The novel Listeria monocytogenes bile sensor BrtA controls expression of the cholic acid efflux pump MdrT


E-mail; Tel. (+1) 773 834 1781; Fax (+1) 773 834 8150.


Mammalian bile has potent anti-microbial activity, yet bacterial pathogens of the gastrointestinal tract and hepatobiliary system nonetheless persist and replicate within bile-rich environments. Listeria monocytogenes, a Gram-positive pathogen, encounters bile at three stages throughout its infectious cycle in vivo: in the gut during initial infection, in the liver where it replicates robustly and in the gallbladder, from which it can return to the intestine and thence to the environment. The mechanisms by which L. monocytogenes senses mammalian bile and counteracts its bactericidal effects are not fully understood. In this report, we have determined the L. monocytogenes bile-induced transcriptome, finding that many critical virulence factors are regulated by bile. Among these, the multidrug efflux pumps MdrM and MdrT, previously shown to be critical for the bacterial provocation of a pathogenesis-promoting host innate immune response, are robustly and specifically induced by the bile component cholic acid. This induction is mediated by BrtA, the first identified L. monocytogenes sensor of bile, which loses the ability to bind to and repress the mdrT promoter in the presence of cholic acid. We show that MdrT can export cholic acid, and that ΔmdrT bacteria are significantly attenuated both in vitro when exposed to cholic acid or bile, and in vivo in the gallbladders and livers of infected mice.


All successful bacterial pathogens of the mammalian gastrointestinal tract (GIT) and hepatobiliary system must have mechanisms for surviving exposure to bile. Bile is synthesized in the liver, and then excreted to the intestine, with a fraction being diverted to the gallbladder as a reservoir for future use (Hofmann, 1998). In addition to its role in emulsification of dietary fats, bile has many properties or components that potently inhibit microbial growth (Hofmann, 1999; Begley et al., 2005a). Bile acids, which are sterol acids derived from cholesterol, can disaggregate lipid bilayers, and thus solubilize bacterial membranes (Merritt and Donaldson, 2009). While Gram-positive bacteria are particularly sensitive (MacConkey, 1908; Tatum, 1916), bile acids can also kill Gram-negative bacteria (Drew et al., 2008; Merritt and Donaldson, 2009). Bacteria have been coevolving with mammalian hosts for millennia, however, and many enteric bacteria have evolved sophisticated mechanisms for detoxifying bile acids or otherwise reducing their toxic effects (Thanassi et al., 1997; van Velkinburgh and Gunn, 1999; Dussurget et al., 2002; Lin et al., 2002; Begley et al., 2005a).

In addition, bacterial pathogens can take advantage of the unique environmental information provided by bile, in order to co-ordinate their virulence strategies. Whereas elevated temperature can indicate residence within an animal host (Hoe and Goguen, 1993; Johansson et al., 2002), certain pathogens can more precisely ascertain their location within the GIT by sensing bile. For example, expression of virulence determinants in Vibrio cholerae, Vibrio hepatolytica and Salmonella enterica serovar Typhimurium is regulated by exposure to bile (Osawa and Yamai, 1996; Gupta and Chowdhury, 1997; Prouty and Gunn, 2000). Bile acids undergo modifications during their passage through the hepatobiliary system and GIT, including amino acid conjugation in the liver, and deconjugation and dehydroxylation by commensal bacteria in the intestine (Hofmann, 1999; Ridlon et al., 2006). Thus, bile acids are attractive candidate molecules for bacterial sensing of bile, as the composition and concentration of bile acids encodes location-specific information. Indeed, MarR- and TetR-like proteins that sense bile acids have been identified in multiple bacterial species, including Salmonella enterica Typhimurium, Campylobacter jejuni and V. cholerae, with bile acid-induced changes in protein dimer conformation leading to de-repression of target genes (Prouty et al., 2004; Lin et al., 2005; Cerda-Maira et al., 2008).

Listeria monocytogenes is a facultative, intracellular pathogen causing significant mortality and morbidity, particularly among neonates and the elderly, and causing abortion in pregnant women (Vazquez-Boland et al., 2001). L. monocytogenes is a food-borne pathogen, and during infection initially resides within the GIT (Portnoy et al., 1992), but, following invasion of the intestinal epithelium, the bacteria disseminate rapidly to the liver and spleen (Gaillard et al., 1991). Recently, the gallbladder lumen has been identified as an additional site of L. monocytogenes infection, with the bacteria existing extracellularly amid concentrated bile (Hardy et al., 2004; Dowd et al., 2011). Thus, the ability to persist and replicate within bile-rich environments is critical for pathogenesis of L. monocytogenes. Multiple L. monocytogenes genes have been identified that contribute to survival in bile, including bile salt hydrolases that metabolize bile to potentially less toxic forms, and bile exclusion systems that expel bile (Dussurget et al., 2002; Begley et al., 2003; 2005b; Sleator et al., 2005; Watson et al., 2009; Dowd et al., 2011). However, no studies to date have yet identified the mechanisms by which L. monocytogenes senses bile, whether this bacterium utilizes the location-specific information encoded in bile to regulate virulence factor expression, or bacterial genes contributing to gallbladder colonization in vivo.

In this report we have comprehensively determined the transcriptional response of L. monocytogenes to mammalian bile, using DNA microarrays. We find that many of the PrfA-regulated virulence factors, required for intracellular invasion and replication, are regulated by bile. Additionally, we find that two multidrug resistance (MDR) efflux pumps of the major facilitator superfamily, MdrM and MdrT, are strongly induced by bile, and more specifically by the bile component cholic acid. This induction is mediated by a transcriptional regulator of the TetR family, which has been previously been shown to be important for L. monocytogenes virulence in a mouse infection model (Crimmins et al., 2008). Here, we establish TetR as the first identified L. monocytogenes bile sensor, with TetR binding to the mdrT promoter significantly diminished in the presence of cholic acid, resulting in de-repression of mdrT expression. Accordingly, we propose that this protein be renamed from TetR to bile-regulated transcription factor A (BrtA), to reflect this newly discovered activity.

