INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Infections caused by Neisseria meningitidis remain an important cause of mortality and morbidity worldwide (14). It is the organism responsible for the majority of childhood cases of bacterial meningitis in the United Kingdom, and in patients presenting with severe shock, mortality may be as high as 50%. Although prompt recognition, early treatment and intensive care has reduced this figure in recent years (18), survivors may have extensive tissue injury, sometimes requiring amputation and/or skin grafting.

Capillary leak and intravascular thrombosis are serious consequences of meningococcal sepsis and are indicative of widespread vascular endothelial injury (23). Histological studies of meningococcal disease show that cutaneous lesions contain large numbers of organisms that are associated with the vascular endothelium (11). Recent studies have also shown that meningococci have the capacity to bind endothelial cells in a receptor-ligand-specific fashion (33, 34) and indicate that bacterium-endothelium contact may itself be critical in mediating the vascular injury seen in this disease (29). There is evidence that meningococci, both alone and in the presence of neutrophils, can lead to endothelium damage (20, 32). However, there is still very little information on how meningococci may themselves modulate the influx of neutrophils into inflammatory sites.

Expression of adhesion molecules by the vascular endothelium is a critical step in the inflammatory response. Leukocyte adhesion occurs through a complex and multistep process involving initial tethering and then rolling of leukocytes by low-avidity interactions with mainly the selectin family of cell adhesion molecules (e.g., CD62E/E-selectin). This is followed by firmer adhesion, which is mediated largely by higher-affinity interactions involving the members of the immunoglobulin Ig superfamily (e.g., ICAM-1 and VCAM-1) on endothelial cells (30). After firm adhesion, transendothelial and subendothelial migration may occur, a process also involving leukocyte integrins and complex cross talk among leukocytes, the endothelium, cytokines, and chemokines. The initial inflammatory stimulus to this activation cascade is critical since it determines which leukocytes will participate in the subsequent inflammatory response (4). The pattern of endothelial activation seen in response to the proinflammatory cytokines tumor necrosis factor alpha, interleukin-1, and CD40 (17) and bacterial endotoxin has been described (5, 39). There is very limited information on the response of endothelial adhesion molecules to live meningococci.

We have previously shown that encapsulation and lipooligosaccharide (LOS) structure influence the host inflammatory response to N. meningitidis (20). In this study, we have used isogenic mutants of N. meningitidis B1940 to investigate the relationship between bacterial structure and the expression of adhesion molecules on cultured human endothelial cells. We have also explored the effect of human recombinant bacterial/permeability increasing protein (rBPI₂₁) (2) to modulate endothelial activation by these organisms.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Bacterial strains. The parent organism, N. meningitidis B1940, and three isogenic mutants derived from it have been described previously (7). The capsule-deficient (siaD), mutant of B1940 was constructed by using insertional inactivation of the polysialyltransferase gene. Inactivation of the galE gene by replacement of the cpsD region with a chloramphenicol resistance marker produces a capsulated mutant that expresses a truncated LOS that cannot be sialylated. In the cps mutant, the whole cps gene complex is missing, and so it has both defective capsule expression and a truncated LOS. A fourth mutant, the lst mutant of B1940, has a deleted -2,3-sialyltransferase gene and cannot sialylate terminal lacto-N-neotetraose of its LOS (36). The parent bacterium, B1940, and its derived mutants express pili and both Opa and Opc as indicated by electron microscopy and immunoreactivity with specific monoclonal antibodies (20).

Materials. rBPI₂₁, a recombinant, modified amino-terminal fragment of bactericidal/permeability increasing factor, was a kind gift from XOMA (US) LLC, Berkeley, Calif. Escherichia coli lipopolysaccharide (LPS) (protein content, <1%), serotype O111:B4, was purchased from Sigma, Poole, United Kingdom. LOS from N. meningitidis serogroup B (strain 44/76) was prepared as previously described (1). Briefly, LPS was extracted by hot aqueous phenol extraction, ultracentrifugation, gel filtration, and cold-ethanol-NaCl precipitation (31). The final product contained <0.3% protein and was without detectable nucleic acids.

Bacterial culture. All the above strains were grown on gonococcal agar (Difco) supplemented with Vitox (Oxoid) and cultured in 6% CO₂ in air at 36°C. In all experiments, the bacteria used were subcultured at least once and were used after 18 h. Suspensions of bacteria were prepared in RPMI 1640 medium with no phenol red (Gibco, Paisley, United Kingdom), and their optical density was measured at 540 nm. Bacterial viability counts were measured by a modification of Miles and Misra technique (24). In some experiments, nonviable bacteria were used. These were killed either with 0.5% paraformaldehyde or by heating at 56°C for 30 min.

