Effect of Lung Surfactant Protein SP-C and SP-C-Promoted Membrane Fragmentation on Cholesterol Dynamics

Abstract

To allow breathing and prevent alveolar collapse, lung surfactant (LS) develops a complex membranous system at the respiratory surface. LS is defined by a specific protein and lipid composition, including saturated and unsaturated phospholipid species and cholesterol. Surfactant protein C (SP-C) has been suggested to be an essential element for sustaining the presence of cholesterol in surfactant without functional impairment. In this work, we used a fluorescent sterol-partitioning assay to assess the effect of the surfactant proteins SP-B and SP-C on cholesterol distribution in membranes. Our results suggest that in the LS context, the combined action of SP-B and SP-C appears to facilitate cholesterol dynamics, whereas SP-C does not seem to establish a direct interaction with cholesterol that could increase the partition of free cholesterol into membranes. Interestingly, SP-C exhibits a membrane-fragmentation behavior, leading to the conversion of large unilamellar vesicles into highly curved vesicles ∼25 nm in diameter. Sterol partition was observed to be sensitive to the bending of bilayers, indicating that the effect of SP-C to mobilize cholesterol could be indirectly associated with SP-C-mediated membrane remodeling. Our results suggest a potential role for SP-C in generating small surfactant structures that may participate in cholesterol mobilization and pulmonary surfactant homeostasis at the alveolar interfaces.

Introduction

From the first moment the lungs of a newborn are exposed to air, lung surfactant (LS) complexes secreted by pneumocytes rapidly cover the alveolar air-liquid interface. There, LS layers prevent alveolar collapse by lowering the surface tension to values close to 0 mN/m during expiration, thereby minimizing the work of breathing and allowing the lungs to expand and compress normally (reviewed in (1)).

Lipids account for 90% of the LS total mass. The exact proportion of saturated/unsaturated phospholipids and cholesterol is still a matter of discussion, especially with regard to the human lung. However, it is widely accepted that in most mammals, disaturated phospholipids such as dipalmitoylphosphatidylcholine (DPPC) represent ∼40% of LS lipids. These saturated phospholipids are responsible for the low surface tension values observed upon expiration. Other PC species are also present in LS, as well as phospholipids such as phosphatidylinositol and phosphatidylglycerol. In LS neutral lipids, cholesterol accounts for ∼8–10% by mass of the whole lipid fraction (1). Specific surfactant proteins (SP-A, SP-B, SP-C, and SP-D) represent <10% of the surfactant mass but are crucial components for the proper function of LS. Surfactant optimal biophysical function is essentially achieved by the presence of the small hydrophobic polypeptide SP-B, while the transmembrane protein SP-C serves to enhance lipid adsorption and modulate the activity of cholesterol-containing films. SP-B and SP-C each account for <1% of surfactant mass.

LS membranes and monolayers have a complex and dynamic lateral structure that is determined by its lipid composition and includes segregation of liquid-ordered (Lo) and -disordered (Ld) fluid phases (2). The proteins SP-B and SP-C do not appear to contribute to lateral phase segregation (2), but they do contribute to the formation of membrane stacks that generate highly cohesive multilayer assemblies (3). However, cholesterol has been shown to be essential for lateral phase segregation, especially at near-physiological temperatures (2, 4). Removal of cholesterol substantially alters phase coexistence in LS membranes (2), but functionally its incorporation above a certain threshold has been reported to be deleterious for the proper surface activity of some clinical surfactant preparations (5). However, it has also been shown that cholesterol improves the spreading capability of model surfactant lipids (6). These contradictory findings suggest that the role of cholesterol is subtly related to the architecture and properties of LS at the air-water interface, and that it could critically depend on the presence of other LS components and/or LS structures assembled in the native complexes.

Surfactant protein SP-B is essential for LS function. This protein exists naturally as a covalent dimer supporting the stability of multilayered films at the air-liquid interface. SP-B also improves lipid adsorption and the respreading ability of surfactant material along the breathing cycles (reviewed in (7)). Previous evidence showed that SP-B alone is not enough to sustain the functionality of cholesterol-containing mixtures (8). Remarkably, the presence of the lipopeptide SP-C has been shown to prevent the dysfunction of surfactant preparations incorporating cholesterol (8, 9). This small (4.2 kDa) protein increases cholesterol miscibility in surfactant-mimicking membranes (9), and SP-C exposure in bilayers has been reported to be affected by the presence of cholesterol (10). Therefore, SP-C may be an important factor in modulating cholesterol in the LS context, although the molecular mechanisms connecting SP-C and cholesterol remain unclear.

Our aim in this work was to obtain further information about the role of the hydrophobic surfactant proteins SP-B and SP-C in cholesterol distribution in LS membranes. Recently, Nyström and co-workers (11) developed a novel fluorescence-based method to determine the effect of transmembrane peptides on the affinity of sterols for membranes. With this technique, it is possible to follow the partitioning of a fluorescent analog of cholesterol (cholesta-5,7,9(11)-trien-3-β-ol (CTL)) between membranes and a solubilizing agent such as cyclodextrin by measuring the anisotropy of the fluorophore. These measurements allow the determination of a partition coefficient, K_x, which indicates the affinity of a sterol for a given membrane.

