Introduction

Abstract Introduction Materials and Methods Results Discussion References

The primary cause of morbidity and mortality in cystic fibrosis (CF) is respiratory failure as a consequence of chronic suppurative lung disease. The CF airway is peculiarly susceptible to bacterial colonization with a range of organisms, most notably Staphylococcus aureus, Haemophilus influenza, and Pseudomonas aeruginosa. The fundamental reason for this underlying susceptibility to infection remains unresolved. Respiratory epithelium from CF patients with the F508 mutation is reported to facilitate bacterial adherence to a greater degree than epithelial cells from normal subjects (1), and reintroduction of a normal cystic fibrosis transmembrane conductance regulator (cftr) gene restored more normal bacterial adherence. Altered sodium and chloride concentrations in the periciliary fluid in the airways may also affect the activity of defensins, thereby interfering with the host defense (2). However, none of these defects explain why in CF there is an unprecedented host immune response to the bacterial colonization, which in turn causes progressive lung damage. This suggests that there is not only a susceptibility to lung infection but also a defect in modulation of the host immune response.

The presence and potential physiological function of pulmonary surfactant in airways as well as in alveoli has been demonstrated by electron microscopy and by analysis of tracheal washings. Important roles proposed for lung surfactant in addition to prevention of alveolar collapse on expiration include maintenance of patency of terminal bronchioles (3), promotion of ciliary beat frequency (4), and rheological effects on mucus viscosity. Individual components of surfactant may also exert considerable direct immunomodulatory properties. For instance, surfactant phospholipid inhibits T-lymphocyte proliferation (5) and decreases cytokine production by alveolar macrophages (6). In contrast, the hydrophilic surfactant proteins A (SP-A) and D (SP-D) enhance many aspects of lung host defense responses to bacterial infection. It is therefore attractive to hypothesize that defects in the surfactant phospholipid constituents, combined with additional abnormalities in hydrophilic surfactant proteins, explain much of the immunopathology of the lung problems in CF.

Previous studies of sputum or bronchial secretions have reported an abnormal composition and impaired surface tension properties of lung surfactant in CF (7, 8). Increased total phospholipid concentrations in such samples were accompanied by significant reductions in the fractional contents of the surface active components disaturated phosphatidylcholine (DSPC) and phosphatidylglycerol (PG). However, the extensive contribution of leukocyte membrane phospholipid to mucus and sputum in CF subjects with lung infection limits the value of these analyses as indicators of surfactant composition. Moreover, the thin layer chromatography (TLC) techniques employed were inherently too insensitive for the analysis of small volumes of bronchoalveolar lavage fluid (BALF) recovered by fiberoptic bronchoscopy.

In contrast, analysis of BALF by high performance TLC showed no significant increase in total phospholipid content in CF subjects, but did show a decreased fractional content of phosphatidylcholine (PC) and PG with elevated phosphatidylinositol (PI) (9). Additionally, the DSPC content of BALF in infants and young children with CF was decreased in those with lung infection (10). However, the osmium tetroxide oxidation technique used in this study is prone to error and measures all disaturated and some monounsaturated phospholipid species in addition to the major surface active component dipalmitoyl PC (PC16:0/ 16:0) (11).

There are two conflicting reports of the SP-A content of BALF in CF. SP-A was increased in those infants identified by neonatal screening for CF who also had evidence of lung infection (10). By contrast, when measured in older children with more established lung disease, the SP-A content of BALF was decreased (9). There have been no previous reports of SP-D concentration in CF.

Consequently, in this study we have addressed two major aims. First, we have examined by detailed structural analysis the nature and extent of any alteration to surfactant phospholipid composition in CF. Second, concentrations of SP-A and SP-D were measured in BALF to determine any potential contribution of dysfunction of hydrophilic surfactant proteins to the impaired inflammatory response to bacterial lung infection in CF. One major technical advance, employed in this study for the first time, is the use of electrospray ionization mass spectrometry (ESI-MS) as a novel approach to analysis of phospholipid composition. Such ESI-MS measurements are exquisitely sensitive and rapid, and have provided detailed structural information of membrane compositions (12, 13). The availability of this technology enabled direct quantification of surface active phospholipid components of BALF from as little as 1 ml of BALF.

