Phospholipid binding and biophysical activity of pulmonary surfactant-associated protein (SAP)-35 and its non-collagenous COOH-terminal domains.

Surfactant-associated protein of Mr = 35,000, SAP-35, is the major glycoprotein present in mammalian pulmonary surfactants. In this study, canine SAP-35 and several of its COOH-terminal peptides were purified and characterized by amino acid composition and NH2-terminal sequencing analysis. These proteins were then studied in terms of their specific lipid-binding characteristics and surface activity when combined with a synthetic phospholipid mixture, SM, chosen as an approximation of lung surfactant phospholipids. Purified, delipidated SAP-35 bound SM strongly. In contrast, SAP-21 (a non-collagenous fragment generated by collagenase digestion) bound phospholipid weakly; SAP-18 (an acidic COOH-terminal fragment comprising residues Gly-118 to Phe-231) did not bind phospholipid, demonstrating the importance of hydrophobic amino acid residues Gly-81 to Val-117 and the NH2-terminal collagenous domain in interaction of the SAP-35 with phospholipids. In surface activity experiments, purified SAP-35 enhanced the adsorption of SM phospholipids in terms of both rate and overall surface tension lowering. However, the adsorption facility of the SM-SAP-35 mixture did not approach that of either whole surfactant or the surfactant extract preparations, calf lung surfactant extract or surfactant-TA, used in exogenous surfactant replacement therapy for the neonatal respiratory distress syndrome. In addition, the dynamic surface activity of the SM-SAP-35 mixture was well below that of natural surfactant or surfactant extracts. This was also true of mixtures of SM phospholipids combined with the SAP-18 and SAP-21 fragments of SAP-35.

____ major pulmonary surfactant-associated protein and has been detected in numerous mammalian surfactants (2-10). However, the existence of other surfactant apoproteins bas also been reported in a variety of studies (I-lO), and it is now clear that smaller s u r f a c t~t -a s s o c j a~d proteins of M , = approximately 6,000, which are clearly not fragments of SAP-35, are also prominent in m a m m~l i~n pulmonary surfactant extracts (11). The various roles played in lung surfactant activity by SAP-35 and other surfactant-associated proteins are currently a subject of much debate and interest. For example, a number of studies have demonstrated that lipidrich extracts of natural lung surfactant (such as calf lung surfactant extract, CLSE or calf lung lipid (12)(13)(14)(15)(16) and surfactant-TA (17)(18)(19)) have equivalent biophysical or physiological activity to whole surfactant. However, recent evidence suggests that such preparations lack SAP-35 and are enriched in other surfac~nt-associated proteins which may confer surface activity to the mixtures (11,(20)(21)(22)(23).
SAP-35, first described by King and Clements (I), is now known to be synthesized primarily by pulmonary Type II epithelial cells, and in the rat and dog, is derived from a primary translation product of M, = 26,000 which is highly glycosylaied by the addition of asparagine-linked complex carbohydrate (9,24). The protein has been referred to as apolipoprotein A, glycoprotein A, pulmonary surfactant apolipoprotein, and SAP-35 by various investigators. The entire amino acid sequence of this protein has been derived from cDNA sequence and has been recently reported for both canine and human 26). Sulfhydryl-dependent oligomerization of the various glycosylated and unglycosylated forms present in canine and rat surfactant occur in the N-terminal region of the molecule and the s u l~y d~l -d e p e n dent interchain cross-linking is lost by treatment of the molecute with bacterial collagenase selective for the Gly-X-Y amino acid repeats in the collagen-like region of the molecule (8,27). In the presence of calcium, SAP-35 binds to phospho-Iipids forming vesicles and aggregates with some of the features of natural surfactant (28,29).
In the present study, we have carried out specific studies of the binding of SAP-35, and several weli-defined segments of SAP-35, with a synthetic phospholipid mixture containing an acyl chain distribution and headgroup constituents (zwitterionic, anionic) related to lung surfactant phospholipids. These lipid-binding studies are then complemented with direct determinations of the surface activity in vitro of the 'lipidapoprotein recombinants in terms of their ability to adsorb to the air-water interface and to lower surface tension under dynamic compression in an oscillating bubble at physiologically relevant temperature, humidity, and cycling rate. In addition to addressing potential contributions of SAP-35 t.o lung surfactant activity, the experiments here also test whether specific regions of the molecule might confer particular lipid-binding characteristics, or generate particular surface active properties, in combination with phospholipids.