These MDRs have recently been demonstrated in L. monocytogenes to transport cyclic di-AMP (c-di-AMP), a small non-DNA nucleic acid capable of eliciting robust host innate immune responses during intracellular infection (Crimmins et al., 2008; Woodward et al., 2010). While MDR efflux pumps can often transport multiple different substrates, these pumps usually transport exogenous, rather than endogenous molecules (Van Bambeke et al., 2000), suggesting that they may have other physiologically relevant substrates. However, other functions and substrates of MdrM and MdrT have not been previously identified. Here, we demonstrate that MdrT is an efflux pump for the host bile acid cholic acid. Additionally, we show a central role for MdrT, and a synergistic role for MdrM, in L. monocytogenes survival to cholic acid and bile in vitro, and in bacterial colonization of the liver and gallbladder in a mouse i.v. infection model.


Determination of the L. monocytogenes bile-induced transcriptome

To initiate a dissection of the bacterial response to conditions encountered during infection of the GIT and hepatobiliary system, we first determined the L. monocytogenes transcriptional response to mammalian bile in broth culture. Bacteria were exposed to 1% porcine bile for 30 min, and the resulting transcriptome was determined using DNA microarrays. Dual criteria were used to establish the L. monocytogenes bile-induced transcriptome: (i) genes must be statistically differentially expressed during bile exposure, as determined by Statistical Analysis of Microarrays (SAM) (Tusher et al., 2001) with a false discovery rate of 1%, with (ii) at least a twofold change in expression. By these criteria, we find that 391 genes, or over 13% of all L. monocytogenes genes, are regulated by bile (Table S1). Genes encoding transport/binding proteins and lipoproteins and metabolism of carbohydrates are highly represented in this transcriptome, indicating a significant change in cellular membranes and metabolism (Table 1). Additionally, genes involved in RNA synthesis are highly represented, indicating a robust transcriptional response to bile. With significant but partial overlap to the transcriptomes controlled by either the stress-specific sigma factor σB (Raengpradub et al., 2008) or the major L. monocytogenes virulence-specific transcription factor PrfA (Milohanic et al., 2003) (21% and 7% overlap, respectively), the response to bile is likely controlled by stress sensors and transcriptional activators in addition to these two previously described pathways.

Table 1.  Description of the L. monocytogenes bile-induced transcriptome.
FC #Functional Category (FC)a% of bile-induced transcriptome# of bile-regulated genes in this FC% of bile-regulated genes in this FC also regulated by PrfAb% of bile-regulated genes in this FC also regulated by Sigma Bc
  • a. 

    Functional categories as determined by ListiList ( for the corresponding gene in strain EGDe.

  • b. 

    PrfA regulation determined by Milohanic et al. (2003).

  • c. 

    Sigma B regulation determined by Raengpradub et al. (2008). For full array data, see Table S1. Data are from biological replicates (N = 3).

  • Dual criteria were applied for inclusion in the LM bile-induced transcriptome: (1) bile-induced changes in expression must be significant, as determined by SAM, with false discovery rate of 1%, and (2) genes be induced or repressed ≥ twofold.

1.1Cell envelope and cellular processes (cell wall)2.811027
1.2Cell envelope and cellular processes (transport/binding proteins and lipoproteins)20.5801019
1.3Cell envelope and cellular processes (sensors /signal transduction)0.3100
1.4Cell envelope and cellular processes (membrane bioenergetics)2.610020
1.5Cell envelope and cellular processes (motility and chemotaxis)0.83033
1.7Cell envelope and cellular processes (cell division)1.0400
1.8Cell envelope and cellular processes (cell surface proteins)2.082563
2.1Intermediary metabolism (metabolism of carbohydrates and related molecules)12.047930
2.2Intermediary metabolism (metabolism of amino acids and nucleic acids)4.317629
2.3Intermediary metabolism (metabolism of nucleotides and nucleic acids)1.87014
2.4Intermediary metabolism (metabolism of lipids)2.610200
2.5Intermediary metabolism (metabolism of coenzymes and prosthetic groups)1.561750
3.2Information pathways (DNA restriction/modification and repair)1.0400
3.3Information pathways (DNA recombination)0.5200
3.5Information pathways (RNA synthesis)9.035311
3.6Information pathways (RNA modification)0.3100
3.7Information pathways (protein synthesis)0.3100
3.8Information pathways (protein modification0.3100
4.1Other functions (adaptation to atypical conditions)1.561767
4.2Other functions (detoxification)1.352040
4.3Other functions (phage-related functions)1.3500
4.5Other functions (miscellaneous)2.082525
5.1Similar to unknown proteins (from Listeria)2.08013
5.2Similar to unknown proteins (from other organisms)20.580521
6No similarity7.931310

Interestingly, virtually the entire PrfA regulon, including many virulence factors critical for intracellular invasion and replication (Portnoy et al., 1992), was repressed in response to bile under these experimental conditions (Table 2). These results were confirmed with real-time quantitative PCR (RT-qPCR) analysis for three representative PrfA-regulated genes, as we have observed that nucleotide differences between strains of L. monocytogenes can result in gene-specific under- and over-estimation of expression changes using microarrays. Among the PrfA regulon genes, haemolysin (hly), encoding the pore-forming toxin used by the bacterium to escape the phagolysosomal compartment during intracellular infection (Portnoy et al., 1988), was among the most repressed genes in the transcriptome. PrfA-regulated genes lying outside the pathogenicity island were also downregulated, including Internalins A and B, which are utilized by the bacterium for invasion of the gut epithelium and liver, respectively (Gaillard et al., 1991; Engelbrecht et al., 1996), as well as for crossing the fetoplacental barrier (Disson et al., 2008). Transcriptional of PrfA itself, however, was not repressed, indicating that bile-mediated repression of PrfA regulon genes may proceed by a mechanism other than negative regulation of PrfA expression. As PrfA activity can be modulated by various small molecules (Milenbachs et al., 1997; Stoll et al., 2008), it is possible that constituents of bile negatively regulate PrfA activity.

