You are on page 1of 11

BBA - Biomembranes 1865 (2023) 184113

Contents lists available at ScienceDirect

BBA - Biomembranes
journal homepage: www.elsevier.com/locate/bbamem

Pro-inflammatory protein S100A9 alters membrane organization by


dispersing ordered domains
Rimgailė Tamulytė a, Evelina Jankaitytė a, Zigmantas Toleikis b, Vytautas Smirnovas b,
Marija Jankunec a, *
a
Institute of Biochemistry, Life Sciences Center, Vilnius University, Saulėtekio av. 7, Vilnius LT-10257, Lithuania
b
Institute of Biotechnology, Life Sciences Center, Vilnius University, Saulėtekio av. 7, Vilnius LT-10257, Lithuania

A R T I C L E I N F O A B S T R A C T

Keywords: Pro-inflammatory, calcium-binding protein S100A9 is localized in the cytoplasm of many cells and regulates
Aggregation several intracellular and extracellular processes. S100A9 is involved in neuroinflammation associated with the
Atomic force microscopy pathogenesis of Alzheimer’s disease (AD). The number of studies on the impact of S100A9 in co-aggregation
S100A9
processes with amyloid-like proteins is increasing. However, there is still a lack of data on how this protein
Lipid membrane remodeling
interacts with lipid membranes. We employed atomic force microscopy (AFM), dynamic light scattering (DLS),
Phase separation
Tethered lipid bilayers and fluorescence measurements (Laurdan and Thioflavin-T) to study the interaction between protein and the
membrane surface. We used lipid vesicles in bulk and planar tethered lipid bilayers as biomimetic membrane
models. We demonstrated that the protein accumulates on negatively charged lipid bilayers but with no further
loss of the bilayer’s integrity. The most important result is that the initial adsorption and accumulation of apo-
form of S100A9 on the lipid membrane surface is lipid phase-sensitive. The breaking down of raft-like and
disappearance of gel-like domains indicate that protein incorporates into the hydrophobic part of the lipid
bilayer. We observed the most noticeable loss of integrity in lipid bilayers constructed from a lipid mixture (brain
total lipid extract). Understanding the function and interactions of these proteins in cellular environments might
expand the development of new diagnostic and therapeutic approaches for AD or other related diseases.

1. Introduction are maintained low, roughly 0.1–0.5 μM [19]. The extracellular free
calcium concentration typically ranges from 1.5 to 2.0 mM [20]. In the
Inflammation is an organism’s natural defensive response to harmful presence of calcium ions, S100A9 changes the conformation to expose a
stimuli [1]. Lately, it has been named as a leading cause of Alzheimer’s more hydrophobic surface to solvent [15], facilitating its translocation
disease (AD) pathogenesis and as a factor promoting the progression and into the extracellular space [21]. When the extracellular S100 proteins
severity of this illness [2,3]. Notably, a pro-inflammatory S100A9 pro­ have all calcium-binding sites occupied, the protein’s structure is sta­
tein has been a subject of high interest [4–7]. This calcium-binding bilized [15], and interactions with target proteins or membranes are
protein contains two helix-loop-helix (EF-hands) motifs and belongs to promoted [17]. Under the cell resting conditions, when the cytosolic
the S100 protein family. The protein is permanently expressed in free calcium concentration is maintained at the nanomolar level, S100
monocytes, neutrophils, dendritic cells, and other cells as macrophages proteins occupy a closed, relatively hydrophilic conformation [16].
upon activation [8–10]. Furthermore, S100A9 is highly expressed and The structural changes of S100 proteins upon binding calcium are
plays a prognostic role in various cancers, including lung [11], gastric reflected in the affinity toward lipid bilayers. One example is S100D, of
[12], and others [13,14]. The S100A9 proteins are calcium sensors which calcium-free form is associated with detergent dodecyl phos­
changing their conformation in response to calcium influx [15,16]; this phocholine (DPC) micelles, but site-specific binding with individual DPC
mediates interactions with membrane-associated molecular targets molecules occurs when Ca2+ ions are bound [22]. Another study showed
[17,18]. Intracellular free Ca2+ concentration widely varies depending that both forms of S100A12 protein interact with both zwitterionic and
on its location. Typically, in the mammalian brain, the concentrations negatively-charged membranes [23]. The interactions with ions and

* Corresponding author.
E-mail address: marija.jankunec@gmc.vu.lt (M. Jankunec).

https://doi.org/10.1016/j.bbamem.2022.184113
Received 31 May 2022; Received in revised form 6 December 2022; Accepted 16 December 2022
Available online 23 December 2022
0005-2736/© 2022 Elsevier B.V. All rights reserved.
R. Tamulytė et al. BBA - Biomembranes 1865 (2023) 184113

liposomes might work as a molecular triggering mechanism for the ~13.2 kDa) was dissolved in PBS buffer at pH 7.2, with all solutions
protein to adopt adequate conformation and to find its targets on the remaining on ice. No further actions were conducted before the analysis.
membrane surface. Increasing evidence shows that protein interaction Lipid compositions used in this work: DOPC/Chol 60/40 mol% —
with cell membranes plays a critical role in the accumulation and zwitterionic lamellar phase; DOPC/DOPS 60/40 mol% — negatively
growth of various amyloidogenic proteins [24,25]. The heterogeneous charged anionic liquid-disordered (Ld) phase; DOPC/DPPC 60/40 mol%
aggregation promoted by protein crowding on the membrane surface — Ld/gel phase (diphasic system); neuronal membrane biomimetic,
increases the local concentration of proteins and shapes the lipid lipid raft-like system — DOPC/DOPE/bSM/Chol/GM1 (44/20/30/5/1
membrane [26]. Since S100A9 possesses amyloid-like properties [27], mol%) and brain total lipid extract.
the lipid bilayer’s closeness might affect the protein’s aggregation rates.
However, the data for this protein is lacking in the literature. One study
showed that the mature S100A9 aggregates were co-localized with 2.2. Methods
monosialotetrahexosylganglioside-1 (GM1) in neuroblastoma SH-SY-5Y
cells in vitro [28]. These data support the hypothesis that the binding 2.2.1. Preparation of multilamellar and unilamellar vesicles
sites of S100A9 amyloids to the cell membrane occur at the raft domains. Initial lipid mixtures were dissolved in chloroform and dried under
These observations also suggest the critical role of GM1 in protein- N2 for 40 min. After solvent traces removal and formation of thin lipid
membrane association, most likely due to the clusters of sialic acid film, lipids (the final concentration of 1 mM) were rehydrated: i) in
moieties, producing negatively charged regions at the cell surface. In saline solution (pH 4.5) to obtain multilamellar vesicles for AFM ex­
another study, the formation of large aggregates of the S100A8/S100A9 periments; and ii) in PBS-buffer, subjected five times to freeze-thaw
complex upon incubation with the model phosphatidylcholine/choles­ cycles, ultrasonicated for 1 h and subsequently extruded 21 times
terol lipid bilayer was observed [29]. This complex translocates into a through 100 nm polycarbonate membranes with a mini-extruder
detergent-resistant lipid structure only after calcium activation of the (Avanti Polar Lipids Inc., USA) to obtain unilamellar vesicles (LUVs).
neutrophils. However, the S100A9/A8 membrane association is The extrusion was done above the lipids’ melting temperature. The LUVs
cholesterol and sphingolipid-independent, not restricted to lipid rafts size distribution was determined by dynamic light scattering using a
[18]. Malvern Zetasizer μV apparatus (Malvern Panalytical, UK). Z-average ±
This work aims to study the affinity of the apo-form (calcium-free) of SEM calculated based on size distribution per mass; n = 3, the number of
pro-inflammatory S100A9 protein toward lipid bilayers. The surface runs per sample is 4.
sensitive-technique atomic force microscopy (AFM), dynamic light
scattering (DLS), and Thioflavin-T (ThT) fluorescence assay allowed us 2.2.2. Laurdan spectroscopy
to study the accumulation and aggregation of S100A9 protein and Prepared LUVs were mixed with Laurdan (dissolved in dimethyl
morphological changes in the affected lipid bilayer surface. Two cell sulphoxide) to give a final probe-to-lipid ratio of 1/200. Samples were
membrane biomimetic platforms were used. Tethered bilayer lipid pre-incubated 1 h before fluorescence measurements on ClarioStar Plus
membranes (tBLMs) were used as a model on a solid substrate, while (BMG Labtech, Germany) plate reader. Before measurements, liposome
unilamellar liposomes represented the lipid bilayer in bulk solution. The suspensions were allowed to equilibrate for 10 min at each temperature.
initial stage of protein-lipid interactions is influenced by electrostatic Laurdan emission spectra were recorded from 400 to 600 nm using a
interaction between hydrophilic parts of protein and the charge of the 340 nm excitation wavelength. The excitation and emission band-pass
headgroup of lipids. The fluidity of the lipid bilayer could be modulated bandwidth was 10 nm. The generalized polarization (GP) quantifies
by including sterols and saturated and unsaturated backbone lipids. the Laurdan emission spectrum changes. In analogy with fluorescence
Thus our chosen lipid membranes contain 60 mol% of unsaturated lipid polarization:
dioleoylphosphatidylcholine (DOPC). The other lipids with a negatively I440 − I490
charged headgroup (dioleoylphosphatidylserine, DOPS) and saturated GP = , (1)
I440 + I490
acyl chains (dipalmitoylphosphatidylcholine, DPPC), or cholesterol,
would cover the rest 40 mol%. Next, lipid bilayers composed of brain where I440 and I490 are the fluorescence intensities registered at the
total lipid extract combine both (fluidity and negative charge) properties maximum emission wavelength in the ordered (440 nm) and disordered
in one. Additionally, we used more complex lipid compositions to mimic (490 nm) phases [32]. All experiments were performed at least three
the neuronal membrane — DOPC/dioleoylphosphatidylethanolamine times with triplicate assays.
(DOPE)/Chol/brain sphingomyelin (bSM)/GM1 [30]. Here, for the
latter composition, we expect the formation of lipid-raft-like domains 2.2.3. Formation of tethered lipid bilayer membranes (tBLMs)
that are enriched with cholesterol, sphingomyelin, and, in particular, Freshly cleaved mica (grade IV, SPI Supplies, USA) sheets with
GM1 [31]. annealed gold layers were prepared by magnetron sputtering using a
PVD75 system (Kurt J. Lesker Co., USA). The typical thickness of the
2. Experimental gold layer was 100 nm, while the adhesive Cr layer was 10 nm. Then,
modification with self-assembled monolayers (SAMs) via incubation in
2.1. Materials the ethanolic solution of WC14/βME (20-tetradecyloxy-
3,6,9,12,15,18,22 heptaoxahexatricontane-1-thiol/β-mercaptoethanol,
All lipids were purchased from Avanti Polar Lipids (Alabaster, USA), in the molar ratio 35/65, the total thiol concentration 0.05 mM) for 12 h.
including 1,2-dioleoyl-sn-glycero-3-phosphocholine (DOPC), 1,2-dio­ The details on the synthesis and properties of the self-assembled
leoyl-sn-glycero-3-phospho-L-serine (sodium salt) (DOPS), 1,2-dipalmi­ monolayers used in this work are described elsewhere [33]. After in­
toyl-sn-glycero-3-phosphocholine (DPPC), GM1 ganglioside (brain, cubation, the samples were rinsed with ethanol, dried under a nitrogen
ovine‑sodium salt), brain sphingomyelin (bSM, porcine), brain total stream, and stored dry until further use. Tethered lipid bilayers (tBLMs)
lipid extract (BTLE, porcine, composition by weight: PC (9.6 %), PE were formed via incubation of multilamellar vesicles (0.1 M NaCl, 0.01
(16.9 %), PS (10.6 %), PI (1.6 %), PA (2.8 %), unknown 58.7 %), 1,2- M NaH2PO4, pH 4.5) of various lipid compositions for 0.5 h on SAMs-
dioleoyl-sn-glycero-3-phosphoethanolamine (DOPE). PBS buffer (pH covered mica substrates [34]. The incubation was done above the
7.2) containing 130 mM NaCl, 2.7 mM KCl, 10 mM Na2HPO4, and 1.8 melting temperature of the lipids at 50 ◦ C. After tBLMs formation,
mM KH2PO4 (Sigma Aldrich, Hamburg, Germany). The saline solution samples were thoroughly rinsed with lipid-free PBS buffer (pH 7.2).
containing 0.1 M NaCl, 0.01 M NaH2PO4 (Roth, Frederikssund, Fig. S6 (see Supplementary Materials) shows a schematic view of tBLMs
Denmark), pH 4.5. S100A9 protein (1.9 mg/ml, molecular weight is constructs.

