Surfactant Proteins A and D Bind CD14 by Different Mechanisms*

Surfactant proteins A (SP-A) and D (SP-D) are lung collectins that are constituents of the innate immune system of the lung. Recent evidence (Sano, H., Sohma, H., Muta, T., Nomura, S., Voelker, D. R., and Kuroki, Y. (1999) J. Immunol. 163, 387–395) demonstrates that SP-A modulates lipopolysaccharide (LPS)-induced cellular responses by direct interaction with CD14. In this report we examined the structural elements of the lung collectins involved in CD14 recognition and the consequences for CD14/LPS interaction. Rat SP-A and SP-D bound CD14 in a concentration-dependent manner. Mannose and EDTA inhibited SP-D binding to CD14 but did not decrease SP-A binding. The SP-A binding to CD14 was completely blocked by a monoclonal antibody that binds to the SP-A neck domain but only partially blocked by an antibody that binds to the SP-A lectin domain. SP-A but not SP-D bound to deglycosylated CD14. SP-D decreased CD14 binding to both smooth and rough LPS, whereas SP-A enhanced CD14 binding to rough LPS and inhibited binding to smooth LPS. SP-A also altered the migration profile of LPS on a sucrose density gradient in the presence of CD14. From these results, we conclude that 1) lung collectins bind CD14, 2) the SP-A neck domain and SP-D lectin domain participate in CD14 binding, 3) SP-A recognizes a peptide component and SP-D recognizes a carbohydrate moiety of CD14, and 4) lung collectins alter LPS/CD14 interactions.

main, a coiled-coil motif neck domain, and a carbohydrate recognition domain (CRD) (2). The CRD region of SP-A is essential for dipalmitoylphosphatidylcholine and galactosylceramide binding, liposome aggregation, the inhibitory effect on lipid secretion, and the augmentation of lipid uptake by alveolar type II cells (4 -11). Likewise, the CRD region of SP-D functions in the recognition of the ligands phosphatidylinositol and glucosylceramide (12,13). In addition to their interaction with lipids and alveolar type II cells, lung collectins interact with macrophages (14 -16) and enhance phagocytosis of a wide spectrum of microorganism (17)(18)(19)(20)(21)(22)(23)(24)(25). Lung collectins are now thought to be important components of the innate immune system of the lung (26 -28).
Studies with transgenic mice provide strong support for a role of pulmonary collectins in host defense properties. Mice homozygous for null alleles of SP-A exhibit increased susceptibility to group B streptococcal and Pseudomonas aeruginosa infections (29 -31). These mice clear the bacteria from the lungs at a slower rate than wild-type mice. Phagocytosis of P. aeruginosa by alveolar macrophages in SP-A Ϫ/Ϫ mice is also significantly decreased (30). Coadministration of SP-A with bacteria into the airway of SP-A Ϫ/Ϫ mice enhance phagocytosis of the bacteria by alveolar macrophages. Secretion of proinflammatory cytokines into the alveolar space is significantly elevated in SP-A Ϫ/Ϫ mice compared with SP-A ϩ/ϩ mice after intratracheal challenge with P. aeruginosa. These studies suggest that SP-A modulates innate immune responses by several different mechanisms. Although these in vivo studies explicitly indicate that SP-A plays a crucial role in host defense of the lung, the molecular basis of SP-A-mediated modification of inflammatory responses remains to be elucidated.
Lipopolysaccharide (LPS), derived from Gram-negative bacteria, is a potent stimulator of inflammation (32). Smooth LPS is composed of O-antigen, core oligosaccharides, and lipid A, while rough LPS lacks O-antigen and a part of the core oligosaccharides (33). The cellular responses to physiological amounts of LPS depend on membrane CD14 that is phosphatidylinositol-anchored to the plasma membrane of myeloid cells (34). A soluble form of CD14 which exists in serum also facilitates the responsiveness of the cells to LPS (35,36). The principal role of CD14 is to bind LPS, but how CD14 acts in transmitting LPS signal remains to be resolved. Recently, Tolllike receptors have been implicated in signaling by LPS and CD14 (37)(38)(39).
SP-A and SP-D bind to rough LPS (24,40). We have recently shown that human SP-A and its collagenase-resistant fragment bind to lipid A and rough LPS but not to smooth LPS (41). In addition, human SP-A interacts directly with CD14, and modulates the LPS-induced cellular responses (41). In the macrophage-like cell line U937 and alveolar macrophages, SP-A inhibits expression and secretion of tumor necrosis factor ␣ induced by smooth LPS. Preincubation of CD14 with SP-A prevents CD14 binding to smooth LPS. The disruption of smooth LPS-CD14 interaction by SP-A may play an important role in reducing inflammatory responsiveness of the lung to smooth LPS. In contrast, the binding of CD14 to rough LPS is significantly increased by the preincubation of CD14 with SP-A. Moreover, the cellular response to rough LPS can be enhanced by SP-A. From the results, we hypothesize that rough LPS, which is a ligand for both CD14 and SP-A, might more effectively bind to the SP-A-CD14 complex.
The purpose of the current study was to extend our examination of lung collectin interactions with LPS and CD14 by investigating: 1) the interaction of SP-D with CD14, 2) the structural determinants that participate in SP-D/CD14 recognition, and 3) the structural determinants involved in SP-A/ CD14 interaction. These studies provide clear evidence that SP-A and SP-D bind CD14 by different mechanisms and alter LPS/CD14 interaction.