Endothelial-cell culture. Human umbilical vein endothelial cells (HUVEC) were obtained by collagenase type 2 (Gibco) digestion as described previously with some modifications (21). Cells in primary culture were grown in MCDB 131 medium (Gibco) supplemented with 2 mM L-glutamine, penicillin, streptomycin, and 20% heat-inactivated fetal calf serum (Gibco) in 25-cm² tissue culture flasks (Becton Dickinson, Oxford, United Kingdom). The cells were then passaged into 24-well plates, previously treated with endothelial attachment factor, by using trypsin-EDTA (Sigma). The cells were grown to confluence and then washed thoroughly, 24 to 48 h prior to experiments, in antibiotic-free RPMI 1640 medium supplemented with 25 mM HEPES buffer, 2 mM L-glutamine (Gibco), and 20% heat-inactivated fetal calf serum (Gibco).

Incubation of N. meningitidis with HUVEC. Initial experiments were conducted with either heat-inactivated or 0.5% paraformaldehyde-fixed bacteria. In subsequent experiments, HUVEC were stimulated with endotoxin or live meningococci and then incubated for up to 24 h at 37°C. Each well of confluent HUVEC contained approximately 10⁵ cells, so that the ratio of bacteria to HUVEC ranged from 1:0.01 to 1:100. In some experiments, HUVEC were exposed to bacteria or LPS before being washed in fresh RPMI medium containing 20% fetal calf serum, and incubated for a further 5 h. The cells were then removed from culture plates by incubation in Puck's A saline followed by gentle mechanical scraping. Adhesion molecule expression was detected by incubation with mouse monoclonal antibodies to human ICAM-1, CD62E, and VCAM-1 (Serotec, Oxford, United Kingdom) followed by goat anti-mouse F(ab')₂-phycoerythrin conjugate (Dako, High Wycombe, United Kingdom). Nonspecific binding of antibodies was controlled for by inclusion of an irrelevant isotype-matched control (Dako). Samples were washed, resuspended in Cellfix (Becton Dickinson), and analyzed by flow cytometry (FACScalibur; Becton Dickinson) with Cell Quest software (Becton Dickinson). The endothelial-cell population was identified by its forward-scatter and side-scatter position and by expression of CD31 (Serotec). For each sample, 5,000 events were collected within the Endothelial gate.

Statistics. All experiments were done at least three times on independent primary cell cultures from different sources. Where statistical analysis is shown, the results are expressed as mean median fluorescence intensity and standard error of the mean. A comparison of ranking order of cellular adhesion molecule expression by parent and cpsD, siaD, and cps mutant strains was analyzed by a Kruskall-Wallis one-way analysis of variance.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

HUVEC adhesion molecule expression induced by N. meningitidis B1940. Incubation of endothelial cells with the parent organism induced the expression of CD62E, VCAM-1, and ICAM-1. With 10⁶ organisms per ml, CD62E expression was first detected at 2 h, reached maximal levels at 5 h, and diminished markedly by 24 h. VCAM-1 expression was not detected until 4 h, and it reached maximal levels between 12 and 18 h. Similar levels were still seen at 24 h. ICAM-1 was expressed on resting HUVEC, but the levels were seen to increase by 4 h and were still rising by 24 h. Similar levels and kinetics of these adhesion molecules were seen with 10 ng of E. coli LPS per ml.

Influence of capsulation and LOS structure on endothelial cell adhesion molecule expression. To examine the influence of capsulation and LOS structure on HUVEC adhesion molecule expression, HUVEC were exposed to the B1940 isogenic mutant organisms. When the endothelial cells were incubated with bacteria for 5 h, the magnitude of CD62E, ICAM-1, and VCAM-1 expression induced by the mutants was similar to that seen with the parent (capsulated, sialylated LOS) organism. However, when HUVEC were exposed to bacteria for shorter periods, differences in adhesion molecule expression became apparent. This was most marked after only 15 min of exposure. Under these conditions, the siaD (unencapsulated, sialylated LOS) and cps (unencapsulated, truncated LOS, nonsialylated) mutants consistently induced higher levels of all three adhesion molecules than did the parent. This was particularly marked for CD62E and VCAM-1 (Fig. 1). Expression induced by the cpsD (capsulated, truncated LOS, nonsialylated) organism was always higher than that induced by the parent but not usually to the levels observed with the siaD and cps mutants (Fig. 1). Interestingly, in separate experiments comparing the parent and the cpsD and lst (capsulated, nonsialylated LOS) mutants, the level of cell adhesion molecules seen with the lst organisms was between those seen with the parent and the cpsD mutant. Similar results were obtained when live meningococci were replaced with bacteria that were either heat inactivated or fixed in 0.5% paraformaldehyde (results not shown).

