Stabilization of Phospholipid Multilayers at the Air−Water Interface by

9190

Langmuir 2004, 20, 9190-9197

Stabilization of Phospholipid Multilayers at the Air-Water Interface by Compression beyond the Collapse: A BAM, PM-IRRAS, and Molecular Dynamics Study J. Saccani,† S. Castano,† F. Beaurain,‡ M. Laguerre,‡ and B. Desbat*,† Laboratoire de Physico-Chimie Mole´ culaire, UMR 5803-CNRS, 351 cours de la Libe´ ration, 33405 Talence Cedex, France, and Institut Europe´ en de Chimie et Biologie, UMR 5144-CNRS, 2 rue Robert Escarpit, 33607 Pessac Cedex, France Received April 21, 2004. In Final Form: June 29, 2004 Compression beyond the collapse of phospholipid monolayers on a modified Langmuir trough has revealed the formation of stable multilayers at the air-water interface. Those systems are relevant new models for studying the properties of biological membranes and for understanding the nature of interactions between membranes and peptides or proteins. The collapse of 1,2-dimyristoyl-sn-glycero-3-phosphocholine (DMPC), 1,2-di[cis-9-octadecenoyl]-sn-glycero-3-[phospho-L-serine] (DOPS), 1,2-di[cis-9-octadecenoyl]-snglycero-3-phosphocholine (DOPC), and 1,2-di[cis-9-octadecenoyl]-sn-glycero-3-[phospho-1-rac-glycerol] (DOPG) monolayers has been investigated by isotherm measurements, Brewster angle microscopy (BAM), and polarization modulation infrared reflection-absorption spectroscopy (PM-IRRAS). In the cases of DMPC and DOPS, the collapse of the monolayers revealed the formation of bilayer and trilayer structures, respectively. The DMPC bilayer stability has been analyzed also by a molecular dynamics study. The collapse of the DOPC and DOPG systems shows a different behavior, and the Brewster angle microscopy reveals the formation of luminous bundles, which can be interpreted as diving multilayers in the subphase.

Introduction Langmuir monolayers have been used as models of biological membranes for a long time. Easy to realize, these systems are an interesting alternative for studying membrane properties and their interactions with peptides or proteins.1-4 Indeed, the amphipathic phospholipid molecules are the main component of membranes and are well adapted to form monomolecular films with compositions close to those of natural membranes. The fluidity of those films can be controlled with the temperature, the nature of the lipid, and the compression barrier of the Langmuir trough that allows variation of the surface pressure. Moreover, the aqueous subphase environment can be modulated to reproduce the physiological conditions of cells and real hydration of polar heads. The study of Langmuir films can be performed with many surface techniques such as isotherm5 and surface potential measurements6 and various optical techniques such as Brewster angle microscopy (BAM),7-10 fluorescence * Corresponding author. Phone: 33 5 40 00 63 64. Fax: 33 5 40 00 66 45. E-mail: [email protected]. † Laboratoire de Physico-Chimie Mole ´ culaire, UMR 5803-CNRS. ‡ Institut Europe ´ en de Chimie et Biologie, UMR 5144-CNRS. (1) Gericke, A.; Flach, C. R.; Mendelsohn, R. Biophys. J. 1997, 73, 492-499. (2) Flach, C. R.; Prendergast, F. G.; Mendelsohn, R. Biophys. J. 1996, 70, 539-546. (3) Cornut, I.; Desbat, B.; Turlet, J. M.; Dufourcq, J. Biophys. J. 1996, 70, 305-312. (4) Castano, S.; Desbat, B.; Laguerre, M.; Dufourcq, J. Biochim. Biophys. Acta 1999, 1416, 176-194. (5) Gaines, G. L. Insoluble monolayers at liquid-gas interfaces; WileyInterscience: New York, 1966. (6) Dynarowicz-Latka, P.; Dhanabalan, A.; Oliveira, O. N. Adv. Colloid Interface Sci. 2001, 91, 221-263. (7) He´non, S.; Meunier, J. Rev. Sci. Instrum. 1991, 62 (4), 936-939. (8) Ho¨nig, D.; Mo¨bius, D. J. Phys. Chem. 1991, 95 (12), 4590-4562. (9) Rivie`re, S.; He´non, S.; Meunier, J.; Schwartz, D. K.; Tsao, M. W.; Knobler, C. M. J. Chem. Phys. 1994, 101 (11), 10045-10051. (10) de Mul, M. N. G.; Mann, J. A. Langmuir 1998, 14, 2455-2466.

