Download Cytoskeleton remodelling of confluent epithelial cells cultured on

Downloaded from http://rsif.royalsocietypublishing.org/ on August 3, 2017
rsif.royalsocietypublishing.org
Research
Cite this article: Rother J, BüchsenschützGöbeler M, Nöding H, Steltenkamp S, Samwer
K, Janshoff A. 2015 Cytoskeleton remodelling
of confluent epithelial cells cultured on porous
substrates. J. R. Soc. Interface 12: 20141057.
http://dx.doi.org/10.1098/rsif.2014.1057
Received: 22 September 2014
Accepted: 11 December 2014
Subject Areas:
biomechanics, biophysics, biomaterials
Keywords:
porous substrates, MDCK-II cells,
microrheology, atomic force microscopy,
tension, cytoskeleton
Author for correspondence:
Andreas Janshoff
e-mail: [email protected]
Electronic supplementary material is available
at http://dx.doi.org/10.1098/rsif.2014.1057 or
via http://rsif.royalsocietypublishing.org.
Cytoskeleton remodelling of confluent
epithelial cells cultured on porous
substrates
Jan Rother1, Matthias Büchsenschütz-Göbeler2, Helen Nöding1,
Siegfried Steltenkamp3, Konrad Samwer2 and Andreas Janshoff1
1
Institute of Physical Chemistry, University of Goettingen, Tammannstrasse 6, Göttingen 37077, Germany
I. Institute of Physics, University of Goettingen, Friedrich-Hund Platz 1, Göttingen 37077, Germany
3
Research Center CAESAR, Ludwig-Erhard-Allee 2, Bonn 53175, Germany
2
The impact of substrate topography on the morphological and mechanical properties of confluent MDCK-II cells cultured on porous substrates was scrutinized
by means of various imaging techniques as well as atomic force microscopy
comprising force volume and microrheology measurements. Regardless of the
pore size, ranging from 450 to 5500 nm in diameter, cells were able to span
the pores. They did not crawl into the holes or grow around the pores. Generally,
we found that cells cultured on non-porous surfaces are stiffer, i.e. cortical tension rises from 0.1 to 0.3 mN m21, and less fluid than cells grown over pores.
The mechanical data are corroborated by electron microscopy imaging showing
more cytoskeletal filaments on flat samples in comparison to porous ones. By
contrast, cellular compliance increases with pore size and cells display a more
fluid-like behaviour on larger pores. Interestingly, cells on pores larger than
3500 nm produce thick actin bundles that bridge the pores and thereby
strengthen the contact zone of the cells.
1. Introduction
Structure and function of eukaryotic cells heavily depend on their cellular microenvironment leading to a close coupling between cellular properties and the
surroundings. Sensing of the environment is usually accomplished through a
defined molecular contact between a protein network—the extracellular matrix
(ECM)—and specified transmembrane proteins such as integrins that connect the
ECM network to the cytoskeleton allowing the transmission of force. Adhesion
of cells is the initial step that precedes cell spreading, proliferation, differentiation
and cell–cell contact formation. Cellular adhesion determines the function and
fate of eukaryotic cells to a much larger extent than initially expected. In particular,
environmental cues such as those emanating from the substrate itself, like topography, elasticity or surface functionalization, govern a large number of cellular
responses encompassing cell growth, gene expression, apoptosis all accompanied
by substantial cytoskeletal remodelling [1]. Strikingly, also the differentiation
of stem cells is guided by mechanical and adhesive properties of the culture
dish [1]. Cells are capable of sensing the underlying substrate and respond to variations in elasticity or topography as first shown by Pelham & Wang [2]. Since then,
many studies have shown an influence of substrate rigidity on cellular migration,
proliferation, cell stiffness and even differentiation [3–5]. Cells gather information
about mechanical properties of the surroundings via mechanotransduction [6–8],
in which they sense their mechanical environment mostly via cell–cell or cell–
substrate connections. These mechanosensory elements rely predominantly on
the actin cytoskeleton. The actin cytoskeleton is tightly linked to the membrane
and is able to generate forces via motor proteins of the myosin family. It is therefore
conceivable that the amount of tension that can be generated due to the deformability of the substrate is responsible for the integration of the mechanical signal. While
the impact of substrate stiffness on cell morphology and more particularly on
cell adhesion has been well examined, less attention has been paid to substrate
& 2015 The Author(s) Published by the Royal Society. All rights reserved.
Downloaded from http://rsif.royalsocietypublishing.org/ on August 3, 2017
2.1. Substrates
Substrates with pore diameters of 0.45, 0.8, 1.2 and 2 mm were purchased from fluXXion B.V. (Eindhoven, The Netherlands). The
porous membrane of these substrates consists of silicon nitride
2.2. Cell culture
All experiments were carried out using epithelial Madin – Darby
canine kidney cells (MDCK-II cells) purchased from the Health
Protection Agency (Salisbury, UK). The cells were cultured in
minimal essential medium with Earle’s salts (Biochrom, Berlin,
Germany) containing 4 mM L-glutamine, 2.2 g l21 NaHCO3 and
10% FCS (M10F medium) at 378C in a 5% CO2-humidified
incubator. Confluent cell layers were subcultured weekly by
trypsinization (Biochrom).
For experiments, substrates were placed into a Petri dish
(2.5 cm, TPP, Switzerland). Cells were seeded in a density of
500 000 cells per Petri dish and incubated for 2 days at 378C in
a 5% CO2-humidified incubator. Experiments were carried out
in M10F medium supplemented with penicillin – streptomycin
and HEPES. For actin depolymerization experiments, the
medium additionally contained 2 mM cytochalasin D (see the
electronic supplementary material).
2.3. Atomic force microscopy
Atomic force microscopy (AFM) experiments were carried out
using a MFP-3D (Asylum Research, Santa Barbara, CA, USA)
set-up equipped with a BioHeater mounted on an inverted
Olympus IX 51 microscope (Olympus, Tokio, Japan). MLCT cantilevers (C-lever, nominal spring constant 10 pN nm21, length
200 mm, tip height 8 mm, Bruker, Camarillo, CA, USA) with a
pyramidal tip were used for imaging and force spectroscopic
experiments. Prior to each experiment, the spring constant of
the cantilever and the hydrodynamic drag force acting on the
cantilever were determined on a flat glass slide. The spring constant was calibrated using the thermal noise method [17]. The
hydrodynamic coefficient was calculated by a method previously
described by Alcaraz et al. [18]. Afterwards, the glass slide
was replaced by the substrates used for cell culturing. A homemade holder of spring steel fixed the substrates during
measurement inside the BioHeater. The temperature was set to
378C throughout measurement. Before each force spectroscopic
measurement, the area of interest was imaged in contact mode
(constant force, see figure 1).
Force spectroscopy and frequency-dependent rheological data
were acquired by the cantilever approaching the surface with a
velocity of 3 mm s21. When the pre-set cantilever deflection was
reached, the z-piezo movement was stopped for 0.5 s before it
was excited to oscillate with frequencies between 5 and 100 Hz
at small oscillation amplitudes (Ad ¼ 40 nm, peak to peak). After
an additional quiescent period of 0.5 s, it was retracted from the
surface. Per area of interest, 1024 of these force–distance curves
where recorded in a 32 32 point grid, thus the individual positions where the force curves have been measured have a
distance of 2 mm. Each experiment has been independently
conducted at least two times probing several cells.
2
J. R. Soc. Interface 12: 20141057
2. Material and methods
and displays a regular hexagonal pore pattern (figure 1, insets).