MATERIALS AND METHODS
Surfactants for Binding or Surface Activity Studies-A number of synthetic phospholipids and other surfactants were used in this study in addition to the surfactant-associated protein SAP-35 and its various molecular segments described below. Synthetic phospholipids were purchased from the Sigma and were used without further purification after verification of content by thin layer chromatography using the system of Touchstone et al. (30). For binding and surface activity studies of synthetic phospholipids and surfactant-associated proteins, a specific phospholipid mixture used was 65% dipalmitoyl phosphatidylcholine, 20% egg-phosphatidylcholine, 7.5% egg-phosphatidylglycerol, and 7.5% soy-phosphatidylinositol. This synthetic phospholipid mixture (SM) was chosen because its fatty acyl chain composition is relatively reflective of that in lung surfactant phospholipids, and the major zwitterionic and anionic headgroups are also closely relevant for natural surfactant.
For comparative standards representing highly active surfactants, both whole lung surfactant (LS) and organic solvent extracts of lung surfactant (CLSE and surfactant-TA) were used. Natural surfactant was obtained by bronchoalv~lar lavage of intact calf lungs followed by centrifugation, as detailed by Notter et ai. (13). CLSE was obtained from this whole calf LS by ch1oroform:methanol extraction following the method of Bligh and Dyer (31). The composition of natural LS and CLSE obtained by these methods has previously been reported (13)(14)(15)(16)32). Surfactant-TA, which is an organic solvent extract of minced bovine lungs supplemented with dipalmitoyl phosphatidylcholine and palmitic acid, was provided by Abbott with a composition similar to that reported by Vidyasagar et a1. (19). Both CLSE and surfactant-TA have been shown to be biophysically and physiologically active lung surfactants (13-16, 18, 19), and have been used successfully to treat premature infants with the neonatal respiratory distress syndrome (17,33,34).
Delipidation and Initial Purification of SA P-35i"ongrel dogs were sacrificed by pentobarbital-KC1 injection. The lungs were rapidly removed and placed on ice. The trachea was cannulated and the lung infused with 0.9% NaCI, 1 mM phenylmethylsulfonyl fluoride. Surfactant from lavage of intact dog lung was collected and placed on ice. Samples containing visible blood were not utilized. Lung lavage was then centrifuged twice at 1000 X g for 5 min at 4 "C to remove cells and debris. Supernatant was then centrifuged at 30,000 X g for 15 min, and the dense pellet was resuspended in 0.9% NaCl, 1 mM phenylmethylsulfonyl fluoride and sonicated on ice by three 10-s bursts with a Fisher model 300 Dismembranator. The material was again pelleted at 30,000 X g, resuspended in 0.9% NaC1, 1 mM phenylmethylsulfonyl fluoride, and stored at -80 "C.
For delipidation and purification of surfactant-associated protein, the pelleted material from each dog lung was resuspended in 20 ml of ether/ethanol (3:l) and incubated at -20 "C overnight. The extract was centrifuged at 10,000 X g for 10 min. The pellet was re-extracted first with ether/ethanol for 3 h and finally with pure ether for 1 h at -20 "C. The final pellet was air dried and resuspended in 10 ml of 100 mM Tris-C1, pH 7.4, 6 M guanidine hydrochloride. Solubilized protein was then dialyzed twice against 1-liter volumes of 4 M electrophoresis grade urea (Bio-Rad) at 4 "C to give a delipidated protein product. Insoluble material was removed by centrifugation at 30,000 X g for 20 min. This delipidation and resolubilization technique is similar to that described by Cardin et al. (35) for plasma lipoproteins.
Preparatiue Isoelectric Focusing for SAP-35 from Delipidated Protein-For further purification of delipidated surfactant-associated protein, the method of isoelectric focusing was used. This involved a slight modification of procedures described hy Vesterberg (36) and employed an LKB 110-ml capacity column (LKB). The dense cathode solution contained 60% sucrose, 4 M urea, and 0.3 M ethylenediamine in a total volume of 21 ml. Delipidated surfactant protein resuspended in 4 M urea was applied within a linear 0-50% sucrose gradient containing 4 M urea and 2.5% (v/v) LKB carrier a m p h o l~ mixture, composed of 20% pH 3.5-5, 25% pH 4-6,35% pH 5-7, and 25% pH 6-8 ampholines. The anode solution contained 1% sulfuric acid. Electrophoresis was carried out at 200 V for 6 h, 400 V for 12-16 h, and 600 V for 24 h. Two-mi fractions were collected and the pH measured. Twenty microliters of selected fractions were mixed di-rectly with 75 pl of SDS-PAGE buffer for analysis. Fractions from the isoelectric focusing column containing SAP-35 as determined by SDS-PAGE analysis were pooled, dialyzed extensively in 0.3 M ammonium bicarbonate, pH 7.8, and lyophilized. Finally, it was found that additional purification of SAP-35 could be accomplished by another chromatography step, where the protein samples were applied to a Cibacron-Blue Sepharose CL-GB column (d = 1 em, E = 15 cm) equilibrated with 0.25 M sodium phosphate, pH 7.0, 1 mM dithiothreitol. Typically 5-10 mg of protein were added per milliliter of column running buffer. Fractions (2 ml) were collected at a flow rate of 12 ml/h and monitored by UV adsorbance at 280 nm.
