EVERYONE NEEDS TO READ THIS, BPI IS NEAR THE END OF THIS POST.
Endotoxin or LPS comprises the major component of the outer membrane of all gram-negative bacteria. Therefore, the recognition of and response against this glycolipid by the host will provide protection against a wide range of microorganisms. The host can be exposed to LPS due to translocation of gut bacteria into the circulation, as a result of mucosal epithelium dysfunction or due to severe trauma, whereas another route of exposure is the inhalation of air-borne LPS (Chapter 1).
LPS interacts with a broad range of humoral and cellular systems of the host. After entering the circulation, LPS binds to lipoproteins in plasma, resulting in a reduction of its biological activity. Furthermore, LPS activates the complement system, and induces coagulation, whereas also the liver and the neuro-endocrine system are also activated, resulting into metabolic alterations. In parallel, cells of the immune system, like monocytes and PMN, but also endothelial and epithelial cells, which perform besides their barrier function also an immune function, are activated. These cellular activation processes are mediated via LBP and CD14. The diversity of physiological responses act in concert and are directed at the elimination of free LPS and gram-negative bacteria, and subsequently at the repair of damaged tissues.
However, under certain conditions, like the continuous presence of high concentrations of LPS, or as a result of enhanced sensitivity for LPS, the actions of these normally well regulated defense systems, become noncontrollable. As a consequence, instead of contributing to protection against microorganisms, these responses will be detrimental for the host. More insight in the pathways of LPS action will help to develop therapeutic strategies to intervene in LPS-mediated pathology. To this end, a series of studies was performed analyzing several aspects of host response to LPS.
The sensitivity for LPS, which varies considerably across species (1), is strongly increased due to exposure to environmental agents, like lead (2). The mechanisms which underlie this phenomenon are not yet fully understood. Since TNF is an important mediator of LPS effects (3), the role of this pro-inflammatory cytokine in this phenomenon was investigated (Chapter 2).
In this study evidence was provided that TNF plays an important role in the mechanism of the enhanced sensitivity of animals to endotoxin after exposure to lead. It was shown that mice after exposure to lead become more sensitive to TNF, in a similar way as to LPS. Furthermore, lead treated mice had lower serum TNF levels two hours after LPS exposure, whereas serum TNF levels at six and eight hours after LPS were enhanced as compared to mice not exposed to lead. Finally, a monoclonal antibody (mAb) directed against mouse TNF prevented lethality induced by the combined administration of lead and LPS.
Other studies confirmed the central role of TNF in the enhanced susceptibility for LPS, after exposure to hepatotoxic agents like lead and D-galactosamine (4,5). Recently, it was demonstrated that the enhanced sensitivity to LPS induced by D-galactosamine is caused by TNF-induced apoptosis in hepatocytes with transcriptional arrest (6). This indicates that lead most probably induces sensitization via a similar pathway, although this had not yet been confirmed.
Furthermore, a series of studies was preformed to analyze the regulation of cellular responses to LPS. Studies by Wright et al. demonstrated that CD14, an antigen present on monocytes but also at low levels at the surface of PMN (7), functions as a receptor for LPS. LPS was shown to bind to CD14, a process mediated by the plasma protein LBP, whereas blockade of CD14 prevented LPS-induced TNF release by monocytes (8,9). Next to TNF, the pro-inflammatory cytokine IL-6 and the chemokine IL-8 are important mediators of LPS-induced biological effects (10,11). Therefore, it was analyzed whether the release of these cytokines was also mediated via CD14 (Chapter 3).
First, a series of anti-CD14 mAb were tested for their ability to block LPS- induced TNF release. Two out of the five anti-CD14 mAb tested, namely MEM-18 and Cris6 which recognize the same epitope (7) inhibited LPS-induced TNF release by monocytes, indicating that a specific CD14 epitope is involved in CD14-LPS recognition. Evidence that CD14 is also involved in the LPS-induced IL-6 and IL-8 release was obtained from the observation that anti-CD14 mAb MEM-18 blocked the LPS-induced, TNF independent, release of IL-6 and IL-8. This inhibition was most pronounced at low LPS concentrations, and only partial at higher LPS concentrations. A possible explanation for this observation is that at high LPS concentrations cytokine release is induced via both CD14-dependent and independent pathway.
