ABSTRACT
Listeria monocytogenes is an environmental saprophyte and intracellular bacterial pathogen. Upon invading mammalian cells, the bacterium senses abrupt changes in its metabolic environment, which are rapidly transduced to regulation of virulence gene expression. To explore the relationship between L. monocytogenes metabolism and virulence, we monitored virulence gene expression dynamics across a library of genetic mutants grown under two metabolic conditions known to activate the virulent state: charcoal-treated rich medium containing glucose-1-phosphate and minimal defined medium containing limiting concentrations of branched-chain amino acids (BCAAs). We identified over 100 distinct mutants that exhibit aberrant virulence gene expression profiles, the majority of which mapped to nonessential metabolic genes. Mutants displayed enhanced, decreased, and early and late virulence gene expression profiles, as well as persistent levels, demonstrating a high plasticity in virulence gene regulation. Among the mutants, one was noteworthy for its particularly low virulence gene expression level and mapped to an X-prolyl aminopeptidase (PepP). We show that this peptidase plays a role in posttranslational activation of the major virulence regulator, PrfA. Specifically, PepP mediates recruitment of PrfA to the cytoplasmic membrane, a step identified as critical for PrfA protein activation. This study establishes a novel step in the complex mechanism of PrfA activation and further highlights the cross regulation of metabolism and virulence.
INTRODUCTION
Upon infection of host cells, bacterial pathogens reside in a confined metabolic environment that is shared with their host. From that moment, the host and the bacteria compete for nutrients and the metabolic state of each continuously changes. These changes serve as metabolic cues, for both the pathogen and the host, to sense the state of infection and to respond accordingly. Listeria monocytogenes, a ubiquitous saprophyte and intercellular pathogen, invades directly the cytosol of mammalian cells; there it replicates efficiently and utilizes host nutrients to support growth (its intracellular growth rate being comparable to that in rich medium) (1, 2). To date, several host-derived metabolic signals had been shown to be sensed by L. monocytogenes and to stimulate virulence gene expression via the regulation of the major virulence activator protein, PrfA. For example, in the host intracellular environment the carbon sources glycerol and hexose phosphates (e.g., glucose-1-phosphate [G1P]) are both utilized by L. monocytogenes and serve as niche-specific signals that report the intracellular location of the bacteria and trigger transcription and activity of PrfA and thereby induce the virulence state (3). These nutrients are imported into L. monocytogenes via GlpF and UhpT, respectively, which are dedicated transport systems and part of the L. monocytogenes arsenal of virulence factors regulated by PrfA (4–7). Conversely, in extracellular niches (e.g., soil and vegetation), carbon sources such as glucose, mannose, and cellobiose are available, which repress prfA transcription and repress the virulent state (4, 5, 8). Similarly, sensing the availability of amino acids within the intracellular niche was also shown to regulate prfA. For example, in response to scarce amounts of isoleucine, leucine, and valine in the host cell cytosol, L. monocytogenes triggers their de novo synthesis (9–11). This metabolic adaptation was shown to be directly linked to the induction of prfA transcription via the action of the global regulator and branched-chain amino acid (BCAA) sensor CodY (10, 12, 13). Additional host-derived signals known to directly regulate PrfA include glutathione, which was recently shown to act as an allosteric activator of PrfA (14), temperature, which was shown to regulate PrfA translation via an RNA-based thermosensor (15), and S-adenosylmethionine, which regulates PrfA via a trans-acting riboswitch (16). Taken together, these studies support the premise that L. monocytogenes senses various metabolic and physical signals within the host and uses this information to regulate virulence gene expression.
To further study the impact of L. monocytogenes metabolism on virulence, we performed two independent genetic screens looking for mutants that exhibit aberrant virulence gene transcription under metabolic growth conditions known to activate the virulent state: glucose-1-phosphate in a charcoal-treated Luria-Bertani–morpholinepropanesulfonic acid (LB-MOPS) medium and minimal defined medium containing limiting concentrations of BCAAs. Notably, we found over 100 mutants that display differential virulence gene expression, many mapping to metabolic genes that alter virulence gene transcription in a multidimensional manner. Further we identified one mutant as an aminopeptidase, PepP, and show that it is critical for PrfA protein activation. Surprisingly, this function requires the localization of PrfA to the cytoplasmic membrane.
RESULTS
In vitro metabolic screens.The aim of this study was to identify genes involved in the cross talk between L. monocytogenes metabolism and virulence. The L. monocytogenes 10403S strain harboring the integrative pPL2 plasmid containing luxABCDE genes under the control of the hly promoter (pPL2-Phly-lux) was used as a parental reporter strain for generation of a Himar1-mariner transposon mutant library (hly encodes listeriolysin O, one of the major L. monocytogenes virulence factors [17, 18]). Mutants were screened individually under two metabolic growth conditions known to trigger PrfA activation: glucose-1-phosphate (G1P) in a charcoal-treated LB-MOPS medium (G1P medium) and minimal defined medium containing limiting concentrations of BCAAs (LBM medium) (6, 10, 19). Luminescence, indicative of hly promoter activity, and optical density (OD), indicative of bacterial growth, were continuously monitored over 16 h of growth. Representative luminescence and growth profiles are shown for wild-type (WT) and ΔprfA mutant bacteria, demonstrating PrfA-dependent activation of the hly promoter under the two conditions (Fig. 1A) and PrfA-dependent growth of L. monocytogenes in G1P medium. Notably, growth in G1P medium represents a classic diauxic growth (showing two distinct phases of active growth), as this medium contains glucose and G1P as carbon sources. Under this condition, the bacteria first utilize the glucose in the medium and only when exhausted switch to utilize G1P, leading to the second growth phase. Utilization of G1P is dependent on the expression of the G1P transporter, UhpT, which is under the regulation of PrfA. PrfA itself becomes active only when the glucose in the medium is depleted, resulting in activation of virulence genes, including uhpT, in the second growth phase (6, 7). In the LBM medium, the ΔprfA mutant exhibits enhanced growth in comparison to WT bacteria, most likely due to the lack of virulence gene expression, which can pose a significant fitness burden (20, 21).
