Correlation between the Hydrogen-Bond Structures and the C O

In this study, the correlation between the H-bond structures and the C═O stretching (νC═O) frequencies of COOH groups was studied using density f...
0 downloads 0 Views 585KB Size
J. Phys. Chem. B 2008, 112, 6725–6731

6725

Correlation between the Hydrogen-Bond Structures and the CdO Stretching Frequencies of Carboxylic Acids as Studied by Density Functional Theory Calculations: Theoretical Basis for Interpretation of Infrared Bands of Carboxylic Groups in Proteins Ken-ichi Takei, Ryouta Takahashi, and Takumi Noguchi* Institute of Materials Science, UniVersity of Tsukuba, Tsukuba, Ibaraki 305-8573, Japan ReceiVed: February 7, 2008; ReVised Manuscript ReceiVed: March 13, 2008

Carboxylic groups (COOH) of Asp and Glu side chains often function as key components in enzymatic reactions, and identifying their H-bond structures in the active sites is essential for understanding the reaction mechanisms. In this study, the correlation between the H-bond structures and the CdO stretching (νCdO) frequencies of COOH groups was studied using density functional theory calculations. The νCdO frequencies and their shifts upon OH deuteration were calculated for model complexes of acetic acid and propionic acid H bonded at different sites with various compounds. Calculation results together with some experimental data showed that, upon direct H bonding at the CdO group, the νCdO frequencies downshift from the free value (1770-1780 cm-1 in an Ar matrix) to 1745-1760 cm-1, while H bonding at the OH hydrogen induce even larger downshifts to provide the frequencies at 1720-1745 cm-1. In contrast, when the COH oxygen is H-bonded, the νCdO frequencies upshift to 1785-1800 cm-1. In double and multiple H-bond forms, H-bonding effects at individual sites are basically additive, and complexes in which the CdO and the OH hydrogen are simultaneously H bonded exhibit significantly low νCdO frequencies at 1725-1700 cm-1, while complexes H bonded at the oxygen of the COH in addition to either at the CdO or the OH hydrogen exhibit medium frequencies of 1740-1765 cm-1. The νCdO frequencies linearly correlate with the CdO lengths, which are changed by H bonding at different sites. Upon OH deuteration, all the complexes showed νCdO downshifts mostly by ∼10 cm-1 and in some cases as large as ∼20 cm-1, and hence deuteration-induced downshifts can be a good indicator, irrespective of H-bond forms, for assignments of the νCdO bands of carboxylic groups. The results in this study provide the criteria for determining the H-bond structures of Asp and Glu side chains in proteins using their νCdO bands in Fourier transform infrared spectra. Introduction Carboxylic groups (COOH) in Asp and Glu play crucial roles in various enzymes.1–6 In particular, these residues are found as major components in proton-transfer pathways, because their (de)protonation states can be changed by pKa shifts induced by interactions with surrounding molecules in protein environments. Such proton-transfer pathways are formed by H-bond networks connecting polar residues and water molecules. A COOH group has three sites for H-bond interactions: the hydrogen atom of the OH as a donor site and the oxygen atoms of the CdO and COH as acceptor sites. Hence, this group can take various H-bonding forms in interactions with surrounding groups, the property that is crucial for the formation of proton-transfer pathways. Thus, identifying the H-bond structures of carboxylic groups in the active sites and proton pathways in proteins is essential for understanding the reaction mechanisms at the molecular level. Vibrational spectroscopy is a powerful method to investigate the structures and reactions of proteins. In particular, reactioninduced Fourier transform infrared (FTIR) difference spectroscopy has been extensively used to study the reactions of various enzymes.7–11 This method is complementary to X-ray crystallography and is very powerful to detect the protonation and H-bond structures of cofactors and amino acid side groups in the active sites of proteins. The CdO stretching (νCdO) bands * To whom correspondence should be addressed. Phone: +81-29-8535126. Fax: +81-29-853-4490 . E-mail: [email protected].

of the COOH groups of Asp and Glu are readily identified in FTIR spectra, because their frequency region of 1700-1790 cm-1 is rather isolated from the vibrations of other amino acids and main chains.12,13 Also, a downshift by ∼10 cm-1 upon OH deuteration1 can discriminate the νCdO band of COOH from the ester νCdO bands of lipids and cofactors that overlap the same frequency region. The νCdO frequencies of the COOH groups are known to be sensitive to H-bond interactions.10,13 It has been reported that the νCdO frequencies downshift by ∼25 cm-1 upon formation of a single H-bond and by ∼50 cm-1 by strong H-bonds.10,13 In addition, it was proposed that νCdO bands underwent small downshifts or even upshifts upon OH deuteration when a H-bond was formed at the OH hydrogen.2 However, clear criteria for interpretation of the νCdO bands to identify the H-bonded sites among many possibilities (Hbonding at the CdO, the OH hydrogen, and the COH oxygen, and these combinations) have yet to be established. In this study, we have systematically investigated the effects of H-bonding at the different sites of a COOH group on the νCdO frequencies using the density functional theory (DFT) calculations. Vibrational analyses were performed on model H-bonded complexes of acetic acid and propionic acid in different H-bond forms with various H-bond donors/acceptors, most of which represent amino acid side chains. The DFT analysis using model H-bonded complexes is a quite powerful method for studying the effects of H-bonding, because it is almost free to manipulate the H-bonding sites and partners.14–17 Vibrational analyses of carboxylic acids using quantum chemical

