In recent years, a push towards in vitro models for toxicology and drug development
has arisen partly from the cost associated with animal models, but also from regulation
that requires the reduction and replacement of animal experimentation. To date, the
main focus has been on development of cellular models to replace animal studies where
possible, using traditional cell-biology methods, such as fixed-cell imaging or quantification
of biomolecules from dead cells. These methods are rarely amenable to high-throughput
screening, are generally high-cost and usually require the use of labels such as chromophores
and fluorophores. There is a definite need for the development of new technologies
and materials for monitoring in vitro cell models.
One particularly promising whole-cell based system amenable to high-throughput and
label-free testing for in vitro toxicology and drug screening, is electrical impedance
sensing (or electrochemical impedance spectroscopy (EIS)), and a number of commercially
available EIS-based systems are now on the market.[1] EIS is a technique which has
long been used to determine mechanisms of reactions occurring at electrode/electrolyte
interfaces.[2] EIS was first demonstrated for monitoring cell attachment and spreading
on electrodes in the 1980s, first at fixed frequencies, but later by scanning a range
of frequencies, and may simply be explained as measurement of changes in impedance
due to the presence of cells on or near electrodes.[3–5] Initial work on electrical
cell-substrate impedance sensing (ECIS) demonstrated that this technology could be
used to determine such parameters as distance from the substrate,[6] cell adhesion
and proliferation[4,7] and thus could be used in applications like wound healing,[8]
and cell spreading.[9] Variations on this technology have also been developed by a
number of different groups and have shown significant potential for medium to high-throughput
monitoring of cells in vitro.[10–12] Two different formats exist in EIS-based devices;
the first, the nonplanar format, involves cells that are grown on filters between
the electrodes, whereas the second, the planar format, has cells grown directly on
the electrodes. Although the filter format is compatible with transport assays used
in drug development as it allows sampling from the lower compartment, it is very difficult
to image cells on the filters.[10] Indeed, although all the variations on EIS demonstrate
the potential of electronic monitoring for assessing cell integrity and function,
in general these systems do not allow high-resolution imaging and are thus incompatible
with microscopy systems used for life-sciences research, still recognized as the gold
standard for monitoring cell health and function, particularly at the molecular level.[13]
This highlights a particular drawback with electrical monitoring, which is the inability
to unambiguously correlate a measured electrical signal, especially in non-electrogenic
cells (where a passive electrical resistance is being measured rather than for example
an action potential seen in electrogenic cells such as nerve cells), with an event
occurring in the life cycle of the cell. Advances in time-lapse microscopy and availability
of reagents that allow live–cell imaging have been made that allow for more relevant
capturing of dynamic events occurring in the cell, hitherto impossible with static
imaging techniques;[14] however, a number of disadvantages of optical monitoring technologies
still remain, such as the use of costly chromo-/fluorophores, and labor intensive
protocols. The ability to simultaneously monitor cells both optically and electrically
in vitro is thus a very attractive proposition, with high enough resolution to not
only relate electronic signals to events occurring in the cell, but also to exploit
the complementarity of these techniques, to monitor events that may easily be captured
electronically and not optically or vice versa.
We previously developed an organic electrochemical transistor (OECT) based on poly(3,4-ethylenedioxythiophene):poly(styrenesulfonate)
(PEDOT:PSS), capable of monitoring the integrity of epithelial barrier tissue cells
grown on filters, with unprecedented temporal resolution and high sensitivity.[15]
This device has been validated against existing toxicology techniques[16,17] and has
been transitioned towards more high-throughput systems capable of long-term operation
under physiological conditions.[18] However, this device was originally configured
as a non-planar transistor with a commercially available Transwell filter, and was
therefore incompatible with high-resolution optical monitoring. Herein, we describe
the development of a planar all-polymer transistor based on the conducting polymer
PEDOT:PSS, compatible with low-cost production techniques such as ink-jet printing,
or roll-to roll processing. Further, we take advantage of the optical transparency
of PEDOT:PSS to demonstrate simultaneous electronic and optical monitoring of epithelial
cells in vitro, with acquisition of high-resolution time-lapse images. We highlight
the complementarity of these techniques, illustrating the advantage of electronic
monitoring over optical monitoring in certain cases, and finally, demonstrate an acute
example of how these complementary techniques lead us from the observation of an electronic
signal in sub-confluent (before the cells have completely covered the surface) epithelial
cell monolayers, to investigate the formation of structures important for ion-regulation
in these cell types.
