Coupling of Complex Aromatic Ring Vibrations to Solvent through

Mar 11, 2008 - Nathaniel V. Nucci,*, J. Nathan Scott, andJane M. Vanderkooi. Department of Biochemistry and Biophysics, University of Pennsylvania, 42...
0 downloads 0 Views 892KB Size
4022

J. Phys. Chem. B 2008, 112, 4022-4035

Coupling of Complex Aromatic Ring Vibrations to Solvent through Hydrogen Bonds: Effect of Varied On-Ring and Off-Ring Hydrogen-Bonding Substitutions Nathaniel V. Nucci,* J. Nathan Scott, and Jane M. Vanderkooi Department of Biochemistry and Biophysics, UniVersity of PennsylVania, 422 Curie BouleVard, Philadelphia, PennsylVania 19104 ReceiVed: July 25, 2007; In Final Form: December 18, 2007

In this study, we examine the coupling of a complex ring vibration to solvent through hydrogen-bonding interactions. We compare phenylalanine, tyrosine, L-dopa, dopamine, norepinephrine, epinephrine, and hydroxylDL-dopa, a group of physiologically important small molecules that vary by single differences in H-bonding substitution. By examination of the temperature dependence of infrared absorptions of these molecules, we show that complex, many-atom vibrations can be coupled to solvent through hydrogen bonds and that the extent of that coupling is dependent on the degree of both on- and off-ring H-bonding substitution. The coupling is seen as a temperature-dependent frequency shift in infrared spectra, but the determination of the physical origin of that shift is based on additional data from temperature-dependent optical experiments and ab initio calculations. The optical experiments show that these small molecules are most sensitive to their immediate H-bonding environment rather than to bulk solvent properties. Ab initio calculations demonstrate H-bond-mediated vibrational coupling for the system of interest and also show that the overall small molecule solvent dependence is determined by a complex interplay of specific interactions and bulk solvation characteristics. Our findings indicate that a full understanding of biomolecule vibrational properties must include consideration of explicit hydrogen-bonding interactions with the surrounding microenvironment.

Introduction There are millions of biologically active small organic molecules, both natural and synthetic. Unraveling the intricacies of how these small molecules are recognized, bound to specific receptors, and lead to biological activity is one of the central questions in current physical biochemistry research. Understanding the complex interplay between small molecules and the solvating environment is an important piece of this puzzle.1-5 Coupling of vibrational modes through H bonds has been a topic of great interest over the last several years. Most examinations of vibrational coupling through H bonds have focused on simple X-H stretching modes of atoms directly engaged in the H bond. Time-resolved spectroscopic methods have been used to directly measure resonance through H bonds for simple vibrational modes of carboxylic acid dimers and H-bonded dipeptides.6-11 Linear IR spectroscopy, with and without temperature excursion, has also been used to characterize H-bond-mediated vibrational coupling of simple stretching and bending modes in carbohydrates12,13 and in amino acid crystals.14 Extensive spectral analysis of H-bonding effects on molecules in vapor clusters has been performed, but these studies, too, have focused primarily on simple stretching modes.15-20 These cluster studies have clearly shown vibrational coupling effects of H bonding on the O-H stretch and the C-O stretch of hydroxylated aromatic molecules, but they have not evaluated effects on complex ring modes of these molecules. Computational analysis of more complex, cooperative vibrational modes of peptides has been performed by Dannenberg and co-workers,21-25 but experimental evidence for coupling of complex modes through H-bonding interactions has not been * To whom correspondence should be addressed. E-mail: nvnucci@ mail.med.upenn.edu.

shown to date. In this study we focus on the complex vibrational modes of aromatic ring-based small molecules and their coupling to solvent through H-bonding interactions. This study focused on a group of molecules structurally related to a molecule used in previous examples of vibrational temperature independence, tyrosine.26 The full group of molecules used is given in Figure 1. This family of molecules is based on varied substitutions about a benzene scaffold; each varies from another by a single functional group substitution, and all substitutions alter the degree and/or location of Hbonding interactions with solvent. This group of molecules also permits comparison of both on-ring and off-ring H-bonding substitutions, and all molecules are of physiological importance. Tyrosine and phenylalanine are incorporated into proteins and are vital for protein structure and function. Dopamine, norepinephrine, and epinephrine are important neurotransmitters that regulate much of the brain’s control of endocrine functions.27 The minor structural differences between these molecules produce surprisingly large differences in physiological activity. For example, differences in binding affinity between dopamine and norepinephrine can be up to 2 orders of magnitude or more.27,28 The addition or loss of a single hydrogen-bonding interaction between protein and ligand does not account for this enormous difference in binding, so this group of molecules is an intriguing model for understanding more subtle physical properties that may have importance for biological function of small molecules. Temperature-excursion infrared spectroscopy (TEIR) is a widely used technique for examining the behavior of H-bonded systems.26,29,30 Traditionally, vibrational modes that involve the motion of atoms not engaged in H bonding with solvent have been thought to be temperature independent over large temperature ranges. The model for H-bond-induced temperature

10.1021/jp0758770 CCC: $40.75 © 2008 American Chemical Society Published on Web 03/11/2008

Solvent Dependence of Aromatic Ring Vibrations

Figure 1. Schematic representation of the molecules examined in this study. The numbered ring and table show the molecules used in this study as varied substitutions about a benzene scaffold.

dependence is simple; the presence of an H-bond to an X-H chemical group causes a strain on the X-H bond, lengthening it and shifting the X-H stretching mode to lower frequency. The magnitude of the shift is dependent on the strength of the H-bonding interaction.30 This simple relationship has been used extensively to investigate protein structure as well as the properties of H-bonded polymers, including many plastics,26,31-33 and as addressed above, such vibrational coupling through hydrogen bonds has recently been directly demonstrated and measured.6-11 The central questions we address in this study are: do complex modes show similar coupling to solvent as seen for simple X-H stretching modes? If so, what is the relative extent of coupling, and what are the physical origins of that coupling? To answer this question, we examined modes 19a/b over a wide temperature range for all of the molecules shown in Figure 1. This vibrational mode is readily resolvable from other solute modes, as well as from solvent modes. Modes 19a/b are principally composed of motion of atoms not directly engaged in H-bonding with solvent, so they are a convenient probe of how a complex vibration can be affected by solvent H-bonding of peripheral chemical groups. Because of the selection rules for the aromatic ring vibrations, mode 19a is the resolvable mode for phenylalanine, tyrosine, and hydroxy-DL-dopa (OH-DL-dopa), while the observable absorption for L-dopa, dopamine, norepinephrine, and epinephrine is from mode 19b. Modes 19a and 19b are a degenerate pair and have been assigned according to the system laid out by Wilson34 and applied to a wide variety of benzene derivatives by Varsanyi.35 The atomic motions of mode 19a and 19b for benzene are shown in Figure 6 using vector diagrams; the measured vibrations of the solute molecules in this study are analogous to the benzene motions shown. To gain insight on the physical reasons for the results of the TEIR experiments, the temperature dependence of the fluorescence and low-temperature phosphorescence of these molecules was also examined as a function of temperature. The temperature dependence of tyrosine and phenylalanine fluorescence properties have been used before to investigate protein structure and hydrogen-bonding interactions,36-41 and the fluorescence properties of the catecholamine neurohormones dopamine, norepinephrine, and epinephrine have been utilized to design assays for their quantitation in biological samples.42-44 The phosphorescence lifetime was also measured for all molecules. The fluorescence and phosphorescence results are used to develop an understanding of the solvent dependence of the small molecule electronic properties as a function of H-bonding chemical substitution. Interpretation of the infrared and fluorescence results require consideration of the temperature-dependent changes in water/ glycerol solvation properties. The temperature and concentration dependent behavior of water/glycerol mixtures has been studied extensively. We have used temperature-dependent infrared spectroscopy and simulations to examine the behavior of the H-bond network of water/glycerol.45,46 These studies have shown