The porosity varies between 20% and 30%. To produce substrates
with larger pore diameters (3.5 mm, 5.5 mm, we used (1-0-0)
oriented SOI wafers as described previously [16]. To achieve a
hydrophilic surface, the pores were covered with a SiO2 layer of
500 nm thickness by means of wet-thermal oxidation. After cleaning the substrates in argon plasma, they were coated with a 30–
35 nm thick gold layer by argon sputtering (Balzers, BAE 250 coating system) or were first sputtered with a thin titanium layer
(Cressington 108auto Sputter coater, Watford, UK) and goldcoated afterwards (BAL-TEC MED 020 Coating System). Before
use, the substrates were sterilized in pure ethanol and incubated
with cell culture medium before inoculation of the cells.
rsif.royalsocietypublishing.org
topography of rigid surfaces. Especially in the context of cell
attachment and growth on implant materials, surface roughness
and topography are decisive parameters that might foster or
reduce the degree of differentiation and polarization of epithelial
cells. Numerous hard materials ranging from metals to silicon
are used by the healthcare industry to serve as artificial joint
replacements, stents or dental implants. Surface treatment to
change the roughness has been proved to modulate adhesion
of cells, cytokine release and gene expression of osteoblastic
cells [9]. Spatz and co-workers [10] used colloidal lithography
to provide specific attachment sites for integrins in a defined
geometry, thus addressing topographical effects on the nanoscale. The study helped to shine light on the universal length
scale that defines the optimal spacing of RGD sequences
found in ECM proteins such as collagen to match the intrinsic
spacing of integrins in the basal cell membrane. Recently, it
has also been found that topographical cues can determine
the fate of stem cell differentiation [11,12]. Additionally, the
response of cells to their environment depends heavily on
the cell type. So far, mostly mesenchymal cells such as fibroblasts
have been studied. These cells have been demonstrated to bridge
non-adhesive areas via actin stress fibres and myosin II [13].
Similarly, growths of epithelial cell monolayers over large nonadhesive gaps simulating wound healing of skin have been
shown to rely on tension generated by the actomyosin network
[14]. Supports of microneedles or pillars used for traction force
microscopy also show that cells can span large non-adhesive
areas connected via contractile elements [15]. Consequences of
substrate topography for cell morphology, polarity and mechanics in established cell monolayers have only sparsely been
addressed, although surface topology is crucial to identify the
right scaffolds for epithelial tissue engineering that requires a
fundamental understanding how cell adhesion and mechanics
are coupled to environmental cues and how cells respond
collectively to changes in substrate topography.
Here, we systematically investigated the morphological
and viscoelastic properties of MDCK-II cells grown to confluence on hard porous substrates with varying pore sizes. We
found that cells generally appear softer and more liquid-like
with increasing pore size up to 5.5 mm in diameter and remodel their actin cytoskeleton to span larger pores. Additionally,
cells generally become smaller and higher when grown on
pores. Cultured on substrates with pore sizes larger than
5 mm in diameter, MDCK-II cells mirror the cubic organization of the underlying substrate and also display a larger
amount of excess surface area and localization of ezrin only
at the apical membrane of the cells. The study shows how
subtle changes in substrate topography substantially influence the morphology and mechanics of epithelial cell layers
by forcing the cell to remodel the actin cytoskeleton in
response to the altered environment. As softness and morphology of the cells can be adjusted by changing the
porosity and pore size of the surface, it is conceivable to
use this technique also in implants, stents and wound healing
to adapt to the biological requirements.
Downloaded from http://rsif.royalsocietypublishing.org/ on August 3, 2017
0.45 mm
plain
3
0.80 mm
rsif.royalsocietypublishing.org
2
0
–2
2
0
–2
2
0
–2
height (mm)
1.20 mm
3.50 mm
2
0
–2
5.50 mm
2
0
–2
0
10
20 30 40
position (mm)
50
60
J. R. Soc. Interface 12: 20141057
height (mm)
10 mm
0
10
20 30 40
position (mm)
50
60
2
0
–2
0
10
20 30 40
position (mm)
50
60
Figure 1. Morphology of MDCK-II cells grown on substrates with different pore sizes. Images show AFM deflection maps of the cell surface of living MDCK-II cells.
The diagrams below are height profiles of the cells shown in the picture (red line, average over five scan lines). Inlets show 10 10 mm2 AFM height images of the
porous substrates (approx. twofold magnification compared with the corresponding deflection images of the cell surface). (Online version in colour.)
2.3.1. Tension model
A selection of force–indentation curves, obtained from the centre
of the cell, was chosen from the overall 1024 force–distance curves
per force map. The selection was necessary to avoid artefacts from
the underlying substrate and cell boundaries. Additionally, force–
distance curves, which show a mechanical instability, were also
excluded from the analysis. Each force curve was subject to fitting
with the parameters of the liquid droplet model as detailed
previously [19,20]. In brief, the shape parameters R1, which
describe the morphology of the cell (subconfluent state) or the
apical cap of the cell (confluent state), were obtained from AFM
height images before indentation (see also electronic supplementary material, figure S3). The obtained parameters are compiled
in the electronic supplementary material, table S1. The actual
shape of the parametrized cell surface was computed at every
indentation depth assuming constant volume. The fitting procedure relies on a Simplex algorithm and provides the overall
tension T0 comprising membrane and cortical tension and the
apparent area compressibility modulus K~A that reflects excess
membrane area, i.e. smaller values indicate the presence of a
higher amount of excess membrane area.
amplitude (40 nm) at frequencies ranging from 5 to 100 Hz.
The resulting force response F(v) and the corresponding
change of indentation depth d(v) are measured. Applying a linearized contact mechanical model, the complex shear modulus
G*(v) can be calculated from the ratio of F(v) and d(v)
1n
F(v)
(2:1)
G (v) ¼
ivb0 ,
3 d0 tan (fhalf ) d(v)
where n is the Poisson ratio, d0 is the indentation depth before the
oscillation, fhalf is the half opening angle of the cantilever tip and
b0 is the drag coefficient.
The resulting values for the complex shear modulus were
fitted using the power-law structural damping model [21]
a
2
v
G (v) ¼ G0 1 þ i tan
þ ivm,
(2:2)
pa
v0
with the scaling factor G0, the power-law coefficient a and an
Newtonian viscosity m and frequency v0 that was kept constant
at a value of 1.
2.4. Fluorescence staining
2.4.1. Actin staining
2.3.2. Microrheology
Rheology maps of confluent MDCK-II cells cultured on substrates with different pore sizes were obtained using a method
previously described by Alcaraz et al. [21]. During indentation,
the cantilever is excited sinusoidally at its base with small
MDCK-II cells grown on different substrates were washed in
phosphate buffered saline (PBS) and afterwards fixed for
15 min at room temperature using a 4% paraformaldehyde solution in PBS (FLUKA, Switzerland). After fixation, the samples
were washed two times with PBS and stored under PBS at 48C
Downloaded from http://rsif.royalsocietypublishing.org/ on August 3, 2017
The cells were grown for 2 days on the corresponding substrate and
then fixed as described in §2.4 using paraformaldehyde (PFA).
Prior to labelling, the samples were washed two times with
washing buffer (0.1% BSA (Sigma-Aldrich) in PBS) and incubated
for 30 min in blocking buffer (5% BSA, 0.3% Triton-X-100 in PBS)
at room temperature. After rinsing the sample three times with
PBS, cells were incubated with mouse anti-ezrin antibody
(5 mg ml21, BD Biosciences) for 1 h at room temperature. In the
next step, the cells were again rinsed three times for at least 5 min
using PBS and then treated with the secondary antibody
(5 mg ml21 AlexaFluor546-coupled goat anti-mouse, Invitrogen)
for 1 h. After staining ezrin and rinsing three times, ZO-1 was
stained using an AlexaFluor488-coupled mouse anti-ZO-1 antibody
(5 mg ml21, Invitrogen, 339188). Finally, the sample was washed
three times and DNA was stained using a 50 ng ml21 solution of
DAPI. All antibodies and DAPI were solved in a PBS buffer containing 1% BSA, 0.3% Triton-X. Electronic supplementary material,
figure S7, shows MDCK-II cells grown in part on the porous
region of the substrate and in part on the flat region.
