!"#$%&'#(!"#$%&'( )$*+,-,(.

#(/0%'%10#(2,"-3,-"4'#((56786(9&#'

weff

;4*,(4$.(*#$*0+<#( w1

&0%=%'#3-'4"(>?@(2A#3,"%*3%AB

/#"$:4".(/"-,*3:#"

!"#$#%&'(%)*+,-.+/0&'1*#2'#03+4*#(0
)*+,-$(&./0(,"#$%&'(

EMBO course, Basel, July 2013

C-,'0$#(%D('#3,-"#

1 Multidimensional NMR: short introduction

2 Fast-pulsing techniques
- Longitudinal 1H relaxation enhancement
- BEST techniques
- 15N polarization enhancement & BEST-TROSY
- Ernst-angle excitation & SOFAST-HMQC
3 Applications
- NMR quality control: probing structural compactness
- Real-time NMR studies of protein folding
- NMR studies of highly disordered proteins
 ?-'+.0=#$*0%$4'(A"%,#0$(>?@(*A#3,"%*3%AB

1D

1H

2D

15 N
15N
1H
3D

13C

1H
1 15 N
H
4D, 5D,

Site-resolved information on molecular structure & dynamics

/0%=%'#3-'4"(>?@E(=4F%"('0=0,4+%$*

(1) low sensitivity (2) long experimental times

B0
t1 mix t2 mix tacq Recycle delay

1017 molecules time scale: !1-2 seconds

- Only 1 out of ~10000 molecules

contributes to NMR signal. Sampling grid: t2

t1
- Signal loss during various coherence
transfer and frequency editing steps.
 ?-'+.0=#$*0%$4'(>?@E(+=#("#G-0"#=#$,*

Example: 4D

t1 mix t2 mix t3 mix tacq

t3

Each point corresponds to

1 signal acquisition

t2
t1
tim e

Time requirements for nD NMR experiments:

1D 2D 3D 4D 5D
seconds minutes hours weeks years

9HA#"0=#$,4'(+=#("#G-0"#=#$,*

Standard NMR techniques 1D 2D 3D 4D 5D

seconds minutes hours weeks years

Fast-pulsing NMR BEST

Sparse data sampling

2D 3D 4D
seconds minutes hours
;4*,(>?@E(A"0$30A4'(=%+<4+%$*

NMR studies of short lived samples

e.g. folding intermediates, in-cell studies,

Optimized use of (expensive) NMR spectrometer time

do not waste time where sensitivity is abundant, spend time where
sensitivity is needed (quantitative measurements).

NMR studies of large, or intrinsically unstructured proteins

making high-resolution 3D, 4D, experiments feasible
that otherwise require unreasonably long acquisition times.

Time-resolved NMR studies of kinetic molecular processes

at atomic resolution.

;4*,(>?@E(*A#3,"%*3%A03(,%%'(&%H

Polarization-enhanced
fast-pulsing techniques

Longitudinal Heteronuclear
1
H relaxation polarization
enhancement enhancement

Optimized flip
angle (Ernst
angle)

Spectral
folding
Spatial
encoding

Projection Hadamard
spectroscopy spectroscopy
Non-uniform data
sampling
& processing

Sparse-sampling techniques
8%'4"0I4+%$7#$:4$3#.(D4*,7A-'*0$1(,#3:$0G-#*

Gain in speed and sensitivity from simple modification

of standard pulse sequences

BEST

Gain in Gain in speed

sensitivity = reduced
acquisition time

J%$10,-.0$4'("#'4H4+%$K(0$,#"7*34$(.#'4B(L(*#$*0+<0,B

wait...
z-magnetization

~0.1 s 1-2s

scan time Longitudinal relaxation

0 1 2 3
Sensitivity

S/N ~ !(number of scans)

~1/!(scan time)
S/N

1 2
scan time / seconds
0 longitudinal
Enhanced 1 relaxation:
2 3
! Shorter optimalScan
repetition
time [sec]rates
! increased experimental sensitivity
 J%$10,-.0$4'(MN("#'4H4+%$(0$(=43"%=%'#3-'#*

Solomon equations:

I1z-I1z0 "j#1j $12 $13 ... $1n I1z-I1z0

d I2z-I2z0 $21 "j#2j $23 ... $2n I2z-I2z0
= ...
dt ... ...
$ $ n2 $n3 "j# nj Inz-Inz0
Inz-Inz0 n1

Example: amide 1H in proteins

The spin-lattice relaxation of a
1H spin depends on the initial

C
spin-state of the surrounding 1H.
N
H H (nOe or spin diffusion effect)
C H C
H
H H
C C

In addition: chemical exchange with water 1H

J%$10,-.0$4'(MN("#'4H4+%$(#$:4$3#=#$,

Selective excitation of a small number of protons, e. g. amide 1H

Aliphatic 1H saturated (!)

