Influence of Side Chain–Backbone Interactions and Explicit Hydration on Characteristic Aromatic Raman Fingerprints as Analysed in Tripeptides Gly-Xxx-Gly (Xxx = Phe, Tyr, Trp)

Abstract

Because of the involvement of π-electron cyclic constituents in their side chains, the so-called aromatic residues give rise to a number of strong, narrow, and well-resolved lines spread over the middle wavenumber (1800–600 cm⁻¹) region of the Raman spectra of peptides and proteins. The number of characteristic aromatic markers increases with the structural complexity (Phe → Tyr → Trp), herein referred to as (F_i = 1, …, 6) in Phe, (Y_i = 1, …, 7) in Tyr, and (W_i = 1, …, 8) in Trp. Herein, we undertake an overview of these markers through the analysis of a representative data base gathered from the most structurally simple tripeptides, Gly-Xxx-Gly (where Xxx = Phe, Tyr, Trp). In this framework, off-resonance Raman spectra obtained from the aqueous samples of these tripeptides were jointly used with the structural and vibrational data collected from the density functional theory (DFT) calculations using the M062X hybrid functional and 6-311++G(d,p) atomic basis set. The conformation dependence of aromatic Raman markers was explored upon a representative set of 75 conformers, having five different backbone secondary structures (i.e., β-strand, polyproline-II, helix, classic, and inverse γ-turn), and plausible side chain rotamers. The hydration effects were considered upon using both implicit (polarizable solvent continuum) and explicit (minimal number of 5–7 water molecules) models. Raman spectra were calculated through a multiconformational approach based on the thermal (Boltzmann) average of the spectra arising from all calculated conformers. A subsequent discussion highlights the conformational landscape of conformers and the wavenumber dispersion of aromatic Raman markers. In particular, a new interpretation was proposed for the characteristic Raman doublets arising from Tyr (~850–830 cm⁻¹) and Trp (~1360–1340 cm⁻¹), definitely excluding the previously suggested Fermi-resonance-based assignment of these markers through the consideration of the interactions between the aromatic side chain and its adjacent peptide bonds.

1. Introduction

A pioneering work published by Lord and Yu in 1970 [1] emphasized for the first time the importance of aromatic residues in structuring the Raman scattering profiles of peptide chains. In their work, off-resonance Raman spectra of lysozyme compared with those recorded from constituting amino acids showed that aromatic vibrations are responsible for strong, narrow, and well-resolved lines distributed across the middle wavenumber region. Since then, numerous investigations, using both off- and on-resonance Raman spectroscopy, have been devoted to the spectral analysis and environmental sensitivity of aromatic Raman markers [2,3,4,5,6,7,8,9,10,11,12,13], as well as to their use as protein fingerprints in Raman microscopy [14,15].

This work is limited to the study of the origin and conformational dependence of three aromatic residues, phenylalanine (Phe), tyrosine (Tyr), and tryptophan (Trp) (Figure 1). During recent years, simple molecular compounds, such as free amino acids [16,17,18] and dipeptides [19,20,21], have been used to analyze aromatic vibrational motions. However, it progressively appeared that these short motifs were insufficient to thoroughly reproduce different nonbonded interactions in a peptide chain, especially those between an aromatic side chain and its adjacent peptide bonds. As a result, a tripeptide model with an aromatic residue located at its second position, seemed to be a suitable choice. Herein, we consider the most structurally simple tripeptides with the primary sequence Gly-Xxx-Gly (Xxx = Phe, Tyr, Trp) (Figure 1), because the two flanking Gly residues at either side of the aromatic residue do not provide characteristic vibrations overlapping with the aromatic Raman markers [22,23]. On the other hand, the structuring trend of tripeptides in aqueous solution has received attention during the two past decades. Among these studies, a phenomenological approach applied to homo- and hetero-tripeptides revealed the capability of these compounds to form ordered secondary structures, such as β-like, helical, polyproline-II (pP-II), and γ-turn folds [24,25,26,27,28,29,30]. To explore as accurately as possible the structural and vibrational features of short peptides’ quantum mechanical approaches, and more commonly those based on the density functional theory (DFT), [31] has been applied. However, the main objective in DFT calculations is to choose suitable functionals [32]. Although semiempirical (parameterized) functionals have shown a higher flexibility to account for the experimental data, the shortcomings of the widely used B3LYP functional [33,34], albeit a three-parameter functional, has been revealed in relation to the treatment of noncovalent interactions in molecular systems. A systematic investigation based on the use of a series of hybrid functionals [23,35] has recently shown that improved hybrid functionals, such as ωB97XD [36,37,38] and M062X [39], provide better structural stability in tripeptides. However, it has also been evidenced [23] that the relative energies obtained by these functionals are systematically higher than those predicted by the Møller–Plesset (MP2) method [40] taken as a reference. To account for the solvent effects on the structural features, several hydration models were considered, either by embedding the Gly-rich tripeptides [23,41,42,43] in a polarizable solvent continuum (PCM) [44,45], or by surrounding the cationic Ala-rich [35,46] and Gly-rich [35,43,47] tripeptides with a number of explicit water molecules.

