Pathogenetic Aspects of Sepsis and Possible Targets for Adjunctive Therapy

O. Rigato, E. Silva, E.G. Kallas, M.K.C. Brunialti, P.S. Martins and R. Salomao*

Division of Infectious Diseases, Universidade Federal de São Paulo – UNIFESP, Rua Pedro de Toledo, 781 15^o andar, 04039-032 São Paulo SP, Brazil

*Address correspondence to this author at the Division of Infectious Diseases, Universidade Federal de São Paulo – UNIFESP, Rua Pedro de Toledo, 781 15^o andar, 04039-032 São Paulo SP, Brazil, FAX and Phone: 0055-11-5081 5394, E-mail: rsalomao-dipa@pesquisa.epm.br

 

Abstract:The outcome of patients with sepsis arises from multiple factors affecting both the host and the invading microorganisms. Age, presence of underlying disease, source of infection and some specific etiological agents have been related to prognosis.

Appropriateness of antimicrobial therapy, considering the in vitro susceptibility tests for the infecting bacteria, has been strongly associated with the outcome. Therefore even after the cascade of sepsis has been triggered, the control of bacteria growth is still fundamental for the outcome of the infection. This is a major distinction point from experimental studies in which whole killed bacteria and their products are used as model of sepsis. However, even within the setting of adequate antimicrobial use, patients still die of sepsis. Thus, strategies focusing on further therapy targets are an important area of interest for basic and clinical research.

Although such adjunctive sepsis therapy has failed to achieve consistent better survival rates so far, nevertheless, it improved our understanding of the pathophysiological events seen in sepsis that the possibility that a new and effective treatment may arise has been warmly considered.

In this paper we aim to review some aspects of the pathogenesis of sepsis, focusing on recent advances and on possible targets for adjunctive therapy. Published clinical trials and experimental data supporting such trials are commented on.

I. INTRODUCTION

     The pathogenesis and clinical manifestation of sepsis are the results of an impressive interaction between an infecting microorganism and the host. Evolving in different stages of a continuum process it places the physician to one of its major challenges, a medical emergency associated with extremely high morbidity and mortality [1], in which the healing and deleterious response are often part of the same process.

     A dramatic increase in the incidence and a continuous change in the etiology of bloodstream infections (BSI) have been seen during the last century [2-9]. Incidence and etiology are under the influence of host – such as age and the presence of underlying disease – and hospital factors, with higher attack rate in major teaching hospitals. Therapy for the underlying diseases and invasive life support procedures in intensive care units are also importantly related to BSI.

     It is well established that the presence of bacteria in the bloodstream is neither mandatory and nor enough to define sepsis and its sequelae. In an attempt to better define the spectrum of disorders associated with sepsis, Bone et al. have proposed a diagnostic system based on progressive stages of sepsis disease: bacteremia (presence of positive blood cultures); sepsis (clinical evidence suggestive of infection plus signs of a systemic response to the infection); sepsis syndrome (clinical diagnosis of sepsis as outlined above, plus the evidence of altered organ perfusion) and septic shock (clinical diagnosis of sepsis syndrome plus hypotension) [10]. The pathophysiologic events seen in sepsis may also be triggered by non-infectious stimuli, such as trauma, burns and acute pancreatitis. In a SCCM/ACP Consensus Conference, the term "systemic inflammatory response syndrome" (SIRS) was proposed to embrace the common inflammatory pathways involved in all these different events, while the term sepsis is to be restricted to SIRS caused by infection [11].

     A number of factors have been related to poor outcomes in sepsis: extremes of age, nosocomial-acquired infections, source of infection, presence of multiple organ dysfunction, in particular shock, severity of the underlying disease and inappropriate use of antimicrobial agents [3, 4; 12-16]. Two of them deserve special attention: the presence of underlying disease and appropriateness of antimicrobial therapy. Both are extremely relevant for the clinical care of patients with sepsis. They are also important areas of investigation, that is, there is a need to better define the relevance of underlying disease on host-response and susceptibility to infection, and, of paramount importance, to develop new therapeutic agents to treat infections caused by increasingly resistant microorganisms.

     Before further considerations on perspectives for adjunctive therapy of sepsis it is noteworthy to emphasize the importance of appropriate antimicrobial therapy. Despite the fact that the use of appropriate antimicrobial therapy is not restricted to the results of in vitro susceptibility tests, patients with isolated bacteria resistant in vitro to the ongoing antimicrobial therapy have a much higher mortality rate when compared to patients with in vitro susceptible isolated bacteria [4, 12]. In one series of 362 patients we found a mortality rate of 57.1% for those with inappropriate therapy (in vitro resistant bacteria), 21% for those with appropriate therapy [in vitro susceptible bacteria) and 34.1% for those whose antimicrobial agents were changed based on the sensitivity of the isolated organism [17]. These data provide strong evidence of the importance of antimicrobial therapy in sepsis. Thus, clinical and epidemiological data have to be considered for the appropriate choice of an empirical antimicrobial regimen. The efforts for the rational use of antimicrobial agents to control the emergence of resistance and the development of new antimicrobial drugs are important but beyond the scope of this review. Nevertheless even when the cascade of sepsis has been triggered, the control of bacteria growth is still fundamental for the outcome of the infection. This is a major distinction point from experimental studies in which whole killed bacteria or their products are used as model of sepsis.

     Although the role of antimicrobial therapy has been well established, an adjunctive therapy for modulating the inflammatory response in sepsis is imperative, once sepsis remains with an unacceptable high mortality [18]. To reach this goal an evolving understanding of pathogenesis, and experimental and clinical trials designed to evaluate pathogenesis-oriented targets for candidate adjunctive therapy are in progress. In the following sections we will review some aspects on this field.

II. PATHOGENESIS

     The pathogenesis of sepsis involves a complex interaction between the host and the infecting microorganism [19]. Bacteria and their products may trigger in the host a cascade of cellular responses that involves several cell types and cellular pathways. The inflammatory response is essential for the control of infection but also contributes to the deleterious sequelae seen in sepsis. A better knowledge of bacterial biologically active components and their interaction with host cells is crucial to improve our understanding of the pathogenesis of sepsis.

     One of the most studied bacterial components, the lipopolysaccharide (LPS), present in the outer membrane of the gram-negative bacteria cell wall, is a pivotal molecule in the induction of pathophysiologic events during gram-negative sepsis, leading to several reactions in the susceptible host, including fever or hypothermia, diarrhea, intravascular coagulation, hypotension, shock and death [20]. It is generally accepted that the toxic effects of LPS are ultimately a consequence of a host response and macrophages have been demonstrated to be the primary targets of LPS (for a review see Salomao and associates [19] and Galanos [21, 22]. Therefore, LPS has been a very interesting tool for the investigation of pathogenesis and potential adjunctive therapy in sepsis.