EXPERIMENTAL PROCEDURES
Reagents-Rough LPS from Salmonella minnesota Re595 was purchased from Sigma. 3 (7) that had been given an intratracheal instillation of silica in saline 4 weeks before the lavage (42). The surfactant was delipidated with 1-butanol. SP-A was isolated and purified from the delipidated surfactant by mannose-Sepharose 6B column chromatography followed by gel filtration over Bio-Gel A-5m as described previously (43). SP-D was also purified from rat lung lavage as described previously (44,45).
For biotinylation, 150 g of SP-A was incubated for 30 min at room temperature in 0.5 ml of 0.1 M NaHCO 3 buffer (pH 8.3) containing N-hydroxysulfosuccinimidobiotin (Sulfo-NHS-biotin) (Pierce) at a final concentration of 0.5 mg/ml. The protein was then dialyzed against 5 mM Tris buffer (pH 7.4) and stored at Ϫ20°C.
Monoclonal antibodies 1D6 and 6E3 were prepared against rat SP-A as described previously (46). The monoclonal antibodies recognize epitopes in the polypeptide portion of SP-A and have nearly equivalent affinity for the SP-A antigen. The epitopes for antibodies 1D6 and 6E3 have been localized at the CRD and the neck domain, respectively ( Fig.  1) (12).
Recombinant SP-A, SP-D, and SP-A/SP-D Chimera-The isolation and sequencing of the 1.6-kilobase cDNA for rat SP-A was described previously (47). The isolation of the 1.2-kilobase cDNA for rat SP-D was also previously reported (48). The SP-A/SP-D chimera ad5 containing the SP-A region Asn 1 -Met 134 and the SP-D region Cys 261 -Phe 355 was also constructed as described previously (13). Fig. 1 shows a schematic representation of SP-A, SP-D, and chimera ad5. Chimera ad5 exhibits defects of binding dipalmitoylphosphatidylcholine and galactosylceramide and interacting with alveolar type II cells but exhibits the phosphatidylinositol and glucosylceramide binding property of SP-D (13). The proteins were expressed in the baculovirus expression system by the methods of O'Reilly et al. (49), using monolayers of Spodoptera frugiperda cells and Trichoplusia ni (Tni) cells as described previously (13).
For labeling SP-A with Trans 35 S-label, the infected cells were incubated in serum-free medium containing 500 Ci/ml Trans 35 S-label TM . After the incubation for 3 days, the medium was applied to a mannose-Sepharose column in the presence of 2 mM CaCl 2 . The labeled protein was eluted with 5 mM Tris buffer (pH 7.4) containing 2 mM EDTA. The specific activity of 35 S-labeled SP-A was approximately 25 cpm/ng. Electrophoretic analysis revealed that the forms of 35 S-labeled SP-A correlated well with the forms of unlabeled protein.
Recombinant CD14 Expressed in Insect Cells-The construction of recombinant soluble CD14H in which the carboxyl end and the glycosylphophatidylinositol-anchoring site of CD14 were replaced with a protein kinase A site and a 6-histidine tag, was performed by a modified method based on that described by Tapping and Tobias (50). The expression of CD14H protein in the baculovirus expression system was performed by the method (13) described above for recombinant collectins. After incubation of Tni cells with recombinant viruses for 3 days, the medium was collected after low-speed centrifugation and dialyzed against 0.1 M Tris buffer (pH 8.0) containing 0.3 M NaCl. The dialyzed medium was filtered and CD14H was purified by a column of nickelnitrilotriacetic acid beads (Qiagen, Santa Clarita, CA) according to the manufacturer's instruction. The purified protein was finally dialyzed against 5 mM Tris buffer (pH 7.4) containing 0.15 M NaCl and stored at Ϫ20°C. Approximately 1 mg of CD14H protein was obtained from 230 ml of medium.
Recombinant CD14 Expressed in CHO Cells-The cDNA for CD14H was also ligated into a pEE14 plasmid vector (51). The recombinant CD14H was expressed in CHO cells using the glutamine synthetase amplification system (51). The pEE14 vector containing CD14H cDNA was transfected into CHO-K1 cells using LipofectAMINE (Life Technologies, Inc.). Transfected cells were incubated in glutamine-free Glasgow minimum essential medium supplemented with 10% dialyzed fetal bovine serum (complete Glasgow minimum essential medium-10) in the presence of 25 M methionine sulfoximine (Sigma) for 2 weeks and resistant cell lines were cloned. A stable cell line that secretes CD14H was obtained and maintained in the presence of 250 M methionine sulfoximine. To produce soluble CD14H, the cloned cells (0.5 ϫ 10 6 / 100-mm dish) were incubated in complete Glasgow minimum essential medium-10 for 4 days until reaching 70% confluency and then changed to serum-free media ExCell 302 (JRH Biosciences). After 3-4 days incubation, the medium was collected, dialyzed, and applied to a column of nickel-nitrilotriacetic acid beads. The CD14H was then isolated as described above. Approximately 1.5 mg of CD14H protein was obtained from 200 ml of medium.
Analysis of SP-A, SP-D, and Recombinant CD14 -Protein concentrations were estimated by the BCA assay (Pierce) using bovine serum albumin (BSA) as a standard. Protein samples were separated by SDSpolyacrylamide gel electrophoresis (13%) by the method of Laemmli (52) and stained with Coomassie Brilliant Blue.
