Probing local changes to α-helical structures with 2D IR spectroscopy and isotope labeling

Abstract

α-Helical secondary structures impart specific mechanical and physiochemical properties to peptides and proteins, enabling them to perform a vast array of molecular tasks ranging from membrane insertion to molecular allostery. Loss of α-helical content in specific regions can inhibit native protein function or induce new, potentially toxic, biological activities. Thus, identifying specific residues that exhibit loss or gain of helicity is critical for understanding the molecular basis of function. Two-dimensional infrared (2D IR) spectroscopy coupled with isotope labeling is capable of capturing detailed structural changes in polypeptides. Yet, questions remain regarding the inherent sensitivity of isotope-labeled modes to local changes in α-helicity, such as terminal fraying; the origin of spectral shifts (hydrogen-bonding versus vibrational coupling); and the ability to definitively detect coupled isotopic signals in the presence of overlapping side chains. Here, we address each of these points individually by characterizing a short, model α-helix (DPAEAAKAAAGR-NH₂) with 2D IR and isotope labeling. These results demonstrate that pairs of ¹³C¹⁸O probes placed three residues apart can detect subtle structural changes and variations along the length of the model peptide as the α-helicity is systematically tuned. Comparison of singly and doubly labeled peptides affirm that frequency shifts arise primarily from hydrogen-bonding, while vibrational coupling between paired isotopes leads to increased peak areas that can be clearly differentiated from underlying side-chain modes or uncoupled isotope labels not participating in helical structures. These results demonstrate that 2D IR in tandem with i,i+3 isotope-labeling schemes can capture residue-specific molecular interactions within a single turn of an α-helix.

Significance

α-Helices are a common secondary structure responsible for a host of diverse biological functions. However, α-helices can denature or exhibit structural transformations that trigger loss of activity or induce new, toxic behaviors. Therefore, revealing specific regions of α-helices that undergo structural changes is crucial to better understand their structure-function dependence. Here, we implement a sensitive isotopic labeling scheme at various positions in a model α-helix to probe local changes in helicity with 2D IR spectroscopy. These findings demonstrate the sensitivity of 2D IR spectroscopy with isotope labeling to local molecular interactions within an α-helical peptide. The methods established here can be applied broadly to α-helical systems to correlate detailed structural changes to functional alterations.

Introduction

Secondary structures impart unique mechanical and physiochemical properties to proteins, allowing them to act as efficient molecular machines (1,2,3). In particular, α-helices are the most common secondary structural motif adopted by polypeptides (4). This secondary structure is critical for a wide array of molecular functions, including membrane insertion (5,6,7), protein-protein and protein-DNA interactions (8,9,10), and intra/intermolecular allostery (11,12,13). These diverse biological activities have raised interest in α-helical peptides as drug candidates and novel peptide nanomaterials (1,2,14,15). However, denaturation or transformation of α-helices into other secondary structures may lead to loss of native function or gain of toxic function (16,17,18). For example, α-helical denaturation, whether through glycation (19), sequence mutations (20), or adsorption onto nanoparticle surfaces (21,22,23,24), has been directly correlated with loss of enzymatic activities. Thus, the crucial relationship between highly abundant α-helical motifs in peptides and proteins and their biological function makes revealing structural changes of great importance.

Two-dimensional infrared (2D IR) spectroscopy can provide exquisite insight into protein secondary structure (25,26,27,28,29,30,31), including details that are inaccessible via 1D methods such as Fourier transform infrared spectroscopy (FTIR) or transient absorption spectroscopy (25,28,29,30,32,33). By adding another frequency dimension, spectral resolution is improved, inhomogeneous and homogeneous contributions to the linewidths are decoupled (32,33,34,35), and crosspeaks are revealed that can be used to determine the relative distance and orientation between coupled vibrational modes (26,32,33,36,37,38,39). Finally, signal strength scales as |μ|⁴ in a 2D IR spectrum but only |μ|² in an FTIR spectrum. This signal enhancement provides 2D IR with improved sensitivity compared with FTIR and thus enables detection of peaks that are unresolved in linear spectra (28), such as spectral features arising from a single isotope-labeled residue. Furthermore, FTIR is largely insensitive to changes in transition dipole strength (TDS) caused by vibrational coupling (40), but 2D IR is highly sensitive to such changes and the integrated peak areas increase as coupling increases. Thus, comparison of 2D IR peak areas can report on the relative amount of coupling between oscillators and the ratio of 2D and linear IR signals can be used to calculate absolute TDS. Such analysis can prove invaluable as studies have demonstrated that the TDS of a vibrational mode is often a more sensitive measure of coupling than mode frequency (41,42,43).

To date, most 2D IR studies of peptides and proteins have examined the amide I′ transition (which comprises primarily backbone carbonyl stretching) because it is sensitive to secondary structure (32,33,36). When coupled with isotope labeling methods, 2D IR is capable of single-residue structural resolution (34,35,37,38,39,44,45,46,47,48). Although the vast majority of these studies have focused on β-sheet peptides, several reports have employed 2D IR and isotope labeling to characterize helical peptides (49,50,51,52). It is important to note that both hydrogen-bonding interactions and vibrational coupling affect the observed frequency of the amide I′ vibrational transition (32,38,53,54,55,56). Hydrogen-bonding weakens the carbonyl double bond, lowering its vibrational frequency relative to carbonyls that are not hydrogen bonded (53,57). On the other hand, vibrational coupling occurs via either mechanical (through-bond) or electrostatic (through-space) interactions (57). Peptides experience both kinds of vibrational coupling, with the strongest mechanical interactions occurring between adjacent residues (46,50,56). When peptides exhibit a higher degree of order, patterns of vibrational couplings between individual transition dipoles lead to delocalization of vibrational modes over the extended structure, giving rise to distinctive spectral signatures (frequencies, intensities, lineshapes) in an IR spectrum (32,33).