	Materials and Methods

Abstract Introduction Materials and Methods Results Discussion References

Study Population

BALF was collected from children undergoing fiberoptic bronchoscopy for diagnostic reasons, including 13 children with cystic fibrosis, 11 children with a variety of underlying problems associated with lung infection (four had delayed resolution of pneumonia, three had gastroesophageal reflux associated with various congenital abnormalities, two had bronchiectasis of unknown cause, and two were investigated for unexplained chronic cough), and 11 children with suspected structural abnormalities of the airways and no discernible infection (seven had laryngo- or tracheomalacia, one had a unilateral vocal cord palsy, one had a bronchial branching defect, and two had no detectable abnormality). These children, without significant infection or inflammatory lung disease, were regarded as a control group for the purposes of this study. The presence, type, and extent of any respiratory infection was determined by standard microbiological assessment of colony-forming units. The age ranges and microbiological assessment, where appropriate, of these three groups of children are summarized in Table 1. This table shows that the age ranges of both infected groups of children were reasonably similar, but that of the control group was significantly lower. As expected, P. aeruginosa was the most common bacteria found in the children with CF.

BALF Sample Collection and Processing

All bronchoscopies were performed for clinical purposes as outlined in Table 1. Parents gave informed written consent for the procedure, which routinely includes BAL and the retention of a lavage aliquot for research purposes. The bronchoscopies were performed with the Olympus BF3C30 fiberoptic bronchoscope (Key Med, Southend-on-Sea, UK), which has a 3.6-mm outer diameter and a 1.2-mm suction channel. The procedure was conducted under the influence of local anesthesia with heavy sedation using midazolam and pethidine, and in general followed the procedure described by Raine and Warner (14). After the bronchoscope was inserted through one nostril, it was guided into the airway and preferentially wedged into the right middle lobe bronchus, unless clinical factors dictated lavage in another lobe. In infants younger than 3 yr of age, lavages of 3 × 10 ml were instilled and aspirated, and in infants more than 3 yr of age, lavages of 3 × 20 ml. These specimens were immediately transferred into a tissue culture grade centrifuge tube and stored on ice for processing within 30 min of collection. The safety of this procedure has now been well established with an accumulated experience from 51 European centers of 2,231 BALs in 12 mo (15). Individual lavage fractions were centrifuged separately at 400 × g × 10 min to remove cells for differential leukocyte analysis of bronchial and bronchoalveolar compartments, pooled and stored at 80°C for analysis.

Extraction and Analysis of Total Phospholipid

Total phospholipid concentration of BALF was determined as inorganic phosphate using Bartlett's method (16). Briefly, lipid was extracted from an aliquot of BALF using chloroform/methanol according to Bligh and Dyer (17) and dried under N₂ gas, and distilled water was added until a volume of 60 µl was reached. Inorganic phosphate was measured after digestion of organic compounds at 180°C for 40 min with 60% perchloric acid (wt/vol), and addition of 5 µl of 30% hydrogen peroxide (wt/vol) to oxidize any remaining carbon. Standard curves of dimyristoylphosphatidylcholine (PC14:0/14:0) and dipotassium hydrogen orthophosphate were constructed between 0 and 60 nmol. Phosphate concentration was calculated from the absorbance at 830 nm after further addition of ammonium molybdate and Fiske and Subbarow reagent and heating at 100°C for 10 min (16).

A second aliquot of BALF containing approximately 50 nmol phospholipid was then extracted with chloroform/ methanol for ESI-MS analysis. PC14:0/14:0 (10 nmol) and dimyristoyl PG (PG 14:0/14:0; 2 nmol) were added before extraction as internal standards, after confirming that neither compound was detectable as endogenous components in BALF samples. Just before analysis, total lipid extracts were dissolved in 25 µl methanol/chloroform (2:1, vol/vol) containing 5 mM NaOH. Dissolved samples were centrifuged at 13,000 × g × 1 min to remove any particulate matter before injection into the ESI-MS.

