Transcript Applications of high-field NMR

(Bio)-applications of highfield NMR
1
Aims
• To give an overview of the capability of NMR
to answer biological questions
• To make aware of limitations
• To give a basic idea about structure
determination by NMR
• To make aware of NMR sample requirements
• To enable you to decide whether NMR would
be a useful method in your research
2
Outline
• Introduction to biological applications
of NMR
• Basics of solution structure
determination of proteins
• Heteronuclear NMR
• NMR of nucleic acids
• NMR and dynamic phenomena
• (some more applications)
3
What can NMR do for biology ?
• 3D Structure determination of proteins and
nucleic acids
• Assess stability and folding of proteins
• Binding studies (Proteins, DNA, Ligands)
• Protein dynamics and “reactions”: possible
to look at timescales between ps and days
• Elucidation of structure of biomarkers,
metabolites, and synthetic pathways
• NMR of bio-fluids and tissues
• In vivo NMR
• Magnetic Resonance Imaging
4
3D Structure determinations
GSDIIDEFGTLDDSATICRVCQKPG
DLVMCNQCEFCFHLDCHLPALQD
VPGEEWSCSLCHVLPDLKEEDVDL
QACKLN
Protein sequence
Acquire NMR spectra
Express and
purify protein
(or isolate from
natural source)
Initial characterisation
- Identity, composition
- Concentration
- Stability (buffers, salt,
pH, temperature)
Evaluation:
3D structure
Sequential Assignment
Extraction of distance restraints
and other structural data
5
3D Structure determinations
1. The sample
In solution:
•
•
•
•
•
ca. 0.2-1 mM protein solution (ca. 200-500 mL)
Smaller than 35 kDa
Preferentially in native form, not aggregated....
Usually nothing paramagnetic (e.g. Cu(II), Fe(II)
or Fe(III), …
Recombinant expression necessary for proteins
> 8kDa (for isotopic labelling with 13C and 15N)
6
3D Structure determinations
2. The spectra
7
Fourier Transform pulse sequences
• The simplest 1D experiment:
1. Radiofrequency pulse with high power
2. Recording of the free induction decay (FID)
0.10
0.10
0.20
0.20
0.30
0.30
0.40
0.40
0.50
0.50
0.60
0.60
0.70
0.70
Acquisition
Repeat - but need to make sure that
excitation from previous scan has
completely vanished  relaxation delay
8
1D NMR
“Free induction
decay” (FID)
0.10
0.20
0.30
0.40
0.50
0.60
0.70
Time domain
(s)
Fourier Transformation
Frequency domain
1D NMR spectrum
(s-1)
9
Typical 1H NMR spectrum of a small
molecule
Recorded in 90% H2O/10% D2O
8H
aliphatic
16H (aromatic)
H2O
Low
field
10
High
field
4H
9
8
7
6
5
4
3
2
1
d 1H (ppm)
Aromatic protons are affected by electron cloud (“ring current”) of aromatic
ring (deshielded – the field experienced by aromatic protons is weaker than
10
B0, consequently the resonance frequency is lower
1H
NMR spectrum of a 55 amino acid protein
C225H356N70O80S9
a
NH
H
N
H
O
C
C
aliphatic side-chain
Backbone
CH2 b
N
H e
NH
CH(a)
Hd
Side-chain
H2O
NH and aromatic
10
9
8
7
6
d
5
4
(ppm)
1H
3
2
1
0
11
NMR spectrum of a 66 kD protein:
Size limitation
- Heavy overlap
- Broad lines
10
9
8
7
6
5
4
d 1H (ppm)
3
2
1
0
12
Relaxation
• Relaxation is the process that brings the
excited system (e.g. after the rf pulse) back
to its equilibrium state
• Transversal (T2): “spin-spin”
• Longitudinal (T1): “spin-lattice”
• Line-width of signal is reciprocally related to
T2: fast relaxation  broad lines
• Both T1 and T2 are dependent on molecular
motions, e.g. for proteins molecular
“tumbling” (correlation time tc (1/tumbling
rate: large molecules have long tc) and
backbone dynamics/conformational
13
fluctuations
Factors influencing the quality of
NMR spectra: pH
• Backbone amide protons very important
for structure determination
• But: Can dissociate and hence exchange
with bulk protons (from water)
• Exchange leads to loss of signal intensity
• Exchange rates usually most favourable at
pH 3-5
14
Factors influencing the quality of
NMR spectra: ions
• salt and buffer
• proteins usually require the presence of buffers
and/or salt
• but: salt and buffer ions add to spectral noise
 loss of signal intensity
• Usually not more than 50-100 mM total
• NB: Buffer must not contain non-exchangeable
protons (otherwise need deuterated compound)
e.g.:
H2C
HO
CH2
C
OH
D2C
NH2
DO
CD2
C
ND2
D2C
H2C
OD
OH
Tris
OD
Tris-d11
15
Water suppression
• Proteins are usually studied in aqueous
solution: 90% H2O/10% D2O.