SDS-PAGE Analysis-One-millimeter thick 13% polyacrylamide gel slabs were prepared as described by Laemmli (37) using the limited SDS conditions suggested by Wyckoffe et al. (38). Sample buffer contained 125 mM Tris-HCl, pH 6.8, 10% glycerol, 2% SDS, and 1% p-mercaptoethanol. Mark VII-L molecular weight markers (Sigma) were used as standards. Electrophoresis was carried out at a constant 30 mA/slab through a 3% polyac~lamide stacking gel. Amperage was raised to 40 mA/slab when the bromphenol blue dye front had entered the separating gel.
Two-dimensional Gel Electrophoresis-Two-dimensional IEF-polyacrylamide gel electrophoresis analysis of surfactant proteins was carried out using the method of Garrison and Wagner (39) with the sample buffer as described by Anderson and Anderson (40). Isoelectric focusing was performed using carrier ampholytes (final concentration of 2%) mixed as 2 parts pH 6-8, 2 parts pH 5-7, and 1 part pH 3.5-10. Isoelectric-focusing gels were layered on top of the 13% ~lyacrylamide, 0.1% SDS gel slabs for separation of proteins on the basis of molecular weight. Slab gels were analyzed using a modification of the silver-stain technique described by Sammons et aZ. (41). SAP-35 had a characteristic salmon pink stain using this procedure.
Isolation of the Acidic CO~H-term~nal F r~~e n t , SAP-18-The preparative isoelectric-focusing procedure used also yields a major SAP-35 subfragment referred to here as SAP-18. Because of the marked acidity of the SAP-18 peptide, which migrated with pi 3.5-4.0 during isoelectric focusing, virtually homogeneous preparations of the peptide were isolated from the acidic end of the preparative isoelectric-focusing column described earlier. Fractions containing only SAP-18 were dialyzed extensively in 0.3 M ammonium bicarbonate, pH 7.8, and lyophilized. Purity was assessed by silver-stain analysis after SDS-PAGE.
Endoglycosidose F Treatment of SAP-18-In order to better estimate the polypeptide molecular weight of SAP-18, the peptide was resuspended (0.2 mg/ml) in 0.1 M sodium phosphate, pH 6.1, 10 mM disodium EDTA, and 0.1% Nonidet P-40. Endoglycosidase F from New England Nuclear was added at a final concentration of 2 units/ ml and the sample incubated at 37 "C.
HPLC Separation of Tryptic Fragments-In order to compare their polypeptide structural similarities, purified SAP-35 and SAP-18 were reduced and carbox~ethylated and digested with trypsin-TPCK (Cooper Biomedical, Malvern, PA). Tryptic fragments were separated using an Ultrapore analytical Cs column (4.6 mm x 75 mm, Beckman Instruments) as previously described (27). Details of the elution procedure are provided in the legend to Fig. 3 later under "Results." E n~y m e -l i n~d I m m u n o s o r~n t Assay (ELISA) of Canine SAP-35-Monospecific antisera were prepared against canine SAP-35 in both goats and rabbits by standard immunization protocols using Freund's complete adjuvant. Resultant antisera were fractionated with ammonium sulfate to obtain immunoglobulin for use in a capture ELISA assay. ELISA assays followed the general procedure described by Katyal and Singh (42). Plastic plates were coated with goat anti-SAP-35 (1:lOO) in 0.9% NaCl, 50 mM Tris-HC1, pH 7.4, overnight at 4 "C. Unknown samples were added in the same buffer containing 0.5% Nonidet P-40 and incubated for 2 h at 37 "C. Rabbit anti-SAP-35 was then added (1:lOOO) followed by specific goat anti-rabbit IgG conjugated to horseradish peroxidase (Behring Diagnostics). Color development was assessed by absorbance after addition of 0.15% H,O, using ff-phenylenediamine as the substrate. Both antisera were reactive against glycosylated and non-glycosylated SAP-35 forms, detecting only SAP-35 in immunoblots of whole canine surfactant, and were unreactive to canine serum proteins. Both antisera detected only the M, = 26,000 translation products after in vitro translation of adult canine lung poly(A+) mRNA (9). Standard curves were constructed with purified SAP-35. Protein was assessed by the method of Lowry et al. (43) using bovine serum albumin as the standard.