The planar OECT used for these studies is illustrated in Figure
1
. In the current configuration both the channel and the gate consists of PEDOT:PSS,
although gold was used as a contact material (Figure 1a). Wells were either 3D printed
or made of polydimethylsiloxane (PDMS) and affixed to the device for cell growth.
The mechanism of the OECT for monitoring barrier tissue integrity has been described
previously;[15] however, in the previous case the barrier tissue cells were grown
on a filter positioned between a top Ag/AgCl gate and the channel consisting of PEDOT:PSS.
In the current device, barrier tissue cells were seeded directly into a well on top
of the planar device. The well was designed to cover not only a section of the channel
but also a section of the gate (Figure 1a). Upon application of a positive gate voltage
(V
GS = 0.3V), the cell monolayer covering the channel reduces the flux of cations from
the media to the channel. This restriction of cation flow results in slower de-doping
of the channel and thus a modification of the channel current (Figure 1b).[15] In
contrast to the previous configuration, the gate is now located below the cell monolayer.
In agreement with previous studies using ECIS, we hypothesize that due to the high
resistance of the nanometer-sized electrolyte filled cleft between the cells and the
substrate, the gate current passes through the cell monolayer into the medium and
then back through the cell monolayer into the channel, rather than laterally between
the gate and channel.[19] In the present system, eight OECTs were fabricated on a
single glass slide to allow for simultaneous measurements of multiple devices. The
glass slide is positioned inside a 3D-printed holder with spring-loaded contact pins
for connections (Figure 1d). This holder is in turn placed on a microscope stage (Figure
1c). The microscope is equipped with a sealed chamber to mimic the conditions found
inside a cell culture incubator.
Figure 1
Simultaneous optical and electronic recording. a,b) Schematic of device which consists
of a PEDOT:PSS channel and gate patterned onto a glass slide. Cells and media were
contained inside a 3D-printed/PDMS well. Under applied gate voltage, cations from
the media penetrate through the cell layer and de-dope the channel. c) Photograph
of the optical microscope equipped for long-term live-cell culture. d) 8 OECTs connected
via a 3D-printed holder adjusted for the microscope, pictured here on the microscope
stage. Connections are made by spring-loaded contacts.
Taking advantage of the optical transparency of the PEDOT:PSS film, high-resolution
images of the cells can be recorded. PEDOT:PSS has an extinction coefficient below
0.3 in the visible range[20] (for comparison, the extinction coefficient of gold is
between 2 and 5 in the visible range depending on the wavelength.[21] In contrast
to existing technologies, high-resolution images can easily be produced in bright
field or fluorescence (see work by Strakosas et al.[22] and Figure S1 in the Supporting
Information for representative images). In contrast, imaging on gold electrodes requires
the use of an upright microscope with a long distance objective, due to the inefficient
transmission of the low intensity fluorescence signal through the gold layer.[23]
We recorded the electrical characteristics of the cell monolayer using the OECT device
(under physiological conditions), while simultaneously collecting optical images in
time-lapse mode. This technique allows precise correlation between the function of
the cell monolayer (electrical characteristics) and the morphology of the cells (optical
characteristics).
Madin Darby Canine Kidney (MDCK-I) cells are a well characterized cell-line which
has been shown to develop properties of transporting epithelia when they grow on membranes[24]
and on solid supports.[25] Using a more-traditional chopstick electrode set-up, MDCK-I
cells have been shown to reach resistances of ca. 1000 Ω cm2 as early as 4–5 days
after seeding.[26] In the case of the OECT, there is no direct measurement of the
ion conductance (or resistance) across the cell monolayer. Rather, the current through
the OECT channel is linearly proportional to the number of ions that have entered
the channel, which in turn determines the speed at which the transistor reaches steady
state.[27] The time constant, τ, is used as a figure of merit for the electrical characterization
of the ion flow through epithelia as measured by the OECT. Extended information on
determination of τ has been reported previously.[18] For electrical characterization
of MDCK-I cells seeded onto the OECT, the channel current was measured upon application
of a positive square-pulse gate voltage. Similar to the charge and discharge of a
capacitor in an RC electronic circuit, the channel current can be fit by an exponential
equation (see materials and methods) and the τ value extracted as a fit parameter.
