SOLID STATE NUCLEAR MAGNETIC RESONANCE STUDIES OF STRUCTURES
AND MEMBRANE LOCATIONS OF PEPTIDES
By
Li Xie

A DISSERTATION
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of
Chemistry - Doctor of Philosophy
2014

ABSTRACT
SOLID STATE NUCLEAR MAGNETIC RESONANCE STUDIES OF STRUCTURES
AND MEMBRANE LOCATIONS OF PEPTIDES
By
Li Xie
Solid state nuclear magnetic resonance (SSNMR) can be used to study the structures
of molecules such as small peptides and large proteins. Different structures correspond
to different chemical environments and thus yield different chemical shifts in NMR
spectra. In addition, SSNMR can also be used to probe membrane locations of peptides
and proteins and provide insights into their biological functions. This dissertation mainly
focuses on the structural and membrane location studies of peptides by rotational-echo
double-resonance (REDOR) SSNMR, which is a technique for measuring distances
between two coupled hetero nuclei such as

13

C and

15

N.

Influenza fusion peptide (IFP) is the N-terminal peptide of the HA2 subunit of the
influenza hemagglutinin (HA) protein and this peptide plays an important role in the
membrane fusion between the virus and the endosome of the host cell. There are 15
different HA subtypes. I studied the structure of membrane-associated H3 subtype IFP
(H3_IFP) by

13

15

C- N REDOR SSNMR. SIMPSON simulations of the data indicate that

H3_IFP adopts predominantly closed and semi-closed structures in membranes.
Similar to IFP, human immunodeficiency virus (HIV) fusion peptide (HFP) is the
N-terminal peptide of the viral gp41 fusion protein and this peptide plays a key role in
the HIV-host cell membrane fusion. HFP adopts major antiparallel and minor parallel β

sheet structures in membranes. Earlier fluorescence spectroscopy and SSNMR studies
support a strong positive correlation between the membrane insertion depth and
fusogenicity of HFP. However, the

19

F labeling in earlier membrane location studies of

HFP may perturb the membrane bilayer integrity. Due to this concern,

13

2

C- H REDOR

was developed to detect the residue-specific membrane location of HFP, where one
residue of HFP is backbone

13

CO labeled and the lipid is either perdeuterated or
1

2

selectively deuterated in its acyl chains. Since H and H are chemically equivalent,
there is no perturbation on membrane bilayer integrity as well as peptide-lipid interaction
regardless of what fraction of deuterated lipid is used. The

13

2

C- H REDOR pulse

sequence was optimized using the setup peptide I4.
The membrane locations of two peptides, HFP and KALP, were studied by
2

13

C-

H REDOR, where KALP is a designed transmembrane α helical peptide. Fitting of the

REDOR data revealed that both peptides have multiple locations within the membrane
hydrocarbon core. The multiple locations are attributed to the snorkeling of lysine
sidechains on both termini for KALP and to the distribution of antiparallel β sheet
registries for HFP. HFP has a ~0.7 fraction of deep insertion and a ~0.3 fraction of
shallow insertion in the membrane. The predominant deep insertion of HFP may
significantly perturb the membrane bilayer structure and lower the activation energy of
membrane fusion, which is consistent with the observed positive correlation between
the membrane insertion depth and fusion rate for different HFP constructs.

Dedicated to Qingwei Li and Elaine Q. Xie

iv

ACKNOWLEDGMENTS

First of all, I would like to thank my research advisor, Prof. David Weliky, for his
tremendous guidance, support, and encouragement throughout my Ph.D. career. He
taught me not only NMR principles but also how to think as a real scientist. He has been
very patient and I do not remember for how many times that he has spent a couple of
hours teaching me in his office over the past few years. He always gave me helpful
suggestions and encouragement while I was stuck in NMR troubleshooting. He has
been a very supportive advisor and I really appreciate this.
I would also like to thank my committee members, Prof. Daniel Jones, Prof. John
McCracken, and Prof. James Geiger, for their valuable inputs and guidance over the
past years. Prof. Daniel Jones also gave me very helpful suggestions for job searching,
and he has always been generous with his time and insights.
I am grateful for the great help from Dr. Daniel Holmes and Mr. Kermit Johnson at
the Max T. Rogers NMR facility center at MSU. Dr. Daniel Holmes has been a very
good source for discussing solution state NMR and troubleshooting the instrument.
I am also grateful for the support and help from both the former and current
Weliky group members, Dr. Kelly Sackett, Dr. Charles Gabrys, Dr. Matthew Nethercott,
Dr. Scott Schmick, Dr. Erica Vogel, Koyeli Banerjee, Punsisi Ratnayake, Ujjayini Ghosh,
Shuang Liang, Lihui Jia, and Robert Wolfe. They have been very helpful not only in
research but also in many other aspects.
Finally, I would like to thank my wife, Qingwei Li, for her support and unselfish
love during my Ph.D. career. It is her support and encouragement that have made my
v

graduate study much easier than it would have been on my own. I would also like to
thank my family and friends for their support and help in many different ways.

vi

TABLE OF CONTENTS

LIST OF TABLES..............................................................................................................x
LIST OF FIGURES...........................................................................................................xi
KEY TO ABBREVIATIONS............................................................................................xxi
Chapter 1 - Introduction.................................................................................................1
1.1 NMR background......................................................................................................1
NMR interactions....................................................................................................1
Zeeman interaction............................................................................................2
Interaction with B1 field (r.f. pulses)...................................................................5
J-coupling interaction.........................................................................................7
Isotropic and anisotropic chemical shift interactions........................................10
Dipolar coupling interaction.............................................................................14
Quadrupolar interaction...................................................................................17
Rotating frame......................................................................................................22
Magic angle spinning (MAS).................................................................................26
REDOR.................................................................................................................27
1.2 Influenza and HIV fusion peptides........................................................................34
Influenza fusion peptide (IFP)...............................................................................34
HIV fusion peptide (HFP)......................................................................................36
1.3 Membrane insertions of peptides.........................................................................41
Earlier studies of membrane insertion depths of IFP and HFP.............................41
Techniques for measuring membrane insertion depths of peptides.....................42
(a) Fluorescence resonance energy transfer (FRET)......................................42
(b) Paramagnetic relaxation enhancement (PRE) SSNMR.............................43
(c) Spin diffusion..............................................................................................44
(d) REDOR SSNMR........................................................................................46
(e) Electron paramagnetic resonance (EPR)...................................................48
REFERENCES...............................................................................................................49
Chapter 2 - Structural studies of Influenza fusion peptide (IFP)..............................55
vii

2.1 Background.............................................................................................................55
2.2 Sample preparation................................................................................................59
Peptide sequence and synthesis..........................................................................59
REDOR sample preparation.................................................................................60
2.3

13

15

C- N REDOR experiments................................................................................61
REDOR pulse sequence and parameters............................................................61
REDOR spectra and dephasing curve of ∆S/So vs τ............................................64

2.4 Result discussion and conclusions......................................................................65
exp

Natural abundance calibration of (∆S/So)
lab

SIMPSON simulations of (∆S/So)

........................................................65

vs τ..............................................................69

Intramolecular vs intermolecular dipolar interactions............................................73
Conclusions..........................................................................................................76
REFERENCES...............................................................................................................79
Chapter 3 - Development of

13

2

C- H REDOR..............................................................83

3.1 Probe design...........................................................................................................83
Low-power and high-power tuning........................................................................83
Tuning configuration.............................................................................................84
3.2 Setup peptide I4......................................................................................................88
Criteria for REDOR setup compound...................................................................88
I4 peptide sequence and isotopic labels...............................................................90
3.3 Pulse sequence optimization................................................................................90
(a) MAS setup using KBr......................................................................................93
(b)

13

C chemical shift referencing using adamantane..........................................94

(c) CP optimization...............................................................................................97
(d)

13

C π pulse optimization................................................................................100

2

(e) H π pulse optimization.................................................................................102
Comparison of REDOR pulse sequences..........................................................104
3.4 I4 data fitting.........................................................................................................109
exp

Natural abundance calibration of (∆S/So)
lab

SIMPSON simulations of (∆S/So)
lab

Exponential fitting of (∆S/So)

......................................................109

vs τ............................................................110

vs τ...................................................................112

REFERENCES.............................................................................................................113
viii

Chapter 4 - Membrane location studies of KALP and HIV fusion peptide (HFP)..115
4.1 Background...........................................................................................................115
4.2 Peptide synthesis and sample preparation........................................................118
Peptide sequence and synthesis........................................................................118
REDOR sample preparation...............................................................................119
4.3

13

2

2

C- H REDOR experiments and H T1 measurements....................................120
Part I:

13

2

C- H REDOR experiments...................................................................120

KALP_A11C samples.....................................................................................121
HFP_F8C samples.........................................................................................124
HFP_G5C samples........................................................................................126
2

Part II: H T1 measurements..............................................................................128
2
2

H spectra under static conditions (no spinning) ............................................129
H spectra under 10 kHz MAS........................................................................134

Data fitting......................................................................................................136
4.4 Result discussion and conclusions....................................................................143
Membrane locations of KALP.............................................................................145
Membrane locations of HFP...............................................................................150
Conclusions........................................................................................................158
REFERENCES.............................................................................................................159
Chapter 5 - Summary and future work......................................................................162
Summary of experimental results and future work of IFP...................................162
Summary of experimental results of KALP and HFP and future work of HFP....165
REFERENCES.............................................................................................................168
APPENDICES...............................................................................................................170
APPENDIX A NMR file locations.................................................................................171
APPENDIX B Fmoc solid phase peptide synthesis (SPPS)........................................176
APPENDIX C HPLC purification of peptides...............................................................183
APPENDIX D NMR troubleshooting............................................................................189
REFERENCES.............................................................................................................195

ix

LIST OF TABLES

Table 2.1

13

15

CO- N distances in the closed structure of H1_23...................................68
2

Table 4.1 Best-fit lipid H T1 in ms with uncertainty in parentheses ...........................143
Table 4.2

13

2

C- H REDOR fitted parameters................................................................146

x

LIST OF FIGURES

Figure 1.1 The two spin states of a spin-1/2 nucleus in the static magnetic field Bo. The
field Bo is along z axis. The two magnetic moments µα and µβ (blue) are under Larmor
precession about the field Bo with an angular frequency of ωo =
γBo....................................................................................................................................4
Figure 1.2 Rabi precession of the magnetization M (initially along z axis) about the B1
field of a 90x pulse (a) and a 180x pulse (b)....................................................................7
Figure 1.3 J-coupling mechanism. (a) Nuclear and electron spins in the X-Y bond and
the spin energy diagram of nucleus X with (black) and without (blue) the X-Y J-coupling
interaction. (b) Peak splitting of nucleus X in the spectrum due to the J-coupling
between X and Y, where Y is a spin 1/2 nucleus..............................................................9
Figure 1.4 Principal axis frame (PAF) and the shielding tensor (red). The two polar
angles defining the orientation of Bo in PAF are θ and φ, where θ is the angle between
PAF

Bo and the principal axis z
PAF

(xy)

and φ is the angle between the projection of Bo in the
PAF

plane and the principal axis x

. The three principal values associated with

the three axes in PAF are σxx, σyy, and σzz , which also correspond to the three
principal chemical shifts in the CSA (chemical shift anisotropy) powder pattern. The PAF
frame at the lower right corner displays the angles α, β, and γ used in equation 1.15...11
Figure 1.5 (a) The PAF frame of backbone

13

PAF

CO in a peptide or protein. The z
PAF

PAF

is perpendicular to the C—CO—N plane. The x
PAF

CO—N plane with x

and y
PAF

perpendicular to C=O and y

axis

axes are both in the C—

parallel to C=O. (b)

13

CO CSA
PAF

powder pattern. The three chemical shifts are σxx = 247 ppm with Bo parallel to x
,
PAF
PAF
σyy = 176 ppm with Bo parallel to y
, and σzz = 99 ppm with Bo parallel to z
......14
Figure 1.6 Definition of the angle θ and distance r for the dipolar interaction between
nucleus I and S. Bo is the external magnetic field which is along the laboratory frame z
axis.................................................................................................................................15
Figure 1.7 Electric properties of a quadrupolar nucleus. (a) prolate charge distribution.
(b) oblate charge distribution. (c) electric quadrupole moment of (a) with a positive sign
xi

(e.g. it is 2.8 x 10

-31

2

2

m for H). (d) electric quadrupole moment of (b) with a negative
-30

sign (e.g. it is -2.6 x 10

2

m for

17

O)............................................................................18
2

2

Figure 1.8 Orientation dependence of the H spectra for aliphatic C- H bonds. (a) θ =
2

0o. (b) θ = 54.7o. (c) θ = 90o. (d) H Pake doublet for all possible θ values, where θ refers
2

to the angle between the C- H bond and Bo field..........................................................22
Figure 1.9 Quadrature detection scheme in the rotating frame for all modern NMR
spectrometers. The device ADC refers to analogue to digital converter and is used to
convert the NMR signal from a voltage to a binary number which can be stored in the
computer memory...........................................................................................................26

Figure 1.10 The geometry of a

13

15

C- N internuclear vector in a solid state sample

o

under MAS where α equals 54.7 (magic angle)............................................................27
Figure 1.11

13

15

C - N REDOR pulse sequence.............................................................31

Figure 1.12 Dipolar coupling evolution as a function of rotor period during the
dephasing time τ in REDOR. The dipolar interaction is averaged out over each rotor
period by MAS in So experiment but not in S1 experiment.............................................33
Figure 1.13 REDOR universal dephasing curve of ∆S/So vs λ for a spin 1/2 - spin 1/2
pair (panel a) and for a spin 1/2 - spin 1 pair (panel b), respectively. Panel a is edited
from reference [12] and panel b is obtained from SIMPSON simulations......................34
Figure 1.14 Infection model of influenza virus. The virus recognizes the host cell
(respiratory epithelial cell) via the viral HA1 binding protein and then enters the cell by
endocytosis. The cell physiology pumps protons into the newly-formed endosome and
lowers the pH to ~5 to trigger the fusion between the viral and endosomal membranes.
The viral nucleocapsid is then released to the cell cytoplasm........................................36

Figure 1.15 (Left) HIV infection model. (Right) HIV infection process illustrated by
electron microscopy. (a) HIV-host cell membrane binding. (b) hemi-fusion. (c) complete
fusion with the formation of fusion pore. (d) entry of the viral nucleocapsid into the host
cell..................................................................................................................................38

xii

Figure 1.16 (a) HIV-host cell membrane fusion model: (i) native-state trimers of gp120
and gp41 prior to fusion, (ii) pre-hairpin intermediate (PHI), and (iii) final hairpin state. (b)
Different segments of the HIV gp41 fusion protein.........................................................39

Figure 1.17 Crystal structure of the HIV gp41 ectodomain lacking fusion peptide
(gp41528-683, residues are numbered according to their positions in the gp160 complex).
(a) α helical structure of the gp41 ectodomain, including polar region(blue, 528-540),
NHR(blue, 541-581), loop region(blue,582-627), CHR(blue,628-666), and MPER
(grey,667-683). The structure is 88 Å long. (b) Close-up of the MPER region showing
the exposure of aromatic side chains of Trp 678, Trp 680 and Tyr 681 towards the
membrane, which indicates the potential membrane insertion of MPER residues
adjacent to the transmembrane domain of gp41............................................................40

Figure 1.18 Ribbon representation of the HFP30, which is the N-terminal 30 residues
of the HIV gp41 fusion protein, in an SDS micelle..........................................................41
1

Figure 1.19 X-nucleus detected, H spin diffusion pulse sequence. The sequence starts
with an initial saturation (destroy) of any X magnetization by several 90

o

pulses,

1

followed by a dipolar dephasing period. After the H excitation, a T2 filter is applied
1
o
during the period τd to select the H magnetization which has a long T2. The 180 pulse
in the middle of the period τd is applied to refocus chemical shift and field
o

1

inhomogeneity. The 90 pulse at the end of τd stores the remaining H magnetization
with a long T2 on the z axis, followed by a spin diffusion period with a variable tm (e.g.
1

0-100 ms). During the mixing time tm, the existing H z-magnetization is transferred to
1 1
1
nearby protons via H- H dipolar coupling. In the end, a H→X CP is applied and then
1

the X signal is detected with H decoupling...................................................................45

Figure 1.20 (a)

13

13

31

C- P and

13

19

exp

C- F REDOR experimental dephasing (∆S/So)

as a

function of the CO-labeled residue. (b) Membrane insertion models of the anti-parallel
β sheet HFPmn_V2E, HFPmn, and HFPtr, respectively.................................................47

Figure 2.1 Structures of the influenza hemagglutinin fusion peptides H3_20 (a) and
H1_23 (b) in DPC detergents.........................................................................................57
Figure 2.2 N-helix/turn/C-helix structures of the H1_23_G8A mutant in DPC micelles at
pH 7................................................................................................................................58
xiii

Figure 2.3 Backbone structures of the membrane-associated H3_20 peptide at pH 5.0.
Residue E11 is in green, N12 is in red, the hydrophobic residues L2, F3, I6, F9, and I10
are in gold.......................................................................................................................59

Figure 2.4
Figure 2.5
13

13

15

C- N REDOR pulse sequence................................................................62

13

15

15

C- N REDOR dephasing curves of ∆S/So vs τ simulated by SIMPSON

for C- N dipolar couplings of 2 Hz (black) and 52 Hz (red), where 2 Hz corresponds
to an open structure and 52 Hz corresponds to a closed structure of H3_20,
respectively.....................................................................................................................63
13

Figure 2.6 (a) Experimental curve of ∆S/So vs τ of the H3_20 peptide; (b) CO peaks
in the So (black), S1 (red), and ∆S (green) spectra for τ = 32 ms...................................65

∆S lab
∆S sim
) vs τ (black) and best-fit (
)
vs τ for 32 Hz (purple)
So
So
which corresponds to the "single closed structure" model, 51 Hz x 0.64 (blue) which
corresponds to the "closed/open" model, and 65 Hz x 0.41 + 21 Hz x 0.59 (red) which
corresponds to the "closed/semi-closed" model, respectively........................................72

Figure 2.7 Plots of (

Figure 2.8 Backbone structural models of H3_20 in lipid membranes at pH 5.1. The
carbon, nitrogen, and oxygen atoms are in green, blue, and red, respectively. The G16
13
15
CO-F9 N internuclear distance is 3.6 Å in the closed structure (a) and 5.2 Å in the
semi-closed structure (b)................................................................................................73
Figure 2.9 (a) Possible aggregates of H3_20 for the dimer model for the "50% labeled +
13
15
50% unlabeled" sample. The G16 CO and F9 N labeled H3_20 is in red while the
13

15

unlabeled H3_20 is in black. (b) Theoretical
CO- N REDOR dephasing plots of
∆S lab
(
) vs τ for the 100% labeled H3_20 sample (red) and the "50% labeled + 50%
So

∆S lab
) for the 100% labeled sample is twice
So
that for the 50% labeled sample for any given dephasing time τ....................................77
unlabeled" H3_20 sample (black). The (

∆S lab
)
vs τ after n.a. correction for the
So
membrane sample containing 100% labeled H3_20 (red) and the one containing 50%
Figure 2.10

13

15
CO- N REDOR plots of (

xiv

labeled and 50% unlabeled H3_20 (black) at pH 5.0 (courtesy from Ujjayini Ghosh, the
REDOR data for the two samples were acquired under the same optimized
experimental conditions which were different than those for Figure 2.6)........................78
Figure 3.1 Cable connections for low-power tuning. Only part of the sweeper and
oscilloscope are displayed. For the low-power tuning, the oscilloscope needs to be
adjusted such that all the red circled lights are on..........................................................84
Figure 3.2 (a) Triple-resonance solid state NMR probe. (b) Tune tube and tuning rod. (c)
Series plug-in and trap plug-in........................................................................................86
Figure 3.3 Three commonly used REDOR pulse sequences. Each sequence starts with
1
CP from H to the observed nucleus S to enhance the S signal, followed by a
dephasing period (only ten rotor cycles are shown here) and at last an acquisition
1
period. H decoupling pulses are applied throughout the dephasing and acquisition
periods............................................................................................................................91
o

Figure 3.4 Br spectrum (64 scans) after MAS setup, spinning rate = 4 kHz, -50 C. The
spectrum was processed with 20 Hz Gaussian line broadening....................................94
Figure 3.5

13

o

C spectrum of adamantane under 4 kHz MAS, -50 C. The left peak at -

119.0 ppm corresponds to the methylene
corresponds to the methine
Figure 3.6

13

13

13

C whereas the right peak at -127.9 ppm

C.....................................................................................96

C spectra of I4_A8DA9C for "aH" array from 0.50 to 0.70 in an increment
o

of 0.02, 10 kHz MAS, -50 C..........................................................................................98
Figure 3.7
o

C. The A9

13

C spectrum of I4_A8DA9C after CP optimization under 10 kHz MAS, -50

13

CO chemical shift after referencing is 178.8 ppm..................................101
1

Figure 3.8 Pulse sequence of cp_zfilter. The phase cycle is "x, -x" for H π/2 pulse, "-y,
-y, -x, -x" for

13

C CP pulse, "x, x, -y, -y" for

13

C π/2 pulse, and "-x, -x, y, y" for

13

C

1

π pulse, respectively. The H CP and decoupling pulses have a fixed phase of "y"....102
13

2

C- H REDOR experimental dephasing plot of ∆S/So vs τ of I4_A8DA9C
o
under 10 kHz MAS and -50 C.....................................................................................104

Figure 3.9

xv

13

2

13

2

Figure 3.10 C- H REDOR experimental dephasing plots of ∆S/So vs τ of I4_A8DA9C
for the "redorxy8xypi_pm" (black) and "redorxy8ypi_pm" (red) pulse sequences under
o
10 kHz MAS and -50 C...............................................................................................106
Figure 3.11

C- H REDOR So spectra of I4_A8DA9C for the "redorxy8xypi_pm" (a)
o
and "redorxy4xpi_pm" (b) pulse sequences under 10 kHz MAS and -50 C. The number
of acquisitions is 400 and 1000 for the 16 ms and 24 ms So spectra in panel a and 6875
and 13331 for the 16 ms and 24 ms So spectra in panel b. All spectra are processed
with 20 Hz Gaussian line broadening and baseline correction.....................................107
Figure 3.12 Evolution of two individual

13

C magnetic dipole moments a and b due to

anisotropic chemical shift interactions under MAS when applying

13

C π pulses in the

middle and at the end of each rotor period Tr...............................................................108
13

Figure 3.13

2

C- H REDOR dephasing curves of ∆S/So vs τ of I4_A8DA9C for the
o

pulse sequence "redorxy8xypi_pm" under 10 kHz MAS and -50 C. Black squares are
labeled ∆S/So for A9

13

CO after removal of the natural abundance contribution. Red

triangles are SIMPSON simulated ∆S/So for d = 37 Hz. Red line is the best-fit curve
using the fitting function Ax(1-e

-βτ

)...............................................................................111

Figure 4.1 (a) Regular lipid bilayer (left) consisting of unlabeled DPPC and
1
19
interdigitated bilayer (right) consisting of labeled DPPC that contains a H→ F
19

substitution at C16 in one of the two palmitoyl chains. (b) Structure of C16- F labeled
DPPC lipid....................................................................................................................116
Figure 4.2 (a) D54: dimyristoylphosphatidylcholine perdeuterated in the myristoyl
chains. (b–d) D4, D8, and D10: dipalmitoylphosphatidylcholine deuterated at palmitoyl
carbons 2; 7 and 8; and 15 and 16, respectively..........................................................117
Figure 4.3

13

2

C- H REDOR pulse sequence................................................................121

13

2

C- H REDOR So (black) and S1 (colored) spectra for τ = 40 ms for
KALP_A11C in D4, D8, D10, and D54 membranes, respectively. All spectra were
processed
using
100
Hz
Gaussian
line
broadening
and
baseline
correction......................................................................................................................122
Figure 4.4

xvi

13

2

C- H REDOR So (black) and S1 (purple) spectra for τ = 2 ms (a) and τ =
40 ms (b) for KALP_A11C in D4 lipid membrane.........................................................123
Figure 4.5

Figure 4.6

13

2

C- H REDOR dephasing plots of ∆S/So vs τ for KALP_A11C in D4

(purple), D8 (blue), D10 (red), and D54 (green) membranes, respectively. The ∆S/So at
13
each τ was calculated using So and S1 CO intensities determined over a 3.0 ppm
integration width............................................................................................................124
Figure 4.7

13

2

C- H REDOR So (black) and S1 (colored) spectra for τ = 40 ms for

HFP_F8C in D4, D8, D10, and D54 membranes, respectively. All spectra were
processed using 100 Hz Gaussian line broadening and baseline correction...............125
Figure 4.8

13

2

C- H REDOR dephasing plots of ∆S/So vs τ for HFP_F8C in D4 (purple),

D8 (blue), D10 (red), and D54 (green) membranes, respectively. The ∆S/So at each τ
13

was calculated using So and S1 CO intensities determined over a 3.0 ppm integration
width.............................................................................................................................126
13

2

13

2

C- H REDOR So (black) and S1 (colored) spectra for τ = 40 ms for
HFP_G5C in "100% D4" (a), "80% D8 + 20% DTPG" (b), and "80% D10 + 20% DTPG"
(c) membranes, respectively. All spectra were processed using 100 Hz Gaussian line
broadening and baseline correction..............................................................................127
Figure 4.9

Figure 4.10 C- H REDOR dephasing plots of ∆S/So vs τ for HFP_G5C in "100% D4"
(purple), "80% D8 + 20% DTPG" (blue), and "80% D10 + 20% DTPG" (red) membranes,
13
respectively. The ∆S/So at each τ was calculated using So and S1 CO intensities
determined over a 3.0 ppm integration width................................................................128
2

Figure 4.11 "t1D_ir" pulse sequence used for H T1 measurements. The phase/phase
cycle is "x" for the π pulse, "x, -x, y, -y" for the first π/2 pulse, and "y, y, x, x" for the
second π/2 pulse...........................................................................................................129
o

Figure 4.12 "t1D_ir" experiments of HFP_F8C in D8 membrane at -50 C under static
2

condition. For each τ1, the number of acquisition = 16000. (a) H FID for τ1 = 0.1 ms
2
and 200.1 ms; (b) H spectra for τ1 = 0.1 ms through 400.1 ms in an increment of 40 ms.
All spectra were processed using 2000 Hz Gaussian line broadening and data shift of -

xvii

12. For simplicity, spectra for τ1 = 20.1 ms through 380.1 ms in an increment of 40 ms
are not shown here.......................................................................................................131
o

Figure 4.13 "t1D_ir" experiments of HFP_F8C in D10 membrane at -50 C under static
2

condition. For each τ1, the number of acquisition = 3000. (a) H FID for τ1 = 0.1 ms and
2

200 ms; (b) H spectra for τ1 = 0.1 ms through 100 ms. All spectra were processed
using 2000 Hz Gaussian line broadening, data shift of -11, and baseline correction of
order 3. For simplicity, spectra for τ1 = 120 ms through 300 ms are not shown...........132
o

Figure 4.14 "t1D_ir" experiments of HFP_F8C in D54 membrane at -50 C under static
2

condition. For each τ1, the number of acquisition = 2000. (a) H FID for τ1 = 0.1 ms and
2

200 ms; (b) H spectra for τ1 = 0.1 ms through 100 ms. All spectra were processed
using 2000 Hz Gaussian line broadening and data shift of -12. The outer horns (1 and 4)
and inner horns (2 and 3) were diagnostic of -CD3 and -CD2- in D54 membrane,
respectively. For simplicity, spectra for τ1 = 120 through 300 ms are not shown.........133
o

Figure 4.15 "t1D_ir" experiments of HFP_F8C in D8 membrane at -50 C under 10 kHz
2

MAS. For each τ1, the number of acquisition = 15000. (a) H FID for τ1 = 0.1 ms and
2
200.1 ms; (b) H spectra for τ1 = 0.1 ms through 140.1 ms. All spectra were processed
using 200 Hz Gaussian line broadening, data shift of -32, and baseline correction of
order 3. For simplicity, spectra for τ1 = 160.1 through 400.1 ms are not displayed......134
o

Figure 4.16 "t1D_ir" experiments of HFP_F8C in D10 membrane at -50 C under 10
2
kHz MAS. For each τ1, the number of acquisition = 5000. (a) H FID for τ1 = 0.1 ms and
2

200.1 ms; (b) H spectra for τ1 = 0.1 ms through 100.1 ms. All spectra were processed
using 200 Hz Gaussian line broadening, data shift of -31, and baseline correction of
order 10. For simplicity, spectra for τ1 = 120.1 through 300.1 ms are not displayed....135
o

Figure 4.17 "t1D_ir" experiments of HFP_F8C in D54 membrane at -50 C under 10
2
kHz MAS. For each τ1, the number of acquisition = 3000. (a) H FID for τ1 = 0.1 ms and
2

200.1 ms; (b) H spectra for τ1 = 0.1 ms through 100.1 ms. All spectra were processed
using 200 Hz Gaussian line broadening and data shift of -12. For simplicity, spectra for
τ1 = 120.1 ms through 300.1 ms are not displayed......................................................136

xviii

Figure 4.18 "t1D_ir" experimental (black squares with uncertainties) and best-fit (red
2
line) plots of H intensity vs τ1 under static conditions for CD2 in D8 sample (a), CD3 in
D10 sample (b), and CD2 and CD3 in D54 sample (c, d), respectively. a.u. ≡ arbitrary
unit................................................................................................................................139
Figure 4.19 "t1D_ir" experimental (black squares with uncertainties) and best-fit (red
2
2
line) plots of H intensity vs τ1 under 10 kHz MAS for H in D8 (a), D10 (b), and D54
samples (c), respectively..............................................................................................142
Figure 4.20 (a) Experimental (points with uncertainties) and best-fit (solid line) plots of
∆S/So vs τ for KALP_A11C in membranes. The D4 data were not fitted. (b) Membrane
locations of KALP with major (left) and minor (right) populations. The colored bands,
2
13
orange dots, and brown lines represent H positions, A11 CO nuclei, and lysine
sidechains, respectively. The thickness of the palmitoyl region of the DPPC membrane
is ~31 Å.........................................................................................................................148
Figure 4.21 Models of α helical KALP in the DPPC membrane. The green ribbon is the
KALP backbone and the vertical lines are the full van der Waals extent of the helix
including leucine sidechains. The horizontal black dashed lines are the boundaries of
2
the ~31 Å-thick palmitoyl region of the bilayer. The blue and red bands are the H
locations in the D8 and D10 membranes, respectively. Model (a) is the major KALP
location with a = 2.4 Å calculated using the best-fit r (D10) = 4.5 Å. Model (b) is the
minor KALP location with (9 Å – b) = 7.8 Å calculated using the best-fit r (D8) = 4.0
Å...................................................................................................................................149
Figure 4.22

