NO Is a Macrophage Autonomous Modifier of the Cytokine Response to Streptococcal Single-Stranded RNA

Streptococci are constituents of the normal mucosal flora. However, once they become invasive, they are a significant threat to life. Resident tissue macrophages are the outposts of innate immunity. Their primary role is that of sentinel cells. They orchestrate the inflammatory response to single streptococci that have broken through the mucosal surface and reach the subepithelial space. Accordingly, a balanced macrophage response is pivotal for host–microbe coexistence.

Streptococci are typical extracellular bacteria. Yet, the predominantly extracellular lifestyle does not mean that these bacteria are recognized at the cytoplasmic membrane. Clearly, free streptococcal lipoproteins activate the cell surface receptors TLR2/6 and CD14 (1, 2). In contrast, streptococcal organisms as particulate matter do not essentially engage any of the single surface TLRs known to instruct the inflammatory cytokine response to other bacteria (1, 3). Moreover, streptococcal ssRNA, and not the “visible” bacterial surface molecules, such as capsular polysaccharides, lipoteichoic acid, or lipoproteins, is the dominant cytokine-inducing molecule in streptococcal organisms (4).

The observation that uptake and lysosomal processing are prerequisites for the transcriptional activation of inflammatory cytokines (5, 6) suggests that the modification of ligand–receptor interactions by the microenvironment of the maturing phagosome is important for inflammatory gene activation. The sampling process of bacterial particles is not a static event: it is a dynamic process of reciprocal modifications on both sides of the host–microbe interface.

An initiating event of bacteria–macrophage interaction in macrophages is the closure of the phagocytic cup. Then follows the formation of specific endosomal pattern recognition receptors and cytoplasmic multiprotein complexes (7). At the same time, host and bacterial molecules are modified, in particular through the three principal activities comprising pH shift and activation of lysosomal proteases and the oxidative forces, reactive oxygen intermediates and reactive nitrogen intermediates. These activities are closely interdependent. For example, reactive oxygen intermediates indirectly liberate lysosomal proteases from the endosomal proteoglycan matrix (8) and thereby propagate their activity. These processes are likely to transfer specific functional flavors to receptors expressed on the endosomal membranes and ligands being accessible for these receptors. It has been reported that inducible NO synthase positively regulates the formation of inflammatory cytokines (9). However, the molecular link between lysosomotropic modification of the host–bacteria interface and regulation of inflammatory cytokine genes is still not completely understood.

In this study, we report that the type I macrophage response to Gram-positive bacteria is dependent on phagolysosomal processing of bacterial RNA by a mechanism involving the MyD88-dependent formation of NO and the subsequent acidification of the phagolysosome. This is accompanied by processing of streptococcal RNA in the endosomal compartment, which finally results in the transcriptional activation of inflammatory cytokine genes.

Mice lacking MyD88 and those lacking TLRs were previously described (10–15). Mice with an UNC-93B-H412R (3D) mutation were kindly provided by Bruce Beutler (La Jolla, CA). Mice lacking inductible nitric oxide synthase (iNOS), gp91^Phox, and C57BL/6J mice were purchased from The Jackson Laboratory. The mice used in the study were back-crossed onto C57BL/6J mice according to Refs. 10–15.

The Group B streptococci (GBS) type III strain COH1, originally isolated from a newborn infant with sepsis, was cultured to exponential growth phase in chemically defined medium and heat fixed, as previously described (6). Macrophage differentiation and maintenance of immortalized cell line were done according to Ref. 16. Before infection, nonadherent cells were removed by washing with PBS, and the medium was replaced by medium supplemented with 10% FBS, with or without GBS. Samples were analyzed at different time points postinfection (up to 48 h).

For live bacterial infection, GBS were grown in chemically defined medium up to exponential growth phase. After extensive washing with PBS, bacteria were added to monolayers at the indicated multiplicity of infection (MOI) (bacteria/cell). Monolayers were washed extensively after 30 min at 37°C and incubated for various time intervals in presence of antibiotics (100 μg/ml gentamicin, 5 U/ml penicillin, and 5 μg/ml streptomycin). Bacterial count was determined by colony counting on blood agar according to standard procedures. Cell viability was determined by trypan blue exclusion and was not affected by any of the pretreatments used in some of the experiments.

