Bacterial stain and culture conditions. Lactobacillus crispatus CIP104459 was obtained from the Institut Pasteur Collection (CRBIP-Microorganism biobank catalogue, Paris, France). It was isolated in 1955 from a non-menopausal woman vaginal swab in the “La Croix-Rousse” maternity center (Lyon, France). In our laboratory, this strain was stored at -140°C in a cryofreezer (Thermo Fisher Scientific, Waltham, Massachusetts, USA). The draft genome sequence of L. crispatus CIP 104459 was recently determined and deposited in the DDBJ/ENA/GenBank under accession number VOMA0000000028. This bacterium was grown anaerobically in De Man, Rogosa and Sharpe (MRS) medium (VWR, Fontenay-sous-Bois, France) at 37°C under static conditions into conical 15 ml tubes (Corning®, Thermo Fisher Scientific, Waltham, Massachusetts, USA) filled with medium to maximal capacity. Culture stocks in glycerol / MRS 30% (v/v) were made and stored at −80°C before use. Pre-cultures were prepared anaerobically in MRS at 37°C for 48 h to reach the stationary growth phase. The density of the bacterial suspensions was determined by absorbance at 600 nm using a spectrophotometer (ThermoSpectronics, Cambridge, UK). Absence of contamination was controlled by plating onto MRS agar Petri dishes (VWR, Fontenay-sous-Bois, France). Growth of L. crispatus CIP104459 was monitored over 48 h
For monitoring of the growth kinetic, bacteria were layered in microplates, at an initial OD600nm = 0.08 and incubated in anoxic conditions over 48 h at 37°C with shaking (360 rpm) for ten seconds before each measure point using a multimode microplate reader (Tecan Group Ltd., Männedorf, Switzerland). Growth curves were obtained by automatic measurement every 30 min. Generation time and lag phase were calculated from the software of the microplate reader.
Tested molecule. 17β-estradiol (Sigma-Aldrich, Saint-Quentin-Fallavier, France) being poorly soluble in aqueous media, a stock solution was made by dilution in ethanol 100%. For bacterial treatment, this solution was dissolved in MRS so that the final concentration of ethanol in MRS was always kept at 0.1% v/v. Controls were performed using the same percentage of ethanol in MRS. Preliminary studies allowed to verify the absence of effect of ethanol 0.1% on L. crispatus CIP104459.
Evaluation of bacterial membrane fluidity. L. crispatus CIP104459 was grown in the absence or presence of 17β-estradiol over 18 h from the onset to the end of the experiment. After incubation, bacteria were collected by centrifugation (7,500 ×g, 10 min) and washed two times at room temperature in 10 mM MgSO4. The pellets were resuspended in the same solution and the OD600nm was adjusted to 0.1. In each 1mL aliquot of the bacterial suspension, 1 µL of 1,6-diphenyl-1,3,5-hexatriene (DPH) 4 mM in tetrahydrofuran (Sigma-Aldrich, Saint-Quentin-Fallavier, France) was added. Aliquots were then incubated in the dark for 30 min at 37 °C to allow incorporation of the probe into the bacterial membrane. Fluorescence polarization was measured using a Spark 20 M multimode microplate reader, equipped with an active temperature regulation system (Te-CoolTM, Tecan Group Ltd., Männedorf, Switzerland). Excitation and emission wavelengths were set at 365 and 425 nm, respectively. Each measure was realized in triplicate. Membrane anisotropy (r) was calculated according to Lakowicz52. Data were analyzed using the SparkControlTM software 2.1 (Tecan Group Ltd., Männedorf, Switzerland). The relation between fluorescence polarization and membrane fluidity is inverse. When anisotropy values increase that is corresponding to a decrease of membrane fluidity and vice versa. In a second series of experiments bacteria were grown for 6, 18 or 24 h in normal MRS medium collected and rinsed as previously describes and exposed to 17β-estradiol after incorporation of the fluorescent probe. Evolution of fluorescence polarization was measured over 3 h using the same equipment.
