Tài liệu Báo cáo khoa học: Structure and membrane interaction of the internal fusion peptide of avian sarcoma leukosis virus pdf

Structure and membrane interaction of the internal fusion peptide
of avian sarcoma leukosis virus
Shu-Fang Cheng, Cheng-Wei Wu, Eric Assen B Kantchev and Ding-Kwo Chang
Institute of Chemistry, Academia Sinica, Taipei, Taiwan, Republic of China
The structure and membrane interaction of the internal
fusion peptide (IFP) fragment of the avian sarcoma and
leucosis virus (ASLV) envelope glycoprotein was studied by
an array of biophysical methods. The peptide w as found to
induce lipid mixing of ves icles more stron gly than the f usion
peptide derived from the N-terminal fusion peptide of
influenza virus (HA2-FP). It was observed that the helical
structure was enhanced in association with the model
membranes, p articularly in the N -terminal portion of the
peptide. According to the infrared study, the peptide inserted
into the membrane in a n oblique orientation, but less deeply
than the influenza HA2-FP. Analysis of NMR d ata in
sodium dodecyl sulfate micelle suspension revealed that
Pro13 of the peptide was located near the micelle–water
interface. A type II b-turn was deduced from NMR data for
the peptide in aqueous medium, demonstrating a c onforma-
tional flexibility of t he IFP in a nalogy to the N -terminal FP
such as that of gp41. A loose and multim odal self-assembly
was deduced from the rhodamine fluorescence self-quench-
ing experiments for the peptide bound to the membrane
bilayer. Oligomerization o f the peptide and its variants can
also be observed in the electrophoretic experiments, sug-
gesting a property in common with other N-terminal FP of
class I fusion pr oteins.
Keywords: membrane fusion; conformational change;
insertion d epth; self-assembly; fluorescence self-quenching.
Entry of e nveloped viruses into the host cells is mediated

by the viral envelope glycoproteins [1], which in most
cases are cleave d by proteolysis to yield the transmem-
brane (TM) [ 2,3] subunit responsible for membrane fusion
and the surface (SU) subunit f or receptor binding. For a
majority of the class I fusion proteins, a region in t he TM
protein crucial for binding to and destabilizing target
membranes, termed fusion peptide (FP), is located a t the
N-terminal region, while others have the internal fusion
peptide (IFP) domain [ 4]. Avain sarcoma/leucosis virus
(ASLV) is a prototype retrovirus [5], the envelope
glycoprotein of which uses IFP for fusion to target cells
[6,7]. A proline is often found near the centre of many of
the viral IFP sequences [1]. Delos et al. [8] have shown
that the central proline of the FP of ASLV subtype A
plays important roles in forming a native e nvelope protein
(EnvA) structure a nd in membrane fusion. It is thought
that the envelope protein undergo es conformational
change triggered by i ts binding to the receptor on the
target cell surface (e.g. Tva f or ASLV-A), exposing the
hydrophobic FP domain to destabilize t he cell membrane
preceding the membrane fusion [9] similar to influenza
haemagglutinin and HIV-1 gp41. As the majority of
studies were performed on the N-terminal FP, it would
be of interest to compare the structure of the internal FP
and i ts interaction w ith membrane bilayer, including in
particular the structural influence of proline. Consistent
with other c lass I viral fus ion proteins, the IFP o f ASLV
inserts into the membrane primarily as a h elix in contrast
to the I FP of class II fusion protein which uses a Ôcd loopÕ
to insert into the target membrane in the fusion process

[10,11]. In the following, a variety of physical properties of
the putative IFP of ASLV are reported and differences
between N-terminal and internal FP a re compared. The
pH dependence of s ome o f the properties is discussed in
regard to the experimental observation that ASLV
induced hemifusion, but not complete fusion, at neutral
pH [12].
Experimental procedures
All chemicals and solvents were used without further
purification. N-a-(9-Fluorenylmethoxycarbonyl) ( Fmoc)-
protected amino acids were products of Anaspec (San
Jose, CA, USA) or Bachem (Bubendorf, Switzerland).
1,2-Dimyristoyl-sn-glycero-3-phosphocholine (DMPC) and
Correspondence to D K. Chang, Institute of Chemistry, Academia
Sinica, Taipei, Taiwan 115, Republic of China.
Fax: + 886 2 27831237, Tel.: + 886 2 2 7 898594,
E-mail:
Abbreviations: ASLV, avain/sarcoma leucosis virus; ASLV-A, ASLV
subtype A; ATR-FTIR, attenuated total reflectance-FTIR; DG, dis-
tance geometry; DMPC, 1,2-dimyristoyl-sn-glycero-3-phosphocho-
line; DMPG, 1,2-dimyristoyl-sn-glycero-3-phosphoglycerol; EnvA,
native envelope protein; FP, fusion peptide; HA2-FP, N-terminal
fusion peptide of influenza virus; IFP, internal fusion peptide; NBD,
7-nitrobenz-2-oxa-1,3-diazole; NBD-PE, N-(7-nitrobenz-2-oxa-1,3-
diazol-4-yl)-1,2-dihexadecanyol-sn -glycero-3-phosphoethanolamine;
rhodamine, 5(6)-carboxytetramethylrhodamine; Rh-PE, Lissa-
mine
TM
rhodamine B 1,2-dihexadecanoyl-sn-glycero-3-phospho-
ethanolamine, triethylammonium salt; SA, simulated annealing;

SU, surface; TM, transmembrane.
(Received 2 6 May 2004, revised 6 October 2004,
accepted 13 October 2004)
Eur. J. Biochem. 271, 4725–4736 (2004) Ó FEBS 2004 doi:10.1111/j.1432-1033.2004.04436.x
1,2-dimy ristoyl-sn-glycero-3-phosphoglycerol (DMPG)
were obtained from A vanti Polar L ipids (Alabaster,
AL, USA). 7-Nitrobenz-2-oxa-1,3-diazole (NBD) and
proteinase K were purchased from Sigma (St. Louis, MO,
USA). 5(6)-Carboxytetramethylrhodamine (TAMRA),
N-(7-nitrobenz-2-oxa-1,3-diazol-4-yl)-1,2-dihexadecanyol-
sn-glycero-3-phosphoethanolamine (NBD-PE) and
Lissamine
TM
rhodamine B 1,2-dihexadecanoyl-sn-glycero-
3-phosphoethanolamine, triethylammonium salt (Rho-PE)
were purchased from Molecular Probes, Inc. (Eugene,
OR, USA). S DS and d
25
-SDS were acquired from Boeh-
ringer Mannheim (Mannheim, Germany) and Cambridge
Isotope (Andover, MA, USA), respectively. Solutions con-
taining vesicles were p repared by solubilizing the lipids i n
chloroform/methanol (4 : 1, v/v) mixture and drying the
sample under n itrogen stream b efore dissolving i n buffer
solution. Peptide/SDS mixtures a nd peptide/phospholipid
mixtures were sonicated for 30–60 min before measure-
ments.
The internal fusion peptide Ac-GPTARIFASILAPG
VAAAQALREIERLA-NH
2

(IFP–wt), residues 16–43 of
the native envelope protein from the avian sarcoma and
leukosis virus s ubtype A , a nd its P 13F ( IFP–F13), P 13V
(IFP–V13) variants were assembled by Fmoc/t-Bu solid
phase peptide synthesis as C-terminal carboxyamides on
Fmoc-Rink amide Resin using a peptide syn thesizer (Pro-
tein Technologies, Tucson, AZ, USA, model Rainin PS3)
operated in the manual mode. The N-terminal free peptides
were labelled with TAMRA following the standard Fmoc-
amino a cid coupling protocol with a coupling t ime of 10 h.
Labelling with NBD was achieved according t he procedure
of Rapaport and Shai [ 13] with modifications as described in
a recent s tudy [14]. Cleavage and purification of the peptide
were as described previously [14,15].
The N-terminal peptide corresponding to residues 1–25
of HA2 (strain X31) of influenza virus, HA2[1–25], w as also
synthesized [15] for comparison of structure and f unction.
Circular dichroism experiments
CD measurements were carried out on a Jasco 720
spectropolarimeter. E ach o f the peptides tested was incu-
batedwithNaCl/P
i
or DMPC/DMPG (1 : 1) vesicular
suspension, at pH 5.0 or 7 .4 to give a final concentration of
30 l
M
of peptide in NaCl/P
i
or of peptide/DMPC/DMPG
(12 l

