Side Chain Dependence of Intensity and Wavenumber Position of

Mar 30, 2005 - wavenumber, while those containing polar side chains experienced an ... amide I′ depolarization ratio and wavenumber which reflects t...
1 downloads 0 Views 502KB Size
Download PDF
J. Phys. Chem. B 2005, 109, 8195-8205

8195

Side Chain Dependence of Intensity and Wavenumber Position of Amide I′ in IR and Visible Raman Spectra of XA and AX Dipeptides Thomas Measey,† Andrew Hagarman,† Fatma Eker,‡ Kai Griebenow,‡ and Reinhard Schweitzer-Stenner*,† Department of Chemistry, Drexel UniVersity, 3141 Chestnut Street, Philadelphia, PennsylVania 19104 and Department of Chemistry, UniVersity of Puerto Rico, Rı´o Piedras Campus, San Juan, Puerto Rico 00931 ReceiVed: September 17, 2004; In Final Form: February 8, 2005

A series of AX and XA dipeptides in D2O have been investigated by FTIR, isotropic, and anisotropic Raman spectroscopy at acidic, neutral, and alkaline pD, to probe the influence of amino acid side chains on the amide I′ band. We obtained a set of spectral parameters for each peptide, including intensities, wavenumbers, half-widths, and dipole moments, and found that these amide I′ parameters are indeed dependent on the side chain. Side chains with similar characteristic properties were found to have similar effects on the amide I′. For example, dipeptides with aliphatic side chains were found to exhibit a downshift of the amide I′ wavenumber, while those containing polar side chains experienced an increase in wavenumber. The N-terminal charge causes a substantial upshift of amide I′, whereas the C-terminal charge causes a moderate decrease of the transition dipole moment. Density functional theory (DFT) calculations on the investigated dipeptides in vacuo yielded different correlations between theoretically and experimentally obtained wavenumbers for aliphatic/aromatic and polar/charged side chains, respectively. This might be indicative of a role of the hydration shell in transferring side chain-backbone interactions. For Raman bands, we found a correlation between amide I′ depolarization ratio and wavenumber which reflects that some side chains (valine, histidine) have a significant influence on the Raman tensor. Altogether, the obtained data are of utmost importance for utilizing amide I as a tool for secondary structure analysis of polypeptides and proteins and providing an experimental basis for theoretical modeling of this important backbone mode. This is demonstrated by a rather accurate modeling for the amide I′ band profiles of the IR, isotropic Raman, and anisotropic Raman spectra of the β-amyloid fragment Aβ1-82.

Introduction The amide I band of the IR and, to a lesser extent, the Raman spectra of polypeptides and proteins is frequently used to determine their secondary structure composition.1,2 The structural sensitivity of the amide I mode, which is predominantly a CO stretching vibration,3 is generally attributed to the influence of hydrogen bonding on the force constant of the carbonyl bond4 and to the structure sensitivity of vibrational mixing between amide I modes in a polypeptide chain.3,5,6 A variety of techniques have been developed over the last 30 years for decomposing the amide I band profile of proteins and peptides into its secondary-structure-representing components. All these methods have in common that they are based on the simplifying assumption that neither the IR oscillator strength, the Raman cross-section, nor the wavenumber of the local amide I mode of a residue depends on the side chain of the adjacent amino acid residues.1,2 However, to our best knowledge, this has never been checked thoroughly for a representative set of amino acid side chains. It is well-established that the superior structural sensitivity of amide I stems, to a major extent, from its delocalization in polypeptide chains owing to significant through-bond and * Corresponding author. Phone: (215) 895-2268. Fax: (215)895-1265. E-mail: [email protected]. † Drexel University. ‡ University of Puerto Rico.

through-space coupling between adjacent local oscillators.3-6 In principle, these modes of interpeptide coupling render the amide I intensity distribution of a polypeptide a function of the dihedral angles of the respective amino acid residues. Only recently, spectroscopists have begun to exploit this peculiarity of amide I for a more detailed structure analysis. Hamm and associate, for instance, employed femtosecond two-dimensional IR to determine the structure of small model peptides by measuring the interpeptide coupling energy of amide I modes.7,8 Their approach has recently been extended by Hochstrasser and associates to explore the conformations of β-amyloid aggregates and alanine-based peptides.9,10 The Keiderling group generally combines ab initio-based calculations with IR and vibrational circular dichroism (VCD) profiles of peptide spectra for structure analysis.11-13 Schweitzer-Stenner and co-workers used a classical excitonic coupling model to determine the conformation of tri- and tetrapeptides as well as a β-amyloid fragment from the amide I′ profile in the respective IR, isotropic Raman, anisotropic Raman, and VCD spectra.14-17 Their modeling of the amide I′ band shape is empirical in that it is based on spectroscopically determined parameters.18 Its extension to longer and biologically relevant peptides and proteins requires a detailed knowledge of the parameters governing the intrinsic local wavenumber of an amide I mode. This issue has been addressed only recently on a theoretical basis by Cho and coworkers, who used Hartree-Fock and density functional theory (DFT) calculations to show that the wavenumbers of local

10.1021/jp045762l CCC: $30.25 © 2005 American Chemical Society Published on Web 03/30/2005

8196 J. Phys. Chem. B, Vol. 109, No. 16, 2005

Measey et al.

