Biochimica Et Biophysica Acta: Kamil Wojciechowski, Marta Orczyk, Thomas Gutberlet, Thomas Geue

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 11

Biochimica et Biophysica Acta 1858 (2016) 363–373

Contents lists available at ScienceDirect

Biochimica et Biophysica Acta

journal homepage: www.elsevier.com/locate/bbamem

Complexation of phospholipids and cholesterol by triterpenic saponins in


bulk and in monolayers
Kamil Wojciechowski a,⁎, Marta Orczyk a, Thomas Gutberlet b, Thomas Geue c
a
Faculty of Chemistry, Warsaw University of Technology, Noakowskiego 3, 00-664 Warsaw, Poland
b
Helmholtz-Zentrum Berlin für Materialien und Energie GmbH, Hahn-Meitner-Platz 1, 14109 Berlin, Germany
c
Laboratory for Neutron Scattering and Imaging, Paul Scherrer Institute, WHGA/110, 5232 Villigen — PSI, Switzerland

a r t i c l e i n f o a b s t r a c t

Article history: The interactions between three triterpene saponins: α-hederin, hederacoside C and ammonium glycyrrhizate
Received 16 October 2015 with model lipids: cholesterol and dipalmitoylphosphatidylcholine (DPPC) are described. The oleanolic acid-
Accepted 1 December 2015 type saponins (α-hederin and hederacoside C) were shown to form 1:1 complexes with lipids in bulk, character-
Available online 2 December 2015
ized by stability constants in the range (4.0 ± 0.2)·103–(5.0 ± 0.4)·104 M−1. The complexes with cholesterol are
generally stronger than those with DPPC. On the contrary, ammonium glycyrrhizate does not form complexes
Keywords:
Biosurfactant
with any of the lipids in solution. The saponin–lipid interactions were also studied in a confined environment
Saponin of Langmuir monolayers of DPPC and DPPC/cholesterol with the saponins present in the subphase. A combined
α-Hederin monolayer relaxation, surface dilational rheology, fluorescence microscopy and neutron reflectivity (NR) study
Hederacoside C showed that all three saponins are able to penetrate pure DPPC and mixed DPPC/cholesterol monolayers. Overall,
Ammonium glycyrrhizate the effect of the saponins on the model lipid monolayers does not fully correlate with the lipid–saponin complex
Surface pressure formation in the homogeneous solution. The best correlation was found for α-hederin, for which even the pref-
Dilatational surface rheology erence for cholesterol over DPPC observed in bulk is well reflected in the monolayer studies and the literature
data on its membranolytic activity. Similarly, the lack of interaction of ammonium glycyrrhizate with both lipids
is evident equally in bulk and monolayer experiments, as well as in its weak membranolytic activity. The com-
bined bulk and monolayer results are discussed in view of the role of confinement in modulating the saponin–
lipid interactions and possible mechanism of membranolytic activity of saponins.
© 2015 Elsevier B.V. All rights reserved.

1. Introduction microscopy observations of saponin-treated biological cell membranes


[12,13], were however more consistent with the hypothesis of penetra-
Saponins are glycosidic biosurfactants produced by numerous or- tion of saponins into the lipid layers and subsequent disturbance of
ganisms, mainly plants [1,2]. Despite a great variety of chemical struc- cholesterol's interactions with other lipids (phospho- and
tures, the common feature of all saponins is the presence of at least sphingolipids) [14]. On the other hand, some reports suggest that cho-
one glycoside bond linking an aglycone part (steroid or triterpenoid) lesterol is not crucial for the saponins' membranolytic activity [15] or
with a glycone (mono- or oligosugar) part. The biological role of sapo- that their membrane activity is related to interactions with specific re-
nins is still largely unknown. However, the fact that they are stored in ceptors [16].
parts of a plant most susceptible to attack by predators and are often Despite some claims [17], usually no clear correlation can be found
produced in response to threats [3,4] suggests that they play active between the surface activity of saponins and their membranolytic activity.
role in plant's self-defense [1,3]. Many saponins display membranolytic Moreover, the ability to lower surface tension (hence the detergent activ-
activities towards bacteria, viruses and fungi, and some are even capable ity) of saponins is rather weak at biologically-relevant concentrations. For
of lysing the red blood cell (erythrocyte) membranes [5]. The mem- example, typical concentrations for 50% cytotoxic activity, IC50, for most
brane activity of both triterpenoid and steroid saponins has been tradi- hemolytic saponins are in the range of 10 μM [18]. Our recent studies
tionally associated with their ability to interact with membrane-bound with Quillaja bark saponins (QBS) clearly show that interaction of sapo-
cholesterol [6]. Following the observations of Ransom [7] and Windaus nins with lipid membranes is much more complex than one would expect
[8], saponins were initially believed to simply remove cholesterol from from a simple surfactant activity [19–21].
biological membranes [9], leaving the cytosol-permeable pores. The To the best of our knowledge no successful attempts have been
works of Schulman and Rideal [10,11], as well as subsequent electron described so far to systematically compare the strength of interac-
tions between different saponins and membrane lipids, especially
⁎ Corresponding author. phospholipids and cholesterol. Some preliminary information can
E-mail address: kamil.wojciechowski@ch.pw.edu.pl (K. Wojciechowski). be deduced from molecular dynamics (MD) investigation which

http://dx.doi.org/10.1016/j.bbamem.2015.12.001
0005-2736/© 2015 Elsevier B.V. All rights reserved.
364 K. Wojciechowski et al. / Biochimica et Biophysica Acta 1858 (2016) 363–373

showed that the binding free energy for dioscin–cholesterol in dec- saponin complex stability and the extent of lipid monolayer penetration
ane is significantly more favorable (− 19.10 kcal/mol) than for cho- by the given saponin.
lesterol–cholesterol complexes (− 0.77 kcal/mol) [14].
Three triterpenoid saponins were selected for this study (Fig. 1): α- 2. Experimental
hederin, hederacoside C and glycyrrhizic acid (ammonium salt). The
first two originate from the same plant (Hedera helix) and share the For all experiments, α-hederin (Cat. No. 9970.1), hederacoside C
same aglycone (hederagenin, belonging to the oleanolic acid-type (Cat. No. 6775.1) and ammonium glycyrrhizate (ammonium salt of
triterpenes). They differ in the extent of sugar substitution: α-hederin glycyrrhizic acid, Cat. No. 6247.1) were purchased from Carl Roth
is a monodesmoside with one disaccharide (α-L-rhamnose-(1 → 2)- and were used as received. Chloroform (CHROMASOLV for HPLC),
α-L-arabinose) group attached at the C3 position of the aglycone, ethanol (pure p.a.) and dimethylsulphoxide (pure p.a.) were pur-
while hederacoside C is a bidesmoside, additionally substituted at C28 chased from Sigma-Aldrich and POCh. The 1,2-dipalmitoyl-sn-
with a trisaccharide group. Both saponins are active components of glycero-3-phosphocholine (DPPC) with purity of 99% and cholester-
cough syrups, where they act as mucolytic and secretolytic agents. ol with purity of 99.5% were purchased from Sigma-Aldrich. Milli-Q
The aglycone part of glycyrrhizic acid (from Glycyrrhiza glabra L) is of water (Millipore) was used to prepare all aqueous solutions. All
glycyrrhetinic acid type and is substituted only at C3 position glassware for Langmuir trough experiments was cleaned with ace-
(monodesmoside) with a disaccharide moiety. Because of its low toxic- tone and Hellmanex II solution (Hellma Worldwide) and subse-
ity, it finds many applications in healthcare and food industries, e.g., as a quently rinsed with copious amounts of Millipore water.
low calorie sweetener [22].
The three saponins differ significantly in hemolytic activity. α- 2.1. Surface pressure
Hederin is one of the most hemolytic triterpenoid saponins, with HD50
and HD100 (referring to the concentrations causing 50% and 100% hemo- All measurements were carried out with a KSV NIMA small Lang-
lysis, respectively) of 11.3 and 28.8 μM. For comparison, hederacoside C muir–Blodgett trough (77.5 cm2) equipped with two movable barriers.
shows HD50 N 100 μM [23]. Glycyrrhizic acid shows no hemolytic activ- Each measurement was performed at least in duplicate; in case of any
ity at least up to 48 μM (the highest concentration tested in [24]). In significant discrepancies it was repeated further. The surface pressure
membrane toxicity study by Böttger et al. [17] using human urinary was measured using a Wilhelmy plate made of filter paper (Whatman
bladder epithelial carcinoma cells (ECV-304), the monodesmoside Chr1) connected to an electro balance. The speed of barrier movement
α-hederin was found by far the most cytotoxic (IC50 = 29 μM). Its was set to 10 mm/min, as suggested by the manufacturer. For the sur-
bidesmoside analog, hederacoside C, was much less toxic to the same face pressure vs. time measurements, the trough was filled with an ap-
cells (IC50 N 238 μM). Glycyrrhizic acid was not toxic either, at least up propriate saponin solution (10−4 M) in Milli-Q water (with addition of
to the maximum studied concentration (IC50 N 164 μM). The membrane 4% DMSO in the case of α-hederin for poor solubility reasons) and the
toxicity of the three saponins in ECV-304 cell is closely related to their surface pressure was recorded for 1 h. Temperature was controlled at
effect on membrane cholesterol: glycyrrhizic acid and hederacoside C 21 °C by water circulation from a thermostat. The Wilhelmy balance
did not affect its content in the cells, while α-hederin significantly was zeroed on pure Milli-Q water just before pouring the saponin
reduced it [9]. These observations suggest that both the glycone and solution.
aglycone parts of a saponin molecule play important roles in its
membranolytic activity. 2.2. Subphase exchange
The limited knowledge about the mechanism of saponin–lipid inter-
actions prompted us to investigate in more detail the stoichiometry and The subphase exchange setup consisted of a Gilson's MINPULS 3
stability of the three triterpenoid saponin–lipid complexes using choles- peristaltic pump and Teflon tubings. The in and out tubings were im-
terol and 1,2-dipalmitoylphosphatidylcholine (DPPC) as model mem- mersed in the subphase on opposite sides of the trough to introduce the
brane lipids. The common aglycone moiety was chosen in order to concentrated saponin solution and to recirculate the subphase, in order
allow for general conclusions concerning the role of nature of glycones to achieve the final saponin concentration of 10−4 M. Hederacoside C
in interactions between saponins and lipids. To the best of our knowl- and ammonium glycyrrhizate were dissolved in Milli-Q water, and for
edge chemical interactions have not been considered so far as an impor- α-hederin 4% of DMSO was added to improve its solubility. Accordingly,
tant contribution to the membranolytic activity of saponins. In this the subphase consisted of either pure Milli-Q water or water with
contribution we seek for a possible correlation between the lipid– addition of DMSO (4%). The lipid (DPPC and DPPC/cholesterol 10:9)

