WHHll”IEHINIWIMHWW!HilWlHlthHl

     
   
   

137
692
TH .

n .
c‘ pr. I

5:3- ‘795 .“f

This is to certify that the
thesis entitled

TEMPERATURE DEPENDENCE, STRUCTURAL
PLASTISCITY , AND RESONANCE ASSIGNMENT OF
UNIFORMLY LABELED HIV-1 FUSION PEPTIDES
ASSOCIATED WITH MEMBRANES

presented by

Michele L. Bodner

has been accepted towards fulfillment
of the requirements for the

M. 8. degree in Chemistry

 

flmzfl p [Qua/co,

Major Professor’s Sfihature
26¢ ' /o'1 960.6

Date

MSU is an Affirmative Action/Equal Opportunity Institution

‘7 ‘ -—~ ~ —.

'V v' «.-

 

LIBRARY
Michigan State
University

 

 

 

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 c:/CIRC/DateDue.p65-p.15

 

TEMPERATURE DEPENDENCE, STRUCTURAL PLASTICITY, AND
RESONANCE ASSIGNMENT OF UNIFORMLY LABELED HIV-1 FUSION
PEPTIDES ASSOCIATED WITH MEMBRANES

By

Michele L. Bodner

A THESIS

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

MASTER OF SCIENCE

Department Of Chemistry

2003

ABSTRACT
TEMPERATURE DEPENDENCE, STRUCTURAL PLASTICITY, AND
RESONANCE ASSIGNMENT OF UNIFORMLY LABELED HIV-1 FUSION
PEPTIDES ASSOCIATED WITH MEMBRANES
By
Michele L. Bodner

The HIV-l viral fusion peptide serves as a biologically relevant model for
viral/target cell membrane fusion and in my work, the structure of the membrane-
associated peptide was probed by solid state NMR MAS l3C chemical shift
measurements. Solution NMR studies have shown that the peptide is predominantly
helical in detergent micelles and this was correlated with solid state NMR l3C chemical
shifts in frozen detergent. Large shift changes (2-4 ppm) were observed for the peptide in
a mixture whose lipid headgroup and cholesterol composition reflects the membranes of
host cells of the virus. In this more biologically relevant composition, the chemical shifts
are consistent with predominant non-helical structure. NMR spectra were compared at
—50 0C and at 20 °C. Similar peak chemical shifts were observed at both temperatures,
which indicates that cooling the sample does not significantly change the peptide
structure. Relative to -50 °C, the 20 °C signals were narrower and had lower intensity,
which is consistent with greater motion at higher temperature. l3C/‘3C correlation
experiments were performed on a sample in which the peptide was U-“C/‘SN labeled
over three or twelve sequential residues. The resulting 2D spectra were used to assign the
13C chemical shifts in the labeled residues and the shifts were consistent with beta strand

structure. l5N-‘3C correlation experiments were also done on a uniformly 13C/ 15N N-

Acetyl-leucine model compound sample.

To Mom and Dale

iii

Acknowledgements

I would like to thank my advisor, Dr. David Weliky, and the Weliky group for
their help and support during the last couple of years. I also want to thank my family and

friends for all their support.

iv

TABLE OF CONTENTS

Pages

LIST OF TABLES .................................................................................. vi

LIST OF FIGURES ................................................................................ vii

LIST OF ABBREVIATIONS ..................................................................... x
Chapter 1 Introduction ....................................................................... 1
References ......................................................................... 7

Chapter 2 Experimental: 1D REDOR Filtering and 2D ”C-”C Correlation. . 9

Materials ........................................................................... 9

Peptides ............................................................................ 9

Lipid Preparation ................................................................. 11
Cross-Linking ..................................................................... 12
Solid State NMR Sample Preparation ......................................... 12
Magic Angle Spinning ........................................................... 13
I3C CP MAS Experiments ....................................................... l3
Filtering Methods ................................................................. 15
NMR Spectroscopy-Experimental Details ..................................... 15
1D REDOR Experimental Details .............................................. 17
2D Experimental Details ......................................................... 18
References ......................................................................... 23

Chapter 3 Results and Discussion: 1D REDOR Filtering and 2D ”C-”C

Correlation ...................................................................... 25
Utility of REDOR Filtering .................................................... 25
Comparison of HFP Spectra in Detergent and Membranes ................ 27
Temperature Dependence of Spectra .......................................... 29
2D Correlation Spectra ......................................................... 29
References ........................................................................ 43
Chapter 4 l5N-”C Correlation Experiments ............................................ 45
Introduction ........................................................................ 45
N-Acetyl-leucine Synthesis and Crystallization .............................. 45
1D 15N-”C Correlation Experiments ........................................... 46
Results and Discussion .......................................................... 46
Conclusions ....................................................................... 50
References ........................................................................ 51
Chapter 5 Conclusions and Future Work .............................................. 52
Temperature Dependence of Spectra ......................................... 52
2D ”C-”C and 15N-”C Correlation ........................................... 52
References ........................................................................ 55

LIST OF TABLES

Pages
Tablel. l3C chemical shift assignment for LM3-associated HFP-U3 ....................... 34
Table2. l3C chemical shift assignment for LM3-associated HFP-U12 dimer .............. 42

vi

LIST OF FIGURES
Pages

Figure]. Model (left) and Electron Microscopy (right) of the HIV virus (a) binding to
host cell (b) fusion of viral and host cell membranes (c,d) formation of large pore and
infection of host cell. The Triangle represents the viral RNA that enters the host cell. . .. 2

Figure 2. Model of HIV Infection. FP= Fusion Peptide. Time Sequence: Left to Right.
.......................................................................................................... 3

Figure 3. Model for HIV-l/Host cell fusion. In the left-most figure, a gp120/gp40 trimer
is displayed with the balls representing gp120 and rods representing gp41. F represents
the fusion peptide and A represents the transmembrane anchorage of gp41. Fusion
proceeds temporally from left to right with (i) initial state, (ii) receptor binding and fusion
peptide membrane insertion, (iii) gp41 conformational change, and (iv) membrane fusion.

.......................................................................................................... 4
Figure 4. Chemical structure of HFP-U3 showing 13C and 15N labeling .................... 10
Figure 5. ”C CP MAS sequence. Magnetization was transferred from protons to ”C
during CP. lH continuous wave decoupling was applied during acquisition .............. 14
Figure 6. 1D ”C REDOR pulse sequence. CP transferred proton magnetization to ”C.
”C magnetization is dephased (reduced) by l3C-”N dipolar coupling mediated by two
equally spaced 15N 180° pulses per rotor period. A ”C 180° pulse in the middle of the
dephasing period refocuses the ”C chemical shift ............................................. 16

Figure 7. 2D ”C-”C correlation PDSD sequence. Magnetization was transferred from 1H
to ”C during CP. Continuous wave (CW) decoupling is applied after CP on the protons
during t1 and t2, but not during I. tl was the evolution time, 1: was the magnetization
exchange time, and t2 was the acquisition time ................................................ 20

Figure 8. 2D l3C-”C correlation RFDR sequence. CP transferred magnetization from 1H
to 13C and evolves during t,. The 1: exchange period consists of ”C 180° pulses every
other rotor period. Acquisition was done during t2 ............................................ 21

Figure 9. ”C solid-state NMR spectra of HFP-U3 peptide associated with LM3 lipid
mixture at —50°C. (top) Unfiltered REDOR spectrum where signal from both the lipids
and peptide are observed. (bottom) Filtered spectrum obtained from the difference of So-
Sl FIDs. Because of the short 1.6 ms dephasing time only the backbone carbonyl and Ca

signals from Phe-8, Leu-9, and Gly-lO are observed. * = Carbonyl spinning sidebands.
...................................................................................................... 26

vii

Figure 10. ”C solid state NMR spectra of HFP-U3 associated with DPC at -80°C (top)
and LM3 lipid mixture at ~50°C (bottom). The peptide had uniform ”C , ”N labeling at
the Phe-8, Leu-9, and Gly-lO residues. Because of the lms REDOR filter, the spectra are
dominated by backbone carbonyl signals from Phe-8 and Leu-9 and backbone Ca. signals
from Phe-8, Leu-9, and Gly-lO. For each spectrum, the spinning speed was 8 kHz and 50
Hz gaussian line broadening was applied. The large differences between chemical shifts
in (top) and (bottom) spectra are consistent with a change from helical structure in
detergent to non-helical structure in LM3 ...................................................... 28

Figure 11. ”C solid state NMR spectra of HFP-U3 associated with LM3 lipid mixture at
—50 °C (top) and 20 °C. The chemical shifts in the spectra are consistent with non-helical
structure at both —50 °C and 20 °C. However, the -50 0C spectrum has about three times
more integrated signal per scan than the 20 °C spectrum, which can be explained by
greater motion at 20 °C ............................................................................ 30

