NMR: Practicalities and
Applications
BIOC 530
Learning Goals
• Have a better understanding of NMR data
in publications
• Determine how/ if NMR can be useful in
your project
3D structures by solution NMR
SH3 domain ensemble: Nature Methods 7, S42 - S55 (2010)
Why use NMR to solve structures?
3D structures by solution NMR
Steps:
1. Over-expression, isotope enrichment, and purification of
the protein of interest;
2. Sample optimization of buffer, salt, and pH to maximize
protein solubility and minimize oligomerization;
3. Assignment of side-chain and backbone NMR
resonances;
4. Determination of structural constraints: NOE (Nuclear
Overhauser Enhancement) cross-relaxation rate
constants), J-coupling constants, RDCs (residual
dipolar couplings), and isotropic chemical shifts;
5. Structure calculation and refinement; structure
validation.
1) Protein sample preparation
• Overwhelming majority of the proteins studied by NMR
are over-expressed in and purified from E. Coli
• M9 (minimal media) with 13C-enriched glucose and 15Nenriched ammonium chloride as sole carbon and
nitrogen sources is used for 13C/15N labeling
• E. Coli growth in D2O is used to introduce deuterium into
non-exchangeable protein sites. Partial deuteration is
useful for NMR studies of proteins > 25 kDa
• Insect cell medium and in-vitro translation systems
enriched with stable isotopes are available; but still
prohibitively expensive
2) Optimization of sample
conditions
• Buffers with non-negligible temperature dependence of
pH (e.g. Tris) should be avoided.
• pH < 7 is preferred, as it minimizes the loss of 1H
sensitivity due to exchange with water protons.
• The protein must be in a well-defined oligomeric state
• 0.5 - 1 mM is the optimum protein concentration for
structural and dynamical studies
• The NMR sample should be stable over periods of time
required to collect the NMR data
– days > binding studies
– weeks > assignments or dynamics
– months > all atom assignments / full dynamics characterization
3) Assignments
4)Structural restraints in NMR
• Short Distance Restraints
– NOEs (from crosspeak intensities in NOESY spectra)
• Dihedral angle restraints/ secondary structure
– J-coupling constants, Chemical Shifts
• Long Distance Restraints
– Residual dipolar couplings (RDCs), paramagnetic spin labels
• Hydrogen Bond restraints
– Amide (NH) hydrogen/deuterium exchange rates, temperature
dependence
NOE distance constraints: 1H
NOESY spectra
Cross-peaks
• NOE cross-peak intensities relate to the distance between 1H in space.
• Can also be done with filtering to obtain 15NH or 13CH NOEs etc.
Structural restraints: NOEs
• Initial rate approximation: NOE cross-peak intensities are
proportional to the cross-relaxation rate constants
• Cross-relaxation rate constants are proportional to (1/r6),
where r is the distance between the two 1H that are close in
space
tyrosine
rref(Sref/Si)1/6
ri =
where Si and Sref are
the volumes of crosspeaks measured in
NOE spectra. “ref”
stands for the reference
proton pair with a welldefined geometry
d-e distance
always the same = 2.5 Å H
H
Short distances found in protein
structures
Structural restraints: NOEs
• Hundreds of NOEs are required to obtain a good
NMR structure
• You don’t want to use the exact distance
calculated as a restraint. Instead provide a range.
class
restraint
description
*for protein w/ Mr<20 kDa
strong
medium
weak
1.8-2.7 Å
1.8-3.3 Å
1.8-5.0 Å
strong intensity in short mixing time NOESY
weak intensity in short mixing time NOESY
only visible in longer mixing time NOESY
Structural constraints: Dihedral
angles from J-coupling constants
Structural constraints: Dihedral
angles from J-coupling constants
Horst Joachim Schirra's PPS2 project
Measuring 3JHN-Ha: 3D HNHA spectra
ratio of crosspeak
and diagonal peak intensities
can be related to 3JHN-Ha
J large
J small
HN to Ha
crosspeak
HN diagonal
peak
this is one plane of a 3D spectrum
of ubiquitin. The plane
corresponds to this 15N chemical
shift
Archer et al. J. Magn. Reson.
95, 636 (1991).