Determination of bacterial surface polarity. The surface polarity and Lewis acid-base properties of L. cristapus CIP104459 exposed or not to 17β-estradiol were studied using the microbial adhesion to solvents (MATS) technic53. Bacteria were grown in MRS and harvested by centrifugation (7,500 ×g, 10 min) after 18 h of incubation. The pellets were washed twice in phosphate buffer saline (PBS) (Lonza™, Thermo Fisher Scientific, Waltham, Massachusetts, USA) to remove traces of culture medium. Two solvent couples: chloroform/hexadecane and ethyl acetate/n‐decane were employed. For each condition, 1.2 mL of bacterial suspension at OD400nm = 0.8 was mixed for 60 s with 0.2 mL of each solvent. After incubation for 15 min and separation of the aqueous and organic phases, the OD400nm of the aqueous phase was measured. The percentage of bacteria accumulating in each organic compartment (solvent phase) was calculated using the following equation with [AO] = OD400nm of the aqueous phase without solvent and [A] = OD400nm of the aqueous phase after exposure to the solvent:
% solvent affinity = (1-A) / A0 × 100
All experiments were conducted in triplicate or more.
Investigation of bacterial aggregation and morphology. Formation of aggregates by L. crispatus CIP104459 was studied by the sedimentation technique described by Vandevoorde et al.30. Briefly, cultures of L. crispatus CIP104459 grown in MRS for 18 h under anoxic static conditions at 37°C and in the absence or presence of 17β-estradiol, were harvested by centrifugation (7,500 xg, 10 min, room temperature), washed twice in PBS, and resuspended in 10 mL of the same medium. After 1.5 min of agitation (vortex) which defined the T=0 of the experiment, the variation of OD600nm of the suspension was monitored over 30 min using a spectrophotometer (Thermo Fisher Scientific, Waltham, Massachusetts, USA). The percentage of auto‐aggregation after 30 min was calculated by the following equation:
% auto-aggregation = ([(OD0min - OD30min)/OD0min])/0D0min × 100
Where OD0min is the initial OD600nm at T = 0 and OD30min is the final OD600nm after 30 min.
The potential auto-aggregation and structure of L. crispatus CIP104459 was also studied by flow cytometry using a CytoFlex S flow cytometer (Beckman coulter Life science, Indianapolis, USA) and the CytExpert software. After 18 h of culture in the absence or presence of 17β-estradiol, the bacterial suspension was harvested by centrifugation as previously described and resuspended in PBS. Bacteria were not stained and aliquots (200 µL) were distributed in 96 wells microplates (Thermo Fisher Scientific, Waltham, Massachusetts, USA). After incubation in static condition for 30 min the samples were submitted to flow cytometry analyses. A minimum of 10,000 events at OD488 ± 4 nm (SSC channel) and OD525 ± 20 nm (FSC channel) were recorded at a flow rate of 10 µL.min-1 in each condition. Data were analyzed using the Cytexpert software. The percentage of the population organized as aggregates was corresponding to the fraction of events appearing in the Q1-UR quarter of the graph (red zone). The size is given by the FSC-A (horizontal) axis and the percentage of small bacteria appeared in the Q1-UL and Q1-LL zones. The surface granulometry is plotted in the SSC-A (vertical) axis. The increase of complexity appears in the Q1-UL and Q1-UR areas of the graph. Measures were realized in triplicate.
The precise morphology of L. crispatus CIP104459 was studied by scanning electron microscopy using a TENEO VolumeScope microscope (FEI, Hillsboro, OR, USA) under 10 kV. For that, L. crispatus grown for 18 h in the absence (control) or presence of 17β-estradiol were collected by centrifugation (7,500 xg, 10 min). For fixation, bacterial pellets were immersed in 1 mL 2.5% glutaraldehyde in 1 M phosphate buffer, pH 7.1 for 1 h. Sample preparation was carried out as previously described54. After HMDS treatment (hexamethyldisilazane), the surface of the filter was coated with an electrically conductive 25 nm thick layer of platinum alloy coating in a sputter-coater vacuum chamber (LEICA EM ACE600 sputter coater, Wetzlar, Germany).
