in!
. a. b} .‘
".3”: l Tum.
an?) Rafi;
‘ 1.er
, .mwxhafiw

1.31.

. 3. pk
f.§$:$?f. .
115.5. :51.

an.

3
an“:
a.-.
5..

{nil-u...
c .12".
arhbiaiz I:
\11tx1r 1
2.. I\ 1..
.: ‘..R v I ,
. ix . .
4.. €333.41“? 15:1.
3:556 2: 31.
5:1. .1393:
. 2 .5!!!
.3. v... I

2‘1

11

A11.

_ gun...
a
gammy“...
a. .. .2...
.. .t. E.
.‘ Inn-1':
3%.“.stth
a... ! ii.
$3...

lthi

halffk
tialilk
I:
i

 

 

1000

 

 

LIBRARY
Michigan State
University

 

 

This is to certify that the
dissertation entitled

SECONDARY STRUCTURE AND MEMBRANE INSERTION OF
THE MEMBRANE-ASSOCIATED INFLUENZA FUSION PEPTIDE
PROBED BY SOLID-STATE NUCLEAR MAGNETIC RESONANCE

PhD.

 

presented by
Yan Sun

has been accepted towards fulfillment
of the requirements for the

degree in Department of Chemistry

[magi i2 Laid/Lg

 

 

Major Professor’s Signattlre

9M ///L Z/fflq
I/

Date

MSU is an Affirmative Action/Equal Opportunity Employer

Il-O-O-l-O:I-l-0-Q-O-D-I-I'D-0-.-.-I-.-'-O-I-O-I-O-O-U-O-I-O-.-I-I-I-l-D-I-O-I-C-O-O-O-I-'-l-I-I-O-—A“-h-W

 

PLACE IN RETURN BOX to remove this checkout from your record.
TO AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DAIEDUE

DATEDUE

DATEDUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

5/08 K:IProj/Acc&Pres/ClRC/DateDm.indd

SECONDARY STRUCTURE AND MEMBRANE INSERTION OF THE
MEMBRANE-ASSOCIATED INFLUENZA FUSION PEPTIDE PROBED BY
SOLID-STATE NUCLEAR MAGNETIC RESONANCE
By

Yan Sun

A DISSERTATION
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of
DOCTOR OF PHILOSOPHY
Chemistry

2009

ABSTRACT
SECONDARY STRUCTURE AND MEMBRANE INSERTION OF THE
MEMBRANE-ASSOCIATED INFLUENZA FUSION PEPTIDE PROBED BY
SOLID-STATE NUCLEAR MAGNETIC RESONANCE
By
Yan Sun

An initial step in infection by the influenza virus is joining or “fusion” of the
membrane of the virus with the membrane of the endosome of the host cell and
consequent release of the viral nucleocapsid into the host cell cytoplasm. Fusion
is mediated by the influenza viral hemagglutinin protein (HA) which is activated
by the low pH of the endosome. The ~20 N-tenninal amino acids of the HA2
domain of HA are known as the ‘fusion peptide’ (IFP) which binds to the
endosomal membrane and plays a critical role in fusion catalysis. The chemically
synthesized IFP induces vesicle fusion in a pH-dependent manner and is a
model system for understanding some aspects of viral fusion.

In this work, solid-state NMR was used to probe the structure of
membrane-associated IFP and its correlation with fusogenic function. It was
observed that IFP has predominant helical conformation in membranes lacking
cholesterol and B strand conformation in membranes that contained cholesterol.
For either membrane composition, the overall IFP conformation has little
dependence on pH. Low pH triggered IFP- induced fusion between vesicles that
did not contain cholesterol or between vesicles that contained cholesterol. The

combination of structural and functional data suggested that both the helical and

the B strand conformations of IFP could induce vesicle fusion.

ln membranes which lacked cholesterol, IFP was determined to have the
helix-tum-helix motif with the turn formed around Glu-11 and/or Asn-12. The N-
terminal helix extends from Leu-2 to lle-1O or from Leu—2 to GIu-11. The first
conformation was observed at both pHs and the latter one was only observed at
pH 5.0. The 13CO—31P and 13CO—19F distance measurements by the REDOR
technique indicated that both the IFP N- and C-termini have close contact to the
lipid phosphate headgroups and the middle region of IFP is inserted into a single
leaflet of the membrane. Based on these data, we proposed an inverted
boomerang IFP structural model in membranes without cholesterol. More
contacts between IFP labeled 13COS and the membrane bilayer center were
detected for membrane-associated IFP at pH 5.0 which may be correlated to the
more deeply inserted or more population of inserted IFP at fusogenic pH 5.0
relative to the non-fusogenic pH 7.4.

In addition, static 15N chemical shifts and 15N-‘H dipolar couplings were
used to probe the tilt angle of the IFP N-terminal helix relative to the bicelle
normal. The data were well fitted with a model of the fast rotation of the IFP N-
tenninal helix about its own axis with a tilt angle of ~45° from the bicelle normal.
The motion and orientation of the IFP N-terminal helix are intrinsic properties of
the IFP sequence and are independent of the sample pH and the mutation of

Gly-1 to Ser or Val.

Dedicated to Wei Qiang

iv

ACKNOWLEDGEMENTS

It is a great pleasure to express my gratitude to all the people who have
helped me to complete the dissertation.

First and foremost, I would like to sincerely thank Prof. David Weliky for
his guidance, understanding, encouragement and incredible patience. He has
been a very supportive advisor and provided me lots of good ideas throughout
my graduate career, yet he has always given me great freedom to do research
which allowed me to grow into an independent scientist. He always has faith in
my ability and he’s willing to share his own tough or happy experience with me to
encourage me to overcome difficulties and believe in myself. When he reviewed
my first paper, he asked me to sit at his side and discussed with me how to
revise each paragraph. That was the time when I tremendously improved my
scientific writing ability and start to think and write as a reviewer.

I’d also like to thank all my committee members, Prof. John McCracken,
Prof. Gavin Reid and Prof. James Geiger, for their input, valuable discussions
and guidance over the years. My gratitude also goes to Prof. Michael Feig, who
is generous with his time and insights, both on my research and on my career.
He has been a good source in discussions about my research project and his
ideas greatly inspired me.

I am grateful for the help from the Max T. Rogers NMR facility, MSU Mass
Spectrometer Facility and the CTA Biomolecular NMR Facility. I especially
appreciate the expertise and patience of Mr. Kermit Johnson and Dr. Daniel

Holmes. I would like to acknowledge and thank Dr. Jiadi Xu and Dr. Jochem

Struppe for the help they gave me on the 900 MHz Bruker spectrometer. Dr. Jiadi
Xu kindly assisted me in the static 20 NMR experiments and generously passed
me his own experience with NMR hardware and software. I also extend my
gratitude to Afra Panahi for her warm heart and the useful discussions on
research.

I would like to take the opportunity to thank the former and current
members in Prof. David Weliky’s group: Dr. Charles Gabrys, Dr. Kelly Sackett, Dr.
Jaime Curtis-Fisk, Matthew Nethercott, Erica Schwander, Scott Schmick, Jacob
Billcheck, Dr. Zhaoxiong Zheng, Dr. Michele Bodner and Angela Karst. They are
a wonderful group of people to work with. I am particularly thankful to Dr.
Zhaoxiong Zheng, not only being a great senior to learn NMR from, but also
being a good friend who always provides me good advices in so many ways.

Finally, and most importantly, I would like to thank my husband Wei Qiang,
as the dearest friend and an amazingly supporting colleague. It is his support,
encouragement and unwavering love which have kept me walking all the way up
to where I am today. I thank my parents for their love and indulgence, allowing
me to pursuit anything I want. Also, I thank Qiang’s parents, who have provided
me unending support and encouragement and showed me the true value of hard

working.

TABLE OF CONTENTS

LIST OF TABLES ................................................................................. ix
LIST OF FIGURES .............................................................................. x
LIST OF ABBREVIATIONS ................................................................... xx
Chapter1 Introduction ..................................................................... 1
1.1 Background .................................................................. 1

1. Structural Biology of Viral Fusion Proteins ........................ 2

2. Membrane bilayers as model biomembrane systems ......... 10

3. Solid-State NMR measurements of membrane-bound FP...13

1 .2 References .................................................................. 1 7

Chapter2 Materials and Methods ...................................................... 23
2.1 Materials ..................................................................... 23

2.2 Sample preparation ...................................................... 24

1. IFP and HFP Synthesis ............................................. 24

Chapter 3

2. Membrane sample preparation for MAS solid state NMR...27
3. Membrane sample preparation for static solid state NMR...28

4. N-acetylvaline Sample Preparation .............................. 30
2.3 MAS solid-state NMR ................................................... 30
1. General MAS NMR spectroscopy ................................. 30

2. CP MAS spectroscopy .............................................. 31

3. REDOR spectroscopy ................................................ 32

4. prTDQBU spectroscopy ............................................. 35

5. PDSD spectroscopy ................................................... 37

6. Experimental data analysis .......................................... 38
2.4 Static solid-state NMR ................................................... 41
1. General static NMR Spectroscopy ................................. 41

2. 2H and 31P spectroscopy ............................................. 42

3. 15N chemical shift measurements ................................. 44

4. 15N chemical shift & 15N—1H dipolar coupling correlation
spectroscopy ............................................................. 45

5. Static 15N NMR analysis ........................................... 46
2.5 References .................................................................. 53
Conformational Studies of Membrane-associated IFP ............... 56
3.1 Background ............................................................... 56
3.2 Results ..................................................................... 59
1. 13CO Chemical Shift ................................................... 59

2. Ala-7 13CO—GIy—8 13CO distance measurements ............... 65

3. REDOR Experiments ................................................ 68

4. IFP mutants ............................................................ 74

5. PDSD Experiments ..................................................... 77

3.3 Discussion .................................................................. 93

3.4 References .................................................................. 99

Chapter 4 Studies of Influenza Fusion Peptide Membrane Location ......... 104
4.1 Background ............................................................ 104

4.2 Results ..................................................................... 106

1. Static 3‘P spectra ...................................................... 106

2. Membrane location of pH 5.0 samples ........................... 108

3. Membrane location of pH 7.4 samples ........................... 118

4.3 Discussion ............................................................... 128

4.4 References ............................................................... 133

Chapters Solid-State NMR Studies of the IFP N-terminal Helix in Oriented
Lipid Bilayers .................................................................. 135

5.1 Background ............................................................... 135

5.2 Results ..................................................................... 142

1. Bicelle alignment ...................................................... 142

2. 1D experiment ......................................................... 144

3. 20 experiment ......................................................... 155

4. Comparison between 9.4 T and 21.4 T data ................... 160

5. Comparison of the N-terminal helix orientation between IFP

and its mutants ........................................................ 163

6. Comparison of the N-terminal helix orientation between IFP

and HFP ................................................................ 167

5.3 Discussion ............................................................... 168

5.4 References ............................................................... 175

Chapter 6 Summary and Future Directions ........................................... 178
Appendix I Natural Abundance Corrections for (AS/So)” ......................... 184
Appendix II Some Additional Data for IFP .............................................. 191
Appendix III Alternative way of fitting 13c-3‘P and ‘30-‘9F data ................... 201
Appendix IV FMOC protection of amino acids and IFP synthesis ............... 205
Appendix V Location of NMR data ...................................................... 208

viii

LIST OF TABLES
Table 1. Labeling scheme of IFPS and HFPs ........................................... 25
Table 2. Peak 13CO chemical Shifts for membrane-associated IFP .................. 63

Table 3. Ala-7 1300 - GIy-8 13CO dipolar couplings and distances and Ala-8
dihedral angle (a in membrane-associated IFP at pH 5.0 ............................... 66

Table 4. 13CO""15N dipolar couplings and distances in membrane-associated IFP
with comparison to distances in detergent-associated IFP ............................ 69

Table 5. Measured 130 chemical shifts for lFP-I10E11U and IFP-N12G13u
samples compared to the corresponding chemical shifts from the Ref-DB ....... 79

Table 6. 13‘C chemical Shifts in ppm for the membrane-associated IFP at pH 5.0
with the first 10 N-terminal residues uniformly labeled ................................. 83

Table 7. go and III angles in degrees of residue Gly-1 to Gly-20 from membrane-
associated IFP at pH 5.0 and detergent-associated IFP. Standard deviations are

in the parentheses. .............................................................................. 91
Table 8. Peak 13CO chemical shifts in ppm for IFP samples at pH 5.0 ............ 113
Table 9. Best-fit “CO-31F and ‘3CO-‘9F(C16) distances for IFP samples at pH

5.0. ................................................................................................. 118
Table 10. Peak 13CO chemical shifts in ppm for IFP samples at pH 7.4 .......... 119

Table 11. ‘3CO—3‘P and 13CO—‘9F(C16) distances for IFP samples at pH 7.4-127

Table 12. 15N chemical shifts (a) and the corresponding orientational analysis for
the aligned IFP samples ...................................................................... 151

Table 13. N-H dipolar couplings (v) and corresponding orientational analysis for
the aligned IFP samples ....................................................................... 160

Table 14. 15N chemical shifts (O), NH dipolar couplings (v) and corresponding
orientational analysis for the aligned IFP mutant samples ........................... 165

ix

LIST OF FIGURES

Figure 1. HIV Infection: (Left) Model of infection process. (Right) Freeze fracture
electron micrography of a virion ................................................................. 3

Figure 2. The first half of the influenza viral life cycle. The virus enters the cell
by endocytosis. The pH drops in the endosome, initiating conformational
changes of the hemagglutinin protein on the viral membrane surface. These
conformational changes ultimately lead to fusion of the viral and endosomal
membranes and release of the viral contents into the cell ............................... 4

Figure 3. Ribbon diagram of Influenza hemagglutinin glycoprotein at (a) pH 7.4
(PDB ID: 1qu1)1 and (b) pH 5.0 (PDB ID: 1htm)2. The structure at pH 7.4
included residues 23-185 of the HA2 chain. The structure at pH 5.0 Shows
residues 12-16 of HA1 and residues 40-153 of HA2 for one monomer; residues
11—16 of HA1 and residues 40-162 of HA2 for another monomer; and residues
10-17‘ of HA1 and residues 40-162 of HA2 for the third
monomer .............................................................................................. 8

Figure 4. Proposed mechanism for Influenza membrane fusion.3 (a) HA protein
conformation at pH 7.4. (b) Upon exposure to the low pH found in endosomes,
HA forms an extended conformation and the N-terminal fusion peptide inserts
into the target membrane. (c) Several HA proteins are thought to be involved. (d)
Protein refolding begins and the released free energy cause the membranes to
bend to each other. (e) Formation of the hemifusion intermediate in which the
outer leaflets of membranes mix. (f) Protein refolding completes and the fusion
pore is formed. The fusion peptide and the transmembrane domain are in the
same membrane and interact with each other. Only the HA structures in (a) and
(f) have been observed by crystallography but many biological data support other
proposed steps ...................................................................................... 8

Figure 5. Solution NMR showed that IFP in DPC micelles has hellx-tum-(310-helix)
structure at (a) pH 5.0 (PDB ID: 1ibn) and has helix-tum-extended structure at (b)
pH 7.4 (PDB ID: 1ibo). The N-terminal helix has an insertion angle of ~38° and
~23° with respect to the membrane surface at pH 5.0 and pH 7.4,
respectively ......................................................................................... 12

Figure 6. Representation of aligned bicelles with their normals (a) perpendicular
to the extemal magnetic field (Bo) or (b) parallel to Bo due to the addition of
YbCl3 ................................................................................................. 13

Figure 7. Construct of HFP dimer. Two HFP monomers cross linked at Cys to
form dimer .......................................................................................... 26

Figure 8. Pulse sequences (a) Cross-polarization. Magnetization is transferred
from 1H to 13C or 5N during OF. 1H continuous wave (CW) decoupling was

applied during a uisition. (b) 13C REDOR. CP transfers the 1H transverse
magnetization to 1 C. 13C magnetization is dephased (i.e., reduced) by the
dipolar coupling between 13C and 15N, 31F or 19F nuclei mediated by the 11 pulses
per rotor period. 13C chemical shift is refocused by multiple 11
pulses ................................................................................................ 26

Figure 9. prTDQBU pulse sequence. (a) The displayed prTDQBU pulse
sequence included: (1) CP from 1H to 13C; (2) a constant~time (CT) period of (L +
M + N)TR duration where L, M and N were integral multiples of 8, and TR was the
duration of a single rotor period; and (3) acquisition period. Continuous wave 1H
decoupling was applied during the CT and acquisition periods. A pair of back-to-
back 13C TT/2 pulses separated the Up and MTR periods and another pair
separated the MTR and MR periods. (b) Pulses during LTR, MTR and NTR periods.
There was one 17 pulse per rotor cycle. The 130 CP rf phase was 4' + rT/2 and the
phase of each 13C n/2 and 11 pulses is noted above the pulse. For C = y or —y, the
evolution of the transverse 13C magnetization due to 130-130 dipolar coupling was
averaged to zero and 80 data were obtained and for 4' = x or -x, the evolution
was not averaged to zero and 81 data were obtained. The duration of 1‘°’C
recoupling or dephasing period T = 2LrR, and data with increasing r were
obtained by incrementing L and decrementing N while keep constant M and
constant (L + M + N) ............................................................................. 36

Figure 10. PDSD pulse sequence. Magnetization was transferred from 1H to 130
during the CP period. CW decoupling was applied after CP on the 1H during t1
and t2, but not during 1. t1 was the evolution time, T was the magnetizatiOn
exchange time, and t2 was the acquisition time ........................................... 43

Figure 11. Pl-WlM-z pulse sequence. After the magnetization transfer from 1H to
15N during the CP period, the rT/2 pulses on each channel flip 1H and 15N
magnetization to opposite directions to achieve polarization inversion. The
multiple back-to-back rT/2 pulses during the WIM period enable the spin
exchange between the 1H and 15N nuclei and suppress the 1H-‘H dipolar
coupling and chemical shift evolution. TPPM decoupling was applied during the
acquisition period ................................................................................. 43

Figure 12. (a) The 15N chemical shift tensor relative to the peptide plane. Axes
011 and 033 are in the peptide plane with 033 pointing between the N-H and N-CO
bonds, ~17° away from N-H vector and 011 painting to N-Cd. Axis 022 is
perpendicular to the peptide plane. (b) Definition of helix rotation (p) and tilt (T)
angles. The helix axis is defined by ha and M is the radial axis that passes
through Ca carbon and varies for different residues. The helix tilt angle, T, is the
angle between the helix axis and the magnetic field (Bo). The helix rotation p is
the angle between M and the projection of Bo on to the plane perpendicular to ha.
(c) Definition of angles in the system in which the helix has fast rotation around a
fixed axis. Angles a and B are the Euler angles of the rotation axis and the
chemical shift principal axis system (PAS). The angle between the N-H vector

xi

and 033, the N-H vector and the rotation axis, the N-H vector and Bo, and the
rotation axis and Bo are represented by C, n, r} and 6,
respectively ......................................................................................... 49

Figure 13. Dependence of 13CO MAS NMR spectra on pH, membrane
cholesterol content, and temperature. The 13CO labeled positions are noted
above each set of spectra. All spectra were obtained with -50 °C cooling gas
temperature except the third spectra in (b) and (c). All samples contained ~0.8
umol IFP and membranes composed of either ~16 umol DTPC and ~4 umol
DTPG for the spectra in the first two rows or ~16 umol DTPC, ~4 umol DTPG,
and ~10 umol cholesterol for the Spectra (except b and c) in the last two rows.
The pH was labeled for each row of spectra. There is little pH dependence of the
peak shift in the spectra and a strong dependence of spectra on membrane
cholesterol content. The observed chemical shifts of Leu-2, Phe-3, Ala-5, lle-6,
Ala-7, Gly-8, Phe-9, Gly-13, Gly—16 and Gly-20 are consistent with predominant
helical conformation in membranes without cholesterol. The chemical shifts of
Gly-1, Ala-5, Ala-7, Gly-8 and Phe-9 are consistent with predominant [3 strand
conformation in membranes that contained significant cholesterol. For Spectra f,
the downfield peak is assigned to Ala-7 13CO and the upfield peak is assigned to
Gly-8 13CO. The third spectra in (b) and (c) were taken with samples contained
~0.8 umol IFP and membranes composed of ~16 umol POPC and ~4 umol
POPG with cooling gas temperature 0°C. There were very similar peak Shifts in
spectra taken for samples at different temperatures which indicates that cooling
the sample does not change the peptide structure in membranes without
cholesterol. The third and fourth Spectra in (f) were obtained using a 4 mm
diameter rotor and 12.0 kHz MAS frequency. Other Spectra were obtained using
a 4 mm diameter rotor and 8.0 kHz MAS frequency. Each Spectrum was
processed with 200 Hz Gaussian line broadening and was based on 200 - 10000
scans .................................................................................................. 64

Figure 14. Ala-7 13CO — GIy-8 13CO distance measurements that probe the IFP
conformational change associated with membrane cholesterol content. The
prTDQBU spectra at 8 ms dephasing time are displayed for IFP2-A7cG8c
associated with (a) DTPC/DTPG (4:1) or (b) DTPC/DTPG/CHOL (8:2:5)
membranes at pH 5.0. Panel (c) displays plots of prTDQBU AS/So vs dephasing
time for the samples containing DTPC/DTPG (darker symbols) or
DTPC/DTPG/CHOL (lighter symbols). Experimental data are represented as
lines with error bars and best-fit simulated data are represented as diamonds.
The displayed experimental data were based on integrations of the 1300 regions
of the So and S1 spectra and have been adjusted for the ~12% contribution of the
natural abundance 13CO Signal. For the DTPC/DTPG sample, the best-fit Ala-7
13CO — Gly-8 1“CO distance was 2.8 A and was consistent with a backbone
dihedral angle (p = -30° and helical conformation and for the DTPC/DTPG/CHOL
sample, the best-fit distance of 3.5 A was consistent with (p = -121° and with [3
sheet conformation. The DTPC/DTPG sample contained ~16 umol DTPC and ~4
umol DTPG and the DTPC/DTPG/CHOL sample contained ~16 umol DTPC, ~4

xii

umol DTPG, and ~10 umol cholesterol. Each sample also contained ~0.8 umol
lFP2-A7cG8c and the sample cooling gas temperature was —50 °C. Each 80 or
81 spectrum was processed using 100 Hz Gaussian line broadening and was the
sum of: (a) 9000 or (b) 10000 scans. The best-fit )3? were: DTPC/DTPG sample,
36; and DTPC/DTPG/CHOL sample, 32 .................................................... 67

Figure 15. 13CO"'1"’N REDOR measurements that probe the pH dependence of
helicity of IFP associated with DTPC/DTPG. Data are presented for (a-d) IFP-
A5cF9N, (e-h) lFP-F9cG13N, and (H) lFP—G13cM17N and at (a, c, e, g, i, k) pH
5.0 or (b, d, f, h, j, l) pH 7.4. The So and 81 spectra at 16 ms dephasing time are
displayed for (a, b) IFP-A50F9N, (e, f) IF P-FQCG1 3N, and (i, j) IFP-G130M17N, and
the corresponding plots of (AS/So) vs dephasing time are displayed below each
set of spectra with the pH3 5. 0 plot on the left and the pH 7. 4 plot on the right. For
each sample, the peak 3C0 Shift was independent of dephasing time. In each
plot, experimental data are represented as lines with error bars and the best-fit
simulated data are represented as diamonds. The displayed experimental data
were based on integrations of the 3CO regions of the So and S1 spectra and
have been adjusted for the ~22% contribution of the natural abundance 13CO
signal. Each sample contained ~16 umol DTPC, ~4 umol DTPG and ~0. 8 umol
IFP and the sample cooling gas temperature was —50 °C. Each 80 or 81
spectrum was processed using 200 Hz Gaussian line broadening and was the
sum of: (a, b, i, j) ~eooo; (e) ~1oooo; or (f) 16080 scans. The best-fit 22 were: (c)
15; (d) 8; (g) 24; (h) 20; (k) 42; and (I) 21 ................................................... 70

Figure 16. 13CO—13CO distance measurements that probe the conformation of
residues 11 through 13 of lFP-E11VN12A and lFP-N12A. The spectra at 24 ms
dephasing time are displayed for membrane-associated (a) IFP- E11VN12A and
(b) lFP-N12A at pH 5.0. Panel (c) displays the magnitude of dephasing (AS/So)
over dephaisng time, 6, 12 and 24 ms are displayed for lFP-E11VCN12AC, IFP-
N12AcG13c and lFP-A7cG8c associated with DTPC/DTPG (4: 1) membrane or
IFP-A7cG8c associated with and DTPC/DTPG/CHOL (8: 2: 5) membrane a1t3 pH
5. 0. The displayed experimental data were based on integrations of the13CO
regions of the So and 81 spectra. The measured values from IFP-A7CGBC
samples in membranes with or without cholesterol represent data for B strand or
a helix structure, respectively. The data from IFP mutants are more consistent
with data from DTPC/DTPG/CHOL-associated lFP-A7c68c and therefore more
consistent B strand structure ................................................................... 72

Figure 17. Ala-5 13CO Phe—9 15N distance measurements that probe the helicity
of IFP mutants associated with DTPC/DTPG at pH 5.0. The So and S1 spectra at
16 ms dephasing time are displayed for (a) lFP-E11VN12A and (b) lFP-N12A.
The peak chemical shifts of lFP-E11VN12A and lFP—N12A correspond to B
strand and a helix conformation, respectively. For the lFP-N12A sample, there is
also one peak that corresponds to [3 strand conformation. Panel (c) displays plots
of REDOR AS/So vs dephasing time for the samples containing lFP-E11VN12A
(square), IFP-N12A (triangle) or wild-type IFP (circle). Each of the peptides was

xiii

13CO labeled at Ala-5 and 15N labeled at Phe-9. The displayed experimental data
were based on integrations of the whole 1300 regions of the So and 81 spectra
and are represented with error bars. The typical uncertainty is 20.02. Each
sample contained ~0.8 umol peptide, ~16 umol DTPC and ~4 umol DTPG and
the sample cooling gas temperature was -50 °C. Each 80 or 81 spectrum was
processed using 200 Hz Gaussian line broadening and was the sum of ~8000
scans ................................................................................................. 73

Figure 18. 2D 130-130 PDSD spectra of membrane-associated (a) IFP-I10E11u
and (b) lFP—N12F13u at pH 5.0. Each sample contained 16 umol DTPC, 0.4 umol
DTPG and ~0.8 IF P. The data were collected with 10 ms exchange time and total
signal averaging time of ~1.5 days. The MAS frequency was 10 KHz and the
temperature of the sample cooling gas was -50 °C. Spectra were processed with
200 Hz Gaussian line broadening in both dimensions. Peak assignments are
shown using the convention of assignment in f1 (vertical axis)/f2 (horizontal
axis) ................................................................................................... 80

Figure 19. IFP backbone structure based on (a) Glu-11 Shift set A or (b) GIu-11
shift set B. All the hydrophobic residues (Leu-2, Phe-3, lle-6, Phe-9 and lle-10) at
N-temlinal helix are shown in gold in (a-b). Residue Glu-11 is in green and
residue Asn-12 is in red. The N-terminal helix is from residues 2-11 in (a) or 2-10
in (b). The C-terminal conformation was based on the detergent-associated IFP
structure at pH 5.0. In (c) and (d), structure A (red) and'B (green) are respectively
overlaid on top of the solution NMR structure (blue) which is lowest in energy at
pH 5.0. .............................................................................................. 87
Figure 20. 2D 130—130 PDSD spectra of membrane-associated IFP-UI1OE11U at
(a) pH 4.0; (b) pH 5.0; and (c) pH 7.4. The acquisition and processing
parameters are the same as the spectra in Figure 18. Grey arrows point to the
crosspeaks of Glu Cv/CO(COO') (fr/f2). This crosspeak is absent in spectrum (a).
Black arrows point to the crosspeaks of Glu Cy/CO(COOH) (f1/f2) which are
overlapped with other CO peaks. Hollow arrows point to crosspeaks of Glu
Cd/CO (f1/f2) A (left) and B (right). The crosspeak B is absent in spectrum
(0) ...................................................................................................... 88

Figure 21. PDSD Spectra for membrane-associated IFP-I10E11u at (a, b) pH 7.4;
(0) pH 5.0 and (d) pH 4.0 and for membrane-associated lFP-N12G13u at (e, f) pH
7.4 and (g) pH 5.0. The membrane composition was (a, d, e) DTPC/DTPG (4:1)
and (b, c, f, g) POPC/POPG (4:1 ). The samples were cooled with nitrogen gas at
(a, d, e) -50 °C and (b, c, f, g) 0 °C. The spectra have no temperature
dependence from — 50 °C to 0 oC. The spectra were processed with 200 Hz
Gaussian line broadening. The total number of scans was (a-c, e-g) ~100000,
and (d) ~200000. Some of the peak assignments were shown using the
convention of assignment in f1 (vertical axis)/f2 (horizontal axis) ...................... 89

Figure 22. 31F spectra of membrane samples (DTPC/DTPG, 4:1) that (a, b)
contained IFP or (c) had no bound IFP at 35 °C. The peptide to lipid mol ratio is

xiv

0.04. Each spectrum was processed with 200 Hz Gaussian line broadening and
was the sum of 300 to 1000 scans .......................................................... 107

Figure 23. (a) Model of a membrane bilayer with a peptide inserted into a Single
leaflet and with the positions of 31F, 19F(CS), 1°F(C16) and peptide backbone
1300 labeled. The blue balls represent the phosphate headgroups and the gray
lines represent the hydrocarbon chains of lipids. The approximate dimension in A
of the membrane bilayer is shown by the scale bar on the right. The thickness of
the membrane bilayer is ~50 A and the phosphate headgroup is ~8 A in diameter.
The 13CO—19F(C5) and 1300—19F(C16) distances are shown by the black lines. All
the 19FS(C16) are at the bilayer center and have Similar distances to the labeled
13CO. The peptide labeled 1 CO has a shorter distance to the 19F(CS) located at
the same leaflet relative to the 19F(C5) located at the different leaflet. The 1300—
1‘1F(C5) REDOR data contain information from the 13CO-19F(C5) pair and is
dominated by the pair with the shorter distance. The effective concentration of
19F(C5) is half of its real concentration for the REDOR measurement. Panels (b)
and (0) show the structures of 5-19F-DPPC and 16-19F-DPPC molecules
respectively ....................................................................................... 107

Figure 24. REDOR 13C So and S1 NMR spectra at long dephasing time for
membrane-associated IFP samples at pH 5.0. The experiment type and the
dephasing time are labeled on the top of each group of spectra and the 1300
labeled residues are also labeled above each set of So and S1 spectra. Each
sample contained 16 umol DTPC, 4 umol DTPG and 0.8 umol IFP. The samples
used to take spectra (d), (e) and (j-n) contained 9 mol% 5-F-DPPC lipid and the
samples used to take spectra (o-u) contained 9 mol% 16-F-DPPC lipid. Each
spectrum was processed with 200 Hz Gaussian line broadening and was the
sum of 20000 - 30000 scans ................................................................. 110

Figure 25. Summary of experimental REDOR dephasing (AS/So)” for the
spectra displayed in Figure 24. The (AS/So)” values are shown as bars for
different residues and a typical uncertainty is 10.01002 ............................. 113

Figure 26. 13CO—31P REDOR (dark solid line), 13CO—19F(05) REDOR (gray solid
line) and 13CO—19F(C16) REDOR (dotted line) experimental dephasing curves for
IFP samples at pH 5.0. The uncertainties are represented by the error bars and
are typically tom-0.02. The 13CO labeled residues are labeled on top of each
spectrum. The samples used were the same as the ones used to take the
corresponding spectra in Figure 24 ......................................................... 115

Figure 27. Plots of (AS/So)” (circles) and (AS/So)“ (solid lines) vs. dephasing
time for 13CO-31P and 1300— 9F(C16) REDOR of IFP samples at pH 5.0. The
13CO labeled residues are labeled at the top of each spectrum ..................... 116

Figure 28. REDOR 130 So and $1 NMR spectra at long dephasing time for
membrane-associated IFP samples at pH 7.4. The experiment type and the

XV

dephasing time are labeled on the top of each group of Spectra and the 1300
labeled residues are also labeled above each set of So and S1 spectra. Each
sample contained 16 umol DTPC, 4 umol DTPG and 0.8 umol IFP. The samples
used to take spectra (d), (e) and (j-n) contained 9 mol% 5-F-DPPC lipid and the
samples used to take spectra (o-u) contained 9 mol% 16-F-DPPC lipid. Each
spectrum was processed with 200 Hz Gaussian line broadening and was the
sum of 20000 — 30000 scans ................................................................. 120

Figure 29. Summary of experimental REDOR dephasing (AS/So)” for the
spectra displayed in Figure 28. The (AS/So)°"” values are shown as bars for
different residues and a typical uncertainty is :l:0.01-0.02 ............................. 124

Figure 30. 1300-31P REDOR (dark solid line), 13CO—19F(C5) REDOR (gray solid
line) and 13CO—19F(C16) REDOR (dotted line) experimental dephasing curves for
IFP samples at pH 7.4. The uncertainties are represented by the error bars and
are typically 20.01002. The 1300 labeled residues are labeled on top of each
Spectrum. The samples used were the same as the ones used to take the
corresponding spectra in Figure 28 ......................................................... 125

Figure 31. Plots of (AS/So)” (circles and (AS/So)” (solid lines) vs. dephasing
time for (a-h) 1300— 1P and (i) 1300- 9F(C16) REDOR of IFP samples at pH 7.4.
The 1300 labeled residues are labeled at the top of each spectnIm ............... 126

Figure 32. Insertion models for IFP in DTPC/DTPG membranes. The lipid head
groups are shown in blue and the alkyl chains are shown in gray. The gray balls
represent the labeled IFP backbone 1300. (a) IFP inserted into the membrane
bilayer with the 13CO labeled residues indicated by the corresponding residue
number. (b) IFP located at the membrane surface and water layer interface. The
~10 A water layer is above the membrane surface and is not shown in the
model ............................................................................................... 127

Figure 33. A sample powder spectrum of 1300 with the principal values labeled
and the corresponding CO bond orientation. Chemical shift anisotropy arises
from the non-spherical distribution of electron density around a nucleus. The
degree of Shielding or the effect of electron density on chemical shifts depends
on the orientation of the electronic cloud. The electron cloud is shown by the
ellipsoids and the relative orientation of the ellipsoid is Shown for each principal
value ................................................................................................ 139

Figure 34. (a) structure of DMPCd54. (b—g and k) are 2H spectra obtained with the
quadrupolar- echo pulse sequence at 40 °C and processed with (b—g ) 25 Hz or
(k) 100 Hz Gaussian line broadening. (h—j) are 3”P spectra obtained with the
1pda pulse sequence at (h, i) 40 °C or (j) 35 °C and processed with (h, i) 0 Hz or
(k) 100 Hz Gaussian line broadening. The peaks in the 2H spectra represent the
signals of 2 mol% of DMPCd54. The peaks in the 31P spectra represent the 31P
signals of the lipid phosphorus headgroups. Compared to the (k) 2H or (j) 31P

xvi

powder spectra, the spectra in (b—i) have sharp peaks and Show the good
alignment of pure bicelles (b, c) or bicelles with incorporated IFP (d-i). Spectra (b,
d, f, h) are for the unflipped bicelle samples which have their normals
perpendicular to the magnetic field. Spectra(c, e, g, i) are for the flipped bicelle
samples which have their normals parallel to the magnetic field. The sample
composition used to obtain each spectrum was: (b, c) DTPC/DMPCd54/DHPC
(53:1:17 umol); (d-i) 0.7 mol IFP and DTPC/DMPCd54lDHPC (53:1:17 umol); (j)
0.8 umol IFP and DTPC/DTPG (16:4 umol); (k) 1.1 mol IFP and
DTPC/DMPCa54 (53:1 umol). The total number of scans was: (b—e, k) ~8000; (f, g,
j) ~2000; (h) 3; (i) ~100 ........................................................................ 140

Figure 35. 15N NMR Spectra of IFP which probe the orientation of the N-tenninal
helix axis relative to the membrane bilayer normal. Panels a-c are 15N static
spectra of IFP-UN that provide information about 15N chemical shift tensor
principal values and IFP motion in (a, b) hydrated membrane dispersions or (c)
lyophilized dry peptide without membranes. The similar appearances of spectra a,
b, and c suggests that there is not large amplitude motion of the N-terminal helix
in membranes with respect to the membrane bilayer normal at either (a) pH 5.0
or (b) pH 7.4. Panel d is a 15N MAS spectrum of lyophilized lFP2-UN that was
used to determine 15N CSA principal values. Panels e-m display 15N static
spectra of IFPS in the aligned bicelle samples. The incorporated IFP and the
sample pH are labeled above each set of spectra. For each labeled IFP, sharp
15N Signals were observed and there was a significant change in peak chemical
shift as a function of bicelle orientation (lFP-G4N and IFP-GBN have less change
in the 15N chemical shifts and the possible reasons are explained in the text).
Both of these observations were consistent with a well-defined alignment of the
N-ten'ninal helix axis of IFP relative to the bicelle normal. For each labeled IFP,
there was little change in 15N chemical shift as a function of pH which indicated
little change in average helix axis orientation with pH. The samples used to
obtain spectra a and b contained ~1 umol IFP and ~50 umol DTPC. The samples
used to obtain Spectra e-m contained DTPC/DMPCd54/DHPC (53:1 :17 mol) and
0.7 mol IFP. All the spectra were obtained with 1H-15N ramped cross-
polarization followed by 15N detection with 1H decoupling. For spectrum d, the
MAS frequency was 1.5 kHz. The temperature of the gas which flowed around
each sample was 40 °C. Spectra were processed using Gaussian line
broadening of magnitude (a, b) 500 Hz, (c, d) 100 Hz, (e-m) 50—200 Hz. The
number of acquisitions summed for each spectrum was 20000 -
60000 .............................................................................................. 152

Figure 36. 2D 15N chemical shift and N-H dipolar coupling correlation spectra of
bicelle-associated IFP that probe the orientation of the N-terminal helix axis
relative to the membrane bilayer normal. The samples used to obtain the spectra
contained (a, b) IFP-UN, (c, d) IFP-F3NA7N, (e, f) IFP-G4N. (g) IFP-A5”, (h) lFP-
I6N, (i) lFP-G8N and (j) IFP-I10N at (a, c, e, g-j) pH 5.0 or (b, d, f) pH 7.4. All the
samples have the composition of DTPC/DMPCd54/DHPC (53:1:17 umol) and 0.7
mol IFP. Each panel is separated into two parts by the dotted square with the

xvii

inside part representing the unflipped bicelle sample and the outside part
representing the flipped bicelle samples. For the spectra of lFP-F3NA7N, peaks
were assigned based on the measured chemical shifts of lFP-F3N and IFP-A7N in
Figure 33. All the spectra were obtained with Pl-WlM-z pulse sequence. Spectra
c and d were obtained with “Efree” free probe on the 21.4 T spectrometer. Other
spectra were obtained with the Varian Biostatic 1H/X probe on the 9.4 T NMR
spectrometer. The temperature of the gas which flowed around each sample was
40 °C. The number of acquisitions for each spectrum was ~100000 .............. 157

Figure 37. 1D and 2D spectra taken with either a 9.4 T (left) or a 21.4 T (right)
spectrometer. The 1D spectra were taken with the unflipped bicelle samples
containing IFP-UN at pH 5.0. The sensitivity of the two spectrometers is estimated
to be ~1:4.5 (9.4 T spectrometer:21.4 T spectrometer) based on the signal-to-
noise ratios and the numbers of scans of the two spectra. The peak width is
reduced from ~6.5 ppm to ~37 ppm. The 2D spectra were taken with the flipped
bicelle samples containing lFP-F3NA7N at pH 7.4. The linewidth is narrowed for
the spectra taken with 9.4 T spectrometer compared to the spectra taken with
21.4 T spectrometer. All the spectra were processed without line broadening.
The number of acquisitions for each spectrum was (a) 5048; (b) 1042; (c) 6144
(12) x 20 (f1); (d) 1280 (12) x 92 (f1) ........................................................ 161

Figure 38. 2D 15N chemical Shift and N-H dipolar coupling correlation spectra of
bicelle-associated (a, b) IFP-G18-F3NA7N and (c, d) IFP-G1V- F3NA7N IFP at pH
5.0 (a, c) and pH 7.4 (b, d), Each panel represents the composite spectra of
unflipped (< 130 ppm) and flipped (> 130 ppm) bicelle samples containing same
labeled IFP at the same pH. The spectra for lFP-G1S and lFP-G1V resemble
each other and resemble the spectra of wild lFP-F3NA7N which suggest the N-
terrninal helix of these two mutants and wild IFP have Similar motions and tilted
angle. All the spectra were taken with PI-WIM-z sequence. The samples have
the same composition as the samples in Figure 23. The number of acquisitions
for each spectrum is ~100000 ................................................................ 162

Figure 39. 2H quadrupolar splitting spectra of HFP in (a) unflipped and (b) flipped
bicelles show the good alignment of bicelles with incorporated HF P. 15N chemical
shift spectra of bicelle-associated HFP (c-i) have Similar line shapes to the
powder spectrum of HFP (j) and suggest HFP not aligned relative to the bicelle
normal. The HFP used for each sample is labeled above each spectrum. All the
spectra in the left panels and spectrum (h) were taken with unflipped bicelle
samples. The spectra in the first three rows of the right panels were taken with
flipped bicelle samples. The spectrum (j) was taken with an unoriented
membrane bilayer sample. All spectra were taken with samples that contain 0.7
mol HFPmn or 0.4 mol HFPdm and DTPC/DMPCd54lDHPC (53:1:17 umol).
The temperature of the gas which flowed around each sample was 40 °C.
Spectra were processed using Gaussian line broadening of magnitude (a, b) 25
Hz, (c-f, j) 300 Hz, (9, ,h) 400 Hz, and (i) 500 Hz. The number of acquisitions

xviii

summed for each spectrum was: (a) 8000; (b) 3426; (c, f, i) ~26000; (d, g, h, j)
~10000 ............................................................................................. 166

xix

HIV
AIDS

IFP
HFP
CD
ESR
FTIR
NMR

RF

CP
REDOR
r

d
prTDQBU
PDSD
PISEMA
FMOC
DTPC
DTPG
POPC

LIST OF SYMBOLS AND ABBREVIATIONS

Human immunodeficiency virus

Acquired immunodeficiency syndrome

Hemagglutinin

Influenza fusion peptide

HlV fusion peptide

Circular dichroism

Electron spin resonance

Fourier transform infrared

Nuclear magnetic resonance

Magic angle spinning

Radlofrequency

Cross polarization

Rotational-echo double resonance

Distance

Dipolar coupling

Constant-time double-quantum buildup with finite pulses
Proton-driven spin diffusion

Polarization-inversion spin exchange at the magic angle
9-fluorenylmethoxycarbonyl

1 ,2-di-O-tetradecyI-sn-glycero-3-phosphocholine

1 ,2-di-O-tetradecyl-sn-gchero-3-[phospho-rac—(1-glycerol)]

1 -palmitoyl-2-oleoyl-sn-gchero-3-phosphocholine

 

DHPC
16-19F-DPPC

POPG
DMPCd54
5-19F-DPPC

HEPES
MES
HFPmn
HFPdm
CHOL
NAV
TPPM
CW
Pl-WIM-z

CSA

YNv YH

I". (I!

01 1 9 022’ 033

1 ,2-Di-O-Hexyl-sn-Glycero-3-Phosphocholine

1 -palmitoyl-2-( 1 6-fluoropalmitoyl )-sn-glycero-3-
Phosphocholine

1 -palmitoyl-2-oleoyl-sn—glycero-3-[phospho-rac-(1 -glycerol)]
1 ,2-Dimyristoyl-DS4-sn—Glycero-3-Phosphocholine

1-palmitoyl-2-(5-fluoropaImitoyl)-sn-g|ycero-3-
Phosphocholine

N-(2-hydroxyethyl)-piperazine-N’-2-ethanesulfonic acid
2-(N-morpholino)—ethanesulfonic acid

HFP monomer

HFP dimer

Cholesterol

N-acetylvaline

two- pulse phase modulation

Continuous wave

polarization inversion - windowless isotropic mixing -
polarization exchange between z-component of
magnetization

Dephaslng time

Chemical shift anisotropy

Gyromagnetic ratios of 15N and 1H

Chemical shift tensor

Angle between N-H vector and the magnetic field

Peptide dihedral angles

Principal values of the chemical shift tensors

VII

6’

Rigid limit 15N-1H dipolar splitting
Rotational angle about the helix axis
Tilt angle of the helical axis with respect to the magnetic field

Euler angles between the helix rotational axis and the
principal axes of the rigid lattice chemical shift tensors

Angle between the helix rotational axis and the external
magnetic field

Angle between the N-H vector and the helix rotational axis
Angle between the N-H vector and the principal axis 033

Angle between the bicelle normal and the external magnetic
field

Angle between the helix axis and the bicelle normal

xxii

Chapter 1 Introduction

1.1 BACKGROUND

Enveloped viruses such as influenza, chicken pox, measles (Rubeola),
and AIDS (HIV) are encapsulated by a membrane which is acquired upon
budding from an infected cell. Penetration of viruses into the target cells requires
the joining or “fusion” of viral and host cell membranes.1'3 After the fusion
process, a fusion pore is formed across the two membranes which allows the
viral gene to enter its host, as evidenced by the electron micrography of HIV viral
infection (cf. Figure 1).4 Although the membrane free energies are approximately
equal before and after fusion has occurred, membrane fusion rates are typically
very slow in the absence of a catalyst. For this reason, enveloped viruses contain
“fusion proteins” which catalyze the fusion process. For a subgroup of enveloped
viruses termed type I such as HIV and influenza, their fusion proteins share
similar structural and folding motifs, suggesting a comparable fusion
mechanism.5 Therefore the clarification of the structural features of fusion
proteins and the consequent understanding of the fusion mechanism have been
increasingly viewed as bases of rational antiviral drug designs for type I vimses
in general. In addition, the mechanism of viral/host cell fusion induced by fusion
proteins resembles that of intracellular vesicle fusion mediated by the SNARE
proteins,6 so investigating the function of viral fusion proteins may provide
additional insight into the understanding of cellular transport processes.

For all the fusion proteins, the N-temlinal region, termed the fusion peptide

(FP), is relatively apolar and plays a critical role in initiating fusion. Synthetic FP

analogs have been reported to induce vesicle fusion and the red blood cell fusion
in the absence of the rest of fusion proteins.”9 The site directed mutation/fusion
activity relationship are Similar for viral/cell fusion and FP-induced vesicle
fusion.1‘*14 Therefore, understanding the structure of fusion peptides should be
important for understanding the fusion mechanism.

The overall goal of our research is to understand some aspects of the
fusion peptide-induced membrane fusion. I choose influenza fusion peptide (IFP)
to carry out most of my research on the studies of FF structure and the
FP/membrane interaction. Because influenza viral fusion is induced by a simple
change of pH (cf. Figure 2) rather than binding to the host cell membrane
proteins (as in the case of HIV),15 influenza served as the most studied system
for fusion research. My approach has been to study most of the critical aspects of
the membrane-associated IFP, including the secondary structure, IFP membrane

location and the insertion angle of helical form of IFPS.

