The use of Deuteration for the Structural

Study of Larger Proteins

EMBO Practical Course: Structure, dynamics and function of

biomacromolecules by solution NMR
Biozentrum Basel
20 27 July 2013

Daniel Nietlispach
University of Cambridge
Department of Biochemistry
1. Larger proteins: What are the problems ?

nuclei relax faster due to slower tumbling:

broader lines
lower sensitivity of NMR experiments

relaxation time T2
1
linewidth: "# =
$T2
c [ns] ~ 0.4 MW [kDa]

c : 4 ns
MW : 8 kDa ! correlation time c
8 ns
16 kDa 12 ns
24 kDa
25 ns
50 kDa
 number of signals increase with higher MW:

increased signal overlap

2D NOESY

8 kDa (Tendamistat) 21 kDa (Cdc42)

increasing amounts of protein in NMR sample lead to solubility issues:

may have to reduce concentrations:

probably ok if oligomeric protein, but difficulties if monomeric proteins
 Improvements that facilitate larger MW studies

isotope labeling:
deuteration reduction of proton density:
selective protonation slow down relaxation
simplify spectra
all protons

pulse sequences:
better sensitivity
relaxation compensation TROSY
new approaches

alternative types of restraints: remove 1H(C):

RDC HN, NH2
cross-correlation
chemical shifts
PRE, PCS

hardware development:
higher magnetic fields
cryoprobes
only 1H from
Ile, Leu, Val
 Effect of deuteration on sensitivity of 3D experiments

(ppm)
(ppm)

better sensitivity

13C
13C

increased resolution

1H (ppm) 1H (ppm)
N N

F
protonated sample sidechain deuterated sample
1.0
Data 1
coherence transfer relaxation
relative transfer efficiency

0.8
0.8
HNCA (protonated sample) Sensitivity nsin(J) mcos(J) exp(-R2 ) }
0.6 active passive
0.6

0.4
0.4 Relative transfer efficiency for
HNCA on a protonated sample
0.2
0.2 as correlation time increases.

0
0 5 10-9 1 10-8 1.5 10-8 2 10-8 2.5 10-8 3 10-8
0 5 10 15 20 25 30
correlation time c [ns]
2. Relaxation in solution

B0 S
Main mechanisms contributing to relaxation in solution are:
dipole-dipole interaction
chemical shift-anisotropy ~ r-3
I
R1,2DD( XY ) ~ (K/r6) S(S + 1) (xy )2 " J(i)
i
dipole-dipole interaction

!
S = 1/2 for 1H !
S = 1 for 2H (D) DD( HX ) / DD( DX )
= 16
H = 6.5 D

Transverse relaxation times as a function of molecular tumbling

(600 MHz). Relaxation contributions of DD and CSA interactions
are considered.
2.1 Impact of deuteration on relaxation contributions

Dipolar relaxation

Removable contributions to transverse relaxation

internal DD (protein tumbling with c ~ 12ns)

contributions

H
H
H
external DD
X contributions

X = 13C,15N H

13C(D) 15N(H) H(13C) H(15N)

Deuteration allows to reduce internal and external contributions to dipole-dipole relaxation.

 Perdeuteration
what is it:
removal of all sidechain protons H D ( > 95%)
deuteration level is as high as possible
C
identical labeling pattern for all molecules
remaining protons:
HN C H
sidechain H(N) of Asn, Gln, Arg
labile OH
C N
benefits: N C
strong reduction of external relaxation contributions
removal of internal contributions for 13CHn.
increase in T2 and T1 of 13CHn and 1HN H O
removal of J(H,H) sharper lines
16 =D
!
14

T2(CD) / T2(CH)
12
J(0) dominates
10 coherence transfer relaxation
ratio

8 J(C D) n
sensitivity ! # sin(! J")# cos(! J")exp(R j $ " i )
6 T1(CD) / T1(CH) m n i =1

2
0 5 10 15 20

rotational correlation time [ns]

 Random fractional deuteration
what is it:
statistical removal of sidechain protons to a certain
percentage in a random fashion. H D (0 - 80%)
mixture of different isotopomers (varying local
environment)

benefits:
increase in T2 and T1 of HN and to lesser extent H
C

(and
also sidechain H s)
CH2CH
increasing level of fractional deuteration
CHDCH

CH2CD

C CHDCD

CD2CH

C H CD2CD
0 5 10 15 20 25 30

isotopomer population

C N
N C

H O
correlation time c = 18 ns. DD and CSA are taken into account.
(600 MHz). Relaxation rates with increasing levels of random
= 1H or D statistical sidechain deuteration. Increasing deuteration level
removes external relaxation contributions.
3. Experimental considerations
2H decoupling during 13C transverse periods
remove scalar coupling

reduce effects from scalar relaxation 2nd kind and

dynamic frequency shifts due to quadrupolar
relaxation of deuterium

Editing methods for incomplete deuteration

selection for C-D or e.g. CH2D
d = 1/4JCH
1H suppress all CH
isotopomers
d TCd TC
13C

y y

2H
WALTZ-16x

= x , x rec. = x , x
1/2JCH suppress CHD1,2 and CH3
1H isotopomers

CT-HN(CO)CA 2D 1HN/13C planes with CT=28 ms.

13C
above: with 2H decoupling; below: no decoupling
4. Approaches to structure determination of larger proteins

tentative classification of available labeling strategies for different protein sizes:

performance will vary from protein to protein and depend on the NMR techniques used, e.g. conventional vs.
TROSY implementations etc.





