|Home | About | Journals | Submit | Contact Us | Français|
During advanced stages of inhalation anthrax, Bacillus anthracis accumulates at high levels in the bloodstream of the infected host. This bacteremia leads to sepsis during late-stage anthrax; however, the mechanisms through which B. anthracis-derived factors contribute to the pathology of infected hosts are poorly defined. Peptidoglycan, a major component of the cell wall of Gram-positive bacteria, can provoke symptoms of sepsis in animal models. We have previously shown that peptidoglycan of B. anthracis can induce the production of proinflammatory cytokines by cells in human blood. Here, we show that biologically active peptidoglycan is shed from an active culture of encapsulated B. anthracis strain Ames in blood. Peptidoglycan is able to bind to surfaces of responding cells, and internalization of peptidoglycan is required for the production of inflammatory cytokines. We also show that the peptidoglycan traffics to lysosomes, and lysosomal function is required for cytokine production. We conclude that peptidoglycan of B. anthracis is initially bound by an unknown extracellular receptor, is phagocytosed, and traffics to lysosomes, where it is degraded to a product recognized by an intracellular receptor. Binding of the peptidoglycan product to the intracellular receptor causes a proinflammatory response. These findings provide new insight into the mechanism by which B. anthracis triggers sepsis during a critical stage of anthrax disease.
Inhalation anthrax, caused by Bacillus anthracis, is an insidious disease with a poor prognosis. Infections occur when endospores of the bacterium are phagocytosed by macrophages, carried to regional lymph nodes, and escape from the lung environment. During the course of infection, the bacteria grow rapidly and enter the bloodstream. In an acute infection, levels of bacteria as high as 108 bacteria per ml of blood have been recorded (7). If the disease is unabated, vegetative bacteria can be found in almost every organ, presenting a life-threatening situation that is difficult to treat (15, 19). Indeed, death is thought to occur due to toxemia (50) and/or septicemia (47) that results from a high bacterial load in the infected host. Despite exceptional advances in characterizing the early stages of anthrax infection and the effects of the toxin, very little is known about the B. anthracis factors which contribute to sepsis as the disease progresses.
In humans, sepsis is caused by an exaggerated proinflammatory response. For Gram-negative organisms, the proximal cause is the ability of an endotoxin, a lipopolysaccharide (LPS), to stimulate innate immune cells via Toll-like receptor 4 (TLR4) (51). Gram-positive bacteria also cause sepsis, and the proximal cause is often attributable to superantigens derived from Staphylococcus or Streptococcus species, which stimulate polyclonal T-cell activation through their antigen receptors to cause toxic shock syndrome (11). More recent studies of the causes of Gram-positive sepsis have revealed roles for lipoteichoic acid (LTA) (21), a Toll-like receptor 2 (TLR2) agonist (38), triacylated lipopeptides (17), an agonist for TLR2, TLR4, and TLR6 (44, 49), and peptidoglycan (PGN) (22), whose products bind to nucleotide oligomerization domain (NOD) receptors (4, 12-14). The PGN products include N-acetylated muramic acid-alanine-glutamine-diaminopimelic acid (M-Tri-DAP) and N-acetylated muramic acid-alanine-glutamic acid (MDP). Thus, PGN from B. anthracis is a prime candidate for the factor that causes sepsis during advanced stages of systemic anthrax.
We previously reported that B. anthracis cell wall extracts were able to induce production of proinflammatory cytokines in cells in human blood (26). The cell wall extracts lacked detectable LTA, and the biological activity was sensitive to lysozyme, suggesting that the biologically active material was PGN. Here, we used B. anthracis PGN whose purity was very high to examine the molecular and cellular mechanism by which this molecule stimulates inflammatory responses. Our findings indicate that B. anthracis PGN interacts with an extracellular receptor on select hematopoietically derived cells and that receptor endocytosis or phagocytosis is required for the induction of proinflammatory cytokines. Endocytosed or phagocytosed PGN was found to traffic to lysosomes, and an acidified lysosome was also required for PGN biological activity. In contrast to these results for human cells, mouse macrophages were found to be less dependent on internalization of PGN and of acidified lysosomes to mount a proinflammatory response, and NOD-deficient and wild-type macrophages responded equivalently. The data are consistent with a model in which B. anthracis PGN is engaged by human cells, internalized, and degraded in order to trigger a potent inflammatory response. In addition, critical differences in this process between human and mouse cells suggest that more focused studies directed at primate and human models are important for understanding this aspect of anthrax disease in humans.
