Lipid-Mediated Clusters of Guest Molecules in Model Membranes and

May 3, 2017 - To our knowledge, formation of nanoscale clusters of guest amphiphilic molecules in biological membranes is a new phenomenon suggested ...
1 downloads 0 Views 1MB Size
Download PDF
Article pubs.acs.org/JPCB

Lipid-Mediated Clusters of Guest Molecules in Model Membranes and Their Dissolving in the Presence of Lipid Rafts Maria E. Kardash and Sergei A. Dzuba* Institute of Chemical Kinetics and Combustion, Russian Academy of Sciences, 630090 Novosibirsk, Russia Department of Physics, Novosibirsk State University, 630090, Novosibirsk, Russia ABSTRACT: The clustering of molecules is an important feature of plasma membrane organization. It is challenging to develop methods for quantifying membrane heterogeneities because of their transient nature and small size. Here, we obtained evidence that transient membrane heterogeneities can be frozen at cryogenic temperatures which allows the application of solid-state experimental techniques sensitive to the nanoscale distance range. We employed the pulsed version of electron paramagnetic resonance (EPR) spectroscopy, the electron spin echo (ESE) technique, for spin-labeled molecules in multilamellar lipid bilayers. ESE decays were refined for pure contribution of spin−spin magnetic dipole− dipolar interaction between the labels; these interactions manifest themselves at a nanometer distance range. The bilayers were prepared from different types of saturated and unsaturated lipids and cholesterol (Chol); in all cases, a small amount of guest spin-labeled substances 5-doxyl-stearic-acid (5-DSA) or 3β-doxyl5α-cholestane (DChl) was added. The local concentration found of 5-DSA and DChl molecules was remarkably higher than the mean concentration in the bilayer, evidencing the formation of lipid-mediated clusters of these molecules. To our knowledge, formation of nanoscale clusters of guest amphiphilic molecules in biological membranes is a new phenomenon suggested only recently. Two-dimensional 5-DSA molecular clusters were found, whereas flat DChl molecules were found to be clustered into stacked one-dimensional structures. These clusters disappear when the Chol content is varied between the boundaries known for lipid raft formation at room temperatures. The room temperature EPR evidenced entrapping of DChl molecules in the rafts.

■

INTRODUCTION The lipid content of plasma membranes is assumed to have a compositional heterogeneous structure. The growing evidence suggests that the plasma membrane is highly compartmentalized, thus allowing lipids and proteins to be organized in specific regions of varying size and composition.1−5 This membrane compartmentalization has been widely discussed, especially in relation to the concept of lipid raftsthe microand nanoscale assemblies of lipids, cholesterol (Chol), and proteins assumed to appear in the membranes because of the liquid ordered−disordered phase segregation of lipids.1−10 Lipid rafts are postulated to be involved in different functions in cellular membranes, such as trafficking, signal transduction, and membrane protein activity.6−10 It is a challenging task for developing methods to quantify membrane heterogeneities owing to their transient nature and small size. Super-resolution microscopy techniques can attain the 10−100 nm lateral resolution.1,4 Small-angle neutron scattering11,12 and fluorescence resonance energy transfer13,14 are used to investigate distance ranges ≤10 nm. Electron paramagnetic resonance (EPR) of spin labels can deliver information on the molecular scale. The EPR spectra of spinlabeled phospholipids can be used to study lipid raft formations14−18 providing the detailed phase diagram that determines coexisting liquid ordereddisordered phases.17,18 © 2017 American Chemical Society

The pulsed version of EPR, the electron spin echo (ESE) technique,19 was found to be useful for detecting nanoscale heterogeneities of spin-labeled cholestane20,21 and stearic acid22 molecules in lipid bilayers. Compared with conventional continuous wave (CW) EPR spectroscopy, the inhomogeneous line broadening disappears in the ESE technique;19 therefore, this technique is much more sensitive to the static spin−spin magnetic dipole−dipolar (d−d) interaction between the spin labels. The d−d interactions contribute to the ESE signal decay for labels separated by the distances at the nm scale; therefore, the ESE technique is a promising tool for searching nanoscopic heterogeneities in biological membranes. However, the ESE technique can be applied to the frozen state only. Although in this state the experimental difficulties related to the transient nature of membrane heterogeneity are lifted, it is unclear to what extent the structure of the frozen bilayer replicates the structure of the bilayer under physiological conditions. Nevertheless, some important hints might be obtained for the possible properties of the system under physiological conditions. Received: February 17, 2017 Revised: April 27, 2017 Published: May 3, 2017 5209

