MARCO is required for TLR2- and NOD2-mediated responses to Streptococcus pneumoniae and clearance of pneumococcal colonization in the murine nasopharynx

Abstract

Streptococcus pneumoniae is a common human pathogen that accounts for over a million deaths every year. Colonization of the nasopharynx by S. pneumoniae precedes pulmonary and other invasive diseases, and is therefore a promising target for intervention. Since the receptors Scavenger Receptor A (SRA), Macrophage Receptor with Collagenous Structure (MARCO), and Mannose Receptor (MR) have previously been identified as non-opsonic receptors for S. pneumoniae in the lung, we utilized scavenger receptor knock out mice to study the roles of these receptors in the clearance of S. pneumoniae from the nasopharynx. MARCO^−/−, but not SRA^−/− or MR^−/−, mice had significantly impaired clearance of S. pneumoniae from the nasopharynx. In addition to impairment in bacterial clearance, MARCO^−/− mice had abrogated cytokine production and cellular recruitment to the nasopharynx following colonization. Furthermore, macrophages from MARCO^−/− mice were deficient in cytokine and chemokine production, including type I interferons, in response to S. pneumoniae. MARCO was required for maximal TLR2- and NOD2-dependent NF-κB activation and signaling that ultimately resulted in clearance. Thus, MARCO is an important component of anti-S. pneumoniae responses in the murine nasopharynx during colonization.

Introduction

Streptococcus pneumoniae (the pneumococcus) is one of the most prevalent human pathogens and causes over a million deaths each year, most of which are young children (1). The bulk of these deaths occur due to pneumococcal pneumonia, in which bacteria spread from their preferred niche, the nasopharynx, to the lungs. Pneumococci are armed with a plethora of colonization factors that allow them to establish a ‘carrier state’ in the nasopharynx of approximately 10% of the adult population at any given time, with higher rates in young children under 5 years of age (2). Colonization events are sequential and each lasts days to weeks before clearance. Despite the use of antibiotics and the introduction of polysaccharide-based vaccines, antibiotic resistance and serotype replacement have resulted in continued challenges in managing this pathogen (3–6).

The initial immune response to pneumococcal colonization is characterized by a brisk, yet ineffective, neutrophil response (7). The resolution of the carrier state appears to require the subsequent recruitment of monocytes from the blood into the nasal interstitial spaces, where they differentiate into macrophages, which are then thought to phagocytose and destroy the bacteria. These monocytes are attracted by the chemokine CCL2 which is produced in response to the recognition of bacterial peptidoglycan by Nucleotide-binding Oligomerization Domain-containing (NOD)2 (8). Signaling through NOD2 also drives the expression of type I interferons (IFNs) which contributes to clearance (9). Additionally, the influx of monocytes is affected by TLR2, though the mechanism of this interaction has not been fully elucidated (7). These innate pathways involved in clearance of pneumococcal colonization all require prior of the pneumococcus by resident cells, yet this initial interaction remains elusive.

It has also been shown that the pneumococcus, through the expression of its capsular polysaccharide and other mechanisms, is able to avoid opsonization by antibodies (10, 11) as well as complement components (12–14). While it is not known which macrophage receptors recognize the pneumococcus in the low-opsonic environment of the nasopharynx, previous studies have shown that Mannose Receptor (MR) (15), SIGNR1 (16), and the class A Scavenger Receptors (SRs) known as Scavenger Receptor A (SRA) (17) and Macrophage Receptor with Collagenous structure (MARCO) (18) are capable of binding to S. pneumoniae via non-opsonic mechanisms.

The class A SRs are known for their broad ligand specificity and phagocytic functions (19). They have also been shown to directly and indirectly modulate TLR and NLR signaling (20, 21). MARCO is constitutively expressed only on specific subsets of macrophages (22). However, MARCO expression can also be induced on macrophages at sites of inflammation as a result of bacterial infection, thus reinforcing the notion that it is important in immune defense. In contrast, SRA is constitutively expressed on most, if not all, macrophages as well as other cells such as endothelial cells. MARCO and SRA have been shown to directly bind and phagocytose S. pneumoniae in a murine pneumonia model, however the bacterial ligands were not identified (23, 24). The role of scavenger receptors in colonization is not known and is the focus of this study.

In the current study, we show that MARCO is a key component in the macrophage response to S. pneumoniae during colonization. We demonstrate that MARCO is involved in the timely clearance of pneumococcal colonization from the nasopharynx of mice by augmenting the production of pro-inflammatory cytokines and chemokines. In vitro analyses confirm MARCO’s role in TLR2 and NOD2 signaling pathways that lead to the production of these soluble mediators. We also show, for the first time, a connection between SR activity and the production of type I IFNs in response to extracellular bacteria.

