Bacterial peptidoglycan-associated lipoprotein is released into the bloodstream in gram-negative sepsis and causes inflammation and death in mice.

Gram-negative bacterial sepsis commonly causes organ dysfunction and death in humans. Although circulating bacterial toxins trigger inflammation in sepsis, little is known about the composition of bacterial products released into the blood during sepsis or the contribution of various bacterial components to the pathogenesis of sepsis. We have shown that diverse Gram-negative bacteria release bacterial peptidoglycan-associated lipoprotein (PAL) into serum. The present studies explored release of PAL into the blood during sepsis and tested the hypothesis that PAL contributes to bacterial virulence and inflammation in Gram-negative sepsis. Released PAL was detected in the blood of 94% of mice following cecal ligation and puncture. Picomolar to nanomolar levels of PAL stimulated macrophages and splenocytes from lipopolysaccharide-hyporesponsive (C3H/HeJ) mice. Injection of PAL into C3H/HeJ mice stimulated production of serum cytokines and increased pulmonary and myocardial expression of inflammatory markers. PAL caused death in sensitized C3H/HeJ mice. Mutant Escherichia coli bacteria with reduced levels of PAL or truncated PAL were less virulent than wild-type bacteria, as indicated by higher survival rates and lower circulating levels of interleukin 6 and bacteria in a model of peritonitis in lipopolysaccharide-responsive mice. The studies suggest that PAL may be an important bacterial mediator of Gram-negative sepsis.

Gram-negative sepsis (GNS) 1 is a devastating consequence of Gram-negative bacterial infection that frequently causes severe respiratory failure and cardiovascular dysfunction and is a common cause of death in hospitalized patients (1)(2)(3). In sepsis, interactions between microorganisms and host cells trigger inflammatory responses that include release of soluble mediators such as cytokines and nitric oxide, expression of cell surface receptors and adhesion molecules, and recruitment of inflammatory cells into organs (4 -9). The similarity in these responses caused by microorganisms as diverse as Gram-negative and Gram-positive bacteria, fungi, and viruses suggests that multiple microbial components may stimulate common inflammatory signaling pathways and contribute to the pathogenesis of sepsis. The concept that multiple bacterial factors are active in sepsis is supported by studies indicating that multiple bacterial components, including lipopolysaccharide (LPS) (10,11), lipoteichoic acid and peptidoglycan (12,13), several outer membrane proteins (14 -18), flagellin (19), and several lipoproteins (20 -22), activate common inflammatory responses. These bacterial products signal through different Toll-like receptors (TLRs) (21)(22)(23)(24)(25)(26)(27)(28)(29) and nuclear factor B (30).
Although LPS from Gram-negative bacteria has been shown to circulate in GNS and to stimulate inflammation, little is known about the composition of bacterial products that are released into the blood during GNS or the contribution of different Gram-negative bacterial products to the pathogenesis of GNS. Previous studies indicated that antiserum raised to the rough mutant strain Escherichia coli J5 (J5 antiserum) improved survival in experimental GNS and in humans with GNS (31)(32)(33). Although J5 antiserum was believed to protect through anti-LPS antibodies, our prior studies indicate that J5 antiserum contains high titers of IgGs that bind bacterial peptidoglycan-associated lipoprotein (PAL), suggesting that anti-PAL IgG may also have contributed to the protection (34 -37). We have also found that PAL is released into human serum by heterologous Gram-negative bacteria in vitro and into the blood of burned rats with E. coli sepsis and that a proportion of PAL is released in fragments that also contain LPS and additional outer membrane proteins (36 -39).
PAL is the product of the ExcC gene located at 17 min on the E. coli map and is part of the system of cytoplasmic membrane, periplasmic and outer membrane proteins involved in maintaining cell wall integrity (40 -44). The precursor protein is composed of 173 amino acids. During posttranslational processing, a 21-amino acid N-terminal signal sequence is removed, and glyceride and fatty acid groups are added to the N-terminal cysteine (45)(46)(47). PAL mutants that either lack PAL or have abnormal PAL are hypersensitive to detergents (42). There is a high degree of homology in PAL among enteric and nonenteric Gram-negative bacteria (44, 48 -51).