Fluorochrome-conjugated antibodies to human CD14, CD3, and CD19 were purchased from Invitrogen, while CD14 conjugated to phycoerythrin was purchased from eBioscience. Antibody to human CD16b was purchased from Accurate Chemical & Scientific Corporation, and lysosome-associated membrane protein 1 (LAMP1) conjugated to Alexa Fluor 647 was purchased from Santa Cruz. Human IgG was purchased from Lampire Biological Laboratories. A monocyte isolation II kit was purchased from Miltenyi Biotech. Phalloidin conjugated to Alexa Fluor 633 was purchased from Molecular Probes. The enzymes proteinase K and DNase I and a SuperScript III first-strand synthesis kit were purchased from Invitrogen. RNase A was purchased from Qiagen, and Sybr green PCR master mixture was obtained from Applied Biosystems. Brefeldin A was purchased from eBioscience, and a Cambrex Limulus amebocyte lysate assay kit was purchased from Lonza. Ultrapure lipopolysaccharide (LPS), saponin, guanidine HCl, iodoacetamide, and heparin were purchased from Sigma. A silkworm larvae plasma test kit was purchased from Wako Chemicals.
Peptidoglycan was isolated from B. anthracis strain ΔSterne, which lacks the capsule and toxins, as described by Langer et al. (26). Briefly, PGN was isolated from overnight cultures of the bacteria grown on tryptic soy broth plates. The bacteria were boiled in 8% SDS for 30 min, which was followed by centrifugation. The pellet obtained was washed with endotoxin-free water and subjected to DNase I and RNase A treatment for 15 min at room temperature. The sample was boiled again in 4% SDS for 30 min and then washed three times with endotoxin-free water. The pellet was treated with 2 M NaCl and then washed six times with endotoxin-free water. The pellet was then dried, weighed, and resuspended in endotoxin-free water.
PGN was treated with HF to remove the polysaccharide associated with the PGN as described previously (8). Following HF treatment, the water-washed PGN was subjected to treatment with a denaturing buffer (50 mM Tris [pH 8.0], 6 M guanidine HCl, 25 mM dithiothreitol [DTT]) at 60°C for 1 h. Iodoacetamide was added to a final concentration of 75 mM and the preparation was incubated for 15 min in the dark to alkylate Cys residues. The reaction was stopped by adding DTT to a final concentration of 40 mM. The PGN was resuspended in a buffer containing 50 mM Tris (pH 7.5), 1 M guanidine HCl, and 5 mM CaCl2 and was treated with 20 μg proteinase K (added every 12 h for 36 h) at 50°C for 36 h. The PGN was washed three times with endotoxin-free water, dried, weighed, and resuspended in endotoxin-free water.
For amino acid analysis, 200 μg of PGN was dried with norleucine as an internal standard and treated with 6 N HCl in sealed vacuum tubes for 20 to 24 h at 110°C. The hydrolysate was then subjected to cation-exchange high-performance liquid chromatography (HPLC) with ninhydrin detection to determine the amino acid content of PGN. The amino acid analysis was performed at the Molecular Biology-Proteomics Facility at the University of Oklahoma Health Science Center.
For labeling with fluorescein, PGN was treated with fluorescein isothiocyanate (FITC) at a ratio of 1:2 (wt/wt) in 100 mM NaHCO3 at pH 8.0. After 30 min of incubation at room temperature, the labeled PGN was washed thoroughly with endotoxin-free water and resuspended in water.
Heparinized peripheral blood (PB) was obtained by venipuncture after signed consent of healthy human volunteers; the venipuncture protocol was approved by an internal review board. The blood was diluted 1:3 with Dulbecco's modified Eagle's medium and cultured in non-tissue-culture 24-well plates. PB mononuclear cells were obtained by Ficoll-Hypaque density gradient centrifugation. Neutrophils were separated from the red blood cells by brief incubation in ACK lysis buffer (155 mM ammonium chloride, 10 mM KHCO3, 0.1 mM EDTA). Monocytes were purified from PB mononuclear cells by using a monocyte isolation II kit according to the manufacturer's instructions.