Evaluation of biosurfactant production. Cultures on MRS Petri dishes showed that L. crispatus CIP104459 grown in the presence of 17β-estradiol were spreading and flowing on the surface of the culture medium suggesting biosurfactant production. Then, MRS Petri dishes were inoculated in surface by spreading 100 µL of control or 17β-estradiol treated bacterial suspensions to obtain a continuous bacterial mat. As described by Meylheuc et al.55, after 24 h of anaerobic culture at 37°C, the bacterial mat was gently scraped and resuspended in 15 mL Volvic water (selected for its neutral effect on surface tension). Then, the suspension was shaken by vortex for 3 min and centrifuged two times for 30 min at 4°C (10,000 xg) to remove all bacteria and fragments. The supernatant was collected and stored at 4°C.
The presence of biosurfactant in solution was first investigated by the sessile drop technique56 by deposit on a polystyrene surface of 20 µL drops of supernatant and visualization of the contact angle with the surface. Measurement of this contact angle can be used to calculate the surface tension between the solution and the surface. However, this value being influenced by the properties of the surface, another approach, the pendant drop method was preferred. The shape of the drops of supernatant was analyzed using a drop shape analyser DSA30 temperature controlled tensiometer equipped with a video camera (Kruss, Hamburg, Germany). The surface tension, or interfacial tension, was calculated using the tensiometer drop shape analysis software based on the Young-Laplace equation57. In order to take in consideration any potential variation due to the bacterial biomass collected, calculated surface tension values were correlated to the OD600nm of the bacterial suspension measured immediately after scrapping of the mat on the Petri dishes.
Measure of bacterial adhesion to vaginal cells. Adhesion of L. crispatus CIP104459 to vaginal cells was studied in vitro using the VK2/E6E7 cells line (ATCC CRL-2616). This vaginal cell line was developed from the vaginal mucosal tissue of a healthy pre-menopausal women. It was maintained and propagated using keratinocyte serum-free medium (KSFM) (Thermo Fisher Scientific, Waltham, Massachusetts, USA), supplemented with 0.05 mg/mL bovine pituitary extract, 0.1 ng/mL human recombinant EGF and additional calcium chloride 44.1 mg/mL as recommended by the provider. Cells were replicated into fresh medium every 3 days after reaching confluence. For evaluation of bacterial adhesion, VK2/E6E7 cells were exposed to L. crispatus CIP104459 grown for 18h in MRS medium with or without 17β –estradiol at an MOI of 100 bacteria /cell. Experiments were performed using KSFM without antibiotics and VK2/E6E7 cells reaching a minimum of 80% confluence. After 1 h of bacterial interactions, the medium was removed slowly to withdraw planktonic bacteria and rinsed carefully with KSFM without antibiotics. Then VK2/E6E7 cells were disrupted by addition of 0.1% triton X-100 in PBS. The lysate was diluted in MRS medium and plated on MRS agar Petri dishes. The number of cells adherent bacteria was deduced from direct counting of growing L. crispatus colonies after 48 h culture at 37°C in anoxic conditions.
Determination of biofilm formation activity and structure. Biofilm formation by L. crispatus CIP104459 was initially studied using the crystal violet technique according to a procedure adapted from O’Toole58. Bacteria grown in MRS for 48h were adjusted to OD600nm = 0.1 in a final volume of 1 mL and distributed in sterile 24-well polystyrene plates (Falcon®, Durham, USA). Plates were incubated anaerobically at 37°C without agitation for 48 hours in a Whitley A85 Workstation. At the end of the incubation period, the medium was discarded by aspiration and non adhered bacteria were removed by washing twice with physiologic water (PW, NaCl 0.9%). Biofilms were stained by incubation with crystal violet (0.1% w/v in sterile pure 18.2MΩ water) for 10 minutes at room temperature. The excess of dye was removed by three washing steps in pure 18.2MΩ water. The crystal violet adsorbed in the biofilm matrix and on bacteria was dissolved with absolute ethanol and the OD595nm of the solution was determined using an automated plate reader (Te-CoolTM, Tecan Group Ltd., Männedorf, Switzerland).
