Complement Activation Induced by Purified Neisseria meningitidis Lipopolysaccharide (LPS), Outer Membrane Vesicles, Whole Bacteria, and an LPS-Free Mutant

Abstract

Complement activation is closely associated with plasma endotoxin levels in patients with meningococcal infections. This study assessed complement activation induced by purified Neisseria meningitidis lipopolysaccharide (Nm-LPS), native outer membrane vesicles (nOMVs), LPS-depleted outer membrane vesicles (dOMVs), wild-type meningococci, and an LPS-free mutant (lpxA⁻) from the same strain (44/76) in whole blood anticoagulated with the recombinant hirudin analogue. Complement activation products (C1rs-C1 inhibitor complexes, C4d, C3bBbP, and terminal SC5b-9 complex) were measured by double-antibody EIAs. Nm-LPS was a weak complement activator. Complement activation increased with preparations containing nOMVs, dOMVs, and wild-type bacteria at constant LPS concentrations. With the same protein concentration, complement activation induced by nOMVs, dOMVs, and the LPS-free mutant was equal. The massive complement activation observed in patients with fulminant meningococcal septicemia is, presumably, an indirect effect of the massive endotoxemia. Outer membrane proteins may be more potent complement activators than meningococcal LPSs

The complement system is a key component of innate immunity and plays a central role in defense against invading bacteria. Complement activation can be triggered by antigen-antibody complexes, C-reactive protein, lipid A of lipopolysaccharides (LPSs), and other agents via the classical pathway, the mannan-binding lectin (MBL) pathway, or the alternative pathway, by bacteria or by LPSs (figure 1) [1–4]. Although most persons who contract meningococcal disease are complement sufficient [5, 6], deficiencies in the complement system are a risk factor for the development of meningococcal disease, which indicates that activation of the complement system is an essential defense against Neisseria meningitidis [7, 8]. Predominantly, deficiencies of the terminal complement complex system have been found. Patients with these deficiencies often have mild recurrent infections [9, 10]. Classical and alternative pathway deficiencies are less frequent; in these patients, the disease severity varies [8, 11, 12]. Genetically determined low levels of MBL in circulation increase the susceptibility to meningococcal disease in children [13]; however, in adolescents with meningococcal infections, the MBL levels did not differ from those of a control group of healthy blood donors [14]

Figure 1

Schematic presentation of complement system. Activation products analyzed in this study are in boxes. MASPs, mannan-binding lectin (MBL)–associated serine proteases; TCC, terminal SC5b-9 complement complex (see Materials and Methods for further description of activation products)

The outcome of systemic meningococcal disease (SMD) correlates closely with the plasma concentration of LPSs and with the degree of complement activation in meningococcal disease [15]. Patients with mild SMD (with low or undetectable LPS plasma levels) have low levels of complement activation, whereas those with septic shock and purpura are characterized by high circulating levels of LPSs and by excessive complement activation [15]. The close quantitative relationship among disease severity, plasma LPS levels, and complement activation has not been reported in other gram-negative infections. Both the classical and the alternative pathways appear to be activated in patients with meningococcal infections [1, 16]. The finding of very high levels of complement activation products and low levels of vitronectin or S protein and clusterin (SP-40,40/cytolysis inhibitor), which inhibit complement-mediated cytolysis, suggests that extensive complement-mediated damage to the host’s endothelial cells may occur during meningococcal disease [17]

During growth in vitro, meningococci release large amounts of LPS combined with outer membrane proteins (OMPs) in the form of outer membrane vesicles (OMVs), which are often referred to as “blebs” [18, 19]. In addition, meningococci may lyse, leading to release of outer membrane fragments. The existence of native outer membrane vesicles (nOMVs) and membrane fragments in vivo was documented by electron microscopy studies of plasma from a patient [20]. The contribution of bacterial OMVs in the pathophysiologic process is increasingly recognized [21], and we recently showed that meningococcal LPSs bound to nOMVs in an in vitro model induce release of proinflammatory cytokines in a dose-dependent manner [22]

