Bacterial endotoxin, long recognized as a potent pro-inflammatory mediator in acute infectious processes, has more recently been identified as a risk factor for atherosclerosis and other forms of cardiovascular disease [^1-3]. In endotoxemia, one of the first cells to elicit a pro-inflammatory response is the circulating monocyte. Monocytes are known to respond to extremely low levels of endotoxin (pg/ml) [³], producing a wide array of chemokines and cytokines, and releasing cytotoxic levels of reactive oxygen species (ROS) via the respiratory burst. The resulting inflammatory process is initially beneficial to the host as a means of protection against invading microorganisms; however, if not resolved, chronic damage to host tissues can ensue. In this respect, a growing body of evidence suggests that the persistent low-level inflammation associated with chronic, subclinical infections (e.g., periodontitis, diverticulitis, or smoker's bronchitis) may also play a role in exacerbation of vascular diseases such as atherosclerosis [¹].

The classic endotoxin signaling pathway has been shown to involve LBP, CD14, MD-2, and Toll-like receptor 4 (TLR4)[^3-6]. It was commonly believed that all of these elements are required for initiation of endotoxin signaling[⁷], Although this pathway has been extensively documented in many model systems, there is evidence that alternate signaling pathways may exist in some cells. For example, endotoxin concentrations >100 ng/ml have been shown to activate host cells by mechanisms independent of the CD14-TLR4 pathway [^8-10] Moreover, rapid production of ROS in LPS-simulated (100 ng/mL) macrophages has been shown to be partly dependent on the activation of cytoplasmic GTPase Rac1, a known activator of NOX-1 oxidase enzyme activity[¹¹]. The above findings may be pertinent to mechanisms of LPS signaling in the setting of acute sepsis. In contrast, relatively little is known about potential alternative signaling pathways in humans activated by lower levels of endotoxin pertinent to chronic inflammatory diseases such as atherosclerosis.

To address this question, we designed experiments using freshly isolated human peripheral blood mononuclear cells (PBMC) to compare and contrast the effects of low levels of endotoxin (≤1 ng/ml) on two inflammatory responses, IL-8 release and superoxide production. Our data suggest that two distinct signaling pathways are involved in eliciting these responses. The presence of these two pathways in monocytes raises the possibility of targeting therapeutic interventions to modulate specific pro-inflammatory responses involved in chronic inflammatory diseases such as atherosclerosis.

Taken together, our results suggest that two important pro-inflammatory responses to low level endotoxin, IL-8 release and superoxide production, are activated by two distinct signaling pathways in monocytes freshly isolated from healthy human donors (Table 1). One is the rapid (minutes) endotoxin mediated oxidative burst leading to production of the reactive oxygen species superoxide anion by the NADPH oxidase complex. The other is release of the pro-inflammatory cytokine IL-8, involving activation of multiple transcription factors including NF-κB. We found that the endotoxin-mediated respiratory burst is LBP-dependent yet an anti-TLR4 antibody had no effect on the endotoxin-mediated respiratory burst as measured here. As inhibition by the CD14 blocking antibody was somewhat variable, there may a component of the activation pathway that is CD14 independent. Contrastingly, in our experimental conditions IL-8 release appears to be LBP-independent, CD14- and TLR4-dependent.

There is ample evidence for multiple endotoxin signaling pathways. Best characterized is the pathway in which LBP facilitates the transfer of endotoxin monomers to CD14, which in turn presents the endotoxin monomer to the MD-2/TLR4 complex [^28-32]. The transmembrane domain of TLR4 then transmits the signal via a series of adapter proteins and kinases (e.g., MyD88, IRAK, TRAF6, NIK, IκB, and NFκB) to the interior of the cell and ultimately to the nucleus, resulting in gene expression. However, a number of reports in recent years have suggested the existence of alternate signaling pathways[³³].

