Spin-labeled Stearic Acid Behavior and Perturbations on the Structure

Dec 4, 2008 - In one configuration, the doxyl ring lays at the lipid−lipid interface, shifting its polar head toward the bilayer center. The other e...
0 downloads 0 Views 2MB Size
16830

J. Phys. Chem. B 2008, 112, 16830–16842

Spin-labeled Stearic Acid Behavior and Perturbations on the Structure of a Gel-Phase-Lipid Bilayer in Water: 5-, 12- and 16-SASL Martı´n R. Vartorelli,† Alberto S. Garay,† and Daniel E. Rodrigues†,‡,* Departamento de Fı´sica, Facultad de Bioquı´mica y Ciencias Biolo´gicas, UniVersidad Nacional del Litoral and INTEC (UNL-CONICET), C.C. 242, Ciudad UniVersitaria, C.P. S3000ZAA, Santa Fe, Argentina ReceiVed: July 22, 2008; ReVised Manuscript ReceiVed: October 21, 2008

We have studied the effect of the insertion of spin-labeled molecules n-doxyl-stearic acid (n-SASL, n ) 5, 12, 16) on the structure and dynamics of a model lipid bilayer in gel-like phases using molecular dynamics simulations. We have studied the atomic density depth profiles and configurations of the labeled molecules in a host hydrated stearic acid bilayer system. We have found that the 5-SASL label positions its paramagnetic group at the water-lipid interface, and its polar head builds H bonds to neighboring lipids and to the solvent. 16-SASL positions its paramagnetic group at the lipid-lipid interface. The 12-SASL label presents two configurations at high lateral pressure. In one configuration, the doxyl ring lays at the lipid-lipid interface, shifting its polar head toward the bilayer center. The other equilibrium configuration of 12-SASL presents its paramagnetic group laying in the center of the compact hydrophobic region of the layer (erected configuration). It was determined that the coexistence of these two configurations is governed by the polar head-water interaction. We have found that the insertion of the labeled molecules at the concentrations used in the present work (0.36 mol %) do not perturb global properties like area per lipid, tilt angle, or order parameters. Nevertheless, there are local perturbations of the host system that are confined to a 10 Å neighboring shell around the spin label molecule. To study the interactions that determine the position of the labeled molecules in the bilayer, we performed simulations at different lateral pressures, which allowed us to extract important conclusions. Introduction One of the most-used experimental tools to characterize the dynamics and structure of lipid biomembranes during the last 20 years has been the electronic paramagnetic resonance (EPR) spectroscopy of spin labels.1 This technique has been used to obtain relevant data for lipid bilayers: dipalmitoylphosphatidylcholine (DPPC)/water model,1,2 fatty acid films, 3-5 and biological membranes.6-9 The high sensitivity of the EPR spectroscopy makes it particularly suitable for the study of interfacial phenomena if compared with other techniques such as nuclear magnetic resonance. The drawback of this experimental tool is the necessary requirement of paramagnetic centers in the system under study. These paramagnetic centers must be added by doping the sample with spin-labeled groups if, as it is the case in biomembranes, the native system does not have unpaired spins. Since for the nitroxide group of n-SASL labels the anisotropic part of the Zeeman contribution is small, the EPR spectral features derived from the energy level hyperfine splitting are directly related to the dynamics of its molecular orientation.2 The EPR spectra line shape can be related with the order parameter tensor S of the spin labeled group2 in the magnetic field frame of reference (as defined by Saupe10 to characterize the molecular order in liquid crystals). The shape of the EPR spectra contributed by each spin label molecule is mainly determined by the orientational distribution of its magnetic π-like orbital relative to the magnetic field direction and by the rotational dynamics of these orientations. The second * Corresponding author e-mail: [email protected]; phone/ fax: +54 (342) 4575-213. † Universidad Nacional del Litoral. ‡ INTEC, CONICET-UNL.

factor leads to an averaging effect of the hyperfine derived spectral features when the motion is fast enough. The narrowing of each of the hyperfine components is associated heuristically with a faster dynamics of the labeled molecule, and it is used to quantify this property. In the case of samples where the normal to the biomembrane does not possess a unique orientation, a further spatial convolution of the spectra over different spin label molecules must be done. It is usually assumed that the dynamics of the molecular environment of the labeled molecule is surveyed by the EPR spectral features. In this context we thought it was relevant to perform a detailed molecular study of the behavior of different, commonly used spin-labeled molecules in a lipid bilayer environment. Previous experimental studies have shown that the gel-liquid crystalline phase transition temperature of DMPC measured by EPR using a labeled stearic acid (n-SASL) probe is changed by the doping.11 Hakansson et al.12 have also called attention over some differences between the dynamics of the spin-label molecule and that of the lipids where they are dissolved. The state of the art of the physical chemistry of biomembranes1-4 makes necessary a more throughout understanding of how the EPR spin labels sense and perturb their environment. It is particularly relevant to determine the position where the paramagnetic group of the label located in the lipid bilayer and how it interacts with its environment. To the best of our knowledge, only two recent works deal with this subject for n-SASL labels.13,14 Aside from our own previous work, the other is a simulation of the liquid crystalline phase of a mixture of DPPC and labeled stearic acid molecules, and both are therefore complementary to the present study.

10.1021/jp806476a CCC: $40.75  2008 American Chemical Society Published on Web 12/04/2008

Spin-labeled Stearic Acid Behavior of a Lipid Bilayer Molecular dynamics (MD) simulations are a useful tool to study these effects since they can provide atomic-scale detailed information about the behavior of the layer and the spin label molecules.15-17 We have chosen as our host model system a simple stearic acid (SA) bilayer in water that has been shown14 to be in a gel-like phase at the conditions studied in the present work. SA layers are relevant on their own since the stratum corneum of the skin, the outermost layer of the epidermis, are composed of stacked fatty acid layers.18,19 Experimentally, it is also known that SA forms Langmuir-Blodgett films, which have attracted considerable interest over the past years20 in connection with highly specific sensors or devices based on molecular electronic. Multilayers of pure SA have been well-characterized experimentally in Langmuir-Blodgett films and show a regular arrangement of the fatty acid molecules.21,22 Moreover, SA is the lipid carrier of some of the most popular spin labels used to characterize lipid membranes, therefore the comparison of the variables associated with the environment and the marker are straightforward. Despite its simplicity, the SA molecule is an amphiphile with a small polar head and a long alkyl hydrophobic chain. Conversely, it has a polar head much smaller than that of phospholipids, which lead to important differences in its phase diagram. First we study the unlabeled host lipid system under different surface tension (lateral pressure) conditions, which will give us the opportunity to change the environment seen by the labeled molecules when it is doped. For pure SA layers there is experimental information from the lateral pressure-area isotherms in monolayers and X-ray diffraction of Langmuir-Blodgett (LB) films.5-8 We used in the present study as spin labels SA molecules with the doxyl ring (nitroxide 2-doxylpropane (4,4′-dimethyloxazolidine-N-oxyl)) in positions 5 (5-SASL), 12 (12-SASL), and 16 (16-SASL), which are some of the most-used doxyllabeled-SA. We have simulated the labeled molecule in its protonated or neutral state. Under experimental conditions the charge state of the spin labels depends on the pH of the surrounding solvent and can be controlled by using different buffers. The simulation of a charged state in our case where the host system is also SA would lead to several spurious problems, making it useless for the present purposes. We use the R chiral form for the configuration of the n-SASL molecules. The arrangement and dynamics of the spin label molecule in a compact system like gel-phase self-assembled lipid layers have been the subject of some quantitative and semiquantitative analysis. Wisniewska et al.11 have measured the gel-LC transition temperature in DMPC samples doped with 5-, 9-, 12- and 16-doxyl-labeled SA by EPR spectroscopy and also in unlabeled samples using calorimetric measurements. They found that the labeled DMPC samples have a lower gel-LC transition temperature as determined by EPR than that obtained from differential scanning calorimetry. The largest disagreement appears for samples labeled with the 9-SASL. It has also been suggested that in ordered phases the labeled molecules could be segregated.23 Risse et al. have studied SA Langmuir-Blodgett films of labeled and unlabeled samples using near edge X-ray absorption fine structure (NEXAFS) and EPR spectroscopy.3,4 They found that the presence of the doxyl-labeled molecules alters the structure and dynamics of the layers, although they were not able to precisely determine the label concentration in the samples. They found inconsistencies in the interpretation of the NEXAFS results for labeled layers that were concealed using qualitative arguments about the dynamics of the spin label molecule. They conclude that the motion of the labeled molecule