Luminescence and growth profiles of select mutants in G1P and LBM media. (A) L. monocytogenes (Lm) WT and ΔprfA mutant bacteria harboring pPL2-Phly-lux grown in G1P and LBM media. Luminescence (relative luminescence units [RLU]), indicative of hly transcription, and growth profiles (optical density [OD]) were detected during 16 h of growth. (Measurements were taken every 15 min using a plate reader.) (B) Luminescence and growth profiles of select mutants identified in the G1P or LBM screens. The mutants presented in this figure exhibited a phenotype only under one of the conditions described above. WT bacteria are presented in blue, and mutants in red. Of note, the y axis values vary in each graph due to the different luminescence and growth profiles of each mutant. The x axis values are constant in most graphs, representing 0 to 16 h of growth. Changes in WT bacterial profiles are a result of using different medium preparations on different days of experiments. The graphs shown are representative of six biological repeats.
Using this experimental setup, 6,000 mutants were screened three times, yielding a total of 118 mutants exhibiting aberrant luminescence profiles under at least one of the two conditions (see Tables S1, S2, and S3 in the supplemental material). Some mutants exhibited a phenotype under only one condition (examples are presented in Fig. 1B), while others exhibited a phenotype under both conditions, although not necessarily the same phenotype (examples are presented in Fig. 2). The variety of luminescence profiles among the 118 mutants indicates that hly activation is highly plastic (Fig. 1 and 2). The luminescence profiles varied across two dimensions—intensity (high to low) and time (early to delayed)—with some mutants even exhibiting persistent profiles. Sequence analysis indicated that most of the mutants map to metabolic genes (∼60%), with the rest mapping to genes encoding transport proteins (with some linked to metabolic pathways), transcription regulators, or unknown proteins (Tables S1 to S3). Many of the mutated metabolic genes were not essential for growth under the tested conditions (as indicated by growth curves) and were distributed across diverse metabolic pathways, such as carbon, nucleotide, amino acids, riboflavin, and fatty acid metabolism.
Luminescence and growth profiles of select L. monocytogenes (Lm) mutants exhibiting phenotypes in both G1P and LBM media. Luminescence (relative luminescence units [RLU]), indicative of hly transcription, and growth profiles (optical density [OD]) were detected during 16 h of growth in G1P and LBM media. (Measurements were taken every 15 min using a plate reader.) The mutants presented in this figure exhibited phenotypes under both conditions, although not necessarily the same phenotype. WT bacteria are presented in blue, and mutants in red. Of note, the y-axis values vary in each graph due to the different luminescence and growth profiles of each mutant. The x-axis values are constant in most graphs, representing 0 to 16 h of growth. Changes in WT bacterial profiles are a result of using different medium preparations on different days of experiments. The graphs shown are representative of six independent biological repeats.
Examples of mutant phentoypes include an early response in G1P medium due to mutated mannose 6-phosphate isomerase, a delayed response in G1P medium due to mutation in 1-phosphofructokinase or superoxide dismutase, high responses in LBM medium mapping to mutation in cysteine synthase or ribose 5-phosphate isomerase (rpiB), a high response in G1P but low response in LBM exhibited by an acetyl coenzyme A (acetyl-CoA) carboxylase mutant, and a delayed response in G1P and very low response in LBM due to mutation in phosphoglycerate dehydrogenase.
Examples of transport proteins exhibiting distinctive mutant phenotypes include LMRG_02271, representing a glutamine ABC transporter exhibiting a low response in LBM (22), and LMRG_01497, representing another ABC transporter exhibiting an early response in LBM medium.
Several transcription regulators were found to display mutant phenotypes, such as a LacI/PurR-like regulator (LMRG_02460) exhibiting a delayed response in G1P medium, a Crp regulator (LMRG_00280) exhibiting a high response in LBM and low response in G1P medium, and a DeoR-like regulator (LMRG_01144) that overall exhibited a high and persistent response under both conditions.
Among all of the mutants, three exhibited a null response similar to ΔprfA bacteria: an X-prolyl aminopeptidase (LMRG_00804, named PepP), an arsenate reductase (LMRG_1822), and the G1P transporter UhpT, which as expected exhibited a phenotype only in G1P medium (Fig. 1 and 2). In summary, this screening strategy revealed new metabolic and regulatory factors that likely play a role in the cross talk between L. monocytogenes metabolism and virulence. Furthermore, the findings afford new insights into the impact of L. monocytogenes metabolism on virulence and highlight a remarkable plasticity in virulence gene regulation.
PepP plays a role in virulence gene expression and infection in vivo.To further study the role of the identified genes in L. monocytogenes infection, intracellular growth analysis in bone marrow-derived (BMD) macrophages was performed for ∼60 mutants (indicated in Tables S1 to S3), representing regulatory factors, transporters, and metabolic and unknown genes. Among the tested mutants the Tn::pepP mutant (LMRG_00804) displayed the most dramatic intracellular growth defect. To validate this observation, a complete pepP gene deletion (ΔpepP) mutant was generated. The ΔpepP mutant exhibited a similar growth defect in BMD macrophage cells, which was complemented by addition of a copy of pepP gene on the integrative pPL2 plasmid (pPL2-pepP) (Fig. 3A). Of note, the ΔpepP mutant grows like WT bacteria in the rich brain heart infusion (BHI) medium (see Fig. S1 in the supplemental material). Next, we examined if the ΔpepP mutant exhibited the same luminescence phenotype as the Tn::pepP mutant. To this end, the ΔpepP mutant was conjugated with pPL2-Phly-lux, and luminescence and growth profiles were monitored during growth in G1P medium and LBM medium. As expected, the ΔpepP mutant behaved comparably to the Tn::pepP and ΔprfA strains under each of these conditions (Fig. 3B and C and Fig. 2). Since glutathione was recently shown to be a metabolic signal that activates PrfA, we examined the luminescence and growth profiles of WT and ΔpepP bacteria in minimal defined medium supplemented with or without glutathione (14). Notably, while WT bacteria induced hly transcription in response to the presence of glutathione in the medium (see Fig. S2A and B in the supplemental material), ΔpepP bacteria failed to display a similar response (Fig. 3D). Taken together, these observations support a role for PepP in virulence gene activation that is downstream of various metabolic signals, including glutathione.
