Staphylococcus aureus Elicits Marked Alterations in the Airway Proteome during Early Pneumonia

Abstract

Pneumonia caused by Staphylococcus aureus is a growing concern in the health care community. We hypothesized that characterization of the early innate immune response to bacteria in the lungs would provide insight into the mechanisms used by the host to protect itself from infection. An adult mouse model of Staphylococcus aureus pneumonia was utilized to define the early events in the innate immune response and to assess the changes in the airway proteome during the first 6 h of pneumonia. S. aureus actively replicated in the lungs of mice inoculated intranasally under anesthesia to cause significant morbidity and mortality. By 6 h postinoculation, the release of proinflammatory cytokines caused effective recruitment of neutrophils to the airway. Neutrophil influx, loss of alveolar architecture, and consolidated pneumonia were observed histologically 6 h postinoculation. Bronchoalveolar lavage fluids from mice inoculated with phosphate-buffered saline (PBS) or S. aureus were depleted of overabundant proteins and subjected to strong cation exchange fractionation followed by liquid chromatography and tandem mass spectrometry to identify the proteins present in the airway. No significant changes in response to PBS inoculation or 30 min following S. aureus inoculation were observed. However, a dramatic increase in extracellular proteins was observed 6 h postinoculation with S. aureus, with the increase dominated by inflammatory and coagulation proteins. The data presented here provide a comprehensive evaluation of the rapid and vigorous innate immune response mounted in the host airway during the earliest stages of S. aureus pneumonia.

Staphylococcus aureus is a leading cause of hospital-acquired and health care-associated pneumonia and may be increasing in importance as a cause of severe community-acquired pneumonia. In the inpatient setting, it is the most common gram-positive bacterium implicated in cases of ventilator-associated and hospital-acquired pneumonia (1, 9, 31). In addition, S. aureus is a frequent cause of health care-associated pneumonia occurring in residents of long-term-care facilities, individuals recently discharged from acute-care hospitals, and patients receiving outpatient treatment at hospitals and dialysis centers (1, 27, 30). A steady increase in the isolation of methicillin-resistant strains of S. aureus from patients with hospital-acquired pneumonia and, more recently, community-acquired pneumonia underscores the importance of identifying host and bacterial factors that facilitate the progression of staphylococcal pneumonia.

Mice have been used extensively to study pneumonia caused by a variety of bacteria (2, 6, 26, 35, 36, 45, 55, 63, 64). Murine models of airborne infection with S. aureus have been useful in characterizing host responses during the first 4 to 8 h of lung infection but do not mimic the natural route of infection and result in self-limited disease, even in immunocompromised animals (28, 53, 56). In these studies, proinflammatory cytokines and chemokines were released and neutrophils (polymorphonuclear leukocytes [PMNs]) were rapidly recruited to the site of infection; however, the mice were able to clear the infection within 24 to 36 h (53). Bolus infection models in which mice are challenged by intratracheal or intranasal (i.n.) inoculation have been more successful in producing intrapulmonary bacterial replication and host mortality (13, 17, 23, 32, 42, 60). Heyer et al. utilized an infant mouse model of staphylococcal pneumonia, which mimics disease in immunocompromised individuals, in which the mice were anesthetized and infected i.n., leading to 100% morbidity and 30% mortality following inoculation with virulent strains of S. aureus (23). They observed an increase in granulocyte-macrophage colony-stimulating factor (GM-CSF) and an influx of PMNs in the airway. Earlier studies established a lethal S. aureus pneumonia model in adult mice; however, they infected the mice intratracheally, which introduces the additional factor of surgical trauma (13, 42). One goal of the present study was to develop a staphylococcal pneumonia model in immunocompetent adult mice by using a nasal inoculation and aspiration approach that mimics a common route of natural infection in order to provide a system in which to define the earliest events in the host immune response to S. aureus in the airway. Similar models were developed simultaneously by other groups to study the requirement for specific S. aureus virulence factors in pneumonia (32, 60).

Shotgun proteomics has proven to be a very useful tool for determining the global protein profile in a particular organ or body fluid in the context of various disease states. A study by Guo et al. utilized one-dimensional (1D) electrophoresis with mass spectrometry (MS) and two-dimensional liquid chromatography-MS (LC-MS) to define the airway proteome of a healthy mouse (20). In addition, a proteomics approach has been used to define the proteins present in the airway in patients with a variety of conditions (3, 4, 7, 15, 16, 39, 41, 44, 46, 50, 59, 61, 62, 65, 70). However, little is known about the effects of acute infection on the airway proteome and the ways these effects change over time. We hypothesized that early host responses to S. aureus infection of the lung, including changes in the airway proteome, could be critical determinants of the course and severity of pneumonia. To address this, we developed a mouse model of acute staphylococcal pneumonia and utilized cell biological, immunological, and proteomics techniques to examine the host response and changes in the airway proteome during the first 6 h of S. aureus pneumonia. We demonstrate that S. aureus elicits a vigorous airway inflammatory response characterized by the rapid release and influx of inflammatory mediators during the first 6 h of pneumonia. Further, we show that this inflammatory response causes significant changes in the host airway proteome during the development of pneumonia.

MATERIALS AND METHODS

Bacteria and growth conditions.

S. aureus strains RN6390 and JP1 were used in these studies. RN6390 is a commonly used laboratory strain (49) that was kindly provided by David Heinrichs (University of Western Ontario), and JP1 is a human blood isolate obtained from the microbiology laboratory of the Veterans Affairs Puget Sound Health Care System (53). S. aureus was grown in Luria-Bertani (LB) broth at 37°C under aerobic conditions. For mouse infections, bacteria were grown from a frozen stock for 6 to 10 h in LB broth and then diluted 1/100 into fresh LB broth (4:1 flask-to-medium ratio) and grown for an additional 16 to 18 h with shaking (180 rpm). Stationary-phase bacteria were harvested by centrifugation at room temperature, washed twice with endotoxin-free phosphate-buffered saline (PBS) (Mediatech, Herndon, VA), and resuspended in endotoxin-free PBS to the desired concentration as estimated by optical density and confirmed by quantitative plate counting.

