Broad-Bandwidth Chiral Sum Frequency Generation Spectroscopy for Probing the Kinetics of Proteins at Interfaces.

The kinetics of proteins at interfaces plays an important role in biological functions and inspires solutions to fundamental problems in biomedical sciences and engineering. Nonetheless, due to the lack of surface-specific and structural-sensitive biophysical techniques, it still remains challenging to probe protein kinetics in situ and in real time without the use of spectroscopic labels at interfaces. Broad-bandwidth chiral sum frequency generation (SFG) spectroscopy has been recently developed for protein kinetic studies at interfaces by tracking the chiral vibrational signals of proteins. In this article, we review our recent progress in kinetic studies of proteins at interfaces using broad-bandwidth chiral SFG spectroscopy. We illustrate the use of chiral SFG signals of protein side chains in the C-H stretch region to monitor self-assembly processes of proteins at interfaces. We also present the use of chiral SFG signals from the protein backbone in the N-H stretch region to probe the real-time kinetics of proton exchange between protein and water at interfaces. In addition, we demonstrate the applications of spectral features of chiral SFG that are typical of protein secondary structures in both the amide I and the N-H stretch regions for monitoring the kinetics of aggregation of amyloid proteins at membrane surfaces. These studies exhibit the power of broad-bandwidth chiral SFG to study protein kinetics at interfaces and the promise of this technique in research areas of surface science to address fundamental problems in biomedical and material sciences.

Introduction

Protein kinetics at interfaces plays an important role in biology and biomedical sciences and engineering. For instance, life processes rely heavily on the dynamic interactions between proteins and cell membrane surfaces, such as molecular transportation,[1] cell adhesion,[2] and signal transduction.[3] In addition, the kinetics of <span class="Disease">protein aggregation at membrane surfaces is associated with the progression of neurodegenerative diseases.[4] Furthermore, proteins used as therapeutics can be denatured during storage and manufacturing upon interactions with surfaces of piping and receptacles;[5] therefore, understanding how fast the interaction alters protein structures is crucial to the optimization of shelf life. While proteins and peptides can self-assemble into well-defined structures at various interfaces, the rate of the self-assembly process dictates the stability and synthesis method for manufacturing functional biomimetic materials.[6] Moreover, enzymes have been developed into biocatalysts for the transformation of organic compounds and for biofuel cells,[7] and their functions and activities are highly correlated to their real-time conformations at interfaces. Hence, in situ and real-time methods for monitoring the kinetics of protein conformational changes at interfaces are crucial in many research fields across various disciplines in surface science.

The requirement for the selectivity of molecules at interfaces without interference from bulk media or the use of spectroscopic labels renders it challenging to probe protein kinetics at interfaces. Among the available biophysical techniques, fluorescence spectroscopy often requires labeling,[8] potentially perturbing the behaviors of proteins. Nuclear magnetic resonance (NMR) spectroscopy,[9] circular dichroism (CD) spectroscopy,[10] and dynamic light scattering (DLS)[11] are all excellent techniques for probing protein kinetics in bulk solutions but are unable to obtain structural information at interfaces. Atomic force microscopy (AFM)[12] and thin-film X-ray diffraction[13] can reveal the morphology of protein layers at interfaces but fall short of molecular information about chemical structures. Although conventional vibrational spectroscopy techniques, such as infrared reflection absorption spectroscopy (IRRAS)[14] and surface-enhanced Raman spectroscopy,[15] are able to probe protein kinetics in real time, the overlap of characteristic peaks from α-helical and disordered secondary structures in the amide I region (both centered at ∼1654 cm–1)[16,17] poses challenges in the quantitative interpretation of kinetic data, especially for complex native proteins. Hence, a complementary method is desired to carry out kinetic studies of proteins at interfaces.

Chiral sum frequency generation (SFG) spectroscopy is a noninvasive and label-free technique that is able to provide surface-specific and structurally sensitive information about proteins in situ and in real time. Previously, several groups have conducted in situ chiral SFG studies on proteins at interfaces using scanning SFG spectrometry.[18−21] However, the acquisition of SFG spectra requires the scanning of vibrational frequencies, making real-time kinetic studies difficult. Since the setup of a broad-bandwidth SFG spectrometer in our laboratory in 2008, we have been implementing the chiral SFG method to probe the kinetics of protein structures at interfaces.[22−26] The broad-bandwidth spectrometer has allowed us to obtain SFG spectra shot-by-shot without spectral scanning,[27] thereby advancing SFG spectroscopy into a real-time method for monitoring protein kinetics at interfaces for a time scale of 1–10 min.[22−25] With its surface specificity and structural selectivity, chiral SFG is able to characterize different secondary structures that are difficult to distinguish using conventional vibrational methods, such as α-helices and disordered structures,[22] allowing kinetic studies of protein conformational changes exclusively at interfaces both in situ and in real time.[22,23] Moreover, its chirality-selective characteristic makes chiral SFG a background-free technique since it eliminates the O–H stretch background of water and allows the use of the peptide amide N–H stretch as a new vibrational tool for studying protein structures.[22,24] Furthermore, chiral SFG is silent to achiral molecular structures, thus making possible the use of the chiral C–H stretch for protein characterization without interference from lipid molecules and other organic compound at interfaces.[25] Given the higher laser power used in the C–H stretch vibrational regions, the spectral qualities can be improved; therefore, a higher temporal resolution for kinetic studies can be achieved.[25] Altogether, the above developments have made broad-bandwidth chiral SFG spectroscopy a promising complementary technique for studying the kinetics of protein conformational changes at interfaces.

In this article, we will review the recent progress in the kinetic studies of proteins performed at interfaces. Following a brief introduction of chiral SFG spectroscopy, we will first discuss the development of broad-bandwidth SFG spectrometers, which makes it possible to acquire spectra in real time, vital to performing kinetic studies. Then, we will discuss our kinetic studies of proteins using chiral SFG in three vibrational regions: (a) the C–H stretch, which can reveal structures of side chains in proteins, (b) the N–H stretch, which can be used to monitor proton exchange in the protein backbone, and (c) the <span class="Chemical">amide I region, which can be used to distinguish secondary structures of proteins at interfaces. We will then discuss the outlook of chiral SFG spectroscopy in terms of method development and potential applications in the broader field of surface science.