The present studies were performed to explore the release of PAL into the blood in a cecal ligation and puncture (CLP) model of polymicrobial sepsis in mice and to test the hypotheses that PAL has inflammatory effects and may contribute to bacterial virulence during GNS. Released PAL was present in the blood of 94% of mice following CLP. Low concentrations of PAL stimulated splenocytes and macrophages in vitro and stimulated multiple responses in vivo, including induction of cytokines in the blood, transcription of proinflammatory pulmonary and myocardial genes, and death in sensitized mice. Survival was higher, and circulating levels of IL-6 and bacteria were lower in mice infected with PAL mutant versus wild-type bacteria in a peritonitis model of GNS. These data indicate that PAL is released by Gram-negative bacteria into the bloodstream during polymicrobial sepsis, that PAL induces inflammation, and that PAL may be an important factor in the development and severity of GNS induced by Gram-negative bacterial peritonitis.
Animals-The Institutional Animal Care and Use Committee at the Massachusetts General Hospital approved the animal studies. Release of PAL was studied in 20 -25-g C3H/HeOuJ mice (Jackson Laboratories, Bar Harbor, ME) and C3H/HeN mice (Charles River Laboratories, Wilmington, MA). Inflammatory and lethal effects of purified PAL were studied in 20 -25-g C3H/HeJ mice (Jackson Laboratories), which have a point mutation in the TLR4 gene that renders them hyporesponsive to the effects of LPS (23). Inflammatory and lethal effects of wild-type and PAL mutant bacteria were studied in LPS-responsive C3H/HeN mice. Rabbit IgGs were prepared using female 2-3-kg New Zealand White rabbits (ARI Breeding Laboratories, East Bridgewater, MA).
Antibodies-Mouse monoclonal anti-PAL IgG (mouse anti-PAL IgG) was produced as described previously (37,38). Rabbit polyclonal anti-PAL IgG (rabbit anti-PAL IgG) was prepared by immunizing rabbits with purified PAL in incomplete Freund's adjuvant. Immunoblot analysis revealed that anti-PAL-specific IgGs do not react with components of mouse serum.
PAL Release in the CLP Model of Sepsis-PAL release was studied in a CLP model of sepsis as described by Wichterman and others (52,53). Mice were anesthetized and given a 1-ml subcutaneous normal saline bolus, and a laparotomy was performed. The cecum was ligated, and both walls were punctured with a 19-gauge needle distal to the ligature. A small amount of fecal material was expressed through each puncture site, the bowel was returned to the abdomen, and the wound was closed. Bupivacaine was placed between the fascia and skin during closure, and buprenorphine was administered subcutaneously for analgesia. Control (sham) mice underwent an identical operation, but without ligation or puncture of the cecum.
At 20 h, blood was drawn via the tail artery, and the concentration of Gram-negative bacteria was determined by culturing serial dilutions of the blood on MacConkey agar plates. Mice were then given an intramuscular injection of ceftazidime (80 mg/kg), and blood was collected by cardiac puncture 2 h later. PAL release was assessed using previously described methods (36,38). Briefly, plasmas were prepared and filtersterilized to remove intact bacteria. Filtered plasmas (250 l) were diluted 4-fold and incubated overnight with mouse anti-PAL IgG that was covalently conjugated to magnetic beads (200 g IgG/ml). The beads were then washed, and PAL was eluted from the beads by heating in SDS (36). PAL was detected by immunoblotting half of the eluted material using rabbit anti-PAL IgG as the primary antibody. Immunoblots were developed using biotin-conjugated anti-rabbit IgG (Vectastain; Vector Laboratories, Burlingame, CA).
Purification of PAL-Total bacterial membranes were prepared from 20-liter overnight cultures of E. coli K12 CH202(pRC2) as described by Nikaido (54) and processed using differential extraction with SDS essentially as described by Mizuno (48). LPS and non-PAL proteins were then removed by size separation using continuous gel electrophoresis on a 14% SDS-PAGE tube gel, (Prep Cell model 491; Bio-Rad) and 1% SDS in 25 mM Tris, pH 8.3, and 192 mM glycine as the elution buffer.
