SENSITIZATION BY LUMAZINE PROTEINS OF THE BIOLUMINESCENCE EMISSION FROM THE REACTION OF BACTERIAL LUCIFERASES

Abstract— Lumazine protein from Photobacterium phosphoreurn blue shifts the in vitro bioluminescence spectra in the reactions using each of the 4 main types of bacterial luciferases: P. phosphoreum, P. leiognathi, Vibrio harveyi and V. fischeri . For the reaction initiated with FMNH₂ and tetradecanal at 2°C, this "sensitizing" property of lumazine protein differs quantitatively between the luciferases. An interaction constant characterizing each type of luciferase may be derived from a reciprocal plot of the spectral shift against the lurnazine protein concentration. The weakest interaction constant is in the V. fischeri reaction, 180 μM. For the V. harveyi reaction the interaction is in the range 6–9 μ M , and for both Photobacterium reactions it is 2–3 μM. A concentration of only 0.6 μ M of lumazine protein is sufficient to cause an observable change in the Photobacterium bioluminescence spectra. For the V. harveyi case the interaction constant is near to the equilibrium K _d for the luciferase-lumazine protein complex, observed directly by Visser and Lee. Both constants are decreased markedly by increase in phosphate concentration so that it is concluded that, with V. harveyi luciferase, sensitization occurs within this protein-protein complex. For P. phosphoreum luciferase, however, the equilibrium complex is too weak to correspond to the sensitizing interaction and it is concluded that the rate-limiting process is a protein-protein bimolecular collision. As judged from their molecular weight around 20000, spectral properties, and ability to blue shift the bioluminescence spectra, lumazine proteins are identified in a second strain of P. phosphoreum and in P. leiognathi .     

Random isotopolog libraries for protein perturbation studies. 13C NMR studies on lumazine protein of Photobacterium leiognathi

[graph: see text] Lumazine proteins of luminescent bacteria are paralogs of riboflavin synthase which are devoid of catalytic activity but bind the riboflavin synthase substrate, 6,7-dimethyl-8-ribityllumazine, with high affinity and are believed to serve as optical transponders for bioluminescence emission. Lumazine protein of Photobacterium leiognathi was expressed in a recombinant Escherichia coli host and was reconstituted with mixtures (random libraries) of 13C-labeled isotopologs of 6,7-dimethyl-8-ribityllumazine or riboflavin that had been prepared by biotransformation of [U-(13)C6]-, [1-(13)C1]-, [2-(13)C1]-, and [3-(13)C1]glucose. 13C NMR analysis of the protein/ligand complexes afforded the assignments of the 13C NMR chemical shifts for all carbon atoms of the protein-bound ligands by isotopolog abundance editing. The carbon atoms of the ribityl groups of both ligands studied were shifted up to 6 ppm upon binding to the protein. Chemical shift modulation of the side chain and chromophore carbon atoms due to protein/ligand interaction is discussed on the basis of the sequence similarity between lumazine protein and riboflavin synthase.     

19F NMR ligand perturbation studies on 6,7-bis(trifluoromethyl)-8-ribityllumazine-7-hydrates and the lumazine synthase complex of Bacillus subtilis. Site-directed mutagenesis changes the mechanism and the stereoselectivity of the catalyzed haloform-type reaction

The riboflavin synthase/lumazine synthase complex of Bacillus subtilis catalyzes the last two steps in riboflavin biosynthesis. The protein comprises a capsid of 60 beta subunits with lumazine synthase activity and a core of three alpha subunits with riboflavin synthase activity. The beta subunits catalyze the formation of 6,7-dimethyl-8-ribityllumazine (3) from 5-amino-6-ribitylamino-2,4(1H,3H)-pyrimidinedione (1) and 3,4-dihydroxy-2-butanone 4-phosphate (2). Complexes of recombinant lumazine synthase (beta(60) capsids) with 6-trifluoromethyl-7-oxo-8-ribityllumazine (10) as well as 7S- or 7R-6,7-bistrifluoromethyl-8-ribityllumazine hydrate (11) were studied by (19)F NMR spectroscopy. Despite the large molecular weight of approximately 960 kDa of the protein, spectra with separated signals of free and bound ligand could be obtained. An unusually large shift difference of 8 ppm was observed between the 7-trifluoromethyl signals of free and bound ligand for epimer B of 11 and the enzyme. The signal is sensitive to the replacement of amino acid residues F22 and H88. Lumazine synthase catalyzes the elimination of the 7-trifluoromethyl group of R-diastereomer epimer A in a haloform-like reaction. The elimination reaction is also catalyzed by F22 mutants. The H88R mutant displays an opposite stereoselectivity for epimer B and a greatly enhanced reaction rate. From a model of the epimers in the active site of the protein, the main function of the side chain of F22 seems to be to keep the substrate ring in the correct position. H88 is in a position suited to act as proton acceptor in both the physiological as well as the haloform reaction. A different mechanism of the haloform-reaction is proposed in the case of the H88R mutant, initiated by hydrogen bonding of the 7-trifluorormethyl group and the guanidinium group of the arginine residue.     