Using the CTL partitioning assay, we prepared model and native surfactant membranes in the presence or absence of SP-B and SP-C to assess their effects on the affinity of sterols for such membranes. These surfactant proteins contributed substantially to alter CTL distribution in a specific manner both together and separately. Moreover, our results show that membrane curvature has a significant influence on CTL’s affinity for bilayers, which may have important physiological consequences for the regulation of cholesterol levels in membranes. We conclude that SP-B, SP-C, and membrane curvature alter cholesterol distribution in surfactant membranes, which may have important implications for the maintenance of surfactant function and homeostasis.

Materials and Methods

Materials

Lipids were purchased from Avanti Polar Lipids (Alabaster, AL). LS was isolated from porcine lungs as described previously (12) and subjected to an organic extraction (13). LS hydrophobic constituents were obtained upon extraction with chloroform/methanol according to the classical Bligh and Dyer (13) method. The LS organic extract (OE) was then subjected to chromatographic separation in a gel filtration resin (LH-20; GE Healthcare, Little Chalfont, United Kingdom) equilibrated with chloroform/methanol (2:1, v/v) (14). The hydrophobic protein fraction (Prot), containing only the surfactant proteins SP-B and SP-C in their original proportions, was obtained in the first peak. The second fraction (PL) corresponded to surfactant phospholipids, and the third fraction (N) consisted of neutral lipids. These fractions were combined, keeping their volumetric ratio to maintain their native proportions, and tested for effects on CTL membrane partitioning. Fractions containing phospholipids were quantified for phosphorus content (15).

As a model of LS membranes, we used the lipid mixture DPPC/1-palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine (POPC)/1-palmitoyl-2-oleoyl-sn-glycero-3-phosphoglycerol (POPG) (50:25:15, w/w/w), which simulates the physiological proportions of the saturated/unsaturated and zwitterionic/anionic phospholipids found in most lung surfactants.

Palmitoyl sphingomyelin (PSM) was purified from egg yolk sphingomyelin by reverse-phase high-performance liquid chromatography (HPLC) using methanol/water (95:5, v/v) as the eluent. CTL was synthesized as described by Fischer et al. (16) and purified by reverse-phase HPLC with a mobile phase composed of acetonitrile/methanol (7:3, v/v). 18:1-DPH-PC (1-oleoyl-2-propionyl-DPH-sn-glycero-3-phosphocholine) was prepared from 1-oleoyl-2-OH-sn-glycero-3-phosphocholine and DPH-propionic acid (Setareh Biotech, Eugene, OR) according to established methods (17, 18). The probe was purified by preparative HPLC (Discovery C18 column; Supelco, Bellefonte, PA) with methanol as the eluent. The purity and identity of DPH-PC were assessed by electrospray ionization mass spectrometry and analytical HPLC. The concentrations of the fluorophore stock solutions were determined based on their extinction coefficients (DPH-PC: 88,000 cm⁻¹M⁻¹ at 350 nm in methanol; CTL: 11,250 cm⁻¹M⁻¹ at 324 nm in ethanol).

Phospholipid stock solutions were prepared in hexane/2-propanol (3:2, v/v). The solutions were stored at −20°C and warmed to room temperature before use. Methyl-β-cyclodextrin (mβCD) was purchased from Sigma-Aldrich (St. Louis, MO).

The surfactant proteins SP-B and SP-C were isolated from minced porcine lungs as previously described (14). Briefly, the lavage of minced lungs was subjected to an organic extraction and the organic phase, containing lipids and proteins SP-B and SP-C, was loaded onto a gel filtration chromatographic column (LH-20; GE Healthcare). The protein fraction (SP-B and SP-C) was obtained from the lipids that were eluted in the first peak. To obtain SP-B and SP-C separately, the protein fraction was subjected to a second chromatographic step through a gel filtration column (LH-60; GE Healthcare).

Recombinant SP-C (rSP-C) was overexpressed in Escherichia coli and purified according to the method of Lukovic et al. (19) with modifications (20). The rSP-C sequence consisted of 37 residues: GPFGIPCCPVHLKRLLIVVVVVVLIVVVIVGALLMGL. Protein purity was assessed by SDS-PAGE and the concentration was determined by amino acid analysis.