1. Structural Biology of Viral Fusion Proteins

Membrane fusion is characterized by the mixing of lipid molecules of viral
and host cell membranes and the lipid bilayers eventually merge into one united
whole membrane. There are four proposed steps in viral/host cell membrane
fusion:16 (1) Viral/host cell bindings; (cf. Figure 1a) (2) formation of small fusion
pores through which electrolytes can pass; (3) mixing of viral and host cell lipids;
(cf. Figure 1b) (4) formation of a large fusion pore through which large molecules

can pass and creation of a single virus/host cell moiety. (of. Figure 1c and d)

Steps (1), (3) and (4) were experimentally observed as Shown by the electron
micrograph in Figure 1. There is no experimental proof for the step (2). The
fusion process is controlled on a temporal and molecular basis such that the host

cell remains intact and can protect the viral genetic material until

 

 

 

 

 

 

 

 

 

 

 

Fl'Qure 1. HIV Infection: (Left) Model of infection process. (Right) Freeze fracture
eIsotron micrography of a virion.

 

 

 

I Hemagglutinin 3
‘L
Coated pit .,
\
Coated vesicle
Endosome

H+ ”

.\

Endosome

Nucleus

 

 

 

Figure 2. The first half of the influenza viral life cycle. The virus enters the cell
by endocytosis. The pH drops in the endosome, initiating conformational
changes of the hemagglutinin protein on the viral membrane surface. These
conformational changes ultimately lead to fusion of the viral and endosomal
membranes and release of the viral contents into the cell.

infection. The activation energy barrier between the initial two separate bilayers
and the merged state is hypothesized to be provided by the conformational
change of fusion proteins.

For the influenza virus, the fusion protein is hemagglutinin (HA) and is
abundant on the surface of the virus. HA consists of a HA1 domain that lies
outside the virus and a HA2 domain which is composed of a ~185-residue
ectodomain that is outside the virus, a ~25-residue transmembrane domain, and
a ~10-residue endodomain that is inside the virus.17 Prior to infection of a
respiratory epithelial host cell, the pH of the virus exterior is 7.4 (cf. Figure 2) and
the HA1 domain and the HA2 ectodomain are joined by a single disulfide bond.
There is a crystal structure of HA1 and the HA2 ectodomain at pH 7.5 which
showed that a HA trimer is formed by association of three I-lA2 ectodomains and
that the three HA1 partners are located on the outside of the HA2 core (cf. Figure
3a)." The structure undergoes a drastic change in the process of fusion which is
shown in Figure 318 and 419. The fusion starts from binding of the HA1 domains
to the terminal sialic acids of glycoproteins and glycolipids on the surface of a
host cell which leads to endocytosis of the virus into the cell. The cell physiology
results in pumping of protons into the endosome and lowering of the pH ofithe
endosome to ~52 (cf. Figure 2). At this pH, HA has a different equilibrium
structural state which is characterized by cleavage of the disulfide bond between
HA1 and HA2, release of HA1 from the HA2 trimer, and a major conformational
change of the HA2 molecules within the trier (cf. Figure 4). Some of these pH-

dependent structural differences have been understood from comparison of the

previously discussed pH 7.5 HA1/HA2 structure and a pH 5 structure of a HA2
trimer for which residues 38-175 of HA2 were resolved as shown in figure 3b.18
Fusion between the viral membrane and the endosomal membrane occurs only
after these structural changes have occurred and leads to release of the viral
nucleocapsid into the host cell cytoplasm and subseqUent insertion of the viral
genetic material into the host genome.15' 2°

The ~20 N-terrninal residues of HA2 domain are relatively apolar and are
known as the fusion peptide or IFP. The IFP plays a critical role in fusion, as
evidenced by disruption of fusion activity for HA with point mutations in the IFP.
In the pH 7.5 HA1/HA2 structure, the IFP was buried within the HA1/HA2 trimer.
The pH 5.0 HA2 structure did not include the IFP or the transmembrane domain
but did suggest that these regions are near the same end of the molecule. Other
experiments have shown that after fusion has occurred, the IFP and
transmembrane domain of HA2 are the only HA2 regions which are deeply
membrane-inserted?“25 The overall literature data support a model in which the
lowered pH leads to exposure of the IFP followed by IFP insertion into the
endosomal membrane and then membrane fusion. The role of the IFP in
membrane fusion catalysis is probably related to its perturbation of the
endosomal membrane.

Peptides with the IFP sequence have been studied as models to
understand the role of the IFP in influenza viral fusion and there is experimental

evidence which supports the utility of the peptide model system. For example,

introduction of the IFP into a vesicle solution results in mixing of lipids between

vesicles. Such mixing is one characteristic of vesicle fusion. There has also been
investigation of IFPs with the same point mutations as were studied in the IFP
domains of whole HA proteins. There are good correlations in the mutation-
activity relationships of IFP-induced vesicle fusion and HA-catalyzed fusion of
cell membranes.10 Another interesting aspect of IFP is that much greater vesicle
fusion is induced at pH 5 than at pH 7.4. A direct comparison cannot be made to
HA-induced fusion because the IFP is buried in HA at pH 7.4 whereas the IFP
peptide is initially free in solution at both pHS. However, the pH-dependence of
IFP-induced vesicle fusion has still been studied in part because comparison of
IFPS at different pHs is more straightforward than comparison of wild-type and
mutant IFPS.15' 23'” It is interesting that pH-dependent vesicle fusion has also
been observed for HA2 constructs which contain the IFP including a full
ectodomain “FHA2”construct .29 At pH 5 and lower HA2:lipid ratios, FHA2
induces ~10-fold greater vesicle fusion than does IFP.

In the pH 7.5 HA1/HA2 crystal structure, the buried IFP has an extended
conformation.“ 3° There have been NMR structures of the IFP peptide in
detergent micelles at pH 5 which showed a N-tenninal helix from residues 2-10
followed by a turn followed by'a C-terrninal 31o-helix from residues 13-18. The
angle between the axes of the two helices was ~105° (cf. Figure 5a). The IFP
structure in detergent at pH 7.4 showed a N-terminal helix from residues 2-9
followed by a turn and a C-terminal region with mostly extended secondary
structure (cf. Figure 5b).31' 32 For pH 5 samples containing IFP associated with

membranes which lack cholesterol, IFP structure has been studied by circular

(b)

   

Figure 3. Ribbon diagram of Influenza hemagglutinin glycoprotein at (a) pH 7.417
and (b) pH 5.018. The structure at pH 7.4 included residues 4-328 of HA1 and
residues 1-175 of HA2. The structure at pH 5.0 shows residues 12-16 of HA1
and residues 40-153 of HA2 for one monomer, residues 11-16 of HA1 and
residues 40-162 of HA2 for another monomer, and residues 10-17 of HA1 and
residues 40-162 of HA2 for the third monomer.

(a) (b) Fusion peptide (0)
Cellular membrane T ~ ,

1 I . I I ‘ .
a... II I
HA... 37 .
HA1»
Viral membrane _ " . ’qIIIII‘D
(d) (e) i 1 (f) V Transmembrane

domain

I

E as; as

Figure 4. Proposed mechanism for Influenza membrane fusion.19 (a) HA protein
conformation at pH 7.4. (b) Upon exposure to the low pH found in endosomes,
HA forms an extended conformation and the N-terminal fusion peptide inserts
into the target membrane. (c) Several HA proteins are thought to be involved. (d)
Protein refolding begins and the released free energy cause the membranes to
bend to each other. (e) Formation of the hemifusion intermediate in which the
outer leaflets of membranes mix. (f) Protein refolding completes and the fusion
pore is formed. The fusion peptide and the transmembrane domain are in the
same membrane and interact with each other. Only the HA Structures in (a) and
(f) have been observed by crystallography but many biological data support other
proposed steps.

dichroism (CD), infrared, electron spin resonance (ESR), and solid-state nuclear
magnetic resonance (NMR) spectroscopies.3‘°”3‘5 There is general agreement that
IFP adopts helical form at pH 5.

As noted above, the IFP induces much greater vesicle fusion at pH 5.0
than at pH 7.4 and there have been some studies of membrane-associated
structure at the two pHs to detect structural differences which may correlate with
the different functional activities. In membranes without cholesterol, there appear
to be similar fractions of helical conformation at the two pHs as determined by
analysis of CD and infrared spectra?3 The pH dependence of the membrane
location of the N-terminal IFP helix has also been investigated and was
motivated by the idea that the IFP membrane location is related to membrane
perturbation and fusion activity. For example, infrared spectra have provided
information about the angle between an IFP helix axis and the membrane normal
and the derived angles have been 45° or 65°, both independent of pH.” 3° The
membrane insertion depths of particular IFP residues have been investigated
with analysis of effects of relaxation agents on the ESR linewidths of Spin-labeled
IFPS, which can provide information on the IFP helix orientation relative tothe
membrane bilayer normal. In one study, a 15° degree change was observed for
the helix formed by the N-terminal first ten residues (cf. Figure 5).19 In some
contrast, only a 3° helix orientational change was detected in similar ESR study
using the IFP region of a HA2 ectodomain construct.

There have also been molecular dynamics investigations of IFP in

37-42

detergent and in membranes . Several of these studies Showed structures

similar to those observed by NMR in detergent 37' 40' 41. In some contrast, another
study found that the IFP is a single continuous helix located near the phosphate
headgroups with an angle of 78° between the helix axis and the membrane
normal ‘12. In this study, the pH dependence was also examined by using different
protonation states of glutamic acids or aspartic acids and it is shown that there
was little pH dependence of the equilibrium structure or membrane location. This
study also considered an IFP trimer because of the trimeric oligomelization of
HA2 and found a variety of stable membrane locations and tilt angles for the
three IFP helices. For example, one configuration had the three helices located
close to the phosphate headgroups with tilt angles of 80°, 70°, and 60° and
another configuration had two of the IFP helices located near the phosphate
headgroups with tilt angles of 60° and 50° while the third IFP helix had its
deepest insertion 12 A from the phosphate headgroups and a tilt angle of 35°.
Another simulation also predicted that the monomeric IFP lies at the water-
bilayer interface and is parallel to the membrane surface.38

Those discrepancies indicate that the IFP structure and membrane _
location merit further investigation, especially in a more biologically relevant
environment with a higher resolution method. In our research, a detailed study of
the IFP stnlcture was carried out, which considers the effect of membrane
cholesterol and pH. The pH-dependent membrane location and orientation of IFP

helix were also explored in a systematic way at an atomic resolution level.

2. Membrane bilayers as model biomembrane systems

10

A biomembrane consists of a fluid phospholipid bilayer intercalated with
proteins, carbohydrates, and their complexes. Phospholipids are molecules that
are composed of a polar head group and a hydrophobic acyl tail. The major lipid
component of biomembranes includes phosphatidylcholine, phosphatidylserine,
phosphatidylinositol, phosphatidylethanolamine, phosphatidylglycerol and
sphingomyelin. Upon suspension in water, phospholipids themselves can form
liposomes which are closed bilayer vesicles. They're generally several hundred A
in diameter and ~50 A in thickness, depending on the preparation protocol and
the lipids in use. The self-sealed bilayer system is an excellent model to study
membrane-bound proteins and peptides.

In my research, phosphatidylcholine and phosphatidylglycerol are mainly
used due to the significant amount of phosphatidylcholine and the presence of
negatively charged components in the real biomembranes of both influenza virus
and infected respiratory epithelial cells.43"1°

Addition of a certain amount of detergent or short chain lipids to the
phospholipids can drive the formation of uniform sized bilayers which are rimmed
by the short chain Iipids.‘17'49 These disk-like bilayers, termed bicelles,
automatically align in the presence of the external magnetic field with their bicelle
normal perpendicular to the magnetic field (cf. Figure 6a).‘1°'°°52 The uniform
alignment is due to the interaction of the anisotropy of the bicelle magnetic
susceptibility and the magnetic field. This alignment can be flipped 90°, where
the bicelle normal is parallel to the magnetic field (cf. Figure 6b), by adding

lanthanide ions into the bicelle solution which changes the anisotropy of bicelle

11

magnetic susceptibility.53 The bicelle system is ideal for the study of the relative
orientation of peptide residing in bicelles. It is important that bicelles are intact in
the presence of the fusion peptides to maintain their alignment in the magnetic
field. My studies with both influenza and HIV fusion peptides showed that the
alignment of bicelles is not affected by the fusion peptides and the IFP has good

alignment when bound with bicelles.

(a)

 

(b)

 

Figure 5. Solution NMR showed that IFP in DPC micelles has helix-tum-(31o-helix)
structure at (a) pH 5.0 (PDB ID: 1ibn) and has helix-tum-extended structure at (b)
pH 7.4 (PDB ID: 1ibo). The N-tenninal helix has an insertion angle of ~38° and
~23° with respect to the membrane surface at pH 5.0 and pH 7.4, respectively.

12

 

Figure 6: Representation of aligned bicelles with their normals (a) perpendicular
to the external magnetic field (Bo) or (b) parallel to B0 due to the addition of YbCl3.
3. Solid-State NMR measurements of membrane-bound FPS

Solid-state nuclear magnetic resonance (NMR) spectroscopy is a unique
method to determine the atomic-level structure of large biological systems. Since
the problems with broad lines and low sensitivity have been mitigated by the
Introduction of magic angle Spinning (MAS) and multiple radiofrequency (RF)
pulse sequences, solid-state NMR has been especially useful for systems which
are difficult to crystallize for X-ray analysis or too big to have fast tumbling motion
for solution NMR. Useful parameters such as chemical Shifts and magnetic
dipolar coupling measured by solid-state NMR provide valuable information for
the structure determination of solid systems. There are two principle advantages
of solid-state NMR over the more established X-ray crystallography and solution
NMR spectroscopy: (1) crystals are not required and (2) large (>30,000
molecular weight) systems can be studied.“ 55 Recently, systems including the
B-amyloid fibrils implicated in Alzheimer’s disease56' 57, the E. coli serine
receptor”, and a HIV-1 peptide/neutralizing antibody complex59 have been

studied by solid state NMR.

Several techniques have been applied in my research. (1) Cross
polarization (CP) and MAS were generally used in most of the solid-state NMR
techniques to increase the signals.°° For most static solid samples, NMR peaks
are broadened by anisotropic effects, such as dipolar and quadrupolar
interactions, due to the slow molecular tumbling rates. These effects can be
greatly reduced and peaks can be subsequently narrowed by MAS, whereby the

NMR sample is rotated at kHz frequencies about an axis tilted at the “magic

angle” tan'1J2 , or 54.7° relative to the static external magnetic field direction.
Cross-polarization increases the signal of nuclei with low gyromagnetic ratio such
as 13C, 15N by transferring the spin polarization from species with high
gyromagnetic ratio, e.g. 1H. (2) Rotational-echo double resonance (REDOR)
under MAS condition permits the detection of heteronuclear magnetic dipole-
dipole couplings.°1 The distances (r) between two heteronuclei such as ”CO—
15N, 1‘°’CO—31P and 130—19F can be determined from the measured magnetic
dipole-dipole couplings (d) based on r = (C/ d)1’3, where the value of constant C
is calculated from the nuclear isotopes. (3) Likewise, the constant-time double-
quantum buildup with finite pulses (prTDQBU) technique permits the detection
of homonuclear magnetic dipole-dipole coupling and the distance between two
homonuclei such as 130—13C can be determined based on a similar formula
between r and d.”67 Those distance measurements provide a lot of valuable
infonTIation on membrane-associated IFP conformation and the membrane
location of IFP. The distance measurements between 13CO—15N and 130-13C

nuclei allow us to discriminate different IFP conformations in a more quantitative

l4

way. The 13CO and 15N labels are carefully chosen so that the distances between
13CO—15N and 130-13C are shorter in a helical conformation compared to the
distances between the same labeled pair in an extended conformation. In
addition the dipolar-coupling measurements between 13C0 of IFP and 31P or 19F
of phospholipids allow the measurement of the distances between the IFP
backbone and the membrane surface (31P) or interior (19F) which can be used to
study membrane location of IFP. (4) The proton-driven spin diffusion (PDSD)
technique detects the 13C—13C correlation of the membrane-associated IFPS.63 A
short mixing time is used so that only intra-residue crosspeaks are observed
which provides chemical shift information of 1°C labeled residues as the basis for
the secondary structure determination of IFP. (5) The PISEMA-type (polarization-
inversion spin exchange at the magic angle) technique detects the chemical
shifts and heteronuclear magnetic dipole-dipole coupling for static (non-MAS)
samples, eg. 15N chemical shifts and 15N—1H dipolar coupling of selectively 15N
labeled bicelle-associated lFP.°°' 69 These values are anisotropic, i.e. the values
depend on the orientation of the N-H vector, which can be correlated to the
orientation of the peptide in the bicelle system. i
The utilization of these innovative techniques enables a detailed
investigation of the structure of the membrane-associated IFP. The chemical shift
measurements combined with 13CO—‘5N and 13CO—13CO dipole—dipole
measurements revealed that IFP adopts a major helical conformation in the
membranes lacking cholesterol and B strand conformation in cholesterol-

containing membranes. For IFP in membranes lacking cholesterol, the helical

15

conformation does not change as a function of pH, i.e. both the N-terrninal and
C-terminal residues are helical at pH 5.0 and pH 7.4. This result is different from
the measurements of micelle-associated IFP using solution NMR as shown in
Figure 5.31 The break in the middle of the IFP helix in this model of micelle-
associated IFP was also observed for the membrane-associated IFP by the 13C—
130 chemical shifts. But the observed break region tumed out to have two
conformations instead of the simple turn suggested by the solution NMR
measurement.

This work also focused on the investigation of the membrane location and
tilt angle of IFP helix relative to the membrane bilayer normal. The 15N chemical
shift and N-H dipolar coupling measurements suggested that the N-temlinal helix
of IFP has a fast rotational motion relative to the helix axis and the angle
between the helix axis and the membrane bilayer normal is ~45°. Both the motion
and the helix tilt angle is independent of mutations to Gly-1 and to pH change.
The distance measurement between IFP backbone 13C0 and 31P on the
membrane bilayer surface or 19F in the membrane interior showed that IFP is
inserted into the membrane hydrophobic interior. An inverted boomerang
stmcture was proposed with the N- and C-terminal regions of IFP in close contact
with the membrane phosphorus headgroups and the middle region inserted into
the outer leaflets of membranes. The inserted IFP has no gross positional

change for samples at different pHS.

16

1.2 REFERENCES

1. Blumenthal, R.; Dimitrov, D. S.; Hoffman, J. F.; Jamieson, J. C.,
Membrane fusion In Handbook of Physiology, Section 14: Cell Physiology.
Oxford: New York, 1997, 563-603.

2. Dimitrov, D. 8., Cell biology of virus entry. Cell 2000, 101, (7), 697-702.

3. Eckert, D. M.; Kim, P. S., Mechanisms of viral membrane fusion and its
inhibition. Annu. Rev. Biochem. 2001, 70, 777-810.

4. Hernandez, L. D.; Hoffman, L. R.; Wolfsberg, T. G.; White, J. M., Virus-cell
and cell-cell fusion. Annu. Rev. Cell. Dev. Biol. 1996, 12, 627—661.

5. Kielian, M.; Rey, F. A., Virus membrane-fusion proteins: more than one
way to make a hairpin. Nature Reviews Microbiology 2006, 4, 67-76.

6. Weber, T.; Zemelman., B. V.; McNew, J. A.; Westerrnann, B. Gmachl, M.;
Paliati, F.; SOllner, T. H.; Rothman, J. E., SNAREpins: Minimal Machinery for
Membrane Fusion. Col/1998, 92, 759-772.

7. Durell, S. R.; Martin, l.; Ruysschaert, J. M.; Shai, Y.; Blumenthal, R., What
studies of fusion peptides tell us about viral envelope glycoprotein-mediated
membrane fusion. Mol. Membr. Biol. 1 997, 14, (3), 97-1 12. ‘

8. Rafalski, M.; Lear, J. D.; DeGrado, W. F., Phospholipid interactions of
synthetic peptides representing the N-tenninus of HIV gp41. Biochemistry 1990,
29, (34), 7917-7922.

9. Slepushkin, V. A.; Melikyan, G. B.; Sidorova, M. S.; Chumakov, V. M.;
Andreev, S. M.; Manulyan, R. A.; Karamov, E. V., Interaction of human
immunodeficiency virus (HIV-1) fusion peptides with artificial lipid membranes.
Biochem. Biophys. Res. Commun. 1990, 172, (2), 952-957.

10. Steinhauer, D. A.; Wharton, S. A.; Skehel, J. J.; Wiley, D. 0., Studies of
the membrane-fusion activities of fusion peptide mutants of influenza-virus
hemagglutinin. J. Virol. 1995, 69, (11), 6643-6651.

11. Freed, E. O.; Myers, D. J.; Risser, R., Characterization of the fusion
domain of the human immunodeficiency virus type 1 envelope glycoprotein gp41.
Proc. Natl. Acad. Sci. U.S.A. 1990, 87, (12), 4650-4654.

12. Freed, E. O.; Delwart, E. L.; Buchschacher, G. L., Jr.; Panganiban, A. T.,
A mutation in the human immunodeficiency virus type 1 transmembrane
glycoprotein gp41 dominantly interferes with fusion and infectivity. Proc. Natl.
Acad. Sci. U.S.A. 1992, 89, (1), 70-74.

17

13. Schaal, H.; Klein, M.; Gehrmann, R; Adams, 0.; Scheid, A., Requirement
of N-terrninal amino acid residues of gp41 for human immunodeficiency virus
type 1-mediated cell fusion. J. Virol. 1995, 69, (6), 3308-3314.

14. Delahunty, M. D.; Rhee, I.; Freed, E. O.; Bonifacino, J. S., Mutational
analysis of the fusion peptide of the human immunodeficiency vims type 1:
identification of critical glycine residues. Virology 1996, 218, (1 ), 94-102.

15. Murata, M.; Sugahara, Y.; Takahashi, S.; Ohnishi, S., PH-Dependent
Membrane-Fusion Activity of a Synthetic 20 Amino-Acid Peptide With the Same
Sequence As That of the Hydrophobic Segment of Influenza-Virus Hemagglutinin.
J. Biochem. 1987, 102, (4), 957-962.

16. Bentz, J., Membrane fusion mediated by coiled coils: a hypothesis.
Biophys. J. 2000, 78, (2), 886-900.

17. Wilson, I. A.; Skehel, J. J.; Wiley, D. C., Structure of the haemagglutinin
membrane glycoprotein of influenza virus at 3 A resolution. Nature 1981, 289,
(5796), 366-373.

18. Bullough, P. A.; Hughson, F. M.; Skehel, J. J.; Wiley, D. 0., Structure of
influenza haemagglutinin at the pH of membrane fusion. Nature 1994, 371,
(6492). 37-43.

19. Jardetzky, T. 8.; Lamb, R. A., Virology: A class act. Nature 2004, 427,
307-308.

20. Hughson, F. M., Structural characterization of viral fusion proteins. Current
Biology 1995, 5, (3), 265-274.

21. Gething, M. J.; Doms, R. W.; York, 0.; White, J., Studies On the
Mechanism of Membrane-Fusion - Site-Specific Mutagenesis of the
Hemagglutinin of Influenza-Virus. J. Cell Biol. 1986, 102, (1 ), 11-23.

22. Daniels, R. 8.; Downie, J. C.; Hay, A. J.; Knossow, M.; Skehel, J. J.; Wang,
M. L.; Wiley, D. 0., Fusion Mutants of the influenza-vims hemagglutinin
glycoprotein. Cell 1985, 40, (2), 431-439.

23. Orlich, M.; Rott, R., Therrnolysin activation mutants with changes in the
fusogenic region of an influenza-virus hemagglutinin. J. Viro. 1994, 68, (11),
7537-7539.

24. Durrer, P.; Galli. C.; Hoenke, S.; Corti, C.; Gluck, R.; Vorherr, T.; Brunner,
J.; Walker, J. A.; Kawaoka, Y.; importance of conserved amino-acids at the
cleavage site of the hemagglutinin of a virulent avian influenza-a virus. J. gene.
Virol. 1993, 74, 31 1-314.

18

25. Durrer, P.; Galli. C.; Hoenke, S.; Corti, C.; Gluck, R.; Vorherr, T.; Brunner,
J., H"-induced membrane insertion of influenza virus hemagglutinin involves the
HA2 amino-tenninal fusion peptide but not the coiled coil region. J. Biol. Chem.
1996, 271, (23), 13417-13421.

26. Qiao, H.; Armstrong, R. T.; Melikyan, G. 3.; Cohen, F. S.; White, J. M., A
specific point mutant at position 1 of the influenza hemagglutinin fusion peptide
displays a hemifusion phenotype. Mol. Biol. Cell 1999, 10, (8), 2759-2769.

27. Wharton, S. A.; Martin, S. R.; Ruigrok, R. W. H.; Skehel, J. J.; Wiley, D. C.,
Membrane-Fusion By Peptide Analogs of Influenza-Virus Hemagglutinin. J. Gene.
Virol. 1988, 69, 1847-1857.

28. Lear, J. D.; Degrado, W. F., Membrane-binding and conformational
properties of peptides representing the NHz terminus of influenza HA2. J. Biol.
Chem. 1987, 262, (14), 6500-6505.

29. Curtis-Fisk, J.; Preston, 0.; Zheng, Z. X.; Worden, R. M.; Weliky, D. P.,
Solid-state NMR structural measurements on the membrane-associated
influenza fusion protein ectodomain. J. Am. Chem. Soc. 2007, 129, (37), 11320.

30. Chen, J.; Skehel, J. J.; Wiley, D. C., N- and C-terrninal residues combine
in the fusion-pH influenza hemagglutinin HA(2) subunit to form an N cap that
terminates the triple-stranded coiled coil. Proc. Natl. Acad. Sci. U S A 1999, 96,
(16), 8967-8972.

31. Han, X.; Bushweller, J. H.; Cafiso, D. S.; Tamm, L. K., Membrane
structure and fusion-triggering conformational change of the fusion domain from
influenza hemagglutinin. Nat. Struct. Biol. 2001, 8, (8), 715-720.

32. Lai, A. L.; Park, H; White, J. M.; Tamm, L. K., Fusion Peptide of Influenza
Hemagglutinin Requires a Fixed Angle Boomerang Structure for Activity. J. Biol.
Chem. 2006, 281, (9), 5760-5770.

33. Luneberg, J.; Martin, l.; Nussler, F.; Ruysschaert, J. M.; Herrmann, A.,
Structure and Topology of the Influenza-Virus Fusion Peptide in Lipid Bilayers. J.
Biol. Chem. 1995, 270, (46), 27606-27614.

34. Macosko, J. C.; Kim, C. H.; Shin, Y. K., The membrane topology of the
fusion peptide region of influenza hemagglutinin determined by spin-labeling
EPR. J. Mol. Biol. 1997, 267, (5), 1 139-1 148.

35. Wasniewski, C. M.; Parkanzky, P. D.; Bodner, M. L.; Weliky, D. P., Solid-
state nuclear magnetic resonance Studies of HIV and influenza fusion peptide
orientations in membrane bilayers using stacked glass plate samples. Chem.
Phys. Lipids 2004, 132, (1 ), 89-100.

I9

36. lshiguro, R.; Kimura, M; Takahashi, S., Orientation of fusion-active
synthetic peptides in phospholipid bilayers: Determination by Fourier transform
infrared spectroscopy. Biochemistry 1993, 32, (37), 9792-9797.

37. Vaccaro, L.; Cross, K. J.; Kleinjung, J.; Straus, S. K.; Thomas, D. J.;
Wharton, S. A.; Skehel, J. J.; Fratemali, F., Plasticity of Influenza Haemagglutinin
Fusion Peptides and Their Interaction with Lipid Bilayers. Biophys. J. 2005, 88,
25—36.

38. Bechor, D.; Ben-Tal, N., Implicit solvent model studies of the interactions
of the influenza hemagglutinin fusion peptide with lipid bilayers. Biophys. J.
2001, 80, (2). 643—655.

39. Spassov, V. 2.; Yan, L.; Szalma, 8., Introducing an implicit membrane in
generalized Born/solvent accessibility continuum solvent models. J. Phys. Chem.
B 2002, 106, (34), 8726-8738.

40. Huang, 0.; Chen, C. L.; Herrmann, A., Bilayer Conformation of Fusion
Peptide of Influenza Virus Hemagglutinin:A Molecular Dynamics Simulation
Study. Biophys. J. 2004, 87, (1 ), 14-22.

41. Lague, P.; Roux, B.; Pastor, R. W., Molecular dynamics simulations of the
influenza hemagglutinin fusion peptide in micelles and bilayers: Conformational
analysis of peptide and lipids. J. Mol. Biol. 2005, 354, (5), 1129-1141.

42. Sammalkorpi, M.; Lazaridis, T., Configuration of influenza hemagglutinin
fusion peptide monomers and oligomers in membranes. BBA-Biomembranes
2007, 1768, (1 ), 30-38.

43. Woman, H. J.; Brasitus, T. A.; Dudeja, P. K.; Fozzard, H. A.; Field, M.,
Relationship Between Lipid Fluidity and Water Permeability of Bovine Tracheal
Epithelial-Cell Apical Membranes. Biochemistry 1986, 25, (7), 1549-1555.

44. Scheiffele, P.; Rietveld, A.; Wilk, T.; Simons, K., Influenza Viruses Select
Ordered Lipid Domains during Budding from the Plasma Membrane. J. Biol.
Chem. 1999, 274, (4), 2038-2044.

45. Lenard, J.; Compans, R. W., The membrane structure of lipid-containing
viruses. BBA-Biomembranes 1974, 344, (1 ), 51-94.

46. Blom, T. S.; Koivusalo, M.; Kuismanen, E.; Kostiainen, R.; Somerharju, P.;
lkonen, E., Mass Spectrometric Analysis Reveals an Increase in Plasma
Membrane Polyunsaturated Phospholipid Species upon Cellular Cholesterol
Loading. Biochemistry 2001, 40, (48), 14635—14644.

20

47. Sanders, C. R.; Landis, G. C., Reconstitution of membrane proteins into
lipid-rich bilayered mixed micelles for NMR studies. Biochemistry 1995, 34, (12),
4030-4040.

48. Sanders, C. R.; Prosser, R. S., Bicelles: a model membrane system for all
seasons? Structure 1998, 6, (10), 1227-1234.

49. Sanders, C. R.; Hareb, B. J.; Howardc, K. P.; Prestegardc, J. H.,
Magnetically-oriented phospholipid micelles as a tool for the study of membrane-
associated molecules. Prog. Nucl. Magn. Reson. Spec. 1994, 26, 421-444.

50. Forrest, B. J.; Reeves, L. W., New lyotropic liquid crystals composed of
finite nonspherical micelles. Chem. Rev. 1981, 81, (1), 1—14.

51. Davis, J. H., Deuterium nuclear magnetic resonances of exchange-labeled
gramicidin in an oriented lyotropic nematic phase. Biochemistry 1988, 27, (1),
428-436.

52. Sanders, C. R., Solid state 13C NMR of unlabeled phosphatidylcholine
bilayers: spectral assignments and measurement of carbon-phosphorus dipolar
couplings and 130 chemical shift anisotropies. Biophys. J. 1993, 64, (1 ), 171 -1 81.

53. Howard, K. P.; Opella, S. J., High-Resolution Solid-State NMR Spectra of
Integral Membrane Proteins Reconstituted into Magnetically Oriented
Phospholipid Bilayers. J. Magn.. Reson. B 1996, 112, 91-94.

54. Griffin, R. G., Dipolar recoupling in MAS spectra of biological solids. Nat
Struct. Biol. 1998, 5 Suppl, 508-512.

55. Marassi, F. M.; Opella, S. J., NMR structural studies of membrane
proteins. Curr. Opin. Struct. Biol. 1998, 8, (5), 640-648.

56. Benzinger, T. L.; Gregory, D. M.; Burkoth, T. S.; Miller-Auer, H.; Lynn, D.
G.; Botto, R. E.; Meredith, S. C., Propagating structure of Alzheimer‘s b-
amyloid(10-35) is parallel b-sheet with residues in exact register. Proc. Natl.
Acad. Sci. U.S.A. 1998, 95, (23), 13407-13412.

57. Lansbury, P. T.; Costa, P.R.; Griffiths, J. M.; Simon, E. J.; Auger, M.;
Halverson, K. J.; Koclsko, D. A.; Hendsch, Z. S.; Ashbum, T. T.; Spencer, R. G.;
Tidor, B.; Griffin, R.G., Structural Model for the Beta-amyloid Fibril Based on
lnterstrand Alignment of an anti-parallel-sheet Comprising a C-terminal Peptide.
Nature Struc. Biol. 1995, 2, (11), 990-998.

58. Wang, J.; Balazs, Y. S.; Thompson, L. K., Solid-state REDOR NMR

distance measurements at the ligand site of a bacterial chemotaxis membrane
receptor. Biochemistry 1 997, 36, (7), 1699-1 703.

21

59. Weliky, D. P.; Bennett, A. E.; Zvi, A.; Anglister, J.; Steinbach, P. J.; Tycko,
R., Solid-state NMR evidence for an antibody-dependent conformation of the V3
loop of HIV-1 gp120. Nat. Struct. Biol. 1999, 6, (2), 141 -5.

60. Pines, A.; Gibby, M. G.; Waugh, J. S., Proton enhanced nuclear magnetic
resonance of dilute spins in solids. J. Chem. Phys. 1973, 59, 569-590.

61. Gullion, T., Introduction to rotational-echo, double-resonance NMR.
Concepts Magn. Reson. 1998, 10, (5), 277-289.

62. Balbach, J. J.; Petkova, A. T.; Oyler, N. A.; Antzutkin, O. N.; Gordon, D. J.;
Meredith, S. C.; Tycko, R., Supramolecular structure in full-length Alzheimer‘s b-
amyloid fibrils: Evidence for a parallel b-sheet organization from solid-state
nuclear magnetic resonance. Biophys. J. 2002, 83, (2), 1205-1216.

63. Bennett, A. E.; Ok, J. H.; Griffin, R. G.; Vega, 8., Chemical-shift correlation
spectroscopy in rotating solids - radio frequency-driven dipolar recoupling and
longitudinal exchange. J. Chem. Phys. 1992, 96, (11), 8624-8627.

64. Gullion, T.; Vega, 8., A simple magic angle spinning NMR experiment for
the dephasing of rotational echoes of dipolar coupled homonuclear Spin pairs.
Chem. Phys. Lett. 1992, 194, (4-6), 423-428.

65. Bennett, A. E.; Rienstra, C. M.; Griffiths, J. M.; Zhen, W. G.; Lansbury, P.
T.; Griffin, R. G., Homonuclear radio frequency-driven recoupling in rotating
solids. J. Chem. Phys. 1998, 108, (22), 9463-9479.

66. Bennett, A. E.; Weliky, D. P.; Tycko, R., Quantitative conformational
measurements in solid state NMR by constant-time homonuclear dipolar
recoupling. J. Am. Chem. Soc. 1998, 120, (19), 4897-4898.

67. lshii, Y., 13C-1°C dipolar recoupling under very fast magic angle spinning in
solid-state nuclear magnetic resonance: Applications to distance measurements,
spectral assignments, and high-throughput secondary-structure determination. J.
Chem. Phys. 2001, 114, (19), 8473-8483.

68. Dvinskikh, S. V.; Durr, U.; Yamamoto, K.; Ramamoorthy, A., A high-
resolution solid-state NMR approach for the structural studies of bicelles. J. AM.
Chem. Soc. 2006, 128, 6326-6327.

69. Dvinskikh, S. V.; Yamamoto, K.; Ramamoorthy, A., Separated local field

NMR spectroscopy by windowless isotropic mixing. Chem. Phys. Lett. 2006, 419,
168-1 73.

22

Chapter 2 Materials and Methods

2.1 MATERIALS

Wang resins and 9-fluorenylmethoxycarbonyl (FMOC) amino acids were
obtained from Peptides lntemational (Louisville, KY), Calbiochem-Novabiochem
(La Jolla, CA), and Advanced Chemtech (Louisville, KY). lsotopically labeled
amino acids were obtained from Cambridge Isotope Laboratories (Andover, MA)
and were Fmoc-protected using literature procedures.“ 2 The lipids 1,2-di-O-
tetradecyI-sn-glycero-3-phosphocholine (DTPC), 1,2-di-O-tetradecyl-sn-glycero-
3-[phospho-rac-(1-glycerol)] (DTPG), 1-palmitoyl-2-oleoyI-sn-glycero-3-
phosphocholine (POPC), 1,2-Di-O-Hexyl-sn-Glycero-3-Phosphocholine (DHPC),
1 -palmitoyl-2-(1 6-fluoropalmitoyl)-sn-glycero-3-phosphocholine (16-19F-DPPC),
1-paImitoyl-2-oleoyl-sn-glycero-3-[phospho-rac-(1-glycerol)] (POPG) and 1,2-
DimyristoyI-D54-sn-Glycero-3-Phosphocholine (DMPCds4) were obtained from
Avanti Polar Lipids (Alabaster, AL). 1-paImitoyl-2-(5-fluoropalmitoyl)-sn-glycero-
3-phosphocholine (5-19F-DPPC) was custom synthesized by Avanti Polar Lipids
(Alabaster, AL). N-(2-hydroxyethyl)-piperazine-N’-2-ethanesulfonic acid (HEPES)
and 2-(N-morpholino)-ethanesulfonic acid (MES) were purchased from Sigma-
Aldrich (St. Louis, MO). The buffer solution used in the study contained 10 mM

HEPES/5 mM MES at pH 5.0 or 7.4 with 0.01% (w/v) NaN3 preservative.

23

2.2 SAMPLE PREPARATION
1. IFP and HFP Synthesis

All IFP contained the sequence GLFGAIAGFI_E_NGWEGMIDGGGKKKKG-
NH2 in which the underlined part is the 20 N-tenninal residues of the influenza A
X31 strain hemagglutinin fusion protein. All HFPmn had the sequence
AVGIGALFLGFLGAAGS-TMGAfiWKKKKKG-NHz and the underlined part is
the 23 N-terrninal residues of the HIV gp41 fusion protein. The HFPdm, the
abbreviation of a HFP dimer, had additional cysteines in the above sequence
which are used to cross-link HFP monomers into a dimer. The construction of a
HFP dimer is shown in Figure 7. A set of peptides were synthesized in order to
probe the conformation of the fusion peptide region. The labeling schemes of
synthesized IF Ps are listed in Table 1. The non-native Iysines increased aqueous
solubility and resulted in monomeric peptide in the buffer solution prior to
membrane binding.3 Peptides were either synthesized using an ABI 431A
peptide synthesizer (Foster City, CA) or synthesized manually in 5 mL
polypropylene columns from Pierce (Rockford, IL) and FMOC chemistry.
Peptides were cleaved from the resin for 2-3 hours using a mixture of
TFA:thioanisole:1,2-ethanedithiolzanisole in a 90:4:4:2 volume ratio. TFA was
removed from the cleavage filtrate with nitrogen gas and peptides were
precipitated with cold t-butyl methyl ether. The HFPdm was formed by cross-
linking two HFP monomers with a disulflde bond between the cysteines of two
peptide strands. All peptides were purified by reversed-phased high performance

liquid chromatography using a semi-preparative C13 column and a water-

24

Table 1. Labeling scheme of IFPS and HFPs

 

Peptide Labeled residues

lFP-G1c GIy-1 13co

IFP-L20 Lou-2 13cc

lFP-F3c Phe-3 13CO

lFP—A7cG8c Ala-7 and Gly-8 13co

lFP-G16c GIy-16 13co

IFP-6200 Gly-20 13co

lFP-A5cF9N Ala-5 13CO and Phe-9 15N

IFP-F9cG13N Phe-9 1"‘co and GIy-13 15N

IFP-G13cM17N GIy-13 13CO and Met-17 15N

IFP-G1” Gly-1 15N

IFP-LZN Lou-2 15N

lFP-F3N Phe-3 15N

lFP-G4N GIy-4 15N

lFP-A5N Ala-5 15N

IFP-I6” lie-6 15N

lFP-A7N Ala- 15N

IFP-G8" GIy-8 15N

IFP-FQN Phe-9 15N

IFP-HON lIe-10 15N

IFP-UN 15N for residues 1 - 10

lFP-F3NA7N Phe-3 and Ala-7 15N

lFP-(G1S)-F3NA7N IFP G1S mutant with Phe-3 and Ala-7 15N labels
lFP-(G1V)-F3NA7N IFP G1V mutant with Phe-3 and Ala-7 15N labels
IFP-I10E11u lie-10 and Glu-11 uniform 13c, 15N label
lFP-N12613u Asn-12 and Gly-13 uniform 13'c, 15N label
IFP-E11VcN12Ac IFP E11VN12A mutant with Val-11 and Ala-12 13CO labels
lFP-N12AcG13c IFP N12A mutant with Ala-12 and GIy-1 3 13CO labels
IFP-(E11VN12A)-A5cF9N IFP E11VN12A mutant with Ala-5 13CO and Phe—9 15N labels
lFP-(N12A)-A5cF9N IFP N12A mutant with Ala-5 13CO and Phe-9 15N labels
HFPmn-UN HFP monomer with 15N labels for residue 1 - 14
HFPmn-GSA6L7N HFP monomer with , Gly—5, Ala-6, Lou-7 15N labels
HFPmn-A14A15G16N HFP monomer with , Ala-14, Ala-15, Gly-16 15N labels
HFPdm-AGN HFP dimer with Ala-6 15N labels

 

25

acetonitrile gradient containing 0.1% TFA. Mass Spectroscopy was used for

peptide identification.

AVGIGALFLGFLGAAGSTMGARSWKKKKKCG
AVGIGALFLGFLGAAGSTMGARSWKKKKKCG

Figure 7. Construct of HFP dimer. Two HFP monomers cross linked at Cys to
form dimer.

 

(a) 1t/2
CW decoupling

 

1H Hopi

 

 

1t/2 ,

1H flan-l TPPM decoupling I
7t 1t 1t it it

. Acquisition

II II II II II a A ..=

rtnnnrtrt1VV

IIIIIII

I l I l I I 7 I I I
ROtOI' 0 1 2 3 4 5 6 7 8

. TC
15N/19F/31P E [I

Figure 8. Pulse sequences (a) Cross-polarization. Magnetization is transferred
from 1H to 13C or 5N during OF. 1H continuous wave (CW) decoupling was
applied during a uisition. (b) 13C REDOR. CP transfers the 1H transverse
magnetization to 1 C. 13C magnetization is dephased (i.e., reduced) by the
dipolar coupling between 13C and 1°N, 31P or 19F nuclei mediated by the 11 pulses
per rotor period. 13C chemical shift is refocused by multiple IT pulses.

26

2. Membrane sample preparation for MAS solid state NMR

Most vesicle samples were made with DTPC/DTPG lipids in a 4:1 mol
ratio. The reason for the major DTPC fraction was the Significant quantity of
phosphatidylcholine lipids in membranes of host cells of the influenza virus.4
Incorporation of the minor DTPG fraction was based on the small quantity of
negatively charged lipids in these membranes.4 In addition, PC/PG mixtures of
this approximate composition have been previously used in structural and
functional studies of the influenza fusion peptide and proteins; 11 Use of ether-
linked rather than ester-linked lipids Simplified the data analysis which was based
on the 11‘00 signal from the labeled IFP. Ether-linked lipids don’t have carbonyl
groups and the natural abundance 13CO Signal was therefore minimized.12 Some
samples were prepared with a “DTPC/DTPG/CHOL” (8:2:5 mol ratio) membrane
composition which reflected the significant fraction of cholesterol in the
membranes of the influenza virus and its host cells.“ 13 A few samples were
prepared with a “POPC/POPG” (4:1). Samples for the 130-1 9F REDOR
experiments were prepared with “DTPC/DTPG/16-19F-DPPC (or 5-19F-DPPC)”
(8:2:1 mol ratio) membrane composition. The incorporation of fluorinated lipids
provided Spin labels for the detection of the location of IFPS relative to the
membrane interior. The samples with 100% 1"F-DPPC lipid were not used
because a 100% 16-19F-DPPC lipid sample forms a non-bilayer structure.14
Studies with HFP and various mol fraction of 5-19F-DPPC Showed that 13CO—1"F

dephasing reaches a constant and maximum value over 0.07-0.14 mol fraction

27

range. Static 31P Spectra were consistent with overall bilayer structure in samples
containing 9 mol% 5-19F-DPPC and HFPS.1°

The sample preparation protocol began with dissolution in chloroform of
lipids (20 umol total) or lipids (20 umol total) and cholesterol. The chloroform was
removed under a stream of nitrogen followed by overnight vacuum pumping. The
lipid film was suspended in 2 mL aqueous buffer solution and was homogenized
with ten freeze-thaw cycles. Large unilamellar vesicles were formed by extrusion
through a 100 nm diameter polycarbonate filter (Avestin, Ottawa, ON).
Quantitation of IFP was done using A230 and 8230 = 5700 cm‘1 M‘ 1 and 0.8 umol
IFP was dissolved in 2 mL aqueous buffer solution which was then added to the
vesicle solution. The pH of the IFP/vesicle solution was adjusted to either 7.4 or
5.0 and the mixture was gently vortexed for a couple of hours and then
ultracentrifuged at ~1500009 for five hours. The membrane pellet with associated
bound IFP was transferred to a 4 mm diameter magic angle Spinning (MAS)
NMR rotor. The pH of the hydrated pellet was also checked before and after the

NMR experiments using pH paper.

3. Membrane sample preparation for static solid state NMR

Oriented bicelle samples were also prepared using predominant ether-
rather than ester-linked lipids because of the greater chemical stability of the
ether-linked lipids.1° DTPC (52.9 umol), DMPCd54 (1.1 pmol), and DHPC (16.9
umol) lipids were dissolved in chloroform in a round bottom flask. Bicelle

alignment was monitored using 2H NMR of the DMPCd54 lipids. IFP (1.1 umol) or

28

HFP (0.6 umol) was dissolved in a mixture of trifluoroethanol/chloroform (4:1) in
another round bottom flask and vortexed to ensure full solubilization before being
added to the lipid solution. Solvents were removed by a gentle nitrogen gas
stream followed by vacuum over night. Deuterium depleted water or aqueous
buffer solution (150 uL) was added to the dry IFP/lipid mixture. The pH was
adjusted to pH 7.4 or pH 5.0 with small aliquots of 1 M HCI or NaOH solutions.
The pH was checked before and after NMR measurement using pH paper and
did not change with time. Several cycles of vortex/freeze/thaw were carried out
until the solution tumed clear and flowed freely after being chilled in an ice-bath
for 5 minutes. Samples prepared by this protocol were transferred into glass
tubes with 5 mm diameter, 1 cm length, and ~120 uL volume. The tubes were
sealed with custom-machined Kel-F inserts covered with parafilm or rubber
plungers from 1 mL disposable syringes which can be pierced by a syringe to
remove air bubbles from the sample and ensure a tight seal. Samples placed in
the NMR Spectrometer had their bicelle normals aligned perpendicular to the
magnetic field direction at the temperature of 40 °C. Samples with bicelle normals
aligned parallel to the magnetic field direction were obtained with Yb3+ at 7 mM
concentration which was attained by adding ~5 uL 200 mM YbCl3 solution to the
bicelle solution.

Unaligned “powder" samples were prepared in a manner similar to that
used for the bicelle samples. The final sample composition was ~1 umol IFP or

HFP, ~54 umol DTPC, and ~150 uL water, and the pH was adjusted to 5.0 or 7.4.