Electrospray Ionization Mass Spectrometric Analysis of BALF Phospholipid

Molecular species of phospholipid were analyzed on a Quattro II triple quadrapole mass spectrometer fitted with an electrospray interface (Micromass UK Ltd., Manchester, UK). Samples (2 µl) were introduced by Rheodyne valve injection into the methanol/chloroform/water (80:10:10, vol/vol) mobile phase, which was pumped at 10 µl/min into the capillary inlet of the mass spectrometer. Inlet temperature was maintained at 100°C and the inlet solvent stream was nebulized using N₂ gas. Spectral information was collected under conditions of positive and negative ionization on consecutive scans. Scan time was 4 s, resolution was 0.1 mass units, and data were collected over the 2-min period of sample introduction. Capillary and cone voltage settings were, respectively, 3 kV and 48 V in positive mode and 2.5 kV and 77 V in negative mode. For the results presented here, all data were collected at the MS1 detector after single quadrapole analysis. Controlled fragmentation of selected mass ions by tandem MS/MS analysis ions with collection of data at the MS2 detector was performed to confirm identities of individual species (results not shown) (12, 13).

Spectral data was obtained over the mass:charge (m/z) range 400 to 1,000, which, for singly charged phospholipid species, corresponded to molecular mass. PC and sphingomyelin species were preferentially detected as their sodiated adducts under positive ionization conditions (M + 22), whereas PG and PI species were preferentially detected under conditions of negative ionization (M  1). The optimal negative ionization conditions for PG and PI detection resulted in a much lower signal response for PE species. Because of the large relative concentration of PC in these samples, a variety of mass peaks under negative ionization were derived from PC (m/z = 718, 768), but these did not interfere with quantification of PG and PI species.

Spectra were collected as continuum data onto a computer using MaxLynx software (Micromass), over the region of maximal total ion current response (typically 1 to 2 min), and were combined and integrated after transformation to area centroid format. For each molecular ion, the contribution of the ¹³C isotope effect was calculated using an iterative macro program on a Microsoft Excel spreadsheet. Under these conditions, the signal intensity response of the mass spectrometer to the largest component of these spectra was linear with concentration after correction for the response to the internal standard PC14:0/14:0. Typical examples of ESI-MS analysis of total lipid extracts of a BAL sample under conditions of positive and negative ionization are given in Figures 1A and 1B, respectively. Contributions from phosphatidylserine and phosphatidylethanolamine to the negative ionization spectra were minor. These phospholipid classes were not included in the ESI-MS analysis both to simplify data presentation and because they have no significant proposed role in surfactant function. Consequently, phospholipid classes (PC, PG, and PI) data were calculated as the sum of their respective individual species, and expressed relative to each other.

View larger version (36K): [in this window] [in a new window]   Figure 1.   Analysis of BALF phospholipid by electrospray ionization mass spectrometry. A chloroform:methanol total lipid extract from BALF was dissolved in 25 µl of methanol:chloroform (2:1, vol/ vol) containing 5 mM NaOH. An aliquot (2 µl) was introduced into the electrospray interface of a Quattro II mass spectrometer. The representative averaged spectrum in A was obtained under positive ionization conditions and displays the PC molecular species composition. Alternate scanning under negative ionization conditions during the same data acquisition gave the spectrum in B, which displays predominantly PG and PI molecular species compositions.

Where sample amounts permitted, molecular species of PC in BALF were also analyzed by reverse phase high performance liquid chromatography (HPLC) for comparison with the ESI-MS results. Intact PC species were resolved on an Apex I ODS column (Jones Chromatography, Hengoed Glamorgan, UK), and detected and quantified by postcolumn fluorescence derivatization with 1,6-diphenyl-1,3,5-hexatriene as previously described (18).