J. Phys. Chem. B, Vol. 112, No. 51, 2008 16831 is a cooperative effect with its surrounding and suggest an induced increase of the area per lipid. We performed MD simulations of the systems under consideration to obtain a detailed atomic scale description of the properties of the labeled and unlabeled lipid layers. The aims of this work are (a) to study how the different spin labeled molecules arrange into the lipid environment, (b) to understand how the labels perturb the bilayer behavior, and (c) to correlate the changes of the bilayer properties with those surveyed by the spin labeled molecule. We study the labeled and unlabeled systems as a function of lateral pressure at T ) 300 K and for some systems also at 315 K. The analysis of the SA bilayer properties as a function of lateral pressure allows us to characterize the structural changes of the host system. Then we study how these changes are surveyed by the spin-labeled molecule. In this work we analyze several structural and dynamic features of the labeled stearic acid molecule (n-SASL) and its molecular environment. We calculate the following properties to characterize the results of the simulated systems: (a) the average structure factor that brings information about the long-range two-dimensional periodicity; (b) the atomic density profiles that characterize the depth position of the spin labeled molecules relative to the lipids in the host bilayer; (c) the lipids and labeled molecule order parameter profiles (S) that survey the orientational order of the alkyl chains at different depths; and several other properties (e.g., averaged molecular areas, inclination angle, fraction of gauche defects) that allow us to present a coherent picture of the systems. Our results show that the studied spin label species (5-SASL, 12-SASL, and 16-SASL) behave very different from one another in the gel-like host system analyzed here. The doxyl paramagnetic group positions itself at depths in the lipid bilayer that in many cases are very different from the naive pictures usually assumed in the interpretation of experimental data. The position of the magnetic group is also at variance with that found in MD simulations of similar spin labels in the liquid crystalline phase of DPPC. Altogether these studies allow for a more throughout interpretation of the experimental information derived from the EPR spectroscopy of lipid bilayer systems. Methods The parameters used in the MD for SA were GROMOS96 43a2x, extended to better reproduce the behavior of long-tail lipids.24 Explicit atoms and hydrogens were used for all polar groups. United atoms were used for the hydrocarbons. For the CH2 and CH3 groups of the fatty acids, the Ryckaert-Bellemans torsional potentials were used, which appear to be well suited for membrane simulations. The SPC model was used for water.25 The forcefield parametrization for the labeled SA molecule have been taken from ref 14. All bonds were constrained by the LINCS algorithm.9 Periodic boundary conditions were applied in all directions. All MD simulations were performed using the GROMACS 3.3 software package.26,27 Long-range electrostatic interactions were calculated using particle mesh ”Ewald” method with a 15 Å cutoff. We use a time step of 2 fs for integration. The neighbor list was updated each 10 integration steps (20 fs). The properties calculations were done with the GROMACS package analysis tools and with FORTRAN custom-made software. The initial configuration of the system was an equilibrated bilayer of SA in water from a previous work.14 This starting system has 70 lipids in each leaflet and 1435 water molecules in the tetragonal simulation cell. In all the MD simulations

16832 J. Phys. Chem. B, Vol. 112, No. 51, 2008 performed in this work we used the Berendsen thermostat28,29 to control the temperature with a coupling time of 0.1 ps. A semi-isotropic pressure bath with a coupling time of 4 ps and a compressibility constant of 6 × 10-5 bar-1 was used. For the systems with one labeled molecule in each leaflet, the quantity of lipid was increased to 280 per layer in order to work with a lower concentration (0.36 mol %). The larger bilayer was built by joining 4 of the original cells of 70 lipids and then performing an equilibration step (2 ns) to eliminate any residual correlation originated from the starting periodicity. The labeled molecules (5-, 12-, or 16-SASL) were inserted one in each leaflet within a previously generated hole with the shape of its molecular surface to avoid large equilibration steps.30 These labeled and unlabeled larger size systems were run only at values of γ ) 1206 and 2000 bar nm. The choice of the most realistic boundary conditions for lipid membrane simulations has been largely under debate in the literature. The constant surface tension ensemble has emerged as a very convenient alternative to determine structural parameters from simulations that employ frequently used system sizes.31 Nevertheless, the appropriate value of the surface tension parameter is not an experimentally accessible quantity.32 Moreover the most convenient value of the surface tension parameter seems to be dependent on the size of the simulated membrane.33 In this work we chose values of the surface tension that lead to areas per lipid in the range of those experimentally measured in SA monolayers isotherms.34 It is well-known that the behavior of lipid monolayers can be very different from that of bilayers, particularly in the gel phase.35 Nevertheless, for some cases, such as DMPC in gel phase, the value of the area per lipid obtained from neutron scattering measurements in bilayers36 agrees with that from direct monolayer measurements.37 Regardless, this choice will approximately reproduce the average intermolecular distances found in the experimental systems. The determination of the equilibration stage is a crucial concern in membrane simulations. It is well-known that bilayers simulations started from a nonequilibrated structure can require very long times to reach the production stage,38 particularly if they started from the compact periodic structures of the crystalline phase. Our simulations have been started from an equilibrated bilayer of SA used in a previous work.14 At the new surface tension and temperature conditions studied here the equilibration was analyzed by following the evolution of the area per lipid and of the order parameter calculated in contiguous 1 ns time intervals. Nevertheless, for the labeled systems these average variables are not very sensitive to the local equilibrium around the marker due to its low concentration. For this reason the local equilibrium condition around the marker was controlled also by following the time evolution of the order parameter profiles of the labeled molecule in contiguous 1 ns intervals. The systems with labeled molecules deserve particular consideration. It is important to recall that there is only one labeled molecule in each leaflet of the bilayer. The simulation time required for these molecules to laterally diffuse to any of the allowed local environments in the layers was beyond our computational resources. Therefore, our approach was to evaluate only the behavior and properties of the labeled molecules locally averaged due to the short time scale movements in their local environments. This approximation means that the calculated properties of the labeled molecule should be interpreted by taking into account this assumption. As the averaged area and global properties of the system are well-converged under

Vartorelli et al.

Figure 1. Time evolution of the average area per lipid in unlabeled bilayers at several surface tension values (γ).

the conditions of the chosen NPZγT ensemble, these conditions are equivalent to mimic the behavior of lipid patches around spin-labels in different conformations that are in equilibrium in real systems. This approach possesses some limitations on the accuracy of the calculated averages over spin label molecule properties. Nevertheless, in spite of this limitation we think we were able to draw a significant picture of the behavior of the spin label in these systems that has not been previously reported. An alternative approach to surmount this problem has been adopted in the study of spin labels in liquid crystalline DPPC bilayers.13 These authors choose to work with a higher concentration of labeled molecules (11 mol %) to obtain an average behavior of them in different local environments. The drawback of that approach is that interactions among labeled molecules are unavoidable. Moreover, in real systems with spin label concentrations higher than 1% mol, there are magnetic interactions among probes that strongly distort the spectral EPR features used to extract information of the bilayer. These facts lead us to adopt the low concentration approach used in the present work despite the above-mentioned limitations. Results We started the work by characterizing the pure system that serves as host for the spin-labeled molecules that are studied in the sections that follow. The next subsection deals with the position and configurations of the n-SASL in the marked systems. The last two subsections of the results are dedicated to the characterization of the perturbing effects introduced by the labeled molecule and to the change of the lateral pressure in these systems. Pure SA Bilayers. The pure SA bilayer system with 70 molecules in each leaflet was simulated at surface tension values γ ) 50, 600, 1206, 1500, 1750, 1800, 1850, 1900, and 2000 bar nm at T ) 300 and 315 K. All simulations have been started from an equilibrated bilayer at 300 K and γ ) 1206 bar nm from a previous work.14 Figure 1 shows the time evolution of the mean area per lipid at 300 K for these surface tension values. For the last 2 ns of each simulation the mean area per lipid appears to be stabilized. To evaluate the arrival to the equilibrium condition, the order parameter profiles SZ,Z(i)44 have been calculated in contiguous 1 ns intervals. Their temporal evolution are shown in Figure 2 for γ ) 1900 bar nm, which seems to be the system with the longest relaxation time. The average fractions of dihedral angles in trans conformation over contiguous 1 ns intervals also show evolutions that reach stability during at least the last 2 ns of the simulations (results not shown). The last 2 ns of each simulation was therefore taken into account as the production stage to obtain averaged properties based on the above analysis.