Partitioning of CTL between mβCD and phospholipid membranes

The partitioning assay is based on cyclodextrin extraction of cholesterol/CTL from membranes. The efficiency of this extraction depends on the lipid composition, reflecting the different cholesterol/CTL affinities for each lipid system. These affinities can be assessed by incubating lipid vesicles with increasing amounts of mβCD and measuring how much cholesterol/CTL remains in the membrane. Within a phospholipid bilayer, CTL motion is restricted, which translates into a certain anisotropy value. However, when CTL is forming part of the CTL-mβCD complex, CTL is more isotropic and consequently the anisotropy value decreases substantially. The anisotropy value can be related to the CTL concentration. Determining the amount of CTL in the bilayers after exposure to different mβCD concentrations allows the determination of the CTL molar fraction-partitioning coefficient (K_x) between the membrane and mβCD. Higher K_x values indicate a higher affinity of CTL for the bilayer as compared with cyclodextrin.

Lipid vesicles were prepared by mixing phospholipids and CTL (2 mol %) in organic solvent. The solvent was evaporated under a stream of nitrogen at 40°C. The dried films were thoroughly redissolved in chloroform/methanol (2:1, v/v) and the indicated amount of protein was added. All of the LS components were mixed in a single step and dried in the same manner. After an additional solvent evaporation cycle, the samples were maintained under vacuum overnight to eliminate solvent traces. Multilamellar suspensions were obtained by rehydrating lipid or lipid/protein films in Tris buffer (5 mM Tris (pH 7) 150 mM NaCl) at a temperature above the gel-to-fluid transition temperature of phospholipids and vortexing. Large unilamellar vesicles (LUVs) were formed by extrusion through polycarbonate membranes with 200 nm pores. For vesicle curvature experiments, the extrusion was performed through polycarbonate membranes with 30, 50, 100, and 200 nm pore sizes.

For the experiments with synthetic lipid mixtures including the native protein fraction (SP-B+SP-C), we increased the proportion of protein/phospholipid fivefold compared with the native concentrations to enable comparison with the assays performed with each protein alone.

For the partition assay, LUVs were dispensed into 100 nmol lipid fractions, resulting in 10 tubes (final lipid concentration of 40 μM). One tube was used to measure CTL anisotropy in each lipid system in the absence of mβCD, reflecting the motion state of CTL for each particular lipid composition. Higher anisotropy values indicate less motion of the CTL molecule, whereas the lower the anisotropy, the more freely the molecule is moving. Increasing concentrations of mβCD were added to the remaining tubes (0.04–1 mM) to calculate the partition coefficient, as detailed below. LUV dispersions were checked for vesicle size using a Z-Sizer (Malvern, Worcestershire, UK).

For POPC and POPC/PSM vesicles, samples were saturated with argon to prevent the oxidation of fluorophores, and incubated overnight at room temperature or for 2 h at 37°C to allow the partitioning to reach equilibrium. For the rest of the lipid systems tested, the incubation and measurements were carried out at 37°C. Then, the steady-state anisotropy of CTL (at 23°C or 37°C) was measured using a PTI Quantamaster 1 spectrofluorimeter (Photon Technology International, Birmingham, NJ) operating in the T-format, with both the excitation and emission slits set to 5 nm. For CTL anisotropy, the samples were excited at 324 nm and the emission was measured at 390 nm. The molar concentration of CTL $\left(C_{CTL}^{LUV}\right)$ for each sample was calculated from the measured anisotropies (11):

$$C_{CTL}^{LUV}=C_{CTL}\frac{\left(r_i-r_{m\beta CD}\right)}{\left(r_{LUV}-r_{m\beta CD}\right)},$$ (1)

where $C_{CTL}$ is the total concentration of CTL in the samples, $r_{LUV}$ is the anisotropy of CTL in the specific phospholipid bilayer in the absence of cyclodextrin, $r_i$ is the CTL anisotropy in each measured sample, and $r_{m\beta CD}$ is the anisotropy of CTL in the CTL-mβCD complex. The anisotropy of the CTL-mβCD complex was determined to 0.175 at room temperature and 0.170 at 37°C. According to this equation, for the sample tube with no mβCD, $C_{CTL}^{LUV}$ equals the total concentration of CTL used for these experiments.

The CTL concentrations calculated above were plotted against the mβCD concentration assuming a 2:1 CTL/cyclodextrin stoichiometry in the complexes as previously described (21). To calculate the molar fraction partition coefficient, K_x, these curves were fitted to Eq. 2 using the software Origin 7.5 (OriginLab, Northampton, MA):

$$C_{CTL}^{LUV}=\frac{C_L-C_{CTL}+\frac{{\left(C_{CD}\right)}^n}{K_X}}{2}\left(\sqrt{1+4\frac{C_L{C}_{CTL}}{{\left[C_L-C_{CTL}+\frac{{\left(C_{CD}\right)}^n}{K_X}\right]}^2}}-1\right),$$ (2)

where $C_L$ is the concentration of phospholipid, $C_{CTL}$ is the concentration of CTL at the beginning of the assay, n is the stoichiometry of the CTL-mβCD complex, C_CD is the concentration of cyclodextrin, and K_x is the partition coefficient. Representative fittings are shown in Fig. S1 in the Supporting Material.