Fig. 1. Chemical structures of saponins (upper row, from left to right): α-hederin, hederacoside C and ammonium glycyrrhizate, as well as lipids (bottom row, from left to right): choles-
terol and 1,2-dipalmitoylphosphatidylcholine (DPPC).
K. Wojciechowski et al. / Biochimica et Biophysica Acta 1858 (2016) 363–373 365

monolayers were spread first on pure water as chloroform solutions and trough. The scattered neutrons were recorded with a 3He-single detec-
after the solvent evaporated, the monolayers were compressed to the ini- tor tube in time-of-flight mode requiring typically 8 h of beam time.
tial surface pressure of 32.5 mN/m. Then the concentrated saponin solu- The experimentally obtained reflectivity curves were analyzed by
tion was introduced at the flow rate of 2 ml/min. The flow and the applying the standard fitting routine, using the Parratt 32 software
barrier position were maintained for 1 h and the changes in surface pres- [26]. It determines the optical reflectivity of neutrons from planar sur-
sure, Π(t), induced by the presence of saponins were monitored. faces using a calculation based on Parratt's recursion scheme for strati-
fied media [27]. The reflecting interface is modeled as consisting of
2.3. Dilatational rheology layers of specific thickness, scattering length density and roughness,
which are the fitting parameters. The neutron scattering length density
Mechanical properties of the Langmuir monolayers and adsorbed (SLD) is defined by the sum of the bound coherent scattering length bc of
layers were probed using the Langmuir trough. At the end of each Π the reflecting material normalized by the volume, v, as SLD = ∑bc/v.
vs. time measurement harmonic oscillations of the area per molecule The model reflectivity profile is calculated and compared to the mea-
were applied by moving both barrier positions around the initial loca- sured data, then the model is recursively adjusted by a change in the
tion with the frequency of 0.1 Hz and amplitude of 2%, which is within fitting parameters to best fit the data.
the viscoelastic linear regime for the system under study. Each measure-
ment consisted of several harmonic compression/decompression cycles 3. Results
for 300 s. The two moduli were obtained from the Fourier transform of
harmonic oscillations of the surface pressure in response to the applied 3.1. Stoichiometry and stability constants of binary complexes of triterpene
strain (area oscillations) using the KSV NIMA software. saponins with cholesterol and DPPC
The dilatational visco-elasticity modulus, | E(ω)|, is defined as
follows: The actual location of the saponin–lipid complex formation in real
biological systems is not a priori known. As described in the
dπ Introduction, the complexes could form within the membrane (pene-
jEðωÞj ¼ − A tration of a saponin), in the surrounding aqueous phase (solubilization
dA
of cholesterol) or at the interface between the polar aqueous phase sur-
where A is the molecular area at a given surface pressure π. rounding the lipid membrane and the non-polar lipid layer. Attempts
The visco-elasticity modulus E(ω) is a complex quantity and consists were first made to predict the most probable complex location by
of two frequency-dependent contributions: storage (E′) and loss (E″) using extraction procedures. Unfortunately, liquid–liquid extraction
moduli: using an aqueous solution of saponin and a tetradecane solution of
lipids could not be used for this purpose, because of technical problems
with separating the phases after extraction (emulsion formation).
EðωÞ ¼ E0 ðωÞ þ iE″ ðωÞ:
Instead, a solid–liquid extraction was employed: solid cholesterol
was extracted with an aqueous solution of QBS (10−3 M), and a solid
QBS — with a tetradecane solution of cholesterol (10−3 M). After 24 h
2.4. Fluorescence microscopy extraction with the respective solids, the liquid phases (aqueous or
tetradecane) were assessed by UV/Vis. No noticeable changes in their
In order to observe the topography of the monolayers exposed to the spectra could be found, suggesting that under these conditions the sa-
action of triterpenic saponins an OLYMPUS BX51WI inverted fluores- ponin–cholesterol complex is formed neither within the membrane,
cence microscope was used. The microphotographs were taken during nor in the surrounding aqueous phase. Therefore, to mimic an interme-
the subphase exchange for the appropriate saponin solution, as diate polarity of the membrane-aqueous interface, ethanol/water mix-
described for Π(t) measurements. As a fluorescent dye TopFluor PC (1- tures were used (95:5 for α-hederin and hederacoside C, and 90:10
palmitoyl-2-(dipyrrometheneboron difluoride)undecanoyl-sn-glycero- for ammonium glycyrrhizate).
3-phosphocholine) was used (1 mol% vs the lipid). The dye was of The complex stoichiometry was assessed by employing a continuous
N99% purity and was supplied by Avanti Lipids. Fluorescence micrographs variation method (“Job's method”). For this purpose the UV/Vis spectra
from the samples were collected using a MPLFN10X objective and the im- of saponin–lipid mixtures with continuously varying molar ratio but
ages were processed using cellSens software. constant total concentration were acquired, as described in the
Experimental part. A representative curve (so called “Job plot”) for α-
2.5. Neutron reflectivity hederin/cholesterol mixtures is shown as an inset in Fig. 2, with a char-
acteristic maximum at the molar ratio = 0.5, indicative of 1:1 complex
In a reflectivity measurement the beam intensity, R, reflected stoichiometry. Similar curves were obtained for α-hederin and
from a surface, normalized to the incoming intensity at a certain hederacoside C mixtures with both cholesterol and DPPC.
angle of incidence, θ is recorded. In case of a neutron reflectivity In order to determine the stability constants of the respective sapo-
(NR) experiment, R defined as above is measured as a function of nin–lipid complexes, UV/Vis titrations were performed with constant
scattering wave vector, q, expressed as the incidence angle θ normal- concentration of the lipid and increasing concentrations of saponins
ized to the corresponding wavelength, λ, as qz = 4π sin(θ)/λ. NR (see Fig. 2 for a representative set of spectra). For both hederagenin sa-
measurements were performed at the time-of-flight neutron ponins, clear inflection points around the saponin-to-lipid ratio of one
reflectometer AMOR (Paul Scherrer Institute (PSI), Villigen, were observed, consistent with the Job method results. The correspond-
Switzerland) [25]. The time-of-flight method allows the recording ing titration curves could be fitted using a 1:1 complexation model. The
of a wide range of q values in a single angular setting. global fitting procedure (simultaneously at typically two wavelengths)
The NR experiments at air/water interface were performed at three provided the stability constants of α-hederin and hederacoside com-
angles of incidence (0.25°, 0.65° and 1.4°) using a wavelength band plexes as collected in Table 1. On the other hand, for ammonium
from 3 to 12 Å, covering the necessary q range for the experiments of glycyrrhizate, the Job plots for both DPPC and cholesterol were almost
qzmin = 0.01 Å−1 to qzmax = 0.15 Å−1. The resolution was set by a slit flat, suggesting weak interaction of both lipids with this saponin in the
system on the incident side and the time-of-flight parameters to mixed ethanol:water solvent. This was further confirmed by UV/Vis ti-
Δqz = 0.006 Å− 1. A beam of rectangular cross-section 1 × 35 mm2 tration, where no inflection point could be observed on an absorbance
(for low angles) impinged on the samples at the air/water Langmuir vs saponin-to-lipid ratio.
366 K. Wojciechowski et al. / Biochimica et Biophysica Acta 1858 (2016) 363–373

