J. Phys. Chem. B 2007, 111, 1377-1383
1377
Heterogeneity of Water at the Phospholipid Membrane Interface Victor V. Volkov,*,† D. Jason Palmer,† and Roberto Righini†,‡ European Laboratory for Nonlinear Spectroscopy (LENS), UniVersity of Florence, Via Nello Carrara 1, 50019 Sesto Fiorentino, Italy, and Department of Chemistry, UniVersity of Florence, Via della Lastruccia 3, 50019 Sesto Fiorentino, Italy ReceiVed: September 9, 2006; In Final Form: December 11, 2006
Femtosecond infrared (IR) two-color pump-probe experiments were used to investigate the nonlinear response of the D2O stretching vibration in weakly hydrated dimyristoyl-phosphatidylcholine (DMPC) membrane fragments. The vibrational lifetime is comparable to or longer than that in bulk D2O and is frequency dependent, as it decreases with increasing probe frequency. Also, the lifetime increases when the water content of the sample is lowered. The measured lifetimes range between 903 and 390 fs. A long-lived spectral feature grows in following the excitation and is attributed to photoinduced D-bond breaking. The photoproduct spectrum differs from the steady state difference Fourier transform infrared (FTIR) spectrum, showing that the full thermalization of the excitation energy happens on a much longer time scale than the time interval considered (12 ps). Further evidence of the inhomogeneous character of the water residing in the polar region of the bilayer comes from the spectral anisotropy. The water molecules absorbing on the low frequency side of the absorption band show no decay at all of the anisotropy, while an ultrafast partial decay appears when the high frequency side of the spectrum is probed. The overall behavior differs remarkably from that observed with similar experiments in bulk water and in water segregated in inverse micelles. In weakly hydrated phospholipid membranes, water molecules are present mostly as isolated species, prevalently involved in strong, rigid, and persistent hydrogen bonds with the polar groups of the bilayer molecules. This specific character appears to have a direct effect on the structural stability and thermal properties of the membrane.
Introduction It is known that the degree of hydration and the structural dynamics of water molecules at the cellular envelope polar interface play very important roles in a membrane’s structural stabilization and in the definition of its biological activity.1-6 Therefore, understanding of the basic mechanisms of native membrane functioning at a molecular level requires certain knowledge of the organization of water molecules in the polar compartments of bilayers and the characterization of their dynamics: energy flow processes and physical motion. The available experimental tools for this purpose are rather limited; neutron diffraction4,7 and NMR8-12 measurements show certain distributions of water molecules above the membrane surface, within the polar interface, and even in the hydrophobic compartment of phospholipid bilayers. Unfortunately, neither technique provides deep insight about local molecular arrangement and dynamics in those distributions. The complexity arises from the fact that in each of these spatial domains water is expected to have specific organization and dynamics. It is therefore difficult to access the gradual structural variance, progressing from the continuous water phase to solitary molecular species in the hydrophobic compartment. The dynamics in bulk water is the first point of reference to consider. Owing to rapid progress both in methods of ultrafast optical spectroscopy and in computing power, there is good knowledge of the vibrational lifetime, energy transfer, and orientational dynamics as well as of the intermolecular connectivity (making and breaking of hydrogen bonds) in the bulk liquid phase.13-25 † ‡
European Laboratory for Nonlinear Spectroscopy (LENS). Department of Chemistry.
Recently, several papers have also been published concerning interfacial and confined water mostly in nanosize materials.26-37 Since surfactant micelles are considered to be good membrane mimetic systems, it is particularly interesting to review the results of Fayer’s33-36 and Bakker’s37 groups about the dynamics of water confined in nanometric reverse micelles: the vibrational relaxation and the orientational dynamics show a clear slowing down with decreasing size of the water cluster, accompanied by a blue shift of the OD stretching infrared absorption. In those systems, water clusters show less and less bulk liquid character as their size decreases, with a growing role played by the interfacial molecules. However, it is our aim to show that the confinement of water molecules in phospholipid membranes is of a different type. Molecular dynamics (MD) simulations38-42 confirmed and substantiated the picture (provided by NMR8-12 and neutron scattering4,7 experiments) that water molecules permeate the bilayer interface with a steeply decreasing concentration as one proceeds toward the inner hydrophobic region of the bilayer. The majority of the water molecules residing in the polar region of the membrane are H-bonded to the phosphate groups, with a smaller fraction binding the carbonyl groups located deeper in the bilayer.38-41 These effects are evidenced in the intensity profile of the infrared (IR) spectrum of the D2O-hydrated dimyristoylphosphatidylcholine (DMPC) membrane in Figure 1. In comparison to the bulk D2O spectrum, the high frequency side of the stretching band (generally associated with weakly D-bonded and solitary water molecules) shows a reduction in intensity. The spectroscopic data and the MD simulations on hydrated membranes are then consistent with the picture of water
10.1021/jp065886t CCC: $37.00 © 2007 American Chemical Society Published on Web 01/24/2007
1378 J. Phys. Chem. B, Vol. 111, No. 6, 2007
Figure 1. (A) FTIR spectra of stretching modes in neat D2O (blue), in DMPC membrane fragments (black), and in reverse AOT micelles when D2O/AOT ≈ 20 (magenta). The optical density scale is shown for the AOT sample, while the spectra of neat D2O and of D2O in membrane fragments are properly scaled. (B) Spectra of the pump pulse according to the excitation conditions in time-resolved experiments.
