Electrostatic screening and backbone preferences of amino acid residues in urea-denatured ubiquitin

Characterizing local structures of denatured proteins is crucial for understanding the protein folding (Baldwin 1986) and misfolding processes. Chemically denatured proteins unfolded in 8 M urea or GdmCl are generally highly opened and solvent-exposed polypeptides with no detectable αR-helix or β-sheet secondary structures. Proteins under such strong denaturing conditions are considered to be completely unfolded (Dill and Shortle 1991). It was commonly assumed that these proteins are true random coils that can be described by Flory's random coil model (Flory 1969). The random coil model predicts that residues in unfolded proteins populate mainly the broad β region of φ–ψ space, which contains about equally probable β (φ ≈ −120° and ψ ≈ 120°) and PII (φ ≈ −65°, ψ ≈ 145°) conformations. A growing number of experimental studies show that chemically denatured proteins display a certain degree of order even under the most severe denaturing conditions. The coupling constants 3J(Hα,HN) and chemical shifts of chemically denatured proteins in general deviate little from their values in small peptides in aqueous solution (Arcus et al. 1995; Frank et al. 1995; Schwalbe et al. 1997; Fong et al. 1998; Meekhof and Freund 1999; Peti et al. 2000; Bai et al. 2001; Schwarzinger et al. 2002). Small peptides in water were presumed to be completely unstructured (random coil); therefore, it was construed that chemically denatured proteins are also in a random coil state. However, recent experimental studies have shown that small peptides adopt relatively stable structures in aqueous solutions (Woutersen and Hamm 2000; Schweitzer-Stenner et al. 2001; Eker et al. 2002, 2003, 2004a,b; Shi et al. 2002; Ding et al. 2003; Weise and Weisshaar 2003; McColl et al. 2004; Schweitzer-Stenner et al. 2004). The combined polarized Raman, Fourier transform IR spectroscopy, and vibrational CD measurements have shown that VVV and AVA tripeptides in aqueous solutions adopt predominantly the β-strand conformation (Eker et al. 2002, 2004a). The conformational preferences of residues in small peptides are not significantly affected by urea or GdmCl (Plaxco et al. 1997; Schwarzinger et al. 2000). The NMR study has shown that the alanine residue of GGAGG adopts mainly the PII conformation (φ ≈ −65°, ψ ≈ 145°) in water (Ding et al. 2003). Small deviations of the NMR parameters of chemically denatured proteins from their values in small peptides thus actually indicate that better ordered local structures exist in these proteins. Distinct patterns of backbone conformational preferences and flexibility, which sometimes coincide with secondary structures in the native states, have been observed in chemically denatured proteins (Arcus et al. 1995; Frank et al. 1995; Farrow et al. 1997; Schwalbe et al. 1997; Fong et al. 1998; Meekhof and Freund 1999; Peti et al. 2000; Bai et al. 2001; Schwarzinger et al. 2002; Mohana-Borges et al. 2004). These patterns indicate that a certain level of cooperativity exists between nearest-neighbor residues in chemically denatured proteins. It has been shown recently that blocked amino acids (dipeptides) display diverse intrinsic preferences for certain φ,ψ backbone angles (Avbelj et al. 2006) found in short peptides (Bundi and Wuthrich 1979; Merutka et al. 1995; Plaxco et al. 1997; Schwarzinger et al. 2000; Eker et al. 2004a), chemically denatured proteins (Penkett et al. 1997; Peti et al. 2000), and native proteins (Serrano 1995; Smith et al. 1996). Strong backbone electrostatic interactions occur in dipeptides (Avbelj et al. 2006) and polypeptides (Avbelj and Moult 1995; Avbelj and Fele 1998b; Avbelj 2000). The side chain–dependent screening of these interactions by water dipoles (the electrostatic screening model of conformational preferences [ESM]) has been suggested to cause the conformational preferences of residues in blocked dipeptides, model peptides, and unfolded and native proteins (Avbelj and Moult 1995; Avbelj and Fele 1998a,b; Avbelj 2000; Avbelj and Baldwin 2002, 2003; Avbelj et al. 2006). Strong electrostatic coupling among closely spaced and large backbone dipole moments between neighboring residues has been introduced in the strand-coil transition model (Avbelj and Fele 1998b) and electrostatic screening model of unfolded proteins (Avbelj 2000; Avbelj and Baldwin 2003). The energetics of a residue is in the ESM determined mainly by the local backbone electrostatic energy (E local) and its screening by backbone solvation. The β-strand is a low-energy conformation of a polypeptide chain in vacuo, because the alignment of neighboring peptide dipole moments is antiparallel (Avbelj and Moult 1995). The difference in the electrostatic energy E local of a residue in the higher-energy αR-helix (parallel backbone dipole moments) and the β-strand conformations is very large (≈5 kcal/mol). The difference in the values of E local between the polyproline-helix and the β-strand conformations is smaller but still significant (≈1.3 kcal/mol). In water, however, the free energy difference between these two conformations is smaller than the difference in the values of E local and may even be zero. The reason lies in the extraordinary ability of water to screen electrostatic interactions. The change in the value of E local for a particular conformational transition of a residue is partly or completely compensated (screened) by the backbone solvation (ESF). Screening of E local depends on the sizes and shapes of nearby side chains, which causes distinct preferences of residues for backbone conformations. It has been shown also that the screening of backbone electrostatic interactions by water is the main cause of the neighboring residue effect on peptide conformation (Avbelj and Baldwin 2004). The ESM predicts existence of fluctuating β-strands in completely denatured proteins under the drive of electrostatic interactions even in the absence of stabilizing long-range tertiary interactions (Avbelj and Fele 1998b). Fluctuating β-strands should form in those regions of an unfolded polypeptide chain that contain mainly nonpolar residues, because large nonpolar side chains shield the backbone and prevent screening by water dipoles. The local backbone electrostatic interactions are thus stronger, stabilizing the β-strand conformation. The electrostatic screening effect should persevere even at very high concentrations of denaturant; however, fluctuating β-strands in chemically denatured proteins have not been demonstrated. To show the presence of these β-strands in chemically denatured proteins, we need to determine the populations of residues in the most populated backbone conformations along a polypeptide chain. A major problem in determining these populations is the overlap of φ and ψ dihedral angles in the most important backbone conformations of a residue: β (φ ≈ −120°, ψ ≈ 120°), PII (φ ≈ −65°, ψ ≈ 145°), and αR (φ ≈ −60°, ψ ≈ −40°). The PII conformation has the ψ of the β conformation but the φ of the αR conformation. In this study, we estimated these populations by measuring the ratios of NMR NOE connectivities, Q NOE, and the residual dipolar couplings, D NH, of urea-denatured ubiquitin (8 M urea at pH 2). We also used its coupling constants, 3 J(Hα,HN), and cross-correlated relaxation rates, ΓHN,CαNα c, measured by Peti et al. (2000). Under these conditions, ubiquitin contains no detectable αR-helix or β-sheet secondary structures, but the NMR data are shown to be consistent with the predictions of the ESM. Fluctuating β-strands are predicted using the strand-coil transition model (Avbelj and Fele 1998b).