Fig. 2. UV/Vis spectra for 2·10−4 M cholesterol solution titrated with 4·10−3 M α-hederin solution (the red arrow shows a direction of changes with increasing saponin concentration).
Inset shows a Job's plot for α-hederin/cholesterol system at λ = 212 nm. All spectra recorded in 95% ethanol.

3.2. Saponin–lipid interactions at the water–air interface high values of both moduli. This feature is typical for monodesmosidic
saponins capable of interacting by means of both hydrogen bonding
3.2.1. Surface pressure and hydrophobic interactions [28]. The values for ammonium
Spontaneous adsorption of the triterpene saponins on a bare water– glycyrrhizate are also high, but significantly lower than for α-hederin.
air interface (formation of Gibbs layers) was studied using a Langmuir In agreement with the surface pressure results, the bidesmosidic analog
trough. The ability to lower surface tension of aqueous solutions of a-hederin (hederacoside C) shows negligible values of both moduli.
(i.e., to increase surface pressure) for pure saponins at the same concen- In the next step, the interaction of the triterpene saponins with
tration (c = 10−4 M) is compared in Fig. 3. The monodesmosidic α- DPPC and cholesterol at the model membrane–aqueous interface
hederin increases surface pressure by nearly 20 mN/m more than its was studied using Langmuir monolayers. Pure DPPC and DPPC/cho-
bidesmosidic analog, hederacoside C. This suggests that the presence lesterol (10:9 mol/mol) mixed monolayers were first spread on
of the second glycone group located at the opposite end of the Milli-Q water and compressed to 32.5 mN/m. Next, the subphase
triterpene backbone with respect to C-3 (at C-28) disturbs the was exchanged with the appropriate triterpene saponin solution
lipophilic-hydrophilic balance of the molecule and renders it more to give the final concentration of 10 − 4 M, as described in the
water-soluble, but also less amphiphilic. Neutralization of the negative Experimental part. The surface pressure vs time curves are shown
ionic charge of a free carboxylic group (in α-hederin) by its esterifica- in Fig. 3. Note that the time evolution of surface pressure starts at
tion (in hederacoside C) might also play a role in lowering the surface the same time as the subphase exchange and that the complete sub-
activity. In another monodesmosidic saponin used in this study, ammo- phase exchange in the employed setup takes about 15 min [19]. For
nium glycyrrhizate, the presence of a ketone group and the location of DPPC monolayers the effect of the saponins follows the same order
the free carboxylic acid group (C-30 vs C-28 in α-hederin) apparently as for the respective Gibbs layers on bare water/air interface. This
disfavor adsorption at the water–air interface. Overall, the trend of suggests that it is not the interaction of saponins with DPPC but rath-
lowering ability to increase the surface pressure (decrease surface ten- er their intrinsic amphiphilicity that drives their penetration into
sion) follows the order: α-hederin N ammonium glycyrrhizate N DPPC monolayers. The weakly adsorbing hederacoside C has practi-
hederacoside C. cally no effect on surface pressure of the DPPC monolayer (the relax-
A similar trend can be observed for surface dilational storage and ation curve is very similar to that on pure water, without subphase
loss moduli of the Gibbs layers after 1 h adsorption (Table 2). Because exchange, not shown). In contrast, the presence of α-hederin sharply
of different surface pressures achieved after 1 h adsorption for each sa- increases the surface pressure to about 45 mN/m, i.e., 12 mN/m
ponin, the values cannot be directly compared. Nevertheless, it is clear above the surface pressure value prior to the subphase exchange.
that the layers formed by α-hederin are characterized by exceptionally For ammonium glycyrrhizate the increase of surface pressure is
much slower than for the Gibbs layer on bare water/air interface.
Table 1 The kinetics of glycyrrhizate adsorption is thus slowed down when
Stability constants for saponin–lipid binary complexes determined from UV/Vis titration the lipid layer is present, or that the latter undergoes slow rearrange-
in 90% (for α-hederin and hederacoside C) or 95% (for ammonium glycyrrhizate) ethanol. ment in the presence of ammonium glycyrrhizate.
Saponin Lipid Ka, M−1 Similar relaxation behavior can be observed for this saponin also
when cholesterol is present in the mixed lipid monolayer. This might
α-Hederin Cholesterol (5.0 ± 0.4)·104
DPPC (5.0 ± 0.2)·103 suggest that ammonium glycyrrhizate does not differentiate between
Hederacoside C Cholesterol (9.0 ± 0.3)·103 DPPC and cholesterol. Despite distinct differences between ammonium
DPPC (4.0 ± 0.2)·103 glycyrrhizate and hederacoside C on a free water surface and on DPPC
Ammonium glycyrrhizate Cholesterol No complex formation monolayers, in the presence of cholesterol their effect on surface pres-
DPPC No complex formation
sure is rather similar. Among the three investigated saponins, α-
K. Wojciechowski et al. / Biochimica et Biophysica Acta 1858 (2016) 363–373 367

Fig. 3. Surface pressure vs time curves for Gibbs layers formed from aqueous solutions (c = 10−4 M) of pure saponins (□) and for Langmuir monolayers during the subphase exchange
following compression to 32.5 mN/m: DPPC ( ) and DPPC/cholesterol 10:9 ( ). Left panel: ammonium glycyrrhizate, middle panel: hederacoside C, and right panel: α-hederin (inset
shows expansion of the curve for DPPC/cholesterol on α-hederin). The green dashed line corresponds to the initial value of surface pressure, Π0 = 32.5 mN/m.