Materials & Methods

Mice

C57Bl/6 mice were obtained from Charles River Laboratories (Wilmington, MA) and Jackson Laboratories (Bar Harbor, ME) at 6–8 weeks of age. Original MARCO^−/− and SRA^−/− mice were generously provided by the laboratory of Professor Siamon Gordon (University of Oxford, UK) and bred at the McMaster Central Animal Facility. MR^−/− mice were from University of Oxford, UK, C3^−/− mice by John Lambris (U. Pennsylvania), and MAC1^−/− mice were purchased from Jackson Laboratories. All animals were used at either 9–11 weeks of age or when they reached 20g in weight, whichever occurred first. WT mice were sexed to match the knock out groups. All procedures performed were done so in accordance with the McMaster Animal Research Ethics Board guidelines and Institutional Animal Care and Use Committee protocols at the University of Pennsylvania.

Bacterial Strains and Culture Conditions

S. pneumoniae strain P1121 (a clinical isolate, serotype 23F) (25) was used for all assays and inoculations. TIGR4 (a clinical isolate, serotype 4) was also used for colonization studies of SR knockout mice. Bacteria were propagated in tryptic soy broth (Gibco) at 37°C and 5% CO₂ until cultures reached log phase, OD₆₀₀ between 0.45 and 0.50. In vitro and ex vivo experiments were performed with lysozyme-digested bacteria as this enzyme is abundant on the mucosa and during macrophage processing. Lysozyme treatment does not lyse the pneumococcus but releases peptidoglycan fragments that promote innate immune signaling (8). Lysozyme-digested pneumococcus was prepared as follows; bacteria were heat killed by incubation at 65°C for 10 min, followed by incubation for 18 h in the presence of 0.5 µg/mL recombinant human lysozyme (Cedarlane, Burlington, ON, Canada) at room temperature, vortexing every few hours.

Murine Model of Pneumococcal Colonization

Pneumococcus was grown to log phase, concentrated 10 fold in PBS and stored on ice. Unanaesthetized mice were inoculated intranasally with 10 µL of bacterial suspension containing ~1×10⁷ CFUs. At the time indicated, mice were sacrificed, the trachea was cannulated and 200 µL PBS was instilled. Lavage fluid was collected from the nares and serially diluted in PBS and plated on tryptic soy plates containing 5% sheep’s blood and neomycin (5 µg/mL for TIGR4 and 10 µg/mL for P1121). Colonies were counted after overnight incubation at 37°C and 5% CO₂.

Macrophage Culture and Stimulation

All macrophages were cultured in a humidified environment at 37°C with 5% CO₂ in RPMI 1640 supplemented with 10% FCS, penicillin (100U/mL), streptomycin (100 µg/mL), and 10 mM L-glutamine. For assays involving stimulation of cultured cells, lysozyme-digested P1121 was added in RPMI 1640 containing 1% FCS and L-glutamine. In order to study recruited, or “elicited” macrophages, BioGel elicitation was performed. Mice were injected intraperitoneally with 1 mL of 2% (w/v) BioGel P100 45–90 µm diameter microbeads (Bio-Rad). Peritoneal lavages were performed 5 days later with 10 mL of ice-cold PBS. Cells were then washed once with RPMI 1640 and resuspended in RPMI 1640 supplemented with 10% FCS, penicillin (100 U/mL), streptomycin (100 µg/mL), and 10 mM L-glutamine and incubated in 24-well tissue culture plates at a concentration of 5×10⁵ cells per well in 1 mL of media. Cells were allowed to adhere for 2 hours then non-adherent cells were removed by washing with warm media and adherent cells were incubated at 37°C with 5% CO₂ overnight. The following day cells were stimulated with 20 µg/mL of recombinant murine interferon gamma (rIFNγ) (Peprotech) for 24 h to upregulate NOD2 expression. Elicited macrophages were then stimulated with lysozyme-digested S. pneumoniae at an MOI of 25, 10 ng/mL LPS (Sigma) together with 5 µg/mL MDP (Sigma), or media alone for 16h. Supernatants were collected and either used immediately for cytokine ELISA or stored at −80°C for later use.

RNA Extraction & RT-PCR

Total RNA was extracted from nasal lavages collected with RLT Lysis Buffer (Qiagen) containing 0.7% 2-mercaptoethanol. An RNA extraction Micro kit (Ambion) was used following the manufacturer’s directions. Complementary DNA was synthesized using MMLV reverse transcriptase (Invitrogen) following manufacturer’s directions. Quantitative real time PCR was performed on cDNA with SYBR green (Promega) following manufacturer’s directions and results were compared to a GAPDH control gene. The primers used for each gene can be found in Table I.

Table I.