• D2O required for “lock”: ensures stable field
• Typically, protein concentration a few mM
• Ca. 100 M protons from water (i.e. a 100000-fold
excess)
• Various ways for “getting rid” of water signal:
– “Presaturation”: Irradiation of water resonance at low
power before high-power rf pulse (during the
relaxation delay)
– “Watergate”: Selective pulse flanked by gradient
pulses
– DPFGSE (Double Pulsed Field Gradient Spin Echo) 16or
“Excitation sculpting” (AJ Shaka)
Principles of 2D NMR
• 2D NMR experiments are composed of
a series of 1D experiments
• Involves
– Irradiation of a nucleus (as in 1D)
– “Incremented delay” (different for each 1D
experiment) (also called “evolution”)
– Magnetisation transfer to other nucleus
that is “coupled” to irradiated nucleus
Signal detection (as in 1D)
• Results in information on correlations
between nuclei
17
Principles of 2D NMR
1D NMR:
acquisition
preparation
e.g. relaxation
delay and rf
pulse
t2
2D NMR are a series of 1D experiments:
acquisition
preparation
evolution
mixing
t1
This time period changes between the
various individual 1D experiments
 gives a second time domain
t2
What is detected depends
on what happens during
mixing time (spin coupling)
18
Principles of 2D NMR
Generated from FID as in 1D
1st dimension:
0.10
0.10
0.20
0.20
0.30
0.30
0.40
0.40
Last FID; incremented delay =
0.5 s (e.g.)
time (s)
2nd dimension:
Repeated several hundred
times with different
evolution times t1 (also
called
incremented
delay)
0.50
0.60
0.70
0.50
0.60
0.70
time (s)
Etc....
2nd FID; incremented delay = 10 us
1st FID; incremented delay = 0
Frequency (Hz or ppm)
Fourier Transformation of the second dimension gives the second frequency axis
19
Principles of 2D NMR: Fully FT
transformed spectrum
1st dimension (F2)
(the third dimension is the
peak intensity)
2nd dimension (F1)
20
The “FID” for the second dimension is generated by the “incremented delay”
The mixing time
• Correlation between nuclei happens
during the mixing time
• Reciprocal relationship to observed
coupling
– large couplings - short mixing time
– Difficult to detect small couplings, as
mixing takes too long, and at end of
mixing time no magnetisation left (due to
relaxation)
– If coupling through space: Long range long mixing time
21
Homonuclear 2D NMR
• Typical experiments:
– DQF-COSY (double-quantum-filtered correlation
spectroscopy: up to 3-bond coupling
– TOCSY (total correlation spectroscopy): entire
residues
– NOESY (nuclear Overhauser enhancement
spectroscopy): through space
• COSY and TOCSY are based on scalar
coupling (through bonds), NOESY on dipolar
coupling
22
Identification of spin systems
E.g. Valine:
• Protons have
0 ppm
characteristic shifts
• Tabulated
• Each amino acid has
a characteristic
“pattern” in the
various 2D spectra
F1
H 3C
N
H
H
C
C
H
CH 3
O
C
10
10
Expected TOCSY spectrum
F2
0 23
ppm
2D NMR techniques: TOCSY and
COSY in proteins
0 ppm
Ala10
H(b)
H(a)
O
H
N
C
TOCSY
C
H
C H3
F1
H(a)
H(b)
TOCSY and COSY help
identifying the type of
residue
COSY
amide
10
10
24
F2
0 ppm
Regions in 2D spectra
