INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Pulmonary surfactant covers the alveolar surface, reduces the surface tension at this interface and thereby prevents alveolar collapse. Three major density fractions of extracellular surfactant material have been identified: ultraheavy, heavy, and light (1). The ultraheavy and heavy fractions, comprised in large surfactant aggregates (LA) (2), consist mainly of lamellar bodies, tubular myelin, and large multilamellar vesicles, and are highly surface active. The light fraction, which is also referred to as small surfactant aggregates (SA), appears ultramorphologically as small, unilamellar vesicles (1) and shows poor surface activity both in vitro (2) and in vivo (3). Pulse-chase and turnover studies suggest a metabolic cycle from LA to SA within the alveolar space (4): the LA fraction provides the interfacial monolayer and the SA fraction is subjected to reuptake and recycling by alveolar type II cells and to phagocytosis by alveolar macrophages. A decrease in the percentage of LA and an increase in that of SA were recently noted in experimental lung injury (5) and under conditions of the acute adult respiratory distress syndrome (ARDS) (6).

Exposing bronchoalveolar lavage fluid (BALF) to cyclic variation of the air-fluid interface in vitro has been shown to result in the conversion of LA to SA in a time-dependent manner (1, 2, 4). Loss of surface tension-reducing properties under these conditions was ascribed to the proportional increase in the SA at the expense of the LA fraction. Using rabbit BALF for cycling in vitro, we found that, in parallel with the LA-to-SA transition, a marked impairment of surface activity also occurs in the remaining LA fraction, and that both events appear to be intimately linked to a dramatic loss of the hydrophobic surfactant apoprotein-B (SP-B) in the LA subtype. Interestingly, we also found a corresponding correlation between reduction of SP-B in the LA fraction, loss of surface activity of this functionally decisive fraction, and increased appearance of SA at the expense of LA in BALF samples from patients with ARDS requiring mechanical ventilation. Such disturbances of surfactant homeostasis may thus contribute to the alveolar instability and impairment of gas exchange encountered in severe respiratory failure.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Materials

Mouse monoclonal antibody against porcine SP-B (8B5E, cross-reactive with human, bovine, rabbit, or canine SP-B) was kindly provided by Y. Suzuki of the Chest Disease Research Institute, Kyoto, Japan. This antibody recognizes both dimeric and monomeric SP-B (monomeric SP-B will be detected with 10% efficacy as compared to dimeric SP-B). Human and rabbit SP-B dimers were isolated from pooled BALF of healthy human volunteers and of healthy rabbits through LH60-chromatography as described previously (6), yielding a protein of > 95% purity. Bovine serum albumin (BSA) was purchased from Paesel and Lorei GmbH (Frankfurt/Main, Germany). Microtiter plates (Polysorp with certification) were obtained from Nunc (Wiesbaden, Germany). Biotinylated sheep antimouse antibody was obtained from Amersham Buchler GmbH (Braunschweig, Germany). Avidin-biotin-horseradish-peroxidase (AB-complex) was obtained from Dako GmbH (Hamburg, Germany). 2,2'-Azino-bis-(3-ethyl-benzthiazoline-6-sulfonic acid) (ABTS) was obtained from Boehringer GmbH (Mannheim, Germany). Tween-20, primuline ("direct yellow 59"), and various fatty acid methyl ester (FAME) standards were purchased from Sigma (Munich, Germany). Silica 60 plates and all other chemicals were from Merck (Darmstadt, Germany).