M
:0.8m
M
:0.8m
M
). All samples were measured
with a 1.0-mm path length cell, at 37 °C. The spectra were
recorded from 260 to 190 nm at a scanning rate of
50 nm Æmin
)1
with a time constant of 2 s, step resolution
of 0.1 nm, and bandwidth of 1 nm. The final spectra were
taken from the average of five s cans. The V
ARSELEC
program was used for the secondary structure prediction
as described previously [16,17].
Fluorescence spectrometry
All fluorescence experiments were performed on a Hitachi
F-2500 Fluorescence Spectrometer at 37 °Cusinga1cm
2
semimicro quartz cuvette with stirrer. The response time
was set at 0.08 s, slit bandwidth for excitation and emission
was 10 nm. A scan rate of 300 nmÆmin
)1
was used for the
wavelength scans.
Membrane binding and depth of immersion
of the peptide probed by N-terminally labelled NBD
The NBD-labelled peptide was used to m onitor the
interaction between the peptide and lipid vesicles of

DMPC/DMPG (1 : 1, molar ratio) with a suspension
containing 0.06 l
M
and 300 l
M
of the NBD-labelled
peptide and phospholipid, respectively, at pH 5.0 or 7.4
[18]. T he excitation and emission wavelengths w ere set at
467 and 530 nm, respectively, for time scan measure-
ments. Spectra in the 500–650 nm range were collected in
the wavelength scan experiments. The digestive enzyme,
proteinase K (60 lgÆmL
)1
in final concentration), was
added to vesicles loaded with NBD-conjugated peptide
to investigate the extent of protection from the
enzyme action by the membrane on each of the peptides
tested.
For Co
2+
quenching experiments, the NBD-labelled
peptide was added to a cuvette containing DMPC/DMPG
vesicles at pH 5 or pH 7.4 and measurement was taken
until the fluorescence signal attained a steady value. The final
concentrations of peptide/DMPC/DMPG w ere 0.06 : 150 :
150 l
M
. A n i ncremental amount of CoCl
2
stock s olution

(0.1
M
) was then injected into the cuvette to give final
concentrations in the range of 0.02–2.0 m
M
. Corrections due
to dilution were made to the observed fluorescence intensi-
ties. T he data were analyzed using the Stern–Volmer
equation:
F
0
=F ¼ 1 þ K
sv
[Q]
where F
0
and F are the intensity o f NBD fluorescence
before and after addin g a given amount of CoCl
2
solution, respectively, [Q] is the concentration of the
quencher and the slope K
SV
is the Stern–Volmer
constant.
Lipid mixing assay by fluorescence resonance energy
transfer
Membrane fusion assay used in t he study is based on
the measurement of FRET from NBD to rhodamine
[19]. Specifically, two lipid suspensions were prepared,
one unlabelled ( DMPC/DMPG 2 50 : 2 50 l

M
)and
one labelled (DMPC/DMPG/NBD-PE–Rho-PE 250 :
250:5:5l
M
), using NaCl/P
i
buffer at n eutral or acidic
pH. A 9 : 1 molar ratio o f unlab elled to l abelled
liposomes (total volume 1 mL) was used in the assay;
hence the final DMPC/DMPG to NBD-PE molar ratio is
1000 after the lipid mixing. Various aliquots of 1 m
M
peptide stock solution dissolved in dimethylsulfoxide
(DMSO) were injected into the liposome mixture. As a
control, 20 lL DMSO was used. As the fusion peptide is
added to induce lipid mixing, the fluorescent probe is
diluted by m ixing of t he unlabelled a nd labelled v esicles,
resulting in r educed energy transfer efficiency and an
increase in the fluorescen ce intensity of the ener gy donor,
NBD-PE. T o monitor the NBD probe, the excitation and
the e mission wavelengths were set at 467 nm and 530 nm,
respectively. The fluorescence inte nsity after the a ddition
of Triton X-100 (0.2% v/v) was referred to as 100%,
respectively.
4726 S F. Cheng et al. (Eur. J. Biochem. 271) Ó FEBS 2004
Self-association tendency of IFP–wt by N-terminally
labelled rhodamine fluorophore
The rhodamine self-quenching experiments were carried out
to examine the propensity of self-association o f the peptides

in NaCl/P
i
and in t he vesicular su spension. Briefly, the
rhodamine-labelled peptide (0.1 l
M
) in a queous buffer at
pH 5.0 or 7.4 was mixed with DMPC/DMPG (1 : 1, molar
ratio) vesicles (lipid concentration 200 l
M
). To monitor t he
rhodamine probe, the excitation and emission wavelengths
were set at 530 and 578 nm, r espectively. Digestion of the
labelled peptides by proteinase K (50 lgÆmL
)1
)leadsto
disassembly of the membrane-associated oligomeric fusion
peptides, resulting in dequenching of rhodamine fluores-
cence. The 100% reference intensity was taken from the
fluorescence measured in the peptide/lipid dispersion solu-
bilized with 0.2% (v/v) Triton X-100.
In the experiments on the composition variation of
rhodamine-labelled peptide, the total (labelled plus unla-
belled) peptide c oncentration w as kept cons tant at 0.01 l
M
while the fraction of labelled peptide, x , w as varied from
0.05 to 1 i n DMPC/DMPG (150 : 150 l
M
)vesicular
suspension. The normalized emission intensity I
x

/x was
plotted against x [20].
It is noted that intra-trimeric interaction is detected
for x values near 1 since nearly all peptide molecules are
labelled and quenching therefore arises predominantly from
the close neighbours within the same trimer. In contrast, for
low x values, t he probability of find ing a pair of labelled
peptides is slim and h ence quenching arises mainly fr om
labelled peptides in nearby t rimers.
SDS/PAGE experiments to examine the oligomerization
of IFP–wt, –F13 and –V13 in the membranous setting
A PhastSystem
TM
(Pharmacia Biotech, Sweden) accom-
panied with PhastGel
Ò
high density a nd PhastGel
Ò
SDS
buffer strips was used for SDS/PAGE experiments, which is
particularly suitable for molecules in the molecular mass
range 1000–20000. Peptide samples were added to 6% SDS,
10% glycerol and 10 m
M
Tris-buffered solutions at pH 6.8
andheatedat55°C for 10 min. IFP analogues and markers
(0.5 m
M
; Pharmacia Biotech MW marker kit, code no.
80-1129-83) were loaded in a P hastGel

Ò
sample applicator
8/1 (code no. 18-1816-01). The running condition and
staining method followed the procedures given in Phast-
System handbook (ref. no. 80-1312-29 and 80-1312-30,
respectively). Compositions of the buffer s ystem in the gel,
buffer strips and solutions used for development can be
found in the homepage .
Attenuated total reflectance-FTIR measurements
Polarized ATR-FTIR spectra were recorded on a Boman
DA8.3 spectrometer with a KBr beamsplitter and a liquid
nitrogen-cooled MCT detector according to procedures
described previously [20]. Each of the studied peptides
(20 lg) and DMPC/DMPG (1 : 1, molar ratio) were mixed
in chloroform/meth anol (1 : 1 , v /v) s olution and equili-
brated with sodium pho sphate buffer at pH 5.0 or 7 .4 to
give a final peptide/lipid molar ratio of 1 : 50. The sample
was carefully spread on the germanium surface until solvent
had e vaporated. T he ATR sample c overed with a home-
made box was kept in full D
2
Ohydration(D
2
O/lipid ratio
> 35) based on infrared absorbance ratio of D-O/C-H
stretch peaks.
Three hundred scans w ere collected at a resolution of
2cm
)1
with triangular apodization and incoming r adiation

was polarized with a g ermanium single diamond polarizer
(Harrick, Ossining, NY, U SA). Before depositing sample,
the 45° germanium ATR-plate (2 · 5 · 50 mm) was
cleaned by a plasma cleaner (Harrick, Ossining, N Y,
USA). Analysis of ATR-FTIR data was performed in
accordance with a previous study [21].
NMR experiments
NMR samples were p repared by dissolving the IF P–wt
powder at 1 m
M
concentrationinH
2
O/D
2
O9:1(v/v)
aqueous buffer or 100 m
M
d
25
-SDS micellar solution.
Dilute H Cl or N aOH solution was used to adjust pH to
5.0. One- and two-dimensional
1
H N OESY and TOC SY
NMR experiments were performed on a Bruker AMX-500
spectrometer at 298 K (in SDS m icellar solu tion) or 278 K
(in aqueous solution), as described previously [22]. In
deuteron/hydrogen (D/H) e xchange experiments, the pep-
tide incorporated into SDS sample was lyophilized three
times with pure H