LPS Binding-The binding of recombinant SP-A and SP-D to Re-LPS was performed by the method described previously (41). The protein binding to LPS was detected using anti-SP-A IgG or anti-SP-D IgG and horseradish peroxidase (HRP)-labeled anti-rabbit IgG. The peroxidase reaction was performed by using o-phenylenediamine as a substrate and the absorbance was measured at 492 nm. 35 S-SP-A was used to determine the effect of monoclonal antibodies on the binding of SP-A to LPS. 35 S-SP-A (5 g/ml) was preincubated at 37°C for 1 h in the presence or absence of 50 g/ml monoclonal antibody 1D6 or 6E3 in 20 mM Tris buffer (pH 7.4) containing 0.15 M NaCl, 1 mg/ml BSA, and 5 mM CaCl 2 . The mixture of SP-A and antibody (50 l) was then incubated with Re-LPS coated onto microtiter wells. After incubation for 5 h, the wells were washed, and 100 l of 0.1 N NaOH solution was added to the wells. The suspended NaOH solution was transferred to vials and mixed with scintillation liquid. The radioactivity of each vial was measured using a ␤-radiation counter.
CD14 Binding-The binding of SP-A and SP-D to CD14H coated onto microtiter wells was performed by the method described previously (41). The protein binding to the CD14 was detected using anti-SP-A IgG or anti-SP-D IgG and HRP-labeled anti-rabbit IgG. Anti-SP-D IgG was used to detect chimera ad5 binding to CD14H since this polyclonal antibody binds chimera ad5 strongly (13).
The ligand blot analysis was carried out by the method (41) described previously with minor modifications. CD14H (1.5-4 g) was electrophoresed and transferred to PVDF membrane. The nonspecific binding was blocked with 20 mM Tris buffer (pH 7.4) containing 0.15 M NaCl, 5 mM CaCl 2 , 5% (w/v) BSA, and 1% (w/v) polyvinylpyrrolidone (binding buffer). The membrane was then incubated at room temperature overnight with 0.5-20 g/ml native SP-A or SP-D in the binding buffer. The membranes were then washed and incubated with anti-SP-A IgG or anti-SP-D IgG for 90 min, followed by the incubation with HRP-labeled anti-rabbit IgG (1:1000) for 90 min. SP-A or SP-D that had bound to the PVDF membrane was visualized by using chemiluminescence reagent (NEN Life Science Products Inc.).
In some experiments 10 g/ml biotinylated SP-A which had been preincubated at room temperature for 2 h in the absence or presence of monoclonal antibody 1D6 or 6E3, or control antibody 6B2 (100 g/ml) was incubated with CD14H transferred onto the PVDF membrane at room temperature overnight. The membrane was washed and further incubated with HRP-conjugated streptavidin D (1:1000) for 20 min. After washing with phosphate-buffered saline containing 0.1% (v/v) Triton X-100, the peroxidase reaction was performed by using diaminobenzidine tetrahydrochloride as a substrate.
Sucrose Density Gradient Centrifugation-The 5-30% (w/v) sucrose gradient was prepared by overlaying 0.25 ml of each step varying by 2.5% sucrose concentration above a bottom cushion of 0.5 ml of 40% sucrose. Forty ng of 3 H-LPS was incubated in the absence or presence of wild type rat SP-A (20 g) and/or CD14H (20 g) at 37°C for 1 h in 200 l of 20 mM Tris buffer (pH 7.4) containing 0.15 M NaCl and 2 mM CaCl 2 . Then 50 l of 12.5% sucrose was added to each incubated sample to adjust a final sucrose concentration at 2.5%, and the samples were applied to the top of the gradient. The gradients were centrifuged at room temperature at 30,000 rpm for 90 min by using RPS 56T rotor (Hitachi Koki Co., Tokyo, Japan). The fractions were collected from the top of the gradient by careful vacuum aspiration. The collected fractions were mixed with scintillation liquid, and the radioactivity of each fraction was measured using a ␤-counter. Recovery of 3 H-LPS ranged from 70 to 95% under all conditions.

RESULTS
Characterization of SP-A, SP-D, Chimera ad5, and CD14 -We have previously characterized wild type SP-A, SP-D, and chimera ad5 synthesized in insect cells using recombinant baculoviruses (13,53,54). These proteins retain the properties of the authentic proteins in interacting with carbohydrates, phospholipids, glycosphingolipids, and alveolar type II cells (13,53,54). In this study we used an invertebrate expression system to produce recombinant SP-A and SP-D and CD14. Recombinant wild type SP-A (wt SP-A) and SP-D (wt SP-D) were purified by affinity chromatography on mannose-Sepharose 6B. The cDNA for human CD14 was isolated from differentiated U937 cells and was engineered to replace the carboxyl end and the putative glycosylphophatidylinositol anchoring site with a protein kinase A site and a 6-histidine tag by the method described by Tapping and Tobias (50). Recombinant CD14 (CD14H) was purified by affinity chromatography on nickel-nitrilotriacetic acid beads. The purified proteins were analyzed by SDS-polyacrylamide gel electrophoresis (Fig. 2). The wt SP-A migrated as a polymorphic band at approximately 30 kDa under reducing conditions. Recombinant SP-A produced in insect cells typically migrated faster than native SP-A derived from lung lavage due to different post-translational modifications as described previously (53). Chimera ad5 migrated as bands at 27-30 kDa as described previously (13). Native and recombinant SP-Ds migrated as bands at approxi-mately 40 -43 kDa, as described previously (13,55). Purified CD14H produced by insect cells migrated as bands at approximately 40 -43 kDa, which is consistent with the result described by Tapping and Tobias (50). All isoforms of SP-As, SP-Ds, and CD14H on the gels were demonstrated to represent immunoreactive proteins by immunoblotting analysis (data not shown). Anti-rat SP-A monoclonal antibody 6E3 recognizes both wt SP-A and chimera ad5 (13).