Structural restraints – another option
for secondary structure
• CSI (chemical shift index) - establishes the secondary
structure of proteins based on chemical shift differences
with respect to some predefined “random coil” values. It
can be applied from the measured HA, CA, CB and CO
chemical shifts for each residue in a protein.
0 = random coil
chemical shift
Structural restraints: bond
orientations
• Residual dipolar couplings (RDCs)
1. Intrinsic anisotropy
2. External liquid crystalline
medium (sterics and/or charge)
•
Bicelles
•
Phage
•
Polyacrylamide gels
•
C12E5 PEG + hexanol
Structural restraints: RDCs
• Measured for a pair of covalently-linked NMR-active
nuclei in partially aligned molecules
• Examples: 15N-1H, 13Ca-15N,13CO-15N RDCs
• RDCs depend on the orientation of the bond vector
relative to the molecular alignment frame
B
Aligned sample splitting = JNH+DNH
θ
H
r
N
DNH =
ħ gN gH
4 p rNH3
(1 – 3 cos2q)
PREs
• long distance restraints – 15-24Å
Paramagnetic DNA or Membrane
Chem. Rev. 2009, 109, 4108–4139
Step 5: Structure calculation and
refinement
• NMR data do not uniquely define the 3D structures,
because the restraints are included as a range of
allowed values
• A good “ensemble” of structures minimizes violations of
the input restraints and the RMSD between the members
of the ensemble
Step 5: Structure calculation and
refinement
NMR
experimental
data
Experimental
restraints
Assignment and
conversion
Structure
ensemble
Structure calculation
and selection
Restraint violation
and error analysis
Validated
structure
data
Structure quality
checks and statistics
“Validation of NMR-derived protein structures”, Chris Spronk, Centre for
Molecular and Biomolecular Informatics, University of Nijmegen,
The Netherlands.
Ensembles of NMR structures
Stereo view of the
conformational ensemble
• The precision of an NMR
structure is determined by
the root-mean-square
deviation (RMSD)
between the backbone
atoms of the
conformational ensemble
Stereo view of
representative structures
NMR structure of RNA-binding SAM (sterile alpha motif) protein,
from Edwards et al, J. Mol. Biol. (2006), 356, 1065-1072.
Making structural models with
limited NMR data
• CS Rosetta - Makes empirical correlations
gained from mining chemical shift data
deposited in the BMRB (Biological Magnetic
Resonance Bank) database
• structure prediction based on chemical shift
data, chemical shift perturbations, PREs, RDCs,
SAXS using ROSETTA
• Fragment based modeling based on known
structures in PDB
Conformation b RDC (Hz)
Limited data refinement example from a zinc
coordinating kinase regulatory domain
Conformation a RDC (Hz)
B
Aligned sample splitting = JNH+DNH
θ
H
r
N
DNH =
ħ gN gH
4 p rNH3
(1 – 3 cos2q)
Limited data refinement example from a zinc
coordinating kinase regulatory domain
NMR is good for more than just
structure determination!
Timescales of binding in NMR
kex>>
A
k1
k-1
Fast exchange
B
kex=
kex=k1 + k-1
kex<<
Frequency (Hz)
Slow exchange
Titration of a membrane bound second
messenger, diacylglycerol, into a signaling protein
Wild-type signaling protein
Fast exchange
Tighter binding mutant
slow exchange
Titration of a membrane bound second
messenger, diacylglycerol, into a signaling protein
Wild-type signaling protein
Fast exchange
Tighter binding mutant
slow exchange
pH dependent conformational exchange
Protonation = fast
Conformational
exchanage = slow
Double-stranded RNA binding innate immune protein
Studying ligand binding in a large
unassigned protein
Met 572
• Voltage gated K+ channel
(HCN2)
• Heart - pace making
• Brain - chronic pain
• Two activating ligands
cAMP
cAMP
fisetin
Carlson et
al. (2013)
13C-HSQC
resonances
13C-HSQC
methyls
13C-HSQC
of HCN2
M572
Carlson et al. (2013)
Assignment by mutagenesis
M572T
Carlson et al. (2013)
Met Scanning TechniqueStructure 20: 4, p573–581, 4 April 2012
Time scales for protein dynamics
Side Chain Rotations
Folding
Loop and Domain Motions
Enzyme Catalysis
10-12
10-9
R1, R2, NOE,
