Báo cáo Y học: Biophysical characterization of the interaction of high-density lipoprotein (HDL) with endotoxins doc

Biophysical characterization of the interaction of high-density lipoprotein (HDL) with endotoxins Klaus Brandenburg 1 , Gudrun Ju¨ rgens 1 ,Jo¨ rg Andra¨ 1 , Buko Lindner 1 , Michel H. J. Koch 2 , Alfred Blume 3 and Patrick Garidel 3 1 Forschungszentrum Borstel, Biophysik, Borstel, Germany; 2 European Molecular Biology Laboratory, Hamburg Outstation, EMBL c/o DESY, Hamburg, Germany; 3 Martin-Luther-Universita ¨ t Halle/Wittenberg, Institut fu ¨ r Physikalische Chemie, Halle, Germany The interaction of bacterial endotoxins [lipopolysaccharide (LPS) and the Ôendotoxic principleÕ lipid A], with high-den- sity lipoprotein (HDL) from serum was investigated with a variety of physical techniques and biological assays. HDL exhibited an increase in the gel to liquid crystalline phase transition temperature T c and a rigidification of the acyl chains of the endotoxins as measured by Fourier-transform infrared spectroscopy and differential scanning calorimetry. The functional groups of the endotoxins interacting with HDL are the phosphates and the diglucosamine backbone. The finding of phosphates as target groups is in accordance to measurements of the electrophoretic mobility showing that the zeta potential decreases from )50 to )60 mV to )20 mV at binding saturation. The importance of the sugar backbone as further target structure is in accordance with the remaining negative potential and competition experiments with polymyxin B (PMB) and phase transition data of the system PMB/dephosphorylated LPS. Furthermore, endo- toxin binding to HDL influences the secondary structure of the latter manifesting in a change from a mixed a-helical/ b-sheet structure to a predominantly a-helical structure. The aggregate structure of the lipid A moiety of the endotoxins as determined by small-angle X-ray scattering shows a change of a unilamellar/inverted cubic into a multilamellar structure in the presence of HDL. Fluorescence resonance energy transfer data indicate an intercalation of pure HDL, and of [LPS]–[HDL] complexes into phospholipid liposomes. Fur- thermore, HDL may enhance the lipopolysaccharide-bind- ing protein-induced intercalation of LPS into phospholipid liposomes. Parallel to these observations, the LPS-induced cytokine production of human mononuclear cells and the reactivity in the Limulus test are strongly reduced by the addition of HDL. These data allow to develop a model of the [endotoxin]/[HDL] interaction. Keywords: endotoxin conformation; high density lipopro- teins (HDL); lipopolysaccharides; Fourier-transform infra- red spectroscopy. Bacterial lipopolysaccharides (LPS) belong to the most potent stimulators of the immune system and play an important role in the pathogenesis and manifestation of Gram-negative infections, in general, and of septic shock, in particular, and are thus called endotoxins. The mechanism of endotoxin interaction with different target cell structures are still largely unknown and only limited data are available on the detailed mode of binding of endotoxins to various endogenous proteins, which are important with regard to combat invading microorgan- isms and to transport and neutralize free endotoxin. Among the humoral factors which are important LPS- binding molecules are serum lipoproteins. It was sugges- ted that sequestering of LPS by lipid particles may form an integral part of humoral detoxification [1]. Lipo- proteins are water-soluble complexes with a neutral core, surrounded by a phospholipid layer that contains cholesterol and one or more ÔapolipoproteinsÕ.Theyserve as ligands for cell membrane receptors, as cofactors for enzymes, and can dock lipopolysaccharide-binding pro- teins. They are classified as very-low density, low-density and high-density lipoproteins (HDL) according to their buoyant density. The primary function of these lipo- proteins is to transport lipids, cholesterol and cholesteryl esters in blood and the lymphatic system. HDL moreover plays a role in binding and neutralizing bacterial lipopolysaccharide and decrease the immunostimulatory action of LPS. In particular, a drastic reduction of the LPS-induced cytokine production [tumor necrosis factor- a, interleukin (IL)-1, IL-6] due to HDL binding was observed [2–4]. Furthermore, it was demonstrated that lipopolysaccharide-binding protein (LBP) increased the uptake of LPS by reconstituted HDL (R-HDL) particles derived from either LPS ÔmicellesÕ or LPS–sCD14 com- plexes, and in this process LPS molecules are exchanged with phospholipids [5]. Here, we report on the interaction of HDL with deep rough mutant LPS Re and the Ôendotoxic principleÕ, lipid A applying a variety of physical and biological techniques. With Fourier-transform infrared spectroscopy (FTIR) the phase transition behavior of the acyl chains of the Correspondence to K. Brandenburg, Forschungszentrum Borstel, Biophysik, Parkallee 10, D-23845 Borstel, Germany. Fax: +49 4537 188632, Tel.: + 49 4537 188235, E-mail: kbranden@fz-borstel.de Abbreviations: ATR, attenuated total reflectance; FTIR, Fourier- transform infrared spectroscopy; HDL, high-density lipoprotein; IL, interleukin; LAL, Limulus amebocyte lysate; LBP, lipo- polysaccharide-binding protein; LPS, lipopolysaccharide; PMB, polymyxin B; PtdSer, phosphatidylserine. (Received 2 September 2002, revised 18 October 2002, accepted 24 October 2002) Eur. J. Biochem. 269, 5972–5981 (2002) Ó FEBS 2002 doi:10.1046/j.1432-1033.2002.03333.x endotoxins in absence and presence of HDL as well as the effect of HDL on functional groups of the endotoxins were observed for the latter, using the attenuated total reflectance (ATR) method. To obtain information about the phase transition enthalpy changes of the endotoxins, differential scanning calorimetry in the absence and presence of HDL was carried out. Also, with FTIR the influence of endotoxin binding on the secondary structure of the protein part of HDL, apolipoprotein A-I (apoA-I) was observed. The effect of HDL on the surface charge of the endotoxin aggregates was studied by applying zeta potential measurements, which also enabled an estimate for the binding saturation to be made. The aggregate structure and, with that, the conformation of the lipid A part of LPS was studied by small-angle X-ray diffraction. With fluorescence resonance energy transfer experiments, information about the influence of HDL on the inter- calation of LPS and LBP, and the intercalation of the lipoprotein itself into phospholipid target membranes could be given. Finally, in