Basic Principles of Chiral SFG

SFG Method

SFG is a second-order spectroscopic technique pioneered by Shen’s group.[28] Rigorous treatment of the SFG theory has been described in excellent books.[29,30] In this section, we focus on vibrational SFG that uses two pulsed laser sources, one at IR frequency ωIR and the other at visible frequency ωvis (Scheme A). When these two beams spatially and temporally overlap at an interface, a new laser beam with the sum of the incident frequencies (ωIR+ ωvis) is generated. When the IR frequency (ωIR) is in resonance with molecular vibrations, the SFG signal is enhanced, thereby providing vibrational information about molecules.

Scheme 1

Vibrational Sum Frequency Generation Spectroscopy

(A) Illustration of vibrational sum frequency generation. (B) Illustrations of chiral (C) and achiral (C) surfaces. The blue susceptibility elements are nonzero at both chiral and achiral surfaces, and the red ones are nonzero only at chiral surfaces. (C) Illustrations of polarization settings for achiral and chiral SFG experiments: the projection of the electric field of p-polarized and s-polarized light onto the laboratory coordinates. (Left) ssp (achiral) polarization setting. (Middle) psp (chiral) polarization setting. (Right) Definition of polarization.

Since the development of chiral SFG, there have been discussions about whether chiral SFG is surface-specific. We recently discussed in a review article, from first principles, how to consider the chiral SFG contributions from the bulk versus interface.[26] Our theoretical analysis concluded that when an anisotropic interface is situated between two isotropic bulk media under the conditions that electronic resonance is absent, chiral vibrational SFG is surface-specific.[26]

More specifically, using the <span class="Chemical">dipole approximation, second-order polarization, P(2), can be induced at the sum frequency (ωIR+ ωvis) to coherently generate the SFG electric fieldwhere x̂, ŷ, and ẑ are unit vectors; I, J, and K (= x, y,and z, laboratory coordinates) specify the vector direction; χIJK(2) is a second-order susceptibility tensor element; and Evis, EIR, and ESFG are the optical fields. The second-order susceptibility of an interface contains a nonresonant term, χNR(2), which is independent of IR frequencies, and a sum of resonance terms, χ(2)where A, Γ, and ωIR are the amplitude, damping coefficient, and incident IR frequency, respn>ectively, and ω is the vibrational frequency of the qth vibrational mode. As seen in eq , when the IR frequency (ωIR) coincides with the vibrational frequency (ω), χ(2) is maximized, yielding an amplified SFG electric field according to eq .

Chiral SFG

Here, we discuss how chiral SFG differs from achiral SFG. As discussed by Simpson and co-workers,[31,32] an achiral surface adopts C symmetry, leaving seven χ elements nonzero (χ = χ, χ = χ, χ = χ, and χ), while a chiral surface adopts C symmetry (Scheme B), adding six extra χ nonvanishing elements (χ, χ, χ, χ, χ, and χ). When I ≠ J ≠ K, the second-order susceptibility elements exhibit the characteristics of a chiral Cartesian coordinate system, where the three axes x, y, and z are orthogonal to one another. The orthogonal χ (I ≠ J ≠ K) elements are representations of a chiral surface, which can be measured by using specific polarization settings in an SFG experiment.

SFG experiments are often performed using linearly s- or p-polarized IR and visible beams, while measuring s- or p-polarized SFG signals. While chiral SFG signals can be indirectly measured with the interference method using a mixed polarization of s and p,[33] we used the psp polarization setting (p-polarized SFG, s-polarized visible, and p-polarized IR) to directly observe chiral SFG signals (Scheme C, middle). We also used the ssp polarization setting (s-polarized SFG, s-polarized visible, and p-polarized IR) to observe achiral SFG signals (Scheme C, left). The SFG intensity measured using these settings can be expressed aswhere ISFG, Ivis, and IIR are the intensities and χssp(2) and χpsp(2) are the effective second-order susceptibilities, which can be expressed as linear combinations of χ(2) of the interface (Scheme C, left and middle).[26]where α is the incident/reflected angle of the ith laser beam and L(ω) is the Fresnel factor. Scheme C shows that the χpsp(2) susceptibility is related to orthogonal tensor elements of χ and χ, which are nonzero for a chiral interface (C symmetry). Under the condition of no electronic resonance, χ = 0, and the psp polarization setting measures only χ, which contains structural information about a chiral interface.

Broad-Bandwidth SFG for Studies of Protein Kinetics at Interfaces

Two types of spectrometers, scanning and broad-bandwidth, have been used in SFG experiments. The IR bandwidth of the two systems can be determined by the uncertainty principle, Δv̅ ∝ (1/Δt), where Δt is the pulse width and Δv̅ is the bandwidth. In a scanning spectrometer, picosecond/nanosecond pulsed laser sources are used to generate both IR and visible beams with a narrow bandwidth. For example, the bandwidth of a 2 ps IR pulse is 10–15 cm–1. Spectra are thus acquired stepwise by scanning the IR frequencies across a vibrational region, such as amide I (1600–1700 cm–1) and the C–H stretch (2800–3000 cm–1). In a broad-bandwidth spectrometer, a femtosecond IR beam and a picosecond visible beam are used. As an example, our broad-bandwidth spectrometer (Scheme )[27] uses a 100 fs mid-IR laser pulsed beam spanning a bandwidth of 200–300 cm–1, which enables shot-by-shot acquisitions of spectra in a specific vibrational region without scanning the IR frequencies. In the meantime, the bandwidth of the visible beam is narrowed by a pulse shaper to ∼10 cm–1, corresponding to picosecond pulses, enabling high spectral resolution. The combination of femtosecond IR and picosecond visible beams reduces the acquisition time of full spectra and thus facilitates the studies of kinetic processes at interfaces.

