Listeriolysin S Is a Streptolysin S-Like Virulence Factor That Targets Exclusively Prokaryotic Cells In Vivo

ABSTRACT Streptolysin S (SLS)-like virulence factors from clinically relevant Gram-positive pathogens have been proposed to behave as potent cytotoxins, playing key roles in tissue infection. Listeriolysin S (LLS) is an SLS-like hemolysin/bacteriocin present among Listeria monocytogenes strains responsible for human listeriosis outbreaks. As LLS cytotoxic activity has been associated with virulence, we investigated the LLS-specific contribution to host tissue infection. Surprisingly, we first show that LLS causes only weak red blood cell (RBC) hemolysis in vitro and neither confers resistance to phagocytic killing nor favors survival of L. monocytogenes within the blood cells or in the extracellular space (in the plasma). We reveal that LLS does not elicit specific immune responses, is not cytotoxic for eukaryotic cells, and does not impact cell infection by L. monocytogenes. Using in vitro cell infection systems and a murine intravenous infection model, we actually demonstrate that LLS expression is undetectable during infection of cells and murine inner organs. Importantly, upon intravenous animal inoculation, L. monocytogenes is found in the gastrointestinal system, and only in this environment LLS expression is detected in vivo. Finally, we confirm that LLS production is associated with destruction of target bacteria. Our results demonstrate therefore that LLS does not contribute to L. monocytogenes tissue injury and virulence in inner host organs as previously reported. Moreover, we describe that LlsB, a putative posttranslational modification enzyme encoded in the LLS operon, is necessary for murine inner organ colonization. Overall, we demonstrate that LLS is the first SLS-like virulence factor targeting exclusively prokaryotic cells during in vivo infections.

L isteria monocytogenes is a foodborne pathogen and a facultative intracellular bacterium capable of causing severe disease in humans and animals. Upon ingestion of contaminated food, L. monocytogenes colonizes the intestine and crosses the intestinal barrier, disseminating via the blood to the liver, spleen, brain, and placenta (1,2). The listeriosis fatality rate is estimated to be 20 to 30% of infected individuals despite antibiotic treatment (1). The most severe human listeriosis outbreaks are associated with a subset of L. monocytogenes lineage I strains that harbor a gene cluster encoding the bacteriocin and hemolytic factor listeriolysin S (LLS) (3,4). Interestingly, the LLS gene cluster is absent from the most commonly studied L. monocytogenes lineage II strains EGD, EGDe, and 10403S (3), and its contribution to the intracellular lifecycle of L. monocytogenes is unknown (5).
LLS is homologous to streptolysin S (SLS [encoded by sagA in the sag operon]), a potent cytolytic toxin produced by most group A Streptococcus pyogenes (GAS) strains (6,7). SLS is naturally expressed in vitro and is responsible for the classical ␤-hemolytic phenotype of S. pyogenes on blood agar plates (8). Using live-cell imaging, it has been shown that SLS activates the major erythrocyte anion exchange protein band 3 and favors a rapid influx of Cl Ϫ ions into red blood cells (RBCs), leading to cellular rupture (9). In HEK-293 cells, SLS causes lactate dehydrogenase (LDH) release into the medium, massive cytoskeletal disassembly, loss of focal contacts, and detachment from the tissue culture plates (7). SLS promotes resistance to phagocytic killing in whole-blood killing assays and activates an inflammatory programed cell death pathway in macrophages (10,11). In vivo studies have shown that SLS is required for S. pyogenes infection of skin and soft tissues (9,12). The toxin encoded by an SLS-like gene cluster in Clostridium botulinum is named clostridiolysin S (CLS), and similarly to SLS, it is hemolytic and cytotoxic in HEK-293 cells (7,13).
LLS, CLS, and SLS belong to the family of thiazole/oxazole-modified microcins (TOMMs) and are encoded by biosynthetic gene clusters characterized by the presence of cyclodehydratases/dehydrogenase genes (7). In the case of the LLS gene cluster, the genes llsB, llsY, and llsD code for putative posttranslational modification (PTM) enzymes that modify the product of the llsA gene coding for the structural LLS toxin (6,7). Attempts to uncover the structural characteristics of the mature SLS-like toxins have been unsuccessful (7,12,13). The only direct structural insight available suggests that SLS contains two oxazole moieties observed at positions Ser 46 and Ser 48 and that CLS contains a methyloxazole at position Thr 46 (13).
