DSpace at VNU: Orientation determination of interfacial bent alpha-helical structures using Sum Frequency Generation vibrational spectroscopy

Chemical Physics 447 (2015) 15–21

Contents lists available at ScienceDirect

Chemical Physics
journal homepage: www.elsevier.com/locate/chemphys

Orientation determination of interfacial bent a-helical structures
using Sum Frequency Generation vibrational spectroscopy
Khoi Tan Nguyen ⇑
Department of Applied Chemistry, School of Biotechnology, International University, Vietnam National University of HCMC, Ho Chi Minh City 70000, Viet Nam
School of Chemical Engineering, University of Queensland, St. Lucia, QLD 4072, Australia

a r t i c l e

i n f o

Article history:
Received 29 July 2014
In final form 26 November 2014
Available online 6 December 2014
Keywords:
Sum Frequency Generation
Melittin
Bent helix
Supported lipid bilayers
Antimicrobial peptides

a b s t r a c t
Sum Frequency Generation (SFG) has been shown to be a powerful and versatile technique in studies of
proteins/peptides at surfaces and interfaces. Recently SFG was successfully applied in studies of

interfacial macro-molecules with increasing size and complexity. In this report we continued to employ
bond additivity model and group theory to demonstrate the importance of both the inter-helical tilt angle
and the lengths of the helical segments assembling the structures being studies. Specifically, a newly
improved SFG data analysis of multiple a-helical structures on melittin was used to interpret the SFG
experimental observation and also verified the findings with the recent insights brought by other
spectroscopic techniques.
Ó 2014 Elsevier B.V. All rights reserved.

1. Introduction
Sum Frequency Generation (SFG) signal in the amide I band of
the a-helical secondary structure was first observed in 2005 by Chen
and coworkers [1]. Since then SFG data analysis methods have been
continuingly developed and applied to a variety of biological
molecules of different levels of complexity including melittin, magainin 2, cecropin P1, fibrinogen, alamethicin, MSI-78, tachyplesin I,
cytochrome, human islet amyloid polypeptide, and heterotrimeric
G-protein [2–7]. The orientation analysis of simple biological molecules consisting of a single a-helical segment has been successfully
carried out on small peptides such as magainin 2, alamethicin and
MSI-78 [3–5]. Due to their simple structural properties and interaction schemes, these analyses were carried out using just the SFG signals in ssp and ppp polarization combinations. For more complex
structures/interaction schemes which inherently consist of more
orientational parameters, SFG ssp and ppp signals are not sufficient
to provide target orientational information because these two pieces
of information can only be used to solve for a single orientation
parameter if the absolute peptide coverage is unknown. In such
cases, complementary spectroscopic techniques (typically ATRFTIR), as well as additional computational ab-initio simulation and
mathematical approaches/information theories, have been required
to provide extra orientational information about the interfacial species [8–11]. Even though these complementary approaches have
⇑ Address: School of Chemical Engineering, University of Queensland, St. Lucia,
QLD 4072, Australia. Fax: +61 7 3365 4199.
E-mail address:
/>0301-0104/Ó 2014 Elsevier B.V. All rights reserved.

certainly been useful in the interpretation of various biological secondary structures at the interfaces, they all have intrinsic limitations
in the computational algorithms or methods of approximation. For
instance, the transition Raman polarizability of an oscillator can
only be approximated due to its indefinite number of virtual excited
states in the Raman process.
A systematic study employing group theory and the bond
additivity model in the orientation determination of single ahelical structures of different lengths using ssp and ppp SFG
amide I signals has been previously reported [12]. However, to
my knowledge, there have been no published reports of direct
and formal determination of the interfacial orientation of a general multi-helical structure. There have been a few recent SFG
studies on bent helical structures [13–15]; however, the computational details the successive Euler’s transformation implementations in the calculation of the net dipole moment and the
polarizability tensor were not explicitly discussed. In addition,
these studies assume strong vibrational coupling among all the
backbone C@O in the protein molecules; which may not be true
if there are substantial symmetry point group interruptions
among helical units or when they are distant from each other.
Although the orientations of cytochrome b5 in model lipid bilayers have been reported recently, the validity of the analysis
relied primarily on the assumed internal cancelation of the SFG
hyperpolarizability which arises from helical segments pointing
in opposite directions [11]. Unfortunately, a general multi-helical
structure may not possess such conformation, notably pardaxin,
melittin, and cytochrome P450.


