SP-A Permeabilizes Lipopolysaccharide Membranes by Forming Protein Aggregates that Extract Lipids from the Membrane

Materials

Rough LPS from Salmonella minnesota (serotype Re 595) was purchased from Sigma (St. Louis, MO, USA). The fluorescent labeling chemical sulforhodamine 101 sulfonyl chloride, Texas Red (TR), as well as the fluorescent lipid probe 1,6-diphenyl-1,3,5-hexatriene (DPH) were obtained from Molecular Probes Inc. (Eugene, OR, USA). The organic solvents (methanol and chloroform) used to dissolve rough LPS were HPLC-grade. All other chemicals were obtained from Sigma.

DPH concentration was determined spectrophotometrically by absorbance using a molar extinction coefficient in methanol of = 88000 M⁻¹cm⁻¹. Re-LPS concentration was assessed by quantification of 2-keto-3-deoxyoctulosonic acid (KDO) (8). Water used in all experiments and analytical procedures was deionized and doubly distilled in glass, the second distillation being from dilute potassium permanganate solution.

Isolation and labeling of SP-A

SP-A was isolated from bronchoalveolar lavage of patients with alveolar proteinosis using a sequential butanol and octylglucoside extraction. The purity of SP-A was checked by one-dimensional SDS-PAGE in 12% acrylamide under reducing conditions and mass spectrometry (8). Quantification of SP-A was carried out by amino acid analysis in a Beckman System 6300 High Performance analyzer. The oligomerization state of SP-A was assessed by electrophoresis under nondenaturing conditions, electron microscopy, and analytical ultracentrifugation as reported elsewhere (9,10).

Fluorescently labeled SP-A was prepared as described by Ruano et al. (11). Briefly, SP-A in 5 mM Tris-HCl buffer, pH 8.3, was incubated with 1 mM TR (SP-A/TR molar ratio of 6:1) for 90 min in darkness at room temperature. To remove unreacted fluorescent reagent, the mixture was exhaustively dialyzed against 5 mM Tris-HCl, pH 7.4. Activity of labeled TR-SP-A compared to that of native protein was assayed by testing its ability to self-associate and to induce aggregation of DPPC and Re-LPS in the presence of calcium at 37°C as described elsewhere (8–10,12,13). The effect of TR-SP-A on π-A isotherms of Re-LPS monolayers was essentially identical (within our ability to measure) to that of nonconjugated SP-A.

Differential scanning calorimetry

For differential scanning calorimetry (DSC) measurements the lipid samples were prepared as aqueous dispersions. For this, rough LPS was suspended directly in 5 mM Tris-HCl, 150 mM NaCl, pH 7.4 (buffer A), containing either EDTA or CaCl₂, and allowed to swell for 1 h at a temperature above the gel-to-liquid phase transition temperature of the Re-LPS vesicles. Calorimetric measurements were performed as previously reported (14–16) in a Microcal VP differential scanning calorimeter (Microcal Inc., Northampton, MA, USA) at a heating rate of 0.5°C/min. Re-LPS aqueous dispersions (0.4 mM), in the absence and presence of different amounts of SP-A, were loaded in the sample cell of the microcalorimeter with 0.6 mL of buffer A in the reference cell. Four calorimetric scans were collected from each sample between 10°C and 50°C. The standard Microcal Origin software was used for data acquisition and analysis. The excess heat capacity functions were obtained after subtraction of the buffer-buffer baseline.

DPH fluorescence measurements

Steady-state fluorescence measurements were carried out using an SLM-Aminco AB-2 spectrofluorimeter equipped with a thermostatted cuvette holder (± 0.1°C) (Thermo Spectronic, Waltham, MA, USA), with 5 × 5 mm path-length quartz cuvettes, as previously described (14). The required amount of DPH dissolved in methanol was added to Re-LPS dissolved in chloroform/methanol/water 17:7:1 (v/v) (probe/Re-LPS molar ratio of 1:200, final Re-LPS concentration of 1 mg/mL). The solvent was removed under a stream of nitrogen before addition of buffer A, which contained either 150 μM EDTA or 150 μM CaCl₂. The sample was allowed to hydrate for 1 h at a temperature above the gel-to-liquid phase transition temperature of Re-LPS suspension (45°C). To reduce the size of Re-LPS particles, the sample was sonicated at 45°C for 4 min at 390 W/cm² (bursts of 0.6 s, 0.4 s between bursts) in a UP 200S sonifier, with a 2-mm microtip. Exposure to light was minimized throughout the preparation of Re-LPS aqueous dispersions doped with DPH. An appropriate amount of SP-A in buffer was added to each aliquot of Re-LPS dispersions to give the desired final concentration of protein, with less than a 5% increase in volume for each aliquot. For DPH fluorescence intensity measurements, the emission spectra were recorded with the emission polarizer set at the magic angle (m = 54.7°) relative to the vertically polarized excitation beam. This was done to reduce contributions from scattering due to Re-LPS dispersions and to avoid intensity artifacts due to molecular rotation during the lifetime of the excited state (17). Moreover, background intensities in probe-free samples were subtracted from each recording of fluorescence intensity. Measurements were performed at 15°C, 25°C, and 37°C.

