Pulmonary Surfactant Protein D Inhibits Lipopolysaccharide (LPS)-induced Inflammatory Cell Responses by Altering LPS Binding to Its Receptors*

Pulmonary surfactant protein D (SP-D) is a member of the collectin family that plays an important role in regulating innate immunity of the lung. We examined the mechanisms by which SP-D modulates lipopolysaccharide (LPS)-elicited inflammatory cell responses. SP-D bound to a complex of recombinant soluble forms of Toll-like receptor 4 (TLR4) and MD-2 with high affinity and down-regulated tumor necrosis factor-α secretion and NF-κB activation elicited by rough and smooth LPS, in alveolar macrophages and TLR4/MD-2-transfected HEK293 cells. Cell surface binding of both serotypes of LPS to TLR4/MD-2-expressing cells was attenuated by SP-D. In addition, SP-D significantly reduced MD-2 binding to both serotypes of LPS. A chimera containing the N-terminal region and the collagenous domain of surfactant protein A, and the coiled-coil neck and lectin domains of SP-D, was a weak inhibitor of LPS-induced cell responses and MD-2 binding to LPS, compared with native SP-D. The collagenase-resistant fragment consisting of the neck plus the carbohydrate recognition domain of SP-D also was a very weak inhibitor of LPS activation. This study demonstrates that SP-D down-regulates LPS-elicited inflammatory responses by altering LPS binding to its receptors and reveals the importance of the correct oligomeric structure of the protein in this process.

Pulmonary surfactant protein D (SP-D) 3 is a member of the collectin protein family that also includes surfactant protein A (SP-A) and mannose binding lectin (1,2). The structure of the collectins is characterized by four domains consisting of: 1) an N terminus involved in interchain disulfide bonding, 2) a collagen-like domain, 3) a coiled-coil neck domain, and 4) a carbohydrate recognition domain (CRD) (3). SP-A and mannose binding lectin contain collagenous domains consisting of 23 and 19 repeating Gly-X-Y triplets, respectively, with an interruption at the middle of the collagenous sequence (4,5). In contrast, SP-D possesses a longer collagenous tail composed of 59 Gly-X-Y repeats without an interruption (6). These differences cause distinct oligomeric organization, with SP-A and mannose binding lectin exhibiting bouquet-like structures consisting of either six or four trimeric subunits (7) and SP-D exhibiting cruciform structures composed of four trimeric subunits (8).
Lipopolysaccharide (LPS) is a principal component of the outer membrane of Gram-negative bacteria that activates macrophages and induces a variety of inflammatory mediators, including TNF-␣, IL-1, IL-6, IL-8, and interferon (9). LPS composed of O-antigen, core oligosaccharide, and lipid A is named smooth LPS, and LPS lacking O-antigen and a part of the core oligosaccharides is named rough LPS (10). Toll-like receptor 4 (TLR4) plays a critical role in recognition and signaling by LPS (11,12). MD-2 binds directly to LPS and is required for TLR4mediated signaling induced by LPS (13,14). Structural examination of the TLR4-MD-2 complex revealed that MD-2 binds to the concave surface of the N-terminal and central domains of TLR4 (15). A study with recombinant soluble forms of the extracellular TLR4 domain (sTLR4) and MD-2 (sMD-2) from our laboratory indicates the importance of the N-terminal region of TLR4 in MD-2 binding (16).
Engineered genetic defects in the pulmonary collectins of mice have revealed their important functions in protecting the lung from microbial infections and inflammation. SP-D-null mice infected with group B Streptococcus or Haemophilus influenza by intra-tracheal instillation show increased inflammation and inflammatory cell recruitment in the lung (17). Increased pulmonary inflammation in LPS (Escherichia coli O55:B5, smooth serotype)-instilled SP-D Ϫ/Ϫ mice and wildtype mice was decreased by intratracheal administration of SP-D and pulmonary surfactant (18). Intratracheal recombinant SP-D prevents endotoxin shock in the newborn preterm lamb (19). These in vivo studies provide compelling evidence that SP-D dampens pulmonary inflammation induced by LPS.
We have previously shown that SP-A modulates LPS-induced inflammatory cell responses by direct interaction with TLR4 and MD-2 (20,21). SP-A attenuates or enhances LPSinduced inflammatory responses in a ligand-specific manner. For example, SP-A inhibits TNF-␣ secretion elicited by the smooth serotype of LPS, which is not a ligand for SP-A. However, SP-A can also enhance TNF-␣ secretion stimulated by rough LPS, which is a ligand for the protein. We have recently reported that SP-D directly binds to the extracellular domains of TLR4 and TLR2 through its CRD (22).
The purposes of this study were: 1) to determine whether SP-D binds a complex of TLR4 and MD-2 using sTLR4 and sMD-2, 2) to determine whether SP-D affects cytokine secretion and signaling induced by distinct serotypes of LPS, 3) to investigate the mechanism by which SP-D alters TLR4-mediated LPS signaling, and 4) to examine the role of the native oligomeric structure of SP-D in modulating LPS-induced inflammation. We now demonstrate that SP-D down-regulates inflammatory cell responses elicited by smooth and rough serotypes of LPS by altering LPS-receptor interaction.