The phospholipid concentration of the samples was corrected after the preparation of each mixture for the CTL partitioning assay by phosphorus content determination.

Determination of DPH-PC anisotropy in LS membranes

We prepared 100 nmol lipid films by mixing the appropriate amount of the different LS fractions and DPH-PC (1 mol %). The films were dried as described in the previous section and maintained under vacuum overnight. The dried films were hydrated as described above and extruded through 200 nm pores in 1 mL of Tris buffer. An additional milliliter of buffer was added and the steady-state anisotropy at 37°C was measured in the same instrument with the slits at 5 nm. All samples were excited at 358 nm and the fluorescence emission was measured at 430 nm.

Cryo-electron microscopy of SP-C-containing vesicles

Sample LUVs made of DPPC/POPC/POPG (50:25:15, w/w/w), in the absence or presence of SP-C, were applied to holey carbon copper grids (R2/2; Quantifoil, Großlöbichau, Germany) and vitrified using a Cryoplunge (Gatan, Pleasanton, CA). The grids were imaged in a JEM-1230 transmission electron microscope (JEOL, Tokyo, Japan) operated at 100 kV and recorded at a final magnification of 5.68 Å/px using an F416 CMOS camera from TVIPS (Gauting, Germany).

Statistical significance

Data were analyzed using an unpaired Student’s t-test or ANOVA with a post hoc Student-Newman-Keuls test (^∗p < 0.05, ^∗∗p < 0.01, ^∗∗∗p < 0.001) from the corresponding control lipid systems using the software GraphPad Prism 5.

Results

CTL partitioning between mβCD and LS-derived membranes

As an indirect measurement of CTL distribution in LS membranes, we measured CTL affinity for LS-derived membranes using LUVs made of the most physiologically relevant lipid mixtures. The OE of surfactant constituted the most representative LS sample, since it contained all of the hydrophobic components of LS in their native proportion. These components were first separated into protein (Prot), phospholipid (PL) and neutral lipid fraction (N), and later recombined for testing by the CTL partitioning assay. Fig. 1 a shows CTL anisotropy for each combination as a function of the concentration of mβCD. Higher anisotropy values indicate less motion for CTL, meaning it is in a more rigid environment, whereas higher K_x values indicate a higher affinity of CTL for that particular membrane. For no mβCD and every mβCD concentration added, the anisotropy achieved in the presence of the hydrophobic surfactant proteins (PL+Prot) was below the anisotropy reached for samples containing only LS phospholipids. This tendency was maintained when the neutral lipid fraction was added (PL+N versus PL+Prot+N). Lastly, the OE behaved similarly to the samples containing all the components of surfactant (PL+Prot+N), as expected. Thus, the anisotropy data suggest an alteration of CTL distribution to a more fluid environment (lower anisotropy) depending on the presence of the proteins.

Figure 1.

CTL anisotropy and partitioning into LS-derived membranes. (a) CTL anisotropy is shown as a function of mβCD concentration at 37°C. (b) The partition coefficients were computed from the CTL anisotropy values for each LS combination. Error bars represent the standard deviation (SD) of three independent experiments.

A significant decrease in K_x values was also observed in the presence of the proteins (PL versus PL+Prot and PL+N versus PL+Prot+N), suggesting that the proteins increase the susceptibility of cholesterol to be removed from the membrane by mβCD (Fig. 1 b).

From these results, it is clear that in the presence of surfactant proteins, CTL acquires increased motion and a decreased affinity for LS-derived membranes. SP-B and SP-C have been widely reported to perturb lipid membranes (22, 23), and their effect on cholesterol-induced membrane phases has been also assessed (24). The effect we observed could be explained by an indirect consequence of the membrane perturbation induced by SP-B and SP-C, or by a direct alteration mediated by the proteins.

Regarding the effect of the neutral lipid fraction, slightly higher anisotropy values were found when it was included in the vesicle composition (PL versus PL+N) even in the presence of the proteins (PL+ Prot versus PL+Prot+N), especially for mβCD concentrations beyond 0.15 mM (Fig. 1 a). This is reflected by the observed increase in K_x for samples including the neutral fraction (PL versus PL+N) (Fig. 1 b). As can be deduced from the graph in Fig. 1 a, the same thing occurred when K_x was calculated in the presence of the proteins (PL+N versus PL+Prot+N). It is important to note that CTL and cholesterol compete for mβCD, and that could be the reason for these apparently higher K_x values.