View larger version (29K): [in this window] [in a new window]   FIG. 1.   Expression of ICAM-1 (A), E-selectin (B), and VCAM-1 (C) on HUVEC at 5 h after incubation with the N. meningitidis parent strain and cps, siaD, and cpsD mutants at 106 organisms per ml. HUVEC were washed thoroughly in fresh medium after 15 min of incubation with organisms. E. coli LPS was added at 10 ng per ml, and the cells were left unwashed. Results here are expressed as means and standard errors of the mean. #, P = 0.03; *, P = 0.015; +, P = 0.045.

View larger version (21K): [in this window] [in a new window]   FIG. 2.   Differential induction of CD62E on HUVEC by the N. meningitidis B1940 parent and the cpsD and siaD mutants occurs over a range of bacterial concentrations. Bacteria were added at 108 to 104 organisms per ml. After 15 min of incubation, HUVEC were washed thoroughly in fresh medium and incubated for a further 5 h. CD62E expression was determined at 5 h by flow cytometry. This is representative of at least three experiments yielding similar results.

E. coli LPS and meningococcal LOS cause a different pattern of adhesion molecule expression to N. meningitidis B1940 and isogenic mutants. When HUVEC were incubated with purified E. coli LPS or meningococcal LOS, the profiles of CD62E, ICAM-1, and VCAM-1 expression observed by flow cytometry were similar (Fig. 3). In comparison, N. meningitidis induced greater expression of CD62E than did purified endotoxin from either bacterial species, even when very high endotoxin doses (100 ng/ml) were used. Endotoxin from either source was at least as effective at up-regulating ICAM-1 or VCAM-1 as were the bacteria (Fig. 3).

View larger version (38K): [in this window] [in a new window]   FIG. 3.   Flow cytometric analysis of expression of CD62E, ICAM-1, and VCAM-1 on HUVEC following 5 h of incubation with 100 ng of E. coli O111:B4 LPS, 100 ng of meningococcal LOS, and 107 N. meningitidis B1940 parent and siaD mutant organisms per ml. Shaded areas represent CD62E, ICAM-1, and VCAM-1 staining in response to stimuli; the solid line represents cell adhesion molecule staining in unstimulated cells; and the dotted line represents staining with an irrelevant mouse immunoglobulin G1 antibody. Numbers on each plot are median fluorescent intensities and percent positive events for stimulated cells. Data presented here is representative of three experiments yielding similar results.

View larger version (19K): [in this window] [in a new window]   FIG. 4.   Brief exposure of the unencapsulated siaD mutant to HUVEC is a potent inducer of expression of CD62E. (A) N. meningitidis B1940 siaD mutant at 106 organisms per ml or 10 ng of E. coli LPS was added to HUVEC, which were then washed thoroughly with fresh medium at 5, 15, 60, 120, and 240 min. (B) N. meningitidis B1940 siaD and parent organisms were incubated at 106 organisms per ml and washed at 15, 60, 120, and 240 min. CD62E expression was determined by flow cytometry after 5 h of incubation. The results shown here are representative experiments from at least three separate experiments that yielded similar results.

Bacterial concentration, LOS structure, and capsulation influence the inhibitory effect of rBPI₂₁ on the induction of HUVEC adhesion molecule expression by meningococci. The effects of the endotoxin antagonist rBPI₂₁ on the induction of HUVEC adhesion molecule expression were investigated. When 10 µg of rBPI per ml was added prior to or at the same time as 100 ng of E. coli LPS per ml, adhesion molecule expression was completely inhibited (Fig. 5). However, when rBPI₂₁ was added to meningococci, the pattern of cell adhesion molecule expression was more complicated. First, the bacterial concentration was found to be important. When 10 µg of rBPI₂₁ per ml was added to either the parent B1940 or siaD mutant at 10⁴ organisms/ml, there was a marked reduction in adhesion molecule expression. However, when the bacterium was present at 10⁵ organisms/ml, inhibition was only between 20 and 50%. Little effect was observed at 10⁶ organisms/ml (Fig. 6). Second, LOS structure and encapsulation also influenced the efficacy of rBPI₂₁. When 10⁵ organisms of N. meningitidis B1940 or the isogenic mutants per ml were incubated with rBPI₂₁, the levels of CD62E expression were more efficiently inhibited for the cps and cpsD mutants than for either the parent or siaD mutant (Fig. 7A). In separate experiments, the level of inhibition seen with the lst mutant was similar to that observed with the cps and cpsD mutants (Fig. 7B), indicating that LOS sialylation may be the major determinant of this effect. rBPI₂₁-mediated inhibition of either ICAM-1 or VCAM-1 up-regulation by meningococci was similar to that seen with CD62E expression (results not shown).

