Pulmonary surfactant protein-A (SP-A) restores the surface properties of surfactant after oxidation by a mechanism that requires the Cys6 interchain disulfide bond and the phospholipid binding domain.

Reactive oxygen species produced by activated leuko-cytes in the alveolar epithelial lining fluid have been implicated in the inactivation of pulmonary surfactant and the impairment of lung function. Oxidation of bovine lipid extract surfactant (BLES), a therapeutic surfactant, with hypochlorous acid (H-BLES) or the Fenton reaction (F-BLES) led to temporary increases in conjugated dienes and formation of malondialdehyde and 4-hydroxy-2-nonenal. Electrospray ionization mass spectrometry revealed the appearance of lipid hydroperoxides, peroxides, lysophospholipids, and free fatty acids. Captive bubble tensiometer studies of H-BLES demonstrated prolonged adsorption times, film instability at low surface tensions during film compression, and reduced respreadability during film expansion. F-BLES exhibited prolonged adsorption times, a marked effect on increasing compressibility during compression, and a lesser effect on reducing respreadability on expansion. Addition of native bovine or rat surfactant-associated protein A (SP-A) reversed the effects of oxidation on surfactant biophysical properties. Studies using mutant recombinant rat SP-As indicated that an intact carbohydrate recognition domain and disulfide-dependent oligomeric assembly are critical for these effects, but the collagen-like region is not required. We conclude that SP-A can reverse the detrimental effects of surfactant oxidation on the biophysical properties of surfactant, by a mechanism that is dependent on interchain disulfide bond formation and the C-terminal domains of the protein.

Pulmonary surfactant is essential for normal lung function. By lowering surface tension (␥) 1 during inspiration, surfactant reduces the work of breathing. Through its ability to generate very low tensions during expiration, surfactant stabilizes the alveoli at low lung volumes and limits transudation of interstitial fluid into the airspaces. The high spreadability of surfactant also limits differences between adjacent alveolar regions (1,2). In addition to its surface-active properties, surfactant performs host defense and anti-inflammatory functions (3,4).
Surfactant lipids and proteins in the alveolus are exposed to oxidizing conditions, including direct exposure to oxidative air pollutants and to hyperoxia during supplemental oxygen therapy, as well as to endogenous reactive oxygen species from activated neutrophils and macrophages in the subphase. Although surfactant is resistant, to some extent, to oxidation because of the predominance of saturated PL (ϳ30 -45% DPPC), unsaturated PLs represent ϳ55-70% of the weight of surfactant, including ϳ12% with two or more double bonds (8,9). Whereas unsaturated PL species are essential to the fluidity of surfactant, the initial formation of the surface film, and the reincorporation of material into the film during expansion in the breathing cycle (1,2,8), they are also distinctively susceptible to oxidation. Surfactant oxidation and dysfunction likely contributes to the pathophysiology of several diseases including adult respiratory distress syndrome (10,11), asthma (12,13), chronic obstructive pulmonary disease (12), and infections (10). In addition to direct effects on the surface properties of surfactant, lipid peroxidation products can inhibit the synthesis and secretion of new surfactant (14) and reduce the viability of alveolar macrophages, and type II pneumocytes (15).
In vitro studies indicate that the most abundant surfactantassociated protein, SP-A, facilitates the surface tension-lowering properties of surfactant PLs (16 -18) and counteracts protein inactivation of surfactant in vitro (17,19).
Although not yet clearly delineated in vivo, these studies suggest that SP-A may enhance surfactant function in vivo under stress conditions associated with lung injury and proteinaceous pulmonary edema.
The structural domains of SP-A required for surface tension lowering effects have been mapped by targeted mutagenesis, which has revealed the importance of intact oligomeric assembly and the PL binding function of the COOH-terminal domains. The supraquaternary structure of SP-A consists of six trimeric subunits. Each monomer has four distinct domains: a short NH 2 -terminal domain containing one or two cysteines that function to stabilize the octadecamer by forming disulfide bonds between trimers; a collagen-like domain, a neck region, and a globular C-terminal domain. The collagen and neck portions of the monomers combine as triple helices to form trimers, and six trimers associate to form an octadecameric "flower bouquet-like" structure that resembles C1q, the first component of complement. The COOH-terminal domain contains both a carbohydrate recognition domain (CRD), and an overlapping PL binding domain (4,20).
In the present study we hypothesized that SP-A may restore the surface-active ability of surfactant inactivated through oxidation. The effects of SP-A and mutant recombinant SP-As on the biophysical function of bovine lipid extract surfactant (BLES) were tested in the captive bubble tensiometer (CBT) before and after oxidation with Fenton chemistry or hypochlorous acid. Our results demonstrated that oxidative modification of surfactant results in loss of surface properties that can be restored to preoxidation levels by the addition of SP-A, through a mechanism that requires the carbohydrate-recognition domain and intact oligomeric assembly.
In Vitro Oxidation-Reaction mixtures composed of 10 mg/ml surfactant lipids, plus Fenton reagents, HOCl/ Ϫ OCl or controls in buffer A (0.150 mM NaCl, 2 mM Tris-HCl, 1.5 mM CaCl 2 , pH 7.4) were incubated at 37°C in a shaking water bath for 24 h. Hypochlorous acid was purchased as sodium hypochlorite (NaOCl, 10 -13%) from Sigma. Hypochlorite concentration was determined spectrophotometrically immediately before use by diluting the NaOCl stock solution 1:10 in 1 M NaOH and applying a molar extinction coefficient for pH 12 of 43.6 mol Ϫ1 cm Ϫ1 at 240 nm. Treatment with HOCl/ Ϫ OCl was carried out at a final concentration of 0.5 mM at pH 7.4. Because hypochlorous acid has a pK a of 7.46, the term HOCl/ Ϫ OCl used in this paper represents the mixture of hypochlorite and its corresponding free acid. For oxidation by the Fenton reaction, BLES was incubated with 0.65 mM FeCl 2 , 0.65 mM sodium EDTA, and 30 mM H 2 O 2 in buffer A. These conditions approximate those used by a number of other investigators who conclude that they represent high physiological ranges (21)(22)(23).