ELISA for SP-A and SP-D

Native human SP-A and SP-D were purified by maltose agarose affinity-chromatography of BALF obtained from alveolar proteinosis patients (19). Polyclonal antibodies to SP-A were raised in both chickens and rabbits, and to SP-D in rabbits. Concentrations of SP-A and SP-D were measured using sandwich enzyme-linked immunosorbent assays (ELISAs). Microtiter plates (Limbro; ICN Biomedicals Ltd., High Wycombe, Bucks, UK) were coated with either chicken antihuman SP-A immunoglobulin G (IgG) or rabbit antihuman SP-D IgG (10 µg/ml in 35 mM Na₂CO₃, pH 9.6) at 4°C overnight. The nonspecific binding sites were blocked with TBS-NTC (50 mM Tris; 50 mM NaCl; 2 mM CaCl₂; 0.05% [v/v] Tween-20, 0.05% [w/v] NaN₃, pH 7.4) containing 1 mg/ml bovine serum albumin (BSA) for 1 h at 37°C. Each assay plate included a standard curve generated with purified human SP-A (10 to 1,000 ng/ml) or SP-D (1.6 to 1,000 ng/ml) and duplicate serial dilutions of each BALF sample. Samples and standards were incubated for 2 h at 37°C, followed by a further incubation for 2 h at 37°C with biotinylated antihuman SP-A or SP-D (100 µl, 50 µg/ml). After addition of streptavidin-alkaline phosphatase conjugate (Sigma Chemical Co. Ltd., Poole, Dorset, UK) diluted 1:10,000 in TBS-NTC buffer containing 1 mg/ml BSA for 1 h at room temperature, plates were washed with TBS-NTC buffer and then incubated with p-nitrophenyl phosphate substrate for 30 min at 37°C. The reaction was stopped by addition of 1 N NaOH and absorbance of each well measured at 405 nm. Correlation coefficients of standard curves were generally 0.90 to 0.95.

Protein Analysis

Total protein concentration was measured using the phenol:Ciocalteau reagent (20).

Statistical Analysis

Statistical evaluation of the differences between the subject groups was made using the Mann-Whitney U test.

	Results

Abstract Introduction Materials and Methods Results Discussion References

Total phospholipid concentration in the postcell supernatant of BALF from children with cystic fibrosis was marginally but not significantly higher than that measured in children with no significant lung pathology (Table 2). This lack of increase in BALF phospholipid concentration was not due to selection of a group of CF patients with less severe lung disease. All CF subjects underwent the bronchoscopy procedure for diagnostic reasons, and only one of these 13 CF subjects had no evidence of current lung infection (Table 1). This result agrees with a recent report of phospholipid analysis in BALF from young CF subjects, which also demonstrated no increased concentration compared with a control group of infants (10). When calculated as a percentage composition of the sum of their concentrations, the contributions of PC, PG, and PI were not altered compared with the control group of children in either the infection or CF groups (Table 2). There was a tendency for PG to be lower in CF, but this was not significant.

When sufficient sample was available (n = 5), we compared the PC molecular species composition of BALF analyzed by ESI-MS and by reverse phase HPLC (18) (Figure 2). Although results from the two methodologies were not identical, this comparison demonstrated a comparable pattern of molecular species compositions. The greater resolution and sensitivity of ESI-MS (Figure 1A) resulted in more PC species being quantified than was achieved by HPLC analysis, but the five major PC species were reported in equivalent proportions by both techniques. Importantly, there were reasonable correlations between HPLC and ESI-MS analyses of the fractional concentrations of PC16:0/16:0 and PC16:0/18:1 measured in the same samples (Figure 3).

View larger version (33K): [in this window] [in a new window]   Figure 2.   Comparison of analysis of BALF PC by HPLC and by electrospray ionization mass spectrometry. Results are presented as means + standard deviations of analysis of the same samples (n = 5) by both techniques.

View larger version (9K): [in this window] [in a new window]   Figure 3.   Correlation of analysis of individual PC species by HPLC and ESI-MS. Results are presented as percentages of total PC for (A) dipalmitoyl PC (PC16: 0/16:0) and (B) palmitoyloleoyl PC (PC16:0/18:1).