Epifluorescence microscopy

Epifluorescence microscopy measurements were performed on a surface balance whose construction and operation have been described previously (18). Monolayers were prepared by spreading Re-LPS onto a buffer A subphase containing either 150 μM CaCl₂ or 150 μM EDTA, with or without 0.08 μg/mL of TR-SP-A, as reported elsewhere (19). The use of low Ca²⁺ concentrations prevents extensive protein self-aggregation that occurs at higher Ca²⁺ concentrations (13). Extensive SP-A self-aggregation hampers the interaction of SP-A with lipid films (20).

After a 1-h period, which allowed for solvent evaporation and penetration of the protein into the gas or gas-liquid expanded coexistence phases, the monolayer was compressed. At selected surface pressures, a video recording of TR fluorescence was made for a 1-min period. The video images were obtained with a CCD camera, which records in black and white. Images were analyzed with digital image processing using JAVA 1.3 software (Jandel Scientific, San Rafael, CA) as discussed elsewhere (11,18).

Pressure-area isotherms and relaxation kinetics of Re-LPS monolayers

Surface pressure-area isotherms of Re-LPS films were measured using a thermostated Langmuir-Blodgett trough (102M micro Film Balance, NIMA Technologies, Coventry, United Kingdom) equipped with an injection port and magnetically stirred. The trough, of a total area of 100 cm², is equipped with two symmetrical movable barriers controlled by an electronic device, which allows barrier movement at constant speed. Re-LPS dissolved in chloroform/methanol/water 17:7:1 (v/v) was spread onto a buffer A subphase containing either 150 μM CaCl₂ or 150 μM EDTA. The lipid monolayers were allowed to equilibrate for 15 min before the injection of SP-A into the subphase to yield a final protein concentration of 0.1 μg/mL (SP-A/Re-LPS weight ratio of 1%). After a 1-h period, which allowed for solvent evaporation and penetration of the protein into the gas or gas-liquid expanded coexistence phases, the monolayer was compressed. A barrier speed of 50 cm²/min has been chosen as the optimum velocity for the film compression. Our preliminary experiments showed rather poor reproducibility of Re-LPS monolayers when a lower compression speed (10 and 35 cm²/min) was used, whereas a higher velocity (60 cm²/min) profoundly affected the collapse characteristics. The surface areas in the trough before and after compression were 76 and 22 cm², respectively. The data shown represent the average of seven measurements. All measurements were performed at 25°C.

To determine SP-A's effects on the relaxation kinetics of Re-LPS monolayers, the Re-LPS organic solution was spread onto a buffered saline subphase by microsyringe. After 15 min the monolayer was compressed to a preset surface pressure kept constant by automatically adjusting the surface area of the trough through the movement of barriers. Once the desired surface pressure was reached, SP-A was injected into the subphase. A relaxation curve was obtained by recording the trough surface area during the relaxation period.

SP-A decreases van der Waals interactions between Re-LPS acyl chains

To gain insight into the mechanism of SP-A-promoted bacterial membrane permeabilization, we explored SP-A's effects on the physical properties of rough lipopolysaccharide membranes using DSC. To obtain further information on the role that calcium ions play in Re-LPS/SP-A interaction, aqueous dispersions of Re-LPS were prepared in buffer A containing either EDTA or calcium. In the presence of 150 μM EDTA, the Re-LPS thermogram was characterized by a phase transition with a gel-to-liquid phase transition characterized by a melting temperature (T_m) of 24.5°C and a transition enthalpy of 4.2 kcal/mol (Fig. 1 and Table 1). When 1 wt % SP-A was added to Re-LPS membranes, no SP-A-induced effects on the T_m were observed, but the overall transition enthalpy of Re-LPS membranes decreased by ∼21% (Fig. 1 and Table 1). Remarkably, greater amounts of SP-A (2.5 wt %) had no effect on either the T_m, the onset temperature, or the transition enthalpy with respect to Re-LPS alone. This might be because of electrostatic repulsions between negative charges in the SP-A surface and the negatively charged Re-LPS molecules at such protein concentrations, which would prevent the interaction of the protein with Re-LPS. In this regard, we have previously shown that at low SP-A concentrations the protein interacts with the negatively charged DPPC/1-palmitoyl-2-oleoyl-phosphatidylglycerol/palmitic acid vesicles (16). However, as the amount of SP-A increases, the interaction of the protein with these membranes is hindered by electrostatic repulsions between SP-A and negatively charged lipids (16).