J. Phys. Chem. B, Vol. 112, No. 13, 2008 4023 that the average hydrogen-bond strength increases greatly over the liquid temperature range between room temperature and the glass transition temperature. Several groups have examined the temperature dependent behavior of water/glycerol with respect to its dielectric properties.47-51 It has been known for many years that the relative static permittivity (dielectric constant) of glycerol/water solutions drops dramatically with temperature, going from ∼40 at room temperature to ∼90 at 200 K for the mixture used in the present study.48 Most recently, Feldman and co-workers have produced a new model that offers a unified explanation for the complex behavior of water/glycerol mixtures as a function of temperature and concentration.47,50,51 In short, this model describes the dielectric relaxation properties as a function of local H-bond heterogeneity and demonstrates that for glycerol-rich mixtures, such as the one used in the present study, the dynamic relaxation slows dramatically with reduced temperature. In light of these previous findings, we can interpret temperature-dependent infrared and optical spectra in terms of changing solvation environment with respect to increased H-bonding strength, increased polarity, and decreased dielectric relaxation of the solvent as temperature is reduced. By combination of temperature-dependent infrared and fluorescence measurements with ab initio calculations, we demonstrate that increased H-bonding interactions with solvent lead to more extensive coupling of complex vibrational modes to solvent motions and that this coupling is dependent on both on-ring and off-ring substitutions. This is the first time, to our knowledge, that vibrational coupling to complex modes through H-bond interactions has been shown. Our findings indicate that a full understanding of biomolecule vibrational properties must include consideration of explicit hydrogen-bonding interactions with the surrounding microenvironment. Materials and Methods Materials. D2O and D8-glycerol were purchased from Cambridge Isotope Laboratories, Inc. (Andover, MA). Solutes and glycerol were purchased from Sigma Aldrich (St. Louis, MO). Water was deionized and distilled. Sample Preparation. Samples were prepared under argon (Airgas, Inc.; Radnor, PA), and all solvents were argon-purged prior to sample preparation for infrared experiments or vacuum degassed for fluorescence experiments in order to prevent oxidation of the solute molecules. Samples were also protected from light to prevent photo-oxidation. All TEIR samples were prepared in 40/60 (v/v) D2O/D8glycerol. All fluorescence and phosphorescence lifetime samples were prepared in 40/60 (v/v) H2O/glycerol. Sample concentrations for TEIR were as follows: phenylalanine (Phe), saturated (∼ 80 mM); tyrosine (Tyr), saturated (∼10 mM); hydroxy-DLdopa (OH-DL-dopa), saturated (∼10 mM); L-dopa, saturated (∼10 mM); dopamine, 65 mM; norepinephrine, 200 mM; epinephrine, 200 mM. Concentrations for fluorescence spectra were: phenylalanine, saturated (∼ 80 mM); tyrosine, saturated (∼10 mM); hydroxy-DL-dopa, saturated (∼10 mM); L-dopa, saturated (∼10 mM); dopamine, 25 mM; norepinephrine, 25 mM; epinephrine, 25 mM. Concentrations for phosphorescence lifetime measurements were: phenylalanine, 5 mM; tyrosine, saturated (∼10 mM); hydroxy-DL-dopa, saturated (∼10 mM); L-dopa, saturated (∼10 mM); dopamine, 8 mM; norepinephrine, 8 mM; epinephrine, 10 mM. For dopamine, norepinephrine, and epinephrine, the hydrochloride salts were used without buffer or pH adjustment. As a result, these three solutes were positively charged and included dissolved chloride as the counterion. All other solutes were zwitterionic.

4024 J. Phys. Chem. B, Vol. 112, No. 13, 2008

Nucci et al.

Figure 2. TEIR spectra. Infrared absorptions of mode 19a or 19b of each molecule examined are shown at 20 K intervals from 290 to 30 K. Extreme temperature curves are indicated. Note the lack of frequency shift for phenylalanine, the slight shift for tyrosine, and the larger shifts for all other molecules.

Spectroscopy. Infrared spectra of each sample were recorded in a transmission cell between CaF2 windows with a 50- or 100µm Teflon spacer. Spectra were recorded using a Bruker IFS

66 FTIR spectrometer (Bruker Optics; Billerica, MA) with 2-cm-1 resolution. For TEIR experiments, temperature was controlled with an Omniplex closed-cycle Helitran cryostat from

Solvent Dependence of Aromatic Ring Vibrations

Figure 3. Example fits to infrared spectra. Examples of the spectral fits for tyrosine, OH-DL-dopa, and norepinephrine. Spectra were fit to a complex combination of Voigt functions to represent the baseline with one or two Voigt functions representing the absorption of mode 19a or 19b. Spectral raw data is shown as data points; components and sum of components are shown as solid lines. Baseline contributions were removed for clarity.

Advanced Research Systems, Inc. (Macungie, PA). TEIR spectra of each sample were taken at 20 K intervals from 290 to 30 K. Fluorescence emission spectra were measured with a Fluorolog-3-21 Jobin-Yvon Spex Instrument SA (Edison, NJ) equipped with a 450-W xenon lamp for excitation and a cooled R2658P Hamamatsu photomultiplier tube for detection. Slit width was set to provide a band pass of 1, 2, or 3 nm for excitation and 3 nm for emission. Temperature was varied using an Omniplex cryostat with the sample held between quartz windows separated by a 200-µm spacer for Tyr and OH-DLdopa, 100-µm spacer for Phe, and a 25-µm spacer for L-dopa, dopamine, norepinephrine, and epinephrine. Emission spectra were recorded at 20 K intervals from 290 to 30 K. Excitation wavelengths were: phenylalanine, 266 nm; tyrosine, 275 nm; hydroxy-DL-dopa, 290 nm; L-dopa, dopamine, norepinephrine, and epinephrine, 280 nm. Phosphorescence lifetime measurements were recorded with a Fluorolog-3-21 Jobin-Yvon Spex Instrument SA equipped with a 450-W xenon lamp for excitation and a cooled R2658P Hamamatsu photomultiplier tube for detection. Slit width was set to provide a band pass of 1 nm for excitation and 10 nm for

J. Phys. Chem. B, Vol. 112, No. 13, 2008 4025 emission. For tyrosine and phenylalanine, signal was integrated at 0.25 s intervals. For all other samples, signal integration time was 0.01 s. Samples were held in quartz tubes and equilibrated at 77 K in liquid nitrogen prior to measurement. Samples were excited for 50 s prior to gating the excitation beam and recording phosphorescence emission decay over time. Excitation wavelengths were the same as for temperature-dependent fluorescence emission spectra. Recorded phosphorescence emission decay wavelengths were: phenylalanine, 380 nm; tyrosine, 387 nm; hydroxy-DL-dopa, L-dopa, dopamine, norepinephrine, and epinephrine, 340 nm. Spectral Analysis. All infrared spectra were manipulated for atmospheric and baseline correction using Opus 5.0. (Bruker Optics). All spectra were analyzed and reported without smoothing. For TEIR spectra, the peak at ∼1500 cm-1 was evaluated for each solute. This peak corresponds to mode 19a for phenylalanine, tyrosine, and OH-DL-dopa. For L-dopa, dopamine, norepinephrine, and epinephrine, it corresponds to mode 19b. Because several of the molecules had low solubility, solvent modes sometimes interfered slightly with the 19a/b modes of these solutes. This necessitated the use of spectral fitting to evaluate temperature-dependent frequency shifts. To evaluate these shifts, the region of the infrared spectrum corresponding to modes 19a/b was fit using PeakFit 4.11 (Systat Software, Inc.; San Jose, CA). Each spectrum was fit to a combination of Voigt functions such that the baseline was fit to many functions and the 19a/b ring mode was fit to a single Voigt in all cases, except norepinephrine and epinephrine, in which case, mode 19a/b was fit to a combination of two Voigt functions. Justification for treatment of norepinephrine and epinephrine spectra is addressed in Results. Fluorescence emission spectra and phosphorescence decay data were used without adjustment. Phosphorescence decays were fit to single-exponential functions for phenylalanine and tyrosine and to double exponential functions for OH-DL-dopa, L-dopa, dopamine, norepinephrine, and epinephrine. All fits achieved R2 g 0.999. Quantum Chemical Calculations. Ab initio calculations were performed using Gaussian 03.52 Optimized geometries and vibrational frequencies were calculated using B3LYP model chemistry with the 6-311G(2d,p) basis set.53-65 This model chemistry has been shown to be appropriate for calculating the typically problematic ring vibrational frequencies for small aromatic molecules.66,67 Phenylalanine, tyrosine, and OH-DLdopa geometries were first individually optimized in vacuum, and vibrational frequencies of the vacuum-optimized isolated geometry were evaluated. These geometries were then used without optimization in frequency calculations with dielectric representation in order to evaluate dielectric effects on vibrational frequency. The SCIPCM model68 was used with all parameters set to their default values for water solvation except the dielectric, which was set to 4 or 78.39. A single water molecule was introduced near the C4 ring hydroxyl substitution for both tyrosine and OH-DL-dopa, and the geometries of these two pairs of molecules were optimized in vacuum. These optimized geometries were then used for frequency calculations with the SCIPCM representation of dielectric at 4 and at 78.39, as performed above for the isolated optimum geometries. Finally, the water molecule was removed, leaving the aromatic molecule in the optimized geometry calculated with the water molecule, and vibrational frequencies were again calculated in vacuum and in a dielectric of 4 and 78.39. By comparison of the results of these frequency calculations, we sought to isolate,

4026 J. Phys. Chem. B, Vol. 112, No. 13, 2008

Nucci et al.