Cholesterol apparently partitions into ordered domains in LS membranes at physiological temperatures (2). Given previous evidence (25, 26), we assume that CTL partitions analogously to cholesterol in different lipid environments within the membrane, although its lateral distribution is difficult to predict in these complex membranes. To better assess the effects on CTL motion of adding protein or cholesterol (N) to each lipid system, we analyzed CTL anisotropy in the absence of cyclodextrin (Fig. 2). Again, our results indicate that the effect of the proteins is to increase CTL motion, as detected by a decrease in anisotropy, which could be accomplished by favoring CTL distribution to a more fluid environment. The addition of neutral lipids (PL+N) did not significantly change the apparent anisotropy of CTL in the pure lipid system (PL), as seen from the anisotropy data (Fig. 2). This could be interpreted as indicating that the CTL environment was not disturbed by solely cholesterol addition. In membranes perturbed by the action of SP-B and SP-C (PL+Prot), the presence of cholesterol partially counteracted the decrease in CTL anisotropy induced by the proteins (PL+Prot+N). A possible explanation for this could be a cholesterol-induced membrane reorganization, which could modify the proteins’ density and distribution. In this sense, the motion of CTL would be more restricted, as indicated by its higher anisotropy.

Figure 2.

CTL and DPH-PC anisotropy in LS-derived membranes. Anisotropy measurements are shown for CTL (left panel) and DPH-PC (right panel) in membranes prepared from each indicated combination of LS fractions including LS OE, in the absence of mβCD, and measured at 37°C. Average ± SD of n = 3.

To determine whether the effect on CTL anisotropy was mediated directly by the proteins, the anisotropy of another fluorescent lipid probe, DPH-PC, was measured in the same samples. The measurements were conducted at 37°C, a temperature at the end of the native surfactant phase transition (2), where surfactant membranes are extremely dynamic. DPH-PC partitions preferentially into Ld phases (27), where both SP-B and SP-C are also preferentially located, whereas CTL would likely be distributed into Lo domains. It is expected that a nonselective perturbation of the membrane by the proteins would affect both probes with a similar magnitude, and therefore the changes observed for CTL anisotropy could be mirrored by the effects on DPH-PC. Fig. 2 compares the anisotropy of DPH-PC with that of CTL in the same samples. We observed that there were no significant differences in the anisotropy of DPH-PC in lipid systems alone compared with the same combinations in the presence of the proteins (PL versus PL+Prot, PL+N versus PL+Prot+N). However, as described above, the presence of the proteins caused a significant decrease in CTL anisotropy. This supports the idea that the presence of SP-B and SP-C alters the behavior of CTL in as yet unknown specific manner. Thus, we can conclude that the combined action of SP-B and SP-C results in an increase in CTL motion and a decrease in CTL affinity for LS-derived membranes.

Effect of SP-C and SP-B on CTL distribution

Once we confirmed that the hydrophobic protein fraction could be directly responsible for the alteration in CTL motion and the affinity of CTL for LS-derived membranes, the effect of SP-C on CTL distribution became of special interest because of previous evidence connecting SP-C with the structure and activity of surfactant preparations in the presence of cholesterol (8, 9, 10). To assess the specific action of SP-C on the affinity of CTL for membranes, we tested 200 nm LUVs of different compositions in the absence or presence of equimolar proportions of SP-C and CTL. Fig. 3 summarizes the effects of SP-C on CTL anisotropy in the pure lipid system (in the absence of mβCD) and the K_x values for two of the lipid systems analyzed. The protein caused a substantial reduction in CTL anisotropy for POPC membranes, suggesting again an increase in CTL motion. No significant differences were found for CTL anisotropy in the surfactant-mimicking mixture DPPC/POPC/POPG (50:25:15 w/w/w) in the presence of SP-C. A significant decrease in K_x for both POPC and DPPC/POPC/POPG membranes was observed, suggesting that CTL had a reduced affinity for the two types of membranes in the presence of SP-C.

Figure 3.

CTL anisotropy and partitioning into SP-C-containing membranes. (a) Comparison of CTL anisotropy in 200 nm vesicles made of POPC or DPPC/POPC/POPG (50:25:15 w/w/w), in the absence (control) or presence of SP-C, with no mβCD in the medium. (b) K_x was calculated from the anisotropy values applied to the model as explained in Materials and Methods. Average ± SD of n = 3.

We previously reported that SP-C could cause membrane blebbing in giant unilamellar POPC vesicles (23). The increase in the dynamic character of these vesicles caused by the protein, with membranes showing more frequent undulations of higher amplitude, occasionally resulted in the formation of smaller liposomes. Taking into account this previous observation, we checked the size of the vesicles including SP-C by dynamic light scattering (DLS). Remarkably, the diameters obtained for the population of liposomes were greatly reduced compared with the 200-nm-pore vesicles that were originally formed by extrusion (Table 1). This SP-C-induced fragmenting effect was also confirmed by cryo-electron microscopy (cryo-EM). A high proportion of very small and highly curved vesicles was observed in the presence of the protein (Fig. 4 a), whereas vesicles with no protein showed a size distribution centered around 130 nm, which followed the typical Gaussian distribution obtained upon vesicle extrusion (Fig. 4 b). This effect was comparable in the two lipid systems studied.

Table 1.