biological experiments the ability of the endotoxin and [endotoxin]/[HDL] complexes to induce cytokine production in mononuclear cells and to activate the Limulus amebocyte lysate (LAL) clotting cascade was measured. Thus, it was possible to charac- terize the binding of HDL to the endotoxins profoundly and to get insight into the mechanisms of the reduction of the LPS-induced cytokine production in human mono- nuclear cells. MATERIALS AND METHODS Lipids and reagents Lipopolysaccharide from the deep rough mutant Re Salmonella minnesota (R595) was extracted by the phenol/ chloroform/petrol ether method [6] from bacteria grown at 37 °C, purified, and lyophilized. Free lipid A was isolated by acetate buffer treatment of LPS R595. After isolation, the resulting lipid A was purified and converted to its triethyl- amine salt. The known chemical structure of lipid A from LPS R595 was checked by the analysis of the amount of glucosamine, total and organic phosphate, and the distribution of the fatty acid residues applying standard procedures. The amount of 2-keto-3-deoxyoctonate never exceeded 5 weight %. Dephospho-LPS Re was prepared from LPS deep rough mutant F515 from Escherichia coli byHFtreatmentatlow temperature (4 °C). The detailed procedure is described elsewhere [7]. High-density lipoprotein (HDL) from human plasma was purchased from Fluka (Deisenhofen, Germany). It was essentially free of contaminants, in particular of LPS, which was examined by applying the Limulus test (see later). Lipopolysaccharide-binding protein (LBP) was a kind gift of S. F. Carroll (XOMA corporation, Berkeley, CA, USA). Sample preparation The lipid samples were usually prepared as aqueous disper- sions at high buffer content, i.e. above 60% using 20 m M Hepes (pH 7). For this, the lipids were suspended directly in buffer, sonicated and temperature-cycled several times between5and70°C and then stored for at least 12 h before measurement. For the elucidation of the protein secondary structure in the absence and presence of endo- toxins, HDL was prepared in buffer made either from H 2 O or D 2 O incubated at 37 °C for 30 min, and lipid dispersions prepared as described above were added in appropriate amounts, and further incubated at 37 °C for 15 min. Afterwards, 10 lL of these dispersions were spread on a CaF 2 infrared window, and the excess water was evaporated slowly at 37 °C. FTIR spectroscopy The infrared spectroscopic measurements were performed on a 5-DX FTIR spectrometer (Nicolet Instruments, Madison, WI, USA) and on an IFS-55 spectrometer (Bruker, Karlsruhe, Germany). The lipid samples were placed in a CaF 2 cuvette with a 12.5-lm Teflon spacer. Temperature-scans were performed automatically between 10 and 70 °C with a heating-rate of 0.6 °CÆmin )1 .Every 3 °C, 50 interferograms were accumulated, apodized, Fou- rier transformed and converted to absorbance spectra. For strong absorption bands, the band parameters (peak position, band width, and intensity) were evaluated from the original spectra, if necessary after subtraction of the strong water bands. In the case of overlapping bands, in particular for the analysis of amide I-vibration mode, curve fitting was applied using a modified version of the CURFIT program obtained by D. Moffat, NRC, Ottawa, Canada. An estimate of the number of band components was obtained from deconvolution of the spectra [8] and the curve was fitted to the original spectra after subtraction of base lines resulting from neighboring bands. The bandshapes of the single components are superpositions of Gaussian and Lorentzian. Best fits were obtained by assuming a Gauss fraction of 0.55–0.60. The precision of the curve fit procedure is approximately 3%. ATR The lipids were prepared as oriented thin multilayers as described previously [9] by spreading a 1-m M lipid suspen- sion, which was temperature-cycled between 5 and 70 °C several times prior to spreading, in Hepes buffer on a ZnSe ATR crystal and evaporating the excess water by slow periodic movement under a nitrogen stream at room temperature. The lipid sample was placed in a closed cuvette, and the air above the sample was saturated with water vapor to maintain full hydration. Infrared ATR spectra were recorded with a mercury–cadmium–telluride detector with a scan number of 1000 at a resolution of 2cm )1 . The measurements were performed at 26 °C, the intrinsic instrument temperature, in some cases also at 37 °C. Differential scanning calorimetry LPS was dispersed in buffer at a concentration of 1mgÆmL )1 . A liposomal lipid dispersion was obtained by sonication for 10 min at 40 °C. After cooling to room temperature, a defined amount of HDL was added to 1 mL lipid dispersion and the sample was gently vortexed until Ó FEBS 2002 HDL interaction with endotoxins (Eur. J. Biochem. 269) 5973 HDL was completely dissolved [9]. Differential scanning calorimetry measurements were performed with a MicroCal VP scanning calorimeter (MicroCal, Inc., Northampton, MA, USA). The heating and cooling rate was 1 °CÆmin )1 . Heating and cooling curves were measured in the tempera- ture interval from 10 to 100 °C. Three consecutive heating and cooling scans were measured [10]. X-ray diffraction X-ray diffraction measurements were performed at the European Molecular Biology Laboratory (EMBL) outsta- tion at the Hamburg synchrotron radiation facility HASY- LAB using the double-focusing monochromator-mirror camera X33 [11]. Diffraction patterns in the range of the scattering vector 0.07 < s <1nm )1 (s ¼ 2sinhÆk )1 ,2h scattering angle and k the wavelength ¼ 0.15 nm) were recorded at 40 °C with exposure times of 2 or 3 min using a linear detector with delay line readout [12]. The s-axis was calibrated with tripalmitate, which has a periodicity of 4.06 nm at room temperature. Details of the data acquisi- tion and evaluation system can be found elsewhere [13]. The diffraction patterns were evaluated as described previously [14] assigning the spacing ratios of the main scattering maxima to defined 3D structures. The lamellar and cubic structures are most relevant here. They are characterized by the following features: (a) lamellar: The reflections are grouped in equidistant ratios, i.e. 1, 1/2, 1/3, 1/4, etc. of the lamellar repeat distance dL; (b) cubic: The different space groups of these nonlamellar 3D structures differ in the ratio of their spacing. The relation between reciprocal spacing s hkl ¼ 1/d hkl and lattice constant a is s hkl ¼ [(h 2 +k 2 +l 2 )/a] 1/2 , where hkl are Miller indices of the corresponding set of plane. Zeta potential Zeta potentials were determined with