(uncoupled) amide I modes are significantly affected by the peptide-peptide interactions, so that they depend on the distance between the considered peptides and their nearest neighbors, and thus also on the dihedral angle of the respective amino acid residues.19,20 The present study is aimed at providing experimental information about the parameters governing the spectral properties of amide I′. It focuses on a yet unresolved issue, namely the determination of the influence of adjacent side chains on wavenumber, IR oscillator strength, and half-width. To this end, we measured and analyzed the amide I′ band of a series of XA and AX dipeptides in D2O, where X denotes a representative series of amino acid side chains. We confined ourselves to D2O as a solvent to avoid vibrational mixing between amide I and H2O bending vibrations.21,22 DFT calculations for the respective peptides in vacuo were carried out to facilitate the interpretation of the experimental data. Additionally, we measured and analyzed the amide I′ band of isotropic and anisotropic Raman spectra of a limited subset of the investigated peptides in order to explore the influence of side chains on the depolarization ratio, which reflects the ratio of anisotropic and isotropic scattering, and thus the properties of the Raman tensor. The knowledge of the side chain dependence of Raman scattering is necessary for accurate modeling of the amide I band shapes in the isotropic and anisotropic Raman spectra of polypeptides. Even though only the anionic state of the investigated dipeptides can be considered to be a suitable model system for the nonresidual peptide groups of polypeptides, we also investigated the zwitterionic and cationic states to infer how the terminal charges affect the amide I wavenumber and its side chain dependence. In this regard, dipeptides are suitable model systems for which experimental data can guide theoretical calculations exploring the influence of charges on vibrational modes. The results of our study provide ample material for the experimental check of (quantum chemical) calculations and significantly increase the set of experimentally determined parameters usable for the simulation of amide I band profiles of polypeptides. This is demonstrated by a modeling of the respective amide I profiles of the β-amyloid fragment Aβ1-28, which yielded a more accurate reproduction of the experimental spectra than a rather qualitative calculation reported earlier, which was based on the assumption that all amide I wavenumbers of nonresidual peptides are identical.16 Theory IR Absorption. The amide I′ oscillator strengths were directly calculated from the FTIR amide I′ band profile by using the equation23

Da01 ≈

9.2 × 10-39 ν0

∫band (ν) dν

(1)

where ν0 is the amide I′ wavenumber and (ν) is the molar absorptivity. The absolute value of the transition dipole moment was calculated as the square root of the oscillator strength and expressed in units of esu‚cm. Determination of the Amide I′ Raman Tensor. As a first approximation, one can generally assume that even with faroff resonance excitation the Raman intensity of amide modes stems predominantly from Franck-Condon-type vibronic coupling to in-plane π f p* transitions of the peptide group with predominant contributions from π f π1* (NV1) and π f π2* (NV2) electronic transitions.24 It is therefore appropriate to select a molecular coordinate system that is coplanar to the peptide

plane. As described and visualized in earlier publications, we identify the x-axis with the NCR bond at the C-terminal. The y-axis is in the peptide plane and points toward the carbonyl oxygen. The z-axis is perpendicular to the peptide plane.25 For this coordinate system, the Raman tensor can be written as

( )

a c 0 Rˆ ) c b 0 0 0 0

(2)

For zwitterionic and anionic dipeptides, a small out-of-plane component can be admixed to the tensor by vibronic coupling to an n(COO-) f π1 charge transfer transition from the carboxylate to the peptide group.26,27 We approximate the corresponding Raman tensor by solely considering on-diagonal zz-contributions17,28

( )

a c 0 Rˆ ′ ) c b 0 0 0 d

(3)

Of course, the off-diagonal element c depends on a and b, and the corresponding relationship can be obtained by rotating the tensor into a coordinate system in which the tensor has only diagonal elements. As shown earlier, this yields 25

c)-

b-a 9.3

(4)

The depolarization ratio of Raman lines can be calculated as a function of a, b, and d by using the well-know equation

F)

3γ2aniso Iy ) Ix 45β2 + 4γ2 iso

(5)

aniso

where Ix and Iy are the scattered intensities measured parallel and perpendicular to the polarization of the exciting laser beam. γ2aniso and β2iso are both proportional to the tensor elements and are defined as follows

1 β2iso ) (TrRˆ )2 9

(6a)

1 γ2aniso ) [(Rxx - Ryy)2 + (Ryy - Rzz)2 + (Rzz - Rxx)2] + 2 3 [(R + Ryx)2 + (Ryz + Rzy)2 + (Rxz + Rzx)2] (6b) 4 xy Because we are not interested in obtaining absolute values for the tensor elements, we normalized the tensors to ayy, setting b ) 1, as was done in an earlier publication.28 Materials and Methods Materials. L-alanyl-L-phenylalanine (AF), L-phenylalanyl-Lalanine (FA), L-alanyl-L-glycine (AG), L-glycyl-L-alanine (GA), L-alanyl-L-lysine (AK), L-lysyl-L-alanine (KA), L-alanyl-Lleucine (AL), L-leucyl-L-alanine (LA), L-alanyl-L-serine (AS), L-seryl-L-alanine (SA), L-alanyl-L-valine (AV), L-valyl-L-alanine (VA), L-alanyl-L-isoleucine (AI), L-isoleucyl-L-alanine (IA), L-alanyl-L-threonine (AT), L-threonyl-L-alanine (TA), L-histidylL-alanine (HA), L-methionyl-L-alanine (MA), L-aspartyl-Lalanine (DA), L-alanyl-L-proline (AP), L-prolyl-L-alanine (PA), L-glutamyl-L-alanine (EA), L-arginyl-L-alanine (RA), L-tryptophyl-L-alanine (WA), L-tyrosyl-L-alanine (YA), L-alanyl-Lalanine (AA), and L-alanyl-L-glutamine (AQ) were purchased