The samples were imaged using a LSM710 confocal laser
scanning microscope (Zeiss, Göttingen, Germany) equipped
with a 63 objective (Zeiss) and an Argon-LASER (LASOS
Lasertechnik GmbH, Jena, Germany). The open source software
FIJI (homepage: www.fiji.sc) was used for image representation.
2.5. Scanning electron microscopy and image analysis
MDCK-II cells grown on different substrates were washed with
PBS before they were treated for 2 min with a 4% Triton X-100
solution in PBS. After rinsing the samples in PBS, the cells
were fixed using a 2.5% glutardialdehyde solution (Acros Organics, Belgium) in PBS. The samples were incubated for 1 h at room
temperature. Then cells were again washed in PBS and dehydrated using an ethanol series with an increasing ethanol
proportion (50%, 70%, 80% and 90%). The samples were incubated in each ethanol solution for 1 h at room temperature and
stored overnight in ethanol p.a. at 48C. Finally, the samples
were dried using a gentle nitrogen stream and subsequently
coated with a 5 nm gold layer by argon sputtering.
Line densities of filaments were determined by analysing
SEM images of the same category. For each pore size, a spatially
linear sequence of at least 20 images was recorded. Furthermore,
lines with an average length of 2 mm and random orientation
were drawn into each image. Finally, the number of single filaments crossing an individual line was counted yielding the line
densities in 1 nm21. This was repeated for at least 100 lines
resulting in line density distributions for the different pore
sizes (see figure 3 and electronic supplementary material).
2.6. Finite-element simulation
In fluorescence images of the actin cytoskeleton of MDCK-II cells
grown on 5.5 mm pores, we observed that many pores were
3. Results
MDCK-II cells were grown to confluence on substrates displaying regular pores ranging from 0.45 to 5.5 mm in
diameter. Morphology, cytoskeleton organization and viscoelasticity of MDCK-II cells were investigated by techniques
with high spatial resolution to quantify how substrate topography is mirrored in cellular structure and mechanics.
Notably, MDCK-II cells grow on all substrates to confluence
bridging the pores.
3.1. Morphology of MDCK-II cells grown on flat and
porous substrates
MDCK-II cells belong to normal epithelia. They polarize
when cultured on a Petri dish and exhibit a high density of
cell –cell contacts (tight junctions), which is also expressed in
a high barrier resistance (RB) of 30 V cm2 measured by electric cell –substrate impedance sensing [22]. Figure 1 shows the
influence of substrate topography on the morphology of
confluent MDCK-II cells. MDCK-II cells grown on the goldcoated, non-porous glass substrate possess an ordinary
cobblestone-like morphology with well-developed cell –cell
contacts and have a size of approximately 20 20 mm2.
ECM proteins deposited on a preformed cysteine monolayer
mediate adhesion of cells. The line profile shows rather flat
cells with a height difference of 1.5 mm from cell– cell contacts to their highest point at the centre of the cell. With
increasing pore size from 0.45 to 5.5 mm, an increase in the
average height of the apical cap and a decrease in the spreading area of the MDCK-II cells can be observed. If cells are
grown on a substrate with a pore size of 5.5 mm in diameter,
they exhibit a height difference from their lowest point in the
cell periphery to their highest point of more than 3 mm. At
the same time, the cell’s footprint decreases to an area of
13 13 mm2. A similar trend has been observed for vascular
endothelial cells [23]. It should be noted that it is not possible
to determine the overall cell height from substrate to apex with
AFM-contact imaging but only the height of the apical cap of
the cell. However, confocal laser scanning microscopy images
(z-stacks) reveal that MDCK-II cells cultured on pores are generally higher than those cultured on flat substrates (see also
table 1) [24].
3.2. Structure of the actin cytoskeleton
Actin filaments are stained using AlexaFluor546-labelled
phalloidin and imaged by confocal laser scanning microscopy.
Figure 2 shows the structure of the actin cytoskeleton at the
basal level of MDCK-II cells cultured on substrates with different pore sizes. On the flat surface, the actin cytoskeleton is well
developed (figure 2a). A large amount of stress fibres, which
traverse the entire length of the cells, is observable. By contrast,
4
J. R. Soc. Interface 12: 20141057
2.4.2. Ezrin and ZO-1 staining
interconnected by two distinct thick actin bundles. To find an
explanation for this effect, we conducted finite-element analysis.
For numerical simulations, we used COMSOL multiphysics
(Göttingen, Germany). A long, 200 nm thin sheet crossed by two
orthogonal sheets simulated the actin cytoskeleton. The sheet’s
elastic modulus was set to 1 MPa and the Poisson ratio to 0.49.
To account for the pores, we drew circular holes at the two crossing points of the three sheets. An isotropic inwards-directed
pressure of 5 Pa was applied to the boundaries of the pores.
rsif.royalsocietypublishing.org
for a maximum of 3 days. Prior to labelling, the samples were
washed two times with washing buffer (0.1% bovine serum albumin (BSA, Sigma-Aldrich, Munich, Germany) in PBS) and
incubated for 45 min in blocking buffer (5% BSA, 0.3% TritonX-100 in PBS) at room temperature. To stain actin filaments,
samples were incubated for 1 h at room temperature with
165 nM solution of AlexaFlour546-labelled phalloidin (Invitrogen, Germany) in dilution buffer (1% BSA, 0.3% Triton-X-100).
After incubation, cells were rinsed two times for 5 min in washing buffer. To label DNA, samples were treated with a
25 ng ml21 solution of 40 ,6-diamidino-2-phenylindole (DAPI,
Sigma-Aldrich) in PBS for 10 min at room temperature. Finally,
samples were washed two times with PBS for 5 min.
Downloaded from http://rsif.royalsocietypublishing.org/ on August 3, 2017
Table 1. Cell height of MDCK-II cells cultured on substrates with different
pore size (diameter) measured by confocal laser scanning microscopy.
5.6 + 0.9
5.9 + 0.7
0.80 mm (n ¼ 20)
1.20 mm (n ¼ 20)
4.6 + 0.8
6.5 + 0.9
3.50 mm (n ¼ 20)
6.3 + 2.2
5.50 mm (n ¼ 20)
5.5 + 0.9
cells grown on pores show reduced fluorescence intensity on
the porous regions (figure 2b–f ). The number of stress fibres
traversing the cell also decreases with increasing pore size.
Strikingly, however, we observe an accumulation of f-actin
inside the largest pores (5.5 mm and occasionally also in
3.5 mm pores), which is expressed in periodically occurring
bright spots indicating the position of the pores (figure 2f ).
Pores covered by a single cell are interconnected by thick
actin fibres producing a square pattern, where the pores are
located at the corners. Many of the interconnections have
been found to consist of two thick actin bundles (figure 2h,i).
Finite-element simulations show an increased stress at the
outer regions of an interconnection between two pores
giving a possible explanation for the observed effect. We
assume that tension of the actomyosin network needed to
span the pores leads to formation of two actin bundles on
the pore rim to sustain the stress. The actomyosin filaments
inside/over the pores might also release tension in the membrane by pulling the attached membrane to the pore centre
as opposed to the gradient in free energy that generates
stress in the opposite direction. Interestingly, at subconfluent
regions, we did not observe actin aggregates inside large
pores in cells close to cell-free areas of the substrate (see electronic supplementary material, figure S8). Instead, actin
fibres were spanning through the whole cell. This might be
an indication for the incomplete process of polarization.
To confirm and quantify our findings of a reduced or
remodelled actin cytoskeleton in cells cultured on porous
substrates, we used scanning electron microscopy. By rinsing the samples in a detergent solution before fixation and
dehydration, we were able to uncover the cytoskeleton of
MDCK-II cells. Analysis of the filament thickness indicates
that predominantly actin filaments are visible in the electron
micrographs. Microtubules were not found, although we
cannot exclude that intermediate filaments are still present.