Although complement activation in meningococcal disease has been extensively studied, the precise mechanisms of complement-mediated bacterial killing and the importance of different pathways remain to be clarified [1, 16]. We designed an experimental model by using whole blood anticoagulated with the recombinant hirudin analogue, lepirudin, to elucidate the relative contribution of the LPSs versus the OMPs and to study how different complement pathways contribute to meningococcal disease. Lepirudin is a highly specific thrombin inhibitor that does not interfere with complement activation, unlike other anticoagulants, and has no effect on cytokine release or on procoagulant activity [23]. We compared complement activation of different presentation forms of N. meningitidis: purified N. meingititidis LPS (Nm-LPS), nOMVs, LPS-depleted outer membrane vesicles (dOMVs), whole bacteria, and an LPS-free mutant, lpxA⁻, from the same strain. In addition, complement activation of purified Nm-LPS and nOMVs was compared with activation of LPS from Escherichia coli

Materials and Methods

Bacterial strain.N. meningitidis strain 44/76 was isolated from a culture of blood from a patient with fulminant septicemia [24]. Strain 44/76 is the production strain of the Norwegian group B OMV vaccine and an international reference strain. The strain belongs to the ET-5 clone [25] and has been serologically characterized with monoclonal antibodies (MAbs) as B:15:P1,7,16, with the immunotype L3,7,9. The strain also produces small amounts of shorter LPSs of immunotype L8, as determined by SDS-PAGE and anti-L8 MAb

LPS mutantAn LPS-deficient mutant was isolated after mutagenesis of the lpxA gene in strain 44/76. The enzyme LpxA is required for adding the O-linked 3-OH fatty acid to the UDP-N-acetylglucosamine, which is the first step in the lipid A biosynthesis [26]. In this way, the completely LPS-deficient mutant of 44/76 (lpxA⁻ 44/76) still expresses the immunodominant OMPs in normal amounts [26]. The OMV preparation was analyzed by SDS-PAGE and silver staining, and no LPS was detected

Meningococcal LPS (Nm-LPS).Nm-LPS was extracted from strain 44/76 and was purified, as described elsewhere [20, 27]. LPS content was quantitated by gas chromatography by use of 3-hydroxy lauric acid (3-OH-12:0) as a marker substance, as described elsewhere [20]. LPS also was analyzed by SDS-PAGE with silver staining [28] and with anti-LPS MAbs on immunoblots. No protein (<0.3%) was detected in the purified LPS [27]

nOMVsnOMVs from strain 44/76 and from the lpxA⁻ knockout mutant were prepared by extraction of the bacterial cells with LiCl/NaAc buffer at pH 5.8, as described elsewhere [29, 30]. The membrane pellets (nOMVs) were dissolved in distilled water, and the protein concentration was determined by a modified Lowry technique (DC-BioRAD) according to the manufacturer’s instructions. LPS content of nOMVs was measured by KDO (2-keto-3-deoxyoctonate) analysis [31] and for both nOMV and nOMVlpxA⁻ on silver-stained SDS-PAGE gels [28], with the purified Nm-LPS preparation used as a standard

dOMVsdOMVs were prepared by extraction with deoxycholate from the native N. meningitidis 44/76 and from the lpxA⁻ knockout mutant, as described elsewhere [32]. The dOMVs were similar to those used in the Norwegian group B vaccine. The LPS content of dOMVs was measured by gas chromatography [20], and the LPS content of dOMV and dOMVlpxA⁻ was determined by SDS-PAGE with silver staining [28], as reported above

Whole cell preparations.N. meningitidis 44/76 was grown overnight on chocolate agar plates in 5% CO₂ in air at 37°C and was harvested in Hanks’ balanced salt solution with 0.1% (wt/vol) bovine serum albumin. The number of colony-forming units was determined by plating 10-fold dilutions of the suspension. The colony-forming units were calculated to 6×10¹⁰. The cells were killed by heating for 30 min at 60°C and were stored at 4°C in PBS with 0.02% (wt/vol) sodium azide. LPS content was determined by SDS-PAGE gels with silver staining [28] and by use of KDO [31]. LPSs from E. coli strain O55:B5 were purchased from BioWhittaker (table 1). Protein concentration in the preparations was determined by a modified Lowry technique (DC-BioRAD) per the manufacturer’s instructions

Table 1

Lipopolysaccharide (LPS) and protein concentrations in different preparations.

