Introduction

Top Abstract Introduction Historic Perspectives Molecular Structure Genomic Organization Biosynthesis and Cellular Airspace Distribution of Lung Biochemical Properties In Vitro Interactions with Interactions of Lung Collectins Interactions with Nonmicrobial Interactions of Lung Collectins Altered Leukocyte Function Complexities Relating to the Important Questions and Summary References

Pulmonary Host Defense

The respiratory system is challenged by a constant onslaught of inhaled toxic substances and infectious agents. For this reason, the upper airways and lung have evolved a complex and multilayered system of defense that involves mechanical, reflex, and cellular mechanisms, as well as locally synthesized and systemically derived defense molecules (Table 1). It has been appreciated for many years that non-immune mechanisms must be important components of this defense system, particularly in early development, in the interval between exposure and the development of specific immunity, and in states of impaired immune function (1). It is also appreciated that individuals with similar immune status and exposure history can show marked differences in their susceptibility to pulmonary infection. Recent studies have drawn attention to the probable roles of surfactant-associated proteins, specifically surfactant protein A (SP-A) and surfactant protein D (SP-D), in this innate, natural, and nonclonal defense system (4).

Collectins: A Family of Collagenous Lectins

SP-A and SP-D are members of a family of collagenous carbohydrate binding proteins (collagenous C-type lectins), now commonly known as collectins (2, 7, 8). These include serum mannose binding protein (MBP or MBL) and two bovine serum lectins, conglutinin and CL-43. MBP and conglutinin are recognized acute-phase proteins that have been strongly implicated in various aspects of the systemic response to microbial challenge.

Both SP-A and SP-D show specific interactions with various microorganisms and leukocytes in vitro. Recent data further suggest that lung collectins are a part of a more generalized response of the lung to acute lung injury, and that they modulate local inflammatory/immune responses (9). Collectins may also participate in the recognition or clearance of other complex organic materials, such as pollens (12) and dust mite allergens (13). Thus, acquired or genetically determined differences in pulmonary collectin activity may in part account for the varied susceptibility of individuals to microbial challenge, particularly in the setting of inadequate or impaired specific immunity, or contribute to the pathogenesis of certain immunologically mediated lung disorders, such as allergic asthma and hypersensitivity pneumonitis. Some disorders associated with an increased risk of pneumonia (e.g., diffuse alveolar damage, chronic bronchitis, cystic fibrosis, diabetes mellitus, congestive heart failure) may also be associated with acquired defects in collectin function.

This review focuses on the structure-function correlations of the two known pulmonary collectins. Biochemical properties of potential functional significance, and the interactions of these proteins with specific organisms and host cells, are systematically compared and contrasted. Potential methodologic difficulties are discussed, and questions for guiding future research are presented. For completeness, recent findings relating to the regulation of collectin biosynthesis and secretion and the molecular regulation of collectin gene expression are also concisely reviewed. The structural and functional properties of the serum collectins are thoroughly discussed in several recent reviews (2, 14).

	Historic Perspectives

Top Abstract Introduction Historic Perspectives Molecular Structure Genomic Organization Biosynthesis and Cellular Airspace Distribution of Lung Biochemical Properties In Vitro Interactions with Interactions of Lung Collectins Interactions with Nonmicrobial Interactions of Lung Collectins Altered Leukocyte Function Complexities Relating to the Important Questions and Summary References

Surfactant Protein A

SP-A was first identified by King and colleagues in 1972 (17, 18). In the late 1980s, protein, cDNA, and genomic sequencing demonstrated the presence of collagenous sequences and a carboxyl-terminal C-type lectin motif (19- 22), predicting the subsequent demonstration of a triple helical collagen domain (18) and carbohydrate binding activity (23, 24).

The best-characterized ligands for SP-A are lipid in nature. Purified SP-A binds to specific surfactant-associated phospholipids in vitro, primarily dipalmitoylphosphatidylcholine (DPPC) (25). These and numerous other observations have until recently suggested that SP-A's major function in vivo is to regulate the production and/or metabolism of the airspace lining material.

Novel studies by Tenner and coworkers published in 1989 were the first to suggest a more diverse range of biologic functions (28). Purified SP-A was shown to enhance FcR and CR1-mediated phagocytosis by monocytes/ macrophages, indicating modulation of leukocyte function. Subsequent studies demonstrated a variety of interactions with microorganisms in vitro and effects on leukocyte function and antimicrobial activity (6). In fact, recent studies suggest that the primary function of SP-A relates to lung defense. The most compelling evidence to date is that otherwise healthy transgenic mice lacking a functional SP-A gene, SP-A (/), do not demonstrate obvious abnormalities in normal respiratory function or surfactant lipid metabolism (29, 30). Furthermore, these animals demonstrate increased bacterial proliferation, more intense lung inflammation, and an increased incidence of splenic dissemination following intratracheal inoculation with the Group B streptococcus, a major pulmonary pathogen in the neonatal period (31). These studies also suggested defective clearance of Staphylococcus aureus and Pseudomonas aeruginosa.

Surfactant Protein D

SP-D was originally characterized in 1988 as one of several collagenous glycoproteins (CP4) secreted in cultures of freshly isolated rat type II cells (32). Subsequent studies demonstrated the presence of SP-D in bronchoalveolar lavage (BAL), and in association with crude surfactant (33). In 1992 Kuan and coworkers demonstrated lectin-mediated binding of SP-D to gram-negative bacteria and resultant bacterial aggregation, first suggesting possible roles for SP-D in pulmonary host defense (34).