There is some debate on which type of underlying molecular interactions dominate the spectral features observed in 2D IR spectra of isotope labeled in α-helices. Fang and co-workers employed mixed ¹³C and ¹³C¹⁸O probes spaced one, two, or three residues apart in a stable α-helix to determine how the relative positions of the isotope labels influenced vibrational signatures (50). Experimental spectra demonstrated varying shifts of the labeled amide I′ modes and differing intensities for the crosspeaks between the ¹³C and ¹³C¹⁸O modes depending on the relative label positions. When supplemented with extensive calculations, these results yielded the interresidue coupling constants (β) in an α-helix, with the strongest vibrational coupling occurring between adjacent residues (β₁₂ = 8.5 ± 1.8 cm⁻¹) followed by the i,i+3 positions (β₁₄ = −6.6 ± 0.8 cm⁻¹). They attributed coupling arising from adjacent residues primarily to through-bond vibrational coupling, whereas electrostatic coupling between the i and i+3 residues, which corresponds to the α-helical repeat length of 3.6 residues, should be unique to this particular secondary structure.

A subsequent study by Backus et al. utilized 2D IR with ¹³C¹⁸O isotope labeling to probe local molecular interactions within a photoswitchable α-helix (51). This study revealed that even single ¹³C¹⁸O isotope labels experienced significant frequency shifts as the peptide switched between α-helical and denatured states. As vibrational coupling between unlabeled ¹²C¹⁶O and single ¹³C¹⁸O amide I′ modes has a negligible effect on their respective frequencies due to the large energy difference, they attributed frequency shifts to intramolecular hydrogen-bonding between the i and i+4 residues. An entirely novel labeling scheme reported by Maekawa et al., which utilized ¹⁵N- and ¹³C¹⁸O-labeled residues, led to the formation of crosspeaks between the amide I′ and IIʹ modes in 2D IR spectra of a model 3₁₀-helix (58). Such crosspeaks indicate vibrational coupling between the labeled residues. While this study was performed on a 3₁₀-helix rather than an α-helix, and thus the repeat length along the helical axis differs, the conclusions should prove general to all helical peptides.

Despite many recent advances in this field, there are currently limitations to probing the structure and dynamics of α-helical motifs with 2D IR and isotopic labeling methods. One major drawback is side-chain interference in the isotopic region (∼1550–1600 cm⁻¹) (59). For example, aspartate (ε ∼ 820 M⁻¹ cm⁻¹), glutamate (ε ∼ 830 M⁻¹ cm⁻¹), and arginine (ε ∼ 500 M⁻¹ cm⁻¹) side chains contain IR-active functional groups that absorb between ∼1560 and 1610 cm⁻¹ in deuterated solvent (59). While the extent of side-chain IR absorption in polypeptides is dependent upon many factors, including solvent exposure, protonation state, and chain length, these molecular vibrations coincide with the ¹³C¹⁸O-edited amide I′ transition (59,60). Some previous 2D IR studies have circumvented this complication by mutating native residues to those that do not contain side chains with overlapping vibrational signatures, which was successfully demonstrated in a study of the influenza A M2 proton channel (61). However, the need to employ mutants limits our understanding of the structure and dynamics of the native peptide or protein. Another drawback of probing α-helical structures with isotopes is that the effects of vibrational coupling in α-helices are inherently weaker compared with the more commonly studied extended β-sheet structures (25,38,41,46,50,56). The local amide backbone modes in β-sheets experience both intra- and intermolecular couplings. The nearest neighbor (intrastrand) couplings are positive in sign but relatively small, so the large, negative coupling between strands (interstrand) dominates the amide I′ spectral shift (25,38,56). In addition, extended β-sheet structures contain large numbers of oscillators, leading to greater delocalization of the vibrational modes and thus greater changes in spectral features. In contrast, α-helices are typically smaller and exhibit competing vibrational couplings as large, positive couplings between adjacent residues oppose the large, negative couplings between turns, resulting in smaller frequency shifts (41,46,50). Furthermore, as α-helices predominately experience intramolecular couplings, unless they are involved in higher-order structures, such as the M2 proton channel from influenza A (61), multiple labels are required to detect these intramolecular contacts and delocalization is limited to the few labeled residues.

In this study, we aim to resolve the remaining challenges for studying α-helical structures with 2D IR and isotope labeling by addressing the following questions: How sensitive are isotope-labeled modes to subtle changes in α-helicity, such as when an α-helix begins to unravel slightly? Is hydrogen-bonding or vibrational coupling responsible for the characteristic signatures of α-helical isotopologues? Can we definitively distinguish coupled isotopic signals from overlapping side-chain modes? To answer these questions, we applied 2D IR spectroscopy to study systematic changes in a model peptide with a solvent-dependent secondary structure, which provided a way to precisely control peptide helicity (62). Both single and paired ¹³C¹⁸O probes were used to monitor the structural change from disordered to α-helical in the model system. The findings presented here suggest that both vibrational coupling and hydrogen-bonding contribute to the spectral features, with hydrogen-bonding being the major determinant of the red shift in mode frequency while vibrational coupling leads to greatly enhanced mode intensity as the organized structure forms. The resulting change in integrated peak area intensity provides another avenue to differentiate coupled isotopes from overlapping side-chain modes.

