ABSTRACT
The Panton-Valentine leukocidin (PVL) is a cytotoxin expressed by many methicillin-resistant Staphylococcus aureus (MRSA) strains that cause community-acquired infections (CA-MRSA). Its role in virulence however, is controversial, with clinical data suggesting that PVL-producing strains may cause less severe disease in humans. PVL is capable of lysing human white blood cells, but at sublytic amounts, PVL can activate protective host immunity in the absence of cell damage. The concentration-dependent reactions it elicits from host cells could be the reason for seemingly contradictory results about PVL's role in virulence. We hypothesized that a key to understanding PVL's action on host cells and, possibly, outcomes from infection is the amount of toxin present, a hypothesis previously supported in studies using a low-inoculum skin infection model, where low levels of PVL augmented innate immune resistance to infection. Here, we present additional data supporting this hypothesis using a mouse model of MRSA pneumonia, wherein we found increased virulence of isogenic Δpvl strains and further confirmed PVL's capacity to activate proinflammatory responses from mouse and human neutrophils and pulmonary cells. Activation was measured as the production of phosphorylated p38 mitogen-activated protein kinase (MAPK) and proinflammatory cytokines interleukin-8 (IL-8) and KC (from human and mouse cells, respectively), as well as the release of antibacterial factors. Conversely, PVL lowered the levels of tumor necrosis factor alpha (TNF-α) produced in active pulmonary infection, while low doses induced apoptosis, suggesting that PVL also has the capacity to regulate inflammation. Our data indicate that, independent of its cytotoxic effects, PVL also plays an important and positive immunomodulatory role during MRSA infections.
INTRODUCTION
Community-acquired methicillin-resistant Staphylococcus aureus (CA-MRSA) infections are widespread and are increasingly causing life-threatening conditions even in young, healthy individuals. The most common manifestations of CA-MRSA infections are skin and soft tissue infections and pneumonia (14, 16, 17, 35). A high percentage of CA-MRSA isolates recovered from patients express Panton-Valentine leukocidin (PVL) (30, 35), but the role of PVL in pathogenesis is controversial due to conflicting findings in animal studies regarding its modulation of the outcomes from infection as reported by multiple research groups (2, 3, 27, 41, 44, 46). Of note, studies of humans with S. aureus infections find either no difference in clinical outcomes from PVL-positive MRSA infections or a reduced virulence of these strains (1, 8, 9, 28, 36). The discrepancies in outcomes, particularly in experimental animal studies, may be explained, in part, by the variable effects of PVL on the host, which depend on the amount of PVL present. Beyond a threshold concentration, PVL is able to lyse cells in the white blood cell lineage, including polymorphonuclear neutrophils (PMNs), of a subset of mammals that includes humans by forming pores in the membranes of these immune cells (23). Below the threshold needed for pore formation (sublytic), PVL can activate neutrophils and granulocytes, stimulating the release of proinflammatory cytokines, such as interleukin-8 (IL-8) and leukotriene B4 (19, 24, 25), which might favorably dispose the host to resist infection.
We recently reported findings from a low-inoculum skin abscess model where MRSA strains with deletions of the two genes encoding the two-component PVL toxin, lukF and lukS (Δpvl strains), were more virulent than their isogenic wild-type PVL-producing parental strains, and neutralization of PVL toxin by antibody also increased bacterial virulence (46), consistent with a beneficial release of protective inflammatory responses from PVL's action on host cells. Studies by others have also shown an increase in the virulence of Δpvl MRSA strains in a mouse model of pneumonia (3), although the reproducibility and breadth of this outcome using multiple strains was not reported. We therefore sought to test the hypothesis that elaboration of PVL might actually be beneficial to the host by activating the innate immune response. We evaluated the virulence of multiple wild-type (WT) and isogenic Δpvl MRSA strains in a murine model of pneumonia. The resistance of mouse cells to lysis by PVL makes murine models of infection well suited for the study of the inflammatory response to PVL and how it might influence the course of disease independently of its cytolytic effects. We determined that PVL can regulate the inflammatory response, activating immune and lung epithelial cells while downregulating potentially damaging tumor necrosis factor alpha (TNF-α) and inducing apoptosis. Taken together, we documented the potential for a desirable PVL-dependent host response that augments innate resistance to CA-MRSA pneumonia.
MATERIALS AND METHODS
Bacterial strains.S. aureus strains MW2 (NRS123), NRS193, and NRS194 were obtained from the Network on Antimicrobial Resistance in S. aureus (NARSA). Isogenic Δpvl constructs of strains MW2, NRS193, and NRS194 were generated by allelic replacement with the PVL-encoding genes lukF and lukS that were disrupted by introducing a cassette conferring resistance to erythromycin (erm) into the pvl locus (Δpvl::erm), as previously described (46). Strains with increased PVL expression were constructed using a multicopy plasmid containing the clone's pvl genes along with its native promoter. The tandem lukSF (pvl) genes were amplified from the genome of S. aureus MW2 by PCR (5′ primer BamHIuplukSF5, GGA TCC GTG CGA TTC ATG GTA AGC CTA TAG GTG G, and 3′ primer PstI-rlukF-R, GC CTG CAG TTA GCT CAT AGG ATT TTT TTC CTT AG; restriction enzyme sequences for BamHI and PstI, respectively, are in boldface) and cloned into plasmid pOS1 (kindly provided by Olaf Schneewind, University of Chicago) that had been digested with BamHI and PstI restriction enzymes, giving rise to pOS1-pvl. The plasmid was transformed into Escherichia coli TOP10, and isolates harboring pOS1-pvl were selected on LB agar plates with 100 μg ampicillin/ml. The plasmids pOS1 and pOS1-pvl were next electroporated into S. aureus RN4220 and selected on tryptic soy agar (TSA) containing 10 μg chloramphenicol/ml and then phage transduced into target MRSA strains using phage 85 (6).