It has been suggested that given the genetic similarities of the TOMM operons in Gram-positive pathogens, it is likely that these TOMMs contribute to the pathogenic potential of each bacterial producer, similarly to SLS for S. pyogenes (6,14). However, it is currently unknown if all SLS-like molecules of bacterial pathogens are cytotoxins, which play a role in tissue injury and contribute to virulence by targeting eukaryotic cells. By using an llsB deletion mutant, it has been previously concluded that LLS is a hemolysin and a cytotoxin that contributes to L. monocytogenes virulence in an intraperitoneal murine model of infection (3). We recently described that in orally infected mice, LLS behaves as a bacteriocin favoring L. monocytogenes gut colonization through modulation of the intestinal microbiota (4).
In the present study, we sought to deepen our insight into the functions of LLS during listeriosis, with a particular focus on its effects once the intestinal barrier has been crossed. Here, we show that while LLS kills prokaryotic targets, it does not impact infection of eukaryotic host cells in vitro and in vivo. Our present results demonstrate that, apart from its role as a bacteriocin in the host intestine, the structural product of the llsA gene has negligible activity on eukaryotic host cells, while the llsB gene product contributes to colonization of deep organs.

RESULTS
LLS causes weak RBC hemolysis in vitro and does not alter RBC counts in vivo. By growing L. monocytogenes colonies that constitutively express LLS in blood agar plates, or by using cell-free supernatants obtained from bacteria activated by the LLS inducer yeast RNA core, LLS has been previously shown to display hemolytic activity (3). In the case of SLS, real-time imaging indicated that the secreted toxin induces hemolysis of RBCs in 30 s (9). We investigated whether LLS exerts hemolysis as efficiently as SLS. First, we determined in brain heart infusion (BHI) agar plates with human or mouse blood the hemolytic activity of the L. monocytogenes F2365 wild type (WT), F2365 Δhly, F2365 ΔllsA PrfA* (a strain that constitutively expresses listeriolysin O [LLO]), and F2365 Δhly(pHELP::llsA) (a strain that does not produce LLO but expresses LLS). F2365 Δhly was not hemolytic on human (Fig. 1A) or mouse (see Fig. S1 in the supplemental material) BHI blood agar plates. On the other hand, the F2365 WT and ΔllsA PrfA* strains caused hemolysis of both RBCs due to LLO activity, while the Δhly(pHELP::llsA) strain also caused hemolysis in human and mouse RBCs due to LLS activity ( Fig. 1A; Fig. S1). Second, we performed live-cell microscopy to visualize human RBC hemolysis caused by the same F2365 strains. Real-time imaging indicated 100% RBC hemolysis in 3 min due to LLO activity when incubated with the F2365 WT (Fig. 1B, upper panel, and C) or in 1 min when incubated with F2365 PrfA* ΔllsA (Fig. 1C). Hemolysis was marked by the visual rupture of the RBC membranes, as previously described for SLS (9). F2365 Δhly caused no hemolysis, highlighting the potent activity of LLO (Fig. 1C). Surprisingly, only 15% of RBCs were lysed after 60 min of incubation with F2365 Δhly(pHELP::llsA) (Fig. 1B, Mice were killed at 48 h p.i., and a blood sample was taken to assess the number of RBCs and the hemoglobin concentration. Mice whose blood contained more than 300 L. monocytogenes CFU/ml are colored to facilitate comparison of blood parameters. LLS Targets Exclusively Prokaryotic Cells In Vivo ® lower panel, and C), suggesting that the hemolytic activity of LLS is much less efficient than that of LLO, as shown in our experiments and as reported in the literature for SLS. In order to reveal the potential effects of the LLS hemolytic activity during L. monocytogenes infection in vivo, we infected intravenously conventional BALB/c mice with the F2365 WT, ΔllsA mutant strain, or its complemented strain and quantified the concentration of hemoglobin in blood, the number of RBCs, and the bacterial burden in blood at 48 h postinfection (p.i.). Consistent with real-time imaging performed in vitro, no significant changes in hemoglobin concentration (Fig. 1D, left panel) or RBC counts (Fig. 1D, center panel) were observed in mice infected with these three strains. Importantly, no correlation was observed between the number of L. monocytogenes cells in the blood (Fig. 1D, right panel) and the number of RBCs or hemoglobin ( Fig. 1D; see Table S1 in the supplemental material). Taken together, these results show that LLS hemolytic activity is weak in vitro and that this activity does not affect RBC numbers during L. monocytogenes infection in vivo.