IIa. LPS-Cell Binding and Internalization

     Several substances present in the serum are involved in recognition, transport and LPS-binding to a cell-associated receptor and consequent cell activation. An important protein that interacts with LPS is the LPS-binding protein (LBP) [23], facilitating the LPS binding to CD14 molecule on the target cells. CD14 is the main receptor of LPS and it is found as soluble (sCD14) and membrane-bound receptor (mCD14) [24-26]. LBP enhances LPS-induced macrophage activation and subsequent production of pro-inflammatory cytokines, such as TNF-a [23]. Another protein, the bactericidal permeability-increasing protein (BPI), also interacts with LPS, and despite 45% homology to LBP, has an antagonist effect, inhibiting LPS-induced cellular activation [27, 28].

     Binding of LPS to isolated monocytes/ macrophages [29-31] and neutrophils [32-33] is followed by intracellular transport within minutes of incubation. Fluorescein isothiocyanate -conjugated LPS (FITC-LPS) and ¹²⁵I-labelled LPS added to whole leukocyte preparations or isolated cellular populations showed that monocytes bound to LPS much more efficiently than did neutrophils and lymphocytes [23]. Using leukocyte preparations, obtained from whole blood, it has been shown that the binding of FITC-LPS was obtained only in the presence of normal human serum and that sCD14 in addition to mCD14 enables LPS to bind to monocytes [35].

     The internalization of LPS has been claimed to be an important step in cell activation [31, 32, 34], while others consider internalization of LPS and cell activation as two independent events [24, 36] .Thus, it has been shown that LPS from Rhodobacter spheroides (RsLPS) [37], which does not induce monocyte activation, binds to monocytes and neutrophils but is not transported into the intracellular space. Moreover, RsLPS blocks the induction of IL-6 by smooth LPS. Chlorpromazine-induced RsLPS internalization resulted in IL-6 production [34]. On the other hand, Lentschat et al [38], studying binding and internalization of LPS with different stages of core completion, could not correlate the kinetics of internalization with those of TNF-a release. A possible detoxifying mechanism for internalization has been considered.

     Several systems have been developed to evaluate the LPS and cell interaction, and its consequences. In most of these studies isolated mononuclear cells are used, which are frequently found to be already activated during the isolation procedure. Therefore, induction of cytokine production by LPS in whole blood has drawn increasing interest [39, 40]. Blood is not only easily accessible but also represents the transport vehicle of LPS in the organism, and the interaction of LPS with humoral and cellular components of the blood is decisive for its subsequent fate and activity [41]. Thus, the study of LPS effects in whole blood represents an in vitro system that preserves as far as possible the milieu of LPS-cell interaction, as it occurs in vivo. However, information at the cell level is usually lacking in this system. We used biotinylated LPS (LPSb) combined with subsequent FACS analysis to evaluate LPS binding, internalization and cell activation in whole blood, at the single cell level. LPS enhanced the expression of HLA-DR in monocytes and induced the expression of CD69 in T and B lymphocytes. Intracellular TNF-a was detected in monocytes following six hour of incubation. LPS-binding was not influenced by EDTA or heparin, while internalization seems to occur at earlier stages with EDTA. Cellular activation was more efficient in heparinized-blood (Brunialti et al, submitted).

IIb. LPS-Induced Cellular Activation

     The LPS-CD14 monocyte receptor binding elicits a cascade of events in the cytoplasm that culminates with the production of cytokine mRNA, such as TNF-a and IL-1. Since CD14 has no intracellular domain to transduce the LPS signal, a putative co-receptor for LPS was previously proposed [24, 36]. A breakthrough in this field came with the identification of a human homologue of Drosophila Toll, toll-like receptor 4 (TLR-4), as the LPS transmembrane signal transducer [42]. TLR4-deficient mice have been shown to be hyporesponsive to LPS [43]. Also TLR2 has been implicated in LPS-induced cell activation [44, 45], but it is more likely that TLR2 functions as a receptor for components of gram-positive bacteria, mycobacteria, yeast and other microbial pathogens [reviewed in Aderem 46]. TLR4 and other TLRs have a cytoplasmatic domain with a high degree of homology to IL-1 receptor (Toll-IL-1 receptor, TIR). Intracellular activation follows similar pathway as IL-1, involving an adapter protein and protein kinases which will result in the phosphorylation of the inhibitor of kappa B (IkB) and its dissociation of nuclear factor kappa B (NFkB), that translocates to the nucleus and induces cytokines and costimulatory molecules gene expression [47-49].

     Apart from NFkB, MAP-kinase cascades are also activated, such as ERK, JNK and p38, that are involved in transcription of cytokine mRNA [46, 48]. The role of MAP-kinase in LPS-activated cell was evidenced by the inhibition of MAP-kinase induction after blockade of LPS-CD14 binding by anti-CD14 antibodies [50]. The induction of MAP-kinase after LPS stimulus was inhibited by herbimicin A, a tyrosine-kinase antagonist [51]. Interestingly, herbimicin A did not inhibit MAP-kinase induction after stimulus with phorbol-myristate acetate (PMA), denoting specificity of tyrosine-kinase as a pathway of cell activation by LPS. In a model of lung vascular endothelial cells stimulated with LPS, the use of a specific p38 inhibitor blocked IL-8 production, a cytokine involved in the pathogenesis of acute respiratory distress syndrome (ARDS) [52]. Similarly to the inhibition of tyrosine-kinase by herbimicin A, the suppression of IL-8 production through p38 blockade occurs when the neutrophils are stimulated by LPS, but not by non-specific stimulus like PMA or ionomycin, showing distinct pathways of cellular activation [53].

IIc. LPS-Induced Cytokine Production

     Among LPS-induced cytokines, TNF-a plays a pivotal role in endotoxic shock [19, 22]. It has been shown that antibodies to TNF-a protects experimental animals from the lethal toxicity of LPS [54, 55] and recombinant TNF-a mimics many pathological events seen in sepsis [56, 57]. TNF-a is subject to complex mechanisms of up and down regulation by natural-killer (NK) and T-cell derived cytokines. In this context, IFN-g and TNF-a act synergistically by activating macrophages, while IL-4 and IL-10 inhibit macrophage activation and the subsequent production of cytokines [58-64]. Interestingly, monocytes are also a source of IL-10, exerting an autoregulatory role [61].

     Other cytokines like IL-1 and IL-6 are also induced by LPS and, together with TNF-a, are referred to as pro-inflammatory cytokines. There is considerable overlapping of biological activities between TNF and IL-1 [65].

IId. Inflammatory Response in Experimental and Clinical Sepsis

     LPS-induced cellular activation may be modulated in vivo, and hyperresponse and hyporesponse (tolerance) may be induced in a susceptible host [22]. The lethal effect of endotoxin in experimental animals can be enhanced under a variety of conditions, such as infections [66] or the presence of growing tumors [67]. Under these conditions hypersensitive animals produce enhanced amounts of cytokines (TNF-a) upon LPS challenge. The hypersensitivity to LPS induced by infection is mediated by IFN-g [22, 68, 69]. Sensitization to endotoxin may also be induced by a number of hepatotoxic agents, such as D-galactosamine (D-GaIN) [70]. On the other hand, minute amounts of LPS render the animals hyporesponsive to a subsequent LPS challenge, a phenomenon known as tolerance [71].

