Antimicrobial properties of Lactobacillus cell‐free supernatants against multidrug‐resistant urogenital pathogens

Abstract The healthy vaginal microbiota is dominated by Lactobacillus spp., which provide an important critical line of defense against pathogens, as well as giving beneficial effects to the host. We characterized L. gasseri 1A‐TV, L. fermentum 18A‐TV, and L. crispatus 35A‐TV, from the vaginal microbiota of healthy premenopausal women, for their potential probiotic activities. The antimicrobial effects of the 3 strains and their combination against clinical urogenital bacteria were evaluated together with the activities of their metabolites produced by cell‐free supernatants (CFSs). Their beneficial properties in terms of ability to interfere with vaginal pathogens (co‐aggregation, adhesion to HeLa cells, biofilm formation) and antimicrobial activity mediated by CFSs were assessed against multidrug urogenital pathogens (S. agalactiae, E. coli, KPC‐producing K. pneumoniae, S. aureus, E. faecium VRE, E. faecalis, P. aeruginosa, P. mirabilis, P. vulgaris, C. albicans, C. glabrata). The Lactobacilli tested exhibited an extraordinary ability to interfere and co‐aggregate with urogenital pathogens, except for Candida spp., as well as to adhere to HeLa cells and to produce biofilm in the Lactobacillus combination. Lactobacillus CFSs and their combination revealed a strong bactericidal effect on the multidrug resistant indicator strains tested, except for E. faecium and E. faecalis. The antimicrobial activity was maintained after heat treatment but decreased after enzymatic treatment. All Lactobacilli showed lactic dehydrogenase activity and production of D‐ and L‐lactic acid isomers on Lactobacillus CFSs, while only 1A‐TV and 35A‐TV released hydrogen peroxide and carried helveticin J and acidocin A bacteriocins. These results suggest that they can be employed as a new vaginal probiotic formulation and bio‐therapeutic preparation against urogenital infections. Further, in vivo studies are needed to evaluate human health benefits in clinical situations.


| INTRODUC TI ON
Lactobacilli are important members of the human gastrointestinal, oral, and vaginal microbiota and are gaining great interest for their healthpromoting effects in the host both on direct interactions between cells and indirectly through their released metabolites, thus making them suitable to be used as probiotic strains (Reid et al., 2011). Over the last few years, the search for probiotic strains possessing innovative functional characteristics and formulations has been evolving and is an attractive goal in therapeutic strategies to restore the natural microbiota. Antibiotic treatment is the main approach used to fight bacterial infections (Aslam et al., 2018), but excessive and inappropriate use in both hospital and community settings has been one of the main factors of the onset of antibiotic resistance, and urogenital tract infections (UGTIs) are the most common infections in which many multidrugresistant (MDR) pathogenic strains are recorded due to the abuse of antibiotic therapy (Matulay et al., 2016).
Lactobacilli dominate the healthy vaginal microbiota and are considered gatekeepers of this ecosystem, maintaining a healthy state and impeding the growth of pathogens (Bautista et al., 2016;Martin, 2012;Ravel et al., 2011). Recent studies have focused on the vaginal microbiome in healthy reproductive-aged women by 16S rRNA gene sequencing showed at least 5 community state types (CSTs), in which four were dominated by L. crispatus (CST-I), L. gasseri (CST-II), L. iners (CST-III), L.
Perturbations of this highly regulated ecosystem occur during urogenital tract infections (UGTIs), as well as urinary tract infections (UTIs), bacterial vaginosis (BV), and during antimicrobial therapy, resulting in an even greater aberration of the microbiota and, eventually, in the extension of an infectious state (Donders et al., 2017;Eryilmaz et al., 2018;Matulay et al., 2016). Restoration of vaginal homeostasis, driven by Lactobacilli, may be accomplished through numerous mechanisms: (i) "competitive exclusion," the first critical line of defense against local pathogens, which is the ability of bacteria to adhere to vaginal epithelial cells competing for nutrients and adhesion receptors (Liu et al., 2008), (ii) "co-aggregation," the assembly of microbial communities into distinct, interlinked structures (Pino et al., 2019); in addition, (iii) an intense production of antimicrobial compounds such as lactic acid, hydrogen peroxide (H 2 O 2 ), bacteriocin-like substances, and biosurfactants may inhibit pathogen growth (Petrova et al., 2015).
In vitro and in vivo studies have indicated the use of probiotics as an alternative approach for restoring healthy vaginal microbiota by interfering with potential pathogens. Although the use of live microorganisms is currently widely employed, safety issues remain a matter of debate, mainly for vulnerable subjects (Borges et al., 2014;Ravel et al., 2011;Reid et al., 2011). To overcome these issues, in the last decade, the use of non-live microorganisms such as heat-killed probiotics, microbial extracts, and cell-free supernatants has been growing in interest for their applications in therapeutic strategies also considering that they can confer relevant beneficial effects (Piqué et al., 2019).
In this study, we characterized three vaginal Lactobacilli, L. gasseri 1A-TV, L. fermentum 18A-TV, and L. crispatus 35A-TV from healthy vaginal microbiota for their probiotic properties mainly focusing on their antimicrobial activity against the most common MDR UGTI pathogens Al-Zahrani et al., 2019), considering also both adhesive properties and inhibitory substances released in their cell-free supernatants (CFS). Generating Pouch Systems (BD). All Lactobacilli were taxonomically identified at the species level by amplification and sequencing of the tuf and 16S rRNA genes for accurate identification. Genomic DNA was extracted from overnight cultures of isolates in 5 ml of MRS and the tuf and16S rRNA genes were amplified. All PCR products obtained were purified using the QIAquick PCR gel extraction kit (Qiagen) and sequenced (Hütt et al., 2016;Marchisio et al., 2015;Ventura et al., 2003). Sequence analyses were performed using Gapped BLAST (Altschul et al., 1997