Figure 12. (a) 2D ”C-”C contour plot spectrum of HFP-U3 associated with LM3 lipid
mixture at —50 0C and (b) f2 slices from this spectrum. The peptide: lipid mol ratio was
~0.04, the buffer pH was 7.0, and the sample volume in the 4mm diameter rotor was ~30
pl. The peptide had uniform ”C, ”N labeling at the Phe-8, Leu-9, and Gly-lO residues.
The 2D data were obtained with a proton-driven spin diffusion sequence and the total
signal averaging time was ~54 hours. The spectrum displayed in (a) was processed with
200 Hz Gaussian line broadening in f1 and 150 Hz line broadening in f2. Ten contours are
shown with each increasing contour representing 1.5 times greater signal intensity. In the
upper slice of (b), cross peaks to the Phe-8 aromatic Cl (fl = 137.6 ppm) are displayed.
From left to right, they represent magnetization in f2 on the following Phe-8 nuclei: CO,
aromatic C1(diagonal), aromatic C2-C6, Ca, and Ca. In the lower slice of (b), cross peaks
to the Leu-9 CO (fl = 172.1ppm) are displayed. From left to right, they represent
magnetization in f2 on the following Leu-9 nuclei: CO (diagonal), CO (m =-1 spinning
sideband), Ca, C,3 , CV with C6 shoulder, and CO ( m = -2 spinning sideband) ............. 31

Figure 13. 2D ”C-”C contour plot spectrum of HFP-U3 associated with LM3 lipid
mixture. The 2D data set was obtained with a radio frequency- driven recoupling
sequence using a 4 ms RFDR mixing time ...................................................... 33

Figure 14. Secondary ”C chemical shifts for Phe-8, Len-9, and Gly-lO of HFP-U3
associated with LM3 lipid mixture. Vertical bars represent the secondary shifts for each
residue. For these three residues, the pattern shown above is indicative of a B strand

structure .............................................................................................. 35

Figure 15. 2D ”C-”C contour plot spectrum of HFP-U3 associated with LM3 lipid
mixture at —50 °C. The 2D data was obtained with a proton-driven spin diffusion
sequence using a lOOms exchange time. The cross peak between Phe ar. 2-6 and Leu CY
is observed because of the longer exchange time .............................................. 37

Figure 16. ”C solid state NMR spectrum of HFP—Ul2 dimer associated with LM3 at
—50°C. Due to a 1 ms REDOR filter time only the backbone carbonyl and Ca signals are

viii

observed. Other experimental conditions were: 4 mm rotor diameter; 8 kHz MAS
frequency; and 50 Hz line broadening .......................................................... 39

Figure 17. 2D ”C-”C contour plot spectrum of HFP—U12 dimer associated with LM3
lipid mixture. The 2D data was obtained using a proton-driven spin diffusion sequence
with a 10 ms exchange time ....................................................................... 40

Figure 18. 2D ”C-”C contour plot spectrum of HFP—U12 dimer associated with LM3
lipid mixture. The 2D data was obtained using a radio frequency-driven recoupling
sequence with a 4 ms RFDR mixing time ...................................................... 41

Figure 19. Chemical structure of U-N-Acetyl-leucine showing ”C and ”N labeling..... 47

Figure 20. 1D ”N-”C correlation sequence. CPI transfers magnetization from protons to
”N. CP2 transfers magnetization from ”N to ”C followed by ”C detection .............. 48

Figure 21. Crystalline N-Acetyl-Leucine model compound uniformly ”C/ ”N labeled.
The spectra were obtained at -50 0C, MAS frequency of 6.8 kHz, and 32 scans of signal
averaging. (bottom) ”C CP MAS all labeled carbons were observed. (middle) 1D 1H-
”N -”C selective CP spectrum dominant signal from the carbonyl directly bonded to ”N.
(Top) 1D lH-”N-”C selective CP spectrum dominant signal from the Ca ...................... 49

Figure 22. 2D I“IN-”C correlation sequence. CPl transferred magnetization from protons

to ”N, which then evolved during t,. CP2 transferred magnetization from ”N to ”C and
was detected during t2 .............................................................................. 54

ix

AIDS

CP

CW

DPC

FMOC

FWHM

HEPES

HIV

HPLC

LUV

MALDI

MAS

NMR

PDSD

PI

POPC

POPE

POPS

PPm

REDOR

RF

RFDR

LIST OF ABBREVIATIONS

Acquired Immune Deficiency Syndrome

Cross Polarization

Continuous Wave

Dodecylphosphocholine

9-Fluorenylmethoxycarbonyl
Full-Width-at-Half-Maximum
N-2-Hydroxyethylpiperazine-N’-2-Ethanesulfonic acid
Human Immunodeficiency Virus

High Performance Liquid Chromatography

Large Unilamellar Vesicles

Matrix Assisted Laser Desorption Ionization

Magic Angle Spinning

Nuclear Magnetic Resonance

Proton Driven Spin Diffusion

Phosphatidylinositol

l -Pa1mitoyl—2-Oleoyl-sn- glycero-3-Phosphocholine

l -Palmitoyl-Z-Oleoyl-sn-glycero-3-Phosphoethanolamine
1-Palmitoyl-Z-Oleoyl-sn-glycero-3-[Phospho-L-Serine]
parts per million

Rotational Echo Double Resonance

Radio Frequency

Radio Frequency Driven Recoupling

SSNMR Solid State Nuclear Magnetic Resonance
TFA Trifluoroacetic Acid

TPPM Two Pulse Phase Modulation

xi

Chapter 1

Introduction

Fusion between cells and cellular components plays an important role in such
significant physiological processes as egg fertilization and synaptic transmission in the
nervous system. Membrane fusion is also an important step in viral infection for such
widespread and serious diseases as measles, influenza and AIDS.” Understanding viral
fusion is important as a key step in the viral life cycle and as a possible target for anti-
viral therapeutics.

Many viruses are “enveloped”, i.e. they are enclosed by a membrane. To initiate
infection of a new cell, the membranes of the virus and cell must fuse so that the viral
nucleic acid can enter the host cell." 2'4' 5 Figure 1‘5 illustrates the three sequential steps of
fusion: binding of the .two membranes, mixing of lipid membranes, and formation of a
large pore through which the contents of both the virus and the host cell mix.7 There is a
high activation barrier to membrane fusion and in the absence of a catalyst, the viral
fusion rate is usually negligible. Fusion is also very slow between unilamellar liposomes,
which often serve as a model membrane system for viruses or cells. To increase the
fusion rate, many viruses such as HIV-l employ a “fusion peptide” which represents a
~20-residue apolar domain at the N-terminus of the viral envelope fusion protein.3' 8

Current models of HIV- 1/host cell infection include interaction of the fusion
peptide with the host cell membrane as displayed in figures 29 and 3”. Fusion and

infection are initiated by strong interactions of two viral enveloped proteins (gp41 and

 

 

 

 

 

 

 

 

 

 

 

 

Figure 1. Model (left) and Electron Microscopy (right) of the HIV virus (a) binding to
host cell (b) fusion of viral and host cell membranes (c,d) formation of large pore and
infection of host cell. The triangle represents the viral RNA that enters the host cell.

 

 

Figure 2. Model of HIV Infection. FP= Fusion Peptide. Time Sequence: Left to Right

(i) (ii) (iii) (iV)

iiiiiiiiiiiii iiii as W W
Itiuitiiiiii mi tuiiiiiiiiiggt,’

   
     

 

-—"——-“-:.‘ 0:.

WWII??? "‘
WWW” .‘zi.

 

 

 

 

 

 

 

 

iiiiiiiiiiitiiv: 1...... my;

 

Figure 3. Model for HIV-l/Host cell fusion. In the left-most figure, a gpl20/gp40 trimer
is displayed with the balls representing gp120 and rods representing gp41. F represents
the fusion peptide and A represents the transmembrane anchorage of gp41. Fusion
proceeds temporally from left to right with (i) initial state, (ii) receptor binding and fusion
peptide membrane insertion, (iii) gp41 conformational change, and (iv) membrane fusion.

gp120) with the CD4 and chemokine (e. g. CXCR4) receptors of human T and
macrophage cells.” ‘2 gp41 traverses the HIV-1 membrane and the fusion peptide region
is located at the N-terminus of the gp41 extraviral ectodomain.

Because of the hydrophobic nature of this region and the presence of the lipid
membrane, the fusion peptide region has not been studied in membranes with atomic—
level structural techniques, such as crystallography or solution nuclear magnetic
resonance. Solid state NMR, however, can be applied to obtain atomic-resolution
structural information about the fusion peptide in a fully hydrated lipid membrane
environment, and the structural information can be related to function.

Solid state NMR of peptides and proteins has traditionally relied on selective ”C
and ”N labeling and by this approach the determination of a complete structure requires
at least one selectively labeled sample per residue. Preparation of such a large number of
samples is time consuming and complete structures of only two membrane-bound
peptides have been reported with this methodology.” '4 Recently, there has been an
effort to develop an alternate solid state NMR approach in which a complete structure is
derived from a single unifomtly ”C/ ”N labeled sample. This methodology requires
multidimensional NMR techniques for uniquely assigning the chemical shifts of each ”C
and ”N nucleus and multidimensional NMR techniques to determine internuclear
distance and angle constraints.”20 A whole protein structure would then be developed
through molecular dynamic simulations, which incorporate the constraints. The overall
approach is analogous to the well-developed NMR methods for proteins and nucleic acids
in aqueous solution and has the potential to provide high—resolution structures for small

proteins in intact biological membranes. The solid state NMR (SSNMR) approach will

be complementary to solution NMR and crystallographic methods for membrane
proteins, which are typically done in detergent environments. This thesis describes
application of this type of SSNMR approach to the membrane-associated HIV-1 fusion

peptide.