     Thus, the cellular response to LPS, and possibly the inflammatory response to other bacterial products, in experimental and clinical sepsis may be anticipated to be distinct from the normal host.

     Earlier studies in septic patients have found increased circulating levels of inflammatory cytokines, such as TNF-a, which was also associated with poor outcomes, supporting a role for these cytokines also in clinical setting. This was clear in meningococcal disease [72] and leptospirosis [73]. On the other hand, other observations have failed to correlate the TNF-a plasma levels with poor outcomes and, sometimes, have shown to be related to a better prognosis in septic patients [74-75].

     Induction of cytokines by LPS and killed whole bacteria was evaluated in experimental and clinical sepsis. As commented above and reviewed elsewhere [76] bacteria sensitizes animals to a challenge with LPS, a mechanism found to be dependent on IFN-g and LBP. Cellular response in septic patients has been usually evaluated using peripheral blood mononuclear cells (PBMC) or whole blood. The use of whole blood is very attracting because it preserves the milieu of LPS-cell interaction. Suppression of LPS-induced inflammatory cytokines in whole blood from septic patients has been demonstrated [39-40]. We confirmed the suppressed LPS-induced TNF-a production and found a preserved LPS-induced IL-10 production in whole blood from septic patients (Rigato & Salomão, manuscript in preparation).

     These findings suggest that the host is counter-regulating the inflammatory response. However, it is important to consider that this response may be distinct for different stimuli and may depend also on the source of host cells. It is likely that there is a differential cellular adaptation for LPS and other bacterial products. Thus, Mitov et al have found a suppressed LPS-induced TNF-a and an enhanced response to killed gram-negative and gram-positive bacteria in whole blood from septic patients [40]. Supporting differential regulation of cellular response to LPS and other bacterial products is the induction of TNF-a by gram-positive and gram-negative bacteria in LPS-resistant mice [22]. More recently, it has been shown that inactivated whole gram-negative bacteria, but not LPS, may induce TNF-a in CD14-/-macrophage [77]. Regarding the cell source one may consider that peripheral blood or PBMC may be not representative of the distinct cellular compartments of the host. Even in peripheral blood different cell types may have diverse adaptive response. It is known that LPS interacts with neutrophils and that these cells are very sensitive to LPS stimulation [78-79]. Some studies have found that the oxidative burst is down regulated in neutrophils from septic patients [80-82]. In our hands, however, bacterial phagocytosis and LPS-induced oxidative burst was enhanced in neutrophils from patients with severe sepsis and septic shock (PS Martins et al, manuscript in preparation).

     Down regulation of LPS-induced cellular activation has been extensively studied in experimental models of LPS-induced tolerance. Little is known about the underlying mechanisms leading to hyporesponse in patients with sepsis. We used biotinilated LPS and flow cytometry to evaluate LPS binding and cellular activation in whole blood of septic patients. (R Salomao et al, unpublished observation). Our results indicate that the detection of intracellular TNF-a in monocytes from septic patients is lower than in healthy controls, despite LPS binding and internalization are not impaired.

     Two other points should be considered in the cellular response to LPS in the clinical setting: the genetic background and the presence of underlying disease. It has been recently shown that TNF-a genomic polymorphism may be related to LPS-response, septic shock susceptibility and mortality. A polymorphism located at nucleotide position –308, which resulted in 2 allelic forms TNF-1 and TNF-2, was shown to be related to TNF-a production. In a study with severe septic patients TNF-2 was associated with increased TNF-a plasma levels and higher mortality [83]. A case-control study have shown that TNF-2 was also associated with increased susceptibility to septic shock and to higher mortality [84], but no differences in TNF-a levels was observed. Despite these evidences, the precise role of this polymorphism and linked disequilibrium with other immune response genes is still not clear [85].

     Directly or mediated by cytokines, LPS can further induce the release by other target cells of mediators such as the arachidonic acid metabolites, pro-coagulant factors, nitric oxide (NO) and oxygen intermediate metabolites. This cascade has interdependent pathways and complex regulatory mechanisms [86, 20].

IIe. Activation of the Coagulation Cascade in Sepsis

     The coagulation system is activated by LPS through the intrinsic and extrinsic pathways [20]. Simultaneously, there is a depression of inhibitory systems. These two mechanisms increase fibrin deposition and can be responsible for multiple organ dysfunction.

     Several mediators have been reported as potential inducers of coagulation disturbances. For instance, proinflammatory cytokines, such as TNF-a, IL-1 and IL-6 can activate secondary mediators such as the coagulation, fibrinolytic, and contact systems. Conversely, thrombin can stimulate the inflammatory cytokine production. In addition, thrombin promotes platelet aggregation throughout the microcirculation and activated platelets to express P-selectin which is responsible for leukocyte-endothelial cell rolling. Finally, activated neutrophils, in turn, release elastase, which destroys anti-thrombin III (AT III), an important endogenous anticoagulant.

     Data from experimental septic models have demonstrated the critical role of TF as responsible for initiating the extrinsic pathway of coagulation [87, 88]. This molecule is constitutively expressed by different cell types and is also induced in monocytes and endothelial cells. TF is a highly thrombogenic molecule so that a very small amount is able to generate clotting activation. Antibodies blocking TF fully neutralized endotoxin clotting activation [89].

     The source of TF in sepsis is not clear. More likely, circulating blood cells provide the main source as demonstrated in volunteers challenged with endotoxin. Monocytes would probably be these cells, but granulocytes have been also investigated. Other possibility is that TF would be expressed by endothelial cells.

     TF forms a catalytic complex with factor VII (a), which cleaves factors IX and X (TF/FVIIa/FXa complex), activating the common coagulation pathway. In turn, factor X (a) converts prothrombin to thrombin. The recent formed thrombin activates factor V, which becomes a potent cofactor for factor X (a) and greatly enhances the conversion of prothrombin to thrombin by factor X (a). Thrombin generation by TF mediated coagulation pathways results in the formation of fibrin clots, endothelial activation to produce proinflammatory mediators, and microvascular thrombosis [90].

     Physiologically, the generation of procoagulant mediators is counteracted by anticoagulant systems. The main endogenous anticoagulants are anti-thrombin III, protein C and tissue factor pathway inhibitor (TFPI). These proteins modulate the clotting formation, attenuating tissue perfusion impairment.

     The TF mechanism is inhibited by the natural anticoagulant, the TFPI. TFPI downregulates the TF/FVIIa/FXa complex formation decreasing the thrombin generation. However, TF/FVIIa also activates factor IX, and together with factor VIII (a), formed by traces of thrombin, takes over the function of TF/FVIIa to activate additional factor X and propagate thrombin generation. This amplification of TF/FVIIa-triggered factor X activation is essential to allow for effective clotting and normal haemostasis.

