4.
Wu . x, .
Surf. Fummmwwwgwm‘.

3. has... n.
eth : J. . ,9
..

 

 

 

 

! .15” b
i 3.2... .
.. s. .1
{.31).

4

a.
z
2 -
5236 h
:95: a: .6. l. .

I I 1.: 1er .1
5 $3. .2...

a. A

mu...‘ .. “ﬂaw

r. :t,

it, 4...... . 3.1.5 .J 0+5?

 

 

 

 

 

.1313. away?
3: ..H.;...L..

K
'

 

THESIS

2
3 mm LIBRARY
5% é a I ’7 49 Michigan State

University

This is to certify that the
dissertation entitled

SOLID-STATE NUCLEAR MAGNETIC RESONANCE
STRUCTURAL STUDIES OF THE HIV-1 FUSION
PEPTIDE IN THE MEMBRANE ENVIRONMENT

presented by

Jun Yang

has been accepted towards fulfillment
of the requirements for the

Ph.D. degree in Chemistry

 

 

’8ch IO w; L“

Major Professor’s Signature
7// 21 o 3

Date

 

MSU is an Affirmative Action/Equal Opportunity Institution

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.

 

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

6/01 cJClRC/DateDuepGS—p. 15

 

SOLID-STATE NUCLEAR MAGNETIC RESONANCE STRUCTURAL STUDIES
OF THE HIV-1 FUSION PEPTIDE IN THE MEMBRANE ENVIRONMENT

By

Jun Yang

A DISSERTATION

Submitted to
Michigan State University
In partial fulfillment of the requirements
For the degree of

DOCTOR OF PHILOSOPHY
Department of Chemistry

2003

ABSTRACT

SOLID-STATE NUCLEAR MAGNETIC RESONANCE STRUCTURAL STUDIES
OF THE HIV -1 FUSION PEPTIDE IN THE MEMBRANE ENVIRONMENT

By

Jun Yang

Solid-state nuclear magnetic resonance (NMR) spectroscopy was applied to
understand the structures of a synthetic peptide representing the 23-residue N-terminal
fusion peptide domain of the HIV-1 gp4l envelope glycoprotein in the membrane
environment. Numerous biochemical and biophysical studies have shown that the free
peptide by itself can cause the fusion of liposome without the rest of the glycoprotein.
Hence, the information provided by the free peptide should be applicable to learn the
mechanism of HIV-1 viral fusion with the host cells.

The wild peptide and a series of its variant peptides were synthesized and
examined for their physical properties and their biological activities toward large
unilamellar vesicles that were made with different lipid compositions. These
measurements provided evidence that the probed peptide structures were indeed
biologically relevant.

Peptides were l3C carbonyl carbon labeled and/or 15 N amide nitrogen labeled at
different residues according to the requirement of specific solid-state NMR experiments.
Sample preparation studies provided the protocol so that only the contribution from
membrane-bound fusion peptide is observed in the solid-state NMR measurement. The

initial linewidth measurements of singly l3C carbonyl labeled peptides showed that a

signiﬁcant population of the membrane-bound peptide is well structured in its N-terminal
and central regions while the C-terminus has more disorder. There was some dependence
of linewidth on lipid composition, with narrower linewidth and hence greater structural
order observed for a lipid composition comparable to that found in the virus-infected host
cells. In the ordered N-terminal and central regions of the peptide, the 13C carbonyl
chemical shifts are consistent with a B strand membrane-bound conformation. Additional
evidence for a B strand membrane-bound conformation was provided by analysis of two-
dimensional rotor-synchronized magic angle spinning NMR spectra of doubly 13C
carbonyl labeled peptides.

The B strand structure was further investigated in terms of its oligomeric state by
applying solid-state NMR rotational-echo double-resonance spectroscopy (REDOR).
Comparison of REDOR spectra on samples that were labeled at different residue
positions suggests that there are both parallel and anti-parallel arrangements of peptide
strands in the oligomers. In the parallel arrangement, inter-peptide hydrogen bonding
decreases toward the C-terminus while in the anti-parallel arrangement, hydrogen bonds
are observed along the entire length of residues that was probed (Gly-S to Gly—16). For
the parallel arrangement, these observations are consistent with the model in which the
apolar N-terminal and central regions of the peptides penetrate into the membrane and
hydrogen bond with one another while the polar C-terminus of the peptide is outside the
membrane and hydrogen bonds with water. These measurements show that at least at the

end state of fusion, the peptide can adopt an oligomeric B strand structure.

Dedicated to my family, especially my lovely wife.

iv

ACKNOWLEDGMENTS

This work would not have been possible if not for the help and support of many
people. First and foremost I would like to thank my advisor, Dr. David P. Weliky.
Through his guidance and example I have learned much of what it means to be a
scientist, teacher and mentor.

I also want to thank my committee members, Drs. Gregory. L. Baker, John. L.
McCracken and James Geiger for being supportive and providing me helpful critiques of
my dissertation. I would like to thank Dr. Robert E. Maleczka, Jr. for lending the advice
and support throughout my graduate career and Drs. Babak Borhan, Honggao Yan and
David L. DeWitt for giving me the opportunities to run experiments in their labs.

The Max T. Rogers NMR Facility and the MSU Mass Spectrometry Facility all
provided me the indispensable technical support for my work. I sincerely thank the
faculties and staffs there for their assistance.

I would like to thank all the past and present members of the Weliky group for
their friendship and partnership in science. Special thanks to Dr. Charles Gabrys, for his
patience and selfless help during my research, and Paul Parkanzky, Christian Canlas,
Rong Yang, Chris Wasniewski, Michele Bodner, Zhaoxiong Zheng (Norm), Vamshi
Cotla and Robert Boes for sharing the expertise of science and happiness of life with me
and making the years at MSU such an enjoyable experience to me. I wish all of you good
luck in your NMR adventure!

I cannot thank life enough for bringing me the most wonderful friends! Ouyang,

Yinghui, Yang and Wei, thank you for sharing with me such a loyal friendship. Janathon

and Wendy for your selﬂess help over the years. And for all my dear friends, thank you
all for creating a positive and uplifting friend circle and being my cheerleaders whenever
need. My life is so blessed with all of you!

As always, the past five years’ journey will not be successful without the support
from my family, my mom, dad, little sister, and grandparents. I can never tell you how
much your love means to me. With you, my faith in the virtues of human nature will
never die. Finally, my deep thanks go to Zhengyi for being a considerate and supportive
wife. There is not an easy way to tell you my appreciation of the love, encouragement
and trust that you have translated in a unique way. I am so glad that we will take the

many new journeys ahead together.

vi

TABLE OF CONTENTS

 

 

 

 

 

Page

LIST OF TABLES ............................................................................................................. ix
LIST OF FIGURES ............................................................................................................. x
LIST OF SYMBOLS AND ABBREVIATIONS ............................................................. xv
CHAPTER I. INTRODCUTION ................................................................................... 1
BACKGROUND ..................................................................................... 2

REFERENCE .......................................................................................... 8

CHAPTER H. PHYSICAL AND BIOLOGICAL PROPERTIES OF THE FUSION
PEPTIDE ............................................................................................... 12
BACKGROUND ................................................................................... 13

MATERIALS AND METHODS .......................................................... 13

RESULTS AND DISCUSSION ........................................................... 18

Peptide Aggregation in Aqueous Solution ...................................... 18

Peptide Overall Structure in Solution .............................................. 26

Peptide-Induced Fusion of LM Vesicles ......................................... 28

CONCLUSION ..................................................................................... 32

REFERENCE ........................................................................................ 34

CHAPTER HI. CONFORMATIONAL STUDIES OF THE MEMBRANE-
ASSOCIATED FUSION PEPTIDE BY SOLID-STATE ONE-
DIMENTIONAL NMR MEASUREMENTS ....................................... 36
BACKGROUND ................................................................................... 37

METHODS ............................................................................................ 38

RESULTS AND DISCUSSION ........................................................... 42

Peptide binding to Lipids ................................................................ 42

Dependence of 1D NMR Spectra on Sample Conditions ............... 44

1D NMR Spectral Scanning down the FP23 Backbone .................. 51

Correlation of Carbonyl Chemical Shifts with Secondary Structure 53

Structure Similarity of Membrane—Associated FP, FPK3, FPW, and

FPK3W ............................................................................................ 54

CONCLUSION ..................................................................................... 56

REFERENCE ........................................................................................ 57

CHAPTER IV. CONFORMATIONAL STUDIES OF THE MEMBRANE-

 

ASSOCIATED FUSION PEPTIDE BY SOLID-STATE TWO-
DIMENTIONAL EXCHANGE NMR MEASUREMENTS ................ 58

vii

BACKGROUND ................................................................................... 59

 

 

 

 

 

 

 

METHODS ............................................................................................ 61
RESULTS AND DISCUSSION ........................................................... 63
Model Studies .................................................................................. 63

2D Exchange Measurements for Reﬁned Secondary Structure
Determination of Membrane-Associated Fusion Peptide ............... 66
CONCLUSION ..................................................................................... 70
REFERENCE ........................................................................................ 71
CHAPTER V. DESIGN AND OPTIMIZATION OF SOLID-STATE ROTATIONAL—
ECHO DOUBLE-RESON AN CE (REDOR) MEASUREMENTS TO
PROBE DISTANCES BETWEEN CARBONYL CARBONS AND
AMIDE NITROGENS .......................................................................... 73
BACKGROUND ................................................................................... 74
MATERIALS AND METHODS .......................................................... 75
RESULTS AND DISCUSSION ........................................................... 78
Quality of REDOR 7t Pulses ............................................................ 78

IH Decoupling Sequence and Strength in REDOR Experiments 85
Selection of REDOR Sequences ..................................................... 91
CONCLUSION ................................................................................... 104
REFERENCE ...................................................................................... 106
CHAPTER VI. OL_IGOMERIZATION STUDIES OF THE MEMBRANE-
ASSOCIATED FUSION PEPTIDE BY REDOR MEASUREMENTS:
PARALﬂ-EL VS. ANTI-PARALLEL STRAND ARRANGEMENTS 108
BACKGROUND ................................................................................. 109
MATERIALS AND METHODS ........................................................ 111
RESULTS ............................................................................................ 1 13
REDOR Simulations of Multiple 15N Coupled to 13c ................... 113
REDOR Control Experiments ....................................................... 117
REDOR Experiments to Probe Inter-Peptide Hydrogen Bonding 121
DISCUSSION AND CONCLUSION ................................................. 128
REFERENCE ...................................................................................... l3 1
CHAPTER VI. SUMMARY AND FUTURE DIRECTIONS OF THE RESEARCH 134

 

REFERENCE ...................................................................................... 140

viii

LIST OF TABLES

 

 

 

 

Page
CHAPTER II
Table 1. FP Peptide Aggregation in Different Aqueous Buffer Solutions ....................... 21
CHAPTER 111
Table 2. Fusion Peptides Binding to Membranes ............................................................ 43
Table 3. Structural Correlations from Carbonyl Chemical Shift Analysis ...................... 53
CHAPTER IV
Table 4. Structural Correlations from 2D Exchange Measurements ............................... 69
CHAPTER V
Table 5. Pulse Length at Different Positions in the 4 mm Rotor ..................................... 80
Table 6. Pulse Length at Different Positions in the 6 mm Rotor ...................................... 81
Table 7. Pulse Proﬁle Adjustment and Corresponding REDOR Dephasing Ratio .......... 84

Table 8. Dependence of REDOR AS/So on the ‘H Decoupling/”N Tl: Pulse Field Ratio 86

 

CHAPTER VI
Table 9. REDOR Control Experiments .......................................................................... 119
Table 10. REDOR Measurements to Probe Inter-Peptide Hydrogen Bonds ................. 123

ix

LIST OF FIGURES

Page
CHAPTER I
Figure 1. Model (left) and freeze fracture electron microscopy (right) of HIV virion (a)
binding to, (b) fusion with, (c-d) forming large pore, and infecting host cell .................... 3
Figure 2. Model of HIV fusion. FP: Fusion Peptide. Time sequence: left to right ............ 3
CHAPTER H

 

Figure 3. Solution amide lH NMR spectra of (a) PP at pH 6.7, (b) FP at pH 3.7, (c) FPK3
at pH 6.4, and (d) FPK3 at pH 3.7 .................................................................................... 24

Figure 4. Sedimentation equilibrium experiments of (a) 80 11M FPK3W and (b) 80 11M
FPW in buffer. In (a), the main panel shows the absorbance at 280 nm as a function of the
centrifugal radius after reaching the equilibrium in 20 hours at 52,000 rpm. The best ﬁt to
the model for a single species was obtained with a molecular weight of ~2630 (shown as
a solid line through the experimental points). The upper panel shows the residuals
between the data and the fit. In (b), the logarithm of 280 nm absorbance is plotted as a
function of (radius)2 at 5,000 rpm. The data are represented as dark circles and the dark
dotted and light-dashed lines represent the expected results for FPW lOO-mers and
monomers, respectively .................................................................................................... 25

Figure 5. CD spectra of four 20 11M fusion peptide constructs in unbuffed water: FF (1),
FPK3 (H), FPW (1H), and FPK3W (IV); and in 5 mM HEPES buffer: FP (I’), FPK3 (H’),
FPW (IH’), and FPK3W (IV’) .......................................................................................... 27

Figure 6. CD spectra of 20 11M FP peptide in unbuffered water (I), and a series of
acidified water solutions at pH 5.0 (H), pH 4.1 (HI), and pH 3.1 (IV). The pH is adjusted
by adding a small amount of 30 mM HCl solution .......................................................... 27

Figure 7. FP-induced lipid mixing (top) and FP-promoted aqueous contents leakage
(bottom) as observed by ﬂuorescence assays at 37 °C. For the lipid-mixing assay,
1501tM LM-3 lipid was extruded into 100nm diameter vesicles. Lipid mixing was
induced with (I) 7.5 HM and (H) 1.5 uM FP. For the aqueous contents leakage assay,
80 11M LM-3 lipid was extruded into 100 nm diameter vesicles loaded with ANT S/DPX.
Leakage was promoted by FP at (HI) 4 “M and (IV) 1 uM concentrations .................... 29

Figure 8. Lipid mixing induced by four fusion peptide constructs. In the assay, 150 [AM
LM3 was extruded into 100 nm diameter vesicles and mixing was induced by (I) 3 11M

and (I’) 1.5 uM FP; (H) 3 [AM and (H’) 1.5 MM FPK3; (IH) 3 uM and (IH’) 1.5 uM
FPK3W; and (IV) 3 11M and (IV’) 1.5 MM FPW ............................................................. 31

CHAPTER HI

Figure 9. 13C solid-state NMR spectra of (a) 5% polycrystalline AGG-(A1G2)C peptide;
(b) 20 mM frozen aqueous solution of HGRVGIYFGMK-FSC epitope peptide; (c) 7.5
mM frozen aqueous solution of FP-(L7F8)c; and (d) FP-(L7F8)C/DTPC made by
preparation method (2) at 1:20 peptide:lipid molar ratio. The AGG spectrum was taken at
room temperature and the other spectra were taken at —50 °C ........................................ 45

Figure 10. 13 C solid-state NMR spectra of samples made from FP-F8C bound to (a) DTPC
dispersion, (b) LM-l dispersion, (c) LM-3 dispersion, and (d) 100 nm diameter LM-3
vesicles with a peptidezlipid molar ratio of (a), (b) 1:200 and (c), (d) 1:80; and spectra of
samples made from FP-(G10F11)C bound to LM-3 with a peptidezlipid molar ratio of (e)
1:100 and (f) 1:20 ............................................................................................................. 48

Figure 11. 13C solid-state NMR spectra of samples made from FP-(L7F8)c/DTPC with a
peptidezlipid molar ratio of 1:200 by (a) method 1 and (b) method 2; and spectra of
sample made from FPK3-L7NF8c/LM-3 with a peptidezlipid molar ratio of 1:100 by (c)
method 1 and ((1) method 3 .............................................................................................. 50

Figure 12. 13C solid-state NMR difference spectra of LM-bound (a) FP-Alc, (b) FP-V2c,
(C) FP-Fgc, (d) FP-GIOC, (e) FP-FI lc, (f) FP-Al4c, (g) FP-AISC, and (h) FP-A21c.
Samples were prepared by method 1 with a peptidezlipid molar ratio of 1:100 .............. 52

Figure 13. 13C solid-state NMR spectra of membrane-associated fusion peptide samples
with (a) FP-F8CL9N, (b) FPK3-F8CL9N, (c) FPW-F8CL9N, and (d) FPK3W-F8CL9N. Each
sample contained ~ 0.3 limo] peptide and 40 umol LM3. The displayed REDOR-filtered
difference spectra are dominated by the Phe-8 carbonyl signals ..................................... 55

CHAPTER IV

Figure 14. 2D 13C solid-state NMR exchange spectra of (a) 5% polycrystalline AGG-
(A1G2)c peptide, and (b) Melittin-(G3A4)c/DTPC with a peptidezlipid molar ratio of
1:20. The corresponding secondary structure analyses are shown in (c) and ((1),
respectively. These analyses are displayed as contour plots of the total squared deviation
x2 (normalized to spectral noise) between experimental and calculated off-diagonal
crosspeak intensities for a grid of dihedral angles. The darkest region in each plot
represents the best global fitting ....................................................................................... 65

Figure 15. 2D 13C solid-state NMR exchange spectra of (a) FP-(A1V2)c/LM-3, (b)
FP—(L7F8)c/DTPC, (c) FP-(G10F11)c/LM-3, and (d) FP-(A14A15)c/LM-3. Each
spectrum represents the summation of 10-15 twelve-hour data blocks. Secondary
structure analyses for (a)-(d) are shown respectively in (e)—(h). These analyses are
displayed as contour plots of the total squared deviation x2 (normalized to spectral noise)

xi

between the experimental and calculated off-diagonal crosspeak intensities for a grid of
dihedral angles. The darkest regions represent values of x2 less than 6, 9, 12, and 12 for
(e)-(h), respectively. Each lighter contour level represents an increase of two units in x2 68

CHAPTER V

Figure 16. l3C-observed REDOR dephasing curve of calibrated AS/So for the 5% 1-13C,
l5N L—lcucine sample placed at different positions in the 4 mm rotor. A is the product of
dipolar coupling D (195 Hz, calculated from the 2.5 A l3C—ISN distance) and dephasing
time T. All experiments used the pulse lengths from the central position. The ﬁlled
triangles represent the results from two central 1/6 positions and the open triangles
represent the results from two outer 1/6 positions. The results from the two outmost 1/6
positions are not shown because the signals are signiﬁcantly attenuated ........................ 80

Figure 17. 13C-observed REDOR dephasing curve of calibrated AS/So for the 10% 1-‘3C,
15N glycine sample placed at different positions in the 6 mm rotor. A 195 Hz dipolar
coupling D from 2.5 A 13C—ISN distance was applied. All experiments used the pulse
lengths from the central position. The ﬁlled triangles represent the results from two
central 1/6 positions and the open triangles represent the results from two outer 1/6
positions. The half-filled triangles represent the results from two outmost 1/6 positions 81

Figure 18. l3C-observed REDOR dephasing curve of calibrated AS/So for the
MB(i+4)EK-A9CA13N peptide in full and 2/3 full 4 mm rotor (open and ﬁlled upper
triangles “A”), and in full and 2/3 full 6 mm rotor (open and ﬁlled lower triangles “V”).
100 kHz and 65 kHz TPPM decoupling were applied for the 4 mm and 6 mm rotors,
respectively ....................................................................................................................... 82

Figure 19. l3c—observed REDOR (a) so intensity and (b) calibrated AS/So for the
MB(i+4)EK-A9CA13N peptide in the 4 mm rotor. A 42 Hz dipolar coupling D from a 4.2
A 13C—ISN distance was applied. The filled and open triangles represent the results with
TPPM and CW decoupling sequences, respectively. The upper (A) and lower (V)
triangles represent the results with 100 kHz and 70 kHz decoupling field, respectively 88

Figure 20. l3c-observed REDOR calibrated AS/So for (a) 10% 1-‘3c, 15N Glycine with
196 Hz dipolar coupling D (2.50 A), and (b) FP-F8cL9N peptide with 1165 Hz dipolar
coupling D (1.38 A) in the 2/3 full 4 mm rotor. The upper (A) and lower (V) triangles
represent the results with 100 kHz and 65 kHz TPPM decoupling ﬁeld, respectively 89

Figure 21. l3C-observed REDOR calibrated AS/So for FP-F8CL9N peptide spinning at 16
kHz (open lower triangles) and 8 kHz (open upper triangles), 10% 1-13C, lsN Glycine
spinning at 16 kHz (half-ﬁlled lower triangles) and 8 kHz (half-ﬁlled upper triangles)
(half—ﬁlled triangles), and MB(i+4)EK-A9CA13N peptide spinning at 8 kHz (ﬁlled lower
triangles) and 4 kHz (filled upper triangles). All the experiments use 65 kHz TPPM
decoupling. At each dephasing time for the same sample, the experiment running at lower
speed contains half the amount of 15N n pulses as the one at higher speed ..................... 90

xii

Figure 22. Three REDOR sequences with (a) “all 15N 1: pulse” version, (b) “alternating

l5N/13C 17. pulse” version, and (c) “all 13C it pulse” version. XY-8 phase cycling is shown
is (a) and (b) and XY-4 phase cycling is shown in (c) ..................................................... 92

Figure 23. 13C-observed REDOR (a) So intensity and (b) calibrated AS/So for the 10% 1-
13C, ISN Glycine in the 2/3 full 4 mm rotor with all 15N 1: pulse version (ﬁlled triangles),
alternating 15N/BC 1: pulse version (half-ﬁlled triangles), and all 13C it pulse version
(open triangles). All REDOR experiments use 100 kHz TPPM decoupling. In (b), a
calculated REDOR dephasing curve based on 196 Hz dipolar coupling (2.50 A l3C-15N
distance) is represented by the solid line .......................................................................... 93

Figure 24. l3C-observed REDOR calibrated AS/So for the FP-F8cL9N peptide in the 2/3
full 4 mm rotor using all 15N 1t pulse version with 100 kHz (upper triangles) and 65 kHz
(lower triangles) TPPM decoupling. The calculated REDOR dephasing curve based on
1165 Hz dipolar coupling (1.38 A ' C—ISN distance) is represented by the solid line ..... 96

Figure 25. 13C-observed REDOR calibrated AS/So for the MB(i+4)EK—A9CA13N peptide
in the 2/3 full 4 mm rotor (ﬁlled triangles) and 6 mm rotor (open triangles). All 15N n:
pulse version REDOR with 100 kHz (upper triangles) and 70 kHz (lower triangles)
TPPM decoupling was used for the samples in the 4 mm rotor. All 15N 71: pulse version
(lower triangles) and alternating 15N/13C 1: pulse version (upper triangles) REDOR with
65 kHz TPPM decoupling were used for the samples in the 6 mm rotor. The calculated
REDOR dephasing curve based on 42 Hz dipolar coupling (4.20 A 13C-ISN distance) is
represented by the solid line ............................................................................................. 97

Figure 26. l3C-observed REDOR (a) So intensity and (b) calibrated AS/So MB(i+4)EK-
A9cA13N peptide in the 2/3 full 6 mm rotor with alternating 15N/‘3C 1: pulse REDOR
using 8n (upper triangles) and 8n + 2 (lower triangles) rotor periods .............................. 99

Figure 27. l3C-observed REDOR calibrated AS/So for the MB(i+4)EK-A9CA13N peptide
in the 2/3 full 6 mm rotor using 40 kHz 15N 1: pulse ﬁeld (lower triangles) and 20 kHz
l5N 1t pulse ﬁeld (upper triangles). The all 15N 1t pulse REDOR with 65 kHz TPPM
decoupling was used. The calculated REDOR de hasing curves based on 4.2 A (I), 4.3 A
(H), 4.4 A (111), 4.5 A (IV), 4.6 A (V), and 4.7 (v1) 3C—‘5N distances are represented
by solid lines ................................................................................................................... 101

CHAPTER VI

Figure 28. Structure of the gp41 soluble ectodomain trimer. (A) Ribbon drawing of the
trimer. (B) An end-on view looking down the 3-fold axis of the trimer ......................... 109

Figure 29. REDOR simulation from the SIMPSON pro am (triangles) and numeric
calculation (solid line). A 44.4 Hz dipolar coupling D (4.1 ) was used in both cases 114

Figure 30. (a) Coordinates of one 13C coupled to three adjacent ‘5 N in parallel and anti-
parallel segments of streptococcal protein G B1 domain with the carbonyl carbon

xiii

hydrogen bonded to the middle amide nitrogen. (b) SIMPSON simulated REDOR
dephasing ratios of the 13C dephased by only hydrogen bonded 15N (open circles), or two
more adjacent 15N in parallel strand arrangement (open upper triangles), or two more
adjacent 15N in anti-parallel strand arrangement (open lower triangles). The ﬁlled
symbols represent the corresponding simulation results when all the distances are
extended by a factor of 4.6/4.1 as the hydrogen bond distance expands from 4.1 A to 4.6
A based on the REDOR result of MK(i+4)-A9CA13N peptide at 24 ms dephasing time 116

Figure 31. 13c solid-state NMR spectra of the FPK3-(GSA6L7)c/FPK3—
(A14A15G16)N/LM-3 sample from (a) CP experiment and (b) So REDOR experiment
with a 24.0 ms dephasing time. The spectral intensities in each experiment were divided
by the number of scans in that experiment. The vertical scale in (b) is 8 times that in (a),
showing that the signal/scan in (b) is ~ 1/8 as large as that in (a). The (b) spectrum
appears better resolved than the (a) spectrum because the natural abundance lipid
carbonyls make a relatively smaller contribution to the So signal than to the CP signal 117

Figure 32. 13C solid-state NMR So (left) and 81 (right) REDOR spectra of membrane-
associated fusion peptides that had (a) ~ 0.8 umol each of FP-(F8L9GlO)c and FP-
unlabeled, (b) ~ 0.8 limo] each of FP-(F8L9GlO)N and FP—unlabeled, (c) ~ 1.6 umol
FPK3-unlabeled, and ((1) ~ 1.6 umol FP-(F8L9G10)N. Each of the samples in (a-d) also
contained 40 umol lipid and 20 umol cholesterol. For samples (a-c), LM-3 lipid mixture
was used and for sample ((1), LM-4 lipid mixture was used .......................................... 120

Figure 33. '3C solid-state NMR S0 and S1 REDOR spectra of membrane—associated
fusion peptides. Each sample was made with LM-3 lipid mixture, ~ 0.8 umol peptide
with selective l3C carbonyl labeling and ~ 0.8 mol peptide with selective 15N amide
labeling. The speciﬁc labeling schemes were: (a) FPK3-(G5A6L7)c/FPK3-(G5A6L7)N,
(b) FP-(F8L9G10)C/FP-(F8L9G10)N, (c) FP-(Fl1L12G13)c/FP-(F11L12G13)N, (d)
FPK3-(A14A15G16)c/FPK3-(A14A15G16)N, (e) FPK3-(G5A6L7)c/FPK3-
(A14A15G16)N, (f) FP-(F8L9G10)C/FP-(F11L12G13)N, (g) FP—(F11L12Gl3)C/FP-
(F8L9G10)N, and (h) FPK3-(A14A15Gl6)c/FPK3—(G5A6L7)N. Samples (a)-(d) primarily
probed the presence of parallel strand arrangements and samples (e)-(h) primarily probed
the presence of anti-parallel strand arrangements. The dotted lines are included for visual
comparison of the spectral intensities ............................................................................ 124

Figure 34. Peptide oligomerization and membrane insertion models for the (a) parallel
strand arrangement, and (b) anti-parallel strand arrangement. The arrows indicate the
peptide direction from N-terminus to C-terminus while the numbers label the indices of
individual residues. In the parallel arrangement, inter-peptide hydrogen bonding was
observed at N-terminal and central regions, and considerably weaker toward the C-
terrninus, which is consistent with an apolar region of peptide insertion into membrane
interior model. In the anti-parallel arrangement, inter—peptide hydrogen bonding was
detected along the whole length of the probed region. Since this arrangement would place
the polar region of one peptide and the apolar region of the other peptide at the same side,
it is unclear whether it is a result of peptides binding on the surface of membrane or
peptides inserting through the membrane ...................................................................... 125

xiv

1D/2D
AIDS
ANTS

CD

CP

CSA

cw

DMPC
DMPS
DMSO
DPPC-l-‘3C
DPX

DTPC
FWHM

HEPES

HPLC

LUV

MALDI

MAS

LIST OF SYMBOLS AND ABBREVIATIONS

One—Dimensional / Two-Dimensional

Acquired Immune Deﬁciency Syndrome
8—AminoNaphthalene-l,3,6-TriSulfonic acid

Circular Dichroism

Cross-Polarization

Chemical Shift Anisotropy

Continuous Wave

1 ,2-DiMyristoyl-sn-glycero-3-PhosphoCholine
1,2-DiMyristoyl-sn-glycero-3-[Phospho-L—Serine]
DiMethyl SulfOxide
1,2-DiPalmitoyl[1-13C]-Sn-glycero-3~PhosphoCholine
p-xylene-bis-pyridinium bromide
Di-o-Tetradecyl-sn-glycero-3-PhosphoCholine
Full-Width-at-Half-Maximum
N-2-HydroxyEthylPiperazine-N’-2-EthaneSulfonic acid
Human Immunodeficiency Virus

High Performance Liquid Chromatography

InfraRed

Large Unilamellar Vesicles

Matrix Assisted Laser Desorption Ionization

Magic Angle Spinning

XV

N-NBD-PE

N-Rh-PE

PI

POPG

POPC

POPE

POPS

PPm
REDOR

TFA

TPPM

N-(7-Nitro-2,1,3-Benzoaniazol—4-yl)-PhosphatidylEthanolamine
N-(lissamine Rhodamine B sulfonyl)-PhosphatidylEthanolamine
Nuclear Magnetic Resonance

Phosphatidlenositol
1-Palmitoyl-2-Oleoyl-sn-glycero—3-[Phospho-rac-(l-Glycerol)]
1-Palmitoyl-2-Oleoyl-sn-glycero-3-PhosphoCholine
1-Palmitoyl-2-Oleoyl-sn-glycero-3-PhosphoEthanolamine
1-Palmitoyl-2-Oleoyl-sn-glycero-3—[Phospho-L—Serine]

parts per million

Rotational-Echo DOuble-Resonance spectroscopy
RadioFrequency

TriFluoroacetic Acid

Two-Pulse Phase-Modulation

xvi

CHAPTERI

INTRODUCTION

BACKGROUND

Fusion between cells and cellular components has an essential role in organismal
life and is an integral part of critical processes such as egg fertilization and synaptic
transmission in the nervous system. Membrane fusion is also an important step in viral
infection for the widespread diseases measles, inﬂuenza and most seriously, acquired
immunodeﬁciency syndrome (AIDS)’. AS a crucial step in the viral life cycle, viral
fusion can be the potential target for anti-viral therapeutics. In March 2003, Roche and
Trimeris, Inc. introduced the first U. S. Food and Drug Administration (FDA) approved
anti-fusion treatment for human immunodeficiency virus type-l (HIV-1), FUZEONTM
(enfuvirtide), which is a peptide inhibitor that prevents fusion of the viral and cell
membranes“. This drug also obtained the approval of European Commission to be used
against HIV in the European Union. T-1249, a second-generation fusion inhibitor through
the Roche/Trimeris partnership, is also in Phase I/H clinical testing.