2
O. D
2
O/H
2
O 9 : 1 (v/v) was added
immediately before acquiring NMR data at 298 K a nd
pH 5.0. To measure the effect of Mn
2+
ions on the
relaxation behaviour o f IFP–wt protons, MnCl
2
dissolved
in H
2
O was added to IFP–wt micellar solution to give a fi nal
molar ratio of 0.696 (Mn
2+
/IFP–wt). Mn
2+
is an aqueous
ionic probe which is excluded from t he apolar core of the
micelle. Because the relaxation rate enhancement varies
inversely with the distance between the proton a nd the
probe, t he protons located more d eeply in the micellar
interior will be less affected by the spin p robe. The fraction
of attenuated backbone amide proton s ignal is taken as the
fractional intensity difference in the cross-peaks of NH/ aH
or side-chain protons of a given res idue obtained for the
protonated and deuterium-exchanged peptide (for exchange

experiments) or before and a fter introduction of Mn
2+
to
d
25
-SDS micellar s olution (for relaxation enhancement
measurements).
Structure calculations
Using d istance geometry (DG)/simulated annealing (SA)
protocols of
BIOSYM
programs
INSIGHTII
,
DISCOVER
and
NMRCHITECT
(version 2000.1) from Accelrys Inc. ( San
Diego, CA, U SA), 360 constraints were e mployed i n the
structural computations (Table 1). The intermolecular
NOE r estraints were classified semiquantitatively into three
categories: strong (less than 2.6 A
˚
), medium (2.6–3.6 A
˚
)or
weak (3.6–4.6 A
˚
). A range of 0.6–2.0 A
˚

was allowed to
vary in the distance constraints. In the SA protocol, the
temperature was raised to 1000 K in f our s teps followed by
amoleculardynamicsrunfor30pstoallowmore
conformational space to be explored. The system was
subsequently annealed to 300 K in 10 steps for a total of
55 ps and minimized by the steepe st-descent and conju-
gated-gradients m ethods before final refined structures
were obtained.
Ó FEBS 2004 Structural study of internal fusion peptide of ASLV (Eur. J. Biochem. 271) 4727
Results
CD experiments on IFP–wt and -F13 indicate helix
enhancement of the two peptides upon associating with
lipid bilayer
Figure 1 displays CD data on the two internal FP analogues
in aqueous s olution and i n DMPC/DMPG (1 : 1) v esicular
dispersions at pH 5.0 and 7.4 at 3 7°C. No sign ificant change
in the s econdary structure o ccurs upon acidification fr om
pH 7.4–5.0 for both analogues in DMPC/DMPG disper-
sion; the only peculiarity is a dramatic increase in helicity
with acidic pH for the F13 analogue in a queous solution.
On the other hand, the helix content is increased when IFP–
wt is transferred from aqueous to vesicular suspension, in
analogy to the N-terminal FP such as that of gp41 [22] and
of influenza HA2 [23], demonstrating the c onformational
plasticity of the v iral fusion pepti des. H owever, the helix
population in the membranous e nvironment is h igher for
IFP–wt than for these two N-terminal FPs. The h igh helix
content of F13 varian t in aqueous solution at acidic pH may
reflect the c ritical effect of proline on the helical propensity

of the fusion peptide sequence, but this structural effect
diminished on as sociating with t he membrane. A s will be
further demonstrated by NMR r esults, helix is induced in
the N-terminal portion of the peptide in the membranous
environment.
Binding of IFP–wt to membrane bilayer is detected
by NBD-labelled peptide
To investigate the natur e of me mbrane interaction o f I FP–wt
as manifested in the secondary structure change o f the
peptide upon binding to the membrane (Fig. 1), we utilized
IFP–wt with an N-terminally attached NBD which exhibits
greatly increased fluorescence emission in a less polar
environment. As illustrated in Fig. 2A, the fluorescence
intensity increased several f old when the labelled peptide was
transferred from aqueous buffer to vesicle dispersions,
providing direct evidence that the peptide (or at least its
N-terminal porti on) penetrates i nto t he membrane apolar
interior. To further investigate t he insertion depth, Stern–
Volmer constant K
SV
obtained from N BD quenching b y
Co
2+
was utilized. The K
SV
was calculated from the linear
part of Fig. 2B (in the range of [Co
2+
] ¼ 0–0.3 m
M

). The
data reveal that acidification of vesicular dispersion results in
deeper immersion of the peptide , at least at the N-terminus,
as reflected by a smaller K
SV
. In the bottom panel of Fig. 2B,
data from HA2(1–25) were used to compare the insertion
depth between the N-terminal and internal fusion peptides.
The N-terminal region of the N-terminal fusion peptide is
seen to penetrate more deeply than that of IFP–wt. The result
Table 1. C onstraints used for molecular simulation calculations on
IFP–wt in sodium dod ecyl sulfate micelle and the deviations f rom the
average stru ctures.
Constraint type
Total no.
constraints
No.
constraints
Constraints
Chiral 30
Dihedral 26
Distance 304
Subtype
Sequential 139
Medium
i,i+2: 54
i,i+3: 55
i,i+4: 52
Long
a

4
RMSD type
Deviation
range No. of deviation
Distance 0.5–1.0 A
˚
17
> 1.0 A
˚
0
Dihedral > 5° 19
>10° 7
>15° 7
>20° 0
a
d(i,j) constraints where j 3 i+5.
Fig. 1. Far-UV CD spectra of ASLV IFP–wt and IFP–F13 in aqueous
buffer and in DMPC/DMPG vesicle media at pH 5.0 (top panel) and 7.4
(bottompanel)at37°C. Helix content estimated from the ellipticity
value a t 222 n m is en hanced considerably as IF P–wt was trans ferre d
from aqu eous to vesicular solution at both pHs; this is true for IFP–F13
only at pH 7.4. The CD spectra of IFP–V13 are similar to those of IFP–
F13 under the same conditions. The insets to the panels display the
results of the secondary s tructure analysis with the
VARSELEC
program.
4728 S F. Cheng et al. (Eur. J. Biochem. 271) Ó FEBS 2004
will be discussed i n c onjunction with NMR data (Fig. 7)
obtained in SDS micellar solution. Furthermore, greater
insertion depth is observed for both fusion peptides at acidic

pH than at neutral p H. This may h ave implications on the
fusion pheno type of the viral fusion proteins since the fusion
function of both e nvelope proteins is sensitive t o pH [24].
Fluorescence resonance energy transfer data
demonstrate the lipid mixing activity of IFP–wt
To investigate the fusogenicity of the peptide, NBD- and
Rho-labelled PE w ere used a s the donor and acceptor,
respectively, of fluorescence energy transfer. Dilution of the
fluorescent-labelled PE loaded into v esicles by the unloaded
vesicles via membrane fusion induced by the fusion peptide
results in a reduction in the fluorescence energy transfer
efficiency, hence d equenching of the donor fluorescence. As
illustrated in Fig. 3, lipid mixing as de fined in Experimental
procedures is plotted against peptide to lipid ratio i ndicates
that the peptide is capable of promoting lipid mixing of
vesicles consisting of DMPC and DMPG. Additionally, a
peptide d erived from randomized sequence of IFP–wt (see
Supplementary data) exhibited insignifi cant lipid mixing
activity, demonstrating the specificity of the activity of the
IFP–wt sequence. Inspection of Fig. 3 revealed that lipid
mixing activity is higher for IFP–wt than the N-terminal
fusion peptide, HA2[1–25], of influenza v irus at the same
peptide-to-lipid ratio. The result is analogous to the data on
the Sendai viru s observed by Peisajovich et al. [ 25] showing
that the IFP has higher fusion activity than t he N-terminal
FP.
Self-quenching of rhodamine fluorescence indicates
self-assembly of IFP–wt in the lipid bilayer
To further examine the organization of the fusion peptide in
the membrane, the peptide was l abelled with rhodamine

