EMBO Practical Course NMR, Biozentrum Basel 2013

Design, principles and building blocks of

heteronuclear NMR pulse sequences

Michael Sattler

http://www.nmr.ch.tum.de
http://www.helmholtz-muenchen.de/stb/

Prog. NMR Spectrosc. (1999) 34, 93-158.

Contents

Heteronuclear NMR: motivation

Basic pulse sequence elements and 2D correlations
RF pulses: calibration, selective pulses
Water-flip back, sensitivity enhancement, coherence selection
Triple resonance experiments
[Isotope editing and filtering]
[Large proteins]
Structure determination by NMR

NOE-based assignment strategies

 1J- and 2J-couplings in proteins

Speed/efficiency of magnetization transfer

Homonuclear Heteronuclear
magnetization transfer magnetization transfer
via 3J coupling via 1J couplings

1H 1H 1H 1H
10Hz

| | | |
140Hz

140Hz

C C 13C 13C
35Hz

100 ms 35 ms

S/N ~ N exc det3/2 B03/2 NS1/2 T2

Isabella Felli
lecture
Insensitive nuclei enhanced by polarization transfer INEPT, but see also:
 Assignment based on J-correlations

Multi-dimensional NMR experiments

To resolve signal overlap with increasing molecular weight
1/2 loss of S/N per indirect dimension
potential problems: time needed for sampling

3D FT NMR I S T I S
S T S
preparation mixing t1 mixing t2 mixing t3 detection

13C

15N

15N
2

3
1H 1H
 Product operator formalism
Iz Izt Iz
Chemical shift
Ix,y Ix,y cos t Iy,x sin t

Iz Iz cos Ix,y sin

Pulses Ix,y I y,x Ix,ycos + Iz sin
rft

Iy,x Iy,x

Iz Ix,y Pulse Educt Product

/2 I y,x
90 pulse x y z & cyclic
Ix,y + Iz x z y permut.

Iz Iz
IxSy JIS2IzSz t IxSy
Scalar coupling
Ix,y Ix,y cosJISt 2Iy,xSz sinJISt
2Ix,ySz 2Ix,ySz cosJISt Iy,x sinJISt

Ix,y Ix,y cos t Iy,x sin t

 y,x Flip angle
Iz Iz cos Ix,y sin
rf p B1p
90 pulses:
90 pulse:
pulse educt product
x y z p90 = /2 /(2rf)
y x -z Radial velocity: [rad/s] = 2
x z -y = 1/(rf)
Frequency: [cycles/s=Hz]
y z x

2I1yI2z

Iz = (I I)
1 = I + I
Iz = (I I)
1 = I + I

Chemical shift and J-coupling evolution

 How to analyze pulse sequences

180o pulses can be combined if coherence order

magnitude is preserved inbetween
( i.e. no 90o pulse applied)

Heteronuclear polarization transfer

= 1/(2J)

S/N ~ N exc det3/2 B03/2 NS1/2 T2

Basic heteronuclear correlations: HMQC

(1H): refocused

(13C): evolution during t1

J(1H,13C): active during

J(H,H): active !

J(C,C): active !

Relaxation during t1: multiple quantum line-narrowing

Basic heteronuclear correlations: HSQC

(1H): refocused

(13C): evolution during t1

J(1H,13C): active during

J(H,H): not active

J(C,C): active !

Relaxation during t1: T1(1H), T2(13C)

 Basic building blocks: heteronuclear correlation
Relaxation during t1

T2MQ (IxSy) > T2(S)

Methyl TROSY

T1 I-spin (1H)
T2 S-spin (13C)

synchronous decoupling!

T2 S-spin (13C)

Bax et al JMR (1990) 86, 304-318

Transfer amplitudes for antiphase/in-phase

conversion in CH, CH2 and CH3 spin systems

[ms]
J-coupling evolution during the CH: 2HzCy Cx sin(1JH,C )
second half of a refocused INEPT: CH2: 2HzCy Cx sin(1JH,C ) cos(1JH,C )
CH3: 2HzCy Cx sin(1JH,C ) cos2(1JH,C )
 Constant-time HSQC
Set 2T = n/JCC to refocus evolution of homonuclear C,C couplings during 2T

J-coupling evolution: cos(JCC2T)n = 1n

Folding and aliasing in multidimensional NMR spectra

cos[(k/(2SW)+)+] = cos[(k/(2SW)+)]
DW=t=1/(2SW) = cos[4SW-)(k/(2SW)+)2/(t)]

phase difference folded vs. not folded signals:

