Chiral Water Superstructures Around Anti-parallel β-sheets Observed by Chiral Vibrational Sum Frequency Generation Spectroscopy

Abstract

Hydration modulates every aspect of protein structure and function. However, studying water structures in hydration shells remains challenging mostly due to overwhelming background from bulk water. We used vibrational sum frequency generation (SFG) spectroscopy to characterize hydrated films of an anti-parallel β-sheet peptide (LK₇β) adsorbed on glass slides. The hydrated films give chiral SFG response from water only when the peptide self-assembles into anti-parallel β-sheets. Experiments of isotopic labeling, isotopic dilution of water, and H₂O-D₂O exchange kinetics corroborate the assignments of the chiral SFG response to water stretching modes. Since individual water molecules are achiral, the chiral SFG response indicates formation of chiral superstructures of water around the anti-parallel β-sheet, implying that a protein secondary structure can imprint its chirality onto the surrounding water. This result demonstrates chiral SFG spectroscopy as a promising tool for probing water structures in protein hydration and addressing fundamental questions of protein structure-function.

Graphical Abstract

Water molecules are integral parts of proteins, influencing protein structures, dynamics, and functions. (1–3) Probing waters in protein hydration remains challenging due to background from a large population of water molecules in bulk solution. X-ray crystallography has shown that ordered water structures occur in the first hydration shell of proteins, (4) and that ordered water H-bonding networks can be found in protein reaction centers that are critical for protein function. (5,6) However, these ordered water structures were not observed under ambient conditions. NMR, (7) neutron scattering, (8) and terahertz spectroscopy (9,10) can report on the collective motions of waters near proteins but may not be sensitive to the architectures of waters surrounding proteins. Molecular dynamics (11) and quantum mechanics (12) simulations have yielded insights into water structures and dynamics at protein surfaces. Still, additional experimental validations of computational results are needed.

Here, we provide experimental evidence that vibrational sum frequency generation (SFG) can probe water superstructures around protein. SFG is a non-linear optical spectroscopy that selectively probes the second-order susceptibility χ⁽²⁾ of non-centrosymmetric media, such as interfaces. (13) SFG has become one of the major tools for investigating water structures and dynamics at biological interfaces. (14–18) Chiral SFG is a variant of SFG spectroscopy that is selective for the vibrational modes of chiral macromolecular structures. Theoretical treatments of chiral SFG suggest achiral molecular entities that form extended chiral macromolecular or supermolecular structures can give chiral SFG response. (19) Chiral SFG has been increasingly used to study biomacromolecules, e.g., distinguishing the handedness of dsDNA molecules (20) and characterizing protein secondary structures at interfaces. (21–24)

In two recent studies, Petersen and co-workers used chiral SFG and observed a chiral “spine of hydration” of water molecules in the minor groove of the DNA double helix, (25) and chiral “water wires” in artificial channels spanning supported lipid bilayers. (26) The chiral SFG technique allowed for detecting this water superstructure without bulk water background. The existence of a chiral water superstructure around DNA engenders foundational research questions. (27) Most pressing is whether chiral water superstructures also exist around proteins and other biomacromolecules.

In this study, we present evidence for chiral water superstructures surrounding the LK₇β peptide. This peptide has the sequence Acetyl-LKLKLKL-NH₂. (28) It forms amphiphilic antiparallel β-sheet with hydrophobic leucine and hydrophilic lysine side chains residing on opposite sides of the β-sheet. This peptide has been characterized by vibrational circular dichroism (VCD), attenuated total reflectance-infrared (ATR-IR), (28) achiral SFG, (29,30) and chiral SFG. (30,31) Here, we present studies of this peptide using chiral SFG in conjunction with isotopic labeling, isotopic dilution, and kinetic measurements of H₂O-D₂O exchange. The results support the assignments of the O-D and O-H stretches of water in the chiral SFG spectra. Since individual water molecules are achiral, the chiroptical responses of the water stretching modes indicate water molecules form chiral supermolecular structures around the anti-parallel β-sheet. Therefore, the results introduce a new concept that protein secondary structures can orient the surrounding water molecules into chiral superstructures. Since water stretching modes are highly sensitive to H-bonding environments, chiral SFG can be a promising tool for probing in situ and real time local H-bonding interactions of water with proteins under ambient conditions. Thus, it can potentially be used to study the interplay of water structures in hydration with protein structures and dynamics, addressing fundamental questions of how local water structures modulate protein folding, aggregation, and denaturation, as well as protein structural-functional correlations.

