Early Strides in NMR Dynamics Measurements

Abstract

The study of protein dynamics using the measurement of relaxation times by NMR was based on a set of studies in the mid-20^th century that outlined theories and methods. However, the complexity of protein NMR was such that these simple experiments were not practical for application to proteins. The advent of techniques in the 1980s for isotopic labeling of proteins meant that pulse sequences could now be applied in multi-dimensional NMR experiments to enable per-residue information on the local relaxation times. One of the earliest advances was published in Biochemistry in 1989. The paper “Backbone dynamics of proteins as studied by ¹⁵N inverse detected heteronuclear NMR spectroscopy: application to staphylococcal nuclease” by Lewis Kay, Dennis Torchia and Ad Bax, delineated a set of pulse sequences that are used with minor modifications even today. This paper, with others from a limited number of other labs, forms the basis for the experimental determination of the backbone dynamics of proteins. The biological insights obtained from such measurements have only increased in the last 30 years. Sometimes the best and perhaps only way to advance a field is an advancement in the technical capabilities that allows new perspectives to be reached.

Graphical Abstract

Using relaxation times measured by NMR to infer information about the local motions of molecules has a long history. Beginning in the 1950s and 60s, physicists developed the underlying theory.^1–4 Applications to biological molecules were slow to come, as the NMR methods in the 1960s did not include either FT methods or multi-dimensional spectroscopy. Even with the addition of two-dimensional proton spectroscopy, the complexity of the spectra precluded application of spin relaxation methods to molecules of sizes greater than a few amino acids. The use of ¹³C and ¹⁵N provided a tangible advantage in spectral range,^5–7 but was limited due to the low sensitivity of these nuclei compared to protons.

The widespread use of isotopically-labeled proteins was well under way in the late 1980s.^8–10 Uniform isotopic labeling with ¹³C and ¹⁵N not only allows for additional dimensions with spectral characteristics widely different from the overlapped proton spectra, but also allows advances in indirect detection, where the sensitivity of the proton was leveraged by transferring the magnetization from the heteronucleus to the attached proton for detection. The stage was set for the study of heteronuclear relaxation with residue specificity, the point at which the paper “Backbone dynamics of proteins as studied by nitrogen-15 inverse detected heteronuclear NMR spectroscopy: application to staphylococcal nuclease” by Lewis Kay, Dennis Torchia and Ad Bax (Biochemistry 28, 8972–8979, 1989)¹¹ was published.

The paper begins by pointing out the utility of relaxation measurements for the estimation of the internal dynamics of proteins. Molecular dynamics calculations showed that proteins have the potential to undergo picosecond-nanosecond fluctuations in solution, a time scale that is sampled by the R₁ and R₂ relaxations of the ¹³C and ¹⁵N heteronuclei and the heteronuclear NOE. The authors point out that while such measurements had been made in the past, they were limited to one-dimensional measurements on natural-abundance or singly-labeled samples, and by the sensitivity of the heteronuclear detection. This paper was also the first to point out the higher sensitivity to ps-ns motions of the heteronuclear NOE, compared to R₁ or R₂. The paper uses a uniformly-labeled monomeric protein of molecular weight 18 kDa, Staphylococcal nuclease, whose resonances had previously been assigned.¹² Pulse sequences are presented that take advantage of uniform isotopic labeling and the use of indirect ¹H detection. Magnetization must be transferred between the heteroatom and the attached proton; this is achieved using two INEPT sequences,^13,14 which “are employed to transfer magnetization from the directly bound protons to the low γ heteronucleus and back to protons for detection. In this way the detected signal intensity is independent of the gyromagnetic ratio (γ) of the heteroatom, providing an almost 30-fold increase in sensitivity relative to ¹H-¹⁵N correlations recorded with ¹⁵N detection (Bax et al., 1989a).”^11,15

True to the tradition of papers published in Biochemistry, this paper provides a comprehensive analysis not only of the NMR spectroscopic techniques and methods, but also of the treatment of the data and the biochemical conclusions that can be derived from the relaxation data. The authors give a succinct description of the methods of Abragam⁴ and Lipari and Szabo^16–18 to estimate the overall correlation time τ_m and the order parameter S², a measure of the amplitude of the local motion. Although the authors assume that the ¹⁵N relaxation of the backbone NH is governed primarily by the ¹H-¹⁵N dipolar interaction and the chemical shift anisotropy, they note that chemical exchange can affect the R₂ relaxation rate and that the addition of a series of pulses, termed a Carr-Purcell-Meiboom-Gill^19,20 (CPMG) pulse train in the T₂ measurement can minimize the effects of chemical exchange and other interactions. The CPMG sequence is the one now routinely used to quantify the effects of chemical exchange and give direct information on the biologically-relevant microsecond-millisecond time-scale motions of proteins.²¹

If the three-dimensional structure of the protein is known (it may be determined by X-ray crystallography or by NMR), then the per-residue dynamic behavior can be directly mapped onto the structure, to give insights into the regions of the protein where local motion is predominant. In the case of Staphylococcal nuclease (S.Nase), the protein studied by Kay et al.,¹¹ the predominance of secondary structure would raise the expectation that the molecule would be relatively rigid. Nevertheless, there are several long loops in S.Nase. The S² values from the paper¹¹ are mapped onto the X-ray crystal structure of the protein (PDB 1STN²²) in Figure 1.

Figure 1.

A. Cartoon of the X-ray crystal structure of Staphylococcal nuclease (PDB 1STN¹⁹), colored from blue at the N-terminus to red at the C-terminus. Secondary structure elements are labeled. B. Plot of S² as a function of residue number (from Kay et al.¹¹), with secondary structure elements shown. The horizontal line in each panel represents the average S² calculated for the whole molecule (except for Leu-7, His-8 and Gly-148). Figure 1B adapted with permission from reference¹¹.

The majority of the data shown in Figure 1B lies within 10% of the mean S² value (0.86¹¹). As stated in the paper, the interesting conclusion here is that the loops are as rigid as the helices and sheets, which means that the loops are likely tethered to the core of the protein through hydrogen bonds or other interactions.

This paper¹¹ represents one of the first attempts to quantify backbone motion on a per-residue basis. Many subsequent papers have explored the ps-ns dynamics of a wide variety of proteins, giving important information on, for example, the molecular mechanisms of enzymes.²³ The accurate measurement of slower time-scale processes using relaxation dispersion measurements has provided an even richer source of information on enzyme mechanism.^24,25

A measure of the impact of this paper on the field is given by its citation list, currently numbering over 1600. Recent papers citing Kay et al.¹¹ include studies from a wide variety of areas: disordered proteins,²⁶ protein folding mechanisms,²⁷ phase separation,²⁸ ligand binding,²⁹ chaperone behavior,³⁰ protein aggregation,³¹ and motor protein mechanism.³² The fundamental groundwork for multi-dimensional NMR relaxation measurements with indirect detection, leading to mechanistic interpretation of molecular motion, was laid down in seminal papers such as the subject of this perspective.