S. Typhimurium wild-type (WT) strain ATCC 14028s and its ΔramR, ΔramR/pramR, and ΔramRA::kan/pramA derivatives were used in this study. Deletion mutants were constructed using the Datsenko and Wanner inactivation gene method as previously described (Datsenko and Wanner, 2000; Abouzeed et al., 2008). Complementation plasmids carrying the ramR gene (pramR) or the ramA gene (pramA) were previously described (Abouzeed et al., 2008; Nikaido et al., 2008). Bacterial strains were grown at 37°C in Luria–Bertani broth (LB, pH 7.5) supplemented with 25.6 g/L bile or with 5 mM of individual bile salts where appropriate. Bile used in this work was a crude ox-bile extract which contains the main bile sodium salts of taurocholic, glycocholic, deoxycholic, and cholic acids purchased under the label “sodium choleate” (Sigma–Aldrich, Steinheim, Germany). Physiological concentrations of bile salts encountered by bacteria in the intestinal lumen are variable with high and low concentrations in the anterior small bowel and distal ileum, respectively, estimated in the millimolar range that is consistent with their critical micellar concentrations (e.g., 6–10 mM for taurocholic acid) (Martinez-Augustin and Sanchez de Medina, 2008). In a previous study, we showed that a bile concentration of 25.6 g/L allowed normal growth (i.e., similar to growth control in LB medium) of S. Typhimurium isolates (Baucheron et al., 2005). Individual bile salts (Sigma–Alrich, Steinheim, Germany), also allowed normal growth of the tested strains when used at 5 mM.

Invasions assays were performed as previously described (Rosselin et al., 2010). Caco-2 intestinal epithelial cells were grown in Dulbecco's modified Eagle medium (DMEM) supplemented with 10% inactivated fetal bovine serum, 1% non-essential amino acids, and 1% antibiotic solution (Gibco, Invitrogen), at 37°C under 5% CO₂. Cells were seeded at a density of 2 × 10⁵ cells/well in a 24-well plate (Falcon) and grown until confluence in the same medium. Antibiotic was removed 24 h before the invasion assays. Bacteria grown to an OD₆₀₀ of 0.6 in LB broth were inoculated on Caco-2 monolayers at a multiplicity of infection (MOI) of 30. After a 30 min incubation, the bacteria-containing medium was removed from the wells, and the cells monolayers were washed with phosphate buffered saline (PBS). For adhesion assays, cells were then lysed for 30 min with sterile ultrapure water and serial dilutions of lysates were plated on LB agar. For invasion assays, cells were further incubated for 1.5 h with DMEM supplemented with gentamicin at 100 μg/mL. After washing with PBS, cells were lysed with sterile ultrapure water and serial dilutions of lysates were plated on LB agar. The percentage of penetrating bacteria was calculated as the ratio of the counted colony forming units (cfu) to the bacterial inoculum. All assays were repeated at least twice, with three replicates for each tested condition. Data presented correspond to mean values of at least six replicates for invasion and adhesion assays.

Bacteria were grown in 20 ml liquid cultures (standard LB, 1% NaCl), in 125 mL Erlenmeyer flasks, under shaking at 180 RPM, for about 150 min, until they reached an OD₆₀₀ value of 0.6. Culture samples were pelleted by centrifugation, stabilized with RNAprotect Bacteria Reagent (Qiagen) and stored at −80°C until use. Total RNA was extracted using the RNeasy Mini kit (Qiagen) following the manufacturer's instructions. Residual genomic DNA was removed using the Turbo DNA-free kit (Ambion). Total RNA (1.5 μg) was reverse-transcribed using random hexamers and the Superscript III First Strand Synthesis System (Applied Biosystems). The expression level of each gene was calculated from three independent cDNA samples. For each cDNA sample and each gene, qRT-PCR runs were performed in duplicated wells. Primers and cycling conditions used for qRT-PCR were previously described (Giraud et al., 2013). The relative quantities of transcripts were normalized against the geometric mean of three reference genes (gmk, gyrB, rrs). Statistical significance was assessed at a P-value of <0.05 using a two-tailed Student's t-test.

The qRT-PCR assays of this study confirmed that crude ox bile extract at 25.6 g/L increased ramA transcript levels about 20-fold, as previously described (Figure 1A) (Baucheron et al., 2014). To address the effects of individual bile salts on ramA expression, primary bile salts and their derivatives (dehydroxylated and glycine- or taurine-conjugated) were tested at the concentration of 5 mM. Chenodeoxycholate showed the most important effect, by increasing ramA transcript level ~7- vs. 4-5-fold for its taurine- and glycine- conjugated derivatives in the S. Typhimurium strain ATCC 14028s (Figure 1A). Cholate also increased ramA mRNA levels by a 4-fold factor, i.e., notably more than its dehydroxylated derivative, deoxycholate (2.5-fold). In contrast, no ramA-inducing activity could be detected for both cholate conjugates, taurocholate and glycocholate (Figure 1A).