Procedures

BALF from ARDS patients (BALF_ARDS). The study included 10 patients (seven male and three female) aged 48 ± 5 yr (mean ± SD) who had ARDS defined according to the recent American European Consensus Conference (7). Underlying diseases were sepsis (n = 6), severe pneumonia (n = 2), shock (n = 1), and pancreatitis (n = 1). Bronchoscopy was performed for diagnostic purposes from 12 to 72 h after the beginning of mechanical ventilation; the mean Pa_O2/FI_O2 ratio at this time was 148 ± 23 mm Hg. A fiberoptic bronchoscope was wedged into one segment of the lingula or the right middle lobe and 10 aliquots of 20 ml each of sterile saline were injected and immediately removed by gentle suction (recovery: 55 to 70%). The lavage fluid was filtered through sterile gauze, and the filtrate was collected on ice and immediately centrifuged at 200 × g (4° C, 10 min) to sediment cellular material. The supernatant was frozen in liquid nitrogen and kept at 85° C until further analysis. Bronchoscopy and lavage were performed in the same manner on 11 healthy control subjects. The study was approved by the local ethics committee, and informed consent was obtained from the closest relatives in each case.

Preparation of a rabbit BALF pool (BALF_rab). Rabbits of either sex were killed by intravenous application of a lethal dose of pentobarbital/ketanest. A catheter was immediately placed into the trachea and the lungs were lavaged three times with 50 ml saline. After filtration of the lavage fluid through sterile gauze and sedimentation of cells (200 × g, 4° C, 10 min), supernatants originating from 15 animals were pooled on a single day and aliquoted in 5-ml fractions in polypropylene tubes (15 ml Falcon blue cups, No. 2096; Becton Dickinson, Meylan, France) with continuous stirring. The aliquots were then stored at 85° C until further experimental use.

In vitro cycling of rabbit BALF. Aliquots of the pooled rabbit BALF were thawed and upon reaching room temperature were fixed on a rotating disk in an incubator at 37° C. As described by Gross and colleagues (4), cycling was then performed at a rate of 32 ×/min for various time periods (10, 20, 30, 60, 120 and 240 min), with changes in surface area of  8-fold (1.65 cm² in the vertical versus 13 cm² in the horizontal direction). Uncycled controls were sham incubated for 240 min at 37° C.

Surfactant subtype separation. Unless otherwise stated, separation of LA from SA was done through high-speed centrifugation (48,000 × g, 4° C, 1 h) using a Sorvall centrifuge (SS34 rotor; DuPont, Bad Homburg, Germany). This centrifugation scheme differed slightly from that used by Veldhuizen and colleagues (40,000 × g, 15 min, 4° C) (2, 8). The LA-containing pellets were resuspended in saline with 3 mM CaCl₂ for biophysical and biochemical characterization. The relative LA content was assessed by relating the fraction of pelleted phospholipid (PL) to the total amount of PL. In additional experiments the technique of Magoon and coworkers (9) was used, with a 1,000 × g centrifugation done first (20 min, 4° C), and followed by a 60,000 × g centrifugation of the 1,000 × g supernatant (60 min, 4° C), and a final 100,000 × g centrifugation of the 60,000 supernatant (16 h, 4° C). The 1,000 × g centrifugation was done in a Sorvall centrifuge (SS34 rotor; DuPont). The 60,000 × g and 100,000 × g centrifugation steps were done with an Omega 70 ultracentrifuge (W60 and W44A1 rotors; Heraeus, Hanau, Germany). The 1,000 × g, 60,000 × g, and 100,000 × g pellets were resuspended for biochemical and biophysical analysis in saline with 3 mM CaCl₂.

Analysis of phospholipid content and phospholipid classes. Original BALF or isolated SA or LA fractions were extracted according to Bligh and Dyer (10), and organic phosphorus was determined in each fraction as described (6). High-performance thin-layer chromatography (HPTLC) was used for separation of PL classes as previously described (6). Briefly, 30 µg of extracted phospholipids or seven different concentrations of a standard phospholipid mixture containing eight different phospholipids were applied to Silica 60 plates, with a Linomat IV applicator (Camag, Muttenz, Switzerland), separated with a mobile phase of chloroform:methanol:acetic acid:distilled water (50: 37.5:3.5:2; vol/vol), stained with molybdenum blue reagent, and quantified by densitometric scanning at 700 nm, using a TLC scanner II (Camag). The detection limit was 200 ng PL. The variance of this technique depended on the relative amount of PL, with values averaging 0.745% for phosphatidylcholine (PC), 1.16% for phosphatidylglycerol (PG), and 7.34% for lysoPC (6).