View larger version (25K): [in this window] [in a new window]   FIG. 5.   Effect of rBPI21 on E. coli O111:B4 LPS-induced HUVEC expression of CD62E, ICAM-1, and VCAM-1. E. coli LPS (100 ng/ml) was added to HUVEC preincubated with 10 µg of rBPI21 per ml. Cell adhesion molecule determination was determined by flow cytometry after 5 h of incubation. The results shown are representative of multiple experiments with same level of inhibition.

View larger version (22K): [in this window] [in a new window]   FIG. 6.   Effect of rBPI21 on HUVEC expression of CD62E by various doses of the N. meningitidis B1940 siaD mutant. rBPI21 (10 µg/ml) or medium was added to HUVEC, and then either 100 ng of E. coli O111:B4 LPS or various concentrations of bacteria were added as shown. CD62E expression was determined after 5 h. The data presented here are histograms of fluorescence intensity of phycoerythrin fluorochrome plotted against the number of events. The dotted line represents resting CD62E expression; the shaded area represents CD62E expression with LPS or bacteria; and the continuous black line represents CD62E expression when LPS or bacteria were preincubated with rBPI21.

View larger version (20K): [in this window] [in a new window]   FIG. 7.   Effect of rBPI21 on induction of CD62E expression on HUVEC in response to N. meningitidis B1940 parent and cps, siaD, and cpsD mutants (A) and B1940 parent and the lst mutant (7B). HUVEC were pre-incubated with 10 µg of rBPI21 per ml, and then 105 organisms/ml were added. CD62E expression was measured after 5 h by flow cytometry. The results shown are the mean median fluorescence intensity (MFI) and standard error of the mean from three separate experiments.

The influence of rBPI₂₁ on induction of HUVEC adhesion molecule expression after exposure to N. meningitidis or endotoxin. The capacity of rBPI₂₁ to influence adhesion molecule expression after bacterial exposure was investigated. HUVEC were exposed to either the parent (nonadherent) or the siaD mutant (adherent) for 15 min before washing. rBPI₂₁ was then added at 15, 60, and 120 min after bacterial exposure. A reduction of between 40 and 60% in the expression of CD62E was observed when rBPI₂₁ was added after a 15-min exposure to the bacteria (Fig. 8). This level of inhibition was similar even if rBPI₂₁ was added at 60 or 120 min. While the siaD mutant induced higher levels of CD62E than the parent organism, the relative effect of rBPI₂₁ was similar for both organisms.

View larger version (29K): [in this window] [in a new window]   FIG. 8.   Effect of time exposure of rBPI21 on induction of CD62E on HUVEC after 5 h of stimulation with 10 ng of E. coli O111:B4 LPS or N. meningitidis B1940 parent and siaD mutant strains. Bacteria were added at 106 per ml and washed off after 15 min. Then 5 µg of rBPI21 was added at 15, 60, and 120 min. E. coli O111:B4 LPS (10 ng) was added, and 5 µg of rBPI21 was added at the same time points. The results shown are a representative of three experiments with similar results.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

The nature and degree of the host inflammatory response to N. meningitidis may determine the fate of an infected individual. While host factors such as cytokine gene polymorphisms are likely to be determinants of this response, there is increasing recognition that bacterial properties are also important. This study provides evidence that bacterial composition can influence host endothelial cell responses. In view of the importance of vascular injury in this condition, these findings may be pertinent to understanding the pathophysiology of meningococcal disease.

