Organ Injury and Cytokine Release Caused by Peptidoglycan Are Dependent on the Structural Integrity of the Glycan Chain

Abstract

Several studies have implicated a role of peptidoglycan (PepG) as a pathogenicity factor in sepsis and organ injury, in part by initiating the release of inflammatory mediators. We wanted to elucidate the structural requirements of PepG to trigger inflammatory responses and organ injury. Injection of native PepG into anesthetized rats caused moderate but significant increases in the levels of alanine aminotransferase, aspartate aminotransferase, γ-glutamyl transferase, and bilirubin (markers of hepatic injury and/or dysfunction) and creatinine and urea (markers of renal dysfunction) in serum, whereas PepG pretreated with muramidase to digest the glycan backbone failed to do this. In an ex vivo model of human blood, PepG containing different amino acids induced similar levels of the cytokines tumor necrosis factor alpha (TNF-α), interleukin-6 (IL-6), IL-8, and IL-10, as determined by plasma analyses (enzyme-linked immunosorbent assay). Hydrolysis of the Staphylococcus aureus cross-bridge with lysostaphin resulted in moderately reduced release of TNF-α, IL-6, IL-8, and IL-10, whereas muramidase digestion nearly abolished the ability to induce cytokine release and IL-6 mRNA accumulation in CD14⁺ monocytes compared to intact PepG. However, additional experiments showed that muramidase-treated PepG synergized with lipopolysaccharide to induce TNF-α and IL-10 release in whole blood, despite its lack of inflammatory activity when administered alone. Based on these studies, we hypothesize that the structural integrity of the glycan chain of the PepG molecule is very important for the pathogenic effects of PepG. The amino acid composition of PepG, however, does not seem to be essential for the inflammatory properties of the molecule.

In groups of surgical patients, severe inflammation leading to aberrant regulation of host defense systems may result in significant hemodynamic disturbances and subsequent sepsis with multiple organ dysfunction syndrome (MODS), the major cause of death in surgical intensive care units (3, 9). In gram-negative bacterial infections, lipopolysaccharide (LPS) is the main initiator of sepsis and organ injury (1, 6, 43), in part by triggering the release of the inflammatory cytokines tumor necrosis factor alpha (TNF-α) and interleukin-6 (IL-6) from peripheral blood mononuclear cells and tissue macrophages. Production of the chemokine IL-8 is partly triggered by TNF-α (21) and serves to recruit phagocytes to an infectious site. The anti-inflammatory cytokine IL-10 inhibits production of TNF-α, IL-6, and IL-8 in vitro (15, 17), and IL-10 has been shown to be a functional repressor of monocyte activation in blood from sepsis patients (7).

It has become increasingly clear that LPS cannot fully reproduce the aberrant signaling and clinical features associated with the development of sepsis and MODS, implicating the involvement of other factors. A number of studies suggest that the bacterial component peptidoglycan (PepG) may contribute to this fatal condition (14, 35, 44, 49). PepG may reach the circulation by translocation from the intestine and by bacterial breakdown during gram-positive infections (25, 33, 37). Moreover, some antibiotics may enhance the release of PepG from bacteria (41). We have recently demonstrated that injection of Staphylococcus aureus PepG alone to rats causes injury and/or dysfunction to various organs, which correlates with the induction of both local and systemic inflammation (44).

Consisting of large networks of alternating glycan moieties of N-acetylglucosamine acid (GlcNAc) and N-acetylmuramic acid (MurNAc), cross-linked by short peptides, PepG is the major cell wall component of gram-positive bacteria, and a thin layer of this molecule is also found in gram-negative bacteria. In human monocytes and whole blood, PepG has been shown to initiate the production of inflammatory mediators (27, 29, 39, 45). There is also evidence that PepG may synergize with lipoteichoic acid and LPS to cause organ injury and hemodynamic shock in rodents (14, 24, 28, 49). It has been argued that only part of the PepG molecule is responsible for the inflammatory properties of PepG (24, 26), but the structural requirements of this wall component in initiating or enhancing septic responses are not well defined. The present study was designed to elucidate this important aspect.