Fusion between two membrane-bound bodies such as cells, viruses or vesicles is a
protein-mediated process and is generally separated into three sequential steps8: (1)
binding of the two bodies; (2) mixing of their membrane lipids; (3) formation of a large
fusion pore through which the contents of virus and cell can mix. Figure 1 illustrates a
series of freeze fracture electron micrographs that follow the time evolution of I-HV
virion infection of a host cellg.

In AIDS, fusion and infection are mediated by the heavily glycosylated,
predominantly extraviral envelope proteins gp120 and gp41 which form a non-covalently
bound complex. The complex is attached to the virus through a transmembrane segment

in gp41. Conserved regions in the gp120 subunit are responsible for the binding of the

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Figure 1. Model (left) and freeze fracture electron microscopy (right) of I-HV virion (a)
binding to, (b) fusion with, (c-d) forming large pore, and infecting host cell.

 

Figure 2. Model of HIV fusion. FP: Fusion Peptide. Time sequence: left to right.

virus to the CD4 and chemokine (e.g. CXCR4) receptors of human T and macrophage
cells prior to fusion”’ 11. This interaction is followed by a series of poorly understood
conformational changes which are believed to eventually lead to the exposure of a
conserved segment of about twenty amino acids at the N-terminus of gp41]2 (Figure 2).
This region has been proposed to have an integral role in the fusion mechanism of I-HV
and thus is given the name “Fusion Peptide (FP)”. The fusion peptide of gp41 shares
homology with regions of other enveloped viruses, such as ortho- and paramyxoviruses”.
A typical fusion peptide consensus sequence is the 23-mer N-terrninal residues of gp41:
Ala-Val-Gly-Ile-Gly—Ala-Leu-Phe-Leu-Gly-Phe-Leu-Gly-Ala-Ala-Gly-Ser-Thr-Met-Gly-

Ala—Arg-Ser.

The importance of this region in membrane fusion was initially shown by
mutation/fusion activity studies on Hela cells engineered to express the envelope protein.
Hydrophobic to polar mutations throughout the fusion peptide region decreased fusion
activity by more than 50 times but had no effect on binding to the CD4 receptor“.
Significantly, the Val-2 to Glu-2 mutation was trans-dominant, i.e. cells expressing ten
times more wild-type than mutant protein had only 40% of the fusion activity of those
expressing only wild-type protein”. Further studies showed that deletion of four or more
N-terrninal amino acids in the fusion peptide completely disrupted fusion activity”.
Fusion activity was also decreased at least twenty-fold with point mutants of Gly—lO
and/or Gly-13 to Val". Both of these glycine residues are found in the strongly conserved
FLGFLG motif at the center of the fusion peptide.

Another important aspect is that even in the absence of the rest of gp41, the 20-

residue fusion peptide at ~10 11M concentration and 1:20 — 1:100 peptidezlipid molar

ratios causes rapid fusion and/or leakage of lipid vesicles or erythrocytes. In these
studies, fusion is detected ﬂuorimetrically with assays which monitor (1) mixing of lipids
from different vesicles, (2) mixing of contents between different vesicles, and/or (3)
mixing of vesicle contents with the extravesicular medium. In addition, the growth in the
size of vesicles or erythrocytes in the presence of fusion peptide has been monitored with
either light scattering or electron n1icroscopy18'26. Also the Site-directed mutation/fusion
activity or deletion/fusion activity studies showed the same pattern as is formed in the
intact protein. In particular, (1) deletion of three or four N—terminal amino acids

eliminates intervesicle lipid mixing and vesicle leakage” 23; (2) hydrophobic to polar

4
2 , 25, 27; and

point mutations also eliminate intervesicle lipid mixing and vesicle leakage
(3) the Val-2 to Glu-2 point mutation peptide had a trans-dominant effect in mixtures
with wild-type peptide. In these latter studies, lipid mixing was inhibited by a factor of
four relative to having only wild-type peptide. The observed trans-dominant effect was
attributed to the inability of the mutated peptide to form aggregates larger than dimer”.
These data provide strong evidence that the free fusion peptide is a good model system to
probe biologically relevant fusion.

Despite its biological importance, there is no atomic-level structure of the fusion
peptide region in membranes. The hydrophobicity of this region impairs the solubility
requirement of traditional high-resolution structural techniques, such as crystallographic
and solution nuclear magnetic resonance (NW) analysis, and neither technique is suited
to studies in intact membrane systems. Other techniques (e.g. circular dichroism, infrared,

ﬂuorescence, electron spin resonance, etc.) have provided general structural information

about the membrane-bound fusion peptide. However, due to the difference in sample

preparation protocols, the results from these studies are complex and sometimes
contradictory.
Circular dichroism (CD) measurements on the I~HV -1 fusion peptide show

24, 28-31

significant helical character for samples in organic solvent , in sodium dodecyl

9 3 . .
24' 2 ' 2, and in an envrronment of

sulfate (SDS) detergent at a 1:200 peptidezlipid ratio
negatively charged vesicles with a 1:200 peptidezlipid molar ratio”. NlVTR studies on the
peptide in SDS micelles are also consistent with a helical central region”. By contrast, at
1:10 peptide2SDS molar ratio” or 1:30 peptidezlipid molar ratio”, there is signiﬁcant B
strand character. Similar results were found in 1:1 mixtures of neutralznegatively charged
lipids”.

With neutral lipid vesicles or erythrocyte ghosts at peptidezlipid molar ratios of
~ 1:200, there are two infrared reports of predominantly helical structure” 29, three

18, 25, 34

reports of predominantly B structure , and one report of mixed helical and B

Structure”. Two investigators report that the peptide conformation changes from helical

23 29
0 l

to B as the peptidezlipid molar ratio is increased from 1:200 to 1:3 while two others

report that the B conformation does not change with these ratios's' 25. These differences in
structure may have to do with differences in peptide sequence, lipid composition, sample
preparation, or hydration level. A recent infrared study of the peptide in a neutral
membrane monolayer was also consistent with a B structure”.

Fluorescence and ESR measurements on derivatized HIV -1 fusion peptides in
membrane vesicle environments provide evidence that the N-terminal and central regions

are inserted into the membrane and the C-terminus is outside the membrane“ 29. In some

contrast, a H/D NMR exchange study in SDS micelles suggests that the N-terminus is

outside the micelle36.

These complicated structure results suggest that residue-specific structural
information is necessary because different regions of the peptide may adopt different
structures. Solid-state NMR is suitable to this kind of structural measurement because
unlike the X-ray technique, crystals are not required and the sample can be Studied in a
fully hydrated membrane environment. In solid-state NME, orientational information is
also retained in the system.

The goal of this study is to apply solid-state NlVHl methodologies to determine the
structure of synthetic fusion peptide in a membrane environment. We will exploit
isotopic labeling at speciﬁc positions to probe the secondary structures as well as the
peptide oligomerization Status of the peptide. These results will be incorporated with the
orientation measurements of the peptide with respect to the membrane and the peptide
effect on lipid dynamics and structure to understand PHV-l viral fusion at a molecular

level.

REFERENCE

10.

11.

Pic], J. (1988) The Science of AIDS, W. H. Freeman and Company, New York.

Wild, C. T., Shugars, D. C., Greenwell, T. K., McDanal, C. B., and Matthews, T.
J. (1994) Peptides corresponding to a predictive alpha-helical domain of human
immunodeﬁciency virus type 1 gp41 are potent inhibitors of virus infection. Proc.
Natl. Acad. Sci. U. S. A. 91, 9770-9774.

Kilby, J. M., Hopkins, 3., Venetta, T. M., DiMassimo, B., Cloud, G. A., Lee, J.
Y., Alldredge, L., Hunter, E., Lambert, D., Bolognesi, D., Matthews, T., Johnson,
M. R., Nowak, M. A., Shaw, G. M., and Saag, M. S. (1998) Potent suppression of
I-HV-l replication in humans by T-20, a peptide inhibitor of gp41-mediated virus
entry. Nat. Med. 4, 1302-1307.

Kilby, J. M., Lalezari, J. P., Eron, J. J., Carlson, M., Cohen, C., Arduino, R. C.,
Goodgarne, J. C., Gallant, J. E., Volberding, P., Murphy, R. L., Valentine, F.,
Saag, M. 8., Nelson, E. L., Sista, P. R., and Dusek, A. (2002) The safety, plasma
pharrnacokinetics, and antiviral activity of subcutaneous enfuvirtide (T-20), a
peptide inhibitor of gp41-mediated virus fusion, in I-HV-infected adults. Aids Res.
Hum. Retrovir. 18, 685-693.

Barretina, J ., Blanco, J ., Armand-Ugon, M., Gutierrez, A., Clotet, B., and Este, J.
A. (2003) Anti-PHV -1 activity of enfuvirtide (T-20) by inhibition of bystander cell
death. Antivir. Ther. 8, 155-161.

Cooleya, L. A., and Lewin, S. R. (2003) PHV-l cell entry and advances in viral
entry inhibitor therapy. J. Clin. Virol. 26, 121-132.

Gulick, R. M. (2003) New antiretroviral drugs. Clin. Microbiol. Infect. 9, 186-
193.

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

Grewe, C., Beck, A., and Gelderblom, H. R. (1990) I-HV: early virus-cell
interaction. J. Acq. Immun. Def. Synd. 3, 965-974.

Bleul, C. C., Farzan, M., Choe, H., Parolin, C., ClarkLewis, I., Sodroski, J ., and
Springer, T. A. (1996) The lymphocyte chemoattractant SDF-l is a ligand for
LESTR/fusin and blocks I-HV-l entry. Nature 382, 829-833.

Oberlin, E., Amara, A., Bachelerie, F., Bessia, C., Virelizier, J. L.,
ArenzanaSeisdedos, F., Schwartz, 0., Heard, J. M., ClarkLewis, 1., Legler, D. F.,
Loetscher, M., Baggiolini, M., and Moser, B. (1996) The CXC chemokine SDF-l

12.

13.

14.

15.

16.

17.

18.

19.

20.

21.

22.

is the ligand for LESTR/fusin and prevents infection by T-cell-line-adapted MV-
1. Nature 382, 833-835.

Balter, M. (1996) New hope in IHV disease. Science 274, 1988-1989.

Gallaher, W. R. (1987) Detection of a fusion peptide sequence in the
transmembrane protein of human-immunodeﬁciency-virus. Cell 50, 327-328.

Freed, E. 0., Myers, D. J ., and Risser, R. (1990) Characterization of the fusion
domain of the human-immunodeﬁciency-virus type-1 envelope glycoprotein
Gp41. Proc. Natl. Acad. Sci. U. S. A. 87, 4650-4654.

Freed, E. 0., Delwart, E. L., Buchschacher, G. L., and Panganiban, A. T. (1992)
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. 89, 70-74.

Schaal, R, Klein, M., Gehrrnann, R, Adams, 0., and Scheid, A. (1995)
Requirement of N-terminal amino-acid-residues of gp41 For human-
immunodeficiency-virus type l-mediated cell fusion. J. Virol. 69, 3308-3314.

Delahunty, M. D., Rhee, 1., Freed, E. 0., and Bonifacino, J. S. (1996) Mutational
analysis of the fusion peptide of the human immunodeﬁciency virus type 1:
Identification of critical glycine residues. Virology 218, 94-102.

Rafalski, M., Lear, J. D., and Degrado, W. F. (1990) Phospholipid interactions of
synthetic peptides representing the N-terminus of IHV gp41. Biochemistry 29,
7917-7922.

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

Martin, I., Defrisequertain, F., Decroly, E., Vandenbranden, M., Brasseur, R., and
Ruysschaert, J. M. (1993) Orientation and structure of the NH2-terminal I-HV-l
gp41 peptide in fused and aggregated liposomes. Biochim. Biophys. Acta 1145,
124-133.

Nieva, J. L., Nir, S., Muga, A., Goni, F. M., and Wilschut, J. (1994) Interaction of
the I-HV -1 fusion peptide with phospholipid vesicles - different structural
requirements for fusion and leakage. Biochemistry 33, 3201-3209.

Mobley, P. W., Lee, H. F., Curtain, C. C., Kirkpatrick, A., Waring, A. J., and
Gordon, L. M. (1995) The arnino-terminal peptide of PHV-l glycoprotein-41
fuses human erythrocytes. Biochim. Biophys. Acta-Mal. Basis Dis. 1271, 304-314.

23.

24.

25.

26.

27.

28.

29.

30.

31.

32.

Martin, I., Schaal, H., Scheid, A., and Ruysschaert, J. M. (1996) Lipid membrane
fusion induced by the human immunodeﬁciency virus type 1 gp41 N-terminal
extremity is determined by its orientation in the lipid bilayer. J. Virol. 70, 298-
304.

Kliger, Y., Aharoni, A., Rapaport, D., Jones, R, Blumenthal, R., and Shai, Y.
(1997) Fusion peptides derived from the I-HV type 1 glycoprotein 41 associate
within phospholipid membranes and inhibit cell-cell fusion - Structure-function
study. J. Biol. Chem. 272, 13496-13505.

Pereira, F. B., Goni, F. M., Muga, A., and Nieva, J. L. (1997) Permeabilization
and fusion of uncharged lipid vesicles induced by the HIV-1 fusion peptide
adopting an extended conformation: Dose and sequence effects. Biophys. J. 73,
1977-1986.

Nieva, J. L., Nir, S., and Wilschut, J. (1998) Destabilization and fusion of
zwitterionic large unilamellar lipid vesicles induced by a beta-type structure of the
I-HV-l fusion peptide. J. Liposome Res. 8, 165-182.

Pereira, F. B., Goni, F. M., and Nieva, J. L. (1995) Liposome destabilization
induced by the IHV-l fusion peptide effect of a single amino-acid substitution.
FEBS Lett. 362, 243-246.

Slepushkin, V. A., Andreev, S. M., Sidorova, M. V., Melikyan, G. B., Grigoriev,
V. B., Chumakov, V. M., Grinfeldt, A. E., Manukyan, R. A., and Karamov, E. V.
(1992) Investigation of human-immunodeficiency-virus fusion peptides - analysis
of interrelations between their structure and function. Aids Res. Hum. Retrovir. 8,
9-18.

Gordon, L. M., Curtain, C. C., Zhong, Y. C., Kirkpatrick, A., Mobley, P. W., and
Waring, A. J. (1992) The arnino-terrninal peptide of I-HV-l glycoprotein-41
interacts with human erythrocyte membranes - peptide conformation, orientation
and aggregation. Biochim. Biophys. Acta 1139, 257-274.

Chang, D. K., Chien, W. J., and Cheng, S. F. (1997) The FLG motif in the N-
terminal region of glucoprotein 41 of human immunodeficiency virus type 1
adopts a type-I beta turn in aqueous solution and serves as the initiation site for
helix formation. Eur. J. Biochem. 247, 896-905.

Waring, A. J ., Mobley, P. W., and Gordon, L. M. (1998) Conformational mapping
of a viral fusion peptide in structure- promoting solvents using circular dichroism
and electrospray mass spectrometry. Proteins, 38-49.

Chang, D. K., Cheng, S. F., and Chien, W. J. (1997) The arnino-terminal fusion
domain peptide of human immunodeficiency virus type 1 gp41 inserts into the

10

33.

34.

35.

36.

sodium dodecyl sulfate rnicelle primarily as helix with a conserved glycine at the
rnicelle-water interface. J. Virol. 71, 6593-6602.

Mobley, P. W., Waring, A. J., Sherman, M. A., and Gordon, L. M. (1999)
Membrane interactions of the synthetic N-terminal peptide of IHV-l gp41 and its
structural analogs. Biochim. Biophys. Acta 1418, 1-18.

Peisajovich, S. G., Epand, R. F., Pritsker, M., Shai, Y., and Epand, R. M. (2000)
The polar region consecutive to the PHV fusion peptide participates in membrane
fusion. Biochemistry 39, 1826-1833.

Taylor, S. E., Desbat, B., Blaudez, D., Jacobi, S., Chi, L. F., Fuchs, H., and
Schwarz, G. (2000) Structure of a fusion peptide analogue at the air-water
interface, determined from surface activity, infrared spectroscopy and scanning
force microscopy. Biophys. Chem. 87, 63-72.

Chang, D. K., and Cheng, S. F. (1998) Determination of the equilibrium micelle-
inserting position of the fusion peptide of gp41 of human immunodeﬁciency virus
type 1 at amino acid resolution by exchange broadening of amide proton
resonances. J. Biomol. NMR 12, 549-552.

11

CHAPTER H

PHYSICAL AND BIOLOGICAL PROPERTIES OF THE FUSION PEPTIDE

l2

BACKGROUND

The first step of the project involves the examination of the physical properties of
the fusion peptide in aqueous solution before it binds to the membrane. The biological
functionality of the peptide is also investigated.

In order to investigate the peptide structure in its native membrane environment,
we need to ensure that the system is biologically relevant. As discussed in the previous
chapter, the free fusion peptide without the rest of the gp41 protein has similar biological
function as the intact protein. Therefore, a series of fusion assays were applied to the
synthetic fusion peptide and the results were compared to those from other groups to
prove that our sample system is indeed biologically relevant.

It is important that we can compare the physical properties of the fusion peptide
before and after binding to the membrane and correlate any physical property differences
with the biological functionality. Therefore we also examined the physical properties of

the fusion peptide prior to membrane association.

MATERIALS AND METHODS

Materials. Rink amide resin was purchased from Advanced Chemtech
(Louisville, KY), and 9-ﬂuorenylmethoxycarbonyl (FMOC)-amino acids were obtained
from Peptides International (Louisville, KY). l3C Carbonyl labeled amino acids were
purchased from Icon Services Inc. (Summit, NJ) and the FMOC group was added using
literature methods”. 1,2-Dimyristoyl-sn-glycero-3-phosphocholine (DMPC), 1,2-
dimyristoyl-sn-glycero-3-[phospho-L-serine] (DMPS), 1,2-dipalmitoyl[1-13C]-sn-

glycero-3-phosphocholine (DPPC-l-BC), di-o-tetradecyl-sn-glycero-3-phosphocholine

l3

(DTPC), 1-palmitoyl-2—oleoyl-sn-glycero-3-[phospho-rac-(1-glycerol)] (POPG), l-
palmitoyl—2-oleoyl-sn-glycero-3-phosphocholine (POPC), 1-palmitoy1—2-oleoyl-sn-
glycero-3-phosphoethanolamine (POPE), 1-palmitoyl-2-oleoyl-sn-glycero-3-[phospho-L-
serine] (POPS), phosphatidylinositol (PI), sphingomyelin, N-(7-nitro-2,l,3-
benzoxadiazol-4-yl)-phosphatidylethanolamine (N-NBD-PE), N-(lissamine Rhodamine B
sulfonyl)-phosphatidylethanolarnine (N-Rh-PE) and cholesterol were purchased from
Avanti Polar Lipids, Inc. (Alabaster, AL). 8-aminonaphthalene-l,3,6-trisulfonic acid
(ANTS) and p-xylene-bis-pyridinium bromide (DPX) were purchased from Molecular
Probes (Eugene, OR). The Micro BCA (bicinchoninic acid) protein assay was obtained
from Pierce (Rockford, E). N-2-hydroxyethylpiperazine-N’-2-ethanesulfonic acid
(HEPES) and Triton X-100 were obtained from Sigma. All other reagents were analytical
grade.

Peptides. FP peptides corresponding to the 23 N-terminal residues
(AVGIGALFLGFLGAAGSTMGARS) of the LAV;al strain of HIV-1 gp41 were
synthesized as their C-terrninal amides using a peptide synthesizer (ABI 431A, Foster
City, CA) equipped for FMOC chemistry. FP was either unlabeled (denoted as “FP”) or
singly, doubly or triply l3C carbonyl or 15N amide labeled (denoted as “FP-AngN” where
“A” and “n” are the one-letter abbreviation and the index of the labeled residue and the
subscript indicates whether the carbonyl carbon or amide nitrogen is 13C or 15N labeled).
Other FP peptide analogs were synthesized in a similar approach, including FPK3
(AVGIGALFLGFLGAAGSTMGARSKKK) with three-lysine addition at the C-terminus
to improve aqueous solubility, FPW (AVGIGALFLGFLGAAGSTMGARSW) with one-

tryptophan addition at the C-terminus as a UV/Vis chromophore, and FPK3W

14

 

(AVGIGALFLGFLGAAGSTMGARSKKKW) to take advantage of both aspects. The
unlabeled and labeled peptides are denoted in the same pattern as the FP peptide.
Peptides were cleaved from the resin in a three hour reaction using a mixture of
triﬂuoroacetic acid (TFA):H202phenol:thioanisolezethanedithiol in a 33:2:2:2:l volume
ratio. Peptides were subsequently puriﬁed by reverse-phase HPLC using a preparative
C13 column (Vydac, Hesperia, CA) and a water/acetonitrile gradient containing 0.1%
TFA. Matrix assisted laser desorption ionization (MALDI) mass spectroscopy was used
to demonstrate the success of the synthesis.

Lipid Preparation. Samples were either prepared using single lipids or using a
lipid/cholesterol mixture reﬂecting the approximate lipid and cholesterol content of the
HIV-1 virus infected host cells3. Three such mixtures were used: (a) “LM-l” had POPC,
POPE, DMPS, and cholesterol in a 10:6:4:10 molar ratio; (b) “LM-2” had POPC, POPE,
DMPS, sphingomyelin and cholesterol in a 10:6:4:2:10 molar ratio; (c) “LM-3” had
POPC, POPE, POPS, sphingomyelin, PI and cholesterol in a 10:5:2:2:1:10 molar ratio.
Lipid and cholesterol powders were dissolved together in chloroform. The chloroform
was removed under a stream of nitrogen followed by overnight vacuum pumping. Lipid
dispersions were formed by addition of water or buffer containing 0.01% NaN3 followed
by homogenization with ten freeze-thaw cycles. Large Unilamellar Vesicles (LUV) of
100 nm diameter were prepared by extruding the lipid dispersions ~30 times through two
stacked 0.1 pm polycarbonate ﬁlters" (Avestin, Inc., Ottawa, ON, Canada).

Peptide Aggregation. Peptide aggregation in aqueous solution was estimated from
measurements of peptide concentration (by the BCA assay) in solution before and after

centrifugation. Centrifugation was either done on a Sorvall RC-SB centrifuge (Newtown,

15

CT) with a GSA rotor (14,000 x g, 90 minutes) or on a Sorvall Ultra 80 ultracentrifuge
(Newtown, CT) with a SW50.1 or SW25.1 rotor (100,000 - 150,000 X g, 120-240
minutes).

Solution NMR Spectroscopy. The NMR measurements were carried out on a
Varian INOVA-600 spectrometer at 25 °C. The sample was dissolved in a mixture of 700
u] HZO/DZO (6:1 v/v) and acidiﬁed with 30 mM HCl solution in some experiments. The
strong H20 proton signal was suppressed by applying the binomial pulse sequence5.

Sedimentation Equilibrium. Sedimentation equilibrium experiments were
performed at 20 0C on a Beckman XL—I analytical ultracentrifuge using an An-60 Ti
rotor. The measurements were carried out using six-channel centerpieces loaded with 110
1.1.1 of peptide solution against 110 u] of the buffer. Samples were spun between 5,000
rpm and 52,000 rpm and the 280 nm absorbance was recorded as a function of the
centrifugal radius. For each speed, data were collected three times at every 4 h and the
proﬁles at the 8m and 12‘h hours were identical, conﬁrming that equilibrium had been
reached. Multiple records at different speed were subjected to the non-linear regression
analysis using the software from Beckman/MicroCal to obtain the best ﬁt for the
molecular weight (M) of the solute by equation (1):

A. = A0 exp[M x (l—yp) x (rz—roz) x (m2 / 2RT)] (1)
where A, and A0 are the absorbance at radius r and at the reference radius r0, 3 is the
partial specific volume, p is the buffer density, (1) is the angular velocity, R is the gas
constant, and T is the temperature.

Circular Dichroism (CD) Spectroscopy. CD spectra were recorded on a JASCO

J 810 Spectropolarimeter (Tokyo, Japan) at 20 0C using a quartz optical cell with a 10 mm

16

 

pathlength. The spectra were recorded from 190 nm to 250 nm at a scanning rate of 100
nm/min with a step resolution of 0.1 nm and bandwidth of 1 nm. Each set of CD data was
obtained from an average of eight scans. Samples were dissolved in water or 5 mM
HEPES (pH 7.0). The observed ellipticity (0) was converted to mean residue ellipticity
[0] to allow direct comparison among peptides by the following equation:
[0]=0/(lxcxN) (2)

where l is the cell length in mm, c is the molar concentration of the sample, and N is the
number of amino acid residues in the peptide.

Lipid Mixing Assay for Membrane Fusion. The resonance energy transfer assay of
Struck, et al. was used to monitor membrane fusion6. Two types of LUV's were prepared.
One set contained 2 mole percent of the ﬂuorescent lipid N-NBD-PE and 2 mole percent
of the quenching lipid N-Rh-PE while the other set contained only the unlabeled lipids.
Fluorescently labeled and unlabeled vesicles were mixed in a 1:9 ratio. Following
addition of the peptide, lipid mixing between labeled and unlabeled vesicles caused
dilution of the labeled lipids with a resulting increase of ﬂuorescence. Fluorescence was
recorded using 4 nm bandwidth on a Instruments S. A. Fluoromax-2 spectroﬂuorimeter
operating at excitation and emission wavelengths of 465 nm and 530 nm, respectively. A
siliconized glass cuvette was used with continuous stirring in a thermostated cuvette
holder. Measurements were carried out at 37 °C with 2 ml of 150 MM LUV in 5 mM
HEPES (pH 7.0). A small quantity of peptide solution was added to achieve the desired
peptidezlipid molar ratio, and the ﬂuorescence of the sample (F,) was monitored as a
function of time following this addition. The initial residual ﬂuorescence intensity was

recorded as F0. The maximum ﬂuorescence intensity, Fm“, was obtained following

17

addition of 20 it] of 10% Triton X-100. Ft and Fmax were corrected for the small
ﬂuorophore dilution resulting from the added volumes of peptide and Triton X-100.
Percent lipid mixing at time t (M,) was given by [(F, -Fo)/(Fmax- 0)] x 100%. With
sufﬁcient time, Mt reaches a constant value which is denoted as Mr, the ﬁnal extent of
lipid mixing. For a given set of labeled and unlabeled vesicles, Mr had i2%
reproducibility between different assay trials. In this study, Mf is used as a measure of
peptide fusogenicity.

