INTERFACIAL WATER AT SYNTHETIC AND NATURAL LIPID BILAYERS PROBED BY VIBRATIONAL SUM-FREQUENCY GENERATION SPECTROSCOPY

• Users may download and print one copy of any publication from the public portal for the purpose of private study or research. • You may not further distribute the material or use it for any profit-making activity or commercial gain • You may freely distribute the URL identifying the publication in the public portal ? Take down policy If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately and investigate your claim.

Biological interfaces and water make an indivisible pair, where the physicochemical properties and biological processes of biointerfaces are intimately linked to the structure and the behaviour of water. Among biological interfaces, the cell membrane is an ultimate interfacial system, because it is that two-dimensional region making the barrier between the inner and the outer cell environments. Many physiological processes occurring within the membrane barrier are carried and triggered by water, such as protons and ions transport, metabolites adsorption and translocation, or proteins folding, for instance. [1,2] This is why the properties and the behaviour of this small and ubiquitous molecule have attracted the interest of researchers from different disciplines, from biology and medicine, to chemistry and physics. A crucial point is that water close to biological interfaces is far from behaving as bulk disordered water, while it owns specific order and organization. [3,4] Indeed, upon an electrical potential, such as the one generated by charged interfaces, water dipoles orient along preferential directions, thus forming an organized layer of water, as long as the electric field associated to the electrical potential is non-vanishing. The degree of organization and the thickness over which the organization take place will depend on the physicochemical properties of the interface, i.e., net molecular charge, charge density per surface area, molecular orientation, structure, and hydrophilicity, as well as pH and ionic strength of the liquid medium. [5] Whatever the process that would modify the interfacial properties, this latter may affect the organisation of such a water layer. The structure, the order and the behaviour of interfacial water close to membrane interfaces have been extensively modelled with theoretical approaches, while its direct measure remained … long-time unavailable. This limitation has been definitely encompassed with the advent of vibrational sum frequency generation (SFG) spectroscopy. Indeed, being based on a second order nonlinear optical (NLO) phenomenon, which is possible only in non-centrosymmetric environments -like anisotropic bulk materials or interfacial systems -, [6,7] SFG spectroscopy is ideally able to probe an organized layer of water, because this latter owns a noncentrosymmetric structure, contrary to bulk water. Moreover, SFG spectroscopy probes molecular vibrations, so that it provides the chemical fingerprint of interfacial water. These skills made SFG spectroscopy a favorite technique to probe water close to membrane systems. [8][9][10][11][12][13][14][15][16][17][18][19][20][21][22] The investigation of the SFG response around membrane models demonstrated that the surrounding structure of water strongly depends on the physicochemical properties of the interface, and that is tuned by the interactions and the processes occurring within these biological systems.
Here, we will probe the vibrational NLO response of interfacial water at different models of lipid membranes, made of synthetic or natural lipids, supported on CaF2. Indeed, the average charge properties of the interfaces trigger the signal of interfacial water accordingly. The goal is testing the potentiality of using the SFG response of interfacial water to identify the underlying lipid bilayer, and then explore new solutions to make SFG spectroscopy a biodiagnostic tool.