This procedure was repeated twice to maximize removal of LPS and protein impurities. Trace amounts of a contaminating protein were then removed by anion-exchange chromatography using DEAE-Sephacel (Amersham Biosciences). PAL was eluted from the DEAE column with 80 mM NaCl in 10 mM Tris-HCl, pH 8.4, 2% Triton X-100, and 6 M urea and precipitated with 75% chilled acetone and 10 mM MgCl 2 . PAL was then resuspended in and dialyzed extensively against 50 mM sodium phosphate, pH 7.4 (carrier). A single protein band was present on gold stains of the final product and identified as PAL by immunoblotting. The protein concentration was determined using the BCA assay (Pierce). LPS levels were quantitated using the Limulus-ameobocyte lysate assay (Associates of Cape Cod, Falmouth, MA) (55). All PAL preparations contained Ͻ 400 pg LPS/g protein.
Mitogenic Assay-Splenocytes were prepared from C3H/HeJ mice and maintained in RPMI 1640 medium (Mediatech Cellgro) supplemented with L-glutamine, 5% heat-inactivated fetal calf serum, and gentamicin (50 g/ml) as described previously (56). Cells were placed in wells of 96-well microtiter plates (4 ϫ 10 5 cells/well), treated with PAL and controls in replicates of 6 wells/condition for 48 h, and then pulsed for 18 h with [ 3 H]thymidine (PerkinElmer Life Sciences; 1 Ci/well). The cells were harvested (PHD Cell Harvester; Cambridge Technology, Inc.), and [ 3 H]thymidine incorporation into DNA was assessed by liquid scintillation counting. In some experiments, cells were treated with PAL plus polymyxin B or with LPS alone to test the potential effects of the low levels of LPS that co-purified with PAL.
Macrophage Assays for Cytokine and Nitrite Production-Thioglycollate-elicited C3H/HeJ peritoneal macrophages were prepared as described previously (56). Four days after intraperitoneal injection of 3% thioglycollate, macrophages were obtained by peritoneal lavage with Dulbecco's modified Eagle's medium (Mediatech Cellgro) supplemented with glucose, L-glutamine, 10% heat-inactivated fetal calf serum, penicillin, and streptomycin. Cells were added to wells of 96-well plates at a concentration of 10 6 cells/ml and incubated for 2 h to allow macrophage adherence. Adherent cells were incubated with PAL in replicates of 6 wells/concentration of PAL. Levels of TNF-␣, IL-1␤, and IL-6 were quantitated in the supernatants by enzyme-linked immunosorbent assay after 18 h (Quantikine enzyme-linked immunosorbent assay kits; R&D Systems, Minneapolis, MN). Nitrite levels were measured in supernatants of macrophages treated with mouse IFN-␥ (10 units/ml; R&D Systems) and PAL or media using the Greiss reagent reaction and nitrite standards (56). In some experiments, cells were treated with PAL plus polymyxin B or with LPS alone to test the potential stimulatory effects of the low levels of LPS that co-purified with PAL.
PAL and Expression of Inflammatory Markers in Blood and Organs-C3H/HeJ mice received an intravenous injection of PAL (10 g) or an equivalent volume of carrier (50 mM sodium phosphate, pH 7.4). At specified time points, mice were euthanized, and blood and organs were collected. Blood was obtained by cardiac puncture up to 16 h after injection, and serum was prepared and analyzed for TNF-␣, IL-1␤, and IL-6 by enzyme-linked immunosorbent assay (n ϭ 5-6 mice/group).
Lungs and hearts (myocardia) were collected up to 4 h after injection and blotted free of blood. Pulmonary and myocardial cytokine expression and adhesion molecule expression were assessed using RNA blot hybridization and random primed cDNA templates (n Ն 4/group). RNA was isolated using homogenization in guanidine isothiocyanate and centrifugation on a CsCl cushion (57). The murine TNF-␣ probe was generated by Bruce Beutler (University of Texas Southwestern Medical Center, Dallas, TX). The IL-1␤ probe was made using polymerase chain reaction, sense (5Ј-atgaaagacggcacacccac-3Ј) and antisense primers (5Јcccacacgttgacagct-3Ј), and cDNA generated from reverse transcription using 2 g of RNA isolated from LPS-treated mouse lungs, Moloney murine leukemia virus reverse transcriptase, RNasin, and the antisense primer, as described previously (58). The ICAM probe was made using HinDIII and a murine expressed sequence tag fragment cloned into pBluescript SK (American Type Culture Collection). To confirm that equal quantities of RNA were analyzed, the blots were stripped of the labeled cDNA probes and then hybridized with a 15 M excess of an oligonucleotide corresponding to rat 18 S ribosomal RNA (5Ј-acggtatctgatcgtcttcgaacc-3Ј) that had been end-labeled with [␥-32 P]ATP and T4 polynucleotide kinase (59).
