Lipopolysaccharide and ceramide docking to CD14 provokes ligand-specific receptor clustering in rafts

2.1 Ligand-induced clustering of CD11b and CD14 within rafts

In a previous study, co-association of CD11b and CD14 on LPS-activated neutrophils was demonstrated using FRET technology 18. Therefore, human monocytes were stimulated in vitro with LPS, fMLP and PMA and FRET between CD11b and CD14 was measured. Co-assembly was induced by LPS, but not by fMLP or PMA (Fig. 1a). This effect reached a maximum near 20 ng/ml of LPS (Fig. 1b). On resting monocytes, in contrast, neither FRET analysis nor confocal microscopy revealed proximity between CD14 and CD11b (Fig. 1c). Microscopy confirmed the co-localization of CD14 and CD11b following activation by LPS (Fig. 1d).

Ceramide-stimulation partially mimics LPS-responses in different cellular systems 8. In our experiments ceramide induced FRET between CD11b and CD14 similar to LPS (Fig. 1b) and saturation was reached near 2 nM. Confocal microscopy again confirmed the co-association (Fig. 1e). The potent LPS-antagonist compound-406 19 which did not induce clustering of CD14 and CD11b (Fig. 1f) significantly reduced the co-association of CD14 and CD11b induced by either LPS or ceramide indicating a similar mechanism of interaction. Further evidence for an interaction of ceramide and CD14 was provided by the specific binding of [14C]-N-Palmitoyl-D-Sphingosine (C16-ceramide) to CD14-transfected CHO-fibroblasts (Fig. 1g).

Cytochalasin D, a potent inhibitor of LPS internalization 20, inhibited the uptake of a fluorescent LPS derivative (FITC-LPS) by 59±8% as assessed by flow cytometry. The LPS- or ceramide dependent co-localization of CD14 and CD11b, in contrast, was not attenuated (Fig. 1f), suggesting that receptor clustering was probably not dependent on ligand internalization.

To further identify agonists inducing co-assembly of CD14, monocytes were incubated with lipoteichoic acid, sulfogalactosylceramide, and different glycero-phospholipid liposomes. Only lipoteichoic acid, phosphatidylethanolamine and phosphatidylinositol induced a co-assembly of CD11b and CD14 (Fig. 2a).

Recently, HSP70 was found to stimulate cytokine production through a CD14 dependent pathway 10. HSP70 in its native form in our experiments induced co-assembly of CD11b and CD14 (Fig. 2b), while the delipidated form (dHSP70) was not effective. Since HSP70 binds sulfogalactosylceramide 9, monocytes were incubated with dHSP70 recombined with this ceramide derivative. Complexes of sulfogalactosylceramide and dHSP70 had the same effect on clustering of CD11b and CD14 as native HSP70 (Fig. 2b), while sulfogalactosylceramide alone was not effective (Fig. 2a).

CD14 has been shown to be associated with lipid-enriched low density domains resembling rafts in monocytic THP-1 cells 12. This was further analyzed by modification of plasma membrane cholesterol composition with cyclodextrins 21. Cholesterol loading led to decreased expression of CD14 and CD11b (Fig. 3a) and completely inhibited the LPS induced co-assembly of CD14 and CD11b (Fig. 3b). A relatively milder cholesterol depletion, in contrast, did not affect the expression densities of CD14 and CD11b. The co-association of both receptors upon stimulation with LPS or ceramide, however, was significantly reduced. Finally, raft specific staining (Fig. 3c–e) revealed that cholesterol loading strongly increased the size and number of rafts while cholesterol depletion did not affect raft structure.

2.2 Characterization of surface receptors in the proximity of CD14

Because large signaling receptor complexes with numerous protein components have been reported for example for T cells 22, we now investigated the sterical association of further receptors with CD14 on LPS- or ceramide-activated monocytes. In resting cells, CD14 was found to be clustered with the GPI-R decay accelerating factor (CD55), a complement regulatory protein, the Fcγ-receptors CD32 and CD64, and with the integrin associated protein CD47 (Table 1). LPS stimulation, in comparison, induced clustering of CD14 with CD11b and CD18, TLR-4, the Fcγ-receptor CD16a and the scavenger receptor CD36. CD47 was no longer associated with CD14 following LPS stimulation but the tetraspanin CD81 appeared in the complex. LTA induced the clustering of the same receptor complex components as LPS.