Immunoblot Assay-Surfactant proteins were separated by one-or two-dimensional SDS-gel electrophoresis and transferred electrophoretically to nitrocellulose. The nitrocellulose sheet was treated with rabbit anti-SAP-35 antisera followed by horseradish peroxidase conjugated goat anti-rabbit IgG. The blocking buffer described by Johnson et al. (44) was included in the incubation to reduce nonspecific background. Color was developed as described by Towbin et al. (45) except that 4-chloro-napthol was used as a substrate.
Amino Acid Composition Annlysis-Details of amino acid analytical procedures for surfactant-associated proteins are given by Ross et al. (27). Purified SAP-35 samples and the various peptides above were hydrolyzed in 300 pl of 5.7 N HCI, 0.01% phenol, and 0.1% pmercaptoethanol a t 110 "C under vacuum. Free amino acids were resolved using a Beckman 6300 amino acid analyzer. Automated Edman degradation was performed using an Applied-Biosystems model 470A vapor phase protein sequencer.
Phospholipid Binding Studies-The binding of SAP-35, and specific molecular segments of SAP-35, to phospholipids was assessed using both the ELISA assay and silver-stain analysis. The phospholipids studied were the SM described earlier. Phospholipid vesicles were prepared by sonication at 45 "C of SM phospholipids, dried to a thin film under nitrogen, into a buffer composed of 20 mM Tris-HC1, pH 7.4, 0.1 M NaC1, and 5 mM CaC12. Purified SAP-35 or its SAP-21 and SAP-18 fragments were then added to the phospholipids. The dispersion was incubated for 2 h at 37 "C with rotation, and bound protein was then separated from soluble protein by centrifugation at 10,000 X g for 15 min. Recovery of ['4C]phosphatidylcholine, added to assess the recovery of lipid in the presence of each peptide, was always greater than 99% in the 10,000 X g pellet. SAP-35 binding was assessed using two-antibody ELISA and SDS-PAGE analysis of the resultant lipid pellet and aqueous fractions. Phospholipid binding of SAP-18 and SAP-21 was analyzed directly by immunoblot analysis after SDS-PAGE, since the capture ELISA did not detect the COOHterminal fragments.
Interfacial Biophysical Methods-Biophysical activity of lipid-protein mixtures was evaluated by measurements of adsorption facility in the absence of diffusion resistance (12)(13)(14) and dynamic surface tension-lowering ability on an oscillating bubble apparatus (46). The phospholipids used were the same as in the binding studies above. To form lipid-protein mixtures, SM phospholipids were dried from chloroform under nitrogen into the wall of a test tube, and a solution of surfactant-associated protein in 0.15 M NaCl and 1.4 mM CaC12, pH 6, was added to a final lipid/protein ratio in the range 982 to 99.5:0.5 mg/mg. The lipids were then dispersed into the protein-containing aqueous solution using a vortex mixer (Scientific Industries, Inc., Bohemia, NY) for 1 h at room temperature. The dispersed proteinlipid mixture was then added at time zero to a stirred 70-ml subphase of 0.15 M NaCl, 1.4 mM CaClz in a Teflon evaporating dish, and surface pressure ( x ) (surface tension lowering) measured as a function of time from the force on a Wilhelmy (platinum, sandblasted) slide dipped into the interface. As detailed previously (12)(13)(14), this procedure defines a x-T adsorption isotherm in the absence of diffusion resistance. Subphase temperature used was 35 +. 2 "C for adsorption experiments.