Animals.

Specific-pathogen-free male and female C57BL/6 mice, aged 9 to 11 weeks, were purchased from the Jackson Laboratory (Bar Harbor, ME). Animals were group housed in filtered, ventilated cages containing autoclaved bedding and were permitted ad libitum access to sterile food and water. Cage changes and animal handling occurred in a laminar flow hood. All experimental procedures were approved by the Institutional Animal Care and Use Committee of the University of Washington.

Mouse model of pneumonia and tissue harvest.

Mice were anesthetized with isoflurane, held vertically, and inoculated i.n. with S. aureus in 50 μl endotoxin-free PBS. Occasionally, lightly anesthetized mice flipped their heads during inoculation, which may have affected bacterial deposition; these mice were removed from the study. To determine the dose at which S. aureus would replicate in the lungs, 12 mice each were inoculated with 3 × 10⁷, 1 × 10⁸, or 3 × 10⁸ CFU JP1 and monitored at least twice daily. At 30 min after inoculation (all doses), at 24 h, 48 h, and 96 h after inoculation (3 × 10⁷ and 10⁸ CFU), or when mice reached a moribund state (3 × 10⁸ CFU), defined by hunched posture, piloerection, labored breathing, immobility, and loss of resistance to handling, mice were euthanized by intraperitoneal injection of an overdose of pentobarbital. Both lungs were harvested and homogenized for quantitative culture as described previously (53). For analysis of the host response to S. aureus, 7 or 10 mice were inoculated with a dose of 3 × 10⁸ to 5 × 10⁸ CFU JP1 or with endotoxin-free PBS, as described above. At 30 min and 6 h postinoculation, the mice in each group were euthanized, and bronchoalveolar lavage (BAL) was performed as described previously (53, 54). Lungs were inflated in situ to approximately 15 cm pressure with 4% paraformaldehyde and stored at 4°C in the same fixative.

BAL cultures and differential cell counts.

An aliquot of BAL fluid from each animal was removed for quantitative culture, cytokine analysis, and differential counts; the remaining BAL fluid was centrifuged at 300 × g, and the supernatants were frozen at −80°C. The cell pellets were resuspended in RPMI 1640 containing 10% heat-inactivated fetal calf serum (HyClone Laboratories, Logan, UT), and cells were counted with a hemacytometer. Differential cell counts were determined from cytocentrifuge specimens stained with Diff-Quik (Dade-Behring, Dudigen, Switzerland).

Measurement of cytokines.

Levels of immunoreactive tumor necrosis factor alpha (TNF-α), interleukin-1β (IL-1β), macrophage inflammatory protein 2 (MIP-2), keratinocyte-derived chemokine (KC), monocyte chemotactic protein 1, IL-6, IL-10, IL-12p70, IL-17, and GM-CSF were measured with antibody-coated microbeads (R&D Systems, Minneapolis, MN) and a BioPlex analyzer (Bio-Rad, Hercules, CA).

Statistical analysis.

Data are expressed as means ± standard errors of the means. The Mann-Whitney test was performed to determine whether the median times to death for mice at each dose were statistically different. Statistical analysis of cytokines and BAL fluid cells was performed using the Kruskal-Wallis test with Dunn's posttest. A P value of <0.05 was considered significant.

Histopathology.

Paraformaldehyde-fixed lung tissue was embedded in paraffin, sectioned, and stained with hematoxylin and eosin (H&E). A veterinary pathologist examined two to four sections from each lung of mock-infected and infected mice in a manner blinded to time after inoculation and inoculum.

Depletion of BAL fluid.

Twenty mice were inoculated with PBS or S. aureus as described above, and BAL was performed on 10 mice per treatment at 30 min and 6 h postinoculation; the experiment was performed twice. Eukaryotic cells were removed by centrifugation at 300 × g, and BAL fluids were pooled by treatment and time point and frozen at −80°C. After the BAL fluid was thawed, Triton X-100 (TX-100) was added to 0.2%, each sample was vortexed for 15 s, and bacteria were removed by centrifugation at 10,000 × g for 10 min at room temperature. The amount of total protein in each pooled BAL sample was determined by a bicinchoninic acid (BCA) assay (Pierce, Rockford, IL). Each sample was concentrated and exchanged into Agilent buffer A (Agilent Technologies, Inc., Santa Clara, CA) containing 0.2% TX-100 by use of an Amicon Ultra 5000 nominal molecular weight limit spin concentrator (Millipore, Billerica, MA). Two hundred micrograms of protein from each sample was depleted of albumin, transferrin, and immunoglobulin (Ig) by use of a mouse multiple affinity removal system (Agilent). The depletion was performed according to the manufacturer's recommendations except that Agilent buffer A was replaced with Agilent buffer A containing 0.2% TX-100 to reduce protein aggregation. Each depleted BAL sample was concentrated and exchanged into 50 mM ammonium bicarbonate by use of an Amicon Ultra 5000 nominal molecular weight limit spin concentrator (Millipore) and frozen at −80°C.

Fractionation and LC-MS-MS analysis of depleted BAL samples.

Depleted BAL samples were fractionated by strong cation exchange (SCX) or 1D sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) prior to LC-tandem MS (LC-MS-MS). Prior to SCX fractionation, depleted BAL samples were lyophilized and then dissolved in 0.5 ml 50 mM ammonium bicarbonate and reduced in 5 mM Tris(2-carboxyethyl) phosphine (Sigma) for 30 min at 50°C. Cysteines were alkylated with 20 mM iodoacetamide (Sigma) for 60 min at room temperature in the dark. The alkylation reaction was quenched with 20 mM dithiothreitol (Sigma) for 5 min at room temperature. Each sample was digested with 1 μg of sequencing-grade trypsin (Promega, Madison, WI) at pH 8 for 18 h at 37°C. Digestion was confirmed using SDS-PAGE. Each digested sample was fractionated using SCX (polysulfolethyl A; PolyLC, Inc., Columbia, MD). Each sample was brought up to 1 ml of buffer A (5 mM KH₂PO₄, 25% acetonitrile, pH 2.7), and the pH was adjusted to 2 with 10% phosphoric acid. A 40-min gradient was run from 100% buffer A to 100% buffer B (5 mM KH₂PO₄, 25% acetonitrile, 0.35 M KCl, pH 2.7), and absorbance was recorded at 214 nm and 284 nm. Fractions were collected every 2 min and were combined into a total of seven final fractions based upon UV absorbance signal. The seven fractions were desalted using C₁₈ ultramicrospin columns (The Nest Group, Inc., Southborough, MA).