The ceramide-induced cluster differs from that induced by LPS and includes: CD14, CD11b/CD18, CD55, CD47, CD32 and CD64, and CD36 (Table 1). Complexes of sulfogalactosylceramide and HSP70 induced a receptor cluster which was similar to that induced by ceramide.

Thus, LPS, LTA, ceramide, and sulfogalactosylceramide/HSP70 induce the partitioning of membrane receptors into two different clusters, despite ligation of the same receptor, CD14. As control for unspecific Fc-receptor binding, energy transfer was measured between CD16a, CD32, CD64 and irrelevant (non-)monocytic monoclonal antibodies (CD3, Vß8, CD25 and CD91). Furthermore, stimulation of cells with fMLP or PMA as a control did not lead to a significant energy transfer between any pair of the antibodies (data not shown).

The different patterns of receptor co-clustering which are detected by FRET in LPS- or ceramide-stimulated monocytes may represent both minor changes in the intermolecular distances of receptors within a given membrane microdomain as well as lateral sorting of receptors between membrane compartments. Therefore, rafts were isolated from monocytes as Triton X-100-insoluble complexes that float to a low density (0.65 and 0.5 M sucrose) during density gradient centrifugation 17. Both CD11b and CD14 floated in these low-density fractions independent of the activation state of monocytes (Fig. 4). CD16a, however, was detected in these compartments only upon stimulation with LPS.

2.3 Co-assembly of monocyte receptors with CD14 in inflammatory diseases

CD14 activation plays a role in the pathogenesis of infectious diseases and other inflammatory diseases such as coronary artery disease (CAD). Therefore, FRET analysis was performed on monocytes from patients suffering from sepsis, stroke and CAD (Table 2a). Monocytes from all patients but not the control group showed a co-association of CD11b and CD14 without prior in vitro stimulation.

While LPS represents a candidate for monocyte activation in sepsis, it is unlikely to interact with monocytes in the two groups of patients with inflammatory disorders of the vessel wall which are linked to disturbances in lipid metabolism. Since atherogenic lipoproteins have been reported to contain elevated levels of ceramide 23, plasma ceramide levels of patients and controls were determined by tandem mass spectrometry. Significantly different plasma concentrations of the ceramide species C22, C23, C24:0, and/or C24:1 were found in CAD, stroke and sepsis patients as compared to controls (Table 2b). Sepsis showed the most pronounced alteration in ceramide species with elevated C24:1 and C22, and decreased C24:0 and C23 levels.

Finally, the receptor clusters on monocytes in patients with sepsis and CAD were analyzed in more detail. Based on a cell-by-cell determination of FRET efficiency, co-assembly of CD11b and CD14 was observed on the majority of monocytes of patients with sepsis and CAD (Fig. 5a). Next we measured the co-association of CD81 with CD14 which had differed in vitro between LPS and ceramide stimulation. A co-assembly of CD81 and CD14 was only observed in the sepsis patients but not in CAD patients (Fig. 5b).

4.1 Blood samples

Heparinized blood samples were obtained from healthy volunteers. After informed consent and with the approval of the local Ethical Committee (No. 00/33), blood samples were also obtained from patients with sepsis, coronary artery disease and stroke. The diagnosis of sepsis was based on the consensus criteria published by the American College of Chest Physicians and the Society of Critical Care Medicine. The diagnosis of infection was based on the criteria of the Center of Disease Control, Atlanta, GA. Coronary artery disease (CAD) was defined by coronary angiography and blood samples were obtained the day following acute myocardial infarction or during angina. Blood samples of patients with acute stroke were obtained within 12 h after infarction. All patients presented middlecerebral artery territory infarction proven by cranial computer tomography. Blood samples of all patients were analyzed for LPS in a chromogenic limulus amebocyte lysate (LAL) assay (QCL-100, BioWhittaker, Walkersville, MD). Half of the sepsis patients showed a positive result whereas all other samples were negative for LPS.

4.2 Preparation of phospholipid liposomes

Phosphatidylethanolamine (PE), phosphatidylcholine (PC), phosphatidylinositol (PI) and phosphatidylserine (PS) were obtained from Sigma (Deisenhofen, Germany) as chloroform solutions. Lipids were dried under vacuum and resuspended in Dulbecco's modified phosphate buffer saline without Ca²⁺ and Mg²⁺ (PBS). Lipid content was determined by thin layer chromatography.