H
H H
Mz = 0 HN H%
Mz = 0 Mz = 0
0 1 2 0 1 2
Recovery time Recovery time
[sec] [sec]

Aliphatic 1H in equilibrium (")

H
H H
Mz = Meq HN H%
Mz = 0 Mz = Meq
0 1 2 0 1 2
Recovery time Recovery time
Pervushin et al., JACS (124) 2002, 12898. [sec] [sec]
 J%$10,-.0$4'(MN("#'4H4+%$(#$:4$3#=#$,

Magnetic field (B0) dependence

selective non-selective

non-selective

selective

/92O(9HA#"0=#$,*(

BEST: Band-Selective Excitation Short-Transient

aliphatic

amide Replace all 1H hard pulses by

band-selective or pairs of
broadband inversion pulses.

10 8 6 4 2 0 ppm
Example: HNCO and HNCA

Schanda et al., JACS (128) 2006, 9042; Lescop et al., JMR (130) 2007, 5014.
 /92O(9HA#"0=#$,*(

Examples of band-selective shaped pulses

* BURP family (Geen & Freeman, early 90s)

* Gaussian pulse cascades (Emsley & Bodenhausen, early 90s)
* SNOB family (Kupce, Campbell, early 90s)
* PC: Polychromatic pulses (Kupce & Freeman, early 90s)
* E-pulses: High-performance selective excitation pulses
(Veshtort & Griffin, 2004)
Spin polarization (Iz)

REBURP UBURP
EBURP2 E400B

Frequency offset (kHz) o.k. Do not use !

/92O(9HA#"0=#$,*(

Problems related to the use of shaped pulses

* Some pulses are very sensitive to B1-field inhomogeneity

* Most shaped pulses perform a specific task: excitation, inversion

* Homo- and heteronuclear coupling evolution during pulse ???

=> pulse length can not be neglected with respect to spin evolution

* Realization of several parallel coherence transfer pathways ????

 /92O(9HA#"0=#$,*(

shaped pulse: suite of delay and universal rotation pulse

Simulations performed with SPINEVOLUTION

!A powerful tool for the simulation of solid- and liquid-state
Lescop et al., JMR (203) 2010, 190.
NMR experiments! (Veshtort & Griffin, JMR 2OO6, 178, 248)

/92O(9HA#"0=#$,*(

INEPT
Basic building blocs

BEST-HSQC

BEST-TROSY

E: Excitation pulse (EBURP2)

FB: Flip-back pulse (time-reversed EBURP2)
R: Refocusing pulse (REBURP)
/92O(9HA#"0=#$,*(

high-speed and optimal-sensitivity regime

BEST
4
Sensitivity [a.u.]

~ 30-100%
3 more sensitive

2
standard
1

0
0 1 2
Recovery time [sec]

Sensitivity gain of a factor of 2-5

/92O(9HA#"0=#$,*
600 MHz, 25C
BEST 1H-15N HSQC 8.6 kDa

12 kDa

BEST 1H-15N TROSY

21 kDa
 MP>(A%'4"0I4+%$(#$:4$3#=#$,(0$(/92O7O@C2Q

TROSY spectroscopy
CSA-DD cross correlation
1H line width 15N line width

1.0-1.2 GHz 800-1000 MHz

Pervushin et al., PNAS (94) 1997, 12366.
Yao LS, Grishaev A, Cornilescu G, Bax A, J. Am. Chem. Soc. 2010.
Pervushin, Quarterly Reviews of Biophys. 2000.

Factor of 2 signal loss due to single-transition selection

Main field of application: large perdeuterated proteins
High magnetic field strength advantageous

MP>(A%'4"0I4+%$(#$:4$3#=#$,(0$(/92O7O@C2Q

TROSY: detection of different pathways

1H pathway: H z "INEPT (
""# 2H z N x = H$ N x % H & N x "ST"2%PT )
"# N & H x

N z " ""# N x = ( H$ N x + H % N x ) " ""# N % H x

15N INEPT ST 2PT
pathway:

!
 MP>(A%'4"0I4+%$(#$:4$3#=#$,(0$(/92O7O@C2Q

Adding 1H and 15N steady-state polarization

&0 = y

(
Equilibrium conditions (Trec > 5 T1): Ptoteq = " H 1+ " N " H )
~ 10% signal enhancement &0 = 'y

!
10 8
1H
6
(ppm)
Steady-state (fast pulsing) conditions:
&0 = y
~ no signal enhancement

&0 = 'y

10 1H
8 6
(ppm)

MP>(A%'4"0I4+%$(#$:4$3#=#$,(0$(/92O7O@C2Q

Situation changes with additional relaxation delay

Trec = 200ms ; ( = 100 ms

&0 = y

&0 = 'y

significant (> 10%) signal enhancement !

