Biophysical Journal Volume 108 April 2015 1577–1579
1577
New and Notable Catching Protein Structural Dynamics by Two-Dimensional Infrared Spectroscopy Chungwen Liang1,* 1 Department of Chemistry and Biochemistry, University of California, San Diego, San Diego, California
Unveiling the principles of protein conformational changes is one of the most challenging research topics over the past few decades. Because proteins’ three-dimensional structures are highly flexible and sensitive to the environment, a wide range of motions with timescales from femtosecond to millisecond is responsible for the overall conformational changes. Characterizing functional states of proteins and determining transition rates between them is necessary to understand how the environment initiates protein conformational changes. Therefore, an advanced approach with both high spatial and temporal resolution is indispensable for probing protein structures at the atomic level and dynamics on a short timescale. The report by Baiz and Tokmakoff (1) in this issue of Biophysical Journal is an excellent demonstration of the combination of advanced experimental and theoretical approaches to characterizing protein conformational heterogeneity on subpicosecond timescales. Two-dimensional infrared (2D IR) spectroscopy (2) (a schematic illustration of which is shown in Fig. 1) is one of the most powerful approaches to studying biological systems. It makes use of a series of femtosecond laser pulses with carefully selected time delays between them to probe system structure and dynamics. This information can then be extracted by
Submitted January 2, 2015, and accepted for publication March 4, 2015. *Correspondence: [email protected] Editor: H. Jane Dyson. Ó 2015 by the Biophysical Society 0006-3495/15/04/1577/3 $2.00
monitoring the spectral changes as a function of the time delays. Because the time resolution of 2D IR is measured on subpicosecond timescales, the quickest conformational changes of biomolecules appear static, and thus can be probed. The optically active amide I vibration (C¼O stretching), which is naturally present in peptides and proteins, is one of the most popular excitations probed by 2D IR spectroscopy. The 2D IR spectrum carries information about the magnitude and timescale of the electric field fluctuations in the environment of amide I groups. Due to the complexity of the 2D IR spectrum, it is hard to directly translate the spectral information into structural and dynamical information at the atomic level. Computational spectroscopy is a state-of-the-art spectrum modeling approach, which can be used to interpret experimental results. For modeling the 2D IR spectrum of protein, a series of molecular dynamics (MD) simulations (3,4) are first performed to obtain an ensemble of protein structures. Then, a quantum Hamiltonian, which describes the amide I vibrational degree of freedom, is constructed from molecular configurations based on electronic structure methods. Using time-dependent perturbation theory, the applied fields (the laser pulses in the experiment) can be treated as perturbations that interact with the transition dipole of amide I groups. Afterwards, the simulated 2D IR spectrum can be obtained by Fourier transformation of the nonlinear optical response function. Previous studies have successfully demonstrated the capability of 2D IR spectroscopy to study protein structure and dynamics since the late 1990s. Several small peptides in solution were first investigated by 2D IR experimentally (2,5) and theoretically (3,4). Applications were then extended to more complex systems, such as soluble polypeptides (6,7), membrane channel proteins (8,9), and protein aggregates (10).
The study by Baiz and Tokmakoff (1) illustrates how 2D IR spectroscopy probes the conformational heterogeneity of the 39-residue protein L9. By isotopic labeling of the amide I groups of several selected residues, the local structural signatures can be probed directly, due to the fact that the vibrational frequencies of the isotopelabeled bonds are red-shifted, which results in isolated peaks. As mentioned above, the amide I vibrational frequency is sensitive to the magnitude and timescale of the electric field fluctuation in the environment. This fluctuation is governed by the degree of solvent exposure (water hydrogenbonding to amide I groups), the geometry of the protein (hydrogen-bond formation between protein backbone N-H and amide I groups), and the torsion angles between neighboring amide I units. While the majority of the absorptions remain difficult to analyze by linear FTIR due to background absorptions, the essential information is available from diagonal slices of the 2D IR spectra. Therefore, the degree of hydration and local conformation of each isotope-labeled bond can be revealed by analyzing the peak position and linewidth of the 2D IR spectrum. In particular, the authors found that V9 in the b-turn region is more flexible and solvent-exposed than V3 in one of the b-strands, which can be deduced from the red-shifted and broader peak of V9 in both experimental and simulated spectra. However, quantitative agreement is not always found between experiments and simulations. Analyzing the experimental V9 and V9G13 spectra showed that the b-turn region is more flexible and solvent-exposed than the prediction of MD simulations. For the highly solvent-exposed residues G16 and G24, broad peaks with low amplitudes appear in both experimental and simulated spectra, which is attributed to water hydrogen-bonding to the amide I
http://dx.doi.org/10.1016/j.bpj.2015.03.007
1578
Liang
FIGURE 1 A schematic illustration: a 2D IR spectrum of protein can be measured by time-resolved spectroscopic techniques, or be modeled by computational approaches. This allows a direct comparison between experiment and simulation. To see this figure in color, go online.