The data is then normalized between 0 and 1, where 1 is the highest τ value obtained
(corresponding to the lowest amount of ion flux across a functional cell monolayer),
and 0 is the lowest value τ obtained (corresponding to a device with no cells). After
seeding of the MDCK-I cells, the rise of τ (Figure
2
a) started as early as 6 h and plateaued between 35–50 h. Within this time frame,
optical images of the cell monolayer (Figure 2b) showed the development of a regular,
confluent monolayer on the device. Subsequently, the τ value decreased to 0 at approcimately
95 h, where correlation of the signal with the optical images (see the Supporting
Information for a video) shows that many of the cells are in suspension and no longer
attached to the substrate. This is anticipated since the media remains unchanged throughout
the duration of the experiment. A very noticeable and, to our knowledge, previously
unknown phenomenon was the onset of an increase of τ before the cells had completely
covered the surface (were sub-confluent). Indeed, as seen in Figure 2b, the full coverage
of MDCK-I cells on the channel occurred after t = 18 h, corresponding to a normalized
τ of approximately 0.5. These data suggest that the structures which regulate ion
flow in epithelial cells are functional before the cell monolayer is confluent.
Figure 2
Electrical and optical characterization of MDCK-I cells on the planar OECT. MDCK-I
were seeded at 105 cells/cm2 and monitored for 4.5 days. a) Micro-optical images of
MDCK-I on top of the channel area. The darker horizontal line in the middle of the
picture corresponds to the PEDOT:PSS channel. b) Electrical characteristics with a
measurement taken every 3 h. c) Immunofluorescence images taken on sub-confluent MDCK-I
monolayers show presence of ZO-1 and E-cadherin at the cell periphery very early on.
Occludin and Claudin-1 localize in the cytoplasm and just in the cell contacts the
first 3 days. At day 5 all the proteins are present at the cell borders.
The exact sequence of events that occur in the development of the structures that
regulate ion flow in epithelia after the initial contact formed between adjacent cells
remains unknown.[28] The formation of junctions between cells is a highly regulated
process; in epithelial cells, neighboring cells are known to form tight contacts,
necessary for tissue formation. It is thought that spot-like adhesions form first,
at the tip of the initial cell-cell contacts, which then develop into belt-like adherens
junctions (AJ),[29,30] followed by formation of tight junctions (TJ) at the apical
side of the AJ.[31] These TJs have a barrier function to selectively limit the flow
of ions and macromolecules. ZO-1 is a cytoplasmic actin binding protein that plays
an important role in both the AJ and the TJ, being present with E-cadherin in the
AJ and then segregating to co-localize with other proteins such as Occludin and Claudin-1
in the TJ.[32–34] To study TJ function, cells are usually seeded at high concentrations,
and their resistance is measured at designated time points post confluence, which
may be a matter of hours if the density is high enough for the cells to cover a surface
without dividing.[35,36] In the typical Transwell filter set-up, measurement of paracellular
resistance is only possible post-confluency, as otherwise the gaps between groups
of cells provide easy pathways for ion flux. In contrast, in a planar configuration
made possible using the device reported here, it is possible to measure resistance
to ion-flow before the cell monolayer becomes confluent since the cells are grown
directly on the active area of the device. Reports using the ECIS device appear to
indicate that a ‘fully established cell matrix is required before the formation of
tight junctions’, however this is difficult to establish without recourse to simultaneous
optical and electronic monitoring.[19] Our data clearly show that although this may
be true locally (in isolated clusters of cells), functional TJs can form before confluence.