Determination of neutral lipid profile. Neutral lipids (NL) were first separated from PL by TLC on Silica 60 plates, with chloroform:methanol:H₂O (65:35:6; vol/vol) serving as the mobile phase. NL were visualized with primulin: spots were then scraped off and NL were extracted with chloroform. In a second HPTLC step on Silica 60 plates, with hexane:diethylether:formic acid (80:20:2, vol/vol) as solvent, the NL fraction was completely separated into mono-, di-, and triglycerides; cholesterol; cholesterol ester; and free fatty acids. After staining with primulin, quantification was done by densitometric scanning at 254 nm with coprocessing of seven reference concentrations of each NL class on the same plate. Under these analytical conditions, the influence of the variability in fatty acid chain length and degree of saturation was found to be negligible.

Characterization of fatty acid profiles. The total lipid fraction as well as the isolated, predominant PC and PG classes of PL in original BALF and in the SA and LA fractions of pooled BALF were treated with 2 N hydrochloric acid/methanol for 10 h at 100° C. The resultant FAME were isolated with hexane and purified by TLC on Silica 60 plates (with toluene as the mobile phase). The FAME were then analyzed with a Carlo Erba Fractovap 2150 gas chromatograph equipped with a capillary column (CP-Sil 88, 50 m × 0.25 mm I.D.; Chrompack) and flame ionization detection, with the temperature set at 200° C.

Quantification of SP-B. SP-B was quantified through a solid phase adsorption-enzyme linked immunosorbent assay (ELISA) technique recently described in detail (11). Briefly, human or rabbit SP-B dimer and SP-B contained in original BALF LA and SA fractions was transferred and fixed to Polysorp microtiter plates through the use of different organic solvents (diisopropylether:butanol 3/2, vol/vol; trifluoroethanol). Free binding sites were saturated with bovine serum albumin (BSA) and SP-B was detected with monoclonal anti-SP-B antibody (8B5E) and a second monoclonal, biotinylated antimouse antibody. Horseradish peroxidase-labeled avidin-biotin complex was used for signal amplification, ABTS served as the substrate, and reading was done at 405 nm.

Assessment of surface activity. Surface activity of the resuspended surfactant pellets was measured at a constant PL concentration of 2 mg/ml by means of a pulsating bubble surfactometer as previously described (6). The technique provided readings of the surface tension after 12 s of film adsorption (_ads) and the surface tension at a minimum bubble radius after 5 min of film oscillation (_min). In separate experiments, LA obtained from cycled rabbit BALF (120 min) were recombined with purified dimeric SP-B. For this purpose, 18 µg of dimeric rabbit SP-B (in CHCl₃/CH₃OH, corresponding to the tenfold greater quantity of SP-B that should have been present in an uncycled LA sample) were transferred to a glass vial, dried under nitrogen, and resuspended with the cycled LA sample (600 µg PL total) by ultrasonic treatment for 1 min (50 W, 25 kHz). The sample was then incubated for 30 min at 37° C prior to measurement of surface activity with the bubble surfactometer.

Electron microscopy. Cycled (240 min) or sham-incubated, uncycled rabbit BALF was centrifuged at 48,000 × g as described earlier. For electron microscopy, a recently described method was adopted (12). For this purpose, the 48,000 × g pellets were resuspended in 2 ml fixation buffer (2.5% glutaraldehyde, 1% tannic acid in sodium cacodylate, pH 7.4). The tubes were then again ultracentrifuged at 100,000 × g for 10 min in a swinging bucket rotor (Omega 70, S40 rotor) and were stored overnight in the fixation buffer. After washing with 0.1 M sodium barbital buffer, pH 7.4, the pellets were fixed in 1% OsO₄ (in the same sodium barbital buffer).

Following rapid dehydration in cold graded acetone solution, the pellets were embedded in Epon. Sections were stained in 5% uranyl acetate and alkaline lead citrate, and were examined in a transmission electron microscope (Leo-906; Zeiss, Oberkochen, Germany).

