- Top of page
- Experimental procedures
- RESULTS and DISCUSSION
Lipopolysaccharide (LPS) represents a major virulence factor of Gram-negative bacteria (‘endotoxin’) that can cause septic shock in mammals including man. The lipid anchor of LPS to the outer membrane, lipid A, has a peculiar chemical structure, harbours the ‘endotoxic principle’ of LPS and is responsible for the expression of pathophysiological effects. Chemically modified lipid A can be endotoxically inactive, but may express strong antagonistic activity against LPS, a property that can be utilized in antisepsis treatment. We show here that these different biological activities are directly correlated with the molecular shape of lipid A. Only (hexaacyl) lipid A with a conical/concave shape, the cross-section of the hydrophobic region being larger than that of the hydrophilic region, exhibited strong interleukin-6 (IL-6)-inducing capacity. Most strikingly, a correlation between a cylindrical molecular shape of lipid A and antagonistic activity was established: IL-6 induction by enterobacterial LPS was inhibited by cylindrically shaped lipid A except for compounds with reduced headgroup charge. The antagonistic action is interpreted by assuming that lipid A molecules intercalate into the cytoplasmic membrane of mononuclear cells, and subsequently blocking of the putative signaling protein by the lipid A with cylindrical shape.
Lipopolysaccharides (LPS, endotoxin) exert a variety of biological (endotoxic) activities in mammals including pathophysiological effects such as fever, tachycardia, tachypnoe, leukopenia, and hypotension, the hallmarks of sepsis and septic shock . The molecular mechanisms of these effects are unknown and comprise the activation of various host cells, in particular monocytes and macrophages, leading to the secretion of nitric oxide, vasoactive lipids, cytokines such as interleukin (IL)-1, IL-6, IL-12, and tumor necrosis factor α. Studies of the chemical prerequisites for endotoxic activity of LPS have revealed that the activation of monocytes/macrophages depends on a peculiar primary structure of lipid A present in enterobacterial genera . Thus, lipid A of the biologically most potent enterobacterial LPS consists of a β-1,6-linked d-glucosamine (GlcN) disaccharide carrying six saturated fatty acids and two negatively charged phosphates at defined locations. Variations of this structural arrangement such as a reduction of the number of charges or the number of acyl chains or a change in their distribution or degree of saturation results in a dramatic reduction in biological activity . These observations were interpreted as indicating an influence of a variation in the primary structure of endotoxin molecules on their physicochemical behavior [3,4].
One step towards an understanding of the molecular mechanisms of cellular activation (agonistic activity), but also of its inhibition (antagonistic activity), by endotoxins of different chemical structure is the correlation of their molecular properties with different activities. The shape of individual lipid A molecules has been deduced from their aggregate structures, and is either conical (with a larger cross-section of the hydrophobic region than of the hydrophilic part) in the case of nonlamellar inverted aggregate structures (cubic Q or hexagonal HII), or cylindrical (both cross-sections are identical) in the case of lamellar (L) structures . Experimental evidence for the crucial role of the molecular shape of endotoxin molecules in the expression of biological activity has been provided recently in a study that showed that only conical molecules exhibited endotoxicity, whereas cylindrical molecules were inactive .
One concept for therapeutic intervention in septic shock is the application of endotoxin antagonists [1,6–8]. The tetraacyl compound 406 (corresponding to lipid A precursor Ia or lipid IVa), the first synthetic endotoxin partial structure with antagonistic properties, does not induce cytokines in human monocytes and exhibits its antagonistic action only in the human system, and not in rodents , thus excluding animal trials. Other antagonistic compounds have been described, such as the pentaacyl lipid A of Rhodobacter capsulatus and Rhodobacter sphaeroides and their synthetic analogs which exert antagonistic activity in humans and rodents when delivered before or together with activating endotoxin [7,9,10]. The antagonistic activities of lipid A of other nonenterobacterial species with very low or even no biological activity, like those from Campylobacter jejuni and Rhodospirillum fulvum, have not yet been investigated.
RESULTS and DISCUSSION
- Top of page
- Experimental procedures
- RESULTS and DISCUSSION
We determined the molecular shapes of lipid A of various enterobacterial and nonenterobacterial species and correlated these with their biological activities (for chemical structures see Fig. 1). The molecular shapes of the individual lipid A molecules were determined from synchrotron radiation X-ray diffraction patterns of their supramolecular aggregates at physiological water content . As a typical indicator of biological activity (agonistic activity), lipid A-induced IL-6 production in whole blood was determined after stimulation ex vivo. Also, enterobacterial hexaacyl bisphosphoryl and monophosphoryl lipid A and lipid A from R. fulvum, for which respective data have been published previously , were included as standards and links to previously published data on agonistic lipid A. Furthermore, the antagonistic activities of agonistically inactive lipid A preparations, i.e. the inhibition of IL-6 production, were determined by adding them to human peripheral blood mononuclear cells prior to stimulation with agonistic enterobacterial S-form LPS of Salmonella enterica sv. Friedenau.