PHC0 = 2PHC1 = 2/(t)
set PHC1=0, PHC0=90 => folded signals negative

= initial delay for t1(0) determines 1st order phase

= 0th order phase in t1

DW=t=1/SW

exp[i(k/SW+)+i]
= exp[i(SW(k/SW+)iiSW]

phase difference aliased vs. unaliased signals:

PHC0 = 0 PHC1 = 2SW
set PHC1=180 => folded signals negative
 Phase cycling - selection of 13C bound signals
x
rec = 1/(2*1JCH) Phase cycle:
1H
x x x
13C rec x x
0 180

rec
90x, 90 =x (13C)
2IxSz Iy
90x (1H) = (Iy)*(+1) +(Iy)*(1)) = 2Iy
1H-13C: Iz Iy 2IxSz
2IxSz Iy
90x, 90 =x (13C)

rec

90x (1H)
180x (1H);
1H-12C: Iz Iy Iy =0
90x, 90 (13C)

Sample with natural abundance of 13C:

1H-12C:
1J Int = 99%
CH

1H-13C:

Int = 0.5%

with phase cycle without phase cycle

Pulse calibration: considerations

What sample to use for pulse calibration H2O, protein, urea?

H2O: large signal, very sensitive, optimize on FID
radiation damping, different NMR properties than biomolecule
protein: NMR signals of interest
poor S/N
urea: high sensitivity, isolated signals
different NMR properties than actual sample

Calibration as 90, 180, 360 pulse?

Problems: radiation damping, B1 inhomogeneity, amplifier power drop

Problems with cryoprobes

salt concentration, B1 inhomogeneity, water suppression
 Pulse calibration: B1 inhomogeneity
Pulse calibrated on bulk
Calibration of 180 or 360 pulse
off-resonance effects
B1 inhomogeneity
180
Pulse calibration:
Center: [Z-shim detuned] Bulk [Z tuned] ERROR
90: 61us (expected) (app. 90)
180: 123us (2*90=122us) 132us (66.0us) 8%
360: 248us (4*90=244us) 257us (64.3us) 5%
360

180

1H nutation experiment
Z-shim detuned
0.1% Ethylbenzene
1800
500 MHz Cryoprobe
360

sample
NMR tube

sample
Jerschow & Bodenhausen JMR (1999) 137, 108-115.

Pulse calibration: recommendations

calibrate 360 pulse for 1H

bulk pulse closer to correct pulse at the center of rf coil

use Shigemi (solvent susceptibility matched) tubes

reduced effects of B1 inhomogeneity

reduced problems of higher salt concentrations

optimize individual pulses in multipulse sequences

i.e. p2 p1*2,

Nutation experiments (Wu & Otting, JMR (2005) 176,115119)

 Pulse calibration

Problem: no phase cycle Phase cycling

1H-12C signals not suppressed! 1H-12C signals are suppressed!

1D 13C spectrum of a protein

C' Carom. Caliph.

13C 160.0 120.0 80.0 40.0 [ppm]

 Selective RF pulses: rectangular pulses

Flip angle of RF pulse:

13C 13C
= 1p = B1p

pulse:
eff = 4 = (1eff)2 = (41)2 = 12 + ()2 1 = /

p(90) = /1 = / ()

pulse:
eff = 2 = (1eff)2 = (21)2 = 12 + ()2 1 = /

p(180) = /1 = / ()

Radial velocity: [rad/s] = 2

Frequency: [cycles/s=Hz]

Band-selective RF pulses: shaped pulses

Rectangular pulse Shaped pulse

(Ai p/NP)

p = pulse width
B1max /2 NP = number of points in shape
B1rect/2 Ai = relative intensity, 01

p p

= B1rect p = B1max Ai p
= B1 rect/2 p = B1rect Ai p
 Band-selective RF pulses: rectangular vs. shaped
pulses

Selective excitation Selective inversion

1.0 1.0
90o 0.8
Caliph. rectangular 0.6 C'
0.8
pulse
0.4 180o
0.2 rectangular
0.6 pulse
Mxy Mz 0
G4-pulse
0.4 0.2
0.4
0.2 C'
0.6 Caliph.
0.8 G3-pulse
0 1.0
25 20 15 10 5 0 -5 -10 -15 -20 -25 25 20 15 10 5 0 -5 -10 -15 -20 -25