Growth of S. aureus strains.For mouse infections, S. aureus strains were grown in tryptic soy broth (TSB) or yeast extract-Casamino Acid-sodium pyruvate (YCP) broth at 37°C with gyratory shaking to mid-late logarithmic phase (optical density at 650 nm of ∼0.8 to 0.9), washed twice, and suspended in 1/100 of the volume of the original growth medium. Aliquots were frozen at −80°C until used for mouse infections, at which time they were thawed and diluted in growth medium to the desired inocula. The actual CFU injected were confirmed by plate counts of the inocula. When generating MRSA supernatants to be incubated with cell lines, S. aureus cells were grown in RPMI with 1% Casamino Acids.
Mouse model of S. aureus pneumonia.FVB mice aged 3 to 5 weeks purchased from Jackson Laboratories (Bar Harbor, ME) were anesthetized with a mixture of ketamine (90 μg/kg) and xylazine (10 μg/kg) prior to infection with S. aureus strains. Ten microliters of S. aureus was instilled onto each nostril of a mouse with the mouse placed on its back after sedation. Following aspiration of the S. aureus inocula into the lungs, the mice were returned to their cages. Each mouse was inoculated with S. aureus NRS193 and NRS194 (as well as the Δpvl derivative strains) at 2 × 109 CFU, while MW2 strains were inoculated at 5 × 108 CFU. For survival studies and analysis of bacterial burden, animals were monitored for signs of illness over a period of 48 h, at the end of which surviving mice were sacrificed and their lungs harvested for determination of the levels of S. aureus cells in the lungs. Moribund mice and mice that succumbed to S. aureus pneumonia in between routine observations (2 to 4 times a day) during the 48-h time course of infection were removed from their cages and promptly frozen at −20°C until lungs could be harvested. Plating of the S. aureus Δpvl strains was on antibiotic selective medium (3 μg erythromycin/ml). In the mouse pneumonia experiments comparing WT PVL-positive (PVL+) MRSA harboring pOS1 and strains harboring pOS1-pvl that encoded increased expression of PVL, mice were given drinking water containing 10 μg/ml chloramphenicol for maintenance of the pOS1 plasmids starting 24 h prior to infection, with antibiotic selection maintained through the duration of the experiment. Animals were monitored for signs of illness over a period of 96 h. Animal experiments were conducted in accordance with the guidelines of and under a study approved by the Harvard Medical Area Institutional Animal Care and Use Committee.
Production of purified PVL, antibody to PVL, and evaluation of potential effects from endotoxin in recombinant PVL preparations.Purification of PVL and production of antibody for the detection of PVL have been described previously (46). Purified PVL preparations were treated with an endotoxin removal kit (Genscript, Piscataway, NJ). PVL preparations denatured by boiling no longer had any effect on mammalian cells, providing further verification of the absence of endotoxin contamination at a level sufficient to activate or affect mammalian cells.
Detection of PVL in infected murine pulmonary tissue.Mice were infected with 2 × 108 CFU of WT or Δpvl strain MW2 and, 24 h later, euthanized, and lungs were removed, weighed, and homogenized in phosphate-buffered saline (PBS). Cells and debris were removed by centrifugation. Microtiter plates were sensitized with 0.5 μg/ml purified LukF and LukS proteins, prepared as described previously (46). Dilutions of clarified, homogenized lung tissue were made in PBS containing 0.1% Tween 20 and then applied to the plate, along with primary rabbit antibody to recombinant LukS and LukF proteins (46) mixed at equal volumes to achieve a final dilution of 1:1,000 for each antiserum. A standard curve was constructed by adding known amounts of purified PVL (in place of lung samples) as an inhibitor of binding of the antiserum. A horseradish peroxidase-conjugated antibody to rabbit IgG was used as the secondary antibody. Following the addition of a colorimetric reagent, signal intensities were measured by reading the optical density (OD) values of the plate at 450 nm. A standard curve was constructed plotting the known concentrations of PVL with the OD values; this was used to determine the levels of PVL in the lung samples.
Detection of cytokines from MRSA-infected mice.Pulmonary tissues were harvested 12, 18, and 24 h postinfection with WT or Δpvl S. aureus strain MW2. Cytokine analyses were carried out with the Luminex system (Affymetrix, Santa Clara, CA). The cytokines tested included IL-1α, IL-1β, IL-2, IL-5, IL-6, IL-10, IL-12p40, IL-12p70, IL-13, IL-17, IL-33, gamma interferon (IFN-γ), TNF-α, monocyte chemoattractant protein 1 (MCP1), and granulocyte-macrophage colony-stimulating factor (GM-CSF).