Our recent DFT calculations on both Gly-rich and Ala-rich tripeptides [35,43] have shown that the structural stability of all types of conformers, especially those with an inverse γ-turn conformation, is preserved upon geometry optimization by means of the M062X functional, provided that (i) they are explicitly hydrated by a minimal number of water molecules interacting preferentially with the polar sites of backbone and aromatic side chains and (ii) the whole cluster formed by each tripeptide conformer and its hydrating molecules is embedded in a solvent continuum. Our objective in the present work was to extend this protocol to the three tripeptides Gly-Xxx-Gly (Xxx = Phe, Tyr, Trp) upon consideration of different types of secondary structures (β-like, helix, pP-II, and γ-turn) and plausible side chain orientations. The structural and vibrational data of the optimized conformers, as well as the corresponding relative energies, were subsequently used to calculate the thermal averaged Raman spectra through multiconformational analysis. The reliability of the theoretical spectra was checked through their comparison with the experimental ones obtained from the solution samples. It was shown that beyond the validation of the used theoretical methods, the confrontation of the experimental and calculated Raman spectra leads to obtaining insight into the energetic and conformational landscape of each tripeptide in an aqueous environment.

2. Results

2.1. Observed Solution Raman Spectra

Buffer- and TFA-subtracted Raman spectra recorded from solution samples of the three tripeptides are displayed in Figure 2 within the middle wavenumber spectral region. Because of the shortness of the backbone, including only three amide bands (two inter-residue and another located at the C^ter terminus) (Figure 1A), only a weak, broad amide-I band centred at ~1692 cm⁻¹ was observed in all tripeptides (Figure 2). The number of aromatic markers, giving rise to intense and narrow Raman lines, increases with the aromatic ring structural complexity (phenyl → phenol → indole), referred to as (F_i = 1, …, 6) in Gly-Phe-Gly (Figure 2A), (Y_i = 1, …, 7) in Gly-Tyr-Gly (Figure 2B), and (W_i = 1, …, 8) in Gly-Trp-Gly (Figure 2C). See also Table 1 for the corresponding wavenumbers. Isotopic shifts in aromatic markers upon H-D exchange on labile hydrogen atoms were previously reported [41,42,43].

2.2. Theoretical Conformers and Raman Spectra

Tables S1–S3 (Supporting Information) provide the energetic and structural data, i.e., relative energy (ΔE) and backbone and side chain torsion angles of optimized conformers, as well as intramolecular H-bond lengths (d_HB) corresponding to folded structures. The energy interval, within which the ΔE values are located, increases with the structural complexity of the aromatic ring: 0 ≤ ΔE ≤ 7.66 kcal/mol in Gly-Phe-Gly (Table S1); 0 ≤ ΔE ≤ 8.43 kcal/mol in Gly-Tyr-Gly (Table S2); and 0 ≤ ΔE ≤ 10.72 kcal/mol in Gly-Trp-Gly (Table S3). All conformers, initially having a β-strand structure, were transformed into extended chains, i.e., with a quite flat and quasi-planar backbone. Helical, pP-II, and γ-turn structures are maintained upon geometry optimization, as confirmed by the (Φ₂,ψ₂) torsion angles corresponding to the middle (aromatic) residue.

The experimental and calculated Raman spectra of Gly-Phe-Gly are compared in Figure 3A,B. The calculated spectrum is the thermal (Boltzmann) average of the individual spectra obtained from 15 optimized conformers (Table S1, Supporting Information). To better emphasize the main contributing conformers, the thermal-weighted calculated spectra corresponding to the low-energy conformers with a ΔE ≤ 2 kcal/mol are displayed in Figure 3C. This proves that the contribution of the higher-energy conformers (with ΔE > 2 kcal/mol) to the global Raman intensity is negligible. The graphical representation of the seven low-energy conformers (ΔE < 2 kcal/mol) is displayed in Figure 4.

Similarly, the comparison between the observed and calculated (thermal average of 30 conformers) Raman spectra of Gly-Tyr-Gly is shown in Figure 5A,B. Only three conformers were found within the ΔE ≤ 2 kcal/mol range. The thermal corrected Raman spectra of these three conformers are shown in Figure 5C (see graphical representation of these conformers in Figure 6).

At last, the comparison between the observed and calculated (thermal average of 30 optimized conformers) Raman spectra of Gly-Trp-Gly is shown in Figure 7A,B. Only two conformers are located within the ΔE ≤ 2 kcal/mol range (Figure 7C). See Figure 8 for the graphical representation of these conformers.

3. Discussion

3.1. Conformational and Energetic Landscapes of Tripeptides

The relative energy (ΔE) of a conformer was shown to depend on its backbone structure, as well as on the aromatic side chain type (phenyl, phenol, indole) and orientation (Tables S1–S3, Supporting Information). For instance, in Gly-Phe-Gly, the lowest-energy conformer (ΔE = 0) is an extended chain with a g⁻g^± side chain orientation (Figure 4), whereas in Gly-Tyr-Gly, an extended chain with a tg⁻ side chain corresponds to the lowest energy. In Gly-Trp-Gly, the lowest-energy conformer was found to have a pP-II backbone with a g⁺g⁺ side chain (Figure 8). However, in all tripeptides, classic γ-turn conformers provide the highest ΔE values, whereas helical and inverse γ-turn structures form the intermediate relative energies.