Wild-type mice with a C57BL/6 background were used for experiments. Bones were obtained from NOD1−/− (4), NOD2−/− (25), and NOD1−/− NOD2−/− (41) mice that had a C57BL/6 background. Bone marrow was flushed from the femurs and tibias and cultured in RPMI 1640 medium supplemented with 10% fetal bovine serum, macrophage colony-stimulating factor (M-CSF), penicillin, streptomycin, and glutamine for 7 days as described previously (36). From day 7 on bone marrow-derived macrophages (BMDM) expressed the mature macrophage markers Mac-1 and FcγRII/III, and they were cytokine starved prior to treatment with PGN (10 μg/ml) or LPS (1 μg/ml).
Defibrinated sheep blood was infected with a single mucoid colony of B. anthracis strain Ames 34 and incubated at 37°C overnight in the presence of 20% CO2 to obtain actively growing capsulated bacteria. The following day, the blood was diluted 1:10 in fresh defibrinated blood and incubated for an additional 6 h at 37°C. Serum was collected after incubation by centrifugation, filtered through a 0.22-μm filter, and heated at 80°C for 10 min. The serum was then incubated with silkworm larva plasma reagent at 30°C for 90 min. Peptidoglycan was detected spectrophotometrically at 650 nm and was quantitated using a standard curve.
For detection of cytokines and chemokines by flow cytometry, diluted blood was treated with PGN (10 μg/ml) or LPS (1 μg/ml) for 6 h in the presence of brefeldin A (3 μg/ml) to inhibit secretion of cytokines and chemokines. Following treatment, the cells were suspended in 100 μl of fluorescence-activated cell sorting buffer (1× phosphate-buffered saline [PBS], 3% fetal bovine serum) containing brefeldin A and incubated with human IgG for 10 min on ice to block Fc receptors (36). Cells were then incubated with appropriate cell surface antibodies for 20 min on ice. The samples were treated with ACK lysis buffer for 5 min at room temperature to lyse the red blood cells. The white blood cells (WBC) were then washed with PBS containing brefeldin A, fixed with 2% formaldehyde, and permeabilized with 0.5% saponin. The permeabilized cells were stained with antibodies against tumor necrosis factor alpha (TNF-α), interleukin-1β (IL-1β), or IL-8. The cells positive for the cytokines or chemokines were identified by flow cytometry analysis by collecting 200,000 events per sample. In experiments in which inhibitors were used, the diluted blood was first preincubated with the inhibitors for 30 min and then treated with PGN or LPS.
To observe internalization of PGN in WBC, diluted blood was treated with 10 μg/ml of fluorescein isothiocyanate (FITC)-conjugated PGN (PGN-FITC) for 30 min at 37°C. The WBC were stained with unique cell surface markers, fixed in 2% formaldehyde, and sorted by flow cytometry to obtain >98% pure populations of monocytes, neutrophils, and B lymphocytes. The cells were permeabilized and stained with phalloidin conjugated to Alexa Fluor 633 and 4′,6′-diamidino-2-phenylindole (DAPI). The cells were observed with a Zeiss inverted deconvolution microscope or a Zeiss LSM 510 confocal microscope using a 63× oil objective.
For colocalization experiments, diluted blood was treated with 10 μg/ml of PGN-FITC for 60 min at 37°C. The WBC were stained with unique cell surface markers, fixed in 2% formaldehyde, and sorted by flow cytometry to obtain >98% pure populations of monocytes and neutrophils. The cells were then permeabilized, blocked with 3% bovine serum albumin (BSA) in PBS, and stained with an antibody against LAMP1 conjugated to Alexa Fluor 647. The cells were then observed with a Zeiss LSM 510 confocal microscope. All microscopy images were prepared using Adobe Photoshop.
