The soluble form of the endotoxin receptor CD14 is required for the LPS-induced activation of cells lacking membrane-bound CD14. It has been shown that a deletion mutant of human CD14 consisting of the N-terminal 152 amino acids has the capacity to mediate the stimulation of different cell types by LPS. To identify the structural domains of the molecule related to this functional property, we screened a set of alanine substitution mutants using CD14-negative U373 astrocytoma cells. We show that 3 of 18 soluble mutants of human CD14 failed to mediate the LPS-induced IL-6 production in U373 cells. These mutants were located in two regions of the molecule (aa 9–13 and 91–101) that are not essential for LPS binding. In addition, the mutants had a reduced capacity to mediate LPS-stimulated IL-6 production in human vascular endothelial and SMC. In contrast, the potential of sCD14(91–94,96)A, and sCD14(97–101)A to signal LPS-induced activation of human PBMC was not significantly reduced. These results show that the regions 9–13 and 91–101 are involved in the sCD14-dependent stimulation of cells by LPS but that the mechanisms by which different cell types are activated may not be identical.
After a pathogen has gained entry into a host, the host’s immune system responds by a complex network of reactions that has been termed “inflammatory response.” Initiation of an inflammatory response relies on recognition of pathogen-derived compounds by a relatively small number of receptor molecules that are either expressed on immune cells or present as soluble proteins in body fluids.
One of these molecules is the pattern recognition receptor CD14 (1). It is expressed as a glycosylphosphatidylinositol-anchored surface protein (mCD14)3 on myeloid lineage cells (2) and occurs in soluble form (sCD14) in serum (3). In vitro (4, 5, 6) and in vivo (7), experiments indicate that membrane-bound CD14 is the key receptor for Gram-negative bacterial LPS on monocytes/macrophages and is required for an inflammatory response triggered by low amounts of endotoxin.
Nevertheless, it is common knowledge that many different cells respond directly to LPS without possessing a high affinity LPS receptor. Various reports (8, 9, 10, 11) indicated that LPS-induced activation of HUVEC and the epithelial cell line U373 is mediated through an sCD14-dependent pathway. Further studies demonstrated that this pathway is also used by smooth muscle cells (SMC) (12) and CD14-negative monocytes derived from patients suffering from paroxysmal nocturnal hemoglobinuria (13, 14, 15). None of these cellular responses is strictly dependent on LPS-binding protein (LBP), an acute phase reactant originally discovered as a catalyst of CD14-LPS interaction (4). Thus, it appears that in general lack of mCD14 may be compensated for by the soluble receptor and that sCD14 mediates transfer of LPS to a putative signal transducer on cellular surfaces. However, the importance of this pathway for the induction of local and systemic inflammatory reactions has not yet been clarified. Notably, Hailman et al. (16) found that, in the absence of LBP, even CD14-bearing macrophages and polymorphonuclear neutrophils may become activated by LPS-sCD14, and Blondin et al. (17) provided evidence for a direct, mCD14-independent, interaction of LPS-sCD14 complexes with a molecule on normal human monocytes.
Various efforts have been made to identify the signal transducing molecule. Schletter et al. (18) isolated an 80-kDa protein (LMP80) from membrane preparations that interacts with LPS only in the presence of sCD14 and LBP. This protein has recently been identified as CD55 (19). Zarewych et al. (20) showed that LPS in the presence of serum or LBP triggers the association of CD14 with complement receptor 3, and, in line with this, Ingalls et al. (21) provided evidence that complement receptor 3 may present LPS to a downstream signal transducer. More recently Vita et al. (22) described a 216-kDa receptor for sCD14 that is present on several cell types, including monocytes, endothelial cells, and epithelial cells. Yang et al. (23) demonstrated that the human Toll-like receptor (TLR) 2 is able to signal LPS-induced responses that depend on LBP and can be enhanced by CD14. These results were corroborated by Kirschning et al. (24), who found that expression of CD14 increases LPS signal transmission through TLR2 and that sCD14 can replace serum to support activation of TLR2 by LPS. Finally, Poltorak et al. (25) and Qureshi et al. (26) independently discovered that mutations of another TLR, TLR4, selectively impede LPS signal transduction in endotoxin-resistant C3H/HeJ and C57BL/10ScCr mice. As an alternative pathway, Joseph et al. (27) proposed that, with the help of CD14, LPS mimics the second messenger function of ceramide and stimulates cells by interaction with a ceramide-activated kinase located within the cell membrane.
