PcpA Promotes Higher Levels of Infection and Modulates Recruitment of Myeloid-Derived Suppressor Cells during Pneumococcal Pneumonia

Streptococcus pneumoniae, an aerotolerant Gram-positive anerobe, is the leading cause of pneumonia, bacteremia, otitis media, sinusitis, and meningitis worldwide. S. pneumoniae is responsible for ∼850,000 deaths per year in children under the age of 5 y, with its greatest impact in developing countries (1). S. pneumoniae normally colonizes individuals asymptomatically and is considered a commensal in the upper respiratory tract. During the disease process, pneumococci are aspirated into the lower respiratory tract, which can lead to the development of pneumonia (2–4). The exact mechanism of how S. pneumoniae changes from a state of asymptomatically colonizing the upper airways of individuals to causing pneumonia has not been elucidated.

Several surface Ags associated with the pneumococcus aid in immune evasion (5) during pneumococcal infection. Although the roles of capsule and other pneumococcal virulence factors in the host response during infection have been investigated, little is known about the role of PcpA during pneumonia and invasive pneumococcal disease. PcpA (6–9), a choline-binding protein, is considered an excellent candidate for inclusion into a pneumococcal protein vaccine (10). This surface Ag is not produced under the high-Mn²⁺ concentrations present in the nasopharynx (8), and Ab to PcpA does not protect against colonization (9). Thus, immunity to it would not have an effect on colonization and thus would not open up a niche for an opportunistic pathogen lacking PcpA. PcpA contains several leucine-rich repeat regions, which may contribute to its mechanism of virulence. Previous studies have shown that PcpA contributed to adherence of pneumococci to an immortalized alveolar epithelial cell line in vitro (11) (M.M.W.R. Walker, M. Coats, S. Mirza, D. Glover, L. Myers, M. Ochs, and D.E. Briles, manuscript in preparation); our in vivo and in vitro results reported in the present study identify another possible activity of PcpA, the modulation of innate immunity to the benefit of the pneumococcus.

During the initial stage of pneumococcal pneumonia, ineffective phagocytosis by alveolar macrophages and differential proinflammatory cytokine release in lung tissues and airways are observed (12, 13). Pneumococci then rapidly multiply in the alveoli (4–24 h), followed by increased influx of neutrophils into the tissue accompanied by elevated cytokine and leukotriene levels in both the airways and lung tissue. The bulk of the damage to the lung is seen during the third stage (24–48 h), which presents as alveolar injury and interstitial edema and is marked also by regeneration and resolution. This is the window during which bacteria invade the bloodstream, followed by an increase in alveolar monocyte and lymphocyte activity. When bacterial infection is not sufficiently controlled during the final stage (72–96 h), there is further bacterial multiplication, massive tissue damage, and high mortality (12–16).

Another hallmark of the immune response during pneumococcal pneumonia is the rapid induction of apoptosis in macrophages (17). Phagocytosis of apoptotic macrophages is critical for limiting excessive tissue damage and inflammation during infection (18–25). The inhibition of induction of apoptosis in macrophages during bacterial infections has been shown to decrease bacterial clearance. Recently, Steinwede et al. (17) have shown that pneumococcal infection with serotype 19F strain EF3030 induces production of TRAIL by neutrophils and neutrophil-dependent upregulation of DR5 (the mouse homolog to human TRAIL receptor R2) expression on alveolar macrophages. The subsequent binding of TRAIL to the DR5 receptor ultimately led to apoptosis of alveolar macrophages (17). When alveolar macrophages underwent apoptosis instead of necrosis, the environment of the lungs was more anti-inflammatory than proinflammatory, which proved to be protective for the host during pneumococcal pneumonia (17).

Recent studies have also provided evidence for the role of immature myeloid-derived suppressor-like cells (MDSCs) in the resolution of bacterial pneumonia (26). Although MDSCs are known for immune suppression via amino acid depletion and cytokine/reactive free radical–mediated modulation of T cell function, MDSCs have recently been implicated in the efferocytosis of apoptotic neutrophils in a Klebsiella pneumoniae infection model, suggesting an association of their infiltration with resolution of inflammation in bacterial pneumonia. Subsets of MDSC-like cells have also recently been identified as important regulators of allergic airway inflammation (27). However, the defined role for MDSCs in the resolution of S. pneumoniae–associated inflammation has not yet been investigated.