10.1021/jp801151k CCC: $40.75  2008 American Chemical Society Published on Web 05/02/2008

6726 J. Phys. Chem. B, Vol. 112, No. 21, 2008

Takei et al.

calculations have been performed by many researchers but mostly on a monomer and a cyclic dimer.18–22 Gao and Leung16 studied the H-bond interactions in acetic acid monohydrates and dihydrates using the DFT method, but systematic calculations focusing on different H-bond forms with various H-bond partners have not been performed yet. The results obtained in this study provided general criteria for determining the H-bonded structures of carboxylic side groups in proteins using the νCdO bands in FTIR spectra. Materials and Methods Molecular orbital calculations were performed using the Gaussian03 program package.23 The B3LYP functional24,25 with the 6-31++G(d,p) basis set was used for geometry optimization and subsequent calculation of the normal-mode frequencies. The frequencies of model complexes of deuterated acetic and propionic acids (CH3COOD, CD3COOH, CD3COOD, C2H5COOD) were calculated using the optimized geometries of nondeuterated complexes. The potential energy distributions (PED) of free acetic acid were calculated using a modified program of NCTB.26,27 For FTIR measurements, acetic acid-d3 (CD3COOH: >99 atom% D, Isotec, Miamisburg, OH), acetic acid-d4 (CD3COOD: >99 atom% D, Cambridge Isotope Laboratories, Andover, MA), propionic acid, and propionic acid-OD (C2H5COOD: 98 atom% D, Icon Services Inc., Mt. Marion, NY) were dissolved in dioxane and pyridine, which were dried beforehand with molecular sieves 3A, at a concentration of 1 M. For aqueous solutions, acetic acid-d3 and propionic acid were dissolved in H2O or D2O (99.9 atom% D, Aldrich) at a concentration of 100 mM. FTIR spectra were measured on JEOL JIR6500 and Bruker IFS-66/S spectrophotometers using an attenuated total reflection (ATR) accessory (DurasamplIR II, Smiths Detection, Danbury, CT) with a three reflection silicon prism (3 mm in diameter). FTIR spectra were measured at room temperature at 2-cm-1 resolution. For H2O solutions, the spectrum of H2O was subtracted from the original spectra to remove the H2O bending band around 1640 cm-1. Results Figure 1 shows the optimized structures of model complexes of acetic acid with a single H-bond at the oxygen atom of the CdO (A: CdO · · · ), the hydrogen atom of OH (B: OH · · · ), and the oxygen atom of the COH (C: CO(H) · · · ). H-bond donors and acceptors adopted for calculations were water (1, 10), ethanol (2, 11), p-cresol (3), acetamide (4, 6, 12), imidazole (5, 7, 13), dioxane (8), and pyridine (9). Ethanol and p-cresol represent the side chains of alcoholic (Ser, Thr) and phenolic (Tyr) amino acids; acetamide corresponds to amide side chains (Asn, Gln); imidazole represents the side chain of His. Complexes with dioxane and pyridine were also calculated to compare with experimental data (see below). It is noted that in most of the complexes, weak CH · · · O interactions together with an intended strong H bond determine the orientation of the H-bond partner and hence the H-bond angle is often slightly distorted (e.g., complexes 1, 10, and 11). Figure 2 shows the optimized structures of double H-bonded complexes in the forms of CdO · · · /OH · · · (A), CdO · · · / CO(H) · · · (B), and OH · · · /CO(H) · · · (C) and of a multiple H-bonded complex in the CdO · · · /OH · · · /CO(H) · · · form (D). In the double H-bonded complexes, two H-bond partners were selected from those used for single H-bonded complexes: two water molecules (14, 18), imidazole/water (16, 19, 21), aceta-

Figure 1. Optimized structures of model complexes of acetic acid with a single H bond. (A) Complexes H-bonded at the CdO with water (1), ethanol (2), p-cresol (3), acetamide (4), and imidazole (5). (B) Complexes H-bonded at the hydrogen atom of the OH with acetamide (6), imidazole (7), dioxane (8), and pyridine (9). (C) Complexes H-bonded at the oxygen atom of the COH with water (10), ethanol (11), acetamide (12), and imidazole (13). Hydrogen, carbon, oxygen, and nitrogen atoms are expressed with white, gray, red, and blue colors, respectively.

mide/water (20), and water/pyridine (17). In complex 15, the amide group of acetamide functions as a hydrogen bond donor (-NH2) and an acceptor (CdO) simultaneously. A multiple H-bonded complex (22) was formed with three water molecules and can be a model of acetic acid in an aqueous solution. Table 1 shows H-bond-induced changes in the CdO, CsO, and OsH lengths (rCdO, rCsO, rOsH, respectively) in the optimized geometries of the model complexes with respect to the values of free acetic acid: 1.213, 1.360, and 0.972 Å for rCdO, rCsO, and rOsH, respectively. These calculated bond lengths of free acetic acid well reproduced the experimental values of gaseous acetic acid: 1.214, 1.364, and 0.97 Å for rCdO, rCsO, and rOsH, respectively.27–28 When the H bond is formed at the CdO, the rCdO was lengthened, as expected, by 0.006-0.008 Å. Interestingly, the rCsO was shortened to a larger extent (0.008-0.011 Å), whereas the rOsH was little affected. Upon H bonding at the OH hydrogen, the rOsH was significantly increased by 0.022-0.039 Å, and as a result the rCsO was shortened (by 0.018-0.023 Å). Notably, the rCdO increased by 0.008-0.010 Å, the extent which is even larger than the increase by direct interaction at the CdO. In the CO(H) · · · form, the rCsO was lengthened by 0.012-0.016 Å, while the rOsH was almost unaffected. The rCdO was shortened by 0.002-0.004 Å, contrary to the effects of H-bonding at the CdO and OH. The effects of double and multiple H-bonds on the bond lengths were basically explained by additive changes by

