15558
J. Phys. Chem. B 2009, 113, 15558–15562
Conformational Changes of Lipids in Bilayers at the Dynamical Transition Near 200 K Seen by Raman Scattering N. V. Surovtsev*,†,‡ and S. A. Dzuba‡,§ Institute of Automation and Electrometry, and Institute of Chemical Kinetics and Combustion, Russian Academy of Sciences, NoVosibirsk, 630090, Russia, and NoVosibirsk State UniVersity, NoVosibirsk, 630090, Russia ReceiVed: August 6, 2009; ReVised Manuscript ReceiVed: October 7, 2009
Dynamical transition in biomolecules (proteins, membranes, etc.) is the phenomenon of increase of mean square atomic displacements, when temperature increases above 180-230 K. This increase is seen in neutron scattering and is ascribed to the increase of anharmonical motion of hydrogen atoms in biomolecules. In this work, Raman scattering study was performed for model synthetic membranes of dipalmitoylphosphatidylcholine in the spectral range of C-C vibrations (1050-1150 cm-1). It was demonstrated that dynamical transition is accompanied by appearance of newly achievable conformations of the lipid tails. In particular, the portion of the all-trans conformations decreases above the temperature of the dynamical transition. The data are compared with simple theoretical models. Introduction The phenomenon of “dynamical transition” in biomolecules (proteins, biomembranes, etc) is the change of dynamical response, when the temperature increases above 180-230 K. This transition is evidenced as a deviation of the mean square displacement of hydrogen atoms seen in the neutron scattering from the linear temperature dependence, expected from the harmonic approximation.1-6 Also, the dynamical transition is seen in Mo¨ssbauer absorption,7 infrared,8 and Raman9 spectroscopy and pulsed EPR of spin probes and labels.10 In spite of the numerous studies of the dynamical transition phenomenon, its nature is still not clear. In particular, one of the most debated topic concerns the problem of whether a biomolecule is a “slave” of the solvent.2-6 In the case of slaving, the dynamical transition phenomenon relates to the solvent dynamics rather than to dynamics of a biomolecule itself. Indeed, since water is an essential part of biological substances, determining the conformational states of biomolecules, it is hardly possible to separate the contribution of the dynamics of a biomolecule from the solvent dynamics. The study of dynamical transition in synthetic lipid bilayers could provide important new information. First, lipids are characterized by well-defined conformational states. Second, the dynamical transition in the bilayer interior is related mainly to dynamics of nonpolar lipid chains, which have no contact with polar solvent (water). Thus, in the case of lipid bilayers one can study the dynamical transition without “solvent effect”. The synthetic lipid bilayers undergo dynamical transition around 200 K. In our previous work9 we found that the Raman spectrum in the C-H stretching range (2800 - 3000 cm-1) is sensitive to the dynamical transition, which may have two possible interpretations: (i) the appearance of a static conformational disorder (or a disorder fluctuating at the time scale * To whom correspondence should be addressed. E-mail address: lab21@ iae.nsk.su. Fax: (7)-3833-333863. † Institute of Automation and Electrometry, Russian Academy of Sciences. ‡ Novosibirsk State University. § Institute of Chemical Kinetics and Combustion, Russian Academy of Sciences.
longer than picoseconds) and (ii) the appearance of relaxation at the picosecond time scale without change of the conformational state. It is known that C-C vibrations are sensitive to the conformational state of lipid chain (see, e.g., ref 11). In the present work we study the Raman spectrum of a lipid bilayer in the range of C-C vibrations and in a wide temperature range. A similar study was made previously in ref 11, but at that time the phenomenon of dynamical transition had not yet been known and the experimental data of ref 11 were too noisy in the temperature range of interest. Our experimental data will be compared with the theoretical model of ref 11, and also a simple phenomenological description will be proposed. Experiment Sample. Dipalmitoylphosphatidylcholine (DPPC, from Avanti Polar Lipids) was dispersed in water by vortex mixing with heating up to 55 °C (the temperature Tm of the reversible gelto-fluid phase transition of DPPC is 41 °C). The sample was made from 24 mg of DPPC and 50 mg of H2O and sealed in a glassy tube. This amount of water per lipid is higher than can be bound by lipid molecules. In this case the phenomenon of dynamical transition does not depend on particular water concentration.9 Raman Experiment. Raman spectra of an opaque white sample were measured in nominally right-scattering angle by a triple-grating TriVista 777 spectrometer. The image of the laser trace was projected onto the spectrometer entrance slit. The diaphragm selected only the image of the scattering volume from the sample itself, excluding the scattering light from the sample tube. A spectral slit of 2 cm-1 (fwhm) was used. The spectral range from 675 to 1170 cm-1 was studied. Raman scattering was excited by a solid-state laser (Millenia, Spectra Physics) with wavelength of 532 nm and power of 400 mW. Wavelength calibration of the spectrometer was done by a neon-discharge lamp. An optical closed-cycle helium cryostat was used, in which the sample tube was attached to a coldfinger through an indium layer. The sample temperature was measured by silicon thermosensor with accuracy better than 1 K.