In a helical structure, an intramolecular H-bond is formed between the backbone C=O (Gly¹) and one of the hydrogen atoms involved in the C^ter amide group (Figure 2 and Figure 4). Its length (d_HB) increases with the aromatic ring’s structural complexity, i.e., 2.09–2.12 Å in Gly-Phe-Gly, 2.07–2.24 Å in Gly-Tyr-Gly, and 2.02–2.27 Å in Gly-Trp-Gly. In γ-turn conformers, the intramolecular H-bond, also referred to as the “turn closing H-bond”, is formed between the backbone C=O (Gly¹) and N-H (Gly³) (Figure 2 and Figure 4). Surprisingly, the d_HB values relative to classic γ-turn folds are generally shorter than those corresponding to inverse γ-turn folds: 1.91–2.01 Å (classic γ-turn) versus 2.02–2.27 Å (inverse γ-turn) in Gly-Phe-Gly; 1.90–2.07 Å (classic γ-turn) versus 2.00–2.27 Å (inverse γ-turn) in Gly-Tyr-Gly; and 1.78–2.00 Å (classic γ-turn) versus 2.07–2.37 Å (inverse γ-turn) in Gly-Trp-Gly. As already mentioned, conformers with a classic γ-turn fold have systematically higher ΔE values than those assigned to inverse γ-turn structures. This means that the shortness of intramolecular H-bonds cannot solely explain the higher stability of conformers, and favourable (versus unfavourable) nonbonded interactions occurring between the aromatic ring and the backbone of a tripeptide play the key role in forming the energy landscape of its conformers.

The influence of an aromatic ring’s orientation on relative energies can be better described by considering the role of the side chain torsion angles (χ₁, χ₂) (Figure 1A). χ₁ orientation places the C_γ atom of an aromatic ring with respect to the backbone. As a result, with a χ₁(g⁻) orientation, the aromatic ring is placed onto the peptide bond located at the N^ter side of the aromatic residue, whereas with a χ₁(t), the opposite orientation is favoured, enabling aromatic ring interactions with the peptide bond at its C^ter side. At last, a χ₁(g⁺) orientation brings the aromatic ring between the two aforementioned peptide bonds (Figure 4). The role of χ₂ is simply maintaining the aromatic ring stacked with the backbone. Consequently, the side chain orientation defined by the pair (χ₁, χ₂) leads to the optimization of the interactions between the aromatic ring and its adjacent peptide bonds. On the other hand, the orientation change in the aromatic ring brings drastic modification to the location of the water molecules, thus affecting the relative energies of hydrated conformers.

Supposing that the population of a given secondary structure can be identified by its thermal (Boltzmann) weight, Figure 9 displays the histograms representing the normalized populations (expressed in percent) of each backbone type (extended chains, pP-II, helical, classic and inverse γ-turn), as averaged from all possible aromatic side chain orientations. Quite similar distributions are obtained for Gly-Phe-Gly (Figure 9A) and Gly-Tyr-Gly (Figure 9B), presenting extended chains as major contributions (71% and 84%, respectively), and pP-II, helical, and inverse γ-turn as minor ones. However, subtle differences are found in minor populations in going from one tripeptide to another. In Gly-Trp-Gly, the major population (85%) is ascribed to the pP-II secondary structure, making only one remarkable minor population (14%) assignable to extended chains (Figure 9C).

3.2. Assignment of Aromatic Raman Markers

The agreement between the observed and calculated spectra within the middle wavenumber region (Figure 3, Figure 5 and Figure 7) justifies both the consistency of the used theoretical level and the adequacy of the multiconformational approach to calculate the relative energies and Raman intensities of conformers.

Table 1 gives the assignments of aromatic markers derived from the vibrational calculations on the lowest-energy (ΔE = 0) conformers. As can be seen, the three high-wavenumber markers located above 1200 cm⁻¹ in Phe (F₁, F₂ and F₃) and Tyr (Y₁, Y₂ and Y₃) originate from quite similar vibrational motions, i.e., basically from the bond stretching motions occurring in phenyl and phenol moieties, respectively. However, substantial differences appear in the markers located below 1200 cm⁻¹, among which are the two strong Phe markers (F₄ and F₅), as well as the Tyr marker (Y₄). This effect can be attributed to the contribution of the phenol hydroxyl vibrational motions. Both components of the well-known characteristic Tyr doublet (Y₅–Y₆) originate from the fundamental phenol ring vibrational motions. While the higher-wavenumber component of this doublet (Y₅) results from the bond-stretching motions, the lower-wavenumber one (Y₆) arises from the out-of-plane bending of the C-H bonds. Nevertheless, the similarity between the vibrational motions responsible for the lowest-wavenumber markers in Phe (F₆) and Tyr (Y₇) is to be emphasized. Six out of eight Trp Raman markers (W₁, …, W₆) originate from the indole ring’s bond-stretching motions. Particularly, both components of the widely discussed Trp doublet (W₄–W₅) are assigned to the indole ring’s fundamental vibrations. The two lowest-wavenumber Trp markers are ascribed to either the in-plane bending (W₇) or the out-of-plane bending (W₈) vibrations of the aromatic ring.

Table 1. Aromatic Raman markers and their assignments.