	Molecular Structure

Top Abstract Introduction Historic Perspectives Molecular Structure Genomic Organization Biosynthesis and Cellular Airspace Distribution of Lung Biochemical Properties In Vitro Interactions with Interactions of Lung Collectins Interactions with Nonmicrobial Interactions of Lung Collectins Altered Leukocyte Function Complexities Relating to the Important Questions and Summary References

Domain Structure

The lung and serum collectins are assembled as oligomers of trimeric subunits. Each subunit consists of four major domains: a short cysteine-containing NH₂-terminal cross-linking domain (N); a triple helical collagen domain of variable length; a trimeric coiled-coil linking domain (L; sometimes referred to as the neck); and a carboxyl-terminal, C-type lectin carbohydrate recognition domain (CRD) (Figures 1 and 2; Table 1). Interactions between the amino-terminal domains of SP-D subunits have been shown to be stabilized by interchain disulfide bonds (35, 36), and similar mechanisms stabilize the oligomerization of SP-A (37) and most other collectins.

View larger version (17K): [in this window] [in a new window]   Figure 1.   Domain structures of SP-A and SP-D. Note the similar size and distribution of the amino-terminal, L-peptide, and carbohydrate recognition domains (CRD). Differences in monomer size result from differences in the length of the collagen domain (diagonal lines). The Asn-linked oligosaccharide in SP-D is located in the collagen domain, whereas the Asn-linked sugar in human SP-A is located within the CRD.

View larger version (16K): [in this window] [in a new window]   Figure 2.   Molecular organization of pulmonary collectins. The predominant molecular forms of SP-A and SP-D are compared assuming maximal spatial separation of the CRDs. The Asn-linked oligosaccharides of SP-D are probably less accessible to lectins expressed by microorganisms than the sugars of SP-A.

SP-A structure. SP-A (26-35 kD, reduced) is predominantly assembled as octadecamers consisting of six trimeric subunits (18 chains) with relatively short collagen domains (Figure 2). In these respects, SP-A is very similar to serum MBP. However, SP-A lacks hydroxylysine, and shows no O-linked glycosylation of hydroxylysyl residues in the collagen domain. Although rat SP-A contains two Asn-linked oligosaccharides, one in the amino-terminal peptide domain and one in the CRD, human SP-A contains a single Asn-linked oligosaccharide within the CRD. Alternative proteolytic processing of the amino-terminal peptide of SP-A has been reported to influence glycosylation and the formation of disulfide cross-linked oligomers (38). Human SP-A molecules can be assembled as homotrimers or as heterotrimers derived from two genetically different chain types (39). The relative proportions of homo- and heterotrimers accumulating in the lung have not been established, and it is not yet known whether the two forms are synthesized by the same cell type or accumulate at the same site in vivo.

SP-D structure. SP-D (43 kD, reduced) is predominantly assembled as dodecamers consisting of four homotrimeric subunits (12 chains) with relatively long triple helical arms (35, 40) (Figure 2). With respect to these features, SP-D is most similar to conglutinin (8), which has no other known human homolog. However, SP-D is distinguished from conglutinin by an uninterrupted, cysteine-free collagen domain that contains the single site of Asn-linked glycosylation (Asn₇₀). By contrast with SP-A, the collagen domain of SP-D contains hydroxylysine and hydroxylysyl-glycosides (33). Although natural and recombinant rat SP-Ds are almost exclusively assembled as dodecamers, preparations of natural human and bovine SP-D can include a high proportion of trimers (14, 42). It is unclear whether the accumulation of these forms reflects differences in the efficiency of intracellular multimerization or in the stability of secreted dodecamers.

Higher-order Oligomerization of Lung Collectins

SP-A multimers. In the setting of alveolar proteinosis, SP-A octadecamers can self-associate to form multimolecular complexes (45, 46). Under some circumstances the aggregated molecules may become cross-linked through disulfide interchange and the formation of other covalent bonds (47). The functional significance of SP-A multimerization is uncertain. However, differences have been observed between proteinosis SP-A and less highly multimerized preparations of natural or recombinant proteins in several biologic assays. For example, the multimerized form shows lower affinity binding to type II cells and is less potent as an inhibitor of lipid secretion (45). On the other hand, proteinosis SP-A is more potent than recombinant human SP-As in enhancing bacterial phagocytosis (50). Likewise, proteinosis SP-A is more effective than natural or recombinant SP-A in enhancing the adherence and phagocytosis of mycobacteria by macrophages (51).

SP-D multimers. SP-D dodecamers can self-associate at their amino-termini to form much more highly ordered, stellate multimers with peripheral arrays of trimeric CRDs (35, 44, 52) (Figure 3). Natural SP-D from human alveolar proteinosis and bovine lavage, and recombinant human SP-D contain a high proportion of these multimers with up to eight (or possibly more) SP-D dodecamers. The multimers are not dissociated by ethylenediamenetetraacetic acid (EDTA) or competing sugars, and are cross-linked by disulfide and nondisulfide bonds. SP-D multimers show higher apparent binding affinity to a variety of ligands and are considerably more potent on a molar or weight basis in mediating microbial aggregation and aggregation-dependent interactions with leukocytes (44, 53).

View larger version (22K): [in this window] [in a new window]   Figure 3.   Alternative molecular forms of SP-D. Monomeric CRDs, trimeric CRDs, the single-arm RrSP-Dser15/20 mutant, and SP-D multimers are compared with natural or recombinant rat SP-D (RrSP-D) dodecamers. Monomeric CRDs show very weak binding affinity to various ligands. The minimum structure required for high-affinity binding is the trimeric CRD. Trimeric CRDs and single-arm mutant forms of SP-D can inhibit bridging interactions mediated by SP-D dodecamers or multimers. A similar situation exists for SP-A; however, SP-A multimers are less highly ordered.