Macrophages were grown overnight on glass coverslips. Cells were stimulated and fixed with 3% paraformaldehyde at room temperature for 15 min. Hoechst (Molecular Probes) staining of nuclei was carried out by adding a 1:10,000 dilution in PBS. Staining of bacteria was carried out by using wheat germ agglutinin conjugated to tetramethylrhodamine isothiocyanate (WGA-TRITC) (Molecular Probes) or wheat germ agglutinin conjugated Alexa 488 (WGA-488) (Molecular Probes). Phalloidin-TRITC (Molecular Probes) was used to stain actin cytoskeleton. Staining of nitrated proteins was carried out by using anti-3NTyr Ab (Enzo). Confocal microscopy images were collected using the LSM 710 Laser Scanning Microscope (Carl Zeiss, Jena, Germany).

The assay was adapted from Ref. 17. In short, dilutions of heat-inactivated GBS were made in DMEM and stained by Oregon green. Samples of 2.5 × 10⁶ GBS were added to macrophages and were allowed to incubate for 30 min at 37°C with shaking. After washing with ice-cold DMEM, the cells were transferred to FACS tubes, fixed with paraformaldehyde (2%) in PBS, and analyzed in a flow cytometer (FACScan; Becton Dickinson). The percentage of Oregon green-positive macrophages was used as a measure of the phagocytic activity. Percentages of Oregon green-positive cells were corrected with the negative control. Results are expressed as percent phagocytosis, resulting from Oregon green-positive macrophages. In similar setup of experiment for intracellular iNOS staining, anti-iNOS (C-11) PE (Santa Cruz Biotechnology) Ab was used.

The ratio of bacterial DNA to mouse DNA was used to monitor bacterial DNA processing in bone marrow-derived macrophages (BMDM). Similarly, the ratio of bacterial cDNA to mouse cDNA was used to monitor bacterial RNA processing in BMDM. Purity of RNA was verified with non-reverse-transcribed RNA control to exclude contamination with residual DNA. The GBS-specific caf gene was used for determination of GBS DNA or RNA from respective Ct values. Mouse-specific gapdh gene was used for determination of macrophage DNA or RNA from its raw Ct values. For determination of GBS DNA or RNA processing, Ct curves were generated by using the 2^–ΔCt method, where ΔCt is the difference between raw Ct values of the bacteria-specific gene and the raw Ct values of the BMDM-specific gene. The levels of GBS DNA or RNA are relative to the level of BMDM, which was set at 1 (18).

Total RNA was extracted from samples using the RNeasy Mini Kit according to the instruction manual (Qiagen). For quantitative two-step RT-PCR, 2 μg total RNA was reverse transcribed to first-strand cDNA with the iScript cDNA synthesis kit (Bio-Rad, Hercules, CA). Aliquots of 20 ng of first-strand cDNA were subsequently used as a template for quantitative PCR with gene-specific primers. The mouse-specific gapdh gene served as a control for constitutive gene expression. Gapdh expression was unchanged after bacterial inoculation when compared with the amount of 18S rRNA. Amplifications were performed in 20 μl SYBR Green JumpStart Taq ReadyMix (Sigma-Aldrich) with 350 nM oligonucleotides using an Eppendorf Realplex Thermal Cycler (Eppendorf). After an initial activation step at 95°C for 7 min, 40 cycles (94°C for 30 s, 60°C for 30 s, 72°C for 30 s, and 82°C for 15 s) were performed, and a single fluorescent reading was obtained after the 82°C step of each cycle. A melting curve was determined at the end of cycling to ensure amplification of only a single PCR product. Ct values were determined with the Realplex V2.0 software. Comparative expression levels (2^ΔCt) were calculated according to Ref. 18. Expression levels are relative to the level of gapdh expression, which was constant in all RNA samples used and was set to 1. Values are the representative of six samples of two biological experiments assayed by quantitative PCR in triplicate.

The nitrite accumulation in macrophage supernatant was measured as an indicator of NO production based on the Griess reaction. Briefly, 100 μl cell culture medium was mixed with 100 μl Griess reagent [equal volumes of 1% (w/v) sulfanilamide in 5% (v/v) phosphoric acid and 0.1% (w/v) naphthylethylenediamine-HCl], incubated at room temperature for 10 min. Then, the absorbance at 540 nm was measured in a Sunrise Microplate Reader (Tecan). Fresh culture medium was used as the blank in all experiments. The amount of nitrite in the samples was compared with a sodium nitrite serial dilution standard curve, and nitrite production was calculated.

Real-time live cell imaging of NO induction was performed using the fluorescent stain 4-amino-5-methylamino-2′,7′-difluorescein diacetate (DAF-FM DA; Molecular Probes). BMDM were preloaded with DAF-FM DA according to the product manual followed by stimulation with rhodamine-labeled GBS at MOI 50. Phagocytosis and NO induction was imaged in real-time up to 300 min using a confocal laser scanning microscope (Zeiss 710) equipped with a live cell imaging station.

