Characterization and selection of vaginal Lactobacillus strains for the preparation of vaginal tablets


Correspondence to: Dr P. Mastromarino, Istituto di Microbiologia, Università‘La Sapienza’, Piazzale Aldo Moro 5, I-00185 Roma, Italy (e-mail:


Aims: To characterize and select Lactobacillus strains for properties that would make them a good alternative to the use of antibiotics to treat human vaginal infections.

Methods and Results: Ten Lactobacillus strains belonging to four different Lactobacillus species were analysed for properties relating to mucosal colonization or microbial antagonism (adhesion to human epithelial cells, hydrogen peroxide production, antimicrobial activity towards Gardnerella vaginalis and Candida albicans and coaggregation with pathogens). The involvement of electrostatic interactions and the influence of bacterial metabolic state in the binding of lactobacilli to the cell surface were also studied. Adherence to epithelial cells varied greatly among the Lactobacillus species and among different strains belonging to the same Lactobacillus species. The reduction in surface negative electric charge promoted the binding of several Lactobacillus strains to the cell membrane whereas lyophilization reduced the adhesion capacity of many isolates. The antimicrobial activity of lactobacilli culture supernatant fluids was not directly related to the production of H2O2.

Conclusions: Three strains (Lactobacillus brevis CD2, Lact. salivarius FV2 and Lact. gasseri MB335) showed optimal properties and were, therefore, selected for the preparation of vaginal tablets. The selected strains adhered to epithelial cells displacing vaginal pathogens; they produced high levels of H2O2, coaggregated with pathogens and inhibited the growth of G. vaginalis.

Significance and Impact of the Study: The dosage formulation developed in this study appears to be a good candidate for the probiotic prophylaxis and treatment of human vaginal infections.


Lactobacilli are the dominant bacteria of a healthy human vagina and their presence and number are influenced by oestrogen production, which undergoes age- and menstrual cycle-dependent changes (Keane et al. 1997). The function of lactobacilli is to maintain an environment that limits the growth of pathogenic micro-organisms. Variations in a normal female genital microflora, and in particular the loss of H2O2-producing lactobacilli, cause an increase in genital and urinary infections (Redondo-López et al. 1990; Klebanoff et al. 1991; Hawes et al. 1996). Moreover, recent studies have demonstrated a significant association between infection with HIV-1 and the depletion of vaginal lactobacilli, particularly among women with severe bacterial vaginosis (Martin et al. 1999). These observations have led to research on the strains and properties of vaginal lactobacilli, which may be responsible for the maintenance of a pathogen-free environment in the urogenital tract. Lactobacilli are able to interfere with genitourinary pathogens by different mechanisms, including competitive exclusion of pathogens from the cell surface, production of antimicrobial compounds, competition for nutrients and stimulation of the immune system (Reid et al. 1987; Charteris et al. 1997). There is a growing interest in the use of lactobacilli of human origin as probiotics against urogenital tract infections (McGroarty 1993; Elmer et al. 1996; Charteris et al. 1997). The term ‘probiotics’ was redefined by Havenaar et al. (1992) as ‘a viable mono or mixed culture of micro-organisms which, applied to animal or man, beneficially affects the host by improving the properties of the indigenous microflora’.

In this study, 10 different Lactobacillus strains were analysed for properties relevant to mucosal colonization or antagonism (adhesion to epithelial cells, hydrogen peroxide production, antimicrobial activity towards Gardnerella vaginalis and Candida albicans and coaggregation with pathogens). The purpose of the work was to characterize and select Lactobacillus strains for the preparation of tablets to be used to treat vaginal infections as an alternative to antibiotics. The capacity of the strains to maintain their biological characteristics during manufacturing of the vaginal tablets was also studied. The effect of lyophilization and of technological processing on the viability and adhesiveness of each strain was evaluated. Three strains (Lactobacillus brevis CD2, Lact. salivarius ssp. salicinius FV2 and Lact. gasseri MB335) were selected for their optimal properties to design a product for local application to the vaginal tract. Vaginal tablets containing a mixture of the three strains of lactobacilli were produced and tested for the capacity to interfere with G. vaginalis and C. albicans adhesion to the cell surface, for the production of antimicrobial compounds and for the maintenance of lactobacilli viability.

Materials and methods

Strains and culture conditions

Ten Lactobacillus strains of vaginal origin were studied. They were identified as Lact. brevis (CD2), Lact. salivarius ssp. salicinius (FV2), Lact. salivarius ssp. salivarius (FV3), Lact. crispatus (FV4) and Lact. gasseri (MB 331, MB 332, MB 333, MB 334, MB 335 and MB 336) by standard biochemical tests (Hammes and Vogel 1993) and API 50 CHL system (BioMérieux, Marcy l'Etoile, France). The organisms were inoculated from frozen (− 80 °C) vials onto de Man-Rogosa-Sharpe (MRS) broth (Oxoid). Following 48 h of culture at 37 or 30 °C (Lact. brevis) in anaerobic conditions (Gas-Pak System; BBL, Becton Dickinson Biosciences, Milan, Italy) the organisms were checked for purity and subcultured in MRS broth.