For all SFG measurements, LK₇β was dissolved in water solvent and adsorbed onto glass slides by evaporation, giving a hydrated thin film (see Experimental Methods). We first obtained the chiral amide I spectrum of LK₇β using H₂O on glass to examine the folding of LK₇β. The spectrum (red, Figure 1) shows an intense vibrational band at ~1618 cm⁻¹. This band is characteristic of the B₂ modes of antiparallel β-sheets. (31) This band is significantly reduced when the LK₇β sample was prepared using HCl aqueous solution at ~pH 2 as solvent (blue, Figure 1), indicating acid denaturation. Moreover, instead of using LK₇β made of native L-form amino acids, we also used the LK₇β peptide made of mixed L-leucine and D-lysine amino acids. Since proteins made of alternate L- and D-forms of amino acids cannot form antiparallel β-sheet structures, (32) the chiral amide I spectrum is silent (green, Figure 1). These results support that the LK₇β peptide made of native L-form amino acids is in antiparallel β-sheet and the observed chiral SFG signals are correlated with formation of β-sheets.

Figure 1.

Chiral SFG amide I spectra of the LK₇β thin films on glass slides: L-form LK₇β prepared using H₂O as solvent (red), L-form LK₇β prepared using HCl_(aq) at pH 2 as solvent (blue), and mixed (L-Leu)/(D-Lys) LK₇β prepared using H₂O as solvent (green). The fitting parameters and assignments are given in Table S1.

We obtained the chiral SFG spectra in the O-H stretch and O-D stretch regions for the hydrated thin films of LK₇β made with the native L-form amino acids in the H₂O and D₂O solvents, respectively (Figures 2a and 2b). Each spectrum exhibits a major and a minor vibrational band. These vibrational bands are muted for the hydrated thin film of LK₇β made with the mixed (L-Leu)/(D-Lys) residues (Figure S2). To fit the spectra (Figure 2), we used Eq. 1 and obtained the fitting results presented in the left-hand column of Table 1. We further performed three experiments: (i) O¹⁸-isotopic shifts, (ii) H₂O/D₂O isotopic dilution, and (iii) kinetic studies of H₂O-D₂O exchange. Based on the results of these three experiments, we assign the vibrational bands (left-hand column, Table 1). In the LK₇β-H₂O spectrum (Figure 2a), the peak at 3269 cm⁻¹ is due to N-H stretch of the LK₇β peptide and the peak at 3191 cm⁻¹ to the O-H stretch of H₂O. In the LK₇β-D₂O spectrum (Figure 2b), the peaks at 2401 cm⁻¹ and 2454 cm⁻¹ are both due to O-D stretches of D₂O. Below, we provide details of the three experiments and discuss the results in the context of these spectral assignments.

Figure 2.

Chiral SFG spectra of (a) LK₇β-H₂O and (b) LK₇β-D₂O on glass slides. The fitting parameters and assignments are given in Table 1. Fits to equation 1 are shown in black lines with component peaks in solid grey lines.

Table 1.

Fitting parameters and vibrational mode assignments for the spectra in Figures 2 and 3.