One depleted BAL sample pooled from 10 mice that were mock infected for 30 min was fractionated by 1D SDS-PAGE rather than SCX due to the presence of interfering and unidentifiable contaminants that made LC-MS-MS analysis impossible. The sample was separated in a Novex NuPAGE 4 to 12% bis-Tris gel (Invitrogen, Carlsbad, CA) and stained with Coomassie blue. The lane was excised into seven slices: 10 to 20 kDa, 20 to 40 kDa, 40 to 50 kDa, 50 to 60 kDa, 60 to 85 kDa, 85 to 120 kDa, and 120 to 190 kDa. Each slice was cut into ∼1-mm³ pieces, washed three times with water followed by 50% acetonitrile, and then dehydrated with pure acetonitrile. The Coomassie stain was removed with two washes with 100 mM ammonium bicarbonate mixed 1:1 with acetonitrile. The gel slices were reduced with 10 mM dithiothreitol at 50°C and alkylated with 55 mM iodoacetamide for 45 min in the dark at room temperature. The pieces were dried, rehydrated with 1 μg sequencing-grade trypsin (Promega) in 50 mM ammonium bicarbonate, and incubated for 18 h at 37°C. The digested peptides were extracted from the gel using 20 mM ammonium bicarbonate and acetonitrile washes followed by 5% acetic acid and acetonitrile washes.

An LTQ linear ion trap mass spectrometer (Thermo Finnigan, San Jose, CA) was used with an in-house-fabricated micro-electrospray ionization source and an HP1100 nanoflow solvent delivery system (Agilent). Samples were automatically delivered by an Agilent microwell plate autosampler to a 100-μm-internal-diameter fused silica capillary precolumn packed with 2 cm of 200-Å-pore-size Magic C18AQ material (Michrom Bioresources, Auburn, CA), as described elsewhere (68). The samples were washed with solvent A (0.1% formic acid, 5% acetonitrile) on the precolumn, eluted with a gradient of 10 to 35% solvent B (100% acetonitrile) over 30 min to a 75-μm by 10-cm fused silica capillary column packed with 100-Å-pore-size Magic C18AQ material (Michrom Bioresources), and then delivered into the mass spectrometer at a constant column tip flow rate of 250 nl/min. Eluting peptides were analyzed by micro-LC-MS and data-dependent micro-LC-MS-MS acquisition, selecting three precursor ions for MS with a dynamic exclusion of 1 (21).

Proteomics data analysis.

The MS-MS scans from each LC-MS-MS run were converted from the .RAW file format to mzXML files by use of the program ReAdW.exe (version 1.0; Institute for Systems Biology, Seattle, WA). The database search program X!Tandem (12), included in the Computational Portal and Analysis System (CPAS version 1.4) (48), was used for peptide identification of the MS-MS spectra. The Comet scoring function (38) was used in place of the default X!Tandem scoring function. The following parameters were used in the database search: trypsin enzyme specificity, peptide mass tolerance of 2.5 Da, fragment ion tolerance of 0.5 Da, monoisotopic molecular weight for both peptide and fragment ion masses, b/y ion search, variable modification at M of +15.995, and static modification at C of +57.1. The database was searched against a combined database consisting of the mouse International Protein Index (IPI) version 3.22, S. aureus COL (version NC_002951.1), and a list of contaminants. In addition, randomly reshuffled versions of each database were appended. This resulted in a database of 113,804 sequences being searched.

A composite peptide identification score was generated from the X!Tandem output based upon a combination of the Comet score, delta (relative difference to the second-best match), expectation value, percentage of matching ions, charge state, peptide length, and delta mass (difference between the observed and theoretical masses) by use of the logistic identification of peptide sequences (LIPS) model (24). Experiment-specific peptide identification probabilities were generated from the distribution of reshuffled peptide matches (25). A minimum of 90% identification certainty was used to accept a peptide spectrum identification. This resulted in an overall peptide false-positive identification rate of 0.8% (with a 95% confidence interval of 0.7 to 0.9%) based on reshuffled database matches. Protein identification for each experimental condition was based on four levels of certainty (very high, high, medium, and low). All proteins identified by two or more unique peptides were classified with a “very high” level of certainty. Single-hit proteins were classified in the remaining three certainty levels based on the LIPS model by using peptide identification probability and peptide length (25). The estimated false discovery rates for the four categories are 0.1%, 1.5%, 24%, and 48% (with 95% confidence intervals of 0.0 to 0.4%, 0.3 to 2.9%, 13 to 36%, and 40 to 58%, respectively).

The IPI mouse database contains a large number of redundant peptide sequences, and these redundant sequences generate different randomly reshuffled sequences, resulting in the reshuffled database having more unique sequences. Therefore, the false-positive error rates need to be multiplied by the ratio of unique peptide sequences in the target database to the number of unique sequences in the reshuffled database; otherwise, the number of false positives will be overestimated. The ratio of unique peptide sequences in the target database to the number of unique sequences in the reshuffled database was estimated to be 60%, so the false discovery rate is estimated to be the number of reshuffled peptides divided by the number of target peptides multiplied by the percentage of unique sequences. Confidence intervals were generated by assuming Poisson distributions for the numbers of reshuffled and false target peptide or protein identifications.

