Letter pubs.acs.org/JPCL
Cooperative Hydrogen-Bond Dynamics at a Zwitterionic Lipid/Water Interface Revealed by 2D HD-VSFG Spectroscopy Ken-ichi Inoue,† Prashant C. Singh,†,# Satoshi Nihonyanagi,†,‡ Shoichi Yamaguchi,†,§ and Tahei Tahara*,†,‡ †
Molecular Spectroscopy Laboratory, RIKEN, 2-1 Hirosawa, Wako, Saitama 351-1098, Japan Ultrafast Spectroscopy Research Team, RIKEN Center for Advanced Photonics (RAP), 2-1 Hirosawa, Wako, Saitama 351-1098, Japan § Department of Applied Chemistry, Graduate School of Science and Engineering, Saitama University, 255 Shimo-Okubo, Sakura, Saitama 338-8570, Japan ‡
S Supporting Information *
ABSTRACT: Molecular-level elucidation of hydration at biological membrane interfaces is of great importance for understanding biological processes. We studied ultrafast hydrogen-bond dynamics at a zwitterionic phosphatidylcholine/water interface by two-dimensional heterodyne-detected vibrational sum frequency generation (2D HDVSFG) spectroscopy. The obtained 2D spectra confirm that the anionic phosphate and cationic choline sites are individually hydrated at the interface. Furthermore, the data show that the dynamics of water at the zwitterionic lipid interface is not a simple sum of the dynamics of the water species that hydrate to the separate phosphate and choline. The center line slope (CLS) analysis of the 2D spectra reveals that ultrafast hydrogenbond fluctuation is not significantly suppressed around the phosphate at the zwitterionic lipid interface, which makes the hydrogen-bond dynamics look similar to that of the bulk water. The present study indicates that the hydrogen-bond dynamics at membrane interfaces is not determined only by the hydrogen bond to a specific site of the interface but is largely dependent on the water dynamics in the vicinity and other nearby moieties, through the hydrogen-bond network.
B
spectroscopy is the extension of HD-VSFG to 2D spectroscopy, and it has been applied to the study of aqueous interfaces recently.13−18 2D HD-VSFG spectra represent the pump frequency dependence of the IR pump-induced change of Im χ(2) (ΔIm χ(2)), which is an interfacial analogue of 2D IR spectra for bulk. In a previous study, we carried out 2D HD-VSFG experiments for charged lipid/water interfaces using a cationic lipid (1,2-dipalmitoyl-3-trimethylammonium propane; DPTAP) and an anionic lipid (1,2-dipalmitoyl-sn-glycero-3phosphorylglycerol; DPPG) (Figure 1a).17 Although the steady-state Im χ(2) spectra of the two interfaces show similar broad OH bands (except for their opposite signs due to the opposite net orientation of interfacial water molecules),10 2D HD-VSFG revealed that hydrogen-bond dynamics of interfacial water is markedly different: The interfacial water at the cationic DPTAP interface shows “bulk-like” ultrafast hydrogen-bond fluctuation, whereas corresponding dynamics is largely suppressed at the anionic DPPG interface.17 The distinct hydrogen-bond dynamics was explained in terms of the
iological membranes maintain appropriate environments inside cells. It is known that the functional activity of a membrane-bound protein and the membrane itself is highly dependent on the hydration structure at membrane/water interfaces.1,2 Therefore, molecular-level elucidation of the interfacial hydration is critically important. Furthermore, because the hydrogen bond is highly dynamic, clarification of its dynamics is crucial for understanding physicochemical properties of biological membranes. Vibrational spectra of the OH stretch region provide rich information about the structure and dynamics of water because they are very sensitive to the strength and fluctuation of the hydrogen bond.3 The steadystate and dynamic properties of water in various environments have been extensively studied with IR spectroscopy, particularly with time-resolved and two-dimensional (2D) IR spectroscopy recently.4−8 Heterodyne-detected vibrational sum frequency generation (HD-VSFG) spectroscopy provides interface-specific vibrational spectra, and it is an ideal tool for probing water at the membrane/water interfaces.9−12 HD-VSFG can provide spectra of the imaginary part of the second-order nonlinear susceptibility (Im χ(2)), which exhibits a simple absorptive band shape for the vibrational resonance, and hence, they are free from spectral distortion from which the |χ(2)|2 spectra obtained with conventional VSFG suffer. 2D HD-VSFG © XXXX American Chemical Society
Received: August 5, 2017 Accepted: October 4, 2017
5160
DOI: 10.1021/acs.jpclett.7b02057 J. Phys. Chem. Lett. 2017, 8, 5160−5165
Letter
The Journal of Physical Chemistry Letters
OH stretch region. Nevertheless, the signal in the 3400−3500 cm−1 region is noticeably weak compared to that of the DPPG/ water interface which only has an anionic phosphate in the headgroup. This spectral difference was attributed to the contribution of the negative signal due to coexisting H-down water around the cationic choline site.12 To clarify the hydrogen-bond dynamics at the zwitterionic lipid interface, we carried out time-resolved HD-VSFG experiments by exciting six different regions of the broad OH stretch band with femtosecond IR pump pulses. The set of obtained time-resolved ΔIm χ(2) spectra were represented as 2D spectra, in which the vertical axis represents the frequency of the IR pump pulse (ωpump) and the horizontal axis denotes the frequency of the IR probe pulse (ωIR) in the HD-VSFG measurement. The obtained 2D HD-VSFG spectra at the delay time from 0.0 to 0.7 ps are shown in Figure 2. The 2D
Figure 1. (a) Chemical structures of a zwitterionic lipid (DPPC), a cationic lipid (DPTAP), and an anionic lipid (DPPG). (b) Real (black) and imaginary (red) parts of the steady-state χ(2) spectrum of the DPPC/HOD-D2O interface.