Electron microscopyThe preparations from strain 44/76 were diluted in distilled water. Droplets (5 μL) of the solutions were applied to glow-discharged carbon film grids for 1 min and were negatively stained with 0.5% (wt/vol) phosphotungstic acid, pH 7.0, for another minute. The specimens were examined by a transmission electron microscope operated at 100 KeV (JEM 1010; Jeol)

Whole blood modelVenous blood from healthy volunteers was collected without stasis into LPS-free tubes and was anticoagulated by the recombinant hirudin analogue, lepirudin (Refludan; 50 mg/mL; final concentration 50 μg/mL). The blood was divided into 0.6-mL aliquots in sterile cryotubes (Nunc) and was incubated with purified LPS, nOMVs, dOMVs, whole wild-type bacteria, or lpxA⁻ mutant extracted in 2 different ways (nOMVs or dOMVs) from strain 44/76. The samples were incubated at 37°C and were gently shaken at 50 rpm. At time 0 and after 1 h of incubation, the appropriate tubes were removed, and further complement activation was stopped immediately by addition of EDTA to a final concentration of 15 mM. The tubes were centrifuged (1400 g, 10 min, 37°C), and plasma was stored in cryotubes at −70°C until analyzed. Each experiment was performed with a negative control (i.e., anticoagulated whole blood without any added form for LPS or OMPs) and was done ⩾3 times with different donors

Study designWe analyzed complement activation induced by different preparation forms of N. meningitidis 44/76 incubated in whole blood. The study was divided into 4 parts as follows: I, Comparison of complement activation at equal LPS concentrations (10 or 25 μg/ mL) of purified Nm-LPS, nOMVs, and whole bacteria; II, comparison of complement activation at equal LPS concentrations (1 or 10 μg/mL) of purified Nm-LPS, nOMVs, and dOMVs; III, comparison of complement activation at equal protein concentrations (25 μg/mL) of nOMVs, dOMVs, nOMVlpxA⁻, and dOMVlpxA⁻, and IV, comparison of complement activation at equal concentrations (1, 10, or 100 μg/mL) of purified LPSs from E. coli, Nm-LPS, and nOMVs

Complement activation productsThe following activation products were measured: C1rs-C1 inhibitor complexes (C1rs-C1inh; classical pathway), C4d (classical and lectin/MBL pathway), C3bBbP (alternative pathway), and TCC (terminal SC5b-9 complement complex pathway; figure 1). The analyses were performed with double-antibody EIAs. Except for C4d, the results are given in arbitrary units per milliliter related to a standard of activated serum defined to contain 1000 AU/mL. Results are given in AU because many of the complexes and activation products are heterogeneous with respect to molecular weights, and SI units are difficult to calculate. For C1rs-C1inh, this standard was made by activating serum with heat-aggregated IgG; for the other products, the standard was serum activated with zymosan. Procedures for production of standards and general assay performances are described elsewhere [33]

C1rs-C1inhWe measured the C1rs-C1inh by using the KOK-12 MAb (gift of C. E. Hack, Amsterdam), which is specific for a neoepitope in C1 inhibitor when it is in the complex with the protease. The assay is described in detail elsewhere [34]. In brief, plates were coated with the KOK-12 antibody, which reacted with plasma and control samples. The complex was detected by use of a cocktail of anti-C1r and anti-C1s antibodies

C4dSamples were measured by commercial EIA kit (Quidel) according to the manufacturer’s instructions. Results are given as micrograms per milliliter

C3bBbPThe alternative pathway convertase, composed of C3b, Bb, and properdin, was detected as follows: Capture antibody was mouse monoclonal anti–human properdin (no. 2; Quidel) diluted 1:1000. Test samples were diluted 1:25. Detection was made by polyclonal rabbit anti–human C3c (Behringwerke) diluted 1:1000 and peroxidase conjugated anti–rabbit immunoglobulin (Amersham International) diluted 1:1000. The rest of the assay was done as described elsewhere for complement activation assays [33]

TCCTCC was quantified by using the neoepitope-specific MAb aE11 as capture antibody. aE11 recognizes an epitope exposed in C9 when C9 is incorporated into the TCC complex. This epitope is not exposed in native C9. The assay has been described in detail elsewhere [33]

Statistical analysisAll data are expressed as mean±range. Statistical evaluations used Student’s paired t test for paired data. Significance was set at P<.05

Results

LPS quantitation by SDS-PAGE and KDO analysisOn silver-stained SDS gels, purified Nm-LPS from strain 44/76 showed one major LPS band and one minor, slightly faster moving band. The major band represents the L3,7,9 component and the minor band the L8 component of LPS, respectively. The dOMVs and nOMVs from 44/76 contained qualitatively the same LPS components as purified Nm-LPS, but the amount of LPS relative to protein (wt/wt) was about 5% in dOMVs and 50% in nOMVs. In the lpxA⁻ 44/76 knockout mutant, no LPS was detected. LPS in whole cells grown on agar plates did not express the lower (L8) band, only the L3,7,9 band. Their LPS content in the whole cell was 50 ng/10⁷ cfu (table 1). Table 1 also lists the protein concentration by preparation