Samples	Parameters	Fittings	Assignments	Samples	Parameters	Fittings	Assignments
LK7β-H2O (2a)	${\mathit{\chi }}_{\mathit{N}\mathit{R}}^{(\mathbf{2})}$	0.01 ± 0.01	-	LK7β-H218O (3a)	${\mathit{\chi }}_{\mathit{N}\mathit{R}}^{(\mathbf{2})}$	0.02 ± 0.01	-
	ω1 (cm−1)	3191 ± 1	O-H (H2O)		ω1 (cm−1)	3179 ± 3	O-H (H218O)
	A1 (a.u.)	1.59 ± 0.11			A1 (a.u.)	1.06 ± 0.23
	Γ1 (cm−1)	38.62 ± 2.14			Γ1 (cm−1)	33.80 ± 6.95
	ω2 (cm−1)	3269 ± 1	N-H (LK7β)		ω2 (cm−1)	3267 ± 1	N-H (LK7β)
	A2 (a.u.)	5.82 ± 0.06			A2 (a.u.)	9.86 ± 0.12
	Γ2 (cm−1)	28.31 ± 0.21			Γ2 (cm−1)	32.03 ± 0.33
LK7β-D2O (2b)	${\mathit{\chi }}_{\mathit{N}\mathit{R}}^{(\mathbf{2})}$	0.02 ± 0.01	-	LK7β-D218O (3b)	${\mathit{\chi }}_{\mathit{N}\mathit{R}}^{(\mathbf{2})}$	0.02 ± 0.01	-
	ω1 (cm−1)	2401 ± 1	O-D (D2O)		ω1 (cm−1)	2390 ± 1	O-D (D218O)
	A1 (a.u.)	4.27 ± 0.04			A1 (a.u.)	5.75 ± 0.05
	Γ1 (cm−1)	28.18 ± 2.19			Γ1 (cm−1)	28.82 ± 1.33
	ω2 (cm−1)	2454 ± 4	O-D (D2O)		ω2 (cm−1)	2442 ± 3	O-D (D218O)
	A2 (a.u.)	0.64 ± 0.03			A2 (a.u.)	0.88 ± 0.06
	Γ2 (cm−1)	22.82 ± 9.42			Γ2 (cm−1)	26.54 ± 9.84

In the first experiment, we used H₂¹⁸O and D₂¹⁸O as solvent to isolate the water vibrational bands. The frequency of the water O-H (or O-D) stretch overlaps with the peptide N-H (or N-D) stretch. Thus, both water and the peptide can contribute to the chiral SFG spectra (Figure 2). However, the greater atomic mass of ¹⁸O is expected to red shift the O-H or O-D stretch by ~12 cm⁻¹, (33)³³ but not the peptide N-H and ND stretches. (34) Thus, the ¹⁸O-labeled water will allow for unambiguous assignments of water vibrational modes.

We used H₂¹⁸O as solvent to prepare the thin film of LK₇β on a glass slide and obtained the chiral SFG spectrum. This LK₇β-H₂¹⁸O spectrum (Figure 3a) exhibits a major peak at 3267 cm⁻¹ and a minor peak at 3179 cm⁻¹ (right-hand column, Table 1). Relative to the LK₇β-H₂O spectrum (Figure 2a; left-hand column, Table 1), the ¹⁸O substitution does not shift the major peak (within error ± 1 cm⁻¹), but red-shifts the minor peak by 12 cm⁻¹ (Table 1). Thus, the minor band is assigned to water O-H stretch and the major band is assigned to peptide N-H stretch.

Figure 3.

Chiral SFG spectra of (a) LK₇β-H₂¹⁸O and (b) LK₇β-D₂¹⁸O on glass slides. The fitting parameters and assignments are given in Table 1. Fits to equation 1 are shown in solid black lines with component peaks in solid grey lines.