Scheme 2

Illustration of a Broad-Bandwidth SFG Spectrometer

The spectrometer comprises a 6 W regenerative amplifier (120 fs) seeded by a Ti:sapphire oscillator and pumped by two Nd:YLF lasers. One half of the 6 W output passes through a home-built pulse shaper to produce narrow-bandwidth 800 nm pulses, and the other half pumps an automated optical parametric amplifier (OPA) to generate broad-bandwidth IR pulses. The power of the 800 nm beam at the sample stage is ∼135 mW, and the power of the IR beam is ∼10 mW. The repetition rate is 5 kHz. The incident angles for the 800 nm and IR beams are 56° and 69°, respectively. The SFG signal generated from the sample is filtered and then dispersed by a monochromator before being detected by a CCD.

The spectrometer comprises a 6 W regenerative amplifier (120 fs) seeded by a Ti:sapphire oscillator and pumped by two Nd:YLF lasers. One half of the 6 W output passes through a home-built pulse shaper to produce narrow-bandwidth 800 nm pulses, and the other half pumps an automated optical parametric amplifier (OPA) to generate broad-bandwidth IR pulses. The power of the 800 nm beam at the sample stage is ∼135 mW, and the power of the IR beam is ∼10 mW. The repetition rate is 5 kHz. The incident angles for the 800 nm and IR beams are 56° and 69°, respectively. The SFG signal generated from the sample is filtered and then dispersed by a monochromator before being detected by a C<span class="Disease">CD.

The broad-bandwidth SFG (BBSFG) spectrometer was first developed by Stephenson’s group.[34] Since then, broad-bandwidth SFG has been used as a tool for the characterization of molecular kinetics at interfaces. Allen’s group compared the bandwidths that can be covered by 80 fs and 1.6 ps IR pulses (Figure A)[35] and demonstrated the possibility of acquiring broad-bandwidth IR profiles in the frequency range of 1000–3500 cm–1,[36] rendering it possible to obtain high-quality spectra with a spectral resolution of <5 cm–1.[37] They obtained full spectra of surfactants at <span class="Chemical">aqueous interfaces with a short acquisition time (500 ms) in the C–H stretch region.[36] Our group extended the frequency range of broad-bandwidth IR to 900–3800 cm–1 (Figure B).[27] Borguet’s group compressed the IR pulse to ∼20 fs and achieved an ultrabroad bandwidth of IR profiles that cover over 1000 cm–1 (Figure C).[38] Benderskii’s group incorporated heterodyne detection to accurately subtract the nonresonant SFG background and to further improve the signal-to-noise ratio (Figure D1–D3),[39] obtaining phase information to extract molecular-level information such as absolute orientation.[39] Bonn’s group and Tahara’s group[40] further applied heterodyne detection to the studies of molecular orientation and dynamics at interfaces. Recently, Wang’s group developed a broad-bandwidth spectrometer with a subwavenumber spectral resolution (<1 cm–1, Figure E),[41] demonstrating the potential of broad-bandwidth spectrometers to probe the kinetics of ultrafine structural changes at interfaces. All of the recent developments of broad-bandwidth SFG spectrometers have widened the applications of SFG and can potentially be implemented in kinetic studies of proteins at interfaces to improve spectral and temporal resolutions.

Figure 1

(A) Comparison of 80 fs and 1.6 ps IR profile bandwidths. Reprinted with permission from ref (35). Copyright 2001, The Japan Society for Analytical Chemistry. (B) IR profiles of a typical femtosecond BBSFG spectrometer in the mid-IR 900–3800 cm–1 region. Reprinted with permission from ref (27). Copyright 2009, Springer. (C) IR profiles of an ultra-BBSFG spectrometer. Reprinted with permission from ref (38). Copyright 2011, Optical Society of America. (D1) Homodyne-detected and (D2) heterodyne-detected SFG C–H stretch spectra of saturated octanol/d-octanol monolayers (octanol mole fraction indicated) at different surface coverages obtained using a 100 s acquisition time, with (D3) enlargement of the heterodyne-detected spectrum of the 6% monolayer. (D1–D3) Reprinted with permission from ref (39). Copyright 2008, American Chemical Society. (E) Spectrum of the 800 nm visible picosecond (ps) pulse with a fwhm of 0.55 ± 0.01 cm–1. Reprinted with permission from ref (41). Copyright 2013, AIP Publishing LLC.

(A) Comparison of 80 fs and 1.6 ps IR profile bandwidths. Reprinted with permission from ref (35). Copyright 2001, The Japan Society for Analytical Chemistry. (B) IR profiles of a typical femtosecond BBSFG spectrometer in the mid-IR 900–3800 cm–1 region. Reprinted with permission from ref (27). Copyright 2009, Springer. (C) IR profiles of an ultra-BBSFG spectrometer. Reprinted with permission from ref (38). Copyright 2011, Optical Society of America. (D1) Homodyne-detected and (D2) heterodyne-detected SFG C–H stretch spectra of saturated <span class="Chemical">octanol/d-octanol monolayers (octanol mole fraction indicated) at different surface coverages obtained using a 100 s acquisition time, with (D3) enlargement of the heterodyne-detected spectrum of the 6% monolayer. (D1–D3) Reprinted with permission from ref (39). Copyright 2008, American Chemical Society. (E) Spectrum of the 800 nm visible picosecond (ps) pulse with a fwhm of 0.55 ± 0.01 cm–1. Reprinted with permission from ref (41). Copyright 2013, AIP Publishing LLC.

Kinetic Studies on Proteins at Interfaces Using Chiral Broad-Bandwidth SFG

In this section, we will discuss the applications of broad-bandwidth chiral SFG for probing the kinetics of (1) the protein assembly process using the chiral signal of the C–H stretch,[25] (2) H/D exchange using the chiral signal of the N–H stretch,[24] and (3) early-stage amyloidogenesis using the chiral signals of <span class="Chemical">amide I and the N–H stretch.[22,23]

Chiral C–H Stretch for Probing the Kinetics of Self-Assembly of Proteins at the Interface

The self-assembly of biomacromolecules at interfaces is important in many research fields, such as polymer sciences, material sciences, and supramolecular chemistry.[42] To monitor the self-assembly processes of proteins or peptides, the <span class="Chemical">amide I vibrational signal is often used.[16,17] However, the amide I vibration does not reveal structural information about side chains, hampering the comprehensive understanding of self-assembly mechanisms. In addition, many biomacromolecules, such as DNA, RNA, and synthetic biopolymers, do not have amide groups. Given the abundance of hydrocarbyl groups in various chiral biomacromolecules, it is useful to establish C–H stretch vibrations as a probe for the self-assembly process of chiral molecules at interfaces using chiral SFG spectroscopy.