Preparation of human and mouse neutrophils.Human PMNs were purified from fresh human blood obtained from healthy adult volunteers who provided informed consent under a protocol approved by the Partner's Healthcare Institutional Review Board. Neutrophils were harvested after separation of cells by gradient centrifugation with Polymorphprep (Axis-Shield, Oslo, Norway).
Mouse PMNs were harvested from the bone marrow of FVB mice. Upon euthanasia, bilateral tibias and femurs were aseptically excised from mice, followed by removal of the epiphyses. A small-bore needle was inserted into the bone shafts to flush out the bone marrow with Hanks' balanced salt solution (HBSS). Bone marrow suspensions were separated by centrifugation through a gradient of layered Histopaque-1077 and Histopaque-1119 solutions (Sigma-Aldrich, St. Louis, MO) to obtain a highly enriched (>95%) neutrophil population.
Cell lines.Human alveolar basal epithelial cells (A549) and FVB mouse lung epithelial cells (MLE 12) were obtained from the American Type Culture Collection (ATCC) and maintained in Dulbecco's modified Eagle's medium (MEM) containing 10% or 2% fetal bovine serum, respectively. Both cell lines were grown at 37°C with 5% atmospheric carbon dioxide.
Detection of phosphorylated p38, activated caspase 3, and cytokines.For detection from cultured cell lines of phosphorylated p38 mitogen-activated protein kinase (MAPK) (phospho-p38), activated caspase 3, and cytokines, cells were weaned off serum the day before the assays as it interfered with the analysis for production of these factors. The following day, purified PVL or S. aureus supernatants diluted 1 in 10 were added to the cells for 2 to 6 h. Human and mouse neutrophils were used as described above. Cell lysates were assayed by Western blotting for phospho-p38 after a 2-h incubation, while assays for activated caspase 3 were carried out after 6 h. Supernatants were collected between 3 and 6 h for cytokine analysis. Baseline cytokine readings (where no PVL was added) were set as background and subtracted from experimental samples containing PVL. Antibodies (Cell Signaling, Danvers, MA) and cytokine assaying kits (R&D Systems, Minneapolis, MN) were used according to the manufacturers' instructions.
Assays for antibacterial activity.Approximately 104 A549 or MLE 12 cells were seeded per well of a 96-well plate the day prior to an experiment. On the day of an experiment, culture medium was replaced with fresh medium and allowed to incubate for an additional 2 h, at which point purified PVL was added. For assays with human and mouse neutrophils, freshly harvested cells were suspended to 5 × 106/ml in MEM containing 1% bovine serum albumin. All of these cell types were stimulated with PVL for 4 h, and the supernatants collected for bacterial killing assays. The relative amounts of antibacterial activity in cell supernatants were measured by comparison of S. aureus viability in each supernatant with its viability in supernatants from cells not exposed to PVL, as previously reported (46).
Histopathology.Histopathology of lung sections from S. aureus-infected mice was performed by the Dana-Farber/Harvard Cancer Center (DF/HCC) Research Pathology Cores.
Statistical analysis.Overall survival was compared by chi-square analysis. Pairwise comparisons were derived using t tests on parametric or log-transformed data, and for multiple-group comparisons, P values were corrected using the Bonferroni correction. Survival curves were compared using the log-rank tests. All calculations were performed using GraphPad Prism 4 software.
RESULTS
Isogenic Δpvl mutants were more virulent than PVL-producing MRSA in a mouse pneumonia model.In the mouse pneumonia model, the overall inoculum of S. aureus was large, but for the strains tested, the lowest possible doses were used where the majority of the animals would be expected to have signs of illness and high levels of bacteria in their lungs. Doses just 50% lower were rapidly cleared, with few CFU present in the lungs as soon as 5 min postinfection, and no significant signs of infection were noted (4; unpublished observations). The rates of mortality of mice infected with all Δpvl mutants were consistently higher than those for mice infected with each corresponding PVL-producing parental MRSA strain, despite infections with the same doses (Fig. 1A), even though the overall low mortality obtained with strain NRS193 did not reach statistical significance for a difference in mortality. Restoring PVL expression in the MW2 Δpvl strain again reduced the virulence of that strain. Notably, the CFU counts of viable S. aureus bacteria measured in the infected mouse lungs were not significantly different among mice infected with WT, Δpvl, or pvl-complemented S. aureus despite the differences in outcome (Fig. 1B). The histopathologic analysis of lung sections from mice infected with isogenic WT and Δpvl S. aureus MW2 revealed that infection with the PVL-producing parental strain resulted in more-inflamed tissue with a significant influx of immune cells (Fig. 1C), indicative of a robust host response. This was associated with a lower rate of mortality in murine pneumonia. Using an enzyme-linked immunosorbent assay (ELISA), we found that PVL was expressed at 18 h postinfection in the pulmonary tissues of WT-MW2-infected mice at 19.41 ± 2.57 (mean ± standard error of the mean) ng/g lung tissue. By comparison, lung sections from mice infected with S. aureus MW2 Δpvl exhibited little inflammation, with a notable absence of immune cells, suggesting that the loss of PVL production led to the loss of beneficial host inflammatory response; this was associated with increased lethality from MRSA infection in murine pneumonia but with comparable levels of bacteria present in the lung.