Properties of Collectin Binding to LPS-Human SP-A binds to rough LPS and causes calcium-induced aggregation of rough LPS (41). Lectin blot analysis performed by Kuan et al. (24) demonstrates that SP-D binds to Rc and Rd strains of rough LPS, and they have described that the binding is inhibited by 100 mM maltose or EDTA (24). Thus, we first examined whether EDTA and excess mannose affect the binding of rat SP-A and SP-D to rough LPS. The wt SP-A and wt SP-D bound to Re-LPS coated onto microtiter wells in the presence of 5 mM CaCl 2 (Fig. 3, A and B). When 20 g/ml wt SP-A or SP-D was incubated with Re-LPS in the presence of 5 mM EDTA, the binding of the collectins to rough LPS was negligible. The binding of SP-A and SP-D to rough LPS in the presence of 0.2 M mannose was significantly reduced to 44 -56% of the levels of those without mannose. These results indicate that the interaction of lung collectins with rough LPS depends on calcium, and the lectin property of lung collectins takes part in the binding to rough LPS.
A previous study from this laboratory indicates that not only SP-A but also the collagenase-resistant fragment of human SP-A bind to rough LPS (41). These findings implicate the neck plus CRD domains in LPS binding. In order to determine which domain of SP-A is involved in the binding of SP-A to rough LPS, the effect of anti-SP-A monoclonal antibodies on the binding of rough LPS was investigated. The epitopes for anti-rat SP-A monoclonal antibodies 1D6 and 6E3 are located at the CRD and the neck domain, respectively ( Fig. 1) (12). 35 S-Labeled SP-A was incubated with Re-LPS coated onto microtiter wells in the absence or presence of antibody 1D6 or 6E3. The binding of 35 S-SP-A to LPS was determined by measuring radioactivities associated with solid phase LPS. As shown in Fig. 3C, antibody 1D6 significantly inhibited the binding of 35 S-labeled SP-A to Re-LPS. In contrast, preincubation of 35 S-labeled SP-A with antibody 6E3 resulted in enhanced LPS binding. These results clearly indicate that the CRD of SP-A is involved in the binding of SP-A to rough LPS.
Binding of Rat Lung Collectins to CD14 -We have recently shown that human SP-A binds CD14 (41). Since SP-D is homologous to SP-A, the possibility that SP-D also binds CD14 was tested. We investigated the binding of rat lung collectins to CD14 expressed in insect cells by ligand blot and microtiter well binding. When CD14H was electrophoresed and transferred to PVDF membrane, it was visualized as bands with apparent molecular mass of 40 -43 kDa by Coomassie Brilliant Blue staining (Fig. 4A). For the ligand blot analysis, the membrane was incubated with native rat SP-A, SP-D, or BSA and probed with anti-SP-A or anti-SP-D IgG. SP-A or SP-D that had bound to the membrane was detected as a band corresponding to that of CD14H (Fig. 4A), demonstrating that lung collectins bind to CD14. Furthermore, we examined the binding of the collectins to CD14H coated onto microtiter wells. Both SP-A and SP-D bound to CD14H in a concentration-dependent manner (Fig. 4B). From these results, we conclude that lung collectins bind CD14 expressed in insect cells.
Properties of Collectin Binding to CD14 -sCD14 expressed using the baculovirus/insect cell system has been shown to be active in stimulating nonmyeloid cells in the presence of LPS (50,56). We next examined the binding of the collectins to CD14 in the presence of EDTA or excess mannose, and compared it to the binding of those to rough LPS. In the presence of Ca 2ϩ wt SP-A bound to CD14H coated onto microtiter wells (Fig. 5A). Inclusion of 1 mM EDTA instead of CaCl 2 in the binding buffer failed to attenuate the binding of wt SP-A to CD14H. The SP-A binding was enhanced rather than inhibited in the presence of 0.2 M mannose. These results are quite different from the binding of SP-A to rough LPS, and indicate that the interaction of SP-A with CD14 does not require calcium and may be independent of the lectin property of SP-A.
The wt SP-D bound to CD14H coated onto microtiter wells in the presence of calcium (Fig. 5B). However, in contrast to the findings with SP-A, the binding of wt SP-D to CD14H was prevented by 0.2 M mannose or 1 mM EDTA. The results clearly indicate that the lectin property of SP-D is involved in SP-D-CD14 interaction. Chimera ad5 also bound to CD14H in the presence of 5 mM CaCl 2 and the binding was clearly inhibited by the presence of EDTA or mannose (Fig. 5C). The results demonstrate that chimera ad5 behaves like SP-D in the binding to CD14H. These findings identify the CRD of SP-D as the likely domain involved in CD14 recognition. The stark contrast in CD14 binding characteristics between SP-A and SP-D demonstrates that the mechanisms of CD14 recognition must be different.