NH-to-sidechain crosspeaks
aromatic
H(a)-to-H(b)
H(a)-to methyl
(Ala, Thr,Leu,
Val, Ile)
H(b)-to-methyl (Leu,Val,
Ile)
TOCSY spectrum of a
decapeptide (Luteinising
hormone releasing
25
hormone)
Sequential assignment
NOESY connects residues that are adjacent to each other
0 ppm
H
H
N
C
A10
O
C
C H3
N
H
O
C
C
H
COSY
TOCSY
H C H
C11
F1
S
Cd
Intra-residue Cys11
Intra-residue Ala10
Inter-residue,
10
sequential
10
NOESY
26
F2
0 ppm
Overlay of TOCSY with NOESY
H(a)
Sequential
assignment
27
Amide H
Break
28
Recognising secondary structure:
Chemical shift index
• Shifts of backbone atoms are sensitive towards
secondary structure (a helix, b sheet etc)
• Comparison of experimental shifts with
tabulated “random coil” shifts (one for each
amino acid)
• Quick and robust method, 95% accuracy
• Can utilise H(a) protons (13C backbone shifts
also useful)
• Each residue with a shift larger than expected
gets an index of 1
• Each residue with a shift smaller than expected
gets an index of -1
• Residues within random coil shift get a 0
29
Chemical shift index: Example
No recognisable
secondary structure
b strands
a helix
MTKKIKCAYHLCKKDVEESKAIERMLHFMHGILSKDEPRKYCSEACAEKDQMAHEL
-----HHHEE---------HHHHHHHHH--------------HHHHHHHHHH---(secondary structure prediction by jpred)
C
30
N
Secondary Structure
Can also Be
Characterised by
Regular Patterns of
NOEs
H(a) of residue 47
NH of residue 50
NH of residue 51
Strong NOEs between
NH’s of adjacent
residues
NOE between Ha(i) and
NH(i+3)
a helix
31
• Very strong
sequential NOEs
(from H(a) to NH of
next residue)
• Also information on
tertiary structure:
Strong NOEs
between
neighbouring strands
b sheet
32
Recognising the fold: Analysis of
backbone NOEs
Backbone trace
Residue number
50
C
40
b hairpin
30
20
a helix
N
10
0
0
10
20
30
40
Residue number
50
Antiparallel
b sheet
(Predicted by homology modelling,
consistent with CSI and fold
analysis)
33
Distance restraints from NOESY
• The NOE is a dipolar interaction: Through space
• A cross peak between two nuclei means that
magnetisation transfer through dipolar interactions
between two neighbouring spins must have taken
place during the mixing time. This means that the
two nuclei are close together in space.
• The cross peak intensity is defined as follows:
• I = k g12g22 r-6 S J(w)
34
Real-world example: 100 ms 2D NOESY of a 55 aa protein
356 protons
Ca. 2000 peaks
• Intra-residue
• Sequential
• Long-range
35
NMR restraints

evaluated ca 1000
1H peaks

600 peaks
unambiguously
assigned

extracted about
300 relevant
distance restraints
(3-5 Å)
36
Use of coupling constants to gain
structural information
•
3J-scalar
coupling constants (extracted from
dedicated NMR spectra) are dependent on dihedral
(or torsional) angles
B
B
:
A
A
B
Dihedral angle
A
dihedron
X
X
37
Coupling constant 3J
Karplus relationship
Dihedral angle
= a cos2 a - b cos a + c;
a, b, c are empirical parameters - tabulated for
various combinations of nuclei
3J
38
Structure calculations
• A number of programs available, most popular: XPlor, Cyana,
CNS...
