Download Lecture Notes on Introduction to 3D NMR | ANES E501 and more Study notes Geriatrics in PDF only on Docsity!
Introduction to 3D NMR
Obstacles to complete 2D NMR analysis of proteins > 100 residues: spectral overlap & sensitivity
Linewidth increases with size (½ ~ 1e
3
*MW)
Number of resonances in same spectral region increases linearly with size
Efficiency of magnetization transfer goes down with decreasing T 2
A partial solution: increase resolution by going from 2D experiments to 3D experiments
Analogous to going from 1D to 2D experiments
First 3D experiment was homonuclear NOESY TOCSY
Oschkinat et al, Nature 332, 374 (1988)
Worked, but:
Poor signal to noise
Limited experimental options
Dizzying number of cross peaks
Within the year (1989), two labs developed 3D
methods based on the amide
1 H
N 15 N coupling
Bax @NIH J ACS 111, 1515 (1989) &
Biochemistry 28, 6150 (1989)
Fesik & Zuiderweg @Abbott Labs
Biochemistry 28, 2387 (1989)
Rationale for the 3D
1 H
15 N methodology:
Assignments are already based mainly on
amide H
N NOEs
So, if increase the resolution of the H
N
, will
increase the size of a protein which can be
assigned (up to ~150 residues)
Resolve the H
N
by the frequency of the
attached
15 N (amide
15 N chemical shift range is
~35ppm, from 100ppm to 135ppm = a good
spread)
Need uniformly
15
N labelled protein (e.g. from
growth of E. coli on
15
NH 4 Cl as Ns ource)
Use [
1
H
N
to all
1
H TOCSY] separated by
15
N
for type/spin system assignments
Use [
1 H
N to all
1 H NOESY] separated by
15 N
for sequential assignments
Figure 1 : Overlapped peaks in a 2D NOESY experiment
get resolved by adding the amide
15 N chemical shift as a
third dimension.
On to 3D:
3D pulse sequence is obtained by concatenating two 2D sequences. Combination of TOCSY with
HSQC gives the TOCSYH SQC, whereas NOESY with HSQC yields the NOESYH SQC (see also
Cavanagh, pp. 447 457):
2D (
1 H
1 H) TOCSY:
What does the TOCSYH SQC do?
Magnetization transfer pathway: All
1
H excited, their chemical shift is labeled in t 1 evolution, then
TOCSYt ransfer, magnetization is transferred from
15
Na ttached protons to nitrogen via INEPT
transfer,
15 N chemical shift labeled during t 2
evolution, reverse inept back to H
N and detection (t 3
t 1
and t 2
are the incremented delays, giving rise the the 2
nd
and 3
rd
dimensions – one can increment
both, either, or neither
OH
H - C - H
I
t
3
t
1
t
2
NOESYH SQC:
Magnetization transfer pathway: All
1 H excited, their chemical shift is labeled in t
1 evolution, then
NOESYt ransfer, magnetization is transferred from
15
Na ttached protons to nitrogen via INEPT
transfer,
15 N chemical shift labeled during t
2 evolution, reverse inept back to H
N and detection (t
3 )
Rationale for the 3D
1
H
15
N methodology:
Assignments are already based mainly on amide H
N
NOEs
So, if increase the resolution of the H
N
, will increase the size of a protein which can be assigned
(up to ~150 residues)
Resolve the H
N by the frequency of the attached
15 N (amide
15 N chemical shift range is ~35ppm)
Need uniformly
15
N labeled protein (e.g. from growth of E. coli on
15
NH 4 Cl as Ns ource)
Use [
1 H
N to all
1 H TOCSY] separated by
15 N for type/spin system assignments
Use [
1
H
N
to all
1
H NOESY] separated by
15
N for sequential assignments
N – C – C
H H O
OH
H - C - H
I
t
3
t
1
t
2
N - H
H - C
Summary of assignment procedure using HSQC, TOCSYH SQC and NOESYH SQC:
- For each crosspeak in the HSQC spectrum, identify the corresponding spin system in the TOCSY
HSQC ⇒ identification of amino acid type and assignment of spin system
- Look at the same strips from NOESYH SQC spectrum: additional crosspeaks that are not present
in TOCSYH SQC are likely sequential contacts: H
N
N
(and H
b
i 1 – H
N
i)^ in^ a^ helices, and
H
a
–H
N in extended conformations.
