Bovine bile (reference B3883, Sigma-Aldrich and reference 212820, Difco, BD Diagnostic Systems) was prepared at 100 mg/ml stock solution in distilled water, sterilized through 0.2 μm filters and stored at −20°C. Pure bile salts: glycocholate (GC), taurocholate (TC), glycodeoxycholate (GDC), taurodeoxycholate (TDC), glycochenodeoxycholate (GCDC), and taurochenodeoxycholate (TCDC), bile salt mix or pure deconjugated bile salts: cholate (C), deoxycholate (DC), and chenodeoxycholate (CDC), and fusidic acid were from Sigma-Aldrich and dissolved in distilled water to 12 mg/ml stock solutions, filtered through 0.2 μm pores and kept at −20°C. Choloylglycine Bile Acid Hydrolase or Bile Salt Hydrolase (BSH, EC 3.5.1.24) from Clostridium perfringens (reference C4018, Sigma-Aldrich) was prepared at 10 U/ml in distilled water and stored at −20°C. Fetal calf serum (FCS, reference A15-101, PAA Laboratories, GE Healthcare), and bovine calf serum (reference B9433, Sigma-Aldrich) were used. The NEFA-C kit used for quantitative determination of non-esterified fatty acids (NEFAs) was from Biolab, WAKO Diagnostics.

Trophozoites of G. duodenalis strains WB (clone C6, ATCC 30957) and HP1 (Portland-1, gift of Jan Tackezy, Charles University of Prague, Czech Republic) from assemblage A were routinely maintained in axenic culture in TYI-S-33 Keister's medium (KM) as previously described (Keister, 1983). Parasites were grown in 15 ml tubes filled with 12 ml culture medium. They were regularly subcultured at a density of 5 × 10⁴ cells per tube (12 ml) from log phase parasites chilled on ice for 10 min and centrifuged at 700 × g, 5 min. For bioassays, parasites were maintained in culture in KM adjusted to pH 6.0, supplemented with 10% heat-inactivated FCS (Paget et al., 2004) and 0.6 g/L bovine bile (Carnaby et al., 1994).

L. johnsonii La1 (CNRZ 1897, NCC533) was kindly provided by Pascal Quénée (INRA Jouy en Josas, Equipe Atalis, France) and was isolated from LC1 product in 1996 (Chambourcy, France). The presence of L. johnsonii La1 prophage, as previously described (Ventura et al., 2004; Denou et al., 2008), was checked by PCR using two primer pairs: 1DF/1B2R resulting in a 1500 bp amplicon for L. johnsonii “La1-like” (with the prophage) and 1EF/1B2R resulting in a 268 bp amplicon as a positive control for L. johnsonii strains. PCR reactions were run in a final volume of 25 μl containing 200 ng of bacterial DNA extracted with a standard protocol (Zhong et al., 1998), 200 nM of each primer (1DF: AGA CTT TTT GGC AGG CAA AGG, 1EF: ACA AAC CAC CAG TGC CTA AGG, 1B2R: GCT CTT CGA GAT CAC TGG GC), 200 nM of dNTP, and 1U of Taq DNA polymerase (NEB M0273S). Reactions were initiated with an initial denaturation for 5 min at 95°C followed by 35 cycles at 95°C for 30 s, 50°C for 30 s and 72°C for 1 min. PCR products were visualized on 0.8% agarose gel containing 0.5 ng/μl of BET and images were captured with a camera coupled with Biocapt software, Vilber Lourmat.

L. johnsonii La1 stock cultures were kept frozen in de Man Rogosa Sharpe (MRS) Broth (Sigma-Aldrich) media with 15% glycerol. Bacteria were subcultured in MRS or modified TYI-S-33 medium (MTYI) (Pérez et al., 2001) and incubated anaerobically for 12–18 h at 37°C. Bacteria were subsequently grown in MTYI or KM supplemented with or without 10% heat-inactivated FCS for 12–16 h in the presence or the absence of 0.6 g/L bovine bile. After centrifugation (3000 × g, 10 min) and 0.2 μm filtration, the pH of the L. johnsonii La1 supernatant was adjusted to 6.0 with 5 N NaOH. For some experiments, pH was adjusted to 6.2, 6.7, 6.9, or 7.2. Appropriated controls were prepared as follows: lactic acid produced during growth was quantified from aliquots of supernatants (Enzytec™ kit, R-Biopharm) and an equivalent amount of lactic acid was added to fresh medium before pH adjustment.