a Zeta-Sizer 4 (Malvern Instr., Herrsching, Germany) at a scattering angle of 90° from the electrophoretic mobility by laser-Doppler anemometry as described earlier [15]. The zeta potential was calculated according to the Helmholtz-Smoluchovski equa- tion from the mobility of the aggregates in a driving electric field of 19.2 VÆcm )1 . It was determined for the endotoxins (0.5 m M ) at different HDL concentrations. Isothermal titration calorimetry Microcalorimetric experiments of HDL-binding to endo- toxins were performed on an MCS isothermal titration calorimeter (Microcal Inc., Northampton, MA, USA). The endotoxin samples at a concentration of 0.25 mgÆmL )1 , prepared as described above, were filled into the microca- lorimetric cell (volume 1.3 mL), and HDL at concentrations up to 12 mgÆmL )1 were loaded into the syringe compart- ment, both after thorough degassing of the suspensions. After temperature equilibration, the HDL was titrated in 5 lL portions every 10 min into the endotoxin-containing cell, and the heat for each injection measured by the ITC instrument was plotted vs. time. The total heat signal from each experiment was subsequently determined by integra- ting the individual peaks and plotted against the [HDL]/ [endotoxin] weight ratio. Fluorescence resonance energy transfer The fluorescence resonance energy transfer assay was per- formed as described earlier [16,17]. Briefly, phospholipid liposomes from phosphatidylserine (PtdSer) were doubly labeled with the fluorescent dyes N-(7-nitrobenz-2-oxa-1,3- diazol-4yl)-phosphatidylethanolamine and N-(lissamine rhodamine B sulfonyl)-phosphatidylethanolamine (Rh-PE) (Molecular Probes, Eugene, OR, USA). Intercalation of unlabeled molecules into the doubly labeled liposomes leads to probe dilution and thus inducing a lower fluorescence resonance energy transfer efficiency: the emission intensity of the donor increases and that of the acceptor decreases (for clarity, only the quotient of the donor and acceptor emission intensity is shown here). In all experiments, doubly labeled PtdSer liposomes were prepared and after 50, 100, and 150 s recombinant LBP, LPS, and HDL were added in different order, and the NBD donor fluorescence intensity at 531 nm was monitored for at least 300 s. LBP, HDL and LPS were added in the weight ratios 0.5 : 1 : 1. Stimulation of human mononuclear cells by LPS Re For an examination of the cytokine-inducing capacity of the [endotoxin]/[HDL] mixtures, human mononuclear cells were stimulated with the latter and the IL-6 production of the cells was determined in the supernatant. Mononuclear cells were isolated from heparinized (20 IEÆmL )1 ) blood taken from healthy donors and processed directly by mixing with an equal volume of Hank’s balanced solution and centrifugation on a Ficoll density gradient for 40 min (21 °C, 500 g). The layer of mononuclear cells was collected and washed twice in Hank’s medium and once in serum-free RPMI 1640 containing 2 m ML -glutamine, 100 UÆmL )1 penicillin, and 100 lgÆmL )1 streptomycin. The cells were resuspended in serum-free medium and their number was equilibrated at 5 · 10 6 cellsÆmL )1 . For stimu- lation, 200 lLÆwell )1 mononuclear cells (5 · 10 6 cellsÆmL )1 ) were transferred into 96-well culture plates. The stimuli were seriallydilutedinserum-freeRPMI1640andaddedtothe cultures at 20 lL per well. The cultures were incubated for 4 hat37 °C under 5% CO 2 . Supernatants were collected after centrifugation of the culture plates for 10 min at 400 g and stored at )20 °C until determination of cytokine content. Immunological determination of IL-6 in the cell super- natant was performed in a sandwich-ELISA as described elsewhere [18]. Ninety-six-well plates (Greiner, Solingen, Germany) were coated with a monoclonal (mouse) human IL-6 antibody (clone 16 from Intex AG, Switzerland). Cell culture supernatants and the standard (recombinant human IL-6, Intex) were diluted with buffer. After exposure to appropriately diluted test samples and serial dilutions of standard rIL-6, the plates were exposed to peroxidase- conjugated (sheep) anti-human IL-6 antibody. The plates were shaken 16–24 h at room temperature (21–24 °C) and washed six times in distilled water to remove the antibodies. Subsequently the color reaction was started by addition of tetramethylbenzidine/H 2 O 2 in alcoholic solution and stopped after 5–15 min by addition of 0.5 molÆL )1 sulfuric acid. In the color reaction, the substrate is cleaved enzymatically, and the product was measured photometri- cally on an ELISA reader (Rainbow, Tecan, Crailsham, 5974 K. Brandenburg et al. (Eur. J. Biochem. 269) Ó FEBS 2002 Germany) at a wavelength of 450 nm and the values were related to the standard. IL-6 was determined in duplicate at two different dilutions and the values were averaged. Determination of endotoxin activity by the chromogenic Limulus test Endotoxin activity of [LPS]–[HDL] mixtures at concentra- tions between 10 lgÆmL )1 and 10 pgÆmL )1 was determined by a quantitative kinetic assay based on the reactivity of Gram-negative endotoxin with LAL [19], using test kits from LAL Coamatic Chromo-LAL K (Chromogenix, Haemochrom). The standard endotoxin used in this test was from E. coli (O55:B5), and 10 EUÆmL )1 corresponds to 1ngÆmL )1 . In this assay, saturation occurs at 125 endotoxin units EUÆmL )1 , and the resolution limit is £ 0.1 EUÆmL )1 (maximum value for ultrapure water from embryo-transfer, Sigma). RESULTS Measurements of hydrated LPS–HDL complexes Infrared-ATR experiments were performed with hydrated LPS multilayers in the absence and presence of different HDL concentrations. In these measurements, the LPS concentration was held constant and the spectra were normalized by taking the band intensity of the symmetric stretching vibration m s (CH 2 ) as standard. In Fig. 1, a change in the band contours in the range of the two phosphate and the diglucosamine vibrations, m as (PO 2 ) 1270–1250 cm )1 , and m as (PO 2 ) hydr. 1230–1220 cm )1 and m as (diglucosamine) 1180–1150 cm )1 , can be seen; the addition of HDL leads to an intensity decrease in the band contours proportional to the HDL concentration. From Fig. 1 it can be taken that especially the intensity of the band component around 1190 cm )1 increases as compared with that at lower wavenumbers and becomes sharper. Additionally, the component at 1170 cm )1 for pure LPS is shifted to approximately 1177 cm )1 in the presence of HDL. These results indicate that (i) besides the phosphate groups, the sugar diglucosamine part in lipid A are also binding-sites for HDL, and (ii) these vibrational