CD14 Employs Hydrophilic Regions to “Capture” Lipopolysaccharides

Reagents, cells, and buffers

Reagent grade chemicals were obtained from Sigma (St. Louis, MO). HUVEC growth medium consisted of medium 199 (Life Technologies, Gaithersburg, MD0 containing 4 mM l-glutamine, 90 μg/ml heparin, 1 mM sodium pyruvate, 30 μg/ml endothelial cell growth stimulant (Biomedical Products, Bedford, MA), and 20% FBS (HyClone, Logan, UT). Medium 199 stimulation medium consisted of medium 199 containing 4 mM l-glutamine, 90 μg/ml heparin, 1 mM sodium pyruvate, 1 mg/ml human serum albumin, and 5% pooled normal human serum (Gemini Bioproducts, Calabasas, CA). RPMI 1640 medium was obtained from Life Technologies (Grand Island, NY). IL-8 levels in supernatants were measured using Cytoscreen IL-8 Immunoassay Kit (BioSource, Camarillo, CA). Pooled human sera were obtained from Gemini Bioproducts and were depleted of sCD14 as previously described (20). Protein A-Sepharose was obtained from Repligen (Cambridge, MA). HUVEC were obtained from Clonetics (San Diego, CA) and maintained in HUVEC growth medium. Cells were used at the fourth passage for all experiments described in this report.

Bacterial strains and purification of LPS

Helicobacter pylori ATCC 43504 was obtained from the American Type Culture Collection (Manassas, VA). Porphyromonas gingivalis ATCC 33277 was provided by Aaron Weinberg, Department of Periodontics, University of Washington (Seattle, WA). E. coli BMS A016 (serogroup O6) is a human blood culture isolate provided by Marie Coyle, Harborview Medical Center (Seattle, WA). Strains were examined for purity, properly identified, and stored at −70°C. Cultures were made from frozen bacterial stocks to avoid repetitive subculture. Bacterial cells were grown for LPS isolation as follows. P. gingivalis was grown anaerobically at 37°C for 2–3 days in large batch cultures by Lee Laboratories (Grayson, GA) using enriched Trypticase soy broth (Becton Dickinson, Cockeysville, MD) and was provided as frozen cell pellets. H. pylori ATCC 43504 was grown for 3–4 days on H. pylori agar medium under microaerophilic conditions at 37°C. E. coli was grown overnight in Trypticase soy broth at 37°C with shaking in room air.

Purified LPS from P. gingivalis ATCC 33277 and H. pylori ATCC 43504 were prepared by the cold MgCl₂-ethanol procedure (21). Purified LPS from E. coli A016 was obtained by the phenol-water method (22). All LPS preparations were suspended in distilled water and were found to be free from contaminating nucleic acid and protein by measuring OD at 280 and 260 nm (OD₂₈₀ and OD₂₆₀). Gas chromatographic analysis for sugar and fatty acid composition revealed that the compositions of these purified LPS agreed with published data (23, 24, 25) and allowed calculation of average m.w. The m.w. for E. coli, P. gingivalis, and H. pylori LPS were estimated to be 9170, 7042, and 5405, respectively.

Construction and expression of sCD14:Rg

A cDNA fragment encoding the extracellular domain of wild-type CD14 was subcloned upstream of a cDNA fragment encoding the H, CH2, and CH3 domains of human IgG1 in the mammalian expression vector CDM7B as previously described (20). Similarly produced CD40 recombinant globulin (CD40:Rg) was used as a control.