LLS expression does not confer resistance to phagocytic clearance. The weak hemolytic activity of LLS prompted us to evaluate its contribution to L. monocytogenes survival in blood. In the case of GAS, WT strains are able to proliferate in human blood, while SLS-negative mutants are cleared, demonstrating a crucial role for SLS in the resistance to phagocytic killing in human blood (10). To explore whether LLS contributes to phagocytic clearance, we incubated fresh human blood from three donors with L. monocytogenes F2365 WT, F2365 ΔllsA, and F2365(pHELP::llsA) strains. No significant difference could be observed between the three strains in phagocytic clearance ( Fig. 2A). GAS can kill macrophages through the activation of an inflammatory programed cell death pathway mediated by SLS and streptolysin O (SLO) (11). Moreover, SLS induces alterations in keratinocyte inflammatory signaling cascades during GAS infection (15). These SLS properties prompted us to investigate a hypothetical role of LLS in cytokine secretion by macrophages infected by L. monocytogenes. We used an array kit detecting 111 different cytokines (Proteome Profiler mouse XL cytokine array) to compare the inflammatory responses of RAW264.7 macrophages infected with the F2365 WT strain (which does not express LLS in vitro, as shown above) to the responses Quereda et al.
® of cells infected with the ΔllsA deletion mutant or with the pHELP::llsA strain (which constitutively expresses LLS). No significant difference between the three groups (WT, ΔllsA, and pHELP::llsA strains) could be observed for secreted cytokine levels, including interleukin-1 (IL-1), IL-6, and tumor necrosis factor alpha (TNF-␣), whose role is essential in the response to L. monocytogenes infection (see Fig. S2 in the supplemental material) (16), suggesting that LLS does not alter cytokine production in macrophages. Overall, our results indicate that LLS does not contribute to L. monocytogenes survival during the blood stage. LLS expressed by intracellular L. monocytogenes is cytotoxic for neither macrophages nor epithelial cells. The absence of a role for LLS in macrophages contrasts with the previously published role of this molecule as cytotoxic for diverse cell lines, including professional phagocytes (3). To revisit this concept, we evaluated LLS cytotoxicity by measuring lactate dehydrogenase (LDH) release from RAW264.7 macrophages and enterocyte-like Caco-2 cells. No differences in LDH release into culture supernatants were observed between the WT, ΔllsA, and pHELP::llsA strains in RAW264.7 and Caco-2 cells (Fig. 2B). Taken together, these results suggest that LLS secreted by intracellular L. monocytogenes is not cytotoxic for eukaryotic host cells.