Table 2.  Comparison of microarray- and RT-qPCR-determined bile-induced expression changes for select L. monocytogenes virulence factors.
GroupORF DesignationChange in expression by microarrayChange in expression by RT-qPCRHomolog/NameAssignment
PrfA-regulated genesLMO02001.6 PrfAPositive regulatory factor A
LMO0201−3.1 PlcAPhosphatidylinositol-specific Phospholipase
LMO0202−11.3−14.6LLO (Hly)Listerolysin O/Hemolysin
LMO0203−6.0 MplZinc metalloproteinase precursor
LMO0204−2.5 ActAActin-assembly inducing protein
LMO0205−3.4 PlcBPhospholipase C
LMO0433−3.7−8.0InlAInternalin A
LMO0434−7.8−7.2InlBInternalin B
Lipoteichoic acid modificationLMO0974−7.1 DltAD-alanine-activating enzyme
LMO0973−5.0 DltBD-alanine esterification protein
LMO0972−8.1−7.0DltCD-alanyl carrier protein
LMO0971−8.9−6.4DltDD-alanine esterification protein
Efflux pumps (and cognate regulators)LMO16173.05.1MdrMMFS-family efflux pump
LMO16182.1 MarRMarR-family transcription factor
LMO25883.032.9MdrTMFS-family efflux pump
LMO25894.9 TetRTetR-family transcription factor

Interestingly, significant induction was observed for genes encoding the efflux pumps mdrM and mdrT, which transport the bacterial nucleic acid c-di-AMP to the cytosol of infected cells (Crimmins et al., 2008; Woodward et al., 2010). Transcriptional induction of MDRs often indicates the presence of a substrate, as the substrate will also bind a pump-specific transcriptional repressor, leading to the transcriptional de-repression of both the pump and the regulator (Grkovic et al., 2002; Schumacher and Brennan, 2002). C-di-AMP levels have no measurable effect on transcription of mdrM and mdrT (J.J. Woodward, C.E. Witte and D.A. Portnoy, pers. comm.), suggesting that it is likely a promiscuous substrate of these MDRs. In contrast, in response to bile, we noted the robust transcriptional induction of mdrM, mdrT, and their cognate transcriptional regulators, marR and brtA (previously tetR) respectively. Thus, we hypothesized that some component of mammalian bile might be the natural substrate for MdrM and MdrT, and that MarR and/or BrtA might be L. monocytogenes sensors of bile. We sought to determine these substrates, the mechanism by which these pumps are regulated, and their role in L. monocytogenes pathogenesis.

Cholic acid is a potent inducer of mdrT and mdrM transcription

Defined constituents of bile, and physiological properties of bile, were individually assayed for their capacity to transcriptionally induce mdrM and mdrT. Tested components and properties included cholic acid, cholesterol, high osmolarity and detergent. Of all the conditions tested, only the bile component cholic acid was able to elicit transcriptional induction of both mdrT (Fig. 1A) and mdrM (Fig. 1B), and this induction was comparable with that observed with bile treatment. To test if the induction of the MDRs was specific to cholic acid, or was an activity of bile acids in general, seven bile acids were individually assayed. Bile acids common to mammalian bile (Ridlon et al., 2006; Hofmann and Hagey, 2008) were tested, including glycocholic acid (GCA), glycodeoxycholic acid (GDCA), deoxycholic acid (DCA), glycochenodeoxycholic acid (GCDCA), chenodeoxycholic acid (CDCA), lithocholic acid (LCA), and taurocholic acid (TCA). We found that only cholic acid was capable of inducing robust MDR expression (Fig. 1C and D). Deoxycholic acid, which differs from cholic acid by only the removal of a single hydroxyl group, exhibited no MDR-inducing capacity, and glycocholic acid, the glycine-conjugated form of cholic acid, was similarly inert. The induction of the MDRs was unlikely to be in response to membrane damage, as deoxycholic acid is significantly more potent at solubilizing bacterial membranes than is cholic acid (Stacey and Webb, 1947), yet did not induce mdrT and mdrM.

Figure 1.

Cholic acid transcriptionally induces L. monocytogenes mdrT and mdrM.
A and B. WT bacteria were treated as indicated with the following: bile (1% porcine bile), cholic acid (10 mM), high osmolarity (250 mM sucrose), detergent (0.1% Triton) or cholesterol (1 mM). The induction of mdrT (A) and mdrM (B) was assayed by RT-qPCR. MDR induction is relative to expression in bacteria grown in BHI, and all values are normalized to 16S RNA. Data are the average of biological replicates (n = 2), with error bars depicting the standard error of the mean. The asterisk indicates a value statistically different from that of cells grown in BHI, with a P-value ≤ 0.05 (see Experimental procedures).
C and D. As in A and B, above, but cells were treated with 10 mM of the indicated bile acid for 30 min. GCA, glycocholic acid; GDCA, glycodeoxycholic acid; CA, cholic acid; DCA, deoxycholic acid; GCDCA, glycodeoxycholic acid; CDCA, chenodeoxycholic acid; LCA, lithocholic acid; TCA, taurocholic acid. Data are the average of biological replicates (n = 2), with error bars depicting the standard error of the mean. The asterisk indicates a value statistically different from that of cells grown in BHI, with a P-value ≤ 0.05.

Transcriptional regulation of mdrT and mdrM by cholic acid is BrtA-dependent

To confirm that this induction reflected a specific substrate-efflux pump-transcription factor regulatory network, rather than a general stress response, we investigated the involvement of BrtA and MarR in cholic acid-induced expression of the MDRs. Bacteria lacking BrtA (brtA−) are constitutively de-repressed for mdrT expression, and ΔmarR bacteria are similarly de-repressed for mdrM expression (Crimmins et al., 2008); thus, we tested if mdrT and mdrM could be further induced in response to cholic acid in brtA− and ΔmarR cells. No further increase in mdrT expression was observed in brtA− bacteria in response to cholic acid (Fig. 2A), consistent with the hypothesis that cholic acid regulates mdrT expression through BrtA. The slight reduction in mdrT expression in brtA− bacteria observed during exposure to cholic acid was consistently observed, and is of unknown cause. One possibility is that the mdrT promoter is additionally regulated by additional components of bile, by a BrtA-independent mechanism.