TNF levels in macrophage culture medium were quantified by ELISA kits according to the manufacturer’s instructions (R&D Systems).

All of the experiments were performed at least three times. Statistical significance was evaluated using the Student t test for selected experiments.

Oligonucleotides used in this study are as follows: Streptococcus agalactiae cAMP gene (caf) (accession number X72754), 5′-TAATCAAGCCCAGCAAATGG-3′ and 5′-GTTGGCACGCAATGAAGTCT-3′; Mus musculus GAPDH (accession number NM_008084), 5′-TCTCCATGGTGGTGAAGACA-3′ and 5′-ACTCCACTCACGGCAAATTC-3′; Mus musculus TNF (Tnf) (accession number NM_013693), 5′-TCGTAGCAAACCACCAAGTG-3′ and 5′-CCTTGTCCCTTGAAGAGAACC-3′; Mus musculus NO synthase 2 (iNOS) (accession number NM_010927), 5′-GCTTCACTTCCAATGCAACA-3′ and 5′-GGCTGGACTTTTCACTCTGC-3′.

The type I cytokine response to GBS organisms requires recognition of bacterial ssRNA in phagosomes. This is mediated by the TLR adapter protein MyD88 (4). Accordingly, we questioned whether MyD88 was involved in the processing of phagosomes containing bacteria. As depicted in Supplemental Fig. 1, wild-type (WT) and MyD88-deficient macrophages were indistinguishable in GBS phagocytosis, both with respect to the kinetics of the process and the amount of ingested bacteria. Next, we carefully analyzed the kinetics of GBS processing in macrophages by confocal microscopy for up to 48 h after exposure to GBS. Thirty minutes postinfection, many GBS were found in phagocytic vacuoles (Fig. 1A). Analysis at subsequent intervals from 1 to 4 h after inoculation revealed increasing numbers of intracellular bacteria. After 24 h, extensive loss of nucleic acids from the GBS particles occurred, as revealed by the loss of Hoechst 33258 staining of the bacteria (Fig. 1A). Expression of MyD88 essentially contributed to the loss of phagosomal GBS nucleic acids (Fig. 1A). Analysis by quantitative PCR revealed that loss of both macrophage-associated GBS DNA and RNA was reduced in MYD88Ko macrophages compared with WT macrophages (Fig. 1B, 1C). These findings strongly suggested that MyD88 was modulating the phagosomal processing of bacterial nucleic acids.

GBS and other Gram-positive bacteria such as staphylococci and streptococci induce inflammatory cytokines in macrophages in strict dependence on phagocytosis and MyD88. However, the signaling events underlying regulation of phagosomal properties upon processing of bacteria by macrophages are less well established. In this study, we assessed how the inducible lysosomotropic NO was regulated, both with respect to uptake and requirement of individual TLRs and TLR adapters. MyD88 was absolutely required for NO formation, whereas the individual TLRs 2, 4, 7, 8, and 9 were dispensable (Fig. 2A, 2B). Furthermore, TLR3 signaling was not involved, as deletion of the essential TLR3 adapter TIR-domain-containing adapter-inducing INF-β did not impair the NO response (Fig. 2B). UNC-93B, which is used primarily by endosomal nucleic acid-recognizing TLRs, has recently been demonstrated to play an important role in cytokine production in response to GBS (4). In accordance with these data, we found that BMDM from UNC-93B mutant mice were unable to mount an NO response to GBS, whereas the response to the Gram-negative rod Escherichia coli and to its TLR4 ligand LPS were independent of UNC-93B (Fig. 2C). In addition to MyD88 and UNC-93B, phagocytosis was essential for GBS-induced NO formation, as indicated by the dose-dependent inhibitory effect of cytochalasin D, which interferes with F-actin polymerization (Fig. 2E). Ablation of phagocytosis by cytochalasin D treatment was verified by microscopy and FACS analysis (data not shown). These data were in line with previous observations (19). In addition, acidification of the GBS-containing phagosome was necessary for NO induction, as clearly indicated by the inhibitory effect of chloroquine on this process (Fig. 2D).

Next, we determined the kinetics of iNOS and NO formation and its effects on host structures. We found that inos transcription was induced as early as 20 to 30 min after initial contact with GBS. This was accompanied by induction of NO at the GBS-containing phagosomes, as determined by time-resolved live cell microscopy using the sensitive NO indicator DAF-FM DA and rhodamine-labeled bacteria. In addition, formation of iNOS protein was measured 1 h after stimulation by immunostaining and FACS analysis (Fig. 3A–D and Supplemental Videos 1 and 2). NO potentially modulates protein function through posttranslational nitration of amino acids (7). In macrophages processing GBS, we found nitration of tyrosine residues in the vicinity of the GBS-containing phagosomes of BMDM from WT, but not iNOSKo or MYD88Ko, mice (Fig. 3E–L and data not shown). Accordingly, GBS rapidly induced NO, which in turn modified the protein scaffold of the bacteria-bearing phagosome.