[26]. Self-association of the molecules is monitored by the
Fig. 2. Mem brane binding and insertion depth o f IFP–wt p robed by
NBD fluorescen ce. (A)LipidbindingofNBD-IFP–wtatpH7.4and
37 °C. I ncreased intensity and the blue-shift of flu orescence of NBD
attached to the N-terminus o f the peptide in dicate emb ed ding o f the
peptide i n the apolar milieu of membrane bilayer. As a control, pro-
teinase K digestion of the peptide disrupts membrane binding releasing
bound NBD an d thus re du ces fluore scence of the fluorophore. (B)
Stern–Volmer plot of cobalt quenching of NBD-IFP–wt to probe the
immersion dept h in D MPC/DMPG vesicular suspension at pH 5.0
and 7.4. K
SV
values of HA2(1–25), the fusion peptide of influenza virus
are compared to IFP–wt. The calculated K
SV
values (based on the data
in the range of [Co
2+
] ¼ 0–0.3 m
M
) were sho wn in the inset. For both
fusion peptides, K
SV
is smalle r a t lower pH, indicating deeper penet-
ration than at neutral pH. At the s ame pH, larger K
SV
for IFP–wt
shows shallower immersio n o f t he peptid e i n t he vesicl e t han t he
N-terminal FP.
Fig. 3. Lipid mixing induced by IFP–wt as probed b y FRET at 37 °C.

NBD-andRho-labelledPEwereincorporatedinthevesiclestowhich
wereaddedfusionpeptideandunlabelled lipid disper sion. Mixing
rates are plotted against peptide-to-lipid ratios for I FP–wt and the
N-terminal FP of influenza virus. Fusion activity exhibits strong
pH de pendenc e for the IFP–wt. Lipid mixing rate is larger fo r IFP–wt
than the i nfluenza fusion peptide at a given P/L-value . D equenc hing
of the donor N BD b y dilution of the acceptor Rho resulting
from p eptide-me diated membrane fusion is normalized with respect
to the intensity obtained from lysis of vesicles w ith 0.2% Triton
X-100.
Ó FEBS 2004 Structural study of internal fusion peptide of ASLV (Eur. J. Biochem. 271) 4729
self-quenching of rhodamine fluorescence. The result shown
in Fig. 4A for IFP–wt i n DMPC/DMPG vesicular suspen-
sion indicates a moderate (approximately 40%) dequench-
ing after addition of the d etergent (Triton X-100) to the
suspension, suggestive of a loose a ssociation for the peptide
in the membrane bilayer.
Self-assembly can also b e analyzed by compositional
variation of rhodamine-labelled p eptide, keeping the total
concentration of labelled and u nlabelled peptides constant.
In the experiments le ading to Figs 4B, 0.01 l
M
of total
peptide was incorporated in DMPC/DMPG (150 : 150 l
M
)
vesicle dispersion for IFP–wt, with composition of the
labelled peptide (x) varying from x ¼ 1–0.05. All three
peptides exhibit c haracteristic of m ultimeric species in the
lipid bilayer, as displayed b y self-quenching of rhodamine at

x ¼ 1 compared to intensity at x ¼ 0.05 (for example,
quenching efficiency in excess of 3 for IFP–wt) and the
shape o f I
x
/x vs. x plot. Specifically, the initial slow rise of
the l atter plot (from x ¼ 1) reflects that the self-association
of the peptide is not tight ( on the scale of self-quenching
distance of rhodamine,  15 A
˚
[17]). The deviation of the
observed profiles from calculated ones based on homogen-
eous clustering of various multimers ( N ¼ 1, 2, 4 and 8 )
indicates more than one mode of association of the FP
molecules, involving m ore tightly packed oligomeric (such
as trimer as in other class I fusion proteins) subunits
interacting loosely with neighbouring oligomers. Thus, the
sharp i ncrease in I
x
/x near the x ¼ 0 r egion shows
dequenching of the probe and hence a tendency toward
random distribution of t he peptide in t he membrane in the
long distance range (> 30 A
˚
), indicating no large scale
aggregation occurs for the peptide.
SDS/PAGE experiments suggest propensity of
self-association of IFP–wt in the membrane-mimic
environment
Self-assembly of IFP–wt in the membranous environment
canalsobediscernedbySDS/PAGEdatashowninFig.5,

which also displays the results for IFP–F13 and IFP–V13.
All peptide analogues exhibit diffuse bands and migrate
roughly as dimeric species, with F13 and V13 variants
forming oliogomers of slightly higher mass. The data are in
qualitative agreement with the rhodamine self-quenching
result for IFP–wt ( Fig. 4). All peptides exhibit little
tendency of forming a large and tight molecular cluster.
As SDS micelle is considered to be strongly disruptive o n
Fig. 4. Self-a ssembly of the internal fusion peptide analogues i n associ-
ation with DMPC/DMPG vesicles at pH 7.4 and 37 °C. (A) Relative
intensity of Rho-labelled IFP–wt, –F13 and –V13 i n aqueous buffer
(unfilled bar), DMPC/D MPG vesicle (grey bar) a nd vesicle t reated
with proteinase K (black bar). The results indicate that IFP–wt has
lower propensity of forming oligome r in the vesicular me dium than the
other two varian ts and is probably monomeric in aq ueous solution as
Rho self-quenc hing is le ss than in the presence of v esicles for the
labelled peptide. (B) Normalized Rho emission intensity as a function
of th e fraction o f labelled peptide as a probe for self-aggregation. As
indicated in the plot, monome ric and dimeric species fo r th e p eptide
are re presented by the h ypoth etical h or izontal and diagonal lines,
respectively. The nearly unchanged I
x
/x in the h igh x regio n reflects a
tight shorter range, probably intratrimeric, packing but the sharp rise
in the low x region indicates a loose association for longer r ange inter-
trimeric interaction. This interpretation is based on the fact that, at
low x limit, th e probability of finding a labelled FP within a trimer for
a g iven rhodamine p robe is low thus the result emphasizes inter-trimer
interaction; the r everse is true in the x ¼ 1 limit. M oreover, the d evi-
ation of t he experimental curves f rom those calculated by assuming a

single species of a ssociation of N monomers indicates a m ulti-mode
association for the peptide analogues in the membrane bilayer. The
data therefore imply a heterogeneous distribution of FP molecules,
which can be interpreted by a more close-packed trimers interacting
loosely with a djacent trimers.
Fig. 5. SDS/PAGE measurements on the molecular association for the
three IFP analogues. In accord with the data o f Fig. 4A, IFP –wt has
lower propen sity of forming high order oligomer th an t he o ther two
analogues.
4730 S F. Cheng et al. (Eur. J. Biochem. 271) Ó FEBS 2004
the non–covalent interaction, the data of Fig. 4 provide
indirect evidence for p ropensity of IFP–wt and analogues
for self-assembly in the membranous medium.
Helical structure and insertion angle
of the membrane-associated fusion peptide
as measured by FT-IR spectroscopy
The secondary structure of peptides can also be quantitated
by infrared spectroscopy. As shown i n F ig. 6 , the helical
content calculated f rom t he band at 1655 cm
)1
after
deconvolution is 60% (Table 2), i n agreement with that
observed in the CD data of the wild-type peptide in the
vesicular s uspension at pH 7.4 (cf. Figure 1). The helix
population i n IFP–wt is h igher, while the b-sheet content is
lower, than that in the fusion peptide of influenza virus [20].
Slightly higher helix content was found for IFP–F13 (64%,
Table 2), and IFP–V13 (66%, data not shown), suggesting
helix structure i s not a sufficient determinant f or the f usion
activity of the internal fusion peptide of ASLV. Secondary