2
R. Tamulytė et al. BBA - Biomembranes 1865 (2023) 184113

2.2.4. Expression and purification of recombinant S100A9 protein scan speeds of 0.3–0.5 kHz. After pre-visualization, tBLMs were exposed
The plasmid containing human S100A9 coding DNA (construct to S100A9 solution in PBS (pH 7.2) (the final protein concentration was
described previously [35]) was chemically transformed into E. coli BL21 5 μM; 37 μl of protein stock solution was added into AFM liquid cell
(DE3) one-star strain. A single colony from transformed cells was filled with 2 ml buffer) and imaged immediately. As the injection is done
selected and grown overnight in 100 ml LB medium supplemented with locally near the scanning area, the local concentration of peptides may
100 mg/ml ampicillin at 37 ◦ C. The cells were grown until the culture vary due to uneven distribution in the fluid cell. The lag time between
reached an optical density of 0.6 at 600 nm, and then 3 ml of this culture adding the protein to the tBLM and starting the first image acquisition is
was transferred to 300 ml of ZYM-5052 auto-inductive medium with at least 10 min. Membrane imaging was conducted at ambient room
ampicillin (100 μg/ml) for protein expression. The culture was grown temperature in a liquid cell in PBS, pH 7.2, with hydration of the
overnight at 37 ◦ C, 220 RPM; cells were harvested by centrifugation at membrane maintained throughout imaging. Successive AFM images
5000 ×g for 15 min at 4 ◦ C and frozen at − 20 ◦ C. were recorded for several hours at the exact area unless otherwise
Before the purification, the cells pellets were thawed and resus­ stated. Tethered lipid bilayers (tBLMs) formed from respective multi­
pended in 20 mM Tris buffer (pH 8.0) with 0.5 mM EDTA and 0.5 mM lamellar vesicles on the SAM-covered gold-coated mica sheets are me­
DTT buffer. Homogenized and sonicated for 20 min (50 % amplitude, chanically stable over a long period. The average image acquisition time
cycles of the 30 s ON, 30 s OFF). The pellet was collected by centrifu­ is 20 min. Thus the first complete AFM image is recorded 30 min after
gation for 25 min at 20,000 ×g, 4 ◦ C, and the supernatant was discarded. protein injection. All AFM measurements were performed at room
The washed pellets containing S100A9 protein were dissolved in 8 M temperature (22 ◦ C). First, we imaged a 5 μm × 5 μm area for at least an
urea solution and refolded at 4 ◦ C via dialysis against 20 mM Tris buffer, hour to verify the stability of the prepared bilayer and that no defects
containing 0.5 mM EDTA and 0.5 mM DTT, pH 8.0. The dialysis buffer were present within the deposited membrane. Each time, the lipid
was changed 4 times during 16 h of refolding. After refolding, the pre­ bilayer formation was confirmed by a ramp to obtain force curves.
cipitant was removed by centrifugation for 25 min (20,000 ×g, 4 ◦ C). The image’s surface roughness is expressed by the arithmetic average
The supernatant was mixed with a weak anion exchange chromatog­ (Image Ra) value. Image Ra was assigned to each represented AFM image
raphy resin (DEAE Sepharose Fast Flow, GE Healthcare) and loaded on a in this work to show changes in a particular area of interest. Image Ra is
column. S100A9 was eluted when a linear gradient reached 0.25 mM an average of the absolute values of the surface height deviations
NaCl concentration. The fractions containing S100A9 protein were measured from the mean plane:
concentrated to 7 mg/ml using Amicon®Ultra-15 centrifugal filters
1 ∑N ⃒⃒ ⃒⃒
(Millipore) with 10 kDa molecular weight cut-off (MWCO) membranes Ra = Zj . (2)
N j=1
and purified further by the size exclusion chromatography (Superdex 75,
GE Healthcare) in PBS. The fractions with S100A9 protein were ali­ Multiple images of different scanning areas were taken for each
quoted to tubes by 0.5 ml and stored at − 80 ◦ C freezer. sample. The images were flattened and analyzed using NanoScope
Analysis 1.9 (Bruker, USA) and WSxM 4.0 [37] software.
2.2.5. S100A9 aggregation and ThT fluorescence assay
Fluorescent dye Thioflavin-T (ThT) (Sigma-Aldrich, Germany) binds 2.3. Statistical analysis
to the β-sheet containing amyloid structures [36]; this leads to an in­
crease in its fluorescence and thus enables the kinetics of amyloid for­ Statistical analyses and the Gauss distribution were performed using
mation to be followed and quantified. ThT was dissolved in PBS buffer GraphPad Prism version 8.4.3 for Windows (GraphPad Software, USA),
and filtered through a 0.2 μm syringe filter. Then the concentration of and data were analyzed using a t-test. Significance was ascribed at P <
ThT was determined using an extinction coefficient of 36 mM− 1 cm− 1 at 0.05. The data of different parameters were represented as the means ±
412 nm. The aggregation of 50 μM S100A9 in PBS (pH 7.2, containing standard deviation if stated otherwise.
100 μM ThT (protein/ThT ratio of 1/2)) in the absence or presence of the
lipid membrane of various compositions (LUVs concentration was 500 3. Results and discussion
μM) was monitored by ThT fluorescence in 96-well microplate (Corning
Inc., USA) on a reader ClarioStar Plus (BMG Labtech, Germany). The This study aimed to assess calcium-unrelated effects and eliminate
volume of the reaction well was 100 μl. The ThT fluorescence intensity factors that impact the properties of the lipid bilayer to highlight only
of each sample was recorded every 5 min with 440 nm excitation and protein-driven interactions. Calcium ions interact with lipids’ head­
490 nm emission filters set at 37 ◦ C. The excitation and emission band- groups, causing dehydration of the phosphate [38], changing the
pass bandwidth was 10 nm. The sample was constantly shaking at 400 thickness [39], and increasing the ordering [40] of phospholipid bi­
rpm between measurements. layers. It is known that the binding of Ca2+ ions with negatively charged
lipids (ex., phosphatidylserine) and hydration effect create local changes
2.2.6. AFM Imaging in the charge via the formation of negatively and positively charged
For AFM sample preparation in the air, the S100A9 stock solution patches [41] or induce phase separation [42]. All these factors could
was diluted to 100 nM or 5 μM with PBS (pH 7.2), and 20 μl of the so­ impact the lipid bilayer’s permeability or cause deformation not
lution was placed onto a freshly cleaved mica (grade 4, SPI Supplies, detectable by atomic force microscopy (AFM) or fluorescence tech­
USA) surface. After 2 min incubation, the sample was rinsed several niques. These local deviations in the integrity of the lipid bilayer could
times with MilliQ-H2O and dried with a nitrogen airflow. AFM images be primary targets of protein binding.
were obtained using Dimension Icon (Bruker, USA) AFM, operating in
tapping mode. For imaging in the air, tapping mode silicon nitride 3.1. The morphology of initial protein preparation
probes FESP (Bruker, USA) with a nominal spring constant of 2.8 N/m
and resonance frequency of 75 kHz were used, nominal tip radius of 8 Characterization of freshly thawed S100A9 protein (114 amino
nm. Height images were collected (512 × 512 pixels) at scan speeds of acids) was assessed by AFM, allowing us to confirm the monomeric/
0.6–0.8 kHz. oligomeric state of the protein at the starting point of the protein-
For imaging of tBLMs, tapping-mode or PeakForce QNM were used in membrane interaction experiment. Fig. 1a shows an AFM topography
PBS buffer at room temperature. AFM imaging in the fluid was per­ image of the freshly thawed S100A9 protein solution (5 μM) adsorbed on
formed using triangular silicon nitride probes SNL-C (Bruker, USA) with freshly cleaved mica and imaged in the air. The height and diameter
a nominal spring constant of 0.24 N/m and resonance frequency of 56 distributions (Fig. 1b and c panels) show the presence of particles with a
kHz in air. Height and phase images were collected (512 × 512 pixels) at mixed size distribution, preferentially observed with sizes ranging from