 MP>(A%'4"0I4+%$(#$:4$3#=#$,(0$(/92O7O@C2Q

".relaxation
"" "# H z "ST"
2$PT
"# $N z

contributes to 15N pathway in next scan

!
Relaxation induced signal loss is (partly) recovered
via this 3rd coherence transfer pathway !

MP>(A%'4"0I4+%$(#$:4$3#=#$,(0$(/92O7O@C2Q

What we can get ?

N N (( ( ))
" = P ss P eq =1 # 1 # $ H $ N 1 #exp ( #% T1 H ) exp ( #Trec T1 N ) )
Trec = 50 - 500 ms
T1H (non-sel)
!
T1H(non-sel) > T1N

T1N
T1H ( sel) < T1N T1N 1s
T1H (sel) 200ms
T1H (sel)

!
1H Larmor frequency (MHz)
 MP>(A%'4"0I4+%$(#$:4$3#=#$,(0$(/92O7O@C2Q

2D BEST-HSQC 2D BEST-TROSY

Average gain: ~ 20%

Ubiquitin (5C, 800 MHz, )c 8 ns)

MP>(A%'4"0I4+%$(#$:4$3#=#$,(0$(/92O7O@C2Q

i-HNCA BEST-HSQC BEST-TROSY

(( " 100ms) Average gain: ~ 110%

S/N Gain

Correlation peak

HNCO BEST-HSQC BEST-TROSY

(( " 60ms) Average gain: ~ 70%

S/N Gain

Correlation peak
 MP>(A%'4"0I4+%$(#$:4$3#=#$,(0$(/92O7O@C2Q

Application to a 138-residue (16 kDa) protein

BEST-TROSY versus BEST-HSQC BEST versus standard

(iHNCA, 800 MHz, (iHNCA, 800 MHz,

Trec= 200 ms) Trec= 200 ms)
Intensity gain

Peak intensity [a.u.] Peak intensity [a.u.]

BEST-TROSY versus standard

Sensitivity enhancement of up to an order of magnitude !!

Weak signals are enhanced more !!

RAA'034+%$(,%($-3'#03(430.*E(#H4=A'#(N7&%$.(,"4$*D#"

800 MHz, coldprobe, 25C

concentration: 0.9 mM
Exp. time: 20 minutes

tRNAVal
 /92O7,BA#(#HA#"0=#$,*(S4'"#4.B(0$(,:#(,%%'(&%HT

2D BEST-HSQC, 2D BEST-TROSY
(Schanda et al., JACS (128) 2006, 9042; Farjon et al., JACS (131) 2009, 8571;
Favier & Brutscher, J Biomol NMR (49) 2011, 9)
#Sample quality control, titrations, chemical shift mapping
3D & 4D BEST-HNC triple resonance experiments
(Schanda et al., JACS (128) 2006, 9042; Lescop et al., JMR (187) 2007, 163;
Rasia et al, J Biomol NMR (51) 2011, 369)
# Sequential resonance assignment, backbone RDCs
BEST-TROSY-HNN
(Farjon et al., JACS (131) 2009, 8571)
# trans-H-Bond couplings in RNA/DNA
BEST-HMQC2
(Schanda et al., J Biomol NMR (38) 2007, 47)
# measurement of HN-HN RDCs
BEST-15N-relaxation dispersion and BEST-EXSY experiments
(Kern et al., unpublished)
# conformational exchange

MN(A%'4"0I4+%$(#$:4$3#=#$,(0$(N?UV(#HA#"0=#$,*

Ernst-angle excitation

% HMQC-1: %opt =180-%

90
HMQC-2: %opt =2*JHX(
60
% opt
40 Ernst-angle excitation
20 cos %opt = exp(- trec/ T1)

0.5 1.0 1.5 2.0 2.5

trec / T1
 O:#(2C;R2O(N?UV(#HA#"0=#$,

SOFAST: Band-Selective Optimized-Flip-Angle Short-Transient

Combining the advantages of

Longitudinal-relaxation enhancement
and Ernst-angle excitation !
6

Sensitivity [a.u.]
+ > 90 180 4

H
standard se-HSQC
2 SOFAST 90
N
SOFAST 120
Gz SOFAST 150

0
0 1 2
scan time [sec]