• Randomised starting structures
• Use distance restraints (+ various other experimental data)
together with generic atom masses, chirality, electric charges,
Van der Waals radii, covalent bond lengths and angles, peptide
geometry… (constraints)
• Several methods:
1) Distance geometry (DG): calculation of distance constraint
matrices of for each pair of atoms (older method)
2) Restrained Molecular Dynamics (MD): Simulate molecular
motions (e.g. torsions around bonds)
3) Simulated Annealing (SA): heat to a high temperature
(e.g. 3000 °K) followed by slow cooling steps
• Methods 2 and 3 work towards the energetically favourable
final structure under the influence of a force field derived from
39
the restraints and constraints
There is always more than one solution to the
parameter set: The results of an NMR structure
determination are presented as an “ensemble of
conformers”
40
20, structures, all atoms
The ensemble (20 structures)
Backbone traces
41
Average structure
Ensembles are awkward to handle, if one
wants to inspect the structure,
therefore calculation of an
average structure is useful.
“Sausage”: Backbone
representation of
average structure;
thickness of tube
indicates deviations
between individual
conformers
42
Final average structure
Initial average structure is only mean between positions
of individual atoms in different conformers - bonds and
angles strongly distorted - need to do force-field based
energy minimisation.
Newer approach: Select representative conformer
43
Validation
• Structural statistics:
– Violation of restraints
– root-mean-square deviations between individual
conformers and the mean structure
• Back-calculations: does the structural model
give rise to a NOESY spectrum that
resembles the experimental data ?
• Is the structure physically reasonable ?
 Comparison of the resulting structure with
empirical parameters:
– E.g. Whatcheck and Procheck : Look at bond
lengths, angles, dihedrals, van-der-Waals
contacts, stereochemistry.....
44
Heteronuclear NMR
• Common nuclei: 15N, 13C
• Usually requires uniform labelling
 expression of protein in cells that live on
15NH Cl as single nitrogen source, and (e.g.)
4
13C-glucose as single carbon source
• Other nuclei:
• 31P (the only stable isotope) : useful for DNA
• 113Cd or 111Cd : Cd has eight stable isotopes needs enrichment
45
Labelling strategies
•
•
•
•
Uniform
Selective, e.g. all histidine residues
Chain selective (for hetero-oligomers)
Partial
– e.g. deuterate only aliphatic protons
– For solid-state NMR: Use only x % isotopically labelled
nitrogen or carbon source: “dilute spins”
– Or: Mix uniformly-labelled with unlabelled protein
– Or: use differentially labelled 13C sources
• Differential labelling (mixture of 2 compounds,
observe signals of only one): Useful for
protein/protein or protein/DNA interactions
46
15N
• Natural abundance: 0.368%
• Spin ½
• Receptivity relative to 1H: 0.00000384
 Need isotopic labelling
• Recombinant protein expression in minimal
medium with 15NH4Cl as single nitrogen
source
• Relatively cheap: ca. 15£/l culture (which can
be enough for one NMR sample)
47
1H,15N
correlation (HSQC)
d 15N
N
H
O
C
C
105
H
110
C H3
115
120
125
130
135
9.0
d 1H
7.0
48
Advantages of 15N labelling:
Quick way to explore folding
well folded
Unfolded/random coil
49
HSQC spectra taken from NMR pages of the Max-Planck-Institut für Biochemie, Martinsried.
Advantages of labelled proteins
Isotope editing
15N
1H
1H
3D [1H,15N,1H] HSQC-TOCSY and HSQC-NOESY
50
Advantages of labelled proteins
TOCSY
NOESY
Many overlapped peaks
51
1H,1H
plane from 3D [1H,15N,1H] HSQC-TOCSY and HSQC-NOESY
 Peak overlap has been remedied
52
Advantages of labelled proteins:
Chemical shift perturbation
(or shift mapping)
• Universally applicable to study
anything that interacts with proteins
– small molecules (drugs, metabolites)
– other proteins
– DNA and RNA
– metal ions
– ...