Biochemical Analysis-Conjugated dienes formed during oxidation were measured via spectrophotometric monitoring of the absorbance at 235 nm (A 235 ) after exposure to HOCl/ Ϫ OCl or Fenton reagents, as above. Aliquots of the oxidized BLES were diluted to 0.25 mg/ml surfactant and A 235 determined every 10 min for 5.5 h against a buffer blank. Data are expressed as the changes in UV absorbance at 235 nm across time. The lipid peroxide content of the surfactant formed during oxidation was measured by using a lipid peroxidation assay kit BIOXYTECH TM LPO-586 (OXIS International, Inc.), which is based on the formation of a colored adduct that absorbs at A 586 , from the reaction of a chromogenic reagent with either malondialdehyde (MDA) and/or 4-hydroxy-2-nonenal (HNE). Data are expressed as nanomoles of MDA and HNE/mg of surfactant PLs.
Mass Spectroscopy-Molecular species of surfactant PLs were analyzed by electrospray ionization mass spectroscopy (ESI-MS) with a triple quadrupole ESI-MS (model PI 365, Sciex Co., Concord, ON, Canada) using positive mode for PCs and negative mode for PGs employing a slight modification of the method of Postle et al. (13,24). Briefly, the surfactant lipids were extracted with chloroform:methanol (1:1, v/v) by the method of Bligh and Dyer (25). The lower chloroform layers were dried under N 2 and the samples were stored in choroform: methanol (1:1, v/v). Dimyristoyl-PC and dimyristoyl-PG (10 g, 10:1), from Avanti Polar Lipids (Birmingham, AL), were added as internal standards to aliquots of each surfactant preparation containing ϳ80 g of surfactant. Samples were dried under nitrogen and dissolved in 0.5 ml of methanol:chloroform (2:1, v/v) containing 5 mM NaOH. PC molecular species were detected as sodiated adducts under positive conditions, (M ϩ Na) ϩ , whereas PG and PI molecular species were detected as their molecular ions under negative conditions, (M Ϫ H) Ϫ . Spectra were obtained by signal averaging (typically 50 scans, 1 min) and processed by use of software supplied by the manufacturer.
SP-A-Bovine SP-A was extracted from natural pulmonary surfactant and purified by chromatography on mannose columns as previously described by Cockshutt et al. (17). Protein purity was tested by SDS-PAGE (12% gel) under reducing conditions. Rat SP-A was isolated from surfactant obtained from rats exposed to silica, using chromatography on mannose-Sepharose 6B columns. The various rat recombinant SP-As (rSP-A) were prepared as previously described (26 -31) and their properties are listed in Fig. 1. Wild-type recombinant SP-A (wt rec) lacks proline hydroxylation as it is produced in insect cells. The recombinant SP-A ⌬N1-P80 (⌬N1-P80) lacking the NH 2 -terminal and collagenlike regions, is trimeric (31). The rSP-A ⌬G8-P80 (⌬G8-P80) is a collagen deletion mutant containing the NH 2 terminus, the neck region, and the CRD (29). The mutant rSP-A hyp-C6S (C6S), which has Cys 6 replaced by a serine, forms trimers but not larger oligomeric forms (29). The two CRD mutant proteins, rSP-A hyp-E195A (E195A) and rSP-A hyp-D215A (D215A), bear point mutations in amino acids predicted to contribute to coordination of a calcium ion involved in PL and carbohydrate binding (28). Although these SP-As have reduced ligand binding, they are isolated by mannose chromatography and consequently retain some carbohydrate binding activity (Fig. 1).
Captive Bubble Tensiometry-CBT assays were performed in triplicate, using 0.05 mg/ml samples diluted in buffer (0.150 mM NaCl, 2 mM Tris-HCl, 1.5 mM CaCl 2 ). The effect of SP-A was tested by adding natural bovine or recombinant rat SP-A at 5% (w/w) relative to surfactant phospholipids. Samples were incubated in a shaking water bath for 1 h prior to measurements. All the experiments were performed at 37°C except those involving the recombinant rat SP-As because the lack of hydroxyprolines makes these proteins less stable at temperatures above 30°C (32,33). Therefore the effect of the recombinant SP-As was studied at 25°C.
After filling the CBT chamber with a suspension of the desired surfactant in buffer, an air bubble, 5-8 mm diameter, was introduced. The change in bubble shape was recorded to monitor the adsorption of the surface-active materials to the air-saline interface. After (␥ eq , mN/m) was attained, the chamber was sealed, and quasi-static or dynamic compression-expansion cycles were performed. Changes in bubble area were recorded during each individual experiment and the bubble shapes analyzed using custom designed software to calculate the surface tension of the film (34,35).
For quasi-static experiments, the film was allowed to adsorb to ␥ eq before cycling. Quasi-static compression-expansion curves were generated by increasing the pressure in the sample chamber by slowly turning the plunger 5-10°to generate ϳ10% changes in bubble volume per step, and allowing 10 s for equilibration after each step. Five quasistatic cycles were conducted with an inter-cycle delay of 1.5 min on return to the original volume. Minimum surface tension (␥ min ) was established either as the lowest ␥ attainable without bubble clicking (observed as an instantaneous change to a small surface area with a higher ␥) or ␥ near 1.0 mN/m. In the case of F-BLES, compression was discontinued at ϳ50% of the original surface area to avoid breaking the chamber by the high hydrostatic pressure.
For dynamic cycling, the surfactant was allowed to adsorb to near ␥ eq . A quasi-static compression-expansion cycle was conducted to establish the surface area reduction required to attain ␥ min . Dynamic modes were implemented by cycling the bubble between 100 and 110% of the original surface area and the area required to achieve ␥ min at 30 cycles per minute.