bands are immobilized due to HDL binding. Gel to liquid crystalline (b«a) phase behavior The b«a gel to liquid crystalline acyl chain melting behavior was investigated with FTIR by evaluating the peak position of the symmetric stretching vibration m s (CH 2 ), which is a measure of acyl chain order. HDL induces a slight rigidification in particular in the liquid crystalline (a) phase of the acyl chains of LPS Re, as deduced from a decrease in wavenumber values at a given temperature, and a significant increase in the phase transition T c from 31 °C for pure LPS to 40 °C for an [LPS]–[HDL] mixture at a weight ratio of 1 : 4. Also, pure HDL exhibits a signal in this wavenumber range due to its phospholipid moiety. This, however, is much higher with an only weak temperature dependence in the wavenumber range 2852.5–2853.5 cm )1 (Fig. 2). These values are indicative of acyl chains with a large amount of gauche conformers. Importantly, the interaction of HDL with LPS leads to a reduction of the wavenumber by more than one unit (see vertical line at 37 °C), i.e. a strong rigidification of the lipid A acyl chains. This holds true also for lipid A even although higher amounts of HDL are required to induce a significant increase in T c . Thus, at a weight ratio[lipid A]/[HDL] 1 : 3 the phase transition at T c ¼ 45 °C of pure lipid A is shifted to 50 °C (data not shown). This observation reflects the different number of negative charges and monosaccharide units (LPS Re has four negative charges and four sugar units, lipid A two of each) which may be connected with different conformations of the molecules. Differential scanning calorimetry measurements of the interaction of LPS with HDL (Fig. 3) shows for pure LPS a phase transition in accordance to that observed in Fig. 2. Fig. 1. Infrared-ATR spectra in the range of the antisymmetric stretching vibration of the negatively charged phosphate groups m as (PO 2 – )1210–1260 cm )1 ) and the diglucosamine ring vibration (see arrows) of LPS at different [LPS]/[HDL] weight ratios. The spectra were normalized by taking the band intensity of the symmetric stretching vibration of the methylene groups m s (CH 2 )asstandard. Fig. 2. Peak position of the symmetric stretching vibration of the methylene groups m s (CH 2 ) vs. temperature for a 10-m M LPS Re pre- paration at different HDL concentrations. In the gel (b) phase of the acyl chains, the peak position lies at 2850 cm )1 , in the liquid crystalline (a) phase at 2852.5 cm )1 . Ó FEBS 2002 HDL interaction with endotoxins (Eur. J. Biochem. 269) 5975 The phase transition in the first heating scan is characterized by a coexistence region between 22 and 37 °C(T 1/2 ¼ 4.5 °C) and the maximum of the heat capacity curve is found at 31 °CwithDH C ¼ 38 kJÆmol )1 . The succeeding cooling scan reveals only a very small hysteresis for the re-crystallization of the acyl chains from the liquid crystal- line to the gel phase. The maximum of the heat capacity curve of the 1st cooling scan is observed at T ¼ 28 °Cwith DH ¼ )39 kJÆmol )1 . A shoulder at  23 °C is observed in the first and succeeding cooling scan. The thermograms of the succeeding heating scan are slightly broader compared with the 1st heating scan (Fig. 3A). HDL was added to LPS at different concentrations {[LPS]/[HDL] 1 : 0.25, 1 : 0.45, 1 : 0.6 and 1 : 1 (w/w)}. In Fig. 3(B) representative thermograms for the sample at a LPS/HDL 1 : 1 (w/w) ratio are plotted. The phase trans- ition temperature of LPS is shifted from 31 °Cto 33 °C, the half-width of the phase transition is increased (T 1/2 ¼ 7 °C) and the phase transition enthalpy is decreased by  22%. The presence of HDL induces a broadening of the coexistence range of the phase transition, especially for the offset temperature which is shifted above 42 °C. The phase transition as derived from the IR spectra from the temperature dependence of m s (CH 2 ) of the [LPS]/[HDL] 1 : 0.5 system revealed similar data: T c ¼ 34 °Cand T 1/2 ¼ 8.5 °C. The heat-capacity curve of LPS/HDL ratio develops a shoulder starting at  20 °C in the gel phase indicating that HDL interacts with the gel phase LPS. This is observed for all four investigated LPS/HDL concentra- tion ratios. A second peak with a very small enthalpy contribution at higher temperature (T  63 °C, DH ¼  8kJÆmol )1) corresponds to the denaturation peak of pure HDL, because the maximum of the heat capacity curve of pure HDL is observed at  63 °C (Fig. 3C). Thus, additional HDL does not interact with the LPS membrane but acts like pure protein. Heating of the sample above  70 °C leads to complete and irreversible denaturation of HDL (data not shown). Parallel to the measurements of LPS Re, differential scanning calorimetry measurements of the phase behavior of lipid A indicated a similar increase in T c ,andthe evaluation of the phase transition enthalpy (peak area) showed a value of 14 kJÆmol )1 which in the presence of HDL is reduced to 12 kJÆmol )1 , i.e. a reduction by 15%. These data indicate that the binding of HDL to LPS and lipid A leads to a disturbance of the hydrophobic moiety. Inhibition experiments were performed with the polycat- ionic peptide polymyxin B (PMB), which binds strongly to the lipid A phosphates [20]. At a [LPS]/[PMB] weight ratio of 1 : 0.24, PMB alone causes a drastic fluidization of LPS, while HDL leads to a rigidification of LPS at a weight ratio of [LPS]/[HDL] 1 : 1.5 (Fig. 4). Addition of HDL to the preincubated [LPS]–[PMB] complex leads to almost the same result as without HDL, and addition of PMB to preincubated [LPS]–[HDL] causes a slightly attenuated fluidizing effect as compared with LPS with PMB alone. PMB, which binds much stronger to the LPS phosphates than HDL, may displace HDL molecules from their binding site, the lipid A phosphates. These results are complemented by the data of the dephospho-LPS Re and HDL systems (Fig. 5). Dephos- pho-LPS Re has a T c of 45 °C,andinthecaseof phosphates as the primary binding site no change of the phase behavior of dephospho-LPS Re would be expected. However, addition of HDL causes a fluidization parti- cularly in the gel phase and in the transition range at a Fig. 3. Differential scanning calorimetry heat capacity curves of pure LPS Re (A), a mixture of [LPS]/[HDL] at 1.1 : 1 w/w (B), and for pure HDL (C). Heating and cooling curves were measured in the temperature interval 10– 100 °C. Three consecutive heating and cooling scans are presented (A,B) (h.s. heating-scan, c.s. cooling scan) and first heating scan (C). Fig. 4. Peak position of the symmetric stretching vibration of the methylene groups m s (CH 2 ) vs. temperature in competition experiments with LPS Re, PMB and HDL in different sequences. 