16

K.T. Nguyen / Chemical Physics 447 (2015) 15–21

There have also been a number of excellent SFG studies on peptides/proteins using both the amide I band and the C-H (or C-D)

vibrational modes in the 2800–3100 cmÀ1 (or 2000–2200 cmÀ1)
regime. These studies tackle the orientation problem from different
aspects such as investigating certain particular amino acid residues
[7] or by studying the hyperpolarizability of the backbone C@O
bonds. The latter approach can be performed either qualitatively or
quantitatively depending on the information being sought [6,16,17].
For quantitative studies, the analyses may become unexpectedly complicated and the level of accuracy depends significantly
on the models being implemented in the calculation. Theoretically,
the a-helical amide I band was shown to consist of two orthogonal
vibrational modes A and E [18]. The double mode added to the
complexity of calculating the multi-helical molecular hyperpolarizability. This study successfully investigates the significant impact
of the molecular twist angle on the SFG amide I band signal
obtained while this twist angle has been omitted in a few
over-simplified SFG analyses of bent-helical structures [10,19]. In
addition, the inter-helical twist angle was confirmed to be not a
crucial factor contributing to the SFG Ippp/Issp value, which is commonly used in the orientation analysis of helical structures. Finally,
we took into account the length of the helical motifs in the data
analysis to characterize the molecular orientation of melittin in
lipid bilayers and verify the findings with current literature. The
molecular orientation of melittin is of great interest since it gives
direct hints to the peptide mode of action, which is still being hotly
debated. Traditionally, two main peptide-lipid interaction models
have been proposed: barrel stave [20] and carpet/toroidal pore
models [21,22]. More recent studies observed a more sophisticated
interaction picture in which melittin adopts a dual orientation
distribution [10,23] or the pore formation occurs as a transient
process [24,25]. In this present study, an orientation distribution
that aligns with recent studies on the mode of action of melittin
was proposed, on the basis of the kink in the molecular structure
of melittin being taken into account and the experimental evidence

that the peptide adopts a single d orientation distribution reported
recently using dual-fluorescence spectroscopy [26]. It is worth
emphasizing that the analysis in this present study may not be able
to describe the peptide/lipid interaction in more complex cases in
which the single d orientation distribution condition is not met; in
such cases, further parameters can be sought using other optical
spectroscopic techniques such as ATR-FTIR or FWM.
2. SFG data analysis – normal vibrational modes of multiple ahelical structures
In fact, this paper is an advancement of a previously developed
theoretical background [12,27]; interested readers may find these
materials particularly helpful in providing fundamental details
for the present study. Therefore, full details on the SFG orientation
analysis of single Pauling’s a-helical structures of different chain
lengths will not all be reiterated here. However, certain concepts
which are believed to be crucial for the development of this paper
will be selectively discussed.
A sum frequency process can be considered as a hyper-Raman
process of which hyperpolarizability is described by a third rank
tensor. This hyperpolarizability can be calculated using the transition dipole moment and the transition polarizability tensors.
Hence, the normal vibrational modes of the infrared absorption
and Raman scattering processes should be thoroughly studied.
The hyperpolarizability tensor of the SFG process can be
expressed as a combination of the IR absorption (x2) and the
Raman scattering (x1):

bijk ¼

X
v

~ ij ðxSF Þjv ihv jl
~ k ðx2 Þjgi
hgja

ð0Þ
g

ð0Þ

q À qv
x2 À xv þ icv

ð1Þ

where xSF ¼ x1 þ x2 ; xv is the frequency at which vibrational resonant transition occurs from jg i to jv i. The quantity

qgð0Þ Àqð0Þ
v
x2 Àxv þicv

appears in the expression as a line-shape function. The q values
are the fractional populations at the vibrational states; while c dictates the line width of the spectral peak corresponding to the indicated vibrational transition.
The SFG macroscopic susceptibility tensor element vijk ði; j; k ¼ x;
y; zÞ is related to the SFG molecular hyperpolarizability tensor element blmn ðl; m; n ¼ a; b; cÞ by an Euler angle projection [18,27–29]:

vijk;q ¼ Ns

X

^Án

b k
^ Þiblmn;q
hð^i Á ^lÞð^j Á mÞð

ð2Þ

l;m;n

Eq. (1) reveals the dependence of the hyper-Raman tensor on
the IR absorption and Raman scattering. Each hyperpolarizability
tensor corresponds to a vibrational mode that is both active among
IR and Raman normal modes. Therefore, the number of normal
modes of the SFG process is reduced from three to two: A
(symmetric, along the z axis) and E (asymmetric, in the xy plane)
modes. To illustrate this, the A and E modes are presented
graphically in Fig. 1.
It can be seen from Fig. 1 that the overall molecular symmetric
A mode is directly dictated by the relative tilt angle between the
two adjacent helical segments; whilst it is unlikely that the
molecular E mode will be influenced by the corresponding twist
angle. The process of calculating the molecular hyperpolarizability
tensors could be simplified if these two vibrational modes are well
separated spectrally; unfortunately, the vibrational energy of the A
and E modes are inherently only a few wavenumbers apart in the
SFG spectra. For this reason, the SFG signals of the A and E modes
cannot be resolved due to the limited resolution of a typical SFG
system. Thus both modes are likely to be contributing to the amide
I signal as shown below [10]:

