Materials and Methods

Bacterial strains and culture.  S. aureus strains were selected on the basis of clinical relevance and genotype, including multilocus sequence type (MLST or ST), spa type, agr type, staphylococcal cassette chromosome (SCC) mec type, and the presence or absence of PVL (table 1). Unless specified, bacteria were grown in tryptic soy broth (Becton Dickinson) that contained 0.25% glucose. Overnight cultures were diluted 1&rcolon;200 and incubated at 37°C with shaking (250 rpm). S. aureus were harvested at the midexponential phase of growth (OD₆₀₀, 0.75), washed in Dulbecco’s PBS (DPBS; Sigma‐Aldrich), and resuspended in RPMI 1640 medium (Invitrogen) buffered with 10 mmol/L HEPES (RPMI/H [pH 7.2]) or DPBS.

Construction of isogenic lukF/S‐PV deletion mutants.  Standard molecular biological methods, such as the isolation of plasmid DNA and transformation into Escherichia coli, were performed as described elsewhere [15]. Staphylococcal plasmid DNA was isolated using a Qiafilter Plasmid Midi Kit (Qiagen) in accordance with the manufacturer’s instructions, except that bacteria were incubated in P1 buffer (Qiagen) that contained 25 μg/mL lysostaphin for ∼15 min at 37°C. To delete lukF/S‐PV in S. aureus strains MW2 and LAC, polymerase chain reaction (PCR)–amplified regions flanking the lukF/S‐PV locus were cloned into plasmid pBT2spec that contained a spectinomycin resistance cassette [16]. DNA fragments upstream (PCR fragment 1; 800 bp) and downstream (PCR fragment 2; 1900 bp) of lukF/S‐PV were amplified with Ready‐To‐Go PCR beads (Amersham Biosciences) in accordance with the manufacturer’s instructions (figure 1). PCR‐amplified regions flanking the lukF/S‐PV loci and a spectinomycin resistance cassette were cloned into plasmid pBT2 and verified by DNA sequencing (Applied Biosystems). The resulting plasmid, pBTΔlukF/S‐PV, was transformed sequentially into E. coli, S. aureus strain RN4220, and MW2 or LAC. The MW2 and LAC constructs were used for allelic replacement as described elsewhere [16]. The absence of lukF/S‐PV genes and transcripts in lukF/S‐PV isogenic MW2 (MW2Δpvl) and LAC (LACΔpvl) mutant strains were verified using PCR (figure 1) and TaqMan real‐time PCR, respectively.

Mouse infection models.  Female CD1 Swiss and Crl:SKH1‐hrBR hairless mice (outbred and immunocompetent) were purchased from Charles River Laboratories. Mice were housed in microisolator cages and received food and water ad libitum. All studies conformed to guidelines set forth by the National Institutes of Health and were reviewed and approved by the Animal Use Committee at Rocky Mountain Laboratories, National Institute of Allergy and Infectious Diseases (NIAID).

S. aureus (table 1) were cultured to the midexponential phase of growth, washed twice with sterile DPBS, and resuspended in DPBS at 5×10⁷ or 1×10⁸ cfu/100 μL. For the mouse bacteremia studies, CD1 Swiss mice were inoculated with the indicated dose of S. aureus or sterile DPBS via tail‐vein injection [8]. Criteria for determining morbidity or sickness in mice included hunched posture, decreased activity, ruffled fur, and labored breathing. Mice were euthanized if they were unable to eat or drink or if they became immobile. All surviving mice were euthanized at 72 h. At the time of death, selected tissues were fixed in 10% neutral‐buffered formalin, processed for 12 h in an Excelsior Tissue Processor (ThermoShandon), and dehydrated using a graded series of ethanol. Tissues were cleared with Histosolve (ThermoShandon) and infiltrated and embedded in paraffin. Blocks were sectioned at 5 μm, and slides were stained with hematoxylin‐eosin. Samples were examined by a veterinary pathologist (D.J.G.).

