Vibrational Probes: From Small Molecule ... - ACS Publications

Mar 27, 2017 - Republic of Korea. ‡. Department of ... The University of Texas at Austin, 105 E. 24th Street, STOP A5300, Austin, Texas 78712, Unite...
0 downloads 0 Views 3MB Size
Article pubs.acs.org/accounts

Vibrational Probes: From Small Molecule Solvatochromism Theory and Experiments to Applications in Complex Systems Bartosz Błasiak,†,‡ Casey H. Londergan,§ Lauren J. Webb,∥ and Minhaeng Cho*,†,‡ †

Center of Molecular Spectroscopy and Dynamics, Institute of Basic Science (IBS), 145 Anam-ro, Seongbuk-gu, Seoul 02841, Republic of Korea ‡ Department of Chemistry, Korea University, 145 Anam-ro, Seongbuk-gu, Seoul 02841, Republic of Korea § Department of Chemistry, Haverford College, Haverford, Pennsylvania 19041-1392, United States ∥ Department of Chemistry, Center for Nano- and Molecular Science and Technology, and Institute for Cell and Molecular Biology, The University of Texas at Austin, 105 E. 24th Street, STOP A5300, Austin, Texas 78712, United States S Supporting Information *

CONSPECTUS: The vibrational frequency of a chosen normal mode is one of the most accurately measurable spectroscopic properties of molecules in condensed phases. Accordingly, infrared absorption and Raman scattering spectroscopy have provided valuable information on both distributions and ensemble-average values of molecular vibrational frequencies, and these frequencies are now routinely used to investigate structure, conformation, and even absolute configuration of chemical and biological molecules of interest. Recent advancements in coherent time-domain nonlinear vibrational spectroscopy have allowed the study of heterogeneous distributions of local structures and thermally driven ultrafast fluctuations of vibrational frequencies. To fully utilize IR probe functional groups for quantitative bioassays, a variety of biological and chemical techniques have been developed to site-specifically introduce vibrational probe groups into proteins and nucleic acids. These IR-probelabeled biomolecules and chemically reactive systems are subject to linear and nonlinear vibrational spectroscopic investigations and provide information on the local electric field, conformational changes, site−site protein contacts, and/or function-defining features of biomolecules. A rapidly expanding library of data from such experiments requires an interpretive method with atom-level chemical accuracy. However, despite prolonged efforts to develop an all-encompassing theory for describing vibrational solvatochromism and electrochromism as well as dynamic fluctuations of instantaneous vibrational frequencies, purely empirical and highly approximate theoretical models have often been used to interpret experimental results. They are, in many cases, based on the simple assumption that the vibrational frequency of an IR reporter is solely dictated by electric potential or field distribution around the vibrational chromophore. Such simplified description of vibrational solvatochromism generally referred to as vibrational Stark effect theory has been considered to be quite appealing and, even in some cases, e.g., carbonyl stretch modes in amide, ester, ketone, and carbonate compounds or proteins, it works quantitatively well, which makes it highly useful in determining the strength of local electric field around the IR chromophore. However, noting that the vibrational frequency shift results from changes of solute−solvent intermolecular interaction potential along its normal coordinate, Pauli exclusion repulsion, polarization, charge transfer, and dispersion interactions, in addition to the electrostatic interaction between distributed charges of both vibrational chromophore and solvent molecules, are to be properly included in the theoretical description of vibrational solvatochromism. Since the electrostatic and nonelectrostatic intermolecular interaction components have distinctively different distance and orientation dependences, they affect the solvatochromic vibrational properties in a completely different manner. Over the past few years, we have developed a systematic approach to simulating vibrational solvatochromic data based on the effective fragment potential approach, one of the most accurate and rigorous theories on intermolecular interactions. We have further elucidated the interplay of local electric field with the general vibrational solvatochromism of small IR probes in either solvents or complicated biological systems, with emphasis on contributions from non-Coulombic intermolecular interactions to vibrational frequency shifts and fluctuations. With its rigorous foundation and close relation to quantitative interpretation of experimental data, this and related theoretical approaches and experiments will be of use in studying and quantifying the structure and dynamics of biomolecules with unprecedented time and spatial resolution when combined with time-resolved vibrational spectroscopy and chemically sensitive vibrational imaging techniques.



INTRODUCTION Molecular vibrational spectroscopy plays an increasingly important role in modern chemistry and biology due to its simplicity and remarkable sensitivity to the changes in external environment and molecular structure.1−4 Infrared and Raman © 2017 American Chemical Society

spectroscopies including rapidly developing techniques such as two-dimensional (2D) vibrational spectroscopy provide Received: January 1, 2017 Published: March 27, 2017 968