PepG isolated from different species of gram-positive bacteria (S. aureus, Bacillus subtilis, and Curtobacterium flaccumfaciens) were compared with respect to cytokine responses, and selective hydrolysis of specific bonds within the PepG molecule was performed with produce substructures likely to be found in vivo (26). The significance of these structures to (i) cause MODS in the anesthetized rat, (ii) initiate cytokine release and mRNA accumulation in whole blood, and (iii) act as potential enhancers or suppressors of cytokine production was assessed.

MATERIALS AND METHODS

Purification of PepG.

B. subtilis, C. flaccumfaciens, and S. aureus PepG were isolated from bacterial cell walls as previously described for B. subtilis (18). Covalently attached proteins were removed by treatment with pronase (2 mg/ml) for 1 h at 60°C (5). Anionic polymers were removed from the PepG by the treatment of purified cell walls (10 mg [dry weight]/ml) with hydrofluoric acid (48% [vol/vol]) for 24 h at 4°C. The insoluble PepG was then washed by centrifugation (14,000 × g, 5 min) and resuspension, once in 100 ml of Tris-HCl (pH 8.0) and five times in distilled water, until the pH was neutral. The PepG was then recovered by centrifugation as described above and resuspended in saline (0.9% [wt/vol]) prior to sterilization by autoclaving and storage at −20°C. Extracts of PepG were subjected to sodium dodecyl sulfate-polyacrylamide gel electrophoresis, with no evidence of protein whatsoever. The PepG was also enzymatically digested, and we obtained the expected reversed-phase high-pressure liquid chromotography muropeptide profile with no spurious products. The PepG was dispersed by sonication (3,000 Hz, 3 × 10 s) prior to the experiments. The PepG preparations were analyzed for the presence of LPS by Limulus amebocyte lysate test (COAMATIC Chromo-LAL; Chromogenix, Falmouth, Mass.) and were shown to contain 35 pg of LPS/mg of PepG. Moreover, pretreatment with the LPS inhibitor polymyxin B or the CD14 antibody 18D11 (which specifically blocks LPS-induced signaling) had no effect on the production of TNF-α induced by PepG (45).

Enzymatic hydrolysis of PepG.

Cellosyl (Hoechst AG, Frankfurt, Germany), a muramidase produced by Streptomyces coelicolor (8), catalyzes the breaking of the bond between MurNAc and GlcNAc in the PepG glycan backbone, resulting in disaccharide subunits substituted with peptide side chains. A proportion of these are linked together by PepG cross-bridging (Fig. 1). Lysostaphin from Staphylococcus simulans (Sigma, St. Louis, Mo.) has bacteriolytic properties specific for S. aureus (40) and hydrolyzes the pentaglycine cross-bridge in S. aureus PepG (Fig. 1). The purity of both enzymes was >99% as seen by sodium dodecyl sulfate-polyacrylamide gel electrophoresis.

FIG. 1.

Illustration of PepG partial structures obtained by digestion of S. aureus PepG with two hydrolases.

The PepG polymer (10 mg/ml) was incubated for 5 h at 37°C with the two different enzymes at an end concentration of 150 μg/ml in 20 mM Kphos buffer (pH 6.5). The reaction was stopped by boiling the sample for 3 min. Insoluble material was removed by centrifugation (14,000 × g, 8 min, room temperature). This procedure is sufficient to completely digest the PepG polymer (>95%) as seen by a decrease in the optical density at 450 nm. The resulting cortex hydrolysate was stored at −20°C at a concentration of 10 mg/ml.

In vivo experiments.