Influence of Side Chains on IR and Raman Spectra

J. Phys. Chem. B, Vol. 109, No. 16, 2005 8197

from Bachem Bioscience, Inc. With the exception of AF and KA (>96% purity), all the peptides contained >98% purity, and all were used without further purification. NaClO4 was purchased from Sigma-Aldrich Chemical Company (St. Louis, MO) and was of analytical grade. The peptides were dissolved in D2O (Aldrich) at concentrations of 0.1-0.2 M for both the IR and Raman experiments. The concentrations of all peptides were determined by using their molecular weight. To check the accuracy of this approach, we measured the 1H NMR spectrum of L-alanyl-L-alanine at different concentrations. We plotted the intensities I’s of several signals as a function of concentration c and found that all these data could be satisfactorily fitted to a linear function I ) a‚c (a ) const), indicating that decontamination of the lyophilized peptide samples (e.g., by water) is negligible. The pH values of the solutions were adjusted to 1, 6, and 12 by adding small aliquots of DCl and NaOD (Acros Organics) to obtain the cationic, zwitterionic, and anionic states of the peptides, respectively. The obtained pH values were converted to pD values by using the method of Glasoe and Long.29 Methods. Raman Spectroscopy. The Raman measurements were carried out in the Biospectroscopy laboratory at the Department of Chemistry of the University of Puerto Rico (Rı´o Piedras, PR). The Raman spectra were obtained with 442-nm (65-mW) excitation from an HeCd laser (model IK 4601R-E, Kimmon Electric U.S.). The equipment and experimental setup used for the Raman measurements have been described elsewhere.15 IR Spectroscopy. The IR spectra of L-alanyl-L-alanine, Ltyrosyl-L-alanine, L-tryptophyl-L-alanine, L-arginyl-L-alanine, L-glutamyl-L-alanine, L-prolyl-L-alanine, L-alanyl-L-proline, Lmethionyl-L-alanine, L-threonyl-L-alanine, l-alanyl-L-threonine, L-isoleucyl-L-alanine, L-alanyl-L-isoleucine, L-valyl-L-alanine, L-alanyl-L-valine, L-alanyl-L-phenylalanine, and L-phenylalanylL-alanine were measured with a Perkin-Elmer model 1600 FTIR absorption spectrometer using an ICL precision demountable cell equipped with CaF2 windows and a 15-µm Mylar spacer. IR spectra of the remaining dipeptides were measured using a Nicolet Magna-IR System 560 optical bench as described elsewhere.15 For these experiments, a Spectra Tech liquid cell equipped with CaF2 windows and a 12-µm Mylar spacer was used. Each IR spectrum was obtained by averaging 256 scans at 2-cm-1 resolution. Spectral Analysis. All IR and Raman spectra were analyzed using the program MULTIFIT.30 The Raman spectra were normalized to the internal standard (i.e., the ClO4- band at 934 cm-1). Solvent spectra were measured for IR and both perpendicular and parallel polarized Raman. All sample spectra were corrected for solvent contributions. For each dipeptide, IR and Raman spectra were both fitted with the same amide I′ band half-widths for consistency. The isotropic and anisotropic Raman intensities were calculated as follows

4 Iiso ) Ix - Iy 3

(7a)

Ianiso ) Iy

(7b)

Density Functional Theory Calculations. All calculations were carried out with the TITAN program, a joint product from Wavefunction, Inc. (Irvine, CA) and Schro¨dinger, Inc. (Portland, OR). Equilibrium geometries of cationic and anionic deuterated dipeptides assignable to the upper left-hand quadrant of the Ramachandran plot were obtained at the B3LYP/6-31G** level of theory. The calculations for the zwitterionic state were found

to be unreliable, most likely because of the strong Coulombic interactions between terminal charges. We obtained the amide I′ frequencies, dihedral angles, dipole moments, and CdO bond lengths for L-alanyl-L-histidine, L-alanyl-L-tyrosine, L-alanyl-Ltryptophan, L-alanyl-L-methionine, L-alanyl-L-arginine, and Lalanyl-L-glutamic acid, in addition to all the dipeptides listed in the Materials and Methods section. Results and Discussion This section of the paper is organized as follows: First, the results obtained from the spectral analysis of the amide I′ in the IR spectra of XA and AX peptides are described and discussed. The second section presents the results of DFT calculations and their comparisons with the spectroscopic data. In this context, we focus on identifying possible relationships between amide I′ parameters and the dihedral angles of the X residue. In the third section, the Raman depolarization ratios of the amide I′ band of a selected set of dipeptides is used to evaluate the influence of side chains on Raman scattering. Side Chain Dependence of Amide I′ Wavenumbers. Figure 1 exhibits the IR spectra measured between 1500 and 1700 cm-1 of anionic XA and AX peptides in D2O. All spectra display two bands resulting from the antisymmetric carboxylate stretching mode at 1590 cm-1 and the amide I′ band around 1630 cm-1. The perpendicular line reflects the respective wavenumber position for AA, which is used as the reference. Spectra for the cationic and zwitterionic states31 have also been recorded. The spectra are not shown, but the results of their analysis are considered in detail in the text. The anionic peptides can be regarded as the best representation of the respective amino acids in polypeptides because of the elimination of the N-terminal proton, which has a significant influence on the amide I wavenumber. The influence of the negative charge of the C-terminal is much weaker.32 All spectra were decomposed into Voigtian profiles as described in the Materials and Methods section. Satisfactory fits of the spectra were generally obtained, but some peptides apparently exhibit slightly asymmetric amide I′ band profiles, which are not fully reproduced by a symmetric Voigtian profile. An asymmetric band profile results from solvent fluctuations coupled to the amide I vibration.33 For some of the investigated peptides, the amide I′ overlaps at least partially with bands assignable to side chain vibrations (X ) D-, E-, Q, Y, F, W).34,35 Bands assignable to the antisymmetric COO- and the CO stretching mode of D and E appear clearly discernible from the amide I′ band at all three pH values used in this study (cf. the spectrum of DA in Figure 1, where the strong band at 1590 cm-1 results from an overlap of the COO- antisymmetric stretching bands of D and the C-terminal). Bands from the aromatic-ring modes of the aromatic side chains in the region between 1580 and 1610 cm-1 appear mostly in the anisotropic Raman spectrum and can also be clearly separated from the respective amide I′ band. The ND2 (ND3) deformation modes appear at much lower wavenumbers. Only for one of the dipeptides investigated, namely L-alanylL-glutamine (AQ) at anionic pD, a side chain band (arising from the CO stretching mode of Q) significantly overlaps with amide I′. Figure 2 illustrates how we eliminated this band from the spectrum of the anionic peptide. The side chain band, which is clearly visible and distinct from amide I′ at acidic pD, and to a sufficient extent, also at neutral pD, was fitted and the thusobtained band profile subtracted from the anionic spectrum. Figure 2b shows the amide I′ band of anionic AQ with the side chain band, as well as the amide I′ band with the side chain band subtracted. Figure 2c shows the IR spectrum of anionic

8198 J. Phys. Chem. B, Vol. 109, No. 16, 2005

Measey et al.

