Lewis E Kay

Three-Dimensional Triple-Resonance NMR Spectroscopy of Isotopically Enriched Proteins

correlates the amide ‘H and “N shifts with the 13C shift of the carbonyl resonance of the preceding amino acid. A second experiment (HNCA) correlates the intraresidue amide ‘H and 15N shifts with the CLY chemical shift. This experiment often also provides a weak correlation between the amide NH and 15N resonances of one amino acid and the Ca resonance of the preceding amino acid. A third experiment (HCACO) correlates the Ha and GY shifts with the intraresidue carbonyl shift. Finally, a 3D relay experiment, HCA( CO)N, correlates Ha and Cal resonances of one residue with the “N frequency of the succeeding residue. The principles of these experiments are described in terms of the operator formalism. To optimize spectral resolution, special attention is paid to removal of undesired J splittings in the 3D spectra. Technical details regarding the implementation of these triple-resonance experiments on a commercial spectrometer are also provided. The experiments are demonstrated for the protein calmodulin ( 16.7 kDa).

Comparison of different modes of two-dimensional reverse-correlation NMR for the study of proteins

Different two-dimensional NMR schemes for generating ‘H-detected ‘H-“N and ‘H13C correlation spectra are compared. It is shown that the resolution in the dimension that represents the “C or “N chemical shift depends on the type of correlation scheme used. For “N NMR studies of proteins, it is found that experiments that involve “N single-quantum coherence offer improved resolution compared to multiple-quantum correlation experiments, mainly because the ‘H- ‘H dipolar broadening of the multiplequantum coherence is stronger than the heteronuclear dipolar broadening of “N, but also because ofthe presence ofunresolved Jsplittings in the F, dimension ofthe multiplequantum correlation spectra. For 13C, the heteronuclear dipolar interaction is much larger and the ‘H-13C multiple-quantum relaxation is slower than the “C transverse relaxation; however, because of the presence of ‘H- ‘H Jcouplings in the F, dimension of such spectra, in practice the multiplequantum type correlation experiments often offer no gain or even a small loss in resolution, compared to experiments that use transverse 13C magnetization during the evolution period. A modified pulse scheme that increases F, resolution by elimination of scalar relaxation of the second kind is proposed. Experiments for the proteins calmodulin, uniformly enriched with “N, and staphylococcal nuclease, labeled with 13C in the Ca position of all Pro residues are demonstrated.

Pulse sequences for removal of the effects of cross correlation between dipolar and chemical-shift anisotropy relaxation mechanisms on the measurement of heteronuclear T1 and T2 values in proteins

The effects of cross correlation between dipolar and chemical-shift anisotropy relaxation interactions on the measurement of heteroatom T1 and T2 relaxation times in proteins is considered. It is shown that such effects can produce errors of approximately 25% in the measurement of 15N transverse relaxation times at a field strength of 11.8 T. Cross correlation has a less significant effect on the measurement of 15N spin-lattice relaxation rates and for proteins the errors in T1 decrease as a function of increasing molecular weight. Nevertheless, for T1 measurements at 11.8 T errors of approximately 15 and 5% are calculated for proteins with correlation times, τc, of 5 and 9 ns, respectively. Pulse sequences which eliminate dipolar and chemical-shift anisotropy cross-correlation effects are described. These sequences are used to make more accurate measurements of 15N T1 and T2 values of staphylococcal nuclease and to determine errors in these parameters that result when cross correlations are present.

New methods for the measurement of NHCαH coupling constants in 15N-labeled proteins

Abstract Several new techniques, requiring 15 N incorporation, are described for measuring NHCαH J couplings in proteins. 1 H-detected heteronuclear 1 H 15 N multiple-quantum correlation spectra retain the homonuclear J coupling information. Because of the favorable relaxation properties of 15 N 1 H zero- and double-quantum coherences, significant line narrowing occurs in the F 1 , dimension compared to the regular NH 1 H linewidth, permitting high accuracy measurements of J splittings, even for medium sized proteins. Methods for convenient analysis of such coupling information are described, correcting for linewidth and dispersion mode contributions. The new approach is demonstrated for the protein staphylococcal nuclease (18 kDa), complexed with pdTp and calcium.