C < 12 ns ( ~20 kDa protein at 298K): 13C, 15N labeling should be enough

C < 18 ns ( ~35 kDa protein at RT): fractional deuteration with 13C, 15N

C > 18 ns ( > 35 kDa protein): - perdeuteration (> 95%) with 13C, 15N
- selective protonation and background deuteration with 13C, 15N
- selective protonation/reverse labeling (12C); background
deuteration 13C, 15N

Backbone assignment
Sidechain assignment
NOE distance information
Dipolar coupling information
4.1 Backbone assignment strategies

Perdeuteration
Maximizes sensitivity thanks to very high level of background deuteration (the higher the more sensitive)
Strong reduction of R2(C) and R2(HN): increased sensitivity, sharper lines
Up to 45 kDa, constant time 13C-evolution periods (CT=1/JCaCb): high resolution
HN back exchange required: sometimes difficult, unfold/refold protocol: will loose some HN
R1(HN) are reduced: slower pulse repetition
Out-and-back triple-resonance experiments:
in pairs: 3D HNCA/ HN(CO)CA
3D HN(CA)CB/ HN(COCA)CB e.g. HN N CO CA CB (t1) CA CO N HN
3D HN(CA)CO/ HNCO
3D intra-HN(CA)CO/ HNCO
3D intra-HNCA/ DQ-HNCA

further: 4D HN(COCA)NH
3D HN(CACB)CG
Increased resolution using 4D approach: HNCOCA/ HNCACO (e.g. MSG 723 AA)
Combine experiments with H/N TROSY transfer/detection scheme. For > 50 kDa probably rather 4D than 3D
 Example: backbone assignment strategy for MSG

Malate synthase G from E. coli (MSG):

723 Amino acids, 81 kDa, correlation time 37 ns @ 37 C

4D TROSY HNCOCA, HNCACO and HNCOi1CAi (to

help resolve ambiguities) shift matching via 13CO and 13C

4D HN,HN NOESY start with Ala-HNCACB to get starting

points

start with Ala-selective 2D HN( CACB) to find starting

points (Alai and some Alai1) Ala ~ 10% of residues in MSG

-sheets and loops: sequential HNHN NOE cannot be

detected. In cases of chemical shift degeneracy, use 13C shifts
from 3D TROSY experiments

Correct 13C shifts for deuterium isotope shifts before

predicting 2nd structure e.g using CSI

Tugarinov, V.; Muhandiram, R.; Ayed, A.; Kay, L. E. J. Am. Chem. Soc. 2002, 124,
10025-10035.
Tugarinov, V.; Kay, L. E. J. Mol. Biol. 2003, 327, 1121-1133.
Tugarinov, V.; Kay, L. E. J. Am. Chem. Soc. 2003, 125, 13868-13878.
Example: Backbone resonance assignment of a 502 residue protein (56 kDa) 3D CT-13C H/C/N experiments

Perdeuteration ( > 98%)

3D TROSY CT-13C experiments TROSY-CT-HNCA

1H-15N TROSY-HSQC

Department of Biochemistry
Away Day 2003
Example: Sequential assignment using 3D intra-HNCA and DQ-HNCA

H O H O
Ras (1-171).GDP @ 4 C
N
C
C C
C
N
C
C C
C
~ 75 kDa . 800 MHz
Assignment setup
for intra-HNCA / DQ-
DQ-HNCA
O
intra-HNCA
O
HNCA
D154 A155 F156 Y157
DQ- intra- DQ- intra- DQ- intra- DQ- intra-
52 HNCA
116.8
HNCA
116.8
HNCA
124.3
HNCA
124.3
HNCA
112.8
HNCA
112.8
HNCA
119.7
HNCA
119.7 !(15N)
13C(i-1)+13C(i)
[ppm]

DQ-HNCA
54

56
intra-HNCA
58
+
60 K104 DQ-HNCA
62 13C(i)
64
increased resolution intra-HNCA
8.14 8.14 8.50 8.50 7.13
1H(F3) [ppm]
7.13 9.55 9.55
and sensitivity

Cross peak is shifted

HN(CO)CA
D154
HNCA HN(CO)CA
A155
HNCA HN(CO)CA
F156
HNCA HN(CO)CA
Y157
HNCA
on-the-fly to its
52 116.8 116.8 124.3 124.3 112.8 112.8 119.7 119.7 !(15N) correct position.
[ppm]
54
?
56 HNCA
i-1
58 +
60 HN(CO)CA
62
? K104

64
13C(i-1) 13C(i)
8.14 8.14 8.50 8.50 7.13 7.13 9.55 9.55
1H(F3) [ppm]

Nietlispach et al., J. Am. Chem. Soc. 2002, 124, 11199.

Transfer efficiency for some triple-resonance TROSY experiments

fast 13CO relaxation with increasing c

and at high magnetic fields

better sensitivity for experiments that

achieve inter-residue correlation without
transffer via 13CO

Field dependence of transfer efficiency Transfer efficiency at 800 MHz

intra-HNCA TROSY

NxC(i)zC(i-1)zCz

8NzC(i)zC(i-1)zC(i-1)y
4NzC(i)zC(i-1)x

selective refocusing of
intra contribution

DQ-HNCA TROSY

DQCx(i)Cx(i-1)

Peak intensity comparison for residues 4-120 of H-Ras(1-171) at 4C (c ~

28ns): intra-HNCA, DQ-HNCA, HN(CO)CA, sequential-HNCA and
conventional HNCA. For large proteins and at high magnetic fields DQ-
HNCA becomes more sensitive than the HN(CO)CA.

J. Am. Chem. Soc. 2002, 124, 11199.

 intra- and inter-residue connectivity using 13C shift information

HNCA HN(CO)CA intra-HNCA

interresidue
correlation
intraresidue
correlation

W31/
A32 F47/
L48

F178/
E179

S76/
N5/ Y77
R6
G152/
Q153

A89/ N219/
C90 E220
N114/ Q136/
15N: 123ppm F115 F137 15N: 123ppm 15N: 123ppm
Example: 3D intra-HNCACB
reduced overlap for 13C resonances

HN(CA)CB HN(COCA)CB intra-HN(CA)CB

E173

N174
Example: Sequential assignment using 3D intra-HN(CA)CO and HNCO

O H

J (CO,C)
C J (N,C)
C
N

C N C
J (N,C) J (N,CO)
J (CO,C) intra-HN(CA)CO

H O

HNCACO HNCO
intra + inter inter

HN(CA)CO
assignment based on matching 13CO shifts



reduced overlap in intra-HN(CA)CO
2D 1HN/13CO projections of the selective intra-
MQ-HN(CA)CO increases signal intensity for
(red) and the conventional 3D TROSY-
HNCACO (blue) experiments recorded on a 80
Ser, Thr, Gly
kDa protein
 sequential

sequential

Nietlispach, J. Biomol. NMR 2004, 28, 131-136

 Random fractional deuteration

Sidechain HC resonances can be observed Hside, Cside assignment

Reduction of R2(C,,...) and R2(HN); smaller reduction for R2(H)
improved sensitivity
Statistical reduction of 1H population
Sensitivity improvement is smaller than with perdeuteration
One sample for backbone, sidechain and NOE
Useful up to ~35 kDa (c ~ 18ns)
Much less problems with HN back exchange
Mixture of various H/D isotopomers: 13C isotope shift effects
limits available resolution in 3D experiments
instead use 4D experiments. Keep lower resolution in each dimension
Out-to-stay triple-resonance experiments: HC HN
4D HBCB/HACANH and HBCB/HACA(CO)NH

Out-and-back triple-resonance experiments e.g HNCA work too:

requires suppression of CH isotopomer
sensitivity reduction by a statistical factor ~ % H2O level in growth condition
 50 60% random fractional deuteration gives increased sensitivity
HBCB/HACA(CO)NH

0% 50% 75%
HBCB/HACANH HBCB/HACA(CO)NH

2D 1H/13C projection plane of the 4D HBCB(CACO)NH for the

deuteration levels 0%, 50% and 75%.