In this study, we show that the expression of the virulence factor PcpA had a strong impact on host defense in two distinct models of pneumococcal lung infection. In a mouse model of bacteremic pneumonia, PcpA expression was associated with increased bacterial burden in the lung, blood, spleen, and liver, increased inflammation, increased IL-6 production in the airways, and decreased neutrophils and recruitment of MDSCs to the lungs during infection. We also observed a significant decrease in TRAIL production, DR5 expression, and apoptosis of macrophages in the presence of PcpA. Taken together, our studies suggest that PcpA is likely to have a role in innate immune modulation during pneumococcal pneumonia.

All mice used in this study were CBA/CaHN-Btk^xid/J (CBA/N) (The Jackson Laboratory, Bar Harbor, ME). Female mice were purchased at 6–8 wk of age and rested for 7–14 d before use in specific pathogen-free facilities at the University of Alabama at Birmingham. All studies were carried out under the supervision of veterinarians using protocols approved by the Institutional Animal Care and Use Committee of the University of Alabama at Birmingham.

S. pneumoniae strains TIGR4, EF3030, and their respective pcpA⁻ mutants, JEN11 and JEN18 (8), were grown to an OD₆₀₀ of ∼0.3 at 37°C in Todd–Hewitt broth with 0.5% yeast extract (THY; Fisher Scientific, Pittsburgh, PA) or on blood agar plates (Becton Dickinson, Sparks, MD). Previous expression studies have shown that an OD₆₀₀ of ∼0.3 yields optimal PcpA expression. All infection stocks were started from plates of −80°C parent stocks of each strain in 10% glycerol. PcpA expression of TIGR4, EF3030, and their respective mutants was assayed by staining with polyclonal anti-PcpA Ab, and flow cytometry analysis was performed. Flow cytometry analysis (11, 28) confirmed expression of PcpA by both wild-type (WT) strains and its absence in the mutant strains. Supplemental Fig. 1 shows representative data for TIGR4 and EF3030 and their pcpA mutants JEN11 and JEN18, respectively. The WT and mutants were grown in THY or blood agar plates with the addition of erythromycin (Fisher Scientific) for growth of the mutants at a concentration of 0.3 μg/ml. The lists of WT and mutant bacterial strains as well as the plasmid constructs are provided in Supplemental Fig. 1. The construction of the mutant strains has been described previously (8).

Pneumococci were grown in high- and low-Mn²⁺ THY for these studies. PcpA is under control of PsaR, a manganese-dependent repressor, and therefore pneumococci were grown in low-Mn²⁺ THY to induce expression of PcpA prior to infection (8, 9). High Mn²⁺ THY is defined as regular Todd–Hewitt with 0.5% yeast extract. Low Mn²⁺ THY was made by depleting manganese out of the media and adding it back at a known concentration. Low Mn²⁺ was prepared according to the manufacturer’s instructions with Chelex 100 (2% w/v; Sigma-Aldrich, St. Louis, MO) being added prior to autoclaving. After autoclaving, low-Mn²⁺ solution was stirred overnight at room temperature and then filter sterilized. Prior to use, ZnCl₂, MgCl₂, CaCl₂, and FeSO₄ were added at a concentration of 1 mM each and MnSO₄ (Fisher Scientific) was added to a concentration of 0.1 μM. Infection stocks were made by growing strains to 0.3 OD₆₀₀ and storing bacteria in 10% glycerol in cryovials at −80°C. At least 2 wk after freezer stocks were made, CFU in a representative vial were quantitated by serial dilution and plating on blood agar.

The expression of PcpA in different manganese conditions was assessed through flow cytometry. TIGR4, EF3030, JEN11, and JEN18 were grown up in low- and high-Mn²⁺ media to an OD₆₀₀ of 0.2, diluted back to an OD₆₀₀ of 0.1, and grown up to an OD₆₀₀ of 0.3. Bacterial cultures were then aliquoted into 250-μl samples and spun at 8000 rpm for 5 min. After removal of the supernatant, the pellet was resuspended in 100 μl primary Ab diluted 1:100 in 1% BSA in PBS. Cells were incubated with polyclonal anti-rabbit PcpA for 30 min at room temperature.

After incubation, cells were washed with 750 μl 1% BSA in PBS, spun down, and supernatant was removed. The pellet was resuspended in 100 μl FITC-labeled secondary Ab diluted 1:100 in 1% BSA in PBS and underwent incubation for 30 min at room temperature. After incubation, cells were washed with 1% BSA in PBS, spun down, and supernatant was removed. Pellets were resuspended in 125 μl 1% BSA in PBS and 125 μl 2–4% paraformaldehyde in a FACS tube. Samples were analyzed within 2 d using a BD LSR II (BD Biosciences). Expression of PcpA was described as fold change in the mean fluorescence intensity (MFI) of the sample over MFI of the negative control, which were cells that had only the secondary Ab (FITC-labeled goat anti-rabbit) added to them. FlowJo (Tree Star, Ashland, OR) was used to analyze dot plots and statistics on MFI of samples.