Statistics

All data are given as mean ± SE. Statistical significance for differences between controls and ARDS patients was obtained by application of the H-test followed by the Mann-Whitney U-test. For assessing the statistical significance of correlations, Student's t test for paired samples was used.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

In Vitro Cycling of BALF_rab

The 48,000 × g centrifugation step resulted in the separation of LA from SA in both uncycled as well as cycled rabbit BALF. Figure 1 shows representative electron-microscopic pictures of 48,000 × g pellets of uncycled and cycled (240 min) rabbit BALF. As can be seen in these pictures, the pelleted surfactant material from uncycled rabbit BALF consisted mainly of large, multilamellar vesicles and tubular myelin (Figure 1, left panel ). In 48,000 × g pellets from cycled rabbit BALF (240 min), the tubular myelin fraction almost completely disappeared, but the bulk of surfactant still consisted of large, multilamellar vesicles (Figure 1, right panel ). Small unilamellar vesicles were not observed in either of the samples. Thus, the appearance of the 48,000 × g (1 h) centrifugation preparation was similar to that in previous studies done with centrifugation at 40,000 × g (15 min) (2, 8).

View larger version (166K): [in this window] [in a new window]   View larger version (141K): [in this window] [in a new window]   View larger version (153K): [in this window] [in a new window]   View larger version (154K): [in this window] [in a new window]   Figure 1.   Electron-microscopic examination of 48,000 × g pellets of rabbit BALF, showing morphologic appearance of 48,000 × g pellets of uncycled (left panel, A and B) and cycled (240 min; right panel, C and D) rabbit BALF. Original magnification: top panels, ×40,000 (A, C ); bottom panels, ×60,000 (B, D). For details see METHODS.

The percentage of LA in fresh BALF_rab and in noncycled (sham-incubated) controls was ~ 85% (Figure 2). In vitro cycling resulted in a rapid conversion of LA to SA, with the relative amount of LA decreasing to ~ 23% within 120 min, with subsequent plateauing of values (Figure 2). Concurrently, marked loss of surface activity of the remaining LA fraction was noted, despite study at standardized quantities of phospholipid (2 mg/ml PL). LA obtained from noncycled BALF_rab reached near zero minimum surface tension, with adsorption surface tension values approximating 25 mN/m (Figure 2). Upon cycling, _min of the LA fraction increased to  22 mN/m within 120 min, with subsequent plateauing of values, and _ads increased to  45 mN/m. The decrease in the proportional appearance of LA and the loss of surface activity of the remaining LA fraction were significantly correlated (Figure 3). The SA fraction, whether obtained from fresh BALF or from cycled samples, never reduced min values below 20 mN/m (data not given in detail).

View larger version (21K): [in this window] [in a new window]   Figure 2.   Influence of in vitro cycling of rabbit BALF on LA-to-SA conversion and LA surface activity. The percentage of LA in relation to total PL (left panel ), and the surface activity of standardized quantities (2 mg/ml PL) of the LA fraction (right panel ), are given for various periods of in vitro cycling (10 to 240 min). Time zero represents controls, sham-incubated for 240 min in the absence of cycling. Mean ± SE values of five or more independent experiments each are depicted. No error bar is given when range of SE falls within symbol.

View larger version (19K): [in this window] [in a new window]   Figure 3.   Correlation between minimum surface tension and relative LA content of cycled rabbit BALF (left panel ) and BALF from ARDS patients (right panel ). The surface activity was assessed by employing standardized quantities of the respective LA fractions (2 mg/ml PL). For rabbit BALF, the corresponding mean ± SE values from five or more independent experiments are shown for each cycling period. For ARDS samples, the mean of two measurements for each patient is given. No error bar is displayed when range of SE falls within symbol. **p < 0.01.