Marker	Exp.	Assignment
Phe
F1	1605	ν(Cδ1-Cε1); ν(Cδ2-Cε2)
F2	1586	ν(Cε1-Cζ); ν(Cε2-Cζ); ν(Cγ-Cδ2); ν(Cγ-Cδ1)
F3	1207	ν(Cβ-Cγ); Cβ-H2 rock.;ν(Cγ-Cδ1); ν(Cγ-Cδ2)
F4	1032	ν(Cε2-Cζ); ν(Cε1-Cζ); δ(Cζ-Cε1-H); δ(Cζ-Cε2-H)
F5	1003	δ(Cγ-Cδ1-Cε1); δ(Cδ1-Cε1-Cζ); δ(Cζ-Cε2-Cδ2); δ(Cγ-Cδ2-Cε2)
F6	628	δ(Cζ-Cε2-Cδ2); δ(Cδ1-Cε1-Cζ); δ(Cγ-Cδ1-Cε1); δ(Cγ-Cδ2-Cε2)
Tyr
Y1	1616	ν(Cδ1-Cε1); ν(Cδ2-Cε2); ν(Cγ-Cδ1); ν(Cε2-Cζ)
Y2	1600	ν(Cε1-Cζ); ν(Cγ-Cδ2); ν(Cε2-Cζ); ν(Cγ-Cδ1)
Y3	1209	ν(Cβ-Cγ); Cβ-H2 rock.; ν(Cδ1-Cε1); ν(Cδ2-Cε2); ν(Cγ-Cδ2); δ(Cδ1-Cγ-Cδ2)
Y4	1178	δ(Cγ-Cδ1-H); δ(Cε1-Cδ1-H); δ(Cζ-Oη-H); δ(Cζ-Cε2-H); Cβ-H2 twist.; δ(Cδ2-Cε2-H); δ(Cδ1-Cε1-H)
Y5	851	ν(Cγ-Cδ1); ν(Cβ-Cγ); ν(Cε1-Cζ); ν(Cε2-Cζ); ν(Cγ-Cδ1); ν(Cζ-Oη)
Y6	828	ω(Cε2-H); ω(Cδ2-H); ω(Cδ1-H)
Y7	643	δ(Cγ-Cδ1-Cε1); δ(Cγ-Cδ2-Cε2); δ(Cδ1-Cε1-Cζ); δ(Cδ2-Cε2-Cζ)
Trp
W1	1620	ν(Cε2-Cζ2); ν(Cε3-Cζ3)
W2	1578	ν(Cδ2-Cε3); ν(Cζ2-Cη); ν(Cζ3-Cη); ν(Cδ2-Cε2)
W3	1552	ν(Cγ-Cδ1); ν(Cε2-Cζ2); ν(Cβ-Cγ)
W4	1360	ν(Cγ-Cδ2); Cβ-H2 twist.
W5	1340	ν(Cζ2-Cη); ν(Cζ3-Cη); ν(Cε2-Cζ2); ν(Cδ2-Cε2)
W6	1012	ν(Cζ3-Cη); ν(Cε3-Cζ3); ν(Cζ2-Cη)
W7	880	δ(Cε2-Nε1-Cδ1); Cβ-H2 wagg.
W8	759	ω(Cζ2-H); τ(Cζ2-Cη); ω(Cη-H); ω(Cζ3-H); τCδ2-Cε3); τ(Cε2-Cζ2)

Marker: notations used for characteristic aromatic Raman lines are reported. (F_i = 1, …, 6) for Phe, (Y_i = 1, …, 7) for Tyr, and (W_i = 1, …, 8) for Trp. Exp.: positions of the main peaks (cm⁻¹) taken from the room-temperature aqueous solution Raman spectra of the tripeptides Gly-Phe-Gly (Figure 2A), Gly-Tyr-Gly (Figure 2B), and Gly-Trp-Gly (Figure 2C). Assignments: obtained from the vibrational data of the lowest-energy conformers (ΔE = 0) of the cationic species of Gly-Phe-Gly (Figure 4), Gly-Tyr-Gly (Figure 6), and Gly-Trp-Gly (Figure 8). Based on the potential energy distribution (PED) matrix derived from the vibrational calculations. ν, δ, ω, and τ designate bond stretch, angular bending, out-of-plane bending, and torsion internal coordinates. Rock. (rocking), Wagg. (wagging), and twist. (twisting) reflect the symmetrical coordinates associated with the CH₂ groups at the β position of the aromatic side chain. See also Figure 1A,B for atomic nomenclature of backbone and side chain.

Table 1. Aromatic Raman markers and their assignments.

Marker	Exp.	Assignment
Phe
F1	1605	ν(Cδ1-Cε1); ν(Cδ2-Cε2)
F2	1586	ν(Cε1-Cζ); ν(Cε2-Cζ); ν(Cγ-Cδ2); ν(Cγ-Cδ1)
F3	1207	ν(Cβ-Cγ); Cβ-H2 rock.;ν(Cγ-Cδ1); ν(Cγ-Cδ2)
F4	1032	ν(Cε2-Cζ); ν(Cε1-Cζ); δ(Cζ-Cε1-H); δ(Cζ-Cε2-H)
F5	1003	δ(Cγ-Cδ1-Cε1); δ(Cδ1-Cε1-Cζ); δ(Cζ-Cε2-Cδ2); δ(Cγ-Cδ2-Cε2)
F6	628	δ(Cζ-Cε2-Cδ2); δ(Cδ1-Cε1-Cζ); δ(Cγ-Cδ1-Cε1); δ(Cγ-Cδ2-Cε2)
Tyr
Y1	1616	ν(Cδ1-Cε1); ν(Cδ2-Cε2); ν(Cγ-Cδ1); ν(Cε2-Cζ)
Y2	1600	ν(Cε1-Cζ); ν(Cγ-Cδ2); ν(Cε2-Cζ); ν(Cγ-Cδ1)
Y3	1209	ν(Cβ-Cγ); Cβ-H2 rock.; ν(Cδ1-Cε1); ν(Cδ2-Cε2); ν(Cγ-Cδ2); δ(Cδ1-Cγ-Cδ2)
Y4	1178	δ(Cγ-Cδ1-H); δ(Cε1-Cδ1-H); δ(Cζ-Oη-H); δ(Cζ-Cε2-H); Cβ-H2 twist.; δ(Cδ2-Cε2-H); δ(Cδ1-Cε1-H)
Y5	851	ν(Cγ-Cδ1); ν(Cβ-Cγ); ν(Cε1-Cζ); ν(Cε2-Cζ); ν(Cγ-Cδ1); ν(Cζ-Oη)
Y6	828	ω(Cε2-H); ω(Cδ2-H); ω(Cδ1-H)
Y7	643	δ(Cγ-Cδ1-Cε1); δ(Cγ-Cδ2-Cε2); δ(Cδ1-Cε1-Cζ); δ(Cδ2-Cε2-Cζ)
Trp
W1	1620	ν(Cε2-Cζ2); ν(Cε3-Cζ3)
W2	1578	ν(Cδ2-Cε3); ν(Cζ2-Cη); ν(Cζ3-Cη); ν(Cδ2-Cε2)
W3	1552	ν(Cγ-Cδ1); ν(Cε2-Cζ2); ν(Cβ-Cγ)
W4	1360	ν(Cγ-Cδ2); Cβ-H2 twist.
W5	1340	ν(Cζ2-Cη); ν(Cζ3-Cη); ν(Cε2-Cζ2); ν(Cδ2-Cε2)
W6	1012	ν(Cζ3-Cη); ν(Cε3-Cζ3); ν(Cζ2-Cη)
W7	880	δ(Cε2-Nε1-Cδ1); Cβ-H2 wagg.
W8	759	ω(Cζ2-H); τ(Cζ2-Cη); ω(Cη-H); ω(Cζ3-H); τCδ2-Cε3); τ(Cε2-Cζ2)