Comparison of the virulence of PVL-producing MRSA strains with that of their respective isogenic Δpvl mutants in a mouse pneumonia model. (A) Comparison of the percent mortality at 48 h in mice infected with three different MRSA strains, their Δpvl isogenic strains, and in the case of strain MW2, the PVL-complemented strain (pvl comp). P values were determined by chi-square analysis. Differences with strain NRS193 were not significant. (B) CFU in lungs of moribund mice 48 h after infection. No significant differences in CFU counts were found by analysis of variance (ANOVA) or t tests. (C) Pathology of selected murine lung sections from MRSA strain MW2 and its isogenic Δpvl mutant. The number of mice included in each group is noted on the individual graphs. Bars represent geometric means, and error bars denote standard errors of the means (SEM).
Increased expression of PVL reduces the virulence of MRSA in a mouse pneumonia model.The increased in vitro expression of PVL from MRSA strains harboring the plasmid pOS1-pvl was verified by Western immunoblot analysis (Fig. 2A). Although the pvl genes were under the control of their own promoter in pOS1-pvl, increased expression of PVL protein was probably due to the presence of the pOS1 plasmid in multiple copies (6). Mice infected with all three strains of MRSA with increased PVL expression (MRSA/pOS1-pvl) had better overall survival rates and/or survived longer than mice infected with the same three strains carrying the empty vector (MRSA/pOS1) that expressed lower levels of PVL from the single copy of pvl genes in the chromosome (Fig. 2B). Measurement of PVL in lungs of mice infected 18 h earlier with S. aureus MW2 carrying only the cloning vector pOS1 had a mean of 6.78 ± 5.22 ng PVL/lung, whereas mice infected with S. aureus MW2/pOS1-pvl had 309.03 ± 1.98 ng PVL/lung (Fig. 3B).
Increased production of PVL decreases the virulence of MRSA strains in a mouse pneumonia model. (A) PVL production, as determined by immunoblot analysis of culture supernatants, by three strains of MRSA carrying empty vector pOS1 or vector pOS1 containing the PVL genes lukF and lukS. (B) Survival curves comparing outcomes in mice infected with WT PVL+ MRSA or isogenic strains expressing higher levels of PVL from pOS1-pvl. Infecting doses: S. aureus strains NRS193 and NRS194, 2 × 109 CFU per mouse; MW2 strains, 5 × 108 per mouse. Infected mice were monitored for survival over 4 days. P values were determined by log-rank test.
PVL production by MRSA strains in vitro (A) and in vivo (B and C). (A) PVL production by MW2 strains after overnight growth in TSB broth (medium for strains carrying pOS1 plasmids contained 20 μg/ml chloramphenicol). (B) PVL production in pulmonary tissue of mice after 18 h of infection. Pulmonary tissues harvested 18 h postinfection were tested for PVL levels by ELISA using ≥four animals per group. (C) Trend correlating levels of PVL production and survival in a mouse model of S. aureus pneumonia.
Interestingly, the MW2 strain carrying the empty pOS1 plasmid exhibited reduced virulence in mice compared with the virulence of the MW2 strain with no plasmid (20% survival versus 5%, respectively, at 48 h postinfection). This is consistent with the reduced levels of PVL detected from in vitro-grown MW2/pOS1 versus the levels detected in MW2 (Fig. 3A). The increased burden of maintaining a high-copy-number plasmid may have been a factor in the decreased levels of PVL produced. Nevertheless, the three MW2 strains form a decreasing gradient of PVL production (MW2/pOS1-pvl > MW2 > MW2/pOS1) that correlates overall with an increase in virulence (Fig. 3C).
WT-PVL-producing MRSA elicits lower levels of TNF-α than its isogenic Δpvl counterpart.Analyses of cytokines produced in the lungs 12, 18, or 24 h after intranasal infection of mice with isogenic WT or Δpvl MRSA strain MW2 showed lower levels of TNF-α produced at all three time points by mice infected with WT S. aureus (Fig. 4A). No significant or consistent differences were found in the levels of IL-1α, IL-1β, IL-2, IL-5, IL-6, IL-10, IL-12p40, IL-12p70, IL-13, IL-17, IL-33, IFN-γ, MCP1, or GM-CSF. Given the potential for TNF-α to have a toxic/lethal effect on the host (40), it was not surprising that the higher overall mouse mortality seen in animals infected with the Δpvl strains corresponded with increased levels of TNF-α. However, cytokines and/or chemokines not included in the panel and other host factors could also have contributed to the differences in outcomes following infection with WT or Δpvl MRSA.