The compressibility (C) and respreadibility (R) of the film at a desired ␥ were calculated using {C/R (␥) ϭ 1/A (␦A/␦␥)} for the compression or expansion curve, respectively, where A is surface area and ␦A/␦␥ is the slope at a desired ␥. C refers to the ability of the film to reduce ␥ during surface area reduction. C ϭ 1.0 indicates no reduction in ␥ during compression. R refers to the ability of the film to maintain low ␥s during surface area expansion. This property represents the elasticity of the film, plus the insertion or reinsertion of phospholipid molecules into the expanding film. Insertion occurs through adsorption while reinsertion occurs from the surface-associated surfactant reservoir (2,36). R ϭ 1.0 would indicate no increase in ␥ during expansion. For samples unable to reach surface tension values Յ10 mN/m, ␥ at 10 mN/m was obtained by extrapolation of the curve.
Statistical Analysis-All experiments were performed at least three times with individual freshly prepared samples. Standard error of the mean was obtained from samples of n ϭ 3 or greater. Statistical comparisons were conducted using SPSS Inc. (Chicago, IL) software. Comparisons between two groups were made using paired Student's t test and comparisons between larger groups were conducted by analysis of variants followed by a Tukey post-hoc test. Probability values below 0.05 are considered significant.

Effect of Free Radical Exposure on Surfactant
Phospholipids-Both hypochlorous acid and the Fenton reaction were effective in oxidizing BLES, as shown by increased primary (conjugated dienes, Fig. 2A) and secondary lipoperoxidation products (MDA and HNE, Fig. 2B). The conjugated dienes formed during oxidation ( Fig. 2A) were monitored by reading the absorbance at 235 nm after the exposure of BLES to oxidation. In the absence of oxidants (closed circles), in the presence of Fe 2ϩ /EDTA (closed triangles) or H 2 O 2 (open circles) small, similar increases in A 235 were observed that remained relatively constant over the 2-h period shown in Fig. 2A. The levels did not change further over the remaining 3.5 h of the experiment (data not shown). In the presence of HOCl/ Ϫ OCl 0.5 mM (closed squares) and the Fenton reagents (FeCl 2 :EDTA: H 2 O 2 , 0.65:0.65:30 mM, open triangles), A 235 increased rapidly, reaching maximal levels at 20 -30 min, after which it declined. Higher differences in A 235 were evident with the Fenton reaction compared with hypochlorous acid. Differences between the oxidation reactions and the control incubations disappeared at ϳ60 min for BLES:HOCl/ Ϫ OCl and ϳ90 min for BLES:Fenton. The decrease in absorbance, observed between 90 and 100 min, is consistent with the consumption of the susceptible unsaturated fatty acids and the production of secondary products.
The secondary peroxidation products formed during oxidation were measured after 24 h incubation as the content of MDA and HNE in the surfactant sample (Fig. 2B). The Fenton reaction, which acts through generation of reactive hydroxyl radicals (37), generated 4.5 times more MDA and HNE than the control while there was a 2.0-fold increase after reaction FIG. 1. Recombinant rat SP-As assessed in this study. These SP-As, including the wild-type recombinant SP-A, lack proline hydroxylation because they are produced in insect cells. The recombinant SP-A ⌬N1-P80 (⌬N1-P80) lacking the NH 2 -terminal and collagen-like regions is trimeric. The rSP-A ⌬G8-P80 (⌬G8-P80) is a collagen deletion mutant containing the NH 2 terminus, the neck region, and the CRD. The mutant rSP-A hyp-C6S (C6S), which has Cys 6 replaced by a serine, forms trimers but not larger oligomeric forms. The two CRD mutant proteins, rSP-A hyp-E195A (E195A) and rSP-A hyp-D215A (D215A), bear point mutations in amino acids predicted to contribute to coordination of a calcium ion involved in phospholipid and carbohydrate binding.
with HOCl/ Ϫ OCl. These results are in overall agreement with the kinetic analysis of the conjugated diene formation ( Fig. 2A), where the increased absorbance produced by the Fenton reaction on BLES was approximately double H-BLES and triple the BLES control (BLES versus F-BLES p Ͻ 0.001, BLES versus H-BLES p Ͻ 0.001). These findings are consistent with and expand on previously published results reporting elevations in primary (conjugated dienes) and secondary lipid peroxidation products (thiobarbituric acid-reactive substances) after incubation of various surfactants with in vitro reactive oxygen species generating systems (21,22,38,39).
ESI-MS Analysis-The effects of oxidation on lipid composition were examined using ESI-MASS in positive (Fig. 3, A-C) and negative (Fig. 4, A-C) modes. The lipid profile for BLES PLs before oxidation (Mass spectra, Fig. 3A) was typical of previous reports on lung surfactant (6,7,40). The spectrum is clearly dominated by diacyl-PCs detected as sodiated adducts (M ϩ 22) ϩ , with DPPC (m/z 256) being the major component. The presence of the quaternary ammonia group in PC results in highly efficient detection. PEs at m/z 710 -740 were detectable only to a very limited extent.
Major peaks observed for control BLES at the higher m/z ratios (500 to 850) were 16 It is important to note that ESI-MS analysis determines molecular mass/charge and not PL molecular structure, but interpretation in the context of known surfactant composition strengthens the molecular species analysis in terms of fatty acyl distribution. For example, the ion detected at m/z 808 is probably a mixture of 18:1/18:1-PC and 18:0/18:2-PC because these species have identical molecular masses, but the ion at m/z 756 must be 16:0/16:0-PC because other possibilities, such as 14:0/18:0-PC, are not present in surfactant at appreciable levels (6,7,9,40,41). In addition, although the location of specific unsaturated fatty acids cannot be assigned by ESI-MS, it is known that unsaturated fatty acids predominantly occupy the sn-2 position (9).
Oxidation by either of the systems led to the appearance of new species, indicated by ions that appear between m/z 700 and 900, such as 16:0/peroxy 16:1 PC (m/z 786.8); 16:0/hydroxy 18:1 PC (m/z 798.1); 16:0/peroxy 18:1 PC (m/z 814.8), and 18:0/ peroxy 18:2 PC (m/z 840.9). The emergence of these new species was accompanied in some cases by a decrease in the intensity of the peaks of their corresponding unsaturated PLs ( Fig. 3B; Table I). The Fenton reaction, as was expected, generated greater increases in these hydroperoxide products and larger reductions in the parent species. The effects of hypochlorite on the peroxide formation from PL moieties were weaker.
