From more than 100 lactic acid bacteria, Weissella kimchii PL9023 was selected as producing the most hydrogen peroxide and was characterized as a probiotic for women. W. kimchii PL9023 inhibited the growth and adherence of vaginal isolates of Candida albicans, Escherichia coli, Staphylococcus aureus and Streptococcus agalactiae. The surface components involved in adherence of W. kimchii PL9023 to vaginal epithelial cells appeared to be glycoproteins, as determined by susceptibility to proteinase K, heat, or periodate. W. kimchii PL9023 did not produce harmful metabolites or enzymes. The results of this study suggest that W. kimchii PL9023 is a potential probiotic for vaginal health.
Lactic acid-producing bacteria (LAB) have long been considered to constitute the primary microbiological barrier to infection by urogenital pathogens. They exert a protective role mainly through a combination of steric exclusion and inhibitory substance production . An average healthy premenopausal woman has 107–108 CFU of LAB in 1 g vaginal fluid. Lactobacillus crispatus, L. fermentum, L. gasseri, L. iners, L. jensenii, L. pentosus, L. plantarum, L. vaginalis, and Weissella viridescens have been detected in Western women [2–4]. Most of these species of LAB have rarely been found in Korean women suggesting regional differences in vaginal microflora. To develop a probiotic for Korean women, it was hypothesized that LAB isolated from Korean women would easily colonize in Korean women. Since hydrogen peroxide-producing LAB are predominant in the vaginal tract of healthy women [5,6], a vaginal isolate from Korean women that produced the most hydrogen peroxide was selected and characterized in terms of various activities pertinent to its application as a probiotic for vaginal health.
2 Materials and methods
2.1 Isolation of hydrogen peroxide-producing LAB
Hydrogen peroxide-producing LAB were detected using a modified version of the method developed by Rabe and Hillier . LAB were inoculated on TMB-Plus agar containing horseradish peroxidase and 3,3′,5,5′-tetramethylbenzidine dihydrochloride (TMB) and incubated at 37 °C for 48–72 h in an anaerobic environment. After incubation, the plate was exposed to air for 30 min; then, the oxidation of TMB by peroxidase in the presence of hydrogen peroxide produced blue pigments on the colonies. Blue colonies were selected, and the amount of hydrogen peroxide produced by each colony was measured. After culture overnight in Man–Rogosa–Sharpe (MRS) broth (Difco, Sparks, MD), bacterial cells were collected by centrifugation, washed with Brucella broth (Difco), and suspended in Brucella broth to an OD of 4 (at 600 nm). After incubation for 4 h, the amount of hydrogen peroxide produced by 50 μl of the culture supernatant was assayed using an Amplex Red Hydrogen Peroxide/Peroxidase Assay kit (Molecular Probes, Eugene, OR). The isolate (PL9023) producing the most hydrogen peroxide was selected and characterized as a potential probiotic.
2.2 Identification of Weissella kimchii PL9023 and preparation of bacterial cells
W. kimchii PL9023 was originally isolated from the vagina of a healthy premenopausal woman. This isolate was identified as L. coprophilus (73.3%) according to Bergy's manual of systematic bacteriology using an API50CH kit (Bio-Merieux, Marcy-l'Etoile, France). Sequencing of its 16S rDNA was performed using universal primers  as described previously . Since W. kimchii has been recently branched from W. confusa , PL9023 was identified again as W. kimchii with amplified rRNA restriction analysis (ARDRA) following the procedure of others . Vaginal isolates of Candida albicans (CCARM 14021), Staphylococcus aureus (CCARM 3707), and Escherichia coli (CCARM 1428) were grown in potato dextrose agar (Difco), mannitol salt agar (Mast Group, Merseyside, UK), and MacCONKEY agar (Mast Group), respectively. Vaginal isolates of Streptococcus agalactiae (CCARM 4506) were cultured in blood agar (Difco) in a 10% CO2 incubator.