Oligonucleotide-directed mutagenesis

Mutations were introduced into CD14:Rg by a two-step PCR procedure (18). Overlapping PCR fragments were first generated in separate reactions and then reannealed and extended in a second PCR. In the first step, oligonucleotides (complement and reverse complement) containing the desired codon changes flanked by 18 nucleotides on each side were synthesized for each mutation. Two fragments were then synthesized by PCR using T7 (in pcDNA3) and the reverse complement oligonucleotide to generate the 5′ fragment and CD14R5 (5′-CAAGTCCTGTGGCTTCCA GAG-3′) and the complementary oligonucleotide to generate the 3′ fragment. After gel purification, ∼10 ng of each fragment was mixed and used as template in a second PCR reaction using T7 and CD14R5 oligonucleotides. The resulting 480-nucleotide PCR fragment was gel purified, digested with HindIII and XhoI, and used to replace the 5′ HindIII/XhoI fragment from the wild-type CD14:Rg/pcDNA3 plasmid. All nucleotide changes were confirmed by DNA sequencing. The fusion proteins were produced by transient expression in COS cells and were purified from the spent COS cell supernatants using protein A-Sepharose chromatography. Fusion protein stock solutions were found to be free of endotoxin by the Limulus lysate assay obtained from BioWhittaker (Walkersville, MD).

Generation of mouse mAbs

mAbs to human sCD14 were cloned by standard methods after fusion of a murine myeloma line with spleen cells from mice immunized with purified sCD14:Rg as previously described (20). Throughout the cloning procedure, culture supernatants were assayed for Abs using the ELISA method, with purified sCD14:Rg as the solid phase Ag. Select mAbs of the appropriate specificity were produced in high titer culture supernatants. The Abs were purified from the culture supernatants by protein A chromatography, extensively dialyzed into PBS, filter-sterilized, and stored at 4°C. The purified mouse anti-human sCD14 mAbs used in these experiments were 2G7 and 6C8. A commercial mouse anti-human CD14 mAb (MY4) was obtained from Coulter Immunology (Hialeah, FL).

The mouse anti-LPS mAbs used for detection of LPS binding to sCD14:Rg were 3B4 (IgM; anti-E. coli A016), 5B9 (IgM; anti-P. gingivalis ATCC 33277), and 7A9 (IgG1 anti-H. pylori ATCC 43504). They were cloned as described above using spleen cells from mice immunized with purified LPS.

ELISA for detection of LPS binding to immobilized sCD14:Rg

LPS binding to wild-type and mutations of sCD14 were measured in a capture ELISA format (20). Briefly, Immulon II (Dynatech, Chantilly, VA) 96-well plates were coated overnight at 4°C with 0.05 μg/well SCD14 wild-type or point mutation protein. The next day, the wells were emptied, and nonspecific binding sites were blocked with 100 μl/well of 0.5% BSA in PBS for 30 min at 37°C. The plates were then washed three times with PBS. Fifty microliters of LPS diluted in 1% CD14-depleted serum (source of LBP) was added per well, and the plates were incubated at 37°C for1 h. Unbound LPS was removed with four PBS washes, and bound LPS was detected with specific mouse anti-LPS mAb at 0.5 μg/ml in 0.5% BSA in PBS as previously described (20).

ELISA for detection of anti-CD14 mAb binding to immobilized sCD14:Rg

Immunlon II plates were coated with 0.05 μg/well of sCD14 wild-type or point mutation fusion proteins as previously described (18), and the amount of fusion protein that was captured was monitored with goat anti-human IgG, which reacts with the CD14:Rg fusion protein. mAbs MY4, 2G7, and 6C8 (0.02 μg/ml) were added, and ELISA detection was performed as previously described (20) with goat anti-mouse IgG (Sigma).

IL-8 functional assay

Primary HUVEC cultivation and the IL-8 ELISA were performed as described previously (18, 26), except that 1% sCD14-depleted serum, prepared as previously described (20), instead of 5% complete serum was used in the medium 199 stimulation medium.

LPS blocking experiments

The ability of the CD14 mAbs to block LPs binding to CD14 was examined. CD14 was captured on a 96-well ELISA plate as previously described (27), and 30 min before the addition of E. coli LPS (100 ng/ml), anti-CD14 mAbs or a negative control Ab (anti-CD40) were added. E. coli LPS binding was detected as previously described (20), except that the anti-E. coli LPS mAb was biotinylated (used at 3 μg/ml), because the usual second-step goat anti-mouse would react with the mouse mAbs.

