Published online before print March 27, 2007
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
,1
* Department of Laboratory Medicine, Nordland Hospital, Bodø, Norway;
Institute of Medical Biology, University of Tromsø, Tromsø, Norway;
Tanox Inc., Houston, Texas, USA; and
Institute of Immunology, Rikshospitalet University Hospital and University of Oslo, Oslo, Norway
1 Correspondence: Department of Laboratory Medicine, Nordland Hospital, N-8092 Bodø, Norway. E-mail: ole.lars.brekke{at}nlsh.no
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Key Words: sepsis • endotoxin • complement receptor 3
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
E. coli entering the bloodstream are opsonized rapidly by complement including C1q, mannose-binding lectin (MBL), C4b, and C3b/iC3b [8 ], in addition to Igs [9 ]. The terminal C5b-9 complex (TCC) induces lysis of certain complement-sensitive bacterial strains. The opsonization of bacteria by complement also facilitates the binding of the bacteria to the adherence complement receptor 1 (CR1) and the integrin and phagocytosis receptor CR3 (CD11b/CD18) on blood leukocytes [10 ]. Complement activation in the fluid phase with the release of C5a up-regulates CD11b rapidly and induces oxidative burst [11 ]. C5a is also involved in the defect phagocytosis and oxidative burst in granulocytes observed in the cecal ligation/puncture (CLP)-induced sepsis in rats [12 ]. Furthermore, C5a mediates a number of other detrimental effects during sepsis including enhanced synthesis of inflammatory mediators and degranulation of granulocytes [13 ]. Thus, complement is a double-edged sword and plays a dual role during sepsis.
LPS is a major constituent of the membrane in gram-negative bacteria and may be released from the bacteria when they multiply or lyse [14
]. LPS induces a septic-like condition in vivo including fever, increased body temperature, and synthesis of pro- and anti-inflammatory cytokines by different cell types including macrophages [14
]. LPS binds to soluble or membrane CD14, which functions as a receptor for LPS [15
], after binding of LPS to a LPS-binding protein [16
]. CD14 in monocyte/macrophage cell membranes are attached through a GPI linkage [16
], and the transmembrane signal is mediated through TLR4 [17
] and MyD88 [18
]. The nonreceptor-mediated signaling by LPS is also mediated through CD14 [19
]. LPS and CD14 are therefore important molecules involved in the inflammation induced during gram-negative sepsis. However, blocking of CD14 using mAb in animal models inhibits LPS effects [20
] but has no effect on the mortality during the CLP model of sepsis in mice [21
]. It is interesting that LPS has been reported to activate the transcription factor NF-
B through CR3, indicating a possible link between complement and LPS signaling [22
].
We have shown previously that a specific C5a receptor antagonist (C5aRa) reduced the E. coli-induced expression of the adhesion molecule CD11b dose-dependently [11 ]. It is notable that the C5aRa did not inhibit E. coli-induced CD11b up-regulation completely, especially in monocytes, suggesting that other signaling pathways might be involved.
We hypothesized that a combined inhibition of complement and CD14 inhibition would reduce leukocyte activation. Thus, in the present study, the role of complement C3 opsonization, C5a, and LPS/CD14 on E. coli-induced CD11b up-regulation, phagocytosis, and oxidative burst was studied using a novel human whole blood model, where anticoagulation does not interfere with complement activation [11 ]. The results showed that the E. coli-induced granulocyte activation was more dependent on complement, and monocyte activation was more dependent on CD14. It is interesting that the activation of both cell types was blocked efficiently by combining complement and CD14 inhibition.
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Antibodies and inhibitors
The murine IgG1 mAbs to human C2 (clone 175-62), Factor D (Clone 166-32) [23
], anti-C5 (137-30), blocking the cleavage of C5, anti-C5/C5a (clone 137-26) [24
], and the isotype-matched control anti-HIV-1 gp120 (clone G3-519) were from the laboratory of M. Fung (Tanox Inc., Houston, TX, USA). Murine anti-human CD14 (clone 18D11) F(ab')2 and a control F(ab')2 were obtained from Diatec AS (Oslo, Norway). The C5aRa and a corresponding control peptide were from the laboratory of M. Fung. The maximal inhibitory dose of each mAb and inhibitor was determined in initial dose-response experiments. All antibodies were checked for LPS contamination using an endotoxin assay described below and purified using END-X®B15 (Associates of Cape Cod Inc.) if necessary.