To determine in vivo expression of PcpA, a PcpA capture ELISA assay was carried out as follows. Microtiter 96-well plates (Nunc, Weisbaden, Germany) were coated with an optimized combination of three mouse mAbs (20.10, 163, and 2.16, all of isotype IgG1) to PcpA (R.W. Widener, J.A. Grubbs, J.D. King, P. Schachern, W.E. Swords, and D.E. Briles, manuscript in preparation). The plates were coated with 2 μg/ml total mAb in 1× PBS (Fisher Scientific) and incubated at 4°C overnight. As negative controls each assay contained mAb-coated wells to which no sample was added. To evaluate the specificity of PcpA detection, PcpA-uncoated wells were included. After coating, plates were washed three times with ELISA wash buffer (1× PBS with 0.05% Tween 20). Each well was blocked with 150 μl 1% BSA/0.2% casein in 1× PBS for 1 h and then washed three times with ELISA wash buffer. Samples of bronchoalveolar lavage (BAL) fluid from infected mice and a standard with a known concentration of rPcpA were then added to the plate and incubated at room temperature for 1 h and washed three times with ELISA wash buffer. Plates were next incubated at room temperature for 1 h with rabbit anti-PcpA IgG and then washed three times with ELISA wash buffer. Plates were then incubated at room temperature for 1 h with biotin-conjugated donkey anti-rabbit Ig IgG (H+L) (SouthernBiotech, Birmingham, AL) and then washed three times with ELISA wash buffer. Plates were then incubated at room temperature for 1 h with strepavidin–alkaline phosphatase (SouthernBiotech) and washed three times with ELISA wash buffer. The plates were next developed with p-nitrophenyl phosphate (Sigma-Aldrich) and absorbance was read at 405 nm. Data were reported as nanograms PcpA per milliliter.

The bacteremic pneumonia model was established by intratracheally (i.t.) infecting 8- to 12–wk-old male CBA/N mice with the capsular type 4 pneumococcal strain, TIGR4. This model is characterized by severe lung infection, development of bacteremia and sepsis within 24 h, and mortality within 48 h postinfection.

The focal pneumonia model was established by i.t. infecting 8- to 12-wk-old male CBA/N mice with the capsular type 19F pneumococcal strain, EF3030. This model is characterized by self-resolving of lung infection generally lacking any detectable bacteremia (29).

Mice were anesthetized with isoflurane (Henry Schein Animal Health, Dublin, OH). After anesthetization, mice were i.t. infected with between 3.0 and 5.0 × 10⁶ CFU TIGR4, EF3030, or their respective pcpA⁻ mutants grown in low manganese in 40 μl lactated Ringer’s solution (Hospira, Lake Forest, IL). At 12, 24, 36, and 96 h postinfection, mice were euthanized, and BAL fluid, lung tissue, blood, liver, and spleen were harvested. Prior to whole lungs being harvested, the trachea was cut at the top of the larynx and lungs were lavaged with 1 ml lactated Ringer’s solution. Lungs, spleen, and liver were then harvested and homogenized in 1 ml Ringer’s solution in stomacher bags. The BAL and homogenized tissue samples were serially diluted and plated on blood agar plates with gentamicin (Lonza, Walkersville, MD) or erythromycin. Plates were incubated at 37°C under 5% CO₂ overnight and used to calculate CFU recovery.

Eight- to 12-wk-old CBA/N mice were infected as previously described and were euthanized at 6, 12, 24, and 48 h time points after infection. Prior to removal of the lung tissues, 1 ml 10% formalin solution (Fisher Scientific) was introduced into the lungs via the cannulated trachea. Inflated lungs were surgically removed, immediately submerged into 10% formalin, and, after fixation, embedded in paraffin. Lung sections were stained with H&E (Fisher Scientific), mounted, and examined microscopically with a ×20 objective. Lung sections were scored for inflammation and congestion. Mouse lung tissues were fixed in formalin, processed, sectioned (5 μm thick), and stained with H&E. Each lung lobe tissue was examined microscopically by a pathologist for the presence of inflammatory infiltrate and congestion, and further graded. Inflammation was characterized by the presence of lymphocytic infiltrates: grade 1, scattered lymphocytes in perivascular and peribronchial area; grade 2, dense lymphocytes in perivascular and peribronchial area; grade 3, diffuse lymphocytic infiltrate involving portion of a lobe. Congestion was characterized by dense alveoli with blood-filled capillaries: grade 1, focal congestion (less than half of lobe); grade 2, diffuse congestion (more than half of lobe).