The most intense peaks in the low mass region of BLES spectra corresponded to lyso-PCs, including 16:0 lyso-PC (m/z 518.5) as well as 18:1 lyso-PC (m/z 544.6), which were detected with low intensity. Oxidation of BLES by HOCl/ Ϫ OCl or Fenton reaction led to increased generation of lyso-PC. That the tendency for the formation of lyso-PL was a consequence of the PL oxidation was reinforced by the fact that the Fenton reaction enhanced the appearance of lyso-PLs to a greater extent than HOCl/ Ϫ OCl.
ESI-MS negative mode analysis of BLES revealed palmitoyl, oleoyl-PG, and DPPG as the major peaks with palmitoyl, palmitoleoyl-PG, stearoyl, linoleoyl ϩ oleoyl, oleoyl-PG, and stearoyl, oleoyl-PG at significant levels (Fig. 4A, Table I). Small amounts of unsaturated species of PI and PS were also detected in negative-mode ESI-MS but not quantified.
BLES oxidation with HOCl/ Ϫ OCl or the Fenton reaction resulted in the appearance of peroxy derivatives of palmitoyl, palmitoleoyl-PG, and palmitoyl, oleoyl-PG. Fenton oxidation also resulted in a decline in dienoic and, to a lesser extent, monoenoic species.
As in the case of the neutral PCs, significant levels of snshortened chain PGs were not detected. Oxidation did result in a relative increase in lyso-PGs, particularly with the Fenton reaction. Of additional interest was the increase in free fatty acids, again to a higher degree with the Fenton reaction. The ESI-MS intensity yield of lyso-PLs and fatty acids relative to PLs under our conditions has not been established, although it is apparent free fatty acids can "fly" to a greater extent than PLs. Consequently, with these particular compounds, these studies provide only a qualitative indication of the underlying chemistry.
Effect of Free Radical Exposure on Surfactant Function-The  Table I. alveolar mechanics during respiration. Changes in bubble shape are analyzed to obtain the surface tension of the film during surface area manipulation and to generate isotherms. Figs. 5-7 represent adsorption times and quasi-static and dynamic isotherms. Oxidation by exposure to either of the two free radical systems resulted in retardation of initial film formation such that times to attain equilibrium ␥ (23 mN/m) were increased from ϳ10 min (control) to ϳ120 min (H-BLES) and ϳ140 min (F-BLES). Addition of SP-A to oxidized BLES improved the adsorption rates to levels similar to the control BLES (Fig. 5).
Although breathing is a dynamic process, quasi-static cycling experiments provide important information about film behavior, by making it possible to study re-spreading and/or re-  Table I. adsorption. During the inter-cycle delay, phospholipids reinsert into the film, promoting a drop in ␥ allowing the film to return to near equilibrium. This phenomenon can be studied only in the quasi-static mode. Quasi-static compression of BLES (control) adsorbed at 0.05 mg/ml results in an initial highly compressible "squeeze-out" plateau to ϳ18 mN/m, followed by a steep decline to near 0 mN/m. Compressibility at ␥ 20 (C 20 ) was 0.078 Ϯ 0.020. As reported previously, control BLES adsorbed at 50 g/ml required ϳ60% surface area reduction to lower ␥ from equilibrium (23 mN/m) to near zero during the first compression (16,35,42).
Upon surface area expansion of compressed BLES films, ␥ increases sharply with a respreadability at ␥ 10 (R 10 ) of 0.011 Ϯ 0.002, which improves above 15 mN/m to a R 10 of 0.080, mean Ϯ S.E. Surface tension remains close to equilibrium during the latter stages of expansion until the original surface area is reached. The difference in respreadability during film expansion relative to compressibility during surface area reduction resulted in a moderate hysteresis. By the fifth quasistatic compression, the squeeze-out plateau was barely evident. Because of this alteration in surface properties, ␥ could be reduced to near zero with ϳ20% surface area reduction. Compressibility at ␥ 10 was improved to 0.009 Ϯ 0.000, although with cycles 1 and 5, there is an increase in compressibility at very low surface tensions, resulting in a slight hook to the isotherm. Upon bubble expansion, ␥ increases sharply, although R 10 was slightly higher than the 0.011 observed during the first cycle. The increase in respreadability at ␥ 10 indicates an improved elasticity of the film, because neither insertion of PL through adsorption from bulk phase nor lipid reinsertion from the surface-associated reservoir should occur at ␥ values below equilibrium (2).