microscopy,11-15 ellipsometry,16 X-ray17-22 and neutron reflectivity,23-25 and Raman,26-28 IR,29-36 and ultravioletvisible6 spectroscopy. Although there are numerous ad(11) Moy, V.; Keller, D. J.; Gaub, H. E.; McConnell, H. M. J. Phys. Chem. 1986, 90, 3198-3202. (12) Mohwald, H. Annu. Rev. Phys. Chem. 1990, 41, 441-476. (13) Moore, B. G.; Knobler, C. M.; Akamatsu, S.; Rondelez, F. J. Phys. Chem. 1990, 94, 4588-4595. (14) McConnell, H. M. Annu. Rev. Phys. Chem. 1991, 42, 171-195. (15) Qiu, X.; Ruiz-Garcia, J.; Stine, K. J.; Knobler, C. M. Phys. Rev. Lett. 1991, 67 (6), 703-706. (16) Ducharme, D.; Max, J. J.; Salesse, C.; Leblanc, R. M. J. Phys. Chem. 1990, 94, 1925-1932. (17) Dutta, P.; Peng, J. B.; Lin, B.; Ketterson, J. B.; Prakash, M.; Georgopoulos, P.; Ehrlich, S. Phys. Rev. Lett. 1987, 58, 2228-2231. (18) Kjaer, K.; Als-Nielsen, J.; Helm, C. A.; Laxhuber, L. A.; Mo¨hwald, H. Phys. Rev. Lett. 1987, 58, 2224-2227. (19) Tippmann-Krayer, P.; Mohwald, H. Langmuir 1991, 7, 23032306. (20) Durbin, M. K.; Malik, A.; Ghaskadvi, R.; Shih, M. C.; Zschack, P.; Dutta, P. J. Phys. Chem. 1994, 98, 1753-1755. (21) Kaganer, V. M.; Mo¨hwald, H.; Dutta, P. Rev. Mod. Phys. 1999, 71 (3), 779-819. (22) Lavoie, H.; Desbat, B.; Vaknin, D.; Salesse, C. Biochemistry 2002, 41, 13424-13434. (23) Bradley, J. E.; Lee, E. M.; Thomas, R. K.; Willatt, A. J.; Penfold, J.; Ward, R. C.; Gregory, D. P.; Waschkowski, W. Langmuir 1988, 4, 821-826. (24) Grundy, M. J.; Richardson, R. M.; Roser, S. J.; Penfold, J.; Ward R. C. Thin Solid Films 1988, 159, 43-52. (25) Penfold, J.; Thomas, R. K. J. Phys.: Condens. Matter 1990, 2, 1369-1412. (26) Chamberlain, J. R.; Pemberton, J. E. Langmuir 1997, 13, 30743079. (27) Kawai, T.; Umemura, J.; Takenaka, T. Chem. Phys. Lett. 1989, 162, 243-247. (28) Castaings, N.; Blaudez, D.; Desbat, B.; Turlet, J.-M. Thin Solid Films 1996, 284-285, 631-635. (29) Dluhy, R. A. J. Phys. Chem. 1986, 90, 1373-1379. (30) Dluhy, R. A.; Wright, N. A.; Griffiths, P. R. Appl. Spectrosc. 1988, 42, 138-142. (31) Hunt, R. D.; Mitchell, M. L.; Dluhy, R. A. J. Mol. Struct. 1989, 214, 93-109. (32) Dluhy, R. A.; Reilly, K. E.; Hunt, R. D.; Mitchell, M. L.; Mautone, A. J.; Mendelsohn, R. Biophys. J. 1989, 56, 1173-1181. (33) Blaudez, D.; Buffeteau, T.; Cornut, J.-C.; Desbat, B.; Escafre, N.; Pe´zolet, M.; Turlet, J. M. Appl. Spectrosc. 1993, 47, 869-874. (34) Sakai, H.; Umemura, J. Langmuir 1998, 14, 6249-6255.

10.1021/la0489920 CCC: $27.50 © 2004 American Chemical Society Published on Web 09/08/2004