Analysis of purified surfactant preparations by ESI-MS demonstrated more than 20 molecular species of PC (Figure 1A), but the 11 species reported in Table 3 were the most abundant and could be detected in all samples measured. One point to emphasize is that single quadrapole ESI-MS analysis only resolves molecular mass, which, in terms of phospholipid structure, defines the number of carbon atoms and double bonds in the molecule. For example, the sodiated ion detected at m/z 808 is probably a mixture of PC18:1/18:1 and PC18:0/18:2 because these species have identical molecular masses, but the ion at m/z 756 must be PC16:0/16:0 because the only alternative, PC14:0/18:0, is not present in surfactant (21). The results presented in Table 3 demonstrate clearly that that there were no significant alterations to the PC molecular species compositions of BALF from either the infection or CF groups compared with the control group.

The molecular species compositions of both PG and PI in BALF were very different from those of PC (Figure 1B, Table 4). These analyses are the first full reports of PG and PI molecular compositions of human surfactant. Analysis of purified surfactant preparations detected at least 17 molecular species of both PG and PI. However, because of their considerably lower concentration in BALF compared with PC (Table 2), only seven PG species (Table 4) and four PI species (Table 5) were routinely quantified in clinical samples. This low concentration of anionic phospholipids in BALF precluded any direct comparison of their analysis by ESI-MS and HPLC. However, analysis of surfactant purified by density gradient centrifugation from pooled BALF from adult lung demonstrated a preponderance of PG16:0/18:1, PG18:0/18:1, and PG18:1/18:1 by both techniques (results not shown). In contrast to its predominance in the PC composition of BALF, the contribution of the 16:0/16:0 species to the spectrum of PG (m/z = 723) or PI (m/z = 809) (Figure 1B) was very low. Instead, PG and PI were characterized largely by monounsaturated and diunsaturated molecular species, with the same three or four species contributing about 80% of the total composition in each case (16:0/18:1, 18:1/18:1 + 18:0/18:2, 18:0/ 18:1). This result contrasts with a previous report of rabbit lung surfactant compositions, which, on the basis of a higher fractional content of the 16:0/16:0 species in PG compared with PI, suggested that PG and PI were derived from distinct CDP:phosphatidate substrate pools (22). Our results provide little support for this hypothesis in human lung.

The PI results provide good evidence for relatively little contamination of these BALF samples with phospholipid derived from inflammatory cell membranes. The 18:0/20:4 species represents about 70% of human neutrophil PI (our unpublished observations), and gross contamination of BALF with such cell membranes would lead to a large increase in the 18:0/20:4 component of surfactant PI. PI18:0/20:4 content tended to increase in the CF group, consistent with increased inflammation, but the effect was modest and not significant. As with the PC compositions, there were no significant differences in either the PG (Table 4) or PI (Table 5) molecular species compositions between any of the subject groups.

In contrast to the unaltered phospholipid composition of BALF from CF children, concentrations of hydrophilic surfactant proteins in BALF were greatly decreased in CF (Table 6). Although there was no difference in median SP-A concentration between the control and infection groups, this value was significantly decreased in the children with CF (P < 0.001). SP-A concentration was below the ELISA detection limit in one of the 13 CF children, but there was considerable overlap between SP-A values for control and CF children (Figure 4A). Similarly, median BALF SP-D concentration was significantly decreased (P < 0.001) in children with CF compared with children without respiratory infection (Table 6), but not with children in the infection group. There was less overlap in values between the various groups for SP-D than for SP-A; the decreased SP-D concentration in the CF children compared with all other conditions was dramatic, with 7 of the 13 CF children having values for SP-D below the ELISA detection limits (Figure 4B). We felt that it was important to establish that the extent of these decreases of SP-A and SP-D in CF was not due in part to low recovery of surfactant in those BALF samples. Consequently, values for SP-A and SP-D were also calculated relative to total BALF phospholipid (Table 6). However, such correction for surfactant phospholipid content accentuated the differences between the CF and other groups. Additionally, the decreased concentrations of SP-A and SP-D were not due to an overall decrease of protein content of BALF in CF. Total protein in BALF from the CF children was marginally higher (median 1.91, 95% CI 1.64 to 2.13 mg/ml, n = 11) than in the control group (median 1.64, 95% CI 1.48 to 1.94 mg/ml), but this difference was not significant.