After bacterial challenge as described in the in vivo mouse studies, CBA/N mice were anesthetized using a 50–100 μl 1:1 ketamine (VetOne, University of Alabama at Birmingham Veterinary Care)/xylazine (Lloyd Laboratories, Shenandoah, IA) mixture. Mice were euthanized by cardiac perfusion with 5–7 ml chilled 1× PBS (Fisher Scientific) followed by bilateral thoracotomy. The BAL was produced by lavaging lungs four times with 1 ml chilled PBS. The first 1 ml lavage fluid collected was centrifuged and the clarified supernatant was used for quantitating CFU recovery and measuring cytokine production. We enumerated infiltrating immune cells recruited to the airways during infection by pooling cellular components from all the lavage aspirates. The whole lungs were then harvested, placed in complete Iscove’s media (Mediatech, Manassas, VA), and minced into fine pieces. Collagenase B (Roche Diagnostics, Indianapolis, IN) and DNAse (Sigma-Aldrich) were added to the minced lung mixture and incubated at 37°C for 30 min as described before (27). After incubation, digested lung tissue extract was then strained through a 40-μm cell strainer as previously described (27, 30).

Both the BAL and lung homogenates were spun down at 1400 rpm for 5 min at 4°C. Supernatants were removed and the pellet was washed twice with 1 ml 1× PBS. Samples were then blocked with anti-CD16/CD32 (2.4G2, Thermo Fisher Scientific, Waltham, MA) in 1 ml 3% BSA (Fisher Scientific) in PBS for 30 min at 4°C. Cells were stained with rat anti-mouse Abs for CD11b (M1/70, eBioscience, San Diego, CA), Ly6G (Gr-1) (RB6-8C5, eBioscience), Ly6C (HK1.4, eBioscience), F4/80 (BM8, eBioscience), and annexin V⁺ (eBioscience). Lung cells and BAL were first gated on live cells. Lung and BAL neutrophils were identified as Gr-1⁺CD11b⁺Ly-6G⁺F4/80⁻. Alveolar macrophages were identified as cells that were CD11b⁺CD11c⁺F4/80⁺. Polymorphonuclear cells (PMNs) were defined as cells that were CD11b⁺, Ly-6G⁺, F4/80⁻. MDSCs were identified as Gr-1⁺CD11b⁺Ly-6C⁺F4/80⁺ cells. Samples were fixed with 2–4% paraformaldehyde (Fisher Scientific), and data acquisition for flow cytometry was performed using a BD LSR II flow cytometer (BD Biosciences). Flow cytometry data were imported and differences in cell recruitment were analyzed using FlowJo (Tree Star). Absolute cell numbers were calculated as total cells in tissue × percentage gated × percentage positive.

The effects of in vivo depletion of MDSCs were examined in the bacteremic pneumonia model. After bacterial challenge with TIGR4 and JEN11 strains as described in the in vivo mouse studies, CBA/N mice received anti–Gr-1 Ab (150 μg/100 μl) (31) or IgG control (Bio X Cell, West Lebanon, NH) i.p. 3 h postinfection. At 24 h postinfection, BAL fluid and lung homogenates were harvested from mice and CFU were quantitated as described before.

The BAL fluid from the in vivo mouse experiments taken at various time points was used in ELISA experiments to quantitate IL-6, TNF-α, and TRAIL levels in the airways of CBA/N mice in our bacteremic and focal pneumonia models. IL-6, TNF-α, and TRAIL levels were measured using an IL-6 Quantikine ELISA (R&D Systems, Minneapolis, MN), a TNF-α Quantikine ELISA (R&D Systems), and a TRAIL-R1 ELISA (MyBioSource, San Diego, CA) following the manufacturers’ instructions.

Statistical analysis was performed using GraphPad Prism software. All animal experiments contained five to seven mice per group. Each figure represents the pooled data from at least two independent experiments. Differences in the median between two groups were compared by the Mann–Whitney two-sample rank test. A p value <0.05 was considered significant.