Figure 4. Temperature-excursion fluorescence spectra. Fluorescence emission spectra are shown for each molecule tested at 20 K intervals from 290 to 30 K. Extreme temperature spectra are indicated. Emission of phenylalanine from top left is shown zoomed in at top right to make the fluorescence emission and vibrational resolution of fluorescence visible. Each full emission spectrum also shows low-temperature phosphorescence emission at higher wavelength. Note the good vibronic resolution of phenylalanine, the weaker vibronic resolution of tyrosine, and the lack of vibronic resolution for all other molecules.

to some extent, the effects on vibrational frequencies of varied dielectric vs changing strength of hydrogen bonds in our

experiments. This approach and its rationale are discussed further in Results.

Solvent Dependence of Aromatic Ring Vibrations

J. Phys. Chem. B, Vol. 112, No. 13, 2008 4027 TABLE 1: Infrared Absorption Maxima of the Spectral Fits of the Aromatic Ring C-C Stretching Modes at High and Low Temperatures phenylalanine tyrosine hydroxy-DL-dopa L-dopa dopamine norepinephrine 1 norepinephrine 2 epinephrine 1 epinephrine 2

290 K

30 K

shift

1497.5 1515.6 1519.6 1520.3 1521.2 1515.5 1522.7 1514.4 1522.5

1497.5 1516.4 1522.6 1522.3 1522.5 1517.4 1524.6 1516.3 1524.5

0 0.8 3 2 1.3 1.9 1.9 1.9 2

TABLE 2: Fluorescence Emission Peak Wavelengths for Each Molecule at High and Low Temperature Figure 5. Phosphorescence decay of L-dopa. Phosphorescence emission of L-dopa at 340 nm is shown as a function of time after excitation beam gating. This decay was fit to a double exponential decay. Raw data is shown as points with the fit function shown as a line. Fit values are given in Table 3.

temperature (K)

290

30

shift

phenylalanine tyrosine OH-DL-Dopa L-Dopa dopamine norepinephrine epinephrine

280, 288 303 343.0 314 315 314 313

280, 288 296 325 305 305 304 304

0, 0 -7 -18 -9 -10 -10 -9

TABLE 3: Phosphorescence Lifetimes of Small Molecules Based on Single or Double Exponential Fits of the Decay Curves (R2 ) 0.999)

phenylalanine tyrosine L-dopa hydroxy-DL-dopa dopamine norepinephrine epinephrine

T1 (s)

(

8.503 3.094 0.219 0.388 0.186 0.212 0.210

0.004 0.014 0.002 0.008 0.004 0.005 0.006

% contribution T2 (s) 100.0 100.0 24.1 29.9 26.0 20.9 19.2

0.907 1.716 0.841 0.909 0.921

(

% contribution

0.005 0.033 0.014 0.013 0.013

75.9 70.1 74.0 79.1 80.8

conditions (phenylalanine, tyrosine, and OH-DL-dopa isolated in vacuum and tyrosine and OH-DL-dopa with a water molecule in vacuum) are provided in Supporting Information in z-matrix format. Results

Figure 6. Vibrational modes evaluated from ab initio calculations. The vibrations of benzene, to which the evaluated vibrational modes are analogous, are shown with vector diagrams. Plus and minus signs indicate deflections out of the plane of the figure. The calculated vibrations of the aromatic molecules tested in this study were identified by their similarity to the modes shown.

Facio69 was used to construct initial geometries and to visualize calculated vibrational modes. Molekel 5.2.0 was used to evaluate the electrostatic potential at the solvent accessible surface for each calculation. Solvent probe size was 1.4 Å, and the standard MEP calculation included in Molekel was used with a cube file of the potential generated from Gaussian 03. Facio was used to create the vibrational vector diagrams in Figure 9. Electron density distributions were visualized using gOpenMol from cube files of the density generated from Gaussian 03. Vibrational assignments were performed by visual inspection and comparison to the analogous benzene ring vibrational modes as described by Varsanyi,35 based on the work of Wilson.34 Frequencies are reported without scaling because relative frequencies, rather than absolute frequencies, are of relevance for this study. Optimized geometries for all five

TEIR Spectroscopy. The temperature dependence of the 19a or 19b ring modes of phenylalanine, tyrosine, hydroxy-DL-dopa, L-dopa, dopamine, norepinephrine, and epinephrine are shown in Figure 2 over a temperature range of 260 K. From inspection of the spectra, it is clear that ring modes 19a/b, a set of coupled complex vibrations principally of the aromatic ring, exhibit temperature-dependent frequency shifts for molecules that have peripheral H bonding to solvent. To quantitatively compare the temperature-dependent shifts, spectra were fit such that solvent contributions were represented by a complex combination of many Voigt functions, while the ring mode was represented by a single Voigt peak for all solutes except norepinephrine and epinephrine, in which case the ring mode was represented by a combination of two Voigt functions. Examples of the spectral fits at high and low temperature are shown in Figure 3 with baseline contributions removed for clarity, and peak frequencies from the fits are shown in Table 1. The 19b vibrations of norepinephrine and epinephrine could not reasonably be approximated as a single component, so each was fit to a combination of two components. Dopamine, norepinephrine, and epinephrine have been shown to have multiple conformations in solution that are separated by distinct energy barriers to rotation about the CR-Cβ bond, resulting in a mixture of trans

4028 J. Phys. Chem. B, Vol. 112, No. 13, 2008

Nucci et al.

Figure 7. Electron distributions from ab initio calculations. The electron distribution contours are shown for all calculations from the top (7a) and bottom (7b) of the ring. Top and bottom were arbitrarily assigned. Contours represent electron densities at 0.1 (blue), 0.5 (green), 0.25 (yellow), and 0.01 (white). The only appreciable changes in electron distribution are in the presence of specific H bonding to water; dielectric change has no effect on electron distribution.

and gauche conformers.70-76 We attribute the two-component nature of the 19b mode for norepinephrine and epinephrine to the mixture of conformations present in solution for these molecules. A similar character should be seen for the 19b mode of dopamine, but dopamine’s poor solubility results in the 19b absorption being too weak to resolve multiple components (not shown). Comparison of the temperature-dependent frequency shifts for phenylalanine, tyrosine, L-dopa, and OH-DL-dopa showed that temperature dependence of the observed ring mode increases with increasing substitution of H-bonding groups on the aromatic ring. Phenylalanine exhibited no solvent dependence of ring mode 19a. With increasing addition of hydroxyl groups to the ring, solvent dependence increased from 0.8 cm-1 for tyrosine to 2 cm-1 for L-dopa to 3 cm-1 for OH-DL-dopa, showing an approximately additive effect for each additional hydroxyl group on the aromatic ring. By comparison of the ring-mode frequency shifts for L-dopa, dopamine, norepinephrine, and epinephrine, the effects of off-ring substitutions were observed. Mode 19b of L-dopa shifted more than that of dopamine, demonstrating that the presence or absence of the carboxylic acid on the R-carbon affects the degree of solvent dependence of the ring mode. Mode 19b of norepinephrine shifted more than that of dopamine showing that presence or absence of a β-carbon H-bonding moiety affects ring-mode solvent dependence. The shifts of norepinephrine and epinephrine are identical, showing that addition of a methyl group to the terminal amine has no measurable effect on the temperature dependence of the observed ring mode. The results of this set of experiments demonstrate that complex vibrational modes can be affected by H bonding of nearby chemical groups to solvent, even if these modes involve

only very slight motion of H-bonded atoms. This is the first time, to our knowledge, that H-bonding substitution has been shown to impart solvent dependence on complex, many-atom vibrational modes. The solvent dependence of the observed complex mode is dependent on the degree of H-bonding substitution both on- and off-ring. On-ring H-bonding substitutions have a stronger effect than off-ring substitutions, and the effects of off-ring substitutions appear to have some dependence on distance from the ring. Further investigation of other small aromatic molecules will be required to determine the extent of this distance dependence as well as how different H-bonding substitutions may alter this effect. The present findings clearly show, however, that complex modes can be coupled to solvent and appear to be coupled to the solvent as a function of H-bonding capacity. The question remains, however, whether this effect is the result of vibrational coupling between the aromatic small molecule and water via the H bond or the result of less specific bulk solvation effects such as the changing dielectric constant. In our experiments, the temperature variance from 290 to 30 K induces three primary known changes in the solvation environment, all of which occur mainly between 290 K and the glass transition temperature. The glass transition is estimated to be near 150 K for the water/glycerol mixture used here. As temperature drops in the liquid phase, there is an increase in the linearity (and therefore the strength) of the hydrogen-bonding environment.46 This change in the degree of hydrogen bonding greatly increases the relative static permittivity.48 At the same time the dielectric constant is increasing, the dielectric relaxation is slowing dramatically as the solution cools, effectively going to zero at the glass transition nears 150 K.47,50,51 Infrared measurements alone cannot resolve these different effects.