2.3 Overlay method to detect the inhibitory activity on growth of vaginal isolates
An aliquot (10 μl) of an extract of ethyl acetate of the culture supernatant of W. kimchii PL9023, avoiding the effect from pH and hydrogen peroxide, was laid on each pathogen in the early stationary phase, after inoculation by a swab with Muller Hinton agar (Difco). After culture overnight, the growth-inhibition zone was measured. The extract was prepared by extracting equal volumes of culture supernatant and ethyl acetate, which then underwent evaporation and dispersion in distilled water (1/100 of the original volume).
2.4 Preparation of vaginal epithelial cells
Vaginal epithelial cells were prepared as described by Wood et al. . In brief, vaginal epithelial cells were obtained from three consecutive scrapings from the lateral vaginal wall, using a new pop brush for each scraping, from a healthy premenopausal woman. Cells were dispersed in 10 ml minimal essential medium (MEM; GIBCO BRL, Grand Island, NY) at various pHs. Indigenous bacteria were removed from cells by shaking in 10 ml MEM for 30 s and by centrifugation (750g) for 5 min three times. Then, cells were filtered through a membrane filter (8 μm, mixed cellulose acetate and nitrate; Millipore Corporation, Bedford, MA) and dispersed in MEM, at a final concentration of 2 × 104 cells ml−1.
2.5 Adherence of W. kimchii PL9023 to vaginal cells
Adherence of W. kimchii PL9023 to vaginal cells was assayed as described by Boris et al. . The same volumes of W. kimchii PL9023 cells (106 cells ml−1 in MEM) and vaginal cells (2 × 104 cells ml−1 in MEM) were mixed in an orbital shaker (100 rpm) for 30 min at 37 °C. The mixture was passed through a membrane filter (8 μm) and washed with the same volume of MEM. Cells retained on the filter were placed on an albumin-coated microscope slide (Fisher) and were fixed with 97% methanol. Cells were observed under a light microscope after Gram staining. The number of bacterial cells bound to one vaginal cell was calculated by averaging the numbers of bacterial cells that had adhered to >60 vaginal cells. To observe the effect of pH on adherence, the pH of the medium was adjusted to 4.4, 6.0 and 7.2 with 0.1 N HCl.
2.6 Hydrophobicity determination
The hydrophobicity and ionic characteristics of the cell surface of W. kimchii PL9023 were determined by the bacterial adhesion to hydrocarbons (BATH) method  and the microbial adhesion to solvents (MATS) method . In brief, bacterial cells at the stationary phase were harvested by centrifugation at 5000g for 10 min and were washed twice and suspended in 0.1 M KNO3 (pH 6.2) to an OD of 1.0 (at 600 nm) (∼108 CFU ml−1). The cell suspension (1.2 ml) was mixed with 0.2 ml of each solvent for 2 min. After phase separation (∼15 min), the OD of the aqueous phase was measured at 600 nm. The relative hydrophobicity (RH) of cells was calculated using the following equation:
(ODafter indicates the OD at 600 nm after extraction, and ODbefore indicates the OD at 600 nm before extraction).
2.7 Biochemical treatment to determine the cell-adhesion factor
W. kimchii PL9023 cells were treated for 1 h at 37 °C with 1 mg ml−1 (final concentration) proteinase K to remove proteins or with 1 mg ml−1 (final concentration) lipase to remove lipids and then were washed twice with PBS (pH 7.2). Alternatively, cells were treated with 100 mM (final concentration) sodium periodate in the dark at 4 °C overnight to block carbohydrate moieties, and then washed three times with phosphate buffered saline (PBS, pH 7.2). Bacterial cells were heated in boiling water for 10 min and were washed twice with PBS to denature the proteins. Treated bacterial cells were added to vaginal cells, and the number of adhered W. kimchii PL9023 cells was counted as described above.
2.8 Aggregation test
W. kimchii PL9023 and C. albicans were prepared as described above and suspended in pH-adjusted MEM. Incubation was in an orbital shaker (100 rpm) at 37 °C. After 4 h, bacterial cells were Gram-stained and observed under a light microscope.