Stabilization of Phospholipid Multilayers

vantages of Langmuir monolayers, their major drawback with respect to natural membranes is their asymmetric organization related to their thickness being two times smaller. Consequently, the repartition of the hydrophobic and hydrophilic domains is different between the two systems. For those reasons, the stabilization of phospholipid bilayers or trilayers at the air-water interface will be a more adequate model to study, with the same techniques, the interactions of membranes with molecules and peptides or proteins. In this paper, we present a new method to realize, at the air-water interface, stable phospholipid multilayers that reproduce membrane thickness and organization. To achieve this challenge, we used the properties of Langmuir monolayers to collapse at high surface pressure. This phenomenon, whose mechanism is not yet completely elucidated, is associated with the transition from a bidimensional to a tridimensional system. Up to now, the two main collapse mechanisms described at the air-water interface were the formation of aggregates on the water surface6,37-39 and the formation of folds in the subphase.40-42 In our case, we analyzed the compression of various phospholipid monolayers beyond the collapse up to very low molecular areas on a modified Langmuir trough. Indeed, to avoid rupture of the interface at very high surface pressures, this method requires a new compression barrier which allows one to keep the water level below the trough edges. Our study has been focused on the 1,2-dimyristoyl-sn-glycero-3-phosphocholine (DMPC), 1,2-di[cis-9-octadecenoyl]-sn-glycero-3-[phosphoL-serine] (DOPS), 1,2-di[cis-9-octadecenoyl]-sn-glycero-3phosphocholine (DOPC), and 1,2-di[cis-9-octadecenoyl]sn-glycero-3-[phospho-1-rac-glycerol] (DOPG) systems in the fluid phase. Brewster angle microscopy (BAM), polarization modulation infrared reflection-absorption spectroscopy (PM-IRRAS), and molecular dynamics simulations have been used to characterize the thickness, organization, and molecular orientation of the multilayer films formed. Experimental Section 1,2-Dimyristoyl-sn-glycero-3-phosphocholine (DMPC), 1,2-di[cis-9-octadecenoyl]-sn-glycero-3-phosphocholine (DOPC), and 1,2-di[cis-9-octadecenoyl]-sn-glycero-3-[phospho-1-racglycerol] (DOPG) were purchased from Sigma (Saint Louis, MO). 1,2-Di[cis-9-octadecenoyl]-sn-glycero-3-[phospho-L-serine] (DOPS) was purchased from Avanti Polar Lipids (Alabaster, AL). The solutions of phospholipids were prepared in chloroform (SDS, France) at a concentration of 5 × 10-3 M. The H2O subphase was obtained from a MilliQ (Millipore, Molsheim, France) system with a nominal resistivity of 18.2 MΩ‚cm. Melittin was purchased from SERVA (SERVA Electrophoresis, Heidelberg, Germany). The compression of phospholipid monolayers was performed on a modified Langmuir trough (type 611, Nima Technology, Coventry, England) equipped with a new in-lab-manufactured compression barrier. The flat normal barrier lies on the trough (35) Flach, C. R.; Gericke, A.; Mendelsohn, R. J. Phys. Chem. B 1997, 101, 58-65. (36) Dicko, A.; Bourque, H.; Pe´zolet, M. Chem. Phys. Lipids 1998, 96, 125-139. (37) Friedenberg, M. C.; Fuller, G. G.; Frank, C. W.; Robertson, C. R. Langmuir 1994, 10, 1251-1256. (38) de Mul, M. N. G.; Mann, J. A. Langmuir 1994, 10, 2311-2316. (39) Xue, J.; Jung, C. S.; Kim, M. W. Phys. Rev. Lett. 1992, 69, 474477. (40) Blaudez, D.; Turlet, J. M.; Dufourcq, J.; Bard, D.; Buffeteau, T.; Desbat, B. J. Chem. Soc., Faraday Trans. 1996, 92 (4), 525-530. (41) Kale´, L.; Skeel, R.; Bhandarkar, M.; Brunner, R.; Gursoy, A.; Krawetz, N.; Phillips, J.; Shinozaki, A.; Varadarajan, K.; Schulten, K. J. Comput. Phys. 1999, 151, 283-312. (42) Jorgensen, W. L.; Chandrasekhar, J.; Madura, J. D.; Impey, R.; Klein, M. L. J. Chem. Phys. 1983, 79, 926-935.

Langmuir, Vol. 20, No. 21, 2004 9191

Figure 1. Schematic of a Langmuir trough and the location of the water level with (a) a normal compression barrier and (b) a new embedded barrier. edges (Figure 1a). The new one (Figure 1b), with two notches, can be embedded in the trough. Thus, the water level can be adjusted beneath the trough edges. This configuration prevents the rupture of the liquid meniscus at very high surface pressures. The speeds of compression were 5 × 10-2 and 2 × 10-2 Å2/(molecule‚s) for DMPC and DOPS and DOPC and DOPG, respectively. The surface pressure (π) was monitored by a Wilhelmy surface balance using a filter paper plate (Whatman). The experiments were carried out at 28 °C for DMPC and at 24 °C for DOPS, DOPC, and DOPG. The trough was thermostated by a water circulating bath (Bioblock Scientific, Illkirch, France). To obtain the initial monolayers, the phospholipid solutions were spread at the air-water interface using a Hamilton microsyringe according to the Langmuir method. The surfaces of the films were observed during the compression using the Brewster angle microscope BAM2plus (NFT, Go¨ttingen, Germany) equipped with a frequency doubled Nd:YAG laser with a wavelength of 532 nm and a charge-coupled device (CCD) camera with a ×10 magnification lens. The exposure time (ET), depending on the image luminosity, was adjusted from 20 to 2 ms to avoid saturation of the camera. The spatial lateral resolution of the Brewster angle microscope is 2 µm, and the image size is 400 × 650 µm. The BAM images are coded in gray level. To determine the thickness of the layer at the surface, we used the calibration procedure of the BAM software that determines the linear function between the reflectance and the gray level. This function is established by comparison between the experimental curve of the gray level as a function of the incidence angle and the Fresnel curve (curve of the reflectance as a function of the incidence angle) that can be fitted by a parabola around the Brewster angle minimum. From the reflectance value, the BAM thickness model allows evaluation of the thickness of the layer at the surface with the knowledge of the experimental Brewster angle and the optical index of the film. This model is based on the proportionality relation between the reflectance and the square of the interfacial film thickness when the optical index of the film is assumed constant.10 Moreover, with Brewster angle microscopy, information on the fluidity of the film can be obtained by observing the geometry of the domains at the water surface. The PM-IRRAS spectra were recorded on a Nicolet 870 Fourier transform infrared (FT-IR) spectrometer with a spectral resolution of 8 cm-1 and a two-level zero-filling. The details of the optical setup, the experimental procedure, and the two-channel processing of the detected intensity have been already described.40 Minimization and molecular dynamics were performed running NAMD, version 2.5.41 The construction of monolayer and bilayer models has been carried out in CHARMM, version 27b1. Analyses were performed within the DeCipher and analysis modules of INSIGHT II, and dihedral angles were measured with an in-house Fortran program. To set up monolayers and bilayers at the air-water interface, we used the CHARMM27 model and topology file for the DMPC molecules along with the CHARMM modified TIP3P water model for the water solvent.42 A (6 × 6) DMPC monolayer and a 2(6 × 6) DMPC bilayer were built on a hexagonal lattice with an initial molecular area of 45 Å2. The DMPC molecules facing the water interface were soaked with water (at least 25 H2O molecules per lipid). In the case of the bilayer, the DMPC molecules facing the air interface were hydrated with around four H2O molecules per polar head. The