In addition to adsorption facility, surface tension-lowering ability during dynamic cycling was also measured, since lung surfactant behavior under dynamic conditions is critical for its physiologic activity (47). Dynamic surface tension measurements were made at 37 "C and 100% humidity in an oscillating bubble apparatus developed by Enhorning (46). This apparatus measures a combination of dynamic surface tension lowering, respreading, and adsorption, at cycling rates (20 cycle/min) and surface area changes (50%) relatively reflective of the lung in vivo. In previous studies, the in vitro biophysical activity of surfactants on the oscillating bubble apparatus has been found to correlate well with their physiologic efficacy in lungs (16). Details of oscillating-bubble methods applied to pulmonary surfactant and phospholipid-containing systems have been given previously (14,16,32).
In the biophysical experiments here, bubble measurements were made on the same lipid-apoprotein dispersions studied above for adsorption facility. As standards representing highly active surfac-tants for comparative purposes, additional adsorption and oscillatingbubble experiments also investigated the surface activity of natural LS, and the organic solvent surfactant extracts CLSE and surfactant-TA. For these experiments, CLSE was dispersed by vortexing as above for the lipid-apoprotein mixtures. Surfactant-TA was dispersed by sonication (Heat Systems sonicator, model W-220F, microtip, power 25 watts), and natural surfactant was simply resuspended in the standard calcium-supplemented saline solution used in all interfacial biophysical experiments. For all biophysical experiments, phospholipid concentration was defined by phosphate determinations using the method of Chen et al. (48). mixture of glycosylated and unglycosylated proteins which copurify (27). A second acidic protein migrating with PI 3.5-3.8, M , approximately 18,000, was also detected and is displayed in Fig. 1A. The most acidic surfactant-associated protein fractions (PI < 3.8) provided an essentially homogeneous preparation of this polypeptide (designated SAP-18) for further analysis.
Comparison of the structural features of SAP-35 and SAP-18 was achieved by analysis of tryptic fragments contained within each polypeptide. The SAP-18 elution pattern (Fig. 3) shares identical tryptic peaks with the SAP-35 profile previously reported by our laboratory (27)  Approximately 350 mg of purified protein was reduced and alkylated and digested with trypsin TPCK. Separation was achieved by reverse phase chromatography using an Ultrapore Cs column. Sample was applied onto the column equilibrated in 0.1% trifluoroacetic acid at a flow rate of 1 ml/min. Fragment elution was achieved via a linear acetonitrile gradient (0-35%, 1% per min) and monitored by absorbance at 225 nm.

TABLE I Amino acid composition of dog SAP-35, SAP-21, and SAP-18
The polypeptides were hydrolyzed in 300 pl5.7 N HCl, 0.3% phenol, 0.01% P-mercaptoethanol at 110 "C under vacuum. Free amino acids were resolved using a Beckman 6300 amino acid analyzer. See "Materials and Methods" for details.  In summary, composition and sequence data indicate that SAP-18 represents a non-collagenous fragment of SAP-35, starting with the glycine residue at position 118 and extending to the COOH terminus. It contains 1 asparagine-linked oligosaccharide likely at position Asn-190 and is significantly acidic in nature (PI < 3.8).
Production and Analysis of a Collagenase-resistant Fragment SAP-21-We have recently reported an initial characterization of the major non-collagenous domain of . This major segment, SAP-21, is a collagenase-resistant fragment of SAP-35 and incorporates the SAP-18 fragment above. Specifically, following treatment of SAP-35 with bacterial collagenase, a large resistant fragment of M, = 21,000, containing asparagine-linked carbohydrate and having an apparent polypeptide molecular weight of 16,000 is produced, herein termed SAP-21. The amino acid composition of this collagenase-resistant peptide, isolated by HPLC gel filtration using a Du Pont GF-250 column, is also listed in Table I Table I for this collagenase-resistant fragment closely matches the corresponding composition predicted by the SAP-35 sequence. It can be inferred, therefore, that the collagenase-resistant fragment, SAP-21, contains within its structure the acidic SAP-18 sequence.
Phospholipid Binding-As one measure of the interactions of SAP-35 in lung surfactant, specific studies were carried out on the binding of this surfactant-associated protein and its specific fragments (SAP-18 and SAP-21) with a complex mixture of synthetic phospholipids reflective of those in natural surfactant. The results of the lipid-binding studies are shown in Figs. 4 and 5. The overall protein, SAP-35, had a clear affinity for the SM dispersed by sonication into vesicles. SAP-35 was precipitated with lipid from aqueous solution by centrifugation (see "Materials and Methods"). SAP-35 alone did not pellet at 10,000 x g in the absence of lipids. Binding of SAP-35 to synthetic phospholipid vesicles was rapid (complete at 1 min) at 37 "C and apparently nonsaturable up to 1 mol of protein/300 mol of phospholipid (Fig. 4). Results using silver staining after SDS-PAGE (Fig. 4A) or the ELISA assay ( Fig. 4B) to determine binding were comparable. The SAP-21 fragment, however, bound phospholipid weakly, while the SAP-18 fragment failed to bind and was completely recovered in the supernatant of the reconstitution assay (Fig. 5). This latter finding is of particular interest since, as shown below, SAP-18 had equivalent effects to SAP-35 in enhancing the adsorption of SM phospholipids.