The structure of filaments found in our measurements
resembles strongly the structure of actin filaments found by
Svitkina using a similar preparation protocol including actin
stabilization by phalloidin [25]. The filaments have a thickness
of 13 + 4 nm, which corresponds well to the expected thickness of actin fibres covered by a thin gold layer of approx.
5 nm on the filaments (see electronic supplementary material,
figure S1). Figure 3 shows representative scanning electron
micrographs of MDCK-II cells grown on a substrate with
0.8 mm pores. In figure 3b,c, the perinuclear region of a cell
grown on the flat part of the substrate and of a cell grown on
the porous part can be seen. A dense network of filaments traverses the whole cell. Magnifications are depicted
3.3. Cellular mechanics
To determine the influence of different pore sizes on the cellular
mechanics, we conducted AFM force–indentation curves
and AFM-based microrheological experiments. Force–distance
curves were described quantitatively using a modified tension
model first introduced by Sen et al. [19] and refined by Pietuch
and colleagues [20,28] as this model has been shown to be
indenter invariant and delivers more universal mechanical
parameters compared with the Hertz, Sneddon or comparable
contact mechanical models neglecting the shell structure of the
cells [19,20,28]. The tension model describes the cell as an isotropic elastic shell with a constant surface tension. The model
assumes that the restoring force originates solely from a tension T, which is the sum of the cortical and the membrane
tension T0 and a dynamic contribution from stretching of the
plasma membrane (equation (3.1)). The contribution of stretching to the tension is dependent on the projected cell surface
area A0 and the area compressibility modulus KA, which
needs to be replaced by the apparent area compressibility
~ A if the projected cell surface area is smaller than
modulus K
the actual cell surface area due to folds and wrinkles in the
membrane in the nanometre scale
T ¼ T0 þ KA
DA
A0
(3:1)
and
~ A ¼ KA
K
A0
:
A0 þ Aex
(3:2)
A0 is the projected cell surface area before indentation, DA is
the change of surface area due to stretching and Aex is the
area of the excess membrane. If the excess cell membrane
~A
stored in folds like caveolae or microvilli is very small, K
approaches KA. T0 dominates the tension at low indentation
depth, while at large strains stretching of the membrane
becomes the main contributor to the overall tension—a
consequence of the inextensibility of lipid bilayers. As a consequence, the restoring force increases nonlinearly with the
indentation depth. Recently, Pietuch et al. [29] showed that isolated apical membrane sheets display the same mechanical
J. R. Soc. Interface 12: 20141057
flat (n ¼ 20)
0.45 mm (n ¼ 17)
5
rsif.royalsocietypublishing.org
h + s.d. (mm)
in figure 3d,e. In the case of the cell grown on the porous
part, the network is less dense (figure 3c,e). Additionally, we
determined the line density of the actin network. The
distribution of line densities as a function of substrate topography is shown in figure 3f. The line density allows a rough
estimate of the network density. We observe a minimum in
the line density of actin filaments for cells grown on the
0.8 mm pores. A value of 0.014 nm21 found for cells grown
on flat surfaces corresponds to an average filament-to-filament
distance of 71 nm, which is in good agreement with mesh sizes
of the actin network found for other cells [26]. Cells cultured
on a flat support exhibit a significantly denser network compared with cells grown on porous supports. Micrographs of
cells grown on the different substrates can be found in electronic supplementary material, figure S2. The images show
that in all cases, the actin network covered the pores. In the
case of the larger pores, the actin is located slightly under
the plane of the pore rims. The reason for this might be that
the cells extend into the pore interior ( partial wetting the
pore interior) if pores are large enough (see also Scheme in
figure 7) [27].
Downloaded from http://rsif.royalsocietypublishing.org/ on August 3, 2017
(a)
(b)
porous
non-porous
(d)
3.50 mm
(e)
5.50 mm
(f)
non-porous
non-porous
porous
porous
(g)
(h) (i)
(j)
intensity (arb. units)
120
100
80
60
40
20
0
–2
–1
0
x-position
1
2
Figure 2. (a – f ) Confocal micrographs of MDCK-II cells grown on substrates with different pore sizes. Images show the actin cytoskeleton (Alexa-Fluor546-labelled
phalloidin, Invitrogen) of the cells at the level of the pores ( pseudocoloured) and corresponding orthogonal views below (scale bar, 20 mm). (g) Confocal micrograph of the actin cytoskeleton of MDCK-II cells grown on 5.5 mm substrates above pore level (grey scaled). (h) Magnification showing an area of six pores (green
filled circles). (i) Line profiles of the interconnections between pores shown in H (grey lines) and mean intensity value (red squares, mean + s.d.). ( j ) Finiteelement simulation of a thin elastic sheet representing the actin cytoskeleton between two pores. Pores are represented by holes in the sheet at the crossing. Black
lines indicate shape of the sheet before deformation. An inward-directed pressure (arrows) is applied to the pore boundaries causing deformation and occurrence of
stress compensating tension in the membrane. Blue colour indicates low stress, green and yellow intermediate stress and red high stress). (Online version in colour.)
properties in response to indentation as living cells. Apical
membranes were removed from confluent living MDCK-II
cells and placed on a porous mesh. The plasma membrane
patches were subject to indentation with an AFM tip and the
force curves fitted with a tension model neglecting bending
resistance. Area compressibility modules were similar than
those found for living cells suggesting that the tension model
used in this study is appropriate for polarized confluent
cells. It was also demonstrated that contact models such as
those based on Hertzian mechanics fail to deliver results independent of indenter geometry. Indenting one and the same cell
with two different indenters (sphere and pyramid) attached to
an AFM cantilever produced apparent Young’s modules originating from fitting the contact models to the indentation
curves that are over one order of magnitude apart. By contrast,
applying the tension model to fit the indentation curves gave
almost identical results (pre-stress and area compressibility modulus) for both indenter geometries [29]. Schneider et al. [20]
J. R. Soc. Interface 12: 20141057
non-porous
1.20 mm
6
rsif.royalsocietypublishing.org
porous
(c)
Downloaded from http://rsif.royalsocietypublishing.org/ on August 3, 2017
counts
counts
(d)
30
20
10
0
30
20
10
0
30
20
10
0
(e)
E
2 mm
450 nm
800 nm
1200 nm
3500 nm
5500 nm
0
0.005
0.010 0.015 0.020
line density (nm–1)
0.025
Figure 3. Scanning electron micrographs of the cytoskeleton of MDCK-II cells grown on a porous substrate (0.8 mm pores). (a) Overview of the area. White rectangles mark the areas shown in (b,c). Panels (d ) and (e) show the area marked with the rectangle in (b) and (c), respectively. (f ) Line densities of actin
cytoskeleton for MDCK-II cells grown on substrates with different pore sizes obtained from scanning electron micrographs. (Online version in colour.)
(a)
(b)
400
0.35
T0 = 0.30 mN m–1
~
KA = 0.12 N m–1
0.30
T0 (mN × m–1)
F (pN)
300
200
0.20
0.15
T0 = 0.11 mN m–1
100
0.25
~
KA = 0.06 N m–1
0.10
0
0
0.4
0.8
d (mm)
1.2
1.6
0
1
2
3
4
pore diameter (mm)
5
Figure 4. (a) Exemplary force – distance curves of MDCK-II cells grown on flat substrates (grey dots) and on porous substrates (magenta triangles, 5.50 mm pores).
Force distance curves were fitted using the liquid droplet model (continuous lines). (b) Pre-stress T0 as a function of pore diameter. (Online version in colour.)
as well as Pietuch et al. [28] found that individual cells not being
part of a confluent monolayer display different mechanical
properties as those found for confluent cells. In single cells, the
whole cell seems to participate in the mechanical response and
stress fibres generate appreciably higher tension. The tension
measured for single cells was found to be almost one order of
magnitude larger than the cortical tension measured for cells
within a confluent monolayer.