13

2

C- H REDOR experimental (points with uncertainties) and best-fit (solid

line) plots of ∆S/So vs τ for (a) HFP_F8C and (b) HFP_G5C in membranes. (c, d) Deep
and shallow membrane insertion of HFP. Residues A1 and A14 are close to the
13
31
membrane surface in concurrence with peptide CO-lipid P distances of ~5 Å for
these residues. For clarity, only one β strand is displayed............................................152
13

2

Figure 4.23 C- H REDOR experimental plots ∆S/So vs τ for IFP_L2C in either D8
(blue) or D10 (red) membrane......................................................................................154
Figure 4.24

13

2

C- H REDOR experimental plots ∆S/So vs τ for the L9R mutant of

HFP_G5C in "D4+DPPG" (4:1, purple), "D8+DPPG" (4:1, blue), and "D10+DPPG" (4:1,
red) membranes, respectively......................................................................................155

xix

2

Figure 4.25 Lipid H Pake doublets acquired by "quecho" experiments under static
conditions for (a) KALP_A11C in D4 and (b-d) HFP_F8C in D8 (blue), D10 (red), and
D54 (green) membranes, respectively. The number of acquisitions was 14482 for (a),
8000 for (b), 13591 for (c), and 3000 for (d), respectively. All spectra were processed
using 2000 Hz Gaussian line broadening and baseline correction...............................157
13

Figure 5.1 Membrane insertion model of the N-terminal helix of IFP. The CO labeled
residue is numbered N and the residue in the membrane headgroup region is numbered
No. θ is the tilt angle of the N-terminal helix of IFP to the membrane surface, r is the
distance between the labeled

13

CO nucleus and the lipid headgroup, and d is the

distance from residue N to residue No along the N-terminal helix axis........................163

Figure 5.2 Structure of cholesterol-26,26,26,27,27,27-d6............................................166
Figure C1 HPLC chromatogram of HFP purification....................................................187
Figure C2 MALDI mass spectrum of HFP after purification. The theoretical molecular
weight of HFP is 3150.8 Da..........................................................................................188
Figure D1 Typical bearing parameters in the ACC panel for a 4 mm MAS probe.......191
1

1

Figure D2 Little screw (red circled) on the H match that plays a key role for H
tuning............................................................................................................................193
Figure D3 (a) REDOR So and S1 FIDs in the presence of severe probe arcing; (b)
Stator showing the carbonized rim near one leg of the solenoid coil where the arcing
occurs...........................................................................................................................194

xx

KEY TO ABBREVIATIONS

CHCA: α-cyano-4-hydroxycinnamic acid
CHR: C-terminal heptad repeat region
CSA: chemical shift anisotropy
CP: cross polarization
DCM: dichloromethane
DHB: 2,5-dihydroxybenzoic acid
DIEA: N,N-diisopropylethylamine
DMF: N,N-dimethylformamide
DMPC: dimyristoylphosphatidylcholine
DPC: dodecylphosphocholine
DPPC: dipalmitoylphosphatidylcholine
DPPG: dipalmitoylphosphoglycerol
DTPC: 1,2-di-O-tetradecyl-sn-glycero-3-phosphocholine
DTPG: 1,2-di-O-tetradecyl-sn-glycero-3-[phospho-(1'-rac-glycerol)]
EPR: electron paramagnetic resonance
ESI: electrospray ionization
Fmoc: 9-fluorenylmethyloxycarbonyl
FID: free induction decay
FRET: fluorescence resonance energy transfer
xxi

HA: hemagglutinin protein
HBTU: 2-(1H-Benzotriazole-1-yl)-1,1,3,3-tetramethyluronium hexafluorophosphate
HEPES: 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid
HFP: HIV fusion peptide
HIV: human immunodeficiency virus
HOBt: 1-hydroxybenzotriazole
HPLC: high pressure liquid chromatography
HSQC: heteronuclear single quantum correlation
IFP: influenza fusion peptide
IR: Infrared spectroscopy
LWHM: line width at half maximum
MALDI: matrix-assisted laser desorption ionization
MES: 2-(N-morpholino)ethanesulfonic acid
MPER: membrane-proximal external region
n.a.: natural abundance
NHR: N-terminal heptad repeat region
NMR: nuclear magnetic resonance
NOE: Nuclear Overhauser effect
PAF: principal axis frame
POPC: 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine
POPG: 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphoglycerol
xxii

PHI: pre-hairpin intermediate
REDOR: rotational-echo double-resonance
r.f. (or RF): radio frequency
SHB: six helix bundle
SPPS: solid phase peptide synthesis
SSNMR: solid state nuclear magnetic resonance
t-Boc: tert-butyloxycarbonyl
TFA: trifluoroacetic acid
TIS: triisopropylsilane
TPPM: two pulse phase modulation
XRD: X ray diffraction

xxiii

Chapter 1 - Introduction
1.1 NMR background
NMR interactions
For any NMR-active nuclei with nuclear spin quantum number I ≠ 0, they can
interact with the electric and magnetic fields and the interactions can be divided into two
categories, external and internal. The external interactions are between the nuclear spin
and external fields such as the static magnetic field Bo (Bo = Boz) and radio frequency
a

(r.f.) pulse field B1. The internal interactions are between the nuclear spin and the
intrinsic fields such as J-coupling field

BJ and chemical shift field BCS. The total

Hamiltonian of the spin is
ˆ = H
ˆ +H
ˆ = (H
ˆ +H
ˆ ) + (H
ˆ +H
ˆ +H
ˆ +H
ˆ )
H
t
ext
int
Z
RF
J
CS
D
Q

(1.1)

where
ˆ is the Hamiltonian of Zeeman interaction between the spin and Bo field;
H
Z
ˆ is the Hamiltonian of the spin interaction with the B1 field;
H
RF
ˆ is the Hamiltonian of J-coupling interaction (indirect nuclear dipole-dipole coupling)
H
J

between two spins;

______________________
a

In this dissertation, letters or symbols representing vectors are displayed in bold, and
letters or symbols representing quantum mechanical operators have a "^" above them.

1

ˆ
H
CS is the Hamiltonian of the spin interaction with chemical shift field (electronic

shielding field) induced by Bo field;
ˆ is the Hamiltonian of direct nuclear dipole-dipole coupling between two spins;
H
D
ˆ is the Hamiltonian of quadrupolar interaction between the electric quadrupole
H
Q

moment of a quadrupolar nucleus (I > 1/2) and the surrounding electric field gradient.
Zeeman interaction
For a nucleus having a spin quantum number I (I ≠ 0), the nuclear spin energy
splits into (2I + 1) levels in the static magnetic field Bo. This interaction is referred to as
Zeeman interaction and the Hamiltonian is[1]
ˆ = −µˆ ⋅ Bo
H
Z

(1.2)

where µˆ is the nuclear magnetic moment operator and is given by
µˆ = γ ˆI = γ [i ˆIx + j ˆIy + k ˆIz ]

(1.3)

where γ is the nuclear gyromagnetic ratio,  is the Planck constant divided by 2π, ˆI is the
nuclear spin operator, and i, j, and k are the unit vectors in the x, y, and z axes,
respectively.
Incorporation of equation 1.3 into equation 1.2 yields
ˆ = −γ ˆI Bo
H
Z
z

(1.4)

ˆ are the wavefunctions describing the (2I + 1) states of the spin
The eigenfunctions of H
Z

2

ˆ are the energies associated with different states
in the Bo field. The eigenvalues of H
Z

of the spin and are obtained through
ˆ | I,m〉 = −γ Bo ˆI | I,m〉
H
Z
z

(1.5)

where | I,m〉 is an eigenfunction of ˆIz with an eigenvalue of m, i.e.
ˆI |=
m | I,m〉
Z I,m〉

(1.6)

Incorporation of equation 1.6 into equation 1.5 yields
ˆ | I,m〉 = −γ Bo m=
| I,m〉 EI,m | I,m〉
H
Z

(1.7)

where EI,m is the energy associated with the spin state m. For a spin-1/2 nucleus such
as

13

C, possible values of m are +1/2 and -1/2. For the m = +1/2 state, i.e. α state, the

spin energy Eα = -

1
1
γ  Bo. For the m = -1/2 state, i.e. β state, the spin energy Eβ = γ
2
2

 Bo. The energy difference between α and β states is
∆E = Eβ - Eα = γ  Bo

(1.8)

The nuclear magnetic moments associated with the α and β states are displayed in
Figure 1.1. Each magnetic moment µ experiences a torque Γ = µ x Bo from the Bo field
and precesses about Bo clockwise with an angular frequency of γBo. This precession is
called Larmor precession and the corresponding frequency is called Larmor frequency.
At the thermal equilibrium, the magnetization M, which is the vector sum of all individual

3

magnetic moments of all nuclei of a specific type (e.g.

13

C), is along the direction of Bo,

i.e., z axis.
Since the static magnetic field Bo is usually orders of magnitude greater than the
internal magnetic fields such as dipolar field and chemical shift field, Zeeman interaction
is much stronger than the internal interactions. In this case, the internal interactions are
treated as perturbations of Zeeman interaction. For any type of internal interaction, its
Hamiltonian can be divided into two parts, the secular part which commutes with
Zeeman Hamiltonian and the non-secular part which does not. Since only the secular
part affects the observable spectrum to the first order, we do not consider the nonsecular part, i.e., the non-secular part of an internal Hamiltonian is truncated. This type
of treatment is referred to as secular approximation.

Figure 1.1 The two spin states of a spin-1/2 nucleus in the static magnetic field Bo. The
field Bo is along z axis. The two magnetic moments µα and µβ (blue) are under Larmor
precession about the field Bo with an angular frequency of ωo = γBo.
4

Interaction with B1 field (r.f. pulses)
In a SSNMR probe, the solenoid coil generates a B1 field when applying r.f.
pulses. Take a 90x pulse for example, the B1 field is
B1 (t) = B1cos(ωt)x

(1.9)

where ω = 2πν and ν is the frequency of the 90x pulse set by the spectrometer, and x is
the unit vector along x axis. This oscillating B1 field can be divided into two components,
and non-resonant part Bnon-res
:
the resonant part Bres
1
1
B1res =

1
B1[cos(ωt)x - sin(ωt)y]
2

=
Bnon-res
1

(1.10)

1
B1[cos(ωt)x + sin(ωt)y]
2

(1.11)

The Bres
component rotates clockwise in the xy plane and Bnon-res
rotates
1
1
counterclockwise. As previously discussed, the magnetic moment of each spin state is
precessing clockwise about the Bo field. Therefore, only the Bres
component affects the
1
nuclear spins since it rotates in the same sense as the spin precession (Larmor
precession) while the Bnon-res
component does not.
1
In SSNMR, we usually run experiments under magic angle spinning (refer to the
o

"MAS" section for details), where the solenoid coil in the probe has an angle of 54.7

5

(magic angle) with the Bo field. In this case, for a B1 field generated along the solenoid
o

coil axis, only a ~0.4 fraction (0.5 x sin54.7 ) of it is utilized in the experiments.
For a magnetization M under the B1 field of a r.f. pulse, it experiences a torque
Γ = M x B1 and precesses about the B1 field with an angular frequency of γB1. Such a
precession is called Rabi precession and the frequency is called Rabi frequency. The
direction of the torque Γ can be determined using the right-hand rule: first point the four
fingers of your right hand along M, then curl your fingers toward the direction of B1, the
thumb is now pointing along the torque Γ. For instance, if M is along z axis, B1 is along
x axis, then Γ is along y axis. For any r.f. pulse, its flip angle θ = γΒ1τp, where τp is the
duration of the pulse. θ = π/2 for a 90x pulse and π for a 180x pulse, where x is the r.f.
pulse phase and refers to a B1 field along the x axis. Figure 1.2 displays the effects of a
90x pulse and a 180x pulse on the magnetization M which is initially along z axis at the
thermal equilibrium. For an x pulse, its Hamiltonian is
ˆ = -  ω1 ˆI = - γ Β1 ˆI
H
RF
x
x

(1.12)

Note that equation 1.12 corresponds to the r.f. pulse Hamiltonian in the rotating frame
(refer to the "Rotating frame" section for more details) since the term B1 is static, as
opposed to equation 1.9, which corresponds to the laboratory frame where the B1 field
is oscillating as a function of time.

6

Figure 1.2 Rabi precession of the magnetization M (initially along z axis) about the B1
field of a 90x pulse (a) and a 180x pulse (b).
J-coupling interaction
J-coupling, sometimes also known as scalar coupling, is the indirect nuclear
dipole-dipole interaction through bonding electrons. There are several mechanisms
proposed for J-coupling, of which the most important one is the Fermi contact
mechanism[2].
According to this mechanism, both bonding electrons in the X-Y bond (X and Y are both
NMR-active, i.e. I ≠ 0) spend a finite time at the point of nuclei X and Y, in which case
the electron and the nucleus are said to be in Fermi contact. In the X-Y bond, it is more
stable (i.e. lower energy) when nucleus X (or Y) and the electron in Fermi contact have
7

antiparallel angular momentum. Since the gyromagnetic ratio is usually positive for a
nucleus (e.g.

13

1

C, H, etc) and negative for an electron, it means it is more stable when

the magnetic moments of the nucleus and electron are parallel. Therefore, when
nucleus Y is spin-up, the electron in Fermi contact will be also up. According to the Pauli
exclusion principle, the other electron in the X-Y bond must be down since the two
electrons occupy the same molecular orbital. In this case, the more favorable spin state
of nucleus X will be down. Similarly, when Y is spin-down, the more favorable spin state
of X will be up. This tells us the α state of X is more favored (i.e. lower energy) when Y
is in β state and less favored (i.e. higher energy) when Y is in α state due to the X-Y Jcoupling interaction. Similarly, the β state of X is more favored when Y is in α state and
less favored when Y is in β state. The nuclear and electron spins in the X-Y bond and
the spin energy diagram of X with and without the X-Y J-coupling interaction are
displayed in Figure 1.3a. If Y is a spin 1/2 nucleus, the X-Y J-coupling will result in two X
peaks in the spectrum with a spacing equal to the J-coupling constant, JXY (Figure
1.3b).

8

Figure 1.3 J-coupling mechanism. (a) Nuclear and electron spins in the X-Y bond and
the spin energy diagram of nucleus X with (black) and without (blue) the X-Y J-coupling
interaction. (b) Peak splitting of nucleus X in the spectrum due to the J-coupling
between X and Y, where Y is a spin 1/2 nucleus.
The Hamiltonian of J-coupling between two spins Ij and Ik is
ˆ = 2π Jjk ˆI ⋅ ˆI
H
j
k
J

(1.13)

where ˆIj and ˆIk are the nuclear spin operators and Jjk is the J-coupling tensor. Jjk is a 3
x 3 matrix depending on the molecular orientation but it reduces to a number in isotropic
liquids. That is why sometimes J-coupling is also called scalar coupling.
J-coupling is the most important spin interaction for solution NMR but not
considered in most cases for SSNMR since it is usually the weakest spin interaction
existing in solids and is sometimes eliminated by decoupling pulses.

9

Isotropic and anisotropic chemical shift interactions
In a molecule, the electrons are either surrounding the nuclei or located at the
chemical bonds. For a molecule in the presence of the external magnetic field Bo, the
electrons which are moving about a nucleus or within a chemical bond will also move
under the Lorentz force from Bo. This type of electron motion generates a magnetic field
called shielding field or chemical shift field. Since different molecules in a sample have
different orientations with respect to the Bo field, the directions of the induced electric
current and shielding field will also vary from molecule to molecule. Although the
shielding field could be in different directions relative to Bo, only the component along
Bo is relevant according to the secular approximation. In most cases, the shielding field
decreases the magnitude of Bo experienced by a nucleus and thus moves the chemical
shift of the nucleus to a higher value. However, in a paramagnetic species which
contains unpaired electron(s), the shielding field may increase (add to) the magnitude of
Bo experienced by a nucleus. The Hamiltonian of shielding interaction (chemical shift
interaction) acting on a spin I is
ˆ = - γ Bo σ ⋅ ˆI
H
CS

(1.14)

where ˆI is the nuclear spin operator and σ is the shielding tensor which is a secondrank tensor and can be represented by a 3 x 3 matrix. To understand σ, we can take a
look at the electron distribution around a nucleus in a molecule. In general, the electron
distribution is not spherically symmetric. In this case, the size of the shielding depends
10

on the molecular orientation with respect to the external magnetic field Bo. The
shielding tensor σ is then used to describe how the size of shielding varies with the
molecular orientation. There is an axis frame associated with the shielding tensor σ,
PAF

called principal axis frame (PAF), in which there are three principal axes, x
PAF

and z

PAF

,y

,

(Figure 1.4). The principal values associated with the three axes in PAF are

σxx, σyy, and σzz , respectively. The orientation of PAF is determined by the electronic
structure of the part containing the nucleus of interest (e.g.

13

C in the

13

CO group) in a

molecule and is fixed with respect to the molecule.

Figure 1.4 Principal axis frame (PAF) and the shielding tensor (red). The two polar
angles defining the orientation of Bo in PAF are θ and φ, where θ is the angle between
PAF

Bo and the principal axis z
PAF

(xy)

and φ is the angle between the projection of Bo in the
PAF

plane and the principal axis x

. The three principal values associated with

the three axes in PAF are σxx, σyy, and σzz , which also correspond to the three
principal chemical shifts in the CSA (chemical shift anisotropy) powder pattern. The PAF
frame at the lower right corner displays the angles α, β, and γ used in equation 1.15.
11

In general, the chemical shift of a nucleus depends on the orientation with respect
to the external magnetic field Bo. To better understand this, let us take
example. The backbone

13

13

CO for

CO chemical shift of a peptide or protein depends on the

orientation of its shielding tensor σ relative to Bo. The three PAF axes of σ are
PAF

displayed in Figure 1.5a. Assuming the angles between Bo and the three axes x
PAF

y

PAF

, and z

are respectively α, β, and γ (refer to the PAF frame at the lower right

corner of Figure 1.4), the
2

,

2

13

CO chemical shift is then defined as:
2

σ = σxxcos α + σyycos β + σzzcos γ

(1.15)

where σxx, σyy, and σzz are the three principal values associated with PAF and are also
the three principal chemical shifts displayed in Figure 1.5b. For solids under MAS (refer
to the "MAS" section for details) or solutions with rapid molecular tumbling, the isotropic
chemical shift results and is given by
σiso =

1
(σxx + σyy + σzz)
3

(1.16)

2

2

2

This is because the time-average values <cos α> = <cos β> = <cos γ> =

1
under
3

MAS or rapid molecular tumbling. Without MAS or rapid molecular tumbling, the
backbone

13

CO of a peptide or protein may yield a CSA powder pattern similar to the

one in Figure 1.5b (the red spectrum).

12

The chemical shift field (shielding field) can be divided into two components,
isotropic field BCS,iso and anisotropic field BCS,aniso. The total chemical shift
Hamiltonian is
ˆ = {-σiso γ Bo + 1 δcs  [3cos2θ - 1 - ηcssin2θ cos(2φ)]} ˆI
H
CS
z
2

(1.17)

where σiso is the isotropic chemical shift, δcs = -γBo(σzz - σiso), ηcs = (σyy -σxx)/(σzz σiso) and is called asymmetry parameter, θ and φ are the polar angles of Bo in the PAF
frame (refer to Figure 1.4)[3]. The first term in equation 1.17 corresponds to isotropic
chemical shift interaction whereas the second term corresponds to anisotropic chemical
shift interaction. When the nucleus is at a site of axial symmetry, the shielding tensor
PAF

has a PAF in which the z

axis coincides with the symmetry axis and thus

σxx = σyy ≠ σzz. In this case, the asymmetry parameter ηcs = 0.

13

Figure 1.5 (a) The PAF frame of backbone

13

PAF

CO in a peptide or protein. The z
PAF

PAF

is perpendicular to the C—CO—N plane. The x
PAF

CO—N plane with x

and y
PAF

perpendicular to C=O and y

axis

axes are both in the C—

parallel to C=O. (b)

13

CO CSA
PAF

powder pattern. The three chemical shifts are σxx = 247 ppm with Bo parallel to x
PAF
PAF
σyy = 176 ppm with Bo parallel to y
, and σzz = 99 ppm with Bo parallel to z
.

,

Dipolar coupling interaction
When two nuclear spins are close to each other in space, the magnetic moment
of one spin interacts with the magnetic field generated by the other spin, and vice versa.
This type of interaction between two spins is called dipolar coupling and the field is
3

called dipolar field. The dipolar field has a distance dependence of 1/r

14

and an

2

orientation dependence of (3cos θ-1) for the secular component, where r is the
internuclear distance between the two spins and θ is the angle between Bo and the
internuclear vector (Figure 1.6).

Figure 1.6 Definition of the angle θ and distance r for the dipolar interaction between
nucleus I and S. Bo is the external magnetic field which is along the laboratory frame z
axis.
There are two types of dipolar interaction: homonuclear and heteronuclear dipolar
coupling. The Hamiltonian of homonuclear dipolar coupling between two spins I1 and I2
of the same isotope is
2
ˆ = − µo  γ 1 (3cos2θ - 1)(3 ˆI ˆI - ˆI ⋅ ˆI )
H
II
1z 2z 1 2
4π r 3 2

where µo = 4π x 10

-7

(1.18)

H/m and is the permeability of free space, θ and r are defined in

Figure 1.6, and ˆI1 and ˆI2 are two vector operators of spin 1 and 2, respectively. The
Hamiltonian of heteronuclear dipolar coupling between two spins I and S of different
isotopes is
15

ˆ = − µo  γI γ S 1 (3cos2θ - 1)2 ˆI Sˆ
H
IS
z z
4π r 3 2

where the term

(1.19)

µo γI γ S
is called dipolar coupling constant (in units of rad/s). For the

4π r 3

dipolar coupling constant between I and S in units of Hz, it is defined as
d=(

µ hγ γ
µo γI γ S
 3 )/2π = o 3I 3S
4π r
16π r

Using equation 1.20, we can derive that
and

13

(1.20)
13

15

3

C- N dipolar coupling in Hz equals 3066/r

2

3

C- H dipolar coupling in Hz equals 4642/r , where r is the distance in Å.

Notice that the spin part of equation 1.18 differs from that of equation 1.19, to
understand this, we can first take a look at the interaction between Zeeman Hamiltonian
and the homonuclear dipolar Hamiltonian. The Zeeman Hamiltonian for two same spins
ˆ = - γ Bo( ˆI + ˆI ). We can do the calculation to see whether or not the
I1 and I2 is H
z
1z
2z
ˆ :
operator ˆI1 ⋅ ˆI2 commutes with H
z
ˆ , ˆI ⋅ ˆI ] = - γ Bo [(Iˆ + ˆI ),(Iˆ ˆI + ˆI ˆI + ˆI ˆI )]
[H
z 1 2
1z
2z
1x 2x
1y 2y
1z 2z

= - γ Bo {Iˆ1z (Iˆ1x ˆI2x + ˆI1y ˆI2y + ˆI1z ˆI2z ) + ˆI2z (Iˆ1x ˆI2x + ˆI1y ˆI2y + ˆI1z ˆI2z )}
= - γ Bo {[Iˆ1z ,Iˆ1x ]Iˆ2x + [Iˆ1z ,Iˆ1y ]Iˆ2y + [Iˆ1z ,Iˆ1z ]Iˆ2z
+[Iˆ2z ,Iˆ2x ]Iˆ1x + [Iˆ2z ,Iˆ2y ]Iˆ1y + [Iˆ2z ,Iˆ2z ]Iˆ1z }

= - γ Bo {i ˆI1y ˆI2x − i ˆI1x ˆI2y + 0 + i ˆI2y ˆI1x − i ˆI2x ˆI1y + 0}

16

=0

(1.21)

ˆ for homonuclear dipolar coupling.
Therefore, the operator ˆI1 ⋅ ˆI2 commutes with H
z

However, for heteronuclear dipolar coupling, since the two spins I and S have different
gyromagnetic ratios, the γ in equation 1.21 can no longer be put right in the front of the
equation. As a result, the component ( ˆI1x ˆI2x + ˆI1y ˆI2y ) of the operator ˆI1 ⋅ ˆI2 no longer
commutes with Zeeman Hamiltonian of the two spins. According to the secular
approximation discussed earlier, ˆI1x ˆI2x + ˆI1y ˆI2y is the non-secular part of the heteronuclear
dipolar Hamiltonian and is thus truncated. That is why the spin part of homonuclear
dipolar Hamiltonian is 3Iˆ1z ˆI2z − ˆˆ
I1I2 (equation 1.18) while for heteronuclear dipolar
Hamiltonian it is 2IˆzSˆ z (equation 1.19). The heteronuclear dipolar coupling will be
discussed in more details in the "REDOR" section.
Quadrupolar interaction
If a nucleus has a spin quantum number greater than 1/2, it is referred to as a
2

quadrupolar nucleus, e.g. H (I = 1) and

23

Na (I = 3/2). The charge distribution of a

quadrupolar nucleus is displayed in Figure 1.7 and results in a non-zero electric
quadrupole moment eQ, where e is the charge carried by a proton and Q is specific to
the nuclear isotope. The electric quadrupole moment interacts with the electric field
gradient (resulting from the non-uniform electron density distribution) at the location of
the nucleus and this interaction is known as quadrupolar coupling. The magnitude of
quadrupolar coupling depends on the magnitudes of the electric quadrupole moment of
the nucleus and the electric field gradient. For a quadrupolar nucleus, the quadrupolar

17

interaction affects the nuclear spin energy levels as other previously discussed
magnetic interactions do.

Figure 1.7 Electric properties of a quadrupolar nucleus. (a) prolate charge distribution.
(b) oblate charge distribution. (c) electric quadrupole moment of (a) with a positive sign
-31 2
2
m for H). (d) electric quadrupole moment of (b) with a negative
(e.g. it is 2.8 x 10
-30

sign (e.g. it is -2.6 x 10

2

m for

17

O)[4].

The quadrupolar Hamiltonian is
ˆ =
H
Q

eQeq 1
2
2
1
[3cos θ - 1 - ηQsin θ cos(2φ)] [3Iˆz2 − I(I + 1)]
2I(2I − 1) 2
2

(1.22)

where e is the elementary charge, Q is the magnitude of the electric quadrupole
moment, q is a value associated with the electric field gradient tensor, I is the nuclear
spin quantum number, θ and φ are the polar angles of the Bo field in the PAF, ηQ is the
asymmetry parameter of the electric field gradient tensor, and ˆIz is the z-component of
the nuclear spin operator. Since the electric field gradient tensor has an orientation
2

2

dependence of [3cos θ - 1 - ηQsin θ cos(2φ)], the quadrupolar Hamiltonian also has this
18

dependence. The term

eQeq
in equation 1.22 is called quadrupolar coupling constant


(QCC) and is in units of rad/s. As for the QCC in units of Hz, it is defined as
χ=

eQeq
h

(1.23)

2

2

For H (I = 1) in aliphatic C- H bonds, χ is ~170 kHz[3]. To better understand the
2

2

quadrupolar interaction, we will take H in aliphatic C- H bonds as an example and
2

discuss in some detail. The H quadrupolar energy is given by

EQ =

=

eQeq 1
2
2
2
1
[3cos θ - 1 - ηQsin θ cos(2φ)] [3m - I(I + 1)]
2I(2I − 1) 2
2
2
2
2
π
χ [3cos θ - 1 - ηQsin θ cos(2φ)](3m - 2)
4

(1.24)

2

where χ is H QCC and m has three possible values of -1, 0, and 1 which correspond to
2

the three different spin states of H. For a fixed orientation (i.e. fixed θ and φ), different
spin states (m = ±1 vs m = 0) have different nuclear charge distributions and thus
different electric quadrupole moments, which cause different quadrupolar interactions
2

2

and energies. For H in aliphatic C- H bonds, ηQ ≈ 0. In this case, equation 1.24 can be
simplified to
EQ =

2
2
π
χ (3cos θ - 1)(3m - 2)
4

(1.25)

19

where θ is one of the two polar angles of Bo in the PAF and refers to the angle between
2

the C- H bond and Bo. Next we will discuss the orientation dependence of the
2

quadrupolar energy EQ and how it affects the H resonance frequency.
2

(a) When θ = 0o, 3cos θ - 1 = 2, EQ =

respectively E-1 =

2
π
χ (3m - 2). For m = -1, 0, and 1, EQ is
2

2
π
π
χ , E0 = - πχ , and E1 = χ . There are two transitions for H,
2
2

one is m = 1 → m = 0 and the other is m = 0 → m = -1. Assuming the Zeeman energy
for m = -1, 0, and 1 is +E', 0, and -E', respectively, in this case, the energy difference for
the m = 1 → m = 0 and m = 0 → m = -1 transitions is ∆E1→0 = E' -

= E' +

3π
χ and ∆E0→−1
2

3π
χ , respectively. Therefore, the resonance frequency for the m = 1 → m = 0
2

and m = 0 → m = -1 transitions is ν1→0 = ν' where ν' = E'/h is the

2

3
3
χ and ν 0→−1 = ν' + χ , respectively,
4
4

H Larmor frequency without considering the quadrupolar

2

interaction. If we set the H transmitter frequency at ν', we will observe two signals in
2
3
the H spectrum, one at - χ (in Hz, not in ppm) resulting from the m = 1 → m = 0
4

transition and the other at

3
χ resulting from the m = 0 → m = -1 transition (Figure
4

1.8a).

20

2

(b) When θ = 54.7o, 3cos θ - 1 = 0, EQ = 0 for all m values. In this case, the m = 1
2

→ m = 0 and m = 0 → m = -1 transitions yield a signal at the same frequency in the H
spectrum (Figure 1.8b).
2
2
π
(c) When θ = 90o, 3cos θ - 1 = -1, EQ = - χ (3m - 2). For m = -1, 0, and 1, EQ
4

π
π
π
is respectively E-1 = - χ , E0 = πχ , and E1 = - χ . The energy difference for the
4
2
4

m = 1 → m = 0 and m = 0 → m = -1 transitions is respectively ∆E1→0 = E' +

∆E0→−1 = E' -

3π
χ and
4

3π
χ . Therefore, the resonance frequency for the m = 1 → m = 0 and m
4

= 0 → m = -1 transitions is ν1→0 = ν' +

3
3
χ and ν 0→−1 = ν' - χ , respectively. For this
8
8
2

orientation, we will observe two signals in the H spectrum with one at

3
χ resulting
8

3
from the m = 1 → m = 0 transition and the other at - χ resulting from the m = 0 → m =
8

-1 transition (Figure 1.8c).
2

When we consider all the possible θ values, i.e. all the possible C- H bond
2

orientations with respect to Bo, we will observe a H quadrupolar powder pattern, which
is also referred to as Pake doublet, as displayed in Figure 1.7d, where the black and
blue component spectra correspond to the m = 0 → m = -1 and m = 1 → m = 0
transitions, respectively.