• Very rapid method: spectra can be recorded
in few minutes
53
chemical shift perturbation
d 15N
110
Effect of copper binding
T9
on a 64 aa protein
Peaks can
-Stay the same
-Shift
-Split (multiple conformers)
-(Dis)appear
T31
T6
G60
C12
E26
C15
E50
120
E13
T42
A14
H61
Q51
R53
E49
S45
D47
I10
A16
V63
A55
130
V41
I3
A28
A11
V7
E64
d
1H
54
Chemical shift perturbation
5
4.5
4
3.5
Δδ  δ 1 H
2
1
2
 δ 15 N
7
3
Weighted mean deviations
2.5
2
1.5
1
0.5
I5
6
S5
8
G
60
E6
2
E6
4
P8
I1
0
C
12
A1
4
A1
6
A1
8
T2
0
A2
2
Q
24
E2
6
A2
8
A3
0
V3
2
V3
4
L3
6
S3
8
K4
0
T4
2
T4
4
A4
6
G
48
E5
0
D
52
T5
4
T2
Q
4
T6
0
55
Triple-resonance-experiments
(1H,15N and 13C)
• For facilitating sequential assignments
• Example: HNCA
56
Triple-resonance-experiments
The more experiments, the less
ambiguity
Automated sequential assignment
possible
But: NMR instrument time is precious
57
Nucleic acids NMR
• Same principles as in
protein NMR
200 ms NOESY of octanucleotide
d(CGCTAGCG)
O
8
N
4’
CH2
N
O
HC
3’
CH
O-
C
NH
C
C
HC
5’
-O
C
CH
CH2
N
1’
2’
H2’ and 2”
NH2
H3’, 4’ and 5’
{
H1’
{
Guanosine
Aromatic
H8, H6, H2
http://nmr.chem.sdu.dk/dna/noesy_ba.htm
58
Nucleic acids NMR
Sequential assignment:
Correlation between sugar H1’ and aromatic base protons
d(CGCTAGCG)
T4
T4_H1’
H6
H1’
C1
H8
G2
A5_H1’
A5_H8
A5
H1’
C7
G6
C3
T4_H6
G8_H8
59
H8, 6
Break
60
Recent advances
• TROSY: Transverse-Relaxation Optimized
Spectroscopy: enables study of larger proteins
than previously (record so far: 9 megadalton)
• Use of aligned media:
Induces dipolar coupling
– Novel sequences to measure
these residual dipolar couplings
– Gives information on bond
orientation
– Can be used as additional
information for structure
determination
• Partial labelling
61
Example of partial labelling:
Bacterial growth on partially labelled 13C source
• E.g. Glycerol:
OH
H2C
OH
CH
CH2
OH
[1,3-13C]
OH
H2C
OH
CH
OH
Castellani et al, Nature 2002.
CH2
[2-13C]
62
Why partial labelling ?
• Partial 13C labelling:
– No scalar 13C,13C coupling:
– Spectra become less crowded, can
concentrate on dipolar couplings for
structural information
– Avoid dipolar truncation effects (polarization
transfer between two nuclei is cut off in the
presence of a third nucleus)
• 2H: reduce overlap and dipolar couplings
between 1H and 13C or 15N
63
13C
distance restraints from protondriven spin diffusion
64
Kinetics by NMR
65
The NMR time-scale
• NMR is a relatively slow technique
• If there is more than one conformation in
solution, two sets of peaks can be observed,
providing the two species “live” for long
enough to be detected
• Otherwise, averaging occurs
• the "NMR time-scale" for averaging of two
peaks is the reciprocal of the difference in
frequency of the peaks
66
Chemical exchange
• Any process in which nucleus changes
between different environments
• E.g.