molecules sparsely distributed in the proximity of the polar groups, to which they form rather strong hydrogen bonds. The occurrence of water-water H-bonded interactions is definitely less frequent, and equally low is the number of solitary, nonH-bonded water molecules inside the polar crust. In this paper, we present the results of pump-probe infrared absorption experiments performed on a sample consisting of weakly hydrated DMPC membrane fragments in D2O. The broadband infrared pulse excites the D2O stretching vibrations around 2450 cm-1; the time-dependent changes induced in the IR absorption band are measured by a second broad pulse in the same spectral region. The time evolution of the differential spectra and of their anisotropy allows the measurement of the vibrational lifetimes, the orientational relaxation, and the energy transfer rate, which represent important parameters for the characterization of the water-membrane structural and dynamical correlations. Experimental Section 1,2-Dimyristoyl-sn-glycero-3-phosphocholine (DMPC) was obtained from SIGMA (product number P2663). In order to prepare membrane fragments, first, we dissolved DMPC in chloroform and dried it on a glass plate. Second, after addition of D2O, the phospholipid suspension underwent mechanical mixing until the sample (10 µm film between two CaF2 windows) demonstrated proper optical quality. Using the optical density of CdO and OD stretching modes, we prepared samples in which the lipid-to-water ratios were 1:2, 1:1, and 2:1. It is worth noting that the prepared membrane samples are therefore rather dry and that this level of hydration is characteristic of biological systems. Using a Fourier transform infrared (FTIR) spectrometer (Shimadzu 8400S), we measured the steady state spectra of 0.1 M reverse sodium bis(2-ethylhexyl) sulfosuccinate (AOT) micelles in toluene for D2O/AOT ratios of 1, 5, 10, and 20. We detected the ultrafast infrared response of water using a spectrometer based on a Ti:sapphire laser-amplifier system (Thales Laser, Paris) producing a 1 kHz train of 60 fs, 620 µJ pulses at 800 nm. This output is equally split to generate a 2 µJ mid-IR pump pulse (vertical polarization) in an optical parametric amplifier (OPA) (TOPAS, Light Conversion Ltd., Vilnius, Lithuania) and a 1.5 µJ mid-IR probe pulse in a homebuilt OPA.43 The spectral width of both OPAs is about 200 cm-1 in the middle IR spectral range. The pump pulse traverses a
Volkov et al. variable delay line and a wave-plate (to control the polarization state of the pump radiation) before it is focused with the probe pulse on the sample. Before it reaches the sample, a fraction of the probe pulse is split off and used as a reference. After the sample, both probe and reference pulses are spectrally dispersed in a spectrometer (TRIAX 180, HORIBA Jobin Yvon, Milano, Italy) and imaged separately on a double 32-channel mercury cadmium telluride detector (InfraRed Associates Inc., Florida). The ratio of the probe to the reference spectra gives the differential absorption spectrum which is recorded as a function of the pump-probe time delay. The ab initio calculations of the structure and of the vibrational frequencies of water coordinated to a phosphate cation were carried out using the B3LYP/6-31G(d) density functional as implemented in Gaussian 98.44 Results In Figure 1A, we show the optical density of OD stretching modes in a phospholipid membrane sample where the lipid-towater ratio is 1:2. For comparison, the spectra of neat deuterium oxide and of water in reverse AOT micelles (for a D2O/AOT ratio of ≈20) are also shown. In all cases, the transitions are dominated by inhomogeneous broadening due to the variance in the degree of coordination with neighboring molecules.15,19,22-24 In order to distinguish the physical nature of the involved substates, we performed pump-probe experiments under three excitation conditions; see Figure 1B: (i) both pump and probe pulses identically cover the full bandwidth of D2O absorption at 2450 cm-1 (“full” excitation); (ii) the probe spectrum is unchanged, while the pump spectrum is tuned to the low frequency edge of the water absorption band (“red” excitation); (iii) similar to scheme ii but with the pump pulse tuned to the high frequency side of the water band (“blue” excitation). Figure 2 shows the transient spectra measured at different time delays, with pump and probe polarizations at the magic angle, for the 1:2 sample with the three excitation schemes described above. At short delay times, the spectral evolution is dominated by the population relaxation of the vibrational excited states. For longer delays, similar to what is observed in bulk water,45 a broad, predominantly negative spectral feature is observed which persists for several picoseconds, that is, at time delays much longer than the OD vibrational lifetime. We attribute this time-dependent spectral feature to the decrease of the extinction coefficient of the OD stretching mode due to the breaking of the D-bonds. The ensemble of liquid structures with a temporarily disrupted hydrogen/deuterium bonded network has been referred to in previous publications as a photoproduct.45 The presence of this spectral component complicates the analysis of the “true” time evolution of the OD stretching excited states. We can obtain the photoproduct-free spectral dynamics by removing the contribution of the photoproduct from the spectra measured under full excitation. To evaluate this contribution, we analyze the kinetics at a “zero-crossing” frequency, that is, a frequency at which, at very short delay, when the photoproduct contribution is absent, the excited state signal is zero balanced with the bleach contribution (label 1 in Figure 2A). This procedure is applicable provided that the energy transfer contribution to the overall signal is small. For long delays (>3 ps), when the population of the OD excited states has completely decayed, the measured signal gives directly the profile of the photoproduct spectrum (Figure 3A). The experimental data show that its profile is essentially unchanged from 1.8 to 12 ps; the time dependence of its intensity gives directly the time evolution of the photoproduct
Heterogeneity of Water
J. Phys. Chem. B, Vol. 111, No. 6, 2007 1379
Figure 3. (A) Time-resolved spectra under full-width excitation at five delay times: 1.8, 2.6, 3, 5, and 10 ps. (B) Steady state FTIR temperature difference spectrum obtained from data recorded at 300 and 302 K. Figure 2. FTIR spectrum of OD stretching modes of D2O in DMPC membrane fragments (panel A). Time-resolved spectra of the stretching modes at 0.2, 0.5, 1, and 4 ps (black, red, green, and blue lines, respectively), when the excitation spectrum covers the full bandwidth (panel B), the low frequency red side (panel C), and the high frequency blue side (panel D) of the absorption spectrum, according to the excitation schemes (see the Experimental Section).