In contrast to their effects on ramA expression, bile and individual bile salts globally showed a repressive effect on hilA expression in S. Typhimurium ATCC 14028s (Figure 1B). Furthermore, this repressive effect on hilA expression appeared correlated to the ramA-inducing effect. Indeed, crude bile extract, which showed the strongest ramA-inducing effect, decreased hilA transcripts to very low levels (about 1.5% that of the control) (Figure 1B). By contrast, bile salts with low (deoxycholate) or no (taurocholate, glycocholate) ramA-inducing activity, showed no significant repressive effects on hilA expression. Other bile salts, which induced ramA between 4- and 7-fold (see above), decreased hilA transcript levels 3–4-fold (chenodeoxycholate and taurochenodeoxycholate), 2.3-fold (cholate), and 1.6-fold (glycochenodeoxycholate) (Figure 1B). The repressive effects exerted on hilA by crude bile extract and by chenodeoxycholate and its conjugates were also observed for invA, which encodes another SPI-1 positive regulatory protein (Supplementary Figure S1). However, no significant effects of cholate and its derived salts were observed on invA expression (Supplementary Figure S1).

Further to above data, we hypothesized that the repression exerted by some bile salts on hilA expression could result into decreased cell invasion efficiency of S. Typhimurium ATCC 14028s. Therefore, gentamicin protection assays were performed to address the effects of bile and individual bile salts on adhesion to and invasion of Caco-2 intestinal epithelial cells (Figures 1C, D). Crude bile extract, chenodeoxycholate and its two conjugates, taurochenodeoxycholate and glycochenodeoxycholate, decreased adhesion to the Caco-2 cells by about 20% (Figure 1C). Crude bile extract decreased S. Typhimurium ATCC 14028s invasion about 5-fold, and chenodeoxycholate and taurochenodeoxycholate (i.e., the bile salts which most efficiently repressed hilA expression), decreased its invasion by about 2.5-fold (Figure 1D). These decreased invasions may be explained partly by the decreased adhesion of the strain mentioned above. It is also possible that the 20% decrease of invasion observed with glycochenodeoxycholate was mostly due to defective adhesion of the strain. Bile salts that had no significant effect on hilA expression did not repress invasion of the S. Typhimurium strain ATCC 14028s.

These data suggest that some individual bile salts actively participate to the repression of S. Typhimurium cell invasion by bile. They also reveal that the specific structure of bile salts may determine their activity as environmental signals to regulate gene expression and cell invasion of S. Typhimurium. Nevertheless, since only one S. Typhimurium strain was investigated, these data must be taken with caution, to avoid any overinterpretation regarding S. Typhimurium as a pathogenic serovar, or more generally Salmonella as a pathogen, since as previously published distinct genetic lineages of serovar Typhimurium may behave differently regarding cell invasion and its regulation (Giraud et al., 2013).

In our experimental conditions, chenodeoxycholate seemed to be the most active bile salt and its conjugation to notably glycine, appeared to decrease its activity. Although they do not establish any causality relationship between ramA activation and hilA repression, our results indicate that ramA overexpression is associated to decreased TTSS-1 genes expression and to decreased invasion of the S. Typhimurium strain studied. These different correlations led us to examine further the actual involvement of the ramRA regulatory locus in the bile-mediated repression of S. Typhimurium ATCC 14028s cell invasion.

A previous study suggested that, whereas ramA is activated by bile mainly depending on ramR, another undetermined ramR-independent pathway also contributes to the up-regulation of ramA by bile in S. Typhimurium (Baucheron et al., 2014). Here, we assessed whether the major individual bile salts (chenodeoxycholate, cholate and deoxycholate) could induce ramA expression in S. Typhimurium ATCC 14028s by different, ramR-dependent and/or ramR-independent pathways.