groups of these residues. The advantage of simulation is that one can quantify the degree of hydration by analyzing the average total number of protein-water and protein-protein hydrogen bonds. Then, the degree of hydration can be correlated with the spectral signature. Utilizing 2D IR spectroscopy to study protein conformation allows a direct comparison between experiment and simulation; however, several challenges must still be overcome to achieve quantitative agreement. From a theoretical point of view, sufficient sampling of protein conformational space, accurate prediction of the population of protein functional states, and advanced models that fully characterize the spectroscopic properties of protein, are the key factors determining the quality of simulation results. Using massively parallel supercomputers, atomistic simulations of small proteins on the millisecond timescale are possible. But, in order Biophysical Journal 108(7) 1577–1579
to construct a reliable Markov state model of a protein and to predict accurate transition rates between each state, the required samplings are often beyond the currently accessible simulation timescales. Modern classical MD force fields are capable of predicting the geometry of protein structures by empirical parameterization from a wide range of experimental data. However, the accurate estimation of protein functional state populations is still a work in progress. The spectroscopic properties of biomolecules are often missing in the description of classical MD force fields. Establishing advanced quantum mechanical models to fully characterize the vibrational frequency/coupling and transition dipole/polarizability of amide I groups is the basis of spectrum modeling. These approaches generally treat amide I groups as quantum oscillators in a classical environment, so that the quantum degree of freedom of the solvent or of
other protein functional groups is missing. Future theoretical development for treating vibrational couplings between amide I and other important vibrational modes, such as waterbending or protein amide II modes, can be used to interpret the dual-frequency 2D IR measurements that study the coupling and energy transfer between different vibrational modes. This will give insight into how water molecules interact with different protein functional groups, changing the local and overall protein conformation. In addition, the nonequilibrium properties of protein structure are crucial for understanding how the environment initiates structural changes at very early stages. These properties can be probed by transient 2D IR spectroscopy, such as temperature-, pH-, and ion-concentrationjumps experiments. Understanding these molecular processes in detail will shine light on the mystery of protein folding.
2D IR Spectra of Proteins
ACKNOWLEDGMENTS The author thanks Prof. Dr. T.L.C. Jansen, Dr. S. Roy, Dr. Z. Terranova, and D. Wilkins for useful discussions.
REFERENCES 1. Baiz, C. R., and A. Tokmakoff. 2015. Structural characterization of folded protein ensembles: isotope-edited 2D IR spectroscopy and spectral simulations based on a Markov state model. Biophys. J. 108:1747–1757. 2. Hamm, P., M. H. Lim, and R. M. Hochstrasser. 1998. Structure of the amide I band of peptides measured by femtosecond nonlinear-infrared spectroscopy. J. Phys. Chem. B. 102:6123–6138.
1579 3. Scheurer, C., A. Piryatinski, and S. Mukamel. 2001. Signatures of b-peptide unfolding in two-dimensional vibrational echo spectroscopy: a simulation study. J. Am. Chem. Soc. 123:3114–3124. 4. la Cour Jansen, T., W. Zhuang, and S. Mukamel. 2004. Stochastic Liouville equation simulation of multidimensional vibrational line shapes of trialanine. J. Chem. Phys. 121:10577–10598. 5. Zanni, M. T., J. Stenger, ., R. M. Hochstrasser. 2001. Solvent dependent conformational dynamics of dipeptides studied with two-dimensional infrared spectroscopy. Biophys. J. 80:8A–9A, Abstract. 6. Lessing, J., S. Roy, ., A. Tokmakoff. 2012. Identifying residual structure in intrinsically disordered systems: a 2D IR spectroscopic study of the GVGXPGVG peptide. J. Am. Chem. Soc. 134:5032–5035.
7. Meuzelaar, H., K. A. Marino, ., S. Woutersen. 2013. Folding dynamics of the Trpcage miniprotein: evidence for a native-like intermediate from combined time-resolved vibrational spectroscopy and molecular dynamics simulations. J. Phys. Chem. B. 117:11490–11501. 8. Manor, J., P. Mukherjee, ., I. T. Arkin. 2009. Gating mechanism of the influenza A M2 channel revealed by 1D and 2D IR spectroscopies. Structure. 17:247–254. 9. Liang, C., M. Louhivuori, ., J. Knoester. 2013. Vibrational spectra of a mechanosensitive channel. J. Phys. Chem. Lett. 4: 448–452. 10. Buchanan, L. E., E. B. Dunkelberger, ., M. T. Zanni. 2013. Mechanism of IAPP amyloid fibril formation involves an intermediate with a transient b-sheet. Proc. Natl. Acad. Sci. USA. 110:19285–19290.