3
R. Tamulytė et al. BBA - Biomembranes 1865 (2023) 184113

Fig. 1. The morphology of the freshly dissolved S100A9 deposited on freshly cleaved mica. (a) AFM topography image visualized in air. The protein concentration is
5 μM. The scan size is 1 μm × 1 μm. Gauss distributions for height (b) and diameter (c) are 0.89 ± 0.23 nm (n = 649) and 25.4 ± 3.7 nm (n = 681), respectively. The
estimated protein’s hydrodynamic radius in solution by DLS (d).

15 to 45 nm in diameter and 0.6–1.5 nm in height. The height and binding with cell membranes. Moreover, the lipid bilayer’s biophysical
diameter Gauss distributions (GraphPad Software, Inc.) are 0.89 ± 0.23 characteristics impact protein accumulation and its ability to penetrate
nm and 25.4 ± 3.7 nm, respectively. Fig. 1S shows a close look at in­ the lipid bilayer. Three of the five lipid compositions used in this work
dividual proteins and their profiles. Here we have used a geometrical are well-characterized membrane models: i) at room temperature, the
method to estimate the protein’s volume. The volume was calculated by binary lipid system of zwitterionic DOPC and cholesterol (60/40 mol%)
considering the protein as a spherical cap. The height and the full width based on X-ray [46] and NMR [47] data is in a liquid-ordered phase; ii)
at half maximum were measured via cross-section analysis [43]. The the zwitterionic/anionic DOPC/DOPS (60/40 mol%) system consists of
dimensions of a single S100A9 monomer structure were approximately two unsaturated lipids of the same acyl chain, with phase transition
7.90–13.62 nm wide and 0.56–0.71 nm high, giving an experimental temperatures − 17 and − 11 ◦ C [48], respectively, forms liquid-
molecular volume for the monomer as 26–31 nm3. These results agree disordered (Ld) phase; and iii) DOPC/DPPC (60/40 mol%) liposomes,
favorably with the calculated molecular volume of the S100A9 mono­ whereas DOPC is unsaturated and DPPC is a saturated lipid, form phase-
mer of 25 nm3 and are in line with previously reported values [44]. The separated regions of Ld and gel (Lβ) phases [49]. Nevertheless, there is a
tip convolution was not taken into account. Based on the assumption lack of data for complex BTLE and five-component lipid bilayers. To
that the S100A9 monomer’s height is 0.50–0.75 nm, we conclude that address these issues, we used the Laurdan spectra to identify the mem­
24.8 % of observed protein particles are monomers. brane lipid packing. Laurdan fluorescence emission spectra and gener­
AFM in the air visualized the morphology of adsorbed molecules; alized polarization (GP) values for all well-characterized compositions
however, due to adsorption and surface-induced aggregation, molecules are available in Supplementary Materials Fig. S7.
tend to deform. Therefore, dynamic light scattering (DLS) data repre­ Fig. 2 displays the evaluation of the lipid packing of BTLE and five-
sents average size distribution in bulk. The estimated hydrodynamic component (DOPC/DOPE/Chol/bSM/GM1) large unilamellar vesicles
radius of S100A9 protein is 2.4 ± 0.5 nm (see Table 1, Fig. 1d), and (LUVs) monitored at 22, 37, and 60 ◦ C by the Laurdan fluorescence
polydispersity is 0.22. The obtained particle size is close to the theo­ assay. After excitation at 390 nm, the values for the solid and fluid
retical value of 2.1 nm for folded (globular) protein based on its mo­ phases were recorded at 440 nm and 490 nm, respectively. The wave­
lecular weight [45]. We conclude that derivatives of low lengths of interest are marked in Fig. 2a with dash lines. To compare the
oligomerization dominate the initial protein stock solution. These data lipid packing/order of these bilayers, we measured the GP values ac­
allow us to understand what structures might populate the membrane’s cording to Eq. 1. The measured mean GP value for BTLE composition at
surface. All protein-lipid membrane interaction measurements were room temperature is 0.37 ± 0.01 (n = 4), which decreases to 0.32 ±
performed on a rough gold-coated surface. Therefore, very small olig­ 0.02 at 37 ◦ C and 0.18 ± 0.01 when liposomes are heated up to 60 ◦ C.
omers may be overlooked. The registered values are the highest among studied compositions,
indicating that prepared BTLE liposomes possibly are in a solid state.
The GP value for the five-component system is 0.10 ± 0.02 at room
3.2. Determination of membrane lipid packing temperature, which slightly decreases to 0.04 ± 0.01 and − 0.07 ± 0.02
at 37 and 60 ◦ C, respectively. The obtained GP value for the five-
Disordered proteins can undergo structural rearrangements upon component liposomes is higher than for the homogeneous Ld phase
(DOPC/DOPS) but lower than measured for coexisting Ld-Lβ system
Table 1 (DOPC/DPPC).
DLS data of liposomes incubated with S100A9 protein for 24 h, n = 3. The reduction in lipid packing with temperature increase indicates
Time Composition Zaverage ± SEM (nm) the transition from a solid to a fluid state. However, the GP is an average
value, and the coexistence of different phases might be misinterpreted
0h S100A9 alone 2.4 ± 0.4
BTLE 127.4 ± 30.2
[50]. Bagatolli et al. [51] showed that Laurdan distribution in the lipid
DOPC/Chol 133.8 ± 21.0 membranes is heterogeneous, whereas dye molecules partition prefer­
24 h, 37 ◦ C S100A9 alone 43.8 ± 4.6 entially into one of the phases of the lipid mixtures. Also, Laurdan
S100A9 + BTLE 137.4 ± 11.8 fluorescence behavior is less sensitive toward GM1-containing compo­
S100A9 + DOPC/Chol 135.6 ± 12.6
sitions due to the complexity of the polar head group [51].

4
R. Tamulytė et al. BBA - Biomembranes 1865 (2023) 184113

Fig. 2. The normalized Laurdan fluorescence emissions of LUVs recorded at 22, 37, and 60 ◦ C temperature: (a) brain total lipid extract; (b) DOPC/DOPE/bSM/Chol/
GM1 (44/20/30/5/1 mol%); and (c) the generalized polarization (GP). The Laurdan probe to lipid ratio was 1/200.