     The Protein C is one of the primary anticoagulation pathways initiated with the binding of thrombin to thrombomodulin. The thrombomodulin-thrombin complex activates protein C, which in turn inactivates factors Va and VIIIa on the phospholipid surface. Further, binding of thrombin to thrombomodulin prevents thrombin from activating fibrinogen or factors V and VIII. The extent of thrombomodulin-thrombin binding is dependent on local release of inflammatory cytokines [91]. Protein C is a vitamin K-dependent protein synthesized in the liver and released into the plasma. Activated protein C has a half-life of 20 minutes and is cleared by the liver or neutralized by complexing with several protein C inhibitors, plasminogen activator inhibitor-3, alpha-1-plasminogen inhibitor, and alpha-2-macroglobulin. Activated protein C is also known to enhance clot lysis by neutralizing plasminogen activator inhibitor-1 (PAI-1), and decrease local inflammatory-induced coagulation by inhibiting monocyte production of tumor necrosis factor-alpha and E-selectin-mediated leukocyte adhesion to injured endothelial cells [92].

     As mentioned, AT III is the major inhibitor of most of the serine proteases generated during the coagulation process, which include kallikrein, XIa, Xa, IXa, and IIa. AT III complexes are cleared from the circulation by the liver. Clot-bound serine proteases, especially factor Xa, are protected from inactivation by AT III. AT III complexing with serine proteases is increased several thousand fold by exogenous heparin or heparan sulfate-bearing proteoglycans in vessel walls. AT III-heparin complex also neutralizes factor VIIa when it is bound to TF. However, the most powerful inhibitory effect of AT III is against thrombin and factor Xa. AT III levels are rapidly depleted in early severe sepsis and is related to outcome [93]. This depletion is related to an acute consumption, reflected by increase in circulating thrombin-antithrombin complexes observed in some studies. Other secondary and less proven mechanisms implicated are elastase inactivation, decreasing in hepatic synthesis and secondary to vascular leak phenomenon. Fig. (1) illustrates the inhibitory mechanisms which are activated after coagulation cascade has begun [94].

 

    

 In conclusion, a considerable progress has been made in the understanding of mechanisms regulating cellular response to LPS and other bacterial products. There is still much to be done at the interface between basic research findings and clinical practice. By far the knowledge on pathogenesis of sepsis have an encouraging perspective for the development of pathogenesis-oriented target adjunctive therapy.

III. PATHOGENESIS-ORIENTED TARGETS FOR ADJUNCTIVE THERAPY

     Adjunctive therapy is aimed both, at interfering with the ongoing pathophysiological response to infection or at preventing its development. Results of preliminary investigations are encouraging, although the clinical results are still conflicting. Several strategies have been applied as adjunctive therapy: neutralizing LPS/lipid A, blocking its binding to macrophages and inhibiting the release or action of cytokines, such as TNF-a and IL-1 and cytokine-induced mediators thereof, such as NO. The better understanding of the coagulation dysfunction along with its influence on outcomes have prompted investigators to conduct clinical trials to interfere with the coagulation cascade in sepsis.

IIIa. Therapies Directed to Interfere with LPS-Cell Interaction

     The lipid A/inner core region of the LPS molecule is antigenically preserved among many pathogenic gram-negative bacteria and is responsible for endotoxic activity [95]. The prospects of preparing antibodies directed against common structures, that were capable of neutralizing the toxic effects of LPS of serologically unrelated gram-negative bacteria, was very exciting [19].

     Antibodies to lipid A, obtained for the first time in 1971 [96], were found to exhibit a wide cross-reactivity with the free lipid A of unrelated bacteria [95, 97]. However, if these antibodies were able to recognize and bind to the lipid A/inner core as a part of the intact LPS molecule, thus protecting experimental animals against LPS toxicity and bacterial infections, or if the protection was related to other mechanisms triggered by immunization remains still controversial.

     Promising results were initially obtained with anti-lipid A/core policlonal antibodies. In one prospective, randomized, double-blind study, the use of anti-E. coli J5 antibodies resulted in a 37% reduction on mortality in patients with gram-negative bacteremia compared to placebo [98]. However, these results were not confirmed in a larger clinical trial [99]. The same was true for anti-lipid A monoclonal antibodies. A protective effect was demonstrated in two clinical trials [100], but further studies failed to confirm this benefit [101].

     LPS toxicity may be further inhibited by antagonizing effect of LBP or by blocking the CD14 receptor. Some experimental studies have shown that monoclonal antibodies anti-CD14 impair the binding of LPS-LBP complex to this receptor, decreasing the binding of LPS to macrophages, neutrophils and even endothelial cells [102, 103]. Antibodies to LBP were reported to have a protective effect against the toxic effects of LPS and lipid A in animal models. This effect is probably due to a significant decrease in TNF-a production [104].

     Besides neutralizing LPS, BPI is cytotoxic against gram-negative bacteria. Injection of rBPI₂₃ protected mice from LPS-induced tissue damage and death (93% survival in rBPI₂₃ group vs 13% in control group), and reduced LPS-induced TNFa and IL-1 production [105]. Although there are no clinical studies in humans conducted so far, the rBPI₂₃ infusion was related to better outcome in baboon sepsis model. The group that received rBPI₂₃ presented significant improvement in organ dysfunction (liver, kidney and adrenal) and survival (37% vs.100%). Interestingly, rBPI₂₃ infusion reduced bacteremia and circulating TNFa without reducing LPS levels [106].

IIIb. Blocking Inflammatory Cytokines

     Corticosteroids are potent inhibitors of TNFa production, both at the pre- and post-transcriptional level. Although corticosteroids have protective effect in experimental studies, prospective, double-blind clinical trials using high-doses of corticosteroids did not improve outcome in septic patients [107]. A possible explanation for these divergent findings is the timing of corticosteroid administration. Experimentally, protective effects are seen when corticosteroids are injected before or concomitantly to LPS. In the clinical setting, septic patients receive corticosteroid administration when inflammatory cascade has already been triggered. Galanos et al [108] found that dexamethasone afford complete protection against the toxic effects of LPS in D-GalN sensitized mice only when it was administered before LPS. No protection was seen when corticosteroid is injected 15 minutes or longer after LPS. Pre-treatment with dexamethasone affords protection against LPS by inhibiting TNFa production, but has no protective effect against TNFa toxicity. Recently, clinical trials enrolling a small number of patients have reevaluated the effect of low doses of corticosteroids in septic patients. Even without improvement in survival, some beneficial effects were observed, such as earlier discharge of vasoactive drugs in patients that survived [109, 110].

     Beside corticosteroids, pentoxifylline, a methylxanthine derivative, is another important cytokine inhibitor that acts mainly in a pre-transcriptional level [111]. Pentoxifylline has been used in experimental studies in the last two decades. It was shown to be a promising therapy in a well-designed clinical trial with 100 prematurely delivered infants with sepsis confirmed by positive blood culture. One of 40 infants died in pentoxifylline group, whereas 6 of 38 infants died in placebo group (p= 0,046). A significant increase in organ dysfunction was seen in placebo group [112].