hederin is clearly the most sensitive to the presence of cholesterol. responsible for the high visco-elasticity of its layers on water, are there-
During the first minutes of subphase exchange (even before it is fore weakened by the saponin molecules penetrating the monolayer,
completed) the surface pressure rises very steeply and few sudden especially in the case of hederacoside C. The latter observation is sur-
jumps can be noticed (see the inset in Fig. 3, right panel). These pres- prising in view of the weak effect of this saponin on relaxation behavior
sure jumps might suggest a partial collapse of the lipid monolayer as well as on the neutron reflectivity curve (see below) of the DPPC
penetrated by α-hederin. However, if the lipid layer were replaced monolayer.
by a detergent (as it was observed e.g., for ionic synthetic surfac- On the other hand, the effect of α-hederin on dilatational surface rhe-
tants: SDS or CTAB), the surface pressure would decrease to the ological response of DPPC monolayers is markedly different (note that
levels comparable with those for the respective Gibbs layers of the these experiments were performed on water containing 4% DMSO for sol-
pure surfactant [19]. In the case of α-hederin and DPPC/cholesterol, ubility reasons). The storage modulus, E′ = 188.6 mN/m, seems not to be
the monolayer replacement by the saponin can be excluded given significantly affected by the saponin penetration (for DPPC on water/
the consecutive increase of surface pressure and its stabilization at DMSO 4%, E′ = 181.8 mN/m, close to the value for DPPC on pure water,
about 50 mN/m, i.e., over 17 mN/m higher than the initial value be- E′ = 176.5 mN/m). However, the viscous component (loss modulus) is
fore the subphase exchange, and 25 mN/m higher than α-hederin drastically reduced as compared to the starting DPPC monolayer on
on bare water/air interface. water/DMSO 4% (E″ = 8.5 mN/m vs E″ = 111.2 mN/m). Interestingly, E
″ for the penetrated layer is also smaller than that for the α-hederin's
3.2.2. Surface dilational rheology Gibbs layer. Thus, in contrast to the bare DPPC monolayers and α-
The observations regarding the effect of the three saponins on relax- hederin's Gibbs layers, the α-hederin-penetrated DPPC monolayer
ation of DPPC and DPPC/cholesterol mixed monolayers are also con- behaves as almost purely elastic body. The fact that the loss modulus
firmed by the results of the surface dilational rheology measurements decreases upon penetration by all three saponins despite an increase of
(Table 2). For DPPC layers exposed to both ammonium glycyrrhizate surface pressure indicates that packing of the molecules is different than
and hederacoside C, the storage and loss moduli take values significant- it would be in an analogous monolayer compressed by mechanical
ly higher than those for the bare saponins, yet smaller than for the means. The molecules in the mixed DPPC–saponin monolayers have
starting DPPC layers (E′ = 176.5 mN/m, E″ = 105.0 mN/m [19]). The at- less possibilities to relax the mechanical stress by molecular reorientation
tractive interactions between the closely-packed DPPC molecules, and rearrangements, than in pure DPPC.
When cholesterol is present in addition to DPPC, the effect of
Table 2 α-hederin is again different: both E′ and E″ of the penetrated mixed
Dilational surface visco-elasticity moduli for Gibbs layers of the triterpene saponins on wa- lipid layer are close to the values for the Gibbs layer of α-hederin on
ter–air interface (10−4 M) and for the Langmuir monolayers of DPPC and DPPC/cholester-
free water surface, despite markedly different monolayer relaxation be-
ol (10:9 mol/mol) penetrated with these saponins (10−4 M). “n.d.” refers to the cases
where surface rheological response could not be determined because of a brittle nature havior. Unfortunately, the surface dilational visco-elasticity of the pene-
of cholesterol-rich monolayers. trated DPPC/cholesterol monolayers values cannot be compared
quantitatively with those for the corresponding monolayer on pure
Subphase Gibbs layer DPPC DPPC/cholesterol
water, because of a loss of material during surface area oscillations for
E′ E″ E′ E″ E′ E″
the latter (see Supporting Information, Fig. S1.). This phenomenon orig-
Ammonium glycyrrhizate 21.1 4.3 134.6 62.0 n.d. n.d. inates probably from a very high rigidity of the cholesterol-rich mono-
Hederacoside C 2.6 0.8 84.1 42.1 n.d. n.d. layers on water. For the same reasons, the surface dilational
α-Hederin 242.3 147.0 188.6 8.5 303.2 146.0
rheological response could not be retrieved for the mixed DPPC/
368 K. Wojciechowski et al. / Biochimica et Biophysica Acta 1858 (2016) 363–373

cholesterol monolayers in contact with the two remaining saponins, of the stripe pattern might be formation of folds during compression,
which interact only weakly with the monolayers. As far as the me- but at the present stage we cannot distinguish between the two possibil-
chanical response to the monolayer area perturbation is concerned, ities. After introduction of the saponins, the contrast between dark and
the DPPC/cholesterol mixed layers penetrated by the weakly- bright stripes first increases, and with time the inhomogeneity gets less
interacting saponins (hederacoside C and ammonium glycyrrhizate) visible (Fig. 5). This suggests eventual mixing of the phases, possibly
can be considered similar to those not penetrated. due to the flow induced by introduction of saponins.
Surprisingly, the two saponins for which the least changes were
3.2.3. Fluorescence microscopy noticed in surface pressure showed also the highest incompatibili-
Fluorescence microscopy is a well-established tool to investigate the ties with the penetrated mixed monolayer. The circular bright do-
phase behavior of amphiphilic systems at the air–water interface [29– mains seen in microphotographs for ammonium glycyrrhizate and
32]. Here it was employed to highlight possible effects of the triterpene especially hederacoside C (Fig. 5) correspond probably to a
saponins penetration on lipid monolayers initially compressed to Π0 = cholesterol-poor phase [36], separated as a consequence of the
32.5 mN/m (Fig. 4). saponin-DPPC and/or saponin-cholesterol complex formation. Inter-
During compression, bare DPPC monolayers showed typical bean- estingly, α-hederin–lipid complexes are rather well miscible with
shaped LC domains growing within the LE phase [33,34] (see the lipid matrix of the DPPC/cholesterol monolayer, which is consis-
Supporting Info, Fig. S2). Above surface pressure of 10 mN/m the LC do- tent with the fact that the α-hederin-penetrated mixed monolayer
mains filled almost an entire space and the fluorescence microscopy im- spontaneously approaches a surface pressure of about 50 mN/m
ages did not change significantly upon further compression (Fig. 4). (Fig. 3). The α-hederin-penetrated monolayers were also distinct
Introduction of the saponins at Π0 = 32.5 mN/m did not change the sit- in their sharply increasing viscosity, which practically blocked any
uation (not shown), despite a clear increase of surface pressure, even up movement due to the subphase exchange after 20 min. No fluores-
to 45 mN/m (for α-hederin, see Fig. 3). This suggests that all three sapo- cence changes were noticed during the sudden surface pressure
nins are miscible with DPPC, independently of the strength of their in- drops observed in the first 3–5 min of penetration, confirming that
teraction with the lipid. the pressure jumps are not related to any processes within the
In binary lipid monolayers, the phase transitions are additionally com- monolayer (e.g., a collapse), but are rather artifacts due to partial
plicated by mutual solubility and complex formation between the lipids wetting of the trough's Teflon rim.
[35,36]. Keller et al. [37] in a systematic study of mixtures of cholesterol
with variable-length phosphatidylcholines (PC) nicely showed that phos- 3.2.4. Neutron reflectivity
pholipids with chain-melting temperatures above 23 °C exhibit two The effect of the three triterpene saponins on the structure of DPPC
upper miscibility critical points, depending on the amount of cholesterol. and DPPC/cholesterol monolayers was investigated using neutron re-
At lower surface pressures, the PC:cholesterol complexes phase separate flectivity (NR). The lipid chain-perdeuterated DPPC (DPPC-d62) and
from the pure PC and cholesterol, forming the cholesterol-poor (α) and protonated cholesterol were used for this purpose. In order to highlight
cholesterol-rich (β) regions in the phase diagram [38]. Even though the the changes in the lipid layer induced by the saponin, the scattering
existence of a critical point in the β region of DPPC/cholesterol mixtures length density (SLD) of the subphase was matched to that of air (so-
has been initially neglected [39], recent works clearly confirm remixing called null reflecting water, NRW). The reflectivity curves for bare
at higher surface pressures [40,41]. Similar features have been observed DPPC-d62 and DPPC-d62/cholesterol (10:9 mol/mol) monolayers on
in other cholesterol/phospholipid mixtures [42–44]. Also our present pure NRW and NRW containing dissolved α-hederin, hederacoside
fluorescence microscopy results confirm remixing of the DPPC/cholester- and ammonium glycyrrhizate at a concentration of 10−4 M are shown
ol (10:9 mol/mol) monolayer above Π0 = 3 mN/m (see Supporting Info, in Fig. 6. The saponins were introduced to the subphase in the same
Fig. S3). Therefore, the DPPC/cholesterol monolayers prior to saponin in- way as for the surface pressure measurements, by the subphase ex-
troduction (Π0 = 32.5 mN/m) likely contain a liquid-ordered (lo) change following compression to 32.5 mN/m. For comparison, also the
mixed phase (Fig. 4). The alternate bright and dark stripes (not to be corresponding reflectivity curves for the Gibbs layers of each saponin
confused with the stripe phase often observed near critical points [45]) at the same concentration at the NRW/air interface are shown.
suggest formation of cholesterol-rich/cholesterol-poor, few micrometer- All three saponins spontaneously form Gibbs layers on free NRW
thick linear domains of very low line tension [30,42]. The increase of the surface, as evidenced by an increased reflectivity of neutrons (NRW
number of dark stripes with increasing cholesterol concentration (data without saponins was not reflecting neutrons at all). The NR curves
not shown) suggests that they may correspond to the condensed phase for the saponins' Gibbs layers could be fitted assuming a 51.0 Å-
and/or cholesterol-rich regions. An alternative explanation for the origin thick (for ammonium glycyrrhizate and hederacoside C) or a 19.4

Fig. 4. Fluorescence microphotographs of DPPC (left) and mixed 10:9 (mol/mol) DPPC/cholesterol (right) Langmuir monolayers compressed to Π0 = 32.5 mN/m.
K. Wojciechowski et al. / Biochimica et Biophysica Acta 1858 (2016) 363–373 369

Fig. 5. Fluorescence microphotographs of mixed DPPC/cholesterol (10:9) Langmuir monolayers penetrated by ammonium glycyrrhizate (left), hederacoside C (middle) and α-hederin
(right) after 2 min (top), 10 min (middle) and 30 min (bottom). The monolayers were initially compressed to 32.5 mN/m.

