The in vitro activity of vaginal Lactobacillus with probiotic properties against Candida.

Lactobacilli, the predominant vaginal microorganisms in healthy premenopausal women, control other members of the vaginal microflora and thus protect against bacterial vaginosis and urinary tract infections. It has been claimed that some lactobacilli are also protective against Candida vaginitis. Little is known, however, about the mechanisms by which these lactobacilli can control vaginal populations of Candida and prevent vaginitis. To address this question, vaginal Lactobacillus strains with known antagonistic properties against bacteria were tested for their cell surface properties, adhesion to vaginal cell lines in vitro and antagonistic activities against Candida. A small proportion of the lactobacilli tested adhered strongly to cultured vaginal epithelial cells and inhibited growth of Candida albicans but not of C. pseudotropicalis. This anticandidal activity was in some Lactobacillus strains related to hydrogen peroxide (H2O2) production, but catalase treatment did not suppress this activity in other Lactobacillus strains, suggesting alternative mechanism(s). Moreover, tested vaginal Candida strains were resistant to relatively high concentrations of H2O2 that markedly exceeded those produced by even the most active Lactobacillus strains.


Introduction
Lactobacilli form a critically important component of the vaginal ecosystem, and are the predominant (10 7 to 10 8 CFU/g of vaginal fluid) vaginal microorganisms in healthy premenopausal women. Certain factors that increase the risk of bacterial vaginosis (BV) and urinary tract infections (UTIs) are associated with decreased vaginal Lactobacillus populaions [1,2]. It has also been claimed that oral intake of Lactobacillus acidophilus is protective against Candida vaginitis (CV) [3]. The protective role that lactobacilli play in UTI and BV seems to be based upon two mechanisms, namely specific adherence by selected Lactobacillus species [4] to vaginal epithelium, leading to intensive colonization of this surface by formation of microcolonies and biofilm, and control of the remaining vaginal microflora by production of active metabolites, including acidic products, bacteriocinlike substances and hydrogen peroxide (H 2 O 2 ) [1,5,6]. Little is known, however, about the mechanisms by which some Lactobacilli can control vaginal populations of Candida and thus prevent CV.
The aim of this study was to test vaginal Lactobacillus strains with known antibacterial effects for their cell surface properties, adhesion toward vaginal cell lines in vitro and antagonistic activities directed against yeasts, with special regard to H 2 O 2 production.

Bacteria
A total of 111 Lactobacillus strains were randomly selected from a larger collection of strains with known antagonistic properties exerted against bacteria associated with BV, namely Gardnerella vaginalis, Peptostreptococcus anaerobius and Prevotella bivia, as described previously [7]. The strains were isolated from vaginas of healthy women attending the Correspondence: P. B. Heczko, Department of Microbiology, Jagiellonian University Medical College, 18 Czysta Street, 31-121 Cracow, Poland. Tel: + 48 12 633 25 67. Fax: + 48 12 423 39 24. E-mail: mbheczko@cyf-kr.edu.pl outpatient clinic for students in Cracow, Poland. All were premenopausal and were not menstruating at the time of collection. All of the women studied were clinically normal on vaginal examination, had a normal Nugent's score [8] and a low (4.0 to 4.5) pH of the vaginal fluid. They had neither received antibiotics nor used hormonal contraceptives during the previous 3 months. They presented a high level of personal hygiene and reported high-quality hygiene practices. They did not use vaginal douching. They were free of known STD as checked by routine methods. Informed consent was obtained from those agreeing to participate.
The strains were cultured on MRS agar [Oxoid, Basingstoke, UK]. They were initially identified using API 50 CHL (BioMerieux, Marcy l'Etoile, France) and then speciated using a polymerase chain reaction (PCR) technique and species-specific primers [9]. The strains were then stored on glass beads in MRS broth with 10% glycerol at 7708C, so that their genotypic and phenotypic properties were well preserved. Before the experiments, four to five glass beads coated with lactobacilli were transferred to 2 ml liquid MRS, then incubated at 378C for 48 h in strictly anaerobic conditions using BioMerieux gas boxes with hydrogen and carbon dioxide gas generators [BioMerieux].

Surface properties
Hydrophobicity was estimated by the salt aggregation test (SAT) according to Jonsson and Wadströ m [10]. Bacterial cell suspensions were mixed with equal volumes of ammonium sulphate of various molarities (0.01 to 4.0 mol/l 71 ). The lowest concentration of ammonium sulphate that produced visible aggregation was scored as the SAT hydrophobicity value. The aggregation in saline solution used as a control was regarded as auto-aggregation. Slime production was demonstrated using a qualitative method of Christensen et al [11]. The tested strain was grown in glass tubes in 3 ml MRS broth at 378C for 24 h. After incubation, the culture was discarded and the tube filled again with 1 ml 0.1% solution of safranin [Poch, Glinerice, Poland] and rotated for 60 s. The staining solution was decanted, and the tubes inverted and dried at room temperature. The density of biofilm covering the tube walls was estimated using a semi-quantitative scale (-to + + + ).