29

4. N-acetyivaline Sample Preparation

Synthesis of 15N or 13CO-Iabeled N-acetylvaline (1°N-NAV or 13CO-NAV)
began by dissolving 1 mmol selectively labeled valine and 2 mmol acetic
anhydride in a mixture of 0.5 mL acetic acid and 0.5 mL water. The solution was
sonicated for 2 minutes. Additional cycles were done of addition of 2 mmol acetic
anhydride and 2 minutes of sonication. After each cycle, the presence of valine
was qualitatively monitored with addition of ninhydrin to a small aliquot of the
solution. The reaction was considered complete when the solution remained
clear after ninhydrin addition. Excess acetic anhydride and solvents were
removed with nitrogen gas and cyclohexane was added during this process to
facilitate the removal of water. The remaining solid 15N-NAV or 13CO-NAV was
dissolved in acetone. The container was wrapped in aluminum foil and after Slow
evaporation of the acetone, single crystals of 15N-NAV or 13CO-NAV were formed

with ~2 x 2 x 2 mm dimensions.

2.3 MAS SOLID-STATE NMR
1. General MAS NMR spectroscopy

Most experiments were done on a 9.4 T solid-state NMR spectrometer
(Varian Infinity Plus, Palo Alto, CA) using a triple-resonance MAS probe in
double resonance 1H/1‘°’C configuration for prTDQBU and PDSD experiments, in
triple resonance 1H/11‘C/15N configuration for 130—15N REDOR experiments and in
triple resonance 1H/31P/13C configuration for 130-31P REDOR experiments. For

the 130-19F REDOR experiment, a quadruple-resonance MAS probe was used in

30

triple resonance 1H/19F I13C configuration. Both probes were equipped for 4 mm
diameter rotors. The 1H, 19F, 13C, 31P and 15N frequencies were 398.7, 375.1,
161.5, 100.3, and 40.5 MHz respectively. The 13C and 15N shifts were externally
referenced to the methylene resonance of adamantane at 40.5 ppm and to
(15NH4)ZSO4 resonance at 26.5 ppm respectively. The 1°C referencing allowed
direct comparison with 130 shift databases derived from liquid-state NMR
assignments of proteins.“ 1° These databases are appropriate for solid-state
NMR data as evidenced by similar 130 shifts observed for the same protein in
either aqueous solution or the microcrystalline state.19'21 Samples were typically
cooled with nitrogen gas at -50 °C because 13C signals were larger at lower
temperature and motional averaging of dipolar couplings was reduced. Some
experiments were done at 0°C or ambient temperature to investigate the

temperature dependence of chemical shifts or internuclear dipolar coupling.

2. CP MAS spectroscopy

The CP MAS experiment provided chemical shift information in samples
that contain selectively 13CO or 15N labeled IFP. The CP pulse sequence is
displayed in Figure 8a. Simultaneous rf radiation of both the 8 (1H) and I(130 or
15N) nuclei such that both isotopes effectively precess at the same frequency
allows the transverse polarization transfer of 8 spin to l spin through
heteronuclear dipolar coupling.

For membrane bilayer samples that contained 1300 labeled IFPS, the 1H

and 130 pulse lengths were approximately obtained by direct pulsing on

31

adamantane. The CP matching condition was obtained by running ramped CP on
the lyophilized “l4” peptide which had sequence AcAEAAAKEAAAKEAAAKA-
NH2 and which was 13CO labeled at Ala-9 and 15N labeled at Ala-13.22 Calibration
of the 1H rT/2 pulse and 13C HQ and 11 pulses was done with the CP "Z-filtel"
sequence (CP — rT/2 - Tz — TT - acquisition). The parameters generally used were
~45 kHz rT/2 pulse, ~54 kHz CP and ~60 kHz CW decoupling for 1H and a 58-69
kHz ramped CP field for 111C. The contact time and recycle delay were 1 ms and
1 s respectively and phase cycling included 1H 11/2, x, —x, x, —x; 1H CP, y, y, y, y,
130 CP, —y, —y, —x, —x; receiver, —x, x, y, —y.

For samples that contained 15N labeled IFPS, the 1H and 15N pulse lengths
were approximately obtained by direct pulsing on 1°(NH4)2$O4 and the CP
matching condition was obtained by running ramped CP on 15N-NAV. The 1H rT/2,
15N rT/2 and 11 pulses were calibrated with CP ”Z-filter" sequence using the 15N-
NAV sample. The general parameters included ~51 kHz 1H rT/2 pulse, 1.0 ms
ramped cross-polarization with 23—26 kHz1°N and ~20 kHz 1H Rabi frequencies,
~50 kHz 1H decoupling. The recycle delay was 4 S and phase cycling was the

same as listed above for the CP MAS experiments of 1300 labeled samples.

3. REDOR spectroscopy

The REDOR experiment provided structural information in samples that
contained a 13CO and a heteronuclear spin (e.g. 15N, 19F or 31P) labels. The
REDOR sequence is shown in Figure 8b, which includes: (1) a 1H rr/2 pulse; (2)

a cross-polarization period on-resonance 1H and 13C fields; (3) a dephasing

32

period of duration rduring which there is one 1°C 11 pulse at the beginning of
each rotor cycle except for the first cycle and for some acquisitions, a
heteronuclear spin (e.g. 15N, 19F or 31F) 11 pulse in the middle of each cycle; and
(4) 13C detection. For each sample and eacht; two spectra were acquired. The
heteronuclear pulses were absent during the dephasing period of the “So”
acquisition and during each rotor cycle, MAS averaged to zero the 13C evolution
due to heteronuclear dipolar coupling (e.g. 130—15N, 130—19F or 13C—31P). During
the “81” acquisition, incorporation of one heteronuclear 17 pulse per rotor cycle
disrupted the MAS averaging and resulted in reduction of Signals of 130 nuclei
close to heteronuclei. Spectra were acquired for different 1; and XY-8 phase
cycling (x, y, x, y, y, x, y, x) was used for the heteronuclear 11 pulses and for all of
the 13C TT pulses except the final pulse. Individual So or 81 transients were added
with phase cycling: 1H rT/2, x, —x, x, -x; 1H CP, y, y, y, y, 1°C CP, —y, -y, -x, -x,
final 130 1T pulse, -y, -y, -x, -x; receiver, —x, x, y, —y. The differences in the 13C
Signal intensities of the So and 81 Spectra as a function of 1' were used to
determine the dipolar coupling (d) and the distance (r) between the labeled nuclei.

For the 130-15N REDOR experiments, the 1H, 1°C and 15N pulse lengths
and the 1H-13C cross-polarization matching condition were calibrated with the
lyophilized “I4” peptide mentioned before. The typical experimental parameters
included 8000 Hz MAS frequency, 44 kHz 1H n/2 pulse, 54 kHz 1H field and 58-
69 kHz ramped ‘30 field during the CP period, 62 kHz 130 rrpulse, 28 kHz 1°N 7r

pulse and 93 kHz two- pulse phase modulation (TPPM) 1H decoupling during the

33

dephasing and detection periods. The durations for the cross-polarization period
and recycle delay were 1 ms and 1 s, respectively.

The 13C—31P REDOR experiments provided information on the proximity of
the IFP backbone relative to the membrane surface which has 31F nuclei. The CP
matching condition and the 1H 1T/2, 13C TT/2 and 17 pulses were calibrated with the
lyophilized “I4” peptide. The 31P n pulse length was set with the CP "Z-filter"
using the 31P signals from the phosphate headgroups of the membrane samples.
The typical parameters included 8000 Hz MAS frequency, ~50 kHz 1H rT/2 pulse,
52 kHz 1H field and 58-69 kHz ramped 130 field during the 1 ms CP period, ~50
kHz ‘30 rrpulse, ~60 kHz 31P rrpulse, ~1oo kHz TPPM 1H decoupling during the
dephasing and detection periods and 1 S recycle delay.

The 130—19F REDOR experiments provided information on the proximity of
IFP backbone relative to either the membrane center (16-19F) or the midpoint
between the membrane surface and membrane center (5-19F). In these
experiments, the membrane bilayers were incorporated with some DPPC lipid
which has 19F substituted for a 1H at either the terminal C16 or the CS position of
the sn2 acyl chain. The 16-19F is located at the membrane center while the 5-19F
is located at the midpoint of the bilayer longitude between the membrane surface
and bilayer center. The experimental parameters were validated using a
lyophilized sample containing helical peptide which had the sequence
EQLLKALEFLLKELLEKL with 13CO label at Leu-10 and Phe-9 was substituted
with p-fluorophenylalanine.23 The typical parameters included 8000 Hz MAS

frequency, ~50 kHz 1H "/2 pulse, ~50 kHz 1H field and 55-66 kHz ramped 13C

34

field during the 1 ms CP period, ~50 kHz 13C rrpulse and ~33 kHz 19F 7r pulse
during the dephasing period, ~95 kHz TPPM 1H decoupling during the dephasing

and detection periods and 1 S recycle delay.

4. prTDQBU spectroscopy

The prTDQBU experiment was used to determine the 13CO-13CO dipolar
coupling and distance in samples in which the IFP was 1300 labeled at two
adjacent residues. The prTDQBU sequence can be represented as CPgog —
(prFDR)L — TT/2; — 11/20 - (prFDR)M — "/2130 - 1T/2go - (prFDR)~ - acquisition
where L, M and N refer to the number of rotor periods in each prFDR period
and other subscripts refer to the TI phases (Figure 9). The sequence was
implemented in a manner Similar to the REDOR sequence with So and $1 spectra
and a dephasing period r= 2L1p where TR was the duration of a Single rotor cycle.
Increasing values of rwere obtained by incrementing L and decrementing N by
the same number while keeping M constant. For each value of rand each C = 0,
90, 180, or 270, a distinct Spectrum was obtained and the So and S1 Spectra were

30 = Sg=go + ngzm and S1 = Sg=o 4' Sg=130.22

35

(a)

 

1t/2

 

 

 

 

 

 

 

 

 

 

 

 

 

 

CW dec I'
1H OF 0UP '09 CW decoupling
(1x '1? 3 :
LtR MT R N1: R Acqursrtion
13C CP :\ :
I I) A ;
!: CT :! V V

(b)
X Y X y y x y x

 

Rotor [I I l A I l | I l I]
1 2 3 4 5 6 7 8 9
TR L/8, M/8 or N/8 repeats

Figure 9. prTDQBU pulse sequence. (a) The displayed prTDQBU pulse
sequence included: (1) CP from 1H to 13C; (2) a constant-time (CT) period of (L +
M + MTR duration where L, M and N were integral multiples of 8, and TR was the
duration of a single rotor period; and (3) acquisition period. Continuous wave 1H
decoupling was applied during the CT and acquisition periods. A pair of back-to-
back 13C rT/2 pulses separated the LTR and MTR periods and another pair
separated the MTR and NTR periods. (b) Pulses during LTR, MTR and NTR periods.
There was one 11 pulse per rotor cycle. The 13C OP rf phase was 6 + rT/2 and the
phase of each 13C rT/2 and 11 pulses iS noted above the pulse. For 4’ = y or —y, the
evolution of the transverse 13C magnetization due to 130-13C dipolar coupling was
averaged to zero and 80 data were obtained and for 4’ = x or —x, the evolution
was not averaged to zero and 81 data were obtained. The duration of 1°C
recoupling or dephasing period T = 2LTR, and data with increasing T were
obtained by incrementing L and decrementing N while keep constant M and
constant (L + M + N).

36

For the experimental parameter calibrations of prTDQBU, a
polycrystalline N-acetyl leucine (NAL) sample was used which was 130 labeled at
the acetyl and carboxyl positions. Typical experimental parameters included
8000 Hz MAS frequency, 46 kHz 1H rT/2 pulse, 1.0 ms cross-polarization period
with 54 kHz 1H field and 58-69 kHz ramped ‘30 field, 22 KHz13C 11 pulses with
XY-8 phase cycling, continuous wave 1H decoupling during the evolution and
acquisition periods with Rabi frequencies of ~70 and ~60 kHz, respectively, and
1.5 s recycle delay. The experiment was done with “constant time” so that for all

values of r, M= L + N.

5. PDSD spectroscopy

The proton-driven spin diffusion (PDSD) experiment provided 130/130
correlation Spectra for membrane-associated IFPS. The 13C chemical shifts from
the cross peak assignments can be used to determine the secondary structure of
IFP. The pulse sequence is Shown in Figure 10 which contains an initial 1H TT/2
pulse followed by a 1H—13C CP, an evolution period t1, a 13C TT/2 pulse that
rotated the 13C transverse magnetization to the longitudinal axis, a spin diffusion
period T during which 13C magnetization was mixed among nearby nuclei, a
second 13C rT/2 pulse that rotated the 13C magnetization back to the transverse
plane, and a detection period t2. A ~70 KHz 1H decoupling field with TPPM was
applied during t1 and t2, but not during 7.24 The 1H and 13C pulses were calibrated
with a NAL sample which was uniformly 13C labeled. The following parameters

were typical for PDSD experiments: 10000 kHz MAS frequency, 36-42 kHz ramp

37

on the ‘30 CP rffield, 56 kHz 1H CP rf field, 25 us 11 dwell time, 200 11 values, 20
us t2 dwell time, 256 ii points, 10 ms exchange time, and 1 S recycle delay.
Hypercomplex data were obtained by acquiring two individual FIDS for each t1
point with either a 13C (iT/2)x or (Tr/2)y pulse at the end of the t1 evolution period.
For the first of these t1 F IDS, individual transients had the following phase cycling
scheme: first 13C rT/2 pulse, Ix, -x, x, -x, x, —,x x,-x,second130 TT/2 pulse, x, x, y,
y, —x, -x, -y, —y, receiver, y, —y, -x, x, —y, y, x, —x. For the other t1 FIDS, the first

1°C IT/2 pulse followed y, -y, y, —y, y, —y, y, —y cycling.

6. Experimental data analysis

For the REDOR and prTDOBU experiments, the differences in the 13C
signal intensities of the So and 81 spectra at different 1 can be used to calculate
the internuclear dipolar coupling and distance which are quantitatively related to
each other. For a single spin labeled pair, the dipolar coupling d in Hz and

distance r in A are related by the following equations:

rC~=—C(3100/d~()1/31°C—1°N pair) (2.1)
rep: -C(12250/d p)1/3( 130—31P pair) (2.2)
rep: —C(28540/d F)1/3( 130—19F pair) (2.3)
’00: —C(7700/d C)(1/3 130—130 pair) (2-4)

These determinations relied on integrations of the 1300 regions of the So
and 81 spectra and the resulting integrated intensities were denoted “So” and “81”

respectively. An experimental fractional dephasing

38

(AS/so)” = (33” _ 3fo VS?” (2.5)

was calculated for each 1'. Determination of d also relied on calculation of
theoretical dephasing (AS/So)°”" for different values of d and 1: For the REDOR

data, the theoretical dephasing was calculated:

(AS/80f“ =1’[Jo (722)]? +225: [Jk (NE/1)]

k=1 16k2-1

 

(2.6)

where zl=dr and Jr, is the k1h order Bessel function of the first kind.25 Eq. 2.6 is
based on a model of a Single spin pair separated by a single distance. This was a
reasonable model for the labeled spin pair which made the dominant contribution
to (AS/So)”.

For the prTDQBU data, (AS/So)” values were calculated using the
SIMPSON program, which is a computer program for a fast and accurate
numerical simulations of multiple-pulse NMR experiments. It functions as a
“computer Spectrometer" by using NMR concepts such as spin systems, nuclear
Spin interactions, phase cycling and by using specific NMR acquisition and
processing parameters. The simulation also uses a Single Spin-label pair model
and literature-based 13CO chemical shift anisotropy (CSA) principal values.” 2°

Quantitative comparison between (AS/So)” and (AS/So)°”" at different

values of d and 1' provide information on the best-fit internuclear distances.

However, there were also contributions to (AS/So)” from natural abundance

12, 22

nuclei. AS described in the appendix and other publications, models have

been developed to calculate these natural abundance contributions. For each

39

value of (AS/So)”, a “corrected” (AS/So)” was calculated that only reflected the
labeled pair.

AS an example, calculations of (AS/So)” for IFP-A7cG8c were based on
the following parameters/approximations:
(1 ). There is 99% labeling of the Ala-7 and Gy-8 13CO Sites.
(2). 81 = 0 for Ala-7 13CO/natural abundance 13C or Gly-8 13CO/natural
abundance 13C separated by one or two bonds. The 81 is not affected by other
natural abundance 1°Cs.
(3). For natural abundance backbone 13CO Sites, 81 = 80.

Each (AS/So)” value is calculated using the parameters Uc, Ac, n, m and
m; where: Uc is the fraction of Ala-7 12CO or Gly-8 12CO sites; Ac is the fractional
13C natural abundance; n is the number of unlabeled CO sites; and m and m
are the number of natural abundance sites which satisfies (2) for Ala-7 and Gly-8
respectively. AS described in the appendix as well as in the literature, for each
value of (AS/So)°"", a “corrected” (AS/So)” was calculated that only reflected the
labeled pair.
[Airy = 2 " ”c + "Ac [Era - "11’40 1 ”2A0 (2.7)

S0 (1'Uc)(2‘m1Ac—m2Ac) S0 2-m1Ac+m2AC

The values of Uc, Ac, n, m and M2 are 0.01, 0.011, 26, 3 and 2, respectively,
and numerical evaluation of Eq. 2.7 yields:

cor exp
[E] = 1.177 [AS—J - 0.028 (28)
S0 So

40

The calculations of (AS/So)” for the REDOR data are detailed in the appendix.
The (AS/So)cor has the general form ax(AS/SO)°"”—b, where a and b are

positive numbers.
The quantitative comparison between (AS/So)” and (AS/So)°”" requires
the knowledge of experimental uncertainties. Uncertainties of So and 81 spectra,

050 and as] , were calculated as the root-mean-squared deviations of integrated

intensities in 12 regions of the So and Si spectra without signal. The uncertainties

in (AS/So)°"° were calculated:

 

 

 

 

2 2 2 2 2
exp = 031 SI 0'30 = fl 031 0'30
0' 2 1' 4 2 + 2 (2'9)
S0 S0 S0 S1 30

The values of 0'°°’ were calculated from the 0'1”“D and had the form of axa°xP .

The (AS/So)” could be directly compared to (AS/So)” using a ,1? analysis:

AS cor AS slm 2
s71 ‘ [EWI- I
(M

where the sum is over the experimental r values. The best-fit distance d

(2.10)

 

220mg“

corresponded to the minimum x2(d) in Eq. 2.10. The uncertainty in the best-fit
value was calculated using the d values at f = (16min + 5 which is a generous

statistical criterion.

2.4 STATIC SOLID-STATE NMR

1. General static NMR spectroscopy

41

Alignment of the bicelles was probed with measurement of the anisotropic
quadrupolar splittings of 2H nuclei in the lipid acyl chains or anisotropic 31P
chemical Shift of the lipid head group. Alignment of the IFP helix axis relative to
the bicelle normal was probed with measurements of the anisotropic chemical
shifts of labeled 1"’N nuclei in the IFP backbone or the dipolar coupling of the
directly bonded 1°N-1H pair. Most of the NMR Spectra were obtained with the
aforementioned 9.4 T Spectrometer and a Varian Biostatic 1H/X probe with 5 mm
coil which was designed for minimizing RF heating in the sample. Some of the
chemical shift measurements were done with a regular static 1H/X probe which
requires a much longer recycle delay compared to the Biostatic probe in order to
reduce the RF heating. The 2H frequency was 61.2 MHz. For some of the
SXperimentS, a 21.1 T solution/solid state NMR Spectrometer (Bruker Avance,
Billerica, MA) equipped with “Efree” static 1H/15N (31F) probe was used. The
Bruker “Efree” probe is similar to the Varian Biostatic probe but uses a different
design to minimize RF heating. The 1H and 15N frequencies were 899.9 and 91.2
MHz respectively. For all the experiments, the 13C, 15N and 31P Shifts were
respectively externally referenced to the methylene resonance of adamantane at
40.5 ppm, to (15NH4)2SO4 at 26.5 ppm or to H3PO4 at 0 ppm. The bicelle samples
were typically heated with nitrogen gas at 40 °C which is near the middle of the

temperature range of stable bicelle structure.

2. 2H and 31P spectroscopy

42

 

Tt/2

1H II CP | CWdecoupling I

Tt/23 rt/23

 

 

CW decoupling

 

 

 

' Acquisition
13° 61 11 11 I I] AZ '
“ V VA"

Figure 10. PDSD pulse sequence. Magnetization was transferred from 1H to 13C
during the CP period. CW decoupling was applied after CP on the 1H during t1
and t2, but not during T. t1 was the evolution time, T was the magnetization
exchange time, and t; was the acquisition time.

Tt/2

“/2
1” “GP “HIHIIIIIH TPPMdecoupling
7t/2 2'

op lllllllllll \ 21°:

Figure 11. Pl-WlM-z pulse sequence. After the magnetization transfer from 1H to
15N during the CP period, the 77/2 pulses on each channel flip 1H and 15N
magnetization to opposite directions to achieve polarization inversion. The
multiple back-to-back 77/2 pulses during the WIM period enable the spin
exchange between the 1H and 15N nuclei and suppress the 1H-1H dipolar
coupling and chemical Shift evolution. TPPM decoupling was applied during the
acquisition period.

 

 

 

 

 

 

 

 

43

The 2H spectra were obtained without 1H decoupling and with a (rt/2)o — 1'1
— (rt/2m — 12 - acquisition solid echo sequence. The sequence was used to
minimize effects from probe ring-down. The phase of the first rri2 pulse was x
and the phase of the second 77/2 pulse alternated between y and —y.
Experimental parameters included 33 kHz 2H Rabi frequency, 50 us 11, 25 12 and
0.5 S recycle delay. The 77/2 pulse was calibrated with a water sample containing
10 % of deuterium oxide.

For 31P experiments, Bloch decay Spectra were taken using a recycle

delay of 3 S and CYCLOPS phase cycling. The 31P Rabi frequency was 65 kHz

and was calibrated with a sample of phosphoric acid.

3. 15N chemical shift measurements

The 1"’N spectra were obtained with a sequence that contained 1H—15N
cross-polarization followed by 15N detection with 1H decoupling. The rf fields were
calibrated using a 1°N-NAV Single crystal. For experiments performed with the
regular static probe on the 9.4 T spectrometer, typical experimental parameters
included a ~50 kHz 1H 77/2 pulse, 1.0 ms ramped cross-polarization with 18—22
kHz 15N and ~20 kHz 1H Rabi frequencies, ~40 kHz 1H decoupling, and 7 s
recycle delay. The long recycle delay reduced rf heating of the samples.

For experiments done with the Biostatic probe on the 9.4 T Spectrometer,
a ~42 kHz 1H 77/2 pulse, a 1 ms cross polarization with ~ 40 kHz 1H field and 38-
44 kHz 15N field and a ~42 kHz 1H decoupling field during acquisition were used.

The recycle delay was 1 S.

For experiments done with the 21.1 T solution/solid state NMR
spectrometer, the parameters included ~38 kHz 1H 77/2 pulse, 1.0 ms cross
polarization with ~33 kHz 1H and 15N Rabi frequencies, and ~28 kHz SPINAL-16
decoupling. During the CP period, a ramp 80.100 was added on the 1H channel.

The recycle delay was 1 S.

4. 15N chemical shift 8. 15N-1H dipolar coupling cone/etion spectroscopy
The heteronuclear dipolar spin interactions such as 1H-15N dipolar
coupling can provide valuable information for the structural and dynamic
investigations of anisotropic solid samples. The high-resolution dipolar coupling
spectrum can be obtained by 2D PISEMA-type experiments which are based on
the heteronuclear spin exchange via the local field with efficient homonuclear
proton decoupling. In my research, the 20 correlation spectra were acquired by
the 2D PI-WlM-z (polarization inversion - windowless isotropic mixing —
polarization exchange between z-component of magnetization) technique which
was designed to have high tolerance towards chemical Shift and frequency
offsets and iS more suitable for measurements of motionally reduced dipolar
couplings.27 The sequence is shown in Figure 9. After the regular cross
polarization period, a 77/2 pulse is applied to 1H and 15N respectively to prepare
the two spin polarizations in opposite directions to achieve polarization inversion
(Pl) which can enhance the experimental sensitivity. The WIM-z sequence

(WIM24) is composed of on-resonance back-to-back 77/2 pulses with phase

cycling. -y. X. -y. -y. X. -y. y. X. y. y. X. y. -y. -X. -y. -y. -X. -y. y. -X. y. y. -X. y. which

45

are simultaneously applied to both nuclei. The spin exchange between the 1H
and 15N magnetizations occurs during this period in much the same way that CP
transfers the spin locked components and the oscillation frequency was
determined by the 15N—‘H dipolar coupling. At the same time, the homonuclear
dipolar coupling and chemical shift evolution were suppressed by the WIM-z
sequence. The resulting scaling factor of the dipolar coupling is 0.67. in this
experiment, the t1 increment is the duration of the 24 11/2 pulses of the WIM-z
sequence which provides a ~7 kHz spectral window to record the dipolar splitting.
This width can be doubled by setting t1 increment to half of the WlM24 cycle, i.e.,
WlM12.

The 1H and 15N pulses were calibrated with a N15-NAV single crystal
sample. The following parameters were typical for Pl-WlM-z experiments
performed with the 9.4 T spectrometer: ~42 kHz 1H and 15N 11/2 pulses, 1 ms
cross polarization with ~ 40 kHz 1H field and 38-44 kHz 15N field, ~42 kHz 1H
decoupling field and 1 s recycle delay. For most of the experiments, WIM12 was
generally used. For some experiments, WIM24 was also used to indentify
aliasing. The parameters for experiments done with the 21.1 T spectrometer are:
~38 kHz 1H and 15N 17l2 pulses, 1.0 ms cross polarization with ~33 kHz 1H and
15N Rabi frequencies (ramp on the 1H frequency channel with ramp80.100), ~28
kHz SPINAL-16 decoupling and 1 s recycle delay. WIM24 was used for the

experiments.

5. Static 15N NMR analysis

46

Nuclear spin interactions are anisotropic which provides the basis for the
extraction of the peptide orientation in a lipid bilayer. Specifically, the chemical
shift and dipolar coupling frequencies depend on the relative orientation of the
corresponding tensors (i.e., chemical shift and dipolar coupling tensors) and the
external magnetic field.

The major spin interactions for the 15N nucleus in the rigid lattice are

chemical shift anisotropy (HCSA) and 15N-1H dipolar coupling (Horp), which can be

written as:28

H=HCSA +HolP (2-11)
where

HCSA = 7~hB - (l—a) - S (2.12)
and

H0... = %¥%H-(1-30082 :9)szrz (2.13)

B is the magnetic field, N and y" are the gyromagnetic ratios of 15N and 1H, and
S and I represent the 1 N and 1H spin angular momentum. The chemical shift
tensor, 0, in the principal axis system is given by

011 0 O
0 = 0 0'22 0 (2.14)
O 0 0'33

The unit matrix 1 is

(2.15)

l...
ll
00-8
O-BO
400

The 15N-1H internuclear distance is r, and N-H vector makes an angle of dwith

the magnetic field.

47

For an ideal a-helical structure with dihedral angles (,0 = -65° and w = -40°,
the chemical shift and dipolar coupling frequencies can be calculated using the

following equations:29

0( p, r) = 011(—0.828cos,osinz' + 0.558 sin psinr - 0.047cosz')2
+ 022 (0.554cos p sinr + 0.803 sin psinr — 0.2200037)2 (2.16)
+ 033 (-0.08800$psinr - 0.206 sin p sinr — 0.975 cos r)2

V
v( p, r) = 311(3(—0.32600$psinr — 0.034sin p sinr — 0.946cosr)2 — 1) (2.17)

where 011, 022 and 033 are the principal values of the chemical shift tensor and v“
is the rigid limit 15N-1H dipolar splitting, or dipolar coupling constant v“ =
(npd4anNyH/r1), which is 22.6 kHz assuming an internuclear separation of 1.024
A. In the calculations, standard values for the relative orientation of the dipolar
and chemical shift tensors 0 was assumed (cf. Figure 12a). The axis 033 is in the
peptide plane and makes an angle of ~17° from the N-H vector and points
between the NH and N-CO bonds; 011 is perpendicular to 033 and points
between the NH and N-Ca bonds; 022 is perpendicular to the peptide plane and
follows the right hand rule. The two Euler angles, p and 1' (cf. Figure 12b)”,
represent the rotation of a particular residue about the helix axis and the tilt of the
helical axis with respect to the magnetic field, respectively. For an ideal helix with
a periodicity of 3.6 residues per turn, the increment of p is 100° for each

consecutive residue.

48

(a) 033 f?“ (b) helix

   

 

  

 

 

axis
H.~'
3 N *022 ‘
2/0'15 V-
ca E.
: >—
rotation ‘go
axis - : ““
Ca

 

 

 

011

Figure 12. (a) The 15N chemical shift tensor relative to the peptide plane. Axes
011 and 033 are in the peptide plane with 033 pointing between the N-H and N-CO
bonds, ~17° away from N-H vector and 011 pointing to N-Ca. Axis 022 is
perpendicular to the peptide plane. (b) Definition of helix rotation (p) and tilt (r)
angles. The helix axis is defined by h3 and hr is the radial axis that passes
through Ca carbon and varies for different residues. The helix tilt angle, 1, is the
angle between the helix axis and the magnetic field (B0). The helix rotation p is
the angle between m and the projection of 80 on to the plane perpendicular to h3.
(c) Definition of angles in the system in which the helix has fast rotation around a
fixed axis. Angles a and [3 are the Euler angles ,of the rotation axis and the
chemical shift principal axis system (PAS). The angle between the N-H vector
and 033, the N-H vector and the rotation axis, the N-H vector and Bo, and the
rotation axis and Bo are represented by C, n, 19 and 0, respectively.

49

In a real biological system, there are generally a lot of motions associated
with the proteins or membranes, which cause the motional average of spin
interactions when the rates of motions are faster than those of the NMR
interaction time scale (~10'5 sec based on the typical magnitude of NMR
interaction, ~10 kHz). There are two basic motions of membrane-associated
protein, fast rotation and libration. In the case of fast rotation around a fixed axis,

the rigid lattice chemical shift tensor will be averaged to an axially symmetric

tensor ages,28
0 L 0 0 -1 0 0
JEFF = O 01 0 = Oil '1' (1/3) AO'EFF 0 —1 0 (2.18)
0 0 0'” 0 0 2

where 01 and 0H are the components perpendicular and parallel to the rotational

axis. The isotropic chemical shift 0. is
1 1
0'] =§(0'11+O'22+033)=3(20'_L+0'”) (2.19)

And the effective chemical shift anisotropy A05”: can be written as28

1 1
AUEFF = 0” - 0t = *(30032 fl—1)(033 “-(0'11+ 0'22 ))
2 2
3 (2.20)
+z(011—022)sin2 ficosZa
where a and B are the Euler angles between the rotational axis (cf. Figure 12c)
and the principal axes of the rigid lattice chemical shift tensor.

Therefore, HCSA in the case of fast rotational motion is
HCSA = YNhB ' (1 ’ aEFF) ' S (2'21)

and yields the chemical shift28

50

2
0(6) = EAO'EFFPZ (C086) 4' (TI-so (2.22)

where 9 is the angle between the rotational axis and the extemal magnetic field

and P2(cos 0) is a Legendre polynomial

P2(cosl9) = (3cos2 9—1) (2.23)

l
2

Similarly, the observed dipolar coupling can be calculated as

v(6l) = v” %(30032 n-1)P2 (ooso) (2.24)
where
cosn = cos flcosg“ - sin flsin {cosa (2.25)

Angle n is between the N-H vector and the rotational axis; C is between the N-H
vector and 033.

All the experimental data for aligned samples were taken with bicelles
which are considered to have rapid small angle “wobbling” of the bicelle normal
with respect to the magnetic field direction. A parameter describing the
magnitude of wobbling, Swobb, was estimated as the ratio of a 2H quadrupolar
splitting of DMPC354 in a bicelle sample relative to the splitting in a mechanically
aligned bilayer sample that lacks wobbling motion. The value of 8...,» is 0.8.30
This wobbling motion will scale down the effective chemical shift anisotropy and
the maximum dipolar coupling. In addition, there is a scaling factor (K = 0.67)
associated with the homonuclear decoupling sequence used in the 2D Pl-WlM-z
experiment, which also reduces the dipolar coupling. Therefore, the 15N chemical

shift and dipolar coupling in a wobbling bicelle become:28

51

2
d 0 =—. e
( ) 3 AOEFF P2(0039)‘Swobb + O'iso (2-25)

1
V 6 = I I- 2 —
( ) K' v" 2(3cos n 1)P2(cos0)-Swobb (2.27)

52

2.5 REFERENCES

1. Chang, C. D.; Waki, M.; Ahmad, M.; Meienhofer, J.; Lundell, E. 0.; Haug,
J. 0., Preparation and properties of N-a-9- fluorenylmethyloxycarbonylamino
acids bearing tert-butyl side chain protection. Int. J. Pept. Protein Res. 1980, 15,
(1). 59-66.

2. Lapatsanis, L.; Milias, G.; Froussios, K.; Kolovos, M., Synthesis of N-
2,2,2-(trichloroethoxycarbonyl)-L-amino acids and N-(9-
fluorenylmethoxycarbonyl)-L-amino acids involving succinimidoxy anion as a
leaving group in amino-acid protection. Synthesis-Stuttgart 1983, (8), 671 -673.

3. Yang, J.; Prorok, M.; Castellino, F. J.; Weliky, D. P., Oligomeric beta
structure of the membrane-bound HIV-1 fusion peptide formed from soluble
monomers. Biophys. J. 2004, 87, 1951 -1 963.

4. Worman, H. J.; Brasitus, T. A.; Dudeja, P. K.; Fozzard, H. A.; Field, M.,
Relationship Between Lipid Fluidity and Water Permeability of Bovine Tracheal
Epithelial-Cell Apical Membranes. Biochemistry 1986, 25, (7), 1 549-1 555.

5. Han, X.; Bushweller, J. H.; Cafiso, D. S.; Tamm, L. K., Membrane
structure and fusion-triggering conformational change of the fusion domain from
influenza hemagglutinin. Nat. Struct. Biol. 2001, 8, (8), 715-720.

6. Macosko, J. C.; Kim, C. H.; Shin, Y. K., The membrane topology of the
fusion peptide region of influenza hemagglutinin determined by spin-labeling
EPR. J. Mol. Biol. 1997, 267, (5), 1139-1148.

7. Li, Y.; Han, X.; Tamm, L. K., Thermodynamics of Fusion Peptide-
Membrane Interactions. Biochemistry 2003, 42, 7245-7251.

8. Han, X.; Tamm, L. K., pH—dependent self-association of influenza
hemagglutinin fusion peptides in lipid bilayers. J. Mol. Biol. 2000, 304, (5), 953-
965. .

9. Bodner, M. L.; Gabrys, C. M.; Struppe, J. 0.; Weliky, D. P., 13C—13C and
15N—13C correlation spectroscopy of membrane-associated and uniformly
labeled human immunodeficiency virus and influenza fusion peptides: Amino
acid-type assignments and evidence for multiple conformations. J. Chem. Phys.
2008, 128, 052319.

10. Parkanzky, P. D., Solid state nuclear magnetic resonance studies of the

influenza fusion peptide associated with membrane bilayers. Ph. D. thesis,
Michigan State University: East Lansing, 2006, 80.

53

11. Wasniewski, C. M.; Parkanzky, P. D.; Bodner, M. L.; Weliky, D. P., Solid-
state nuclear magnetic resonance studies of HIV and influenza fusion peptide

orientations in membrane bilayers using stacked glass plate samples. Chem.
Phys. Lipids 2004, 132, (1 ), 89-100.

12. Qiang, W.; Yang, J.; Weliky, D. P., Solid-state nuclear magnetic
resonance measurements of HIV fusion peptide to lipid distances reveal the
intimate contact of beta strand peptide with membranes and the proximity of the
Ala-14-GIy-16 region with lipid headgroups. Biochemistry 2007, 46, (17), 4997-
5008.

13. Lenard, J.; Compans, R. W., The membrane structure of lipid-containing
viruses. BBA-Biomembranes 1974, 344, (1 ), 51-94.

14. Hirsh, D. J.; Lazaro, N.; Wright, L. R.; Boggs, J. M.; McIntosh, T. J.;
Schaefer, J.; Blazyk. J., A New Monofluorinated Phosphatidylcholine Forms
lnterdigitated Bilayers. Biophys. J. 1998, 75, (4), 1858-1868.

15. Pereira, F. B.; Valpuesta, J. M.; Basanez, G.; Goni, F. M.; Nieva, J. L.,
lnterbilayer lipid mixing induced by the human immunodeficiency virus type-1

fusion peptide on large unilamellar vesicles: the nature of the nonlamellar
intermediates. Chem. Phys. Lipids 1 999, 1 03, (1 -2), 1 1-20.

16. De Angelis, A. A.; Nevzorov, A. A.; Park, S. H.; Howell, S. C.; Mrse, A. A.;
Opella, S. J., High-resolution NMR spectroscopy of membrane proteins in
aligned bicelles. J. Am. Chem. Soc. 2004, 126, (47), 15340-15341.

17. Morcombe, C. R.; Zilm, K. W., Chemical shift referencing in MAS solid
state NMR. J. Magn. Reson. 2003, 162, 479-486.

18. Zhang, H. Y.; Neal, 8.; Wishart, D. S., RefDB: A database of uniformly
referenced protein chemical shifts. J. Biomol. NMR 2003, 25, (3), 173-195.

19. Igumenova, T. l.; Wand, A. J.; McDemlott, A. E., Assignment of the
backbone resonances for microcrystalline ubiquitin. J. Am. Chem. Soc. 2004,
126, (16), 5323-5331 .

20. Franks, W. T.; Zhou, D. H.; Wylie, B. J.; Money, B. G.; Graesser, D. T.;
Frericks, H. L.; Sahota, G.; Rienstra, C. M., Magic-angle spinning solid-state
NMR spectroscopy of the beta 1 immunoglobulin binding domain of protein G
(GB1): N-15 and C-13 chemical shift assignments and conformational analysis. J.
Am. Chem. Soc 2005, 127, (35), 12291-12305.

21. Marulanda, D.; Tasayco, M. L.; Cataldi, M.; Arriaran, V.; Polenova, T.,
Resonance assignments and secondary structure analysis of E-coli thioredoxin

54

by magic angle spinning solid-state NMR spectroscopy. J. Phys. Chem. B 2005,
109, (38), 18135-18145.

22. Zheng, Z.; Yang, R.; Bodner, M. L.; Weliky, D. P., Conformational flexibility
and strand arrantments of the membrane-associated HIV fusion peptide trimer
probed by solid-state NMR spectroscopy. Biochemistry 2006, 45, 12960-12975.

23. Taylor, K. 8.; Lou, M. 2.; Chin, T. M.; Yang, N. C.; Garavito, R. M., A novel,
multilayer structure of a helical peptide. Protein Sci. 1996, 5, (3), 414-421.

24. Bennett, A. E.; Rienstra, C. M.; Auger, M.; Lakshmi, K. V.; Griffin, R. G.,
Heteronuclear decoupling in rotating solids. J. Chem. Phys. 1995, 103, (16),
6951-6958.

25. Mueller, K. T., Analytical Solutions for the Time Evolution of Dipolar-
dephasing NMR Signals. J. Magn. Reson. Series A 1995, 1 13, (1 ), 81 -93.

26. Bak, M.; Rasmussen, J. T.; Nielsen, N. C., SIMPSON: A general
simulation program for solid-state NMR spectroscopy. J. Magn. Reson. 2000,
147, (2), 296-330.

27. Dvinskikh, S. V.; Yamamoto, K.; Ramamoorthy, A., Separated local field
NMR spectroscopy by windowless isotropic mixing. Chem. Phys. Lett. 2006, 419,
168-173.

28. Hemminga, M. A.; Cullis, P. R., 31P NMR Studies of Oriented
Phospholipid Multilayers. J. Mag. Reson. 1982, 47, 307-323.

29. Wang, J.; Denny, J.; Tian, C.; Kim, 8.; Mo, Y.; Kovacs, F.; Song, 2.;
Nishimura, K.; Gan, Z.; Fu, R.; Quine, J. R.; Cross, T. A., Imaging membrane
protein helical wheels. J. Magn. Reson. 2000, 144, (1), 162-167.

30. Angelis, A. A. D.; Opella, S. J., Bicelle samples for Solid-State NMR of
membrane proteins. Nature protocol 2007, 2, (10), 2332-2338.

55

Chapter 3 Conformational Studies of Membrane-associated Influenza
Fusion Peptide

3.1 BACKGROUND

The secondary structure of IFP associated with membranes or detergents
has been the subject of considerable studies. However, variations in sample
preparation and lipid composition are common, and there are some
discrepancies in the data found in the literature. For membrane-associated lFP,
although a major helical conformation was detected by FTIR1'3, CD3 and ESR“,
there has been detection of an extended form of IFP when lFP:lipid > 0.15 at pH
7.4 or when ~33 mol% of cholesterol is incorporated into the membrane bilayer.6

The solution NMR structure of detergent-associated IFP indicated that
residues 13-18 adopt a helical form at pH 5.0 and have mostly extended
structure at pH 7.4. These structural differences have been thought to be
correlated to the different functional activities at different pHs.7 In the same study,
IFP was also shown to have a kink around residue GIu-11 at both pHs, which
was proposed to be important for the fusion peptide to be active. From the
measurements, this kink is stabilized by the hydrogen bonds between Glu-11 NH
and GIy-8 CO and between Asn-12 NH and Phe-9 CO at both pHs. Additional
hydrogen bonds between side chain NH of Asn-12 and Gly-8 CO and between
Trp-14 NH and Phe-9 CO were also observed in 50% of the conformers of the
pH 5 structure and may further stabilize the kink. The proposed “kinked

boomerang” structure helps the formation of a hydrophobic pocket in the inner

56

face of the kink which can potentially disrupt the membrane and trigger the
membrane fusion.

This chapter presents a detailed investigation of the secondary structure
of membrane-associated IFP using a variety of solid-state NMR methods. The
first part considers the effect of membrane cholesterol on local conformation
using measurement of the 1300 chemical shifts and Ala-7 13CO—GIy-8 13CO
distance. These studies were mainly based on the previous observations that the
related fusion peptide of the HIV virus adopted [3 strand conformation in
cholesterol-containing membranes?“ Measurement of local conformation and
fusion activity of IFP as a function of cholesterol and pH could provide insight into
the relationship between fusion activity and a specific conformation.

The next set of solid-state NMR data measured the more subtle pH-
dependent changes in the structure of helical IFP in membranes without
cholesterol. The goal was to test existing structural models which attempt to
explain the much greater IFP-induced fusion at pH 5.0 than at pH 7.4. For
example, intemuclear13CO-15N distance measurements were carried out to test
whether the pH-dependent structures observed for IFP in detergent micelles
were also present in membranes. f

The final set of solid-state NMR measurements focused on the middle
region of the IFP sequence including residues lle-10 through GIy-13. It is
essential to know the conformation of this region of membrane-associated IFP,
which can be compared to the detergent-associated IFP structure determined by

solution NMR and provides valuable information on the IFP structure/function

57

model. The conformations of IIe-10, Glu-11, Asn-12 and Gly-13 were determined
by 2D 130-130 correlation spectroscopy of membrane-associated IFP with scatter
uniform labeling. Attempts to determine the conformation of those residues have
also been made by using internuclear 13CO—13CO distance measurement of IFP
mutants, lFP-E11VN12A or IFP-N12A. HA protein with point mutation of Asn-12
to Ala was determined to have similar fusion activity with wild-type HA at pH
5.0.14 The HA-E11VE15V mutant (HA mutant with mutations of both Glu-11 and
GIu-15 to Val) was shown to induce red blood cell lipid mixing with very similar
rates to wild-type HA protein at both pH 5.0 and pH 7.4.15 However, a similar HA
mutant, HA-E11A, does not have the ability to induce red blood cell content
mixing even though it can induce the cell lipid mixing.14 In addition, the vesicle
fusion induced by IFP and lFP—E11VE15V were compared. lFP-E11VE15V was
shown to induce the lipid vesicle fusion in a greater rate and extent (~4 fold)
relative to the wild IFP. The fusion activity of IFP-E11VE15V was barely affected
by a pH change to 7.4.15

The backbone 13CO and sidechain Ca and CB chemical shifts are
correlated to the local conformations of proteins. The empirical correlation
databases have been established by liquid-state NMR assignments of proteins.16'
17 These databases are appropriate for solid-state NMR data as evidenced by
similar 13C shifts observed for the same protein in either aqueous solution or the
microcrystalline state.111'19 These chemical shift measurements relied on the IFPs
with specific 13C labeling which allows the observation of specific 13C chemical

shifts.

58

The 13CO—13CO distance (rcc) were determined with the prTDQBU
technique which allows the detection of homonuclear dipolar coupling (doc) under
MAS and MAS was used to obtain sharper and more intense 130 signals.”25 For
IFP with two 13CO labels at adjacent residues in the sequence, measurement of
the 1300—1300 magnetic dipole-dipole coupling discriminated between
separation of the two labeled nuclei by ~2.9 A and separation by ~3.7 A. These
different distances were respectively consistent with helical and B-strand
conformations. The 130-15N distance measurements rely on the application of
the REDOR technique which is an approach to measure heteronuclear dipolar
coupling.”33 For IFP with a 13CO label and an 15N label at different residues
separated by four amino acids in the sequence, the 13CO—15N magnetic dipole-
dipole coupling measurement distinguishes between the helical and more
extended conformations, which have ~4 A internuclear separation of the two
labeled nuclei or > 9 A separation respectively.

The 130—13C correlation spectra were obtained by the PDSD pulse
sequence with samples uniformly 13C and 15N labeled at selectively residues.21
The polarization transfer between backbone 130s within a residue enables the
13C assignment. The 13C chemical shifts were analyzed using the TALOS
program which was designed to derive backbone dihedral angles of proteins

based on the aforementioned empirical databases?"1

3.2 RESULTS
1. 13CO chemical shifts

59

For membrane-associated IFP samples with IFP with specific 1300
labeling, solid-state NMR spectra of the 13CO region provided information about
the local conformations of the labeled residues. For most of the samples, the
lipids were ether-linked and therefore lacked CO functionalities. The unfiltered
spectrum did contain natural abundance peptide 13CO signals which represented
~20% of the integrated signal intensity for singly labeled IFP and ~10% of the
intensity for doubly labeled IFP. Relative to the contribution from the labeled
13003, the natural abundance contribution likely had a broader linewidth because
it was due to many different amino acid types. Most of the spectra are displayed
in Figure 13 and peak 1300 shifts from all of the samples are summarized in
Table 2. The typical full-width at half-maximum linewidth was 2-3.5 ppm which
indicated a narrow conformational distribution.35 The exceptions were IFP-G163
and IFP-G20c which had 5-6 ppm linewidths. For the IFP-L23 and lFP-F3c
samples, spectra were also obtained with sample cooling gas temperatures of —
50 and 0 °C and the spectra were very similar at the two temperatures, cf. table 2.
This result indicated that at least the IFP conformation at Leu-2 and Phe-3 did
not change when the sample was frozen. Most of the spectra were acquired at
the lower temperature to obtain higher signal-to-noise. I

The interpretation of the peak 1300 chemical shifts was based on the well-
known correlation between the 13CO chemical shift and the local conformation of
a residue.17'1° In particular, the database distributions of the 13CO shifts of Leu,

Phe, Ala, He, and Gly residues in helical conformation are 178.53 :I: 1.30, 177.13

:I: 1.38, 179.40 :I: 1.32, 177.72 i 1.29, and 175.51 :I: 1.23 ppm and the

60

corresponding distributions in B strand conformation are 175.67 1 1.47, 174.25 :I:
1.63, 176.09 1 1.51, 174.86 :I: 1.39, and 172.55 :I: 1.58 ppm. For all residues
except Gly-1 in samples whose membranes lacked cholesterol, the 13CO shifts
were more consistent with helical conformation than with B strand conformation.
For these samples, the difference in 13CO shift between the pH 5.0 and pH 7.4
samples was s 0.3 ppm which indicated that helical conformation was pH-
independent for these residues. The non-helical shift of Gly-1 is presumably
related to its N-tenninal location. The liquid-state NMR derived structures of IFP
in DPC detergent showed helical conformation for residues Leu-2 to lle-10 in the
pH 5.0 sample, helical conformation for Gly-13 and Gly-16 in the pH 5.0 sample,
and non-helical conformation for Gly-13 and Gly-16 in the pH 7.4 sample.7 There
was therefore agreement between the solid- and liquid-state NMR results in the
pH 5.0 samples but disagreement for Gly-13 and Gly-16 in the pH 7.4 samples.
Helical 13CO shifts were also observed for Gly-13 and Gly-16 in pH 7.4 PC/PG
membrane samples which disagreed with the non-helical conformation proposed
for these residues in the pH 7.4 liquid-state NMR detergent samples. The pH-
dependence of the liquid-state NMR conformation in this region was an important
part of the structural model proposed to explain the pH-dependence of IFP-
induced vesicle fusion. The pH-independent conformation of membrane-
associated IFP indicated by the solid-state NMR 13CO shifts shows that the
conformation in this region as well as the structure-function model merits further

investigation.