O-Nucleoside, S-nucleoside, and N-nucleoside probes of lumazine synthase and riboflavin synthase

Lumazine synthase catalyzes the penultimate step in the biosynthesis of riboflavin, while riboflavin synthase catalyzes the last step. O-Nucleoside, S-nucleoside, and N-nucleoside analogues of hypothetical lumazine biosynthetic intermediates have been synthesized in order to obtain structure and mechanism probes of these two enzymes, as well as inhibitors of potential value as antibiotics. Methods were devised for the selective cleavage of benzyl protecting groups in the presence of other easily reduced functionality by controlled hydrogenolysis over Lindlar catalyst. The deprotection reaction was performed in the presence of other reactive functionality including nitro groups, alkenes, and halogens. The target compounds were tested as inhibitors of lumazine synthase and riboflavin synthase obtained from a variety of microorganisms. In general, the S-nucleosides and N-nucleosides were more potent than the corresponding O-nucleosides as lumazine synthase and riboflavin synthase inhibitors, while the C-nucleosides were the least potent. A series of molecular dynamics simulations followed by free energy calculations using the Poisson-Boltzmann/surface area (MM-PBSA) method were carried out in order to rationalize the results of ligand binding to lumazine synthase, and the results provide insight into the dynamics of ligand binding as well as the molecular forces stabilizing the intermediates in the enzyme-catalyzed reaction.     

Activities of the bimodal fluorescent protein produced by Photobacterium phosphoreum strain bmFP in the luciferase reaction in vitro

The activity of the bimodal fluorescent protein (bmFP) (lambda max, 488 and 517 nm) in the in vitro luciferase reaction has been studied. The bmFP that is produced by Photobacterium phosphoreum strain bmFP is a dimer of two homologous subunits binding four riboflavin 5'-phosphate (FMN)-myristate chromophores. The addition of bmFP to the luciferase reaction in the presence of the lumazine protein prevented the lumazine protein-induced blue shift in the emission band. The bmFP reduced electrochemically serves as a substrate in the luciferase reaction in the absence of added FMN, resulting in light emission with a single maximum at about 487 nm. The bmFP was also active in lieu of FMN in the NADH/FMN oxidoreductase (flavin reductase)-luciferase coupled bioluminescence reaction in the absence of added FMN. In the coupled reaction, bioluminescence with the isolated bmFP chromophore was weaker than that with the holo-bmFP. After bmFP was used in luciferase reactions initiated either chemically or electrochemically, it was still capable of emitting bimodal fluorescence.     

Properties of the bimodal fluorescent protein produced by Photobacterium phosphoreum

A fluorescent protein isolated from the deep-sea luminous bacterium Photobacterium phosphoreum strain bmFP has been purified, cloned and sequenced. The protein is 96.5% identical in amino acid sequence to FP390, the weakly fluorescent flavoprotein encoded by the luxF gene characteristic of Photobacterium species. Similar to FP390, bmFP is a dimer of two homologous subunits binding four FMN-myristate chromophores but has the distinctive feature of emitting a bimodal fluorescence with maxima at about 488 and 517 nm, hence the name bmFP. For both bands of this fluorescence, the excitation spectrum exhibits a peak at 336 nm, not corresponding to its flavin-like absorption spectrum. Heating of bmFP in urea resulted in a decrease in the intensity of the 488 nm band along with the appearance of a new fluorescence peaking at 423 nm, partially reversible upon the removal of the urea. Upon complete denaturation, either by heat or guanidium chloride at 65 degrees C, fluorescence characteristic of both free flavin and this 423 nm species appears. It is speculated that chromophores in different states of protonation, associated with a single protein, are responsible for the unusual spectral properties of bmFP.     