Fig. 6. NR curves for DPPC-d62 ( ) and DPPC-d62/cholesterol (10:9 mol/mol, ) monolayers on NRW containing 10−4 M of: (A) ammonium glycyrrhizate, (B) hederacoside and α-hederin
(C). For comparison, the NR curves for Gibbs layers of each saponin on the free surface of NRW (■), as well as for DPPC-d62 ( ) and DPPC-d62/cholesterol (10:9 mol/mol, ) monolayers on
pure NRW are shown. Solid lines correspond to the best fits from Tables 3 and 4. All experiments were performed at 21 °C.
370 K. Wojciechowski et al. / Biochimica et Biophysica Acta 1858 (2016) 363–373

Å-thick (α-hederin) layers (see Table 3). The two former saponins Table 4
seem to favor formation of thicker Gibbs layers, with adsorbed mol- The best-fit parameters (for neutron reflectivity of bare DPPC-d62 and DPPC-d62/cho-
lesterol (10:9 mol/mol) monolayers on pure NRW and NRW containing α-hederin,
ecules oriented presumably perpendicular to the surface. For hederacoside C and ammonium glycyrrhizate (10−4 M). Surface roughness was fixed
bidesmosidic hederacoside C, this orientation is probably forced by at 2 Å in all fittings.
the presence of two glycone groups attached to the opposite sides
Subphase DPPC-d62 DPPC-d62/chol
of the triterpenoid aglycone (C3 and C28 atoms), while for ammoni-
um glycyrrhizate, a hydrophilic free carboxylic group (see Fig. 1) Layer thickness Scattering Layer thickness Scattering
(fitted), length (fitted), length
probably plays the decisive role in forcing the perpendicular orienta-
τ (Å) density τ (Å) density
tion. In contrast, much smaller thickness of the monodesmosidic α- (fitted), (fitted),
hederin suggests an orientation parallel to the surface, which is SLDlayer SLDlayer
probably favored by the presence of only one glycone group and (10−6/Å2) (10−6/Å2)
the free carboxylic group attached to C-28. The corresponding best- α-Hederin 26.2 5.80 23.2 2.70
fit values of SLD for the Gibbs layers vary among the saponins, – – 15.0 0.90
reflecting different volume fractions in the layer (φsaponin) and differ- – – 33.9 0.54
Hederacoside C 26.2 5.56 23.8 3.02
ent adsorbed amounts (Γsaponin ). The two parameters calculated
38.0 0.15 15.0 0.49
using Eqs. 1 and 2 are collected in Table 3. Ammonium 26.2 5.56 23.8 3.02
glycyrrhizate 36.3 0.24 10.2 0.76
SLDlayer ¼ SLDsaponin  φsaponin ð1Þ NRW 26.2 5.56 23.8 3.02
NRW/DMSO 28.0 5.56 27.1 3.02
τ
Γ saponin ¼ φ ð2Þ
NAv V saponin saponin
hydrated glycone parts extending beyond the lipid layer towards the
where SLD layer is the scattering length density obtained from the aqueous phase, hence its low thickness and SLD.
best-fit of the NR curve, SLDsaponin and Vsaponin are theoretical values For α-hederin the situation is different. Despite a clear increase of
of the scattering length density, and molar volume of the given sapo- surface pressure, the NR curve for DPPC-d62 penetrated with this saponin
nin. NAv is the Avogadro's number. lays slightly below that for the bare monolayer on pure NRW. This is
The adsorbed amount calculated from the NR data is the highest for α- probably caused by the presence of only one glycone chain, which can
hederin (3.0·10−6 mol/m2) and the lowest for hederacoside C be accommodated almost entirely within the DPPC layer, only slightly
(0.8·10−6 mol/m2), in good agreement with the order of their ability to changing its thickness and SLD. Unfortunately, the resolution of NR
increase surface pressure in the Langmuir trough measurements measurements is not sufficient to quantify these small changes of SLD
(Fig. 3). The distinctly different Γα-hederin is a consequence of much lower for α-hederin penetrated DPPC layers. It should be noted here that all
extent of the α-hederin layer hydration (φ = 1), as compared to the the measurements for α-hederin were performed in the presence of
other two saponins (φ = 0.2 and 0.3). DMSO in the subphase (4% v/v). As shown by Dabkowska et al., at such
The NR curve for bare DPPC monolayer is consistent with the literature low DMSO content, the structure of the lipid layer is not significantly
results [46,47]. The NR results confirm that the DPPC monolayers com- affected [48]. Nevertheless, the effect of DMSO was taken into account
pressed to 32.5 mN/m can resist the detergent activity of all three sapo- by using the reference measurements for lipid layers spread also on
nins. Thanks to perdeuteration of the alkyl chains in DPPC used for our NRW containing DMSO. The NR curves for DPPC-d62 in the absence of
NR experiments, any removal of the highly neutron-reflecting DPPC-d62 α-hederin are shifted slightly upwards with respect to those on pure
would immediately result in a decrease of reflectivity. With the progres- NRW (see Supporting Information, Fig. S3). Our results show that in the
sive loss of deuterated material from the monolayer to the subphase, presence of DMSO (4% v/v), thickness of the lipid layers slightly increased
the resulting reflectivity curve would then tend towards that for the re- (by 3.3 Å, see Table 4).
spective Gibbs layer on bare NRW surface. No such trends were observed Similarly to the pure phospholipid monolayers, the mixed DPPC-d62/
even on the eight-hour timescale of the NR experiment. On the contrary, cholesterol (10:9 mol/mol) ones are not solubilized by hederacoside C
for hederacoside C and ammonium glycyrrhizate the reflectivity curves or ammonium glycyrrhizate. The reflectivity curves are shifted upwards,
were even slightly shifted upwards, suggesting deposition of additional as for the pure DPPC-d62 layers, suggesting accumulation of additional
neutron-reflecting material (with SLD slightly higher than that of NRW). adsorbed material. Alternatively, because the lipid layers contain now a
For both saponins, an additional low-SLD layer had to be included to fit protonated cholesterol (SLDcholesterol = 0.21·10−6 Å2), the upwards shift
the NR curves for DPPC-d62 after the subphase exchange (Table 4). In could be explained by the replacement of cholesterol with other mole-
both cases, the additional layer is thinner and has lower SLD than the cor- cules of higher SLD (e.g., with saponins, see Table 3). However, the NR
responding freely adsorbing Gibbs layer. This effect could be explained by curve cannot be fitted with a single layer, and the presence of an addition-
partial inclusion of the hydrophobic triterpenoid part of the adsorbing sa- al thin layer of 10.2 and 15.0 Å (for ammonium glycyrrhizate and
ponin molecules into the topmost DPPC-d62 monolayer. Such a partial in- hederacoside C, respectively) had to be assumed. In analogy to the results
clusion could also explain the increase of surface pressure observed in the for pure DPPC, they might correspond to the glycone parts of both sapo-
Langmuir trough measurements during the subphase exchange. The ad- nins extending towards the aqueous phase, while the less hydrated agly-
ditional layer seen in NR is probably composed mostly of the NRW- cone parts would be intermixed within the lipid layer. The reduced depth

Table 3
The best-fit parameters for neutron reflectivity of Gibbs layers of free saponins on pure NRW and NRW containing α-hederin, hederacoside C and ammonium glycyrrhizate (10−4 M).
Surface roughness was fixed at 2 Å in all fittings.