Both mCD14 and sCD14 are attractive potential targets of therapies directed against LPS-induced septic shock. Therefore, several approaches have been used to identify the structural basis for the functional properties of human CD14. Juan et al. (28) showed that an N-terminal fragment of human CD14 consisting of 152 amino acids is a functional soluble LPS receptor. The same group reported that a CD14 mutant with a deletion of aa 57 to 64 does not interact with LPS (29, 30) and, in a further study, that CD14(7–10)A lacks the capacity to activate cells (31). To extend and complete this work, we have generated a set of alanine substitution mutants covering aa 1–152 of human CD14. In each mutant, a block of five consecutive amino acids, except cysteines, prolines, and consensus motifs for N-linked glycosylation, was exchanged to alanine. Of 23 constructs generated so far, 21 mutants were expressed as membrane-bound proteins on the surface of stably transfected Chinese hamster ovary (CHO) cells. We found that only CD14(39–41,43,44)A was unable to recognize LPS and Escherichia coli (32).
In this study we have used the soluble forms of these alanine substitution mutants to investigate their potential to mediate LPS-induced activation of different cells including U373 cells, HUVEC, SMC, and human PBMC. We show that 3 of 18 soluble mutants located in aa regions 9–13 and 91–101 fail to mediate activation of U373 astrocytoma cells by LPS. In addition sCD14(9–13)A, sCD14(91–94,96)A, and sCD14(97–101)A have a reduced capacity to signal an LPS response in vascular endothelial and SMC. In contrast, using PBMC, the potential of two mutants, sCD14(91–94,96)A, and sCD14(97–101)A, to mediate cellular activation is comparable to the wild-type molecule. These data indicate that the mechanisms by which different cell types are activated by LPS via sCD14 may not be identical.
Materials and Methods
CHO cells transfected with CD14 wild-type and CD14 alanine substitution mutants were prepared and cultured as previously described (32). All cell cultures were performed at 37°C in a humidified atmosphere containing 5% CO2. For preparation of sCD14, supernatants were collected and stored at −20°C until use.
Binding of [3H]LPS to mCD14 mutants and determination of Kd
Purification of sCD14 mutants from culture supernatants
Human recombinant sCD14 was purified by affinity chromatography from pooled serum-free supernatants of transfected CHO cells. The anti-CD14 mAb biG-2 (Biometec, Greifswald, Germany) was coupled to a HiTrap N-hydroxy-succinamide- activated Sepharose column (Pharmacia Biotech, Uppsala, Sweden). After passage of the supernatant, the column was washed with 10 ml PBS, and sCD14 was eluted with 0.1 M glycine buffer (pH 3.0). The pH was immediately adjusted to 8.0 by adding 0.5 M Tris-HCl (pH 8.0). The material was then concentrated by ultrafiltration using a Centricon 30 concentrator (Amicon, Beverly, MA) and equilibrated with PBS. Purity of sCD14 preparations was >95% as determined by SDS-PAGE. Protein concentrations were determined by Protein Assay ESL (Boehringer Mannheim, Mannheim, Germany) using BSA as standard and calculated assuming a carbohydrate content of 20%.
Binding of LPS to sCD14
Native PAGE and Western blotting.
Purified sCD14 or sCD14 mutants (0.5 μg) were mixed with 1 μg LPS (E. coli LCD 25; List Biological Laboratories) and incubated overnight at 37°C in a final volume of 10 μl PBS. Reaction mixes were loaded onto 4–15% Tris-glycine Gels (pH 8.8) (Bio-Rad Laboratories, Munich, Germany) and separated by native PAGE. sCD14 was detected by Western blotting using a rabbit anti-human CD14 antiserum (34).