21

2

2

Figure 1.8 Orientation dependence of the H spectra for aliphatic C- H bonds. (a) θ =
2

0o. (b) θ = 54.7o. (c) θ = 90o. (d) H Pake doublet for all possible θ values, where θ refers
2

to the angle between the C- H bond and Bo field.
Rotating frame
In the laboratory frame, the B1 field of a pulse rotates with an angular velocity of
ω in the xy plane, where ω is the transmitter frequency of the pulse. We call it
transmitter frequency for the reason that a radiofrequency transmitter is used to
22

generate the pulse. However, in the rotating frame, the xy coordinate system rotates
with an angular velocity of ωr, where ωr is the so-called rotating frame frequency. When
we set the transmitter frequency the same as the rotating frame frequency (as is usually
the case), the B1 field of a pulse appears to be static, i.e. the time dependence of the
field is removed. This makes it much simpler to work out the effect of the B1 field on a
nuclear magnetization. In the rotating frame, the apparent precession rate of a
magnetization is (ωo - ωr), where ωo = γBo is the Larmor frequency. This difference
frequency, ωo - ωr, is often referred to as the resonance offset frequency ωr.o., i.e.
ωr.o. = ωo - ωr

(1.26)

According to the relationship ω = γB, the resonance offset field is given by
Br.o. = (ωr.o./γ)z

(1.27)

where the field Br.o. is along the z axis and refers to the secular component of the
apparent magnetic field experienced by a nucleus in the rotating frame. If we set the
transmitter frequency ω close to the Larmor frequency ωo and set ω the same as the
rotating frame frequency ωr, the resonance offset frequency ωr.o. will be close to zero
and thus the field Br.o. will be very small. As the magnitude of Br.o. is small relative to
that of the r.f. pulse field B1, the r.f. pulses can work efficiently on the nuclear
magnetization and cause rapid spin transitions. However, for quadrupolar nuclei which
have a broad Larmor frequency distribution, no matter where we set the transmitter
23

frequency, there are large resonance offsets for some of the nuclei and thus Br.o./B1
may not be close to zero. For such nuclei, the spin transitions caused by the r.f. pulses
may be slow. Since the r.f. pulse duration is usually very short (in the order of µs), the
nuclei with large resonance offsets may not be affected by the pulses. To solve or
minimize this issue, a strong pulse field B1 is usually applied to compensate the large
resonance offsets of quadrupolar nuclei. In the next paragraph, we will discuss in some
detail about how can a spectrometer detect an NMR signal (the xy component of a
magnetization, i.e. the transverse magnetization) with a resonance offset frequency ωr.o.
in the rotating frame while the transverse magnetization actually precesses with a
Larmor frequency ωo.
In the NMR spectrometer, a mixer is used to mix two input signals together. One
is the NMR signal (i.e. transverse magnetization of the observed nuclei) from the probe
and the other is a signal generated by the local oscillator. Suppose the NMR signal from
the probe is Acosωot where A is the overall signal intensity and ωo is the Larmor
o

frequency. The local oscillator generates two types of signals which are 90 out of
phase: cosωrt and -sinωrt, where ωr is the receiver frequency and is usually the same as
the transmitter frequency[5]. When the local oscillator generates the signal cosωrt, the
mixer multiplies together the NMR signal and the local oscillator signal and yields:
Acosωot x cosωrt =

1
A[cos(ωo + ωr)t + cos(ωo - ωr)t]
2

24

(1.28)

After the mixing process, a low-pass filter is used to filter out the term cos(ωo + ωr)t and
thus only the term cos(ωo - ωr)t survives. In this case, the output signal of the mixer is
cosine modulated and has a frequency of ωo - ωr which is exactly the resonance offset
frequency ωr.o. in the rotating frame. This cosine modulated signal is equivalent to the x
component of the magnetization detected in the rotating frame and is referred to as the
real part of the FID (free induction decay). Next let us consider the other case in which
the local oscillator signal is -sinωrt. The output signal of the mixer is given by

Acosωot x -sinωrt =

1
A[sin(ωo - ωr)t - sin(ωo + ωr)t]
2

(1.29)

where the term sin(ωo + ωr)t is removed by the low-pass filter and only the term sin(ωo ωr)t remains. Now the output signal is sine modulated and also has the resonance offset
frequency. This sine modulated signal is equivalent to the y component of the
magnetization detected in the rotating frame and is referred to as the imaginary part of
the FID. The real and imaginary parts of the FID are detected separately and such a
scheme is called quadrature detection, as displayed in Figure 1.9.

25

Figure 1.9 Quadrature detection scheme in the rotating frame for all modern NMR
spectrometers. The device ADC refers to analogue to digital converter and is used to
convert the NMR signal from a voltage to a binary number which can be stored in the
computer memory[5].

Magic angle spinning (MAS)
In liquids, the rapid molecular tumbling averages out the anisotropic interactions
such as CSA, dipolar, and quadrupolar interactions. As a result, sharp peaks with good
resolution are usually observed in liquid-state NMR spectra. However, in solids, there is
no such rapid molecular tumbling and the anisotropic interactions cause severe line
broadening in the static NMR spectra. To solve this problem, magic angle spinning
(MAS) was developed[6]. By spinning the sample under MAS, one can average out the
anisotropic interactions and thus isotropic chemical shifts result. The geometry of MAS
is displayed in Figure 1.10, where the angle between the rotor axis (i.e. sample spinning
o

axis) and the external field Bo equals the so-called magic angle of 54.7 . To better
understand the effect of MAS on anisotropic interactions, let us take the
26

13

15

C- N dipolar

coupling as an example. Assume θ is the angle between Bo and the

13

15

C- N

internuclear vector, α is the angle between Bo and the sample spinning axis, and β is
the angle between the

13

15

o

C- N vector and the spinning axis. Under MAS, i.e. α = 54.7 ,
2

we can mathematically show that the average of 3cos θ(t) - 1 over each rotor period is
2

<3cos θ(t) - 1> =

2
2
1
(3cos α - 1)( 3cos β - 1) = 0
2

(1.30)

2

Since the heteronuclear dipolar Hamiltonian has an orientation dependence of 3cos θ 1(refer to equation 1.19), the

13

15

C- N dipolar interaction is averaged out by MAS.

Nowadays MAS has been widely used in various SSNMR experiments.

Figure 1.10 The geometry of a

13

15

C- N internuclear vector in a solid state sample

o

under MAS where α equals 54.7 (magic angle).
REDOR
Rotational-echo double-resonance (REDOR) is a SSNMR technique originally
developed by Gullion and Schaefer[7]. It utilizes rotor-synchronized π pulse trains to
27

recouple the heteronuclear dipolar interaction under MAS. Since the dipolar coupling d
is proportional to

1
, the distance r between two nuclei can be obtained from REDOR
r3

measurement. Next we will take

13

15

C - N REDOR for example to discuss in some detail

about the spin interactions in REDOR under MAS. Figure 1.11 shows the
1

13

15

C - N

o

REDOR pulse sequence. In the beginning of the sequence, a H 90 pulse is applied to
1

1

rotate the H magnetization from the z axis (Bo axis) to the xy plane. After that, H and
13

C cross polarization (CP) pulses are applied simultaneously to transfer magnetization
1

from H to

13

1

13

C via H - C dipolar coupling to enhance the

13

C signal intensity. CP

1

occurs in a so-called doubly rotating frame with one in which the H B1 field is static and
the other in which the

13

1

C field B1 is also static. The magnitudes of the H and

13

C CP

pulse fields, B1,H and B1,C, must fulfill the Hartmann-Hahn matching condition[8]:
γHB1,H = γCB1,C

(1.31)

This is the condition for ideal δ-function pulses for which no resonance offset field in the
rotating frame needs to be considered. However, in practice, we need to take into
account the resonance offset field Br.o. since the r.f. pulses are applied with limited
power. As we consider Br.o., the effective magnetic field Beff experienced by a nucleus
is the vector sum of B1 and Br.o. and the matching condition is now given by
1

13

γHBeff( H) = γCBeff( C)

(1.32)
28

Note that equation 1.31 and 1.32 describe the Hartmann-Hahn matching condition for
1

13

the H→ C CP in a static sample. Under MAS, this matching condition needs to be
modified since the sample rotation affects dipolar interactions. As previously discussed,
the

1

13

H→ C CP is achieved via

backbone

13

1

13

H - C dipolar coupling. For a peptide with a

1

1

13

1

13

CO label, the closest H is ~2 Å apart. Therefore, the largest H- CO

dipolar coupling is ~4 kHz. As we spin the sample at a rate of 10 kHz, the H- CO
dipolar coupling is supposed to be averaged out by MAS. However, this is not true due
1

1

to the H- H dipolar coupling. In peptides and other organic compounds, all the protons
are dipolar-coupled as a network. There is rapid flip-flop (also known as
1

1

1

H spin

1

diffusion) between protons via H- H dipolar coupling, that is, when one H changes its
1

spin state from β to α, another H will change from α to β. The flip-flop rate is roughly
equal to the

1

1

H- H dipolar coupling, which is typically in the 10-50 kHz range[9].
1

Therefore, under 10 kHz MAS, H will change its spin state in a timescale comparable
1

to or even shorter than the rotor period. In this case, the H flip-flop will disrupt the
1

13

1

coherent averaging of the H- CO dipolar field/coupling by MAS. As a result, the H13

CO dipolar coupling in the peptide is not averaged out over each rotor period. This is
1

13

1

13

why you can achieve the H→ CO CP via H- CO dipolar coupling under 10 kHz

29

1

13

MAS. For the H→ C CP under MAS, the Hartmann-Hahn matching condition splits
into a series of new matching conditions[10]:
1

13

γHBeff( H) = γCBeff( C) + nωmas

(1.33)

where ωmas is the angular frequency of MAS and n = ±1, ±2, etc which corresponds to
the nth spinning sideband in the

13

C spectrum. Recall that the nuclear spin energy E = -

µ ⋅ B where µ is the nuclear magnetic dipole moment and the energy splitting between

the α and β states of a spin 1/2 nucleus is ∆E = γ B. Therefore, for the CP under static
1

13

1

condition where γHBeff( H) = γCBeff( C), the energy splitting of H is equal to that of
13

1

C in the rotating frame. As one H changes the spin state from α to β, the energy

required can be exactly compensated by a transition of
1

13

C from β to α which releases

13

the same amount of energy. In other words, the H- C dipolar interaction causes a
1

redistribution of energy between H and
13

13

1

C spins but the total energy of the H and

C spin system is conserved during the static CP. However, for the CP under MAS, the
1

13

total spin energy is no longer conserved since γHBeff( H) ≠ γCBeff( C), as shown in
equation 1.33.
Due to the different orientations of molecules, chemical bonds, and internuclear
vectors relative to Bo, the spin interactions such as shielding and dipolar coupling for a
specific nucleus will vary from molecule to molecule. As a result, there is usually a
distribution of the Larmor frequency as well as the resonance offset frequency, i.e. there
30

is a distribution of the resonance offset field Br.o.. Therefore, a ramp is usually applied
to either

1

H or

13

C CP pulse field amplitude to increase the efficiency of the
1

magnetization transfer from H to

Figure 1.11

13

13

C under MAS.

15

C - N REDOR pulse sequence.

After CP, the rotor-synchronized π pulses are applied in two different ways which
correspond to the so-called So and S1 experiments in REDOR. In the So experiment,
only

13

C π pulses are applied at the end of each rotor period except for the last one. In

this case, the

13

15

C - N dipolar interaction is averaged out over each rotor period by

MAS (Figure 1.12). In addition, other anisotropic interactions such as
eliminated by MAS. The
applied to refocus the

13

13

13

C CSA is also

C π pulses at the end of each odd-numbered rotor period are

C transverse magnetization at the end of each even-numbered

31

rotor period under the effect of isotropic chemical shift field. In the S1 experiment,
π pulses are applied at the end of each rotor period while
the middle of each rotor period. In this case, the
under MAS (Figure 1.12). Due to the
the dipolar field from

15

N spins, the

13

13

13

15

13

C

N π pulses are applied in

15

C - N dipolar coupling is recoupled

15

C - N dipolar interaction and the distribution of

C transverse magnetization decays as a function

of the dephasing time τ, where τ refers to the time period after CP but before FID
acquisition. Therefore, the

13

C signal intensity in S1 experiment is weaker than that in

So experiment. Suppose the

13

C signal intensity in the So and S1 experiments is

respectively So and S1, the REDOR dephasing is defined as
∆S/So = (So - S1)/So

(1.34)

After the dephasing buildup curve of ∆S/So vs τ is achieved, the

13

15

C - N dipolar

coupling can be obtained by comparison with SIMPSON simulation, which is a quantum
mechanics-based numerical simulation program for SSNMR[11]. Once the
dipolar coupling d is known, one can calculate the
3

13

15

15

C - N

C - N internuclear distance r

through the equation d = 3066/r , where d is in Hz and r is in Å.

32

13

Figure 1.12 Dipolar coupling evolution as a function of rotor period during the
dephasing time τ in REDOR. The dipolar interaction is averaged out over each rotor
period by MAS in So experiment but not in S1 experiment.
For REDOR between a spin 1/2 and spin 1/2 pair such as

13

C and

15

N, there is a

so-called universal dephasing curve of ∆S/So vs λ, where λ is the product of the dipolar
coupling d and dephasing time τ (Figure 1.13a)[12]. For REDOR between a spin 1/2
and spin 1 pair such as

13

2

C and H, there is also a universal dephasing curve of ∆S/So

vs λ according to the SIMPSON simulations (Figure 1.13b). However, the maximum
dephasing for

13

2

2

C - H REDOR is only ~2/3 without considering T1 relaxation of H.
2

This is because the H π pulses only cause spin transitions between m = 1 and m = -1

33

2

states whereas the m = 0 state which accounts for 1/3 population of the H spins
2

remains unchanged. In practice, we need to take into account T1 relaxation of H.

Figure 1.13 REDOR universal dephasing curve of ∆S/So vs λ for a spin 1/2 - spin 1/2
pair (panel a) and for a spin 1/2 - spin 1 pair (panel b), respectively. Panel a is edited
from reference [12] and and panel b is obtained from SIMPSON simulations.

1.2 Influenza and HIV fusion peptides
Influenza fusion peptide (IFP)
Influenza virus is a type of retrovirus (i.e. RNA virus) enveloped by a lipid
membrane and it infects the respiratory epithelial cell. There are several different
subtypes of influenza virus based on two proteins on the surface of the virus. The two
viral proteins are hemagglutinin (HA) and neuraminidase (NA). There are 18 HA
subtypes and 11 NA subtypes and many different combinations of HA and NA proteins
are possible[13]. However, only some of the subtypes (e.g. H1N1, H3N2) are in general
circulation among people.
34

During the infection process, the influenza HA protein plays an important role,
where the HA protein is organized as a homotrimer in the viral membrane and consists
of two subunits, HA1 (binding subunit) and HA2 (fusion subunit). HA1 stays completely
outside the virus. HA2 has a ~185 residue N-terminal ectodomain outside the virus, a
~25 residue transmembrane domain, and a ~10 residue C-terminal endodomain (i.e.
cytoplasmic domain) inside the virus. There is a single disulfide bond between cysteine
14 in HA1 and cysteine 137 in HA2[14].
The infection (Figure 1.14) starts with the binding of HA1 to the sialic acids which
are attached to the glycoproteins or glycolipids on the host cell surface, followed by the
endocytosis of the virus into the host cell. After that, protons are pumped into the newlyformed endosome via the cell physiological processes to lower the endosomal pH to ~5.
The low pH then triggers a dramatic conformational change of the HA protein[15], which
is cleaved to yield HA1 and HA2. The two subunits remain linked via a single disulfide
bond. During the conformational change, the ~20 residue fusion peptide (IFP) at the Nterminus of HA2 is exposed. The IFP plays a vital role in the fusion between the viral
and the host cell endosomal membranes, as evidenced by the disruption of the fusion
activity of HA2 with point mutations in the IFP region[16]. Substitution of glutamic acid
for glycine residue right at the amino-terminus of HA2 even completely abolishes its
fusion activity. Earlier studies show that the IFP and transmembrane domain are the
only two regions of HA2 which are deeply inserted in the membrane after the fusion
between the viral and target membranes[17].

35

Figure 1.14 Infection model of influenza virus. The virus recognizes the host cell
(respiratory epithelial cell) via the viral HA1 binding protein and then enters the cell by
endocytosis. The cell physiology pumps protons into the newly-formed endosome and
lowers the pH to ~5 to trigger the fusion between the viral and endosomal membranes.
The viral nucleocapsid is then released to the cell cytoplasm[18].
More background and

13

15

C- N REDOR studies of the structures of membrane-

associated IFP will be discussed in Chapter 2.
HIV fusion peptide (HFP)
Similar to the influenza virus, human immunodeficiency virus (HIV) is another
type of retrovirus and is also enveloped by a lipid membrane. A mature HIV virus has a
diameter of 110 to 128 nm[19]. HIV targets human T-cells through the viral envelope
glycoprotein (Env). The Env protein is also known as gp160 and consists of two
subunits, gp120 and gp41, which are non-covalently associated with each other. Earlier
electron microscopy tomography studies have shown that the wild-type Env protein is
trimeric and each HIV virion has ~14 Env trimers[20, 21]. When HIV infects a host cell
(T cell), the viral gp120 binding protein first recognizes the host cell through the CD4
receptor on the cell surface. After that, gp120 leaves gp41 and binds to both the CD4
receptor and CXCR4 (or CCR5) coreceptor[22]. The exposed gp41 fusion protein binds

36

to the host cell membrane and initiates the HIV-host cell membrane fusion. The HIV
infection process on T cells is displayed in Figure 1.15[23].
During the HIV-host cell membrane fusion process, the HIV gp41 protein
undergoes three states sequentially: (i) pre-fusion native state, (ii) pre-hairpin
intermediate (PHI) state, and (iii) final hairpin state[24]. These three different states are
displayed in Figure 1.16a. The sequence of gp41 fusion protein consists of 345 residues
and can be divided into the N-terminal fusion peptide (FP), N-terminal heptad repeat
region (NHR), loop region, C-terminal heptad repeat region (CHR), membrane-proximal
external region (MPER), transmembrane domain, and the endodomain (Figure 1.16b).
All the segments outside the viral membrane form the so-called ectodomain of gp41.
The heptad unit in NHR and CHR regions contains seven residues a, b, c, d, e, f, and g,
where residues a and d are hydrophobic and residues b, c, e, f, and g are
hydrophilic[25]. In the hairpin state, three NHR helices and three CHR helices form a
hairpin-like six helix bundle (SHB).

37

Figure 1.15 (Left) HIV infection model. (Right) HIV infection process illustrated by
electron microscopy. (a) HIV-host cell membrane binding. (b) hemi-fusion. (c) complete
fusion with the formation of fusion pore. (d) entry of the viral nucleocapsid into the host
cell[23].
The crystal structure of HIV gp120 binding protein has been elucidated by X-ray
diffraction (XRD)[26]. However, there has been no crystal structure for the full-length
gp41 fusion protein. Earlier X-ray crystallography studies have determined the crystal
structures of the hairpin (i.e. SHB) and the gp41 ectodomain lacking the fusion peptide
(Figure 1.17)[25, 27]. For the hairpin structure (SHB), an essential function of it is to
stabilize the membrane fusion pore and prevent it from being collapsed or closed during
the HIV infection process. Previous fluorescence spectroscopy studies show that the
addition of gp41-derived peptides which inhibit the SHB formation (e.g. C34 and N36)
38

can cause the cold-arrested fusion pore to quickly and irreversibly close, indicating that
the SHB formation is not complete by the time when a fusion pore has formed[28]. Once
the SHB formation is complete, this structure stabilizes the membrane fusion pore and
ensures its growth.

Figure 1.16 (a) HIV-host cell membrane fusion model: (i) native-state trimers of gp120
and gp41 prior to fusion, (ii) pre-hairpin intermediate (PHI), and (iii) final hairpin state[24].
(b) Different segments of the HIV gp41 fusion protein.

39

Figure 1.17 Crystal structure of the HIV gp41 ectodomain lacking fusion peptide
(gp41528-683, residues are numbered according to their positions in the gp160 complex).
(a) α helical structure of the gp41 ectodomain, including polar region(blue, 528-540),
NHR(blue, 541-581), loop region(blue,582-627), CHR(blue,628-666), and MPER
(grey,667-683). The structure is 88 Å long. (b) Close-up of the MPER region showing
the exposure of aromatic side chains of Trp 678, Trp 680 and Tyr 681 towards the
membrane, which indicates the potential membrane insertion of MPER residues
adjacent to the transmembrane domain of gp41[27].
Earlier solution NMR studies of micelle-associated HIV fusion peptide (HFP),
which is the N-terminal ~20-30 residues of the gp41 fusion protein, have shown that
HFP adopts a predominantly α helical structure in detergents such as DPC and SDS
(Figure 1.18)[29]. For HFP in lipid membranes, previous SSNMR and IR studies have
both shown that the membrane-associated HFP majorly adopts an oligomeric antiparallel β sheet structure[30, 31].

40

Figure 1.18 Ribbon representation of the HFP30, which is the N-terminal 30 residues
of the HIV gp41 fusion protein, in an SDS micelle[29].
Membrane binding and lipid mixing assays indicate that HFP binds to lipid
membranes and induces lipid mixing (i.e. vesicle fusion)[32]. On the other hand, lipid
mixing assays of HFP23 (the N-terminal 23 residues of gp41), gp41e (the ectodomain of
gp41 excluding HFP23), and gp41e-HFP (the ectodomain of gp41 including HFP23)
indicate that both HFP23 and gp41e-HFP have a much higher fusion activity than
gp41e[33]. All the above data imply that HFP may play an important role in the HIV-host
cell membrane fusion during the infection process.
1.3 Membrane insertions of peptides
Earlier studies of membrane insertion depths of IFP and HFP
The insertion depth of protein and peptide residues in lipid membranes is an
important feature of membrane-bound proteins and peptides. For instance, the
membrane insertion depth of viral fusion peptides such as IFP and HFP is crucial for
their membrane fusion activities[34-36]. Earlier electron paramagnetic resonance (EPR)
41

studies of IFP reveal that the residues in the N- and C-terminal regions penetrate the
membrane 3~6 Å more deeply at the fusogenic pH 5.0 than at the non-fusogenic pH
7.4[34]. For HFP, previous lipid mixing assays and REDOR SSNMR studies together
have shown a strong positive correlation between the membrane insertion depth and
fusion activity of the peptide[35, 37, 38].
Techniques for measuring membrane insertion depths of peptides
A variety of techniques are available to probe the residue-specific locations of
proteins and peptides in membranes. In the next paragraphs, we will discuss some of
these techniques in some detail.
(a) Fluorescence resonance energy transfer (FRET)
This technique is based on the energy transfer from the fluorophore such as
tryptophan to the quencher such as acrylamide or bromolipid. The fluorescence
quenching effect has a fluorophore-quencher distance dependence and can be
described by the Stern-Volmer equation:
Fo/F = 1 + KSV[Q]

(1.35)

where Fo and F represent the fluorescence intensities in the absence and presence of
the quencher Q, respectively, KSV is the Stern-Volmer quenching constant which is
6

proportional to 1/r

and r is the fluorophore-quencher distance, and [Q] is the

concentration of the quencher Q. The membrane insertion depth of HFP has been
studied by the FRET technique[39, 40]. The residue Phe8 of HFP was mutated to Trp
(i.e. F8W mutant) to detect the membrane location of the Trp indole group. The

42

quencher was either acrylamide in aqueous solution or bromine atom attached to the
lipid acyl chain. Comparison of the Trp fluorescence intensities with and without the
LUVs (large unilamellar vesicles) in the aqueous solution containing the quencher
acrylamide showed that the Trp residue was embedded in the membrane interior. A
more precise membrane location of the Trp residue was probed by using the bromolipid
which was either 6,7 Bromo-PC (1-palmitoyl-2-stearoyl (6-7 dibromo)-sn-glycero-3phosphocholine) or 11,12 Bromo-PC (1-palmitoyl-2-stearoyl (6-7 dibromo)-sn-glycero-3phosphocholine). The quenching data indicated that the Trp residue was located closer
to the C6-C7 than to the C11-C12 position in the membrane hydrocarbon core.
(b) Paramagnetic relaxation enhancement (PRE) SSNMR
3+

This technique involves the use of paramagnetic ions such as Dy

and Mn

2+

[41,

42]. The unpaired electron(s) in a paramagnetic ion can significantly enhance the T1
and T2 relaxation rates of nuclei, which is known as the PRE effect. Due to the faster T2
relaxation, the signal of the observed nuclei decays faster in the presence of
6

paramagnetic ions. The PRE effect has a distance dependence of 1/r , where r is the
distance between the unpaired electron and the nuclear spin. This distance dependence
can be described by the Solomon-Bloembergen equation. Below is a modified version
of the equation used in biomolecular NMR[43].

=
r [

3 τc
K
(4τc +
)]1/6
R2
1 + ωh2 τc2

(1.36)

43

where r is the distance between the unpaired electron and the nuclear spin, R2 is the
nuclear T2 relaxation rate solely from the PRE contribution (i.e. the intrinsic T2
relaxation rate is not included in equation 1.36), τc is the correlation time for the
electron-nuclear interaction, ωh is the nuclear Larmor frequency, and K is a constant
given by
2 2 2

K = S(S+1)γ g β /15

(1.37)

where S is the nuclear spin quantum number, γ is the nuclear gyromagnetic ratio, g is
the electronic g factor, and β is the Bohr magneton. To calculate the distance r, an
approximation could be made that τc is equal to the correlation time of the protein or
peptide[43].
Since the aqueous paramagnetic ions are either in the bulk solution or at the
water-membrane interface, one can probe the location of a protein or peptide with
respect to the membrane surface.
(c) Spin diffusion
13

X-nucleus ( C,

15

N, etc)

detected,

1

H spin diffusion is another recently

developed SSNMR technique for membrane location measurement of peptides and
proteins[44-46]. The experiment proceeds in a time sequence as (1) select the

1

H

magnetization from water or the mobile lipid chain-end methyl group using a T2 filter; (2)
1

allow a mixing time tm for the H spin diffusion from water or lipids to a rigid protein or

44

1

13

1

15

peptide; (3) apply H→ C or H→ N cross polarization (CP) where the protein or
peptide is

13

C or

15

N labeled; and (4) detect the

13

C or

15

N signal. This technique can

semiquantitatively measure the insertion depth of the protein or peptide relative to the
1

membrane surface or bilayer center. The pulse sequence of the X-nucleus detected, H
spin diffusion experiment is displayed in Figure 1.19[44].

1

Figure 1.19 X-nucleus detected, H spin diffusion pulse sequence[44]. The sequence
o

starts with an initial saturation (destroy) of any X magnetization by several 90 pulses,
1

followed by a dipolar dephasing period. After the H excitation, a T2 filter is applied
1
o
during the period τd to select the H magnetization which has a long T2. The 180 pulse
in the middle of the period τd is applied to refocus chemical shift and field
o

1

inhomogeneity. The 90 pulse at the end of τd stores the remaining H magnetization
with a long T2 on the z axis, followed by a spin diffusion period with a variable tm (e.g.
1

0-100 ms). During the mixing time tm, the existing H z-magnetization is transferred to
1 1
1
nearby protons via H- H dipolar coupling. In the end, a H→X CP is applied and then
1

the X signal is detected with H decoupling.
13

In the X-nucleus ( C,

15

N, etc) detected,

1

H spin diffusion experiments, the

signal intensity of X is detected as a function of the mixing time tm. The X signal
intensity at each tm is then normalized to that at the final mixing time and plotted as a

45

1/2

function of (tm)

. Finally, the best-fit simulation of the buildup curve of the normalized
1/2

X signal intensity versus (tm)

1

1

yields a semiquantitative H- H distance, where the

1

former H refers to those which are either directly bonded to or right next to (~2 Å) the
1

X-nucleus in the protein or peptide and which are the major H source for the CP while
1

the latter H refers to those from water or the mobile lipid chain-end methyl group which
have a long T2 time and thus their magnetization remains after the T2 filter in the pulse
sequence.
(d) REDOR SSNMR
As previously discussed in the "REDOR" section, this technique is a very useful
tool for measuring heteronuclear dipolar couplings from which the internuclear distances
can be calculated. There are several different types of REDOR such as
19

F, and

13

13

31

C- P,

13

C-

2

C- H REDOR which can be used to probe the membrane locations

(insertion depths) of peptides and proteins[35, 47, 48]. For instance,
can probe the membrane insertion depth of the

13

13

31

C- P REDOR

CO-labeled residue of a peptide

relative to the membrane surface since the membrane headgroup region contains
For a peptide-membrane sample in which the peptide contains one

13

31

P.

CO-labeled

residue and the membrane contains ~10% (mole fraction) 1-palmitoyl-2-[1619

fluoropalmitoyl]-phosphatidylcholine (16- F-DPPC) lipid,
to probe the membrane insertion depth of the
46

13

13

19

C- F REDOR can be used

CO-labeled residue with respect to the

membrane bilayer center. Three HFP constructs, which are respectively the V2E
mutated HFP monomer (HFPmn_V2E), wild-type HFP monomer (HFPmn), and wildtype HFP trimer (HFPtr), have been studied by

13

31

C- P and

13

19

C- F REDOR and the

data show that the membrane insertion depths of the three HFP constructs have an
increasing order of HFPmn_V2E < HFPmn < HFPtr (Figure 1.20)[35].

13

31

C- P and

13

19

exp

C- F REDOR experimental dephasing (∆S/So)
as a
13
function of the CO-labeled residue. (b) Membrane insertion models of the anti-parallel
β sheet HFPmn_V2E, HFPmn, and HFPtr, respectively[35].

Figure 1.20 (a)

47

(e) Electron paramagnetic resonance (EPR)
EPR is a commonly used technique for distance measurement which requires
site-specific mutagenesis. One example of such mutagenesis is the introduction of a
cysteine into a protein or peptide to replace the residue of interest. The thiol group of
cysteine can specifically react with a spin label which contains a paramagnetic center
(e.g. nitroxyl radical) and thus attach the spin label to the protein or peptide. The
distance measurement by EPR is based on the electron-electron spin interaction where
2+

one electron is from the spin label and the other could be from metal ions such as Ni .
EPR has been used to detect the membrane insertion depths of both IFP and HFP[34,
49, 50]. For the membrane-associated IFP, the residues in the N- and C-terminal
regions penetrate the membrane 3~6 Å more deeply at the fusogenic pH 5.0 than at the
non-fusogenic pH 7.4[34]. For HFP in the membrane containing 80% (mole fraction)
POPC (1-palmitoyl-2-oleoyl-sn-gycero-3-phosphocholine) and 20% POPG (1-palmitoyl2-oleoyl-sn-gycero-3-phosphoglycerol), EPR studies have shown that residues 4, 7, 12,
and 15 penetrate into the membrane to a similar depth which is ~8 Å below the level of
the lipid phosphate groups at the membrane surface[50].
The membrane location studies of peptides by
discussed in more detail in chapter 4.