The addition of SP-A, at a wt ratio of 5% PL, resulted in a FIG. 5. Effect of bovine SP-A on the initial film formation by oxidized BLES. Bar graph represents the time required for the film to achieve equilibrium surface tension. All measurements were performed in triplicate at 37°C for 0.05 mg/ml samples reconstituted in buffer A, as indicated under "Experimental Procedures." The effect of SP-A was tested by adding natural bovine SP-A at 5% (w/w) relative to surfactant PLs. Comparisons among the samples were conducted using Tukey's studentized range (highly significant difference) test. Letters represent comparisons among samples. Means with the same letter are not significantly different. slight improvement in film surface properties during the first cycle (Fig. 6, top right, BLES:SP-A). The squeeze-out plateau observed during the initial stage of film compression was reduced so that Ͻ40% surface area reduction was sufficient to reduce ␥ to near zero. Compressibility and respreadability at ␥ 10 were also slightly improved compared with those in the absence of SP-A, and hysteresis was reduced compared with cycle 1 in the absence of SP-A. By the 5 th compression-expansion cycle, little difference could be noted between samples with and without SP-A. These observations reinforce the concept that SP-A is not essential during normal surface activity manipulations. The addition of Fe 2ϩ /EDTA alone, or H 2 O 2 alone, had no observable effect on quasi-static and dynamic profiles for any of the samples studied (data not shown). Incubation of surfactant with HOCl/ Ϫ OCl (Fig. 6, middle panel) or complete Fenton reagents (Fig. 6, bottom panel) caused a reproducible deterioration in surface activity, resulting in significantly higher ␥ max and ␥ min . With H-BLES, the squeeze-out plateau during the first compression was similar to or smaller than control. However, C 10 was significantly increased to 0.029 and the lowest ␥ min attainable was elevated to ϳ5 mN/m and increased further during cycling. Attempts at lowering ␥ further through surface area reduction resulted in "clicking." Clicking refers to an almost instantaneous decrease in bubble surface area, resulting in an increase in ␥ that may be due to the loss of PL from the bubble surface (34). During the initial surface expansion with H-BLES, ␥ increases sharply to ϳ25 mN/m. This indicates that, compared with control, film elasticity and respreading are extremely poor. Whether the decrease in slope at ␥ ϳ 25 mN/m results from PL respreading or adsorption of surfactant from the bulk phase cannot be concluded directly from the expansion curve. However, it is clear from the adsorption studies that adsorption is slow at this particular ␥, suggesting that respreading is the major mechanism. As a result of the poor respreadability, ␥ rises to Ͼ40 mN/m. Poor insertionreinsertion of PL molecules into the expanding film results in a large hysteresis. Because of the poor PL reincorporation, compression 5 initiates with ␥ Ϸ 35 mN/m. After a relatively sharp decline, a small squeeze-out plateau initiating at ␥ ϭ 25 mN/m is observed, with a shape similar to that of the first compression cycle. The C 10 was 0.0048 Ϯ 0.001, which is superior to that noted with the first compression. However, the film remains unstable and cannot achieve ␥s below 10 mN/m. R 10 during surface area reduction increases to 0.003 and an improvement in lipid incorporation is noted at ␥ Ϸ 15 mN/m, so the respreadibility at ␥ 20 improves to 0.037 Ϯ 0.01. However, little further change is observed in respreadability at ␥ Ϸ 25 and ␥ increases almost linearly to ϳ40 mN/m, again suggesting poor PL insertion even above equilibrium surface tension. Nevertheless, because of the improved characteristics during film compression and expansion, minimal hysteresis is evident.
Addition of bovine SP-A to the HOCl/ Ϫ OCl-treated sample leads to a marked improvement in film performance. The length of the squeeze-out plateau is markedly reduced so that ␥ decreases more readily during compression (Fig. 6, middle,  right panel). The presence of SP-A improves the ability to achieve low surface tensions, although with the initial compression there is a slight tendency toward higher compressibility. Inclusion of SP-A also improves respreadability at ␥ 10 to 0.001 and PL reinsertion increases markedly at ␥ ϭ 15, so that ␥ max is lowered from Ͼ40 to Ͻ30 mN/m at maximum surface area. The overall compression-expansion pattern approaches that of BLES:SP-A (Fig. 6, upper right), although hysteresis remains somewhat increased.
The ability of SP-A to restore surface activity with H-BLES is also evident for the 5 th quasi-static cycle, which has a shape close to that of BLES:SP-A, except that enhanced PL incorporation into the monolayer during expansion is not observed until equilibrium ␥ is attained, leading to a slight increase in hysteresis.
Subjecting BLES to the Fenton reaction (Fig. 6, lower left panel) results in a greater deterioration in surface activity than HOCl/ Ϫ OCl. The squeeze-out plateau is broadened considerably so that a 50% reduction in surface area results in lowering ␥ to only ϳ10 mN/m. Respreadability is not hampered as much as with the HOCl/ Ϫ OCl treatment so that, although film expansion initiates from a higher ␥, the final ␥ attained remains below 40 mN/m. This ␥ is still significantly higher than that observed with BLES alone. By the 5 th cycle, there is a decrease in the squeeze-out plateau so that ␥ min remains at ϳ10 mN/m, despite being initiated at ␥ ϭ 38 mN/m. However, due the high compressibility during adsorption and poor PL insertion/reinsertion above equilibrium, hysteresis remains high.
As in the case of H-BLES, addition of bovine SP-A to F-BLES results in a large improvement in surface activities (Fig. 6,  lower right panel). The sample recovers its ability to attain low surface tensions with ϳ60% surface area reduction in the first quasi-static cycle and improved respreadability is also observed. A further small improvement in surface properties is observed during the 5 th cycle. Nevertheless, hysteresis remains somewhat higher than observed with control BLES:SP-A.
Dynamic cycling of surfactant films in the CBT more closely mimics normal respiratory compression-expansion within the alveolar airspace. For these experiments, adsorbed films were cycled at a rate of 30 cycles/min between 100 and 110% of the original bubble surface area and the area required to achieve ␥ min during the first quasi-static compression. Overcompression of the bubbles at low ␥s was avoided. The data were averaged from three separate experiments, and isotherms for cycles 1 and 21 are shown in Fig. 7. In general, the overall results obtained from dynamic cycling were similar to those obtained in quasi-static cycling. However, the ␥ max levels observed during film expansion were higher. This was more evident with the hypochlorous oxidized samples; the ␥ max values attained during expansion increased to ϳ60 mN/m, and a marked deterioration was observed on consecutive dynamic cycles. These observations are consistent with the reduced respreadability observed with HOCl/ Ϫ OCl during quasi-static cycling. Addition of natural bovine SP-A to the oxidized BLES restored the BLES ability of both oxidized samples to attain surface tensions near zero with less surface area reduction during the film compression and remain close to equilibrium during the film expansion (middle and bottom right panels). The overall pattern of compression-expansion isotherms of these samples approached those for non-oxidized BLES.