O:#(2C;R2O(N?UV(#HA#"0=#$,

Variable-flip angle band-selective excitation

* Most pulse shapes optimized for 90 or 180 flip angles

Example: E400B

45 45
90 90
Hz 135
Hy 135

- typical behavior of shaped pulses for changing B1 field

O:#(2C;R2O(N?UV(#HA#"0=#$,

Variable-flip angle band-selective excitation

Exception : PC9 excitation pulse (Kupce)

30

50
Iz polarization

70

90

110

130

Hz 150
170

offset (kHz)

O:#(2C;R2O(N?UV(#HA#"0=#$,

standard SOFAST ST-SOFAST

Schanda & Brutscher, JACS (127) 2005, 8014. Gal et al, J Biomol NMR (43) 2009, 1.
Schanda et al, J Biomol NMR (33) 2005, 199.

- highest sensitivity - no 15N decoupling, 2 sensitivity loss

- gradient coherence selection

FTA-SOFAST J-SOFAST
Kupce & Freeman, MRC (45) 2007, 2.
Schanda et al, PNAS (104) 2007, 11257. Mueller, J Biomol NMR (42) 2008, 129.

- for use with a fast mixing device - J-coupling-based selection

- small flip angle - natural abundance samples
 2C;R2O(N?UV(7(WX(*A#3,"4(0$(4(D#Y(*#3%$.*

Ubiquitin (8.6 kDa, 0.2 mM)

tRNAVal (26 kDa kDa, 0.9 mM)

3 sec.
4 sec.

TET-2 (486 kDa, 80 M, ~ 1 mM protomer concentration)

3.4 sec.

;4*,7A-'*0$1(,#3:$0G-#*(7(*-==4"B

BEST
!
most sensitive and fastest 2D experiment
Sensitivity [a.u.]

4
but: MQ line widths, no extension to 3D !
3 ! BEST-HSQC
2 # 3D experiments possible
SQ line widths
1
conventional optimal performance at < 800 MHz
0 ! BEST-TROSY
0 1 2 # 3D experiments possible
Scan time [s]
highest spectral resolution
Sensitivity gain optimal performance at ! 800 MHz
of a factor of 3-10
2,"-3,-"4'(3%=A43,$#**(4'%$1(A#A+.#(3:40$*(

Towards a quantitative measure of structure:

what distinguishes structured and unstructured
polypeptide chains ?

- high proton density - low proton density

- low water accessibility - high water accessibility

9HA'%0+$1(.0Z#"#$3#*(0$(A"%,%$(OM(

non-selective ( )
selective ( )

Amide 1H relaxation enhancement depends on :

* aliphatic protons close in space
* local dynamics
* water-amide 1H exchange
 N9O#"%1#$#0,B(7(2C;R2O[/92O(#HA#"0=#$,*(

Quantification of proton density and water accessibility

along polypeptide chains

180 ref-spectrum sat-spectrum

Hwater/aliph tc

Hamide SOFAST/
BEST
, = Isat/Iref

water
1.0
aliphatic
, Increasing
amide structure
0.0

* Schanda, Forge & Brutscher,

10 8 6 4 2 0 ppm Magn. Reson. in Chem. (2006) 44, S177.

WX(N9O72C;R2O(>?@(
ref-spectrum Aliph-sat-spectrum

Ubiquitin (2 mM, 25C, 600 MHz)

Acquisition time: 10 s / spectrum

Structural and dynamic heterogeneities along polypeptide chains

with assigned 1H, 15N resonances

Similar information to heteronuclear NOE

but obtained in significantly reduced experimental time
 WX(N9O72C;R2O(>?@(

Structural and dynamic heterogeneities along polypeptide chains

prior to resonance assignment

,noe dispersion profiles

,noe

MX(N9O72C;R2O(>?@(
Following subtle structural changes upon metal binding

HETnoe: Changes in spin diffusion efficiency

Ca2+
Ca2+

%-lactalbumin
The structure of apo and holo
forms are very similar HETex: Changes in solvent accessibility
Metal binding was reported to
induce some structure
rigidification

HET-SOFAST NMR
experiments can be used
to follow small changes of
protein compactness
 8"%,#0$(D%'.0$1(A4,:Y4B*(4$.(D%'.0$1(0$,#"=#.04,#*(

Unfolded state (U state)