Figure 2.

BrtA is required for the cholic acid-induced expression of both mdrT and mdrM.
A–D. The indicated strains were treated with 10 mM cholic acid for 30 min. The induction of mdrT (A and C) and mdrM (B and D) was assayed by RT-qPCR analysis. Data are the average of biological replicates (n = 2). The asterisk indicates a value statistically different from that of cells grown in BHI, with a P-value ≤ 0.05.

As expected, the induction of mdrT in response to cholic acid was independent of MarR, the mdrM-specific transcriptional regulator. Surprisingly, while ΔmarR bacteria exhibited no further increase in mdrM expression in response to cholic acid, cholic acid-induced expression of mdrM was BrtA-dependent (Fig. 2B). This cross-talk between the regulation of mdrT and mdrM expression was dependent on MarR, as brtAΔmarR bacteria were still constitutively de-repressed for mdrM expression. Regulation of mdrT and mdrM by cholic acid was also independent of the stress-specific alternate sigma factor, σB, as cells lacking σBsigB) were still able to induce the MDRs under these conditions (Fig. 2C and D).

Cholic acid disrupts BrtA binding to the mdrT promoter

To investigate the mechanism by which cholic acid was able to transcriptionally induce mdrT expression in a BrtA-dependent manner, recombinant BrtA protein was expressed and purified. Purified protein was incubated with a fluorescently labelled DNA probe composed of the 100 base pairs directly upstream of the mdrT initiator ATG codon (Fig. 3). At increasing concentrations of protein, higher molecular weight species were observed, indicating binding of BrtA to the mdrT promoter (lanes 2–5). Protein affinity for this DNA was specific, as it could be competed away by an excess of unlabelled but otherwise identical DNA (lane 6), but not by DNA of the same GC content with scrambled sequence (lane 7). The interaction between BrtA and the mdrT promoter was disrupted by the addition of 10 mM cholic acid (compare formation of DNA-protein complexes in lanes 2–5 with lanes 8–11). This effect of cholic acid on BrtA binding to the mdrT promoter was specific, as the binding of an unrelated DNA binding protein from Brucella abortus to its target DNA was largely unaffected by the addition of cholic acid (compare lanes 13–17 with 18–22).

Figure 3.

BrtA affinity for the mdrT promoter is disrupted by cholic acid. The indicated amount of purified BrtA protein was incubated with a fluorescent mdrT promoter DNA probe, and additionally, where indicated, a specific (S) or non-specific DNA competitor (NS) (lanes 1–12). Cholic acid was added to the binding reaction (lanes 8–12) to disrupt the BrtA–DNA interaction. An unrelated B. abortus DNA binding protein (‘Ba protein’) was incubated with its cognate DNA target sequence (‘Ba probe’) with (lanes 13–17) or without (lanes 18–22) cholic acid, as a control (see Experimental procedures).

MdrT transports cholic acid

Given the transcriptional induction of mdrT and mdrM by cholic acid, we hypothesized that cholic acid might be an efflux substrate for either or both of these pumps. The capacity of the MDRs to transport cholic acid was determined using an in vitro assay with a 14C-radiolabelled cholic acid substrate, similar to previous MDR efflux assays (Thanassi et al., 1997; Sleator et al., 2005). Log-phase wild type (WT), ΔmdrT and ΔmdrM cells were incubated with 14C-cholic acid (uptake period), and then diluted into buffer containing an excess of non-labelled cholic acid (efflux period). At various time points post-uptake, cells were separated from buffer, and the cell-associated radiolabel was determined using scintillation counting. While initial 14C-cholic acid uptake was identical for all strains tested (6037 ± 7 CPM for WT, 6015 ± 17 CPM for ΔmdrT), throughout the efflux period ΔmdrT L. monocytogenes consistently and significantly retained more radiolabel than WT cells (Fig. 4). The difference in WT and ΔmdrT 14C-cholic acid retention was constant over time, similar to what has been previously reported for the transport of 14C-chenodeoxycholic acid by the L. monocytogenes MDR BilE (Sleator et al., 2005). The reason for increasing cell-associated radiolabel during the efflux period, after which cells have been diluted in excess non-labelled cholic acid, is unclear, but may reflect continued incorporation of loosely associated 14C-cholic acid during the efflux period after dilution. Interestingly, we were unable to demonstrate any role for MdrM in transport of cholic acid under these experimental conditions (data not shown).

Figure 4.

MdrT exports cholic acid. Cells of the indicated genotype were incubated for 30 min with 1 µCi 14C-cholic acid, and then diluted into an excess of buffer containing unlabelled cholic acid. At the indicated times, cells were spun through silicone oil, and the cell pellets assayed for retained radiolabel by liquid scintillation counting. Data are the average of biological replicates (n = 3). The asterisk indicates a value statistically different between the WT and ΔmdrT samples, with a P-value ≤ 0.05.

Cholic acid is toxic to L. monocytogenes lacking MdrT in vitro

We next tested if MdrT was required for L. monocytogenes viability in the presence of bile and cholic acid, as bile acids can damage both cellular membranes and nucleic acids (Bernstein et al., 2005; Merritt and Donaldson, 2009). In a plate-based assay measuring growth over 24 h, WT, ΔmdrM, ΔmdrT and ΔmdrM ΔmdrT strains grew identically under non-selective conditions (Fig. 5A). Addition of either 1% porcine bile or 10 mM cholic acid resulted in markedly diminished growth for the ΔmdrT and ΔmdrM ΔmdrT strains, while WT and ΔmdrM growth was not significantly impaired. While no inhibition was observed upon addition of 1 mM cholic acid at neutral pH, at pH 6.0 the growth of the ΔmdrT and ΔmdrM ΔmdrT strains was specifically inhibited. The pKa of cholic acid is ∼ 5.2, and at lower pH the anti-microbial effects of cholic acid are likely to be enhanced (Begley et al., 2005a).