48

13

2

C- H REDOR SSNMR will be

REFERENCES

49

REFERENCES

1.

Duer, M.J., Solid-State NMR Spectroscopy: Principles and Applications. 2002.

2.

Lambert, J.B. and E.P. Mazzola, Nuclear Magnetic Resonance Spectroscopy: An
Introduction to Principles, Applications, and Experimental Methods. 2004.

3.

Schmidt-Rohr, K. and H.W. Spiess, Multidimentional Solid-State NMR and
Polymers. 2005.

4.

Harris, R.K., Nuclear Magnetic Resonance Spectroscopy. 1986.

5.

Keeler, J., Understanding NMR Spectroscopy. 2002.

6.

Andrew, E.R., A. Bradbury, and R.G. Eades, NUCLEAR MAGNETIC
RESONANCE SPECTRA FROM A CRYSTAL ROTATED AT HIGH SPEED.
Nature, 1958. 182(4650): p. 1659-1659.

7.

Gullion, T. and J. Schaefer, Rotational-Echo Double-Resonance NMR. Journal of
Magnetic Resonance, 1989. 81(1): p. 196-200.

8.

Hartmann, S.R. and E.L. Hahn, NUCLEAR DOUBLE RESONANCE IN
ROTATING FRAME. Physical Review, 1962. 128(5): p. 2042-&.

9.

Yannoni, C.S., HIGH-RESOLUTION NMR IN SOLIDS - THE CPMAS
EXPERIMENT. Accounts of Chemical Research, 1982. 15(7): p. 201-208.

10.

Meier, B.H., CROSS POLARIZATION UNDER FAST MAGIC ANGLE SPINNING
- THERMODYNAMICAL CONSIDERATIONS. Chemical Physics Letters, 1992.
188(3-4): p. 201-207.

11.

Bak, M., J.T. Rasmussen, and N.C. Nielsen, SIMPSON: A general simulation
program for solid-state NMR spectroscopy. Journal of Magnetic Resonance,
2000. 147(2): p. 296-330.

12.

Gullion, T., Introduction to rotational-echo, double-resonance NMR. Concepts in
Magnetic Resonance, 1998. 10(5): p. 277-289.
50

13.

CDC, Types of influenza viruses. http://www.cdc.gov/flu/about/viruses/types.htm,
2014.

14.

Chen, J., J.J. Skehel, and D.C. Wiley, N- and C-terminal residues combine in the
fusion-pH influenza hemagglutinin HA(2) subunit to form an N cap that
terminates the triple-stranded coiled coil. Proceedings of the National Academy
of Sciences of the United States of America, 1999. 96(16): p. 8967-8972.

15.

Hughson, F.M., STRUCTURAL CHARACTERIZATION OF VIRAL FUSION
PROTEINS. Current Biology, 1995. 5(3): p. 265-274.

16.

Gething, M.J., et al., STUDIES ON THE MECHANISM OF MEMBRANE-FUSION
- SITE-SPECIFIC MUTAGENESIS OF THE HEMAGGLUTININ OF INFLUENZAVIRUS. Journal of Cell Biology, 1986. 102(1): p. 11-23.

17.

Durrer, P., et al., H+-induced membrane insertion of influenza virus
hemagglutinin involves the HA2 amino-terminal fusion peptide but not the coiled
coil region. Journal of Biological Chemistry, 1996. 271(23): p. 13417-13421.

18.

Curtis-Fisk, J., Structural studies of the Influenza and HIV viral fusion proteins
and bacterial inclusion bodies. Ph.D. Thesis, Michigan State University, 2009.

19.

Gentile, M., et al., DETERMINATION OF THE SIZE OF HIV USING
ADENOVIRUS TYPE-2 AS AN INTERNAL LENGTH MARKER. Journal of
Virological Methods, 1994. 48(1): p. 43-52.

20.

Zhu, P., et al., Electron tomography analysis of envelope glycoprotein trimers on
HIV and simian immunodeficiency virus virions. Proceedings of the National
Academy of Sciences of the United States of America, 2003. 100(26): p. 1581215817.

21.

Zhu, P., et al., Distribution and three-dimensional structure of AIDS virus
envelope spikes. Nature, 2006. 441(7095): p. 847-852.

22.

White, J.M., et al., Structures and mechanisms of viral membrane fusion proteins:
Multiple variations on a common theme. Critical Reviews in Biochemistry and
Molecular Biology, 2008. 43(3): p. 189-219.

23.

Grewe, C., A. Beck, and H.R. Gelderblom, HIV - EARLY VIRUS-CELL
INTERACTIONS. Journal of Acquired Immune Deficiency Syndromes and
Human Retrovirology, 1990. 3(10): p. 965-974.
51

24.

Sackett, K., et al., Hairpin Folding of HIV gp41 Abrogates Lipid Mixing Function
at Physiologic pH and Inhibits Lipid Mixing by Exposed gp41 Constructs.
Biochemistry, 2009. 48(12): p. 2714-2722.

25.

Chan, D.C., et al., Core structure of gp41 from the HIV envelope glycoprotein.
Cell, 1997. 89(2): p. 263-273.

26.

Kwong, P.D., et al., Structure of an HIV gp120 envelope glycoprotein in complex
with the CD4 receptor and a neutralizing human antibody. Nature, 1998.
393(6686): p. 648-659.

27.

Buzon, V., et al., Crystal Structure of HIV-1 gp41 Including Both Fusion Peptide
and Membrane Proximal External Regions. Plos Pathogens, 2010. 6(5).

28.

Markosyan, R.M., F.S. Cohen, and G.B. Melikyan, HIV-1 envelope proteins
complete their folding into six-helix bundles immediately after fusion pore
formation. Molecular Biology of the Cell, 2003. 14(3): p. 926-938.

29.

Jaroniec, C.P., et al., Structure and dynamics of micelle-associated human
immunodeficiency virus gp41 fusion domain. Biochemistry, 2005. 44(49): p.
16167-16180.

30.

Nieva, J.L., S. Nir, and J. Wilschut, Destabilization and fusion of zwitterionic large
unilamellar lipid vesicles induced by a beta-type structure of the HIV-1 fusion
peptide. Journal of Liposome Research, 1998. 8(2): p. 165-182.

31.

Schmick, S.D. and D.P. Weliky, Major Antiparallel and Minor Parallel beta Sheet
Populations Detected in the Membrane-Associated Human Immunodeficiency
Virus Fusion Peptide. Biochemistry, 2010. 49(50): p. 10623-10635.

32.

Buzon, V., E. Padros, and J. Cladera, Interaction of fusion peptides from HIV
gp41 with membranes: A time-resolved membrane binding, lipid mixing, and
structural study. Biochemistry, 2005. 44(40): p. 13354-13364.

33.

Cheng, S.F., et al., The fusion peptide domain is the primary membrane-inserted
region and enhances membrane interaction of the ectodomain of HIV-1 gp41.
Molecular Membrane Biology, 2010. 27(1): p. 31-44.

34.

Han, X., et al., Membrane structure and fusion-triggering conformational change
of the fusion domain from influenza hemagglutinin. Nature Structural Biology,
2001. 8(8): p. 715-720.
52

35.

Qiang, W., Y. Sun, and D.P. Weliky, A strong correlation between fusogenicity
and membrane insertion depth of the HIV fusion peptide. Proceedings of the
National Academy of Sciences of the United States of America, 2009. 106(36): p.
15314-15319.

36.

Tamm, L.K. and X. Han, Viral fusion peptides: A tool set to disrupt and connect
biological membranes. Bioscience Reports, 2000. 20(6): p. 501-518.

37.

Yang, R., et al., A trimeric HIV-1 fusion peptide construct which does not selfassociate in aqueous solution end which Has 15-fold higher membrane fusion
rate. Journal of the American Chemical Society, 2004. 126(45): p. 14722-14723.

38.

Gabrys, C.M., et al., Solid-State Nuclear Magnetic Resonance Measurements of
HIV Fusion Peptide (CO)-C-13 to Lipid P-31 Proximities Support Similar Partially
Inserted Membrane Locations of the alpha Helical and beta Sheet Peptide
Structures. Journal of Physical Chemistry A, 2013. 117(39): p. 9848-9859.

39.

Agirre, A., et al., Interactions of the HIV-1 fusion peptide with large unilamellar
vesicles and monolayers. A cryo-TEM and spectroscopic study. Biochimica Et
Biophysica Acta-Biomembranes, 2000. 1467(1): p. 153-164.

40.

Haque, M.E., et al., Properties and structures of the influenza and HIV fusion
peptides on lipid membranes: Implications for a role in fusion. Biophysical
Journal, 2005. 89(5): p. 3183-3194.

41.

Grobner, G., C. Glaubitz, and A. Watts, Probing membrane surfaces and the
location of membrane-embedded peptides by C-13 MAS NMR using lanthanide
ions. Journal of Magnetic Resonance, 1999. 141(2): p. 335-339.

42.

Su, Y., R. Mani, and M. Hong, Asymmetric insertion of membrane proteins in lipid
bilayers by solid-state NMR paramagnetic relaxation enhancement: A cellpenetrating peptide example. Journal of the American Chemical Society, 2008.
130(27): p. 8856-8864.

43.

Battiste, J.L. and G. Wagner, Utilization of site-directed spin labeling and highresolution heteronuclear nuclear magnetic resonance for global fold
determination of large proteins with limited nuclear overhauser effect data.
Biochemistry, 2000. 39(18): p. 5355-5365.

53

44.

Gallagher, G.J., M. Hong, and L.K. Thompson, Solid-state NMR spin diffusion for
measurement of membrane-bound peptide structure: Gramicidin A. Biochemistry,
2004. 43(24): p. 7899-7906.

45.

Huster, D., X.L. Yao, and M. Hong, Membrane protein topology probed by H-1
spin diffusion from lipids using solid-state NMR spectroscopy. Journal of the
American Chemical Society, 2002. 124(5): p. 874-883.

46.

Wang, T., H. Yao, and M. Hong, Determining the depth of insertion of
dynamically invisible membrane peptides by gel-phase H-1 spin diffusion
heteronuclear correlation NMR. Journal of Biomolecular Nmr, 2013. 56(2): p.
139-148.

47.

Xie, L., et al., Residue-specific membrane location of peptides and proteins using
specifically and extensively deuterated lipids and C-13-H-2 rotational-echo
double-resonance solid-state NMR. Journal of Biomolecular NMR, 2013. 55(1): p.
11-17.

48.

Toke, O., et al., Secondary structure and lipid contact of a peptide antibiotic in
phospholipid bilayers by REDOR. Biophysical Journal, 2004. 87(1): p. 662-674.

49.

Gordon, L.M., et al., THE AMINO-TERMINAL PEPTIDE OF HIV-1
GLYCOPROTEIN-41
INTERACTS
WITH
HUMAN
ERYTHROCYTEMEMBRANES
- PEPTIDE
CONFORMATION,
ORIENTATION AND
AGGREGATION. Biochimica Et Biophysica Acta, 1992. 1139(4): p. 257-274.

50.

Lai, A.L., et al., Fusion Activity of HIV gp41 Fusion Domain Is Related to Its
Secondary Structure and Depth of Membrane Insertion in a CholesterolDependent Fashion. Journal of Molecular Biology, 2012. 418(1-2): p. 3-15.

54

Chapter 2 - Structural studies of Influenza fusion peptide (IFP)
2.1 Background
Influenza virus is a type of retrovirus (i.e. RNA virus) enveloped by a lipid
membrane and it infects the respiratory epithelial cell. During the infection process, the
influenza fusion peptide (IFP), which is the N-terminal ~20 residues of the HA2 subunit
of the hemagglutinin protein, plays a vital role in the fusion between the viral and the
host cell endosomal membranes, as evidenced by the disruption of the fusion activity of
HA2 with point mutations in the IFP region[1]. Substitution of glutamic acid for glycine
residue right at the amino-terminus of HA2 even completely abolishes its fusion activity.
The structure of IFP depends on the sequence and whether it is in detergents or
lipid membranes. For instance, the H3_20 peptide, which has the sequence
GLFGAIAGFIENGWEGMIDGGCGKKKK with the underlined part representing the wild
type 20-residue IFP of serotype H3, adopts open structures at both pH 5.0 and 7.4 in
detergents

whereas

the

H1_23

peptide,

which

has

the

sequence

GLFGAIAGFIEGGWTGMIDGWYGSGKKKKD with the underlined part representing the
wild type 23-residue IFP of serotype H1, adopts a closed N-helix/turn/C-helix hairpin
structure at pH 7.4 in detergents and a similar structure at pH 4.0 according to the
similar HSQC (heteronuclear single quantum correlation) spectra at both pHs (Figure
2.1)[2, 3]. The H3_20 sequence differs from H1_23 in that the 12th and 15th residues
are Asn and Glu in H3_20 and Gly and Thr in H1_23, respectively. Besides, H1_23 has
three extra C-terminal residues, WYG. Although the H3_20 peptide has open structures
in detergents at both pH 5.0 and 7.4, there are some conformational changes between
the two pHs according to the study by Han and Tamm[2]. At pH 5.0, H3_20 adopts an
55

N-terminal α helix from residue Leu2 to Ile10, followed by a turn which redirects the Cterminal region of the peptide and is stabilized by hydrogen bonds from the amide NHs
of Glu11 and Asn12 to the carbonyls of Gly8 and Phe9, and a C-terminal 310 helix. At
pH 7.4, the N-terminal α helix ranges from residue Leu2 to Phe9, followed by a turn
stabilized by hydrogen bonds from the amide NHs of Glu11 and Asn12 to the carbonyls
of Gly8 and Phe9. The C-terminal region forms an extended structure. Interestingly, a
very recent study by Lorieau and Bax has shown that the H1_23 peptide adopts a
helical structure at both N- and C-termini at pH 7.4[4]. This study also reports that the
H1_20 peptide is highly dynamic and in equilibrium between a ~11% population of
closed structure and a ~89% population of open structure when solubilized by DPC
(dodecylphosphocholine) micelles at pH 7.4. Comparison of the major open structure of
H1_20 with the predominant closed structure of H1_23 indicates that the three Cterminal residues, WYG, play a key role in stabilizing the closed structure of H1_23 in
detergents.

56

Figure 2.1 Structures of the influenza hemagglutinin fusion peptides H3_20 (a) and
H1_23 (b) in DPC detergents[2, 3].
Additional liquid state NMR studies have been conducted on the wild type H1_23
and its mutant H1_23_G8A where the residue Gly8 was substituted by Ala[5].
Relaxation data indicate that the wild type H1_23 adopts a closed structure (helical
hairpin) in DPC micelles at pH 7. However, at pH 4, H1_23 adopts a major (~80%
population) closed structure and a minor (~20% population) open structure where the
two structures are rapidly exchanging with each other at a rate of ~40 kHz. For the
mutant H1_23_G8A in detergents at pH 7, it adopts a small fraction (~15%) of closed
structure and a large fraction of at least two open structures which are classified as
extended and L-shaped, respectively (Figure 2.2). The topology of these N-helix/turn/Chelix structures can be described by the interhelical angle which is defined in such a

57

way that each helix axis is represented as a vector from the N- to the C-terminus and
the angle between the two vectors refers to the interhelical angle. The closed, extended,
o

o

o

and L-shaped structures of the H1_23_G8A have interhelical angles of 159 ± 1 , 73 ±
o

o

o

11 , and 110 ± 6 , respectively[5]. It has previously been reported that cells expressing
the G8A mutant of HA2 had little membrane fusion activity[6].

Figure 2.2 N-helix/turn/C-helix structures of the H1_23_G8A mutant in DPC micelles at
pH 7[5].
For the H3_20 peptide in lipid membranes, previous SSNMR studies have shown
that it adopts predominantly helical structure in membranes without cholesterol and
predominantly β sheet structure in membranes containing ~30% (mole fraction)
cholesterol (e.g. LM3 membrane)[7]. In membranes without cholesterol, H3_20 adopts
an N-helix/turn/C-helix motif at both pH 5.0 and 7.4. Two different conformations of the
turn are observed at pH 5.0 since there are two distinct

13

C chemical shift sets for

Glu11 in the turn region (Figure 2.3)[8]. The fusion activity of IFP has a strong pH
dependence. Previous fluorescence and hemolytic studies indicate that IFP induces
more rapid and extensive lipid mixing (membrane fusion) at pH 5.0 than at physiological
pH 7.4 for cholesterol-free liposomes[9-11]. However, for cholesterol-containing

58

liposomes, IFP induces lipid mixing at low pHs such as 5.0 but not at high pHs such as
7.4[9, 11].

Figure 2.3 Backbone structures of the membrane-associated H3_20 peptide at pH 5.0.
Residue E11 is in green, N12 is in red, the hydrophobic residues L2, F3, I6, F9, and I10
are in gold[8].
In the present study,

13

15

C- N REDOR SSNMR is employed to quantitatively

probe the population and interhelical proximity of the H3_20 structures in lipid
membranes.
2.2 Sample preparation
Peptide sequence and synthesis
The

H3_20

peptide,

which

has

the

sequence

GLFGAIAGFIENGWEGMIDGGGKKKKG with the underlined part representing the wild
type 20-residue IFP of serotype H3, was chemically synthesized by Fmoc solid phase
peptide synthesis (SPPS) with a specific

13

CO label at residue G16 and a specific

15

N

label at residue F9[12]. The Fmoc SPPS protocol is described in detail in Appendix II.
After the synthesis, H3_20 was purified by reversed-phase high pressure liquid
chromatography (HPLC) using a preparative C4 column. The peptide purity was
checked by mass spectrometry such as matrix-assisted laser desorption ionization
59

(MALDI) and electrospray ionization (ESI). Comparison of peak intensities in the mass
spectrum indicated the peptide was >95% pure.
REDOR sample preparation
The REDOR samples were prepared in a time sequence of (1) dissolving 40 µmol
of 1,2-di-O-tetradecyl-sn-glycero-3-phosphocholine (DTPC) and 10 µmol of 1,2-di-Otetradecyl-sn-glycero-3-[phospho-(1'-rac-glycerol)] (DTPG) lipids in 3~4 mL chloroform;
(2) removing the chloroform using N2 gas in the fume hood; (3) drying the lipid film in a
vacuum desicator overnight; (4) suspending the lipid film in ~2 mL pH 5.1 buffer which
contains 5 mM 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES) and 10 mM
2-(N-morpholino)ethanesulfonic acid (MES); (5) performing ten cycles of "freeze/thaw",
where the "freeze" in liquid N2 breaks apart the lipid vesicles and the "thaw" in water
bath at room temperature drives the lipid molecules to reform the vesicles; (6)
performing ~20 extrusions using a filter paper with pores of 100 nm in diameter (each
time only 1 mL lipid solution was extruded due to the volume limit of the syringe); (7)
dissolving 2 µmol of the H3_20 peptide in ~30 mL pH 5.1 buffer; (8) adding the peptide
solution to the extruded lipid vesicle solution dropwise with gentle shaking (avoid violent
shaking in case of peptide oxidation at the methionine site); (9) mixing the peptide/lipid
solution overnight on a rotator; (10) ultracentrifugation at ~100,000 g for 4 hours; (11)
lyophilizing the pellet overnight; (12) adding 10 µL pH 5.1 buffer to the bottom of a 4mm
rotor, packing the lyophilized pellet in the rotor, adding another 10 µL buffer on top of
the sample, and finally allowing overnight sample hydration before running the REDOR
experiment. The DTPC:DTPG (4:1) lipid composition reflects a small fraction of

60

negatively charged lipids and a significant fraction of phosphatidylcholine lipids in the
membranes of host epithelial cells of the influenza virus[13]. Ether- rather than esterlinked lipids were used to reduce the natural abundance (n.a.)

13

CO signal in the

REDOR spectra.
2.3

13

15

C- N REDOR experiments

REDOR pulse sequence and parameters
The
1

13

15

1

C- N REDOR pulse sequence consists of (1) H π/2 pulse which rotates
1

H magnetization from z axis to the xy plane, (2) simultaneous H and
1

which transfer magnetization from H to
13

13

C to enhance the

13

13

C CP pulses

C signal intensity, (3)

C π pulses at the end of each rotor period except the last one without (So experiment)

and with (S1 experiment)

15

N π pulses in the middle of each rotor period, in the
1

meantime, two pulse phase modulation (TPPM) H decoupling π pulses are turned on
till the end of
13

13

C signal acquisition (Figure 2.4). The period right after CP but prior to

C acquisition is called dephasing time τ. The REDOR experimental conditions
o

included 10 kHz MAS frequency, sample cooling gas temperature of -50 C with actual
o

1

sample temperature of approximately -30 C, 50 kHz H π/2 pulse and CP pulse, 80
1

kHz H decoupling pulses during both the dephasing and acquisition periods, 61-65 kHz
ramped

13

C CP pulse, 8.0 µs

13

C π pulses, and 10.0 µs

61

15

N π pulses. The pulse delay

was 1 s for τ = 2 ms, 8 ms, and 16 ms, 1.5 s for τ = 24 ms and 32 ms, and 2 s for τ = 40
ms and 48 ms, respectively. The data collection time is typically 2~3 hours for τ = 2 and
8 ms, 6~12 hours for τ = 16 ms, ~1 day for τ = 24 ms, ~1.5 days for τ = 32 ms, ~2 days
for τ = 40 ms, and 2~3 days for τ = 48 ms, respectively. The pulses were optimized
using

the

lyophilized

setup

peptide

I4

which

had

the

sequence

Ac-

AEAAAKEAAAKEAAAKA-NH2 with N-terminal acetylation and C-terminal amidation.
The I4 peptide was synthesized with a

13

CO label at residue A9 and an

15

N label at

residue A13. Previous SSNMR studies have shown that I4 is predominantly (83±6%)
α helical at A9[14]. The internuclear distance between A9
13

13

CO and A13

15

15

N of I4 is

~4.1 Å in the α helical structure which corresponds to a C- N dipolar coupling of ~44
3

Hz (d(Hz) = 3066/r (Å))[15].

Figure 2.4

13

15

C- N REDOR pulse sequence.
62

As mentioned before, the H3_20 peptide was chemically synthesized with a
specific

13

CO label at G16 (G16C) and a specific

15

N label at F9 (F9N). This labeling

scheme is based on earlier solution NMR studies. The internuclear distance between
F9N and G16C is ~11.5 Å in the open structure of H3_20 at pH 5.0 in detergents and
~3.9 Å in the closed structure of H1_23 at pH 4.0 in detergents[2, 3]. If H3_20 adopts an
open structure in the membrane where the F9N and G16C are ~11.5 Å apart with a
corresponding

13

15

C- N dipolar coupling d ≈ 2 Hz, there will be no

13

15

C- N REDOR

buildup of ∆S/So vs τ. However, if H3_20 adopts a closed structure in the membrane
where the F9N and G16C are ~3.9 Å apart with a corresponding

13

15

C- N dipolar

coupling d ≈ 52 Hz, significant buildup of ∆S/So vs τ from 2 ms to 48 ms will be
observed, as displayed in Figure 2.5.

Figure 2.5
13

15

13

15

C- N REDOR dephasing curves of ∆S/So vs τ simulated by SIMPSON

for C- N dipolar couplings of 2 Hz (black) and 52 Hz (red), where 2 Hz corresponds
to an open structure and 52 Hz corresponds to a closed structure of H3_20, respectively.
63

REDOR spectra and dephasing curve of ∆S/So vs τ
To detect whether H3_20 adopts predominantly open or closed structures in the
membrane,

13

15

C- N REDOR was conducted under experimental conditions described

earlier. The experimental dephasing curve of ∆S/So vs τ was displayed in Figure 2.6a.
The So, S1, and ∆S spectra for τ = 32 ms were displayed in Figure 2.6b, where ∆S
spectrum refers to the difference spectrum of So - S1. All the So and S1 spectra at each
τ were processed with 20 Hz exponential line broadening and baseline correction. The
exp

(∆S/So)

were calculated from the So and S1 intensities integrated in a 3 ppm window

centered at the

13

CO peak shift. The G16

13

CO peak shift of 177.1 ppm is consistent

with helical conformation of H3_20[16]. The experimental errors of So and S1 were
calculated as the standard deviation of the integrals of twelve noise regions with each in
a 3 ppm window in the So and S1 spectra, respectively.

64

Figure 2.6 (a) Experimental curve of ∆S/So vs τ of the H3_20 peptide; (b)

13

CO peaks

in the So (black), S1 (red), and ∆S (green) spectra for τ = 32 ms.
2.4 Result discussion and conclusions
exp

Natural abundance calibration of (∆S/So)

exp

The significant buildup of (∆S/So)
predominantly closed

vs τ of H3_20 supports that H3_20 adopts

rather than open structure. SIMPSON simulations were

performed to quantitatively analyze the structure models of H3_20 in the membrane. To
make the data analysis more accurate, natural abundance (n.a.) contributions to

65

exp

(∆S/So)

from unlabeled

13

lab

CO signals were removed to yield (∆S/So)

to ∆S/So only for the labeled G16

13

CO signal. The n.a. correction was done as follows:

Soexp = Solab + Sona = 0.99 + 0.011x 26 = 1.276
lab
where So refers to the labeled G16

to the n.a.

13

which refers

13

(2.1)

na
CO So signal which equals 0.99 and So refers

CO So signal for the 26 unlabeled residues in H3_20.
26

S1exp = S1lab + S1na = S1lab + ∑ S1na
k

(2.2)

k =1

where k represents the number of the 26 unlabeled residues in H3_20.
For each unlabeled residue,

Sona − S1na
0.011 − S1na
∆S na
=
(
)
(=
)
So
0.011
Sona

S1na 0.011 − 0.011x(
=

(2.3)

∆S na
)
So

(2.4)

For all the 26 unlabeled residues,
26

26

na
∑ S1k =
∑ [0.011 − 0.011x(

=
k 1=
k 1

26 ∆S
∆S na
) ]=
0.286 − 0.011x ∑ (
)kna
So
=
k 1 So

(2.5)

Substitution of equation 2.5 into 2.2 yields
26

S1exp =S1lab + 0.286 − 0.011x ∑ (
k =1

∆S na
)k
So

(2.6)

Therefore,

66

26

∆S exp
(
)
=
So

− S1exp
=
Soexp

Soexp

1.276 − S1lab − 0.286 + 0.011x ∑ (
k =1

∆S na
)
So k

1.276
26

0.99 − S1lab + 0.011x ∑ (
k =1

=

∆S na
)
So k

(2.7)

1.276

Rearrangement of equation 2.7 yields

S1lab =
0.99 − 1.276 x(

26 ∆S
∆S exp
) + 0.011x ∑ (
)na
k
S0
k =1 S0

(2.8)

Finally,

∆S lab Solab − S1lab 0.99 − S1lab 1.276 ∆S exp 0.011 26 ∆S na
x(
x∑(
(
)
) −
)k
=
=
=
0.99
0.99
0.99 k =1 So
So
So
Solab

To simulate (

(2.9)

∆S na
)k by SIMPSON, the 13CO-15N dipolar coupling needs to be
So

provided and thus the

13

15

exp

CO- N distance needs to be estimated. Since the (∆S/So)

vs τ of H3_20 has a significant buildup (Figure 2.6a), the peptide may adopt a closed
structure in which the labeled G16

13

CO and F9

15

N nuclei are close to each other.