DOI: 10.1021/acs.accounts.7b00002 Acc. Chem. Res. 2017, 50, 968−976

Article

Accounts of Chemical Research invaluable insights into the molecular granularity and the associated dynamics down to the femtosecond time scale.5 Direct probing of an IR peak position can often indicate the extent of exposure to water6−8 and reveal the distinct conformations of biomolecules that could be otherwise difficult to observe by using other experimental techniques with relatively low time resolution.6,9 However, quantitative interpretation of experimental spectra is still a difficult challenge not only because of the topological complexity of the studied model or biological systems but also because of the intricate interplay of the diverse physical phenomena governing the vibrational solvatochromism at the molecular level.10−18 Despite the existence of very simple and systematic empirical calibrations of the vibrational frequency with respect to sets of certain solute- or solvent-specific parameters,19,20 interpreting the frequency is not always clear and does not provide detailed insights at the microscale. Over the past decade, a number of theories have been developed13,16,21 based mainly on electrostatic interactions, considering the local electric potential or its gradient around infrared (IR) probe groups. IR probes are indeed very useful in site-specific detection of both the local and nonlocal structural variations of the environment and ultrafast solvent dynamics, thus providing a direct access to the nature of biomolecules in water. Apart from a few exceptions such as OH stretch mode of water molecule that has been used to probe water H-bonding network structure and dynamics, most vibrational probes widely used in vibrational spectroscopy and imaging studies have frequencies that are in the spectrally transparent window (from 2000 to 2600 m−1) of aqueous biomolecule solutions, e.g., nitrile (−CN) stretch, thiocyanate (−SCN) stretch, asymmetric azide (−N3) stretch, isonitrile (−NC) stretch, C−D stretch, O−D stretch, S−H stretch, −CC− stretch, and so on. There are important anomalies in this family of vibrational probes that seem to follow quite complicated rules and deviate from a purely electrochromic behavior.4,17 In Figure 1a, we summarize a list of useful vibrational probes for not just spectroscopic but also chemical imaging applications. Although a simple rule-of-thumb guidance on which effects are important for a particular case is difficult to draw at this point, it is clear that electric field is not the only and decisive factor in these cases. Therefore, it is of tremendous importance to thoroughly understand the origin of vibrational frequency shifts and accompanying line shape changes of vibrational reporters to further utilize them in complicated and heterogeneous systems such as proteins, lipid membranes, and nucleic acids.



Figure 1. Vibrational solvatochromism. (a) List of popular vibrational probes. (b) Nitrile vibrational frequencies of MeCN in various solvents of varying polarity and proticity at room temperature. In the cases of MeCN solutions in methanol, ethanol, and propanol, the CN stretch IR bands of MeCN exhibit two peaks and their frequencies are plotted here. Data were taken from ref 17.

(X = O, S, and Se) stretch, and azide asymmetric stretch frequencies become blue-shifted upon forming a specific H-bond with protic solvent molecules.23,24 Moreover, the CN frequency is also red-shifted by increased polarity, whereas N3 and NC frequencies are relatively insensitive to solvent polarity.23,24 This remarkable difference between these IR probes is a direct consequence of the differences in the intermolecular interactions they make with molecules in their immediate environments. A linear relationship between the vibrational absorption energy and the local solvent electric field was found for CO stretches, regardless of their H-bonding status. This linear relationship between field and frequency should not be considered as a general rule for other vibrations, however, even though such linear correlations were also found for nitrile probes in aprotic solvents. Recently, a few modified models were reported, which approximately and empirically take into account H-bonding effects on vibrational frequency shifts in an ad hoc manner.13,25 However, it should be clearly noted that the nonCoulombic contributions to vibrational solvatochromism cannot be described in a unified manner with theoretical models based only on Coulomb interactions.

VIBRATIONAL SOLVATOCHROMISM

Vibrational Solvatochromism Experiments with Infrared Probes

Molecular vibrational frequencies depend on many environmental factors that can be grouped into two categories: (i) electrostatic interaction of distributed polarizable charge densities of solute and solvent molecules and (ii) the remaining short-range solute−solvent interactions, including hydrogen bonding. These factors can be relatively easily varied by choosing a solvent with desired characteristics11,17,22,23 (Figure 1b). It was noticed that the vibrational frequency of carbonyl (CO) stretch modes are generally red-shifted when increasing the polarity of the medium as well as the H-bond strength.1 However, nitrile (CN) and isonitrile (NC) symmetric stretches, X−CN

Vibrational Solvatochromism or Electrochromism?

Electrostatic interactions have long been considered the dominant driving force in electronic as well as vibrational solvatochromisms. In particular, the vibrational Stark effect (VSE) has become the 969

DOI: 10.1021/acs.accounts.7b00002 Acc. Chem. Res. 2017, 50, 968−976

Article

Accounts of Chemical Research tool that potentially provides a direct link between the vibrational frequency and the external electric field. Over the last 15 years, more sophisticated semiempirical electrostatic models of vibrational solvatochromism have been developed.13,16,21,26,27 The vibrational frequency shift in all these models (including the simple first-order VSE) can be described by the following equation28 Δω =

∑ x

{l ϕ(r ) − L ·F(r ) − 13 Λ : ∇ ⊗ F(r ) + ...} x

x

x

x

x

x

Figure 2. Schematic of solvation-induced vibrational frequency shift. Methyl thiocyanate (MeSCN) molecule is chosen as an example of IR probe. Molecular structure in solution differs from its gas-phase structure due to solute−solvent interaction and potential anharmonicity. The second derivative of the solute−solvent interaction potential with respect to the normal mode (the first term in eq 3) changes its force constant, and solvation-induced structural distortion also results in the change of IR probe frequency.

(1)

where ϕ is the electrostatic potential and F is the electric field, all evaluated at distributed sites of the solute molecule.29 The parameters lx, Lx, and Λx can be determined from fitting of the benchmark vibrational frequencies to the electric potentials, electric fields, and even higher electric field-gradients, but they can also be computed from first principles.16 If eq 1 is approximated by considering the lowest-order electric field at molecular center r0,30 the vibrational frequency shift is simply given as Δω = −L ·F(r0) ≅ −μ VSE F(r0)

The solute−solvent interaction potential can be decomposed into five fundamental contributions34 U ≈ U Coul + U Ex − Rep + U Ind + U Disp + U CT

(2)