L. crispatus CIP104459 grown in MRS was unable to adhere and form biofilm on glass surfaces, as required for optical microscopy. A more complete medium, specific to microorganisms of the vaginal microflora, designated as “simulating genital tract secretion” (SGTS) medium was produced according to Geshnizgani and Onderdonk33 (Table 1). Then, L. crispatus CIP104459 pre-culture were realized in MSR and bacteria were transferred in SGTS medium for biofilm studies. 17β-estradiol, or an equivalent amount of ethanol in water, was administered from the onset of the biofilm formation study. The absence of effect of 17β-estradiol on L. crispatus CIP104459 growth in SGTS medium was controlled in preliminary studies (Suppl Fig. 3). For confocal microscopy, bacteria were harvested by centrifugation (7,500 xg, 10 min) and re-suspended in SGTS medium at OD600nm = 0.1. Suspension aliquots (1 mL) containing or not 17β-estradiol were layered in 24-well microplates with flat glass bottom (Sensoplate, Greiner Bio-One, Germany) and incubated for 48 h in static and anoxic conditions to allow biofilm formation. Wells were washed twice with PW to remove remaining planktonic bacteria and biofilms were stained with SYTO 9 Green Fluorescent Nucleic Acid Stain (Thermofisher, Waltham, Massachusetts, USA). Stained samples were examined under an LSM 710 inverted confocal laser scanning microscope (Zeiss, Marly-le-Roi, France) using the Zen 2009 software package (version 188.8.131.522). Images were reconstructed and analyzed using the COMSTAT2 software. Biofilm average thickness (μm), mean biomass volume (μm3/μm2) and roughtness coefficient were calculated over a minimum of 30 observations in each condition. All experiments were repeated at least of three times.
Bioinformatic studies. As a steroid, 17β-estradiol is an amphiphilic molecule and is not showing reactive groups allowing to form directly covalent conjugates with dyes or tracers. In addition, because of its small size, minor chemical modifications can deeply affect its functions. Incorporation of isotopes remains the sole technique to investigate experimentally its binding to potential receptors, but these techniques are submitted to heavy regulatory constraints. Then we decided to use bioinformatics approaches to investigate potential 17β-estradiol binding site(s) in L. crispatus. As previously mentioned, the draft genome of L. crispatus CIP104459 was recently published28. This draft genome was aligned by BLASTp on ExPASy (https:// web.expasy.org/blast) with the annotated sequence of L. crispatus CO3MRSI134 to improve its definition. The FASTA amino-acid sequence of the L. crispatus CIP104459 genome was then aligned with the sequence of known human estrogens receptors listed in Table 2. A bacterial protein showing high homology with human estrogens receptors was identified and submitted to molecular docking in silico using 17β-estradiol as potential ligand. L. crispatus proteins 3D models were generated using RaptorX Structure59 and visualized using Python Molecular Viewer V1.5.6. Prediction by alignment on the crystalized structure of the corresponding eukaryotic ortholog. The potential binding of 17β-estradiol to the identified sensor protein was studied using AutoDock 4.235. Binding values were generated using the Lamarkian Genetic Algorithm of AutoDock 4.2. All calculations were performed using a DELL Precision T7610 computer equipped with four hard disks (4 To each, for a total of 12 To under RAID5).
Statistical analysis. Statistical significance of experimental values was evaluated using the Prism GraphPad online tool (https://www.graphpad.com/quickcalcs/ttest1/). Data were analyzed using unpaired (two sample) two-tailed t test to calculate p values. Mean with standard error of the mean (SEM) were calculated and plotted.