Bacterial strains and counting
Heat-inactivated E. coli and FITC-labeled E. coli Strain LE392 were from Orpegen Pharma. In some experiments, E. coli Strain LE392 (ATCC 33572) from American Type Culture Collection (Manassas, VA, USA) was grown in Luria-Bertani medium (1% Tryptone, 0.5% yeast extract, 1% NaCl in H2O) overnight and then washed once in PBS (3220 g, 10 min) at +4°C. After resuspension in PBS, E. coli was heat-inactivated at 60°C for 60 min. E. coli was then washed at +4°C in PBS without Ca2+ and Mg2+ and stored at –70°C in PBS. E. coli was counted after staining with Syto BC using a bacteria counting kit (Molecular Probes). To obtain absolute counts, an aliquot of the bacteria was counted on the flow cytometer using Trucount tubes from Becton Dickinson, and 2500 beads were acquired. Disrupted E. coli bacteria were obtained by pulse sonication on ice (0°C) using a Bandelin Sonopuls HD2070 sonicator equipped with a MS 72 probe from Bandelin Electronic Gmbh and Co. (Berlin, Germany). After sonication, bacterial counting on flow cytometry was used to verify that all bacteria were disrupted.
Whole blood model of sepsis
Blood from healthy donors was obtained after informed consent according to guidelines from the local ethics committee. Experiments with each blood donor were performed as single experiments at different time-points. In brief, fresh human whole blood was collected in polypropylene tubes containing lepirudin (50 mg/L). Aliquots of whole blood in sterile polypropylene tubes were then preincubated for 4 min in a water bath at 37°C with PBS, antibodies, or inhibitors in PBS (14.3% of total vol, v/v) as indicated. E. coli or PBS (14.3% of total vol, v/v) was then added, and samples were incubated 10, 20, or 30 min further as indicated. After incubation, samples were processed immediately for flow cytometry. Aliquots for assay of complement activation products were placed on ice (0°C) immediately and supplemented with EDTA (10 mM final concentration) to stop further activation. Samples were centrifuged (3220 g, 15 min at 4°C), and plasma was stored at –70°C until further analysis.
LPS activation of complement in lepirudin plasma and serum
Lepirudin-anticoagulated whole blood and clotted whole blood were centrifuged at 3220 g for 15 min at 4°C. Plasma or serum from three or four different donors were pooled and frozen at –70°C until the experiments were performed. Plasma or serum was preincubated with PBS for 5–6 min and then incubated with E. coli LPS in tenfold titrations from 1000 to 0.001 mg/L for 30 min at 37°C. E. coli bacteria were similarly added in equivalent LPS concentrations. Immediately after incubation at 37°C, EDTA (10 mM final concentration) was added to stop further complement activation, the samples were centrifuged at +4°C, and the supernatants were stored at –70°C. The experiment was performed three times using three different plasma and serum pools.
Flow cytometric analysis
Granulocyte and monocyte CD11b expression was analyzed using anti-CD11b PE (Becton Dickinson, San Jose, CA, USA) after fixation of the cells with 0.5% (v/v) paraformaldehyde as described previously [11
]. Whole blood was stained with the nuclear dye LDS-751 [fluorescence 3 (FL3)] to separate leukocytes from RBC. Monocytes and granulocytes were gated separately in a plot displaying side-scatter versus FL1 [anti-CD14 FITC (Becton Dickinson, San Jose, CA, USA)]. Phagocytosis of FITC-labeled E. coli was analyzed by flow cytometry after 10 min incubation using the Phago test kit (Orpegen Pharma) and quenching solution to avoid interference of surface-bound bacteria as described previously [11
]. Phagocytosis was expressed as median fluorescence intensity (MFI) of the whole granulocyte or monocyte population. In some experiments, phagocytosis was also expressed as percent-positive cells using a marker set on the upper fluorescence limit in the negative control consisting of E. coli Strain LE392 without FITC labeling incubated 10 min in whole blood. Cells with a higher fluorescence were regarded as positive cells.
Oxidative burst
Oxidative burst was measured using the Burst test (Orpegen Pharma). Samples were processed immediately (baseline sample T0) or after 10 min incubation by transferring 100 µL blood to 5 mL polypropylene tubes (Becton Dickinson, Franklin Lakes, NJ, USA), and 20 µL substrate solution containing dihydrorhodamine 1,2,3 was added. All tubes were then incubated 10 min at 37°C, and samples were lysed and washed according to kit instructions. Within 1 h, monocytes and granulocytes in PBS were gated separately in a forward/side-scatter dot-plot using flow cytometry. Results were expressed as MFI of the whole granulocyte or monocyte population.