We used the abscess model of Bunce et al. [17] to evaluate skin and soft‐tissue infections. Crl:SKH1‐hrBR mice were anesthetized with isoflurane and inoculated by subcutaneous injection in the right flank with 50 μL of DPBS that contained 1×10⁷ S. aureus. The optimal dose of S. aureus for the abscess model was determined by preliminary studies in which small groups of mice (n=3) were inoculated by subcutaneous injection with 1×10⁵ , 1×10⁶ , and 1×10⁷ cfu of LAC (USA300; PVL⁺) and BK11540 (USA300; PVL⁻). Inoculation with 1 × 10⁷ cfu produced consistent abscesses within 4 days that resolved 2 weeks after infection. In addition, the dose chosen for the experiments was at the threshold for inducing dermonecrosis for some of the strains. Mice were weighed before inoculation and at 24‐h intervals for 14 days. The size of abscesses was calculated using the formula for a spherical ellipsoid [υ= (π/6)×l×w ²; where l is length and w is width [17]. Area of abscesses with dermonecrosis was calculated using the formula l×w.

Fifteen mice were used for each strain and/or control in bacteremia and abscess models, unless indicated otherwise, and the MW2 and LAC wild‐type (wt) infections were repeated for the parallel infections with Δpvl strains. One mouse infected subcutaneously with strain BK2394 developed unusual pathological signs and was not included in the data analysis.

Human neutrophil assays.  Polymorphonuclear leukocytes (PMNs or “neutrophils”) were isolated from venous blood from healthy individuals in accordance with a protocol approved by the Institutional Review Board for Human Subjects, NIAID, as described elsewhere [8]. Informed consent was obtained from each subject who participated in the study. Killing of S. aureus by human PMNs was determined as described elsewhere [8]. PMNs (1×10⁶ ) were combined with S. aureus (1×10⁷ ) in 24‐well tissue‐culture plates and centrifuged at 380 g for 8 min to synchronize phagocytosis. Cells were incubated for up to 180 min at 37°C with 5% CO₂. At specified times, PMNs were lysed with 0.1% saponin (for 20 min on ice), plated onto tryptic soy agar, and incubated overnight at 37°C. The percentage of S. aureus survival was calculated using the equation (cfu_+PMN at t/cfu_+PMN at t₀&parr0;×100. The assay measures the total number of viable ingested and uningested bacteria.

Lysis of PMNs by S. aureus was determined as described elsewhere [8, 18]. Phagocytosis was synchronized, and cells were incubated for up to 9 h at 37°C. At the desired times, the lysis of PMNs was determined by the release of lactate dehydrogenase (LDH), using the Cytotoxicity Detection Kit (Roche Applied Sciences). Alternatively, S. aureus were grown in RPMI/H for 6 h at 37°C and centrifuged at 10,000 g for 20 min to remove bacteria. Clarified culture medium was sterilized by filtration and mixed with PMNs (1×10⁶ ) in 96‐well tissue‐culture plates, and PMN lysis was evaluated.

Damage to plasma membranes of human neutrophils (pore formation) was assessed by uptake of ethidium bromide (EtBr) as described by Gauduchon et al. [19]. Briefly, strains were cultured for 18 h at 37°C in 50 mL of YCP medium (3% [wt/vol] yeast extract, 2% Bacto‐Casamino acids, 2% sodium pyruvate, 0.25% Na₂HPO₄, and 0.042% KH₂PO₄ [pH 7.0]) with vigorous shaking using a 500‐mL flask (the flask’s volume&rcolon;culture volume ratio was 10&rcolon;1). Bacteria were removed by centrifugation (2800 g for 10 min at 4°C), and culture medium was sterilized by filtration and stored at −80°C in aliquots until it was used. Culture medium was diluted in RPMI/H and mixed with human PMNs (1×10⁶ ) in the presence of 4 μmoL of EtBr, as described elsewhere [19]. At the desired times, PMNs were analyzed by flow cytometry (FACSCalibur; BD Biosciences). Alternatively, culture medium was tested for its ability to cause the eventual release of neutrophil LDH.