PepP plays a role in virulence gene regulation downstream to various metabolic conditions. (A) Intracellular growth analysis of the L. monocytogenes (Lm) WT, the ΔpepP mutant, and a ΔpepP complemented strain harboring pPL2-pepP in BMD macrophages. Error bars represent the standard deviation from triplicate trials. The data represent three independent biological repeats (n = 3). (B). Luminescence (RLU) and growth (OD600) profiles of the L. monocytogenes WT and ΔpepP bacteria harboring pPL2-Phly-lux plasmid in G1P medium, LBM medium (C), and minimal defined medium supplemented with reduced l-glutathione (20 mM) (GSH medium) (D). The graphs shown are representative of three independent biological repeats. Error bars represent the standard deviation from triplicate trials and are hidden by the symbols.
To corroborate the premise that PepP acts in a pathway that modulates virulence, we performed direct reverse transcription-quantitative PCR (RT-qPCR) analysis of the three virulence genes hly, actA, and plcA in WT and ΔpepP bacteria grown to the mid-exponential phase in G1P medium, LBM medium, and glutathione-supplemented minimal medium (GSH medium). As shown in Fig. 4, ΔpepP bacteria exhibited reduced transcription levels of the three tested genes under all growth conditions (Fig. 4A to C). Next, we examined if reduced virulence gene expression was evident also during infection of BMD macrophage cells by the ΔpepP mutant. For this purpose, the promoter of plcA (encoding the virulence factor phospholipase C) was fused to 3 consecutive yfp genes in the pPL2 plasmid (pPL2-PplcA-3-yfp), which was then conjugated to WT, ΔpepP, and ΔprfA bacteria. Yellow fluorescent protein (YFP) fluorescence was detected at 4 h postinfection of BMD macrophage cells, indicative of PlcA intracellular expression, and was found to be reduced by 40% in ΔpepP bacteria in comparison to WT bacteria, with ΔprfA bacteria exhibiting no florescence (Fig. 5A). Further characterization of the ΔpepP mutant using different ex vivo experimental assays uncovered significant defects in attachment to Caco2 cells, in escape from BMD macrophage phagosomes, and in virulence in vivo in mice (with the latter also shown in reference 23) (Fig. 5B to D). Taken together, these findings suggest a role for PepP in virulence gene regulation in vitro and in vivo during infection.
Transcription analysis of the main virulence genes in L. monocytogenes (Lm) WT and ΔpepP mutant bacteria grown in G1P, LBM, and glutathione-supplemented media. Shown are the results from RT-qPCR analysis of expression of hly, plcA, and actA in WT L. monocytogenes and ΔpepP mutant bacteria grown to the mid-exponential phase in G1P medium (A), LBM (B), and glutathione-supplemented medium (GSH medium) (C). Bacterial growth under each condition is similar to what is shown in Fig. 3B to D. Samples were taken at 9 h for the G1P and GSH conditions and at 4 h for the LBM condition. Transcription levels are represented as the relative quantity (RQ) relative to the level in WT bacteria. The data represent three biological repeats (n = 3). Error bars represent the 95% confidence interval (*, P < 0.01).
PepP plays a role in L. monocytogenes infection in vivo. (A) Measurements of plcA intracellular expression. L. monocytogenes (Lm) WT and ΔpepP bacteria harboring pPL2-PplcA-3-yfp were grown in BMD macrophage cells. Infected cells were fixed at 3 h postinfection (p.i.) and stained for bacterial DNA and macrophage nuclei with DAPI. YFP fluorescence was measured per bacterium using a fluorescence microscope. The data are based on measurements of 100 intracellular bacteria. The P value was calculated using the t test (P < 0.05). (B) Attachment assay of WT L. monocytogenes and ΔpepP bacteria to Caco2 cells. The ΔflaA mutant serves as a negative control. Error bars represent the standard deviation from three independent biological repeats. The P value was calculated using the t test (P < 0.05). (C) Bacterial phagosomal escape assay of the L. monocytogenes WT and ΔpepP mutant at 2.5 h p.i. of BMD macrophage cells. The results are based on 10 microscope frames from three independent biological repeats for each strain. (In each frame, between 50 and 100 bacteria were counted.) The P value was calculated using a chi-square test. (D) Intravenous infection of C57BL/6 female mice with the L. monocytogenes WT and ΔpepP bacteria. Bacterial CFU were numerated at 72 h postinfection from livers and spleens taken from 4 to 6 mice for each strain. Mann-Whitney tests were performed to calculate statistical significance (*, P < 0.05).
PepP mediates PrfA association with the cytoplasmic membrane.Since PepP influences the transcription of several virulence genes, we hypothesized that it controls PrfA transcription, translation, or processing and thereby affects PrfA activity. To investigate this, bacteria were grown in G1P medium (with growth curves shown in Fig. S3 in the supplemental material) and total mRNA and proteins were extracted for RT-qPCR and Western blot analyses, respectively. As shown in Fig. 6A and B, PrfA transcription and protein levels were similar in WT and ΔpepP bacteria. PrfA protein levels were detected in total cell extracts using two types of antibodies: polyclonal antibodies generated directly against PrfA and anti-His antibodies detecting a chromosomally encoded C-terminus 6-His-tagged PrfA (C′-6His-PrfA) (Fig. 6B).