To model the force –distance curves with the tension
model above, the projected cell surface area needs to be calculated using the parametrization described by Sen et al. [19]
The radius of the cap (apical surface of cells), R1 and the contact angle w are determined from AFM– images (see figure 1
and electronic supplementary material, figure S3). Assuming
that both, the curvature and the volume stay constant during
indentation, one can calculate the restoring force F for different indentation depths. The mechanical parameters T0 and
~ A can thus be obtained by fitting the result to the measured
K
force –indentation curves. Figure 4a exemplarily shows two
force –indentation curves for MDCK-II cells cultured on flat
substrates (grey circles) and substrates with 5.5 mm pores
(magenta triangles). We generally observed that cells grown
on the flat surface show a steeper increase of the force with
increasing indentation depth compared with cells on
porous substrates. At small indentation, depth cells grown
on larger pores show a weaker increase of force and therefore
a lower cortical tension. This observation is reflected in
changes of the tension T0, which is the sum of membrane
and cortical tension as the dominating contribution and
thus the response to externally applied forces (equation
7
J. R. Soc. Interface 12: 20141057
(c)
30
20
10
0
counts
2 mm
30
20
10
0
counts
5 mm
plain
20
10
0
rsif.royalsocietypublishing.org
counts
( f ) 30
D
counts
(b)
(a)
Downloaded from http://rsif.royalsocietypublishing.org/ on August 3, 2017
(a)
(b)
(c)
8
3.0
rsif.royalsocietypublishing.org
2.5
103
(G¢¢/G¢)
G¢¢ (Pa)
G¢ (Pa)
103
b = 1.00
b = 0.23
102
2.0
1.5
1.0
0.5
102
10
100
1
10
f (Hz)
100
1
f (Hz)
10
100
f (Hz)
Figure 5. (a,b) Real part G 0 (a) and imaginary part G 00 (b) of the complex shear modulus as a function of frequency f. The grey circles represent data from MDCK-II
cells cultured on flat substrates, while the magenta triangles are obtained from samples with 5.5 mm-sized pores. Lines show the fits of the frequency-dependent
complex shear modulus by the power-law structural damping model. Parameters are compiled in table 2. (c) Loss tangent (G 00 /G 0 ) of cells cultured on the two
samples (grey circles: flat substrate; magenta triangles: 5.50 mm pores) as a function of oscillation frequency. Data (median values) are obtained from more than
100 curves. The corresponding lines are the fits of the power-law structural damping model. (Online version in colour.)
Table 2. Mechanical parameters of MDCK-II cells grown on porous substrates with different pore size (diameter).
liquid droplet model
AFM-based microrheology
image analysis
T0 + s.d. (mN m21)
KA,app + s.d. (Nm21)
G0 + s.e. (Pa)
a + s.e.
m + s.e. (Pa s)
line dens. + s.d.
(nm21)
flat
0.31 + 0.07
0.15 + 0.11
188 + 21
0.25 + 0.02
1.9 + 0.1
0.014 + 0.003
0.45 mm
0.28 + 0.08
0.05 + 0.04
128 + 9
0.25 + 0.01
1.81 + 0.04
0.011 + 0.002
0.80 mm
1.20 mm
0.28 + 0.08
0.18 + 0.08
0.05 + 0.05
0.14 + 0.11
127 + 26
137 + 27
0.31 + 0.04
0.30 + 0.03
1.7 + 0.2
2.7 + 0.2
0.008 + 0.002
0.010 + 0.002
2.00 mm
3.50 mm
0.19 + 0.07
0.21 + 0.08
0.07 + 0.04
0.08 + 0.10
180 + 9
134 + 15
0.21 + 0.01
0.23 + 0.02
2.55 + 0.03
1.72 + 0.06
n.a.
0.011 + 0.002
5.50 mm
0.11 + 0.05
0.05 + 0.02
120 + 37
0.26 + 0.05
2.0 + 0.2
0.011 + 0.002
(3.1)) at low indentation depth. T0 decreases monotonically
when cells are cultured on substrates with increasing pore
diameter (figure 4b). Also, the apparent area compressibility
~ A decreases with increasing pore size, indicative
modulus K
of the presence of excess membrane area (table 2). This behaviour can be directly guessed from the fact that the slope of the
force –indentation curves at high indentation depths for cells
on larger pores is lower than that of cells cultured on the
plane surface (figure 4a).
Time- and frequency-dependent mechanical data are
obtained by AFM-based microrheological experiments. The
method first described by Shroff et al. [30] uses small amplitude oscillations of the cantilever, which is in contact with
the sample. By application of a contact mechanical model
the complex shear modulus G* can be calculated [31,32]. The
real part of the complex shear modulus G 0 , the storage modulus, represents the energy stored in the sample during
oscillation of the cantilever. The energy that is dissipated
during the measurement is represented by the imaginary
part of the complex shear modulus G 00 and is therefore called
loss modulus. In figure 5a,b, G 0 and G 00 are exemplarily shown
as a function of the oscillation frequency f for MDCK-II cells
cultured on flat substrate (grey circles) compared to cells
cultured on substrates exhibiting 5.5 mm pores (magenta
triangles). At all frequencies, the cells cultured on flat substrates appear stiffer than those grown on porous samples. In
general, G 0 increases according to a weak power law with
the frequency (linear increase in an double logarithmic diagram), whereas G 00 shows a stronger dependency on the
oscillation frequency, which leads to a crossing point of G 0
and G 00 at a certain frequency. At this crossing point, the oscillation of F(v) and d(v) are 458 (or p/2) out of phase and both
quantities, G 0 and G 00 , exhibit the same value. Therefore, the
loss tangent (G 00 /G 0 ) equals 1 at this frequency. A loss tangent
smaller than 1 means that G 0 . G 00 and the elastic properties of
the sample dominate the rheological behaviour. Inversely, if
G 0 , G 00 , the loss tangent exceeds 1 and viscous properties
dominate the rheology of the cells showing a more fluid-like
behavior. For MDCK-II cells grown on a flat surface the crossing point of G 0 and G 00 is found at a frequency of around 40 Hz
(figure 5c, grey), while cells grown on large pores the crossing
point is found already at 20 Hz (magenta). Generally, we
found that for cells cultured on porous substrates the frequency, at which G 0 and G 00 match, is smaller as G 00 is largely
unaffected by the substrate, especially at higher frequency.
Thus, cells cultured on porous substrates show a more fluidlike behaviour at smaller frequencies compared with cells on
a non-porous support. We used the power-law structural
J. R. Soc. Interface 12: 20141057
0
1
0.016
180
0.014
160
0.012
140
0.010
120
0.008
100
line density (nm–1)
200
0.006
1
2
3
4
pore diameter (mm)
5
Figure 6. G0(Pa) from the power-law structural damping model (blue circles)
and the measured line density (nm21) of the cytoskeleton (green squares) as
a function of pore size. The dotted line is just to guide the eye. (Online
version in colour.)
damping model, first used by Fabry and colleagues to analyse
the spectra of G* in cellular microrheology [21,33,34]. Table 2
summarizes all parameters of the power-law structural damping model obtained from fitting the spectra as described
recently [32]. Explicitly, we find a power-law coefficient a
between 0.2 and 0.3 for all samples (table 2) [33]. The scaling
factor G0 shows a similar trend as the storage modulus. Cells
on a flat substrate exhibit the highest value (188 + 21 Pa),
cells grown the substrate with the 5.5 mm pores exhibit the
lowest value (120 + 37 Pa).
Figure 6 shows that a correlation (Pearson correlation
coefficient r ¼ 0.81) between the scaling factor G0 and the
line density of actin filaments obtained from SEM analysis
exists. The expression level of the actin network is obviously
responsible for the reduced stiffness found in microrheology
experiments of cells cultured on pores.
4. Discussion
In this study, we investigated the influence of macroporous
materials on confluent epithelial cells. These materials either
simulate the influence of an ECM consisting of adhesive fibres
separated by micrometre-sized gaps or ceramic or metallic
implant materials on the response of adherent cells to structural
heterogeneity of their environment [13]. First, we investigated
the impact of the substrate on the morphology of cells grown
to a confluent monolayer. Figure 1 shows the apical surface of
MDCK-II cells on substrates with varying pore sizes imaged
by AFM. The line profiles show a height increase of the apical
cap when cells are cultured on macroporous substrates.