     Several studies have been performed in order to block the effects of TNF in severe septic patients using more specific antagonists. Two anti-TNF strategies have been evaluated in those trials: anti-TNF antibody (monoclonal and polyclonal) and TNF soluble receptor.

     The efficacy of anti-TNF monoclonal antibody (TNF-a Mab) was evaluated in a phase III randomized clinical trial (NORASEPT I) [113]. This study enrolled 994 severe septic (516) and septic shock (478) patients. The patients were randomized to receive a single infusion of TNF-a Mab or placebo. There was no difference in all-cause mortality among patients who received placebo as compared with those who received TNF-a Mab. In septic shock patients, there was a trend toward a reduction in mortality rate.

     Six months after the start of NORASEPT, another large phase III clinical trial called INTERSEPT (International Sepsis Trial) was begun [114]. This study enrolled 564 patients from 40 centers in 14 countries. Eleven patients did not receive the study drug for different reasons. Moab was given as a 15-mg/kg dose to 205 (37.1%) patients and as a 3-mg/kg dose to 181 (32.7%) patients, while 167 (30.2%) patients received placebo. The overall mortality rate was 39.5% in the placebo group, compared to 31.5% in the 3 mg/kg group and 42.4% in the 15 mg/kg group. Neither of these changes were statistically significant. There were no significant differences between the survival rate curves of the three groups (log-rank analysis 3 mg/kg vs. placebo, p = .19).

     In 1998, the NORASEPT II [115] was published. This study enrolled 1900 patients only with septic shock, based on the previous NORASEPT I results. No improvement in survival was found in septic shock patients treated with the TNF-a Mab compared to those who received placebo.

     In 1996, Reinhart et al. [116] had published data from another clinical trial involving 122 patients with severe sepsis and septic shock. The author also utilized TNF-a Mab (MAK 195F). Patients received three different doses and placebo, with no significant mortality rate variation in those groups. However, increased circulating IL-6 concentrations (> 1000 pg/ml) were found to be an important prognostic factor. In these patients, MAK 195F appeared to be benefit in a dose-dependent fashion.

     These results elicited a recent large clinical trial, designed to assess the efficacy and safety of a Moab anti-TNF-a (afelimomab) compared to placebo in patients with severe sepsis and septic shock with highly elevated IL-6 levels (IL-6 _ 1000 pg/mL) called “positive SEPTEST” [117]. In 2634 patients enrolled for this study, 998 had a positive SEPTEST and were randomized to receive afelimomab (n=498) and placebo (n=510). The 28-day mortality rate was 43.6% in afelimomab group and 47.6% in placebo group (p < 0.05). By the manuscript sending, this data has not been published yet.

     As mentioned, the effects of TNF-a may be blocked by soluble TNF receptor infusion. Three clinical trials have addressed this potential therapy. In the first of these studies, the molecule used consisted of the extramembrane components of the human type II (p75) receptor joined to the Fc portion of a human IgG1 antibody molecule [118]. Patients (n = 141) with septic shock were randomized to receive either increasing doses of p75 or placebo. There was a dose dependent increase in mortality rate (from 30% in the placebo group to 53% in the highest dose group, p = 0.014). The other two clinical trials have examined the role of a p55 TNF receptor fusion protein construct in septic patients. In the first trial, 498 patients were randomized considering the presence of severe sepsis with or without early septic shock and refractory septic shock [119]. The doses of the p55 TNF receptor complex used in this study were substantially lower than those administered in the p75 TNF receptor complex clinical trial. Reduction of mortality rate (36%; p = 0.07) was observed in patients with severe sepsis and early septic shock. Despite this apparent benefit, a 1340 patients trial did not confirm this observation (unpublished data). The protein used in this trial has a lower TNF neutralizing effect than that used in the former one. It is not clear whether this change was responsible for the worst results.

     The interleukin-1 receptor antagonist (IL-1ra) is a naturally occurring inhibitor of IL-1, which competitively binds to IL-1 receptor [120]. Data from experimental studies show that the infusion of IL-1ra decrease mortality rate even when administered shortly after the endotoxin challenge [121].

     In 1994, Fisher et al published a Phase II trial in 99 septic patients which demonstrated a significant mortality rate reduction [122]. Unfortunately, two large subsequent phase III trials did not confirm these preliminary data [123]. The most important study was published 3 years later [124] in which 696 patients with severe sepsis and septic shock were randomized to receive a 72-hours IL-1ra infusion versus placebo. The 28-day all-cause mortality rate was 33.1% in the rhIL-1ra treatment group, while the mortality rate in the placebo group was 36.4%.

     Some speculations tried to explain why IL-1ra has not altered the septic patient outcome. First, IL-1 would not play a pivotal role in perpetuating the inflammatory response. Second, the timing of administration may not have been early enough to block the initial inflammatory response. Third, few septic patients have elevated plasma levels of IL-1 and the actual tissue expression of IL-1 is not known. In summary, it is necessary to accumulate more clinical experience with IL-1ra showing a protective role in septic patients before its routine clinical use.

IIIc. Therapy Directed to LPS and Cytokine-Induced Systemic Effects

     NO, a potent vasodilator, plays a pivotal role in septic shock. It is produced by NO synthase (NOS), which is found in constitutive (type I and type III) and inducible (type II) isoforms. Constitutive isoforms (cNOS) are expressed in neuronal (type 1) or endothelial (type III) cells and inducible isoforms (iNOS) are expressed in many cells, including macrophages [125]. While cNOS are responsible for the reactions of organic homeostasis, iNOS is involved in situations of stress, such as those induced by LPS or cytokines [126]. Non-selective inhibition of NOS appears to explain the lack of beneficial effects seen in experimental studies using NO synthase inhibitors [127]. One inhibitor, N-monomethyl-L-arginin (L-NMMA), has been used in clinical trials. In 12 patients with sepsis, LNMMA induced a significant increase in mean arterial pressure and systemic and pulmonary resistance, but also caused a reduction in cardiac output and tissue oxygen uptake [128]. More recently, similar results were seen with the use of another non-selective NOS inhibitor, NG-nitro-L-arginine methyl ester (L-NAME), in a continuous infusion in 11 septic patients [129]. In a phase II study using the non-selective NOS inhibitor 546C88, Grover et al. found improvement in the mean arterial pressure and reduction or elimination of the requirement for norepinephfrine therapy in 32 septic shock patients [130]. Although small clinical trials have demonstrated benefic hemodynamic effects of non-selective NOS inhibitors in septic patients, experimental studies have shown better effects with the use of selective iNOS inhibitors. S-methylisothiourea (SMT) and aminoguanidine, two inhibitors of NOS with selectivity of iNOS, prevented rats from pulmonar transvascular flux after injection of LPS [131]. Prolonged survival of rats infected with E. coli was observed with SMT, while a high dose of L-NAME, an inhibitor of constitutive and inducible forms of NOS, significantly shortened survival [132]. Enhanced lethality with L-NAME was also observed in mice injected intraperitoneally with E. coli [133]. However, even S-methylisothiourea and 2-amino-5,6dihydro-6-methyl-4H-1,3-thiazine, two iNOS inhibitors, were shown to increase lethality of rats exposed to LPS in one study [134]. Pretreatment with L-arginine could prevent mortality without affecting the iNOS-dependent NO production, thus suggesting that toxicity was due to inhibition of other NOS isoforms (neuronal or endothelial). It has been suggested that INOS inhibitors of greater selectivity may be needed for therapeutic use [134]. Unfortunately, a large phase III study was interrupted due to higher mortality rate in treatment group. The complete data has not been published yet.