Leakage Studies. Aqueous content leakage from liposomes was monitored by the
ANT S/DPX assay7. LUV was prepared in 5 mM HEPES (pH 7.0) solutions which
contained the ﬂuorescent molecule ANTS and quenching molecule DPX at 12.5 mM and
45 mM, respectively. ANTS and DPX were then removed from the extravesicular
medium by dialysis against 65 mM HEPES solution. After addition of a small quantity of
peptide solution to achieve the desired peptidezlipid molar ratio, vesicle leakage was
monitored by the increase in ANTS ﬂuorescence. Fluorescence measurements were
performed at excitation and emission wavelengths of 355 nm and 520 nm, respectively.
Measurements were made at 37 °C using 2 ml of 80 11M LUV in 65 mM HEPES (pH

7.0). Percent leakage was defined in the same way as was used for the lipid mixing assay.

RESULTS AND DISCUSSION

Peptide Aggregation in Aqueous Solution. Fusion peptide has been studied in
aqueous solution at acidic conditions. However, in neutral pH PBS or saline buffered
D20 where the fusion of I-HV virion takes place, at least two groups has reported the

formation of large fibrillar fusion peptide aggregates by electron microscopy9 ’10.

18

Furthermore, the fusion activity of the peptide was significantly reduced due to the
aggregation"). To alleviate this problem, fusion peptide was often prepared in an organic
solvent such as dimethyl sulfoxide (DMSO) or triﬂuoroethanol and then transferred into
the aqueous media for further studies. Although control experiments generally showed
that the organic solvent by itself did not cause fusion in the functional assays, it was not
clear whether the peptide molecules collide and form aggregates before they bind to
membranes. In order to ﬁnd non-aggregating conditions without introducing organic
component into the sample environment, we carried out a systematic investigation on
wild type 23-residue FP peptide in the absence of membrane.

Soluble peptide was separated from aggregated peptide through centrifugation.
The concentration of the soluble peptide was measured using the highly sensitive Micro
BCA protein assay. Results from these studies are presented in Table 1, with
uncertainties of i 10% in the peptide concentration measurements. In unbuffered
solution, the fusion peptide formed a clear solution, even at mM concentrations. After
addition of phosphate buffer or NaCl, the solution became cloudy. Formation of large
aggregates was conﬁrmed by signiﬁcant loss of peptide concentration after
centrifugation. Large aggregates were not formed in 100 mM acetate buffer pH 4.5 or in
5 mM HEPES buffer pH 7.0. Thus, there appear to be pH and ionic strength effects on
aggregation. The differences in aggregation between HEPES and phosphate or NaCl may
be a result of the differences in their molecular sizes. Based on these results, most solid-
state NMR samples were made using 15 — 150 MM peptide concentration in 5 mM pH 7.0
HEPES buffer, which is close to the physiological pH of 7.4 for I-HV -1/host cell fusion.

Although physiological solutions contain ~140 mM NaCl, it was omitted from the

19

sample preparation because it causes aggregation. It would be possible to add NaCl after
the peptide has bound to the membrane.

A more soluble fusion peptide construct was introduced based on the study of
inﬂuenza fusion peptide by Tamm and coworkers. In this study, addition of a lysine-rich
“host sequence” at the C-terminus of the peptide greatly improves the peptide solubility
in aqueous solution“. For some of our studies, a similar modiﬁcation was made for the
I-HV -1 fusion peptide (FPK3 and FPK3W).

The oligomerization of viral envelope proteins is well known and there is some
evidence that fusion peptide oligomerization is a structural requirement for membrane
fusion”. Fusion peptide oligomers might be pre-formed in solution and bind as a unit to
the membrane or oligomers might only form during interaction with the membrane.
Using a simplified model for ultracentrifugation, our data for the 15 11M peptide
concentration in 5 mM pH 7.0 HEPES demonstrate that the solution contains no more

than ~900 FP molecules in any aggregate”.

20

Table l. FP Peptide Aggregation in Different Aqueous Buffer Solutions.

 

 

Initial Conc. (uM) Solution Conditions % Soluble Peptide“‘b

6200 Unbuffered H20 90

210 100 mM acetate, pH 4.5 >90

1300 5 mM HEPES, pH 7.0 88

1100 5 mM HEPES, pH 7.0, sitting for 2 days 80
15 5 mM HEPES, pH 7.0, sitting for 1 day >90

""""""" 3 010mMphosphatepH7016

22 10 mM phosphate, pH 6.0 60

1100 5 mM HEPES, pH 7.0, 20 mM NaCl 18

 

a In rows 1-4 and 6-9, the concentration was measured after centrifugation at 14,000 x g
for 90 min. In row 5, the concentration was measured after centrifugation at 100,000 x
g for 120 min.

b Estimated error: i10%.

Solution NMR linewidths also provide some information about aggregate size. At
all pH, there is a molecular tumbling contribution to the observed linewidth that is
proportional to molecular weight”. Figure 3 displays the amide region of the 600 MHz
solution 1H NMR spectra of 80 pM FP and FPK3 solution at acidic and neutral pH. The
narrowest linewidths for the neutral pH FF and FPK3 spectra are ~ 16 Hz, which
correspond to ~ 25 KD molecular weight or ~ 10-12 peptide molecules. In these neutral
pH spectra, signals from individual amide proton are not completely resolved and it is
possible that the true linewidths and oligomer sizes are less than these estimates. In
addition, in the amide region, exchange with the solvent may broaden the observed

lines”. The exchange rate is significantly reduced at lower pH, which is consistent with

21

the observation of narrower linewidths of ~ 5 Hz in both spectra at pH 3.7. This linewidth
is consistent with a monomer or oligomer of two or three molecules. Since both fusion
peptide sequences have no acidic residues and their C-termini are amidated, they should
have similar type of charge in both pHs and it’s not clear that lower pH should alter
oligomerization. The exchange rate at different pH is likely the reason of different
linewidths. Further study to observe the linewidth of non-exchangeable Ca protons at
different pHs would give a better understanding of the oligomerization dependence on pH
change.

A better way to estimate the molecular weight of solute in solution is through the
sedimentation equilibrium experiment by analytical ultracentrifugation. In the
experiment, when sedimentation and diffusion come to equilibrium, no apparent
movement of solute occurs and its molecular weight can be derived from the equilibrium
concentration distribution”. Because of background absorbance at 220 nm from the
buffer, the initial experiments with 50 11M of both FP and FPK3 peptide in unbuffered
water or under acidic conditions did not provide useful information, so we added a Trp
residue at the C-terminus of the peptide to introduce a chromophore that allowed us to
measure absorbance at 280 nm with less interference. Figure 4a displays the results of the
sedimentation equilibrium experiment at 20 °C for a sample made with 80 11M FPK3W
peptide in 5 mM HEPES pH 7 buffer. A total of 4 such data sets at both 45,000 and
52,000 rpm were included into the data analysis. Each data set gave random variations of
the residuals similar to the one shown, which indicates that the global ﬁt matched the
model for a single Species. With a partial speciﬁc volume of 0.763 ml/g that was

calculated by the method of Durschschlag" and a solvent density of 1.0 g/ml, the optimal

22

ﬁt to all 4 data sets, determined by a non-linear least-squares method, was a molecular
weight of 2610 (2570 — 2650), where the values in parentheses are the 95% conﬁdence
limits from this ﬁtting routine. Since the molar mass of the FPK3W peptide is ~2690, this
result indicates that the peptide is a monomer in the buffer. The experiments were
repeated in another instrument and a similar result was obtained, Showing that the
FPK3W peptide is truly a monomeric peptide in the buffer.

Figure 4b displays the data with 80 MM FPW peptide in the buffer solution.
Peptide pellets were observed at speed above 15,000 rpm, so the displayed data were
obtained at 5,000 rpm. A mean value of M ~ 230,000 was estimated from the fitting
which corresponds to ~ 100 peptide molecules. However, the experimental data has
substantial curvature, which indicates that there is a signiﬁcant distribution of oligomer
sizes.

In terms of their overall solubility and the rates with which they dissolve in
buffer, the peptides can be ordered FPW < FP < FPK3W, FPK3. This order together with
the sedimentation equilibrium data and previous linewidth analysis in solution NMR
experiments suggest that FPW is predominantly associated as large oligomers, and PP is
mainly associated as small oligomers. Because FPK3 contains the same number of
charged residues as FPK3W, we expect that FPK3 is the same as FPK3W and is

predominantly a monomer in the buffer solution.

23

(a)

 

IIIIIIIIIITIITIIIIIIIIIITﬂTrleIIlIIII'IIIIIIIII'

8.4 8.3 8.2 8.1 8.0 7.9 7.8 7.7 ppm

(b)

 

IﬁjlllIllIIII1—IrIrITIII'IIIII—ITIr'jITI'rIIIlIIII]

8.4 8.3 8.2 8.1 8.0 7.9 7.8 7.7 ppm

(0)

 

rIerIII IfII Illj IIII IITI IIII IIII IIII IIIT
I I I I I I I I I I

8.4 8.3 8.2 8.1 8.0 7.9 7.8 7.7 ppm

(d)

 

IIIIIIIIIIIIIIIIIII'IIII'IIIIIIIIIllIIIjIIII'IIIII

8.4 8.3 8.2 8.1 8.0 7.9 7.8 7.7 ppm

Figure 3. Solution amide 1H NMR spectra of (a) FP at pH 6.7, (b) FP at pH 3.7, (c) FPK3
at pH 6.4, and (d) FPK3 at pH 3.7.

24

 

 

Residuals

 

 

A 280

 

 

 

 

 

(b)

In(Azso)

 

. 8.3....

.p

 

 

 

IIIIIIII'IIIIIIIITTIIII

I I I
35.0 35.5 36.0 36.5 37.0 37.5
Radius2 (cm?)

Figure 4. Sedimentation equilibrium experiments of (a) 80 11M FPK3W and (b) 80 BM
FPW in buffer. In (a), the main panel shows the absorbance at 280 nm as a function of the
centrifugal radius after reaching the equilibrium in 20 hours at 52,000 rpm. The best ﬁt to
the model for a single species was obtained with a molecular weight of ~2630 (shown as
a solid line through the experimental points). The upper panel shows the residuals
between the data and the ﬁt. In (b), the logarithm of 280 nm absorbance is plotted as a
function of (radius)2 at 5,000 rpm. The data are represented as dark circles and the dark
dotted and light-dashed lines represent the expected results for FPW 100-mers and

monomers, respectively.

25

Peptide Overall Structure in Solution. The overall secondary structures of the
peptides prior to associating with membrane were evaluated from their CD spectra.
Figure 5 displays the results of four 20 11M fusion peptide constructs in water and 5 mM
HEPES pH 7 buffer. In unbuffered water, both FP and FPK3 peptides form random coil
structures as indicated by their characteristic negative molar ellipticity at wavelengths
below 210 nm’s. The results of FPW and FPK3W peptides are less clear and may be a
mixture of several different secondary Structures. In 5 mM HEPES buffer, the CD spectra
of all four peptides change to more or less the same shape except for their intensities.
Also there is greater noise in the spectra. Efforts to deconvolute each Spectrum into
individual secondary structure components (i.e. (ll-helix, B-sheet, random coil, etc.) were
not successful, which may indicate that there is no overall secondary structure through
the entire peptide.

Since we observed a better-resolved solution NMR spectrum at low pH, we also
examined whether low pH also leads to a better structure elucidation. A series of FP
peptide CD spectra at different pH were obtained by gradually adding small amount of 30
mM HCl solution into the unbuffered 20 BM solution. Figure 6 Shows the result and

demonstrates that the overall structure is insensitive to changes in pH to 3.1.

26

 

[9] 103 deg cmz/ dmol

 

 

 

 

190 200 210 220 230 240 250
Wavelength (nm)

Figure 5. CD spectra of four 20 BM fusion peptide constructs in unbuffed water: FP (I),
FPK3 (H), FPW (III), and FPK3W (IV); and in 5 mM HEPES buffer: FF (1’), FPK3 (H’),
FPW (HI’), and FPK3W (IV ’).

 

 

 

 

 

2
a 1-
E
'O
\
NE 0 4
O
8’
'c -1 ~
co
2
§ -2-
-3 r , , 1 r I , f T I ,
1 90 200 21 0 220 230 240 250

Wavelength (nm)

Figure 6. CD spectra of 20 11M FP peptide in unbuffered water (I), and a series of

acidiﬁed water solutions at pH 5.0 (H), pH 4.1 (TH), and pH 3.1 (IV). The pH is adjusted
by adding a small amount of 30 mM HCl solution.

27

 

Peptide-Induced Fusion of LM Vesicles. In order to test whether the peptide is
fusogenic, measurements were first made on the FP peptide-induced lipid mixing
between vesicles, and aqueous contents leakage from vesicles. Figure 7 demonstrates
both effects and gives results comparable to those obtained by other investigatorsm’ 19. At
low peptide:lipid molar ratios (S 1:80), lipid mixing and leakage only proceed to ~ 10%
of their maximal values. At a peptide:lipid molar ratio of 1:20, fusion and leakage are
much more efficient, which suggests that a vesicle's fusogenicity is related to the number
of incorporated peptide molecules. This is consistent with the “all-or-none” hypothesis
that fusion requires a critical number of fusion peptides per vesiclezo. A model is
demonstrated below and a normal distribution of the number of peptide molecules per

vesicle is assumed. According to the hypothesis, only those vesicles that have the same

 

r f t 1 number or more than the critical

critical local [peptide]

’ \

a number of peptide per vesicle

     
 

 

 

 

 

O\° ’II “\ I .
8 l - (the dotted line) would promote
a- . fusion; i.e., the amount of fusron
8 {I ‘. ------ Low peptide dose

H'gh pept'dem is proportional to the area to the

l

 

 

\‘n--- x right of the dotted line. At low

I

 

[Peptide] per ve5lcle peptide close, this area is rather

small; at high peptide dose, the area increases signiﬁcantly and significantly more fusion
will be observed. This hypothesis also predicts that at constant peptide:lipid molar ratio,
fusion will be greater for larger vesicles because the population of vesicles that have the
required number of fusion peptides increases with larger vesicle size. However, we

observed that at low peptide:lipid molar ratios, the percentage of lipid mixing is

28

independent of vesicle size between 100 nm and 400 nm. This result suggests that instead
of local number, it may be the local concentration of membrane-associated peptide which

is the critical factor. This concentration is constant with respect to vesicle size.

 

80

3

% Lipid Mixing
8

 

 

 

 

 

 

 

 

20 i I,
A V _ f .~ A—“vv.‘-‘:VAV-'L~_A_‘ : Adv—#AWAW
O I TI I
0 1 00 200 300 400
Time (sec)
60
III
a. 40 -
(U
x
CB
0)
_l
o\° 20 i
IV
0 MW
0 100 200 300 400
Time (sec)

Figure 7. FP-induced lipid mixing (top) and FP-promoted aqueous contents leakage
(bottom) as observed by ﬂuorescence assays at 37 °C. For the lipid-mixing assay,
150 uM LM-3 lipid was extruded into 100nm diameter vesicles. Lipid mixing was
induced with (I) 7.5 1.1M and (H) 1.5 uM FP. For the aqueous contents leakage assay,
80 11M LM-3 lipid was extruded into 100 nm diameter vesicles loaded with ANTS/DPX.

Leakage was promoted by PF at (IH) 4 11M and (IV) 1 uM concentrations.

29

Lipid mixing assays were also carried out to assess the fusogenicities of the four
fusion peptide constructs. Figure 8 displays plots of M, vs. time for each of the constructs
with peptide:lipid molar ratios of 1:100 and 1:50. All peptides induce signiﬁcant lipid
mixing. At a peptide:lipid molar ratio of 1:100, the ordering of Mf values for the different
peptides is FPW < FPK3, FPK3W < FP while at a peptide:lipid molar ratio of 1:50, the
ordering is FPW < FPK3W < FP, FPK3. Overall, FPW appears to be the least fusogenic
while FP is the most fusogenic. As discussed in peptide aggregation section, the likely
solution oligomerization states of the peptides are: FPK3W, FPK3 - monomeric; FP —
mixture of small and large oligomers; and FPW - large oligomers. Thus, there does not
appear to be a linear relationship between solution oligomerization state and fusogenicity,
although it may be that small oligomers (FP) promote lipid mixing more efﬁciently than

monomers (FPK3W, FPK3) or large oligomers (FPW).

30

 

40

30‘

. ' l , l. . -
wrimliimmtllll'l'lW‘ll'” "Ml“!!!

 

°/e Llpld Mlxmg
8

10-
I

 

 

 

o 100 200 300 400 500 600
Time (sec)

Figure 8. Lipid mixing induced by four fusion peptide constructs. In the assay, 150 1.1M
1M3 was extruded into 100 nm diameter vesicles and mixing was induced by (I) 3 11M
and (I’) 1.5 uM FP; (H) 3 uM and (H’) 1.5 uM FPK3; (H1) 3 11M and (HI’) 1.5 MM
FPK3W; and (IV) 3 11M and (IV’) 1.5 uM FPW.

31

CONCLUSION

The fusion peptide is derived from the fusogenic region of the I-HV -1 viral
envelope protein gp41. Numerous studies have shown that the free peptide has an integral
role in the fusion process. Hence the study of the free fusion peptide provides a good
model to understand the mechanism of HIV -1 viral fusion with the host cells.

Before we investigate the peptide structure in the membrane environment using
solid-state NMR, we need to ensure that the system that we examine is biologically
relevant. This is important because the conﬂicting structural results of fusion peptide
studies from other groups suggest that the sample preparation may have a signiﬁcant
effect on the experimental observations. In this study, we conducted a series of
experiments to verify the validity of our measurements.

The oligomerization of fusion peptide in aqueous solution was the first issue we
considered. There was evidence that in neutral pH phosphate buffer the peptide forms
large fibrillar aggregates that reduce the fusion activity. After testing a series neutral
buffer solutions, we found that 5 mM HEPES buffer at pH 7 was a good choice Since
>90% of peptide was soluble at the concentration used for solid-state NMR samples.
Later, a more soluble construct was utilized by adding a lysine-rich “host sequence” at
the C-terminus of the peptide. The oligomerization status of the peptides was estimated
by solution NMR and sedimentation equilibrium experiments. The results show that most
of the fusion peptide constructs are monomeric or form small oligomers of 10 ~ 20
molecules, except for FPW which forms oligomers over 100 molecules. CD
measurements suggest that there is no global structure for the entire peptide in the

absence of membrane.

32

The fusion activity of the peptide was the next issue we examined. The results of
FP-induced lipid-mixing and aqueous contents leakage assays were comparable to those
obtained by other investigators. We also compared the fusogenicities among the four
fusion peptide constructs. While FP, FPK3 and FPK3W have about the same fusion
activity, there was signiﬁcantly less lipid mixing induced by FPW peptide. Considering
the sizes of the peptides in solution, FPK3W and FPK3 are monomeric, FP forms small
oligomers, and FPW forms large oligomers, there is not a linear relationship between the
solution oligomerization state and fusogenicity, although it appears that the small
oligomers (FP) promote lipid mixing more efﬁciently than monomers (FPK3, FPK3W) or
large oligomers (FPW).

In summary, a series of experiments confirm that we can prepare and examine
biologically relevant samples and that the peptide structure we obtain should reﬂect at

least the ﬁnal stage of the fusion process.

33

REFERENCE

10.

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

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

Aloia, R. C., Tian, H., and Jensen, F. C. (1993) Lipid composition and ﬂuidity of
the human immunodeficiency virus envelope and host cell plasma membranes.
Proc Natl Acad Sci U SA 90, 5181-5185.

Hope, M. J ., Bally, M. B., Webb, G., and Cullis, P. R. (1985) Production of large
unilamellar vesicles by a rapid extrusion procedure - characterization of size
distribution, trapped volume and ability to maintain a membrane-potential.
Biochim. Biophys. Acta 812, 55-65.

Hore, P. J. (1983) Solvent suppression in fourier-transform nuclear magnetic
resonance. J. Magn. Reson. 55, 283—300.

Struck, D. K., Hoekstra, D., and Pagano, R. E. (1981) Use of resonance energy
transfer to monitor membrane fusion. Biochemistry 20, 4093-4099.

Ellens, H., Bentz, J ., and Szoka, F. C. (1985) H+- and Ca2+-induced fusion and
destabilization of liposomes. Biochemistry 24, 3099-3106.

Chang, D. K., Chien, W. J., and Cheng, S. F. (1997) The FLG motif in the N-
terrninal region of glucoprotein 41 of human immunodeficiency virus type 1
adopts a type-I beta turn in aqueous solution and serves as the initiation site for
helix formation. Eur J Biochem 247, 896-905.

Slepushkin, V. A., Andreev, S. M., Sidorova, M. V., Melikyan, G. B., Grigoriev,
V. B., Chumakov, V. M., Grinfeldt, A. E., Manukyan, R. A., and Karamov, E. V.
(1992) Investigation of human immunodeficiency virus fusion peptides. Analysis
of interrelations between their structure and function. AIDS Res Hum Retroviruses

8, 9-18.

Pereira, F. B., Goni, F. M., Muga, A., and Nieva, J. L. (1997) Permeabilization
and fusion of uncharged lipid vesicles induced by the HIV-1 fusion peptide

adopting an extended conformation: dose and sequence effects. Biophys. J. 73,
1977-1986.

34

11.

12.

13.

14.

15.

16.

17.

18.

19.

20.

Han, X., and Tamm, L. K. (2000) A host-guest system to study structure-function
relationships of membrane fusion peptides. Proc. Natl. Acad. Sci. U. S. A. 97,
13097-13102.

Kliger, Y., Aharoni, A., Rapaport, D., Jones, R, Blumenthal, R., and Shai, Y.
(1997) Fusion peptides derived from the HIV type 1 glycoprotein 41 associate
within phospholipid membranes and inhibit cell-cell Fusion. Structure- function
study. J. Biol. Chem. 272, 13496-13505.

Yang, J ., Gabrys, C. M., and Weliky, D. P. (2001) Solid-state nuclear magnetic
resonance evidence for an extended beta strand conformation of the membrane-
bound HIV -1 fusion peptide. Biochemistry 40, 8126-8137.

Cavanaugh, J., Fairbrother, W. J., Palmer, A. G., and Skelton, N. J. (1996)
Protein NMR Spectroscopy Principles and Practice, Academic Press, San Diego.

Connelly, G. P., Bai, Y., Jeng, M. F., and Englander, S. W. (1993) Isotope effects
in peptide group hydrogen exchange. Proteins: Struct. F unct. Genet. 1 7, 87-92.

Bowen, T. J. (1970) An Introduction To Ultracentrtﬁtgation, John Wiley and Son,
London.

Durschschlag, H. (1986) Specific volumes of biological macromolecules and some
other molecules of biological interest, Springer-Verlag, Berlin.

Cantor, C. R., and Shimmel, P. R. (1980) Biophysical Chemistry, W. H. Freeman,
New York.

Peisajovich, S. G., Epand, R. F., Pritsker, M., Shai, Y., and Epand, R. M. (2000)
The polar region consecutive to the HIV fusion peptide participates in membrane
fusion. Biochemistry 39, 1826-1833.

Nieva, J. L., Nir, S., and Wilschut, J. (1998) Destabilization and fusion of

zwitterionic large unilamellar lipid vesicles induced by a beta-type structure of the
HIV-1 fusion peptide. J. Liposome Res. 8, 165-182.

35

CHAPTER III

CONFORMATION AL STUDIES OF THE MEMBRANE-ASSOCIATED FUSION

PEPTIDE BY SOLID-STATE ONE-DIMENTION AL NMR MEASUREMENTS

36

BACKGROUND

For biophysical studies, including solid state NMR spectroscopy, it is often easier
to work with smaller systems than the full envelope proteins. In this project, we work
with peptides containing the ~ 20-residue N-terrninal fusion peptide domain of the gp41
envelope protein. There is body of biochemical and biophysical evidence which
demonstrate that the free ~ 20-residue fusion peptide by itself has biological relevance in
a similar manner as the intact protein, and therefore provides important information about
the mechanism of viral envelope protein induced membrane fusion.

In the previous chapter, we examined the biological activity of the synthetic
fusion peptide. In this chapter, we ﬁrst verified that the structure we would probed with
solid-state NMR techniques is indeed the membrane-associated structure. Furthermore,
since we prepared our sample in the similar way as we conducted the fusion assay, the
structure probed would also represent the one at least at the end stage of fusion.

Because of the low natural abundance of interesting nuclei such as 13C and 15N,
we generally need specific isotopic labeling to obtain adequate signal-to-noise. Isotopic
labeling is in fact advantageous because the signals from the isotopic labels dominate the
spectrum and allow us to focus on the structure in the vicinity of the labeled nuclei,
which makes solid-state NMR a powerful structural approach for residue-specific
structural studies.

It is well known that the useful information contained in an NMR spectrum comes
from the different local field experienced by each nuclear spin. In solution NMR, because
of the rapid molecular tumbling, the nucleus experiences a homogenous local ﬁeld and a

very sharp peak is observed. In solid state, this molecular tumbling is significantly

37

slower, and the orientation dependence is not averaged out. The local field is dominated
by the chemical shift anisotropy (CSA) and by the dipole-dipole coupling between
spatial-nearby nuclei. The CSA is the dependence of chemical shift on the relative
orientation between the functional group and the external magnetic ﬁeld. These two
factors will signiﬁcantly broaden the linewidth. Fortunately, we can apply magic angle
spinning (MAS) to improve the resolution by spinning the sample about an axis tilted at
54.7° relative to the external magnetic ﬁeld direction. As a result, instead of a broad peak,
the spectrum will contain a much narrower isotropic peak and a series of spinning
sidebands separated by the spinning frequency. By examining the linewidth and chemical
shift of the isotropic peak, we can obtain information about the structural heterogeneity

and the secondary structure near the labeled nuclei in the peptide.

METHODS

Solid-State NMR Sample Preparation. Samples were typically prepared using
0.01% (w/v) N aN3 in either unbuffered water or 5 mM HEPES buffer (pH 7.0). Samples
were prepared by three different methods. Method 1: Mixtures of peptide (15 - 150 1.1M)
and either a lipid dispersion or vesicles (1 - 30 mM) in 4 ml or 30 ml total volume were
kept at room temperature overnight to ensure maximum peptide/lipid binding.
Subsequent centrifugation of the peptide/lipid complex, typically at 100,000 - 150,000 x
g for two to four hours, pelleted down the complex and left unbound peptide in the
supernatant. The peptide/lipid pellet formed after ultracentrifugation was transferred by
spatula to a 7 mm or 6 mm diameter MAS NMR rotor. Method 2: Mixtures of peptide

(1 -10mM) and lipid dispersion (ZOO-400 mM) were mixed to form ~ 200 pl total

38

 

volume. The mixtures were transferred directly to the MAS rotor. Method 3: Peptide and
lipid were codissolved in 6 ml mixture of chloroform:triﬂuoroethanol:hexafluoropropanol
(322:2 v/v). The solvent was removed under a stream of nitrogen and dried in vacuum
overnight. The mixture of peptide and lipid was then hydrated with 5 mM pH 7.0 HEPES
buffer followed by homogenization with ten freeze-thaw cycles. The sample was diluted
with buffer to ~ 35 ml total volume and kept at room temperature overnight. Subsequent
centrifugation of the peptide/lipid complex at 100,000 x g for four hours pelleted down
the complex and left unbound peptide in the supernatant. The peptide/lipid pellet formed
after ultracentrifugation was transferred by spatula to a 6 mm diameter MAS rotor.

Measurement of Peptide/Lipid Binding. For solid-state NMR preparation method
1, peptide/lipid binding was determined through BCA assay of the peptide solution prior
to addition of lipid and after ultracentrifugation. After ultracentrifugation, all lipid was
pelleted and there was no interference from unpelleted lipid in the BCA assay. Controls
were run with peptide-only samples to ensure that the peptide alone did not pellet.

For solid-state NMR preparation method 2, ~ 800 pl water was added to the
peptide/lipid dispersion, the sample vortexed, and the peptide/lipid mixtures were spun
down at 10,000 - 14,000 x g. The unbound peptide concentration in the supernatant was
measured by BCA assay. Under these centrifugation conditions, the entire lipid was
pelleted, and there was no interference from unpelleted lipid in the BCA assay.