Purified monocytes, neutrophils, B lymphocytes, and T lymphocytes were obtained by flow cytometry and were lysed in Trizol. Total RNA was isolated, and cDNA was synthesized from 1 μg of RNA using the Superscript First Strand synthesis system for reverse transcription-PCR (RT-PCR). Semiquantitative PCR was performed using Sybr green PCR master mixture and primers corresponding to the actin, NOD1, or NOD2 sequence. The following primers were used for amplification of actin: forward primer 5′-ACAACGGCTCCGGCATGTGCAA-3′ and reverse primer 5′-CATGTCGTCCCAGTTGGTGACGAT-3′. The NOD primers have been described previously (24). Their sequences are as follows: for NOD1, forward primer 5′-TCCAAAGCCAAACAGAAACTC-3′ and reverse primer 5′-CAGCATCCAGATGAACGTG-3′; and for NOD2, forward primer 5′-GAAGTACATCCGCACCGAG-3′ and antisense primer 5′-GACACCATCCATGAGAAGACAG-3′. DNA was amplified with the 7500 real-time PCR system (Applied Biosystems, Foster City, CA) using the following parameters: 10 min at 95°C, followed by 40 cycles of 95°C for 15 s and 60°C for 1 min. Analysis of relative gene expression was performed as described by Livak and Schmittgen (29).
All statistical analyses were performed using Microsoft Excel. Statistical significance was determined by using a t test, and a P value of <0.05 was considered statistically significant.
Since the poly-d-glutamic acid capsule forms the outermost layer of B. anthracis, PGN may not be biologically available to stimulate immune cells. Yet, it is also formally possible that PGN is released during growth and division of B. anthracis. To distinguish between these possibilities, we inoculated defibrinated sheep blood with B. anthracis strain Ames, a fully encapsulated anthrax strain, and incubated the bacteria for 6 h. The blood was centrifuged, and the serum was collected and filtered to remove vegetative bacteria. Encapsulation of the growing bacteria was confirmed by microscopy (data not shown). We determined the presence of PGN in the serum by using a commercial reagent that can detect the presence of peptidoglycan based on the insect prophenol oxidase signaling cascade (54). As shown in Fig. Fig.1,1, PGN was detected readily in the serum of sheep infected with B. anthracis Ames and minimally in the uninfected blood sample. These findings show that, despite the presence of the capsule, PGN is shed and hence is accessible to immune cells during infection.
We reported previously that B. anthracis-derived PGN was a potent inducer of proinflammatory cytokines from cells in whole human blood (26). Although the PGN biological activity was sensitive to lysozyme and not sensitive to proteases, analysis of the amino acid composition of the PGN revealed a complex array of amino acids, suggesting that there was contamination with bacterially derived protein. To improve the purity of B. anthracis PGN, we treated the SDS-extracted material with hydrofluoric acid (HF) to remove the bound polysaccharide, as described previously (8), and treated the HF-digested product with proteinase K. The amino acid compositions of the PGN before and after proteinase K treatment are shown in Fig. Fig.2A.2A. We noted that the amount of pimelic acid did not decrease after this treatment, while the amounts of the vast majority of the other amino acids were greatly decreased. The major amino acids that were present in the proteinase K-treated sample were pimelic acid, alanine, and glutamic acid/glutamine, which is consistent with the amino acid residues in the PGN stem peptide of members of the family Bacillaceae (45). We next compared the ability of the proteinase K-treated PGN to elicit proinflammatory cytokines to the ability of the protein-contaminated material that we used in our previous study to elicit proinflammatory cytokines. For this experiment, we treated human blood with the PGN preparations for 6 h, fixed and permeabilized the white blood cells, and measured the TNF-α in the cells by flow cytometry. The data are expressed here as the percentage of CD14+ monocytes that were positive for TNF-α. The results (Fig. (Fig.2B)2B) show that PGN induced monocytes to make TNF-α and that there was a minimal decrease in the percentage of monocytes that were positive for TNF-α production following proteinase K treatment of the stimulating PGN. Lastly, we used the proteinase K-treated material in a dose-response experiment to determine the minimal amount needed to elicit TNF-α production by human blood monocytes. We found (Fig. (Fig.2C)2C) that TNF-α was readily detected with doses of PGN greater than 5 μg/ml, which is similar to the dose needed for the undigested material and reported previously (26). This dose is comparable to the doses of PGN used for other bacterial strains in studies evaluating the inflammatory role of PGN (5, 52). The data are consistent with our previous conclusion that the biological activity in a B. anthracis PGN preparation is due to PGN itself and not to any contaminating material.