4.3 Delipidation of HSP70

HSP70 was obtained from Stressgen (Victoria, Canada). HSP70 (200 μg) was delipidated in ethanol/diethylether (3:1) at –20°C overnight. After centrifugation (15 min at 1,000 g) the pellet was delipidated in diethylether at –20°C for 4 h. After centrifugation (15 min, 1,000) delipidated HSP70 (dHSP70) was transferred into 50 nM Tris-HCl pH 7.5, 100 mM NaCl, 1 mM dithiothreitol and 0.1 mM phenylmethylsulfonyl fluoride.

4.4 Stimulation of blood samples

Whole blood aliquots (100 μl) were incubated for 15 min at 37°C with the following stimulatory compounds: LPS from Salmonella minnesota (1–100 ng/ml), ceramide containing primarily stearic and nervonic acids (Cer, 1 nM – 0.5 μM); lysophosphatidylcholine (10 μM); phorbol 12-myristate 13-acetate (PMA, 1 μM); N-formyl-L-methionyl-L-leucyl-L-phenylalanine (fMLP, 1 μM); PE, PC, PI and PS-liposomes (10 μM) prepared as described above. All reagents were obtained from Sigma. Cells were also incubated with LTA (1 μg/ml), HSP70 (7 nM), dHSP70 (7 nM) and sulfogalactosylceramide (Sigma, 40 nM) together with dHSP70 (7 nM). LTA (bacillus subtilis) was prepared as described before 30. After incubation the cells were washed with cold PBS containing 0.1% NaN₃. The synthetic LPS antagonist compound-406 (1–400 ng/ml) was kindly provided by Dr. Shoichi Kusumoto. The cells were preincubated with the LPS antagonist for 15 min at 37°C and then stimulated with LPS or ceramide without washing. All compounds despite LPS used for cellular stimulation were negative for LPS in the LAL assay.

4.5 Inhibition of LPS internalization

Whole blood was incubated with cytochalasin D (Sigma, 3 μM) for 30 min at 37°C, then further stimulated with LPS (40 ng/ml) and ceramide (40 nM) as described and the energy transfer between CD14 and CD11b was measured. Cells were also further incubated with FITC-LPS (Sigma, 400 ng/ml) and internalization was measured by flow cytometry using trypan blue as quencher of extracellular fluorescence.

4.6 Binding of [¹⁴C]N-palmitoyl-D-sphingosine to CHO-cells

CHO cells transfected with human CD14 or with pPOL-DHFR vector were cultured and prepared for labeling as previously described 31. CHO-cells (4×10⁵) were preincubated on ice in the presence of a 50-fold excess of unlabeled N-Palmitoyl-D-Sphingosine (Sigma) for 15 min. Then [¹⁴C]N-palmitoyl-D-sphingosine (American Radiolabeled Chemical Inc., St. Louis, MO) was added and cells were incubated for another 15 min. Cells were washed and the pellet was dissolved in 100 μl of 1% SDS, 10 mM EDTA and added to 3 ml of scintillation fluid. Cell-bound radioactivity was determined by liquid scintillation counting.

4.7 Immunostaining of cells

Stimulated or control washed whole blood samples (200 μl) were incubated for 15 min on ice with saturating concentrations of the fluorochrome conjugated or biotinylated mAb as shown in Table 3. Similar results were obtained for both antibodies against CD81. Lysis of erythrocytes and washing were performed as described 3. After the last washingstep, samples were divided and one part of each sample was incubated with a saturating amount of SA-Cy5 (0.6 μg/ml) for 15 min on ice and washed before flow cytometric measurement. For analysisof receptor co-association by confocal multiparameter microscopy, lysed cells were seated on borosilicated coverslips (Nunc, Wiesbaden, Germany).

4.8 Modifications in the raft structure

Mononuclear cells were isolated from blood by Histopaque 1077 (Sigma) density gradient centrifugation. Methylated β-cyclodextrin (Sigma) was used for cholesterol loading of the cell membrane 21. Aliquots of mononuclear cells containing 5×10⁵ cells/μl were re-suspended in 3 ml of PBS and incubated for 30 min at 37°C with complexes of cholesterol (100 μg/ml) and methyl-β-cyclodextrin (4.6 mg/ml) prepared as described 21. Cells were incubated with 2-hydroxy-propyl-β-cyclodextrin (Sigma, 0.15 mg/ml) for 30 min at 37°C to deplete cholesterol. After incubation with cyclodextrins the cells were washed, immunostained as described above, and the expression of the antigens was measured by flow cytometry. For energy transfer experiments 200-μl aliquots of whole blood were treated with cholesterol/methyl-cyclodextrin (C/C) (100 μg/ml / 4.6 mg/ml) or 2-hydroxy-propyl-β-cyclodextrin (0.15 mg/ml) at 37°C for 30 min, then stimulated with LPS (40 ng/ml) or ceramide (40 nM) as described and the energy transfer between CD14 and CD11b was measured. For microscopic determination of raft size and structure, monocytes from healthy volunteers were isolatedby leukapheresis followed by elutriation. Cells were cultured on borosilicated coverslips (Nunc, Wiesbaden, Germany) in macrophage SFM medium (Gibco BRL, Karlsruhe, Germany) overnight. Cells were loaded with cholesterol or cholesterol was depleted for 12 h as described above.

