News & Views | Published:

Structural biology

Protein dynamics from disorder

Nature volume 459, pages 10631064 (25 June 2009) | Download Citation

Subjects

The functions of proteins are often crucially dependent on how they move, but measuring the absolute magnitudes of protein motions hasn't been possible. A spectroscopic method looks set to change all that.

In the half-century since our first glimpse of a protein structure at atomic resolution, structural biology has had a fundamental role in advancing our understanding of biological function at a molecular level. Yet although it is generally agreed that proteins must have some flexibility in order to perform their roles, uncertainty remains about how well static structural models capture the essential features underlying function. A picture may be worth a thousand words, but does it tell the entire story?

This is a surprisingly difficult question to answer unambiguously, because the picture of protein dynamics available from established experimental methods is incomplete. But Salmon et al.1 report in Angewandte Chemie the absolute magnitude of motions spanning the entire submillisecond timescale for the peptide backbone of the protein ubiquitin. In effect, this work provides a quantitative measurement of how 'blurry' the picture of a protein is as a result of that protein's motion.

The requirement for protein flexibility was recognized early on by Max Perutz, who noted that there was no obvious route for ligand molecules to enter the active site of haemoglobin2. Since then, the need for protein flexibility has been demonstrated repeatedly by observations that some proteins adopt different structures when in complex with different ligands, or as a result of differences in the conditions used to crystallize them. Yet efforts to obtain a direct, quantitative description of protein dynamics in solution have generally suffered from limited spatial resolution or timescale sensitivity. It is now well established that all proteins undergo fast, picosecond-timescale motions of limited amplitude, but little is known about their motions on longer timescales (with the exception of a few specific cases on microsecond or slower timescales that involve large-scale motions)3,4.

Nuclear magnetic resonance (NMR) spectroscopy is a powerful tool for studying the structure and dynamics of small proteins at atomic resolution. When this technique is used to study protein molecules that are weakly aligned in the magnetic field of the spectrometer, previously unobservable interactions, such as couplings between nuclear magnetic dipoles, manifest themselves as a splitting of the peaks in the spectrum. The splitting is proportional to the degree of alignment.

In the absence of dynamics, these so-called residual dipolar couplings (RDCs) between covalently bonded pairs of atoms depend on only two things: the extent of the molecule's overall alignment with the magnetic field, and the projection of the bond onto the direction of maximum molecular alignment with respect to that field (Fig. 1a). In the presence of dynamics, the movements of the bond occurring on a submillisecond timescale act as a source of disorder, diminishing the effective magnitude of alignment for that bond (Fig. 1b). But this second case is complicated because one cannot readily distinguish between the orientational and dynamic components of RDC measurements.

Figure 1: Molecular alignment enables the detection of protein dynamics.
Figure 1

Weak alignment of dissolved protein molecules in a magnetic field leads to the incomplete averaging of dipolar interactions between pairs of atoms, represented by the numbered vectors. These unaveraged interactions appear in NMR spectra as residual dipolar couplings (RDCs) — splitting of peaks in the spectrum that scale proportionally with the alignment strength, and that depend on the individual projections of dipolar interaction vectors onto the direction of maximum molecular alignment. a, In static molecules, RDC measurements of covalently bonded pairs of chemically identical nuclei depend only on the angle between the direction of maximum alignment and the specific bond. b, For dynamic molecules, submillisecond molecular motions introduce local disorder, which appears as an apparent reduction in alignment strength, as indicated by the shortened vectors. The resulting convolution of orientational and dynamic contributions to the RDC measurements can be untangled by performing several experiments in different solvents that cause the molecules to align in different ways. Salmon et al.1 use this deconvolution approach in their method for analysing protein dynamics.

Fortunately, the orientational and dynamic components of RDCs can be separated by collecting several sets of data, with the protein aligned differently for each set5,6. But although this allows the relative magnitude of motions to be established, the absolute magnitude remains undetermined because of uncertainty in the overall magnitude of molecular alignment. Any underestimation of the magnitude of molecular alignment translates into an underestimation of the amplitude of internal motions.