On the other hand, in the presence of 150 μM CaCl₂, the Re-LPS thermogram was characterized by a complex double peak, indicative of calcium-induced lipid phase separation (Fig. 1 and Table 1). When SP-A was added, the overall transition enthalpy sharply decreased (∼67% at SP-A/Re-LPS weight ratio of 1% and ∼78% at 2.5 wt % SP-A) (Fig. 1 and Table 1). In the presence of 2.5 wt % SP-A there was a very small increase in the transition onset temperature, which was also reflected in the T_m. At saturating calcium concentrations (5 mM), the calorimetric endotherm of Re-LPS was characterized by a single gel-to-liquid crystalline phase transition, which became narrower and shifted to higher temperatures (T_m = 36.4 ± 0.2°C) (data not shown). This indicates stabilization of the gel phase. Under these conditions, extensive self-aggregation of SP-A induced by calcium might prevent their interaction with the membrane, since 1 wt % SP-A decreased the transition enthalpy by only 24% (data not shown). On the other hand, the greater effect of SP-A at low (150 μM) rather than high (5 mM) calcium concentrations might also be because of the presence of lipid phase separation within the Re-LPS membrane at 150 μM Ca²⁺. The coexistence of gel/fluid lipid phases might confer susceptibility to SP-A-mediated membrane disruption since SP-A can insert into or perturb the membrane at the dislocations of the domain boundaries.

Together, these results show that SP-A decreases the transition enthalpy of Re-LPS membranes in both the absence and presence of calcium. This might be explained by a reduction in van der Waals interactions between LPS acyl chains. The SP-A effect on the transition enthalpy of Re-LPS membranes seemed to be stronger in membranes with lipid phase separation induced by calcium.

SP-A favors the penetration of water molecules in Re-LPS membranes

To determine whether the SP-A-induced reduction of LPS-LPS interactions described above is related to the permeabilization of Re-LPS membranes, we explored SP-A/Re-LPS interactions using the fluorescent dye DPH. DPH fluorescence is very sensitive to the polarity of the membrane environment. This probe has a low quantum yield and a very short lifetime when exposed to water. Thus, its fluorescence is sensitive to the amount of water that penetrates into the lipid bilayer. Since SP-A's effects on the thermotropic properties of Re-LPS membranes were studied in the absence and presence of calcium ions, we first examined calcium effects on the fluorescence intensity of DPH embedded in Re-LPS aqueous dispersions. Fig. 2 A shows that in the presence of 150 μM calcium (discontinuous line), the fluorescence intensity of DPH in Re-LPS suspensions was lower than in its absence (continuous line). This effect might be due to the calcium-induced lipid phase separation described above, which would favor the accessibility of water molecules to the probe. As a result, DPH fluorescence is quenched. In fact, at saturating calcium concentrations, the fluorescence of DPH increased with respect to that obtained in the presence of 150 μM CaCl₂ because of the more tight packing of the Re-LPS acyl chains plus the lack of domain separation with attendant boundary dislocations (data not shown).