5976 K. Brandenburg et al. (Eur. J. Biochem. 269) Ó FEBS 2002 weight ratio [dephospho-LPS]/[HDL] 1 : 4.5. As a control, the effect of PMB on dephospho-LPS Re was monitored. It is found that PMB causes a slight decrease in T c , but no change in fluidity takes place (data not shown). Therefore, the phosphates can be assumed not to be the only binding- sites for HDL. LPS and lipid A aggregate structures Synchrotron radiation X-ray small-angle diffraction was performed with lipid A at 40 °C and at different concentrations of HDL. The diffraction patterns of pure lipid A (Fig. 6, top) are indicative of a superposition of a unilamellar with a cubic inverted structure in accordance to former results [21], which can be deduced from the occurrence of the broad interference maximum between 0.1 and 0.4 nm )1 superimposed by diffraction maxima at 8.20 nm ¼ 18.4 nmÆÖ5, 5.31 nm ¼ 18.4 nmÆÖ12, 4.08 nm ¼ 18.4 nmÆÖ20 of a periodicity at a Q ¼ (18.3 ± 0.3) nm (the latter is expressed only very weakly). In the presence of HDL, this mixed structure converts into a multilamellar one, which can be deduced from the occurrence of reflections at equidistant ratios, d | ¼ 5.13 nm and 2.60 nm ¼ d | /2 and 1.74 nm ¼ d | /3 (Fig. 6, bottom). From these data an approximation of the molecular shape of lipid A is possible: In the absence of HDL, it is conical with a higher cross-section of the hydrophobic than the hydrophilic moiety, and is conver- ted into a cylindrical one in the presence of HDL. HDL secondary structure The secondary structure of the apolipoprotein (apoA-I) part of HDL was determined by IR-spectroscopy by analyzing the amide I-vibration (predominantly C¼O stretching vibration) in the spectral range 1700–1600 cm )1 in H 2 O-containing as well as D 2 O-containing buffer. IR spectra are given in the range 1700–1400 cm )1 at a [LPS]/ [HDL] ratio of 1 : 0.5 weight percentage in D 2 O(Fig.7A) exhibiting the amide I¢-vibration centered around 1653 cm )1 , but only a very weak amide II-vibration due to H/D exchange [22]. The evaluation of the amide I¢ vibrational band shows that for HDL in the presence of Re-LPS ([HDL]/[Re-LPS] 1 : 0.5 weight ratio) in Fig. 7B the b-turn/antiparallel b-sheet components of the protein’s secondary structure is changed in favor of the a-helical component (a detailed assignment of the different secon- dary structures is presented in the legend of Fig. 7). For pure HDL the a-helical portion is approximately 34% and for the complexes ([HDL]/[Re-LPS] 1 : 0.5 weight ratio) approximately 44%. From the broadening of the 1653 cm )1 band it becomes obvious that a more hetero- geneous population of a-helical structures emerge as consequence of binding to LPS. The occurrence of unordered (random coil) structures, which in D 2 Oare located in the range 1640–1645 cm )1 , can be excluded, as measurements in H 2 O, for which the unordered structures are found around 1660 cm )1 , exhibited a similar band contour except for the fact that the peak position of the amide I vibration is shifted to approximately 1658 cm )1 . Zeta potential The zeta potential as an indicator for accessible surface charges was determined for LPS Re and lipid A in the presence of increasing amounts of HDL. From Fig. 8 it can be deduced that the pure endotoxins have a high negative surface charge corresponding to potential values of )50 to Fig. 5. Peak position of the symmetric stretching vibration of the methylene groups m s (CH 2 ) vs. temperature for a 10-m M dephospho-LPS Re preparation at different HDL concentrations. Fig. 6. Synchrotron radiation X-ray diffraction patterns of lipid A (top) and a mixture of lipid A and HDL (bottom, weight ratio 1 : 0.5) at 90% water content. The diffraction pattern of the aggregate structure of lipid A indicates the existence of a superposition of a unilamellar with a cubic inverted structure, that of the mixture a multilamellar structure. Ó FEBS 2002 HDL interaction with endotoxins (Eur. J. Biochem. 269) 5977 )60 mV, which is increasingly compensated by the addition of higher amounts of HDL. However, the charge compen- sation seems to be completed at a weight ratio [endotoxin]/ [HDL] 1 : 1 at a remaining potential of )20 mV. Therefore, HDL does not compensate the negative charges of the endotoxins completely. Isothermal titration calorimetry With ITC an estimate of the stoichiometry of HDL–LPS binding can be obtained. For this, a LPS dispersion (0.25 mgÆmL )1 ) within the calorimeter cell was titrated with a HDL solution (5 lLof12mgÆmL )1 every 10 min). The titration yields a negative enthalpy change DH c of the LPS– HDL binding corresponding to an exothermic reaction (data not shown). A maximum of DH c ¼ )14 kJ is observed at a weight ratio of 1 : 1. At higher [HDL] contents, the DH c values decrease to )6 kJ at weight ratios [HDL]/[LPS] ¼ 4 : 1–6 : 1, but do not decrease to zero. Unfortunately, the HDL amounts available did not allow to realize higher HDL concentrations, i.e. to determine the saturation of binding, which therefore must be significantly higher than a weight ratio of [HDL]/[LPS] ¼ 6:1. Intercalation into phospholipid liposomes It has been shown that LBP mediates the transport of LPS into negatively charged liposomes [17] which seems to be an important step in cell activation. Here, the LBP-mediated transport of LPS into PtdSer as example of negatively charged phospholipids was determined by fluorescence resonance energy transfer spectroscopy in the absence and presence of HDL (Fig. 9). The addition of LPS at t ¼ 50 s indicates that LPS itself does not intercalate into the PtdSer liposomes, the following addition of LBP at t ¼ 150 s leads to an rapid increase in NBD-fluorescence intensity corres- ponding to the LBP-mediated intercalation of LPS and LBP into the PtdSer liposomes (Fig. 9A). The addition of HDL at t ¼ 50 s leads to an increase in the NBD-fluorescence intensity indicating an intercalation of HDL into the PtdSer liposomes, the subsequent addition of LPS at t ¼ 100 s apparently leads to an HDL-mediated transport of LPS into the target cell membrane (Fig. 9B), as the addition of pure buffer instead of LPS at this time causes a reduction of the fluorescence intensity due to dilution (data not shown). The final addition of LBP at t ¼ 150 s leads to another increase in the NBD-fluorescence intensity caused by intercalation of pure LBP and LBP-mediated intercalation of LPS into the PtdSer liposomes (Fig. 9B). In Fig. 9C, the addition of LPS first and then of HDL again showed no intercalation of LPS