Electron microscopyThe purified LPS was observed as long, thin, and smooth membranes (figure 2A). In contrast, the preparations containing protein showed a more or less vesicular shape (figure 2B and 2C). The nOMV with the highest LPS content in relation to protein appeared as membranes with vesicular to elongated shapes, whereas dOMV appeared mainly as smooth, round vesicles. No differences were observed between dOMVlpxA⁻ and nOMVlpxA⁻. These preparations contained no LPS and did not form vesicles easily. The few vesicles observed were irregular in shape and size (figure 2D)

Figure 2

Electron microscopy views of negatively stained preparations showing different forms of meningococcal lipopolysaccharide (LPS) and protein. A Purified Neisseria meningitidisLPS (Nm-LPS); B native outer membrane vesicles (nOMVs); C OMVs depleted of LPS (dOMVs); D vesicles isolated from LPS-free mutant (nOMVlpxA⁻). Magnification, ×125,000

Complement activation in relation to LPS concentration (parts I and II)The study was performed with 2 different LPS concentrations: part I with 10 and 25 μg/mL and part II with 1 and 10 μg/mL. Both experiments demonstrated a dose-dependent increase in complement activation. Data from the experiments with the highest LPS concentrations are shown in figures 3 and 4

Figure 3

Complement activation in whole blood model of equal lipopolysaccharide (LPS) concentrations (25 μg/mL) presented as purified LPS (Neisseria meningitidis [Nm] LPS), native outer membrane vesicles (nOMVs), and whole bacteria, compared with a negative control (n=3; mean±range). Complement activation products are indicated on vertical axes of panels A, B, C,and D. TCC, terminal SC5b-9 complement complex (see Materials and Methods for further description of activation products). *P<.05; **P<.01; ***P<.001 vs. negative control

Figure 4

Complement activation in whole blood model of equal lipopolysaccharide (LPS) concentrations (10 μg/mL) presented as purified LPS (Neisseria meningitidis [Nm] LPS), native outer membrane vesicles (nOMVs), and OMVs depleted of LPS (dOMVs) vs. negative control (n=3, mean±range). Complement activation products are indicated on vertical axes of panels A, B, C,and D. TCC, terminal SC5b-9 complement complex (see Materials and Methods for further description of activation products). *P<.05; **P<.01 vs. negative control

In both series of experiments, purified Nm-LPS was a weak complement activator (figures 3 and 4). No significant differences could be observed between the negative control and purified Nm-LPS at any level of the complement cascade. Complement was not activated in vitro even at high LPS concentrations (10 and 25 μg/mL) of purified Nm-LPS in these experiments. In contrast, whole bacteria with the same LPS concentration as purified Nm-LPS induced a substantial complement activation when compared with the negative control at all levels, except for initiation of the classical pathway (C1rs-C1inh; figure 3). The increase was statistically significant for C3bBbP (P=.02), C4d (P=.03), and TCC (P<.001; figure 3B–3D). Similarly, nOMVs induced a marked complement activation, which reached significance for C3bBbP (P=.04, parts I and II) and for TCC (P<.01 and .01, respectively, parts I and II) in both series of experiments (figure 3B and 3D and figure 4B and 4D). As for C4d, significance was reached only in part II (P=.006; figure 4C). Also, dOMVs induced a substantial complement activation that was statistically significant for C3bBbP (P<.01), C4d (P=.014), and TCC (P=.03; figure 4B–4D). No significant complement activation was seen for whole bacteria and nOMVs at C1rs-C1inh levels (figure 3A)

Complement activation as related to protein content in preparations (part III)No significant difference was seen in complement activation among the various preparations when adjusted for protein content of nOMVs, dOMVs, nOMVlpxA⁻ and dOMVlpxA for the activation products C1rs-C1inh, C3bBbP, or TCC (figure 5A5B and 5D). However, C4d increased significantly more by nOMVlpxA⁻ than by dOMV and dOMVlpxA⁻ (P=.002 and P=.04, respectively; figure 5C). Compared with the negative control, significant activation was seen for C3bBbP and TCC for all preparations, whereas this was not observed for C1rs-C1inh or C4d, as mentioned above (figure 5)