3.3. S100A9 self-assembly in the presence of liposomes We would expect that the size of the liposome should be affected by
the protein’s binding. The DLS data (see Supplementary materials
The ability of S100A9 to self-assemble into fibrils may lead to the loss Fig. S8 and Table S2) indicate that liposomes in a protein-free envi­
of signaling functions and acquired amyloid cytotoxicity [2]. The ronment remain stable over 72 h. An exception is the DOPC/DPPC
Thioflavin-T (ThT) assay is a standard method to quantify the growth of composition, for which a steady state was reached after the increase of
fibrillary aggregates during amyloid formation [36,52]. ThT, as an the diameter of liposomes by 36 nm in the first 24 h. However, we
extrinsic fluorescent probe molecule, once bound to the protein’s observe a 20-fold increase in the protein’s hydrodynamic diameter,
β-structure, gives a characteristic fluorescent emission at ~480 nm (λexc while changes in BTLE and DOPC/Chol liposome size are insignificant
440 nm). The ThT intensity is related to the growth and amount of (see Table 1). The surface presence should increase local concentration,
fibrillar structures. Fig. 3 displays the increase in fluorescence intensity promote binding of the material, and further accumulation. Important to
of S100A9 bound-ThT over time in the presence or absence of liposomes. mention that factors such as curvature, lipid phase, and electrostatic
We incubated 100 μM of S100A9 protein w/o of 500 μM unilamellar interaction are also involved in this process. However, a lower number
liposomes at 37 ◦ C pH 7.2 under orbital agitation to evaluate if the of S100A9 aggregates in the presence of liposomes reveals that the
presence of the lipid surface promotes the protein accumulation. The nucleation process is disturbed. Likely, the binding to the lipid surface is
high protein concentrations were needed to accelerate the slow aggre­ limited due to the hydrophilic surface of S100A9 in a calcium-free
gation process. Still, the protein concentration was probably insufficient environment.
as the plateau was not reached within our experiment’s timeframe (75 The observed (Fig. 3) short and not-well-defined lag phase represents
h). the time required for the nuclei formation. Our results agree with pub­
The changes in ThT fluorescence intensity indicate a slight depen­ lished reports that in vitro S100A9 protein alone self-assembles into
dence of the number of formed fibrillary structures on the lipid amyloids by the nucleation-dependent polymerization mechanism
composition. The sequence from the slowest to the fastest aggregation is [53,54]. This unusual behavior for other amyloid-like proteins indicates
DOPC/DOPS < DOPC/DOPE/Chol/bSM/GM1; DOPC/DPPC; DOPC/ that nucleation sites are small and fast-forming. In comparison to the
Chol < S100A9 alone < BTLE. At the initial stages (first 24 h) of accu­ S100A9 published data [53,54], our measurements were performed
mulation, all lipid compositions had lower ThT intensity than protein under similar physiological conditions: PBS buffer (pH 7.2 vs. 7.4),
alone. However, after 35 h of incubation, the intensity data for BTLE temperature (37 vs. 42 ◦ C), and calcium-free environment. The main
liposomes increased significantly and stood out from the rest composi­ difference is that we promoted aggregation by external agitation. It
tions. While BTLE liposomes promoted protein aggregation, the aggre­ should be noted that calcium ions significantly slow down the S100A9
gation of S100A9 in the presence of negatively charged liposomes was oligomerization [55] (see Supplementary materials Fig. S9).
reduced. Meanwhile, zwitterionic, two-phase, and five-component lipid
compositions slightly inhibit fibrillation compared to the S100A9 pro­
tein alone.

Fig. 3. Monitored aggregation of S100A9 (100 μM)


by Thioflavin-T (ThT) fluorescence: (a) for S100A9
protein alone (black) and in the presence of lipid
vesicles composed of DOPC/Chol (blue), DOPC/
DOPS (cyan), DOPC/DPPC (purple), DOPC/DOPE/
bSM/Chol/GM1 (red) and BTLE (green); monitoring
time 75 h; (b) enlarged inset from the (a) panel;
monitoring time 24 h. Experimental details: PBS
buffer (pH 7.2), at 37 ◦ C, w/o 500 μM of lipid. Con­
trol: protein samples without lipids but under other­
wise identical conditions. Each dot represents the
average ± S.D. from triplicate measurements.

5
R. Tamulytė et al. BBA - Biomembranes 1865 (2023) 184113

3.4. Visualization of S100A9-induced lipid membrane remodeling by Table 2


AFM The changes of the arithmetic average roughness (Image Ra) of repre­
sentative AFM images of tBLMs incubated with S100A9.
Our ThT fluorescence data indicates that the zwitterionic DOPC/ Sample Ra, nm
cholesterol 60/40 mol% lipid bilayer does not affect or slightly inhibit DOPC/Chol (Fig. 4) 0.35
the aggregation. Due to high cholesterol content and membrane local + S100A9 (0.5 h) –
stiffening [56], tethered bilayer lipid membranes (tBLMs) would form a + S100A9 (2 h) 0.35
lamellar liquid-ordered (Lo) phase [46,47]. The influence of membrane DOPC/DOPS (Fig. 4) 0.38
+ S100A9 (0.5 h)
fluidity on lipid-protein interaction is the object of interest as mem­ –
+ S100A9 (2 h) 0.40
branes’ fluidity decreases with cells’ age [57]. Fig. 4a–c panels show a DOPC/DPPC (Fig. 6) 0.46
series of AFM topographic images of prepared tBLMs recorded before + S100A9 (0.5 h) 0.47
and after 2 and 24 h incubation with S100A9 protein. No significant + S100A9 (2 h) 0.54
morphology changes in the lipid layer were observed even after incu­ DOPC/DOPE/Chol/bSM/GM1 (Fig. 7) 0.91
+ S100A9 (0.5 h) 0.75
bation for 24 h (Fig. 4c). Probably due to too low protein concentration, + S100A9 (1 h) 0.68
slower aggregation at room temperature, or localized accumulation
overlooked by AFM imaging. The roughness value (Image Ra) for the
area imaged in Fig. 4a is 0.35 nm and remains constant over 2 h negatively charged surface. However, the driving force of this effect is a
(Table 2). lower surface pH compared to the bulk solution. The attraction of
Next, we prepared lipid bilayers with negatively charged phospha­ counterions as H+ could lower the local pH by up to 2-fold [59,60]. The
tidylserine (PS). PS lipid is actively confined to reside solely in the inner local decrease in pH was assumed to denature the native conformation
cytoplasmic leaflet of the plasma membrane lipid bilayer. It is only of proteins (reducing solubility), inducing the formation of acidic
transferred to the outer lipid layer in conditions like apoptosis and molten globule states and their interaction with or insertion into
cancer [58]. Fig. 4d–f panels show AFM topographic images of tBLMs membranes [60,61]. These results contradict ThT fluorescence data, as
formed from MLVs of DOPC/DOPS (60/40 mol%). In contrast to DOPC/DOPS liposomes had the lowest intensity values and the number
cholesterol, PS accelerates the accumulation of S100A9 protein on of aggregates suggesting that protein might be spontaneously incorpo­
membrane surfaces, as seen in the AFM images (Fig. 4f and g panels). rated into the bilayer [62]. Therefore, protein is no more accessible for
After 24 h of incubation in the presence of S100A9, the visible white nucleation. In this study, it was impossible to distinguish whether it was
blobs indicate globular aggregates of 4–10 nm height (Fig. 4h). The a partial or complete insertion. At room temperature, DOPC/DOPS li­
calculated pI value for S100A9 protein is 5.7 and at pH 7.2 has a posomes are in a liquid-disordered phase [48]; in comparison, DOPC/
negative net charge of − 6.11, suggesting no possible interactions with a Chol liposomes are in a liquid-ordered phase. Nevertheless, the

Fig. 4. AFM topography of pristine tBLMs and incubated with 5 μM of S100A9. Lipid bilayers were formed from DOPC/Chol 60/40 mol% (a) and DOPC/DOPS 60/
40 mol% (d) multilamellar vesicles. AFM images show changes in topography following incubation for 2 h (b and e) and 24 h (c and f), respectively. The scale bar is
400 nm. In a magnified view (g) from (f), the scale bar is 140 nm. The (h) panel represents cross-section profiles (1 and 2).

6
R. Tamulytė et al. BBA - Biomembranes 1865 (2023) 184113

morphology observable in the AFM images showed that these two- distal leaflet. Since we observed that the initial depth increases from
component liposomes form homogeneous tethered lipid bilayers. Thus, 1.31 ± 0.23 nm to 2.2 ± 0.29 nm, it shows the progression of membrane
the differences observed between these two compositions might also be compression [67] by increasing the protein’s local concentration over
related to the fluidity of the lipid bilayer and reduced barrier between time. Here an evident lateral expansion of lipids, with further insertion
the acyl chains, which presumably affects initial protein binding. into the membrane, is associated with a decrease in the thickness of the
Cell membranes have highly heterogeneous lipid compositions. To bilayers. The significant increase in protein aggregates in the presence of
better simulate the healthy neuronal membrane, we used a membrane brain total lipid extract liposomes compared to pure S100A9 solution
comprised of natural brain total lipid extracts (BTLE), consisting of over (Fig. 3) is somehow related to these drastic changes in the bilayer
58 % unidentified lipids. The registered GP value at room temperature integrity.
for BTLE is slightly higher than for DOPC/cholesterol (see Supplemen­ In eukaryotic cells, fluidity across the membrane is not homogenous.
tary materials Fig. S7); we assume BTLE liposomes have a similar lipid Lipid mixtures with different melting temperatures induce the formation
packing. Therefore, Laurdan fluorescence and AFM data (Fig. 5a) indi­ of localized lipid packing [68] and mimic liquid-ordered (Lo) and
cate that the lipid bilayer formed from the extract is in a monophasic disordered (Ld) regions on a solid support lipid bilayer. Domains of
liquid-ordered phase. Also, 15 wt% of the extract’s lipids have anionic different fluidity show reduced or increased thickness. The cantilever
headgroups. Likely, as in the DOPC/DOPS case, the protein accumula­ tip-induced compression of the lipid bilayer in AFM topographies ob­
tion will be accelerated based on reduced local pH. The highest regis­ tained in the conventional tapping mode enhances the height difference
tered ThT fluorescence confirmed that. However, the topography of [69]. Fig. 6 shows topography and phase images of S100A9 protein
lipid BTLE bilayers showed apparent changes after adding S100A9 incubated with tBLMs, formed from DOPC and a saturated lipid DPPC
protein (Fig. 5). Following the protein incubation, spontaneous defects (60/40 mol%) MLVs. As shown in Fig. 6a, the topographic image ob­
were formed. The depressed regions’ average depth is 1.31 ± 0.23 nm tained for the protein-free bilayer revealed the coexistence of two phases
(n = 12) (Fig. 5g). Following the longer incubation times, as seen in at room temperature. The DOPC acyl chains are fluidic at room tem­
Fig. 5 e and f panels, the expansion of these reduced step heights is perature and are in the liquid-crystalline phase (Lα or Ld). The melting
observed. The depth increases up to 2.20 ± 0.29 nm (n = 12) (Fig. 5h), temperature of acyl chains for saturated lipid DPPC is 41 ◦ C. Thus, the
and for a better overview, the large-scale AFM images are presented in higher areas are assigned to DPPC-enriched domains at room tempera­
Fig. S4 (see Supplementary materials). The thickness of the initial ture and are in the gel state (Lβ or So) [49]. The differences between
bilayer is approx. 5.2–5.3 nm (Fig. S3), in agreement with previous re­ these two phases are also visible in the phase images (Fig. 6, e–h panels),
ports for BTLE-supported lipid bilayers (SLBs) [63]. The depth of formed where the stiffer domains are brighter than the surrounding. The
pits is less than one leaflet of a lipid bilayer. Thus some structural observed step height difference between phases of different mechanical
changes and the formation of compressed layers occurs. A similar properties is 1.04 ± 0.30 nm (n = 11), comparable to the confirmed
decrease in depth by ~1–2 nm was reported for α-synuclein [64,65] and height difference from AFM data for SLBs (0.8–1.1 nm) [70,71]. Before
antimicrobial peptide melittin [66], as well as alcohols [67] incubated the protein’s addition, the area of interest was scanned for at least an
with lipid bilayers. The cause is either the removal of the upper leaflet by hour to confirm the layer’s morphological stability (see Fig. S10).
a detergent-like effect or thinning of acyl lipid chains and the formation S100A9 protein induced a slow disappearance of the gel phase domains
of interdigitated domains caused by peptide binding. Removal of a distal (brighter regions in AFM image), which was completed after 2 h. The
leaflet and exposure of hydrophobic tails into the water environment is erosion of Lβ domains started from the edges as it was more exposed to
energetically unfavorable and is stabilized by protein coverage of the environment.
exposed lipid acyl chains. However, the formation of detectable protein Interestingly, no further morphological alterations (i.e., holes) were
blobs within compressed valleys was not observed. More likely, S100A9 observed apart from the slow disappearance of these protrusions.
induces local membrane thinning but not the removal of lipids from the Similar lipid bilayer remodeling was observed for the dicationic