Domains of SP-A Required for Enhancing the Surface Properties of Oxidized Surfactant-The structural basis for the restorative effects of SP-A on the surface activity of oxidized BLES was assessed using a set of six mutant recombinant rat SP-As produced in insect cell systems. These include the wt rec protein, and mutant recombinant SP-As containing a C6S substitution that disrupts the critical NH 2 -terminal interchain disulfide bond (C6S), a deletion of the collagen domain (⌬G8-P80), a deletion of the collagen-like domain and the NH 2terminal segment (⌬N1-P80), and point mutations in residues that are critical for coordination of calcium, carbohydrate, and PL binding (E195A and D215A). Cross-linking analyses and gel filtration chromatography under physiologic ionic strength conditions have revealed that the ⌬N1-P80 protein is trimeric (31), the ⌬G8-P80 and C6S proteins are predominantly trimeric and hexameric, and the wt rec, E195A, and D215A proteins are composed of trimers, hexamers, and nonamers (not shown).
Bar graphs summarizing the adsorption times required to achieve equilibrium ␥, generated for control BLES and oxidized BLES alone and in the presence of native rat and the six rSP-As, are presented in Fig. 8. Overall effects of SP-As on BLES oxidized by HOCl/ Ϫ OCl or the Fenton reaction were very similar. Native rat SP-A was comparable with bovine SP-A in improving the adsorption time of oxidized BLES (Figs. 5 versus  8). The recombinant wild type also decreased the adsorption rates of oxidized BLES, albeit less effectively than the natural SP-As, from ϳ110 min for control BLES to ϳ35 min for F-BLES and ϳ45 min for H-BLES. The collagen deletion mutant ⌬G8-P80 (31) was the most active in restoring a rapid adsorption time to oxidized BLES, whereas ⌬N1-P80 (31), which differs from the ⌬G8-P80 only in the absence of the NH 2 -terminal segment and interchain disulfide bonds, was ineffective in promoting initial film formation. Isolated disruption of the C6S interchain disulfide bond (29) also reduced restoration of adsorption, as did point mutations replacing amino acids thought to be important in calcium and carbohydrate binding (E195A and D215A) (43).
Investigation of the ability of rat native and recombinant SP-As to reverse the effects of HOCl/ Ϫ OCl and the Fenton reaction on BLES was extended to expansion-compression cycling of surfactant films under quasi-static and dynamic modes (Fig. 9, A and B). As in the case of film formation, the ability of the various SP-As to counteract the effects of BLES oxidation on ␥ max and ␥ min by either system showed a number of simi- larities and so these results will be described together. Recombinant wild-type and ⌬G8-P80 were as effective as native rat SP-A in restoring the ability of oxidized BLES to achieve ␥s below 5 mN/m during the initial film compression. This activity was maintained throughout successive cycles, undergoing either quasi-static or dynamic manipulation. However, the ability of these SP-As to restore ␥ reduction to near 0 mN/m was better with H-BLES than F-BLES.
The ⌬N1-P80 was much less effective than ⌬G8-P80 in restoring surface properties to oxidized BLES, highlighting the importance of the NH 2 -terminal segment and interchain disfulfide bond to this activity. Furthermore, the ␥ min and ␥ max values increased during quasi-static and dynamic cycling with ⌬N1-P80, as occurs with oxidized BLES by itself. This effect was more evident with H-BLES where ⌬N1-P80 exhibited only a small effect on ␥ min during quasi-static cycling. Nevertheless, samples containing ⌬N1-P80 demonstrate significantly better ␥ min and ␥ max values with F-BLES. C6S restored ␥ min to low values with H-BLES but had less effect on ␥ max particularly with dynamic cycling. The lack of effectiveness of C6S in restoring ␥ max was also apparent in experiments using F-BLES.
The E195A and D215A mutations reduced the restorative activity of SP-A on the surface properties of F-BLES and H-BLES during quasi-static or dynamic cycling, although both mutant recombinant SP-As improved the ability of the oxidized surfactant to attain lower ␥ min and ␥ max values. E195A protein tended to show better overall effects, but the differences were small.
Comparison of the percentage of surface area compressions required to achieve ␥ near zero during quasi-static cycling (Fig.  10, A and B) provides a highly discriminative method for grading the abilities of native and rSP-As to restore the surface activity of oxidized BLES. In some cases such as oxidized BLES control, it was necessary to extrapolate the experimental values to obtain the values plotted. Addition of native rat SP-A restored the % area reductions required for H-BLES to attain values similar to control BLES, but recovery was less complete with F-BLES. The rat recombinant wild-type and the collagendeletion mutant ⌬G8-P80 were as active as rat SP-A in reestablishing the ability of BLES oxidized by either system to achieve low ␥s during quasi-static cycling. With H-BLES little or no protective effect was observed with rSP-As containing NH 2 -terminal segment mutations that affect oligomeric assem- bly (⌬N1-P80, C6S) or wild-type CRD (E195A, D215A). Except for ⌬N1-P80, all of these mutants showed intermediate abilities to restore deterioration induced by the Fenton reaction. Collectively, these data indicate that the CRD and the Cys 6 interchain disulfide bond are critical for the salutary effects of SP-A on the surface activity of oxidized surfactant.

Effect of Reactive Oxygen Species on Pulmonary Surfactant-
Considerable evidence accumulated through extensive in vitro and in vivo investigations has shown that reactive oxygen species such as HOCl/ Ϫ OCl, H 2 O 2 , superoxide radical, and singlet oxygen radical, released in the alveolar lining layer by activated inflammatory cells, can react with and inhibit pulmonary surfactant (44,45). The present studies confirm that exposure of BLES, a therapeutic modified natural surfactant, to HOCl/ Ϫ OCl and the Fenton reaction at concentrations employed by others with other surfactants and considered in the high physiological range (21)(22)(23) generates primary conjugated dienes ( Fig. 2A), secondary lipoperoxidation (MDA, HNE, Fig.  2B) products despite the relatively low levels of polyunsaturated PL. In contrast to low and high density lipoproteins, where prolonged lag phases in diene formation are observed (46,47), peroxidation of BLES occurs rapidly. This can be attributed to the lack of endogenous antioxidants, including ascorbic acid and ␣-tocopherol, which are removed from BLES during processing (9).