We reported previously that blood cells produced large amounts of TNF-α, IL-6, IL-1β, and IL-8 and that blood monocytes were the sole source of the TNF-α (26), yet the cell types that released the other cytokines were not determined. We resolved the different hematopoietic populations in blood using unique cell surface markers (monocytes [CD14+ CD16b−], neutrophils [CD14− CD16b+], B lymphocytes [CD19+ CD3−], and T lymphocytes [CD19− CD3+]), combined with forward and side scatter properties. Using this gating strategy together with intracellular cytokine staining, we examined which blood cell population produced TNF-α, IL-6, IL-1β, and IL-8. The primary data for cytoplasmic staining of IL-8 in the four gated blood cell populations are shown in Fig. 3A, and a summary for all of the cytokines and cell populations is shown in Fig. Fig.3B.3B. We found that only monocytes produced IL-6 and IL-1β, while both monocytes and neutrophils produced IL-8 in response to PGN. Neither lymphocyte population responded to PGN treatment.
We hypothesized that lymphocytes do not respond to B. anthracis PGN because the cells do not bind PGN. We labeled the PGN with fluorescein isothiocyanate (FITC) to determine which blood cell population was able to bind PGN. First, to test whether FITC labeling of PGN altered its biological activity, we compared unlabeled PGN and FITC-labeled PGN for induction of TNF-α. We found (Fig. (Fig.4A)4A) that the percentages of TNF-α-expressing monocytes were similar (~40%) for the two forms of PGN, confirming that the labeling procedure did not render the PGN inactive.
We evaluated the binding of PGN-FITC to WBC by incubating diluted peripheral blood with PGN-FITC for 30 min at 4°C to prevent internalization. Following PGN incubation, the cells were stained with the distinguishing cell surface markers, and PGN-FITC binding to each individual population was determined by flow cytometry. We found that in the 30-min assay, PGN bound about 40 to 50% of the neutrophils (Fig. (Fig.4B)4B) and monocytes (Fig. (Fig.4C)4C) and about 30% of the B lymphocytes (Fig. (Fig.4E),4E), while T lymphocytes were not bound (Fig. (Fig.4D).4D). A summary of the binding properties of these cells is shown in Fig. Fig.4F4F.
To test if the same cells that bind PGN-FITC also make TNF-α, we electronically gated on CD14+ monocytes and measured the levels of PGN-FITC and cytoplasmic TNF-α (Fig. 4G to I). We found (Fig. (Fig.4I)4I) that the vast majority (91%) of TNF-α-producing monocytes also bound PGN (3% PGN− TNF-α+ cells, compared with 31% PGN+ TNF-α+ cells). Overall, the data show that the cells that produce proinflammatory cytokines (monocytes and neutrophils) bind PGN, while a significant fraction of B lymphocytes bind PGN yet fail to produce any detectable cytokine.
PGN binding alone or binding followed by internalization might induce the production of proinflammatory cytokines in blood cells. To determine whether PGN internalization was needed for a biological response, we treated the whole blood with an inhibitor of actin reorganization (cytochalasin D or latrunculin A) and measured the percentage of TNF-α+ cells (for monocytes) or IL-8+ cells (for neutrophils). We used LPS as a control since LPS responses are due to stimulation of surface CD14 and TLR4 and do not require endocytosis. The results shown in Fig. Fig.5A5A and and5B5B reveal that both of the actin reorganization inhibitors were able to block PGN-triggered cytokine production, while neither inhibitor was able to block the response to LPS. These findings show that internalization of the PGN is critical for production of TNF-α and IL-8.