TaqMan real‐time reverse transcriptase–PCR.  MW2 was opsonized in human serum for 30 min and used in PMN phagocytosis assays as described above. At specified times, RNA was isolated using a method published elsewhere [8]. Contaminating chromosomal DNA was removed by DNase treatment (Qiagen). Purified S. aureus RNA was subjected to One‐Step TaqMan real‐time PCR analysis using an ABI 7500 thermocycler (Applied Biosystems). The relative quantification of S. aureus genes was determined by the change in expression of target transcripts relative to gyrB (housekeeping or calibration gene), in accordance with the manufacturer’s protocol (Applied Biosystems Relative Quantification Manual, available at: http://docs.appliedbiosystems.com/genindex.taf, document index number 4347824). Data are expressed as the change in S. aureus transcripts during or after neutrophil phagocytosis relative to S. aureus alone. The sequences of primers and probes for the TaqMan experiments are provided in table 2.

SDS‐PAGE and immunoblotting.  Proteins present in YCP culture medium (prepared for use in the neutrophil pore formation assays as described above) were resolved by 12.5% SDS‐PAGE (Bio‐Rad) and transferred to nitrocellulose (Bio‐Rad). Immunoblots were blocked overnight at 4°C with 10% normal goat serum in DPBS and then for 2 h at room temperature. Rabbit polyclonal antibodies specific for lukS‐PV and lukF‐PV (each at 1 μg/mL) were incubated with immunoblots for 1 h, followed by washing (3 times in DPBS that contained 1% Tween 20) and incubation with donkey anti–rabbit IgG conjugated to horseradish peroxidase (1&rcolon;2500 dilution; BioRad) for 1 h. PVL subunits were visualized with enhanced chemiluminescence (SuperSignal West Pico) and Kodak X‐Omat film (Eastman Kodak).

Statistical analyses.  Mouse survival statistics were performed with a log‐rank test using 12–15 mice/group. Statistics for abscess volume or area of dermonecrosis were performed using a 1‐way analysis of variance (ANOVA) and Bonferroni’s post test for multiple comparisons. A Mann‐Whitney rank sum test was used to compare the number of mice with dermonecrosis over the course of 14 days. Statistics for PMN bactericidal activity and the lysis of human neutrophils were determined using repeated‐measures ANOVA and Bonferroni’s post test. Results are expressed as means ± SEs, unless otherwise indicated. Analyses were done using GraphPad Prism software (version 4.0 for Windows; GraphPad).

Results

PVL and CA‐MRSA virulence.  As a step toward understanding the molecular basis of the pathogenesis of CA‐MRSA, we evaluated the virulence of selected CA‐MRSA or CA–methicillin‐susceptible S. aureus in a mouse sepsis model (figure 2A ). The strains selected were among the most representative pulsed‐field types associated with CA S. aureus infections in the United States [6, 7, 12, 13], including LAC (USA300), MW2 (USA400), and BK648 (USA1000) (table 1). Where possible, PVL⁺ strains were paired with another of similar genetic background (similar spa and agr types and MLST) that lacked PVL or were of special clinical interest (table 1 and figure 2A ). MW2 was the most lethal strain tested. This finding is consistent with the severity of infections often associated with this PVL⁺ CA‐MRSA strain [13, 14, 20]. On the other hand, 3 of 4 PVL⁻ strains tested were either more virulent than or at least as lethal as a genetically similar PVL⁺ strain in the sepsis model (figure 2A ). For example, BK648, a PVL⁻ strain, was significantly more virulent than BK2394 (P<.001, log‐rank test) (figure 2A ). With 1 exception, the survival of mice infected with BK648 was comparable to that of mice infected with MW2 (figure 2A ). We also note that BK9918 produced 10‐fold more lukF‐PV transcript than MW2 in vitro [18] but was far less virulent in the mouse model (0/15 vs. 9/15 mice survived after infection with MW2 and BK9918, respectively) (figure 2A ). Mortality after infection was mirrored closely with gross dissemination to major organs, including the heart and lungs, regardless of PVL status (figure 3). In the aggregate, there was essentially identical mouse survival after bloodstream infection with PVL⁺ and PVL⁻ CA S. aureus strains (figure 2B ).