Solvent Dependence of Aromatic Ring Vibrations

J. Phys. Chem. B, Vol. 112, No. 13, 2008 4029

Figure 8. Electrostatic potential mapped to the solvent accessible surface from ab initio calculations. Electrostatic potential was mapped to the solvent accessible surface from each calculation using the MEP function in Molekel 5.2.0 with a probe radius of 1.4 Å. Surfaces are shown from the top (a) and bottom (b) of the ring. Molecular orientations are the same as for parts a and b of Figure 7, respectively. Dielectric variation alters the magnitude of the electrostatic potential distribution, but not the relative distribution, itself. Presence of specific H bonding does alter the electrostatic potential distribution.

Figure 9. Vector diagrams of vibrational coupling for two aromatic ring modes from ab initio calculations. Modes 3 and 19a of tyrosine with one water molecule are shown from the results of the ab initio calculations at each dielectric condition tested. Coupled motion of the water molecule is dependent on the dielectric constant for both vibrations, though this dependence is reversed between the two modes. Motion of the aromatic molecule is not dependent on the dielectric, and motion of the water molecule is present at all dielectric constants for both modes.

Indeed, these properties of the solvation environment are intimately linked, so separating their respective influences on molecular vibrations is difficult. We attempted to do so,

however, by examining the temperature dependence of fluorescence and phosphorescence emission properties of these molecules and through quantum calculations.

4030 J. Phys. Chem. B, Vol. 112, No. 13, 2008 Temperature Dependence of Small Molecule Fluorescence. Temperature-dependent spectra of the fluorescence and phosphorescence emission for all molecules tested in this study are shown in Figure 4. Peak emission wavelength shifts over the temperature range tested are summarized in Table 2. As was seen for the infrared ring mode, phenylalanine fluorescence displayed no dependence on temperature; the spectral resolution improved as temperature dropped, but the peak emission wavelength did not change. Similar to the infrared behavior, solvent dependence of the maximum emission wavelength increased with increasing H-bonding substitution on the ring, as evidenced by the peak emission wavelength shifts of tyrosine, L-dopa, and OH-DL-dopa. Unlike the infrared behavior, however, the offring substitutions did not show any measurable effect on fluorescence emission; temperature dependence of L-dopa, dopamine, norepinephrine, and epinephrine fluorescence were the same. In addition to the differences in emission wavelength temperature dependence, the degree of vibronic resolution also changed with hydrogen-bonding substitution (Figure 4). Phenylalanine showed considerable resolution of vibrational states. Addition of the single hydroxyl group in tyrosine nearly abolished this vibronic structure in the emission spectrum. All molecules with two or more hydroxyl substitutions on the ring showed no appreciable vibronic resolution. As was seen with the fluorescence emission wavelength temperature dependence, no effect was seen for off-ring hydrogen-bonding substitution changes. Phosphorescence Changes as a Function of H-Bonding Substitution. All of the molecules tested exhibited lowtemperature phosphorescence that also lent useful insight into how H-bonding substitution alters solvent dependence of the aromatic ring. The spectra in Figure 4 show evolution of a broad emission at higher wavelengths corresponding to this lowtemperature phosphorescence. In the case of phenylalanine, strong vibronic resolution of both the fluorescence and the phosphorescence was seen. The vibronic resolution of phosphorescence was drastically reduced for tyrosine and was completely lost for all other solutes. This loss of vibronic resolution is indicative of faster relaxation mechanisms from the triplet state. This trend correlates with the degree of onring H-bonding substitution in this group of molecules. As with the temperature dependence of the fluorescence emission wavelength, no differences were seen in the degree of vibronic resolution for off-ring substitutions. This correlation between loss of vibronic resolution and degree of on-ring H-bonding substitution predicted that the phosphorescence lifetime is shorter. To test this prediction, we measured the phosphorescence decay for each molecule. The phosphorescence decay curve for L-dopa at 340 nm is shown in Figure 5 as an example. The calculated phosphorescence lifetimes are given in Table 3. Phenylalanine and tyrosine fit well (R2 g 0.999) to single-exponential decays, but all other molecules did not (R2 < 0.985). As a result, they were fit to double-exponential decays, which achieved the desired goodness of fit (R2 g 0.999). As with some of the infrared results, we expect that the multiple relaxation times for these molecules are likely due to multiple conformational states that are interconverting on very long time scales as has been previously described for dopamine, norepinephrine, and epinephrine.70-76 Phenylalanine had the longest lifetime, 8.5 s, while the lifetime of tyrosine was considerably slower, 3.1 s. Both of these measurements agree well with previous measurements of phenylalanine and tyrosine phosphorescence decay.77 With increasing H-bonding substitution on the ring, phosphorescence

Nucci et al. lifetime decreased considerably going from phenylalanine to tyrosine and from tyrosine to L-dopa. The lifetimes for dopamine, norepinephrine, and epinephrine were essentially the same as those for L-dopa, showing that off-ring substitutions had no measurable effect on phosphorescence decay within the resolution of this measurement. Phosphorescence lifetimes for OHDL-dopa are slightly longer than those of L-dopa. This slight increase may be due to the ability of OH-DL-dopa to easily form intramolecular H bonds. These data are again indicative of considerable effects of peripheral H bonding to solvent on the aromatic ring’s degree of solvent dependence. Additional insight on the effects of the temperature-dependent changes in solvation on these molecules was gained from ab initio calculations. Quantum Chemical Calculations. To examine the effects of changing dielectric vs the effect of specific H-bonding interaction, the vibrational frequencies, electron distributions, and electrostatic surfaces were compared for phenylalanine, tyrosine, and OH-DL-dopa under various conditions. Each of these three molecules was examined individually under three different dielectric conditions to observe the changes in electron distribution and vibrational frequency as a function of the dielectric of the molecular environment. Tyrosine and OH-DLdopa were also examined under H-bonding conditions achieved by inclusion of a single water molecule in the calculation. The optimized geometry with a water molecule was then re-evaluated without the water molecule to examine the frequency and electron distribution changes that were the result of the change in the aromatic molecular geometry compared to the isolated calculation, rather than the result of H bonding to the water molecule. The primary goal of these calculations was to distinguish between bulk dielectric effects on vibrational frequency and specific H-bonding effects; thus the main results of interest are the vibrational frequency calculations. For examination of vibrational effects, we selectively analyzed those modes that were readily identifiable as analogous to the classical modes of benzene, as described by Wilson and as examined in a wide variety of benzene derivatives by Varsanyi.34,35 The calculated frequencies of these modes are listed in Table 4, and vector diagrams of the benzene modes to which they are analogous are shown in Figure 6. The electron distributions for each calculation are shown using contour representation in Figure 7. The electrostatic potentials mapped to the solvent-accessible surfaces from each calculation are shown in Figure 8. It is clear from the frequency calculations that the effects of changing dielectric and specific H-bonding interaction are quite variable and apparently are dependent on the vibration in question. In all cases, the effect of the changing dielectric was small. In many cases, the presence of a specific H-bonding interaction was also small, but in some cases this effect was quite large. By examination of the frequency calculations in the presence of the water molecule vs the calculated frequencies with the water molecule removed, the effect of the changed geometry of the aromatic molecule can be isolated from the effect of the H bond to water. The group of frequencies examined includes representatives of several of the classes of aromatic ring vibrations. The results of the frequency calculations are examined with respect to these vibrational classes. C-X In-Plane Bending Vibrations. Mode 3 is one of the C-X in-plane bending vibrations. According to the results of our calculations, this mode was most affected by the presence of H bonding to a water molecule for both tyrosine and for OH-DLdopa. The frequency of this mode was also affected by the changing dielectric but to a much lesser extent. The frequency

Solvent Dependence of Aromatic Ring Vibrations

J. Phys. Chem. B, Vol. 112, No. 13, 2008 4031

TABLE 4: Calculated Frequencies for Phenylalanine, Tyrosine, and OH-DL-Dopa Ring Vibrations with and without a Water Molecule Hydrogen Bonded to the Terminal Hydroxyl Group(s) (Except Phenylalanine) in Vacuum and with SCIPCM Dielectric Modeling at E ) 4 and E ) 78

vacuum E)4 E ) 78 vacuum E)4 E ) 78 vacuum E)4 E ) 78 vacuum E)4 E ) 78 vacuum E)4 E ) 78 vacuum E)4 E ) 78 vacuum E)4 E ) 78 vacuum E)4 E ) 78

mode

phenylalanine (cm-1)

tyrosine (cm-1)

tyrosine + H2O (cm-1)

tyrosine + H2O (cm-1) H2O removed

OH-DL-dopa (cm-1)