vzzz ¼ vE;zzz þ vA;zzz

ð3Þ

vyyz ¼ vxxz ¼ vE;yyz þ vA;yyz

ð4Þ

vyzy ¼ vxzx ¼ vzxx ¼ vzyy ¼ vE;yzy þ vA;yzy

ð5Þ

The calculation of the hyperpolarizability tensors using the
bond additivity model and group theory has previously been systematically reported [12]. In this model, the hyperpolarizability
tensors are first calculated in the molecular fixed frame, which
takes into account the relative positions of the amino acid units.
Hence, this analysis is highly dependent on the structural properties of the molecule, as will be discussed later. The microscopic
hyperpolarizability tensors are then calculated and transformed
into measurable macroscopic quantities in the laboratory fixed
frame using the appropriate Euler’s transformations [28]. This

A mode

z

z
E mode

y

y

x

x
A mode
Fig. 1. A (left) and E (right) vibrational mode illustrations of an a-helix.


17

K.T. Nguyen / Chemical Physics 447 (2015) 15–21

transformation basically relates the intrinsic molecular hyperpolarizability tensors to the varying behaviors of the molecule in
the specific coordinate system.
In this report, the hyperpolarizability tensors of the individual
helical segments will be calculated and subsequently summed
together with their corresponding modes. The validity of this treatment depends heavily on the presumed rigidity in the molecular
conformation of the peptide/protein being studied. Not being a
high resolution technique, SFG analyses typically require the
molecular conformation acquired by higher resolution means. For
that reason, SFG typically works with other established high resolution techniques, such as X-ray crystallography and nuclear magnetic resonance, in boosting the power of spectroscopy and
thereby providing insights into their biological functions.
In this study, the hyperpolarizability tensors of the two helical
segments of melittin were calculated separately then subsequently
transformed into the laboratory fixed frame using the z–x–z convention as described previously [12]. The tilt and twist angles of
the first helical segment were simple chosen to be the same as
those of the peptide molecule. The set of tilt and twist angles
(h2, W2) of the second helical segment requires careful treatment
due to the shift in the molecular frame of the second segment
caused by the rotations of the first helical segment. The angles
(h2, W2) were then calculated to be:

2

À1

h2 ¼ h þ h12 þ sin

 p
for W 2 0;
2

3

cosh12
6
7
4sinh12 ð1 À cos WÞ Â qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi5
2
2
2
sin h12 cos W þ cos h12
ð6Þ
2

À1

h2 ¼ h À h12 þ sin
for W 2

p

2

;p



3

The asymmetric E mode of melittin can then be calculated as:

vE;zzz ¼

X

X

b ^z Á n
^ ÞibE1;lmn þ
hð^z Á^lÞð^z Á mÞð

0
c0 Þð^z Á nb0 ÞibE2;l0 m0 n0
hð^z Á b
l Þð^z Á m

l0 ;m0 ;n0

l;m;n

ð12Þ

vE;xxz ¼ vE;yyz ¼

X

X

b ^z Á n
^ ÞibE1;lmn þ
hð^x Á^lÞð^x Á mÞð

l;m;n

l

0

0
c0 Þð^z Á nb0 ÞibE2;l0 m0 n0
hð^x Á b
l Þð^x Á m

;m0 ;n0

ð13Þ

vE;xzx ¼ vE;zxx ¼

X

b ^x Á n
^ÞibE1;lmn
hð^x Á ^lÞð^z Á mÞð

l;m;n

X

þ
l

0

0
c0 Þð^x Á nb0 ÞibE2;l0 m0 n0
hð^x Á b
l Þð^z Á m

ð14Þ

;m0 ;n0

As discussed previously in Eqs. (2)–(4), the zzz, xzx and xxz
hyperpolarizability tensors are then calculated as:

vzzz ¼

X

b ^z Á n

^ ÞibA1;lmn
hð^z Á ^lÞð^z Á mÞð

l;m;n

X

þ

0
c0 Þð^z Á nb0 ÞibA2;l0 m0 n0
l Þð^z Á m
hð^z Á b

l0 ;m0 ;n0

þ

X

b ^z Á n
^ ÞibE1;lmn
hð^z Á ^lÞð^z Á mÞð

l;m;n

X

þ
l

0

vxxz ¼ vyyz ¼
X

þ

cosh12
6
7
4sinh12 ð1 À cos WÞ Â qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi5
2
sin h12 cos2 W þ cos2 h12