SP-D-The 1.181-kb cDNA for human SP-D was inserted into a pEE14 plasmid vector, and recombinant human SP-D was expressed in CHO-K1 cells using the glutamine synthetase gene amplification system (25). CHO-K1 cells expressing human SP-D were grown in glutamate-free Glasgow minimum essential medium (Invitrogen) containing dialyzed fetal calf serum and 25 M methionine sulfoximine for gene amplification. For protein purification, the cells were transferred into serum-free EXCELL 302 medium (SAFC Biosciences, Lenexa, KS) and incubated for 3-4 days. The medium was collected, and four additional harvests were performed, allowing 24 -48 h of culture between harvests. The medium adjusted to pH 7.4 with 1 M Tris buffer (pH 9.0) was finally filtered with a 0.45-m filter and applied to mannose-Sepharose 6B column. The proteins were eluted with 20 mM Tris buffer (pH 7.4) containing 2 M NaCl, 50 mM EDTA, and dialyzed against 5 mM Tris buffer (pH 7.4) containing 0.15 M NaCl for 2 days.
Chimeric Protein-We constructed a chimeric protein with SP-A and SP-D (A/D chimera) in which the N-terminal region and the collagenous domain of SP-D were replaced with those of SP-A, as represented schematically in Fig. 6A. The A/D chimera consists of Glu 1 -Pro 80 of SP-A and Asp 203 -Phe 355 of SP-D. The cDNA for the chimera was constructed by using PCR and the overlap extension method (26) using the cDNAs for SP-A and SP-D. The two primers used at the SP-A/SP-D splicing junction were 5Ј-GGAGGTCCCGAAGGTCTACAACG-AAGAGA-3Ј and 5Ј-CCTCCAGGGCTTCCAGATGTTGCT-TCTC-3Ј. An SP-A sense and an SP-D antisense primers were used as follows: 5Ј-AAGCTTATGTGGCTGTGCCCTCTGG-CCC-3Ј and 5Ј-GAACACCAGACGCTCAAGACTAGATCT-3Ј, respectively. The BamHI and XbaI sites were incorporated into the flanking 5Ј and 3Ј primers, respectively. The construct was inserted into the pEE14 plasmid vector using the BamHI and XbaI sites. The organization of the recombinant plasmid and sequence of the cDNA insert was confirmed by a combination of restriction enzyme mapping and DNA sequencing. The recombinant A/D chimera was expressed using the glutamine synthetase gene amplification system (25), and the recombinant protein was produced and purified by affinity chromatography as described above for SP-D production.
Collagenase Treatment-Because the yield of the A/D chimera is greater than that of SP-D, the chimeric protein was used to obtain the collagenase-resistant fragment (CRF). The A/D chimera was digested with collagenase III from Clostridium histolyticum at 37°C for 22 h. The CRF was isolated by gelfiltration chromatography using a Superose 610/300 GL (GE Healthcare Bio-Science AB). The N-terminal sequence of the isolated CRF was determined by using an Applied Biosystems (Foster City, CA) amino acid sequencer. The sequences obtained were GPPGLP or GLPDVA, corresponding to the end of the collagenous domain of human SP-A (Gly 75 -Prp-Pro-Gly-Lys-Pro 80 ) or the SP-A/SP-D splice junction (Gly 78 -Lys-Pro 80 of SP-A/-Asp 203 -Val-Ala 205 of SP-D), respectively. This indicates that the N-terminal region and the collagenous domain of SP-A were removed from the A/D chimera.
Endotoxin Removal-Endotoxin in the protein preparations was removed using polymixin-agarose and octyl-␤-D-glucopyranoside as described elsewhere (27). The endotoxin level determined by Limulus amebocytes lysate assay was below 1.41 pg/g of protein in the preparations of SP-D, the A/D chimera, and CRF.
Western Blot-A 100-ng sample of the protein was subjected to 13% SDS-PAGE under reducing and denaturing conditions, and transferred onto a polyvinylidene difluoride membrane.
The membrane was immunoprobed with anti-human SP-D polyclonal antibody (5 g/ml), or anti-human SP-A polyclonal antibody (5 g/ml), or anti-human SP-D monoclonal antibody 7C6 or 7A10 (2 g/ml), followed by incubation with HRP-conjugated anti-rabbit IgG or HRP-conjugated anti-mouse IgG. The proteins that reacted with the antibodies were visualized by using a chemiluminescence reagent (SuperSignal, Pierce) according to the manufacturer's instructions.
Electron Microscopy-SP-D, the chimeric protein, and the CRF were diluted to 10 g/ml in 50% glycerol and 20 mM ammonium bicarbonate, sprayed onto freshly cleaved mica, and rotary-shadowed molecules were observed under an H-7650 electron microscope (Hitachi Co. Ltd, Tokyo, Japan) operated at 75 kV.