| In vitro safety assessment of Lactobacillus strains
i. Antibiotic susceptibility testing and detection of hemolytic activity.
The antibiotic susceptibility profiles of the three Lactobacilli were determined by the Kirby-Bauer diffusion and E-test methods on MRS agar at 37°C for 48 h under anaerobic conditions (Charteris et al., 2007). The following antibiotics were tested: penicillin, ampicillin, amoxicillin-clavulanic acid, vancomycin, gentamicin, streptomycin, tetracycline, chloramphenicol, erythromycin, clindamycin, trimethoprim-sulfamethoxazole, rifampicin, ciprofloxacin, levofloxacin, and metronidazole. The antimicrobial susceptibility profiles were analyzed according to the interpretative standard of the European Union Commission recommendations for probiotic safety (Authority EFS, 2012).
ii. The hemolytic activity of Lactobacilli was visually verified on Columbia agar base supplemented with 5% sheep and horse blood (Oxoid) after 24 h and 48 h of incubation under anaerobic conditions at 37°C (Maragkoudakis et al., 2006). Streptococcus pyogenes, strain ATCC 19615, was used as a positive control. Both experiments mentioned above were performed in triplicate.

| Determination of antagonistic activity
The MDR indicator strains, S. agalactiae, E. faecalis VRE, E. faecium, S. aureus, P. aeruginosa, P. mirabilis, P. vulgaris, E. coli, KPC-producing K. pneumoniae, C. albicans, and C. glabrata, were used for detecting the antimicrobial activity of Lactobacilli. The inhibitory activity of vaginal strains was determined by the deferred antagonism test and quantified by the agar spot test with some modifications (Santagati et al., 2012;Siroli et al., 2017). In addition, for the evaluation of Lactobacillus combination, L. gasseri 1A-TV, L. fermentum 18A-TV, and L. crispatus 35A-TV were grown in MRS broth for 48 h at 37°C under anaerobic conditions, using the GasPakEZ Gas Generating Pouch System (BD, New Jersey, USA) and approximately 2 × 108 CFU/ml of each Lactobacillus culture in a 1:1:1 ratio were used. Briefly, for the deferred antagonism assay, the test strain was inoculated diametrically across MRS agar with the addition of 0. the plates were incubated for 18 h at 37°C to examine the interference zones with the indicator. Lactobacillus isolates that inhibited the growth of an indicator strain were considered inhibitory for that species (Maragkoudakis et al., 2006). For the agar spot test, the Lactobacillus cultures were spotted (5 µl) on the surface of MRS agar (1.2%) (20 ml) and incubated anaerobically for 48 h at 37°C.

| Auto-aggregation and co-aggregation assays
Auto-aggregation assays were performed according to Kos et al. (Kos et al., 2003). The auto-aggregation percentage is expressed as  in 20 random microscopic fields to obtain Lactobacillus counts and averages. The adhesion indexes (ADI; the number of bacteria/100

| In vitro adhesion test
HeLa cells) were expressed as strong adhesion: ADI >2500; good adhesion: good adhesion: ADI between 2500 and 500, weak adhesion between 500 and 100, no adhesion, ADI <100. Bacterial adhesion to the HeLa cell layer was also evaluated by viable counts. After incubation, supernatants were discarded and non-adherent bacteria were removed by washing each well twice with PBS and after the detachment by 1 ml of PBS with 0.1% Triton X-100 (Sigma-Aldrich, USA). The viable counts of adherent lactobacilli were evaluated by CFU/ml on MRS agar plates after incubation anaerobically for 48 h at 37°C (Santagati et al., 2012).