We used two models of lung infection, septic pneumonia and focal pneumonia, to investigate the effects of PcpA expression in vivo. In the septic pneumonia model, CBA/N mice were i.t. infected with 10⁶ CFU TIGR4 (serotype 4) or JEN11 (a pcpA mutant of TIGR4) grown in low-Mn²⁺ THY. S. pneumoniae is thought to generally be a commensal of the human nasopharynx. The exact mechanism of how this microorganism transitions from the nasopharynx, where it asymptomatically colonizes, to a more invasive site is unknown. The concentration of manganese within the mucosa tends to be in the micromolar range (32) and has been measured in the saliva at 36 μM. In the internal sites such as the blood and lungs, the concentration of manganese is in the nanomolar range (32). The lung and blood have lower concentrations of manganese, and PcpA expression is de-repressed in this particular environment, allowing PcpA to be a virulence factor in invasive infection in the lung and blood (9). Our preliminary studies indicated that within the first 12 h, the numbers of CFU in the lung, blood, spleen, and liver remained low and were not affected by mutations in pcpA. In this model WT TIGR4 pneumococci usually produced lung inflammation and damage, and from 40 to 72 h most of the mice became moribund or died. At the 12 h time point, we looked at the effects of PcpA on the bacterial burden recovered in the lung tissue of pneumococcal-infected mice and found no significant differences in CFU recovered between TIGR4 (PcpA⁺) and JEN11 (PcpA⁻) (Fig. 1A). We also found that at 12 h postinfection, there was no significant difference between WT and pcpA⁻ TIGR4 pneumococci in terms of bacterial load in the airways (Fig. 1B). However, at 24 h after pneumococcal challenge we saw a significant increase (**p = 0.0012, *p = 0.0157) in bacterial burden in both the lungs and airways in the presence of PcpA (Fig. 1). This difference remained significant in the lung tissue at 36 h postinfection (*p = 0.0262) but not in the BAL (Fig. 1).

The focal lung infection model in which CBA/N mice were infected i.t. with EF3030 strain produced a self-resolving focal pneumonia infection with only rare incidences of bacteremia (29). It is characteristic of 19F strains such as EF3030 not to cause sepsis (33) and, when present in the lung, to be restricted to the lungs and upper respiratory tract (29). Because the lung disease was less severe in this model, we were able to analyze the lungs at later time points than with the TIGR4 infections. We infected CBA/N mice i.t. infected with 10⁶ CFU EF3030 (WT) or JEN18 (pcpA⁻ mutant) and harvested the lungs and BAL at 12, 24, and 96 h postinfection. Surprisingly, the PcpA-dependent effect on bacterial burden of EF3030 in the lungs appeared within the first 12 h after challenge with S. pneumoniae. At the 12 h time point we saw greater clearance of bacteria from the lung tissue (**p = 0.0022) and the BAL (**p = 0.0022) in the absence of PcpA (Fig. 2). This difference between EF3030 and its pcpA mutant remained significant at 24 h (*p = 0.0157) but not at 96 h postinfection (Fig. 2), as the infections were largely resolved by that time. Even though the course of disease with both pathogens was very different, our data revealed a consistent effect with higher CFU levels with each WT strain compared with their pcpA mutants, suggesting that this effect was PcpA-dependent.

There was no effect of the pcpA mutation on the survival time of EF3030 because mice do not die at the challenge dose used; instead, they have a self-resolving infection in the lungs. All mice infected with TIGR4 died. The median survival time of five mice infected with TIGR4 was 29.5 h and the longest surviving mouse died at 68 h. For JEN11-infected mice the median survival was 168 h. One of the five mice died at 101 h and four died at 168 h (p = 0.0075, Mann–Whitney U test).

To more completely understand the role of PcpA during infection, we looked at its effects on levels of the bacteria in the lungs to blood, liver, and spleen. CBA/N mice were infected i.t. with 10⁶ CFU TIGR4 or JEN11 (pcpA⁻) in our bacteremic pneumonia model, and the bacterial burdens in the blood, liver, and spleen were quantitated at 12, 24, and 36 h postinfection. Twenty-four hours after pneumococcal challenge, in the presence of PcpA expression, we observed a significant increase in CFU recovered from blood (*p = 0.0111), spleen (***p = 0.0019), and liver (***p = 0.0021) (Fig. 3). These observations were consistent with the findings at the same time points in the lung tissue and BAL of these same mice (Fig. 1). Thus, PcpA expression strongly enhanced the levels of pneumococci in the lungs, which may have accounted for their higher levels in the blood, liver, and spleen.