MATERIALS AND METHODS
Lipid bilayers. We prepared lipid bilayers adsorbed on CaF2 prisms from different lipids Figure  1), namely1,2-dihexanoyl-sn-glycero-3-phosphoethanolamine (6:0 PE or DHPE), 1,2dipalmitoyl-sn-glyero-3-phosphoethanolamine (16:0 PE or DPPE), 1,2-dioleoyl-sn-glycero-3phosphoethanolamine (18:1 Δ9-Cis PE or DOPE), 1,2-dioleoyl-sn-glycero-3-phospho-L-serine (18:1 PS or DOPS), 1,2-dioleoyl-sn-glycero-3-[phospho-rac-(3-lysyl(1-glycerol))] (18:1 Lysyl PG), E. Coli Cardiolipin (E. Coli CA), E. Coli Total Lipid Extract (ETLE) and Yeast Total Lipid Extract (YTLE). All lipid molecules were purchased from Avanti ® Polar Lipids, Inc. (Alabaster, Alabama, USA). SSLBs were obtained through the method of the spontaneous fusion of lipid vesicles on CaF2 (Crystran, UK). CaF2 prisms were cleaned by immersion in a diluted piranha solution (H2SO4:H2O2, 2:1 diluted 10 times) for 1 minute and then rinsed thoroughly with Milli-Q water (18.2 MΩ.cm, pH = 5.5). Lipid vesicles were prepared by i) solubilizing 2 mg of lipid powder in 300 µL of chloroform (CHCl3), ii) evaporating CHCl3 under N2 flow, and iii) adding 1 mL of Milli-Q water. The obtained solutions were sonicated for 10 min and then centrifuged for 10 min at 6000 rpm. The vesicle solutions were injected into a Teflon cell containing 1 mL of milliQ water (18.2 MΩ cm -1 ) and holding the CaF2 substrate, so that the vesicles concentration during the fusion process at the surface were half of the native concentration. After 3 hours of contact with the prism, SSLBs were expected to be formed. [21,23,24] The remaining vesicles in solution were fully removed by exchanging the solution with Milli-Q water. This procedure was carried out by adding 0.5 mL of Milli-Q water and then removing 0.5 mL of the diluted solution. These steps were repeated at least 20 times in order to leave the SSLBs in contact with an almost pure Milli-Q water solution. SFG setup. The SFG response of water at the interface was recorded with a homemade SFG spectrometer. Two pulsed incident beams (15 ps, 25 Hz), namely, an IR beam tuned between 3600-2800 cm -1 (average power = 25 mW) and a visible (Vis) beam fixed at 532 nm (or 18796.99 cm -1 , average power = 10 mW), were superimposed spatially and temporally at the CaF2/water interface in a total internal reflection geometry through the prism ( Figure 2). An SFG beam was generated with a frequency equal to the sum of the two incident beam frequencies. All measurements were performed at room temperature (22.5°C). To perform comparisons between samples, the SFG responses were normalized by the IR and Vis beams, and by the SFG response of the neat CaF2/water interface, which was measured before each lipid bilayer adsorption. All experiments were repeated at least twice. … Figure 2. Schematic representation of a SFG measurement at the CaF2/lipid/water interface in a Total Internal Reflection (TIR) geometry of the incident IR and Vis beams and of the generated SFG beam.
Theoretical background. Vibrational SFG spectroscopy is based on a 2 nd order NLO process, in which the the three photon mixing process enables probing the vibrational response of chemical groups in a nonlinear regime. In a SFG process (Figure 2), one tunable infrared ( ) and one fixed visible ( ) beams are focussed at the interface, where their interaction with the second order susceptibility ( (2) ) of matter generates a new beam at their sum frequency (SFG, = + ). The key selection rule of 2 nd order NLO processes is that the (2) susceptibility tensor, in the dipole approximation, is zero for centrosymmetric media, while it is non-zero where the inversion symmetry is broken, like at interfaces. It follows that SFG spectroscopy provides the vibrational signature of interfacial regions only, while it is insensitive to their surrounding bulk environments, either solid, liquid or gaseous. In addition to the interfacial sensitivity, the SFG response of a vibrational mode of the interface requires this latter is simultaneously IR and Raman active. Definitely, once the tunable beam matches an SFG active vibrational mode of the interface, then there is an enhancement of the SFG response, accordingly to the following relations: where and are the intensities of the incident visible and infrared beams, (2) is the nonresonant part of the effective second order susceptibility that can eventually come from the electronic properties of the close bulk environment. The (2) resonant contribution is related to the active molecular vibrations as follows: , where and are the frequencies of the ℎ molecular mode and of the incident infrared beam, respectively, Γ is the damping factor, and , is the oscillator strength of the ℎ molecular mode, which is the macroscopic response individual oscillators for which the molecular second order hyperpolarisability (2) is linked to the variations of the dipole moment and of the polarizability. rotates the molecular coordinates into the interface coordinates , and the braket indicates the average over the molecular orientation distribution and and are the variation of the dipole moment (IR activity) and of the polarizability (Raman activity), respectively, of the ℎ vibrational mode. [25,26] /// 33 In summary, the strict selection rulesthat is (2) ≠ 0, ≠ 0 and ≠ 0 −, combined to the 2 dependence of the SFG intensity and to the coherent nature of the SFG process, make SFG spectroscopy ideally able to probe a thin interfacial layer of water molecules close to an interface, with a chemical selectivity.

RESULTS AND DISCUTION
The SFG spectrum of the neat CaF2/water interface (Figure 3, top) was systematically recorded before adsorption of lipid vesicles. The broad band centered at 3125 cm -1 is due to OH stretching vibrations of interfacial water. [20][21][22][23] Indeed, the surface of CaF2 is known to own a slightly positive charge (its zeta potential in pure water is about +20 mV), which drives the organization of a thin layer of water, possibly with the H atoms pointing out from the surface (as shown in the simplified picture in the bottom of Figure 3, left part). [27][28][29] The corresponding maximum SFG intensity was about 0.22 arbitrary units (arb. units) ( Table 1). This value will represent the reference signal to which all following CaF2/lipid/water interfaces will refer. The adsorption and fusion of lipid vesicles at the CaF2/water interface was monitored in real time, by recording the SFG intensity at 3125 cm -1 for 3 hours of adsorption. Then, the remaining lipid vesicles in solution were removed, and the SFG maximum intensity of the CaF2/lipid/water interfaces was measured. Negatively PS and positively Lysyl PG charged lipids. DOPS and Lysyl PG are negative and positive charged lipids, respectively, with one net charge per molecule. DOPS holds two negative and one positive charges, while DOPG holds one negative and two positive charges, which are localized on the polar heads, respectively. Both lipids have the same aliphatic chains, with one unsaturated bond in each aliphatic chain, giving rise to a liquid-like conformation of the corresponding bilayers at room temperature. Starting from the intensity of the CaF2/water signal (0.22 arb.units), once DOPS vesicles were injected in the cell to adsorb and fuse on the CaF2 surface and form a lipid bilayer, the SFG intensity decreased to zero in a few seconds, then increased to 1.50 arb. units in the next minute, and remained stable at this value for the next 3h. The passing from zero of the SFG intensity correspond to the flip-flop of water molecules from the up to the down conformation (Figure 3, bottom). [12,31] Indeed, DOPS vesicles carried their negative charge close to the CaF2 surface, and screened their positive charge. The further saturation of the surface by DOPS vesicles led to a strongly negatively charged interface, so that the SFG signal increased back, which correspond to an organisation of the water dipoles close to the lipid interface. A different behaviour was observed during the adsorption of Lysyl PG vesicles at the CaF2/water interface. Here, the SFG signal showed a fast increase of the SFG signal to 0.60 arb.units directly, without passing from zero, in agreement with the positive net charge carried by the lipids polar heads, which is of the same sign of the neat CaF2 surface. Definitely, at the DOPS/water interface, water dipoles are expected to be oriented with a down conformation (Figure 3 bottom right), while for Lysyl PG/water interface they are expected to be predominantly oriented with an up conformation (Figure 3, bottom left).