Binding of SP-D to LPS-The binding of the biotinylated SP-D to Re595, Rc, or O26:B6 LPS (5 g/well) coated onto microtiter wells was performed as described previously (22). To determine the effects of anti-SP-D monoclonal antibodies on the SP-D binding to LPS, the protein (2 g/ml) was first incubated with anti-SP-D monoclonal antibody 7C6 or 7A10 at 37°C for 1 h. The mixture of SP-D and antibody was then added to the LPS-coated wells (1 g/well), and the mixture was incubated at 37°C for 3 h.
Binding of SP-D to an sTLR4-sMD-2 Complex-A soluble form of recombinant extracellular TLR4 domain (sTLR4: Met 1 -Lys 631 ) containing a His 6 tag at the C-terminal end and a soluble form of recombinant MD-2 (sMD-2) containing the C-terminal fusion V5 tag and a His 6 tag were expressed using a baculovirus-insect cell expression system as described previously (24). sTLR4 and sMD-2, co-expressed in insect cells, were purified from the medium using a nickel-chelating HisTrap column (HP 5 ml, GE Healthcare). The complex of sTLR4 and sMD-2 was purified by gel-filtration chromatography on a Superose 610/300 GL column as described above and eluted with an apparent molecular mass of 110 kDa.
The biotinylated SP-D (100 ng) and/or an sTLR4-sMD-2 complex (100 ng) were incubated for 2 h at 37°C in 100 l of 10 mM Hepes buffer (pH 7.4) containing 0.15 M NaCl, 2 mM CaCl 2 , and 5% (w/v) bovine serum albumin (buffer A). After the incubation, the volume of the mixture was adjusted to 500 l by the addition of the buffer A and further incubated with anti-V5 antibody-conjugated agarose for 2 h at 4°C, with gentle shaking for immunoprecipitation. As a control experiment, affinity-purified mouse IgG2a (eBioscience, San Diego, CA) and protein G-Sepharose were used. For the affinity adsorption assay with the biotinylated SP-D, 800 ng of the biotinylated SP-D and 100 ng of an sTLR4-sMD-2 complex were incubated with streptavidin-conjugated agarose for 30 min at 4°C. The beads were sedimented by centrifugation and washed four times with 10 mM Hepes buffer (pH 7.4) containing 0.15 M NaCl, 2 mM CaCl 2 , and 0.1% (v/v) Triton X-100. The final pellet obtained was suspended in SDS-sample buffer and subjected to SDS-PAGE. Western blotting was performed as described above with HRPlabeled streptavidin, anti-V5 antibody, or anti-sTLR4 polyclonal antibody, to detect biotinylated SP-D, sMD-2, or sTLR4.
TNF-␣ Secretion from Alveolar Macrophages-Alveolar macrophages were obtained by bronchoalveolar lavage of Sprague-Dawley rats. The lungs were lavaged with pyrogen-free saline (Otsuka Pharmaceutical Co., Tokyo, Japan) containing 1 mM EDTA, and alveolar macrophages were sedimented by centrifugation at 150 ϫ g for 10 min. Isolated macrophages were plated at 1 ϫ 10 5 cells/well in a 96-well plate (Falcon) in Dulbecco's modified Eagle's medium containing 10% fetal calf serum, and next incubated for 2 h at 37°C in a 5% CO 2 atmosphere to allow adherence.
The indicated concentrations of SP-D, the chimera, or the CRF were incubated with macrophages for 30 min at 37°C and subsequently incubated with the cells for 5 h at 37°C, after the addition of LPS. The medium was collected and the concentrations of secreted TNF-␣ were determined by using an OptEIA rat TNF set (BD Pharmingen, San Diego, CA) according to the manufacturer's instructions.
Binding of sMD-2 to LPS-One microgram of Re595 LPS, or Rc LPS in 20 l of ethanol, or 5 g of O26:B6 LPS, or O111:B4 LPS, in 50 l of distilled water were added to microtiter wells (Immulon 1B, Thermo, for rough LPS, MultiSorp, Nunc, for smooth LPS) and the solvent was evaporated in the ambient air. Nonspecific binding was blocked with 20 mM Tris buffer (pH 7.4) containing 0.15 M NaCl and 3% (w/v) bovine serum albumin. The sMD-2 (1 or 2 g/ml) was preincubated with SP-D (20 or 40 g/ml), the A/D chimera, or the CRF, and the mixture was added to the wells and further incubated for 3 h at 37°C. The wells were washed with PBS containing 0.1% (v/v) Triton X-100 and 3% (w/v) skim milk, and incubated with an anti-His tagpolyclonal antibody at 37°C for 1 h, followed by incubation with HRP-labeled anti-rabbit IgG. The peroxidase reaction was performed by using o-phenylenediamine as a substrate after washing the wells with PBS containing 0.1% (v/v) Triton X-100. The binding of sMD-2 to LPS was finally detected by measuring the absorbance at 492 nm.