Here, the proportionality constant, μVSE, is called the vibrational Stark tuning rate and its magnitude (but not sign) can be measured with VSE spectroscopy. If spatial variation of the electric field are on the molecular length scale, the approximate result in eq 2 is not applicable. Nonetheless, the more general eq 1 is still based on the assumption that the solvation effect can be perfectly described by a collection of solvent point charges producing a spatially nonuniform electric potential around the IR probe. Within this model of eq 1, the vibrational frequency shift is nothing but a solvation-induced electrochromism, not the general vibrational solvatochromism induced by total solute−solvent interactions. However, we showed that the solvent electric field is not necessarily the dominant factor determining the vibrational frequency shifts of a variety of IR probes in condensed phases and biological systems. Other factors could be included in the fitting scheme with eq 1 in an ad hoc manner, which was our earlier theoretical attempts to quantitatively describe vibrational solvatochromism.13 In addition, polarization and dispersion effects were empirically treated by extensive force field-derived parametrizations.31,32 Recently, we have extended our theory beyond the electrostatic multipole picture and developed an ab initio approach to vibrational solvatochromism by taking into account the electrostatic and nonelectrostatic effects altogether.14,15

Each term in eq 4 has its distinct distance/orientationdependence and strength. The first term, UCoul, which accounts for the Coulombic interactions between nonpolarizable solute and solvent charge distributions, is often considered to be the dominant term since it is strongly dependent on relative orientations of solute and interacting molecules or chemical groups around the IR probe.35 The nonelectrostatic repulsion, UEx−Rep, originates from the Pauli exclusion principle, whereas the attractive induction and dispersion interactions, UInd and UDisp, are due to the induced multipoles and associated charge penetration effects. The energetic cost of the electron population flow between molecules is included in the charge transfer term, UCT. It is natural to expect that, in addition to the long-range Coulomb (first) term in eq 4, the other non-Coulomb contributions cannot be ignored, especially at short intermolecular separations. Inserting eq 4 into eq 3, we have Δω ≈ ΔωCoul + ΔωEx − Rep + Δω Ind + ΔωDisp + ΔωCT (5)

Indeed, Δω , Δω , and Δω contributions were shown to be large even for amide I modes (mainly CO stretch of peptide bond),15 nitrile stretches including thiocyanate,17,24 and the isonitrile stretch.24 We found that the ensemble-averaged Coulombic, induction, and dispersion contributions usually cause frequency red-shifts whereas the exchange-repulsion induces a frequency blue-shift.15,17 An exception is the isonitrile reporter, where we observed blue shifting Coulombic effects due to the opposite direction of the solvatochromic dipole moment as compared to that of the CN reporter.24 We also examined the distance ranges of these contributions to vibrational solvatochromism. The Coulomb interaction is a long-range effect beyond the second and 3rd solvation shells; the induction and dispersion contributions are medium range effects originating from solvent molecules in not just the first but also the second and even third solvation shells; and the exchange-repulsion is a short-range contribution from molecules in the first solvation shell and in immediate contact with the probe atoms. The detailed characteristics of the IR probe then strongly depend on the relative strengths of such distinctively different short-range interactions. In Figure 3, we show the distance- and orientation-dependence of the amide I mode frequency shift in N-methylacetamide Ex−Rep

Interaction Energy-Based Interpretations of the Vibrational Solvatochromism

According to perturbation theory with respect to both solute− solvent interaction and potential anharmonicity,33 the vibrational frequency is related to the curvature of the solute−solvent interaction potential along the normal modes and is given by28 Δωj =

⎡ 1 ⎢ ∂2 − 2Mjωj ⎢⎣ ∂Q j 2

∑ i

gijj Miωi 2

⎤ ∂ ⎥ U ∂Q i ⎥⎦

(4)

(3)

where U denotes the interaction potential, Mi and ωi are the solute’s reduced mass and vibrational frequency associated with the ith normal mode Qi, and gijj is the cubic anharmonic constant. The first term in eq 3 represents the vibrational frequency shift due to the change of harmonic force constant when the solute interacts with surrounding solvent molecules, whereas the second in eq 3 originates from a solute−solvent interactioninduced structural distortion of the solute (IR probe; Figure 2). 970

Ind

Disp

DOI: 10.1021/acs.accounts.7b00002 Acc. Chem. Res. 2017, 50, 968−976

Article

Accounts of Chemical Research

relevant when one considers complicated heterogeneous molecular environments such as interiors of proteins, in which the average intermolecular distances could become shorter due to tight packing and crowding effects. Recently, we studied the vibrational solvatochromism of the CN mode embedded in the Ral guanine nucleotide dissociation stimulator (RalGDS) protein with SCN groups introduced via cyanylation of selected cysteine residues (Figure 4).17 It was found that the

Figure 3. Distance- and orientation-dependence of the amide I frequency shift of N-methylacetamide interacting with a single water molecule. Frequency shifts were calculated by using SolEFP method. Req denotes the O···O distance in the energy-optimized dimer obtained with HF/6-311++G** method. (a) Definitions of interatomic distance Rd and angle θ. (b) Amide I mode frequency shift relative to its gas-phase value. (c) Contributions from Coulomb (red), exchange-repulsion (blue), induction (green), and dispersion (orange) interactions.