	Genomic Organization

Top Abstract Introduction Historic Perspectives Molecular Structure Genomic Organization Biosynthesis and Cellular Airspace Distribution of Lung Biochemical Properties In Vitro Interactions with Interactions of Lung Collectins Interactions with Nonmicrobial Interactions of Lung Collectins Altered Leukocyte Function Complexities Relating to the Important Questions and Summary References

The human SP-A and SP-D genes have been localized to the region of 10q22.2-23.1 (54). SP-A is encoded by two genes, SP-A1 and SP-A2 (20, 59) with several alleles at each locus. Most species (e.g., rat, mouse, rabbit, dog) have only a single gene. Human SP-D is also encoded by a single gene (54). However, protein, cDNA, and genomic sequencing together suggest the existence of a number of SP-D alleles, some of which are characterized by amino-acid substitutions in the coding region (43, 54). Each SP-A gene encodes untranslated exons at the 5'-end, and all but the first of these are subject to gene-specific alternative splicing (59, 61). The two human genes differ by approximately 21 nucleotides within the coding region but have more numerous base substitutions in the upstream and downstream untranslated regions (60). These changes result in only a few amino-acid substitutions, including the substitution of cys for arg at residue 85 within the collagen sequence, which has been implicated in the cross-linking of heterotrimers (63). The SP-D gene encodes at least one, and probably two, untranslated exons at the 5'-end of the gene, similar to SP-A (54).

	Biosynthesis and Cellular Metabolism of SP-A and SP-D

Top Abstract Introduction Historic Perspectives Molecular Structure Genomic Organization Biosynthesis and Cellular Airspace Distribution of Lung Biochemical Properties In Vitro Interactions with Interactions of Lung Collectins Interactions with Nonmicrobial Interactions of Lung Collectins Altered Leukocyte Function Complexities Relating to the Important Questions and Summary References

SP-A and SP-D are both synthesized and secreted by alveolar type II and nonciliated bronchiolar epithelial cells (64). However, there is evidence for considerable cell-to-cell variation in the relative production or accumulation of SP-A and SP-D by these cells (69). Furthermore, SP-A and SP-D show differences in developmental expression and different patterns of regulation by glucocorticoids, cytokines, and other factors in rat and human lung explants.

Developmental Regulation

In the rat and human lung the accumulation of SP-A (relative to total protein or DNA) increases dramatically during late gestation, levels off or slightly declines in the early postnatal period, and then increases to reach maximal levels in the adult (70). The expression of the two human SP-A genes is differentially regulated during development with a predominance of SP-A1 transcripts (65%) in mid-gestation fetal lung, while the majority of the transcripts in the adult lung are derived from the SP-A2 gene (62). The administration of dexamethasone in utero accelerates the appearance of SP-A in rat lung (73).

The accumulation of SP-D in the rat lung also increases abruptly in late gestation, slightly later than SP-A (65, 76, 77). Unlike SP-A, SP-D mRNA and protein levels continue to increase during the early postnatal period, eventually reaching their highest levels in the adult lung. As for SP-A, the administration of dexamethasone in utero accelerates the appearance of SP-D-producing cells and increases the cellular level of SP-D mRNA (76, 78, 79). In the human, SP-D mRNA is first detected at low levels (4 to 13% of adult) in the second trimester and message levels rise steadily during late fetal lung and postnatal lung development (80).

Regulation of Collectin Production in Explant Culture

Human fetal lung explants demonstrated stimulatory effects of cyclic adenosine monophosphate (cAMP) analogs, -interferon, and epidermal growth factor; and inhibitory effects of indomethacin, insulin, phorbol myristate acetate (PMA), tumor necrosis factor alpha (TNF-), tumor growth factor beta, and lipopolysaccharides (LPS) (70, 80). There are also concentration-dependent effects of glucocorticoids on SP-A mRNA levels in human lung explants (80). The effects of hormones and various cytokines appear to involve both transcriptional and/or post-transcriptional regulatory mechanisms. In human fetal lung organ culture, SP-A2 is preferentially upregulated by cAMP and inhibited by glucocorticoids, whereas SP-A1 appears to be constitutively expressed (62).

Sequences upstream from the start site of SP-D transcription include a conserved canonical AP-1 element, several AP-1 and CRE-like sequences, as well as sequences similar to those identified in conglutinin and other acute phase proteins (NF-IL6, PEA3, APF-1). Studies using human fetal lung explants have confirmed the upregulatory effects of glucocorticoids (80), but show no evidence of modulation by several agents known to alter SP-A expression, including -interferon, PMA, LPS, and TNF- (80).

Cellular Pathway of Assembly and Secretion

Collectin biosynthesis and assembly are complex processes. Recent studies of recombinant SP-D assembly by CHO-K1 cells suggest that folding of the CRD, trimerization of monomers, triple helix formation, the amino-terminal association of trimeric subunits, and the formation of interchain disulfide cross-links occur in the rough endoplasmic reticulum, and that oligosaccharide maturation occurs in the Golgi immediately prior to secretion (83). Although the pathway of SP-A secretion has not been fully elucidated, the maturation of the N-linked sugars in SP-A also occurs late and shortly prior to secretion (84). Investigators initially assumed that SP-A was subject to regulated secretion in association with lamellar bodies (LB). However, in human lung explants, all but a small fraction of the newly synthesized protein is secreted via a constitutive pathway and independent of LB (85, 86). It is unclear to what extent bronchiolar cells contribute to the non-LB- associated fraction, and whether there are structural differences in the molecules targeted to these organelles.