Gardnerella vaginalis and C. albicans, isolated from vaginal swab samples, were identified by standard techniques and cultured on Gardnerella Selective Agar (GSA; BBL) and Sabouraud's Dextrose Agar (SDA), respectively.

Lactobacillus species identification using polymerase chain reaction

The taxonomic identification of the Lactobacillus strains was confirmed by molecular characterization, using polymerase chain reaction (PCR)-mediated amplification with species-specific primer sets based on the 16S, 16S-23S and the flanking 23S ribosomal RNA regions. Primer sets Lgas-1/Lgas-2, Lsal-1/Lsal-2, Lcri-1/Lcri-2 (Song et al. 2000) and Lbre-1/Lbre-2 (Guarneri et al. 2001), specific for the species Lact. gasseri, Lact. salivarius, Lact. crispatus and Lact. brevis, respectively, were employed in the PCR conditions reported by the authors. All amplification reactions were carried out in a Thermal Cycler II (Biometra, Göttingen, Germany) and Dynazyme II (Celbio, Milan, Italy) was used as thermostable polymerase in the condition suggested by the supplier. The total volume of each reaction mixture was 25 µl and cells of Lact. salivarius, Lact. crispatus and Lact. gasseri grown on agar plates were used directly as template, whereas total chromosomal DNA from Lact. brevis was extracted according to Guarneri et al. (2001). In the amplification reactions, Lact. gasseri ATCC 33323, Lact. salivarius ssp. salicinius ATCC 11742 and Lact. crispatus ATCC 33197 were used as reference strains. Aliquots (5–10 µl) of the amplified products were subjected to gel electrophoresis in 2% agarose gel and visualized by ethidium bromide staining.

Cell and tablet production

For the batch production of lactobacilli, the strains were grown in a 12-l fermenter jar equipped with a combination pH electrode connected to an automatic pH controller (Applikon dependable Instruments, Schiedam, the Netherlands). Prior to use, the entire fermenter assembly was autoclaved for 15 min at 121 °C. Ten litres of medium (containing (g l−1): whey permeate, 40; casein peptone, 20; yeast extract, 5; glucose, 20; Tween 80, 1; sodium acetate, 5; (NH4)2SO4, 2; sodium citrate, 2; MgSO4.7H2O, 0·2; MnSO4.4H2O, 0·05) were added to the sterilized fermenter jar and all was sterilized for 20 min at 115 °C. The temperature of the medium was set at the appropriate value, the pH adjusted to 6·5 with 30% NaOH and the automatic pH controller adjusted to maintain the growth medium at a pH of 6·5 with a 30% NaOH neutralizer. The medium was then inoculated with 500 ml of MRS lactobacilli culture, previously incubated for 16 h at 37 or 30 °C (Lact. brevis). After 8 h of fermentation, the fermenter was refrigerated at 4 °C and cells harvested by centrifugation for 20 min at 6700 g at 4 °C. The pellet was resuspended at 50% in cryoprotecting medium (skim milk powder, 10% (w/v); malt extract, 10% (w/v)), previously sterilized for 20 min at 115 °C and the pH adjusted to 6·4 with 30% (w/v) NaOH. The concentrated cultures were then freeze dried. For each Lactobacillus strain, a batch of vaginal tablets (containing approx. 109 viable bacteria per tablet) was produced as previously described (Maggi et al. 2000).

Antimicrobial supernatant fluid activity

Lactobacilli were tested for their ability to produce inhibitory substances by growing strains in MRS broth for 36 h at 37 or 30 °C (Lact. brevis). The supernatant fluids were separated by centrifugation, neutralized and filter sterilized. Agar plates of GSA and SDA were sown with the pathogens G. vaginalis and C. albicans, respectively and 20 µl of supernatant fluids were tested by disc diffusion assay. After incubation of G. vaginalis plates at 37 °C and C. albicans plates at 28 °C for 48 h, the diameters of the inhibition haloes were measured.

Determination of H2O2 production

Lactobacillus strains were tested for their ability to produce H2O2 in a semiquantitative assay on tetramethylbenzidine agar plates using the method described by Eschenbach et al. (1989) using Brucella agar (Difco) containing 0·001% (w/v) horseradish peroxidase (Sigma), 0·023% (w/v) tetramethylbenzidine (Sigma) and 1% (w/v) starch. This medium was supplemented with 0·5 mg of bovine haemin (Sigma) and 0·1 mg of vitamin K1 (Sigma) per 100 ml medium. Serial dilutions of lactobacilli were inoculated onto this medium and incubated in anaerobic conditions (Anaerobic System Mod. 1024; Forma Science, Marietta, OH, USA) at 37 °C for 72 h. Plates were then exposed to ambient air and H2O2-producing colonies appeared blue. The time taken for the blue colouration to appear was recorded and used as an indication of the quantity of H2O2 produced (Rosenstein et al. 1997).