(NMA) interacting with a water molecule. Analyzing model systems of this kind is of central importance because IR probe groups embedded in macromolecules might experience similar interactions with water. Clearly, all four contributions play nonnegligible roles in shifting the vibrational frequency within the solute−solvent separations that are typical for H-bonding. The exchange-repulsion contribution is especially large because it is a direct effect of H-bonding leading to the pronounced frequency blue shifts. This major determiner of the frequency cannot be described by electric field theory at all and needs to be considered explicitly via ab initio fragment-based simulations for not only amide I modes but also nitrile stretch modes. Interestingly, we found that the Coulombic contribution induces a frequency red-shift for the quasi-linear CN···HOH configuration.17 Therefore, if a theoretical model purely based on the electrostatic term is used to describe the frequency of a nitrile stretch mode upon H-bonding interaction with protic solvent, it will lead to an incorrect interpretation since the full QM calculations predict a blue-shift (in agreement with experiment), not a red-shift. The dispersion-induced vibrational frequency red-shift of CN mode is yet another crucial effect that cannot be described by the distribution of the electric field around an IR probe.17 The dispersion term is particularly substantial in nonpolar CCl4. Interestingly, we found that in the case of CN and amide I mode solvatochromism, the dispersion term strongly modulates the vibrational frequency but does not make much difference in line width. On the contrary, the distributions of the exchangerepulsion frequency shifts are very broad and asymmetric, which could significantly contribute to the inhomogeneous line broadening of the probe’s absorption spectrum. The contributions to CN frequency shift from the exchangerepulsion and dispersion interactions may become more practically

Figure 4. Short-range frequency shift (in cm‑1) components of the CN stretch mode of SCN probe incorporated at the N29 site of the RalGDS (Ral guanine nucleotide dissociation stimulator) protein in water. (a) Free RalGDS protein. (b) RalGDS noncovalently bound to Ras (human oncoprotein p21Ras) protein mutant.

impact of the non-Coulomb contributions to CN frequency shift is non-negligible at the interface between RalGDS and another protein. Of particular interest was that the exchangerepulsion blue-shifts can be caused not only by H-bonding with water molecules or protic residues, but also by purely steric interactions with surrounding aliphatic or aromatic side chains in such spatially congested protein environments. Nonetheless, as can be seen in Figure 3 only the Coulomb term, ΔωCoul, sensitively depends on the relative orientation of interacting solvent molecules or neighboring chemical groups. Such orientation-dependence of the Coulombic solvatochromism, described by VSE, could make it still a dominant factor in those cases when molecular environment change does not cause any significant redistribution of H-bonding interaction leading to the change in ΔωEx−Rep. This can be the case when an IR probe is placed in a structurally rigid biomolecular framework with conserved H-bond structure. For instance, the green fluorescent protein (GFP) labeled by cyanophenylalanine IR probe was recently studied with VSE and fluorescence spectroscopy (Figure 5).36 971

DOI: 10.1021/acs.accounts.7b00002 Acc. Chem. Res. 2017, 50, 968−976

Article

Accounts of Chemical Research

Figure 5. CN stretch frequency versus local electric field in GFP protein. (a) Schematic of nitrile and electronic chromophores in green fluorescent protein (GFP); nitrile absorption energy; and electronic chromophore fluorescence as a function of mutations at position 203. (b) Change in emission energy due to position 203 mutation plotted against the change in field of the B state of the fluorophore, with (squares and diamonds) and without (circles) pCNF probes. The values of ΔF∥ on the x-axis are electric field projections calculated with eq 2. On the y-axis, FP stands for fluorescent protein. Colors represent the type of mutation at position 203. Best-fit lines are shown in black for wild type (solid line; r = 0.960), pCNF 145 (dashed line; r = 0.965), and pCNF 165 (dotted line; r = 0.989). Error bars represent the maximum error based on standard deviations in published Stark tuning rates. Reproduced from ref 36. Copyright 2016 American Chemical Society.

responsible for spectral line broadening, line shape, and peak position. To approximately calculate IR pump−probe or twodimensional IR spectra, one needs to calculate a bit complicated nonlinear response functions.39 To obtain such a vibrational frequency trajectory {δω(t)} over time, it is customary to conduct molecular dynamics (MD) simulations and vibrational analyses. Since direct computation of the vibrational frequencies from ab initio quantum chemistry or density functional theory (DFT) methods is typically prohibitive for large systems and for long simulation times, it has been worthwhile to develop robust semiempirical models. For example, the PM3 Hamiltonian can be reparameterized as proposed by the Skinner group in their OQM/MM (optimized quantum mechanical/molecular mechanical) model that is capable of accurately predicting single-mode IR absorption spectra.40 Notwithstanding the fact that such approaches are very useful in spectral simulations of complex systems,41 it is also important to understand the absorption line shapes of widely used IR probes specifically in terms of their interactions with the local environment. This would be of enormous help for experimentalists to quantitatively interpret vibrational spectroscopic data, frequency shifts, and line shape changes in complicated chemical and biological environments.

Although the CN group forms H-bonds with a few nearby water molecules, the vibrational frequency was found to be linearly proportional to the electric field at the midpoint of the CN bond, which was estimated from MD trajectories. This suggests that the specific quadrupolar (or higher multipolar) and non-Coulombic contributions can be constant even in varying molecular environments with different electric fields. Only then does the vibrational frequency of the IR probe linearly correlate with local electric field. Therefore, it would be useful to determine the conditions in which the exchange-repulsion and dispersion terms, which are often difficult to estimate, could be more or less separated in the analysis of experimental data using the VSE approximation.