The surgical procedure was carried out on male Wistar rats (200 to 300 g; Tuck, Ltd., Rayleigh, United Kingdom), receiving a standard diet and water ad libitum, in adherence to appropriate institutional guidelines (Home Office Guidance on the Operation of Animals [Scientific Procedure Act 1986], published by HMSO, London, United Kingdom); the experiments were approved by the Home Office of the United Kingdom and are in accordance with the relevant project license. The animals were anesthetized with sodium thiopentone (Intraval Sodium, 120 mg/kg) intraperitoneally, and anesthesia was maintained by supplementary injections as required. The trachea and urinary bladder were cannulated to facilitate respiration and urine flow, respectively. The right jugular vein was cannulated for administration of PepG or vehicle (saline). Cardiovascular parameters were allowed to stabilize for 10 to 15 min, followed by slow injection of either saline (n = 14), S. aureus PepG (10 mg/kg, n = 12), or Cellosyl-treated S. aureus PepG (10 mg/kg, n = 8) over a 10-min period. Throughout the experimental period, rectal temperature was maintained at 37°C by using a homeothermic blanket system (BioScience, Sheerness, Kent, United Kingdom). After 6 h, the animals were given an overdose of sodium thiopentone to end the experiment.

Measurement of markers of organ injury and/or dysfunction.

At 6 h after injection of PepG/saline, blood was collected from the carotid artery and centrifuged at 1,610 × g for 3 min to separate serum. All serum samples were analyzed within 24 h by a contract laboratory for veterinary clinical chemistry (Vetlab Services, Sussex, United Kingdom). The following marker enzymes were measured in the serum as biochemical indicators of multiple organ injury or dysfunction: hepatic injury was assessed by measuring the rise in the levels of specific markers for hepatic parenchymal injury (alanine aminotransferase [ALT] and γ-glutamyl transferase [γ-GT]), a nonspecific marker for hepatic injury (aspartate aminotransferase [AST]), and an indicator of hepatic excretory function (bilirubin) in serum. Renal dysfunction was assessed by measuring the the levels of an indicator of reduced glomerular filtration rate, and hence, renal failure (creatinine), and an indicator of impaired excretory function of the kidney and/or increased catabolism (urea) in serum.

Whole-blood experiments.

The human whole-blood model was used as previously described (47), with some modifications. In brief, venous blood from healthy volunteers was anticoagulated with heparine (30 IU/ml of blood; Leo, Ballerup, Denmark) and incubated in 0.5-ml thin-wall tubes (ABgene House, Epsom, United Kingdom) at 37°C with slow rotation in the absence or presence of 1 μg of PepG/ml or 1 μg of hydrolyzed PepG material (at a hydrolase concentration of 15 ng/ml)/ml. In experiments aimed at investigating costimulatory effects, blood was incubated with both hydrolyzed PepG (0.1, 1, or 10 μg/ml) and LPS (Escherichia coli serotype O26:B6; Difco Laboratories, Detroit, Mich.) or native PepG. Samples were removed for analysis after 6 and 18 h.

RT-PCR.

After incubation with PepG for 2 h, CD14⁺ cells (monocytes, macrophages, and a subset of granulocytes) were isolated from whole blood as previously described (34) by using Dynabeads M-450 CD14 (Dynal, Oslo, Norway). Isolation of mRNA was carried out by using oligo(dT)₂₅-coated Dynabeads (Dynal) and used directly for reverse transcription-PCR (RT-PCR). Semiquantitative analysis of IL-6 mRNA expression was performed by using a PCR cycler (Genius; Techne, Princeton, N.J.). Synthesis of cDNA was performed by RT directly on the mRNA attached to the oligo(dT)₂₅-beads by using GeneAmp Gold RNA PCR kit (PE Biosystems, Foster City, Calif.). Subsequently, the cDNA pool was analyzed by PCR for cDNA specific for IL-6 with specific primers (34).

ELISA assays.

Plasma was removed by centrifugation at 2,500 × g for 5 min and stored at −20°C in 96-well microplates (Greiner) for later analyses by an enzyme-linked immunosorbent assay (ELISA) specific for TNF-α, IL-6, IL-8, and IL-10 and in accordance with the manufacturers' instructions. (CLB [Amsterdam, The Netherlands] and R&D [Minneapolis, Minn.]).

Statistical evaluation.

Data are presented as mean values ± the standard errors of the mean (SEM). Differences between study groups were analyzed by using one-way analysis of variance (ANOVA) with Tukey's post hoc assessment. P values of <0.05 were considered significant.

RESULTS

Influence of S. aureus PepG and Cellosyl-treated PepG on indicators of MODS.