The increase of ramA expression in the WT S. Typhimurium ATCC 14028s strain, in the presence of bile or of the three tested salts, were similar to those reported above (Figure 2A). As expected, in control cultures (LB alone), the ΔramR mutant expressed ramA transcript levels about 10-fold higher than those of the WT strain, and complementation with a functional ramR gene (using pramR) restored WT ramA expression levels. Chenodeoxycholate increased ramA transcript levels by 7.2-fold in the WT strain, probably in a fully ramR-dependent manner, since no increase was observed in the ΔramR mutant. Cholate increased ramA transcript level 5.6-fold in the WT strain. However, in contrast to chenodeoxycholate, cholate also increased ramA expression about 1.6-fold in the ΔramR mutant compared to LB medium alone, indicating that it might also use a minor ramR-independent pathway to activate ramA expression. Lastly, the slight effects of deoxycholate on ramA transcript level appeared similar in the WT and in the ΔramR background (1.7- and 2.7-fold increase, respectively), indicating that deoxycholate likely uses mainly a ramR-independent pathway to achieve this effect. Irrespective of the tested bile salt, complementation using the pramR multicopy plasmid resulted in ramA transcript levels similar, or even lower, than those observed in the WT strain. These results indicate that the individual bile salts tested differ not only in the magnitude of their effects on ramA expression, but also in their relative use of the ramR-dependent and ramR-independent pathways to achieve this effect. Considering these results, the bile-mediated increase of ramA expression that we observed, here and before, can be interpreted as being, at least partly, the intricate result of the individual effects of bile salts, which would explain the implication of both ramR-dependent and ramR-independent mechanisms. It remains also possible that bile components other than the tested bile salts also participate to ramA up-regulation, depending or not on ramR.

In view of the negative correlation between ramA and hilA expression levels, we also tested to what extent hilA repression by chenodeoxycholate, cholate and deoxycholate depended on ramR. In control cultures (LB alone), hilA expression levels in the ΔramR mutant were 8-fold lower than in the WT strain, in agreement with previously reported results (Giraud et al., 2013). In the WT strain, the hilA transcript levels were significantly decreased in the presence of chenodeoxycholate (6.8-fold) or cholate (2.5-fold), whereas deoxycholate had no effect. In the ΔramR mutant (where ramA is overexpressed by 4–8-fold, Figure 2A), the tested bile salts did not further decrease the hilA transcript levels (Figure 2B). Possibly the ramA-mediated repression of hilA is saturated in the ΔramR background, explaining why further increase of ramA expression does not result in further hilA repression.

Irrespective of the tested bile salt, complementation using a pramR multicopy plasmid resulted in hilA transcript levels higher (for chenodeoxycholate and cholate) or similar (for deoxycholate) than those observed in the WT strain (Figure 2B). These results suggest that hilA downregulation by chenodeoxycholate and cholate is dependent on ramR, and via the upregulation of ramA. This observation is congruent with our previous study on the crystal structure of RamR and its interaction with bile acids (Yamasaki et al., 2019). Both cholic and chenodeoxycholic acids, but not deoxycholic acid, were indeed shown to bind to RamR, and more precisely through four hydrogen bonds with RamR, and to induce ramA expression.

Of note, crude bile extract decreased hilA expression to undetectable levels as well in the ΔramR mutant as in the WT strain (Figure 2B). Although it is below the detection level in our experimental conditions, it may suggest that another ramR-independent mechanism, possibly induced by bile components other than bile salts, acts also in the bile-mediated repression of hilA.

Altogether, the results described above suggested a possible involvement of the ramRA locus in the bile-mediated repression of invasion, partly through RamA overexpression, resulting in repression of the master regulator hilA of TTSS-1 genes. To further confirm this hypothesis, the role of RamA in bile-mediated repression of invasion was tested in gentamicin protection assays, using a S. Typhimurium ATCC 14028s mutant overexpressing RamA (ΔramRA/pramA). First, invasion efficiency was confirmed to be decreased about 5-fold by bile in the WT strain (Figure 3, WT). In absence of bile (LB medium alone), ectopic overexpression of RamA in the mutant ΔramRA/pramA resulted in a slight increase of adhesion and in a 2-fold decrease of invasion. Using this mutant, bile further decreased invasion efficiency of an additional 5-fold (Figure 3, ΔramRA/pramA). These results strengthened the hypothesis that the repression of invasion occurs via different pathways dependent or not on RamA overexpression.

In sum, these results suggest that the bile-induced repression of invasion may be driven by a ramRA-dependent mechanism, via ramA overexpression, as well as through other pathways independent of the ramRA regulators which can alternatively have an additive effect to repress invasion. Nevertheless, the observed results must be taken with caution, because of the experimental conditions (e.g., ectopic overexpression) and the use of a single S. Typhimurium strain. At least, ramRA alone seems not entirely responsible for bile-mediated repression of S. Typhimurium cell invasion, and thus other genetic factors need to be further investigated.