FROM VIBRATIONAL SOLVATOCHROMISM TO LINEAR AND NONLINEAR IR SPECTRA

Linear and Nonlinear IR Spectra

The interpretational power of IR spectra in terms of the change of the local structure and the underlying dynamics can be already seen in the following approximate relation governing the linear absorption spectrum of the localized vibrational chromophore: ∞

I(ω) ∝ Re

∫−∞ ei(ω−⟨ω ⟩)t ⟨exp[−i ∫0 01

t

δω(τ )dτ ]⟩dt

EFP2//SolEFP Simulation of IR Absorption Spectra

(6)

where, δω(τ) = ω(τ) − ⟨ω01⟩, ω01 is the vibrational angular frequency, and the angle bracket represents the classical ensemble average. In eq 6, the ultrafast changes of dipole moment direction and strength that are due to reorientational motion and nonCondon effect, respectively, are not included, because their contributions to line-broadening are often negligible except for the notable cases of OH or OD stretch modes of water.37,38 Furthermore, since the fluctuating frequency is treated as a classical variable, eq 6 does not include vibrational Stokes shift contribution either. The vibrational lifetime broadening can be included simply by multiplying an exponentially decaying function that includes the vibrational lifetime. From eq 6, it is clear that the vibrational frequency shift and fluctuation are

To address this issue, we have carried out MD simulations of MeSCN in solvents CHCl3 and H2O that represent two different molecular environments: aprotic and weakly polar vs highly protic and polar. Our simulations consisted of one MeSCN molecule in either 122 CHCl3 or 213 H2O molecules in a periodic box. The systems were initially equilibrated at 300 K and subsequently ab initio Effective Fragment Potentials34 (EFP2) polarizable force field was employed for further simulations with the EFPMD program.42 Then, we used our Solvatochromism theory with Effective Fragment Potential (SolEFP) approach15 to compute the Coulombic, induction, dispersion, and exchange-repulsion frequency shifts of the CN stretch mode relative to the gas phase value. We obtained a red-shift of −2.2 cm−1 in 972

DOI: 10.1021/acs.accounts.7b00002 Acc. Chem. Res. 2017, 50, 968−976

Article

Accounts of Chemical Research

Figure 6. EFP2//SolEFP calculation results. EFP2 MD simulations of MeSCN in chloroform (upper panels) and water (lower panels) solutions at 300 K and SolEFP calculation for MeSCN provide the distributions of SCN stretch vibrational frequency shifts (a and d) from its gas-phase value, the auto- and cross-correlation functions of fluctuating vibrational frequencies (b and e), and the simulated IR spectra (c and f). Our simulated IR spectra are frequency-shifted to make direct comparisons of our results with experimentally measured ones (see the Supporting Information).

CHCl3 and −6.2 cm−1 in water, which are close to those computed by using the PM3/Parm99//SolEFP method.17 For details about these EFP2//SolEFP calculations, see the Supporting Information. To simulate the IR spectra we used the second-order cumulant expansion method based on spectral density calculations.43 If the static distribution of the frequency shifts is approximately a Gaussian and time-dependent third-order cumulant terms are vanishingly small, the line shape and width are governed by the frequency-frequency autocorrelation function (FFCF) defined by M(t ) = ⟨δω(t ) δω(0)⟩

wall at short intermolecular separations, causes an ultrafast collisioninduced memory loss of the initial excitation frequency of the IR probe, which is manifested by the large negative cross-correlation, which also leads to a notably large motional narrowing. The same short-range interaction produces a very broad distribution of blueshifted frequencies, which contributes to significant line broadening. In aqueous solutions, due to H-bonding interaction with water molecules, the frequency shift distribution is much broader than that in CHCl3 by roughly 10 cm−1 (Figure 6a and d). In summary, our EFP2//SolEFP simulations for the first time clarify the line broadening process in terms of separate contributions from distinctively different solute−solvent interactions. The SolEFP calculation for a single solute molecule combined with EFP2 MD simulations is clearly less expensive than full ab initio MD simulations of a whole solution or protein system. Since SolEFP is one of the most systematic and rigorous approaches reported so far, it allows one to study the underlying nature of complicated vibrational solvatochromism of large molecular systems. Nonetheless, since our EFP2//SolEFP approach is based on a perturbation theory in vibrational anharmonicity and solute−solvent interaction potential, a further improvement of the theory by taking into account short-distance charge penetration effect and strong structural distortion effect beyond the first-order (in vibrational anharmonicity) perturbation theory is needed to make it a versatile and transferable model for describing vibrational solvatochromism of any arbitrary IR probe in highly crowded environments such as those incorporated in core regions of proteins or at interfacial regions of protein−protein contacts.

(8)

Since the vibrational lifetime of the CN stretch mode in thiocyanate-derivatized compound in water and chloroform is typically longer than 40 ps,44 we could safely ignore any vibrational lifetime-broadening. The resulting frequency distributions, FFCFs, and absorption line shapes superimposed onto the experimental spectra are shown in Figure 6. Although the cumulant approximation cannot reproduce the slightly asymmetric line shape of the CN stretch band in CHCl3, the simulated spectrum is in excellent agreement with experimental result. It should be emphasized that the spectra simulated with only the Coulomb term (red lines in Figure 6c and f) are too narrow, especially in water. Indeed, the exchange-repulsion, induction, and dispersion effects all contribute significantly to the line-broadening. However, to our surprise, the decay of the total FFCF is largely dictated by the exchange-repulsion term (Figure 6b and e). Clearly, FFCFs computed by using ΔωSolEFP (black solid line) can be well reproduced by only considering the autocorrelation obtained from ΔωEx−Rep(t) (blue dashed line) as well as its cross-correlation with the sum of the Coulombic, induction, and dispersion frequency shifts (blue dash-dotted line). This can be understood by noting that the exchange-repulsion interaction, due to its strength and stiffness in the repulsive potential



VIBRATIONAL SOLVATOCHROMISM: BEYOND IN VITRO IR ABSORPTION While polar IR probe groups like nitriles and azides have seen much recent use in vitro, there are other functional groups with 973