Figure 5.

MdrT is required for survival of cholic acid stress.
A. Stationary-phase cells of the indicated genotypes were serially diluted and then spotted on BHI agar with the indicated supplements. Colony growth was visualized after 24 h at 37°C.
B. Stationary-phase cells of the indicated genotypes were back-diluted to an OD600 = 0.025 into BHI media supplemented with varying concentrations of cholic acid (CA) or deoxycholic acid (DCA). Plotted are the fastest doubling time observed for each strain under each condition, over a 7 h growth period. Expression of mdrT from the p(mdrT) complementing plasmid was induced by the addition of 1 mM IPTG. Data are the average of biological replicates (n = 3).

Similar results were observed in a liquid growth assay, measuring the doubling time of WT and ΔmdrT strains across a range of cholic acid concentrations (Fig. 5B). With as little as 1 mM cholic acid, the ΔmdrT strain exhibited growth rates significantly lower than WT bacteria, with the defect becoming more pronounced at higher cholic acid concentrations. No growth was observed for the ΔmdrT strain above 8 mM cholic acid, while WT continued to grow up to 16 mM cholic acid. The defect of the ΔmdrT strain was specific to cholic acid, as WT and ΔmdrT strains displayed identical growth kinetics in deoxycholic acid. Complementation of the cholic acid-specific ΔmdrT growth defect was observed in a strain containing IPTG-inducible MdrT. Incomplete restoration of growth by this complementing plasmid, at cholic acid concentrations above 4 mM cholic acid, was perhaps attributable to sub-WT expression of mdrT from this construct, as has been observed previously (Crimmins et al., 2008).

To determine the basis for this growth defect, electron microscopy was used to examine the cellular structure of L. monocytogenes strains exposed to cholic acid. Both WT and ΔmdrT bacteria had identical morphologies and membrane structures when grown on BHI (Fig. 6, panels 1–4). During exposure to cholic acid, the membranes of WT bacteria appeared relatively unperturbed, displaying only a modest increase in thickness and electron density (panels 5 and 6). In contrast, multiple lesions were observed in the membranes of ΔmdrT cells grown in the presence of cholic acid, with what appears to be portions of membrane or cellular content leaking out (panels 7 and 8; arrowheads mark lesions). Such lesions were never observed in WT cells. In addition, under these conditions ΔmdrT cells were often substantially elongated, compared with WT cells (note relative cell length in panels 5 and 7).

Figure 6.

Bacteria lacking MdrT exhibit membrane lesions upon exposure to cholic acid. WT (panels 1–2, 5–6) and ΔmdrT (panels 3–4, 7–8) cells were grown in BHI media (panels 1–4) or BHI media supplemented with 10 mM cholic acid (panels 5–8) for 16 h and analysed by transmission electron microscopy. Even-numbered panels are magnified images of the corresponding previous odd-numbered panels, with black boxes indicating the magnified region. Scale is as indicated in each panel. Arrows in panel 8 denote observed membrane lesions.

MDRs are required for virulence in vivo

To examine the role of MdrM and MdrT in bile tolerance in vivo, multiple mouse models of infection were utilized. An intra-gastric infection model was utilized to examine the role of MdrM and MdrT in the GIT (Fig. 7A). Mutations in mouse E-Cadherin restrict L. monocytogenes Internalin A-mediated escape from the GIT (Lecuit et al., 1999; 2001), and hence limit the utility of the mouse oral model for examining disseminated disease. However, bacterial survival within the GIT is presumed to be unaffected by this change in mouse E-Cadherin, and L. monocytogenes survival of GIT conditions is directly tested in this infection model. L. monocytogenes GIT tract survival was assayed by enumerating bacterial CFU in the faeces of infected mice over a 3 day period, with no difference observed among WT and mutant strains. Bile acids are significantly more bactericidal under acidic conditions such as those found in the gut, as the pH approaches the pKa of certain bile acids (Hofmann, 1999; Begley et al., 2005b). However, the rapid metabolism of cholic acid by commensal microbiota may limit its abundance in the GIT (Begley et al., 2005a; Jones et al., 2008).

Figure 7.

L. monocytogenes MdrM and MdrT are required for virulence in vivo.
A. C57BL/6 mice were infected intra-gastrically with 1 × 1010 CFU of the indicated strains. Bacterial CFU in the faeces were enumerated on the indicated days post infection. Five mice were used for each experiment.
B–D. C57BL/6 mice were infected retro-orbitally with 1 × 104 CFU of the indicated strains. Bacterial CFU in the liver (A), spleen (B), and gallbladder (C) were enumerated 4 days post infection. Five mice were used for each experiment, and the solid black bar represents the mean.

In an intravenous infection model, bacterial abundance within organs was determined 4 days post infection. Bacterial replication within the liver was similar for WT and ΔmdrM bacteria, while ΔmdrT cells were approximately 10-fold attenuated, and ΔmdrM ΔmdrT bacteria 100-fold attenuated, suggesting a synergistic requirement for both MdrM and MdrT (Fig. 7B). In the spleen, WT, ΔmdrM and ΔmdrT bacteria replicated to similar levels, while ΔmdrM ΔmdrT bacteria exhibited an approximately 10-fold defect in growth (Fig. 7C). As the spleen is not enriched in bile, the observed defect in growth for the ΔmdrM ΔmdrT bacteria in this organ may reflect decreased trafficking of ΔmdrM ΔmdrT bacteria from the liver. In the gallbladder, the ΔmdrT and ΔmdrM ΔmdrT strains were 100-fold attenuated, suggesting that only MdrT, and not MdrM, is required for maximal bacterial virulence in this organ (Fig. 7D).