As expected, virtually no CFU were noted in the blood, spleen, and liver of mice in the focal pneumonia model at any time point (data not shown). Only 1 mouse out of 36 infected with EF3030, or its pcpA⁻ mutant JEN18, yielded any detectable CFU, and these were seen only in the liver (log 3.3 CFU) at the 12 h time point (data not shown). We further validated in vivo PcpA expression using a PcpA capture ELISA assay. As shown in Supplemental Fig. 2, PcpA was detected in the BAL of mice infected with TIGR4 and EF3030 pneumococci but not in the BAL of mice infected with the mutants JEN11 or JEN18. PcpA is not detectable by our PcpA ELISA in lysates of JEN11 or JEN18, nor is it detectable on the surface of these strains of pneumococci with rabbit anti-PcpA sera and secondary FITC-labeled goat-anti-rabbit antiserum by flow cytometry when these strains are grown in normal THY media or low-manganese THY. It is measurable by these methods on the TIGR4 and EF3030 background strains.

The above studies led us to conclude that PcpA had no significant effects in vivo during the first few hours of lung infection, at least in the case of infection with the TIGR4 strain. We then evaluated whether PcpA expression modulated overall inflammation in the lungs of mice during pneumococcal infection using our bacteremic pneumonia model in which CBA/N mice were i.t. challenged with 10⁶ CFU TIGR4 or JEN11. To ensure that the time course of PcpA-mediated effects on inflammation was adequately studied, we examined 6, 12, 24, and 48 h time points. This time frame allowed us to observe when mice were first developing signs of pneumonia before they start to become septic at ∼40 h. Studies of later time points were not possible because many of the mice infected i.t. with TIGR4 started to die shortly after 48 h postinfection. At a sublethal dose of 5 × 10⁵ CFU TIGR4, we did not detect any sign of infection and inflammation within either group (data not shown).

However, using a challenge dose of 1.0 × 10⁶ CFU/mouse, histological evidence of lung infection became apparent (Fig. 4A). At the 6 h time point inflammation was apparent, but there were no observable differences between the mice infected with the WT and mutant TIGR4 pneumococci (Fig. 4B). By 12 h, lung infection was more pronounced in the mice infected with TIGR4 than those infected with the pcpA mutant of TIGR4 (Fig. 4). This was apparent from the higher histologic grade of inflammation (increased cellular infiltrates containing mainly lymphocytes and neutrophils) (Fig. 4B) and the presence of more inflamed lung lobes per lung when the infection was with TIGR4 versus its pcpA mutant (Fig. 4C, *p = 0.0248). At 24 h, the WT-infected mice continued to have more cellular infiltrates than did the pcpA mutant–infected mice but the difference was not statistically significant (Fig. 4B). At 24 h there was also no longer a difference in numbers of lobes infected by TIGR4 versus its pcpA mutant (Fig. 4C).

Another parameter we used to evaluate the effects of PcpA expression on lung disease during pneumonia was measurement of alveolar congestion, defined as the thickening and filling with blood of alveolar septa usually occurring during the initial stages of pneumonia. Despite the increased cellular infiltration at 12 h and possibly at 24 h (Fig. 4B) in the presence of PcpA, there was significantly less alveolar congestion at 24 h with WT than the pcpA mutant challenge strain (Fig. 4D, *p = 0.0397).

TRAIL is indicative of a potential for triggering anti-inflammatory apoptosis in macrophages (17). Because we saw an increase in inflammation in the presence of PcpA, we then wanted to determine whether the reduced lung inflammation in infections with the pcpA mutant was associated with levels of TRAIL in their lungs. Such an association could be indicative of differences in the potential for triggering anti-inflammatory apoptosis of macrophages in the presence or absence of PcpA. By 12 h postinfection, in the absence of PcpA (in JEN11-infected mice) there were significantly higher levels of TRAIL (**p = 0.0023) in the airways than in mice infected with TIGR4 (Fig. 5A). At 24 h, the levels of TRAIL were reduced to baseline in the airways in both two groups. As PMNs have been reported to be significant contributors of TRAIL, we quantitated the absolute numbers of infiltrating PMNs in the lung tissue. The peak of TRAIL production in the airways appears to positively correlate with the peak of PMN recruitment in the lung tissue (Fig. 5B), which is consistent with previous studies using WT pneumococci (17).