Additionally, cells on larger pores occupy a smaller area,
which might be due to a decrease in spreading rate due to the
limited surface area and/or an increase in proliferation rate
like it has been previously shown for hepatocytes cultured on
mesoporous anodized aluminium oxide [35]. Hoess et al. used
substrates with pore sizes ranging from 57 to 213 nm in diameter
and found that the cells on larger pore diameters show a faster
proliferation rate, which might also be the case in our experiments although the pores used in this study are in the
macroporous range. An explanation for increased proliferation
could be a more in vivo like situation concerning the additional
9
J. R. Soc. Interface 12: 20141057
0
nutrient supply also from the basal membrane when cells are
cultured on porous material. The decrease in spreading area
and the overall shape of the cell found in our experiments
might be explained by surface energy considerations. On a flat
hydrophilic surface, the reduction in surface free energy of a
cell will be large. Thus, the contact angle will be small, which
facilitates spreading of the cell. By removing up to 30% of the
hydrophilic surface, the gain in free energy is reduced and the
contact angle will become larger, which in turn leads to a
higher cell body occupying a smaller area if the volume is conserved. Furthermore, we find that cells grown on a cubic pore
pattern reproduce this pattern in cell layer organization. The
long-range cubic pattern of the cells can also be confirmed
by two-dimensional fast Fourier transform analysis of the
corresponding AFM height image of the image in figure 1
(electronic supplementary material, figure S4).
The obvious changes in cell morphology are accompanied
by substantial changes in the cytoskeletal arrangement. On flat
substrates MDCK-II cells form an f-actin network with thick
stress fibres traversing the entire cell. Cells on porous substrates, however, show a reduced number of stress fibres,
which are also shorter. But nevertheless, up to a pore size of
3.5 mm, the cells are still able to bridge the pores without tremendous remodelling of the actin network. Interestingly, the
organization of the actin cytoskeleton changes dramatically
from 1.2 to 5.5 mm pores. Cells cultured on 5.5 mm pores
show dense actin aggregates inside pores (figure 2f). Occasionally, this effect is also found for cells cultured on 3.5 mm pores.
We attribute this behaviour to an increase in tension of
the actomyosin network, which is needed to span the pores.
A similar behaviour was observed previously by Rossier
et al. [13], who researched the growth of single fibroblasts on
micropatterend substrates. The authors found that bridging between adhesive contacts is achieved by formation of
large stress fibres. This bridging is dependent on active nonmuscle myosin II. Like in a chain bridge, it needs a certain
tension of the actin fibres generated by myosin motors to
span the distance from one side (of the pore) to the other
(see also figure 2j). The larger the distance, the higher becomes
the tension acting on the filaments. As a consequence, the cell
suspends the pores by assembling a large number of actin filaments into strong networks. Additionally, thick actin bundles
connect to the aggregates inside the pores on the pore rims.
Most of them reach directly from one pore to another. Numerous aggregates are interconnected by exactly two thick actin
bundles (figure 2g–i). To explain this phenomenon, we conducted finite-element simulations (figure 2j). The actin
cytoskeleton is simulated by a cross shaped elastic material.
The porous region is represented by the absence of the
material. When applying an inwards-directed homogeneous
pressure to the pore boundaries, the edges of the interconnections are the regions, which exhibit the highest stress values.
This means for the natural system that these areas need to
be strengthened by additional filaments. This supports our
previous hypothesis that there is strong inwards-directed
force, which might be produced by the contractile force
exerted by myosin motorproteins. This high tension also
requires a strong adhesion of the cells to the substrate in proximity to the pores. In epithelial cells or fibroblasts adhesion to
the substrate is mainly realized by focal adhesion. Wu et al.
[36] as well as Rossier et al. [13] observed an accumulation of
focal adhesion contacts in regions next to pores or nonadhesive regions du Roure et al. [15] reported that MDCK
rsif.royalsocietypublishing.org
G0 /Pa
Downloaded from http://rsif.royalsocietypublishing.org/ on August 3, 2017
Downloaded from http://rsif.royalsocietypublishing.org/ on August 3, 2017
10
J. R. Soc. Interface 12: 20141057
previously observed this effect. The nutrient supply might
enhance the functionality and polarity of epithelial cells,
which in turn might lead to a larger number of microvilli on
the apical membrane. Comparison of scanning ion conductance
micrographs that allow real non-contact measurements of
PFA-fixed MDCK-II cells grown either on a Petri dish or on a
substrate with 1.2 mm pores also reveals an increased surface
roughness of the cells grown on the solid substrate (see
electronic supplementary material, figure S5) supporting the
theory of an increased number of microvilli (for description of
the experiment, see also the electronic supplementary material)
or more precisely membrane protrusions and ruffles [37].
Additionally, we also compared the distribution of ezrin, a
protein connecting the actin cytoskeleton and the membrane
at the apical membrane of epithelial cells, in cells grown on
0.8 mm pores with cells on flat parts of the same substrates.
Ezrin is found predominantly and strongly at the apical membrane if cells grow on the porous area, while it is more
dispersed in cells cultured on flat substrates, i.e. ezrin is also
found at the cell–cell contacts and at the basal side of cells. Furthermore, a large amount of ezrin is found in point-like
structures in cells grown on the porous part indicative of microvilli. However, distribution of ZO-1 seems unaffected by the
nature of the substrate (see electronic supplementary material,
figure S7).
Apart from force– indentation experiments, we also conducted microrheological experiments to capture changes in
the viscoelasticity of the cells in response to substrate porosity
as a function of frequency. At all measured frequencies, cells
cultured on the non-porous support show the highest values
in the real part of the complex shear modulus G0 , which is a
measure for the elastically stored energy. MDCK-II cells
grown on porous supports exhibit lower values in G 0 compared with cells grown on a flat surface. This observation
corresponds to the lower actin densities found in SEM
images for cells on porous supports and is along the same
lines as the indentation data discussed above. Interestingly,
the viscous properties of cells are less affected by growth
on porous substrates. As a consequence, cells on porous substrates behave more fluid-like than cells on a flat support as
the loss tangent (G 00 /G 0 ) is higher at all frequencies.
Additionally, the observed line densities of actin filaments
~ A and G 0 , leading to
(figure 3f ) show a similar trend as K
the conclusion that the actin cytoskeleton is the main contributor to both mechanical parameters (figure 6). Notably,
although G 0 of MDCK-II decreases, when cultured on
porous substrates, the power-law exponent does not show a
clear trend. One might expect that a decrease in cell stiffness
is accompanied by a higher power-law exponent as it has
been observed for several cell types treated with various
drugs [33,38]. In our experiments, however, there’s a maximum power-law coefficient of 0.3 at 0.8 and 1.2 mm sized
pores, while larger pores show smaller values comparable
to those found on a flat substrate. This might be due to
severe cytoskeleton remodelling (figures 2 and 3).
In order to show that the contributions of the actin cytoskeleton are responsible for the observed effects, we treated
cells on pores and on a flat substrate with cytochalasin D
(see electronic supplementary material, figure S6). By administration of cytochalasin D, the actin cytoskeleton is
disrupted. As expected for this case, G 0 decreases and
shows a stronger dependency with the frequency. G 00 is also
reduced at all frequencies. Comparison of cells grown on
rsif.royalsocietypublishing.org
cells cultured on micropillars indeed span the region between
the posts. Interestingly, in confluent monolayers, the cells do
not deflect the posts noteably. By use of a mild detergent, we
were also able to uncover the cytoskeleton and visualize it
using scanning electron microscopy (figure 3; and electronic
supplementary material, figure S2). The images showed a
reduction of cytoskeletal filaments in cells grown on porous
substrates.