Biophysical Surface Actiuity Results-Adsorption results for SAP-35 and its fragments combined with SM phospholipids are shown in Table I1 and Fig. 6. The data indicate that mixtures of synthetic phospholipids with SAP-35 and its fragments do not exhibit adsorption facility comparable to whole LS or to the surfactant extracts CLSE and surfactant-TA. The maximum adsorption surface pressure reached by the SAP-35-phospholipid mixture in Table I1 was  Twenty-five-microliter aliquots were electrophoresed through a 13% acrylamide gel and silver stained. The lipid to protein molar ratios during the incubation ranged from lo5 to 1 ( l a n e a) to 3.3 X 10' to 1 ( l a n e I ) . B, ELISA assay of SAP-35 binding to phospholipid. SAP-35 was incubated at 37 "C with the synthetic phospholipid mixture SM (1 mg/ml). The lipids were separated from soluble protein by centrifugation. SAP-35 was assessed in both the supernatants (A) and pellets (0) by a two-antibody capture ELISA assay, described under "Materials and Methods." As in A , the lipid to protein molar ratios ranged from 10' to 1 to 3.3 X 10' to 1.  Surface pressure T is the amount of surface tension lowering below that of the pure subphase. The T value at 0 min is that measured within 10 s after addition of a bolus of surfactant dispersion to the stirred subphase at time zero. Values given are the means of 4-10 experiments with S.E. always <3 dynes/cm, except for SAP-18 data which are the mean of two experiments with a deviation of ? 0.5 dynes/cm about the mean. * Delipidated proteins studied are SAP-35, SAP-21, and SAP-18 (see "Materials and Methods"). These are combined with SM composed of dipalmitoyl phosphatidylcholine/egg-phosphatidylcholine/ egg-phosphatidylglycerol/soy-phosphatidylinositol 65:207.5:7.5. The surfactant extracts CLSE and surfactant-TA, along with natural LS, represent optimally active surfactants; ovalbumin is used as a nonspecific protein control. Phospholipid concentration for all adsorption experiments was 0.063 mg/ml 0.15 M NaCl + 1.4 mM CaC1,; temperature = 35 & 2 "C. For cases 2-5, the lipid/protein ratio was (wt/wt) 991-982. For comparison, the protein content of CLSE is 1% (12)(13)(14)(15)(16) while that of surfactant-TA is 1-2% (18)(19). All mixtures dispersed by vortexing at room temperature except surfactant-TA (sonicated) and natural LS (resuspended). adsorption was significantly enhanced by SAP-35 or its purified fragments SAP-18 and SAP-21. There is even a rapid rise segment present in some of the representative K-7 adsorption isotherms shown in Fig. 6. However, this enhancement of adsorption did not correlate with the phospholipidbinding studies. For example, as alluded to above, SAP-18 failed to bind phospholipid (Fig. 5), but its ability to enhance adsorption was equivalent to SAP-35 which was completely associated with phospholipid.  Table 11. Surface pressure-time ( T -T ) adsorption isotherms are shown in panek A-C for synthetic phospholipids combined with SAP-35 or its fragments, as representative for cases 3 to 5 in Table 11  2 <1 Surfactant dispersions were studied on the oscillating bubble at 37 "C, 100% humidity, cycling rate 20 cpm and area compression 50%. Subphase and dispersion methods used were those in Table 11, and concentration was 1 or 2 mg of lipid/ml of bubble subphase as noted. See "Materials and Methods" and (14,16,32,46) for further oscillating bubble methodology. Natural LS and the extracts CLSE and surfactant-TA represent active lung surfactant preparations against which to compare the lipid-protein combinant mixtures of cases 3-5. See text for details.