llsA promoter activity is undetectable during infection of eukaryotic cells in vitro or infection of deep organs, such as spleen or liver, in vivo. Transcriptional fusion of the LLS promoter to a lux reporter plasmid previously demonstrated that LLS expression is specifically triggered in the intestine of orally infected mice and undetectable in other organs, such as liver and spleen (although these organs contained higher bacterial counts) (4). Bioluminescent signals in the intestine of germfree mice could also be detected, indicating that the intestine itself and not the gut microbiota activates the LLS promoter (4). Different compounds naturally present in the intestine, like mucin, gastric fluid, pepsin, NaHCO 3 , short-chain fatty acids, ethanolamine, or even 6% O 2 or intestinal content of mice added ex vivo could not induce bioluminescence under in vitro conditions (4). In the present study, we investigated whether macrophages or epithelial cells could trigger LLS expression. We generated a transcriptional fusion of the LLS promoter to the promoterless green fluorescent protein gene (gfp), which is more sensitive as a reporter than the luciferase systems in other Gram-positive bacteria (17). A similar green fluorescent protein (GFP) transcriptional reporter was used for hly, a gene essential for the intracellular lifestyle of L. monocytogenes. We next infected Caco-2 epithelial cells and RAW264.7 macrophages with F2365 InlB (pAD-PllsA-GFP) or F2365 InlB (pAD-Phly-GFP) for 2, 6, and 24 h. Figure 3 shows intracellular bacteria that polymerize host cell actin (arrows) and bacteria not associated with actin (arrowheads), presumably inside vacuoles after 5 h of gentamicin treatment. Intracellular F2365 InlB (pAD-Phly-GFP) produced a green fluorescence at 2, 6, and 24 h p.i. in both cell lines tested, indicating that the GFP reporter is functional and that the hly promoter is active ( Fig. 3A and B; see Fig. S3 and S4 in the supplemental material). Interestingly, the activity of the llsA promoter was undetectable at 2, 6, or 24 h p.i. in the cell lines tested ( Fig. 3A and B; Fig. S3 and S4). No signal was observed for the llsA promoter at 2, 6, and 24 h p.i. in other cell lines (LoVo, HeLa, and Jeg-3) where the hly promoter was active (data not shown).
Furthermore, we investigated whether L. monocytogenes inoculated intravenously expresses LLS in deep organs, such as the spleen or the liver. We used the L. monocytogenes F2365 llsA::lux strain in which we fused the LLS promoter to the lux reporter plasmid (4). Upon intravenous infection of BALB/c mice with 10 4 cells of F2365 llsA::lux , a bioluminescent signal was detected in the abdomen of infected animals at 72 and 96 h p.i. To uncover the origin of the bioluminescent signals, abdominal skin and peritoneum dissection was performed. The ex vivo images of the gastrointestinal system and the livers and spleens removed from the body are shown in Fig. 3C. Interestingly, bioluminescent signals were only detected in the stomach and the intestine of infected mice, while being absent from the liver and spleen (which are the main organs targeted by L. monocytogenes). Importantly, although the bioluminescent signals were specifically expressed in the gastrointestinal system, the liver and spleen contained at least more than 100-fold CFU than the stomach, small intestine, cecum, or colon (Fig. 3D).
Altogether, these results demonstrate that the llsA promoter is not active or is expressed at very low levels during infection of host cells, suggesting that the activation of LLS expression in the gastrointestinal system is not triggered by the infection of eukaryotic cells. LLS does not contribute to cell infection and is not sufficient for L. monocytogenes vacuolar escape. Since we could not exclude that LLS produced during intestinal infection could impact the capacity of bacteria to infect eukaryotic cells, we investigated the potential contribution of LLS to infection. No differences in intracellular CFU counts were observed between the F2365 WT, F2365 ΔllsA, and F2365(pHELP:: llsA) strains at 2, 6, or 24 h p.i. in RAW264.7, HD11, and Caco-2 cells (Fig. 4A). These results show that LLS is not required during cellular infection by L. monocytogenes.
If LLS does not play a role in cell infection, this means that it should not affect vacuolar escape by L. monocytogenes. The capacity of LLS to mediate vacuolar rupture and to allow intracellular growth was evaluated in RAW264.7 macrophages. Extracellular and total L. monocytogenes numbers were distinguished by using inside-outside staining of fluorescently labeled bacteria, and cytosolic microorganisms were identified by actin staining. Approximately 80% of F2365 WT cells escaped from vacuoles and polymerized actin ( Fig. 4B and E). As expected, LLO was required for vacuolar escape in macrophages, as no escape events were observed upon cell infection with F2365 Δhly (Fig. 4C and E). Importantly, no F2365 Δhly(pHELP::llsA) cells were found associated with actin ( Fig. 4D and E), suggesting that these bacteria were trapped in phagosomes and that LLS expression is not sufficient to rupture the bacterial internalization vacuole in order to facilitate access to the host cytoplasm. Overall, our results indicate that LLS does not target eukaryotic membranes during infection.