To further investigate the origin of the electrical signal observed with the OECT,
MDCK-I cells were seeded at low densities to achieve incomplete coverage (quantification
of sample coverage is shown in Figure S2 in the Supporting Information: 2% at day
1, 21% at day 3 and 100% at day 5). Representative immunofluorescence images from
cells stained for ZO-1, Claudin-1, Occludin and E-cadherin at days 1, 3 and 5 post
seeding, are shown in Figure 2c. As expected, E-cadherin is present at cell-cell contacts
from day 1, and, although there is also some cytoplasmic staining, it is absent at
the leading or outside edges of the cells. By day 5, E-cadherin appears only at the
cell periphery. A similar pattern is seen for the TJ proteins Occludin, Claudin-1
and ZO-1, although the disappearance of the cytoplasmic fraction appears to be most
rapid for ZO-1, where no cytoplasmic staining is visible at day 3, followed by Occludin
and then Claudin-1. Together with the electrical monitoring data, we believe this
provides clear evidence that ion flow is regulated as soon as the key tight TJ proteins
are present between adjacent cells, and does not depend on a confluent monolayer of
cells to begin regulating ion flow. This does not preclude that the function of the
TJ may evolve post-confluence, and indeed TER measurements are often seen to peak
before stabilizing, possibly indicating a balance between ‘tight’ and ‘leaky’ claudin
expression.[37,38]
The majority of work in the literature on electrical measurements of non-electrogenic
cells grown on surfaces comes from ECIS measurements, which determines the impedance
of cell monolayers grown on electrodes, and can correlate the signal at different
frequencies to a variety of cell parameters by modelling the different elements in
an equivalent circuit.[3,7] The most basic information from ECIS is related to the
coverage of cells on the electrodes which results in changes in impedance.[9] One
possibility for the increase in τ observed during growth of MDCK-I cells on the planar
OECT, is that the signal is due to coverage of the cells on the surface of the device
which could modify the ionic resistance and change the surface capacitance due to
the adhesion of the cells on PEDOT:PSS. To test whether the changes in τ measured
by the OECT upon growth of the cells on the device are attributable to the barrier
function of the cells, or simply to a coverage effect, different types of epithelial
cells were seeded onto devices and allowed to grow to confluence and beyond. As before,
the OECT was used to monitor the electrical behavior of the cells, with simultaneous
optical monitoring using the microscope. Three different cell lines were compared,
Caco-2 cells (a human colon-cancer epithelial cell-line), HEK-293 cells (human embryonic
kidney epithelial cells) and HeLa cells (a human cervical-cancer epithelial cell-line).
Neither HeLa nor HEK-293 have ever, to our knowledge, been reported to have a measurable
TER. HEK-293 cells have been reported to have no expression of Claudins 1–5, nor any
evidence of TJ strands.[39] Caco-2 cells are known to differentiate and form TJs;
however they require approximately 3 weeks to generate a stable TER of ca. 400 Ω cm2
on Transwell filters.[17]
It is clear from Figure
3
a that neither HEK-293 cells nor HeLa cells show a significant increase in the τ value
regardless of the surface coverage. HEK-293 cells show complete, even coverage of
cells on the device by day 3, however there is no change in the τ value. The same
is true for HeLa cells, although the cell monolayer is not as regular as with the
HEK-293 cells, and some gaps are visible between cells. For Caco-2 cells, the τ value
does not change significantly for the first 5 days. At day 6 however, there is a noticeable
upward trend until the end of the experiment at day 8. This is consistent with the
behavior observed when these cells are grown on Transwell filters, reaching average
TER values of 100 Ω cm2 after 5 days, and 300 Ω cm2 after 10 days.[17] An interesting
corollary from this experiment is that Caco-2 cells can differentiate with respect
to their TJ function even on a solid support, without the cues that can be provided
from below thanks to a permeable membrane such as the Transwell filter, something
that had previously been shown for MDCK-I cells.[25]
Figure 3
Electrical signal measured by OECT is related to barrier function of cells. a) Electrical
characteristics of various cell lines measured over 8 days: dark round markers for
Caco-2 cells, dark-grey diamonds markers for HEK-293 cells, and grey triangle markers
for HeLa cells. b) Corresponding micro-optical images of cell coverage over the channel
area.
As a further test of whether the changes in τ measured by the OECT upon growth of
the cells on the device are attributable simply to coverage of the cells on the surface
or could be ascribed to blocking of ions by TJs, we used EGTA (ethylene glycol tetraacetic
acid), a specific Ca2+ chelator which is known to compromise the integrity of TJs
and result in a rapid decrease in the resistance of epithelia to ion flow.[16,24]
Figure
4
a shows the electrical measurement of an MDCK-I monolayer after addition of varying
concentrations of EGTA. At low concentrations of EGTA (1 mM), there was no reduction
of τ and thus the resistance to ion flow was still intact. Since the concentration
of Ca2+ in media is usually 1–2 mM, there may still be enough Ca2+ present for the
barrier function to be maintained. However, for concentrations of 5 mM, 10 mM and
100 mM EGTA, the barrier function of the cells is seen to decrease rapidly, reaching
a normalized value of 0 within 10 min at 100 mM EGTA. In contrast, the corresponding
optical images (Figure 4b) show no visible change in the cell monolayer, regardless
of the concentration of EGTA. Even after 14 min of 100 mM EGTA treatment the cell
monolayer looks completely unchanged and confluent on the device. We conclude thus
that the normalized τ value is attributable to the barrier function of the cells,
and not simply to growth/coverage of the cells on the device (see the Supporting Information
for videos of the data shown here).