Analysis of E. coli C3 opsonization in lepirudin and EDTA plasma
Lepirudin-anticoagulated whole blood and EDTA whole blood were centrifuged at 3220 g for 15 min at 4°C, and plasma was separated from blood cells. EDTA plasma and lepirudin plasma were preincubated 4 min at 37°C with PBS or antibodies, respectively. E. coli was added, and samples were further incubated 10 min, washed twice (3220 g, 10 min, 4°C), and resuspended with PBS containing 0.1% (w/v) BSA.
C3 opsonization of whole E. coli bacteria was detected by flow cytometry using FITC-conjugated rabbit anti-human C3c (F0201, Dako, Glostrup, Denmark) and FITC-conjugated rabbit anti-mouse Ig (F0261, Dako) as isotype control. Samples were incubated 15 min at 20°C, and PBS was added before analysis on the flow cytometer. Results were expressed as MFI.
Deposition of donor serum IgG, IgM, C1q, and C3 on E. coli
Heat-inactivated E. coli was incubated 10 min at 37°C in heat-inactivated serum from the healthy blood donors. E. coli antibodies were analyzed by flow cytometry using FITC-conjugated rabbit anti-human IgG (F0056, Dako) or anti-human IgM (F0058, Dako) and an isotype control (X0929, Dako). Serum from a patient with newly diagnosed E. coli sepsis using standard blood culture techniques was used as positive control and was obtained after informed consent according to guidelines from the local ethics committee. Results were expressed as MFI.
The correlation between anti-E. coli IgM and IgG antibodies and complement activation was investigated in the same sera. Heat-inactivated E. coli was incubated 10 min at 37°C in serum from eight healthy blood donors. PBS or heat-inactivated serum was used as a negative control. C1q and C3 opsonization of whole E. coli bacteria was detected by flow cytometry using FITC-conjugated rabbit anti-human C1q (F0254, Dako) or C3d (F0323, Dako). FITC-conjugated rabbit anti-mouse Igs (F0261, Dako) were used as isotype control. Results were expressed as MFI.
Endotoxin assay
Measurements of LPS in buffers, reagents, E. coli, and E. coli supernatant were carried out as an endpoint test using a Limulus amoebocyte lysate-Pyrochrome® kit (Associates of Cape Cod Inc.) or a QCL-1000 kit from Bio-Whittaker (Walkersville, MD, USA). Before the total LPS concentration in whole E. coli bacteria was measured, they were disrupted by sonication as described above. When the free LPS concentration was measured in the supernatants from whole E. coli, the bacteria were centrifuged at 2300 g for 10 min, and the supernatant was filtered through a 0.2-µm Pall filter (Pall Corp., Ann Arbor, MI, USA) to remove whole bacteria.
Assay of complement activation
Activation of the classical pathway was measured as C1rs-C1 inhibitor complexes and analyzed in an enzyme immunoassay (EIA) as described previously [25
]. The alternative convertase complex C3b-Bb-properdin was measured by EIA as described previously [11
]. Activation of the final common pathway was quantified by the C3 activation product C3bc [26
]. C5a was analyzed in an EIA using the neo-epitope-specific mAb 4A2E10E2 as capture antibody and 3G3C4 as secondary antibody as described previously [27
]. The TCC was measured by EIA using the specific mAb aE11 as described previously [28
].
Statistical analysis
The statistical analysis was performed using SigmaStat Version 3.1 for Windows from SPSS Science Software GmbH (Erkrath, Germany). Friedman repeated measures ANOVA on ranks was used followed by multiple comparison versus E. coli plus PBS using Dunn’s method. P < 0.05 was considered statistically significant.
