Structural Biology: Chapter 4 — NMR of macromolecules

🔬 Key Illustrated Scientific Figure

Full-Screen Zoom

Click to enlarge on this page

4.1 What the spectrum reports #

4.1.1 Spin and chemical shift #

Spin-1/2 nuclei, chiefly 1H, 13C, and 15N, are the reporters. The chemical shift, in parts per million, is the resonance frequency relative to a reference, so it does not depend on field strength. Magnets and pulses are Biophysical Techniques. The shift depends on the local electrons: for the α proton, a value below the random-coil position is typical of helix and a value above it of a β strand. Chemical-shift perturbation marks residues whose environment changes on binding. It does not prove a direct contact, because allostery shifts distant residues too. 13C and 15N are enriched because they are scarce at natural abundance, and deuteration removes protons that would broaden the rest. Enrichment is a fact about the sample, not a protocol given here.

4.1.2 Through bonds and through space #

Scalar J coupling travels through bonds, groups the spins of one residue, and walks the backbone from residue to residue. The coupling between amide and α proton also depends on φ, so it restrains a dihedral angle. Through-space experiments give the NOE. Covalent geometry fixes the local chemistry; short distances say which remote parts of the chain are close. Neither is an image. Both are restraints.

4.2 Building an ensemble #

4.2.1 The NOE as a distance restraint #

The NOE transfers magnetisation through a dipole–dipole coupling and falls as the sixth power of distance. A pair near 5 Å is much weaker than the same pair near 3 Å, and by about 6 Å the effect is generally unusable. NOEs are therefore short upper bounds, often binned rather than treated as exact rulers. Two well-defined domains can still be wrongly placed if no restraint crosses the gap. Ambiguous peaks, many protons at similar shifts, are tested by whether an assignment fights the rest of the data. A paramagnetic tag can reach roughly 20 Å; that is not an NOE, and the classical restraint remains the short one.

4.2.2 Why the result is an ensemble #

The deposit is a bundle of models, often ten to twenty, consistent with the restraints and with covalent geometry. A tight region with many restraints is pinned down. A splayed region may be mobile or merely under-determined. Precision is not accuracy: a consistent misassignment can be tight and wrong. Dynamics are measured separately. Order parameters report picosecond-to-nanosecond motion; relaxation dispersion reports microsecond-to-millisecond exchange; slower events appear as hydrogen exchange or as two peak sets. The pulse sequences are not this course. In a weakly aligning medium, dipolar couplings do not average exactly to zero, and the residual dipolar coupling reports the orientation of a bond relative to a common frame, which is long-range angular information that local NOEs do not provide. The bundle, not a naive average of it, is the result.

4.3 How large, and when to choose it #

4.3.1 The size limit, TROSY, and methyl labelling #

Slow tumbling broadens lines, and overlap grows with the number of peaks. A full structure of a protonated protein is routine below about 25 kDa and progressively harder towards 40–50 kDa. TROSY selects the slowly relaxing component of a 15N–1H signal, where two relaxation paths partly cancel, and extends backbone spectra well past that limit. It does not by itself assign the chain. Methyl labelling keeps isoleucine, leucine, and valine methyls protonated in an otherwise deuterated protein. Methyl-TROSY spectra then reach assemblies of several hundred kilodaltons, including the proteasome, usually as probes of binding and motion rather than as a complete all-atom model. Solid-state NMR, with magic-angle spinning, can structure specimens that do not tumble, including amyloid fibres and some bilayer samples.

4.3.2 When NMR is the better method, and when it is not #

NMR wins in solution, without a crystal, for disorder (sharp random-coil signals rather than missing density), for dynamics on a stated timescale, and for weak complexes. Concentrations are often about 0.1–1 mmol per litre, so a weak interaction can be over-represented relative to the cell. It loses when the goal is a first all-atom model of a large asymmetric assembly, where cryo-EM has the mass on its side, and it loses to a good crystal on an ordered water or a hole in a ring. It is impractical if the protein cannot be labelled or kept soluble. Wüthrich’s solution structures, recognised in the 2002 Nobel prize, put the method beside crystallography. A disagreement between them is data.

4.4 Key terms, summary, and self-check #

Key terms. Chemical shift; J coupling; NOE; ensemble; residual dipolar coupling; TROSY; methyl labelling; chemical-shift perturbation.

Summary. Chemical shifts report local conformation, and J couplings assign atoms and restrain φ. The NOE falls as the sixth power of distance and reaches about 5–6 Å. The deposit is a bundle, mixing dynamics with missing restraints, so motion is measured separately. A residual dipolar coupling adds long-range bond orientation. Above a few tens of kilodaltons, traditional structures fail; TROSY and methyl labels extend spectra, usually as probes of large particles rather than as complete models. NMR is the choice for solution dynamics, disorder, and weak binding.

Self-check.

1. Why is an NOE a distance restraint, and about how far is it useful?

2. Why is the deposit a bundle, and why is a tight bundle not automatically correct?

3. In one sentence, what does a residual dipolar coupling add?

4. What does molecular mass do to the lines, and what did TROSY and methyl labelling each change?

5. Name one question NMR answers better than crystallography, and one it does not.

—

—

—