Lung MPO levels were measured 4 h after injection of PAL (n ϭ 7/group), essentially as described previously (60,61). The right lung was weighed, homogenized in 50 mM sodium phosphate (pH 7.4), and centrifuged (12,000 ϫ g). Lung homogenates were resuspended in 0.5 ml of hexadecyl-trimethyl-ammonium bromide (Sigma) in 50 mM sodium phosphate (pH 6.2), homogenized, sonicated, freeze-thawed three times, and sonicated again. The solution was centrifuged, and MPO levels were measured in the supernatants (62). Twenty l of the super-natants were mixed with 70 l of o-dianisidine dihydrochloride (0.357 mg/ml; Sigma) in Hanks' balanced salt solution and 60 l of 0.1% hydrogen peroxide in 50 mM sodium phosphate (pH 6.2), and absorption was measured at a wavelength of 450 nm (reference, 620 nm) after 15 min. MPO concentrations (units/g lung) were determined using human MPO standards (Sigma).
PAL-induced Mortality in D-Galactosamine-sensitized Mice-Lethal effects of PAL were studied in D-galactosamine-sensitized mice (n ϭ 7-8 mice/group) (63). D-Galactosamine (800 mg/kg) was administered to mice by intraperitoneal injection 5 min before intravenous injection of 1, 10, and 100 g of PAL or an equivalent volume of carrier. Survival was assessed up to 96 h after treatment.
Inflammatory and Lethal Effects of PAL Mutants in E. coli Peritonitis and Sepsis-E. coli K12 wild-type (p400 and JC1129) and PAL mutant (CH202 and JC2721) strains were grown to mid-log phase and washed in sterile normal saline, and the optical density at 550 nm was adjusted to ϳ0.8 using normal saline. Bacteria were then pelleted and resuspended in normal saline at one-twentieth the original volume, and quantitative cultures were performed to ensure that mice received comparable amounts of wild-type and PAL mutant bacteria. Live bacteria were administered to C3H/HeN mice by intraperitoneal injection. Bacterial doses were 3.7 ϫ 10 8 (Ϯ 1.9 ϫ 10 7 ) and 3.4 ϫ 10 8 (Ϯ 7.3 ϫ 10 7 ) CFU for the PAL-deficient strain and its progenitor wild-type strain, respectively, and 1.7 ϫ 10 8 (Ϯ 5.9 ϫ 10 6 ) and 1.7 ϫ 10 8 (Ϯ 7.4 ϫ 10 6 ) CFU for the PAL nonsense strain and its progenitor wild-type strain, respectively. Blood was collected via the tail artery at 20 h, and IL-6 levels were measured, and quantitative blood cultures were performed. Survival was followed for 72 h. Data were compiled from three separate experiments for each PAL mutant strain (n ϭ 18/group) and paired wild-type control (n ϭ 12-18/group). The data from the two wild-type E. coli K12 strains were combined into one group for the figures and for statistical analysis because mortality, IL-6 levels, and degree of bacteremia were not statistically different between these groups (final wildtype n ϭ 30).
Statistical Analysis-Representative data from at least three experiments are presented in the figures. Error bars in the figures represent S.D. A two-way Fisher's exact test was used to analyze PAL release in the CLP model. One-way ANOVA followed by Dunnett's post hoc test was used to assess macrophage and splenocyte responses to purified PAL and plasma IL-6 levels in mice infected with PAL mutant and wild-type E. coli K12 strains. Two-way ANOVA followed by Bonferroni's post hoc test was used to assess serum cytokine levels over time in mice treated with purified PAL or carrier. Lung MPO data and quantitative blood cultures from the PAL mutant peritonitis experiments were analyzed by a two-tailed Mann-Whitney test. Survival studies were analyzed by the log-rank test, and the LD 50 of purified PAL was calculated at 96 h according to the method of Reed and Muench (64). p values Ͻ 0.05 were considered statistically significant. In the figures, ‫,ء‬ ‫,ءء‬ and ‫ءءء‬ signify p values of Ͻ0.05, Ͻ0.01, and Ͻ0.001, respectively.