We also evaluated the subcellular location of PGN in these cells by microscopy. Peripheral blood cells were treated with PGN-FITC for 30 min at 37°C to allow complete internalization in the majority of cells. We then sorted monocytes (CD14+ CD16b−), neutrophils (CD14− CD16+), and B lymphocytes (CD19+) by flow cytometry. The sorted cells were fixed, permeabilized, and stained with Alexa Fluor 633-conjugated phalloidin and DAPI and observed by microscopy. One of the optical sections (Fig. (Fig.5C5C upper panels) showed that the PGN-FITC was internalized by the monocyte and neutrophil populations. More than 80% of these cells internalized at least one PGN particle. We also found PGN-FITC in ~80% of B lymphocytes (data not shown), but due to the very narrow cytoplasm we cannot be certain that the PGN-FITC was truly internalized. Treatment of the cells with cytochalasin D resulted in disruption of the actin cytoskeleton, as expected, and blocked internalization of PGN-FITC (Fig. (Fig.5C,5C, lower panels). Nevertheless, PGN-FITC was still able to bind to monocytes and neutrophils in the presence of cytochalasin D (Fig. (Fig.5D).5D). These results are consistent with the data shown in Fig. Fig.5A5A and and5B5B indicating that the induction of cytokines requires PGN internalization.
Because the PGN particle is internalized, we hypothesized that PGN traffics to the lysosome for hydrolysis and that PGN hydrolysis is required for production of proinflammatory cytokines by the responding cells. To test this hypothesis, we used two lysosomotropic agents, ammonium chloride and chloroquine, and examined their effects on cytokine production. As shown in Fig. 6A and B, both lysosomotropic agents caused significant reductions in the percentage of monocytes expressing TNF-α and the percentage of neutrophils expressing IL-8 following treatment with PGN. There was no significant reduction in TNF-α production in monocytes that were treated with LPS in the presence of these two lysosomotropic agents, supporting the notion that LPS induces biological responses via surface receptors. However, we did observe a significant reduction in IL-8 production in neutrophils following LPS treatment in the presence of chloroquine but not in the presence of ammonium chloride. Chloroquine can inhibit the activity of mitogen-activated protein kinases in some cells (53). Thus, it is possible that neutrophils are more sensitive to the effect of chloroquine than monocytes.
The results described above suggest that, after internalization, PGN traffics to the lysosome for hydrolysis. To test this possibility, we treated cells with PGN-FITC for 1 h. Monocytes and neutrophils that bound PGN-FITC were sorted by flow cytometry by gating on the appropriate surface-stained populations. The cells were then fixed, permeabilized, stained with an antibody against lysosome-associated membrane protein 1 (LAMP1), and observed with a confocal microscope. We found that ~70% of the FITC+ monocytes (Fig. (Fig.6C)6C) and ~50% of the FITC+ neutrophils (Fig. (Fig.6D)6D) showed colocalization of PGN-FITC in LAMP1-containing vesicles. Thus, after its internalization, PGN traffics to the lysosome, and both the trafficking and lysosome function are required for production of TNF-α in monocytes and for IL-8 production in neutrophils.
Our data suggest that internalization and degradation of PGN are required for recognition by an intracellular receptor. The cytoplasmic NOD receptors bind to lysosomal degradation products of PGN to stimulate cytokine production in the blood cells and are obvious candidates for this process. We first measured the levels of NOD1 and NOD2 mRNA in WBC by semiquantitative PCR. Peripheral blood mononuclear cells were isolated from blood by density gradient centrifugation and purified further by flow cytometry. Based on surface marker expression, we obtained >97% pure populations of monocytes, neutrophils, B lymphocytes, and T lymphocytes. RNA extracted from these cells was used to determine the levels of NOD1 and NOD2 mRNA by semiquantitative real-time PCR using actin as a control. We found (Fig. (Fig.7)7) that B lymphocytes and T lymphocytes had significantly higher levels of NOD1 than monocytes and neutrophils. Conversely, monocytes and neutrophils had significantly higher levels of NOD2 than the lymphocyte populations. We tried to evaluate the role of NODs in cytokine and chemokine production by silencing the expression of NODs using small interfering RNA or short hairpin RNA in monocytes. Unfortunately, transfection of either RNA interference reagent resulted in upregulation of the levels of both NODs following transfection (data not shown), which may have been a cellular response to exogenous nucleic acids.