LlsB, but not LLS, has a role in spleen and liver colonization. SLS causes extensive tissue disruption, inflammation, and necrosis of skin lesions and is required for S. pyogenes infection of skin and soft-tissues (9,10,12). Regarding LLS, as mentioned above we have shown that this molecule is specifically expressed in the intestine of intravenously and orally infected mice, where it alters the host intestinal microbiota and increases L. monocytogenes persistence (Fig. 3C and 4). LLS function in the intestine requires the activity of the LlsB enzyme, which is by homology with SLS putatively involved in the posttranslational modification of LLS. As reported, deletion of LlsB completely inactivates LLS hemolytic activity in vitro and decreases L. monocytogenes virulence in a mouse intraperitoneal infection model (3). To determine the impact of LLS on virulence once L. monocytogenes has crossed the intestine, we intravenously infected mice with the F2365 WT, F2365 ΔllsA, F2365 ΔllsB, F2365 ΔllsA, and F2365 ΔllsB complemented strains, and quantified the bacterial burdens in the liver and spleen. In agreement with the inactivity of LLS promoter in liver and spleen (4), and with the absence of a role for LLS in the blood and in the cell lines used in the present study, the ΔllsA mutant strain did not display significantly different bacterial loads in the spleen or liver compared with the WT strain (Fig. 5A). However, significant differences were observed between the WT and ΔllsB strains in the number of CFU isolated from these organs (P Ͻ 0.05) (Fig. 5A). Furthermore, histopathologic assessment of the liver and spleen of mice infected with WT, ΔllsA mutant, and ΔllsA complemented strains showed no differences in the numbers, locations, and inflammatory cell types of necrotic foci (Fig. 5B and C). Together, these results demonstrate that LLS does not play an essential role in L. monocytogenes deep organ colonization once the intestine has been crossed and strongly suggest that LlsB performs additional important functions for virulence apart from putatively driving LLS posttranslational modification.
LLS promotes killing of target bacteria. Our results suggest that during in vivo infections, LLS does not target eukaryotic cells. Our previous results (4) indicate that LLS is able to modulate the growth of Lactococcus lactis, Staphylococcus aureus, and L. monocytogenes lineage II strains as well as representatives of the Allobaculum and Alloprevotella genera during in vivo infection (4). To investigate a direct cytotoxic effect of LLS on target bacteria, diluted overnight cultures of L. monocytogenes F2365 strains (WT and the pHELP::llsA strain) and the target L. lactis were grown in coculture. After 3 h of coculture, there was a reduction in the growth of L. lactis only when cocultured with L. monocytogenes F2365(pHELP::llsA) (Fig. 6A). The effect of LLS on L. lactis growth was even higher at 6 and 9 h of coculture (Fig. 6A). Under transmission electron microscopy, L. lactis cultured alone or cocultured with the L. monocytogenes F2365 WT (which does not produce LLS under in vitro conditions [4]) for 3, 6, or 9 h showed the typical structure of Gram-positive cocci, with a thick, uniform, smooth cell wall and intact Differential immunofluorescence staining for identification of extracellular versus total L. monocytogenes numbers was performed. Extracellular bacteria (Extra) were labeled with a secondary goat anti-rabbit Alexa 647 (blue), and total bacteria (total) were labeled with a secondary goat anti-rabbit Alexa 488 (green) after cell permeabilization. Actin was labeled by using phalloidin conjugated to Alexa 546 (red). Nuclei were stained with Hoechst (gray). Bars, 5 m. (E) Quantification of the number of bacteria of the L. monocytogenes strains used in panels B, C, and D that access the host cytosol. cytoplasmic membrane attached to the wall ( Fig. 6B and C). The cytoplasm was granular and evenly distributed in the cell. Some of the cells demonstrated a dividing septum, indicative of bacterial growth ( Fig. 6B and C). After 3 h of coculture with L. monocytogenes F2365(pHELP::llsA), disruption of L. lactis cell wall integrity was observed (Fig. 6C). At 6, 9, and 24 h of coculture with L. monocytogenes F2365(pHELP:: llsA), increasing numbers of L. lactis cells showed more drastic changes, including cell wall wrinkles and even cellular lysis (Fig. 6C). These results confirm that LLS is the first SLS-like virulence factor of a bacterial pathogen able to promote death of target bacteria.