Studies using LBP knockout mice are consistent with a physiologically relevant LBP-independent inflammatory response. While whole blood from these animals was 1000-fold less responsive to endotoxin in vitro studies, nevertheless when the mice were injected with LPS, no significant differences between wild-type and LBP knockout mice (as measured by TNF release) were observed [³⁴]. In addition, responses to LPS can also be affected by LBP concentration and the presence or absence of other serum proteins[²⁰,^35-38]. Thompson et al. demonstrated in THP-1 cells that adding high-density lipoprotein (HDL) augmented the response to LPS in otherwise inhibitory LBP concentration conditions. Interactions may also take place between TLR4 and HMGB1 as indicated by Youn et al.[³⁹]. In our specific set of experimental circumstances, we found that LPS-induced IL-8 release by monocytes was LBP-independent (Table 1), although in repeated experiments, this finding was in contrast to results with differentiated macrophages from the same donor. This might be explained by the limiting stoichiometry and molar relationships between relative amounts of endotoxin, LBP, CD14 and TLR4[⁴⁰]. For instance, as the concentration of LPS increases, the ratio of LBP to LPS decreases, and activation increases. It is possible that in our monocyte isolation, inadequate CD14 or TLR4 expression may have limited cellular activation and IL-8 release. Furthermore, this would also explain why results from differentiated macrophages, with likely a greater expression of CD14, contrasted with the previous results in monocytes from the same donor. We also observed decreasing LPS-induced IL-8 release with decreasing albumin concentrations (data not shown), suggesting that the human serum albumin in our incubation buffer (and in plasma) was facilitating endotoxin-binding and delivery in our experimental model[²¹]. It is possible from what is known from our studies as well as others that serum proteins may variably contribute or inhibit LPS monomer delivery and hence LPS stimulatory activity, depending on the experimental and/or physiologic conditions.

To determine whether CD14 mediated LPS-induced IL-8 release or the respiratory burst in freshly isolated PBMC, we performed studies with the anti-CD14 blocking antibody MEM-18, which binds to an epitope on the endotoxin-binding site of CD14. MEM-18 completely inhibited IL-8 release; however, the ability of the antibody to block the respiratory burst was highly variable (ranging from approximately 10% to 100%) and appeared to be specific to individual donors (Table 1). Other investigators have reported evidence of CD14-independent signaling, particularly with relatively high (> 100 ng/ml) concentrations of endotoxin [^8-10]. Human and murine responses to LPS differ in that the concentrations of endotoxin challenge necessary to elicit responses in mice are much greater[⁴¹]. For instance, Maitra et al. demonstrated the contribution of IRAK-1, a known activator of NOX-1 enzymatic activity, in the rapid generation of ROS induced by LPS (100 ng/mL) in murine cells[¹¹]. Perera et al. found a CD14-independent signaling pathway in macrophages from CD14-knockout mice [⁴²]. In this model system, cytokine release in response to low doses of endotoxin (1 or 10 ng/ml) showed an absolute requirement for CD14 (either the membrane-bound or soluble form, as found in serum); however, at concentrations of 1 μg/ml or higher, endotoxin induced equal or greater production of TNF-α and IL-1β in CD14-knockout as compared to wild-type mice. Kimura et al. also found evidence of a CD14-independent endotoxin signaling pathway in a murine B cell line [⁴³]. In these experiments, IL-6 mRNA expression as well as ³H-thymidine incorporation as a measure of cellular proliferation were increased upon stimulation with endotoxin at 1 μg/ml even in the presence of anti-murine CD14 monoclonal antibody.

The potential complexity of endotoxin signaling is suggested by the model of a CD14 receptor complex proposed by Triantafilou et al. [⁸,⁴⁴,⁴⁵]. Using fluorescence recovery after photobleaching method (FRAP), these authors showed that endotoxin initially binds to CD14, a GPI-anchored receptor, before being transferred to other receptors such as MD-2. The MD-2:LPS complex associates with the TLR4 ectodomain for signal transduction[⁵,³²]. Interestingly, there is evidence to suggest that additional receptors may play a role in the transmembrane signal transduction. These include CD11/CD18 integrins, the CD55 receptor, and a receptor cluster consisting of four different molecules (Heat shock proteins 70 and 90, chemokine receptor 4, and growth differentiation factor 5).