Therefore, a reasonable way to estimate the distance between the n.a.
labeled
The

13

15

13

CO and

N in H3_20 is to measure the distance in the closed structure of H1_23[3].
15

CO- N distance can be either mathematically calculated using the atomic

coordinates of H1_23 or directly measured using a structure visualizing software such
as PyMol. Table 2.1 displays the distances calculated using the atomic coordinates of
67

H1_23. For other

13

the corresponding (

Table 2.1

13

15

CO- N distances not listed in Table 2.1, they are beyond 8 Å and

∆S na
) ≈ 0.
So

15

CO- N distances in the closed structure of H1_23[3].
13

a
b

15

CO- N spin pair

r (Å)

F3CO-F9N
G4CO-F9N
A5CO-F9N
I6CO-F9N
A7CO-F9N
G8CO-F9N
F9CO-F9N
I10CO-F9N
E11CO-F9N
a
G12CO-F9N
G13CO-F9N
W14CO-F9N
b
T15CO-F9N
M17CO-F9N
I18CO-F9N
D19CO-F9N
G20CO-F9N

7.38
5.40
3.81
3.74
3.14
1.33
2.45
4.69
5.64
5.27
5.23
6.58
6.52
5.36
7.73
7.47
7.28

The 12th residue is Gly in H1_23 and Asn in H3_20.
The 15th residue is Thr in H1_23 and Glu in H3_20.
The

13

15

3

CO- N dipolar coupling d in Hz is calculated via the equation d = 3066/r ,

where r is the

13

15

CO- N distance in Å. After obtaining the dipolar coupling d, SIMPSON

simulations were performed to produce the (

∆S na
) from τ = 2 ms to τ = 48 ms for each
So

68

13

15

15
CO- N spin pair listed in Table 2.1. The sum of (

∆S na
) for each τ for all the 13COSo
26

N spin pairs in Table 2.1 approximately equaled the ∑ (
k =1

the experimental value of (

equation 2.9 to yield the (

∆S na
)k in equation 2.9. At last,
S0

26 ∆S
∆S exp
)na
)
k were plugged in
and simulated value of ∑ (
So
k =1 S0

∆S lab
∆S lab
) for each τ. The buildup curve of (
) vs τ after the
So
So

n.a. correction is displayed in Figure 2.7. The (

∆S exp
∆S lab
)
) was typically higher than (
So
So

but the difference was less than 10% for each τ from 2 ms to 48 ms.
lab

SIMPSON simulations of (∆S/So)

Quantitative analysis of (

vs τ

∆S lab
) vs τ was first done for a single closed structure
So

model. SIMPSON simulations were performed for a range of

13

15

CO- N dipolar coupling

2
d. For each d, the χ (d ) was calculated via

[(

χ 2 (d ) = ∑
τ

∆S lab
∆S sim
) (τ ) − (
) (d ,τ )]2
So
So

(2.10)

σ (τ )2

2
where χ (d ) was summed over the seven values of τ from 2 ms to 48 ms and σ (τ )

was the uncertainty of (

∆S lab
) . The best-fit d was 32 Hz with a minimum χ 2 (d ) of 436.
So

2
The χ (d ) of 436 was much larger than the number of degrees of fitting ν of 6 where ν

69

= the number of fitted data points - the number of fitting parameters. The poor fitting of
this structure model is displayed in Figure 2.7.
Secondly, a "closed/open" structure model was considered. In this model, H3_20
adopts an open structure with a fraction of fo and a
a closed structure with a fraction of fc and a
fc = 1 and do ≈ 0 with consequent (

for a range of

13

13

13

15

CO- N dipolar coupling of do and

15

CO- N dipolar coupling of dc, where fo +

∆S o
) ≈ 0. SIMPSON simulations were performed
So

15

CO- N dipolar coupling dc and a range of the fraction fc from 0.01 to

2
1.00 in an increment of 0.01. For each dc and fc, the χ (dc , fc ) was calculated via

[(

χ 2 (dc , fc ) = ∑
τ

∆S lab
∆S c
) (τ ) − fc x (
) (dc ,τ )]2
So
So

(2.11)

σ (τ )2

2
where χ (dc , fc ) was summed over the seven values of τ. The best-fit dc = 51 Hz, fc =
2
2
0.64, χ (dc , fc ) = 33. However, the χ = 33 was still significantly larger than the

number of degrees of fitting ν = 5. Therefore, the "closed/open" model was also
statistically unreasonable. The poor fitting of this model is displayed in Figure 2.7.
At last, a "closed/semi-closed" structure model was considered. In this model,
H3_20 adopts a closed structure with a fraction of fc and a
dc and a semi-closed structure with a fraction of fs and a

70

13

13

15

CO- N dipolar coupling of
15

CO- N dipolar coupling of

ds and fc + fs = 1. SIMPSON simulations were then performed for a range of dc, ds , and
2
fc. The χ (dc , ds , fc ) was calculated via

[(

χ 2 (dc , ds , fc ) = ∑

∆S lab
∆S c
∆S s
) (τ ) − fc x (
) (dc ,τ ) − (1 − fc )x (
) (ds ,τ )]2
So
So
So

σ (τ )2

τ

(2.11)

2
The best-fit simulation yielded dc = 65.4 Hz, ds = 21.2 Hz, fc = 0.41, and χ (dc , ds , fc ) =
2
5.7. The χ = 5.7 was close to the number of degrees of fitting ν = 4. The good fitting of

this model can be also visualized in Figure 2.7. One measure of the uncertainty of the
2
best-fit value is the deviation which causes an increase in χ by 1. For instance, the
2
uncertainty of dc was the deviation from 65.4 Hz that led to χ = 6.7 while keeping ds =

21.2 Hz and fc = 0.41. Application of this method yielded fc = 0.41 ± 0.01, dc = 65.4 ±
1.7 Hz with corresponding

13

15

CO- N distance rc = 3.61 ± 0.03 Å, and ds = 21.2 ± 0.5

Hz with corresponding rs = 5.25 ± 0.04 Å. These results indicate that the membraneassociated H3_20 peptide adopts ~40% closed structure and ~60% semi-closed
structure at pH 5.1[17]. The two structures are displayed in Figure 2.8. To check the
reproducibility of the results, a second sample was prepared in the same way and the
new REDOR data were acquired under the same experimental conditions. SIMPSON
2
simulations yielded the best-fit χ = 0.83, dc = 63.2 Hz with rc = 3.65 Å, ds = 19.0 Hz

with rs = 5.44 Å, and fc = 0.38. These values were comparable to those for the first
sample.
71

∆S lab
∆S sim
) vs τ (black) and best-fit (
)
vs τ for 32 Hz (purple)
So
So
which corresponds to the "single closed structure" model, 51 Hz x 0.64 (blue) which
corresponds to the "closed/open" model, and 65 Hz x 0.41 + 21 Hz x 0.59 (red) which
corresponds to the "closed/semi-closed" model, respectively.
Figure 2.7 Plots of (

72

Figure 2.8 Backbone structural models of H3_20 in lipid membranes at pH 5.1. The
carbon, nitrogen, and oxygen atoms are in green, blue, and red, respectively. The G16
13
15
CO-F9 N internuclear distance is 3.6 Å in the closed structure (a) and 5.2 Å in the
semi-closed structure (b).
Intramolecular vs intermolecular dipolar interactions
In the above data analysis, the

13

15

CO- N dipolar coupling was assigned to be an

intramolecular interaction. This assignment is supported by several pieces of evidence.
First, the best-fit results of the REDOR data of H3_20 indicate that ~40% of the peptide
molecules have a G16

13

CO-F9

15

N distance of 3.6 Å which is very close to the

intramolecular distance (3.9 Å) observed in the closed structure of H1_23 in
detergents[3]. Second, an earlier solid-state NMR study has supported an Nhelix/turn/C-helix structure of H3_20 in membranes[8]. If the experimentally observed
large dephasing of H3_20 were due to intermolecular dipolar interactions, large
aggregates of H3_20 would probably be required. However, another solid-state NMR
study of H3_20 indicates that the peptide molecules have significant motion at the

73

ambient temperature which is not consistent with a large aggregate of the peptide[18].
o

As we freeze the sample in liquid N2 and then keep cooling it with -50 C N2 gas
throughout the REDOR experiments, we do not expect the formation of large
aggregates of H3_20 during the experiments. Third, although the formation of
antiparallel homodimers of H3_20 could explain the significant dephasing of H3_20 in
membranes, this model is probably incorrect since there is strong evidence for a turn in
the middle of the H3_20 sequence, as discussed earlier. Furthermore, previous solution
NMR studies of H1_23 in detergents have ruled out the antiparallel homodimeric
arrangements of the peptide molecules since the experimentally observed interhelical
nuclear Overhauser effects (NOEs) between the N-terminal and C-terminal residues
have been proven to be intramolecular rather than intermolecular[3]. Fourth, earlier EPR
studies have been done on a construct consisting of residues 1-127 of the HA2 subunit
of the influenza hemagglutinin protein[19]. This construct is trimeric and induces vesicle
fusion under acidic conditions similar to the full length HA. EPR spectra of the spinlabeled construct showed that the fusion peptide region is likely to exist as a monomer
in the membrane at both neutral and acidic pH conditions.
To further confirm whether the

13

15

CO- N dipolar coupling is intra- or

intermolecular, a control REDOR experiment was run with a sample containing 50%
labeled and 50% unlabeled H3_20 in the membrane at pH 5.0. If the

interaction is predominantly intramolecular, the dephasing buildup of

13

15

CO- N dipolar

∆S
vs τ for the
So

100% labeled H3_20 sample will be comparable to that for the 50% labeled H3_20

74

sample. By contrast, if the

13

15

CO- N dipolar interaction is predominantly intermolecular,

the dephasing buildup rate and extent for the 100% labeled sample will be different than
those for the 50% labeled sample. For instance, as we consider a dimer model in which
two peptide molecules aggregate as a dimer and the G16
coupled to the F9

15

13

CO in molecule is dipolar

N in the other molecule, there will be three different types of dimer

consisting of (i) two labeled molecules; (ii) one labeled molecule and one unlabeled
molecule; and (iii) two unlabeled molecules, respectively (Figure 2.9a). The probability
of (i), (ii), and (iii) is 25%, 50%, and 25%, respectively. Suppose there are 25 dimers of
(i), 50 dimers of (ii), and 25 dimers of (iii), the fractional contribution to the overall G16
13

CO signal intensity will be 0.50 from (i) and 0.50 from (ii). If the

interaction is predominantly intermolecular, then ~1/2 of the G16

13

13

15

CO- N dipolar

CO signal from (i)

will be dephased in the REDOR experiment, i.e. ~1/4 of the overall G16

13

CO signal will

be dephased, as opposed to the 100% labeled sample for which ~1/2 of the overall G16
13

CO signal is dephased. In this case, the (

∆S lab
) for the 50% labeled sample will be
So

approximately half of that for the 100% labeled sample for any given dephasing time τ,
where (

∆S lab
) is the dephasing for the labeled G16 13CO signal only (Figure 2.9b).
So

However, the REDOR data show that the 50% and 100% labeled samples had very

75

similar buildups of (

∆S lab
) vs τ (Figure 2.10), supporting that the 13CO-15N dipolar
So

coupling is predominantly intra- rather than intermolecular.
Conclusions
In summary, our REDOR studies show that H3_20 adopts majorly closed rather
than open structures in the membrane. It is important to distinguish between the open
and closed structures since they lead to different peptide-lipid interaction models. For
the open "boomerang" structure, the hydrophobic side chains are majorly in the interior
of the structure, creating a hydrophobic pocket that allows a deep insertion of the fusion
peptide into the membrane hydrocarbon core[2]. However, for the closed structure, the
hydrophobic side chains are majorly on one side which interacts with the lipid
membrane while the Gly residues are on the other side which is exposed to the
solvent[3].

76

Figure 2.9 (a) Possible aggregates of H3_20 for the dimer model for the "50% labeled +
13
15
50% unlabeled" sample. The G16 CO and F9 N labeled H3_20 is in red while the
13

15

unlabeled H3_20 is in black. (b) Theoretical
CO- N REDOR dephasing plots of
∆S lab
(
) vs τ for the 100% labeled H3_20 sample (red) and the "50% labeled + 50%
So

∆S lab
) for the 100% labeled sample is twice
So
that for the 50% labeled sample for any given dephasing time τ.
unlabeled" H3_20 sample (black). The (

77

∆S lab
)
vs τ after n.a. correction for the
So
membrane sample containing 100% labeled H3_20 (red) and the one containing 50%
labeled and 50% unlabeled H3_20 (black) at pH 5.0 (courtesy from Ujjayini Ghosh, the
REDOR data for the two samples were acquired under the same optimized
experimental conditions which were different than those for Figure 2.6).
Figure 2.10

13

15
CO- N REDOR plots of (

78

REFERENCES

79

REFERENCES

1.

Gething, M.J., et al., STUDIES ON THE MECHANISM OF MEMBRANE-FUSION
- SITE-SPECIFIC MUTAGENESIS OF THE HEMAGGLUTININ OF INFLUENZAVIRUS. Journal of Cell Biology, 1986. 102(1): p. 11-23.

2.

Han, X., et al., Membrane structure and fusion-triggering conformational change
of the fusion domain from influenza hemagglutinin. Nature Structural Biology,
2001. 8(8): p. 715-720.

3.

Lorieau, J.L., J.M. Louis, and A. Bax, The complete influenza hemagglutinin
fusion domain adopts a tight helical hairpin arrangement at the lipid:water
interface. Proceedings of the National Academy of Sciences of the United States
of America, 2010. 107(25): p. 11341-11346.

4.

Lorieau, J.L., J.M. Louis, and A. Bax, Impact of Influenza Hemagglutinin Fusion
Peptide Length and Viral Subtype on Its Structure and Dynamics. Biopolymers,
2013. 99(3): p. 189-195.

5.

Lorieau, J.L., et al., pH-triggered, activated-state conformations of the influenza
hemagglutinin fusion peptide revealed by NMR. Proceedings of the National
Academy of Sciences of the United States of America, 2012. 109(49): p. 1999419999.

6.

Steinhauer, D.A., et al., STUDIES OF THE MEMBRANE-FUSION ACTIVITIES
OF FUSION PEPTIDE MUTANTS OF INFLUENZA-VIRUS HEMAGGLUTININ.
Journal of Virology, 1995. 69(11): p. 6643-6651.

7.

Wasniewski, C.M., et al., Solid-state nuclear magnetic resonance studies of HIV
and influenza fusion peptide orientations in membrane bilayers using stacked
glass plate samples. Chemistry and Physics of Lipids, 2004. 132(1): p. 89-100.

8.

Sun, Y. and D.P. Weliky, C-13-C-13 Correlation Spectroscopy of MembraneAssociated Influenza Virus Fusion Peptide Strongly Supports a Helix-Turn-Helix
Motif and Two Turn Conformations. Journal of the American Chemical Society,
2009. 131(37): p. 13228-13229.

80

9.

Korte, T., et al., Role of the Glu residues of the influenza hemagglutinin fusion
peptide in the pH dependence of fusion activity. Virology, 2001. 289(2): p. 353361.

10.

Han, X. and L.K. Tamm, A host-guest system to study structure-function
relationships of membrane fusion peptides. Proceedings of the National
Academy of Sciences of the United States of America, 2000. 97(24): p. 1309713102.

11.

Wharton, S.A., et al., MEMBRANE-FUSION BY PEPTIDE ANALOGS OF
INFLUENZA-VIRUS HEMAGGLUTININ. Journal of General Virology, 1988. 69: p.
1847-1857.

12.

Chan, W.C. and P.D. White, Fmoc Solid Phase Peptide Synthesis: A Practical
Approach. 2000: Oxford University Press.

13.

Worman, H.J., et al., RELATIONSHIP BETWEEN LIPID FLUIDITY AND WATER
PERMEABILITY OF BOVINE TRACHEAL EPITHELIAL-CELL APICAL
MEMBRANES. Biochemistry, 1986. 25(7): p. 1549-1555.

14.

Long, H.W. and R. Tycko, Biopolymer conformational distributions from solidstate NMR: alpha-helix and 3(10)-helix contents of a helical peptide. Journal of
the American Chemical Society, 1998. 120(28): p. 7039-7048.

15.

Zheng, Z., et al., Conformational flexibility and strand arrangements of the
membrane-associated HIV fusion peptide trimer probed by solid-state NMR
spectroscopy. Biochemistry, 2006. 45(43): p. 12960-12975.

16.

Zhang, H.Y., S. Neal, and D.S. Wishart, RefDB: A database of uniformly
referenced protein chemical shifts. Journal of Biomolecular Nmr, 2003. 25(3): p.
173-195.

17.

Ghosh, U., L. Xie, and D.P. Weliky, Detection of closed influenza virus
hemagglutinin fusion peptide structures in membranes by backbone (CO)-C-13N-15 rotational-echo double-resonance solid-state NMR. Journal of Biomolecular
NMR, 2013. 55(2): p. 139-146.

18.

Bodner, M.L., et al., Temperature dependence and resonance assignment of C13 NMR spectra of selectively and uniformly labeled fusion peptides associated
with membranes. Magnetic Resonance in Chemistry, 2004. 42(2): p. 187-194.

81

19.

Macosko, J.C., C.H. Kim, and Y.K. Shin, The membrane topology of the fusion
peptide region of influenza hemagglutinin determined by spin-labeling EPR.
Journal of Molecular Biology, 1997. 267(5): p. 1139-1148.

82

Chapter 3 - Development of

13

2

C- H REDOR

3.1 Probe design
There are different types of REDOR experiments such as
31

13

15

C- N,

13

2

C- H, and

19

P- F REDOR. As noted earlier, REDOR is a very useful SSNMR technique for

measuring distances between two heteronuclei and was first developed by Gullion and
Schaefer[1]. In the past, our group has been using

13

15

C- N,

13

31

C- P, and

13

19

C- F

REDOR to study the structure, membrane topology, and insertion depth of viral fusion
peptides in the membrane[2-4]. To the date I joined the group, there was no probe for
13

2

C- H REDOR. The goal of developing

the membrane locations of peptides using

13

2

C- H REDOR in our laboratory is to study

13

2

CO-labeled peptide and H-labeled lipid or
2

cholesterol. One significant advantage of the H labeling scheme is that there will be no
perturbation of the membrane bilayer no matter what fraction of lipids and/or cholesterol
1

2

in the membrane are deuterated since H and H are chemically equivalent.
Low-power and high-power tuning
As we build a new multi-channel probe, the first step is to get all the desired
channels tunable, e.g., for a 400 MHz NMR spectrometer, the

13

C channel should be

tunable at ~100 MHz. There are two types of tuning with one called low-power tuning
and the other called high-power tuning. Low-power tuning is used to check the tunable
frequency range of a specific channel without applying r.f. pulses to the probe whereas
high-power tuning refers to the tuning when applying r.f. pulses to the probe. The cable
83

connections for low-power tuning are displayed in Figure 3.1. Besides, the "SEC/DIV"
knob on the oscilloscope needs to be adjusted to change the time scale to "CH1 X"
such that the resonance peaks at different frequencies can be displayed on the
oscilloscope.

Figure 3.1 Cable connections for low-power tuning. Only part of the sweeper and
oscilloscope are displayed. The probe is connected to the sweeper via a hybrid coupler
(also known as reflection bridge or magic T). For the low-power tuning, the oscilloscope
needs to be adjusted such that all the red circled lights are on.
Tuning configuration
For each probe, there are specific tuning configurations for different NMR
experiments, e.g. for the probe (probe # 7058 from Chemagnetics) we used to develop
13

2

1

C- H REDOR, the triple resonance tuning configuration for H,

by the manufacturer is as follows:
84

13

2

C, and H provided

33 pf series plug-in
15 pf, 6 t trap plug-in
SC Low channel receiver platform (Y channel)
36 pf Mid channel receiver platform (X channel)
2.2" Low tune tube (Y channel)
5.4" Mid tune tube (X channel)
where pf represents picofarad and is a unit of capacitance, 6 t refers to 6 turns in the
trap plug-in, and SC represents short circuit. For a tune tube, it consists of two parts
with part (i) made of copper and part (ii) made of both copper and dielectric plastic. In
the above tuning configuration, 2.2" and 5.4" refer to the length of the copper region in
part (ii) of the tune tube. Figure 3.2 displays the pictures of probe, tune tube, tuning rod,
series plug-in, and trap plug-in.

85

Figure 3.2 (a) Triple-resonance solid state NMR probe. (b) Tune tube and tuning rod. (c)
Series plug-in and trap plug-in.
Since there are usually several capacitors and inductors in a probe, each channel
in the probe can be considered as an LC circuit. As a r.f. pulse passes through an LC
circuit, the capacitive reactance XC, inductive reactance XL, and total impedance Z the
pulse experiences are defined by

XC = -(

1
2πν C

)i

(3.1)

X L = (2πν L )i

(3.2)

86

=
Z XC + X L = (2πν L -

1
2πν C

)i

(3.3)

where ν is the frequency of the pulse, C is the capacitance of the capacitor, L is the
inductance of the inductor, and i refers to the square root of -1. According to equations
3.1-3.3, we can mathematically show that the total impedance Z = 0 when ν =

where

1
2π LC

1
2π LC

,

is the so-called resonance frequency of an LC circuit. To minimize the

total impedance Z that the pulse experiences, specific components such as plug-ins and
tune tube need to be selected such that the corresponding capacitance C and
inductance L will fulfill the condition of ν =

1
2π LC

, where ν is the transmitter frequency

of the pulse. Although the tuning configuration for each specific NMR experiment is
provided by the probe manufacturer, the final configuration that works may be different
1

than the one provided. For instance, the tuning configuration for H,
really worked as we developed the

13

2

13

C- H REDOR probe was as follows:

82 pf series plug-in
15 pf, 6 t trap plug-in
33 pf Low channel receiver platform (Y channel)
36 pf Mid channel receiver platform (X channel)
3.9" Low tune tube (Y channel)
5.4" Mid tune tube (X channel)

87

2

C, and H that

which was substantially different than the one provided by the probe manufacturer. Note
that the tuning configuration may change again as you modify the probe, e.g.
replacement of the copper ribbon connecting the stator and

13

C channel (X channel)

may cause detuning of the channel and thus a new series plug-in may be needed to get
the channel tunable again.
During the low-power tuning, we adjust the tuning rod and match to make the
inductance L and capacitance C fulfill the resonance condition of ν =

1
2π LC

to

minimize the reflected voltage Vr from the probe at a specific frequency ν and detect Vr
on the oscilloscope. If a channel can be tuned at a specific frequency (e.g. the
transmitter frequency to be set on the spectrometer) to produce a deep and sharp peak
near the zero voltage line on the oscilloscope at the low-power tuning, it means the
channel can be also well tuned for that specific frequency at the high-power tuning, e.g.
the forward-to-reverse voltage ratio of the pulse after the high-power tuning can be 10
or higher. Note that the forward voltage of the pulse can be accurately measured only
after the channel has been well tuned.
3.2 Setup peptide I4
Criteria for REDOR setup compound
To optimize the REDOR pulse program, a proper setup compound is required.
There are several criteria for an ideal REDOR setup compound described as follows:
(a) The setup compound should have the same type of observed nucleus as the
samples to be studied. Since the peptide samples to be studied by
88

13

2

C- H REDOR are

13

CO labeled, the setup compound should be also

compound is

13

13

13

13

13

CO labeled samples since

C chemical shift anisotropies (CSA). The amide

~150 ppm whereas
pulse power for

CO labeled. If the setup

CH3 labeled, the optimized pulse parameters for the setup compound

are not necessarily optimized for the
have different

13

13

CH3 and

CH3 has a CSA of ~30 ppm. As a result, an optimized
13

CO

CO has a CSA of

13

CH3 is not necessarily sufficient for

13

13

C r.f.

CO.

(b) The setup compound should have a single spin pair with a known internuclear
distance, in which case the data analysis as we optimize the pulses will be simplified.
We can first calculate the dipolar coupling using the internuclear distance and then run
SIMPSON simulations to get the theoretical dephasing buildup plot for that dipolar
coupling. At last, we compare the theoretical dephasing buildup with the experimental
one to tell whether the pulses (particularly the π pulses during the dephasing time) have
been optimized or not.
(c) The dipolar interaction for the setup compound should be predominantly intra- rather
than intermolecular. In the latter case, it could be challenging to calculate the
intermolecular distance and there could be a distribution of intermolecular distances,
which complicate the data analysis as we optimize the pulse program.
(d) It is advisable to choose a setup compound which has an intermediate dipolar
coupling, e.g., 30 or 40 Hz. If we choose one with a very small dipolar coupling such as
6 Hz, the SIMPSON-simulated

13

2

C- H REDOR dephasing is only 10% for dephasing

time τ = 40 ms. By contrast, if we choose one with a very large dipolar coupling such as
89

200 Hz, the SIMPSON-simulated

13

2

C- H REDOR dephasing is 65% for dephasing time

τ = 16 ms, which is already reaching the maximum dephasing of 2/3. In this case, the
dephasing buildup may not be sensitive to the pulse optimization, i.e., even if the pulses
(particularly the π pulses during the dephasing time) are not optimized, there might be a
significant dephasing buildup.
I4 peptide sequence and isotopic labels
Based on the above criteria, we synthesized an I4 peptide as the setup
compound for

13

2

C- H REDOR. The I4 peptide was synthesized by Fmoc SPPS using

rink amide resin and thus the C-terminus of the peptide was amidated. I4 has a
sequence of AEAAAKEAAAKEAAAKAW with a

13

2

CO label at residue A9 and a Cα- H

label at residue A8. Earlier SSNMR studies of the lyophilized I4 peptide have shown
that A9 has an α-helicity of (83 ± 6)%[5]. In the α-helical structure, the internuclear
2

distance between A8 Cα- H and A9
dipolar coupling of ~37 Hz. The

13

13

CO is ~5.0 Å which corresponds to a

13

2

C- H

2

C- H REDOR pulse program optimization using the

setup peptide I4 is described in detail in Section 3.3. For clarification purposes, the
setup peptide I4 is denoted as I4_A8DA9C in the following sections and chapters.
3.3 Pulse sequence optimization
In general, there are three commonly used REDOR pulse sequences (Figure
1

3.3)[6]. For each REDOR pulse sequence in Figure 3.3, it starts with CP from H to
nucleus S to enhance the S signal, then a dephasing period during which the dipolar

90

coupling is decoupled in the So experiment and recoupled in the S1 experiment, and at
1

last an acquisition period. Windowless H decoupling pulses (i.e. there is no interpulse
time delay) are applied throughout the dephasing and acquisition periods.
For the pulse sequence in Figure 3.3a, a single S π pulse is applied in the middle
of the dephasing period to refocus the isotropic chemical shift and produce an echo at

Figure 3.3 Three commonly used REDOR pulse sequences. Each sequence starts with
1
CP from H to the observed nucleus S to enhance the S signal, followed by a
dephasing period (only ten rotor cycles are shown here) and at last an acquisition
1
period. H decoupling pulses are applied throughout the dephasing and acquisition
periods[6].

91

the beginning of the acquisition period. For the So experiment, there is no π pulse
applied on the I channel and thus the dipolar coupling is decoupled by MAS. For the S1
experiment, in addition to the single π pulse applied on the S channel in the middle of
the dephasing period, there are also π pulses applied on the I channel in the middle and
at the end of each rotor cycle during the dephasing period except for the end of the rotor
cycle when the S-channel π pulse is applied. Application of the π pulse train on the I
channel recouples the dipolar coupling and thus causes dipolar dephasing of the signal
during the S1 experiment.
For the pulse sequence in Figure 3.3b, π pulses are applied on the S channel at
the end of each rotor cycle during the dephasing period. The I-channel π pulses in the
middle of each rotor cycle are applied during the S1 experiment but not during the So
experiment.
For the pulse sequence in Figure 3.3c, π pulses are applied on the S channel in
the middle and at the end of each rotor cycle except for the midpoint of the dephasing
period. The I-channel π pulse in the midpoint of the dephasing period is applied during
the S1 experiment but not during the So experiment. This pulse sequence can be used
when the I spin has a very large anisotropic interaction (e.g. quadrupolar interaction).
The I-channel π pulse can be a composite π pulse which may better compensate for the
limited r.f. pulse power and large nuclear resonance offsets than a simple π pulse do,
where a composite π pulse consists of multiple simple pulses with different phases and
o

the net flip angle of the composite pulse is 180 . This pulse sequence has been used
92

for

13

2

2

C- H REDOR and the advantage of applying a single rather than multiple H
2

π pulses is that it can minimize the effect of H pulse imperfections[7].
All the three REDOR pulse sequences in Figure 3.3 have the same optimization
procedure described as follows (take

13

2

C- H REDOR for example):

(a) MAS setup using KBr
pulse program = "1 pulse" (one pulse with a phase cycling of x, -x, y, -y), Br transmitter
o

frequency = 99.8942000 MHz, spinning rate = 4 kHz, temperature = -50 C.
We need to first tune the Br channel and then adjust the magic angle rod. As we adjust
the magic angle rod, we can either click "Repeat Scan" in the Spinsight window and
then keep adjusting the magic angle rod until we observe as many and strong rotary
echoes as possible in the Br FID or click "RS and Process" and then keep adjusting the
magic angle rod until we observe as many and strong peaks as possible in the Br
spectrum (Figure 3.4).

93

o

Figure 3.4 Br spectrum (64 scans) after MAS setup, spinning rate = 4 kHz, -50 C. The
spectrum was processed with 20 Hz Gaussian line broadening.
(b)

13

C chemical shift referencing using adamantane

pulse program = "1 pda" (one pulse with a phase cycling of x, -x, y, -y and
decoupling),

13

1

H

1

C transmitter frequency = 100.2677200 MHz, H transmitter frequency =

398.6976190 MHz, spinning rate = 4 kHz (other spinning rates such as 8 and 10 kHz
also work and they produce very similar
o

temperature = -50 C. The
13

C peaks of interest (e.g.

13

13

C chemical shifts within 0.1 ppm deviation),

C transmitter frequency is chosen in such a way that the

13

CO) of the samples to be studied (not adamantane) are
94

close to 0 ppm (e.g. 15 ppm) before doing the chemical shift referencing, in which case
the
13

13

C resonance offsets are small and thus the

13

C pulses can work efficiently. The

C spectrum of adamantane under 4 kHz MAS is shown in Figure 3.5. The peak at -

119.0 ppm refers to the methylene
methine

13

13

C peak and the peak at -127.9 ppm refers to the

C peak. For all the membrane samples of peptides to be studied, the

13

C

shifts will be externally referenced to the methylene peak of adamantane at 40.5 ppm
such that they can be directly compared with the
solution[8, 9]. Since the methylene
the corresponding

13

13

C shifts of proteins in aqueous

C shift of adamantane in Figue 3.5 is -119.0 ppm,

13

C chemical shift referencing value = 40.5 ppm - (-119.0 ppm) =

159.5 ppm. In this case, if the peptide

13

CO peak shift is 20.0 ppm before referencing,

the peak shift after referencing will be 179.5 ppm.

95

Figure 3.5

13

o

C spectrum of adamantane under 4 kHz MAS, -50 C. The left peak at -

119.0 ppm corresponds to the methylene
corresponds to the methine
To correctly view the

13

13

13

C whereas the right peak at -127.9 ppm

C.
C shifts before referencing, we need to make sure 0 ppm

is right in the middle of the chemical shift axis in the Spinsight window, if not, click
"Analysis"→"Reference", then click the spectrum to pop up the "Set Reference" window
in which you set the "Reference point" as 0.5 and "Reference value" as 0, and at last
click "Set"→ "OK" to finish. To correctly view the

13

C shifts after referencing, click

"Analysis"→"Reference" in the Spinsight window, then click the spectrum to pop up the
"Set Reference" window where you set the "Reference point" as 0.5 and "Reference
96

value" as the referencing value of adamantane (e.g. 159.5), and finally click "Set"→
"OK" to finish.
(c) CP optimization
sample = I4_A8DA9C, pulse program = "cp_ramp",

13

C transmitter frequency =

1

100.2677200 MHz, H transmitter frequency = 398.6976190 MHz, spinning rate = 10
o

kHz, temperature = -50 C.
1

step 1: H π/2 pulse optimization
1

1

H π/2 pulse is the first pulse in the REDOR sequence and it rotates the H
1

magnetization from z axis to the xy plane. We first set the H π/2 pulse width (pw90H)
as 10 µs and then array the pulse power aH from 0.50 to 0.70 in an increment of 0.02.
Comparison of the

13

C spectra (Figure 3.6) indicates that pw90H of 10 µs and aH of

0.58 correspond to a

1

H π pulse. The

1

H π pulse produces a zero transverse

1

1

magnetization of H and thus there is no magnetization transfer from H to
CP, which results in a zero

13

13

C during

1

CO signal in the spectrum. To set the H π/2 pulse, we fix
1

aH at 0.58 and change pw90H from 10 µs to 5 µs. The Rabi frequency of the H π/2

pulse =

γ B1
1
=
= 50 kHz (the pulse flip angle = γB1 x (pw90H) = π/2, refer
4x(pw90H)
2π
1

to Chapter 1 for more details). An alternative way to set the H π/2 pulse is to fix the
pulse power aH and array the pulse width pw90H.
97

Figure 3.6

13

C spectra of I4_A8DA9C for "aH" array from 0.50 to 0.70 in an increment
o

of 0.02, 10 kHz MAS, -50 C.
1

step 2: H and

13

C CP pulse optimization
1

1

After optimizing the H π/2 pulse, we then set the H CP pulse power (aHcp) the
same as aH. Although aHcp can be different than aH, we usually set them the same
1

such that we can easily measure the forward voltage of the H π/2 pulse. If we set them
1

differently, it will be difficult to distinguish the H π/2 pulse from the CP pulse due to the
short pulse width (5 µs) compared to the much longer CP contact time (1~2 ms) and
thus difficult to measure the forward voltage of the π/2 pulse. The forward voltage of the
98

1

1

H π/2 pulse is important since we need it to set the H decoupling power, which will be

discussed later in more details.
As we set aHcp = aH = 0.58, the next step is to array the
(aXcp) to find out what power produces the maximum
aXcpmod is arrayed to find out the best
1

magnetization transfer from H to

13

13

13

13

C CP pulse power

CO signal intensity. After that,

C CP ramp which can maximize the

C (refer to Chapter 1 for more details about CP

ramp). The optimized aXcp and aXcpmod are 0.40 and 0.03, respectively.
step 3: CP contact time optimization
The last step of CP optimization is to array the contact time to maximize the

13

CO

1

signal intensity. If the CP contact time is too short, the magnetization transfer from H to
13

C will be incomplete which results in weaker

time is too long, the

13

13

CO signal intensities. If the CP contact

CO signal intensities will also decrease due to the

1

H T1ρ

1

relaxation which refers to the H magnetization decay from Mo to (B1/Bo)Mo, where Mo
1

is the H magnetization magnitude established under the external magnetic field Bo and
1

1

(B1/Bo)Mo is the H magnetization magnitude at the thermal equilibrium under the H
CP pulse field B1. Comparison of

13

CO signal intensities indicates that the optimized

CP contact time is 1.5 ms.