DOI: 10.1021/acs.jpcb.7b01561 J. Phys. Chem. B 2017, 121, 5209−5217

Article

The Journal of Physical Chemistry B

Therefore, p depends on B1 and the spectral density at the selected field position. The described mechanism of the echo decay is known as the “instantaneous diffusion mechanism”29 because the application of the second microwave pulse instantly changes the d−d interaction between the ith and jth spins, so the spin precession frequency for the ith spin is also instantly changed. Equation 1 predicts that the typical distances where the d−d interactions matter are around the value (g2μB2τ/ℏ)1/3 ∼ 5−10 nm (for τ of the order of 10−6 s that is typical for organic solids). Therefore, the ESE decays are sensitive to the nanoscale distance range. Equation 1 is to be averaged over all spatial positions that are available for the spin system. For a uniform spatial distribution, when probability for the selected spin to be located in a small elementary space element is proportional to its value, dr, the averaging provides the echo signal as37,38

In order to gain insight into the formation and properties of membrane heterogeneity, model membranes having simplified lipid compositions should be studied. The bilayers used here were prepared from fully saturated lipids 1,2-dipalmitoyl-snglycero-3-phosphocholine (DPPC), monounsaturated lipids 1palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine (POPC), doubly unsaturated lipids 1,2-dioleoyl-sn-glycero-3-phosphocholine (DOPC), or an equimolecular mixture DOPC/DPPC with Chol added in different contents. The DOPC/DPPC/Chol mixture is believed to mimic the cell membranes and therefore is often used as a model for the investigation of lipid raft formation.8,11,23−26 For the purpose of investigating the cluster formation of guest lipid-like molecules, the bilayers contained spin-labeled stearic acid, 5-doxyl-stearic-acid (5-DSA), or spinlabeled cholestane, 3β-doxyl-5α-cholestane (DChl). Stearic acid mimics fatty acids that are known to participate in membrane lipid homeostasis,27,28 whereas DChl has a chemical structure that is similar to Chol. Pure d−d contribution from the experimental ESE signal decay was extracted using the so-called “instantaneous spectral diffusion” effect.19,29−32 This approach has already been known for the detection of nanoscale clusters of paramagnetic centers of different origin in different media,20−22,31−36 including lipidmediated clusters of spin-labeled molecules found recently in frozen lipid bilayers. 20−22 A brief description of the phenomenon and results of model calculations are given below. ESE Decays Induced by Static d−d Interactions. Let us consider a spin system in an external magnetic field B0. Spins are subjected to an echo-forming microwave two-pulse sequence, which in the case of on-resonance excitation is denoted as π/2−τ−π−τ−echo. We select an arbitrary spin i in the system coupled by d−d interaction with another arbitrary spin j. Under the approximation of nonequivalent spins (which means that, for their resonance field positions, B0i and B0j, the gμ inequality holds r 3B ≪ |B0i − B0j |, where g is the g-factor and

⎛ E(τ ) = E0 exp⎜⎜ −C loc p ⎝

⎡ ⎤⎞ g 2μB 2 2 ⎢1 − cos ⎥⎟⎟ (1 − 3 cos θ ) τ ⎢⎣ ⎥⎦⎠ ℏr 3

⎞ ⎛ 8π 2 g 2μ 2 B E3D(τ ) = E0 exp⎜⎜ − Cpτ ⎟⎟ ⎠ ⎝ 9 3 ℏ

ℏrij 3

where C is the concentration taken in cm units. Spin labels in membranes are expected to have a twodimensional space distribution. For a uniform spin distribution in an infinite plane oriented in the magnetic field, an analytical formula can be obtained by replacing Cloc in eq 3 with the local surface concentration, σloc, and using the elementary surface area as dr. Subsequently, it is easy to obtain from eq 3 that the echo decay is proportional in that case to exp(−constσlocp(g2μB2τ/ℏ)2/3), with const depending on the plane orientation. For superposition of the planes randomly oriented in the space and for a finite size of a cluster of spin labels located in the plane, numerical averaging can be done. We performed these calculations for a circular cluster of radius R with labels uniformly distributed in the cluster. The g value was taken as g = 2.0060 which is a typical averaged value for nitroxides. In calculations, the distance of closest approach between the neighboring spin labels, Rmin, was also varied. Figure 1a shows the results. One can see that, for an infinite cluster, which is modeled by a large R value (R = 200 nm), and for Rmin = 0.5 nm, the echo decay may be fairly well presented by an approximating formula

(1 − 3 cos2 θij)τ (1)

where θij is the angle between the vector rij joining two spins and the external magnetic field B0 and the dimensionless parameter p reflects the off-resonance effects for excitation of the partner spin j (these effects are typical for spin-label EPR) p=

∫ g (B ) d B (B

0

B12 − B)2 + B12

⎞ ⎛ gμ B t p sin 2⎜ (B0 − B)2 + B12 ⎟ ⎠ ⎝ 2ℏ

where B1 and tp are the second microwave pulse amplitude and duration, respectively, and g(B) is the EPR spectral density (i.e., EPR line shape). The g(B) function is assumed to be normalized, ∫ g(B) dB = 1. Note that p < 1. At a selected field position, B0, the maximum p-value is attained when gμBtpB1/ℏ = π, fulfilled for the chosen pulse sequence. For B1 much smaller than the EPR line width, p may be roughly assessed as p ≈ 2B1g (B0 )