DISCUSSION
In 2008, SLS-like gene clusters were discovered in clinically relevant Gram-positive pathogens (including S. aureus and C. botulinum) and other nonpathogenic bacteria (7), leading to the identification of the LLS gene cluster in L. monocytogenes. SLS is a potent membrane-damaging agent and a major virulence factor contributing to GAS infection through rapid destruction of eukaryotic cells and tissue damage (6, 9-11, 15, 18-20). It has been proposed that SLS-like virulence factors from other Gram-positive pathogens also behave as potent cytotoxins (6,14). Interestingly, functional experimental data of LLS activity on eukaryotic cells are scarce (3) cytogenes clinical isolates) and that it has been related to the L. monocytogenes infectious potential in epidemiological and comparative genomic studies (21,22). In the present work, we aimed to characterize the extent to which LLS potentially contributes to host infection by directly performing damaging activities on eukaryotic cells and tissues.
Using an array of molecular, cell biology, and histology techniques, we demonstrate that unlike SLS, (i) LLS causes very weak RBC hemolysis, (ii) it does not confer resistance to phagocytic clearance, (iii) it does not affect the levels of secreted cytokines by cells infected by L. monocytogenes, (iv) when expressed under the control of its native promoter or expressed through a constitutive promoter by intracellular L. monocytogenes, it is not cytotoxic for epithelial cells and macrophages, (v) its constitutive expression by L. monocytogenes in the confined space of a phagocytic vacuole is not sufficient to rupture this membrane compartment, (vi) its expression is undetectable within host cells due to inactivity of its promoter under both in vitro and in vivo conditions, (vii) it does not contribute to eukaryotic cell infection, and (viii) it does not contribute to virulence in an intravenous infection murine model. Altogether, our results clearly demonstrate that the biological activity of LLS is distinct from that of SLS or CLS, showing that LLS does not target eukaryotic host cells and is not involved in inner organ infection during systemic stages of listeriosis.
It has been previously proposed that LLS is hemolytic on blood agar plates and cytotoxic for epithelial and phagocytic cell lines (3). Importantly, experimental conditions in vitro used a ratio of 100 bacteria per cell for 6 h (without explicit use of gentamicin) to demonstrate the cytotoxic effect of LLS on J774, C2-Bbe, and CT26 cells (3). The fact that we did not observe the same cytotoxic effect on Caco-2 or RAW264.7 cells after the same infection period (making use of gentamicin after 1 h p.i.), using an MOI of 5 for Caco-2 cells or an MOI of 2 for RAW264.7 cells, leads us to believe that experimental conditions used in the previous study (3) may have influenced the in vitro cell system conditions, finally leading to an increase of Ϸ20 to 30% of LDH release to the medium due to LLS and also bacterial exposure. Moreover, the cytotoxic effect was only demonstrated by using a L. monocytogenes strain in which LLS was constitutively expressed using the pHELP promoter (3). It has also been claimed that LLS contributes to virulence in a murine intraperitoneal model of infection (3). This conclusion was based on the reduced F2365 ΔllsB CFU numbers in the livers and spleens relative to those in the corresponding F2365 WT-infected mice. However, when we compared the virulence of the F2365 ΔllsA and F2365 ΔllsB mutants to that of the F2365 WT strain, we discovered an unexpected difference between these two deletion mutants: llsB but not llsA (the gene coding for the toxin LLS) contributed to virulence in our mouse intravenous model of infection, indicating that LlsB performs additional functions apart from the putative posttranslational modification of LLS. llsB is therefore the first gene from a dehydratase/dehydrogenase TOMM complex reported to perform additional functions apart from putative posttranslational modifications of the cluster-encoded toxin. An attractive hypothesis that remains to be validated is that LlsB participates in the posttranslational modification of another molecule outside the LLS operon.