Coaggregation assay

Lactobacillus strains were tested for their capacity to coaggregate with the pathogens. The assay was performed as previously reported by Reid et al. (1990), with some modifications. Briefly, 1 ml of each Lactobacillus suspension (109 ml−1 phosphate-buffered saline (PBS)) was mixed with 1 ml of C. albicans or G. vaginalis suspension (109 ml−1 PBS) on a vortex mixer for at least 10 s and then incubated in a 12-well tissue culture tray for 4 h at 37 °C, under agitation. The suspensions were then observed by inversion light microscopy to evaluate the aggregation degree and scored according to a scale described elsewhere (Reid et al. 1990), from 0 for no aggregation to 4 for maximum aggregation. In addition, the aggregates were observed under a phase-contrast microscope after Gram staining.

Adhesion test

Cells.  HeLa cells were grown in 75-cm2 flasks (Falcon, Becton Dickinson Biosciences, Milan, Italy) to confluent monolayers in 5% CO2 at 37 °C in Eagle's Minimal Essential Medium (Gibco, Invitrogen S.r.l., San Guiliano Milanese, Milan, Italy) containing 6% foetal bovine serum (FBS), 2 mmol l−1 glutamine, 100 IU ml−1 penicillin G and 100 µg ml−1 streptomycin.

Preparation of micro-organisms for adherence assay. Lactobacilli were stored at 4 °C as a lyophilized powder in anaerobic conditions until use. Before the adherence assay, all bacterial strains were cultured in MRS broth at 37 or 30 °C (Lact. brevis) in an anaerobic atmosphere. Candida albicans, from fresh overnight SDA subcultures incubated at 28 °C, was grown in liquid yeast extract medium at 28 °C for 18 h under constant shaking (Segal et al. 1982). These conditions yield cultures composed primarily of blastospores at the late exponential growth phase. Gardnerella vaginalis, stored at −80 °C, was grown in protease peptone maltose dextrose broth containing 10% FBS (Catalanotti et al. 1994), in a 10% CO2 atmosphere.

After 18 h incubation, micro-organisms were refreshed in newly made medium and incubated overnight in the same experimental conditions. Cultures of micro-organisms were then washed twice at 2200 g for 10 min in PBS, pH 7·2. The working dilution of the micro-organism suspensions was determined by performing sequential measurements of optical densities (O.D.) of cultures at 600 nm and quantification of viable micro-organisms by colony counts. For viable counts, tenfold dilutions of micro-organism suspension were spread in triplicate onto appropriate agar plates. Colonies were enumerated after 48 h incubation of plates in appropriate atmosphere and temperature conditions. For each strain, the correlation between the O.D. of micro-organism dilutions and colony-forming units was established.

In some experiments the lactobacilli from prepared vaginal tablets were used. Each tablet was dissolved in 10 ml of PBS by vortexing; lactobacilli were washed twice with large quantities of PBS to remove excipients used for tablet manufacture (Maggi et al. 2000) and directly employed in the adherence assay, after quantification by measurement of the O.D. In other experiments the washed lactobacilli from tablets were cultured overnight in MRS broth before the adherence assay.

Adherence assay.  The adhesion reaction was performed in a 24-well tissue culture plate containing a sterile coverslip in each well. HeLa cell suspension (1 ml), at a concentration of 4·5 × 104 cells ml−1, was seeded onto each well and incubated in a 5% CO2 atmosphere at 37 °C. After 48 h, the cells, grown to approx. 60% confluence, were washed twice with PBS and a 0·1-ml suspension of lactobacilli at a concentration of 5 × 109 bacteria ml−1 added. The plates were then incubated for 1 h at 37 °C in microaerophilic conditions to allow attachment. Cell monolayers were then washed several times in PBS, fixed with 0·4 ml May–Grünwald per well for 4 min, washed with water, Giemsa stained for 15 min, water-washed again and air dried. By microscopy, each HeLa cell was scored for the presence and number of bacteria attached. Each adherence assay was conducted in duplicate and 500 randomly chosen cells were evaluated for micro-organism adhesion.