Table S1 in the supplemental material contains all of the proteins identified from MS-MS spectra using X!Tandem. Because the same protein can have multiple IPI entries, the information in Table S1 in the supplemental material was condensed into Table S2 in the supplemental material. Condensation of the protein list in Table S1 in the supplemental material was accomplished by searching the following online databases: the mouse IPI database (http://www.ebi.ac.uk/IPI/IPIhelp.html), the Swiss-Prot and TrEMBL databases (www.expasy.org), the Gene Ontology (GO) database (www.geneontology.org), and the PubMed database (www.ncbi.nlm.nih.gov). When a single protein had multiple IPI entries, all of the lines were combined into a single line entry in Table S2 in the supplemental material. The numbers of unique peptides and total peptides identified for each protein were combined in Table S2 in the supplemental material, and the confidence of the protein identification was adjusted if necessary. Only proteins that were identified with high or very high confidence under at least one treatment condition (30 min or 6 h, mock infected or infected) were retained in Table S2 in the supplemental material. All keratin identifications were eliminated as well, because they are likely a result of keratin contamination during sample processing. By use of these criteria, all S. aureus proteins identified in the airway were eliminated due to low confidence of protein identification.

SDS-PAGE and Western immunoblotting.

A 30-μl aliquot of each pooled BAL sample (30 min and 6 h mock infected, 30 min and 6 h infected) was mixed with 10 μl 4× Laemmli buffer (33) and boiled for 5 min. The samples were separated by SDS-10% PAGE. Gels were stained for 16 h with Sypro ruby (Bio-Rad) and destained in methanol-acetic acid-water (10:7:83) for at least 1 h prior to visualization using a gel documentation system (Bio-Rad Laboratories, Inc., Hercules, CA). For Western blotting, proteins were transferred to nitrocellulose membranes by use of a semidry transblotter (Bio-Rad). All incubations were carried out with 5% skim milk and 0.05% Tween 20 (Fisher Scientific, Pittsburgh, PA) in PBS at room temperature. Detecting antibodies were IRDye800-conjugated goat anti-mouse IgG (Rockland Immunochemicals, Inc., Gilbertsville, PA), rabbit anti-human transferrin (Research Diagnostics, Inc., Concord, MA), rabbit anti-mouse matrix metalloproteinase 9 (MMP-9) (Affinity Bioreagents, Golden, CO), rabbit anti-mouse plasminogen (Molecular Innovations, Southfield, MI), and goat anti-mouse C3 (Bethyl Laboratories, Inc., Montgomery, TX). The secondary antibodies were goat anti-rabbit Ig-Alexa Fluor 680 and donkey anti-goat Ig-Alexa Fluor 680 (Invitrogen). Fluorescence was detected using an Odyssey infrared imaging system (LI-COR Biotechnology, Lincoln, NE).

Gelatin zymography.

A 30-μl aliquot of each BAL sample (30 min and 6 h mock infected, 30 min and 6 h infected) was mixed with 10 μl 4× Laemmli buffer without reducing agent (33). Samples were separated by SDS-PAGE in 10% gels containing 1% gelatin (Bio-Rad, Hercules, CA). Following electrophoresis, gels were washed twice at room temperature with 2.5% TX-100 for 30 min each and then incubated for 16 to 18 h at 37°C in buffer composed of 50 mM Tris, pH 7.5, 10 mM CaCl₂, and 150 mM NaCl. Gels were stained with 0.5% Coomassie brilliant blue (Bio-Rad), destained briefly with 40% methanol and 10% acetic acid, and imaged using the gel documentation system described above.

RESULTS AND DISCUSSION

S. aureus replicates in the lungs of mice.

To explore the early host-pathogen interactions that occur during the development of acute S. aureus pneumonia, we sought to develop a mouse model in which the bacteria actively replicate in the lungs to cause pneumonia. Mice were infected i.n. under anesthesia with 3 × 10⁷, 1 × 10⁸, or 3 × 10⁸ CFU of JP1 to identify a dose that would result in bacterial replication in the lungs. All of the mice exhibited signs of illness, including hunched posture, piloerection, labored breathing, immobility, and loss of resistance to handling, by 6 h postinoculation at each dose tested. Mice inoculated with doses of 3 × 10⁷ CFU and 1 × 10⁸ CFU were ill for 24 to 36 h and then cleared the infection. In contrast, a dose of 3 × 10⁸ CFU caused mortality in 71% of the mice, with a median time to death of 32.5 h (P = 0.0008 compared to mice inoculated with 3 × 10⁷ or 1 × 10⁸ CFU). Lungs were harvested from mice sacrificed 0.5 h, 24 h, 48 h, and 96 h postinoculation (doses of 3 × 10⁷ CFU and 1 × 10⁸ CFU) or from mice sacrificed 0.5 h postinoculation or following death due to S. aureus infection (dose of 3 × 10⁸ CFU). Enumeration of S. aureus bacteria in the lungs at 30 min postinoculation revealed deposition of 3 to 10% of the i.n. inoculum (Fig. 1). Bacterial replication was observed in the lungs of mice infected with 3 × 10⁸ CFU, reaching a mean density of 1.54 × 10⁸ ± 1.44 × 10⁸ CFU/lungs at death (compared to a mean deposition of 2.91 × 10⁷ ± 1.29 × 10⁷ CFU/lungs) (Fig. 1). In contrast, bacteria were gradually cleared from the lungs of mice inoculated with 3 × 10⁷ or 1 × 10⁸ CFU (Fig. 1). Similar levels of bacterial clearance, morbidity, and mortality were observed when mice were inoculated with the laboratory strain RN6390 at the same doses (data not shown), indicating that the results obtained with JP1 were not strain specific. These data demonstrate that a dose of 3 × 10⁸ CFU of JP1 or RN6390 was sufficient to cause pneumonia that could result in mortality, while a dose of 1 × 10⁸ CFU or lower resulted in symptoms of pneumonia that lasted for 24 to 36 h, followed by bacterial clearance and disease resolution. This dose-related mortality is similar to what has been reported with intratracheal inoculation of adult mice (13, 42) and, more recently, with i.n. inoculation using exponentially growing S. aureus bacteria (32, 60). A dose of 3 × 10⁸ CFU JP1 was used for all subsequent experiments.

FIG. 1.