2.9 Interference assay
After W. kimchii PL9023 (108× CFU ml−1) and vaginal epithelial cells (2 × 104 cells ml−1) were mixed for 30 min, each pathogen (1–2 × 106 CFU ml−1) was added to the mixture, which was incubated for another 30 min to assay the interference activity of W. kimchii PL9023 on pathogen binding to vaginal epithelial cells. To assay competition activity, W. kimchii PL9023, each pathogen, and vaginal epithelial cells were mixed at the same time and incubated for 30 min. To assay displacement activity, each pathogen was incubated with vaginal epithelial cells 30 min before W. kimchii PL9023 was added. After incubation, vaginal epithelial cells were treated and observed as described above.
2.10 Statistical analysis
Each experiment was performed in triplicate and statistical analyses were performed using SPSS, version 10.0 (SPSS Inc., Chicago, IL).
2.11 In vitro safety test of W. kimchii PL9023
Published methods were used to detect various virulence factors, such as the production of amine , ammonia , and indole , the degradation of gelatin  and the presence of β-glucuronidase  and nitroreductase .
3.1 Isolation and identification of W. kimchii PL9023
Of >100 LAB isolates, several produced blue-colored colonies on TMB-Plus medium. Of these, every vaginal isolate of W. kimchii produced hydrogen peroxide in the quantitative assay and W. kimchii PL9023 produced the largest amount of hydrogen peroxide (96.86 μM) while the type strain (W. kimchii CHJ3 KCTC3746) produced 38.71 μM. W. kimchii PL9023 was originally isolated from a healthy premenopausal woman's vagina and was identified as L. coprophilus with API 50CH (bioMérieux, Marcy l'Etoile, France). The sequence of 16S rRNA matched 100% with W. kimchii (GenBank accession no. AY190629) and was deposited in the Korea Agriculture Culture Collection as KACC 91139. L. coprophilus was a synonym of L. confusus  and was then renamed as Weissella confusa , which was recently branched into W. kimchii .
3.2 Biochemical properties of W. kimchii PL9023
An extract of the spent culture supernatant of W. kimchii PL9023 inhibited the growth of various pathogens, including E. coli, S. aureus, Strep. agalactiae, and C. albicans (data not shown).
When the RH of the bacterial cell surface was assayed by the BATH method, RH values were 95.77 ± 0.32% in hexadecane and 85.34 ± 4.79% in xylene, suggesting high hydrophobicity of the bacterial cell surface. When the ionic characteristics were determined by the MATS method, a value of 94.42 ± 0.92% in chloroform showed that W. kimchii PL9023 has a strong affinity to chloroform-acidic electron acceptors. A small RH value (45.70 ± 2.27%) in ethyl acetate confirmed the non-acidic and electron donor properties of the cell surface of W. kimchii PL9023.
W. kimchii PL9023 was negative for gelatin liquefaction, ammonia production, indole production, phenylalanine degradation and the presence of β-glucuronidase and nitroreductase.
3.3 Adhesion properties of W. kimchii PL9023 to vaginal cells
W. kimchii PL9023 bound to vaginal epithelial cells at various pHs. The number of adhered bacterial cells (dark purple color) to vaginal cells (light pink color with a dark pink colored nucleus) increased at a lower pH (Fig. 1). Lipase treatment did not affect the binding of W. kimchii PL9023 to vaginal epithelial cells. When protein was removed by proteinase K or by denaturation with heat, the binding decreased significantly. Periodate treatment to modify carbohydrate moieties also decreased binding, suggesting that glycoprotein is the cell-adhesion factor (Fig. 2).
Effect of various treatments on the adherence of Weissella kimchii PL9023 to vaginal epithelial cells. Treatment of W. kimchii PL9023 was as follows: □, control; □, proteinase K; □, lipase; □, sodium periodate; □, heat. ***, P < 0.001.