To assess the influence of lactobacilli on the capacity of vaginal pathogens to adhere to host cells, interference experiments were performed with C. albicans and G. vaginalis. For the exclusion test, HeLa cell monolayers were inoculated with 0·1 ml suspension of Lactobacillus (5 × 109 bacteria ml−1) from overnight culture of a vaginal tablet. After 1 h incubation at 37 °C in microaerophilic conditions, cells were extensively washed and then added with 0·1 ml of C. albicans (109 yeasts ml−1) or G. vaginalis (5 × 109 bacteria ml−1). After a further 1 h incubation, cells were washed and stained as described above. For the displacement test, the pathogens were incubated with cells before the addition of lactobacilli.


Lactobacillus strain identification

Pure cultures of 10 strains of Lactobacillus isolated from the human vagina were identified at species level by using standard biochemical tests, such as the carbohydrate fermentation patterns and the API 50 CHL system (BioMérieux). The identification was confirmed by molecular analysis based on the 16S, 16S-23S and flanking 23S rDNA PCR amplification. Figure 1 shows the different amplification patterns of the tested strains obtained by using the different primer sets, Lgas-1/Lgas-2, Lsal-1/Lsal-2, Lcri-1/Lcri-2 and Lbre-1/Lbre-2, specific for the species Lact. gasseri, Lact. salivarius, Lact. crispatus and Lact. brevis, respectively. Lactobacillus gasseri strains MB 331, MB 332, MB 333, MB 334, MB 335 and MB 336 produced an amplification fragment of the right size (360 bp) with Lgas-1/Lgas-2 primers. Both Lact. salivarius ssp. salicinius FV2 and Lact. salivarius ssp. salivarius FV3 strains provided the peculiar 411-bp amplicon with Lsal-1/Lsal-2 set. The Lact. crispatus-specific 522-bp amplicon was obtained in Lact. crispatus FV4 amplification with the Lcri-1/Lcri-2 primers and the expected product of 1340 bp was recovered by amplification of Lact. brevis CD2 with the Lbre-1/Lbre-2 primer set.

Figure 1.

Amplification patterns of Lactobacillus strains isolated from the human vagina by using the primer sets Lgas-1/Lgas-2, Lsal-1/Lsal-2, Lcri-1/Lcri-2 and Lbre-1/Lbre-2 specific for the species Lactobacillus gasseri, Lact. salivarius, Lact. crispatus and Lact. brevis, respectively. Lanes: 1, molecular weight marker 100-bp DNA ladder; 2–8, Lact. gasseri strains ATCC 33323, MB 331, MB 332, MB 333, MB 334, MB 335 and MB 336 amplified with Lgas-1/Lgas-2; 9–11, Lact. salivarius ssp. salicinius ATCC 11742, Lact. salivarius ssp. salicinius FV2 and Lact. salivarius ssp. salivarius FV3 amplified with Lsal-1/Lsal-2; 12–13, Lact. crispatus strains ATCC 33197 and FV4 amplified with Lcri-1/Lcri-2; 14–15, Lact. brevis strains ATCC 4006 and CD2 amplified with Lbre-1/Lbre-2; 16, λ DNA/EcoRI + HindIII marker

Adhesion of Lactobacillus strains to HeLa cells and the effect of DEAE dextran

The adhesion of micro-organisms to epithelial cells represents an essential step for colonization and persistence in a specific site. Since adhesive properties vary considerably between Lactobacillus strains (Reid et al. 1987; Chauvière et al. 1992; Andreu et al. 1995), we first studied the adhesion capacity of the strains of lactobacilli to HeLa cells, a cell line that originated from a human carcinoma of the cervix. The adhesion of the different strains, reported in Table 1, is expressed as the percentage of colonized cells and the average number of adherent bacteria per colonized cell or per cell. The last value was calculated on the basis of the data obtained by microscopic examination reported in the first two columns (according to the formula: mean no. of adherent bacteria per infected cell × % cells with adherent bacteria × 10−2) and represents a global evaluation of the adhesion. The results indicated that adherence to epithelial cells varied greatly among the Lactobacillus strains studied. Four strains (Lact. salivarius FV2 and FV3, Lact. crispatus and Lact. gasseri MB 333) adhered at low levels (one to three adherent bacteria per cell), two strains of Lact. gasseri (MB 332 and MB 336) showed an intermediate adhesiveness (six to nine bacteria per cell), whereas Lact. brevis and three different strains of Lact. gasseri (MB 331, MB 334 and MB 335) were highly adhesive (20–28 bacteria per cell).

Table 1.  Adhesion *of Lactobacillus strains to HeLa cells and effect of DEAE dextran (DD) †
Lactobacillus strain% Cells with adherent bacteriaMean no. of adherent bacteria/ infected cellMean no. of adherent bacteria/ cell (s.d.)
  • *

    Quantitative bacterial adhesion was analysed on 500 randomly chosen cells in three independent experiments conducted in duplicate.

  • †DD was added to the adhesion medium at a concentration of 1 mg ml−1.