- Identify which residues these crosspeaks are to (from TOCSYH SQC spin systems), and assign
adjacent amino acids.
- Repeat the same process starting from the newly identified neighboring amino acid.
- When the connectivity breaks down, move to a new starting point.
- Once a substantial number of residues are assigned, ambiguities can be resolved and gaps filled in
(hopefully)
Introduction to 3D “Triple Resonance” NMR
Reading: Evans, pp. 791 03
Ikura, Kay & Bax Biochemistry 29 , 4659 (1990)
Cavanagh, pp. 478528 (r eference, for triple resonance methods).
Last time the assignment process using 3D
15
N experiments (TOCSYH SQC and NOESYH SQC was
presented. It was a big improvement over 2D methods, but:
- The experiments don’t help with sides ide chain NOEs
- The method relies on HN X NOEs for assignments, which
- are conformation dependent
- may be weak (and gets worse with larger proteins, esp. for TOCSYH SQC)
- may be to a degenerate
1 H frequency (e.g. to one of several H
a at 4.50ppm)
A parallel effort was underway in the late 1980’s that used heteronuclear labeling and 2D NMR for
the same goal. The doublel abeling approach was used to make direct scalar coupling connections
directly through the protein backbone using
13 C and
15 N: Oh et al, Science 240, 908 (1988); Stockman
et al, Biochemistry 28,230 (1989). The large singlebond c ouplings give efficient transfer through
the backbone (e.g.
13 CO –
15 N). Importantly, the Ca and C’ can be treated as separate nuclei, since
their chemical shifts are so different (Ca~50 70 ppm; C’ ~1701 80ppm) – this lets one manipulate
them independently to direct transfer. NOEs were no longer needed for making sequential
assignments(!).
15 N
13 C
H O
13 C
13 C
15 N
H H H O
13 C
13 C
13 C
15 Hz 11 Hz 55 Hz
30-40 Hz
95 Hz 140 Hz
4-9 Hz
15
N
13
C coupling
Constants
a lesser degree, HACACO are still used extensively.
First example, the HNCA: Correlates the HN with the intrar esidue C
a , and usually gives a 2
nd
weaker crosspe ak to the preceding C
a .
The pulse sequence:
Here = 1/(4x
1
J
HN
) = 2.5ms for the INEPT and reverse INEPT transfers between
1
H and
15
N,
and d = 1/(2x
1
J
NCa
) = ~22 msec to develop antiphase
15
N
13
C magnetization (NyNzCz)
A.) INEPT transfer
1 H
15 N, giving 2H z
N
y
B.) t 1 evolution on
15
N, decoupling
1
H and
13
C with 180 degree pulses, giving 2HzNycos(wNt 1 )
C.) Develop H z
N
x
C
z
antiphase magnetization at end of d.
D.) 90Hx, 90Cx > HyNxCycos(wNt 1 )
E.) t 2
, evolve on Ca, decoupling
1 Hα,
15 N, and
13 CO with 180° pulses, giving H y
N
x
C
y
cos(w N
t 1
)cos(w C
t 2
Now, back out to H
N
:
F.) 90H
x
, 90C
x
> H
z
N
x
C
z
cos(w N
t 1
)cos(w C
t 2
) (just like D., but now modulated by both w N
and w C
G.) HzNycos(wNt 1 )cos(wCt 2 )
H.) reverse inept back to
1 H
N : H y
cos(w N
t 1
)cos(w C
t 2
I.) Detect (t 3 ), modulated by cos(wNt 1 ) and cos(wCt 2 ).
This experiment correlates the
1 H
N ,
15 N
H , and
13
C
a
resonances within each residue, giving
one intra residue cross peak for each non
proline. Due to the smaller, but significant
coupling constant between
15
Ni
H
, and
13
C
a
i
1
2 J
NCa 1 ~ 49H z) one also gets weak cross
peaks to the C
a
i 1 (which are useful).
H
N
t
t
Ca
t
CO
d d
HSQC
HNCA
110
115
120
9 8 7
120 115 110
9 7 8
N
H
H
N
C
H
C
=N
Current Triple Resonance Strategy:
- The backbone assignments are made by pairs of 3D experiments
- First correlates one or more spins (e.g. Ca) within the residue to its HN
- The second correlates the same spins (e.g. Ca) in the previous (i1) r esidue to the HN (i).