In addition, we used D₂¹⁸O as solvent in the sample preparation and obtained the chiral SFG spectrum. The resulting LK₇β-D₂¹⁸O spectrum (Figure 3b) exhibits a major peak at 2390 cm⁻¹ and a minor peak at 2442 cm⁻¹ (right-hand column, Table 1). Relative to the LK₇β-D₂O spectrum (Figure 2b), both peaks are red-shifted by 11–12 cm⁻¹ (left-hand column, Table 1). Thus, both peaks are assigned to O-D water stretching modes (Table 1). Our assignments agree with previous achiral SFG studies of peptide at the air-D₂O interface. (35) Thus, the results of ¹⁸O isotopic shifts allow assigning both the major and minor bands of the LK₇β-D₂O spectrum (Figure 2b) to water stretching modes (Table 1).

The spectral assignments of water vibrational bands in the LK₇β-H₂O and LK₇β-D₂O spectra (Figure 2) are further supported by the second experiment of isotopic dilution. Isotopic dilution is carried out using mixtures of H₂O and D₂O at various ratios (see Experimental Methods). It is often used in vibrational studies to modulate intramolecular vibrational coupling of water molecules. (36) Mixing H₂O and D₂O at a ratio of x:y introduces HOD at an anticipated ratio of [H₂O]:[HOD]:[D₂O] = x²:2xy:y². (37) In HOD, the difference in the stretching frequencies of O-H and O-D removes the intramolecular vibrational coupling that otherwise occurs between two O-H bonds in pure H₂O, or two O-D bonds in pure D₂O. Although intermolecular couplings cannot be discounted under isotopic dilution, if water stretching contributes to the chiral SFG spectra of the hydrated LK₇β thin films (Figure 2), the spectra should be changed by isotopic dilution.

In the isotopic dilution experiments, we mixed H₂O and D₂O at volumetric ratios (v/v) of 1:1, 4:1, and 1:4, which result in [H₂O]:[HOD]:[D₂O] = 1:2:1, 16:8:1, and 1:8:16, respectively. We used these mixtures as solvent to prepare the LK₇β thin film and obtained the chiral SFG spectra in the O-H stretching region (Figure 4a) and O-D stretching region (Figure 4b). We present these spectra in the order of the HOD content relative to H₂O (D₂O) that contributes to the O-H (O-D) stretching region as indicated in Figure 4a (Figure 4b). Both Figures 4a and 4b show that higher HOD content results in larger deviation from the pure H₂O or D₂O spectrum. For example, in the spectrum (Figure 4a, blue) with the highest relative HOD content, the N-H stretch peak at 3269 cm⁻¹ shows lower SFG intensity. This decrease in intensity is expected, as the protons along the peptide backbone can be exchanged to deuterons. Moreover, fitting this spectrum (Figure S3 and Table S2) reveals a third peak emerging at 3304 ± 5 cm⁻¹, which was previously assigned by Shen and co-workers to the O-H stretch of HOD. (38) Altogether, the results support that water stretching modes contribute to the chiral SFG spectra of the hydrated LK₇β thin films (Figure 2), in agreement with the spectral assignment of water stretching bands in Figure 2 and Table 1.

Figure 4.

Isotopic dilution of the LK₇β water superstructures probed by chiral SFG. Chiral SFG spectra of samples with increasing HOD content in the (a) O-H and (b) O-D stretching regions. The dashed lines correspond to peak positions (Table 1) of the pure H₂O and pure D₂O samples.

Finally, the third experiment involved kinetic studies of H₂O-D₂O exchange in the hydrated LK₇β films, providing results in agreement with the spectral assignments of the chiral water stretching modes (Figure 2 and Table 1). We monitored the SFG spectra in the O-H and O-D stretching regions upon exposing the LK₇β-H₂O thin film to D₂O vapor (see Experimental Methods and Figure S4). Figures 5a show the chiral SFG spectra in the O-H stretching region before and after exposure to D₂O vapor. The major peak at 3269 cm⁻¹ remains prominent while the minor peak at 3191 cm⁻¹ disappears over time (Figure 5a). To analyze the kinetics, we extract the amplitudes (A_q) (see Experimental Methods) from the time-dependent chiral SFG spectra. Figure 5b plots the amplitudes for the minor and major vibrational bands against time. The amplitude of the minor band at 3191 cm⁻¹ undergoes a single exponential decay with a time constant (τ_minor) of 0.6 ± 0.1 min (Figure 5b, top), while the amplitude of the major band at 3269 cm⁻¹ fluctuates around its initial value (Figure 5b, bottom). Since the two vibrational bands do not show the same kinetics, they must originate from two different molecular species. The result therefore agrees with the assignments of the minor band to water stretching modes and the major band to the N-H stretching mode of the LK₇β peptide (Figure 2a and Table 1).