Marker: notations used for characteristic aromatic Raman lines are reported. (F_i = 1, …, 6) for Phe, (Y_i = 1, …, 7) for Tyr, and (W_i = 1, …, 8) for Trp. Exp.: positions of the main peaks (cm⁻¹) taken from the room-temperature aqueous solution Raman spectra of the tripeptides Gly-Phe-Gly (Figure 2A), Gly-Tyr-Gly (Figure 2B), and Gly-Trp-Gly (Figure 2C). Assignments: obtained from the vibrational data of the lowest-energy conformers (ΔE = 0) of the cationic species of Gly-Phe-Gly (Figure 4), Gly-Tyr-Gly (Figure 6), and Gly-Trp-Gly (Figure 8). Based on the potential energy distribution (PED) matrix derived from the vibrational calculations. ν, δ, ω, and τ designate bond stretch, angular bending, out-of-plane bending, and torsion internal coordinates. Rock. (rocking), Wagg. (wagging), and twist. (twisting) reflect the symmetrical coordinates associated with the CH₂ groups at the β position of the aromatic side chain. See also Figure 1A,B for atomic nomenclature of backbone and side chain.

3.3. Wavenumber Dispersion of Aromatic Raman Markers

The average wavenumbers accompanied by their standard deviations calculated across all optimized conformers are reported in

Table 2. Calculated average values and standard deviations of the wavenumbers of aromatic Raman markers.

Marker Phe a	av ± sd	Marker Tyr b	av ± sd	Marker Trp c	av ± sd
F1	1643 ± 2	Y1	1621 ± 4	W1	1639 ± 2
F2	1623 ± 2	Y2	1598 ± 7	W2	1594 ± 3
F3	1219 ± 5	Y3	1198 ± 8	W3	1560 ± 5
F4	1046 ± 3	Y4	1164 ± 17	W4	1359 ± 4
F5	1003 ± 1	Y5	848 ± 8	W5	1335 ± 6
F6	626 ± 4	Y6	830 ± 10	W6	1013 ± 3
		Y7	639 ± 8	W7	868 ± 9
				W8	760 ± 4

All DFT calculations were performed by means of M062X functional and 6-311++G(d,p) basis set. Average (av.) wavenumbers (cm⁻¹) are separated from the corresponding standard deviations (sd.) by “±” symbol. ^a Calculated on 15 optimized conformers of Gly-Phe-Gly surrounded by five water molecules and embedded in a solvent continuum. See Table S1 (Supporting Information) for energetic and geometrical parameters of the geometry-optimized conformers. Raw calculated wavenumbers (ν_calc) were scaled by the affine relation ν_scal = aν_calc + b, with (a,b) = (0.974, 11.19). ^b Calculated on 30 optimized conformers of Gly-Tyr-Gly surrounded by six water molecules and embedded in a solvent continuum. See Table S2 (Supporting Information) for energetic and geometrical parameters of the geometry-optimized conformers. Raw calculated wavenumbers (ν_calc) were scaled by the affine relation ν_scal = aν_calc + b, with (a,b) = (0.956, 11.19). ^c Calculated on 30 optimized conformers of Gly-Trp-Gly surrounded by seven water molecules and embedded in a solvent continuum. See Table S3 (Supporting Information) for energetic and geometrical parameters of the geometry-optimized conformers. Raw calculated wavenumbers (ν_calc) were scaled by the affine relation ν_scal = aν_calc + b, with (a,b) = (0.965, 11.19).

. Average wavenumbers depend on, and reflect, the used theoretical level, whereas standard deviations may bring information on the variation in wavenumbers upon conformational transitions between conformers. Phe markers present the lowest dispersion values (≤5 cm⁻¹), whereas larger values are revealed for Tyr markers, among which the most important values are predicted for Y₄ (17 cm⁻¹) and Y₆ (10 cm⁻¹). Comparatively, Trp markers all present a medium wavenumber dispersion, not exceeding 9 cm⁻¹.