Binding of FITC-labeled LPS to sCD14 coupled to ELISA-plates.
Purified wild-type or mutant sCD14 (5 μg/ml in 50 μl/well 0.1 M carbonate buffer (pH 9.6)) was coupled to 96-well plates (Maxisorp, Nunc, Wiesbaden-Biebrich, Germany) by overnight incubation at 4°C. Plates were washed with PBS/0.1% Tween 20 and blocked by incubation with PBS/1% Tween 20 for 1 h. Then FITC-labeled LPS (E. coli, O55:B5, 3 μg FITC/mg LPS; Sigma, Munich, Germany) in PBS/1% Tween 20 was added and incubated overnight. Control reactions were performed in the presence of a 50-fold excess of unlabeled LPS (E. coli, O55:B5; Sigma) or in the presence of 2 μg/ml Fab fragment of the anti-CD14 Ab biG 14. After washing four times, bound LPS was detected by a 1-h incubation with a mouse monoclonal anti-FITC Ab conjugated with alkaline phosphatase (clone FL-D6, Sigma; 1:2000 dilution). Finally, after washing again 4-nitrophenyl phosphate was added as substrate, and OD405 was measured.
Stimulation of U373 cells
U373 cells (ATCC HTB-17) were obtained from the American Type Culture Collection (Manassas, VA) and cultured in MEM containing Earle’s salts and l-glutamine supplemented with 0.1 mM nonessential amino acids, 1 mM sodium pyruvate, 200 U/ml penicillin, 200 μg/ml streptomycin, and 10% FBS (medium and supplements from Life Technologies, Eggenstein, Germany). For experiments, cells were trypsinized, washed in medium, and counted. Approximately 37,500 cells/well were seeded into 48-well plates (Corning-Costar, Corning, NY) and covered with 0.5 ml complete medium. Twenty-four hours later, the medium was removed, and the cells were washed three times with HBSS (Life Technologies). Cells were stimulated in the presence of serum-free medium in a final volume of 0.2 ml. LPS from Salmonella minnesota Re 595 and E. coli O55:B5 were obtained from Sigma, and LPS from Salmonella friedenau was a kind gift of Helmut Brade (Research Center, Borstel, Germany). LPS and sCD14 were added as indicated. After 24 h, supernatants were collected and immediately frozen at −70°C.
Stimulation of HUVEC
Primary HUVEC were purchased from Clonetics (San Diego, CA) and cultured in EGM-2 medium (Clonetics). A 70–90% confluent culture flask (83-cm2 growth area; Nunc) containing cells in the 2nd or 3rd passage was thoroughly trypsinized. Fifteen thousand cells per well were seeded into 48-well plates and covered with 0.5 ml complete medium. After 48 h, medium was aspirated, and the cells were washed once with HBSS (Life Technologies). Cells were stimulated with LPS and sCD14 as indicated in the presence of serum-free EGM-2 medium in a final volume of 0.2 ml. After 24 h, supernatants were collected and stored at −70°C.
Stimulation of SMC
SMC were isolated from unused portions of human saphenous veins obtained following bypass surgery and characterized as described earlier (35, 36). The cells were cultured in DMEM (Biochrom, Berlin, Germany) containing 10% FBS (Linares, Bettingen, Germany), 1 g/L glucose, 2 mM l-glutamine, 100 U/ml penicillin, and 100 μg/ml streptomycin. For experiments, 10,000 cells/cm2 were seeded into flat-bottom 96-well plates (Nunc) and grown for 24 h. Stimulation was performed in the absence of serum and in the presence or absence of LPS and sCD14 for another 24 h. After addition of 5% FBS, medium was aspirated and frozen at −70°C.
Stimulation of PBMC
Measurement of LPS-induced oxidative burst response.
PBMC were prepared by density gradient centrifugation on ficoll (1.077 g/cm3; Biochrom) of heparinized venous blood obtained from healthy volunteers. Cell number was adjusted to 2.5 × 106 cells/ml in PBS and 200 μl luminol (final concentration 1.66 × 10−4 M; Sigma) was added per 1 ml cell suspension. The luminol/cell mixture was incubated for 2 h at 37°C.