/2 [kHz] /2 [kHz]

Universal rotations time-reversed pulses

Universal rotations: same behavior as hard pulse, independent of x,y,z direction
Point-to-point: optimized for a specific trajectory

e-Burp2 4.94 ms 829.2 Hz G4 7.82 ms 598.5 Hz

100

Mz -> My 50

My -> Mz
0

time rev. -50

My -> Mz
[%]

u-Burp 4.66 ms 2174.2 Hz Q5 6.18 ms 742.1 Hz

100

50

-50

[%]

[kHz] 2 1 0 -1 -2 -3 [kHz] 2 1 0 -1 -2 -3

W. Bermel, Bruker
 Band-selective RF pulses

Selective excitation Selective inversion, Adiabatic inversion

(90) refocusing (180)

Gaussian I-BURP Hyperbolic

(90 / 270) secant

E-BURP RE-BURP WURST

G4 G3 CHIRP

Q5 Q3

References (shaped pulses and band-selective decoupling):

JMR (1991) 93, 93; Chem. Phys. Lett. (1990) 165, 469; JMR (1992) 97, 135;
JMR (1992) 100, 604; JMR (1993) A102, 364; JMR (1995) A115, 273; JMR (1996) A118, 299 .

Band-selective RF pulses - adiabatic pulses

Adiabatic fast (relative to T1, T2) passage:
keep magnetization and rf field colinear: |d/dt| eff
frequency sweep: d/dt (non-linear pulse phase modulation)

broad-band inversion with low power
insensitive to B1 inhomogeneity
adiabaticity requires long P

Amplitude Phase

WURST

G3, 250s, B1max/2 14kHz

Mz

WURST, 2.7ms, B1max/2 3kHz

[Hz]
 Adiabatic pulses

Magnetization
trajectory

eff

Cavanagh, Palmer book

Off-resonance pulses: phase- and amplitude modulation

G3, 500s Mz

no modulation

Phase modulation:
Mz

imod = i (2 pi/NP)

Amplitude modulation:
Mz

Aimod = Ai 2cos(2 pi/NP)

[Hz]
 Bloch-Siegert phase shifts (BSP)

13C 13C

180-pulse:
Mx 1

1 eff

1 0 1
Off-resonance: 1
BSP(0) = * = 3 1 * /1 = 3 48.2
Precession around 1eff :
off-reson <1eff> p = eff
(13C)
Instead of precession
around z-axis with:
free = p BSP on 13C
transverse
BSP = ( <1eff>) p magnetization

Chem. Phys. Lett. (1990) 168, 297

Bloch Siegert shift: inversion simulation

Mz -> Mz
Q3 (2 msec, 1650.4Hz, freq: 2kHz, -3kHz)
100

50

-50

[%]

[kHz] 4 0 -4 -8

W. Bermel, Bruker
 Bloch Siegert shift: inversion simulation
Mz -> Mz
Q3 (2 msec, 3300.8Hz, freq: 2kHz, -3kHz)
100

50

-50

[%]

[kHz] 4 0 -4 -8

W. Bermel, Bruker

BSP compensation
120
100
80
60
Bloch-Siegert phase
for a band-selective BSP 40 BSP = 12/(2*P
( 5kHz) G3 pulse [] 20
0
Calculate BSP for each subpulse
-20
of a shaped pulse and add it to
-40
the phase to compensate for the
-60
BSP.
-80
10 12 14 16 18 20 22 24 26 28 30

[kHz]

Determine phase empirically Use amplitude-modulated Use additional pulse Intrinsic correction
(0th and 1st order) inversion pulse (0th order) (all orders are corrected) (all orders are corrected)

+BSP(0)
t1 T T t1 t1 T T t1 t1 T T t1 C t1 T T t1
2 '
2
2 2 C 2 2
C 2
2

C' BSP Single shape with two

excitation bands and intrinsic
G3, 250 s, 18 kHz G3, 700 s, 18 kHz
BSP correction
phase modulation amplitude modulation

JMR (1992) 100, 604; JMR (2000) 146, 369.