Processing multimode binding situations in simulation-based prediction of ligand-macromolecule affinities

The linear response (LR) approximation and similar approaches belong to practical methods for estimation of ligand-receptor binding affinities. The approaches correlate experimental binding affinities with the changes upon binding of the ligand electrostatic and van der Waals energies and of solvation characteristics. These attributes are expressed as ensemble averages that are obtained by conformational sampling of the protein-ligand complex and of the free ligand by molecular dynamics or Monte Carlo simulations. We observed that outliers in the LR correlations occasionally exhibit major conformational changes of the complex during sampling. We treated the situation as a multimode binding case, for which the observed association constant is the sum of the partial association constants of individual states/modes. The resulting nonlinear expression for the binding affinities contains all the LR variables for individual modes that are scaled by the same two to four adjustable parameters as in the one-mode LR equation. The multimode method was applied to inhibitors of a matrix metalloproteinase, where this treatment improved the explained variance in experimental activity from 75% for the unimode case to about 85%. The predictive ability scaled accordingly, as verified by extensive cross-validations.     

Protein dynamics and cytochrome c: correlations between ligand vibrations and redox activity

A recently developed method to directly observe specific protein vibrations, based on deuteration, has been employed to examine the redox-dependent structural and fluctional properties of cytochrome c. The dynamics of the protein-based methionine heme ligand were examined by selectively deuterating the ligand's methyl group. The frequency and line width of the C-D bonds were easily observable and shown to be sensitive to mutation-induced changes in the protein redox potential. However, of seven mutants examined, the C-D line widths were independent of the redox-state of the protein. Therefore, although the ligand dynamics depend on the protein's redox state, there are no detected differences in protein dynamics of the oxidized and reduced proteins.     

Direct calculation of the binding free energies of FKBP ligands

Direct calculations of the absolute free energies of binding for eight ligands to FKBP protein were performed using the Fujitsu BioServer massively parallel computer. Using the latest version of the general assisted model building with energy refinement (AMBER) force field for ligand model parameters and the Bennett acceptance ratio for computing free-energy differences, we obtained an excellent linear fit between the calculated and experimental binding free energies. The rms error from a linear fit is 0.4 kcal/mol for eight ligand complexes. In comparison with a previous study of the binding energies of these same eight ligand complexes, these results suggest that the use of improved model parameters can lead to more predictive binding estimates, and that these estimates can be obtained with significantly less computer time than previously thought. These findings make such direct methods more attractive for use in rational drug design.     

Adiabatic electrochemical electron-transfer reactions involving frequency changes of inner-sphere modes

A Hamiltonian model is formulated for electrochemical electron-transfer reactions involving frequency changes of the inner-sphere modes. The adiabatic free energy surface is calculated and it is shown that this nonlinear coupling leads to a transfer coefficient different from 1/2, and that the transfer coefficient can no longer be identified with the electron occupation probability in the transition state saddle point, as in the linear coupling case.     

Ultrafast Structural Fluctuations of Myoglobin‐Bound Thiocyanate and Selenocyanate Ions Measured with Two‐Dimensional Infrared Photon Echo Spectroscopy

Structural dynamics within the distal cavity of myoglobin protein is investigated using 2D‐IR and IR pump–probe spectroscopy of the N≡C stretch modes of heme‐bound thiocyanate and selenocyanate ions. Although myoglobin‐bound thiocyanate group shows a doublet in its IR absorption spectrum, no cross peaks originating from chemical exchange between the two components are observed in the time‐resolved 2D IR spectra within the experimental time window. Frequency–frequency correlation functions of the two studied anionic ligands are obtained by means of a few different analysis approaches; these functions were then used to elucidate the differences in structural fluctuation around ligand, ligand–protein interactions, and the degree of structural heterogeneity within the hydrophobic pocket of these myoglobin complexes.     

Alchemical absolute protein–ligand binding free energies for drug design

The recent advances in relative protein–ligand binding free energy calculations have shown the value of alchemical methods in drug discovery. Accurately assessing absolute binding free energies, although highly desired, remains a challenging endeavour, mostly limited to small model cases. Here, we demonstrate accurate first principles based absolute binding free energy estimates for 128 pharmaceutically relevant targets. We use a novel rigorous method to generate protein–ligand ensembles for the ligand in its decoupled state. Not only do the calculations deliver accurate protein–ligand binding affinity estimates, but they also provide detailed physical insight into the structural determinants of binding. We identify subtle rotamer rearrangements between apo and holo states of a protein that are crucial for binding. When compared to relative binding free energy calculations, obtaining absolute binding free energies is considerably more challenging in large part due to the need to explicitly account for the protein in its apo state. In this work we present several approaches to obtain apo state ensembles for accurate absolute ΔG calculations, thus outlining protocols for prospective application of the methods for drug discovery.