Effect of Anti-SP-A Monoclonal Antibodies on CD14 Binding-Biotinylated SP-A was preincubated alone or with the antibodies 1D6, 6E3, or control antibody 6B2, and the preincubated samples were further incubated with a PVDF membrane containing transblotted CD14. When biotinylated SP-A was incubated with the membrane in the absence or the presence of antibody 6B2, the bands corresponding to those of CD14 were clearly detected (Fig. 6), indicating that biotinylated SP-A bound to CD14. The binding of biotinylated SP-A to CD14H was completely inhibited by the preincubation of biotinylated SP-A with antibody 6E3. Antibody 1D6 decreased, but did not completely inhibit the binding of SP-A to CD14H. The results indicate that surface determinants within or adjacent to the neck domain of SP-A are predominantly involved in the interaction with CD14. The results are consistent with those deexpressed as % of the binding obtained by 35 35 S-SP-A (5 g/ml) was preincubated at 37°C for 1 h in the presence or absence of antibody 1D6 or 6E3 (50 g/ml), and the mixture of SP-A and antibody (50 l) was incubated with Re-LPS coated onto microtiter wells at 37°C for 5 h. The binding of 35 S-SP-A to LPS was detected by measuring radioactivity as described under "Experimental Procedures." The results are scribed in Fig. 5, demonstrating that EDTA and mannose do not inhibit SP-A/CD14 interactions.
We further examined whether the SP-A neck domain present in wt SP-A or chimera ad5 participates in the binding to CD14 using monoclonal antibody 6E3, which binds to chimera ad5 nearly as well as wt SP-A (13). When wt SP-A bound to CD14H coated onto microtiter wells, it was readily detected by anti-SP-A polyclonal IgG (Fig. 7A). However, antibody 6E3 essentially failed to detect the wt SP-A that bound to CD14H. This finding is consistent with the idea that SP-A interacts with CD14 via the neck domain. In contrast, antibody 6E3 effectively recognized chimera ad5 that had bound to CD14H (Fig.  7B). The absorbance obtained by using antibody 6E3 was almost comparable to the level that was detected by anti-SP-D polyclonal IgG. This finding indicates that SP-D type binding is dominant in the ad5 chimera. This conclusion is consistent with the finding that ad5 behaves like SP-D upon interaction with CD14 (see Fig. 5). The results implicate the SP-D CRD region of Cys 261 -Phe 355 in the binding of SP-D to CD14. Collectively, these results demonstrate that the neck domain is involved in the binding of SP-A to CD14, and that the CRD is required for the binding of SP-D to CD14.
Binding of Lung Collectins to Deglycosylated Insect Cellexpressed CD14 -Since CD14 possesses N-linked carbohydrate (57, 58), we next examined whether the carbohydrate moiety of CD14 is required for the binding of the collectins. When CD14H (1.5 g/lane) was electrophoresed and stained with Coomassie Blue, it exhibited three bands at 40 -43 kDa with a main band of 41 kDa (Fig. 8, Coomassie stain/gel), which is in agreement with the result described by Tapping and Tobias (50). The N-glycosidase F-treated CD14H migrated as a band at 39 kDa. Glycosylated and deglycosylated CD14H were electrophoresed and transferred onto PVDF membrane and ligand blot analysis was performed using native SP-A and SP-D. SP-A and SP-D were detected as bands corresponding to those of control CD14H (Fig. 8, ligand binding). SP-A appeared to bind untreated CD14H species of smaller molecular size (40 -41 kDa), whereas SP-D bound CD14H species of larger molecular size (41)(42)(43). SP-A bound deglycosylated CD14 more strongly than the untreated protein. Remarkably, SP-D exhibited almost no binding to deglycosylated CD14.
Binding of Lung Collectins to CD14 Expressed in Mammalian Cells-Proteins that are N-glycosylated in mammalian cells are generally also glycosylated in insect cells. However, the additional complex oligosaccharide synthesis does not ap-pear to occur in insect cells, although the pentasaccharide core common to N-glycoproteins such as Man3GlcNAc2 is synthesized in insect cells like in mammalian cells (49). In addition to N-linked glycosylation, human sCD14 and recombinant sCD14 produced by CHO cells contains O-linked glycosylation (59). Thus, we expressed sCD14 in CHO cells and examined whether lung collectins bind mammalian CD14. sCD14 produced by CHO cells (CHO CD14) migrated as broad bands with 46 -56 kDa when analyzed by electrophoresis under denaturing and reducing conditions (Fig. 9A, Coomassie stain). The predominant forms of CHO CD14 exhibited molecular mass of 51-53 kDa. Ligand blot analysis revealed that both SP-A and SP-D bound CHO CD14 (Fig. 9A, ligand binding). We also performed the ligand binding using CHO CD14 treated with N-glycosidase F, neuraminidase, and O-glycosidase. The enzyme treatment resulted in the reduction of molecular mass to 39 kDa (Fig. 9B, Coomassie stain), which is in agreement with the results described by Stelter et al. (59). The results obtained by ligand blot analysis with the deglycosylated form of CHO CD14 are different between the collectins. SP-A clearly bound deglycosylated CHO CD14. However, SP-D failed to bind the deglycosylated form. The results obtained by mammalian cell-expressed CD14 are consistent with those found with insect cellexpressed CD14.