99

In the

13

2

1

C- H REDOR pulse sequence, H decoupling pulses are applied right
1

after CP to decouple all H-related couplings such as
1

13

1

C- H J-coupling,

1

13

1

C- H dipolar

1

coupling, and H- H dipolar coupling, etc. We can set the H decoupling power (aHdec)
1

1

based on the forward voltage of the H π/2 pulse. For example, if the H decoupling
power is set as 75 kHz, we need to adjust the parameter aHdec until the forward
1

1

voltage of the H decoupling pulses is 1.5 times that of the H π/2 pulse which has a
Rabi frequency of 50 kHz. For the forward voltage of a pulse, recall that it can be
correctly measured only after the channel has been well tuned. Figure 3.7 shows the
13

1

C spectrum of I4_A8DA9C after optimizing CP and setting H decoupling power as 75

kHz. The A9
(d)

13

13

CO shift of 178.8 ppm is consistent with the α-helical structure of I4[9].

C π pulse optimization

sample = I4_A8DA9C, pulse program = "cp_zfilter",

13

C transmitter frequency =

1

100.2677200 MHz, H transmitter frequency = 398.6976190 MHz, spinning rate = 10
o

kHz, temperature = -50 C. The pulse sequence of cp_zfilter is shown in Figure 3.8.
When both the

13

C π/2 and π pulses are set correctly, a zero

observed. However, when the

13

13

C signal will be

C π pulse is set as a π/2 pulse, a maximum

100

13

C signal

Figure 3.7
o

C. The A9

13

C spectrum of I4_A8DA9C after CP optimization under 10 kHz MAS, -50

13

CO chemical shift after referencing is 178.8 ppm.

will be observed. To optimize the

13

C pulses, we first set the

13

C π/2 pulse width

(pw90X) and π pulse width (pw180X) both as 4.0 µs and then array the power (aX) from
0.30 to 0.45 in an increment of 0.01. The maximum

13

CO signal intensity is observed

for aX = 0.39, i.e. the pulse width 4.0 µs and power 0.39 correspond to a

13

C π/2 pulse.

The next step is to fix pw90X = 4.0 µs and aX = 0.39 and then arry pw180X. A zero
13

CO signal intensity is observed for pw180X = 8.3 µs. Therefore, the pulse width 8.3

µs and power 0.39 correspond to a

13

C π pulse.

101

1

Figure 3.8 Pulse sequence of cp_zfilter. The phase cycle is "x, -x" for H π/2 pulse, "-y,
-y, -x, -x" for

13

C CP pulse, "x, x, -y, -y" for

13

C π/2 pulse, and "-x, -x, y, y" for

13

C

1

π pulse, respectively. The H CP and decoupling pulses have a fixed phase of "y".
2

(e) H π pulse optimization
1

sample = I4_A8DA9C, pulse program = "redorxy8xypi_pm" (refer to Figure 3.3b), H
transmitter frequency = 398.6976190 MHz,

13

C transmitter frequency = 100.2677200

2

MHz, H transmitter frequency = 61.2030000 MHz, dephasing time τ = 32 ms (other
o

2

values such as 24 ms also work), spinning rate = 10 kHz, temperature = -50 C. The H
2

transmitter frequency is set as 61.2030000 MHz such that the lipid H quadrupolar
2

spectra are centered at ~0 ppm. To optimize the H π pulses, we first set the pulse
power (aY180) as 0.90 and then array the pulse width (pw180Y) to find out which value
produces the maximum dephasing. For τ = 32 ms, the maximum dephasing, 0.464 (6),
was observed when pw180Y = 4.7 µs. Therefore, pw180Y and aY180 are set as 4.7 µs
and 0.90, respectively. To check whether pw180Y = 4.7 µs and aY180 = 0.90
2

2

correspond to a H π pulse or not, H "1pulse" experiment of D2O at room temperature
(18.9

o

C) without spinning is carried out to calibrate the
102

2

H pulses. For direct

2

comparison, the H amplifier for "1pulse" experiment of D2O is the same as that for
13

2

2

C- H REDOR. As we set pulse power as 0.90, the maximum H signal of D2O is

observed for pulse width of 2.5 µs, i.e. the pulse power of 0.90 and pulse width of 2.5 µs
2

correspond to a H π/2 pulse. Therefore, pw180Y = 4.7 µs and aY180 = 0.90 do not
2

o

o

exactly correspond to a H 180 pulse but a 169 pulse. This deviation could be due to
2

o

the temperature change since the H pulse is calibrated using D2O at 18.9 C whereas
the

13

2

o

C- H REDOR pulses are optimized at -50 C.
At last, we set pw180Y = 4.7 µs and aY180 = 0.90 and then run

13

2

C- H REDOR

experiments of I4_A8DA9C for dephasing time τ = 2, 8, 16, 24, 32, 40, 48, 56, 64, 72,
and 80 ms, respectively. The experimental dephasing buildup of ∆S/So vs τ is shown in
Figure 3.9.

103

Figure 3.9

13

2

C- H REDOR experimental dephasing plot of ∆S/So vs τ of I4_A8DA9C
o

under 10 kHz MAS and -50 C.
Comparison of REDOR pulse sequences
As discussed earlier, there are three commonly used REDOR pulse sequences
(Figure 3.3). For the

13

2

C- H REDOR experiments of I4_A8DA9C described above, the

pulse sequence is "redorxy8xypi_pm" (Figure 3.3b), where "xy8" refers to the X/Y (i.e.
13

2

1

C/ H) π pulse phase cycle of "x, y, x, y, y, x, y, x" and "pm" refers to TPPM H

decoupling. The π pulse phase cycling is applied to compensate for pulse imperfections
and produce REDOR dephasings which are not sensitive to the resonance offsets[10].
In addition to "redorxy8xypi_pm", we also tested the "redorxy8ypi_pm" (Figure 3.3a) and

104

"redorxy4xpi_pm"

(Figure

3.3c)

pulse

sequences

and

compared

them

to

"redorxy8xypi_pm" in terms of signal sensitivity and REDOR dephasing buildup. Figure
3.10 shows the REDOR dephasing buildups of I4_A8DA9C for "redorxy8xypi_pm" (black)
and "redorxy8ypi_pm" (red). Compared to "redorxy8xypi_pm", "redorxy8ypi_pm"
2

produces a slower dephasing buildup. The reason for that could be due to the H pulse
2

imperfections including limited pulse power relative to the large resonance offsets of H.
2

For a given dephasing time τ, the number of H π pulses for "redorxy8ypi_pm" is twice
2

that for "redorxy8xypi_pm". Therefore, the accumulated H π pulse imperfections are
2

more significant for "redorxy8ypi_pm" which cause a less efficient inversion of H spins
and consequently a slower dephasing buildup. Figure 3.11 shows the REDOR So
13

spectra ( CO region) of I4_A8DA9C for "redorxy8xypi_pm" and "redorxy4xpi_pm". The
13

C

signal

sensitivity

for

"redorxy4xpi_pm"

is

much

lower

than

that

for

"redorxy8xypi_pm". One possible reason for it could be that the anisotropic chemical
shift interactions are not averaged out by MAS due to the application of

13

C π pulses in

the middle and at the end of each rotor period Tr. As shown in Figure 3.12, assuming
the

13

C magnetization is initially along the x axis for dephasing time τ = 0 (i), the two

individual

13

C magnetic dipole moments a and b will precess under anisotropic

chemical shift fields from τ = 0 to τ = Tr/2 (ii). When a

105

13

C πx pulse is applied at τ = Tr/2,

the magnetic moments a and b will be flipped to the opposite side about the x axis (iii).
In the meantime, the anisotropic chemical shift field directions will be inverted at τ = Tr/2

13

2

Figure 3.10 C- H REDOR experimental dephasing plots of ∆S/So vs τ of I4_A8DA9C
for the "redorxy8xypi_pm" (black) and "redorxy8ypi_pm" (red) pulse sequences under
o
10 kHz MAS and -50 C.

106

Figure 3.11

13

2

C- H REDOR So spectra of I4_A8DA9C for the "redorxy8xypi_pm" (a)
o
and "redorxy4xpi_pm" (b) pulse sequences under 10 kHz MAS and -50 C. The number
of acquisitions is 400 and 1000 for the 16 ms and 24 ms So spectra in panel a and 6875
and 13331 for the 16 ms and 24 ms So spectra in panel b. All spectra are processed
with 20 Hz Gaussian line broadening and baseline correction.
due to MAS. Therefore, the magnetic moments a and b will precess in the opposite
direction from τ = Tr/2 to τ = Tr as they do from τ = 0 to τ = Tr/2 (iii). As a result, the
magnetic moments a and b will not be refocused along the x axis at τ = Tr (iv). As

107

dephasing time τ goes on, the individual

13

C magnetic dipole moments will become

randomly oriented and thus their vector sum (i.e.

13

C magnetization) will decrease in

magnitude as a function of τ. This is probably why the

13

C signal sensitivity for

"redorxy4xpi_pm" is so poor compared to that for "redorxy8xypi_pm".

Figure 3.12 Evolution of two individual

13

C magnetic dipole moments a and b due to

anisotropic chemical shift interactions under MAS when applying

13

C π pulses in the

middle and at the end of each rotor period Tr.
In summary, the best pulse sequence for

13

2

C- H REDOR in terms of signal

sensitivity and dephasing buildup is "redorxy8xypi_pm" and this pulse sequence is used
for the peptide

13

2

CO to lipid H REDOR experiments to be discussed in chapter 4. For

2

2

the H π pulses, although composite pulses may better compensate for the large H
quadrupolar anisotropy than simple pulses do, the ~100 kHz fields of
pulses used in our
2

13

2

H simple π

2

C- H REDOR experiments are strong enough to produce efficient
2

H spin inversion even in the presence of non-motionally averaged H quadrupolar
108

anisotropy (see Figure 3.13 and more discussion in chapter 4). For the
2

"redorxy4xpi_pm" pulse sequence, it could be advantageous in that only a single H π
pulse is applied during the dephasing time and thus there is no accumulated pulse
imperfection. However, this pulse sequence may be beneficial for alkyl
2

to H REDOR but not for
~30 ppm for

13

13

2

CO to H REDOR since alkyl

CH3) whereas

13

13

13

C (e.g.

13

CH3)

C has a small CSA (e.g.

CO has a large CSA (~150 ppm). The anisotropic

chemical shift interactions that are averaged out under MAS will be reintroduced by the
13

C π pulses applied in the middle and at the end of each rotor period during the

dephasing time. As a result, the

13

CO signal sensitivity is very low due to the large CSA

(Figure 3.11).
3.4 I4 data fitting
exp

Natural abundance calibration of (∆S/So)

In section 3.3, we have discussed the

13

2

C- H REDOR pulse sequence

optimization using the setup peptide I4_A8DA9C. The experimental dephasing curve of
∆S/So vs τ is shown in Figure 3.9. To obtain ∆S/So vs τ for the labeled A9

13

CO only,

the natural abundance contribution to the experimental ∆S/So is removed in a similar
way as described in chapter 2. The labeled ∆S/So vs τ for A9

13

CO is shown in Figure

3.13 (black squares). For all τ, the difference between the experimental and labeled
∆S/So is ≤ 0.05.
109

lab

SIMPSON simulations of (∆S/So)

vs τ

In the α-helical structure of I4_A8DA9C, the internuclear distance r between A8
2

Cα- H and A9

13

CO is ~5.0 Å which corresponds to a

13

2

C- H dipolar coupling d ≈ 37

3

Hz via the relationship d(Hz)=4642/r (Å)[10]. The theoretical ∆S/So vs τ for d = 37 Hz is
2

2

simulated by SIMPSON for realistic MAS frequency, H r.f. field, and H quadrupolar
coupling, etc. As shown in Figure 3.13, the simulated ∆S/So buildup for d = 37 Hz has a
sigmoidal shape whereas the labeled ∆S/So buildup for A9

13

CO has an exponential

shape. In addition, the simulated ∆S/So has a rapid buildup with a maximum value of
2/3 whereas the labeled ∆S/So has a slower buildup with values greater than 2/3 at
longer τ. The maximum value of 2/3 for simulated ∆S/So reflects 1/3 fractional
2

2

population of H spins in the m=0 state which cannot be inverted by H π pulses and
thus have no contribution to the dipolar dephasing of

110

13

CO.

Figure 3.13

13

2

C- H REDOR dephasing curves of ∆S/So vs τ of I4_A8DA9C for the
o

pulse sequence "redorxy8xypi_pm" under 10 kHz MAS and -50 C. Black squares are
labeled ∆S/So for A9

13

CO after removal of the natural abundance contribution. Red

triangles are SIMPSON simulated ∆S/So for d = 37 Hz. Red line is the best-fit curve
using the fitting function Ax(1-e

-βτ

).

The differences between simulated and labeled ∆S/So buildups could be due to
2

the T1 relaxation between H spin states in the experiment which is not considered in
the SIMPSON simulation. In the absence of relaxation, a

13

experiences dipolar evolution for the full dephasing period τ whereas a
2

a m=0 H experiences no evolution. With relaxation, a

111

13

2

CO coupled to a m=±1 H
13

CO coupled to
2

CO coupled to a m=±1 H

experiences dipolar evolution for part of τ but no evolution for the remainder of τ due to
2

the relaxation that changes the H in the m=±1 state to the m=0 state. The relaxation
will result in a slower ∆S/So buildup with long-τ values close to 1, which is reflected by
the ∆S/So buildup of I4_A8DA9C (Figure 3.13).
lab

Exponential fitting of (∆S/So)

vs τ

Given the exponential buildup shape, the labeled ∆S/So vs τ of I4_A8DA9C is
fitted using the function Ax(1-e

-βτ

). A is the fraction of

13

CO nuclei within the r < 9 Å
13

2

detection limit and β = αd, where α is a coefficient and d is the CO- H dipolar coupling.
The best-fit values are A = 0.87(5) and β = 24(2) Hz. Since d = 37 Hz for I4_A8DA9C,
the coefficient α = β/d = 0.65(5). Since ∆S/So of a spin 1/2-spin 1/2 pair has a single
universal dependence on λ where λ = τd, we expect that ∆S/So of a spin 1/2-spin 1 pair
also has a universal dependence on λ and thus the coefficient α = 0.65 for I4_A8DA9C
may be also used for other different samples[10].

112

REFERENCES

113

REFERENCES

1.

Gullion, T. and J. Schaefer, Rotational-Echo Double-Resonance NMR. Journal of
Magnetic Resonance, 1989. 81(1): p. 196-200.

2.

Ghosh, U., L. Xie, and D.P. Weliky, Detection of closed influenza virus
hemagglutinin fusion peptide structures in membranes by backbone (CO)-C-13N-15 rotational-echo double-resonance solid-state NMR. Journal of Biomolecular
NMR, 2013. 55(2): p. 139-146.

3.

Schmick, S.D. and D.P. Weliky, Major Antiparallel and Minor Parallel beta Sheet
Populations Detected in the Membrane-Associated Human Immunodeficiency
Virus Fusion Peptide. Biochemistry, 2010. 49(50): p. 10623-10635.

4.

Qiang, W., Y. Sun, and D.P. Weliky, A strong correlation between fusogenicity
and membrane insertion depth of the HIV fusion peptide. Proceedings of the
National Academy of Sciences of the United States of America, 2009. 106(36): p.
15314-15319.

5.

Long, H.W. and R. Tycko, Biopolymer conformational distributions from solidstate NMR: alpha-helix and 3(10)-helix contents of a helical peptide. Journal of
the American Chemical Society, 1998. 120(28): p. 7039-7048.

6.

Gullion, T., Rotational-Echo, Double-Resonance NMR. Modern Magnetic
Resonance, 2008: p. 713-718.

7.

Cady, S.D., et al., Structure of the amantadine binding site of influenza M2 proton
channels in lipid bilayers. Nature, 2010. 463(7281): p. 689-U127.

8.

Morcombe, C.R. and K.W. Zilm, Chemical shift referencing in MAS solid state
NMR. Journal of Magnetic Resonance, 2003. 162(2): p. 479-486.

9.

Zhang, H.Y., S. Neal, and D.S. Wishart, RefDB: A database of uniformly
referenced protein chemical shifts. Journal of Biomolecular Nmr, 2003. 25(3): p.
173-195.

10.

Gullion, T., Introduction to rotational-echo, double-resonance NMR. Concepts in
Magnetic Resonance, 1998. 10(5): p. 277-289.
114

Chapter 4 - Membrane location studies of KALP and HIV fusion peptide (HFP)
4.1 Background
The insertion depth (i.e. membrane location) of protein and peptide residues in
lipid membranes is an important feature of membrane-bound proteins and peptides. For
instance, the membrane insertion depth of HFP is crucial for its membrane fusion
activity[1]. Earlier lipid mixing assays and REDOR SSNMR studies together have shown
a strong positive correlation between the membrane insertion depth and fusion activity
of HFP[1-3]. The residue-specific locations of proteins and peptides in membranes can
be detected by a variety of techniques including fluorescence spectroscopy, EPR, and
SSNMR (e.g. PRE, spin diffusion, and REDOR), which have been discussed in detail in
Chapter 1. In this chapter, we will focus on the membrane location studies of peptides
using REDOR SSNMR.
As previously described in Chapter 3, our group have been using
13

31

19

13

C- F REDOR to probe the membrane locations of peptides. For peptide

31

C- P and

13

C to lipid

P REDOR, it detects membrane locations of peptides with respect to the membrane

surface. For peptide
with

respect

to

13

C to lipid

the

19

F REDOR, it detects membrane locations of peptides

membrane

center

when

using
1

the

19

C16- F

labeled

19

dipalmitoylphosphatidylcholine (DPPC) lipid that contains a H→ F substitution at C16
in one of the two palmitoyl chains (Figure 4.1b). However, a disadvantage of using the
19

C16- F labeled DPPC lipid is that the membrane bilayer integrity may be disrupted.

115

1

19

Earlier studies have shown that 100% monofluorinated DPPC lipids with a H→ F
substitution at C16 in one of the two acyl chains form an interdigitated bilayer rather
than a regular bilayer (Figure 4.1)[4]. The C-F bond at C16 in DPPC has a very strong
electric dipole that is energetically more stable in a polar environment and drives the
formation of the interdigitated lipid bilayer where the C-F bond may contact the polar
headgroup region of the DPPC lipids.

Figure 4.1 (a) Regular lipid bilayer (left) consisting of unlabeled DPPC and
1
19
interdigitated bilayer (right) consisting of labeled DPPC that contains a H→ F
19

substitution at C16 in one of the two palmitoyl chains. (b) Structure of C16- F labeled
DPPC lipid.

116

Figure 4.2 (a) D54: dimyristoylphosphatidylcholine perdeuterated in the myristoyl
chains. (b–d) D4, D8, and D10: dipalmitoylphosphatidylcholine deuterated at palmitoyl
carbons 2; 7 and 8; and 15 and 16, respectively[5].
By contrast, for peptide

13

C to lipid

2

H REDOR, there is no membrane

perturbation no matter what fraction of lipids are deuterated since

1

H and

2

H are

chemically equivalent. For the lipids in Figure 4.2, the perdeuterated DMPC lipid (D54)

117

is used to qualitatively determine whether the

13

CO labeled residue is inserted in the

membrane hydrocarbon core whereas the selectively deuterated DPPC lipids (D4, D8,
and D10) are used to quantitatively detect the residue-specific membrane locations of
peptides.
4.2 Peptide synthesis and sample preparation
Peptide sequence and synthesis
The membrane-associated peptides KALP_A11C, HFP_G5C, and HFP_F8C were
manually synthesized by Fmoc SPPS. KALP is a designed membrane-spanning αhelical peptide and KALP_A11C has the sequence GKKLALALALALALALALALKKA
with a

13

CO label at A11, acetylation at the N-terminus, and amidation at the C-

terminus[6].

HFP_G5C

and

HFP_F8C

AVGIGALFLGFLGAAGSTMGARSWKKKKKKA with a

have
13

the

sequence

CO label at either G5 or F8.

The underlined residues are the 23 N-terminal residues of HIV gp41 fusion protein. A
tryptophan residue is added as a 280 nm chromophore for peptide quantification and a
non-native six-lysine tag is added to increase the aqueous solubility of HFP. The
synthesized peptides were purified by reversed-phase HPLC using a preparative C4
column. The peptide purities were checked by mass spectrometry (MALDI and ESI).
Comparison of peak intensities in the mass spectra indicated the peptides were >95%
pure.

118

REDOR sample preparation
For the REDOR samples, each KALP sample was prepared using 1 µmol KALP
and 50 µmol lipids whereas each HFP sample was prepared using 2 µmol HFP and 50
µmol lipids. The KALP_A11C and HFP_F8C samples were prepared in a time sequence
of (1) dissolving the peptide and lipid (either D4, D8, D10, or D54) together in a mixture
of 2,2,2-trifluoroethanol, chloroform, and 1,1,1,3,3,3-hexafluoroisopropanol in a 2:3:2
volume ratio. The initial co-solubilization in organic solvents minimized the fraction of
kinetically trapped peptide on the membrane surface; (2) removing the organic solvents
using N2 gas in the fume hood; (3) drying the peptide-lipid film in a vacuum desicator
overnight; (4) suspending the peptide-lipid film in 2~3 mL pH 7.4 buffer which contains 5
mM HEPES and 10 mM MES; (5) performing ten cycles of "freeze/thaw", where the
"freeze" in liquid N2 breaks apart the lipid vesicles and the "thaw" in water bath at room
temperature drives the lipid molecules to reform vesicles in the buffer; (6) adding more
buffer until the centrifuge tube is full (note that a balance centrifuge tube is needed and
the mass difference between the balance and sample tubes should be within 100 mg);
(7) ultracentrifugation at 270,000g for 4 hours (UV measurement at 280 nm showed little
peptide in the supernatant liquid after ultracentrifugation); (8) lyophilizing the pellet in a
freeze dryer overnight; (9) adding 10 µL pH 7.4 buffer to the bottom of a 4mm rotor,
packing the lyophilized pellet in the rotor, adding another 10 µL buffer on top of the
sample, and finally allowing overnight sample hydration before running
experiments.

119

13

2

C- H REDOR

The HFP_G5C samples were prepared in the same procedure as described
above except that the lipid membrane consisted of either 100% D4 or 80% D8 (or D10)
+ 20% DTPG, where the PC:PG (4:1) lipid composition reflected a small fraction (~10%)
of negatively charged lipids and a significant fraction (~85%) of zwitterionic lipids in HIV
host cell membranes[7].
4.3

13

2

2

C- H REDOR experiments and H T1 measurements

Part I:

13

2

C- H REDOR experiments

The

13

2

1

C- H REDOR pulse sequence (Figure 4.3) consists of (1) H π/2 pulse
1

1

which rotates H magnetization from z axis to the xy plane, (2) simultaneous H and
13

1

C CP pulses which transfer magnetization from H to

intensity, (3)

13

13

C to enhance the

13

C signal

C π pulses at the end of each rotor period except the last one without (So
2

experiment) and with (S1 experiment) H π pulses in the middle of each rotor period
during the dephasing time τ, and (4)

13

C signal acquisition. TPPM

1

H decoupling

π pulses are applied during τ and the acquisition period. The REDOR experimental
o

conditions included 10 kHz MAS frequency, sample cooling gas temperature of -50 C
o

1

with actual sample temperature of approximately -30 C, 5.0 µs H π/2 pulse, 1.5 ms
1

CP with 50 kHz H and 62-66 kHz ramped

13

C fields, 8.3 µs

13

2

C π pulses, 5.1 µs H π

1

pulses, and 75 kHz TPPM H decoupling fields. The recycle delay was 1 s for τ = 2 ms,
8 ms, and 16 ms, 1.5 s for τ = 24 ms and 32 ms, and 2 s for τ = 40 ms and 48 ms,
120

respectively. The pulses were optimized using the setup peptide I4_A8DA9C as
described in chapter 3.
KALP_A11C samples
Figure 4.4 displays the

13

CO regions of the So and S1 spectra of KALP_A11C for

τ = 40 ms in D4, D8, D10, and D10 lipid membranes, respectively. The strong signal at
178.7 ppm is dominated by the labeled A11

Figure 4.3

13

13

CO (~0.80 molar fraction) and this peak

2

C- H REDOR pulse sequence.

shift is consistent with helical Ala in a peptide or protein[8]. The fraction 0.80 is
determined by spin counting. For KALP_A11C, the labeled
unlabeled (i.e. natural abundance)

13

contribution from the labeled A11

13

CO signal is 0.99 and the

CO signal is 0.24 (0.011 x 22), thus the fractional

13

CO is ~0.80. The shoulder at 175.7 ppm is

121

dominated by natural abundance lipid

13

CO nuclei. Comparison of the So and S1

spectra for τ = 40 ms with those for τ = 2 ms for KALP_A11C in D4 membrane (Figure
4.5) shows that the lipid signal (175~176 ppm) is less intense relative to the peptide
signal (~179 ppm) at longer τ, which indicates a shorter lipid T2[5]. For the D4
membrane, there is a significant lipid signal dephasing even for the very short τ = 2 ms
since the carbonyl carbon (C1) is very close to the deuterons at C2 (~2 Å apart) in each
acyl chain.

13

2

C- H REDOR So (black) and S1 (colored) spectra for τ = 40 ms for
KALP_A11C in D4, D8, D10, and D54 membranes, respectively. All spectra were
processed using 100 Hz Gaussian line broadening and baseline correction.
Figure 4.4

122

Figure 4.5

13

2

C- H REDOR So (black) and S1 (purple) spectra for τ = 2 ms (a) and τ =

40 ms (b) for KALP_A11C in D4 lipid membrane.
The

13

2

C- H REDOR experimental dephasing plots of ∆S/So vs τ for KALP_A11C

in D4, D8, D10, and D54 membranes are displayed in Figure 4.6 and will be discussed
in section 4.4.

123

Figure 4.6

13

2

C- H REDOR dephasing plots of ∆S/So vs τ for KALP_A11C in D4

(purple), D8 (blue), D10 (red), and D54 (green) membranes, respectively. The ∆S/So at
13
each τ was calculated using So and S1 CO intensities determined over a 3.0 ppm
integration width.
HFP_F8C samples
Figure 4.7 displays the

13

CO regions of the So and S1 spectra of HFP_F8C for τ =

40 ms in D4, D8, D10, and D10 lipid membranes, respectively. The F8

13

CO peak at

174 ppm is consistent with β-sheet structure and is partially overlapped with the natural
abundance lipid

13

CO signal at 175~176 ppm[8]. The experimental dephasing plots of

∆S/So vs τ for HFP_F8C in membranes are displayed in Figure 4.8.

124

13

2

C- H REDOR So (black) and S1 (colored) spectra for τ = 40 ms for
HFP_F8C in D4, D8, D10, and D54 membranes, respectively. All spectra were
processed using 100 Hz Gaussian line broadening and baseline correction.
Figure 4.7

125

13

2

C- H REDOR dephasing plots of ∆S/So vs τ for HFP_F8C in D4 (purple),
D8 (blue), D10 (red), and D54 (green) membranes, respectively. The ∆S/So at each τ

Figure 4.8

was calculated using So and S1
width.

13

CO intensities determined over a 3.0 ppm integration

HFP_G5C samples
Figure 4.9 displays the

13

CO regions of the So and S1 spectra for τ = 40 ms for

HFP_G5C in D4 (100%), D8+DTPG (4:1), and D10+DTPG (4:1) membranes,
respectively. The peak at 171 ppm is dominated by the labeled G5

13

CO and is

consistent with β-sheet G5 in the peptide. The shoulder at 175 ppm in the So spectrum
of Figure 4.9a may have contributions from labeled

126

13

CO of helical G5, n.a.

13

CO of

unlabeled residues, and n.a.

13

CO of D4 lipid. The experimental dephasing plots of

∆S/So vs τ for HFP_G5C in membranes are displayed in Figure 4.10.

Figure 4.9

13

2

C- H REDOR So (black) and S1 (colored) spectra for τ = 40 ms for

HFP_G5C in "100% D4" (a), "80% D8 + 20% DTPG" (b), and "80% D10 + 20% DTPG"
(c) membranes, respectively. All spectra were processed using 100 Hz Gaussian line
broadening and baseline correction.

127

13

2

Figure 4.10 C- H REDOR dephasing plots of ∆S/So vs τ for HFP_G5C in "100% D4"
(purple), "80% D8 + 20% DTPG" (blue), and "80% D10 + 20% DTPG" (red) membranes,
13
respectively. The ∆S/So at each τ was calculated using So and S1 CO intensities
determined over a 3.0 ppm integration width.
2

Part II: H T1 measurements
To investigate the relaxation effects on

13

2

C- H REDOR dephasing buildups of

2

∆S/So vs τ, the lipid H T1 relaxation times were measured by t1D_ir under both static
and 10 kHz MAS conditions for HFP_F8C samples in D8, D10, and D54 membranes,
respectively. The t1D_ir pulse sequence is inversion-recovery followed by a quadrupolar
2

echo, π – τ1 – (π/2)x – τ2 – (π/2)y – τ3 – detect (Figure 4.11). H spectra were acquired
128

for different τ1 with fixed τ2 and τ3 (τ2 was set as 100 µs such that the two π/2 pulses
were rotor-synchronized under 10 kHz MAS).

2

Figure 4.11 "t1D_ir" pulse sequence used for H T1 measurements. The phase/phase
cycle is "x" for the π pulse, "x, -x, y, -y" for the first π/2 pulse, and "y, y, x, x" for the
second π/2 pulse.
2

H spectra under static conditions (no spinning)
2

o

The static H spectra were acquired at -50 C for τ1 = 0.1 ms through 300 ms in
an increment of 20 ms for HFP_F8C in D8, D10, and D54 membrane, respectively. The
2

number of acquisitions was set the same for each τ1. Ideally, all the H spectra for
2

different τ1 should be acquired in the same array. H spectra were also acquired for
HFP_F8C in D4 but the signal-to-noise was very poor, which was probably due to (a)
small population of deuterons compared to the D8, D10, and D54 samples and (b) large
quadrupolar anisotropy causing spectral broadening and thus lower signal-to-noise ratio.
2

2

As we process the H FID data, we need to do "data shift" to move the H
quadrupolar echo which is the most intense signal in the FID to time zero before
performing Fourier transformation (FT). For instance, if the dwell time which is the
2

interval between two nearby sampling points is 2 µs and the H quadrupolar echo in the
129

FID appears at 22 µs, we need to set "data shift" = -11 (i.e. shift the FID to the left by 11
data points) in the process panel before performing FT. For the same sample (e.g.
2

HFP_F8C in D10), all H spectra were processed using the same parameters (e.g.
2

zero- and 1st-order phasing, data shift, and line broadening). The static H spectra for
HFP_F8C in D8, D10, and D54 membranes are displayed in Figure 4.12, 4.13, and 4.14,
respectively.