One-Dimensional (ID) Solid-State NMR Experiments. Measurements were made
on two 9.4 T spectrometers, Varian VXR and Varian Inﬁnity Plus (Palo Alto, CA). For
the Varian VXR spectrometer, a double resonance MAS probe was used with 7 mm

diameter rotors. The NMR detection channel was tuned to '3C and the decoupling

39

channel was tuned to 1H. Experiments were carried out using a spinning speed between
3.5 and 4.0 kHz. Application of 1 ms of cross-polarization (CP) at 47 kHz was followed
by signal detection with decoupling at 75 kHz. The recycle delay was 0.5 s. For typical
samples containing ~ 0.5 “mole of labeled peptide, a single spectrum was the average of
80,000 - 160,000 scans. All chemical shifts were externally referenced to the methylene
carbon resonance of adamantane (38.2 ppm). Spectra were processed with 25 Hz line
broadening and all refer to the isotropic carbonyl regions.

Samples were cooled to —50 °C either slowly in the NMR probe or by fast
freezing in liquid nitrogen. Both freezing techniques gave comparable spectra. Although
physiological fusion occurs at 37 °C, we chose this low temperature because relative to
room temperature, there is about three times greater signal per 13C at the lower
temperature. This is likely due in part to more efﬁcient cross-polarization (CP) because of
reduced motion at lower temperature. 13C backbone chemical shifts were similar at both
temperatures which suggests that cooling the sample does not cause signiﬁcant peptide
structural changes1 .

In most cases, the samples contained one or two 13C peptide carbonyl backbone
labels and the NMR signals from these labeled sites were of the greatest interest. The
labeled site signals were often unresolved from large natural abundance signals from lipid
carbonyls and from unlabeled peptide carbonyls. To uniquely observe the labeled site
signals, spectra were also taken of unlabeled peptide/lipid samples that were prepared in
the same manner as those with labeled peptide. Subtraction of the natural abundance

spectra from the labeled spectra yielded the difference spectrum of the labeled site(s). For

40

most samples, subtraction was considered to be optimal when natural abundance lipid
glycerol signal at ~ 130 ppm was minimized in the difference spectrum.

A more efﬁcient approach of ﬁltering out the natural abundance signals involves
the synthesis of directly bonded l3C carbonyl and 15N amide labeled peptides and utilizes
the technique of rotational-echo double-resonance spectroscopy (REDOR). The details of
this technique will be discussed in chapter ﬁve. In general, REDOR ﬁltering has several
advantages including: (1) better ﬁltering of natural abundance signals; (2) requiring one
rather than two samples; and (3) better compensation for spectrometer and probe driftsz.

The REDOR filtering measurements were taken on the Varian Inﬁnity Plus
spectrometer using a triple resonance MAS probe equipped with 6 mm diameter rotors.
The NMR detection channel was tuned to 13C, the decoupling channel was tuned to 1H,
and the third channel was tuned to 15N. 13C and 15N chemical shifts were externally
referenced to the resonance of methylene carbon in adamantane at 38.2 ppm and
(NH4)2SO4 at 20 ppm, respectively. Experiments used a MAS frequency of 8.000 :t 0.002
Hz. In REDOR experiments, application of 1 ~ 2 ms of CP at 50 kHz was followed by a
1.0 ms dephasing period and then direct l3C detection. A single 50 kHz l3C refocusing 1:
pulse was placed at the center of the dephasing period and 1H TPPM decoupling of 65
kHz was applied during both dephasing and detection3. The '3 C transmitter was set to 155
ppm and the 15N transmitter was set to 115 ppm. For the $1 acquisition, the dephasing
period contained a 40 kHz 15N 1t pulse at the middle and end of each rotor cycle, while
the So acquisition did not contain these pulses. XY—8 phase cycling was used for the 15N
pulses4' 5. The REDOR filtered spectrum was processed with 50 Hz line broadening of the

difference FID (SO — 51).

41

RESULTS AND DISCUSSION

Peptide binding to Lipids. Since we can prepare fusogenic fusion peptide
solutions that do not pellet under ultracentrifugation, it is possible to examine the binding
strength of peptide to lipid. This is an important issue because the NMR signals are
observed for fusion peptide in all types of environment, i.e. membrane-bound, aggregated
or free. Since we are particularly interested in the membrane-bound peptide, the
contribution from the other two sources must be eliminated. The aggregated peptide can
be avoided by selecting an optimal buffer solution as described in the previous chapter.
The unbound free peptide can be separated by centrifugation after mixing peptide and
lipid since the membrane-bound peptide will stay in the pellet with the lipid. We can then
estimate the percentage of binding by comparing the initial peptide concentration to that
found after addition of lipid and subsequent centrifugation.

Table 2 clearly illustrates that there is strong binding between fusion peptide and
membrane lipids. Binding to LM is quantitative and binding to pure neutral or negatively
charged lipids is 75 — 90%. Therefore we can use centrifugation to ensure that we are

examining only membrane-bound fusion peptides.

42

Table 2. Fusion Peptides Binding to Membranes.

 

 

Lipid Peptide (mM): Lipid (mM) ratioa % Bindingb
DT PC 0.50: 100 75
DMPC 0.50: 100 80
POPG 0.50: 100 90
DTPC 0.12525 80
LM-l 0.125225 99
LM-2 0.125: 12.5 95
LM-3 0.125: 10 97
LM-3c 0. 125: 10 96
LM—3 0.014:1.1 88
LM-3d 0.014:1.1 88

 

a The binding percentage for the ﬁrst three samples were estimated by method 2 and
those of the rest of the samples were estimated by method 1.

b Estimated error: i 10%.

° Measurements were made with lipid dispersion samples except for these last three
entries which was made with 100 nm diameter vesicles.

d Measurements were made with FP peptide except for this entry which was made with
FPK3 peptide.

43

Dependence of ID NMR Spectra on Sample Conditions. Figure 9 displays the 1D
MAS spectra of different 13C carbonyl labeled peptides. Generally, the carbonyl
linewidth is a marker of the structural heterogeneity in the vicinity of the labeled nucleus
or nuclei. For example, Figure 9a displays the spectrum of polycrystalline AGG-(A1G2)C
while Figure 9b displays the spectrum for an unstructured singly 13C carbonyl labeled
epitope peptide HGRVGIYFGMK in frozen solution. The observed full-width-at-half-
maximum (FWHM) linewidths of ~ 1 ppm and ~ 6 ppm mark the two extrema of highly
structured and unstructured residues, respectively.

Figures 9c and 9d respectively display the 1D spectra of FP-(L7F8)C in 7.5 mM
frozen solution and hydrated lipid DTPC environments (1:20 peptide:lipid molar ratio).
The lipid sample in 9d was prepared by method 2 using 100 pl of the 9c solution. DTPC
was initially chosen for these studies because it is ether- rather than ester-linked and
hence has no natural abundance carbonyl background. Thus, in Figure 9c and 9d the
labeled carbonyl carbons contributed 90% of the observed intensity. The signals from the
two l3C labeled carbonyls were unresolved. In the presence of DTPC, the linewidth is ~ 2
ppm FWHM and is narrower than that found in frozen solution. These data indicate that:
(1) the L7F8 region of FF interacts with lipid; and (2) in the presence of lipid, the L7F8

region is reasonably well structured, although not crystalline.

(a) (b)

 

 

180 170 160 ppm 180 170 160 ppm

(C) (d)

 

Lm

180 170 160 ppm 180 170 160 ppm

Figure 9. 13C solid-state NMR spectra of (a) 5% polycrystalline AGG-(A1G2)C peptide;
(b) 20 mM frozen aqueous solution of HGRVGIYFGMK-FBC epitope peptide; (c) 7.5
mM frozen aqueous solution of FP-(L7F8)c; and (d) FP-(L7F8)c/DTPC made by
preparation method (2) at 1:20 peptide:lipid molar ratio. The AGG spectrum was taken at
room temperature and the other spectra were taken at —50 °C.

45

The dependence of the 1D spectrum on lipid composition was also investigated.
As displayed in Figure 10, signiﬁcantly different spectra were obtained for FP-F8c for
samples made either with (a) pure DTPC or with (b) LM-l dispersions. The 10a and 10b
samples were made using preparation method 1 at 1:200 peptide:lipid molar ratio. In 10a,
the raw spectrum is displayed while in 10b, a difference spectrum between the
FP-F8C/LM-1 sample and a pure LM-l sample is displayed. Figure 10b is best described
by a single component with FWHM linewidth ~ 2.7 ppm, and reﬂects a relatively
structured F8 residue. In Figure 10a, the spectrum is best described by two partially
resolved components which likely correspond to at least two distributions of structures,
each with significant population. It is unlikely that either component in 10a has a large
contribution from the ~ 20% natural abundance peptide carbonyl signal because an
unlabeled FP/DTPC sample at 1:80 peptide:lipid molar ratio generated a broad carbonyl
absorption which ranged from 167 ppm to 174 ppm.

The different spectral lineshapes in Figure 10a and 10b demonstrate that there is
some dependence of peptide structure on lipid composition. Although we have not yet
understood the precise lipid compositional factors which cause the structural differences
reﬂected in the two spectra, we have observed spectra similar to 10b at ~ 1:100
peptide:lipid molar ratio for samples made with LM-2 and LM-3. These mixtures differ
from LM-l only through addition of sphingomyelin, and sphingomyelin and PI,
respectively. Two representative FP-F8C/LM-3 difference spectra are displayed in
Figure 10c and 10d, and were obtained from 1:80 peptide:lipid molar ratio samples made

with preparation method 1 using respectively LM-3 lipid dispersion or 100 nm diameter

46

LM-3 vesicles. These difference spectra were obtained from subtraction between FP-F8c
and unlabeled FP samples and thus are nearly pure F8 signals.

The spectra from LM samples appear to be fairly independent of initial peptide
concentration in the range of 12 - 120 “M and also independent of the peptide:lipid molar
ratio in the range of 1:200 to 1:20. For example, little difference was found between
Figure 10b and 10c in which the samples were respectively made with 120 “M and
12 11M initial peptide concentrations and 1:200 and 1:80 peptide:lipid molar ratios.
Spectra from samples made with FP-(GlOFll)C/LM-3 at 1:100 (Figure 10e) and 1:20
(Figure 10f) peptide:lipid molar ratio are also comparable. Although all samples were
made with pH 7.0 HEPES, similar spectra were also obtained when unbuffered water
(pH ~ 6) was used.

Spectra of samples made with DTPC also had little dependence on initial peptide
concentration but had some dependence on peptide:lipid molar ratio. For example, with
both 1:200 peptide:lipid molar ratio, the two-component spectrum in Figure 10a is similar
in appearance to the one in Figure 11a from FP-(L7F8)c/DTPC sample. For this latter
sample, the initial peptide concentration was 2.5 mM, which is more than an order-of-
magnitude greater than the 120 uM concentration used for sample 10a. However, as
displayed in Figure 9d, a sample made with DTPC, 7.5 mM initial peptide concentration
and 1:20 peptide:lipid molar ratio gives a signiﬁcantly different spectrum than that in
Figure 10a. The single-component 9d spectrum closely resembles the Figure 10b-d

spectra of samples made with LM.

47

(a) (b)

 

 

 

180 170 160 ppm 180 170 160 ppm
(C) (d)

180 170 160 ppm 180 170 160 ppm
(6) (f)

 

180 170 160 ppm 180 170 160 mm

Figure 10. 13C solid-state NMR spectra of samples made from FP-F8c bound to (a) DTPC
dispersion, (b) LM-l dispersion, (c) LM-3 dispersion, and (d) 100 nm diameter LM-3
vesicles with a peptide:lipid molar ratio of (a), (b) 1:200 and (c), (d) 1:80; and spectra of
samples made from FP-(G10F11)c bound to LM-3 with a peptide:lipid molar ratio of (e)
1:100 and (f) 1:20.

48

In addition, spectra showed little dependence on how the samples were prepared.
Figure 11 compares results using the three preparation methods. As discussed in the
description of peptide binding strength measurements, methods 1 and 2 differ in that 1
allows us to eliminate the contribution from unbound peptide. However, if the peptide
binds strongly to the membrane and the amount of unbound peptide can be neglected,
there will be little difference between these two methods. This is indeed what we found
when we compared the spectra of samples made with FP-(L7F8)c/DTPC at 1:200
peptide:lipid molar ratio using method 1 (Figure 11a) and method 2 (Figure 11b). The
similarity of the two spectra is consistent with our binding measurements (cf. Table 2)
and therefore suggests that the spectra we collected from samples prepared by method 2
are also dominated by the membrane-associated peptide structure. Method 3 examines
whether pre-rnixing the peptide and lipid in organic solvent has any impact on the final
structure of the membrane-associated peptide. Figures 11c and 11b are the REDOR
ﬁltered spectra of samples made with FPK3-F8CL9N/LM-3 at a 1:100 peptide:lipid molar
ratio, and show that as long as the organic solvent is removed and the peptide/lipid
mixture is hydrated, the membrane-associated peptide prepared by method 3 will adopt a
similar structure as obtained by method 1. Besides sample preparation methods, little
difference was found whether samples were slowly cooled during ~ 1 minute while
spinning in the NMR probe or were frozen in ~ 1 sec by immersion of the sample rotor in
liquid nitrogen. Similar spectra were also obtained after freeze/thaw cycling and after
incubation of the sample at 37 °C for 1.5 hours (data not shown). All these results
indicate that the observed membrane-associated peptide structure is very stable and likely

represents the thermal-equilibrium favored structure.

49

 

 

(a) :J
W
130 170 150 PP!“ 180 170 160 ppm

l
W/

160 ppm 80 170 160 ppm

 

 

 

Figure 11. 13C solid-state NMR spectra of samples made from FP-(L7F8)c/DTPC with a
peptide:lipid molar ratio of 1:200 by (a) method 1 and (b) method 2; and spectra of
sample made from FPK3-F8CL9N/LM-3 with a peptide:lipid molar ratio of 1:100 by (c)
method 1 and (d) method 3.

In summary, our 1D spectral data suggest: (I) there is some dependence of
peptide structure on lipid composition; (2) little dependence of structural distributions on
initial peptide concentration for either LM or DTPC samples; (2) little dependence of
structural distributions on peptide:lipid molar ratio for samples made with LM; (3)
signiﬁcant dependence of structural distributions on peptide:lipid molar ratio for samples

made with DTPC; and (4) once the peptide binds to the membrane, the peptide structure

is the same with different preparation methods and is stable under various thermal

50

conditions. These data strongly suggest that fusion peptide/lipid samples should be
prepared with LM rather than single lipids.

ID NMR Spectral Scanning down the FP23 Backbone. Difference spectra for
LM-bound FP peptide are displayed in Figure 12a-h and are composed of carbonyl
signals from labeled Ala-1, Val-2, Phe-8, Gly-10, Phe-ll, Ala-14, Ala-15, and Ala-21,
respectively. The samples were made by preparation method 1 with a peptide:lipid molar
ratio of 1:100. The Val-2, Phe-8, Phe-ll, Ala-15, and Ala-21 spectra are from samples
made with LM-2 while the Ala-1, Gly-10, and Ala-14 spectra are from samples made
with LM-3. Signals for labels between Ala-l and Ala-15 all contained a relatively sharp
feature with linewidth of 2 - 3.5 ppm, which suggests that a large population of
membrane-bound peptide molecules share a common structure at each labeled site. The
approximately symmetric lineshapes for Phe-8, G1y-10, Phe-ll, and Ala-14 suggest that
the peptide center has a high degree of structural order in a membrane environment. By
contrast, the ~ 5 ppm linewidth of FP-A21C indicates a higher degree of structural
disorder at this site. Overall, these data are consistent with a model of insertion of the
N-terminal and central residues of the peptide into the membrane and the C-terminus
outside the membrane. In this model, there is greater structure for the residues within the
membrane than outside the membrane. This is a reasonable premise because within the
membrane, hydrogen bonding must be either inter- or intra-peptide, whereas outside the
membrane, peptide-H20 hydrogen bonds may be present and can allow for greater

structural disorder.

51

 

 

l;

 

185 170 ppm 185 170 ppm
185 170 ppm 185 170 ppm

 

KM

 

 

l

185 170 ppm 185 170 ppm
185 170 ppm 185 170 ppm

Figure 12. 13C solid-state NMR difference spectra of LM-bound (a) FP-Alc, (b) FP-VZC,
(c) FP-F8C, (d) FP-GlOc, (e) FP-Fl 1C, (f) FP-Al4c, (g) FP—A15c, and (h) FP-A21c.
Samples were prepared by method 1 with a peptide:lipid molar ratio of 1: 100.

52

Correlation of Carbonyl Chemical Shifts with Secondary Structure. Secondary
structure at the different labeled carbonyl carbons in the membrane-bound fusion peptide
was characterized using the known experimental correlation between secondary structure
and NMR chemical shift6. For peptide or protein carbonyl carbons, helical secondary
structure generally correlates with downfield (to higher ppm) chemical shift while strand
structure correlates with upﬁeld (to lower ppm) chemical shift. Table 3 lists the peak
experimental shifts for different carbonyl nuclei in the more structured N-terminal and
central regions of the LM-bound fusion peptide, the typical ranges of chemical shifts
observed for different secondary structures for that residue type, and the assignment of
the local secondary structure based on these data. All of the peak chemical shifts are most

consistent with a B strand structure.

Table 3. Structural Correlations from Carbonyl Chemical Shift Analysis.

 

 

Labeled Experimental or Helical CS B Strand CS CS Predicted

Residuea CS (ppm) Range (ppm)b Range (ppm)b Structure
Ala-l 171.9 175.78 - 178.42 172.28 - 175.30 B strand
Val-2 172.2 173.97 — 176.73 171.11 — 173.89 B strand
Phe-8 170.6 173.45 — 176.21 170.32 — 173.58 B strand
Gly-10 168.4 171.98 — 174.44 168.67 -— 171.83 B strand
Phe-ll 170.3 173.45 — 176.21 170.32 — 173.58 B strand
Ala-14 173.4 175.78 — 178.42 172.28 — 175.30 B strand
Ala-15 173.2 175.78 — 178.42 172.28 — 175.30 B strand

 

a The membrane-associated FP was 13C carbonyl labeled with a peptide:lipid molar ratio

of 1:100.

b Each range corresponds to about 90% of the residues found in that conformation.

53

Structure Similarity of Membrane-Associated FP, FPK3, FPW, and FPK3W. The
previous chapter has demonstrated the impact on biological function of different peptide
oligomer sizes in aqueous buffer. The results from the four fusion peptide constructs
showed that FP, FPK3 and FPK3W have comparable fusogenicity while FPW has much
less fusogenicity probably due to the large aggregates it forms in aqueous solution. It is
therefore interesting to see whether their membrane-associated structures are comparable.
Figure 13 displays the REDOR-ﬁltered difference spectra of sample made with LM-3
associated (a) FP-F8CL9N, (b) FPK3-F8cL9N, (c) FPW-F8CL9N, and (d) FPK3W-F8cL9N
at ~ 1:80 peptide:lipid molar ratio. Because of the directly bonded l3C-ISN spin pairs in
the peptides, the Phe-8 carbonyl signals dominate the displayed REDOR-ﬁltered
difference spectra, with quantitative attenuation of the natural abundance 13C signals from
lipid and peptide. In all spectra, the FWHM linewidths are ~ 2.5 ppm, which is consistent
with a relatively narrow conformational distribution at Phe-8, and the peak chemical
shifts are ~ 171 ppm, which is consistent with non-helical local structure6. The overall
similarity of the lipid mixing data and NMR spectra for the FP, FPK3 and FPK3W
peptides suggests comparable interactions with LM-3 membranes, and the possibility that
FPK3 and FPK3W can be used in place of FF in later experiments.

The soluble oligomerization state varies substantially among the four peptides, so
the similarity of the spectra indicates that the membrane-associated peptide secondary
structure is approximately independent of this oligomerization state. In addition, the
similar spectra indicate that the differences in Mf values among the different peptides are
not due to large variations in secondary structure near Phe-8, i.e. non-helical vs. helical

SII'UCIUI'C.

54

 

 

W (m
I l 7 l I I T I

190 180 170 160 190 180 170 160
ppm

my (:NAmM/V
F I l l l l l l

190 180 170 160 190 180 170 160
ppm

 

 

Figure 13. 13C solid-state NMR spectra of membrane-associated fusion peptide samples
with (a) FP-F8CL9N, (b) FPK3-F8CL9N, (c) FPW-F8cL9N, and (d) FPK3W-F8cL9N. Each
sample contained ~ 0.3 umol peptide and 40 umol LM3. The displayed REDOR-ﬁltered
difference spectra are dominated by the Phe-8 carbonyl signals.

55

CONCLUSION

In this chapter we examined the binding strength of the fusion peptide and
membrane lipid. These results allow us to use ultracentrifugation to separate membrane-
associated peptide from unbound peptide so that we only observe the structure of peptide
in the membrane environment. In general, when the peptide binds to the lipid, the
structure is stable and insensitive to various preparation methods and thermal conditions.
There is some dependence on lipid composition. Peptide binding to DTPC is weaker than
to a lipid mixture whose composition is close to that of HIV-1 virus infected host cells.
The peptide also has greater structural homogeneity with LM, and peptide structure was
invariant over a ~1220-200 peptide:lipid molar ratio range. This strongly suggests that
future fusion peptide studies should incorporate these kinds of lipid mixtures.
Investigation of peptide structure across the backbone shows that the N-terminal and
central regions of the peptide are well structured while the C-terminus is more ﬂexible.
Furthermore, correlating the carbonyl chemical shifts with secondary structure suggests
that the N-terminal and central regions have a B strand secondary structure. A more
precise secondary structure measurement will be discussed in next chapter using two-

dimensional exchange spectroscopy.

56

REFERENCE

1. Bodner, M. L., Gabrys, C. M., Parkanzky, P. D., Yang, J., Duskin, C. A., and
Weliky, D. P. Temperature dependence and resonance assignment of 13C NMR
spectra of selectively and uniformly labeled fusion peptides associated with
membranes. Magn. Reson. Chem, in press.

2. Yang, J., Parkanzky, P. D., Bodner, M. L., Duskin, C. G., and Weliky, D. P.
(2002) Application of REDOR subtraction for ﬁltered MAS observation of
labeled backbone carbons of membrane—bound fusion peptides. J. Magn. Reson.
159,101-110.

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

4. Gullion, T., Baker, D. B., and Conradi, M. S. (1990) New, compensated Carr-
Purcell sequences. J. Magn. Reson. 89, 479-484.

5. Gullion, T., and Schaefer, J. (1991) Elimination of resonance offset effects in
rotational-echo, double-resonance NMR. J. Magn. Reson. 92, 439-442.

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

57

CHAPTER IV

CONFORMATIONAL STUDIES OF THE MEMBRANE-ASSOCIATED

FUSION PEPTIDE BY SOLID-STATE TWO-DINIENSION AL

EXCHANGE NMR MEASUREMENTS

58

BACKGROUND

For unoriented static samples, 2D exchange spectroscopy can determine the
relative orientations of two nearby functional groups that are speciﬁcally labeled at the
same kind of nuclei (”C carbonyls in our case). In the experiment, the magnetization
evolves under the anisotropic chemical shift (which is simply the dependence of chemical
shift on the relative orientation between the functional group and the external magnetic
ﬁeld) of one of the labeled sites during t1, diffuses to the second site during the exchange
period 'c, and evolves under the anisotropic chemical shift of the second site during t2. If
all of the molecules in the sample have the same local structure near the two labeled sites,
then the chemical shift anisotropy (CSA) tensors of the two sites will have a well-deﬁned
relative orientation which will be manifested as correlation between the anisotropic
chemical shifts of the two nuclei. This correlation will produce an off-diagonal intensity
pattern in the 2D exchange spectrum which is distinctively characteristic of the relative
CSA tensor orientation and the particular local structure. Because of the low probability
of nearby natural abundance 13C/13C spin pairs, the off-diagonal intensity will be
dominated by these labeled-site signals. In order to relate the off-diagonal pattern to local
structure, simulated 2D spectra must be generated as a function of local structure and
comparison made between the simulated and experimental spectra. The simulations
require knowledge of the CSA principal values of each of the labeled sites as well as the
orientation of the CSA principal axis system of each labeled site relative to its chemical
bonds. In many cases, the principal values can be directly measured from 1D spectra and
the principal axis system orientations are approximately known from studies of model

compounds.

59

2D exchange spectroscopy has been used to investigate the structures of
molecular solids, synthetic polymers and peptide in the manner described above”.
However, in systems where sample quantities are limited and the spectra of the two sites
are not resolved, the application of this measurement may encounter the problems of low
sensitivity and complex data analysis. These problems can be alleviated to a large extent
by applying MAS, which periodically modulates the anisotropic interactions and narrows
the broad peaks to comparatively sharp peaks at the isotropic and spinning sideband
frequencies of the two labeled sites; therefore, it signiﬁcantly improves the sensitivity

and resolution. These frequencies for site n = 1 or 2 can be calculated as VM" 2 vise“ +

Mv,, where visa“ and vr are the isotropic and MAS frequencies, and M = 0, i1, i2,... If the
shift tensors of the two sites are not colinear, then the 2D exchange spectrum will contain
off-diagonal crosspeaks which have frequency vM“ in one dimension and frequency vpy",
(M at M’, n at n’) in the second dimension. If the CSA principal values and principal axis
orientations of each labeled site are known, then the relative off-diagonal crosspeak
intensities can be simulated as a function of local structure as shown by Veeman et al.5,
Spiess et al.6’ 7, and Tycko et al.8' 9

In a peptide, the backbone is comprised of a series of planer peptide bond units
with relative orientation speciﬁed by the dihedral angles 4) and (p. In a peptide sample
with 13C labeling at two carbonyl carbons on adjacent residues, the orientation of the
CSA tensor for each carbonyl 13C is well-established"). Therefore, the local secondary
structure dihedral angles ((p, w) are the main unknown parameters which determine the
relative orientation of the carbonyl CSA tensors and hence the off-diagonal crosspeak

intensities in a 2D exchange spectrum. Using the known bond geometry in peptides and

60

excha

llE'

Obi;
Mt
IOU
CFC
pu

111

the known orientation of the

O

H
(91 gy'cazﬁ‘: [L carbonyl CSA tensor relative
\Cal/ \N2 """" 02/ \Cor/
H

l to its local bonding, formulas

H
0 have been derived for the 2D

exchange crosspeak intensities as a function of (q), w)8' 9. With the additional input of
experimental CSA principal values, the simulated crosspeak intensities can be calculated
and compared to the experimental intensities. The dihedral angle set corresponding to the
best agreement between experiment and simulation will then be the most likely ((1), (p) set

of the more C-terrninal residue.

METHODS

Two-Dimensional (ZD) Exchange Experiments. 2D exchange spectra were
obtained on the Varian VXR spectrometer at 2.5 kHz spinning frequency and -50 °C
with doubly 13C carbonyl labeled peptide/lipid samples. Spectra were acquired using the
rotor-synchronized pulse sequence (CP — t1 —1t/2 — T — Tt/2 — t2), in which (CP) represents
cross-polarization from lH to 13C, t1 is the evolution period, (rt/2) represents a 13C 1I/2
pulse, 1 (500 ms) is the spin diffusion period, and t2 is the 13C detection period. Samples
typically contained 2 limole of labeled peptide. The experiment was run under the
following conditions: (1) 800 us CP was made with 13C RF radiation at 47 kHz and a
linear ramp on 1H between 42 kHz and 52 kHz; (2) 47 kHz ”o 7:12 pulses were applied;
(3) 72 t1 points were taken with an increment of 40 us; (4) signals were acquired for

10 ms during t2; (5) decoupling at 75 kHz was applied during t1 and t2 but not during 1;

61

and (6) the recycle delay was 0.5 s. Complete data sets were collected in twelve-hour
blocks and then summed together, with 10-15 blocks in a final data set.

Spectra were processed using NMRPipe software“ with 150 Hz line broadening
in the t1 and the t2 dimensions. In the 2D spectrum, structural information is contained in
the relative intensities of the off—diagonal crosspeaks between the spinning sidebands of
the labeled carbonyl sites. These intensities depend on the relative orientations of the two
labeled carbonyls' chemical shift anisotropy (CSA) tensors and can be directly related to
the dihedral angles «1) and w of the more C—terrninal labeled residue‘g’ 12.