In this study, we have used HUVEC to investigate the expression of adhesion molecules in response to N. meningitidis. We found this organism to be a potent inducer of the major vascular endothelial cell adhesion molecules, CD62E, ICAM-1, and VCAM-1, even after only limited periods of exposure. Our previous studies have shown that meningococci also markedly enhance the expression of the ₂-integrin CD11b/CD18 and diminish the expression of the selectin CD62L on neutrophils (20). Since these counterreceptors on both cell types are responsible for neutrophil adhesion and transmigration through vascular endothelia, these findings provide a mechanism to account for the presence of neutrophils within meningococcal lesions (29).

Capsulation and LOS structure have been shown to be determinants of meningococcal survival in human serum, in whole blood, and in the infant-rat model of meningococcal disease (16, 36-38). We have been able to show that these bacterial properties can also influence the degree and pattern of endothelial adhesion molecule expression. Interestingly, the organisms that induced the highest levels of adhesion molecules in this study were also those that we have previously shown to cause the largest reduction in neutrophil CD62L expression (20). These experiments were performed with whole blood and appeared to be related to the degree of bacterial killing. In the present study and in a previous study of meningococcal induction of endothelial cell tissue factor (10), bacterial killing was not the explanation for the differing effects of LOS structure and capsulation on endothelial cell activation. The reasons for the influence of these bacterial structures on endothelial cell activation are complex.

Capsulation is an important determinant of B1940 adhesion to HUVEC. Capsule-deficient mutants are more adherent to endothelial cells than are the capsulated parent strain and the nonsialylated, capsulated organism, B1940 cpsD (20). Possession of a truncated, nonsialylated LOS also influences bacterial adhesion to HUVEC, as indicated by enhanced binding of the B1940 cps mutant compared to the B1940 siaD mutant. In this study we show that adhesion molecule expression is directly correlated with the adherence of these bacteria to endothelial cells, which indicates that the adhesive capacity of N. meningitidis is likely to be a major factor in determining endothelial adhesion molecule expression.

One interpretation of our data is that the more adherent organisms effectively provide a higher dose of endotoxin to the endothelial cells. This appeared to be the explanation for the enhanced tissue factor expression seen in response to the siaD mutant compared to the parent strain (10). However, the dose of endotoxin does not fully account for the differences in adhesion molecule expression seen in this study. Meningococci, especially the unencapsulated cps and siaD strains, were more potent inducers of CD62E expression on HUVEC than were high doses of endotoxin. Indeed, the level of CD62E caused by these unencapsulated mutants after only 15 min of exposure was higher than that seen when HUVEC were continually exposed to endotoxin for a full 5 h. This was not observed with either ICAM-1 or VCAM-1, where endotoxin was at least as effective as bacteria at inducing the expression of these two cell adhesion molecules. Our results show that this differential activation is true for both purified, smooth E. coli O111:B4 endotoxin and purified group B meningococcal endotoxin, indicating that this effect is not simply due to structural differences between the endotoxins from the two sources. Why this differential activation should occur is not clear, but it does suggest that the dose of endotoxin may not be the sole determinant of vascular endothelial adhesion molecule expression.

There are a number of bacterial factors that contribute to the attachment of meningococci to endothelial cells (35). Pili appear to be important for initial endothelial-cell adhesion of encapsulated organisms, while firm attachment may be mediated by the outer membrane proteins Opa and Opc, particularly in unencapsulated strains (33). These bacterial components that mediate attachment to and invasion of host cells may also cause activation of important signal transduction pathways involved in transcriptional regulation of genes in the inflammatory response. It has been shown that both pili and Opa from N. gonorrhoeae can activate epithelial cells via the transcription factors NF-B and AP-1 to produce inflammatory cytokines independently of endotoxin (26). These studies also showed no differences in NF-B activation between invading and noninvading strains of N. gonorrhoeae (25). If this were the case, it would provide a putative mechanism by which the more strongly adherent organisms were also capable of inducing the highest levels of adhesion molecule expression. This is presumably because, in unencapsulated strains, there may be enhanced interaction between opacity proteins and their ligands. It may also explain the differential patterns of CD62E, ICAM-1, and VCAM-1 seen in response to the bacteria and purified endotoxin. It has been demonstrated recently that invasive strains of N. gonorrhoeae cause greater expression of epithelial ICAM-1 than do noninvasive strains. Interestingly, this was not associated with an increase in transcription of ICAM-1 mRNA, indicating that ICAM-1 expression can be modulated by bacteria at the translational or posttranslational level (13). We suggest that there are multiple signalling mechanisms that occur when meningococci interact with vascular endothelial cells. Variations in bacterial structure could affect these mechanisms. We are currently undertaking studies to investigate the signal transduction pathways induced in activation of endothelial cells by the meningococcal mutants.