þ

ð7Þ

þ

0
c0 Þhð^z Á nb0 ÞibE2;l0 m0 n0
l Þhð^z Á m
hð^z Á b

ð15Þ

;m0 ;n0

X

b ^z Á n
^ ÞibA1;lmn
hð^x Á ^lÞð^x Á mÞð

l;m;n
0
c0 Þð^z Á nb0 ÞibA2;l0 m0 n0
hð^x Á b
l Þð^x Á m

0

l ;m0 ;n0

X

b ^z Á n
^ ÞibE1;lmn
hð^x Á ^lÞð^x Á mÞð

l;m;n

X

0
c0 Þð^z Á nb0 ÞibE2;l0 m0 n0
l Þð^x Á m
hð^x Á b

ð16Þ

l0 ;m0 ;n0

W2 ¼ W12 þ W

ð8Þ

where h12 is the inter-helical tilt angle between the two segments; h
and W are the molecular tilt and twist angles, respectively.
The transformed SFG susceptibilities of the segments were then
combined together, yielding the molecular SFG susceptibility. Due
to the deviation of the molecular symmetry property, the molecular twist angle is no longer azimuthally symmetric. Consequently,
the molecular SFG susceptibility quantity is a function of both
the molecular tilt and twist angles.
The non-zero hyperpolarizability tensors of the symmetric A
mode of melittin can then be calculated from the segmental
tensors:

vA;zzz ¼

X

b ^z Á n
^ ÞibA1;lmn þ
hð^z Á^lÞð^z Á mÞð

l;m;n

X

0
c0 Þð^z Á nb0 ÞibA2;l0 m0 n0
l Þhð^z Á m
ð^z Á b

l0 ;m0 ;n0

ð9Þ

vA;xxz ¼ vA;yyz ¼
X

þ
l

0

b ^z Á n
^ ÞibA1;lmn
hð^x Á ^lÞð^x Á mÞð

l;m;n

c0 Þð^z Á nb0 ÞibA2;l0 m0 n0
ð^x Á bl 0 Þhð^x Á m

ð10Þ

;m0 ;n0

vA;xzx ¼ vA;zxx ¼
þ

X

X
l0 ;m0 ;n0

X

b ^x Á n
^ ÞibA1;lmn
hð^x Á ^lÞð^z Á mÞð

l;m;n

c0 Þð^x Á nb0 ÞibA2;l0 m0 n0
ð^x Á bl 0 Þð^z Á m

ð11Þ

vxzx ¼ vzxx ¼
X

þ
l

þ

0

X

b ^x Á n
^ ÞibA1;lmn
hð^x Á ^lÞð^z Á mÞð

l;m;n
0
c0 Þð^x Á nb0 ÞibA2;l0 m0 n0
l Þð^z Á m
hð^x Á b

;m0 ;n0

X

b ^x Á n
^ ÞibE1;lmn
hð^x Á ^lÞð^z Á mÞð

l;m;n

þ

X

0

c0 Þð^x Á nb0 ÞibE2;l0 m0 n0
l Þð^z Á m
hð^x Á b

ð17Þ

0

l ;m0 ;n0

3. Results and discussions
3.1. Melittin – a revisited problem
Melittin is an amphipathic peptide consisting of 26 amino acid
residues (GIGAVLKVLTTGLPALISWIKRKRQQ) divided into three
parts, two of which adopt helical secondary structures (amino acid
residues positioned from 1 to11 and 12 to 21) and a non-helical
segment of five amino acid residue long (22–26) (Fig. 2).
Even though the interaction between melittin and lipid bilayers
has been studied extensively by various techniques such as ATRFTIR, Raman, fluorescence, circular and linear dichroism, dielectric
relaxation, NMR and SFG, results are inconclusive and
contradictory. This apparent variation in results may be due to
the molecules bent structure arising from the Pro residue at
position 14 in the amino acid sequence. Some studies suggested
a horizontal orientation of melittin in the lipid bilayers whilst


18

K.T. Nguyen / Chemical Physics 447 (2015) 15–21

Fig. 2. The structure of melittin. Tertiary structure of melittin [30].