Collectin Degradation

Little is currently known about the physiologic turnover of lung collectins or their susceptibility to degradation following injury. Instillation of SP-A into rabbit lungs has shown that the half-life of the solubilized protein is approximately 4.5-6.5 h (87, 88). SP-A can also be rapidly internalized and degraded by macrophages in vitro (89), and accumulates in alveolar macrophages in vivo (90). SP-D is found in both endosomal vesicles and lysosomal granules of alveolar macrophages, indicating that these cells can also internalize SP-D (4).

Modulation of Collectin Production and Accumulation In Vivo

Immunohistochemical and in situ hybridization studies strongly suggest that the production of these molecules is increased in association with acute injury and epithelial activation (69, 93). Acute hyperoxia in rats is associated with differential, time-dependent alterations in expression of SP-A and SP-D by type II and Clara cells (97, 98). Thus, regional concentrations of these molecules may be influenced by the specific cellular responses to various forms of injury.

The production and accumulation of both SP-A and SP-D is rapidly increased following intratracheal instillation of LPS (99). Because the mRNAs for the lung collectins are increased within several hours to a few days following injury, McIntosh and coworkers suggested that they are pulmonary acute-phase proteins, similar to liver-derived MBL and conglutinin, which are systemic acute-phase reactants (99). The mechanism of LPS-mediated increases in lung collectin production is unknown. However, as previously noted, SP-A and SP-D mRNA levels are decreased or unchanged, respectively, by LPS in human lung explants. This suggests the involvement of other inflammatory mediators or cytokines in vivo. As indicated previously, SP-A production is upregulated by -interferon in lung explants (81).

	Airspace Distribution of Lung Collectins

Top Abstract Introduction Historic Perspectives Molecular Structure Genomic Organization Biosynthesis and Cellular Airspace Distribution of Lung Biochemical Properties In Vitro Interactions with Interactions of Lung Collectins Interactions with Nonmicrobial Interactions of Lung Collectins Altered Leukocyte Function Complexities Relating to the Important Questions and Summary References

Very little SP-A is identified in BAL supernatants following high-speed centrifugation. Although detergents or organic solvents have usually been employed for the extraction of SP-A from the surfactant pellet, EDTA has recently been shown to solubilize the majority of the protein (100), suggesting that the conformation of the C-type lectin domain is important for its association with surfactant lipid. Immuno-electron microscopic studies have shown that SP-A is associated with lipid-rich components, particularly tubular myelin (101, 102). Tubular myelin formation is also nearly absent from the lungs of SP-A (/) transgenic mice (29). The molecular orientation of SP-A in relation to the surfactant layer in vivo is not known.

By contrast, the majority of the total immunoreactive SP-D remains in the BAL supernatant following high-speed centrifugation (50 to 90%, depending on the species) (33, 103). Immunologic studies have shown that the insoluble fraction is associated with amorphous granular material, and that the immunoreactive material can be efficiently solubilized with EDTA or specific saccharides (93). No significant amounts of SP-D are associated with surfactant that has been collected or washed in the presence of chelators or purified by sucrose density centrifugation.

Early studies reported that there was much less total SP-D than SP-A in the airspace of rats and in human alveolar proteinosis lavage. However, recent comparative assays by Honda and coworkers gave 3.1 ± 0.4 µg/ml for SP-A and 1.3 ± 0.2 µg/ml for lavage from healthy nonsmokers (104). These studies employed a standard saline lavage procedure, a very brief 250 × g centrifugation to remove cells, and well-characterized immunoassays for both proteins.

	Biochemical Properties

Top Abstract Introduction Historic Perspectives Molecular Structure Genomic Organization Biosynthesis and Cellular Airspace Distribution of Lung Biochemical Properties In Vitro Interactions with Interactions of Lung Collectins Interactions with Nonmicrobial Interactions of Lung Collectins Altered Leukocyte Function Complexities Relating to the Important Questions and Summary References

As previously suggested, there is evidence that the number and spatial distribution of CRDs can influence the binding activities of collagenous lectins. Thus, solubility or other chemical parameters that influence protein aggregation or multimerization could be important determinants of collectin function or influence binding activity in vitro. Biologic activity may also be influenced by changes in the conformation of the CRDs that influence calcium or ligand binding.

Charge Properties

Both SP-A and SP-D are secreted as several distinct isoforms that can be resolved by 2-D isoelectric focusing/sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) (32, 105). The SP-A isoforms are relatively acidic (pI 4.2-5), whereas the SP-D isoforms are basic (pI 5-9) (Table 1). Because the predicted pI of the CRD for both proteins is in the range of 4 to 5, differences in net charge between SP-A and SP-D predominantly reflect differences in the collagen domain, which is very basic (pI ~ 10) in SP-D but weakly acidic in SP-A.

At least part of the heterogeneity of human SP-A results from slight charge differences between the two SP-A gene products (39). However, the remainder of the heterogeneity for both proteins can be attributed to variations in terminal sialylation (105), heterogeneity in the charge of the collagen domain (as observed for other collagenous proteins), and perhaps acetylation (106). Resolution of newly synthesized or natural rat SP-D by diethylaminoethyl chromatography under nondenaturing conditions also partially resolves species of different monomer size attributable to differences in terminal sialylation (32). This observation suggests that, at least for SP-D, similarly modified monomers are preferentially associated to form multimers.