CD2CD

CD2CH

CDCD

CH2CD

CDCH

CH2CH
Magnetization transfer pathway and relative transfer efficiencies for out-to-stay 0 10 20 30 40 50 60 70

experiments as a function of the fractional deuteration level. The best sensitivity

is obtained at 50 60% (depending on the correlation time). Signal contribution of different isotopomers in %

Nietlispach et al. J. Am. Chem. Soc. 1996, 118, 407-415.

4.2 Sidechain assignment

Perdeuteration
Assignment of 13C resonances using 3D C(CCO)NH: increase in T1(13C(D)), H = 0.25 . H ; still, it s quite sensitive !
Low proton density: Assign sidechain HN of Gln, Asn, Arg to increase number of protons
Correction for 13C isotope shifts for e.g.:
2nd structure prediction
to match 13C with protonated samples e.g. for HCCH-TOCSY
Farmer and Venters J. Am. Chem. Soc. 1995, 117, 4187-4188.

Secondary deuterium isotope shifts

Isotope shifts are additive with major contributions from 1-bond to ca. 3-bond:

13C:
1 2H 0.29 0.05 ppm
1H
2 2H 0.13 0.02 ppm
1H
3 1H 2H 0.07 0.02 ppm

15N:
1 1H 2H 0.3 ppm
2 1H 2H 0.05 to 0.1 ppm

Weak dependence on secondary structure: 13C : 0.5 0.08 -helical; -0.44 0.08 -strands
No significant shifts for 1HN and 13CO
( values are based on HCA II (Venters, R. A.; Farmer, B. T.; Fierke, C. A.; Spicer, L. D. J. Mol. Biol. 1996, 264, 1101-1116
Gardner et al., Biochemistry 1997, 36, 1389-1401.))
 Random fractional deuteration
Assignment of H/C using 5060% D sample: 4D HC(CCO)NH same sample as for backbone assignment
HCCH TOCSY:
lower sensitivity due to less protons; sharper lines
longrange correlations benefit and are more sensitive

correlation time c = 18 ns

2D 1H/13C projection of the 4D HC(CCO)NH for

different levels of sidechain deuteration. Best
sensitivity is achieved around 50%.
4.3 NOE distance information

Perdeuteration
HNHN NOE: 4D HNNH NOESY (HMQC, HSQC, TROSY etc.)
(Grzesiek et al. J. Am. Chem. Soc. 1995, 117, 9594-9595; Farmer et al. J. Biomol. NMR. 1996, 7, 59-71.)

typically up to 5 but further possible (slower spin diffusion and longer selective T1 (diagonal signal)
very long mixing times ( < 1.2 s) up to 8 inter-proton distances. (Mal et al., J. Biomol. NMR 1998, 12, 259-276)
Not enough restraints to calculate accurate global folds (RMSD > 8). Particularly poor if large content of -helices.

Additional NOE restraints are required:

sidechain HN: R, N, Q, W are often in interior of protein. However, many are solvent exposed, exchanging rapidly.
sidechain HC: selective protonation approaches

HN/NH2 all protons

H-Ras
 Fractional deuteration

15N separated NOESY benefit from 50% deuteration.

13C separated NOESY loose in sensitivity
deuteration level of 5060% one sample for backbone and sidechain assignment
5060% D is also a reasonable compromise to get NOE information
various isotopomers contribute similarly to diffferent experiments less problems with isotope shifts
clearly not good enough for proteins > 35 kDa instead: perdeuteration, selective protonation

1H 1planes from a 3D NOESY 15N HSQC recorded at

ali/ HN NOE peak intensities as a function of deuteration
0% and 50% fractional deuteration showing the often more level. Relaxation and population effects are taken
intense and better resolved peaks of the deuterated sample. into account.
5. Selective protonation approaches

additional NOE distance information

selective 1H labeling of individual amino acid types or site-specific protonation:
entire A.A.: Smith et al. J. Biomol. NMR 1996, 8, 360-368. Metzler et al. J. Am. Chem. Soc. 1996, 118, 6800-6801.
methyl groups: Gardner et al. J. Am. Chem. Soc. 1997, 119, 7599-7600
highly deuterated background
13C or 12C labeling at protonated positions (reverse labeling) (Vuister et al. J. Am. Chem. Soc. 1994, 116, 9206)
e.g. attractive for aromatic residues

Advantages of selective protonation

protonation sites are part of the protein core
scheme adaptable for the system under study
varying number of residues can be labeled
labeling techniques unproblematic (add precursors: amino acids or -ketoacids)
aromatic residues give many structurally important NOEs

Residue-selective protonation: more NOEs but faster relaxation. Methyl protonation is most sensitive.
Residue-selective
Residue-selectiveprotonation
protonation
protonated
protonatedIle,
Ile,Leu.
Leu.Val,
Val,(Ala)
(Ala)

backbone
backboneassignment
assignmentasasfor
forperdeuterated
perdeuteratedsample
sample
HC(CCO)NH
HC(CCO)NHfor
forprotonated
protonatedresidue
residueassignment
assignment
13
13CCNOESY,
NOESY,CT-1313
CT- C-NOESY
C-NOESY
sensitivity
sensitivitysuffers
suffersfrom
fromintra-residue
intra-residueinteractions
interactions

Phe,
Phe,Tyr,
Tyr,Trp
Trp aromatic
aromaticring-selective
ring-selectiveprotonation
protonation
aromatic
aromaticresidues:
residues:provide
provideadditional
additionalinformation
informationfor
forglobal
globalfold
fold Rajeshetetal.al.J.J.Biomol.
Rajesh Biomol.NMR
NMR2003,
2003,27,
27,81-86.
81-86.
determinatione.g.
determination e.g.for
for-helical proteins
!-helicalproteins

C
C"and
andC
C!positions
positionsare
aredeuterated
deuterated#assignment
assignmentHNCA/CB
HNCA/CB

1212
CCininaromatic
aromaticpositions
positions(slow
(slowrelaxation)
relaxation)

1H amide-rejected 1D spectrum showing the selective aromatic

1H amide-rejected 1D spectrum showing the selective aromatic
ringsidechain
ring sidechainprotonation
protonationthat
thatcan
canbebeachieved
achievedthrough
throughthe
theuse
use
ofofshikimic
shikimicacid
acidininthe
thegrowth
growthmedium.
medium.