Fig. 5. Topographic investigation of BTLE lipid bilayer before (a) and after injection of 5 μM of S100A9. The protein causes the formation of defects (indicated by
arrows in the b panel). Different locations of the same sample were visualized after 0.5 h (b, c), 1 h (d), and 2 h (e and its magnified view in the f panel, the scale bar is
140 nm) of incubation. Profiles in the g and h panels are from the b and f panels, respectively. The image size is 2 μm. The Z-range is 10 nm.

7
R. Tamulytė et al. BBA - Biomembranes 1865 (2023) 184113

Fig. 6. AFM topography (upper panels) and phase (lower panels) images of phase-separated lipid bilayers. Time-lapse images represent bilayer changes after
exposure to S100A9 for 2 h: (a and e) intact tBLM was formed from MLVs DOPC/DPPC 60/40 mol%; (b and f) after 0.5 h; (c and g) after 1 h, (d and h) after 2 h. The
scan size is 3 μm × 3 μm. In AFM topography images, the color scale indicates the height of the observed structures. In the chosen color scheme, the brighter areas
(emerald green) show the protrusion of higher objects. The DOPC-enriched regions are marked as Ld, and the DPPC-rich domains are Lβ.

antibiotic azithromycin [72] or surfactin [71]. This effect was attributed ganglioside (GM1) [27]. Thus, we chose to imitate the outer layer of the
to the disruption of the tight molecular packing of the DPPC molecules. neuronal plasma membrane [30] by forming tBLMs from DOPC/DOPE/
The erosion of the gel phase is reflected in the surface roughness bSM/Chol/GM1 44/20/30/5/1 mol% multilamellar vesicles. This
decrease (Table 2). However, compared to monophasic DOPC/Chol or composition at room temperature should spontaneously lead to liquid-
DOPC/DOPS tBLMs, the Image Ra value for Fig. 6d is higher, 0.35–0.38 ordered–disordered phase separation, with ordered domains enriched in
and 0.46, respectively. In the enlarged views (Supplementary materials brain SM, cholesterol, and GM1 [73]. Fig. 7a represents the AFM
Fig. S11), we observe that after protein addition, not only do the DPPC topography image of two coexisting immiscible areas. The darker
domains tend to dissolve, but small fragments populate the lower lipid brownish areas are assigned to fluid phosphatidylcholine (Ld), and the
bilayer. However, the observed morphological changes were not brighter ones are bSM-cholesterol-GM1-enriched solid domains (Lo).
assigned as lipid surface-bound proteins. Due to the rough tBLMs sur­ The formed domains are higher than the surrounding area by 2.5 ± 0.2
face, only larger structures could be appropriately distinguished. nm (n = 5). Time-scale AFM images presented in Fig. 7 show the shape
The formation of solid domains in living systems is vital for cellular and evolution of these domains over time. It is important to note that the
processes; these structures are also called rafts. It was reported that GM1 residues have a negative charge, and the height of the domains
matured S100A9 oligomers bind with one of the lipid-raft components, could be influenced by the applied force [69] and the electrostatic

Fig. 7. AFM topography and profiles of lipid raft-like domains in tBLMs formed on mica substrate. tBLMs were formed from DOPC/DOPE/bSM/Chol/GM1 44/20/
30/5/1 mol% multilamellar vesicles. AFM image (a) of an intact lipid bilayer and profile of the lipid raft domain (e). Images and profiles (b and f), (c and g), and (d
and h) show topography changes following 0.5 h, 1 h, and 24 h of incubation with S100A9 protein, respectively. Images visualized in PeakForce QNM mode. The
scale bar is 200 nm. The DOPC-enriched regions are marked as Ld, and SM/cholesterol/GM1-rich regions — as Lo.