PepP is involved in PrfA protein activation. (A) Time course RT-q-PCR analysis of prfA in L. monocytogenes (Lm) WT and ΔpepP bacteria grown in G1P medium at 4, 7, and 8 h. (Bacterial growth curves are shown in Fig. S3.) Transcription levels are presented as relative quantity (RQ) to the prfA mRNA level in WT bacteria at 4 h. The data represent three independent biological repeats (n = 3). Error bars represent 95% confidence interval. (B) Western blot analysis of PrfA protein in total cell extracts of L. monocytogenes WT and ΔpepP mutant bacteria harboring a genomic C-terminus 6-His-tagged prfA gene (C′-6His-PrfA). Bacteria were grown in G1P medium for 8 h. PrfA was detected using anti-PrfA and anti-His tag antibodies. GroEL was used as a loading control protein. The blots shown are representative of three biological repeats (n = 3). (C) Growth analysis of L. monocytogenes WT, ΔprfA, ΔuhpT, ΔpepP, and ΔpepP(pPL2-PrpsD-uhpT) bacteria in G1P medium. The growth curves are shown are representative of three independent biological repeats. Error bars represent the standard deviation from triplicate trials and are hidden by the symbols. (D to E) Growth analysis of ΔprfA and ΔpepP bacteria harboring pPL2 plasmids encoding the WT prfA gene and prfA* variants under the prfA native promoter. (F) Intracellular growth analysis in BMD macrophages of L. monocytogenes WT or ΔpepP and ΔpepP bacteria harboring pPL2 plasmids encoding the different prfA* variants under the prfA native promoter. Growth curves are representative of three independent biological repeats. Error bars represent the standard deviation from triplicate trials and are hidden by the symbols.
While both antibodies demonstrated the same results, they further excluded the possibility that PrfA is directly cleaved by PepP in its C terminus, since no short versions of PrfA were observed, nor was there removal of the C-terminus His tag. The possibility that PrfA is cleaved at the N terminus by PepP was also ruled out by purifying intact PrfA protein from G1P medium-grown bacteria, as confirmed by mass spectrometry. In light of these data, we surmised that PepP does not affect PrfA transcription or translation nor directly processes it.
As demonstrated above, the ΔpepP mutant is unable to grow in medium containing G1P, exhibiting a phenotype that is similar to those of the ΔprfA and ΔuhpT mutants (Fig. 6C and Fig. 1 and 2). To investigate if this phenotype is because the PrfA protein does not become active and induce expression of the UhpT transporter, we examined whether the ΔpepP mutant could be complemented with a copy of a uhpT gene under the regulation of the constitutive rpsD promoter (pPL2-PrpsD-uhpT). Notably, these complemented bacteria (ΔpepP pPL2-PrpsD-uhpT) grew even better than WT bacteria in G1P medium, most likely since they could instantly utilize G1P, without the need to go through the transition of PrfA activation, supporting the premise that PepP controls PrfA activity (Fig. 6C). To further examine the mechanism of how PepP affects PrfA activity, we complemented the ΔpepP mutant with a set of constitutively active PrfA protein variants (prfA* mutants), with different mutations known to lock PrfA in its active form: L140F-PrfA*, G145S-PrfA*, and P219S-PrfA* (24–26). To this end, pPL2 plasmids harboring the different prfA* mutated genes, including the WT prfA gene, under the regulation of the native prfA promoter were conjugated to ΔpepP and ΔprfA bacteria, which were then tested for growth in G1P medium (Fig. 6D and E). Notably, while all PrfA variants, including WT PrfA, complemented ΔprfA growth in G1P medium (as expected) (Fig. 6D), only L140F-PrfA* and G145S-PrfA* were able to restore ΔpepP growth in G1P medium, whereas P219S-PrfA* and WT PrfA could not (Fig. 6E). This finding was also observed during intracellular growth of ΔpepP bacteria in macrophage cells, where L140F-PrfA* and G145S-PrfA* restored ΔpepP growth, while P219S-PrfA* did not (Fig. 6F). These observations indicate that certain constitutively active forms of PrfA can bypass the requirement for PepP function, whereas WT PrfA cannot, supporting that PepP plays a role in PrfA protein activation.
To gain additional insight into the mechanism whereby PepP activates PrfA, we analyzed the PepP protein sequence and found hydrophobic domains predicted to be putative transmembrane regions by some algorithms (TMPred, CCTop, and PHD). To examine whether PepP localizes to the cytoplasmic membrane, a 6-His tag was fused to its C terminus, and its cellular distribution was analyzed in cytosolic and membrane fractions prepared from G1P-grown bacteria using Western blot analysis. As a control for membrane fraction isolation, WT bacteria expressing chromosomally encoded C-terminus 6-His-tagged UhpT protein (C′-6His-PepP) were used (Fig. 7A, lower panel). Notably, the data clearly indicated that PepP is associated with the cytoplasmic membrane (Fig. 7A). In light of this finding, we next examined the cellular distribution of PrfA. Here we found that in WT bacteria PrfA is equally distributed between the cytosol and the membrane, whereas in ΔpepP bacteria PrfA association with the membrane was significantly reduced (5- to 20-fold) (Fig. 7B and C to F). Analysis of the membrane association of the different PrfA* proteins in ΔpepP bacteria revealed that only the variants that supported ΔpepP growth in G1P medium were associated with the membrane (i.e., L140F-PrfA* and G145S-PrfA*), whereas P219-PrfA*, which did not support growth, was not (Fig. 7C). Moreover, the level of PrfA association with the membrane directly correlated with the ability of the bacteria to grow in G1P medium (Fig. 7C and D and Fig. 6E). Of note, the total amounts of PrfA* variants were similar, excluding the possibility that the low membrane association of P219S-PrfA* is due to a lower expression level (Fig. 7C). Since in WT bacteria all PrfA variants associated with the membrane to a similar extent (Fig. 7E and F), we concluded that PepP is involved in PrfA recruitment to the cytoplasmic membrane, a process that appears to be critical for PrfA protein activation. To the best of our knowledge, this is the first demonstration of PrfA interacting with the cytoplasmic membrane, revealing a novel step in the complex mechanism of PrfA activation.
