Bacteriocin-Like Inhibitory Substances from Probiotics as Therapeutic Agents for Candida Vulvovaginitis

Probiotics can potentially prevent and treat diseases. We examined the inhibitory activity of bacteriocin-like inhibitory substances (BLISs) from potentially probiotic lactobacilli and streptococci on Candida albicans and non-Candida albicans clinical isolates from women with vulvovaginitis. Using agar well diffusion assays, BLISs inhibited both Candida albicans and non-Candida albicans isolates. The BLIS from L. pentosus isolates had the highest anti-Candida activity (33/45; 73.3%), followed by BLISs from isolates of L. paracasei subsp. paracasei (31/45; 68.9%), L. rhamnosus I (30/45; 66.7%), L. delbrueckii subsp. lactis I (30/45; 66.7%), and S. uberis II (30/45; 66.7%). Upon characterization according to the retained activity under variable physical and chemical conditions, the BLISs showed stability against heat, pH, and surfactants, but were protease-sensitive, which suggests a proteinaceous nature of the active substances. Using crystal violet assays, the BLISs reduced the Candida biofilm biomass significantly as compared to a control group that lacked BLISs. In vivo testing of the antagonistic activity was performed using the Galleria mellonella (G. mellonella) larvae model. BLISs significantly improved survival in G. mellonella larvae treated with Candida isolates on the first, second, and seventh days, as compared to larvae inoculated with Candida only (p < 0.01). The results show that BLISs can be used as biotherapeutic agents in vulvovaginal candidiasis.


Introduction
Vulvovaginal candidiasis (VVC) is a common condition that is typically mitigated by over-the-counter topical or systemic antifungal drugs. About 75% of women experience one or more VVC attacks, whereas 10% experience recurrent VVC attacks [1]. Pathogenic Candida species have developed resistance to many antifungal agents [2]. This highlights the need to develop alternative antimycotic agents to control pathogenic Candida infections [3]. In addition to resistance issues, many currently available antifungal drugs have narrow spectra and cause side effects [4].
Lactobacilli are dominant in the vagina of the vast majority of healthy premenopausal women [5]. As part of the microbiota, lactobacilli can prevent vulvovaginal infections [6]. The antifungal activities are not always explained by hydrogen peroxide, a typical antimicrobial factor in lactobacilli [7]. The potent antifungal activity of probiotic strains of Lactobacillus can be partially explained by the low pH and the production of organic acids [8]. In addition, lactic acid bacteria such as Enterococci and Streptococci can produce antimicrobial

Isolation and Identification of Potential Probiotics
Fifty potential probiotic isolates were obtained from samples of raw milk, fermented milk, yogurt, cheese, meat products, and mixed pickles. Lactobacillus isolation on de Man, Rogosa, Sharpe (MRS) agar (Oxoid, Basingstoke, UK), and Streptococcus isolation on M17 agar (Oxoid, Basingstoke, UK) were performed [21]. To identify the isolated bacteria, standard microbiological methods (phenotypic, morphological, and biochemical techniques) were used. All isolates were grown at 15 • C and 45 • C. Gram staining, catalase tests, and glucose fermentation tests were performed [22]. Carbohydrate fermentation patterns for the isolated lactobacilli and streptococci were determined using the API Rapid CH fermentation strips (Biomèrieux, Marcy l'Etoile, France) in CHL medium and the API 20 Strep kit (Biomèrieux, Marcy l'Etoile, France) respectively.

Isolation and Identification of Candida Isolates
Vaginal swabs from females with vulvovaginitis were inoculated on Sabouraud dextrose agar (SDA) (Oxoid, Basingstoke, UK) and incubated for 24-48 h at 37 • C. Isolated colonies were identified using standard microbiological methods (colony characteristics, Gram staining, urea hydrolysis, and germ tube test) [23]. The creamy yeast-like colonies that were Gram-positive showed pseudohyphae under a light microscope, and negative results with the urea hydrolysis test were further examined for Candida species identification [23]. Carbohydrate assimilation was tested using API Candida (BioMèrieux, Marcyl'Etoile, France). Forty-five different clinical isolates were retrieved, including twenty-five C. albicans isolates (CA 1-25) and 20 non-C. albicans Candida isolates (NCAC 1-20).