PAL Circulates in Mice after CLP-
The CLP model approximates human sepsis resulting from intestinal disruption (52,53). PAL was affinity-purified from sterile-filtered plasmas collected from CLP and sham mice using mouse anti-PAL IgG and detected by immunoblotting with rabbit anti-PAL IgG. A band with a molecular mass of 18 -19 kDa, which is identical to that of full-length PAL, was detected in 15 of 16 (94%) of the CLP mice but was not detected in any of the 15 sham mice (p Ͻ 0.001). A representative immunoblot from one of three experiments is shown in Fig. 1. Using quantitative cultures, Gramnegative bacteria were detected in the blood of 9 of 16 (56%) CLP mice immediately before administration of ceftazidime (median, 1.7 ϫ 10 3 CFU/ml; range, 0 -3.8 ϫ 10 5 CFU/ml), whereas no bacteria were detected in the blood of the 15 sham mice. Follow-up blood cultures performed on a limited number of CLP mice (n ϭ 5) 2 h after antibiotic administration confirmed the anticipated reduction in CFU/ml (data not shown).
PAL Stimulates Splenocyte Proliferation-Proliferation assays were performed to assess the effects of PAL on lymphocyte function. PAL caused a dose-dependent increase in splenocyte proliferation (p Ͻ 0.001, Fig. 2) at PAL concentrations of Ն80 ng/ml. Proliferation was not stimulated by LPS at concentrations that were 10-fold in excess of the LPS in the purified PAL ( Fig. 2), and co-incubation of PAL with polymyxin B (5 g/ml) did not attenuate the proliferative response to PAL (data not shown).
PAL Increases the Production of TNF-␣ and IL-6 but not IL-1␤ by Cultured Peritoneal Macrophages-PAL caused a dosedependent increase in production of TNF-␣ and IL-6 by macrophages at PAL concentrations Ն 5 ng/ml (280 pM) (p Ͻ 0.001, Fig. 3, A and B). IL-1␤ was not detected when macrophages were treated with up to 5 g/ml PAL (data not shown). TNF-␣ and IL-6 were not detected when macrophages were treated with LPS at concentrations up to 100 ng/ml (Ն 50 -fold in excess of the LPS in the purified PAL) and polymyxin B (5 g/ml) did not inhibit macrophage responses to PAL (data not shown).

PAL Induces Nitrite Production by Cultured Peritoneal Macrophages in the Presence of IFN-␥-Expression of inducible
nitric oxide synthase is up-regulated during sepsis. This may increase nitric oxide production and contribute to vasodilatation and shock (65,66). Accumulation of nitrite, a marker of nitric oxide production, was measured in macrophage supernatants. Although neither PAL (up to 5 g/ml) nor IFN-␥ (10 units/ml) alone increased nitrite levels, nitrite levels were increased by the combination of Ն500 pg/ml PAL and IFN-␥ (p Ͻ 0.001; Fig. 3C). The requirement for IFN-␥ for PAL-induced nitrite production observed in these studies is consistent with results obtained by others (13,19).
PAL Increases Expression of TNF-␣, IL-1␤, and ICAM-1 mRNA in Lungs and Hearts-Pulmonary and myocardial tis- 1. PAL is released into the blood in a CLP model. Plasma was obtained from mice 24 h after CLP (lanes 1-3) or sham (lanes 4 -6) operation and filtered to remove intact bacteria. The released PAL was affinity-purified from plasma filtrates using mouse anti-PAL IgG and detected by immunoblotting with rabbit anti-PAL IgG. A representative immunoblot from one of three experiments is shown. Each well represents one-half of the material that was affinity-purified from 250 l of plasma. Purified PAL (16 ng) is shown at the right. sue were analyzed by RNA blot hybridization 1, 2, and 4 h after administration of PAL. TNF-␣, IL-1␤, and ICAM-1 mRNA levels were all increased in the lungs and hearts of mice treated with PAL but not with carrier. Expression was increased by 1 h and remained above baseline through 4 h (Fig. 5).