Materials and methods

Materials

All reagents were used as purchased, excluding the modifications made to L-alanine (1-¹³C) and L-glycine (1-¹³C), as described in the section on preparing ¹³C¹⁸O-labeled amino acids. Fmoc-protected amino acids, Oxyma, and Rink Amide ProTide resin (high loading, 0.70 meq/g) were purchased from CEM (Matthews, NC). Isotope reagents, including L-alanine (1-¹³C, 99%), L-glycine (1-¹³C, 99%), and ¹⁸O-H₂O (98% enriched), were purchased from Cambridge Isotope Laboratories (Tewksbury, MA). Piperazine and N-methylacetamide (NMA) were purchased from Alfa Aesar (Tewksbury, MA). N,N″-Diisopropylcarbodiimide (DIC), N-methyl-2-pyrrolidone (NMP), and N-(9-fluorenylmethoxycarbonyloxy)succinimide (Fmoc-Osu) were purchased from Oakwood Chemical (Estill, SC). Sodium deuteroxide (NaOD) (30 wt %), deuterium chloride (DCl) (20 wt %), triisopropyl silane (TIPS), and 2,2,2-trifluoroethanol (TFE) (99.8% extra pure) were purchased from Acros Organics (Geel, Antwerp, Belgium). Sodium bicarbonate (NaHCO₃), potassium bisulfate (KHSO₄), dimethylformamide (DMF), diethyl ether, acetone, acetonitrile, hydrochloric acid (12.1 N), and methanol were purchased from Fisher (Saint Clair Shores, MI). All other reagents were purchased from Sigma Aldrich (St. Louis, MO).

Preparation of ¹³C¹⁸O-labeled amino acids

L-Alanine-1-¹³C and L-glycine-1-¹³C were individually Fmoc protected using a 1:1:1 mol ratio mixture of amino acid, NaHCO₃, and Fmoc-OSu dissolved in a 1:1 mixture of water and acetone. The reaction mixture was stirred at room temperature for 24 h and quenched with 2 M KHSO₄ until the solution reached pH 2 to precipitate the Fmoc-protected amino acid. The product was washed sparingly with ice-cold deionized water by vacuum filtration to remove residual salts present, then lyophilized overnight to remove residual water.

Fmoc-alanine-1-¹³C¹⁸O¹⁸O and Fmoc-glycine-1-¹³C¹⁸O¹⁸O were individually prepared by an acid-catalyzed ¹⁸O-exchange using Schlenk techniques under a nitrogen atmosphere, as reported previously (63,64). In brief, 1 g of Fmoc-1-¹³C amino acid was dissolved in a mixture of 8 mL dioxane and 4 mL 4 N HCl in dioxane. The reaction mixture was heated to reflux at 150°C before injecting 1 mL of ¹⁸O-enriched water. After allowing the reaction to reflux for ∼4 h, the solvent was removed under vacuum. The ¹⁸O-exchange reaction was repeated twice more to achieve >90% ¹⁸O-labeling efficiency, as determined by electrospray-ionization mass spectrometry (ESI-MS) (Orbitrap XL Penn, Thermo Fisher Scientific, Wilmington, MA). Before characterization, the amino acid was extracted in cold diethyl ether to remove undesired organic salts formed during the ¹⁸O-exchange reaction, then dried overnight. Fmoc-alanine-1-¹³C¹⁸O¹⁸O and Fmoc-glycine-1-¹³C¹⁸O¹⁸O were found to have 96 and 95% ¹⁸O-labeling efficiency, respectively.

Solid-phase peptide synthesis and purification

The 12-residue model α-helical peptide (MAHP, DPAEAAKAAAGR-NH₂) was obtained through solid-phase peptide synthesis using Fmoc chemistry with piperazine deprotection and DIC/Oxyma activation on a Liberty Blue microwave-assisted peptide synthesizer (CEM). Rink Amide ProTide resin was used as the solid support to yield an amidated C-terminus. Peptide cleavage from the resin and side-chain-protecting group removal was achieved with a cleavage cocktail of 95% TFA, 2.5% TIPS, and 2.5% deionized H₂O. After 3 h, the resin was filtered from the peptide solution and the peptide was precipitated in cold diethyl ether. The crude peptide was dissolved in high-purity H₂O (∼2 mg/mL), filtered, and then purified by reversed-phase high-performance liquid chromatography (Ultimate 3000, Thermo Fisher Scientific) using an XBridge BEH preparative column (Waters Corporation, Milford, MA). A binary gradient of pure water (Solvent A) to 90% acetonitrile in water (Solvent B) with 0.045% HCl (v/v) was implemented for purification. The gradient was varied from 0 to 40% Solvent B over 20 min; the peptide eluted from the column at ∼11 min, as indicated by ultraviolet absorption at 214 nm. The eluted fractions were lyophilized and stored at −20°C. Peptide molecular weight and purity were confirmed by ESI-MS. Peptide isotopologues were synthesized and purified in the same manner, substituting ¹³C¹⁸O-enriched Fmoc-protected amino acids at appropriate positions (Ala3, Ala5, Ala6, Ala8, and Gly11) during peptide synthesis. ESI-MS confirmed that doubly labeled peptide samples (A3A6, A5A8, and A8G11) contained ∼92% peptide with two ¹³C¹⁸O labels and ∼8% peptide with mixed ¹³C¹⁸O/¹³C¹⁶O labels. Singly labeled peptide samples (A5) contained 96% peptide with one ¹³C¹⁸O label and 4% with one ¹³C¹⁶O label.