in this time interval. The photoproduct spectrum is characterized by a negative band that grows in the first 2 ps and then decays to a lower intensity which remains constant over the 12 ps time window considered in our experiments. On the basis of the assumptions outlined above, for short delay times, we attribute the intensity measured at the zerocrossing frequency entirely to the photoproduct: its time dependence then reflects directly its formation kinetics. Thus, we can completely reconstruct the time evolution (rise and decay, see Figure 4) of its spectrum and subtract it, for each delay time, from the measured spectra. From the photoproduct-free signal, we can measure the dynamics of the OD stretching excitation at different probe frequencies. In order to avoid complications due to the overlapping with the excited state absorption, we restricted our analysis to three frequencies: 2496, 2572, and 2630 cm-1 (corresponding to labels 2, 3, and 4 in Figure 2A) in the high frequency half of the absorption band. In Figure 5A, we report the population decays under the magic angle condition for the 1:2 sample. Table 1 provides the population lifetimes for the three samples we prepared. The use of polarized pump and probe beams allows the measurement of the spectral anisotropy as a function of the delay time. The anisotropy is calculated according to
r(τ) )
I|(τ) - I⊥(τ) I|(τ) + 2I⊥(τ)
where the | and ⊥ symbols describe parallel and perpendicular mutual orientations of the pump and probe polarization directions. The anisotropy has its maximum at τ ) 0 (the theoretical value is r(0) ) 0.4). There are in general two mechanisms by which the initial anisotropy is lost: rotational dynamics (includ-
Figure 4. Kinetics detected at 2500 cm-1 (red line) and photoproduct contribution (black line) derived as described in the text. The inset expands the region above the dotted line in the panel.
ing librational inertial motion and orientational diffusion) of the excited oscillators and resonant energy transfer between different oscillators. In Figure 5B, we report the time-dependent anisotropy measured for the 1:2 sample. Discussion In Figure 1A, we compare the spectrum measured in the phospholipid membrane with those of bulk water and of water segregated in inverse AOT micelles. It is quite evident that the three spectra differ for the relative intensities of their high and low frequency components. The enhancement of the blue part of the spectrum, in comparison to bulk water, has been widely discussed in the case of reverse micelles,33-37 where it was attributed to the frustration of the aqueous network due to the confinement that arises below a certain size of the internal cavity hosting the water droplet. Figure 1A shows that, on the contrary, in the weakly hydrated membrane, we have a depletion of the high frequency part of the spectrum. In order to interpret and discuss the results outlined above, one has to consider the present knowledge about the water binding and organization in the polar part of the bilayer. NMR8-12 and neutron scattering4,7 experiments demonstrate that water is largely present in the polar region and permeates even the hydrophobic compartment of phospholipid membranes; of course, the water concentration decreases steeply as one moves toward the inner part of the
1380 J. Phys. Chem. B, Vol. 111, No. 6, 2007
Volkov et al.
Figure 5. (A) Photoproduct-free population relaxation under magic angle polarization conditions. (B) Anisotropy decay detected at the spectral frequencies indicated by arrows 2 (solid circles), 3 (open circles), and 4 (open squares) in Figure 2.