- Typically 12 pa irs of backbone experiments are needed for sequential assignments,
- 12 s ide chain experiments for side chain
1
H,
13
C assignments,
- and 2 3D (or 4D!) NOESY experiments for NOE constraints.
Taking this one step at a time, the companion experiment for the HNCA is the HN(CO)CA:
HN(CO)CA:
It looks like the HNCA, but with only one crosspe ak for each
NH:
For each HN, identify the i1 Ca in the HN(CO)CA, then find the previous residue by scanning for
the matching Ca in the HNCA.
HSQC
HN(CO)CA
110
115
120
9 8 7
120
9
N
H
H
N
C
H
C
=N 120
9
=N
HN(CO)CA HNCA
HN(CA)CO HNCO
TOCSYH SQC H(CCO)NH
15
N
13
C
H O
13
C
13
C
15
N
H H H O
13
C
13
C
13
C
15 Hz 11 Hz
95 Hz
55 Hz (^) 15
N
13
C
H O
13
C
13
C
15
N
H H H O
13
C
13
C
13
C
15 Hz 11 Hz
95 Hz
55 Hz
15
N
13
C
H O
13
C
15
N
H H H
13
C
13
C
13
C
95 Hz
15
N
13
C
H O
13
C
15
N
H H H
13
C
13
C
15 Hz 11 Hz
95 Hz
55 Hz
30-40 Hz
H
2
H
x
13
C
13
C
H
2
H
x
13
C
The last previous experiments give starting points for the side chain assignments (
13 C
a ,
13 C
b ,
1 H
a ,
1 H
b
chemical shifts). Side Chain
1
H and
13
C assignments are then completed using one or two more 3D
experiments, the HCCHC OSY and/or HCCH TOCSY:
For example:
15
N
13
C
H O
13
C
15
N
H H H
13
C
13
C
30-40 Hz
13
C
H
2
H
x
13
C
INEPT
TOCSY or COSY
HCCH-TOCSY
H
N
C
H
70 =C
H
HCCH-COSY
HCCH-COSY
H
=C 70
H
HCCH-TOCSY
2.1 1.
H
a H
b H
g H
g
H
a H
b
Can also measure using the HNHA experiment:
Intensity of CrossPeak / Intensity of Diagonal = t an
2
(2p
3
J
HNHA
)
The intensity of the crosspe ak builds up over transfer time, faster with larger coupling constant
Also, many specific pulse sequences for other coupling constants (and hence, angles)
- Chemical Shifts of C
a
and C
b
The deviations from random coil
13
C chemical shifts for these atoms come from the backbone
dihedral angles. Modern structure calculation programs can refine directly against chemical shifts.
- Hydrogen Bonds
You have heard about hydrogen bonding constraints before. The difficulty is in assigning the two
partners. Perhaps surprisingly, hydrogen bonds have enough covalent character that there is a small
coupling constant across them. A specific version of the HNCO has been developed to observe
cross peaks from these weak couplings – which assign the two partners in an Hbond:
HSQC
HNHA
110
115
120
9 8 7
N
H
H
N
H
H
N H
A
H
120
9
=N
HNHA
9
4
H
N H
N
HSQC
HNCO-J
110
115
120
9 8 7
N
H
H
C
H
CO
HB
13 C
120
9
=N
55
48
CO
HNCO-J
4. NOE
Want semiq uantitative distances from
15
N and
13
Cbond ed
1
H.
A.
1 H
15 N NOESYH SQC: We’ve already talked about the 3D
1 H
15 N NOESYH SQC for NOEs to
15 N
bound protons:
See
1
H N OE>
1
H
N 15
N> cross peaks
1
H unambiguous, all is well
1
H is ambiguous, and is an H
N
, check the
15
N planes corresponding to the possibilities, looking
for a crosspe ak back to the
1 H
15 N
1
H is ambiguous, and is a
13
Ca ttached proton, resolve with next experiment...
B.
1
H
13
C NOESYH SQC:
Looks just like the
15
N experiment, but using
13
C pulses.
Observe
1 H all
NOE>
1 H
13 C
1
H
13
C pair usually unique, but 2
nd 1
H may not be.