Screening for Anti-Candida Activity
Agar well diffusion was used to assess the BLIS inhibitory potential against Candida isolates causing vulvovaginitis [24]. Potential probiotics showing anti-Candida activities against at least one of two indicator organisms: C. albicans (ATCC 90028) and C. glabrata (ATCC 90030), were selected for further processing after confirmation by API (as described in Section 2.1.1). BLISs were extracted from potential probiotics (Section 2.2) and subsequently examined for anti-Candida activity. BLISs with the strongest activities were further characterized (Section 2.3).

Preparation of BLISs
Overnight cultures of the potential probiotic isolates cultivated on broth medium were diluted with fresh medium (inoculum size 1% v/v) at 32 • C for 18 h, and the concentration was adjusted to an optical density of 1.6 at 600 nm (~1 × 10 8 cells/mL). MRS broth (Oxoid, Basingstoke, UK) was used for the cultivation of lactobacilli, and M17 broth (Oxoid, Basingstoke, UK) was used for the cultivation of streptococci. Cultures were centrifuged at 10,000× g for 15 min at 4 • C and the resulting supernatant was designated as a crude cellfree culture supernatant (CCFCS). To neutralize hydrogen peroxide, 1 mg/mL of bovine catalase (Sigma-Aldrich, St. Louis, MO, USA) was added. The pH of each CCFCS was adjusted to 6.5 with 1 mol/L NaOH (Sigma-Aldrich, St. Louis, MO, USA) [21]. The treated supernatant was designated a BLIS. The 10-fold concentration of the BLIS was obtained using a vacuum rotary evaporator at 40 • C, filtered through a sterile 0.2 µm syringe filter, and stored at −20 • C until further use.

Physicochemical Characterization of BLISs
The five BLISs with the most potent anti-Candida activity were selected for further characterization and assessment of the antimicrobial activity. The residual anti-Candida activity of treated BLISs under different conditions was determined by agar diffusion and compared with the positive control (untreated, 100% activity).
(1) Effect of heating: BLISs were incubated in a water bath at 60 • C, 80 • C, and 100 • C for 10, 30, 60 min and at 121 • C for 10, 15, and 20 min, and then cooled on ice.

Protein Purification and Separation from BLIS
The protein content of the cell-free BLIS was isolated by overnight precipitation with ammonium sulphate at a saturation level of 70% with slight agitation at 4 • C [25]. The anti-Candida activity of the ammonium sulfate-precipitated BLIS was determined by the agar well diffusion assay using indicator strains. The BLIS was purified before measuring its protein concentration in the supernatant [26]. The molecular mass of the partially purified BLIS was estimated by SDS-PAGE (sodium dodecyl sulfate-polyacrylamide gel electrophoresis) with 4% stacking and 10% polyacrylamide gel [27].

In Vitro Anti-biofilm Activity of BLIS
The antibiofilm activities of the BLISs from five potential probiotics with the strongest activities against Candida, as detected by agar well diffusion, were tested. BLISs from L. pentosus, L. rhamnosus I, L. paracasei subsp. paracasei II, L. delbrueckii subsp. lactis I, and S. uberis II. were freshly prepared, and viable counts were determined to verify their cellfree status before each experiment. They were then tested against biofilms formed by C. albicans ATCC 90028, C. albicans (CA 1), and C. glabrata (belongs to NCAC 1) clinical isolates. A standard inoculum of 1 × 10 6 cells from the overnight culture of each fungal strain was used to form the biofilm. In each experiment, 40 µL per well of the tested BLISs was added to wells of sterile flat-bottomed polystyrene 96-well microtiter plates, followed by the addition of the tested Candida culture (160 µL) to a final volume of 200 µL/well [20].
For each experiment, two control groups were prepared: One that had wells with phosphate-buffered saline (PBS) only, and one that had wells with standardized BLISs (without C. albicans) from each of the five forementioned probiotics. After daily media change, plates were incubated for 48 h at 37 • C, with shaking at 75 rpm. After biofilm formation, biofilm biomasses with and without BLISs were quantified using crystal violet assays [28]. The absorbance was quantified at 540 nm. The reference strain C. albicans DAY185 was used as a positive control [29].