PAL Increases MPO Levels in Lungs-Levels of MPO, a neutrophil granule protein, were quantitated to assess infiltration and/or activation of neutrophils in the lung. MPO levels were nearly 6-fold higher in PAL-treated mice (mean, 39.9 units/g; range, 7.4 -111.7 units/g) than in carrier-treated mice (mean, 7.4 units/g; range, 2.1-11.4 units/g) (p ϭ 0.007).
PAL Causes Death in D-Galactosamine-sensitized Mice-The number of mice surviving was determined up to 96 h after treatment of D-galactosamine-sensitized mice with purified PAL. PAL caused death at all doses tested (p ϭ 0.0001; Fig. 6). The LD 50 of PAL was 3 g/mouse (167 pmol).
PAL Mutants Are Less Virulent than Wild-type Controls during E. coli Peritonitis-Two different PAL mutants were used for these experiments. The PAL-deficient strain contains markedly reduced levels of PAL by immunoblotting with anti-PAL IgG (37). The PAL nonsense strain contains a mutation in the PAL gene with a stop codon that results in production of a truncated protein (42). Overall survival was markedly increased in the PAL mutant groups versus the wild-type controls FIG. 5. PAL modulates cytokine and adhesion molecule mRNA levels in lungs and hearts (myocardia) of C3H/HeJ mice. RNA was extracted from the lungs and myocardia 1, 2, and 4 h after intravenous injection with either carrier or PAL (10 g), fractionated using formaldehyde-agarose gel electrophoresis, and transferred to charged membranes. After the membranes were exposed to 32 P-labeled restriction fragments of specific cDNAs that encode murine TNF-␣, IL-1␤, and ICAM-1, they were washed at high stringency and subjected to autoradiography. After stripping the membranes with formamide, rehybridization with an excess of 32 P-end-labeled 18 S oligonucleotide and autoradiography confirmed equal RNA loading. A representative RNA blot from one of three experiments is shown. (Fig. 7A). Survival was 7% in the wild-type group as compared with 33% and 100% in the PAL-deficient and PAL nonsense groups, respectively (p Ͻ 0.001). Plasma IL-6 concentrations (Fig.  7B) and levels of bacteremia (Fig. 7C) were reduced in the PAL mutant groups as compared with the wild-type group (p Ͻ 0.001). DISCUSSION We have demonstrated that PAL is released into the blood by Gram-negative bacteria during experimental GNS and that PAL has potent inflammatory effects and contributes to the virulence of E. coli K12 bacteria during peritonitis and sepsis.
Previously, we found that PAL is released into the blood in an infected wound model of monomicrobial sepsis in rats (38,39). The present studies indicate that PAL is also released into the blood in a polymicrobial model of sepsis caused by intestinal disruption. Although these studies were not specifically designed to quantitate PAL release, a rough estimate of the concentration of PAL circulating in the plasma can be made based on the immunoblots for PAL. In some CLP samples, the intensity of staining for PAL was similar to that for a 16-ng standard of purified PAL, as can be seen in Fig. 1 (lanes 1 and  2). Based on the quantity of affinity-purified CLP sample loaded per well, PAL levels in CLP plasma are at least 128 ng/ml. This concentration is well within the range that stimulates splenocyte and macrophage responses in our studies. Although we have been unable to find reports in the literature that other bacterial membrane proteins are released into the bloodstream at concentrations that stimulate inflammatory responses, we have previously detected released murein lipoprotein in the blood of septic rats (39). Prior studies have focused substantially on LPS. It seems likely that multiple other bacterial products, including proteins and lipids, are released into the blood during infection and contribute to the inflammation in sepsis.
Gram-negative bacteria have been reported to contain between 10 4 and 1.2 ϫ 10 5 molecules of PAL/cell, depending on the genus and species of the bacteria (47,49). The highest level of bacteremia in the present studies was 3.8 ϫ 10 5 CFU/ml, which should contain 3.8 ϫ 10 9 to 4.5 ϫ 10 10 molecules/ml or 0.1-1.3 ng PAL/ml, depending on the bacteria involved in this polymicrobial infection. Based on these rough estimates, released PAL levels were 100 -1000-fold higher than predicted by the culture data. This may be due to shedding of PAL by live bacteria during GNS and/or slow clearance of released PAL and raises the possibility that released PAL may remain in the circulation at levels that stimulate inflammatory responses after sterilization of the infection.
