The Major Lung Surfactant Protein, SP 28-36, Is a Calcium-dependent, Carbohydrate-binding Protein*

SP 28-36, a major protein of pulmonary surfactant, has striking amino acid sequence homology with soluble mannose-binding proteins isolated from rat liver and contains residues common to the carbohydrate-binding domains of other mammalian lectins. We have used carbohydrate-affinity chromatography to inves-tigate carbohydrate-binding properties of SP 28-36 isolated from canine and human (alveolar proteinosis patients) lung lavage. SP 28-36 binds to immobilized D-mannose, L-fucose, D-galactose, and D-glucose. The protein binds only weakly to N-acetyl-D-galactosamine and N acetyl-D-glucosamine. Binding is Ca2+-depend-ent. The threshold Ca2+ concentration is 0.6 mM and maximal binding occurs with 1 mM Ca2+. Bound protein is quantitatively recovered by elution with 2 mM EDTA. Ba2+, Sr2+, and Mn2+, but not M 8 + , can substitute for Ca2+. Unlike some other mammalian lectins, SP 28-36 binds to carbohydrate at pH 5.0. Recombi- SP 28- 36 aqueous buffers at a large

3 To whom correspondence should be addressed.
The abbreviations used are: S P 28-36, surfactant protein group with a molecular weight of 28,000 daltons; SDS, sodium dodecyl sulfate; Hepes, 4-(2-hydroxyethyl)-l-piperazineethanesulfonic acid. 4). Each protein consists of a collagen-like NHp-terminal domain and variable N-linked glycosylation of the COOHterminal region (3-5). SP 28-36 binds readily to phospholipids and causes phospholipid aggregation in the presence of calcium (6-8). In concert with smaller hydrophobic surfactantassociated proteins, SP 28-36 promotes the calcium-dependent adsorption of surfactant phospholipids to an air-fluid interface (9). In a recent study it was found that SP 28-36 enhances the uptake of surfactant-like lipids by freshly isolated type I1 cells (lo), which suggests that this protein may regulate recycling and surfactant homeostasis.
The complete primary structures of two mannose-binding proteins from rat liver have been determined (11). Like SP 28-36 these proteins contain collagenous NHp-terminal segments linked to noncollagenous COOH-terminal domains. In addition to this overall organizational homology, mannosebinding proteins A and C can be aligned (by introduction of three and seven gaps, respectively) to show a striking 30% identity in amino acid sequence with SP 28-36. This led to the suggestion that SP 28-36 may have carbohydrate-binding activity (11). In this article we report the investigation of the carbohydrate-binding properties of SP 28-36 isolated from canine and human lung lavage.

EXPERIMENTAL PROCEDURES
Isolation of SP 28-36"Pulmonary surfactant was isolated by bronchoalveolar lavage of adult dogs (8) and patients with alveolar proteinosis (12). The surfactant in water (1.5 mg of protein/ml) was extracted in 1-butanol (1:50, v/v) at room temperature (13). The surfactant/butanol mixture was centrifuged twice at 10,000 X g. , for 20 min. The precipitated protein was dried under nitrogen and washed twice in 20 mM octyl-/3-D-glucopyranoside, 100 mM NaCl, 10 mM Hepes (pH 7.4). The proteins which were insoluble in this buffer were suspended in 5 mM Hepes, pH 7.4, and dialyzed against the same buffer for 48 h. The insoluble material was removed by centrifugation at 100,000 X g. , for 30 min and the supernatant, containing the purified SP 28-36, was stored in small aliquots at -20 "C. Detailed procedures for the expression and isolation of human recombinant S P 28-36 will be published elsewhere. Analyses-Sodium dodecyl sulfate (SDS)-polyacrylamide slab gel electrophoresis was performed according to Laemmli (14). Dithiothreitol (50 mM) was added to each of the samples. Protein determinations were carried out according to Bohlen et al. (15), using bovine serum albumin as a standard.
Reduction and Alkylation of SP 28-36"Purified SP 28-36 (0.5 mg) was incubated for 30 min at 37 "C in 3 ml of 5 mM Tris-C1, pH 7.4, 50 mM dithiothreitol. Iodoacetamide (1 M, in ethanol) was added to the mixture to a final concentration of 100 mM. The mixture was kept for 1 h at room temperature in the dark, dialyzed against 20 mM Tris-C1, pH 7.4, and finally dialyzed against 5 mM Hepes, pH 7.4.
Carbohydrate-binding Assay-Affinity chromatography was used to assess the carbohydrate-binding properties of SP 28-36. The assay was carried out at 4 'C. To prevent aggregation of the protein at this temperature, low ionic strength buffers were used that contained 0.1% (v/v) Triton X-100. Purified SP 28-36 (30 pg) in 5 mM Hepes, pH 7.4, 1 mM CaCl,, 0.1% (v/v) Triton X-100 (total volume, 1 ml), was applied to columns containing a 1-ml gel of immobilized monosaccharides (Selectins, Pierce Chemical Co.). After loading, the columns were eluted with 5 ml of 5 mM Hepes, pH 7.4, 1 mM CaCl,,