Here, we describe our studies of real-time kinetics of the LK7β peptide at the air/water interface using the chiral C–H stretch signals from the leucine side chains.[25] The LK7β peptide is an amphiphilic peptide with a sequence of LKLKLKL. It forms antiparallel β-sheets at the air/water interface at neutral pH,[43] as evidenced by the characteristic ∼1619 cm–1 (B2 mode) and ∼1680 cm–1 (B1 mode) peaks in the amide I region (Figure ).[24] Using chiral SFG, we observed in real time the self-assembly of LK7β at the air/water interface.[25] We combined these kinetic results with the surface-pressure measurement and proposed a molecular mechanism for the self-assembly process of LK7β at the air/water interface, as described below.[25]

Figure 2

Chiral amide I spectra of LK7β at the air/H2O interface (blue) and air/D2O interface (red) at pH 7.4. Reprinted with permission from ref (24). Copyright 2013, American Chemical Society.

In the kinetic studies, we first acquired the in situ chiral SFG spectra of LK7β (∼25 μM concentration) at the air/water interface at different pH values of the bulk solution using an acquisition time of 1 min (Figure ).[25] At pH ∼7.4, LK7β exhibits strong chiral C–H stretch signals, confirming the formation of chiral secondary structures at the interface. Figure shows that the chiral C–H signal intensity plummets to a negligible level after the addition of <span class="Chemical">HCl to bring the bulk pH down to ∼1.2, which is due to the denaturation of ordered chiral structures into disordered achiral structures. Then, Figure also shows the recovery of the chiral SFG signal ∼3 h after the addition of NaOH solution to bring the pH back to ∼7.4. The recovered intensity level is slightly lower than that of the initial state. The lower intensity is likely due to the dilution of the LK7β bulk concentration from pH adjustment, which leads to a lower surface density at the interface in equilibrium. The recovery of the chiral C–H signals indicates the reassembly of LK7β at the interface.

Figure 3

In situ chiral SFG spectra of LK7β at the air–water interface in the C–H stretch region. Spectral deconvolution is provided for the assembled and the reassembled states. The insets show the spectral fitting in the region from 2850 to 2920 cm–1. Reprinted with permission from ref (25). Copyright 2013, American Chemical Society.

In situ chiral SFG spectra of LK7β at the air–<span class="Chemical">water interface in the C–H stretch region. <span class="Chemical">Spectral deconvolution is provided for the assembled and the reassembled states. The insets show the spectral fitting in the region from 2850 to 2920 cm–1. Reprinted with permission from ref (25). Copyright 2013, American Chemical Society.

Hence, we could use the chiral C–H stretch signals to monitor the kinetics of the self-assembly process of LK7β at the air/<span class="Chemical">water interface in real time (Figure A).[25] At time = 0 min, we added <span class="Chemical">NaOH to bring the pH back to ∼7.4. The signal remains almost silent for the first 60 min, followed by the steady increase during the next 70 min. The signal holds at the maximum level for at least 30 min before we ended the experiment. To confirm the kinetics, we repeated the experiments four additional times and monitored the intensity of the highest peak (2958 cm–1) in the spectra (Figure B), where the red curve corresponds to the result in Figure A. While the time taken for the recovery of the chiral SFG signal in the five experiments ranges from ∼70 to ∼137 min, it is still of the same order of magnitude. Most importantly, the kinetic curves reveal a lag phase before the buildup of chiral SFG intensities during the self-assembly of LK7β into chiral structures at the interface.

Figure 4

(A) Time-dependent chiral SFG spectra showing the self-assembly of LK7β at the interface in the C–H stretch region. At t = 0, NaOH is added to change the pH back to ∼7.4. (B) Normalized intensity of the CH3 AS peak (2958 cm–1) as a function of time during the self-assembly process of LK7β at the air–water interface. The experiment is repeated five times. (C) Surface-pressure measurement for the denaturation and reassembly of LK7β at the air/water interface. (D) Illustration of the LK7β self-assembly mechanism at the air/water interface. (A–C) Reprinted with permission from ref (25). Copyright 2013, American Chemical Society.

To explore the molecular mechanism for the lag phase, we conducted surface-pressure measurements of the self-assembly process (Figure C).[25] We first set the surface pressure of <span class="Chemical">water to zero. Once LK7β is added to the buffer solution at pH ∼7.4, the surface pressure immediately rises. Because surface pressure is positively correlated with molecular surface density, this observation indicates a fast adsorption of LK7β at the interface. Upon the change of pH to ∼1.2, the surface pressure drops due to desorption (Figure C). After pH neutralization to ∼7.4, surface pressure gradually increases and reaches a plateau at ∼76 min, followed by another increase until attaining equilibrium.

Combining the results of chiral SFG studies in the C–H stretch region and surface pressure measurements, we proposed a two-stage self-assembly mechanism for LK7β at the air/water interface (Figure D).[25] Upon neutralization of pH, LK7β starts to adsorb at the interface, which increases the surface pressure rapidly until reaching a plateau. In this stage, LK7β is largely <span class="Disease">disordered, thereby exhibiting no chiral C–H SFG signal. At ∼1 h after the neutralization of pH, LK7β in disordered structures continues to populate the interface and reach a critical point that triggers the self-assembly of disordered structures into antiparallel β-sheet structures. This is energetically favorable because the folded antiparallel β-sheet structures are more compact, which allow further adsorption of the amphiphilic LK7β peptide onto the air/water interface. Correspondingly, the increases in both the chiral C–H stretch signals and the surface pressure at ∼1 h after pH neutralization are attributed to the simultaneous adsorption and self-assembly of LK7β into antiparallel β-sheet structures at the interface (Figure D).