61

The conformation of IFP associated with membranes that contain
cholesterol is of some interest because membranes of respiratory epithelial cells
infected by the influenza virus contain ~30 mol% cholesterol.3‘1'37 For IFP 13CO
labeled at Ala-5, Ala-7 and Gly-8, Phe-9 or Gly-13, the spectra were significantly
different in samples with membranes that contained cholesterol relative to
samples in which cholesterol was absent, cf. Figure 13 and Table 2. The peak
1300 shifts of these residues in cholesterol-containing samples were more
consistent with [3 strand conformation than with helical conformation and the B
strand conformation was observed at both pH 5.0 and pH 7.4. For the samples
containing IFP-A50 and IFP-F90 in cholesterol-containing membranes at pH 7.4,
there was also a downfield shoulder which was consistent with a minor
population of IFP with helical conformation at the labeled positions. The
cholesterol-dependent and pH-independent structure of IFP can be correlated
with the functional results. In particular, pH-dependent and pH-triggered IFP
fusion was observed both for vesicles which contained cholesterol and for
vesicles in which cholesterol was absent.38 A reasonable overall interpretation of
the structural and functional data is that IFP in both the helical and the [3 strand

conformations can induce vesicle fusion.

62

Table 2. Peak 1300 chemical shifts for membrane-associated IF P ‘11

 

 

Residue Membrane or detergent composition pH 5.601%"? 4
Gly-1 DTPC/DTPG 171.2 171.3
Gly-1 DTPC/DTPG/CHOL 170.2 170.4
Lou-2 POPC/POPG 177.4 177.4
Lou-2 DTPC/DTPG 177.9 178.1
Phe-3 POPC/POPG 178.0 n.d. 1’
Phe-3 DTPC/DTPG 178.1 178.2
Ala-5 DTPC/DTPG 179.5 179.5
Ala-5 DTPC/DTPG/CHOL 175.0 174.4 °
lie-e DTPC/DTPG 177.8 177.9
Ala-7 DTPC/DTPG 179.3 179.0
Ala-7 DTPC/DTPG/CHOL 175.3 175.5
Gly-8 DTPC/DTPG 175.4 175.5
Gly-8 DTPC/DTPG/CHOL 170.3 170.6
Phe-9 DTPC/DTPG 178.6 178.8
Phe-9 DTPC/DTPG/CHOL 171.9 172.4 d
Gly-13 DTPC/DTPG 175.3 175.0
Gly-13 DTPC/DTPG/CHOL 173.5 n.d.
Gly-16 DTPC/DTPG 175.2 175.3
Gly-20 DTPC/DTPG 174.7 174.8
Gly-20 DTPC/DTPG/CHOL 175.0 n.d.

 

‘1 All the spectra were obtained with lFPztotal lipid ~ 0.04 and cholesterol was not considered to
be a lipid. Spectra for the POPC/POPG membrane samples were obtained with sample cooling
gas at 0 °C. For all the other samples, the gas temperature was -50 °C. The uncertainties in peak
shifts are :l: 0.2 ppm as determined from measurements on different samples that contained
peptide with the same labeled residue.

bn.d. a not determined.
c There is a shoulder (~20 %) around 178.8 ppm.
‘1 There is a shoulder (~30 %) around 178.0 ppm.

63

A
N
v
A
5'
v
A
O
v
A

d)

C)
A
P
N
'n
w
>
0"
7?
V
3;

t
tit
rrr

err

 

 

 

pH74‘J/\\~‘ ”AM «AN :-/\'~.~ WAN.
PH50MJKAA——A——m-/\gfi 1901170 {Ad‘-
WW. wf/LW 190 170 190 ‘70

 

13C chemical shift (ppm)

llllllllllll

190 170 190 170
13C chemical shift (ppm)

Figure 13. Dependence of 1300 MAS NMR spectra on pH, membrane
cholesterol content, and temperature. The 13‘00 labeled positions are noted
above each set of spectra. All spectra were obtained with -50 °C cooling gas
temperature except the third spectra in (b) and (c). All samples contained ~0.8
umol IFP and membranes composed of either ~16 umol DTPC and ~4 umol
DTPG for the spectra in the first two rows or ~16 umol DTPC, ~4 umol DTPG,
and ~10 umol cholesterol for the spectra (except b and c) in the last two rows.
The pH was labeled for each row of spectra. There is little pH dependence of the
peak shift in the spectra and a strong dependence of spectra on membrane
cholesterol content. The observed chemical shifts of Leu-2, Phe-3, Ala-5, lie-6,
Ala-7, Gly-8, Phe-9, Gly-13, Gly-16 and Gly-20 are consistent with predominant
helical conformation in membranes without cholesterol. The chemical shifts of
Gly-1, Ala-5, Ala-7, Gly-8 and Phe—9 are consistent with predominant B‘strand
conformation in membranes that contained significant cholesterol. For spectra f,
the downfield peak is assigned to Ala-7 13‘CO and the upfleld peak is assigned to
Gly-8 13CO. The third spectra in (b) and (c) were taken with samples contained
~0.8 umol lFP and membranes composed of ~16 umol POPC and ~4 umol
POPG with cooling gas temperature 0°C. There were very similar peak shifts in
spectra taken for samples at different temperatures which indicates that cooling
the sample does not change the peptide structure in membranes without
cholesterol. The third and fourth spectra in (f) were obtained using a 4 mm
diameter rotor and 12.0 kHz MAS frequency. Other spectra were obtained using
a 4 mm diameter rotor and 8.0 kHz MAS frequency. Each spectrum was
processed with 200 Hz Gaussian line broadening and was based on 200 — 10000
scans.

64

2. Ala-7 13CO—GIy-8 1300 distance measurements

For samples prepared with lFP-A7cG83, the Ala-7 13CO—Gly-8 13CO
dipolar coupling was probed with prTDQBU experiments. The goal was to
provide additional evidence for the hypothesis that helical conformation was
predominant in DTPC/DTPG membranes and that B strand conformation was
predominant in DTPC/DTPG/CHOL membranes. For regular or helical structure,
the Ala-7 13CO—Gly-8 13co distance will be ~29 A and will result in a ~300 Hz
dipolar coupling while for regular [3 sheet structure, the distance will be ~3.7 A
with corresponding ~150 Hz dipolar coupling.7' 3"

Figure 14a displays representative prTDQBU spectra of the pH 5.0
DTPC/DTPG sample and Figure 14b displays corresponding spectra of the
DTPC/DTPG/CHOL sample. The labeled 1300s contribute ~90% of the
integrated signal intensity of the So spectra. For each sample, characteristic
residue-type chemical shift tables were used to assign the higher shift peak to
the Ala-7 13CO and the lower shift peak to the Gly-8 1300. In both samples, the
S1 intensity of each peak was lower than the corresponding 80 intensity. This
difference was expected because of the 13CO—13CO dipolar evolution during the
S1 acquisition and the refocusing of this evolution during the So acquisition.
Relative to the DTPC/DTPG/CHOL sample, there was a smaller S1/So intensity
ratio in the DTPC/DTPG sample which is qualitatively consistent with a shorter
13co—‘ico distance in the DTPC/DTPG sample.

Figure 14c shows plots of the “AS/So” = ($0 - 81)/So intensity ratio vs

dephasing time 1. Lines with error bars are (AS/So)” and were derived from

65

(AS/So)” with adjustment for the ~10°/o contribution from natural abundance
1300 signals. The (AS/So)”’/(AS/So)°"p ratio had minor dependence on
dephasing time with a typical range 0.9 < (AS/So)“'/(AS/So)°"p < 1.2. Relative to
the DTPC/DTPG/CHOL sample, there was much faster buildup of (AS/So) for the
DTPC/DTPG sample which qualitatively corresponds to a shorter 131CO — 13CO
distance in the DTPC/DTPG sample. The diamonds show the best-fit (AS/so)”
and corresponded to Ala-7 13CO—Gly-8 13CO distances (rcc) of ~2.8 and ~3.5 A
in the DTPC/DTPG and DTPC/DTPG/CHOL samples, respectively, cf. Table 3.
Those 13CO—13CO distances in A are directly dependent on the backbone

dihedral angle rp in degree of the more C-terrninal labeled residue:

 

rCC(¢) = J10.6—3.Zcosqi (3.1)

The above equation was derived based on standard bond lengths and bond
angles in literature.‘10 Therefore the distances can be used to determine angle (p
which is correlated to the local secondary structure of a protein. The above

calculated distances, 2.8 and 3.5 A, correspond to Gly-8 backbone to dihedral

Table 3. Ala-7 13co — Gly-8 13cc dipolar couplings and distances and Ala-8 dihedral angle go in
membrane-associated IFP at pH 5.0

 

membrane composition d (Hz)a HA) (degree)

DTPC/DTPG 336(1 5) 2.8(0.1) -30(30)
DTPC/DTPG/CHOL 176(10) 3.5(0.1 ) -121 (1 7)

 

‘1 Uncertainties are given in parentheses and were calculated from the values within the )2 =
Xmin2 + 5 region.

66

190170 "1901701 "190717017 11907170 ‘
130 chemical shift (ppm)
(C)

 

 

 

 

1.21 0 §
~it
00.8-
(I)
a 3
q 6
0.4- L I
1 O
t
0.0 O .
0

10
dephasing time (ms)

Figure 14. Ala-7 13CO — Gly-8 1300 distance measurements that probe the IFP
conformational change associated with membrane cholesterol content. The
prTDQBU spectra at 8 ms dephasing time are displayed for lFP2-A7cG8c
associated with (a) DTPC/DTPG (4:1) or (b) DTPC/DTPG/CHOL (8:2:5)
membranes at pH 5.0. Panel (c) displays plots of prTDQBU AS/So vs dephasing
time for the samples containing DTPC/DTPG (darker symbols) or
DTPC/DTPG/CHOL (lighter symbols). Experimental data are represented as
lines with error bars and best-fit simulated data are represented as diamonds.
The displayed experimental data were based on integrations of the 13CO regions
of the So and S1 spectra and have been adjusted for the ~12% contribution of the
natural abundance 13CO signal. For the DTPC/DTPG sample, the best-fit Ala-7
1300 — Gly-8 13CO distance was 2.8 A and was consistent with a backbone
dihedral angle (9 = -30° and helical conformation and for the DTPC/DTPG/CHOL
sample, the best-fit distance of 3.5 A was consistent with (p = -121° and with [3
sheet conformation. The DTPC/DTPG sample contained ~16 umol DTPC and ~4
pmol DTPG and the DTPC/DTPG/CHOL sample contained ~16 pmol DTPC, ~4
umol DTPG, and ~10 umol cholesterol. Each sample also contained ~0.8 umol
lFP2-A7cG8c and the sample cooling gas temperature was -50 °C. Each 80 or
81 spectrum was processed using 100 Hz Gaussian line broadening and was the
sum of: (a) 9000 or (D) 10000 scans. The best-fit 2’1 were: DTPC/DTPG sample,
36; and DTPC/DTPG/CHOL sample, 32.

67

angles of —30° and —121°, which are consistent with on helical conformation near

Ala-7 and Gly—8 in the DTPC/DTPG sample and [3 sheet conformation in the

DTPC/DTPG/CHOL sample.

3. ”CO-”N REDOR Experiments

For IFP associated with DTPC/DTPG membranes, the peak 1300
chemical shifts at selected residues between Leu-2 and Gly-20 were consistent
with helical conformation at both pH 5.0 and pH 7.4, of. Table 2. The pH 5.0
results are generally consistent with the liquid-state NMR structure of IFP in
detergent micelles at pH 5.0 but the pH 7.4 results are in some disagreement
with the detergent structure at pH 7.4. In particular, the structure in detergent
shows predominant extended conformation for residues between Gly-13 and Gly-
20 at pH 7.4. The extent of at helical conformation was therefore studied more
quantitatively using REDOR measurements of 13CO"'15N distances in
DTPC/DTPG samples containing lFP-A5cF9N, IFP-F90G13N or lFP-G13cM17N.
For regular or helical structure, the labeled 13CO'"15N distance in these samples
would be ~4.1 A and would result in a ~45 Hz dipolar coupling while for regular [3
sheet structure, the distance would be ~11 A with corresponding ~3 Hz dipolar
coupling.

Figure 15 displays REDOR data and best-fit simulations for the three
samples at both pHs and Table 4 provides the best-fit 13CO"'15N distances and
dipolar couplings. One important conclusion is that for a given labeled IFP at

each dephasing time, (AS/So)°"p (pH 5.0) = (AS/So)” (pH 7.4). The best-fit

68

11°‘CO"°15N distances in the region covering Ala-5 to Met-17 were therefore
approximately independent of pH and the conformation in this region is also likely
independent of pH. For the IFP-A53F9N samples, the best-fit Ala-5 13co---Phe-9
15N distance was 4.0 :l: 0.1 A at both pHs which was consistent with 0 helical
conformation in this region and with the 3.9 — 4.2 A range of distances of the
detergent structures. For the IFP-F93G13N samples, the best-fit Phe-9 111CO"°GIy-
13 15N distance is 3.6 1 0.1 A at pH 5.0 and 3.7 s 0.1 A at pH 7.4. These
distances are a little shorter than the expected distance in regular or helical
conformation. In the detergent-associated IFP structures, this region forms a turn
at both pHs with a 3.8 - 5.4 A range of distances. For the lFP-G13cM17N

samples, the best-fit Gly-1"CO"°Met-17 15N distance is 4.5 :l: 0.1 A at pH 5.0 and

Table 4. 13CO""15N dipolar couplings and distances in membrane-associated IFP with
comparison to distances in detergent-associated IFP ‘1

 

1300""15N pH 5.0 pH 7.4
membrane detergent membrane detergent
R93‘d"°s d (Hz)" r(A) r(A)c d (Hz)" r(A) r(A)

 

Ala-5""Phe-9 49(3) 4.0(0.1) 4.0-4.2 46(2) 4.0(0.1) 3.9-4.1
Phe-9""Gly-13 66(6) 3.6(0.1) 3.8-5.4 59(3) 3.7(0.1) 3.8-5.2
Gly-13""Met-17 35(2) 4.5(0.1) 4.4-5.7 31(3) 4.6(0.1) 9.0-10.2

 

1' Membrane data were obtained with solid-state NMR REDOR experiments on samples
containing IFP:DTPCIDTPG, IFPztotal lipid ratio of ~0.04, and cooling gas temperature of -50 °C.
Detergent data were obtained from the liquid-state NMR pdb structures, 1ibn (pH 5.0) and 1ibo

7
pH 7.4).

2Uncertainties are given in parentheses and were calculated from the values within the x2 = ern2
+ 5 region.

1’ The displa ed ranges are for twenty structures. For these twenty structures, (d) was calculated
and (1/(d))13 was calculated for comparison to the membrane derived r. For the Phe-9“'Gly-13
data, (1/(d))1/3 was 4.3 and 4.4 A at pH 5.0 and 7.4, respectively, and for the GIy-13"'Met-1 7 data,
(1/(d))1/3 was 5.2 and 9.3 A at pH 5.0 and 7.4, respectively.

69

pH 5.0 pH 7.4
lFP2-A5cF9N

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

a b
1.1.31.0. ......... $1 ..... 1.1.89 ......... $1 .....
200 170 200 170 200 170 200 170
13C chemical shift (ppm)
(e) (d)
1.2
. t 0
80.8 ’ . .
F) : g . .
<‘ 0.4- «
0.0; ’ ., 1
10 T 30 1'0 ' 3'0
dephasing time (ms)
IFPZ-FQcG13N
15.1.3.0. ......... $1 ..... j; ..... $1 .....
200 170 200 170 200 ’1'701200'1'701
130 chemical shift (ppm)
(9) (h)
1.2
(50.8: o .
(I)
<1 0.4.
. ‘I’ . i
00; ..
1'0 ' 3'0 1'0 ' 3'0

dephasing time (ms)

Figure 15. 1"’CO"'15N REDOR measurements.

70

 

IFP2-G13CM17N

 

 

(I) 30 81 (l) 80 S1
200' no '200 '170' ' 200 '1'70 '2'0071'70 '

13C chemical shift (ppm)

('0 (I)
115
é?01¥

< 0.4-

 

 

6 e
I

0.01 ° ‘ '
10 ' 30 10 f 30
dephasing time (ms)

 

 

 

 

 

 

Figure 15. 11‘CO'"1"’N REDOR measurements that probe the pH dependence of
helicity of IFP associated with DTPC/DTPG. Data are presented for (a-d) IFP-
A5cF9N, (e-h) IFP-F90G13N, and (i-I) lFP-G13cM17N and at (a, c, e, g, i, k) pH
5.0 or (b, d, f, h, j, l) pH 7.4. The So and 81 spectra at 16 ms dephasing time are
displayed for (a, b) IF P-A5cF9N, (e, f) lFP-F9cG13N, and (i, j) IFP-G13cM17N, and
the corresponding plots of (AS/So) vs dephasing time are displayed below each
set of spectra with the pH 5.0 plot on the left and the pH 7.4 plot on the right. For
each sample, the peak 1300 shift was independent of dephasing time. In each
plot, experimental data are represented as lines with error bars and the best-fit
simulated data are represented as diamonds. The displayed experimental data
were based on integrations of the 1300 regions of the So and S1 spectra and
have been adjusted for the ~22% contribution of the natural abundance 13CO
signal. Each sample contained ~16 pmol DTPC, ~4 umol DTPG and ~0.8 pmol
IFP and the sample cooling gas temperature was —50 °C. Each So or 81
spectrum was processed using 200 Hz Gaussian line broadening and was the
sum of: (a, b, i, j) ~8000; (e) ~10000; or (f) 16080 scans. The best-fit i were: (c)
15; (d) 8; (g) 24; (h) 20; (k) 42; and (I) 21.

71

 

 

I I I I I I I I I I I I ' ‘ 1 r I IIIIIII

190170 190170 190170 190170
13C chemical shift (ppm)

(C)

 

 

 

 

 

 

 

 

 

 

1 .2
F
0.8 ‘
a?
a
<1 7
i...
0.4 - fall,
fit
It.
’8
It.
i ’94
I:
re
gs
O '1""'""'—" It}! I
O 10 20 30
dephasing time (ms)
I] lFP-A7CGBC in DTPC/DTPG 9 IF P-E11VCN12AC

I lFP-A7CGBC in DTPC/DTPG/CHOL lFP-N12ACG13C

Figure 16. 1300—1300 distance measurements that probe the conformation of
residues 11 through 13 of IFP-E11VN12A and IFP-N12A. The spectra at 24 ms
dephasing time are displayed for membrane-associated (a) IFP- E11VN12A and
(b) IFP-N12A at pH 5.0. Panel (c) displays the magnitude of dephasing (AS/So)
over dephaisng time, 6, 12 and 24 ms are displayed for lFP-E11VcN12Ac, IFP-
N12AcG13c and IFP-A7cGBC associated with DTPC/DTPG (4:1) membrane or
lFP-A7cG8r; associated with and DTPC/DTPG/CHOL (8:2:5) membrane at pH
5.0. The displayed experimental data were based on integrations of the 1"‘1CO
regions of the So and 81 spectra. The measured values from IFP-A7CGBC
samples in membranes with or without cholesterol represent data for [3 strand or
Cl helix structure, respectively. The data from IFP mutants are more consistent
with data from DTPC/DTPG/CHOL-associated IFP-A7cG83 and therefore more
consistent B strand structure.

72

(a) 30 lb) 30

'190 170 ' 190 170 ' 190 170 190 170
130 chemical shift (ppm)

(6)

 

 

 

 

‘L2
- 9
(18‘ °
033 A
53 - 0
<1 8
04-
4 A D O
6 O
(10- 0*
9
I I I
0 10 20 30 40

dephasing time (ms)

Figure 17. Ala-5 13CO Phe-9 15N distance measurements that probe the helicity
of IFP mutants associated with DTPC/DTPG at pH 5.0. The So and 81 spectra at
16 ms dephasing time are displayed for (a) lFP-E11VN12A and (b) IFP-N12A.
The peak chemical shifts of lFP-E11VN12A and IFP-N12A correspond to B
strand and a helix conformation, respectively. For the IFP-N12A sample, there is
also one peak that corresponds to B strand conformation. Panel (c) displays plots
of REDOR AS/So vs dephasing time for the samples containing lFP-E11VN12A
(square), IFP-N12A (triangle) or wild-type IFP (circle). Each of the peptides was
3CO labeled at Ala- 5 and15N labeled at Phe—9. The displayed experimental data
were based on integrations of the whole13CO regions of the So and S1 spectra
and are represented with error bars. The typical uncertainty is :I:O. 02. Each
sample contained ~0.8 pmol peptide, ~16 pmol DTPC and ~4 umol DTPG and
the sample cooling gas temperature was -50 °C. Each 80 or S1 spectrum was
processed using 200 Hz Gaussian line broadening and was the sum of ~8000
scans.

73

4.6 :I: 0.1 A at pH 7.4 which are longer than the expected distance in regular 01
helical conformation. In the detergent-associated IFP structures at pH 5.0, this
region forms a 310 helix with a 4.4 - 5.7 A range of distances. The REDOR
distance is generally consistent with this range. However, the detergent-
associated IFP stnictures at pH 7.4 show extended conformation in this region
and the REDOR distance is inconsistent with the detergent structure 9.0 — 10.2 A
range of distances. This result is of some significance because the pH-
dependent difference in structure in this region in detergent samples is a critical

part of the structure-function model for pH dependence of IFP-induced fusion.

4. IFP mutants

An attempt was made to determine the conformation of residues Glu-11
through Gly-13 using backbone 1300—1300 distance measurements of
adjacently labeled residues. As described in part 2, the distance is useful in
determining the dihedral angle rp which has different characteristic values for a
helix and B strand conformations. In the IFP/detergent model determined by
solution NMR, the GIu-11, Asn-12 residues have dihedral angles (0 that reflects B
strand conformation at both pHs. The IFP forms a turn structure at residues Glu-
11 to Asn-12. The distances between Glu-11 and Asn-12 COs are ~3.6 A at pH
5.0 and ~3.7 A at pH 7.4 and the distances between Asn-12 and Gly-13 00s are
~37 A at pH 5.0 and ~34 A at pH 7.4.“- 42 Both the dihedral angle rp and the
distance data suggest that the conformation of residues Glu-11 and Asn-12 has

the characteristic of an extended structure. But some other previous data have

74

shown that IFP adopts a continuous helix and these residues are helical.4 In a
helical structure, the distances between Glu-11 and Asn-12 COs and between
Asn-12 and Gly-13 COs will be ~2.8 A. Therefore the distance measurements
between the above listed residues are useful in distinguishing the two different
structural models and provide valuable insight in correlating the IFP structure and
its function. Our first approach is to use two mutants of IFP, IFP-E11VN12A and
IFP-N12A to detect the intemulcear distances. The reason to choose these
mutants instead of wild-type IFP is that the selectively labeled Glu or Asn were
not commercially available at the time when those experiments were conducted.
In addition, the mutations of Glu-11 to valine and of Asn-12 to alanine do not
deteriorate the fusion activity of IFP.1‘1' 15

The prTDOBU experiments were carried out on samples containing IFP-
E11VcN12Ac or lFP-N12AcG13r; at pH 5.0 using various dephasing times (7).
The obtained (AS/So)” values at different 7 were compared to the corresponding
(AS/So)” of lFP-A7cG8c samples. The data from lFP-A7cG80 samples in
membranes that lacked or contained cholesterol served as standards for a helical
or B strand structures, respectively. Figure 16 displays representative results of
the lFP-E11VcN12Ac or lFP-N12AcG13c samples compared to lFP-A7cG8c
samples. The spectra of lFP-E11VcN12Ac and lFP-N12AcG13c are shown in
Figures 16a and 16b. The peak shifts in the So spectra are 174.5 and 175.0 ppm
for lFP-E11VcN12Ac and lFP-N12AcG13c samples respectively and the
linewidth of each spectrum is ~5 ppm. The peaks for the two labeled 00s in each

sample are not well resolved. For the Val and Ala residues, the database

75

distribution of the 1300 shifts are 177.65 1 1.38 and 179.40 at: 1.32 for those
residues in the helical conformation and the corresponding distribution in B strand
conformation are 174.8 :I: 1.39 and 176.09 :I: 1.51 ppm. The peak shifts of the two
samples are more consistent with B strand conformations for the labeled regions
of the mutants.

The magnitude of dephasing by 13CO—13CO dipolar coupling, (AS/So)”,
over different dephasing time for the two mutant samples is displayed in Figure
16c and compared to the corresponding dephasing in IFP-A7cG80 samples.
Relative to the data from lFP-A7cG83 in non-cholesterol containing samples, the
(AS/So)” values of both mutant samples are more comparable to the ones from
lFP-A7cG8c in cholesterol containing samples and consistent with B strand
structure. These results further suggest that both IFP mutants have extended
structure at the labeled region. The extended conformation is consistent with the
wild-type IFP structure in the middle region determined by solution NMR.
However the overall structures of the IFP mutants were unknown. The extended
structure of the middle region of IFP mutants may suggest the existence of a tum
similar to the one in the solution NMR structure of the wild-type IFP or be due to
the overall B strand conformation of the mutants. Therefore it is important to
know that whether the mutations cause an overall structural change of IFP. The
next set of experiments was designed to determine the conformations of the N-
terrnini of the IFP mutants and can provide information on the effect of mutations

on the overall structure of IFP.

76

As described in parts 1 and 3, REDOR experiments provide chemical shift
information and 1300—15N distances, both of which are useful in determining the
peptide secondary structure, especially helical forms. Both mutants were 13CO
labeled at Ala-5 and 15N labeled at Phe-9 and were examined with 1300—15N
REDOR. Figure 17a and 17b show the example spectra of lFP-E11VN12A and
IFP-N12A samples, respectively. For the lFP-E11VN12A sample, the chemical
shift of Ala-5 13CO is 171.6 ppm which is more consistent with the shift
representing the B strand structure and the (AS/So)” is ~18 % at r = 24 ms
which is smaller than the corresponding value of a typical helix (~60 %), e.g., the
wild-type IFP with the same labels, lFP-A5cF9N. For the IFP-N12A sample, there
are two peaks with one centered at 177.8 ppm and (AS/So)” ~ 60 % and the
other one centered at 172.5 ppm and (AS/So)” ~ 30 %, corresponding to helical
and more extended structures respectively. Figure 17c compares the dephasing
of those two mutants to that of the wild-type IFP which serves as a model for the
standard helical conformation. The dephasing from the IFP mutants is
significantly smaller than the dephasing from the wild-type IFP. Therefore, the
secondary structures of IFP mutants are strongly affected by the mutations and
not comparable to the structure of the membrane-associated wild-type IFP. The
observed extended conformation of the middle region of IFP mutants, at residue
11 and 12, is probably due to the mutations and can not be correlated to the

conformation of the middle region of the wild-type lFP.

5. PDSD Experiments

77

Since the IFP mutants failed to conserve the N-ten'ninal helical structure
that is characteristic of the wild-type IFP, wild-type lFP was studied with uniform
‘30 and 15N labels at Glu-11 and Asn-12. For samples containing IFP-I10E11u or
lFP-N12G13u, 20 130-13C correlation spectra were acquired and analyzed to
obtain the backbone 13CO, 1300, 130B, 130)], and 1306 chemical shifts of the
labeled residues. The obtained chemical shifts of 13co, 13Co and 1308 were
compared to the characteristic shift values that correspond to the different
secondary structures of an individual residue in the Ref-DB17 and TALOS
databases“. Both databases are based on the observation that chemical shifts
are highly correlated with the local protein secondary structure. By comparing to
the chemical shift values of a specific type of amino acid in Ref-DB, qualitative
knowledge about the protein local secondary structure can be obtained. The
current Ref-DB has 1591 protein chemical shift files. More quantitative
predictions can be made by the TALOS approach which uses chemical shift and
sequence information to predict the protein backbone dihedral angles (p and tp. In
practice, TALOS uses chemical shift data for three consecutive residues to make
predictions for the central residue in the triplet. It searches its database for the 10
best matches for a given triplet in the target protein and when the 10 matches
indicate consistent values for (p and iii of the Ramachandran map TALOS uses
their averages and standard deviations as predictions. The TALOS database
contains 186 proteins and provides more than 24,000 residue triplets. The
predicted dihedral angles were then used to build a model of the protein

secondary structure.

78

The 2D 130—130 correlation spectra were generated with a PDSD pulse

sequence using a 10 ms spin diffusion period during which only intra-residue

cross peaks were observed.‘13 The inter-residue cross peaks are not observed in

these short mixing time spectra. Because only two residues were labeled for

each sample, the unambiguous assignments can be easily made using

characteristic connecting patterns for each individual residue.

Table 5. Measured 13C chemical shifts for lFP-I10E11u and IFP-N12G13u samples compared to
the corresponding chemical shifts from the Ref-DB database"

 

Measured chemical shifts (ppm) ‘1'

 

 

 

co Co ce Cv co
lie-10 1’ 178.0 65.1 38.2 30.0, 17.9 c 15.2
d { 178.7 58.8 28.9 37.2
Glu-1 1 . 181 .9
174.5 54.0 32.0 37.9
Asn-12b 175.5 51.4 39.8 175.0
Gly-13” 174.5 45.8
Chemical shifts from the Ref-DB database (ppm) 9
"e a helix 177.72(129‘) 64.57(1.74) 37.60(1.15)
8 strand 174.86(1.39) 60.05(1.57) 39.86(1.98)
Glu a helix 178.61(1.21) 59.11(1.16) 29.37(0.99)
8 strand 175.35(1.40) 55.52(1.67) 32.01(1.98)
Asn a helix 176.91(1.55) 55.45(1.42) 38.61(1.31)
B strand 174.64(1.65) 52.74(1.47) 40.12(2.07)
Gly a helix 175.51(123) 46.91(1.10)
B strand 172.55(1.58) 45.22(1.17)

 

‘1 Typical uncertainties are :I:O.2 ppm.
b For lie-10, Asn-12 and Gly-13, the chemical shifts are the same for pH 5.0 and pH 7.4 samples.
c v-CH2 and v-CH3 respectively.

Two sets of crosspeaks are present for samples at pH 5.0 and only shift set A is observed for

samples at pH 7.4.

‘1 Standard deviations of distribution are in parentheses.

79

 

  
   

 

 

(a)
Glu Cy/C8 Glu CB/C0l(A) (3) lie CB/Cy(CH2)
/ lle Cy(CH3)/CO “e Cy(CH3)/Cc\. / . ..
/ lle Cy(CH2)/CO and
Glu CB/CO lle Cy(CH2)/Cot ’3
fi "6 CB’CO Ile CB/Ca \ lle CB/Cy(CH3)
3% “and lIe CB/Ct‘l -
’-\lle chco .3 .
‘9 lle CB/Cy(CH2) and
Glu Cy/CB
9
. Glu CO/Cu (B)
/ .
’ I Q 9 . .
Glu CO/Ca (A)
200 160 120 80 40 0
13C chemical shift (ppm)

 

.b
O

a)
O

'_.
m
o

-160

200

Figure 18. 2D 130-130 PDSD spectra of membrane-associated lFP-I10E11U

80

130 chemical shift (ppm)

 

 

 

 

 

I b)
- 0
Asn CB/CO and
Asn CB/Cy
Gly Ca/CO _ 40
E
Asn Cot/CO and 8'.
. Asn Cat/Cy z
- 80 5
To
.9
.5.
o h 120 6
C. 0
5'2
' ° , . - 160
.0 I m
F T i I ‘ l r 1 f I . 200
200 160 120 80 40 0

130 chemical shift (ppm)

Figure 18. 2D 130-130 PDSD spectra of membrane-associated (a) IFP-I10E11u
and (b) IFP-N12F13u at pH 5.0. Each sample contained 16 umol DTPC, 0.4 umol
DTPG and ~0.8 IFP. The data were collected with 10 ms exchange time and total
signal averaging time of ~1.5 days. The MAS frequency was 10 KHz and the
temperature of the sample cooling gas was -50 °C. Spectra were processed with
200 Hz Gaussian line broadening in both dimensions. Peak assignments are
shown using the convention of assignment in f1 (vertical axis)/f2 (horizontal axis).

81

The spectra of membrane-associated lFP-I10E11u and lFP-N12G13u at
fusogenic pH are displayed in Figure 18a and 18b, respectively and the 13C shifts
are listed in table 5 and are compared to the characteristic chemical shifts of
corresponding residues in a helical or B strand conformations. The comparison
showed that the labeled region had two conformations rather than a single
structure. The shifts of lie-10 and Gly-13 1300 are more consistent with local
helical structures, while Asn-12 has shifts that have better agreement with a B-
strand structure. Two distinct sets of crosspeaks were identified for Glu-11, one
of which (A) has shifts that agree with helical conformation while the other shifts
set (B) is closer to those expected in B—strand conformation. The ratio of
intensities of the two peak sets AB is ~3:1. The ZD 130—130 correlation spectrum
was previously obtained for membrane-associated IFP at pH 5.0 that was
uniformly 13C labeled at the first ten N-temlinal residues43 and the 13C chemical
shifts are listed in table 6. The 130 chemical shifts of lie-10 from table 5 and table

6 were compared and the variations between the two measurements were 50.5

ppm.

82

Tablet. ”c
Nieiminal r1

 

 

alisls‘lgilm
litre not a
”The Pile
weleirom

W9 6V
mmge
using it
90800
made ll
Wsspé
and G).
Shifts C
IESldug
Weiep
Union!
meas;

CO

 

Table 6. 13C chemical shifts in ppm for the membrane-associated IFP at pH 5.0 with the first 10
N-terrninal residues uniformly labeled43 a

 

 

co Co CB Cy co C2-C6

Ala 179.4 55.5 18.6

(1) 175.9 47.4

(2) 170.6 43.6
G” (3) 170.0 46.4

(4) 176.1 41.0
lle 177.8 65.6 38.1 29.6, 18.0 14.7
Leu-2 177.5 58.7 42.4 26.7 15.6
Phe 178.0, 178.6 " 60.5 39.7 131.8

 

‘1 Assignments could only be done based on amino acid type. Peaks for each specific residue
were not assigned.

The Phe 1300 chemical shifts, which were not available in the previous 2D measurements”,
were from table 2. The two shifts were for Phe-3 and Phe-9 respectively.

The previous shifts ‘13 (cf. table 6) and the shifts derived from Figure 18
were evaluated using TALOS34 and the dihedral angles of Leu-2 through Gly-13
were generated and listed in table 7. All the dihedral angles were generated
using the measured CO, Ca and CB chemical shifts. The assignments of the
previously obtained chemical shifts for residues Gly-1 through Phe-9 were only
made based on amino acid type. For glycines in the sample, four sets of
crosspeaks were observed (of. table 6). The choices of chemical shifts for Gly-1
and Gly-8 out of these four sets were based on the measured carbonyl chemical
shifts of Gly-1 (Table 2). Both shift sets (2) and (3) were used for Gly-1. For
residue Gly-4 and Gly-8, only the shift set (1) was tested because those shifts
were more consistent with database shifts for helix, which agreed better with the
uniform helical conformation for the N-terrninal region. This is evidenced by the
measured chemical shifts of other residues in the region and the measured Ala-5

13CO Phe-9 15N distance which was consistent with a helical structure. The

83

different values input in the TALOS program had minor effect on the predicted
dihedral angles (Predicted values from one set were within the range of error of
the values from the other set.). The chemical shifts of Ala and Phe in table 6
were used for both Ala-5 and Ala-7 and for both Phe-3 and Phe-9, respectively.
The chemical shifts of lie in table 6 were used for lie-6. For residues lie-10
through Gly-13, the measured chemical shifts in table 5 were used.

The listed values of (0 and qr angles in table 7 are the averages with
standard deviations of the 10 best database matches predicted by the TALOS
program. For residues except Leu-2, Glu-11, Asn-12 and Gly-13, all the 10 best
matches have dihedral angles that fall into a small defined region of the
Ramachandran map which has consistent secondary structures (of. table 7).
Residues Leu-2 through Phe-9 have dihedral angles independent of Glu-11
chemical shifts. Residues lie-10, Glu-11 and Asn-12 have different go and qr
angles corresponding to the A and B Glu-11 chemical shift sets.

The dihedral angles from IFP structures determined by solution NMR are
also listed in table 7. The values for IFP structures at pH 5.0 and pH 7.4 are all
included. These rp and (u angles are based on the average values with standard
deviations obtained from 20 IFP structures which are lowest in energy. The
dihedral angles of residues Leu-2 through lie-10 for both membrane-associated
and detergent-associated lFPs at both pHs are generally consistent with each
other and consistent with a helix structure. Dihedral angles of Gly-13 in
membrane-associated IFP are within the error range of dihedral angles of

detergent-associated IFP, whose values are not typical for a residue with a

84

helical conformation (e.g. Gly-4). Gly-13 was determined to be the start of the
IFP C-terrninal helix in membranes at both pHs and in micelles at pH 5.0. The
determined dihedral angles are therefore reasonable considering the flexibility of
Gly-13 as the start of a helix. For residues Glu-11 and Asn-12, distinct dihedral
angles were obtained corresponding to different shift sets of Glu-11. The dihedral
angles corresponding to Glu-11 shift set A in the membrane samples are
consistent with the values obtained from IFP in micelles at pH 5.0 by solution
NMR. The dihedral angles rp of Glu-11 and Asn-12 in membrane bilayer samples
corresponding to Glu-11 shift set B have agreement with the rp angles obtained
for detergent-associated IFP sample at pH 5.0, but the dihedral angles tp are
very different between samples containing membrane and detergent. The shifts
set A generated (a = -69° and tp = -27° for Glu-11 which agreed with a helical
conformation and rp = -96° and I]! = 8° for Asn-12 which did not agree with
helical or B-strand conformation very well. The shifts set B generated rp = ~-120°
and tp = ~150° for both residues, consistent with a B-strand conformation.
These two sets of dihedral angles were both used to build IFP structures by the
MOLMOL program. In the structures, only the average values of rp and (u angles
from table 7 were used. For residue Gly-1 andresidues Trp-14 through Gly-20,
dihedral angles from detergent-associated IFP at pH 5.0 were used because
these angles are not available from the membrane-associated IFP. For the IFP
C-tenninal region, the solid-state NMR measurements of Gly-13 and Gly-16
chemical shifts and of Gly-13 1300 Met-17 15N distance (described in the

chemical shift and the REDOR experiment parts of this chapter) showed that this

85

region has a helical structure similar to the detergent-associated IFP stnicture at
pH 5.0 determined by solution NMR. At pH 7.4, this region of detergent-
associated IFP has extended conformation. The two different sets of dihedral
angles resulted in two different structures (of. Figure 19). In both structures, the
membrane-associated IFP adopted a helix-tum-helix conformation. The N-
terrninal helix ends at Glu-11 or lie-10 in the structures corresponding to shifts A
and B, respectively. In structure A, all the hydrophobic residues at the N-tenninal
helix are at the bottom face of the stnicture; in some contrast, all the hydrophobic
residues of structure B face to the top side of the structure. Compared with the
middle kink of detergent-associated IFP determined by solution NMR,7 the break
in helicity of membrane-associated IFP is more complicated due to the fact that
Glu-11 and Asn-12 have two conformations. The conformation corresponding to
shift set B is only present for samples at lower pH and absent for samples at
poorly fusogenic pH 7.4, as shown in Figure 20. The population of B
conformation is ~25% and ~20% for samples at pH 5.0 and pH 4.0, respectively;
For the IFP sample at pH 7.4, the population of B conformation is less than 10%
because the peak corresponding to the B conformation is not observed in the

spectrum and the signal-to-noise of the spectrum is ~10.

86

(a) (b)

 

Figure 19. IFP backbone structure based on (a) Glu-11 shift set A or (b) Glu-11
shift set B. All the hydrophobic residues (Leu-2, Phe-3, lle-6, Phe-9 and Ile-10) at
N-terminal helix are shown in gold in (a-b). Residue Glu-11 is in green and
residue Asn-12 is in red. The N-tenninal helix is from residues 2-11 in (a) or 2-10
in (b). The C-terrninal conformation was based on the detergent-associated IFP
structure at pH 5.0. In (c) and (d), structure A (red) and B (green) are respectively
overlaid on top of the solution NMR structure (blue) which is lowest in energy at
pH 5.0.

87

 

 

 

 

 

 

 

 

 

 

- 20
(bl

E
O.
3
E
.C
(D
r 45 13
E
G)
.C
O
o

:2

, 70

 

 

 

 

I
155 175 185 175
130 chemical shift (ppm)

Figure 20. ZD 130-130 PDSD spectra of membrane-associated IFP-UI1OE11U at
(a) pH 4.0; (b) pH 5.0; and (c) pH 7.4. The acquisition and processing
parameters are the same as the spectra in Figure 18. Grey arrows point to the
crosspeaks of Glu Cv/05(COO') (ft/f2). This crosspeak is absent in spectrum (a).
Black anows point to the crosspeaks of Glu Cv/CtS(COOH) (f1/f2) which are
overlapped with other CO peaks. Hollow arrows point to crosspeaks of Glu
Ca/CO (f1/f2) A (left) and B (right). The crosspeak B is absent in spectrum (c).

88

 

 

 

 

 

POPC/POPG f9

 

 

 

DTPC/DTPG ’

(a) PH 74,-50°C (bl PH 7.4,o°c
DTPC/DTPG 1. ’7 3. POPC/POPG ° '° ’
- ° ° _ 0' .40
Glu Cy/CS(COO ) Glu Cy/CS(COO )
'9 c
?\G|UCa/CO(A)/ e. e fi‘GIuCa/CO(A)/ e. e 80
.120
I 0
.160 E
E
I I a fi 0 I I a m . E
200 5
(c) PH50 0°C ‘ (d) PH4.0 -5o°c ‘3 0 E
. 00 . 89¢ g
.C
0
0
2

 

 

Lem Cor/CO (A) '. .. ‘ Q;/<3lu Cojco (A) '. g 40
’\GIU CG/CO (B) / .9. 0° wGIU CG/CO (B) / 0.9, 0-
’ e I e ' 80
- 120
' s ' r
. , -160
I I d: tir- I , 9 et-
. . . . . . . e 200
200 160 120 80 40 0 200 160 120 80 40 0
130 chemical shift (ppm)
Figure 21. PDSD spectra for membrane-associated IFP-I10E11u at (a, b) pH 7.4;

(c) pH 5.0 and (d) pH 4.0.

89

 

 

 

 

 

 

 

 

 

(e) PH7.4,-50°C , a (f) PH7.4,0°C
DTPC/DTPG POPC/POPG
g f 3 -40 E
’ ’15? r ‘ -80 3%
U)
R
.2
.120 ,E,
I o 6
m0
b160‘_
q I .. o . do
0 200
(9) PH 500% a
POPC/POPG
a '40 A
E
Q.
.9.
’ 80 E
(D
.8
~120 ’QE,
0 .C
0
cWC.)
460-
a I m
200

 

200 160 120 80 40 0
13C chemical shift (ppm)

Figure 21. PDSD spectra for membrane-associated IFP-I10E11u at (a, b) pH 7.4;
(c) pH 5.0 and (d) pH 4.0 and for membrane-associated IFP-N12G13u at (e, f) pH
7.4 and (9) pH 5.0. The membrane composition was (a, d, e) DTPC/DTPG (4:1)
and (b, c, f, g) POPC/POPG (4:1). The samples were cooled with nitrogen gas at
(a, d, e) -50 °C and (b, c, f, g) 0 °C. The spectra have no temperature
dependence from — 50 °C to 0 °C. The spectra were processed with 200 Hz
Gaussian line broadening. The total number of scans was (a-c, e-g) ~100000,
and (d) ~200000. Some of the peak assignments were shown using the
convention of assignment in f1 (vertical axis)/f2 (horizontal axis).

90

Table 7. rp and lp angles in degrees of residue Gly-1 to Gly-20 from membrane-associated IFP at
pH 5.0 and detergent-associated IFP 7. Standard deviations are in the parentheses.

 

Mem brane-associated IFP Detergent-associated IFP

 

 

 

 

 

 

91

dNumber of matches within error ranges is: b 9; c 8; d 7 out of 10.

41 42

9’ ll! ¢pH5.0 (p ¢pH7.4 w
Gly-1 N/Aa NIA N/A -160(1) N/A 154(4)
Leu-2 39(2)” 33(8)" 47(1) -14(1) 436(1) 49(1)
Phe-3 -63(8) -38(12) -52(2) -35(1) -64(0.1) 46(1)
Gly-4 -64(7) 42(7) -67(1) -35(3) -53(1) -51(1)
Ala-5 -66(3) -37 (7) -72(2) -36(3) -62(1) 40(1)
lie-6 -68(5) 45(5) -58(1) 41(1) -64(0.1) -63(1)
Ala-7 -62(6) -40(8) -66(1) -35(5) -57(2) -31(3)
Gly-8 -62 (8) -39(6) -54(5) -52 (6) -63 (3) -53 (2)
Phe-9 -67(5) -37(8) -61(5) 44(4) -53(3) -30(3)

Shift set A Shift set B

_ .............. ‘P _. ._. ‘t’ ........ 50.. ‘l’ _
lie-10 -63(5) 42(7) -70 (11) -34 (13) 49 (3) -32 (10) —66(4) -14(4)
Glu-11 -69(11)° -27(13)° -126(29) 156(13) -98(13) 25(4) -121 (7) 5(4)
Asn-12 -96(13) 8 (12) -113(18)" 125(27)” -137(24) 34(39) -146(6) 52(63)
Gly-13 87(11)d 10(9)” 87(9)d 10(11)” 130(53) 5(15) 119(73) 7(6)
Tip-"14 ---------------------------------------------------------- 40(3) 42(4) 48(9) -123(83)
Glu-15 -53(4) 43(4) -103 (52) 12(80)
Gly-16 -70(6) -18(9) -121 (76) -19(23)
Met-17 WA WA -98(11) 41(4) -113(22) -2(10)
lle-18 -71(6) 46(9) -88(18) -1 (46)
Asp-19 42 (46) 151 (77) -85 (56) 110(66)
Gly-20 78 (67) N/A -116 (83) N/A
iii/A a: not available.