Initial analysis revealed three different groups of CD14 mutation proteins based upon their LPS binding characteristics

Initially a survey was performed with nearly all the new CD14 mutation proteins. Statistical analysis was not performed, nor were additional assays conducted with all the mutation proteins. Rather, this analysis was used to initially identify those new CD14 mutation proteins that may have an effect on LPS binding. Analysis of the data revealed that three different patterns of LPS binding to the CD14 mutation proteins were observed (Table I⇓). In the majority of mutation proteins (group 1) no effect on LPS binding was observed (26 of 36 examined). In this group, consistent with the initial, more limited analysis, none of the serine replacement mutations reduced LPS binding (a total of 19 were examined). The second group (group 2, indicated in bold in Table I⇓) consisted of those mutation proteins that demonstrated a reduced binding with either P. gingivalis or H. pylori, but not E. coli LPS (E4K, D9K, E11K, D12K, R53E, D59K). All these were charge reversal mutations. Interestingly, no mutation proteins examined demonstrated reduced E. coli, yet had normal P. gingivalis or H. pylori LPS binding. It is suspected that this may relate to the higher affinity for CD14 displayed by E. coli LPS (20), although this possibility has not been formally examined. The third group (group 3, indicated by underlining in Table I⇓) consisted of two CD14 mutation proteins that displayed reduced binding to all three LPS ligands examined (D10K and R14E).

Additional binding assays confirmed that charge reversal mutations in select CD14 residues reduced LPS binding

Additional, more comprehensive binding assays were performed with representative CD14 mutation proteins from each group. These assays confirmed the three different patterns of LPS binding to the CD14 mutation proteins mentioned above (see Table I⇑ and Fig. 2⇓). The serine replacement mutation proteins in group 1 did not significantly reduce the binding of any LPS species (D10S, E11S, D12S, R14S, R53S, D57S, and D59S; p > 0.1, not significant by two-sample t test). The charge reversal mutations in group 3 reduced the binding to all the LPS species examined to near background levels (D10K and R14E; p < 0.05 or 0.001, significant or highly significant compared with wild-type CD14 depending upon the species examined by two-sample t test). However, similar to our previous results (18), not all charge reversal mutations reduced LPS binding (for example, D57K).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 2.

The abilities of LPS from three different species of bacteria to bind CD14 mutation proteins. The binding assays were performed as described in Materials and Methods and Table I⇑. Each assay was performed on at least four separate occasions in duplicate. The mean for each LPS compared with wild-type CD14 is presented.

The CD14 mutation protein representatives from group 2 were of particular interest, because they displayed preferential reductions in LPS binding, suggesting that specific residues may be more important for some LPS species. For example, the binding of E. coli LPS was not significantly reduced (p > 0.7, by two-sample t test), whereas the binding of both P. gingivalis (p < 0.05, by two-sample t test) and H. pylori LPS (p < 0.05, by two-sample t test) was significantly reduced when binding to wild-type CD14 and that to mutation protein E11K were compared. A preferential reduction in LPS binding was also observed for mutation protein D59K (E. coli LPS, p > 0.3, not significant; P. gingivalis LPS, p < 0.001, highly significant; H. pylori LPS, p > 0.1, not significant; by two-sample t test; when the binding of LPS to CD14 mutation protein D59K was compared with wild-type CD14). The binding of all three LPS species to CD14 mutation protein D12K was significantly reduced compared with that to wild-type CD14 (p < 0.05, significant by two-sample t test). However, the reduction of P. gingivalis LPS was significantly greater than that observed with E. coli (p < 0.05, significant by two-sample t test, when P. gingivalis and E. coli LPS binding to D12K were compared with wild-type CD14). Although all LPS species demonstrated reduced binding to CD14 mutation protein R53E, the reduction was only significant (p < 0.05, by two-sample t test) for P. gingivalis compared with wild-type CD14.