3.4 Aggregation test
Self-aggregation has been suggested to be one of the factors in the constitution of the microflora . As shown in Fig. 3(a), W. kimchii PL9023 showed the best self-aggregation at pH 4.0. When W. kimchii PL9023 was mixed with various uropathogenic pathogens in cell culture medium at various pHs, it aggregated with C. albicans at all three pHs. Large purple cells of C. albicans in yeast form, as well as hyphal form, that were surrounded by W. kimchii PL9023 were easily observed under the light microscope (Fig. 3(b)). Whereas W. kimchii PL9023 did not aggregate significantly with other uropathogenic pathogens.
Aggregation of Weissella kimchii PL9023 with Candida albicans. (a) Self-aggregation of W. kimchii PL9023. (b) Aggregation of W. kimchii PL9023 with C. albicans.
3.5 Interference assay
As shown in Fig. 4, W. kimchii PL9023 significantly decreased (P < 0.001) the adherence of C. albicans to vaginal cells at pH 4.4, 6.0 and 7.2 in all three tests. Regarding the adherence of E. coli and S. agalactiae, a significant decrease induced by W. kimchii PL9023 was observed only at pH 6.0. Since neither E. coli nor S. agalactiae bound well to vaginal cells at pH 4.4, it was difficult to observe any change. W. kimchii PL9023 significantly decreased (P < 0.001) the adherence of S. aureus at pH 4.4 in the competition and exclusion tests. W. kimchii PL9023 significantly decreased (P < 0.001) the adherence of S. agalactiae at pH 6.0 in the exclusion and displacement tests.
The adherence of various urinary tract pathogens in the presence of Weissella kimchii PL9023. The number of each pathogen adhered to one vaginal cell in the presence of Weissella kimchii PL9023 was counted. (□, Control; □, competition; exclusion;, displacement ***P < 0.001).
It has been well documented that specific strains of LAB exhibit antagonistic activity toward various human intestinal pathogens, producing antimicrobial compounds such as lactic and acetic acids, hydrogen peroxide and bacteriocins, as well as compounds not yet identified . The capability of LAB to adhere to vaginal epithelial cells is also an important factor in the formation of a barrier that prevents the colonization of pathogenic bacteria at these sites [26,27]. The binding of LAB to cells involves non-specific factors, especially hydrophobicity , surface charge , and specific moieties on the surface such as protein and carbohydrates . The basic properties of the cell surface can create Lewis acid-based interactions with acceptors on vaginal epithelial cells .
Hydrogen peroxide producing activity has been suggested as the main protective factor from vaginal pathogens . Every vaginal isolate of W. kimchii in this study produced hydrogen peroxide suggesting hydrogen peroxide producing activity may be a common characteristic of W. kimchii. In addition to the hydrogen peroxide producing activity, W. kimchii PL9023 produced bacteriocin-like substances which inhibited the growth of various vaginal pathogens. W. kimchii PL9023 aggregated with C. albicans at various pHs and prevented C. albicans from binding to vaginal cells. Coaggregation must be advantageous for W. kimchii PL9023 to exert a direct inhibitory effect on C. albicans.
High RH and basic characteristics of the cell surface of W. kimchii PL9023 could have an advantage in binding to vaginal epithelial cells, especially at low pH. The adhesion factor for W. kimchii PL9023 is thought to be glycoprotein, since binding decreased when protein was removed or denatured and carbohydrate moieties were modified.
Since W. kimchii PL9023 can inhibit the growth of various vaginal pathogens while producing no harmful metabolites or enzymes, it can be included in a formula for local application in the vaginal tract  or in an oral probiotic .
The author thanks Hak-mi Lee for her excellent technical assistance. This work was supported by a Bahrom grant from Seoul Women's University.
(1999) Cell surface-associated lipoteichoic acid acts as an adhesion factor for attachment of Lacto bacillus johnsonii La1 to human enterocyte-like Caco-2 cells. Appl. Environ. Microbiol. 65, 1071–1077.