Practical aspects of 3D heteronuclear NMR of proteins

Abstract Practical aspects regarding the acquisition and processing of 3D heteronuclear data sets are discussed, with particular emphasis on the 3D NOESY-HMQC experiment which combines the 2D NOE and the heteronuclear multiple-quantum coherence (HMQC) experiments. Slices through the 3D spectrum are equivalent to 15 N-filtered 2D NOESY spectra and exhibit sensitivity similar to that obtained in regular 2D NOE experiments. We discuss experimental procedures for obtaining maximum resolution with a relatively small number of t 1 and t 2 increments. In addition, a simple and efficient procedure for performing the third Fourier transformation that permits use of standard 2D software for processing in the other dimensions is described. Other important aspects of the data processing concern optimal digital filtering, noninteractive phasing, and minimization of the space needed for the processed data.

Three-dimensional NOESY-HMQC spectroscopy of a 13C-labeled protein

The analysis of NMR spectra of med ium-size ( lo-20 kDa) proteins is often difficult because of severe signal overlap. A number of isotope-edited 2D experiments (I7) and, more recently, 3D NMR experiments have been used to alleviate this problem (8-12). Spreading spectral information in three independent frequency dimensions greatly reduces spectral overlap and thereby simplifies the process of analysis. The heteronuclear 3D experiments (I.?, 10-12) are particularly useful in this respect, because the total number of resonances remains unchanged relative to the corresponding homonuclear ‘H 2D spectrum; the chemical shift of the heteronucleus is merely used to disperse the resonances in the regular 2D spectrum along a third axis. All applications of heteronuclear 3D experiments to proteins publ ished to date combine the commonly used homonuclear ‘H experiments NOESY (14,15) and HOHAHA ( 16, 17) with the ‘H-detected heteronuclear mu ltiplequantum correlation (HMQC) experiment ( 18-20). Recently, it has been shown that the HMQC experiment does not provide as high a resolution as some more complex pulse sequences (21). However, the resolution in the dimension of the heteronuclear chemical shift is usually lim ited by digital resolution, and the potentially lower resolution does not play a role. So far, all protein applications of heteronuclear 3D NMR have utilized 15N as the heteronucleus. Here we report the first use of heteronuclear 3D protein NMR using the r3C chemical shift to spread the ‘H resonances. Although at first sight, the change from “N to 13C may appear simple, there were a number of problems that had made it uncertain whether this approach would be successful. F irst, the heteronuclear dipolar l ine-broadening effect of 13C on the resonance of its directly attached proton(s) is severe, typically causing a twofold increase of the regular proton linewidth. This sharply reduces sensitivity in those 3D experiments where one of the magnetization transfer steps relies on ‘H‘H Jcoupling. Our experiments with fully (about 95%) ‘3C-labeled protein also showed a reduction in the proton T, by about 30% relative to the unlabeled protein, and it was not certain a priori whether this strong heteronuclear dipolar relaxation would have a negative influence on the quality of the NOESY spectrum. A second problem arises from the presence of relatively large r Jcc couplings, varying from about 60 Hz for C,-carbonyl and aromatic couplings to about 35 Hz for couplings between aliphatic carbons. Attempting to resolve these splittings in the 3D experiment would increase the number of resonances and thus decrease signal-to-noise. As shown by Markley and co-workers ( 22-24)) using a relatively low level of 13C labeling reduces the * 3Cr3C broaden-

Practical aspects of proton-carbon-carbon-proton three-dimensional correlation spectroscopy of 13C-labeled proteins