One milliliter of trophozoite suspension (1 × 10⁵ parasites/ml in KM, pH 6, supplemented with 10% FCS without bovine bile) was mixed with either 500 μl of L. johnsonii La1 supernatant or one, or two, units of commercial bile salt hydrolase (BSH) from C. perfringens, chromatography purified fractions or appropriate control medium. These suspensions were further mixed with different concentrations of either bovine bile, mixed bile salts, conjugated, or deconjugated bile salts or appropriate control medium. Samples were then incubated for 24 h at 37°C and chilled on ice for 10 min to dislodge trophozoites from tube wall. Trophozoites were counted using a Malassez cell chamber and were considered alive when parasites showed the typical pear shape and signs of flagella mobility (Video S1). Morphologically altered parasites with non-mobile flagella were considered “dead” (Video S2). Estimates of parasite viability on these criteria correlate with parasite viability measured by propidium iodide staining and flow cytometry analysis (Barbosa et al., 2008; Figure S1). Multiplication factor [i.e., number of total trophozoites at the end of the experiment/number of trophozoites at time zero], survival rate [i.e., (number of living cells/total number of trophozoites) × 100], and inhibition percentage [i.e., (100-number of living cells in the presence of tested compounds/number of living cells in control) × 100] were calculated.

Supernatants from L. johnsonii La1 cultures in KM, adjusted to pH 6.0, were concentrated up to 30-fold by ultrafiltration using the 10 kDa Centriprep centrifugal filter unit (Millipore). After 0.2 μm filtration, the concentrated supernatants were loaded onto a Sephacryl S300 column 16/100 (GE Healthcare), previously equilibrated with 20 mM ammonium sulfate (pH 6.0), in a cold room and were eluted with the same buffer at a flow rate of 2.0 ml/min. Effluent fractions of 12 ml were collected, concentrated 4-fold by ultrafiltration on a 10 kDa Centriprep and tested for in vitro Giardia growth inhibition in the presence of bovine bile or bile salts. Fractions obtained by similar processing of elution buffer alone and control media containing lactic acid (see above) were used as controls. Column calibration was carried out with ribonuclease A (13.7 kDa) and bovine serum albumin (67 kDa).

The molecular size of inhibitory molecule(s) present in L. johnsonii La1 supernatant was assessed by ultrafiltration using 10, 30, and 50 kDa Centriprep centrifugal filter units. Thermal stability was tested by heating the bacterial supernatant to 90°C for 10 min. Preservation of L. johnsonii La1 supernatant activity upon dialysis was checked by dialyzing twice (at 4°C for 2 and 15 h, respectively) the supernatant against 100 volumes of KM supplemented with 10% FCS or against GKN solution (NaCl, 8 g/l; KCl, 0.4 g/l; glucose, 2 g/l; NaH₂PO₄, H₂O, 0.69 g/l; Na₂HPO₄, 1.57 g/l; pH 7.2–7.4) (Pérez et al., 2001) using a dialysis membrane with a molecular weight cut-off of 3.5 kDa (Spectrum Laboratories). The dialyzed supernatant was then filtrated through a 0.2 μm membrane and kept frozen at −80°C prior to G. duodenalis inhibition assays.

The biochemical nature of inhibitory molecule(s) was investigated by preincubating the 5-fold concentrated L. johnsonii La1 supernatant obtained by ultrafiltration (>10 kDa) with different enzymes coupled to beads. Briefly, proteinase K (Invitrogen), pronase and catalase (Merck) were coupled to CNBr-activated Sepharose™ 4B beads (GE Healthcare) following manufacturer's instructions. Five milliliters of the 5-fold concentrated L. johnsonii La1 supernatant or the 5-fold concentrated fresh control medium were incubated for 6 h at room temperature (RT) in the presence of 100 μl of packed enzyme-coupled beads. Beads were removed by centrifugation (4000 × g, 5 min) before parasite growth inhibition assays.