(4) −3

μB is the Bohr magneton), and in the absence of any other magnetic interactions in the system, the echo signal amplitude for the ith spin is varying with τ as37 g 2μB 2

(3)

where Cloc is the local spin concentration in the space available for the spin labels, which may be a nanoscopic compartment in a larger system, and integration is performed over this available space. In three-dimensional infinite space, eq 3 immediately results in the exponential ESE decay19,30

ij

ei(τ ) = 1 − p + p cos

∫ dr

E2D(τ ) ≅ E0 exp( −3.21σloc p(g 2μB 2 τ /ℏ)2/3 )

(5)

−2

for σloc taken in cm units. If R < 10−20 nm, the decays in Figure 1a are retarded as compared with the τ2/3 dependence. Data in Figure 1a also show that decays with larger Rmin (=2 nm) show some induction period in which the signal decays only slowly. However, as the ESE decays can be measured at τ delays larger than the dead-time limit (typically 0.1 μs), this

(2) 5210

DOI: 10.1021/acs.jpcb.7b01561 J. Phys. Chem. B 2017, 121, 5209−5217

Article

The Journal of Physical Chemistry B

Subsequently, the solvent was removed by nitrogen flow, followed by the storage for 12 h under a vacuum (10−2 bar). The obtained thin dry films were hydrated for 2 h at room temperature by adding bidistilled water, resulting in the formation of large multilamellar vesicles. For low-temperature measurements, the samples were put in glass tubes of 3 mm outer diameter and frozen by immersion in liquid nitrogen. For measurements at room temperature, the samples were put in glass tubes of 1 mm outer diameter and studied immediately after preparation. Stable nitroxide free radical 2,2,6,6-tetramethyl-4-oxo-piperidin-1-oxyl (TEMPONE) was purchased from Sigma-Aldrich and purified by recrystallization from hexane. It was dissolved in toluene at different concentrations varying between 10−3 and 5 × 10−3 M. The solutions were put into EPR tubes and then quickly frozen at liquid nitrogen. The samples were transparent glasses. The solvents hexane and toluene were obtained from commercial sources and purified by distillation. A Bruker ELEXSYS E580 9 GHz FT-EPR spectrometer (Bruker, Germany) was used, equipped with either a split-ring resonator (Bruker ER 4118X-MS3) or a dielectric resonator (Bruker ER 4118X-MD5) inside an Oxford Instruments CF 935 cryostat. The cryostat was cooled by flowing cold nitrogen gas, and the sample temperature was controlled at an accuracy of ±0.5 K. The split-ring resonator was used in pulsed experiments, whereas the dielectric resonator was used in CW experiments. To ensure the absence of EPR saturation, the incident microwave power in CW experiments was controlled. In pulse experiments, the resonator was overcoupled to provide a short ring time (∼100 ns). A two-pulse ESE sequence was employed. The pulse durations for the first and second pulses normally were 16 and 32 ns, respectively, and those for specially indicated cases were 8 and 16 ns, respectively. The pulse amplitudes were adjusted to provide a maximum of the echo signal. The time delay τ was scanned with a 4 ns step. All of the calculations were performed with originally developed computer programs using the PascalABC.NET computer software.

Figure 1. Simulation of echo decay for spin labels randomly distributed (a) on a surface within a circle of the radius R and (b) in a linear segment of length L. Squares, R(L) = 200 nm; circles, R(L) = 20 nm; upward triangles, R(L) = 10 nm; downward triangles, R(L) = 6 nm. The parameters used in calculations are (a) surface density σloc = 5 × 1012 cm−2, excitation parameter p = 0.2; (b) linear density λloc = 0.38 nm−1, p = 0.2. The solid lines present approximations given by eqs 5 and 6 for parts a and b, respectively. Rmin in all cases is 0.5 nm, except for the downward-shifted curve in part a where Rmin = 2 nm.

induction cannot manifest itself experimentally and the measured ESE decays may be considered as independent of Rmin. For a one-dimensional spin distribution, when spins are arranged in a line, the analogous approach immediately results in an expression for the echo decay as exp(−constλlocp(g2μB2τ/ ℏ)1/3), where λloc is the linear spin density in cm−1 units. We performed numerical calculations for the superposition of randomly oriented linear clusters in which labels are uniformly distributed. The results for different linear size L of the cluster are shown in Figure 1b. For an almost infinite cluster (L = 200 nm), the echo decay may be presented with good accuracy by an approximation E1D(τ ) ≅ E0 exp( −1.88λloc p(g 2μB 2 τ /ℏ)1/3 )