130

o

Figure 4.12 "t1D_ir" experiments of HFP_F8C in D8 membrane at -50 C under static
2

condition. For each τ1, the number of acquisition = 16000. (a) H FID for τ1 = 0.1 ms
2
and 200.1 ms; (b) H spectra for τ1 = 0.1 ms through 400.1 ms in an increment of 40 ms.
All spectra were processed using 2000 Hz Gaussian line broadening and data shift of 12. For simplicity, spectra for τ1 = 20.1 ms through 380.1 ms in an increment of 40 ms
are not shown here.
131

o

Figure 4.13 "t1D_ir" experiments of HFP_F8C in D10 membrane at -50 C under static
2

condition. For each τ1, the number of acquisition = 3000. (a) H FID for τ1 = 0.1 ms and
2

200 ms; (b) H spectra for τ1 = 0.1 ms through 100 ms. All spectra were processed
using 2000 Hz Gaussian line broadening, data shift of -11, and baseline correction of
order 3. For simplicity, spectra for τ1 = 120 ms through 300 ms are not shown.

132

o

Figure 4.14 "t1D_ir" experiments of HFP_F8C in D54 membrane at -50 C under static
2

condition. For each τ1, the number of acquisition = 2000. (a) H FID for τ1 = 0.1 ms and
2

200 ms; (b) H spectra for τ1 = 0.1 ms through 100 ms. All spectra were processed
using 2000 Hz Gaussian line broadening and data shift of -12. The outer horns (1 and 4)
and inner horns (2 and 3) were diagnostic of -CD3 and -CD2- in D54 membrane,
respectively. For simplicity, spectra for τ1 = 120 through 300 ms are not shown.

133

2

H spectra under 10 kHz MAS
2

The H spectra under MAS for HFP_F8C in D8, D10, and D54 membranes were
acquired under the same experimental conditions as those for static

2

H spectra

2

described above except for 10 kHz MAS. The H spectra under MAS for HFP_F8C in
D8, D10, and D54 membranes are displayed in Figure 4.15, 4.16 and 4.17, respectively.

o

Figure 4.15 "t1D_ir" experiments of HFP_F8C in D8 membrane at -50 C under 10 kHz
2

MAS. For each τ1, the number of acquisition = 15000. (a) H FID for τ1 = 0.1 ms and
2
200.1 ms; (b) H spectra for τ1 = 0.1 ms through 140.1 ms. All spectra were processed
using 200 Hz Gaussian line broadening, data shift of -32, and baseline correction of
order 3. For simplicity, spectra for τ1 = 160.1 through 400.1 ms are not displayed.

134

o

Figure 4.16 "t1D_ir" experiments of HFP_F8C in D10 membrane at -50 C under 10
2

kHz MAS. For each τ1, the number of acquisition = 5000. (a) H FID for τ1 = 0.1 ms and
2

200.1 ms; (b) H spectra for τ1 = 0.1 ms through 100.1 ms. All spectra were processed
using 200 Hz Gaussian line broadening, data shift of -31, and baseline correction of
order 10. For simplicity, spectra for τ1 = 120.1 through 300.1 ms are not displayed.

135

o

Figure 4.17 "t1D_ir" experiments of HFP_F8C in D54 membrane at -50 C under 10
2

kHz MAS. For each τ1, the number of acquisition = 3000. (a) H FID for τ1 = 0.1 ms and
2

200.1 ms; (b) H spectra for τ1 = 0.1 ms through 100.1 ms. All spectra were processed
using 200 Hz Gaussian line broadening and data shift of -12. For simplicity, spectra for
τ1 = 120.1 ms through 300.1 ms are not displayed.
Data fitting
(1) under static condition

136

(1a) For HFP_F8C in D8 membrane, there were two horns (Pake doublet, see Chapter
2

2

1 for more details) in each H spectrum, one at 62 kHz and the other at -62 kHz. The H
intensity for each τ1 was calculated as the sum of the integrations over a 100 ppm (~6
2

kHz) width for both horns. The H signals correspond to the -CD2- deuterons in D8
membrane. The spectral error σ was calculated as the standard deviation of twelve
2

noise integrals over a 100 ppm width for each and the uncertainty of the H signal (i.e.
sum of the two horn integrals) for each τ1 was calculated as

2 σ.
2

(1b) For HFP_F8C in D10 membrane, there were also two horns in each H spectrum
(Figure 4.12), one at ~10 kHz and the other at ~-20 kHz, which correspond to the -CD3
deuterons in D10 membrane. The

2

H intensity and uncertainty for each τ1 were
2

calculated in the same way as described in (1a). For the -CD2- deuterons, the H
intensity for each τ1 was not calculated due to the poor resolution and low signal-tonoise ratio.
(1c) For HFP_F8C in D54 membrane, there were two outer horns corresponding to the
2

-CD2- deuterons and two inner horns corresponding to the -CD3 deuterons in each H
2

spectrum (Figure 4.13). The H intensity for -CD2- for each τ1 was calculated as the
sum of the integrations over a 100 ppm (~6 kHz) width for the two outer horns and the
intensity for -CD3 was calculated as the sum of the integrations over a 100 ppm (~6 kHz)

137

2

width for the two inner horns. The uncertainty of each H intensity was calculated in the
same way as described in (1a).
2

For each sample, the integrated H intensity vs τ1 was fitted by
I(τ1) = I0 + {∆I × [1 – exp(–τ1 /T1)]}

(4.1)

where I0, ∆I, and T1 are fitting parameters and respectively correspond to I(τ1 = 0), [I(τ1
2

= ∞) – I(τ1 = 0)], and 1/(longitudinal H relaxation rate). The best-fit plots are displayed
in Figure 4.18 and best-fit T1 values are listed in Table 4.1.

138

Figure 4.18 "t1D_ir" experimental (black squares with uncertainties) and best-fit (red
2
line) plots of H intensity vs τ1 under static conditions for CD2 in D8 sample (a), CD3 in
D10 sample (b), and CD2 and CD3 in D54 sample (c, d), respectively. a.u. ≡ arbitrary
unit.
(2) under 10 kHz MAS
(2a) For HFP_F8C in D8 membrane, each

2

H spectrum was split into a series of
2

spinning sidebands with equal spacing of 10 kHz. The H intensity for each τ1 was
calculated as the sum of integrations of the centerband and 12 dominant sidebands
over a 12 ppm integration width for each (the line width at half maximum (LWHM) for
139

each sideband is ~12 ppm). The spectral error σ was calculated as the standard
deviation of twelve noise integrals over a 12 ppm integration width for each and the
2

uncertainty of the H intensity (i.e. sum of the centerband and 12 sideband integrals) for
each τ1 was calculated as 13 σ.
2

(2b) For HFP_F8C in D10 membrane, each H spectrum was also split into a series of
2

spinning sidebands with equal spacing of 10 kHz. The H intensity for each τ1 was
calculated as the sum of integrations of the centerband and 6 dominant sidebands over
a 6 ppm integration width for each (the LWHM for each sideband is ~6 ppm). The
spectral error σ was calculated as the standard deviation of twelve noise integrals over
2

a 6 ppm integration width for each and the uncertainty of the H intensity (i.e. sum of the
centerband and 6 sideband integrals) for each τ1 was calculated as

7 σ.

2

(2c) For HFP_F8C in D54 membrane, the H intensity for each τ1 was calculated as the
sum of integrations of the centerband and 10 dominant sidebands over a 6 ppm
integration width for each (the LWHM for each sideband is ~6 ppm). The spectral error
σ was calculated as the standard deviation of twelve noise integrals over a 6 ppm
integration width for each and the uncertainty of the

2

H intensity (i.e. sum of the

centerband and 10 sideband integrals) for each τ1 was calculated as

11 σ.

2

For the D8 and D54 samples, the integrated H intensity vs τ1 was fitted by
2

equation 4.1. For the D10 sample, the integrated H intensity vs τ1 was fitted by

140

I(τ1) = I0 + {0.4 × ∆I × [1 – exp(–τ1/T1(CD2))]} + {0.6 × ∆I × [1 – exp(–τ1/T1(CD3))]} (4.2)
where separate –CD2 and –CD3 contributions to I(τ1) were considered with distinct
T1(CD2) and T1(CD3). The best-fit plots are displayed in Figure 4.19 and best-fit T1
values are listed in Table 4.1.

141

Figure 4.19 "t1D_ir" experimental (black squares with uncertainties) and best-fit (red
2
2
line) plots of H intensity vs τ1 under 10 kHz MAS for H in D8 (a), D10 (b), and D54
samples (c), respectively.
142

2

Table 4.1 Best-fit lipid H T1 in ms with uncertainty in parentheses
Sample

Static

10 kHz MAS

CD3: 57(2)
90(4)

D54 + HFP
CD2: 91(6)
CD3: 43(1)

CD3: 58(3)

CD2: not determined

CD2: 92(8)

CD2: 148(9)

CD2: 153(3)

D10 + HFP

D8 + HFP

4.4 Result discussion and conclusions
Initial fitting of the REDOR data to determine

13

2

C- H dipolar coupling d was done
exp

using SIMPSON simulations. However, similar to the (∆S/So)
which contained a single
exp

the (∆S/So)

13

buildup of I4_A8DA9C

2

CO- H spin pair with r = 5.0 Å and d = 37 Hz (Figure 3.9),

buildups of all KALP and HFP samples had exponential shape which
sim

contrasted with the sigmoidal shape of (∆S/So)
exp

chapter 3, the differences between (∆S/So)

buildups. As previously discussed in
sim

and (∆S/So)

buildups could be due to

2

the H T1 relaxation in the experiments which was not considered in the SIMPSON
simulations. This possibility was supported by the ∆S/So buildup of I4_A8DA9C (see
2

chapter 3 for more details) and the lipid H T1 values in the 40-150 ms range that were

143

exp

comparable to the larger values of τ in experiments (Table 4.1). All (∆S/So)
were fitted well by the function Ax(1-e

-βτ

). A is the fraction of

13

buildups

CO nuclei within the r <

9 Å detection limit and β = 0.65xd, where 0.65 is a coefficient determined from the
known d = 37 Hz for I4_A8DA9C (see chapter 3). The best-fit parameters are
summarized in Table 4.2 and best-fit plots of ∆S/So vs τ for KALP_A11C, HFP_F8C,
and HFP_G5C are displayed in Figure 4.20a, Figure 4.22a, and Figure 4.22b,
1/3

respectively. For the distance r = (4642/d)

, its uncertainty listed in Table 4.2 is

calculated by

σ (r ) = 46421/3 xσ (d −1/3 )
1/3

= 4642

x ( −1/ 3)xd −4/3 xσ (d )

1/3
−1
= ( −1/ 3)x (4642 / d ) xd xσ (d )

= ( −r / 3)x

σ (d )

(4.3)

d

where σ(d) is the uncertainty of dipolar coupling d and the - sign can be removed since
the uncertainty is usually reported as a positive number (magnitude). The derivation of
n

σ(r) is based on the general equation to calculate the uncertainty of A :
n

σ(A ) = n x A

n-1

x σ(A)

(4.4)

where σ(A) is the uncertainty of A. According to equation 4.3, the relative uncertainty of
r is only 1/3 of that of d, i.e. σ(r)/r = (1/3) x σ(d)/d.
144

Membrane locations of KALP
The A(D54) = 0.96 and r(D54) = 3.3 Å for KALP_A11C in perdeuterated D54
membrane support a transmembrane KALP topology with van der Waals contact
between KALP and the membrane hydrocarbon core. This topology is further supported
by (∆S/So) ≈ 0 for the D4 sample where the

2

Hs are close to the membrane

headgroups. For a single membrane location of KALP, we would expect A(D8) ≈ A(D10)
and possibly different values of d(D8) and d(D10) and therefore different r(D8) and
r(D10). However, the experimentally derived A(D8) and A(D10) were very different while
r(D8) ≈ r(D10) ≈ 4 Å. We were unable to interpret these results quantitatively and selfconsistently using a single membrane location of α helical KALP with sidechains. We
were successful using two distinct membrane locations with major and minor
2

populations. The major population has A11 contact with D10 H nuclei whereas the
2

minor population has A11 contact with D8 H nuclei (Figure 4.20b). The distance r(D8)
and r(D10) were converted to the membrane locations of A11

13

CO relative to the

bilayer center using the membrane location models of KALP in Figure 4.21. In an α
helical peptide structure, the backbone radius of the helix is 2.3 Å. For KALP with
sidechains, the overall α helix radius is approximately 2.3 Å + 3.8 Å = 6.1 Å, where 3.8
2

2

Å is the estimated length of leucine sidechain. The distance between D8 H and D10 H
in the DPPC membrane is ~ 9 Å. For the major KALP location (Figure 4.21a), a is the
distance between A11

13

CO nucleus and DPPC membrane bilayer center, r is the

closest distance between A11

13

2

2

2 1/2

CO and lipid H nuclei, and r = r(D10) = (3.8 + a )
145

=

Table 4.2

13

2

C- H REDOR fitted parameters

Sample

A

β(Hz)

d(Hz)

r(Å)

I4_A8DA9C

0.87(5)

24(2)

37

5.0

KALP_A11C in D8

0.15(2)

47(10)

72(15)

4.0(3)

KALP_A11C in D10

0.48(4)

34(5)

52(7)

4.5(2)

KALP_A11C in D54

0.96(1)

85(4)

131(12) 3.3(1)

HFP_F8C in D8

0.21(1)

71(10)

109(17) 3.5(2)

HFP_F8C in D10

0.82(20) 16(5)

25(8)

HFP_F8C in D54

0.99(1)

122(1)

188(14) 2.9(1)

HFP_G5C in D8+DTPG

0.45(5)

27(5)

42(8)

4.8(3)

HFP_G5C in D10+DTPG

0.85(3)

37(3)

57(6)

4.3(2)

5.7(6)

4.5 Å which corresponds to a = 2.4 Å. For the minor KALP location (Figure 4.21b), r =
2

2 1/2

r(D8) = (3.8 + b )
the minor A11

13

= 4.0 Å which corresponds to b = 1.2 Å and 9 Å – b = 7.8 Å, i.e.

CO location is 7.8 Å from the DPPC membrane center.

For the membrane location models in Figure 4.21, KALP was considered as a
monomer in the membrane. Although earlier molecular dynamics simulations have
suggested that KALP and WALP, which is similar to KALP in both sequence and
146

topology except that the flanking residues are tryptophan rather than lysine, could
aggregate and form oligomers, this possibility was ruled out in our experiments because
a rapid and complete buildup of ∆S/So vs τ was observed for KALP in D54 membrane
(Figure 4.6)[9, 10]. For an oligomer population, we expect that there would be some
A11 residues in the oligomer interior. This population would not have van der Waals
2

contact with lipid H and would dephase slowly.
One possible reason for multiple membrane locations of KALP may be the
hydrophobic mismatch. The hydrophobic length of KALP is ~26 Å (L4 to L20) whereas
the hydrophobic thickness of the DPPC membrane is ~31 Å. To compensate for the
hydrophobic mismatch, the lysine sidechains of KALP could snorkel into the DPPC
membrane headgroup region to increase the hydrophobic length of KALP, which is
consistent with the snorkeling model proposed for charged residues such as lysine and
arginine in transmembrane peptides[11, 12]. The different KALP locations may be
correlated with different snorkeling geometries in the membrane (Figure 4.16b). There
could be additional membrane locations for KALP because the sum A(D8) + A(D10) =
0.65 is significantly smaller than A(D54) = 0.96. Although a fraction of KALP molecules
trapped on the membrane surface may explain why A(D8) + A(D10) < 1, this possibility
was ruled out since a complete (i.e. 100%) dephasing was observed for KALP in the
D54 membrane (Figure 4.6).

147

Figure 4.20 (a) Experimental (points with uncertainties) and best-fit (solid line) plots of
∆S/So vs τ for KALP_A11C in membranes. The D4 data were not fitted. (b) Membrane
locations of KALP with major (left) and minor (right) populations. The colored bands,
2
13
orange dots, and brown lines represent H positions, A11 CO nuclei, and lysine
sidechains, respectively. The thickness of the palmitoyl region of the DPPC membrane
is ~31 Å[5].

148

Figure 4.21 Models of α helical KALP in the DPPC membrane. The green ribbon is the
KALP backbone and the vertical lines are the full van der Waals extent of the helix
including leucine sidechains. The horizontal black dashed lines are the boundaries of
2
the ~31 Å-thick palmitoyl region of the bilayer. The blue and red bands are the H
locations in the D8 and D10 membranes, respectively. Model (a) is the major KALP
location with a = 2.4 Å calculated using the best-fit r (D10) = 4.5 Å. Model (b) is the
minor KALP location with (9 Å – b) = 7.8 Å calculated using the best-fit r (D8) = 4.0 Å.
149

Membrane locations of HFP
The A(D54) = 0.99 and r(D54) = 2.9 Å for HFP_F8C support van der Waals
contact between F8 and the DMPC membrane hydrocarbon core for all HFP molecules.
Since the structure of DMPC is quite similar to that of DPPC except that DPPC has two
more carbons in each acyl chain (Figure 4.2), F8 should also have van der Waals
contact with the DPPC membrane hydrocarbon core. Similar contact is expected for G5
because A(D10) = 0.85 and A(D8) + A(D10) > 1. For either HFP_F8C or HFP_G5C,
there are very different values of A(D8) and A(D10), and for HFP_G5C, r(D8) ≈ r(D10 ) ≈
4.5 Å. These trends support at least two distinct membrane locations of HFP. The larger
A(D10) values are attributed to a major HFP population with deep membrane insertion
2

and HFP contact with D10 H nuclei and the smaller A(D8) values are attributed to a
2

minor population with shallow HFP insertion and HFP contact with D8 H nuclei (Figure
4.22c, d). The major-to-minor population ratio is ~7:3 as calculated from the τ = 48 ms
(∆S/So)D10-to-(∆S/So)D8 ratio for either the HFP_F8C or HFP_G5C samples. There is
negligible HFP localized to the membrane surface, as evidenced by (∆S/So) ≈ 0.1
without buildup for either HFP_F8C or HFP_G5C bound to the D4 membrane (Figure
4.8 and 4.10). The multiple membrane locations of HFP are attributed to the distribution
of antiparallel β sheet registries[13]. Specifically, the membrane insertion depth of a
HFP registry likely depends on the lengths of its contiguous hydrophobic regions and
these lengths vary among registries. Deep and shallow insertions may also have a
distribution of membrane locations of HFP. The predominant deep insertion of HFP
150

could significantly perturb the membrane bilayer and lower the activation energy of
membrane fusion. This is consistent with the observed strong positive correlation
between membrane insertion depth and fusion rate for several HFP constructs[1].

151

Figure 4.22

13

2

C- H REDOR experimental (points with uncertainties) and best-fit (solid

line) plots of ∆S/So vs τ for (a) HFP_F8C and (b) HFP_G5C in membranes. (c, d) Deep
and shallow membrane insertion of HFP. Residues A1 and A14 are close to the
13
31
membrane surface in concurrence with peptide CO-lipid P distances of ~5 Å for
these residues[1]. For clarity, only one β strand is displayed.
152

It is noticeable that the best-fit A values are in an increasing order of A(D4) <
A(D8) < A(D10) < A(D54) for both KALP and HFP, where the D4 data were not fitted but
A(D4) is expected to be the smallest. One possible reason for the increasing order of A
2

could be that there is an increasing order of mobility of H from D4 to D54 which causes
2

2

an increasing order of H spin inversion efficiency from D4 to D54. In recent years H
composite π pulses have been used to compensate for pulse imperfections and
2

incomplete spin inversion due to large H quadrupolar resonance offsets. Sack and
2

coworkers showed that H composite π pulses introduced a two times faster

15

2

N- H

2

REDOR dephasing buildup rate than simple π pulses did, where the H quadrupolar
splitting was ~146 kHz and B1 field was ~40 kHz[14]. On the other hand, Cady and
2

coworkers showed that the H spin inversion efficiency in

13

2

C- H REDOR was only
2

about 70% (i.e. maximum dephasing was ~70% x 2/3) when applying H composite
2

π pulse with a B1 field of ~40 kHz[15]. In contrast to these studies, H simple π pulses
with a B1 field of ~100 kHz were used in our

13

2

2

C- H REDOR experiments. Although H

pulse imperfections and incomplete spin inversion can affect the

13

C dipolar dephasing,

the main reason for the increasing order of A(D4) < A(D8) < A(D10) < A(D54) for both
KALP and HFP is the multiple membrane locations of the peptide rather than an
increasing order of deuteron mobility from D4 to D54, as supported by the following
experimental observations:

153

(I) Buildups of ∆S/So vs τ for other membrane peptide systems do not show an
increasing order of A(D4) < A(D8) < A(D10). For example, IFP_L2C which has the
sequence of GLFGAIAGFIENGWEGMIDGGGKKKK with a

13

CO label at residue L2

shows an order of A(D10) < A(D8) (Figure 4.23) whereas HFP_G5C with a point
mutation of L9→R shows an order of A(D8) < A(D4) < A(D10) (Figure 4.24). The order
of A for the L9R mutant of HFP_G5C supports at least two membrane locations of G5,
one close to the membrane headgroup region (resulting in the D4 buildup) and the other
close to the membrane center (resulting in the D10 buildup).

13

2

Figure 4.23 C- H REDOR experimental plots ∆S/So vs τ for IFP_L2C in either D8
(blue) or D10 (red) membrane.

154

Figure 4.24

13

2

C- H REDOR experimental plots ∆S/So vs τ for the L9R mutant of

HFP_G5C in "D4+DPPG" (4:1, purple), "D8+DPPG" (4:1, blue), and "D10+DPPG" (4:1,
red) membranes, respectively.
(II) There is a rapid buildup of ∆S/So vs τ for the lyophilized I4 peptide with long-time
2

2

∆S/So > 2/3 (Figure 3.13), indicating an efficient H spin inversion caused by the H
simple π pulses with B1 field of ~100 kHz in the presence of a large non-motionally
2

2

averaged quadrupolar coupling from the rigid C- H bond in I4. Therefore, efficient H
spin inversions are also expected for the hydrated membrane samples.
o

(III) All samples were cooled by N2 gas at -50 C with actual sample temperature near o

30 C to minimize motion that would average the
2

13

2

CO- H dipolar coupling. At this

temperature, the H quadrupolar splittings were ~30 kHz for -CD3 in D10 and D54

155

samples and ~120 kHz for -CD2- in D4, D8, D10, and D54 samples (Figure 4.25), which
2

indicated similar -CD2- mobilities among these lipids. The lipid H quadrupolar splittings
were measured under static conditions by "quecho" experiments using the pulse
sequence (π/2)x – τ1 – (π/2)y – τ2 – detect.

156

2

Figure 4.25 Lipid H Pake doublets acquired by "quecho" experiments under static
conditions for (a) KALP_A11C in D4 and (b-d) HFP_F8C in D8 (blue), D10 (red), and
D54 (green) membranes, respectively. The number of acquisitions was 14482 for (a),
8000 for (b), 13591 for (c), and 3000 for (d), respectively. All spectra were processed
using 2000 Hz Gaussian line broadening and baseline correction.
157

(IV) Both D10 and D54 samples contain all the -CD3 groups. However, D10 contains
only -CD2- at C15 whereas D54 contains all the -CD2- (Figure 4.2). The dephasing
buildup is much faster in D54 than that in D10 for both KALP_A11C and HFP_F8C
(Figures 4.6 and 4.8). These results indicate that the buildup rate for
proximity is comparable to that for

13

13

CO-CD2

CO-CD3 with a similar proximity even if -CD2- is

much more rigid than -CD3.
Conclusions
In summary,

13

2

CO- H REDOR SSNMR reveals multiple membrane locations for

both the α helical KALP and β sheet HFP peptides. The KALP locations are attributed to
the hydrophobic mismatch and consequent snorkeling of lysine sidechains to the
membrane headgroup region. The HFP locations are attributed to the distribution of
antiparallel β sheet registries.

13

2

C- H REDOR is expected to be a general method for

determining the residue-specific distribution of membrane locations of peptides and
proteins and for developing membrane topology models.

158

REFERENCES

159

REFERENCES

1.

Qiang, W., Y. Sun, and D.P. Weliky, A strong correlation between fusogenicity
and membrane insertion depth of the HIV fusion peptide. Proceedings of the
National Academy of Sciences of the United States of America, 2009. 106(36): p.
15314-15319.

2.

Yang, R., et al., A trimeric HIV-1 fusion peptide construct which does not selfassociate in aqueous solution end which Has 15-fold higher membrane fusion
rate. Journal of the American Chemical Society, 2004. 126(45): p. 14722-14723.

3.

Gabrys, C.M., et al., Solid-State Nuclear Magnetic Resonance Measurements of
HIV Fusion Peptide (CO)-C-13 to Lipid P-31 Proximities Support Similar Partially
Inserted Membrane Locations of the alpha Helical and beta Sheet Peptide
Structures. Journal of Physical Chemistry A, 2013. 117(39): p. 9848-9859.

4.

Hirsh, D.J., et al., A new monofluorinated phosphatidylcholine
interdigitated bilayers. Biophysical Journal, 1998. 75(4): p. 1858-1868.

5.

Xie, L., et al., Residue-specific membrane location of peptides and proteins using
specifically and extensively deuterated lipids and C-13-H-2 rotational-echo
double-resonance solid-state NMR. Journal of Biomolecular NMR, 2013. 55(1): p.
11-17.

6.

de Planque, M.R.R., et al., Sensitivity of single membrane-spanning alpha-helical
peptides to hydrophobic mismatch with a lipid bilayer: Effects on backbone
structure, orientation, and extent of membrane incorporation. Biochemistry, 2001.
40(16): p. 5000-5010.

7.

Brugger, B., et al., The HIV lipidome: A raft with an unusual composition.
Proceedings of the National Academy of Sciences of the United States of
America, 2006. 103(8): p. 2641-2646.

8.

Zhang, H.Y., S. Neal, and D.S. Wishart, RefDB: A database of uniformly
referenced protein chemical shifts. Journal of Biomolecular NMR, 2003. 25(3): p.
173-195.

160

forms

9.

Kandasamy, S.K. and R.G. Larson, Molecular dynamics simulations of model
trans-membrane peptides in lipid bilayers: A systematic investigation of
hydrophobic mismatch. Biophys. J., 2006. 90(7): p. 2326-2343.

10.

Im, W. and C.L. Brooks, Interfacial folding and membrane insertion of designed
peptides studied by molecular dynamics simulations. Proceedings of the National
Academy of Sciences of the United States of America, 2005. 102(19): p. 67716776.

11.

Strandberg, E., et al., Lipid dependence of membrane anchoring properties and
snorkeling behavior of aromatic and charged residues in transmembrane
peptides. Biochemistry, 2002. 41(23): p. 7190-7198.

12.

Segrest, J.P., et al., Amphipathic Helix Motif - Classes and Properties. ProteinsStructure Function and Genetics, 1990. 8(2): p. 103-117.

13.

Schmick, S.D. and D.P. Weliky, Major Antiparallel and Minor Parallel beta Sheet
Populations Detected in the Membrane-Associated Human Immunodeficiency
Virus Fusion Peptide. Biochemistry, 2010. 49(50): p. 10623-10635.

14.

Sack, I., et al., Deuterium REDOR: Principles and applications for distance
measurements. Journal of Magnetic Resonance, 1999. 138(1): p. 54-65.

15.

Cady, S.D., et al., Structure of the amantadine binding site of influenza M2 proton
channels in lipid bilayers. Nature, 2010. 463(7281): p. 689-U127.

161

Chapter 5 - Summary and future work
During the past four years, I have been working on two projects, with one focused
on structural characterization of serotype H3 IFP (H3_20, see chapter 2 for the
sequence and other details) in membranes and the other focused on membrane
location detection of peptides including KALP and HFP.
Summary of experimental results and future work of IFP
For the IFP work,

13

15

C- N REDOR was used to quantitatively measure the

interhelical proximity of H3_20 in membranes at the fusogenic pH 5. SIMPSON
simulations of the experimental ∆S/So vs τ after n.a. calibration have show that H3_20
adopts ~40% closed structure and ~60% semi-closed structure in membranes at pH
5.1[1]. The results are contradictory to the previously reported open structure of H3_20
in membranes[2]. For future work, it will be interesting to measure the interhelical
proximity of H3_20 in membranes at the physiological, non-fusogenic pH 7.4. We can
correlate the structure with fusogenicity by comparing the structures of H3_20 at pH 5.1
and 7.4. Another interesting study is the interhelical proximity measurement of H3_23 in
membranes at pH 5.1 and 7.4. Earlier liquid-state NMR studies of H1_20 and H1_23 in
detergents have shown that H1_20 adopts predominantly open structure whereas
H1_23 adopts predominantly closed structure, indicating that the residues 21 through
23 (i.e. WYG) play a key role in stabilizing the closed structure of the peptide[3]. We can
determine whether the residues WYG also cause a greater population of closed
structure of IFP in membranes by comparing the structures of H3_20 with those of
H3_23 at either pH 5.1 or pH 7.4. Furthermore, lipid mixing assays using fluorescence

162

spectroscopy may be performed to investigate the fusogenicities of H3_20 and H3_23
at both pH 5.1 and 7.4. The impact of peptide length and structure of IFP on its
membrane fusogenicity may be elucidated based on the results of REDOR proximity
measurements and lipid mixing assays of H3_20 and H3_23.
In addition to the structural and functional studies of IFP, another interesting work
could be the insertion depth and tilt angle measurements of IFP in membranes. The
residue-specific insertion depth of IFP can be probed by
lipid and IFP which is

13

13

2

C- H REDOR using the D10

CO labeled at a specific residue (e.g. Leu 2, Ala 5). After the

insertion depths of IFP at different residues have been measured, the tilt angle of IFP
with respect to the membrane surface can be calculated using the membrane insertion
model of IFP in Figure 5.1.