To assess the presence of free fatty acids in the complex medium inducing G. duodenalis growth inhibition, FCS, bile, and L. johnsonii La1 supernatant were analyzed for non-esterified fatty acid (NEFA) content either alone or in combination in KM. Respective concentrations of the components in samples were as follows: FCS 10% (v/v), bile 0.6 g/L and L. johnsonii La1 supernatant 33.3% (v/v) final concentration in KM, pH 6. Samples (0.5 ml) were kept on ice before being incubated for 24 h at 37°C in the presence of 4.8 × 10⁴ trophozoites or without parasites. At the end of the incubation period, tubes were chilled on ice, centrifuged at 700 × g, 10 min at RT. The supernatants were then taken and frozen at −80°C before NEFA measurement. Numbers of living and “dead” trophozoites (as defined above) were determined using a Malassez cell chamber. NEFAs were quantified using the spectrometric NEFA-C kit method, based on Duncombe (1964) and Itaya and Ui (1965) studies, following manufacturer's instructions. Oleic acid was used as a standard and NEFAs were expressed as oleic acid equivalents (Eq).

After gel filtration, the eluted fractions were concentrated 10-fold by dialysis against 20 mM ammonium acetate buffer containing 2 M sucrose, pH 6.0, using a 3.5 kDa molecular weight cut-off membrane (Spectrum Laboratories). Bile salt GDC was used for enzymatic assays. BSH activity was monitored by measuring glycine liberation from conjugated bile salts following the protocol described by Grill et al. (2000). Briefly 100 μl of effluent fractions or 1 unit of BSH or elution buffer were mixed with 100 μl of 2.4 g/l of GDC and incubated for 24 h at 37°C. Controls were performed in the absence of the bile salt. The enzymatic reaction was stopped by addition of an equal volume of 15% TCA. Precipitated proteins were pelleted by centrifugation at 20,000 × g for 15 min and the supernatant was carefully removed. To 80 μl of supernatant, 680 μl of 0.3 M borate buffer, 1% SDS, pH 9.5 and 80 μl of 0.3% picrylsulfonic acid solution (Sigma-Aldrich) were added. The mixture was incubated for 30 min in the dark and 800 μl of 1 mM HCl was added to stop the reaction. Glycine concentration was measured at 416 nm using an Uvikon spectrophotometer 930 (Kontron Instruments). A standard curve was established using free glycine (Sigma-Aldrich).

L. johnsonii La1 supernatant prepared in KM, pH 6.0, supplemented with 10% FCS and 1 g/L bovine bile was incubated overnight at 37°C. Two different bile batches and two different L. johnsonii La1 supernatant preparations were tested. Controls constituted of KM alone and heat-treated L. johnsonii La1 supernatant (90°C, 10 min). The sample preparation and LC-MS protocol were adapted from the conditions reported in the literature on the LC-MS analysis of bile (Mitamura et al., 2011) and bile acids (Perwaiz et al., 2001; Griffiths and Sjövall, 2010). After incubation, the samples were diluted 4-fold in Milli-Q water and subjected to solid-phase extraction using Oasis® HLB cartridges (30 mg solid phase). After conditioning with 3 ml methanol and 3 ml Milli-Q water, the cartridges were loaded with 1 ml of the 4-fold diluted samples, washed with 2 ml Milli-Q water and eluted with 2 ml methanol. The eluted fractions were vacuum dried and resuspended in 500 μl Milli-Q water/acetonitrile 90:10 (v/v). 5 μl of each resuspended sample was analyzed by liquid chromatography coupled electrospray ionization-mass spectrometry (LC/ESI-MS) on an Ultimate U3000 chromatographic system (Thermo Scientific) connected to a Q-STAR Pulsar Qq-TOF mass spectrometer equipped with an ionspray source (Sciex). The LC separation was achieved by an Interchrom Strategy C18-2 micro column (5 μm, 150 × 1 mm, 100 Å, Interchim). The elution gradient was 10% mobile phase B (acetonitrile) to 70% B against mobile phase A (5 mM ammonium formate/formic acid, pH 6) over 45 min, at a flow rate of 40 μl/min. The MS data were acquired in negative mode in the range m/z 250–1200. Each LC/ESI-MS experiment was conducted twice. Data-dependent LC/ESI-MS/MS experiments were also conducted on each sample, alternating 1-s full-scan MS followed by 2-s product ion collision induced dissociation of the major ions detected at the first step, using a −50 V collision voltage.

The raw LC/ESI-MS data were converted into Network Commun Data Form (NetCDF) using the translation tool provided by Sciex and processed with the XCMS package (Smith et al., 2006), a software implemented in the freely available R environment (www.r-project.org). This analysis allows automatic retention time alignment, matched filtration, peak detection, and peak matching.