In conclusion, these experiments suggested that the rapid NO response to GBS required phagocytosis of the bacteria and MyD88 expression of the macrophages.

Because MyD88 was involved in both NO formation and lysosomal processing of nucleic acids from GBS, we addressed whether both processes were directly interlinked; that is, whether MyD88 exerted its lysosomotropic, cytokine-modulating properties at least partially via NO. First, we assessed whether NO was involved in the intracellular killing of GBS in macrophages from iNOSKo mice. We found that NO was an important component of the killing machinery of GBS, as the average GBS half-life was 11 ± 1 min in iNOSKo macrophages as opposed to 7.5 ± 2 min in WT macrophages (Fig. 4A and data not shown). Notably, as early as 1 h after macrophage infection, bacterial loads were higher in iNOSKo compared with WT macrophages. This is in line with the observation presented earlier that formation of GBS-induced NO is a rapid process, which can be detected as early as 20 min after initial stimulation. It remains to be established whether differences in basal iNOS between iNOSKo and WT mice additionally contribute to differences in killing. We further determined whether NO plays a role in the processing of GBS nucleic acids. Notably, iNOSKo macrophages were substantially retarded in phagosomal GBS RNA processing compared with WT macrophages (Fig. 4B). In contrast, macrophages from gp91Ko mice, which do not form reactive oxygen species (ROS) in response to GBS, were normal in their ability to process RNA (Fig. 4B). To independently substantiate whether NO played a direct role in RNA processing, we preincubated GBS with the potent nitrosating agent sodium nitroprusside (SNP) (20) for 30 min at room temperature in the dark before adding the bacteria to the macrophages. Addition of SNP propagated processing of GBS RNA in macrophages in a dose-dependent manner (Fig. 4C). This finding suggested a role of NO in the processing of GBS nucleic acids in the phagosome, whereas inducible ROS appeared to be less important. Next, we asked whether the propagation of nucleic acid processing was part of a broader role of NO in phagosomal maturation. Accordingly, we measured the acidification of bacteria-containing phagosomes in iNOS-deficient and -sufficient macrophages. Specifically, we assessed the number of GBS-containing vesicles with either pH 4–5 or pH 5–6 compared with pH standards (Fig. 4D). Vesicles in at least 50 cells per condition in each experiment were counted, and the data from three independent experiments were combined. We found that after 24 h of incubation, 70% of GBS were localized in acidic phagosomes in WT macrophages, whereas the same was true for only ∼30% of GBS in iNOS-deficient macrophages (Fig. 4E).

The impact of both MyD88 and NO on RNA processing was intriguing, as streptococcal ssRNA is essential for the activation of type I inflammatory cytokines by whole streptococci in macrophages (4). In contrast, depletion of DNA with DNAse remained without effect on cytokine formation by GBS (4). Furthermore, the NO response was dependent on streptococcal ssRNA (Fig. 4F). Hence, it was important to address whether NO modulated cytokine formation through its effects on streptococcal RNA.

We compared the TNF response of WT, gp91^PhoxKo, iNOSKo, and MYD88Ko macrophages upon stimulation with various concentrations of GBS. Cathepsin B, which has previously been suggested to be involved in phagosomal processes leading to the activation of the inflammasome (21), was not critical for cytoplasmic recognition of GBS ssRNA. Inhibition of cathepsin B remained without effect on GBS-induced cytokine formation (Supplemental Fig. 3). Whereas gp91^PhoxKo macrophages exhibited a normal TNF response to GBS, iNOSKo macrophages had an ∼50% reduced TNF response compared with WT macrophages (Fig. 5A). This effect of iNOS was not limited to TNF but was also found with respect to formation of IL-6 and MIP-1α (Fig. 5C). The contribution of iNOS to cytokine formation was not only evident in the context of fixed GBS, which corresponds to a predominantly particle-driven activation mode, but was also observed upon infection of primary macrophages with live GBS (Fig. 5B). This seemed important, as GBS survived longer in iNOSKo macrophages than in WT macrophages. Therefore, the phagosomal presence of metabolic GBS products in iNOSKo macrophages likely exceeds that in WT macrophages. Based on this observation, we analyzed whether the enzyme iNOS or its product NO contributed to the TNF response. We found that NO was an essential intermediate in the GBS-induced cytokine formation, as the NO donor SNP fully reconstituted the TNF response in iNOS-deficient cells. Importantly, SNP alone did not affect TNF formation (Fig. 5D). Therefore, the iNOS product NO in the phagosome appeared to propagate cytokine formation. To determine whether the compensatory effect of SNP occurred at the transcriptional level, we analyzed TNF gene expression in SNP-treated samples. Similar to protein formation, activation of tnf by GBS was iNOS dependent and was compensated for by SNP in iNOSKo macrophages (Supplemental Fig. 2). It appears that the effect of iNOS on phagosomal acidification was the putative link to its effect on TNF formation, as the alkalinizing agent chloroquine inhibited GBS-induced TNF formation in a dose-dependent manner (Fig. 5E and the model in Fig. 6). Importantly, SNP did not reconstitute the cytokine response to GBS in MYD88Ko macrophages (data not shown). Therefore, NO is an important but not a sufficient downstream intermediate of MyD88 in the cytokine response to GBS.