Figure 21 shows the effect of pH and temperature on the PDSD spectra.
The spectra 21 c and g were taken for samples made with POPC/POPG lipids
with sample cooling gas at 0 °C and are comparable to the spectra taken for
samples containing DTPC/DTPG lipids with sample cooling gas at -50 °C (cf.
Figure 18). The membrane prepared from POPC/POPG is in the liquid crystalline
phase at 0 °C and the membrane prepared from DTPC/DTPG is in the gel phase
at -50 °C. The similar appearance of the spectra taken at different temperatures
indicated the local conformation of IFP is not affected when the membrane
bilayer undergoes a phase transition. Most of the other spectra were taken at -50
°C to enhance sensitivity.

Compared to the spectra taken for samples at pH 5.0 (of. Figure 18 and
Figure 21 c, g), the spectra for samples at pH 7.4 are very similar (of. Figure 21 a,
b, e, f), except for the lack of shift set B discussed in the previous paragraph (cf.
table 5). This similarity suggested that there is no drastic conformational change
for IFP samples at fusogenic pH and neutral pH. For samples at pH 7.4, IFP also
has the helix-tum-helix secondary structure.

The 2D 1110—130 spectra can also provide information on the protonation
states of the Glu-11 side chain carboxylic group. The 13CO chemical shift of the
COOH group is generally shifted upfield compared to the 000' group. In Figure
20, the crosspeaks marked by the grey arrows are due to the correlation between
the side chain 000' and Cy. The COOH/Cy crosspeaks are overlapped by other
crosspeaks marked by the black anows. It is difficult to determine the exact

COO'ICOOH ratio for samples at each pH because of the overlap. However,

92

information on the relative COO'ICOOH ratios at different pHs can be estimated.
The intensity of the COO'ICy crosspeaks increased as the sample pH was 1
increased and the intensity of the peaks marked by the black arrows decreased
suggesting a decrease in the COOH/Cy crosspeak intensity with the increasing
sample pHs. The increasing COO'ICv crosspeak intensity with increasing pH and
increasing COOH/Cy crosspeak intensity with decreasing pH indicated a higher
population of Glu-11 000' for samples at pH 7.4, a higher population of Glu-11
COOH at pH 4.0 and the intermediate population of both 000' and COOH for
samples at pH 5.0, from which the trend of COO’ICOOH can be estimated as: pH
7.4 > pH 5.0 > pH 4.0. This suggests that Glu-11 is located at the surface of the

membrane and has contact with surrounding water and its pH.

3.3 DISCUSSION

The IFP region of the HA2 domain of the influenza hemagglutinin protein
has been the focus of considerable studies because of its importance in viral
fusion. For membranes which lack cholesterol, previous studies have shown that
the IFP mainly adopts an a helical conformation.1"1'6 A population of B strand IFP
was also observed for high lFP:lipid mol ratio (~0.1) with 1 M NaCl in the buffer
solution at pH 7.4.5 For IFP associated with membranes without cholesterol at
lFP:lipid e 0.04, the present study provides a large number of 1300 chemical
shifts and 13CO"'15N and 1:300—1300 distance measurements that are consistent

with predominant 01 helical conformation between residues Leu-2 and Met-17.

However, in membranes containing cholesterol, there is a large fraction of B

93

strand IFP and the magnitude of this fraction varied from ~0.7 at pH 7.4 to ~1 at
pH 5.0. Cholesterol-containing membranes are likely physiologically relevant
because the membranes of the respiratory epithelial cells infected by influenza
virus contain ~30 mol% cholesterol.” 37 The correlation of the B strand
conformation with membrane cholesterol content has also been observed for the
fusion peptide from HIV (HFP).9 However, even in membranes lacking
cholesterol, there is a significant population of B strand HFP for HFleipid 2 0.02
which contrasts with the predominant helical IFP conformation observed in
samples with lFP:lipid = 0.04.

The correlation of membrane cholesterol content and B strand
conformation is not well understood and may be related to the loss of membrane
porosity associated with cholesterol.“ The helical IFPs are likely monomeric
whereas the B strand IFPs are likely associated as larger B sheet oligomers
through inter-peptide hydrogen bonding. Membrane insertion of the apolar N-
terminal region of IFP likely contains a negative free energy term due to
hydrophobic IFP/lipid interactions and a positive term due to disruption of
membrane packing. The latter term likely becomes larger for membranes
containing cholesterol and needs the compensation from the larger more
hydrophobic B sheet oligomer.

The IFP induces much greater vesicle fusion at pH 5.0 than at pH 7.4 and
additional fusion can be triggered by lowering the pH of the vesicle solution from
pH 7.4 to pH 5.0.38 These functional effects are observed both for vesicles which

contain cholesterol and for vesicles which lack cholesterol. In addition, the B

94

strand conformation is detected for IFP in cholesterol-containing membranes at
both pHs. This latter result combined with observation of triggered fusion in both
types of vesicles is most consistent with fusogenic activity of both helical and B
strand lFPs. Fusogenicity of both conformations might be correlated with
enhanced viral fusion and infectivity because it suggests that infection does not
depend on the cholesterol content of the host cell membrane. Corollary support
for this hypothesis is the previous observation that influenza viral fusion is
independent of membrane cholesterol content.‘15

Solid-state NMR detection of predominant helical IFP conformation in
membranes lacking cholesterol is consistent with previous data from CD, infrared,
and ESR spectroscopies. In addition, the solid-state NMR measurements of 13CO
chemical shifts and 13CO"'15N internuclear distances in samples at pH 5.0 and
pH 7.4 provided information about putative conformational changes in IFP which
may be related to the large difference in fusion activity at the two pHs. In
particular, liquid-state NMR spectra of IFP in detergent micelles suggested that
the Gly-13 to Met-17 region adopted helical conformation at pH 5.0 and extended
structure at pH 7.4.7 In some contrast, the membrane data of the present study
show that both the Gly-13 and Gly-16 13co chemical shifts and the Gly-13
13CO Met-17 15N distance are independent of pH and consistent with helical
structure.

This chapter also presents the detailed study of the conformation of the
IFP middle region in membranes lacking cholesterol. The first approach was to

use IFP mutants, IFP-E11VN12A and IFP-N12A, for the sake of inexpensive

95

isolop
PCi’Pl
Howe
helica
Slfanl
in the
that '
chole
com;

then

was
amir
had
5.0

aIOI

I83).

(he!

 

isotopic labeling. The 13CO—1‘1CO distance measurements of those mutants in
PC/PG membranes suggested that the middle region of IFP is not helical.
However, the mutation of Glu-11 to Val or Asn-12 to Ala altered the overall
helical structure of IFP in membranes without cholesterol to partial or complete B
strand structure. As discussed previously, B strand IFP structure is fusion active
in the cholesterol-containing membrane. The results of the mutants suggested
that the B strand structure is also fusion active in the membranes lacking
cholesterol because IFP has respective higher or comparable fusion activity
compared to wild-type IFP with the mutation of glutamic acids to valines1‘1'15 or
the mutation of Asn-12 to Ala1‘1'15.

The second approach to detect the conformation of the IFP middle region
was to use 2D 130—13C correlation spectroscopy and unifome 13C, 15N-labeled
amino acids. The chemical shifts derived from the spectra indicated that Asn-12
had more extended structural characteristics for membrane samples at both pH
5.0 and pH 7.4. The overall structure of IFP is helix-tum-helix with the turn
around residue Glu-11 and/or Asn-12 at both pHs. For samples at pH 5.0, the
residue Glu-11 exhibited two distinct conformations with the relative ratio of ~3:1
(helix/tum). For the sample at pH 7.4, only one conformation was detected which
corresponds to the major population of the two conformations (helix) observed at
pH 5.0. The termination of the IFP N-terminal helix and the formation of a middle
turn are consistent with some previous results. Solution NMR experiments on
detergent-associated IFP showed that IFP has the helix-tum-helix structure in

micelles.7 Hsu et al. studied an IFP analog which had five glutamic acids in the

96

sequence, at residue 4, 8, 11, 15, 19 respectively, by solution NMR. This IFP E5
' analog exhibits similar fusion activity to the native fusion peptide and also has a
helix-tum-helix structure in micelles at pH 5.0.“113 An IFP mutant G13L was shown
to interact with membrane bilayers by fluorescence studies but have no fusion
activity. The loss of fusion activity was proposed to be correlated to the loss of
the ability to terminate the IFP N-tenninal helix by Gly-13 due to the mutation.‘17
Unlike these studies, either with IFP analogues or using micelles which does not
resemble native cell membranes, our studies utilize the native IFPs in the
membrane bilayer system and have more biological relevance. A fixed angle
boomerang structure of IFP was previously suggested to be necessary for IFP to
be fusion active and the kink was observed to be at residues Glu-11 and Asn-
12.14 Our observation of the IFP helix-tum-helix structure is consistent with this
result and the additional observation of the two conformations of Glu-11 and Asn-
12 in the turn region at pH 5.0 may be correlated to the higher fusogenicity of IFP
at pH 5.0. The observation of the two local conformations at an atomic resolution
level is important because it provides the direct evidence for the existence of
complexity even in a small biological system. It is also important for comparison
with simulations which detect the full conformational distributions. In some
contrast, the conformation B was not observed by solution NMR which may be
due to the lack of IFP B conformation or the rapid motional averaging over the
distribution for IFP in detergent micelles.

The information on the protonation state of the Glu-11 side chain

carboxylic group was also obtained from the 2D spectra. The data suggested that

97

the protonation state of the side chain carboxylic group of Glu-11 is affected by
the pH of the solution and Glu-11 is likely in contact with water. For the IFP
samples at pH 5.0, both 000' and COOH are present. For the samples at pH 4.0
and pH 7.4, the respective dominant states of Glu-11 are COOH and 000‘. The
relative population of COO'ICOOH for samples at different pHs, pH 7.4 > pH 5.0
> pH 4.0, can be compared to the relative population of conformation
Alconforrnation B, pH 7.4 (>90 %) > pH 4.0 (~80 %) ~ pH 5.0 (~75 %), and
suggests that the existence of dual protonation states of Glu-11 for samples at
pH 5.0 or pH 4.0 is possibly not correlated to the two conformations of this
residue. Molecular dynamic studies will benefit from the information on the
protonation states of Glu-11 as the detailed knowledge of the charged states of
glutamic acids is critical for the simulations. Different protonation states of the
acidic residues, e.g. Glu-11, resulted in very different secondary structures and

membrane locations of IFP in the previous molecular dynamic studies.“ ‘19

98

3.4 REFERENCES

1. Luneberg, J.; Martin, l.; Nussler, F.; Ruysschaert, J. M.; Herrmann, A.,
Structure and Topology of the Influenza-Virus Fusion Peptide in Lipid Bilayers. J.
Biol. Chem. 1995, 270, (46), 27606-27614.

2. lshiguro, R.; Kimura, N.; Takahashi, S., Orientation of fusion-active
synthetic peptides in phospholipid bilayers: Determination by Fourier transform
infrared spectroscopy. Biochemistry 1993, 32, (37), 9792-9797.

3. Gray, C.; Tatulian, S. A.; Wharton, S. A.; Tamm, L. K., Effect of the N-
terminal glycine on the secondary structure, orientation, and interaction of the
influenza hemagglutinin fusion peptide with lipid bilayers. Biophys. J. 1996, 70,
(5), 2275-2286.

4. Macosko, J. C.; Kim, C. H.; Shin, Y. K., The membrane topology of the
fusion peptide region of influenza hemagglutinin determined by spin-labeling
EPR. J. Mol. Biol. 1997, 267, (5), 1139—1148.

5. Han, X.; Tamm, L. K., pH-dependent self-association of influenza
hemagglutinin fusion peptides in lipid bilayers. J. Mol. Biol. 2000, 304, (5), 953-
965.

6. Wasniewski, C. M.; Parkanzky, P. D.; Bodner, M. L.; Weliky, D. P., Solid-
state nuclear magnetic resonance studies of HIV and influenza fusion peptide
orientations in membrane bilayers using stacked glass plate samples. Chem.
Phys. Lipids 2004, 132, (1 ), 89—100.

7. Han, X.; Bushweller, J. H.; Cafiso, D. S.; Tamm, L. K., Membrane
structure and fusion-triggering conformational change of the fusion domain from
influenza hemagglutinin. Nat. Struct. Biol. 2001, 8, (8), 715-720.

8. Bodner, M. L., Solid state nuclear magnetic resonance of the HIV-1 and
influenza fusion peptides associated with membranes. Ph. D. thesis, Michigan
State University: East Lansing, 2006; p 122.

9. Qiang, W.; Yang, J.; Weliky, D.P., Solid-state nuclear magnetic resonance
measurements of HIV fusion peptide to lipid distances reveal the intimate contact
of beta strand peptide with membranes and the proximity of the Ala-14-Gly-16
region with lipid headgroups. Biochemistry 2007, 46, (17), 4997-5008.

10. Yang, J.; Gabrys, C. M.; Weliky, D. P., Solid-state nuclear magnetic

resonance evidence for an extended beta strand conformation of the membrane-
bound HIV-1 fusion peptide. Biochemistry 2001, 40, (27), 8126-8137.

99

11. Yang, J.; Prorok, M.; Castellino, F. J.; Weliky, D. P., Oligomeric beta
structure of the membrane-bound HIV-1 fusion peptide formed from soluble
monomers. Biophys. J. 2004, 87, 1951-1963.

12. Yang, J.; Weliky, D. P., Solid state nuclear magnetic resonance evidence
for parallel and antiparallel strand arrangements in the membrane-associated
HIV-1 fusion peptide. Biochemistry 2003, 42, 11879-11890.

13. Zheng, Z.; Yang, R.; Bodner, M. L.; Weliky, D.P., Conformational flexibility
and strand arrantments of the membrane-associated HIV fusion peptide trimer
probed by solid-state NMR spectroscopy. Biochemistry 2006, 45, 12960-12975.

14. Lai, A. L.; Park, H.; White, J. M.; Tamm, L. K., Fusion Peptide of Influenza
Hemagglutinin Requires a Fixed Angle Boomerang Structure for Activity. J. Biol.
Chem. 2006, 281, (9), 5760-5770.

15. Korte, T.; Epand, R. F.; Epand, R. M.; Blumenthal, R., Role of the Glu
Residues of the Influenza Hemagglutinin Fusion Peptide in the pH Dependence
of Fusion Activity. Virology 2001, 289, (2), 353-361.

16. Morcombe, C. R.; Zilm, K. W., Chemical shift referencing in MAS solid
state NMR. J. Magn. Reson. 2003, 162, 479-486.

17. Zhang, H. Y.; Neal, 8.; Wishart, D. S., RefDB: A database of uniformly
referenced protein chemical shifts. J. Biomol. NMR 2003, 25, (3), 173-195.

18. Kiicheldorf, H. R.; Muller, D., Secondary structure of peptides. 3. 130 NMR
cross polarization/magic angle spinning spectroscopic characterization of solid
polypeptides. Macromolecules 1983, 16, (4), 615-623.

19. Saito, H., Conforrnation-dependent 13C chemical shifts - a new means of
conformational characterization as obtained by high-resolution solid-state 13C
NMR. Magn. Reson. Chem. 1986, 24, (10), 835-852.

20. Balbach, J. J.; Petkova, A. T.; Oyler, N. A.; Antzutkin, O. N.; Gordon, D. J.;
Meredith, S. C.; Tycko, R., Supramolecular structure in full-length Alzheimer’s b-
amyloid fibrils: Evidence for a parallel b-sheet organization from solid-state
nuclear magnetic resonance. Biophys. J. 2002, 83, (2), 1205-1216.

21. Bennett, A. E.; Ok, J. H.; Griffin, R. G.; Vega, 8., Chemical-shift correlation
spectroscopy in rotating solids - radio frequency-driven dipolar recoupling and
longitudinal exchange. J. Chem. Phys. 1992, 96, (11), 8624-8627.

22. Gullion, T.; Vega, 8., A simple magic angle spinning NMR experiment for

the dephasing of rotational echoes of dipolar coupled homonuclear spin pairs.
Chem. Phys. Lett. 1992, 194, (4-6), 423-428.

100

23. Bennett, A. E.; Rienstra, C. M.; Griffiths, J. M.; Zhen, W. G.; Lansbury, P.
T.; Griffin, R. G., Homonuclear radio frequency-driven recoupling in rotating
solids. J. Chem. Phys. 1998, 108, (22), 9463-9479.

24. Bennett, A. E.; Weliky, D. P.; Tycko, R., Quantitative conformational
measurements in solid state NMR by constant-time homonuclear dipolar
recoupling. J. Am. Chem. Soc. 1998, 120, (19), 4897-4898.

25. lshii, Y., 13C-13C dipolar recoupling under very fast magic angle spinning in
solid-state nuclear magnetic resonance: Applications to distance measurements,
spectral assignments, and high-throughput secondary-structure determination. J.
Chem. Phys. 2001, 114, (19), 8473-8483.

26. Gullion, T.; Schaefer, J., Rotational-echo double-resonance NMR. J. Magn.
Reson. 1989, 81, (1), 196-200.

27. Anderson, R. C.; Gullion, T.; Joers, J. M.; Shapiro, M.; Villhauer, E. B.;
Weber, H. P., Conformation of [1 -130,15N]acetyl-L-camitine - rotational- echo,
double-resonance nuclear-magnetic-resonance spectroscopy. J. Am. Chem. Soc.
1995, 117, (42), 10546-10550.

28. Kimura, S.; Naito, A.; Saito, H.; Ogawa, K.; Shoji, A., Characterization of
a-helix structures in polypeptides, revealed by 13C=O---H-15N hydrogen bond
lengths determined by 13C REDOR NMR. J. Mol. Struct. 2001, 562, (1-3), 197-
203.

29. Murphy, O. J., 3rd; Kovacs, F. A.; Sicard, E. L.; Thompson, L. K., Site-
directed solid-state NMR measurement of a ligand-induced conformational
change in the serine bacterial chemoreceptor. Biochemistry 2001, 40, (5), 1358-
1366.

30. Smith, S. O.; Eilers, M.; Song, 0.; Crocker, E.; Ying, W. W.; Groesbeek,
M.; Metz, G.; Ziliox, M.; Aimoto, S., Implications of threonine hydrogen bonding in
the glycophorin A transmembrane helix dimer. Biophys. J. 2002, 82, (5), 2476-
2486.

31. Nishimura, K.; Kim, S. G.; Zhang, L.; Cross, T. A., The closed state of a H‘
channel helical bundle combining precise orientational and distance restraints
from solid state NMR. Biochemistry 2002, 41, (44), 13170-13177.

32. Toke, O.; Maloy, W. L.; Kim, S. J.; Blazyk. J.; Schaefer, J., Secondary

structure and lipid contact of a peptide antibiotic in phospholipid Bilayers by
REDOR. Biophys. J. 2004, 87, (1 ), 662-674.

101

33. Long, J. R.; Shaw, W. J.; Stayton, P. S.; Drobny, G. P., Structure and
dynamics of hydrated statherin on hydroxyapatite as determined by solid-state
NMR. Biochemistry 2001 , 40, (51 ), 15451 -1 5455.

34. Comilescu, G.; Delaglio, F.; Bax, A., Protein backbone angle restraints
from searching a database for chemical shift and sequence homology. J. Biomol.
NMR 1999, 13, (3), 289-302.

35. Yang, J.; Parkanzky, P. D.; Khunte, B. A.; Canlas, C. G.; Yang, R.; Gabrys,
C. M.; Weliky, D. P., Solid state NMR measurements of conformation and
conformational distributions in the membrane-bound HIV-1 fusion peptide. J. Mol.
Graph. Model. 2001, 19, (1), 129-135.

36. Worrnan, H. J.; Brasitus, T. A.; Dudeja, P. K.; Fozzard, H. A.; Field, M.,
Relationship Between Lipid Fluidity and Water Permeability of Bovine Tracheal
Epithelial-Cell Apical Membranes. Biochemistry 1986, 25, (7), 1549-1555.

37. Scheiffele, P.; Rietveld, A.; Wilk, T.; Simons, K., Influenza Viruses Select
Ordered Lipid Domains during Budding from the Plasma Membrane. J. Biol.
Chem. 1999, 274, (4), 2038-2044.

38. Parkanzky, P. D., Solid state nuclear magnetic resonance studies of the
influenza fusion peptide associated with membrane bilayers. Ph. D. thesis,
Michigan State University: East Lansing, 2006; p 80.

39. Longhi, S.; Czjzek, M.; Lamzin, V.; Nicolas, A.; Cambillau, C., Atomic
resolution (1.0 A) crystal structure of Fusarium solani cutinase: stereochemical
analysis. J. Mol. Biol. 1997, 268, (4), 779-799.

40. Voet, D.; Voet, J. G., Biochemistry. John Wiley & Sons, Inc.: New York,
1995.

41. Han, X.; Bushweller, J. H.; Cafiso, D. S.; Tamm, L. K., NMR structure of
hemagglutinin fusion peptidein DPC micelles at pH 5. P03 #: 1ibn 2001.

42. Han, X.; Bushweller, J. H.; Cafiso, D. 8.; Tamm, L. K., NMR structure of
hemagglutinin fusion peptidein DPC micelles at pH 7.4.. PDB #: 1ibo 2001.

43. Bodner, M. L.; Gabrys, C. M.; Struppe, J. O.; Weliky, D. P., 130-13C and
15N—13C correlation spectroscopy of membrane-associated and uniformly
labeled human immunodeficiency virus and influenza fusion peptides: Amino
acid-type assignments and evidence for multiple conformations. J. Chem. Phys.

2008, 128, 052319.

102

44. Nicol, F.; Nir, S.; Szoka, F. 0., Effect of cholesterol and charge on pore
formation in bilayer vesicles by a pH-sensitive peptide. Biophys. J. 1996, 71, (6),
3288-3301.

45. White, J.; Kartenbeck, J.; Helenius, A., Membrane fusion activity of
influenza virus. EMBO J. 1982, 1, (2), 217-222.

46. Hsu, C. H.; Wu, S. H.; Chang, D. K.; Chen, C. P., Structural
characterizations of fusion peptide analogs of influenza virus hemagglutinin -
Implication of the necessity of a helix-hinge-helix motif in fusion activity. J. Biol.
Chem. 2002, 277, (25), 22725-22733.

47. Matsumoto, T., Membrane destabilizing activity of influenza virus
hemagglutinin-based synthetic peptide: implications of critical glycine residue in
fusion peptide. Biophys. Chem. 1999, 79, (2), 153-162.

48. Sammalkorpi, M.; Lazaridis, T., Configuration of influenza hemagglutinin
fusion peptide monomers and oligomers in membranes. BBA-Biomembranes
2007, 1768, (1), 30-38.

49. Feig, M., personal discussion with Dr. Michael Feig.

103

Chapter 4 Studies of Influenza Fusion Peptide Membrane Location

4.1 BACKGROUND

It has been proposed that the structural bases for the fusion activity of
fusion peptides include the interaction of peptides with membranes and the
membrane location of fusion peptides is an important stnictural component to
understand this interaction. Some previous ESR studies have suggested that the
different membrane locations of IFP are correlated to its different fusogenic
activities at different pHs.1 The membrane location of IFP has been probed by
fluorescence and ESR spectroscopies. While some studies detected a pH-
dependence of the membrane location of IFP and its analogues, other studies
supported an insertion depth which is independent of pH.1'5

A fluorescence study indicated that the IFP is near the hydrocarbon-
phosphate interface of the membrane and no gross positional change of the
peptide between different pH values was detected.2 Three different groups
studied the IFP membrane location with ESR and different results were obtained:
(1) Ltineberg et al. used the 20-amino acid length synthetic peptide
corresponding to HA2 and suggested that the N-terrninal helix of the fusion
peptide is located near the hydrophobic acylchain-phosphate headgroup
interface and the location is independent of pH;3 (2) Macosko et al. expressed a
major portion of HA2 domain (FHA2, residue 1-127) which was then spin-labeled
and studied by ESR and found that the fusion peptide region is inserted into the
membrane with a tilt angle of ~75° from the membrane normal with lie-6 most

deeply inserted. The insertion is also pH-independent.4 (3) Han et al. studied the

104

synthetic IFP which was linked to a short hydrophilic peptide (GGCGKKKK) and
the results suggested that the fusion peptide is inserted into the membrane in a
pH-dependent way and the residue with the deepest insertion is Phe-3 or lie-6 at
pH 5.0 or pH 7.4, respectively.1 This pH-dependent insertion was also observed
for some IFP analogues by another group using tryptophan fluorescence
emission experiments. In this study, Trp-14 in IFP analogues was used and the
observation of a blue shift in the steady-state fluorescence emission spectra for
samples at pH 5.0 relative to samples at pH 7.4 indicated a deeper membrane
insertion of IFP analogues at pH 5.0.5 In addition to these different experimental
results, molecular dynamics studies predicted distinct insertion as well. Two
simulations suggested that IFP is at the membrane-water interface, exposing the
polar sidechains to water and the non-polar sidechains to the hydrophobic core.6'
7 Another two theoretical studies predicted an IFP location at the amphipathic
interface between the lipid headgroups and hydrocarbon chains.8' 9 One
simulation proposed that micelle-associated IFP lies at the micelle surface, while
for the membrane-associated IFP, the N-terrninal region deeply inserts into the
membrane bilayer.1o

I also studied the membrane location of IFP using solid-state NMR. The
solid-state NMR measurements can detect the membrane location at an atomic-
resolution level without introducing any extra group which may potentially disturb
the IFP membrane location. The method is based on the 13O—31P or 130-19F
dipolar coupling measurements corresponding to the distances between IFP

backbone 13CO and lipid phosphate groups or between IFP backbone 13CO and

105

19F in the acylchains in the membrane interior, respectively. The IFP was labeled
with 13CO at various positions and incorporated into the membranes that lack
cholesterol at pH 5.0 or pH 7.4. As discussed in Chapter 3, IFP adopts helical
structure in those membranes based on chemical shift measurements. The 13C-
31P and 130-19F REDOR experiments showed that the N and C-terrninal regions
of IFP are in close contact with phosphate head groups and residues Phe-3, lie-6
and Phe-9 have contacts with F(C16) (F labeled at the DPPC C16 position). An
inverted boomerang structure of membrane-associated IFP was proposed based
on data of the IFP conformational studies in chapter 3 and these 13C-31P and 13C-
19F REDOR data. Compared to the IFP samples at pH 5.0, the samples at pH 7.4
have less IFP population inserted into the membrane bilayer and some IFP

molecules were proposed to lie in the water layer above the membrane bilayer.

4.2 RESULTS
1. Static 31F spectra

Prior to the development of structural models for IFP/membrane
interaction it is essential to know the lipid phase for membranes with bound IFP.
In this section, 31P spectroscopy was applied to study the structure of
membranes for membrane-associated IFP samples at different pHs. Figure 22a-
b shows the representative 31P spectra of IFP membrane samples containing
DTPC/DTPG (4:1) at pH 5.0 or pH 7.4. These spectra have similar lineshapes to

the spectrum of a DTPC/DTPG membrane sample without bound IFP (of. Figure

106

(a) DTPC/DTPG w/ IFP (b) DTPC/DTPG w/ IFP
5.0

M“ pH 7.4
100 5'0 0 -50 300

(c) DTPC/DTPG w/o IFP 31 p chemical shift (ppm)
pH 5.0

 

 

100 5'0 0 -50 -100
31 P chemical shift (ppm)
Figure 22. 31P spectra of membrane samples (DTPC/DTPG, 4:1) that (a, b)
contained IFP or (c) had no bound IFP at 35 °C. The peptide to lipid mol ratio is
0.04. Each spectrum was processed with 200 Hz Gaussian line broadening and
was the sum of 300 to 1000 scans.

(a)

  

19 “ "“’ '
F(016) 19F(C$)

Figure 23. (a) Model of a membrane bilayer with a peptide inserted into a single
leaflet and with the positions of 31P, 111F(C5), 19F(Ci6) and peptide backbone
13CO labeled. The blue balls represent the phosphate headgroups and the gray
lines represent the hydrocarbon chains of lipids. The approximate dimension in A
of the membrane bilayer is shown by the scale bar on the right. The thickness of
the membrane bilayer is ~50 A and the phosphate headgroup is ~8 A in diameter.
The 13CO—19F(CS) and 13CO-19F(C16) distances are shown by the black lines. All
the 19Fs(C16) are at the bilayer center and have similar distances to the labeled
13CO. The peptide labeled 1 CO has a shorter distance to the 19F(CS) located at
the same leaflet relative to the 19F(CS) located at the different leaflet. The 13CO—
19F(C5) REDOR data contain information from the 13CO—19F(C5) pair and is
dominated by the pair with the shorter distance. The effective concentration of
19F(CS) is half of its real concentration for the REDOR measurement. Panels (b)
and (c) show the structures of 5-19F-DPPC and 16-19F-DPPC molecules
respectively.

107

22c) and are consistent with a predominant lamellar phase.11 Therefore, the lipid
phase of membranes are not affected by addition of IFP and are independent of
pH. The analyses of the following REDOR data were based on the lamellar

phase of membranes.

2. Membrane location of pH 5.0 samples

The goal of the IFP insertion studies is to provide a structural model of the
location of IFP in membranes at a high resolution level. The studies were carried
out with IFP samples at both pH 5.0 and pH 7.4 and the IFP locations in
membranes at different pHs can be correlated to their functional activities. In this
section, the data of the IFP samples at fusion active pH 5.0 will be displayed and
analyzed. Residues throughout the IFP sequence were selectively 13CO labeled
in order to study the insertion of the different regions of IFP. In the studies, two
fluorinated lipids, 16-19F-DPPC and 5-19F-DPPC, were used in order to detect the
distances of the IFP backbone to the center of the membrane bilayer and to the
midpoint between the phosphate headgroups and the bilayer center, respectively
(of. Figure 23). The REDOR experiments, denoted as 13CO—31P, 1300-11’F(C16)
and 13‘CO—19F(C5) detected the distances between selectively labeled IFP
backbone 13CO and 31P, 16-19F and 5-19F respectively.

Figure 24 shows the REDOR spectra at long dephasing time (32 ms for
the 130—31P REDOR experiments and 24 ms for the 13CO—19F REDOR
experiments) for IFP samples at pH 5.0. The peak 13C0 chemical shifts from So

spectra are listed in table 8. All the shifts are consistent with the shifts listed in

108

table 2 of chapter 3 and the shifts from residues Leu-2 through Gly-20 are
consistent with a helix structure. For membrane samples containing fluorinated
DPPC (CS or C16), most of the spectra have two peaks with the downfield one
correlated with the a helical conformation and the upfield one correlated with the
B strand conformation. The relative population of the two conformations can be
estimated based on the relative intensities of the two peaks. The helical IFP is

the predominant population for all the samples.

109

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

50 $1 $0 $1

(a) G'Y-1 (b) Len-2
A W “(Ax/w MM
(0) Phe-3 (d) Ala-5

(e) lie-6 (f) Phe-9

(9) Gly-13 (h) Gly-16

I I I I 190 170 190 170

13C Chemical Shift (ppm)

(I) Gly-20
w‘/\"% W

190 170 190 170

13C Chemical Shift (ppm)

Figure 24 130-31P REDOR spectra with 32 ms dephasing time.

110

13C-19F(C5) 24 ms

 

 

 

 

 

 

 

 

 

 

 

s0 s1 s0 S1
(j) Phe-3 Ala-5
(I) Ila-6 Phe-9
I I I I 190 170 190 170

13C Chemical Shift (ppm)
Gly-13

170 ' 170

013C Chemical Shift 1(ppm)

 

 

Figure 24 13C-19F(05) REDOR spectra with 24 ms dephasing time.

111

13C-19F(C16) 24 ms

 

 

 

 

 

 

 

 

 

 

 

 

80 81 80 S1
Gly-1(9) Leu-2
Phe-3 (r) lie-6
Phe-9 Gly-13

 

 

«AMA. JLJL

O13C70 Chemical Shift 1(ppm)17

 

 

 

 

Gly-16

9013C Chemical Shift 1(ppm)17

 

 

Figure 24. REDOR 13C So and 81 NMR spectra at long dephasing time for
membrane-associated IFP samples at pH 5. 0. The experiment type and the
dephasing time are labeled on the top of each group of spectra and the1aCO
labeled residues are also labeled above each set of So and 81 spectra. Each
sample contained 16 pmol DTPC, 4 pmol DTPG and 0.8 pmol IFP. The samples
used to take spectra (d), (e) and (j-n) contained 9 mol% 5-F-DPPC lipid and the
samples used to take spectra (o-u) contained 9 mol% 16-F-DPPC lipid. Each
spectrum was processed with 200 Hz Gaussian line broadening and was the
sum of 20000 - 30000 scans.

112

13C-31P 32ms 13C-1°F(C5) 24ms 13C-1°F(C16) 24ms

 

 

 

 

 

 

 

 

 

 

0.3
7 (a) (b) (cl
2
/
0.8- 7
Q %/' é 0'2- -
O é7/ /
A /¢¢ ; \\ “Q
0 ¢// / \ s .\
(I) /%% 2 V \ 9
a 04. Z22 2 g g s
- /¢% 2 01- - V \ V
d /// / - \ \ \
/ \ \
V /22 // \ \ e
/// 7 \ \ \
2f? / / / / 2 § § § \
222 Z 4 3 f 2 sis s s %
0_01// 22 2 Z 2 2, 001 Eq.. 3 a 83 s .9: x
5

 

0 5 10 1 5 20 0 5 10 15 20
residue

Figure 25. Summary of experimental REDOR dephasing (AS/So)°"" for the
spectra displayed in Figure 24. The (AS/So)” values are shown as bars for
different residues and a typical uncertainty is :I:0.01-0.02.

Table 8. Peak 13co chemical shifts in ppm for IFP samples at pH 5.0 a

 

Gly-1 Lou-2 Phe-3 Ala-5 lie-6 Phe-9 Gly-13 Gly-16 Gly-20

 

171.2 178.0 178.6 179.4 177.9 178.6 175.5 175.2 174.7

 

‘1 Typical uncertainties in peak shifts are :I: 0.5 ppm as determined from the measurements on
samples that contained peptide with the same labeled residue but different membranes.

Figure 25 summarizes the dephasing, (AS/So)”, corresponding to Figure
24. Compared to the So 1300—31P spectra of each labeled IFP, the intensities of
the S1 spectra are greatly reduced (> 60%) for the residues at N- and C-terrninal
regions, i.e., Gly-1, Leu-2, Phe-3 and Gly-20. For most residues at the middle
region of IFP, Ala-5, Phe-9, Gly-13 and Gly-16, the intensities of the 81 spectra
are only reduced by ~20 %. In contrast, the intensities of the 81 11"CO—111F(C16)
spectra are the same as the So spectra for residues Gly-1, Leu-2 and Gly-20 and
are reduced by ~20 % relative to the So spectra for residues Phe-3, lie-6 and
Phe-9. This suggests that the N- and C- terminal regions of IFP are in close
contact to the membrane phosphate headgroups and the middle region of IFP

has some contact to the membrane bilayer center suggesting an insertion of the

113

IFP middle region into the membrane bilayer. Similar conclusions can be drawn
from the experimental dephasing curves shown in Figure 26. All the values were
obtained based on integrations of 1 ppm of the downfield peaks corresponding to
the helical IFPs. More 13CO—31P dephasing was observed for residues at the N-
and C- terrninal regions than for the residues in the middle region and the 1300-
19F(C16) dephasing was only observed for residues in the middle of the
sequence. For all the labeled residues,. little 13CO—‘gF(C5) (< 10 % at 24 ms
dephasing time) was detected and the greatest observed 13CO—“’F(C16)
dephasing is small (< 20 % at 24 ms dephasing time), which is probably due to
the small amount of labeled 19F-DPPQ used (~9 mol % of total lipids). The 19F
atoms at the C16 positions of both layers in a membrane bilayer are located at
the center of the membrane and have similar dipolar interaction to the same
13CO of an lFP. The 19F atoms at the C5 positions of different layers in a
membrane bilayer have different distances and dipolar interactions to the same
1300 in an IFP and the bigger dipolar interaction makes the predominant
contribution to the 13CO—‘QF REDOR experiment (cf. Figure 23). Therefore, only
half of the 5-19F-DPPC is effective for the 13CO—"’F(C5) REDOR measurements
and the actual effective concentration of 5-19F-DPPC is only ~4.5 mol %
compared to the 16-19F-DPPC concentration (~9 mol %). It is reasonable that the
observed 13CO—19F(C5) dephasing is smaller than the observed 13CO—“’F(C16)

dephasing at a specific dephasing time.

114

1.2

 

 

 

 

 

 

 

 

 

 

 

1*) Gly-1 (b) Leu-2 (c) Phe-3
0.8-
0.4-
0.0- /——*
1.2 -
(d) Ala-5 (e) Ila-6 (f) Phe-9
ioa- '
2°
3 0.4 -

 

 

 

 

 

 

 

 

1.2

 

(9) Gly-13 (h) Gly-16 (i) Gly-20
0.81 ' .._——x

 

 

 

 

 

 

 

 

I I I I

0 1o 20 30 0 1o 20 30 0 1o 20 30
dephasing time

Figure 26. ‘3co—3‘P REDOR (dark solid line), ”CO—”F(05) REDOR (gray solid
line) and 13CO—19F(C16) REDOR (dotted line) experimental dephasing curves for
IFP samples at pH 5.0. The uncertainties are represented by the error bars and
are typically :l:0.01-0.02. The 1300 labeled residues are labeled on top of each
spectrum. The samples used were the same as the ones used to take the
corresponding spectra in Figure 24.

115

13C-31P REDOR

 

1.2 a

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

(a) Gly-1, , (p) Leu-2 (c) Phe-3
0.8" ° 0 o
0.4“ 0 .
o.oe ° .
1'2 (a) Ala-5 (e) Phe-9 (f) Gly-13
0.84
0.4+
é 0.0-4 M fi—u/ M
<5
g; 1.2
v (g) Gly-16 (h) Gly-20
0.8“ 0 0
0.4-
0.0- r—r/
13C-19F(C16) REDOR
1'2 a) Phe-3 (i) lIe-6 (k) Phe-3
0.8‘
0.4‘
0.0- b/ ‘/ L—I/

 

 

 

 

 

 

 

O 10 20 30 0 10 20 30 0 1O 20 30
dephasing time

Figure 27. Plots of (AS/So)” gcircles) and (AS/So)“ (solid lines) vs. dephasing

time for 13co-3‘lD and 13co— 9F(C16) REDOR of IFP samples at pH 5.0. The
13CO labeled residues are labeled at the top of each spectrum.

116

A more quantitative analysis of the results can be obtained by fitting
(AS/So)” to a theoretical dephasing (AS/So)“m according to Eq. 2.6 and 2.10,
where (AS/So)” was calculated from (AS/So)” by removing the contributions
from natural abundance nuclei as described in chapter 2 and in the appendix. A
best-fit 13CO—31P or 13CO—‘9F dipolar coupling frequency can be obtained and
correlated to the internuclear distances, rep or rap, through Eq. 2.2 and 2.3.
Figure 27 displays the fitting for all the build up curves with (AS/So)”’> 0.1 at
long dephasing time (T = 32 ms for 13co—“P REDOR and r = 24 ms for ”CO—19F
REDOR). The distances from the fitting are listed in table 9. In the REDOR data
fitting, a single membrane location of IFP was assumed and a 13C-X two-spin
system was used, where X represents 31P or 19F(C16). From the data fitting, the
Gly-1 and Gly-20 13COs were determined to be respectively 4.5 and 6.0 A away
from the phosphorus and the Leu-2 and Phe-3 1:‘COs are ~7 A away from the
phosphorus. These data and reasonable values of van der Waals radii (~2 A for
1300 and ~4 A for the phosphorus headgroups) are consistent with the close
contact of residues Gly-1, Gly-20, Leu-2 and Phe-3 with the phosphorus
headgroups. Residues Phe-3, lle-6 and Phe-9 are calculated to be ~12 A away
from 19F(C16). Molecular dynamics simulations have determined that the
distance between 31F and the 19F(C16) of a DPPC molecule in the gel phase is
~24 A}? The ~12 A fitted ”CO—”F(C16) distance indicated that these middle
residues are inserted into the membrane bilayer. These distances combined with
the fitted 13CO—3‘P distances of most of the middle residues from the

experimental data (~10 A) suggest that the middle region of IFP is half-way

117

inserted into a single leaflet of the membrane. The lle-6 13CO—31P dipolar
coupling was not detected which is reasonable because of the relative location of
lle-6 to the phosphorus headgroups. The van der Waals radius of a helix
backbone is ~4 A and a residue located at the bottom side of the helix (e.g. lle—6)
will be ~8 A further away from the. phosphorus above the helix compared to the
residue located at the top side (e.g. Lou-2) (cf. Figure 32) and is very likely to be

too far away to be detected (1300—31P REDOR detectable limit ~11 A).

Table 9. Best-fit 13CO-31P and 13CO—1QF(C16) distances for IFP samples at pH 5.0 a

 

Gly-1 Lou-2 Phe-3 Ala-5 Ila-6 Phe-9 Gly-1 3 Gly-16 Gly-20

 

31
,P 4.5(0.1) 6.7(0.2) 7.2(0.2) 10.1(1.4) >11b 10.1(0.5) 10.1(1.4) 10.1(1.0) 6.0(0.2)
distance

(A)
x2 1570 31 23 14 N/Ad 48 6 16 22

 

 

19
F(C16) >14” >14 12.1(o.9) NlA 11.5(04) 11.5(02) >14 >14 N/A
distance

(A)
x2 N/A N/A 4 NIA 6 25 N/A N/A NlA

 

‘3 Uncertainties are given in parentheses and were calculated from the values within the xY = x

mmz +5 region.

b Calculated from the REDOR universal dephasing curve with ASlSo = 0.1 at long dephasing time

13 = 32 ms for 13CO—3‘1P or r = 24 ms for 13CO—19F) based on the consideration that only the
C05 with (AS/80f” > 0.1 at long dephasing time are fittable.

c Regardless of the big x2, the fitting is still reasonable and consistent with the result of Gly-1

being close to 31P. The x2 may be reduced by using multi-spin (e.g., X-13CO-X) for the fitting.

d MIA 5 not available.

3. Membrane location of pH 7.4 samples
In this section, the data of the IFP samples at non-fusogenic pH 7.4 will be
displayed and analyzed. The same residues as in the studies of IFP pH 5.0

samples were selectively 13CO labeled and the same experiments were carried

118

out in order to have a direct comparison between the results for samples at
different pHs. A

Figure 28 shows the REDOR spectra at long dephasing time (1 = 32 ms
for the 13c-3‘P REDOR and r = 24 ms for the ”CO—19F REDOR experiments) for
lFP samples at pH 7.4. The peak 13CO chemical shifts from So spectra are listed
in table 10 and are consistent with the shifts obtained for IFP pH 5.0 samples as
listed in table 8. Unlike the spectra for pH 5.0 samples, most of the spectra only
contain single peaks that correspond to the a helical conformation and lack the
upfield peaks which correlate with the [3 strand conformation observed for
membrane samples containing fluorinated DPPC (CS or C16) at pH 5.0.

Table 10. Peak 1300 chemical shifts in ppm for IFP samples at pH 7.4 a

 

Gly-1 Lou-2 Phe-3 Ala-5 Ila-6 Phe-9 Gly-1 3 Gly-16 Gly-20

 

171.3 178.0 178.6 179.5 177.9 178.4 175.1 175.4 174.6

 

1Typical uncertainties in peak shifts are 1 0.5 ppm as determined from the measurements on
samples that contained peptide with the same labeled residue but different membranes.

The experimental dephasing, (AS/So)”, corresponding to Figure 28 are
summarized in Figure 29 and the experimental dephasing curves are displayed
in Figure 30. Compared to the results of IFP pH 5.0 samples, a similar trend was
observed that the residues at N- and C- terminal regions (Gly-1 and Gly-20) have
larger 13CO—‘MP dephasing relative to the residues in the middle region and
residue lle-6 has 13CO—"’F(C16) dephasing. These observations suggest a
generally similar insertion of IFP in membranes at pH 7.4, i.e., the N- and C-
terrninal residues are in close contact to the phosphate headgroups and the
middle region of IFP is inserted into the membrane bilayer. However, there is

some difference between the results of IFP samples at different pHs. For

119

l 7

l(

I)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

S0 S1 S0 $1
(a) G'Y'1 (b) Leu-2
(c) Phe-3 (d) Ala-5
(e) IIe-6 (f) Phe-9
(9) Gly-13 (h) Gly-16
I I ' ' 180 170 180 1%
13C Chemical Shift (ppm)
(i) Gly-20
150 150 180 170

13C Chemical Shift (ppm)

Figure 28 ‘3c-3‘P REDOR spectra with 32 ms dephasing time.

120

13c-191=(05) 24 ms

 

 

 

 

 

 

 

 

 

 

 

 

 

 

So ' 81 80 S1
0') Phe-3 Ala-5
(I) lle-6 Phe-9
I I I I 190 170 1 '90 170
130 Chemical Shift (ppm)
('1) Gly-13 -
150 170 1 50 1 70

13C Chemical Shift (ppm)

Figure 28 13’C-"’F(C5) REDOR spectra with 24 ms dephasing time.

121

13C-19F(C16) 24 ms

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Gly-1 (p) Leu-2
Phe-3 (r) Ile-6
Phe-9 (t) Gly-13
150 170 180 150

13C Chemical Shift (ppm)
Gly-16

190 170 190 170

13C Chemical Shift (ppm)

 

 

 

Figure 28. REDOR 13C So and $1 NMR spectra at long dephasing time for
membrane-associated IFP samples at pH 7. 4. The experiment type and the
dephasing time are labeled on the top of each group of spectra and the 300
labeled residues are also labeled above each set of So and 81 spectra. Each
sample contained 16 pmol DTPC, 4 pmol DTPG and 0.8 umol IFP. The samples
used to take spectra (d), (e) and (j-n) contained 9 mol% 5-F-DPPC lipid and the
samples used to take spectra (o-u) contained 9 mol% 16-F-DPPC lipid. Each
spectrum was processed with 200 Hz Gaussian line broadening and was the
sum of 20000 — 30000 scans.

122

samples at pH 7.4, the residues Leu-2 and Phe-3 have less 13CO-3‘P dephasing
at each dephasing time and the residues Phe-3 and Phe-9 have little 13CO—
19F(C16) dephasing (< 10 %) at r = 24 ms compared to corresponding dephasing
of the pH 5.0 samples. The latter observation may suggest either a shallower
insertion of lFP or less population of inserted IFP for the pH 7.4 samples relative
to the pH 5.0 samples. The justification of these statements is that the measured
dephasing «AS/80)”) for a system with multiple 13CO—X distances is the
combination of the AS/So due to each 13CO—X distance. If two different locations
of IFP are present in a membrane-associated IFP system, with one location
inserted into the membrane bilayer and the other one located at the water layer
above the surface of the membrane bilayers (cf. Figure 32), the measured 13CO—
”F(C16) (AS/So)“ will be smaller relative to the ”CO—”F(C16) (AS/So)” for
100% inserted lFP.