Lact. brevis CD296·128·827·7 (6·21)
+ DD99·350·049·6 (15·21)
Lact. salivarius FV229·5 8·0 2·4 (1·43)
+ DD53·215·2 8·1 (2·17)
Lact. salivarius FV320·8 7·1 1·5 (0·15)
+ DD72·221·515·5 (5·34)
Lact. crispatus FV431·6 6·8 2·1 (1·40)
+ DD43·8 9·6 4·2 (2·11)
Lact. gasseri MB 33183·625·421·2 (7·33)
+ DD94·023·221·8 (7·42)
Lact. gasseri MB 33268·613·2 9·0 (3·11)
+ DD77·710·8 8·4 (2·39)
Lact. gasseri MB 33336·7 6·9 2·5 (1·04)
+ DD44·3 5·2 2·3 (0·95)
Lact. gasseri MB 33491·926·224·1 (8·88)
+ DD89·026·523·6 (8·81)
Lact. gasseri MB 33585·423·219·8 (7·91)
+ DD99·133·132·8 (9·74)
Lact. gasseri MB 33661·310·2 6·2 (4·00)
+ DD91·312·911·8 (3·96)

To achieve the best degree of lactobacilli adhesion to the cell surface, the effect of DEAE dextran (Sigma) on the binding of the micro-organisms to the cell membrane was evaluated. This compound was chosen to verify the involvement of electrostatic interaction in the binding of lactobacilli to HeLa cells. Different studies suggested that charge interactions mediate the attachment of oral lactobacilli to tissues and surfaces (Harty et al. 1993; McGrady et al. 1995). In the presence of DEAE dextran (Table 1) six of the 10 strains (Lact. brevis, Lact. salivarius FV2 and FV3, Lact. crispatus, Lact. gasseri MB 335 and MB 336) showed an increased adhesion capacity, including either highly, intermediate or poorly adhesive micro-organisms. The improved capacity related either to the percentage of cells with adherent bacteria or the number of adherent bacteria per infected cell. Improvement was particularly significant for Lact. salivarius FV3 that showed a 10-fold increase in the adhesion to the cell membrane. On the other hand, the same compound did not affect the adhesiveness of the other strains.

Influence of metabolic state on lactobacilli adhesion

Tablets were obtained from the freeze-dried preparation of a single Lactobacillus strain. To evaluate the effect of the bacterial metabolic condition and technological processing on the binding properties, the adhesiveness of lactobacilli obtained from solubilization of lyophilized powder or tablets was compared with the adhesion capacity of cultured micro-organisms. The results reported in Table 2 demonstrated that, immediately after solubilization, most strains of lactobacilli are in a metabolic condition that is unfavourable to adhesion. For all the strains, except Lact. salivarius FV2, Lact. gasseri MB 331 and MB 332, there was a great reduction in the attachment to the cell surface. A dramatic loss of adhesion occurred for Lact. brevis CD2, showing more than 90% reduction in the adhesion, while Lact. crispatus and Lact. gasseri (strains MB 333, MB 335 and MB 336) lost 50–70% of their adherence ability. The reduced adhesion of freeze-dried powder and tablet micro-organisms indicates that the lyophilization process affected the adhesion properties of some isolates, whereas pharmaceutical formulation and tablet production did not further influence the attachment capacity of the micro-organisms. The loss of adhesion was not associated with a reduction in lactobacilli viability measured in both preparations (Table 3). In fact, although a similar reduction in the adhesiveness (more than 60%) was observed for both Lact. gasseri MB 333 and MB 335, these strains showed a great difference in viability. It is possible to hypothesize that the lyophilization process could modify the conformation of surface bacterial adhesins, thereby limiting their function. When the tablets were cultured overnight in MRS broth, the adhesion capacity of the strains was restored to the level achieved with cultured lyophilized micro-organisms.

Table 2.  Influence of metabolic state on the adhesion of Lactobacillus strains to HeLa cells *
Lactobacillus strainMean no. of adherent bacteria/cell (s.d.)
Lyophilized powderCultured lyophilized powderTabletCultured tablet
  1. *See footnote to Table 1.