10.

11.

12.

References

Blumenthal, R. and BS. Dimitrov, Membrane Fusion, in Handbook of
Physiology, Section 14: Cell Physiology, J.F. Hoffman and J .C. J amieson, Editors.
1997, Oxford: New York. p. 563-603.

Dimitrov, D.S., Cell biology of virus entry. Cell, 2000. 101(7): p. 697-702.

Eckert, D.M. and PS. Kim, Mechanisms of viral membrane fitsion and its
inhibition. Annual Review of Biochemistry, 2001. 70: p. 777-810.

Blumenthal, R., et al., Membrane fusion. Chemical Reviews, 2003. 103(1): p. 53-
69.

Epand, R.M., Membrane fusion - Overview. Bioscience Reports, 2000. 20(6): p.
435-441.

Grewe, C., Beck, A., Gelderblom, H.R., HIV: early virus-cell interactions.
Journal of Acquired Immune Dificiency Syndromes, 1990. 3: p. 965.

Hernandez, L.D., eta1., Virus-cell and cell-cell fusion. Annual Review of Cell
Developmental Biology, 1996. 12: p. 627-61.

Durell, S.R., et al., What studies of fusion peptides tell us about viral envelope
glycoprotein-mediated membrane fusion (review). Molecular Membrane Biology,
1997. 14(3): p. 97-112.

Breakthrough of the Year: New Hope in HIV Disease. Science, New Series. 274:
p. 1988-1991.

Weissenhom, W., Dessen, A., Harrison, S.C., Skehel, J.J., Wiley, D.C., Atomic
Structure of the Ectodomain from HIV-1 gp41. Nature. 387: p. 426.

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

Oberlin, E., Amara, A., Bachelerie, F., Bessia, C., Virelizier, J. L.,
ArenzanaSeisdedos, F., Schwartz, 0., Heard, J.M., ClarkLewis, I., Legler,
D.F.,Loetscher, M., Baggiolini,M., Moser, B., The CXC chemokine SDF -1 is the
ligand for LES TR/.F usin and prevents infection by T-cell-line adapted HIV-1.
Nature, 1996. 382: p. 833-835.

13.

14.

15.

16.

17.

18.

19.

20.

Ketchem, R.R., W. Hu, and TA. Cross, High-resolution conformation of
gramicidin A in a lipid bilayer by solid-state NMR. Science, 1993. 261(5127): p.
1457-60.

Opella, S.J., et al., Structures of the M2 channel-lining segments from nicotinic
acetylcholine and NMDA receptors by NMR spectroscopy. Nature Structural
BiolOgy, 1999. 6(4): p. 374-9.

Rienstra, C.M., et al., 2D and 3D N-15-C-I3-C-I3 NMR chemical shift
correlation spectroscopy of solids: Assignment of MAS spectra of peptides.
Journal of the American Chemical Society, 2000. 122(44): p. 10979-10990.

McDermott, A., et al., Partial NMR assignments for uniformly (C-13, N-I5)-
enriched BPTI in the solid state. Journal of Biomolecular NMR, 2000. 16(3): p.
209-219.

Pauli, J ., et al., Backbone and side-chain C -I 3 and N-I5 signal assignments of the
alpha-spectrin SH3 domain by magic angle spinning solid-state NMR at 17.6
tesla. Chembiochem, 2001. 2(4): p. 272-281.

Egorova-Zachemyuk, T.A., et al., Heteronuclear ZD-correlations in a uniformly
[ C -I 3, N-15] labeled membrane-protein complex at ultra-high magnetic fields.
Journal of Biomolecular NMR, 2001. 19(3): p. 243-253.

Castellani, F., et al., Structure of a protein determined by solid-state magic-angle-
spinning NMR spectroscopy. Nature, 2002. 420(6911): p. 98-102.

Detken, A., et al., Methods for sequential resonance assignment in solid,
uniformly C -I 3, N-I5 labelled peptides: Quantification and application to
antamanide. Journal of Biomolecular N MR, 2001. 20(3): p. 203-221.

Chapter 2

Experimental: 1D REDOR Filtering and 2D ”C-”C Correlation

Materials

Rink amide resin was purchased from Advanced Chemtech (Louisville , KY), and
9-fluorenylmethoxycarbonyl (FMOC)-amino acids were obtained from Peptides
International (Louisville, KY). Isotopically labeled amino acids were purchased from
Cambridge (Andover, MA) and were FMOC-protected using literature procedures.”2 1-
palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine (POPC), 1-palmitoyl-2-oleoyl-sn-
glycero-3-[phospho-L-serine] (POPS), l-palmitoyl-2-oleoyl-sn-glycero-3-
phosphoethanolamine (POPE), phosphatidylinositol (PI), sphingomyelin, and
dodecylphosphocholine (DPC) were purchased from Avanti Polar Lipids, Inc. (Alabaster,
AL). N-2-hydroxyethylpiperazine-N’-2-ethanesulfonic acid (HEPES) was obtained from
Sigma (St. Louis, MO). All other reagents were analytical grade.
Peptides

HFP-U3 fusion peptide (sequence AVGIGALFLGFLGAAGSTMGARSKKK)

(figure 4) was synthesized with the 23 N-terminal residues of the LAVla strain of the

HIV-1 gp41 envelope protein followed by three additional lysines for improved
solubility. HFP-U12 fusion peptide (sequence

AVGIGALFLGFLGAAGSTMGARSCKKKKKKW) was synthesized with the 23 N-
terminal residues of the LAVla strain of the HIV-1 gp41 envelope protein followed by a

cysteine to allow for cross-linking of the peptide. The whole gp41 protein is believed to

0 H
l I
AVGIGALEN 341—33 3N 3’; _‘30' EN .36; BLFLGAAGSTMGARSKKK

Figure 4. Chemical structure of HFP-U3 showing ”C and ”N labeling.

10

be trimeric with the C-termini of three fusion peptides in close proximity (Figure 3), so
the topology gained through peptide cross-linking is likely similar to the topology which
exists in the fusogenic form of gp41.3 As in HFP-U3, the six additional lysines are to
enhance solubility. A single tryptophan was added as a 280 nm chromophore for peptide
quantitation. Both peptides were synthesized as their C-terminal amides using a peptide
synthesizer (ABI 431A, Foster City, CA) equipped with FMOC chemistry. HFP-U3 had
uniform ”C and ”N labeling at Phe-8, Len-9, and Gly-10 and HFP-U12 had uniform ”C
and ”N labeling over the twelve residues between Gly-5 and Gly-l6. Peptides were
cleaved from the resin in a three hour reaction using a mixture of trifluoroacetic acid
(TFA):H20:phenol:thioanisole2ethanedithiol in a 33:2:2:2:l volume ratio. Peptides were
subsequently purified by reversed-phased HPLC using a preparative C18 column (Vydac,
Hesperia, CA) and a water/acetonitrile gradient containing 0.1% TFA. Matrix assisted
laser desorption ionization (MALDI) mass spectroscopy was used to determine the
peptide products.
Lipid Preparation

Samples were prepared using a lipid/cholesterol mixture reflecting the
approximate lipid and cholesterol content of the HIV-1 virus infected host cells.4 This
lipid mixture consists of POPC, POPE, POPS, PI, sphingomyelin, and cholesterol in a
10:5:2: 1 :2: 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 buffer
containing 0.01% N aN 3 followed by homogenization with ten freeze-thaw cycles. Large

Unilamellar Vesicles (LUV) of 100 nm diameter were prepared by extruding the lipid

ll

dispersions ~30 times through two stacked 0.1 pm polycarbonate filters5 (Avestin, Inc.,
Ottawa, ON, Canada).
Cross-Linking

The additional cysteine in HFP-U12 allows covalent cross-linking of two peptides
through a disulfide bond. Peptide was dissolved into a 0.5 M dimethylaminopyridine
buffer at pH 8.40. This solution was vortexed for 24 hours open to the air. Purification
of the peptide dimer was carried out as described above for the monomeric peptides.
Solid State NMR Sample Preparation

Samples were prepared using 0.01% (w/v) NaN3 in 5 mM HEPES buffer (pH
7.0). A solution was made containing 0.4 — 2 pmol peptide in 30 ml volume and a
solution was made containing 100 nm diameter LUV in 5 ml volume. The LUV were
made with 40 umol lipid and 20 pmol cholesterol. The peptide and LUV solutions were
then mixed and kept at room temperature overnight. The solution was then centrifuged at
100,000 — 130,000 * g for four hours to pellet down the LUV and associated bound
peptide. Nearly all peptide bound to LUV under these conditions and unbound peptide
does not pellet. The peptide/LUV pellet formed after ultracentrifugation was transferred
by spatula to a 4 mm magic angle spinning (MAS) NMR rotor.

Peptide/detergent samples were prepared with 1.7 — 2 mM peptide in 200 mM
DPC detergent with a total volume of 250 pl. There was no ultracentrifugation step and
the liquid was transferred by pipet to the 6 mm MAS NMR rotor.