Fig. 2 B shows the fluorescence intensity of DPH embedded in Re-LPS membranes as a function of SP-A concentration at 25°C, in both the absence and presence of calcium. The addition of very small amounts of SP-A (0.5 wt %) to aqueous suspensions of Re-LPS lacking calcium decreased the fluorescence of the dye (Fig. 2 B, open triangles). However, with increasing SP-A concentrations, a dose-dependent recovering of DPH fluorescence was observed. The absence of SP-A effect on DPH fluorescence at high protein concentrations (2.5 wt %) might be because of increased electrostatic repulsions between SP-A and Re-LPS molecules as described above. On the other hand, the addition of increasing amounts of SP-A to Re-LPS vesicles in the presence of 150 μM CaCl₂ led to a dose-dependent decrease in the fluorescence intensity of DPH (∼70% at SP-A/Re-LPS weight ratio of 2.5%) (Fig. 2 B, solid circles). This decrease in DPH fluorescence intensity may be due to water penetration in the Re-LPS bilayer as a consequence of SP-A-induced Re-LPS membrane destabilization. At saturating calcium concentrations, the addition of increasing amounts of SP-A led to a decrease of the fluorescence signal by only ∼30% (data not shown). To test whether the coexistence of solid and fluid phases in LPS membranes confers susceptibility to SP-A-mediated quenching of DPH fluorescence, we measured the effect of SP-A on DPH fluorescence in the presence of 150 μM CaCl₂ at different temperatures. We found that DPH fluorescence was less affected by SP-A at temperatures below (15°C) and above (37°C) the T_m of Re-LPS vesicles rather than close to it (25°C) (data not shown). Thus, it is reasonable to conclude that SP-A-mediated water penetration in Re-LPS model membranes was favored by the coexistence of gel and fluid lipid phases in these membranes. This is consistent with the finding that the presence of lipid phase separation within the Re-LPS membrane conferred susceptibility to SP-A-mediated membrane disruption as described above.

SP-A forms interconnected aggregates on the surface of Re-LPS monolayers in the presence of calcium

To gain insight into SP-A/LPS interaction, we performed epifluorescence measurements with a fluorescent derivative of SP-A (TR-SP-A). In the absence of calcium, TR-SP-A fluorescence appeared below the monolayers at low surface pressures (4.5 mN/m) (Fig. 3). Above this pressure the fluorescence of TR-SP-A was randomly distributed in the film, forming fluorescent structures with amorphous morphology adsorbed at the polar surface of LPS monolayers (Fig. 3). The interactions between the lipids and the protein led to the formation of an unstable lipoprotein film that can be observed in compression isotherms (Fig. 4 A).

On the other hand, in the presence of 150 μM calcium, TR-SP-A fluorescence appeared in the film at lower surface pressures (2.8 mN/m) than in its absence (Fig. 3), which indicates that this cation modified the interaction of SP-A with Re-LPS and allowed the association of SP-A with the monolayer at lower surface pressures. At 3.9 mN/m, bright fluorescent regions of TR-SP-A aggregates appeared in Re-LPS monolayers. As the pressure increased, the intense rings of TR fluorescence became interconnected. These results show that SP-A has different modes of interaction with Re-LPS monolayers in the absence and presence of calcium. Calcium promoted the self-aggregation of the protein (13), which allowed the achievement of high local protein concentrations.

SP-A destabilizes Re-LPS monolayers

Fig. 4 shows the surface pressure-area (π-A) isotherms obtained at 25°C for monolayers of Re-LPS in the absence (A) and presence (B) of calcium, and with and without SP-A. In the absence of calcium (Fig. 4 A), the compression isotherms of Re-LPS monolayers revealed smooth π-A curves lacking indications for structural transitions (Fig. 4 A, solid line). This might suggest that the films are in the liquid expanded (LE) state. However, coexisting LE and liquid condensed (LC) domains in Re-LPS monolayers have been observed by atomic force microscopy (21). At a surface pressure of 56 mN/m, Re-LPS films collapsed. When 0.1 μg/mL SP-A (1 wt %) was present in the subphase, an expansion of the isotherm was observed (Fig. 4 A, solid circles), which indicates perturbation of lipid packing by the protein. The collapse of the Re-LPS/SP-A film took place at surface pressures lower than that of pure Re-LPS. This suggests that SP-A likely promoted membrane destabilization.

In the presence of 150 μM calcium (Fig. 4 B), the compression isotherm of Re-LPS was shifted to lower molecular areas with respect to that obtained in the absence of the divalent cation (Fig. 4 A). The compression isotherm of Re-LPS films showed a kink at 36 mN/m (Fig. 4 B, continuous line), which suggests the appearance of a phase transition (LE-LC). Calcium might facilitate the transition from a disordered phase (LE) to a more ordered phase (LC). This might be attributed to the electrostatic interactions between the negatively charged groups of Re-LPS and calcium ions, as well as to dehydration effects induced by the presence of CaCl₂. Attempts to visualize this transition by epifluorescence microscopy of Re-LPS monolayers containing NBD-PC did not succeed, since epifluorescence images show that the fluorescence of NBD-PC was uniform over the monolayer surface (data not shown). This would signify that solid domains have much smaller dimensions than the optical resolution. With respect to the collapse pressure of Re-LPS films, we found that it was higher in the presence of calcium (60 mN/m) than in its absence (56 mN/m). This indicates that calcium induces stabilization of the monolayer or that EDTA destabilizes the monolayer due to the EDTA-induced weakening of LPS-LPS interactions (22).