Salmon et al.1 report a method that overcomes this problem, and that allows the absolute amplitude of submillisecond motions of the peptide backbones of proteins to be established. Their approach uses RDC data from different bonds within each peptide unit, enabling the orientation and asymmetric motions of each individual peptide unit to be determined. Because of the asymmetry of the peptide motions, RDC data aren't compatible with any arbitrarily chosen magnitude of molecular alignment. The authors therefore determine the absolute amplitudes of peptide motions by using a cross-validation analysis in which the goodness of fit of the determined dynamic and orientational parameters with the overall magnitude of molecular alignment is gauged by its agreement with some fraction of the RDC data that has been excluded from the analysis.

The authors applied their method to ubiquitin, an unusually stable protein that lacks a catalytic function but binds to many other proteins. For much of the protein, the newly determined motional amplitudes1 are in general agreement with the known amplitudes of picosecond dynamics previously obtained by NMR7. The newly determined amplitudes also agree well with a molecular-dynamics simulation of the protein performed by the authors to validate their results.

More excitingly, using their technique, Salmon et al. observed motions that occur on a longer timescale (> 400 nanoseconds) than could be accessed by the molecular-dynamics simulation, specifically for a β-turn region and for a loop that lies close to ubiquitin's protein-binding interface. These motions could not have been observed using previously available NMR methods. The authors' results reinforce a picture of ubiquitin as a protein that has a rigidly packed, stable fold in which most of the flexibility is concentrated in regions responsible for its functional roles — protein recognition and assembly into covalently linked polyubiquitin chains8,9,10 that serve as cellular signals. This defuses a long-standing controversy about whether globular, non-enzymatic proteins such as ubiquitin undergo any motions on timescales longer than nanoseconds.

Salmon and colleagues' method1 opens the door to studies of other proteins that are expected to be more flexible than ubiquitin. At present, the main obstacle to using the authors' technique is that many different solutions of the protein must be prepared, each providing environments that cause the protein molecules to align differently. This can be quite challenging. Nevertheless, the method enables a straightforward average of the atomic coordinates of proteins to be recorded for movements occurring over the full submillisecond timescale, which clearly has important implications. Knowledge of the extent of protein movements could enable the determination of ensembles of structures that represent the conformational fluctuations of proteins, help to guide the design of ligands, and advance our understanding of protein catalysis and allostery (the regulation of protein activity by ligand binding to sites other than the functional binding site).

Dynamics is often used to explain apparent discrepancies between structural models of proteins and their known functions, yet there are few cases in which direct links between dynamics and function have been made. As experimental techniques for determining dynamics continue to improve, we can look forward to gaining deeper insight into the question of whether dynamics is more often essential or extraneous to protein function.

References

  1. 1.

    et al. Angew. Chem. Int. Edn 48, 4154–4157 (2009).

  2. 2.

    & J. Mol. Biol. 21, 199–202 (1966).

  3. 3.

    Annu. Rev. Biochem. 77, 45–50 (2008).

  4. 4.

    & Science 312, 224–228 (2006).

  5. 5.

    , , , & J. Am. Chem. Soc. 123, 6098–6107 (2001).

  6. 6.

    J. Am. Chem. Soc. 124, 12020–12030 (2002).

  7. 7.

    , , & J. Am. Chem. Soc. 117, 12562–12566 (1995).

  8. 8.

    & J. Am. Chem. Soc. 125, 10164–10165 (2003).

  9. 9.

    et al. Science 320, 1471–1475 (2008).

  10. 10.

    , , & J. Am. Chem. Soc. 124, 5822–5833 (2002).

Download references

Author information

Affiliations

  1. Joel R. Tolman is in the Department of Chemistry, Johns Hopkins University, 3400 North Charles Street, Baltimore, Maryland 21218, USA.  tolman@jhu.edu

    • Joel R. Tolman

Authors

  1. Search for Joel R. Tolman in:

About this article

Publication history

Published

DOI

https://doi.org/10.1038/4591063a

Further reading

Comments

By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.

Newsletter Get the most important science stories of the day, free in your inbox. Sign up for Nature Briefing