One possible mechanism for the partially CD14-independent signaling we observed is through the leukocyte integrin transmembrane receptor CD11/CD18 family [^46-49]. There are three subtypes of this receptor. CD11a/CD18 participates in leukocyte-endothelial cell interaction and is found in all white cells. CD11b/CD18 is found mainly on monocytes and neutrophils and functions as complement receptor for C3b. The activity of this subgroup of receptors is enhanced in human PMN when exposed to endotoxin in a CD14-dependent manner [⁵⁰]. The CD11c/CD18 complex, though function is unclear, is found in many cells including monocytes and may be the main alternative to CD14 involved in the identification of endotoxin [⁵¹]. CHO cells transfected with CD11c/CD18 were able to recognize and respond to endotoxin independent of CD14 even at low concentration in the range of 1 ng/ml [⁴⁶]. However this is not a necessary condition since cells with mCD14 can release pro-inflammatory cytokines when exposed to endotoxin in the absence of CD18 receptors [⁵²]. To further support a potential contribution of CD11/CD18 receptors to CD14-independent signaling, we found decreased superoxide production in response to PBMC exposure to endotoxin only with the combined use of CD11b/CD18 and CD11c/CD18 antibodies, but not when used individually. Further studies are warranted to elucidate the contributory mechanisms of specific serum proteins and receptor subtypes in CD14-dependent and independent endotoxin signaling.

The endotoxin-CD14 complex initiates signaling by engaging TLR4 and the co-receptor MD-2 [⁵,³²,⁵³]. Cellular responses to endotoxin are partly determined by concentration and type of LPS (rough or smooth), length of exposure, as well as the affected cell type. Together these factors determine cell surface receptor expression, cytokine production, and ROS generation [⁵⁴,⁵⁵]. Recent evidence suggests that TLR4 plays an important role in atherogenesis [⁵⁶]. Specifically, the Asp299Gly polymorphism, which attenuates endotoxin signaling, is associated with a decreased risk of atherosclerosis [⁵⁷]. TLR4 function has been widely studied in monocytes and macrophages and was recently examined in human blood vessels [²⁷] and human coronary artery endothelial and smooth muscle cells [³] by our laboratory. Endotoxin stimulates ROS production in non-immune cells through a direct interaction between TLR4 and NOX4 [⁵⁸,⁵⁹]. However, we found that while LPS-induced IL-8 release by human monocytes was TLR4-dependent (i.e., blocked by HTA-125), superoxide formation was largely TLR4-independent (Table 1). This suggests that activation of the NADPH oxidase by LPS is mediated through alternative receptors in human monocytes. A potential candidate pathway includes the MAC1 receptor, which is expressed by monocytes [⁶⁰] and mediates endotoxin-dependent ROS formation in microglial cells [⁶¹]. Table 1 also summarizes the effects of lovastatin treatment. Our studies with lovastatin provide additional evidence for the anti-inflammatory effects of these compounds. Given the important role that monocyte activation is believed to play in acute coronary syndromes, our findings may also help to explain the significant reductions in coronary events observed in patients taking statins.

Taken together, our results suggest that two important pro-inflammatory responses to low concentrations of endotoxin, IL-8 release and superoxide production, are activated by two distinct signaling pathways in human PBMC. To our knowledge, this is the first study that evaluates the presence of alternative signaling pathways in freshly isolated human PBMC from healthy donors that are exposed to clinically relevant endotoxin levels.

The authors declare that they have no competing interests.

All contributors to this manuscript have made substantive intellectual contributions to this study. Each author has participated sufficiently in this work to take public responsibility for appropriate portions of the content. Each author has reviewed and given approval for this version to be published.

Specifically, AB prepared the manuscript for submission and participated in the analysis and interpretation of the data. LS conceived of the study, performed necessary assays, and wrote the first draft of the manuscript. WS coordinated the sample acquisition from human volunteers and drafting of the manuscript. SM carried out the assays and analyzed the data. ED analyzed data and helped draft the manuscript. NW participated in the design, coordination of the study and helped draft the manuscript. GD participated in the design of some experiments, analyzed data, prepared figures and assisted in the first draft of the manuscript.

This work was supported by grants HL49264 and HL62984 (NLW) from the National Institutes of Health, by a Merit Review grant from the Dept. of Veteran's Affairs (GMD), and by a Grant-in-Aid from the American Heart Association National Center (GMD). These funding agencies have had no role in the collection, analysis, or interpretation of data; in the writing of the manuscript; and in the decision to submit this article for publication.