Aromaticregion
Aromatic regionofofYUH1
YUH1from
fromamide-rejected
amide-rejectedhomonuclear
homonuclear2D 2D
1H-TOCSY together with strips from 3D
1H-TOCSY 1515N-separated NOESY-
together with strips from 3D N-separated NOESY-
HSQCwith
HSQC withNOEs
NOEsbetween
betweenamide
amideprotons
protonsand
andaromatic
aromaticprotons.
protons.
 Methyl protonation Ile (1), Leu, Val (ILV methyl protonation)

protonated methyl groups of Ile (1 only), Leu, Val with highly deuterated background
well resolved
sharp lines O O O

interior of protein H15N 13CD 13

C O H15N 13CD 13
C O H15N 13CD 13
C O

V,L most common a.a. at molecular interfaces 13CD 13CH

3
13CD
2
13CD CD3

uniform 13C labeling 13CH

3
13CD 13CH
3
13CD
2

13CH 13CH
3 3

backbone: assignment as for perdeuterated sample (out-and-back) precursors for ILV incorporation

sidechain: assignment of methyl groups connect to backbone

3D methyl(H)C(CCO)NH and methylH(CCO)NH
(Gardner, K. H.; Konrat, R.; Rosen, M. K.; Kay, L. E. J. Biomol. NMR 1996, 8, 351-356.)

NOE: too low resolution in methyl-region of conventional 13C NOESY

Val, Ile selective NOESY 3D (HM)CMCB(CMHM) NOESY
(Zwahlen et al. J. Am. Chem. Soc. 1998, 120, 4825-4831)

methyl-selective 13C,13C NOESY

(Zwahlen et al. J. Am. Chem. Soc. 1998, 120, 7617-7625)

methyl-selective HQQF NOESY

(Nietlispach, 1998)
 Methyl protonation of Alanine

protonated methyl group of Ala with highly deuterated background

Abundance: Alanine is one of the most abundant amino acids (8.3%)

abundant in protein hydrophobic core
present in molecular interfaces

Dynamics on ps/ns and s-to-ms timescales

Ala reports on backbone dynamics (Reduced flexibility
Complementarity of
when compared to ILV). ILV + A
ILV provides sidechain dynamics information

very large MW studies with methyl-TROSY (= HMQC !)

structural information
NOEs Ala-13CH3, highly deuterated protein
1) Isaacson et al. JACS 2007, 129, 15428
RDCs of CH3 groups 2) Ayala et al. JBNMR 2009, 43, 111
 Example: Improved resolution in 3D NOESY using CT-13C methyl selective experiment

0
10

12 1
HSQC! (a) HQQF L165a
14
HQQF ! 
I117 I46
2
E156a1
I117a1

16 I126 a2 a2
A142
C [ppm]"

a2
`
E156a2
18
V77 V85
I4 3
a1
a2 (b) HQQC I117`
C (ppm)

a2
20
13 13

4
22 C117_
(c) DEPT-HQQC

[ppm]"
24 L165
b2 5

26 (c) QQF-HSQC
L112 E156_/

1H
b2 L112 C157_
b1 6
28
1.6 1.4 L1651.2 L53 1.0 0.8 0.6 0.4 0.2 0.0
b2 1.5 1.0 0.5 0.0 -0.5
b1 1H (ppm) 1
1H H (ppm)
[ppm]"
Fig.3. Resolution comparison: H/C 2D correlation experiments of a 0.8 mmolar Fig.2. Sensitivity comparison of four 1D versions of CT methyl- 7
sample of doubly labelled CDC42-GMPPNP recorded at 298 K on a Bruker selective experiments: (a) the new QQ filtered experiment (according
AMX600. The conventional 13C HSQC (black) overlayed with the methyl-selective to Fig. 1b), (b) shared-time HQQC (according to [8]), (c) DEPT-HQQC
left: Resolution enhancement obtained in the methyl selective HQQF experiment
(CT-HQQF) experiment (red). The peak heights in the two spectra have been
scaled to approximately the same intensity. The increase in resolution is clearly
(according to Fig. 1a [3,4]), (d) QQF-HSQC with coherence order
selection (according to [9]). The spectra were recorded on a 1 mmolar
I117n
13 8
compared with the C HSQC. right: Sensitivity improvement obtained with
visible. Some of the regions which could not be resolved in the HSQC experiment sample of doubly labelled Ubiquitin at 298 K (CT= 12m, 48 scans). The
are indicated by arrows. new sequence shown in Fig. 1b is by far the most sensitive. The
comparison of (a) and (c) show clearly the improvement achieved by the
HQQC compared with other methyl selective experiments. more favourable relaxation properties of DQ/ZQ over QQ coherence.
For a 2D comparison the S/N in spectrum (d) would improve by 32
relative to the 1D comparison. 9
C157n

10
0.8 0.7 0.85 0.75 0.8 0.7 0.85 0.75 0.9 0.8 1.1 1.0 0.75 0.65
Ile 4 Ile 46 Val 77 Val 85 Ile 117 Ile 126 Ala 142
a2 a2 a1 a2 a2 a2 `

Fig. 4. 3D NOESY-HQQF experiment

1H (CH ) [ppm] " sample of the 21 kDa protein
of 3a 0.8 mmolar
CDC42-GMPPNP recorded at 298 K on an AMX600. The H/H strip-plots show the
strongly overlapped region with bc from 16.2 to 18.5 ppm and bH from 0.7 to 1.0 ppm
left: Pulse sequence of the heteronuclear quadruple quantum
(see Fig. 3). Some assignments are shown for Ile 117a2-CH3 with long-range NOE's to
Glu-156, Cys-157 and Leu-165. The experiment was recorded with 24 scans per
increment13
filtered CT- C HQQF experiment. (CT-period = 6 = 24 ms).
with 90 complex points in F1(H) (t1max=12m) and 50 complex points in F2(C)
(t2max=22m) giving a total experiment time of 5 days.
above: Strip plots (F2 = methyl 13C) from a 3D NOESY 13C
HQQF recorded on 50% fractionally deuterated
Cdc42.GPPNHP.
Proton density comparison for different protonation levels

p21 H-Ras (21 kDa)

Ile, Val, Leu

methyl groups aromatic F, W, Y

HN/NH2 HN, methyl, aromatic all protons

Example: Long range NOEs in ILV (methyl) - FYW (aromatic)-1H YUH1

13C-13C strips from 3D 13C-separated (t1,t2) 1H-1H Strips from 3D 15N- or 13C-separated (t2) NOESY showing aromatic-HN,
NOESY HN-HN, methyl-HN, aromatic-methyl and methyl-methyl NOEs.