Figure 4
Effect of EGTA and trypsin-EDTA on MDCK-I barrier function. a) Electrical characteristics
of MDCK-I barrier tissue measured by an OECT. Normalized τ value of 1 corresponds
to an intact cell monolayer and 0 to a disrupted one. At t = 0 min, EGTA was added
according to the following concentrations: 1 mM for dark round markers, 5 mM for dark-grey
triangle markers, 10 mM for grey diamond markers, and 100 mM for light-grey star markers.
b) Corresponding micro-optical images for EGTA = 100 mM and EGTA = 1 mM. c,d) Effect
of Trypsin-EDTA on MDCK-I barrier: c) Electrical characteristics of MDCK-I monolayers
measured by an OECT. Normalized τ of 1 corresponds to an intact cell monolayer and
0 to a disrupted one. At t = 0 min, Trypsin-EDTA was added according to the following
concentrations: 0.25X for dark round markers, 0.5X for dark-grey triangle markers,
0.75X for grey diamond markers, and 1X for light-grey star markers. d) Corresponding
micro-optical images for trypsin-EDTA 0.5X and 1X.
The integrin proteins, which are the major cell-substrate binding proteins, are known
to be dependent on Mg2+ and therefore the addition of EGTA is unlikely to affect substrate
binding.[40] The addition of trypsin-EDTA however, is known to detach cells from surfaces.
Trypsin is a protease that is routinely used to detach cells from surfaces, by non-specifically
cleaving proteins (at lysine or arginine residues), while EDTA chelates metal ions
such as Ca2+ and Mg2+. As shown in Figure 4c, after trypsin-EDTA treatment the barrier
function of the MDCK-I cells decreases rapidly, with the time required for complete
removal of the barrier properties decreasing with increasing trypsin-EDTA concentration.
In contrast to the EGTA treatment, the addition of trypsin-EDTA results in a detachment
of the cell monolayer, clearly visible from the optical images (Figure 4d). Additionally,
the destruction of the barrier function of the cells precedes the detachment of the
cell monolayer; at 1X trypsin-EDTA, the first gap in the cell monolayer is visible
at 3 min, corresponding to a decrease in the normalized τ value of 25%, whereas the
decrease in electrical signal clearly begins in under a minute.
The OECT is a versatile, robust, low-cost and fully transparent device which can be
used to record barrier tissue properties. The ability to simultaneously record electrical
and optical characteristics is greatly beneficial for monitoring the integrity ofin
vitro models of cells. In this report, we show evidence of one situation in which
the combined electronic and optical information provide new information on the growth
and behavior of cells in culture: notably we have demonstrated that the formation
of TJs in MDCK-I cell layer occurs before full confluence of the cells. This observation
is made possible in the device shown here which allows continuous measurements of
ion flow in epithelial cells pre- and post-confluency coupled with image capture of
cell coverage at high resolution. We clearly show that the measured electrical signal
is due to TJ-related barrier tissue formation rather than simple cell coverage; the
presence of cells on the active area of the OECT does not modify the transistor response
to gate pulse voltage unless the cells present barrier properties. Measurement of
barrier properties by OECTs is a rapid, and potentially low-cost method and thus represents
real future potential as a new technology for high-throughput dynamic monitoring of
live-cell toxicology.
Experimental Section
OECT Fabrication: The conducting polymer formulation consisted of PEDOT:PSS (Heraeus,
Clevios PH 1000), with ethylene glycol (Sigma–Aldrich, 0.25 mL for 1 mL PEDOT:PSS
solution), 4,dodecylbenzenesulfonic acid (DBSA) (0.5 μL/mL), and 3-glycidoxypropyltrimethoxysilane
(GOPS) (10 mg/mL). On a clean glass substrate (75 mm × 25 mm), gold source and drain
contacts were patterned via photolithography, thermal evaporation, and subsequent
lift-off. Photoresist S1813 (MicroChem Corp.) was spin-coated at 3000 rpm for 30 s
on the glass substrate. Patterns were defined by photolithography (Chrome mask and
Mask Aligner). MF-26A was used as developer. Then, 5 nm and 100 nm of chromium and
gold respectively, were evaporated. Finally, the photoresist was lifted-off in an
acetone bath under sonication for 1 h, which left the substrate with the source and
drain Au contacts only. PEDOT:PSS channel dimensions were patterned using a parylene
peel-off technique described previously,[41] resulting in a PEDOT:PSS channel with
width and length of 200 μm and 4 mm, respectively, and thickness of 460 nm. The PEDOT:PSS
gate was 1 mm width and 4 mm length. Following PEDOT:PSS deposition, devices were
baked for 1 h at 140 °C under atmospheric conditions. Either 3D printed or PDMS wells
of 0.5 cm2 (hole diameter of 0.8 cm) defined the cell-growth area.