OH-DL-dopa + H2O (cm-1)

OH-DL-dopa + H2O (cm-1) H2O remove d

3 3 3 4 4 4 5 5 5 10a 10a 10a 16a 16a 16a 16b 16b 16b 19a 19a 19a 19b 19b 19b

356.1 354.3 350.7 755.9 752.4 751.0 964.8 965.9 966.4 852.4 847.5 845.8 416.7 413.0 411.4 535.9 534.9 533.6 1534.6 1528.7 1526.7 1493.0 1486.9 1484.7

430.3 421.5 418.4 730.8 732.1 732.4 930.2 934.2 934.4 812.4 811.4 810.6 424.6 428.1 432.8 491.7 491.6 491.1 1551.1 1543.9 1541.0 1475.8 1468.3 1465.6

456.4 451.8 450.6 734.4 734.4 731.3 941.7 938.9 936.6 836.0 828.4 825.7 427.6 425.2 424.6 493.8 490.8 489.0 1554.3 1547.7 1543.6 1490.4 1484.4 1482.0

435.0 428.6 425.9 730.7 727.4 725.8 931.2 932.8 931.2 843.4 836.9 834.5 424.9 424.4 409.2 491.1 490.3 489.5 1551.9 1544.4 1541.3 1479.2 1470.1 1466.8

619.4 607.5 604.5 689.7 690.2 690.3 841.4 837.1 835.7 825.3 818.3 815.3 458.9 457.1 453.5 522.8 518.5 514.5 1549.6 1540.6 1536.8 1512.6 1504.1 1501.3

645.9 643.8 641.4 684.3 679.4 673.1 840.7 842.2 842.4 825.9 820.4 817.8 464.6 465.3 463.2 521.5 521.7 518.3 1548.3 1540.5 1537.3 1507.2 1500.2 1497.8

619.4 608.3 606.1 690.0 690.3 690.0 844.1 843.3 842.9 825.5 816.8 812.7 465.4 454.1 465.0 523.7 520.2 516.9 1547.9 1539.7 1536.2 1501.5 1494.9 1492.7

shift in the presence of H bonding to a water molecule was eliminated by removal of the water molecule in all dielectric representations, demonstrating that the shift was a result of the H-bonding interaction, rather than of a change in the geometry of the aromatic molecule, itself. Out-of-Plane Skeletal Vibrations. Modes 4 and 16a/b are classified as out-of-plane skeletal vibrations. These three modes showed quite variable results as a function of the solvation environment in our calculations. Neither mode was drastically affected by changing dielectric. The effect of the water molecule H bond was inconsistent between the two molecules and in the various dielectric representations. All of the calculated effects of dielectric and specific H bonding were small. C-H Out-of-Plane Vibrations. Modes 5 and 10a fall into this class of vibrations. Like the out-of-plane skeletal vibrations, the effects of the varied solvation conditions were small in all cases. The effect of the presence of the water molecule was again inconsistent between tyrosine and OH-DL-dopa in the various dielectric representations. C-C Stretching Vibrations. Modes 19a and 19b are representative of the various carbon-carbon stretching vibrations of the ring. These stretching vibrations are all in-plane motions. Changing dielectric had a small effect on these modes, but this effect was consistently a shift to lower frequency with increased dielectric. The effect of the water molecule on tyrosine is clearly a small shift to higher frequency, as was seen in our experiments. For OH-DL-dopa, however, the presence of the water molecule had an inconsistent effect. The position of the water molecule in the optimized OH-DL-dopa + H2O geometry was such that the water molecule formed a strong H bond with one hydroxyl group of the ring and a weak H bond with another ring hydroxyl group. The optimized OH-DL-dopa geometry also contained an intramolecular H bond between the carbonyl oxygen and the hydroxyl group on the 2-carbon. This intramolecular H bond is unlikely to be highly populated in solution and, as can be seen from visual inspection of the structures, produces a considerably strained ring structure. These structural factors may be responsible for the lack of a consistent effect of the water molecule

on modes 19a/b for OH-DL-dopa. For tyrosine, however, the observed experimental result, a small shift to higher frequency for mode 19a, was seen in our calculations. The predicted effect of changing dielectric for all three molecules, however, is in the opposite direction, a shift to lower frequency. It is notable that we did not observe a shift of this mode to lower frequency for phenylalanine in our experiments. Electron Distribution. Electron distributions for each calculation are shown using contour maps in Figure 7. It is clear from these data that dielectric had no appreciable effect on the electron distribution. This finding is consistent with previous analyses of dielectric effects on electron distributions in hydrocarbons.78 The presence of H bonding, however, clearly did have an effect on electron distribution as seen for the 0.25 (yellow) and 0.1 (white) contours for both tyrosine and OH-DL-dopa. In both cases, at all dielectric representations, there is an extension of electron distribution between the aromatic molecule and the water molecule. This is evidence that specific H-bonding interactions have a more significant effect on the electron distribution of the ring system than the dielectric. Electrostatic Potential at the SolVent-Accessible Surface. The electrostatic potential is mapped to the solvent-accessible surface for each calculation in Figure 8. The dielectric variance produced only a change in the magnitude of the electrostatic potentials but no difference in the distribution. This is expected, based on the effects of dielectric on the electron distribution. Also consistent with the electron distribution results, the only effect on electrostatic potential distribution was seen in the presence of specific H-bonding interaction with water. Visualization of Vibrational Resonance Via Specific H Bonding. As described above, the only consistent effects of the presence of specific H bonding to water that were seen in our calculations were for modes 3 and 19a/b. Visual inspection of these modes revealed that the ring motions for these modes are clearly coupled to the H-bonded water molecule, as evidenced by motion of the water molecule in resonance with the complex ring vibration. This is shown using vector diagrams in Figure 9 for tyrosine mode 3 and 19a. Resonance is clear for mode 3 in

4032 J. Phys. Chem. B, Vol. 112, No. 13, 2008 vacuum but is barely visible for mode 19a in vacuum. As dielectric was increased, the magnitude of the resonant motion of the water molecule was reduced with mode 3 and increased dramatically with mode 19a. This appears to be direct computational evidence for the model proposed to explain the observed frequency shifts in our infrared experiments; hydrogen bonding to solvent around the periphery of the aromatic ring mediates vibrational coupling of complex modes to solvent. It is interesting that our calculations also predict an effect of dielectric on the magnitude of the resonant motion of the water molecule, despite the fact that there does not seem to be a dielectricdependent effect on the calculated frequency shift for either mode 3 or mode 19a. It is also important to note that the changing dielectric has no effect on the magnitude of any atomic motions of either mode in the absence of a water molecule for either geometry. This predicts an intricate interplay between bulk shielding effects and specific H-bonding interactions in modulating complex vibrational modes of small molecules. Discussion The present findings demonstrate that complex vibrations can couple to solvent through H bonds. We have examined an aromatic C-C stretching vibration, mode 19a, in a group of small aromatic molecules structurally related to the amino acids phenylalanine and tyrosine, each of which differs by a single hydrogen-bonding substituent. TEIR spectroscopic measurement of mode 19a over a 260 K temperature range showed that this mode is temperature independent in phenylalanine, but as additional hydrogen-bonding substituents are added to the aromatic ring, solvent dependence is imparted on this complex ring mode. Mode 19a is one of the most frequency-insensitive vibrations to substitution about the aromatic ring.35 The frequency of this mode at room temperature varied little between the molecules examined in this study, despite their considerably different substitution. For this reason, we can reasonably conclude that the shifts of this mode with temperature are indicative of strong effects on the ring vibrations. The largest observed shift with temperature is small, only 3 cm-1, but this is more than 10% of the room-temperature frequency difference between phenylalanine and OH-DL-dopa, which differ by three hydroxyl groups covalently bonded to the ring. It is well-established that the H-bonding behavior of water/ glycerol is quite temperature dependent. As temperature is reduced, the average strength and linearity of H bonds in this solvent increase dramatically.45,46 We have previously shown that stretching vibrations of H-bonding substituent groups of biomolecules in water/glycerol are coupled to solvent and shift to lower frequency with reduced temperature as a result of strengthened H-bonding interactions with solvent.26,29,45,79-82 The present study examines more complex vibrational modes, the response of which is difficult to predict in response to changes in H bonding. All of the spectral shifts observed in this study occur almost completely in the liquid regime of the temperature range. The water/glycerol mixture used in this study has a glass transition temperature of approximately 150-160 K. Infrared spectra only slightly sharpened below this glass transition temperature; absorption frequencies did not shift between this transition and 30 K. Mode 19a shifted to a higher frequency in all cases where a shift was observed, but there is no established model for interpreting such a shift. In addition to the strengthening of the average H bond in water/glycerol as temperature drops, the relative static permittivity and the dielectric relaxation behavior