The differential abilities of E. coli, P. gingivalis, and H. pylori LPS to bind mutant CD14 proteins D12K and D59K were further investigated (Fig. 3⇓). Concentration-dependent increases in the binding of all three LPS species examined were observed with both mutant CD14 proteins. Consistent with earlier analysis (Fig. 2⇑), E. coli LPS bound to D59K to nearly wild-type levels, and that to D12K was slightly reduced (a significant reduction in E. coli LPS binding to CD14 mutation protein D12K was only observed at 30 ng/ml LPS). In contrast, P. gingivalis and H. pylori LPS binding to CD14 mutation protein D12K was significantly reduced at all LPS concentrations examined (p > 0.05, by two-sample t test). Examination of LPS binding to CD14 mutation protein D59K revealed that P. gingivalis (p < 0.05, by two-sample t test), but not E. coli or H. pylori, LPS was reduced compared with wild-type CD14. These analyses are consistent with the observation that charge reversal mutations in CD14 residues D12 and D59 have preferential effects on LPS ligand binding

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 3.

The ability of increasing concentrations of E. coli (A), P. gingivalis (B), and H. pylori (C) LPS to bind wild-type CD14 and CD14 mutation proteins D12K and D59K. In creasing concentrations of LPS were mixed with 1% CD14-depleted serum and added to plates previously coated with 0.5 μg/well wild-type sCD14:Rg (♦), D12K (▴), D59K (X), or CD40:Rg (▪). After 1 h of incubation at 37 C, LPS bound to sCD14:Rg was detected with mouse anti-LPS mAbs (see Materials and Methods). The experiments were performed on three separate occasions, and the averages from duplicate wells from a representative experiment are shown.

The ability of mutant CD14 protein to activate IL-8 secretion is consistent with E. coli LPS binding

The ability of the CD14 single-point mutation proteins to activate IL-8 secretion in HUVECs with E. coli LPS was examined. An assay that employed CD14-depleted serum that was supplemented with the different CD14 mutation proteins was used to examine the ability to elicit IL-8 secretion from human endothelial cells (Fig. 4⇓). Several mutation proteins were examined that did not demonstrate significantly lower E. coli LPS binding (E11S, D57K, D59S) and also did not display a significant difference in IL-8 secretion. Three separate experiments were performed, and these CD14 single-point mutation proteins did not alter the amount of IL-8 secreted by human endothelial cells in response to 30 ng/ml E. coli LPS by >20% that of wild-type protein (data not shown). Fig. 4⇓ presents the data obtained with those mutant CD14 proteins that displayed significant reductions in either E. coli LPS binding or the binding of P. gingivalis or H. pylori (group 1, D10K and R14E; group 2, E11K, D12K, R53E, and D59K from Fig. 2⇑). The effect of increasing LPS concentrations was examined. Each CD14 mutant was able to elicit IL-8 secretion; however, the amount of LPS necessary to facilitate secretion varied significantly among the CD14 mutants. Mutants D10K, D12K, and R14E that displayed the poorest ability to bind E. coli LPS were also the poorest activators of IL-8 secretion. Little or no IL-8 secretion occurred with these CD14 mutations proteins at concentrations of LPS <3 ng/ml. In contrast, E11K and D59K, two CD14 mutation proteins that did not display a significant reduction in E. coli LPS binding (p > 0.7 and p > 0.3, respectively), facilitated IL-8 secretion similar to wild-type CD14 (Fig. 4⇓). The relative ability of mutant CD14 proteins to bind E. coli LPS was similar to their respective ability to facilitate E. coli LPS-mediated IL-8 secretion. One exception to this pattern is that mutant D12K bound E. coli LPS slightly better than R14E (Fig. 2⇑); however, R14E was slightly better at facilitating IL-8 secretion. Interestingly, this mutation occurs in the hydrophilic region previously identified as being involved in LPS signaling (17). However, the differences in relative binding and activation were small between these CD14 mutation proteins and do not represent a significant alteration in the general pattern that the amount of LPS binding correlates to the degree of cellular activation. The correlation of LPS binding to activation is consistent with the idea that a major conformational change that independently affected CD14 function did not occur in these CD14 mutation proteins.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 4.

Human endothelial cell IL-8 secretion in response to CD14 mutation proteins and E. coli LPS. Increasing concentrations of E. coli LPS (as indicated on the x-axis) in 1% CD14-depleted sera were mixed with 0.5 μg/ml of thrombin-cleaved wild-type CD14 and the CD14 mutation protein D10K, E11K, D12K, R14E, R53E, or D59K and added to endothelial cells as described (see Materials and Methods). The experiments were performed on three separate occasions, and the averages from duplicate wells from a representative experiment are shown.