Effect of Surfactant Proteins on the Size of Lipid Vesicles as Measured by DLS

Lipid Mixture	Protein	Vesicle Diameter (nm)a
	–	106 ± 25
POPC	+ SP-C	24 ± 2
	+ SP-B/SP-C	83 ± 11
	–	114 ± 9
DPPC/POPC/POPG	+ SP-C	29 ± 3
	+ SP-B/SP-C	101 ± 8
POPC/PSM	+ SP-C	27 ± 2
POPC/Chol	+ SP-C	24 ± 1

^aValues are presented as means ± SD of n = 3.

Figure 4.

Effect of the presence of SP-C on the fragmentation of LUVs as observed by cryo-EM. (a) Representative cryo-EM micrographs illustrate differences in vesicle size in samples of LUVs reconstituted in the absence (upper panels) or presence (lower panels) of SP-C. Scale bars, 100 nm. (b) Size distribution of vesicles obtained by quantifying the diameter of the vesicles in the absence (upper panel) or presence (lower panel) of SP-C. n = 389 for pure lipid LUVs; n = 422 for SP-C-containing LUVs.

In an attempt to prevent the breaking effect of SP-C, we assayed stiffer membranes. Interestingly, the POPC/PSM (80:20 mol/mol) vesicles showed the same fragmentation as those prepared from POPC and 30 mol % cholesterol (Table 1). These results suggest that, under the conditions tested, the high ability of SP-C to fragment large membranes into small nanovesicles was independent of the lipid composition.

A recombinant version of SP-C lacking the two palmitic chains at the N-terminal segment was also incorporated into POPC vesicles. Preliminary results showed that this recombinant version of the protein also produced the conversion of large membranes into small vesicles (data not shown). Consequently, it seems that the fission activity of SP-C was not dependent on the palmitoylation state.

The effect of SP-B on membranes was different from that observed for SP-C. DLS measurements indicated an extremely high polydispersity index for the samples. For this reason, although CTL anisotropy seemed to be reduced when compared with the pure lipid system (data not shown), the data could not be fitted to obtain a K_x within an acceptable error caused by sample scattering. This result is consistent with previous studies that reported a fusogenic activity of SP-B (28, 29), which could generate different vesicle entities in the sample.

Lastly, we wanted to analyze the effect of the simultaneous presence of the two hydrophobic surfactant proteins, SP-B and SP-C. The hydrophobic protein fraction isolated from LS includes the precise relative proportion of SP-B/SP-C encountered in native surfactant membranes (∼1:1 by weight). For this reason, samples including this fraction were assayed. The simultaneous presence of the two proteins reduced both the anisotropy and the calculated K_x (data not shown). However, the size of these vesicles was maintained at the same diameter of the extrusion membrane pores, as confirmed by DLS measurements (Table 1). This indicates that the fission/fusion activities of the two proteins are canceled out when they are together.

There is evidence indicating that in highly curved vesicles, the looser packing of lipids increases cholesterol exchange from small unilamellar vesicles toward other small unilamellar vesicles or LUVs (30). Thus, it is very likely that CTL experiences a similar effect; however, the influence of membrane curvature on CTL partitioning has not been reported. Taking this into account, we cannot discard the possibility that the observed effect of SP-C on CTL partitioning is due to the formation of highly curved membranes. For this reason, we also analyzed the influence of membrane curvature on CTL partitioning.

Effect of membrane curvature on CTL partitioning

To gain insights into how cholesterol distribution is affected by membrane curvature, we performed a CTL partitioning assay using vesicles of increasing diameters (30, 50, 100, and 200 nm) and two different lipid systems. As shown in Fig. 5 a, the lowest anisotropy values were found for the smallest vesicles. This was observed for the two lipid systems tested at the temperatures measured. The K_x values also correlated with the anisotropy decrease for the more curved membranes (Fig. 5 b). A more pronounced membrane curvature seemed to facilitate CTL extraction from the membrane by mβCD. This might be explained by an increase in CTL exposure to the media, which would improve its accessibility to mβCD, or by a weakening in CTL interactions with surrounding lipids, resulting in higher off-rates. Thus, it is possible that the conformational freedom of this molecule is intrinsically increased in more curved membrane environments.

Figure 5.

CTL anisotropy and partition coefficients as a function of membrane curvature. (a) CTL anisotropy in POPC membrane vesicles of increasing diameters measured at 23°C. (b) Partition coefficients calculated for the two lipid systems in vesicles of different diameters. Actual vesicle size values are presented as means ± SD; n = 3.

Taken together, our results suggest that membrane curvature significantly alters the distribution and diffusion of CTL in model membranes. This finding could be extrapolated to cholesterol, with important physiological consequences.