Other Methods-Polyclonal antibodies against human SP-A and SP-D were raised in rabbits by immunizing with purified recombinant SP-A and SP-D, respectively (21,22). Protein concentrations were estimated by the BCA assay (Pierce) using bovine serum albumin as a standard.

RESULTS
SP-D Binds to an sTLR4-sMD-2 Complex-We have previously shown that SP-D interacts with TLR4 (22) through the CRD of the collectin in a Ca 2ϩ -dependent manner. We isolated a complex of sTLR4 and sMD-2 by gel-filtration chromatography and examined whether SP-D binds to an sTLR4-sMD-2 complex. When sMD-2 was immunoprecipitated by anti-V5 antibody-conjugated agarose, SP-D as well as sTLR4 was coprecipitated (Fig. 1A, WB:streptavidin and ␣sTLR4). The biotinylated SP-D also co-sedimented with sTLR4 and sMD-2 by an affinity adsorption assay with streptavidin-conjugated agarose (Fig. 1B, WB:␣sTLR4 and ␣V5). These results indicate that SP-D can bind a complex of TLR4 and MD-2.
SP-D Binds to Re595 LPS and Rc LPS by Different Mechanisms-We have previously shown that SP-A binds to Re595 LPS and Rc LPS but not to O26:B6 LPS (20). SP-D has been reported to bind to Rc LPS and Rd LPS but not to Ra LPS and smooth LPS (28). In this study we also examined the binding of SP-D to rough LPS (Rc LPS from E. coli and Re595 LPS from S. minnesota) and smooth LPS (O26:B6 LPS from E. coli) coated onto microtiter wells. SP-D bound to Re595 LPS and Rc LPS in the presence of Ca 2ϩ in a concentration-dependent manner (Fig. 2, A and B). SP-D failed to bind solid-phase O26:B6 LPS. The SP-D binding to Re595 LPS was not significantly attenuated by the presence of EDTA ( Fig. 2A), whereas EDTA completely blocked the binding of SP-D to Rc LPS (Fig.  2B). Therefore, the SP-D binding to Rc LPS but not to Re595 LPS was Ca 2ϩ -dependent.
The effect of anti-SP-D monoclonal antibodies on the LPS binding was also investigated. Antibody 7C6 completely blocked the binding of SP-D to Re595 LPS, and antibody 7A10 partially inhibited this binding (Fig. 2C). In contrast, antibody 7A10 but not antibody 7C6 completely inhibited the SP-D binding to Rc LPS (Fig. 2D). Because the epitopes for antibodies 7A10 and 7C6 are located at the CRD and the neck region, respectively (22), the results suggest that the neck region and the CRD are involved in the binding of SP-D to Re595 LPS and Rc LPS, respectively. These results are consistent with those obtained from the experiments demonstrating Ca 2ϩ dependence, or independence for binding different ligands. The data confirm that rough LPS but not smooth LPS is an SP-D ligand and indicate that SP-D binds to Re595 and Rc LPS by different mechanisms. The isolated sTLR4-sMD-2 complex (100 ng) was incubated with biotinylated SP-D (100 ng) at 37°C for 2 h. The sTLR4-sMD-2 complex was immunoprecipitated with anti-V5 antibody-conjugated agarose, and the immunoprecipitate was subjected to SDS-PAGE (7.5-15% polyacrylamide gel) under reducing conditions. As a control experiment, affinity-purified mouse IgG and protein G-Sepharose were used (normal IgG). The Western blot was then performed by using anti-V5 antibody for sMD-2, anti-sTLR4 polyclonal antibody for sTLR4 and HRP-streptavidin for biotinylated SP-D, respectively, as described under "Experimental Procedures." B, affinity adsorption assay of SP-D. Biotinylated SP-D (800 ng) was incubated with an sTLR4-sMD-2 complex (100 ng) at 37°C for 2 h. The biotinylated SP-D was pulled down by streptavidin-conjugated agarose, and the precipitates were subjected to SDS-PAGE (7.5-15% polyacrylamide gel) under reducing conditions. The Western blot was then performed as described above. After the incubation, the wells were washed and further incubated with HRP-conjugated streptavidin. The binding of SP-D to LPS was detected by measuring the absorbance at 492 nm, as described under "Experimental Procedures." C and D, the biotinylated SP-D (2 g/ml), which had been preincubated with 40 g/ml of anti-human SP-D monoclonal antibody 7C6 or 7A10, or control mouse IgG at 37°C for 1 h, was incubated with Re595 LPS (C) or Rc LPS (D) coated onto microtiter wells at 37°C for 3 h. After the incubation, the wells were washed and further incubated with HRP-conjugated streptavidin. The binding of SP-D to LPS was detected as described above. The data shown are the means Ϯ S.D. from three separate experiments. *, p Ͻ 0.05 and **, p Ͻ 0.01 when compared with control IgG. DECEMBER 19, 2008 • VOLUME 283 • NUMBER 51