Nucci et al. also change dramatically.47,50,51 Dielectric effects on vibrational frequencies are well-known, particularly small shifts as a function of solvent polarity. The dielectric constant of water/ glycerol approximately doubles over the liquid regime of the experimental temperature range,48 so in the absence of additional data, dielectric effects are certainly a potential explanation of the infrared results. Determination of whether the evolution of temperature dependence with increasing H-bond substitution is a function of vibrational coupling through H-bonds required investigation beyond the infrared spectra alone. Temperature-dependence of the fluorescence emission and low-temperature phosphorescence were examined to add further insight on the solvent dependence of these small molecules. As with the infrared results, a clear correlation was seen between the degree of H-bonding substitution and the solvent dependence of the maximum fluorescence emission wavelength, the vibrational structure visible in the fluorescence emission, the phosphorescence vibronic resolution, and the phosphorescence lifetime. Solvent dependence of the emission wavelength at room temperature is indicative of the differences in solvent relaxation around the excited state. When the dipole of the excited state is larger than that of the ground state, a redshift of the emission spectrum is seen with more polar solvent. The excited state of phenol is known to have a larger dipole difference from the ground state than is seen for the excited versus ground state of benzene; thus the emission spectrum of phenol redshifts more with more polar solvent, whereas the spectrum of benzene does not.83 In water/glycerol, the polarity, as described by the dielectric constant, increases dramatically with reduced temperature.48 If the temperature-dependent spectral shift were a function of solvent polarity alone, a redshift should be seen for tyrosine as temperature drops where none should be seen for phenylalanine. Instead, a blueshift was seen for all molecules except phenylalanine. The observed shift is better understood in terms of the local dielectric relaxation than as a function of the bulk solvent polarity. As temperature drops, the ability of the solvent to reorient around the excited-state slows, as it must based on the fact that the dielectric relaxation of glycerol/water falls with decreasing temperature.47,50,51 As a result, the energy of the excited state at emission gets higher as temperature drops because the solvent cannot completely relax within the excited-state lifetime, causing a blueshift of the emission. This effect has been seen before for other fluorescent molecules, e.g.,84 and the present data suggests that hydrogen-bonding substitutions make the dipole change between ground and excited states larger. This model for the spectral shift is consistent with previous analyses of excited-state dipole changes for substituted benzenes, so it is reasonable to conclude that the same relationship would exist for the present family of molecules.83 The difference in vibrational substructure in the emission spectra between the different molecules is also indicative of local H-bonding interactions. Phenylalanine shows considerable vibrational substructure; tyrosine shows very slight substructure; none of the other molecules show any appreciable vibronic resolution. This spectral feature is readily accounted for by the differences in H-bonding character of the molecules. It has been known for many years that H-bonding interactions by fluorophores produce phonon distortions of the emission spectra, effectively leading to broadening and loss of vibronic resolution.85 The extent of vibronic resolution in the low-temperature phosphorescence of these molecules also correlates with the degree of H-bonding substitution about the aromatic ring. The

Solvent Dependence of Aromatic Ring Vibrations loss of vibronic resolution in phosphorescence spectra is most often due to an increase in the speed of relaxation processes of the triplet state. Our measurements of the phosphorescence lifetimes show that the lifetime shortens dramatically with increased H-bonding substitution on the ring, suggesting that the H-bonding interactions to solvent provide faster relaxation mechanisms. Such correlations between loss of vibronic resolution and H-bonding have been shown before.86 All of the electronic spectral properties point to these small molecules being more sensitive to the local H-bonding environment than to bulk solvent polarity. The lack of an effect of offring substitutions for any of the electronic spectral properties supports the interpretation of the above-mentioned spectral features as being a function of hydrogen bonding of the small molecules to solvent about the periphery of the aromatic ring. This aspect of the solvent dependence also suggests that the vibrational character of these molecules is more dependent on solvation environment than the electronic properties since offring effects were seen in the infrared studies but not in the fluorescence or phosphorescence spectra. The combined results of the temperature dependence of the infrared ring mode, the temperature dependence of the fluorescence emission, and the phosphorescence vibronic resolution and lifetime changes with H-bonding substitution all point to local H-bonding interactions as the primary causative factor for solvent dependence of these small molecules. Despite the correlative agreement of these various measurements, none of them definitively rules out dielectric effects in driving the frequency shifts of the complex vibrational mode measured. Examination of the ab initio calculation results, however, resolves this problem. Quantum chemical calculations afford the opportunity to isolate the effects of bulk dielectric changes from those of specific H-bonding interaction. The set of calculations employed here were designed to do just that in as unambiguous a way as possible. Only a single solvent molecule was used in our calculations to limit the extent of specific H-bonding interactions considered. Two dielectric constants were evaluated and compared to vacuum. The variance of the dielectric in the calculations is fairly similar in scale to what is seen in our experiments but was also chosen for biological relevance. The high dielectric, 78, is representative of bulk water solvation conditions while the low dielectric, 4, is generally representative of the environment in a protein binding pocket or within a membrane. Clearly, the presence of a single molecule specifically interacting with the aromatic molecules in a static H-bonding conformation was not fully representative of the solution conditions in our experiments. This set of calculations did, however, allow us to establish predictions for the vibrational frequency shifts induced by changing dielectric conditions and specific H-bonding interactions. The results of the quantum calculations predict very small effects of drastically changing dielectric on vibrational frequencies. They also predict significant effects of specific H-bonding interactions on only two of the evaluated modes. Both of these modes, however, are in-plane vibrations, while all modes for which little specific H-bonding effect is predicted are out-ofplane vibrations. These in-plane vibrations are known to be more sensitive to the molecular top character of the molecule than out-of-plane modes, so specific H-bonding to solvent significantly affecting only these modes is consistent with a model of vibrational coupling through the H bond. The finding that neither the electron distribution nor the electrostatic potential distribution changed with varying dielectric, while both changed in the

J. Phys. Chem. B, Vol. 112, No. 13, 2008 4033 presence of a specific H-bonding interaction, also supports the argument that H bonding is what drives the solvent dependence of the observed vibrational mode. The predicted effect of varying dielectric on the experimentally observed vibrational mode, mode 19a, is a shift to lower frequency with increased dielectric, corresponding to reduced temperature in our experiments. The predicted effect of specific H-bonding is a shift to higher frequency. A shift to higher frequency was observed in the experiment, so the H-bonding explanation of this shift is consistent, while a dielectric explanation is not. Finally, visible resonance in the vibrational modes that show frequency effects of H bonding to solvent (Figure 9) clearly demonstrate that the experimentally observed effects are the result of vibrational coupling between solute and solvent via H bonding. Resonant motion was only seen for the two complex ring vibrations, modes 3 and 19a, and for the O-H stretching motions of the water molecule and the aromatic hydroxyl group H-bonded to the water (not shown). The magnitude of the resonant motion showed dependence on the dielectric for both complex modes, though the order of dependence was reversed between these two modes. Resonance increased for mode 19a of tyrosine with increasing dielectric, as was seen in the experimental measurement of this vibration. Resonance for mode 3, conversely, decreased with increasing dielectric. This result, while offering confirming evidence that the experimentally observed effect of temperature excursion was most likely the result of throughH-bond vibrational coupling between solute and solvent, strongly suggests that there is an intricate interplay between specific H-bonding interactions and bulk solvent properties. This finding is not unexpected, of course, but it does offer a complicated picture of the solvent dependence of low-frequency solute modes that will require further investigation to fully describe. Implications of Complex Mode Coupling through H Bonds. Coupling of complex vibrational modes of small molecules through H-bonding interactions shows that such molecules are intimately sensitive to their solvation microenvironment. The fact that their low-frequency motions can be dependent on this microenvironment suggests the possibility that these low-frequency modes may have relevance with respect to biomolecular function. The question of whether lowfrequency modes of biological molecules have a role in structure-function relationships is currently a subject of much investigation and debate. The central points of this debate center on the continuing effort to develop our understanding of the role of dynamic processes in how one molecule specifically binds to another. The present findings expand an already active area of investigation, namely, how vibrational coupling may play a role in biological function. The dynamic processes involved in molecular recognition remain poorly understood, though much progress has recently been made on this frontier. Vibrational coupling has been measured across H bonds between helices in a protein.87 The role of protein entropy, an important element of dynamic behavior, in protein-peptide binding has been described, demonstrating that protein entropy and system entropy are intimately tied in specific binding interactions.88 Vibrational excitation has even been demonstrated to play a role in directing chemical reactions, modulating the specific location of substitution reactions.89 Much of this work has come with the use of time-resolved spectroscopic methods, but computational methods have also offered many important insights into dynamic processes. As mentioned above, complex backbone modes have been modeled by Dannenberg and co-workers.21-25 They have shown