Hydrogen-Bond Structure of Carboxylic Acid

Figure 2. Optimized structures of model complexes of acetic acid with double and multiple H bonds. (A) Complexes doubly H bonded at the CdO and the hydrogen atom of the OH with two water molecules (14), acetamide (15), imidazole and water (16), and water and pyridine (17). (B) Complexes doubly H bonded at the CdO and the oxygen atom of the COH with two water molecules (18) and imidazole and water (19). (C) Complexes doubly H bonded at the hydrogen and oxygen atoms of the COH with acetamide and water (20) and imidazole and water (21). (D) Complexes with multiple H bonds with three water molecules (22). Hydrogen, carbon, oxygen, and nitrogen atoms are expressed with white, gray, red, and blue colors, respectively.

individual H-bonds (Table 1). For example, upon CdO · · · / OH · · · double H bonds, the rCdO increased by 0.013-0.019 Å, in good agreement with the added sum (0.014-0.018 Å) of the increases by the CdO · · · (by 0.006-0.008 Å) and OH · · · (by 0.0008-0.0010 Å) H bonds. Also, the rCdO increases by 0.006-0.007 Å in the OH · · · /CO(H) · · · complexes result from the increases by the OH · · · H bond (by 0.008-0.010 Å) and the decreases by the CO(H) · · · H bond (by 0.002-0.004 Å). Table 2 shows the calculated CdO stretching (νCdO) frequencies of free acetic acid and its model H-bonded complexes. The frequencies were all scaled with a scaling factor of 0.9764 to adjust the calculated frequency of free acetic acid to 1779 cm-1, the experimental value in an Ar matrix.29 In the CdO · · · single H-bonded complexes, the νCdO frequencies were found at 1748-1757 cm-1 as a result of downshifts by 22-31 cm-1. In the complexes with an OH · · · form, even larger downshifts by 35-47 cm-1 were obtained and the νCdO vibration occurred at 1732-1744 cm-1. On the contrary, in the CO(H) · · · form, the νCdO frequencies upshifted by 9-17 cm-1, providing the values of 1788-1796 cm-1. The H-bonding effects on the νCdO shifts in the double and multiple H-bonded complexes were found additive again (Table

J. Phys. Chem. B, Vol. 112, No. 21, 2008 6727 2) similarly to the case of the bond lengths (Table 1). In the CdO · · · /OH · · · complexes, the downshifts were 57-79 cm-1, which is very similar to 58-78 cm-1 as an added sum of the downshifts by the CdO · · · (23-31 cm-1) and OH · · · (35-47 cm-1) H bonds. As a result, the νCdO vibrations were found at significantly low frequencies of 1700-1722 cm-1. In the CdO · · · /CO(H) · · · complexes, the downshifts were 16-17 cm-1, which are similar to the values of 14-15 cm-1 obtained by addition of the shifts by individual single H-bonds with the corresponding H-bond donors (H2O/H2O and imidazole/H2O), while the OH · · · /CO(H) · · · complexes showed downshifts by 30-38 cm-1, which are also similar to 29-34 cm-1 as additional shifts by corresponding acceptor/donor (acetamide/ H2O and imidazole/H2O). In addition, the CdO · · · /OH · · · / CO(H) · · · complex H bonded with three water molecules showed a downshift by 57 cm-1, which is smaller than the downshift (65 cm-1) of the double (CdO · · · /OH · · · ) H-bond complex with two water molecules by 8 cm-1, due to the effect of opposite shift by additional CO(H) · · · H-bonding. The νCdO shifts upon deuteration of OH (∆OD) were also calculated for free acetic acid and its H-bonded complexes. Free acetic acid showed a downshift by 7 cm-1, which is caused by a weak coupling of the COH bending (δCOH) vibration and its decoupling by deuteration. All of the calculated H-bonded complexes also showed negative deuteration shifts. In the single H-bond complexes in the CdO · · · and CO(H) · · · forms, the downshifts (7-9 cm-1) were mostly similar to the free acetic acid, while in the OH · · · complexes, slightly larger downshifts (7-13 cm-1) were obtained. These larger downshifts are mainly caused by the upshifting tendency of the δCOH vibration by a strong H-bond at the OH hydrogen resulting in a larger coupling with the νCdO vibration. For example, complex 9 having a H-bond with pyridine (∆OD ) -13 cm-1) showed a δCOH vibration at 1468 cm-1, while a free acetic acid (∆OD ) -7 cm-1) and complex 1 having a H bond with a water at the CdO (∆OD ) -8 cm-1) showed δCOH vibrations at 1176 and 1289 cm-1, respectively. The double and multiple H-bond complexes also showed similar or larger downshifts (6-20 cm-1). Significantly large downshifts in some complexes (e.g., complexes 17, 21, 22) are due to even larger upshifts of the δCOH vibrations by H-bonding at the CdO or CO(H) in addition to a strong H bond at the OH. For example, complex 21 having H bonds with imidazole and H2O (∆OD ) -20 cm-1) showed a very high δCOH frequency of 1487 cm-1. Figure 3 (solid lines) shows the νCdO bands in the FTIR spectra of acetic acid-d3 (CD3COOH) (a, d) and propionic acid (C2H5COOH) (b, c, e) dissolved in dioxane (a, b), pyridine (d), and water (d, e). Spectra of their deuterated (OD) species (CD3COOD, and C2H5COOD) were also presented (dotted lines). Acetic acid-d3 was used instead of unlabeled acetic acid, because the νCdO band of the latter species often splits by the Fermi resonance most likely with the νC-C mode.30–32The spectra of acetic acid-d3 in pyridine were not involved in Figure 3, because the νCdO band also spits into two peaks. The relatively low concentration of 100 mM was used for aqueous solutions to avoid dimerization.32,33 The observed νCdO and ∆OD values are summarized in Table 3 together with the data of free acetic acid, acetic acid-d3, and propionic acid in an Ar matrix in the literature.22,29 In this table, the experimental data were also compared with the calculated values of corresponding model compounds. The H-bonded complexes of acetic acid-d3 and propionic acid with dioxane, pyridine, and water were modeled in the same H-bond confor-