CA‐MRSA is known for its ability to cause skin and soft‐tissue infections—manifestations typically associated with PVL [7, 9, 12, 21]. Therefore, we evaluated the ability of the CA S. aureus strains to cause abscesses and/or dermonecrosis in a mouse skin‐infection model (figure 4). Unexpectedly, the average abscess volume in mice infected with PVL⁻ strains was slightly greater than that of mice infected with PVL⁺ strains (P<.01 on days 8 and 11) (figure 4C ). Notably, MW2 and LAC, which together cause the greatest number of CA infections, produced, on average, slightly smaller abscesses than PVL⁻ strains (figure 4D ). There were generally more mice with dermonecrotic abscesses after infection with LAC and BK2394, which are both PVL⁺ strains (figure 4E ). However, this observation was offset by the striking finding that the other PVL⁺ strains, including MW2, caused little or no dermonecrosis (figure 4E ). In this regard, differences in the nature of abscesses caused by MW2 and LAC were dramatic (figure 5). Taken together, these data suggest that strain background, rather than the presence of PVL, is the underlying basis for the type and severity of CA‐MRSA infections.

Virulence of isogenic lukS/F‐PV knockout strains (Δpvl) of MW2 and LAC.  Given that genetic makeup of each strain appeared to be crucial to differences in pathogenesis in the mouse infection models, we generated isogenic lukS/F‐PV knockout strains (Δpvl) in LAC and MW2 to directly test the role that PVL plays in disease (figure 1). Consistent with the data from the genetically similar strains (figure 2B ), mouse survival curves for those infected with each wt and Δpvl strain pair were strikingly similar (figure 5A ). That comparable levels of fatal sepsis were caused by MW2 and LAC wt and isogenic Δpvl strain pairs underscores the importance of virulence factors other than PVL in pathogenesis.

We next compared the ability of wt and Δpvl strains to cause abscesses and tissue necrosis directly (figure 5B –5D ). As with the data for PVL⁺ and PVL⁻ strain pairs (figure 4), abscess volumes in wt‐ and Δpvl‐infected mice were similar (figure 5C ), if not slightly larger in the Δpvl‐infected mice (P<.05). Furthermore, numbers of mice with dermonecrosis and areas of abscesses with dermonecrosis were comparable in each group infected with wt and Δpvl strain pairs (figure 5D ). These observations are remarkable, given the accepted view that PVL is the underlying determinant of skin infections caused by CA‐MRSA [7, 9, 12, 21].

Destruction of human neutrophils by Δpvl strains of MW2 and LAC.  PMNs are the key effector cells for defense against S. aureus infection. As such, lysis of human neutrophils by PVL has been proposed as a mechanism for the pathogenesis of CA‐MRSA [19, 22]. To test the involvement of PVL in granulocyte destruction, we measured lysis of human PMNs in vitro after the phagocytosis of wt and isogenic Δpvl strains of MW2 and LAC (figure 6A and 6B ). In accordance with data from the mouse pathogenesis model, there was identical lysis of neutrophils by wt and isogenic Δpvl strains (figure 6A and 6B and figure 7). Notably, PMN lysis was the same with wt and Δpvl strains over time or with varied numbers of bacteria (figure 6A and 6B ). These results were reflected by the ability of conditioned neutrophil assay medium (RPMI/H) from wt and isogenic Δpvl strains to cause comparable neutrophil lysis (figure 6C ). Conditioned neutrophil assay media from MW2 and LAC had varied capacities to cause host cell lysis, which provides support to the idea that factors other than PVL are key to the types of disease caused by each strain. This notion is expanded below (figure 8). Consistent with the neutrophil lysis data, there was similar survival of LAC and MW2 wt and isogenic Δpvl strains after phagocytic interaction with human PMNs (figure 6D ). The finding that wt and Δpvl strains caused identical PMN lysis was not due to a lack of lukS‐PV and lukF‐PV transcripts under our assay conditions, given that there was rapid up‐regulation of each within 5 min after exposure to human neutrophils (transcripts were up‐regulated 2.0‐ and 4.6‐fold for lukS‐PV and lukF‐PV, respectively) (figure 6E ). Notably, genes encoding several other cytolytic toxins, including hlgA, hlgB, hlgC, lukD, and lukE, were up‐regulated after phagocytosis by neutrophils (figure 6E ).