4.9 Measurement of FRET by flow cytometry

A dual-laser FACSCalibur flow cytometer and Cellquest software (Becton Dickinson, San José, CA) were used for the measurements. Measurements were performed without compensation. 50,000 leukocytes were acquired in list mode using forward scatter (FSC) as the triggering signal.

4.10 Determination of FRET efficiency (acceptor sensitized emission or donor quenching)

The principle of FRET first described by Förster in the late 1940s allows the measurement of distances between surface molecules 32. The energy transfer parameter (ET_p), which is proportional to FRET efficiency (ET), was calculated according to (1) where A is acceptor, D is donor, FL2 is mean fluorescence in channel 2 (488 → 530 nm) (donor, R-PE), FL3 is mean fluorescence in channel 3 (488 → 585 nm), FL4 is mean fluorescence in channel 4 (633 → 670 nm, acceptor, Cy5) (each value obtained after autofluorescence subtraction):

ET_p = FL3(D,A) – FL2(D,A)/a – FL4(DA)/b FL3(D,A) (1)

a = FL2(A)/FL3(A)

b = FL4(D)/FL3(D)

FRET efficiency (ET) was calculated from quenching according to (2):

ET = FL2(D) – FL2(D,A) FL2(D) (2)

According to the publication by Szöllösi et al. 33 an ET = 5% was defined as the threshold level for significant transfer efficiency in preliminary experiments. This level was experimentally confirmed as the detection limit for the proportional ET_p value (data not shown).

4.11 Measurement of plasma ceramide levels

Plasma ceramide levels were quantified by Tandem mass spectometry as described previously 34. The measurement was performed with an API 365 triple quadrupol system (Perkin Elmer, Norwalk, CT) equipped with a turbo ion spray interface.

4.12 Microscopic characterization of raft structure

For analysis of raft structure samples were labeled with 50 μg/ml raft-specific 1,2-dimyristoyl-sn-glycero-3-phospho-ethanolamine)-Cy5 (DMPE-Cy5) and analyzed in epi-illustration (Axiovert 135TV, Zeiss, Jena, Germany) as described earlier 16.

A TCS 4D inverted confocal laser scanning microscope (Leica Lasertechnik, Heidelberg, Germany), with an argon-crypton mixed ion laser was used for multiparameter cell analysis. Monocytes were selected according to positive CD14-expression. Overlay images were generated after separate analysis of R-PE (CD11b) and Cy5 (CD14) dependent fluorescence.

4.13 Cell harvesting

For raft-specific microscopy and western blot analysis peripheral blood monocytes from healthy volunteers were separated by leukapheresis as described above and cultured in ultra-low attachment plates (Corning, Wiesbaden, Germany) in macrophage SFM medium over night and stimulated with LPS (40 ng/ml) or ceramide (40 nM) in the presence of 5% human serum for 15 min at 37°C. Cells were washed twice in 50 ml ice-cold TNE-buffer (150 mM NaCl, 50 mM Tris-HCl, pH 7.5, 5 mM EDTA and 1 mM PMSF) and pelleted at 500 g for 5 min.

4.14 Detergent lysis and sucrose flotation gradients

Triton X100 lysis and sucrose density gradient floatation was performed as described 17 using a SW55Ti ultracentrifuge (Beckman/Coulter/Immunotech).

4.15 Immunoblotting

Immunoblotting was performed using the ECL plus detection system (Amersham Pharmacia Biotech, Braunschweig, Germany). The following antibodies were used: anti-CD14 (goat polyclonal, clone N-15, Santa Cruz, Heidelberg, Germany), anti-CD11b (goat polyclonal, clone C-19, Santa Cruz), anti-CD16a (mouse monoclonal, clone 3G8, Dianova).

4.16 Statistical analysis

Data are given as mean ± standard deviation (SD). For statistical comparison Wilcoxon signed-ranks test was used. (*) designates p<0.05.