Molecular dynamics based absolute protein–ligand binding free energies can be calculated accurately and at large scale to facilitate drug discovery.     

Free enthalpies of replacing water molecules in protein binding pockets

Water molecules in the binding pocket of a protein and their role in ligand binding have increasingly raised interest in recent years. Displacement of such water molecules by ligand atoms can be either favourable or unfavourable for ligand binding depending on the change in free enthalpy. In this study, we investigate the displacement of water molecules by an apolar probe in the binding pocket of two proteins, cyclin-dependent kinase 2 and tRNA-guanine transglycosylase, using the method of enveloping distribution sampling (EDS) to obtain free enthalpy differences. In both cases, a ligand core is placed inside the respective pocket and the remaining water molecules are converted to apolar probes, both individually and in pairs. The free enthalpy difference between a water molecule and a CH₃ group at the same location in the pocket in comparison to their presence in bulk solution calculated from EDS molecular dynamics simulations corresponds to the binding free enthalpy of CH₃ at this location. From the free enthalpy difference and the enthalpy difference, the entropic contribution of the displacement can be obtained too. The overlay of the resulting occupancy volumes of the water molecules with crystal structures of analogous ligands shows qualitative correlation between experimentally measured inhibition constants and the calculated free enthalpy differences. Thus, such an EDS analysis of the water molecules in the binding pocket may give valuable insight for potency optimization in drug design.     

Influence of the "innocent" ligands on the MLCT excited-state behavior of mono(bipyridine)ruthenium(II) complexes: a comparison of X-ray structures and 77 K luminescence properties

The variations in the nonchromophoric ligands of [Ru(L)4bpy]2+ complexes are shown to result in large changes in emission band shapes, even when the emission energies are similar. These changes in band shape are systematically examined by means of the generation of empirical reorganizational energy profiles (emreps) from the observed emission spectra (Xie, P.; et al. J. Phys. Chem. A 2005, 109, 4671), where these profiles provide convenient probes of the differences in distortions from the ground-state structures of the 2,2-bipyridine (bpy) ligands (for distortion modes near 1500 cm(-1)) in the metal-to-ligand charge-transfer (MLCT) excited states for a series of complexes with the same ruthenium(II) bipyridine chromophore. The bpy ligand is nearly planar in the X-ray structures of the complexes with (L)4 = (NH3)4, triethylenetetraamine (trien), and 1,4,7,10-tetraazacyclododecane ([12]aneN4). However, for (L)4 = 5,12-rac-5,7,7,12,14,14-hexamethyl-1,4,8,11-tetraazacyclotetradecane, the X-ray crystal structure shows that the bpy ligand is twisted in the ground state (a result of methyl/bpy stereochemical repulsion) and the emrep amplitude at about 1500 cm(-1) is significantly larger for this structure than for the complex with (L)4 = 1,4,8,11-tetraazacyclotetradecane, consistent with larger reorganizational energies of the bpy distortion modes in order to form a planar (bpy(-)) moiety in the excited state of the former. The trien and [12]aneN4 complexes have very nearly the same emission energies, yet the 40% smaller vibronic sideband intensity of the latter indicates that the MLCT excited state is significantly less distorted; this smaller distortion and the related shift in the distribution of distortion mode reorganizational energy amplitudes is apparently related to the 36-fold longer lifetime for (L)4 = [12]aneN4 than for (L)4 = trien. For the majority (77%) of the [Ru(L)4bpy]2+ complexes examined, there is a systematic decrease in emrep amplitudes near 1500 cm(-1), consistent with decreasing excited-state distortion, with the excited-state energy as is expected for ground state-excited state configurational mixing in a simple two-state model. However, the complexes with L = [12]aneN4, 1,4,7,10-tetraazacyclododeca-1-ene, and (py)4 all have smaller emrep amplitudes and thus less distorted excited states than related complexes with the same emission energy. The observations are not consistent with simple two-state models and seem to require an additional distortion induced by excited state-excited state configurational mixing in most complexes. Because the stereochemical constraints of the coordinated [12]aneN4 ligand restrict tetragonal distortions around the metal, configurational mixing of the 3MLCT excited state with a triplet ligand-field excited state of Ru(II) could account for some of the variations in excited-state distortion. The large number of vibrational distortion modes and their small vibrational reorganizational energies in these complexes indicate that a very large number of relaxation channels contribute to the variations in 3MLCT lifetimes and that the metal-ligand skeletal modes are likely to contribute to some of these channels.     