S. aureus replicates in the lungs of mice infected with 3 × 10⁸ CFU. Twelve mice at each dose were inoculated i.n. with 3 × 10⁷ CFU (▪), 1 × 10⁸ CFU (▴), or 3 × 10⁸ CFU (▾) S. aureus JP1. Three mice inoculated with 3 × 10⁷ and 1 × 10⁸ CFU were sacrificed at each of four time points (0.5, 24, 48, and 96 h postinoculation). Three mice inoculated with 3 × 10⁸ CFU were sacrificed 0.5 h postinoculation; the remaining mice succumbed to the infection. Bacteria were enumerated from homogenized lungs. Each symbol represents three mice, except 3 × 10⁸ CFU at 17 h (n = 4), 24 h (n = 2), 41 h (n = 2), and 46 h (n = 1).

Proinflammatory cytokines and chemokines recruit neutrophils during early infection.

The concentrations of airway cytokines and chemokines were determined as a measure of the initial inflammatory response to inoculation with S. aureus or PBS. The levels of proinflammatory TNF-α, MIP-2, and KC in the BAL fluid were elevated 30 min postinoculation with S. aureus compared to levels for the mock-infected controls. In addition, the levels of TNF-α, KC, MIP-2, IL-1β, IL-6, and GM-CSF in the airways of infected mice were significantly higher 6 h postinoculation than the levels in BAL fluid from 30-min-infected and 6-h-mock-infected animals (Fig. 2A to F). In contrast, levels of anti-inflammatory IL-10, IL-12p70, IL-17, and gamma interferon were not increased significantly above background in any of the groups during the first 6 h of infection (data not shown). These data show that a measurable proinflammatory cytokine and chemokine response was initiated by 30 min and increased significantly by 6 h postinoculation with S. aureus.

FIG. 2.

Proinflammatory cytokines and chemokines are released and recruit PMNs to the airway in response to S. aureus. Each symbol represents one mouse. The data are combined from three independent experiments with seven to eight mice per experiment. The bar for each data set represents the median value for 21 or 22 mice per condition. Statistical comparisons were made using the Kruskal-Wallis test with Dunn's posttest. SA, S. aureus; MN, mononuclear cells. *, P < 0.05; †, P < 0.01; #, P < 0.001.

One of the major functions of proinflammatory cytokines and chemokines is to recruit PMNs from the bloodstream to the site of an infection. To determine the kinetics of the PMN influx during early S. aureus pneumonia, total cell counts and differential counts of the BAL fluids from infected and mock-infected mice were performed. The total number of BAL fluid cells increased 10-fold by 6 h postinoculation with S. aureus (Fig. 2G) as a result of PMN influx (Fig. 2H). In contrast, the total number of BAL fluid cells from mock-infected mice did not change significantly during the first 6 h, despite a modest PMN response after instillation of PBS into the lung. The numbers of mononuclear cells remained relatively constant (Fig. 2I), regardless of treatment or time point. These data indicate that mice respond rapidly to S. aureus airway challenge by releasing proinflammatory cytokines and chemokines, which act to recruit PMNs to the affected area.

S. aureus causes consolidated pneumonia.

To assess the consequences of S. aureus infection of the airway histologically, lung specimens were stained with H&E and examined microscopically in a blind manner. H&E-stained lung sections taken from mice at 30 min postinoculation with either PBS or S. aureus were histologically similar except for the presence of a few, widely scattered intra-alveolar macrophages containing S. aureus in infected mice (Fig. 3, 30 min S. aureus). Otherwise, the lungs of the 30-min-mock-infected and 30-min-infected mice were normal. In the 6-h-mock-infected animals, minimal neutrophilic inflammation was observed in widely scattered locations (Fig. 3, 6 h PBS). In contrast, lungs from 6-h-infected mice had multiple, frequently confluent foci of inflammation, with various degrees of severity (Fig. 3, 6 h S. aureus). The influx of PMNs into small vessels and capillaries was pronounced (Fig. 3, 6 h S. aureus, ×40 inset), leading to the thickening of alveolar walls and, in more severely affected areas, the diffuse accumulation of PMNs within alveolar spaces. In some areas, consolidation of the air spaces with concomitant loss of alveolar detail was observed. In the most severe foci, fibrin accumulation, thrombosis, and necrosis were evident, with an increase in the number of free S. aureus bacteria. Our data demonstrate that S. aureus infection of the airway results in the rapid development of consolidating pneumonia.

FIG. 3.

Histopathology shows signs of consolidated pneumonia in infected animals but not mock-infected animals. Representative low (×10)- and high (×40)-power histologic sections of lungs from mice infected with S. aureus for 30 min and 6 h or mock infected (PBS) for 6 h. The inset (6 h S. aureus, ×40) shows a smaller vessel that is thrombosed. Bar = 100 μm.

The composition of the airway proteome changes during early pneumonia.

One of the hallmarks of an inflammatory response is an increase in the amount of total protein present in the infected area as a result of the local production and influx of inflammatory mediators in response to cytokine and chemokine recruitment. To determine whether this occurred in our system, the total protein contents of BAL fluids from mock-infected and infected mice were measured using a BCA assay. As shown in Fig. 4A, the amounts of protein in BAL samples at 30 min, regardless of treatment, and in 6-h-mock-infected BAL samples were between 82 and 104 μg/ml. In contrast, more than double that amount (213 μg/ml) was present in the 6-h-infected animals (P < 0.05). SDS-PAGE analysis of BAL samples from mock-infected (30 min and 6 h) and infected (30 min and 6 h) mice indicated the presence of 50-, 67-, and 78-kDa proteins in high abundance (Fig. 4B, lane 1). Based upon molecular mass and immunoblot analyses, these proteins were predicted to be immunoglobulin (Fig. 4C), albumin, and transferrin (Fig. 4D), respectively. An affinity removal system was used to deplete the BAL samples of these proteins, which represent 75 to 85% of the total BAL protein (3), so as to increase the probability of identifying the less-abundant proteins during subsequent MS sampling. SDS-PAGE (Fig. 4B, lane 2) and Western blot (not shown) analyses of native and depleted samples demonstrated that depletion removed all of the detectable immunoglobulin, albumin, and transferrin.

FIG. 4.