Adherence to vaginal cells
The ability of Lactobacillus strains to adhere to the human vaginal cell line A431 (kindly provided by Professor D. Dus, Polish Academy of Sciences) was checked by in vitro assay according to Gaffney et al., the degree of adhesion being evaluated according to a semi-quantitative score system (-to + + + ) [12]. Briefly, a 24-h culture of A431 cells at 1 x 10 5 cells/ ml, cultured in wells of a 12-well flat-bottom tissue culture plate (Iwaki, Fuuahashi, Japan) on Eagle's 1959 medium (MEM) with L-glutamine and sodium bicarbonate (IITD, Wroclaw, Poland) containing 5% fetal calf serum (Sigma-Aldrich, St Louis, USA) and antibiotics (penicillin 100 UI/ml, streptomycin 100 UI/ml, neomycin 200 mg/ml) (Sigma-Aldrich) was washed twice with PBS. Overnight cultures of bacteria were appropriately diluted with MRS + MEM (1:1) to give a concentration of approximately 10 8 CFU/ml, and was used to inoculate cells in wells; the pH of the mixture ranged from 5.0 to 5.6. After incubation at 378C for 30 min, wells were washed four times with PBS to release unbound bacteria. The cells were then fixed with 3.7% formaldehyde for 1 h, washed twice with PBS and stained with crystal violet stain (Merck, Darmstadt, Germany), washed, dried and examined under the microscope (magnification x 100). According to our laboratory standard procedure, the adherent lactobacilli were counted in 20 random microscopic fields.

Antagonism against yeasts
From among the Lactobacillus strains, 14 strains showing most surface activity were compared, using a modified agar overlay method according to Fitzimmons and Berry [13], against Candida albicans and Candida pseudotropicalis. The yeast strains were freshly isolated from clinical cases of CV confirmed by physical examination and routine culture, identified phenotypically using API 20C AUX system (BioMerieux) and then stored at 7708C as described for lactobacilli. The examined Lactobacillus strains were cultured on agar plates containing 10 ml MRS, in the form of a stripe 2 cm wide across the plate. The plates were incubated in anaerobic conditions at 378C for 24 h. The Lactobacillus growth was then overlaid with melted MYPG agar (Bacto YM Agar, Difco, Detroit, USA), MYPG with sodium thiocyanate (0.81 g/l) (POCH), MYPG with sodium thiocyanate (0.81 g/l) (POCH) and 20 mg/ml catalase (Sigma-Aldrich); the pH was adjusted to 6.2. After solidification of the agar overlay, overnight culture of C. albicans or C. pseudotropicalis in Sabouraud broth (Oxoid), diluted with PBS to 0.5 MacFarland's score, was streaked over the agar surface with a cotton swab. The plates were preincubated at 48C for 4 h in a refrigerator, then kept at 378C for 24 h in aerobic conditions and left for 24h at room temperature before reading. This culture method best displays the inhibition zones.
After adequate growth, the zones of inhibition of Candida were measured using a semi-quantitative scale on which (7) = lack of Candida growth inhibi-tion over Lactobacillus culture ( Figure 1); ( + /7) = minimal inhibition of Candida growth over Lactobacillus culture; ( + ) = partial inhibition of Candida growth over Lactobacillus culture; ( + + ) = total inhibition of Candida growth over Lactobacillus culture ( Figure 2); and ( + + + ) = total inhibition of Candida growth beyond Lactobacillus culture.
Semi-quantitative agar plate method for H 2 O 2 detection H 2 O 2 production by the 14 selected strains of vaginal lactobacilli was examined using a method described by Eschenbach et al. [14], modified by Fontaine and Taylor-Robinson [15]. The examined Lactobacillus strains were cultured in a reductive manner on MRS plates containing 4 mg/ml 3-ethyl-benzthiazoline-6sulfonic acid (ABSA) (Sigma-Aldrich) and 0.2 mg/ ml horseradish peroxidase (HRPO, Sigma-Aldrich). The plates were then incubated in anaerobic conditions at 378C for 48 h, after which they were exposed to air. The HRPO oxidised ABSA in the presence of H 2 O 2 , forming a blue pigment in and around the colonies. The first reading was performed after 1 h, the last after 24 h. The level of H 2 O 2 produced was marked according to the intensity of blue coloration, and the results were calculated by a semi-quantitative scale on which (7) = lack of colony coloration; ( + / 7) = minimal blue colour change of the colony; ( + ) = small and limited blue colour change of the colony; ( + + ) = large but incomplete blue colour change of the colony; and ( + + + ) = clearly visible complete blue colour change of the colony.