13877
Surfactant Protein SP [28][29][30][31][32][33][34][35][36] Is a Carbohydrate-binding Protein 0.1% (v/v) Triton X-100 and subsequently with 5 ml of the same a reversible equilibrium in binding ability. Delipidated SP 28buffer containing 2 mM EDTA instead of 1 mM CaC12. Fractions of 1 36 in aqueous buffers at physiolo~cal ionic is a large ml were collected. Aliquots of the fractions were taken for SDSpolyacrylamide gel electrophoresis and for protein determinations. oligomeric protein of unknown structure (7). We found that EDTA wash was calculated as a percentage of the total protein when the NaCl concentration was increased beyond 20 mM recovered.
at 4 "C even in the presence of Triton X-100. Even at the low ionic strength used in this study, it is likely that variable

RESULTS AND DISCUSSION
oligomerization of the protein occurs in the Dresence of Ca2+.
To express binding, the amount of protein which appears in the canine and sp 28-36 was precipitated by 1 mM ca2+ SP 28-36, isolated from canine and human lung lavage, binds to a number of immobilized monosaccharides under the conditions tested. Binding requires Ca2+ (1 mM) and the proteins are eluted from the affinity columns in the presence of 2 mM EDTA. Examples of gels on column fractions are shown in Fig. 1, and the results of numerous column runs are summarized in Table I. Both proteins were retained nearly quantitatively by immobilized mannose and fucose. Protein bound to the mannose column could be eluted by a 20 mM but not 10 mM mannose solution in the presence of 1 mM Ca2+. Canine SP 28-36 consistently bound only partially to immobilized glucose and galactose. When either the glucosebound or unbound protein was reapplied to immobilized glucose, the same distribution of binding was found, suggesting

Man GlcNAc
The extent of protein aggregation may diffe; between species (17). Variable aggregation may influence protein binding to the immobilized sugars. Both human and canine SP 28-36 bound poorly to immobilized N-acetyl-sugars. SP 28-36 did not bind to the column matrix with or without Ca2+. Although these results suggest different affinities for the various sugars, it should be emphasized that the column assay cannot be used to quantify affinities. The use of low ionic strength buffers in this study may also have affected relative binding potencies. Further studies will be required to establish the relative specificity of carbohydrate binding to various sugars at physiological ionic strength.
The binding of SP 28-36 to immobilized monosaccharides is Ca2+-dependent, just as for some other mammalian lectins (18)(19)(20)(21)(22). The threshold Ca2+ concentration for binding is 0.6 mM and maximal binding occurs with 1 mM Ca2+ (Fig. 2). S3+, Mn2+ and Ba2+ could substitute for Ca2+, although Mn2+, and particularly, Ba2+, seem to be less effective (Fig. 2). Mg+, in concentrations up to 2 mM, could not substitute for Ca2+. Unlike the membrane-bound hepatic receptors (18)(19)(20) and the soluble mannose-binding proteins (22), SP 28-36 retains its Ca2+-dependent binding activity at pH 5. Partial proteolysis of rat (23) and chicken (24) hepatic lectin and collagenase treatment of soluble mannose-binding protein from rat liver (11) yield COOH-terminal fragments that are able to bind to carbohydrate-affinity columns. This suggests that the mem-    . 2. Binding of human SP [28][29][30][31][32][33][34][35][36] to immobilized mannose as a function of the divalent cation concentration. SP 28-36 was applied to the affinity columns in 1 ml of 5 mM Hepes, pH 7.4,0.1% (v/v) Triton X-100, and different concentrations of divalent cations (as chlorides). The columns were eluted with 5 ml of the same buffer and were subsequently eluted with 5 ml of 5 mM Hepes, pH 7.4,0.1% (v/v) Triton X-100, and 2 mM EDTA. The amount of bound protein that eluted in the presence of 2 mM EDTA was expressed as the percentage of the total recovered protein.  brane-spanning and collagen-like portions of these respective proteins are not required for carbohydrate binding. Cysteine residues and residues located near cysteine residues in the carbohydrate-recognition domains are highly conserved (11), suggesting that disulfide bond formation may be necessary to produce an active binding domain in all of these proteins (23).
As shown in Table 11, reduction of SP 28-36 results in the complete loss of binding activity. This is consistent with the suggestion that the homologous COOH-terminal portions of the human and dog SP 28-36 and the hepatic lectins are folded into similar disulfide-bonded domains.
In spite of the similarities in the structures of SP [28][29][30][31][32][33][34][35][36] and the mannose-binding proteins, we were not able to demonstrate carbohydrate binding of either the human or canine collagenase-resistant fragment. We have previously reported the NH2-terminal amino acid of this fragment as Gly-78 in the dog (3). The failure to bind the collagenase-resistant fragment to the monosaccharide columns may be the result of nonspecific cleavage or other modifications within the binding domain as a result of the collagenase treatment but it is interesting that the binding capacity of SP 28-36 could also be destroyed by heat treatment for 10 min at 50 "C but not at 45 "C (  Fig. 3. No protein of this molecular weight is obtained by affinity chromatography of media obtained from untransfected cells (data not shown).
This study shows that properties of proteins may sometimes be predicted from sequence homologies with other proteins. SP 28-36 shares a common carbohydrate-recognition domain with other proteins and was found to bind to immobilized monosaccharides in a Ca2+-dependent fashion. In order to elucidate the biological significance of this finding, other assays will be needed to determine the specific binding affinities of SP 28-36.