There are several possible explanations for the observation that PAL was detected in 94% of the CLP mice, whereas Gram-negative bacteria were detected in only 56% of the CLP mice. First, some mice may have been bacteremic, but at a lower level than the Ն100 CFU/ml required for detection. Second, PAL may be released at the site of infection and then diffuse into the blood. Third, PAL may be released into the blood and accumulate during transient bacteremic events that were missed by our single culture. The high rate of detection of released PAL suggests that PAL may be a sensitive marker for GNS.
PAL potently stimulates multiple immune effector cells, including macrophages, lymphocytes, and neutrophils. Picomolar levels of PAL stimulated production of TNF-␣, IL-6, and nitrite by peritoneal macrophages. Injection of PAL into mice stimulated responses that are characteristic of early sepsis, including induction of serum cytokines, increased pulmonary and myocardial expression of cytokine and adhesion molecule mRNA, and accumulation and/or activation of neutrophils within the lungs.
Respiratory failure, myocardial dysfunction, and vasodilatation frequently occur during sepsis. Inflammatory responses Equivalent doses (ϳ10 8 bacteria as described under "Experimental Procedures") of wild-type (Ⅺ) and PAL mutant bacteria that contained reduced amounts of PAL (f) or contained truncated PAL on the basis of a nonsense mutation in the PAL gene (q) were injected intraperitoneally into C3H/HeN mice at t ϭ 0 h. Survival was assessed for 72 h (A). Plasma IL-6 levels (B) and blood bacteria levels (C) were measured at t ϭ 20 h. The horizontal bars represent the median for each group. within organs include increased expression of cytokines and vascular adhesion molecules such as ICAM and up-regulation of inducible nitric oxide synthase leading to increased nitric oxide production (2,6,9,67). The increased myocardial and pulmonary expression of these genes, the elevated pulmonary MPO levels, and the increased macrophage production of nitrite induced by PAL suggest that PAL may contribute to the acute respiratory failure, myocardial dysfunction, and shock observed in GNS.
The notion that PAL may be important in GNS is further supported by the reduced mortality, lower plasma IL-6 levels, and lower levels of bacteremia induced by bacterial mutants defective in PAL as compared with wild-type strains of bacteria in LPS-responsive mice. There are several potential mechanisms for the decreased virulence of the PAL mutant bacteria. Abnormalities of this structural cell wall component may render the bacteria more sensitive to host defenses, may influence the quantity of other cell wall constituents that may also be important in the pathogenesis of GNS, and/or may influence the pattern of release of bacterial components such as LPS during infection. In addition, PAL may contribute more directly to the pathogenesis of GNS by triggering inflammation either on its own or in conjunction with other bacterial or host mediators or by facilitating movement of the bacteria from the peritoneum into the bloodstream.
Additional studies will be required to define the cellular mechanisms of PAL-induced inflammation. The broad inflammatory responses elicited by PAL suggest a nuclear factor B-dependent pathway (30). TLR2 is a candidate receptor for PAL because it is a receptor for lipoproteins from other bacteria (24,27,28). TLR4 is not required for the responses observed in these studies because C3H/HeJ mice lack functional TLR4 (23).
Lipoproteins from spirochetes, Mycoplasma, and Mycobacteria and murein lipoprotein from Gram-negative bacteria have been shown to activate nuclear factor B-mediated inflammatory responses (20 -30). However, unlike PAL, most of these lipoproteins are derived from bacteria that do not cause Gramnegative sepsis, and they have not been reported to circulate in the blood separately from bacteria during active infection.
The presence of PAL in the circulation at concentrations that stimulate inflammatory cells, the potent inflammatory and toxic effects of PAL, and the reduced injury induced by PAL mutant bacteria as compared with wild-type control bacteria in a model of infection all support the hypothesis that PAL may play an important role in GNS. If PAL is an important mediator of GNS, the high degree of homology between PAL from diverse Gram-negative bacteria suggests that it may be a suitable bacterial target for future antisepsis therapies.