Similar to the lFP pH 5.0 samples, the values of (AS/So)°"” for IFP pH 7.4
samples were also corrected and the values of (AS/So)” were fitted to the
theoretical values, (AS/so)”. The corrected data and the best-fit simulated build-
up curves are shown in Figure 31 and the best-fit distances are listed in table 11.
In the REDOR data fitting, a single membrane location of IFP was assumed and
a 13C-X two-spin system was used as for the IFP pH 5.0 samples. The Gly-1
1300 was determined to be 4.6 A away from the phosphorus and the middle
residues, Ala-5, Phe-9 and Gly-13 13008 have distances of ~10 A from the 31P,
similar to the pH 5.0 samples. These distances and the flat 13CO—3'P dephasing

curve observed for lle-6 of pH 7.4 samples are consistent with the results from

123

IFP pH 5.0 samples. For residues Leu-2, Phe-3 and Gly-20, the 13CO—3‘P
distances for pH 7.4 samples are longer than the corresponding distances of the
pH 5.0 samples. For 100% inserted IFP, the longer distances suggest a deeper
insertion of IFP in the pH 7.4 samples and a generally shorter 13CO—19F(C16)
distance or bigger 13CO—19F(C16) dephasing will be expected. However, for the
pH 7.4 samples, only lle-6 was detected to be in contact with 19F(C16). A more
reasonable explanation for the longer 13CO—31P distances of the pH 7.4 samples
is therefore the existence of two different locations of IFPs in the membranes
with one in the membrane bilayer and the other one in the water layer above the
membrane surface. IFPs with the latter location may be loosely attached to the
membrane surface and not close to the phosphorus, which will result in longer
average 13CO—31P distances compared to the 100% inserted IFP. The models of

these two IFP locations will be discussed and pictorialized in the discussion

 

 

 
    

 

 

 

 

section.
‘30-3‘P 32ms 13C-19F(05) 24ms 13C-19F(C16) 24ms
0.3
(a) (b) (c)

Q 08- 0.2_ i
“A 2
ZS 04- 2 0.14 ~

0.04 2 ’1" 2 0.0- . . 4

0 5 10 15 20 0 5 10 15 20 0 5 10 15 20

 

 

 

  

residue

Figure 29. Summary of experimental REDOR dephasing (AS/So)” for the
spectra displayed in Figure 28. The (AS/So)MD values are shown as bars for
different residues and a typical uncertainty is :l:0.01-0.02.

124

1.2

 

 

(a) Gly-1 (b) Lou-2 (c) Phe-3
0.8 -

0.4 -

 

 

 

 

 

 

 

 

 

 

(AS/So)exP

 

 

 

 

 

 

 

 

1.2

 

0.8 "

0.4 . /

o.o~ % M =/' /

I I I I

0 1o 20 30 o 10' 20 30 0 1o 20 30
dephasing time

 

 

 

 

 

 

 

 

Figure 30. ‘3co—3‘P REDOR (dark solid line), 13co—‘9F(cs) REDOR (gray solid
line) and 13CO—‘9F(C1 6) REDOR (dotted line) experimental dephasing curves for
IFP samples at pH 7.4. The uncertainties are represented by the error bars and
are typically :l:0.01-0.02. The 13CC) labeled residues are labeled on tOp of each
spectrum. The samples used were the same as the ones used to take the
corresponding spectra in Figure 28.

125

(AS/so)cor

Figure 31. Plots of
time for (a-h) 13CO—

1.2

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

(a) Gly-1 (b) Leu-2 (C) Phe-3
0.84 °
0.4-
0.0. i/ o
1.2

(d) Ala-5 (e) Phe-9 (f) Gly-13
0.8- '
0.4-
0.0- O—Tf/ r—J/ /
1.2

(9) Gly-16 (h) Gly-20 (i) Ila-6
0.8-
0.4-
0.0« ’ L - 4 °

 

 

 

 

 

 

 

 

0 10 20 30

dephasing time

 

(AS/so)cor (circles and (AS/so)“ (solid lines) vs. dephasing
1P and (i) 1300— 9F(C16) REDOR of IFP samples at pH 7.4.

The 13CO labeled residues are labeled at the top of each spectrum.

126

 

 

 

(a) (b)

1.4

 

inserted IFP model surface attached IFP model

Figure 32. Insertion models for IFP in DTPC/DTPG membranes. The lipid head
groups are shown in blue and the alkyl chains are shown in gray. The gray balls
represent the labeled IFP backbone 13CO. (3) IFP inserted into the membrane
bilayer with the 13CO labeled residues indicated by the corresponding residue
number. (b) IFP located at the membrane surface and water layer interface. The
~10 water layer is above the membrane surface and is not shown in the model.

Table 11. 13CO—31P and 1300—19F(C16) distances for IFP samples at pH 7.4 a

 

Gly-1 Leu-2 Phe-3 Ala-5 lle-6 Phe-9 Gly-13 %y' Gly-20

 

“00:15: b
distance 4.6(0.1) 9.1(02) 8.4(0.2) 10.4(o.7) >11 10.4(03) 9.3(05) >11 7.0(02)
(A)

x2 59 6 54 6 N/Ac 12 4 18 44

 

 

133°—
_F(C16) >14” >14 >14 N/A 11.3(o.2)>14 >14 >14 N/A
distance

(A)
x2 N/A N/A N/A NIA 36 N/A N/A N/A N/A

 

a Uncertainties are given in parentheses and were calculated from the values within the x2 = x

mm +5 region.

b Calculated from the REDOR unlversal dephasing curve with AS/So = 0.1 at long dephasing time

(T = 32 ms for CO— P or r = 24 ms for CO— F) based on the consideration that only the
3005 with (AS/So)°xp > 0.1 at long dephasing time are fittable.

c N/A E not available.

127

4.3 DISCUSSION

The 13‘CO-31P and 13CO—19F(C16) distance measurements suggest that
both the N- and C- tenninal regions of IFP are close to the membrane
phosphorus headgroups (13CO-3‘P distance 57 A) relative to the middle region
of IFP (”CO—3‘P distance > 10 A) in a membrane with DTPC/DTPG composition.
The structure of IFP in this kind of membrane was determined to have the helix-
tum-helix motif in the chapter 3. These data can be combined to developt the
insertion models of DTPC/DTPG membrane-associated IFP. The IFP can either
adopt an inverted boomerang structure inserted into the membrane bilayer as
shown in Figure 32a or a boomerang structure located at the membrane
surface/water interface as shown in Figure 32b. In these insertion models, the N-
terminal region of IFP has a helix structure and a turn is formed around residue
Asn-12 followed by a short helix over residues 13-18 followed by some extended
structure. These structural characteristics were determined before and discussed
in chapter 3 and were retained at both pH 5.0 and pH 7.4. For the IFP samples at
pH 5.0, 13CO—“’F(C1 6) dephasing was detected for residues Phe-3, Ile—6 and
Phe-9 and data fitting indicated ~ 12 A 13CO—"’F(C16) distances for these
residues. IFP is more likely inserted into the membrane bilayer with the inverted
boomerang structure at pH 5.0. In this inserted model, Phe-3, lle-6 and Phe-9 are
located at the bottom of the helix and close to the membrane bilayer center and
lle-6 may have the deepest insertion compared to the other labeled residues
because no 13CO—3‘P contact was detected for lle-6. For the IFP pH 7.4

samples, 13CO—“’F(C1 6) dephasing was detected for only one residue, lle-6, and

128

the data fitting also suggested ~12 A 13CO—‘9F(C1 6) distance for lle-6. It is very
likely that there is at least some population of IFP that has the inverted
boomerang structure. However, little 13CO—19F(C16) dephasing was observed for
residues Phe-3 and Phe-9 for the pH 7.4 samples. The magnitude of 13CO-3’1P
dephasing at 32 ms is also less for some residues (Lou-2, Phe-3, Gly-16 and
Gly-20) of the pH 7.4 samples and the data fitting indicated longer 13CO—31P
distances compared to the corresponding pH 5.0 samples. A reasonable model
for these data is that IFP may have two different locations in the membrane at pH
7.4. As shown in Figure 32a and 32b, IFP may adopt both insertion models in the
membrane at pH 7.4. In the latter model, only Gly-1 and Gly-20 are close to the
phosphate headgroups and even Gly-20 is a little bit further away from the
membrane surface compared to the Gly-20 in pH 5.0 samples. This suggests
that IFP is more loosely associated to the membranes at pH 7.4 relative to pH
5.0 samples. Previous studies showed that a ~10 A water layer is associated with
the membrane surface.12 The other residues including Leu-2 and Phe-3 may be
located at this water layer above the membrane surface for IFP samples at pH
7.4. This explains the observed smaller 13CO-3‘P dephasing and longer 13CO—
31P distances for these residues. The less inserted IFP at pH 7.4 may Correlate to
its lower fusogenic activity at pH 7.4 and is probably associated with the multiple
acidic residues in the IFP sequence (Glu-11, Glu-15 and Asp-19), which will be
more negatively charged at pH 7.4 and repelled from the negatively charged

membrane.

129

Membrane lipids are very mobile and flexible. For the inserted IFP model,
IFP may disrupt the membrane bilayer and push the lipid molecules aside and
lipids are not aligned as in a non—disrupted bilayer. In such a system, the
measured 13CO-31P and 13CO—19F(C16) distances will not reflect the real
distances of labeled 13CO to the membrane surface or the membrane center but
rather the distances between the labeled 13CO and nearby 3‘P or 19F(C16) nuclei.
Even for residues that are close to the membrane surface, the detected 13CO—
31P REDOR dephasing may be smaller than expected since the lipids are pushed
away from the whole IFP. This may explain that the residues Gly-13 and Gly-16
have small magnitude of 13CO—mP dephasing and little 13CO—19F(C16)
dephasing. The plasticity of the membrane bilayer may also explain the
observation in the Chapter 3 of Glu-11 being in contact with water. There are two
additional charged residues at the C-tenninus cf IFP, Glu-15 and Asp-19, which
also have preference to stay in the water layer. Therefore, the membrane may be
disrupted by IFP in a way that the whole C-terrninal arm is exposed to water
while the N-terrninal helix remains in the membrane interior. In this case, the
lipids around the C-terminus will be highly disordered and water molecules will
leak into the membrane and have contact with the IFP C-tenninus. -

The V-shaped IFP insertion model combined with the IFP backbone
structure A and B determined in Chapter 3 (cf. Figure 19) can be correlated to
the IFP fusion activity at pH 5.0 and pH 7.4. With the B structure in the
membrane bilayer, the hydrophobic residues at the IFP N-terminal helix face the

membrane interior and have more favorable interaction with membrane bilayers

130

compared to the membrane-associated A structure in which the N-terminal
helical hydrophobic residues face to the membrane bilayer surface. Since the B
structure was only observed at low pHs (pH 4.0 and pH 5.0) and not observed at
pH 7.4, the better interaction between the B structure and the membrane bilayer
relative to the A structure explains the higher fusion activity of IFP at pH 5.0
compared to pH 7.4.

Our insertion model of IFP generally supports some of the previous
experimental results studies by ESR" 4 and some theoretical work10 that the
influenza fusion peptide is inserted into the hydrophobic region of membrane
bilayers. In the ESR studies of “FHA2” done by Macosko, et al., residues from
Ala-5 to Gly-8 and from Asn-12 to Trp-14 were studied and the residue with the
deepest insertion in their data fitting was IIe-6. In our study, the residue lle-6 also
seems to insert deeper than other 1300 labeled residues. However our results do
not agree with the straight helix structure proposed in this study and also do not
agree with the inserted boomerang structure suggested by Han, et al. using ESR
studies. An inverted boomerang stnlcture of IFP inserted into the membrane
bilayer is more consistent with our data. For samples with different pHs, our
results do not support a significant positional change of IFP in the membrane. In
my opinion, the membrane-associated IFP system is more complicated than a
single structure or a single IFP membrane location. We propose here that IFP
may have two different locations in the membrane bilayer with one having the
inverted boomerang structure inserted into the membrane and one in the water

layer above the membrane surface. We can not determine whether the IFP pH

131

5.0 samples have some IFP population with the latter location. However, it
seems like the IFP samples at pH 7.4 have a larger population of IFP with the
latter membrane location.

The HIV fusion peptide (HFP) was also studied using the solid-state NMR
methods in our group. It will be interesting to compare the insertion model of IFP
with that of HFP because both of them have similar biological function and both
HIV and influenza viruses belong to type-l enveloped viruses which share a
similar fusion mechanism”. The structure of HFP in DTPC/DTPG membranes is
only partially helical with a significant population of B strand conformation. This
complicated the structural analysis of HFP and the helical part of HFP was not as
well determined as IFP. The 13CO chemical shifts studies showed that HFP has
helical structure from residue lle-4 through residue Ala-14. The 13CO-31P and
1300—19F(C16) REDOR measurements showed that the helical part of HFP is
also inserted into a single membrane leaflet. The overall HFP has very similar
structure and location14 as IFP with the N- and C- terminal regions in close
contact to the phosphate headgroups and the middle region inserted into the

membrane hydrophobic region with contact with 19F(C16).

132

4.4 REFERENCES

1. Han, X.; Bushweller, J. H.; Caflso, D. S.; Tamm, L. K., Membrane
structure and fusion-triggering conformational change of the fusion domain from
influenza hemagglutinin. Nat. Struct. Biol. 2001, 8, (8), 715-720.

2. Clague, M. J.; Knutson, J. R.; Blumenthal, R.; Herrmann, A., Interaction of
influenza hemagglutinin amino-tenninal peptide with phospholipid vesicles: a
fluorescence study. Biochemistry 1991, 30, (22), 5491—5497.

3. Luneberg, J.; Martin, I.; Nussler, F.; Ruysschaert, J. M.; Henrnann, A.,
Structure and Topology of the Influenza-Virus Fusion Peptide in Lipid Bilayers. J.
Biol. Chem. 1995, 270, (46), 27606-27614.

4. Macosko, J. C.; Kim, C. H.; Shin, Y. K., The membrane topology of the
fusion peptide region of influenza hemagglutinin determined by spin-labeling
EPR. J. Mol. Biol. 1997, 267, (5), 1139-1148.

5. Hsu, C. H.; Wu, S. H.; Chang, D. K.; Chen, C. P., Structural
characterizations of fusion peptide analogs of influenza virus hemagglutinin -
Implication of the necessity of a helix-hinge-helix motif in fusion activity. J. Biol.
Chem. 2002, 277, (25), 22725-22733.

6. Efremov, R. G.; Nolde, D. E.; Volynsky, P. E.; Chemyavsky, A. A.;
Dubovskii, P. V.; Arseniev, A. 8., Factors important for fusogenic activity of
peptides: molecular modeling study of analogs of fusion peptide of influenza virus
hemagglutinin. FEBS Lett. 1999, 462, 205-210.

7. Bechor, D.; Ben-Tal, N., Implicit solvent model studies of the interactions
of the influenza hemagglutinin fusion peptide with lipid bilayers. Biophys. J. 2001,
80, (2), 643-655.

8. Huang, 0.; Chen, C. L.; Herrmann, A., Bilayer Conformation of Fusion
Peptide of Influenza Virus Hemagglutinin:A Molecular Dynamics Simulation
Study. Biophys. J. 2004, 87, (1 ), 14-22.

9. Sammalkorpi, M.; Lazaridis, T., Configuration of influenza hemagglutinin
fusion peptide monomers and oligomers in membranes. BBA-Biomembranes
2007, 1768, (1), 30-38.

10. Lague, P.; Roux, 8.; Pastor, R. W., Molecular dynamics simulations of the
influenza hemagglutinin fusion peptide in micelles and bilayers: Conformational
analysis of peptide and lipids. J. Mol. Biol. 2005, 354, (5), 1129-1 141.

133

11. Tilcock, C. P. S.; Cullis, P. R.; Gmner, 8., On the validity of 31 P-NMR
determinations of phospholipid polymorphic phase behavior. Chem. Phys. Lipids
1986, 40, 47-56.

12. Venable, R. M.; Brooks, B. R.; Pastor, R. W., Molecular dynamics
simulations of gel (Ll) phase lipid bilayers in constant pressure and constant
surface area ensembles. J. Chem. Phys. 2000, 1 12, (10), 4822-4832.

13. Jardetzky, T. 8.; Lamb, R. A., Virology: A class act. Nature 2004, 427,
307-308.

14. Qiang, W., Solid-state nuclear magnetic resonance studies of the
structures and membrane insertion of HIV fusion peptide. Ph.D. thesis, 2009.

134

Chapter 5 Solid-State NMR Studies of the IFP N-tennlnal Hellx in Orlented
Lipid Bilayers

5.1 BACKGROUND

The relative orientations of membrane proteins and peptides with respect
to the membrane surface or membrane normal are generally interesting topics for
research and can be correlated to their biological functions. As an example, there
has been a lot of effort to correlate the IFP helix insertion angle to its fusogenicity
and it has been proposed that an oblique insertion of the IFP helix in the
membrane plays an essential role in the fusion event.1 Some groups also
suggested that the lFP helix has different orientation in membranes at different
pHs which is associated with its pH-dependent fusogenicity? 3 However, results
from different groups have not reached a consensus. There are two major
techniques that have been used to study the tilted angle of IFP helix in the
membranes, ATR-FT IR and ESR spectroscopies. Two groups utilized the former
technique and proppsed that IFP and its fusion-active analogues had orientations
independent of pH but with different insertion angles relative to the bilayer normal,
~45° from one group1 and ~65° from the other group“. In another study using the
same technique and a peptide designed to resemble IFP, the peptide is helix with
an almost parallel-to-surface orientation at fusogenic pH 5.0 and either a random
or 55° insertion angle relative to the membrane normal at non-fusogenic pH 7.4.2
Two different groups also carried out ESR studies to develop a structural model
for the observation that IFP is much more fusogenic at acidic pH than at neutral

pH. One group used a major portion of HA2 domain (FHA2, residue 1-127) and

135

found the fusion peptide portion was a helix tilted 62° from the membrane normal
at pH 5 and 65° at pH 7.5 The other group used just the fusion peptide region
linked to a short hydrophilic peptide (GCGKKKK) and observed a 52° tilted angle
of the N-terrninal IFP helix with respect to the membrane normal at pH 5.0 and a
67° tilted angle at pH 7.4.3 There have also been efforts from molecular dynamics
studies to shed light on the membrane insertion of IFP. There are two studies
which supported that the IFP helix is parallel to or slightly tilted from the
membrane surface.°' 7 Other studies showed that the IFP helix is obliquely
inserted into the membranes and has a tilt angle either ~65° or ~30° with respect
to the membrane norrnal.°'10 In addition, the 61V mutant of IFP is fusion-inactive
and an IR study from one group supported a membrane surface orientation for
the peptide while a study from another group supported a transmembrane
orientation.” ‘2

In this chapter, we investigate the IFP orientational distributions in
membrane samples at different pHs using solid-state NMR. As in the
conformational studies, measurements at different pHs enable the correlation of
IFP insertion angles to its functionality. Experiments were also conducted on
membrane samples containing HFP or IFP mutants, lFP-G1V and IFP-G18, in
order to compare the insertion angle of IFP with other fusion peptides. The N-
terrninus of IFP is highly conserved and particular critical for HA function. The
functional activity of Ser HA (point mutation of HA Gly-1 to Ser) was tested with
red blood cells and the mutant was found to possess only hemifusion activity, i.e.,

only the outer leaflets of membranes are fused together with no contents mixing.

136

In the same study, Val HA (point mutation of HA Gly-1 to Val) was discovered to
have no fusion activity.13 IFP with the mutation of Gly-1 to Val was also probed
for is ability to induce liposome fusion and this mutant was found to lack the
ability to induce membrane vesicle lipid mixing.12 It will be interesting to compare
the orientations relative to the membrane normal of these mutants to the wild-
type IFP because different orientations of these peptides were proposed to be
associated with their different fusion activities. Studies from FI'IR showed that
wild-type IFP has an oblique orientation while the GIV mutant has a
transmembrane insertion.12 ESR studies on IFP GIV mutant indicated an
insertion angle of ~66° with respect to the membrane normal.11 In addition to
these IFP mutants, the insertion angle of HFP relative to the membrane normal
was also compared to that of IFP. HFP shares common features with IFP of
being the N-tenrlinus of the enveloped protein and glycine-rich. Comparison
between different native fusion peptides is potentially useful in generating a
common structure/function model.

In our study, aligned membranes were obtained by placing bicelles in the
magnetic field. Bicelles are generally composed of lipid and detergent molecules
in a certain ratio (roughly 2/1 — 5/1), which form rather uniform sized units.” ‘5
The long chain phospholipids such as DTPC form planar bilayers and addition of
detergent or short chain lipids such as DHPC can break up the bilayers into small
disks and form the disc-shaped phospholipid bilayers (of. Figure 6).16 These disc-
shaped bicelles can be automatically aligned in the magnetic field. Depending on

the lipid/detergent ratio, bicelles are normally 250 - 400 A in diameter and ~50 A

137

in thickness.” 18 Compared to lipid vesicles which are solely composed of lipids
or to micelles which are solely composed of detergents, bicelles are small
enough to be dissolved in the aqueous solution but big enough to not have fast
isotropic motions which are not favored by solid-state NMR measurement.
Bicelles have many advantages over the classical glass-sandwiched lipid
bilayers in which uniaxial orientation of membrane bilayers is achieved by
sandwiching phospholipid bilayers between stacked glass plates. Bicelles are
easier to prepare, adequately hydrated and easier to adjust pH. In addition, glass
plates waste a lot of space in the coil by filling the coil with glass rather by
sample, which leads to significantly lower sensitivity.

For peptides embedded in uniaxially oriented bicelles, NMR constraints
can be obtained to determine the orientations of the peptides relative to the
magnetic field and bicelles. The NMR constrains include anisotropic chemical
shifts, dipolar couplings and quadrupolar couplings. These values vary as a

function of the orientation of the molecule with respect to the external magnetic

field and follow the relation, al(¢)oc %(300s2¢—1), where al(¢) is the resonance

frequency or internuclear coupling, ¢ is the angle between the concerned tensor

(e.g. chemical shift tensor or dipolar coupling tensor) and the external magnetic
field. For a liquid sample, isotr0pic molecular motions average the anisotropic
interactions and therefore isotropic sharp peaks can be obtained; while for
unaligned solid samples, the NMR spectra are distributions of the resonances
due to all alignments and are called powder patterns. These powder patterns

reflect all the possible molecular orientations existed in a sample (cf. Figure 33).

138

However, for a single molecular orientation, a single peak will be observed and
the measured value (e.g. chemical shift or dipolar coupling) that corresponds to
this peak will change as the alignment is changed. Thus anisotropic NMR
constraints can provide valuable information on the orientation of the molecule
with respect to the magnetic field. The linewidth of the peak reflects the
orientational distribution of a sample.

In this chapter, data on chemical shift, dipolar coupling and quadrupolar
coupling measurements of bicelle samples are presented. The quadrupolar
coupling of 2H and 31F chemical shift were used to determine the alignment of the
bicelles. The 15N chemical shift and 15N—‘H dipolar coupling were used to

determine the orientation of fusion peptides relative to the bicelle normal.

 

 

 

 

 

 

 

511 022 C33

Figure 33. A sample powder spectrum of 1““CO with the principal values labeled
and the corresponding CO bond orientation. Chemical shift anisotropy arises
from the non-spherical distribution of electron density around a nucleus. The
degree of shielding or the effect of electron density on chemical shifts depends
on the orientation of the electronic cloud. The electron cloud is shown by the
ellipsoids and the relative orientation of the ellipsoid is shown for each. principal
value.

139

Figure 34. (a) structure of DMPCd54. (b—g and k) are 2H spectra obtained with the
quadrupolar- echo pulse sequence at 40 °C and processed with (b—g ) 25 Hz or
(k) 100 Hz Gaussian line broadening. (h—j) are 31P spectra obtained with the
1pda pulse sequence at (h, i) 40 °C or (j) 35 °C and processed with (h, i) 0 Hz or
(k) 100 Hz Gaussian line broadening. The peaks in the 2H spectra represent the
signals of 2 mol% of DMPCd54. The peaks in the 31F spectra represent the 31F
signals of the lipid phosphorus headgroups. Compared to the (k) 2H or (j) 31P
powder spectra, the spectra in (b—i) have sharp peaks and show the good
alignment of pure bicelles (b, c) or bicelles with incorporated IFP (d-i). Spectra (b,
d, f, h) are for the unflipped bicelle samples which have their normals
perpendicular to the magnetic field. Spectra(c, e, g, i) are for the flipped bicelle
samples which have their normals parallel to the magnetic field. The sample
composition used to obtain each spectrum was: (b, c) DTPC/DMPCd54/DHPC
(53:1:17 umol); (d—i) 0.7 mol IFP and DTPC/DMPCd54/DHPC (53:1:17 umol); (j)
0.8 mol IFP and DTPC/DTPG (16:4 umol); (k) 1.1 mol IFP and
DTPC/DMPCd54 (53:1 umol). The total number of scans was: (b—e, k) ~8000; (f, g,
j) ~2000; (h) 3; (i) ~100.

140

02 Dz Dz D; 02 02 O

03WOYO’P‘OwN,’
02 02 02 02 02 02 0' .\

D213202132 D202 ss H
(3
D3

02 02 02 Dz D2 D2 0

     

=fél§lifflii§i§lili€t

(b) (C)
{I II A pure bicelle h“ 1' I I II “I
(d) (e)
,\ l l /‘ bicelle WWW
pH 7.4

(fl (9)
bicelle with IFP
pH 5.0
30 2'0 1'0 0 -1'0 20 3030 20 1'0 0 -‘1'0 -2'0 -30
2H quadrupolar coupling (kHz)

0!) (i)
bicelle with IFP
pH 5.0“ M

 

 

 

 

 

 

100 50 6 50 400100 5'0 6 50 400
31 P chemical shift (ppm)
(1') (k)

  
  
 

  
     

unoriented membrane
dispersion with IFP
pH 5.0

 

 

100 5'0 0 -50 40030 20 1'0 0 -1'0 -2'0 -30
31F chemical shift (ppm) 2H quadrupolar coupling (kHz)

141

5.2 RESULTS
1. Bicelle alignment

It is important that bicelles are intact in the presence of fusion peptides in
order to use bicelles as the model system to investigate the peptide orientation in
membranes. Potential perturbations to bicelles include the alteration of bicelle
diamagnetic susceptibility, distortion of the disc shape of bicelles and disc
aggregation or fusion, all of which are likely associated with the loss of alignment
of bicelles in the magnetic field and can be manifested by NMR because loss of
alignment leads to broadening in the NMR spectrum. The alignments of bicelles
were demonstrated with 2H NMR spectra of bicelles containing 2 mol%
perdeuterated DMPC and with or without 2 mol% IFP (cf. Figure 34). The
difference between the samples in Figure 34b and 34c was the respective
absence or presence of 3 mM Yb°+ ions. As shown in Figure 34b, each site in the
perdeuterated chain (cf. Figure 34a) gives rise to a doublet which is symmetric
with respect to 0 kHz. The broad doublet with the largest splitting results from the
relatively immobile methylene deuterons near the ester linkage. The doublets
with smaller splittings result from the deuterons near the terminal portion of the
acyl chain which have bigger mobility. The very mobile terminal methyl groups
give rise to the most intense doublets with the smallest quadrupolar coupling.
Each of the splittings double when the orientation of the bicelle is flipped by
adding paramagnetic lanthanide as shown in Figure 34c. The good resolution of
peaks for each 2H labeled site is consistent with good alignment of the bicelles.

For an unoriented membrane dispersion, a powder pattern will be obtained in

142

which no individual peak for an individual 2H labeled site can be seen (cf. Figure
34k). The doubling in the quadrupolar splitting of flipped bicelles is consistent
with the change of bicelle orientation where bicelle normal changes from being
perpendicular to parallel to the magnetic field and is the result of a change in

order parameter from 822 = - 1/2 to Szz = 1. The order parameter is defined as

822 =%(3COSZ/1—1), where A is the angle between bicelle normal and the

external magnetic field.19 The spectra in Figure 34b and 34c were taken for
bicelle samples without fusion peptide and resemble the literature 2H spectra of
bicelles which have corresponding alignment in the magnetic field.19 The spectra
for bicelle samples containing fusion peptide at non-fusogenic pH (cf. Figure 34d
and 34e) were very similar to the ones for samples lacking fusion peptide and
indicated the good alignment of fusion peptide-incorporated bicelles. This
alignment was independent of pH as shown in Figure 34f and 349. Therefore the
bicelle morphology was not altered by having fusion peptide included even at
fusogenic pH.

The alignment of bicelles can also be detected by using 31P spectroscopy.
There is abundant 31F present in the lipid phosphate headgroups. Figure 34h and
34f show the 31P spectra of aligned bicelle samples containing IFP. The good
alignment of the bicelles was evidenced by the upfield or downfield dependence
of the peak shifts on respective perpendicular or parallel orientations of the
bicelles relative to the magnetic field direction. The spectrum of the unflipped
bicelle sample (cf. Figure 34h) resembles the 31F spectrum in the literature of a

sample with similar bicelle composition and the downfield peak in the spectrum

143

was shown to correspond to the 31Ps of DHPC headgroups in the bicelles”.
Compared to the linewidths of this spectrum, the linewidth for the flipped bicelle
sample is broad (cf. Figure 34i) because the added Yb3+ is likely associated with
the phosphorus headgroups and the paramagnetic ions broaden the peak as
traditional broadening reagents. The alignment was stable for the entire
measurement time in the NMR spectrometer demonstrated by the 2H or13‘P
spectra after other NMR acquisitions. In addition, the samples could be stored in

a -20 °C freezer, melted, and then re-aligned in the NMR spectrometer.

2. 1 D experiments

There has been disagreement in the literature about whether or not there
is significant pH dependence to the orientation of the N-terrninal IFP helix (Leu-2
to lle-10) relative to the membrane bilayer nonnal.3' 5 This disagreement is of
some relevance because a significant change in orientation plays an important
role in one of the structure-function models for the pH-dependence of IFP-
induced fusion. This section describes the study of IFP orientation using
anisotropic 15N chemical shifts.

There appeared to be good alignment of the first ten residues of IFP
relative to the bicelle normal as evidenced by much narrower linewidths In bicelle
samples containing IFP-UN, cf. Figure 35e, relative to membrane dispersion
samples containing IFP-UN, cf. Figure 35 a-b. Additional evidence for alignment
is the large difference in peak 15N chemical shift of IFP-UN in unflipped and

flipped bicelles, i.e. ~110 and ~150 ppm, respectively. Because the resonance of

144

each individual 15N site was not well resolved in these spectra, samples
containing singly labeled IFP were also studied. Spectra for these IFP samples
are shown in Figure 35 f-m. Each 15N spectrum of unflipped bicelles containing
singly labeled IFP is composed of a single narrow peak with ~5 ppm full-width at
half-maximum linewidth (except for IFP-A511). Observation of narrow 15N peaks in
unflipped bicelles is consistent with the expected fast rotation of the bicelle about
its normal axis. The peak 15N chemical shifts in flipped bicelles containing IFPs
with single Ala, Phe or Ile 15N labels are very different than the shifts in the
respective unflipped bicelles, cf. Table 12. This difference is consistent with the
observations in the IFP-UN samples. For IFPs with a single Gly 15N label, the
peak 15N shift has 514 ppm change for unflipped and flipped bicelle samples,
less than the change for other labeled residues which have ~ 30-40 ppm
changes in peak 15N shifts. The small 15N chemical shift anisotropy (difference in
peak 15N chemical shifts between flipped and unflipped bicelle samples) of Glys
relative to other residues was also observed by another group20 and could result
from near isotropic motions of Glys and/or very distinct chemical shift anisotropy
(CSA) principal values compared to other residues“ 22 which will be described
later in this chapter. For each flipped bicelle sample, the linewidth (~10 ppm) is
larger than in the respective unflipped sample. This difference can be partly
explained by the contribution to the linewidth from the mosaic spread of
alignment which reflects the errors in alignment and can be interpreted as the
variation or probability distribution of the bicelle normal directions in the magnetic

field”. The mosaic spread in alignment is two times larger for flipped bicelles

145

which is consistent with the order parameter change from Szz = - 1/2 to Szz = 1,
cf. Figure 34b and 34c. Another possible contribution to the linewidth of a flipped
bicelle sample is an increased transverse relaxation rate due to the paramagnetic
vb3+ ions.

The peak 15N chemical shifts of all the singly labeled IFP samples are
listed in Table 12. For most samples containing either unflipped or flipped
bicelles, the differences between 15N shifts at pH 5.0 and pH 7.4 are $3 or 56
ppm, respectively. These small differences are consistent with only minor pH
dependence of IFP N-terminal helix orientation relative to the bicelle normal. This
conclusion is further supported by very similar IFP-UN spectra at the two pHs, cf.
Figure 35 e, g and Figure 35 f, h.

A more detailed analysis of IFP orientation based on chemical shifts can
provide quantitative information on the IFP N-tenninal helix axis orientation with
respect to the bicelle normal. The measured chemical shift anisotropies of a
molecule can be greatly affected by the molecular motions and quantitative
analysis of the IFP N-terminal helix orientation from the 15N chemical shift data
must take into account all the motions of IFP. Possible motions considered in the
present work include rotations around fixed axes, e.g. helix axis or membrane
bilayer normal. Such motions, especially for the rotation around membrane
bilayer normal, can be partly assessed by comparing spectra of IFP-UN in
hydrated membrane dispersions to the spectrum of lyophilized IFP-UN which
serves as a standard powder pattern of IFP. The theoretical scaled chemical shift

anisotropy (A0) of a powder spectrum due to the rotational motions can be

146

calculated using Eq. 2.20 and then compared to the expen'mental Ad to get an
idea on the reasonable Euler angle values (cf. 0 and B in Figure 12). The spectra
in dispersions at pH 5.0 or 7.4 are very similar to the spectrum of lyophilized
material (cf. Figure 35 a-c). The 15N chemical shift anisotropy principal values
used in data analysis were determined from magic angle spinning spectra of
lyophilized IFP-UN, cf. Figure 35d, and were: 011, 65:2 ppm; 022, 781:2 ppm; and

033, 225:1:2 ppm resulting in chemical shift anisotropy (A0) of 154 ppm using

022 +011

A0=033— 2

(5.1)

The theoretical scaled A0 was calculated for each set of a and [3 values which
vary from 1° to 360° with an interval of 1° and compared to the measured A0:
The variation in a values has little effect on A0; however, variation in 6 causes
some change in the value of A0; e.g. a change of [3 from 0° to 10° results in the
change of A0 from 154 ppm to 146 ppm. The similarity of the powder spectra
between membrane-associated IFP and lyophilized IFP indicated B, the angle
between the principal axis 033 and rotational axis is small. Therefore the principal
axis 033 is approximately parallel to the rotational axis. This suggests that IFP
does not have fast rotation about the membrane bilayer normal because the
measured anisotropic 15N chemical shifts do not correspond to a parallel insertion
of the IFP helix relative to the bicelle normal. For such an orientation, the 15N
chemical shifts from flipped bicelle samples should be close to the downfield

edge of the powder pattern, determined from Eq. 2.22 and Eq. 5.1,

147

o(0)=§Aa-%(300520—1)+o,so=a33 , and this was not observed in our

experiments.

For a helix that is tilted from the bicelle normal, the distribution of the
measured anisotropic chemical shifts of different 15N labeled residues in flipped
bicelle samples is broad, ~30-80 ppm for a helix tilted >10° from the bicelle
nonrlal.24 This distribution can be substantially reduced for molecules with fast
motions on the time scale of NMR interactions (i.e., rate of motion >> NMR
interaction strength, ~10‘5 s'1 based on ~10 kHz CSA and dipolar coupling
interaction). As shown in Table 12, the measured chemical shifts of all the IFP
non-glycine residues fall into a very narrow (58 ppm) range, centered on ~150
ppm for the flipped samples. This narrow distribution suggests an IFP structural
model with significant mobility. The residues Gly-4 and Gly-8 are not considered
because they have different CSA principal values compared to other residues
and will have very different anisotropic chemical shifts even with the similar
orientation and mobility as other residues.

A reasonable model of motion for the bicelle-associated IFPs is the fast
rotation of IFP about its own N-terrninal helix axis. For IFP molecules with such
motions in membrane dispersion samples, the chemical shift tensors will be
axially symmetric leading to the powder spectra with steep edges. However, the
powder spectra shown in Figure 35a and b are from the 15N labeled first ten
residues at the IFP N-terminus and processed with 500 Hz Gaussian line
broadening. It is difficult to observe the characteristic symmetric powder patterns

in spectra with overlapping signals from 10 residues and broad line broadening.

148

Using the model of IFP with fast rotation around its helix axis, the angle
between the helix axis and the magnetic field, 9, can be calculated based on the
measured 15N anisotropic chemical shifts and Eq. 2.22. In this analysis, the
standard 15N CSA tensor orientations were assumed as shown in Figure 12a and
the CSA principal values were the aforementioned ones derived from the MAS
spectrum of lyophilized IFP. These values are generally consistent with the 15N
CSA principal values in the literature.24 For glycines, literature values of chemical
shift tensors were used, 011, 41 ppm, 022, 64 ppm and 033, 211 ppm.“22 In this
analysis, the bicelle motion relative to the magnetic field direction was also taken
into account, including the wobbling motion discussed in chapter 2 and a rapid
tumbling motion of bicelle and embedded IFP about the bicelle normal. Based on
the Eq. 2.26, the 15N chemical shift of a bicelle sample which rotates rapidly with
respect to the average normal direction is:25

2 1
0(9) = 5 AGEFF ' P2 (0080) ' SWODD "5(30082 I1. *1) '1' 0,80 (5.2)

where II is the angle between the bicelle normal and the magnetic field, which is
90° for the unflipped bicelles and 0° for the flipped bicelle samples. In the
calculation, 0.8 was used for Swobb; Aagpp was 159 ppm for Gly and 154 ppm for

other residues and 0130 was 105 ppm for Gly and 122 ppm for other residues. The
angle 0’ between the helix axis and the bicelle normal (6’ = 90°—t9 for the
unflipped bicelle samples and 6' = 6 for the flipped bicelle samples) from the

measured 15N chemical shift of each labeled residue in unflipped and flipped

bicelle samples is listed in Table 12.

149

Based on the analysis, the IFP N-terminal helix is determined to be tilted
~45° from the bicelle normal. For IFP samples at pH 5.0 with 15N labeled Phe-3,
Ala-5, Ila-6, Ala-7 or lle-10, Ia'unflrppad - a’mpped | 3 7° which supports the accuracy
of the analysis. Among these samples, Ala-5 and Ala-7 have the best agreement
in the calculated 6’ between the unflipped and flipped bicelle samples with
la'unfirpped — fl'flrpped | = 1° and 3° respectively; Ile—6 and lIe-10 have some
agreement in 6’ with la’unmpped — €111,0de = 6° and 5° respectively; Phe-3 has the
worst agreement with I6'unnippsd — B'flrpped | = 6°. For samples at pH 7.4, the
calculated 6' values are also more consistent between the unflipped and flipped
bicelle samples for Ala-7 than for Phe-3, with Ifi’unmppod — 6'11er | = 2° or 10°
respectively. This is probably because different residues generally have slightly
different CSA principal values and the difference between different types of
amino acids is even bigger“. While a small variation in A0 only results in a minor
change in 9’ (e.g., 9' only changes 102° when Ao changes 1:2 ppm), a similar
variation in org, causes a much greater change in 9’ (e.g., 9’ changes $2° when
also changes :2 ppm). In the analysis, only a single set of CSA principal values
were used, which were obtained from an lFP-UN sample. The obtained values
may agree better with the CSA principal values of Ala and have more deviation
from those of Phe-3, which leads to a bigger la'unflrppad — 6’11;de | for Phe-3. For
samples containing IFP with "N labeled Gly-4 or Gly-8, lamp,” - 8,1,,pr is 15-
21°. This is probably due to the inaccuracy of principal values used and/or
existence of additional motions of glycines which were not considered in the

calculation. However, this analysis shows a general consistency of Glys” 49's with

150

other residues. In addition, IB’pH 7,4 - 6'pH 5.0I 5 3° for all samples which suggests

that there is little change in N-terminal helix orientation as a function of pH.

Table 12. 15N chemical shifts (o) and the corresponding orientational analysis for the aligned IFP

 

 

samples a
pH 5.0 pH 7.4

Residue Unflipped bicelle Flipped bicelle Unflipped bicelle Flipped bicelle

0 (ppm) b 9' °‘ d 0 (ppm) 9' °' d 0 (ppm) 9' 0 (ppm) 9'
Phe-3 107 49 148 42 1 05 51 1 53 41
Gly-4 1 1 1 29 1 16 50 108 32 122 47
Ala-5 1 14 43 148 42 1 16 41 n.d. n; d.
lie-6 1 10 47 1 52 41 109 48 n.d. n. d.
Ala-7 116 41 144 44 114 44 150 42
Gly-8 109 31 1 1 0 52 n. d. n. d. n. d. n. d.
Phe-9 111 46 n.d.° n.d. n.d. n.d. n.d. n.d.
lIe-10 1 1O 47 1 50 42 n. d. n. d. n.d. n. d.

 

a Data were obtained with bicelle-associated IFP, IFP:DTPC ~0.02, and 40 °C nitrogen gas
temperature.

b The uncertainty in each chemical shift is estimated to be :l:1 ppm as determined from the
repetition of measurements.

c 0' is in degrees and is the angle between the helix axis and the bicelle normal. 0' = 90°—6i for
the unflipped bicelle samples; 6' = afor the flipped bicelle samples, where 0is the angle between
the helix axis and the magnetic field.

d The uncertainty in (9' is estimated to be 23° based on the uncertainty in the chemical shift.

° not determined

151

Figure 35. 15N NMR spectra of IFP which probe the orientation of the N-terrninal
helix axis relative to the membrane bilayer normal. Panels a-c are 15N static
spectra of IFP-UN that provide information about 15N chemical shift tensor
principal values and IFP motion in (a, b) hydrated membrane dispersions or (c)
lyophilized dry peptide without membranes. The similar appearances of spectra a,
b, and c suggests that there is not large amplitude motion of the N-terminal helix
in membranes with respect to the membrane bilayer normal at either (a) pH 5.0
or (b) pH 7.4. Panel d is a 15N MAS spectrum of lyophilized lFP2-UN that was
used to determine 15N CSA principal values. Panels e-m display 15N static
spectra of IFPs in the aligned bicelle samples. The incorporated IFP and the
sample pH are labeled above each set of spectra. For each labeled IFP, sharp
15N signals were observed and there was a significant change in peak chemical
shift as a function of bicelle orientation (lFP-G4N and lFP-G8N have less change
in the 15N chemical shifts and the possible reasons are explained in the text).
Both of these observations were consistent with a well-defined alignment of the
N-terrninal helix axis of IFP relative to the bicelle normal. For each labeled IFP,
there was little change in 15N chemical shift as a function of pH whiCh indicated
little change in average helix axis orientation with pH. The samples used to
obtain spectra a and b contained ~1 timol IFP and ~50 umol DTPC. The samples
used to obtain spectra e-m contained DTPC/DMPCd54/DHPC (53:1:17 mol) and
0.7 mol IFP. All the spectra were obtained with 1H-‘5N ramped cross-
polarization followed by 15N detection with 1H decoupling. For spectrum d, the
MAS frequency was 1.5 kHz. The temperature of the gas which flowed around
each sample was 40 °C. Spectra were processed using Gaussian line
broadening of magnitude (a, b) 500 Hz, (6, d) 100 Hz, (e-m) 50—200 Hz. The
number of acquisitions summed for each spectrum was 20000 - 60000.

152

(a) (b)
IFP in unoriented
pH 5'0 membrane bilayer pH 7'4
(C) (d)
lyophmzeleP
efizéo....6.....2.50....(.)...

15N chemical shift (ppm)
Bo

l

(e) lFP-UN

l 11
l A

(f) lFP-F3N
W LNWPH 5W
MW MPH 7%

(9)1FP-G4N

l,___

   

 

 

200 ' 100 ‘ '7 200 r 100
15N chemical shift (ppm)

153

(h) IFP-A5N

(DIFP46N

WH M

(j) IFP-A7"

 

 

(k)IFP-GBN
WWI-1 W
(I) IFP-F9"

(m)IFP-I10N
' 200 ' 100 ' ' . 200 ' 100 ' '

15N chemical shift (ppm)

Figure 35. 15N NMR spectra of IFP

154

3. 20 experiments

The orientation of the IFP N-terrninal helix was also assessed by ZD 15N
chemical shift and N-H dipolar coupling correlation spectroscopy, which is a
better way for orientational determination because the unique axis of the dipolar
frame is rigorously along the N-H vector. For the bicelle-associated IFP system,
the ZD spectroscopy can also provide additional support for the fast rotational
motion of IFP about its helix axis because dipolar coupling will be scaled by the
fast rotational motions and the dipolar splitting is determined by the angle
between the rotational axis and the magnetic field.

The 20 spectra were acquired for IFP samples with different 15N labels in
both unflipped and flipped bicelles at different pHs and are shown in Figure 36.
For each IFP sample with a single or double 15N labeled residue(s), a single peak
with a relatively narrow linewidth (100 Hz for IFP-F3NA7N and 400-500 Hz for
other samples) was observed for each 15N labeled site (of. Figure 36c-j). This is
consistent with the good alignment of the IFP N-terminal helix in membrane
bicelles and consistent with the chemical shift measurements. The big dipolar
coupling change (almost doubles for most of the residues) of the IFP with the
same 15N label in the unflipped and flipped bicelle samples provides additional
evidence for the IFP good alignment. For the spectra of IFP-UN, peaks of each
individual residue are not resolved and the observed narrow distribution of peaks
and the two-fold increase in dipolar coupling from the unflipped bicelle samples
to the flipped bicelle samples agree with results from the selectively labeled IFPs.

Comparison between the measured dipolar couplings for the sample at different

155

pHs shows only small changes and supports my hypothesis that the orientation
of the IFP N-terrninal helix has minor pH dependence (of. Table 13). Similar to
the 1D data, the experimental distribution of crosspeaks can be compared to the
theoretical distribution of a tilted helix and can provide some information on the
IFP motions. For a tilted helix with a tilt angle of > 10° relative to the bicelle
normal, a > 3 kHz distribution of dipolar couplings and > 30 ppm distribution of
chemical shifts are expected as calculated from Eq. 2.16 and 2.17. Particularly,
for the 15N—Hs of a helix tilted 40 — 50° from the bicelle normal, the theoretical
dipolar couplings and chemical shifts fall in a very broad range (0 - 8 kHz and
100 — 200 ppm respectively) and the magnitude of the dipolar splitting or the
chemical shift depends on the residue’s relative location in the helix.24 Both the
broad spread of chemical shifts and dipolar coupling can be greatly reduced by
molecular motions. As shown in Figure 36 and table 13, the distribution of
crosspeaks for different 15N labeled IFP is narrow, clustered at ~110 ppm, ~1 kHz
for IFPs in the unflipped bicelles, and at ~150 ppm, ~2 kHz for IFPs in the flipped
bicelles. This is best manifested by the 2D spectra from the IFP-UN samples
which have only one peak for the unflipped bicelle samples and one peak with a
small shoulder for the flipped bicelle samples (cf. Figure 36a and b). This nanow
distribution of crosspeaks from bicelle-associated IFP samples provides
additional support for an IFP structural model with large mobility, possibly a fast
rotation of IFP helix about its own helix axis which fits well with the experimental

data.