RT-PCR Primer Sequences

Gene	Sense Primer	Anti-Sense Primer
marco	GGCACCAAGGGAGACAAA	TCCCTTCATGCCCATGTC
cd45	CAGAGCATTCCACGGGTATT	GGACCCTGCATCTCCATTTA
ccl2	GTCTGTGCTGACCCCAAGAAG	TGGTTCCGATCCAGGTTTTTA
il6	ATACCACTTCACAAGTCGGAGGC	CTCCAGAAGACCAGAGGAAATTTTC
tnfa	CAAAGGGAGAGTGGTCAGGT	ATTGCACCTCAGGGAAGAGT
il1b	GCCTCGTGCTGTCGGACCCATA	GATCCACACTCTCCAGCTGCAGG
ifnb	GCACTGGGTGGAATGAGACT	AGTGGAGAGCAGTTGAGGACA
gapdh	TGTGTCCGTCGTGGATCTGA	CCTGCTTCACCACCTTCTTGA

Immunofluorescent Staining

Nasal lavage fluid was applied to Colorfrost Plus Microscope Glass Slides using the Shandon Cytospin 3 cytocentrifuge (Thermo Scientific) at 230g for 10 minutes. The cytospin preparations were air dried briefly before fixing in acetone. Samples were blocked with 10% donkey serum before addition of primary antibody. Signal was detected with Cy2- or Cy3- conjugated species specific secondary antibody (Jackson ImmunoResearch, West Grove, PA) incubated at 1:500 dilution in block for 2hr at room temperature. After washing with PBS followed by dH2O, sections were counterstained with DAPI (4′,6-diamidino-2-phenylindole) (Molecular Probes, Invitrogen, Carlsbad, CA) diluted 1:10,000 in dH2O. All image analysis was carried out using iVision-Mac (BioVision Technologies, Exton, PA).

For positive control of scavenger receptor staining spleens were dissected from naïve mice and fresh frozen in Tissue-Tek O.C.T. embedding medium (Miles) in a Tissue-Tek Cryomold. Five-micrometer-thick sections were cut and stored at −80°C. Tissue sections were stained as above.

Flow Cytometry

Nasal lavage samples were stained with the following fluorescent antibodies: Ly6C:FITC, Ly6G:PE, F4/80:APC, and CD45:PacificBlue (eBioscience). Staining was completed by incubating cells with the fluorophore-conjugated antibodies after blocking with the 2.4G2 antibody (eBioscience) at 4°C. Cells were then fixed with 1% paraformaldehyde (PFA) and assayed with a BD LSRII flow cytometer the following day. Data was gathered using FACSDiva software (BD) and analyzed using FlowJo (Tree Star Inc.)

Cytokine ELISA

ELISAs for TNFα, IL-1β, CCL2, and IL-6 were performed as per manufacturer’s directions (eBioscience). Plates were read on a Safire plate reader within 20 minutes of the addition of H₂SO₄.

Interferon Bioassay

Elicited macrophages were prepared as described above and were stimulated with lysozyme-digested P1121 at an MOI of 25 for 24h. Supernatants were removed from the macrophages, serially diluted, and added to a confluent monolayer of L929 cells overnight at 37°C and 5% CO₂. IFNα was used as a concentration standard. The next day the media was removed and replaced with 30 µL of 1×10⁵ pfu/mL of a GFP-expressing vesicular stomatitis virus in serum-free media for 24h as in (26). If the supernatants contained any type I IFN (ie. if the macrophages produced IFN in response to the bacterial stimulation) then the virus would not be able to replicate and no fluorescence would be seen. The fluorescence signal given off by GFP was measured the next day using a Typhoon Trio variable mode imager and quantified using ImageQuant software (ImageMaster).

NF-κB Luciferase Assay

Human TLR2, CD14, TLR9, and NF- κB luciferase (NF-κB-luc) plasmids were provided by Dr. Cynthia Leifer (Cornell University), human NOD2 plasmid was provided by Dr. Dana Philpott (University of Toronto), and β-galactosidase (β-gal) plasmid was provided by Dr. Brian Lichty (McMaster University). All plasmids were amplified by Escherichia coli DH5-α cells and purified using HiPure Plasmid Filter Midiprep Kit (Invitrogen).

Low passage (P ≤ 4) HEK293T cells were seeded at 5×10⁵ cells per well in 3 ml of DMEM per well in a 6 well plate overnight. HEK293T cells were transfected with NF-κB-luc (100 ng), β-gal (100ng), and optimal combinations of hMARCOI (300 ng), TLR2 (30 ng), CD14 (30 ng), and NOD2 (50 ng). The total amount of DNA was brought to 1 µg by transfecting empty pcDNA3.1 vector. Transfections were performed using GeneJuice transfection reagent as per the manufacturer’s instructions (Novagen). At five hours post transfection serum-free DMEM media was replaced with complete DMEM media. 24 hours later transfected cells were stimulated with an MOI of 25 lysozyme-digested P1121 in 3 ml of serum-free DMEM media. After 48h, the lysates were collected using Reporter Lysis Buffer (Agilent) and were analyzed for luciferase (Agilent) and β-galactosidase (Clontech) activity using a luminometer (Turner Biosystems).