Figure 1. FTIR spectra of the indicated AX and XA dipeptides dissolved in D2O between 1500 and 1700 cm-1. All spectra were recorded at pD 12 to obtain the anionic state with respect to the terminal charges. The solid line and the band profiles result from a curve-fitting process described in the Materials and Methods section. The vertical lines represent the amide I′ wavenumber position of AA.

Figure 2. (a) FTIR spectra of the amide I′ band of AQ at all protonation states. (b) Amide I′ band of anionic AQ both with and without the side chain band. (c) Fitted amide I′ band of anionic AQ.

QA after the subtraction of the side chain band. This procedure was based on the assumption that the protonation of the terminal

group affects neither the position nor the profile of the side chain band, a notion corroborated by the practically identical

Influence of Side Chains on IR and Raman Spectra

J. Phys. Chem. B, Vol. 109, No. 16, 2005 8199

Figure 3. Amide I′ wavenumber histograms of XA and AX dipeptides obtained from FTIR spectra measured at acidic, neutral, and alkaline pD.

wavenumber position of this band in the spectra measured at neutral and acidic pD. Figure 3 depicts histogram plots of the observed amide I′ wavenumbers for all (terminal) protonation states of the peptides investigated. The peak positions are listed in Table 1. The wavenumbers for the XA series were found to be distributed over a range of approximately 12.8, 12.0, and 10.9 cm-1 for the cationic, zwitterionic, and anionic state, respectively. The corresponding distribtutions of the AX peptides are significantly narrower, namely, 7.0, 7.6, and 8.5 cm-1, indicating that the C-terminal residue has a lesser influence on the amide I′ wavenumber than the N-terminal residue. Interestingly, the XA distributions are all significantly narrower than the spread of amide III wavenumbers recently obtained for a more limited set of X-residues (24 cm-1).36 On the contrary, amide III appeared to be nearly independent of X in AX peptides. This indicates that the influence of side chains on amide III and amide I′ is governed by at least a partially different mechanism. We will address this issue again in the text when we discuss the results of our DFT calculations. The ordering of the XA peptides with respect to the amide I′ wavenumber is consistent among the cationic and zwitterionic protonation states only, with KA and EA the only exceptions. That is, the amide I′ wavenumber of the XA peptides show the

same deviation from AA, which is considered our reference peptide. However, the anionic XA peptides, in which the N-terminal nitrogen is deprotonated, show a significantly different ordering. For the AX peptides, on the other hand, a nearly perfect consistency among all protonation states was obtained. The corresponding wavenumber hierarchy reads as AQ, AG > AS, AT > AF > AK, AA > AL > AV > AI > AP. This clearly shows that the influence of the N-terminal proton on the amide I′ wavenumber is side-chain- and thus conformation-dependent. This result is certainly of relevance for future computational analyses of the influence of adjacent charges on local amide I wavenumbers in polypeptides. We found some correlations between both the XA and AX series of peptides in relation to the amide I′ wavenumber position. Dipeptides containing valine (V), leucine (L), and isoleucine (I) exhibit a downshift of the amide I′ position (with respect to AA), while peptides containing glycine (G) and serine (S) depict an upshifted amide I′ band. As shown by Cho and co-workers, the amide I′ wavenumber changes in the presence of electrostatic interactions between the peptide atoms and the environment, which, in our case, is constituted by the terminal groups, the side chains, and the solvent molecules.6,37 The direction of the wavenumber shift depends on the sign of the partial charges involved. Apparently, the V, L, and I side chains

8200 J. Phys. Chem. B, Vol. 109, No. 16, 2005

Measey et al.

TABLE 1: Amide I′ Wavenumber Positions for XA and AX Inferred from the Respective FTIR Spectra Measured at Alkaline, Neutral, and Acidic pD Amide I′ Wavenumber [1/cm] -- XA peptide

acidic

neutral

basic

AA DA HA GA SA RA EA TA FA YA MA KA PA WA LA IA VA

1667.67 1674.71 1674.07 1672.80 1671.54 1670.51 1669.68 1668.32 1668.24 1667.63 1666.87 1666.04 1665.84 1665.32 1662.86 1662.01 1661.94

1662.50 1664.54 1669.60 1667.98 N/A 1663.48 1662.34 1663.38 1663.47 1663.38 1663.23 1663.44 1660.83 1661.96 1658.34 1657.64 1658.74

1629.54 1634.09 1633.69 1632.49 1634.79 1628.40 1632.33 1633.84 1630.67 1630.70 1632.60 1635.12 1628.66 1627.52 1629.64 1624.18 1626.17

Amide I′ Wavenumber [1/cm] -- AX peptide

acidic

neutral

basic

AA AQ AG AT AS AF AK AL AV AI

1667.67 1672.64 1672.32 1670.92 1670.44 1669.48 1668.23 1667.48 1665.45 1665.29

1662.50 1668.12 1667.33 1664.54 1665.38 1663.70 1663.02 1662.50 1661.85 1660.56

1629.54 1632.80 1635.62 1631.76 1632.74 1631.24 1629.34 1628.47 1628.41 1627.10

TABLE 2: Section A. Transition Dipole Moments as Obtained from the Integrated Amide I′ Band Profile in the FTIR Spectra of XA and AX Dipeptides Measured at Acidic, Neutral, and Alkaline Protonation States. Section B. Average Transition Dipole Moments of the Investigated AX and XA Dipeptides for the Three Protonation States Investigated Amide I′ Dipole Moment [Debye ) 1018 esu cm-1] Section A acidic

neutral

basic

0.26 0.31 0.31 0.32 0.28 0.25 0.29 0.31 0.29 0.24 0.17 0.24 0.20 0.25 N/A 0.26 0.25

0.27 0.33 0.24 0.29 0.29 0.25 0.28 0.27 0.28 0.25 0.26 0.22 0.30 0.28 0.28 0.27 0.23