Discussion

In this work, we used a cholesterol fluorescent analog, CTL, to examine how the hydrophobic proteins of LS influence cholesterol distribution in membranes. Although the role of cholesterol in LS has been discussed extensively, to date, its specific function is poorly understood. Cholesterol is a crucial actor in defining the lateral structure that LS layers adopt at physiological temperatures, and it has been shown that its removal has remarkable effects on their lateral structure (2). Together with the hydrophobic surfactant proteins, cholesterol confers the proper fluidity and viscosity to LS (31). In doing so, it contributes greatly to the unique dynamic characteristics that support the biophysical function of LS. This notion is supported by the fact that the proportion of cholesterol is subjected to a tight regulation in surfactant from heterothermic mammals, which are able to live at different body temperatures (4). Paradoxically, it has been demonstrated that an excess of cholesterol inhibits surfactant function (32, 33, 34), and cholesterol is usually removed from some clinical surfactant preparations used to initiate pulmonary function in preterm babies (5).

Cholesterol regulation in LS is very tight and independent of that of phospholipids (31). Changes in cholesterol levels enable LS structures to adapt to diverse environmental changes involving temperature, pressure, or mechanical demands, as has been demonstrated for torpid and diving mammals (31, 35). However, it is not clear whether the hydrophobic protein component of surfactant, SP-B and/or SP-C, is controlled to match cholesterol changes, or whether the proteins play an important role in sensing or modulating cholesterol needs in surfactant.

SP-B and SP-C increase CTL motion in surfactant membranes

In this work, the hydrophobic proteins SP-B and SP-C were shown to increase CTL mobility in surfactant-derived membranes, as detected by a decrease in anisotropy. This suggests a protein-induced change in CTL distribution toward more fluid phases. We are aware that the application of the CTL mβCD-partitioning model to membranes with phase coexistence might be controversial if CTL is segregated differently between phases. However, we think that our observation that proteins SP-B and SP-C alter the mobility and partitioning of CTL is reliable because 1) it was previously reported that the proteins do not alter the lateral structure of surfactant membranes (2), and 2) the proteins caused a consistent reduction in anisotropy and the CTL partition coefficient both in simple membrane systems and in the complex LS membranes, which could exhibit marginal segregation of ordered and disordered phases.

We have observed that SP-B and SP-C are efficient in facilitating CTL mobility within the membrane, a property that was also exhibited for other lipid probes, such as spin-labeled lipids (36). The increase in motion for CTL seems to be related to lower K_x values, reflecting a higher susceptibility of CTL to be removed from the membrane. This increment in mobility could be particularly essential at the interfacial film, where cholesterol has to be squeezed out during compression for the film to reach the minimum tension required to sustain breathing (37, 38). We have observed that the effects of the proteins in the most simple lipid systems resemble what happens in the native lipid-protein complex. The protein-promoted increase in CTL motion and the diminished affinity for the membrane support the notion that SP-B and SP-C play a major role in modulating the dynamic properties of surfactant lipids. Given the similarities between CTL and cholesterol (26), our findings could be extrapolated to the behavior of cholesterol in the presence of the proteins.

In this study, we observed that SP-B and SP-C have a carefully balanced combined action. The behavior of the whole hydrophobic protein fraction differed greatly from that of each independent protein. DLS experiments showed that vesicles incorporating SP-B and SP-C were stable in size. SP-B increased the variability in vesicle size in the absence of SP-C, which is consistent with the previously reported SP-B-induced membrane fusion (28, 39). SP-C, however, exhibited the opposite effect as observed by DLS and cryo-EM. These observations illustrate how the dynamic structure of surfactant complexes at the alveolar spaces could be fine-tuned by the subtle and integrated regulation of SP-B and SP-C.

SP-C-induced membrane-fragmenting effect and cholesterol homeostasis in LS

We have shown here that SP-C disrupts lipid bilayer integrity by forming very small and highly curved vesicles, which is consistent with previous observations (23). These small vesicles generated in SP-C-enriched environments could be the basis for SP-C to participate in mechanisms to refine the membrane composition, with possible implications for LS function.

It is known that an increased level of cholesterol impairs the proper biophysical behavior of LS (8), as occurs in some respiratory disorders. Patients with acute respiratory distress exhibit high levels of cholesterol in surfactant (34), as do some patients with pulmonary alveolar proteinosis (40). Thus, LS refinement from less functional components could be essential to maintain LS homeostasis and functionality.

It has been demonstrated that SP-C restores the function of cholesterol-containing surfactant preparations (8, 9). Our results for different vesicle diameters proved that in small and highly curved vesicles, CTL anisotropy was increased and partitioning toward cyclodextrin was favored (reduced K_x). Considering that SP-C alone caused a similar effect on both CTL anisotropy and K_x, and given its membrane-fragmenting effect, we suggest that CTL formed part of those SP-C-generated vesicles.

Taking into account the selective CTL mobilization we observed here, we speculate that SP-C-promoted membrane fragmentation could be involved in cholesterol regulation with regard to its recycling and removal from LS membranes. This could provide a mechanistic framework for the pivotal role played by cholesterol levels in the adaptation of surfactant to different environmental conditions (4). Indeed, the SP-C-generated small vesicles could also be related to the conversion of large, functional surfactant aggregates into less active, small aggregates, which have been reported to be produced as a consequence of surfactant extracellular metabolism (41, 42, 43).