Cell Association/Internalization Assay

Møs or HEK293T cells were suspended in 1mL of Hank’s Balanced Salt Solution (HBSS) at a concentration of 1×10⁶ cells/mL. Live P1121 was added at an MOI of 10 and the solution was mixed on a nutating mixer at 37°C for 1h. The cells were then separated from unbound bacteria by centrifuging at 1500rpm for 5min. To measure cell association, cells were washed once in HBSS and then lysed in H₂O. Serial dilutions were performed in H₂O and plated on sheep’s blood agar supplemented with 10µg/mL neomycin. Colonies were counted the next day. In order to measure uptake directly, extracellular bacteria were removed by adding 25µg/mL gentamycin for 10min at 37°C. Cells were then washed with HBSS and lysed in H₂O.

Statistics

Statistical analyses were carried out using the unpaired Student’s t-test (GraphPad) except where indicated. Results were considered statistically significant if p≤0.05, as indicated in the figure legends.

Results

Clearance of S. pneumoniae colonization does not require opsonins

Although antibodies have been shown to be dispensable for pneumococcal clearance in the nasopharynx (7, 10), it is not clear whether other opsonins, such as complement, could play a role in the clearance of pneumococcal colonization. To address this, mice lacking Complement Receptor 3 (CR3, also known as MAC-1) or Complement component 3 (C3) were intranasally colonized with S. pneumoniae isolate P1121. The bacterial burden in the nasopharynx was measured in WT, MAC-1^−/−, and C3^−/− mice at 21 days post-inoculation (Fig. 1A). The clearance of S. pneumoniae was not significantly different among the three different genotypes, demonstrating that complement opsonization is not necessary for the clearance of pneumococcal colonization.

Figure 1. Clearance of pneumococcal colonization is complement-independent and MARCO-dependent.

Mice were intranasally inoculated with S. pneumoniae and colonization assessed at indicated time points. (A) WT, C3^−/−, and MAC-1^−/− mice were colonized with P1121 for 21 days (n≥10, per group) and bacterial burden in nasal lavages was determined. (B) Bacterial burden in nasal lavages was determined for WT, MARCO^−/−, and SRA^−/− mice colonized with P1121 for 1, 3, 7, 14, and 21 days post-inoculation (p.i) (n≥6, per group). Data are plotted as mean ± SEM. (C) WT and MR^−/− mice were colonized with P1121 for 21 days (n≥7, per group) and bacterial burden in nasal lavages was determined. (D) WT, MARCO^−/−, SRA^−/− and MR^−/− mice were colonized with TIGR4 for 21 days (n≥7, per group) and bacterial burden in nasal lavages was determined. Box-and-whisker plots indicate high and low values, median and interquartile ranges. ** = p ≤ 0.005, *** = p ≤ 0.001 when comparing WT and MARCO^−/− mice, by Mann-Whitney U test.

MARCO and SRA have previously been shown to be opsonin-independent phagocytic receptors important in innate immune pneumococcal surveillance (17, 18). MR was also previously shown to bind to pneumococcal polysaccharides (27). In order to determine the importance of these receptors in the clearance of P1121 from the nasopharyngeal passage, we inoculated mice with 10⁷ CFU intranasally and assessed bacterial numbers at various timepoints post-inoculation. We compared pneumococcal CFUs in the nasal lavages of WT, MARCO^−/−, SRA^−/−, and MR^−/− mice. MARCO^−/− mice, but not SRA^−/− or MR^−/− mice, were significantly impaired in clearing the bacteria beginning at day 14, with an even greater deficit seen at day 21 post-inoculation, at which point clearance was completed in WT mice (Fig. 1B and C). The role of MARCO in clearance of pneumococcal colonization was confirmed using TIGR4, an isolate which expresses a different capsular polysaccharide (Fig. 1D), thus demonstrating the effect of MARCO is likely conserved across pneumococcal serotypes.

RNA transcripts of SIGNR1, an additional non-opsonic receptor that has been implicated in anti-pneumococcal immune responses (16), were undetectable in colonized nasal lavages (data not shown), effectively ruling it out as an important receptor in our model.

Expression of class A SRs in the nasopharynx

To determine the cell population expressing MARCO in the URT, nasal lavages were analyzed for MARCO transcripts at 30 min, as well as days 3 and 7, when the effector monocytes/macrophages reach a maximum level in the URT. CD45 transcript increased at day 3 and day 7, correlating to the influx of effector cells (Fig. 2A). However, although MARCO transcript was detected in nasal lavages at all time points, the amount of transcript did not increase (Fig, 2A). This suggests that MARCO is present on resident cells in the nasopharynx, but not the recruited effector cells. To further confirm this we stained cytospin preparations of nasal lavages from colonized mice at the same time points. While scavenger receptors CD68 and MR were both detected on recruited lumenal cells, MARCO was not detected on this population (Fig. 2B).