by itself, an intercalation of HDL as found already in Fig. 9(B), and the final strong increase of the NBD-fluorescence intensity indicates the intercalation of LBP and the [LPS]–[LBP] complex. In Fig. 9D, after addition of the preincubated complex (LPS + HDL) the increase of the NBD-fluorescence intensity indicates an intercalation of HDL and (LPS + HDL) complex, which is followed by the strong increase due to LBP-induced intercalation. Similar results are obtained when the PtdSer is replaced by phospholipid liposomes corresponding to the composi- tion of the macrophage membrane [16], only the effects are significantly weaker. IL-6-production in mononuclear cells IL-6 production in human mononuclear cells induced by LPS Re (10 ngÆmL )1 ) was investigated at different HDL concentrations. The concentration of LPS Re was Fig. 7. Infrared spectra in the range 1700–1400 cm )1 for a hydrated sample of [LPS]/[HDL] = 1 : 2 weight ratio (A) and in the range of the amide I¢ (predominantly C=O stretch) vibration for hydrated samples of HDL (B, top) and in the presence of Re-LPS ([LPS]/[HDL] = 1 : 2 weight ratio) (B, bottom). The measurements were performed at 37 °CinD 2 O. Band component assignments: 1653 cm )1 , a-helix; 1636–1638 cm )1 , b-sheet; 1667– 1671 cm )1 , b-turns; 1682–1685 cm )1 , b turns and antiparallel b sheet. In the figures, the values of the peak positions and the respective bandwidths (determinedathalfheight,incm )1 ) are listed. The curve fitting was performed by assuming a Gaussian fraction of 0.6 (Lorentzian fraction 0.4). The precision of the secondary structural determination is approximately 3%, obtained from repeated measurements (n ¼ 5). Fig. 8. Zeta potential of 0.5 m M lipid A and LPS Re preparations in dependence on different [endotoxin]/[HDL] weight ratios from the determination of the electrophoretic mobility by laser Doppler anemometry. 5978 K. Brandenburg et al. (Eur. J. Biochem. 269) Ó FEBS 2002 held constant (10 ngÆmL )1 ) while the HDL concentration was increased (10 ngÆmL )1 ,100ngÆmL )1 ,1lgÆmL )1 , 10 lgÆmL )1 ) to produce the weight ratios shown. As plotted in Fig. 10, three types of experiments with LPS Re and HDL have been carried out. Preincubation of the cells withLPSRe(30minat37°C) and following addition of HDL, preincubation of the cells with HDL and following addition of LPS Re, and the incubation of the cells with (LPS + HDL) complexes. Preincubation of the cells with HDL and following addition of LPS Re, and the (LPS + HDL) complex leads in all examined concentrations to a decrease in IL-6 production. Preincubation of the cells with LPS Re and following addition of HDL leads to an insignificant increase in IL-6 production at the lowest HDL concentration ([LPS]/[HDL] 1 : 1 weight ratio), but at higher concentrations of HDL also to a decrease in the IL-6 production, but less as compared with the results in the other experiments. Biological activity in the LAL assay The ability of the LPS–HDL complexes to activate the LAL clotting cascade was measured at LPS Re concentrations of 100 pgÆmL )1 and 1 ngÆmL )1 . As shown in Fig. 11 we have found that HDL reduces the enzymatic activity induced by pure LPS Re at all investigated concentrations. For example, at 1 ngÆmL )1 the activity of 45 EUÆmL )1 for the pure LPS Re is reduced to values in the range 15–25 EUÆmL )1 at all concentrations, starting from [LPS]/[HDL] 1 : 1–1 : 1000 (Fig. 11B). Also, pure HDL was found to be endotoxin-free as deduced from the low values in the LAL test which nearly correspond to the values of pure water. DISCUSSION The results from the biophysical measurements of the binding of HDL to endotoxins indicate a strong interaction, which manifests in a binding to the lipid A backbone, in particular to the diglucosamine-phosphate region (Fig. 1), to an increase of the phase transition temperature of the acyl chains of the endotoxins and a drastic increase in acyl chain order, i.e. a rigidification of the endotoxin aggregates (Fig. 2). As the m s (CH 2 ) signal results from both, the acyl chains of LPS and of the phospholipids from the HDL particles, a pure addition of the signals would lead to a curve somehow in between those of pure HDL and pure LPS, i.e. it would indicate (Fig. 2) a Ôfluidization of LPSÕ.The observation of the rigidification therefore can be assumed to be even stronger than found in Fig. 2 due to the superpo- sitions of the HDL and phospholipid signals. The phase transition enthalpy of LPS with DH ¼ 38 kJÆmol )1 is slightly larger compared with the data reported for phos- pholipids [10], but considering the difference in the number of acyl chains (six for LPS instead of two for phospholipids), it is strongly reduced for LPS as compared with saturated phospholipids. The decrease of the phase transition Fig. 10. LPS-induced IL-6 production of human mononuclear cells by 10 ngÆmL )1 LPS Re and at different [LPS]/[HDL] weight ratios was determined in three types of examinations. Preincubation of the cells with LPS Re (30 min at 37 °C) and following addition of HDL, pre- incubation of the cells with HDL and subsequent addition of LPS Re, and incubation of the cells with [LPS Re]–[HDL] complexes. Fig. 11. Endotoxin activity in the chromogenic Limulus amebocyte lysate assay at two LPS Re concentrations (100 pgÆmL )1 and 1ngÆmL )1 ) and different HDL weight ratios. Fig. 9. Quotient of the fluorescence intensity at 531 nm of doubly labeled liposomes from PtdSer vs. time. After incubation with LPS Re and subsequent addition of LBP (A), after incubation with HDL and subsequent addition of LPS Re and LBP (B), after incubation with LPS Re and subsequent addition of HDL and LBP (C), and after incubation with a preincubated mixture of LPS and HDL and sub- sequent addition of LBP (D). Ó FEBS 2002 HDL interaction with endotoxins (Eur. J. Biochem. 