For both the membrane and detergent samples, the MAS rotors were fitted with

specially machined vespel caps. At liquid nitrogen temperatures, a cap inserted tightly

12

into the rotor. After the cap warmed up, it formed a very tight seal, which minimized

dehydration of the samples.

Magic Angle Spinning (MAS)

In solids, there is negligible molecular tumbling and the NMR signals are
broadened by anisotropic effects. These effects can be reduced and the lines narrowed by
magic angle spinning (MAS), whereby the sample is rotated at kHz frequencies about an
axis tilted at 54.7° relative to the external magnetic field direction. The MAS spectrum
contains signals at the isotropic chemical shifts, i.e. the shifts which would be observed in
a typical liquid state spectrum. If the spinning frequency is slow enough, additional
signals are observed which are separated from the isotropic peaks by integral multiples of

the spinning frequency and are known as spinning sidebands.

”C CP MAS Experiments

Transfer or “Cross polarization” (CP) of magnetization from protons to ”C is a
SSNMR technique which increases ”C signals in ID as well as multidimensional NMR
experiments. CP is accomplished by simultaneous RF radiation of both the 1H and IEC
nuclei such that both isotopes precess at the same frequency. Magnetization transfer is
mediated by heteronuclear dipolar coupling. Figure 5 shows the typical pulse sequence

for a 1D ”C CP experiment including acquisition.

l3

 

Nlél

 

1 CW DECOUPLING
H

CP

 

 

 

 

 

Figure 5. ”C CP MAS sequence. Magnetization was transferred from protons to ”C
during CP. lH continuous wave decoupling was applied during acquisition.

l4

Filtering Methods

In peptide/membrane samples containing specifically ”C labeled peptide, lH-”C
cross-polarized (CP) ”C NMR spectra typically have very large lipid, cholesterol, and
peptide natural abundance signals. Two approaches were taken to filter out the natural
abundance signals. For 1D spectra, a rotational-echo double- resonance (REDOR) NMR
filtering sequence was applied and the resulting spectra were dominated by labeled
backbone ”C with directly bonded ”N .6'7 For example, for the HFP-U3 sample, a clean
spectrum was observed of the Phe-8, Leu-9, and G1y-10 Ca and the Phe-8 and Leu-9 CO
carbons. In the second filtering approach, 2D ”C-”C correlation spectra were obtained on
both the HFP-U3 and HFP-U12 sample and off-diagonal cross peaks were only detected
between labeled ”C which are separated by one or a few bonds; i.e. in a strong ”C-”C
dipole coupling network.

The REDOR technique relies on the difference of two spectra, one with the
REDOR sequence (Figure 6), and one with a REDOR sequence without ”N It pulses.
With ”N pulses (Sl spectrum), there is heteronuclear ”C-”N dipolar coupling and the
signals of ”C directly bonded to ”N are reduced. Without ”N pulses (S0 spectrum), there
is no ”C-”N dipolar coupling and the signals of all ”C nuclei are equally detected. Thus,
the SO — Sl difference spectrum yields only the signals of ”C directly bonded to ”N.
NMR Spectroscopy - Experimental Details

The NMR spectra were taken on 9.4 T spectrometers (Varian Infinity Plus, Palo
Alto, CA) using triple resonance MAS probes equipped for either 4 mm or 6 mm
diameter rotors. The temperature was monitored by a thermocouple located about 1”

from the rotor and in the flow of the cooling nitrogen gas. The actual temperature of the

15

 

TPPM DECOUPLING
lH CP

I'OtOI'

 

 

Figure 6. 1D ”C REDOR pulse sequence. CP transferred proton magnetization to ”C.
”C magnetization is dephased (reduced) by ”C-”N dipolar coupling mediated by two
equally spaced ”N 1800 pulses per rotor period. A ”C 1800 pulse in the middle of the
dephasing period refocuses the ”C chemical shift.

16

sample is likely warmer than the measured temperature because of frictional heating from
sample spinning and from heating due to the radiofrequency (RF) fields. In the NMR
probe circuit, the RF fields are highly attenuated at the ends of the coil and it was
experimentally observed that nearly the entire N MR signal comes from the central 2/3 of
the sample volume specified by the manufacturer. Hence, in the 6 mm rotor, longer
spacers were used to restrict samples to this central 2/3 volume(~160 pl). For the 4 mm
rotor, the total possible rotor sample volume was ~ 70 pl.

The detection channel was tuned to ”C at 100.2 or 100.8 MHz, the decoupling
channel was tuned to 1H at 398.6 or 400.8 MHz, and the third channel for REDOR-
filtered 1D spectra was tuned to ”N at 40.4 or 40.6 MHz. ”C chemical shift referencing
was done using the methylene resonance of solid adamantane at 38.5 ppm.8 The ”N
transmitter was set to ~ 115 ppm using solid (”NH,)ZSO,, as a chemical shift reference at
20 ppm. The ”C It pulse rf field, 1H rt/2 pulse field, lH and ”C CP fields, and ”N 311 pulse
field using leucine which contained 5% U-”C, ”N molecules diluted in natural
abundance material. The MAS frequency was 8000 2 2 Hz for the 1D REDOR
experiments and 6800 Hz for the 2D ”C/ ”C correlation experiments.
1D REDOR Experimental Details

For the REDOR experiments, generation of ”C transverse magnetization was
followed by a REDOR dephasing period and then direct ”C detection. For the HFP-U3
sample, the ”C transmitter was set to 155 ppm at —50 °C and to 100 ppm at 20 °C. ”C
transverse magnetization was generated using lH-”C CP with a 53—57 kHz ”C ramp and
1.8 ms contact time at —50 °C and 3 ms contact time at 20 °C. The dephasing period was

set to 8 rotor periods (1 ms) and contained a single 55 kHz ”C refocusing rt pulse at the

17

center of this period. For the S, acquisition, 45 kHz ”N It pulses were applied at the
middle and end of every rotor cycle during the dephasing period except for the fourth and
eighth cycles. The S0 acquisition did not contain these ”N It pulses. In the dephasing
period, pulse timing was not actively synchronized to the rotor position. Two-pulse phase
modulation (TPPM) 'H decoupling at 100 kHz was applied during both dephasing and
detection with 5.4 ps pulse length and 90° and 105° phases.9 To obtain optimal
compensation of B0, 8,, and MAS frequency drifts, S0 and S, free induction decays
(FIDs) were acquired alternately. The recycle delay was 2 s at —50 °C and 1.3 s at 20 °C.

A Z-filter sequence was used to set the ”C It pulse length and contained the
following sequential elements: lH-”C CP; ”C Jt/2; 10 ms; ”C It; detection. lH decoupling
was applied during pulses and detection. The ”N It pulse length was set by minimization
of S, signals for the model compound and the TPPM pulse length was set by
maximization of the 80 signal for the model compound.
For each S, transient, XY-8 phase cycling was applied to the ”N n: pulses.”ll
Individual S0 or S, transients were coadded with the following phase cycling scheme: lH
rt/2, x, -x, x, -x; ”C CP and ”C it, -y, -y, x, x; receiver, x, -x, y, -y. After completion of
data acquisition, the sum of S, FIDs was subtracted from the sum of S0 FIDs. Spectral
processing was done on the difference FID with a DC offset correction, 25 Hz Gaussian
line broadening, Fourier transform, and baseline correction.
2D Experimental Details

The 2D ” C-13 C correlation spectra were obtained at —50 °C with the probe
configured for double resonance ” C/ 1H Operation. The ”C sensitivity in double

resonance mode was ~ 1.5 times greater than the sensitivity in triple resonance mode. For

18

one data set, correlations were generated by the proton-driven spin diffusion (PDSD)
pulse sequence: CP — t1 —- 1t/2 — 1: — 1t/2 — t2 (figure 7) where t, was the evolution period,
the first tt/2 pulse rotated ”C transverse magnetization to the longitudinal axis, 1: was a 10
ms spin diffusion period during which ”C longitudinal magnetization was transferred
between ”C nuclei connected by a network of direct ” C-” C bonds, the second 1t/2 pulse
rotated ”C longitudinal magnetization to the transverse plane, and t; was the detection
period. Continuous wave (CW) lH decoupling at 100 kHz was applied during the pulse,
t1, and t2 periods, but not during I. In a second data set, longitudinal transfer of ”C
magnetization during ‘r was achieved with use of the radiofrequency-driven dipolar
recoupling (RFDR) method (figure 8).”’” In this approach, a ”C 1: pulse was applied at
the end of rotor cycles 1, 3, 5, ..., 31 during 1:. CW 1H decoupling at 100 kHz was also
applied during 1. The 1: period contained a total of 32 rotor cycles.