Biophysical Journal 108(7) 1577–1579
1577
New and Notable Catching Protein Structural Dynamics by Two-Dimensional Infrared Spectroscopy Chungwen Liang1,* 1 Department of Chemistry and Biochemistry, University of California, San Diego, San Diego, California
Unveiling the principles of protein conformational changes is one of the most challenging research topics over the past few decades. Because proteins’ three-dimensional structures are highly flexible and sensitive to the environment, a wide range of motions with timescales from femtosecond to millisecond is responsible for the overall conformational changes. Characterizing functional states of proteins and determining transition rates between them is necessary to understand how the environment initiates protein conformational changes. Therefore, an advanced approach with both high spatial and temporal resolution is indispensable for probing protein structures at the atomic level and dynamics on a short timescale. The report by Baiz and Tokmakoff (1) in this issue of Biophysical Journal is an excellent demonstration of the combination of advanced experimental and theoretical approaches to characterizing protein conformational heterogeneity on subpicosecond timescales. Two-dimensional infrared (2D IR) spectroscopy (2) (a schematic illustration of which is shown in Fig. 1) is one of the most powerful approaches to studying biological systems. It makes use of a series of femtosecond laser pulses with carefully selected time delays between them to probe system structure and dynamics. This information can then be extracted by
Submitted January 2, 2015, and accepted for publication March 4, 2015. *Correspondence: [email protected] Editor: H. Jane Dyson. Ó 2015 by the Biophysical Society 0006-3495/15/04/1577/3 $2.00
monitoring the spectral changes as a function of the time delays. Because the time resolution of 2D IR is measured on subpicosecond timescales, the quickest conformational changes of biomolecules appear static, and thus can be probed. The optically active amide I vibration (C¼O stretching), which is naturally present in peptides and proteins, is one of the most popular excitations probed by 2D IR spectroscopy. The 2D IR spectrum carries information about the magnitude and timescale of the electric field fluctuations in the environment of amide I groups. Due to the complexity of the 2D IR spectrum, it is hard to directly translate the spectral information into structural and dynamical information at the atomic level. Computational spectroscopy is a state-of-the-art spectrum modeling approach, which can be used to interpret experimental results. For modeling the 2D IR spectrum of protein, a series of molecular dynamics (MD) simulations (3,4) are first performed to obtain an ensemble of protein structures. Then, a quantum Hamiltonian, which describes the amide I vibrational degree of freedom, is constructed from molecular configurations based on electronic structure methods. Using time-dependent perturbation theory, the applied fields (the laser pulses in the experiment) can be treated as perturbations that interact with the transition dipole of amide I groups. Afterwards, the simulated 2D IR spectrum can be obtained by Fourier transformation of the nonlinear optical response function. Previous studies have successfully demonstrated the capability of 2D IR spectroscopy to study protein structure and dynamics since the late 1990s. Several small peptides in solution were first investigated by 2D IR experimentally (2,5) and theoretically (3,4). Applications were then extended to more complex systems, such as soluble polypeptides (6,7), membrane channel proteins (8,9), and protein aggregates (10).
The study by Baiz and Tokmakoff (1) illustrates how 2D IR spectroscopy probes the conformational heterogeneity of the 39-residue protein L9. By isotopic labeling of the amide I groups of several selected residues, the local structural signatures can be probed directly, due to the fact that the vibrational frequencies of the isotopelabeled bonds are red-shifted, which results in isolated peaks. As mentioned above, the amide I vibrational frequency is sensitive to the magnitude and timescale of the electric field fluctuation in the environment. This fluctuation is governed by the degree of solvent exposure (water hydrogenbonding to amide I groups), the geometry of the protein (hydrogen-bond formation between protein backbone N-H and amide I groups), and the torsion angles between neighboring amide I units. While the majority of the absorptions remain difficult to analyze by linear FTIR due to background absorptions, the essential information is available from diagonal slices of the 2D IR spectra. Therefore, the degree of hydration and local conformation of each isotope-labeled bond can be revealed by analyzing the peak position and linewidth of the 2D IR spectrum. In particular, the authors found that V9 in the b-turn region is more flexible and solvent-exposed than V3 in one of the b-strands, which can be deduced from the red-shifted and broader peak of V9 in both experimental and simulated spectra. However, quantitative agreement is not always found between experiments and simulations. Analyzing the experimental V9 and V9G13 spectra showed that the b-turn region is more flexible and solvent-exposed than the prediction of MD simulations. For the highly solvent-exposed residues G16 and G24, broad peaks with low amplitudes appear in both experimental and simulated spectra, which is attributed to water hydrogen-bonding to the amide I
http://dx.doi.org/10.1016/j.bpj.2015.03.007
1578
Liang
FIGURE 1 A schematic illustration: a 2D IR spectrum of protein can be measured by time-resolved spectroscopic techniques, or be modeled by computational approaches. This allows a direct comparison between experiment and simulation. To see this figure in color, go online.