10.1021/jp907586p CCC: $40.75 2009 American Chemical Society Published on Web 10/20/2009
Dynamical Transition in Biomolecules
Figure 1. Raman spectra of lipid-water bilayers at selected temperatures in the spectral range of skeletal mode vibration.
Raw spectra of inelastic light scattering were corrected for detector dark counts and for photoluminescence background in linear approximation. The Raman mode near 722 cm-1, corresponding to C-N stretching vibration, was used as the reference mode, whose contribution is temperature independent according to ref 12 (and references therein). The intensity of our Raman spectra was normalized by the integrated intensity of the 722 cm-1 mode. Results and Discussion Raman Spectrum and Line Interpretation. In Figure 1 there are shown Raman spectra of lipid DPPC-water bilayers at different temperatures in the range of 1050-1150 cm-1, which corresponds to the skeletal mode vibrations (C-C stretch) of the hydrocarbon chains. There are three main lines in the spectrum at T ) 50 K, centered at 1062, 1130, and 1100 cm-1. They are attributed to the all-trans state.11,13 The all-trans band at 1100 cm-1 is close to the frequency of the C-C stretch vibration near 1090 cm-1, which is ascribed to the chains containing gauche conformations.13 The 1090 cm-1 band dominates at T ) 316 K, where the lipid bilayer becomes fluid (Figure 1). Sometimes, this closeness provokes the erroneous interpretation of the band at 1100 cm-1 as corresponding to the gauche conformation of hydrocarbon chains, as in ref 12 and, importantly, in recent works (for example, ref 14). On the other hand, this closeness makes this mode inconvenient for quantitative analysis. The similar problem concerns the mode near 1062 cm-1, since it overlaps with the band of C-C vibration modes of chains with gauche conformations.13,15 The mode seen at 1130 cm-1 (Figure 1) seems to be most suitable for quantitative estimation of the temperature dependence of conformational states.15 It corresponds to the all-trans conformation state and is rather well separated from the other bands. The only remark is that in the fluid state there is a weak line at 1124 cm-1, corresponding to a “melted” state, but this mode is relatively weak and, in principle, can be separated from the 1130 cm-1 mode. Note that in ref 11 the authors did not provide a separation between these two modes, and the fact of nonzero intensity in the fluid state served as an argument that chains with gauche conformations also can contribute to the 1130 cm-1 line. Here we stick to the interpretation that only all-trans states contribute to the 1130 cm-1 line. Qualitative Consideration. From Figure 1 it is seen that the Raman spectrum does not show notable changes in the temperature range from 50 to 195 K. There is only slight gradual decrease seen here for the 1062 and 1130 cm-1 lines with temperature increase, accompanying by their slight broadening. Starting from ∼200 K, these changes become pronounced, and
J. Phys. Chem. B, Vol. 113, No. 47, 2009 15559
Figure 2. Differences between Raman spectra at neighboring temperatures of Figure 1.
at the temperature of gel-to-fluid transition, Tm ) 314 K, the discontinuous change of the Raman spectrum takes place. As 200 K is the temperature of dynamical transition, it could be suggested that the onset of the notable changes in the Raman spectrum seen near 200 K is related to this phenomenon. Note that the difference between the Raman spectrum at T > 200 K and the low-temperature spectrum qualitatively is similar to that seen above and below the gel-to-fluid transition (Figure 1). For illustration of this similarity, the differences between spectra at neighboring temperatures of Figure 1 are shown in Figure 2. One can see the decrease of the intensity of all-trans modes accompanying the increasing intensity of other modes. Note that the integral of the Raman spectrum (data not given) over 995-1150 cm-1 is temperature independent with a precision of ∼10% up to 300 K and with a precision of ∼25% up to 323 K (above Tm). It means that the main change in the Raman spectrum is the redistribution between different vibrational modes. In the gel-to-fluid transition, the lipid chains become less ordered, possessing a greater number of achievable conformations. Therefore, the stated above qualitative similarity between Raman spectrum changes at the gel-to-fluid transition and the dynamical transition leads to the suggestion that dynamical transition is accompanied by changes in the conformational state of hydrocarbon chains. So, the phenomenon of the dynamical transition may be related to the appearance of a static conformational disorder (or a disorder fluctuating at the time scale longer than picoseconds) at T ) Td and its increasing at T > Td (Td is the temperature of dynamical transition). Temperature Dependence of the 1130 cm-1 Mode. As it was pointed out above, for quantitative description of the conformational states the 1130 cm-1 line suits well. The intensity of the 1130 cm-1 line is known to be proportional to the concentration of the all-trans conformations.13,15 To extract the parameters of this line, we described the experimental spectrum in the range of 1030-1150 cm-1 by a sum of three Lorentzians. An example of such fitting is shown in Figure 3. The fit allows getting parameters of the 1130 cm-1 line in spite of the contributions from numerous modes in this spectral range. Note that above 314 K the closely located line at 1124 cm-1 corresponds to the melted state of conformations. In Figure 4 the best-fitted 1130 cm-1 parameters are shown for whole temperature range studied from 11 to 323 K. It is seen from Figure 4a that above ∼200 K a decrease of the integral intensity of the 1130 cm-1 line, Is, takes place. Is is scaled to unity at low temperatures. The temperature dependence of Is (Figure 4a) evidences the decrease of the relative portion of all-trans conformations above 200 K.