Sodium lauroyl sarcosinate (NLS, 0.1%, Sigma-Aldrich) was added to L. johnsonii La1 supernatant. After mixing, trichloroacetic acid (TCA) was added to a final 7.5% concentration and the solution was precipitated on ice overnight. Mixed protein-detergent precipitate was collected by centrifugation (10,000 × g, 10 min, 4°C). The supernatant was carefully removed and the pellet was washed twice with 2 ml of precooled tetrahydrofuran (Sigma-Aldrich). The pellet was then resuspended in 8 M urea in 20 mM triethylammonium bicarbonate and incubated for 1 h at 20°C with 20 mM DTT, followed by incubation with 50 mM iodacetamide for 1 h at 20°C in the dark. The sample was incubated with 0.05 U of endoproteinase Lys-C (Wako Pure Chemical Industries) for 18 h at RT. Then, the samples were diluted with 20 mM triethylammonium bicarbonate to a final concentration of 1 M urea and trypsin (Promega) digestion was performed with 2 μg of enzyme for 4 h at 20°C and terminated with a final concentration of 0.5% trifluoroacetic acid. The sample was passed sequentially through two home-made Poros Oligo-R3 (PerSeptive Biosystems) microcolumns packed (~1 cm) on p200 tips over 3MM C18 material plug. Loaded resin was washed with 100 μl 0.1% trifluoroacetic and peptides were eluted with 100 μl 70% acetonitrile/ 0.1% trifluoroacetic, then 20 μl 100% acetonitrile. The desalted sample was dried down, resuspended in 50% acetonitrile, dried, and stored at −80°C. Part of the sample (10%) was collected for amino acid analysis using a Biochrom 30 amino acid analyzer.

Samples (3 μg per run) were analyzed by an EASY-nano LC system (Proxeon Biosystems) coupled online to an LTQ-Orbitrap Velos mass spectrometer (Thermo Scientific). Peptides were loaded onto a 18 cm fused silica emitter (75 μm inner diameter) packed in-house with reverse phase capillary column ReproSil-Pur C18-AQ 3 μm resin (Dr. Maisch GmbH) and eluted using a gradient from 100% phase A (0.1% formic acid) to 35% phase B (0.1 formic acid, 95% acetonitrile) for 180 min, 35 to 100% phase B for 5 min and 100% phase B for 8 min (a total of 23 min at 250 nl/min). After each run, the column was washed with 90% phase B and re-equilibrated with phase A. Mass spectra were acquired in positive mode applying data-dependent automatic survey MS scan and tandem mass spectra (MS/MS) acquisition. Each MS scan in the orbitrap (mass range of m/z of 400–1800 and resolution 100,000) was followed by MS/MS of the 15 most intense ions in the LTQ. Fragmentation in the LTQ was performed by collision-induced dissociation and selected sequenced ions were dynamically excluded for 25 s.

Raw data were viewed in Xcalibur v.2.1 (Thermo Scientific) and data processing was performed using Proteome Discoverer v.1.3 (Thermo Scientific). The generated raw files were submitted to searching using Proteome Discoverer with an in house Mascot v.2.3 algorithm against a database of proteins predicted from the L. johnsonii isolates NCC533 and FI9785 and the L. johnsonii prophage Lj965 (http://www.ncbi.nlm.nih.gov/bioproject). Contaminant proteins (human keratins, BSA and porcine trypsin) were also added to the database and all contaminant proteins identified were manually removed from the result lists. The searches were performed with the following parameters: ms accuracy 10 ppm, MS/MS accuracy 0.6 Da, trypsin digestion with one missed cleavage allowed, fixed carbamidomethyl modification of cysteine and variable modification of oxidized methionine and N-terminal protein acetylation. The number of proteins, protein groups, and number of peptides were filtered for False Discovery Rates (FDR) less than 1%. Only peptides with rank 1 and minimal of 2 peptides per protein were accepted for identification using Proteome Discoverer (Charneau et al., 2007; Queiroz et al., 2013).

Significant differences between treatments were tested using the “Kruskal-Wallis” test in http://marne.u707.jussieu.fr/biostatgv/. Results were considered significant at P < 0.05.