OECT Operation: All measurements were done using the PEDOT:PSS film as gate electrode
and cell media (as described below) as the electrolyte. Experiments were performed
in an XL multi SI humidified incubator from PECON GmbH mounted on a microscope Axio
Observer Z1 from Carl Zeiss MicroImaging GmbH. The temperature and CO2 level were
at 37 °C and 5%, respectively. The measurement parameters were chosen to avoid exposing
the cell monolayers to a voltage drop above 0.5 V, as high voltages have been shown
to damage bilayer membranes.[42] The recording of the OECTs was performed using a
Keithley 2612 Source Meter and customized Labview software. OECT data were collected
using the following parameters: V
DS = −0.2 V, V
GS = 0.3 V, V
GS on time = 2 s, off time = 28 s.
Data Analysis: Data analysis was performed using a customized Matlab program to isolate
and fit the time constant for each pulse as described previously.[18] In Figure 2
and 4, the data are then normalized using the following equation: NR = (τ
no cells – τ)/(τ
no cells – τ
cells), where τ
cells refers to the tau value in response to the application of the gate voltage of
a barrier forming monolayer, and τ
no cells refers to the tau value in response to the application of the gate voltage
of no barrier, with the dataset subsequently normalized to [0,1] scale, where 1 corresponds
to intact cell monolayer and 0 to a disrupted one or absence of barrier monolayer.
Cell Culture: MDCK-I cells, Caco-2 cells, and HeLa cells were obtained from HPA culture
collection. Cells were routinely maintained at 37 °C in a humidified atmosphere of
5% CO2, in Dulbecco's modified Eagle mediumDMEM) low glucose (Advanced DMEM Reduced
Serum Medium 1X, Invitrogen) with 2 mM Glutamine (Glutamax-1, Invitrogen), 10% fetal
bovine serum (FBS) (Invitrogen), 0.5% PenStrep (PenStrep 100X, Invitrogen) and 0.1%
Gentamicin (Gentamicin 100X, Invitrogen). HEK-293 cells were a kind gift from Marc
Borsotto (IPMC, Valbonne).
EGTA and Trypsin-EDTA Experiments. MDCK-I cells were grown directly on top of the
OECT. They were maintained inside the incubator for 4 days until full confluence.
Then, the substrate bearing the OECTs seeded with cells was placed onto the stage
of the microscope. After 1 h required for stabilization, the recording of the electrical
signal of the OECT and time-lapse optical recording was started. After approximately
15 min and while continuing the recording, the desired concentration of trypsin-EDTA
or EGTA was added inside the well, which corresponds to time t = 0 min in Figure 4.
Note that trypsin-EDTA 1X corresponds to 0.05% of trypsin with 0.02% EDTA.
Immunofluorescence: Cells were washed three times with cold PBS 1X (Ca2+/Mg2+) pH
7.4, then fixed and permeabilized with a mix of 4% paraformaldehyde, 0.1% Triton X-100
during 3 min at 4 °C. Cells were then incubated with 4% paraformaldehyde for 15 min
at 4 °C, then washed with cold PBS 1X (Ca2+/Mg2+) pH 7.4 and incubated in blocking
solution (3% bovine serum albumin (BSA), 0.3 M Glycine) for 1 h at room temperature.
This was followed by addition of primary antibodies, rabbit anti-ZO-1, rabbit anti-Claudin-1,
mouse anti-E-cadherin, mouse anti-Occludin (Invitrogen) in blocking solution for 1
h at 37 °C. They were washed three times with cold PBS 1X (Ca2+/Mg2+) pH 7.4 and incubated
with the secondary antibodies Alexa Fluor 488 donkey anti-mouse and Alexa Fluor 568
donkey anti-rabbit (Molecular Probes) for 1 h at 37 °C. Finally, the cells were incubated
for 5 min at room temperature with Fluoroshield with DAPI (Sigma–Aldrich), mounted,
and examined using a fluorescence microscope.