structure composition of the peptide varied little, if any,
with pH between 5.0 and 7.4 (C.W. Wu and D.K. Chang,
unpublished observation).
The insertion angle of t he peptide helix deduced from the
polarized ATR FT-IR result (Table 2) is  53° with respect
to the membrane normal. Compared to N-terminal f usion
peptides such as that of influenza virus [20], the result
indicates that IFP–wt associates with the bilayer in a more
shallow fashion. Still shallower insertion was obtained for
the IFP–F13 variant. There is an insignificant change in
insertion angle as pH varies from 5.0 to 7.4.
Secondary structure change and insertion depth for
membrane-bound fusion peptide can be determined
on the residue level by NMR measurements
The secondary structure and membrane insertion of the
fusion peptide can be examined at the atomic level by NMR
spectroscopy using a micelle as a model. Figure 7 A
summarizes the NOE interactions from proton p airs with
distance ¼ 5A
˚
in SDS micellar dispersion and in aqueous
buffer solution at p H 5.0. T he d
aN
(i,i+3) and d
aN
(i,i+4)
interactions, c haracteristic of helix structure, can be found in
the residues 1–10 and residues downstream of Pro13 in SDS
micelles. In accord with NOE data, helical segments, which
Fig. 6. ATR-FTIR results on IFP–wt in DMPC/DMPG vesicles at

pH 7 . 4. The top panel displays data from polarized FT-IR experi-
ments; the ratio of th e parallel to perpe ndicular components relates to
the angle of insertion of t he FP molecules into the bilayer. || and ^
represent parallel and perpendicular polarized light, respectively. The
bottom panel displays the result of deconvolution of the IR absorption
spectrum of the peptide f or analysis of the sec ondary struc ture. The
dominant pe ak at 1655 cm
)1
is attributed to the helical structure. The
dottedandsolidtracesareexperimentalanddeconvoluteddata,
respectively.
Table 2. S econdary structure and helix orientation of IFP–wt and -F13 in DMPC/DMPG vesicular dispersion as deduced from ATR-FTIR data.
a
The
peptide-to-lipid molar r atio was 1 : 50.
b
H values are angles between the helix axis and the bilayer normal.
pH 5.0 pH 7.4
IFP–wt IFP–F13 IFP–wt IFP–F13
Secondary structure percentage
a-Helix 57 ± 3 66 ± 3 60 ± 4 64 ± 3
Disorder 12 ± 2 11 ± 3 12 ± 3 11 ± 3
b-Sheet 24 ± 1 20 ± 2 22 ± 1 22 ± 1
b-Turn 7 ± 1 3 ± 2 6 ± 1 3 ± 1
Helix axis orientation
Order parameter S
amide I¢
0.032 ± 0.02 )0.19 ± 0.03 0.095 ± 0.04 )0.14 ± 0.06
H 53° 63° 51° 60°
Ó FEBS 2004 Structural study of internal fusion peptide of ASLV (Eur. J. Biochem. 271) 4731

are characterized by consecutive r esidues with spin–spin
coupling constant
3
J
aN
¼ 5 Hz, can be identified for t he
residues 3–10 and 14–28 as shown i n Fig. 7B. Also plotted
in Fig. 7B are
3
J
aN
data in aqueous buffer, indicating
residues 16–21 are more helical while residues 6–10 have
more b-strand character. C omparing
3
J
aN
values in aqueous
buffer and in mice llar s uspension, it is clear that helix is
induced in the N-terminal half f or the peptide in association
with SDS micelle. Thus helix enhancement o f t he peptide on
binding to the membrane shown in Fig. 1 is corroborated
by the data of Fig. 7.
The backbone amide deuterium/proton (D/H) exchange
can be used to p robe the immersion of the peptide in SDS
micelle with the notion that slower exchange correlates with
deeper penetration into the micelle for the residue under
consideration. Another method to gauge the insertion depth
is a r eduction of resonance intensity resulting from r elax-
ation enhancement by the aqueous spin probe, manganese,

the extent of which is highly dependent on the Mn
2+
–
proton distance [27]. The results o f these two experiments as
summarized in Fig. 7C are consistent in that the region
encompassing Pro13 is near the micellar headgroup region
and the residues 16–21 penetrate more d eeply into the
micellar interior than o ther regions of the peptide. Accord-
ing to Fig. 7C, the backbone of Arg22 resides near the
apolar–headgroup interface of the micelle; its side chain is
likely extended to t he headgroup region of the micelle
resulting in neutralization o f the positive ch arge by the
sulfate group.
Fig. 7. Sec ondary structure an d t opology of IFP–wt in SDS micelles
determined by NMR spectroscopy. (A) NOE interactions of IFP–wt in
aqueous buffer ( top) an d i n a ssociation w ith SD S mic elles ( bottom).
(Upper) Folded structure can be discerned in the region 10–18 and 20–
26. (Lower) Helical segments can b e observed i n 1–11 a nd 14–28 by
contiguous (i,i +3) interactions. The absence of (i,i +3) cross-peaks
between L eu 11 and Gly14 is consistent with a distorted h elix for the
region aro und Pro13. (B)
1
H sp in–spin coupling constants (
3
J
aN
)for
the residues of I FP–wt me asured i n aqu eous solutio n and S DS mice llar
suspension. Residues of the helical and b-sheet structure are charac-
terized b y values smaller than 5 Hz and greater than 8 Hz, respect-

ively. The tran sformation into helix of the N-terminal regions is
obviousasIFP–wtistransferredfromaqueoustomicellarmedium.
(C) Attenuation o f the intensity o f NH -aHcross-peaksinD/H
exchange and Mn
2+
relaxation enhancement experiments. The
standard error in computing signal attenuation is 5%. Mo re exposed
backbone amide proton results in faster ex cha nge between deuterium
and proton and hence smaller cro ss-peak reten tion. The relaxation
enhancement o f the backbone proto ns is inversely proportional to the
sixth p ower of Mn
2+
-proton distance; larger signal retention therefore
represents greater dep th of insertion of the amino a cid residue into the
micelle. Taken together, the two sets of data indicate t hat the stretch
around Pro13 is closer to the surface and the C-terminal half penetrates
more deeply than t he N-terminal half of IFP–wt.
Fig. 8. NH- aH and NH -side chain proton region of
1
HNOESYspec-
trum of IFP–wt in the aqueous medium. Interactions between the resi-
dues around Pro13 used to determine the type II b-turn are indicated.
Particularly not eworthy are the cross-peaks attributed to bHofAla12
and NH o f Val15, and aH of P ro13 and NH o f Val15.
4732 S F. Cheng et al. (Eur. J. Biochem. 271) Ó FEBS 2004
Conformational alteration of IFP–wt occurs when the
peptide is transferred from aqueous solution to SDS
micellar dispersion, particularly in the region near Pro13
NMR data of IFP–wt suggest an overall lower helicity in
aqueous solution compared to that in SDS micellar

solution. Because of the helix-disruptive property of proline,
we further e xamined the structure o f the region adjacent t o
Pro13 of IFP–wt. Several lines of evidence from NMR data
shown in Fig. 8 suggest that Pro13-Gly14 essentially form a
type II b-turn in aqueous solution, while a slightly kinked
helix is deduced in the presence of micelles. First, in aqueous
medium coupling constant
3
J
aN
for Gly14 is  6 Hz, closed
to 5 H z in a type II b-turn. Second, the r esult that
d
aN
(Pro13,Gly14) is stronger than d
aN
(Gly14,Gly14) indi-
cates t hat a type II instead of t ype 1 turn is adopted for the
Ala12-Pro13-Gly14-Val15 stretch. Third, the relationship of
interproton distance a lso provides evidence of the claimed
turn structure. Thus d
NN
(Gly14,Val15) is short for the
strong inten sity o f the cross-peak between the a mide
protons of the two residues; d
aN
(Gly14,Val15) is shorter
than in traresidue d
aN
(Val15,Val15); a prominent d

bN
(Ala12,Val15) peak is observed but not other d
bN
(i,i +3)
for i ¼ Ser9, Ile10 and Leu11, suggesting a b-turn exists in
the segment Ala12-Pro13-Gly14-Val15. A turn-to-helix
conformational change is a lso manifested in the upfield
shift of Val15 C
a
H from 4 .05 to 3 .85 p.p.m. as SDS micelles
are added to the medium although the helix is distorted near
Pro13 in the presence of the micelle. Another r esult in
support of the contention that Val15 is the fourth residue of
a t urn is a fforded by the small t emperature coefficient of
NH of Val15, namely 2.5 · 10
)3
p.p.m.Æ°C
)1
(compared to
8.5 · 10
)3
p.p.m.Æ°C
)1
for G ly14 NH) in aqueous solution,
suggesting that its N H p articipates i n hydrogen bonding.
The latter notion is also corroborated by a distinct
d
aN
(Pro13,Val15) cross-peak.
A kink at Pro13 is visualized by the structural calculation