PepP recruits PrfA to the cytoplasmic membrane. (A) Western blot analysis of C-terminus 6-His-tagged PepP (C′-6His-PepP) in cytosolic and membrane fractions prepared from G1P medium-grown ΔpepP bacteria expressing C′-6His-PepP from the pPL2 plasmid (pPL2-6his-pepP). GroEL was used as a protein loading control. Chromosomally expressed C-terminus 6-His-tagged UhpT (C′-6His-UhpT) was used as a membrane protein marker in WT bacteria, validating membrane fraction isolation. Blots shown are representative of three independent biological repeats. (B) PrfA cellular distribution analyzed by Western blotting using anti-PrfA antibodies. L. monocytogenes (Lm) WT and ΔpepP bacteria, chromosomally expressing C′-6His-UhpT, were grown in G1P medium, and cytosolic and membrane fractions were isolated at 8 h. GroEL was used as a protein loading control and 6-His-tagged UhpT was used as a membrane protein marker. Blots are representative of three biological repeats. (C) Membrane association of PrfA* variants in ΔpepP bacteria. Membrane fractions were prepared from G1P medium grown bacteria. PrfA was detected using anti-PrfA antibodies. GroEL was used as a protein loading control. Total PrfA protein levels are also shown using anti-PrfA antibodies. Blots shown are representative of three biological repeats. (D) Quantification of PrfA variants' membrane association in ΔpepP bacteria, using Image Studio Lite software, was based on three Western blot analyses representing 3 biological repeats. The P value was calculated using the t test (*, P < 0.05). (E) Membrane association of PrfA* variants in WT L. monocytogenes bacteria. Membrane fractions were prepared from G1P medium-grown bacteria. PrfA was detected using anti-PrfA antibodies. GroEL was used as a protein loading control. The blots shown are representative of three biological repeats. (F) Quantification of PrfA variants' membrane association in L. monocytogenes WT bacteria based on three Western blot analyses representing 3 biological repeats using Image Studio Lite software. The P value was calculated using the t test (*, P < 0.05).
DISCUSSION
The virulent traits of bacterial pathogens enable access to new nutrient resources via invasion of mammalian cells. Together with the capability to exhibit virulence, specific metabolic traits have evolved that ensure the bacteria are capable of efficiently utilizing these nutrients, particularly those specific to the mammalian niche. This coevolution hypothesis explains why metabolic and virulence pathways are so interlinked and why they are often coregulated (27, 28). During infection, both the bacterial and host metabolisms change constantly, such that hundreds of reactions and metabolites are modulated, which in turn affects both the bacteria and the host cell. Tools to measure these global metabolic changes are still limited, as is our ability to integrate them into coherent schemes of gene regulation. Therefore, to gain insight into the response of bacterial pathogens to metabolic changes and signals, such as those encountered during infection, one approach has been to study the bacteria under in vitro conditions that mimic the intracellular niche. In this way, several metabolic signals (or growth conditions) have been identified that trigger L. monocytogenes virulence (e.g., glucose-1-phosphate, charcoal, BCAAs, glutathione, and S-adenosylmethionine) (6, 10, 14, 16). Nevertheless, how these multiple metabolic signals interact with one another and converge on PrfA regulation during in vivo infection remains unclear. Here we demonstrate another level of regulation in the complex relationship between L. monocytogenes metabolism and virulence, whereby PrfA subcellular location is controlled to modulate the activity of this master regulator of virulence.
The previously unknown regulation of PrfA described here was uncovered as a result of a screen designed to identify mutants with aberrant virulence phenotypes. The high number of mutations identified in metabolic genes was striking. Many of them had no apparent effect on growth, but a large effect on virulence gene expression, supporting the hypothesis that there is a close relationship between metabolism and virulence. For example, several enzymes involved in core metabolic pathways were found to modulate virulence gene expression but not growth, including cysteine synthase, acetyl-CoA carboxylase, 5′-nucleotidase, and 2-oxoisovalerate dehydrogenase E1 (Fig. 1 and 2). Notably, some of these pathways were shown previously to affect L. monocytogenes virulence (i.e., cysteine, BCAAs, and fatty acids) (10, 29, 30). These data support the hypothesis that while bacteria adjust their core metabolism to bypass a metabolic block (e.g., starvation for a nutrient or in this case a mutated enzyme), the consequent metabolic rearrangements are detected and transduced into regulation of virulence gene expression. Such a mechanism may involve metabolic sensors that respond to the presence or absence of specific metabolites and then interact with global regulators that modulate gene expression, as was shown for BCAAs and CodY (12). The data emerging from the genetic screens also highlighted the fructose-mannose pathway as a player in the relationship between metabolism and virulence, with seven distinctive mutants identified, among them mannose 6-P-isomerase (LMRG_01264), 1-phospho-fructokinase (LMRG_01507), and ribose 5-phosphate isomerase (LMRG_00036), which are involved in the production of fructose-6P and fructose-1,6P2, the latter an important intermediate of glycolysis (see Fig. S4 in the supplemental material). Notably, fructose-1,6P2 directly affects components of the carbon catabolic repression (CCR) system, which was suggested previously to regulate PrfA activity (4, 8, 31). Each one of the three mutants exhibits a distinct phenotype—an early, persistent, and extremely high luminescence profile, respectively—suggesting a complex interaction of this pathway (or related pathways) with PrfA. Indeed, carbon metabolism has long been known to regulate L. monocytogenes virulence, although the exact mechanism is still not clear. The identification of additional genes involved in this interaction may help to achieve a better understanding.