15560
J. Phys. Chem. B, Vol. 113, No. 47, 2009
Figure 3. Example of Raman spectrum fitting in the C-C range by three Lorentzians.
Figure 4. Temperature dependence of Lorentzian’s parameters for the line 1130 cm-1: (a) integral intensity (circles), (b) width (fwhm), (c) position. Temperature dependence of the ratio of the intensity of the 2880 cm-1 antisymmetric C-H vibration line and the 2850 cm-1 symmetric one is shown in the top part of the figure by triangles.
Also, the temperature behavior of the line width (Figure 4b) points out that anharmonicity of effective potentials for vibrations increases with temperature. The same conclusion can be made from the temperature dependence of the line position. It is very weak below 200 K and demonstrates significant shift above 200 K. And, finally, the discontinuous breaks at 314 K of temperature dependences of the integral intensity and position (Figure 4a, c) are related to gel-to-fluid transition, when line at 1130 cm-1 belonging to the all-trans conformations is replaced by the lowintensity line at 1124 cm-1 arising from melted configuration of hydrocarbon chains. Data of Figure 4 are to be compared with the previously obtained data on temperature dependence of the Raman scattering spectra at the spectral range of the CH2-stretching modes.9 The temperature dependence of the ratio of the intensity of the 2880 cm-1 antisymmetric C-H vibration line and the 2850 cm-1 symmetric one is shown in Figure 4a for comparison. It is seen that both sets of data are the same in the temperature range of the dynamical transition. We may conclude therefore that conformational changes of the lipids are the essential constituent of the dynamical transition near 200 K.
Surovtsev and Dzuba
Figure 5. Temperature dependence of the scaled interal intensity of the line 1130 cm-1 and the temperature dependence expected from the Pink model (dotted line) and from the two-excited-state model (solid line).
Our data for the integral intensity of 1130 cm-1 line (Figure 4a) are approximately in agreement with early data of ref 11. In spite of higher statistical error in the data,11 we should note that in the temperature range 220-303 K our data are rather systematically lower by 4-9% than the data of ref 11, being in better agreement outside of this temperature range. We attribute this to the absence of fair line decomposition in ref 11, where lines were “decomposed by eye”. Conformational States by the Pink Model. In the work of ref 11, a theoretical model was suggested (the Pink model), which describes the temperature dependence of the portion of different conformational states of hydrocarbon chains. This model uses a triangular lattice with multistate conformation of the chains, where each acyl chain is treated independently.11 In the framework of the Pink model, there are 10 allowed conformational states of the chain. One of the states is an alltrans conformation, and another one is a highly disordered melted state. Other intermediate states have effective energy barrier from Eg to 3Eg, where Eg is the internal energy needed to form a single gauche bond. In the ref 11 the value of Eg/kB ) 325 K was used. The Pink model does not provide an analytical expression but can be calculated numerically. From data of Figure 5 of ref 11, we digitalized the model’s prediction for the 1130 line intensity for a hydrocarbon chain to be in its all-trans state. It is given in Figure 5, which is to be compared with the temperature dependence of the integral intensity of the 1130 cm-1 line. One can see a rather good agreement between the Pink model and our data. This agreement supports the interpretation of the dynamical transition in lipid-water bilayers as a consequence of release of non-all-trans conformations of hydrocarbon chains. Two-Excited-State Model. The problem of the Pink model is multiplicity of the parameters and absence of the convenient analytical expression. We suggest here a simple adaptation of the Pink model. In Figure 5 of ref 11, the probability for a hydrocarbon chain to adopt the all-trans state, or any kink state, or its excited melted state is shown versus temperature. From this figure it is seen that the decrease of the probability of alltrans states as temperature increases is accompanied by the corresponding increase of the probabilities for kink and excited melted states. The sum of these three probabilities is approximately unity, with precision about 10-20%. From Figure 5 of ref 11, one may conclude that kink states appear above 150-200 K and excited melted states appear above 270-280 K. We suggest restricting the consideration to three states of the hydrocarbon chain only, instead of multiple possibilities of the
Dynamical Transition in Biomolecules
J. Phys. Chem. B, Vol. 113, No. 47, 2009 15561
Figure 6. Temperature dependence of the parameter IS-1(T) - 1 in Arrehenius plot and single thermoactivated law (dashed line) and the sum of two thermoactivated laws (solid line).