Figure 5

Complement activation in whole blood model of equal protein concentrations (25 μg/mL) presented as native outer membrane vesicles (nOMVs), OMVs depleted of lipopolysaccharide (LPS; dOMVs), nOMVs prepared from LPS-free mutant (nOMVlpxA⁻), and dOMVs prepared from LPS-free mutant (dOMVlpxA⁻), compared with negative control (n=3, mean±range). Complement activation products are indicated on vertical axes of panels A, B, C,and D. TCC, terminal SC5b-9 complement complex (see Materials and Methods for further description of activation products). *P<.05 vs. dOMV and dOMVlpxA⁻; **P<.01 vs. negative control

Comparison of complement activation of LPSs from E. coli, Nm-LPS, and nOMVs (part IV)The relationship between the LPS from E. coli and meningococcal LPS in 2 different preparation forms was studied by use of 3 LPS concentrations: 1, 10, and 100 μg/mL. Complement activation was analyzed by measuring TCC. As with the previous experiments, there were a dose-dependent increase in complement activation and reproducible results. LPS from E. coli triggered more complement activation than purified meningococcal LPS and Nm-LPS but less than nOMVs (figure 6)

Figure 6

Comparison of complement activation of the terminal SC5b-9 complement complex (TCC) pathway in whole blood model of equal lipopolysaccharide (LPS) concentrations (10 μg/mL) among LPS from Escherichia coli purified LPS (Neisseria meningitidis [Nm]LPS), and native outer membrane vesicles (nOMVs; n=3; mean±range)

Discussion

In this whole blood in vitro model, purified meningococcal LPS was a weak complement activator, compared with the more complex neisserial structures such as OMVs and whole meningococci. No significant difference in complement activation was detectable in plasma containing purified N. meningitidis LPS or a negative control plasma. Neisserial structures more complex than the LPS seem to activate complement more readily, as reflected by the substantial increase in activation by nOMVs, dOMVs, and whole bacteria compared with the same concentration of purified LPS

Previous in vitro studies have shown that LPS from other gram-negative bacteria may induce complement activation by direct reaction of lipid A with C1q independent of specific antibodies in the classical pathway and that the polysaccharide chain activates the alternative pathway directly [4, 35, 36]. These experiments indicate that the complement system has the capability to be activated immediately by intruding gram-negative bacteria, without the presence of preformed antibodies. This and other studies have shown that different preparations of LPS vary in their capacity to activate complement [3, 37]. In this study, LPS from E. coli activated complement to a greater extent than Nm-LPS but less than a preparation containing OMPs, such as nOMVs, dOMVs, or whole bacteria (figure 6). Recently, MBL was suggested to play an important role as a complement activator and, as such, in protecting humans against invasive meningococci. However, invasive strains of N. meningitidis are encapsulated and usually harbor sialylated lacto-N-neotetraose in their LPS [38]. Such strains bind MBL negligibly [39]. Thus, the role of MBL in invasive meningococcal disease remains elusive

By using purified meningococcal LPS, nOMVs, dOMVs, and inactivated whole meningococci and adjusting the preparations to the same level of LPS, we could compare the ability of LPS versus other components in the outer membrane to activate complement. Increasing complexity of the structures led to an increasing activation of complement. Because, dOMVs contained about 5% LPS relative to protein and nOMV contained about 50% LPS, equal LPS amounts of dOMVs and nOMVs implied nearly 10 times more bacterial proteins when used in the whole blood model. This observation suggested that components other than LPS, presumably OMPs, contributed considerably more than LPS to the complement activation

To further study the influence of LPSs relative to proteins, we made use of the recently developed N. meningitidis 44/76 mutant completely lacking LPS [26]. nOMVs and dOMVs from this knockout strain and preparations extracted by the same methods from the 44/76 wild-type strain were compared in a series of experiments. By using the same protein concentration, we examined the ability of the various LPS-positive and -negative preparations to activate complement. There was no significant difference between the positive and negative LPS OMVs, which supports the role of OMPs as complement activators (figure 5)

The importance of OMPs, particularly porins, from other bacterial species in complement activation has been studied. Complement-activated porins from the outer membrane of Salmonella typhimurium, S. minnesota, Klebsiella pneumoniae and Legionella pneumophila [40–43] all bind C1q and activate the classical pathway. Our results support the assumption that OMPs contribute significantly to complement activation. This may be particularly evident for bacteria with short-chained LPSs (ligo-oligosaccharides) as the principal glycolipid in the outer membrane, as is the case in N. meningitidis. In addition, to induce complement activation, porins may activate adhesion molecules, induce leukocyte transmigration through human endothelial cells in vitro [44], stimulate cytokine release [45], and increase the surface expression of CD86 and the T lymphocyte costimulatory ability of B lymphocytes [46]