TABLE 1: Population Lifetimes for the Three Samples We Prepared frequency lipid-to-water ratio
2496 cm-1
2572 cm-1
2630 cm-1
1:2 1:1 2:1
620 fs 700 fs 903 fs
490 fs 510 fs 780 fs
390 fs 380 fs 620 fs
bilayer, as a consequence of the hydrophobic nature of the hydrocarbon tails. Besides the few available experimental data, a number of molecular dynamics (MD) simulations have been published that propose a detailed picture of the structure and dynamics, at the molecular level, of hydrated phospholipid membranes. A recent paper on this subject by Klein and co-workers41 gives a detailed analysis of the hydrogen bonds between water molecules and the polar groups of DMPC. According to their findings, and in full agreement with previous simulations,38-40 water is preferentially bonded to the phosphate groups: practically all the double-bonded oxygens of the PO4 moiety form at least one hydrogen bond, and about 75% of them form two H-bonds. The less accessible carbonyls of the tails are also H-bonded, with a probability which is about 4 times smaller with respect to that of the phosphate groups. The hydrogen bond between water and the oxygens of the phosphate and carbonyl groups is a very strong one and is strictly directional:41 in a very recent paper, Chanda et al.42 reported a binding energy of 11.7 kcal/mol for the PO4 group, about 3 times larger than that of the waterwater bond in the bulk liquid. The choline group is the most exposed to the external water; although there is no evidence of H-bond formation with the nitrogen atom, a well-defined solvation shell around this group is revealed by the simulation.41 It is important to stress that all of the simulations published to date have been done with a large excess of water; instead, we performed our experiments on weakly hydrated membranes, with a phospholipid/water ratio ranging between 0.5 and 2. It is then quite probable that, in the dehydration process, the less strongly bonded water molecules, that is, those of the liquid film on top of the bilayer surfaces and those solvating the choline groups, are lost first. On the basis of our FTIR data and of the predictions of the MD simulations, we can qualitatively state that in our samples the majority of the D2O molecules are specifically bonded to phosphate and carbonyl moieties of the
phospholipid membrane. The presence of such a direct and strong interaction between water and carbonyl oxygens in DMPC bilayers was very recently demonstrated with a twocolor pump-probe experiment.46 The low value of the water/ DMPC ratio suggests that the carbonyl-bound molecules are essentially solitary, with a low probability of binding to other water molecules. Obtaining quantitative information on the configuration of the water molecules and correlating it to the spectral features is definitely more arduous. In the hydrated sample of the simulation in ref 38, there are about 13 water molecules coordinated around each DMPC molecule. On average, 4.8 of them are H-bonded to the phospholipid and almost 1 is simultaneously bonded to two nearby polar groups, thus contributing to the stabilization of the layer. In our sample, there are about two water molecules (or less) per DMPC unit. We can easily admit that the one water molecule forming two D-bonds to the DMPC groups, predicted by the MD simulation, will survive the dehydration process: it is then possible that about half of the water molecules in our sample (approximately 1 out of 2 for each DMPC unit) have both deuterium atoms involved in bridging two phospholipid polar groups. These molecules will then absorb in the red and central parts of the band. The water molecules with only one deuterium engaged in bonding to DMPC will contribute both in the central part and in the blue part, where the dangling OD bonds absorb. Finally, although the limited number of D2O molecules in the sample and the competing attraction of the polar groups do not favor strong water-water interactions, we cannot rule out the presence of frustrated or weakly bound dimers, trimers, and other very small clusters. In agreement with the comparative enrichment of D-bonded species upon dehydration, the spectral dynamics under full-band and red excitation (Figure 2B and C) confirms that the prepared system is dominated by deuterium-bonded strong waterphospholipid interactions. Here, we observe a decay of intensity without a noticeable spectral evolution. In contrast, Figure 2D demonstrates that, under blue excitation, the bleaching signal experiences an intensity redistribution from the high to the low frequency side in less than 1 ps. We attribute this effect to an essentially intramolecular energy transfer involving the water molecules forming just one D-bond with DMPC: the vibrational energy is transferred from the blue-absorbing OD dangling bond to the OD oscillator engaged in the D-bond to the DMPC polar moiety, whose absorption falls in the central red region of the spectrum. This picture implies that for this fraction of water molecules the two OD bonds are essentially independent oscillators, with a rather large frequency separation. The assumption is confirmed by the ab initio density functional theory (DFT) calculation (data not shown) that we performed on a very simplified system, consisting of one water molecule bonded to a phosphate group: the two calculated stretching normal modes of the D2O molecule have a frequency separation of about 250 cm-1, and each of them is almost perfectly localized on one of the two bonds. A similar energy transfer has been observed by Bakker’s group in a recent infrared pump-probe experiment47 on a single water molecule enclosed by two acetone molecules. Their system has some analogies with the present one, and also in that case, the transient spectra clearly show that a blue-to-red energy transfer takes place. The presence of this ultrafast energy transfer introduces a source of error in the evaluation of the photoproduct contribution at short times. However, considering the relative contribution of the blue-absorbing states, and the competition with the
Heterogeneity of Water
J. Phys. Chem. B, Vol. 111, No. 6, 2007 1381 spectrum (600 fs, see below). Figure 6A shows that a satisfactory agreement can be obtained with the following parameters: vibrational lifetime (1/k1) ) 600 fs; hydrogen-bond breaking time constant (1/k2) ) 450 fs; hydrogen-bond recovery time (1/k3) ) 3.4 ps. The treatment just described does not consider the possible contribution of energy transfer to the rise of the photoproduct signal at early times. To describe the energy transfer contribution adopting the following approximation expression
Figure 6. Rise and partial decay of the photoproduct signal. In both panels, the open square black line represents the experimental data. (A) Red full line, calculated (see text) not including the energy transfer. (B) Red full line, calculated including the energy transfer; blue line, energy transfer contribution. In both panels, the red dashed line is the nondecaying contribution to the photoproduct signal calculated according to eq A1 of the Appendix.