A further explanation for our results comes from the recent discovery of human Toll-like receptors that transduce signals in response to endotoxin in both CD14⁺ and CD14 cells (43). There are at least five human homologues of the Toll receptor, and two of these, TLR2 and TLR4, transduce endotoxin signals (28). The demonstration of these receptors has highlighted the potential complexity of endotoxin signalling to host cells. Initial findings indicate that different Toll receptors may have different properties and may relate to observed differences in signal transduction (19). It is becoming clear that the location and form of endotoxin are central to how it interacts with host endotoxin recognition molecules (42). This may be critical in how host cells interact with whole bacteria such as N. meningitidis. Our results demonstrate that LOS structure is an important factor in determining the pattern of vascular endothelial adhesion molecule expression. The cpsD mutant, which possesses a truncated LOS, and the lst mutant, which lacks just the terminal sialic acid of the -oligosaccharide chain, induced higher levels of adhesion molecules than did the parent organism. This indicates that even subtle changes in LOS structure can influence endothelial-cell activation, possibly by influencing the interactions between the bacteria and endothelial-cell endotoxin recognition receptors.

In the light of these findings, we investigated the capacity of rBPI₂₁, a recombinant form of BPI, a host defense protein with antibacterial and antiendotoxin activities, to modulate the endothelial adhesion molecule response to these different bacterial strains. rBPI₂₁ kills gram-negative bacteria and decreases tumor necrosis factor alpha production induced by gram-negative bacteria or LPS in whole blood (41). rBPI₂₁ also binds to purified endotoxin and abrogates the degree of endothelial-cell activation (2). In our study, we found that 10 µg of rBPI₂₁ per ml could completely abolish the induction of cell adhesion molecules on HUVEC when given prior to or very early following exposure to high doses of purified E. coli endotoxin. However, the pattern seen with organisms was more complex. Our results show that the effectiveness of rBPI₂₁ in blocking the induction of cell adhesion molecules by bacteria was dependent not only on the dose of infecting bacteria but also on the structure of the LOS; in particular, LOS sialylation may influence rBPI₂₁ activity. Previous studies have shown that BPI and its N-terminal fragment bind to the lipid A portion of endotoxin, resulting in endotoxin neutralization (6, 22). How LOS sialylation could influence rBPI₂₁ binding to the lipid A component of LPS is not known. Structure of polysaccharide can have an effect on the inhibitory effects of rBPI₂₁ on some LPSs (3). It has been demonstrated that mutants of E. coli that have incomplete LPS are much more sensitive to bactericidal permeability-increasing protein than are E. coli strains with intact LPS. It has been proposed that charged sugar moieties could impede the action of cationic proteins like BPI (40). Our results indicate that just the presence of sialic acid on the terminal lacto-N-neotetraose is sufficient to influence the inhibitory effects of BPI. This is presumably due to the effect of sialic acid on LOS-BPI interactions. This could be important in vivo, since most pathogenic meningococci have LOS structures (immunotypes) that can be sialylated (15). These findings may also indicate that the dose of rBPI required to inhibit the inflammatory effects of gram-negative bacteria may have to be optimized for different bacteria strains.

One of the disappointing features of most anti-inflammatory agents is that they are effective only when administered prior to or at the same time as the inflammatory stimulus. Our results indicate that this was not the case with rBPI₂₁ in vitro. We found that rBPI₂₁ could influence adhesion molecule expression even when added several hours after the organisms. This was true whether the organism was adherent or nonadherent. These results are consistent with previous studies involving E. coli LPS, which have shown that rBPI₂₁ can reverse the LPS-induced expression of CD62E (E-selectin) after several hours of continuous exposure (12).

Taken together, these results indicate that LOS structure and the presence of a capsule can influence the level, kinetics, and profile of endothelial adhesion molecule expression induced by N. meningitidis. The apparent ability of pathogenic Neisseria strains to down-regulate their capsule and desialylate the LOS in natural meningococcal infections would indicate that the endothelial-cell response to invading organisms may be variable even for a single strain (8, 9). It would appear that even limited exposure to some organisms might be able to induce a rapid influx of inflammatory cells. This may be beneficial in killing invading organisms but could influence the degree and extent of endothelial cell injury from activated inflammatory cells (27). Strategies to combat meningococcal disease should take into account the multiple signalling pathways that may be activated during the course of this potentially fatal disease.