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
 View larger version (14K): [in a new window] |
Figure 1. Effect of complement inhibition on granulocyte activation in human whole blood. The effect of a C5aRa and mAbs to C5 (a-C5), C2 (a-C2), and factor D (a-D) on E. coli-induced C5a formation, CD11b up-regulation, phagocytosis, and oxidative burst in granulocytes after 10 min incubation was examined. Complement activation measured as plasma C5a formation was analyzed using EIA and expressed as arbitrary units (AU)/mL. CD11b up-regulation, phagocytosis, and oxidative burst were measured by flow cytometry and expressed as MFI. The inhibitor concentrations used were C5aRa (7.1 mg/l), anti-C5 137-30 blocking C5 cleavage (71 mg/L), anti-C2 (71 mg/L), anti-factor D (36 mg/L), and combined anti-C2 and anti-factor D mAb (a-C2+a-D; 71+36 mg/L). Controls were an isotype-matched control mAb (Ctr. ab; 71 mg/L) and a control peptide (Ctr. pep.; 7.1 mg/L). Results are given as the median and 25–75 percentile of six different blood donors. *, P < 0.05, compared with E. coli alone (71x106/mL) using Friedman repeated measures ANOVA on ranks followed by one-way multiple comparison versus E. coli using Dunn’s method. There was no statistical difference between the negative controls T0 (baseline sample) and T10 (baseline sample incubated for 10 min without E. coli), and the latter was used for statistical comparison with test samples.
|
 View larger version (18K): [in a new window] |
Figure 2. Effect of complement inhibition on monocyte activation in human whole blood. The effect of complement inhibition on E. coli-induced CD11b up-regulation, phagocytosis (n=5), and oxidative burst in monocytes in the same whole blood samples as described for granulocytes with identical experimental conditions, inhibitors, concentrations, readouts, data presentation, and statistics as described in the legend to Figure 1
. *, P < 0.05 compared to E. coli alone (71 x 106/mL) using Friedman repeated measures analysis of variance on ranks followed by one way multiple comparison versus E. coli using Dunn’s method.
|
 View larger version (16K): [in a new window] |
Figure 3. Effect of complement inhibition on the degree of phagocytosis of E. coli expressed using percent-positive cells as read-out. Granulocyte (left panel) and monocyte (right panel) phagocytosis of FITC-labeled E. coli was measured by flow cytometry after 10 min incubation and calculated as percent-positive cells. Experimental conditions, inhibitors, concentrations, and controls are as described in Figure 1
. Results are given as the median and 25–75 percentile of six and five different blood donors for granulocytes and monocytes, respectively. *, P < 0.05, compared with E. coli alone.
|
|
View this table: [in a new window] |
Table 1. Effect of Complement Inhibition on E. coli C3 Opsonization in Lepirudin Plasma
|
 View larger version (17K): [in a new window] |
Figure 4. Time course of E. coli-, disrupted E. coli-, and purified LPS-induced leukocyte CD11b up-regulation and oxidative burst in human whole blood, which was incubated for 10, 20, or 30 min with whole E. coli ( ), disrupted E. coli ( amp;), purified LPS L-8274 ( ), CVF ( ), and PBS control ( ). Granulocyte (left panels) and monocyte (right panels) CD11b expression (upper panels) and oxidative burst (lower panels) were examined using flow cytometry. Whole E. coli (71x106/mL), disrupted E. coli, and purified LPS were added in equivalent LPS concentrations (0.5 mg/L). CVF (5 U/mL), which activates the C3 convertase directly, was included as control for the effect of pure fluid-phase complement activation. Data from one of three experiments with virtually identical results are shown.
|
 View larger version (16K): [in a new window] |
Figure 5. Dose-response effect of whole E. coli, disrupted E. coli, and purified LPS on complement and leukocyte activation in human whole blood, which was incubated for 10 min with whole E. coli (left panels), equivalent amounts of disrupted E. coli (second panels from left), and purified E. coli LPS L-8274 (second panels from right). CVF (right panels) was included as a control for pure, fluid-phase complement activation. The former three preparations all contained 0.5 mg LPS/L at the highest concentration. Complement activation was measured as TCC formation (top panels; ) and is expressed as AU/mL. CD11b expression (middle panels) and oxidative burst (bottom panels) in granulocytes ( ) and monocytes () were measured using flow cytometry, and results are given as MFI. Data from one of three experiments with virtually identical results are shown.
|
All four preparations up-regulated monocyte CD11b after 10 min incubation (Fig. 5 , middle panels). In comparison, LPS did not up-regulate CD11b in granulocytes, in contrast to the other three preparations, where a dose-dependent effect was observed. The effect of whole bacteria and disrupted E. coli bacteria was comparable on CD11b up-regulation on monocytes and granulocytes.