0.37 0.27 0.29 0.31 0.27 0.26 0.25 0.27 0.26 0.26 0.26

0.32 0.33 0.30 0.31 0.26 0.21 0.27 0.30 0.27 0.27 0.25

XA DA LA TA VA EA WA IA YA GA PA HA MA KA RA SA AA FA

0.38 0.34 0.33 0.33 0.32 0.32 0.30 0.30 0.29 0.29 0.29 0.28 0.28 0.28 0.28 0.27 0.24

AP AV AQ AT AI AG AL AK AA AS AF

0.30 0.34 0.33 0.32 0.31 0.30 0.28 0.28 0.27 0.27 0.26

AX

Section B XA AX

Figure 4. Correlation between the amide I′ wavenumber of XA and AX dipeptides and the hydropathy index of the side chains obtained from ref 38.

put enough atoms in close proximity to the peptide group to substantially affect the amide I′ wavenumber. Glycine might constitute a reference system in which there is no side chain influence on amide I′ at all. Serine has a polar side chain with substantial partial charges, which apparently causes an upward shift of the amide I′ wavenumber. Figure 4 exhibits the correlation between amide I′ wavenumbers and the hydropathy index of amino acids reported by Kyte and Doolittle.38 It is a measure of their respective hydrophobicity, so that nonpolar side chains have a positive hydropathy index value, while polar and charged side chains display negative values. Interestingly, we found good correlations between our experimentally obtained amide I′ wavenumber and this hydrop-

acidic

neutral

basic

0.30 ( 0.03 0.29 ( 0.02

0.23( 0.04 0.27 ( 0.04

0.27 ( 0.02 0.27 ( 0.04

athy index for all protonation states of AX dipeptides, whereas only intermediate correlations were obtained for XA peptides. This indicates that the N- and C-terminal side chains affect the amide I′ mode by different mechanisms. In concrete terms, our findings suggest direct noncovalent interactions for the former, whereas solvent-mediated interactions should govern the influence of the C-terminal side chain. The influence of the N-terminal side chain is discussed in more detail later on the basis of DFT calculations. Side Chain Dependence of the Amide I′ Dipole Moment. The dipole moments of the amide I′ modes were obtained from the integrated IR absorption of the amide I′ band profiles. Figure 5a,b shows histograms of the absolute values of the dipole moments for both XA and AX peptides, respectively. The values are listed in Table 2, section A. The average values and the respective standard deviations of the XA and AX series are presented in Table 2, section B for all three protonation states. While the average dipole moments of the cationic and anionic states of the XA peptides are identical in the limit of statistical accuracy, the corresponding value of the cationic state is larger, indicating a systematic influence of the C-terminal charge on the transition dipole moment. For the cationic state, we found the transition dipole moments of DA (0.39 D), LA (0.33 D), and FA (0.24 D) to exhibit a statistically significant deviation

Influence of Side Chains on IR and Raman Spectra

J. Phys. Chem. B, Vol. 109, No. 16, 2005 8201

Figure 6. Correlation between amide I′ wavenumbers of the cationic and anionic XA dipeptides obtained from a normal-mode calculation based on a DFT force field and from FTIR spectra.

Figure 5. Amide I′ dipole moment histograms of AX and XA dipeptides at all protonation states.

from the average value. This is also the case for VA, TA, LA, and YA (0.30 - 0.32 D), HA (0.17 D), and KA(0.20 D) in the zwitterionic case. Apparently, the positive charges of H and K decrease the transition dipole moment. For the anionic state V, L, and Y exhibit transition moments which are significantly larger than the average dipole moment. These data suggest a slight enhancement of the dipole moment by adjacent aliphatic and aromatic side chains. With the exception of FA, AS, AP, and YA, all of the dipeptides exhibit the strongest dipole moments in the cationic state. For the AX peptides, AT and AQ exhibit the largest dipole moments for all protonation states, while the amide I′ dipole strengths of LA and VA were consistently among the strongest for the XA peptides. Hahn et al.39 recently calculated the amide I′ dipole strengths for different structures of the neutral, deuterated acetylproline (AP) dipeptide by ab initio calculations. Two dipole strengths were calculated, corresponding to the higher and lower wavenumber amide I′ vibrations. The values reported were found to be of the same order of magnitude as the values we obtained for deuterated AX and XA dipeptides. This finding reflects the accuracy of our experiment and calculations. Density Functional Theory Calculations. DFT calculations were carried out for the cationic and anionic protonation states of all dipeptides investigated in this study, with the inclusion of AH, AY, AW, AM, AR, and AE. Figure 6 shows the correlations between the experimental and theoretical amide I′ wavenumbers for XA peptides. The obtained correlations are graded as very good for 0.9 e R e 1.0, as good for 0.8 e R
HA > AA > VA, and for the AX peptides, AS > AA > AV. Moreover, all amide I′ Raman bands could indeed be fitted with the same parameters as the respective FTIR amide I′ bands. Second, we used the depolarization ratios of amide I′ to check whether the Raman tensor depends on the side chain. To this end, we determined the depolarization ratios from the respective x- and y-polarized amide I′ Raman bands, which are listed in Table 3. Interestingly, we found a moderate correlation between the depolarization ratio and the amide I′ wavenumber, as displayed in Figure 9.

Influence of Side Chains on IR and Raman Spectra

J. Phys. Chem. B, Vol. 109, No. 16, 2005 8203

Figure 9. Correlation between the Raman depolarization ratio and the amide I′ wavenumber for the anionic XA and AX peptides.