4034 J. Phys. Chem. B, Vol. 112, No. 13, 2008 that substantial cooperativity exists in these backbone modes and that these cooperative motions are intimately dependent on coupling through H bonds.21-25 Perhaps the most widely applied computational method over the past two decades for understanding dynamic processes of proteins has been analysis of protein normal modes.90 Normal-mode analysis consists of mapping protein motions computationally by simulating protein motion using only a handful of the protein’s lowest frequency vibrational modes. Several models have been developed for such analysis and have had considerable success. Conformational pathways from inactive to active states of many proteins, including soluble proteins such as calmodulin, myosin, and hemoglobin,91 have been mapped using normal-mode analysis methods. Recently, conformational change of a membrane protein, the glycine R1 receptor, was also successfully mapped.92 On the basis of the success of these methods, the relevance of low-frequency modes in biological systems seems clear: lowfrequency motions of biological molecules seem to be the vehicle of molecular activation/inactivation processes. Within the context of this knowledge, the present findings suggest that the environmental sensitivity of ligand vibrational modes may also be of importance in biological activity. The idea of small molecule vibrations playing a role in binding specificity is by no means a new one. This concept was first proposed in the 1930s by Dyson,93 who suggested that the vibrational properties of odorant molecules plays a role in their detection by olfactory receptors. Dyson’s ideas were revisited in the 1970s by Wright and colleagues94 and more recently by Turin.95 Turin has argued that vibrations play the principle role in odorant detection via vibration-driven electron tunneling. This idea has been both supported96,97 and refuted.98 On the basis of the considerable success that computational molecular docking methods have offered over the past two decades, surface complimentarity and shape factors are clearly the principle driving factors in small molecules binding to proteins, but they do not account for all of the binding energy since these docking methods cannot predict binding interactions with high accuracy. The present data surely raise the question of what roles ligand low-frequency modes may play in the subtle determinants of specificity. Conclusions We have demonstrated that complex vibrational modes can couple to solvent through H-bonding interactions. The degree of H-bonding substitution, both on- and off-ring, affects the magnitude and extent of environmental dependence of the small aromatic molecule vibrational modes. These vibrational data were supported by similar solvent dependence in the fluorescence spectra, phosphorescence vibronic resolution, and phosphorescence lifetimes of the small molecules tested. Quantum calculations demonstrated that the observed temperature dependence of a complex ring mode is the result of through-Hbond vibrational coupling. The results of the calculations also revealed, as one might expect, that the solvent dependence of these small molecules is the result of a complex interplay of specific H-bond interactions and bulk solvation effects. These findings suggest that there may be important implications of small aromatic molecule sensitivity to the solvation microenvironment in their biological activity. Extensive further experiment will be required to clarify if small molecule low-frequency modes play a role in binding specificity and, if so, to tease out the significance of these likely subtle, cooperative effects. Acknowledgment. We thank Drs. Josh Wand, Kim Sharp, Jeff Saven, Bogumil Zelent, and Jennifer Greene for helpful

Nucci et al. discussions. Funding was provided by NIH P01GM48103 to J.M.V. and NIH F31NS53399 to N.V.N. Supporting Information Available: The optimized geometries of tyrosine and OH-DL-dopa are provided in Z-matrix format both in the presence and absence of a water molecule hydrogen bound to the terminal hydroxyl(s). The optimized geometry of phenylalanine alone in vacuum is also provided. This material is available free of charge via the Internet at http:// pubs.acs.org. References and Notes (1) Gilson, M. K.; Zhou, H.-X. Annu. ReV. Biophys. Biomol. Struct. 2007, 36, 21. (2) Lemieux, R. U. Acc. Chem. Res. 1996, 29, 373. (3) Levy, Y.; Onuchic, J. N. Annu. ReV. Biophys. Biomol. Struct. 2006, 35, 389. (4) Mobley, D. L.; Dumont, E.; Chodera, J. D.; Dill, K. A. J. Phys. Chem. B 2007, 111, 2242. (5) Plumridge, T. H.; Waigh, R. D. J. Pharm. Pharmacol. 2002, 54, 1155. (6) Gundogdu, K.; Bandaria, J.; Nydegger, M.; Rock, W.; Cheatum, C. M. J. Chem. Phys. 2007, 127, 044501/1. (7) Heyne, K.; Huse, N.; Nibbering, E. T. J.; Elsaesser, T. Chem. Phys. Lett. 2003, 382, 19. (8) Huse, N.; Bruner, B. D.; Cowan, M. L.; Dreyer, J.; Nibbering, E. T. J.; Miller, R. J. D.; Elsaesser, T. Phys. ReV. Lett. 2005, 95, 147402/1. (9) Madsen, D.; Stenger, J.; Dreyer, J.; Hamm, P.; Nibbering, E. T. J.; Elsaesser, T. Bull. Chem. Soc. Jpn. 2002, 75, 909. (10) Nandi, C. K.; Hazra, M. K.; Chakraborty, T. J. Chem. Phys. 2005, 123, 124310/1. (11) Rubtsov, I. V.; Wang, J.; Hochstrasser, R. M. J. Phys. Chem. A 2003, 107, 3384. (12) Carmona, P.; Molina, M.; Aboitiz, N.; Vicent, C. Biopolymers 2002, 67, 20. (13) Gallina, M. E.; Sassi, P.; Paolantoni, M.; Morresi, A.; Cataliotti, R. S. J. Phys. Chem. B 2006, 110, 8856. (14) Jarmelo, S.; Reva, I.; Rozenberg, M.; Carey, P. R.; Fausto, R. Vib. Spectrosc. 2006, 41, 73. (15) Asselin, P.; Goubet, M.; Lewerenz, M.; Soulard, P.; Perchard, J. P. J. Chem. Phys. 2004, 121, 5241. (16) Ebata, T.; Fujii, A.; Mikami, N. Int. ReV. Phys. Chem. 1998, 17, 331. (17) Ebata, T.; Nagao, K.; Mikami, N. Chem. Phys. 1998, 231, 199. (18) Hartland, G. V.; Henson, B. F.; Venturo, V. A.; Felker, P. M. J. Phys. Chem. 1992, 96, 1164. (19) Tanabe, S.; Ebata, T.; Fujii, M.; Mikami, N. Chem. Phys. Lett. 1993, 215, 347. (20) Zwier, T. S. Annu. ReV. Phys. Chem. 1996, 47, 205. (21) Chen, Y.-F.; Viswanathan, R.; Dannenberg, J. J. J. Phys. Chem. B 2007, 111, 8329. (22) Viswanathan, R.; Asensio, A.; Dannenberg, J. J. J. Phys. Chem. A 2004, 108, 9205. (23) Kobko, N.; Dannenberg, J. J. J. Phys. Chem. A 2003, 107, 10389. (24) Wieczorek, R.; Dannenberg, J. J. J. Am. Chem. Soc. 2003, 125, 14065. (25) Dannenberg, J. J.; Kobko, N. Abstracts of Papers, 226th ACS National Meeting, New York, United States, September 7-11, 2003. (26) Nucci, N. V.; Vanderkooi, J. M. Temperature Excursion Infrared Spectroscopy. In Methods in Protein Structure and Stability Analysis: Vibrational Spectroscopy; Uversky, V. N., Ed.; Nova Science Publishing: Hauppauge, NY, 2007. (27) Norman, A. W.; Litwack, G. Hormones, 2nd ed.; Academic Press: Boston, MA, 1997. (28) U’Prichard, D. C. Direct Binding Studies of R-Adrenoreceptors. In Adrenoreceptors and Catecholamine Action; Kunos, G., Ed.; John Wiley & Sons: New York, 1981; Vol. A. (29) Vanderkooi, J. M.; Dashnau, J. L.; Zelent, B. Biochim. Biophys. Acta 2005, 1749, 214. (30) Jeffrey, G. A. An Introduction to Hydrogen Bonding; Oxford University Press: New York, 1997. (31) Sato, H.; Murakami, R.; Zhang, J.; Mori, K.; Takahashi, I.; Terauchi, H.; Noda, I.; Ozaki, Y. Macromol. Symp. 2005, 230, 158. (32) Taguet, A.; Ameduri, B.; Boutevin, B. AdV. Polym. Sci. 2005, 184, 127. (33) Gruebele, M. Spectrum 2002, 15, 13. (34) Wilson, E. B., Jr. Phys. ReV. 1934, 45, 706. (35) Varsanyi, G. Assignments for Vibrational spectra of seVen hundred benzene deriVatiVes; John Wiley & Sons: New York, 1974; Vol. 1.