View larger version (11K): [in this window] [in a new window]   Figure 4.   Concentration of hydrophilic surfactant proteins in BALF samples. Results are presented relative to total phospholipid (ng protein/nmol phospholipid) for (A) SP-A and (B) SP-D. Open symbols for children in the infection and cystic fibrosis groups denote samples in which active infection with P. aeruginosa was confirmed by microbiologic analysis.

The possibility that SP-D might preferentially bind to inflammatory cells in BALF from CF subjects was addressed by immunoblot analysis of SP-D distribution in cell pellets and supernatants. Although trace amounts of SP-D could be detected in all cell pellets, it was present at no higher concentration in the CF subjects, despite a considerably higher inflammatory cell count in BALF (results not shown). Given the increased concentration of neutrophils in the lungs of CF subjects, the possible preferential association of SP-D with neutrophils was also assessed. However, there was no correlation of neutrophil concentrations in BALF from children in the control and infection groups with corresponding values for SP-A and SP-D. Consequently, it is unlikely that the low concentration of SP-D in CF was due to its association with inflammatory cells and removal by centrifugation at 400 × g.

There were no significant correlations between the SP-A and SP-D results and any measured parameter of clinical status, including genotype, lung function measurement, and radiographic lung score. There was a suggestion that colonization with P. aeruginosa may have been one determinant of SP-D concentration (Figure 4B), but numbers were too small for this to be significant. The eight children with diagnosed Pseudomonas infection (six in the CF and two in the infection group) all had values of SP-D:phospholipid below the 5% percentile of the control group. There was no such suggestion for the distribution between subject groups of SP-A:phospholipid in respect of Pseudomonas colonization.

	Discussion

Abstract Introduction Materials and Methods Results Discussion References

The decreased concentration of both SP-A and SP-D in BALF from children with CF was dramatic, whether expressed as absolute concentration or relative to total phospholipid or protein (results not shown), and may have profound consequences for the host defense response of the lungs to bacterial infection. The SP-A results confirm the recent observations by Griese and coworkers (9) but diverge from those of Hull and his colleagues (10), who reported increased SP-A in BALF from CF subjects with lung infection. This discrepancy between studies may be due either to the much lower age of the children in Hull's study, who were identified from a neonatal screening program rather than recruited from a diagnostic bronchoscopy service, or to duration of disease and infection. Additionally, samples collected in Hull's study by small volume lung lavage may have been more representative of bronchial rather than of alveolar material. The lower age range of the control group of children in the present study was an unavoidable consequence of the population dynamics of children presenting for bronchoscopy to diagnose structural abnormalities. Hull's study suggests strongly that there is no fundamental absence of SP-A in very young children with CF, and that any subsequent deficiency is probably a consequence of the history of infection in older children. Such a response to infection may underlie the foregoing observation (Figure 4B) that BALF SP-D:phospholipid tended to be lower in children with Pseudomonas infection of the lungs.

This conclusion is supported by analysis of our results (Table 6, Figure 3), which show SP-A concentration marginally decreased in the infected, non-CF group, but only significantly lower in the CF group, who had a history of chronic rather than of acute lung infection. This discrimination was shown even more strikingly for the SP-D results, for which the differences between groups was much greater than for SP-A (Table 6, Figure 4). The SP-D:phospholipid ratio was significantly decreased in the infected, non-CF group, but was dramatically lower in the children with CF. Diverse effects of infection history are suggested by other studies, which generally have shown BALF SP-A to be decreased in established lung infection (23) but increased in rat lung in response to inhalation of bacterial lipopolysaccharide (24).

The consistently decreased concentrations in BALF of hydrophilic surfactant proteins, especially SP-D, have considerable implications for prolongation if not initiation of bacterial colonization of the lungs in CF. SP-A and SP-D interact with bacterial and viral pathogens through lectin-mediated binding to carbohydrate recognition domains, and both contribute to the innate defense system of the lungs. Consequently, absence of SP-D could result directly in impaired presentation of Pseudomonas to a variety of host-defense systems in the lungs.