| Biofilm formation assay
Biofilm production was tested in MRS broth. as a positive control strain as it was a good biofilm producer (Lebeer et al., 2007), and MRS medium without inoculum was included as a negative control. As a selection criterion for biofilm formation, a cutoff OD (ODc) for the test was defined as three standard deviations above the mean OD of the negative control. The strains were considered non-biofilm producers (OD_ODc); weak biofilm producers (ODc<OD_2_ODc); moderate biofilm producers (2_ODc<OD_4_ ODc); strong biofilm producers (4_ODc<OD_8_ODc); and very strong biofilm producers (8_ODc<OD). These experiments were performed in triplicate.

| Assessment of in vitro antimicrobial activity of Lactobacillus cell-free supernatants
Cell-free supernatants (CFSs) of L. gasseri 1A-TV, L. fermentum 18A-TV, and L. crispatus 35A-TV and the CFS of the Lactobacilli combination were prepared as previously reported (Parolin et al., 2015). Each Lactobacillus culture was centrifuged at 7000 × g for 30 min at 4°C, and their supernatants were filtered through a 0.2 μm membrane, and pH values were measured by a pH meter (pH50+DHS Bench pH meter). For the CFS combination, each Lactobacillus culture at 2 × 108 CFU/ml, after the filtration step, was mixed in a 1: 1:1 ratio.  The bactericidal activity was defined as a reduction of at least 99.9% (≥3 log 10 ) (NCCLS, 1999). This experiment was repeated in triplicate.

| Evaluation of the antimicrobial activity of CFSs after pH, heat, catalase, and proteolytic enzymatic treatment
The effects of heat treatment, catalase, and proteolytic enzymatic treatments were evaluated for all CFSs.  (Oliveira et al., 2017). After these treatments, the antibacterial activity of the CFSs was determined by antagonism experiments in 96-well plates and expressed as total (+++), good (++), partial (+), and no inhibition (-). The effects of pH were tested at pH 5.5, 6.5, and 7.5 adjusted by 10 N NaOH, and untreated cell-free supernatants were used as controls. The antagonism experiments were performed in a sterile 96-well plate (Corning ® Incorporated Life Sciences) using the indicator strains at 3 × 105 CFU/ml as described above. After incubation for 6 and 24 h at 37°C, the results were estimated by the growth rates of the indicator strains measured by a turbidimetric method with Microplate Reader (BioTek Synergy™ H1) system using OD 600 for bacterial strains and OD 530 for Candida spp. (Yang et al., 2018). All experiments were repeated three times.

| Determination of hydrogen peroxide production, lactic dehydrogenase activity, L-and D-lactic acid production, and the presence of bacteriocin genes
The production of H 2 O 2 was tested by the Eschenbach method (Eschenbach et al., 1989) using the scale previously reported by Parolin et al. (Parolin et al., 2015). All strains were scored as low  (Kasai et al., 2019). The enzyme assay was performed at 30°C, and 1 U of the enzyme was defined as the amount of enzyme that catalyzes the degradation of 1 µmol of NADH per minute (Sung et al., 2004).
The production of D-and L-lactic acid produced by Lactobacillus  Table A1. PCR was performed as previously published (Santagati et al., 2012).

| Statistical analysis
Statistical analyses were performed using GraphPad Prism 6 software (GraphPad Software Inc.), and results were expressed as mean ±standard deviation (SD) of 3 independent experiments. For the coaggregation assays, ANOVA with Fisher's least significant difference (LSD) test was used to determine significant differences (p < 0.05).

| Evaluation of Lactobacillus antagonistic activity against multidrug-resistant clinical isolates
The antagonistic activity of L. gasseri 1A-TV, L. fermentum 18A-TV, Lactobacilli antagonized C. albicans, showing inhibition zones between 6 and 10 mm (+ + +), and exerted a partial inhibition versus C. glabrata, with inhibition zones between 1 and 3 mm. They also showed good inhibition versus P. aeruginosa, P. mirabilis, and P. vulgaris with diameters between 3 and 6 mm (++). The same results were obtained with the combination (1:1:1 ratio) of the three Lactobacilli (Table 2).
on sheep and horse blood agar. The in vitro safety assessment of vaginal Lactobacilli isolates is given in Table A2.