These tremendous changes in cytoskeletal arrangement in
response to substrate properties also have strong impact on
cellular mechanics as shown in our force– indentation and
microrheolgical experiments. The cortical tension of cells
obtained from indentation experiments decreases with
increasing pore diameter. Compared with cells cultured on
5.5 mm pores, the tension T0 of cells on a flat surface is
increased by a factor of 3 to 4. The found tension value T0
of 0.30 mN m21 observed for cells grown on a flat surface
is in good agreement with the tension value of 0.37 +
0.04 mN m21 found by us in previous experiments for
MDCK-II cells cultured on a Petri dish [20]. The reduction
of tension in cells cultured on pores could be the result of
the reduced spreading area and the accompanying actin
cytoskeleton rearrangement (reduced line densities in SEM
images). While a well-spread cell exhibits a relatively high
tension, cells showing a disturbed spreading behaviour
would not be able to generate a high contractile force [28].
In this previous work, we could show that the cortical tension
of freshly seeded cells is first low to facilitate spreading. With
time, the cells started to spread and thereby the tension of the
cell increased up to a constant value. This also fits to the
observation that stress fibres, which are also important for
the overall contractile tone of the cell, seem to be shorter or
less pronounced on porous substrates.
However, considering that the adherens belt efficiently
separates apical from basolateral side, it is conceivable that
stress fibres of a fully polarized cell do not contribute substantially to the mechanical response to indentation of the
apical cap. A dominating effect might be a higher degree of
polarization and changes in the cell’s size as mentioned
above. Along the same lines, the findings of Pietuch et al.
[29] investigating isolated apical membrane sheets deposited
on a porous mesh suggest that contributions from stress fibre
contraction to the mechanical response of the apical site
might be of second order. Actin remodelling might also comprise removal of actin from the cortex and reassembly across
pores. Taken together, the observed softness of cells cultured
on porous substrate in comparison to confluent cells on stiff
continuous substrates might originate from both a softer
apical cap due to rearrangement of cortical actin as well as
a less tension generated by the shorter and fewer stress fibres.
~ A,
Looking at the apparent area compressibility modulus K
we found a largely constant value of around 0.05 N m21
for cells grown on porous substrates. Cells grown on a flat
surface exhibit a significantly higher value of 0.14 N m21.
As discussed by Pietuch et al., K~A could be a measure for
additional membrane area stored in folds, protrusions and
wrinkles like caveolae (basal) or microvilli (apical). The cell
itself uses these membrane reservoirs to increase its reactive
surface on the one side but also to regulate its tension T as a
response to changed environmental conditions and to prevent
bilayer rupture on the other side. A decrease in K~A could be elicited by an increased number of microvilli, when MDCK-II cells
are grown on porous substrates. Hoess et al. [35] have
Downloaded from http://rsif.royalsocietypublishing.org/ on August 3, 2017
actin line density
f-actin
nucleus
tension over pores
3.5
1.2
0.8
0.45
non-porous
Figure 7. Scheme compiling the effects of culturing MDCK-II cells on various macroporous substrates. Cells decrease their spreading area with increasing pore
diameter. This goes along with a reduction in cortical tension, which might be a consequence of a reduced contractile tone of the cell due to fewer and shorter
stress fibres. Furthermore, on large pores the cells produce thick actin aggregates over the pores, which generate high tension. (Online version in colour.)
the flat surface and on the porous support indicates that G 0 is
reduced in both cases. Especially at low frequencies, both
samples show comparable values in G 0 . At higher frequencies, cells on flat supports show a reduced storage modulus
compared with cells on a porous substrate. We conclude
that the main effect on cell mechanics that we observe,
when cells are cultured on solid supports, results from remodelling of the actin cytoskeleton also affecting the availability
of membrane reservoirs to buffer tension exerted on the cells.
5. Conclusion
The extracellular environment is an essential mediator of cellular structure and function by providing biochemical and
mechanical stimuli to influence single and collective cell
behaviour. Through adhesion, cells sense the mechanical
properties of the environment and translate this information
into a signalling cascade, which regulates cellular responses
such as spreading, migration and growth. While abnormal
mechanotransduction is known to be responsible for a
number of diseases, the robustness of how cells respond collectively to such stimuli is largely unexplored. Here, we
provide structural and mechanical information about the
impact of pore size on the mechanical response of confluent
epithelial cells to substrates with defined topography.
Interestingly, we found that the cells can distinguish even
subtle changes in pore size and develop strategies to remodel
their cytoskeleton in order to explore substrates with large
pores that would prevent sufficient adhesion area otherwise
by spanning and reinforcing the cytoskeleton of the free
standing part.
Cells on larger pores up to 5.5 mm respond to the reduced
adhesion area and strain in the membrane by a comprehensive
remodelling of the cytoskeleton especially in the vicinity of the
pores. The cells invade into the pores, form a contractile actin
aggregates and thereby manage to maintain an intact cell
monolayer covering the entire area. Interestingly, the cell size
and arrangement adapt to the underlying porous structure,
which we attribute to a self-organization driven by minimizing
the area of freestanding membrane over pores. The effect of
culturing MDCK-II cells on substrates with different pore
sizes are also illustrated in figure 7. A simple change in topography by adding pores to the surface results in a rich variety
of phenotypes that partly also mimic softer material and might
become an alternative to self-organize cells to a predefined
morphology, which also effects cellular elasticity, without
using biochemical cues or modifying the stiffness or surface
functionalization of the substrate.
Funding statement. Financial support of the DFG through CRC 937 (A14)
is gratefully acknowledged.
References
1.
2.
3.
Levental I, Georges PC, Janmey PA. 2007 Soft
biological materials and their impact on cell
function. Soft Matter 3, 299 –306. (doi:10.1039/
B610522j)
Pelham RJ, Wang YL. 1998 Cell locomotion and
focal adhesions are regulated by substrate flexibility
(vol. 94, pg 13661, 1997). Proc. Natl Acad. Sci. USA
95, 12070.
Trichet L, Le Digabel J, Hawkins RJ, Vedula RK, Gupta
M, Ribrault C, Hersen P, Voituriez R, Ladoux B. 2012
Evidence of a large-scale mechanosensing mechanism
for cellular adaptation to substrate stiffness. Proc. Natl
Acad. Sci. USA 109, 6933–6938. (doi:10.1073/Pnas.
1117810109)
4.
5.
6.
7.
Tee S-Y, Fu JP, Chen CS, Janmey PA. 2011 Cell shape
and substrate rigidity both regulate cell stiffness.
Biophys. J. 100, L25 –L27. (doi:10.1016/j.bpj.2010.
12.3744)
Engler AJ, Sen S, Sweeney HL, Discher DE. 2006
Matrix elasticity directs stem cell lineage
specification. Cell 126, 677–689. (doi:10.1016/j.cell.
2006.06.044)
Geiger B, Bershadsky A. 2002 Exploring the
neighborhood: adhesion-coupled cell
mechanosensors. Cell 110, 139–142. (doi:10.1016/
S0092-8674(02)00831-0)
Geiger B, Spatz JP, Bershadsky AD. 2009
Environmental sensing through focal adhesions.
Nat. Rev. Mol. Cell Biol. 10, 21 –33. (doi:10.1038/
nrm2593)
8. Ingber DE. 2006 Cellular mechanotransduction:
putting all the pieces together again. FASEB J. 20,
811–827. (doi:10.1096/fj.05-5424rev)
9. Ross AM, Jiang ZX, Bastmeyer M, Lahann J. 2012
Physical aspects of cell culture substrates:
topography, roughness, and elasticity. Small 8,
336–355. (doi:10.1002/smll.201100934)
10. Arnold M, Cavalcanti-Adam EA, Glass R, Blummel J,
Eck W, Kantlehner M, Kessler H, Spatz JP. 2004
Activation of integrin function by nanopatterned
adhesive interfaces. Chemphyschem 5, 383–388.