Figure 2. MARCO is not expressed on the effector cells recruited to the nasopharynx.

(A) QRT-PCR on RNA lysis buffer nasal lavages at 30 mins (black bars), 3 days (grey bars) and 7 days (white bars) post-inoculation with P1121. Bars represent relative mRNA levels (GAPDH/Target) n ≥ 3. (B) Upper panels: Immunofluorescent staining of cytospin preparations from nasal lavages. Representative images from day 3. Images at 400× magnification. Inset in MARCO panel demonstrates positive staining in alveolar macrophages. Lower panels: Fresh frozen spleen sections stained as positive controls for scavenger receptor antibodies. Images at 100× magnification.

Cellular recruitment to the URT is hindered in MARCO^−/− mice

It has been previously shown that effector cells are recruited to the nasopharynx after pneumococcal colonization and are required for clearance (7). Thus, we examined whether MARCO^−/− mice were deficient in neutrophil, macrophage, and monocyte recruitment to the nasopharynx during colonization. We utilized a highly sensitive flow cytometry assay to quantitate the low numbers of cells recruited to the murine nasopharynx during pneumococcal colonization. Neutrophil (Ly6G⁺Ly6C⁻) counts remain high in the WT lavages until day 21, when the majority of bacteria have been cleared, whereas neutrophil levels remain low in MARCO^−/− mice. Monocyte (Ly6C⁺Ly6G⁻) recruitment also remains at basal levels throughout colonization in the MARCO^−/− mice, whereas WT mice show robust levels of monocytes at day 14. Interestingly, there is an influx of recruited macrophages (F4/80⁺) by day 14 post-inoculation in the WT mice, whereas this recruitment is delayed in the MARCO^−/− mice by day 14, possibly accounting for the lack of bacterial clearance in these mice at this timepoint (Fig. 3C). In summary, MARCO^−/− mice have delayed recruitment of leukocytes in response to pneumococcal colonization.

Figure 3. Recruitment of leukocytes to the nasopharynx is impaired in MARCO^−/− mice.

WT and MARCO^−/− mice were colonized with S. pneumoniae for 3, 14, and 21 days p.i. Nasal lavages were stained and analyzed for cell surface markers by flow cytometry. These markers were used to determine the number of (A) neutrophils (Ly6G+Ly6C−), (B) monocytes (Ly6C+Ly6G−), and (C) macrophages (F4/80+) in each population (n≥5 for each group). Data are plotted as mean percentage of total cells ± SEM. * = p ≤ 0.05, *** = p ≤ 0.001.

MARCO enhances S. pneumoniae-induced chemokine and cytokine responses

During the course of pneumococcal colonization, macrophage-mediated detection of bacteria or bacterial components drives an inflammatory response that is required for clearance. It has previously been shown that class A SRs are able to modulate cytokine production (20). In order to determine whether MARCO contributes to the production of inflammatory cytokines, we examined expression of TNFα, IL-1β, and IL-6 mRNA in the nasopharynx of WT and MARCO^−/− mice during the course of colonization (Fig. 4A) as well as cytokine production resulting from stimulation of Møs with S. pneumoniae ex vivo (Fig. 4B). MARCO^−/− mice colonized with P1121 exhibited significantly delayed TNFα, IL-6 and IL1-β mRNA transcription when compared to WT mice throughout colonization, with the largest differences seen at day 7. Correspondingly, Møs from MARCO^−/− mice produced significantly less of all three cytokines, compared to WT Møs, when stimulated with bacteria ex vivo. In fact, in most cases MARCO-deficient macrophages did not produce any cytokines when stimulated with S. pneumoniae compared to stimulation with media alone.

Figure 4. MARCO enhances the production of cytokines and chemokines.

(A) WT and MARCO^−/− mice were intranasally inoculated with S. pneumoniae, sacrificed at days 1, 3, 7, 14 and 21 p.i. and RNA was isolated by intratracheal nasal lavage. cDNA was analyzed by semi-quantitative RT-PCR (n≥3, per time point). (B) Møs were isolated from WT and MARCO^−/− mice and stimulated with lysozyme-digested S. pneumoniae preparations and controls ex vivo. Cytokine production was measured by ELISA (n≥6 for each group). Data are presented as mean ± SEM. * = p ≤ 0.05, *** = p ≤ 0.001.