Furthermore, MEM-18 also blocked cytokine release by monocytes stimulated under serum-free conditions, indicating that LPS is able to interact directly with CD14, as was confirmed in other studies (12,13). However, as the production of cytokines induced by LPS via CD14 is generally higher in the presence of serum as compared to serum-free conditions, serum factors like LBP, seem to enhance the affinity of LPS to membrane CD14. In addition, the LPS-induced cytokine release by alveolar macrophages was also blocked by MEM-18. This indicates that on these differentiated mononuclear phagocytes CD14 also functions as a receptor for LPS. From this study was concluded that CD14 is involved in the LPS-induced release of TNF, IL-6 and IL-8 by monocytes and alveolar macrophages.
Besides monocytes, endothelial cells function as immunologically active cells and respond strongly to LPS (14). Whereas, as described in chapter 3, CD14 is involved in LPS-induced monocyte activation, the mechanisms of LPS-induced endothelial cell activation were not elucidated. This prompted us to analyze the role of CD14 in LPS-induced endothelial cell activation (Chapter 4).
This study demonstrated that, comparable to monocytes, CD14 is involved in LPS-induced endothelial activation. The anti-CD14 mAb MEM-18 blocked the LPS-induced human endothelial cell (HUVEC) activation, as measured by the novo expression of the adhesion molecule E-selectin, and the release of IL-6. This inhibition was most pronounced at low LPS concentrations. Messenger RNA encoding E-selectin and IL-6 was reduced in parallel, indicating that MEM-18 blocked these LPS effects at the pretranscriptional level. MEM-18 did not affect endothelial cell activation induced by IL-1, TNF, and phorbol myristate acetate (PMA), indicating that anti-CD14 mAb specifically inhibited HUVEC responses to LPS.
Since IL-6 release and E-selectin expression on HUVEC were strongly reduced when LPS activation was performed in the absence of serum, the involvement of serum components in LPS activation of HUVEC was indicated. Nevertheless, MEM-18 also blocked LPS-induced HUVEC activation under serum-free conditions, suggesting that LPS is able to interact directly with the endothelial cell membrane associated CD14. This hypothesis however, could not be confirmed, since no indications for expression of CD14 on endothelial cell surface were obtained, neither by flow cytometry, nor by surface labeling.
Although the involvement of CD14 in LPS-induced endothelial cell activation was clearly shown, the role of serum components was not completely elucidated. Further studies by others showed that the plasma protein LBP mediates binding of LPS to sCD14, also present in plasma. The LPS-sCD14 complex activates cells that do not express membrane CD14, like endothelial cells, epithelial cells, and astrocytes, most probably via an yet unidentified receptor on these cells (15,16).
The pro-inflammatory cytokine TNF, mediating many LPS effects, activates cells via interaction with membrane receptors. Most cells express two types of TNF receptors namely TNFR55 and TNFR75, which can be shed from the cell surface after activation, resulting into soluble forms (17-19). sTNFR probably play a role in inactivation of circulating TNF, by preventing the interaction with membrane TNFR, although low sTNFR concentrations were reported to stabilize the trimeric structure of TNF, so reducing its decay of activity (20,21). Since monocytes are a primary target for LPS, it was investigated whether LPS also affected TNFR shedding or expression, whereas also the role of CD14 in this process was analyzed (Chapter 5).
LPS was shown to induce release of sTNR75, but not of sTNFR55, reaching plateau levels after approximately 48 hours. The induction of TNFR75 release was mediated via LBP and CD14, since anti-CD14 mAb inhibited LPS-induced TNFR75 shedding, whereas the LPS-induced TNFR75 shedding, which was strongly reduced under serum-free conditions could be restored by addition of LBP. Furthermore, BPI, a PMN protein known to possess LPS neutralizing capacity (22), blocked LPS-induced TNFR75 shedding.
Also the cell surface expression of both TNFR was affected by LPS. On freshly isolated monocytes the expression of both TNFR was similar, which increased gradually during culture of monocytes. In contrast, after 1 hour exposure to LPS, the expression of both receptors was completely reduced, followed by enhanced re-expression after 24 hours of stimulation. This LPS effect was enhanced by LBP, and blocked by BPI.