TABLE 4: Raman Tensor Elements of Amide I′ Obtained from the Experimentally Observed Depolarization Ratios as Described in the Text. Values Listed in Section A Were Obtained for d ) 0, while Section B Lists the d Values Obtained by Assuming the Same In-(Peptide)Plane Contributions, a and b, for All Dipeptides Section A anionic XA AA GA SA AS VA AV HA

Raman Tensor Element a b 0.34 0.32 0.32 0.32 1 0.94 0.17

1 1 1 1 1 1 1

c

d

-0.071 -0.073 -0.072 -0.072 0 -0.007 -0.09

0 0 0 0 0 0 0

Section B anionic XA AA GA SA AS VA AV HA

Raman Tensor Element a b c 0.34 0.34 0.34 0.34 0.34 0.34 0.34

1 1 1 1 1 1 1

-0.07 -0.07 -0.07 -0.07 -0.07 -0.07 -0.07

d 0.00 -0.02 -0.01 -0.01 0.20 0.22 -0.12

To determine the relative values of the Raman tensor (for b ) 1) for the anionic state, we utilized eqs 2-6 to calculate the depolarization ratio as a function of a, to obtain the best reproduction of the depolarization ratio. We first assumed the Raman tensor to be solely governed by vibronic coupling to in-plane π f p* transitions of the peptide group, so that Rzz ) d ) 0. Table 4, section A lists the tensor elements obtained. The a values of GA, SA, and AS can be considered as identical in the limit of experimental accuracy. The values for AA are slightly larger, reflecting the somewhat lower depolarization ratio. For AV and VA, we obtained a values of ∼1 (i.e., nearly identical xx and yy components). Next, we considered out-ofplane contributions, which could result from coupling to n(COO-) f π* (peptide) charge transfer transitions (which would depend on the relative orientation of the carboxylate group)26,27,32 from some σ f σ* transition in the far UV and, in the case of histidine, from the π f π* transition of its imidazole ring. Thus, we assumed that Rzz ) d * 0. Anionic L-alanyl-L-alanine (AA) was used as our reference system for which we assumed d ) 0. The corresponding a and c values

(0.34 and -0.071) were then used as fixed parameters in our simulation for the remaining peptides. This approach is at least partially experimentally justified, because the weak intensity of the band from the symmetric carboxylate stretching mode in UV,43 as well as in the visible Raman spectra of AA15 in H2O, is indicative of weak oscillator strength of the respective n(COO-) f π* (peptide) charge transfer transition. Table 4, section B lists the d values obtained from this procedure. They are all small and negative for GA, SA, and AS, in accordance with what one would expect for a contribution from an n(COO-) f π*(peptide) transition for a COO- orientation similar to that of diglycine.27 VA and AV behave differently. The corresponding zz-component is positive and comparatively large. We can exclude virbronic coupling to the n(COO-) f π*(peptide) transition as a possible explanation, because the similarity of the visible Raman spectra of AA, VA, and AV in H2O in the wavenumber region between 1390 and 1420 cm-1 suggests a weak oscillator strength for this transition.36 Alternatively, one might invoke coupling to some intrinsic side chain transitions of valine, but this is unlikely, because something similar could be expected for alanine and serine. We therefore conclude that the low depolarization ratios observed for VA and AV do not reflect any substantial zz contribution to the Raman tensor. Instead, we interpret it as indicating that the ratio of xx and yy scattering is affected, in accordance with the first simulation. That would reflect a change of the principal axis of the Raman tensor (PART). Pajcini et al.27 obtained a value of 30° for the angle between the amide I PART and the carbonyl bond of diglycine with 488-nm excitation. The PART shows only a small out-of-plane component. The corresponding angle with the x-axis of our coordinate system is 96°. It is most likely somewhat larger for 442-nm excitation, but the y-component is certainly predominant as assumed for our simulation. A decrease of the depolarization ratio reflects an increase of the angle which gives rise to a PART with a larger x-component. If the Raman tensor is indeed dominated by contributions from NV1 and NV2 Franck-Condon transitions, this would reflect rotations of the corresponding electronic transition dipole moments. HA exhibits a particularly large depolarization ratio, which can be interpreted either in terms of a comparatively large negative zz-component (Table 4, section B) or a significantly reduced xx-component of the Raman tensor (Table 4, section A). In view of the fact that the electron circular dichroism (ECD) spectra of HA are indicative of electronic coupling between the π f π* transition of the imidazole ring41 and the NV1 transition of the peptide, it is much more likely that the amide I′ mode gets vibronically coupled to the imidazole transition, which brings about a zz-component of the Raman tensor. When taken together, our results show that the amino acid side chain on both sides of the peptide linkage can influence the Raman tensor of the amide I′ mode. Our data indicate a correlation between the amide I′ wavenumber and the depolarization ratio. That makes physical sense, because a wavenumber change is indicative of CO bond length variation and, correspondingly, a change of the normal-mode composition. Simulation of Amide I′ Band Profiles of Aβ1-28. We have recently developed an excitonic coupling model for amide I′ modes in polypeptides which allows us to simulate the corresponding band profiles in the Raman, IR, and VCD spectra of polypeptides.18 A similar approach has recently been utilized to simulate the IR spectra of globular proteins.44 We used this model to identify a significant polyproline II fraction in the monomeric state of the β-amyloid fragment Aβ1-28. Even though the results of the simulations allowed us to discriminate between

8204 J. Phys. Chem. B, Vol. 109, No. 16, 2005

Figure 10. Simulation of the IR, isotropic Raman, and anisotropic Raman amide I′ band profiles (solid lines) of the β-amyloid fragment Aβ1-28 obtained by employing a refined excitonic-coupling model based on the parameter values obtained from XA and AX peptides. The experimental profiles were obtained from spectra reported by Eker et al.16 We eliminated overlapping bands in the depicted spectral region by subtracting their band profiles from the spectrum. The dashed lines in all three spectra depict the result of an earlier simulation based on the assumption that the amide I′ vibrations of nonterminal peptide linkages are degenerated (the anisotropic band has been scaled up to account for the correct depolarization).16 The red line also visualizes a calculation based on the refined excitonic-coupling model, for which we disregarded a conformational influence on the intrinsic wavenumbers of the uncoupled amide I vibrations. The inset in the lower panel shows the depolarization dispersion in the spectral region of amide I′.

different extended conformations, the obtained agreement between experimental and theoretical profiles was only qualitative.16 The modeling was based on the simplifying assumption that the amide I modes of nonterminal peptide linkages have identical wavenumbers. As shown in Figure 10 (dashed line), the simulation particularly failed to reproduce the asymmetry of the IR and isotropic Raman profile. We have now performed a more refined simulation based on the results of the present study. The recently determined structural propensities of amino acid residues in water suggest that the most likely structure of the peptide is PPII with locally more β-strand-like segments (E3, H6, V12-H14, V18-F20, V24). To derive the local wavenumbers of the amide I vibrations, we used the recently obtained value of 1657 cm-1 for alanine15 as a reference value.