269) 5979 enthalpy by approximately 22% at a weight ratio of [LPS]/ [HDL] 1 : 1 (Fig. 3) indicates the significant influence of the acyl chain moiety of LPS in the interaction which might be connected with the observation that the acyl chain melting in the presence of HDL does not take place completely. This may be taken from the wavenumber values of the LPS– HDL sample above T c (Fig. 5) which are by 0.3 cm )1 lower than the pure LPS sample, whereas the wavenumber values in the gel phase below T c aremoreorlessthesame. Beside the lipid A phosphate groups as target structures (Figs 1, 4, 5 and 8) the change of the phase transition of dephospho LPS (Fig. 4) and the remaining zeta potential after binding saturation (Fig. 8) give a hint that HDL binds also to other target structures in the endotoxins, for example to the sugar part of the endotoxins as deduced from the band intensity decrease of the diglucosamine ring mode (Fig. 1). This interpretation is strongly supported by the biological data: Coincubated (LPS + HDL) com- plexes lead in both test systems to a significant decrease of the signals (IL-6 production and LAL coagulating acti- vity). It has been reported for synthetic endotoxins that LAL activity is highest for preparations with a digluco- samine backbone including the 4¢-phosphate (compound 504), whereas the sample without 4¢-phosphate but with 1-phosphate (compound 505) was less active by one order of magnitude [23]. Thus, the binding of HDL to LPS must comprise at least the diglucosamine backbone inclusive the 4¢-phosphate (see also Fig. 1) which inhibits the activity in the LAL at all concentrations (Fig. 11). In previous papers, we have reported that binding of various proteins (hemoglobin, lactoferrin, recombinant human serum albumin) lead to systematic changes (increase or decrease in dependence on the protein) in cytokine induction, but there was no corresponding behavior in the LAL test [9,15,21]. This can now be interpreted as resulting from different target structures (epitopes) of the proteins as found here for HDL. Concomitant with the binding of HDL to the endotoxins, a reorientation of the lipid A aggregate structure from inverted cubic [21] to a multilamellar one (Fig. 6), and a slight change of the secondary structure of HDL from a mixed a-helical/b-sheet to a predominantly a-helical struc- ture (Fig. 7) take place. From this, a model of the LPS– HDL interaction can be deduced. The binding of HDL takes place essentially to the diglucosamine sugar backbone and the 4¢-phosphate of lipid A. The binding-places within the HDL moiety at present cannot be given. A further possibility of LPS binding to HDL, an incorporation of the LPS bilayer into the HDL interior, the phospholipid bilayer [24], does not seem to be probable as this would not explain the data for the 1-phosphate and diglucosamine groups as binding sites as well as the rigidification of the acyl chains. Still unclear is the binding stoichiometry of the LPS– HDL system. The data from the zeta potential (Fig. 8) indicate a value around 1 : 1 weight ratio. At this ratio, from an estimate of the molecular weights some hundreds LPS molecules per HDL apolipoprotein can be calculated. This is, however, far below saturation. Isothermal titration calorimetric (ITC) experiments showed up to a weight ratio of 1 : 6 [LPS]/[HDL] still no saturation. This is also in accordance to the biological data that a high excess weight ratio of HDL to LPS is necessary for a saturation of the tumor necrosis factor-a production (Fig. 10). The comparison of the results of the interaction of LPS with HDL to those published for the interaction of the former with another serum protein, albumin, indicates a completely different characteristic: Albumin (in its recom- binant form) compensates the phosphate charges to an only very low degree, the zeta potential remains lower than )40 mV, which seems to be connected with the observation that albumin does not reduce the immunostimulatory activity of LPS, rather a slight increase is observed [9]. Together with the data for hemoglobin, for which also no binding to the phosphate groups of LPS and no reduction of the immunostimulatory activity can be found [21], it may be hypothesized that a basic prerequisite for a decrease of the endotoxicity of LPS is the neutralization of the its negative charges. The binding process of HDL to the endotoxins is accompanied by a dramatic decrease of the LPS immu- nostimulatory activity which is strongest when HDL is added before LPS to the cells (Fig. 10). One possible explanation is the change of the aggregate structure from a mixed unilamellar/cubic into a multilamellar one (Fig. 6). In the former structures, the binding structures (epitopes) may be accessible to proteins. Within the multilamellar stacks, in contrast, the epitopes of the endotoxins are more or less hidden, thus leading to a considerable decrease of interacting molecules such as LBP, soluble (s) or membrane-bound (m) CD14 (sCD14 and mCD14), or other receptor proteins on the cell surface [25–30]. Another pathway, however, is also probable. HDL by itself incorporates into phospholipid liposomes (Fig. 9B,C) which is also valid for the [LPS]– [HDL] complex (Fig. 9D), that means there is some similarity to the action of LBP [17]. After incorporation of these molecules into target membranes, a process which is also enhanced by the action of LBP (Fig. 9A), the decrease of cell activation may be understood in the light of our conformational concept [31]. Only those LPS with a conical shape of their lipid A moiety, corresponding to an inverted (cubic, H II ) aggregate structure, represent a sufficiently high sterical stress at the site of a signaling protein such as the ion channel Maxi K [32] to induce cell signaling and, with that, cytokine induction. LPS with a lipid A moiety having a cylindrical shape are not able to induce this stress. They are therefore agonistically inactive, but may block the action of active endotoxins by occupying the binding-sites [33]. According to this model and the present data, the reaggregation of the lipid A moiety due to HDL binding from a cubic into a multi- lamellar structure would correspond to a change from a conical into a cylindrical molecular conformation, and would thus explain the loss of its ability to induce cytokine production. ACKNOWLEDGMENTS We are indebted to C. Hamann, and U. Diemer for performing fluorescence spectroscopic and Limulus amebocyte lysate measure- ments, respectively. The expert help of B. Fo ¨ lting for performing the differential scanning calorimetry experiments is kindly acknowledged. We thank S.D. Carroll (XOMA Corporation, Berkely, CA, USA) for the kind gift of LBP. This work was financially supported by the Deutsche Forschungsg- emeinschaft (projects Br 1070/3–1 and SFB 367/B8). 5980 K. Brandenburg et al. (Eur. J. Biochem. 269) Ó FEBS 2002 REFERENCES 1. Levels, J.H.M., Abraham, P.R., Van den Ende, A. & van Deventer, S.J.H. (2001) Distribution and kinetics of lipoprotein- bound endotoxin. Infect. Immun. 69, 2821–2828. 2. Casas, A.T., Hubsch, A.P., Rogers, B.C. & Doran, J.E. (1995) Reconstituted high-density lipoprotein reduces LPS-stimulated TNFa. J. Surg. Res. 59, 544–552. 3. Pajkrt, D., Doran, J.E., Koster, F., Lerch, P.G., Arnet, B., van de Poll, T., ten Cate, J.W. & van Deventer, S.J.H. (1996) Antiin- flammatory effects of reconstituted high-density lipoprotein dur- ing human endotoxemia. J. Exp. Med. 184, 1601–1608. 