vibrational relaxation, we estimate that the energy transfer effect comprises only a small fraction of the photoproduct signal. In addition, the energy transfer cannot be effective for time delays significantly longer than the (sub-picosecond) lifetime of the OD stretching modes. Thus, only the build-up kinetics is affected (dotted segment in the inset of Figure 4). This figure shows (black line) that the formation of the photoproduct is followed by a partial decay, after which the signal remains constant. The time scale and the time constant of the decay observed after ∼2 ps are inconsistent with the much faster vibrational relaxation of the OD stretching vibration. It may be attributed to the partial recovery of the coordination between polar components of the lipid and favorably distributed D2O molecules. As shown by Steinel et al.,45 the kinetics of the process leading to the formation of the photoproduct can be described by three coupled rate equations which involve the initially excited molecules (Ne), the relaxed ones (Ng), and the photoproduct (Np):
dNe(t) ) -k1Ne(t) dt
(1a)
dNg(t) ) k1fNe(t) - k2Ng(t) dt
(1b)
dNp(t) ) k2Ng(t) - k3Np(t) dt
(1c)
Here, k1 is the relaxation rate of the D2O stretching vibration, k2 is the rate of D-bond breaking, and k3 is the decay rate of the photoproduct; the f factor in eq 1b represents the fraction of the initially excited molecules that relax to Ng. The first two equations are identical to those used in ref 45; in eq 1c, the last term on the right has been added to describe the decay of the photoproduct signal shown in Figure 4. The solutions of the coupled equations for Np, with and without the last term in eq 1c, can be easily obtained by substitution: they are summarized in the Appendix. The bimodal time dependence of the photoproduct signal at delays longer than 2 ps is interpreted as the superposition of a quasi-constant term and of a decaying contribution. The former corresponds in our kinetic model to the solution given in ref 45 (eq A1 in the Appendix), while the latter is described by eq A5. The optimal values of the kinetic parameters appearing in those expressions can then be obtained by fitting the experimental curve of Figure 4. In the fitting procedure, we fixed k1, the inverse lifetime of the D2O stretching vibration, to the average value measured experimentally in the red part of the
dNt(t) ) ktsNe(t) - k1Nt(t) dt
(2)
where Nt are the excited molecules produced, with rate kt, by energy transfer from a fraction, s, of the initially excited Ne molecules; these excited molecules decay with the same time constant, 1/k1, already introduced. The solution for Nt has the form of eq A3 in the Appendix; it can then be summed with an appropriate weight to eqs A1 and A5 and used to fit the experimental curve. Figure 6B shows that the final agreement is comparable with that obtained without including the energy transfer contribution (Figure 6A). The corresponding best fitting parameters are the following: 1/k1 ) 600 fs, 1/k2 ) 800 fs, and 1/k3 ) 3.6 ps. The energy transfer time is 1/kt ) 600 fs. Not surprisingly, the only relevant difference between the two models concerns the value of the D-bond breaking time 1/k2: the presence of energy transfer is an obstacle to the accurate determination of the photoproduct rise time (largely dependent on k2), but it does not affect significantly the rest of the kinetic curve. Since the relative weight of the energy transfer contribution cannot be determined experimentally, we can only estimate that the deuterium-bond breaking time for water in membranes ranges between 450 and 800 fs. The effect of energy transfer is undoubtedly that of making the rise of the photoproduct signal apparently faster: the true value of the D-bond breaking time is then expected to be close to the upper limit. We recall that 800 fs is the value measured for bulk water by Fayer’s group.45 The overall behavior in the picosecond time scale (see Figure 4) differs from that observed in reverse micelles;33-37 the most striking peculiarity, however, concerns the shape of the photoproduct spectrum. As shown in Figure 3, the transient spectra (A) are very different from the temperature difference spectrum (B) of the DMPC membrane measured at 300 and 302 K. The photoproduct spectrum is essentially unchanged in the entire 12 ps range; only a very limited broadening is observed (for instance, if one compares the spectra at 1.8 and 10 ps in Figure 3A). In the case of OH in bulk D2O,45 the photoproduct spectrum and the temperature difference spectrum are instead practically identical (only a small frequency shift is detected at very short delay times). In that case, in fact, the energy deposited in the OH vibration rapidly dissipates into the H-bonded network, causing the partial disruption of the network itself. The same effect is produced in bulk water by a temperature increase. Our data then show that the process of energy dissipation from the initially excited water molecules to the surrounding moieties of the bilayer is much slower. The perturbed local structures produced by the optical excitations have little relation with the changes produced in the membrane by a temperature increase, which very likely involve local structural rearrangements, including the slow motions of the phospholipid units. The heterogeneity of water in terms of the variance of bonded and unbonded forms at the bilayer interface is reflected directly in the distribution of photoproduct-free vibrational lifetimes reported in Figure 4A and in Table 1. It is well-known that the cooperative effects due to the extended hydrogen-bond network
1382 J. Phys. Chem. B, Vol. 111, No. 6, 2007 play an important role in determining the extremely short vibrational lifetime (190 fs) of the stretching vibration in pure H2O.24 Until now, no data have been published on the corresponding lifetime in neat D2O: preliminary results obtained in our laboratory give an approximate value of 350 ( 50 fs (consistent with the lower anharmonicity of the deuterated compound and with the less complete overlap of the stretching band with the overtone of the D2O bending mode). The peculiar organization of the water molecules in semidry membranes is very different from that of the networked system, and correspondingly, the vibrational lifetime increases. As shown in Table 1, the lifetime measured at the maximum of the FTIR absorption is 620 fs for the sample with a phospholipid/water ratio of 1:2. The measured decay time is clearly frequency dependent: it decreases at higher probe frequencies. This behavior is the opposite of that observed experimentally48 and theoretically predicted49 for OH impurity