Lact. brevis CD2 2·3 (0·8)25·7 (6·4) 1·1 (0·6)26·0 (7·1)
Lact. salivarius FV2 1·7 (0·9) 2·1 (1·3) 1·3 (0·4) 1·9 (0·9)
Lact. salivarius FV3 1·1 (0·4) 1·3 (0·3) 0·7 (0·4) 1·4 (0·5)
Lact. crispatus FV4 1·2 (0·6) 2·6 (1·2) 1·3 (0·7) 1·9 (0·9)
Lact. gasseri MB 33123·2 (8·1)24·2 (6·5)21·9 (7·3)25·8 (9·3)
Lact. gasseri MB 332 7·9 (2·6) 8·9 (2·7) 6·2 (2·2) 7·5 (2·9)
Lact. gasseri MB 333 0·9 (0·5) 2·8 (1·3) 0·8 (0·5) 1·9 (0·6)
Lact. gasseri MB 33413·1 (3·7)20·5 (8·5)11·6 (4·9)19·7 (7·3)
Lact. gasseri MB 335 7·6 (2·0)19·9 (6·9) 6·4 (2·2)18·7 (8·2)
Lact. gasseri MB 336 3·0 (1·2) 6·8 (3·7) 3·2 (1·5) 7·5 (3·6)
Table 3.  Viability of Lactobacillus strains in lyophilized powder and tablet *
Lactobacillus strainViable count (log cfu mg−1)
Lyophilized powder (cells mg−1)Tablet (cells mg−1)
  1. *The number of viable lactobacilli was evaluated soon after preparation of lyophilized powder and tablets (t0) and immediately after solubilization for the adhesion assay (tA).

  2. †Mean ± s.d.

Lact. brevis CD210·21 ± 0·09 9·28 ± 0·04 8·72 ± 0·08 8·64 ± 0·04
Lact. salivarius FV210·31 ± 0·1010·07 ± 0·1310·01 ± 0·21 9·96 ± 0·06
Lact. salivarius FV310·44 ± 0·0910·09 ± 0·08 9·53 ± 0·06 9·36 ± 0·04
Lact. crispatus FV4 8·52 ± 0·06 7·78 ± 0·08 6·64 ± 0·08 6·34 ± 0·04
Lact. gasseri MB 331 9·76 ± 0·09 9·11 ± 0·20 8·57 ± 0·05 8·28 ± 0·07
Lact. gasseri MB 33210·72 ± 0·1110·67 ± 0·0810·56 ± 0·0810·51 ± 0·09
Lact. gasseri MB 33310·42 ± 0·09 9·60 ± 0·06 9·77 ± 0·09 9·04 ± 0·06
Lact. gasseri MB 334 9·80 ± 0·07 9·50 ± 0·04 9·60 ± 0·05 8·97 ± 0·07
Lact. gasseri MB 335 9·69 ± 0·04 9·43 ± 0·09 9·52 ± 0·11 9·32 ± 0·09
Lact. gasseri MB 33610·71 ± 0·1510·10 ± 0·10 9·59 ± 0·09 8·98 ± 0·07

Antimicrobial activity and H2O2 production

The strains of lactobacilli were assayed for their ability to produce inhibitory substances against the growth of G. vaginalis and C. albicans. Gardnerella vaginalis was inhibited by different Lactobacillus strains (Table 4) whereas no inhibition was found against C. albicans (data not shown). Culture supernatant fluids of some isolates of Lact. gasseri (strains MB 331, MB 332, MB 333 and MB 335) were able to inhibit the growth of G. vaginalis, showing the largest halo diameter.

Table 4.  Growth inhibition of Gardnerella vaginalis and H2O2 production by Lactobacillus strains
Lactobacillus strainDiameter of inhibition haloes (mm) H2O2 production
  1. –, Non-producer of H2O2; +, weak producer of H2O2;+ +, strong producer of H2O2.

Lact. brevis CD2
Lact. salivarius FV2 90+ +
Lact. salivarius FV3+ +
Lact. crispatus FV4+
Lact. gasseri MB 331100+
Lact. gasseri MB 332100+ +
Lact. gasseri MB 333100
Lact. gasseri MB 334 90+ +
Lact. gasseri MB 335100+ +
Lact. gasseri MB 336 80

The production of H2O2 by lactobacilli is shown in Table 4. The results obtained demonstrated that there was no direct correlation between H2O2 production and the activity of lactobacilli culture supernatant fluids against G. vaginalis. Indeed, Lact. salivarius FV3, a high H2O2 producer, did not inhibit the growth of G. vaginalis. On the contrary, Lact. gasseri MB 333 showed good inhibiting activity against G. vaginalis without demonstrable H2O2 production. The lactobacilli culture supernatant fluids were neutralized, demonstrating that the antimicrobial activity of the non-H2O2-producing strains was not related to the reduction in pH due to the production of lactic acid.

Coaggregation experiments

The coaggregation assay provides a measurement of the interaction between different micro-organisms. Coaggregation experiments showed that the capacity of lactobacilli to bind to a pathogen varies according to the single Lactobacillus strain and the pathogen involved (Table 5). Lactobacillus salivarius FV2 was able to coaggregate very efficiently (score 3) with both pathogens to such an extent that no isolated lactobacilli were observed. In contrast, Lact. salivarius FV3 showed a lower coaggregation activity with C. albicans (score 1) and G. vaginalis (score 2) with aggregates of small dimensions and a great number of non-adherent lactobacilli. Among the different strains of Lact. gasseri, the interaction with pathogens varied from none (strains MB 331 and MB 334) to a good coaggregation capacity (strains 333 and MB 335). As an example, Fig. 2 shows the microscopic appearance of the aggregates formed between Lact. salivarius FV2 and C. albicans (a) or G. vaginalis (b), which contrasts with the presence of isolated Lact. gasseri MB 331 and G. vaginalis cells (c).