– Conformational equilibria
– Binding of small molecules to
macromolecules
– Protonation/deprotonation equilibria
– Isotope exchange processes
67
Exchange regimes
Slow exchange between 2
species
• Lifetime of individual species
decreases
• Exchange rate increases
• Can be achieved by raising the
temperature
Intermediate; Coalescence
Fast exchange
68
http://tesla.ccrc.uga.edu/~jhp/nmr_04/notes/bcmb8190_042604.pdf
H/D exchange
• Dissolve protein in 100 % D2O
• Backbone amide H (and other NHx or OH
groups) exchange with solvent deuterons.
• Exchange is fast when H is solvent exposed
or in a flexible region (loop)
• Exchange is slow when H is buried and/or
involved in H-bond (eg in b sheets or a
helices)
69
Ligand binding studies
• With “small” proteins: can look at
protein and map binding site (1H,15N
HSQC) via chemical shift changes
• With big proteins: observe ligand
spectrum (1H), check qualitatively
whether ligand interacts with protein:
can do rapid screening
• Advantage: Binding does not need to
be strong
70
Ligand binding: Transferred NOE
• Allows observation of ligand conformation
bound to protein
• Principle: Detect NOEs arising from bound
state in unbound ligand
• Conditions:
– Only works for weakly-binding ligands (ligand
must dissociate faster than NOE decays)
– Good if protein is very big (so protein signals
don’t interfere with ligand spectrum)
• Advantage: Sharp signals, as detection
happens in the unbound form
71
Protein motions
• Not all parts of protein have same flexibility
• On 15N-labelled proteins, relaxation rates can
be measured to derive time-scales for motion
of whole molecule, or individual parts, e.g.
backbone dynamics
• Can also estimate correlation time (molecular
tumbling) and infer molecule size and shape
(monomer/dimer, aggregation)
Residue number
Region with high flexibility
72
Metabolomics and -nomics
• Structure elucidation of novel natural compounds
– Combination of NMR with chromatography and mass
spectrometry
• Elucidation of biochemical pathways:
– protein function and mechanisms
– use of labelled precursors, e.g. 13C-labelled acetate, NMR
analysis of products gives information on how metabolites
are synthesised
• Metabonomics looks at complex mixtures such as
body fluids or tissues
– With or without prior separation (chromatography)
– Analysis via comparison with known spectra
– Can be used in diagnosis of diseases
73
Various rat cells
and tissues
Magic-angle
spinning NMR
Vast differences
between tissues
http://www.bbriefings.com
/pdf/855/fdd041_metabom
74
etrix_tech.pdf
MRI: Magentic resonance imaging
MRI scanners
• B0 field horizontal
• 0.5-3 Tesla
• Also uses radiofrequency pulses
• Observed nuclei are the
water protons
• Contrast is achieved by
different relaxation
properties of protons in
different tissues
• Gradient magnets for
spatial information
75
Typical images obtained by MRI
76
In vivo NMR spectroscopy (MRS)
• Diagnostic method
• Looks directly at metabolites in the
body of a living patient (or animal)
• Examples:
• 31P in muscles
• Brain diseases (Alzheimer)
77
In vivo 31P NMR of carp muscle
Normal conditions
Anoxic conditions
http://143.129.203.3/biomag/bil_bio1_spectra/bil-bio1.html
78
In vivo NMR
Energy metabolism in microorganisms
In vivo 31P NMR spectrum of Corynebacterium glutamicum
http://www.fz-juelich.de/ibt/genomics/coryne-phosphorus.html
79
o
s
c
o
p
i
c
S
t
u
d
y
o
f
R
e
v
e
r
s
i
b
l
e
G
l
o
b
a
l
C
e
r
e
b
2D In vivo NMR of brain
S. Brulatout et al.
J. Neurochem. 66
80
2491(1996).
Summary/Outlook
• NMR has a lot to offer for elucidating the
structure and function of biomolecules
• Complementary method to X-ray
crystallography for structure determination
• Can now also do membrane proteins
• Size limitation is still a problem
• Much more than a tool for structure
elucidation (Kinetics/dynamic phenomena,
biomolecular interactions, metabolomics and
-nomics...)
81