Oxidative burst was induced only by whole E. coli bacteria (Fig. 5 , bottom panels), more pronounced in granulocytes than in monocytes, indicating that complement activation in the absence of whole E. coli bacteria is not sufficient to stimulate oxidative burst. Furthermore, complement activation in the fluid phase induced by CVF is associated with CD11b up-regulation but is not sufficient to stimulate oxidative burst in the absence of whole E. coli bacteria.
Complement activation induced by purified LPS and whole E. coli bacteria
The effect of purified LPS and whole E. coli bacteria on complement activation in the fluid phase measured as TCC formation was then investigated (Fig. 6
). Whole E. coli bacteria at equivalent LPS concentrations above 0.1 mg/L enhanced TCC formation rapidly in plasma and serum. In comparison, three different E. coli LPS preparations did not activate complement at concentrations up to 1 mg/L. However, LPS at concentrations from 10 mg/L induced TCC formation in plasma and serum. In conclusion, LPS at concentrations present in whole E. coli bacteria is not responsible for E. coli-induced complement activation. LPS at high concentrations activates the complement system, but the free LPS concentrations measured in the whole E. coli preparations used in the present study were 3–5 µg/L, which is far below the complement-activating level.
 View larger version (12K): [in a new window] |
Figure 6. Effect of whole E. coli and three different E. coli LPS preparations on complement activation in human plasma and serum. Human lepirudin-anticoagulated, pooled plasma (upper panel) or pooled serum (lower panel) was incubated for 60 min with whole E. coli () or equivalent amounts of purified E. coli LPS L-8274 (), L-2880 ( ), or L-2630 (). Complement activation was measured as TCC formation and is expressed as AU/mL. Data from one of three experiments with virtually identical results are shown.
|
 View larger version (30K): [in a new window] |
Figure 7. Effect of inhibition of C5a and CD14 and a combination thereof on leukocyte CD11b up-regulation in human whole blood. Dose-response effect of the anti-C5a neutralizing mAb 137-26 (left panel), F(ab')2 fragments of an anti-CD14 mAb (middle panel), and a combination of these (right panel) on CD11b up-regulation in granulocytes (upper panels) and monocytes (lower panels) was examined using flow cytometry. Human whole blood was stimulated with whole E. coli (), disrupted E. coli (amp;), purified LPS (0.5 mg/L; ), or CVF () for 10 min. Data from controls incubated 10 min with PBS only are also indicated (). Data from one of three experiments with virtually identical results are shown.
|
The role of C5a in up-regulation of CD11b and oxidative burst
The effect of rhC5a on CD11b up-regulation in human whole blood was examined. RhC5a up-regulated CD11b rapidly and dose-dependently on granulocytes and monocytes (Fig. 8
, left panel). The rhC5a-induced CD11b up-regulation was blocked completely by the C5aRa in both cell types, supporting the essential role of C5a in CD11b expression (Fig. 8
, right panel). It is interesting that rhC5a did not induce any oxidative burst in granulocytes or monocytes (data not shown). This further supports that complement activation in the fluid phase with release of C5a is necessary, but not sufficient, in E. coli-induced oxidative burst, whereas C3 opsonization of whole E. coli bacteria is necessary for phagocytosis and the subsequent oxidative burst.
 View larger version (11K): [in a new window] |
Figure 8. Effect of rhC5a on granulocyte and monocyte CD11b expression in human whole blood. rhC5a up-regulated CD11b (left panel) dose-dependently on granulocytes () and monocytes () after 10 min incubation. CD11b up-regulation was measured by flow cytometry and is expressed as MFI. The effect of adding increasing concentrations of C5aRa to human whole blood stimulated with rhC5a (5 mg/L) is shown in the right panel with the same symbols as described for the left panel. Data from one of three experiments with similar results are shown.