LPS (0.1 μg) and 0.5 μg sCD14 were mixed in a final volume of 10 μl and incubated overnight at 37°C. This reaction mix was diluted 1:100 with PBS, and 20 μl was added to 200 μl of the luminol/cell mixture. Reaction vials were immediately placed in an AutoLumat LB 953 luminometer (Berthold, Munich, Germany), and chemiluminescence was measured for a time period of 60 min.
LPS-induced production of IL-6.
PBMC (4 × 105) in 150 μl serum-free RPMI supplemented with gentamicin (50 μg/ml) and l-glutamine (0.3 g/L) were seeded into 48-well plates (Corning-Costar) and incubated for 2 h at 37°C. LPS (0.01–1 ng) was preincubated overnight with 0.5 μg sCD14 in a final volume of 10 μl PBS. Reaction mixes were diluted with RPMI 1640 to yield final concentrations of 0.1–0.001 ng/ml LPS and 50 ng/ml sCD14. Fifty microliters of diluted reaction mix was added per well, and cells were stimulated for 24 h. Supernatants were collected, centrifuged, and stored at −70°C until use.
IL-6 levels in culture supernatants of U373 cells and HUVEC were measured by standard ELISA technique using a pair of monoclonal rat anti-human IL-6 Abs and recombinant human IL-6 (PharMingen, Hamburg, Germany) as standard protein. All measurements were done in duplicate.
Detection of IL-6 activity by 7TD1 bioassay.
IL-6 activity in supernatants of stimulated SMC was determined by 7TD1 bioassay. Cells of the murine B cell line 7TD1 (37) were cultured in DMEM (Biochrom) containing 4.5 g/L glucose, 10% FBS (Linares), 2 mM l-glutamine, antibiotics, 2 × 10−5 M 2-ME, and 100 pg/ml human recombinant IL-6 (Boehringer Mannheim). Test samples were prepared by serial 3-fold dilution with medium lacking IL-6. Washed 7TD1 cells were adjusted to 40,000 cells/ml, and 50 μl of this suspension was mixed with 50 μl of the diluted sample. Cells were incubated for 72 h. Then 10 μl/well MTT (2-(4,5-dimethyl-thiazol-2-yl)-2,5-diphenyl tetrazolium bromide; 0.5 mg/ml in PBS) was added for 4 h. Finally, stop solution (5% SDS, 50% dimethylformamide; 100 μl/well) was added for 2 h, and absorption was read at 550 nm. IL-6 activity of samples was calculated by comparison with an IL-6 standard (10 ng/ml IL-6). The data are derived from triplicate cultures and are given as mean ± SD.
LPS binding to mutant and wild-type sCD14 and sCD14-mediated cellular responses in HUVEC, SMC, and U373 cells were compared by ANOVA with two factors (concentration and type of mutation). Because equal variances were not assumed, the post hoc test was performed according to Tamhane. A p < 0.05 was considered significant. For comparison of cellular responses, values at a concentration of 0.008 μg/ml of recombinant protein were excluded.
Amino acids 9–13 and 91–101 of human CD14 are required for LPS-induced IL-6 release by U373 cells but are not essential for binding to LPS
To identify structural domains related to functional properties of human CD14, 23 alanine substitution mutants have been cloned and stably transfected into CHO cells (32). Eighteen of the transfected cell lines released sCD14 into the culture supernatant. In a first screen, we tested the capacity of serum-free supernatants containing these mutants to mediate LPS-induced IL-6 production in the human astrocytoma cell line U373. Three mutants, sCD14(9–13)A, sCD14(91–94,96)A, and sCD14(97–101)A, repeatedly failed to activate U373 cells in this assay (data not shown). Subsequently, these mutants were purified by affinity chromatography and tested over a concentration range of 8–1000 ng/ml. Fig. 1 shows that the capacity of the mutants to signal different types of LPS to U373 cells is strongly reduced in comparison with the wild-type protein. A concentration as low as 40 ng/ml of sCD14 is sufficient to mediate the IL-6 production by U373 cells in response to rough type LPS from S. minnesota Re 595 (Fig. 1, upper panel, filled columns). In contrast, the cells are almost entirely unresponsive if stimulated in the presence of 1 μg/ml of mutant proteins. With smooth-type LPS (E. coli O55:B5 and S. friedenau; Fig. 1, middle and lower panel), sCD14(9–13)A has again almost no stimulating activity. The effects of the mutations in region 91–101 are less pronounced as compared with Re-LPS: sCD14(97–101)A is more active than sCD14(91–94,96)A but still approximately 5-fold less than wild-type sCD14.