4. Flegel, W.A., Baumstark, M.W., Weinstock, C., Berg, A. & Northoff, H. (1993) Prevention of endotoxin-induced monokine release by human low- and high-density lipoproteins and by apolipoprotein A-I. Infect. Immun. 61, 5140–5146. 5. YuB.,Hailman,E.&Wright,S.D.(1997)Lipopolysaccharide binding protein and soluble CD14 catalyze exchange of phos- pholipids. J. Clin. Invest. 99, 315–324. 6. Galanos, C., Lu ¨ deritz, O. & Westphal, O. (1969) A new method for the extraction of R. lipopolysaccharides. Eur. J. Biochem. 9, 245–249. 7. Schromm, A.B., Brandenburg, K., Loppnow, H., Za ¨ hringer, U., Rietschel, E.Th., Carroll, S.F., Koch, M.H.J., Kusumoto, S. & Seydel, U. (1998) The charge of endotoxin molecules influences their conformation and IL-6 inducing capacity. J. Immunol. 161, 5464–5471. 8. Kauppinen, J.K., Moffat, D.J., Mantsch, H.H. & Cameron, D.G. (1981) Fourier self-deconvolution: a method for resolving intrinsically overlapped bands. Appl. Spectrosc. 35, 271–276. 9. Ju ¨ rgens, G., Mu ¨ ller, M., Garidel, P., Koch, M.H.J., Nakakubo, H., Blume, A. & Brandenburg, K. (2002) Investigation into the interaction of recombinant human serum albumin with Re-lipopolysaccharide and lipid A. J. Endotoxin Res. 8, 115– 126. 10. Blume, A. & Garidel, P. (1999) Lipid model membranes and biomembranes. In The Handbook of Thermal Analysis and Calorimetry (Gallagher, P.K. & Kemp, R.B., eds), pp. 109–173. Elsevier,Amsterdam,theNetherlands. 11. Koch, M.H.J. & Bordas, J. (1983) X-ray diffraction and scattering on disordered systems using synchrotron radiation. Nucl. Instr. Methods 208, 461–469. 12. Gabriel, A. (1977) Position-sensitive X-ray detector. Rev. Sci. Instrum. 48, 1303–1305. 13. Boulin, C., Kempf, R., Koch, M.H.J. & McLaughlin, S.M. (1986) Data appraisal, evaluation and display for synchrotron radiation experiments: hardware and software. Nucl. Instr. Methods A249, 399–407. 14. Brandenburg, K., Funari, S.S., Koch, M.H.J. & Seydel, U. (1999) Investigation into the acyl chain packing of endotoxins and phospholipids under near physiological conditions by WAXS and FTIR spectroscopy. J. Struct. Biol. 128, 175–186. 15. Brandenburg, K., Ju ¨ rgens, G., Mu ¨ ller, M., Fukuoka, S. & Koch, M.H.J. (2001) Biophysical characterization of lipopolysaccharide and lipid A inactivation by lactoferrin. Biol. Chem. 382, 1215– 1225. 16. Schromm, A.B., Brandenburg, K., Rietschel, E.Th., Flad, H D., Carroll, S.F. & Seydel, U. (1996) Lipopolysaccharide binding protein (LBP) mediates CD14-independent intercalation of lipo- polysaccharide into phospholipid membranes. FEBS Lett. 399, 267–271. 17. Gutsmann, T., Schromm, A.B., Koch, M.H.J., Kusumoto, S., Fukase, K., Oikawa, M., Seydel, U. & Brandenburg, K. (2000) Lipopolysaccharide-binding protein-mediated interaction of lipid A from different origin with phospholipid membranes. Phys. Chem. Chem. Phys. 2, 4521–4528. 18. Gallati, H. (1982) Interferon: wesentlich vereinfachte, enzym- immunologische Bestimmung mit zwei monoklonalen Antiko ¨ rp- ern. J. Clin. Chem. Clin. Biochem. 20, 907–914. 19. Friberger, P., So ¨ rskog, L., Nilsson, K. & Kno ¨ s, M. (1987) The use of a quantitative assay in endotoxin testing. Prog. Clin. Biol. Res. 231, 149–169. 20. Morrison, D.C. & Jacobs, D.M. (1976) Binding of polymyxin B to the lipid A portion of bacterial lipopolysaccharides. Immunochemistry 13, 813–818. 21. Ju ¨ rgens, G., Mu ¨ ller, M., Koch, M.H.J. & Brandenburg, K. (2001) Interaction of hemoglobin with enterobacterial lipopolysaccharide and lipid A: physicochemical charcterization and biological activity. Eur. J. Biochem. 268, 4233–4242. 22. Arrondo, J.L.R., Muga, A., Castresana, J. & Goni, F.M. (1993) Quantitative studies of the structure of proteins in solution by Fourier-transform infrared spectroscopy. Prog. Biophys. Molec. Biol. 59, 23–56. 23. Takada, H., Kotani, S., Tanaka, S., Ogawa, T., Takahashi, I., Tsujimoto,M.,Komuro,T.,Shiba,T.,Kusumoto,S.& Kusunose, N. (1988) Structural requirements of lipid A species in activation of clotting enzymes from the horseshoe crab, and the human complement cascade. Eur. J. Biochem. 175, 573–580. 24. Phillips, J.C., Wriggers, W., Li, Z., Jonas, A. & Schulten, K. (1997) Predicting the structure of apolipoproteins A-I in reconstituted high-density lipoprotein disks. Biophys. J. 73, 2337–2346. 25. Gallay,P.,Heumann,D.,LeRoy,D.,Barras,C.&Glauser,M.P. (1993) Lipopolysaccharide-binding protein as a major plasma protein responsible for endotoxemic shock. Proc. Natl Acad. Sci. USA 90, 9935–9938. 26. Hailman, E., Lichenstein, H.S., Wurfel, M.M., Miller, D.S., Johnson,D.A.,Kelley,M.,Busse,L.A.,Zukowski,M.M.& Wright, S.D. (1994) Lipopolysaccharide (LPS)-binding protein accelerates the binding of LPS to CD14. J. Exp. Med. 179, 269– 277. 27. Delude, R.L., Savedra, R., Zhao, H., Thieringer, R., Yamamoto, S., Fenton, M.J. & Golenbock, D.T. (1995) CD14 enhances cel- lular responses to endotoxin without imparting ligand-specific recognition. Proc.NatlAcad.Sci.USA92, 9288–9292. 28. Wright, S.D., Ramos, R.A., Tobias, P.S., Ulevitch, R.J. & Mathison, J.C. (1990) CD14, a receptor for complexes of lipo- polysaccharide (LPS) and LPS binding protein. Science 249, 1431–1433. 29.Chow,J.C.,Young,D.W.,Golenbock,D.T.,Christ,W.J.& Gusovsky, F. (1999) Toll-like receptor-4 mediates lipopoly- saccharide-induced signal transduction. J. Biol. Chem. 274, 10689– 10692. 30. Kopp, E.B. & Medzhitov, R. (1999) The Toll-receptor family and control of innate immunity. Curr. Opin. Immunol. 11, 13–18. 31. Seydel, U., Schromm, A.B., Blunck, R. & Brandenburg, K. (2000) Chemical structure, molecular conformation, and bioactivity of endotoxins. Chem. Immunol. 74, 5–24. 32. Blunck, R., Scheel, O., Mu ¨ ller, M., Brandenburg, K., Seitzer, U. & Seydel, U. (2001) New insights into endotoxin-induced activation of macrophages: Involvement of a K + channel in transmembrane signaling. J. Immunol. 166, 1009–1015. 33. Schromm, A.B., Brandenburg, K., Loppnow, H., Moran, A.P., Koch, M.H.J., Rietschel, E.Th & Seydel, U. (2000) Biological activities of lipopolysaccharides are determined by the shape of their lipid A portion. Eur. J. Biochem. 267, 2008–2013. Ó FEBS 2002 HDL interaction with endotoxins (Eur. J. Biochem. 269) 5981 . classified as very-low density, low-density and high-density lipoproteins (HDL) according to their buoyant density. The primary function of these lipo- proteins. hysteresis for the re-crystallization of the acyl chains from the liquid crystal- line to the gel phase. The maximum of the heat capacity curve of the 1st cooling