■

(6)

RESULTS AND DISCUSSION CW EPR Spectra at Room Temperature in DOPC/ DPPC/Chol Bilayer. Figure 2 shows representative CW EPR spectra for 5-DSA (a) and DChl (b) in DOPC/DPPC and DOPC/DPPC/Chol bilayers at 27 °C. These spectra are similar to those published previously for 5-DSA16 and DChl15 in phospholipid bilayers. The spectra were normalized by the amplitude of the left component. One can see that the 5-DSA spectra do not exhibit dependence on the Chol content, whereas a noticeable line broadening is observed for DChl with the increasing Chol content. For the width of the left component in the latter case, ΔB, Figure 2c shows a detailed dependence on the Chol content. For the ΔB dependence in Figure 2c, the maximum of this dependence takes place between 10 and 25 mol % of Chol content. The EPR line shape is determined by molecular motion and molecular ordering of individual spin labels14−18 and by d−d interaction between different spin labels.21 The data in Figure 2c may be explained by a lipid raft formation because at this temperature it is known to occur between approximately 8 and ∼30 mol % of Chol content8,23−25 (2 mol % of DChl may be added to the Chol content for the phase diagram8,23−25 in our case because of the closeness of chemical structures between Chol and DChl). As the rafts are considered

When L < 10−20 nm, the data in Figure 1b decay slower than the τ1/3 dependence. In order to refine the pure contribution of the d−d mechanism from the total echo decay, which is also caused by other different mechanisms such as nuclear spin flips, experiments must be performed in which the p parameter is varied. This could be done in two wayssee eq 2. First, ESE decays can be measured with different pulse amplitudes B1. Second, the experimental ESE time traces can be compared for the two field positions in which g(B0) is different. In the second approach, the condition of sufficiently low temperature must be fulfilled so that molecular librations are frozen39 and do not affect the ESE decays.

■

EXPERIMENTAL SECTION The lipids DPPC, POPC, DOPC, and Chol were obtained from Avanti Polar Lipids (Birmingham, AL, USA). 5-DSA and DChl were obtained from Sigma-Aldrich (Saint Louis, MO, USA). The lipid bilayers were prepared either from pure lipids or from DOPC/DPPC mixture (1:1) with Chol added at different proportions. In all cases, the spin-labeled substance, 5DSA or DChl, was added in a small amount, not larger than 2 mol %, and the chemicals were codissolved in chloroform. 5211

DOI: 10.1021/acs.jpcb.7b01561 J. Phys. Chem. B 2017, 121, 5209−5217

Article

The Journal of Physical Chemistry B

Figure 3. EPR spectra of 5-DSA in the POPC bilayer taken at concentrations of 0.1 mol % (thick line) and 1.4 mol % (thin line). Spectra are recorded at 190 K and normalized by the second integral. The dotted line presents the spectrum taken at 0.1 mol % and numerically convoluted with the d−d line shape, for σ = 5 × 1012 cm−2.

increasing concentration from 0.1 to 1.7 mol %. This broadening can be readily explained as arising because of the magnetic d−d interaction between the spin labels. Importantly, the EPR spectra in Figure 3 show that large broadening typical for the immediate association of spin labels40 is not observed. Therefore, the labels at all concentrations are diluted in their lipid surrounding. To simulate the EPR spectra broadening for a 5-DSA concentration of 1.4 mol % in Figure 3, the 5-DSA EPR spectrum taken at low concentration (0.1 mol %) was convoluted with the d−d line shape theoretically predicted by the Fourier transform data similar to those shown in Figure 1a. There was a rather good agreement between the simulated and experimental 5-DSA EPR spectra at a concentration of 1.4 mol %, for σ = 5 × 1012 cm−2 (it is presented in Figure 3 as a dotted line). Note that results of simulation of EPR spectra were found to depend strongly on the Rmin parameter: for larger Rmin, the larger σ was required for good agreement between calculation and experiment (data not shown). Therefore, CW EPR data cannot provide separate information on σ and Rmin. The host lipid lateral density, σ0, may be assessed from the known area that lipids occupy on the bilayer surface. At room temperature, it is approximately 60 Å2,41−43 resulting in the estimation that σ0 ≈ 1.7 × 1014 cm−2. Then we note that the 5DSA concentration of 1.4 mol % mentioned above corresponds to the lateral density σ = 2.4 × 1012 cm−2 which is twice lower than the value obtained in the fitting described above. This mismatch could indicate clustering of 5-DSA molecules; however, it is difficult to discuss this effect because the experimental spectra in Figure 3 differ only slightly and because of the strong influence of Rmin on the calculated spectra. As is seen below, more definite information on this mismatch can be obtained from the ESE decays. ESE Decays for 5-DSA in Chol-Free Bilayers. In order to extract the pure contribution of d−d interaction, ESE decays taken at 80 K were compared for two field positions of remarkably different spectral density g(B0) and, consequently, possessing remarkably different p-values (see eq 2). Figure 4 shows the representative results of the division of ESE decays at the positions indicated by arrows in the inset. The data are given for different molar concentrations χ of 5-DSA in the POPC bilayer and are plotted in the coordinates convenient for comparison with eq 5. The small oscillations observed on the curves are induced by static electron−nuclear interactions with