|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Inhibition of complement C3 opsonization and C5a formation using anti-C2 and anti-factor D mAb in combination effectively inhibited E. coli-induced phagocytosis and oxidative burst. Although the C5aRa, anti-C5, and anti-C5a mAbs effectively reduced phagocytosis of E. coli measured as MFI, they were less effective than the combination of anti-C2 and anti-factor D mAb when the data were expressed as percent-positive cells, underscoring the critical role for bacterial C3 opsonization in this context. Previous studies using isolated granulocytes support that serum is required for efficient phagocytosis of E. coli and E. coli-induced oxidative burst [29 , 30 ]. The serum factors required for efficient complement opsonization and phagocytosis of E. coli include complement C1q, C3, and the neutrophil bactericidal/permeability-increasing protein [30 ]. Our study extends previous data showing that complement C3 and C5a are critical in the oxidative burst in neutrophils [7 ]. However, the bactericidal function of neutrophils is mediated by enzymes in the phagosomes and not by the oxidative burst itself [31 ], as previously suggested. Our data indicating the important role of complement C3 opsonization in the phagocytosis of E. coli in human whole blood do not rule out the greater importance of LPS in E. coli binding by adherent macrophages through the receptor family CR3/LFA-1/p150,95 in the absence of complement [32 ]. A previous study also indicates that polymorphonuclear neutrophils (PMNs) in solution have a different distribution of CR3 and CD14 compared with those adsorbed to a surface, as only PMNs in solution showed CR3 and CD14 close to each other [33 ].
Complement inhibition with the combination of anti-C2 and anti-factor D mAb blocks the classical, alternative, and MBL pathways. Anti-C2 inhibits the classical and MBL/lectin pathways, and both pathways can be activated by E. coli. In the serum, from our blood donors, anti-E. coli antibodies of IgM and IgG classes were detected. However, no correlation between the antibody levels—C1q deposition on the E. coli surface and oxidative burst—was found, supporting that the classical/MBL pathways have less importance in E. coli-induced complement activation compared with the alternative pathway in our model [11 ]. The significant effect of anti-C2 and anti-factor D could in part be a result of inhibited complement-mediated lysis of E. coli with correspondingly less liberation of several soluble, inflammatory mediators such as LPS. Microvesicles of Neisserria meningitidis induce complement activation [34 ], indicating that activation of the complement system is not limited to whole bacteria. C3b binds covalently to the bacterial surface and is converted rapidly to iC3b [35 ]. As the effect of anti-C2 combined with anti-factor D blocks C3 activation and thereby bacterial opsonization, as we documented in our experiments, the consequent binding to C3 receptors—C3b to CR1 and iC3b to CR3—is abolished. Furthermore, the downstream activation of complement with release of the anaphylatoxins C3a and C5a is blocked simultaneously by these mAbs. A previous report about the phagocytosis and oxidative burst in isolated human granulocytes stimulated with zymosan particles also supports the critical role of C3 opsonization and CD11b [36 ]. It is interesting that E. coli strains with K capsular polysaccharides, which activated complement at a slow rate, were opsonized ineffectively by C3 and were resistant to phagocytosis of human granulocytes, further supporting the important role of C3 opsonization [9 ].
Although the combination of anti-C2 and anti-factor D mAbs blocks complement C3 opsonization completely on the bacterial surface, the bacteria are still opsonized by C1q, MBL, C4b (C4 is upstream to C2), and Igs. It is surprising that anti-C2 and anti-factor D, when used separately, had no effect on E. coli C3 opsonization (Table 1) , and the same inhibitors reduced C5a formation slightly in the fluid phase (Fig. 1) . This may indicate that the classical/lectin (C2) and the alternative (factor D) pathways are sufficient to give maximally detectable C3 deposition on the bacterial surface. However, the relative role of the classical pathway in E. coli-induced complement activation cannot be elucidated from these experiments, as anti-C2 blocks the classical and MBL/lectin pathways.
The initial E. coli-induced CD11b up-regulation in granulocytes after 10 min was mainly C5a-dependent, supporting previous studies [11 , 37 ]. However, CD14 also contributed to the CD11b up-regulation in granulocytes, as only the combination of anti-C5 and anti-CD14 mAbs blocked the initial CD11b up-regulation completely. The role of LPS/CD14 in neutrophil CD11b up-regulation is also supported by previous studies [38 ]. Furthermore, Gordon et al. [37 ] showed that E. coli-induced CD11b up-regulation in isolated granulocytes incubated in buffer in the absence of serum was inhibited completely by polymyxin B, suggesting that LPS mediates E. coli-induced CD11b up-regulation solely in the absence of serum. However, one could not expect a role of complement in serum-free models using isolated cells, emphasizing the advantage of using whole blood models when studying the relative role of candidates in the inflammatory cross-talk. In contrast to granulocytes, monocytes are highly dependent on CD14 for up-regulation of CD11b at low LPS concentrations [39 ]. It is interesting that we found that C5a also contributed significantly to E. coli-induced monocyte up-regulation of CD11b, as only the combination of anti-CD14 and anti-C5a mAb blocked this effect completely. Collectively these data indicate that C5a/C5aR interaction and activation of CD14 are required and sufficient for CD11b up-regulation on granulocytes and monocytes, although the relative importance of these two differs significantly between the two cell types.