The following parameters were common to the PDSD and RF DR data sets: 44-64
kHz ramp on the ”C CP rf field; 2 ms CP contact time; 50 kHz ”C 1t/2 pulse rf field; 25
ps t1 dwell time; 20 ps t2 dwell time; and l s recycle delay. Hypercomplex data were
obtained by acquiring two individual FIDs for each t1 point with either a ”C (rt/2)x or
(tr/2)y pulse at the end of the t1 evolution period. For the first of these t1 FIDs, individual
transients were coadded with the following phase cycling scheme: first ”C 1t/2 pulse, x, -
x, x, -x, x, -x, x, -x; second ”C n/2 pulse, x, x, y, y, -x, -x, -y, -y; receiver, y, -y, -x, x, -y,
y, x, -x. For the other t1 FID, the first ”C tr/2 pulse followed y, -y, y, -y, y, -y, y, -y
cycling. The PDSD data were acquired in ~ 54 hours with 200 t, points, 1024 t2 points,
and 1024 transients per F ID, and the RFDR data were acquired in ~ 60 hours with 200 t,

points, 1024 t2 points, and 512 transients per FID. Both data sets were processed

19

 

 

2

1H CP CW DEC CW DEC

 

.71:

_ E
2 2
t1 7 t2

 

Figure 7. 2D ”C-”C correlation PDSD sequence. Magnetization was transferred from 1H
to ”C during CP. Continuous wave (CW) decoupling is applied after CP on the protons
during t, and t2, but not during 1:. t, was the evolution time, 1: was the magnetization

exchange time, and t, was the acquisition time.

20

 

N|=I

CW DECOUPLING

 

 

Figure 8. 2D ”C-”C correlation RFDR sequence. CP transferred magnetization from 1H
to ”C and evolves during t,. The 1 exchange period consists of ”C 1800 pulses every
other rotor period. Acquisition was done during t,.

21

according to the method of States using nmrPipe software.”’” Processing included zero-

filling, Gaussian line broadening, and baseline correction.

22

10.

References

Chang, C.D., et al., Preparation and Properties of N-Alpha-9-

F luorenylmethyloxycarbonylamino Acids Bearing Tert-Butyl Side- Chain
Protection. International Journal of Peptide and Protein Research, 1980. 15(1): p.
59-66.

Lapatsanis, L., et al., Synthesis of N-2,2,2-( Trichloroethoxycarbonyl )-L-Amino
Acids and N-(9-Fluorenylmethoxycarbonyl)-L-Amino Acids Involving
Succinimidoxy Anion As a Leaving Group in Amino-Acid Protection. Synthesis-
Stuttgart, 1983. 8: p. 671-673.

Yang, R., J. Yang, and DP. Weliky, Synthesis, enhanced fitsogenicity, and solid
state NMR measurements of cross-linked HIV-I fusion peptides. Biochemistry,
2003. 42(12): p. 3527-3535.

Aloia, R.C., H. Tian, and EC. Jensen, Lipid composition and fluidity of the human
immunodeficiency virus envelope and host cell plasma membranes. Proceedings
of the National Academy of Sciences of the United States of America, 1993.
90(11): p. 5181-5.

Hope, M.J., et al., Production of Large Unilamellar Vesicles By a Rapid
Extrusion Procedure - Characterization of Size Distribution, Trapped Volume and
Ability to Maintain a Membrane-Potential. Biochimica Et Biophysica Acta, 1985.
812(1): p. 55-65.

Yang, J ., et al., Application of REDOR Subtraction for Filtered MAS Observation
of Labeled Backbone Carbons of Membrane-Bound Fusion Peptides. Journal of
Magnetic Resonance, 2002. 159(2): p. 101-110.

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

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

Bennett, A.E., et al., Heteronuclear Decoupling in Rotating Solids. Journal of
Chemical Physics, 1995. 103(16): p. 6951-6958.

Gullion, T., D.B. Baker, and MS. Conradi, New, Compensated Carr-Purcell
Sequences. Journal of Magnetic Resonance, 1990. 89(3): p. 479-484.

23

 

ll.

l2.

13.

14.

15.

Gullion, T. and J. Schaefer, Elimination of Resonance Ofiset Efi‘ects in

Rotational-Echo, Double-Resonance NMR. Journal of Magnetic Resonance, 1991.
92(2): p. 439-442.

Bennett, A.E., et al., Chemical-Shift Correlation Spectroscopy in Rotating Solids -
Radio F requency—Driven Dipolar Recoupling and Longitudinal Exchange.
Journal of Chemical Physics, 1992. 96(11): p. 8624-8627.

Bennett, A.E., et al., Homonuclear radio frequency-driven recoupling in rotating
solids. Journal of Chemical Physics, 1998. 108(22): p. 9463-9479.

States, D.J., R.A. Haberkom, and DJ. Ruben, A Two-Dimensional Nuclear
Overhauser Experiment With Pure Absorption Phase in 4 Quadrants. Journal of
Magnetic Resonance, 1982. 48(2): p. 286-292.

Delaglio, F., et al., NMRPipe: a multidimensional spectral processing system
based on UNIX pipes [see comments]. Journal of Biomolecular NMR, 1995. 6(3):
p. 277-93.

24

Chapter 3

Results and Discussion - 1D REDOR Filtering and 2D ”C-”C Correlation

In MAS solid state NMR spectra of membrane-bound peptides and proteins,
signals from specifically labeled nuclei can provide important information about the local
structure and structural homogeneity."4 However, for ”C, these labeled nuclei signals are
usually poorly resolved from large natural abundance signals of lipid and protein. To
filter out these natural abundance signals, REDOR difference spectroscopy was
investigated. REDOR difference spectroscopy was shown to be an easy method to
implement because it only requires only one sample and uses a fairly simple pulse
sequence.

Utility of REDOR Filtering

For the NMR sample HFP-U3 associated with the LM3 lipid/cholesterol mixture
at peptidezlipid mol ratio ~0.04, Figure 9 displays the REDOR S0 and difference spectra.
In the S0 spectrum, there is significant natural abundance signal from the lipid and
peptide. The difference spectrum was taken with a 1.6 ms dephasing time and cleanly
shows the signals of backbone carbons directly bonded to labeled ”N. The isotropic Ca,
carbonyl, and M = +1, -1 , and —2 spinning sidebands of the carbonyls are apparent. In
the filtered spectrum, the isotropic Phe-8 and Leu-9 carbonyl signals at 172 ppm are
unresolved whereas the 40-55 ppm Phe-8, Leu-9, and Gly-10 C0l signals were all
resolved. On the basis of the characteristic chemical shifts of residue-types, we can

tentatively assign the 42.5 ppm signal to Gly-10 Ca”

25

 

250 150 50 ppm

 

250... .. .150.... ....50....pf)m

Figure 9. ”C solid-state NMR spectra of HFP-U3 peptide associated with LM3 lipid
mixture at —50°C. (top) Unfiltered REDOR spectrum where signal from both the lipids
and peptide are observed. (bottom) Filtered spectrum obtained from the difference of S,-
S, FIDs. Because of the short 1.6 ms dephasing time only the backbone carbonyl and Ca
signals from Phe-8, Leu-9, and Gly-10 are observed. * = Carbonyl spinning sidebands

26

In this case, the difference spectrum is also useful for assessing linewidths and
indirectly, the feasibility of doing a full assignment of this peptide with multidimensional
NMR methods. The linewidths are a few times larger than those observed in larger U-”C,
”N crystalline peptides and proteins, but the recent successes in assignment of these
crystalline systems suggest that a fairly long sequence in the membrane-associated fusion
peptide can be assigned, followed by a full structure detemrination."6"°
Comparison of HFP Spectra in Detergent and Membranes

In Figure 10, HFP-U3 was associated with frozen detergent at —80 °C (top) and
frozen LM3 membranes at —50 °C (bottom). In both the carbonyl and Ca regions of the
spectra, there are significant differences in chemical shifts which indicate a change from
helical structure in detergent to non-helical structure in LM3. In the Ca region, signals
are expected from the Phe-8, Leu-9, Gly-10 nuclei, and there are three clear peaks in both
the DPC and the LM3 samples. Although site-specific assignment is not definitive in this
one-dimensional spectrum, a tentative assignment can be made on the basis of
characteristic chemical shifts of particular amino acids. In particular, the Ca shifts for
Phe, Leu, and Gly in random coil structures are 57.4 ppm, 53.6 ppm, and 43.5 ppm,
respectively. When this ordering is used to assign Ca peaks in the spectra, the measured
shifts for the DPC and LM3 samples lie downfield and upfield of the random coil values,
which is consistent with a change for this part of HFP from helical structure in detergent

to non-helical structure in LM3. Thus, it appears that there can be significantly different

structures in detergent and membranes.

27

 

| I F
150 100 50 ppm

Figure 10. ”C solid state NMR spectra of HFP-U3 associated with DPC at -80°C (top)
and LM3 lipid mixture at —50°C (bottom). The peptide had uniform ”C , ”N labeling at
the Phe-8, Leu-9, and Gly-10 residues. Because of the lms REDOR filter, the spectra are
dominated by backbone carbonyl signals from Phe-8 and Leu-9 and backbone Ca signals
from Phe-8, Leu-9, and Gly-10. For each spectrum, the spinning speed was 8 kHz and 50
Hz gaussian line broadening was applied. The large differences between chemical shifts
in (top) and (bottom) spectra are consistent with a change from helical structure in
detergent to non-helical structure in LM3.