Figure 2. (a and b) Representative CW EPR spectra at 27 °C normalized to the same amplitude of the left component in DOPC/ DPPC (49:49) bilayer (dashed line) and DOPC/DPPC/Chol (42:42:14) bilayer (solid line): (a) 2 mol % of 5-DSA is added and (b) 2 mol % of DChl is added. (c) The line width ΔB for DChl as a function of Chol content; the dashed line is drawn to visualize the broad maximum in the dependence.

to be more ordered and denser structures than the surroundings, molecular motions are expected to be slower and more anisotropic within these structures; the EPR left component line broadening for DChl may be explained by trapping these molecules in the rafts. This interpretation is in agreement with the numerous results on EPR of spin-labeled lipids showing a biphasic behavior upon the raft formation,14−18 which was also explained by trapping spin-labeled molecules by the rafts. In the case of 5-DSA, the EPR spectra in Figure 2a do not exhibit a noticeable dependence on the presence of Chol. However, these spectra exhibit a biphasic behavior even in the absence of Chol because of the interplay between motion and ordering,16 possibly obscuring the trapping of these molecules by the rafts. CW EPR Spectra at Low Temperature. The CW EPR spectra for 5-DSA and DChl in all of the bilayers studied at low temperature (190 K) showed only a slight dependence on their concentration. For DChl in the DOPC/DPPC/Chol bilayer, the EPR spectra at 200 K were presented previously;21 they indicate clustering of DChl molecules. Figure 3 shows the representative CW EPR spectra taken for 5-DSA in POPC bilayers. Only a slight line broadening is observed with the 5212

DOI: 10.1021/acs.jpcb.7b01561 J. Phys. Chem. B 2017, 121, 5209−5217

Article

The Journal of Physical Chemistry B

lipid lateral density in the bilayer σ0. The bilayer contraction upon freezing may not be taken into account because, in the calibration experiments (see above), the analogous contraction was also not taken into account; therefore, the obtained σloc + σ value corresponds to that expected for room temperature (under the assumption that, in both cases, the contraction induces similar correction in the densities). In addition, it should be taken into account that σ/σ0 = χ. Figure 5 shows the

Figure 4. Ratio of ESE signal time traces taken for 5-DSA in the POPC bilayer at two field positions in the EPR spectrum as indicated by arrows in the inset. The data are plotted in coordinates convenient for comparison with eq 5. The straight lines present linear approximation. The data are given for the different concentrations χ of 5-DSA indicated in mol %. The temperature is 80 K.

the neighboring proton spins.19 The obtained echo decays obey the τ2/3 time dependence and are strongly concentration dependent. The remarkable concentration dependence of ESE decays seen in Figure 4 is to be compared with only a slight concentration dependence of CW EPR spectra (see Figure 3); this difference is a consequence of purifying the d−d interactions in the ESE experiment. ESE decays in Figure 4 may be directly used for obtaining the local spin label densities, σloc, by applying the theoretical eq 5. However, it should be taken into account that d−d interaction exists not only between spins in a given cluster but also between spins in different clusters. In this study, we use a model in which clusters are randomly distributed in a twodimensional space. The intercluster d−d interaction is expected to result in echo decay described by the expression similar to eq 5, with σloc replaced by the mean surface concentration σ. Therefore, the total echo decay is the product of the two factors, the intra- and intercluster ones, and the tangent of slope of the linear time dependences in Figure 4 must be assigned to the 3.21(σloc + σ)p(g2μB2τ/ℏ)2/3 value. In this model, σloc presents an excess of the surface density in the cluster over the mean surface density σ in the whole bilayer, i.e., in the absence of clustering, σloc = 0. When ESE decays are compared for two field positions, the parameter p in eq 5 is replaced by the difference p1 − p2, where subscripts refer to these positions. The difference p1 − p2 was determined in the calibration experiment performed at 80 K for freshly purified nitroxide TEMPONE dissolved at different concentrations in toluene glass (see the Experimental Section); the data obtained (not shown) were found to well fit the theoretical eq 4. As the EPR line shape g(B) for TEMPONE in toluene glass was found to be similar to that for 5-DSA in lipid bilayer at 80 K, the similar p values in both cases may be suggested. The TEMPONE volume concentration was taken without correction on the sample contraction upon freezing (see also below). It was found in this calibration experiment that p1 − p2 = 0.22 ± 0.02. Subsequently, σloc + σ values were obtained from the line slopes such as those in Figure 4. It is convenient to present the obtained data in dimensionless units, dividing σloc by the host

Figure 5. Normalized local surface concentration of spin-labeled 5DSA molecules, σloc/σ0, obtained from the slope of straight lines such as those in Figure 4, as a function of 5-DSA mole fraction χ in POPC, DOPC, and DPPC bilayers. Squares are data22 corrected for the p value equal to 0.22. The data for different bilayers are shifted by ±2% along the vertical scale for convenience. The solid lines are drawn to guide the eyes.