8
R. Tamulytė et al. BBA - Biomembranes 1865 (2023) 184113

interaction of the AFM tip [74]. Also, in this work, two types of AFM observations, the phase state or charge of the membrane affects protein-
modes were used: tapping and PeakForce QNM. Due to precise force lipid interaction. These results support three models for membrane-
control, the rigid domains would appear with higher height differences mediated S100A9 toxicity: accumulation on a negatively charged sur­
in images obtained by PeakForce QNM than in tapping mode. The face, gel or liquid-ordered phase disruption, and lipid layer thinning.
similar height difference between Ld and GM1-containing Lo phases was The most drastic changes in the loss of lipid bilayer’s integrity were
observed and described before (2.2 ± 0.2 nm [75]; 2.4 ± 0.3 nm [76]). registered for brain total lipid extract lipid bilayers.
The GM1 headgroup has five residue branch oligosaccharides. Given the In conclusion, despite the visible changes in the bilayer’s
size of a glucose molecule, approx. 0.8 nm in the long axis (the average morphology, the presence of membrane did not lead to the severe pro­
carbon‑carbon bond length is 0.13 nm [77]), and the length of the GM1 motion of protein aggregation. The binding of Ca2+ ions changes the
oligosaccharide headgroup can be estimated at roughly 2.5–3.0 nm. The protein’s conformation; thus, future works should focus on more bio­
height difference between DPPC-containing or GM1-containing domains logically relevant conditions to expand the knowledge of the S100A9
and the lower surrounding areas for tBLMs is comparable to data for protein’s mechanistic role in the progression of diseases. Furthermore,
SLBs formed on smooth mica substrate [70,71,75,76]. Also, the di­ the structural data that protein undergoes upon aggregation impacts the
mensions of observed raft-like domains are smaller than those of Lβ interaction with lipid bilayer mechanism are also lacking.
domains in DOPC/DPPC tBLMs. Likely, the appearance of domains in
tBLMs, including size distribution and shape, is highly dependable on CRediT authorship contribution statement
the substrate (i.e., the impact of gold-coated mica). This issue was also
described in detail [78]; the domains formed on glass substrate R.T. carried out the experiments, analyzed the data, and wrote the
compared to structures obtained on smooth mica are smaller with original draft.
irregular boundaries. The molecular diffusion remains the same on the M.J. designed the experiment and wrote the paper.
different substrates, and the authors proposed a significant hindrance of E.J. purified the protein and reviewed the paper.
hydrodynamic lipid flow resulting in restricted domain growth [78]. Z.T. purified the protein and reviewed the paper.
Fig. 7b and c panels show the same area during the first hour of in­ V.S. formal analysis, resources.
cubation of raft-like tBLMs with S100A9 protein. After adding S100A9 All authors read and approved the final manuscript.
protein, we observed a decrease in the domains’ height by 1 nm from 2.5
± 0.2 nm to 1.5 ± 0.2 nm (n = 6). The average step height of the same Declaration of competing interest
sample after incubation with protein for 24 h was 1.5 ± 0.4 nm (n = 6)
(Fig. 7, d and h panels). Fig. S5 (see Supplementary materials) shows The authors declare that they have no known competing financial
differences between the large-scale time-lapse AFM images. We observe interests or personal relationships that could have appeared to influence
not only the reduction in height but surface roughness decreases from the work reported in this paper.
0.91 nm to 0.68 nm for pristine tBLMs and incubated with protein for 1
h, respectively. We hypothesize that these alterations indicate the sol­
Acknowledgments
ubilization of raft domains. Also, we observed that the domains’ sur­
rounding areas changed their appearance. Previously smooth Ld regions
Authors gratefully acknowledge Dr. Darius Šulskis (Institute of
became rougher due to bound protein or the formation of protein-
Biotechnology, Life Sciences Center, Vilnius University) for his help in
induced lipid-raft nanosized domains. A lipid rafts’ fragmentation was
the S100A9 protein purification, and Prof. Gintaras Valinčius (Institute
already demonstrated for human α-synuclein [79] incubated with a
of Biochemistry, Life Sciences Center, Vilnius University) for support.
reconstituted membrane bilayer and was observed for the myristoyl
We acknowledge the author of tBLMs’ schematic view (Fig. S6) — Dr.
modification of tyrosine kinase c-Src [80]. Hence, based on these results,
Tadas Penkauskas (Institute of Biochemistry, Life Sciences Center, Vil­
we speculate that the protein can recruit lipid molecules to avoid un­
nius University).
favorable hydrophobic mismatch.
Finally, in our experiments, we also applied a calcein leakage assay.
Appendix A. Supplementary data
Unfortunately, control liposomes were unstable under experimental
conditions (see Supplementary materials, Fig. S2), with at least 20 %
Supplementary data to this article can be found online at https://doi.
calcein content loss over 4 h of measurement. A similar permeability is
org/10.1016/j.bbamem.2022.184113.
observed for liposomes incubated with protein. Thus, the protein did not
promote any membrane leakage. The hydrodynamic radius of calcein is
References
0.7 nm, and only defects exceeding 1.4 nm in diameter are detectable in
calcein release measurements. [1] R. Medzhitov, Inflammation 2010: new adventures of an old flame, Cell 140 (6)
(2010) 771–776, https://doi.org/10.1016/j.cell.2010.03.006.
4. Concluding remarks [2] C. Wang, I.A. Iashchishyn, J. Pansieri, S. Nyström, O. Klementieva, J. Kara,
I. Horvath, R. Moskalenko, R. Rofougaran, G. Gouras, G.G. Kovacs, S.K. Shankar, L.
A. Morozova-Roche, S100A9-driven amyloid-neuroinflammatory cascade in
Although the role of the S100A9 protein in Alzheimer’s disease re­ traumatic brain injury as a precursor state for Alzheimer’s disease, Sci. Rep. 8
mains debated, the interaction of the protein with cellular membranes (2018) 12836, https://doi.org/10.1038/s41598-018-31141-x.
[3] I. Horvath, X. Jia, P. Johansson, C. Wang, R. Moskalenko, A. Steinau, L. Forsgren,
seems essential for both functionality and toxicity. In this study, we T. Wägberg, J. Svensson, H. Zetterberg, L.A. Morozova-Roche, Pro-inflammatory
successfully applied AFM to investigate the impact of the lipid packing, S100A9 protein as a robust biomarker differentiating early stages of cognitive
the headgroup charge, and the incorporation of GM1 in tethered lipid impairment in Alzheimer’s disease, ACS Chem. Neurosci. 7 (1) (2016) 34–39,
https://doi.org/10.1021/acschemneuro.5b00265.
bilayers on the binding and accumulation of S100A9 in a calcium-free [4] C. Zhang, Y. Liu, J. Gilthorpe, J.R.C. van der Maarel, MRP14 (S100A9) protein
environment. However, the approach used in our work has several interacts with alzheimer beta-amyloid peptide and induces its fibrillization, PLoS
limitations. AFM data is not sufficient to determine whether the ONE 7 (3) (2012), e32953, https://doi.org/10.1371/journal.pone.0032953.
[5] C. Wang, A.G. Klechikov, A.L. Charibyan, S.K.T.S. Wärmländer, J. Jarvet, L.
morphological alterations caused by protein actions affect the lipid bi­
N. Zhao, X.E. Jia, V.K. Narayana, S.K. Shankar, A. Olofsson, T. Brännström, Y.
layer’s permeability. Therefore, more sensitive techniques like electro­ G. Mu, A. Gräslund, L.A. Morozova-Roche, The role of pro-inflammatory S100A9 in
chemical impedance spectroscopy should be applied [81]. Similar Alzheimer’s disease amyloid-neuroinflammatory cascade, Acta Neuropathol. 127
topographical effects may indicate different mechanisms of protein ac­ (4) (2014) 507–522, https://doi.org/10.1007/s00401-013-12-08-4.
[6] I. Horvath, I.A. Iashchishyn, R.A. Moskalenko, C. Wang, S.K.T.S. Wärmländer,
tion. Despite these limitations, we have determined the morphological C. Wallin, A. Gräslund, G.G. Kovacs, L.A. Morozova-Roche, Co-aggregation of pro-
changes associated with exposing lipid bilayers to S100A9. Based on our inflammatory S100A9 with α-synuclein in Parkinson’s disease: ex vivo and in vitro