9192

Langmuir, Vol. 20, No. 21, 2004

Saccani et al.

Figure 3. Isotherm of a DMPC monolayer compressed beyond the collapse. The BAM images of Figure 4 have been recorded at the a, b, c, and d surface pressures. each full electrostatic evaluation, with short-range nonbonded forces being evaluated each time step. Simulations have been performed at constant surface tension using a modified Nose´-Hoover method in which Langevin dynamics is used to control fluctuations in the barostat. The bilayer system was submitted to a simulation with a surface tension target of 48 dynes/cm (experimental value), and the monolayer system was submitted to two simulations with surface tension targets of 48 dynes/cm (in order to compare with the bilayer system) and 30 dynes/cm (experimental value), respectively. The accessory parameters were the following: Langevin piston pressure control is ON, the cell is flexible, the target pressure along the z axis is 1 bar, the oscillation period of the Langevin piston is 200 fs, and the damping time scale of the piston is 100 fs.

Results

monolayer and bilayer were then placed in the centers of 37.5 × 43.3 × 130 Å and 37.5 × 43.3 × 190 Å boxes, respectively (see Figure 2 of the final system for the bilayer). The whole assemblies (13 986 and 23 424 atoms for the monolayer and bilayer, respectively) were treated with periodic boundary conditions and thoroughly minimized using a gradient method during 5000 steps. The resulting systems were then submitted to a 3 ns molecular dynamics run with the following starting conditions: temperature of 300 K and reassignment every 50 steps, cutoff of 14 Å and switching ON, SHAKE43 algorithm is ON but only on water hydrogens, nonbonded interactions were reassigned every 20 steps, a snapshot was written every 1 ps, and a restart file was written every 104 steps. The computing of electrostatic forces was performed using a multiple time stepping integration scheme with a time step of 1 fs and four time steps being used between

DMPC System. The compression isotherm of the DMPC monolayer is reported in Figure 3. From a 100 Å2 up to 40 Å2 molecular area (before the collapse), the surface pressure shows a peculiar progressive increase.44 After the collapse, the isotherm exhibits a plateau at 51 mN/m and the compression is stopped at an apparent molecular area of 14 Å2. In Figure 4 are reported the BAM images corresponding to significant changes of the surface morphology during the compression. Figure 4a shows the aspect of the monolayer at 40 mN/m with an exposure time of 20 ms; before the collapse, the monolayer displays a constant luminosity. This gray level has been progressively reached during the increase of the surface pressure. At low area, the collapse is characterized by the apparition of small homogeneous circular domains (D1 domains) with higher luminosity compared to that of the monolayer (Figure 4b). To avoid saturation of the camera, those domains require a shorter exposure time of 4 ms, and then appear gray dispersed in the black monolayer (Figure 4b-d). Under the compression, they grow up and become larger and larger. After the end of the compression, they coalesce to form still larger domains (Figure 4d), stable during all of the experiment time (several hours). However, the D1 domains do not totally cover the surface and some D2 domains of higher intensity appear. Without doubt, those D2 domains, characteristic of thicker multilayers, prevent the formation of a full bilayer. From the BAM calibration procedure, the reflectance intensity of the circular D1 domains was evaluated around 6 × 10-6. By comparison, the reflectance intensity of a DMPC monolayer at 35 mN/m, in the same experimental

(43) Ryckaert, J.-P.; Ciccotti, G.; Berendsen, H. J. C. J. Comput. Phys. 1977, 23, 327-341.

(44) Mingotaud, A. F.; Mingotaud, C.; Patterson, L. K. Handbook of monolayers; Academic: San Diego, CA, 1993.

Figure 2. Topology of the bilayer system just after the minimization process and before any molecular dynamics study. The DMPC molecules are shown with gray sticks (the hydrogens have been omitted for clarity), and the water molecules are shown with thin black lines. The simulation box is outlined with black lines.