ESI-MS Analysis of Control and Oxidized BLES-ESI-MS is a highly sensitive technique that has recently been applied to the study of PL molecular species in surfactant from a number of animal species, including human (reviewed in Ref. 6). Ap-plying ESI-MS to control and oxidized BLES demonstrate increases in the levels of hydroxide, hydroperoxide, and other truncated PCs and PGs, consistent with previously reported studies with low and high density lipoproteins (48 -50). Higher levels of PL peroxidation products were observed with the Fenton reaction than with HOCl/ Ϫ OCl, but with both systems the amounts were relatively low (Table I). Interestingly, ESI-MS detected lyso-PC, lyso-PG, and free fatty acids (Figs. 3 and 4). Precise ion yields for these derivatives must still be established, but it is evident that these lipid species are markedly elevated over control BLES, particularly with the Fenton reaction. Lyso-PCs can also be generated through a phospholipase A-independent mechanism during myeloperoxidase-mediated PC degradation (38).
The positive and negative ESI-MS spectra presented here demonstrate that reaction of BLES with either HOCl/ Ϫ OCl or Fenton reaction results in the degradation of PCs and PGs. Phospholipids containing two or more double bonds appear particularly susceptible, but it is evident that monoenoic species are also reactive. Fenton reactions generated higher levels of degradation products, consistent with a difference between this system and hypochlorous acid. In contrast, alkyl-ether-PCs were relatively stable, consistent with previous reports on the ability of these lipids to resist oxidation. It has been suggested that plasmenyl ether-PCs may serve to protect surfactant in the alveolar space (51). Taken together with the results from positive mode ESI-MS, it is clear that both hypochlorous acid and the Fenton reaction attack unsaturated PCs and PGs. These observations provide further evidence indicating fundamental differences in the manner in which HOCl/ Ϫ OCl and the FIG. 10. Effect of recombinant rat SP-As on the percent of surface area reduction required to achieve surface tensions near zero of oxidized surfactant during cycling. The bar graph depicts the percentage of surface area compression required for the films to attain a minimum surface tension near 0 from 20 mN/m during quasi-static cycling. Extrapolated compressions of over 100% were estimated for those films that could not attain low minima during compression. Comparisons among the samples were conducted using Tukey's studentized range (highly significant difference) test. Letters represent comparisons among samples within a cycle. Means with the same letter are not significantly different.
Fenton reaction destroy the ability of pulmonary surfactant to maintain surface active properties consistent with normal lung function.
Interfacial Effects of Surfactant Oxidation and Reversal by SP-A-Exposing BLES to reactive oxygen species led to a marked inhibition in surface active properties assessed as (i) a reduced ability of the reacted surfactants to adsorb rapidly to the surface, (ii) elevated ␥ min and ␥ max during quasi-static and dynamic cycling, (iii) increases in the surface area reduction required to achieve ␥ min , and (iv) a marked progressive deterioration in these functions upon cycling, relative to control nonreacted BLES.
The two systems produced markedly different effects during compression-expansion cycling. HOCl/ Ϫ OCl treatment did not affect film compressibility during quasi-static compression but induced film instability at low ␥s. It also resulted in poor respreadability below and above ␥ eq during film expansion. The Fenton reaction greatly increased compressibility at ␥ 10 and also limited PL insertion:reinsertion above equilibrium ␥. Similar overall differences were observed during dynamic cycling. The basis for this dissimilarity in behavior is presently unknown. However, it appears possible that they could arise from differential effects on the two low molecular weight hydrophobic proteins (SP-B and SP-C) and the phospholipids. Studies on the oxidation of low density lipoproteins and high density lipoproteins have shown that HOCl/ Ϫ OCl is effective in oxidizing proteins, whereas the Fenton reaction more specifically oxidizes lipids containing unsaturated fatty acyl groups (48,52). Addition of Fenton-oxidized palmitoyl, linoleoyl-PC, or palmitoyl, linoleoyl-PG to BLES hampers the ability of this clinical surfactant to reduce ␥ to low values during film compression (data not shown). These preliminary results are consistent with both lipid and protein oxidation being involved in disrupting surface activity. Further studies are required to determine the contribution of oxidation to the surfactant proteins SP-B and/or SP-C as opposed to PL oxidation.
Addition of bovine SP-A virtually abolished the effects of oxidation, leading us to conclude that (i) SP-A promotes a decrease in the compressibility of surfactant films by limiting the extent of the squeeze-out plateau and by improving (reducing) compressibility of surfactant films during compression at low ␥s; (ii) SP-A stabilizes surface films at low ␥s, thereby extending the ability to attain ␥ near 0; (iii) SP-A enhances respreadability of surfactant lipids during film expansion, thereby maintaining ␥ max near equilibrium ␥. These contributions of SP-A to surfactant function occur during both quasistatic and dynamic cycling. We anticipate that SP-A improves surfactant function of normal, control surfactant in the same manner, but that these effects are slight and consequently are more readily demonstrable with BLES oxidized with HOCl/ Ϫ OCl or the Fenton reaction.
Role of SP-A Domains in Reversing the Effects of BLES Oxidation-Native and wt rec rat SP-As were capable of reversing most of the effects of oxidation on adsorption, although the recombinant protein was less effective. Overall effects on adsorption with H-or F-BLES showed a striking similarity, consistent with a common mechanism for restoration of surface activity by SP-A that is independent of the method of oxidation. This effect of SP-A appears to be highly specific. SP-D, a collectin that shares many properties with SP-A had no effect on the surface activity of BLES, H-BLES, or F-BLES (data not shown).
Rat SP-A, wt rec SP-A, and ⌬G8-P80 were effective in improving the compressibility of films of oxidized surfactant and their stability at low ␥s. These SP-As also restored film respreadability. The results show that the CRD and the Cys 6 interchains disulfide bond of SP-A are critical for restoring pre-oxidation surface properties to H-BLES and F-BLES, and that the collagen-like region of SP-A is dispensable for the function.