In light of the genetic screens, we characterized PepP, which was selected for study because it exhibited a phenotype similar to that of the ΔprfA mutant. Notably, ΔpepP was previously identified in other genetic screens performed in L. monocytogenes bacteria (17, 23). Those studies looked first for hypohemolytic mutants on blood agar plates and second for mutants defective in intracellular growth in macrophage cells. Both studies indicated a role for PepP in L. monocytogenes virulence, although no further investigation of this protein was performed. In this study, we characterized for the first time the role of PepP in L. monocytogenes virulence and found it to be critical for posttranslational activation of PrfA. Our observation that PepP impacts PrfA activation under different growth conditions representing different metabolic signals (i.e., glucose-1-P, BCAAs, and glutathione) supports the premise that it functions downstream of the sensing of these signals. The ΔpepP mutant was found to exhibit defective induction of virulence genes and to display impaired ability to invade and replicate within macrophage cells, and it demonstrated attenuated infection in mice, which altogether indicate an important role for PepP in L. monocytogenes virulence.
Although we did not completely decipher the mechanism by which PepP activates PrfA, we gained an important insight into the way this key virulence activator is regulated. Our data support that PepP somehow mediates recruitment of PrfA to the cytoplasmic membrane, a critical step in PrfA activation. Notably, to the best of our knowledge, this is the first report that PrfA is distributed between the cytosol and the membrane, and moreover, that alteration of this distribution influences its activity. Of note, it was hypothesized before that PrfA might be sequestered by phosphotransferase system (PTS) sugar permeases to keep it inactive, although it was never directly shown (3, 5, 32). We found the ΔpepP mutant to exhibit reduced levels of PrfA membrane association, without detectable effect on PrfA transcription, translation, or processing, suggesting that PepP mediates PrfA targeting to the membrane in an indirect manner. PepP is predicted to function as an X-prolyl aminopeptidase that selectively removes N-terminal residues from peptides or that hydrolyzes small peptides. Such proteins are abundant in bacteria and are involved in different processes, and so far, we have not managed to identify a PepP substrate or substrates. Nevertheless, we found that PepP associates with the membrane, which has led us to hypothesize that it might be part of a membrane protein complex that interacts with PrfA, a hypothesis that is under investigation.
PrfA belongs to the cyclic-AMP (cAMP) receptor protein (Crp) family of transcription regulators and thus is thought to require/bind a cofactor to gain protein activity (33). This hypothesis is based in part on protein structure similarities between PrfA and a well-studied Escherichia coli Crp, to which cAMP binding was demonstrated (34). Although it has been reported that PrfA does not bind cAMP, it is likely that another cofactor exists for PrfA (35, 36). Indeed, as mentioned, glutathione was recently shown to bind PrfA and serve as an allosteric inducer, supporting the premise that PrfA protein exists in two forms, active (37) and nonactive (36). This hypothesis is also founded on studies in which various prfA star (prfA*) mutations were isolated that lock PrfA protein in an active form, which was assumed to bypass the need for cofactor binding (i.e., G145S, L140F/P, and P219S) (6, 24, 26, 38–40). The crystal structure of G145S-PrfA* protein was shown to demonstrate a conformational change that is similar to an active Crp protein bound to cAMP, explaining the enhanced ability of this variant to bind DNA and activate virulence gene transcription (6, 33, 41). It is noteworthy that both G145S and L140F mutations localize to the same region in PrfA, the hinge α D domain next to the DNA binding motif, thus affecting DNA binding, whereas in contrast, the P219S mutation localizes to a distant region, the C-terminal α G domain, that was shown not to affect DNA binding but still render the protein constitutively active (resulting in a star phenotype). Interestingly, the P219 region does not exist in classical Crp proteins, raising the possibility that PrfA is capable of a distinct mode of regulation (25). Here we provide the first insight into the difference between the two sets of mutations or regions within PrfA (G145S/L140F versus P219S). While G145S-PrfA* and L140F-PrfA* complemented ΔpepP mutant growth in G1P medium, P219S-PrfA* could not. These observations correlate with the ability of the mutated proteins to associate with the membrane, as G145S-PrfA* and L140F-PrfA* could associate with the membrane without PepP, effectively bypassing its function, whereas P219S-PrfA* could not, strengthening the premise that membrane association is critical. Based on these findings, a model can be hypothesized in which PepP is part of a membrane complex that binds and activates PrfA. G145S-PrfA* and L140F-PrfA* mutants bind the membrane complex without the function of PepP and gain activity in addition to their enhanced ability to bind DNA. On the other hand, P219S-PrfA* requires PepP for membrane targeting, suggesting its region, the C-terminal α G domain, may be involved in PrfA association with the membrane. At this point it is not clear whether PrfA binds the membrane in its dimer or monomer form, and additional studies are required to explore this interaction as well as the involvement of additional factors. In summary, this study identifies new players in the cross talk between metabolism and virulence in L. monocytogenes and indicates a novel step in PrfA regulation.
MATERIALS AND METHODS
Ethics statement.Experimental protocols were approved by the Tel Aviv University Animal Care and Use Committee (04-15-057 and 04-13-039) according to the Israel Welfare Law (1994) and the National Research Council (Guide for the Care and Use of Laboratory Animals, 2010).
Bacterial strains and growth media.Listeria monocytogenes 10403S was used as a wild-type and parental strain for the generation of allelic exchange mutants (see Table S4 in the supplemental material). The E. coli XL-1 Blue strain (Stratagene) was used for plasmid cloning, and the E. coli SM-10 strain was used for plasmid conjugation to L. monocytogenes bacteria (42). L. monocytogenes bacteria were grown in brain heart infusion (BHI; Merck) medium or in Luria-Bertani–morpholinepropanesulfonic acid (LB-MOPS) medium preincubated with 0.2% activated charcoal supplemented with 25 mM glucose-1-phosphate (G1P medium) (6) or in minimal defined medium containing low concentrations (10 μg ml−1 of each) of isoleucine, leucine, and valine (LBM medium) (10, 19). For growth in the presence of glutathione, minimal defined medium (14) was supplemented with 20 mM reduced l-glutathione (Sigma).
Detection of L. monocytogenes growth and luminescence profiles.For luminescence assays, L. monocytogenes strains harboring pPL2-Phly-lux were used (18). Overnight bacterial cultures grown in BHI medium were adjusted to an OD600 of 0.03 in fresh BHI, LBM, or G1P medium, and 200 μl was transferred into a clear-bottom white 96-well plate sealed with a Breathe-easy (Diversified Biotech) cover. Plates were then incubated in Synergy HT microplate reader (BioTek) at 37°C for a period of 12 to 24 h. Plates were agitated every 15 min for 1 min, followed by measurement of luminescence (relative light units [RLU]) and absorbance (OD600).