156

       

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Bo Bo
pH50 pH74
(a) lFP-UN (b) lFP-UN '3
% Q .......... b-2
o e --1
a unflipped CL") 0
G ' ‘1
H? ....... . mm _2
y . I pped
flipped flipped .........
3
(c) lFP-F3NA7N (d) lFP-F3NA7N '3
..................... T;
N - ‘ / \ - 1--2 I
2 \ 2A7 1 F3 A7 ' A7 F3 5
F3 A7 . f3 . I ‘ 1 .9
: ‘o l : ; ‘3, :--1 a
. . , :3
: - ‘ 8
g g . .1 3 0° ,10 5
1 : . .3
' . .1. '0
" r - "2102
flipped unflipped ‘ flipped unflipped ‘—
.................. 3
(e) lFP-G4N W IFP-G4N - ’3
0
g ........... --2
: l . --1
. unflipped .0
unflipped:
1 1 -1
......... 0
o .......... _2
flipped flippeg 3
200 150 100 200 150 100

15N chemical shift (ppm)

Figure 36. 20 15N chemical shift and N-H dipolar coupling correlation spectra

157

pH 5.0 pH 5.0

 

 

 

 

 

 

 

 

 

 

 

 

(g) lFP-A5N (h) lFP-I6N '3
(~2
Q l
o :--1
3-0
. 1.1 74‘
l . : g
2.“."T‘IRPPS’. 3- 2 E;
flipped 3 8
-3 3'5.
(l) lFP-G8N (i) IFP-HON .......... g
. - 'o
.__2 I
...9 ...... 9 ,2
e 1 e '-1 .—
l
0 f 0 1-0
unflipped . . j
e I I o .1
.......... . 1
O a .
flipped unflipped 1+2
flipped ..........
I I I I 3
200 150 100 200 150 100

15N chemical shift (ppm)

Figure 36. 20 15N chemical shift and N-H dipolar coupling correlation spectra of
bicelle-associated IFP that probe the orientation of the N-terminal helix axis
relative to the membrane bilayer normal. The samples used to obtain the spectra
contained (a, b) IFP-UN, (c, d) IFP-F3NA7N, (e, f) IFP-G41... (g) lFP-A5N, (h) IFP-
I610, (i) lFP-G8N and (j) lFP-I10N at (a, c, e, g-j) pH 5.0 or (b, d, f) pH 7.4. All the
samples have the composition of DTPC/DMPCd54/DHPC (53:1:17 umol) and 0.7
mol IFP. Each panel is separated into two parts by the dotted square with the
inside part representing the unflipped bicelle sample and the outside part
representing the flipped bicelle samples. For the spectra of lFP-F3NA7N, peaks
were assigned based on the measured chemical shifts of lFP-F310 and IFP-A7N in
Figure 33. All the spectra were obtained with Pl-WlM-z pulse sequence. Spectra
c and d were obtained with “Efree” free probe on the 21.4 T spectrometer. Other
spectra were obtained with the Varian Biostatic 1H/X probe on the 9.4 T NMR
spectrometer. The temperature of the gas which flowed around each sample was
40 °C. The number of acquisitions for each spectrum was ~100000.

158

The quantitative analysis of dipolar splitting can be determined using Equ.

2.27 with adjustment to the bicelle rotation around its normal:

1
V(6) = K'V” ”5(30082 ”-1)P2 (0089)-Swobb '

 

%(30082 ll —1)I (53)

where K = 0.67, v” = 22.6 kHz, Swobb = 0.8, and A is 90° for the unflipped bicelles
and 0° for the flipped bicelle samples. In the analysis, I) = 12° was used based on
the 8° - 15.8° tilt angle (0) of N—H bond relative to the helix axis for a helices in
the literature”. The 14° variations in r7 correspond to only 11° variation in the
calculated 9. The values of 9’ obtained from dipolar couplings are listed in table
13 and are ~45° and comparable to the corresponding angles obtained based on
the anisotropic chemical shifts. For all' the 15N labeled residues, la’unmpped —
49’”;de s 8°. Unlike the chemical shift analyses which were based on chemical
shift principal values with different amino types (especially glycine) having distinct
values, dipolar coupling analyses were based on the dipolar coupling constant
which is only related to the nuclei types involved and keeps constant for all the
residues. Compared to the 9’ values obtained for Gly-4 and Gly-8 from chemical
shifts, the angle 9’ for these two residues from dipolar coupling analyses is more
consistent between the unflipped and flipped samples and also more consistent
with other residues. For an IFP sample at different pHs, the difference between
6', la’pH 7,4 - a’pH ml 5 3° and supports the observations from the chemical shift
measurements that there is little change in the N-tem'rinal helix orientation as a
function of pH. The similarity between the spectra of lFP-UN at different pHs
provides further evidence for this pH-independent orientation of the IFP N-

tenninal helix relative to the bicelle normal.

159

Table 13. N-H dipolar couplings (v) and corresponding orientational analysis for the aligned IFP
samples a

 

 

pH 5.0 pH 7.4

Residue Unflipped bicelle Flipped bicelle Unflipped bicelle Flipped bicelle

v (kHz) " 9' °' d v (kHz) 9' Q d v (kHz) 9' v (kHz) 9'
Phe-3 1.1 43 2.0 48 1.1 43 2.3 47
Gly-4 0.8 41 1.9 48 1.3 44 2.3 47
Ala-5 n.d.° n. d. 1 .6 50 n.d. n.d. n. d. n. d.
lle-6 0.7 40 n.d. n.d. n.d. n.d. n.d. n.d.
Ala-7 1.0 42 2.0 48 1.2 43 2.2 47
Gly-8 0.9 42 1 .6 50 n.d. n.d. n. d. n. d.
Phe-9 n.d. n. d. n.d. n. d. n.d. n.d. n.d. n.d.
Ile-10 0.9 42 1.7 49 n.d. n.d. n.d. n.d.

 

a Data were obtained with the same samples and same experimental conditions as in table 12.
b The uncertainty in each dipolar coupling is estimated to be 1:250 Hz.
c 9' is in degrees and is the angle between the helix axis and the bicelle normal. 0' = 90°—0 for

the unflipped bicelle samples; 6' = 0for the flipped bicelle samples where His the angle between
the helix axis and the magnetic field.

dThe uncertainty in 6' is estimated to be 13° based on the uncertainty in the dipolar coupling.
° not determined

4. Comparison between 9.4 T and 21.4 T data

Some of the 10 15N chemical shift and 2D 15N chemical shift and N-H
dipolar coupling correlation spectra were taken on a 21.4 T spectrometer, which
is located in the biomolecular NMR facility in Michigan State University. The 1D
spectra of lFP-UN sample at pH 5.0 and ZD spectra of lFP-F3NA7N sample at pH
7.4 from 9.4 T and 21.4 T spectrometers are displayed in Figure 37. Based on
the signal-to-noise ratios of the 1D spectra, the sensitivity of the 21.4 T
spectrometer is ~4.5 times higher than the sensitivity of the 9.4 T spectrometer.

The peaks are better resolved in the spectrum taken with the 21.4 T

160

spectrometer. The linewidth of the spectrum taken with the 21.4 T spectrometer
is ~3.7 ppm. The linewidth of the spectrum taken with the 9.4 T spectrometer is
~6.5 ppm. The resolution is also significantly improved for the ZD spectra of the

lFP-F3NA7N sample when using higher field.

 

 

 

 

 

sxrr 21xlT
(a) (b)
260 150 160 56 2'00 150 160 5‘0
-3
(C) (d)
i:
Q ~ l--2 3:"
2’
n4 21
3
8
r l-o a
'5
£4
.1 ‘C
F:
2'
9 l' 22
3

 

 

 

 

 

 

200 150 150 200 150 160
15N chemical shift (ppm)

Figure 37. 1D and 2D spectra taken with either a 9.4 T (left) or a 21.4 T (right)
spectrometer. The 1D spectra were taken with the unflipped bicelle samples
containing lFP-UN at pH 5.0. The sensitivity of the two spectrometers is estimated
to be ~1:4.5 (9.4 T spectrometen21.4 T spectrometer) based on the signal-to-
noise ratios and the numbers of scans of the two spectra. The peak width is
reduced from ~6.5 ppm to ~3.7 ppm. The 2D spectra were taken with the flipped
bicelle samples containing lFP-F3NA7N at pH 7.4. The linewidth is narrowed for
the spectra taken with 9.4 T spectrometer compared to the spectra taken with
21.4 T spectrometer. All the spectra were processed without line broadening.
The number of acquisitions for each spectrum was (a) 5048; (b) 1042; (c) 6144
(12) x 20 (f1); (d) 1280 (12) x 92 (f1).

161

       

pH 5.0 pH 7.4

 

 

 

 

 

 

 

 

 

 

 

 

 

. -3
(a) lFP-G1S-F3NAg7N (b) lFP-G1S-F3NA7'N

Q. g ‘ i --2

0 0 § . 0 --1

a 0 '0 5 I I -o
r 5 -1 7°
3 0 a ' ° e15
3 .2 5:”
D) : ‘ s s
E : 8
:1 «:5
(c) lFP-G1V—F3NA7N (d) IFP-G1V-F3NA7N é.
‘. ‘ “2 VI
‘ 5 102
: . . . I“'1 ‘-

0 3 no 0 N -o

" or .1

' i I § 2

1 E i
'7 : fi . ‘ I 3
200 150 100 200 150 100

15N chemical shift (ppm)

Figure 38. 20 15N chemical shift and N-H dipolar coupling correlation spectra of
bicelle-associated (a, b) lFP-G1S-F3NA7N and (c, d) lFP-G1V- F3NA7N IFP at pH
5.0 (a, c) and pH 7.4 (b, d), Each panel represents the composite spectra of
unflipped (< 130 ppm) and flipped (> 130 ppm) bicelle samples containing same
labeled IFP at the same pH. The spectra for lFP-G18 and lFP-61V resemble
each other and resemble the spectra of wild IFP-F3NA7N which suggest the N-
terrninal helix of these two mutants and wild lFP have similar motions and tilted
angle. All the spectra were taken with Pl-WlM-z sequence. The samples have
the same composition as the samples in Figure 23. The number of acquisitions
for each spectrum is ~100000.

162

5. Comparison of the N-terminal helix orientation between IFP and its
mutants

It was shown that Gly-1 in the lFP sequence is very important and the only
allowed mutation is alanine. HA with a point mutation of HA Gly-1 to Ser can only
induce red blood cells lipid mixing without their contents mixing and HA with a
point mutation of HA Gly-1 to Val can induce neither lipid mixing nor content
mixing.13 It was proposed that a shallower insertion of the N-ten'ninal helix in the
lFP-G1V mutant compared to the wild-type lFP is correlated to this fusion activity
loss. While both the wild-type IFP and the lFP-G18 mutant have a tilt angle of
~50° between the N-tem'linal helix axis and the membrane bilayer normal, this
angle for the lFP-61V mutant is proposed to be ~65°." This big change in the tilt
angle should result in an obvious change in the 15N chemical shifts and the N-H
dipolar couplings for the same 15N labels in the wild-type IFP and IFP mutants,
which can be detected using solid-state NMR.

Figure 38 shows the 20 15N chemical shift and N-H dipolar coupling
correlation spectra for the Phe-3 and Ala-7 15N labeled lFP-G18 or lFP-G1V
mutant samples at pH 5.0 or pH 7.4. The structures of the two IFP mutants in
detergent micelles were determined by solution NMR before and were found to
have the similar helical structures with the micelle-associated wild-type IFP in
their N-temlinal regions.11 Based on the solution NMR results, two assumptions
were made in the analysis of 2D data in Figure 38: (1) The N-terminal regions
(residue Phe-3 through lle-10) of both mutants have helical structures similar to

the wild-type IFP N-terminal helix in bicelles and are independent of pH; (2) The

163

relative positions of the Phe-3 and Ala-7 crosspeaks for both mutants are similar
to that of the wild-type IFP and the assignments of the two crosspeaks for both
mutants were made based on the assignments of the wild-type IFP samples. The
chemical shifts and dipolar couplings derived from Figure 38 are listed in table 14.
For the samples at each pH with each bicelle alignment, both lFP-G18 and IFP-
G1V show very similar spectra compared to the wild-type lFP (cf. Figure 360, d
and 38). The magnitudes of the chemical shifts or the N-H dipolar couplings are
independent of the mutations of Gly-1 or the pH changes of the samples (cf.
table 12, 13 and 14). The two labeled residues, Phe-3 and Ala-7, have very close
chemical shifts and N-H dipolar couplings for the wild IFP and lFP mutants, ~110
ppm and ~ 1 kHz for the unflipped samples and ~150 ppm and ~2 kHz for the
flipped samples. The similar appearances of the corresponding spectra for all the
samples suggest the aforementioned assumptions are reasonable and also
supports similar motional and orientational motifs for the N-terrninal helices of
wild-type IFP and its mutants. Therefore the 15N chemical shifts and the N-H
dipolar couplings were analyzed in the same way as the data from wild-type IFP
using Eq. 5.3 and the obtained 9' are included in table 14. As in the wild lFP,
both the N-tenninal helices of lFP-G18 and lFP-61V mutants were modeled as
experiencing a fast rotation about the helix axis and the helix has a best-fit tilt

angle of ~45° relative to the bicelle normal.

164

Table 14. 15N chemical shifts (a), N-H dipolar couplings (v) and corresponding orientational
analysis for the aligned lFP mutant samples 8'

 

c. de 5.0 0' ‘pH 7.4

 

 

 

Residue 0(Ppm)b 9. v(kHz)° 6' a(ppm) 9' V(kHZ) 9'
lFP-G18 3:331)? 105 52 1'1 43 104 52 1.2 44
Phe-3 2:223: 151 41 2.0 48 153 40 2.2 47
lFP-G18 3:353“ 114 43 1.2 43 111 45 1.3 44
“3'7 23387;“ 147 43 2.2 47 151 41 2.2 47
lFP-61V 335:“ 111 46 0.8 41 110 47 1.0 42
Ph°'° 5:3"? 148 42 2.3 47 142 45 2.0 48
lFP-61V 3:232“ 114 43 1.2 43 116 41 1.0 42
“3'7 33°: 145 44 1.7 49 140 46 1.8 49

 

‘3 Data were obtained with bicelle-associated lFP-G18 or lFP-G1V, IFP mutantzDTPC ~0.02, and
40 °C sample nitrogen gas temperature.

b The uncertainty in each chemical shift is estimated to be :l:1 ppm as determined from the
repetition of measurements.

° 0' is in degree and is the angle between the helix axis and the bicelle normal. 0' = 90°-6 for the
unflipped bicelle samples; 0' = 0for the flipped bicelle samples.

d The uncertainty in 6' is estimated to be 13° based on the uncertainty in the chemical shift or
dipolar coupling.

° The uncertainty in each dipolar coupling is estimated to be 1250 Hz.

165

 

   

“" 19.4 W

30 2'0 1'0 9 30 -2o 730 3'0 2'0 1'0 ('3 -1'0 -20 -30
2H quadrupolar splitting (kHz)

(d)
me;: UN

HFPmn- -GSA6L7N unflipped mHFPmn-A14A15G16N unflipped

WW

(i) HFPdm-AGN unflipped (j) HFPmn-UN unoriented

250'III0'Ij'I250'Ut'01
15N chemical shift (ppm)

 

Figure 39. 2H quadrupolar splitting spectra of HFP in (a) unflipped and (b) flipped
bicelles show the good alignment of bicelles with incorporated HFP. 5N chemical
shift spectra of bicelle-associated HFP (c-i) have similar line shapes to the
powder spectrum of HFP (j) and suggest HFP not aligned relative to the bicelle
normal. The HFP used for each sample is labeled above each spectrum. All the
spectra in the left panels and spectrum (h) were taken with unflipped bicelle
samples. The spectra in the first three rows of the right panels were taken with
flipped bicelle samples. The spectrum (j) was taken with an unoriented
membrane bilayer sample. All spectra were taken with samples that contain 0.7
mol HFPmn or 0.4 mol HFPdm and DTPC/DMPCd54/DHPC (53:1:17 umol).
The temperature of the gas which flowed around each sample was 40 °C.
Spectra were processed using Gaussian line broadening of magnitude (a, b) 25
Hz, (c-f, j) 300 Hz, (9, h) 400 Hz, and (i) 500 Hz. The number of acquisitions
summed for each spectrum was: (a) 8000; (b) 3426; (c, f, i) ~26000; (d, g, h, j)
~10000.

166

 

6. Comparison of the N-terminal helix orientation between IFP and HFP

Both the HIV and influenza viruses belong to type I viruses and their
fusion proteins share similar topology and folding motifs.28 HFP and IFP are
different in sequence, but both of them are rich in glycine. Structural studies
show that both peptides have a major helical structure in membranes lacking
cholesterol at low FP/Iipid ratio and have predominant B strand structures in

membranes containing 33 mol% cholesterol.29

It will be interesting to study and
compare the motional and orientational properties of both peptides which can be
correlated to the fusion activities. The HIV fusion peptide induces fusion at
physiological pH 7.4 upon binding to the target membranes, so experiments were
only carried out with the HFP samples at pH 7.4.

Similar to IFP, HFP does not affect the morphology of bicelles and the
bicelles have good alignments with incorporated HFP (cf. Figure 39a, b). Figure
39c and d show the 15N spectra of bicelle-associated HFPmn with 15N labeling of
the first 14 residues of the N-terminus. Compared to the spectra of lFP-UN, the
spectra of HFPmn-UN are much broader and the linewidths are similar to the
spectrum of HFPmn-UN in unoriented membrane dispersions (of. Figure 39j),
which suggests that HFPmn is not aligned in the bicelle. The alignment of
HFPdm in bicelles was also studied with 15N chemical shifts to examine whether
the greater fusion activity of the HFPdm relative to the HFPmn3° is correlated to
an improved alignment of the HFPdm. Figure 39e and f show the spectra of the

bicelle-associated HFPdm-UN with the same 15N labeling as HFPmn-UN. Those

spectra are comparable to the spectra of HFPmn-UN in either bicelles or

167

 

membrane dispersions and are consistent with the unaligned HFPs in the
membrane bicelles.
HFPs were shown to have both a helical and [3 strand conformations in

l.29 The existence of these two different

the membranes lacking cholestero
conformations of HFPs in bicelles may also cause a broader spread in the
measured anisotropic 15N chemical shifts. In addition, partial alignments or
multiple alignments of HFPs in bicelles are other possible reasons for a broader
distribution of the measured 15N chemical shifts. However, the powder-shaped
spectra of bicelle-associated HFPs are probably mainly due to the unaligned
HFPs based on the following arguments: (1) Multiple sharp lines instead of a
broad peak should be observed for two well-aligned conformations of HFP; (2)
For a sample with partly aligned HFPs, at least some sharp peaks should be
observed or the appearance of the spectra of the unflipped bicelle and the flipped
bicelle samples should be different from each other and from the powder
spectrum; (3) The multiple alignments of HFPs should also result in multiple
sharp peaks instead of a powder-shaped spectrum. In order to get further
evidence for the above arguments, samples of HFPs with selective 1°N labels
were examined, including HFPmn-G5A6L7N, HFPmn-A14A15G16N and HFPdm-
A6N. The spectra are shown in Figure 399-i. All the spectra have low signal-to-

noise ratio due to the broad distribution of chemical shifts and have powder-

lineshapes and are consistent with unaligned HFPs.

5.3 DISCUSSION

168

The disruption of membrane bilayers by fusion peptides is considered to
be important for the membranes to overcome the kinetic barrier to fusion and is
correlated to the insertion of fusion peptides into the membranes. The membrane
insertion of IFP and HFP has been studied before and a lot of these studies have
been focused on the insertion angles of the fusion peptides relative to the
membrane bilayer normal. The solid-state NMR data presented in this chapter
show that the membrane insertion of FPs can be probed with the bicelle
membrane system which is an excellent model system to probe the membrane
peptide/protein orientation. It is significant that the bicelles remain intact and
aligned with incorporated FPs rather than being strongly perturbed by IFP or HFP.
The 2H NMR spectra showed good and controllable alignment of FP-incorporated
bicelles and indicated the feasibility of potential measurements of bicelle-
associated bigger constructs of fusion proteins (HA2 or gp41 for influenza or HIV
virus respectively), which would possibly provide valuable orientational and
motional information for the more biologically relevant fusion proteins.

The orientational information of FPs was explored by 15.N chemical shifts
and 15N-1H dipolar couplings. For the orientation derived from chemical shifts,
thorough knowledge of chemical shift tensors including the tensor orientation and
principal values is required and both can be determined using the 1°N spectra of
the unoriented membrane dispersion samples. However the 15N spectra of the
membrane-associated FPs with good signal-to-noise ratio were difficult to
acquire because of the reduction in GP efficiency due to molecular motions and

the limited FP concentration. The principal values used for non-glycine residues

169

of IFP were derived from the spectra of lyophilized lFP-UN. Two uncertainties
exist from this determination: (1) the spectra of the membrane-associated IFP
have some subtle difference in lineshapes compared to the spectra of lyophilized
IFP and could lead to slightly different principal values; and (2) Each residue may
have somewhat different principal values which can not be obtained from the
spectra of lFP-UN. For IFP Glys, their principal values were not experimentally
determined in my studies and literature values were used directly. The tilt angle
of the IFP N-terminal helix relative to the bicelle normal derived from 15N
chemical shifts is subject to these uncertainties and a ~20° difference in 9’ for
Gly-4 and Gly-8 between the unflipped and flipped bicelles was obtained. Unlike
the chemical shift tensors, the N-H dipolar coupling tensor is axially symmetric
and independent of residue type. For Glys, the results from dipolar coupling
measurements do not have the big deviations as in the results from the chemical
shift measurements. For other residues, the results from both chemical shift and
dipolar coupling measurements agree well with each other. The measured 15N-‘H
dipolar couplings are affected by the distribution of bicelle alignment and some
variations in the bicelle alignment such as the order parameter and the angle of
the bicelle normal relative to the magnetic field (e.g. ASwom = 10.1 and All = 110°
can change the derived 15N-‘H dipolar coupling by Av = 10.2 kHz for the
unflipped bicelle samples and Av = 10.4 kHz for the flipped bicelle samples). The
incorporation of FPs into bicelles and the DTPC/DHPC ratio both affect bicelle
alignment. Different bicelle-associated IFP samples are very likely to have

different Swab), and A which will lead to variations in the measured 15N-‘H dipolar

170

couplings. Therefore, the spectra of lFP-UN measure the 15N-‘H dipolar coupling
of each residue with the same Swab and A and the observation of single peaks
provides the best evidence for a helix with fast rotational motion. The fast
rotational motion of the IFP N-terminal helix can be fitted to the inverted
boomerang structural model proposed in the Chapter 4 considering the dynamic
nature of membrane lipids in the liquid-crystalline phase. As discussed in
Chapter 4, the entire C-terminal region of IFP is likely exposed to water. The
surrounding lipid and water molecules may provide enough interaction to
stabilize the lFP C-terrninus. The IFP N-terminal helix is not restricted by the C-
terminal region and can rotate relatively freely.

The present studies probed the dependence of the IFP N-tenninal helix
orientation on pH and mutations of Gly-1. The data are interpreted as: (1) The
IFP N-temiinal helix has a fast rotation about its axis; (2) The lFP N-terminal helix
has a tilt angle of ~45° relative to the membrane bicelle normal; (3) The angle
between the IFP N-terminal helix axis and bicelle normal is independent of the
sample pH; (4) This angle is also independent of the mutations of Gly-1 to Ser or
Val. The IFP N-tenninal helix orientation and its dependence on pH or Gly-1
mutations can be compared to some previous studies?"11 Our result on the lFP
N-terminal helix orientation is consistent with the ~45° insertion of IFP detected
by ATR-FTIR‘. Two different groups studied the IFP orientational dependence on
pH using ESR and tried to correlate the IFP insertion to its functional activity. The
angle between the lFP N-tenninal helix axis and the membrane bilayer normal

was determined to be 52° at pH 5.0 and 67° at pH 7.4 by one group using a

171

 

similar IFP construct as our group.3 In another study, the IFP part of FHA2
(residue 1-127 of HA2) was found out to be 62° or 65° away from the bilayer
normal for the samples at pH 5 or pH 7.5 The orientation of the lFP N-terminal
helix derived from solid-state NMR (45°) does not agree with these angles
mentioned above and does not agree with the ~15° change in the N-terrninal
helix orientation observed by the first group. If the 15° change was correct, we
would anticipate a ~30 ppm change in the 15N shifts and ~2.3 kHz change in the
1"’N-‘H dipolar splitting for the flipped bicelle samples at different pHs. This is not
inconsistent with our observation of 56 ppm and 50.4 kHz variations in the 15N
chemical shifts and 15N-‘H dipolar couplings respectively. Our results fit better
with the 3° change which would yield ~5 ppm change in the 15N chemical shifts
and ~0.7 kHz change in the 15N-‘H dipolar couplings for the flipped bicelle
samples at different pHs. The 15° difference in insertion angle between the two
pHs has been proposed to be coupled to a difference in the C-terminal
conformation (helical vs extended) described in the chapter 3. Our observation of
little change in either insertion angle or C-terrninal conformation as a function of
pH suggest that other models need to be developed to address the structure-pH-
dependent fusion activity relationship of IFP. Similar previous studies based on
ESR data showed ~15° change in N-terminal helix axis orientation when Gly-1
was mutated to Val. Our data do not support this model based on a similar
argument against the big pH dependence discussed above. We observed a
respective change of 53 ppm or 50.3 kHz in the 15N chemical shifts and 15N-‘H

dipolar couplings between wild-type IFP and lFP-G18 or lFP-G1V mutants. From

172

our studies, the orientation of the IFP helix axis is an intrinsic property of the
peptide sequence. This orientation does not change when the pH of the samples
changes or the single residue Gly-1 is mutated. The change in the fusion activity
is not correlated to a change in the orientation of the lFP N-terminal helix.

Our studies can also be compared to the results from molecular dynamics
simulations. IFP serves as a good model for simulations and extensive work has
been done to predict the insertion of the membrane-bound lFP. In one study, the
IFP was shown to locate near the lipid phosphate headgroups with an almost
parallel alignment to the membrane surface.7 In other simulations, IFP was found
to have oblique insertion relative to membrane bilayer normal. In one study,
simulations were also done to examine the effect of pH on the membrane
location and little dependence was found. Our results on the lFP N-terminal helix
orientation based on solid-state NMR studies provided valuable experimental
data to guide the molecular dynamics studies.

There are relatively fewer previous results on the orientational studies of ‘
HFPs. One polarized ATR spectroscopy study found that the a helical form of
HFP has an insertion angle of ~40° and the B strand form has an insertion angle
of ~30° relative to the membrane bilayer nonnal.31 In our studies, the HFP has
been shown to have complex secondary structures in membranes lacking
cholesterol. The B strand forms of HFP are mostly anti-parallel and have multiple
registries. The helical conformation extends from lie-4 to Leu-12, and residues
around Leu-9 have the best defined a helical conformation. The complexity in the

HFP secondary structures suggests a complicated insertion of HFP which is

173

supported by our observation of powder-shaped spectra for the HFP in aligned
bicelles. In contrast, the N-tenninal region of IFP has a single conformation of an
a helix in the bicelle system without cholesterol and the N-terrninal helix has a

single insertion angle of ~45° relative to the bicelle normal.

174

5.4 REFERENCES

1. Luneberg, J.; Martin, I.; Nussler, F.; Ruysschaert, J. M.; Herrmann, A.,
Structure and Topology of the Influenza-Virus Fusion Peptide in Lipid Bilayers.
Journal of Biological Chemistry 1 995, 270, (46), 27606-27614.

2. lshiguro, R.; Matsumoto, M.; Takahashi, 8., Interaction of Fusogenic
Synthetic Peptide with Phospholipid Bilayers: Orientation of the Peptide a-Helix
and Binding Isotherm. Biochemistry 1996, 35, (15), 4976-4983.

3. Han, X.; Bushweller, J. H.; Cafiso, D. S.; Tamm, L. K., Membrane
structure and fusion-triggering conformational change of the fusion domain from
influenza hemagglutinin. Nat. Struct. Biol. 2001, 8, (8), 715-720.

4. lshiguro, R.; Kiura, N.; Takahashi, S., Orientation of fusion-active synthetic
peptides in phospholipid bilayers: Determination by Fourier transform infrared
spectroscopy. Biochemistry 1993, 32, (37), 9792-9797.

5. Macosko, J. C.; Kim, C. H.; Shin, Y. K., The membrane topology of the
fusion peptide region of influenza hemagglutinin determined by spin-labeling
EPR. J. Mol. Biol. 1997, 267, (5), 1139—1148.

6. Bechor, D.; Ben-Tal, N., Implicit solvent model studies of the interactions
of the influenza hemagglutinin fusion peptide with lipid bilayers. BIOPHYSICAL
JOURNAL 2001, 80, (2), 643-655.

7. Sammalkorpi, M.; Lazaridis, T., Configuration of influenza hemagglutinin
fusion peptide monomers and oligomers in membranes. BIOCHIMICA ET
BIOPHYSICA ACTA-BIOMEMBRANES 2007, 1768, (1 ), 30-38.

8. Efremov, R. G.; Nolde, D. E.; Volynsky, P. E.; Chemyavsky, A. A.;
Dubovskii, P. V.; Arseniev, A. 8., Factors important for fusogenic activity of
peptides: molecular modeling study of analogs of fusion peptide of influenza virus
hemagglutinin. FEBS LETTERS 1999, 462, 205-210.

9. Spassov, V. 2.; Yan, L.; Szalma, 8., Introducing an implicit membrane in
generalized Born/solvent accessibility continuum solvent models. JOURNAL OF
PHYSICAL CHEMISTRY B 2002, 106, (34), 8726-8738.

10. Vaccaro, L.; Cross, K. J.; Kleinjung, J.; Straus, S. K.; Thomas, D. J.;
Wharton, S. A.; Skehel, J. J.; Fratemali, F., Plasticity of Influenza Haemagglutinin
Fusion Peptides and Their Interaction with Lipid Bilayers. Biophysical Journal
2005, 88, 25-36.

11. Li, Y.; Han, X.; Lai, A. L.; Bushweller, J. H.; Cafiso, D. S.; Tamm, L. K.,
Membrane Structures of the Hemifusion-lnducing Fusion Peptide Mutant 618

175

and the Fusion-Blocking Mutant G1V of Influenza Virus Hemagglutinin Suggest a
Mechanism for Pore Opening in Membrane Fusion. JOURNAL OF VIROLOGY
2005, 79, (18), 12065-12076.

12. Epand, R. M.; Epand, R. F.; Martin, l.; Ruysschaert, J. M., Membrane
interactions of mutated forms of the influenza fusion peptide. Biochemistry 2001,
40, (30), 8800-8807.

13. Qiao, H.; Armstrong, R. T.; Melikyan, G. 8.; Cohen, F. S.; White, J. M., A
specific point mutant at position 1 of the influenza hemagglutinin fusion peptide
displays a hemifusion phenotype. Molecular Biology of the Cell 1999, 10, (8),
2759-2769.

14. Sanders, C. R.; Hareb, B. J.; Howardc, K. P.; Prestegardc, J. H.,
Magnetically-oriented phospholipid micelles as a tool for the study of membrane-

associated molecules. Progress in Nuclear Magnetic Resonance Spectroscopy
1994, 26, 421 -444.

15. Sanders, C. R.; Landis, G. C., Reconstitution of membrane proteins into
lipid-rich bilayered mixed micelles for NMR studies. Biochemistry 1995, 34, (12),
4030-40.

16. Sanders, C. R.; Prosser, R. S., Bicelles: a model membrane system for all
seasons? Structure 1998, 6, (10), 1227-1234.

17. Sanders, C. R.; Schwonek, J. P., Characterization of magnetically
orientable bilayers in mixtures of dihexanoylphosphatidylcholine and
dimyristoylphosphatidylcholine by solid-state NMR. Biochemistry 1992, 31, (37),
8898—8898.

18. Forrest, B. J.; Reeves, L. W., New lyotropic liquid crystals composed of
finite nonspherical micelles. chemical review 1981, 81, (1 ), 1—14.

19. Prosser, R. 8.; Hunt, S. A.; DiNatale, J. A.; Void, R. R., Magnetically
Aligned Membrane Model Systems with Positive Order Parameter". Switching the
Sign of $22 with Paramagnetic Ions. J. AM. Chem. Soc. 1996, 118, (1 ), 269—270.

20. Yamaguchi, 8.; Hong, T.; Waring, A.; Lehrer, R. |.; Hong, M., Solid-state
NMR investigations of peptide-lipid interaction and orientation of a ss-sheet
antimicrobial peptide, protegrin. Biochemistry 2002, 41, (31 ), 9852-9862.

21. Lee, D. K.; Wittebort, R. J.; Ramamoorthy, A., Characterization of N-15
chemical shift and H-1-N-15 dipolar coupling interactions in a peptide bond of
uniaxially oriented and polycrystalline samples by one-dimensional dipolar
chemical shift solid-state NMR spectroscopy. J. AM. Chem. Soc. 1998, 120, (34),
8868-8874.

176

22. Oas, T. G.; Hartzell, C. J.; DAHLQUIST, F. W.; DROBNY, G. P., THE
AMIDE N-15 CHEMICAL-SHIFT TENSORS OF 4 PEPTIDES DETERMINED
FROM C-13 DIPOLE-COUPLED CHEMICAL-SHIFT POWDER PATTERNS. J.
AM. Chem. Soc. 1987, 109, (20), 5962-5966.

23. Quine, J. R.; Achuthan, S.; Asbury, T.; Bertram, R.; Chapman, M. S.; Hu,
J.; Cross, T. A., Intensity and mosaic spread analysis from PISEMA tensors in
solid-state NMR. Journal of Magnetic Resonance 2006, 179, (2), 190-198.

24. Marassi, F. M.; Opella, S. J., A solid-state NMR index of helical membrane
protein structure and topology. Journal of Magnetic Resonance 2000, 144, (1),
150-155.

25. Hemminga, M. A.; Cullis, P. R., 31P NMR Studies of Oriented
Phospholipid Multilayers. J. Mag. Reson. 1982, 47, 307-323.

26. Lazo, N. D.; Hu, W.; Cross, T. A., Low-Temperature Solid-State N-15 Nmr
Characterization of Polypeptide Backbone Librations. Journal of Magnetic
Resonance Series B 1995, 107, (1 ), 43-50.

27. Mascioni, A.; Eggimann., B. L.; Veglia, 6., Determination of helical
membrane protein topology using residual dipolar couplings and exhaustive
search algorithm: application to phospholamban. Chemistry and Physics of Lipids
2004, 132, (1), 133-144.

28. Jardetzky, T. 8.; Lamb, R. A., Virology: A class act. Nature 2004, 427,
307-308.

29. Qiang, W.; Weliky, D. P., HIV Fusion Peptide and Its Cross-Linked
Oligomers: Efficient Syntheses, Significance of the Trimer in Fusion Activity,
Correlation of beta Strand Conformation with Membrane Cholesterol, and
Proximity to Lipid Headgroups. Biochemistry 2009, 48, (2), 289—301.

30. Yang, R.; Prorok, M.; Castellino, F. J.; Weliky, D. P., A trimeric HIV-1
fusion peptide construct which does not self-associate in aqueous solution and
which has 15-fold higher membrane fusion rate. J. Am. Chem. Soc. 2004, 126,
(45), 1 4722-14723.

31. Castano, S.; Desbat, 8., Structure and orientation study of fusion peptide
FP23 of gp41 from HIV-1 alone or inserted into various lipid membrane models
(mono-, bi- and multibi—layers) by FT -IR spectroscopies and Brewster angle
microscopy. Biochim. Biophys. Acta-Biomembr. 2005, 1 71 5, (2), 81 -95.

177

Chapter 6 Summary and Future Directions

My work during the past years has been focused on the membrane-
associated influenza fusion peptide (IFP) and extensive aspects of IFP have
been studied including the secondary structure, membrane location and
orientation of IFP. The overall goal is to correlate these structural characteristics
to the functional activity of IFP and to gain some insights into the mechanism of
IFP-induced vesicle and target cell membrane fusion. All the structural
characterizations were done with IFP membrane samples at both pH 5.0 and pH
7.4 in an effort to find the structural basis/bases for the pH-mediated lFP-induced
membrane fusion. In my work, solid-state NMR was mainly used.

The secondary structure of membrane-associated IFP was studied by the
13C chemical shift and internuclear (13CO — 13CO and 13CO - 15N) distance
measurements. It was shown that the physiologically relevant membrane
cholesterol content is correlated with a large population of B strand IFP at both
fusogenic pH 5.0 and non-fusogenic pH 7.4. In membranes that lacked
cholesterol (DTPC/DTPG, 4:1 or POPC/POPG, 4:1), predominant a helical IFP
conformation was observed with C—temlinal helical conformation at both pH 5.0
and 7.4. Specifically, the IFP structure starts with an N-terminal helix ranging
from Leu-2 to Glu-11 followed by a turn at Asn-12 followed by a C-terrninal helix
(structure A). This structure was observed for membrane-associated IFP at both
pHs. An additional similar structure with some variations in the turn region was

observed for the membrane-associated IFP at pH 5.0 with the N-terminal helix

178

 

extending from Leu-2 to lie-10 and the turn at Glu-11 and Asn-12 redirecting the
C-terrninal helix (structure B). The ratio of structure Alstructure B is ~3:1 for
membrane-associated lFP at pH 5.0. The sthctural variation at residue Glu-11 is
responsible for the two different structures, namely two different conformations
were observed for Glu-11 based on two detected chemical shift sets which were
consistent with a helix and [3 strand structures respectively.

The changes in the protonation states of the Glu-11 side chain COOH
were also examined with the 13C chemical shifts for IFP samples at different pHs.
As the pH changed from 7.4 to 4.0, the ratio between the population of the
protonated state and the non-protonated state (COOH/COO‘) was ordered pH
4.0 > pH 5.0 > pH 7.4, which indicated that Glu-11 is sensitive to the solution pH
and in contact with the water layer above the membrane surface.

Because membrane disruption is thought to be important during the
membrane fusion and is correlated to the membrane insertion of fusion peptide,
the membrane location of helical IFP was studied with 13CO - 3‘P and 13CO — 19F
distance measurements. The detectable contact between the lie-6 13CO and 19F
(C16) (~12 A) indicated that the IFP middle region inserts into the membrane
outer leaflets at both pHs. The 13co — 3‘P REDOR data showed that both the N-
and C-terrninal regions of IFP (Gly-1 and Gly-20) 13008 are close to 31Ps in the
lipid phosphate headgroups (57 A). The results of Gly-1 and Gly-20 being close
to the membrane surface and the IFP middle region being inserted into the
membranes are consistent with the helix-tum-helix IFP structure and suggest an

inverted boomerang structure of membrane-associated IFP. Additional contacts

179

between 1300s of Phe-3 or Phe-9 and 19F (C16) (~12 A) for membrane-
associated IFP at pH 5.0 suggested a deeper/greater population of membrane
insertion of IFP at pH 5.0 relative to pH 7.4.

In addition to the membrane location of the helical IFP, I also studied the
IFP N-tenninal helix orientation relative the membrane bicelle normal using
anisotropic 15N chemical shifts and 15N-‘H dipolar couplings. The narrow peaks
(linewidth < 10 ppm for chemical shift and < 500 Hz for 15N-‘H dipolar coupling)
and the big change in peak positions between unflipped and flipped bicelle
samples (~ 40 ppm in chemical shifts and ~ 1.0 kHz in dipolar couplings)
indicated a good alignment of the IFP N-terrninal helix in bicelles. The narrow
distribution of peaks (the range of chemical shifts s 11 ppm for non-glycine
residues and the range of dipolar couplings $0.6 kHz) suggested appreciable
mobility of IFP in membrane bicelles. The 15N chemical shifts and 1"SN-1H dipolar
couplings were analyzed and fitted by a model of fast rotation of the IFP N-
tenninal helix about its helix axis. The analysis showed a ~45° tilt angle of the
IFP N-terrninal helix relative to the bicelle normal. The similar appearance of
spectra for IFP samples at both pH 5.0 and pH 7.4 and for lFP-G18 or lFP-G1V
mutants indicated that the IFP N-terrninal helix has a common motional and
orientational motif regardless of pH or point mutations to Gly-1.

The proposed fast rotational motion of the IFP N-tenninal helix may first
appear surprising and difficult to fit with the inverted boomerang structure.
However, such a rotational motion is possible considering the highly dynamic

nature of lipid molecules in the liquid-crystalline phase and the small size of an

180

lFP molecule (comparable to a lipid molecule). As mentioned previously, the
residue Glu-11 is in contact with water. This fact combine with the inverted
boomerang structure suggests that the membrane bilayers are likely being
dismpted in a way that the entire IFP C-terrninal arm is exposed to water while
the C-temtinus remains associated with the lipid headgroups. The C-terrninal
region may be stabilized by the interaction with lipids and water but not by the
hydrogen bonds between the IFP N- and C-terrninal regions. Therefore the IFP
molecule is possibly flexible enough to enable the rotation of the N-terrninal helix
and retain its secondary structure.

My work features a combination of MAS and static solid-state NMR
studies of membrane-associated IFP. Extensive aspects of lFP structural
characteristics were explored using these methods and demonstrated the power
of this combination. Many remaining questions can be further probed with these
methods. For trans-membrane proteins/peptides, the bicelle system has been
shown to be successful to determine the protein/peptide structure 1'5 and showed
little difference from the results from the mechanically aligned samples (stacked
glass plates)°. However, there is no precedent of its utilization for a peptide with
significant motions which only inserts into a single membrane leaflet. The
motions of bicelles complicate the analysis of the data and the existence of short-
chain lipids may alter the peptide insertion by interacting with the peptide. It will
be interesting to use mechanically aligned samples to study the membrane-
bound IFP. The 20 15N chemical shift and 15N-‘H dipolar coupling correlation

spectra can provide information not only on the IFP orientation but also on the

181

 

IFP secondary structure for the mechanically aligned samples with limited
motions. The IFP orientation from different sample types should be compared in
order to get further support for the proposed IFP motional and orientational

model. The comparison between the results from different methods will also shed

the light on the suitability of bicelle systems for the non-trans-membrane peptides.

In addition, the lipid compositions of bicelles are limited and the lipid
compositions of the membrane bilayers for stacked glass plates can be varied
substantially and can be used to study the system with cholesterol.

Another interesting question is whether the B strand IFP also inserts into
the membrane and has pH dependence of the insertion. Comparison between
helical IFP and helical HFP membrane locations indicated that both peptides
have similar insertion with the N- and C-terrninal regions in close contact to the
lipid phosphorus headgroups and the middle region inserted into the membrane.
The B strand HFPs were shown to have a similar insertion motif and the depth of
insertion is correlated to their functional activities where a deeper insertion
correlates to a higher fusion activity. The interpretation of 13CO — 3‘P and 13CO -
19F data for B strand HFP seemed to be more straightforward relative to the a
helix HFP because the helix forms wheels and the relative locations of residues
on the helix need to be considered and complicate the data interpretation. It will
be interesting to study the insertion of [3 strand IFP in cholesterol-containing
membranes at different pHs and correlate the insertion to its function at different
pHs. Some preliminary data have been obtained for the B strand IFP in

membranes than contained cholesterol and are displayed in the appendix.

182

l

 

REFERENCES

1. Angelis, A. A. D.; Opella, S. J., Bicelle samples for Solid-State NMR of
membrane proteins. Nature protocol 2007, 2, (10), 2332-2338.

2. De Angelis, A. A.; Nevzorov, A. A.; Park, S. H.; Howell, S. C.; Mrse, A. A.;
Opella, S. J., High-resolution NMR spectroscopy of membrane proteins in
aligned bicelles. J. Am. Chem. Soc. 2004, 126, (47), 15340-15341.

3. Howard, K. P.; Opella, S. J., High-Resolution Solid-State NMR Spectra of
Integral Membrane Proteins Reconstituted into Magnetically Oriented
Phospholipid Bilayers. J. Magn. Reson., Series B1996, 112, 91-94.

4. Kochendoerfer, G. G.; Jones, D. H.; Lee, S.; Oblatt-Montal, M.; Opella, S.
J.; Montal, M., Functional characterization and NMR Spectroscopy on full-length
Vpu from HIV-1 prepared by total chemical synthesis. J. Am. Chem. Soc. 2004,
126, (8), 2439-2446.

5. Park, S. H.; Nevzorov, A. A.; Wu, C. H.; Opella, S. J., Three-Dimensional

Structure of the Transmembrane Domain of Vpu from HIV-1 in Aligned
Phospholipid Bicelles. Biophys. J. 2006, 91, (8), 3032-3042.

183

Appendix I

Natural Abundance Corrections for (AS/So)”

This section considers the determination of (AS/So)” for 13C—15N, 13C—31P
and ‘30—‘9F REDOR, the contribution to (AS/So)°"° due only to labeled nuclei.
(AS/So)” is equivalently described by the parameter “f', the si/so ratio for the
labeled 13CO nuclei considering only other labeled nuclei. Comparison of
(AS/So)” to (AS/So)°’”’ yields the labeled nuclei internuclear dipolar couplings
and distances. Contributions to So and 81 are considered for different

experiments.

1.Determination of (AS/So)” for 13co—“N REDOR

Determination of (AS/So)” relied on estimating the natural abundance
(n.a.) contribution to (AS/So)” and this estimation was based on the following
parameters/approximatlons.
A1. There is 99% labeling of the 13CO and 15N nuclei. 81 = So for a labeled 13CO
in IFP with a 13co---“°'N pair.
A2. (1) S1 = 0 for a labeled 13C0 separated by one or two bonds from a natural
abundance 15N. The 81 is not affected by other natural abundance 15N. (2) $1 = 0
for natural abundance backbone 13003 which are separated by one or two bonds
from the labeled 15N. 81 = So for other natural abundance backbone 13CO sites.

Criteria (1) and (2) are based on the close distance (5 2.5 A) and consequent

184

4m. can em 9 2259.528 mg new £26.38 9:32 mg .cozatowoc cam 35 “cocoa—com 5 9m xon some 5 952 Soc.
2? .ztzoflodu. eeo 2.8.5811”: sauce/1.11.8 1081 12839.88 8888 so coco 26.”. .2 2:9“.

j

-- as:
E .
z<§znbai

 

Az<~4xzabai
2.“: m: 3.8: 5:5 1 23 m: >93: 22,

 

 

 

 

 

_ E 1 § -. e- llSli: . j 1 1
m E _ E E E _ :
oimé o<~ 0: , z: , 23.034
23088 9. .350 Waconcmo m: 353: 3:0 N> 2m? _ 233002 223002.
l E l, --.,--._.l. . w.
2 3 1.
2 2 1
0E a ,2

2:098 m: an:

 

 

 

m_>co€mo 8.32m. an: v

r l I ._

2m 2 8:32:88
.8 o. 8:39:88
cos—223a 323.3.
cozatomoo 85

185

strong (2 200 Hz) dipolar coupling of 13'CO and 15N nuclei separated by one or
two bonds.

The (AS/So)” expression is calculated using the following parameters:
U0 and UN, the fractions of 12CO and 14N of the labeling CO"'N sites, respectively;
Ac and AN, the 13C and 15N natural abundances, respectively; and n, the total
number of unlabeled peptide backbone CO sites in the IFP. A flow chart for the
determination of (S1/So)°°’ or “f ” is given in Figure A1.

A complete derivation of (AS/So)” follows:

 

[ASJW = SSW—Sf” (A1)

33' 83"”

88"" is expressed as the sum of contributions from labeled 13CO nuclei ($63”)
and from natural abundance 13CO nuclei (sg-a- ):

83"” = "Masai:=1—uC+nAC (A2)
8?“ is also expressed as the sum of contributions from labeled 13CD nuclei

(S1’ab ) and from natural abundance 1300 nuclei (81M ):

S1exp = S1Iab + S1n.a. (A3)
with:
slab=(1— uC — u~)(1 — 2AN)f + UN (A4)
and:
81": = (n ‘2)Ac (A5)

186

where 1 — UN is the fractional 15N labeling of IFP, AN is the fractional 15N natural

abundance and the parameter 1‘.