On the other hand, the presence of 0.1 μg/mL SP-A in the subphase expanded the isotherm (Fig. 4B, solid circles) to a lesser extent than in the presence of EDTA (Fig. 4 A, solid circles). This might be because of the calcium-induced charge neutralization, which would minimize the electrostatic repulsions between SP-A and Re-LPS. In addition, SP-A decreased the collapse pressure of Re-LPS monolayers (Fig. 4 B, solid circles), which indicates that SP-A destabilized the interfacial film in the presence of calcium.

SP-A induces Re-LPS molecular loss

To characterize the mechanism of Re-LPS membrane destabilization by SP-A, the relaxation time of the surface area of Re-LPS films was measured after compression to different surface pressures, in both the absence and presence of 0.1 μg/mL SP-A. From the DSC and DPH fluorescence measurements described above (Figs. 1 and 2, respectively) we infer that the SP-A-induced destabilization of LPS membranes in the presence of Ca²⁺ is facilitated by the presence of gel/fluid phase coexistence in the membrane. Thus, Re-LPS monolayers at 25°C were compressed to 10 mN/m (LE phase), 36 mN/m (LE/LC phase coexistence), and 45 mN/m (LC phase). Fig. 5 shows the typical relaxation curves for Re-LPS films at these surface pressures in the presence of 150 μM CaCl₂, with (solid circles) and without (open circles) 0.1 μg/mL SP-A. Pure Re-LPS monolayers exhibited an unstable behavior, with considerable area loss during the relaxation period at all surface pressures studied (Fig. 5, open symbols). Thus, within 20 min, the area loss at both 10 and 45 mN/m was ∼25%, whereas at 36 mN/m the rate of area loss increased substantially and films at 36 mN/m were the least stable, with area loss of ∼53% within 20 min.

The presence of 0.1 μg/mL SP-A in the subphase increased the relaxation rate of Re-LPS monolayers at all surface pressures studied (Fig. 5, solid symbols). As a result the area loss at a given time was greater in the presence than the absence of SP-A. This effect was stronger at 36 mN/m, at which LE and LC phases coexist, with an area loss of 40% within 6 min. This supports the finding that the SP-A-induced destabilization of Re-LPS membranes is favored by the coexistence of gel and fluid lipid phases within the LPS membrane. On the other hand, in the presence of 150 μM EDTA, monolayer instability markedly increased (data not shown). The addition of 0.1 μg/mL SP-A in the presence of 150 μM EDTA did not affect the relaxation kinetics of Re-LPS films compressed to 10 mN/m (data not shown). Re-LPS monolayers compressed to higher surface pressures became so unstable that it was not possible to determine SP-A's effects on the relaxation kinetics of Re-LPS monolayers. The fact that SP-A affected the relaxation kinetics of Re-LPS films, compressed to 10 mN/m, in the presence of calcium, but not in its absence, seems to indicate that the SP-A-induced destabilization of Re-LPS membranes is calcium-dependent.

Two possible processes might account for the area loss of Re-LPS monolayers described above: the transformation of the two-dimensional (2D) monolayer material into three-dimensional (3D) aggregates (23,24), and the dissolution of lipid molecules into the subphase through diffusion (25). According to a model proposed by Smith and Berg (25), if the monolayer molecular loss is because of dissolution of Re-LPS molecules into the subphase, a linear relationship between -log(A/A₀) and the square root of time, would be obtained, where A and A₀ are the trough surface areas at a given time, t, and at t = 0, respectively. Thus, to discriminate between desorption and the nucleation and growth of 3D aggregates, the variance of -log(A/A₀) with was plotted for Re-LPS films in the absence and presence of SP-A (Fig. 6). The absence of a linear relationship of Re-LPS data, with and without SP-A, indicates that at all surface pressures studied, the area loss during the relaxation was caused by the formation of 3D aggregates but not by the dissolution of Re-LPS molecules into the subphase. These results are in agreement with previous findings of Garidel and co-workers (26), who showed that unilamellar/cubic inverted aggregate structures of Re-LPS are converted into multilamellar structures by divalent cations. Since in the presence of calcium large protein aggregates were observed on the surface of Re-LPS monolayers (Fig. 3), it is conceivable that the calcium-dependent permeabilization of Re-LPS membranes by SP-A is related to the formation of calcium-mediated lipid-protein aggregates that extract LPS molecules from the membrane. This would mean fewer LPS molecules per unit of area, which would cause the membrane to be less tightly packed and permeable.