I202C 1 I36C 1 I55C 1 L231 H N L231 H 1# L231 H N L231 H 1#

Y33 H #
I202C 1
Y33 H #
I36C 1
I55C 1

W81 H 2

Aromatic 1H-methyl NOESY

Aromatic 1H- H N NOESY

L231 H 1 #

N- NOESY

C NOESY
L231 H 2 #
V207C 1

I55 H1

13
15

V52C 2 V202 H2 #
G232 H N
V52C1
L45C 2 V52C1
G230 H N V52 H1#
L48C 1

N-NOESY

C- NOESY
L39C 1
A32 H N
L48C 2 V52 H2 #

15

13
Example: Global fold of YUH1 calculated from unambiguous NOEs measured
on ILV/FYW- protonated sample

Size: 35 kDa

NOEs
CH3CH3
intra 149
total 268

aroCH3 30

H2NHN and HNHN

intra 181
sequential 672
total 1616

Coordinate RMSD () at initial stage:

backbone heavy atoms (all residues): 2.64 (3.20*)
all heavy atoms (all residues):3.28 (3.86*)
backbone heavy atoms (-sheet region): 1.10 (1.64*)
all heavy atoms ( -sheet region): 1.84 (2.48*)
*without aromatic distance restraints.

Ito et al. 2002

 Example: CH3selective protonation for global fold determination of MBP
Mueller, G. A.; Choy, W. Y.; Yang, D. W.; Forman-Kay, J. D.; Venters, R. A.; Kay, L. E. J. Mol. Biol. 2000, 300, 197-212.

methyl protonated Ile (1), Leu, Val

Global fold of MBP in complex with
-cyclodextrin
(370 residues, 42 kDa)
5.5 RMSD
NOE: HNHN 826
HNCH3 769
CH3CH3 348
H-bonds: 48 N domain
dihedral: 555 C domain

measure 5 dipolar couplings to orient peptide plane

2.2 RMSD
 Example: Alanine methyl protonation for structural studieswww.isotope.com
Cambridge Isotope Laboratories, Inc. APPLICATION NOTE
APPLICATION NOTE

Ala complementarity to ILV labeling

MSG
(Godoy-Ruiz,
JACS 2010, 132,
18340)

Figure 1. Left: Schematic representation of the structure of Malate Synthase G (82-kDa; 73 alanines) with positions of AlaE carbon atoms shown with cyan-colored spheres.
Center: Methyl 1H-13C HMQC correlation map recorded on a2.
0.9 mMSchematic
{U-2H; AlaE-[ CH3]}-labeled MSG
COMMUNICA TIO
of the(37 C; 600 MHz). Right:
withThe zoomed partof
ofAla
theE,correlation
IleG1, LeuGmap
13
Figure Left: representation structure of MSG methyl carbons and ValJ positions shown with cyan-colored spheres. Center: Methyl
highlighted in the center. Selected assignments of separated AlaE cross-peaks are indicated.22 Negative contours are shown
E 13 in red. G1 13
H- C HMQC correlation map recorded on a 0.75 mM {Ala -[ CH3]; Ile -[ CH3]; Leu,Val-[ CH3 / CD3]}-labeled MSG (37 C; 600 MHz). The regions of the map approximately
1 13 13 12

corresponding to IleG1, ValJ and LeuG methyl positions are enclosed in dashed rectangles. The region enclosed in a solid rectangle and highlighted corresponds to AlaE methyl
correlations. Right: The AlaE part of the map zoomed from the region highlighted in the center. Cross-peaks arising from ValJ and LeuG correlations are marked with
Ala labeling for interaction studies asterisks. AlaE assignments are indicated for selected methyls.22

and protonation of AlaE methyls to a level of 95% with minimal In full agreement with previous observations,18 no signs of
background labeling (<1%) in minimal D2O-based bacterial medium scrambling of isotope labels from alanine to other amino-acids have
Interaction of p97 hexamer with Npl4-UBD Proximity
supplemented with large amounts of selectively 13 of alanine side-chains to the protein backbone and their
C-labeled been detected in either sample. In fact, a comparison Appendix of signal 1. Selective labeling of AlaE positions with 13CH3
high degree
D-deuterated alanine and co-addition of three deuterated of order turn Alaintensities of ILV methyls in the {U- H; Ala -[ CH3]; Ile -[ CH3{U-[
E
methyl groups into excellent NMR
2 E 13 groups G1 (the
13
]; 2H]; AlaE-[13CH3]}-labeled sample) has been achieved
compounds: (i) D-ketoisovalerate-D7, (ii) probes
succinate-D for 4aand
number
(iii) of applications:
Leu,Val-[13(i)CH measurements
/ 12
CD ]}-labeled of H-
sample
1
C following the
13 with those obtained in {U-2H;
protocol of Ayala et al.18 using [U-2H]-D-glucose as
Methyl TROSY of p97 ND1 ( CH -Ala, U-2H) 13
L-isoleucine-D10 has been reported by Boisbouvier and co-workers. IleG1-[inCH
3
];E Leu,Val-[
3
CH / 12CD3]}-labeled MSGthe (without Ala E
3 residual dipolar couplings 18
(RDCs) 13
Ala
3 methyls, 13
(ii) characterization
3 main carbon source in E.coli medium and addition of (i) 800mg
This labeling protocol has been closely
2 followed for production of labeling)
slowindicated that the levels of isotope incorporation
of {2-D, 3-into 13 ILV
complexed with Npl4-UBD (U- H)
{2H; AlaE-[13CH3]}-labeled Malate Synthase
of fast (pico-to-nanosecond) and (s-to-millisecond) dynamics C}-L-alanine, (ii) 2.5g of succinate-D4, (iii) 200mg for
atGfunctionally an 82-kDa sitesmethyl
(MSG) important positionsand are (iii)
not methyl-TROSY
compromised to any significant extent by
19,20
of enzymes, D-keto-isovalerate-D 7
and (iv) 60mg of L-isoleucine-D10 to 1 liter
enzyme containing 73 methyl groups. Alanine is the most abundant additions of large amounts ofE labeled Ala to the medium. This is the
NOE spectroscopy that can be performed on {U- H; Ala [ CH3]; 2 13
of D2O-based M9 medium 1 hour prior to induction of protein
residue in MSG comprising 10.1% of the total amino-acid content. direct consequence of the fact that the biosynthetic pathway leading
IleG1-[13CH3]; Leu,Val-[13CH3 / 12CD3]}-labeled samples increasing the overexpression with 1 mM IPTG. Selective 13CH3 labeling of all ILV
Figure 1 shows the methyl-TROSY 1H-13C correlation map of MSG to incorporation of labels into IleG1 positions is short-circuited by
prepared using the protocol described innumber of methyl probes
1. for addition
derivation of distance restraints positions together whilewith Ala methyls ({U-[ H]; Ala -[ CH3]; Ile -
E 2 E 13 G1