Spin-labeled Stearic Acid Behavior of a Lipid Bilayer

J. Phys. Chem. B, Vol. 112, No. 51, 2008 16833

Figure 2. SZ,Z order parameter for the SA pure system at γ ) 1900 bar nm and 300 K, calculated for different 1 ns time intervals.

Figure 3. Snapshot of the pure SA bilayer at the start of the production stage at γ ) 2000 bar nm. Spheres correspond to the positions of the water O. The bond style molecules are the SA. The point of view of the snapshot has been chosen to make evident the tilt in the upper layer.

Figure 3 shows a snapshot of the system at 300 K and γ ) 2000 bar nm at the beginning of the production stage. It is seen from the figure that the system shows a high conformational order of the alkyl chains, a collective inclination angle of the long axis of the lipid molecules, and a periodic arrangement in the directions parallel to the interface. The same features had been found for the system at 300 K and γ ) 1206 bar nm.14 i. Analysis of the AWerage Structure Factor of Unlabeled Bilayers. We calculated the two dimensional (2D) structure factor Fj(q) to characterize the periodicity of the arrangement of the SA molecules at each value of the surface tension.

Fj(q) )

〈| ∑ f (q)e | 〉 iqrj,m 2

j

(1)

m

where rj,m is the position of the jth atom on the mth molecule, q ) (qX, qY, 0) is the wave vector, and the symbol 〈...〉 indicates an ensemble average. This ensemble average is performed by summing up the contribution of all the trajectory frames in a

Figure 4. Module of the two primitive vectors of the direct 2D lattice (left side ordinate axis), and the angle among them (right side ordinate axis) as a function of the surface tension γ (note that the horizontal γ scale increases toward the left, so toward the right the lateral pressure grows).

given temporal interval. The quantity fj(q) is the form factor of C atom j, and for the present purposes we set them all to 1. We calculated Fj(q) for C atoms positioned at different j sites in the chain, but all of them showed the same qualitative behavior under the present simulation conditions. Using the above-defined structure factor patterns (graph shown in Figure A in the Supporting Information) we see that the systems show an average two-dimensional long-range periodicity of hexagonal or oblique symmetry depending of the γ value. It is also possible to calculate the primitive vectors of the direct lattice of such arrangement. Figure 4 shows the module of these 2D primitive vectors as a function of γ, along with the angle between them at T ) 300 K. It is seen that the systems abruptly change their symmetry at a γ in the interval [1850, 1900] bar nm, from a low-pressure oblique to a higherpressure hexagonal phase. We also calculated the three dimensional (3D) structure factor for these systems (not shown here). These 3D structure factors are proportional to the diffraction intensity pattern measured in grazing angle incidence X-ray diffraction experiments. From this quantity the global tilt angle of the alkyl chains and the azimuthal orientation of that inclination can be determined. We found that the azimuthal orientation of the global tilt of the alkyl chains is different in both layers, which is evident from Figure 3. This azimuthal orientations of the global tilt appears to be in a not unique direction relative to the underlying 2-D lattice for different leaflets and surface tension value conditions. This kind of behavior has been previously found in similar systems by experimental and theoretical studies,39,40 and it is indicative that such configurations have similar free energies. ii. AWerage Area Per Molecule and Tilt Angle of Unlabeled Bilayers. Figure 5a shows the average area per lipid as a function of γ as calculated from the dimensions of the simulation box. These results also agree with those calculated from the primitive vectors of the direct lattice. As it is seen from Figure 5, the average area has an abrupt drop in the interval γ:[1850, 1900] bar nm for T )300 K that coincides with the change in the symmetry of the system shown in Figure 4. There is also a further shallow decrease of the average area in the interval γ:[1206, 1500] bar nm that follows the slight lowering in the modules of the primitive vectors in Figure 4. Figure 5b shows the γ dependence of the average tilt angle of the long axis of the lipid molecules. This variable is calculated as the angle among the average C1-C18 vector of the SA molecules and the normal to the layer. The azimuthal angles between the C1-C18 SA vectors and the x-axis of the simulation

16834 J. Phys. Chem. B, Vol. 112, No. 51, 2008

Vartorelli et al.

Figure 7. Largest eigenvalue, Sm(i) (a), and Z component SZ,Z(i) (b) of the order parameter tensor profiles for several surface tensions (γ) at 300 K.

Figure 5. Average area per lipid (a) and average tilt angle (b) as a function of the surface tension γ at T ) 300 and 315 K.

Figure 6. Average tilt angle as a function of the average area per lipid. Note that the area axis is inverted to point in the direction of increasing lateral pressure.

box, show that there is a collective inclination of the long molecular axis toward a common direction. These features are in agreement with those derived from the average 3D structure factor calculations. The average tilt angle shows the same general trends as the average area per lipid with γ (Figure 5). Figure 6 remarks the strong correlation between these variables, showing that the tilt angle is determined by the average area available to the lipids. As the area per lipid grows the alkyl chains increase their tilt angle to favor the attractive interactions among them. iii. Conformational Order of the Alkyl Chains of Unlabeled Bilayers. The conformational order of the SA molecules can be characterized by the average fraction of dihedral angles in the trans conformation and by the order parameter tensor.14 The average fraction of dihedral angles in trans conformation was 98% at 300 K (and 97% at 315 K) and did not present a significant change with γ. These values are equivalent to the average existence of one gauche defect by each 3.3 lipid. This result shows that the alkyl chains maintained a high conformational order for all the γ range.

The largest eigenvalue of the order parameter tensor S(i), that we call Sm(i), characterizes the orientational order of the vector that joins atoms Ci-1 and Ci+1 of the alkyl chain.14 The values of Sm(i) depend on the orientational dispersion of the corresponding vector in time but not on its preferential direction relative to the bilayer normal. The profiles of Sm(i) as a function of C atom position in the chain for several γ values are shown in Figure 7a. Figure 7a presents large values of Sm(i), particularly for C atoms between positions 4 and 15, which means a high orientational order of this portion of the molecules. The values of Sm(i) decrease toward the head and tail of the lipids, showing a larger orientational dispersion of the molecules at these zones. This behavior is characteristic of a very compact liquid condensed phase of fatty acids, a gel-like phase in the terminology used in phospholipids. The orientational order is not very sensitive to the surface tension γ, although the lower pressure phase is slightly more ordered at the plateau region. This last feature is more pronounced at 315 K. Figure 7b shows the Z component of order parameter tensor profiles, SZ,Z(i). This variable is sensitive to the orientational order and also to the preferential orientation of the molecular axis about the Z-axis.14 The appreciable dependence of this profile on the surface tension γ is easily understood in terms of the change of the inclination angles (“tilt”) of the long molecular axis of the alkyl chains (see Figure 5b). In the following subsection we will study the average position and conformations of n-SASL labels inserted in the bilayer. We performed simulations at γ values that correspond to both sides of the structural phase transition (γ ) 1206 and 2000 bar nm). Labeled Stearic Acid Bilayers. Stearic acid bilayer systems of 560 lipids with a labeled molecule inserted in each leaflet were simulated for each of the n-SASL species. The equilibration stages were determined from the evolution of the average area per lipid and from the n-SASL SZ,Z order parameter profiles calculated from contiguous 1 ns intervals (see Figure B in the Supporting Information). For each of the studied labeled systems, MD runs of 20 ns were performed. After the first 16 ns the n-SASL SZ,Z 1 ns profiles started to show fluctuations around their temporal average for all the labels. From the