SP-D Attenuates LPS-induced TNF-␣ Secretion and NF-B
Activation-SP-A, an SP-D homologue, regulates LPS-induced inflammatory responses in a ligand-specific manner. Smooth LPS is not a ligand for SP-A, but the protein potently inhibits TNF-␣ secretion induced by smooth LPS. Conversely, SP-A binds to rough LPS and enhances its inflammatory response (20,21). For comparison, we next investigated whether SP-D regulates TNF-␣ secretion elicited by smooth and rough LPS. SP-D significantly attenuated O26:B6 LPS-stimulated TNF-␣ secretion from alveolar macrophages in a concentration-de-pendent manner (Fig. 3A). Unlike SP-A, SP-D down-regulated TNF-␣ secretion elicited by Re595 LPS and Rc LPS (Fig. 3, B and C), both of which are ligands for SP-D (Fig. 2). The inhibitory effect of SP-D appeared more potent for antagonizing the stimulation elicited with Rc LPS, than that occurring with Re595 LPS. The IC 50 , for SP-D, was 8.16 g/ml with O26:B6 LPS, 19.51 g/ml with Re595 LPS, and 5.77 g/ml with Rc LPS, respectively. We also performed the experiments with SP-A as a control. SP-A down-regulated O26:B6 LPS-induced TNF-␣ secretion. The results expressed as percentages of TNF-␣ secretion in the absence of SP-A were 23 Ϯ 8% (the mean Ϯ S.D., n ϭ 3), 15 Ϯ 10%, and 15 Ϯ 3% at 1.5, 7.5, and 15 g/ml of SP-A, respectively. In contrast, SP-A failed to attenuate Re595 LPS-stimulated TNF-␣ secretion. The percentages of TNF-␣ secretion were 110 Ϯ 49%, 111 Ϯ 21%, and 120 Ϯ 43% at 1.5, 7.5, and 15 g/ml of SP-A, respectively. These results obtained from control experiments with SP-A are consistent with our previous studies (20,21).
We next examined the effect of SP-D on LPS-elicited NF-B activation in HEK293 cells transfected with TLR4 and MD-2. O26:B6 LPS, Re595 LPS, and Rc LPS induced robust NF-B-dependent luciferase activities through TLR4 and MD-2 (Fig. 4). The presence of SP-D significantly inhibited NF-B activation  stimulated by both smooth (Fig. 4A) and rough serotypes (Fig.  4, B and C) of LPS. The NF-B activities in the presence of SP-A (10 g/ml) were 30% (the mean of two experiments) and 119% of those in the absence of SP-A when the cells were stimulated with O26:B6 LPS and Re595 LPS, respectively. These results are consistent with those obtained by examining TNF-␣ secretion.

SP-D Inhibits Cell Surface Binding of LPS to TLR4/MD-2expressing Cells and Alters MD-2-LPS
Interaction-We next examined whether SP-D alters the binding of LPS to TLR4/ MD-2-expressing HEK293 cells by using Alexa488-labeled E. coli O55:B5 LPS (smooth LPS) and Alexa488-labeled S. minnesota LPS (rough LPS). Flow cytometric analysis revealed significant LPS binding to the cell surface (Fig. 5, A and B, upper, gray shadow) when the cells were incubated with the fluorescent LPS at 4°C. The binding of smooth and rough serotypes of Alexa488-conjugated LPS was inhibited by the presence of SP-D (Fig. 5, A and B, upper, dotted line). SP-D significantly . C, the binding of sMD-2 to LPS. sMD-2 (2 g/ml) was preincubated with or without SP-D (40 g/ml) for 1 h at 37°C, and the mixture was further incubated for 3 h at 37°C with Rc LPS or Re595 LPS coated onto microtiter wells. When O26:B6 LPS or O111:B4 LPS was coated onto the wells, 1 g/ml sMD-2 was preincubated with 20 g/ml SP-D for 1 h at 37°C. After the incubation, the wells were washed and incubated with anti-His tag polyclonal IgG for 1 h, followed by incubation with HRP-conjugated anti-rabbit IgG for 1 h. The binding of sMD-2 to LPS was detected by measuring the absorbance at 492 nm, as described under "Experimental Procedures." The results are expressed as the means Ϯ S.D. from three experiments. **, p Ͻ 0.01 and *, p Ͻ 0.05 when compared with SP-D (Ϫ) in each LPS. , chimera (c), and CRF (d) were subjected to 13% SDS-PAGE under reducing and denaturing conditions and transferred onto polyvinylidene difluoride membranes. The membranes were immunoprobed with anti-human SP-D or anti-human SP-A polyclonal antibody (pAb) (5 g/ml), or with anti-human SP-D monoclonal antibody (mAb) 7C6 or 7A10 (2 g/ml), followed by incubation with HRPconjugated anti-rabbit IgG or HRP-conjugated anti-mouse IgG. The proteins that reacted with the antibodies were visualized as described under "Experimental Procedures." DECEMBER 19, 2008 • VOLUME 283 • NUMBER 51 attenuated the mean fluorescence intensity of cell surface binding of both O55:B5 LPS and S. minnesota LPS (Fig. 5, A and B,  lower).