in neat D2O. The peculiar frequency dependence observed in the present case is to be attributed to the intramolecular energy transfer from the high frequency states toward the center of the band, which occurs on a sub-picosecond time scale (see above). The blue spectral region corresponds to dangling OD bonds of single water molecules bonded to DMPC and, possibly, of small aqueous clusters. In both cases, the relatively high mobility allows fast frequency modulation that favors the intramolecular energy transfer to the OD oscillator involved in a strong D-bond and absorbing at lower frequency. This energy pathway is not available for OH in HOD molecules. The increase of the measured lifetime with decreasing degree of hydration is consistent with this picture. A large variance across the bandwidth of the OD stretching is observed also for the decay of the spectral anisotropy. The most striking feature is the constant plateau obtained for r(τ) in all three cases (Figure 5B). When the anisotropy is probed at 2496 cm-1, no decay is observed, and r maintains its maximum theoretical value of 0.4 in the entire time window considered. The value of the constant anisotropy diminishes with increasing probe frequency, and at the same time, a very fast initial decay appears. Constant anisotropy means that for a fraction of the water molecules in the membrane (the totality for those absorbing at the 2496 cm-1 probe frequency) rotational diffusion and resonant energy transfer mechanisms are not in effect in the considered time window. On the other hand, the partial very fast initial decay (definitely shorter than 500 fs) observed at 2572 and 2630 cm-1 is not compatible with the much slower orientational dynamics of water molecules (τOR ) 2.6 ps in bulk45); it can be attributed to ultrafast inertial dynamics and to resonant energy transfer involving the blueabsorbing OD dangling bonds and the weakly bonded water molecules. We notice here that the results obtained for the anisotropy of the less hydrated samples (1:1 and 2:1 phospholipid/water ratios, data not shown) are qualitatively similar to those of the 1:2 sample and that, most relevantly, the fast decaying contribution decreases in comparison to the constant term with the decreasing amount of water in the sample. This confirms that a higher degree of hydration corresponds to a larger fractional component of D2O molecules forming just one D-bond with DMPC and of water molecules involved in small clusters. Here, we recall that ultrafast energy transfer determines the very fast anisotropy decay in the hydrogen-bonded network of neat water. The overall experimental evidence indicates that the pathway of the relaxation for water in membrane is dependent on where the excitation energy is invested, or which aqueous structural
Volkov et al. motifs are excited. In other words, the structural heterogeneity of water rules the pathway of energy relaxation. Conclusions The time evolution of the infrared spectra of heavy water in DMPC phospholipid membrane after resonant excitation of the OD stretching vibration was measured by using a two-color pump-probe setup. Several important results were obtained: (a) The vibrational lifetime is comparable to or longer than that in bulk D2O and is frequency dependent, as it decreases with increasing probe frequency; this effect is a consequence of the fast intramolecular energy transfer characterizing the vibrational states of higher frequency. A dependence on the degree of hydration is also observed: the lifetime increases when the water content of the sample is lowered. The variety of lifetimes measured in our experiment range between 903 and 390 fs. (b) The formation of the photoproduct is completed in about 2 ps; its signal shows a partial decay (3.4-3.6 ps time constant) to a plateau which remains constant in the entire time window considered. The spectral profile differs substantially from the difference spectrum obtained by subtracting the FTIR spectra at two temperatures. This observation indicates that, in contrast to the results in bulk water, the full thermalization of the energy deposited by the excitation pulse happens on a time scale much longer than the time range of our experiment. (c) The time evolution of the spectral anisotropy is also frequency dependent. At the lowest probe frequency considered, it is perfectly constant for the entire 12 ps range. At higher probe frequencies, the amplitude of the constant term diminishes and a very fast subpicosecond initial decay appears. The constant anisotropy is due to water molecules forming two strong D-bonds with the polar groups of the bilayer, whose rotational dynamics is practically inhibited, and to the local OD oscillators engaged in a single D-bond with a DMPC molecule. Clearly, their contribution decreases as the probe frequency moves to the blue. The fast partial anisotropy decay observed for the blue-absorbing oscillators is attributed to the resonant energy transfer and/or to ultrafast inertial relaxation which become possible for the OD dangling bonds and for weakly bonded water molecules. The comparison with the results of molecular dynamics simulations confirms the conclusions outlined above and allows us to propose a picture for the structure and dynamics of water in weakly hydrated phospholipid membranes. Furthermore, we can state that water in such an environment loses most of its bulk properties and shows a high degree of structural heterogeneity. In spite of some apparent similarities, the situation is very different from that of water segregated in reverse micelles, where the population of water molecules can be distinguished in a fraction belonging to the shell next to the micelle surface and in a fraction constituting the bulk core.36 In weakly hydrated membranes, the majority of the water molecules are engaged in strong and persistent bonds with the phospholipid polar groups which largely determine their properties. The resulting main characteristic feature is the extreme rigidity of the local structures involving hydrogen-bonded water and phospholipid molecules. This property is reflected in the dramatic slowing down of the rotational dynamics and in the reduction of the energy transfer rate. The implications of our results for the role of the membrane in the physics of living cells are straightforward: cavitated and functionally engaged water in the interface of a native envelope is an important factor in the interface structural stabilization. In addition, the interface bound water does not represent a good medium for energy transport;46 this phenomenon may play an important role in the stabilization of the enzymatic cellular activity.