6728 J. Phys. Chem. B, Vol. 112, No. 21, 2008

Takei et al.

TABLE 1: Changes in the Calculated Lengths (Å) of the CdO, CsO, and OsH Bonds of Acetic Acid by H-Bond Formationa H-bond form CdO · · ·

OH · · ·

CO(H) · · ·

CdO · · · /OH · · ·

CdO · · · /CO(H) · · · OH · · · /CO(H) · · · CdO · · · /OH · · · /CO(H) · · ·

donor/acceptor (complex no.)b

∆rCdOc

∆rCsOd

∆rOsHe

H2O (1) ethanol (2) p-cresol (3) acetamide (4) imidazole (5) acetamide (6) imidazole (7) dioxane (8) pyridine (9) H2O (10) ethanol (11) acetamide (12) imidazole (13) 2H2O (14) acetamide (15) imidazole/H2O (16) H2O/pyridine (17) 2H2O (18) imidazole/H2O (19) acetamide/H2O (20) imidazole/H2O (21) 3H2O (22)

+0.007 +0.007 +0.008 +0.006 +0.007 +0.010 +0.010 +0.008 +0.010 -0.003 -0.002 -0.002 -0.004 +0.016 +0.017 +0.013 +0.019 +0.004 +0.005 +0.007 +0.006 +0.017

-0.008 -0.007 -0.011 -0.009 -0.011 -0.023 -0.023 -0.018 -0.023 +0.013 +0.012 +0.016 +0.013 -0.032 -0.034 -0.027 -0.033 +0.005 +0.002 -0.013 -0.018 -0.016

+0.000 +0.000 +0.001 +0.001 +0.001 +0.025 +0.037 +0.022 +0.039 +0.000 +0.000 +0.000 +0.001 +0.033 +0.037 +0.022 +0.045 +0.001 +0.001 +0.030 +0.057 +0.024

a Calculations were performed at the B3LYP/6-31++G(d,p) level. b The number of the complexes whose optimized structures are shown in Figure 1. c Change in the CdO bond length from that of free acetic acid (1.213 Å). d Change in the CsO bond length from that of free acetic acid (1.360 Å). e Change in the OsH bond length from that of free acetic acid (0.972 Å).

TABLE 2: Calculated CdO Stretching Frequencies (cm-1) and Deuteration Shifts of H-Bonded Complexes of Acetic Acida H-bond form free CdO · · ·

OH · · ·

CO(H) · · ·

CdO · · · /OH · · ·

CdO · · · /CO(H) · · · OH · · · /CO(H) · · · CdO · · · /OH · · · /CO(H) · · ·

donor/acceptor (complex no.)b

νCdO (∆H-bondc)

∆ODd

H2O (1) ethanol (2) p-cresol (3) acetamide (4) imidazole (5) acetamide (6) imidazole (7) dioxane (8) pyridine (9) H2O (10) ethanol (11) acetamide (12) imidazole (13) 2H2O (14) acetamide (15) imidazole/H2O (16) H2O/pyridine (17) 2H2O (18) imidazole/H2O (19) acetamide/H2O (20) imidazole/H2O (21) 3H2O (22)

1779 (0) 1752 (-27) 1752 (-27) 1748 (-31) 1757 (-22) 1753 (-26) 1738 (-41) 1733 (-46) 1744 (-35) 1732 (-47) 1791 (+12) 1790 (+11) 1788 (+9) 1796 (+17) 1714 (-65) 1718 (-61) 1722 (-57) 1700 (-79) 1763 (-16) 1762 (-17) 1749 (-30) 1741 (-38) 1722 (-57)

-7 -8 -8 -8 -7 -8 -7 -11 -10 -13 -8 -8 -9 -8 -11 -6 -7 -16 -9 -9 -8 -20 -19

a

Calculations were performed at the B3LYP/6-31++G(d,p) level. The computed frequencies were scaled with a uniform scaling factor of 0.9764 so as to adjust the νCdO of free acetic acid to the experimental value of 1779 cm-1 in an Ar matrix.29 b The number of the complexes whose optimized structures are shown in Figure 1. c Difference from the νCdO of free acetic acid. d Frequency shift upon deuteration of OH.