We first generated PepG substructures by selective hydrolysis of native PepG of S. aureus by the muramidase Cellosyl (breaking the glycan chain between the MurNAc and GlcNAc residues) (Fig. 1) and then examined the functional importance of intact PepG glycan chain on its ability to cause organ injury in vivo. We injected either intact S. aureus PepG molecules or Cellosyl-digested S. aureus PepG intravenously into anesthetized rats. Figure 2 shows that native S. aureus PepG caused moderate but significant increases in AST, ALT, γ-GT, and bilirubin (indicative of hepatic injury and/or dysfunction), as well as creatinine and urea (indicative of renal dysfunction), in the rat. In contrast, Cellosyl-digested S. aureus PepG failed to cause increased levels of these indicators of organ dysfunction or injury above baseline levels in serum. The values were significantly reduced compared to native S. aureus PepG, except for urea, where the tendency was clear, although not significant.

FIG. 2.

Effects of native and Cellosyl-treated S. aureus PepG on indicators of organ injury in the anesthetized rat. Anesthetized rats were subjected to surgical procedure, followed by injection of saline (sham, n = 14), PepG (10 mg/kg, n = 12), or Cellosyl-treated S. aureus PepG (10 mg/kg, n = 8). After 6 h, blood samples were drawn from the rats, and the concentrations of AST (A), ALT (B), γ-GT (C), bilirubin (D), urea (E), and creatinine (F) in serum were determined. ✽, A significant difference (P < 0.05) between sham animals and those treated with native PepG; ✽✽, a significant difference (P < 0.05) between animals treated with native PepG and animals treated with Cellosyl-digested PepG.

Impact of stem peptide composition and selective hydrolysis on cytokine release.

We then assessed the importance of amino acid composition of the PepG stem peptide for its inflammatory properties. PepG isolated from S. aureus, B. subtilis, or C. flaccumfaciens, which contain different amino acids in their stem peptides (PepG structure shown in Fig. 3), was added to whole human blood, and cytokine release was measured. As seen in Fig. 4, 1 μg of intact PepG/ml from all three bacteria evoked a substantial and similar release of both TNF-α (Fig. 4A), IL-6 (Fig. 4B), IL-8 (Fig. 4C), and IL-10 (Fig. 4D) in the blood.

FIG. 3.

General structure of PepG from S. aureus (A), B. subtilis (B), and C. flaccumfaciens (C), illustrating the differences in amino acid sequence in the stem peptides. D/L-Ala, d/l-alanine; D-Glu, d-glutamine; Gly, glycine; L-Hsr, l-homoserine; L-Lys, l-lysine; m-DAP, meso-diaminopimelic acid; D-Orn, d-ornithine.

FIG. 4.

Effects of native and hydrolyzed PepG on the release of TNF-α, IL-6, IL-8, and IL-10 in whole human blood. Whole human blood was incubated with intact PepG (1 μg/ml), hydrolyzed PepG (1 μg of PepG/ml with a hydrolase concentration of 15 ng/ml), or hydrolase controls (15 ng/ml) at 37°C with slow rotation. Plasma was isolated by centrifugation and analyzed for TNF-α after 6 h, as well as for IL-6, IL-8, and IL-10 after 18 h. The results are mean values ± the SEM of experiments on blood from 6 donors. ✽, A significant difference (P < 0.05) from values obtained by intact PepG, as analyzed by one-way ANOVA; ✽✽, a significant difference of P < 0.01. Cell, Cellosyl; Lyso, lysostaphin.

We next examined the ability of PepG partial structures generated by selective hydrolysis of native PepG of S. aureus, B. subtilis, or C. flaccumfaciens origin (Fig. 1) by Cellosyl and lysostaphin (breaking the pentaglycine bridge in S. aureus PepG) to induce cytokine production in human blood. Hydrolysis of PepG with the muramidase resulted in a profound and significant reduction in plasma values of TNF-α (Fig. 4A; 65 to 90% reduction), IL-6 (Fig. 4B; 85 to 95% reduction), IL-8 (Fig. 4C; 80 to 99% reduction), and IL-10 (Fig. 4D; 70 to 85% reduction) compared to the cytokine release induced by native PepG. S. aureus PepG digested with lysostaphin, elicited a more moderate (40 to 60%) reduction in cytokine release compared to untreated PepG (Fig. 4A to D).