Heterogeneity of Water
J. Phys. Chem. B, Vol. 111, No. 6, 2007 1383
Acknowledgment. This work has been supported by the European Union Marie Curie program (contract MTKDCT2004-509761). Financial support from EU contract RII3-CT2003-506350 and from the Italian Ministry of University (MIUR) is also acknowledged. Appendix The solution for Np of the system of coupled rate eqs 1a-c without the last term in eq 1c was given in ref 45:
Np(t) )
fNe°
{k1[1 - exp(-k2t)] + k2[-1 +
(k1 - k2)
exp(-k1t)]} (A1) When the term proportional to k3 is included, the coupled rate equations
dNe(t) ) -k1Ne(t) dt
(A2a)
dNg(t) ) k1fNe(t) - k2Ng(t) dt
(A2b)
dNp(t) ) k2Ng(t) - k3Np(t) dt
(A2c)
are equally solved by substituting Ne ) Ne° exp(-k1t) into eq A2b, yielding
k1 [exp(-k1t) - exp(-k2t)] Ng(t) ) fNe° k 2 - k1
(A3)
Substituting eq A3 into eq A2c, we obtain the linear differential equation
{
}
dNp(t) k1 ) k2 fNe° [exp(-k1t) - exp(-k2t)] - k3Np(t) dt k2 - k1 (A4) whose solution is
[(
)
k1k2 1 1 exp(-k3t) + Np(t) ) fNe° k2 - k1 k3 - k2 k3 - k1 1 1 exp(-k1t) exp(-k2t) (A5) k3 - k1 k3 - k2
]
References and Notes (1) Rand, R. P.; Parsegian, V. A. Biochim. Biophys. Acta 1989, 988, 351. (2) Rand, R. P.; Fuller, N. L.; Gruner, S. M.; Parsegian, V. A. Biochemistry 1990, 29, 76. (3) Cojen, H.; Slater, S. J.; Stubbs, C. D. Biochemistry 1995, 34, 6188. (4) Fitter, J.; Lechner, R. E.; Dencher, N. A. J. Phys. Chem. B 1999, 103, 8036. (5) Sotomayor, C. P.; Aguilar, L. F.; Cuevas, F. J.; Helms, M. K.; Jameson, D. M. Biochemistry 2000, 39, 10928. (6) Poolman, B.; Spitzer, J. J.; Wood , J. M. Biochim. Biophys. Acta 2004, 88, 1666. (7) Jacobs, R. E.; White, S. H. Biochemistry 1989, 28, 3421. (8) Yeagle, P. L.; Martin, R. B. Biochem. Biophys. Res. Commun. 1976, 69, 775. (9) Schmidt, C. F.; Barenholz, Y.; Huang, C.; Thompson, T. E. Biochemistry 1977, 16, 3948. (10) Pope, J. M.; Comell, B. A. Chem. Phys. Lipids 1979, 24, 27. (11) Smaby, J. M.; Hermetter, A.; Schmid, P. C.; Paltauf, F.; Brockman, H. L. Biochemistry 1983, 22, 5808. (12) Zhou, Z.; Sayer, B. G.; Hughes, D. W.; Stark, R. E.; Epand, R. M. Biophys. J. 1999, 76, 387.
(13) Graener, G.; Seifert, H.; Laubereau, A. Phys. ReV. Lett. 1991, 66, 2092. (14) Woutersen, S.; Bakker, H. J. Nature 1999, 402, 507. (15) Gale, G. M.; Gallot, G.; Hache, F.; Lascoux, N.; Bratos, S.; Leicknam, J-Cl. Phys. ReV. Lett. 1999, 82, 1068. (16) Stenger, J.; Madsen, D.; Hamm, P.; Nibbering, E. T. J.; Elsaesser, T. Phys. ReV. Lett. 2001, 87, 027401. (17) Crupi, V.; Majolino, D.; Migliardo, P.; Venuti, V. J. Phys. Chem. B 2002, 106, 10884. (18) Lawrence, C. P.; Skinner, J. L. J. Chem. Phys. 2002, 117, 8847. (19) Fecko, C. J.; Eaves, J. D.; Loparo, J. J.; Tokmakoff, A.; Geissler, P. L. Science 2003, 301, 1698. (20) Torre, R.; Bartolini, P.; Righini, R. Nature 2004, 428, 296. (21) Moller, K. B.; Rey, R.; Hynes, J. T. J. Phys. Chem. A 2004, 108, 1275. (22) Asbury, J. B.; Steinel, T.; Kwak, K.; Corcelli, S. A.; Lawrence, C. P.; Skinner, J. L.; Fayer, M. D. J. Chem. Phys. 2004, 121, 12431. (23) Cringus, D.; Yeremenko, S.; Pshenichnikov, M. S.; Wiersma, D. A. J. Phys. Chem. B 2004, 108, 10376 (24) Cowan, M. L.; Bruner, B. D.; Huse, N.; Dwyer, J. R.; Chugh, B.; Nibbering, E. T. J.; Elsaesser, T.; Miller, R. J. D. Nature 2005, 434, 199. (25) Odelius, M.; Ogasawara, H.; Nordlund, D.; Fuchs, O.; Weinhardt, L.; Maier, F.; Umbach, E.; Heske, C.; Zubavichus, Y.; Grunze, M.; Denlinger, J. D.; Pettersson, L. G. M.; Nilsson, A. Phys. ReV. Lett. 