SP-D Inhibits TLR4/MD-2-mediated LPS Signaling
Because MD-2 is critical for TLR4-mediated LPS signaling (13) and MD-2 directly interacts with LPS (14) and its principal core structural constituent, lipid A (29), we examined whether SP-D affects the binding of sMD-2 to LPS coated onto microtiter wells. SP-D significantly inhibited the binding of sMD-2 to different serotypes of LPS (Fig. 5C). The binding of sMD-2 to Re595 LPS and Rc LPS was reduced to near background levels by SP-D. The binding of sMD-2 to O26:B6 LPS and O111:B4 LPS was reduced to ϳ50% by SP-D. The results clearly demonstrate that SP-D alters MD-2-LPS interaction.
Analysis of Chimeric Protein and CRF-SP-A binds to Re595 LPS and Rc LPS and does not bind to O26:B6 LPS or O111:B4 LPS (20). In this study SP-D also binds to Re595 LPS and Rc LPS but not to O26:B6 LPS, and SP-D suppresses inflammatory responses elicited by both serotypes of LPS, unlike SP-A. We thus constructed a chimera composed of the N terminus plus the collagenous domain of SP-A and the neck plus the CRD of SP-D (A/D chimera) to determine whether the different responses between SP-A and SP-D are consequences of different oligomeric structures. Recombinant SP-D and the A/D chimera (Fig. 6A) were expressed in CHO-K1 cells and purified by affinity chromatography using mannose-Sepharose. CRF consisting of the neck region plus the CRD of SP-D was also purified by gel filtration after collagenase digestion. The main protein bands of SP-D and SP-A migrated at the positions of apparent molecular masses of 45 and 36 kDa, respectively, under reducing conditions when analyzed by electrophoresis (Fig. 6B, lanes a  and b). The A/D chimera and CRF exhibited the expected sizes of apparent molecular masses of 32 and 18 kDa, respectively (Fig. 6B,  lanes c and d). We also analyzed these proteins by Western blotting using anti-SP-A polyclonal antibody, anti-SP-D polyclonal antibody, and anti-SP-D monoclonal antibodies 7C6 and 7A10 whose epitopes are located at the neck region and the CRD of SP-D, respectively. SP-D or SP-A was recognized only by anti-SP-D antibody or anti-SP-A antibody, respectively (Fig. 6C, lanes a and b). The chimera was recognized by all four of the antibodies used (Fig. 6C, lane c) and CRF was detected by anti-SP-D polyclonal antibody and antibodies 7C6 and 7A10 (Fig. 6C, lane d). These results indicate that the chimera and the CRF are correctly expressed and secreted.
We further analyzed these proteins by gel-filtration chromatography and by electron microscopy to examine their oligomeric structure in solution. Gel-filtration analysis with Superose revealed that SP-D eluted at void fraction (Fig. 7A), indicating that SP-D is highly oligomeric (molecular mass of ϳ5 MDa). The elution profile of the A/D chimera showed two main peaks with apparent molecular masses of 220 and 620 kDa. CRF eluted at a fraction corresponding to the size of 68 kDa. The ultrastructures of the proteins were observed by electron microscopy using the rotary shadowing technique (Fig. 7B). In the preparation of SP-D, heterogenous particles, presumably consisting of two, or four trimers, or multimers were observed. One subpopulation of SP-D was a cruciform dodecamer. Another subpopulation includes multimerized oligomer con- The molecular mass standards are; 1, thyroglobulin (669 kDa); 2, ferritin (440 kDa); 3, aldolase (158 kDa); 4, ovalbumin (43 kDa); 5, ribonuclease A (13.7 kDa). B, electron micrographs of rotary-shadowed SP-D, A/D chimera, and CRF. SP-D shows the cruciform structures produced by dodecamers. Another SP-D subpopulation includes higher order multimers. The A/D chimera shows the bouquet-like structure with either four or six globular heads. The CRF preparation is composed of one or two globular heads with a short stalk. SP-A shows the bouquet-like structure mainly with six globular heads.
sisting of SP-D molecules associated at their N terminus. These observations are consistent with those of previous studies (8,30). The preparation of the A/D chimera contained two, or four, or six trimers. Structures of four or six globular heads were seen and appeared to form a bouquet-like arrangement, which is typically observed for SP-A (7). The collagenous domain of SP-A consists of 23 Gly-X-Y repeats with an interruption near the midpoint that imparts a kink to the stem-like regions of the bouquet structure (31). The collagenous domain of SP-D contains 59 Gly-X-Y triplets without interruption (6).