PVL stimulates the release of proinflammatory cytokines and phosphorylation of p38 MAPK from human and mouse cells. (A) Detection of TNF-α in murine pulmonary tissues infected with WT or Δpvl MRSA strain MW2 18 h after intranasal infection with 5 × 108 CFU/mouse. (B) Detection of phospho-p38 and murine KC produced by purified PMNs from FVB mice exposed to indicated concentrations of purified PVL. (C) Detection of phospho-p38 and murine KC produced by MLE 12 mouse lung epithelial cells exposed to indicated concentrations of purified PVL. (D) Phosphorylation of p38 MAPK in FVB mouse PMNs exposed to culture supernatants from WT or Δpvl S. aureus strains. (E and F) Detection of phospho-p38 and human IL-8 produced by purified PMNs (E) or cultured A549 alveolar type II pneumocytes (F) exposed to indicated concentrations of purified PVL. (G) Phosphorylation of p38 MAPK in human PMNs exposed to culture supernatants from WT or Δpvl S. aureus strains. Error bars denote SEM. Statistical analysis was performed by using the t test or ANOVA, comparing levels of cytokines detected in response to PVL with background levels in cells not exposed to PVL (***, P < 0.01; **, P < 0.05; *, P < 0.1).
PVL activates mouse and human neutrophils and lung epithelial cells to release KC or IL-8, as well as activating p38 MAPK phosphorylation.In accounting for the increased mortality of mice infected with the Δpvl S. aureus, in addition to increased TNF-α, there was also reduced inflammation in these animals' lungs, suggestive of other changes affecting this aspect of host responses. It has previously been shown that sublytic amounts of PVL (∼5 nM and under) could stimulate human PMNs to phosphorylate MAP kinases and to produce and release cytokines (24, 25, 46). We determined whether PVL could activate mouse cells in a manner similar to its activation of human neutrophils, which would provide a possible explanation for the positive outcomes from infection seen in mice infected with PVL+ S. aureus strains compared with the outcomes following infection with Δpvl strains. We examined the activation of mouse cells by detecting phospho-p38 MAPK and the release of the proinflammatory cytokine KC. p38 MAPK is a signaling molecule that participates in the relay of signals from activating stimuli at the cell surface that culminates in nuclear translocation of transcription factors that stimulate the transcription of proinflammatory cytokine genes and, ultimately, increased cytokine production. We found that purified PVL in the range of 0.8 nM to 13 nM activated mouse neutrophils (Fig. 4B), as well as MLE 12 mouse lung epithelial cells, as evidenced by increased levels of phospho-p38 MAPK (Fig. 4C). Both types of cells were resistant to lysis by PVL at all concentrations tested (not shown), including those that routinely lyse human PMNs (approximately 5 nM and above). In mouse neutrophils, the levels of phospho-p38 increased in a dose-dependent manner following incubation with PVL. All concentrations of PVL tested appeared to induce similar levels of phospho-p38 from MLE 12 cells.
We detected KC release by mouse neutrophils and lung epithelial cells with exposure to increasing amounts of PVL (Fig. 4B and C), the only exception being mouse PMNs incubated with 0.8 nM PVL (Fig. 4B). Interestingly, KC release by MLE 12 lung epithelial cells was approximately 10-fold higher than its release by neutrophils (Fig. 4C and B, respectively). In addition, we also found that culture supernatants from PVL+ MRSA could activate both mouse (Fig. 4D) and human (Fig. 4G) neutrophils similarly, as reflected by the increased amounts of phospho-p38 detected compared with the amounts in mouse neutrophils incubated with supernatants from isogenic Δpvl MRSA strains. The higher levels of phospho-p38 detected in human and mouse neutrophils suggest that PVL retains its immune-activating properties in the context of S. aureus infections and that the effect was not just an artifact seen in experiments with purified proteins. We were not able to readily discern a difference between the levels of KC or IL-8 released after stimulation with supernatants from cultures of WT or Δpvl MRSA strains (not shown), probably due to the presence of other S. aureus factors that could also modulate cytokine release (21, 34).
In parallel, we confirmed that sublytic amounts of purified PVL could activate human PMNs. Overall, we detected robust IL-8 release from PVL-stimulated human PMNs, which rose as the concentration of PVL increased from 0.8 nM to 3.25 nM but began decreasing as PVL was further increased to 6.5 nM and 13 nM (Fig. 4E). The toxicity of 6.5 nM and 13 nM PVL to human PMNs (>50% lysis) probably explains the reduction in IL-8 production. PVL concentrations of 0.8 nM to 3.25 nM were not significantly toxic to human PMNs (<20% maximal lactate dehydrogenase release after 2 h), which is consistent with PMN activation by sublytic doses of PVL. Notably, in contrast with human PMNs, the human type II pneumocytes, A549 cells, were not susceptible to lysis by PVL at all PVL concentrations tested (not shown). While IL-8 release by A549 cells was significantly lower than IL-8 release by human PMNs, IL-8 production from the lung cell line continued to increase in a dose-dependent manner up to 13 nM PVL, the upper limit used in our tests (Fig. 4F). There appeared to be a correlation between the relative amounts of phospho-p38 detected and the amounts of IL-8 released from the corresponding human cells.