The functional significance of this heterogeneity and differences in charge is not known. The sialylated sugars of SP-A, which are located on the CRD, are able to interact with carbohydrate binding proteins associated with certain viruses, and may also contribute to the calcium- dependent self-aggregation of SP-A (44, 107, 108). However, the N-linked oligosaccharides of SP-D dodecamers do not appear to be similarly accessible, and mutant proteins lacking the consensus for N-linked glycosylation are indistinguishable from wild-type SP-D with respect to the aggregation and neutralization of influenza A virus (IAV) and the agglutination of gram-negative bacteria (K. Hartshorn, manuscript submitted).

Solubility

Purified SP-A and SP-D show marked differences in solubility at neutral pH that probably reflect differences in charge. SP-A tends to precipitate at physiologic ionic strength (even in the absence of calcium) and is usually stored in low-salt (e.g., 5 mM Tris) buffers. By contrast, SP-D is fully soluble in saline at neutral pH and precipitates in low-salt buffers. This is consistent with the supposition that the vast majority of SP-A is insoluble and associated with surfactant lipids in vivo, whereas SP-D is preferentially distributed within the "aqueous hypophase." Bound surfactant lipids, or partial denaturation of the protein by organic solvents used to remove these lipids, may influence the solubility of SP-A in vitro. For example, butanol-extracted proteinosis SP-A has been reported to show a greater propensity for self-aggregation than normal SP-A solubilized with EDTA (23).

Binding of Divalent Cations

Calcium binding to the CRD. The saccharide binding activities of SP-A and SP-D are abolished with 2 mM EDTA and can be restored following readdition of excess calcium (24, 109). Biophysical studies of SP-A have identified two calcium binding sites: one low affinity (10³ M) and one high affinity (10⁵ M). The high-affinity binding site has been localized to the collagenase resistant fragment (L+CRD) (110). Calcium binding alters the conformation of this domain as detected by intrinsic fluorescence assays, increased resistance to proteolytic degradation, and alterations in antibody binding. Although similar studies have not been performed for SP-D, neoglycoprotein binding assays are consistent with the presence of at least one high-affinity binding site (10⁶ M) (109). Finally, sequence studies of the CRDs of both proteins and comparison with MBL are consistent with two calcium coordination sites per monomer (111, 112). Together these observations suggest that the calcium binding sites in the CRD are fully occupied at the estimated alveolar fluid calcium concentration of 1.5 mM (110) and help maintain a conformation suitable for lectin activity. Interestingly, other divalent cations show disparate effects on the saccharide binding activities of SP-D and SP-A, presumably through subtle effects on the conformation of the carbohydrate binding site. Although magnesium, strontium, manganese, and barium inhibit the binding of SP-D to -glucosyl sugars, the latter three divalent cations can substitute for calcium in the binding of SP-A to mannose (24).

Calcium-dependent precipitation. SP-A can undergo calcium-dependent self-aggregation (108, 113). Precipitation readily occurs with low calcium concentrations with NaCl concentrations in excess of 20 mM in the presence of non-ionic detergents (24). Because the self-association of human proteinosis SP-A and SP-A-mediated phospholipid vesicle aggregation were blocked following glycosidase digestion, Haagsman and coworkers concluded that the CRD binds to the N-linked oligosaccharide within the CRD of another SP-A molecule, thereby leading to aggregation (113). This conclusion was subsequently questioned because deletion of the consensus for N-linked glycosylation in recombinant rat SP-A synthesized by insect cells did not prevent calcium-dependent lipid aggregation (114). However, this apparent discrepancy could reflect differences in assay conditions and the state of protein oligomerization. For example, the former studies were performed at low ionic strength, whereas the latter experiments were performed in the presence of 0.15 M NaCl.

Effects of pH

Many C-type lectins show reversible loss of binding at mildly acidic pH. This property is critical to the function of endocytic receptors that must release their ligands following internalization. It has been suggested that protonation of carboxylate groups involved in calcium binding decreases calcium binding and is necessary (although not always sufficient) for carbohydrate ligand release. Interestingly, both SP-A and SP-D show relative preservation of saccharide binding activity at pH 5 (24, 109).

Intrachain Disulfide Bonds within the CRD

Intrachain bonds stabilize the conformation of the CRD. Peptide mapping studies of SP-A have shown that the second and third cysteines (cys204-cys218), and the first and fourth cysteines (cys135-cys226), are cross-linked (110), consistent with the crystal structure of serum mannose binding protein. Analogous cross-linking of the cysteines of SP-D (cys261-cys353 and cys331-cys345) is predicted. A variety of observations suggest the intrachain bonds are required for calcium-dependent saccharide binding and confer resistance to thermal denaturation or proteolytic degradation. For example, natural SP-D retains lectin activity in solid-phase neoglycoprotein binding assays following boiling for 10 min at neutral pH. However, activity is irreversibly lost when the protein is heated briefly in the presence of low concentrations of dithiothreitol (DTT).