Isaacson et al., JACS 2007, 129, 15428 more detail in Appendix

compared to conventional ILVthe labeling
of the
pathway
suitably labeled
methodology.
D-ketobutyrate to
leading to labeling of ValJ and LeuG[ sites
13
the medium,
CH3is];short-
Leu,Val-[ CH3 / CD3]}-labeled sample) has been achieved
13 12

To maximize the number of available methyl probes, it is clearly circuited by addition of the D-keto-isovalerate using the Itsame
precursor. mightcarbon sources as above by addition of (i) 800mg
advantageous to combine selective labeling of Ala positions with have been expected that the addition of D-ketobutyrate of {2-D,3- C}-L-alanine; (ii) 2.5g of succinate-D4; (iii) 120mg of
to13the
ILV methyl labeling.21 The use of selectively 13CH3-labeled AlaE ({2-D, medium instead of deuterated isoleucine in order D-ketoisovaleric
to ensure 13CH3 acid, sodium salt (3-methyl-13C: 3,4,4,4-D4) and (iv)
3-13C}-L-alanine) in combination with (i) selectively 13CH3-labeled labeling at IleG1 methyl positions, would result in60mg partialof 13
CHD-Ketobutyric
3 acid, sodium salt (3-methyl-13C; 3,3-D2) to
D-ketobutyrate for labeling of IleG1 poistions, (ii) 13CH3 / 12CD3-labeled labeling at IleJ2 methyl sites (arising from alanine-derived 1 liter of[3- theCH
13
]-
medium
3 1 hour prior to induction. {AlaE-[13CH3]}- and
D-ketoisovalerate for labeling of ValJ and Leu G
Figuresites,1.and
(A(iii)
and B) Alanine pyruvate and 1H-into
labelingentering 13C 2D methyl TROSY spectra
the biosynthetic cycle of IleG instead 13
Eof13(A) of CH theG1
3-Ala,U-
2H labeled Ufd1in UN and (B) 13CH -Ala,U-2H lab
3
{Ala -[ CH ]; Ile -[ CH ]; Leu,Val-[ CH / CD ]}-labeled samples
13 13 12
 the beginning of R-helices in the structure of MSG ), t
of structural environment seen by alanine residues led
overlap in all the four dimensions. Of note, the two m
Example: 4D Ala-HMCBCACO for alanine methyl group assignment shifted Ala methyl peaks in the 1H dimension belong
A585 with 1H(13C) chemical shifts of -0.68(15.6) an
ppm, respectively (outside of the spectral window of F
direct consequence of the ring-current effect induced b
Alanine methyl region can be very overlapped W587 indole rings, respectively.
We anticipate that the developed methodology will

(1H): 1.0 1.7 ppm, (13C): 17 21 ppm use of alanine 13CH3 methyls as probes of molecular s
dynamics in large proteins. The NMR experiment de
is expected to be applicable to protein systems with a
Assignments from 3D HMCBCA(CO) and 3D HMCB(CA)CO of chemical shift degeneracy, such as membrane o
or from 4D HMCBCACO. unfolded proteins as well as large protein-protein
irrespective of the labeling strategy chosen to selective
alanine methyl sites.
Acknowledgment. This work was supported in
Nano-Biotechnology Award (University of Maryland)
authors thank Prof. Lewis Kay (University of Toront
HM-CB slices from 4D useful suggestions.
HMCBCACO
Supporting Information Available: One table listin
CR, and 13CO chemical shifts of alanines in MSG (37 C
13