To test whether failure of mutant proteins to signal LPS to the cells is due to an impaired capacity to bind LPS, we determined the dissociation constants between [3H]LPS (E. coli, LCD 25) and the membrane-bound forms of the three mutants. The Kd values shown in Table I range from 26–33 nM and are similar for the four proteins tested. Because binding of the ligand to the cell surface molecule might not necessarily correlate with the ability of the soluble protein to react with LPS, we also analyzed interaction of sCD14 mutants with LPS in two different assays. First, it can be demonstrated by native PAGE that incubation with LPS causes a mobility shift of the CD14 bands that indicates the formation of stable complexes between sCD14 mutants and LPS (Fig. 2,A). Second, we observe a similar extent of LPS binding to sCD14 mutants coupled onto ELISA plates (Fig. 2 B). The binding of FITC-LPS (E. coli, O55:B5) to CD14 in this assay is specific because it can be blocked by addition of a 50-fold excess of unlabeled LPS and by a Fab fragment of the anti-CD14 mAb biG14. Taken together, these results suggest that the inability of CD14 mutants to activate U373 cells is not due to inadequate recognition of LPS.
|Mutant .||Kd (nM) .|
|CD14 wild type||31.9|
|Mutant .||Kd (nM) .|
|CD14 wild type||31.9|
CHO-cells expressing the membrane-bound proteins were incubated with different concentrations of LPS. The amount of free and bound ligand at each concentration was measured, and Kd values were determined by curve fitting. All measurements were done in triplicate.
Stimulation of human vascular cells by LPS is altered in the presence of sCD14 mutants carrying alanine substitutions in regions 9–13 and 91–101
To analyze whether the inability of sCD14 mutants to mediate activation of U373 astrocytoma cells is not a phenomenon restricted to a particular tumor cell line, we performed similar experiments using CD14-negative primary vascular cells: HUVEC and SMC.
Fig. 3 shows that HUVEC’s IL-6 response to LPS is strongly dependent on sCD14; there is already stimulation at 8 ng/ml with all types of LPS tested. All three mutants are considerably less potent. The capacity of sCD14(91–94,96)A and sCD14(97–101)A to facilitate activation of endothelial cells by LPS is approximately 25-fold reduced as compared with the wild-type molecule; a concentration of 1 μg/ml of mutant protein is required for a response observed with 40 ng/ml of sCD14. As with U373 cells, sCD14(9–13)A is the least activating mutant. There is no obvious difference between the tested types of LPS in the ability of the mutants to mediate stimulation of endothelial cells.
As with endothelial cells, sCD14 facilitates LPS-induced activation of vascular SMC (Fig. 4). In principle, the results resemble those obtained with endothelial cells; in comparison with wild-type sCD14, the mutants have a reduced capability to mediate the LPS-induced secretion of IL-6. Ranking of the mutants reveals the same pattern: sCD14(97–101)A is most active, followed closely by sCD14(91–94,96)A, and again sCD14(9–13)A is least potent. However it should be noted that the absolute difference in IL-6 production between mutants and wild-type is often less than a factor 2 and is not as clearly pronounced as with U373 and endothelial cells.
Mutations of aa 9–13 and 91–101 have a different impact on the response of human PBMC to LPS-sCD14 complexes
Activation of myeloid cells by LPS occurs through engagement of membrane-bound CD14 in a reaction catalyzed by LBP (4, 5, 6). However, LPS-induced reactions in whole blood (13) or PBMC (14, 15) derived from PNH patients require the presence of sCD14, and, furthermore, it appears that this sCD14-dependent pathway is also functional in CD14+ granulocytes (16) and monocytes (17). Therefore we wanted to know whether mutations in regions 9–13 and 91–101 that impair the capacity of sCD14 to mediate activation of CD14− vascular and epithelial cells have a similar impact on LPS-induced stimulation of PBMC.