The macrophage-chemotactic protein CCL2 (also known as MCP-1) has been shown to be of primary importance for the recruitment of monocytes/macrophages to the nasopharynx during S. pneumoniae colonization (8). We investigated CCL2 expression at the RNA level in nasal lavages during colonization of WT and MARCO^−/− mice (Fig. 4A). MARCO^−/− mice colonized with P1121 had significantly delayed CCL2 transcription. At days 1 and 3, there were no significant differences in CCL2 transcription. However, by day 7 there was a significant reduction in CCL2 transcription in the MARCO^−/− mice compared to WT mice. When elicited macrophages (Møs) from MARCO^−/− mice were stimulated ex vivo with S. pneumoniae, the level of CCL2 production did not exceed background levels. However, their WT counterparts produced a robust CCL2 response (Fig. 4B). This reduction in CCL2 production by macrophages was consistent with the diminished macrophage influx seen in MARCO^−/− mice.

Together these results demonstrate that MARCO-mediated recognition drives the inflammatory response during pneumococcal colonization.

MARCO modulates type I IFN production in macrophages

We have recently shown that pneumococcal colonization of the nasopharynx leads to an increase in type I IFN production (9). In order to test whether MARCO was involved in type I interferon production, we first analyzed nasal lavages from WT and MARCO^−/− mice for IFN-β mRNA content. MARCO^−/− mice had significantly less IFN-β mRNA than their WT counterparts at days 1, 3, and 7 post-inoculation (Fig. 5A). We then stimulated Møs with S. pneumoniae for 24 hours, after which the cell supernatants were used in a standard bioassay to measure type I IFN production. Supernatants from WT and MARCO^−/− Møs stimulated with P1121 were transferred to L929 cells which were subsequently infected with a GFP-expressing vesicular stomatitis virus (GFP-VSV). Cell supernatants from WT Møs treated with S. pneumoniae protected L929 cells from GFP-VSV infection, indicating pneumococcal IFN stimulation, as measured by low levels of viral GFP fluorescence. Conversely, supernatants from MARCO^−/− Møs yielded less protection from viral infection, indicating that they produced significantly less type I IFN (Fig. 5B). Thus, MARCO has a role in the production of type I IFNs in response to pneumococcal stimulation.

Figure 5. MARCO modulates the production of type I IFNs.

(A) IFN-β mRNA from nasal lavages of colonized mice. RNA from WT and MARCO^−/− mice was examined by RT-PCR. (B) Møs from WT and MARCO^−/−,mice were isolated and stimulated with lysozyme-digested S. pneumoniae. Resulting supernatants were then added to L929 cells in culture. VSV-GFP was used to infect L929 cells and cell infection was measured by GFP fluorescence. (C) Bacterial association with Møs from WT and MARCO^−/− mice was measured by cellular lysis after 1h of stimulation. GFP fluorescence data, indicating viral infection, are shown as mean fluorescence units across replicates (n = 3) ± SEM. ** = p ≤ 0.005, *** = p ≤ 0.001.

Type I IFN signaling has been shown to require uptake of the bacteria and subsequent intracellular signaling (28). In order to determine whether MARCO’s role in type I IFN production was related to its capacity to bind and internalize the bacteria, we performed a bacterial cell association assay with WT and MARCO^−/− Møs (Fig. 5C). Upon stimulation with live P1121 for 1 hour, bacterial association was reduced by approximately 50% in MARCO-deficient Møs when compared to WT Møs, providing us with a possible mechanism for MARCO’s role in type I IFN production.

MARCO contributes to TLR and NLR signaling pathways

Although MARCO itself has no known signaling capacity, it has been shown to enhance TLR signaling in response to certain bacteria or their components (29). Therefore, we hypothesized that MARCO may be required for bacterial recognition by pattern recognition receptors, such as TLR2 and NOD2, in order to enhance S. pneumoniae-specific responses. In order to discern MARCO’s role in the signaling capacities of these receptors, we used an NF-κB luciferase assay. Cells transfected with MARCO and NOD2 demonstrated significantly more NF-κB activation upon stimulation with P1121 than cells transfected with NOD2 alone (Fig. 6A). The same was true for cells transfected with MARCO, TLR2, and CD14 versus cells transfected with just TLR2 and CD14 (Fig. 6B). Interestingly, cells transfected with TLR2, CD14, and SRA showed a decrease in NF-κB activation when compared to cells transfected with TLR2 and CD14 alone. Cells transfected with MARCO or CD14 alone did not show any NF-κB activation, nor did MARCO enhance NF-κB activation upon stimulation with the TLR2 ligand Pam3Csk4 or the NOD2 ligand MDP (data not shown). In addition, when this assay was performed utilizing plasmids expressing TLR4 or TLR9, MARCO had no effect on NF-κB activation (data not shown).

Figure 6. MARCO affects NOD2 and TLR2 responses to the pneumococcus.