The determination of protein solution structures by NMR requires complete assignment of both the backbone and the amino acid side-chain resonances. For larger proteins, such assignments can be difficult to obtain because of extensive resonance overlap in the 2D ‘H‘H correlation spectra. The overlap problem can be substantially reduced by separating homonuclear correlation spectra in a three-dimensional experiment, combining the homonuclear correlation techniques with a heteronuclear multiple-quantum correlation (HMQC) experiment. Such 3D NOESY-HMQC and HOHAHA-HMQC techniques work well for “N-labeled proteins (1-3). Very recently, it has been demonstrated that the 3D NOESY-HMQC technique can also be used for 13C-labeled proteins (4,5). However, separating homonuclear J-correlation spectra according to the chemical-shift frequencies of attached 13C nuclei, as previously used for smaller molecules ( 6, 7)) is inefficient for macromolecules. The main reason for this stems from the severe heteronuclear dipolar line broadening of the protons attached to 13C, making the JHH transfer very inefficient. Recently, we proposed a new three-dimensional NMR experiment, known as HCCH, in which magnetization is transferred between vicinal protons in a three-step manner (8). In the first step, ‘H magnetization is transferred to its attached 13C nucleus via the large and well-resolved ‘J,-coupling ( 140 Hz). In the second step, 13C magnetization is transferred to its 13C neighbor(s) via the relatively large ’ Jcc coupling (33-45 Hz). Finally 13C magnetization is transferred back to ‘H via the ’ Jcn coupling. To maximize sensitivity, this HCCH experiment requires a high level (>90%) of uniform 13C labeling. Since magnetization is transferred through one-bond couplings that to first order are independent of dihedral angles, the transfer efficiency is independent of local geometry, in contrast to conventional homonuclear correlation methods. Because all three individual transfer steps are highly effective, the HCCH experiment transfers a substantial fraction of magnetization from one proton to its vicinal partners, despite the relatively large ‘H linewidth. Although spectra previously obtained with the HCCH method show many of the desired correlations, they were plagued by relatively high levels oft’ noise and spectral artifacts (8). In the present Communication we address the sources of these undesirable features and demonstrate improvements that greatly reduce their intensity.

Improved three-dimensional 1H-13C-1H correlation spectroscopy of a 13C-labeled protein using constant-time evolution.

SummaryAn improved version of the three-dimensional HCCH-COSY NMR experiment is described that correlates the resonances of geminal and vicinal proton pairs with the chemical shift of the13C nucleus attached to one of the protons. The experiment uses constant-time evolution of transverse13C magnetization which optimizes transfer of magnetization and thus improves the sensitivity of the experiment over the original scheme. The experiment is demonstrated for calmodulin complexed with a 26-residue peptide comprising the binding site of skeletal muscle myosin light chain kinase.

Removal of F1 baseline distortion and optimization of folding in multidimensional NMR spectra

We demonstrate an equally effective approach, requiring that the sampling delay equals exactly half a dwell time. This approach will be illustrated for the simplest case of 1D and 2D NMR, but it is successfully used in our laboratory for the processing of 3D and even 4D spectra

The effects of dipolar cross correlation on 13C methyl-carbon T1, T2, and NOE measurements in macromolecules

Abstract The effects of dipolar cross correlation on 13 C T 1 , T 2 , and NOE values are calculated for methyl groups attached to macromolecules. Using the Woessner model to describe methyl-group internal dynamics, it is found that for macromolecules with tumbling times between 5 and 20 ns and fast internal methyl rotation characterized by a correlation time between 15 and 65 ps, the recovery of longitudinal magnetization differs by less than 10% from that of the case with no cross correlation. Therefore for a large range of values of τ m and τ e , cross-correlation effects on longitudinal relaxation are small, despite the fact that reorientation is highly anisotropic. In contrast, for 13 C spin-spin relaxation the effects of dipolar cross correlation are significant for all values of τ m and τ e in the range examined (2 ns τ m τ e 3 spin system attached to a macromolecule. The effect of cross correlation is calculated to change NOE values by no more than 6–7% in the range examined. The influence of neighboring 1 H spins on 13 C relaxation is assessed by including random-field effects in the calculations. While the effects of cross correlation are attenuated, 13 C longitudinal and transverse relaxation rates can, under certain conditions, still be substantially different from rates obtained in the absence of cross correlation. Although a quantitative description of cross-correlation effects on 13 C relaxation depends on the details of methyl-group internal dynamics, the results derived here using the Woessner model are qualitatively the same as results obtained for different descriptions of the methyl-carbon internal motions.