In this study, we have determined the L. monocytogenes transcriptional response to mammalian bile. As this pathogen infects the GIT and hepatobiliary system, it is certain that bile constitutes an important component of the host–pathogen interaction. We observed that over 10% of all L. monocytogenes genes were significantly and robustly regulated by bile, indicating the importance of bile to the bacterial life cycle. As with other enteric pathogens, many critical virulence factors were regulated by bile (Osawa and Yamai, 1996; Gupta and Chowdhury, 1997; Prouty and Gunn, 2000), including many PrfA-regulated virulence factors. Among these, the bile-mediated repression of Internalins A and B is reminiscent of the decreased invasiveness of S. enterica in response to bile (Prouty and Gunn, 2000). In S. enterica, it has been proposed that this might allow the bacterium to initiate invasion only after successful penetration of the mucous lining of the GIT, where bile concentrations are lower, rather than while still in the bile-rich lumen (Prouty and Gunn, 2000). It is possible that L. monocytogenes employs a similar mechanism of bile-mediated regulation of invasion.

The overlap between the bile-induced and PrfA- and σB-regulated transcriptomes suggests that there exists a cohort of L. monocytogenes proteins induced by and required for survival of multiple stresses. However, the majority of bile-regulated genes have not been described as under the control of either PrfA or σB, indicating that a novel bacterial sensory and signal transduction mechanism exists to respond to mammalian bile.

Consistent with this hypothesis, we have identified BrtA as the first described L. monocytogenes bile sensor. Among BrtA target genes, we have identified the multidrug efflux pump-encoding mdrT as potently induced by cholic acid. Bile acid-responsive transcriptional regulators have been identified in multiple bacterial species (Prouty et al., 2004; Lin et al., 2005; Cerda-Maira et al., 2008), and, as in these other systems, exposure of BrtA to cholic acid results in an inhibition of BrtA binding to the mdrT promoter. While we favour the hypothesis that BrtA directly binds to cholic acid, as has been demonstrated for C. jejuni CmeR (Lei et al., 2011), it is possible that BrtA DNA binding activity is instead attenuated indirectly. For instance, the non-denaturing detergent properties of cholic acid might perturb BrtA structure or activity. Arguing against this alternate non-direct binding mode of action, we have found that the DNA binding activity of an unrelated DNA binding protein from B. abortus is unaffected by cholic acid. Identification of BrtA amino acid residues required for regulation of activity by cholic acid, and precise determination of binding constants, will allow us to dissect the interaction between BrtA and its ligands.

As MDR efflux pumps often promiscuously transport a variety of molecules, it is difficult to determine the bona fide substrate of an MDR, i.e. the compound whose export gave initial selective advantage to those bacteria that had the pump (Van Bambeke et al., 2000; Lewinson et al., 2006). While MDR efflux pumps are most commonly associated with bacterial resistance to antibiotics (Van Bambeke et al., 2000), MDRs have been shown to be crucial for bacterial survival in bile, including AcrAB in E. coli (Thanassi et al., 1997; Drew et al., 2008), CmeABC in C. jejuni (Lin et al., 2002) and BilE in L. monocytogenes (Sleator et al., 2005). BilE, which exports chenodeoxycholic acid, has additionally been shown to be a critical virulence factor for L. monocytogenes in vivo. However, its expression, while dependent on PrfA and σB (Sleator et al., 2005), is not responsive to bile (this study, Table S1). We contend that cholic acid is a ‘natural’ substrate of MdrT, as our data fulfil three major criteria to support this designation. First, cholic acid potently activates the transcriptional induction of mdrT. The activation of mdrT expression by cholic acid is specific, as a variety of extremely structurally similar bile acids exhibit little or no mdrT-inducing activity under the experimental conditions tested. Second, in an in vitro efflux assay, we have demonstrated that MdrT is required for cholic acid export. Third, ΔmdrT bacteria are attenuated for growth when challenged with cholic acid in vitro, and electron microscopic examination of mdrT cells grown in the presence of cholic acid reveals membrane abnormalities. In total, these data argue that cholic acid is a biologically relevant substrate for MdrT.

We have demonstrated that MdrT is required for L. monocytogenes virulence in vivo, most significantly in the bile-rich environments of the liver and gallbladder. A previous report (Crimmins et al., 2008) has described a liver-specific defect of the brtA- L. monocytogenes strain, independent of the role of BrtA in delivering innate immune-stimulating bacterial ligands, highlighting the importance of regulated MDR expression. Interestingly, while the ΔmdrT strain was ∼ 10-fold attenuated in the liver, the double ΔmdrM ΔmdrT strain was 100-fold attenuated. Bile acids can exhibit synergistic bactericidal effects (Begley et al., 2002), and the ΔmdrM ΔmdrT bacteria may be sensitive to low concentrations of a mixture of bile acids present in the liver. While consistent with the notion of a synergistic interaction between MdrM- and MdrT-mediated efflux, the substrate(s) of MdrM are as yet unidentified (see further discussion below).

While bile acids can reach 300 mM in the lumen of the gallbladder, they are significantly lower in the liver (Hofmann, 1999). Furthermore, within the liver a significant fraction of cholic acid is quickly converted to glyco- and tauro-conjugated species, which, in our in vitro experiments, are insufficient to induce the transcription of mdrT or mdrM. It is possible that L. monocytogenes encounters cholic acid at higher concentrations as a result of bacterial conversion of these conjugated bile acids to cholic acid. L. monocytogenes encodes a bile salt hydrolase (bsh), absent from the genome of non-pathogenic Listeria innocua (Dussurget et al., 2002), and strains lacking bsh have previously been shown to be defective for liver colonization in a mouse i.v. infection model (Dussurget et al., 2002).