One table listing acquisition parameters of the 4D ALA-H

experiment. Listed also is the Bruker pulse code for th
HMCMCACO experiment (Figure 1). This material is a
of charge via the Internet at http://pubs.acs.org.
Figure 2. (a) Alanine region of a 2D methyl 1H-13C CT-HMQC spectrum
recorded on a 0.9 mM {U-13C}-Pyruvate-derived sample of MSG (99.9% References
D2O; 600 MHz; 37 C; pH ) 7.1). The spectrum was filtered for 13CH3
(1) Goto, N. K.; Gardner, K. H.; Mueller, G. A.; Willis, R.
(against 13CHD2) methyl groups.12 Assignments of Ala methyls are indicated J. Biomol. NMR 1999, 13, 369374.
with residue numbers. The peaks belonging to Val and Ile2 methyl signals (2) Tugarinov, V.; Kay, L. E. J. Biomol. NMR 2004, 28, 165
are marked with *. (b-c) Selected 2D 1H-13CR slices from the 4D ALA- (3) Tugarinov, V.; Hwang, P. M.; Ollerenshaw, J. E.; Kay, L. E
HMCBCACO data set recorded on the same sample as in (a) using the Soc. 2003, 125, 1042010428.
pulse scheme of Figure 1. The planes are drawn at 13C/13CO chemical (4) Sprangers, R.; Kay, L. E. Nature 2007, 445, 618622.
(5) Hamel, D. J.; Dahlquist, F. W. J. Am. Chem. Soc. 2005, 1
shifts indicated at the top left corner of each panel. Methyl cross-peaks are(6) Sprangers, R.; Gribun, A.; Hwang, P. M.; Houry, W. A.; K
marked with residue numbers, while the peaks whose maxima lie outside Natl. Acad. Sci. U.S.A. 2005, 102, 1667816683.
of the plotted 2D slices are marked with *. (7) Velyvis, A.; Yang, Y. R.; Schachman, H. K.; Kay, L. E. Pr
Sci. U.S.A. 2007, 104, 881520.
observable signal at the end of the scheme. The pulse scheme (8) McCaldon, P.; Argos, P. Proteins 1988, 4, 99122.
utilizes the Methyl-TROSY3 principle where possible by keeping Figure 2. (a)(9) Fernandez,
Alanine region A.;
of a Scott, L. R.;
2D methyl 1 Scheraga,
H-13C CT-HMQC H. A. spectrum
J. Phys. Che
recorded on a 0.999299932.
mM {U-13C}-Pyruvate-derived sample of MSG (99.9%
the methyl ( H - C ) magnetization in a multiple-quantum state
1 13
(10) Tugarinov, V.; Muhandiram, R.; Ayed, A.; Kay, L. E. J. A
D2O; 600 MHz; 37 C; pH ) 7.1). The spectrum was filtered for 13CH3
before the 13C2/Hy pair of pulses and after the 13C4/Hy pair (against
of 13CHD 2002, 124, 1002510035.
12
13 (11)2) methyl
Isaacson,groups.
R. L.; Assignments
Simpson, P.ofJ.;Ala methyls
Liu, are indicated
M.; Cota, E.; Zhang,
pulses (Figure 1). To eliminate CHD2 methyl isotopomer signals with residue numbers.
P.; Matthews, belonging
The peaks S. J. Am. to Val Soc.
Chem. and Ile2
2007, methyl signals
129, 15428154
1 13 R
2
from the spectra, the 180 H pulses in the middle of 4TC and t3are marked with
(12) *. (b-c)
Gardner, Selected
K. H.; 2D H
Rosen, -
M. CK.;slices
Kay, from
L. E.the 4D ALA- 19
Biochemistry
1 HMCBCACO data set recorded on the same sample as in (a) using the
1401.
periods ensure evolution of the JC-D coupling; subsequent applica- Sheppard et al. JACS 2009, 131, 1364
pulse scheme of Rosen,
(13) Figure M.1. The
K.; planes
Gardner,areK.drawn at 13CR.
H.; Willis, /13CO
C.; chemical
Parris, W. E
tion of 2H 90 pulses at time points approximately equal to 0.25/
shifts indicated atKay, L. left
the top E. J.corner
Mol. of each1996,
Biol. panel.263, 627636.
Methyl cross-peaks are
1
J marked with(14)
eliminates most of the deuterium-containing magnetization Venters,
residue R. A.;
numbers, Farmer,
while B. T.;
the peaks Fierke,
whose C. A.;lieSpicer,
maxima outsideL. D
 Sidechain assignment in very large proteins

Malate synthase G from E. coli (MSG): 723 Amino acids, 81 kDa, correlation time 37 ns @ 37 C

Tugarinov, V.; Kay, L. E. J. Am. Chem. Soc. 2003, 125, 13868-13878.

previously:

methyl protonated Ile (1), Leu, Val with 2H background and uniform 13C

TOCSY: HM(CCO)NH and (HM)CM(CCO)NH and [H,N]-TROSY implementations of it

Cme

CM Cb/Cg
beyond 60 kDa sensitivity becomes too low:
Ca HN
due to branching at C (Val), C (Leu): makes TOCSY problematic

Approach 1: COSY type correlation Methyl HN

Ile: 3D (HM)CM(CGCBCA)NH

problematic for Leu, Val due to similar pro-R,S CH3 shifts

Leu, Val: reverse label one methyl group to 12CD3

= linearize CC-spin system
Leu: 3D (HM)CM(CGCBCA)NH
Val: 3D (HM)CM(CBCA)NH
Approach 2: Methyl-detected out-and-back experiments

higher sensitivity than approach 1: 5 10 x

benefits from methyl reverse labeling to 12CD3 for Val, Leu
3D experiments: Cme/Cali/Hme and Cme/CO/Hme
3D HMCM[CG]CBCA
3D HMCM([CG]CBCA)CO
sequence specific CH3 assignments by matching
Tugarinov, V.; Kay, L. E. J. Am. Chem. Soc. 2003, 125, 13868-13878.
13Ca, 13Cb, 13CO with backbone experiments
High-resolution 13C CT-HMQC of methyl region (CT=28ms) Effect of external relaxation sources on 13C HMQC

usage of linearized spin systems in MSG Reduced concentration of methyl protons leads to
higher resolution and sensitivity

Tugarinov, V.; Kay, L. E. J. Am. Chem. Soc. 2003, 125, 13868-13878.

 Methyl-TROSY

Tugarinov V. et al. J. Am. Chem. Soc. 2003, 125, 10420-10428.

Tugarinov V. et al. J. Biomol. NMR 2004, 28, 165-172.

13C: HMQC
8 transitions of which 50% relax slowly
1H: 10 transitions of which 50% relax slowly

TROSY: keep fast and slow relaxing components separate

HMQC transfers 50% slow slow
HSQC transfers 19% slow slow

Cancellation of HH and CH dipolar interactions field independent

for cm >> 1

Remove external HH dipolar interactions that degrade performance:

Ile(1)-13CH3
rest: 2H and 13C for assignment HMQC > HSQC
rest: 2H and 12C for NOE

strong external HH between pro-R,S CH3 in Leu, Val:

Leu, Val mono-methyl 13CH3,12CD3
rest: 2H and 13C for assignment
rest: 2H and 12C for NOE c ~ 118 ns

ZQ-line narrowing in Methyl-TROSY:

reduces intra- and inter-methyl dipolar interactions

Tugarinov et al. J. Am. Chem. Soc. 2004, 126, 4921-4925.