Subphase Gibbs layers

Molar volume Scattering length density Layer thickness Scattering length density Volume Adsorbed amount,
(theoretical), Vsaponin (Å3) (theoretical), SLDsaponin (10−6/Å2) (fitted), τ (Å) (fitted), SLD (10−6/Å2) fraction, φ Γ (mol/m2)

α-Hederin 1250 0.77 22.6 0.77 1 3.0·10−6


Hederacoside C 2030 0.91 51.0 0.20 0.2 8.4·10−7
Ammonium glycyrrhizate 1395 0.99 51.0 0.28 0.3 1.8·10−6
K. Wojciechowski et al. / Biochimica et Biophysica Acta 1858 (2016) 363–373 371

of the second layer (containing mostly the glycone groups) for the surprisingly, a previous electrospray ionization mass spectroscopy
cholesterol-containing lipid layers indicates that the presence of choles- (ESI-MS) study of Lekar et al. showed that in the gas phase cholesterol
terol promotes penetration of these two saponins into the lipid layer. is complexed by α-hederin, but not by hederacoside C [51,52]. However,
This is also in agreement with the more pronounced surface pressure in- as far as the bulk solution interaction is concerned, the location of the
crease observed for cholesterol-containing lipid layer, as compared to carboxylic group seems to be the most important, as proven by the
those composed solely of DPPC (Fig. 2). lack of interaction of DPPC and cholesterol with ammonium
The most pronounced changes in NR curves can be noticed for α- glycyrrhizate. Moreover, the comparison of chemical structures of the
hederin penetrating the mixed DPPC-d62/cholesterol monolayer on three saponins points to an important role of van der Waals interactions
NRW containing DMSO (4%). It should be noted that the NR curves for between the steroid backbone of cholesterol with the triterpene agly-
DPPC-d62/cholesterol in the absence of α-hederin are shifted slightly cones of the saponins in bulk. In view of this, the presence of a ketone
upwards with respect to those on pure NRW, analogously to DPPC-d62 group in ammonium glycyrrhizate may act as a steric hindrance
(see Supporting Information, Fig. S4). Thickness of the lipid layers in- preventing a close approach and consequently efficient van der Waals
creased slightly by 1.8 Å in the presence of DMSO (4% v/v) (Table 4). interaction with cholesterol and DPPC.
To obtain a reasonable fit of the reflectivity for the DPPC-d62/ The ability to form lipid–saponin complexes in the homogeneous
cholesterol/α-hederin system, two additional layers had to be taken bulk solution does not fully correlate with the strength of interactions
into account (Table 4). The first, 15 Å-thick layer with SLD of in a confined environment of lipid monolayers. The best correlation be-
0.9·10− 6 Å− 2 (higher than that for pure cholesterol, SLDcholesterol = tween the bulk complexation and the monolayer activity can be found
0.21·10−6 Å2, or α-hederin, SLDα-hederin = 0.77·10−6 Å2) must contain for α-hederin, for which the highest affinity to the lipids is evident
some higher-SLD material, e.g. DPPC-d62. The second layer of lower SLD from both the bulk and monolayer studies. Even the preference for cho-
(0.54·10−6 Å− 2) and a thickness of about 34 Å probably comprise lesterol over DPPC (Table 1) is well reflected in the monolayer studies in
mostly cholesterol and the saponin (although the presence of DPPC- this case (Fig. 2). The highest value of the complex stability constant as
d62 cannot be excluded). Thus, α-hederin seems to pull part of choles- well as the most pronounced monolayer response correspond well also
terol and DPPC out of the lipid layer. The latter, being less strongly to the highest membranolytic activity of α-hederin. Similarly, the lack of
bound by α-hederin remains partially immersed in the monolayer, interaction of ammonium glycyrrhizate with both lipids is evident
while the stronger α-hederin–cholesterol complex is shifted more to- equally in the bulk and monolayer experiments, and corresponds well
wards the aqueous phase. A similar picture of a digitonin-penetrated to its weak membranolytic activity. However, the relatively high affinity
mixed lipid bilayer was recently provided by Frenkel et al. using X-ray to both cholesterol and DPPC for hederacoside C observed in bulk is nei-
reflectivity (XR) and QCM study [49]. The XR approach, however, does ther reflected in monolayer studies, nor in its membranolytic activity.
not allow for contrast variation, hence the authors could not resolve This again highlights the important role of spatial organization of the
the composition of the observed 45 Å-thick additional layer. lipids in their interactions with saponins. One possible explanation has
been recently provided by Lorent et al. who extensively discussed the
4. Discussion differences between hederagenin, δ-hederin and α-hederin in view of
different curvature of the molecules [53,54]. The failure of our solid–liq-
Even though the saponin–membrane interactions are generally be- uid extraction experiments could also be explained by the crucial role of
lieved to be dominated by cholesterol, our study shows that phospho- the lipid state in its interaction with saponins.
lipids may also actively participate in attracting saponins to biological Hederacoside C and ammonium glycyrrhizate easily penetrate
membranes. The saponin–lipid complexes formed in solution are only monolayers of both DPPC and DPPC/cholesterol without significantly
slightly less stable for DPPC than for cholesterol (Table 1). These obser- disturbing them. This correlates well with their mild activity against
vations are in line with suggestions of Demana et al., who used diffuse both red blood cells (hemolysis) and cancer cells [55]. For α-hederin
reflectance IR (DRIFT) spectroscopy for studying complexes formed be- penetrating the mixed DPPC/cholesterol monolayers, the situation is
tween a saponin mixture from Quillaja saponaria Molina (Quil A) and clearly more complex, in line with its pronounced hemolytic and cyto-
phosphatidylcholine (PC) [50]. They showed that the PC's CO, PO groups toxic activity. Not only the surface pressure response is much more pro-
may form hydrogen bonds with the\\OH groups of the saponin's sugar nounced than for other saponins, but also the structure and mechanical
moieties. In parallel, the free carboxylic acid group of the latter may properties of the penetrated ternary α-hederin/DPPC/cholesterol layer
electrostatically bind with the PC's (CH3)3N+ group. The differences in are distinct. Bare DPPC/cholesterol monolayers are too rigid to reliably
stability constants for the saponin–lipid complexes observed in our determine their surface dilational rheological parameters. Penetration
studies point to the crucial role of proper spatial arrangement of the with α-hederin softens them to the extent allowing for quantitative
complimentary groups in these complexes. Despite several structural measurements without breaking the monolayer. The visco-elasticity
similarities between the three saponins, clear differences in lipid- modulus for the penetrated layers is in fact similar to that for the α-
binding properties can be noticed even in solution, where the effects hederin Gibbs layers. On the other hand, both the surface pressure and
of spatial arrangement of the host and guest molecules are less neutron reflectivity results assure us that the lipid layer is still present
constrained than in the mono- or bilayer. The most striking example is after penetration. This prompts us to propose that α-hederin/DPPC/
provided by comparison of ammonium glycyrrhizate and α-hederin. cholesterol mixtures could form surface-confined structures analogous
Both possess a two-sugar glycone attached to C-3, but only α-hederin to those suggested by Frenkel et al. [49] for digitonin, or to the open
is capable of complexing DPPC or cholesterol. A single carboxylic cage-like colloidal immunostimulatory complex matrices (ISCOM)
group of α-hederin (at C-28 position) is more effective than the three [56]. The latter structures are often found in cholesterol/phosphatidyl-
present in ammonium glycyrrhizate (one attached to the glycone at choline layers hydrated with saponin solutions, especially with purified
C-30 position, and two in the glucuronic acids of the glycone part). Ap- QBS extracts (Quil A) [57]. They were shown to exist in mixtures con-
parently, their spatial arrangements in the latter molecule do not allow taining between 10 and 30% of cholesterol after 1 day of hydration.
for complementarity of hydrogen and ionic bonding as effectively as in However, if hydration was prolonged to 2 months, they could even be
α-hederin or Quil A. observed in mixtures containing up to 70% of cholesterol. Whether the
Some authors noted that the presence of a free\\COOH group is im- α-hederin complexes with cholesterol and DPPC in the monolayer are
portant for lipid binding [23], but the small difference in stability con- of an adduct-type, as suggested by Frenkel et al. [49], or cage-like, re-
stants for DPPC complexation between α-hederin and hederacoside C mains to be determined in the future. In both cases, an additional
observed in the present study suggests that this effect is more likely layer built-up from the saponin and parts of the original lipid layer
linked with spatial confinement in the membrane. Somehow would strongly affect mechanical and physiological properties of the
372 K. Wojciechowski et al. / Biochimica et Biophysica Acta 1858 (2016) 363–373

biological membrane (e.g. in erythrocytes), and can be responsible for Appendix A. Supplementary data
the enhanced membranolytic activity of α-hederin. Despite better
packing of molecules suggested by the increased surface pressure, the Supplementary data to this article can be found online at http://dx.
overall structure might be more fluid and permeable than the original doi.org/10.1016/j.bbamem.2015.12.001.
non-penetrated structure.