 BSP compensation

Bloch Siegert shift: inversion simulation

Mz -> Mz
Q3 (2 msec, 3300.8Hz, freq: 2kHz, -3kHz)
100

50

-50

with BS compens.
without
[%]

[kHz] 4 0 -4 -8

W. Bermel, Bruker

Water-flip-back
Daniel Nietlispach
Lecture

1, -1 Jump-return WATERGATE with water-flip-back

excitation null on-resonance H2O: Iy Iz Iy -Iy Iz
Protein Iz Iy Iy
 HSQC with WATERGATE & water-flip-back

water-flip-back

H2O: z -y y z -z z z

Suppress radiation damping

of H2O signal during t1

Sensitivity enhancement

90x (S) t1 2IzSy cost1 + 2IzSx sint1

RSH
amplitude modulation
90y (S) 2IzSx cost1 + 2IzSy sint1

Coherence order selective

coherence transfer COS-CT

Echo/anti-echo
Phase modulation

Linear combination of echo and antiecho pathways produces absorptive lineshape!

 Sensitivity enhancement / gradient coherence selection

Standard HSQC
no gradient coherence selection
S/N = 1

Standard HSQC
gradient coherence selection
S/N = 1/2

Sensitivity enhanced HSQC

gradient coherence selection
S/N = 2

S/N=: I/n (I: (intensities), n: # signals)

Sensitivity-enhanced HSQC with water-flip-back

Heteronuclear gradient echo: j{ (GzG)j Sk( pkjk) } = 0

water-flip-back
H2O: z -y y z -z -y y y -y -z z

Suppress radiation damping

of H2O signal during t1 GzG(15N)*15N* = GzG(1H)*1H

p15N = 1; p1H = -1
 Water-flip-back

SE HSQC WATERGATE HSQC

[Note: In 2D S/N will be 2 larger]

H2O: +z H2O: +z
H2O: dephased H2O: dephased
H2O: -z H2O: -z

Longitudinal relaxation optimization

faster longitudinal relaxation

band-selective inversion of spins, i.e. amide protons Bernhard Brutscher
Lecture
partially already achieved by water-flip-back
faster repetition rates, i.e. SOFAST experiments

Pervushin et al JACS (2002)124, 12898

Schanda & Brutscher JACS (2005), 127, 8014
Attreya & Szyperski PNAS (2004) 101, 9624.
Deschamps & Campbell JMR (2006) 178, 206
 Sensitivity enhancement

Water-flip-back (for amides affected by solvent exchange)

Longitudinal relaxation optimization
Optimized coherence transfer
sensitivity enhancement (coherence order selective coherence transfer)
double sensitivity enhancement
TROSY (spin-state-selective coherence transfer)
MQ line-narrowing (methyl TROSY = HMQC)
Simultaneous acquisition, i.e. 1H-13C, 1H-15N correlation
Fast data acquisition

Simultaneous 13C/15N,1H HSQC

1H X transfer can be optimized simultaneously for 13C and 15N

but: some relaxation loss for 13C due to longer delay.
poor water-suppression, since E/AE cannot be implemented without sensitivity loss.
building block for simultaneous 3D/4D NOESY experiments.
JBN (1994) 4, 201-213; JMR (1994) B103, 197-201.
Simultaneous sensitivity enhancement

Simultaneous sensitivity enhancement

JBN (1995) 5, 97-102.

Simultaneous sensitivity enhanced HSQC in H2O

JBN (1995) 5, 97-102.

Basic building blocks: spin-state-selective filters

IP AP

JMR (1998) 131 373.

 Basic building blocks: TROSY
Sensitivity Enhanced TROSY

Kay TROSY

Basic building blocks: optimal TROSY

Nietlispach, JBNMR (2005) 31: 161

 Basic building blocks: optimal TROSY
Christian
a) Kay TROSY b) Nietlispach TROSY
Griesinger Lecture
Kay
Nietlispach

Nietlispach, JBNMR (2005) 31: 161

Multi-dimensional NMR experiments

To resolve signal overlap with increasing molecular weight
1/2 loss of S/N per indirect dimension, but increased resolution

3D FT NMR I S T I
S T
S
preparation mixing t1 mixing t2 mixing t3 detection S

13C

15N

15N
2

3
1H 1H
 sin2( 1JH,C ') exp('/T2H)

Transfer amplitude * sin( 1JC,N 2) cos( 2JC,N 21) cos( 1JC,C 21) exp(21/T2C)
including T2 relaxation sin( 1JC,N 22) cos( 2JC,N 22) exp(22/T2N)
* sin2( 1JN,HN) exp(-/T2HN)

sin4( 1JN,HN ) exp(-2/T2HN)