Integrated experimental crosspeak intensities were calculated by summing the
intensities of points in a 1 ppm x 1 ppm area. To determine the most likely values for (l)
and w, experimental crosspeak intensities were compared to simulated crosspeak
intensities calculated for a grid of (1) and 11! values. This grid was made in 5° increments
for —180° 5 (p5 0° and —180° S w _<_ 180°. The simulated crosspeak intensities were
calculated from a computer program supplied by R. Tycko (NIH) and included a small
contribution (< 20%) from longitudinal l4N relaxation, whose characteristic time is
shorter than the 500 ms exchange time. The simulated crosspeak intensity pattern for a
(o, w) pair is the same as that for (—¢, —w). In addition to their dependence on (11 and w,
the simulated intensities depend on the orientation of the carbonyl CSA principal axes
relative to the peptide group and on the carbonyl CSA principal values. The former were
taken from the literature") while the latter were experimentally determined from
measurements of the 1D peak spinning sideband intensities measured at a few different
spinning frequencies between 2 kHz and 4 kHz”. In the FP-(L7F8)C/DTPC sample, the

sideband intensities were measured from raw 1D spectra and contained a small

62

contribution (< 10%) from natural abundance signals of unlabeled peptide carbonyls. For
the LM-3 samples, the sideband intensities were measured from 1D difference spectra
and only contained contributions from the labeled carbonyl nuclei. In all samples, the
chemical shifts of the two labeled carbonyl nuclei were not resolved so the same set of
principal values was assumed for both nuclei.

In the 2D analysis, the total squared deviation )6“), qt) between experiment and
simulation was evaluated as:

x201). w) = z. [13, — to, w) x slap. w)lz / oz . i = 1 to N. (3)
where E, and 8,01), m) are experimental and simulated off-diagonal crosspeak intensities,
0'2 is the mean-squared noise per data point, 1(4), w) is a scaling factor calculated to
minimize x2 at each ((1), w) pair, and N is the total number of data points. Because mirror
image (across-the-diagonal) crosspeaks always have the same simulated intensity, the
experimental mirror image crosspeak intensities were summed, and the twenty off-
diagonal crosspeak intensities (from a 5 x 5 2D spectral array of spinning sidebands)

were reduced to ten data points”.

RESULTS AND DISCUSSION

Model Studies. 2D exchange measurement was tested by measuring the dihedral
angles of Gly-2 in the polycrystalline tripeptide Ala-Gly-Gly. Figure 14a displays the 2D
exchange spectrum, and the arrow indicates an off-diagonal crosspeak. The experimental
data were fitted with simulations which yield a goodness-of—ﬁt parameter x2 for each

possible set of dihedral angles as shown in the contour plot of Figure 14b. The best ﬁt

63

secondary structure is given by the smallest x2. This region is coincident with the
crystallographically determined ((1), (p) values of (—83°, 170°).

A more realistic test was made in membrane-bound melittin, a 26-residue peptide
that is the major protein component of the European honey bee Apis mellifera venom. X-

ray diffraction data shows that soluble melittin is predominantly helical” 16. 1H NMR

17,18 17, 19,20

studies as well as CD and Raman spectroscopic studies also showed a helical
conformation when melittin binds to membranes. Figures 14c and 14d are the 2D solid-
state spectrum and x2 fitting result for membrane-bound melittin having the carbonyls of
Gly-3 and Ala-4 13c labelled. The minima — around the (—40°, 45°) dihedral angle set —

again agree with the typical helix conformation.

64

(a) (c)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

'0 .0 '0 ‘3 O 120 180 v I '-
O
'0 ﬂ '3 ¢ :0 h 140 90 '1
O
\ mm 3
O, . w u o. '0 0
‘ -180 g tp
- 1
a - a 0s -.
§ é ” 200 -90
. I n °. 0. 0. L220
1 I I l l l '180 v t . a v
220 200 180 160 140 120 '180 420 (I) -50 0
F2 ppm
(b) 180ml)
00 s c? '- 120 ‘ ‘
140 . .
° 90 % a r 90
~160 3 J 1
°0 % a % ‘ 3 W0 :
" 180 3 _ -
I?" a % St, oé — 200 -90-
I .9 ° “o ° ,, p220 ‘
l l r l l l '180 w t . r
220 200 180 160 140 120 480 -120 (I) 60 0

F2 ppm

Figure 14. 2D 13C solid-state NMR exchange spectra of (a) 5% polycrystalline AGG-
(A1G2)c peptide, and (b) Melittin-(G3A4)C/DTPC with a peptide:lipid molar ratio of
1:20. The corresponding secondary structure analyses are shown in (c) and (d),
respectively. These analyses are displayed as contour plots of the total squared deviation
x2 (normalized to spectral noise) between experimental and calculated off-diagonal
crosspeak intensities for a grid of dihedral angles. The darkest region in each plot
represents the best global fitting.

65

2D Exchange Measurements for Reﬁned Secondary Structure Determination of
Membrane-Associated Fusion Peptide. Because chemical shifts only give a general
indication of secondary structure (helix vs. strand), more precise measurements were
made using 2D exchange spectroscopy on samples containing doubly carbonyl labeled
peptides. These measurements typically provide dihedral angles with i 20° accuracy and
can hence distinguish between various types of helical, turn, and sheet conformations. In
addition, consistency between chemical shift and 2D exchange measurements provides
much greater confidence in both types of measurements.

Figure 15a-d respectively displays 2D exchange spectra for FP-(A1V2)c/LM-3,
FP-(L7F8)C/DTPC, FP-(G10F11)c/LM-3, and FP-(A14A15)c/LM-3, all made with 1:20
peptide:lipid molar ratio. The first, third, and fourth samples were prepared with
method 1 and the second sample was prepared with method 2. Phosphate buffer (50 mM,
pH 7.0) had been added to the FP-(L7F8)c/DTPC sample after initial peptide/lipid
binding, but this had little effect on the 1D spectrum and presumably the membrane-
associated peptide structure. The 1D lineshapes of the LM-3 samples at 1:20 peptide:lipid
ratio are comparable to their respective lineshapes at 1:100 peptide:lipid ratio with 2 - 3
ppm FWHM linewidths from each labeled carbonyl nucleus. The 1D lineshape of the
FP-(L7F8)C/DTPC sample (prior to phosphate addition) is displayed in Figure 9d with ~
2 ppm linewidth from both nuclei. Thus, in all samples, it is reasonable to conclude that
the labeled nuclei of interest are in a well-structured region of the peptide.

The experimentally determined carbonyl CSA principal values were:
FP-(A1V2)c/LM-3, 246 ppm, 180 ppm, 92 ppm; FP-(L7F8)c/DTPC, 241 ppm, 179 ppm,

93 ppm; FP-(G10Fll)C/LM-3, 240 ppm, 174 ppm, 93 ppm; and FP-(A14A15)c/LM-3,

66

241 ppm, 186 ppm, 94 ppm. The experimental uncertainty in each principal value
determination was typically i 2 ppm. These principal values are comparable to those

[2, 2]

measured in rigid solids which indicates that large peptide backbone motions have

been frozen out at the -50 °C measurement temperature.

The analyses for FP-(A1V2)C/LM-3, FP-(L7F8)c/DTPC, FP-(G10F11)c/LM-3,
and FP-(A14A15)c/LM-3 are displayed in Figure 15e-h, respectively, and represent
contour plots of the x2 difference between the experimental crosspeak intensities and the
simulated intensities calculated for a grid of ((1), m) values. In the contour plots, black
shading represents the lowest (best-ﬁt) )6, increasing lighter shades of gray represent
increasing intervals of two units of )8, and white shading represents all x2 greater than
some value speciﬁed in the Figure 15 legend. One unit of x2 corresponds to about one

122. Only negative values of (I) are displayed because the simulated

conﬁdence leve
crosspeak intensities for any (q), 11!) are the same as those for the corresponding (-¢, -\lt).
For Figure 15e, 15g, and 15h, the non-displayed dihedral angle regions of —90° < (I) < 0°
and —180° < \u < 90° would all be represented by white. Table 4 presents the best-ﬁt (d),
w) for the Val-2, Phe-8, Phe-l 1, and Ala-15 residues of membrane-associated FP. In all
four cases, the best-fit values correspond to the B strand region, which is consistent with
the non-helical chemical shifts presented in Table 3. The corresponding (—¢, —w) values
are sterically disfavored for these non-glycine residues”. Comparison of analyses using

different NMR processing parameters (e.g. line broadening) suggests that each dihedral

angle can be determined with a precision of about i 20°.

67

 

 

 

 

 

 

 

 

(a)
o 0 ﬂ 6 ' —L-
1’ e130
- e a - ﬁ- I
1" I ‘ o . -170
9 y 0 ~O Q :
~21O
’ G I o r-
l l I l l l l l l l
210 170 130
F2 ppm
(b)
0 Q o . l"
‘ -130
o o O o '4 :
O .—
- o _p . 0 ° -170
0 a. O . o :
‘ . . . . :210
I l l l l l l l l l r
210 170 130
F2 ppm

uldd 1:1

 

 

 

 

 

 

 

 

. . - , 180
L’ 1120
:- “-w
:' :° ‘11
- --60
1-
. , . -180
-180 -1Z) (I) -60 0
(f) . . 1 . 180
F ~60
L :0 w
? (CR-Q, :“
_ $52....» “120
' r , , , -180
-180 -120 -60 0

Figure 15. 2D l3C solid-state NMR exchange spectra of (a) FP-(A1V2)c/LM-3, (b)
and (d) FP-(A14A15)c/I_M-3. Each
spectrum represents the summation of 10-15 twelve-hour data blocks. Secondary
structure analyses for (a)-(d) are shown respectively in (e)-(h). These analyses are
displayed as contour plots of the total squared deviation x2 (normalized to spectral noise)
between the experimental and calculated off-diagonal crosspeak intensities for a grid of
dihedral angles. The darkest regions represent values of x2 less than 6, 9, 12, and 12 for
(e)-(h), respectively. Each lighter contour level represents an increase of two units in x2.

FP-(L7F8)c/DTPC, (c) FP-(G10F11)C/LM-3,

68

 

 

 

 

 

 

 

 

(G)
0. ' 0 Q . '—
0’ -130
\ ' of :
~ o f o —170
o p 0 . «o 2 :
' L.
’ o ‘ q. _210
l l l r l l l l l
210 170 130
F2ppm
(d)
O 0 . 0 . '-
_ I -130
o O C .I . :
o o p , o . e170
. I v .. . :
—210
I O o a _
l 1 r1 r1 I | l l
210 170 130
F2ppm

Figure 15 (cont’d).

 

180
- 120

 

 

 

 

L120

 

 

-180

420

8.1

 

Table 4. Structural Correlations from 2D Exchange Measurements.

 

 

Residue Determined Best Fit (1), (p Predicted Structure
Val-2 —160°, 140° B strand
Phe-8 —120°, 115° B strand
Phe-ll —l65°, 140° B strand
Ala-15 —145°, 140° B strand

 

69

CONCLUSION

The reﬁned secondary structure determination by 2D exchange measurements is
consistent with the B-strand structure predicted from the chemical shifts analysis. All
these results provide strong evidence for the presence of an extended B-strand
conformation in the N-terminal and central regions of the membrane-associated peptide.
The C-terminus is more disordered. We interpret these data with a model of insertion of
the N-terminal and central residues of the peptide into the membrane and the C-terminus
outside the membrane. Within the membrane, there is greater peptide structure because
hydrogen bonding must be either inter— or intra-peptide, whereas outside the membrane,
greater conformational disorder is allowed because of the possibility of peptide-H20
hydrogen bonds. N-terrninal membrane insertion of the fusion peptide is an important
element of biological fusion models and is biophysically reasonable because of the apolar
character of the N—terminal residues and the polar/charged character of some of the C-
terminal residues.

In the B-strand conformation, hydrogen bonding may be satisﬁed by a hairpin
structure and/or by a parallel or antiparallel arrangement of peptide oligomers. One
tertiary structure may be strongly preferred or there may be a mixture of populations of
tertiary structures. Although none of our 2D exchange constraints are speciﬁcally
consistent with a turn motif of a hairpin structure, none of the measurements have been
made on any of the glycines that would likely be part of such a turn. In the next chapters,
we will apply solid-state NMR REDOR technique to investigate possible formation of B

sheet oligomer structures via detecting inter-peptide hydrogen bond patterns.

70

REFERENCE

10.

ll.

Edzes, H. T., and Bemards, J. P. C. (1984) Two-dimensional exchange NMR in
static powders - interchain C-l3 spin exchange in crystalline polyethylene. J. Am.
Chem. Soc. 106, 1515-1517.

Henrichs, P. M., and Linder, M. (1984) C-l3 spin diffusion in the determination
of intermolecular structure in solids. J. Magn. Reson. 58, 458-461.

Tycko, R., and Dabbagh, G. (1991) Nuclear-Magnetic-Resonance crystallography
- molecular orientational ordering in 3 Forms of solid methanol. J. Am. Chem. Soc.
113, 3592-3593.

Dabbagh, G., Weliky, D. P., and Tycko, R. (1994) Determination of monomer
conformations in noncrystalline solid polymers by 2-dimensional NMR exchange
spectroscopy. Macromolecules 27, 6183-6191.

Kentgens, A. P. M., Deboer, E., and Veeman, W. S. (1987) Ultraslow molecular
motions in crystalline polyoxymethylene - a complete elucidation using two-
dimensional solid-state NMR. J. Chem. Phys. 87, 6859-6866.

Hagemeyer, A., Schnridt-Rohr, K., and Spiess, H. W. (1989) Two-dimensional
nuclear magnetic resonance experiments for studying molecular order and
dynamics in static and rotating solids. Adv. Magn. Reson. 13, 85-130.

Luz, Z., Spiess, H. W., and Titman, J. J. (1992) Rotor synchronized MAS 2-
dimensional exchange NMR in solids - principles and applications. Isr. J. Chem.
32, 145-160.

T ycko, R., Weliky, D. P., and Berger, A. E. (1996) Investigation of molecular
structure in solids by two-dimensional NMR exchange spectroscopy with magic
angle spinning. J. Chem. Phys. 105, 7915-7930.

Tycko, R., and Berger, A. E. (1999) Dual processing of two-dimensional
exchange data in magic angle spinning NMR of solids. J. Magn. Reson. 141, 141-
147.

Oas, T. G., Hartzell, C. J., McMahon, T. J ., Drobny, G. P., and Dahlquist, F. W.
(1987) The carbonyl C-13 chemical-shift tensors of 5 peptides determined from
N-15 dipole-coupled chemical-shift powder patterns. J. Am. Chem. Soc. 109,
5956-5962.

Delaglio, F., Grzesiek, S., Vuister, G. W., Zhu, G., Pfeifer, J ., and Bax, A. (1995)

NMRPipe: a multidimensional spectral processing system based on UNIX pipes
[see comments]. J Biomol NMR 6, 277-293.

71

12.

l3.

14.

15.

16.

17.

18.

19.

20.

21.

22.

23.

Weliky, D. P., and Tycko, R. (1996) Determination of peptide conformations by
two-dimensional magic angle spinning NMR exchange spectroscopy with rotor
synchronization. J. Am. Chem. Soc. 118, 8487-8488.

Herzfeld, J ., and Berger, A. E. (1980) Sideband intensities in NMR - spectra of
samples spinning at the magic angle. J. Chem. Phys. 73, 6021-6030.

Gabrys, C. M., Yang, J., and Weliky, D. P. (2003) Analysis of local conformation
of membrane-bound and polycrystalline peptides by two-dimensional slow-
spinning rotor-synchronized MAS exchange spectroscopy. J. Biomol. NMR 26,
49-68.

Terwilliger, T. C., Weissman, L., and Eisenberg, D. (1982) The structure of
Melittin in the forrn-I crystals and its implication for Melittins lytic and surface-
activities. Biophys. J. 37, 353-361.

Terwilliger, T. C., and Eisenberg, D. (1982) The structure of Melittin .1. structure
determination and partial reﬁnement. J. Biol. Chem. 257, 6010-6015.

Dawson, C. R., Drake, A. F., Helliwell, J., and Hider, R. C. (1978) Interaction of
bee Melittin with lipid bilayer membranes. Biochim. Biophys. Acta 510, 75-86.

Knoppel, E., Eisenberg, D., and Wickner, W. (1979) Interactions of Melittin, a
pre-protein model, with detergents. Biochemistry 18, 4177-4181.

Vogel, H. (1981) Incorporation of melittin into phosphatidylcholine bilayers -
study of binding and conformational changes. F EBS Lett. 134, 37-42.

Vogel, H. (1987) Comparison of the conformation and orientation of Alamethicin
and Melittin in lipid membranes. Biochemistry 26, 4562-4572.

Weliky, D. P., Bennett, A. E., Zvi, A., Anglister, J ., Steinbach, P. J ., and Tycko, R.
(1999) Solid-state NMR evidence for an antibody-dependent conformation of the
V3 loop of HIV-1 gp120. Nat Struct Biol 6, 141-145.

Bevington, P. R., and Robinson, D. K. (1992) Data Reduction and Error Analysis
for the Physical Sciences, McGraw-Hill, Boston.

Cantor, C. R., and Shimmel, P. R. (1980) Biophysical Chemistry, W. H. Freeman,
New York.

72

CHAPTER V

DESIGN AND OPTIMIZATION OF SOLID-STATE ROTATIONAL—ECHO

DOUBLE-RESON AN CE (REDOR) MEASUREMENTS TO PROBE DISTANCES

BETWEEN CARBON YL CARBONS AND AMIDE NITROGEN S

73

BACKGROUND

Due to the relative simplicity and robustness of the experiment and data analysis,
rotational-echo double-resonance spectroscopy (REDOR) is one of the widely used MAS
solid-state NMR techniques to analyze molecular structures by probing heteronuclear
dipolar couplings and distances”. Heteronuclear dipolar coupling is generally averaged
out by MAS, but we can reintroduce the dipolar coupling by applying two 1t pulses in one
rotor period and still take advantage of the better sensitivity and resolution from MAS.
As the result of the reintroduced dipolar coupling, the magnetization of the observed
nucleus will be reduced or “dephased” to 81 as compared to So from the sequence without
the two 11: pulses in each rotor period. This reintroduced dipolar coupling can be
accumulated over a long dephasing time which consists of many rotor periods. The
amount of normalized reduction ratio, (So — St)/So or AS/So, as a function of dephasing
time, can be used to extract the heteronuclear dipolar coupling, and the corresponding
heteronuclear distance can be obtained by its inversely proportional relationship to the
cube root of the dipolar coupling (distance oc (dipolar coupling) "“3).

REDOR experiments involve at least two channels, the detection channel that
detects the NMR signal and the dephasing channel that reintroduces the heteronuclear
dipolar coupling. In samples with a large number of protons, it is often necessary to
include a third proton decoupling channel to minimize the strong proton dipolar coupling.
For probing the dipolar coupling between 13C carbonyl carbons and 15N amide nitrogens,
the detection channel is tuned to 13C and the dephasing channel is tuned to 15N.

In this study, REDOR was applied to identify inter-peptide hydrogen bonds

through detection of the 40 ~ 45 Hz 13C—15 N dipolar coupling arising from a ~ 4.1 A 13C

74

carbonyl—”N amide distance. In the REDOR experiment, this is a rather weak coupling
and requires a dephasing period of 10 ~ 30 ms duration and hundreds of NMR 113 pulses
for detection. We therefore expended some effort in optimizing the REDOR experiment
for detection of these hydrogen bonds. The compounds used for optimization was the
lyophilized model helical peptide MB(i+4)EK-A9CA13N, the lyophilized 5% 1-‘3c, 15N
L-leucine, the polycrystalline 10% 1-13C, 15N Glycine and the lyophilized FP-F8CL9N
peptide.

The factors that we examined with model compounds include the quality of the

pulses, 1H decoupling sequence and strength, and REDOR sequence variant.

MATERIALS AND METHODS

Materials. A 17-residue acetylated peptide MB(i+4)EK-A9CA13N (Ac-
AEAAAKEAAAKEAAAKA) was synthesized as a C-terrninal amide with 13C carbonyl
labeling at Ala-9 and ‘5N amide labeling at Ala-13. This peptide is predominantly (-
85%) or helical in its lyophilized form so the two labeled nuclei should have a C—N
internuclear distance of 3.9 ~ 4.2 A8. A 1-‘3C, 15N L-leucine sample was prepared by
dissolving a 1:19 mixture of labeledzunlabeled leucine in water. After lyophilization, the
two labeled nuclei should have a C-N internuclear distance of ~ 2.5 A. A 1-13C, 15N
Glycine sample was prepared by dissolving a 1:9 mixture of labeledzunlabeled glycine
followed by crystallization by slow evaporation of the water. The two labeled nuclei
should also have a C—N internuclear distance of ~ 2.5 A. A lyophilized FP-F8cL9N dry
peptide was used as a model compound that has directly bounded 13C-15N pair with a

distance of 1.3 ~ 1.4 A.

75

Solid State NMR Spectroscopy. Experiments were done on the Varian Infinity
Plus spectrometer using either a 4 mm triple resonance MAS probe or a 6 mm triple
resonance MAS probe. The NMR detection channel was tuned to 13C, the decoupling
channel was tuned to 1H, and the third channel was tuned to 15N. 13C and 15N chemical
shifts were externally referenced to the methylene resonance of adamantane at 38.2 ppm
and (NI-14)2SO4 at 20 ppm, respectively. In our discussion of following results, “13C” and
“15 N” refer specifically to carbonyl carbon and amide nitrogen, respectively, and
“intensity” refers to the integration of isotropic l3C carbonyl peak.

In REDOR experiments two spectra were acquired: S0 acquisition without a 13C-
15N recoupling sequence, and 81 acquisition with such a sequence. The general REDOR
sequence starts with l — 2 ms of CP at 50 kHz, followed by a few ms dephasing period
and then direct l3C detection]. Optimization was performed using a variety of REDOR
dephasing sequences, and different lH decoupling sequences and strengths during
dephasing and acquisition. Unless noted otherwise, the XY—8 phase cycling scheme was
used for the it pulsesg’ 10. The 13 C transmitter was set close to the frequency of the labeled
carbonyl or carboxyl carbon and the 15N transmitter was set close to the frequency of the
labeled amide or amine nitrogen. Calibration of the 13C 1t pulse length was done with a
“Z-ﬁlter” sequence (CP-rr/Z-t-rt-acquire) in which lH decoupling was applied during the
pulses and acquisition and for 10 us at the beginning and end of the 1 period. The 15N 71:
pulse was calibrated with minimization of the signal in 81 spectrum at desired dephasing
time. To obtain optimal compensation of B0, B1, and spinning frequency drifts, So and Si
FIDs were acquired alternately. The recycle delay was 1 s. The data were processed with

10 — 100 Hz line broadening and with ﬁfth order polynomial baseline correction.

76

REDOR Data Analysis. The integrated signal intensities in the isotropic carbonyl
regions of the So and St spectra are denoted as “S0” and “Si”. AS/So is a convenient
normalized dephasing ratio for comparison of experiment and simulation and is deﬁned:

AS/So = (So — St)/So = 1—(81/80) (4)
The corresponding dipolar coupling can then be extracted by fitting the experimental data
with theoretical calculations, and related to the 13C—‘SN distance r through the f3
dependence of the dipolar coupling.

To estimate the real dephasing ratio from the speciﬁc labeled nuclear pair, the
experimental AS/So values were multiplied by a correction factor of 1.23 for the 5%
labeled IrLeucine sample and 1.10 for the 10% Glycine sample prior to ﬁtting to
eliminate the contribution of non-dephasing natural abundance carbonyls. For the model
peptide MB(I+4)EK-A9CA13N, the correction factor is about 1.2 ~ 1.3 which accounted
for the ~ 15% population of random coil peptide in the sample, the small populations of
labeled and natural abundance carbonyls which experience complete dephasing from
neighboring natural abundance 15N and labeled 15N, respectively, and the remaining
population of natural abundance carbonyls which do not dephase. For the FP-F8CL9N
peptide, the correction factor is 1.25 and includes only the non-dephasing natural
abundance carbonyls because the dephasing ratio from the directly bonded 13C-ISN pair
dominates the overall results at the examined dephasing time.

The theoretical dephasing ratio was calculated using the formulal’ 3:

(AS/so)T = 1 — (101010“ fo"{cos[2\/2 DT x sin(2B)sinOt]sinB}dBdOL (5)

77

where D is the heteronuclear dipolar coupling in Hz, T is the dephasing time, and (or, B)
are the azimuthal and polar angles describing the orientation of internuclear vector in the

rotor frame.

RESULTS AND DISCUSSION

Quality of REDOR rtPulses. The REDOR sequence mainly consists of it pulses,
and the quality of these it pulses can directly affect the REDOR result. In general, there
are several factors related to the quality of the 1: pulses: ﬂip angle, homogeneity of the
radiofrequency (RF) ﬁeld along the sample, and pulse shape.

The pulse length tolerance of the REDOR experiment was examined with the
MB(i+4)EK-A9CA13N peptide at 24 ms dephasing time where a few hundred pulses were
applied. An array of pulse lengths was tested and the corresponding REDOR dephasing
ratios were used to determine the optimal 7t pulse length range. The results (not shown)
suggest that the REDOR experiment can tolerate pulse length errors of approximately
i10% from the ideal it pulse length. Experimentally it is possible to control the pulse
length variation within this range.

The homogeneity of the RF coil along the rotor was estimated by dividing the
volume of regular rotor into 6 segments (two outmost 1/6, two outer 1/6 and two central
1/6 positions), restricting sample into one of these segment and measuring the 13C and
15N 1t pulse length experienced by the sample. A 5% 1-13C, 15N L—leucine sample was
used for the 4 mm rotor and a 10% 1-13C, 15N Glycine sample was used for the 6 mm

rotor. The results are listed in Table 5 for the 4 mm rotor and Table 6 for the 6 mm rotor,

78

respectively. The corresponding REDOR dephasing curves are displayed in Figure 16
and Figure 17, respectively.

It is clear that the in the NMR probe circuit, the RF ﬁeld is not homogenous along
the rotor. Samples placed at the edge of the rotor experienced a weaker ﬁeld than samples
placed at the center of the rotor. Therefore a longer pulse length is required to reach a
certain ﬂip angle as shown in Tables 5 and 6. Because the REDOR experiment employs
many 1: pulses and the error due to incorrect pulse length setting can accumulate over the
pulses, the inhomogeneity issue will have a signiﬁcant impact on the result. Figures 16
and 17 displays the REDOR dephasing curves of samples placed at different positions in
the 4 mm and 6 mm rotors, respectively. For comparison, the same settings optimized for
samples placed at the central position were used in all other positions. For the 4 mm
rotor, Figure 16 demonstrates that samples placed at the two central 1/6 positions gave
the best results because the correct pulse length was applied. When samples were placed
at the two outer 1/6 positions, the pulse length was too short enough to reach the desired
ﬂip angle and the results were worse. The pulse length experienced by samples placed at
the two outmost 1/6 positions was so far from the correct value that the magnetization
could not refocus, and no signal could be observed. Homogeneity is better in the 6 mm
rotor than in the 4 mm rotor. Figure 17 shows only subtle differences between samples
placed at the four 1/6 positions around the center of the rotor. Again, a significantly
worse REDOR curve was observed when the sample was placed at the two outmost 1/6

positions of the 6 mm rotor.

79

Table 5. Pulse Length at Different Positions in the 4 mm Rotor.

 

 

 

I3 3 15 a
Position in the Rotor (Top to Bottom) C PUISC Length (“8) N PUISC Length (“5)
it 2n 1! 21:
Outmost 1/6 16.8 31.4 18.2 36.2
Outer 1/6 10.8 21.5 13.5 26.5
Central 1/6 10.0 19.7 12.6 25
Central 1/6 9.8 19.7 12.7 25.3
Outer 1/6 11.0 21.8 14.0 28.2
Outmost 1/6 14.4 28.5 18.6 37.4

 

 

a The pulse lengths were measured with same RF ﬁeld strength.

 

 

 

 

1.01 {$.-
t
r txx.¢ilexxgm 54
0.8-1 0 VA A 9 g A V _
. ‘34 v V .
0- v
0.6- 'AA 4
(£0 i 2VV 1
(204- :v n
A
u ' .4
02- §4 ~
' i
00-9 .
r l I 1 ' l '
0 1 2 3 4
A=D*T

Figure 16. 13C-observed REDOR dephasing curve of calibrated AS/So for the 5% 1-13C,
15N L—leucine sample placed at different positions in the 4 mm rotor. A is the product of
dipolar coupling D (195 Hz, calculated from the 2.5 A l3C—ISN distance) and dephasing
time T. All experiments used the pulse lengths from the central position. The ﬁlled
triangles represent the results from two central 1/6 positions and the open triangles
represent the results from two outer 1/6 positions. The results from the two outmost 1/6
positions are not shown because the signals are significantly attenuated.

80

Table 6. Pulse Length at Different Positions in the 6 mm Rotor.