IL-6 and TNF-α are two of the major proinflammatory cytokines that play a role in the pathogenesis of pneumococcal infections in mouse models (12). IL-6 and TNF-α levels are elevated during early infection in the airways, and these levels are reported to increase between 4 and 24 h postinfection (12). We determined whether the decreased TRAIL expression in WT infections would be associated with higher levels of IL-6 and TNF-α. Such a result might be expected in a proinflammatory environment observed in the WT infection groups. At 24 h postinfection, in the presence of PcpA, significantly higher levels of IL-6 (Fig. 5C, ***p = 0.0001) and TNF-α (*p = 0.028, Supplemental Fig. 3) production were observed in the airways of CBA/N mice in the bacteremic model. We observed the same trend in the airways of infected mice in the focal pneumonia model. Mice infected with EF3030 (WT) had higher levels of IL-6 (**p = 0.0022) in their airways at 12 h than did the mutant-infected mice (Fig. 5D). The increased bacterial burden and the resulting lung inflammation may contribute to the increased IL-6 levels in the airways of mice infected with EF3030. The peak of IL-6 production also positively correlated with peak inflammation scores in histological experiments in the lungs of CBA/N mice. Taken together, in the airways of mice infected with pneumococci expressing PcpA, there were significantly lower levels of TRAIL, indicative of less apoptotic potential, and significantly higher levels of IL-6 and TNF-α production, indicative of proinflammatory airway milieu.

Apoptosis of alveolar macrophages in the lung after pneumococcal challenge is a hallmark of the disease process (17, 19–23). The induction of apoptosis in alveolar macrophages limits the inflammatory environment of the infected lungs and promotes tissue repair. Because we observed a significant decrease in TRAIL production in the presence of PcpA, we wanted to determine whether that translated into a decrease in apoptotic immune cells in the lung tissue and airways during pneumococcal pneumonia. At 12 h postinfection, CBA/N mice infected with TIGR4 had reduced infiltration of total macrophages and apoptotic cells as measured by annexin V⁺ staining in lung homogenates compared with JEN11-infected mice, although the reduction was not statistically significant (Fig. 6A, 6B). By 24 h, there were significantly more total macrophages (*p = 0.0152) and apoptotic cells (*p = 0.0152) in the lungs of mutant-infected mice compared with the lungs of TIGR4-infected mice (Fig. 6A, 6B). In the airways, CBA/N mice infected with TIGR4 had a significantly reduced infiltration of macrophages at 12 h (*p = 0.0411) but equivalent numbers at 24 h compared with JEN11-infected mice (Fig. 6C). As observed in the lungs, a significantly higher number (*p = 0.0152) of apoptotic cells was observed in the airways of mutant-infected mice by the 24 h time point (Fig. 6D). We also observed that PcpA expression reduced significantly the total numbers of MDSCs recruited to the lung tissue and airways (*p = 0.0411) (Fig. 6E, 6F). At 24 h postinfection, WT-infected mice had significantly fewer total numbers of DR5⁺ (*p = 0.0152) macrophages in the BAL (data not shown). Taken together, the presence of PcpA in the lungs of CBA/N mice in our bacteremic pneumonia model leads to suppression of immune regulatory cell recruitment and reduction of apoptotic cells, making it more likely to promote an exacerbated proinflammatory environment.

To further investigate whether a significant increase in innate immune cell recruitment, specifically MDSCs, in the lungs of pcpA⁻ mutant–infected CBA/N mice were contributing to clearance of the pneumococci, and to determine whether monocytic or granulocytic MDSCs contribute to the clearance of pneumococci, 3 h postinfection we instilled anti–Gr-1 or an IgG isotype control Ab i.p. in mice infected with either WT pneumococci or the pcpA⁻ mutant. Gr-1 is a GPI myeloid differentiation marker (also known as Ly6G) whose cellular expression is restricted to monocytes in the bone and neutrophils in the peripheral organs (34). MDSCs are heterogeneous immature myeloid cells including immunosuppressive monocytic and granulocytic subsets. Anti–Gr-1 Ab administration depletes both monocytic and granulocytic subsets of MDSCs (34). CBA/N mice were sacrificed 24 h postinfection; lungs were first lavaged to collect CFU counts in the BAL, and lungs were homogenized to quantitate the CFU in the lung homogenates. At 24 h, we found that WT-infected CBA/N mice that were given the IgG isotype control had a significant increase in bacterial burden in the lung tissue (*p = 0.0159) and airways (*p = 0.0002) compared with PcpA⁻ mutant–infected mice (Fig. 7). When we used anti–Gr-1 Ab to deplete MDSCs in the lungs of WT-infected mice, we found that anti–Gr-1 administration had no significant effects on the bacterial burden in the lung tissue and BAL (Fig. 7). Interestingly, in the mutant JEN11 strain–infected mice, depletion of MDSCs led to a significant increase in bacterial burden in both the lung tissue and airways (**p = 0.0079, **p = 0.0025) (Fig. 7). These data suggest that the significant increases in MDSCs that are recruited to the lungs in the absence of PcpA are responsible for the increased bacterial clearance that is observed during infection with the mutant strain.