Spin-labeled Stearic Acid Behavior of a Lipid Bilayer

J. Phys. Chem. B, Vol. 112, No. 51, 2008 16835

Figure 8. Atomic density profiles for the bilayer systems labeled with 5- (a), 12- (b) and 16-SASL (c). Thin lines correspond to the unlabeled SA molecules and thick lines to the labeled ones. The distributions of the Cn atom, where the doxyl ring is attached, are shown in area-filled style. The densities that correspond to the labeled molecules are scaled by 100×.

analysis of these variables we conclude that during the final 4 ns of each simulation the labels have reached the local equilibrium condition referred to in the Method section and have developed a significant sampling of their energy surface neighborhood. The differences between the profiles of both leaflets that are more notable for the 12-SASL (see below), are attributable to various local minima that are not all of them reached by each label and would need longer simulation times to show the expected symmetry. In the case of 12-SASL we have performed simulations with other initial conditions, and we have not found configurations different from those sampled in the presented run. Therefore, we think that the configurations of the two leaflets taken as an ensemble allow a significant sampling of the configurational space. Therefore, using the above-mentioned criteria, the final 4 ns of each simulation were taken as the production stage. 5-SASL Bilayer System. Figure 8 shows the atomic density profiles of several atomic species in the bilayer systems labeled with 5-, 12- and 16-SASL. The simulation box has been divided in 200 slabs parallel to the lipid-water interface (of 0.35 Å width). The labeled molecule atomic density profiles have been scaled to make them appreciable since they are in a ratio of 1:280 with those of the unlabeled SA. The atomic profiles of the C where the doxyl ring is attached has been drawn in painted-area style to highlight their position. Figure 8a shows the density profiles for the system labeled with 5-SASL. We remark the following features in these profiles: (a) the depth distribution of C1 and C5 atoms of the labeled molecules (one in each layer) are strongly overlapped; (b) the peaks of these distributions are located slightly over the z-peak-position of the C1 atoms of the unlabeled SA molecules; (c) the peaks in the C18 atomic density profiles of the labeled molecule are significatively shifted to the lipid-water interface of the bilayer if compared with those of the unlabeled SA molecules. The first feature means that C1 and C5 (the carbon where the doxyl ring is attached in 5-SASL) are positioned at approximately similar depths and are slightly shifted outward from the average C1 atom position of the host system. The terminal methyl groups of the labeled molecules are lifted toward the lipid-water interface due to the position of its head and its ring moiety. We analyze the SZ,Z(i) and Sm(i) profiles of the 5-SASL molecules to characterize the conformations adopted by them

Figure 9. SZ.Z(i) (a) and Sm(i) (b) profiles for the 5-SASL molecule in the lipid bilayer.

in Figure 9. The SZ,Z(i) profiles (see Figure 9a) show that the Ci-1-Ci+1 vectors have an important change in its average orientation for i < 5. If one neglects the effect of the orientational dispersion, values of SZ,Z around -0.3 correspond to angles between the local alkyl axis and the z-axis around ∼69°. Conversely, for i > 5, values of the SZ,Z(i) profile around 0.7 are found. It is important to remark that the C atoms in this portion of the 5-SASL lay near the water-lipid interface. The values of the profiles for these atoms are slightly above that for the unlabeled SA, showing that the average orientation of these segments are closer to the interface normal. As will be shown below, the interaction among the polar heads of the 5-SASL and their surrounding lipids give rise to steric constraints that affect the orientation of the labeled molecule segments near the interface. Again, if one neglects dispersion effects these values correspond to angles between the local alkyl axis and the z-axis around ∼27°, which are close to the global tilt of the unlabeled molecules. As it is seen in Figure 9a, the values of SZ,Z(i) of labeled and unlabeled molecules tend to become equal for C atoms of the 5-SASL that lay deeper in the lipid layer. In summary, the portion of the labeled molecule between C6 and

16836 J. Phys. Chem. B, Vol. 112, No. 51, 2008

Vartorelli et al.

Figure 10. Representative structures of the most statistically significative clusters found in the covariance analysis of the n-SASL trajectories. The analysis is performed individually for each layer (the arrows label each of them). When more than one structure is presented, the most significative is drawn in light blue color and the other in turquoise. Only structures belonging to clusters with more than 20% of presence in the trajectory have been drawn.

TABLE 1: Average Number of Hydrogen Bonds between the Carboxyl Group of the n-SASL Molecule (of Each Leaflet) and Those of Unlabeled SA and Watera molecule unlabeled SA 5-SASL (COOH) 12-SASL (COOH) 16-SASL (COOH)

layer

SA-COOH

water

v V v V v V

0.05 0.7 0.9 0.5 0.2 0.2 0.1

1.6 1.5 1.4 0.4 1.6 1.5 1.5

a The first row correspond to HB of unlabeled SA molecules with water and other SA in the pure system.

C18 that lays within the more hydrophobic and compact region of the bilayer has an orientational order similar to the alkyl chains of the host system. Conversely, the portion of the alkyl chain of the labeled molecule between C1 and C5 that lays at the lipid-water interface, orients nearly perpendicular to the long axis of the host lipids and presents a higher mobility, as shown by the Sm(i) profiles in Figure 9b. We have made a statistical cluster study of the n-SASL molecule geometries along the trajectory time based in a covariance analysis (rmsd cutoff ) 0.1 nm) of the structures after fitting41 to confirm such interpretation. Figure 10 shows representative structures of the most statistically significative clusters found in the analysis for the different n-SASL labels. The structures found for 5-SASL (Figure 10, left side) agree with the picture drawn from the atomic density profiles and the order parameter analysis for this labeled molecule. There are small differences in the configurations of the 5-SASL in both layers that were already present in the atomic density and order parameter profiles. This feature will be discussed in the next paragraph about hydrogen bonding at the polar region of the interface. Table 1 shows the average number of hydrogen bonds (HB) per labeled molecule between different polar groups: unlabeled SA carbonyl (SA-COOH), n-SASL carbonyl (n-SASL-COOH), and water. The carbonyl group of the 5-SASL molecule forms an important average number of HB to the polar head of neighboring unlabeled SA molecules and to water. The first quantity shows a small but significative difference between the down and upper leaflets. This fact accounts for the differences observed in the behavior of the 5-SASL in both layers: (a) a slightly lower dispersion of the C1 atomic density profiles of the 5-SASL for the down layer (see Figure 8a); (b) the lower

Figure 11. Orientational distribution of the normal to the average doxyl plane along the production stage of the simulation for 5-SASL. The polar radii is sin(Θ) for Θ:[0°, 90°]. Plots (a) and (b) correspond to the labeled molecules in the so-called up and down leaflets, respectively. The angle that the preferred orientation forms with the normal to the bilayer is 38°.

orientational dispersion of the segment between C1:C4 surveyed by the Sm(i) profiles for this layer in Figure 9b; and (c) values of the SZ,Z (i) profiles closer to -0.5 for the segment C1:C4 for the same layer as shown in Figure 9a (therefore, angles between the local molecular and z-axis closer to 90°). The case of 5-SASL is the only one among the studied labels whose doxyl ring moiety has an appreciable number of HB with water. Nevertheless, this number is only a 7% of those formed between the 5-SASL COOH group and water. In summary, we found that the 5-SASL positions their doxyl moiety at the water-lipid interface. This configuration allows for the building of HB between the polar head of the 5-SASL and the neighboring unlabeled SA molecules; and also for a lower disruption of the host lattice of unlabeled lipid by the bulky doxyl ring as will be shown in a subsection below. Another property that deserves attention is the analysis of the orientational distribution of the normal to the average doxyl ring plane. This property is relevant since its angular distribution and the time that characterizes its motion, determine the EPR spectral features.2,18,42 To perform this analysis we will study the distribution of this normal vector around its average orientation (we will refer this average orientation as preferred orientation because it is determined using the direction of the eigenvector of the S tensor corresponding to the largest eigenvalue). We will show the distribution as function of two angles: Θ, the angle between the instant normal and the preferred orientation and Φ, the azimuthal orientation of the normal vector around the same direction. The angle comprised between the preferred orientation and the normal to the bilayer (z-axis) will also be quoted (θZ). Figure 11 shows these angular distributions for the 5-SASL in both layers in points-scatter polar graphs. The polar angle of the graphs corresponds to Φ and the radial distance to sin(Θ). Each point in the graphs correspond to an orientation of the normal to the doxyl ring for a given time in the trajectory relative to its preferred direction.