Figure 5.

H₂O-to-D₂O exchange via the vapor phase monitored by chiral SFG. (a) Response in the O-H/N-H region (red) before and (blue) after exposure to D₂O vapor. See Figure S5 and Table S3 for fittings of time-dependent spectra. (b) Amplitudes over time for the minor peak at 3191 cm⁻¹ (top) and major peak at 3269 cm⁻¹ (bottom). (c) Response in the O-D region before and after exposure to D₂O vapor. (d) Amplitudes over time for the minor peak at 2401 cm⁻¹ (top) and major peak at 2454 cm⁻¹ (bottom). Solid lines in (b) and (d) are single exponential fits.

Figure 5c shows the spectra of LK₇β-H₂O in the region of O-D stretch before and after exposure to D₂O vapor. Similarly, we extracted the amplitudes of the major and minor vibrational bands. Figure 5d plots the amplitudes as a function of time. Both the major and minor peaks undergo a single exponential increase with time constants for the minor peak (τ_minor) of 4.4 ± 0.5 minutes and major peak (τ_major) of 3.8 ± 0.4 minutes. These time constants are indistinguishable within one standard deviation, suggesting that both vibrational bands likely originate from a single species, consistent with the assignments of both vibrational peaks in the LK₇β-D₂O spectrum to water stretching modes (Figure 2b and Table 1).

The results of the three experiments – ¹⁸O isotopic shift (Figure 3), H₂O/D₂O isotope dilution (Figure 4), and kinetic studies of H₂O-D₂O exchange (Figure 5) – corroborate the assignments of the minor band (3191 cm⁻¹) in the LK₇β-H₂O spectrum (Figure 3a) to the water O-H stretch and both the major (2401 cm⁻¹) and minor (2454 cm⁻¹) bands in the LK₇β-D₂O spectrum (Figure 3b) to water O-D stretch. The presence of one water band when the peptide is solvated in H₂O (Figure 1a) versus two water bands when the peptide is solvated in D₂O might be because the intense N-H stretch response obscures another H₂O band in the spectrum, or because H₂O and D₂O could be differently structured around the peptide due to different H-bonding strengths. Computational studies in conjunction with experimental work will be needed for further exploration.

We attribute the chiral SFG signals of water stretches to chiral supermolecular structures of water surrounding the β-sheet LK₇β peptide. These chiral water signals cannot be observed using the LK₇β peptide made with mixed L/D-forms of amino acids (Figure S2). Since the mixed (L-Leu)/(D-Lys)-LK₇β peptide contains the same side chains but cannot fold into β-sheet structures, the observed chiral SFG responses of water stretching is not due to local interactions of the side-chains but associated with the antiparallel β-sheet structure. These chiral SFG responses arise from orthogonal second-order susceptibility elements (i.e., ${\chi }_{ijk}^{(2)}$, where 𝑖 ≠ 𝑗 ≠ 𝑘) and these orthogonal elements can be non-zero only for chiral molecular systems. Since individual water molecules are achiral, the chiral SFG signals of water stretching modes must be generated from water molecules arranged in chiral superstructures.