σloc/σ0 ratios for all of the samples studied as a function of the 5-DSA mole fraction χ. Here, data of previous measurements on 5-DSA in POPC bilayer22 also are included; these data however were corrected for the p1 − p2 = 0.22 value (the lower p1 − p2 value22 appeared because of an annoying error in the calibration experiment). The solid straight lines in Figure 5 demonstrate that the obtained σloc/σ0 concentration dependence in all of the cases shows a rapid increase for a small mean concentration χ that is replaced by a weaker dependence, or even decay, at concentrations larger than some critical value, χcrit. The fact that σloc/σ0 is well above zero, and even remarkably exceeds the mean concentration χ, unambiguously evidences cluster formation. This behavior in Figure 5 is similar for the different types of bilayers. The critical concentration χcrit in Figure 5 lies between 0.3 and 1.4 mol %, and the maximal attained σloc/σ0 value lies between 2 and 4%. Taking into account the evaluation σ0 ≈ 1.7 × 1014 cm−1 (see above), we readily obtain that intermolecular distance in the cluster is of the order of 4−5 nm, with typically ∼5 host lipid molecules separating the guest 5-DSA molecules. The clusters are formed due to attractive forces arising between the perturbations induced by guest 5-DSA molecules in their lipid surrounding. The surface density in a cluster, σloc, may increase up to some maximal value which is determined by the equilibrium between the attractive forces and repulsive forces between the guest molecules dominating at small distances. Below χcrit, the guest molecules are self-assembling into clusters, and above χcrit, the clusters are growing in size and in number. For the POPC bilayer, this process appears to include two stages: first, the clusters are formed at the first stage, and then, the more dense assemblies appear. 5213

DOI: 10.1021/acs.jpcb.7b01561 J. Phys. Chem. B 2017, 121, 5209−5217

Article

The Journal of Physical Chemistry B

Chol content, x. One can see that, with the Chol content increase, the σloc/σ0 value decreases, drops down to nearly the zero value, and then increases. The nonzero σloc/σ0 value at low and high Chol concentration unambiguously implies that guest 5-DSA molecules in the DOPC/DPPC/Chol bilayer are selfassembled in clusters, like it was shown above for the DPPC, POPC, and DOPC bilayers. The zero value attained between ∼15 and ∼25 mol % of Chol content means that clusters are dissolved. The similar range of Chol content in the DOPC/ DPPC/Chol bilayer with equimolecular DOPC/DPPC composition is known for the raft formation at physiological temperatures which looks as the ordered−disordered phase coexistence.8,23−25 The two rectangles in Figure 6 (bottom) show the lower and upper Chol concentration boundaries between which the liquid-ordered phase coexists with the liquid−disordered one (and with the gel phase at lower Chol content as well), which were derived from the published ternary phase diagrams.8,23−25 The widths of rectangles reflect scattering of the data obtained in different studies at different temperatures (varying in these studies between 10 and 30 °C). Note that one may suggest that the clusters are emerged during the freezing process. In our opinion, in view of the strong Chol dependence seen in Figure 6, this possibility is unrealistic, because there are no obvious reasons why Chol may influence, and so strongly, this hypothetical process. Most probably, the clusters of 5-DSA molecules are formed at room temperaturefor small and large Chol content, with their dissolving when Chol content varies in the intermediate range, between ∼15 and ∼25 mol %. The reason for the dissolving could be related with lipid rafts, because this Chol content range is characteristic for the raft formation in the bilayer. The destructive effect on the formation of clusters of guest molecules may imply that lipid rafts serve as a “membrane homogenizer”, in relation to such clusters. Also, the data obtained allow one to suggest that lipid rafts formed at physiological temperatures are frozen at cryogenic temperatures. ESE Decays for DChl in Chol-Free DOPC/DPPC Bilayer. Our previous ESE experiments20 with DChl incorporated into Chol-free DOPC/DPPC bilayers have shown that DChl molecules form lipid-mediated clusters with the strong orientational correlation between the neighboring molecules. This interpretation was later confirmed by CW EPR data,21 which were interpreted within the model of single-crystal-like structures formed by DChl molecules in the Chol-free bilayer. The previous ESE experiments20 involved extraction of the d−d contribution by comparing the ESE decays at two field positions. Here, the measurements were performed using the alternative approach, i.e., by comparing the decays for two different microwave amplitudes B1 (see above) with the same field positionat the maximum of the EPR spectrum (position 1 in the inset to Figure 4). Therefore, two different echoforming pulse sequences were employed, 16 ns − τ − 32 ns − τ − echo (the pulse sequence denoted as p.s.1) and 8 ns − τ − 16 ns − τ − echo (p.s.2); in the latter case, B1 is twice larger than in former. The obtained results for different DChl concentrations in Chol-free bilayer are given in Figure 7. These data fit well to the theoretical eq 6, suggesting that DChl molecules are stacked in a one-dimensional cluster (it is a lipid-mediated stacking because CW EPR spectra do not show close aggregation). This stacking is not surprising taking into account the flat structure of the DChl molecule.