Sample preparation

All purified peptide samples were prepared in deuterated solvents to ensure strong water bending modes did not overlap with the amide I′ mode in 2D IR measurements. Peptide solutions (5 mM) were prepared in either pure D₂O or 40% TFE in D₂O by volume. The solution pH was adjusted with NaOD and DCl to either pH 3 or 10.

NMA and NMP were used as model compounds for 2D IR studies of the solvatochromic shift induced by TFE. Both molecules were prepared in either pure D₂O or 40% TFE in D₂O by volume at final concentrations of 68 mM for NMA and 60 mM for NMP.

All samples were vortexed and incubated at room temperature for at least an hour before 2D IR measurements. For 2D IR measurements, ∼10 μL of solution was placed between two CaF₂ windows (Crystran, Poole, Dorset, UK) separated by a 50 μm Teflon spacer. For circular dichroism (CD) experiments, ∼40 μL of peptide solution was sandwiched between two quartz windows (Starna Cells, Atascadero, CA). A 0.1 mm cell pathlength was implemented in CD experiments to enable measurements at the same peptide concentration as 2D IR experiments.

2D IR spectroscopy

2D IR data collection and processing methods are described in detail elsewhere (64). In brief, a one-box ultrafast amplifier (Solstice Ace, Spectra-Physics, Santa Clara, CA) was used to produce 800 nm pulses (7 mJ, 1 kHz, 60 fs). Half of the beam was used to pump an optical parametric amplifier coupled with difference frequency generation (TOPAS-Prime, Spectra-Physics) and produce ultrafast mid-IR pulses (6.1 μm, 30 μJ, 1 kHz, 70 fs), which were directed into a commercial 2D IR spectrometer with pulse shaping (2DQuick IR, PhaseTech Spectroscopy, Madison, WI), which was purged with dry, compressed air. The time delay between the first and second pulses (t₁) was scanned from 0 to 2.54 ps in 23.8 fs steps, yielding a frequency resolution of ∼2.7 cm⁻¹ for the pump, while the delay between the second and third pulses (t₂) was kept constant at 0 ps for all measurements. All spectra were collected using a parallel (ZZZZ) beam polarization. A monochromator (Princeton Instruments, Trenton, NJ) was used to disperse the signal onto a 128-pixel mercury cadmium telluride focal-plane array detector (PhaseTech Spectroscopy). The beam was centered at ∼1610 cm⁻¹ with an FWHM of 280 cm⁻¹and had even intensity across the spectral window of interest (1580–1650 cm⁻¹). Data were collecting using the QuickControl software provided by PhaseTech Spectroscopy and processed using custom scripts in MATLAB (The MathWorks, Natick, MA). The pulse shaper was calibrated using a frequency comb (64), while the monochromator was calibrated using the 1535, 1605, and 1711 cm⁻¹ absorption peaks of 4-nitrobenzaldehyde. Using a 30 groove/mm grating, the frequency resolution was ∼2 cm⁻¹ for the probe. The pump frequencies were determined by setting the diagonal line through the fundamental peaks (blue).

Electronic CD

CD measurements were made on a J-810 spectropolarimeter (Jasco, Easton, MD) purged with N₂ (g) to prevent optic damage. All measurements were conducted at 24°C. Spectra were collected from 190 to 260 nm in 1 nm increments at a scan rate of 100 nm per minute and a standard sensitivity of 100 mdeg. Three spectra were collected per sample, then averaged and smoothed in Microsoft Excel.

Results and discussion

Previous 2D IR studies of α-helices have focused on either stable helical structures or those that could be switched between completely helical and completely disordered states (49,50,51). In this study, we aim to identify and localize structural changes as the peptide sample gains or loses α-helicity. MAHP was selected as a model system based on reports that its helicity could be modulated gradually via pH (62). While disordered under acidic conditions, ionizable side chains deprotonate at higher pH and favorable intramolecular interactions (e.g., salt bridge formation between the E and K residues and deprotonation of the N-terminal D residue) promote and stabilize the α-helical structure of the peptide. Such solvent-dependent behavior enables precise control of the conformation of the peptide while examining isotopic spectral features at each point.