Several studies have addressed the relative importance of the different complement pathways in meningococcal disease. The prevailing view is that the alternative pathway is quantitatively the most important [5, 16], but that the classical pathway also contributes to the activation [16, 47]. Our results support the role of the alternative pathway as a major contributor to the total complement activation and those of the classical pathways as discrete, in accordance with earlier findings from our group [16]. Our data, however, do not exclude the involvement of the MBL pathway, particularly since C4 was substantially activated. Furthermore, in addition to a genuine alternative pathway activation, both the classical and MBL activation pathways are amplified through the alternative pathway, contributing to an increase in the C3bBbP convertase

Patients with fulminant meningococcal septicemia have excessive complement activation. This activation, as evaluated by some of the same methods used in the present study, are quantitatively correlated with plasma levels of LPS [15]. The amount of LPS required to induce complement activation in vitro must be ⩾1000-fold higher and thus by far greater than the levels measured in patients [15]. This supports the assumption that other factors in addition to LPS are of etiologic importance in complement activation in patients with meningococcal disease. The endothelial lining of the vascular bed is altered through the action of various inflammatory mediators (i.e., tumor necrosis factor–α and interleukin-1β, -6, and -8), proteolytic enzymes released from marginated granulocytes, and bradykinin generated by massive activation of the plasma contact system [48–50]. Thus, an altered endothelium exposing subendothelial structures could be a major factor contributing to complement activation. In line with this is the observation that complement activation increases during the first 12–15 h in many patients with meningococcal septicemia before a gradual decline [15]. Thus, complement activation peaks later than most pro- and anti-inflammatory mediators detected in these patients [15]. This does not exclude, however, that the initial complement activation may contribute to induction of several of the other inflammatory arms involved in the pathophysiology of septic shock

The physical conformation of purified LPS may be of importance to induce complement activation. Galanos et al. [51] showed earlier that smooth LPS preparations with high molecular weight in aqueous solutions have greater anticomplementary activity than more soluble triethylamine salt forms. Wilson et al. [52] demonstrated that the alternative pathway is less dependent on the physical state of LPSs than the classical pathway, and Vukajlovich et al. [36] studied the structural requirements for activation of both classical and alternative pathways. In our electron microscopy studies, the long, smooth structures of purified Nm-LPS were found to have a physical configuration different from that of the other preparations and possibly might influence activation of complement (figure 2). Thus, the physical state of the purified Nm-LPS used in our experiments could possibly influence complement activation and could thus explain the low potency. Recent research clearly demonstrates that the presentation form of LPS is crucial for the immune response [53]. During meningococcal septicemia, it is unlikely that LPS exists as aggregates of pure LPS molecules. LPS detached from the bacterial outer membrane will rapidly bind to a variety of proteins, such as LPS-binding protein, lipoproteins, and lactoferrin

Our data suggest that purified Nm-LPS is a weak complement activator, compared with other neisserial membrane components. Complement activation is efficiently induced by such outer membranes even when there are only trace amounts of LPS. The massive complement activation observed in patients with fulminant meningococcal septicemia thus could result from the complex interplay of direct complement activation induced by different outer membrane constituents, such as OMPs in association with LPS, and complement activation secondary to endothelial damage and exposure to subendothelial matrix

Acknowledgments

We thank P. van der Ley (Rijksinstituut voor Volksgezondheit en Milieu, Bilthoven, The Netherlands) for generously providing an lpxA⁻ knockout mutant of 44/76 and Arne Høiby (Department of Bacteriology, National Institute of Public Health, Oslo) for the whole cell preparation of meningococci. We also thank Torill Tangen and Karin Bolstad (Department of Vaccinology, National Institute of Public Health, Oslo), for preparing the various OMV preparations, and Hilde Fure, Grethe Bergseth, and Dorte Christiansen, for excellent technical assistance

Presented in part: 6th Conference of the International Endotoxin Society, Paris, 24–27 August 2000 (abstract P34; J Endotox Res 2000; 6:123)

Financial support: Research Council of Norway; Norwegian Foundation for Health and Rehabilitation