Analytical Merckoquant peroxide test strips for H 2 O 2 production
Test strips (Merck) were used to measure the quantity of H 2 O 2 on a detection scale between 0 and 100 mg/l. Lactobacillus strains were suspended in 1 ml liquid MRS and cultured at 378C in anaerobic conditions for up to 48 h. The final density of Lactobacillus cells was estimated as approximately 1 x 10 9 CFU/ml. The test strip was then immersed in the culture and read against the scale provided by the producer 1 min after removal and several times during the incubation period.

Enzyme microimmunoassay test for H 2 O 2 and O 2 production
The technique used for the measurement of H 2 O 2 and O 2 production by Lactobacillus was an adaptation of Pick and Mizel's phenol red solution method [16] for testing H 2 O 2 production by macrophages. Lactobacillus strains were suspended in 1 ml liquid MRS and cultured at 378C in anaerobic conditions for up to 48 h. The final density of Lactobacillus reached approximately 1 x 10 9 CFU/ml. The cultures were centrifuged at 1200 rpm for 7 min, and a 50 ml sample of each supernate was placed in a well of a 96-well plate (Nunc, Rochester, USA). Following this, 50 ml of reagent 1 ((2 ml of phenol red) (Sigma), 2 ml of HRPO, 46 ml DPBS buffer and 10 ml 1N sodium hydroxide) were added. Absorbance of this mixture was measured at 600 nm. Control consisted of 50 ml DBPS buffer or MRS medium with 50 ml reagent 1 and 10 ml sodium hydroxide. The results, after deduction of control values, were calculated as nmoles or transformed into mg/l to compare with the test strips method. The tests were run in triplicate.

Effect of H 2 O 2 on Candida and Lactobacillus
This was measured by incubating suspensions of randomly selected strains of Candida (approximately 10 7 CFU/ml) and Lactobacillus isolated from vaginal flora (approximately 10 8 CFU/ml) in PBS with decreasing concentrations of H 2 O 2 in PBS. Clinical isolates of Candida and selected Lactobacillus species were used. Initial numbers of tested organisms were measured by counting colonies after serial tenfold dilutions on appropriate solid media (Sabouraud agar for Candida and MRS agar for Lactobacillus). The mixtures were incubated at 378C for 1 h in an aerobic (Candida) or anaerobic (Lactobacillus) atmosphere. After this period, tenfold dilutions in PBS were prepared and plated on appropriate media. The inoculated media were cultured at 378C for 24 h in the appropriate atmosphere and colonies were counted for CFU/ml. Concentrations of H 2 0 2 before and after incubation were controlled by the test strips method.

Results
The surface properties of Lactobacillus strains and their adherence to vaginal cells are shown in Table I.
Of the 111 tested vaginal Lactobacillus strains, most produced extracellular slime-like material, but only about 3% showed higher surface hydrophobicity.
More than 50% of the tested vaginal Lactobacilli adhered to A431 line cells, although at different rates. Only a few strains were strongly adherent, whereas the rest showed moderate-to-low adhesion.
Lactobacillus strains showing high to moderate ability to adhere to vaginal cells (n = 14) were further tested for their anti-candidal activity. The selected strains were of different species, and although all of them inhibited growth of C. albicans to a certain degree, only five strains were distinctly active. This activity was inhibited by catalase incorporated into the test medium, whereas the anti-candidal properties of some less active Lactobacillus strains were not influenced. Addition of thiocyanate to test agar did not alter the degree of inhibition in any of the tested strains. Most lactobacilli were able at least slightly inhibit the growth of C. pseudotropicalis (Table II).
Only two Lactobacillus strains showed a positive reaction for H 2 O 2 1 h after the exposure of the test plates to air, and these results were then confirmed   (Table III). Selected Candida strains belonging to different species were resistant to relatively high concentrations of H 2 O 2 in test conditions. Concentrations of 30 g/l were necessary to kill all Candida cells; 3 g/l was microbicidal for only a minor proportion of Candida cells. No differences among the strains were observed, and concentrations of H 2 O 2 did not change during the incubation period (Table IV).
Lactobacillus strains producing as well as those not producing H 2 O 2 were equally sensitive to 3 g/l and higher concentrations of H 2 O 2 except for L. acidophilus G/12A and L. delbrueckii 117A, which were partially inhibited by the lowest tested concentration (150 mg/l) (Table V).