In an alternative approach, we used mouse bone marrow-derived macrophages (BMDM) from animals lacking NOD1, NOD2, or both NOD1 and NOD2. This model is a compromise since we were not able to cause differentiation of human monocytes to macrophages and could not isolate sufficient mouse blood monocytes. BMDM of wild-type animals responded to B. anthracis PGN by producing TNF-α, although the responses were markedly lower than those in human cells (~20% TNF-α+ murine cells compared with ~50% TNF-α+ human cells) (Fig. (Fig.8A).8A). Surprisingly, analysis of BMDM from NOD1−/−, NOD2−/−, and NOD1−/− NOD2−/− mice showed no difference in the percentage of TNF-α-positive BMDM (Fig. (Fig.8A).8A). This finding suggests that neither NOD1 nor NOD2 plays a role in PGN sensing or that there are other proteins in addition to these NODs that can sense the PGN and generate the production of proinflammatory cytokines. We tested the latter possibility by using cytochalasin D and ammonium chloride with the mouse BMDM and measuring TNF-α production following PGN treatment. We found only a ~50% reduction in the percentage of murine BMDM expressing TNF-α in the presence of either cytochalasin D or ammonium chloride (Fig. (Fig.8B8B and inset), while human cells were ~90% inhibited. These results suggest that mouse BMDM recognize PGN of B. anthracis using an extracellular receptor, similar to LPS, in a process distinct from the process used by human monocytes, which appear to use only a cytoplasmic receptor and not an extracellular receptor.
In inhalation anthrax, B. anthracis produces an extraordinarily high number of bacteria, which is probably facilitated by anthrax toxins and the poly-d-glutamic acid capsule. We reasoned that the high bacterial load should be accompanied by a high level of circulating cell wall components of the vegetative bacteria, including PGN. We reported previously (26) that the PGN of B. anthracis is able to elicit a proinflammatory response in cells in human peripheral blood. We showed that the response was not due to the presence of a PGN-associated polysaccharide, introduced contaminants like Gram-negative endotoxin, bacterial nucleic acids, teichoic acids, or bacterial proteins that may be associated with the PGN. However, the molecular mechanism by which PGN is able to cause such a response is not clear for any Gram-positive organism. In the present study, we expanded the previous findings and examined the molecular and cellular mechanisms through which B. anthracis PGN induces inflammatory responses in human WBC.
Here, we established that PGN within the cell wall of B. anthracis is bioavailable to the host, despite the fact that the organism is surrounded by a poly-d-glutamic acid capsule. We also showed that PGN binds to an unidentified surface receptor on responding cells, but binding is not sufficient to induce cytokine production. PGN internalization, trafficking to lysosomes, and lysosome hydrolysis are required for PGN to induce a biological response. This is the first report to show that PGN from a Gram-positive organism requires internalization and lysosomal trafficking to trigger a host response in the form of production of proinflammatory cytokines and chemokines. Mouse macrophages had a distinct mechanism for activation of proinflammatory cytokines in that the response was less dependent than that of human cells on internalization and acidified lysosomes. While a comparison between mouse macrophages and human monocytes is not ideal, monocytes are the macrophage precursors and the two types of cells express similar types of TLRs (20) and NODs (30). The biological response in humans was associated with cells that express NOD2, and we were not able to determine that any NOD sensor is required for PGN to induce cytokine production in human blood cells. First, RNA interference strategies designed to knock down NOD expression caused upregulation of NOD mRNA instead. Second, NOD-deficient mouse macrophages responded as well as macrophages from wild-type mice, probably because PGN responses are less dependent on a cytoplasmic sensor in mice.
In nature, B. anthracis is known to have a complex cell wall architecture that includes a thick peptidoglycan, a polysaccharide associated with the PGN, S-layer proteins, and a poly-d-glutamic acid-containing capsule (9, 10). Since the capsule forms the outermost layer of the cell wall, it is possible that the PGN may not be accessible to the immune cells of the host. There have been several reports for other members of the family Bacillaceae, like Bacillus subtilis, Bacillus megaterium, and Bacillus cereus, establishing that the organisms shed PGN fragments during the logarithmic growth phase (3, 34, 35). Therefore, it is likely that B. anthracis PGN encounters human blood cells as it grows to produce the high levels of bacteria that have been reported (7).