Spin-labeled Stearic Acid Behavior of a Lipid Bilayer For this case the preferred orientation of the doxyl normal forms an angle of approximately 38° with the normal to both layers ((z-axis). The first feature to note is that the normal distribution does not have axial symmetry around its preferred orientation. The most notable departures from the axial symmetry coincides with the azimuthal directions (Φ) where the Θ angles takes larger values. This fact can be rationalized taking into account that, when the carbonyl group of the labeled molecule forms HB with the head of their unlabeled neighboring counterparts, it brings the normal of the doxyl ring to larger Θ values (see Figure 10 for 5-SASL in the lower panel). Because of the 2D lattice structure of the unlabeled molecule heads the azimuthal orientations of the normal to the ring lose their axial symmetry. This fact is particularly notable for the labeled molecule in the down leaflet (see Figure 11b). The 5-SASL head of this layer has a larger average number of HB with neighboring lipids than the opposite leaflet (see Table 1) as was already analyzed and therefore shows the largest departures from the axial symmetry of the distribution. These facts are worth to take into account to understand how this EPR spin label surveys its environment at the bilayer interface. For the typical time scale of an X-band EPR spectra each point in the distribution of Figure 11 will contribute to the averaged hyperfine interaction tensor. The distribution is anisotropic and has a prevalence of large values of Θ angles with the average orientation of 38° relative to the normal of the layer. For a distribution with such characteristics one can expect that the average hyperfine tensor have all their principal components values lower than the maximum eigenvalue of the molecular A|2 and lead to a narrower line shape. 12-SASL Bilayer System. Figure 8b shows the atomic density profile for the system labeled with 12-SASL (one in each layer). The most striking feature to note is the important differences in the atomic density profiles of both layers. These density profiles show that the labeled molecules adopt different configurations in each layer. Simulations started from different initial conditions lead to the presence of both configurations of the labeled molecule. Experimentally, the presence of both configurations had also been conjectured in the high lateral pressure phases of labeled SA monolayers.43 On the basis of these facts we can safely disregard the hypothesis that these differences arise from an artifact of the simulations. Figure 8b shows that the C12 of the 12-SASL for the upper leaflet has the peak of its atomic density profile that partially overlaps with the distribution of the C18 atoms of the same molecule and those of the unlabeled ones. This fact means that for this leaflet the C12-C18 molecular segment of the 12-SASL positions itself at the lipid-lipid interface with a direction nearly perpendicular to the bilayer normal. This shift of the doxyl moiety of the 12-SASL toward the hydrophobic region lead the head of the labeled molecule (C1) to position below the C1 atoms of its unlabeled neighbors (see Figure 8b). Figure 12a shows the SZ,Z(i) order parameter profile for this 12-SASL label. The profile that correspond to the label in the upper leaflet shows an abrupt change around the position of C12. The SZ,Z(i) values for C2:C11 are similar to the order parameter of the unlabeled SA molecules showing similar orientation and orientational dispersion. Conversely, the segment C13:C17 shows values around -0.3 that correspond to orientations nearly parallel to the interface plane. Figure 12b shows the Sm(i) profile that also indicates a higher orientational dispersion for the segment C13:C17 due to the lower steric constrains at the lipid-lipid interface. These facts also agree

J. Phys. Chem. B, Vol. 112, No. 51, 2008 16837

Figure 12. Order parameters SZ,Z and Sm profiles for the labeled 12SASL and for the unlabeled SA molecules in both leaflets.

with the geometries that emerge from the cluster statistical analysis shown in the upper part of Figure 10 for the 12-SASL. The 12-SASL in the lower leaflet shows important differences with that previously analyzed. The atomic density profile in Figure 8b (lower part) shows that C12 of the 12-SASL positions itself at depths close to those of the unlabeled SA. The same is also true for the rest of the atoms of the 12-SASL molecule relative to their homologous of unlabeled SA. This fact shows that the 12-SASL in this leaflet remains included in the compact hydrophobic region and does not shift toward the polar or lipid-lipid interfaces. The SZ,Z(i) order parameter profile for the 12-SASL molecule in this layer (see Figure 12a) shows that there is an important change in the orientation of the alkyl chain axis for the segment C9:C12. Values of SZ,Z ≈ -0.1 for this segment signal that its orientation forms angles larger than 54° with the normal to the layers. The rest of the alkyl chain atoms of the 12-SASL show values of their SZ,Z(i) profiles similar to those of the unlabeled molecules. The values of the Sm(i) profile for the 12-SASL molecule in this layer (see Figure 12b) are all homogeneously high (>0.8) for all the segments of the chain, with a slight lowering at C9:C10. This fact means small orientational dispersion of all the segments of the 12-SASL chain, with a small increment for the C9:C10. These features of the atomic density and order parameter profiles agree with the representative geometry derived from the covariance analysis of Figure 10 (lower image for 12-SASL). From Table 1 it is clear that the carboxylic group of the 12SASL that remains completely included in the hydrophobic region (lower leaflet) has a notably higher average number of H bonds to water than the other label. Conversely, the carboxylic group of the 12-SASL label, whose doxyl moiety lays at the lipid-lipid interface (upper leaflet), has a lower total average number of H bonds that distributes between water and the polar head of neighboring SA molecules. Figure 13 shows the orientational distribution of the normal to the doxyl ring for both 12-SASL molecules. The normal to the doxyl ring that lays in the lipid-lipid interface has a larger orientational dispersion with a rather axially symmetric distribution around its preferred direction, which forms an angle of g 39° with its respective layer normal. The doxyl moiety

16838 J. Phys. Chem. B, Vol. 112, No. 51, 2008

Vartorelli et al.

Figure 13. Orientational distribution of the normal to the average doxyl plane along the production stage of the simulation for 12-SASL. The polar radii is sin(Θ) for Θ:[0°, 90°]. Plots (a) and (b) correspond to the labeled molecules in the so-called up and down leaflets, respectively. is the angle that the preferred orientation of the normal to the doxyl ring forms with the normal to each layer.

Figure 14. Orientational distribution of the normal to the average doxyl plane along the production stage of the simulation for 16-SASL. The polar radii is sin(Θ) for Θ:[0°, 90°]. Plots (a) and (b) correspond to the labeled molecules in the so-called up and down leaflets, respectively. is the angle that the preferred orientation of the normal to the doxyl ring forms with the normal to each layer.