To confirm that the structure of MAHP changed sufficiently to serve as a model system for this study, CD spectra were obtained across a pH range of 3–10 and the α-helicity quantified using the 222 nm band (62,65). To calculate percent α-helicity, the experimental mean residue ellipticity (MRE) ([θ]) value must be compared with the maximum MRE (${\left[\theta \right]}_H$) possible for a perfect helix of the same length (Eq. 1).(65)

$$\%\mathrm{h}\mathrm{e}\mathrm{l}\mathrm{i}\mathrm{c}\mathrm{i}\mathrm{t}\mathrm{y}=\frac{\left[\theta \right]}{{\left[\theta \right]}_{\mathrm{H}}}\ast 100$$ (Equation 1)

The experimental MRE is calculated from the measured ellipticity ($\theta$) according to the following equation:

$$\left[\theta \right]\left(\frac{{\mathrm{deg}\ast \mathrm{c}\mathrm{m}}^2}{\mathrm{d}\mathrm{m}\mathrm{o}\mathrm{l}}\right)=\frac{\theta }{l\ast \mathrm{C}\ast \mathrm{N}}$$ (Equation 2)

where $\theta$ is measured in millidegrees, l is the pathlength in mm, C is the concentration in mol/L, and N is the number of residues (65). The theoretical maximum MRE value for a perfectly helical peptide with N residues can be calculated according to the following equation:

$${\left[\theta \right]}_{\mathrm{H}}=-\mathrm{40,000}\ast \left(1-\frac{\mathrm{x}}{\mathrm{N}}\right)+100\ast \mathrm{T}$$ (Equation 3)

where x is a chain-length-dependent constant used to correct for nonhydrogen-bonded carbonyls, and T is the temperature (^oC) (66). The constant x has been previously estimated to range from ∼0 to 3 and is inversely dependent upon chain length (66). MAHP is a relatively short peptide with 12 residues, thus x was estimated as 2.5 based on the CD unfolding curves from the study by Scholtz et al. (66). Taken together, Eq. 3 yielded a ${\left[\theta \right]}_H$ value of $-\mathrm{29,667}\frac{{\mathrm{deg}\ast \mathrm{c}\mathrm{m}}^2}{\mathrm{d}\mathrm{m}\mathrm{o}\mathrm{l}}$ at 24°C.

CD measurements confirmed a pH-dependent increase in α-helicity for MAHP (Fig. 1). According to above equations, MAHP exhibited ∼8% helicity at pH 3, indicating that the peptide is primarily disordered. Under this condition, isotopes placed in the center of the helix would be expected to experience negligible coupling. At pH 10, the helicity of the peptide was observed to increase to ∼25%. This observation could reflect two scenarios: 1) ∼25% of the peptides adopt a highly α-helical structure, or 2) the majority of peptides in the sample adopt a partially α-helical structure, with only three residues participating in just under a full helical turn. In either scenario, paired isotope labels would not reliably report coupling. To enhance the peptide’s helicity, a helix-stabilizing cosolvent, TFE, was introduced (67,68,69,70). At pH 10 with 40% TFE (v/v), the peptide exhibited ∼60% helical character. As before, this could reflect two different scenarios: 1) slightly over half of the ensemble adopts a highly α-helical structure, or 2) the majority of peptides in the sample adopt a primarily α-helical conformation, with seven to eight residues participating in approximately two helical turns. Regardless, under this condition, pairs of isotopic labels placed in the center of MAHP would be expected to participate in a well-organized α-helix in a majority of the peptides present in the sample. As such, the ¹³C¹⁸O amide I′ mode should exhibit spectral shifts arising from vibrational coupling and/or the formation of intramolecular hydrogen bonds. In contrast, one or both of the peptide termini likely remain frayed under these conditions and isotopic pairs placed in these regions would be expected to undergo minimal spectral changes. Thus, we anticipate that pairs of isotope labels placed at the center of the helix versus the termini should yield distinct signatures in 2D IR spectra.

Figure 1.

Modulating MAHP α-helicity with solvent conditions. CD measurements revealed an increase in average α-helicity as pH was increased from 3 to 10 (from 8.13 to 25.13%). The addition of 40% TFE (v/v) further enhanced peptide helicity (up to 61.48% at pH 10). To see this figure in color, go online.

While TFE increases α-helicity in MAHP, presumably by creating a more hydrophobic environment that promotes the formation of intramolecular hydrogen bonds (68,69,70,71,72), this environmental change can also result in solvatochromic shifts of IR frequencies (73,74,75). Therefore, before examining 2D IR spectral features that arise as MAHP transitions from a disordered to an α-helical structure, we first must decouple frequency shifts that arise from changes in solvent environment from those that arise from changes in peptide structure. To this end, we measured solvatochromic shifts in NMA and NMP. NMA is an amide-containing small molecule commonly used as peptide bond analog for IR studies (53,76). NMP is a small, cyclic compound that contains a tertiary amide and displays a peak at 1643 cm⁻¹, which corresponds well to the frequencies observed for disordered peptides and some α-helices (77).

2D IR spectra of the model amide-containing compounds were collected in both neat D₂O and in a mixture of 40% TFE in D₂O to mimic the solvent conditions identified in MAHP structural studies. Both NMA and NMP exhibited a 5–6 cm⁻¹ blue shift in 40% TFE compared with neat D₂O (Figs. S1 and S2). The blue shift suggests that the carbonyl bonds are strengthened in the presence of TFE, consistent with a reduction in the available hydrogen-bond partners in solution. While the 5–6 cm⁻¹ blue shift is notable, both vibrational coupling and hydrogen-bonding are expected to red shift the amide I′ frequencies. Therefore, while solvent-induced frequency shifts may dampen structure-induced frequency shifts in the model peptide, they should not create “false positives” or prevent data analysis.

