Cell - 8 September 2016

Resonances are characterized by a chemical shift and

couplings of individual spins. The intensity reects the

number of spins in a specic environment (i.e., C-H vs.

C-H 2 vs. CH 3 ). The chemical shifts, couplings, and

intensity provide information on the number and nature

of functional groups present and how they are linked and

are used to deduce the structure for small molecules.

1D spectrum

Spins exchange magnetization during the experiment via

couplings or relaxation controlled by the pulse sequence.

Exchange results in off-diagonal (cross) peaks, here

indicating pairs of coupled spins. Cross peaks are

characterized by two sets of chemical shifts and couplings,

providing information about which spins are coupled or

how close together they are, needed information for

analyzing structure or dynamics.

2D spectrum Spins excited by a pulse sequence can exchange

magnetization more than once during the experiment,

again controlled by the pulse sequence used. Peaks

are characterized by triples (or higher multiples) of

frequencies for spins that exchanged magnetization

during different parts of the experiment. Visualization

of data requires taking “slices” through the data to

identify groups of coupled spins, or a combination of

coupling and proximity.

3+ D spectrum

Interpretation of NMR data

Small compounds: Chemical shifts reect

the local chemical environment of the

nucleus, and coupling constants reect the

number and types of neighboring spins;

together they enable molecular structures to

be deduced. Mass information may be needed

when non-NMR active nuclei are present.

Biopolymers: For proteins and nucleic acids NMR is often used to

analyze folded conformations, binding interactions, and dynamics.

The rst stage of analysis is assigning resonances to specic atoms

in the molecules, done by systematic identication of cross peaks

(correlations) in multidimensional NMR data, often done using^13 C

and^15 N enrichment and correlations among H, C & N nuclei, e.g.,

HNCO, HNCA, HACACO. By comparing sets of coupled spins

linkages between neighboring residues can be identied.

Structure determination

Once many constraints have been determined for a molecule, they can be

used, together with all known features of the covalent structure (bond lengths

and angles determined with crystallography), in a molecular dynamics

optimization calculation. Treating constraints as pseudo-energy terms, and

introducing molecular dynamics to enable efcient sampling of conformations,

a set of structures with good geometry and agreement with constraints is

generated. The degree to which structures cluster is a measure of how well

the NMR data dene the structure and may vary in different parts of the

molecule. Segments that are not well dened, e.g., loops and termini, often

reect dynamic processes that can also be analyzed by NMR (see below).

Interaction site

NMR can map interaction sites, and hence binding

surfaces, using spectral changes upon binding.

Experiments for different kinetic ranges, as for

dynamics, are available.

(1) Ligand binding causes chemical shift changes of

nearby spins in the target. Weak ligands exchanging

rapidly can be followed in ligand titrations, and

mapped using assigned resonances of the free target.

A caveat is that binding-induced conformational

changes may cause shifts in the target away from

the site of direct interaction.

Dynamic analysis—rapid motions

If spin state populations are perturbed, or a coherent state of the spins is

created, the spins are not at equilibrium. Fluctuations in magnetic environment

from molecular tumbling and internal conformational motion couple spins to

their environment and allow them to return to their equilibrium state with time

constants T 1 and T 2 , respectively. Fast motions affect T 1 most, while slow

motions affect T 2. By measuring the relaxation times for spin pairs

(e.g.,^15 N-^1 H or^13 C-^1 H) the rates of molecular motion can be mapped. If

relaxation times are measured at several different magnetic elds, and/or for

different nuclei, then multiple motions affecting a spin pair can be analyzed.

For example, the rate and amplitude of rapid internal loop motion can be

determined in addition to the rate for overall tumbling of a molecule. Rapid is

functionally dened as faster than the overall tumbling, generally a timescale

of 10–50 ns for molecules studied by NMR.

Structural constraints:

Several different NMR experiments can provide information about

local and global structure of a biological molecule.

(1) Protons in close proximity (distances less than ca. 5Å) induce

transitions of neighboring spins—called the nuclear Overhauser

effect (NOE)—creating cross peaks in multidimensional spectra.

Cross peak intensities scale with the inverse sixth power of distance.