Incorporation of an amide into 5-phosphonoalkyl-6-D-ribitylaminopyrimidinedione lumazine synthase inhibitors results in an unexpected reversal of selectivity for riboflavin synthase vs lumazine synthase

Several analogues of a hypothetical intermediate in the reaction catalyzed by lumazine synthase were synthesized and tested as inhibitors of both Bacillus subtilis lumazine synthase and Escherichia coli riboflavin synthase. The new compounds were designed by replacement of a two-carbon fragment of several 5-phosphonoalkyl-6-D-ribitylaminopyrimidinedione lumazine synthase inhibitors with an amide linkage that was envisioned as an analogue of a Schiff base moiety of a hypothetical intermediate in the enzyme-catalyzed reaction. The incorporation of the amide group led to an unexpected reversal in selectivity for inhibition of lumazine synthase vs riboflavin synthase. Whereas the parent 5-phosphonoalkyl-6-D-ribitylaminopyrimidinediones were lumazine synthase inhibitors and did not inhibit riboflavin synthase, the amide-containing derivatives inhibited riboflavin synthase and were only very weak or inactive as lumazine synthase inhibitors. Molecular modeling of inhibitor-lumazine synthase complexes did not reveal a structural basis for these unexpected findings. However, molecular modeling of one of the inhibitors with E. coli riboflavin synthase demonstrated that the active site of the enzyme could readily accommodate two ligand molecules.     

TAUTOMERISM AND FLUORESCENCE OF LUMAZINE

Abstract— The fluorescence properties of lumazine, 1-methyl-lumazine (1-ML) and 3-methyl-lumazine (3-ML) in aqueous solution are pH dependent. Emissions with the following maxima were attributed to the four ionic species of lumazine: dianion (483 nm), monoanion (467 nm), neutral (380 nm) and monocation (505 nm). Neutral lumazine emitted, besides, a fluorescence at 481 nm with a large Stokes shift (10 000 cm^-1). As a similar emission is observed with 3-ML but not with 1-ML, we suggest as emitting species the N(8)-H-phototautomer resulting from the N(l) to N(8) proton transfer in the excited state. At pH 10, the fluorescence quantum yield of 3-ML, Q_f= 0.24 ± 0.02 was higher than that of 1-ML, Q_f= 0.015 ± 0.002. At this pH, lumazine is a mixture of N(l)-deprotonated and N(3)-deprotonated monoanions, the emitting properties being principally due to the N(l)-deprotonated species, Q_f= 0.24 ± 0.02. The evaluation of the ionization constants in the excited state are discussed in relation to the tautomeric nature of the emitting species.     

Spectroscopic and computational studies of Co2+corrinoids: spectral and electronic properties of the biologically relevant base-on and base-off forms of Co2+cobalamin

Co(2+)cobalmain (Co(2+)Cbl) is implicated in the catalytic cycles of all adenosylcobalamin (AdoCbl)-dependent enzymes, as in each case catalysis is initiated through homolytic cleavage of the cofactor's Co-C bond. The rate of Co-C bond homolysis, while slow for the free cofactor, is accelerated by 12 orders of magnitude when AdoCbl is bound to the protein active site, possibly through enzyme-mediated stabilization of the post-homolysis products. As an essential step toward the elucidation of the mechanism of enzymatic Co-C bond activation, we employed electronic absorption (Abs), magnetic circular dichroism (MCD), and resonance Raman spectroscopies to characterize the electronic excited states of Co(2+)Cbl and Co(2+)cobinamide (Co(2+)Cbi(+), a cobalamin derivative that lacks the nucleotide loop and 5,6-dimethylbenzimazole (DMB) base and instead binds a water molecule in the lower axial position). Although relatively modest differences exist between the Abs spectra of these two Co(2+)corrinoid species, MCD data reveal that substitution of the lower axial ligand gives rise to dramatic changes in the low-energy region where Co(2+)-centered ligand field transitions are expected to occur. Our quantitative analysis of these spectral changes within the framework of time-dependent density functional theory (TD-DFT) calculations indicates that corrin-based pi --> pi transitions, which dominate the Co(2+)corrinoid Abs spectra, are essentially insulated from perturbations of the lower ligand environment. Contrastingly, the Co(2+)-centered ligand field transitions, which are observed here for the first time using MCD spectroscopy, are extremely sensitive to alterations in the Co(2+) ligand environment and thus may serve as excellent reporters of enzyme-induced perturbations of the Co(2+) state. The power of this combined spectroscopic/computational methodology for studying Co(2+)corrinoid/enzyme active site interactions is demonstrated by the dramatic changes in the MCD spectrum as Co(2+)Cbi(+) binds to the adenosyltransferase CobA.     