Stabilization of Phospholipid Multilayers

Langmuir, Vol. 20, No. 21, 2004 9193

Figure 5. PM-IRRAS spectra of crystalline DMPC (dashed line), a DMPC monolayer at 30 mN/m (thin solid line), and a DMPC monolayer compressed at 51 mN/m (thick solid line). Figure 4. Brewster angle microscopy images during compression beyond the collapse of a DMPC monolayer: (a) compact monolayer at 40 mN/m (ET ) 20 ms); (b and c) coexistence of black monolayer domains with gray bilayer circular domains (ET ) 4 ms) after the collapse; (d) coalescence of bilayer domains after the end of compression (ET ) 4 ms).

conditions, was previously evaluated around 10-6. If we assume the same refraction indexes between the monolayer and the D1 domain, we can deduce a thickness ratio of 2.4 from the values of their intensities. This ratio is in favor of the formation of DMPC bilayers domains (D1 domains) at the air-water interface in those collapse conditions. Furthermore, by using the BAM thickness model, with an experimental Brewster angle of 53.05 and an optical index of 1.45 (i.e., the optical index of crystalline DMPC), the thickness of the circular domains is estimated at 39 Å. This value is consistent with literature values determined by diffraction45 and small-angle neutron scattering46 measurements for DMPC bilayers. Moreover, the circular geometry of the domains indicates that DMPC bilayers keep their fluid state (by analogy to fluid spherical tridimensional systems). The PM-IRRAS spectra of DMPC at the air-water interface under the same conditions of surface pressure are presented in Figure 5. The intensities of the antisymmetric (νasCH2) and symmetric (νsCH2) stretching bands, located at 2920 and 2850 cm-1, respectively, are multiplied by ∼1.6 for the system at 51 mN/m with respect to the monolayer at 30 mN/m. This value, slightly inferior to 2, confirms that the trough surface is not totally covered by a bilayer and that monolayer domains subsist. Furthermore, the PM-IRRAS results are in agreement with the BAM results and mean that fluid DMPC bilayers are partially formed at that high surface pressure. Moreover, the small shift of the νsCH2 band to low frequencies (∆ν ) 0.5 cm-1) for the system at 51 mN/m indicates that the DMPC bilayer is slightly more rigid than the monolayer. The main difference between the two infrared spectra, at 30 and 51 mN/m, comes from the apparition at high pressure of a well-defined positive band around 2950 cm-1 corresponding to the antisymmetric methyl stretching vibration (νasCH3) and a negative νsCH3 band located at 2870 cm-1 characteristic of the symmetric methyl stretching vibration. According to the PM-IRRAS selection (45) Nagle, J. F.; Tristan-Nagle, S. Biochim. Biophys. Acta 2000, 1469, 159-195. (46) Balgavy, P.; Dubnickova, M.; Kucerka, N.; Kiselev, M. A.; Yaradaikin, S. P.; Uhrikova, D. Biochim. Biophys. Acta 2001, 1512, 40-52.

Figure 6. Isotherm of a DOPS monolayer compressed beyond the collapse. The BAM images of Figure 7 have been recorded at the a, b, c, and d surface pressures.

rule,33,40 the positive and negative signs of those two bands, respectively, suggest a methyl axis orientation more vertical in the bilayer than in the monolayer. DOPS System. The compression beyond the collapse carried out with DOPS shows a similar behavior to that for DMPC. The isotherm (Figure 6) presents a progressive increase of the surface pressure and a plateau around 44 mN/m before and after the collapse, respectively. Nevertheless, in this case, we observe an abrupt increase of the surface pressure just before the end of the compression. In the BAM images (Figure 7), two differences can be noticed. The circular domains have a higher intensity because they appear medium gray for an exposure time as low as 2 ms, indicating that their thickness is larger than that for the DMPC bilayers. Furthermore, contrary to the DMPC system, entire coverage of the trough surface is observed. The abrupt increase of the surface pressure observed on the isotherm corresponds to this phenomenon. If the compression is stopped at this stage, the system is stable for several hours. For lower molecular areas, we observed the formation of domains still more luminous, corresponding to multilayers with higher thicknesses (results not shown). Using the BAM calibration procedure, we determined a reflectance intensity of 1.6 × 10-5 for the circular domains. This value has to be compared with the 1.4 × 10-6 reflectance intensity of a DOPS monolayer at a surface pressure of 32 mN/m. The ratio of both intensities corresponds to a thickness ratio of 3.4, suggesting that a DOPS trilayer is formed in those collapse conditions. This result was confirmed by PM-IRRAS spectra (Figure 8). The intensities of the νasCH2, νsCH2, and νCdO stretching bands are multiplied by 3 between the monolayer and the

9194

Langmuir, Vol. 20, No. 21, 2004

Saccani et al.

Figure 7. Brewster angle microscopy images during compression beyond the collapse of a DOPS monolayer: (a) initial monolayer at 21 mN/m (ET ) 20 ms); (b) trilayer circular domains at the collapse (2 ms); (c) coalescence of trilayer domains after the collapse (2 ms); (d) the entire trough surface is covered by a trilayer (2 ms).