In this study, we introduce a complex novel model of reciprocal interactions between GBS and macrophages during phagosomal processing of the bacteria, which results in the recognition of GBS ssRNA. During this process, the highly reactive and diffusible NO is formed by iNOS, which is tied to a signaling cascade downstream of MyD88 (22). NO propagates nucleic acid processing and acidification of the bacteria-containing phagosome and mediates inflammatory activation by streptococci (Fig. 6). Streptococci are well adapted to a coexistence with the host, and mucosal sites constitute typical interfaces. GBS colonizes up to 20% of all adults (genital and intestinal tract) and 10% of newborn infants (23). Compared with this physiological situation, invasive infections are relatively rare (for GBS only in 1% of colonized infants). In view of this, it is tempting to visualize a model whereby the colonizing bacterial community exploits the innate immune system as a shepherd that punishes individual bacteria with deviating invasive behavior to guarantee persistence of the colonizing community.

The role of macrophages in the regulation of colonization has been demonstrated by the increase of invasive streptococcal disease in a mouse colonization model where local macrophages are eliminated by intranasal clodronate (24). The potential costs of a highly alert monocytic defense system have been demonstrated with respect to so-called inflammatory monocytes, as overexpression of the chemokine receptor CCR2 in the lung induces chronic inflammatory changes (25). A powerful and timely cellular reaction requires a potent response by cells that have seen low numbers of bacteria (as otherwise the system may be overwhelmed by the bacterial load). Hence, it makes inherent sense that processing of bacterial particles is intercalated into the inflammatory activation of the macrophage.

NO is clearly part of the cellular killing machinery for bacteria, as gp91^phox/iNOS double knockout mice are much more prone to invasive bacterial disease than conventional gp91^Phox knockout mice (26). We show in this article that NO contributes to the bactericidal activity in macrophages. Moreover, we find that NO exerts more global lysosomotropic effects beyond its direct antibacterial function. NO propagates acidification of the bacteria-containing phagolysosome. Modulation of the intraluminal pH has several important direct and indirect implications for phagolysosomal processes such as modification of microbial substructures. As an example, interaction of hypomethylated CpG DNA with TLR9 heavily depends on an acidic intraluminal environment (27). Furthermore, lysosomal proteases exert their function in a pH-dependent fashion (28).

The short-lived NO seems to be an ideal intermediate for regulating the activation level of the single macrophage and the intercellular network in the macrophage vicinity. The rapid induction of NO and accumulation of nitrated proteins at the GBS-bearing phagosome and the role it plays in bacterial processing puts NO very upstream in the signaling propagation cascade, well before secondary products such as TNF will reciprocally modify cellular activity (19). This interdependence of particle processing and inflammatory activation of macrophages may be especially important under non-opsonizing conditions; for example, in children lacking protective Abs and in the serum-starved submucosal tissue where ITAM-dependent activation through Fcγ receptors is poorly developed.

The data presented in this study may provide the molecular basis for the previous observation of an unresolved proinflammatory role of NO in GBS sepsis mortality (9). However, it remains to be established whether NO exerts phagosomal acidification through nitration of phagosomal proteins via chemical properties of the NO product peroxynitrite or any other mechanism. Importantly, ROS are fully dispensable in this context.

In summary, this study identifies NO as a macrophage autonomous signaling modifier in the ssRNA-induced response to streptococci.

We thank S. Bauer and M. Freudenberg for the provision of knockout mice, B. Kremer, K. Bruckner, S.-H. Seibel, L. Fuchs, and A. Imm for outstanding technical assistance, and A.-M. Eades-Perner for help in preparing the manuscript.

The authors have no financial conflicts of interest.