Taken together, the data indicate that SP-A binds the peptide portion of CD14 and that SP-D binds the carbohydrate moiety of CD14. The results clearly demonstrate the different mechanisms utilized by the lung collectins in recognition of CD14.
The Effect of Lung Collectins on LPS-CD14 Interaction-A recent study from this laboratory has shown that human SP-A decreased the binding of CD14 to smooth LPS and increased the binding of CD14 to rough LPS (41). The effect of rat SP-A and SP-D and chimera ad5 on the binding of CD14H to solid phase smooth and rough LPS was examined in this study. As with human SP-A, preincubation of wt rat SP-A with CD14H inhibited binding to smooth LPS but increased the CD14H binding to rough LPS (Fig. 10, A and B). The wt SP-D and chimera ad5 attenuated the binding of CD14H to smooth LPS like wt SP-A (Fig. 10A). However, the preincubation of wt SP-D and chimera ad5 with CD14 also decreased the binding of CD14H to rough LPS (Fig. 10B). These results indicate that SP-A and SP-D have distinct effects upon LPS-CD14 interaction.
Because the recent (41) and present studies indicate that the FIG. 4. Rat lung collectins bind CD14. A, ligand blot analysis of CD14H. Three g of CD14H produced by insect cells was electrophoresed and transferred to PVDF membrane. CD14H on the membrane was visualized by Coomassie Brilliant Blue staining (Coomassie stain). The membrane was also incubated with rat SP-A, rat SP-D, or BSA and probed with anti-SP-A or anti-SP-D IgG, followed by incubation with HRP-labeled anti-rabbit IgG (ligand binding). B, concentration-dependent binding of lung collectins to CD14H. Fifty-l aliquots of 10 g/ml BSA (Ⅺ, E) or CD14H (q, f) were coated onto microtiter wells and incubated with the indicated concentrations of rat SP-A (f, Ⅺ) and rat SP-D (q, E) at 37°C overnight. The binding of collectins to CD14H was detected using anti-SP-A or anti-SP-D IgG as described under "Experimental Procedures." The data are mean Ϯ S.E. of three experiments.
binding of CD14 to rough LPS is apparently enhanced by the preincubation of CD14 with SP-A, we speculate that rough LPS can interact with an SP-A-CD14 complex. In order to gain better insight into the interaction of rough LPS with SP-A and CD14, sedimentation analysis of complex mixtures of SP-A, CD14, and rough LPS was performed by using sucrose density gradient centrifugation. We used 3 H-labeled LPS prepared from a rough strain of bacteria (Rb) which possesses somewhat longer core oligosaccharides than Re-LPS. When 3 H-LPS alone was applied to the top of the gradient and centrifuged, 14% of the total radioactivity was recovered at the bottom fraction (Fig. 11A), indicating the aggregated form of 3 H-LPS. After the preincubation of 3 H-LPS with wt SP-A, 3 H-LPS migrated a short distance into the gradient and was recovered mainly in fraction 3 (Fig. 11B). A similar profile of the migration of 3 H-LPS was observed upon incubation with CD14H (Fig. 11C). In contrast, when 3 H-LPS was preincubated in the presence of both wt SP-A and CD14H, a significant amount of 3 H-LPS remained at the top of the gradient (Fig. 11D). As much as 16% of total radioactivity was detected at the top of the gradient, while only 4.5, 5, or 6% of that was recovered from 3 H-LPS FIG. 5. Effects of EDTA and mannose on the binding of SP-A, SP-D, and chimera ad5 to CD14. CD14H (10 g/ml, 50 l/well) produced by insect cells was coated onto microtiter wells and incubated with 20 g/ml wt SP-A (A), wt SP-D (B), or chimera ad5 (C) at 37°C overnight in 10 mM Hepes buffer (pH 7.4) containing 0.15 M NaCl, 5 mg/ml BSA, and 5 mM CaCl 2 (Ca 2ϩ ). The proteins binding to CD14H were detected using polyclonal antibody as described under "Experimental Procedures." One mM EDTA was also included instead of CaCl 2 (EDTA). In some experiments the binding study was performed in the presence of 0.2 M mannose (mannose). The data shown are mean Ϯ S.E. of three experiments.
FIG. 6. Monoclonal antibody directed against the SP-A neck blocks binding to CD14. CD14H (5 g) produced by insect cells was electrophoresed and transferred to PVDF membrane. The membrane was incubated with 10 g/ml biotinylated SP-A which had been preincubated alone or with anti-SP-A monoclonal antibody 1D6, 6E3, or control antibody 6B2 (100 g/ml). The binding of biotinylated SP-A to CD14H was detected by using HRP-conjugated streptavidin D as described under "Experimental Procedures." FIG. 7. Antibody 6E3 fails to detect wt SP-A but not chimera ad5 bound to CD14. CD14H (10 g/ml, 50 l/well) produced by insect cells was coated onto microtiter wells and incubated with wt SP-A or chimera ad5 at 37°C overnight as described under "Experimental Procedures." A, the binding of wt SP-A to CD14H was detected by using anti-SP-A IgG or anti-SP-A monoclonal antibody 6E3. B, the binding of chimera ad5 to CD14H coated onto microtiter wells was detected by using anti-SP-D IgG or antibody 6E3. The data shown are mean Ϯ S.E. of three experiments. alone, 3 H-LPS plus SP-A or 3 H-LPS plus CD14H, respectively. Taken together, these results show that inclusion of SP-A in the solution containing CD14H and rough LPS altered the migration profile of LPS. This may be explained by assuming that SP-A can simultaneously interact with CD14 and rough LPS. Collectively, these results demonstrate that lung collectins can alter LPS-CD14 interaction. DISCUSSION This study demonstrates that the lung collectins, SP-A and SP-D, bind CD14 by different mechanisms. The structure required for the binding of SP-A to rough LPS is distinct from that for CD14 binding. SP-A interacts with CD14 and rough LPS via the neck domain and the CRD, respectively. In contrast, SP-D interacts with CD14 and rough LPS via the CRD. This work also demonstrates that the structure of CD14 required for SP-A binding is different from that for SP-D binding. SP-A binds the peptide portion of CD14 but SP-D binds its carbohydrate moiety.