Influence of droplet size, capillary-cone distance and selected instrumental parameters for the analysis of noncovalent protein-ligand complexes by nano-electrospray ionization mass spectrometry

It has been suggested in the literature that nano-electrospray ionization (nano-ESI) mass spectrometry better reflects the equilibrium between complex and free protein in solution than pneumatically assisted electrospray ionization (ESI) in noncovalent interaction studies. However, no systematic studies of the effects of ionization conditions have been performed to support this statement. In the present work, different instrumental and sample-derived parameters affecting the stability of noncovalent complexes during analysis by nano-ESI were investigated. In general, increased values of parameters such as drying gas flow-rate, ion-source temperature, capillary tip voltage and buffer concentration lead to a dissociation of ribonuclease A (RNAse)-cytidine 2'-monophosphate (CMP) and cytidine 5'-triphosphate (CTP) complexes. The size of the electrosprayed droplets was shown to be an important issue. Increasing the capillary to cone distance yielded an increased complex to free protein ratio when a hydrophilic ligand was present and the reverse effect was obtained with a hydrophobic ligand. Important in this regard is the degree of sampling of ions originating from late-generation residue droplets, that is, ions present in the droplet bulk. Sampling of these ions increases with longer capillary-cone distance (flight time). Furthermore, when the sample flow-rate was increased by increasing the capillary internal tip i.d. from 4 to 30 microm, a decreased complex to free protein ratio for the RNAse-CTP system was observed. This behavior was consistent with the change in surface to volume ratio for flow-rates between 2 and 100 nl min(-1). Finally, polarity switching between positive and negative ion modes gave a higher complex to free protein ratio when the ligand and the protein had the same polarity.     

The Confine-and-Release Method: Obtaining Correct Binding Free Energies in the Presence of Protein Conformational Change

Free energy calculations are increasingly being used to estimate absolute and relative binding free energies of ligands to proteins. However, computed free energies often appear to depend on the initial protein conformation, indicating incomplete sampling. This is especially true when proteins can change conformation on ligand binding, as free energies associated with these conformational changes are either ignored or assumed to be included by virtue of the sampling performed in the calculation. Here, we show that, in a model protein system (a designed binding site in T4 Lysozyme), conformational changes can make a difference of several kcal/mol in computed binding free energies, and that they are neglected in computed binding free energies if the system remains kinetically trapped in a particular metastable state on simulation timescales. We introduce a general "confine-and-release" framework for free energy calculations that accounts for these free energies of conformational change. We illustrate its use in this model system by demonstrating that an umbrella sampling protocol can obtain converged binding free energies that are independent of the starting protein structure and include these conformational change free energies.     

Elastic network normal modes provide a basis for protein structure refinement

It is well recognized that thermal motions of atoms in the protein native state, the fluctuations about the minimum of the global free energy, are well reproduced by the simple elastic network models (ENMs) such as the anisotropic network model (ANM). Elastic network models represent protein dynamics as vibrations of a network of nodes (usually represented by positions of the heavy atoms or by the C(α) atoms only for coarse-grained representations) in which the spatially close nodes are connected by harmonic springs. These models provide a reliable representation of the fluctuational dynamics of proteins and RNA, and explain various conformational changes in protein structures including those important for ligand binding. In the present paper, we study the problem of protein structure refinement by analyzing thermal motions of proteins in non-native states. We represent the conformational space close to the native state by a set of decoys generated by the I-TASSER protein structure prediction server utilizing template-free modeling. The protein substates are selected by hierarchical structure clustering. The main finding is that thermal motions for some substates, overlap significantly with the deformations necessary to reach the native state. Additionally, more mobile residues yield higher overlaps with the required deformations than do the less mobile ones. These findings suggest that structural refinement of poorly resolved protein models can be significantly enhanced by reduction of the conformational space to the motions imposed by the dominant normal modes.