C5a was important, not only for E. coli-induced CD11b up-regulation but also for phagocytosis and oxidative burst, as supported by the effect of anti-C5a mAb 137-26, which inhibits the effect of C5a by binding to the C5a moiety of C5 without preventing C5 cleavage [24 ]. It is interesting that the inhibitory effect of anti-C5a mAb on phagocytosis of E. coli was more pronounced when it was given as MFI and less pronounced when phagocytosis was expressed as percent-positive cells, suggesting that C5a is involved in the quantitatively most efficient part of phagocytosis, most likely a result of the up-regulation of CD11b, whereas blocking of C3 opsonization was required for complete inhibition. Furthermore, we documented a direct effect of CVF and rhC5a in the expression of CD11b, which was blocked by the C5aRa. Consistently, Sprong et al. [40 ] have shown that neutralization of C5a inhibited CD11b up-regulation and phagocytosis of N. meningitides. Marder et al. [41 ] showed that leukotriene B4 (LTB4) up-regulates granulocyte CD11b, and this up-regulation can be inhibited by a LTB4 receptor antagonist. It is therefore possible that LTB4 acts as a second messenger between C5a and CD11b.
The role of soluble E. coli-derived molecules such as LPS in inflammation has been studied in detail previously [14
]. The present study shows that LPS, at concentrations present in the whole E. coli bacteria preparations, do not activate complement. Similar results were found with N. meningitidis LPS [34
, 42
]. However, LPS at high concentrations induced TCC formation, supporting previous data that LPS might activate complement [43
]. We speculate that the LPS-induced complement activation at high LPS concentrations may be a result of the formation of LPS in micellar forms or binding of LPS to plastic surfaces, as the latter phenomena have been reported to activate the alternative complement pathway [44
]. However, as LPS only induces complement activation at concentrations above the levels found in vivo in E. coli sepsis [44
], it is unlikely that LPS is an important complement activator in vivo. However, LPS might sensitize blood mononuclear cells to the activation by C5a, as previously shown for IL-1ß and TNF-
synthesis [45
]. Furthermore, blocking of CD14 and CD18, the latter being the common ß-chain of CR3/CD11b and CR4/CD11c, inhibits group B streptococci-induced TNF synthesis in monocytes, supporting a possible link between CD14 and ß-integrins [46
].
In conclusion, the present data indicate that activation of CD14 and complement, including C3 opsonization and C5a release, are crucial events, which are both required and sufficient for E. coli-induced activation of granulocytes and monocytes, as judged by CD11b up-regulation, phagocytosis, and subsequent oxidative burst in an in vitro human whole blood model of sepsis. Provided that the bacteria are treated properly with antibiotics, we propose that a combined inhibition of complement and CD14 might be a therapeutic approach to reduce the uncontrolled, inflammatory response in sepsis.
Received August 29, 2006; revised December 23, 2006; accepted January 7, 2007.
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
1-acid glycoprotein, and fibrinogen Semin. Arthritis Rheum. 20,129-147[CrossRef][Medline] by human peripheral blood mononuclear cells exposed to recombinant human C5a Eur. Cytokine Netw. 2,27-30[Medline] B activation and TNF production induced by lipopolysaccharide and group B streptococcal cell walls J. Immunol. 160,4535-4542This article has been cited by other articles:
 |
|
 |
|
  E. B. Thorgersen, A. Pharo, K. Haverson, A. K. Axelsen, P. Gaustad, G. J. Kotwal, G. Sfyroera, and T. E. Mollnes Inhibition of Complement and CD14 Attenuates the Escherichia coli-Induced Inflammatory Response in Porcine Whole Blood Infect. Immun., February 1, 2009; 77(2): 725 - 732. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. A. Flierl, D. Rittirsch, B. A. Nadeau, D. E. Day, F. S. Zetoune, J. V. Sarma, M. S. Huber-Lang, and P. A. Ward Functions of the complement components C3 and C5 during sepsis FASEB J, October 1, 2008; 22(10): 3483 - 3490. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