Finally, we assessed the significance of PepG and its substructures to induce IL-6 mRNA accumulation in CD14⁺ cells after incubation in whole blood. As seen in Fig. 5, Cellosyl-treated S. aureus PepG induced profoundly less IL-6 mRNA than native PepG. The IL-6 mRNA accumulation in response to lysostaphin-treated PepG varied considerably between donors and was inconclusive.

FIG. 5.

Expression of IL-6 mRNA induced by S. aureus PepG and PepG partial structures. Whole human blood was incubated with intact PepG (1 μg/ml), hydrolyzed PepG (1 μg of PepG/ml with an hydrolase concentration of 15 ng/ml), or LPS (10 ng/ml) at 37°C with slow rotation. After 2 h,CD14⁺ cells and their mRNA were isolated by immunomagnetic separation, and RT-PCR was performed to assess IL-6 mRNA levels. The figure shows the DNA-gel depicting the relative amounts of IL-6 mRNA induced by the different bacterial products from one experiments (lower panel) and the means ± the SEM of data obtained from two independent experiments obtained by densitometry scanning (upper panel). cell, Cellosyl; lyso, lysostaphin.

Potential synergistic and/or antagonistic effects of Cellosyl-generated PepG partial structures.

Potential synergistic or antagonistic effects of the PepG partial structures generated by muramidase treatment on the release of TNF-α and IL-10 induced by LPS or native PepG was examined in human blood. As seen in Fig. 6, 10 μg of muramidase-digested S. aureus PepG/ml evoked 76 and 130% increases in TNF-α and IL-10 release induced by LPS, respectively (P < 0.05). In contrast, these PepG fragments slightly decreased the PepG-induced release of TNF-α and IL-10 (26 and 31%). Lower concentrations had no significant effects on cytokine release. This effect was not due to potential Cellosyl activity, since the inactivated enzyme did not influence PepG-induced cytokine release in control experiments (data not shown).

FIG. 6.

Synergistic and/or antagonistic effects of Cellosyl-generated PepG partial structures. Whole human blood was incubated with or without S. aureus PepG (1 μg/ml) or LPS (10 ng/ml) in addition to Cellosyl-digested S. aureus PepG (0.1, 1, or 10 μg/ml) at 37°C with slow rotation. Plasma was isolated by centrifugation and analyzed for TNF-α after 6 h and for IL-10 after 18 h of incubation. The results are mean values ± the SEM of experiments on blood from six donors. ✽, A significant difference (P < 0.05) compared to cytokine values induced by LPS alone, as measured by one-way ANOVA.

DISCUSSION

Several studies have suggested that the integrity of the PepG glycan backbone is important for its ability to initiate the formation of cytokines and other inflammatory mediators in cell cultures (22, 24, 26, 39, 50). This is the first study to report that disruption of the glycan backbone integrity of PepG abolishes its ability to cause organ injury in vivo, supporting the findings by Ganz et al. (19). This is clearly demonstrated by the fact that muramidase-digested PepG completely failed to increase levels of markers of hepatic injury and/or dysfunction (ALT, AST, γ-GT, and bilirubin) in serum or renal dysfunction (creatinine and urea) in serum in the rat. In contrast, intact PepG caused a significant increase in the levels of the these markers in serum, as recently reported (44). The close relationship between pathogenicity and the inflammatory capabilities of PepG was highlighted by the fact that cytokine production and mRNA accumulation in an ex vivo model of whole human blood was also crucially dependent on integrity of the PepG glycan backbone, as muramidase digestion strongly and significantly attenuated the responses.