LPS thus not only activates monocytes to release TNF, but also controls the TNF effects by inducing shedding of TNFR75, possibly inhibiting TNF effects, and by temporary downregulation the expression of both TNFR55 and TNFR75. Moreover, both the release of TNF and the LPS-induced TNFR shedding are mediated via a similar pathway mediated by LBP and CD14, and blocked by BPI.
Studies described in chapter 2 to 5 showed that endogenous proteins strongly affect LPS-induced cell activation, as is confirmed in other studies (11,23,24). LBP and BPI, two proteins with opposing effects, have considerable sequence homology (9), and both bind to the lipid A part of LPS (25,26). This urged us to study the functional relationship between these two proteins (Chapter 6). To this end, the effect of both proteins on LPS-induced cytokine release by monocytes was analyzed.
LBP was demonstrated to enhance not only the LPS-induced TNF release (as reported previous:9,12), but also the IL-6 and IL-8 release by monocytes stimulated under serum-free conditions. In contrast to the enhancing effect of LBP, BPI inhibited the release of all these three cytokines. Furthermore, it was demonstrated that the effects of LBP and BPI on LPS-induced cytokine release by monocytes were counteractive. BPI interfered with the enhancing effect of LBP on the LPS-induced cytokine release. At high LBP to BPI ratios, BPI could no longer inhibit LBP-induced enhancement. In accordance, increasing concentrations of BPI abrogated the LBP effects.
In order to analyze the mechanism of these antagonistic activities, it was studied whether LBP and BPI compete for binding to LPS. To this end, an assay was used that detects binding of free BPI to an anti-BPI mAb, which will be further described below. Presence of LPS prevented binding of BPI to anti-BPI mAb, whereas preincubation of LPS with LBP prevented this inhibition, indicating that LBP and BPI bind on the same site of LPS.
It was therefore concluded that the antagonistic effects of LBP and BPI are due to competition for binding to LPS (Figure 1), as was confirmed by other reports (27,28). These data indicate that the balance between BPI and LBP levels determines LPS activity under physiologic conditions. Although BPI, in contrast to LBP, is not present in blood of healthy individuals, levels can be enhanced during disease, as will be discussed below.
Figure 1. LBP and BPI have antagonistic effects upon LPS-induced monocyte activation. This antagonistic effects are due to competion for interaction with LPS. Whereas the interaction of LPS with LBP results in CD14 mediated monocyte activation, BPI binding blocks LPS interaction with CD14 and, thus prevents cell activation.
The PMN protein BPI, which was originally characterized as a potent bactericidal protein (29), thus appears to be a very important protein neutralizing LPS-induced cell activation. Next, the presence of this protein in health and disease was investigated. Firstly, mAb directed against BPI were developed and characterized (Chapter 7).
Two BPI-specific mAb, named 4E3 (IgG1) and 5D7 (IgG2a) were generated. The specificity of these mAb was shown in several ways. Both mAb reacted with BPI coated on immuno-plates, in a BPI concentration dependent manner. In addition, the anti-BPI mAb coated in wells captured biotinylated BPI. These mAb recognized native BPI, but not BPI denatured by sodium dodecyl sulfate. Furthermore, the anti-BPI mAb did not cross-react with LBP, a protein with high sequence homology to BPI (9).
The use of 4E3 and 5D7 in flow cytometry revealed that PMN contained intracellular BPI, which was confined to the granules in the cytoplasm, as shown by immunofluorescence. These data correspond to previous reports of the presence of BPI in azurophilic granules (30,31). Furthermore, also low PMN surface expression of BPI was detected, in accordance with other studies (32,33).
In addition, both anti-BPI mAb were able to block biological activity of BPI. The inhibition by BPI of LPS-induced limulus amebocyte lysate activation, and of LPS-induced TNF release by monocytes, was prevented by 4E3 and by 5D7. Moreover, LPS was shown to interfere with the interaction of anti-BPI mAb with BPI in an enzyme linked immunosorbent assay (ELISA) system. Together, these data strongly suggest that both mAb recognize an epitope on BPI involved in the BPI-LPS interaction.
In chapter 7 it was described that BPI is present not only inside PMN, but also on the cell surface of PMN. Next it was investigated whether monocytes, which are extremely sensitive for LPS, also express the specifically LPS-binding protein BPI on their cell surface (Chapter 8). Flow cytometric analysis, using anti-BPI mAb 4E3 and 5D7, and anti-BPI mAb kindly provided by M. Marra (Incyte, Palo Alto, Ca) revealed that monocytes, express BPI on their cell surface. However, no indications for intracellular presence of BPI were obtained.