G. mellonella Survival Assay for BLIS Activity Against Candida Isolates (In Vivo Model)
The pathogenicity of C. albicans ATCC 90028, CA 1, and NCAC 1 clinical isolates in the presence or absence of cell-free BLISs was assessed using G. mellonella survival assays. Sixteen G. mellonella larvae in the final larval stage with similar masses (250-350 mg) and sizes were used [30]. Three control groups of noninfected larvae were included as controls: Group 1 was inoculated with PBS to assess for potential physical trauma; Group 2 was inoculated with MRS broth to evaluate for any toxicity on G. mellonella larvae; Group 3 was not injected (control for overall viability). A 5 µL inoculum of the standard Candida isolate suspension (10 5 cells/mL) was injected into the hemolymph of each larva through the last left proleg, and 5 µL of each BLIS was injected into the last right proleg. For the groups infected with Candida isolates only, 5 µL of the microbial suspension was inoculated into the last left proleg and 5 µL of PBS was injected into the last right proleg. In these experiments, Candida isolates were inoculated 1 h before the inoculation of the BLISs. The larvae were incubated at 37 • C. When a larva displayed no movement in response to stimuli or showed dark discoloration of the cuticle, it was considered dead. Dead larvae were counted daily for 7 days [30].

Polymerase Chain Reaction (PCR) Amplification and Identification of Lactobacilli by Partial Sequencing of the 16S rRNA Gene
The genomic DNA of potential probiotics with the strongest in vitro anti-Candida activities was extracted and purified using DNeasy blood and tissue kits (Qiagen Inc., Hilden, Germany). The 16S rRNA gene was amplified using universal bacterial primers (518F/800R) [31]. PCR fragment analysis was done [32]. Sequences were compared, using BLAST, to those in the GenBank database.

Statistical Analysis
Data analysis was done using SPSS 16 (SPSS Inc., Chicago, IL, USA) and the chi-square test. Fisher's exact test was used for two-by-two tables when the expected cell counts were less than 5. The G. mellonella survival curve was analyzed using the log-rank test.

Isolation of Active Lactobacilli and Streptococci and Screening for BLIS Antagonistic Activity
The BLISs of the thirteen probiotic isolates (with activity against one or more indicator organisms) were extracted and tested for anti-Candida activity against twenty-five C. albicans (CA 1-25) and twenty non-C. albicans (NCAC 1-20) ( Table 1). The inhibitory activity against all Candida isolates ranged from 51.1% (23/45 isolates) for the BLIS from Streptococcus uberis I to 73.3% (33/45 isolates) for the BLIS from L. pentosus. The BLIS from L. pentosus showed a significantly higher antagonistic activity against NCAC isolates than against CA isolates (90% versus 60%, p = 0.024). BLISs from other potential probiotic isolates did not show a significant difference in antagonistic activity against NCAC isolates versus CA isolates ( Table 1). The BLIS from L. pentosus isolates had the highest anti-Candida activity (33/45; 73.3%), followed by the BLIS from isolates of L. paracasei subsp. paracasei  Table 1). The previous five BLISs with the most potent anti-Candida activity were selected for further characterization and assessment against C. albicans ATCC 90028 and C. glabrata ATCC 90030 indicator strains.

BLIS Characterization
The five BLISs with the highest antimicrobial activity were reassessed after exposure to heat, enzymes, pH, surfactants, and organic solvents. Variable degrees of stability of BLISs after treatment with these factors were reported (Table 2). All BLISs retained their activity at 10% concentration of the three organic solvents: Chloroform, ethanol, and n-hexane, for 1, 4, and 6 h against both C. albicans and C. glabrata. With chloroform, all BLISs completely lost their activity at 30% concentrations at 1, 4, 6, and 24 h. With ethanol, only the BLISs from L. paracasei subsp. paracasei II retained~50% of their activity against C. albicans at 30% concentration for 1 h. The BLIS from L. pentosus retained 75% of its activity against C. glabrata at 20% concentration of ethanol for 4 h.
BLISs were completely inactivated by pepsin and trypsin but not by α-amylase (Table 2). BLISs retained different levels of activity after treatment with surfactants. In some cases, an enhancement of activity (>100%) was observed (Table 3). Partial purification of BLISs was accomplished by ammonium sulphate precipitation. Precipitates at lower ammonium sulphate saturation had lower anti-Candida activity. Maximal activity was observed at 70% ammonium sulphate saturation. Using SDS-PAGE followed by Coomassie blue staining, the purified BLISs showed protein bands with molecular weights ranging from 2.5 to 10.5 kDa.