 Global fold determination of the 82 kDa MSG (723 residues)

Methyl protonation approach. MSG: 22% I, L, V residues

Global fold based uniquely on NMR data. Tugarinov et al., PNAS 2005, 102, 622-627
Assignment:
backbone and CH3: U-[15N,13C,2H], Ile1-[13CH3], Leu,Val-[13CH3,12CD3]
NOE: 3D [H,N]-& methyl-TROSY, 4D CH3-CH3
HNHN: U-[15N,2H] in H2O
CH3CH3: U-[15N,2H], Ile1-[13CH3], Leu,Val-[13CH3,12CD3] in D2O
HNCH3: U-[15N,2H], Ile1-[13CH3], Leu,Val-[13CH3,12CD3] in H2O

long-range contacts
HNHN: 99 only few as high -helical content
CH3CH3: 386 shows strength of CH3 labeling
HNCH3: 142 rmsd 5.6
total longr.: 627 almost 1 per residue
all NOEs: 1531 -helices ok: i i1, i i3
-sheet: show shorter than in X-ray. Due to 2H no H
, values from chemical shift (TALOS): 1066 contacts between proximal strands!
rmsd: 5.6 : HN(i) HN(j) >4.0; : HN(i) HN(j) >3.3

Inclusion of orientational restraints: rmsd 4.1

RDC (1H-15N): 415
changes in 13CO shifts: 300 Conformational space of 2nd structure elements is
reduced. A few couplings per residue are enough to
achieve correct alignment.
NOE data provides translational information.
SAIL (stereo-array isotope labelling)

Kainosho et al.
Nature 440,
52-57 (2006)

chemical or enzymatic synthesis of

amino acids

followed by cell-free protein

expression (prevent scrambling due to
lack of cell-based metabolic pathways)
 Spectroscopic advantages of SAIL

through bond connectivity maintained backbone and side chain assignment

no need for stereospecific assignment

J-couplings are easier to measure

reduced spin-diffusion improved rHH distances

sharper lines as:

less long-range J
reduced dipolar interaction

dynamics studies:
simple and established for CHD and CHD2
 reduced number of NOEs: 40-45% (difference: mostly fixed or redundant)

reduced overlap

higher sensitivity: T2(H,C) , less long-range J

CH3 vs CHD2 ~ 1:1 (for 1/3 population)
CHD vs CH2 3 - 7x (increases with c)

NOEmax: mix 1.5 - 3x longer than with UL proteins

 SAIL versus universal labelling (UL): e.g. Calmodulin
CHD2 CHD

SAIL

CH3 CH2

UL

SAIL UL
Arginine
N-H

Structure determination of MBP (41 kDa)
MPB (41 kDa) Structure determination:
similar precision and accuracy and statistics to typical small proteins

previously: global fold: RMSD 3.8

SAIL RMSD 2.3 (over N- and C-terminal domains)
6. Practical aspects of producing deuterated proteins

Freshly transformed E.coli (e.g. BL21 (DE3)), minimal medium D2O: growth rate , biomass , protein

Approaches:

1) Quantity of protein more important than deuteration level: max. 75-80% 2H incorporation

A) Plate onto solid H2O-based minimal or rich medium. Increase level of D2O on
plates for gradual adaptation.

B) Growth in solution: Small scale prep. Minimal medium. Adapt bacteria to grow in
deuterated medium by culturing in increasingly higher levels of D2O. OD600 < 0.6
spin, remove cultures. Resuspend in fresh medium (A600 <0.1, log-phase)
=cell resuspension approach. 1H-glucose.

Fractional deuteration: no adaptation, 1H-glucose, ~ % D2O

Per deuteration: 75-80% 2H using approach B).

2) High deuteration levels: Perdeuteration of sidechains (>85%)

D2O, absolutely no H2O. Adaptation procedure for cells (e.g. 10, 25, 60, 90%), 2H-glucose.
Many cells survive and express protein when transferred from H2O D2O without adaptation.

2H-glucose (more reliable for protein expression), 2H-acetate, 2H-glycerol, 2H-succinate, 2H-pyruvate.

D H back exchange incomplete (particularly 2nd structure) unfold/refold protocol.

3) Site-specific protonation, highly deuterated background

Residue protonation: 13C, 2H background

minimal medium, D2O, 2H-glucose(13C), 15NH4Cl
+ protonated amino acids or precursors (> 50mg/l)
but: D2O will replace 30-80% of H.




or: auxotrophic strains
+ e.g.shikimate F,Y,W ring protonation with ,-2H





ILV Methyl protonation, with 13C, 2H background
minimal medium, D2O, 2H-glucose(13C), 15NH4Cl
[2,3-2H] 15N,13C 2-Ketoisovalerate (> 80mg/l) Leu, Val CH3
[3,3-2H] 15N, 13C 2-Ketobutyrate (> 50mg/l) Ile CH3(1 only)

ILV mono-Methyl protonation for linearized spin system approach: 13C,2H background
as above but use:
(13CH3)(12CD3)[2,3-2H] 15N,13C 2-Ketoisovalerate
2H-glucose(13C)

for CH3-TROSY

ILV mono-Methyl protonation for isolated 13CH3 spin system: 12C,2H background
as above but use:
mono-methyl 13CH32-Ketoisovalerate
mono-methyl 13CH32-Ketobutyrate
2H-glucose (12C)


3) Site-specific protonation, highly deuterated background (continued)

Alanine Methyl protonation, with 2H background

{2H, 3-13C}-L-alanine (50 mg/l)
+ deuterated rich medium, D2O

or (less scrambling, <1%):

{2H, 3-13C}-L-alanine (800 mg/l)

+ deuterated ILV precursors (60-120 mg/l)
+ deuterated succinate (2.5 g/l), D2O

ILV + Alanine Methyl protonation, with 2H background

{2H, 3-13C}-L-alanine (800 mg/l)
+ deuterated ILV precursors with according 13C-CH
3 (60-200 mg/l)
+ deuterated succinate (2.5 g/l), D2O

Alanine Methyl protonation, with 13C, 2H background

(U-2H, 13C)-pyruvate, D2O
This leads to CH3 (25%), CH2D, CHD2 but also Ile2, Leu, Val

which are partially protonated
Further reading:

K. H. Gardner, L. E. Kay, Annu. Rev. Biophys. Biomol. Struct. 1998, 27, 357-406.
The use of 2H, 13C, 15N multidimensional NMR to study the structure and dynamics of proteins.

R. A. Venters, R. Thompson, J. Cavanagh, J. Mol. Struct. 2002, 602, 275-292.

Current approaches for the study of large proteins by NMR.

V. Tugarinov, P. M. Hwang, L. E. Kay, Annu. Rev. Biochem. 2004, 73, 107-146.

Nuclear Magnetic Resonance Spectroscopy of high-molecular weight proteins.

L.-Y. Lian, D. A. Middleton, Prog. Nucl. Magn. Reson. Spectrosc. 2001, 39, 171-190.
Labelling approaches for protein structural studies by solution-state and solid-state NMR.