9
R. Tamulytė et al. BBA - Biomembranes 1865 (2023) 184113

studies, J. Neuroinflammation 15 (2018) 172, https://doi.org/10.1186/s12974- [29] S.M. Valenzuela, M. Berkahn, D.K. Martin, T. Huynh, Z. Yang, C.L. Geczy,
018-1210-9. Elucidating the structure and function of S100 proteins in membranes, Proc. SPIE
[7] M.A. Gruden, T.V. Davydova, C. Wang, V.B. Narkevich, V.G. Fomina, V.S. Kudrin, 6036 (2016), 603619, https://doi.org/10.1117/12.638873.
L.A. Morozova-Roche, R.D.E. Sewell, The misfolded pro-inflammatory protein [30] H.I. Ingólfsson, T.S. Carpenter, H. Bhatia, P.-T. Bremer, S.J. Marrink, F.
S100A9 disrupts memory via neurochemical remodelling instigating an C. Lightstone, Computational lipidomics of the neuronal plasma membrane,
Alzheimer’s disease-like cognitive deficit, Behav. Brain Res. 306 (2016) 106–116, Biophys. J. 113 (10) (2017) 2271–2280, https://doi.org/10.1016/j.
https://doi.org/10.1016/j.brr.2016.03.016. bpj.2017.10.017.
[8] J. Edgeworth, M. Gorman, R. Bennett, P. Freemont, N. Hogg, Identification of [31] C. Yuan, J. Furlong, P. Burgos, L.J. Johnston, The size of lipid rafts: an atomic force
p8,14 as a highly abundant heterodimeric calcium binding protein complex of microscopy study of ganglioside GM1 domains in sphingomyelin/DOPC/
myeloid cells, J. Biol. Chem. 266 (12) (1991) 7706–7713, https://doi.org/ Cholesterol membranes, Biophys. J. 82 (5) (2002) 2526–2535, https://doi.org/
10.1016/S0021-9258(20)89506-4. 10.1016/S0006-3495(02)75596-3.
[9] M.M. Averill, S. Barnhart, L. Becker, X. Li, J.W. Heinecke, R.C. Leboeuf, J. [32] T. Parasassi, G. De Stasio, G. Ravagnan, R.M. Rusch, E. Gratton, Quantitation of
A. Hamerman, C. Sorg, C. Kerkhoff, K.E. Bornfeldt, S100A9 differentially modifies lipid phases in phospholipid vesicles by the generalized polarization of Laurdan
phenotypic states of neutrophils, macrophages, and dendritic cells: implications for fluorescence, Biophys. J. 60 (1) (1991) 179–189, https://doi.org/10.1016/S0006-
atherosclerosis and adipose tissue inflammation, Circulation 123 (11) (2011) 3495(91)82041-0.
1216–1226, https://doi.org/10.1161/CIRCULATIONAHA.110.985523. [33] D.J. McGillivray, G. Valincius, D.J. Vanderah, W. Febo-Ayala, T.J. Woodward,
[10] M.A. Ingersoll, R. Spanbroek, C. Lottaz, E.L. Gautier, M. Frankenberger, F. Heinrich, J.J. Kasianowicz, M. Losche, Molecular-scale structural and functional
R. Hoffmann, R. Lang, M. Haniffa, M. Collin, F. Tacke, A.J.R. Habenicht, L. Ziegler- characterization of sparsely tethered bilayer lipid membranes, Biointerphases 2
Heitbrock, G.J. Randolph, Comparison of gene expression profiles between human (2007) 21–33, https://doi.org/10.1116/1.2709308.
and mouse monocyte subsets, Blood 115 (3) (2010) e10–e19, https://doi.org/ [34] T. Ragaliauskas, M. Mickevicius, B. Rakovska, T. Penkauskas, D.J. Vanderah,
10.1182/blood-2009-07-235028. F. Heinrich, G. Valincius, Fast formation of low-defect-density tethered bilayers by
[11] H. Kawai, Y. Minamiya, N. Takahashi, Prognostic impact of S100A9 over fusion of multilamellar vesicles, BBA Biomembr. 2017 (1859) 669–678, https://
expression in non-small cell lung cancer, Tumour Biol. 32 (4) (2011) 641–646, doi.org/10.1016/j.bbamem.2017.01.015.
https://doi.org/10.1007/s13277-011-0163-8. [35] M.J. Hunter, W.J. Chazin, High level expression and dimer characterization of the
[12] B. Fan, L.H. Zhang, Yn. Jia, X.Y. Zhong, Y.Q. Liu, X.J. Cheng, X.H. Wang, X.F. Xing, S100 EF-hand proteins, migration inhibitory factor-related proteins 8 and 14,
Y. Hu, Y.A. Li, H. Du, W. Zhao, Z.J. Niu, A.P. Lu, J.Y. Li, J.F. Ji, Presence of J. Biol. Chem. 273 (1998) 12427–12435, https://doi.org/10.1074/
S100A9-positive inflammatory cells in cancer tissues correlates with an early stage jbc.273.20.12427.
cancer and a better prognosis in patients with gastric cancer, BMC Cancer 12 [36] H. Iii LeVine, Quantification of β-sheet amyloid fibril structures with thioflavin T,
(2012) 316, https://doi.org/10.1186/1471-2407-12-316. Methods Enzymol. 309 (1999) 274–284, https://doi.org/10.1016/S0076-6879(99)
[13] J. Markowitz, W.E. Carson, Review of S100A9 biology and its role in cancer, 09020-5.
Biochim. Biophys. Acta 2013 (1835) 100–109, https://doi.org/10.1016/j. [37] I. Horcas, R. Fernández, J.M. Gómez-Rodríguez, J. Colchero, J. Gómez-Herrero, A.
bbcan.2012.10.003. M. Baró, WSxM: a software for scanning probe microscopy and a tool for
[14] V. Bagheri, V. Apostolopoulos, Pro-inflammatory S100A9 protein: a double-edged nanotechnology, Rev. Sci. Instrum. 78 (2007), 013705, https://doi.org/10.1063/
sword in cancer? Inflammation 42 (2019) 1137–1138, https://doi.org/10.1007/ 1.2432410.
s10753-019-00981-8. [38] A. Melcrová, S. Pokorna, S. Pullanchery, M. Kohagen, P. Jurkiewicz, M. Hof,
[15] S. Bhattacharya, C.G. Bunick, W.J. Chazin, Target selectivity in EF-hand calcium P. Jungwirth, P.S. Cremer, L. Cwiklik, The complex nature of calcium cation
binding proteins, BBA Mol. Cell Res. 1742 (2004) 69–79, https://doi.org/10.1016/ interactions with phospholipid bilayers, Sci. Rep. 6 (2016) 38035, https://doi.org/
j.bbamcr.2004.09.002. 10.1038/srep38035.
[16] A. Hermann, R. Donato, T.M. Weiger, W.J. Chazin, S100 calcium binding proteins [39] K. Balantič, V.U. Weiss, G. Allmaier, P. Kramar, Calcium ion effect on phospholipid
and ion channels, Front. Pharmacol. 3 (2012) 67, https://doi.org/10.3389/ bilayers as cell membrane analogues, Bioelectrochemistry 143 (2022), 107988,
fphar.2012.00067. https://doi.org/10.1016/j.bioelechem.2021.107988.
[17] P. Calissano, S. Alema, P. Fasella, Interaction of S-100 protein with cations and [40] U.R. Pedersen, C. Leidy, P. Westh, G.H. Peters, The effect of calcium on the
liposomes, Biochemistry 13 (22) (1974) 4553–4560, https://doi.org/10.1021/ properties of charged phospholipid bilayers, BBA Biomembranes 1758 (5) (2006)
bi00719a013. 573–582, https://doi.org/10.1016/j.bbamem.2006.03.035.
[18] W. Nacken, C. Sorg, C. Kerkhoff, The myeloid expressed EF-hand proteins display a [41] W.W. Christie, Phosphatidylserine and Related Lipids, Structure, Occurrence,
diverse pattern of lipid raft association, FEBS Lett. 572 (2004) 289–293, https:// Biochemistry and Analysis, The AOC Lipids Library, 2010.
doi.org/10.1016/j.febslet.2004.07.024. [42] C.P.S. Tilcock, P.R. Cullis, S.M. Gruner, Calcium-induced phase separation
[19] E.M. Kawamoto, C. Vivar, S. Camandola, Physiology and pathology of calcium phenomena in multicomponent unsaturated lipid mixtures, Biochemistry 27 (5)
signaling in the brain, Front. Pharmacol. 3 (2012) 61, https://doi.3389/ (1988) 1415–1420, https://doi.org/10.1021/bi00405a004.
fphar.2012.00061, https://doi.3389/fphar.2012.00061. [43] S.W. Schneider, J. Lärmer, R.M. Henderson, H. Oberleithner, Molecular weights of
[20] H.C. Jones, R.F. Keep, The control of potassium concentration in the cerebrospinal individual proteins correlate with molecular volumes measured by atomic force
fluid and brain interstitial fluid of developing rats, J. Physiol. Lond. 383 (1987) microscopy, Pflugers Arch. Eur. J. Physiol. 435 (3) (1998) 362–367, https://doi.
441–453, https://doi.org/10.1113/jphysiol.1987.sp016419. org/10.1007/s004240050524.
[21] E. Källberg, S. Tahvili, F. Ivars, T. Leanderson, Induction of S100A9 homodimer [44] S.M. Valenzuela, M. Berkahn, A. Porkovich, T. Huynh, J. Goyette, D.K. Martin, C.
formation in vivo, Biochem. Biophys. Res. Commun. 500 (3) (2018) 564–568, L. Geczy, Soluble structure of CLIC and S100 proteins investigated by atomic force
https://doi.org/10.1016/j.bbrc.2018.04.086. microscopy, J. Biomater. Nanobiotechnol. 2 (1) (2011) 8–17, https://doi.org/
[22] A. Malmendal, C.W.V. Kooi, N.C. Nielsen, W.J. Chazin, Calcium-modulated S100 10.4236/jbnb.2011.21002.
protein-phospholipid interactions. An NMR study of calbindin D9k and DPC, [45] Fluidic analytic. https://www.fluidic.com/toolkit/hydrodynamic-radius-convert
Biochemistry 44 (17) (2005) 6502–6512, 10.21/bi050088z. er/.
[23] A.F. Garcia, J.L.S. Lopes, A.J. Costa-Filho, B.A. Wallace, A.P.U. Araujo, Membrane [46] T.T. Mills, G.E.S. Toombes, S. Tristram-Nagle, D.-M. Smilgies, G.W. Feigenson, J.
interactions of S100A12 (Calgranulin C), PLosOne 8 (12) (2013), e82555, https:// F. Nagle, Order parameters and areas in fluid-phase oriented lipid membranes
doi.org/10.1371/journal.pone.0082555. using wide angle X-ray scattering, Biophys. J. 95 (2) (2008) 669–681, https://doi.
[24] K.A. Burke, E.A. Yates, J. Legleiter, Biophysical insights into how surfaces, org/10.1529/biophysj.107.127845.
including lipid membranes, modulate protein aggregation related to [47] D.E. Warschawski, P.F. Devaux, Order parameters of unsaturated phospholipids in
neurodegeneration, Front. Neurol. 4 (2013) 17, https://doi.org/10.3389/ membranes and the effect of cholesterol: a 1H–13C solid-state NMR study at
fneur.2013.00017. natural abundance, Eur. Biophys. J. 34 (2005) 987–996, https://doi.org/10.1007/
[25] D. Mrdenovic, I.S. Pieta, R. Nowakowski, W. Kutner, J. Lipkowski, P. Pieta, s00249-005-0482-z.
Amyloid β interaction with model cell membranes – what are the toxicity-defining [48] J.R. Silvius, Thermotropic phase transitions of pure lipids in model membranes and
properties of amyloid β? Int. J. Biol. Macromol. 200 (2022) 520–531, https://doi. their modifications by membrane proteins, in: P.C. Jost, O.H. Griffith (Eds.), Lipid-
org/10.1016/j.ijbiomac.2022.01.117. Protein Interactions vol 2, Wiley-Interscience, New York, 1982, pp. 239–281.
[26] M. Löwe, M. Kalacheva, A.J. Boersma, A. Kedrov, The more the merrier: effects of [49] M.L. Schmidt, L. Ziani, M. Boudreau, J.H. Davis, Phase equilibria in DOPC/DPPC:
macromolecular crowding on the structure and dynamics of biological membranes, conversion from gel to subgel in two component mixtures, J. Chem. Phys. 131
FEBS J. 287 (23) (2020) 5039–5067, https://doi.org/10.1111/febs.15429. (2009), 175103, https://doi.org/10.1063/1.3258077.
[27] J. Pansieri, I.A. Iashchishyn, H. Fakhouri, L. Ostojić, M. Malisauskas, [50] T. Parasassi, G. De Stasio, A. d’Ubaldo, E. Gratton, Phase fluctuation in
G. Musteikyte, V. Smirnovas, M.M. Schneider, T. Scheidt, C.K. Xu, G. Meisl, T.P. phospholipid membranes revealed by Laurdan fluorescence, Biophys. J. 57 (6)
J. Knowles, E. Gazit, R. Antoine, L.A. Morozova-Roche, Templating S100A9 (1990) 1179–1186, https://doi.org/10.1016/S0006-3495(90)82637-0.
amyloids on Aβ fibrillar surfaces revealed by charge detection mass spectrometry, [51] L.A. Bagatolli, B. Maggio, F. Aguilar, C.P. Sotomayor, G.D. Fidelio, Laurdan
microscopy, kinetic and microfluidic analyses, Chem. Sci. 11 (2020) 7031–7039, properties in glycosphingolipid-phospholipid mixtures: a comparative fluorescence
https://doi.org/10.1039/C9SC05905A. and calorimetric study, Biochim. Biophys. Acta 1325 (1) (1997) 80–90, https://
[28] M. Leri, H. Chaudhary, I.A. Iashchishyn, J. Pansieri, Ž.M. Svedružić, S. Gómez doi.org/10.1016/s0005-2736(96)00246-5.
Alcalde, G. Musteikyte, V. Smirnovas, M. Stefani, M. Bucciantini, L.A. Morozova- [52] L.P. Jameson, N.W. Smith, S.V. Dzyuba, Dye-binding assays for evaluation of the
Roche, Natural compound from olive oil inhibits S100A9 amyloid formation and effects of small molecule inhibitors on amyloid (Aβ) self-assembly, ACS Chem.
cytotoxicity: implications for preventing Alzheimer’s disease, ACS Chem. Neurosci. Neurosci. 3 (11) (2012) 807–819, https://doi.org/10.1021/cn300076x.
12 (11) (2021) 1905–1918, https://doi.org/10.1021/acschemneuro.0c00828. [53] J. Pansieri, L. Ostojić, I.A. Iashchishyn, M. Magzoub, C. Wallin, S.K.T.
S. Wärmländer, A. Gräslund, M. Nguyen Ngoc, V. Smirnovas, Ž. Sverdužić, L.
A. Morozova-Roche, Pro-inflammatory S100A9 protein aggregation promoted by