Effect of BLIS on Candida Biofilm Formation In Vitro
C. albicans biofilm formation is a key mechanism for its growth and survival in the host. Using crystal violet assays, BLISs from L. pentosus, L. paracasei subsp. paracasei II, L. rhamnosus I, L. delbrueckii subsp. lactis I, and S. uberis II significantly reduced biofilm formation (expressed as relative absorbance) by C. albicans ATCC 90028, CA 1, and NCAC 1 clinical isolates. A control group that lacked BLISs failed to yield similar results. The BLIS from L. pentosus was most effective, whereas the BLIS from S. uberis II was least effective (Table 4).

BLIS Prolongs the Survival of Candida-Infected G. mellonella Larvae
The pathogenicity of C. albicans (ATCC 90028) and the clinical isolates CA 1 and NCAC 1, in the presence and absence of cell-free BLISs from L. pentosus, L. paracasei subsp. paracasei II, L. rhamnosus I, L. delbrueckii subsp. lactis I, and S. uberis II, was assessed using the G. mellonella killing assay.
Inoculation of the larvae with Candida isolates (in the absence of BLISs) killed 87.5% of the larvae within 24 h when CA 1 was injected, and 81.2% upon injection of NCAC 1 or C. albicans ATCC 90028 (Table 5). Conversely, larvae treated with Candida isolates plus BLISs of L. pentosus, L. delbrueckii subsp. lactis I, and L. paracasei subsp. paracasei II (50 µg/larva) exhibited significantly improved survival on the first, second, and seventh days in comparison to larvae inoculated with Candida isolates alone (p < 0.01) ( Table 5). The effect of the BLIS from L. rhamnosus I on the survival of G. mellonella larvae infected with Candida isolates on the first, second, and seventh days was less significant. Finally, the effect of the BLIS from S. uberis II was not statistically significant as compared to Candida isolates alone (Table 5 and Figure 1). Control larvae (without Candida isolates or BLISs) exhibited 0% mortality on the first, second, and seventh days (Figure 1). These results showed that BLISs of potential probiotics protected G. mellonella larvae from Candida-induced mortality.

Identification of Lactobacilli by Partial Sequencing of the 16S rRNA Gene Sequences
PCR fragment analysis was done for the 16S rRNA gene sequences of potential probiotics with the best in vitro anti-Candida activity. After comparing sequences with those in the GenBank database, the 16S rRNA sequences were deposited in the GenBank database with the accession numbers LC406091, LC406092, LC406093, and LC406094 for L. pentosus, L. rhamnosus, L. paracasei, and L. delbrueckii subsp. lactis, respectively.