Measey et al. The wavenumbers for the other residues were obtained by calculating Ωx(Aβ) ) 1657 cm-1 + (ΩXA - ΩAA), where X denotes the residue and ΩXA is the wavenumber in the spectrum of anionic XA. For XG segments, we assumed an additional upshift of 5 cm-1 to account for the particularly high wavenumber obtained for anionic AG (Figure 3f). Moreover, we assumed a 10-cm-1 downshift of the local wavenumbers of residues in a β-strand conformation, in accordance with recent DFT studies.4 For the IR band simulation, we invoked the oscillator strengths visualized in Figure 5. Because we do not have the data for X ) Q and N (the synthesis of these peptides is extremely expensive), we utilized the respective wavenumbers and dipole strengths of E and D for these residues. For the calculation of the Raman profiles, we used the Raman tensors in Table 4 for the listed residues. The tensor elements of the remaining peptides were estimated by employing the correlation depicted in Figure 10. We assumed a Gaussian half-width of 14 cm-1 for all bands of the involved excitonic vibrations. This value appropriately represents the amide I′ half-widths obtained for the investigated peptides. A calculation based on this input yielded asymmetric IR and isotropic Raman profiles in very good agreement with the experimental band shapes (solid black line in the upper panel of Figure 10). We also improved the reproduction of the anisotropic Raman band, but some intensity at the high wavenumber side is still not accounted for. Interestingly, we obtained an even slightly better reproduction of the two Raman profiles by disregarding the 10-cm-1 downshift for the residues which adopt a β-strand-like conformation (red profiles in Figure 10). We were also able to account for the differences between the peak positions of the IR and the isotropic Raman band, a result which corroborates the validity of our calculation. For the Raman profiles, the simulation could not simultaneously account for the isotropic and anisotropic band profiles, owing to an underestimation of relative contribution of anisotropic scattering. We interpreted that as indicative of higher depolarization of amide I′ in a polypeptide context owing to the vibronic coupling to electronic transitions, which are also delocalized because of very strong dipole-dipole coupling between π f π* backbone transitions.45 Apparently, this brings about a z-component of the dipole moment of the electronic transitions to which amide I is vibronically coupled. This either increases or decreases the anisotropy of Raman scattering. To heuristically account for this effect, we consistently multiplied the Rxx ) a values of all tensors by a factor of 0.9 and additionally introduced an Rzz ) d component of -0.1. Altogether, however, the result of the revised simulation based on the spectroscopic information reported in this study can be considered a substantial improvement compared with what we had obtained previously with a model based on the degeneracy of amide I modes. This strongly underscores the relevance of the present study. Conclusions The data obtained in the present study show that, for alaninebased dipeptides, the wavenumber of amide I′ depends on both side chains framing the peptide linkage. As expected, the side chain influence is more pronounced for the side chain on the N-terminal side. The width of the wavenumber distributions obtained for the investigated AX and XA dipeptides is significant (approximately 7.7 and 11.9 cm-1, respectively) and cannot be neglected in any meaningful modeling of amide I′ band profiles of polypeptides. The (partial) correlation(s) between theoretically and experimentally obtained wavenumbers suggest that the amide I′ wavenumber depends on direct noncovalent

Influence of Side Chains on IR and Raman Spectra interactions between the peptide linkage and the side chain, and thus on the chain conformation. The correlation between the amide I′ wavenumber and the hydropathy index suggests that the C-terminal side chain affects amide I′ via solvent-mediated coupling. Both the C-terminal and the N-terminal side chains can influence the transition dipole moment, indicating that assuming it to be side-chain-independent might be an oversimplification, depending on the amino acid composition. The selected Raman spectra of this study suggest that some side chains affect the amide I′ Raman tensor. A combined application of the Raman and IR parameters was finally used to substantially improve the modeling of the amide I band profiles measured for the β-amyloid fragment Aβ1-28. Acknowledgment. The work carried out at the University of Puerto Rico, Rı´o Piedras, was financially supported from the NIH-COBRE II grant for the Center for Research in Protein Structure, Function and Dynamics (P20 RR16439-01) and from the Fondos Institucionales para la Investigacio´n of the University of Puerto Rico (20-02-2-78-514 to R.S.S.). T.M. amd A.H. did their research in the framework of the Drexel University Coop program. Financial support was obtained from a Petroleum Research grant to R.S.S. (37406-AC). References and Notes (1) Surewicz, W. K.; Mantsch, H. H. In Spectroscopic Methods for Determining Protein Structure in Solution; Havel, H. A., Ed; VCH: New York, 1996; p 135. (2) Infrared and Raman Spectroscopy of Biological Materials; Gremlich, H.-U., Yan, B., Eds.; Marcel Dekker: New York, 2000. (3) Krimm, S.; Bandekar, J. AdV. Protein Chem. 1986, 38, 181. (4) Ham, S.; Cha, S.; Choi, J.-H.; Cho, M. J. Chem. Phys. 2003, 119, 1451. (5) Torii, H.; Tasumi, M. J. Raman Spectrosc. 1998, 29, 81. (6) Choi, S.-J.; Ham, S.; Cho, M. J. Phys. Chem. B 2003, 107, 91329138. (7) Woutersen, S.; Hamm, P. J. Phys. Chem B 2000, 104, 11316. (8) Woutersen, S.; Hamm, P. J. Chem. Phys. 2001, 123, 9628. (9) Paul, C.; Wang, J.; Wimley, W. C.; Hochstrasser, R. H.; Axelsen, P. H. J. Am. Chem. Soc. 2004, 126, 5843. (10) Fang, C.; Wang, J.; Kim, Y. S.; Charnley, A. K.; Barber-Armstron, W.; Smith, A. A., III; Decatur, S. M.; Hochstrasser, R. M. J. Phys. Chem. B 2004, 108, 10415. (11) Kubelka, J.; Keiderling, T. J. Am. Chem. Soc. 2001, 123, 6142. (12) Bour, P.; Kubelka, J.; Keiderling, T. Biopolymers 2002, 65, 45. (13) Bour, P.; Keiderling, T. A. J. Mol. Struct. 2004, 675, 95. (14) Eker, F.; Cao, X.; Nafie, L.; Schweitzer-Stenner, R. J. Am. Chem. Soc. 2002, 124, 14330.