The CRDs of lung collectins appear to be the principal protein domains involved in binding to rough LPS in this study. The requirement for calcium and the lectin property also sup-port the role of CRD in LPS binding. The present results are consistent with the study by Kuan et al. (24) demonstrating that the binding of SP-D to rough strain bacteria was inhibited by the presence of EDTA or maltose. The interaction with rough LPS may also be one of the important physiological functions of the CRD of the lung collectins.
In contrast to the association with rough LPS, the interaction of SP-A with CD14 does not require calcium and is not attenuated in the presence of mannose. Since antibody 6E3 completely inhibited binding of SP-A to CD14 and antibody 6E3 failed to recognize the SP-A that had bound to CD14, the neck domain of SP-A is likely to be directly involved in CD14 binding. Since antibody 6E3 inhibits surfactant lipid uptake by type II cells, the neck domain also plays some role in lipid recognition (12). A previous study using SP-A/SP-D chimeric proteins also suggest that the neck region may account for some but not all of the lipid binding properties of SP-A (55). In addition, the mutants in which the regions in the neck domain of SP-A are deleted fail to bind to the mannose-Sepharose column (12), indicating that the neck domain might be important for the folding of SP-A. The present study demonstrating that SP-A interacts with CD14 via the neck domain may imply a novel function of the SP-A neck domain in protein-protein interactions.
However, the neck domain of SP-A in chimera ad5 does not seem to be sufficient for interacting with CD14. The ad5 chimera binds CD14 via the SP-D CRD but not via the SP-A neck domain, since anti-SP-A monoclonal antibody 6E3 can effectively detect the chimera ad5 that is bound to CD14H. In contrast, the 6E3 epitope becomes unavailable when wt SP-A binds CD14H. EDTA and excess monosaccharide alter the binding of the chimera to CD14H, providing further evidence that the CRD of SP-D is the major site for binding. Although the role of the SP-A neck domain in the chimera ad5 molecule is unclear, the substitution of the SP-D region Cys 261 -Phe 355 for the corresponding SP-A region may alter the structure and the function of the SP-A neck domain.
The neck domain of SP-A contains both an amphipathic helix and a segment of hydrophobic amino acids (60). In contrast, SP-D lacks such a clearly identifiable hydrophobic region (61). The structural differences between SP-A and SP-D probably account for the differences in interacting with CD14. CD14 contains 10 leucine-rich repeat motifs (LXXLXLX) (62). In addition, the NH 2 -terminal 152 amino acids of CD14 are sufficient to bind LPS (63). Furthermore, the region including amino acids 57 to 64 has been recently suggested to be involved in LPS binding (64,65). The SP-A neck domain with amphipathic and hydrophobic regions may interact with the hy- FIG. 8. Ligand blot analysis of N-glycosidase F-treated CD14 reveals carbohydrate is required for SP-D recognition. CD14H (1.5 g) produced by insect cells was treated with 1 unit of N-glycosidase F at 37°C for 2 h as described under "Experimental Procedures." The proteins electrophoresed on the gel and those transferred on the PVDF membranes were visualized by staining with Coomassie Brilliant Blue (Coomassie stain). The PVDF membranes were also incubated with SP-A (2 g/ml) and SP-D (0.5 g/ml) and the binding of SP-A and SP-D to N-glycosidase F-treated and -untreated CD14H was detected (ligand binding) using anti-SP-A and anti-SP-D IgG, respectively, as described under "Experimental Procedures." FIG. 9. SP-D fails to recognize deglycosylated CD14 produced in mammalian cells. CD14H (4 g) produced by CHO cells (CHO CD14) was treated with N-glycosidase F, neuraminidase, and O-glycosidase as described under "Experimental Procedures." CHO CD14 (A) and deglycosylated CHO CD14 (B) electrophoresed on the gels were either stained with Coomassie Brilliant Blue (Coomassie stain) or transferred onto the PVDF membranes. The PVDF membranes were incubated with SP-A (5 g/ml), SP-D (20 g/ml), or BSA as a control, and the binding of SP-A and SP-D to CHO CD14 was detected (ligand binding) using anti-SP-A or anti-SP-D IgG, respectively, as described under "Experimental Procedures." drophobic leucine-rich region of CD14.