Table 5.  Coaggregation between lactobacilli and vaginal pathogens *
Lactobacillus strainCoaggregation score for
Candida albicansGardnerella vaginalis
  1. *The score is based upon a scale described by Reid et al. (1990), from 0 for no aggregation to 4 for maximum aggregation.

Lact. brevis CD210
Lact. salivarius FV233
Lact. salivarius FV312
Lact. crispatus FV411
Lact. gasseri MB 33100
Lact. gasseri MB 33222
Lact. gasseri MB 33333
Lact. gasseri MB 33400
Lact. gasseri MB 33523
Lact. gasseri MB 33612
Figure 2.

Microscopic observations of coaggregation between Lactobacillus salivarius ssp. salicinius FV2 and (a) Candida albicans or (b) Gardnerella vaginalis and (c) the lack of coaggregation between Lact. brevis CD2 and G. vaginalis

Formulation of vaginal tablet and adhesion interference

On the basis of the adhesive properties, H2O2 production, pathogen inhibition and coaggregation with pathogens reported above and the appropriate physical properties previously described (Maggi et al. 2000), the three strains Lact. brevis CD2, Lact. salivarius FV2 and Lact. gasseri MB 335 were chosen for the final formulation of vaginal tablets that were prepared with equal amounts of each Lactobacillus strain. Despite improved lactobacilli adhesion to the cell membrane observed with DEAE dextran, this compound was not included in the formulation of the vaginal tablet since it also enhanced the adhesiveness of vaginal pathogens (data not shown).

All the characteristics previously assayed for the selection of the strains were re-examined in the tablet formulated with the lactobacilli mixture. After overnight culture of the tablet in MRS broth, the mixture of lactobacilli showed good adhesion properties and maintained a high capacity to inhibit the growth of G. vaginalis, produce H2O2 and coaggregate with pathogens. The total number of viable lactobacilli present in the tablet remained stable for a period of 18 months (data not shown).

The capacity of lactobacilli from cultured vaginal tablets to interfere with the attachment to epithelial cells of the adhesive vaginal pathogens C. albicans and G. vaginalis was evaluated. Two proposed protective characteristics of lactobacilli are the abilities to competitively exclude adhesion of pathogens from cells and displace adherent pathogens (Reid et al. 1987). The effect of lactobacilli on the adhesion of C. albicans and G. vaginalis is reported in Table 6. The results demonstrate that, when epithelial cells are coated with lactobacilli, there is a reduction of more than 50% in the adhesion of pathogens. The adhesion interference was greater in the displacement assay, demonstrating that lactobacilli were able to remove more than 60% of C. albicans and G. vaginalis previously attached to epithelial cells.

Table 6.  Effect of lactobacilli from cultured vaginal tablets containing the consortium of selected strains on the adhesion of Candida albicans and Gardnerella vaginalis to HeLa cells *
Strain% Cells with adherent
Mean no. of adherent
Mean no. of adherent
% Reduction
  1. *See footnote to Table 1.

Lactobacillus tablet65·117·011·0 (3·4)
C. albicans36·6 2·96 1·08 (0·29)
Lactobacillus + C. albicans20·3 2·57 0·51 (0·11)52·8
G. vaginalis72·522·516·1 (2·8)
Lactobacillus + G. vaginalis50·413·5 6·8 (1·2)57·7
Lactobacillus tablet53·8 9·0 4·85 (1·4)
C. albicans35·7 2·8 1·0 (0·21)
C. albicans + Lactobacillus15·1 2·4 0·36 (0·09)64·0
G. vaginalis74·022·016·2 (2·6)
G. vaginalis + Lactobacillus41·815·2 6·35 (1·1)60·8


The characteristics needed for a Lactobacillus strain to serve effectively as a probiotic include avid adherence to epithelial cells, interference with the adhesion of pathogens and production of H2O2 and other molecules capable of inhibiting the growth of pathogens. Reid et al. (1987), analysing different characteristics (bacterial size, adherence capability, competitive exclusion and inhibition of pathogenic growth) that could be involved in the inhibition of uropathogens by lactobacilli, concluded that adherence is an essential factor for the antimicrobial activity of lactobacilli. Bacterial adherence has been suggested to be the result of two essentially different mechanisms, specific and non-specific binding (Piette and Idziak 1992). Non-specific binding involves electrostatic or hydrophobic interactions of lower affinity than in specific binding. Piette and Idziak (1992) reported that cell surface charge and hydrophobicity influence the strength of adhesion. We studied the role of electrostatic interactions in lactobacilli adherence by means of DEAE dextran. The results obtained demonstrated that the reduction in surface negative electric charge promotes the attachment of several Lactobacillus strains to the cell membrane, but the involvement of electrostatic interactions varies according to the different strains.