PcpA is a virulence factor in the lung and the blood; however, the direct mechanism of virulence has not yet been identified. For this reason we chose to study the role of PcpA during pneumococcal lung disease in two distinct mouse models.

It was previously shown that the expression of PcpA significantly enhanced adherence to A549 cells in vitro but had no biologically relevant role in initial attachment in a mouse model for pneumococcal pneumonia in vivo (11) (M.M.W.R. Walker, M. Coats, S. Mirza, D. Glover, L. Myers, M. Ochs, and D.E. Briles, manuscript in preparation). The observation of Glover et al. (9) that the pcpA⁻ mutant in the EF3030 background was significantly less recovered compared with the WT strain in the blood 7 d postinfection was consistent with a the late role of PcpA in infection that was observed in the present study. The pathogenesis of pneumococcal pneumonia involves five stages (12–16, 35). The first stage of the immune response during pneumococcal pneumonia is marked by rapid apoptosis of alveolar macrophages and differential proinflammatory cytokine release in lung tissues. The failure to induce apoptosis in macrophages leads to a highly inflammatory environment. Our present studies suggest that the presence of PcpA has negative effects on the first two stages of the innate immune response during pneumococcal pneumonia. Therefore, we chose to study the role of PcpA during pneumococcal infection to tease apart when its expression is critical for virulence in pneumonia. Our current studies have identified a strong association of PcpA expression with increased bacterial burden in the lung tissue, BAL, blood, spleen, and liver by 24 h postinfection in our bacteremic pneumonia model. By 36 h, we observed that mice infected with the pcpA⁻ mutant appear to have a similar bacterial load in each site, leading us to conclude that the increase in mutant pneumococci at that time point is sufficient enough to induce a successful innate immune response. The PcpA-dependent effect on bacterial burden occurring later on in infection seems to suggest that the role of PcpA during pneumococcal pneumonia has little to do with attachment to the lung epithelium and more with modulating the host innate immune response.

We evaluated the role of PcpA in inflammation (where we define it as an influx of neutrophils and lymphocytes) and alveolar congestion after pneumococcal lung infection. In our studies we observed a PcpA-dependent effect on inflammation and alveolar congestion in the lungs during bacteremic pneumonia starting at 12 h postinfection. This association suggests that the absence of PcpA is protective to the lung tissue and promotes tissue repair earlier than in WT-infected mice. We predict that this virulence factor thus promotes lung damage and increased bacterial burden.

We investigated the role of PcpA in recruitment of key players in the innate immune response to bacterial infections. We found that in the presence of pcpA-expressing pneumococci there were significantly fewer professional phagocytes including macrophages and neutrophils aiding in bacterial clearance in the lungs of WT-infected mice. We report in the present study also our novel observation that PcpA suppresses infiltration of MDSCs. We predict that in this environment created by PcpA activity on innate immunity, pneumococci can more rapidly multiply, resulting in collateral lung injury and increased burden. MDSCs are a heterogeneous population of innate cells. These immature myeloid cells suppress host immune responses through various mechanisms, including production of large amounts of reactive oxygen and nitrogen species, immunoregulatory cytokines such as TGF-β and IL-10, and upregulation of immune regulatory enzymatic pathways, including inducible NO synthase and arginase. Although they were discovered first in the microenvironment of tumors, more recently, MDSCs have been shown to play an integral role in the innate immune response during bacterial infection. Those studies showed that lung MDSCs appeared with a delayed kinetics in response to K. pneumoniae infection (26). These cells produced IL-10, which was associated with enhanced efferocytosis of apoptotic PMNs. Our data suggest that expression of PcpA is associated with dysregulation of innate immune mechanisms, including reduced recruitment of MDSCs, PMNs, and macrophages to the lung tissue. Further investigations are necessary to elucidate the direct mechanism of PcpA-dependent suppression of infiltration of MDSCs.

Although PcpA is produced in low amounts by pneumococci, it has been shown to be a virulence factor in the lung and blood. Based on our present results, it seems likely that the mode of action of Ab to PcpA could be inhibition of PcpA’s deleterious effect on innate immunity. PcpA may therefore be a useful candidate for incorporation in a vaccine against invasive pneumococcal infection.

We thank Janice King, Joanetha Y. Hale, and Tong Huan Jin for technical assistance.

The authors have no financial conflicts of interest.