These adsorption results, that SAP-35 or its fragments were not able to generate activity approaching that of natural LS, are in agreement with the dynamic surface tension-lowering studies in Table 111. The data reported there give the minimum surface tension reached by mixtures of SAP-35 or SAP-21 with phospholipids under rapid compression-expansion in an oscillating bubble (see "Materials and Methods"). At. concentrations as low as 0.5 mg of phospholipid/ml of subphase, physiologically active natural surfactant, or CLSE, can lower surface tension to minimum values <1 dynesjcm on dynamic cycling (32); explicit data for such surfactants are given in cases 6-8 in Table I11 at a dispersion concentration of 1 mg/ ml. By contrast, mixtures of phospholipids and SAP-35 did not lower surface tension on the oscillating bubble below 21 dynes/cm at equivalent (or higher) concentrations, for the experimental conditions used here (20 cycles/min compression rate with a 50% area change at 37 "C).

DISCUSSION
SAP-35 is the major surfactant-associated protein in the mammalian alveoli. Significant evidence supports the hypothesis that specific protein components contribute to the biophysical and physiological activity of mammalian surfactants (29). Understanding the molecular mechanisms involved in surfactant protein and phospholipid interactions necessitates stringent characterization of these protein structures. In the present work, we have isolated COOH-terminal fragments of purified canine SAP-35 and measured the binding and biophysical activity of phospholipid mixtures containing SAP-35 and these specific SAP-35 fragments.
Characteristics of SAP-35 and Its Fragments-SAP-35 is heterogeneous with respect to both size and charge in various species (3)(4)(5)(6)(7)(8)(9)(10)24,27). This is due to the presence of sialylated, asparagine-linked carbohydrate chains covalently bound to a primary translation product of M, = 26,000-28,000 (8,9, 271, Larger forms present in lung lavage arise from sulfhydryldependent oligomerization of SAP-35 molecules (8,27). In the present work, an acidic polypeptide (PI < 4.0, M, = 16,000-18,000), which shares antigenic sites with SAP-35 by virtue of reacting with rabbit anti-SAP-35 antisera, was also detected in canine surfactant. This acidic polypeptide, SAP-18, contains a single N-linked carbohydrate chain and an apparent molecular mass of 14,000 daltons. Amino acid composition and sequence analysis identify this peptide to be a fragment of SAP-35, extending from residues Val-117 or Gly-118 to the COOH terminus. Previous work by King et al. (2,49,50) demonstrated the presence of a small (M, = 10,000-12,000) surfactant protein (termed surfactant apolipoprotein B) which shared immunological properties with SAP-35. It was suggested that "apolipoprotein B" might represent a metabolic product of 50). The data presented here demonstrate that SAP-18 is a naturally occurring fragment of SAP-35, and its amino acid composition in Table I closely  residues (glutamate, aspartate), and is clearly distinct from the M, = 6,000 protein rich in hydrophobic amino acids (11,23) and insoluble in water.  Benson et al. (25). The non-collagenous domain (Gly-81 to Phe-231) produced by bacterial collagenase treatment (SAP-21) and the acidic (GIy-118 to Phe-231) CoOH-terminal fragment (SAP-18) isolated from whole surfactant are presented also. The two asparagine-linked glycosylation sites are noted, as is the cysteine residue implicated in intermolecular disulfide bonding (27) and the hydrophobic region of the SAP-35 sequence most probably involved in phospholipid binding, residues Gly-81 to Val-117.
Discussion of Lipid Binding Studies- King et al. (29,51) have demonstrated that reassembly of surfactant apoprotein (apolipoprotein A or SAP-35) with synthetic phospholipid vesicles does occur. They reported that SAP-35 associated with phospholip~d in a nonsaturable manner over a phospholipid/protein molar ratio range of 3201 to 80001. In our studies, intact SAP-35 bound to synthetic phospholipid dispersions in aqueous solution in an apparently similar nonsaturable manner (Fig. 4). However, qualitative SDS-PAGE analysis showed that SAP-18 (Gly-118 to Phe-231), the SAP-35 COOH-terminal domain, did not associate with phospholipid (Fig. 5). This result was anticipated by the lack of the strongIy hydrophobic domain in the SAP-18 primary sequence. The SAP-35 collagenase-resistant domain (SAP-21), extending from residue Gly-81 to Phe-231, associated weakly with phosphoiipid (Fig. 5).