The chiral SFG theory and results of control experiments suggest that the bulk of the LK₇β thin films can contribute to the observed chiral SFG signals. According to the theoretical framework established by Simpson and coworkers, orthogonal χ⁽²⁾ elements can originate from achiral molecular entities arranged in chiral macromolecular structures at anisotropic interfaces, giving rise to strong surface-specific SFG signals. Such surface-specific chiral signals have been increasingly used to study chiral biomacromolecular structures at interfaces. (20,21,23,39,40) Further elaboration concerning the orthogonal ${\chi }_{zyx}^{(2)}$ element based on a symmetry argument is detailed in a recent review. (23) The analysis suggests that under dipole approximation and without electronic resonance an anisotropic chiral bulk system can also give rise to strong chiral SFG signals that are detected in reflection geometry (Figure S6). Such systems can include (i) reflection from surfaces of chiral crystals and (ii) reflection from solid or liquid interfaces adsorbed with layers of anisotropic chiral macromolecules. In these cases, the depth of bulk contribution can be roughly equal to the coherence length. Unlike chiral SFG generated from isotropic bulk media (e.g., chiral liquid) that is generally small and requires electronic resonance for detection, (41,42) these chiral signals generated from anisotropic media can be of a magnitude comparable to or even higher intensity than the surface-specific SFG signal. (43) In addition, the hydrated thin film of LK₇β is indeed anisotropic on glass surface with a preferred orientation of the β-strand axis, as shown by our independent measurements using polarized attenuated total reflectance-Fourier transform infrared (polarized ATR-FTIR) spectroscopy (Figure S7). Hence, the anisotropic bulk of the LK₇β thin film likely contributes to the observed chiral SFG signals.

The observation of water stretching modes by chiral SFG spectroscopy suggests that water molecules can follow the contour of protein secondary structures and form chiral supermolecular structures. Although duplex DNA orienting surrounding water molecules into chiral structures has for decades been a subject of intense interest, (11,44–47) the ability of proteins to imprint their macromolecular chirality onto surrounding waters has not been investigated. In general, there are 1.2 water molecules per amino acid in proteins and around 80% of ordered water structures are found in the first hydration shell of proteins. (4) The chiral supermolecular structures observed by chiral SFG are likely composed of these ordered water molecules. Indeed, X-ray structures of amyloid-like fibrils have revealed ordered arrangements of water oxygen atoms along surfaces of β-sheets.(48–50) Further experimental and theoretical efforts are pertinent to elucidate the molecular arrangements and H-bonding configurations in the chiral water superstructures. For example, temperature-dependent kinetic studies of H₂O-D₂O exchange can potentially reveal the energetics of H-bonding interactions essential for formation of water superstructures.

Our prior study reported chiral SFG spectra of LK₇β at the air-H₂O and air-D₂O interfaces in the O-H and O-D spectral regions, respectively. (22,31) In this prior study, ab initio calculations guided the assignments of all vibrational bands of LK₇β at the air-water interface to the N-H or N-D stretches of the LK₇β peptide. With the current results, these assignments need to be re-examined in future experiments. Nonetheless, the current results confirm the assignment of the major band at 3269 cm⁻¹ to the N-H stretches of the LK₇β peptide, supporting the applications of this vibrational band for characterizing protein structures.(21–23,31,51–54)

The work reported here suggests that SFG can be a promising tool for probing protein hydration under ambient conditions. The water stretching modes are highly sensitive to H-bonding environments. Thus, chiral SFG can be used to solve a wide range of problems related to how H-bonding interactions with water molecules modulate protein structures and dynamics, potentially providing molecular insights into the role of hydration in protein aggregation, folding, and denaturation as well as structural-functional correlation of proteins. As a vibrational method, chiral SFG is label-free, non-invasive, in situ, and real time. It does not require large amount of sample or growth of crystal. Thus, it can be a powerful method complementing the current techniques (e.g., NMR and X-ray crystallography) to probe energetics and dynamics of protein hydration. Being a second-order optical method, chiral SFG offers unique selectivity and high sensitivity for detecting chiral macromolecular structures and can potentially offer new biophysical descriptions of hydration of proteins and other biomacromolecules (e.g., DNA, RNA, and polysaccharides).