The oxygen radical production of PBMC is an immediate response to LPS that requires serum or LBP (15). However, in the absence of LBP, cells may become activated by preformed LPS-sCD14 complexes. LPS and sCD14 mutants were preincubated overnight and used to stimulate human PBMC. The rapid release of reactive oxygen species was measured as chemiluminescence for a time period of 60 min. Maximum stimulation is detected after 15 min and requires both LPS and sCD14. In Fig. 5, the stimulation of PBMC by rough LPS (S. minnesota Re 595; Fig. 5,A) and smooth LPS (E. coli O55:B5; Fig. 5,B) is shown. In the presence of LPS preincubated with sCD14, the production of oxygen radicals is increased in comparison with sCD14 or LPS alone. Wild-type sCD14, sCD14(91–94,96)A, and sCD14(97–101)A are equally able to mediate a cellular response to LPS. It should be noted that, with S. minnesota LPS, the potential of sCD14(91–94,96)A is even greater than that of the wild-type protein. This effect of sCD14(91–94,96)A is restricted to this particular type of LPS and has been consistently observed in five independent experiments. As with other cell types, sCD14(9–13)A is also less potent in activation of PBMC. With smooth-type LPS from (E. coli O55:B5 (Fig. 5,B) and S. friedenau (data not shown), the signal obtained with sCD14(9–13)A + LPS is slightly, but consistently, reduced as compared with the other mutants. With S. minnesota Re 595 LPS (Fig. 5 A), the curve in the presence of sCD14(9–13)A is very similar to that seen with LPS alone. Thus, in contrast to U373 and vascular cells, only mutations in region 9–13 but not in region 91–101 may influence the sCD14-dependent oxidative burst response of mononuclear cells.
However, generation of oxygen radicals by PBMC is an immediate response in contrast to IL-6 production that is measured over a time period of 24 h. Therefore, the difference might be due to the type of response rather than to the type of cells. To exclude this possibility, PBMC were stimulated under serum-free conditions by LPS-sCD14 complexes for 24 h to produce IL-6. In Fig. 6, two experiments with mononuclear cells derived from different donors are shown. In comparison with U373 and vascular cells, PBMC require only low amounts of LPS for induction of IL-6; cells were stimulated with 0.1 and 0.01 ng/ml LPS from S. minnesota Re 595 or 0.01 and 0.001 ng/ml LPS from E. coli O55:B5. sCD14 was present at a final concentration of 50 ng/ml. With both types of LPS, generation of IL-6 depends on the presence of sCD14. With one exception in experiment 1, sCD14(9–13)A has the lowest activating capacity. Obviously, mutations in region 91–101 do not cause a functional loss. There is evidence that, in the presence of 0.01 ng/ml Re-LPS, sCD14(91–94,96)A has a stronger activating capacity than has the wild-type protein. Altogether, our results indicate that sCD14-dependent responses of PBMC are impaired by mutation of aa 9–13 but not by substitutions in region 91–101.
LPS is a major component of the outer cell wall of all Gram-negative bacteria and an important molecule recognized by receptors of the innate immune system. It is known as a strong inducer of proinflammatory responses, including the release of cytokines, lipid mediators, and reactive oxygen species and NO. This response, however, can have both beneficial and detrimental effects on the host’s integrity; it is required to combat an infection but may also lead to a severe systemic dysregulation that in clinical terms is called “septic shock.” Therefore, the understanding of molecular mechanisms underlying the interaction of LPS with host tissues is crucial for the development of new treatment strategies directed against Gram-negative bacterial sepsis and septic shock.