Depletion of BAL fluid removes overabundant proteins. (A) Total protein in pooled BAL samples from mock-infected (white bars) and infected (gray bars) mice was measured by BCA assay. Results are from three pooled BAL samples. *, P < 0.05. (B) SDS-PAGE gel stained with Sypro ruby, showing native BAL fluid (lane 1) and depleted BAL fluid (lane 2) from 6-h-infected mice. Equivalent volumes of each sample were separated in the gel. (C and D) Western blots of BAL samples from 30-min-mock-infected (lane 1), 30-min-infected (lane 2), 6-h-mock-infected (lane 3), and 6-h-infected (lane 4) mice, probed with antibodies against mouse immunoglobulin (C) and human transferrin (D). Equivalent volumes of each native BAL sample were separated in the gel prior to immunoblotting. Molecular masses in kilodaltons are shown on the right side of each panel.

A shotgun proteomics approach was utilized to determine how the composition of the airway proteome changed during the course of early pneumonia. The depleted BAL samples were digested with trypsin, and the peptides were subjected to SCX fractionation and LC-MS-MS. The peptide sequences generated by analysis of the MS-MS data were matched to mouse and S. aureus proteins using the combined databases of the mouse IPI and the S. aureus COL proteome. A total of 1,096 mouse and 19 S. aureus proteins were identified in the airways of mice inoculated for 30 min or 6 h with PBS or S. aureus (see Table S1 in the supplemental material). A few peptides were identified as serum albumin, while no peptides were assigned as transferrin or immunoglobulin, showing that the depletion of these three overabundant proteins was remarkably efficient. One potential concern with depletion is that we might remove a significant number of proteins that are associated with albumin, transferrin, or immunoglobulin (either in the BAL fluid or attached to the cartridge resin). Because our goal in this study was to identify potential targets for the molecular characterization of specific host-pathogen interactions, we determined that the advantages of depleting the overabundant proteins from all BAL samples prior to MS-MS analysis outweighed the potential losses that may have occurred.

The list of proteins in Table S1 in the supplemental material was refined into Table S2 in the supplemental material by combining multiple entries for a given protein into a single line of the table, as described in Materials and Methods, so as to obtain a more biologically useful list of proteins. A total of 727 unique host proteins were identified with high or very high confidence in the airway under one or more of the treatment conditions (see Table S2 in the supplemental material). All of the identified S. aureus proteins were disregarded because the confidence of the identifications for these proteins was below the cutoff for further consideration (identification using a single peptide with low to medium confidence, which resulted in false discovery rates of 48 and 24%, respectively). Of the 727 total proteins, 458 (63%) were identified using two or more peptides, which increases the confidence that the protein to which the peptide was assigned was correct (the false discovery rate for identification using at least two peptides was 0.1%, compared with 1.5% for high-confidence identifications using a single peptide).

Relative levels of abundance of airway cytoplasmic and extracellular proteins are reversed as a result of S. aureus infection.

The identified proteins were assigned to the cellular-component and biological-process GO categories. Many of the mouse proteins identified in this proteomics screen were not assigned by GO to cellular-component or biological-process categories, so the assignments were generated manually during the refinement process described above. We chose to analyze the percentages of proteins present in given GO categories because the raw numbers of identified proteins in those categories were not meaningful, as a result of experimental variability in the total numbers of proteins identified per condition. The relative percentages of proteins in different GO categories were more stable and, therefore, more meaningful to compare. There was no difference in the percentages of total proteins assigned to any of the GO subcategories among both 30-min samples (mock infected and infected) and the 6-h-mock-infected sample (analysis not shown). The similarity of the samples from these treatment conditions suggests that any changes in the airway proteome immediately following inoculation with S. aureus and in the first 6 h following PBS inoculation are too subtle to be detected using current methodologies. In particular, the cytokine response observed within 30 min postinoculation with S. aureus (Fig. 2A to F) was not detected on a proteomic level, most likely due to the low molecular weight and low relative abundance of cytokines and chemokines. Thus, the data from the 30-min-mock-infected, 30-min-infected, and 6-h-mock-infected samples were combined into one “control” group to provide for a more rigorous analysis of the inflammatory response elicited by S. aureus 6 h following inoculation (Fig. 5). We also observed that the airway proteome from uninfected mice was nearly identical to the proteome of samples obtained from mice subjected to any of the treatment conditions in the control group (our unpublished observations). In the airway proteome of the control group, 27% of the proteins were extracellular (Fig. 5A) and the remaining 73% were localized to various compartments within the host cell, including the cell membrane (10%). In contrast, 41% of the proteins identified in the 6-h-infected BAL fluid were extracellular, while only 25% were cytoplasmic (compared to 38% in the control airway proteome). The increase in relative abundance of extracellular proteins between the 6-h-infected and control samples (41% versus 27%) is indicative of the local production and release of proinflammatory proteins, as well as the influx of acute-phase reactants from the blood; both events occur rapidly during acute inflammation. The presence of cytoplasmic and other intracellular proteins in the extracellular milieu of the airway likely results from lysis of host cells as a result of normal cell turnover, apoptosis, necrosis, and/or S. aureus-mediated cytolysis (in the case of the 6-h-infected sample).

FIG. 5.

A total of 727 proteins were identified in the airways of control and/or 6-h-infected mice (see Table S2 in the supplemental material for a complete list of the proteins). Shown are cellular-component GO categories for proteins identified in the airways of control (A) and 6-h-infected (B) mice and biological-process GO categories for proteins identified in the airways of control (C) and 6-h-infected (D) mice. A total of 658 proteins were identified in the control samples, and a total of 396 proteins were identified in the 6-h-infected samples.

Inflammatory and coagulation proteins dominate the early response to S. aureus in the airway.