If the cluster size is small, the simulation presented in Figure 1a shows that deviation of ESE time traces from the τ2/3-type behavior predicted by eq 5 must be expected. As the deviation is not observed in Figure 4, the cluster size is certainly >10 nm. From this estimation and the attained σloc values (Figure 5), it can be inferred that the number of guest molecules in the clusters is >10. ESE Decays for 5-DSA in DOPC/DPPC/Chol Bilayers. DOPC/DPPC/Chol bilayers were studied for DOPC and DPPC taken at equimolecular proportion with varying Chol content and with 5-DSA added at a concentration of 2 mol %. Therefore, the ternary system under investigation was DOPC/ DPPC/Chol with the molar proportion (49 − x/2):(49 − x/ 2):x, with 2 mol % of 5-DSA added (to be more precise, it is a quaternary system) and with the x value varying between 0 and 38 mol %. The contribution of d−d interactions to the ESE decays was obtained by comparing decays at the two field positions (Figure 6). The ESE decays obey the τ2/3 time dependence, with a

Figure 6. Top: The ratio of ESE signal time traces taken at two field positions in the EPR spectrum as indicated by arrows in the inset to Figure 4, for the 5-DSA in DOPC/DPPC/Chol bilayers of (49 − x)/ (49 − x)/x composition, for different x. The 5-DSA concentration is 2 mol %. The data are plotted in coordinates convenient for comparison with eq 5. Bottom: The local surface concentration given in dimensionless units as a function of x. The two rectangles show the low and upper x boundaries obtained in the literature at physiological temperatures;8,23−25 within these boundaries, the ordered−disordered phase segregation of lipids was detected (see text for other details).

somewhat stretched behavior for intermediate Chol concentrations. This stretching may be explained within the model of cluster formation; it is expected for the cluster size 10, and the intermolecular distance in the cluster is of the order of 4−5 nm, with typically ∼5 host lipid molecules separating the guest molecules. To our knowledge, formation of nanoscale clusters of guest amphiphilic molecules in biological membranes is a new phenomenon suggested only recently. For the DOPC/DPPC/Chol bilayer with varying Chol content, the combination of low-temperature ESE data and room-temperature CW EPR data suggests that lipid rafts are frozen when the temperature is dropped to cryogenic values. Because of the freezing, the experimental difficulties related with the transient nature of these heterogeneities are lifted, and the nanoscale-sensitive spin-label ESE technique can be used. These clusters in the DOPC/DPPC/Chol bilayers are destroyed in the presence of Chol, which is observed as a remarkable decrease in the local concentration in the cluster and/or the size of the cluster. This range of Chol content, in which the destruction occurs, lies nearly in the same range as that known for the raft formation at physiological temperatures. Therefore, the destruction of the clusters may be explained by the interaction of these guest molecules with lipid rafts. The CW EPR results at room temperature indicate that raft formation is accompanied by the absorption of the molecules by lipid rafts. The observed destruction−absorption effect provides an additional insight into the mechanism of functioning of lipid rafts as microreactors for biochemical processes.

■

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. ORCID

Sergei A. Dzuba: 0000-0001-8880-6559 Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS The authors thank Derek Marsh for useful suggestions at the beginning of this study and Denis Baranov for assistance in sample preparations. This work was supported by Russian Foundation for Basic Research, project no. 15-03-02186.