Interchain Disulfide Bonds within the Collagen or Amino-terminal Domains

Interchain disulfide bonds are required for the formation of stable trimeric subunits and higher-order oligomers. Biochemical studies are consistent with two classes of interchain disulfide bonds, intersubunit (i.e., between trimers) and intrasubunit (within trimers) (35). The latter are comparatively stable and can form and re-form spontaneously following reduction (37). Thus, low concentrations of sulfhydryl reducing agents can liberate trimeric subunits from collectin molecules, sometimes with partial preservation of intratrimeric disulfide bonds, depending on the concentration of reducing agent and the incubation temperature (35, 37). The interchain bonds (inter- and/or intrasubunit) are also required for normal thermal stability of the collagen triple helix. Complete reduction of SP-A with DTT decreases the melting temperature of the short collagen helix from 41.5 to 28.5°C (37). Elimination of the amino-terminal cysteines of rat SP-D shows a much more modest decrease in melting temperature (36), presumably reflecting an intrinsically greater stability of the long and uninterrupted SP-D helix.

Susceptibility to Proteolytic Degradation In Vitro

SP-A can be degraded in vitro by human leukocyte elastase (115, 116). However, in the presence of physiologic calcium concentrations purified SP-D is not degraded by elastase or a variety of mammalian neutral proteinases (36). Apparently, disulfide cross-linking of the amino-terminal domain, tight folding of the collagen triple helix and coiled-coil domain, and disulfide-stabilized folding of the CRD together exclude proteases from potential internal cleavage sites in natural SP-D at 37°C. Multimerization of collectin subunits may also interfere with proteolytic degradation within the amino-terminal peptide and collagen domains. Oxidant treatments have been shown to enhance the susceptibility of SP-A to cleavage by elastase or trypsin (115- 117), suggesting that lung injury might result in protein modifications that could enhance the susceptibility of collectins to degradation in vivo.

	In Vitro Interactions with Carbohydrate and Lipid Ligands

Top Abstract Introduction Historic Perspectives Molecular Structure Genomic Organization Biosynthesis and Cellular Airspace Distribution of Lung Biochemical Properties In Vitro Interactions with Interactions of Lung Collectins Interactions with Nonmicrobial Interactions of Lung Collectins Altered Leukocyte Function Complexities Relating to the Important Questions and Summary References

Structural Requirement for Ligand Binding

For all of the collectins, the major requirements for specific carbohydrate binding include the conserved C-type lectin motif in the context of a tertiary structure stabilized by calcium binding and intrachain disulfide cross-linking, and the formation of a trimeric molecule with an appropriate spatial distribution of the constituent CRDs.

The C-type lectin domain. Protein and cDNA sequencing studies have shown that the primary sequence of the carboxy-terminal domains of SP-A and SP-D are homologous (~ 40%) and that they both contain characteristic elements of the mannose-type C-type lectin motif (19, 21, 43, 118, 119). Various biochemical and molecular studies have definitively established that these domains are primarily responsible for the carbohydrate binding activity.

Trimeric clusters of CRDs. The linking peptide in conjunction with the CRD domain (L+CRD) can form trimers, and interactions between hydrophobic sequences on L and the CRD determine the spatial distribution of the three CRDs, thereby generating a single, trimeric, high-affinity ligand binding site (122). Approximately 100- to 200-fold higher concentrations of saccharide competitor are required to block the binding of trimeric CRDs (10-20 mM), as compared to monomers (0.1 mM), to neoglycoprotein ligands in solid-phase binding assays. Thus, trimeric clustering of CRD permits collectin binding to complex ligands in the physiologic range of free "sugar" concentrations. At least in the case of MBP, the three saccharide binding sites form a relatively planar array perpendicular to the axis of the trimeric linking peptide. Thus, high-affinity binding usually requires the simultaneous occupancy of two to three saccharide binding sites within a single trimeric subunit in apposition to a surface with a comparable spatial distribution of saccharide ligands.

Spatial distribution of trimeric CRDs. As implied in the previous paragraph, the assembly of collectin monomers to form trimeric clusters of C-type CRDs is necessary and sufficient for high-affinity binding to specific saccharide and lipid ligands. However, the capacity for bridging interactions between spatially separated ligands depends on an appropriate oligomerization of trimeric subunits, resulting in a characteristic spatial distribution of the trimeric CRDs. Thus, trimeric CRDs appear to be functionally univalent with regard to their capacity to participate in bridging interactions between large particulate ligands. The concept of a functionally univalent trimer is consistent with the inability of a single-arm mutant, RrSP-Dser15/20, and bacterially expressed trimeric L+CRDs to cause significant bacterial aggregation and viral precipitation (36). In fact, trimeric CRDs and single-arm forms of the protein can function as competitive inhibitors of SP-D-mediated microbial aggregation.

8

11

In Vitro Binding Activities

SP-A and SP-D show calcium-dependent and saccharide-inhibitable interactions with a wide variety of carbohydrates or carbohydrate-containing ligands, as well as specific phospho- and glycolipids (Tables 2 and 3). The ligands include various neoglycoproteins or saccharide-substituted affinity matrices, purified microbial glycoconjugates, and whole organisms. Given that these studies have employed many different assay systems, from solid-phase or solution binding assays to light-scattering assays of aggregation or precipitation, it is not surprising that there are some apparent discrepancies in the literature. Nevertheless, important generalizations can be made based on the available data, and the remaining discrepancies only serve to emphasize the importance of exercising caution when extrapolating from assay systems to the possible situation in vivo.

Some of the interactions of collectins with lipids involve interactions with saccharide components of complex lipids (e.g., the binding of SP-D to glucosyl-ceramide, or SP-A with galactosyl-ceramide). However, it is likely that interactions with lipid-containing structures in the pulmonary airspace, or hydrophobic interactions with nonpolar moieties expressed on the surface of various microorganisms (e.g., SP-A binding to lipid A), constitute distinct biologic activities of these proteins in vivo. For this reason, carbohydrate and lipid binding activities are discussed separately below.