Slow folding phases

Free energy

Breakage of non-native interactions

Isomerization of S-S bonds
Isomerization of prolines

fast

Long-lived folding
slow Intermediate (I-state) I - and N-states
Different structure
(G Different function
Native state Different pathogenicity
(N-state)

! I-state may be sparsely populated under native conditions

! Importance of obtaining structural information on high-energy states

;%'.0$1(%D(4$(4=B'%0.%1#$03(A"%,#0$E(+W7=03"%1'%&-'0$

Unfolded Intermediate(s) ? Folded

Fibrils

(Dialysis-related
amyloidosis)
 @#4'7+=#(WX(>?@(,%(*,-.B(A"%,#0$(D%'.0$1

(1) The folding process is started by a sudden pH 2

change of the sample conditions, e.g. pH jump

Fast injection N
device I
U
Refolding buffer
Injection of Start of NMR
refolding buffer data acquisition

Protein in pH 7
unfolding buffer
fast
slow

(2) NMR spectra are acquired U

while the folding process is occurring I
N

@#4'7+=#(WX(>?@

0 0 1 2
time [sec]
3 4

1
]
[s

2
me

3
Ti

4
 @#4'7+=#(>?@(*,-.B(%D(/W?(D%'.0$1
pH 2 pH jump pH 7
t=0 Final spectrum

U I N ?

I0

Kinetic trace
I1
for each detected peak

0
10
]

20
[s
me

30
Ti

40 N

;%'.0$1(=%.#'(D%"(/W?(
Fit to experimental data
Head-to-head dimers
I2, IN, N2 kiso = 0.0008 s "1; K I = 2400 M "1; K MIX = 2000 M "1; K N = 360 M "1

NMR

KI " 7 KN !
50 M 150 M

kISO kISO
I2 IN N2

! KI KMIX KN
190 M 380 M 750 M
kISO
I N

Monomeric states
SAXS
I, N 750 M 250 M

! I-state: further characterization

by real-time NMR methods
 XB$4=03(0$D%"=4+%$(D"%=("#'4H4+%$(=#4*-"#=#$,*

(1) HET-SOFAST

N state
I state

! No significant changes between I- and N-states dynamics on sub-ns time-scale

(2) R2 BEST-TROSY

! Conformational exchange dynamics (s to ms time scale) close to PRO32

?4AA0$1()7*,4,#(0$,#"43+%$(Y0,:('014$.*

- ANS
N-state

+ ANS
15N

1H

- ANS T4
I-state

+ ANS Q89

15N

V85
L87

ANS interacts with the FG loop of B2M

1H
in the I-state (but not in the N-state)
ANS (1-Anilino-8-Naphthalene Sulfonate)
 >?@(*A#3,"%*3%AB(%D()X8*

! Limited sample stability !'%&-'4"(A"%,#0$ )X8

! Low concentrated samples

due to aggregation problems

! Reduced spectral dispersion, especially in 1H dimensions

! Chemical exchange of labile (amide) and water 1H
! Repetitive primary sequence containing many PRO residues

! Very limited structural information from 1H-1H NOEs

! Large ensemble of conformations interconverting on
sub-nanosecond time scale instead of single compact structure

MN(J%$10,-.0$4'(@#'4H4+%$(9$:4$3#=#$,(0$()X8*

nonsel T1
pH 2.0, 5C pH 7.5, 5C
WFB T1
Sel T1
1.5
T1 (s)

T1 (s)

1.0

0.5
1H

1H

0
10 45 60 85
residue no

,enh = 3 T1 " 0.9 s ; T1 " 300 ms ,enh = 17 T1 " 1.5 s ; T1 " 90 ms

pH 6.5, 5C pH 7.5, 20C

T1 (s)
T1 (s)

1H
1H

peak no

,enh = 4.5 T1 " 0.9 s ; T1 " 200 ms ,enh = 38 T1 " 2.3 s ; T1 " 60 ms
 /92O7O@C2Q(<#"*-*(/92O7N2UV 800 MHz

! Increased resolution and sensitivity

BT-HNCO BH-HNCO

NS5A BASP-1

Solyom, Schwarten, Geist, Konrat, Willbold, Brutscher,

BT: BEST-TROSY
J Biomol NMR(55) 2013, 311. BH-BEST-HSQC

2#G-#$+4'(>?@(4**01$=#$,(%D()X8*(

12343*5678("(98($:;(&6< 123=34215678("(98($:;(&6<
 /43\&%$#(>?@(4**01$=#$,(%D(WP]7"#*0.-#()X8(
(ppm)
1H

15N (ppm)