groups of these residues. The advantage of simulation is that one can quantify the degree of hydration by analyzing the average total number of protein-water and protein-protein hydrogen bonds. Then, the degree of hydration can be correlated with the spectral signature. Utilizing 2D IR spectroscopy to study protein conformation allows a direct comparison between experiment and simulation; however, several challenges must still be overcome to achieve quantitative agreement. From a theoretical point of view, sufficient sampling of protein conformational space, accurate prediction of the population of protein functional states, and advanced models that fully characterize the spectroscopic properties of protein, are the key factors determining the quality of simulation results. Using massively parallel supercomputers, atomistic simulations of small proteins on the millisecond timescale are possible. But, in order Biophysical Journal 108(7) 1577–1579
to construct a reliable Markov state model of a protein and to predict accurate transition rates between each state, the required samplings are often beyond the currently accessible simulation timescales. Modern classical MD force fields are capable of predicting the geometry of protein structures by empirical parameterization from a wide range of experimental data. However, the accurate estimation of protein functional state populations is still a work in progress. The spectroscopic properties of biomolecules are often missing in the description of classical MD force fields. Establishing advanced quantum mechanical models to fully characterize the vibrational frequency/coupling and transition dipole/polarizability of amide I groups is the basis of spectrum modeling. These approaches generally treat amide I groups as quantum oscillators in a classical environment, so that the quantum degree of freedom of the solvent or of
other protein functional groups is missing. Future theoretical development for treating vibrational couplings between amide I and other important vibrational modes, such as waterbending or protein amide II modes, can be used to interpret the dual-frequency 2D IR measurements that study the coupling and energy transfer between different vibrational modes. This will give insight into how water molecules interact with different protein functional groups, changing the local and overall protein conformation. In addition, the nonequilibrium properties of protein structure are crucial for understanding how the environment initiates structural changes at very early stages. These properties can be probed by transient 2D IR spectroscopy, such as temperature-, pH-, and ion-concentrationjumps experiments. Understanding these molecular processes in detail will shine light on the mystery of protein folding.
2D IR Spectra of Proteins
ACKNOWLEDGMENTS The author thanks Prof. Dr. T.L.C. Jansen, Dr. S. Roy, Dr. Z. Terranova, and D. Wilkins for useful discussions.
REFERENCES 1. Baiz, C. R., and A. Tokmakoff. 2015. Structural characterization of folded protein ensembles: isotope-edited 2D IR spectroscopy and spectral simulations based on a Markov state model. Biophys. J. 108:1747–1757. 2. Hamm, P., M. H. Lim, and R. M. Hochstrasser. 1998. Structure of the amide I band of peptides measured by femtosecond nonlinear-infrared spectroscopy. J. Phys. Chem. B. 102:6123–6138.
1579 3. Scheurer, C., A. Piryatinski, and S. Mukamel. 2001. Signatures of b-peptide unfolding in two-dimensional vibrational echo spectroscopy: a simulation study. J. Am. Chem. Soc. 123:3114–3124. 4. la Cour Jansen, T., W. Zhuang, and S. Mukamel. 2004. Stochastic Liouville equation simulation of multidimensional vibrational line shapes of trialanine. J. Chem. Phys. 121:10577–10598. 5. Zanni, M. T., J. Stenger, ., R. M. Hochstrasser. 2001. Solvent dependent conformational dynamics of dipeptides studied with two-dimensional infrared spectroscopy. Biophys. J. 80:8A–9A, Abstract. 6. Lessing, J., S. Roy, ., A. Tokmakoff. 2012. Identifying residual structure in intrinsically disordered systems: a 2D IR spectroscopic study of the GVGXPGVG peptide. J. Am. Chem. Soc. 134:5032–5035.
7. Meuzelaar, H., K. A. Marino, ., S. Woutersen. 2013. Folding dynamics of the Trpcage miniprotein: evidence for a native-like intermediate from combined time-resolved vibrational spectroscopy and molecular dynamics simulations. J. Phys. Chem. B. 117:11490–11501. 8. Manor, J., P. Mukherjee, ., I. T. Arkin. 2009. Gating mechanism of the influenza A M2 channel revealed by 1D and 2D IR spectroscopies. Structure. 17:247–254. 9. Liang, C., M. Louhivuori, ., J. Knoester. 2013. Vibrational spectra of a mechanosensitive channel. J. Phys. Chem. Lett. 4: 448–452. 10. Buchanan, L. E., E. B. Dunkelberger, ., M. T. Zanni. 2013. Mechanism of IAPP amyloid fibril formation involves an intermediate with a transient b-sheet. Proc. Natl. Acad. Sci. USA. 110:19285–19290.
Biophysical Journal 108(7) 1577–1579