some proposed that melittin forms channels by orienting vertically
in the membrane [23,31–38]. In fact, research about the mode of
action of melittin was fairly fruitful in the years of 2012 and
2013. A series of very recent papers employing X-ray diffraction,
oriented
circular
dichroism,
electrochemical
impedance
spectroscopy as well as various fluorescence spectroscopic and
computational simulation techniques to investigate the mechanism of melittin membrane permeability converged to a general
idea that melittin does not form a typical toroidal pores when disrupting the lipid bilayers as previously believed, but rather via a
transient process in which the peptide does not adopt the transmembrane orientation at equilibrium [23–26,39,40]. Being an
intrinsic surface sensitive technique, SFG was also used by the
Chen group in studying the interaction between melittin and

DPPG/dDPPG lipid bilayer. They proposed that melittin adopts both
vertical and horizontal orientations in DPPG/dDPPG lipid bilayer
[10]. However, by introducing some small modifications into the
data analysis, a description of this peptide-membrane interaction
can be obtained which better aligns with findings reported in the
current literature.
It has been shown by molecular simulation (using CHARMM
force field) that the melittin molecular conformation remains rigid
during the interaction with the lipid bilayer [41]. Using this result,
the molecular SFG hyperpolarizability tensors of melittin can be
calculated. The structure of melittin under different conditions
has been studied using a number of techniques. In particular, the

inter-helical tilt angle h12 was determined to be around
2.48 ± 0.05 (rads) in DTPC lipid [42], 2.25 rads in crystal form
[43], or 2.2 ± 0.3 (rads) for DPC bound melittin [44] and
2.3 ± 0.16 (rads) for melittin lyophilized from methanol [45]. This
inter-helical tilt value was found to be significantly lower to the
value of 1.5 ± 0.6 (rads) for PC bound melittin, based on transferred
NOE [46]. Although being heavily influenced by the surrounding
conditions, the inter-helical angel of melittin was found not to differ greatly among peptide crystals prepared from aqueous solution,
melittin lyophilized from methanol and DPC bound melittin. This
inter-helical angle, however, changes significantly when melittin
binds to DTPC in hydrated gel state and dry form [42]. Therefore,
in studies of interactions between fully hydrated melittin and
model cell membranes, the inter-helical angle has been widely
assumed to be close to its value in the crystal form, which is 2.25
rads.
The detailed expressions of the above modes are relatively
lengthy and therefore are not reported here. Instead, their
relationships with the molecular tilt and twist angles are presented
visually in Fig. 3 (plotted using Eqs (15) and (16)). The calculation
resulted in a possible range of the ratio vzzz/vyyz of [1.4, 2.8], which
is broader than the range calculated previously for melittin [10].
This may be due to the omission of the slight kink at Pro14 that
Chen and colleagues assumed. This new broader range suggests
the possibility that melittin can actually adopt a single orientation
distribution in a 1,2-dipalmitoyl-sn-glycero-3-phosphoglycerol
(DPPG) lipid bilayer.
In Chen et al.’s study, a ratio vzzz/vyyz of 1.4 was experimentally
obtained and it was concluded that melittin adopts two distinct
orientations upon its interaction with the DPPG bilayer at a concentration of 780 nM [10]. They employed a fairly sophisticated
and elegant analysis involving the maximum entropy approach

as well as the assumption of double delta orientation distributions.
However, in their analysis, melittin structure was assumed to be
linear so that its symmetry properties guaranteed the angles W
and u to be azimuthal. This assumption will possibly be valid if
the second helical segment is substantially shorter than the latter
or the inter-helical tilt angle is negligible (close to 0o) since the
molecular twist angle W, directly affects the tilt angle of the second
helical segment which in turn alters the projections of the symmetric and asymmetric vibrational modes in the laboratory fixed
frame.
Based on the analysis presented here, an experimental ratio
vzzz/vyyz of around 1.4 implies a possible range of [0, 0.40 (±0.02)
(rads)] and [1.40 (±0.07), 2.00 (±0.10) (rads)] for the molecular tilt
and twist angles respectively (Fig. 3(c)). It is worth mentioning that
the region when the ratio vzzz/vyyz falls in [1.4,1.5] at h e [1,1.4] is
due to a mathematical artifact in Fig. 3(c) and hence will not be
considered. To assess the effects of broader orientation distributions of the tilt and twist angles on the ratio vzzz/vyyz, one can
assume that these angles adopt a more realistic normal distribution with a fixed standard deviation. Ideally, both the tilt and twist
angles can be assumed as normally distributed and the vzzz/vyyz
ratio can be calculated and plotted accordingly. Unfortunately,


19

K.T. Nguyen / Chemical Physics 447 (2015) 15–21

ð2Þ

ð2Þ

ð2Þ

ð2Þ

(a)

(b)

(c)

(d)