■

REFERENCES

(1) Garcia-Parajo, M. F.; Cambi, A.; Torreno-Pina, J. A.; Thompson, N.; Jacobson, K. Nanoclustering as a Dominant Feature of Plasma Membrane Organization. J. Cell Sci. 2014, 127, 4995−5005. 5216

DOI: 10.1021/acs.jpcb.7b01561 J. Phys. Chem. B 2017, 121, 5209−5217

Article

The Journal of Physical Chemistry B (24) Veatch, S. L.; Soubias, O.; Keller, S. L.; Gawrisch, K. Critical Fluctuations in Domain-Forming Lipid Mixtures. Proc. Natl. Acad. Sci. U. S. A. 2007, 104, 17650−17655. (25) Davis, J. H.; Clair, J. J.; Juhasz, J. Phase Equilibria in DOPC/ DPPC-d62/Cholesterol Mixtures. Biophys. J. 2009, 96, 521−539. (26) Davis, J. H.; Ziani, L.; Schmidt, M. L. Critical Fluctuations in DOPC/DPPC-d62/Cholesterol Mixtures: 2H Magnetic Resonance and Relaxation. J. Chem. Phys. 2013, 139, 045104. (27) Kaneda, T. Iso- and Anteiso-Fatty Acids in Bacteria: Biosynthesis, Function, and Taxonomic Significance. Microbiol. Rev. 1991, 55, 288−302. (28) Zhang, Y.-M.; Rock, C. O. Membrane Lipid Homeostasis in Bacteria. Nat. Rev. Microbiol. 2008, 6, 222−233. (29) Klauder, J. R.; Anderson, P. W. Spectral Diffusion Decay in Spin Resonance Experiments. Phys. Rev. 1962, 125, 912−932. (30) Salikhov, K. M.; Dzuba, S. A.; Raitsimring, A. M. The Theory of Electron Spin-Echo Signal Decay Resulting from Dipole-Dipolar Interactions between Paramagnetic Centers in Solids. J. Magn. Reson. 1981, 42, 255−276. (31) Raitsimring, A. M.; Salikhov, K. M. Electron Spin Echo Method as Used to Analyze the Spatial Distribution of Paramagnetic Centers. Bull. Magn. Reson. 1985, 7, 184−217. (32) Raitsimring, A. In Biological Magnetic Resonance; Berliner, L. J., Eaton, G. R., Eaton, S. A., Eds.; Kluwer/Plenum Publishers: New York, 2002; Vol. 19, pp 461−491. (33) Samoilova, R. I.; Raitsimring, A. M.; Tsvetkov, Y. D. The Structure of Radical Tracks in Methanol Irradiated by Tritium BetaParticles. Radiat. Phys. Chem. 1980, 15, 553−559. (34) Rakvin, B.; Maltar-Strmecki, N.; Nakagawa, K. Pulsed EPR Study of Low-Dose Irradiation Effects in L-Alanine Crystals Irradiated with Gamma-Rays, Ne and Si Ion Beams. Radiat. Meas. 2007, 42, 1469−1474. (35) Marrale, M.; Brai, M.; Barbon, A.; Brustolon, M. Analysis of the Spatial Distribution of Free Radicals in Ammonium Tartrate by Pulse EPR Techniques. Radiat. Res. 2009, 171, 349−359. (36) Raitsimring, A. M.; Tregub, V. V. Electron Spin Echo Decay Kinetics of an Ion Track in Beta-Irradiated Frozen Solution of Sulfuric Acid. Numerical Simulation by the Monte Carlo Method and Experiment. Chem. Phys. 1983, 77, 123−130. (37) Milov, A. D.; Maryasov, A. G.; Tsvetkov, Y. D. Pulsed Electron Double Resonance (PELDOR) and Its Applications in Free-Radicals Research. Appl. Magn. Reson. 1998, 15, 107−143. (38) Abragam, A. The Principles of Nuclear Magnetism; Oxford Clarendon Press: Oxford, U.K., 1961; Chapter 4. (39) Erilov, D. A.; Bartucci, R.; Guzzi, R.; Marsh, D.; Dzuba, S. A.; Sportelli, L. Librational Motion of Spin-Labeled Lipids in HighCholesterol Containing Membranes from Echo-Detected EPR Spectra. Biophys. J. 2004, 87, 3873−3881. (40) Rabenstein, M. D.; Shin, Y.-K. Determination of the Distance between 2 Spin Labels Attached to a Macromolecule. Proc. Natl. Acad. Sci. U. S. A. 1995, 92, 8239−8243. (41) Berger, O.; Edholm, O.; Jahnig, F. Molecular Dynamics Simulations of a Fluid Bilayer of Dipalmitoylphosphatidylcholine at Full Hydration, Constant Pressure, and Constant Temperature. Biophys. J. 1997, 72, 2002−2013. (42) Berkowitz, M. L. Detailed Molecular Dynamics Simulations of Model Biological Membranes Containing Cholesterol. Biochim. Biophys. Acta, Biomembr. 2009, 1788, 86−96. (43) Kučerka, N.; Nieh, M.-P.; Katsaras, J. Fluid Phase Lipid Areas and Bilayer Thicknesses of Commonly Used Phosphatidylcholines as a Function of Temperature. Biochim. Biophys. Acta, Biomembr. 2011, 1808, 2761−2771.

5217

DOI: 10.1021/acs.jpcb.7b01561 J. Phys. Chem. B 2017, 121, 5209−5217