Figure 9. Isotherms of (a) a DOPC monolayer compressed beyond the collapse and (b) a DOPG monolayer compressed beyond the collapse. The BAM images of Figure 10 have been recorded at the a, b, a′, and b′ surface pressures.

Figure 8. PM-IRRAS spectra of a DOPS monolayer at 26 mN/m (dashed line) and of the trilayer (solid line): (a) in the alkyl chain vibration range; (b) in the polar headgroup vibration range.

fully covered multilayer water surface. In addition, in Figure 8b, we observe a significant 3-fold increase of the dip located at 1660 cm-1. This dip is an optical effect peculiar to IRRAS29,40 due to the strong dispersion of the refractive index of water in this spectral range (this dispersion is caused by the bending mode δH2O). Previous numeric simulations have showed that, in first approximation, this dip is proportional to the layer thickness at the water surface.47 Thus, in our case, this increase

confirms the formation of a three-layer system under those collapse conditions. Contrary to the DMPC bilayer, no change in frequency is observed in the characteristic absorption bands (νasCH2, νsCH2, and νCdO) of the trilayer system. This suggests that organizations of DOPS monolayers and trilayers are very close together. DOPC and DOPG Systems. To complete our study, we have performed the same experiments with two other phospholipids, DOPC and DOPG, also in the fluid phase at ambient temperature. Their isotherms (Figure 9) present a behavior close to that of DMPC and DOPS. However, BAM images (Figure 10) show that the collapse of the monolayers leads, whatever the exposure time, to the formation of very luminous bundles flickering at the water surface. The nature of these systems does not allow determination of their thickness and organization by BAM and PM-IRRAS measurements. Nevertheless, the appearance of the luminous bundles at the surface is very similar to the results observed for the collapse of the 2-OH TCA (2-hydroxy tetracosanoı¨c acid).48-50 This analogy suggests that bundles can be assigned to the formation of multilayers in the subphase and flickers are due to reflections on those diving multilayers. Discussion The BAM and PM-IRRAS thickness measurements are consistent with the formation of a fluid DMPC bilayer (47) Buffeteau, T.; Blaudez, D.; Pere, E.; Desbat, B. J. Phys. Chem. B 1999, 103 (24), 5020-5027. (48) Ybert, C.; Lu, W.; Mo¨ller, G.; Knobler, C. M. J. Phys. Chem. B 2002, 106, 2004-2008. (49) Ybert, C.; Lu, W.; Mo¨ller, G.; Knobler, C. M. J. Phys.: Condens. Matter 2002, 14, 4753-4762. (50) Lu, W.; Knobler, C. M.; Bruinsma, R. F. Phys. Rev. Lett. 2002, 89 (14), 146107.

Stabilization of Phospholipid Multilayers

Figure 10. Brewster angle microscopy images during compression beyond the collapse of (a) DOPC and (a′) DOPG monolayers. Very luminous sparkling dots appear after the collapse at the surfaces of the (b) DOPC and (b′) DOPG monolayers.

and a fluid DOPS trilayer at the air-water interface. However, those techniques cannot give information on the relative orientation of molecules in each layer of the multilayer systems. The suggested structures presented below for each system are based on thermodynamical considerations for the DOPS trilayer and molecular dynamics simulations for the DMPC bilayer. In the case of DOPS, isotherms and BAM images are comparable with results obtained on the collapse of the liquid crystal 8-CB (4′-n-octyl-4-cyanobiphenyl). Different studies at the air-water interface on liquid crystal monolayers compressed beyond the collapse37-39 have revealed the formation of a trilayer. The trilayer is composed of a bilayer and a monolayer with tails pointed in the air above. The similarity of the BAM results between DOPS and 8-CB suggests that the DOPS trilayer structure is very close to the 8-CB trilayer structure. Furthermore, this structure can explain the thermodynamic stability of the DOPS trilayer during the experiment. Concerning the DMPC bilayer organization, PM-IRRAS results showed that the formation of a second layer on top of the monolayer leads to a slightly more rigid system. Moreover, additional interaction at the methyl extremities appears, suggesting a model where the alkyl chains of each layer in the bilayer are in contact. This tail-to-tail contact organization seems to be more stable than the organization where the polar headgroups of the top layer are in interaction with the alkyl chains of the bottom layer. Obviously, the tail-to-tail contact organization supposes that a hydration layer exists at the polar head level to minimize repulsion between the polar heads of the top layer and the air. To confirm our hypothesis, molecular dynamics simulations were carried out on a DMPC monolayer at 30 mN/m and a bilayer made up of tail-to-tail contact organization at 48 mN/m. To facilitate comparison, we carried out also