ð2Þ

Fig. 3. (a)–(c): vzzz ; vyyz and vzzz vzzz =vyyz as a function of the molecular titl (h,rad) and twist (W,rad)) angles. In (c), the plot was truncated to the
[1.4,1.5]. (d): Definitions of the molecular tilt (h) and twist (W) angles.

the attempt to calculate the expected value of the hyperpolarizability tensor elements in the whole continuous range of both
the tilt and twist angles failed because it was rather computational
intensive for a personal computer to handle. To simplify this process, either the tilt or twist angle was be held fixed while the other
was assumed to adopt a normal distribution. In particular, the
twist was first fixed at 1.70 rads while the tilt angle was assumed
a normal distribution of mean 0.20 rads and standard deviation of
0.26 rads; a vzzz/vyyz ratio of 1.45 was obtained. Then, the tilt angle
was set to 0.20 rads while the twist angle was assumed a normal
distribution with mean and standard deviation of respectively
1.70 and 0.26 rads; a vzzz/vyyz ratio of 1.43 was accordingly
obtained. As a reference, the ratio vzzz/vyyz was also calculated to
be 1.41 as both the tilt and twist angles were assumed to follow
the d-distribution at 0.20 and 1.70 rads, respectively. These three
values of the vzzz/vyyz are sufficiently close, which validates the

simplification of using the d-distribution for both the tilt and twist
angles in the analysis. As discussed extensively in earlier studies
[27], SFG ssp and ppp signals are only able to provide one piece
of information of structural significance. For this bent helical structure, the orientation description includes both the tilt and twist
angles. Therefore, a second technique which possesses comparable
experimental setup is, in principle, required. However, given the
narrow orientational range obtained above, one is able to conclude
that one of the two helical segments of melittin adopts a vertical
orientation in relative to the membrane surface. This tilt and twist

ð2Þ
ð2Þ
vð2Þ
zzz vzzz =vyyz ratio range of

angle combination supports the peptide penetration model in
which one helical segment orients almost vertically and the other
orients almost horizontally that leads to the coexisting vertical and
horizontal peptide orientation predicted by Chen et al. [10]. The
interpretation of this study is consistent with a recent model of
melittin in lipid bilayers proposed by Postupalenko et al. using
dual-fluorescence spectroscopy. In their study, L9 residue was
found to protrude into the bilayer while the W19 residue was
reported to localize at the interface [26]. Moreover, the interaction
picture presented here is surprising in line with the classical model
Vogel proposed using Raman spectroscopy by the fact that one of
the two helical segments of melittin adopts a vertical and the latter
adopts a horizontal orientation in the lipid membranes [30].
Interestingly, in a recent study using X-ray diffraction and oriented
circular dichroism, Huang et al. recently suggested that when the

pore induction occurs, the ratio between vertical/horizontal helical
content is 0.56 which falls in the range of [0.55,0.68] achieved by
projection calculations [23]. Although the proposed molecular orientation proposed here seems to align well with Huang’s and
Vogel’s, there is a problem relating these pieces of orientation
information into the toroidal pore disruption mode of melittin
because the length of its transmembrane oriented segment is
definitely not sufficient to span the whole lipid bilayer. More
specifically, the thickness of the hydrophobic core of a typical POPC
lipid bilayer is around 4 nm [39], while the transmembrane
segment is only able to penetrate at most 1.5 nm into the bilayer.


20

K.T. Nguyen / Chemical Physics 447 (2015) 15–21

posing lipids [22,34]. The analysis of this study on DPPG bound
melittin explains the dual orientation distribution as well as the
transient pore formation proposed by a number of recent studies
[10,23–26,30,39]. On a broader scale, this study does not only provides new orientational information on DPPG lipid bilayer bound
melittin, but also opens up opportunities for the implementation
of this high throughput technique in structural biology. This report
serves as a foundation for SFG studies on biological systems under
physiological conditions.
Fig. 4. Representative average molecular orientation of melittin in lipid bilayers.

Conflict of interest
Therefore, the pore formation is unlikely to occur; the orientation
of melittin in DPPG lipid bilayers appears as presented in Fig. 4.
A study on the effect of melittin on DPPG/dDPPG lipid bilayer