J. Phys. Chem. B, Vol. 109, No. 16, 2005 8205 (15) Schweitzer-Stenner, R.; Eker, F.; Griebenow, K.; Cao, X.; Nafie, L. J. Am. Chem. Soc. 2004, 126, 2768. (16) Eker, F.; Griebenow, K.; Schweitzer-Stenner, R. Biochemistry 2004, 43, 6893. (17) Huang, Q.; Schweitzer-Stenner, R. J. Raman Spectrosc. 2004 , 35, 586. (18) Schweitzer-Stenner, R. J. Phys. Chem. B 2004, 108, 16965. (19) Choi, J.-H.; Ham, S.; Cho, M. J. Phys. Chem. 2003, 107, 9132. (20) Ham, S.; Hahn, S.; Lee, C.; Kim, T.-K.; Kwak, K.; Cho, M. J. Phys. Chem. B 2004, 108, 9333. (21) Chen, X. G.; Schweitzer-Stenner, R.; Asher, S. A.; Mirkin, N.; Krimm, S. J. Phys. Chem. 1995, 99, 3074. (22) Sieler, G.; Schweitzer-Stenner, R. J. Am. Chem. Soc. 1997, 119, 1720. (23) Nafie, L. A.; Dukor, R. K.; Freedman, T. B. Vibrational Circular Dichroism. Handbook of Vibrational Spectroscopy; Chalmers, J. M., Griffiths, P. R., Eds.; John Wiley and Sons, Ltd.: Chichester, 2002. (24) Schweitzer-Stenner, R. J. Raman Spectrosc. 2001, 32, 711. (25) Schweitzer-Stenner, R. Biophys. J. 2002, 83, 523-532. (26) Chen, X.; Li, P.; Holtz, J.; Chi, Z.; Pajcini, V.; Asher, S.; Kelly, L. J. Am. Chem. Soc. 1996, 118, 9705-9715. (27) Pajcini, V.; Chen, X. G.; Bormett, R.; Geib, S. J.; Li, P.; Asher, S. A.; Lidiak, E. G. J. Am. Chem. Soc. 1996, 118, 9716. (28) Schweitzer-Stenner, R.; Eker, F.; Perez, A.; Griebenow, K.; Cao, X.; Nafie, L. Biopolymers (Peptide Science) 2003, 71, 558-568. (29) Glasoe, P. K.; Long, F. A. J. Phys. Chem. 1960, 64, 188-193. (30) Jentzen, W.; Unger, E.; Karvounis, G.; Shelnutt, J. A.; Dreybrodt, W.; Schweitzer-Stenner, R. J. Phys. Chem. 1996, 100, 14184-14191. (31) Throughout this paper, the terms “cationic”, “zwitterionic”, and “anionic” refer exclusively to the terminal charges. (32) Sieler, G.; Schweitzer-Stenner, R.; Holtz, J. S. W.; Pajcini, V.; Asher, S. A. J. Phys. Chem. B 1999, 103, 372. (33) Kwac, K.; Lee, H.; Cho, M. J. Chem. Phys. 2004, 120, 1477. (34) Venyaminov, S. Y.; Kalinin, N. N. Biopolymers 1990, 30, 1243. (35) Chirgadze, N. Y.; Fedorov, O. V.; Trushina, N. P. Biopolymers 1975, 14, 679. (36) Schweitzer-Stenner, R.; Eker, F.; Huang, Q.; Griebenow, K. H.; Mosz, P.; Kozlowski, P. M. J. Phys. Chem. B 2002, 106, 4294. (37) Kwac, K.; Cho, M. J. Chem. Phys. 2003, 119, 2247. (38) Kyte, J.; Doolittle, R. F. J. Mol. Biol. 1982, 157, 105-132. (39) Hahn, S.; Lee, H.; Cho, M. J. Chem. Phys. 2004, 121, 1849-1865. (40) Han, W.-G.; Jalkanen, K. J.; Elstner, M.; Suhai, S. J. Phys. Chem. B 1998, 102, 2587. (41) Eker, F.; Griebenow, K.; Cao, X.; Nafie, L. A.; Schweitzer-Stenner, R. Proc. Natl. Acad. Sci. U.S.A. 2004, 101, 10054. (42) Schweitzer-Stenner, R.; Sieler, G.; Christiansen, H. Asian J. Phys. 1998, 2, 287. (43) Asher, S. A.; Ianoul, A.; Nix, G.; Boyden, M. N.; Karnoup, A.; Diem, M.; Schweitzer-Stenner, R. J. Am. Chem. Soc. 2001, 123, 11775. (44) Brauner, J. W.; Flach, C.; Mendelsohn, R. J. Am. Chem. Soc. 2005, 127, 100. (45) Moffit, W. Proc. Natl. Acad. Sci. U.S.A. 1956, 42, 736.