MdrT is the first demonstrated L. monocytogenes virulence factor involved in colonization of the gallbladder in vivo, with a 100-fold defect observed for the ΔmdrT and ΔmdrM ΔmdrT strains. The gallbladder has only recently been identified as a site of significant L. monocytogenes infection (Hardy et al., 2004; 2006; Dowd et al., 2011), with still relatively little known about either the mechanisms by which L. monocytogenes is able to survive in this extreme environment, and the role of gallbladder colonization in L. monocytogenes pathogenesis. As more is learned about this stage of L. monocytogenes infection, it will be important to continue to dissect the role of MdrT.

It is unclear what precise role MarR and MdrM play in the L. monocytogenes response to bile. Given the canonical mechanism by which MDRs are regulated (Grkovic et al., 2002; Schumacher and Brennan, 2002), the cholic acid-induced expression of mdrM would suggest both that MdrM exports cholic acid, and that MarR activity is regulated by cholic acid. However, our data indicate that this simple hypothesis may not be correct, as we find no evidence for a role of MdrM in cholic acid export. Furthermore, while it is possible that L. monocytogenes maintains two efflux pumps with identical activities, it is more likely that MdrM and MdrT have differences in either substrate recognition or regulation, which were not revealed in the current study. For instance, MdrM may transport cholic acid only at precise concentrations, or under specific pH conditions. Alternatively, MdrM may have initially evolved for cholic acid transport, and retains transcriptional regulation by cholic acid, but has diverged to transport some other, related bile acid that is present in vivo but not in our in vitro assays, consistent with the observed synergistic role of MdrM in L. monocytogenes colonization of the liver.

Cholic acid and c-di-AMP, while both small molecules, do not share significant structural features. The ability of L. monocytogenes MdrT to transport two distinct molecules, each of which plays a role in bacterial virulence in vivo, may be coincidental; alternatively, this may reflect a sophisticated virulence strategy. The host innate immune response to cytosolic L. monocytogenes has been shown to paradoxically increase bacterial virulence, most notably in the liver (Auerbuch et al., 2004; Carrero et al., 2004; O'Connell et al., 2004). While bile salts are concentrated in extracellular secreted bile, bile acids are also present within the cytoplasm of hepatocytes, where they will be encountered by L. monocytogenes during intracellular infection (Wilton et al., 1994; Trauner and Boyer, 2003). Thus, L. monocytogenes may use cholic acid as a chemical cue to provide information signifying specific location within the liver, so that the release of c-di-AMP can be co-ordinated with other aspects of L. monocytogenes virulence.

Experimental procedures

Bacterial strains, culture and bile exposure

The WT L. monocytogenes strain used in these studies was 10403S (Bishop and Hinrichs, 1987). Mutants containing in-frame deletions of marR, mdrM and mdrT, and the Tn917 transposon insertion mutant in brtA (tetR) have been described previously (Crimmins et al., 2008). The IPTG-inducible mdrT complementing plasmid used in Fig. 5[‘p(mdrT)’] has been described previously (Crimmins et al., 2008), and was induced by the addition of 1 mM IPTG. The brtA::TN917 ΔmarR strain was created by U153 phage transduction of the brtA::TN917 mutation into the ΔmarR strain. The ΔsigB strain has been described previously (Wiedmann et al., 1998). For broth bile exposure experiments, bacteria were grown in brain heart infusion media (Difco) to OD600 ∼ 0.25, and were exposed to 1% porcine bile (Sigma-Aldrich) for 30 min. Error bars represent the standard error of the mean. For samples exposed to defined bile acids or properties, mid-log bacteria were exposed for 30 min to BHI supplemented with the following, all from Sigma-Aldrich: 10 mM of the indicated bile acid, 250 mM sucrose, 0.1% Triton or 1 mM cholesterol.

Microarray analysis

Microarray analysis was as previously described (Leber et al., 2008). Microarrays were printed at the Microarray Core Facility of the Genome Center at Washington University, using the L. monocytogenes oligo set designed and provided by The Institute for Genomic Research. Bacteria RNA was prepared as above, and amplified as previously described (Leber et al., 2008), except that the Message Amp II – Bacteria kit was used (Applied Biosystems). Statistical analysis was performed using the Statistical Analysis of Microarrays (Stanford) plug-in available for Excel, with the false discovery rate set to 1%. All experiments were performed in triplicate. Microarray data have been deposited with GEO under accession # GSE28507.

Gene expression analysis

For analysis of bacterial gene expression, cells were treated as described, and immediately chilled to 4°C in ice water. Bacteria were then harvested by centrifugation, and RNA was extracted with Trizol (Invitrogen) and FastPrep glass bead/mechanical disruption (MP Biomedical). After treatment with TURBO Dnase (Applied Biosystems), RNA was reverse transcribed with iScript Select using random primers (Bio-Rad). Quantitative PCR was carried out with SsoFast EVAgreen supermix (Bio-Rad) on a CFX96 machine (Bio-Rad) using the following gene-specific primers: 16S RNA: For 5′ ACGGCTAACTACGTGCCAG 3′, Rev 5′ ATGACCCTCCCCGGTTAAG 3′; LMO2588 (mdrT): For 5′ CCCCAACCATCATTACCCGCTGAACTAAATCCGTATAG 3′, Rev 5′ GGTTGGATTGTGGATTCGTATGATTGGCGCG 3′; LMO1617 (mdrM): For 5′ GGTATTTTGATTGTTATGCTTATGG 3′, Rev 5′ TTGTAAATCGTTCAATTAAAAAGGC; hly: For 5′ GGCACCAGCATCTCCGCC 3′, For 5′ GCATTATTTTGATTGATGGATTTCTTCTTTTTCTCC 3′; inlA: For 5′ CAGATTTTTACAGACGCAGCTCTAGCGG 3′, For 5′ GGCGTTATATCCGTAAGTTGATTATTGCTG 3′; inlB: For 5′ GTACAAGCGGAGACTATCACCGTGTCAACGCC 3′, Rev 5′ GTTTATTCCCGTTTAAAAATAACTTTGTCACATTGGG 3′; dltC: For 5′ ATGGCTTTTCGTGAAAATGTATTAGAAATATTAGAGG 3′, Rev 5′ CAGGAGTAGCCCATTCGTCGCGG 3′; dltD: For 5′ GAAAAAAAAGCTGTGGATGACATTTGGGCC 3′, Rev 5′ GCTCTGAAGAACCATAAATTGGTAAATAGTTTCC 3′.