based on the proton NOE and coupling constant
constraints
To obtain three-dimensional structure, we employed
INSIGHTII/DISCOVER
and
NMRCHITECT
using constraints
derived from NOE and
3
J
aN
. Figure 9 illustrates the
superposition of 20 structures (PDB ID: 1XNL). Detailed
structural statistics are tabulated in Table 1 . Note that the
helices have a kink of  20° at Pro13, consistent with the
effect of th e proline on the helix summarized by Barlo w and
Thornton [28].
Discussion
We have studied the s tructure of the IFP of ASLV-A and its
membrane interactions to examine the difference in the
biophysical property b etween the N-terminal and internal
FPs, in the hope of improving our understanding of the role
of fusion peptides in the fusion e vent.
Structure and membrane interaction of IFP–wt
Binding of IFP–wt to lipid vesicles as demonstr ated in
Fig. 2 supports the v iew that t he IFP o f A SLV is e ither
responsible for or contributes to the binding of the virus to
liposomes after exposure o f the fusion peptide in response to
the conformational change triggered by interaction with the
specific receptor [8].

As is the case for many IFPs, a proline is located near the
centre of the FP sequence. Both CD and NMR data
indicate a higher helix content than t he N-terminal FP (e.g.
influenza H A2 and HIV-1 gp41) in the membrane-mimic
environment. Figures 1, 7 and 8 also demonstrate b–a
transformation when th e peptide is transferred from aque-
ous to micellar medium; more specifically, NMR results
indicate that the helical structure is in duced mainly in the
region N-terminal to proline and there is a slight kink in the
helix at Pro13. As shown i n Fig. 7C, the region surrounding
Pro13 r esides closest to t he micellar surface. A kinked h elix
structure has also been deduced from an NMR study on the
fusion peptide of HA2 in dodecylphosphocholine micellar
solution [29], indicating that it is therefore probably a
common f eature for the class I viral fusion peptides
embedded i n the membrane and is involved in destabilizing
the membrane during the fusion process. The notion is
supported by a mutational study on ASLV fusion protein in
which the viral infectivity was found largely unaffected by
the substitution of Pro13 with a residue thought to retain the
bending or flexibility of the central region of IFP [30].
Except for t he moderate distor tion on the helical structure,
Pro13 does not significantly reduce helicity of the wt-peptide
in lipid a s observed from the comparison of CD data
(Fig. 1 ) between the p eptide and i ts F13 variant.
The conformational switch for the stretch near Pro13
from a type I I b-turn to a kinked h elix as the f usion peptide
is transferred from aqueous to SDS micellar medium
underscores our previous contention r egarding t he
Fig. 9. Supe rposition of 2 0 structures of IFP–wt calculated with t he

constraints derived from NMR data performed in SDS micelles. Akink
caused by Pro13 s hows that th e a mino acid may have a structural effect
on helix in the membrane-mimic medium s imilar to t hat in a queo us
solution.
Ó FEBS 2004 Structural study of internal fusion peptide of ASLV (Eur. J. Biochem. 271) 4733
N-terminal fusion peptide that the structural plasticity is
germane to its fusogenicity and a proper balance between
helix and b-sheet structure i s important for fusion activity.
It is thus likely that both internal and N-terminal fusion
peptides destabilize the target membrane in a s imilar
fashion, n otwithstanding a shallower penetration of the
IFP into the membrane.
It is noteworthy that substitution of Pro13 by a residue
with intermediate hydrophobicity (Thr for instance) resulted
in an MLV-pseudotyped virus with h igher i nfectivity [8]; the
activity, moreover, did not correlate with the predicted
secondary structure propensity of the substituted amino
acid. The result is in line with our finding that Pro13 is
located in the interfacial region having transitional polarity.
Introduction of a residue with extreme polarity may perturb
the topology o f the fusion peptide and its interaction with
the membrane bilayer. It is of interest to note the ph enotype
of H IV-2 gp41 mutants [31]: t he amino acid substitution of
the fusion peptide sequence that increases hydrophobicity of
the region would enhance syncytium formation, while
mutation that increases charge or polarity of the fusion
peptide domain reduces syncytium-inducing c apacity of t he
virus. Coupled with our previous finding of a deeper
insertion of t he fusion peptide of HIV-1 gp41 into model
membrane [22], the result of Steffy et al. [31] c orroborates

the assertion that Pro13 of IFP–wt re sides near the
interfacial region of the bilayer.
The orientation and i nsertion depth w ere probed b y
FTIR and NMR techniques. The angle between the helix
axis and b ilayer surface deduced from FTIR d ata is
 37° for IFP–wt ( 27° for IFP–F13) but  50° for
HA2-FP [20]. Hence the IFP–wt inserts into bilayer less
steeply than the N-terminal fusion peptide of i nfluenza
virus. The insertion depth obtained from NMR data
(D/H exchange and Mn
2+
probe) in the SDS micelle
suspension reveals that the N-terminal portion of the
N-terminal fusion peptide [15] is embedded mo re deeply
than the C-terminal segment whereas the r everse is true
for IFP–wt. Moreover, the estimated angle of insertion
from the relaxation enhancement measurements is flatter
and closer to the micelle surface for IFP–wt. This
suggests th at IFP has a shallower i nsertion into the
membrane. It also follows that penetration into both
leaflets of the membrane bilayer is not required for the
IFP to exert its f unction. The r esult is reasonable a s the
charged and polar residues flanking the IFP sequence
would p revent the FP from immersing too d eeply into
the apolar core of the membrane. Our data are also
compatible with the observation that truncation of t he
membrane-spanning region of HA2 to the extent that the
resulting segment cannot cross t he bilayer l ed to hemi-
fusion but not complete fusion [32] and with the report
that lipid-anchored influenza haemagglutinin promotes

hemifusion [33]; both imply that the complete fusion
cannot be supported b y t he fusion peptide domain alone
and the transmembrane segment is indispensable. The
deeper penetration at pH 5.0 than at pH 7.4 for both
HA2 and ASLV fusion peptides as revealed by Fig. 2B
may be r elevant i n t he pH dependence of fusion a ctivity
for both v iruses; it may be that further perturbation of
membrane structure with deeper insertion would promote
progress to later s tages of fusion reaction, for example
deeper penetration into membrane at acidic pH may
facilitate p rogress from h emifusion achieved by ASLV at
neutral p H t o pore formation [34]. Comparison o f t he
orientation of IFP–wt and IFP–F13 in association with
phospholipid bilayer (Table 2) also suggests that a small
angle of insertion into the membrane may exist for the
internal fusion peptide [35] to destabilize the membrane.
An oblique insertion of fusion p eptide of SIV gp32 and
influenza HA2 has been deduced by Martin et al.[36]
and by Lu
¨
nerberg et al. [37] from FT-IR measurements.
Lipid mixing activity of IFP–wt and its relationship
to the secondary structure
Lipid m ixing measurements on the internal and N-terminal
fusion peptides indicate that the IFP has higher fusion
activity than N-terminal fusion peptide of influenza v irus
(Fig. 3 ). Both types of fusion peptides insert into the
membrane bilayer obliquely and possess substantial regular
structures (a-helix and b-sheet) in t he membranous medium,
in supp ort of the idea that balanced helix and b-sh eet