of this label is located at the lipid-lipid interface (see Figure 8), a region with higher free space as evaluated in ref 14. The lower motional constrains at this zone allow for the appearance of gauche defects that generates the tumbling motion of the ring around the average orientation imposed by the segment of the alkyl chain that remains included in the compact hydrophobic region. Conversely, the doxyl ring in the other layer moves in a more restrained environment since it lays in the more compact zone of the layer. This normal has a narrower distribution, and the preferred orientation forms an angle of g 56° with its respective layer normal. Below it will be shown that the above- and below-the-ring-moiety alkyl segments of this 12SASL locate at the host lattice positions to minimize its perturbing effect. Therefore, the doxyl ring motion is constrained by this fact. 16-SASL Bilayer System. The right panel of Figure 8 shows the atomic density profiles for the system with a 16-SASL label in each layer. The density profile for C16 of 16-SASL (the atom where the doxyl ring is attached) is shifted toward the center of the bilayer relative to the peak that corresponds to C16 of unlabeled SAs and has strong overlap with the distribution of C18 atoms of labeled and unlabeled molecules. These facts are easy to rationalize in terms of the representative structures at the right side of Figure 10. The labeled molecules shift their doxyl ring toward the lipid-lipid interface, where there is more free space available. The C16-C18 segment of the 16-SASL molecules bent to fit in that region. The SZ,Z(i) and Sm(i) profiles for 16-SASL (Figure C in the Supporting Information) agree with this picture. From Table 1 it is seen that there is a large average number of H bonds of the polar head of 16-SASL to water. The atomic density profiles for C1 of 16-SASL have their peaks approximately at the same depths of those of C1 of unlabeled molecules (Figure 8). Therefore, it is natural that the average number of H bond to the solvent are comparable to those formed

by the polar head of 5-SASL and the 12-SASL label that remains included in the lower leaflet of the bilayer (see Table 1). Conversely, the head of 16-SASL does not build so many H bonds to neighboring unlabeled SA molecules as the 12-SASL that shifts toward the lipid-lipid interface (upper leaflet) does. Figure 14 shows the orientational distribution of the normal to the average plane of the doxyl ring for the 16-SASL labels in both layers. The graphs show an almost axially symmetric orientational distribution of the normal to the ring. This distribution has a dispersion around its average orientation smaller than 20° for both layers. The preferred orientations form small angles with each layer normal (12° and 5°). One can compare these distributions with that corresponding to the 12SASL whose doxyl ring lays at the lipid-lipid interface (upper leaflet in Figure 13a). It is seen that the 12-SASL distribution has a larger dispersion than those of 16-SASL, and also larger angles with the normal to the bilayer (37°), although the depth atomic density profile for the carbon where the doxyl is attached has similar distributions for both labels. We think the reason for this behavior is found in Table 1. The 16-SASL polar head has a slightly lower average number of HB to water than the unlabeled SA and the 12-SASL in the lower leaflet have. Moreover, the average number of 16-SASL HB to polar heads of neighboring SA molecules is comparable to those built among unlabeled molecules, and therefore this interaction does not compensate the absence of water HB. We think that the polar head of 16-SASL positions itself at the water-lipid interface to maximize the number of HB to the solvent, and therefore the doxyl ring is pulled outward to a more constrained environment. The relevance of H bond interactions between the labeled molecule head and water is also involved in the stability of configurations like that of 12-SASL, which remains with its doxyl ring laying in the compact region of the layer (down leaflet). Our own previous results14 show that for 10-SASL the

Spin-labeled Stearic Acid Behavior of a Lipid Bilayer

J. Phys. Chem. B, Vol. 112, No. 51, 2008 16839

Figure 15. Sm(i) profiles of unlabeled SA molecules for differentsized neighboring shells around the n-SASL spin label. Note that the ordinate scale was expanded to begin at 0.5.

configurations with the doxyl ring inserted in the compact hydrophobic region are the most probable. This fact can be rationalized by thinking that the 10-SASL-polar-head H bond interaction to water would be drastically lowered if the ring shifts to the lipid-lipid interface. In the case of 12-SASL, whose doxyl moiety lay at the lipid-lipid interface (upper leaflet), the H bond interactions between its head and water do not decrease so markedly when it shifts toward the membrane center. Moreover, the lowering of these interactions are partially compensated by H bonds to the carboxylic groups of neighboring SA molecules (see Table 1). Perturbing Effects of the n-SASL Labels on the Bilayer Properties. In this section our concern will be to quantify the perturbing effects of the insertion of the n-SASL labels (at the concentration used in our work 0.36 mol %) in the properties of the membrane. To analyze these perturbing effects we calculated the Sm(i) order parameter profiles for several neighboring sets of unlabeled SA molecules around the label in each system layer. We included in each set all the unlabeled SA that were closer to the labeled n-SASL than a cutoff radio R0. Values of R0 ) 5, 10, 15, 20, 25 Å were considered for calculating the Sm(i) along with those calculated for all the SA in the layer that were already shown in previous sections. This way we will be able to quantify the extension of the perturbing effects of the labeled molecule. Figure 15 shows these profiles for all the studied systems labeled with different n-SASL molecules. From Figure 15 we conclude that for all the n-SASL labels the most significative perturbations are localized in a unlabeledSA neighboring shell of 10 Å. It is evident from these graphs that the insertion of the 12-SASL label lead to the strongest orientational perturbations in the host system. These effects were particularly larger for the 12-SASL in the lower leaflet (see Figure 15b, lower panel). That label was the one whose doxyl ring remained within the compact hydrophobic region of the layer. The zone where the order parameter profiles show the largest perturbations coincides with the region of void space left off by the shifting of the n-SASL molecule (in the case of 5-SASL, 16-SASL, and 12-SASL upper leaflet). In the case of 12-SASL lower panel, the doxyl ring remains included in the layer and the portion of neighboring unlabeled SA that were below and above the ring moiety are perturbed. For this configuration the C1-C8 and C13-C18 segments of the 12-SASL alkyl chain adopt orientations similar to the unlabeled SA molecules (see SZ,Z(i) order parameter profiles of 12-SASL in Figure 12a). The segment C9-C12 of 12-SASL orients with a direction nearly perpendicular to the long axis of unlabeled SA

Figure 16. Snapshot of the simulation of the SA bilayer labeled with 12-SASL.

molecules. In the snapshot of Figure 1644 we illustrate such a situation. The C1-C8 and C13-C18 segments of the 12-SASL alkyl chains each align in neighboring planes of the SA periodic average structure. The configuration of the 12-SASL label leaves void spaces above and below the doxyl moiety. These holes enhance the dispersion in orientation of the neighboring SA molecules that lead to the lowering of the Sm(i) profiles in Figure 15b lower panel. Analyzing Sm(i) profiles of individual unlabeled SA molecules (not shown), we see that those perturbed in the upper region are not the same as those perturbed in the lower zone. The local character of the orientational perturbation generated by the n-SASL label brings up the topic about the global changes in properties such as the average area per lipid in the membrane. In a previous work14 we found that the insertion of a 10-SASL spin label in SA membranes at concentrations of 1.4 mol % produces significative changes in the average area per lipid and tilt angle of the system. In the present work, where concentrations of the n-SASL label of 0.36 mol % were used, we have not found any significant change in the area per lipid, periodicity, nor in the tilt angle of the SA molecules. Effects of the Lateral Pressure Change. All the above analyzed systems were also simulated at a surface tension of γ ) 2000 bar nm. At this value of γ in the absence of the n-SASL molecules the system had an average periodic lattice of oblique symmetry, a tilt angle θ ) 37°, and an average area per lipid 14% higher than in the hexagonal γ ) 1206 bar nm phase (see first section of Results). With the insertion of the n-SASL labels in the systems under the present conditions, the global properties of the bilayer (lattice periodicity, average area and tilt) do not experiment any significative change. The Sm(i) profiles of the n-SASL neighboring SA molecules showed local perturbations of magnitude much less than those in Figure 15 for γ ) 1206 bar nm. This fact is easy to rationalize due to the higher area available per lipid. The most notable difference observed at γ ) 2000 bar nm is that the 12-SASL configuration with the doxyl moiety laying in the center of the leaflet was not observed (see Figure 16). The reason for this feature can be trailed by analyzing the data of the average

16840 J. Phys. Chem. B, Vol. 112, No. 51, 2008

Vartorelli et al.