A panel of anti-CD14 mAbs that block LPS binding and bind native tertiary structure was employed to perform epitope-mapping studies

Binding studies with the different CD14 mutation proteins employing three different anti-CD14 mAbs were performed. The generation of these Abs has been previously described (27). These Abs blocked LPS binding (Fig. 5⇓). The blocking of E. coli LPS to CD14 is shown in Fig. 5⇓; however, these Abs also blocked P. gingivalis and H. pylori LPS binding (data not shown). In fact an analysis of 35 master wells with LPS-blocking activity failed to identify any Ab that displayed selective blocking of these three different types of LPS (data not shown). In addition, these Abs bound tertiary epitopes on CD14, as determined by their inability to bind heat-denatured CD14 by ELISA (data not shown). The inability of these Abs to bind denatured CD14 was further analyzed by immunoblot analysis that demonstrated that these Abs did not bind to SDS and heat-denatured CD14 after SDS-PAGE (data not shown). Further studies (23 different anti-CD14 master wells that blocked LPS binding were examined) failed to reveal an Ab that bound a heat-denatured (or presumably linear) epitope by ELISA (data not shown).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 5.

Anti-CD14 mAbs block E. coli LPS binding to CD14. In this experiment the ability of the anti-CD14 mAbs to block E. coli LPS from binding CD14 was examined. CD14 was captured on a 96-well ELISA plate as described in Materials and Methods, and 30 min before the addition of E. coli LPS (100 ng/ml), anti-CD14 mAbs or a negative control Ab (anti-CD40) were added at the concentrations indicated on the x-axis. Two separate experiments were performed with similar results; a representative experiment is shown.

Epitope mapping of mAbs to CD14 revealed that hydrophilic regions 1, 3, and 4 involved in LPS binding are located on the same surface of CD14

Each mAb displayed a different pattern of binding to the CD14 mutation proteins, demonstrating that each Ab bound a different epitope (Table II⇓). In addition, as expected, the mAbs provided a measure of protein conformation independent of LPS binding. For example, CD14 mutation proteins from group 1 (D10S, E11S, D12S, D57K, and D59S), which demonstrated no observable phenotype with respect to LPS binding or host cell activation, demonstrated reduced or no binding to at least one of the mAbs. These data demonstrate that although the CD14 mutation did not influence LPS binding, mAb binding was affected. Conversely, the three CD14 mutation proteins from group 3 all bound at least one CD14 mAb (one of these, R14E, bound all the mAbs; another, E37K, bound two of three of the Abs). These CD14 mutation proteins displayed a significant reduction with all the LPS species examined, making it more difficult to rule out potential conformation effects of these charge reversal mutations. However, based upon the mAb data, the mutations did not induce conformational changes that altered Ab binding. Collectively, these data clearly demonstrate that reductions in mAb binding to CD14 mutation proteins occur independently of their LPS binding properties.

Examination of the CD14 mutation proteins that affected each mAb revealed two different patterns. The first pattern was observed with mAb MY4. Mutation proteins D9K, D10K, D12S, and D12K, but no other mutation proteins, reduced or eliminated MY4 binding. These residues are located at or near hydrophilic region 1 and are located nearly adjacent to each other in the linear sequence of the protein.

m Abs 2G7 and 6C8 represent the second pattern. mAb 2G7 failed to react with CD14 mutation proteins located in hydrophilic region 1, similar to MY4 (D9K, D10K, E11S, E11K, and D12K), although a slightly different pattern was observed. In contrast to MY4, this mAb also failed to bind CD14 mutation protein E37K, a CD14 residue located in hydrophilic region 3 on the linear sequence (see Fig. 1⇑). mAb 6C8 binding was also reduced by mutations located in two different hydrophilic regions found in the amino-terminal sequence of CD14. CD14 mutation proteins R53E, D57K, D59S, and D59K, at or near region 4, and D10S, D10K, and R14E, at or near region 1, all reduced or eliminated mAb 6C8 binding.