Although the TaqMan data provide support to the idea that CA‐MRSA produces PVL during phagocytosis, the data fail to demonstrate directly that the protein is made by MW2 or LAC wt strains under conditions that promote PMN lysis. To that end, we cultured each strain under conditions known to promote the synthesis of PVL and measured protein using immunoblot analysis (figure 8A ). These results demonstrate that lukS‐PV and lukF‐PV are produced by each wt strain and are secreted into YCP culture medium. Because it is possible that slight differences in plasma membrane damage might not be revealed clearly by the PMN lysis assay, we evaluated the formation of neutrophil membrane pores at sublytic concentrations of culture medium using the method of Gauduchon et al. [19]. Consistent with the PMN lysis data (figure 6A –6C ), the formation of pores in the plasma membrane of human neutrophils due to CA‐MRSA culture medium was essentially identical between LAC and MW2 wt and isogenic Δpvl strains (figure 8B ). Moreover, damage to the plasma membrane was reflected by subsequent PMN lysis (figure 8C ).

Discussion

The dramatic increase in the number of CA‐MRSA infections over the past few years has triggered an intense search for the underlying factors. PVL emerged as the likely virulence determinant because of a strong association indicated by epidemiological studies [4, 5]. PVL is found in virtually all MRSA strains that cause CA infections—most commonly, pulsed‐field types USA300 and USA400—but is almost never found in the prominent hospital‐acquired MRSA that harbor SCCmec types I, II, and III [5, 6, 23, 24]. The widespread increase in the severity of CA‐MRSA disease correlates generally with infection by strains that contain PVL [3, 10, 11, 14]. As such, the toxin is strongly associated with necrotizing pneumonia and skin and soft‐tissue infections, which account for the majority of CA‐MRSA infections [7, 9, 11]. Notably, the ability of CA‐MRSA strains to cause necrosis has been linked to the lysis of neutrophils, which are a primary target of PVL [8, 22]. Given that neutrophils are the main cellular defense against bacterial infections, the destruction of these host cells by CA‐MRSA is likely a key component to disease. Recent studies by Voyich et al. [8] and Genestier et al. [22] have provided strong support to this idea. Genestier et al. further suggested that PVL is the primary cause of neutrophil destruction during human infections and thereby promotes disease such as necrotizing pneumonia [22]. Although evidence supporting the involvement of PVL in CA‐MRSA infections has been circumstantial, the notion that PVL is critical for disease or is even the cause of these infections is widely accepted [4, 5, 9, 21, 25].

We evaluated virulence of the most prominent CA‐MRSA strains and isogenic Δpvl strains, to gain insight into the molecular basis of CA disease and to directly address the role that PVL plays in pathogenesis. The type and severity of disease caused by USA300 (ST8), USA400 (ST1), and USA1000 (ST59) varied significantly, regardless of PVL status. For instance, MW2 (USA400; ST1) was more virulent in the sepsis model than BK2394 (USA1000; ST59), whereas the latter caused far more dermonecrosis, and each is PVL⁺ (compare figures 2A and 4E ). The finding that USA1000 strains were highly virulent in the abscess model is consistent with the noted prevalence of these strains in skin infections among injection drug users [26]. There were also differences in virulence within pulsed‐field types—for example, between BK648 and BK2394—in the sepsis model, and such variance was unrelated to presence or absence of PVL (figure 2A ). The dramatic differences in disease noted in our studies indicates clearly that multiple factors in the overall genetic makeup of each strain, rather than any single determinant, such as PVL, promotes CA‐MRSA infection. These observations may explain the results of Genestier et al. [22], who studied the involvement of PVL using a laboratory strain of S. aureus unrelated to CA‐MRSA.

The present data, obtained using isogenic Δpvl strains of USA300 and USA400, demonstrate directly that the toxin is not the major determinant of disease caused by these prominent CA‐MRSA strains. These results were not unique to mouse models of disease, because we made parallel observations using human neutrophils (figures 6 and 8). Thus, although PVL is a useful marker for strains with the potential to cause severe S. aureus infections, its clinical significance has yet to be determined. It may be that PVL contributes to specific pathologic conditions such as necrotizing pneumonia, a disease not tested in our studies. Future efforts to understand the pathogenesis of CA‐MRSA should focus on identifying the factors (other than PVL) directly responsible for the type and severity of disease caused by these emerging human pathogens.