* sin2( 1JC,N 2) cos2( 2JC,N 2) exp(/T2N)
* cos( 1JC,C t2) exp(t2/T2C)
 Out-and-back vs. transfer experiments
13C evolution:
1
0.8
0.6
(cos(J21) exp(1/T2))dt1
0.4
0.2
0
-0.2
-0.4
0 2 1

1
0.8
(cos(Jt2) exp(-t2/T2))dt2
0.6
0.4
0.2
0
-0.2
-0.4
0 21/3 2 1

Semi-constant-time chemical shift evolution

t1 t1
I 2 2 2 2 I t1a t1b t1c
J-coupling =:
Chemical shift =: t1
S S
real time semi-constant- time
t1a t1b t1c = t1
J: t1a t1b t1c= (1)
With the requirement that t1(0) = 0, Eq. (1) yields:
t1a(0) = t1c(0) = /2 and t1b(0) = 0. (2)
For a FID along t1 that is to be digitized by TD data points, two relations can be written for the
increments t1a, t1b and t1c (t1=1/SWH, where SWH is the spectral width in Hz):
t1a t1b t1c = t1
J: t1a t1b t1c= 0 (3)
Since it is required that t1c(TD) 0;. this yields (for t1c(TD) = 0):
t1a = t1/2; t1b = t1/2 t1c and t1c = t1c(0)/TD (4)

Fulfilling Eqs. (2) and (4) assures chemical shift evolution with t1(0) = 0 < t1 < t1max and
evolution of the coupling during for all increments.
Note, that t1c is negative, reflecting the fact t1c is decremented.
 HCCH-TOCSY

H(t1) C(t2) C C H(t3)

usually recorded in D2O for H detection

Double sensitivity enhanced HCCH-TOCSY

Can be recorded in H2O due to excellent water suppression
by heteronuclear gradient echo

J. Biomol. NMR (1995) 6, 11-22.

Relative sensitivity of triple resonance experiments

Experiment Assignment Comment Relative

S/N [%]
2
HNCO H(i), N(i), C(i-1) <20 kD, above use H labeling 100
2
HNCA H(i), N(i), C(i),C(i-1) <20 kD, above use H labeling 50/15
2
HN(CO)CA H(i), N(i), C(i-1) <20 kD, above use H labeling 71
2
HN(CA)CO H(i), N(i), C(i) <20 kD, above use H labeling 13/4

13/9
2
CBCA(CO)NH H(i), N(i), C(i-1), C(i-1) <20 kD, above use H labeling

13/9
2
HBHA(CO)NH H(i), N(i), H(i-1), H(i-1) <20 kD, above use H labeling

4/1.7 (i)
2
CBCANH, H(i), N(i), C(i), C(i), <15 kD, above use H labeling
HNCACB C(i-1),C(i-1) 1.3/0.5
i-1)
2
(H)CC(CO)NH- H(i), N(i), Caliph.(i-1) <15-20 kD, above use H labeling
TOCSY
2
H(CC)(CO)NH- H(i), N(i), Haliph.(i-1) <15-20 kD, above use H labeling
TOCSY
HCCH-TOCSY Haliph., Caliph. <25 kD, - sensitive, but tedious to analyze,
combine with HCCONH type experiments

Heteronuclear NOESY experiments

water-flip-back during NOE mixing time due to radiation damping

1 = 45o to enhance water flipback independent of TPPI on 1
 Heteronuclear NOESY experiments

HMQC rather than HSQC for 13C

edited NOESY to minimize off-
resonance effects for 13C rf pulses
Alternatively use adiabatic
inversion pulses, i.e. at higher fields

Tips and tricks for triple-resonance experiments

Multiple INEPT: ABCDCBA
Impossible to cycle each pulse independently.

instead:

Adapt sequence to J-spin system

Band-selective pulses

Selection of pathway: difference experiment

e.g C D

Cycle 90 pulse where evolution period.

Remove 180 imperfections: gradients (no sensitivity loss)

Purge while along z.

gradient coherence selection in t1 only with SE

Spectral purity higher when gradient selection

 Tips and tricks for triple-resonance experiments

Example 3: 3D CT HNCA: gradient selection with sensitivity enhancement and water flip-back
H2O: +z
flip-back water control

1 = (x, x) select H N transfer

band-selective
pulses 2 = [x, x, x, x] select N C transfer

3 = 4(x), 4(x) reduced EXORCYCLE

rec = [(x, x) (x, x)]