To underscore the advantages of chiral SFG in probing the kinetics of chiral macromolecular structural changes at the interface, we also used achiral SFG to repeat the above experiments. The in situ achiral SFG spectra do not exhibit consistent peak shapes and relative peak intensities for the assembled states (Figure A1,A3) at different probing spots of the air/<span class="Chemical">water interface in the beaker containing the sample (a–c in Figure A1,A3).[25] Moreover, prominent achiral SFG signals can be observed when LK7β is in <span class="Disease">disordered structures at pH 1.2 (Figure A2).[25] As a result, the time-dependent achiral spectra do not show an obvious trend of changes in intensity during the self-assembly process upon pH neutralization. Consequently, the results of the kinetic experiments using achiral SFG are not reproducible (Figure C1,C2).[25]

Figure 5

Achiral and chiral SFG spectra characteristic of assembled, denatured, and reassembled states of LK7β at the air/water interface. (A1–A3) Achiral in situ spectra. (B1–B3) Chiral in situ spectra. Letters a–c represent different probing areas at the interface, and the same letter in ssp and psp spectra represents the same area. The highest peak intensities are scaled to the same value for the convenience of comparison. (C1, C2) Time-dependent achiral (ssp) C–H stretch spectra showing the reassembly of LK7β at the interface in two independent experiments. At t = 0, NaOH was added and the pH was brought back to ∼7.4 to trigger the self-assembly process. Reprinted with permission from ref (25). Copyright 2013, American Chemical Society.

Here, we offered the simplest explanation for the inconsistency of achiral spectra. We attributed the irreproducible achiral SFG spectra to the inhomogeneity at the interface in terms of achiral structures. When the self-assembly reaches equilibrium, there can still be a small proportion of peptide molecules that do not form antiparallel β-sheets. They can exist as either disordered or partially assembled intermediate structures, all of which can contribute to the achiral spectra. The partially assembled intermediates can be largely inhomogeneous in terms of structures and orientations that complicate the achiral spectra. On the other hand, only assembled chiral antiparallel β-sheets can contribute to the chiral spectra. Since the assembled β-sheet structures with the leucine residue point to the air phase and the lysine residue point to the water phase, the assembled chiral structures are relatively homogeneous. Thus, chiral spectra are more reproducible at different probing spots than the achiral ones. On the basis of this interpretation, both achiral and chiral spectra should be reproducible at different probing spots when the interface is covered with only one kind of structure. Indeed, this is what we observed at acidic pH 1.2, where the interface is relatively homogeneous with the coverage of completely disordered structures (Figure A2).

Achiral and chiral SFG spectra characteristic of assembled, denatured, and reassembled states of LK7β at the air/<span class="Chemical">water interface. (A1–A3) Achiral in situ spectra. (B1–B3) Chiral in situ spectra. Letters a–c represent different probing areas at the interface, and the same letter in sspn> and pspn> spectra represents the same area. The highest peak intensities are scaled to the same value for the convenience of comparison. (C1, C2) Time-dependent achiral (sspn>) C–H stretch spectra showing the reassembly of LK7β at the interface in two independent experiments. At t = 0, <span class="Chemical">NaOH was added and the pH was brought back to ∼7.4 to trigger the self-assembly process. Reprinted with permission from ref (25). Copyright 2013, American Chemical Society.

This work exhibits the ability of chiral C–H stretch SFG signals to reveal the kinetics of self-assembly of peptide at interfaces. This study implies that the chiral C–H SFG signals can also be a promising tool for probing a wide range of macromolecules other than proteins abundant in C–H groups at interfaces, including DNA, RNA, synthetic chiral polymers, and macromolecular structures. The study also implies that the chiral C–H stretch SFG signals can be useful in exploring kinetic mechanisms of these molecules at interfaces. The relatively high IR power and no interference from atmospheric water vapor and CO2 in the C–H stretch region contribute to a higher signal-to-noise ratio, shortening the acquisition time and thus improving the time resolution for kinetic measurements. The use of the chiral C–H stretch signal is expected to advance the applications of chiral SFG to address fundamental and engineering problems not only in biological sciences but also in polymer science, supramolecular chemistry, and material science.[42]

Chiral N–H for Probing Hydrogen/Deuterium Exchange in Proteins at the Interface

Hydrogen/deuterium (H/D) exchange is a chemical reaction where a covalently bonded hydrogen atom is replaced by deuterium and vice versa. The H/D exchange has long been useful for the characterization of protein structures and dynamics,[44] providing information about protein stability, protein folding, solvent exposure, and conformational changes.[45] Various techniques have been developed to probe the H/D exchange in the bulk, including NMR,[46] mass spectrometry,[47] and FTIR.[48] However, there is a still a dearth of in-depth studies on protein H/D exchange at interfaces, mostly due to technical challenges of acquiring surface-selective signals. In addition, because of the close proximity in frequency of the N–H (or N–D) and O–H (or O–D) stretches, the N–H (or N–D) stretch signal from protein backbones is often overwhelmed by the broad O–H (or O–D) stretch vibrational band of water (Figure ),[22,26] making it impossible to monitor H/D exchange by probing the N–H or N–D stretch signal using conventional (achiral) SFG. In contrast, chiral SFG can probe H/D exchange without interference from the O–H (or O–D) water background because chiral SFG is sensitive to the N–H stretch of the protein backbone in chiral secondary structures but insensitive to achiral water structures at interfaces.[22,26] In the study described below, we used the chiral N–H/N–D SFG signal to probe the kinetics of H/D exchange of proteins at interfaces.

Figure 6

Achiral SFG spectra in the N–H stretch region are usually broad and featureless due to the interference of the O–H stretch in water molecules. Reprinted with permission from ref (26). Copyright 2014, American Chemical Society.

Achiral SFG spectra in the N–H stretch region are usually broad and featureless due to the interference of the O–H stretch in <span class="Chemical">water molecules. Reprinted with permission from ref (26). Copyright 2014, American Chemical Society.