In order to further characterize this loss of surface activity and the responsible LA subfraction upon in vitro cycling, we additionally separated uncycled and cycled rabbit BALF according to the technique of Magoon and colleagues. As can be seen in Table 1, such separation resulted in a similar appearance of  82% LA in uncycled rabbit BALF (the sum from the 1,000 × g and the 60,000 × g pellets). Both the 1,000 × g and the 60,000 × g pellets displayed excellent surface activity (_min around 0 mN/m), whereas the 100,000 × g pellet showed only limited surface activity (_min  20 mN/m). Upon cycling for 240 min, a decrease in the relative PL content of the 1,000 and 60,000 × g pellets was observed. This decline was reflected by a dramatic loss of surface activity in both of the recovered LA subfractions (see Table 1).

With the 48,000 × g single centrifugation step, no differences were noted in the phospholipid profile of LA and SA obtained from cycled and noncycled BALF_rab (displayed for the major PL classes PC and PG in Table 2). Similarly, the fatty acid profiles of PC and PG did not differ in the LA and SA of the original lavage fluid, and also remained unaffected upon in vitro cycling (indicated for palmitic acid [16:0] and oleic acid [18:1] in Table 2): high percentages of palmitic acid in PC ( 70% of all fatty acids) and PG ( 55% of all fatty acids) were noted throughout. In contrast, the fatty acid profile of the whole lipid fraction (PL and neutral lipids) differed between SA ( 40% palmitic acid residues) and LA ( 60% palmitic acid residues, Table 2). Similarly, the NL profile and NL/PL ratio in LA and SA in uncycled samples of BALF_rab showed marked differences (Table 3). The NL/PL ratio was ~ 0.05 in LA but ~ 0.40 in SA. Within the NL of the LA subtype, cholesterol and free fatty acids represented the largest fractions, whereas triglycerides and free fatty acids were the predominant classes in the NL of the SA fraction. As the relative proportion of SA increased during in vitro cycling, the fatty acid and NL profiles of the SA came to resemble those of the LA fraction (Tables 2 and 3).

In vitro cycling of BALF_rab caused a time-dependent decrease in the total BALF SP-B content as detected by immunoreactivity, which approached  40% of the initial values within 120 min (as evident from the sum of the LA and SA SP-B values in Figure 4). This decrease was particularly prominent in the LA fraction, in which hardly any SP-B remained detectable after 120 min of cycling (Figure 4). In control experiments done with a cycling period of 240 min, less than 1% of the initially provided SP-B could be recovered from the walls of the plastic tubes upon rinsing with propanol (data not given in detail), thus making less likely a significant loss of SP-B due to adsorption to the plastic surface. As related to the PL content of the SA fraction, there was also a decrease in SP-B in this fraction, whereas the total SP-B content of the SA fraction remained roughly constant, presumably owing to cycling-dependent transit of lipids from LA to SA. There were strict correlations between the decrease in the proportional appearance of LA and the decline in SP-B in the remaining LA fraction, as well as between the decline in SP-B and the loss of surface tension-reducing properties of this fraction (Figures 5 and 6).

View larger version (26K): [in this window] [in a new window]   Figure 4.   Dependency on cycling of SP-B content in surfactant aggregate subtypes of rabbit BALF. SP-B in the LA and SA fractions was measured with an ELISA, each fraction being reconstituted to the original fluid volume after separation by centrifugation. The left scale gives absolute SP-B concentrations in the reconstituted volume, the right scale gives the SP-B content in relation to the total PL in each fraction. Data are given for various periods of in vitro cycling (10 to 240 min); time zero represents controls sham-incubated for 240 min in the absence of cycling. Mean ± SE values from five or more independent experiments are depicted. No error bar is given when range of SE falls within symbol.

View larger version (19K): [in this window] [in a new window]   Figure 5.   Correlation between the SP-B content of LA (related to the PL content of this fraction) and the proportional appearance of LA (related to total PL) in cycled rabbit BALF (BALFrab; cycling times between 0 and 240 min) and in BALF of ARDS patients (BALFARDS). For rabbit BALF, the corresponding mean ± SE value from five or more independent experiments is given for each cycling period. For ARDS samples, the mean of two measurements for each patient is given. No error bar is shown when range of SE falls within symbol. **p < 0.01.