2D IR spectra of MAHP were collected at each solvent condition identified in Fig. 1. These spectra (Fig. 2) revealed that, as the peptide’s helicity increases, the amide I′ frequency red shifts. Critically, the amide I′ frequency of MAHP continues to red shift even with added TFE. This suggests that, while TFE blue shifts the amide I′ modes by ∼6 cm⁻¹, the red shift associated with increased vibrational coupling and/or hydrogen-bonding dominates as the overall α-helicity of the sample increases.

Figure 2.

2D IR spectra of unlabeled MAHP. 2D IR spectra of MAHP adopting (A) a disordered structure at pH 3 without TFE, (B) a partially helical structure at pH 10 without TFE, and (C) a predominantly helical secondary structure at pH 10 with 40% TFE by volume. Amide I′ frequencies (red) and percent helicities quantified by CD experiments (black) are noted for each spectrum. The aspartate side chain presents as a shoulder near ∼1610 cm⁻¹ at pH 3 (A) and as a well-defined peak pair at 1580 cm⁻¹ at pH 10 (B and C). Average frequencies and standard deviations (SDs) over three replicates are summarized in Table S1. To see this figure in color, go online.

After establishing solvent conditions that correspond to varying levels of α-helicity in the model peptide, we first investigated whether paired isotope labels were sensitive to local structural changes as the helix folds/unfolds. Dual ¹³C¹⁸O probes were incorporated three residues apart into the sequence at the N-terminus (A3A6), center (A5A8), and C-terminus (A8G11) of MAHP. At the i and i+3 positions, the isotope-edited carbonyls should be stacked along the α-helical axis and exhibit spectral features indicative of vibrational coupling. Fig. 3 shows 2D IR spectra of the three labeled positions under the three distinct helicity conditions. For all isotopologues in the disordered configuration, the heavier ¹³C¹⁸O isotopes red shift the local amide I′ backbones by 55–65 cm⁻¹ and provide a well-resolved isotopic signature (28,64,78,79,80), although it is important to note that an antisymmetric stretching mode of the aspartate side chain appears in this region (Fig. 2) (59), a complication that is addressed in detail later in this report.

Figure 3.

Double ¹³C¹⁸O-labeled MAHP variants exhibit red shifted isotope modes at increased α-helicity. 2D IR spectra of MAHP labeled with two ¹³C¹⁸O probes at (A–C) A3 and A6, (D–F) A5 and A8, and (G–I) A8 and G11. The dual isotopic probes exhibit varying red shifts upon α-helix formation, with frequencies noted in red. Average frequencies and SDs over three replicates are summarized in Table S2. To see this figure in color, go online.

Under the primarily disordered condition at pH 3, A3A6 and A8G11 both exhibit an isotope-labeled peak pair at an average frequency of 1593 cm⁻¹ (Fig. 3, A and G). This frequency is ∼59 cm⁻¹ red shifted from the disordered amide I′ mode at 1652 cm⁻¹, as expected. Interestingly, the isotope-labeled features for A5A8 are slightly red shifted to 1590 cm⁻¹ under these conditions (Fig. 3 D). This shift may indicate that, while the peptides are mostly disordered, some initial structure is present at A5 and A8 leading to either weak vibrational coupling or intramolecular hydrogen-bonding. Under partially helical conditions at pH 10 without TFE, A3A6 and A5A8 both red shift by approximately 10 cm⁻¹ to around 1581 cm⁻¹ (Fig. 3, B and E). In contrast, the A8G11 isotope peak red shifts by only 5 cm⁻¹, to around 1588 cm⁻¹ (Fig. 3 H). Together, these results indicate that the N-terminal and central regions of MAHP have adopted a helical configuration, while the C-terminus remains more disordered. Upon the addition of TFE to further increase helicity, the frequency of the A3A6 and A5A8 peaks remains unchanged, while A8G11 red shifts slightly to an average frequency of 1586 cm⁻¹, indicating that the C-terminus remains less helical than the rest of the peptide. From these results, we conclude that paired isotope labels are capable of capturing subtle changes in local α-helical structures, even within one helical repeat length.

While it is clear that frequency shifts of the isotope-labeled amide I′ mode can localize structural changes, the cause of these shifts is not yet clear; these spectral changes could arise from either vibrational coupling or hydrogen-bonding. As previously discussed, vibrational coupling between the i and i+3 amide I′ modes can red shift the observed frequencies through the delocalization of local transition dipoles, whereas the formation of intramolecular hydrogen bonds can red shift the local frequency due to a weakening of the carbonyl bond. To understand the origin of the isotopic frequency shift, we measured a singly labeled variant (A5) under the same solvent conditions. Since the ∼60 cm⁻¹ energy difference between ¹²C¹⁶O and ¹³C¹⁸O amide I′ modes is nearly an order of magnitude larger any of the vibrational coupling strengths present in α-helices, this labeling scheme should effectively eliminate coupling-induced frequency shifts of the isotope-labeled mode. Therefore, any observed frequency shifts could arise only from hydrogen-bonding interactions. As the peptide’s α-helicity increased, the A5 isotope peak red shifted from 1591 to 1582 cm⁻¹ (Fig. 4), similar to the doubly labeled A3A6 and A5A8 samples (Fig. 3, A–F). Therefore, hydrogen-bonding was identified as the primary determinant of the frequency shift observed for isotope-labeled residues within α-helices.