References
5. Conclusions
[1] J.M. Augustin, V. Kuzina, S.B. Andersen, S. Bak, Molecular activities, biosynthesis and
Adsorption of three triterpene saponins (α-hederin, hederacoside C evolution of triterpenoid saponins, Phytochemistry 72 (2011) 435–457.
[2] T.K. Das, D. Banerjee, D. Chakraborty, M.C. Pakhira, B. Shrivastava, R.C. Kuhad, Sapo-
and ammonium glycyrrhizate) onto a free water surface and onto Lang- nin: role in animal system, Vet. World 5 (2012) 248–254.
muir monolayers of DPPC and DPPC/cholesterol (10:9 mol/mol) was de- [3] A. Osbourn, Sponins and plant defence — a soap story, Trends Plant Sci. 1 (1996)
scribed. A combined monolayer relaxation, surface dilational rheology, 4–9.
[4] A. Szakiel, C. Pa¸czkowski, M. Henry, Influence of environmental biotic factors on the
fluorescence microscopy and neutron reflectivity (NR) study showed
content of saponins in plants, Phytochem. Rev. 10 (2011) 493–502.
that all three saponins are able to interact and partially penetrate the [5] J.H. Lorent, J. Quetin-Leclercq, M. Mingeot-Leclercq, . The amphiphilic nature of sa-
pure DPPC and mixed DPPC/cholesterol monolayers. At the surface of ponins and their effects on artificial and biological membranes and potential conse-
pure water, α-hederin (10−4 M) forms densely packed layers with the quences for red blood and cancer cells, Org. Biomol. Chem. 12 (2014) 8803–8822.
[6] E.A.J. Keukens, T. De Vrije, L.A.M. Jansen, H. De Boer, M. Janssen, A.I.P.M. De Kroon,
thickness of 22.6 Å and decreased surface tension by 25 mN/m. In con- W.M.F. Jongen, B. De Kruijff, Glycoalkaloids selectively permeabilize cholesterol con-
trast, hederacoside C and ammonium glycyrrhizate decrease surface taining biomembranes, Biochim. Biophys. Acta Biomembr. 1279 (1996) 243–250.
tension only slightly and form highly hydrated (70–80%), rather loose [7] F. Ransom, Saponin Und Sein Gegengift, Dtsch. Med. Wochenschr. 13 (1901)
194–196.
layers of 51.0 Å thickness. The surface activity of the saponins correlates [8] A. Windaus, Uber Die Entgiftung Der Saponine Durch Cholesterin, Ber. Dtsch. Chem.
well with their effect on lipid monolayers: α-hederin was shown to af- Ges. 42 (1909) 238–246.
fect the layers to the highest extent, especially when cholesterol is [9] S. Böttger, M.F. Melzig, The influence of saponins on cell membrane cholesterol,
Bioorg. Med. Chem. 21 (2013) 7118–7124.
present. The mixed DPPC/cholesterol monolayers in contact with [10] J.H. Schulman, E.K. Rideal, Molecular interaction in monolayers II. The action of
α-hederin develop an additional thick layer extending towards the sub- haemolytic and agglutinating agents on lipo-protein monolayers, Proc. R. Soc.
phase and consisting probably of α-hederin-DPPC and α-hederin- Lond. Ser. B Biol. Sci. 122 (1937) 46–57.
[11] J.H. Schulman, E.K. Rideal, Molecular interaction in monolayers: I. Complexes be-
cholesterol complexes. We believe that similar structures might be re-
tween large molecules, Proc. R. Soc. Lond. Ser. B Biol. Sci. 122 (1937) 29–45.
sponsible for the pore formation during α-hederin's membranolytic ac- [12] A.D. Bangham, R.W. Horne, Action of saponin on biological cell membranes, Nature
tivity (e.g., during hemolysis). The other two saponins partially 196 (1962) 952–953.
[13] A.M. Glauert, J.T. Dingle, J.A. Lucy, Action of saponin on biological cell membranes,
penetrate the lipid layers, probably only with their hydrophilic glycones
Nature 196 (1962) 953–955.
extending towards the aqueous subphase. [14] F. Lin, R. Wang, Hemolytic mechanism of dioscin proposed by molecular dynamics
For comparison, the saponin–lipid complex formation was also stud- simulations, J. Mol. Model. 16 (2010) 107–118.
ied in ethanol-aqueous homogeneous solutions using UV/Vis spectros- [15] M. Hu, K. Konoki, K. Tachibana, Cholesterol-independent membrane disruption
caused by triterpenoid saponins, Biochim. Biophys. Acta, Lipids Lipid Metab. 1299
copy. The oleanolic acid-type saponins (α-hederin and hederacoside (1996) 252–258.
C) were shown to form complexes with lipids characterized by stability [16] F. Siu, D. Ma, Y. Cheung, C. Lok, K. Yan, Z. Yang, M. Yang, S. Xu, B.C. Ko, Q. He, et al.,
constants in the range (4.0 ± 0.2)·103–(5.0 ± 0.4)·104 M−1. The com- Proteomic and transcriptomic study on the action of a cytotoxic saponin
(polyphyllin D): induction of endoplasmic reticulum stress and mitochondria-medi-
plexes with cholesterol are stronger than those with DPPC. On the con- ated apoptotic pathways, Proteomics 8 (2008) 3105–3117.
trary, ammonium glycyrrhizate does not form complexes with any of [17] S. Böttger, K. Hofmann, M.F. Melzig, Saponins can perturb biologic membranes and
the lipids in solution. Among the studied saponins, only α-hederin in- reduce the surface tension of aqueous solutions: a correlation? Bioorg. Med. Chem.
20 (2012) 2822–2828.
teracts strongly with the lipid monolayers and displays membranolytic [18] C. Gauthier, J. Legault, K. Girard-Lalancette, V. Mshvildadze, A. Pichette, Haemolytic
activity towards red blood cells and cancer cells. Ammonium activity, cytotoxicity and membrane cell permeabilization of semi-synthetic
glycyrrhizate does not interact strongly with the lipids neither in bulk, and natural lupane- and oleanane-type saponins, Bioorg. Med. Chem. 17 (2009)
2002–2008.
nor in monolayers. It also shows negligible membrane activity. These re-
[19] K. Wojciechowski, M. Orczyk, T. Gutberlet, M. Trapp, K. Marcinkowski, T. Kobiela, T.
sults suggest that the chemical interactions of the aglycone part are not Geue, Unusual penetration of phospholipid mono- and bilayers by Quillaja bark sa-
decisive in the strength of interactions with cholesterol, and/or that the ponin biosurfactant, Biochim. Biophys. Acta Biomembr. 1838 (2014) 1931–1940.
[20] K. Wojciechowski, M. Orczyk, K. Marcinkowski, T. Kobiela, M. Trapp, T. Gutberlet, T.
interactions with glycone parts are important. The ability to form a
Geue, Effect of hydration of sugar groups on adsorption of Quillaja bark saponin at
lipid–saponin complex in bulk solution seems to be a necessary but air/water and Si/water interfaces, Colloids Surf. B: Biointerfaces 117 (2014) 60–67.
not sufficient condition for efficient interaction with the same lipids in [21] K. Wojciechowski, Surface activity of saponin from Quillaja bark at the air/water and
a confined environment of the monolayer or real biological membranes. oil/water interfaces, Colloids Surf. B: Biointerfaces 108 (2013) 95–102.
[22] R.A. Isbrucker, G.A. Burdock, Risk and safety assessment on the consumption of lic-
The fact that hederacoside C, despite similar cholesterol-binding ability orice root (Glycyrrhiza sp.), its extract and powder as a food ingredient, with empha-
as α-hederin, is less active in membrane permeabilization might reflect sis on the pharmacology and toxicology of glycyrrhizin, Regul. Toxicol. Pharmacol.
its hindered access to the cholesterol molecules embedded in the 46 (2006) 167–192.
[23] M. Chwalek, N. Lalun, H. Bobichon, K. Plé, L. Voutquenne-Nazabadioko, Structure–
bilayer. activity relationships of some hederagenin diglycosides: haemolysis, cytotoxicity
and apoptosis induction, Biochim. Biophys. Acta Gen. Subj. 1760 (2006) 1418–1427.
Transparency document [24] R. Gilabert-Oriol, K. Mergel, M. Thakur, B. Von Mallinckrodt, M.F. Melzig, H. Fuchs, A.
Weng, Real-time analysis of membrane permeabilizing effects of oleanane saponins,
Bioorg. Med. Chem. 21 (2013) 2387–2395.
The Transparency document associated with this article can be [25] M. Gupta, T. Gutberlet, J. Stahn, P. Keller, D. Clemens, AMOR — the time-of-flight
found, in online version. neutron reflectometer at SINQ/PSI, Pramana 63 (2004) 57–63.
[26] C. Braun, Parratt32 Fitting Routine for Reflectivity Data, Experimental report HMI,
Berlin, 1999.
Acknowledgments [27] L.G. Parratt, Surface studies of solids by total reflection of X-rays, Phys. Rev. 95
(1954) 359–369.
[28] K. Golemanov, S. Tcholakova, N. Denkov, E. Pelan, S.D. Stoyanov, Remarkably high
This work was financially supported by the Polish National Science
surface visco-elasticity of adsorption layers of triterpenoid saponins, Soft Matter 9
Centre, grant no. DEC-2011/03/B/ST4/00780 and COST CM1101 Action. (2013) 5738–5752.
Ms. Katarzyna Krzeminska is acknowledged for assistance with UV/Vis [29] K.J. Stine, R.K. Hercules, J.D. Duff, B.W. Walker, Interaction of the glycoalkaloid
measurements. The work is based on experiments performed at the tomatine with DMPC and sterol monolayers studied by surface pressure measure-
ments and Brewster angle microscopy, J. Phys. Chem. B 110 (2006) 22220–22229.
neutron reflectometer AMOR at Swiss spallation neutron source SINQ, [30] H. Möhwald, Phospholipid and phospholipid-protein monolayers at the air/water
Paul Scherrer Institute, Villigen, Switzerland. interface, Annu. Rev. Phys. Chem. 41 (1990) 441–476.
K. Wojciechowski et al. / Biochimica et Biophysica Acta 1858 (2016) 363–373 373