| Aggregation assays and biofilm formation
Aggregation properties were assayed with the auto-aggregation and co-aggregation tests measuring two characteristics of the strains.
Auto-aggregation can be mediated by intra-species cellular promoting factors and cell-wall hydrophobicity, while co-aggregation is the ability to achieve an adequate mass by co-aggregating other bacterial species, however, the ability of a probiotic to aggregate is a desirable property.
The auto-aggregation rates of L. gasseri 1A-TV, L. fermentum 18A-   Despite the co-aggregation percentage of all bacterial strains being higher than self-aggregation percentages, significant coaggregation (p < 0.05) was found only for S. aureus, P. aeruginosa,

F I G U R E 1 Co-aggregation ability of
and Proteus spp. with 1A-TV, S. agalactiae, E.coli, S. aureus, E. faecium, and P. aeruginosa with 18 A-TV and 35A-TV; in addition, 35A-TV also showed significant co-aggregation with K. pneumoniae (Figure 1).
Regarding the biofilm production, we found different levels: weak for L. gasseri 1A-TV, moderate for L. fermentum 18A-TV while L. crispatus 35A-TV was not a biofilm producer; however, the Lactobacillus combination stood out as being a strong biofilm producer.

| Adhesion test on HeLa cells
L. gasseri 1A-TV, L. fermentum 18A-TV, and L. crispatus 35A-TV were tested for their capability to adhere to HeLa cells. After being extensively washed with PBS, a significant proportion of cells from all bacterial strains remained attached to the HeLa monolayer displaying a strong adhesive phenotype, coinciding with an adhesion index (ADI) greater than 2500, as shown in Figure 2 (a, b). This adhesion in L. crispatus 35A-TV showed an extraordinary ADI of 70000. The

| In vitro antimicrobial activity of Lactobacilli CFSs and their sensitivity to pH, heat, catalase, and proteolytic enzymatic treatment
Cell-free supernatants of L. gasseri 1A-TV, L. fermentum 18A-TV, and L. crispatus 35A-TV, at pH 4.2, 4.3, and 4.8, respectively, were assayed for their ability to inhibit the pathogens by time-killing tests ( Figure 3 and Figure A1).   In Table 2 The pH-dependent effects on antimicrobial activity of CFSs were tested at pH 5.5, 6.5, and 7.5. by measurements of the growth rates (OD) of the indicator strains. All CFS Lactobacilli and their formulation at pH 5.5 maintained their activity up to 6 h and weakly lost their efficiency at 24 h compared to the untreated pH, while the antagonistic activity of CFS at pH 6.5 and 7.5 was lost after 6 h despite the growth of the indicator curve showed a slight decrease in slope compared to controls ( Figure A2).

| Determination of hydrogen peroxide production, lactic dehydrogenase activity, and bacteriocin-encoding genes
L. gasseri 1A-TV produced a higher quantity of hydrogen peroxide with respect to L. crispatus 35A-TV, while L. fermentum 18A-TV did not release this metabolite ( Table 2).
The lactic dehydrogenase activity was evaluated using cell lysates, in particular, L. crispatus 35A-TV had a specific activity of 56.17 U mg/L, higher than that observed in L. gasseri 1A-TV and L. fermentum 18A-TV, which were, respectively, 25.57 U mg/L and 28.27 U mg/L (Table 2).
The detection of bacteriocin-encoding genes revealed helveticin J only in L. gasseri and acidocin A in L. gasseri and L. crispatus. L.
fermentum, despite showing activity against pathogens, was negative for all genes tested (Table 2).

| DISCUSS ION
Several studies have reported beneficial effects exerted by probiotics, and it has been well demonstrated that functional properties are strain-dependent (Borges et al., 2014). In this study, we characterized three Lactobacilli, L. gasseri 1A-TV, L. fermentum 18A-TV, and L. crispatus 35A-TV isolated from the vaginal microbiota, with the activities of their metabolites produced by CFSs for their beneficial features addressed mainly to their antimicrobial activity against multidrug-resistant clinical isolates.
In accordance with the objectives of our study, the selected Lactobacilli were tested in vitro for surface properties to determine their capability to colonize the human vagina. In vitro experiments showed their ability to adhere to HeLa cells, and this is also related to their predisposition to self-aggregate. As is well known, adhesion and auto-aggregation represent the determining factors for the initial development of biofilm, which is a strategy of some organisms to persist in harsh environments promoting microbial resistance to antimicrobial agents, the immune system, and stress conditions (Leccese Terraf et al., 2016). In this regard, our Lactobacilli possessed strong biofilm formation capacity when tested in combination; however, they were poor producers when tested alone. These data make us hypothesize a synergistic interspecific interaction between our Lactobacilli to optimize their living conditions. Biofilm formation is a phenomenon that can promote mucosal colonization and masking epithelial cell receptors, can exert a protective role by interfering with the growth and adhesion of pathogens (Leccese Terraf et al.,