For hexaacyl lipid A from E. coli, an inverted cubic structure was found [4,19], whereas the pentaacyl and tetraacyl lipid A from E. coli formed multilamellar structures. For the latter, a slight tendency also towards a micellar structure was found. Lipid A from C. jejuni adopted a unilamellar structure with a slight tendency towards an inverted cubic structure. All other lipid A preparations exclusively formed multilamellar structures. From these data, the following conformational characteristics of the various lipid A molecules are deduced. The shape of the individual lipid A molecules are conical/concave for enterobacterial hexaacyl, predominantly cylindrical for pentaacyl, and cylindrical with a slight tendency towards conical/convex for tetraacyl lipid A (the cross-section of the hydrophobic being slightly smaller than that of the hydrophilic region). The basic molecular shape of the five investigated nonenterobacterial lipid A samples is cylindrical; for lipid A of C. jejuni, however, there is a slight tendency towards a conical/concave shape.
The IL-6 production in peripheral blood mononuclear cells of different lipid A samples at varying concentrations (Fig. 2) shows that hexaacyl lipid A from E. coli, although less active than the control LPS from smooth strain Salmonella enterica sv. Friedenau in accordance with literature data [4,20], is significantly more active than monophosphoryl lipid A and two to three orders more active than the other lipid A tested. As shown in Figs 2 and 3, the activities of the different lipid A samples differ by orders of magnitude. Therefore, the presence of partly acylated subfractions within an active lipid A sample as found by MALDI-TOF MS do not influence the results significantly.
Figure 2. Interleukin-6 production in human mononuclear cells by various lipid A and control LPS (S-form LPS from Salmonella enterica sv. Friedenau): dependence on endotoxin concentration. As measure of agonistic activity, the amount of endotoxin necessary to induce the production of 10 ng·mL−1 IL-6 was taken (horizontal line, see Fig. 3)
Download figure to PowerPoint
Figure 3. Agonistic and antagonistic activities and molecular shapes of various enterobacterial and nonenterobacterial lipid A. The value found for hexaacyl lipid A from E. coli (1 ng·mL−1 lipid A induce 6 ng·mL−1 IL-6) was set to 100% and the other lipid A were related to this standard (see Fig. 2). The antagonistic activity of agonistically inactive lipid A is expressed as percentage inhibition of IL-6 induction due to preincubated lipid A (100 ng·mL−1) 30 min before stimulation by 1 ng·mL−1 LPS (S-form LPS from S. enterica sv. Minnesota). The inhibition value obtained for lipid A from R. capsulatus (LPS-induced IL-6 production was reduced to 2.3 ± 1%) was set 100% and the data obtained with the other preparations related to this standard. Error bars are SD resulting from three independent experiments.
Download figure to PowerPoint
The agonistically inactive lipid A molecules were investigated with respect to their ability to block agonistic activity of S-form LPS. The results from the conformational study and the agonistic and antagonistic activities are summarized for all investigated lipid A in Fig. 3. The endotoxic, i.e. IL-6-inducing capacity of the lipid A samples is strikingly correlated with their tendency to adopt a particular molecular shape. Strong endotoxic activity was found only for hexaacyl lipid A from E. coli or from S. enterica spp., which exhibit a conical/concave shape. For all endotoxins that exhibit a cylindrical molecular shape, the biological activity was weak or absent. This was observed for enterobacterial structures such as the pentaacyl lipid A from the E. coli Re mutant strain F515 and the synthetic compound 406, as well as for all lipid A from the nonenterobacterial species R. capsulatus, R. vannielii, C. jejuni, C. violaceum, and R. fulvum. With respect to antagonistic activity, both partial structures, pentaacyl and tetraacyl lipid A, of enterobacterial origin were highly active in inhibiting cytokine induction by LPS. Lipid A from the nonenterobacterial species R. capsulatus and C. violaceum also inhibited cytokine induction, whereas those from C. jejuni, R. fulvum, and R. vannielii did not.