DOI: 10.1021/acs.accounts.7b00002 Acc. Chem. Res. 2017, 50, 968−976

Article

Accounts of Chemical Research similar frequencies in the “clear” spectral region that are beginning to see use as imaging probes for sectioned tissue or even in vivo samples.45 Since Raman imaging is more widespread and has finer spatial resolution than IR imaging, understanding the solvatochromism of these strong and unique Raman signals is of great current importance. While carbon-deuterium vibrations can be viewed (with some instrumental difficulty) in both IR absorption and Raman scattering spectra, S−H vibrations (in isolated cases where free thiols are a stable feature of natural systems) and alkyne CC vibrations provide clear signals that are more reliably strong in Raman spectra. The solvatochromism of these groups is less well-established than that of nitriles and azides, and the physical basis for that solvatochromism is also not clear. Figure 7 shows three solvent-dependent spontaneous Raman spectra for the amino acid homopropargyl glycine, which can be

is especially important in designing and interpreting stimulated Raman imaging experiments using Raman probes. The promise of SolEFP is that it directly reveals the physical basis for the solvatochromism of new vibrational probe groups in environments that might be arbitrarily complex. For Raman imaging, this means that probe groups might be used to report on both their spatial locations in a complicated system like a cell or tissue and their immediate molecular environments via solvatochromic changes in their frequencies.



CONCLUDING REMARKS Vibrational spectroscopy can be an effective and useful tool to probe molecular conformations in condensed phases, particularly when small probes are site-specifically incorporated into biomolecular systems. In this Account, we addressed the interpretation of linear and nonlinear IR spectroscopic data, and the possibility of interpreting Raman data as well, in terms of specific intermolecular interactions between the probe and its environment. Such direct connections between spectroscopic data and the locally interacting environment around the probe group is the key for unveiling the unknown topology around the IR probe as well as the equilibrium or dynamic conformation of a given biomolecule in condensed phase, when average frequency shifts and fluctuation dynamics of the probe-incorporated systems change as a result of different solvents or structural changes due to target binding or other biologically relevant processes.



ASSOCIATED CONTENT

* Supporting Information S

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.accounts.7b00002. Details of EFP2//SolEFP simulation method, table of frequency shift values of the coarse-grained SolEFP calculations, and supporting references (PDF)

Figure 7. Spontaneous Raman spectra in the CC stretching region for homopropargyl glycine (structure shown). The corresponding Raman spectra in deionized water, hexane, and dimethyl sulfoxide (DMSO) were collected using 1 s exposure on a home-built instrument with 80 mW 532 nm excitation.



AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. ORCID

recombinantly incorporated into proteins in place of methionine but also here serves as a representative for any aliphatic terminal alkyne probe group. There are clear solvent-dependent differences in both the CC stretching frequency and the line shape, and these changes are quantitatively different from solvatochromism of the CN stretch of nitriles or the symmetric or asymmetric NNN stretch of azides. While some recent spontaneous Raman imaging experiments featured alkyne labels attached to specific biomolecules, stimulated Raman scattering microscopy is a more sensitive Raman imaging technique, where the part of the two-beam experiment that selects the stimulated Raman shifts can be either a ps narrow-band light source46 or a fs broad-band source. The ps narrow-band stimulated Raman scheme selects a very narrow range of Raman shifts to enhance and make up the imaging signal. The data in Figures 1b, 5, and 7 indicate that both nitriles and alkynes display a wide range of solvatochromic frequencies. This large range of possible frequencies means that probes located in specific molecular environments might either be selectively enhanced or missed altogether by the particular choice of Raman shift in a narrowband or a frequency-resolved stimulated Raman experiment. So, understanding vibrational solvatochromic frequency shifts as well as changes in Raman scattering lineshapes, in particular,

Lauren J. Webb: 0000-0001-9999-5500 Minhaeng Cho: 0000-0003-1618-1056 Author Contributions

The manuscript was written through contributions of all authors. Funding

M.C. was supported by IBS-R023-D1. B.B. thanks the Wrocław Centre for Networking and Supercomputing (WCSS) at Wrocław University of Science and Technology in Poland for support. L.J. W. is supported by the Welch Foundation (F-1722). Notes

The authors declare no competing financial interest. Biographies Bartosz Błasiak was born in 1988 and raised in Poland. In 2012, he received his M.S. in chemistry from Wrocław University of Technology at Wrocław, Poland, and in 2016 his Ph.D. in Physical Chemistry from Korea University, Seoul, Korea. He is interested in developing quantum chemistry and theoretical methods for molecular spectroscopy. Casey H. Londergan is an associate professor of chemistry at Haverford College. He obtained a B.A. in chemistry from Williams College (1997) 974