Figure 4.

Frequency shifts of A5 MAHP reveal hydrogen-bonding interactions. 2D IR spectra of A5 MAHP at (A) pH 3 without TFE, (B) pH 10 without TFE, and (C) pH 10 with 40% TFE by volume. The single isotopic probe exhibits a red shift similar to the doubly labeled A3A6 and A5A8 upon increasing sample α-helicity. To see this figure in color, go online.

Finally, we must address the underlying side-chain features that congest the spectral region in which isotope-labeled amide I′ modes appear. Several side chains, including those from aspartate, glutamate, and arginine, contain IR-active functional groups that absorb between 1560 and 1610 cm⁻¹ (59). In the case of MAHP, the antisymmetric stretch of the aspartate side chain appears as a strong mode between ∼1580 and 1590 cm⁻¹ under pH 10 conditions (Fig. 2). Here, we analyze the signal strength of side-chain/isotope modes to determine whether integrated peak areas can be used to distinguish coupled isotopic signals from overlapping side-chain modes. When vibrational modes couple, the delocalization of the local modes to form normal modes redistributes oscillator strength (32,57). While we have shown that hydrogen-bonding is the main cause of frequency shifts in α-helices, coupling between paired isotope labels at the i and i+3 positions still exists. Thus, peaks arising from coupled isotope-labeled modes should appear more intense than peaks arising from uncoupled isotope-labeled modes or from uncoupled side-chain modes. To rigorously quantify the extent of delocalization, and thus the extent of coupling, the TDS of the ¹³C¹⁸O-labeled amide I′ modes for each labeling scheme should be calculated (25,41). However, at the concentrations used in this study, TDS calculations suffer from low signal/noise in the isotopically edited amide I′ region. As an alternative, we calculated peak ratios between the low-frequency side-chain and isotope modes between 1580 and 1595 cm⁻¹ and the high-frequency unlabeled amide I′ modes around 1640–1650 cm⁻¹ for each sample of MAHP. 2D IR signals are highly sensitive to changes in TDS (40,41,42), unlike FTIR, and thus comparison of 2D IR integrated peak areas can report on the relative amount of coupling between oscillators. Integrated peak areas were calculated by fitting 2D IR diagonal slices to a sum of Gaussian functions (Fig. S3). A comparison of the relative peak areas for the unlabeled, single, and double ¹³C¹⁸O-edited peptides as a function of peptide α-helicity is shown in Fig. 5 and Table S3.

Figure 5.

Average ratio of integrated peak areas (low-frequency mode to amide I′ mode) for unlabeled, single, and double ¹³C¹⁸O-labeled MAHP as a function of solvent condition. The unlabeled (red) ratio increases at higher pH due to side-chain ionization. Incorporation of a single ¹³C¹⁸O isotope at A5 (orange) yields a higher peak ratio under all conditions due to the presence of both the isotope-labeled amide I′ mode and the side-chain mode within the low-frequency region. Double ¹³C¹⁸O labels within the C-terminal region (A8G11, blue) yield peak ratios that follow a similar trend to the single label at A5, suggesting that coupling between the A8 and G11 labels is minimal. Double ¹³C¹⁸O labels within the N-terminal (A3A6, yellow) and central (A5A8, green) regions exhibit enhanced peak ratios at high pH conditions compared, indicating significantly greater coupling between paired isotope labels within these regions compared with those at the C-terminus. The error bars indicate SDs over three replicates. To see this figure in color, go online.

In the unlabeled sample, the low-frequency region only contains contributions from the aspartate side chain. The integrated area of this mode increases with pH, from 0.03 to 0.12 of the amide I′ mode, due to deprotonation of the aspartate side chain. When a single ¹³C¹⁸O is added in the A5 labeling scheme, the low-frequency region now contains contributions from both aspartate and the isotope-labeled carbonyl. The integrated peak area of this peak also increases with pH, from 0.12 to 0.24. Based on the observed isotopic frequency shifts in Figs. 3 and 4, the A5 label falls in the middle of the α-helix in MAHP. Incorporation of an isotope label at this position might be expected to reduce the integrated peak area of the unlabeled amide I′ band as it disrupts delocalization of the vibrational mode across the full helical length; however, this is not observed to be a significant effect. The increase in peak ratio for A5 is comparable with that observed for the unlabeled sample, which suggests that aspartate deprotonation remains the primary cause of the intensity change. Paired labels at the C-terminal residues A8G11 followed a similar trend to the single A5 label. While the average peak ratio for A8G11 does reach a maximum of 0.3, higher than in the single label, the difference between A5 and A8G11 cannot be confidently assigned as statistically significant due to the large SD. This high variation in peak area likely reflects higher conformational flexibility, consistent with a frayed C-terminus. Based on the similarity between A5 and A8G11, we conclude that isotope labels placed in the C-terminal region experience minimal vibrational coupling.