Upon introduction of bacteria into mammalian tissue, phagocytes release muramidases as part of the host defense strategies to kill the invading microorganisms. Cellosyl, the muramidase we used to disintegrate the glycan backbone, is a bacterial muramidase produced by S. coelicolor. Human lysozyme possesses similar muramidase activities and functions as a bactericidal effector in vivo. Studies have indeed shown that bacterial killing is enhanced by the expression of lysozyme in the mouse lung (2) and that lysozyme-deficient mice are more susceptible to inflammation induced by gram-positive bacteria and its PepG (19). Lysozyme is one of the principal components of neutrophile granules (11) and a major secretory product of macrophages (20) and is believed to kill bacteria by hydrolyzing the bonds between the repetitive units of PepG, as described in Materials and Methods. However, there is some evidence that lysozyme possesses antibacterial and immunomodulating activities independent of muramidase activity (16). The potential role of muramidases as therapeutic agents has yet to be investigated in more detail. However, like many innate immune components, the role of lysozymes in sepsis might be that of a double-edged sword. While fighting off the infectious challenge by breaking down their cell wall, large amounts of PepG (native and substructures) and other inflammatory substances are released, which can contribute to the overwhelming systemic inflammation and development of MODS in the patient. Some reports indicate that highly active branched-chain PepG fragments can be released from pathogenic bacteria by muramidase digestion (26, 50). The aim of the present study, however, was not to chemically characterize single fragments that might not be of any functional relevance but to elucidate the in vivo significance of specific bonds within the the PepG molecule.

We also found that cutting the S. aureus PepG cross-bridge with its specific hydrolase (lysostaphin) moderately attenuates the cytokine response in whole blood. Lysostaphin is reported to be effective in treating S. aureus-induced experimental endophthalmitis, endocarditis, and burn wound infections in the rabbit (10, 12, 13, 23, 31) and might be of great therapeutic value in the treatment of methicillin-resistant S. aureus infections (10, 12). There exists one report of systemic use of lysostaphin as a therapeutic agent in a neutropenic patient, with uncertain conclusions (36).

Previous studies report that native PepG synergizes with LPS to release TNF-α in whole blood (46) and in the rat (49). We hypothesized that muramidase-digested PepG might also enhance the biological effects of LPS, despite the fact that it seemed to lack proinflammatory activity alone. We found that at high doses, the PepG substructure significantly increased the release of TNF-α and IL-10 induced by LPS. This is in line with several reports showing that muramyl dipeptide (MDP), the minimal bioactive subunit of PepG, enhances LPS-induced organ injury and cytokine release in a number of models (30, 32, 38). Wolfert et al. argue that MDP is in fact responsible for PepG's synergism with LPS by inducing TNF-α mRNA accumulation, which is processed into TNF-α protein by processes initiated by LPS (48). We demonstrate here that muramidase-digested PepG does not induce significant amounts of IL-6 mRNA in monocytes isolated from whole blood but nevertheless seems to synergize with LPS. Wolfert et al. report that MDP also synergizes with native PepG to cause TNF-α release (48), whereas there was no evidence of such interactions between muramidase-digested PepG and native PepG in the present study. Muramidase-digested PepG therefore seems to mimic the synergistic actions of native PepG rather than those of MDP. This may be related to structural differences between MDP and muramidase-treated PepG. Most notably, the latter still contains cross-linked forms, exerting different effects than those exerted by the monomeric dipeptide. Since our results indicated that the substructure shared receptor-binding properties with native PepG, we hypothesized that these fragments would potentially act as receptor antagonists for signaling induced by native PepG. Indeed, the PepG fragments seemed to slightly attenuate the response when administrated in high doses, but the significance of these data are uncertain.

Whereas S. aureus is a widely known pathogen and one of the bacteria most commonly isolated from septic patients (4, 42), B. subtilis and C. flaccumfaciens are not associated with disease in humans. Here, we have demonstrated for the first time that PepG from different bacterial species, with different stem peptide compositions, induce cytokine production in human blood with similar potency. It therefore seems plausible that stem peptides cannot account for the differences in pathogenicity between bacterial species and is not essential for the inflammatory properties of PepG. It might be that PepG, independent of its origin, mainly acts to amplify the response caused by other bacterial components, as suggested by others (24).

Based on these findings, we hypothesize that the structural integrity of the glycan chain of PepG is important for its pathogenicity in sepsis and MODS. Further research is needed to elucidate the in vivo roles of PepG partial structures and bacteriolytic enzymes in gram-positive infections.