difference in the hydrogen bond between water and the lipid head groups, thus demonstrating that the hydrogen-bond dynamics of the water at membrane interfaces is strongly dependent on the lipid headgroup. Zwitterionic phosphatidylcholine is a major constituent of biological membranes, and it is a neutral lipid that has both anionic phosphate and cationic choline in the headgroup (Figure 1a). The water structure of this zwitterionic lipid/water interface has been studied experimentally and theoretically.11,12,19−23 In particular, we studied this interface with HD-VSFG and obtained an Im χ(2) spectrum indicating the coexistence of major “H-up” oriented water and minor “Hdown” oriented water, which were attributed to the water molecules in the vicinity of the phosphate and choline in the headgroup, respectively.12 The positive OH band due to H-up water is much stronger and masks the weak negative OH band due to H-down water, which makes the OH band positive in total. Nevertheless, the OH band exhibits a dent that was attributed to the signature of the H-down oriented OH water. The presence of this minor H-down water around the choline site was reproduced in several MD simulations,21,23 but it was denied or not recognized in other MD simulations.19,20,22 The H-up water around the anionic phosphate has been reproduced in all of the MD simulations. In this study, we investigate the water dynamics at the zwitterionic lipid monolayer/water interface with 2D HDVSFG, which can provide new insight into the interfacial water from a dynamical viewpoint. In fact, in the 2D spectrum, the “hidden” H-down water unambiguously appears as a distinct bleach signal. Furthermore, the 2D spectra clearly show that the hydrogen-bond dynamics of the water around the phosphate at the zwitterionic lipid interface is much faster than that at the phospholipid DPPG interface. This indicates that the hydrogen-bond dynamics at the zwitterionic lipid/water is not the simple sum of the corresponding dynamics of water at the anionic and cationic lipid interfaces. Figure 1b shows the Im χ(2) spectrum of a zwitterionic lipid/ water interface. We use 1,2-dipalmitoyl-sn-glycero-3-phosphocholine (DPPC) as the lipid and isotopically diluted water (HOD-D2O) as water. Isotopically diluted water is used to suppress the intra/intermolecular vibrational couplings, which enables straightforward discussion about the water structure based on the Im χ(2) spectrum.24,25 The overall feature of the Im χ(2) spectrum is essentially the same as that reported previously,12 although the intensity at around 3600 cm−1 in the present spectrum is weaker due to the improved phase accuracy.26 The Im χ(2) spectrum exhibits a positive OH band peaked around 3300 cm−1, indicating that the H-up water around the anionic phosphate site dominates the signal in the
Figure 2. 2D HD-VSFG spectra of the DPPC/water interface from 0.0 to 0.7 ps. Vertical and horizontal axes are the frequencies of the IR pump (ωpump) and the IR probe of VSFG (ωIR), respectively. The color scale of ΔIm χ(2) is shown in the upper right. The steady-state Im χ(2) spectrum is shown at the top. White markers stand for the peak ωIR frequencies of the lobe B at each pump frequency. White solid lines are fits for the white markers and correspond to the center line slope (CLS).