In contrast to the lowered levels of hydrophilic protein, this study demonstrated clearly that any changes to the concentration or composition of the major surfactant phospholipids in CF are only marginal at best. This lack of difference in surfactant phospholipids between CF, infection, and control groups was not due to any lesser severity of disease in this population of children with CF. All the children in this study underwent fiberoptic bronchoscopy for diagnostic reasons, and represented a wide range of chronological age, duration and severity of lung infection, and clinical status. None of these parameters was related to the phospholipid content of BALF in these children.

Previous studies in CF have reported decreased concentrations of PC and PG in bronchial secretion phospholipid with concomitant increases of PE, PI, and SM, changes consistent with membrane phospholipid from inflammatory cells accumulating in infected sputum (25). Although the trend toward increased fractional content of PI18:0/20:4 in the CF group was possibly derived from inflammatory cell membrane, this contribution to total phospholipid was small. Similarly, although sphingomyelin (16:0-sphingomyelin.Na⁺ m/z = 725; 16:0-dehydrosphingomyelin.Na⁺ m/z = 723) could be detected in all of the children with CF but only in 45% and 54% of children in control and infection groups, respectively, it was always a very minor component (Figure 1A). Consequently, no attempt was made to quantify sphingomyelin species in this study. These results, together with the minimal detectable levels of PE and PS, support the concept that the major portion of BALF phospholipid was indeed derived from pulmonary surfactant.

One problem of previous techniques for the analysis of phospholipid species in BALF has been the relative insensitivity of the analytical techniques available, particularly for control subjects. Consequently, although molecular compositions of PC have occasionally been reported (11, 20, 26), there have been no previous descriptions of the molecular species compositions of PG or PI in BALF or purified surfactant from human subjects. A second advantage of ESI-MS over previous techniques for phospholipid analysis is the degree of structural information provided. Direct-injection ESI-MS has previously been used to analyze phospholipid compositions of blood erythrocytes (12) and platelets (27) but has not, to our knowledge, been previously applied to the analysis of lung surfactant phospholipid.

The absence of any substantial alteration to surfactant phospholipids in BALF from CF patients is perhaps not surprising, because alveolar collapse is not a major clinical finding in CF. Any comparison with previous analyses of bronchial surfactant is problematic, given the high concentration of necrotic leukocytes higher up the bronchial tree. In relation to host defense, it is possibly the ratio of phospholipid to protein that is critically important, because many of the stimulatory actions of hydrophilic surfactant proteins on immune and inflammatory cells are regulated in the opposite direction by surfactant phospholipids. For example, surfactant phosphatidylcholine inhibits both neutrophil respiratory burst response (28) and lymphocyte proliferation (12), whereas SP-A potentiates lymphocyte proliferation (29). Consequently, treatment of CF with exogenous surfactant therapy (30) may have been ineffective because the preparation used, lacking SP-A or SP-D, promoted the inhibitory rather than the stimulatory host defense properties of the endogenous surfactant pool.

The concept that the relative proportions of phospholipid and hydrophilic components of surfactant may contribute to the regulation of the response of lung inflammatory cells to bacterial infection is supported by results from transgenic mouse models. In the cftr^m1HGU/m1HGU homozygous mouse, total lavage PC pool was doubled with unchanged SP-A, giving a decreased SP-A:PC ratio in the absence of lung infection (31). Strikingly, lung function was normal in the SP-A deficient mouse, but instead was associated with increased susceptibility to lung infection with group B Streptococcus (32).

In summary, this study has demonstrated lower concentrations of both hydrophilic surfactant proteinsSP-A and SP-Din CF, which could have a significant role in the pathophysiology of the progressive lung disease in these children. Further studies are required to evaluate whether replacement or enhancement of SP-A and/or SP-D functions may have a beneficial effect on reducing bacterial lung colonization in CF. In addition, introduction of electrospray mass spectrometry provides an experimental tool for analysis of surfactant phospholipid in very small volumes of BALF, with evident direct application to many clinical conditions.