Metabolic genetic screens.The L. monocytogenes library of mariner1-Tn mutants was constructed as previously described using pJZ037 with some modifications (17). Since pJZ037 plasmid contains chloramphenicol resistance, we first modified the available pPL2-lux plasmid (18), a generous gift from Collin Hill, by replacing the chloramphenicol resistance gene with a kanamycin cassette (43). The resulting pPL2-Phly-lux was conjugated to the L. monocytogenes 10403S strain and was used as a background for library construction. A total of 6,000 mariner-Tn mutants (in which the transposon inserts randomly) were picked into 96-well plates, with each well containing a single mutant. Mutants were grown overnight in BHI medium at 37°C. The next day, 5 μl of bacterial cultures were added to 200 μl of G1P or LBM medium in white 96-well plates with clear bottoms, which were then sealed with a Breathe-easy cover and incubated in a Synergy HT microplate reader (BioTek) as mentioned above. Mutants that showed differential luminescence profiles from wild-type bacteria were selected for further investigation. A total of 160 mutants were selected following a secondary screen. Transposon insertion sites of 98 selected mutants were successfully identified by PCR product sequencing as previously described (17). Genomes of an additional 40 mutants were fully sequenced using Illumina HiSeq 2500 to identify the transposon insertion sites (performed in the Technion Genome Center, Haifa Israel), yielding a total of 118 different mutants identified.
Generation of gene deletion mutants, 6-His-tagged fusion proteins, and complemented strains.Upstream and downstream regions of the selected gene were amplified using Phusion DNA polymerase and cloned into pKSV7oriT vector (44). Cloned plasmids were sequenced and then conjugated to L. monocytogenes using the E. coli SM-10 strain. L. monocytogenes transconjugants were selected on BHI agar plates supplemented with chloramphenicol and streptomycin and then grown on BHI agar plates supplemented with chloramphenicol alone for 2 days at 41°C to allow plasmid integration into the bacterial chromosome by homologous recombination. Bacteria were then passed several times in fresh BHI medium without chloramphenicol at 30°C to promote plasmid curing and the generation of an in-frame gene deletion. Bacteria were then plated on BHI plates, and chloramphenicol-sensitive colonies were validated for gene deletion by PCR. To construct chromosomal C-terminus 6-His-tagged PrfA (C′-6His-PrfA) and C-terminus 6-His-tagged UhpT (C′-6His-UhpT) translational fusions, a DNA sequence encoding a six-histidine tag was fused in-frame to the end of prfA or uhpT gene. A chromosomal C′-6His-UhpT was constructed in WT and ΔpepP bacteria. Complemented strains of the pepP deletion mutant were generated by introducing a copy of the pepP gene in trans under the control of its native promoter using the pPL2 integrative vector.
Bacterial RNA purification.For bacterial RNA extraction, overnight cultures were adjusted to OD600 of 0.03 in 20 ml of fresh BHI, LBM, G1P, and glutathione-supplemented media and incubated with agitation at 37°C. Bacteria were harvested by centrifugation at the indicated time points and snap-frozen in liquid nitrogen. Prior to RNA extraction, bacteria were thawed on ice and then washed with cold phosphate-buffered saline (PBS).
Total nucleic acids were extracted by a standard phenol-chloroform extraction protocol in the case of G1P medium-grown bacteria and by the RNAsnap method (45) in the case of LBM medium-grown bacteria. In both cases, the samples were then treated with DNase I, and 1 μg of purified RNA was reverse transcribed to cDNA using the QScript reverse transcription kit (Quanta). RT-qPCR was performed on 16 ng of cDNA using SYBR green (Roche) in a Step One Plus real-time PCR system (Applied Biosystems). The transcription level of a gene of interest was normalized to that of a reference gene, rpoD. Statistical analysis was performed using the Step One V2.3 software.
Intracellular growth of L. monocytogenes.For all infection experiments, L. monocytogenes strains were grown overnight in BHI medium at 30°C without shaking. The bone marrow-derived (BMD) macrophages used for infection experiments were isolated from 6- to 8-week-old female C57BL/6 mice (Harlan Laboratories, Israel) as described previously (46). Macrophage cells were cultured in Dulbecco's modified Eagle's medium (DMEM)-based medium supplemented with 20% fetal bovine serum (FBS), sodium pyruvate (1 mM), l-glutamine (2 mM), β-mercaptoethanol (0.05 mM), and monocyte colony-stimulating factor (M-CSF) prepared from L929 cells (BMDM medium). Cell cultures were incubated in a 37°C incubator with 5% CO2. A total of 1 × 105 macrophage cells per well in 100 μl of medium were seeded in a 96-well plate overnight. The next day, macrophage cells were washed twice with PBS and fresh medium was added. Approximately 1.6 × 103 L. monocytogenes bacteria in PBS were used to infect cells in each well. One hour postinfection, macrophage monolayers were washed twice with PBS and fresh medium supplemented with gentamicin (50 μg ml−1) was added to limit bacterial extracellular growth. At each time point, the medium was aspirated and 100 μl of sterile water was added, followed by vigorous pipetting to release intracellular bacteria. Then serial dilutions of cell lysates were plated on BHI agar plates, and CFU were counted after 24 h of incubation at 37°C.