Using known distances as calibrations, unknown interproton

distances can be estimated and used as constraints for structural models.

(2) Through-bond coupling constants depend on the dihedral angles,

dened by three bonds. Coupling constant values can be mapped to

dihedral angles. Both 1-bond and 2-bond coupling values can also

be used to constrain structures.

(3) Most of the chemical shift for an individual spin is determined by

covalent structure. However, local structural features (e.g., secondary

structure in proteins) induce systematic shifts that can be interpreted

to dene local conformations.

(4) If molecules are weakly oriented in a sample, e.g., in a gel or liquid

crystal medium, then direct dipolar couplings (through space rather

than through bonds) are not averaged to zero as they are in isotropic

solution. For spin pairs with known bond lengths (e.g., N-H or C-H)

the RDC values provide a constraint on the angle of the bond vector

in a molecule xed axis system.

Dynamic analysis – slow motions

(1) Moderately slow motions (in the μs to ms range) affect

the T 2 relaxation but not T 1. With special pulse sequences,

the slow processes can be probed in detail and the rate of

exchange and the chemical shifts of the exchanging states

can be derived. Such conformational exchange can

correspond to functionally signicant motions around the

active sites of enzymes, for example.

(2) Conformers exchanging even more slowly give rise to

separate peaks. Exchange between them can be detected

as magnetization transfer between the resonances for the

same spin in the two states when the exchange rate is of

the order of the T 1 relaxation rate (~100 ms to 1 s timescale).

Multidimensional experiments can be used to correlate

shifts of spins in the different conformers.

(3) Rare (infrequent) events can be detected if something happens that changes the NMR

parameters during the event. An example is exchange of^2 H from solvent for^1 H on an amide

of a protein that requires opening of the structure (usually breaking hydrogen bonds). The

open state lives only a very short time, so populations of the open states are very low, but

the^1 H intensity reects the integrated probability that an exchange event occurred.

(2) NOEs can be detected between the ligand protons and nearby target protons.

Distinguishing these from other NOEs is best done by differential labeling, e.g., having 15

N-H or^13 C-H on the ligand but not on the target (or vice versa). The NOE spectra

can be “ltered” to select only the intermolecular NOEs, which identify the binding site.

(3) If the ligand carries a paramagnetic center, such as a metal or nitroxide, then the

position of that group relative to the target spins can be mapped by observing changes

in their relaxation rates (termed paramagentic relaxation enhancement, PRE). Such

relaxation effects may be quite strong, so even low populations of the complex (as

occur for weakly binding ligands) can be studied.

INTENSITY

COUPLING CONSTANT (Hz)

Initial

slope

Mixing time θ Angle (degrees)

(^0180)
Hθ
C C H
rij
∝τ 6
1D NMR data are collected by applying a strong radio frequency pulse near the spins’
frequency to induce a coherent state of the spins and then measuring the transient
response. The Fourier transform of the response gives the normal absorption spectrum.
If more than one pulse is applied, then correlation information is generated.
Multidimensional Pulse-Fourier Transform NMR Spectroscopy
By varying delays between pulses, multidimensional data are generated. The resulting
peaks have frequencies in different dimensions reporting the spins involved in the the
correlations. Correlations can be created using either spin-spin couplings or mutual
relaxation through dipolar couplings, reecting proximity.
(^1) H
(^1) H
(^1) H
(^15) N
t 1 t 1
t 1 t 3
t 2
5.0 4.0 3.0 2.0 1.0 t^2
ppm
3.0
D2 =^15 N
D1=^1 H
D3 =^1 H
5.0 4.0ppm
CA CA
CA
CB
CB
HA
HA
C N C N C C
H
H
H
N
O
O
O
O O
O
S CH 3
R
O N O
H NH N NH
H 3 C H 3 C
OH OH
N
C
τc
τi
N
H
CBHA
8.0 7.0 6.0

SnapShot: Biomolecular NMR

David Wemmer
Department of Chemistry, University of California, Berkeley, Berkeley CA, 94720, USA

See online version for
1600 Cell 166 , September 8, 2016 ©2016 Elsevier Inc. DOI http://dx.doi.org/10.1016/j.cell.2016.08.061 legend and references.