We have recently demonstrated that SP-A modulates the LPS-induced cellular response in a macrophage-like cell line and alveolar macrophages (41). SP-A can inhibit smooth LPSinduced tumor necrosis factor ␣ expression by alveolar macrophages. We have proposed that the binding of SP-A to CD14 prevents the interaction of CD14 with smooth LPS. In contrast, rough LPS-induced tumor necrosis factor ␣ expression is increased by preincubation of alveolar macrophages with SP-A. The binding of CD14 to rough LPS is modestly enhanced by SP-A. Rough LPS, which is a ligand for both CD14 and SP-A, appears to more efficiently associate with an SP-A-CD14 com-plex. The present study strongly supports this idea, since SP-A interacts with rough LPS via its CRD and with CD14 via its neck domain. The distinct structural domain topology of SP-A binding to rough LPS and CD14 may enable the protein to interact simultaneously with both ligands.
Analysis by sucrose density gradient centrifugation revealed that the sedimentation of 3 H-LPS was markedly altered in the presence of SP-A or CD14. One interpretation of this finding is that 3 H-LPS may partially disaggregate in the presence of SP-A or CD14. Incubation of 3 H-LPS with both SP-A and CD14 changes the distribution of 3 H-LPS more dramatically. This migration is quite different from that induced by SP-A or CD14 alone. The SP-A function of dispersing LPS aggregates may be similar to the function described for LPS-binding protein which enables a 3 H-LPS⅐ 35 S-sCD14 complex to form (66,67). This FIG. 10. SP-A and SP-D alter CD14 interactions with rough and smooth LPS. CD14H (5 g/ml) was preincubated alone (control) or with wt SP-A, wt SP-D, or chimera ad5 (50 g/ml) at 37°C for 1 h. The preincubated mixtures were then incubated with smooth (A) or rough (B) LPS coated onto microtiter wells, and the binding of CD14H to LPS was detected by anti-CD14 polyclonal antibody. The results are expressed as relative absorbance (%) compared with that obtained from the binding of CD14H alone to LPS (100%). The absorbance at 492 nm obtained from the binding of CD14 to smooth or rough LPS was 0.236 or 0.176, respectively. The data are the average of two experiments.
FIG. 11. Distribution of 3 H-LPS analyzed by sucrose density gradient centrifugation. 3 H-LPS (40 ng) was incubated alone (A), with 20 g of SP-A (B), with 20 g of CD14H (C), or with both SP-A and CD14H (D) at 37°C for 1 h before application to a 5-30% sucrose density gradient centrifugation as described under "Experimental Procedures." The radioactivity in each fraction is presented as % of total recovered radioactivity. The arrow indicates the fraction containing the maximum radioactivity. The data shown are a representative one of three experiments. study clearly indicates that the physical form of rough LPS in the presence of CD14 can be significantly modified by the addition of SP-A. Collectively, these results support the idea of a ternary complex of SP-A, CD14, and rough LPS.
The ternary interaction of SP-A with CD14 and rough LPS suggests one of the important roles of SP-A in the host defense mechanism of the lung. By associating with rough LPS or dispersing rough LPS aggregates, SP-A may facilitate CD14 interactions with LPS and initiate physiological responses against pathogens. The present results also demonstrate that SP-D inhibits the binding of CD14 to smooth and rough LPS. In contrast, SP-A enhances the binding of CD14 to rough LPS but inhibits the binding to smooth LPS. Thus, this study indicates that SP-D and SP-A act differently on LPS-CD14 interaction. Since the association of LPS with CD14 also plays an important role in LPS clearance (68), these findings are consistent with other studies demonstrating that SP-A but not SP-D enhanced phagocytosis of rough LPS-containing bacteria by alveolar macrophages (69). Since most Gram-negative bacteria colonizing the respiratory tract express rough LPS (70), the increased association of CD14 with rough LPS induced by SP-A may be important to initiate the immune response. However, the interaction of CD14 with LPS also results in the release of numerous inflammatory mediators which can have marked pathological effects on the lung without adequate regulation. SP-D may act as a negative regulator to protect against a chronic inflammatory state. The role of SP-D in modulating LPS-elicited cellular responses is now under investigation.
The characteristics of mice harboring null alleles for SP-A provide strong in vivo evidence that the collectins are important components of the innate immune system of the lung (29 -31). In one recent report (71) production of tumor necrosis factor ␣ and nitric oxide in SP-A Ϫ/Ϫ mice has been shown to significantly increase after intratracheal administration of smooth LPS when compared with wild type mice. Intratracheal administration of SP-A to SP-A Ϫ/Ϫ mice restored the production of proinflammatory cytokines to that of SP-A ϩ/ϩ mice. The study demonstrates that endogenous and exogenous SP-A inhibit LPS-induced cytokine production in vivo. Earlier in vitro studies (41) also demonstrate that SP-A decreases the binding of smooth LPS to CD14 and markedly attenuates inflammatory cellular responses. Taken together, these in vivo and in vitro studies are consistent with the idea that SP-A interacts directly with immune cells to modulate LPS-induced cellular responses, suggesting an anti-inflammatory role of SP-A. The present study provides the molecular and mechanistic details by which lung collectins can modulate the innate immune response to LPS in the lung.
In conclusion, this study provides clear evidence that SP-A and SP-D bind CD14. The neck region of SP-A and the CRD region of SP-D are involved in CD14 recognition. SP-A binding to CD14 enhances rough LPS binding but inhibits smooth LPS binding. SP-D binding to CD14 inhibits smooth and rough LPS binding. The lung collectin-CD14 interactions are likely to play a critical role in modulating the inflammatory response of the lungs to specific pathogens.