Cross-correlation
10-6
10-3
100 s
1H exchange
CPMG, R1
Lineshape analysis
Magnetization transfer
Atomic resolution
Multiple timescales
NMR relaxation parameters
• Steady-state hetero-nuclear NOE
• Transverse (or spin-spin) relaxation rate
constant, R2
– Loss of phase coherence in the transverse plane
• Longitudinal (or spin-lattice) relaxation rate
constant, R1
– Relaxation to Z plane
• All 3 depend on the motion
of your protein in solution
z
y
x
ω = γB
– Size/ shape
B0
– Hydrodynamic radius
– Unstructured regions/ multiple domain proteins
0
Characterization of conformational
exchange = Rex
R2=R20+Rex
A
k1
k-1
B
Invisable Population
kex pA pB ΔωAB2
Rex =
ncpmg2 + kex2
kex = kinetic rate constant = k1 + k-1
pA & pB = populations of A and B
ΔωAB2 = difference in chemical shift
Relaxation Dispersion Experiments
ncpmg
Relaxation dispersion experiments
R2  R  Rex
0
2
kex pA pB ΔωAB2
Rex =
ncpmg2 + kex2
The information about microsecondmillisecond protein dynamics is buried
in Rex. By fitting relaxation dispersion
curves one can extract the ratio of
exchanging populations, chemical shift
differences () between the
exchanging conformers and kinetic
rate constants (kex).
Protein dynamics in enzymatic
catalysis
• Substrate binding and product release are often
accompanied by conformational changes that could be
rate-limiting for the catalytic reaction.
• The catalytic step itself is dynamic, as it involves
fluctuations of atomic coordinates during bond breaking or
formation.
• Examples of some of the protein systems where dynamics
proved to be crucial for catalysis: dihydrofolate reductase,
cyclophilin A, and adenylate kinase.
Dynamic energy landscape of
DHFR
• Boehr et al, Science 313,
1638 (2006)
DHF
• There are 5 known
intermediates; the structures
of these intermediates (or
their models) have been
determined by X-ray
crystallography.
THF
Conformational changes during the
DHFR catalytic cycle
Relaxation dispersion data for each
DHFR intermediate
Red: active site conformation; blue: cofactor binding; and green:
substrate/product binding
Correlation between the chemical shift data
obtained from relaxation dispersion analysis
with that of ground state conformations
From fitting relaxation
dispersion curves
Difference in 15N shifts of
ground state from spectra
Dynamic energy landscape of
DHFR
From fitting relaxation
dispersion curves
From enzyme
kinetics assays
Reference: Boehr et al, Science 313, 1638
(2006)
Side Chain Rotations
Folding
Loop and Domain Motions
10-12
10-9
R1, R2, NOE,
Cross-correlation
10-6
10-3
100 s
1H exchange
CPMG, R1
Lineshape analysis
Magnetization transfer
Solid-state NMR
Solid-state NMR: advantages
• Isotropic-like NMR
spectra with site
resolution
• No solubility problem
• No “tumbling time”
problem
Kaliotoxin-K+ channel interactions
kaliotoxin
K+ channel
• The chemical shifts of kaliotoxin are perturbed as a
result of binding to K+ channel.
Lange et al, Nature (2006), 440, 959-962
Kaliotoxin-K+ channel interactions
Solid-state
structure of
kaliotoxin bound
to K+ channel
Residues whose chemical shifts are
perturbed as a result of binding are
colored red.
Lange et al, Nature (2006), 440, 959-962
Kaliotoxin-K+ channel interactions:
looking at K+ channel
kaliotoxin
K+ channel
• Perturbed and unperturbed
residues of K+ channel are
shown in red and blue,
respectively.
Lange et al, Nature (2006), 440, 959-962
Structural model of kaliotoxin-K+
channel kaliotoxin
• High-affinity binding of
kaliotoxin is accompanied
by an insertion of K27
side-chain into the
selectivity filter of the
channel;
• The binding is associated
with conformational
changes in both
molecules.
Lange et al, Nature (2006), 440, 959-962
K+ channel
selectivity filter
A couple good reviews for your
reference
• An introduction to NMR-based approaches
for measuring protein dynamics:
Biochimica et Biophysica Acta 1814
(2011) 942–968
• Mapping Protein-Protein Interactions in
Solution by NMR Spectroscopy:
Biochemistry 41:1 (2002)