1 H unambiguous, all is well
1
H is ambiguous, and is an H
C
, check the
13
C planes corresponding to the possibilities, looking
for a crosspe ak back to the
1
H
13
C
1 H is ambiguous, and is a
15 Na ttached proton, check the previous experiment
Wouldn’t it be nice of you could get:
1
H
13
C N OE>
1
H
13
C(or
15
N) crosspe aks? Well you can, but
it’s a 4D experiment (either of the above two are possible):
1 H
13 C NOESY-HSQC
H
C
H
H
H
H
a
H
b
H
N
13 C = 55 ppm
Pros:
Cons:
- Decreased resolution (not as many increments possible)
- LONG (81 4 days)
- Hard to check visually; need to use peakpi cking software, which can mistakenly pick artifacts
Note: also are some 4D assignment experiments (e.g. HNCAHA and HN(CO)CAHA)
Range of applicability of 3D experiments (how large can one go?)
Using these methods with
1 H
15 N
13 C proteins, can determine structures of proteins to at least 30kD.
Why the size limit? T 2
relaxation.
Larger proteins, longer correlation times (tc), shorter T 2 (faster relaxation)
- Lose coherence over course of pulse sequence
- Esp. for C
a
(refer to HNCA sequence, where d delay = 22 ms (and two of them))
MW t c
T
2
15 N T 2
13 C
a
10 kD ~ 5 ns 100200 m sec 40 msec
30 kD ~ 15 ns 2040 m sec 15 msec
Practical notes:
Need 300uL sample
of 12 m M
Labeling on ly really routine for expression in E. coli.
Minimal media (M9, M63, etc)
15
NH 4 Cl (~0.51 gm /liter, ~$40/gm)
13
Cg lucose (12 gm /liter, ~$200/gm)
“Intermediate” Media
15
Non ly (~$125/liter)
13 C
15 N (~$750/liter)
Rich Media
15
Non ly (~$800/liter)
13
C
15
N (~$2,500/liter)
Suppliers:
- Cambridge Isotopes http://www.isotope.com
- Isotec http://www.isotec.com
- Spectra Stable Isotopes http://www.spectrastableisotopes.com
Check expression levels on minimal & rich media, then choose most coste ffective
Typically, prepare an
15
N sample first.
Optimize solution conditions
Check T 2 ’s
Trial HSQCs
If looks good,
15 N 3D experiments, and prepare
13 C
15 N sample
If T 2 ’s too short (or in general if protein 30kD), make a deuterated sample
Labeling has also been achieved in yeast, insect, and mammalian cells (usually at much higher cost)
References:
Mossakowska, D. E., & Smith, R. A. G. (1997). “Production and Characterization of Recombinant
Proteins for NMR Structural Studies.” in Protein NMR Techniques, D. G. Reid, ed., Humana,
Towata, 325335.
Muchmore, D. C., McIntosh, L. P., Russell, C. B., Anderson, D. E., & Dahlquist, F. W. (1989).
“Expression and Nitrogen1 5 Labeling of Proteins for Proton and Nitrogen1 5 Nuclear
Magnetic Reson..” in Methods in Enzymology, N. J. Oppenheimer and T. L. James, eds.,
Academic Press, New York, 447 3.
For a good summary of the lecture on 3D protein NMR methods in a reall ife example, take a
look at these early papers:
Driscoll, P. C., Clore, G. M., Marion, D., Wingfield, P. T., & Gronenborn, A. M. (1990). “Complete
Resonance Assignment for the Polypeptide Backbone of Interleukin 1b Using Three
Dimensional Heteronuclear NMR Spectroscopy.” Biochemistry, 29, 35423556.
Clore, G. M., Bax, A., Driscoll, P. C., Wingfield, P. T., & Gronenborn, A. M. (1990). “Assignment of
the Side Chain
H and
C Resonances of Interleukin 1b Using Double a nd Triple
Resonance Heteronuclear ThreeD imensional NMR Spectroscopy.” Biochemistry, 29, 8172
Clore, G. M., Wingfield, P. T., & Gronenborn, A. M. (1991). “HighR esolution ThreeD imensional
Structure of Interleukin 1b in Solution by Three a nd FourD imensional Nuclear Magnetic
Resonance Spectroscopy.” Biochemistry, 30, 23152323