PGN required internalization and trafficking to a lysosomal compartment to induce cytokine production by blood cells. The observation that biological responses to PGN can be blocked by chemical agents that neutralize the lysosome pH suggests that lysosomal enzymes degrade PGN to a more simple moiety that can be recognized by a cytoplasmic sensor and that PGN itself is not the stimulus for the sensor. This notion is consistent with findings that NOD ligands are lysozyme products of PGN digestion, including M-Tri-DAP and MDP (4, 12-14). Innate immune cells like neutrophils (6) and monocytes/macrophages (27) contain lysozyme in their lysosomes. Although there are no recent studies, a survey over 30 years ago of human B-cell lymphomas (Raji, Daudi, and Ramos) and three Epstein-Barr virus-transformed human B cells showed that lysozyme was not present (42). Thus, human B cells may be capable of internalizing PGN and trafficking it to lysosomes but may not be capable of degrading it to a moiety that can be recognized by a cytoplasmic sensor.
The observation that PGN bound some cells but not other cells suggests that there is a receptor on the surface of the fraction of cells that bind. Given the dose relative to the amount of LPS that is needed to elicit proinflammatory cytokines from human cells with a myeloid origin, we speculate that the receptor has lower affinity and may not be specific for PGN. Mouse cells appear to use surface TLR2 to bind PGN from a variety of Gram-positive bacteria (48), although this hypothesis is controversial. Human TLR2 is able to bind PGN (18), but whether TLR2 is an actual PGN receptor is complicated by the fact that previous studies often used PGN contaminated by LTA (28). We tested the role of TLR2 in the TNF-α response to PGN by using blocking antibodies. We found that although neutralizing anti-TLR2 antibodies did not prevent TNF-α production with our PGN preparation, the antibodies also did not block TNF-α production induced by the known TLR2 ligands LTA and Pam3CSK4 (data not shown). However, we observed that there was no increase in PGN binding to HEK293 cells stably expressing TLR2 compared to the binding to untransfected HEK293 cells, suggesting that TLR2 many not be involved in the proinflammatory response to PGN (data not shown). In any case, our finding that the biological response to PGN requires internalization and lysosomal trafficking is inconsistent with the hypothesis that PGN engages a surface TLR that induces intracellular signaling events. This hypothesis for PGN is in contrast to the findings for LPS, which signals via CD14 and TLR4, is known to utilize MyD88, and did not require internalization or lysosomal trafficking (Fig. (Fig.55 and and66).
The cytoplasmic NOD-like receptor (NLR) family members NOD1 and NOD2, which recognize M-Tri-DAP and MDP, are strong candidate cytoplasmic sensors for human neutrophils and monocytes responding to PGN. Recently, another NLR family member, Nalp3, was shown to recognize muramyl dipeptide (32). Nalp3 is expressed in innate immune cells and, interestingly, is not present in B cells (31). Both M-Tri-DAP and MDP are degradation components of peptidoglycan, making these proteins ideal candidates for sensing peptidoglycan. Although the innate immune cells express mRNA corresponding to NOD1 and NOD2, we were unable to inactivate either sensor with RNA interference strategies and could not use a murine model since the cell biological features of PGN responses were different in mice and humans (Fig. (Fig.88).
Death due to anthrax has been considered to be primarily due to toxemia (1). However, since toxin-deficient B. anthracis strains are lethal in animal models (16), it is possible that septic shock may play a larger role. There is abundant evidence that PGN can cause pathological features similar to those caused by Gram-negative endotoxin (for a review, see reference 39). For example, infusion of PGN from several Gram-positive organisms into pigs (43) or rats (40) caused production of proinflammatory cytokines and evidence of the development of tolerance, two features shared by Gram-negative LPS (for a review, see reference 2). PGN rapidly induces procoagulant activity in monocytes by causing the expression of tissue factor (33), a property shared with LPS (37). Likewise, Staphylococcus aureus PGN can directly cause platelet aggregation in vitro (23), a feature that could contribute to disseminated intravascular coagulation, a clinical characteristic of severe septicemia (46). Guinea pigs infused with ~90 μg PGN/100 mg (body weight) showed evidence of disseminated intravascular coagulation, and 25% of the PGN-challenged animals died within 24 h (22). Our finding that PGN internalization and lysosomal trafficking are required for a proinflammatory response reveals new points of intervention and should help future studies of pharmacological intervention in the treatment of B. anthracis infections and of Gram-positive sepsis.
This work was supported by NIH grant 2U19 AI062629.
Editor: S. R. Blanke
Published ahead of print on 22 March 2010.