Previous studies using the pulsating bubble surfactometer demonstrated that ⌬G8-P80 improved the ability of BLES at low concentrations to achieve low ␥ min values on the pulsating bubble surfactometer (29). However, surfactant isolated from transgenic mice engineered to express ⌬G8-P80 in the lungs of SP-A null (SP-A Ϫ/Ϫ ) mice showed high equilibrium ␥, ␥ min , and ␥ max values and were more susceptible to inhibition by serum proteins than surfactant from normal controls (32). Pressure volume relationships of lungs from SP-A Ϫ/Ϫ and SP-A Ϫ/Ϫ,⌬G8-P80 mice were similar to those of SP-A Ϫ/ϪrSP-A mice. The discrepancy between the in vivo and in vitro properties of surfactants containing ⌬G8-P80 may be related to the temperatures at which the biophysical surfactant assays were performed (37 versus 25°C).
In general, the CRD and NH 2 -terminal point mutations studied demonstrated partial activities relative to wt rec in restoring the biophysical activities with either oxidized BLES. A notable exception was ⌬N1-P80, which retained reasonable activity in achieving low ␥ min during cycling with H-BLES, but poorer activity with F-BLES. Altering C6S and removing the entire N-terminal region disrupts oligomeric assembly of trimeric subunits, resulting in a predominance of trimers (53). These results strongly suggest that the oligomeric structure is important for the salutary effects of SP-A on surface activity of BLES oxidized either by HOCl/ Ϫ OCl or the Fenton reaction.
Although initially controversial, considerable evidence has accrued indicating that the CRD of SP-A possesses both the carbohydrate-binding and PL-binding motifs (20). The two CRD mutants E195A and D215A used here reportedly lack the ability to bind or aggregate PL liposomes and exhibit reduced binding of carbohydrate beads or calcium (28). Nevertheless, some carbohydrate binding activity must remain, because the SP-As are purified on the basis of calcium-dependent binding on mannose columns. CRD mutants E195A and D215A possessed intermediate activity in reducing the effects of oxidation by either system. However, these rSP-As tended to show poorer activity toward H-BLES, reinforcing the notion that hypochlorous acid and the Fenton reaction have different effects. The intermediate effects suggest that some PL binding activity is also retained. PL-binding and carbohydrate-binding motifs of SP-A have not been precisely mapped and crystallographic analysis will be necessary to define these regions. Nevertheless, it is clear that considerable overlap occurs between these two regions, because the tandem mutations E195Q,R197D dissociate the lipid binding and aggregating functions of SP-A, and convert SP-A carbohydrate binding specificity from a preference for mannose to galactose (26).
It will be important to determine whether the effects observed here derive from direct interactions between SP-A and surfactant lipids, particularly DPPC, or involve cooperative interactions with other surfactant proteins. Although SP-A has only slight effects in enhancing DPPC adsorption by itself, it can interact with DPPC monolayers to improve compressibility with such films (54,55). SP-A also improves compressibility with DPPC:cholesterol spread films, possibly by reducing DPPC/cholesterol interactions and generating aggregated DPPC domains. SP-A binds DPPC ripple phase and, in the presence of DPPC, can form long strands that could contribute to tubular myelin formation (53,56). SP-A binds to the liquid condensed phase of spread surfactant monolayers (57). Because SP-B and SP-C preferentially locate in liquid expanded phases, this would suggest that SP-A and SP-B could occupy different regions of the surfactant monolayers. These observations would indicate that SP-A could have independent effects on oxidized surfactant films.
In contrast, there is considerable evidence that SP-A can have cooperative effects with SP-B. SP-A can enhance surfactant adsorption, reduce film compressibility, increase respreadability, and limit the surface area reductions required to attain low ␥s with surfactant extracts and model surfactant systems containing SP-B but not with systems containing only SP-C (5,16). SP-A and SP-B, but not SP-C, are required for reconstitution of tubular myelin and SP-A can contribute to the formation and maintenance of large aggregates (5,58). Whether SP-A reversal of the effects of surfactant oxidation requires SP-B or SP-C must be determined through reconstitution studies. In addition to counteracting effects of surfactant PL oxidation, SP-A can also counteract inhibitory effects of non-surfactant proteins (17,18). Studies comparing SP-A domain and structural requirements for these two prosurfactant activities are currently in progress.
Summary and Conclusions-The present study examined the biochemical and biophysical effects of free radical exposure on a lipid extract surfactant, BLES. BLES was oxidized using hypochlorous acid or the Fenton reaction. After oxidation the surfactants displayed biochemical modifications, and primary and secondary lipoperoxidation products were identified. ESI-MS analysis of oxidized BLES revealed the presence of hydroxy and hydroperoxy derivatives of unsaturated PCs and PGs, especially with the Fenton reaction. Oxidation led to an increase in lyso-PC and lyso-PG and free fatty acids, again to a greater extent with the Fenton reaction. Surface activity was markedly impaired. Oxidation by either system greatly hampered surfactant adsorption and the ability to attain low ␥s during film compression. Hypochlorous acid treatment had little effect on compressibility at ␥ 20 and ␥ 10 but increased film instability, limiting the ability to attain ␥s below 7 mN/m during cycling. Exposure to HOCl/ Ϫ OCl markedly hampered respreadability during film expansion, particularly at ␥ 20 and at ␥s above equilibrium. Exposing BLES to the Fenton reaction led to increased compressibility at ␥ 10 such that the oxidized surfactants could not reduce ␥ to below 10 mN/m, even with large compression ratios. These results show that, despite similar apparent effects on PLs, BLES oxidation by HOCl/ Ϫ OCl or Fenton reaction produces distinct effects on the surface properties of surfactant films. Regardless of the difference in the effects on biophysical function, addition of bovine or rat SP-As restored the ability of oxidized surfactants to attain ␥s near zero during film compression and to spread during film expansion. The mechanisms by which SP-A rescues surfactant function are not understood. Mutagenesis studies demonstrate that an intact CRD and Cys 6 disulfide bond are critical for restoration of the surface activity of oxidized surfactant and indicate that the phospholipid binding properties and oligomeric assembly of SP-A are essential for this function.