Study of the SAP-35 sequence reveals amino acid domains which are characteristic of phospholipid (membrane)-bound proteins. The SAP-35 amino acid sequence extending from residues GIy-81 to Ile-101 fits the model for an amphipathic a-helix and is followed by a strong nonpolar region, Leu-102 to Val-117. Amphipathic helices have been shown to facilitate the interaction of membrane-bound proteins including plasma apolipoproteins (52), P-lipotropin (53), and calcitonin (54) to phospholipids. The polar and nonpolar faces of a SAP-35 amphipathic helix could associate with phospholipid head groups and fatty acyl chains, respectively. The adjacent hydrophobic region Leu-102 to Val-117 might then traverse the lipid bilayer, thereby anchoring SAP-35 in the membrane. Contiguous nonpolar amino acid sequences have been demonstrated to anchor immunoglobulin heavy chains (55) and the vesicular stomatitis virus G protein (56) in cellular membranes. Furthermore, basic amino acids have been reported to mark the boundary of transmembrane sequences (56-57). Canine SAP-35 contains residues Lys-119 and Arg-120 immediately following the hydrophobic Leu-102 to Val-117 domain. If the hydrophobic region of SAP-35 (through Val-117) is buried within the lipid bilayer, then the generation of the COOH-terminal Gly-118 to Phe-231 fragment may occur through proteolysis of the molecule at the point where it extends beyond the lipid surface. In any case, the finding that SAP-21 (which in contrast to SAP-18 contains the putative lipid-binding hydrophobic region) binds only weakly to phospholipid, supports the hypothesis that the structural features necessary for phospholipid binding of SAP-35 are also dependent upon the integrity of the collagen-like domain at the NH2-terminal region of SAP-35. Discussion of Interfacial Biophysical Results-The biophysical activity of SAP-35 and its COOH-terminal fragments was assessed in adsorption studies and in studies using an oscillating bubble. King and MacBeth (51) have reported weak biophysical effects of purified SAP-35 on lipids by monitoring SAP-35 association with dipalmitoyl phosphatidylcholine vesicles. Hawgood et al. (28) have also shown that SAP-35 reduces surface tension in adsorption experiments when added to lung surfactant lipid extracts, but interpretations are complicated because such lipid extracts contain other distinct apoprotein components (11,22,23). The present experiments demonstrate relatively low adsorption facility (compared to natural lung surfactant) for mixtures of synthetic phospholipids with SAP-35 and its fragments. Moreover, the results show that SAP-35, its collagenase-resistant COOH-terminal fragment , and the Gly-118 to Phe-231 fragment (SAP-18) induce comparable adsorption when added to a synthetic phospholipid mixture, even though only SAP-35 binds avidly to phospholipid. SAP-18, which binds lipid to a negligible extent, showed equivalent adsorption facility to SAP-35 when combined with SM phospholipids. In addition, the magnitude of the observed adsorption, while significantly greater than that of the nonspecific protein ovalbumin plus lipid, was not as large as found for natural surfactant or the surfactant extract preparations CLSE and surfactant-TA (12)(13)(14)(15)(16)(17)(18). Significantly, these latter surfactant extracts lack SAP-35 or its fragments and contain only smaller hydrophobic protein(s), including SAP-6 and its oligomers as recently reported from our laboratory (11,23). Finally, in oscillating bubble studies of dynamic surface activity, none of the mixtures of SM phospholipid plus SAP-35 and its fragments studied here exhibited dynamic surface tension-lowering ability approaching that of natural LS or surfactant extracts.
In summary, the biophysical data of the present study do not support the concept that SAP-35, combined with phospholipids, can provide the same optimal surface properties of whole LS or surfactant extracts used in exogenous surfactant replacement therapy for respiratory distress syndrome. This does not completely rule out the concept that SAP-35 or specific structural regions of the molecule may be useful in conferring surface active properties to phospholipids. In ad-dition, combinations of SAP-35 with other distinct surfactant apoproteins such as SAP-6 (11,23) may also prove to provide useful biophysical activity in mixtures with lipids. Nonetheless, it is significant that the dynamic surface activity of the phospholipid-SAP-35 mixtures studied here was much lower than that of CLSE and surfactant-TA, neither of which contain SAP-35. Moreover, the effects of SAP-35 on phospholipid adsorption, though reproducible, were apparently independent of the hydrophobic or lipid-binding domain of SAP-35. It remains possible that our isolation procedures have altered SAP-35 to give a falsely low assessment of its effects on lipid surface activity. However, as purified here, SAP-35 is completely soluble, binds phospholipids, confers specific calcium-dependent aggregation to SM phospholipids and inhibits phospholipid secretion from Type I1 cells.4 In addition, since SAP-35 is isolated in the absence of detergents, the observed phospholipid binding and biophysical activity should be specific for the SAP-35 structures studied.