28

Temperature Dependence of Spectra

Although physiological fusion occurs at 37 °C, NMR sensitivity is much better at
lower temperature and spectra are often obtained at these temperatures. It is therefore
important to determine whether lower temperature changes peptide structure. Figure 11
displays the Ca region of the REDOR filtered spectrum of the HIV-1 fusion peptide
sample at (top) —50 °C and (bottom) 20 °C. Because of the 1 ms REDOR filter, the
displayed spectral regions are dominated by backbone Ca signals from Phe-8, Leu-9, and
Gly-10. Corresponding peak chemical shifts agree to within 0.5 ppm at the two
temperatures, indicating that the lower temperature does not induce a large peptide
structural change. The carbonyl region was similarly invariant to temperature. The
linewidths were smaller at 20 °C than at —50 °C. For example, for the Gly-10 peak
centered at 43.0 ppm, the full-width at half-maximum (FW HM) linewidth is ~ 2.6 ppm at
—50 °C and is ~ 1.9 ppm at 20 °C. In addition, the integrated signal-to-noise per ”C per
transient at 20 °C is approximately 1/3 of its value at —50 °C. For hydrated membrane
samples, it is reasonable that motion could increase significantly between —50 °C and 20
°C and this greater motion could explain these experimental observations. For example,
increased motion could reduce inhomogeneous broadening and result in smaller
linewidths. In addition, motion could attenuate dipolar couplings and decrease the
efficiency of lH-”C cross-polarization (CP) and REDOR dephasing and result in lower
REDOR-filtered signal per ”C per transient.
2D Correlation Spectra

Figure 12(a) displays a 2D ”C/ ”C correlation spectrum for the HFP-U3/LM3

sample and figure 12(b) displays f2 slices from this spectrum at f, = 137.9 ppm (upper)

29

 

Figure 11. ”C solid state NMR spectra of HFP-U3 associated with LM3 lipid mixture at
—50 °C (top) and 20 °C. The chemical shifts in the spectra are consistent with non-helical
structure at both —50 °C and 20 °C. However, the —50 °C spectrum has about three times
more integrated signal per scan than the 20 °C spectrum, which can be explained by
greater motion at 20 °C.

30

 

(a)

‘O
O

 

's'e'a eases

‘
8
51:1:

 

 

 

 

A N

150 100 50 ppm
Figure 12. (a) 2D ”C-”C contour plot spectrum of HFP-U3 associated with LM3 lipid
mixture at —50 °C and (b) f2 slices from this spectrum. The peptide: lipid mol ratio was
~0.04, the buffer pH was 7.0, and the sample volume in the 4mm diameter rotor was ~30
p1. The peptide had uniform ”C, ”N labeling at the Phe-8, Leu-9, and Gly-10 residues.
The 2D data were obtained with a proton-driven spin diffusion sequence and the total
signal averaging time was ~54 hours. The spectrum displayed in (a) was processed with
200 Hz Gaussian line broadening in f, and 150 Hz line broadening in f2. Ten contours are
shown with each increasing contour representing 1.5 times greater signal intensity. In the
upper slice of (b), cross peaks to the Phe-8 aromatic C1 (f, = 137.6 ppm) are displayed.
From left to right, they represent magnetization in f2 on the following Phe-8 nuclei: CO,
aromatic C1(diagonal), aromatic C2-C6, Ca, and CB. In the lower slice of (b), cross peaks
to the Len-9 CO (f, = 172.1ppm) are displayed. From left to right, they represent
magnetization in f2 on the following Leu-9 nuclei: CO (diagonal), CO (m =-1 spinning
sideband), Ca, C,3 , CY with C, shoulder, and CO ( m = -2 spinning sideband).

31

and 172.1 ppm (lower). The displayed spectra were generated from —50 0C data and the
correlations were a result of magnetization exchange driven by 10 ms proton-driven spin
diffusion (PDSD). A 2D correlation spectrum with similar appearance (figure 13) was
obtained from data for which the magnetization exchange was generated by a 4 ms
radiofrequency-driven dipolar recoupling (RFDR) sequence.‘ "'2 Measurement of the
cross peak chemical shifts and knowledge of ”C connectivities and characteristic residue-
type chemical shifts made possible a full resonance assignment of all of the labeled ”C in
the peptide. For example, the upper slice in fig. 12(b) indicates the f2 shifts of Phe-8 CO,
C,,, and CO by means of their correlations with the unique f, shift of Phe-8 aromatic C1.
The full assignment is presented in Table 1. Each chemical shift entry in Table 1 is the
average of 4 and 14 f, and f2 shift measurements from PDSD and RFDR spectra. There
was a typical standard deviation of ~ 0.2 ppm in the shift distribution of a single entry.
Figure 14 presents a graphical secondary shift analysis for C0, C0,, and C,3 nuclei
in Phe-8, Leu-9, and Gly-10 where the secondary shift is defined as the difference
between the measured and random coil shifts. For this analysis, the literature random coil
Ca and C5 shifts” and CO shifts” were reduced by 2.1 and 2.0 ppm, respectively, which
accounts for the differences between the solid state NMR referencing to neat TMS and
the solution NMR referencing to ~5 mM 3-(trimethylsilyl)-propionate (TSP) and D88,
respectively.”'” For the HFP-U3 sample, the negative CO and Ca shifts and positive
CB shifts correlate with [3 strand secondary structure over these three residues.”'”'”
In the PDSD and RFDR spectra, (1) only intra-residue cross peaks were
definitively observed and (2) relatively strong cross peaks were observed between ”C

separated by several bonds (e. g. Leu-9 CO/ CY ). These observations are consistent with

32

 

 

 

 

 

i .10
G L,c0toLc, , 1m
K L,c0toLc, . 3 i:
. -ao
Q at?” ;.,
310
cf " :°°
o... . :1”
. , r110
'. ’ :18
«a
'0 . .' . a .m :1“
. :1”
5. ,. :100
am / m 9 /.‘"’°
.10; . v . v 1 - . I - 1 -1”
130 150 150 11'» 100 so no 40
f2ppm

51:“:

Figure 13. 2D ”C-”C contour plot spectrum of HFP-U3 associated with LM3 lipid
mixture. The 2D data set was obtained with a radio frequency- driven recoupling
sequence using a 4 ms RFDR mixing time.

33

 

Table 1. ”C chemical shift assignments for LM-associated HFP-U3

 

 

 

 

C, CB C, C, C arl C ar 2-6 co

Phe-8 53.6 41.6 137.3 129.2 171.0
Leu-9 51.2 44.4 25.3 22.5 171.9
Gly-10 42.7 169.1

 

 

 

 

 

 

 

 

34

 

Phe-8 Leu-9 Gly-10

 

 

 

E3ppm
C. I | '
C, I I
CO I l I

Figure 14. Secondary ”C chemical shifts for Phe-8, Leu-9, and Gly-10 of HFP-U3
associated with LM3 lipid mixture. Vertical bars represent the secondary shifts for each
residue. For these three residues, the pattern shown above is indicative of a B strand

SIIUCIUI'C.

35

the results of other groups using these experiments with short mixing times on U- ”C, ”N
labeled peptides and proteins.7"°'”'” Both PDSD and RFDR are relatively broad-banded
exchange sequences mediated by ”C- ”C dipolar coupling. For two ”C separated by
distance r, the exchange rate with have an approximate r *5 dependence, and direct
exchange between ”C separated by two bonds or three bonds would respectively occur at
~5% or 2% of the rate of exchange between ”C separated by one bond. Thus, it is more
likely that a two-bond or three-bond cross peak is due to multiple steps of exchange
between directly bonded ”C rather than a single step exchange process. With the
assumption that there is a single rate constant for directly bonded ”C exchange, at short
times the ratio of intensities of the two bond/one bond cross peaks will be about the same
as the ratio of three bond/two bond cross peaks and the ratio of four bond/three bond
cross peaks. This model is qualitatively supported by the relative intensities of the Leu-9
CO/Ca, CO/CB, CO/C,f , and CO/C,,, cross peaks (10:6:2:1) and all of the (n+1)/n-bond
cross peak intensity ratios are within a range of 0.3 — 0.6. Inter-residue cross peaks are
likely not observed in these short mixing time spectra because the ”N interrupts the direct
”C bond network. Inter-residue cross peaks are apparent in a 2D PDSD spectrum with a
longer 100 ms mixing time (figure 15), in accord with the experience of other
investigators. '0

The success of the complete assignment of the HFP-U3 uniformly labeled peptide
associated with LM3 prompted the addition of more uniformly labeled amino acid
residues to the sequence. HFP-U12 peptide was synthesized and cross-linked with

twelve sequential uniformly ”C and ”N labeled residues between Gly-5 and Gly-16.

36

 

 

 

 

's's's'e'e's's's”

 

 

é, F,ar 216 toLCY
8 ° "
/ o .0 o i o.
p o
9 fl@@ °
0 O O
y 69°® .
110‘ 1a 130 1a 160 so a to 7 a a"
f2ppm

‘
C

Figure 15. 2D ”C-”C contour plot spectrum of HFP-U3 associated with LM3 lipid
mixture at -50 °C. The 2D data was obtained with a proton-driven spin diffusion
sequence using a 100ms exchange time. The cross peak between Phe ar. 2-6 and Len CY

is observed because of the longer exchange time.