The A/D Chimera and CRF Are Less Potent as Inhibitors for sMD-2 Binding to LPS, and Antagonists for LPS Signaling and TNF-␣ Secretion, by TLR4/MD-2-expressing Cells-
The effects of the A/D chimera and CRF on the binding of sMD-2 to LPS coated onto microtiter wells were examined. SP-D and the chimera significantly attenuated the sMD-2 binding to various serotypes of LPS tested (Fig. 8), but the inhibitory effect of CRF on the sMD-2 binding to LPS was insignificant. Although active, the effect of the chimera as an inhibitor was generally less potent than that of SP-D. The A/D chimera was most different from SP-D at inhibiting sMD-2 binding to Re595 LPS and Rc LPS (Fig. 8, C and D). The activity of the A/D chimera was comparable to SP-D for inhibiting sMD-2 binding to O26:B6 LPS and O111:B4 LPS (Fig. 8, A and B).
LPS-elicited TNF-␣ secretion and NF-B activity were also determined in the presence of SP-D, the chimera, and CRF. Because the oligomerization states are different, the findings are expressed as a function of monomer concentrations for the various recombinant proteins. The A/D chimera was less potent as an inhibitor of LPS-induced TNF-␣ secretion from alveolar macrophages than SP-D (Fig. 9, A-C). CRF was very weak at inhibiting TNF-␣ secretion. When normalized per mole of monomer, the IC 50 was 181.3 Ϯ 50.5 nM (mean Ϯ S.D., n ϭ 3) for SP-D, 239.9 Ϯ 184.4 nM for the chimera, and 347.0 Ϯ 87.4 nM for CRF in O26:B6 LPS-stimulation; 126.8 Ϯ 45.8 nM for SP-D and 341.6 Ϯ 122.9 nM for the chimera in Rc LPSstimulation (p Ͻ 0.05, SP-D versus the chimera); 436.8 Ϯ 226.2 nM for SP-D and 1201.6 Ϯ 75.2 nM for the chimera in Re595 LPS stimulation (p Ͻ 0.01, SP-D versus the chimera). The luciferase assay for LPS-elicited NF-B activation (Fig. 9, D-F) showed that SP-D was the most potent in attenuating LPS signaling and that the chimera and CRF were weak inhibitors in rank order of the chimera Ͼ CRF. These results obtained from cellular experiments are consistent with those obtained from the in vitro sMD-2 binding to LPS, indicating that the inhibitory effect of SP-D is dependent upon the formation of cruciform structure and its multimer.

DISCUSSION
Previous in vivo studies (17)(18)(19)32) have provided compelling evidence that pulmonary collectins modulate pulmonary inflammation caused by microbes and their components. Several mechanisms of collectin-mediated modulation of inflammation have been proposed. Gardai et al. (33) have proposed that, in the absence of microbes, the direct binding of CRD of pulmonary collectins to signal inhibitory regulatory protein ␣ induces the activation of tyrosine phosphatase SHP-1 and blocks the downstream signaling, resulting in inhibiting pulmonary inflammation. However, in the presence of microbes, the binding of the aggregated collagenous tail of SP-A to calreticulin/CD91 stimulates p38 phosphorylation, NF-B activation, and cytokine production. They concluded that SP-A plays dual inflammatory roles by its interaction with signal inhibitory regulatory protein ␣ and calreticulin/CD91. We have previously shown that SP-A inhibits inflammatory responses stimulated with O26:B6 LPS, peptidoglycan, and zymosan by direct interactions with TLR2 and TLR4 (21,34,35) and that SP-A does not attenuate but rather enhances Re595 LPS-induced signaling and TNF-␣ secretion (20). Because Re595 LPS but not O26:B6 LPS, peptidoglycan, or zymosan is a ligand for SP-A, this collectin exhibits dual functions in a ligand-specific manner. However, SP-D appears to exhibit anti-inflammatory effects regardless of ligand or non-ligand, because SP-D binds to Re595 LPS and Rc LPS but not to O26:B6 LPS and inhibits signaling and cytokine secretion stimulated with these different serotypes of LPS by altering LPS binding to its receptor (see Figs. [2][3][4][5]. Gardai et al. (33) did not examine whether SP-D binds to calreticulin/CD91 via its collagenous tail. Because the collagenous tails and the globular domains of SP-A and C1q bind to the calreticulin/CD91 and the microbes, respectively, the orientation of the bouquet-like structure could be important. In contrast, the oligomeric structure of SP-D showing a cruciform is quite dif- ferent from those of SP-A and C1q (see Fig. 7). Thus, it is possible to assume that SP-D might not bind to calreticulin/CD91, which cause pro-inflammatory response. SP-D may contribute to keeping the lung in a relatively uninflamed state by its antiinflammatory functions. These studies support the idea that interactions of pulmonary collectins with cell surface receptors and/or their ligands result in the modulation of pulmonary inflammation.
It is difficult to determine the actual concentrations of pulmonary collectins in vivo, because alveolar hypophase (the epithelial lining fluid of the alveolus) cannot be directly measured. Nevertheless, their concentrations can be estimated based on the recovery of the proteins in the bronchoalveolar lavage fluids and the extrapolated hypophase volume (100 -1000 l/lung) (36,37). The SP-D concentration in the alveolar hypophase can be calculated as ϳ63 g/ml (38) when estimated from the average concentration (0.88 g/ml) in the bronchoalveolar lavage fluids of human healthy volunteers (39). The calculated SP-A concentrations can be in the alveolar hypophase range from 180 g/ml to 1.8 mg/ml (38, 40 -42). Although the concentrations of pulmonary collectins in the hypophase under healthy and diseased states cannot be directly determined, the SP-D concentrations used in this study are within the best estimates of the physiological ranges.