TABLE 2: Average Number of H Bonds to the Polar Head of the Labeled n-SASL at γ ) 2000 bar nm γ ) 2000 bar nm

SA-COOH

water

unlabeled SA 5-SASL (COOH) 12-SASL (COOH) 16-SASL (COOH)

0.03 0.4 0.3 0.2

1.7 1.7 0.8 1.6

number of water H bonds to the polar head of 12-SASL in Table 2. The larger average area per lipid in this phase allows for an enhanced interaction of water with the polar head of 12-SASL, which are in bent configurations with the doxyl moiety at the lipid-lipid interface. See in Table 2 that the average number of HB between water and the COOH group of 12-SASL is 0.8, 100% larger than the value of 0.4 found for γ ) 1206 bar nm for the configuration in which the doxyl ring lays at the lipid-lipid interface (see Table 1). For a closer inspection of this phenomena we have calculated the water O atomic density profile in the direction normal to the layer of the molecules within a cylinder whose axis passes through the C1 atom of the polar head of the labeled molecule. The cylinder axis is tilted following the orientation of the SA molecules of the host system, in such a way that this atomic density profile is sensible to the water penetration inside the hole left over by the 12-SASL that has shifted toward the lipid-lipid interface. Figure 17 shows these profiles at the two lateral pressure conditions studied in this work. Also, the profiles for water cylinders over C1 atoms of unlabeled molecules have been included as reference together with the atomic density profiles of C1 atoms of both species. For γ ) 1206 bar nm there is a long tail of the water distribution in the cylinder over the C1 of 12-SASL (thick blue lines) that extends almost to a depth of 4 Å below the maximum in the C1 of unlabeled SA profile (black thin lines). The water distribution over the 12-SASL can also be compared to the water distribution over other unlabeled SA molecules (blue thin lines). From these observations it is concluded that the water molecules penetrate deeper in the hole left by the 12-SASL than in other zones of the host system. For this case (γ ) 1206 bar nm) it is also observed that the atomic density profile of C1 atoms of 12-SASL has a very asymmetric distribution that tails toward the water-lipid interface. As a consequence, there is an important overlap of both distributions that contributes the

Figure 17. Atomic density profile of water O in the direction normal to the layer (blue lines). The results of γ ) 1206 bar nm (full lines) and γ ) 2000 bar nm (dashed lines) are shown. The profiles of different γ have been aligned in such a way that the C1 distributions of unlabeled SA have maximum overlap (shown in thin black lines). Thick black lines correspond to the distribution of C1 atoms of the labeled 12-SASL molecule. In thin blue lines the profiles of water O atoms in regions far from the labeled molecules are plotted. The C1 profiles have been scaled.

above-mentioned H bond between water and the polar head of 12-SASL (see Table 1). The asymmetry of the 12-SASL C1 distribution suggests that this atom is pulled outward by the H bond energy contribution of water in spite of the steric constraints of its doxyl moiety at the lipid-lipid interface. Also, the neighboring unlabeled SA contribute with H bonds to the polar head of 12-SASL (see Table 1). These facts allow one to understand the reason of the coexistence of these bent configurations along with the erected ones of Figure 16. At γ ) 2000 bar nm the area per lipid is 14% larger than at γ ) 1206 bar nm, and also the thickness of the bilayer system has decreased as a consequence of the larger tilt of the lipid tails. The water penetration over the 12-SASL C1 atom is more pronounced (blue dashed thick lines in Figure 17). The 12-SASL C1 atomic density profile also has shifted toward the outer interface as a consequence of the narrowing of the bilayer thickness. These two facts explain the increase of the average number of water-12-SASL HB (Table 2). The 12-SASL C1 distribution does not exhibit the strong asymmetries that were present at γ ) 1206 bar nm, suggesting that for γ ) 2000 bar nm these bent conformations are not disfavored by the environmental constrains. The enhanced solvent interaction of these configurations favors their prevalence over erect ones (see Figure 16) that were present at γ ) 1206 bar nm; therefore, erect configurations are not observed at γ ) 2000 bar nm. For the case of 5-SASL it is seen that at this lower lateral pressure its polar head builds less H bonds with the unlabeled SA neighbors and more to water. The hydration of the 16-SASL is slightly larger than at γ ) 1206 bar nm due to the larger average area per lipid at the new lateral pressure. The doxyl ring exhibits an enhanced mobility at γ ) 2000 bar nm, particularly for 12-SASL and 16-SASL. This is partially due to the larger available area per lipid that imposes less constraints to the dihedral angle transitions of the n-SASL alkyl chains close to ring zone. Nevertheless, there is also a concurrent solvent-related effect already mentioned. The larger average area per lipid allows a deeper penetration of the water toward the polar heads of 12- and 16-DAE. Therefore, these labeled molecules are not so strongly pulled outward from the surface, and the ring moiety keeps moving at the lipid-lipid interface in a less-constrained environment. Summary and conclusions We have performed MD simulations of stearic acid (SA) spin label molecules labeled at positions 5, 12, and 16 in SA bilayers in gel-like phases, used as a workbench for studying their behavior. We have found that the unlabeled SA system presents a structural phase transition when the lateral pressure (surface tension parameter) is changed. The high lateral pressure phase is characterized by a 2D hexagonal symmetry (γ < 1850 bar nm at 300 K and γ < 1950 bar nm at 315 K). At lower lateral pressure there is a symmetry change to a oblique lattice, and the average area per lipid is increased by a 14%. Therefore, the study of the spin-labeled molecules 5-, 12-, and 16-SASL in this system at both phases allows for analyzing the effects of changing the average area per lipid of the host layer. In the high lateral pressure phase (γ ) 1206 bar nm) we have found that the doxyl ring of the 5-SASL label positions itself at the lipid-water interface. The COOH group of the labeled molecule makes H bonds to the polar heads of the host SA molecules and also to the solvent. This fact gives rise to an axially asymmetric distribution of the normal to the average plane of the ring around its average orientation. The normal to the plane of the doxyl ring is approximately the symmetry

Spin-labeled Stearic Acid Behavior of a Lipid Bilayer axis2,45 of the g and hyperfine interaction A tensors from which the EPR spectra are derived. In this high pressure phase the 12-SASL label has 2 different equilibrium configurations. On the first configuration the 12SASL molecule positions its doxyl moiety at the lipid-lipid interface by shifting its polar head toward the hydrophobic region of the layer. In the second configuration the doxyl ring of 12-SASL positions itself at the central portion of the layer parallel to the planes of unlabeled SA. Although this configuration pays a energetic penalty for introducing gauche defects, it is favored by the attractive interactions of its polar head to water. In this last configuration the motion of the normal to the doxyl ring is highly constrained. Conversely, the distribution of the normal to the paramagnetic group is much more dispersed in the case of the first configuration. In the same phase, the 16-SASL label always positions its doxyl ring at the lipid-lipid interface, which is easy to rationalize since it is not necessary to do a major shift of its head apart from the water-lipid zone to achieve this location. Nevertheless, the mobility of the doxyl ring and its normal are constrained in this configuration by the forces exerted by the attractive polar head-water interactions. At the concentration of labeled molecules studied in this work (0.36 mol %) we have not found significant perturbations on the global properties of the host system (e.g., lattice periodicity, average area per lipid, tilt angle). This is at variance with what we had found at concentrations of 1.4 mol % of spin labels in the same system.14 There are local perturbation effects on the orientational order of the host bilayer that are mostly limited to a first-neighbor shell around the labeled molecule. We have found that the magnitude of the perturbation of different labeled species can be ordered as follow 10-SASL > 12-SASL > 5-SASL > 16-SASL (see discussion of Figure 15, and also results from our own previous work on 10-SASL14 have been included). The origin of this sequence can be explained as follows. The labeled molecule configurations that introduce the largest host perturbations are those of erected type (see Figure 16) since they affect neighboring lipid above and below the ring position. In our study of 10-SASL we only found erected configurations. In the present work, 12-SASL shows coexistance of erected and nonerected configurations at γ ) 1206 bar nm. This fact explains the first part of the perturbation sequence. The shifting of the doxyl ring of 5-SASL toward the lipid-water interface left behind its tail a hole in the compact host structure that generates the perturbation of neighboring unlabeled lipids. Finally, 16-SASL is the label that generates the smaller disruption effect on the host structure to accommodate their doxyl moiety at the lipid-lipid interface. The work of Wisnievska et al.16 that studies the perturbational effects of n-SASL spin labels in the gel-liquid-crystalline transition temperatures of DMPC established the following order 9-SASL > 12-SASL > 5-SASL > 16-SASL. Although the ability of our simple bilayer system for modeling phospholipid bilayers in gel phase can be questioned, it is significative that it can quantify the magnitudes of the perturbational effects of different spin labels in agreement with DMPC experiments. These facts puts forward the question if the same conceptual arguments that explain the magnitudes of the perturbations of different n-SASL in the SA systems could be applied to phospolipid bilayers in gel phase. At the lowest lateral pressure phase the perturbation effects of the insertion of spin labels are smaller due to the larger available area per lipid. This enlargement of the available area per lipid allows us to draw conclusions about the solvent-polarhead interaction consequences. We found that at larger available