Carbohydrate Binding

SP-A. Purified human SP-A binds specifically to carbohydrates as assessed by saccharide competition in solid phase-binding assays using adsorbed mannan as the ligand (23) (Table 2). The order of preference was approximately N-acetylmannosamine > L-fucose > maltose > glucose > mannose with no inhibition by galactose, N-acetylglucosamine, and N-acetylgalactosamine. These results were independent of the source or method of extraction of SP-A, and generally consistent with earlier studies that examined binding to various saccharide-substituted supports (24). However, they are in obvious disagreement with another study that characterized the binding of iodinated butanol-extracted human and dog SP-A to various neoglycoproteins, glycolipids, and neoglycolipids (124). Notably, the latter studies demonstrated a preferential recognition of galactose and an inability to compete for binding with free monosaccharides. The basis for these discrepancies is unclear, but they almost certainly reflect differences in the assay systems and possibly the physical state of the isolated SP-A.

SP-D. Purified natural rat, human, and bovine SP-D, as well as recombinant rat and human molecules, preferentially recognize the -anomeric configuration of nonreducing glucopyranosides (109). The order of preference of human SP-D in solid-phase competition assays using maltosyl-bovine serum albumin (BSA) as the ligand is approximately: maltose (inositol) > glucose, mannose, fucose > galactose, lactose, glucosamine > N-acetylglucosamine (Table 2). This binding specificity is consistent with known interactions of SP-D with the glucose-containing core oligosaccharides of LPS, the mannose-rich N-linked oligosaccharides of the hemagglutinin of influenza A, and the gpA of Pneumocystis carinii (see below).

Summary and important caveats. As shown in Table 2, the carbohydrate selectivities of the two proteins are generally consistent with their subclassification as mannose-type C-type lectins. However, somewhat different profiles in saccharide selectivity have been reported by different investigators for human SP-A, and there are subtle differences between rat and human SP-D. There are notable differences in primary structure of the SP-A and SP-D carboxy-terminal domains that include differences in conserved elements of the CRD which could account for differences in ligand carbohydrate binding. On the other hand, differences in protein solubility and in the valency of trimeric CRDs complicates the comparison of binding affinity and other parameters. Although there is evidence for a small soluble fraction of SP-A, the physiologic relevance of solution phase assays for studies of SP-A function remains uncertain.

Lipid Binding

SP-A. As indicated in the previous section, SP-A specifically binds to DPPC (127, 128) and aggregates surfactant phospholipids in the presence of calcium (108, 113, 129). SP-A also binds to several glycolipids and neutral glycosphingolipids including galactosyl-ceramide, lactosyl-ceramide, and asialo-G_M2, but not to glucosyl-ceramide (124, 130, 131). These interactions appear to involve recognition of both the ceramide and saccharide moieties (124, 131).

SP-D. Purified SP-D shows high-affinity binding to phosphatidylinositol (PI) resolved on thin-layer chromatography (TLC) plates or presented in liposomes (135, 137, 138). PI is also the major surfactant-associated ligand of SP-D. In addition, SP-D binds to glucosyl-ceramide when displayed on TLC plates, but not to galactosyl-ceramide, asialo-G_M2, or other major pulmonary glycolipids (103). The interactions of SP-D with PI and glucosyl-ceramide are calcium-dependent and inhibited by competing sugars, consistent with lectin-dependent binding. PI recognition may predominantly involve the inositol moiety. In this regard, myo-inositol is an efficient competitor of SP-D binding to maltosyl-BSA, and inositol shows homology to -D-glucose in its distribution of hydroxyl groups (138).

Summary. Phospholipid binding by the lung collectins involves interactions with the CRD but also requires the L peptide. It is not known whether the L peptide directly participates in binding in the intact protein, or whether hydrophobic interactions between the L peptide and CRD are required to maintain a specific conformation or spatial distribution of the CRDs. Differences in lipid binding preferences of SP-A and SP-D probably reflect differences in the structure of the CRD, as well as differences in the L peptide, which include a highly hydrophobic sequence following the amphipathic -helix in SP-A (118). Phospholipid aggregation requires the L+CRD, as well as the oligomerization of trimeric subunits.

	Interactions of Lung Collectins with Microbial Ligands

Top Abstract Introduction Historic Perspectives Molecular Structure Genomic Organization Biosynthesis and Cellular Airspace Distribution of Lung Biochemical Properties In Vitro Interactions with Interactions of Lung Collectins Interactions with Nonmicrobial Interactions of Lung Collectins Altered Leukocyte Function Complexities Relating to the Important Questions and Summary References

The lining material of the alveoli and distal airways is ideally positioned to participate in the neutralization and clearance of inhaled microorganisms. Because the large serum collectins are presumed to be present at very low concentrations in the extravascular space in uninjured tissues, SP-A and SP-D probably constitute the major collectin defenses of the lung. Consistent with this hypothesis, lung collectins have been shown to interact with a wide range of microorganisms in vitro (Table 3), and dissemination of at least one organism is enhanced in SP-A (/) mice. In some cases specific microbial ligands have been defined.

Viral Glycoconjugates

The interactions of lung collectins with influenza A virus (IAV) have been extensively characterized, and provide a reasonable model system to examine structure-function relationships. IAV attaches to and infects cells by binding through its hemagglutinin to sialic acid-bearing components on the cell surface, while the neuraminidase is involved in viral production and perhaps inactivation of sialylated host proteins. The collectins are potent inhibitors of hemagglutinin (HA)-mediated agglutination, and may also inhibit neuraminidase activity. For example, the levels of SP-D in lavage fluids are sufficient to account for a significant fraction of the total HA inhibitory activity of the fluids as demonstrated by prior adsorption of SP-D using maltosyl-agarose affinity chromatography (140).