Solvent Dependence of Aromatic Ring Vibrations (36) Chi, Z.; Asher, S. A. J. Phys. Chem. B 1998, 102, 9595. (37) Kodicek, M. FEBS Lett. 1979, 98, 295. (38) Lee, J. K.; Ross, R. T. J. Phys. Chem. B 1998, 102, 4612. (39) Menter, J. M. Photochem. Photobiol. Sci. 2006, 5, 403. (40) Sakurovs, R.; Ghiggino, K. P. Aust. J. Chem. 1981, 34, 1367. (41) Sudhakar, K.; Wright, W. W.; Williams, S. A.; Phillips, C. M.; Vanderkooi, J. M. J. Fluor. 1993, 3, 57. (42) Lelkes, P. I.; Friedman, J. E.; Rosenheck, K. J. Neurosci. Methods 1985, 13, 249. (43) Lapainis, T.; Scanlan, C.; Rubakhin, S. S.; Sweedler, J. V. Anal. Bioanal. Chem. 2007, 387, 97. (44) Yoshitake, M.; Nohta, H.; Yoshida, H.; Yoshitake, T.; Todoroki, K.; Yamaguchi, M. Anal. Chem. 2006, 78, 920. (45) Dashnau, J. L.; Nucci, N. V.; Sharp, K. A.; Vanderkooi, J. M. J. Phys. Chem. B 2006, 110, 13670. (46) Zelent, B.; Nucci, N. V.; Vanderkooi, J. M. J. Phys. Chem. A 2004, 108, 11141. (47) Hayashi, Y.; Puzenko, A.; Feldman, Y. J. Non-Cryst. Solids 2006, 352, 4696. (48) Huck, J. R.; Noyel, G. A.; Jorat, L. J. IEEE Trans. Electr. Insul. 1988, 23, 627. (49) Murthy, S. S. N. J. Phys. Chem. B 2000, 104, 6955. (50) Puzenko, A.; Hayashi, Y.; Feldman, Y. J. Non-Cryst. Solids 2007, 353, 4518. (51) Puzenko, A.; Hayashi, Y.; Ryabov Yaroslav, E.; Balin, I.; Feldman, Y.; Kaatze, U.; Behrends, R. J. Phys. Chem. B 2005, 109, 6031. (52) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb, M. A.; Cheeseman, J. R.; Montgomery, J. A.; Vreven, T.; Kudin, K. N.; Burant, J. C.; Milam, 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.; 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, M.; Johnson, B.; Chen, W.; Wong, M. W.; Gonzalez, C.; Pople, J. A. Gaussian 03, revision C.02; Gaussian, Inc.: Wallingford, CT, 2004. (53) Lee, C.; Yang, W.; Parr, R. G. Phys. ReV. B 1988, 37, 785. (54) Miehlich, B.; Savin, A.; Stoll, H.; Preuss, H. Chem. Phys. Lett. 1989, 157, 200. (55) Becke, A. D. J. Chem. Phys. 1993, 98, 5648. (56) Curtiss, L. A.; Jones, C.; Trucks, G. W.; Raghavachari, K.; Pople, J. A. J. Chem. Phys. 1990, 93, 2537. (57) Pople, J. A.; Head-Gordon, M.; Fox, D. J.; Raghavachari, K.; Curtiss, L. A. J. Chem. Phys. 1989, 90, 5622. (58) Curtiss, L. A.; McGrath, M. P.; Blaudeau, J.-P.; Davis, N. E.; Binning, R. C., Jr.; Radom, L. J. Chem. Phys. 1995, 103, 6104. (59) Binning, R. C., Jr.; Curtiss, L. A. J. Comput. Chem. 1990, 11, 1206. (60) McLean, A. D.; Chandler, G. S. J. Chem. Phys. 1980, 72, 5639. (61) Krishnan, R.; Binkley, J. S.; Seeger, R.; Pople, J. A. J. Chem. Phys. 1980, 72, 650. (62) Wachters, A. J. H. J. Chem. Phys. 1970, 52, 1033. (63) Hay, P. J. J. Chem. Phys. 1977, 66, 4377.

J. Phys. Chem. B, Vol. 112, No. 13, 2008 4035 (64) Raghavachari, K.; Trucks, G. W. J. Chem. Phys. 1989, 91, 2457. (65) McGrath, M. P.; Radom, L. J. Chem. Phys. 1991, 94, 511. (66) Palafox, M. A.; Gill, M.; Nunez, N. J.; Rastogi, V. K.; Mittal, L.; Sharma, R. Int. J. Quantum Chem. 2005, 103, 394. (67) Palafox, M. A.; Nunez, J. L.; Gill, M. J. Mol. Struct. 2002, 593, 101. (68) Foresman, J. B.; Keith, T. A.; Wiberg, K. B.; Snoonian, J.; Frisch, M. J. J. Phys. Chem. 1996, 100, 16098. (69) Suenaga, M. Facio Ver. 10.9.7, 2006. (70) Alagona, G.; Ghio, C. Chem. Phys. 1996, 204, 239. (71) Fausto, R.; Joao, M.; Ribeiro, S.; Pedroso de Lima, J. J. J. Mol. Struct. 1999, 484, 181. (72) Nagy, P. I.; Alagona, G.; Ghio, C. J. Am. Chem. Soc. 1999, 121, 4804. (73) Nagy, P. I.; Alagona, G.; Ghio, C.; Takacs-Novak, K. J. Am. Chem. Soc. 2003, 125, 2770. (74) Nagy, P. I.; Takacs-Novak, K. Phys. Chem. Chem. Phys. 2004, 6, 2838. (75) Nagy, P. I.; Voelgyi, G.; Takacs-Novak, K. Mol. Phys. 2005, 103, 1589. (76) Urban, J. J.; Cramer, C. J.; Famini, G. R. J. Am. Chem. Soc. 1992, 114, 8226. (77) Longworth, J. W. Biochem. J. 1961, 81, 23, 1/2. (78) Karelson, M. M.; Katritzky, A. R.; Zerner, M. C. Int. J. Quantum Chem. 1986, 20, 521. (79) Wright, W. W.; Guffanti, G. T.; Vanderkooi, J. M. Biophys. J. 2003, 85, 1980. (80) Walsh, S. T. R.; Cheng, R. P.; Wright, W. W.; Alonso, D. O. V.; Daggett, V.; Vanderkooi, J. M.; DeGrado, W. F. Protein Sci. 2003, 12, 520. (81) Wright, W. W.; Carlos Baez, J.; Vanderkooi, J. M. Anal. Biochem. 2002, 307, 167. (82) Manas, E. S.; Getahun, Z.; Wright, W. W.; DeGrado, W. F.; Vanderkooi, J. M. J. Am. Chem. Soc. 2000, 122, 9883. (83) Malar, E. J. P.; Jug, K. J. Phys. Chem. 1984, 88, 3508. (84) Zelent, B.; Troxler, T.; Vanderkooi, J. M. J. Fluoresc. 2007, 17, 37. (85) Friedrich, J.; Haarer, D. J. Chem. Phys. 1982, 76, 61. (86) Vieira Ferreira, L. F.; Vieira Ferreira, M. R.; Oliveira, A. S.; Moreira, J. C. J. Photochem. Photobiol., A 2002, 153, 11. (87) Fang, C.; Senes, A.; Cristian, L.; DeGrado, W. F.; Hochstrasser, R. M. Proc. Natl. Acad. Sci. U.S.A. 2006, 103, 16740. (88) Frederick, K. K.; Marlow, M. S.; Valentine, K. G.; Wand, A. J. Nature 2007, 448, 325. (89) Yan, S.; Wu, Y.-T.; Zhang, B.; Yue, X.-F.; Liu, K. Science 2007, 316, 1723. (90) Case, D. A. Curr. Opin. Struct. Biol. 1994, 4, 285. (91) Petrone, P.; Pande, V. S. Biophys. J. 2006, 90, 1583. (92) Bertaccini, E. J.; Trudell, J. R.; Lindahl, E. J. Chem. Inf. Model. 2007, 47, 1572. (93) Dyson, G. M. Chem. Ind. 1938, 647. (94) Wright, R. H. J. Theor. Biol. 1977, 64, 473. (95) Turin, L. Chem. Sens. 1996, 21, 773. (96) Brookes, J. C.; Hartoutsiou, F.; Horsfield, A. P.; Stoneham, A. M. Phys. ReV. Lett. 2007, 98, 038101/1. (97) Haffenden, L. J. W.; Yaylayan, V. A.; Fortin, J. Food Chem. 2001, 73, 67. (98) Keller, A.; Vosshall, L. B. Nat. Neurosci. 2004, 7, 337.