     Because there is a link between systemic inflammation and disseminated coagulation, increased attention is being focused on a new and promising target – the interface between these two pathophysiologic mechanisms in severe ill septic patients. Some substances have been addressed in clinical trials, particularly tissue factor pathway inhibitor (TFPI), protein C and antithrombin III replacement.

     Four different inhibitors of the TF/FVII complex have been used in animal and human models for sepsis including a) TFPI, b) anti-TF Moab, c) anti-FVII monoclonal antibody and d) DEGR-FVIIa, a form of FVIIa blocked in its active center with a chloromethyl ketone. Most studies of TF-pathway inhibitors have been done with TFPI. The recombinant protein, referred to as rTFPI, has been used in severe sepsis animal models [135, 136]. The rTFPI administration even after infectious challenge has been associated with improvement in coagulation disturbances and lower mortality rate. Also, this treatment reduces circulating inflammatory cytokine levels. Studies of TFPI or rTFPI were conducted in healthy volunteers challenged with endotoxin. rTFPI infusion induced a dose-dependent attenuation of thrombin generation as assessed by plasma thrombin-antithrombin and F₁₊₂ complexes. Cytokine levels were not significantly altered by rTFPI [137]. Several small phase I and II studies have investigated the use of rTFPI in normal humans and septic patients. The most important results were demonstrated in the phase II study which involved 210 patients with severe sepsis [138]. The treatment and placebo arms did not differ appreciably with respect to all adverse events or those related to bleeding. There was a trend toward reduction in all causes of mortality in the rTFPI-treated group. These results suggest that rTFPI is of benefit in severely ill, septic patients and have led to the initiation of a large international phase III study of rTFPI for this indication.

     Activated protein C has been administered in patients with a wide range of infections. The promising results are coming from the studies that have evaluated the protein C replacement in purpura fulminans secondary to meningococcemia [139] or other sepsis [140]. A recent observational study in 12 patients with severe meningococcal sepsis showed that no deaths occurred in protein C supplementation group.

     A prospective, randomized, double-blind, placebo-controlled phase II clinical trial evaluated the impact of recombinant human activated protein C in patients with severe sepsis, showing a trend of mortality reduction in the high-dose treatment group (p non-significant). It was associated with a trend toward decreased time on the ventilator, in the intensive care unit and in the hospital, and also associated with lower levels of D-dimers and a larger decrease in IL-6 plasma levels.

     Recently, a large multicenter phase III trial was interrupted because there was a 19.5% relative risk reduction in mortality (p = 0.0054) in favor of rhAPC group. This study enrolled 1690 patients with severe sepsis and septic shock and 850 patients were included in the treatment arm. Three possible pathophysiologic mechanisms have been attributed to explain these results such as clot formation reduction, endogenous fybrinolysis restoration and anti-inflammatory properties [141].

     In experimental models, AT III replacement has decreased mortality both as pretreatment as well as after established sepsis [142]. These preliminary results have motivated some researchers to design clinical trials involving AT III substitution in patients with severe sepsis/septic shock.

     Blauhut et al [143] conducted the first controlled study which enrolled 51 septic patients with disseminated intravascular coagulation (DIC). Among three different therapeutic groups (AT III alone, heparin alone, and a combination of two), there was a shortness of DIC duration only in the first group. However, the mortality rate was not affected by the AT III substitution. In this study [143] and in another Blauhut’s study [144] the association between AT III and heparin was correlated to a higher rate of hemorrhagic complications.

     Inthorn et al [145] have shown that AT III supplementation in severe sepsis is related to systemic inflammatory response attenuation. The authors observed that the plasma concentration of IL-6 decreased and AT III administration prevented the continuous increase in circulating soluble adhesion molecules. This paper advocates that a continuous long-term AT III supplementation can modulate the uncontrolled inflammatory response in sepsis. Despite these promising results on anti-inflammatory properties, there is no further documentation that this mechanism could be responsible for organ dysfunction reversing.

 

    

 The larger study on ATIII replacement was conducted by Baudo et al [146]. This study enrolled 120 patients with sepsis and/or postsurgical complications with an ATIII concentrations lower than 70%. The overall mortality rate was not affected by ATIII infusion.

     Finally, Eisele et al [147] published a randomized, placebo-controlled, double-blind, phase II multicenter clinical trial plus a meta-analysis on all randomized clinical trials with AT III in severe sepsis. They found a mortality rate reduction of 39% in patients treated with AT III (Fig. (2)). These results supported a phase III clinical trial.

     More recently, in the European Intensive Care Congress (2000), the preliminary results of AT III phase III clinical trial were reported. Unfortunately, the overall mortality rate was not affected by AT III replacement. In fact, only a small group of patients with high SAPS II score and with no heparin treatment presented a significant reduction in mortality rate. As in the Blauhut’s study, there was an association between ATIII + heparin and hemorrhagic events, what may have blunted a better result with ATIII substitution.

     Among the mediators involved in the pathogenesis of sepsis, there is a class of molecules derived from the metabolism of membrane phospholipids. The platelet-activating factor (PAF) and a vast array of eicosanoids, which are derived from the metabolism of arachidonic acid, are examples of this class.

     PAF is produced by induction of phospholipase A2 by a broad range of cells including platelets, PMNs, macrophages, endothelial and epithelial cells. A wide variety of mediators stimulate these cells to produce PAF, specially proinflammatory cytokines, thrombin and bradykinin. PAF seems to act by specific membrane receptors as a potent inflammatory and hypotensive agent. For these reasons, many experimental and clinical studies have been designed to block its action in sepsis setting.

     PAF has been shown to enhance the macrophages response to LPS increasing TNF and TF production as well as to enhance IL-1 beta production by human monocytes [148,149]. These responses have been shown to be inhibited by PAF antagonists. In a animal model, PAF infusion led to a shock weakly responsive to fluid replacement [150]. Experimental studies involving PAF antagonists have had conflicting results, showing improvement in hemodynamic profile and in survival, while others have failed to show any difference [150].

     Two prospective, randomized, placebo-controlled trials of PAF inhibition in sepsis have been published. The first involved 132 patients; despite a trend toward improved survival, with a 42% mortality rate in the treatment group vs. 51% for the control group, the difference did not reach statistical significance. The subgroup of patients with only Gram-negative sepsis did have a 42% reduction in mortality [151]. The second trial involved only 12 patients and failed to show a difference in 26-day or 58-day mortality [152]. The timing of PAF inhibition may be a key factor in this approach as it may be most effective in preventing the priming of inflammatory cells and, thus, would have to be given very early in the clinical course to be effective.