HEK293T cells were co-transfected with various combinations of plasmids expressing (A) NOD2, SRA and MARCO or (B) TLR2, CD14, SRA and MARCO, or empty vector. 24 hours post-transfection, cells were infected with lysozyme-digested S. pneumoniae at an MOI of 25. 24 hours post-infection, luciferase activity was measured (n≥6 for each group). This assay was normalized for transfection efficiency by dividing the luciferase activity by the β-galactosidase activity. Average of 3 independent experiments ± SEM. ** = p ≤ 0.01, *** = p ≤ 0.001, by 1-way ANOVA with Bonferroni’s post-test. (C, D) Bacterial association with HEK293T cells transfected with MARCO plasmid or empty vector was measured by cell lysis (C) before or (D) after addition of gentamycin to kill extracellular bacteria. *** = p ≤ 0.001.

In order to determine whether the forced expression of MARCO by the HEK293T cells led to an increased ability of the cells to bind and internalize the bacteria we performed a bacterial cell association assay with HEK293T cells transfected with a MARCO-expressing plasmid or an empty vector. When these cells were incubated with live P1121 for 1 hour there was no significant difference in total cell association (ie. binding and uptake) of the bacteria (Fig. 6C). However, when all extracellular bacteria was eliminated by the addition of gentamycin it was shown that the MARCO-expressing cells were better able to internalize the bacteria (Fig 6D), although the total bacterial numbers were quite low for both sets of cells. Together these results demonstrate that MARCO enhances some, but not all, TLR and NLR responses to pneumococci and that this may be due to its phagocytic capabilities.

Discussion

Since nasopharyngeal colonization precedes pneumococcal disease, it is an attractive therapeutic target and thus it is important to understand host defense at this site, which requires interactions between the effector cells and the bacteria. It has been previously reported that macrophages recruited to the nasal mucosa are important effector cells in the clearance of colonization and that they are able to recognize the bacteria without the aid of opsonizing antibodies (10). Here we have shown that clearance of pneumococcal colonization does not require complement component 3 or its cognate receptor, supporting the idea that macrophages are able to act without the aid of opsonins (Fig. 1A). While this had previously been shown in the lung, we are now able to extend the role of non-opsonic receptors in pneumococcal clearance from the nasopharynx.

MARCO, MR and SRA have previously been shown to be important for clearance of pneumococci in the lungs and the central nervous system (15, 17, 18, 27). This led us to investigate their role in clearance from the nasopharynx. Interestingly, the scavenger receptors appear to play distinct roles in upper versus lower respiratory tract clearance. In the lungs, MARCO and SRA play redundant roles in the recognition, uptake, and subsequent clearance of pneumococci. We can only hypothesize on the importance of this redundancy, though it is likely rooted in the need to overcome the abundance of virulence factors expressed by the pneumococcus in the environment of the lung. Both MARCO and SRA, which are constitutively expressed on alveolar macrophages, have been shown to directly recognize the bacteria and trigger their engulfment by these cells (17, 23). Mannose receptor has also been shown to directly recognize pneumococci (27). Conversely, despite MARCO, MR and SRA (data not shown) being expressed in the nasopharynx, only MARCO enhances clearance of pneumococcal colonization from the nasopharynx (Fig. 1). Also, while MARCO is required for efficient clearance of nasal colonization, we did not find MARCO present on the recruited effector cells in WT mice. This leads us to believe that MARCO’s role may be linked to resident cells in the nasopharyngeal mucosa, which function in immune surveillance (Fig. 2). These cells likely act as indirect mediators of the immune response to S. pneumoniae by contributing to the complex cytokine and chemokine milieu at the site of colonization. The receptor(s) on the effector cells responsible for recognizing the pneumococcus remains unknown. These differences in the role of SRs could explain why pneumococci are able to asymptomatically colonize the nasopharynx, but induce a violent inflammatory state once they gain access to the lungs.

The chemokine CCL2 has been shown to be vital to the recruitment of monocytes/macrophages to the nasopharynx during pneumococcal colonization, in a TLR2 and NOD2 dependent manner (8). We have shown a significant deficiency in the transcription of CCL2 mRNA in the nasopharynx of MARCO^−/− mice at early time points in colonization when compared to WT mice (Fig. 4A). We have also shown that the CCL2 production by macrophages from MARCO^−/− mice in response to pneumococcal stimulation is severely impaired (Fig. 4B). This defect in CCL2 production is likely responsible for the impaired recruitment of myeloid cells, and especially macrophages, to the nasopharynx throughout colonization (Fig. 3). The increased recruitment of macrophages to the nasopharynx in WT mice correlated with decreased bacterial load, which began between days 7 and 14 post-inoculation (Fig. 1B). Neutrophil recruitment was also impaired in MARCO^−/− mice, however, prior studies have shown that while these cells are robustly recruited early during colonization they are not sufficient to clear the pneumococcus (7). It is important to note that the macrophages recruited to the nasopharynx were CD11c⁻MHCII⁻ and CD11b^hi (data not shown) and therefore were more similar to recently recruited monocytes in the process of differentiation to macrophages than they were to alveolar macrophages.