In the study, we used LK7β as a model system for the H/D exchange study. We first obtained the chiral amide I spectra at the air/<span class="Chemical">H2O and air/D2O interfaces (Figure )[24] and confirmed the formation of antiparallel β-sheets with the peaks at ∼1619 cm–1 (B2 mode) and ∼1690 cm–1 (B1 mode).[16,17] Then, we switched the IR frequency to obtain the chiral SFG spectra in the N–H stretch and the N–D stretch regions (Figure ).[24] In the N–H stretch region, we observed a major peak at 3268 cm–1 and a shoulder at 3178 cm–1. In the N–D stretch region, we observed a major peak at 2410 cm–1 and a shoulder at 2470 cm–1. On the basis of normal-mode analysis from ab initio computations[24] and previous empirical results,[49] we assigned the major peaks to the N–H (3268 cm–1) and N–D (2410 cm–1) stretches of the peptide backbone. Here, we eliminate the possibility of contribution from the lysine side chains in LK7β to the two major peaks, mainly based on the analyses of kinetics, peak positions, and ionic states for −NH3+.[24] We also attributed the 3178 cm–1 shoulder to the combination of amide I and amide II vibrations and proposed the 2470 cm–1 shoulder as resulting from the combination band of the C–N stretch and the N–D bend.[24]

Figure 7

Chiral SFG spectra of LK7β (A) in the N–H stretch region at the air/H2O interface and (B) in the N–D stretch region at the air/D2O interface. Reprinted with permission from ref (24). Copyright 2013, American Chemical Society.

Chiral SFG spectra of LK7β (A) in the N–H stretch region at the air/<span class="Chemical">H2O interface and (B) in the N–D stretch region at the air/<span class="Chemical">D2O interface. Reprinted with permission from ref (24). Copyright 2013, American Chemical Society.

We then monitored the H/D exchange kinetics of LK7β at the air/water interface in real time, which was initiated by adding <span class="Chemical">D2O (or H2O) to the solution of LK7β prepared with H2O (or D2O) (Figure A–F).[24]Figure A,B shows a gradual buildup of N–H signal due to the replacement of N–D by N–H. The time for the exchange process is shortened from ∼11 min (Figure A) to ∼7 min (Figure B) when the D2O/H2O ratio changes from 4:1 to 2:1. Similarly, the buildup of N–D signal due to the replacement of N–H by N–D is also accelerated when the D2O/H2O ratio changes from 4:1 to 2:1. The exchange time decreases from ∼53 min (Figure C) to ∼36 min (Figure D). Notably, the rate of exchange from N–D to N–H is faster than that from N–H to N–D for ratios of 2:1 and 4:1 (Figure E,F). The results suggest that the rate-determining step of H/D exchange in LK7β is the breaking of the O–H or O–D bond (Figure G) with the following argument. Because the O–D stretch has a lower zero-point energy than the O–H stretch, it requires higher energy to break the O–D bond in the exchange of N–H for N–D than to break the O–H bond in the exchange of N–D for N–H. The higher energy barrier therefore leads to a slower rate. This also agrees with previous studies of H/D exchange in crystalline α-cyclodextrin by Ribeiro-Claro’s group, which conclude that the N–D to N–H exchange is 7–10 times faster than the N–H to N–D exchange due to the rate-limiting step of breaking the water O–H or O–D bond.[50]

Figure 8

Kinetics of H/D exchange in LK7β at the air/water interface. Time-dependent N–D stretch spectra of LK7β at the air/H2O interface upon addition of (A) H2O and (B) D2O in a ratio of D2O/H2O equal to 4:1. (C) Time dependence of the SFG field in D-to-H exchange with a 4:1 (red) ratio and H-to-D exchange with a 4:1 (blue) ratio. (D–F) Results obtained at the ratio of 2:1. (G) Explanation of the difference in H-to-D and D-to-H exchange kinetics. (A–F) Reprinted with permission from ref (24). Copyright 2013, American Chemical Society.

This work shows the power of chiral SFG to probe the protein kinetic processes at interfaces hardly accessible to conventional surface characterization techniques. Without the <span class="Chemical">water background in the N–H stretch region, the proton exchange kinetics at the interface can be readily monitored by chiral SFG without isotopic labeling. Kinetic studies of H/D exchange in proteins using chiral SFG also hold promise in revealing molecular mechanisms of various biological processes at interfaces, such as the solvent accessibility of proteins in membranes, protein-mediated proton transfer across membranes, and intermolecular and intramolecular <span class="Chemical">hydrogen-bonding interactions in membrane proteins.[51]

Chiral Amide I and N–H Stretches for Probing Protein Folding and Misfolding at the Interface

As shown in the previous section, chiral SFG can selectively probe the N–H stretch signal from protein backbones, without the interference from the water background and side chains. We further showed that the chiral N–H stretch in combination with the chiral <span class="Chemical">amide I signals can be used to characterize protein secondary structures at interface. We took in situ chiral SFG spectra of proteins with known secondary structures at the air/water interface (Figures and 10).[22,26,52]Table shows a summary of the observations. A comparison of spectra for α-helix and disordered structures highlights the advantage of chiral SFG in distinguishing protein secondary structures. As probed by conventional vibrational spectroscopy, both α-helix and disordered structures exhibit amide I vibrational bands that overlap at ∼1654 cm–1. In the applications of chiral SFG, while it is difficult to distinguish the two types of structures by the amide I spectra alone, the prominent chiral N–H stretch peak for α-helices but not for disordered structures can clearly distinguish these two structures (Figure and Table ).[22,26] Therefore, chiral amide I and N–H stretch SFG signals together can be used as optical signatures to distinguish secondary structures that conventional vibrational spectroscopy may find difficult to differentiate.

Figure 9

Chiral (psp) N–H stretch and amide I SFG spectra for a series of model proteins and peptides with schematic structures. Reprinted with permission from ref (26). Copyright 2014, American Chemical Society.

Figure 10

Chiral (psp) SFG spectra of hIAPP (a model peptide with parallel β-sheet secondary structure) at different interfaces. (A) Tilted hIAPP yields prominent chiral amide I but no chiral N–H stretch signal. (B) hIAPP lying flat at the interface yields prominent chiral amide I and N–H stretch signals. Color codes: green, hydrophilic residues; white, hydrophobic residues; and blue, positively charged arginine residues. Reprinted with permission from ref (52). Copyright 2015, American Chemical Society.