2016).
Another mechanism that promotes an exclusion/competition behavior is the ability of beneficial bacteria to co-aggregate with pathogens (Santos et al., 2016). In this regard, our Lactobacillus strains showed a significant capability to co-aggregate with S. agalactiae, E.
faecalis, P. aeruginosa, P vulgaris, and P. mirabilis. This is an important contributing factor to create a microenvironment where pathogens can be exposed to higher concentrations of inhibitory substances or metabolites such as organic acids (e.g., lactic acid) and hydrogen peroxide mainly produced by Lactobacilli strains as the dominant bacterial population in the vaginal ecosystem (Verdenelli et al., 2014).
In this study, we found that cell-free supernatants released from three Lactobacilli as single entities, and their combination, exhibited an antagonistic effect against multidrug-resistant clinical isolates including S. agalactiae, E. coli, KPC-producing K. pneumoniae, S. aureus, E. faecium VRE, E. faecalis, P. aeruginosa, P. mirabilis, and P. vulgaris.
Conversely, the anti-candida activity of the three Lactobacilli showed different behavior with the two approaches: agar diffusion and using cell-free supernatants, which had no growth-inhibitory activity and could maintain the candidal growth almost at the same concentration as the initial inoculum compared to the control (CFS-free).
These conflicting results could be explained by the physical state of the media; the concentration of antimicrobial substances released into the solid and liquid media, and by the environment where the substances exert their effects. Scorzoni et al. also reported that the microdilution test is more sensitive to agar diffusion in the evaluation of anti-candida activity highlighting the need to apply different methods to evaluate in vitro antimicrobial effects of Lactobacilli (Scorzoni et al., 2007).
Notably, the CFS combination maintained the same antagonistic profile of each strain, excluding a possible interference between them.
The activity of Lactobacillus CFSs after the heat and enzymatic treatments was reduced in some cases compared with untreated CFSs hypothesizing the presence of thermostable and thermosensitive substances such as bacteriocins in the supernatants, while the neutralization treatment at pH 6.5 and 7.5 canceled antagonistic effects. These data suggested that the acid environment and antimicrobial metabolites released by our strains such as bacteriocins had synergetic action against the growth of pathogens tested, showing a better antagonistic activity. Further, several reports suggested that the pH-induced alterations of net charge might facilitate the translocation of some bacteriocin molecules through the cell wall (Oliveira et al., 2017) and that an acid environment could interfere with the production and bactericidal activity of several bacteriocins (Yang et al., 2018). pH and lactic acid levels display a strong inverse correlation demonstrating that lactic acid is the main acidifier of the human vagina, increasing its production under hypoxic conditions (Tachedjian et al., 2017), which displays antimicrobial and anti-inflammatory properties. In this context, all three isolated properties of the vaginal microbiota, representing an important nonspecific antimicrobial defense mechanism due to a highly toxic state (Kullisaar et al., 2002;Mijac et al., 2006). Additionally, bacteriocins are believed to contribute to the competitiveness between strains by acting against pathogenic strains; therefore, the production of bacteriocins represents an important antimicrobial factor (Soltani et al., 2020). Among our strains, L. gasseri 1A-TV and L. crispatus 35A-TV are producers of helveticin J and acidocin A, which is a small thermostable peptide with maximum production at pH 5, exerting antagonistic activity versus several bacterial genera, including Lactococcus, Pediococcus Staphylococcus, Enterococcus, Streptococcus, Listeria, Clostridium, and Bacillus (Kanatani et al., 1995). The stability at a high temperature of acidocin A and its bacterial targets suggested its decisive role in the antimicrobial activity exerted by the supernatants of L. gasseri 1A-TV and L. crispatus 35A-TV against the indicators tested, also considering that L. crispatus is considered a major determinant in the stability of the normal vaginal microbiota in women of reproductive age (Miller et al., 2016) Our study strengthens the concept of using probiotic Lactobacillus Future work will also characterize the probiotic potential of these bacteria in the vaginal tract through in vivo studies.

E TH I C S S TATEM ENT
Ethical approval is not required. Written informed consent was obtained from all participants.

DATA AVA I L A B I L I T Y S TAT E M E N T
All data generated or analyzed during this study are included in this published article.  P  AMP  VA  SXT  RD  MTZ  CN  S  TE  C  E  AMC  LEV  CIP  DA 1A-TV L. gasseri