Pink model. The first one is the ground state (all-trans conformation), the second one is the intermediate state (mostly, kinklike conformation, labeled as “k”), and the third one is the excited melted state (labeled as “m”). Let us consider the probability for a hydrocarbon chain to be in the all-trans state, since this value can be directly compared with experimental Raman data Is. Thus, in the present model
Is(T) )
1 gk exp(-Uk /kBT) + gm exp(-Um /kBT) + 1
(1) where gk and gm are the degeneracy of kink and excited melted states, correspondingly, and Uk and Um are their effective barriers, respectively, to the energy level of the all-trans state. It is convenient to consider the value
IS-1(T) - 1 ) gk exp(-Uk /kBT) + gm exp(-Um /kBT)
(2) The experimental value IS-1(T) - 1 is shown in Figure 6 by circles in an Arrhenius plot, in which a single Arrhenius-type temperature dependence would correspond to a straight line. Since from data in Figure 6 it looks like barriers Uk and Um are significantly different (and from data of ref 11 also), at low temperatures the excitation to a kink state is expected to dominate. So, the linear dependence seen in Figure 6 for lowtemperature data may be ascribed to an Arrhenius law
IS-1(T) - 1 ) gk exp(-Uk /kBT)
(3)
with Uk/kB ) 650 K and gk ) 2.5. Using that, fitting data by the expression 2 for the whole temperature range studied results in the parameters Um/kB ) 6000 K and gm ) 1.8 × 108. The calculated from expression 2 curve is shown in Figures 5 and 6. From these data we conclude that the suggested model with only two thermally activated excited states (kink and melted) describes fairly well the Raman data for all-trans states. Estimated energy of the kink states Uk/kB ) 650 K corresponds well to the value of 2Eg/kB from ref 11, as it would be expected. But degeneracy gk is much lower than that predicted in the Pink model (the Pink’s gk ) 20). The effective energy of melted states corresponds to the barrier per C-C bond of DPPC hydrocarbon chain that may be estimated as Um/15kB ) 400 K. It is of the same order as energy Eg for forming a gauche bond,
which is intuitively expected. But degeneracy gm in the twoexcited-state model is much higher than that predicted in the Pink model (in ref 11 the estimation of gm ∼ 3.0 × 105). The simplified two-excited-model slightly better describes Raman data than the original Pink model, as is seen from Figure 6. So, this model may appear helpful for description of the dynamical transition phenomenon in lipid bilayers. The simplified model works slightly better than the Pink model, probably because of varying degeneracy parameters in the simplified model, while in the original Pink model these parameters are fixed. The values of degeneracy parameters are different for the Pink model and the two-excited-model. These parameters are fixed within the Pink model and reflect somehow the specific choice of a triangular lattice. It is expected that the restrictions by triangular lattice lead to underestimation of degeneracy gm. On the other hand, the two-excited-model is a useful approach for extraction degeneracy and effective barrier parameters but does not have the prediction power in comparison with the Pink model. In the framework of the two-excited-state model, only barrier Uk is important for the dynamical transition at 200 K. Indeed, the fit in Figures 5 and 6 corresponds to ∼9% of kink state and negligible contribution ∼10-3% from the melted state at T ) 200 K. The appearance of the melted state of hydrocarbon chains seems to be associated with other processes. Indeed, significant concentration of the melted state arises somewhere in the range 275-285 K (the fit in Figures 5 and 6 corresponds to 7% of the melted states at 280 K). This temperature range is known as subgel-gel phase transition (ref 16 and references therein). Probably, the second barrier Um and melted state should be ascribed to new conformational states released at subgel-gel phase transition. Also, the role of water freezing should be addressed in further study of this problem (see, e.g., ref 17). In the framework of the model, the description by thermal activation law questions the existence of real thermodynamical transition at temperatures around 200-230 K. This finding is in line with the conclusion of ref 18, where it was shown that there is no sudden change in the dynamics of the protein at temperatures around 200-230 K (the range of the dynamic transition for the proteins). Conclusion The Raman spectrum of lipid DPPC-water bilayers was studied in the spectral range of C-C vibrations as a function of temperature. It is found that at temperatures below ∼200 K the Raman spectrum does not change with temperature. Above 200 K the bands, corresponding to the all-trans conformational states, decrease as the temperature increases. And new contributions, corresponding to the C-C vibrations of lipid chains with gauche conformations, increase with temperature increase. The temperature dependence of the portion of the all-trans conformations extracted from the temperature dependence of the 1130 cm-1 line of hydrocarbon chains was compared with the numerical description provided by the Pink model, and a good agreement was found. Also, a simplified two-excited-state model was proposed which has an analytical expression and describes well the Raman data. As a temperature of 200 K corresponds to the temperature of the well-known dynamical transition, we suggest that the phenomenon of the dynamical transition at 200 K in the lipid-water systems is related to the release of conformational freedom degrees. This release changes the all-trans state of the chains to the excited state containing gauche configurations.