     A newly developed PAF antagonist is the cloned human PAF-acetylhydrolase (PAF-AH) which is responsible for endogenous metabolism of PAF. A human study [153] has demonstrated an association between decreased circulating levels of PAF-AH and the development of multiple organ failure syndrome (MOFS) in trauma patients. A phase II clinical trial in patients at risk for MOFS has demonstrated safety of the drug and improvements in MOFS scores (Pribble et al., 6^th International Congress on PAF and related lipid mediators, New Orleans, 1998).

     In conclusion, our understanding of the pathogenesis of sepsis has been greatly improved. In many aspects the research on endotoxin has contributed to this knowledge. The chemical characterization of LPS and its toxic moiety, the lipid A, and their biological activities; the identification of mediator(s) of endotoxic shock, such as TNF-a, followed by the recognition of their dual role in infection, as well as the complex mechanisms they are regulated; and an increasing understanding of LPS recognition and mechanisms of cell activation, are some breakthroughs in this field. Pathogenic-target oriented adjunctive therapy in sepsis moves from optimism to criticism. Very often, promising results of phase I and II clinical trials were not confirmed in phase III trials, which sometimes have generated conflicting results. Many issues have been raised to explain these disappointing data. First, the product used would not be able to fully block the mediator of interest, or this mediator would not have an important role in the sepsis pathogenesis. Considering, for example, the complex network of cytokines and the overlapping of biological activities, blocking of one such cytokine may not achieve the expected biological result. Moreover, in the complex pathogenesis of sepsis, in which healing and deleterious response are part of the same process, there is a lack of surrogate markers for the best selection of patients. Second, these products achieved success in animal models but not in clinical setting. Further, clinical sepsis is a very heterogeneous setting as is the population it affects. The challenge to find the better strategies is intimately related to the progress of knowledge of pathogenesis and improvement of clinical trials. As our comprehension of the host and infecting microorganisms interaction improves new strategies for adjunctive therapy are envisaged as some are revisited in new approaches.

Acknowledgement

     Dedicated to M. Freudenberg and C. Galanos in admiration and friendship. This work was supported by the Ministry of Science and Technology of Brazil, PRONEX 41.96.0943.00, and Conselho Nacional de Desenvolvimento Científico e Tecnológico, Grant 524088 / 96-9.

Reference

[1]  Quartin, A.A.; Schein, R.M.; Kett, D.H.; Peduzzi, P.N. JAMA, 1997, 277, 1058-1063.

[2]  McGowan, J.E.; Barnes, M.W.; Finland, M. J. Infect. Dis., 1975, 132 , 316-335.

[3]  McCabe, W.R.; Jackson, G.G. Arch. Intern. Med., 1962, 110, 847-855.

[4]  Kreger, B.E.; Craven, D.E.; McCabe, W.R. Am. J. Med., 1980, 68, 344-355.

[5]  Schaberg, D.R.; Culver, D.H.; Gaynes, R.P. Am. J. Med., 1991, 91, 72S-75S.

[6]  Weinstein, M.P.; Reller, L.B.; Murphy, J.R.; Lichtenstein, K.A. Rev. Infect. Dis., 1983, 5, 35-53.

[7]  Haug, J.B.; Harthug, S.; Kalager, T.; Digranes, A.; Solberg, C.O. Clin. Infect. Dis., 1994, 19, 246-256.

[8]  Pittet, D.; Wenzel, R.P. Arch. Intern. Med., 1995, 155, 1177-1184.

[9]  Pfaller, M.A. J. Hosp. Infect., 1995, 30 Suppl., 329-338.

[10]      Bone, R. Ann. Intern. Med., 1991, 115, 457-469.

[11]      Bone, R.; Balk, R.; Cerra, F.; Dellinger, R.; Fein, A.; Knaus, W.; Schein, R.; Sibbald, W. Chest, 1992, 101, 1644

[12]      Weinstein, M.P.; Murphy, J.R.; Reller, L.B.; Lichtenstein, K.A. Rev. Infect. Dis., 1983, 5, 54-70.

[13]      Bryan, C.S.; Hornung, C.A.; Reynolds, K.L.; Brenner, E.R. Am. J. Epidemiol., 1986, 123, 113-127.

[14]      Bryant, R.E.; Hood, A.F.; Hood, C.E.; Koenig, M.G. Arch. Intern. Med., 1971, 127, 120-128.

[15]      McCabe, W.R.; Jackson, G.G. Arch. Intern. Med., 1962, 110, 92-100.

[16]      Kreger, B.E.; Craven, D.E.; Carling, P.C.; McCabe, W.R. Am. J. Med., 1980, 68, 332-343.

[17]      Salomao, R.; Castelo, F.A.; Pignatari, A.C.; Wey, S.B. Rev. Paul. Med., 1993, 111, 456-461.

[18]      Friedman, G.; Silva, E.; Vincent, J.L. Crit. Care Med.,1998, 26, 2078-2086.

[19]      Salomao, R.; Rigato, O.; Pignatari, A.; Freudenberg, M.; Galanos, C. Infection, 1999, 27, 1-11.

[20]      Morrison, D.C.; Ulevitch, R.J. Am. J. Pathol., 1978, 93, 526-617.

[21]      Galanos, C.; Freudenberg, M.; Coumbos, A.; Matsuura, M.; Lehmann, V.; Bartholeyns, J. In: Tumor Necrosis Factor/Cachetin and Related Cytokines; Bonavida, B.; Gifford, G.E.; Kirchner, H.; Old, L.J. Eds.; Karger: Basel, 1988; pp 114-127.

[22]      Galanos, C.; Freudenberg, M.; Katschinski, T.; Salomao, R.; Mossmann, H.; Kumazawa, Y. In: Bacterial endotoxic lipopolysaccharides.; Ryan, J.; Morrison, D. Eds. CRC Press: Boca Raton, 1992; pp 75-104.

[23]      Schlichting, E.; Aspelin, T.; Lyberg, T. Scand. J. Clin. Lab. Invest., 1996, 56, 167-176.

[24]      Gegner, J.A.; Ulevitch, R.J.; Tobias, P.S. J. Biol. Chem., 1995, 270, 5320-5325.

[25]      Heinzelmann, M.; Miller, M.; Platz, A.; Gordon, L.E.; Herzig, D.O.; Polk, H.C. Ann. Surg., 1999, 229, 542-550.

[26]      West, M.A.; Clair, L.; Bellingham, J. J. Trauma., 1996, 41, 647-652.

[27]      Tobias, P.S.; Mathison, J.C.; Ulevitch, R.J. J. Biol. Chem., 1988, 263, 13479-13481.

[28]      Heumann, D.; Gallay, P.; Betz-Corradin, S.; Barras, C.; Baumgartner, J.D.; Glauser, M.P. J. Infect. Dis., 1993, 167, 1351-1357.

[29]      Gallay, P.; Jongeneel, C.V.; Barras, C.; Burnier, M.; Baumgartner, J.D.; Glauser, M.P.; Heumann, D. J. Immunol., 1993, 150, 5086-5093.