Discussion
In this study, we assessed the anti-Candida activity of BLISs from food-derived probiotic isolates. Notably, 26% of the isolates showed BLIS-mediated anti-Candida activity. In previous studies, the bacteriocinogenic strains ranged from 0.27% to 20% [33][34][35]. Variations in food samples and media used for the isolation of bacteriocinogenic lactobacilli and streptococci may have contributed to this difference. Results, in the present study, indicated that the BLIS from L. pentosus exhibited the highest anti-Candida activity. Pentocin TV35b, a BLIS isolated from L. pentosus, has also been reported to inhibit the growth of C. albicans [36].
Orally administered lactobacilli reduced vaginal colonization and infection by Candida [37]. Similarly, intravaginal capsules containing lactobacilli had anti-microbial effects against C. albicans and other pathogens [38,39]. The inhibitory activities of vaginal lactobacilli may be synergistically enhanced by the production of antimicrobials by the administered probiotics. The previous findings are consistent with our finding that lactobacilli can prevent VVC. Conversely, other studies have suggested that lactobacilli do not protect against VVC [7,40,41]. The ability of lactobacilli in the vaginal flora to produce potent antimicrobials can be impacted by a variety of dietary and environmental factors [42]. The previously reported potent antifungal activity of Lactobacillus reuteri RC-14 and L. rhamnosus GR-1 strains against C. glabrata causing VVC is consistent with the findings in this study [8].
The heat stability of BLISs reported in this study is consistent with reports of the heat stability of bacteriocin-like antimicrobial substances produced by bacteria in other studies [7,34,43,44]. The anti-Candida activity of BLISs was observed under acidic, neutral, and alkaline conditions (highest activity at pH 7), indicating that this activity was not due to acid production. Another report indicated that bacteriocin was stable in a pH range of 2 to 8 [45]. Other bacteriocins were reported to have high antibacterial activity at acidic pH [46]. Treatment of BLISs with organic solvents led to a reduced activity at high solvent concentrations. This may be due to the lipid moiety in BLISs [47]. Alcohol and chloroform inactivated bacteriocin from L. plantarum, but acetone, hexane, and alcohol (90%) did not inactivate bacteriocin from Pediococcus pentosaceus [46]. Organic solvents did not affect the inhibitory activities of some bacteriocins [47].
In line with previous studies, we observed BLIS inactivation by proteolytic (but not nonproteolytic) enzymes, which suggests a proteinaceous nature [33,43,48,49]. BLISs retained their activity after treatment with surfactants. In some cases, the antimicrobial activity was even enhanced (Table 3). This can be due to surfactants causing dispersion of bacteriocin complexes into active subunits with more lethal power [50]. The treatment of bacteriocins with surfactants may also eliminate some natural defenses of the indicator organisms. Finally, the observed anti-Candida activity of BLISs was not due to hydrogen peroxide, as activity was maintained after treatment with catalase.
Bacteriocins from streptococci have been recovered from S. salivarius, S. mutans, S. pyogenes, S. bovis, and S. rattus [10]. A nonpathogenic oral commensal bacterium, S. salivarius K12, exhibited antagonistic activity against oral C. albicans growth in vitro [51]. To our knowledge, this study is the first to report antagonistic activity of S. uberis against Candida isolates. Antibacterial bacteriocins from S. uberis (nisin U and uberolysin) have been biochemically characterized [52]. Although ubericin A was the first class IIa bacteriocin isolated from streptococci to be characterized [53], its activity against Candida has not been studied. The stronger bioactivity of ammonium-sulphate-purified products, compared to crude extracts, in all tested BLISs may be due to the increased concentration of the proteinaceous compounds [47].
Biofilm formation led to poor clinical outcomes with candidiasis [54]. In this study, BLISs from probiotic lactobacilli and streptococci induced a significant reduction in biofilm biomass. Previous studies showed that the biofilms of C. albicans were reduced by cells and supernatants of lactobacilli [19,20,[55][56][57].
L. acidophilus ATCC 4356 was shown to protect G. mellonella against in vivo experimental candidiasis [55]. Similarly, probiotic bacteria (L. rhamnosus and L. acidophilus) significantly reduced C. albicans oral growth in immunocompromised mice [58]. In this study, we showed that injecting BLISs from L. pentosus, L. delbrueckii subsp. lactis I, and L. paracasei subsp. paracasei II into G. mellonella larvae infected with Candida strains significantly increased the survival of larvae. Another study showed that larvae receiving L. rhamnosus supernatants were more protected against C. albicans than larvae receiving L. rhamnosus cells [20]. This might be attributed to the immediate availability of bacteriocins or organic acids in the supernatant versus cells that still need to grow and produce antimicrobial bacteriocins [20].
The BLIS of S. uberis II significantly reduced the biofilm biomass of Candida in vitro. The in vivo protective activity significantly improved the larvae's seventh-day survival (not the first-or the second-day survival) only when larvae were infected with C. glabrata clinical isolates, but not when infected with C. albicans. BLISs from L. pentosus, L. paracasei subsp. paracasei II, and L. delbrueckii subsp. lactis I had a significant protective action on larvae's survival on all three testing days (1, 2, and 7) when larvae were infected with either C. albicans or C. glabrata clinical isolates. L. rhamnosus and L. casei have been previously reported to exhibit stronger antifungal activity than S. thermophilus and S. salivarius [59]. Parolin et al. showed that C. albicans isolates were more highly suppressed by lactobacilli than other Candida species [60].
Our in vitro and in vivo experiments showed that BLISs derived from potential probiotics have anti-Candida activity and can prevent biofilm formation. Furthermore, BLISs are tolerant to heat, nonproteolytic enzymes, pH, surfactants, and organic solvents. Thus, BLISs should be considered as an alternative or adjunct antimicrobial therapy to currently used antifungal agents for the treatment of VVC. Institutional Review Board Statement: The study was conducted according to the guidelines of the Declaration of Helsinki, and was approved by Fayoum University # D24(4).