Binding of SP-D to IAV. HA inhibition by SP-D involves the binding of SP-D through its CRD to glycoconjugates expressed near the sialic acid binding site on the hemagglutinin (or neuraminidase) of specific strains of IAV (53). HA inhibition assay is inhibited by EDTA or maltose (140), and binding to purified HA is blocked by prior glycosidase digestion. Notably, SP-D has negligible activity against the PR-8 strain of IAV, which lacks carbohydrate attachments on the head region of its HA molecule (141). As illustrated in Figure 4, higher degrees of valency or multimerization among the various SP-D preparations are associated with increased HA inhibitory activity. At least with some strains of IAV there is also binding to glycoconjugates associated with the viral neuraminidase, and it has been suggested that collectin binding to the neuraminidase can sterically interfere with HA activity (142, 143). The possibility that collectins can inhibit the neuraminidase requires further investigation.

View larger version (17K): [in this window] [in a new window]   Figure 4.   Effects of various molecular forms of SP-D on influenza A virus and its interactions with neutrophils. Trimers of CRDs and RrSP-Dser15/20s can bind to the virus as evidenced by inhibition of hemagglutination. However, these forms do not mediate viral aggregation or several activities dependent on viral aggregation, including viral binding to neutrophils, augmentation of the respiratory burst response to bound virus, and protection of neutrophils from viral deactivation. The latter activities involve aggregation-dependent enhancement of the association of IAV particles with their natural, neuraminidase-sensitive binding sites on the neutrophil surface (44, 53, 140).

IAV binding to SP-A. Benne and coworkers demonstrated that the HA binds to sialic-acid residues on SP-A, presumably via the Asn-linked sugar associated with the CRD of SP-A (144). Herpes simplex virus can similarly interact with N-linked oligosaccharides on SP-A (107). By contrast with SP-D, human SP-A also has considerable HA inhibitory activity for PR8 (Hartshorn, manuscript submitted). The inhibitory activity for several strains, including PR8, is not inhibited by EDTA or by sugars expected to block collectin activity. On the other hand, binding of human SP-A to the neuraminidase of the A/X31 strain of IAV was partially inhibited by mannose and EDTA (142), suggesting both collectin-dependent and collectin-independent binding. Thus, it remains unclear as to the extent that the carbohydrate binding activity of SP-A contributes to its interactions with viral particles.

Collectin-mediated viral aggregation. An important aspect of the interaction of collectins with IAV is their ability to cause viral aggregation and to enhance other aggregation-dependent activities (44, 53). Among the collectins, SP-D is the most potent at aggregating IAV particles, and multimers of dodecamers are much more potent than dodecamers (Figure 4). Approximately 10-fold lower concentrations of SP-D dodecamers are needed to achieve maximal aggregation in light-scattering assays, as compared with SP-A or MBP octadecamers (53). SP-D-induced viral aggregates are also much larger than those obtained for SP-A or MBP. They can usually be visualized with the naked eye, and rapidly precipitate under unit gravity (Hartshorn, manuscript submitted). We speculate that the greater IAV-aggregating activity of SP-D results from its longer collagen domain, which theoretically allows binding between ligands separated by distances of approximately 100 nm, as compared with 20 nm for SP-A and MBP. The maximum distance between SP-D CRDs is approximately equal to the diameter of a single viral particle.

Effects of SP-D on infectivity and viral multiplication in vivo. SP-D is a potent inhibitor of IAV infectivity as measured by egg inoculation assay, and more potent at inhibiting IAV infectivity than SP-A (or MBL) (44, 145, 146). Recent studies have also shown that the susceptibility of various IAV strains to neutralization by SP-D in vitro is inversely correlated with pulmonary viral replication in mice following nasal inoculation (Margot Anders, submitted manuscript). The susceptibility to neutralization directly correlates with specific differences in the number of glycoconjugates expressed on the hemagglutinin. Thus, strains with more oligosaccharide chains are uniformly more susceptible to neutralization by SP-D in vitro and show reduced replication in vivo. Strains such as PR-8, which lack sugars on the HA and do not bind to SP-D, show very high levels of replication. Inoculation of mice in the presence of mannan, a saccharide competitor of SP-D binding to IAV, also increases viral replication in mice consistent with involvement of a C-type lectin.

Bacterial Glycoconjugates

The lung collectins bind glycoconjugates expressed by a variety of gram-negative bacteria including specific strains of such important pulmonary pathogens as Klebsiella pneumoniae, P. aeruginosa, Hemophilus influenzae, and E. coli (Table 3). However, their bacterial specificities overlap only partially, and their modes of interaction and the effects of binding on microbial interactions with host defense cells appear distinct.

Gram-negative bacterial LPS. SP-A preferentially binds specifically to the lipid A domain of rough forms of LPS and to purified lipid A (147, 148). The binding to purified rough LPS is calcium-independent and is not inhibited by competing saccharides, but is inhibited or partially reversed by lipid A. Consistent with these findings, human SP-A binds to certain rough, but not smooth, strains of E. coli with resulting opsonization and enhanced phagocytosis and killing (149). However, because lipid A is presumed to be embedded within the bacterial cell wall, it is unclear to what extent SP-A binding to lipid A accounts for binding to the intact organism.

View larger version (50K):