 

 

 

I3 3 15 a
Position in the Rotor (Top to Bottom) C PUISC Length (“8) N PUISC Length (“8)
it 211 11 Zn
Outmost 1/6 13.6 26.0 16.2 31.8
Outer 1/6 11.0 21.6 13.5 26.8
Central 1/6 10.0 19.9 12.5 24.8
Central 1/6 10.0 19.9 12.5 24.7
Outer 1/6 11.0 21.6 13.5 26.5
Outmost 1/6 13.0 25.2 15.8 31.2

 

 

a The pulse lengths were measured with same RF field strength.

 

 

 

 

10- ﬂ 1 ' I # I ' -
' e a 9
0.8- 3% v A V v V X -
J V V A X W X A A A ¢ J
c,os- A -
a) . 9 a
(D v X
<o42 8 A ..
A
. V
02- o 3 -
0.0- ' V -
. , . , . , .
0 1 2 3 4
=D*T

Figure 17. l3C-observed REDOR dephasing curve of calibrated AS/So for the 10% 1-13C,
SN glycine sample placed at different positions in the 6 mm rotor. A 195 Hz dipolar
coupling D from 2. 5 A 13C-—'5N distance was applied. All experiments used the pulse
lengths from the central position. The filled triangles represent the results from two
central 1/6 positions and the open triangles represent the results from two outer 1/6
positions. The half-ﬁlled triangles represent the results from two outmost 1/6 positions.

81

It is of interest to compare the results between a full rotor and a 2/3 full rotor with
the MB(i+4)EK-A9CA13N peptide, since it has the l3C—ISN distance comparable to the
real sample. Figure 18 displays the results of such comparison and suggests that for the 4
mm rotor, restricting the sample to 2/3 full volume significantly improves the REDOR
results. For the 6 mm rotor, the results from the 2/3 full volume are only slightly better
than for a full rotor. In summary, the RF homogeneity is uneven along the 4 mm and 6
mm rotors, and for practical reasons it is advantageous to restrict the sample volume to

the central 2/3 of the rotor volume.

 

 

 

 

1.0 . w . I . l
A A1
08- ‘ A ‘ A A A-l
AA A q
06 A A AAA
_ AA 2
o A V V'V
(204- A :a .38" -
A AA g$v
02 ‘AEAVW
. ‘ %%%¥ "
* rt.
00- ,.AV3 _
v I ' I ' I
0.0 0.4 0.8 1.2

A: D*T

Figure 18. l3C-observed REDOR dephasing curve of calibrated AS/So for the
MB(i+4)EK-A9CA13N peptide in full and 2/3 full 4 mm rotor (open and ﬁlled upper
triangles “A”), and in full and 2/3 full 6 mm rotor (open and ﬁlled lower triangles “V”).
100 kHz and 65 kHz TPPM decoupling were applied for the 4 mm and 6 mm rotors,
respectively.

82

The pulse shape was observed on an oscilloscope. The carbon 1: pulse is almost a
square wave while the nitrogen 71: pulse has signiﬁcant ring-up and ring-down effects at
the beginning and end of the pulse. An attempt to improve the pulse shape was made by
inserting cables with different lengths in the circuit. Figure 19 shows that we can modify
the pulse profile to nearly a square wave, but the corresponding REDOR ratio measured
on the MB(i+4)EK-A9CA13N peptide at 24 ms dephasing time did not improve. This
suggests that other properties (e.g. phase transients) associated with the pulse may also

contribute to the imperfection.

83

Table 7. Pulse Profile Adjustment and Corresponding REDOR Dephasing Ratio.

 

Circuit Modification Pulse Profilea Experimental AS/Sob’ c

 

Original >_ 0.41

 

 

 

 

 

 

 

 

 

Add 40 MHz M4 cable K >_ 0.41
\

Add 76 MHZ 1114 cable < }_ 0.42

Add 160 NIHZ 1J4 cable —U } 0.39
J\

 

 

a The pulse profile was observed on an oscilloscope.

b REDOR AS/So ratio was measured on the MB(i+4)EK-A9CA13N peptide at 24 ms
dephasing time.

c The uncertainty is 0.01.

84

1H Decoupling Sequence and Strength in REDOR Experiments. After
understanding the pulse imperfections in our spectrometer, we turned our focus to the 1H
decoupling sequence and strength during the dephasing and acquisition periods. In
general, the proton decoupling sequence and strength affect the 13C transverse relaxation
and residual coupling to protons. However, in the REDOR experiment these decoupling
effects affect Si and So similarly and cancel out in the dephasing ratio (AS/S) calculation,
so the REDOR results should be independent of proton decoupling as demonstrated by
Schaefer and coworkers’ 1 . However, there are also reports arguing that a strong proton
decoupling ﬁeld is essential in REDOR experiments to avoid cross-relaxation12 and
optimize the observed AS/So ratio”. Since most experiments use the 6 mm probe that has
a relatively low decoupling field limit (~ 65 kHz), it is necessary to test the impact of
limited decoupling in REDOR experiments.

The experiments were conducted on the MB(i+4)EK-A9CA13N peptide in the 2/3
volume restricted 4 mm rotor. Two kinds of proton decoupling sequences were used,
regular continuous wave (CW) decoupling and two-pulse phase-modulation (TPPM)
decoupling”. For each decoupling sequence, two ﬁelds (70 kHz and 100 kHz) were
applied. Figure 19 shows the REDOR S0 intensity and AS/So curve over a series of
dephasing times. The results indicate that the TPPM decoupling can preserve the signal
intensity better than regular decoupling. In Figure 19a, the So intensities at the two ﬁelds
with TPPM decoupling are almost the same and both are much higher than the one with
regular decoupling. TPPM decoupling also improves the AS/So ratio as indicated in
Figure 19b. At the same field strength, TPPM decoupling always gives a better result

than regular CW decoupling. When comparing within the same decoupling sequence, we

85

found that higher decoupling ﬁeld does lead to a better AS/So ratio. The abnormally high
AS/So values with CW decoupling at long dephasing times are associated with a very
large uncertainty due to small signal intensities.

Recently another experiment was conducted to test the existence of cross-
relaxation in the REDOR experiments. Our main concern involves the nitrogen channel

which employs many it pulses simultaneously with proton decoupling. Table 8 displays

the results of applying different nitrogen 1: pulse fields with same proton decoupling at 65

Table 8. Dependence of REDOR AS/So on the ‘H Decoupling/”N 1t Pulse Field Ratio.

 

Ratio of TH Decoupling
1.6 2.1 2.6 3.3 3.9 4.6 5.2 6.5
and 15N 1! Pulse Fieldill

 

Experimental AS/Sob 0.38 0.44 0.47 0.49 0.50 0.47 0.45 0.38

 

 

 

 

 

 

 

 

 

a The 1H decoupling was fixed to 65 kHz and the 15 N 11: pulse field was varied from 40
kHz to 10 kHz.
b The results were obtained at 24 ms dephasing time.

kHz. Ishii and coworker reported that the RF ﬁeld intensity for proton decoupling should
be set to around three times that of the dilute spin pulses to avoid cross-relaxation
between protons and dilute spins when applying multiple pulses of the dilute spin
simultaneously with proton decoupling”. Although in the REDOR experiment, the 13C—
15N dipolar coupling does not depend on the status of nitrogen magnetization, we did
observe a dependence of the REDOR AS/So ratio on the ratio of 1H decoupling to 15N n
pulse field. The optimal ratios were found between 3.3 and 3.9, which suggests that the

cross-relaxation can be important for designing optimal REDOR experiments.

86

To find out whether the dependence of the REDOR dephasing ratio on decoupling
ﬁeld relates to the probed l3C—ISN distance, i.e., the dipolar coupling strength, we
obtained the dephasing curve for two model compounds, 10% 1-13C, lsN Glycine and FP-
F8cL9N peptide, which have shorter l3C—‘SN distances and stronger dipolar coupling than
the MB(i+4)EK—A9CA13N peptide. The results in Figure 20 indicate that for both
samples, the AS/So ratio does not differ greatly for 1H decoupling ﬁeld between 100 kHz
and 65 kHz. This can not be explained by the number of nitrogen 1t pulses involved as
shown in Figure 21, which compares the difference of AS/So ratio at different spinning
speed for all three samples. In these experiments, the low speed is half as fast as the high
speed and the rotor period is twice longer. Therefore, at low speed the experiment
contains only half the number of nitrogen 1t pulses. The results (cf. Figure 21) indicate
only small differences among the samples, which argues against the sensitivity to the
number of nitrogen 1t pulses. The difference also can not be ascribed to the dephasing
time, because when we compared the AS/So difference for 10% l-BC, 15N Glycine and
MB(i+4)EK-A9CA13N peptide at comparable dephasing times, we observed signiﬁcantly
less AS/So ratio difference for the first sample. Therefore, the different dependence of
REDOR dephasing ratio on decoupling ﬁeld must relate to the strength of l3C—‘sN

dipolar coupling examined in the system.

87

 

A
m
v

 

 

 

 

 

 

 

 

25 j t t
X It
20" ‘ '
(B .i A x «1
515- v a ‘ -
.3: v A '
a) x
$10- v A I .
E X
— v A x
V A x x
5- v A x x .
V V A A A ' ‘ I x
V V A A X I g
o- VVveeeeeeo-
r l I 1 ' l
0.0 0.4 0.8 1.2
A=D*T
(b)
1.2 l I l
. A
1.0- -l
A
. A ‘
08- ‘ A ‘ A V -
1 ‘ -‘
0.6- A A '-
g 4 A A 6 V I V
(I) A v
<0.4‘ ‘ A 0 ' g v V d
. A A 3 3 v v i
02 ‘ f ' V-
. '1 ‘ x V V
4 ﬂ ‘ V
i A v v V 1
00- it 6 v v V _
u l ' l ' I
0.0 0.4 0.8 1.2
1=D*T

Figure 19. l3C-observed REDOR (a) So intensity and (b) calibrated AS/So for the
MB(i+4)EK-A9CA13N peptide in the 4 mm rotor. A 42 Hz dipolar coupling D from a 4.2
A l3C—‘SN distance was applied. The ﬁlled and open triangles represent the results with
TPPM and CW decoupling sequences, respectively. The upper (A) and lower (V)
triangles represent the results with 100 kHz and 70 kHz decoupling ﬁeld, respectively.

88

(a)

 

 

 

 

 

 

 

 

I I I
10" ‘ ‘ "
. A A A A v V
08- A c v V v e e 9 6 . V ..
' v
- O
C30.6- . -
9
C304- ‘ -
A
0.2- -
. i
0.0- ‘* -
I I I
O 2 3 4
A=D*T
(b)
f j I I
1.01 A : A ‘ a
v v A A v J
08 ‘ V ¢ 9 v
v
0.6- A -
0 V
(L) .
c20.41 _
- 9
0.24 _
A
0.0- 1
' I ' I j I r
O 1 2 3 4

A=D*T

Figure 20. 13C-observed REDOR calibrated AS/So for (a) 10% 1-‘3c, 15N Glycine with
196 Hz dipolar coupling D (2.50 A), and (b) FP-F8cL9N peptide with 1165 Hz dipolar
coupling D (1.38 A) in the 2/3 full 4 mm rotor. The upper (A) and lower (V) triangles
represent the results with 100 kHz and 65 kHz TPPM decoupling field, respectively.

89

 

 

 

 

1.04 e A -
69%? 3*) A 93"
l o M v
#0
08" 6W 9 .-
, A ‘
A
(5 0.6- 9 0 “AAA"' _
a .. A‘ vvva 4
<1 044 2: ..A$v' .,
, Q 0 a"
02- x‘. -
t
3 ¢ t“'x
00- *t“ e
I 1 I ' I T I
O 10 20 30
T(ms)

Figure 21. l3C-observed REDOR calibrated AS/So for FP-F8CL9N peptide spinning at 16
kHz (open lower triangles) and 8 kHz (open upper triangles), 10% 1—13C, 15N Glycine
spinning at 16 kHz (half-ﬁlled lower triangles) and 8 kHz (half-ﬁlled upper triangles)
(half-filled triangles), and MB(i+4)EK-A9CA13N peptide spinning at 8 kHz (ﬁlled lower
triangles) and 4 kHz (filled upper triangles). All the experiments use 65 kHz TPPM
decoupling. At each dephasing time for the same sample, the experiment running at lower
speed contains half the amount of 15N 1t pulses as the one at higher speed.

90

Selection of REDOR Sequences. Three commonly used REDOR pulse sequences
are shown in Figure 22. All these REDOR experiments rely on application of two 1t
pulses per rotor cycle and these could both be 15N pulses (“all 15N 1: pulse” version,
Figure 22a)’, one 15N and one 13C pulse (“alternating 15N/BC 7t pulse” version, Figure
22b)” or both 13C pulses (“all 13C 1: pulse” version, Figure 22c)”. The all lsN 1: pulse
version is the original REDOR sequence. One of its potential problems is that the 13C It
pulse in the middle of the dephasing time is not only responsible for refocusing the 13C
magnetization, but also adds the dephasing from the second half of the dephasing time to
the ﬁrst half. If the pulse does not have the right length for a 1: pulse (e.g. a 170° or 190°
pulse, instead), it will partially cancel the dephasing from the two halves. For example, a
1r/2 pulse will give only 50% of the dephasing as the 1: pulse. This problem can be
signiﬁcantly improved by the alternating 15N/13C 1: pulse version in which the central 13C
1: pulse is replaced by a 13C 1: pulse train using XY-8 phase-cycling scheme to
compensate the possible width and phase errors of a single 1: pulse and eliminate off-

7 8
resonance CffCCtSl' I

. The all '3 C 1: pulse version is useful when the 15N pulse
imperfection can not be corrected easily, so putting all of the 1: pulses on the ‘3C channel
can improve the REDOR results. The problem with the latter two versions of REDOR is
that the multiple 13C 1: pulses may also refocus the homonuclear dipolar coupling and

therefore reduce the overall intensity. The studies on 10% 1-‘3C, l5N Glycine are good

examples to illustrate the advantages and disadvantages of each REDOR version.

91

(a)
1H
13C

15N

rotor

(b)
130

15N

rotor

(6)

13C

15N
r0t0r

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

cp Decouple
¢ .
CP¢ 1 EAcquire
Y X Y X Y X X Y X Y
I I I I | I I I I I
I I I I I I I I I
0 1 2 3 4 5 6 7 8
x y x y y x ¢ . .
CF, I I I I I I I Em"re
Y Y Y X
J I I I
L I I L L L I I I
o 1 2 3 4 5 6 7 8
y x y y 4 x y x y y : .
0P» I I I I I I I I I I :ware
X
l
I I I I I I I I I
0 1 2 3 4 5 6 7 8

Figure 22. Three REDOR sequences with (a) “all 15N n pulse” version, (b) “alternating
15N/13C 1: pulse” version, and (c) “all 13C 7: pulse” version. XY-8 phase cycling is shown
is (a) and (b) and XY—4 phase cycling is shown in (c).

92

A
m
v

 

 

 

 

 

 

 

 

 

I I I I
70- A ‘ A A -
. A
60 A A A A ‘ A ‘ ‘ A A A A A A
tic)> - A A A A A I
950. A A A A A A A
- A A A 1
I; - A A A A A A A A
9340- A 4
C .
— A
30- -
A
204 A -
, A
1o~ -
I I I i
0 1 3 4
A: D*T
(b)
1.2 l l ‘ l
4 J
1
I I I
O 2 3 4

A: D*T

Figure 23. 13C-observed REDOR (a) So intensity and (b) calibrated AS/So for the 10% 1-
13C, 15N Glycine in the 2/3 full 4 mm rotor with all 15N 1: pulse version (ﬁlled triangles),
alternating lsN/13C 7: pulse version (half-filled triangles), and all 13C 7: pulse version
(open triangles). All REDOR experiments use 100 kHz TPPM decoupling. In (b), a
calculated REDOR dephasing curve based on 196 Hz dipolar coupling (2.50 A 13C—lsN
distance) is represented by the solid line.

93

Figure 23 compares the intensities in So and calibrated AS/So using the three
REDOR sequences. The decay of So intensity as a function of dephasing time is governed
by the spin-spin relaxation time T2, which can be different among different versions of
REDOR sequences. In Figure 23a, the So decays slowly with the all 15N n pulse version
because the T2 time is longer, while in the other two versions, the magnetization decays
faster because the T2 time is significantly shorter presumably due to the reintroduction of
l3C—BC homonuclear dipolar coupling by the application of 3 7t pulse train on the 13C
channel. In the all 13C 1: pulse version, there is an additional factor that can reduce the
intensity, the residual CSA that is not completely refocused due to the slight variation of
spinning speed. In the first two REDOR sequences, the CSA is averaged out in each rotor
period, so a little variation of the spinning speed is tolerable. In the all 13C 1: pulse
version, the CSA can be completely refocused only if the first half of the dephasing time
is exactly the same length as the second half of the dephasing time, which requires a
highly stable spinning speed. Obviously, the longer the dephasing time is, the harder it is
to satisfy this requirement and thus the stronger the residual CSA. In experiments
conducted at 8.000 i 0002 Hz, using the all 13C 7t pulse version, the majority of the
magnetization was lost after 10 ms dephasing time because of incomplete refocusing of
the CSA. On the other hand, when we look at the REDOR AS/So ratio, the reduction of
signal intensities due to residual CSA are cancelled in both So and S; spectra and the
advantage of applying XY—8 phase cycling on 13C channel becomes dominant. The all
13C 1: pulse version now gives the highest dephasing ratio. The signiﬁcant improvement
of the all 13C 1t pulse over the all '5 N 1! pulse version is also consistent with the better l3C

pulse shape compared to the 15N pulse shape.

94

The ultimate goal of REDOR studies on model compounds is to verify the
validity of this technique. For this polycrystalline sample, we obtained a reasonably good
ﬁt when experimental results were compared with the theoretically calculated results
based on data from its crystal structure. However, the polycrystalline morphology is not
easily available for real samples, and the So intensity in real samples may decay faster
than Glycine which has an unusually long T2 relaxation time probably due to the small
number of protons in the molecule. It is therefore necessary to conduct studies on more
realistic samples.

Using the all 15N 7: pulse version, the REDOR results of the FP-F8cL9N peptide
sample can be well ﬁtted into a 1.38 A l3C—‘SN distance which matches the typical amide
bond length (Figure 24). Since this short distance corresponds to a strong dipolar
coupling (1165 Hz), the whole dephasing curve can be obtained with a small number of
rotor periods and 1t pulses, the imperfection of pulses have minor effects, and the two

decoupling ﬁelds show small differences.

95

 

 

 

 

 

I T
I l

}»=D*T2

(13"
A

Figure 24. 13'C-observed REDOR calibrated AS/So for the FP-F8cL9N peptide in the 2/3
full 4 mm rotor using all 15N 1: pulse version with 100 kHz (upper triangles) and 65 kHz
(lower triangles) TPPM decoupling. The calculated REDOR dephasing curve based on
1165 Hz dipolar coupling (1.38 A 1 C-‘SN distance) is represented by the solid line.

96

More challenging is the MB(i+4)EK-A9CA13N peptide that has a 13C—ISN
distance comparable to that of the real fusion peptide sample. The measurement of this
sample is difficult in that the 13C—ISN distance is longer (~ 4.2 A) and the corresponding
dipolar coupling is weaker (~ 42 Hz) and thus more 1t pulses are needed to build up the
REDOR curve for quantitative analysis. Figure 25 displays the results of using the all '5 N

1!: pulse version at two TPPM decoupling fields (70 kHz and 100 kHz) and the alternating

15NIBC 1: pulse version at 65 kHz. Because of signiﬁcant signal loss, the all 13C it pulse

 

 

 

 

 

' l ' l r l
0.0 0.4 0.8 1 .2
A = D‘T

Figure 25. 13C-observed REDOR calibrated AS/So for the MB(i+4)EK-A9CA13N peptide
in the 2/3 full 4 mm rotor (ﬁlled triangles) and 6 mm rotor (open triangles). All 15N 1:
pulse version REDOR with 100 kHz (upper triangles) and 70 kHz (lower triangles)
TPPM decoupling was used for the samples in the 4 mm rotor. All 15N 1: pulse version
(lower triangles) and alternating 15N/I3C 1t pulse version (upper triangles) REDOR with
65 kHz TPPM decoupling were used for the samples in the 6 mm rotor. The calculated
REDOR dephasing curve based on 42 Hz dipolar coupling (4.20 A l3C—ISN distance) is
represented by the solid line.

97

version can not be applied. These results demonstrate that we can obtain results close to
the theoretical calculations with either high decoupling ﬁeld or using the alternating
15N/BC 1t pulse REDOR sequence.

There is another interesting aspect associated with the alternating 15N/13C 1t pulse
REDOR sequence when we use 8n or 8n + 2 rotor periods (n is a natural number). The
difference is that with 8n rotor periods we obtain a high AS/So ratio but with 8n + 2 rotor
periods we obtain high So intensity as shown in Figure 26. The intuitive explanation is
that this difference depends on whether the number of n pulses in a speciﬁc channel
fulﬁlls the complete XY—8 phase cycling. One would postulate that the phase cycling
scheme in the 13C channel determines the S0 intensity while the phase cycling scheme in
the 15N channel determines the AS/So ratio. The XY—8 phase cycle was found to be a
better choice than other phase cycles, e.g. XY—4 or XY-16. Therefore, the best result
was obtained when an XY—8 phase cycle was completed in the channel. With 8n rotor
periods, the ISN channel has a complete XY—8 phase cycling so a better AS/So was
obtained, while with 8n + 2 rotor periods, the 13C channel has a complete set of XY—8
phase cycling and a higher So was observed. To verify this argument, we eliminated the
ﬁrst and last 15N 1: pulses in the sequence so that when the rotor periods was set to 8n + 2,
both channels have completed the XY—8 phase cycling. However, similar results were
observed when 8n or 8n + 2 rotor periods were applied with this modiﬁed alternating
l5N/BC 1: pulse REDOR sequence (data not shown). Therefore, the experimentally
observed variations of the So intensities and AS/So were not correlated with the

completion of the XY—8 phase cycle scheme.

98

(a)

 

 

 

 

 

 

 

 

f I I I
60- ' -
v
'v
- V .4
50 A v
(1‘)3 AA 'v
540- h 'v -
E AA 7'
U) ., AA 7'
530- AAA '7' -
E AA 'v
— A v
AA 'v
20- AAA vvvv -
A V
A Vv
10 AA“:
1 r j l r I
00 04 A Dq_0£ 12
(b)
10 r u v I ‘ ' A
AA
. AAA 1
0.8" A“ -
AA vi
AA 'vv
06- v -
AA 'v"
(E " AA 'v' 4
(7)04- A V _
< AA V
‘A'V' a
0.2- AA" a
A'V
.. ~Av~v II
0.0-1W" -
. I , I 1 j
00 OA 08 12
A=DW

Figure 26. 13C-observed REDOR (a) So intensity and (b) calibrated AS/So MB(i+4)EK-
A9CA13N peptide in the 2/3 full 6 mm rotor with alternating 15N/13C 1t pulse REDOR
using 8n (upper triangles) and 8n + 2 (lower triangles) rotor periods.

99

Although our results have verified that this technique can probe l3C—‘SN distances
comparable to hydrogen bonds (3.9 ~ 4.3 A), it is not suitable for the sample system we
probed. In practice, the 4 mm rotor can not hold enough sample to provide adequate
signal-to-noise in reasonable amount of time. Also, in the real sample we have triply 13 C
labeled peptide so that the strong homonuclear l3C-BC coupling could reduce the
intensity if the alternating 15N/ 13C n pulse REDOR sequence was used. Therefore
analyses using the all 15N 1t pulse REDOR with the 6 mm rotor are preferred. A ﬁtting
study was conducted to see whether the deviated REDOR curve can be ﬁt into a single
distance so that a calibration factor can be derived to reflect the real distance. Figure 27
displays the experimental results with 40 and 20 kHz 15N 1: pulse ﬁelds, and obviously
the latter gave better results. However, neither set yields a single distance, instead, they
span a range of distances. The result of the lower 15N 1t pulse ﬁeld was collected after the
measurements of membrane-associated fusion peptide samples, and to be consistent, we
still focus on the result with the 40 kHz lsN 1t pulse ﬁeld. At ~ 24 ms dephasing time, the
experimental results are best ﬁtted with a 4.6 A distance, which will be used to derive the

calibration factor at the 24 ms dephasing time in the next chapter.

100

1.0 . r . u - u . I

 

 

 

 

I
u
0.8- . R}.
v
. VI 1
0.6- , . . _
<0
<1 0.4-l ..
024 .
/
1 x 1
0.0- ' -
u I I ] ' T ' I
0 8 16 24 32
T(ms)

Figure 27. l3C-observed REDOR calibrated AS/So for the MB(i+4)EK-A9CA13N peptide
in the 2/3 full 6 mm rotor using 40 kHz 15N 1: pulse ﬁeld (lower triangles) and 20 kHz
l5N 7: pulse ﬁeld (upper triangles). The all 15N 1: pulse REDOR with 65 kHz TPPM
decoupling was used. The calculated REDOR de hasing curves based on 4.2 A (I), 4.3 A
(II), 4.4 A (HI), 4.5 A (IV), 4.6 A (V), and 4.7 (VI) 3C—IS’N distances are represented
by solid lines.

101

To find out whether we can combine the advantages of the ﬁrst two versions of
REDOR sequences, we designed a series of REDOR sequences in which the number of
15N 1: pulses between 13C n pulses was set to 1 (the alternating 15N/n‘C 1t pulse version), 3
and 47, respectively. The sequences were examined with a total of 192 rotor periods so
that the corresponding numbers of additional 13C n pulses to compensate pulse
imperfection were 190, 94, and 6, respectively. The idea was to reduce the 13C 7: pulses to
a smaller number so that it would have less l3C—UC recoupling than from the alternating
l5N/13C rt pulse version and still have better compensation of pulse imperfection than the
single 13C 1t pulse in the all 15N 1: pulse version. As the results, with the 6 additional 13C 1:
pulses version we would expect higher 80 intensity than that of the alternating 15N/13'C 1t
pulse version, and better AS/So than that of the all 15N 1: pulse version. The experimental
results showed that we did obtained an increase of the So intensity with the 6 additional
l3C 1t pulses version, but the more important factor, AS/So, was worse (0.34) than the
result of the all 15N 1!: pulse version (~ 0.38). Therefore, the idea of reducing l3C 1t pulses
in the sequence does not give all the expected results. Other groups have reported using a
total number of 8 1t pulses in the observe channel (13C channel in our case) to take

19, 20

advantage of both sequences . We will conduct further studies to investigate this

approach.

1 '5 N rt pulse version has only one refocusing 13C 1t pulse in

Considering that the al
So and a large number of imperfect 15N 7: pulses in 5,, we wondered whether it is possible

to compensate the 15N pulse imperfection by incorporating these pulses in So as shown

below. Since the corresponding dephasing sequence S1 is the same as the all 15N n pulse

102

version, the new so is different from S1 only by the central ‘5 N 1t pulse which cancels the

 

 

 

 

 

 

¢ .
13C CP, 1 E Acquire
x y x y y x y x X X y X Y Y X
15N
llllllllllLllll
I l l 1 I l l I J
rotor 0 1 2 3 4 5 6 7 8

accumulated l3C—ISN dipolar coupling from the first half of the dephasing period.
Therefore the difference between 80 and 81 sequences is quite small as compared to the
original all 15N 7t pulse version where the numbers of 15N 1t pulses are signiﬁcantly
different in the So and S1 sequences. However, further considerations suggest that this
version should not give better results than the original one. The two versions have the
same S1 sequences and different So sequences, so AS/So will only depend on the
difference in So between the two versions. In the modiﬁed So sequence, only when the
central 15N 1t pulse is a perfect 1t pulse can the dipolar coupling be cancelled and give an
So intensity equals to the So intensity in the original sequence where the dipolar coupling
is not introduced at all. Any imperfection of the central 15N 1t pulse will lead to the
generation of residual dipolar coupling, which will reduce the So intensity in the modiﬁed
version, and thus reduce the AS/So ratio. This was indeed consistent with the
experimental data. The result from the original sequence gave AS/So = 0.38, while the
result from the modified sequence gave reduced So and AS/So = 0.28.

The last thing we investigated was rotor synchronization. In the past we relied on
the spinning controller to provide about 8.000 i 0.002 kHz stability. The rotor period

used in the experiment was calculated based on the speed setting, so the actual rotor

103

period can be slightly different from what was set. Active rotor synchronization detects
the rotor spinning speed in real time and adjusts the rotor period accordingly so that the
actual pulse timing matches better with the rotor timing. We implemented the active rotor
synchronized version of all 15N 1t pulse REDOR and alternating 15N/BC 1t pulse REDOR
and compared with the corresponding original versions. Little difference was found

regarding the AS/So ratio and So intensity.