DOI: 10.1021/acs.accounts.7b00002 Acc. Chem. Res. 2017, 50, 968−976

Article

Accounts of Chemical Research

(14) Błasiak, B.; Cho, M. Vibrational solvatochromism. II. A firstprinciple theory of solvation-induced vibrational frequency shift based on effective fragment potential method. J. Chem. Phys. 2014, 140, 164107. (15) Błasiak, B.; Cho, M. Vibrational solvatochromism. III. Rigorous treatment of the dispersion interaction contribution. J. Chem. Phys. 2015, 143, 164111. (16) Błasiak, B.; Lee, H.; Cho, M. Vibrational solvatochromism: Towards systematic approach to modeling solvation phenomena. J. Chem. Phys. 2013, 139, 044111. (17) Błasiak, B.; Ritchie, A. W.; Webb, L. J.; Cho, M. Vibrational solvatochromism of nitrile infrared probes: beyond the vibrational Stark dipole approach. Phys. Chem. Chem. Phys. 2016, 18, 18094−18111. (18) Rey, R.; Hynes, J. T. Vibrational phase and energy relaxation of CN− in water. J. Chem. Phys. 1998, 108, 142−153. (19) Laurence, C.; Legros, J.; Chantzis, A.; Planchat, A.; Jacquemin, D. A Database of Dispersion-Induction DI, Electrostatic ES, and Hydrogen Bonding α1 and β1 Solvent Parameters and Some Applications to the Multiparameter Correlation Analysis of Solvent Effects. J. Phys. Chem. B 2015, 119, 3174−3184. (20) Zhang, W.; Markiewicz, B. N.; Doerksen, R. S.; Smith, A. B., III; Gai, F. CN stretching vibration of 5-cyanotryptophan as an infrared probe of protein local environment: what determines its frequency? Phys. Chem. Chem. Phys. 2016, 18, 7027−7034. (21) la Cour Jansen, T.; Knoester, J. A transferable electrostatic map for solvation effects on amide I vibrations and its application to linear and two-dimensional spectroscopy. J. Chem. Phys. 2006, 124, 044502. (22) Maienschein-Cline, M. G.; Londergan, C. H. The CN Stretching Band of Aliphatic Thiocyanate is Sensitive to Solvent Dynamics and Specific Solvation. J. Phys. Chem. A 2007, 111, 10020−10025. (23) Wolfshorndl, M. P.; Baskin, R.; Dhawan, I.; Londergan, C. H. Covalently Bound Azido Groups Are Very Specific Water Sensors, Even in Hydrogen-Bonding Environments. J. Phys. Chem. B 2012, 116, 1172− 1179. (24) Maj, M.; Ahn, C.; Błasiak, B.; Kwak, K.; Han, H.; Cho, M. Isonitrile as an Ultrasensitive Infrared Reporter of Hydrogen-Bonding Structure and Dynamics. J. Phys. Chem. B 2016, 120, 10167−10180. (25) Bagchi, S.; Fried, S. D.; Boxer, S. G. A Solvatochromic Model Calibrates Nitriles’ Vibrational Frequencies to Electrostatic Fields. J. Am. Chem. Soc. 2012, 134, 10373−10376. (26) Hayashi, T.; Zhuang, W.; Mukamel, S. Electrostatic DFT Map for the Complete Vibrational Amide Band of NMA. J. Phys. Chem. A 2005, 109, 9747−9759. (27) Torii, H. Amide I Vibrational Properties Affected by Hydrogen Bonding Out-of-Plane of the Peptide Group. J. Phys. Chem. Lett. 2015, 6, 727−733. (28) Cho, M. Vibrational solvatochromism and electrochromism: Coarse-grained models and their relationships. J. Chem. Phys. 2009, 130, 094505. (29) Choi, J.-H.; Oh, K.-I.; Lee, H.; Lee, C.; Cho, M. Nitrile and thiocyanate IR probes: Quantum chemistry calculation studies and multivariate least-square fitting analysis. J. Chem. Phys. 2008, 128, 134506. (30) Andrews, S. S.; Boxer, S. G. Vibrational Stark Effects of Nitriles I. Methods and Experimental Results. J. Phys. Chem. A 2000, 104, 11853− 11863. (31) Jansen, T. L. C. Linear Absorption and Two-Dimensional Infrared Spectra of N-Methylacetamide in Chloroform Revisited: Polarizability and Multipole Effects. J. Phys. Chem. B 2014, 118, 8162−8169. (32) Małolepsza, E.; Straub, J. E. Empirical Maps For The Calculation of Amide I Vibrational Spectra of Proteins From Classical Molecular Dynamics Simulations. J. Phys. Chem. B 2014, 118, 7848−7855. (33) Buckingham, A. D. Solvent effects in vibrational spectroscopy. Trans. Faraday Soc. 1960, 56, 753−760. (34) Gordon, M. S.; Smith, Q. A.; Xu, P.; Slipchenko, L. V. Accurate First Principles Model Potentials for Intermolecular Interactions. Annu. Rev. Phys. Chem. 2013, 64, 553−578. (35) Buckingham, A. D.; Fowler, P. W. A model for the geometries of Van der Waals complexes. Can. J. Chem. 1985, 63, 2018−2025.

and a Ph.D. in chemistry from the University of CaliforniaSan Diego (2003) under the supervision of Dr. Clifford P. Kubiak. He was a postdoctoral fellow at the University of Pennsylvania with the late Dr. Robin Hochstrasser from 2003 to 2006 and has been at Haverford College since 2006. His main research interest is the development and application of new vibrational probe groups for biophysical problems, including protein−protein and protein−membrane binding and the conformational distributions of dynamic and disordered proteins. Lauren J. Webb is an associate professor of chemistry at The University of Texas at Austin. She obtained her A.B. in chemistry (music minor) from Bowdoin College in 2000. She earned her Ph.D. in chemistry in 2005 from the California Institute of Technology in the laboratory of Dr. Nathan Lewis, then worked as a postdoctoral fellow with Dr. Steven Boxer in the department of chemistry at Stanford University until 2008, when she moved to UT-Austin. Her research interests are centered on understanding and manipulating the mechanisms of interaction, organization, and self-assembly of biological macromolecules in both natural and artificial environments. Minhaeng Cho received a Ph.D. from University of Chicago in 1993 under the direction of Prof. Graham R. Fleming. After 2 years of postdoctoral training at MIT in Prof. Robert J. Silbey’s group, he has been on the faculty of Korea University since 1996 and has been the director of IBS center for molecular spectroscopy and dynamics since 2015.