37

56-:

Figure 16 displays the REDOR difference spectrum of HFP-U12 associated with LM3 at
peptide: lipid mol ratio of ~0.04. The REDOR difference spectrum was taken with a 1
ms dephasing time and shows the signals of backbone carbons directly bonded to labeled
”N. The isotropic carbonyl signals from the labeled residues at 172.1 ppm and 168.9
ppm are unresolved. This is also true of the Ca region of the spectrum where there are
three peaks at 42.7 ppm, 49.1 ppm, and 50.8 ppm. Because of the linewidths, it appears
from the REDOR difference spectrum that the full assignment of this peptide will be
quite challenging even with the use of multidimensional NMR methods.

Figure 17 displays a 2D ”C-”C correlation spectrum of HFP-U12 dimer
associated with LM3. The displayed spectra were obtained from —50 °C data and the
observed cross peaks were a result of magnetization exchange driven by 10 ms PDSD
exchange time. A 2D correlation spectrum, mediated by RFDR exchange had similar
appearance to (figure18).

In the PDSD and RFDR spectra, intra-residue cross peaks were primarily
observed. Because of amino acid degeneracy in HFP-U12, cross peaks could only be
assigned as residue-type rather than residue-specific. A residue-type chemical shift
assignment of the four different residues is shown in table 2. Higher field may be
beneficial for these uniformly labeled systems. In uniformly labeled samples a 40-100
Hz contribution to the ”C MAS linewidth is due to unresolved ”C-”C J -couplings,
which in ppm units are inversely proportional to the magnetic field.20 At higher field,

linewidths will therefore be narrower and spectral resolution should be improved.

38

 

 

 

 

 

 

 

250 20'0 150 160 56 ppm

Figure 16. ”C solid state NMR spectrum of HFP—U12 dimer associated with LM3 at
—50°C. Due to a 1 ms REDOR filter time only the backbone carbonyl and Ca signals are
observed. Other experimental conditions were: 4 mm rotor diameter; 8 kHz MAS
frequency; and 50 Hz line broadening.

39

 

 

883888883
'5

\
§
5':

P 0/ . 3?. Q30 . l
o.. . -110

 

 

 

.130'100130'120 100 doiio'lo in o
f2ppm

Figure 17. 2D ”C—”C contour plot spectrum of HFP—U12 dimer associated with LM3
lipid mixture. The 2D data was obtained using a proton-driven spin diffusion sequence
with a 10 ms exchange time.

40

 

‘
O

   

'2': at s as as“

150 - 100 ' 1410 150 - 1i» ' do do 4'0 all
f 2 ppm

 

 

 

 

Figure 18. 2D l3C-”C contour plot spectrum of HFP—U12 dimer associated with LM3
lipid mixture. The 2D data was obtained using a radio frequency-driven recoupling
sequence with a 4 ms RFDR mixing time.

41

511-:

Table 2. ”C chemical shift assignment for HFP-U12 dimer associated with LM3.

 

 

 

 

 

 

 

 

 

 

 

 

 

Ca C,3 C, C, C ar 1 C ar 2-6 CO
Ala 49.1 21.8 172.2
Gly 43.0 168.6
Leu 51.0 44.5 21.8 25.2 171.7
Phe 53.8 42.0 136.7 128.60 171.2

 

42

 

10.

References

Smith, 80., CS. Smith, and 3.]. Bormann, Strong hydrogen bonding
interactions involving a buried glutamic acid in the transmembrane sequence of
the neu/erbB-Z receptor. Nature Structural Biology, 1996. 3(3): p. 252-8.

Wang, J ., Y.S. Balazs, and L.K. Thompson, Solid-state REDOR NMR distance
measurements at the ligand site of a bacterial chemotaxis membrane receptor.
Biochemistry, 1997. 36(7): p. 1699-703.

Yang, J ., C.M. Gabrys, and DP. Weliky, Solid-State Nuclear Magnetic
Resonance Evidence for an Extended beta Strand Conformation of the
Membrane-Bound HIV-1 Fusion Peptide. Biochemistry, 2001. 40(27): p. 8126-
37.

Yang, J ., et al., Solid state NMR measurements of conformation and
conformational distributions in the membrane-bound HIV-I fusion peptide.
Journal of Molecular Graphics & Modelling, 2001. 19(1): p. 129-35.

Evans, J .N.S., Biomolecular NMR Spectroscopy. 1995, New York: Oxford.

Rienstra, C.M., et al., 20 and 3D N-15-C-I3-C-I3 NMR chemical shift
correlation spectroscopy of solids: Assignment of MAS spectra of peptides.
Journal of the American Chemical Society, 2000. 122(44): p. 10979-10990.

McDermott, A., et al., Partial NMR assignments for uniformly ( C -I 3, N-I5)-
enriched BPTI in the solid state. Journal of Biomolecular NMR, 2000. 16(3): p.
209-219.

Detken, A., et al., Methods for sequential resonance assignment in solid,
uniformly C-13, N-15 labelled peptides: Quantification and application to
antamanide. Journal of Biomolecular NMR, 2001. 20(3): p. 203-221

J aroniec, C.P., et al., Frequency selective heteronuclear dipolar recoupling in
rotating solids: Accurate C -I 3-N-1 5 distance measurements in uniformly C -I 3,N-
15-labeled peptides. Journal of the American Chemical Society, 2001. 123(15): p.
3507-3519.

Straus, S.K., T. Bremi, and RR. Ernst, Side-chain conformation and dynamics in
a solid peptide: C P-MAS NMR study of valine rotamers and methyl-group
relaxation in fully C-13-labelled antamanide. Journal of Biomolecular NMR,
1997. 10(2): p. 119-128.

43

11.

12.

l3.

14.

15.

16.

17.

18.

19.

20.

Bennett, A.E., et al., Chemical-Shift Correlation Spectroscopy in Rotating Solids -
Radio F requency-Driven Dipolar Recoupling and Longitudinal Exchange.
Journal of Chemical Physics, 1992. 96(11): p. 8624-8627.

Bennett, A.E., et al., Homonuclear radio frequency-driven recoupling in rotating
solids. Journal of Chemical Physics, 1998. 108(22): p. 9463-9479.

Spera, S. and A. Bax, Empirical Correlation Between Protein Backbone
Conformation and C -Alpha and C -Beta C-13 Nuclear-Magnetic-Resonance

C hemical- Shifts. Journal of the American Chemical Society, 1991. 113(14): p.
5490-5492.

Wishart, D.S., et al., H—I, C-13 and N-15 Random Coil Nmr Chemical-Shifts of
the Common Amino-Acids .1. Investigations of Nearest-Neighbor Effects. Journal
of Biomolecular NMR, 1995. 5(1): p. 67-81.

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

Wishart, D.S., et al., H—I, C-13 and N-15 Chemical-Shift Referencing in
Biomolecular Nmr. Journal of Biomolecular NMR, 1995. 6(2): p. 135-140.

Saito, H., Conformation-Dependent C -I 3 Chemical-Shifts - a New Means of
Conformational Characterization As Obtained By High-Resolution Solid-State C -
13 Nmr. Magnetic Resonance in Chemistry, 1986. 24(10): p. 835-852.

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

Pauli, J ., et al., Backbone and side-chain C -I 3 and N-I5 signal assignments of the
alpha-spectrin SH3 domain by magic angle spinning solid-state NMR at I 7.6
tesla. Chembiochem, 2001. 2(4): p. 272-281.

Marassi, F .M.a.O., S.J, Simultaneous resonance assignment and structure
determination in the solid-state NMR spectrum of a membrane protein in lipid
bilayers. Biophysical Journal, 2002. 82: p. 2279.

Chapter 4

”N-”C Correlation Experiments

Introduction

In recent years, the problem of structure determination in uniformly labeled ”C
and ”N biological samples has attracted a considerable amount of interest. Partial and
complete assignment of uniformly labeled samples has been done using 2D ”N/ ”C and
l3C/”C experiments”. For sequential assignment, it is particularly interesting to
selectively transfer polarization from backbone ”N to either directly bonded ”C0, or to
directly bonded ”CO and to therefore correlate nuclei within a residue i or between
adjacent i and i —- 1 residues. Selective lD ”N-”C transfer experiments can be used to
determine the feasibility of implementing these 2D ”N-”C correlation experiments. It is
also useful to first optimize the experiments on a high signal-to—noise model compound
with completely resolved resonances. Our compound choice for these experiments was
crystalline N-Acetyl-leucine.
N-Acetyl-leucine Synthesis and Crystallization

Uniformly 13C/ ”N labeled leucine was added to acetic acid and heated to 100 °C
while stirring. 1-”C-acetic anhydride was added, and was stirred until the solid
dissolved. The solution was cooled to 80 °C and water was added to react with any
excess acetic anhydride. Under reduced pressure the solvent was removed and a viscous
liquid remained. Traces of water were removed by the multiple steps of addition of
cyclohexane followed by solvent removal under reduced pressure. Solvent removal was

considered to be complete when the dry weight no longer changed.6 The white solid was

45

dried further overnight in a vacuum dessicator. Residual acetic acid was then removed
by dissolution in water and lyophilization. Solution NMR was used to determine the
success of the synthesis.