mations as the complexes 8, 9, and 22, respectively, of acetic acid (Figures 1 and 2). Free CD3COOH and C2H5COOH in an Ar matrix showed slightly lower νCdO frequencies by 5 and 3 cm-1, respectively, in comparison with free CH3COOH (1779 cm-1).22,29 This tendency was well reproduced in calculation; the calculated νCdO frequencies of free CD3COOH and C2H5COOH was lower than the frequency of free CH3COOH by 4 and 5 cm-1, respectively. The ∆OD shifts were reported to be -9 and -8 cm-1 for CH3COOH and CD3COOH, 29 which are in good agreement with the calculated shifts (-7 cm-1). FTIR spectra of CD3COOH and C2H5COOH in dioxane showed peaks at the

similar positions of 1737 and 1740 cm-1 (parts a and b of Figure 3), respectively, which were well reproduced by the model H-bonded complexes of these species with dioxane showing 1740 cm-1. The observed ∆OD shifts were -7 and -9 cm-1, which also agree well with the calculated values of -10 cm-1. C2H5COOH in pyridine showed a νCdO frequency at 1720 cm-1 with a ∆OD of -8 cm-1 (Figure 3c), while the calculated νCdO and ∆OD values for the C2H5COOH-pyridine complex were 1729 and -13 cm-1, respectively. Although the discrepancy between the experimental and calculated values were a little larger than the results of dioxane, the tendencies of a significantly low νCdO frequency relative to the free form and

Hydrogen-Bond Structure of Carboxylic Acid

Figure 3. CdO stretching bands of the FTIR spectra of acetic acid-d3 and propionic acid in organic and aqueous solutions. (a) CD3COOH (solid line) and CD3COOD (dotted line) in dioxane; (b) C2H5COOH (solid line) and C2H5COOD (dotted line) in dioxane; (c) C2H5COOH (solid line) and C2H5COOD (dotted line) in pyridine; (d) CD3COOH in H2O (solid line) and CD3COOD (dotted line) in D2O; (e) C2H5COOH in H2O (solid line) and C2H5COOD in D2O (dotted line). The sample concentrations were 1 M in dioxane and pyridine solutions and 100 mM in H2O and D2O. A water band around 1640 cm-1 was subtracted from the original spectra for spectra d and e (solid lines).

a ∆OD shift by approximately -10 cm-1 were satisfactorily reproduced. CD3COOH and C2H5COOH in aqueous solutions showed relatively broad bands at 1713 and 1717 cm-1, respectively (parts d and e of Figure 3). Although a COOH group may take various H-bond conformations in water, the observed νCdO frequencies well agree with the calculated ones (1716 and 1721 cm-1 for CD3COOH and C2H5COOH, respectively) of the model complexes H-bonded with three water molecules (Figure 2, complex 22). In addition, the observed large ∆OD shifts by -17 and -18 cm-1 for CD3COOH and C2H5COOH, respectively, were reproduced in calculations as shifts by -24 and -19 cm-1, respectively. The above calculations were all performed in a vacuum, and the dielectric effect of the solvent was not taken into account. However, calculations involving solvent effects using the selfconsistent isodensity polarized continuum model (SCI-PCM) showed much lower frequencies than the experimental ones, indicating that the interactions with solvent molecules were overestimated. It is thus possible that the explicit interactions with solvent molecules as H-bonded partners can also represent a large part of the dielectric effect of the solvent. Discussion The present DFT calculations on the model H-bonded complexes of acetic acid (Figures 1 and 2) showed the effects of H-bonding at the different sites on the νCdO frequency of the COOH group (Table 2). When a single H-bond is formed at the CdO oxygen, the νCdO frequencies were downshifted by 22-31 cm-1 (Table 2), as naturally expected from the direct

J. Phys. Chem. B, Vol. 112, No. 21, 2008 6729 interaction on the CdO bond. Interestingly, in the complexes singly H-bonded at the OH hydrogen, the νCdO frequencies exhibited even larger downshifts (35-47 cm-1). On the contrary, when the oxygen of the COH is H-bonded, the νCdO was upshifted by 9-17 cm-1. The νCdO shifts by double and multiple H-bond formation showed additive effects of the single H bonding at individual sites (Table 2). Thus, the complexes with a CdO · · · /OH · · · form showed larger downshifts (61-79 cm-1) than each single H-bond form, whereas the CdO · · · / CO(H) · · · and OH · · · /CO(H) · · · forms showed smaller downshifts than the CdO · · · and OH · · · forms, respectively, due to the upshifting nature of the CO(H) · · · H-bond. The observed tendencies of the νCdO downshifts in the CdO · · · and CdO · · · /OH · · · forms and the upshifts in the CO(H) · · · form are consistent with the previous results of DFT calculations on acetic acid monohydrates and dihydrates.16 As shown in Figure 4, the νCdO frequencies are linearly correlated with the rCdO values (Table 1), irrespective of the H-bond types. This observation indicates that the νCdO mode remains to be a highly localized vibration even in different H-bond forms, and the νCdO shift is simply caused by the change in the force constant of the CdO bond. The change in the rCdO by H bonding at the sites other than the CdO can be rationalized by conjugation of the CdO and C-O bonds. When a H bond is formed at the OH hydrogen, the rOsH is lengthened, and as a result, the rCsO is shortened. Under this situation, the electron on the CdO bond is delocalized to the CsO bond, providing lengthened rCdO. On the contrary, when the COH oxygen is H-bonded, the rCsO is increased, and then the conjugation is decreased to shorten the rCdO. The δCOH vibration is weakly coupled to the νCdO vibration (4% PED in free acetic acid). This coupling upshifts the νCdO frequency and hence cause a downshift upon decoupling by deuteration. Free acetic acid and most of the complexes exhibited νCdO downshifts by ∼10 cm-1 upon deuteration, while some complexes showed even larger downshifts (16-20 cm-1) due to stronger couplings with the upshifted δCOH vibrations (Table 2). Credibility of the calculated results was verified by comparison of the experimental data of acetic acid, acetic acid-d3, and propionic acid in an Ar matrix,22,29 dioxane, pyridine, and aqueous solutions (Figure 3) with the calculated data of corresponding model compounds (Table 3). Both experimental νCdO frequencies and deuteration shifts were in good agreement with the calculation results. Also, the tendency of the νCdO upshift upon H bonding at the COH oxygen (Table 2) is consistent with the previous matrix isolation study in which the νCdO frequency of methylacetate was upshifted by H bonding at the oxygen atom of COMe.34 The νCdO frequencies of propionic acid were very similar to those of acetic acid-d3 both in experiments and calculations (Table 3), while they were slightly lower (1-5 cm-1) than those of unlabeled acetic acid. The νCdO frequency of propionic acid in an Ar matrix was lower than that of acetic acid by 3 cm-1, and calculated frequencies for free propionic acid and its complexes with dioxane, pyridine, and three water molecules were lower than the corresponding models of acetic acid by 5, 4, 3, and 1 cm-1, respectively (Tables 2 and 3). Thus, the frequency gap between acetic acid and propionic acid seems to be smaller in stronger or more H-bonding formation. Although it has been suggested that acetic acid has a much higher νCdO frequency (by ∼10 cm-1) than longer carboxylic acids,31 the present study showed that theoretically (and experimentally in an Ar matrix) the frequency difference is at most 5 cm-1. The discrepancy could

