N-H Spin-Spin Couplings: Probing Hydrogen Bonds in Proteins
Sheng-Qi Xiang,Raghavendran L. Narayanan,Stefan Becker,Markus Zweckstetter
DOI: https://doi.org/10.1002/anie.201209641
2013-01-01
Angewandte Chemie
Abstract:Sensitive information: Hydrogen bonds stabilize protein structures and are critical for many chemical and biological phenomena. NMR studies found that the one-bond 1H–15N scalar coupling constant is increased by up to 1.6 Hz when a hydrogen bond is formed. These coupling constants therefore provide a sensitive tool for studying protein structure. Hydrogen bonds are essential for the structure of many biochemical compounds. Protein folding, the formation of amyloid aggregates, enzymatic catalysis, drug–receptor interactions, and many other phenomena are intrinsically connected to hydrogen bonding. NMR chemical shifts,1–3 anisotropies of the chemical shifts,1, 4, 5 and nuclear quadrupole coupling constants1, 6 are influenced by hydrogen bonding, making NMR spectroscopy a key tool for the detection and characterization of hydrogen bonds. The experimental observation of NMR spin–spin coupling constants across hydrogen bonds allows the direct detection of hydrogen bonding in proteins.9 In addition, detailed insight into the nature of hydrogen bonds can be obtained by theoretical calculations.10–12 Hydrogen bonding in proteins involves, in most cases, arrangements of the type CO⋅⋅⋅HN. Six spin–spin coupling constants should be affected by hydrogen bonding, four directly because they are spin–spin coupling constants across the hydrogen bond and two indirectly because they involve atoms participating in hydrogen bonding. However, only the 13C–15N three-bond scalar coupling, 3JNC, across the hydrogen bond and the one-bond 1H–15N scalar coupling, 1JNH, can be currently measured in proteins. The magnitude of the 1JNH spin–spin coupling constant is dependent upon the electronic structure of the molecule.11, 13, 14 Theoretical studies on small model systems have shown that 1JNH is negative and becomes more negative upon the formation of a hydrogen-bond.11, 14 Moreover, conformational changes due to ligand binding can induce changes in 1JNH spin–spin coupling constants.16 Although it is possible to accurately measure 1JNH scalar couplings, they have not been employed for the detection and characterization of hydrogen bonds in biomolecules. We attribute this to the fact that changes in 1JNH coupling constants due to hydrogen-bond formation are predicted to be small,11, 14 a systematic analysis of 1JNH spin–spin coupling constants in disordered proteins, which might serve as reference, is lacking, and the influence of pH and temperature is unknown. To overcome these limitations we have measured a large set of 1JNH coupling constants in intrinsically disordered proteins as a function of pH and temperature and compared these to values in folded proteins. Our study demonstrates that the magnitude of 1JNH spin–spin coupling constants increases by up to 1.6 Hz upon hydrogen-bond formation. Thus 1JNH spin–spin couplings can provide a sensitive tool for the study of hydrogen bonds. We measured 1JNH scalar couplings in the 441 residue protein Tau at a static magnetic field strength of 21.1 T and a temperature of 278 K (see Table S1 in the Supporting Information).17 Tau belongs to the class of intrinsically disordered proteins and displays only transient secondary structure.18 A total of 367 1JNH values were obtained for nonoverlapping residues. Experimental values were observed for all non-proline amino acid types except tryptophan, which is not present in the primary sequence of Tau. 1JNH values belonging to the same amino acid type were grouped and median 1JNH values were calculated (Figure 1 a). Glycine and alanine residues have the most negative 1JNH values (−94.2 Hz and −93.4 Hz, respectively). For other amino acids, values ranged from −93.3 Hz (serine) to −93.0 Hz (valine) (Table S2). a)1JNH spin–spin coupling constants observed in the intrinsically disordered protein Tau using 1JNH-modulated 2D 1H–15N heteronuclear single quantum coherence (HSQC) experiments.7 1JNH values were grouped according to amino acid type and subjected to a box plot analysis. The center of the box indicates the median value for each residue type, while bottom and top correspond to 25th and 75th percentile, respectively. The mean values are outlined with an empty box (□). b) Correlation between 1JNH values in α-synuclein at pH 6.5 and pH 5.7, 278 K. c) Correlation between 1JNH values in α-synuclein at 288 K and 298 K, pH 5.7. Data in (b) and (c) were measured