For crystallization, 16.5 mg of U-N-Acetyl-leucine (figure 19) and 89.4 mg of
unlabeled N -Acetyl-leucine were dissolved into ~ 5ml of water under gentle heating.
After cooling, the solution was filtered using 0.22 pm sterile syringe filter (Millipore,
Bedford) into a clean vial. Crystals were formed upon slow evaporation of the water.
1D ”N-”C Correlation Experiments

The 1D ”N -”C correlation spectra were obtained at -—50 °C and 6.8 kHz MAS
with the pulse sequence: CPl- r - CP2 — acquisition where CP1 was the lH-”N cross
polarization, 1: was 1 ps, and CP2 was ”N-”C cross polarization (figure 20). The ”N
transmitter was set to 115 ppm and ”N transverse polarization was generated by lH-”N
CP with a ”N ramp of 30-49 KHz and a 4.0 ms contact time. 100 KHz CW lH
decoupling was used during 1:, CP2, and acquisition. Polarization was transferred from
”N to ”C during CP2 using a 25 KHz field on ”N with no ramp and 1.5 ms contact time.
When the ”C transmitter was set to 155 ppm, the CP2 transfer was selective for CO and
when the ”C transmitter was set to 60 ppm, the CP2 transfer was selective for C0,. The
average ”C CP2 field was 25 KHz for the selective CO transfer and 24.8 KHz for the
selective Ca transfer.

Results and Discussion

Figure 21 (bottom) displays a ”C CP MAS spectrum of the model compound.

The spectrum was taken at -50 °C using 6.8 KHz MAS and processed using 10 Hz

Gaussian line broadening. Two carbonyl peaks are resolved at 177.2 ppm and 175.4 ppm

46

Figure 19. Chemical structure of U-N-Acetyl-leucine showing ”C and ”N labeling.

47

 

ii.
2

 

 

 

1,, I l CP1 I CW DECOUPLING
17
13C . CP2

 

Figure 20. 1D ”N-”C correlation sequence. CP1 transfers magnetization from protons to
”N. CP2 transfers magnetization from ”N to ”C followed by ”C detection.

48

 

 

 

 

150 100 50 ppm

Figure 21. Crystalline N-Acetyl-Leucine model compound unifomily ”C/ ”N labeled.
The spectra were obtained at —50 °C, MAS frequency of 6.8 kHz ,and 32 scans signal
averaging. (bottom) ”C CP MAS all labeled carbons were observed. (middle) 1D 1H-
”N -”C selective CP spectrum dominant signal from the carbonyl directly bonded to ”N.
(Top) 1D lH-”N-”C selective CP spectrum dominant signal from the Ca.

49

where the upfield shift is due to the carbonyl carbon directly bonded to the ”N and the
downfield shift is due to the carbonyl bonded to the oxygen. Five peaks are observed in
the aliphatic region of the spectrum and had chemical shifts of 53.8, 40.5, 25.5, 23.9, and
20.0 ppm. The signals arose from the Ca, CB,C, ,C,,, and the methyl carbon respectively.
Integrated intensities from the CO directly bonded to the ”N and the alpha carbon were
used to determine the efficiency of the selective 1D ”N-”C experiments.

Figure 21 (middle) displays a selective ”N-”C correlation spectrum. A single
peak was observed at 175.4 ppm, which corresponds to the chemical shift of the CO
carbon directly bonded to the ”N. For selective CP from ”N to the alpha carbon a single
peak at 53.8 ppm was the primary observed (figure 21 top). The chemical shifts of both
selective experiments correspond to chemical shifts in the ”C CP MAS experiment. The
integrated intensities of CO and Ca peaks were measured for both the ”N-”C selective
CP and ”C CP MAS experiments to determine the efficiencies of the selective CP
experiments. Both efficiencies were quite similar with 30% for Ca and 33% for CO.
The observed efficiencies suggest that it is reasonable to apply these experiments to
peptide/lipid samples.

Conclusions

U-N-Acetyl-leucine has been very useful as a model compound to determine the
feasibility of the ”N-”C correlation experiments on biologically relevant samples. An
efficiency of ~30% for both CO and Ca selective experiments was observed and suggests
that with sufficient signal averaging time, ”N-”C experiments could be applied to

membrane-associated fusion peptide samples.

50

References

Rienstra, C.M., et al., 20 and 3D N-15-C-13-C-13 NMR chemical shift
correlation spectroscopy of solids: Assignment of MAS spectra of peptides.
Journal of the American Chemical Society, 2000. 122(44): p. 10979-10990.

McDermott, A., et al., Partial NMR assignments for uniformly (C-13, N-I5)-
enriched BPTI in the solid state. Journal of Biomolecular NMR, 2000. 16(3): p.
209-219.

Pauli, J ., et al., Backbone and side-chain C -I 3 and N-15 signal assignments of the
alpha-spectrin SH3 domain by magic angle spinning solid-state NMR at 17.6
tesla. Chembiochem, 2001. 2(4): p. 272-281.

Castellani, F., et al., Structure of a protein determined by solid-state magic-angle-
spinning NMR spectroscopy. Nature, 2002. 420(691 l): p. 98- 102.

Detken, A., et al., Methods for sequential resonance assignment in solid,
uniformly C -I 3, N-I5 labelled peptides: Quantification and application to
antamanide. Journal of Biomolecular N MR, 2001 . 20(3): p. 203-221.

Williamson, K.L., Macroscale and Microscale Organic Experiments. second ed.
1994, Lexington, Massachusetts, Toronto: Heath and Company. 632-635.

51

Chapter 5

Conclusions and Future Work

Temperature Dependence of Spectra

Viral/target cell membrane fusion occurs at 37 °C, but for membrane-associated
HIV-1 FP samples, the signal-to-noise per peptide ”C per transient is about three times
higher at —50 °C. In order to take advantage of the higher signal at low temperatures, it
is important to demonstrate the biological relevance of the cold samples and to therefore
investigate whether cooling changes the peptide structure. The chemical shifts are
similar at 20 °C and at —50 °C, which suggests that cooling does not change the average
peptide structure. The temperature dependences of intensities and linewidths do suggest
that motion at 20 °C is attenuated at —50 °C, which is a physically reasonable result for
hydrated membrane samples.
2D ”C-”C and ”N-”C Correlation

2D ”C/ ”C correlation spectra of the HFP-U3/LM3 sample were used to give
insight into the secondary structure over the three uniformly labeled residues in the
peptide chain. Both PDSD and RFDR sequences were used to determine the chemical
shift of each labeled carbon nucleus. The difference between the measured and random
coil shifts were used to determine the secondary structure. For the HFP-U3/LM3 sample
negative CO and Ca shifts and positive CCl shifts were calculated. This pattern is

diagnostic of a B strand secondary structure over the three labeled residues.

52

We are working on 2D ”N -”C correlation experiments, which will aid in the
complete structure determination of the uniformly labeled HIV-1 samples. The
correlations are generated by: CP 1— t, — CP2 — t2 (figure 22) where CP1 is cross
polarization from lH-”N, t, is the evolution period, CP2 is cross-polarization from ”N -
”C, and t2 is the acquisition time."5 Cross-polarization between ”N-”C can be applied
selectively, where magnetization is transferred from the ”N nuclei to the carbonyl carbon
or the alpha carbon. By selectively controlling the transfer of magnetization we can
‘walk’ down the backbone of the peptide.

2D ”C-”C correlation spectra of HFP-U12 dimer associated with LM3 showed
that only residue type chemical shifts could be obtained at 400 MHz. However, the
spectra should be better resolved at higher field (for example 600 or 900 MHz) because
of narrower resonances arising from attenuation of ”C-”C J- and dipolar couplings. In
addition, sensitivity should be improved at higher field and will be beneficial for

sequential assignment based on 3D experiments such as ”N -”C-”C correlation.

53

 

 

NH:

 

 

1,, CP, CW DECOUPLING
”N CP1 CP2
tl
l3C CP2

 

Figure 22. 2D ”N -”C correlation sequence. CP1 transferred magnetization from protons
to ”N, which then evolved during t,. CP2 transferred magnetization from ”N to ”C and
was detected during t,.

54

 

References

Rienstra, C.M., et al., 2D and 3D N—15-C-13-C-I3 NMR chemical shift
correlation spectroscopy of solids: Assignment of MAS spectra of peptides.
Journal of the American Chemical Society, 2000. 122(44): p. 10979—10990.

McDermott, A., et al., Partial NMR assignments for uniformly ( C -I 3, N-I5)-
enriched BPTI in the solid state. Journal of Biomolecular Nmr, 2000. 16(3): p.
209-219.

Pauli, J ., et al., Backbone and side-chain C -I 3 and N-15 signal assignments of the I
alpha-spectrin SH3 domain by magic angle spinning solid-state NMR at 17.6 .
tesla. Chembiochem, 2001. 2(4): p. 272-281. ,
Egorova-Zachernyuk, T.A., et al., Heteronuclear 2D-correlations in a uniformly

[C-13, N-I5] labeled membrane-protein complex at ultra-high magnetic fields.

Journal of Biomolecular Nmr, 2001. 19(3): p. 243-253.

 

Detken, A., et al., Methods for sequential resonance assignment in solid,
uniformly C -I 3, N-15 labelled peptides: Quantification and application to
antamanide. Journal of Biomolecular Nmr, 2001. 20(3): p. 203-221.

55

   

lililliiiiiiiiiilTrill