 

cor cor_ cor 00’
f=L=1_S° 81 :1. 3.8. (A6)
88‘” 88“ So

Incorporate Eq. A6 into Eq. A4:

slab = (1 — uC — U~)(1—2A~)[1-[§Ta] + UN

 

(A7)
AS °°'
= (l—uc—u~)(1-2A~>— (1-uc—u~>(1-2A~)[—S;] +u~
UC, UN, and 2A~ are much less than 1 so that:
(1-UC—UN)(1—2AN) E1—UC—UN-2AN (A8)
and:
AS °°r
848‘) .2. 1—UC—2AN— (1—UC—UN— 2AN)(_S_] (A9)
0
Incorporate Eqs. A5 and A9 in Eq. A3:
exp AS cor
S1 =1-UC-2AN—(1-UC—UN—2AN) S— +(n-2)AC (A10)
0
Combine Eqs. A2, A3, A4, and A10:
exp exp AS car
80 —S1 =[1—UC+nAc]— 1—UC—2AN—(1—UC-UN—2AN) ? +(n—2)AC
0
(A11)

and simplify:

187

CO!"
83” — Sfxp = 2Ac + 2A~ + (1 — uC — UN - 2A~)[§] (A12)

Combine Eqs. A2 and A12:

 

 

 

exp 2AC+2A~+ (1-UC—UN-2A~)[A§]
A—3 = 8° (A13)
80 1—UC+nAC
and rewrite:
[332]”: 1-UC+nAc [As)°"p_ 2AC+2A~ (A14)
80 (1‘Uc-UN-2AN) S0 (1-UC'UN-2AN)

Expressions in Eq. A14 were numerically evaluated using Ac = 0.011, AN =

0.0037, n = 27, and Uc = UN = 0.01:

cor exp
[Q] = 1.323 [9% — 0.030 (A15)
80 S0

The uncertainties in (AS/So)” were calculated as:

2 2
am =J300§1+310§0

 

 

83 (A16)
The 0"“ resulted from Eq. A15 was:
0°“ = 1.323 09"” (A17)

2. Determination of (AS/So)” for 1ice—3’1» and 1“co-“F(cni) REDOR
Similar to the 13CO—“r’N REDOR data correction, the natural abundance

correction starts from considering that both So and 81 come from the labeled

188

13CO and the natural abundance 13COs (including unlabeled residues of IFP and

13co from the 9 mol% 16-19F-DPPC):

 

(gym = 330' +sg°(/FP)+sga(DPPC)-s,°°’ -s1"a(IFP)—Si’°(DPPC) (A13)
3 33“ + sgauFP) + sgawPPC)

A few algebraic manipulation leads to

 

00' ne ne exp
[as] =(”so (IFP)+SO (DPPC))_[_A_§]

3—0 8” So

"a na ”a ”a
_§L%P_).[§] (IFP)—§°(eoriC)-[£] (DPPC)
So 80 So 80

(A19)

The terms (AS/So)”, (AS/So)"‘(IFP) and (AS/So)"‘(DPPC) represent the
contribution to REDOR dephasing from the labeled 13CO, the natural abundance
of the unlabeled residues and the natural abundance 1300 from 16-‘9F-DPPC,

respectively. The So term in Eq. A19 has the numerical values

sgor =1
83°(IFP) = 26x0.011= 0.286 (A20)
338(DPPC) = 2.5x2x0.011= 0.055

The values of (AS/So)”a(lFP) were calculated from the experimentally available
(AS/So)” for the sample at different pHs and the values (AS/So)"‘(DPPC) were
calculated using a 1""CO—31P or 13CO—‘S’F two-spin system with literature
internuclear distance of the DPPC molecule (Venable et al., J. Chem. Phys,

2000, 112, 4822-4832). Table A1 and A2 provide the numerical values.

189

 

Table A1. The (AS/So) ”5 (lFP)

 

 

 

13CO_31 P a 13CO-(16-19F) a b
Peptide pH 5.0 pH 7.4 pH 5.0 pH 7.4
0.045 0.018 0.032 0.031
0.160 - 0.118 0.018 0.020
r(ms) 16 0.319 0.204 0.070 0.069
24 0.395 0.266 0.095 0.051
32 0.460 0.339 n.d. °

 

aThe 1300-31P values were based on the (S1/So)°"p of samples labeled at Gly-1, Leu-
2, Phe-3, Ala-5, lie-6, Phe-9, Gly-13, Gly-16 and Gly-20. The 13co-“’l=(C16) values
were based on samples labeled at Gly-1, Lou-2, Phe-3, lie-6, Phe-9, Gly-13 and Gly-

16.

bThe maximum 1' for 13C0-19F experiments was 24 ms.
° n.d. 5 not determined

Table A2. The (AS/So) "8 (DPPC) ’3

 

 

0.025 0.000

0.319 0.000

7 16 0.695 0.000
(ms)

24 0.811 0.000

32 0.907 n.d. °

 

8A distance of 15.2 A was used for 13CO-19F(C16).
b The maximum 1 for 13CO—19F experiments was 24 ms.
c n.d. 5 not determined

190

Appendix II
Some Additional Data for lFP
This section will present some complementary data for the membrane- or
bicelle-associated IFP samples. These data include some 31P spectra of
membrane-associated IFP samples, 13C0-31P and 13CO—19F(C5) REDOR
measurements for lFP membrane samples that contained cholesterol and two
15N chemical shift and 15N-‘H dipolar coupling correlation spectra of the lFP C-

terminus.

1. 31P spectra

The dependence of 31F spectra of membranes with bound IFP on pH,
temperature and membrane composition was tested and the spectra are
displayed in Figure A2. All the spectra have similar appearance with each other
and with the spectra shown in Figure 22. The lineshapes of these spectra agree
with the lamellar phase of the membrane samples and are independent of
temperature, pH and membrane lipid compositions.
2. REDOR measurements

Figure A3 displays the representative 13CO—31P and 1:‘COA—19F(C5)
REDOR spectra at long dephasing time (32 ms for 13CO—31P and 24 ms for
13‘C0—"’F(C5) REDOR measurement) for samples that contained
lFP/DTPC/DTPG/CHOL (O.8:16:4:10) at pH 5.0. For the 13co—1"l=(cs) REDOR
measurements, additional 9 mol% of 5-‘9F-DPPC was also incorporated into the

samples. The peak chemical shifts are summarized in table A3 and compared to

191

the database distribution of the 13CO chemical shifts for corresponding residues
with different secondary structures. In particular, Gly, Ala and Phe have the 13CO
chemical shifts of 175.51 1 1.23, 179.40 1: 1.32, 177.13 1 1.38 ppm for the helical
conformation and 172.55 1 1.58, 176.09 :I: 1.51, 174.25 :1: 1.63 ppm for the [3
strand conformation. The measured chemical shifts of Gly-1, Ala-5 Phe-9 and
Gly-13 have better consistency with the B strand conformation than with the
helical conformation and the peaks for Ala-5 and Phe-9 have relative narrow
linewidths (2 ppm) suggesting that at least the region from Ala-5 to Phe-9 has a
well defined B strand structure. The peak of Gly-2O is broad (~ 5 ppm linewidth)
and agrees better helical conformation.

The 13CO—a‘P REDOR experiments were carried out for IFP samples at
pH 5.0 with 13co labels at Gly-1, Ala-5 or Gly-20. The intensity of the 8, spectra
of Gly-1 and Gly-20 are greatly reduced (> 60%) compared to the corresponding
80 spectra, which suggests a close contact between Gly-1 and Gly-20 13C0s with
the phosphorus. The So and 81 spectra of Ala-5 have about equal intensity and
indicate that Ala-5 is not close to the lipid phosphate headgroups. As displayed in
Figure A4a and b, the experimental dephasing curves of Gly-1 and Gly-20 with
fast build-up provide further evidence for the close contact between the 13COs of
these two residues and the phosphate headgroups. The flat dephasing curve of
Ala-5 (cf. Figure A4c) supports a long distance between the Ala-5 13CO and the

lipid phosphate headgroups.

192

 

Table A3. Peak 13CO chemical shifts in ppm for IFP samples that contained cholesterol at pH 5.0
a

 

Gly-1 Ala-5 Phe-9 Gly-13 Gly-20

 

170.0 175.2 172.0 173.3 175.2

 

a Typical uncertainties in peak shifts are :l: 0.4 ppm as determined from the measurements on
samples that contained peptide with the same labeled residue but different membranes.

 

193

(a) POPC/POPG (p) POPC/POPG

pH 5.0 0 pH 7.4 1

(c) DTPC/DTPG (d) DTPC/DTPG/CHOL
pH 4.0 pH 5.0

(e) DTPC/DTPG/5-19F-DPPC (f) DTPC/DTPG/16-19F-DPPC
pH 5.0 pH 7.4

”A

(9) DTPC/DTPG/CHOL/5-19F-DPPC (h) POPC/POPG

p—-—-~—-—-—~—/H 7.4 " \u—w M

(i) DTPC/DTPG (j) DTPC/DTPG
(k) DTPC/DTPG (l) DTPC/DTPG/5-19F-DPPC
pH74 H74

MJL“ M
(m) DTPC/DTPG/CHOL/5-19F-DPPC 100 50 0 -50 -100

pW 31 P chemical shift (ppm)

100 50, 0 -50 -100
31 P chemical shift (ppm)

 

 

Figure A2. 31P spectra of membranes with bound IFP. The sample composition
and pH are labeled on top of each spectrum. The temperature of the gas that
flowed around each sample was (a-g) 35 °C or (h-m) 10 °C. The general lipid and
peptide ratio was lFP/PC/PG (O.8:16:4 umol). For some samples, 10 umol
cholesterol or/and 9 mol% 5-19F-DPPC was incorporated. Each spectrum was
processed with 200 Hz Gaussian line broadening and was the sum of 400-2000
scans.

194

”co-311° 32 ms

 

 

 

 

 

 

 

 

 

s0 51 S0 31
(3) (SW-1 (b) Ala-5
T I I I 1'90 170 190 170
13C Chemical Shift (ppm)
(0) Gly-20
190 170 190 170

13C Chemical Shift (ppm)

13CO-‘9F(C5) 24 ms

 

 

 

 

 

 

 

 

 

(d) Ala-5 (e) Phe-9
' l I I 190 170 190 170
130 Chemical Shift (ppm)
(0 Gly-13
190 170 190 170

13C Chemical Shift (ppm)

Figure A3. REDOR 13C So and S1 NMR spectra at long dephasing time for
membrane-associated IFP samples at pH 5.0. Each sample had the composition
of DTPC/ DTPG/CHOL (16:4:10 umol) and 0.8 pmol IFP. The samples used to
obtain spectra (d-f) contained 9 mol% 5-‘9F-DPPC lipid. Each spectrum was
processed with 200 Hz Gaussian line broadening and was the sum of (a) 22000,
(b) 25598, (c) 16650, (d) 1200, (e) 15940 and (f) 7390 scans.

195

1303113 REDOR

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

08 (a) Gly-1 (b) Ala-5 (c) Gly-20
2» /-~. ./
“A /
«504-
U)
s.
0.0- /i/'\'/
13c-191=(c5) REDOR
03 (d) Ala-5 (e) Phe-9 (f) Gly-13
% .
2104.
U)
s
(10. \/'-—8 kg Y 44 = :______..._—x
o 10 20 30 0 10 20 30 10 20 30

dephasing time‘

Figure A4. Experimental REDOR dephasing curves corresponding for IFP
samples at pH 5.0. The uncertainties are represented by the error bars and
typically :0.01-0.02. The 1300 labeled residues are labeled on top of each

spectrum. The samples used were the same as the ones used to take the
corresponding spectra in Figure A1.

196

 

 

13CO-31P 32 ms

 

 

 

S 8
la) ° ‘
1:0 1;. 1130 .70

13C Chemical Shift (ppm)

 

(b)

.0
m
L

(AS/So)eXp
O
.3;
l

.o
O
I
~\

 

 

 

0 1 0 20 30

Figure A5. (a) 13co -3‘P REDOR so and 3, NMR spectra at T = 32 ms and (b)
the experimental dephasing curve for membrane-associated lFP-G1c sample at
pH 7.4. The sample used to obtain the data had the composition of DTPC/
DTPG/CHOL (16:4:10 umol) and 0.8 umol IFP. The spectra in (a) were
processed with 200 Hz Gaussian line broadening and were the sum of ~15000

scans. The uncertainties in (b) are represented by the error bars and typically
:l:0.01-0.02.

197

The 13CO—19F(C5) REDOR spectra were taken for lFP samples at pH 5.0
with 13CO labels at Ala-5, Phe-9 or Gly-13. No obvious dephasing was observed
for any of the labeled 13COs, evidenced by the similar intensity of the So and S1
spectra at T = 24 ms and the flat dephasing curves (cf. Figure A3d-f and A4d-f).
This could be due to their long distances relative to 19F(C5) and/or the small
population of 5-19F-DPPC incorporated into the lipids s discussed in Chapter 4.

The 13CO—31P REDOR data are generally consistent between the helical
lFP and B strand lFP that the N- and C-temiini are close to the phosphate
headgroups and the middle region is relatively far away from the phosphorus.
However, further supports are required to determine whether IFP is inserted into
the membrane bilayer.

The 13CO—31P REDOR experiments was also done with an lFP sample
that contained lFP/DTPC/DTPG/CHOL (0.8:16:4:10) at pH 7.4. The chemical
shift is 170.2 ppm, consistent with B strand conformation. The So and S1 spectra
and the experimental dephasing curve are shown in Figure A5. The S1 spectrum
was reduced by ~36% compared to the So spectrum at T = 32 ms and the
dephasing curve has a slower build-up compared to corresponding curve of the
pH 5.0 sample, which suggests a longer distance between Gly-1 13C0 and the

lipid phosphate headgroups relative to the pH 5.0 sample.
3. Static NMR

Figure A6 shows the 15N chemical shift and 15N-‘H dipolar coupling

spectra of the unflipped bicelle-associated IFP-G13CM17N at pH 5.0 and pH 7.4.

198

 

Two crosspeaks were observed for the pH 5.0 sample and may correspond to
two slightly different N-H tensor orientations of Met-17. The detected chemical
shifts and dipolar couplings (cf. table A5) are comparable to the values obtained
for the residues at the lFP N-terminal helix which may be due to the similar
orientations and motions between the N- and C- tenninus. However, the data
from only one residue is not enough to make the conclusion. The measured
chemical shifts and dipolar coupling may also reflect a relative static residue
without significant motions with a very different orientation from the N-tenninus.
The measurement of flipped bicelle sample and/or other residues will enable

some further analyses.

Table A4. 15N chemical shifts (a) and N-H dipolar couplings (v) for the aligned lFP-G13CM17N
samples at pH 5.0 and pH 7.4 ‘3

 

 

pH 5.0 b pH 7.4
o(ppm) ° v(kHz) ° o(ppm) WM
108 0.5
102 0.7
103 0.8

 

:Data were obtained with IFP: DTPC ~0. 02 and 40 °C sample nitrogen gas temperature.
bTwo sets of peak were observed for the pH 5. 0 sample.

°The uncertainty in each chemical shift' IS estimated to be :1 ppm based on the uncertainties of
other similar experiments (cf. table 12).

dThe uncertainty in each dipolar coupling is estimated to be i250 Hz.

199

pH 5.0 pH 7.4

 

 

 

 

 

 

 

_ -3
(a) (b)
a -2 {E
as
.95
O “1 E-
l ‘ 8
oo . - 0 ,3
l. l .3
.. 1 13
VI
. 2 :22
200 1'50 100 200 150 100
15N chemical shift (ppm)

Figure A6. 2D 15N chemical shift and 15N-‘H dipolar coupling correlation spectra
for the unflipped bicelle-associated lFP-G13cM17N at (a) pH 5.0 and (b) pH 7.4.
The samples used to obtain the spectra had the composition of
DTPC/DMPCd54/DHPC (1:53:1217 umol) and 0.7 umol IFP. Both spectra with Pl-
WIM-z sequence and “Efree” probe on the 21.4 T spectrometer. The temperature
of the gas which flowed around each sample was 40 °C. The total signal
averaging time was ~12 hours.

200

Appendix III
Altematlve way of fitting 13C-‘”P and 1"C-“F data

Some of the 13C-‘°’1P and 13C-19F data in Chapter 4 were also fitted with a
fraction parameter, f, which reflects the maximal fractional 13C-3‘P or 13C-‘S’F
dephasing. The logical basis for the f parameter includes: (1) two lFP populations;
(2) the 13C-19F (AS/So)” may not reach 1 because of the 0.09 mol fraction of
fluorinated lipids.

In this analysis, the ,1? analysis for (AS/So)” based on Eq. 2.10 becomes:

AS cor AS Sim
T 157] ”(s—0‘”)

2 = i z
2' (d) I; (aforf

2

 

(A21)

The fraction parameter f represents the fraction of 13CO that is close enough to
the 31P or 19F nuclei to provide non-zero dephasing. This fraction parameter was
considered because (AS/So)” reaches plateau values < 1. The data analysis
was shown by contour plots in Figure A7 and A8.

These data can be explained by using the assumption that both the 13C-
31P and 13C-‘9F dephasing are originated from a single IFP membrane location at
both pHs. In this assumption, the lFP backbone 13COs can have detectable
contact with 31F at the membrane surface or 16F(C16) at the membrane bilayer
center. For residues Leu-2 and Phe-3 at pH 5.0, the best-fit f of 13C-31P data
does not reach 100%, which suggests some undetected population of IFP in the
membrane system. For these two residues, the best-fit f has very narrow

distribution which is peaked at 70%. This indicates ~70% of IFP is associated

201

 

with the membrane. The 13C-‘9F data and the 13C-31P data of Phe-9 are not
contradictory to this explanation because of their broad distributions which cover
most of the f values. The broad distribution is due to the small (AS/So)‘5qu which
can either be fitted with a high fraction factor f and small dipolar coupling or with
a low f but large dipolar coupling. Similarly, for samples at pH 7.4, the best-fit f for
the 13C-31P data of Phe-3 has a narrow distribution peaked at 35%, which
suggests ~35% of lFP is associated with the membrane at pH 7.4.

These data can also be explained by a two-membrane-Iocation-model. In
this model, some population of lFP is deeply inserted into the membrane and
some other population of IFP is closer to the membrane surface. The 13C-3‘P
dephasing is only originated from the latter population and the 13C-19F dephasing
corresponds to the deeply inserted lFP. For samples at pH 5.0, the narrow
distribution of 13C-3‘P data for Leu-2 and Phe-3 peaked at 70% reflects ~70% of
IFP is closer to the membrane surface. The population for the deeply inserted
IFP can not be derived from the 13C-19F contour plots due to their broad
distribution of f values. For samples at pH 7.4, ~35% of IFP is close to the
membrane surface because of the 35% peak value of the best-fit f for Phe-3 13C-

3‘P.

202

 

 

 

200

 

 

 

 

13’ Lou-2 13c-31P (b) Phe-3 13c-31P
150- ................
. xznfin.
100- i *2 i
: +4 3
- *6 3 O
50- +8 .
0
200
(c) Phe-913C-31P (d) Phe-313C-19F(C16)

150‘

100-

 

Dipolar coupling frequency (Hz)

 

 

\_.

 

 

 

 

 

 

 

 

   

 

 

 

1
200~
(e) lie-613C-19F c16
( ) Phe-913C-19F(C16)
150-
1004
50-
0 l I l l I I I I
0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8

Fraction parameter f

Figure A7. Contour plots for the data fittings of the 13C-31P and 13C-‘QF data for
membrane-associated IFP at pH 5.0. The range of x2 used for different region
colored regions is represented by scale bar in (a). The value of szin is: (a) 4; (b)
15; (C) 2; (d) 2; (e) 2; (f) 1-

203

 

   

 

 

 

 

 

 

 

 

 

 

 

 

200
15‘) Len-213C-31P 1“) Phe-313c-31P
150- xzming
+2 i
+4 5
100- 5 +6
’N‘
a \
C
G)
3
3
a: 0
g 200
Q.
g 1°) Phg.9 1313.315: W lie-613c-19F(c16)
E 150-
.5 1
D
100-
504
-
0| I I I I I r I I
0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8

Fraction parameter f

Figure A8. Contour plots for the data fittings of the 13C-3‘P and 13C-‘9F data for
membrane-associated IFP at pH 7.4. The range of )(2 used for different region
colored regions is represented by scale bar in (a). The value of szln is: (a) 2; (b)
2; (c) 1; (d) 1.

204

Appendix IV

FMOC protection for amino acids and IFP synthesis
This section includes the typical experimental procedures for FMOC

protection of amino acids and manual synthesis of IFPs.

1. FMOC protection of amino acids

The amino acid (4 mmol) was dissolved in 20 mL 9% (w/w) sodium
carbonate aqueous solution and the solution was cooled in ice bath for 5 min. N-
(9-Fluorenylmethoxycarbonyloxy) succinimide (Fmoc-Osu) (4.2 mmol) in 30 mL
dimethylforrnamide (DMF) was then slowly added to the amino acid solution. The
resulting mixture was stirred in the icebath and allowed to warm to room
temperature overnight. Water (30 mL) was then added and the aqueous layer
was extracted with ether (30 mL x 3). The combined water layer was acidified to
a pH of 2 with 6 M HCI solution then extracted with ethyl acetate (20 mL x 4).
The combined organic layers were washed with saturated brine (20 mL x 2) and
dried with anhydrous sodium sulfate. The organic solvent was removed under
stream of nitrogen gas and the solid was put under vacuum to remove residual

solvents.

2. Manual solid-phase synthesis of IFPs
In addition to the methods described in Chapter 2 to synthesize lFPs,
manual synthesis of IFP was also performed for the preparation of some IFP

samples.

205

(1) Gly-preloaded Wang Resin (0.05 mmol) was washed with DMF (4 mL) and
then with CHzclz (4 mL). Add another 4 mL of CHzclz and soak the resin for 2 h.
(2) After completely draining the resin, add 4 mL 20 % piperidinelDMF (v/v)
solution, shake for 5 min and drain.

(3) Add another 4 mL 20 % piperidinelDMF (v/v) solution, shake for 20 min.

(4) Activation of Fmoc-amino acid: Dissolve 0.25 mmol Fmoc-amino acid, 0.23
mmol HBTU and 0.23 mmol HOBt in 4 mL DMF and add 175 uL DIPEA to the
resulting solution.

(5) Drain the reaction vessel and wash resin with DMF (4 mL x 6) in order to
completely remove piperidine.

(6) Coupling: Add the solution in (4) to the reaction vessel and shake for 2 h for
Gly and polar amino acids (Glu, Asn, Asp) and 4 h for non-polar amino acid (Leu,
Phe, Ala, lle, Trp, Met).

(7) Drain the reaction vessel and wash resin with DMF (4 mL x 2)

(8) Add 2 mL solution containing acetic anhydride/piperidinelDMF (2:1 :3, v/v/v) to
the reaction vessel and shake for 5 min.

(9) Drain reaction vessel and wash with DMF (4 mL x 3).

(10) Repeat steps (2)-(9) for all the other residues.

(11) For the last residue, repeat steps (2)—(3) and then wash the resin with DMF
(4 mL x 4) and CH2Cl2 (4 mL x 3).

(12) Drain and put the resin under vacuum.

(13) Cleavage: Add 3 mL TFA/thioanisole/EDT/anisole (90:5:322, vIv/v/v) solution

to the resin and shake for 2.5 h.

206

(14) Filter the reaction mixture and put the filtrate under the stream of nitrogen
gas to remove TFA. Then add 50 mL cold ether (diethyl ether or methy-t-butyl
ether) to the residual solution.

(15) Centrifuge the ether/IFP mixture and collect the precipitate.

(16) Remove the residual ether from the precipitate by nitrogen gas flow and
dissolve the peptide in water. Lyophilize the IFP water solution.

(17) Purify lyophilized lFP with HPLC and indentify IFP with MS spectrometry.

Figure A9 shows an example of Mass spectrum of IFP.

2801

2739 5733

 

WW
1 000 2000 3000 4000 5000 6000 7000
m/z

Figure A9. MALDl-TOF MS spectrum for the identification of IFP. The m/z of the
major peaks is labeled in the spectrum. The molecular weight of this IFP sample
is 2739. The peak with m/z = 2801 corresponds to the IFP molecule associated
with one Na+ and one K“. The peak with m/z = 5733 corresponds to insulin
(5733), which was added to the IFP samples as an intemal reference.

207

Appendix V

Location of NMR data

Figure 13

(a): Ihome/sunyan4c/data/lFP/redor_CP/PCPG/G1CpH5/2ms So
Ihome/sunyan4c/data/lFP/redor_CP/PCPG/G1CpH7/2ms So
/home/sunyan4c/data/IFP/redor_CP/PCPGCHOL/G1 C _pH5/2ms So
lhome/sunyan4c/data/lFP/redor_CP/PCPGCHOLIG1 C _pH7/2ms So

(b): /home/sunyan4cldatallFP/redor_CP/PCPG/L2CpH5/2ms So
lhome/sunyan4c/data/lFP/redor_CP/PCPG/L2CpH7/2ms So
/home/sunyan4c/data/lFP/redor_CP/zerodegree/L2CpH5/2ms So

(0): /home/sunyan4c/data/lFP/redor_CP/PCPG/F3CpH5/2ms So
lhome/sunyan4c/data/lFP/redor_CP/PCPG/F3CpH7/2ms
/home/sunyan4c/datallFP/redor_CP/zerodegreelF3CpH5/2ms So

(d): lhome/sunyan4b/data/lFP-redor/5AC-9FN/pH5/redor-2ms-070505 So
/home/sunyan4b/data/lFP-redor/5AC-9FN/pH7/redor-2ms-043007 So
/home/sunyan4c/data/lFP/redor_CP/PCPGCHOL/ASC_pH5/2ms So
/home/sunyan4cldatallFP/redor_CF/5FDPPC/CHOL/ASCpH7/2ms So

(9): /home/sunyan4c/data/lFP/redor_CP/PCPG/l6_pH5/2ms-032109 So
lhome/sunyan4c/data/lFP/redor_CP/PCPG/l6_pH7/2ms-032109 Sn

(f): /home/sunyan4cldatallFP/ctdqbu/A7G8_lFP_pH5/PCPG/MAS_8kHz/L16-s0
/home/sunyan4c/data/lFP/A7GB-pH7/A7G8_pH7_PCPG
lhome/sunyan4c/data/lFP/ctdqbu/A7GB_IFP_pH5/PCPG_chol/MAS_8kHz/
L32-s0
[home/sunyan4c/data/lFP/A7GB-pH7/A7G8_chol_pH7

(g): lhome/sunyan4b/data/lFP-redor/9FC-1 3G NlpH5/2ms-sum So
/home/sunyan4b/datallFP-redor/9FC-13GN/pH7/2ms So
/home/sunyan4c/data/lFP/redor_CF/redor_CF_CS/PCPGCHOUFQCpH5/
2ms
/home/sunyan4c/data/lFP/redor_CF/redor_CF_C5/PCPGCHOL/FQCpH7/
1 6ms

(h): Ihome/sunyan4b/data/lFP-redor/13GC-1 7MN/pH5/redor-2ms-050907 So
/home/sunyan4b/data/lFP-redor/13GC-1 7MN/pH7/redor-2ms-O70507 So
lhome/sunyan4c/data/lFP/redor_CF/redor_CF_C5/PCPGCHOLJG1 3CpH5/
2ms '

208

(i): /home/sunyan4c/data/lFP/redor_CP/PCPG/G16CpH5/2ms So
/home/sunyan4c/data/lFP/redor_CP/PCPG/G16CpH7/2ms So

(j): /homelsunyan4c/datallFP/redor_CP/PCPG/GZOCpH5/2ms So
/home/sunyan4chata/lFPlredor_CP/PCPG/G20CpH7/2ms So
/home/sunyan4c/data/IFP/redor_CP/PCPGCHOUGZOC_pH5

Figure 14

(a): lhome/sunyan4c/data/IFP/ctdqbu/A7G8_lFP_pH5/PCPG/MAS_8kHz/L32-s0

(b): /homelsunyan4c/data/lFP/ctdqbulA7G8_lFP_pH5/PCPG_chol/MAS_8kHzI
L32-s0

Figure 15

(a, c): Ihomelsunyan4b/data/lFP-redor/5AC-9FN/pH5/

(b, d): /home/sunyan4b/data/lFP-redor/5AC-9FN/pH7/redor-16ms-043007
(e, g): /home/sunyan4b/datallFP-redor/9FC-13GN/pH5/sum spectra
(f, h): /home/sunyan4bldata/lFP-redor/9FC-13GN/pH7/

(i, k): /home/sunyan4b/data/lFP-redor/13GC-17MN/pH5/

(j, l): Ihome/sunyan4b/data/lFP-redor/13GC-17MN/pH7/

Figure 16

(a): /home/sunyan4c/data/lFP/ctdqbuN1 1A12-pH5/L144

(b): /home/sunyan4cldatallFP/ctdqbu/A12G13-pH5/032508/L144
Figure 17

(a): /home/sunyan4c/datall FP/mutants-CNREDORN1 1A12/16ms
(b): Ihome/sunyan4c/data/lFP/mutants-CNREDOR/A1261 3/16ms
Figure 18

(a): lhome/sunyan4c/data/lFP/pdsd/U-l10E1 1 _pH5/minus50/sum

(b): /homelsunyan4cldata/IFP/pdsd/U-N12G13-pH5/minus_SOC/pdsd_N12G13
_pH5

209

Figure 20

(a): lhome/sunyan4cldata/lFP/pdsd/U_l10E1 1 _pH4/sum_minus50
(b): the same as figure 18(a)

(c): lhomelsunyan4c/data/lFP/pdsd/U-l10E11_pH7/pdsd-minus50

Figure 21
(a): the same as figure 20(6)
(b): /home/sunyan4c/datallFP/pdsd/U-l10E1 1_pH7/pdsd-I1OE1 1 ph7-081308_OC

(c): lhome/sunyan4c/data/lFP/pdsd/U-l10E1 1_pH5/zerodegree/pdsd-081208-
I10E1 1

(d): the same as figure 20(a)

(e): lhome/sunyan4c/data/lFP/pdsd/U-N12G13-pH7/minus5OC/pdsd_U-N12G13-
pH7-minus50

(f): /home/sunyan4c/datallFP/pdsd/U-N12G13-pH7/pdsd_U-N12G13-pH7-0C

(g): /home/sunyan4c/data/lFP/pdsd/U-N12G13-pH5/POPC_POPG_OC/pdsd_
N1ZG1 3-0C_pH5

Figure 22

(a): lhome/sunyan4c/data/lFP/P31IP31_spectra_membraneleTPCPG_pH5
(b): lhome/sunyan4cldata/IFP/P31IP31_spectra_membranesIDTPCPG_pH7
(c): /home/sunyan4c/data/lFP/P31/P31_spectra_membranes/PCPG_noIFP
Figure 24

(a): lhome/sunyan4c/data/lFP/redor_CP/PCPG/G1CpH5/32ms

(b): /home/sunyan4c/data/lFP/redor_CP/PCPG/LZCpH5/32ms

(c): /home/sunyan4cldata/lFP/redor_CP/PCPG/F3CpH5/32ms

(d): lhome/sunyan4c/data/lFP/redor_CP/PCPG/A5CpH5/32ms

210

(e): /home/sunyan4c/data/lFP/redor_CP/PCPG/l6_pH5/32ms

(f): /home/sunyan4c/data/lFP/redor_CP/PCPG/FQCpH5/32ms

(g): /home/sunyan4c/data/lFP/redor_CP/PCPG/G13CpH5/32ms

(h): lhome/sunyan4c/data/lFP/redor_CP/PCPG/G16CpH5/32ms

(i): /home/sunyan4c/data/lFP/redor_CP/PCPG/G20CpH5/32ms

(j): /home/sunyan4cldatallFP/redor_CF/redor_CF_05/PCPG/F3pH5/24ms

(k): /home/sunyan4c/datallFP/redor_CF/redor_CF_C5/PCPG/A5C-pH5/24ms
(I): lhome/sunyan4c/data/lFP/redor_CF/redor_CF_C5/PCPG/l6_pH5/24ms
(m): /home/sunyan4c/data/lFP/redor_CF/redor_CF_C5/PCPG/FQC-pH5/24ms
(n): lhome/sunyan4c/data/lFP/redor_CF/redor_CF_C5/PCPG/G13-pH5/24ms
(o): /home/sunyan4c/data/lFP/redor_CF/redor_CF_C1 6/PCPG/G1 CpH5/24ms
(p): /home/sunyan4c/data/lFP/redor_CF/redor_CF_C16/PCPG/L2CpH5/24ms
(q): lhome/sunyan4c/datallFP/redor_CF/redor_CF_C16/PCPG/F3_pH5/24ms
(r): /home/sunyan4c/datallFP/redor_CF/redor_CF_C16/PCPG/l6 _pH5/24ms
(s): /home/sunyan4c/datallFP/redor_CF/redor_CF_C16/PCPG/F9_pH5/24ms
(t): lhome/sunyan4c/data/lFP/redor_CF/redor_CF_C1 6/PCPG/G1 3 _pH5/24ms
(u): /home/sunyan4c/datallFP/redor_CF/redor_CF_C16/PCPGIG16CpH5/24ms
F lgure 28

(a): lhome/sunyan4c/data/lFP/redor_CP/PCPG/G1CpH7/32ms

(b): /home/sunyan4c/data/lFP/redor_CP/PCPG/LZCpH7/32ms

(c): /home/sunyan4c/datallFP/redor_CP/PCPGIF3CpH7/32ms

(d): lhome/sunyan4c/data/lFP/redor_CP/PCPG/ASCpH7/32ms

(e): /home/sunyan4c/data/lFP/redor_CP/PCPG/l6_pH7/32ms

211

(f): /home/sunyan4c/data/lFP/redor_CP/PCPGIF9CpH7/32ms
(g): /home/sunyan4c/datallFP/redor_CP/PCPG/G13CpH7/32ms
(h): /home/sunyan4c/datallFP/redor_CP/PCPG/G160pH7/32ms
(i): /home/sunyan4c/data/lFP/redor_CP/PCPG/GZOCpH7/32ms

(j): lhome/sunyan4c/data/lFP/redor_CF/redor_CF_C5/PCPGIF3_pH7/24ms
_031809

(k): Ihome/sunyan4c/data/lFP/redor__CF/redor_CF_C5/PCPGIA5CpH7/24ms
(l): lhome/sunyan4c/data/lFP/redor_CF/redor_CF_C5/PCPG/l6_pH7/24ms
(m): /home/sunyan4c/data/lFP/redor_CF/redor_CF_C5/PCPG/F9_pH7/24ms
(n): /home/sunyan4c/data/lFP/redor_CF/redor_CF_C5/PCPG/G13-pH7/24ms
(o): lhome/sunyan4c/datallFP/redor__CF/redor_CF_C1 6/PCPG/G1 CpH7/24ms
(p): lhomelsunyan4cldata/lFP/redor_CF/redor_CF_C16/PCPG/L2CpH7/24ms
(q): /home/sunyan4cldatallFP/redor_CF/redor_CF_C16/PCPG/F3_pH7/24ms
(r): /homelsunyan4cldata/lFP/redor_CF/redor_CF_C16/PCPG/l6_pH7/24ms
(s): /home/sunyan4c/datallFP/redor__CF/redor_CF_C16/PCPG/F9_pH7/24ms
(t): lhome/sunyan4c/data/lFP/redor_CF/redor_CF_C1 6/PCPG/G1 3 _pH7/24ms
(u): lhome/sunyan4c/data/lFP/redor_CF/redor_CF_C1 6/PCPG/G16_pH7/24ms
F lgure 34

(b): lhome/sunyan4c/data/bicelle/BicelIe-C13/purebicelle-1Ibicelle-nopeptide-3-
1 02605

(c): /home/sunyan4c/datalbicelle/Bicelle-C13/purebicelle-1lflipped/bicelle-
nopeptide-fiipped

(d): /homelsunyan4c/data/bicelle/lFP/A7-N1 5/pH7/unflipped/quecho-1 -1 02706
(e): lhome/sunyan4c/data/bicelle/lFP/A7-N1 5/pH7.4/flipped/quecho-1 -1 10906

212

(f): lhome/sunyan4c/data/bicelle/lFPluniform-N1 5/pH5/unflipped/quecho-1 -
101 1 06

(g): /home/sunyan4c/data/bicellellFP/uniform-N1 5/pH5/flipped/quecho-1-101 106
(h): lhome/sunyan4c/datalbicelle/lFP/Oct08/A5N_pH5_unflip/p31

(i): lhome/sunyan4c/datalbicelle/lFP/Oct08/l10N_pH5_flip/P31

(j): /home/sunyan4c/data/lFP/P31/P31_spectra_membranes/ DTPCPG_pH5

(k): lsunyan4c/data/bicellel H FP/uniform-N1 5-bicelle/powderlquecho-1-100306
Figure 35

(a): /home/sunyan4c/data/bicelle/lFP/uniform-N15-2/powder/ph5-4/sum1

(b): Ihome/sunyan4c/data/bicelle/lFP/uniform-N15-2/powder/pH7-3Isum

(c): lhome/sunyan4c/data/bicelle/IFPluniform-N15-2/drypeptide/ drypep-
cp_ramp-1 10406

(d): lhome/sunyan4c/data/bicelle/lFP/unifonn-N1 5-2/d rypeptidelMAS/cp_ramp-1 -
01 0207

(e): lhome/sunyan4c/data/bicelle/l FP/unifonn-N1 5-2/pH5/unflipped/cp_ramp-1 -
1[11)czir6noe6/sunyan4c/data/bicelle/lFP/uniforrn-N1 5-2/pH5/flipped/cp_ramp-2-
1/l1igr2noe6/sunyan4c/data/bicelle/lFP/unifonn-N1 5-2/pH7/unflipped/cp_ramp-1 -
:IEEEnZLZIsunyan4c/data/bicelle/lFP/uniforrn-N1 5-2/pH7/flipped/cp_ramp-2-

(f): /home/sunyan4c/datalbicelIe/lFP/F3-N15/pH5.0/unflipped/cp_ramp-2-1 12806
/home/sunyan4c/data/bicelle/lFP/F3-N1 5/pH5.0/flipped/cp_ramp-1-120306
lhome/sunyan4c/data/bicelle/lFP/F3-N1 5/pH7.4/unflipped/cp_ramp-2-1 12706
lhome/sunyan4c/data/bicelle/lFP/F3-N1 5/pH7.4/flippedlcp_ramp-2-120106

(g): /home/sunyan4c/datalbicellellFP/Oct08/G4N_pH5_unflip/cp-110308
/home/sunyan4c/data/bicelle/lFP/OctO8/G4N_pH5_flip/cp_ramp_1 12308
Ihome/sunyan4c/data/bicelle/lFP/Oct08/G4N_pH7_unflip/cp-1 10308
lhome/sunyan4c/data/bicelle/lFP/Oct08/G4N_pH7_flip/cp_ramp-1 10508

(h): /home/sunyan4c/data/bicelle/lFP/Oct08/A5N_pH5_unflip/sum
/home/sunyan4c/data/bicelle/lFP/Jan09/A5_pH5_flip_2

213

lhome/sunyan4c/data/bicelle/lFP/Oct08/A5N_pH7_unflip/cp-122308

(i): lhome/sunyan4c/data/bicelle/lFP/Oct08/I6N_pH5_unflip/cp-1 10908
lhome/sunyan4c/data/bicelle/lFP/Oct08ll6N_pH5_flip/cp_1 12208
lhome/sunyan4c/data/bicelle/lFP/Oct08/l6N_pH7_unflip/cp_ramp

(j): lhomelsunyan4c/data/bicelle/lFP/A7-N1 5/pH5/unflipped/cp_ramp—3-102306
lhome/sunyan4c/data/bicelle/lFP/A7-N1 5/pH5/flipped/cp_ramp-2-102906
lhome/sunyan4c/data/bicelle/lFP/A7-N1 5/pH7/unflipped/cp_ramp-2-102505
lhome/sunyan4c/data/bicellellFP/A7-N1 5/pH7/flipped/sum

(k): lhome/sunyan4c/data/bicelle/lFP/Oct08/GBN_pH5_unflip/cp-120408
/homelsunyan4c/data/bicelle/lFP/Oct08lG8N_pH5_flip/sum

(l): lhome/sunyan4c/data/bicelle/lFP/Oct08/F9N_pH5_unflip/sum

(m): /home/sunyan4c/data/bicelle/lFP/Oct08/l10N_pH5_unflip/cp-120808
lhome/sunyan4c/data/bicelle/lFP/Oct08/l10N_pH5_flip/sum

Figure 36

(a): /home/sunyan4c/data/bicellellFP/Oct08/uni_pH5_unflip/piwimz_122508
lhome/sunyan4c/data/bicelle/lFP/Oct08/uni_pH5_flip/piwimz-1 10308

(b): /home/sunyan4c/data/bicelIe/lFP/Oct08/uni_pH7_unflip/piwimz-103108

(c): /home/sunyan4c/datalbicelle/900M_data/lFP_F3A7_pH5_unflip_2D
lhome/sunyan4c/data/bicelle/900M_data/lFP_F3A7_pH5_flip_2D

(d): lhome/sunyan4c/data/bicelle/900M_data/lFP_F3A7_pH7_unflip_2D
lhome/sunyan4cldata/bicelle/900M_data/lFP_F3A7_pH7_flip_2D

(e): lhome/sunyan4cldata/bicelle/lFP/Oct08/G4N_pH5_unflip/piwimz-102508
/home/sunyan4c/data/bicelIe/lFP/Oct08/G4N_pH5__flip/piwimz-1 10808-1

(f): lhome/sunyan4c/data/bicelle/lFP/Oct08/G4N_pH7_unflip/piwimz-102908
/home/sunyan4c/data/bicelle/lFP/OctO8/G4N_pH7_flip/piwimz-1 10508

(g): /home/sunyan4c/data/bicelle/lFP/Jan09/A5_pH5_flip_PIW|MZ
(h): lhomelsunyan4c/data/bicelle/lFP/Oct08/l6N_pH5_flip/piwimz—1 12208

(i): /home/sunyan4c/data/bicelle/lFP/Oct08/G8N__pH5_unflip/piwimz-1 20408
lhome/sunyan4c/data/bicelle/lFP/Oct08/G8N_pH5_flip/piwimz-1 21 908-2

(j): /home/sunyan4c/data/bicelle/lFPIOct08/l10N_pH5_unflip/piwimz-120808

214

lhome/sunyan4c/data/bicelle/IFP/Oct08/l10N_pH5_flip/piwimz_1 21 708
Figure 37
(a): /home/sunyan4c/data/bicelle/lFP/Oct08/uni_pH5_unflip/cp_ramp_al_512
(b): lhome/sunyan4c/data/bicellel900M_data/lFP_uni_pH5_unflip_1D
(c): /home/sunyan4c/datalbicelle/lFP/Oct08/F3A7_flip _pH7/piwimz_1 11308
(d): the same as Figure 36(d)
Figure 38

(a): lhome/sunyan4c/data/bicelle/900M_data/lFPG1S_F3A7_pH5_unflip_2D
lhome/sunyan4c/data/bicelle/900M_data/lFPG1S_F3A7_pH5_flip_2D

(b): lhome/sunyan4cldata/bicelle/900M_data/lFPG1S_F3A7_pH7_unflip_2D
lhome/sunyan4c/data/bicelle/900M_datallFPG1S_F3A7_pH7_flip_2D

(c): [home/sunyan4c/data/bicelle/900M_data/lFPG1V_F3A7_pH5_unflip_2D
lhome/sunyan4cldata/bicelle/900M_data/lFPG1V_F3A7_pH5_flip_ZD

(d): /home/sunyan4c/datalbicelle/900M_data/lFPG1V__F3A7_pH7_unflip_2D
/home/sunyan4c/data/bicelle/900M_data/lFPG1V_F3A7_pH7_flip__2D

Figure 39

(a): lhome/sunyan4cldata/bicelle/HFP/unifonn-N1 5-bicelle/u nflipped/quecho-1 -
1 00306

(b): lhome/sunyan4c/data/bicelle/HFP/uniforrn-N15-bicelle/flipped/quecho-2-
1 00306

(c): Ihome/sunyan4c/datalbicelle/HFP/uniform-N15-
bicelle/pH7/unflipped/cp_ramp-1-101 706

(d): lhome/sunyan4cldata/bicelle/HFP/uniform-N15-bicelle/flipped/cp_ramp-1-
100406

(e): /home/sunyan4c/data/bicelle/HFP/uniforTn-N15-dimer/unflipped/cp__ramp-1
1 10806

(f): lhome/sunyan4c/data/bicelle/HFP/uniform-N15-dimer/flipped/cp_ramp—2-
1 1 1306

215

(g): lhomelsunyan4cldata/bicelle/HFP/FP3K-SGAL-N15/bicelle/cp_ramp-2-
070806/

(h): lhomelsunyan4c/datalbicelle/HFP/FP3K-14AAG-N1 5/unflipped/cp_ramp-4-
082406

(i): lhomelsunyan4c/data/bicelle/HFP/FPdm-A6-N1 5/cp_ramp-3-091506

(j): lhomelsunyan4c/data/bicelle/HFP/uniform-N15-bicelle/powder/cp_ramp-1-
100706

Figure A2

(a): lhomelsunyan4cldata/lFP/P31/P31_spectra_membranes/POPC_PG_pH5
(b): lhomelsunyan4c/data/lFP/P31/P31_spectra_membranes/POPCPG_pH7
(c): /home/sunyan4c/data/lFP/P31IP31_spectra_membranes/DTPCPG_pH4

(d): lhomelsunyan4c/data/IFP/P31IP31_spectra_membrahes/DTPCPG_CHOL
_pH5

(e): lhomelsunyan4c/data/IFP/P31IP31_spectra_membranes/PCPG_pH5_
5F DPPC

(f): lhomelsunyan4c/data/lFP/P31/P31_spectra_membranes/DTPC_PG_PH7_
1 6FDPPC

(g): lhomelsunyan4c/datallFP/P31/P31_spectra_membranes/DTPCPGCHOL_
5FDPPC_pH7

(h): lhomelsunyan4c/data/lFP/P31IP31_spectra_membranes/POPCPG_pH7_
1 Odegree

(i): lhomelsunyan4c/data/lFP/P31IP31_spectra_membranes/DTPCPG_pH4_
1 Odegree

(j): lhomelsunyan4c/data/lFP/P31IP31_spectra_membranes/DTPCDTPG_pH5
_1 0C

(k): lhomelsunyan4c/data/lFP/P31/P31_spectra_membranes/DTPCPG_pH7
_1 CC

(I): lhomelsunyan4c/data/lFP/P31IP31_spectra_membranes/DTPCPG_PH7
_5FDPPC_1 0C

216

(m): lhomelsunyan4c/data/IFP/P31IP31_spectra_membranes/DTPCPGCHOL
_5FDPPC_pH7__10degree

Figure A3

(a): lhomelsunyan4c/data/lFP/redor_CP/PCPGCHOL/G1C_pH5

(b): lhomelsunyan4c/data/lFP/redor_CP/PCPGCHOL/A5C_pH5

(c): lhomelsunyan4c/data/lFP/redor_CP/PCPGCHOLIG20C_pH5

(d): lhomelsunyan4c/data/lFP/redor_CF/redor_CF_C5lPCPGCHOL/A5CpH5
(e): lhomelsunyan4c/data/lFP/redor_CF/redor_CF_CS/PCPGCHOLIF9CpH5
(f): lhomelsunyan4c/data/IFP/redor_CF/redor_CF_C5/PCPGCHOL/G13CpH5
Figure A4

lhomelsunyan4c/datallFP/redor_CP/PCPGCHOLIG1C_pH7

Figure A6

(a): lhomelsunyan4c/data/bicelle/900M_data/lFP_M17N_pH5_unflip_ZD

(b): lhomelsunyan4c/data/bicelle/900M_datallFP_M17N_pH7_unflip_2D

217