[30]      Hodge, S.; Hodge, G.; Flower, R.; Han, P. Scand. J. Immunol., 1999, 49, 548-553.

[31]      Thieblemont, N.; Wright, S.D. J. Exp. Med., 1997, 185, 2095-2100.

[32]      Detmers, P.A.; Thieblemont, N.; Vasselon, T.; Pironkova, R.; Miller, D.S.; Wright, S.D. J. Immunol., 1996, 157, 5589-5596.

[33]      Luchi, M.; Munford, R.S. J. Immunol., 1993, 151, 959-969.

[34]      Thieblemont, N.; Thieringer, R.; Wright, S.D. Immunity, 1998, 8, 771-777.

[35]      Blondin, C.; Le Dur, A.; Cholley, B.; Caroff, M.; Haeffner-Cavaillon, N. Eur. J. Immunol., 1997, 27, 3303-3309.

[36]      Tobias, P.S.; Tapping, R.I.; Gegner, J.A. Clin. Infect. Dis., 1999, 28, 476-481.

[37]      Strittmatter, W.; Weckesser, J.; Salimath, P.V.; Galanos, C. J. Bacteriol., 1983, 155, 153-158.

[38]      Lentschat, A.; El-Samalouti, V.T.; Schletter, J.; Kusumoto, S.; Brade, L.; Rietschel, E.T.; Gerdes, J.; Ernst, M.; Flad, H.; Ulmer, A.J. Infect. Immun., 1999, 67, 2515-2521.

[39]      Ertel, W.; Kremer, J.; Kenney, J.; Steckholzer, U.; Jarrar, D.; Trentz, O.; Schildberg, F. Blood, 1995, 85, 1341-1347.

[40]      Mitov, I.G.; Kropec, A.; Benzing, A.; Just, H.; Garotta, G.; Galanos, C.; Freudenberg, M. Infection, 1997, 25, 206-212.

[41]      Freudenberg, M.; Galanos, C. In: Bacterial endotoxic lipopolysaccharides; Ryan, J.; Morrison, D. Eds. CRC press: BocaRaton, 1992; pp 275-294.

[42]      Poltorak, A.; He, X.; Smirnova, I.; Liu, M.Y.; Huffel, C.V.; Du, X.; Birdwell, D.; Alejos, E.; Silva, M.; Galanos, C.; Freudenberg, M.; Ricciardi-Castagnoli, P.; Layton, B.; Beutler, B. Science, 1998, 282, 2085-2088.

[43]      Hoshino, K.; Takeuchi, O.; Kawai, T.; Sanjo, H.; Ogawa, T.; Takeda, Y.; Takeda, K.; Akira, S. J. Immunol., 1999, 162, 3749-3752.

[44]      Yang, R.B.; Mark, M.R.; Gray, A.; Huang, A.; Xie, M.H.; Zhang, M.; Goddard, A.; Wood, W.I.; Gurney, A.L.; Godowski, P.J. Nature, 1998, 395, 284-288.

[45]      Kirschning, C.J.; Wesche, H.; Merrill, A.T.; Rothe, M. J. Exp. Med., 1998, 188, 2091-2097.

[46]      Aderem, A.; Ulevitch, R.J. Nature, 2000, 406 (6797), 782.-7.

[47]      Hoffmann, J.A.; Kafatos, F.C.; Janeway, C.A.; Ezekowitz, R.A. Science, 1999, 284, 1313-1318.

[48]      O'Neill, L.A.; Greene, C. J. Leukoc. Biol., 1998, 63, 650-657.

[49]      Medzhitov, R.; Janeway, C.A. Cell, 1997, 91, 295-298.

[50]      Weinstein, S.L.; June, C.H.; DeFranco, A.L. J. Immunol., 1993, 151, 3829-3838.

[51]      Han, J.; Lee, J.D.; Tobias, P.S.; Ulevitch, R.J. J. Biol. Chem., 1993, 268, 25009-25014.

[52]      Hashimoto, S.; Gon, Y.; Matsumoto, K.; Maruoka, S.; Takeshita, I.; Hayashi, S.; Asai, Y.; Jibiki, I.; Machino, T.; Horie, T. J. Pharmacol. Exp. Ther., 2000, 293 (2), 370.-5.

[53]      Zu, Y.L.; Qi, J.; Gilchrist, A.; Fernandez, G.A.; Vazquez-Abad, D.; Kreutzer, D.L.; Huang, C.K.; Sha'afi, R.I. J. Immunol., 1998, 160, 1982-1989.

[54]      Beutler, B.; Cerami, A. Adv. Immunol., 1988, 42, 213-231.

[55]      Beutler, B.; Milsark, I.W.; Cerami, A.C. Science, 1985, 229, 869-871.

[56]      Tracey, K.J.; Beutler, B.; Lowry, S.F.; Merryweather, J.; Wolpe, S.; Milsark, I.W.; Hariri, R.J.; Fahey, T.J.; Zentella, A.; Albert, J.D. Science, 1986, 234, 470-474.

[57]      Lehmann, V.; Freudenberg, M.A.; Galanos, C. J. Exp. Med., 1987, 165, 657-663.

[58]      Collart, M.A.; Belin, D.; Vassalli, J.D.; de Kossodo, S.; Vassalli, P. J. Exp. Med., 1986, 164, 2113-2118.

[59]      Hart, P.H.; Vitti, G.F.; Burgess, D.R.; Whitty, G.A.; Piccoli, D.S.; Hamilton, J.AProc. Natl. Acad. Sci. USA, 1989, 86, 3803-3807.

[60]      Wong, H.L.; Lotze, M.T.; Wahl, L.M.; Wahl, S.M. J. Immunol., 1992, 148, 2118-2125.

[61]      de Waal, M.; Abrams, J.; Bennett, B.; Figdor, C.G.; de Vries, J.E. J. Exp. Med., 1991, 174, 1209-1220.

[62]      Fiorentino, D.F.; Zlotnik, A.; Vieira, P.; Mosmann, T.R.; Howard, M.; Moore, K.W.; O'Garra, A. J. Immunol., 1991, 146, 3444-3451.

[63]      Bogdan, C.; Vodovotz, Y.; Nathan, C. J. Exp. Med., 1991, 174, 1549-1555.

[64]      Sher, A.; Gazzinelli, R.T.; Oswald, I.P.; Clerici, M.; Kullberg, M.; Pearce, E.J.; Berzofsky, J.A.; Mosmann, T.R.; James, S.L.; Morse, H.C. Immunol. Ver., 1992, 127, 183-204.

[65]      Dinarello, C.A. Blood, 1991, 77, 1627-1652.

[66]      Matsuura, M.; Galanos, C. Infect. Immun., 1990, 58, 935-937.

[67]      Bartholeyns, J.; Freudenberg, M.; Galanos, C. Infect. Immun., 1987, 55, 2230-2233.

[68]      Galanos, C.; Freudenberg, M.A. Immunobiology, 1993, 187, 346-356.