PVL stimulates the release of antibacterial factors from mouse and human cells.Being the cells at the forefront of encounters with bacteria, activated neutrophils and epithelial cells can release a number of antibacterial factors, such as defensins and cathelicidins, as a means to kill invading pathogens (26). We examined the effect of PVL on the production of antibacterial factors by comparing the rates of survival of S. aureus in supernatants of mouse cells stimulated with various concentrations of PVL. Concentrations of 1.62 nM and 0.8 nM PVL elicited antibacterial activity from mouse PMNs, as evidenced by a reduction in viable bacterial CFU (Fig. 5A), while a similar activity was induced from mouse lung epithelial cells by 1.62 nM PVL (Fig. 5B). Human PMNs released antibacterial factors upon incubation with similar concentrations of PVL (Fig. 5C, 3.25 nM and 1.62 nM). While activation of mouse cells was induced by a narrow range of PVL concentrations, it is consistent with the concentrations of PVL that can activate human neutrophils, indicating that PVL in the approximate range of 0.8 nM to 3.25 nM can activate multiple cell types to produce antibacterial factors. Human lung A549 cells were activated by PVL to release antibacterial factors over the entire range of concentrations tested (Fig. 5D, 1.62 nM to 13 nM). Thus, cells in the lung likely to encounter infecting S. aureus bacterial cells can produce extracellular antibacterial factors in response to low levels of PVL, which could contribute to host resistance to PVL-producing MRSA strains.
PVL induces the release of antibacterial factors from human and mouse cells (A to D), which occurred inversely with the activation of caspase-3 (E to H). (A to D) Percentages of viable MRSA strain MW2 bacteria after addition of supernatants from the indicated cells that were first incubated with purified PVL, compared to the bacterial count in cells lacking exposure to PVL. Bacterial counts from cell supernatant killing assays are averaged from a minimum of 3 independent experiments. Error bars denote SEM. Statistical analyses were performed by using the t test (**, P < 0.05; *, P < 0.01). (E to H) Production of caspase 3, as determined by immunoblot analysis, from indicated cells 6 h after exposure to indicated concentration of PVL. ♦, PVL concentration that stimulates release of antibacterial factors by cells.
PVL stimulation of the release of antibacterial factors from mouse and human cells is inversely correlated with the activation of caspase 3.Following cellular activation leading to the production or release of proinflammatory mediators and antibacterial factors, a successfully resolved inflammatory response includes apoptosis of the responding cells. Prior work reported that sublytic concentrations of PVL caused apoptosis in human neutrophils and macrophages (15, 45). We looked for the presence of activated caspase 3 protein in lysates of human and mouse cells incubated with various concentrations of PVL as an indicator of the induction of apoptosis. In human and mouse neutrophils, incubation with 6.5 nM PVL achieved maximal levels of caspase 3 activation (Fig. 5E and G). Mouse epithelial cells had higher levels of activated caspase 3 upon incubation with PVL at concentrations of 3.25 nM to 13 nM than with lower concentrations of PVL at 1.62 nM and 0.8 nM (Fig. 5F). Interestingly, we could not detect activated caspase 3 from A549 lung cells at any concentration of PVL tested (Fig. 5H). Overall, we noted an inverse correlation between cellular activation of caspase 3 and secretion of antibacterial factors, indicative of the apoptotic cells being part of the resolution of the inflammatory response.
DISCUSSION
The role of PVL in the pathogenesis of CA-MRSA has been studied in mouse and rabbit models of infection, but overall, these studies have produced conflicting results that might be attributed to the dual properties of PVL, both eliciting protective inflammatory responses and having virulence-promoting cytotoxic activities. These different properties of PVL are likely to manifest under different settings and conditions of infection and might also be attributable to species-specific cellular responses to PVL. Previously published results have shown that PVL-expressing S. aureus strains engender worse pathology in some settings (2, 27, 41), a possible explanation being the pore-forming, leukotoxic properties of PVL that were dominant in those animal models. Others report no difference in outcomes in mouse pneumonia models comparing WT and Δpvl strains (3, 5), although mouse white blood cells appear to be resistant to the cytotoxic effects of PVL (11).
In the study reported here, we found that PVL enhanced host responses to infection, as manifested by the increased mortality observed in mice inoculated with Δpvl strains in comparison to the mortality in mice inoculated with the parental strains, and we showed an association of the host responses with the production of protective proinflammatory molecules by both mouse and human lung epithelial cells and PMNs. This result is in concordance with our prior findings in a study using a foreign-body skin infection model of S. aureus pathogenesis, where we also found that the predominant effect of PVL was to elicit a protective host innate response that was lost in Δpvl strains (46). In our prior studies in the skin, we attributed the predominant effect of PVL as an activator of host innate immunity to the use of a low bacterial inoculum wherein the low levels of PVL that activate protective host responses could be a factor in the early phase of the disease. High S. aureus inocula were needed to establish pneumonia in mice due to the fact that there is a rapid clearance of S. aureus from mouse lungs, with two thirds of the inoculum being cleared within minutes (4; unpublished observations). Thus, despite the use of an initial high inoculum in our model of pneumonia, the beneficial properties of PVL still seemed to prevail. We showed, using an infectious dose just twice that needed to clear >99.99% of bacteria by 48 h, that compared with isogenic Δpvl strains, the PVL-producing parental S. aureus strains were less virulent, as measured by host survival. Our detection of small amounts of PVL (70 ± 13 ng/g) in infected lung tissue is within the range that could have mediated the positive immune response to infection.