Langmuir, Vol. 20, No. 21, 2004 9195

a molecular dynamics simulation on a DMPC monolayer at 48 mN/m. Analysis of the dynamics allowed the determination of relevant structural parameters. The simulation results are summarized in Table 1. Only small differences are visible for the average molecular area between the monolayers and the bilayer, suggesting that the formation of a second layer beyond the initial monolayer does not change dramatically the organization of the system. Concerning the alkyl chains, we have considered the following two parameters: the number of gauche conformations per DMPC molecule and the tilt relative to the normal of the interface. To evaluate the number of gauche conformations, we have only considered the 10 equivalent dihedral angles for the two SN1 and SN2 alkyl chains of DMPC without taking into account the dihedral angle adjacent to the carboxyl group, whose structure induces a very different value compared to the others. Under these conditions, the number of gauche conformations is almost identical for the three simulated systems, with an average value of 3.1 ( 0.3 gauche conformations per DMPC molecule (i.e., an average value of 1.55 gauche conformations per chain). This is a low value and very likely a direct consequence of the low specific area value for each DMPC. Experimentally, in a fluid bilayer of DMPC, the number of gauche conformations per chain is found between 2.651 and 3.252 using the FT-IR technique. This result has been also confirmed by a molecular dynamics study of a DMPC bilayer in the fluid phase (2.9 gauche conformations per chain).53 The main difference between the three simulated systems has been found between the layers simulated at 48 mN/m and the one simulated at 30 mN/m. From the graphs in Figure 11, it is obvious that the gauche content profile is almost identical for the three simulated systems except for the last dihedral angle (i.e., the one involving the methyl terminal groups) which exhibits clearly a higher gauche content when the systems are simulated at 48 mN/m. Examination of Table 1 shows a tilt weaker for the bilayer than for the monolayer, meaning that the bilayer alkyl chains are more straightened. This result is in agreement with a straighter alkyl chain direction consistent with a system more rigid for the bilayer. A difference of 1° is also observed for the orientation of the carbonyl group with respect to the interface plane. In view of those results, the molecular orientation differences between the monolayer and the bilayer evaluated from molecular dynamics simulations are very weak, which is in agreement with the experimental PM-IRRAS spectra where only a shift of 1 cm-1 for the νsCH2 band has been observed for the bilayer system. The molecular dynamics study shows especially that the bilayer with a hydration of four H2O molecules per polar head and at a surface pressure of 48 mN/m is stable after a 3 ns simulation run with an organization very close to the monolayer but however slightly more rigid, according to experimental observations. From those preliminary molecular dynamics results, the tail-to-tail contact organization seems to be a reasonable hypothesis for explaining the stability of the

Table 1. Average Values of Calculated Parameters from the Molecular Dynamics Calculation during the Last 1 ns of Each Simulation

monolayer at 30 mN/m monolayer at 48 mN/m bilayer at 48 mN/m

molecular area (Å2) (standard deviation)

tilt (deg) (standard deviation)

CO orientation relative to the plane interface (deg) (standard deviation)

44.72 ( 0.22 44.84 ( 0.20 44.07 ( 0.11

17.9 ( 7.5 15.5 ( 9.3 14.8 ( 8.3

18.2 ( 14.4 20.0 ( 15.1 19.5 ( 15.4

9196

Langmuir, Vol. 20, No. 21, 2004

Saccani et al.

Figure 11. Percentage of gauche conformation on carbon atoms of the SN1 and SN2 chains of DMPC: (b) monolayer at 30 mN/m; (9) monolayer at 48 mN/m; (2) bilayer at 48 mN/m.

observed systems as it would be suggested by the experimental results. The simulations for DOPS were not performed because the serine polar head parameters are not available at the present time for the CHARMM field force. As previously written, those multilayer films formed at the air-water interface by compression beyond the collapse present a very similar organization with respect to natural membranes and offer a high potentiality for investigating the nature of interactions between peptides or proteins with membranes. As a preliminary illustration, we have studied the interaction of the DOPS trilayer with melittin, a widely studied lytic amphipathic peptide from bee venom. BAM images of the trilayer before and after the injection in the subphase of a melittin solution at the concentration of 22 nM are presented in Figure 12 for peptide/lipid molar ratios of 1/300, 1/150, and 1/75. We notice that melittin causes destruction of the trilayer in (51) Casal, H. L.; McElhaney, R. H. Biochemistry 1990, 29, 54235427. (52) Tuchtenhagen, J.; Ziegler, W.; Blume, A. Eur. Biophys. J. 1994, 23, 323-335. (53) Takaoka, Y.; Pasenkiewicz-Gierula, M.; Miyagawa, H.; Kitamura, K.; Tamura, Y.; Kusumi, A. Biophys. J. 2000, 79, 3118-3138.