previously carried out by Chen et al. supports the interpretation
of this study on melittin orientation in DPPG/dDPPG bilayers
[47]. Using isotope labeling, Chen et al. were able to differentiate
the proximal and distal leaflets and thus observe the behaviors of
each leaflet when interacting with melittin at the peptide concentration of 780 nM. They found that the integrity of the proximal
leaflet was maintained during the course of the interaction while
the integrity of the distal leaflet was disturbed significantly, which
agrees well with the presented findings, that melittin cannot span
the whole lipid bilayer. Chen et al. observed that the original spectral features of the proximal leaflet in the CD region 2000–
2300 cmÀ1 remained unchanged upon peptide interaction, while
that of the distal leaflet in the CH region 2800–3100 cmÀ1 completely diminished. Although a decrease of the overall SFG signal
in the CD region was observed, this could only be due to the
increase in the symmetry property of the bilayer caused by the
flip-flopping of the lipid molecules. This observation fully supports
interpretation of the present study (Fig. 4) in which the horizontal
helical segment destroys the integrity of the only one phospholipid
layer yet leaves the other intact. It is worth noting that using
vesicle leakage experiment, Wiedman et al. observed a two phase
interaction between melittin and POPC lipid bilayers. There was
observed a rapid burst of leakage occurred in the first phase followed by a second phase during which leakage slowed down and
then stopped before completion (at relevant peptide:lipid ratios)
[25]. Wiedman et al.’s observation is an excellent indication of
the self-healing ability of POPC lipid bilayer which again does not
support the toroidal pore mode of action of melittin.
At this stage, one may wonder if it is not the toroidal pore mode
of action melittin follows, then how melittin lyses the cells. There
have been both experimental and computational evidence suggesting that melittin creates transient pores in lipid bilayers [24,25].
Even though the proposed orientation information does not
directly suggest a transient behavior, it actually supports this
model by providing the peptide orientation information at equilibrium which implies that the translocation of melittin is required

for the cell lysis to occur.

4. Conclusions
Carrying information on both chemical and orientational
properties of interfacial species, SFG vibrational spectroscopy possesses a rather complicated data analysis. Despite rather lengthy
transformations of the hyperpolarizability tensors, SFG vibrational
spectroscopy was employed to determine a multi a-helical
structure without any of the qualitative simplifications that have
previously been assumed [10,11]. As pointed out in earlier studies,
the complexity of the interaction scheme that melittin adopts,
might vary with factors such as membrane composition, hydration
levels, temperature, and the phase states of the membrane com-

There is no conflict of interest.
Acknowledgements
This research is funded by Vietnam National Foundation for
Science and Technology Development (NAFOSTED) under Grant
number 106.16-2012.67. I sincerely thank Dr Gay Marsden for
her generous assistance with the manuscript preparation and the
proofreading.
References
[1] X.Y. Chen, J. Wang, J.J. Sniadecki, M.A. Even, Z. Chen, Langmuir 21 (2005) 2662.
[2] A.P. Boughton, P. Yang, V.M. Tesmer, B. Ding, J.J.G. Tesmer, Z. Chen, Proc. Natl.
Acad. Sci. USA 108 (2011) E667.
[3] K.T. Nguyen, S.V. Le Clair, S.J. Ye, Z. Chen, J. Phys. Chem. B 113 (2009) 12358.
[4] P. Yang, A. Ramamoorthy, Z. Chen, Langmuir 27 (2011) 7760.
[5] S.J. Ye, K.T. Nguyen, Z. Chen, J. Phys. Chem. B 114 (2010) 3334.
[6] L. Fu, G. Ma, E.C.Y. Yan, J. Am. Chem. Soc. 132 (2010) 5405.
[7] T. Weidner, N.F. Breen, K. Li, G.P. Drobny, D.G. Castner, Proc. Natl. Acad. Sci.
USA 107 (2010) 13288.