BrtA protein expression and purification

The brtA open reading frame was amplified with primers oJL381 (5′ AAAACATATGAAGGAGAAGAAGCAGCGGATTATTAAGTCGGC 3′) and oJL382 (5′ AAAAGGATCCTCAGTGGTGGTGGTGGTGGTGTAAAGCCCGTTCACACAAGCTTCTAATTTATTTACTTC 3′, adding a carboxy-terminal hexa-histidine tag) and cloned into the pET24b expression vector (Novagen). The protein was expressed in E. coli strain BL21 DE3, and purified using Ni-NTA agarose resin (Qiagen) according to the manufacturer's directions.

Electrophoretic mobility shift assay

Electrophoretic mobility shift assay binding was in a total volume of 20 µl, in a buffer including 20 mM Tris pH 8.0, 50 mM NaCl, 5% glycerol and 1000 ng BSA. The indicated concentration of protein was incubated with 1 ng of DNA. The mdrT DNA probe was created by annealing two reverse-complementary oligonucleotides corresponding to the 100 bp immediately upstream of the mdrT initiator ATG codon, synthesized with a single 5′ IRdye700 molecule (Operon). Where indicated, 200 ng specific competitor DNA (identical to probe ds DNA, but without the IRdye700 molecule) or 200 ng non-specific competitor (the reverse of the specific competitor sequence) was additionally added. The B. abortus protein used as a control is a predicted DNA binding complex (Brucella melitensis biovar abortus chromosome 1, co-ordinates 959245–959559), expressed with a hexa-histidine tag, and was a kind gift of Brook Ragle and Sean Crosson (University of Chicago). The 50 bp cognate B. abortus DNA probe corresponds to the B. melitensis biovar abortus chromosome 1, co-ordinates 959802–959851, and was created by the same method as the mdrT DNA probe. For incubations with cholic acid, 10 mM cholic acid was incorporated in the binding reaction and gel running buffer. Gels were imaged with the Licor system (Odyssey).

L. monocytogenes growth kinetics in bile acid-containing media

Cells were grown in low-form 96-well plates at 37°C with constant agitation, in a Synergy HT plate reader/incubator (BioTek). Cell growth (OD600) was monitored every 30 min, for 7 h. Expression of mdrT from the complementing plasmid was induced by the addition of 1 mM IPTG.

Bile acid efflux assay

Bacteria grown in BHI to an OD600 ∼ 1 were centrifuged, washed in 10 ml Buffer N (25 mM potassium phosphate pH 7.0, 0.2% glucose, 1 mM magnesium sulphate), and resuspended in Buffer N to a concentration of 100 ODU ml−1. One µCi of cholic acid [24-14C] (American Radiolabelled Chemicals) was added, to a final concentration of 18 µM, and cells were incubated at 25°C for 30 min. All strains initially incorporated an identical 1.5 × 106 CPM. Cells were then diluted to 10 ODU ml−1 in Buffer N containing 1 mM non-radiolabelled cholic acid, and incubated at 25°C. At the indicated time, 250 µl cells were layered on top of 500 µl Nyosil-M25 silicone fluid (TAI Lubricants), centrifuged at 15 000 g for 1 min, and flash-frozen in liquid nitrogen. The cell pellets were then excised from the bottom of the tubes and solubilized in 1 ml water, followed by 5 ml of ExoScint XR cocktail (National Diagnostics). CPM determination was made on a Beckman-Coulter liquid scintillation counter. The experiment was performed in triplicate.

Electron microscopy

Bacteria were grown on BHI plates with and without cholic acid overnight at 37°C. Bacteria were fixed in 4% paraformaldehyde in PBS for 10 min, washed twice and suspended in 0.1 M HEPES, pH 7.0. Suspended cells were applied to carbon-coated nickel grids and stained with 1% uranyl acetate. Samples were viewed on a Tecnai F30 electron microscope.

Mouse infections

All animal experiments were carried out in accordance with the regulations set forth by the University of Chicago Institutional Animal Care and Use Committee. For i.v. infections, anesthetized 6–8 week old female C57BL/6 mice (Charles River) were infected retro-orbitally with 1 × 104 CFU bacteria resuspended in 100 µl sterile PBS. Spleens, livers and gallbladders were separately collected 4 days post infection, and bacterial CFU were enumerated by plating organ homogenates on agar plates. For oral infections, 6–8 week old female C57BL/6 mice (Charles River) were infected intra-gastrically with 1 × 1010 CFU bacteria resuspended in 200 µl sterile PBS. Faeces were collected at 24 h intervals, on the days indicated, and bacterial CFU determined as described above.


The Student's t-test (single tail, homoscedastic) on the equality of the means was calculated using Excel (Microsoft). In all figures, ‘*’ represents statistical difference, with P-value ≤ 0.05, compared with the uninfected or untreated controls.


JHL is funded by a New Investigator Award from the University of Chicago Digestive Disease Research Core Center (P30 DK42086). We wish to thank: the electron microscopy facility at the University of Chicago; Chris Sawyer and Michael Heinz (Microarray Core Facility at Washington University) for microarray printing; Charlie Sun and Scott Peterson (J. Craig Venter Institute) for L. monocytogenes array oligos; Dr Hiroshi Nikaido (University of California, Berkeley) for advice regarding the bile acid efflux assay; Brooke Ragle (University of Chicago) for help in animal techniques and electrophoretic mobility shift assay techniques; Ben Kline (University of California Berkeley) for help with phage transductions; and Drs Daniel Portnoy (University of California Berkeley) and Kathryn Boor (Cornell University) for kindly providing strains.