structures are necessary for fusion. Indeed, higher helix
content alone does not correlate with fusogenicity as can be
seen from comparison of helical structure between IFP–wt
and IFP–F13.
Organization of IFP–wt in model membranes
An apparently dimeric band can be detected in the S DS/
PAGE measurement for IFP–wt. A species of slightly
higher apparent molecular mass i s observed for each of the
less fusogenic analogues IFP–F13 and IFP–V13, suggesting
that oligomerization of fusion peptide m olecules i n t he
membrane is not sufficient for their fusion activity. This i s i n
contrast with the oligomerization propensity observed for
HA2-FP and its mutants [17]. Both the simple rhodamine
self-quenching and the rhodamine-labelled peptide compo-
sition variation experiments (Fig. 4B) c learly demonstrate
self-assembly for the membrane-bound peptides; the latter
further provided e vidence o f m ultimodal s elf-assembly of
the p eptide, a s deduced, in particular, from a steeper rise in
I
x
/x in the region near x ¼ 0. The oligomeric structure for
the ectodomain of the envelope glycoprotein of ASLV has
also been inferred from sucrose gradient sedimentation
experiments [38]. The results of Fig. 4 lead t o t he view that
loose association of the IFP molecules in the membranous
environment is r elevant t o the lipid mixing activity. In
analogy to t he proposed multi-modal association, the
clustering of trimeric envelope proteins on the surface of
HIV virion has been elegantly shown by Zhu et al. [39] in
their e lectron microscopic s tudy. Our data suggest t hat IFP

contributes to the self-assembly of the envelope glycoprotein
of ASLV bound to membrane.
Effects of pH on the secondary structure and membrane
interaction of IFP–wt
Based o n studies on the molecular clustering of EnvA fusion
protein of ASLV i n response to the receptor binding and
low pH, Matsuyama et al. [24] proposed a fusion mecha-
nism for the virus. Specifically, low pH promotes proceeding
of fusion from outer monolayer mixing to complete fusion.
4734 S F. Cheng et al. (Eur. J. Biochem. 271) Ó FEBS 2004
The result of Fig. 3 suggests that the fusion activity of IFP–
wt is highe r while Fig. 2 shows that t he insertion d epth is
deeper at low pH. Taken together, these data and the model
of Matsuyama et al . [24] are suggestive of the idea that
deeper membrane insertion induced by acidic pH facilitates
fusion of both inner and outer leaflets of fusing membranes.
Comparison between class I and II fusion proteins
The crystal structures of soluble domain of two class II
fusion proteins in association with detergent have been
reported recently [10,11]. The secondary structure of both
proteins is mainly b-sheet in the putative pre- and postfusion
forms. The region corresponding to the fusion peptide o f
class I fusion protein exists in the membranous environment
as a loop (hence t ermed fusion loop), which undergoes
minor conformational change as the region is in association
with phospholipid. In c ontrast, helical structures are g reatly
enhanced in the m embranous environment for both the
N-terminal and internal F Ps of class I fusion proteins as
they are transferred from aqueous medium with a b-turn in
the middle of the sequ ence, as suggested in t he present work

(Fig. 7 ) and for gp41 fusion peptide [12]. F urthermore, the
core structure of the class I fusion proteins, which also
undergoes structural r earrangement on transition t o the
fusogenic s tate, is p redominantly helical in both p re- and
postfusion states. We have previously deduced from our
NMR and fluorescence data that the fusion peptides of gp41
of HIV-1 and HA2 do not span both leaflets of the
membrane bilayer [ 15,22]. The I FP of ASLV-A is shown to
penetrate more shallowly into the membrane than the
N-terminal FP, in analogy to the IFP of class II fusion
proteins [10,11]. In conjunction with the model proposed for
the class II fusion proteins, it is c oncluded t hat all the v iral
FPs and fusion loops examined thus far insert into the outer
leaflet of the membrane.
Our result of variation of fluorescence quenching with the
labelled rhodamine composition (Fig. 4B) i s c ompatible
with the result on t he interactions of soluble fusion protein
fragments of Semliki Forest virus with membrane [10]. Thus
the i nsensitivity of the normalized intensity to variation in
the labelled composition at the high fraction regime
corresponds to intra-trimer interaction, whereas the steep
intensity increase in the low label fr action regime reflect s
inter-trimeric interaction between the peptide molecules in
the a djacent trimers, w hich amounts to a four-fold i ncrease
as judged from the slope of intensity increase. Qualitatively,
this means that the inter-trimer inte raction is not strong and
no tight high-order molecular a ssociation exists for t he
fusion peptide in the bilayer.
A 37-mer peptide that contains cysteine residues a t the
starting and ending positions b ridged by disulfide bond has

been synthesized. It was found that the proton chemical shift
and
3
J
aN
coupling constant values are c lose for the residues
within the 37-mer peptide overlapping with IFP–wt. Fur-
thermore, the longe r peptide exhibits lipid mixing activity
and secondary structure similar t o IFP–wt (see supplement-
ary data). These results suggest that the conformation of the
28-mer IFP–wt is similar to the 37-mer cysteine-containing
peptide i n the membranous environment.
The p resent work represents an extensive biophysical
study on the internal fusion peptide of a prototypic class I
viral fusion protein. We h ave observed s imilarities between
the internal and N-terminal fusion peptides, namely,
substantial helix enhancement on binding to membrane,
loose self-assembly and that the fusion peptides do not
traverse both leaflets of the membrane bilayer. Some distinct
differences between the internal and N-terminal fusion
peptides in the membranous environment are: higher helix
and lower b-sheet contents, shallower insertion for the
internal fusion peptide. M ore study on how the internal
fusion peptide segment along with the heptad repeat regions
and transmembran e anchor promotes membrane fusion
would be very useful in the antiviral strategy. Comparison
of the membrane-fusion mechanism b etween the fusion
proteins using internal a nd N-ter minal fusion peptides to
insert into target cell in terms of structural rearrangement
would be of fundamental and practical importance.

References
1. White, J.M. (1992) Membrane fusion. Science 258, 917–924.
2. Hunter, E. & Swanstrom, R. (1990) Retrovirus envelope glyco-
proteins. Curr. Top. Microbiol. Immunol. 157, 187–253.
3. Moore, J.P., Jameson, B.A., Weiss, R.A. & Sattentau, Q.J. (1993)
Viral F usion M echanisms, p p. 233–289. CR C P ress, Inc, Boca
Raton, FL, USA.
4. Gilbert, J.M., Hernandez, L.D., Balliet, J.W., Bates, P. & White,
J.M. (1995) Receptor-induced conformational changes in the
subgroup A a vian leukosis and sarcoma virus e nvelope glyco-
protein. J. Virol. 69, 7410–7415.
5. Smith, J.G., Mothes, W., Blacklow, S .C. & Cunningham, J.M.
(2004) The mature avian leukosis v irus subgroup a envelope gly-
coprotein i s m etastab le, an d r e folding in duc ed by the synergistic
effects o f re cepto r bind ing and low pH is coupled t o infection.
J. Virol. 78, 1403–1410.
6. Hernadez, L.D. & White, J.M. (1998) Mutational a nalysis of the
candidate internal fusion peptide of the avian leukosis and sar-
coma virus subgroup A envelope glycoprotein. J. Virol. 72, 3259–
3267.
7. Boerkoel, C.F., Federspiel, M.J.,Salter,D.W.,Payne,W.,Crit-
tenden,L.B.,Kung,H.J.&Hughes,S.H.(1993)Anewdefective
retroviral vector sy stem based on the bryan strain of rous sarcoma
virus. Viro logy 195, 669 –679.
8. Delos, S.E., G ilbert, J.M. & White, J .M. (2000) The c entral proline
of a n internal vira l fusion p eptide s erves two i mportant roles.
J. Virol. 74, 1686–1693.
9. Damico, R. & Bates, P. (2000) Soluble receptor-induced r etroviral
infection of receptor-deficient cells. J. Virol. 74, 6 469–6475.
10. Gibbons,D.L.,Vaney,M C.,Rousel,A.,Vigoroux,A.,Reilly,