Assignment of the identified proteins into the biological-process GO category revealed an increase in the percentages of inflammatory and coagulation proteins in the airways of 6-h-infected mice compared to those for mice in the control group (5% and 14%, respectively, at 6 h compared with 2% and 6% in control airways) (Fig. 5), which correlates with the influx of inflammatory cells and mediators triggered by the early cytokine response (Fig. 2). The inflammatory proteins identified in the airway are shown in Table 1. Antimicrobial peptides and peptidoglycan recognition proteins have direct antibacterial properties on the S. aureus membrane and cell wall peptidoglycan, respectively, that can ultimately result in bacterial cell lysis. Multimers of cathelicidin antimicrobial peptide (CRAMP) and myeloid bactenecin bind to bacterial membranes and create pores (47). Peptidoglycan recognition proteins exhibit N-acetyl-muramoyl-l-alanine-amidase activity that cleaves the stem peptide of the peptidoglycan, eventually resulting in bacterial cell lysis (51). Mannose binding lectin A (MBL-A) and MBL-C are lectins that have been shown to opsonize S. aureus to promote phagocytosis (40). Nearly every component of the complement cascade was identified in the airway. Thus, the components required for the activation of complement via the classical (immunoglobulins), alternative (C3, factors B and D, and properdin), and lectin (MBL-A and MBL-C) pathways were present in the airway in response to S. aureus infection. While S. aureus is not susceptible to lysis by the membrane attack complex (11), opsonization by any of these proteins facilitates phagocytic uptake of the bacteria by macrophages and PMNs. Complement-mediated opsonization of bacteria may also be augmented by serum amyloid protein P (69), which was also differentially present in the BAL fluids of infected versus mock-infected mice. In a related study, we found that S. aureus was associated with or internalized by 65% of alveolar macrophages within 30 min of infection and by 25% of PMNs after 6 h in the airway; in addition, we found that C3, Ig, and MBL-C were associated with the surface of the bacteria (58). Previous studies by other groups demonstrated that S. aureus was phagocytosed by alveolar macrophages within 30 min following aerosol inoculation as well (18, 28, 34).

TABLE 1.

Inflammatory proteins present in the airways of infected and/or mock-infected animals

Protein	Level of protein identificationa
	Control mice	6-h-infected mice
Advanced glycosylation end product-specific receptor	++	++
Alpha-1-acid glycoprotein	++	++
C4b-binding protein	−	++
Cathelicidin antimicrobial peptide	−	++
CD44 antigen	+	+
Chitinase 3-like 1	++	++
Chitinase 3-like 3	++	++
Complement C1r	−	++
Complement C1s	−	++
Complement C3	++	++
Complement C4b	++	++
Complement C5	++	++
Complement C5a	++	++
Complement C6	++	++
Complement C7	++	++
Complement C8 alpha	++	++
Complement C8 beta	++	++
Complement C8 gamma	++	++
Complement C9	++	++
Complement factor B	++	++
Complement factor D	++	+
Complement factor H	++	++
Complement factor H-related protein	+	−
Complement factor H-related protein	+	+
Complement factor H-related protein C	−	+
Complement factor I	++	++
Complement factor P/properdin	++	++
Gamma interferon-inducible protein 30	++	−
Leucine-rich alpha-2 glycoprotein	++	++
Lipopolysaccharide-binding protein	+	+
Long palate, lung, and nasal epithelium carcinoma-associated protein 1	++	++
Long palate, lung, and nasal epithelium carcinoma-associated protein 3	+	−
Macrophage migration inhibitory factor	++	−
Macrophage stimulatory protein	−	++
Mannose-binding lectin A	−	+
Mannose-binding lectin C	−	++
Matrix metalloproteinase 9	−	++
Major histocompatibility complex	++	++
Myeloid bactenecin	++	++
Neutrophil collagenase/matrix metalloproteinase 8	−	++
Neutrophil gelatinase-associated lipocalin	++	++
Odorant binding protein 1F	++	++
Odorant binding protein 1A	++	++
Palate, lung, and nasal epithelium clone protein	++	+
Parotid secretory protein	++	++
Peptidoglycan recognition protein 1	−	++
Peptidoglycan recognition protein 2	−	++
Polymeric-immunoglobulin receptor	++	++
Secreted phosphoprotein 1	++	++
Secretoglobin family 3A member 1	++	+
Serum amyloid A-1 protein	−	++
Serum amyloid A-2 protein	−	+
Serum amyloid A-4 protein	−	+
Serum amyloid P component	−	++
Small inducible cytokine B15	++	+
Small inducible cytokine subfamily E, member 1	+	−
Whey acidic protein four-disulfide core domain protein 12	+	−

^aBy use of the LIPS model (25), proteins were identified with very high confidence (++) or high confidence (+) or not identified (−). The total number of inflammatory proteins identified in control mice was 41 (6.2% of the total proteins identified), and the total number for 6-h-infected mice was 54 (13.6% of the total proteins identified). The absolute number of proteins identified in a given category is not as useful an indicator of changes in the proteome as is the percentage of total proteins identified in that category because there were differences in the total numbers of proteins identified as a result of differences in MS sampling.

The primary function of MMP-8 and -9 is to facilitate the degradation of extracellular matrix components formed by the host in response to injury; however, both have also been implicated in acute inflammation. Both MMPs are stored in PMN granules and are released during an acute inflammatory response. MMP-8, also known as neutrophil collagenase, promotes balanced PMN recruitment during acute inflammation and resolution of PMN influx during chronic inflammation (57). The proteolytic activity of MMP-9, also known as gelatinase B, cleaves the proforms of the early proinflammatory cytokines IL-8, TNF-α, transforming growth factor β, and IL-1β to their active forms (8, 43). In addition, MMP-9 is known to form complexes with neutrophil gelatinase-associated lipocalin to prevent the autodegradation of MMP-9 (66). Thus, MMPs play an active role in establishing and maintaining an appropriate inflammatory response.