6730 J. Phys. Chem. B, Vol. 112, No. 21, 2008

Takei et al.

TABLE 3: Experimental νCdO Frequencies (cm-1) and Deuteration Shifts of Acetic Acid, Acetic Acid-d3, and Propionic Acid in Comparison with the Calculated Values calcda

exptl chemical species CH3COOH CD3COOH C2H5COOH CD3COOH C2H5COOH C2H5COOH CD3COOH C2H5COOH

solvent Ar matrix Ar matrix Ar matrix dioxane dioxane pyridine H2O H2 O

νCdO

H-bond form free free free OH · · · OH · · · OH · · · CdO · · · /OH · · · /CO(H) · · · CdO · · · /OH · · · /CO(H) · · ·

c

1779 1774c 1776d 1737 1740 1720 1713 1717

∆ODb

νCdO

∆ODb

-9c

1779 1773 1774 1740e 1740f 1729g 1716h 1721i

-7 -7 -7 -10e -10f -13g -24h -19i

-8c n.d. -7 -9 -8 -17 -18

a Calculations were performed at the B3LYP/6-31++G(d,p) level and computed frequencies were scaled with a scaling factor of 0.9764. Frequency shifts upon deuteration of OH. c Experimental data from FTIR spectra by Berney et al.29 d Experimental data from FTIR spectra by Sander and Gantenberg.22 e Calculated for the CD3COOH-dioxane complex in the same H-bond conformation as complex 8. f Calculated for the C2H5COOH-dioxane complex in the same H-bond conformation as complex 8. g Calculated for the C2H5COOH-pyridine complex in the same H-bond conformation as complex 9. h Calculated for the complex of CD3COOH H-bonded with three water molecules in the same H-bond conformation as complex 22. i Calculated for the complex of C2H5COOH H-bonded with three water molecules in the same H-bond conformation as complex 22. b

Figure 5. The νCdO frequency ranges of COOH groups for different H-bond forms estimated by DFT calculations and experimental data. The ranges are expressed by a 5-cm-1 digit involving the calculated and experimental values. Figure 4. Correlation between the νCdO frequency and the CdO length for the optimized structures of the model H-bonded complexes of acetic acid. Individual symbols express different H-bond forms: free acetic acid (open circle); CdO · · · (closed circle); OH · · · (open triangle); CO(H) · · · (closed triangle); CdO · · · /OH · · · (open square); CdO/CO(H) · · · (closed square); OH · · · /CO(H) · · · (plus); CdO · · · / OH · · · /CO(H) · · · (cross).

partially arise from the fact that the νCdO band of unlabeled acetic acid often shifts by the Fermi resonance with the νCC vibration.30–32 Thus, it is considered that the relationship between the νCdO frequencies and the H-bond forms obtained by the calculations on the acetic acid complexes can be applied to longer carboxylic acids and hence for Asp and Glu side chains in proteins. The νCdO frequency ranges for different H-bond forms deduced from the calculations and experimental data presented in this study are summarized in Figure 5. The ranges were expressed by a 5-cm-1 digit involving the calculated and experimental values. A free form shows the νCdO frequency at 1770-1780 cm-1 in an Ar matrix (Table 3). When a single H-bond is formed at the CdO, the OH hydrogen, and the COH oxygen, the νCdO frequency occurs at 1745-1760, 1720-1745, and 1785-1800 cm-1, respectively. In the CdO · · · /OH · · · double H-bond and CdO · · · /OH · · · /CO(H) · · · multiple H-bond forms, a significantly low νCdO frequency of 1700-1725 cm-1 is predicted. Finally, in the CdO · · · /CO(H) · · · and OH · · · / CO(H) · · · double H-bond forms, the νCdO vibration appears at a medium frequency of 1740-1765 cm-1. A downshift by ∼10 cm-1 (sometimes larger) upon deuteration is a good indicator, irrespective of the H-bond structures, for the νCdO assignment of carboxylic acid in the 1700-1800-cm-1 region,