A model for SP-C-induced membrane budding

At a molecular level, the effect of SP-C to promote surfactant membrane curvature and ultimately membrane fission should be related to the simple but particular structure of this lipopeptide. Here, we observed a membrane-fragmenting effect that seems to be independent of the lipid composition or palmitoylation state of SP-C.

We previously reported that both palmitoylated and nonpalmitoylated SP-C forms are tilted in membranes (20). It has also been proposed that SP-C contains a dimerization motif (29, 44, 45), and there is some evidence supporting the importance of SP-C oligomerization along intracellular pathways of SP-C sorting and trafficking (46). The dimerization of SP-C through its AxxxG motif at its C-terminal segment could generate a dimer with a conical configuration as illustrated in Fig. 6, which could be supported by the findings for SP-C tilting.

Figure 6.

Model of the potential mechanism of membrane curvature and budding induced by SP-C in lipid bilayers. To see this figure in color, go online.

Transmembrane proteins with a conical shape have been shown to generate membrane stress that can be relieved upon induction of curvature (47). Such a process would be consistent with our results for SP-C and the nonpalmitoylated recombinant version, which share the same α-helical structure. This would also explain the potential importance of an extreme rigidity at the SP-C transmembrane helix (48), which is sustained by its unusual and highly conserved polyVal-like sequence. Such a rigid structure might be required to force changes in membrane curvature.

It was previously proposed that the N-terminal segment of SP-C is also able to promote protein-protein interactions (49, 50), which could progressively induce the accumulation of SP-C at high densities, an increase in curvature, and ultimately the segregation of small fissioned vesicles. The invariant size of small vesicles induced by SP-C in all the lipid systems assayed in this work could be a mere consequence of the geometry imposed by protein-lipid segregation. Furthermore, our results suggest that cholesterol formed part of these SP-C-promoted, highly curved structures. This could be the basis for a selective refinement of surfactant; however, the functional or regulatory role played by SP-C in this process remains to be explored.

Here, we observed that when SP-B was present in membranes, the SP-C membrane fragmentation activity was no longer detected, at least with the same efficiency. This could be explained by differences in membrane curvature. The effects of SP-B and SP-C (fusion and budding) would involve the concurrence of opposite curvatures, which would balance out in the membrane. SP-B may therefore counteract the SP-C effect, such that an increase in SP-C levels could be needed, at least locally, to achieve the same behavior as observed when no SP-B is present.

In this study, we used an amount of SP-C that is beyond the natural abundance of this protein in surfactant. On average, SP-C represents 1% of surfactant mass. However, SP-C is not distributed uniformly along the LS structure. SP-C has been shown to be located at the boundaries of Lo regions (51) and liquid-condensed phases in DPPC monolayers (52), although it is also present throughout the Ld and liquid-expanded phases (2, 52). Our experimental conditions would therefore resemble what happens in membrane regions where SP-C is locally confined. In such situations, the association of several dimers could lead to the formation of highly curved structures, which might bud from the membranes or interfacial films with different consequences (Fig. 5). A possible trigger for this local increase in SP-C levels is compression of the interfacial film during exhalation. In this scenario, SP-C could be transiently concentrated until it becomes excluded from the monolayer, when it would transfer either to the bilayers forming part of the surfactant reservoir or to the subphase, where small SP-C-containing vesicles would be prone to recapture by pneumocytes or macrophages.

Membrane curvature has an effect on cholesterol motion

Our results also highlight the importance of membrane curvature for CTL and, by extension, cholesterol distribution and diffusion. This has implications that go beyond the LS system, being especially relevant for membrane trafficking and mechanical events in which marked changes in curvature take place, such as membrane fission and fusion (endo- and exocytosis), lamellipodia formation, and cell migration. In this sense, membrane curvature may constitute a factor that is capable of modulating cholesterol distribution and mobilization, and consequently membrane viscoelastic behavior, with potential significance for the different mechanisms that regulate crucial cell processes.

Conclusions

Taken together, the evidence provided by this work sheds some light on the close relationship between SP-B and SP-C, and the importance of their combined action on the distribution of cholesterol in the context of LS. The SP-C-mediated changes in membrane curvature observed in this study affect cholesterol (by analogy to CTL) distribution significantly, which could be relevant for LS dynamics and refinement. Further research on the effect of membrane curvature on cholesterol diffusion, as well as the molecular mechanism by which SP-C exerts its function, could provide the central piece in the puzzle of how lipid-protein interactions govern LS function.

Author Contributions

N.R. designed and performed the research and analyzed the data. T.K.M.N. and J.P.S. designed and supervised the research, contributed tools, and discussed the data. J.P.-G. and B.G.-A. designed and supervised the research and discussed the data. All authors wrote the manuscript and approved its final version.

Editor: Andreas Engel.