View larger version (19K): [in this window] [in a new window]   Figure 6.   Correlation between SP-B content of LA (related to the PL content of this fraction) and the minimum surface tension of the LA fraction in cycled rabbit BALF (BALFrab; cycling times between 0 and 240 min) and in BALF obtained from ARDS patients (BALFARDS). For rabbit BALF, the corresponding mean ± SE from five or more independent experiments is shown for each cycling period. For ARDS samples, the mean of two measurements for each patient is given. No error bar is shown when range of SE falls within symbol. **p < 0.01.

To address the role of SP-B in the impairment of surface activity, we reconstituted rabbit LA, obtained after 120 min of in vitro cycling, with purified SP-B at a tenfold concentration as measured with the SP-B ELISA in the uncycled but sham-incubated control. Whereas the cycled LA displayed a marked impairment in surface activity (_min ~ 16 mN/m), secondary reconstitution with SP-B produced a far-reaching restoration of the surface activity (_min values near 0 mN/m) of this cycled LA (see Table 4).

Lavage Fluids from ARDS Patients

In healthy controls, the LA fraction comprised 65.7 ± 4.8% of the total PL (mean ± SE, n = 11), whereas the proportional appearance of LA in BALF_ARDS was only 46.6 ± 7.1% of the total PL (n = 10; p < 0.05). The _min values obtained upon investigating standardized quantities of these LA fractions were 0.45 ± 0.22 mN/m in controls and 14.43 ± 2.59 mN/m in BALF_ARDS (p < 0.001); the corresponding _ads values were 22.24 ± 0.55 mN/m and 35.85 ± 2.63 mN/m (p < 0.01). As similarly described for cycled rabbit BALF, the percentage of LA in BALF_ARDS and the _min values of this subtype were significantly correlated (Figure 3). The SP-B content of the LA fraction was 4.94 ± 1.03% (as related to the PL in this fraction) in healthy controls, and 0.38 ± 0.15% in BALF_ARDS (p < 0.001). The decrease in the percentage of LA and the decline in SP-B in the remaining LA fraction were significantly correlated, as was also the decline in SP-B and the loss of surface tension- reducing properties of this surfactant aggregate fraction (Figures 5 and 6).

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Our studies confirm the findings in previous investigations indicating that in vitro cycling leads to conversion of LA to SA in a time-dependent manner, with the resulting SA having poorer surfactant properties both in vitro (2) and in vivo (3). We now extend this observation in showing that the LA-to-SA transit is paralleled by a loss of surface activity of the remaining LA fraction itself. Whereas we noted near zero minimum surface tension and rapid adsorption (criteria for excellent surfactant function [13]) for the LA of noncycled control BALF, both the surface tension-reducing and adsorption capacities are dramatically reduced in the LA fraction of cycled BALF, in strict dependency on the cycling time. A nearly identical interrelation between a decreased percentage of LA and loss of the biophysical properties of this functionally important surfactant fraction was noted for BALF samples from ARDS patients. We focused our attempts to analyze the biochemical changes underlying the loss of biophysical function on the surfactant lipids and apoproteins of the surfactant aggregate fractions.

Lipid Changes

In accordance with earlier studies by Magoon and colleagues employing differential centrifugation (9), the PL profile in our study did not differ between original BALF and the LA and SA of uncycled rabbit BALF, and it did not change upon cycling. This was also true for the fatty acid composition of the major PL classes (PC and PG). In contrast, the NL profile of uncycled BALF_rab differed substantially in the LA (predominance of cholesterol and free fatty acids) and SA (predominance of mono- and triglycerides and free fatty acids) fractions. Although increasing its relative proportion in SA during in vitro cycling, the NL profile increasingly resembled that of the LA fraction, thus reflecting the transition of NL from the LA to the SA fraction. On the basis of this analysis, alterations in the surfactant lipid composition are therefore very unlikely to explain the loss of surface activity of the LA fraction during the process of cycling.