REFERENCES

(1) Kim, H.; Cho, M. Infrared Probes for Studying the Structure and Dynamics of Biomolecules. Chem. Rev. 2013, 113, 5817−5847. (2) Ma, J.; Pazos, I. M.; Zhang, W.; Culik, R. M.; Gai, F. Site-Specific Infrared Probes of Proteins. Annu. Rev. Phys. Chem. 2015, 66, 357−377. (3) Waegele, M. M.; Culik, R. M.; Gai, F. Site-Specific Spectroscopic Reporters of the Local Electric Field, Hydration, Structure, and Dynamics of Biomolecules. J. Phys. Chem. Lett. 2011, 2, 2598−2609. (4) Fried, S. D.; Boxer, S. G. Measuring Electric Fields and Noncovalent Interactions Using the Vibrational Stark Effect. Acc. Chem. Res. 2015, 48, 998−1006. (5) Cho, M. Two-Dimensional Optical Spectroscopy; CRC Press: Boca Raton, FL, 2009. (6) Alfieri, K. N.; Vienneau, A. R.; Londergan, C. H. Using Infrared Spectroscopy of Cyanylated Cysteine To Map the Membrane Binding Structure and Orientation of the Hybrid Antimicrobial Peptide CM15. Biochemistry 2011, 50, 11097−11108. (7) Johnson, M. N. R.; Londergan, C. H.; Charkoudian, L. K. Probing the Phosphopantetheine Arm Conformations of Acyl Carrier Proteins Using Vibrational Spectroscopy. J. Am. Chem. Soc. 2014, 136, 11240− 11243. (8) McMahon, H. A.; Alfieri, K. N.; Clark, K. A. A.; Londergan, C. H. Cyanylated Cysteine: A Covalently Attached Vibrational Probe of Protein−Lipid Contacts. J. Phys. Chem. Lett. 2010, 1, 850−855. (9) Londergan, C. H.; Baskin, R.; Bischak, C. G.; Hoffman, K. W.; Snead, D. M.; Reynoso, C. Dynamic Asymmetry and the Role of the Conserved Active-Site Thiol in Rabbit Muscle Creatine Kinase. Biochemistry 2015, 54, 83−95. (10) Morales, C. M.; Thompson, W. H. Molecular-Level Mechanisms of Vibrational Frequency Shifts in a Polar Liquid. J. Phys. Chem. B 2011, 115, 7597−7605. (11) Reimers, J. R.; Hall, L. E. The Solvation of Acetonitrile. J. Am. Chem. Soc. 1999, 121, 3730−3744. (12) Ben-Amotz, D.; Lee, M. R.; Cho, S. Y.; List, D. J. Solvent and pressure-induced perturbations of the vibrational potential surface of acetonitrile. J. Chem. Phys. 1992, 96, 8781−8792. (13) Lee, H.; Choi, J.-H.; Cho, M. Vibrational solvatochromism and electrochromism. II. Multipole analysis. J. Chem. Phys. 2012, 137, 114307. 975

DOI: 10.1021/acs.accounts.7b00002 Acc. Chem. Res. 2017, 50, 968−976

Article

Accounts of Chemical Research (36) Slocum, J. D.; Webb, L. J. Nitrile Probes of Electric Field Agree with Independently Measured Fields in Green Fluorescent Protein Even in the Presence of Hydrogen Bonding. J. Am. Chem. Soc. 2016, 138, 6561−6570. (37) Schmidt, J. R.; Corcelli, S. A.; Skinner, J. L. Pronounced nonCondon effects in the ultrafast infrared spectroscopy of water. J. Chem. Phys. 2005, 123, 044513. (38) Asbury, J. B.; Steinel, T.; Stromberg, C.; Corcelli, S. A.; Lawrence, C. P.; Skinner, J. L.; Fayer, M. D. Water Dynamics: Vibrational Echo Correlation Spectroscopy and Comparison to Molecular Dynamics Simulations. J. Phys. Chem. A 2004, 108, 1107−1119. (39) Kwac, K.; Lee, H.; Cho, M. Non-Gaussian statistics of amide I mode frequency fluctuation of N-methylacetamide in methanol solution: Linear and nonlinear vibrational spectra. J. Chem. Phys. 2004, 120, 1477−1490. (40) Lindquist, B. A.; Furse, K. E.; Corcelli, S. A. Nitrile groups as vibrational probes of biomolecular structure and dynamics: an overview. Phys. Chem. Chem. Phys. 2009, 11, 8119−8132. (41) Layfield, J. P.; Hammes-Schiffer, S. Calculation of Vibrational Shifts of Nitrile Probes in the Active Site of Ketosteroid Isomerase upon Ligand Binding. J. Am. Chem. Soc. 2013, 135, 717−725. (42) Kaliman, I. A.; Slipchenko, L. V. LIBEFP: A new parallel implementation of the effective fragment potential method as a portable software library. J. Comput. Chem. 2013, 34, 2284−2292. (43) Oh, K.-I.; Choi, J.-H.; Lee, J.-H.; Han, J.-B.; Lee, H.; Cho, M. Nitrile and thiocyanate IR probes: Molecular dynamics simulation studies. J. Chem. Phys. 2008, 128, 154504. (44) Park, K.-H.; Jeon, J.; Park, Y.; Lee, S.; Kwon, H.-J.; Joo, C.; Park, S.; Han, H.; Cho, M. Infrared Probes Based on Nitrile-Derivatized Prolines: Thermal Insulation Effect and Enhanced Dynamic Range. J. Phys. Chem. Lett. 2013, 4, 2105−2110. (45) Palonpon, A. F.; Sodeoka, M.; Fujita, K. Molecular imaging of live cells by Raman microscopy. Curr. Opin. Chem. Biol. 2013, 17, 708−715. (46) Wei, L.; Hu, F.; Shen, Y.; Chen, Z.; Yu, Y.; Lin, C.-C.; Wang, M. C.; Min, W. Live-cell imaging of alkyne-tagged small biomolecules by stimulated Raman scattering. Nat. Methods 2014, 11, 410−412.

976

DOI: 10.1021/acs.accounts.7b00002 Acc. Chem. Res. 2017, 50, 968−976