The coagulation proteins identified in the airways of mice inoculated with PBS or S. aureus are shown in Table 2. Interestingly, in addition to their role in recruiting neutrophils and inflammatory mediators to the site of an infection, the proinflammatory cytokines TNF-α, IL-1, and IL-6 have been shown to activate coagulation pathways and attenuate fibrinolytic activity (52), which are hallmarks of alveolar inflammation (10). IL-6 activates bronchoalveolar coagulation via the tissue factor (extrinsic) pathway (37). The increase in IL-6 that we observed during early S. aureus airway infection (Fig. 2E) corresponds with an increase in the abundance of proteins involved in coagulation (Fig. 5D; Table 2) and the appearance of fibrin deposits in the airway, as evidenced histologically (Fig. 3, 6 h S. aureus). We also identified several of the proteins necessary for fibrin accumulation via the contact factor (intrinsic) pathway, including plasma kallikrein and coagulation factors V and X. Heparin cofactor 2 and antithrombin are downstream proteins in both pathways that are involved in the activation of fibrinogen to fibrin. Plasminogen and alpha-2 antiplasmin are fibrinolytic proteins that serve to balance the formation and dissolution of fibrin clots. S. aureus secretes staphylokinase, a protein that activates plasminogen to plasmin; this process can be augmented by CRAMP (5), which was also identified in the BAL fluid from 6-h-infected animals. Further studies to investigate the involvement of staphylokinase, CRAMP, and plasminogen activation in acute staphylococcal pneumonia are ongoing in our laboratory. In addition to their roles in initiating coagulation, proteases of the coagulation system are active in inducing a proinflammatory response (37). Taken together, these proteomics data show an increase in proteins involved in inflammation and coagulation processes during the first 6 h of S. aureus airway infection.

TABLE 2.

Coagulation proteins present in the airways of infected and/or mock-infected animals

Protein	Level of protein identificationa
	Control mice	6-h-infected mice
Alpha-2 antiplasmin	++	++
Antithrombin III	++	++
Coagulation factor V	−	++
Coagulation factor X	−	++
Coagulation factor XIII	−	++
Factor XII	++	++
Fibrinogen, alpha	++	++
Fibrinogen, beta	++	++
Fibrinogen, gamma	++	++
Heparin cofactor 2	++	++
High-molecular-weight kininogen II	++	++
Hyaluronan-binding protein 2	−	++
Kininogen 1	++	++
Plasma kallikrein	++	++
Plasminogen	++	++
Prothrombin	++	++
Vitamin K-dependent protein S	++	−
Vitamin K-dependent protein Z	+	++

^aBy use of the LIPS model (25), proteins were identified with very high confidence (++) or high confidence (+) or not identified (−). The total number of coagulation proteins identified in control mice was 15 (2.3% of the total proteins identified), and the total number for 6-h-infected mice was 18 (4.5% of the total proteins identified). The absolute number of proteins identified in a given category is not as useful an indicator of changes in the proteome as is the percentage of total proteins identified in that category because there were differences in the numbers of total proteins identified as a result of differences in MS sampling.

Alternative approaches confirm changes in airway proteome composition.

Western blot analysis was performed to confirm the presence and relative levels of abundance of complement component C3, plasminogen, and MMP-9 in the different BAL samples. As seen in Fig. 6A, the amount of C3, which was detectable primarily as C3b and iC3b (based upon molecular mass), in both 6-h samples was greater than that in the 30-min samples. In addition, lower-molecular-mass degradation products of C3b were present in the 6-h-infected sample, indicating that cleavage of C3b occurred in the airways of these animals. Plasminogen and MMP-9, both of which have been shown to increase during active infection (5, 10, 14, 19, 22, 67), were also more abundant in the 6-h-infected samples than in the other samples (Fig. 6B and C). Western blot analyses of these proteins were performed on the BAL fluid of mice inoculated with lower doses of S. aureus (3 × 10⁷ and 1 × 10⁸ CFU), with similar results (data not shown). Further evidence for the presence of MMP-9 in the 6-h-infected sample was obtained using gelatin zymography, which is more sensitive than Western blotting (Fig. 6D). The zymogram showed that the predominant form of MMP-9 in BAL fluid was the proform (92 kDa), which is not uncommon, as the activated form (86 kDa) is typically found tightly associated with extracellular matrix components that are not removed during routine lavage. Several MMP-9 complexes were also observed using zymography; MMP-9 is known to form homomultimers as well as heterodimers with neutrophil gelatinase-associated lipocalin (29), which was identified in our proteomics screen (see Table S2 in the supplemental material). These data confirm the proteomics data showing that C3 and plasminogen were present in the airways of mock-infected mice and were increased in BAL fluid from infected mice and that MMP-9 was present only in the airways of mice infected for 6 h. Further, they demonstrate the utility of a shotgun proteomics approach for characterizing changes in the proteome of a biological fluid during the course of an infection.

FIG. 6.

(A to C) Western blotting confirms the presence and relative levels of abundance of C3 (A), plasminogen (B), and MMP-9 (C). Equal volumes of BAL samples from 30-min-mock-infected (lane 1), 30-min-infected (lane 2), 6-h-mock-infected (lane 3), and 6-h-infected (lane 4) mice were separated by SDS-PAGE and subjected to Western immunoblotting. Molecular mass markers in kilodaltons are shown to the right of each blot. (D) Gelatin zymography of BAL fluid from 6-h-infected mice demonstrates the presence of activated MMP-9 (86 kDa), proform (Pro) MMP-9 (92 kDa), and MMP-9 complexes (>100 kDa).

Conclusions.

In this report, we describe marked alterations in the airway proteome that accompany the inflammatory response during the first 6 h of murine staphylococcal pneumonia. We have combined immunology and cell biology techniques with a proteomics approach to define the initial events in S. aureus pneumonia. The data presented here provide a critical first step toward understanding the complex interactions between S. aureus and the airway at the onset of pneumonia. The use of shotgun proteomics provided us with an unparalleled opportunity to define the protein changes within the airway during the first 6 h following staphylococcal challenge. We have demonstrated for the first time that S. aureus elicits a rapid and vigorous inflammatory response within the first 6 h of infection. In fact, as early as 30 min after bacterial inoculation, we observed the release of proinflammatory cytokines and chemokines, which recruited PMNs and antimicrobial mediators to the airway. Within 6 h postinoculation with S. aureus, the airway proteome was altered dramatically to include an increase in antimicrobial peptides, opsonins, proinflammatory mediators, and coagulation proteins, many of which may play key roles in the pathogenesis of acute bacterial pneumonia. These studies provide the foundation for future analyses investigating specific host-pathogen interactions that occur during the early stages of S. aureus pneumonia.