Furthermore, it was demonstrated that monocyte cell surface BPI binds LPS, because binding of anti-BPI mAb 4E3 (which, as described above, does not react with BPI to which LPS is bound) to cell surface BPI was strongly reduced after preincubation of cells with LPS. However, cell surface BPI did not contribute quantitatively to the interaction of LPS with the monocyte cell membrane, since preincubation of cells with 4E3 did not block binding of LPS to monocytes.
Although we assume that BPI is adsorbed by monocytes from the environment, no indications for this hypothesis were obtained, since incubation of monocytes with biotin labeled BPI, or with human serum, known to contain BPI, did not enhance the cell surface expression of BPI as compared to untreated cells. The origin and the function of BPI present on the surface of monocytes and PMN have therefore to be further elucidated. In a recent report it was shown that the monocytic cell line THP-1 is able to adsorb exogenous BPI, which still exerted biological activity (34).
BPI which was originally shown to exert its bactericidal activity intracellularly (35), was recently shown to be released during pathological conditions and to perform biological activity under these conditions (36). These data indicate that BPI released by activated PMN can contribute to the protection against LPS and gram-negative bacteria. Therefore, the release of BPI induced by physiological agents was studied, using a human whole blood system (Chapter 9). BPI levels as present in plasma were detected using a BPI specific ELISA, which is described in more detail in the next study.
Out of a series of microbial agents tested LPS most potently induced BPI release. Exposure of whole blood to FMLP, serum treated zymosan (STZ) and Lipoteichoic acid (LTA) also lead to BPI release. In contrast, Staphylococcal enterotoxin B did not induce BPI release.
In addition, the inflammatory mediator TNF potently activated PMN, via TNFR55, in whole blood to release BPI, whereas IL-1, IL-8, PAF and complement factor C5a hardly induced the release of BPI. The LPS-induced BPI release was not mediated via production of TNF, since agents blocking biological activity of TNF did not affect LPS-induced BPI release.
Whereas LPS, FMLP and LTA potently induced BPI release in whole blood these stimuli hardly activated isolated PMN to release BPI. This is in agreement with other reports showing that for release of azurophilic granules additional stimulation is required, like adherence to solid phase or stimulation with cytochalasin (37,38). These data further indicate that in whole blood other cellular or humoral factors are stimulated that are involved in the activation of PMN. In contrast, STZ and PMA a protein kinase C activator, seemed to activate PMN directly, since isolated PMN were stimulated to a similar magnitude as PMN in whole blood.
Also, BPI release was compared with release of the well studied azurophilic granule protein elastase (39). Elastase was shown to be present in PMN in considerable higher amount than BPI (approximately 3 g versus 200/106 PMN). Expressing the amount of either protein released as percentage of total amount present, revealed that both proteins were released with similar magnitude. Furthermore, both BPI and elastase were released with similar kinetics, which started within 30 minutes after onset of stimulation and lasted up to 1 or 4 hours, suggesting that the release of both proteins is under similar regulatory control.
Although we demonstrated that LPS and TNF are relatively potent inducers of BPI release, the maximal amount of BPI released was maximally 20% of total BPI content present in PMN. This indicates that major part of BPI will remain intracellular, and confirms other studies that under non-vigorously conditions BPI functions mainly intracellularly (35).
In order to further study the role of extracellular BPI and its impact on disease, BPI levels present in healthy volunteers and in patients with microbial infection were analyzed (Chapter 10). To this end, an ELISA for measurement of human BPI was developed, by using anti-BPI mAb 4E3 as catching reagent, and biotinylated polyclonal rabbit anti BPI IgG as detector. Since the presence of LPS disturbed detection of BPI in this ELISA, 80 mM Mg2+, known to prevent BPI-LPS interaction (40), was used in assay buffers. Furthermore, the use of Mg2+ in assay buffers also prevented aspecific adherence of the cationic BPI to solid phase. This newly developed specific BPI ELISA could be used for detection of BPI in biological samples and had a lower detection level of 200 pg/ml.