We have previously shown that SP-D directly binds to the extracellular TLR4 domain (22). In this study sMD-2 was immunoprecipitated together with sTLR4 and SP-D, and biotinylated SP-D also co-precipitated with sMD-2 and sTLR4; demonstrating that SP-D can bind an sTLR4-sMD-2 complex. Because the formation of a complex with TLR4 and MD-2 is critical for initiating LPS signaling, it is reasonable to assume that interaction of SP-D with a receptor complex may affect LPS signaling. Consistently, SP-D attenuates cell surface binding of Alexa-labeled LPS to TLR4/MD-2-expressing cells. We also examined the effect of SP-D on the binding of sMD-2 to LPS, because lipid A avidly binds to sMD-2 but not to sTLR4 (29). The results indicate that SP-D significantly decreases the sMD-2 binding to LPS. Taken together, these results support the conclusion that SP-D dampens LPS-induced inflammation by altering LPS-receptor interaction.
This study shows that SP-D binds S. minnesota Re595 LPS and E. coli Rc LPS, but not E. coli O26:B6 LPS. The experiments with anti-SP-D monoclonal antibodies indicate that the neck domain and the CRD are involved in the binding of SP-D to Re595 LPS and Rc LPS, respectively. A previous study (28) has shown by lectin blot analysis that SP-D interacts with Rc and Rd LPS, but not with Re LPS or smooth LPS. Although the differences between studies with Re LPS binding may be due to the methods used, this and our previous studies (20,43) using microtiter well assays indicate that SP-A and SP-D exhibit significant binding to Re595 LPS, but not to O26:B6 LPS. Because SP-A inhibits inflammatory responses elicited by O26:B6 LPS but not by Re595 LPS (20), we tested whether SP-D modulates LPS signaling in a ligandspecific manner, as observed in SP-A. Unlike SP-A, SP-D inhibits inflammatory cell responses induced by rough LPS, an SP-D ligand, as well as smooth LPS, which is not a ligand for SP-D. The precise mechanism by which SP-D inhibits rough LPS-elicited signaling remains to be determined. The profiles of concentration-dependent inhibition by SP-D and the SP-A/SP-D chimera indicate that the difference between these molecules in the inhibitor activities on smooth LPSinduced responses appears small (see Fig. 9, A and D). In contrast, the difference of the inhibition of rough LPS-induced responses is more pronounced (see Fig. 9, B, C, E, and F). Because the chimera also binds to Re595 LPS and Rc LPS (data not shown) as well as SP-A and SP-D, the finding that the inhibitory activity of the chimera on rough LPS-induced responses is less potent than that of SP-D may be a consequence of the orientation of the CRD and/or the length of the collagenous tail, when the proteins interact with rough LPS and an LPS receptor.
The recombinant SP-D prepared in this study is found to be highly oligomeric in solution by gel-filtration analysis and is organized as a cruciform dodecamer, or higher order multimers in electron micrographs. Disruption of interchain disulfide bond formation at the N terminus by Cys 3 Ser mutations at Cys 15 and Cys 20 prevents the covalent oligomerization of trimeric subunits, resulting in loss of functions, such as inducing viral aggregation and enhancing interactions of influenza A virus with neutrophils (44). Expression of wild-type SP-D but not of SP-D C15S, C20S in SP-D-null mice corrects the pulmonary phospholipid accumulation and emphysema phenotype of the SP-D Ϫ/Ϫ mice (45). These studies indicate that the supratrimeric oligomerization of SP-D is required for native protein functions. In this study CRF is ineffective in modulating LPS signaling, although CRF possesses the neck plus the CRD, which are the functional domains for interactions with LPS and TLR4 (22). This clearly demonstrates the importance of the oligomerization of SP-D in immunomodulatory functions of the protein. In addition, the SP-A/SP-D chimera that is assembled in a bouquet-like octadecamer characteristic of SP-A is less potent at inhibiting the LPS-induced inflammatory responses. This suggests that the integrity of the cruciform formation is critical for the expression of the full activities of SP-D in modulating LPS-stimulated inflammatory responses.
In conclusion, SP-D can bind a complex of TLR4 and MD-2 and inhibits TLR4-mediated inflammatory responses caused by both smooth and rough serotypes of LPS. SP-D inhibits the cell surface binding of smooth and rough LPS to TLR4/MD-2-expressing cells and attenuates the MD-2 binding to LPS. SP-D is a more potent inhibitor of LPS-elicited inflammation than SP-A/SP-D chimera and CRF. These results clearly demonstrate that SP-D inhibits LPS-induced inflammatory cell responses by altering LPS-receptor interaction and that the organization of the cruciform dodecamer and its multimer is a critical feature of its function.