where the ester CdO groups of lipids and some cofactors (e.g., chlorophyll) show bands. This is in contrast to the proposal by Maeda et al.2 that a COOH group H bonded at the OH hydrogen shows almost no shift or even a slight upshift by deuteration. This discrepancy seems to arise from misinterpretation of the acetic acid spectra, which exhibit a strong Fermi resonance splitting,30–32 as has been pointed out by Dioumaev and Braiman.31 We used acetic acid-d3 and propionic acid for FTIR measurements to avoid this problem. Some caution may be necessary in the usage of the above criteria to determine the H-bond structures of carboxylic groups in real proteins: (1) The frequency ranges in Figure 5 were obtained from a limited number of model complexes and experimental data, and hence they may be actually wider than suggested. (2) The frequencies were calculated for optimized H-bonded structures. In real proteins, however, the positions and orientations of side chains are highly restricted, and hence the H-bonds may be weaker than those in proper conformations. In this case, the frequency shifts by H-bonding should be smaller than expected. (3) Only neutral compounds were assumed as H-bond partners to make the criteria in Figure 5. This assumption may be appropriate because the protonated forms of Asp and Glu most likely exist in rather hydrophobic protein environments. In addition, when a cationic amino acid such as Lys and Arg interacts with a carboxylic side chain, the latter will be deprotonated to form a salt bridge. However, the situation in which a protonated carboxylic acid is H bonded to a cationic side group could be realized in a low pH medium. Interaction with a cationic H-bond donor sometimes forms a very strong H bond because of a positive charge on the hydrogen atom.

Hydrogen-Bond Structure of Carboxylic Acid For example, in the model complex in which the NH3+ group of protonated methylamine (the model of Lys) is H bonded to the CdO of acetic acid, the νCdO frequency was calculated to be 1691 cm-1. This value is much lower than the frequencies of the CdO · · · complexes H-bonded with neutral groups (Table 2, Figure 5). In addition, when a carboxylate anion (COO-) is H-bonded to the OH of a COOH group, the latter was reported to show a very low νCdO frequency of ∼1680 cm-1 because of a significant shift of the hydrogen atom to the COO- side.3 Thus, careful treatment is necessary when H-bond partners are charged groups. (4) It has been shown that the νCdO frequencies of carboxylic acids in non-H-bonding solvents exhibit an inverse correlation with dielectric constants () and can decrease to 1740 cm-1 in highly polar solvents.31 However, polar groups in proteins usually have a H-bonding character and hence a truly non-H-bonding environment may be represented by  ) ∼2 ( ) 2.077 for n-hexane,  ) 2.283 for benzene35). According to the correlation between the νCdO frequency and the Onsager’s parameter, 2( - 1)/(2 + 1), proposed by Dioumaev and Braiman, 31  ) 2 gives 1763 cm-1 as νCdO. Thus, COOH groups free from H-bonding in hydrophobic protein environments may exhibit the νCdO at 1760-1770 cm-1. One example of application of the obtained criteria (Figure 5) to a biological system is given below. In the photosynthetic reaction center of the purple bacterium Rhodobacter sphaeroides, light-induced electron transfer takes place, and the secondary quinone electron acceptor (QB) is reduced. It has been known that upon QB- formation, a nearby Glu residue (Glu-L212) is protonated and exhibits a positive νCdO peak at 1728 cm-1 in QB-/QB FTIR difference spectra.4 According to the criteria in Figure 5, this frequency is involved in the frequency range of a single H-bond at the OH hydrogen. In the X-ray structure of the reaction center with QB- (2.6 Å resolution with no hydrogen atoms),36 one water oxygen is located at 3.13 Å from one of the two oxygen atoms of the Glu-L212 side chain with a ∠C-O · · · O angle of 134°, whereas no H-bond partner is found near the other oxygen atom. Hence, this X-ray structure is in good agreement with the predicted H-bond structure from the νCdO frequency. Thus, it is indicated that upon QB- formation, the COO- group of Glu-L212 is protonated to provide a H-bond to a water molecule, which makes a new H-bond network most likely required for the protonation of QB upon its double reduction. Acknowledgment. The authors thank Dr. Koji Hasegawa (AdvanceSoft Corporation) for his technical support especially in the early stage of this research. This study was supported by Grants-in-Aid for Scientific Research (17GS0314 and 18570145) from the Ministry of Education, Culture, Sports, Science and Technology. References and Notes (1) Engelhard, M.; Gerwert, K.; Hess, B.; Siebert, F. Biochemistry 1985, 24, 400.