Pérez et al. (2001) showed that the inhibitory effect of L. johnsonii La1 supernatant upon in vitro G. duodenalis growth was pH dependent with a more pronounced inhibitory effect at pH 6 than pH 7. Change of culture medium pH had a slight effect on parasite proliferation and adhesion properties. We thus set up our culture assays at pH 6. FCS was used instead of adult bovine serum for better reproducibility of the assay. In our experimental conditions, parasite morphology, motility, and viability measured by propidium iodide staining were unaffected (Video S1, Figure S1).

Most media previously described to support G. duodenalis growth in vitro commonly contains bile as a supply for parasite cholesterol and fatty acid requirements (Farthing et al., 1985; Gillin et al., 1986; Halliday et al., 1995). In our hands, G. duodenalis trophozoites growth was observed in the absence of bovine bile. Bile, at the recommended concentration (0.75 g/L) (Pérez et al., 2001), appeared to reduce trophozoite proliferation when added to KM-FCS (Figure 1A). We next confirmed the in vitro inhibitory effect of L. johnsonii La1 supernatant upon G. duodenalis growth. Interestingly, the inhibitory effect was only observed in the presence of bovine bile. No inhibitory effect was observed in the absence of bovine bile, even after 24 h of culture (Figure 1A). After 10 h in the presence of 0.75 g/L of bovine bile, G. duodenalis trophozoite survival was slightly impacted by L. johnsonii La1 supernatant, but survival was largely affected after 24 h of contact (9.10⁴ and 1.10⁴ trophozoite/ml, respectively, Figure 1A, Video S2, Figure S1). Thus, a 24 h incubation time-period was retained for all subsequent inhibitory growth assays.

Maximal inhibitory effects of L. johnsonii La1 supernatant on parasite growth were observed either in the presence of 0.6 g/L of complete bovine bile or in the presence of 0.032 g/L of bile salt mix (96.3 and 75% of inhibition, respectively, by comparison with control media containing the same amount of lactic acid, bile or bile salt mix, Figure 1B, p < 0.05). Similar inhibitory effects were observed using two isolates of G. duodenalis (WB and HP1), two different commercial origins of complete bovine bile (Figure S2) and two different media compatible with bacteria and parasite growth: MTYI (Pérez et al., 2001) and KM-FCS (Paget et al., 2004). Unless otherwise indicated, subsequent in vitro G. duodenalis inhibition growth assays were performed using KM as medium at pH 6 and 0.6 g/L of complete bovine bile.

Since concomitant addition of bovine bile and L. johnsonii La1 supernatant to the culture medium leads to inhibition of G. duodenalis growth, we assessed whether bile composition might be modified by L. johnsonii La1 supernatant. Bile composition was investigated by LC/ESI-MS after 24 h of incubation with L. johnsonii La1 supernatant (Figures 3A,B). Impacted molecules were identified by their m/z MS/MS fragmentation pattern and comparison with standards. Comparison of bile salt profiles showed a decrease of conjugated salts (GC, TC, GDC, TDC, GCDC, TCDC) in favor of non-conjugated salts. C and DC were the main statistically enhanced non-conjugated salts (Figure 4) and in a minor proportion, CDC. These modifications were not observed in presence of heat-treated L. johnsonii La1 supernatant (Figure 3C).

The inhibition of G. duodenalis growth by pure bile salts (C, DC, and CDC) conjugated to glycine or taurine, or their deconjugated counterparts were investigated in the presence or the absence of L. johnsonii La1 supernatant (Table 3). In the absence of L. johnsonii La1 supernatant, glycyl-, or tauryl-conjugated salts and deconjugated C showed no apparent toxicity at the concentrations tested (IC₅₀ values >400 μM). In contrast, deconjugated DC, and CDC exhibited inhibitory effects on trophozoite growth with IC₅₀ values of 132 μM (DC) and 147 μM (CDC). Interestingly, in the presence of L. johnsonii La1 supernatant, the conjugated bile salts IC₅₀ values fell to a range of values similar to those measured for their pure deconjugated counterparts, i.e., 104 (GDC), 79 (TDC), 110, (GCDC), and 115 μM (TCDC) (Table 3).

These results suggested that a deconjugating process mediated by L. johnsonii La1 supernatant component(s) and production of deconjugated bile salts might be responsible for the inhibitory effect of the association of bile with L. johnsonii La1 supernatant. Such a hypothesis is in line with the previous observation that fusidic acid, an antibiotic with a bile salt-like chemical structure, is toxic to Giardia (see Table 3, IC₅₀value = 26 μM) unless conjugated to taurine or glycine (Farthing and Inge, 1986).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.