3.4. A New Interpretation for Tyr (Y₅–Y₆) and Trp (W₄–W₅) Raman Doublets

During past decades, the spectral shape of the Raman doublets arising from the phenol and indole rings was a subject of debate in several reports devoted to the Raman spectroscopy of peptides and proteins [1,2,3,4,5,6,7,8,9,10,11,12,18,41,42]. The Tyr doublet in 1973 [2], and the Trp doublet in 1986 [3], were successively assigned to Fermi-resonance effects. It was suggested that in Tyr, the interaction of a fundamental (planar) mode (Y₅) with the first overtone of an out-of-plane mode gives rise to the doublet (Y₅–Y₆) observed at ~850–830 cm⁻¹ [2]. A similar type of interaction was suggested in Trp, but between a planar mode (W₄) and the additive combination of two out-of-plane modes, resulting in a doublet (W₄–W₅) observed at ~1360–1340 cm⁻¹ [3]. The reason behind this similarity was in fact due to the simplicity of the used model compounds, and the quite modest computational power available at that time for normal mode calculations. Precisely, aromatic compounds, p-cresol (or methyl-phenol) and skatole (or methyl-indole), were used to interpret the normal modes of Tyr and Trp, respectively. Both molecules were supposed to have a planar (C_s symmetry), rendering possible the interpretation of the fundamental modes Y₅ and W₄ (Raman active), as well as the out-of-plane modes (IR active) contributing to the formation of the doublets via the aforementioned Fermi-resonance effects. It should be remarked that in C_s symmetry, the overtone or the additive combination of IR active modes become Raman-active, justifying the apparition of the Tyr and Trp doublets in Raman spectra. Recent DFT calculations on more sophisticated model compounds [18,41,42], as well as the present one (see Section 4.2 for details), have clearly shown that the components of both doublets, i.e., (Y₅–Y₆) and (W₄–W₅) arise from the aromatic ring’s fundamental modes, definitely excluding the previously suggested Fermi-resonance-based interpretations [2,3].

Furthermore, the observed changes in the relative intensities of the doublets, namely ${\mathrm{\rho }}_Y={\displaystyle \frac{{\mathrm{I}}_{850}}{{\mathrm{I}}_{830}}}$ (for the Tyr doublet) and ${\mathrm{\rho }}_{\mathrm{W}}={\displaystyle \frac{{\mathrm{I}}_{1360}}{{\mathrm{I}}_{1340}}}$ (for the Trp doublet), where I is the Raman intensity, have been reported as indicators of the environmental changes in these two aromatic residues. The variation of ρ_Y was attributed to H-bonding [2]. It was assumed that when phenol hydroxyl acts as H-bond donor, ρ_Y < 1, and in the case where this group becomes a H-bond acceptor, ρ_Y > 1. Previous DFT calculations on free amino acid (Tyr) evidenced that the H-bonding on phenol hydroxyl cannot solely explain the observed reversal of ρ_Y [18]. The present calculations on the tripeptide Gly-Tyr-Gly give us the opportunity to analyze more completely the conformation and hydration dependence of the Tyr doublet. Figure 10 displays the calculated Raman spectra obtained from the 30 conformers of the tripeptide within the 900–800 cm⁻¹ spectral region, drawn by keeping the same scale along the vertical axis for all of them. These spectra are shown as the elements of a 5 × 3 table, where each row corresponds to one of the five considered secondary structures, and each column is relative to the side chain orientations. It is interesting to note that the doublet becomes a singlet or a triplet in certain cases. It can also be perceived that the g⁻ ↔ g⁺ orientation change of the χ₁ torsion angle plays a key role in ρ_Y reversal. More precisely, with a χ₁(g⁻) orientation, ρ_Y < 1, whereas when a χ₁(g⁺) is adopted, ρ_Y > 1. In contrast, with a χ₁(t) orientation, different situations may appear depending on the χ₂ orientation, as well as on the backbone secondary structure.

Presumably inspired by the interpretation of ρ_Y, possible observed ρ_W values were suggested, i.e., ρ_W > 1 versus ρ_W < 1 might be related to the Trp hydrophobic versus hydrophilic environments. Recent preliminary DFT calculations have shown that ρ_W reversal mainly depends on the backbone and aromatic side chain conformation [42]. Figure 11 displays the conformational behaviour of the presently calculated Raman spectra obtained from the 30 conformers of Gly-Trp-Gly within the 1400–1300 cm⁻¹ spectral region. It can be deduced that in linear backbone structures (extended chain or pP-II), ρ_W ≤ 1. The situation becomes different in folded secondary structures (helix, classic and inverse γ-turn), where in many cases the doublet may be transformed into a triplet. These spectra also highlight the combined effects of the side chain χ₁ and χ₂ torsion angles on the spectral shape in the analyzed region. For instance, in a helical backbone, while a g⁻g⁻ (or tg⁻) side chain provides ρ_W < 1, a g⁻g⁺ (or tg⁺) orientation leads to ρ_W > 1. A reverse situation occurs with tg⁻ (ρ_W > 1) and tg⁺ (ρ_W < 1) side chains. Interestingly, it appears that classic and inverse γ-turn folds can be discriminated, because with the same side chain orientation, different spectral shapes are obtained for these oppositely folded secondary structures.

4. Methods

4.1. Experimental Details

Tripeptides, with the primary sequence NH₂-Gly-Xxx-Gly-CONH₂ (Xxx = Phe, Tyr, Trp), were synthesized at the Institut Pasteur (Paris, France) according to the Fmoc/tBu solid-phase strategy [48] from a Rink amide resin on an ABI 433 synthesizer (Applied Biosystems, Foster City, CA). Details concerning the synthesis procedure and purification were recently reported [41]. The purity control of the final peptides was >98%. The experimental masses were acquired in positive-ion mode, and were consistent with the theoretical isotopic values. Gly-Phe-Gly: expected M + H⁺ 279.1452, observed 279.1441; Gly-Tyr-Gly: expected M + H⁺ 295.1401, observed 295.1383; and Gly-Trp-Gly: expected M + H⁺ 318.1561, observed 318.1539.