Table 1

Chiral SFG Spectral Features in the Amide I and N–H Stretch Regions for Different Secondary Structures at Interfaces as Summarized from Figures and 10

model peptide/proteins	secondary structures	amide I signal	N–H stretch signal
tachyplesin I, LK7β	antiparallel β-sheet	yes	yes
rhodopsin, pHLIP, LK14α	α-helix	no	yes (low cm–1)
alamethicin	310-helix	no	yes (high cm–1)
rIAPP	disordered	no	no
hIAPP (lipid/water)	parallel β-sheet (tilted)	yes (with shoulder at 1660–1670 cm–1)	no
hIAPP (glass slide)	parallel β-sheet (lying flat)	yes (with shoulder at 1660–1670 cm–1)	yes

Chiral (psp) N–H stretch and <span class="Chemical">amide I SFG spectra for a series of model proteins and peptides with schematic structures. Reprinted with permission from ref (26). Copyright 2014, American Chemical Society.

We used the chiral SFG vibrational signatures to probe the kinetics of amyloid aggregation at interfaces. <span class="Disease">Amyloid aggregation is associated with various neurodegenerative diseases. The latest clinical studies showed that lipids are reliable biomarkers for detecting amyloid diseases.[53] Also, interactions with the membrane can catalyze aggregation, and small prefibrillar aggregates can perturb the cell membrane and cause cytotoxicity.[54] Thus, a detailed molecular understanding of the role of lipid membranes can potentially change the landscape of future approaches for developing treatments. Nonetheless, most molecular studies of amyloid aggregation have been performed in the solution phase. Thus, there is a gap of knowledge on the early stages of amyloid aggregation on membrane surfaces. This gap is mostly due to the lack of label-free methods that can detect in situ and real-time protein conformational changes on surfaces without the interference of signals from water and proteins in solution. This knowledge gap can be bridged only by the innovation of technology that is sensitive to both protein structures and interfaces. We used the chiral SFG method to monitor early-stage aggregation at the air/water interface in the presence of a lipid monolayer for human islet amyloid polypeptide (hIAPP), a 37-residue peptide for which the aggregation on membrane surface is associated with the onset of type II diabetes.[4]

We performed two experiments in which we took the chiral SFG spectra of hIAPP at the air/<span class="Chemical">water interface in the presence and absence of negatively charged dipalmitoylphosphatidylglycerol (DPPG). In the first experiment, we added hIAPP at the air/water interface and took the chiral SFG spectra in the amide I and N–H stretch regions (right in Figure ).[22] Within 10 h, the spectra showed no appreciable signal in either region, indicating that hIAPP is disordered. In the second experiment, we initially added the same amount of hIAPP at the air/water interface. Once the adsorption of hIAPP reached equilibrium, we added DPPG at the interface to form a monolayer and took the chiral amide I and N–H stretch spectra after ∼10 h (left, Figure ).[22] The N–H stretch spectrum shows no chiral SFG signal, while the amide I spectrum shows a major peak at ∼1622 cm–1 and a shoulder at ∼1660 cm–1. The two peaks in the amide I region corresponds to the amide I A mode (1660 cm–1) and the amide I B mode (1622 cm–1) of parallel β-sheets, respectively, indicating the formation of β-sheet-rich hIAPP aggregates at the interface.

Figure 11

Scheme for the in situ chiral SFG experiments for hIAPP at the air/water interface. Spectra in the absence of the DPPG lipid monolayer (right) and in the presence of the DPPG lipid monolayer (left) are shown for the amide I and the N–H stretch regions 10 h after the addition of hIAPP. Reprinted with permission from ref (22). Copyright 2011, American Chemical Society.

To further examine the pathway through which hIAPP folds into the β-sheet-rich aggregates, we monitored in situ and in real time the chiral SFG signal in the amide I and N–H stretch regions. Figure A[22] shows that after the addition of DPPG to initiate the aggregation, the chiral N–H signal increases and reaches its maximum at around 3 h while the chiral amide I signal remains undetectable. After 3 h, the chiral N–H signal fades away but the chiral amide I signal gradually builds up. This transition persists until ∼10 h, when the chiral N–H signal completely vanishes and the chiral amide I signal reaches a maximum. To confirm these observations, we repeated the experiments three times and obtained three sets of data exhibiting similar kinetics for the chiral SFG signal in the amide I and N–H stretch regions (Figure B).[22] Since the chiral SFG signal close to 3300 cm–1 is characteristic of an α-helix, the observations leads to the conclusion that hIAPP first forms an α-helical intermediate and then aggregates into parallel β-sheets (Figure C).[22] This pathway agrees with a previously proposed working model for hIAPP aggregating on membrane surfaces.[55]

Figure 12

Aggregation of hIAPP. (A) The time-dependent chiral SFG spectra in the vibrational regions of the N–H stretch (left) and amide I (right) after the addition of DPPG. (B) Intensity of the N–H stretch and amide I signals as a function of time. Results of triplicate experiments are shown. (C) Aggregation model of hIAPP on a membrane surface as observed in the SFG experiments: adsorption as a random coil leads to the formation of α-helical intermediates, which are converted to β-sheet aggregates. (D) Time-dependent achiral SFG spectra in the amide I region. (A–C) Reprinted with permission from ref (22). Copyright 2011, American Chemical Society.

Aggregation of hIAPP. (A) The time-dependent chiral SFG spectra in the vibrational regions of the N–H stretch (left) and <span class="Chemical">amide I (right) after the addition of DPPG. (B) Intensity of the N–H stretch and amide I signals as a function of time. Results of triplicate experiments are shown. (C) Aggregation model of hIAPP on a membrane surface as observed in the SFG experiments: adsorption as a random coil leads to the formation of α-helical intermediates, which are converted to β-sheet aggregates. (D) Time-dependent achiral SFG spectra in the amide I region. (A–C) Reprinted with permission from ref (22). Copyright 2011, American Chemical Society.