15562
J. Phys. Chem. B, Vol. 113, No. 47, 2009
Acknowledgment. The authors thank E. S. Salnikov for sample preparation. This work was supported by RFBR Grant No. 09-03-00588, by Siberian Branch of RAS No. 87, and by Ministry of Education and Science of RF, project No 2.1.1/ 1522. References and Notes (1) Doster, W.; Cusak, S.; Petry, W. Nature 1989, 337, 754–756. (2) Fitter, J. Biophys. J. 1999, 76, 1034–1042. (3) Fitter, J.; Lechner, R. E.; Dencher, N. A. J. Phys. Chem. B 1999, 103, 8036–8050. (4) Caliskan, G.; Mechtani, D.; Roh, J. H.; Kisliuk, A.; Sokolov, A. P.; Azzam, S.; Cicerone, M. T.; Lin-Gibson, S.; Peral, I. J. Chem. Phys. 2004, 121, 1978–1983. (5) Caliskan, G.; Briber, R. M.; Thirumalai, D.; Garcia-Sakai, V.; Woodson, S. A.; Sokolov, A. P. J. Am. Chem. Soc. 2006, 128, 3233. (6) Wood, K.; Plazanet, M.; Gabel, F.; Kessler, B.; Oesterhelt, D.; Zaccai, G.; Weik, M. Eur. Biophys. J. 2008, 37, 619–626. (7) Parak, F.; Frolov, E. N.; Mo¨ssbauer, R. L.; Goldanskii, V. I. J. Mol. Biol. 1981, 145, 825–833.
Surovtsev and Dzuba (8) Mallamace, F.; Chen, S. H.; Broccio, M.; Corsaro, C.; Crupi, V.; Majolino, D.; Venuti, V.; Baglioni, P.; Fratini, E.; Vannucci, C.; Stanley, H. E. J. Chem. Phys. 2007, 127, 045104. (9) Surovtsev, N. V.; Salnikov, E. S.; Malinovsky, V. K.; Sveshnikova, L. L.; Dzuba, S. A. J. Phys. Chem. B 2008, 112, 12361–12365. (10) Dzuba, S. A.; Kirilina, E. P.; Salnikov, E.S. J. Chem. Phys. 2006, 125, 054502. (11) Pink, D. A.; Green, T. J.; Chapman, D. Biochemistry 1980, 19, 349–356. (12) Gaber, B. P.; Peticolas, W. L. Biochim. Biophys. Acta 1977, 465, 260–274. (13) Yellin, N.; Levin, I. W. Biochim. Biophys. Acta 1977, 489, 177– 190. (14) Fox, C. B.; Uibel, R. H.; Harris, J. M. J. Phys. Chem. B 2007, 111, 11428–11436. (15) Csiszar, A.; Koglin, E.; Meier, R. J.; Klumpp, E. Chem. Phys. Lipids 2006, 139, 115–124. (16) Tristram-Nagle, S.; Nagle, J. F. Chem. Phys. Lipids 2004, 127, 3– 14. (17) Dioumaev, A. K.; Lanyi, J. K. Photochem. Photobiol. 2009, 85, 598–608. (18) Khodadadi, S.; Pawlus, S.; Roh, J. H.; Garcia Sakai, V.; Mamontov, E.; Sokolov, A. P. J. Chem. Phys. 2008, 128, 195106.
JP907586P