spectrum at 0.0 ps exhibits a positive lobe in the low ωIR region (lobe A), a negative lobe in the middle ωIR region (lobe B), and a positive lobe in the high ωIR region (lobe C). As the time delay increases from 0.1 to 0.3 ps, the lobe B spreads over the OH stretch region, and the lobes A and C disappear. After 0.7 ps, the signal in the high ωIR region gradually becomes positive, which reflects the blue shift of the OH stretch band due to the temperature increase caused by the thermalization process.27 5161
DOI: 10.1021/acs.jpclett.7b02057 J. Phys. Chem. Lett. 2017, 8, 5160−5165
Letter
The Journal of Physical Chemistry Letters
stretch band of the H-up and H-down water are on the order of a few hundred femtoseconds even though the dynamics of the H-up water seems a little slower than that of the H-down water. In Figure 3, the temporal change of the CLS of lobe B is plotted
In the 2D spectrum at 0.0 ps, the positive lobe A and the negative lobe B are assignable to the hot band (ν = 1 → 2 transition) and the bleach of the fundamental transition (ν = 0 → 1) of the positive OH stretch band due to H-up water around the phosphate site. An important additional feature of this 2D spectrum is another positive lobe C in the high ωIR region. A positive lobe can appear as either the hot band of a positive band or the bleach of a negative band. However, the former possibility is readily ruled out because we do not observe any signals attributable to the accompanying bleach band: a hot band always appears with a bleach band appearing in the higher ωIR region. Thus, the lobe C is safely assignable to the bleach of a negative band in the steady-state spectrum. In other words, the observation of the lobe C is clear experimental evidence that the broad OH stretch band of the DPPC/HODD2O interface contains a band having the negative sign, in addition to the positive band due to the predominant H-up water around the phosphate. Therefore, the appearance of the positive lobe C strongly supports the existence of the negative OH band due to the H-down water around the choline,12 and the lobe C is attributable to the bleach of its fundamental transition (ν = 0 → 1). The corresponding hot band is expected to appear in the 2D HD-VSFG spectrum, but it is masked by the lobe B and hence is not separately observed. We note that the lobe C appears in the ωpump region between 3200 and 3500 cm−1, and it does not extend to the ωpump region below 3200 cm−1. This implies that the OH stretch frequency of the H-down water is higher than that of the H-up water around the phosphate and that its frequency distribution is narrower. The higher frequency of the H-down water agrees very well with its assignment to the water around the choline because the interaction of water with the choline is weaker than that with the phosphate.12 The 2D HD-VSFG spectrum observed at 0.0 ps strongly confirms that the anionic phosphate and cationic choline sites are individually hydrated by H-up and H-down water, respectively, at the zwitterionic lipid interface. Quantitative information about ultrafast hydrogen-bond dynamics at the interface is obtained by analysis of the center line slope (CLS),28 which is the slope of the line connecting the peaks of the horizontal cuts at each ωpump frequency. Previously, we reported that CLSs of the bleach lobe in the 2D spectra of the cationic DPTAP and anionic DPPG interfaces are significantly different, particularly at 0.0 ps, which gives initial CLS values of 0.29 ± 0.04 (DPTAP) and 0.80 ± 0.03 (DPPG), respectively.17 This difference was attributed to the difference in the contribution of the ultrafast hydrogen-bond fluctuation that occurs within the time resolution of the measurement (∼200 fs). It was concluded that the water at the DPTAP interface shows “bulk-like” ultrafast hydrogen-bond fluctuation that induces very fast spectral diffusion within 200 fs, giving the low CLS value of 0.29. On the other hand, such fluctuation is largely suppressed at the DPPG interface due to the strong hydrogen bond to the phosphoglycerol moiety, resulting in the large CLS value of 0.8. This conclusion is consistent with the time scale of the ultrafast fluctuation of the hydrogen bond, which has been reported as ∼60 fs in bulk HOD.29 For the zwitterionic lipid/water interface, the CLS analysis of lobes B and C can separately provide information about hydrogen-bond dynamics of water around the phosphate and choline, respectively. Lobe B in the 2D HD-VSFG spectrum at 0.0 ps is slightly tilted, while lobe C is almost vertical. This implies that the time scales of the spectral diffusion in the OH
Figure 3. Temporal change of the CLS of the lobe B (black), DPTAP (red), and DPPG (blue).
and compared with those of the bleach lobes of the OH stretch band of DPTAP and DPPG reported previously.17 As clearly seen, the initial CLS value of lobe B is much smaller than that of the bleach lobe of the DPPG interface although both are attributable to the bleach of the water around the phosphate. Rather, the initial CLS value of lobe B is close to that of the bleach lobe of the DPTAP interface where spectral diffusion is predominantly induced by ultrafast hydrogen-bond fluctuation. This result strongly indicates that at the zwitterionic lipid interface ultrafast hydrogen-bond fluctuation occurring on the