Because of the high pK_a value (~10.5) of the amine terminal group, the cationic form (Figure 1A) remains the major species within a wide pH range (up to ~10) in aqueous solution. Samples containing the tripeptides with trifluoroacetic acid (TFA) as the counterion were prepared at room temperature by dissolving lyophilized powders in water taken from a Millipore filtration system. Upon dissolution, the pH was 6.5 ± 0.1. The final concentration was 20 mM for all tripeptides, i.e., ~5.6 g/L (Gly-Phe-Gly), ~5.9 g/L (Gly-Tyr-Gly), and ~6.3 g/L (Gly-Trp-Gly), leading to a good signal/noise ratio in the Raman spectra.

Then, 50 μL solution samples were placed in a suprasil quartz cell (5 mm path length) and excited by the 488 nm line of an Ar⁺ laser (Spectra Physics), ~200 mW power at the sample. Scattered light at a right angle was analyzed on a T64000 (HORIBA Jobin Yvon, Longjumeau, France) in a single spectrograph configuration with a 1200-groove-per-millimetre holographic grating and a holographic notch filter. Room-temperature Stokes Raman data were collected by means of a liquid nitrogen-cooled charge-coupled device detection system (Spectrum One, Jobin Yvon). The effective spectral slit width was set to ~5 cm⁻¹. An average accumulation time of 1200 s (40 scans of 30 s each one) was chosen for recording the reported Raman spectra. Buffer subtraction and smoothing of the observed Raman spectra [41,42,43] was performed by using GRAMS/32 Z.00 (Thermo Galactic, Waltham, MA, United States).

4.2. Theoretical Details

Seven torsion angles (ψ₁, ω₁, Φ₂, ψ₂, ω₂, Φ₃, ψ₃), where subscripts recall residue numbers (Figure 1A), provide together the secondary structure of a cationic tripeptide backbone. Among these angles, ω, defined around the peptide bond (-CO-NH-), remains close to 180°, whereas the (Φ,ψ) pairs define the local conformation of a given residue [49]. To construct the initial conformers of each tripeptide, five distinct secondary structures were considered as follows: β-strand, in which (Φ,ψ) = (−135°,+135°); pP-II with (Φ,ψ) = (−75°,+150°); and α-helix conformation, where (Φ,ψ) = (−60°,−45°). γ-turn folds were supposed to be centred on the middle residue (Phe², Tyr², Trp²) (Figure 1B), with the values (Φ₂,ψ₂) = (+75°,−65°) and (Φ₂,ψ₂) = (−75°,+65°) for the so-called classic and inverse γ-turn, respectively [50,51]. All other backbone torsion angles in a γ-turn conformer, i.e., those relative to Gly¹ and Gly³, were initially set to those relative to a β-strand. Two other torsion angles (χ₁, χ₂) detail the aromatic side chain orientation (Figure 1B). Furthermore, [49] χ₁(N-C_α-C_β-C_γ), around the single bond (C_α-C_β), can naturally adopt three privileged orientations, namely gauche⁺ (or g⁺) (60° ± 60°), gauche⁻ (or g⁻) (−60° ± 60°), and trans (or t) (180° ± 60°). In contrast, because of the sp² type of the C_γ atom (Figure 1B), χ₂ defined around C_β-C_γ bond is limited to two privileged orientations, i.e., g⁺(90° ± 90°) and g⁻(−90° ± 90°). Starting with the special case of Gly-Phe-Gly, as C_δ1 and C_δ2 atoms located within the phenyl ring are indiscernible, two equivalent definitions become possible for χ₂, namely χ₂⁽¹⁾(C_α-C_β-C_γ-C_δ1) or χ₂⁽²⁾(C_α-C_β-C_γ-C_δ2). One of these two angles is always positive (g⁺), and the other one negative (g⁻), while the absolute value of their difference remains close to 180°. For the sake of clarity, both χ₂ values were herein reported in Table S1 (Supporting Information), and their orientation was referred to by the particular symbol g^±. As a result, three distinct Phe² side chain orientations, i.e., g⁻g^±, g⁺g^±, and tg^±, with respect to (χ₁, χ₂) angles, were considered. Therefore, with the five considered backbone rotamers and the three plausible side chain orientations, 5 × 3 = 15 initial conformers were prepared for Gly-Phe-Gly. The situation is rather different in Gly-Tyr-Gly, where C_δ1 and C_δ2 can be distinguished by the (left/right) orientation of the phenol hydroxyl group (O_η–H). It should be stated that the two most energetically favourable locations of O_η–H are within the phenol ring (Figure 1B). To eliminate any ambiguity, both the χ₂⁽¹⁾ and χ₂⁽²⁾ values were reported in Table S2 (Supporting Information), among which χ₂⁽¹⁾ provides the g⁺ (or g⁻) orientation of χ₂. Considering the five backbone rotamers and the six side chain orientations (namely g⁻g⁻, g⁻g⁺, g⁺g⁻, g⁺g⁺, tg⁻, and tg⁺), 5 × 6 = 30 initial Gly-Tyr-Gly conformers were prepared. In Gly-Trp-Gly, the definition of χ₂(C_α–C_β–C_γ–C_δ1) [49] becomes straightforward because of the asymmetric nature of the indole ring. Nevertheless, to avoid any confusion between the present definition and that previously used by many other authors, i.e., χ^2,1(C_α–C_β–C_γ–C_δ2), the values of both torsion angles (χ₂ and χ^2,1) were herein reported (Table S3, Supplementary Information). Note that when a g⁺ orientation is assigned to χ₂, a g⁻ orientation naturally corresponds to χ^2,1, and inversely. As a result, 5 × 6 = 30 initial conformers were prepared for the Trp-containing tripeptide. In all tripeptides, the initial values given to χ₁ were as follows: +60° (g⁺), −60° (g⁻), and 180° (t), whereas those assigned to χ₂ were +90° (g⁺) and -90° (g⁻).