EXPERIMENTAL METHODS

Sample preparation.

The LK₇β peptide was obtained as a lyophilized powder (GL Biochem Ltd.). The powder was dissolved in H₂O, H₂¹⁸O (Sigma Aldrich; 97 atom % ¹⁸O), D₂O (Cambridge Isotope Laboratories, Inc., Andover, MA; 99.9 atom % D), D₂¹⁸O (Sigma Aldrich; 99 atom % D, 95 atom % ¹⁸O), or a mixture of H₂O and D₂O at 1 mM concentration in 100 μL aliquots. The aliquots were frozen in liquid nitrogen and then stored at −80 °C until time of use. Plain glass microscope slides (Thermo Scientific, Portsmouth, NH) were plasma-cleaned (Harrick Plasma, Ithaca, NY; Plasma Cleaner PDC-32G), and 100 μL of defrosted LK₇β solution was pipetted onto the slide. The slides were then dried in a sealed container with desiccant and purged with dry nitrogen to avoid replacement of water isotopes by atmospheric water. The LK₇β peptide formed a hydrated thin film on the glass slides for the SFG experiments.(55–57)

SFG experiments.

All SFG spectra were collected in reflection geometry using a homebuilt broad-bandwidth SFG spectrometer with ~120-fs pulsed IR and ~2-ps pulsed visible beams at a repetition rate of 5 kHz (complete details about the spectrometer can be found in Ma et al.).(58) The psp and ssp polarization settings were used for obtaining chiral and achiral SFG spectra, respectively (see section III in Supporting Information). The spectra in the O-H stretch and O-D stretch regions were collected with the LK₇β thin film on glass slides facing down in the reflection geometry. The SFG spectra in the amide I region were collected with the LK₇β thin film facing up to avoid attenuation of the IR beam by the glass slides. The spectra were collected with an integration time of 10 minutes.

Kinetics of H₂O-D₂O Exchange.

The glass slide with adsorbed LK₇β thin film was supported by a Teflon beaker on a stage with adjustable height. A plastic cuvette was sitting inside the Teflon beaker. At t = 0 min, 1.6 mL of D₂O was carefully pipetted into the cuvette through a gap between the glass slide and the wall of the Teflon beaker without disturbing the slide (Figure S1). The adsorbed LK₇β thin film was situated approximately 1 cm from the surface of the D₂O. The use of the plastic cuvette was to avoid the artifact of water crawling on the wall of the Teflon beaker that supported the glass slide. Evaporation of D₂O from the plastic cuvette induced H₂O-to-D₂O exchange in the LK₇β thin film. Chiral SFG spectra were collected in the O-H stretch and O-D stretch regions until the SFG intensity reached stable plateaus.

Analysis of spectra.

The raw spectral data were processed as detailed in Ma et al. (58) The resulting SFG spectra were fitted to:

$$I_{SFG}\propto {\left|{\chi }_{NR}^{(2)}+{\sum }_q\frac{A_q}{{\omega }_{IR}-{\omega }_q+i{\mathrm{\Gamma }}_q}\right|}^2$$ (1)

with the SFG intensity I_SFG, the non-resonant second-order susceptibility ${\chi }_{NR}^{(2)}$, the frequency of the incident IR beam ω_IR, and the amplitude, frequency, and damping factor of the q^th resonant vibrational mode, A_q, ω_q, and Γ_q, respectively. To extract the amplitudes (A_q) for H₂O-D₂O exchange kinetics, the highest intensity spectrum was first fit to Eq. 1. After a best fit of the spectrum was achieved, the background ${\chi }_{NR}^{(2)}$, frequency ω_q, and damping Γ_q parameters were fixed for fitting the remaining spectra to obtain the amplitude values for kinetic analyses.