A consequence of pneumococcal activation of TLR2 and NOD2 is a robust pro-inflammatory response. In accordance with this, we have shown that MARCO is important in the pathway leading to NOD2 and TLR2 dependent NF-κB activation, which leads to the production of pro-inflammatory cytokines and chemokines. Nasal lavages from MARCO^−/− mice had significantly lower levels of TNFα, IL-1β, and IL-6 mRNA at day 7 post-inoculation when compared to WT mice (Fig. 4A). This time point coincides with the beginning of bacterial clearance in WT mice (Fig. 1B). Also, Møs from MARCO^−/− mice were unable to produce these proteins following stimulation with pneumococci (Fig. 4B). This is consistent with other studies demonstrating MARCO’s contribution to the pro-inflammatory response by macrophages when stimulated with other pathogenic organisms (18, 23, 29, 30). While it is unclear whether this MARCO-dependent production of pro-inflammatory cytokines is a TLR2-, or NOD2-specific phenomenon, we have provided evidence that MARCO enhances both TLR2-, and NOD2-mediated NF-κB activation. Both of these receptors have been shown to be important in pneumococcal clearance in previous studies (8, 31). We hypothesize that either TLR2 or NOD2, or both, is the primary signaling receptor involved in the production of TNFα, IL-1β, and IL-6 and that MARCO is responsible for enhancing this response during colonization. Interestingly, we have shown that MARCO is also involved in the production of type I IFNs by Møs. Two studies have shown a protective effect for IFNβ in pneumococcal disease (32, 33), while another has shown the opposite in the setting of concurrent influenza A infection (9). Thus, the production of type I IFNs during pneumococcal colonization has profound effects on the disease outcome and is linked to MARCO-mediated signaling.

The mechanism by which MARCO enhances TLR2 and NOD2 signaling has yet to be elucidated. MARCO appears to be essential for, and possibly upstream of, NOD2 and TLR2 signaling in response to the pneumococcus but not to monomeric ligands such as MDP and Pam₃Csk₄ (data not shown). Likewise, many studies have shown that NOD2 and some TLR2 signaling pathways require internalization of ligands (34, 35). Our data showing an increased capacity of MARCO-expressing cells to internalize the bacteria (Fig. 5C, 6D) provide us with evidence that MARCO can increase signaling by intracellular receptors, such as NOD2, by facilitating the transition of the bacteria into the cell. This is especially evident in the NF-κB luciferase assays wherein HEK293T cells not expressing MARCO were nearly completely unable to internalize live pneumococci. However, because overall binding of the bacteria to HEK293T cells is not significantly affected by the expression of MARCO, we do not believe that MARCO’s role in TLR2 signaling is confined to its binding and uptake capacities. Also consistent with the hypothesis that MARCO lies upstream of TLR2 and NOD2, the degree to which the bacteria were able to persist in MARCO^−/− mice was greater than shown in previous studies on NOD2^−/− (8) and TLR2^−/− (10)mice. Previous studies, as well as our work, have shown an inhibitory effect of SRA on TLR2 signaling (Fig. 6) (18, 23, 29, 30). A recent report presented a mechanism for TLR4 inhibition by SRA, which involves direct contact between SRA and the signaling machinery of TLR4 (21). It is possible that a similar mechanism occurs with MARCO, where it can directly bind the TLR2 signaling machinery but in this case to enhance signaling.

The production of type I IFNs in response to extracellular bacteria is also reliant on the ability of the cell to internalize bacterial ligands via endocytosis (28). Therefore, it is probable that MARCO’s role in type I IFN production is linked to its ability to internalize the bacteria, as shown in Figure 5C. This does not preclude roles for other molecules formerly established in other laboratories. For example, it is possible that the uptake of live bacteria into a cell could lead to pneumolysin-dependent rupture of endosomes, leading to bacterial ligands reaching the cytoplasm to be sensed by cytosolic or transmembrane receptors (as proposed in (32, 36)).

In summary, we have shown that MARCO is vital to the clearance of S. pneumoniae colonization from the murine nasopharynx. This is due to its role in non-opsonic recognition of the bacteria, which leads to increased NOD2- and TLR2-dependent chemokine and cytokine production, and ultimately the recruitment of effector monocytes/macrophages. This is, to the best of our knowledge, the first demonstration of MARCO-mediated collaboration with pattern recognition receptor signaling contributing to the clearance of the pneumococcus. Our hope is that targeting antigens to the MARCO-mediated response will provide us with novel, serotype-independent vaccination strategies against the pneumococcus.

Abbreviations

SR

scavenger receptor

MARCO

Macrophage Receptor with Collagenous Structure

SRA

scavenger receptor class A

MR

mannose receptor

SIGNR1

Specific Intracellular adhesion molecule-3 Grabbing Nonintegrin homolog-Related 1

NOD2

Nucleotide-binding Oligomerization Domain-containing receptor 2

MDP

muramyl dipeptide Pam₃Csk₄

Mø

BioGel-elicited macrophage