10
R. Tamulytė et al. BBA - Biomembranes 1865 (2023) 184113

NCAM1 peptide constructs, ACS Chem. Biol. 14 (7) (2019) 1410–1417, https://doi. [68] S.D. Connell, G. Heath, P.D. Olmsted, A. Kisil, Critical point fluctuations in
org/10.1021/acschembio.9b00394. supported lipid membranes, Faraday Discuss. 161 (2013) 91–111, https://doi.org/
[54] I.A. Iashchishyn, D. Sulskis, M.N. Ngoc, V. Smirnovas, L. Morozova-Roche, Finke- 10.1039/C2FD20119D.
watzky two-step nucleation-autocatalysis model of S100A9 amyloid formation: [69] Y.F. Dufrêne, W.R. Barger, J.B.D. Green, G.U. Lee, Nanometer scale surface
protein misfolding as “nucleation” event, ACS Chem. Neurosci. 8 (2017) properties of mixed phospholipid monolayers and bilayers, Langmuir 13 (1997)
2152–2158, https://doi.org/10.1021/acschemneuro.7b00251. 4779–4784, https://doi.org/10.1021/la970221r.
[55] T. Vogl, N. Leukert, K. Barczyk, K. Strupat, J. Roth, Biophysical characterization of [70] L. Redondo-Morata, R.L. Sanford, O.S. Andersen, S. Scheuring, Effect of statins on
S100A8 and S100A9 in the absence and presence of bivalent cations, Biochim. nanomechanical properties of supported lipid bilayers, Biophys. J. 111 (2) (2016)
Biophys. Acta 1763 (11) (2006) 1298–1306, https://doi.org/10.1016/j. 363–372, https://doi.org/10.1016/j.bpj.2016.06.016.
bbamcr.2006.08.028. [71] M. Deleu, J. Lorent, L. Lins, R. Brasseur, N. Braun, K. El Kirat, T. Nylander, Y.
[56] S. Chakraborty, M. Doktorova, R.R. Molugu, R. Ashkar, How cholesterol stiffens F. Dufrêne, M.P. Mingeot-Leclercq, Effects of surfactin on membrane models
unsaturated lipid membranes, PNAS 117 (36) (2020) 21896–21905, https://doi. displaying lipid phase separation, BBA Biomembranes 1828 (2013) 801–815,
org/10.1073/pnas.2004807117. 10.106/j.bbamem.2012.11.007.
[57] M. Levi, P. Wilson, S. Nguyen, E. Iorio, O. Sapora, T. Parasassi, In K562 and HL60 [72] A. Berquand, M.-P. Mingeot-Leclercq, Y.F. Dufrêne, Real-time imaging of drug-
cells membrane ageing during cell growth is associated with changes in cholesterol membrane interactions by atomic force microscopy, BBA Biomembranes 1664
concentration, Mech. Ageing Dev. 97 (1997) 109–119, https://doi.org/10.1016/ (2004) 198–205, https://doi.org/10.1016/j.bbmem.2004.05.010.
S0047-6374(97)00047-X. [73] E. Sezgin, I. Levental, S. Mayor, C. Eggeling, The mystery of membrane
[58] R.F.A. Zwaal, P. Comfurius, E.M. Bevers, Surface exposure of phosphatidylserine in organization: composition, regulation and roles of lipid rafts, Nat. Rev. Mol. Cell
pathological cells, Cell. Mol. Life Sci. 62 (2005) 971–988, https://doi.org/ Biol. 18 (6) (2017) 361–374, https://doi.org/10.1038/nrm.2017.16 (2017).
10.1007/s00018-005-4527-3. [74] D.J. Müller, A. Engel, The height of biomolecules measured with the atomic force
[59] F.G. van der Goot, J.K. González-Mañas, J.H. Lakey, F. Pattus, A ’molten-globule’ microscope depends on electrostatic interactions, Biophys. J. 73 (1997)
membrane-insertion intermediate of the pore-forming domain of colicin A, Nature 1633–1644, https://doi.org/10.1016/S0006-3495(97)78195-5.
354 (1991) 408–410. https://10.1038/354408a0. [75] M. Leri, R. Oropesa-Nuñez, C. Canale, S. Raimondi, S. Giorgetti, E. Bruzzone,
[60] V.E. Bychkova, A.E. Dujsekina, S.I. Klenin, E.I. Tiktopulo, V.N. Uversky, O. V. Bellotti, M. Stefani, M. Bucciantini, Oleuropein aglycone: a polyphenol with
B. Ptitsyn, Molten globule-like state of cytochrome c under conditions simulating different targets against amyloid toxicity, BBA Gen. Subj. 1862 (2018) 1432–1442,
those near the membrane surface, Biochemistry 35 (1996) 6058–6063, https://doi. https://doi.org/10.1016/j.bbagen.2018.03.023.
org/10.1021/bi9522460. [76] E. Canepa, S. Salassi, A.L. de Marco, C. Lambruschini, D. Odino, D. Bochicchio,
[61] D.C. Bode, M. Freeley, J. Nield, M. Palma, J.H. Viles, Amyloid-β oligomers have a F. Canepa, C. Canale, S. Dante, R. Brescia, F. Stellacci, G. Rossi, A. Relini,
profound detergent-like effect on lipid membranes bilayers, imaged by atomic Amphiphilic gold nanoparticles perturb phase separation in multidomain lipid
force and electron microscopy, J. Biol. Chem. 294 (19) (2019) 7566–7572, https:// membranes, Nanoscale 12 (2020) 19746–19759, https://doi.org/10.1039/
doi.org/10.1074/jbc.AC118.007195. D0NR05366J.
[62] R. Sabaté, M. Gallardo, J. Estelrich, Spontaneous incorporation of β-amyloid [77] J.N. Israelachvili, Intermolecular and Surface Forces, 1991, ISBN 978-0-12-
peptide into neutral liposomes, Colloids Surf. A 13–17 (2005) 270–271. https://do 375182-9, https://doi.org/10.1016/C2009-0-21560-1.
i.1016/j.colsurfa.2005.05.031. [78] J.A. Goodchild, D.L. Walsh, S.D. Connell, Nanoscale substrate roughness hinders
[63] J. Lee, Y.H. Kim, F.T. Arce, A.L. Gillman, H. Jang, B.L. Kagan, R. Nussinov, J. Yang, domain formation in supported lipid bilayers, Langmuir 35 (47) (2019)
R. Lal, Amyloid β ion channels in a membrane comprising brain total lipid extracts, 15352–15363, https://doi.org/10.1021/acs.langmuir.9b01990.
ACS Chem. Neurosci. 8 (6) (2017) 1348–1357, https://doi.org/10.1021/ [79] M. Emanuele, A. Esposito, S. Camerini, F. Antonucci, S. Ferrara, S. Seghezza,
acschemneuro.7b00006. T. Catelani, M. Crescenzi, R. Marotta, C. Canale, M. Matteoli, E. Menna,
[64] J. Pan, A. Dalzini, N.K. Khadka, C.M. Aryal, L. Song, Lipid extraction by E. Chieregatti, Exogenous alpha-synuclein alters pre- and post- synaptic activity by
α-synuclein generates semi-transmembrane defects and lipoprotein nanoparticles, fragmenting lipid rafts, eBioMedicine 7 (2016) 191–204, https://doi.org/10.1016/
ACS Omega 3 (2018) 9586–9597, https://doi.org/10.1021/acsomega.8b01462. j.ebiom.2016.03.038.
[65] M.M. Ouberai, J. Wang, M.J. Swann, C. Galvagnion, T. Guilliams, C.M. Dobson, M. [80] M. Dwivedi, T. Mejuch, H. Waldmann, R. Winter, Lateral organization of host
E. Welland, А-synuclein senses lipid packing defects and induces lateral expansion heterogeneous raft-like membranes altered by the myristoyl modification of
of lipids leading to membrane remodeling, J. Biol. Chem. 288 (29) (2013) tyrosine kinase c-Stc, Angew. Chem. Int. Ed. 56 (2017) 10511–10515, https://doi.
20883–20895, https://doi.org/10.1074/jbc.M113.478297. org/10.1002/anie.201706233.
[66] E. Sevcsik, G. Pabst, A. Jilek, K. Lohner, How lipids influence the mode of action of [81] R. Budvytyte, F. Ambrulevičius, E. Jankaityte, G. Valincius, Electrochemical
membrane-active peptides, BBA Biomembranes 1768 (10) (2007) 2586–2595, assessment of dielectric damage to phospholipid bilayers by amyloid β-oligomers,
https://doi.org/10.1016/j.bbamem.2007.06.015. Bioelectrochemistry 145 (2022), 108091, https://doi.org/10.1016/j.
[67] F.W.S. Stetter, T. Hugel, The nanomechanical properties of lipid membranes are bioelechem.2022.108091.
significantly influenced by the presence of ethanol, Biophys. J. 104 (2013)
1049–1055, https://doi.org/10.1016/j.bpj.2013.01.021.

11

You might also like