CONCLUSION

We conducted a series of REDOR experiments to verify the feasibility of this
technique for measuring 13C-‘S’N distance in a peptide. The goal is to detect hydrogen
bonded l3C-‘SN pairs through a 40 Hz coupling which requires a dephasing period of 10
~ 30 ms duration and hundreds of NMR rt pulses in the REDOR experiment. We
therefore expended some effort in optimizing the experiment settings including the
quality of 1t pulse, 1H decoupling sequence and strength, and selection of REDOR
sequence.

Pulse imperfections were found in the 15N channel that were not solely due to the
pulse shape. Also, samples placed at outmost 1/3 layer of the rotor experience
significantly less RF ﬁeld as compared to that of the central 2/3 layer and give worse
REDOR results. This situation is especially poor for a 4 mm rotor. For quantitative
analysis, sample should be restricted into the central 2/3 layer of the rotor.

Although the REDOR experiment should in principle involves only two channels,
adequate proton decoupling is essential to remove the strong proton dipolar coupling and

preserve the 13C magnetization during the dephasing time. To accomplish this goal, the

104

TPPM decoupling sequence was found to be superior relative to regular CW decoupling.
A higher decoupling ﬁeld also gives signiﬁcantly better REDOR dephasing ratio, at least
when a longer heteronuclear distance is probed. It is also important to make the proton
decoupling field at least three times greater than the dilute spin ﬁeld to avoid the cross-
relaxation effect.

In general, the alternating lsN/BC rt pulse REDOR with XY-8 phase cycling
scheme in both channels provides a better way to suppress offset effects and compensate
for pulse imperfections, and gives better results than the all 15 N rt pulse REDOR. The all
13C rt pulse REDOR also gives better AS/So ratio than the all 15N 1: pulse REDOR
probably due to the better pulse perfection in 13C than in 15N. However, the over-
sensitivity to rotor spinning stability reduces the intensity rapidly with an increase in the
dephasing time; thus, this REDOR version is of little use for probing weak dipolar
coupling where a long dephasing time is necessary.

Considering the requirements of our real sample, it is preferable to use a 6 mm
rotor to obtain adequate signal-to-noise in a reasonable amount of time and the
decoupling field is restricted to 65 kHz. Also, since we have adjacent triply labeled l3C
carbonyls in the same peptide, the all 15N 1t pulse REDOR was chosen over the
alternating l5N/13 C rt pulse REDOR to avoid 13C-13 C homonuclear dipolar coupling that
will attenuate the signal. Using these conditions, we obtained a REDOR dephasing curve
smaller than theoretical calculation. In our study, REDOR was used as a semi-
quantitative probe of extent of hydrogen bonding, and therefore the results with all 15N rt
pulse REDOR and 65 kHz TPPM decoupling were used for calibration in the study of

membrane-associated fusion peptide.

105

REFERENCE

10.

11.

12.

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

Hing, A. W., Tjandra, N., Cottam, P. F., Schaefer, J., and Ho, C. (1994) An
investigation of the ligand-binding site of the glutamine-binding protein of
Escherichia coli using rotational-echo double-resonance NMR. Biochemistry 33,
8651-8661.

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

Middleton, D. A., Ahmed, Z., Glaubitz, C., and Watts, A. (2000) REDOR NMR
on a hydrophobic peptide in oriented membranes. J. Magn. Reson. 14 7, 366-370.

Jaroniec, C. P., Tounge, B. A., Herzfeld, J ., and Grifﬁn, R. G. (2001) Frequency
selective heteronuclear dipolar recoupling in rotating solids: Accurate C-13-N-15
distance measurements in uniformly C-l3,N—15-labeled peptides. J. Am. Chem.
Soc. 123, 3507-3519.

Murphy, 0. J., 3rd, Kovacs, F. A., Sicard, E. L., and Thompson, L. K. (2001)
Site-directed solid-state NMR measurement of a ligand-induced conformational
change in the serine bacterial chemoreceptor. Biochemistry 40, 1358-1366.

Thompson, L. K. (2002) Solid-state NMR studies of the structure and
mechanisms of proteins. Curr. Opin. Struct. Biol. 12, 661-669.

Long, H. W., and Tycko, R. (1998) Biopolymer conformational distributions from
solid-state NMR: alpha-helix and 3(10)-helix contents of a helical peptide. J. Am.
Chem. Soc. 120, 7039-7048.

Gullion, T., Baker, D. B., and Conradi, M. S. (1990) New, compensated Carr-
Purcell sequences. J. Magn. Reson. 89, 479-484.

Gullion, T., and Schaefer, J. (1991) Elimination of resonance offset effects in
rotational-echo, double-resonance NMR. J. Magn. Reson. 92, 439-442.

Mehta, A. K., Hirsh, D. J., Oyler, N., Drobny, G. P., and Schaefer, J. (2000)
Carbon-proton dipolar decoupling in REDOR. J. Magn. Reson. 145, 156-158.

Ishii, Y., Ashida, J ., and Terao, T. (1995) 13C-1H dipolar recoupling dynamics in
13C multiple-pulse solid-state NMR. Chem. Phys. Let1246, 439-445.

106

l3.

14.

15.

l6.

l7.

18.

19.

20.

Mitchell, D. J ., and Evans, J. N. S. (1998) Improved heteronuclear decoupling in
REDOR with the use of TPPM. Chem. Phys. Lett. 292, 656-660.

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

Anderson, R. C., Gullion, T., Joers, J. M., Shapiro, M., Villhauer, E. B., and
Weber, H. P. (1995) Conformation of [1-C-13,N-15]Acetyl-L-Carnitine -
rotational-echo, double-resonance nuclear-magnetic-resonance spectroscopy. J.
Am. Chem. Soc. 117, 10546-10550.

Garbow, J. R., and Gullion, T. (1991) Improvements in REDOR NMR
spectroscopy — minimizing resonance-offset effects. J. Magn. Reson. 95, 442-445.

Mehta, A. K., personal communication.

McDowell, L. M., Poliks, B., Studelska, D. R., O'Connor, R. D., Beusen, D. D.,
and Schaefer, J. Rotational-echo double-resonance NMR-restrained model of the
ternary complex of 5-enolpyruvylshikimate-3-phosphate synthase. J. Biomol.
NMR, in press.

O'Connor, R. D., and Schaefer, J. (2002) Relative CSA-dipolar orientation from
REDOR sidebands. J. Magn. Reson. I 54, 46-52.

O'Connor, R. D., Poliks, B., Bolton, D. H., Goetz, J. M., Byers, J. A., Wooley, K.

L., and Schaefer, J. (2002) Chain packing in linear phenol-polycarbonate by C-
l3{H—2} REDOR. Macromolecules 35, 2608-2617.

107

CHAPTER VI

OLIGOMERIZATION STUDIES OF THE MEMBRANE-ASSOCIATED FUSION

PEPTIDE BY REDOR MEASUREMENT S: PARALLEL VS. ANTI-PARALLEL

STRAND ARRANGEMENTS

108

BACKGROUND

In the previous chapters, we demonstrated that the free HIV-1 fusion peptide can
serve as a useful model system for understanding viral/target cell fusion, at least to the
lipid mixing stage. Recent studies suggest that envelope protein regions other than the
fusion peptide also interact with membranes and play a role in fusion”.

For the HIV-1 gp41 envelope proteins, there are atomic-resolution structures of
the “soluble ectodomains” which do not contain the fusion peptide7’“ (cf. Figure 28).
These structures are believed to correspond to the protein conformations after fusion has
occurred and perhaps during fusion as well. In each of these soluble ectodomain
structures, the protein is trimeric and its three N-temiini (corresponding to about residue
30 in the whole envelope protein) are in close proximity at the end of an in-register
helical coiled-coil. These structures end several residues C-terminal of the fusion peptide

and it has therefore been hypothesized that during viral/target cell fusion, at least three

 

Figure 28. Structure of the gp41 soluble ectodomain trimer. (A) Ribbon drawing of the
trimer. (B) An end-on view looking down the 3-fold axis of the trimer.

109

fusion peptides insert into the target cell membrane with their C-termini in close
proximity.

There are ﬂuorescence and ESR data suggesting that the fusion peptides form
oligomeric structures”’”. The possible signiﬁcance of fusion peptide oligomerization is
suggested both by envelope protein trimerization and by experiments and modeling
studies which indicate that the fusion site contains multiple trimers and a corresponding

9' ’0’ I 5 '1 7. In addition, the functionally disruptive Val-2

high fusion peptide concentration
to Glu-Z mutation in the gp41 fusion peptide is trans-dominant, i.e. cells expressing 10%
mutant protein and 90% wild-type protein exhibit only 40% of the fusion activity of cells
with 100% wild-type protein”. One interpretation of these data is that the mutant peptide
disrupts the correct assembly of a functionally important fusion peptide oligomer” 14.

In chapter three and four, we described the solid-state NMR studies on the HIV-1
fusion peptide and demonstrated that the peptide ad0pts a predominantly B strand
structure when associated with membranes whose lipid head group and cholesterol
composition is comparable to that found in the virus infected host cells. In chapter ﬁve,
the solid—state NMR technique, REDOR spectroscopy, was shown to be feasible for
detecting the 13C—lSN distance in a hydrogen bond, at least semi-quantitatively. In the
present study, oligomeric structure in the membrane-bound HIV-1 fusion peptide was
investigated via REDOR measurements which probe inter-peptide hydrogen bonding
through samples containing a 1:1 mixture of 13C carbonyl labeled peptide and 15N amide
labeled peptide. In addition, the relative populations and structures of parallel and

antiparallel strand arrangements in the oligomers were probed using a variety of samples

which differed by the residue positions of the 13C labeling in one peptide and/or the 15N

110

labeling in the other peptide. The REDOR experiments demonstrate the presence of a
large population of membrane-associated fusion peptide oligomers with approximately

equal weighting of parallel and antiparallel strand arrangements.

MATERIALS AND METHODS

Solid-State NMR Sample Preparation. The samples were prepared using method
1. In brief, peptide was dissolved in ~ 35 ml of 5 mM HEPES, pH 7.0. Peptide
concentrations were determined by the BCA assay which had been calibrated using
quantitative amino acid analysis. For samples containing two differently labeled peptides,
each peptide was dissolved separately and the two solutions were then combined. Two
kinds of lipid/cholesterol mixtures reﬂecting the approximate lipid and cholesterol
content of the HIV-1 virus and its target T-cells were used: LM-3 had POPC, POPE,
POPS, sphingomyelin, PI and cholesterol in a lO:5:2:2:l:10 molar ratio, and LM-4 had
the same components except the POPC was replaced by DPPC-l—l3 C. LUVs of LM-3 or
LM-4 were prepared in ~ 2 ml buffer. The peptide and LUV solutions were then mixed
and kept at room temperature overnight to ensure maximum peptide/lipid binding.
Subsequent ultracentrifugation of the solution at 100,000 x g for four hours pelleted the
membrane and associated bound peptide, and left unbound peptide in the supernatant.
Peptide binding was typically quantitative as determined by BCA assay measurements of
negligible peptide concentration in the supernatant. The peptide/lipid pellet formed after
ultracentrifugation was transferred by spatula to a 6 mm diameter MAS NMR rotor.

Solid-State NMR Spectroscopy. Experiments were done on the 9.4 T Varian

Infinity Plus spectrometer using a 6 mm triple resonance MAS probe. The detection

111

channel was tuned to 13C, the decoupling channel was tuned to 1H, and the third channel
was tuned to '5 N. 13C and 15N chemical shifts were externally referenced to the methylene
resonance of adamantane at 38.2 ppm and (NH4)ZSO4 at 20 ppm, respectively.
Experiments used a MAS frequency of 8.000 i 0.002 kHz. In the NMR probe circuit, the
RF ﬁelds are attenuated at the ends of the coil. Hence, for the FPK3 samples, longer
spacers were used to restrict samples to the 2/3 central volume (~ 160 pl). For the other
samples, the full volume of the rotor was used. Because the comparison of these two
volumes in 6 mm rotor showed negligible difference, direct comparison could be made
between all of the samples. The chief advantage of the volume restriction is lower sample
requirement.

In REDOR experiments to detect hydrogen bonds, between 1 and 2 ms of CP at
50 kHz was followed by a 24.0 ms dephasing period and then direct 13C detection. The all
15N 11: pulse REDOR was applied with a single 50 kHz l3C refocusing 1t pulse was placed
at the center of the dephasing period and 1H TPPM decoupling of 65 kHz was applied
during both dephasing and detection. The 13C transmitter was set to 155 ppm and the 15N
transmitter was set to 115 ppm. For the 81 acquisition, the dephasing period contained a
40 kHz 15N 1: pulse at the middle and end of each rotor cycle, while the So acquisition did
not contain these pulses. XY—8 phase cycling was used for the 15N pulses.

The spectrometer parameters were optimized with the lyophilized MB(i+4)EK-
A9CA13N model helical peptide as described in chapter five. The data were processed
with 100 Hz line broadening and with ﬁfth order polynomial baseline correction.

REDOR Data Analysis and Simulations. The experimental AS/So is obtained

following the equation 4. In addition, for each pair of sols, spectra, an experimental

112

uncertainty 0 was calculated as the root-mean-squared-deviation of integrated intensities
in twenty-four regions of the spectra without signal. OAS/so, the uncertainty in AS/So, was

calculated using the formula”:

 

owso = o x sl/s0 x \[(l/So)2 + (1/3,)2 (6)

Simulations of the REDOR experiment were carried out as a function of peptide
structure and dephasing period using the SIMPSON program”. The experimental pulse
sequence, MAS frequency, 13C and 15N RF ﬁelds and pulse lengths, and 15N phases were
incorporated into the simulations. Carbonyl chemical shift anisotropy (CSA) principal
values were set to be 250, 190, and 90 ppm, and the principal axis directions were taken
from the literature with the following orientations: 833 perpendicular to the peptide plane
and 522 at an angle x = 130° from the C—N bond”. 1H were not considered in the

simulations.

RESULTS

REDOR Simulations of Multiple ”N Coupled to 13C. Although it is possible to
obtain the numeric calculation of REDOR dephasing ratio accurately based on equation 5
for a single l3C—lSN pair, it is difﬁcult to calculate the result when one 13C is coupled to
multiple 15N nuclei, as is the case in the real peptide samples. The SIMPSON program, a
simulation program for solid-state NMR was applied to provide information about the
multiple coupled REDOR dephasing ratio.

First we compared the result of single l3C—'5N pair from SIMPSON with that
from a previous numeric calculation and a satisfying agreement was observed in Figure

28, indicating that the program can be used for further investigation. The simulated

113

 

REDOR dephasing curves were found approximately independent of the carbonyl CSA

principal axis geometry, which is also consistent with the simulations in other studies”.

 

 

 

 

r l l r
1.0 - 1
J
0.8 - '-
1
o 0.6- -
(L) .
(D
< 0.4 -i -
0.2 d ..
0.0 1 ..
r r . r I ,
0.0 0.4 0.8 1.2
A = D*T

Figure 29. REDOR simulation from the SIMPSON pro am (triangles) and numeric
calculation (solid line). A 44.4 Hz dipolar coupling D (4.1 ) was used in both cases.

114

 

Next we calculated the case where one 13C is coupled not only to the hydrogen
bonded 15N, but also to the adjacent two other 15N as in the case of triply 15N labeled
fusion peptide samples. Since we are probing the possible hydrogen bonds among
peptides forming parallel or anti-parallel beta sheets, the simulation was conducted with
the coordinates from the crystal structure of the parallel and anti-parallel segments of
streptococcal protein G Bl domain (PDB ﬁle ligd)”. Each simulation also used a deﬁned
register of the 13C and 15N labeled residues in neighboring strands. Figure 29 shows that
according to the typical structure of parallel or anti-parallel arrangements, 13C that
couples with two additional adjacent 15N generally experiences more dephasing than if it
only dephases by the hydrogen bonded '5N (open symbols). Not surprisingly we also
found little difference exists between parallel and anti-parallel arrangements since the
two geometries are quite similar. To get a more realistic estimate, we expanded the
typical hydrogen bond distance from 4.1 A to 4.6 A based on the REDOR result of the
MK(i+4)-A9CA13N peptide and expanded the other two l3C-lf’N distances in parallel or
anti-parallel arrangements with the same ratio (4.6/4.1) and redid the simulation. The
ﬁlled symbols in Figure 29 are the corresponding results. Again we found similar results
as with the regular distances. Using these results, at 24 ms dephasing time, we would
obtain a dephasing ratio of ~ 0.5 exclusively for the 13C coupled to three 15N, which will

be used for the semi-quantitative analysis in real samples.

115

 

(ﬂ

 

 

 

 

 

anm Nwem
0 35° N A c 26° c
(4.1) ’
w° .\7§gNMJM
N (5.5 A) N (6.5 A)
Anti-parallel arrangement Parallel arrangement
(M
T j l l l
1.0- a O 9 3 -
¢°38
08- ﬁcgoO . ~
V 0 I . 0
“too 1‘ 0
o -l ‘ C . -
(D 0.6 a U 0 0 ‘ I .
(7) u o . ‘ o . 1
< 3 o O . I o '
0.4“ a a o I 0 . "
j 300 1‘0.
#30 t‘:..
O ..
0.24 8 g 0 ‘ g : .
. a g g ' ' ' .
OD-tt“ -
l j l ' l ' j
0 8 16 24 32
T (ms)

Figure 30. (a) Coordinates of one 13C coupled to three adjacent 15N in parallel and anti-
parallel segments of streptococcal protein G Bl domain with the carbonyl carbon
hydrogen bonded to the middle amide nitrogen. (b) SIMPSON simulated REDOR
dephasing ratios of the '3 C dephased by only hydrogen bonded 15N (open circles), or two
more adjacent 15N in parallel strand arrangement (open upper triangles), or two more
adjacent 15N in anti-parallel strand arrangement (open lower triangles). The ﬁlled
symbols represent the corresponding simulation results when all the distances are
extended by a factor of 4.6/4.1 as the hydrogen bond distance expands from 4.1 A to 4.6
A based on the REDOR result of MK(i+4)-A9CA13N peptide at 24 ms dephasing time.

116

 

REDOR Control Experiments. We have discussed the optimization of the REDOR
experiment in Chapter V. One important reason to choose the all 15 N 1: pulse REDOR is
the consideration of sensitivity when the sample contains triply 13C labeled peptide.
Figure 30 demonstrates that using this version of REDOR on such a sample, the So signal
at 24.0 ms dephasing time was about 1/8 that obtained with direct CP. With the
alternating 13C/ISN 1: pulse version of REDOR, there was additional attenuation by a
factor of ~ 0.6, which may be a result of it pulse recoupling of l3C—'3C homonuclear
interaction324'26. In our study, REDOR was used as a semi—quantitative probe of the
extent of hydrogen bonding, and we chose to use the higher signal-to-noise all 15N 1:
pulse REDOR. In general, as the dephasing period increases, AS/So also increases,
whereas the overall signal decreases. The chosen 24.0 ms dephasing period represents a

compromise between these two competing effects.

(a) (b)

 

 

I l I l I I I I
190 180 170 160 190 180 170 160
ppm

Figure 31. 13C solid-state NMR spectra of the FPK3-(G5A6L7)c/FPK3-
(A14A15G16)N/LM-3 sample from (a) CP experiment and (b) So REDOR experiment
with a 24.0 ms dephasing time. The spectral intensities in each experiment were divided
by the number of scans in that experiment. The vertical scale in (b) is 8 times that in (a),
showing that the signal/scan in (b) is ~ 1/8 as large as that in (a). The (b) spectrum
appears better resolved than the (a) spectrum because the natural abundance lipid
carbonyls make a relatively smaller contribution to the So signal than to the CP signal.

117

 

Table 9 lists AS/So values from a series of REDOR control experiments and
Figure 31 displays some of the corresponding experimental spectra. The goal of the
control experiments was to quantify (AS/So)backgmund, representing the dephasing ratio in
the membrane-associated peptide samples that was not due to the labeled l3C-lSN
hydrogen bonded pairs. For samples containing only 13C labeled and/or natural
abundance peptide (cf. Figure 31a and 31e), the AS/So value is in the —0.01 ~ 0.03 range
and is due to the 0.0037% 15N natural abundance and the resulting small population of
carbonyls with nearby 15N. For example, a 13C with 15N one or two bonds away on the
peptide backbone will have an individual AS/So z 1 at 24.0 ms dephasing time. For a
sample containing a 1:1 mixture of 15N labeled and unlabeled peptide (cf. Figure 31b),
AS/So ~ 0.2, while for a sample containing only 15N labeled peptide, AS/So ~ 0.4. These
relatively large AS/So values are likely a result of the signiﬁcant fraction of natural
abundance carbonyls which are close to the labeled 15N. For the samples used to probe
fusion peptide strand arrangements, there was a 1:1 mixture of 13C and 15N labeled
peptides and a ~ 10% natural abundance peptide contribution to the total 13C signal. In
these samples, we would expect therefore a ~ 0.02 contribution to AS/So from natural
abundance peptide '3 C.

In samples containing 50% 13C carbonyl labeled lipid (“LM-4”) and either lsN
labeled peptide or unlabeled peptide, AS/So ~ 0.03 or 0.01, respectively (cf. Figure 3d). In
addition, when CP and So intensities of these samples were compared to those of samples
containing unlabeled lipid and 13C labeled peptide, it was observed that the CP signal per
13C is about three times less for a lipid carbonyl than for a peptide carbonyl while the So

signal per 13C is about nine times less for a lipid carbonyl than for a peptide carbonyl. By

118

combining this latter result with the observation that AS/So ~ 0.4 for the sample
containing unlabeled lipid and 15N labeled peptide, it is possible to account for the AS/So
values in the 13C lipid labeled samples by only considering dephasing of the natural

abundance peptide carbonyl signal. This analysis suggests negligible dephasing of the

 

 

lipid carbonyls.
Table 9. REDOR Control Experiments.

Sample“ AS/So” Figure
FP-(F8L9G10)c/FP-unlabeled/LM-3 0.01 (0.01) 31a
FP-(F8L9G10)N/FP-unlabeled/LM-3 0.20 (0.05) 31b

FPK3-unlabeled/LM-3 -0.01 (0.06) 31c
FP—(F8L9G10)N/LM-4 0.03 (0.02) 31d
FP-(F8L9G10)N/LM-3 0.41 (0.05) n.s.“
FP-unlabeled/LM-4 0.01 (0.03) n.s.
FP-(F11L12G13)C 0.03 (0.02) n.s.
FPK3—(A14A15G16)N 0.35 (0.02) n.s.
FPK3-unlabeled 0.00 (0.02) n.s.

 

“ Samples with LM-3 or LM-4 contained peptide bound to hydrated membranes. All
other samples were lyophilized dry peptide. For membrane-associated peptide samples,
the overall peptide:lipid molar ratio was ~ 1:25 and for samples containing two
peptides, the peptidezpeptide molar ratio was ~ 1:1.

b The uncertainty is enclosed in parentheses.

C not shown in a figure.

119

 

 

Len

I I I l I l l I
190 180 170 160 190 180 170 160

PPm

 

 

 

 

Figure 32. 13C solid-state NMR So (left) and S1 (right) REDOR spectra of membrane-
associated fusion peptides that had (a) ~ 0.8 umol each of FP-(F8L9G10)c and PP-
unlabeled, (b) ~ 0.8 umol each of FP-(F8L9G10)N and FP-unlabeled, (c) ~ 1.6 limo]
FPK3-unlabeled, and ((1) ~ 1.6 mol FP-(F8L9G10)N. Each of the samples in (a-d) also

contained 40 umol lipid and 20 umol cholesterol. For samples (a-c), LM-3 lipid mixture
was used and for sample ((1), LM-4 lipid mixture was used.

120

The overall conclusion of the control experiments is that (AS/SO)background ~ 0.03
for the LM-3 samples with a 1:1 mixture of labeled 13C and 15N peptides, which is ~ 2/3
due to dephasing of natural abundance peptide carbonyl by labeled 15N and ~ 1/3 due to
dephasing of labeled peptide carbonyl by natural abundance 15N. The contribution of the
lipid carbonyl to (AS/So)backgmund is negligible.

REDOR Experiments to Probe Inter-Peptide Hydrogen Bonding. Figure 32a—h
displays the So and S1 REDOR spectra of membrane-associated fusion peptide samples
and the associated AS/So values are listed in Table 10. Each sample was made with LM-3
lipid mixture (40 umol lipid and 20 umol cholesterol), ~ 0.8 umol peptide with selective
l3C carbonyl labeling, and ~ 0.8 umol peptide with selective l5N amide labeling.

Four samples (cf. Figure 32a-d) were prepared to probe for the presence of an
approximate in-register parallel arrangement of peptide strands. In each of these samples,
the labeling in the 13C peptide and in the 15N peptide was at the same set of three
sequential residues. With these four samples, inter-strand hydrogen bonding of parallel
strands was probed between Gly-5 and Gly-16. For the samples labeled between Gly-S
and Gly-l3, the AS/So values were in the 0.11 - 0.13 range which is signiﬁcantly above
the 0.03 background value. For the sample containing peptide labeled between Ala-14
and Gly-16, AS/So ~ 0.03 which is the approximate background signal. Thus, it appears
there is a population of peptides with parallel arrangements of strands, and this
arrangement frays apart towards the C-terminal side of the peptide. These results are
schematically presented in Figure 33a and are consistent with a reasonable biophysical
model based on the peptide sequence. In the sequence, the N -terminal and central regions

only contain apolar residues while the C-terminal region contains a substantial fraction of

121

polar or charged residues. For extended peptides in an oligomeric parallel strand
arrangement, the apolar regions of all strands would be on one end of the oligomer and
the polar regions would be on the other end of the oligomer. The apolar end of the
oligomer may insert into the membrane so that in this region, the strands must hydrogen
bond to one another, while the polar end of the oligomer is likely outside the membrane
where the strands would hydrogen bond to water. This model is supported by previous
measurements of carbonyl linewidths along the peptide backbone. In the N-terminal and
central regions, the linewidths are 2 - 3 ppm, which indicates a narrow distribution of
secondary structures with a regular hydrogen bonding pattern whereas in the C-terminal
region, the linewidths are ~ 5 ppm, which indicates random coil structure and irregular
hydrogen bonding.

Four samples (cf. Figure 32e-f) were also prepared to probe for anti-parallel
arrangement of peptide strands. The midpoint of the arrangement was chosen to be
between Gly-lO and Phe-l l, which is about the middle of the peptide. For example, the
sample used for Fig. 32h contained peptide which was l3C labeled at Ala-14, Ala-15, and
Gly-16, and peptide which was 15N labeled at Gly-5, Ala-6, and Len-7. For these four
“anti-parallel arrangement” samples, the AS/So values were in the 0.08 ~ 0.16 range, and
were significantly above the background level. Thus, in addition to the previously
described parallel strand population, there also appears to be an anti-parallel strand
population whose inter-peptide hydrogen bonding continues at least from Gly-5 to Gly-
16. This result is schematically presented in Fig. 33b. For extended peptides in an
oligomeric anti-parallel strand arrangement, there would be both polar and apolar ends of

strands on each end of the oligomer, which is qualitatively different than what is expected

122

for a parallel strand arrangement. Because of this mixed polarity, it is less clear whether

an anti-parallel oligomer would just stay on the surface or insert through the membrane.

Table 10. REDOR Measurements to Probe Inter-Peptide Hydrogen Bonds.

 

Sample " AS/So" Figure

 

FPK3-(G5A6L7)c/FPK3-(G5A6L7)N/LM-3 0.13 (0.01) 32a

FP-(F8L9G10)c/FP-(F8L9Gl0)N/LM-3 0.13 (0.01) 32b
Parallel
Arrangement FP-(F11L12G13)c/FP-(Fl1L12Gl3)N/LM-3 0.11 (0.01) 32c
Probed

FPK3-(A14A15G16)C/FPK3- -01 (093) 32d

 

.11 (0.01) n.s.

FPK3-(G5A6L7)c/FPK3-(A14Al5G16)N/LM-3 ‘ i:
.12 (0.02) 32e

Anti-parallel

Arrangement FP-(F8L9G10)c/FP-(Fl 1L12Gl3)N/LM-3 0.08 (0.02) 32f

Probed
FP-(Fl 1L12Gl3)c/FP-(F8L9G10)N/LM-3 0.09 (0.02) 32g

FPK3-(A 14A 15G16)c/FPK3-(G5A6L7)N/LM-3 0.16 (0.02) 32h

 

“ The overall peptide:lipid mol ratio was ~ 1:25.

b The uncertainty is enclosed in parentheses.

‘ Two samples were prepared and measured separately.
d not shown in a figure.