J. Phys. Chem. B, Vol. 112, No. 21, 2008 6731 (2) Maeda, A.; Sasaki, J.; Shichida, Y.; Yoshizawa, T.; Chang, M.; Ni, B.; Needleman, R.; Lanyi, J. K. Biochemistry 1992, 31, 4684. (3) Iliadis, G.; Zundel, G.; Brzezinski, B. FEBS Lett. 1994, 352, 315. (4) Nabedryk, E.; Breton, J.; Hienerwadel, R.; Fogel, C.; Ma¨ntele, W.; Paddock, M. L.; Okamura, M. Y. Biochemistry 1995, 34, 14722. (5) Barth, A.; Kreutz, W.; Ma¨ntele, W. J. Biol. Chem. 1997, 272, 25507. (6) Okuno, D.; Iwase, T.; Shinzawa-Itoh, K.; Yoshikawa, S.; Kitagawa, T. J. Am. Chem. Soc. 2003, 125, 7209. (7) Ma¨ntele, W. Trends Biochem. Sci. 1993, 18, 197. (8) Vogel, R.; Siebert, F. Curr. Opin. Chem. Biol. 2000, 4, 518. (9) Zscherp, C.; Barth, A Biochemistry 2001, 40, 1875. (10) Barth, A. Biochim. Biophys. Acta 2007, 1767, 1073. (11) Noguchi, T. Coord. Chem. ReV. 2008, 252, 336. (12) Venyaminov, S. Yu.; Kalnin, N. N Biopolymers 1990, 30, 1243. (13) Barth, A. Prog. Biophys. Mol. Biol. 2000, 74, 141. (14) Gonzalez, L.; Mo, O.; Yanez, M. J. Comput. Chem. 1997, 18, 1124. (15) O’Malley, P. J. Biochim. Biophys. Acta 2002, 1553, 212. (16) Gao, Q.; Leung, K. T. J. Chem. Phys. 2005, 123, 074325. (17) Takahashi, R.; Noguchi, T. J. Phys. Chem. B 2007, 111, 13833. (18) Nakabayashi, T.; Kosugi, K.; Nishi, N. J. Phys. Chem. A 1999, 103, 8595. (19) Burneau, A.; Ge´nin, F.; Quile`s, F. Phys. Chem. Chem. Phys. 2000, 2, 5020. (20) Mac¸oˆas, E. M. S.; Khriachtchev, L.; Fausto, R.; Ra¨sa¨nen, M. J. Phys. Chem. A 2004, 108, 3380. (21) Lewandowski, H.; Koglin, E.; Meier, R. J. Vibrat. Spectrosc. 2005, 39, 15. (22) Sander, W.; Gantenberg, M. Spectrochim. Acta A 2005, 62, 902. (23) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb, M. A.; Cheeseman, J. R. ; Montgomery, J. A., Jr.; Vreven, T.; Kudin, K. N.; Burant, J. C.; Millam, J. M.; Iyengar, S. S.; Tomasi, J.; Barone, V.; Mennucci, B.; Cossi, M.; Scalmani, G.; Rega, N.; Petersson, G. A.; Nakatsuji, H.; Hada, M.; Ehara, M.; Toyota, K.; Fukuda, R.; Hasegawa, J.; Ishida, M.; Nakajima, T.; Honda, Y. ; Kitao, O.; Nakai, H.; Klene, M.; Li, X.; Knox, J. E.; Hratchian, H. P.; Cross, J. B.; Bakken, V.; Adamo, C.; Jaramillo, J.; Gomperts, R.; Stratmann, R. E.; Yazyev, O.; Austin, A. J.; Cammi, R.; Pomelli, C.; Ochterski, J. W.; Ayala, P. Y.; Morokuma, K.; Voth, G. A.; Salvador, P.; Dannenberg, J. J.; Zakrzewski, V. G.; Dapprich, S.; Daniels, A. D.; Strain, M. C.; Farkas, O.; Malick, D. K.; Rabuck, A. D.; Raghavachari, K.; Foresman, J. B.; Ortiz, J. V.; Cui, Q.; Baboul, A. G.; Clifford, S.; Cioslowski, J.; Stefanov, B. B.; Liu, G.; Liashenko, A.; Piskorz, P.; Komaromi, I.; Martin, R. L.; Fox, D. J.; Keith, T.; Al-Laham, M. A.; Peng, C. Y.; Nanayakkara, A.; Challacombe, M.; Gill, P. M. W.; Johnson, B.; Chen, W.; Wong, M. W.; Gonzalez, C.; Pople, J. A. Gaussian 03, revision C.02; Gaussian, Inc.: Wallingford, CT, 2004. (24) Becke, A. D. J. Chem. Phys. 1993, 98, 5648. (25) Lee, C.; Yang, W.; Parr, R. G. Phys. ReV. B 1988, 37, 785. (26) Shimanouchi, T. Computer Programs of Normal Coordinate Treatment of Polyatomic Molecules; The University of Tokyo, 1968. (27) Yoshida, H.; Tasumi, M. J. Chem. Phys. 1988, 89, 2803. (28) Derissen, J. L. J. Mol. Struct. 1971, 7, 67. (29) Berney, C. V.; Redington, R. L.; Lin, K. C. J. Chem. Phys. 1970, 53, 1713. (30) Haurie, M.; Novak, A. J. Chim. Phys. 1967, 64, 679. (31) Dioumaev, A. K.; Braiman, M. S. J. Am. Chem. Soc. 1995, 117, 10572. (32) Ge´nin, F.; Quile`s, F.; Burneau, A. Phys. Chem. Chem. Phys. 2001, 3, 932. (33) Tanaka, N.; Kitano, H.; Ise, N. J. Phys. Chem. 1990, 94, 6290. (34) Maes, G.; Zeegers-Huyskens, Th J. Mol. Struct. 1983, 100, 305. (35) Handbook of Chemistry and Physics, 83rd ed.;Lide, D. R., Ed.;CRC Press: Boca Raton, FL, 2002; pp6-153.. (36) Stowell, M. H. B.; McPhillips, T. M.; Rees, D. C.; Soltis, S. M.; Abresch, E.; Feher, G. Science 1997, 276, 812.

JP801151K