at 5× the MIC did not alter the microbiota which remained stable (data not shown) although minor variations in the metabolite production profile ( Table 4). At a nisin concentration of 20× the MIC, total microbiota significantly decreased by 0.7 log 10 (p < 0.008), as shown in Figure 4. Gram-positive bacteria were affected by this higher amount of nisin, with Ruminococcaceae group being the mostly altered (−3.7 log 10 ) after 24 h. In a lesser extent, a reduction of 1.5 log 10 , 1.3 log 10 , and 1 log 10 were recorded for Lachnospiraceae group, Lactobacillaceae/Leuconostocaceae FIGURE 2 | Survival of L. lactis UL719 after its last addition (day 62) in a human colon model. L. lactis UL719 (circle); theoretical washout (square).
Frontiers in Microbiology | www.frontiersin.org group and bifidobacteria, respectively. After 24 h of nisin administration, all bacterial populations recovered their initial counts except Ruminococcaceae group which dropped to its minimum counts. While acetate and butyrate significantly decreased (p < 0.05) from 76.24 and 32.13 mmol/L to 69.12 and 26.29 mmol/L, propionate production increased by 13% (Table 4). Besides, a nisin concentration of 5× did not inhibit C. difficile, which counts remained close to control (C. difficile alone) (Figure 3). Conversely, nisin at 20× was effective at inhibiting C. difficile with a significant reduction (p < 0.001) of 2.3 log 10 at 1 h that lasted for 8 h (Figure 3). C. difficile was not detected after 24 h in this model (data not shown).

Discussion
Previously, we have demonstrated the nisin efficacy against several clinical isolates of C. difficile vegetative cells and spores (Le Lay et al., under revision). In addition, we have observed that L. lactis UL719, a nisin Z producer, was able to survive these GIT stressful conditions, to keep ability to produce its bacteriocin, and to reach the colon in large enough numbers (>10 8 CFU) to comply with the recommended daily dose of 10 8 -10 9 cells delivery to exert a beneficial effect on the host (unpublished data). The aim of this study was to assess L. lactis UL719 competitivity and nisin capacity to inhibit C. difficile ATCC43255 in a model of human colon. In this study, L. lactis UL719 at 10 9 CFU/mL did not induce any significant change neither in the intestinal microbiota composition nor in metabolites production. The strain was monitored by quantification of nisI gene by PMA-qPCR, and found able to survive and proliferate up to 10 8 CFU/mL in our colonic model during the 82 days of fermentation (Figure 2). Unlikely, L. lactis DPC6520 was shown more susceptible to GIT conditions, which cell counts were reduced by 10 000-fold 24 h after its inoculation into a colon model . Likewise, a 19 μmol/L concentration of nisin (corresponding to 5× the MIC vs. C. difficile ATCC43255) did not alter microbiota levels. At a higher concentration of 76 μmol/L (20×), Gram-positive bacteria were affected and Ruminococcaceae group was the mostly altered (−3.7 log 10 ), while increase in Gram-negative population (Bacteroidetes and Enterobacteriaceae) were observed. Nevertheless, the initial bacterial balance was quickly restored within 24 h after the addition of 20× nisin. Previously, we have shown in vitro the sensitivity of colonic Gram-positive bacteria such as B. bifidum DSM 20456, L. fermentum ETHZ, C. clostridioforme DSM933, Eubacterium biforme DSM3989 to nisin (Le Blay et al., 2007). Recently, Rea et al. (2011) reported that lacticin 3147 induce similar variations in microbiota composition, with a decrease in Firmicutes abundance in favor of Proteobacteria. Broad-spectrum antibiotics like vancomycin and metronidazole seems to induce also decrease of Firmicutes and an increase in Enterobacteriaceae and Proteobacteria (Antonopoulos et al., 2009;Rea et al., 2011). More recently, thuricin CD, a narrow spectrum bacteriocin produced by Bacillus thuringiensis, was used in the distal colon model and had no significant impact on the composition of the microbiota . Although its capacity to survive colonic conditions, L. lactis UL719 had no significant effect on C. difficile. Similar results were previously reported with L. lactis DPC6520 (a lacticin 3147 producer) and L. lactis DPC6519 (lacticin non-producer) in an ex vivo human colonic model . Although L. lactis UL719 is able to produce nisin in a Macfarlane medium simulating the nutrient composition of the colon (Fernandez et al., 2013), the lack of effectiveness observed here is likely due to no or a low production of nisin, not sufficient to inhibit C. difficile.
Conversely, L. salivarius UCC118 has demonstrated its capacity to produce the Abp118 bacteriocin in vivo and to protect mice against infection with the invasive foodborne pathogen Listeria monocytogenes. This protection was related to bacteriocin production, and mutant of L. salivarius UCC118 lacking the bacteriocin gene failed to protect mice against infection (Corr et al., 2007). Some similar results were obtained with human L. lactis and Pediococcus acidilactici nisin-and pediocinproducing strains that were able to reduce vancomycin-resistant enterococci intestinal colonization in a mouse model (Millette et al., 2008).
Although L. lactis UL719 had no significant effect on C. difficile in this model of human colon, addition of nisin (in Nisaplin R form) at 76 μmol/L induced a significant reduction of C. difficile. The observed efficacy of Nisaplin R against C. difficile could be due to a synergy between nisin and salt present in the commercial product. At lower concentration of nisin (19 μmol/L), we did not show any significant effect on C. difficile, its rapid adsorption on the surface of the colonic microbiota or its inactivation due to enzymatic activities (proteolysis mainly) could explain this lack of activity . Rea et al. (2011) have reported on the effectiveness of other bacteriocins such as lacticin 3147 and thuricin CD against C. difficile in a distal colon model. Lacticin 3147 (270 μmol/L) and thuricin CD (90 μmol/L) affected the viability of C. difficile (10 6 CFU/mL) with a loss of detection after 12 h and three log 10 reduction after 24 h, respectively . After respective addition of lacticin 3147 (270 μmol/L) and thuricin CD (90 μmol/L), authors have shown a CFU reduction of 4 log 10 and 1.2 log 10 , but lacticin at 90 μmol/L had no significant effect on the C. difficile viability . In this study, nisin was as effective as lacticin 3147 and more efficient than thuricin CD with a CFU reduction of 3.23 log 10 with nisin (76 μmol/L) compared to initial time. Besides, three times addition of vancomycin (90 μmol/L) or metronidazole (90 μmol/L) is required to induce a significant effect on C. difficile after 24 h . A single dose of nisin (76 μmol/L) was as effective as antibiotics traditionally used to treat CDIs.
With increase of failures and recurrences in the treatment of CDIs, development of alternative treatments has become necessary. In recent years, use of probiotic bacteria producing antimicrobial molecules (such as bacteriocins) constitute a promising alternative for prevention and treatment of C. difficile related diseases. In the study, we have shown that nisin-producer L. lactis UL719 was able to survive and proliferate in the human colon model. Although L. lactis UL719 failed to inhibit C. difficile in this model, L. lactis UL 719 had not affected the microbiota. Others studies aiming to increase competitivity and nisin production will be necessary and could include the addition of prebiotics or carbohydrate which stimulate nisin production. Nisin (Nisaplin R ) causes some temporary changes in the microbiota but is effective at killing C. difficile in the human colon model.