An understanding of the molecular mechanisms of cellular activation by endotoxin is an important prerequisite for the development and improvement of therapeutic concepts, including the application of endotoxin antagonists. In our previous studies we have proposed that high biological activity is correlated to a noncylindrical shape of lipid A [4,19]. The present data clearly show that this is a general principle, high biological activity being expressed only by the conical/concave shaped lipid A, like that from E. coli. In contrast, all endotoxins exhibiting a cylindrical molecular shape (Fig. 3) were found to have low or no endotoxic activity. Moreover, the present data verify our previous hypothesis  that lack of agonistic activity, but expression of antagonistic activity, by lipid A is connected with a cylindrical shape. Lack of acyl groups per se is not a prerequisite for antagonistic activity, as could be concluded by comparing the chemical composition of the known antagonistic compounds, e.g. natural and synthetic pentaacyl lipid A from R. capsulatus and R. sphaeroides and the synthetic tetraacyl compound 406, which have been described previously [6–10]. Lipid A from C. violaceum, which carries six symmetrically distributed fatty acids on the glucosamine backbone, also exhibits antagonistic activity. These data concomitantly emphasize the importance of the location of the acyl chains at the diglucosamine backbone, because symmetric (3 + 3) and asymmetric (4 + 2) chain distributions result in different packing densities as well as a different tilt of the lipid A backbone with respect to the fatty acid orientation and the membrane surface (A. B. Schromm, K. Brandenburg, H. Loppnow, A. P. Moran, M. H. J. Koch, E. Th. Rietschel & U. Seydel, unpublished results).
Lipid A from C. jejuni possesses low biological activity and, thus, may be expected to represent an antagonist. However, no such activity could be detected in the biological systems applied. This might be explained by its very slight tendency towards a conical/concave molecular shape, which may also explain the low agonistic activity. The finding that the lipid A from R. fulvum and R. vannielii do not show any antagonistic activity despite their cylindrical molecular shape can be explained by assuming that these lipid A molecules do not associate with cellular binding molecules of immune cells as the initial event and prerequisite for cell activation. It has been demonstrated previously that cell activation is strongly augmented by binding of LPS to serum and membrane proteins. A direct binding of LPS to LPS-binding protein (LBP) with the subsequent transport of endotoxin into phospholipid membranes  and binding of LPS to the phosphatidylinositol-anchored surface antigen CD14, which is augmented by LBP [2,22], has been described. The binding of LPS to these proteins, however, requires a sufficiently high negative charge density on the lipid A backbone [21,23], which is much lower and also sterically different for lipid A from R. fulvum with only one negative charge (GalA in 1-position of GlcN I, Fig. 1), or R. vannielii (no charge) compared with the bisphosphorylated backbone of enterobacterial lipid A.
Thieblemont et al.  have reported that the hydrophobic compound chlorpromazine, when added to an agonistically inactive lipid A, converts this lipid A into an agonistically one. We tested the validity of our conformational concept using this system; the addition of chlorpromazine to the antagonistic pentaacyl lipid A from E. coli with the unilamellar structure (see Fig. 3) in a molar ratio of 2 : 1 led to the conversion into an inverted cubic aggregate structure (data not shown), i.e. to a conversion from an agonistically inactive cylindrical shape into an agonistically active conical/concave shape.
Several other proteins, surface-bound or membrane-spanning, have been reported to be involved in LPS binding to and activation of monocytes/macrophages [25–27]. However, LPS-binding proteins in membranes such as CD14 or the β2-integrins CD11/CD18 probably do not transmit signals to the cytoplasm of cells, the former because of a lack of a transmembrane domain, the latter because its membrane domain is not utilized for endotoxin-initiated signalling . The recently detected transmembrane proteins TLR2 and TLR4, members of the Toll-like receptor family, have been shown to be essential in LPS-mediated activation of cells . Also, the purinergic P2Z/P2X7 receptor can play a role in LPS signaling. Thus, LPS has been reported to interact directly or indirectly with the P2X7 receptor, this way modulating the P2X7 channel activity and controlling the influx of divalent cations that are important in signal transduction .
From our results, we conclude that after specific recognition the direct intercalation of endotoxin into phospholipid membranes  represents an essential step in the signaling mechanism. This would be in agreement with data of Thieblemont et al.  who showed that cells sort agonistic and antagonistic LPS molecules differently according to their molecular shape: biologically active LPS molecules are transported from the plasma membrane into a defined intracellular site, whereas antagonistic LPS structures remain at the cell periphery. Independent of the existence of such transport mechanisms, we propose that cell activation occurs in the plasma membrane by lateral diffusion of the intercalated endotoxin molecules to transmembrane proteins that then initiate signaling by steric stress, which is induced by the conical shape of agonistic LPS molecules. In this model, antagonistically active compounds with a cylindrical molecular shape would not be able to induce this steric stress, but occupy the sites around the sensitive protein, thus protecting it from steric stress exerted by active lipid A shapes. Such a mechanism has been shown, for example, for a K+-channel activated by mechanical stress in the lung .
To our knowledge, this is the first report on a correlation between agonistic and antagonistic conformational properties of LPS structures. The data suggest that the association of a conical shape with endotoxicity and a cylindrical shape with lack of endotoxic, but expression of antagonistic activity constitutes a general principle that could be useful in designing new compounds with respect to their antagonistic suitability and in the further development of new preventive or therapeutically active compounds against septic shock.