monolayer domains, as already observed on erythrocytes and liposomes.54-62 Moreover, the formation of aggregates increases with the molar ratio. The PM-IRRAS spectra shown in Figure 13, recorded for three increasing molar ratios, show an intensification of the amide I band located around 1660 cm-1 in the negative side. From the selection rule of PM-IRRAS, on a water surface, we can deduce that the orientation of the band relative to the baseline corresponds rather to a perpendicular melittin orientation in the trilayer.40 The orientation of this peptide in this (54) Dufourcq, J.; Faucon, J. F. Biochim. Biophys. Acta 1977, 467 (1), 1-11. (55) Dufourcq, J.; Faucon, J. F.; Fourche, G.; Dasseux, J. L.; Le Maire, M.; Gulik-Krzywicki, T. Biochim. Biophys. Acta 1986, 859 (1), 33-48. (56) Portlock, S. H.; Clague, M. J.; Cherry, R. J. Biochim. Biophys. Acta 1990, 1030 (1), 1-10. (57) Hincha, D. K.; Crowe, J. H. Biochim. Biophys. Acta 1996, 1284 (2), 162-70. (58) DeGrado, W. F.; Musso, G. F.; Lieber, M.; Kaiser, E. T.; Kezdy, F. J. Biophys J. 1982, 37 (1), 329-38. (59) Hider, R. C.; Khader, F.; Tatham, A. S. Biochim. Biophys. Acta 1983, 728 (2), 206-14. (60) Tosteson, M. T.; Holmes, S. J.; Razin, M.; Tosteson, D. C. J. Membr. Biol. 1985, 87 (1), 35-44. (61) Yianni, Y. P.; Fitton, J. E.; Morgan, C. G. Biochim. Biophys. Acta 1986, 856 (1), 91-100. (62) Dempsey, C. E. Biochim. Biophys. Acta 1990, 1031 (2), 143-61.

Stabilization of Phospholipid Multilayers

Langmuir, Vol. 20, No. 21, 2004 9197

membranes because, by using a more concentrated melittin solution (0.2 µM), we observe a very fast and astonishing destruction of the trilayer (Figure 14) similar to the behavior reported for hemolysis. Conclusion

Figure 12. Brewster angle microscopy images of the DOPS trilayer (a) before the injection of melittin and after the injection of melittin (c ) 22 nM) in the water subphase for peptide/lipid molar ratios of (b) 1/300, (c) 1/500, and (d) 1/75.

Figure 13. PM-IRRAS spectra of the DOPS trilayer (a) before the injection of melittin and after the injection of melittin for peptide/lipid molar ratios of (b) 1/210, (c) 1/105, and (d) 1/70. The spectra have been translated in the y direction for better display.

Figure 14. Brewster angle microscopy images of the DOPS trilayer (a) 30 and (b) 60 s after the injection of melittin (c ) 0.2 µM) in the water subphase for a peptide/lipid molar ratio of 1/27.

condition is different from the result obtained with the Langmuir DMPC and DMPG monolayers where the melittin was found preferentially oriented parallel to the plane of the monolayers.3 Moreover, the multilayer systems seem more appropriate for describing natural

In this paper, we have demonstrated that stable phospholipid multilayers can be formed at the air-water interface by the compression of monolayers beyond the collapse. BAM and PM-IRRAS results showed the formation of a fluid trilayer composed of a monolayer on top of a bilayer in the case of DOPS and a partial fluid bilayer in the case of DMPC. The thermodynamic consideration and molecular dynamics simulation suggest a tail-to-tail contact organization for this last system. It seems difficult to predict if one phospholipid will adopt a fluid bilayer or a trilayer organization at the air-water interface. It is unfavorable for the charged head polar group to give a bilayer organization because the counterion must be in the air, but obviously the neutral headgroup is not sufficient to obtain the bilayer. If we look more carefully at our results, the DMPC bilayer is stabilized at 28 °C which is only 4 °C above its phase transition temperature; with this condition, the defects which produce the fluidity are mainly situated in the terminal carbon of the alkyl chains. At higher temperatures with a higher percentage of defects in the chains, it is impossible to obtain a bilayer or a multilayer. The same behavior has been observed with DOPC; at room temperature, far from its phase transition temperature (+40 °C), the defects are distributed all along the chains almost like in the liquid phase and the collapse does not give a bilayer or a multilayer. Thus, it seems necessary to control the fluidity of the monolayer if we want to stabilize the bilayer at the airwater interface. For each compound, it is necessary to find the better physical conditions to obtain such organization. The results obtained with DOPC and DOPG, although they do not lead to the formation of multilayers on the water surface at room temperature, as in the cases of DMPC and DOPS, show that different mechanisms of collapse are possible, which seem to depend on the nature of the phospholipids and the physical chemistry conditions. Comparison of the preliminary results obtained for melittin in interaction with the DOPS trilayer with those obtained on a monolayer show that the hydrophobic thickness of the biological membrane plays a determinant role in the nature of the interactions with peptides. Therefore, multilayers at the air-water interface appear as relevant new model systems for the in situ study and understanding of membrane-protein interactions. It is also possible to investigate mixed phospholipid multilayers with a composition still closer to natural membranes. Furthermore, those multilayers with a similar organization compared to the biological membranes at the airwater interface open the way to their characterization by other experimental techniques such as X-ray or neutron reflectivity and ellipsometry to determine accurately their thickness and layer organization. The numeric simulation methods are also a promising way to achieve the determination of multilayer structures. LA0489920