As already mentioned in Section 1, the hydration effects were considered in both implicit and explicit models, which consist of embedding a cluster including the tripeptide in its surrounding water molecules in a polarizable continuum medium, with a relative permittivity corresponding to bulk water (ε_r = 78.39). Belonging to the first hydration, all explicit water molecules are capable of forming H-bonds with the polar sites of backbone and aromatic side chains. Starting with the aromatic rings, a pioneering work on benzene [52] revealed that a water molecule can bind to the phenyl ring through the so-called H_w…π interaction (where H_w is a water hydrogen atom interacting with the π-electron cloud of an aromatic ring). More recently, a theoretical work was conducted to find the privileged hydration sites in natural aromatic acids [53]. The most important results derived from this study can be summarized as follows: (i) the H_w…π interaction was confirmed in the phenyl ring of Phe; (ii) water molecules may bind to the phenol ring in Tyr either through the H_w…π interaction or by direct H-bonding with its hydroxyl group (O_η–H); and (iii) in Trp, up to three water molecules can hydrate the indole ring, two of them adopting the H_w…π interaction type, and a third one by H-bonding to N_ε1-H (Figure 1B). As far as the backbone of the tripeptides is concerned, inspired by the investigations on amino acids [54] and tripeptides [35,43], a minimal number of explicit water molecules was considered. Precisely, only one water molecule was used to hydrate each backbone terminal group (NH₃⁺ or CO-NH₂), while two others were considered to form H-bonds with successive N-H and C=O groups along the backbone chain (Figure 1A). As a result, the hydration number (n) was as follows: n = 5 in Gly-Phe-Gly, n = 6 in Gly-Tyr-Gly, and n = 7 in Gly-Trp-Gly. The initial locations of the hydrating water molecules were those determined by the previous DFT calculations on amino acids [54] and tripeptides [35,43].

Theoretical calculations were performed by means of the Gaussian16 package [55]. The 75 initial conformers for the three tripeptides were submitted to geometry optimization using the DFT approach, M062X hybrid functional, and polarized triple-zeta Gaussian atomic basis set 6-311++G(d,p) [56]. In major cases, optimization was carried out by the default algorithm (eigenvalue following), except in a few cases where a greater number of steps was required to achieve a tight convergence.

Geometry optimization of hydrated conformers was followed by harmonic vibrational calculations. The absence of any imaginary frequency proved the correspondence of an optimized geometry with a local minimum. The energies of the optimized conformers were based on their total energy (E_tot), where E_tot = E_e + ΔG, i.e., the sum of electronic energy (E_e) and free energy correction (ΔG). Therefore, the energy assigned to each conformer includes both enthalpic and entropic contributions. Each optimized conformer was thus characterized by its relative energy (ΔE) compared to that of the lowest-energy conformer, for which the ΔE was set to zero.

Assignment of the vibrational modes was based on the so-called potential energy distribution (PED) [57], as expressed in terms of internal coordinates, i.e., bond stretching (ν), angular bending (δ), out-of-plane bending (ω), and torsion (τ) coordinates. Raman intensities were evaluated based on the calculated Raman activities following the previously reported expression [58]. The calculated Raman intensities, to be compared to experimental spectra, were obtained from the thermal (Boltzmann) average of the spectra derived from different conformers. To scale the raw calculated wavenumbers (ν_calc), an affine relation of the following form: ν_scal = aν_calc + b, was used (where ν_scal represents scaled wavenumbers and (a,b) are constants). The advantages of this scaling procedure versus a widely used simple linear relation (ν_scal = aν_calc) were detailed in our previous reports [42,43]. See the caption of Table 1 for the (a,b) values determined via linear regression (correlation coefficient > 0.995).

5. Concluding Remarks

In the present work, the conformation and hydration dependence of the characteristic aromatic markers were analyzed on the basis of the experimental Raman spectra and the theoretical structural and vibrational data derived from the tripeptides with the primary sequence Gly-Xxx-Gly (where Xxx = Phe, Tyr, Trp). The most striking effect is the structural stability of all considered conformers, especially inverse γ-turn, upon geometry optimization. This is certainly due to the joint use of the M062X functional and explicit backbone hydration, avoiding the opening of folded structures. The influence of the backbone and aromatic ring on the relative energies of conformers has been detailed. Extended chains (β-like) in Gly-Phe-Gly and Gly-Tyr-Gly, and pP-II in Gly-Trp-Gly, were shown to be the most populated conformers in an aqueous environment. The adequacy of the multiconformational analysis of Raman spectra has been highlighted, not only for reproducing the observed data on the tripeptides, but also for predicting the conformation dependence of aromatic markers. On the basis of the theoretical data, a new interpretation was suggested for the characteristic Tyr (~850–830 cm⁻¹) and Trp (~1360–1340 cm⁻¹) Raman doublets, definitely excluding the previously suggested Fermi-resonance-based assignment of these markers through the consideration of the interactions between the aromatic side chain and its adjacent peptide bonds.

For the further use of readers, the atomic cartesian coordinates of the low-energy conformers displayed in Figure 4, Figure 6, and Figure 8 are given in Tables S4–S6 (Supporting Information).