13

Figure 5.1 Membrane insertion model of the N-terminal helix of IFP. The CO labeled
residue is numbered N and the residue in the membrane headgroup region is numbered
No. θ is the tilt angle of the N-terminal helix of IFP to the membrane surface, r is the
distance between the labeled

13

CO nucleus and the lipid headgroup, and d is the

distance from residue N to residue No along the N-terminal helix axis.
163

In Figure 5.1, the distance d in Å between residue N and No along the N-terminal
helix axis is given by
d (Å) = 1.5x(No - N)

(5.1)

where 1.5 Å refers to the rise per residue in an α helix. The distance r between the
labeled

13

CO nucleus of residue N and the lipid headgroup can be estimated by

r (Å) = dsinθ = 1.5x(No - N)sinθ

(5.2)

where θ is the tilt angle of the N-terminal helix of IFP to the membrane surface.
However, equation 5.2 needs to be modified since the distance r is not linearly
proportional to the residue position number N. For the N-terminal helix, r is oscillating as
a function of the residue position number N in a periodicity of 3.6 since there are 3.6
residues per turn in an α helix. Therefore, an oscillating term, Asin[2π(No - N)/3.6],
needs to be introduced to account for the periodicity of r and equation 5.2 is modified as
r (Å) = dsinθ + Asin[2π(No - N)/3.6] = 1.5x(No - N)sinθ + Asin[2π(No - N)/3.6]
where r can be experimentally derived by

13

instance, if the D10 data show that the labeled

(5.3)

2

C- H REDOR using the D10 lipid. For

13

CO nucleus is 5 Å from the membrane

bilayer center, the distance r will be ~11 Å since the hydrophobic thickness of the DPPC
bilayer is ~31 Å[4]. The residue No is the one that produces the most rapid

13

2

C- H

REDOR dephasing buildup in the D4 membrane. To obtain the tilt angle θ, the

164

experimentally derived distance r can be fitted as a function of the position number N of
the

13

CO labeled residue using equation 5.3, where θ and A are both fitting parameters.

Summary of experimental results of KALP and HFP and future work of HFP
For the membrane location studies of KALP and HFP, each peptide was

13

CO

backbone labeled and the lipids were either perdeuterated (D54) or selectively
deuterated (D4, D8, and D10) in their acyl chains.

13

2

C- H REDOR was used to detect

the residue-specific membrane locations of KALP and HFP. The data fitting of
experimental ∆S/So vs τ showed a membrane location distribution for both KALP and
HFP. The multiple locations of KALP are attributed to the hydrophobic mismatch
between the membrane and KALP and consequent flexible snorkeling of lysine
sidechains to the membrane headgroup region[5]. The multiple locations of HFP are
attributed to the distribution of antiparallel β sheet registries in membranes[6]. To our
knowledge, this is the first clear experimental support for a distribution of membrane
locations of a peptide in the membrane hydrocarbon core. For future work, we may
detect the residue-specific membrane locations of HFP in a larger construct (e.g. HFPHairpin including the NHR-loop-CHR region) to see whether there is also a membrane
location distribution, if so, what are the populations of deep and shallow insertion
compared to those for HFP?
In addition to the deuterated lipids, deuterated cholesterol can be also used to
study the membrane locations of peptides. Similar to the D10 lipid, cholesterol
deuterated at the chain end (cholesterol-d6, Figure 5.2) can be used to detect the

165

peptide location with respect to the membrane center. Furthermore, HFP binding
preference to the lipid or cholesterol may be determined by comparison of

13

2

C- H

REDOR data for the D10 and cholesterol-d6 samples. For instance, we can prepare two
membrane samples of HFP_G5C, a and b. For sample a, the membrane consists of 30%
D10 lipid, 40% undeuterated DPPC lipid, and 30% undeuterated cholesterol. For
sample b, the membrane consists of 70% undeuterated DPPC lipid and

Figure 5.2 Structure of cholesterol-26,26,26,27,27,27-d6.
30% cholesterol-d6. After obtaining the experimental

13

2

C- H REDOR ∆S/So buildups of

HFP_G5C in sample a and b, we can fit the plots of ∆S/So vs τ by the function Ax(1-e

-βτ

)

(refer to section 4.4 in Chapter 4 for more details) and compare the best-fit values of A
for sample a and b. If the A value for sample a is comparable to that for sample b (e.g.
65% vs 70%), it supports that HFP has no binding preference to either lipids or
cholesterol. If the A value for sample a is significantly smaller (or larger) than that for
sample b (e.g. 65% vs 85%), it supports that HFP has binding preference to cholesterol

166

(or lipids). To further elucidate the HFP binding preference, we may also prepare four
other membrane samples of HFP_G5C by using 20% D10, 20% cholesterol-d6, 10%
D10, or 10% cholesterol-d6, respectively, and compare their A values.

167

REFERENCES

168

REFERENCES

1.

Ghosh, U., L. Xie, and D.P. Weliky, Detection of closed influenza virus
hemagglutinin fusion peptide structures in membranes by backbone (CO)-C-13N-15 rotational-echo double-resonance solid-state NMR. Journal of Biomolecular
NMR, 2013. 55(2): p. 139-146.

2.

Han, X., et al., Membrane structure and fusion-triggering conformational change
of the fusion domain from influenza hemagglutinin. Nature Structural Biology,
2001. 8(8): p. 715-720.

3.

Lorieau, J.L., J.M. Louis, and A. Bax, Impact of Influenza Hemagglutinin Fusion
Peptide Length and Viral Subtype on Its Structure and Dynamics. Biopolymers,
2013. 99(3): p. 189-195.

4.

Xie, L., et al., Residue-specific membrane location of peptides and proteins using
specifically and extensively deuterated lipids and C-13-H-2 rotational-echo
double-resonance solid-state NMR. Journal of Biomolecular NMR, 2013. 55(1): p.
11-17.

5.

Kandasamy, S.K. and R.G. Larson, Molecular dynamics simulations of model
trans-membrane peptides in lipid bilayers: A systematic investigation of
hydrophobic mismatch. Biophysical Journal, 2006. 90(7): p. 2326-2343.

6.

Schmick, S.D. and D.P. Weliky, Major Antiparallel and Minor Parallel beta Sheet
Populations Detected in the Membrane-Associated Human Immunodeficiency
Virus Fusion Peptide. Biochemistry, 2010. 49(50): p. 10623-10635.

169

APPENDICES

170

APPENDIX A

NMR file locations

171

Figure 2.6
/export/home/hapi0/mb4c/data/Ujjayini/IFP_061312
Figure 2.9
/export/home/khafre0/mb4b/data/Ujjayini/Data_112112/IFP_020413_4
Figure 3.4
/export/home/hapi0/mb4c/data/Li/13C2H/setup_121912/MAS_KBr_122012
Figure 3.5
/export/home/hapi0/mb4c/data/Li/13C2H/setup_121912/adamantane_122012
Figure 3.6
/export/home/hapi0/mb4c/data/Li/13C2H/setup_121912/cpr_aHrf_122012
Figure 3.7
/export/home/hapi0/mb4c/data/Li/13C2H/setup_121912/i4_optimizedCP_122012
Figure 3.9
/export/home/hapi0/mb4c/data/Li/13C2H/i4/i4_071713
Figure 3.10
/export/home/hapi0/xieli4c/data/Li/REDOR/13C2H/i4_031112 (for "redorxy8xypi_pm"
data)
/export/home/hapi0/xieli4c/data/Li/REDOR/13C2H/i4_redorxy8ypi_031912 (for
"redorxy8ypi_pm" data)
Figure 3.11
/export/home/hapi0/xieli4c/data/Li/REDOR/13C2H/i4_031112 (for "redorxy8xypi_pm"
spectra)
/export/home/hapi0/xieli4c/data/Li/REDOR/13C2H/i4_redorxy4xpi_031912 (for
"redorxy4xpi_pm" spectra)
172

Figure 4.4
/export/home/hapi0/mb4c/data/Li/13C2H/KALP/KALP_092412 (KALP_A11C in D4)
/export/home/hapi0/mb4c/data/Li/13C2H/KALP/KALP_032713 (KALP_A11C in D8)
/export/home/hapi0/mb4c/data/Li/13C2H/KALP/KALP_122112 (KALP_A11C in D10)
/export/home/hapi0/mb4c/data/Li/13C2H/KALP/KALP_012213 (KALP_A11C in D54)
Figure 4.5
/export/home/hapi0/mb4c/data/Li/13C2H/KALP/KALP_092412 (KALP in D4)
Figure 4.6
the same file locations as those for Figure 4.4.
Figure 4.7
/export/home/hapi0/mb4c/data/Li/13C2H/HFP/wt_F8C_120712 (HFP_F8C in D4)
/export/home/hapi0/mb4c/data/Li/13C2H/HFP/wt_F8C_112912 (HFP_F8C in D8)
/export/home/hapi0/mb4c/data/Li/13C2H/HFP/wt_F8C_110112 (HFP_F8C in D10)
/export/home/hapi0/mb4c/data/Li/13C2H/HFP/wt_F8C_112512 (HFP_F8C in D54)
Figure 4.8
the same file locations as those for Figure 4.7.
Figure 4.9
/export/home/hapi0/mb4c/data/Lihui/13C2H/HFP/HFP_050813 (HFP_G5C in D4)
/export/home/hapi0/mb4c/data/Lihui/13C2H/HFP/HFP_090313 (HFP_G5C in 80% D8
and 20% DTPG)
/export/home/hapi0/mb4c/data/Lihui/13C2H/HFP/HFP_053113 (HFP_G5C in 80% D10
and 20% DTPG)
Figure 4.10

173

the same file locations as those for Figure 4.9.
Figure 4.12
/export/home/hapi0/mb4c/data/Li/t1D_ir/Lipids/D8_101713/D8_array_110613
Figure 4.13
/export/home/hapi0/mb4c/data/Li/t1D_ir/Lipids/D10_101713/D10_tau0.1_101713
/export/home/hapi0/mb4c/data/Li/t1D_ir/Lipids/D10_101713/D10_tau200_101713
/export/home/hapi0/mb4c/data/Li/t1D_ir/Lipids/D10_101713/D10_array_101713
Figure 4.14
/export/home/hapi0/mb4c/data/Li/t1D_ir/Lipids/D54_102113/D54_tau0.1_102113
/export/home/hapi0/mb4c/data/Li/t1D_ir/Lipids/D54_102113/D54_array_102113
Figure 4.15
/export/home/hapi0/mb4c/data/Li/t1D_ir/Lipids/D8_101713/D8_array_MAS_103013
Figure 4.16
/export/home/hapi0/mb4c/data/Li/t1D_ir/Lipids/D10_101713/D10_array_MAS_102813
Figure 4.17
/export/home/hapi0/mb4c/data/Li/t1D_ir/Lipids/D54_102113/D54_array_MAS_102513
Figure 4.23
/export/home/hapi0/mb4c/data/Shuang/13C2H/IFP/IFP_091113 (IFP in D8)
/export/home/hapi0/mb4c/data/Shuang/13C2H/IFP/IFP_040213 (IFP in D10)
Figure 4.24
/export/home/hapi0/mb4c/data/Lihui/13C2H/HFP/HFP_010214 (L9R mutant of
HFP_G5C in 80% D4 and 20% DPPG)

174

/export/home/hapi0/mb4c/data/Lihui/13C2H/HFP/HFP_011414 (L9R mutant of
HFP_G5C in 80% D8 and 20% DPPG)
/export/home/hapi0/mb4c/data/Lihui/13C2H/HFP/HFP_012714 (L9R mutant of
HFP_G5C in 80% D10 and 20% DPPG)
Figure 4.25
/export/home/hapi0/mb4c/data/Li/quecho/DPPC/D4_112112 (KALP in D4)
/export/home/hapi0/mb4c/data/Li/t1D_ir/Lipids/D8_101713/D8_quecho_102013 (HFP in
D8)
/export/home/hapi0/mb4c/data/Li/quecho/DPPC/D10_112112 (HFP in D10)
/export/home/hapi0/mb4c/data/Li/t1D_ir/Lipids/D54_102113/D54_quecho_102513 (HFP
in D54)

175

APPENDIX B

Fmoc solid phase peptide synthesis (SPPS)

176

Solid phase peptide synthesis (SPPS) is a peptide synthesis technique first
developed by Robert Merrifield[1]. There are two commonly used SPPS protocols,
Fmoc and t-Boc SPPS. Compared to t-Boc SPPS which uses TFA as the deprotecting
reagent, Fmoc SPPS is milder and safer since it uses piperidine as the deprotecting
reagent.
For the Fmoc SPPS protocol, each time we may start the peptide synthesis using
either Fmoc (9-fluorenylmethyloxycarbonyl) Wang resin which produces peptides with
free C-terminal carboxylic group or rink amide resin which produces peptides with
amidated C-terminus. Take the synthesis of 0.2 mmol HFP which has a sequence of
AVGIGALFLGFLGAAGSTMGARSWKKKKKKA as an example, if we start with FmocAla-resin that has a loading capacity of 0.5 mmol/g, we will need to weigh out 0.4 g
resin. The Fmoc SPPS procedure for HFP synthesis is described as follows:
(1) Swell the 0.4 g Fmoc-Ala-resin in dichloromethane (DCM) in a reaction vessel for 2
hours on a vortex mixer (Vortex-Genie 2), drain the DCM, swell the resin in DCM for
another 2 hours, drain the DCM, and then add N,N-dimethylformamide (DMF) to rinse
the resin twice.
(2) Add ~10 mL deprotecting solution to remove the Fmoc protecting group from the
residue for 5 minutes, drain the solution, add DMF to rinse the resin once, add another
~10 mL deprotecting solution to remove residual Fmoc for 20 minutes, and then drain
the solution. The deprotecting solution is prepared using piperidine and DMF in a
volume ratio of 1:4. For instance, a 500 mL deprotecting solution contains 100 mL
piperidine and 400 mL DMF.

177

(3) Rinse the resin using DCM once and then using DMF twice. After that, add ~8 mL
freshly-prepared coupling solution which contains 2 mmol Fmoc-Lys(Boc)-OH, where
the lysine sidechain is protected by the tert-butyloxycarbonyl (t-Boc) protecting group.
To increase the peptide yield, the amount of Fmoc-amino acid added is 10 times the
stoichiometry. For isotopically labeled Fmoc-amino acid (e.g.

13

CO labeled), the amount

added is 5 times the stoichiometry. The coupling solution is prepared using the following
materials:
(a) 2 mmol Fmoc-amino acid;
(b) 2 mmol 1-hydroxybenzotriazole (HOBt), the mass is 270 mg;
(c) 2 mmol 2-(1H-Benzotriazole-1-yl)-1,1,3,3-tetramethyluronium hexafluorophosphate
(HBTU), the mass is 759 mg;
(d) 700 µL N,N-diisopropylethylamine (DIEA);
(e) add DMF until the total volume is 8~10 mL.
(4) Allow at least 4 hours for the coupling reaction. In general, the minimum time
required for coupling reaction is 2 hours for every Fmoc-amino acid except for Cys, Ser,
Thr, Met, Trp, and isotopically labeled (e.g.

13

CO or

15

N labeled) amino acids which

require a minimum coupling time of 4 hours. For the first coupling reaction, no matter
what Fmoc-amino acid is added, allow at least 4 hours to increase the yield. After the
coupling reaction, drain the coupling solution, rinse the resin using DCM once and then
using DMF twice, finally add ~10 mL capping solution to cap the N-terminus of any
unreacted peptide to which the added Fmoc-amino acid is not coupled. The capping

178

reaction prevents the unreacted peptide from reacting in the next coupling reaction. For
200 mL capping solution, it consists of
(a) 9.5 mL acetic anhydride;
(b) 4.5 mL DIEA;
(c) 0.4 g HOBt;
(d) add DMF until the total volume is 200 mL.
(5) Allow 5~10 minutes for the capping reaction, drain the capping solution, and then
rinse the resin using DCM once and DMF twice.
(6) Add ~10 mL deprotecting solution and vortex for 5 minutes, drain the solution, rinse
the resin with DMF once and then add another ~10 mL deprotecting solution and vortex
for 20 minutes. In the meantime, prepare the coupling solution.
(7) Drain the deprotecting solution, rinse the resin using DCM once and then using DMF
twice. Add ~8 mL freshly-prepared coupling solution which contains 2 mmol of the
second residue to be coupled, i.e. Fmo-Lys(Boc)-OH. The coupling solution should be
prepared in the same way as described in step (3).
(8) Repeat steps 4-7 until we get the whole sequence. For the last residue, after the
coupling reaction, perform capping reaction once and then deprotection twice. After that,
perform capping reaction again if the N-terminus of the peptide needs to be acetylated,
otherwise, proceed to rinse the resin using DCM 3~4 times, transfer the resin from the
reaction vessel to the cleavage tube (more than one tube may be needed, depending
on the tube capacity). Use DCM to wash all resin down to the cleavage tube from the
reaction vessel. After that, use gentle N2 gas to remove DCM from the cleavage tube.
179

Be careful not to blow away the resin. Keep the resin in the cleavage tube and store the
tube in a vacuum desiccator overnight to further remove any residual DCM and/or other
organic solvents including DMF.
(9) Add 12 mL cleavage solution to the resin in the cleavage tube(s) (if the resin is split
into 2 or 3 tubes, then add 6 or 4 mL to each) and gently vortex for ~4 hours. If the
cleavage time is too short, incomplete cleavage may result. However, if the cleavage
time is too long, oxidation or other side reactions of the deprotected sidechains may
result. To cleave 0.2 mmol peptide from the resin, 12 mL cleavage solution needs to
be prepared using the following materials:
(a) 600 µL distilled water;
(b) 240 µL triisopropylsilane(TIS);
(c) 240 µL thioanisole;
(d) 240 µL 1,2-ethanedithol;
(e) add trifluoroacetic acid (TFA) until the total volume is 12 mL.
TFA cleaves the peptide from the resin and also removes all sidechain protecting
groups from the peptide. For components a-d, they are called “scavengers” and are
used to protect the peptide from oxidation and other side reactions after all sidechain
protecting groups of the peptide have been removed by TFA. The cleavage solution is
sometimes also called "cocktail solution".
(10) After the cleavage reaction, drain the 12 mL cleavage solution to three 50 mL
conical vials (4 mL for each), add 1~2 mL TFA to wash the resin and then drain the
TFA to the vials.

180

o

(11) Add ~40 mL cold diethyl ether (4 C) to each vial to extract the peptide from the
cleavage solution. A 10:1 volume ratio of cold diethyl ether to cleavage solution is used
to maximize the peptide extraction. After adding the ether, we will observe white
precipitate immediately, i.e. the peptide precipitates out due to a poor solubility in cold
diethyl ether. During the extraction, do not shake the peptide-ether mixture in case of
peptide oxidation (e.g. methionine may get oxidized by O2 or oxidized ether). After the
extraction, centrifuge the peptide-ether mixture at a speed of 4000 rpm (revolutions per
minute) or above for ~5 minutes. After that, dump the supernatant, add ~10 mL fresh
cold diethyl ether to the peptide pellet, perform ~2 min sonication to help dissolve the
ether-soluble impurities, centrifuge for another 5 minutes, dump the supernatant, and
finally put the vials in the hood to let the peptide pellet air dry.
(12) Add 10~20 mL distilled water containing 0.1% (by volume) TFA to each vial to
dissolve the peptide pellet. Again, do not shake the vial in case of peptide oxidation. It
may take a few hours to dissolve the peptide pellet. If the pellet does not get dissolved
after several hours or even a day, add a few mL acetonitrile containing 0.1% TFA to
help dissolve the peptide.
(13) After the peptide has been completely dissolved, perform reversed-phase HPLC
using a C4 or C18 column to purify the peptide.
(14) At last, use MALDI or ESI mass spectrometry to identify the peptide purity and
molecular weight.

181

The following is a list of the mass of 2 mmol Fmoc-amino acid (the mass takes
into account the sidechain protecting group if applicable). The Fmoc-amino acids are
commercially available from either “Peptides International” or “Novabiochem” company.
Lys (K) : 937.2 mg
Trp (W) : 1053.2 mg
Ser (S) : 794.8 mg
Arg (R) : 1297.6 mg
.

Ala (A) : 658.8 mg (monohydrate, i.e. Fmoc-Ala-OH H2O)
Gly (G) : 594.8 mg
Met (M) : 742.8 mg
Thr (T) : 1139.3 mg
Leu (L) : 706.8 mg
Ile (I) : 706.8 mg
Val (V) : 678.8 mg
Phe (F) : 774.9 mg
Glu (E) : 851.0 mg

182

APPENDIX C

HPLC purification of peptides

183

High-performance liquid chromatography (HPLC, sometimes referred to as highpressure liquid chromatography) is a technique used to separate different components
in a mixture and quantify each component. There are two types of HPLC. One is
referred to as normal phase HPLC for which the stationary phase (e.g. unmodified silica
resins) is polar and the mobile phase is non-polar. The other is referred to as reversedphase HPLC for which the stationary phase is non-polar and the mobile phase is polar.
Examples of non-polar stationary phase include C4 and C18 columns where alkyl
chains containing four or eighteen carbons are covalently bonded to the resins to create
a hydrophobic stationary phase. For reversed-phase HPLC, hydrophobic molecules in
the mobile phase tend to bind to the column whereas hydrophilic molecules do not bind
to the column and are eluted first.
In our work, the peptides synthesized by Fmoc SPPS were purified by reversedphase HPLC using a C4 column. The C4 column is filled with resins (5 µm in diameter)
to which butyl groups are covalently bonded. The inner diameter of the column is 19
mm and the length is 300 mm. To purify the peptides, two solvents, A and B, are used
as the mobile phase. Solvent A is degased distilled water containing 0.1% TFA and
solvent B is the mixture of 10% degased distilled water and 90% acetonitrile containing
0.1% TFA. Degased distilled water is used to avoid air bubbles in the pump system. 0.1%
TFA is added to each solvent to increase the peptide solubility in the mobile phase.
Different peptide samples may require different HPLC gradient programs for purification.
An ideal HPLC gradient program should have the following criteria:
(a) The product peak should be well resolved from other peaks. If there is severe peak
overlapping, a slower gradient buildup rate can be used to solve this issue.
184

(b) The product peak should not be too far from the nearby peaks. For instance, if the
product peak is 10 mins apart from its nearest peak, it will be a waste of time to wait
such a long period. If this issue appears, a faster gradient buildup rate can be used to
solve it.
(c) After the product peak ends, the gradient can be increased to the maximum % B (e.g.
75% or 100% B) to elute all residual components from the column. It is unnecessary to
wait a long period (e.g. 10 mins) following the product peak to increase the gradient to
the maximum % B.
(d) After the gradient goes back to the initial % B from the maximum % B, allow 3~5
mins to equilibrate the column.
Take HFP peptide purification using a C4 column for example, the time sequence
of the HPLC gradient program used was as follows:
t = 0.0 min, % B = 37.0%;
t = 2.0 min, % B = 37.0%;
t = 6.0 min, % B = 38.0%;
t = 11.0 min, % B = 38.0%;
t = 20.0 min, % B = 42.5%;
t = 20.5 min, % B = 75.0%;
t = 29.5 min, % B = 75.0%;
t = 30.0 min, % B = 37.0%;
t = 33.0 min, % B = 37.0%.

185

where t was the elution time (sometimes also referred to as retention time) and the flow
rate was 8.0 mL/min. This gradient program is displayed in Figure C1 (blue dotted line).
The HPLC chromatogram of "Absorbance at 214 nm vs Retention time" in Figure C1
(black solid line) was obtained from the purification of HFP which had the sequence of
AVGIGALFLGFLGAAGSTMGARSWKKKKKKA. The UV detection wavelength was
chosen as 214 nm since the peptide amide bonds had strong absorbance at this
wavelength. The wavelength of 280 nm may be also used since tryptophan has strong
absorbance at 280 nm and the corresponding molar absorptivity (also referred to as
-1

-1

extinction coefficient) is ~5600 M cm . In the HPLC chromatogram, the first peak
appeared at t ≈ 6.5 min and corresponded to the unretained components which were
eluted directly with the mobile phase from the column. The time t ≈ 6.5 min (referred to
as dead time) was the time required for the mobile phase to pass through the column.
For the HFP peak, it appeared at t ≈ 19.0 min. Since it took ~6.5 min to pass through
the column to reach the detector, the actual time when HFP was eluted from the column
should be ~12.5 min. According to the linear gradient program described above, % B ≈
38.8 for t ≈ 12.5 min, thus the minimum % B for HFP elution from the column was ~38.8.
For the last peak in the chromatogram, which corresponded to the most hydrophobic
components (e.g. peptide with uncleaved Fmoc group at its N-terminus), it appeared at t
≈ 27.0 min. Given that the dead time was ~6.5 min, the actual elution time of the last
peak was ~20.5 min, which was the time when % B was increased to 75.0%.

186

Figure C1 HPLC chromatogram of HFP purification.
The peak centered at t ≈ 20.0 min in Figure C1 was diagnostic of HFP based on
the MALDI mass spectrum. MALDI is a soft ionization source used in mass
spectrometry. Examples of commonly used matrices for MALDI include 2,5dihydroxybenzoic acid (DHB) and α-cyano-4-hydroxycinnamic acid (CHCA). The
matrices can heavily absorb UV laser light at specific wavelengths (e.g. 337 and 355 nm)
and cause sample desorption, during which a hot plume is produced and the analyte
molecules (e.g. peptides) are ionized (more accurately protonated) in the hot plume[2].
+

For the positive-ion mode of MALDI, it is the protonated species M+H that is detected.
+

+

However, sometimes other species such as M+Na and M+K may be also detected.
The MALDI mass spectrum in Figure C2 shows a predominant peak with m/z = 3148.8,
which

corresponds

to

HFP

having

a

sequence

of

AVGIGALFLGFLGAAGSTMGARSWKKKKKKA and a theoretical molecular weight of
+

3150.8 Da. The peak with m/z = 3171.1 is diagnostic of HFP binding to a Na ion
instead of a proton.
187

Figure C2 MALDI mass spectrum of HFP after purification. The theoretical molecular
weight of HFP is 3150.8 Da.

188

APPENDIX D

NMR troubleshooting

189

1. Spinning problems
(a) If the rotor cannot spin at all or even cannot completely insert in the stator (also
referred to as spinning module), check the alignment of the rotor, front bearing, and
bottom bearing. If the rotor can only insert halfway in the stator, it is either the solenoid
coil is twisted or the bottom bearing is not aligned with the coil. To find out which one is
the case, we need to take apart the stator. If the coil is severely twisted, a new coil is
needed. If the coil is fine but the bottom bearing is not aligned with the coil, we need to
first heat up the leg of the coil using a tip heater and then adjust the leg to realign the
coil and bottom bearing until the rotor can insert in the stator completely and freely.
(b) If the rotor can spin in the stator but the MAS speed reading is not stable in the
"Auto" mode, first switch to the "Manual" mode, if the MAS controller can read the
spinning speed stably, it means there is a communication issue rather than a rotor or
stator issue. To solve the communication issue, we can restart the MAS controller and
reopen the accessory control panel (i.e. ACC panel) on the host computer. If the speed
reading is not stable in the "Manual" mode either, there are several possibilities
described as follows:
(1) The sample is not tightly packed in the rotor. If this is the case, repack the sample
and use bottom and top spacers with proper lengths.
(2) The black marking on the rotor is not clear. If this is the case, remark the rotor.
(3) The drive tip of the rotor is partially damaged. If this is the case, use a new drive tip.
(4) The front or bottom bearing in the stator will get scratched and its inner diameter will
become bigger after long-time use, which may cause unbalancing of the rotor while
spinning. If this is the case, replace the front or bottom bearing with a new one.
190

(5) The tiny holes at the bottom of the stator may get clogged, in which case the drive
gas will be partially blocked and thus unstable rotor spinning will result. To solve this
issue, the holes at the stator bottom need to be cleaned either manually or by
ultrasonication.
(6) The tip of the optical fiber is damaged or improperly positioned. If this is the case,
cut the tip of the optical fiber or reposition it until a stable speed reading is obtained.
(7) The bearing gas parameters in the ACC panel are not properly set. The typical
values for a 4 mm MAS probe are given in Figure D1.

Figure D1 Typical bearing parameters in the ACC panel for a 4 mm MAS probe.
2. Tuning problems
(a) The channel is not tunable for the low-power tuning.

191

If a channel is untunable for the low-power tuning, i.e. no resonance is observed
at the targeted frequency on the oscilloscope, it is definitely untunable for the highpower tuning (i.e. tuning in experiments where high-power r.f. pulses are applied). If a
resonance is observed near the targeted frequency but not at that frequency no matter
how we adjust the broad tuning rod and match, there are several parts including the
series plug-in, trap plug-in, tuning tube, and tuning rod that we may change to move the
resonance to the targeted frequency. For instance, if the targeted frequency is 100 MHz
while the resonance only appears in the frequency range of 90-95 MHz, a series or trap
plug-in with a lower capacitance may be used to move the resonance to 100 MHz. If no
resonance is observed either at or near the targeted frequency, it is likely that there is a
contact problem in that channel. In this case, we need to check whether some
connection is loose, for example, the copper ribbon connecting the solenoid coil in the
1

stator and the H (or X) channel may not be soldered properly and thus resoldering may
1

be needed, probably with a new copper ribbon. For the H channel, there is a little
1

screw on the match rod (Figure D2), if this screw is loose or lost, the H channel will
lose its resonance and become untunable.
(b) The channel is tunable for the low-power tuning but not for the high-power tuning.
If a channel is tunable at the transmitter frequency for the low-power tuning but
not for the high-power tuning, there are usually two possibilities described as follows:
(1) The tuning rod is way off during the high-power tuning. When the tuning rod is way

192

1

1

Figure D2 Little screw (red circled) on the H match that plays a key role for H tuning.
off the correct position, in which case the probe resonance frequency is way off the
transmitter frequency of the r.f. pulse, the transmitter frequency will be beyond the
tunable frequency range achieved by adjusting the match. Therefore, the reflected
voltage (also referred to as reverse voltage) of the r.f. pulse will not go down
significantly no matter how we adjust the match while the tuning rod is way off. In this
case, we can switch back to the low-power tuning and first adjust the tuning rod to move
the resonance back to the transmitter frequency and then adjust the match until the
resonance peak at the transmitter frequency becomes as sharp as possible, in which
case the reflected voltage is close to zero. After that, do the high-power tuning again
and the probe should be tunable now.
193

(2) The probe is arcing. When too much r.f. pulse power is input to the probe, the
significant heating may induce carbonization at or near the connection spots and cause
probe arcing. When severe arcing occurs, sparkle may be observed at the arcing spot
and huge spikes may be observed in the FID as well. Figure D3 displays the REDOR
So and S1 FIDs (panel a) in the presence of severe arcing at the carbonized rim near
one leg of the solenoid coil (panel b). In this case, the rim needs to be cleaned
thoroughly to remove all black carbon. After that, the probe may become tunable again.

Figure D3 (a) REDOR So and S1 FIDs in the presence of severe probe arcing; (b)
Stator showing the carbonized rim near one leg of the solenoid coil where the arcing
occurs.
194

REFERENCES

195

REFERENCES

1.

Merrifield, R.B., SOLID PHASE PEPTIDE SYNTHESIS .1. SYNTHESIS OF A
TETRAPEPTIDE. Journal of the American Chemical Society, 1963. 85(14): p.
2149-&.

2.

http://en.wikipedia.org/wiki/Matrix-assisted_laser_desorption/ionization.

196