Natural Lipid Layers.
Going from the easiest model membranes made of mono-component layers of synthetic lipids of known chemical composition towards more representative models of real systems requires facing that the full composition of the lipid layers is unknown. E.coli Cardiolipin. E. Coli CA is a mixture of different CA structures, all holding the CA polar head, carrying two net negative charges, and different hydrophobic chains, which are in the amount of four per molecule. Starting from a SFG intensity of 0.22 V at the neat CaF2/water interface, the injection of E. coli CA vesicles in the liquid cell decreased the SFG intensity to zero in less than 10 minutes, and then this latter increased back and reached a value as high as 3.75 arb. units after 3h. This behavior indicates that E. coli vesicles screened the initially positive charge of CaF2, and then the corresponding lipid bilayer provided a strongly negative character to the interface, which drove a higher organization of interfacial water. This is possibly due to the structure of CA compounds, which, due to the presence of four hydrophobic chains per molecule owing strong intermolecular interactions, give rise to densely packed structures, leading to a higher density of charges per surface area. Moreover, so densely packed films led to highly organized structures, where the polar heads are predominantly oriented in a preferential direction, which may favor a large alignment of water close to the lipid interface. Total Lipid Extracts: ETLE and YTLE. E. coli Total Lipid Extract is composed of different neutral and charged lipids. The known fraction of ETLE (as provided by the supplier Avanti Polar Lipids), which account for the 82.4% of the total composition, exhibits an average charge of the lipid composition that is negatively charged, while the remaining 17.6% fraction was unknown. The injection of ETLE vesicles at CaF2/water interface induced a decreased of the SFG intensity to zero, and then an increase up to 0.25 arb. units in only one minute. This behavior is in agreement with a very fast screening of the positive charge of CaF2 by ETLE vesicles, which inverted the sign the interfacial charges from positive to negative. This statement enabled unravelling that an inversion of the orientation of the water dipoles occurred, which testified for an average negatively charged character of the ETLE lipid interface. Similarly, the exact charge of YTLE was also unknown (see the YTLE composition as provided by Avanti Polar Lipids). The injection of YTLE vesicles decreased the initial SFG intensity to zero, and then this latter increased back up to 0.31 arb.units in the next 3 hours. This behaviour suggests an average negative charge of the YTLE lipid layer, which triggered an inversion of orientation of the water dipoles from the CaF2/water interface to the CaF2/YTLE/water interface.

CONCLUSION
The SFG response of water was measured at neutral, negative and positive mono-component lipid bilayers, respectively, and showed that the highest signal, corresponding to a high degree of organisation, was reached at the DOPS negative lipid interface. At natural lipid layers interfaces, the SFG response of E. Coli Cardiolipin layers indicated an impressive organization of water close to the interface, since the SFG signal increased by two times larger than at the negative mono-component lipid interface. These behaviours enabled estimating the average charge properties of two natural lipids compounds, namely E. Coli total lipid extract and yeast total lipids extracts, whose full composition was unknow. The SFG response during the adsorption of the corresponding vesicles onto the surface demonstrated that these lipid layer interfaces were in average negatively charged, with a sligthly higher organisation of water at the YTLE than at the ETLE interface. By using synthetic lipid layers of known composition as reference signals, it has been then possible to unravel the average charge properties of natural lipid interfaces. The possibility to distinguish lipid interfaces by the signal of their organized water environment represents a possible solution to be explored for developing new, innovative and label-free biodevices.

ACKNOWLEDGMENT
The author thanks the National Belgian Fund for the Scientific Research F.R.S.-FNRS, the NATO SPS Program (under the grant number G5292 "Biohazards) for financial support, and the "Lasers, Optics and Spectroscopies" technological platform (LOS) of UNamur for instrumental and technical support.