Search

The molecular details of how urea interacts with, and eventually denatures proteins, remain largely unknown. In this study we have used extensive experimental NMR data, in combination with statistical coil ensemble modeling and small-angle scattering, to analyze the conformational behavior of the protein ubiquitin in the presence of urea. In order to develop an atomic resolution understanding of the denatured state, conformational ensembles of full-atom descriptions of unfolded proteins, including side chain conformations derived from rotamer libraries, are combined with random sampling of explicit urea molecules in interaction with the protein. Using this description of the conformational equilibrium, we demonstrate that the direct-binding model of urea to the protein backbone is compatible with available experimental data. We find that, in the presence of 8 M urea, between 30 and 40% of the backbone peptide groups bind a urea molecule, independently reproducing results from a model-free analysis of small-angle neutron and X-ray scattering data. Crucially, this analysis also provides sequence specific details of the interaction between urea and the protein backbone. The pattern of urea-binding along the amino-acid sequence reveals a higher level of binding in the central part of the protein, a trend which resembles independent results derived from chemical shift mapping of the urea-protein interaction. Together these results substantiate the direct-binding model and provide a framework for studying the physical basis of interactions between proteins and solvent molecules.

Current NMR information on side-chain conformations of unfolded protein states is sparse due to the poor dispersion particularly of side-chain proton resonances. We present here optimized schemes for the detection of (3)J(HalphaHbeta), (3)J(NHbeta), and (3)J(C'Hbeta) scalar and (1)D(CbetaHbeta) residual dipolar couplings (RDCs) in unfolded proteins. For urea-denatured ubiquitin and protein G, up to six (3)J-couplings to (1)H(beta) are detected, which define the chi(1) angle at very high precision. Interpretation of the (3)J couplings by a model of mixed staggered chi(1) rotamers yields excellent agreement and also provides stereoassignments for (1)H(beta) methylene protons. For all observed amino acids with the exception of leucine, the chemical shift of (1)H(beta3) protons was found downfield from (1)H(beta2). For most residues, the precision of individual chi(1) rotamer populations is better than 2%. The experimental chi(1) rotamer populations are in the vicinity of averages obtained from coil regions in folded protein structures. However, individual variations from these averages of up to 40% are highly significant and indicate sequence- and residue-specific interactions. Particularly strong deviations from the coil average are found for serine and threonine residues, an effect that may be explained by a weakening of side-chain to backbone hydrogen bonds in the urea-denatured state. The measured (1)D(CbetaHbeta) RDCs correlate well with predicted RDCs that were calculated from a sterically aligned coil model ensemble and the (3)J-derived chi(1) rotamer populations. This agreement supports the coil model as a good first approximation of the unfolded state. Deviations between measured and predicted values at certain sequence locations indicate that the description of the local backbone conformations can be improved by incorporation of the RDC information. The ease of detection of a large number of highly precise side-chain RDCs opens the possibility for a more rigorous characterization of both side-chain and backbone conformations in unfolded proteins.

Intrinsically disordered regions are predicted to exist in a significant fraction of proteins encoded in eukaryotic genomes. The high levels of conformational plasticity of this class of proteins endows them with unique capacities to act in functional modes not achievable by folded proteins, but also places their molecular characterization beyond the reach of classical structural biology. New techniques are therefore required to understand the relationship between primary sequence and biological function in this class of proteins. Although dependences of some NMR parameters such as chemical shifts (CSs) or residual dipolar couplings (RDCs) on structural propensity are known, so that sampling regimes are often inferred from experimental observation, there is currently no framework that allows for a statistical mapping of the available Ramachandran space of each amino acid in terms of conformational propensity. In this study we develop such an approach, combining highly efficient conformational sampling with ensemble selection to map the backbone conformational sampling of IDPs on a residue specific level. By systematically analyzing the ability of NMR data to map the conformational landscape of disordered proteins, we identify combinations of RDCs and CSs that can be used to raise conformational degeneracies inherent to different data types, and apply these approaches to characterize the conformational behavior of two intrinsically disordered proteins, the K18 domain from Tau protein and N(TAIL) from measles virus nucleoprotein. In both cases, we identify the enhanced populations of turn and helical regions in key regions of the proteins, as well as contiguous strands that show clear and enhanced polyproline II sampling.

During oxidative folding, the formation of disulfide bonds has profound effects on guiding the protein folding pathway. Until now, comparatively little is known about the changes in the conformational dynamics in folding intermediates of proteins that contain only a subset of their native disulfide bonds. In this comprehensive study, we probe the conformational landscape of non-native states of lysozyme containing a single native disulfide bond utilizing nuclear magnetic resonance (NMR) spectroscopy, small-angle X-ray scattering (SAXS), circular dichroism (CD) data, and modeling approaches. The impact on conformational dynamics varies widely depending on the loop size of the single disulfide variants and deviates significantly from random coil predictions for both NMR and SAXS data. From these experiments, we conclude that the introduction of single disulfides spanning a large portion of the polypeptide chain shifts the structure and dynamics of hydrophobic core residues of the protein so that these regions exhibit levels of order comparable to the native state on the nanosecond time scale.

The detailed, quantitative characterization of unfolded proteins is a largely unresolved task due to the enormous experimental and theoretical difficulties in describing the highly dimensional space of their conformational ensembles. Recently, residual dipolar coupling (RDC) and paramagnetic relaxation enhancement (PRE) data have provided large numbers of experimental parameters on unfolded states. To obtain a minimal model of the unfolded state according to such data we have developed new modules for the use of steric alignment RDCs and PREs as constraints in ensemble structure calculations by the program XPLOR-NIH. As an example, ensemble calculations were carried out on urea-denatured ubiquitin using a total of 419 previously obtained RDCs and 253 newly determined PREs from eight cysteine mutants coupled to MTSL. The results show that only a small number of about 10 conformers is necessary to fully reproduce the experimental RDCs, PREs and average radius of gyration. C(alpha) contacts determined on a large set (400) of 10-conformer ensembles show significant (10-20%) populations of conformations that are similar to ubiquitin's A-state, i.e. corresponding to an intact native first beta-hairpin and alpha-helix as well as non-native alpha-helical conformations in the C-terminal half. Thus, methanol/acid (A-state) and urea denaturation lead to similar low energy states of the protein ensemble, presumably due to the weakening of the hydrophobic core. Similar contacts are obtained in calculations using solely RDCs or PREs. The sampling statistics of the C(alpha) contacts in the ensembles follow a simple binomial distribution. It follows that the present RDC, PRE, and computational methods allow the statistically significant detection of subconformations in the unfolded ensemble at population levels of a few percent.

A new lanthanide chelating tag (M8) for paramagnetic labeling of biomolecules is presented, which is based on an eight-fold, stereoselectively methyl-substituted DOTA that can be covalently linked to the host molecule by a single disulfide bond. The steric overcrowding of the DOTA scaffold leads to an extremely rigid, kinetically and chemically inert lanthanide chelator. Its steric bulk restricts the motion of the tag relative to the host molecule. These properties result in very large pseudocontact shifts (>5 ppm) and residual dipolar couplings (>20 Hz) for Dy-M8 linked to ubiquitin, which are unprecedented for a small, single-point-attachment tag. Such large pseudocontact shifts should be well detectable even for larger proteins and distances beyond approximately 50 A. Due to its exceptionally high stability and lanthanide affinity M8 can be used under extreme chemical or physical conditions, such as those applied for protein denaturation, or when it is undesirable that buffer or protein react with excess lanthanide ions.