[31] H.M. McConnell, Structures and transitions in lipid monolayers at the air–water in- [45] M. Seul, V.S. Chen, Isotropic and aligned stripe phases in a monomolecular organic
terface, Annu. Rev. Phys. Chem. 42 (1991) 171–195. film, Phys. Rev. Lett. 70 (1993) 1658–1661.
[32] C.W. McConlogue, T.K. Vanderlick, A close look at domain formation in DPPC mono- [46] M. Yaseen, J.R. Lu, J.R.P. Webster, J. Penfold, The structure of zwitterionic
layers, Langmuir 13 (1997) 7158–7164. phosphocholine surfactant monolayers, Langmuir 22 (2006) 5825–5832.
[33] R.M. Weis, H.M. McConnell, Cholesterol stabilizes the crystal–liquid interface in [47] D. Vaknin, K. Kjaer, J. Als-Nielsen, M. Losche, Structural properties of phosphatidyl-
phospholipid monolayers, J. Phys. Chem. 89 (1985) 4453–4459. choline in a monolayer at the air/water interface: neutron reflection study and reex-
[34] B.M. Discher, K.M. Maloney, D.W. Grainger, C.A. Sousa, S.B. Hall, Neutral lipids in- amination of X-ray reflection measurements, Biophys. J. 59 (1991) 1325–1332.
duce critical behavior in interfacial monolayers of pulmonary surfactant, Biochemis- [48] A.P. Dabkowska, L.E. Collins, D.J. Barlow, R. Barker, S.E. Mclain, M.J. Lawrence, C.D.
try (N. Y.) 38 (1999) 374–383. Lorenz, Modulation of dipalmitoylphosphatidylcholine monolayers by dimethyl
[35] S. Subramaniam, H.M. McConnell, Critical mixing in monolayer mixtures of phos- sulfoxide, Langmuir 30 (2014) 8803–8811.
pholipid and cholesterol, J. Phys. Chem. 91 (1987) 1715–1718. [49] N. Frenkel, A. Makky, I.R. Sudji, M. Wink, M. Tanaka, Mechanistic investigation of in-
[36] J.P. Slotte, Lateral domain formation in mixed monolayers containing cholesterol and teractions between steroidal saponin digitonin and cell membrane models, J. Phys.
dipalmitoylphosphatidylcholine or N-palmitoylsphingomyelin, Biochim. Biophys. Chem. B 118 (2014) 14632–14639.
Acta Biomembr. 1235 (1995) 419–427. [50] P.H. Demana, N.M. Davies, S. Hook, T. Rades, Analysis of quil a-phospholipid mix-
[37] S.L. Keller, A. Radhakrishnan, H.M. McConnell, Saturated phospholipids with high tures using drift spectroscopy, Int. J. Pharm. 342 (2007) 49–61.
melting temperatures form complexes with cholesterol in monolayers, J. Phys. [51] A.V. Lekar, L.A. Yakovishin, E.V. Vetrova, M.I. Rudnev, N.I. Borisenko, Mass spectrom-
Chem. B 104 (2000) 7522–7527. etry of the self-association and complexation of triterpene saponins and cholesterol,
[38] B.L. Stottrup, S.L. Keller, Phase behavior of lipid monolayers containing DPPC and J. Anal. Chem. 66 (2011) 1276–1280.
cholesterol analogs, Biophys. J. 90 (2006) 3176–3183. [52] L.A. Yakovishin, N.I. Borisenko, M.I. Rudnev, E.V. Vetrova, V.I. Grishkovets, Self-asso-
[39] T.M. Okonogi, H.M. McConnell, Contrast inversion in the epifluorescence of choles- ciation and complexation of triterpene glycosides and cholesterol, Chem. Nat.
terol-phospholipid monolayers, Biophys. J. 86 (2004) 880–890. Compd. 46 (2010) 49–52.
[40] C. Yuan, L.J. Johnston, Phase evolution in cholesterol/DPPC monolayers: atomic force [53] J. Lorent, L. Lins, O. Domenech, J. Quetin-Leclercq, R. Brasseur, M. Mingeot-Leclercq,
microscopy and near field scanning optical microscopy studies, J. Microsc. 205 Domain formation and permeabilization induced by the saponin a-hederin and its
(2002) 136–146. aglycone hederagenin in a cholesterol-containing bilayer, Langmuir 30 (2014)
[41] C. Yuan, L.J. Johnston, Distribution of ganglioside GM1 in L -a- 4556–4569.
dipalmitoylphosphatidylcholine/cholesterol monolayers: a model for lipid [54] J. Lorent, C.S. Le Duff, J. Quetin-Leclercq, M. Mingeot-Leclercq, Induction of highly
rafts, Biophys. J. 79 (2000) 2768–2781. curved structures in relation to membrane permeabilization and budding by the
[42] D.J. Benvegnu, H.M. McConnell, Line tension between liquid domains in lipid mono- triterpenoid saponins, a- and D-hederin, J. Biol. Chem. 288 (2013) 14000–14017.
layers, J. Phys. Chem. 96 (1992) 6820–6824. [55] L. Voutquenne, C. Lavaud, G. Massiot, L. Le Men-Olivier, Structure–activity relation-
[43] P. Mattjus, R. Bittman, J.P. Slotte, Molecular interaction and lateral domain formation ships of haemolytic saponins, Pharm. Biol. 40 (2002) 253–262.
in monolayers containing cholesterol and phosphatidylcholines with acyl- or alkyl- [56] H. Sun, Y. Xie, Y. Ye, ISCOMs and ISCOMATRIX™, Vaccine 27 (2009) 4388–4401.
linked C16 chains, Langmuir 12 (1996) 1284–1290. [57] P.H. Demana, N.M. Davies, B. Berger, U. Vosgerau, T. Rades, A comparison of pseudo-
[44] K. Tanaka, P.A. Manning, V.K. Lau, H. Yu, Lipid lateral diffusion in ternary diagrams of aqueous mixtures of quil a, cholesterol and phospholipid pre-
dilauroylphosphatidylcholine/cholesterol mixed monolayers at the air/water pared by lipid-film hydration and dialysis, J. Pharm. Pharmacol. 56 (2004) 573–580.
interface, Langmuir 15 (1999) 600–606.

You might also like