Our present results show that although LLS constitutive expression [strain F2365(pHELP::llsA)] can damage RBCs in a blood agar plate or in a phosphate-buffered saline (PBS) suspension, the concentration of LLS produced by the epidemic L. monocytogenes F2365 WT strain during in vivo infection once the intestine has been crossed has minimal toxicity or even no effect on eukaryotic cells. On the contrary, LLS is highly toxic for prokaryotic targets such as L. lactis in vitro and modulates the host microbiota in vivo (4). These data demonstrate that LLS is the first SLS-like virulence factor of a bacterial pathogen that only targets prokaryotic cells in vivo. It is very important to highlight that in our present experiments, LLS expression is detected in the stomach and the intestine of mice after intravenous bacterial inoculation. Indeed, it has been previously shown that L. monocytogenes can be discharged back to the gastrointestinal system from the gallbladder (23,24). Our results suggest that modulation of the host microbiota by LLS takes place not only during oral animal infection but also upon intravenous infection: this fact should be now taken into account during in vivo animal experiments with relevant epidemic L. monocytogenes strains expressing the LLS operon.

MATERIALS AND METHODS
Bacterial strains and cell lines. The epidemic lineage I L. monocytogenes strain F2365 of serotype 4b responsible for the 1985 California listeriosis outbreak (25) was used as parental strain (BUG3012; UIBC bacterial collection). The isogenic mutants and plasmids used in this study are listed in Table 1. Bacteria Cell infection. Prior to infection, 96-well tissue culture plates were seeded with cells to attain 80% confluence on the day of infection. Overnight cultures of bacterial strains were washed three times in PBS and resuspended in infection medium (1% fetal bovine serum [FBS]) at an MOI of 2 (RAW264.7 and HD11) or 5 (Caco-2). Cells were centrifuged for 1 min at 1,000 rpm to synchronize infection. The cells were then incubated with the bacteria for 1 h at 37°C. Following this incubation, the cells were washed, and extracellular bacteria were neutralized by adding complete medium containing 40 g/ml of gentamicin. At 2, 6, or 24 h postinfection, cells were washed with PBS and finally lysed in distilled water containing 0.1% Triton X-100. The number of viable intracellular L. monocytogenes cells was calculated by serial dilution and colony counting on BHI agar plates. These experiments employed 6 technical replicates per bacterial strain and were repeated three times with independent clones of each of the strains. Statistical analyses were conducted by using the Student's t test.
Cytotoxicity LDH release assays and cytokine measurements. RAW264.7 and Caco-2 cells were infected as indicated in the previous paragraph for 24 h (washed after 1 h of infection and with 40 g/ml gentamicin added to kill extracellular bacteria) where the supernatant was recovered and filtered. LDH levels were assayed with the kit LDH BR (Linear Chemicals) according to the kit instructions. Supernatants from RAW264.7 cells were also used to measure cytokine levels by using the cytokine array kit (Proteome Profiler mouse XL cytokine array; Becton, Dickinson) according to the manufacturer's instructions.
Vacuolar escape. RAW264.7 cells were infected as indicated in the previous paragraphs for 6 h (washed after 1 h of infection and with 40 g/ml gentamicin added to kill extracellular bacteria) and fixed with a paraformaldehyde (PFA) solution (4% in PBS) for 15 min at room temperature. Extracellular L. monocytogenes cells were labeled with a primary polyclonal goat anti-Listeria serum and with a secondary chicken anti-goat Alexa 647. Next, cells were permeabilized using 0.1% Triton X-100 for 4 min at room temperature, and total L. monocytogenes cells were labeled with the same primary antibody and a secondary chicken anti-goat Alexa 488 antibody. Actin was labeled by using phalloidin conjugated to Alexa 546. Nuclei were stained with Hoechst 33342 (dilution, 1/1,000). Samples were then rinsed four times in PBS and examined with a Zeiss Axiovert 135 epifluorescence microscope (Carl Zeiss, Inc.) associated with a charge-coupled device (CCD) camera. Images were obtained with a ϫ63 oil immersion objective and processed with MetaMorph software (Universal Imaging). Cytosolic bacteria were considered those stained with the Alexa 546 and Alexa 488 antibody but lacking Alexa 647 antibody labeling. Approximately 100 cells were counted in 3 representative fields to estimate the number of cytosolic L. monocytogenes cells.