using IPAP-HSQC, in which long-range couplings are not suppressed.15 d,e) Deviation of 1JNH spin–spin coupling constants observed in Tau (d) and α-synuclein (e), both at pH 6.0, 278 K, from their amino acid specific random-coil values as a function of residue number. The dashed line indicates the estimated error range of 0.35 Hz for Δ1JNH taking into account uncertainties in the random-coil values (a) and potential influence of pH and temperature (b,c). To systematically investigate the influence of pH and temperature, we measured 1JNH couplings in the intrinsically disordered protein α-synuclein at different temperatures and pH values (pH 5.7: 278 K, 288 K, 298 K; pH 6.0: 278 K); pH 6.5: 278 K, and pH 7.4: 278 K, 288 K; Table S3). As a result of increased amide proton exchange the signal-to-noise ratio decreased and the experimental error in 1JNH increased at higher pH and temperature. However, at all tested temperatures and pH values 1JNH values were highly correlated (Figure 1 b,c and Figure S1), indicating that 1JNH values in α-synuclein are stable against pH and temperature variations. In addition, comparison of 1JNH values in the 130 residue Tau fragment K18 observed at 400 and 600 MHz proton frequency (Figure S2) suggests that the contribution of the dynamic frequency shift7, 19, 20 is negligible in intrinsically disordered proteins. Accordingly, mean values of 1JNH observed in α-synuclein at 700 MHz and in Tau at 900 MHz were identical within experimental error (Figure S3). The intrinsically disordered proteins α-synuclein and Tau populate transient secondary structures; for Tau 428LADEVSASLA437 the contribution of the helical structure can reach 25 %.17 Figure 1 d,e show secondary 1JNH values, Δ1JNH, that is, the difference between experimental 1JNH spin–spin coupling constants and the random-coil values from Table S2. For both intrinsically disordered proteins, Δ1JNH values were below 0.35 Hz, which corresponds to the estimated uncertainty in Δ1JNH due to uncertainties in the random-coil values (Figure 1 a) and any remaining influence of pH and temperature (Figure 1 b,c and Figure S1). Only K96 in α-synuclein showed systematically negative Δ1JNH values at different temperatures and pH values (Table S3). The data demonstrate that 1JNH random-coil values are highly stable. Hydrogen bonds potentially present in transient secondary structures of Tau and α-synuclein do not have sufficiently long lifetimes to lead to a detectable change in the 1JNH coupling constant. Having established a set of robust amino acid specific reference values, we studied the variation of 1JNH coupling constants in folded proteins (Figure 2). Particularly suited for this investigation is the protein ubiquitin as it has been extensively studied by NMR spectroscopy.5, 9, 21 Figure 2 a shows secondary 1JNH values in ubiquitin as a function of residue number. Both negative and positive Δ1JNH were found that cover a range from about −1.3 Hz to +2.3 Hz. Ile23–Asp32 within the α-helix of ubiquitin showed a continuous set of negative Δ1JNH values with an average value of approximately −1.0 Hz (Figure 2 a). Positive Δ1JNH values were predominantly observed in loop regions and at the edges of helices and strands. 1JNH spin–spin coupling constants in globular proteins. a) Deviation of 1JNH spin–spin coupling constants observed in ubiquitin from the amino acid specific random-coil values as a function of residue number. Filled and hatched bars mark amide protons involved in hydrogen bonds to the backbone or a side chain, respectively. Stars indicate spatial proximity (<6 Å) to aromatic rings. Secondary structures identified by DSSP8 in the crystal structure of ubiquitin (pdb code: 1ubq) are marked above. b) Selected part of a BSD-IPAP-HSQC spectrum of ubiquitin. Doublet components are shown in red and blue. c) Expanded region of the crystal structure of ubiquitin highlighting the hydrogen bonds between β-strands 1 and 2. Hydrogen bonds are in green. Amide protons involved in hydrogen bonds are in blue while those not in hydrogen bonds are in purple. The aromatic ring of Phe4 is shown in red and residues in spatial proximity (<6 Å) are shown in yellow. d) Deviation of 1JNH couplings observed in DinI from the amino acid specific random-coil values as a function of residue number. Secondary structures identified by DSSP8 in the 3D structure (pdb code: 1ghh) are marked above. According to theoretical studies on small model systems 1JNH is predicted to become more negative upon hydrogen-bond formation.11, 14 Figure 2 a demonstrates that amide protons with Δ1JNH