Intracellular plcA gene expression and phagosomal escape assay in BMD macrophage cells.A total of 1 × 106 macrophage cells were seeded on 20-mm coverslip slides in BMDM medium and incubated overnight. For intracellular plcA gene expression, WT and mutant strains expressing three consecutive YFPs under the regulation of the plcA promoter (pPL2-PplcA-3-yfp) were used. For each experiment, approximately 2 × 106 L. monocytogenes bacteria were used to infect 1 × 106 macrophage cells (multiplicity of infection [MOI] of 2) or 1 × 107 L. monocytogenes bacteria (MOI of 5) in the case of the ΔpepP mutant. One hour postinfection, macrophage monolayers were washed twice with PBS and fresh medium with gentamicin (50 μg ml−1) was added to limit bacterial extracellular growth. Cells were fixed at 3 h postinfection with a fixative buffer (3.7% paraformaldehyde solution) and permeabilized with 0.05% Triton. Slides were then washed and stained as follows: L. monocytogenes bacteria were stained with fluorescein isothiocyanate (FITC)-conjugated anti-L. monocytogenes antibody (B65420F; Meridian Life Sciences), actin was stained with rhodamine-phalloidin (Biotium), and DNA was stained with DAPI (4′,6-diamidino-2-phenylindole)-containing Vectashield mounting medium. Images were taken using Nikon eclipse Ti-E microscope.
Attachment assay.A total of 2 × 106 Caco2 cells were cultured overnight in 6-well plates in Caco2 medium (minimal essential medium [MEM] with 20% fetal bovine serum [FBS], 1 mM sodium pyruvate, 2 mM l-glutamine, and MEM nonessential amino acid solution [01-3401B; Biological Industries]) supplemented with penicillin and streptomycin in a 37°C incubator with 5% CO2. The next day, the cells were washed twice with PBS, and fresh medium without antibiotics was added. The cells were infected at an MOI of 2 with PBS-washed bacteria. Thirty minutes postinfection, the cells were washed 5 times with PBS and lysed with 1 ml of cold water. Serial dilutions of cell lysates were plated on BHI agar plates, and CFU were counted after 24 h of incubation at 37°C.
In vivo mouse infections.L. monocytogenes bacteria were grown in BHI medium at 30°C overnight without shaking. Bacterial cultures were washed twice with PBS. Six- to 8-week-old female C57BL/6 mice (Harlan Laboratories, Israel) were infected via tail vein injections with 4 × 104 bacteria in 200 μl of PBS. Each group consisted of 5 to 6 mice for every bacterial mutant tested. Animals were observed daily for any signs of illnesses and were euthanized 72 h postinfection. Spleens and livers were harvested and homogenized in 0.2% saponin. The numbers of viable bacteria in each organ were determined by plating serial dilutions of homogenates onto BHI agar plates.
Western blot analysis.L. monocytogenes bacteria were grown in G1P medium at 37°C and harvested at a desired OD. Then the bacteria were lysed by incubation with 5 U of mutanolysin (M9901; Sigma) for 1 h at 37°C, followed by sonication. Cell debris were removed by centrifugation at 3,000 × g. Membranes were collected by centrifugation at 290,000 × g for 15 min (4°C) and further solubilized with 1% SDS. The supernatants were used as the cytosolic fraction. Protein concentrations were quantified by modified Lowry method (47). Samples with equal amounts of total proteins, cytosolic or membrane, were separated by 12.5% SDS-PAGE, electroblotted, and probed either with mouse anti-6-His tag antibodies (Abcam ab18184 [dilution, 1:1,000]) or with rabbit anti-PrfA antibodies (made in this study; Almog Diagnostics [dilution 1:2,500]), followed by horseradish peroxidase (HRP)-conjugated goat anti-mouse/rabbit IgG (Jackson ImmunoResearch, USA), respectively. Anti-GroEL antibodies (a kind gift from A. Azem, Tel Aviv University) were used as a loading control at a dilution of 1:20,000, followed by HRP-conjugated goat anti-rabbit IgG. Six-His-tagged UhpT was used as a membrane protein marker control. Western blots were developed by the ECL enhanced chemiluminescence reaction.
PrfA purification and analysis by mass spectrometry.L. monocytogenes bacteria harboring a chromosomal prfA-6-His gene were grown in LB-MOPS-G1P medium and harvested at OD of 1 by centrifugation (1,900 × g, 10 min). The bacterial pellet was resuspended in 15 ml of cold purification buffer (PB) (50 mM NaH2PO4 [pH 8], 0.3 M NaCl) supplemented with 20 mM imidazole, 10 mM phenylmethylsulfonyl fluoride (PMSF), 10 mM sodium fluoride, and 10 mM sodium pyrophosphate. Bacteria were lysed by 3 passages at 12,000 lb/in2 using an ultrahigh-pressure homogenizer (Stansted Fluid Power). Cell debris was removed by centrifugation at 16,000 × g for 20 min, and lysates were incubated with 1 ml of Ni-nitrilotriacetic acid (NTA) agarose resin (ABT) for 1 h at 4°C with tilting. The Ni-NTA beads were then loaded onto a column and washed with 10 ml of PB supplemented with 20 mM imidazole. PrfA proteins were eluted using PB with 250 mM imidazole. Elution fractions were separated on 12.5% SDS-PAGE gel and stained with Coomassie brilliant blue. Stained bands representing purified 27-kDa PrfA were isolated from the gel and analyzed by peptide mass fingerprinting at the Smoler Protein Research Center at the Technion, Haifa, Israel. Protein samples were digested by trypsin, and the resulting proteolytic peptides were analyzed by liquid chromatography-tandem mass spectrometry (LC-MS/MS) on an Orbitrap XL (Thermo) and identified by Discoverer software version 1.4 against the Listeria UniProt database and against decoy databases (in order to determine the false discovery rate [FDR]) using the Sequest search engine.
ACKNOWLEDGMENTS
This work was supported in part by the INFECT-ERA grant (funded by MOH Israel) and by a research grant (R01AI109048) from the U.S. National Institute of Allergy and Infectious Diseases to A.A.H.
FOOTNOTES
- Received 10 January 2017.
- Returned for modification 3 February 2017.
- Accepted 6 April 2017.
- Accepted manuscript posted online 10 April 2017.
Supplemental material for this article may be found at https://doi.org/10.1128/IAI.00027-17.
- Copyright © 2017 American Society for Microbiology.