Discussion
It is generally believed that vaginal Lactobacillus strains control the vaginal microflora, including Candida albicans [6], by colonizing the vaginal epithelium and inhibiting the growth of the other microorganisms. Therefore Lactobacillus strains, as candidates for vaginal probiotics, are usually tested in vitro for their ability to adhere to the vaginal epithelium, their antimicrobial activity towards Gardnerella vaginalis and C. albicans, and their production of H 2 O 2 which, according to several authors, may be mainly responsible for inhibitory activity [1,17].
Unlike other workers [18], in this study as well as in our previous experiments on Lactobacillus adherence in vitro [4] we used the human vaginal cell line A431. This line is derived from vaginal epithelial epidermoid carcinoma. It is useful for standardizing tested bacterial adherence to vaginal cells, to enable comparison of different research approaches in this  area. Using A431 cells, we found that only a minor proportion of the tested vaginal Lactobacillus strains showed a distinct degree of adherence, in spite of securing test conditions that resembled as much as possible the vaginal milieu (the pH of the reaction mixture during the test was kept at 5.0 to 5.6). By means of a screening agar technique designed to study interactions between selected vaginal Lactobacillus strains and yeasts, we also demonstrated that only a few lactobacilli were able to inhibit C. albicans growth. Although lactic acid production seems to be a major factor in maintaining low vaginal pH and controlling strictly anaerobic bacterial vaginal flora [7,19], it is evident that this product, common to all lactobacilli, cannot be selective for acid-tolerant vaginal yeasts [20]. Our test system, based on a double layer of two media, had a low pH (6.2) initially, which became even more acidic (pH = 4.8) during incubation because of sugar fermentation by both lactobacilli and Candida.
Inhibition of Candida growth by single strains of lactobacilli in vitro was been noted some decades ago [21] and also demonstrated in vivo [22]. However, the mechanism of this growth inhibition has not been fully elucidated. The importance of H 2 O 2 production by lactobacilli, proposed as vaginal probiotics and indicated for prevention of CV, is usually stressed [13,14]. In our studies, only some of the Lactobacillus strains showing inhibitory properties against Candida produced H 2 O 2 in amounts detectable by the screening plate method commonly used in previous studies [14]. Moreover, results obtained by the more exact quantitative method confirmed that only two strains produced H 2 O 2 in amounts exceeding 20 mg/l. To check this possibility, for several Candida strains we estimated minimal microbicidal concentrations of H 2 O 2 which ranged from about 0.3% to 0.03% (3 to 0.3 g/l). These values markedly exceeded concentrations produced by tested lactobacilli. This does not support the hypothesis that for the direct inhibition of several Candida strains H 2 O 2 is solely responsible. However, successful inhibition of Candida by vaginal lactoba-cilli producing H 2 O 2 may occur in vivo, since peroxide interacts with peroxidases and halides present in vaginal secretions [23]. Moreover, it may be hypothesized that H 2 O 2 interacts in vivo with ferrous ions present in vaginal secretions, especially during menstruation, to form highly reactive hydroxyl radicals via the Fenton reaction [24].
Catalase markedly diminished the inhibitory activity of the two most active lactobacilli producing H 2 O 2 , and also the activities shown by other strains that did not produce H 2 O 2 in all three tests. This may suggest that the enzyme interferes with other products of lactobacilli with anti-candidal activity.
Fitzsimmons and Berry [13] postulated that anticandidal activity is based on the production of hypothiocyanate from H 2 O 2 . However, this possibility has not been confirmed in our observations, since addition of thiocyanate to the test medium had no influence on the degree of inhibition.
Several of our Lactobacillus strains that were inhibitory to C. albicans showed that their activity was probably independent of H 2 O 2 production. Sookkhee et al. [25] recently found that the anticandidal activity of some oral Lactobacillus strains was not inhibited by heat treatment, which also suggested alternative mechanisms of inhibition. In fact, several antifungal substances such as cyclic dipeptides, pyroglutamic acid and other substances of low molecular mass, such as lactones, produced by different Lactobacillus species, have recently been characterized [19,26].

Conclusions
In the light of our experiments it seems that anticandidal activity of vaginal Lactobacillus isolates is related to various overlapping mechanisms. Production of H 2 O 2 does not seem to be a major mechanism in the direct inhibition caused by lactobacilli, and it certainly is not the sole active inhibitory product.
Certain strains of Lactobacillus genus, isolated from the vagina and representing major species colonizing vaginal mucosa of healthy women, express distinct adherence and antagonistic properties in vitro which may indirectly indicate their protective role in stabilizing the vaginal microflora and controlling other members of the vaginal flora, including Candida albicans. It seems that inhibition of C. albicans by different Lactobacillus strains is dependent on various mechanisms. Although maintenance or reconstruction of the normal composition of the vaginal microflora by applying properly selected lactobacilli may be of prophylactic value in preventing various genitourinary system infections in women, including CV, more studies based on well standardized methods of testing candidates for vaginal probiotics are necessary, before entering the stage of clinical studies. The in vitro tests used by us could be useful in the selection of candidate Lactobacillus strains for vaginal probiotics.