(doi:10.1002/cphc.200301014)
J. R. Soc. Interface 12: 20141057
5.5
rsif.royalsocietypublishing.org
cortical tension
11
Downloaded from http://rsif.royalsocietypublishing.org/ on August 3, 2017
30.
31.
32.
33.
34.
35.
36.
37.
38.
confluent cell monolayers: impact of tension and
surface area regulation. Soft Matter 9, 11 490–
11 502. (doi:10.1039/c3sm51610e)
Shroff SG, Saner DR, Lal R. 1995 Dynamic
micromechanical properties of cultured rat atrial
myocytes measured by atomic-force microscopy.
Am. J. Physiol. Cell Physiol. 269, C286–C292.
Bilodeau GG. 1992 Regular pyramid punch problem.
J. Appl. Mech. 59, 519 –523. (doi:10.1115/1.
2893754)
Rother J, Nöding H, Mey I, Janshoff A. 2014 Atomic
force microscopy-based microrheology reveals
significant differences in the viscoelastic response
between malign and benign cell lines. Open Biol. 4,
140046. (doi:10.1098/rsob.140046)
Fabry B, Maksym GN, Butler JP, Glogauer M,
Navajas D, Fredberg JJ. 2001 Scaling the
microrheology of living cells. Phys. Rev. Lett. 87,
Artn 148102. (doi:10.1103/Physrevlett.87.148102)
Kollmannsberger P, Fabry B. 2009 Active soft glassy
rheology of adherent cells. Soft Matter 5, 1771–
1774. (doi:10.1039/B820228a)
Hoess A, Thormann A, Friedmann A, Heilmann A.
2012 Self-supporting nanoporous alumina
membranes as substrates for hepatic cell cultures.
J. Biomed. Mater. Res. A 100A, 2230–2238.
(doi:10.1002/jbm.A.34158)
Wu XH, Wang SF. 2012 Regulating MC3T3 –E1 cells
on deformable poly(epsilon-caprolactone)
honeycomb films prepared using a surfactant-free
breath figure method in a water-miscible solvent.
ACS Appl. Mater. Inter. 4, 4966–4975. (doi:10.1021/
am301334s)
Korchev YE, Gorelik J, Lab MJ, Sviderskaya EV,
Johnston CL, Coombes CR, Vodyanoy I, Edwards CRW.
2000 Cell volume measurement using scanning ion
conductance microscopy. Biophys. J. 78, 451–457.
(doi:10.1016/S0006-3495(00)76607-0)
Laudadio RE, Millet EJ, Fabry B, An SS, Butler JP,
Fredberg JJ. 2005 Rat airway smooth muscle cell
during actin modulation: rheology and glassy
dynamics. Am. J. Physiol. Cell Physiol. 289,
C1388–C1395. (doi:10.1152/ajpcell.00060.2005)
12
J. R. Soc. Interface 12: 20141057
20. Schneider D, Baronsky T, Pietuch A, Rother J,
Oelkers M, Fichtner D, Wedlich D, Janshoff A. 2013
Tension monitoring during epithelial-tomesenchymal transition links the switch of
phenotype to expression of moesin and cadherins in
NMuMG Cells. PLoS ONE 8, ARTN e80068. (doi:10.
1371/journal.pone.0080068)
21. Alcaraz J, Buscemi L, Grabulosa M, Trepat X,
Fabry B, Farre R, Navajas D. 2003 Microrheology of
human lung epithelial cells measured by atomic
force microscopy. Biophys. J. 84, 2071–2079.
(doi:10.1016/S0006-3495(03)75014-0)
22. Wegener J, Keese CR, Giaever I. 2000 Electric cellsubstrate impedance sensing (ECIS) as a noninvasive
means to monitor the kinetics of cell spreading to
artificial surfaces. Exp. Cell Res. 259, 158 –166.
(doi:10.1006/excr.2000.4919)
23. Thakur S, Massou S, Benoliel AM, Bongrand P,
Hanbucken M, Sengupta K. 2012 Depth matters:
cells grown on nano-porous anodic alumina respond
to pore depth. Nanotechnology 23, 255101. (doi:10.
1088/0957-4484/23/25/255101)
24. Janshoff A, Lorenz B, Pietuch A, Fine T, Tarantola M,
Steinem C, Wegener J. 2010 Cell adhesion on
ordered pores: consequences for cellular elasticity.
J. Adh. Sci. Technol. 24, 2287 –2300. (doi:10.1163/
016942410X508028)
25. Svitkina T. 2010 Imaging cytoskeleton components by
electron microscopy. In Cytoskeleton methods and
protocols. Methods in molecular biology, vol. 586 (ed.
RH Gavin), pp. 187–206. Clifton, NJ: Humana Press.
26. Salbreux G, Charras G, Paluch E. 2012 Actin cortex
mechanics and cellular morphogenesis. Trends Cell
Biol. 22, 536 –545. (doi:10.1016/j.tcb.2012.07.001)
27. Sandmann R, Henriques SSG, Rehfeldt F, Koster S.
2014 Micro-topography influences blood platelet
spreading. Soft Matter 10, 2365–2371. (doi:10.
1039/C3SM52636D)
28. Pietuch A, Janshoff A. 2013 Mechanics of spreading
cells probed by atomic force microscopy. Open Biol.
3, 130084. (doi:10.1098/rsob.130084)
29. Pietuch A, Bruckner BR, Fine T, Mey I, Janshoff A.
2013 Elastic properties of cells in the context of
rsif.royalsocietypublishing.org
11. McNamara LE, McMurray RJ, Biggs MJP, Kantawong F,
Oreffo ROC, Dalby MJ. 2010 January 1, 2010
nanotopographical control of stem cell differentiation.
J. Tissue Eng. 1, 120623. (doi:10.4061/2010/120623)
12. Teo BKK, Wong ST, Lim CK, Kung TYS, Yap CH,
Ramagopal Y, Romer LH, Yim EKF. 2013
Nanotopography modulates mechanotransduction of
stem cells and induces differentiation through focal
adhesion kinase. ACS Nano 7, 4785 –4798. (doi:10.
1021/Nn304966z)
13. Rossier OM et al. 2010 Force generated by
actomyosin contraction builds bridges between
adhesive contacts. EMBO J. 29, 1055 –1068.
(doi:10.1038/emboj.2010.2)
14. Vedula RK, Hirata H, Nai MH, Brugues A, Toyama Y,
Trepat X, Lim CT, Ladoux B. 2014 Epithelial bridges
maintain tissue integrity during collective cell
migration. Nat. Mater. 13, 87 –96. (doi:10.1038/
nmat3814)
15. du Roure O, Saez A, Buguin A, Austin RH,
Chavrier P, Silberzan P, Ladoux B. 2005 Force
mapping in epithelial cell migration. Proc. Natl
Acad. Sci. USA 102, 2390 –2395. (doi:10.1073/pnas.
0408482102)
16. Frese D, Steltenkamp S, Schmitz S, Steinem C. 2013
In situ generation of electrochemical gradients
across pore-spanning membranes. RSC Adv. 3,
15 752–15 761. (doi:10.1039/C3ra42723d)
17. Butt HJ, Cappella B, Kappl M. 2005 Force
measurements with the atomic force microscope:
technique, interpretation and applications. Surf Sci.
Rep. 59, 1–152. (doi:10.1016/j.surfrep.2005.08.003)
18. Alcaraz J, Buscemi L, Puig-de-Morales M, Colchero J,
Baro A, Navajas D. 2002 Correction of
microrheological measurements of soft samples
with atomic force microscopy for the hydrodynamic
drag on the cantilever. Langmuir 18, 716–721.
(doi:10.1021/la0110850)
19. Sen S, Subramanian S, Discher DE. 2005 Indentation
and adhesive probing of a cell membrane with
AFM: theoretical model and experiments.
Biophys. J. 89, 3203 –3213. (doi:10.1529/biophysj.
105.063826)