The results reported in this paper demonstrate that lyophilization reduces the adhesion capacity and, therefore, the colonization ability of some isolates and this factor must be taken into account in the preparation of probiotic products. In this respect, understanding the role of electrostatic interactions in the binding of lactobacilli to human epithelial cells is also of major significance. The addition, to the pharmaceutical formulation, of compounds capable of reducing the surface negative electric charge would greatly improve the colonization capacity. However, the enhancement of G. vaginalis and C. albicans adhesion to the cell surface observed in the presence of DEAE dextran indicates that any excipients used for tablet manufacture must also be tested on the adhesion capacity of potential pathogens of the district under investigation. Moreover, the development of a tablet formulation containing a high number of viable lactobacilli is of major importance in order to obtain a high rate of multiplication, thus quickly restoring a high adhesion capacity. The specific production process adopted by us allowed the three selected strains of lactobacilli to remain alive for a period of 18 months.

Available commercial products containing lactobacilli for vaginal administration generally include only one Lactobacillus species (usually Lact. acidophilus or Lact. casei), whose properties relevant to vaginal use have never been extensively characterized. Since the number of different species of lactobacilli colonizing the normal vagina varies from zero to four, and the species and their combinations change over time in the healthy vagina (Redondo-López et al. 1990), we decided on a formulation of effervescent vaginal tablets containing three different Lactobacillus species with different relevant characteristics. This dosage form ensures a rapid and complete distribution of viable micro-organisms on the vaginal mucosa (Maggi et al. 1994). The selected Lact. salivarius FV2 and Lact. gasseri MB 335 strains are strong H2O2 producers and Lact. gasseri MB 335 is also strongly adherent to epithelial cells. It has been demonstrated that most women with bacterial vaginosis have a vaginal microflora depleted of Lactobacillus strains producing H2O2 (Hawes et al. 1996), these strains exerting a significant bactericidal activity towards G. vaginalis (Klebanoff et al. 1991), a micro-organism isolated in almost all cases of bacterial vaginosis. Both Lact. salivarius FV2 and Lact. gasseri MB 335 strains were also able to coaggregate very efficiently with G. vaginalis and C. albicans. The coaggregation could be an important factor in establishing and maintaining a healthy urogenital flora because of the production of a microenvironment around the pathogen where the concentration of inhibiting substances produced by lactobacilli is exacerbated.

Lactobacillus brevis CD2, although not producing H2O2, was chosen for its strong adherence capacity and morphology. It has been suggested that a large adherent Lactobacillus strain could exert a better competitive exclusion of pathogens than a small one by masking specific receptor sites for pathogenic micro-organisms on the cell surface by steric hindrance. Furthermore, the CD2 strain of Lact. brevis has recently been shown (Di Marzio et al. 2001) to produce high levels of the enzyme arginine deiminase, which was able to down-regulate polyamine synthesis in Jurkat cells. Arginine deiminase catalyses the irreversible conversion of arginine to citrulline and ammonia, thereby decreasing the availability of medium arginine and consequently of ornithine, the starting material for the polyamine biosynthetic pathway. Polyamines are commonly found in elevated concentrations in vaginal discharges of women with bacterial vaginosis (Chen et al. 1979; Chen et al. 1982) and contribute to the elevated pH of the vaginal microenvironment and also to the clinical symptoms of bacterial vaginosis, in particular the ‘fishy’ odour that is characteristic of vaginal discharges from affected women (Chen et al. 1979; Clay 1982). Therefore, the CD2 strain of Lact. brevis, besides its ‘classical’ good probiotic characteristics, appears to possess metabolic properties able to inhibit opportunistic pathogens through the deprivation of arginine, an important source of carbon, nitrogen and energy for bacteria, and to reduce biochemical markers (polyamine synthesis and elevated pH) observed in bacterial vaginosis. In view of its high arginine deiminase activity, this Lactobacillus strain could represent an important step towards improving the efficacy of bacteriotherapy for bacterial vaginosis.

In conclusion, we designed a pharmaceutical formulation of a probiotic preparation containing a mixture of three strains of lactobacilli with antimicrobial and adhesive properties as well as biochemical characteristics potentially effective in treating the clinical signs and symptoms of vaginal infections. Clinical studies to evaluate the effect of these tablets in women with recurrent bacterial vaginosis are in progress.


This research was supported by grants from MIUR (Italy) and VSL Pharmaceuticals, Fort Lauderdale, FL, USA. Lact. brevis CD2 is proprietary to VSL Pharmaceuticals.