In plasma of healthy volunteers no BPI was detectable. It was shown that PMN can be activated in collection tubes, indicating that proper collection is required, including the use of EDTA as anticoagulant as well as rapid handling of blood-samples. In contrast to absence of BPI in healthy volunteers, in biological fluids of patients BPI was detected. BPI was present in plasma of critically ill intensive care unit patients (n=60; mean 4,8 ng/ml), in bronchoalveolar lavage fluid of patients suspected for pneumonia (n=25; mean 16,2 ng/ml), in wound fluid (n=18; mean 1860 ng/ml) and in pleural fluid (n=44; mean 3,1 ng/ml). In subgroups of patients with culture proven bacteria mean BPI levels were enhanced compared with subgroups without bacteria, although the differences were only significant in plasma of intensive care unit patients.
Although it was shown in chapter 10 that BPI is released during pathological conditions, the biological effects of BPI will be affected by the presence of other endogenous LPS binding proteins like LBP and sCD14 (15,41). Therefore, next the levels of all three of these LPS-activity modulating proteins were analyzed in bacteremic patients, and correlated with clinical parameters (Chapter 11).
BPI and LBP levels were significantly enhanced in bacteremic patients (n=64) as compared to healthy controls (n=49) (mean, 8.4 ng/ml versus <.2 ng/ml; 216 g/ml versus 16 g/ml), in accordance with other study (42,43). sCD14 levels in bacteremic patients were not changed as compared to healthy controls, although in a subgroup of patients with gram-positive bacteria significantly higher sCD14 levels were detected as compared to a subgroup of patients with gram-negative infections (2.7 g/ml versus 2.2 g/ml).
The BPI/neutrophil ratio, which is used as an indicator for neutrophil activation, was significantly associated with the presence of sepsis syndrome and death in bacteremic patients. Furthermore, high peak LBP levels were significantly correlated with sepsis, but not with mortality. On the contrary. patients with sepsis syndrome were observed to have reduced sCD14 levels, as opposed to another study (44).
Analyzing the kinetics of BPI and LBP levels in a subgroup of patients revealed that levels of both proteins were already enhanced before demonstration of positive blood culture. Since LBP and BPI exert antagonistic effects concerning LPS-induced cell activation, the ratio of these proteins was also analyzed, which did not correlate with mortality.
BPI and LBP probably exert their main functions locally at the site of infection, therefore the exact role of enhanced circulating BPI and LBP levels as detected in bacteremic patients remains unclear. It can be speculated that, due to leakage, circulating levels reflect levels locally present. Indications that at site of infection BPI can be present at greater concentrations than LBP, thus able to exert its biological function were provided by studies of Opal et al. showing a BPI/LBP ratio >1 in abscess fluid (45).
In conclusion, a series of studies was performed to obtain more insight in the pathways of LPS action. First, it was demonstrated that TNF is an important intermediate in the phenomenon of the enhanced sensitivity for LPS obtained after exposure to lead. Furthermore, it was shown that monocyte membrane CD14 is a general 'LPS-receptor' since this antigen is involved not only in the LPS-induced release of pro-inflammatory cytokines but also in the LPS-induced shedding and downregulation of TNFR. These LPS activities were enhanced by LBP, facilitating the binding of LPS to CD14, and blocked by the PMN protein BPI. Furthermore, both proteins were demonstrated to have antagonistic effects, due to competition for binding to LPS. Besides, it was revealed that CD14 is also involved in LPS-induced endothelial cell activation.
In addition, studies on the potent LPS neutralizing protein BPI were performed, using newly developed antibodies. It was shown that monocytes, like PMN, express BPI on their cell surface. Furthermore, analysis of BPI release in whole blood revealed that LPS and TNF are relatively potent inducers of BPI release. However, major part of BPI remained inside PMN, indicating that BPI functions mainly intracellularly. In blood of healthy volunteers no BPI was detectable, whereas in biological fluids of several patient groups BPI was present. The biological function of extracellular BPI will depend on the levels of other endogenous LPS binding proteins. It was observed that in infectious lesions like wound fluid, high BPI levels were present, and it can be expected that under these conditions BPI will contribute to fluid phase protection against LPS. However, under less rigid conditions, BPI most probably mainly contributes to LPS defense as an intracellular protein. Further studies are required on the function of this protein, which could possibly serve as therapeutic agent in protection against LPS and gram-negative bacteria.
|