2005, 94, 227401. (26) Polnazek, C. F.; Bryant, R. G. J. Chem. Phys. 1984, 81, 4038. (27) Bellissent-Funel, M. C.; Chen, S. H.; Zanotti, J. M. Phys. ReV. E 1995, 51, 4558. (28) Scodinu, A.; Fourkas, J. T. J. Phys. Chem. B 2002, 106, 10292. (29) Floquet, N.; Coulomb, J. P.; Dufau, N.; Andre, G. J. Phys. Chem. B 2004, 108, 13107. (30) Lee, Y.; Martin, C. D.; Parise, J. B.; Hriljac, J. A.; Vogt, T. Nano Lett. 2004, 4, 619. (31) Liu, Y.; Wang, Q.; Lu, L. Langmuir 2004, 20, 6921. (32) Ruan, C.; Lobastov, V. A.; Vigliotti, F.; Chen, S.; Zewail, A. H. Science 2004, 304, 80. (33) Tan, H.; Piletic, I. R.; Riter, R. E.; Levinger, N. E.; Fayer, M. D. Phys. ReV. Lett. 2005, 94, 057405-1. (34) Tan, H.; Piletic, I. R.; Fayer, M. D. J. Chem. Phys. 2005, 122, 174501. (35) Piletic, I. R.; Tan, H.; Fayer, M. D. J. Phys. Chem. B 2005, 109, 21273. (36) Piletic, I. R.; Moilanen, D. E.; Spry, D. B.; Levinger, N. E.; Fayer, M. D. J. Phys. Chem. A 2006, 110, 4985. (37) Dokter, A. M.; Woutersen, S.r; Bakker, H. J. Phys. ReV. Lett. 2005, 94, 178301. (38) Pasenkiewicz-Gierula, M.; Takaoka, Y.; Miyagawa, H.; Kitamura, K.; Kusumi, A. J. Phys. Chem. A 1997, 101, 3677. (39) Takaoka, Y.; Pasenkiewicz-Gierula, M.; Miyagawa, H.; Kitamura, K.; Tamura, Y.; Kusumi, A. Biophys. J. 2000, 79, 3118. (40) Pasenkiewicz-Gierula, M.; Takaoka, Y.; Miyagawa, H.; Kitamura, K.; Kusumi, A. Biophys. J. 1999, 76, 1228. (41) Lopez, C. F.; Nielsen, S. O.; Klein, M. L.; Moore, P. B. J. Phys. Chem. B 2004, 108, 6603. (42) Chanda, J.; Chakraborty S.; Bandyopadhyay, S. J. Phys. Chem. B 2006, 110, 3791. (43) Hamm, P.; Kaindl, R. A.; Stenger, J. Opt. Lett. 2000, 25, 1798. (44) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb, M. A.; Cheeseman, J. R.; Montgomery, J. A., Jr.; Vreven, T.; Kudin, K. N.; Burant, J. C.; Millam, J. M.; Iyengar, S. S.; Tomasi, J.; Barone, V.; Mennucci, B.; Cossi, M.; Scalmani, G.; Rega, N.; Petersson, G. A.; Nakatsuji, H.; Hada, M.; Ehara, M.; Toyota, K.; Fukuda, R.; Hasegawa, J.; Ishida, M.; Nakajima, T.; Honda, Y.; Kitao, O.; Nakai, H.; Klene, M.; Li, X.; Knox, J. E.; Hratchian, H. P.; Cross, J. B.; Bakken, V.; Adamo, C.; Jaramillo, J.; Gomperts, R.; Stratmann, R. E.; Yazyev, O.; Austin, A. J.; Cammi, R.; Pomelli, C.; Ochterski, J. W.; Ayala, P. Y.; Morokuma, K.; Voth, G. A.; Salvador, P.; Dannenberg, J. J.; Zakrzewski, V. G.; Dapprich, S.; Daniels, A. D.; Strain, M. C.; Farkas, O.; Malick, D. K.; Rabuck, A. D.; Raghavachari, K.; Foresman, J. B.; Ortiz, J. V.; Cui, Q.; Baboul, A. G.; Clifford, S.; Cioslowski, J.; Stefanov, B. B.; Liu, G.; Liashenko, A.; Piskorz, P.; Komaromi, I.; Martin, R. L.; Fox, D. J.; Keith, T.; Al-Laham, M. A.; Peng, C. Y.; Nanayakkara, A.; Challacombe, M.; Gill, P. M. W.; Johnson, B.; Chen, W.; Wong, M. W.; Gonzalez, C.; Pople, J. A. Gaussian 03, revision C.02; Gaussian, Inc.: Wallingford, CT, 2004. (45) Steinel, T.; Asbury, J. B.; Zheng, J.; Fayer M. D. J. Phys. Chem. A 2004, 108, 10957. (46) Volkov, V. V.; Nuti, F.; Takaoka, Y.; Chelli, R.; Papini, A. M.; Righini, R. J. Am. Chem. Soc. 2006, 128, 9466. (47) Gilijamse, J. J.; Lock, A. J.; Bakker, H. J. Proc. Natl. Acad. Sci. U.S.A. 2005, 102, 3202. (48) Gale, G. M.; Gallot, G.; Lascoux, N. Chem. Phys. Lett. 1999, 311, 123. (49) Lawrence, C. P.; Skinner, J. L. J. Chem. Phys. 2003, 119, 3840.