J. Phys. Chem. B, Vol. 112, No. 51, 2008 16841 area per lipid the erected configurations of 12-SASL label disappear. The reason is that configurations with the doxyl ring at the lipid-lipid interface are favored since the enlargement of the area allows for a larger interaction of the shifted polar head with water. These conclusions are in agreement with the hypothesis that has been proposed to explain the experimental results in doped fatty acid films when the lateral pressure is changed.43 The area enlargement also favors the 12- and 16SASL rings that have a larger mobility since their polar heads are not so strongly pulled toward the water-lipid interface. In summary, in this work we have analyzed the behavior of n-SASL in a gel-like lipid bilayer by studying their depth profile, orientational distribution, and perturbing effects. We recognize the limitations of the present model system to mimic phospholipid bilayers. Nevertheless, we think that the conceptual ingredients worked out here are important to the interpretation of EPR experiments in lipid bilayers. Acknowledgment. We are grateful to Professor M.C.G. Passeggi and A.M. Gennaro for useful discussions. D.E.R. is a member of CONICET. This work was supported by CONICET PIP 2000-02559 and 2005-5370, and UNL-CAI+D 3-23-2006. Supporting Information Available: This material is available free of charge via the Internet at http://pubs.acs.org. References and Notes (1) McConnell, H. M. In SPIN LABELING. Theory and Application; Berliner, L.J., Ed.; Academic Press: New York, 1976; Chapter 13. (2) Griffith H.; Jost P. In SPIN LABELING. Theory and Application; Berliner, L.J., Ed.; Academic Press: New York, 1976; Chapter 12. (3) Risse, T.; Hill, T.; Schmidt, G.; Hamman, H.; Freund, H. J. J. Phys.Chem. B. 1998, 102, 2668–2676. (4) Risse, T.; Hill, T.; Schmidt, G.; Hamman, H.; Freund, H. J. J. Chem. Phys. 1998, 108, 8615–8625. (5) Taylor, M. G.; Smith, I. C. Biochim. Biophys. Acta 1983, 733 (2), 256–63. (6) Rivas, M. G.; Gennaro, A. M. Chem. Phys. Lipids 2003, 122, 165– 169. (7) Cassera, M. B.; Silber, A. M.; Gennaro, A. M. Biophys. Chem. 2002, 99, 117–127. (8) Carrer, D. C.; Schreier, S.; Patrito, M.; Maggio, B. Biophys. J. 2006, 90, 2394–2403. (9) Rodi, P. M.; Cabeza, M. S.; Gennaro, A. M. Biophys. Chem. 2006, 122, 114–122. (10) Saupe, A. Z. Naturforsch 1964, 19A, 161–171. (11) Wisniewska, A.; Nishimoto, Y.; Hyde, J. S.; Kusumi, A.; Subczynski, W. K. Biochim. Biophys. Acta. Biomembranes. 1996, 1278 (1), 68– 72. (12) Hakansson, P.; Westlund, P. O.; Lindahl, E.; Edholm, O. Phys. Chem. Chem. Phys. 2001, 3, 5311–5319. (13) Stimson, L.; Dong, L.; Karttunen, M.; Wisniewska, A.; Dutka, M.; Ro´g, T. J. Phys. Chem. B. 2007, 111, 12447–12453. (14) Garay, A. S.; Rodrigues, D. E. J. Phys. Chem. B. 2008, 112, 8057– 8070. (15) Curdova´, J.; Capkova, P.; Pla´sek, J.; Repa´kova´, J.; Vattulainen, I. J. Phys. Chem. B. 2007, 111, 3640–3650. (16) Mravljak, J.; Konc, J.; Hodoscek, M.; Solmajer, T.; Pecar, S. J. Phys. Chem. B. 2006, 10, 25559–25561. (17) Sammalkorpi, M.; Lazaridis, T. Biophys. J. 2007, 92, 10–22. (18) Alonso, A.; Meirelles, N. C.; Tabak, M. Chem. Phys. Lipids 2000, 104, 101–111. (19) Ho¨ltje, M.; Fo¨rster, T.; Brandst, B.; Engels, T.; von Rybinski, W.; Ho¨ltje, H. Biochim. Biophys. Acta. 2001, 1511, 156–167. (20) Kaganer, V.; Mohwald, H.; Dutta, P. ReV. Mod. Phys. 1999, 71, 779–816. (21) Ocko, B. M.; Kelley, M. S. Langmuir 2002, 18, 9810–9815. (22) Peng, J. B.; Barnes, G. T.; Gentle, I. R. AdV. Colloid Interfaces Sci. 2001, 91, 163–219. (23) Vogel, A.; Scheidt, H. A.; Huster, D. Biophys. J. 2003, 85, 1691– 1701. (24) McMullen, T. P. W.; Lewis, R. N. A. H.; McEthaney, R. N. Curr. Opin. Colloid Interface Sci., 8, 459-468. (25) Berger, O.; Edholm, O.; Jahnig, F. Biophys. J. 1997, 72, 2002– 2013.

16842 J. Phys. Chem. B, Vol. 112, No. 51, 2008 (26) Berendsen, H. J. C.; van der Spoel, D.; van Drunen, R. Comp. Phys. Comm. 1995, 91, 43–56. (27) Lindahl, E.; Hess, B.; van der Spoel, D. J. Mol. Mod. 2001, 7, 306–317. (28) Berendsen, H. J. C.; Postma, J. P. M.; DiNola, A.; Haak, J. R. J. Chem. Phys. 1984, 81, 3684–3690. (29) Berendsen, H. J. C. Computer Simulations in Material Science; Meyer, M.; Pontikis, V. Eds.; 1991; pp 139-155. (30) Faraldo-Go´mez, J. D.; Smith, G. R.; Sansom, M. S. P. Eur. Biophys. J. 2002, 31, 217–227. (31) Feller, S. E. Curr. Opin. Colloid Interface Sci. 2000, 5, 218–224. (32) Feller, S. E.; Pastor, R. W. J. Chem. Phys. 1999, 111, 1281–1287. (33) Lindahl, E.; Edholm, O. Biopys. J. 2000, 79, 426–433. (34) Merger, F.; Wood, M. G.; Richardson, S. D.; Zhou, Q. Z.; Elrington, A. R. J. Am. Chem. Soc. 1988, 110, 6797–6803. (35) Nagle, J. F.; Tristram-Nagle, S. Biochim. Biophys. Acta 2000, 1469, 159–195. (36) Kucerka, N.; Kiselev, M. A.; Balgavy´, P. Eur. Biophys. J. 2004, 33, 328–334.

Vartorelli et al. (37) Lairion, F.; Disalvo, A. E. Biochim. Biophys. Acta, Biomembranes. 2007, 1768, 450–456. (38) Falck, E.; Patra, M.; Karttunen, M. Hyv¨; onen, M. T.; Vattulainen, I. Biophys. J. 2004, 87, 1076–1091. (39) Kaganer, V. M.; Osipov, M. A.; Peterson, I. R. J. Chem. Phys. 1993, 98, 3512–3527. (40) Kaganer, V. M.; Indenbom, V. L. J. Phys. II 1993, 3, 813–827. (41) Daura, X.; Gademann, K.; Jaun, B.; Seebach, D.; van Gunsteren, W. F.; Mark, A. E. Angew. Chem. Int. Ed. Engl. 1999, 38, 236–240. (42) Pake, G. E. Paramagnetic Resonance; Benjamin: New York, 1962. (43) Cadenhead, D. A.; Mu¨ller-Landau, F. Bioch. Biophys. Acta. 1976, 443, 10–18. (44) Humphrey, W.; Dalke, A.; Schulten, K. Visual Molecular Dynamics 1996, 14.1, 33–38. (45) Improta, R.; di Matteo, A.; Barone, V. Theor. Chem. Acc. 2000, 104, 273–279.

JP806476A