In this study we focused on the structural basis for the “activating” function of the soluble LPS receptor CD14, i.e., on its capacity to mediate LPS-induced cellular activation independently of the membrane-bound molecule. In an attempt to perform a structure function analysis of CD14, we have previously generated a series of alanine substitution mutants comprising the entire 152 N-terminal amino acids of the mature human protein (32). This region has been shown by Juan et al. (28) to contain both endotoxin-binding and cellular activation domains. From an initial screen, three soluble mutants, sCD14(9–13)A, sCD14(91–94,96)A, and sCD14(97–101)A were selected that did not facilitate activation of the CD14-negative astrocytoma cell line U373. We demonstrate here that this lack of function is of relevance not only for this particular tumor cell line but also for human vascular cells, i.e., endothelial and SMC.
In all experiments done with these different types of CD14-negative cells, we consistently found that sCD14(9–13)A is the least activating mutant. The mutation affects a region that contains a cluster of charged residues (DDED) and therefore probably represents an exposed site of the CD14 molecule. The importance of this region for cellular activation has already been shown by Juan et al. (31), who introduced alanine substitutions at position 7–10. Their mutant, sCD14(7–10)A, had a reduced capacity to mediate stimulation of U373 cells and neutrophils by LPS (31), an observation that is confirmed and extended by the present study. In addition to this, our results indicate that a second region that has not yet been described as part of the “activating” domain of human CD14 is located between aa 91–101. Mutations within this region cause a less severe phenotype in comparison with sCD14(9–13)A, but, still, the “activating” capacity of sCD14(91–94,96)A and sCD14(97–101)A is considerably decreased as compared with the wild-type protein (Figs. 1, 3, and 4).
The inability of mutant proteins to activate cells is not due to an impaired capacity to react with LPS. We have previously shown that the membrane-bound forms of CD14(9–13)A, CD14(91–94,96)A, and CD14(97–101)A are qualitatively able to interact with FITC-labeled LPS and with the Gram-negative bacterium E. coli (32). Furthermore, as shown in Table I, the Kd-values do not show considerable variation and are similar to the value (27 nM) published by Kirkland et al. (38). Because binding of LPS to membrane-bound CD14 might involve secondary interactions with coreceptors or an increase in local receptor concentration due to a clustering of molecules, we separately tested interaction of LPS with the soluble mutants. We demonstrate by gel shift (Fig. 2,A) that mutant proteins form stable complexes with LPS although it should be noted that the technical constraints of this assay oblige us to use concentrations of the reactants much higher than those appropriate in a biological assay. In addition, binding curves of FITC-labeled LPS to sCD14 mutants coupled to a plastic surface (Fig. 2 B) do not differ significantly in a concentration range that comes close to the LPS amounts used in U373, HUVEC, and SMC experiments. Taken together, these results indicate that neither mutations in region 9–13 nor mutations in region 91–101 have any major impact on ligand binding. The reduced capacity of these mutants to activate cells indicates that LPS binding and cellular activation are separate functional properties of CD14.
It should be noted that Juan et al. (30) generated a mutant of human CD14 by deletion of aa 57–64 that did not bind LPS and, consequently, failed to mediate LPS-induced activation of U373 cells. In our initial study (32), we could not confirm their conclusion that region 57–64 is the LPS-binding domain of human CD14. The substitution mutant from our panel, CD14(57,59,61–63)A, was found to be perfectly able to react with LPS (32) and, as soluble form, also to facilitate LPS-induced IL-6 production in astrocytoma cells (data not shown). As discussed elsewhere (32), we believe that this discrepancy to the study of Juan et al. is due to the mutagenesis strategy; alanine substitutions may have a less severe impact on the tertiary structure than a deletion of 8 amino acids.
The functional impact of the mutations in regions 9–13 and 91–101 appears to be different in PBMC in comparison with endothelial, SMC, and U373 cells. With PBMC, only sCD14(9–13)A has a lower activating capacity whereas wild-type sCD14, sCD14(91–94,96)A, and sCD14(97–101)A are all able to facilitate both the immediate (oxidative burst) and delayed response (IL-6 production) to LPS. Remarkably, with Re-LPS, sCD14(91–94,96)A has an even higher stimulatory potential in comparison with the wild-type protein. This effect is seen only with the Re-LPS but not in experiments performed in parallel using smooth LPS from E. coli O55:B5 (Figs. 5 B and 6) and S. friedenau (data not shown). The structural basis for this difference is under investigation.