As a control to confirm the conformational changes of proteins at the interface, we also monitored the achiral SFG spectra in the amide I region in real time (Figure D).[23] Prior to the addition of DPPG, the achiral spectrum shows a peak at ∼1650 cm–1, which can be assigned to disordered structures according to chiral experimental results. Upon addition of DPPG to initiate the aggregation, the achiral SFG spectrum shows a C=O stretch peak at ∼1740 cm–1 due to DPPG. In the subsequent 10 h, the achiral amide spectrum exhibits a gradual shift of peak position to ∼1660 cm–1 accompanied by a slight increase in peak intensity. This observation indicates conformational changes at the interface, supporting the conclusion from chiral SFG experiments. Nevertheless, the 1650 cm–1 achiral SFG peak before the start of aggregation has ambiguous assignments. It can be assigned to either disordered structures or α-helices. Therefore, without spectral information from other vibrational regions, such as the N–H stretch, it is difficult to draw a definitive conclusion about the secondary structures of proteins during the aggregation of hIAPP using achiral SFG alone.

This work has illustrated the application of chiral SFG to probe the aggregation kinetics of amyloid proteins at the membrane surface. With its selectivity to chirality, chiral SFG can provide background-free information without interference from achiral molecules. Due to the silent C=O stretch signal from achiral lipid headgroups, characteristic vibrational signals of proteins in the amide I region can be detected without interference from lipid signals, rendering its application to the study of protein aggregation on lipid membrane surfaces. In addition, the absence of the O–H stretch signal from the water solvent furnishes the opportunity to use the N–H stretch from protein backbones as an extra handle to track real-time conformational changes of proteins at interfaces. The series of chiral SFG spectra in both the amide I and the N–H stretch regions for proteins can serve as vibrational signatures to distinguish secondary structures in proteins. Furthermore, our study also implies the application of chiral SFG in studying the effect of inhibitors or drug candidates on the aggregation of amyloid proteins at membrane surfaces, providing direct information to evaluate drug efficacy and guide the rational design of drugs for amyloid diseases. This method can be generally applied to study early stages of misfolding of other amyloidogenic proteins, such as amyloid-β, α-synuclein, and huntingtin, which are associated with neurodegenrative Alzheimer’s, Parkinson’s, and Huntington diseases, respectively.[4] Our work also shows the promise of chiral SFG for solving problems related to protein folding on 2D surfaces and the functions of intrinsically disordered proteins that change conformation upon interaction with interfaces.

Summary and Outlook

We have discussed the theoretical background of chiral SFG and the development of broad-bandwidth SFG spectrometers for kinetic studies at interfaces. We have also reviewed three applications of chiral SFG spectroscopy on protein kinetics at interfaces: (1) the C–H stretch from protein side chains for monitoring protein assembly processes, (2) the N–H stretch for observing interfacial proton exchange on protein backbones, and (3) the amide I and the N–H stretches for probing the kinetics of the early stages of amyloid protein aggregation at the lipid/water interfaces. These kinetic studies have demonstrated the capacity of chiral SFG as a background-free method for studying protein kinetics without the use of spectroscopic labels in situ and in real time at interfaces.

The chiral SFG method can be further developed to widen its applications to tackle problems related to more complex molecular systems. In terms of instrumentation, the signal-to-noise ratio is a limiting factor in the temporal resolution of kinetic studies and can be improved by the implementation of several techniques, including heterodyne detection,[39,40] a total internal reflection setup,[56] and doubly resonant SFG,[57] although the surface specificity of doubly resonant SFG remains controversial.[26] In terms of data interpretation, various methods have been developed to calculate hyperpolarizabilities for biological macromolecules that allow quantitative spectral interpretation for extracting structural and orientational information.[58−60] Moreover, the autocorrelation of second-order optical signals should, in principle, further enable dynamic studies of biomacromolecules at interfaces, as demonstrated by Conboy’s group[61] and Eisenthal’s group.[62] In terms of sample preparation, instead of simple peptides synthesized by solid-state methods, complex proteins can be expressed and purified for SFG studies[22] to address biologically relevant problem beyond simple model systems. In addition, isotopic labeling can also be implemented to aid spectral assignments to resolve ambiguity in the interpretations of vibrational spectra for kinetic studies.[63]

The above developments can help chiral SFG to address a wider range of problems related to the kinetics of proteins at interfaces. For instance, chiral SFG can be used to study protein folding, protein functions, and protein–ligand interactions at interfaces such as membrane surfaces. The use of peptide backbone N–H/N–D stretches for probing proton exchange can be applied to investigate <span class="Chemical">hydrogen-bonding interactions, solvent accessibility, and protein dynamics at interfaces. Moreover, chiral SFG should be a promising tool to reveal molecular mechanisms of intrinsically <span class="Disease">disordered proteins that change conformation drastically upon binding to their interacting partners, e.g., various biological interfaces. It can also be used to study amphiphilic proteins, such as constituent proteins in bacterial biofilm extracellular matrices for bacterial growth into communities on surfaces.[64] Moreover, the use of dynamic chiral SFG signals can potentially inspire research into problems of the immobilization of proteins on solid substrates,[65] immunological responses of antibodies binding with its antigen,[66] and the biocompatibility of implanted medical devices.[67]

Besides proteins, chiral SFG can also be applied to study the kinetics of other important chiral biomacromolecules and synthetic molecular systems. For instance, chiral SFG has revealed the structures of <span class="Chemical">oligonucleotides tethered at interfaces.[33] Moreover, the characterization of chiral porphyrin aggregates within the hydrophobic region of <span class="Chemical">lipid bilayers in plant photosynthetic processes has provided insights into the relationship between their structures/orientation and remarkable electronic properties in the light-harvesting system.[68] By monitoring the structural changes of polysaccharides on plant cell wall matrices, chiral SFG can be used as a potential tool to understand their influence on the formation of cellulose microfibrils.[69] Chiral SFG is also a useful tool for probing the self-assembly of chiral polymers/biopolymers/biocolloids at interfaces,[70] which provides quality control over the growth of materials through real-time structural characterization. Similar approaches can also be applied to chiral liquids,[71] nanoparticles,[72] and biomimetic materials[73] at interfaces. With a fundamental understanding of interfacial kinetics, chiral SFG can potentially identify new problems and inspire new answers to the questions posed in electronics and engineering, such as the developments of biosensors,[74] DNA microarrays,[75] biofuel cells,[7] semiconductor chips,[76] and surface plasmonic techniques.[77] Hence, further development of the chiral SFG method in conjunction with other characterization methods can introduce opportunities to solve problems in the interest of the greater community of surface science.