B., Kielan, M. & Rey, F.A. (2004) Conformational change and
protein–protein interactions of the fusion protein of Se mliki
Forest virus. Na tur e 427, 320–325.
11.Modis,Y.,Ogata,S.,Clements,D.&Harrison,S.C.(2004)
Structure of the dengue virus envelope protein after m embrane
fusion. Nature 42 7, 313–319.
12. Chang, D.K., Chien, W .J. & Cheng, S.F. (1997) The FLG
motif in the N-te rminal region of glycoprotein 41 of human
immunodeficienc y virus type 1 a dopts a type I b- turn in aqueous
solution and serves as the initiation site for helix formation. Eur. J.
Biochem. 247, 896–905.
13. Rapaport, D. & Shai, Y. (1992) Aggregation and organization of
pardaxin in phospholipid membranes. A fluorescence energy
transfer study. J. Biol. Chem. 267 , 6502–6509.
14. Kantchev, E.A.B., Cheng, S.F., Wu, C.W., Huang, H.J. & Chang,
D.K. (2004) S econdary s tructure, p hospholipid membr ane inter-
actions, and fusion activity of two glutamate-rich analogs of
Ó FEBS 2004 Structural study of internal fusion peptide of ASLV (Eur. J. Biochem. 271) 4735
influenza hemagglutinin fusion peptide. Arch . Bio chem. Biophy.
425, 173–183.
15. Chang, D.K., Cheng, S.F., Trivedi, V.D. & Yang, S.H. (2000) The
amino-terminal region of the fusion p eptide of influenza virus
hemagglutinin HA2 inserts into sodium dodecyl sulfate micelle
with residues 16–18 at the aqueous boundary at acidic p H:
oligomerization and the c onformational flexibility. J. Biol. Chem.
275, 19150–19158.
16. Johnso n, W.C. (1990) Protein secondary structu re and circular
dichroism: a practical guide. Proteins Struct. Funct. Genet. 7,
205–214.
17. Cheng, S.F., K antchev, A.B. & Chang, D.K. (2003) Fluorescence

evidence for a loose self-assembly of t he fusion p ep tide of influenza
virus HA2 in the lipid bilayer. Mol. Memb. Biol. 20, 345–351.
18. Kliger, Y., Aharon, A., Rapaport, D., Jones, P., Blumenthal, R. &
Shai, Y. (1997) Fusion peptides derived from the HIV type 1
glycoprotein 4 1 associate within phospholipid me mbranes and
inhibit cell–cell fusion: structure–function study. J. Biol. Chem.
272, 13496–13505.
19. Struck, D.K., Hoekstra, D. & Pagano, R.E. (1981) Use of
resonance energy transfer to monitor membrane fusion. Bio-
chemistry 20, 4093–4099.
20. Wu, C.W., Chen g, S.F., Huang, W.N., Trivedi, V.D., Veeramuthu,
B.,Kantchev,A.B.,Wu,W.G.&Chang,D.K.(2003)Effectsof
alterations of the amino-terminal gly cine of influenza hema-
gglutinin fusion peptide on its structure, organization and
membrane interactions. Biochim. Biophys. Ac ta 1612, 41–51.
21. Tamm, L.K. & Tatulian, S.A. (1993) Orientation of functional
and nonfunctional PTS permease signal sequences in lipid bilayers:
a polarized attenuated to tal reflection infrared study. Biochemistry
32, 7720–7726.
22. Chang, D.K., C heng, S .F. & Chien, W.J. (1997) The amino-
terminal fusion domain peptide of human immunodeficiency virus
type 1 gp41 inserts into the sodium dod ecyl sulfate micel le pri-
marily as a helix with a conserved glycine at the micelle–water
interface. J. Virol . 71, 6593–6602.
23. Lear, J.D. & DeGrado, W.F. (1987) Membrane binding and
conformational properties of peptides representing the NH
2
ter-
minus of influenza HA-2. J. Biol. Chem. 262, 6500–6505.
24. Matsuyama, S., Delos, S.E. & White, J.M. (2 004) Sequential roles

of receptor binding and low pH in forming conformations of a
retroviral envelope glycoprotei n. J. Virol. 78 , 8201–8209.
25. Peisajovich, S.G., Samuel, O. & Shai, Y. (2000) Paramyxovirus F1
proteinhastwofusionpeptides:implications for the mechanism o f
membrane fusion. J. Mol. Biol. 296, 1353–1365.
26. Blackman, M.J., Cor rie, J.E., Croney, J.C., K elly, G., Eccleston,
J.F. & Jameson, D.M. (2002) Structural and biochemical char-
acterization of a fluorogenic rhodamine-labelled malarial protease
substrate. Biochemistry 41 , 12244–12252.
27. Lindberg, M . & Graslund, A . ( 2001) The position o f t he cell
penetrating peptide pen etratin in SDS micelles determined b y
NMR. FEBS Lett. 497 , 39–44.
28. Barlow, D.J. & Thornton, J.M. (1988) Helix geometry in proteins.
J. Mol. Bio l. 201, 601–619.
29. Han,X.,Bushweller,X.H.,Cafiso,D.S.&Tamm,L.K.(2001)
Membrane structure and fusion- triggering conformational change
of the fusion domain f rom i nfl uenza hemagglutinin. Nat. Struct.
Biol. 8, 715–720.
30. Balliet, J.W., Gendron, K. & Bates, P. ( 2000) Mutational analysis
of th e subgroup A avian s arcoma and leuko sis virus putative
fusion peptide domain. J. Virol. 74 , 3731–3739.
31. Steffy, K.R., Kraus, G., Looney, D.J. & Wong-Staal, F. (1992)
Role of the fus ogenic p eptide s equenc e in sync ytium in duction a nd
infectivity o f human immunodeficiency virus t ype 2. J. Virol. 66,
4532–4535.
32. Bagai, S. & Lamb, R.A. (1996) Truncation of the COOH-terminal
region of the p aramyxovirus SV5 fusion protein leads to he mi-
fusion but not complete fusion. J. Cell Biol . 135, 73–84.
33. Kemble, G.W., Danieli, T. & White, J.M. ( 1994) Lipid-anc hored
influenza hemagglutinin promotes hemifusion, not complete

fusion. Cell 76, 383–391.
34. Earp,L.J.,Delos,S.E.,Netter,R.C.,Bates,P.&White,J.M.
(2003) The avian retrovirus av ian sarcoma/leucosis v irus subtype
A reaches the lipid mixing stage of fusion at neutral pH. J. Virol.
77, 3058–3066.
35. Brasseur, R. (2000) Tilted pe ptides: a motif for membrane desta-
bilization (hypothesis). Mol . Membr. Biol. 17, 31–40.
36. Martin, I ., Dubois, M.C., Defrise-Quertain, F ., Saermark, T.,
Burny, A., B rasseur, R. & Ruysachaert, J.M. (1994) Correlation
between fusogenicity of synthetic modified pep tides correspon ding
to the NH
2
-terminal extremity of simian immun odeficien cy virus
gp32 and their mode of insertion into the lipid bilayer: an infrared
spectroscopy study. J. Virol. 68, 1139–1148.
37. Lu
¨
neberg, J., Martin, I., Nu
¨
ßler, F., Ruysschaert, J.M. &
Herrmann, A. (1995) Structure and topology of the influenza
virus fusion peptide in lipid bilayers . J. Biol. C hem. 270, 27606–
27614.
38. Einfeld, D. & Hunter, E. (1988) Oligomeric s tructure of a proto-
type re trovirus glycoprotein. Proc. Natl Acad. S ci. USA 85, 8688–
8692.
39. Zhu, P., Chertova, E., Bess, J ., Lifson, J.D., Arthu r, L.O., Liu, J .,
Taylor, K.A. & Roux, K.H. (2003) Electron tomography
analysis of envelope glycoprotein t rimers on H IV and sim ian
immunodeficiency v irus virion s. Proc. N atl Acad. Sci. USA 100,

15812–15817.
Supplementary material
The following material is available from http://www.
blackwellpublishing.com/products/journals/suppmat/EJB/
EJB4436/EJB4436sm.htm
Fig. S1. Backbone amide proton chemical shift in SDS
micelles of IFP–wt.
Fig. S2. Comparison of the chemical shift deviation a nd
coupling constant results between IFP–wt and IFP(8–44)
allows examination of the conformation of the two peptide
analogues in SDS micelles.
Fig. S3. FRET and rhodamine self-quenching data of
IFP(8–44) in DMPG/DMPC (1 : 1) at pH 5.0, 37 °Care
shown.
Fig. S4. FRET measu rements on the a ctivity of lipid mixing
induced by an IFP–wt sequence-scrambled peptide, show-
ing the specificity of the fusogenic function for I FP–wt.
4736 S F. Cheng et al. (Eur. J. Biochem. 271) Ó FEBS 2004