123

r F F ‘r

 

F F F %

 

 

 

 

l I l I

I F F I I l I I I I I F
190 180 170 160 190 180 170 160 190 180 170 160 190 180 170 160

ppm ppm

Figure 33. 13C solid-state NMR So and S1 REDOR spectra of membrane-associated
fusion peptides. Each sample was made with LM-3 lipid mixture, ~ 0.8 mol peptide
with selective 13'C carbonyl labeling and ~ 0.8 umol peptide with selective 15N amide
labeling. The specific labeling schemes were: (a) FPK3-(G5A6L7)c/FPK3-(G5A6L7)N,
(b) FP-(F8L9G10)c/FP-(F8L9G10)N, (c) FP-(Fl 1L12G13)c/FP-(F11L12G13)N, (d)
FPK3-(A14A15G16)C/FPK3-(A14A15G16)N, (e) FPK3-(G5A6L7)C/FPK3-
(A14A15G16)N, (f) FP-(F8L9G10)c/FP-(Fl1L12G13)N, (g) FP-(F11L12G13)c/FP-
(F8L9G10)N, and (h) FPK3-(A14A15Gl6)C/FPK3-(G5A6L7)N. Samples (a)-(d) primarily
probed the presence of parallel strand arrangements and samples (e)-(h) primarily probed
the presence of anti-parallel strand arrangements. The dotted lines are included for visual
comparison of the spectral intensities.

124

(a) A160} m
5GALFLGFLG A m
5GALFLGFLGA A

16G

 

(b)

5GALFLG FLGAA'SG :2»
<2 16GAAG LFG LFLAsG

 

Figure 34. Peptide oligomerization and membrane insertion models for the (a) parallel
strand arrangement, and (b) anti-parallel strand arrangement. The arrows indicate the
peptide direction from N-terminus to C-terminus while the numbers label the indices of
individual residues. In the parallel arrangement, inter-peptide hydrogen bonding was
observed at N—terminal and central regions, and considerably weaker toward the C-
terrninus, which is consistent with an apolar region of peptide insertion into membrane
interior model. In the anti-parallel arrangement, inter-peptide hydrogen bonding was
detected along the whole length of the probed region. Since this arrangement would place
the polar region of one peptide and the apolar region of the other peptide at the same side,
it is unclear whether it is a result of peptides binding on the surface of membrane or
peptides inserting through the membrane.

125

Because most samples probing for parallel and antiparallel strand arrangements
had comparable AS/So values, it is likely that there are roughly equal populations of each
arrangement. Such arrangements could exist for fully extended peptides in a [3 sheet or [3
barrel structure or they could exist for peptides in a B hairpin structure. For the present
inter-peptide hydrogen bonding study, we also considered whether AS/So values
predicted for a totally extended oligomeric B strand model matched the experimental
AS/So values. If the experimental values are signiﬁcantly less than the predicted values,
then there may be a population of hairpin peptides with intra-peptide hydrogen bonding.

In the extended peptide model, we considered that the total So carbonyl signal had
contributions from labeled 13C in the peptide, natural abundance 13C in the peptide, and
natural abundance 13C in the lipid. These contributions are denoted Soc, SON, and S01“,
respectively. There are corresponding contributions Slc, SIN, and S11” to the total 81
signal. In quantifying l3C, one 13C labeled peptide, one 15N labeled peptide, and 50 lipid
molecules are considered in the sample. The 13C peptide has three labels and SoC is set to
3.0. The following calculations were made for the other So signals.

SON = (number of unlabeled peptide carbonyl C) x (natural abundance '3 C)

= 49 x 0.011 = 0.539 (7a)
SoL = (number of lipid) x (number of lipid carbonyl C per lipid molecule) x
(signal per lipid 13C/signal per peptide 13C) x (natural abundance 13C)
= 50x2x0.11x0.011= 0.121 (7b)
For the SON calculation, we considered the FPK3 peptide with 26 residues. As described

in the “REDOR Control Experiments” section, SIL : SOL. Considering the spectrum

126

 

 

 

displayed in Fig. 31b, we approximate that SIN z 0.8 x SON = 0.431. Using the
simulation result of the labeled l3C dipole-coupled to three adjacent three 15N, we
approximate that SIC z 0.50 x SOC for the case where all three labeled 13'C on one strand
are hydrogen bonded with I5N on adjacent strands. Because the experimental samples
contain a 1:1 mixture of 13C and 15N labeled peptide, there is only 50% probability that a
13C labeled strand will be adjacent to a 15N labeled strand. Therefore, a corrected SIC z
0.75 x SOC = 2.25 was used in the calculations.

Using all of these results, the total AS/So for the extended strand model was

calculated:

AS/So l —(S1/So)
= 1— (2.25 + 0.431 + 0.121)/(3.0 + 0.539 + 0.121) = 0.23 (8)

As described in the “REDOR Control Experiments” section, (AS/Sonackgmund ~ 0.03, so
the labeled l3C-"sN hydrogen bond contribution to dephasing, (AS/So)mode1, is ~ 0.20.
Comparison with experimental data was made using results from the FPK3-
(G5A6L7)c/FPK3-(G5A6L7)N/LM-3 and FPK3-(G5A6L7)c/FPK3-(A14A15A16)N/LM-
3 samples as measures of (AS/So)exp,,-pmuel and (AS/So)expt,-mpmuel, respectively. After
subtracting (AS/SMbackEmnd from each of the experimental AS/So, we obtain values of
(AS/So)cxpt,-pmucl z 0.10 and (AS/So)cxpt.-antiparallcl z 0.08. The total hydrogen bonding
component of AS/So for the FPK3-(G5A6L7)c peptide, (AS/So)exp,,, is ~ 0.18, which is
comparable to (AS/Sammie]. This result suggests that the extended oligomeric structures
account for much of the membrane-associated peptide population. The 10-15% difference

between (AS/Soho“; and (AS/So)expt_ could be due to uncertainties in the experimental

data and model, different alignments of peptide strands than the fully in-register

127

 

arrangements used for the model, and/or a population of [3 turn peptides with intra-
peptide hydrogen bonding. Intra—peptide hydrogen bonding was not probed in the

experiments of this study.

DISCUSSION AND CONCLUSION

In this chapter, we applied solid-state NMR internuclear distance measurements
to develop a structural model for the non-helical form of the membrane-associated HIV-1
fusion peptide. This work was based on previous solid-state NMR chemical shift and 2D
exchange measurements which showed that the peptide is predominantly non-helical
when associated with membranes whose lipid headgroup and cholesterol composition is
close to that found in the virus and its target T cells. One of the possible non-helical
structures is oligomeric [3 strands held together by inter-peptide hydrogen bonding and we
did experiments to detect this type of structure. In membrane-associated fusion peptide
samples containing a 1:1 mixture of 13C carbonyl labeled and 15N amide labeled peptide,
detection of l3C-ISN proximity by solid-state NMR REDOR measurements strongly
supports this structural model. Moreover, the semi-quantitative agreement between the
AS/So values predicted by the model and the experimental AS/So values suggests that the
oligomeric B strand structure represents a substantial population of the membrane-
associated peptide. Solid-state NMR experiments are currently underway to test whether
all of the peptide is fully extended or whether there is also a population of peptide with
hairpin structure.

A second important result is that the data were consistent with roughly equal

populations of peptide in parallel or antiparallel strand arrangements. Inter-strand

128

hydrogen bonding was observed from Gly-S through Gly-16 in the antiparallel
arrangement but only from Gly-5 through Gly-13 in the parallel arrangement. C-terrninal
fraying in the parallel arrangement is consistent with previous solid-state NMR linewidth
measurements and with a reasonable biophysical model in which the apolar N-terminal
and central regions of the oligomer insert into the membrane while the polar C-terminal
region is outside the membrane. Experiments by other groups also suggest that the C-
terminal region is not membrane-inserted, although some of these measurements suggest
that this region is helical, rather than the random coil structure indicated by our
previously published experimentsz. These different structures may be a result of the
different lengths of the C-terminal regions used in the different studies.

The present study also provides some insight into oligomer formation. In chapter
two, the analytical ultracentrifugation data strongly suggest that the 26-residue FPK3
construct is predominantly monomeric in the buffer solution, so formation of membrane-
associated B strand oligomers can be a direct result of peptide/membrane interaction. In
addition, the analytical ultracentrifugation data suggest that the 23-residue FP construct
forms a mixture of oligomeric states in the buffer solution. In the initial step of PP solid-
state NMR sample preparation, separate solutions of 13C and 15N labeled peptide were
made and contained either 13C or ”N oligomers. After FP/membrane association, the
solid-state NMR REDOR data showed signiﬁcant AS/So values (comparable to those of
FPK3 samples), which means that the membrane-associated FP oligomers were a
molecular mixture of '3 C and 15N labeled peptides. This transformation from separated to
mixed oligomers could have been a result of exchange in aqueous solution. However, the

13C and 15N peptide solutions were mixed only briefly prior to addition of liposomes, and

129

 

we believe that it is more likely that the FP oligomers disassemble and reassemble upon
interaction with the membrane. This disassembly/reassembly model is consistent with the
experimental observation that an inﬂuenza fusion peptide which is oligomeric in aqueous

27, 28

solution can be converted to a helical structure upon membrane binding . It is also

consistent with ideas and experiments from other groups” 29; in particular, the
fusogenically disruptive Val-2 to Glu-2 point mutation in the HIV-1 fusion peptide was
correlated with formation of peptide dimers rather than the larger oligomers found for the
wild-type peptide”. It was therefore hypothesized that the mutant peptide prevented the
proper membrane-mediated assembly of peptides required for fusion activity. This idea
was supported by studies showing that viruses which had 90% wild-type/10% mutant
gp41 had only 40% of the infectivity of viruses which had 100% wild-type gp41”.

In summary, solid-state NMR has been applied to detect oligomeric B strand
structure in the membrane-associated HIV-1 fusion peptide as a result of peptide/lipid
interaction. A roughly equal mixture of parallel and antiparallel strand arrangements was
observed with fraying of the strands towards the C-terminus in the parallel arrangement.

These studies provide a basis to develop a more reﬁned model of the peptide structure

and to understand how it may catalyze membrane fusion.

130

 

REFERENCE

10.

11.

Epand, R. F., Macosko, J. C., Russell, C. J ., Shin, Y. K., and Epand, R. M. (1999)
The ectodomain of HA2 of inﬂuenza virus promotes rapid pH dependent
membrane fusion. J. Mol. Biol. 286, 489-503.

Peisajovich, S. G., Epand, R. F., Pritsker, M., Shai, Y., and Epand, R. M. (2000)
The polar region consecutive to the HIV fusion peptide participates in membrane
fusion. Biochemistry 39, 1826-1833.

Peisajovich, S. G., Samuel, 0., and Shai, Y. (2000) Paramyxovirus F1 protein has
two fusion peptides: implications for the mechanism of membrane fusion. J. Mol.
Biol. 296, 1353-1365.

Suarez, T., Gallaher, W. R., Agirre, A., Goni, F. M., and Nieva, J. L. (2000)
Membrane interface-interacting sequences within the ectodomain of the human
immunodeficiency virus type 1 envelope glycoprotein: putative role during viral
fusion. J. Virol. 74, 8038-8047.

Sackett, K., and Shai, Y. (2002) The HIV -1 gp41 N-terrninal heptad repeat plays
an essential role in membrane fusion. Biochemistry 41, 4678-4685.

Peisajovich, S. G., Blank, L., Epand, R. F., Epand, R. M., and Shai, Y. (2003) On
the interaction between gp41 and membranes: The immunodominant loop
stabilizes gp41 helical hairpin conformation. J. Mol. Biol. 326, 1489-1501.

Chan, D. C., Fass, D., Berger, J. M., and Kim, P. S. (1997) Core structure of gp41
from the HIV envelope glycoprotein. Cell 89, 263-273.

Tan, K., Liu, J., Wang, J., Shen, S., and Lu, M. (1997) Atomic structure of a
therrnostable subdomain of HIV-1 gp41. Proc. Natl. Acad. Sci. U. S. A. 94,
12303-12308.

Weissenhom, W., Dessen, A., Harrison, S. C., Skehel, J. J., and Wiley, D. C.
(1997) Atomic structure of the ectodomain from HIV-1 gp41. Nature 387, 426-
430.

Caffrey, M., Cai, M., Kaufman, J., Stahl, S. J ., Wingﬁeld, P. T., Covell, D. G.,
Gronenbom, A. M., and Clore, G. M. (1998) Three-dimensional solution structure
of the 44 kDa ectodomain of SIV gp41. EMBO J. 17, 4572—4584.

Yang, 2. N., Mueser, T. C., Kaufman, J ., Stahl, S. J ., Wingﬁeld, P. T., and Hyde,

C. C. (1999) The crystal structure of the SIV gp41 ectodomain at 1.47 A
resolution. J. Struct. Biol. 126, 131-144.

131

 

 

12.

13.

14.

15.

16.

17.

18.

19.

20.

21.

22.

Gordon, L. M., Curtain, C. C., Zhong, Y. C., Kirkpatrick, A., Mobley, P. W., and
Waring, A. J. (1992) The arnino-terminal peptide of HIV-1 glycoprotein 41
interacts with human erythrocyte membranes: peptide conformation, orientation
and aggregation. Biochim. Biophys. Acta 1139, 257-274.

Kliger, Y., Aharoni, A., Rapaport, D., Jones, R, Blumenthal, R., and Shai, Y.
(1997) Fusion peptides derived from the HIV type 1 glyc0protein 41 associate
within phospholipid membranes and inhibit cell-cell Fusion. Structure- function
study. J. Biol. Chem. 272, 13496—13505.

Pritsker, M., Rucker, J ., Hoffman, T. L., Doms, R. W., and Shai, Y. ( 1999) Effect
of nonpolar substitutions of the conserved Phell in the fusion peptide of HIV-1

gp41 on its function, structure, and organization in membranes. Biochemistry 38,
11359-11371.

Munoz-Barroso, 1., Durell, S., Sakaguchi, K., Appella, E., and Blumenthal, R.
(1998) Dilation of the human immunodeﬁciency virus-1 envelope glycoprotein
fusion pore revealed by the inhibitory action of a synthetic peptide from gp41. J.
Cell Biol. 140, 315-323.

Bentz, J. (2000) Minimal aggregate size and minimal fusion unit for the ﬁrst
fusion pore of inﬂuenza hemagglutinin-mediated membrane fusion. Biophys. J.
78, 227-245.

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

Freed, E. O., Delwart, E. L., Buchschacher, G. L., Jr., and Panganiban, A. T.
(1992) 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. 89, 70-74.

Bevington, P. R., and Robinson, D. K. (1992) Data Reduction and Error Analysis
for the Physical Sciences, McGraw-Hill, Boston.

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

Oas, T. G., Hartzell, C. J ., McMahon, T. J., Drobny, G. P., and Dahlquist, F. W.
(1987) The carbonyl C-13 chemical-shift tensors of 5 peptides determined from

N-15 dipole-coupled chemical-shift powder patterns. J. Am. Chem. Soc. 109,
5956-5962.

Balbach, J. J., Yang, J., Weliky, D. P., Steinbach, P. J., Tugarinov, V., Anglister,
J., and Tycko, R. (2000) Probing hydrogen bonds in the antibody-bound HIV-1

132

 

 

23.

24.

25.

26.

27.

28.

29.

gplZO V3 loop by solid state NMR REDOR measurements. J. Biomol. NMR 16,
313-327.

Derrick, J. P., and Wigley, D. B. (1994) The third IgG-binding domain from
streptococcal protein G. An analysis by X-ray crystallography of the structure
alone and in a complex with Fab. J. Mol. Biol. 243, 906-918.

Gullion, T., and Vega, S. (1992) A simple magic angle spinning NMR experiment
for the dephasing of rotational echoes of dipolar coupled homonuclear spin pairs.
Chem. Phys. Lett. 194, 423-428.

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

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

Yang, J., Parkanzky, P. D., Bodner, M. L., Duskin, C. G., and Weliky, D. P.
(2002) Application of REDOR subtraction for ﬁltered MAS observation of
labeled backbone carbons of membrane-bound fusion peptides. J. Magn. Reson.
159, 101-110.

Parkanzky, P. D., and Weliky, D. P., unpublished experiments.

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

133

 

 

 

CHAPTER VII

SUMMARY AND FUTURE DIRECTIONS OF THE RESEARCH

134

 

 

The ultimate goal of this research on the fusion peptide is to gain insight into its
role in inducing/catalyzing viral/host cell fusion. The earlier part of the work involves the
examination of the biological functionality of various synthetic fusion peptide constructs
and their structures prior to the binding of membrane lipids. These studies demonstrate
that the free peptide without the rest of the gp41 protein can serve as a good model to
understand the fusion mechanism. Sample preparation studies conﬁrm that the structure
we observed in solid-state NMR is indeed the membrane-associated structure and
biologically relevant.

The structural plasticity of membrane-associated fusion peptides has been
observed by several groups using a variety of experimental probes and it is known that
they can exist in well-structured helical or non—helical forms" 2. Other investigators have
developed structural models for the helical form of the peptides based principally on
solution NMR, ESR, and infrared data”. In the next step of our work, we applied various
solid-state NMR techniques (chemical shift analysis of 1D CP/MAS experiments and
peptide dihedral angle analysis from 2D exchange experiments) that provided strong
evidence for the presence of an extended B strand conformation in the N-terminal and
central regions of the membrane-bound peptide. The C-terminus is more disordered.

In the fusogenic atomic-resolution structure of the gp41 ectodomain, the protein is
trimeric and its three N-terrnini are in close proximity at the end of an in-register helical
coiled-coil. These structures end several residues C-terminal of the fusion peptide and it
has therefore been hypothesized that during viral/target cell fusion, at least three fusion
peptides insert into the target cell membrane with their C-termini in close proximity's'm.

The observation of B strand conformation also suggests that the close proximity may be

135

 

the result of forming peptide oligomers. REDOR spectroscopy, a solid-state NMR
technique, was applied to detect inter-peptide hydrogen bonds in membrane-associated
fusion peptide samples that have equimolar amounts of two peptides, one with selective
l3C carbonyl labeling and the other with selective 15N amide labeling. Comparison of
REDOR spectra on samples that were labeled at different residue positions suggests that
there are both parallel and anti-parallel arrangements of peptide strands in the oligomers
with about equal populations. In the parallel arrangement, inter-peptide hydrogen
bonding decreases toward the C-terminus while in the antiparallel arrangement, hydrogen
bonds are observed along the entire length of residues which was probed (Gly-5 to Gly-
16). For the parallel arrangement, these observations are consistent with the model in
which the apolar N-terminal and central regions of the peptides penetrate into the
membrane and hydrogen bond with one another while the polar C-terminus of the peptide
is outside the membrane and hydrogen bonds with water, which is also consistent with
the previous measurements of carbonyl linewidths along the peptide backbone. In the N-
terminal and central regions, the linewidths are 2 ~ 3 ppm, which indicates a narrow
distribution of secondary structures with a regular hydrogen bonding pattern whereas in
the C-terminal region, the linewidths are ~ 5 ppm, which indicates random coil structure
and irregular hydrogen bonding.

It is useful to consider our results in the context of the membrane fusion. The
samples were made by the insertion of the fusion peptide into the membrane from
aqueous solution, which is likely the means by which the viral peptide domain inserts
into the target cell membrane. The fusion activity results in Chapter II also provide

evidence for the fusogenic nature of the peptide under conditions comparable to those

136

 

 

used in the preparation of solid state NMR samples. Thus, the observed B strand
conformation likely reﬂects at least the end-state fusogenic structure of the peptide
domain.

The oligomerization study also provided information about the expected fusion
peptide topology in the gp41 protein. Structures of the gp41 soluble ectodomain begin
around residue 30 and are trimeric with the three N-termini in close proximity at the end
of an in-register helical coiled-coil. These structures end several residues C-terminal of
the fusion peptide and it has been hypothesized that during viral/target cell fusion, at least
three fusion peptides insert into the target cell membrane with their C-termini in close
proximity6’m. In the HIV -1 fusion peptide model system, the biological topology has
been mimicked with C-terrninal cross-linking, and it has been observed that in many
cases, the cross-linked peptides induce both a greater ﬁnal extent and a more rapid rate of
fusion than their non-cross-linked analogs“. Similar fusogenic enhancements have been
observed for inﬂuenza protein domains for which the fusion peptide is thought to assume
the biologically relevant oligomeric topology” 12'". In these contexts, it is appealing that
a fusion peptide population was indeed observed in the present study with an
approximately in-register parallel strand arrangement. This arrangement seems like a
reasonable continuation of the known ectodomain oligomeric structure and this
arrangement also has a defined apolar region of the fusion peptide oligomer which could
insert into the membrane and catalyze fusion. For the non-cross-linked peptides used for
the present study, a population of peptide with anti-parallel strand arrangement was also
observed. In the anti-parallel arrangement, it is not as clear how the oligomer might insert

into the membrane and catalyze fusion. Thus, the fusogenic enhancement of the HIV-1

137

 

cross-linked peptides may be in part due to their preference for a parallel strand
arrangement.
There are several remaining questions about the structures of membrane-
associated oligomeric B strand HIV-1 fusion peptides including:
(1) the relative populations of hairpin and extended structures;
(2) the effect of the biologically relevant oligomeric topology on the relative
populations of parallel and antiparallel strand arrangements;
(3) the overall oligomer size and the number of peptides with a particular strand
arrangement;
(4) the precise residue register(s) between neighboring strands in a particular
arrangement.
(5) the insertion of the oligomer into the membrane.
Question (5) has been previously investigated with ﬂuorescence and ESR measurements
but the data analyses may have been complicated by the presence of parallel and

18'20. These questions are

antiparallel oligomers which may have different insertion depths
currently investigated using solid-state NMR and other biophysical techniques.

Larger biological questions include the relative fusogenicities of the helical and
non-helical forms of the peptide and the possible role of transient structures in fusion
catalysis. At least for the inﬂuenza fusion peptide, it appears that both helical and non-
helical forms are fusogenic and that there is not a single transient fusogenic structure”.
For the helical form of the inﬂuenza fusion peptide, some of the relationships between

point mutations and fusogenicity have been correlated with structural changes, in

particular angle and depth of membrane insertionS' 22. For the non-helical form of the

138

HIV-1 fusion peptide, the mutation/fusogenicity data have generally been correlated both
with membrane insertion and with oligomer size, but more work needs to be done to
establish the structure-function relationships” 23' 24. Finally, we note that data from our
lab suggest that the relative populations of helical and non-helical peptide structures are
signiﬁcantly affected by the lipid and cholesterol composition of the target membrane”.
Further experiments, computer simulation, and modeling will be necessary to understand

the speciﬁc peptide/peptide, peptide/lipid, peptide/cholesterol interactions which underly

the molecular mechanism of HIV-1 viral fusion.

139

 

REFERENCE

10.

11.

Durell, S. R., Martin, I., Ruysschaert, J. M., Shai, Y., and Blumenthal, R. (1997)
What studies of fusion peptides tell us about viral envelope glycoprotein-mediated
membrane fusion (review). Mol. Membr. Biol. 14, 97-112.

Suarez, T., Gomara, M. J., Goni, F. M., Mingarro, I., Muga, A., Perez-Paya, E.,
and Nieva, J. L. (2003) Calcium-dependent conformational changes of
membrane-bound Ebola fusion peptide drive vesicle fusion. FEBS Lett. 535, 23-
28.

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

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

Epand, R. M., Epand, R. F., Martin, 1., and Ruysschaert, J. M. (2001) Membrane
interactions of mutated forms of the inﬂuenza fusion peptide. Biochemistry 40,
8800-8807.

Chan, D. C., Fass, D., Berger, J. M., and Kim, P. S. (1997) Core structure of gp41
from the HIV envelope glycoprotein. Cell 89, 263-273.

Tan, K., Liu, J., Wang, J., Shen, S., and Lu, M. (1997) Atomic structure of a
therrnostable subdomain of HIV-1 gp41. Proc. Natl. Acad. Sci. U. S. A. 94,
12303-12308.

Weissenhom, W., Dessen, A., Harrison, S. C., Skehel, J. J., and Wiley, D. C.
(1997) Atomic structure of the ectodomain from HIV-l gp41. Nature 387, 426-
430.

Caffrey, M., Cai, M., Kaufman, J., Stahl, S. J., Wingﬁeld, P. T., Covell, D. G.,
Gronenbom, A. M., and Clore, G. M. (1998) Three-dimensional solution structure
of the 44 kDa ectodomain of SIV gp41. EMBO J. 17, 4572-4584.

Yang, Z. N., Mueser, T. C., Kaufman, J., Stahl, S. J., Wingfield, P. T., and Hyde,
C. C. (1999) The crystal structure of the SIV gp41 ectodomain at 1.47 A
resolution. J. Struct. Biol. 126, 131-144.

Yang, R., Yang, J ., and Weliky, D. P. (2003) Synthesis, enhanced fusogenicity,

and solid state NMR measurements of cross-linked HIV-l fusion peptides.
Biochemistry 42, 3527-3535.

140

 

 

12.

13.

14.

15.

16.

17.

18.

19.

20.

21.

22.

23.

Kim, C. H., Macosko, J. C., Yu, Y. G., and Shin, Y. K. (1996) On the dynamics
and conﬁrmation of the HA2 domain of the inﬂuenza virus hemagglutinin.
Biochemistry 35, 5359-5365.

Kim, C. H., Macosko, J. C., and Shin, Y. K. (1998) The mechanism for low-pH-
induced clustering of phospholipid vesicles carrying the HA2 ectodomain of
inﬂuenza hemagglutinin. Biochemistry 3 7, 137-144.

Epand, R. F., Macosko, J. C., Russell, C. J ., Shin, Y. K., and Epand, R. M. (1999)
The ectodomain of HA2 of inﬂuenza virus promotes rapid pH dependent
membrane fusion. J. Mol. Biol. 286, 489-503.

LeDuc, D. L., Shin, Y. K., Epand, R. F., and Epand, R. M. (2000) Factors
determining vesicular lipid mixing induced by shortened constructs of inﬂuenza
hemagglutinin. Biochemistry 39, 2733-2739.

Leikina, E., LeDuc, D. L., Macosko, J. C., Epand, R., Shin, Y. K., and
Chernomordik, L. V. (2001) The 1-127 HA2 construct of inﬂuenza virus
hemagglutinin induces cell-cell hemifusion. Biochemistry 40, 8378-8386.

Epand, R. F., Yip, C. M., Chernomordik, L. V., LeDuc, D. L., Shin, Y. K., and
Epand, R. M. (2001) Self-assembly of inﬂuenza hemagglutinin: studies of
ectodomain aggregation by in situ atomic force microscopy. Biochim. Biophys.
Acta-Biomembr. 1513, 167-175.

Gordon, L. M., Curtain, C. C., Zhong, Y. C., Kirkpatrick, A., Mobley, P. W., and
Waring, A. J. (1992) The amino-terminal peptide of HIV-1 glycoprotein 41
interacts with human erythrocyte membranes: peptide conformation, orientation
and aggregation. Biochim. Biophys. Acta 1139, 257-274.

Kliger, Y., Aharoni, A., Rapaport, D., Jones, R, Blumenthal, R., and Shai, Y.
(1997) Fusion peptides derived from the HIV type 1 glycoprotein 41 associate
within phospholipid membranes and inhibit cell-cell Fusion. Structure- function
study. J. Biol. Chem. 272, 13496-13505.

Saez-Cirion, A., and Nieva, J. L. (2002) Conformational transitions of membrane-
bound HIV-1 fusion peptide. Biochim. Biophys. Acta-Biomembr. 1564, 57-65.

Parkanzky, P. D., and Weliky, D. P., unpublished experiments.

Tamm, L. K., Han, X., Li, Y. L., and Lai, A. L. (2002) Structure and function of
membrane fusion peptides. Biopolymers 66, 249-260.

Pereira, F. B., Goni, F. M., Muga, A., and Nieva, J. L. (1997) Permeabilization
and fusion of uncharged lipid vesicles induced by the HIV-1 fusion peptide

141

24.

25.

adopting an extended conformation: dose and sequence effects. Biophys. J. 73,
1977-1986.

Pritsker, M., Rucker, J ., Hoffman, T. L., Doms, R. W., and Shai, Y. (1999) Effect
of nonpolar substitutions of the conserved Phell in the fusion peptide of HIV -1
gp41 on its function, structure, and organization in membranes. Biochemistry 38,
11359-11371.

Bodner, M. L., Gabrys, C. M., Parkanzky, P. D., Yang, J., Duskin, C. A., and
Weliky, D. P. Temperature dependence and resonance assignment of 13C NMR
spectra of selectively and uniformly labeled fusion peptides associated with
membranes. Magn. Reson. Chem, in press.

142

 

IIIIIIIIIIIIIIIIIIIIIIIIIIIIIIII

llllﬂllllllllllllllllllllelllllllllllllllljlll