[8] L. Fu, D.Q. Xiao, Z.G. Wang, V.S. Batista, E.C.Y. Yan, J. Am. Chem. Soc. 135 (2013)
3592.
[9] D.Q. Xiao, L. Fu, J. Liu, V.S. Batista, E.C.Y. Yan, J. Mol. Biol. 421 (2012) 537.
[10] X.Y. Chen, J. Wang, A.P. Boughton, C.B. Kristalyn, Z. Chen, J. Am. Chem. Soc. 129
(2007) 1420.
[11] K.T. Nguyen, R. Soong, S.C. Im, L. Waskell, A. Ramamoorthy, Z. Chen, J. Am.
Chem. Soc. 132 (2010) 15112.
[12] K.T. Nguyen, S.V. Le Clair, S.J. Ye, Z. Chen, J. Phys. Chem. B 113 (2009) 12169.
[13] Y.W. Liu, T.L. Ogorzalek, P. Yang, M.M. Schroeder, E.N.G. Marsh, Z. Chen, J. Am.
Chem. Soc. 135 (2013) 12660.
[14] P. Yang, A. Boughton, K.T. Homan, J.J.G. Tesmer, Z. Chen, J. Am. Chem. Soc. 135
(2013) 5044.
[15] B. Ding, J.E. Laser, Y.W. Liu, P.R. Wang, M.T. Zanni, Z. Chen, J. Phys. Chem. B 117
(2013) 14625.
[16] L. Fu, J. Liu, E.C.Y. Yan, J. Am. Chem. Soc. 133 (2011) 8094.
[17] K.T. Nguyen, J.T. King, Z. Chen, J. Phys. Chem. B 114 (2010) 8291.
[18] C. Hirose, N. Akamatsu, K. Domen, J. Chem. Phys. 96 (1992) 997.
[19] B. Ding, L. Soblosky, K. Nguyen, J.Q. Geng, X.L. Yu, A. Ramamoorthy, Z. Chen, Sci.
Rep.-Uk 3 (2013).
[20] G. Schwarz, R.T. Zong, T. Popescu, Biochim. Biophys. Acta 1110 (1992) 97.
[21] K. Matsuzaki, K. Sugishita, N. Ishibe, M. Ueha, S. Nakata, K. Miyajima, R.M.
Epand, Biochemistry-Us 37 (1998) 11856.
[22] H. Vogel, Biochemistry-Us 26 (1987) 4562.
[23] M.T. Lee, T.L. Sun, W.C. Hung, H.W. Huang, Proc. Natl. Acad. Sci. USA 110 (2013)
14243.
[24] K.P. Santo, S.J. Irudayam, M.L. Berkowitz, J. Phys. Chem. B 117 (2013) 5031.
[25] G. Wiedman, K. Herman, P. Searson, W.C. Wimley, K. Hristova, BbaBiomembranes 1828 (2013) 1357.
[26] V.Y. Postupalenko, O.M. Zamotaiev, V.V. Shvadchak, A.V. Strizhak, V.G.
Pivovarenko, A.S. Klymchenko, Y. Mely, Bioconjugate Chem. 24 (2013) 1998.
[27] J. Wang, S.H. Lee, Z. Chen, J. Phys. Chem. B 112 (2008) 2281.

[28] A.J. Moad, G.J. Simpson, J. Phys. Chem. B 108 (2004) 3548.
[29] G.J. Simpson, J.M. Perry, C.L. Ashmore-Good, Phys. Rev. B 66 (2002).
[30] H. Vogel, F. Jahnig, Biophys. J. 50 (1986) 573.
[31] K. Hristova, C.E. Dempsey, S.H. White, Biophys. J. 80 (2001) 801.
[32] D. Allende, S.A. Simon, T.J. McIntosh, Biophys. J. 88 (2005) 1828.
[33] I. Constantinescu, M. Lafleur, Bba-Biomembranes 1667 (2004) 26.
[34] S. Frey, L.K. Tamm, Biophys. J. 60 (1991) 922.
[35] A.S. Ladokhin, S.H. White, Bba-Biomembranes 1514 (2001) 253.
[36] C. Steinem, H.J. Galla, A. Janshoff, PCCP 2 (2000) 4580.
[37] S. Toraya, K. Nishimura, A. Naito, Biophys. J. 87 (2004) 3323.
[38] L. Yang, T.A. Harroun, T.M. Weiss, L. Ding, H.W. Huang, Biophys. J. 81 (2001)
1475.
[39] S.J. Irudayam, M.L. Berkowitz, Bba-Biomembranes 1818 (2012) 2975.
[40] S.J. Irudayam, T. Pobandt, M.L. Berkowitz, J. Phys. Chem. B 117 (2013) 13457.


K.T. Nguyen / Chemical Physics 447 (2015) 15–21
[41] M. Andersson, J.P. Ulmschneider, M.B. Ulmschneider, S.H. White, Biophys. J.
104 (2013) L12.
[42] Y.H. Lam, S.R. Wassall, C.J. Morton, R. Smith, F. Separovic, Biophys. J. 81 (2001)
2752.
[43] T.C. Terwilliger, D. Eisenberg, J. Biol. Chem. 257 (1982) 6016.
[44] F. Inagaki, I. Shimada, K. Kawaguchi, M. Hirano, I. Terasawa, T. Ikura, N. Go,
Biochemistry-Us 28 (1989) 5985.

21

[45] R. Bazzo, M.J. Tappin, A. Pastore, T.S. Harvey, J.A. Carver, I.D. Campbell, Eur. J.
Biochem. 173 (1988) 139.
[46] A. Okada, K. Wakamatsu, T. Miyazawa, T. Higashijima, Biochemistry-Us 33

(1994) 9438.
[47] X.Y. Chen, J. Wang, C.B. Kristalyn, Z. Chen, Biophys. J. 93 (2007) 866.