In contrast, paired labels at the N-terminal and central regions of the peptide, A3A6 and A5A8, exhibit markedly different behavior. While both start at peak ratios near 0.2, they increase to values above 0.5 as the peptide α-helicity increases, indicating that these residues experience vibrational coupling that leads to a significant change in TDS. Interestingly, all isotope-labeled variants exhibit the highest peak ratio at the intermediate condition of pH 10 without TFE. Upon cosolvent addition, the peak ratios decrease by 35–40% of the maximum value for A5, A3A6, and A8G11. This decrease can be attributed to further stabilization of the α-helix by TFE, which thus leads to an overall increase in the integrated peak area of the unlabeled amide I′ mode but relatively little change in isotope-labeled modes. Notably, the peak ratio for A5A8 decreases by only 25% of the maximum value once the cosolvent is added, suggesting that the central region may experience a greater increase in organization, and thus coupling, compared with the termini when the α-helix is stabilized by TFE. Thus, the relative intensities of the isotope-labeled peaks can be used to differentiate coupled isotope modes from side-chain modes and provide an additional measure of structural organization.

Conclusions

In this study, we demonstrate that 2D IR supplemented with isotope labeling methods is exquisitely sensitive to local structural changes in a tunable α-helix. The incorporation of dual ¹³C¹⁸O probes three residues apart allowed us to determine that the N-terminal and central regions (A3–A8) of our model peptide readily adopted an α-helical structure as solvent conditions were shifted to favor helicity, while the C-terminus (A8-G11) retained more conformational flexibility. These findings are in good agreement with CD measurements, which determined that MAHP should be predominantly helical at high pH and 40% TFE (v/v). However, CD measurements can only capture total helical content within a peptide sample; CD alone cannot definitively determine which residues fall within specific regions of secondary structure. At pH 10 with 40% TFE, individual MAHP peptides could sample numerous peptide conformations. The two most likely scenarios are as follows: 1) over half the peptides adopt a stable α-helical structure that spans the full length of the peptide, or 2) the majority of the peptides adopt a partially ordered structure in which approximately seven residues participate in an α-helix while the remaining residues are more disordered. 2D IR spectra of isotope-labeled MAHP support the latter, as both frequency shifts and integrated peak areas of isotope-labeled modes indicate that the C-terminal region is less helical than the rest of the peptide. This partially helical structure is also supported by the design of the model peptide, in which the helical structure is stabilized by a hydrogen-bond between the free amino group of the N-terminus with the D1 side chain and by a salt bridge between residues E4 and K7 (62). Single labeling with ¹³C¹⁸O at the central A5 position allowed us to decouple the effects of hydrogen-bonding and vibrational coupling, revealing that hydrogen-bonding was the primary contributor to the ∼10 cm⁻¹ red shift in isotope frequency. In fact, studies have used hydrogen-bond-induced frequency shifts of single isotope labels to probe helical structures (51,61). However, spectral congestion arising from side-chain modes can hinder interpretation of the labeled amide I′ frequency shifts. Some 2D IR studies have circumvented this complication by introducing mutations to remove residues that have IR-active modes within the region of interest (61), but such mutations can affect protein structure and dynamics and thus limit our understanding of the native sequence. An alternate approach could be to isotope label any IR-active side chains (81,82) and thus shift their vibrational modes out of the region of interest without the need for mutation, but side chains are generally more difficult and expensive to isotope label than backbone carbonyls. We seek to circumvent the need for mutations or side-chain labeling by introducing dual backbone labels that only experience coupling within α-helical structures. While hydrogen-bonding remains the primary cause of any frequency shifts, the redistribution of oscillator strength due to vibrational coupling allows coupled ¹³C¹⁸O probe pairs to be readily distinguished from uncoupled single isotopes or side-chain modes.

While this study takes advantage of many of the benefits of 2D IR over linear IR, a completely unique feature of a 2D IR spectrum is the appearance of crosspeaks. Crosspeaks arise from vibrational coupling between reporter groups or chemical exchange processes and are sensitive to both the distance and relative orientation between vibrational modes (32,57). As this study utilized identical site-specific probes, we were not able to take advantage of the additional structural information that crosspeaks would provide. However, mixed labeling schemes, such as the incorporation of ¹³C and ¹³C¹⁸O residues, can provide such information if the crosspeaks are sufficiently resolved (50). However, the frequency difference between ¹³C and ¹³C¹⁸O amide I′ modes is relatively small and, particularly when coupling between labels is weakened due to imperfect helix formation, may require the use of higher-resolution spectrometers and/or the use of window functions to clearly resolve the spectral features (50). Instead, a mixed amide Iʹ/amide IIʹ labeling scheme with ¹³C¹⁸O and ¹⁵N placed one helical turn apart would provide higher resolution for underlying crosspeaks. This novel labeling scheme has been demonstrated successfully for a 3₁₀-helix, but 2D IR data acquisition required the use of nonaqueous solvents to avoid overlapping absorptions from H₂O or D₂O (58). In addition, dihedral indexing between adjacent residues can provide additional insight into secondary structure, as demonstrated in studies of transient intermediates formed by an amyloidogenic peptide (46,83). Nevertheless, the current dual ¹³C¹⁸O labels enabled us to capture changes in both hydrogen-bonding and vibrational coupling that arise due to local changes in α-helical structure. These findings can readily be applied to determine how α-helices in larger proteins are affected by mutations, changes in environment, or adsorption onto a membrane or nanoparticle surface.

Author contributions

K.R.W., K.A.H., and L.E.B. designed the research. K.R.W., K.A.H., A.S., and K.D.S. collected experimental data. K.R.W. and K.A.H. analyzed the data. K.R.W., K.A.H., and L.E.B. wrote and revised the manuscript.

Editor: John Conboy.