Evaluation of llsA promoter expression with a GFP and a luciferase reporter system. A transcriptional fusion of the llsA promoter was created by cloning 500 bp upstream of the ATG of the respective gene with the gene encoding GFP-mut2 (30) (generating PllsA-GFP). This construct was cloned into SalI/SmaI-digested pAD vector (generating pAD-PllsA-GFP). Gene synthesis to construct PllsA-GFP was produced by Genecust (Luxembourg). pAD-PllsA-GFP was electroporated into L. monocytogenes F2365 InlB . hly was used as a control gene, the expression of which is upregulated during infection of eukaryotic cells. The transcriptional fusion of the promoter of hly to GFP-mut 2 (pAD-Phly-GFP) was also cloned into L. monocytogenes F2365 InlB .
For epifluorescence analysis of promoter activity, RAW264.7 and Caco-2 cells were infected for 6 h (washed after 1 h of infection and with 40 g/ml gentamicin added) fixed with a PFA solution (4% in PBS) for 15 min at room temperature, and permeabilized (0.1% Triton X-100 for 3 min in PBS). Samples were then rinsed four times in PBS, incubated with Hoechst and phalloidin conjugated to Alexa 546 for 30 min at room temperature, and rinsed four times in PBS. Samples were examined with a Zeiss Axiovert 135 epifluorescence microscope (Carl Zeiss, Inc.) associated with a charge-coupled device (CCD) camera. Images were obtained with a ϫ63 oil immersion objective and processed with MetaMorph software (Universal Imaging).
For in vivo bioluminescence experiments, 8-week-old female BALB/c mice were infected by intravenous inoculation with 10 4 L. monocytogenes F2365 llsA::lux (BUG3763) cells grown in BHI broth to an optical density (OD) of 1.0 at 37°C. Bioluminescence imaging was accomplished using an IVIS Spectrum in vivo imaging system (Perkin Elmer) with a 5-min exposure time. Mice were anesthetized with isoflurane. For CFU determinations, whole luminal contents from stomach, small intestine, cecum, and colon, as well as tissues from stomach, small intestine, cecum and colon, liver, and spleen, were obtained, homogenized, and serially diluted and plated on Oxford agar plates (Oxoid). In order to determine the CFU numbers in the tissues of the gastrointestinal system, the tissues were washed three times in DMEM, incubated for 2 h in DMEM supplemented with 40 g/ml gentamicin, and finally washed three times in DMEM.
Mouse infections. Six-to 8-week-old female BALB/c mice (Charles River, Inc., France) were injected intravenously with 10 4 CFU of the indicated strain. Mice were sacrificed at 48 h after infection (four mice in each group), and livers and spleens were removed. Half of the organ was used to assess bacterial load, and the other half was used for histological analysis. To assess bacterial load, organs were homogenized and serially diluted. Dilutions were plated onto BHI plates and grown during 24 h at 37°C. Colonies were counted to assess bacterial load per organ. RBC and hemoglobin counts were determined using a Horiba scil Vet abc Plus veterinary hematology blood analyzer. Two independent experiments were carried out. Statistically significant differences were evaluated by the Mann-Whitney test. Pearson's correlation coefficients were computed to measure correlations between blood parameters and L. monocytogenes CFU in blood.
Histological analysis. Liver and spleen tissue sections from mice intravenously infected with the F2365 WT, ΔllsA, and ΔllsA complemented strains sacrificed at 48 h were fixed in 10% neutral buffered formalin and routinely processed for the histopathological analysis. Four-micrometer sections per organ were stained with hematoxylin and eosin (H&E). The number of necrotic foci and the main cell type infiltrating necrotic areas were recorded. All of the slides were internally coded and analyzed blind. Statistically significant differences were evaluated by the Mann-Whitney test.
Ethics statement. This study was carried out in strict accordance with the French national and European laws and conformed to the Council Directive on the approximation of laws, regulations, and administrative provisions of the Member States regarding the protection of animals used for experimental and other scientific purposes (86/609/Eec). Experiments that relied on laboratory animals were performed in strict accordance with the Institut Pasteur's regulations for animal care and use protocol, which was approved by the Animal Experiment Committee of the Institut Pasteur (approval no. 03-49). All human blood samples were anonymized and collected from the French National Blood Service under IRB approval no. HS2008-3470.