Basically, PBMC are different from U373, endothelial, and SMC because the monocytes in the preparation are CD14+ and there is no requirement for sCD14 if LBP is present (15). In the absence of serum or LBP, preformed sCD14-LPS complexes might simply act as an LPS shuttle (16) to replace the LBP function. A direct interaction of sCD14 or an sCD14 mutant with a signal transducer should not be necessary; LPS can first be transferred to membrane-bound wild-type CD14, which then associates with the signal-transducing molecule. This scenario, however, is not entirely consistent with the observation that sCD14(9–13)A has a lower stimulating capacity in comparison with the wild-type protein without having a different affinity to LPS. Also, the higher stimulatory potential of sCD14(91–94,96)A with Re-LPS does not fit into this scheme. Blondin et al. (17) found that, after complete blockage of mCD14, LPS can still be delivered to cells by an sCD14-dependent mechanism. Taken together, there is strong evidence that sCD14-dependent activation of monocytes is at least in part independent of mCD14.
Because different domains are required for sCD14-dependent activation of U373 and vascular cells in comparison with PBMC, our results suggest that the signal-transducing pathways are not identical. This suggestion is in agreement with Haziot et al. (39), who showed that certain anti-CD14 mAbs that block activation of endothelial cells do not affect the monocyte response to LPS. Interestingly, Cunningham et al. (40) demonstrated that LPS derived from Helicobacter pylori and Porphyromonas gingivalis are able to activate monocytes but fail to activate endothelial cells. Two further studies done by the same group revealed (41, 42) that, although P. gingivalis binds to LPS in a different manner than does E. coli-LPS, lack of endothelial cell activation is not due to this. Rather, there is evidence that P. gingivalis LPS and sCD14 form complexes that do not interact with endothelial cells. Thus, it appears that the initial links of the chain that leads to cellular activation by LPS-sCD14 are different in different cell types. These putative coreceptors might recognize different parts of CD14 in conjunction with certain exposed structures of the LPS molecule.
The question whether in vivo the “activating capacity” of sCD14 contributes to LPS-induced pathology is still open. sCD14-mediated activation of endothelial cells by LPS might increase the overall load of proinflammatory and procoagulatory mediators. In addition, SMC, which may become accessible to LPS after damage to the endothelial barrier, have also been shown to produce proinflammatory cytokines such as IL-1 and IL-6 (43, 44). It is conceivable that endothelial and SMC may enhance pathophysiologic processes leading to septic shock and multiorgan failure. On the other hand, recognition of LPS by cells lacking mCD14 may contribute to LPS clearance. Application of sCD14 as a treatment of LPS-induced shock has been shown to decrease lethality in mice (45), but, in our hands (46), without reducing serum TNF-α levels or preventing the clinical expression of shock symptoms. Thus, sCD14 treatment may have both pro- and antiinflammatory effects. Future experiments will show whether application of a defective “nonactivating” mutant is superior to the wild-type protein as a treatment of LPS-induced shock.
We thank Mrs. R. Görlich and Mrs. B. Fürll for excellent laboratory assistance. We thank C. Schwahn for the statistical analysis.
This study was supported by grants from Deutsche Forschungsgemeinschaft (DFG IIB6-Ste 587/2–2, to F.S., and SFB 367, project C5, to A.J.U.), from the Bundesministerium für Bildung, Wissenschaft, Forschung und Technologie (BMBF) (Forschungsverbund Halle, project 06, to H.L. and K.W.), and from Fonds der Chemischen Industrie (to C.S.).
Abbreviations used in this paper: mCD14, membrane-bound CD14; LBP, LPS-binding protein; sCD14, soluble CD14; sCD14(x-y)A, sCD14 containing alanine substitutions at amino acid positions x to y; SMC, human vascular smooth muscle cells; TLR, Toll-like receptor; CHO, Chinese hamster ovary.