SP-B

In vitro cycling resulted in a time-dependent decrease in immunologically detectable SP-B in total rabbit BALF. This finding was even more striking when we analyzed the isolated LA fraction, in which SP-B became virtually undetectable with 60 to 120 min of cyclic variation of surface area, whereas no decrease in SP-B was noted in sham-incubated controls. The decay in the SP-B content of the LA fraction was significantly correlated with both the decrease in the LA/SA ratio and the loss of surface activity of the remaining LA fraction, and corresponding correlations were also noted for BALF samples from ARDS patients. In accordance with this observation, Veldhuizen and associates, using sodium dodecylsulfate-polyacrylamide gel electrophoresis (SDS-PAGE), observed reduced SP-B antigens signals upon in vitro cycling (180 min) of canine BALF fractions (2). In contrast to our finding, these authors found no SP-B in the SA fraction of canine BALF upon gel electrophoresis, but did find considerable amounts of SP-B in the LA fraction upon in vitro cycling (separated by 40,000 × g centrifugation). Our data would suggest that a decrease in the SP-B content of LA during cycling might be responsible both for the formation of SA from LA and for the loss of surface activity of the remaining LA pool. Such reasoning is consistent with known features of SP-B, since this apoprotein is essential for the integrity of tubular myelin and large aggregates (8, 12), and inhibition of SP-B by admixture of anti-SP-B antibodies with natural surfactants inhibits biophysical properties of these surfactants in vitro (14) and in vivo (15). Indeed, reconstituting cycled rabbit LA (120 min) with a tenfold excess of purified rabbit SP-B fully restored the surface activity of this surfactant fraction, thus making very likely a pivotal role of SP-B during the cycling process.

How can cyclic variation in the air-liquid interface result in such a dramatic decrease in SP-B? It has been suggested that the conversion of LA to SA is brought about by a serine- active carboxylesterase contained in the LA fraction (16). In view of the wide range of substrate specificities of carboxylesterases (they may also act as amidases), it will have to be determined in future studies whether SP-B is truly the substrate of the surfactant convertase. In this case, a large increase in surface area, as now used for experimental purposes, would lead to the adsorption of lipids and presumably of hydrophobic apoproteins from complex LA structures to the air-liquid interface, which might render SP-B accessible to convertase attack. Loss of SP-B activity might then result in a failure of the component lipids to reassemble in highly organized LA structures upon surface reduction and reentry into the subphase, thereby leading to reduced surface activity of the LA fraction and to enhanced transit of lipids and other surfactant components into the SA fraction.

The present study was limited in that only SP-B was investigated, and not SP-C or SP-A, owing to the lack of antibodies against the rabbit versions of these apoproteins. In canine BALF, a preceding study (2) found SP-A to be unaltered upon cycling, but SP-C has never been investigated under these conditions on a quantitative basis. Moreover, we may not say whether the loss of SP-B immunoreactivity reflects proteolytic scission of SP-B or some type of conformational change in the epitope targeted by the monoclonal antibody used. Additionally, the  8-fold variation in surface area used for rapid conversion of LA to SA in in vitro studies is far beyond the range of changes in alveolar surface area that occur under physiologic conditions. It should, however, be kept in mind that because of the severe loss of gas-exchange area in ARDS, the remaining alveolar space is overextended and overventilated, a mechanism that has been suggested as contributing substantially to the vicious cycle of ventilator-dependent lung injury in ongoing ARDS ("baby lung" concept; [19]). In addition, increased proteolytic activity has repeatedly been noted in lavage samples from ARDS patients (20, 21). It was indeed an intriguing finding in the present investigation that the main features of the in vitro studies were fully reflected by corresponding changes in BALF samples from ARDS patients. SP-B-related loss of LA integrity and function may thus contribute to impaired surfactant function, alveolar instability, and subsequent abnormalities in gas exchange in patients with ARDS.