Further validation of this finding was obtained in experiments where we increased the production of PVL from the multicopy pOS1-pvl plasmid, which resulted in better or prolonged survival compared with that of mice infected with S. aureus carrying only the cloning vector, pOS1. As the mice infected with S. aureus expressing either indigenous or increased levels of PVL had better survival outcomes than mice infected with S. aureus not producing any PVL, it appears that the inflammatory response associated with the production of PVL was overall beneficial to the host. This benefit was also observed in an analysis of the lung pathology that compared infection with S. aureus strain MW2 that produces PVL to infection with its isogenic Δpvl counterpart, where we found that PVL induced greater inflammation that protected the mice from a lethal outcome.
While the toxicity of PVL to neutrophils from different animal species varies (with human and rabbit neutrophils being highly susceptible to lysis by PVL and mouse neutrophils being resistant in comparison) (31), the results presented here show that multiple cell types from both mice and humans, not limited to primary immune cells, can nonetheless respond to PVL with inflammatory activation. This would suggest that PVL could exert a far wider effect in the host than just lysing PMNs. Interestingly, the responses of the diverse cell types tested here appear to differ from one another, which could be due to properties that are unique to each cell type. We do not contest that the cytotoxicity of PVL to human neutrophils can prove very damaging in certain disease manifestations, but by the same token, we cannot ignore other effects that PVL may have on the host as they could conceivably have a role in disease progression and outcome as well. In support of a possible beneficial role for host reaction to PVL, we demonstrated that PVL is expressed during active infection and can elicit a distinct cytokine response. Additionally, our studies found that the presence of PVL decreased the amount of TNF-α present in infected mouse lungs, which could be a contributing factor in the increased mortality seen in animals infected with the Δpvl strains. The data presented here, together with research from other groups, show that cytokine responses to PVL are predominantly proinflammatory. The PVL-associated downregulation of TNF-α in our model of mouse lung infection is in contrast to the findings of Zivkovic et al. (47), where the administration of purified PVL intranasally to C57BL/6 mice upregulated the production of this cytokine in the lungs. However, the use of large amounts of purified PVL, in contrast to that elaborated by S. aureus during infection, might be expected to result in different cytokine responses. Taken together, these data indicate that the immunomodulatory role of PVL is complex and that responses probably differ when using purified protein versus its occurrence during bacterial infections.
If PVL can augment host immunity under some circumstances, PVL-producing MRSA strains could be less virulent than non-PVL-producing strains in certain disease settings. Indeed, Bubeck Wardenburg et al. also showed that an isogenic Δpvl mutant of S. aureus strain LAC was significantly more virulent than the PVL-producing parent strain in a pneumonia model in BALB/cJ mice (3), whereas in a rat model of S. aureus pneumonia, Montgomery and Daum found a robust induction in the transcription of inflammatory genes following intratracheal instillation of bacteria, although there was no difference between the WT and Δpvl S. aureus LAC strain in this measure (33). The finding of Bubeck-Wardenburg et al. (3) is consistent with our previous findings using a low-inoculum skin abscess model where PVL+ S. aureus bacteria were injected into mice at doses low enough to just produce measurable effects. In that infection model, the proinflammatory properties of PVL appeared to be dominant and we were able to demonstrate the protective properties of PVL (46). A different skin infection model used by other researchers showed that wounds inoculated with clinical S. aureus strains having low PVL expression healed better than strains expressing higher levels of PVL and, interestingly, healed better than wounds in uninfected controls (42). Additionally, a recent report that sublytic amounts of PVL potently induced the formation of neutrophil extracellular traps, along with increased killing of S. aureus, strongly supports our findings as well (38). These experimental data reinforce the idea that the amount of PVL present could determine its effect on host cells and disease progression. Furthermore, maximal PVL expression by S. aureus in the lungs of mice was achieved 72 h postinfection (29), suggesting that the small amounts of PVL present early in infection may function predominantly in activating immunity.
Also in agreement with our data, other researchers have shown that PVL induced visible inflammation in rabbit and mouse lungs, along with the release of various proinflammatory cytokines (10, 47). They attribute the inflammation solely to pulmonary neutrophils and macrophages. However, in two independent studies, only one concentration of PVL was tested in each case (both in the 200 to 300 nM range). This is more than 10-fold in excess of the PVL concentration that would be lytic to rabbit neutrophils, suggesting that cytotoxic effects of PVL will predominate despite an increase in cytokines, thus explaining the increased virulence of PVL+ strains in that rabbit pneumonia model.
As the host immune reaction can serve as a double-edged sword, it will ideally control and eliminate infections and return a tissue to baseline homeostasis, but an overactive response with increased release of toxic substances, such as low-pH vesicles, TNF-α, and reactive oxygen species, can in turn be very damaging to the host, resulting in detrimental outcomes (7, 18). Collectively, the outcomes from multiple PVL models of infections highlight the balance that is needed for efficient infection control with minimal damage to the host, further suggesting an immunomodulatory role for PVL.
We also found that PVL triggered apoptosis in the approximate range of 3.25 nM to 13 nM in 3 of 4 types of cells tested, as had been demonstrated by others using human neutrophils or alveolar macrophages where the release of anti-inflammatory cytokines was also detected (15, 45). Apoptosis is necessary for the resolution of inflammation (13). While neutrophils are essential for the control of staphylococcal infections (39), an overabundance of neutrophils and the resulting increased inflammation is not desirable as it can cause extensive damage to host tissue (7, 18). In addition to destroying excess neutrophils without the release of toxic contents meant for the destruction of phagocytosed pathogens, apoptosis elicits an antagonistic anti-inflammatory signal that further decreases inflammation. Also, apoptosis can be an additional measure of bacterial control. S. aureus internalized within apoptotic bodies can rapidly be destroyed upon phagocytosis by macrophages (12). Overall, these data suggest an immunomodulatory role for PVL by first inducing inflammation by activation of inflammatory responses within host cells, followed by a reduction in inflammation when the activated cells undergo apoptosis. This is consistent with our observation that the lack of cell activation, indicated by an absence of detectable antibacterial factors in cell supernatants, coincided with the activation of caspase 3 in those cells. Strikingly, we found no caspase 3 activation by human cells beyond the higher limit of PVL tested here (13 nM), suggesting that apoptotic induction and inflammation resolution occur in a limited range of PVL concentrations.
With these findings, it might appear difficult to explain the increased occurrence of PVL-producing CA-MRSA in humans and the overall greater virulence of these strains in animal models compared with the virulence of S. aureus strains that do not produce PVL (32) if PVL is mostly beneficial to host resistance. We have hypothesized that antibody to PVL can neutralize its beneficial effects, which we have experimentally verified in our skin infection model, where we found that the antibody to PVL augmented the virulence of PVL+ S. aureus strains but not Δpvl strains, and we also found that it inhibited PVL's activation of antibacterial responses by PMNs (46). As recent findings have shown that the majority of humans develop high levels of cytotoxin-neutralizing antibody to PVL by 7 to 9 years of age, even in the absence of any preceding S. aureus infections (20), it may be that the increased virulence of PVL-producing S. aureus strains results from the host immune response that neutralizes the toxin. Regrettably, we could not adequately test whether antibody to PVL affected immunity in the setting of S. aureus pneumonia, as we were limited by the volume of antisera that could be instilled into the nasal cavities of young mice (typically no more than 20 μl), and antisera delivered by the intraperitoneal route did not effectively reach high levels in the lungs during the early stages of infection when any effect on PVL would be apparent (37). Although Brown et al. (2) reported a protective effect in both lung and skin infection models from immunizing mice against PVL, we do not find these results convincing inasmuch as mice immunized subcutaneously were only protected against skin and not lung infection, while mice immunized intranasally were only protected against lung and not skin infection. This suggests that it was not the systemic antibody or cellular responses generated by immunization that led to protection but, rather, a nonspecific localized effect on activation of innate immunity that lasted sufficiently long to manifest as increased resistance to infection. Along the same lines, these investigators did not report that there was a lack of an effect of immunizing with PVL either against the isogenic Δpvl S. aureus LAC strain or against a non-PVL-producing strain, as we showed in our skin infection model (46).
A PVL concentration gradient is probably present in and around the site of a staphylococcal infection. With its ability to induce a regulated and protective inflammatory response from both PMN and lung cells, as well as its cytotoxic effects on human neutrophils, it appears that PVL has tremendous capacity to modulate host immunity. While the amounts of PVL present can have profoundly different impacts on host cells, an individual's immune reaction in response to PVL will also modulate the course of infection. Indeed, in experiments by others where identical S. aureus infections were established in different breeds of mice, dramatically different results were noted (3, 22, 43). This was in conjunction with vastly different cytokine profiles and levels of neutrophil recruitment elicited from the different mouse strains, which were probably factors that determined the outcomes of infection. Since PVL triggers a reaction from host cells even in the absence of cytolysis, it could have an effect on the outcome of staphylococcal infections by virtue of its ability to influence the cytokine response and downstream immune reactions. Our results indicate that PVL is not a factor in staphylococcal virulence simply by virtue of its cytotoxicity on some cells but, rather, that its immunomodulatory effects can modulate the overall infection, host response, and disease progression in different settings of S. aureus infection.
ACKNOWLEDGMENTS
We thank the Network on Antibiotic Resistance in S. aureus (NARSA) for provision of staphylococcal strains. The following isolates were obtained through the NARSA program, supported under NIAID/NIH contract number HHSN272200700055C: NRS123 (MW2), NRS193, and NRS194. We are extremely grateful to Kathryn Kinzel and Stefanie Gauguet for their contribution. We are also appreciative of D. Missiakas and O. Schneewind for the pOS1 plasmid.
This work was supported by the National Institutes of Health, National Institute of Allergy and Infectious Diseases, grant numbers AI46706 and AI057159, a component of award number U54 AI057159.
The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institute of Allergy and Infectious Diseases or the National Institutes of Health.
FOOTNOTES
- Received 19 January 2012.
- Returned for modification 8 May 2012.
- Accepted 17 May 2012.
- Accepted manuscript posted online 4 June 2012.
- Copyright © 2012, American Society for Microbiology. All Rights Reserved.