Overflow Microfluidic Networks for Open and
Closed Cell Cultures on Chip
Robert D. Lovchik,†Fabio Bianco,‡Noemi Tonna,‡Ana Ruiz,§,|Michela Matteoli,§,|and
IBM Research—Zurich, Sa ¨umerstrasse 4, 8803 Ru ¨schlikon, Switzerland, Neuro-Zone s.r.l., via Fratelli Cervi 93,
20090 Segrate, Italy, Department of Pharmacology, CNR Institute of Neuroscience, University of Milano, via Vanvitelli
32, 20129 Milano, Italy, and Fondazione Filarete, viale Ortles 22/4, 20139 Milano, Italy
Microfluidics have a huge potential in biomedical re-
search, in particular for studying interactions among cell
populations that are involved in complex diseases. Here,
we present “overflow” microfluidic networks (oMFNs) for
depositing, culturing, and studying cell populations, which
are plated in a few microliters of cell suspensions in one
or several open cell chambers inside the chip and sub-
sequently cultured for several days in vitro (DIV). After
the cells have developed their phenotype, the oMFN is
closed with a lid bearing microfluidic connections. The
salient features of the chips are (1) overflow zones around
the cell chambers for drawing excess liquid by capillarity
from the chamber during sealing the oMFN with the lid,
(2) flow paths from peripheral pumps to cell chambers
and between cell chambers for interactive flow control,
(3) transparent cell chambers coated with cell adhesion
molecules, and (4) the possibility to remove the lid for
staining and visualizing the cells after, for example,
fixation. Here, we use a two-chamber oMFN to show the
activation of purinergic receptors in microglia grown in
one chamber, upon release of adenosine triphosphate
(ATP) from astrocytes that are grown in another chamber
and challenged with glutamate. These data validate oM-
FNs as being particularly relevant for studying primary
cells and dissecting the specific intercellular pathways
involved in neurodegenerative and neuroinflammatory
Microfluidics are emerging as invaluable tools for research on
cells for a number of reasons.1-3First, they enable experiments
on cells to be performed with well-defined chemical and topo-
graphical environments.4-9Second, they can reduce the number
of cells needed for experiments.1Third, they can be used for
experiments in which a precise number of cells can be stimulated
and observed accurately.10-12Last, they permit experiments to
be performed in parallel and/or with high throughput.13Despite
these possibilities, microfluidics are not well established for daily
research work on cells, as they lack user friendliness and require
specific and often cumbersome methods for depositing and
culturing cells. This is further complicated when taking into
consideration more relevant cell systems such as primary cells,14,15
which may require longer culturing times before being ready for
specific assays (i.e., primary neurons need to be cultured several
days in vitro (DIV) before establishing mature synaptic contacts).
Although progress on sealing a Petri dish with a hybrid microf-
luidic-vacuum platform was recently demonstrated by Chung et
al.,9interfacing “open” cell cultures with microfluidics for cell
cultures that are only a few microliters in volume remains a great
challenge. Similarly, depositing cells on a treated glass slide and
gluing a lid having ports onto the slide provides a method for
enclosing living cells inside a microfluidic flow path.16Such a
method would be hard to implement with multiple types of cells
and can be simplified if the gluing step could be omitted. Finally,
Shuler et al. developed powerful micro cell culture analogs to
assess the cytotoxicity of drugs on cells.17In this work, cells were
mixed with a gel at a temperature of 4 °C, then added to the micro
cell culture analogs, and entrapped in the gel by raising its
temperature to 37 °C. Using a low temperature (especially if
primary cells are used) as well as challenges with air bubbles
* Towhomcorrespondence shouldbe addressed.E-mail:emd@
§CNR Institute of Neuroscience.
(1) Persidis, A. Nat. Biotechnol. 1998, 16, 488–489.
(2) Folch, A.; Toner, M. Annu. Rev. Biomed. Eng. 2000, 2, 227–256.
(3) Prokop, A.; Prokop, Z.; Schaffer, D.; Kozlov, E.; Wikswo, J.; Cliffel, D.;
Baudenbacher, F. Biomed. Microdevices 2004, 4, 325–339.
(4) Takayama, S.; Ostuni, E.; LeDuc, P.; Naruse, K.; Ingber, D. E.; Whitesides,
G. M. Nature 2001, 411, 1016.
(5) Meyvantsson, I.; Beebe, D. J. Annu. Rev. Anal. Chem. 2008, 1, 423–444.
(6) Breslauer, D. N.; Lee, P. J.; Lee, L. P. Mol. BioSyst. 2006, 2, 97–112.
(7) Figallo, E.; Cannizzaro, C.; Gerecht, S.; Burdick, J. A.; Langer, R.; Elvassore,
N.; Vunjak-Novakovic, V. Lab Chip 2007, 7, 710–719.
(8) El-Ali, J.; Sorger, P. K.; Jensen, K. F. Nature 2006, 442, 403–411.
(9) Chung, B. G.; Park, J. W.; Hu, J. S.; Huang, C.; Monuki, E. S.; Jeon, N. L.
BMC Biotechnol. 2007, 7, 60.
(10) Taylor, A. M.; Rhee, S. W.; Tu, C. H.; Cribbs, D. H.; Cotman, C. W.; Jeon,
N. L. Langmuir 2003, 19, 1551–1556.
(11) Lovchik, R. D.; Bianco, F.; Matteoli, M.; Delamarche, E. Lab Chip 2009,
(12) Lucchetta, E. M.; Lee, J. H.; Fu, L. A.; Patel, N. H.; Ismagilov, R. F. Nature
2005, 434, 1134–1138.
(13) Faley, S.; Seale, K.; Hughey, J.; Schaffer, D. K.; VanCompernolle, S.;
McKinney, B.; Baudenbacher, F.; Unumatz, D.; Wikswo, J. P. Lab Chip
2008, 8, 1700–1712.
(14) Rhee, S. W.; Taylor, A. M.; Tu, C. H.; Cribbs, D. H.; Cotman, C. W.; Jeon,
L. N. Lab Chip 2005, 5, 102–107.
(15) Millet, L. J.; Stewart, M. E.; Sweedler, J. V.; Nuzzo, R. G.; Gilette, M. U.
Lab Chip 2007, 7, 987–994.
(16) Morel, M.; Bartolo, D.; Galas, J.-C.; Dahan, M.; Studer, V. Lab Chip 2009,
(17) Sung, J. H.; Shuler, M. L. Lab Chip 2009, 9, 1385–1394.
Anal. Chem. 2010, 82, 3936–3942
10.1021/ac100771r 2010 American Chemical Society
Published on Web 04/14/2010
Analytical Chemistry, Vol. 82, No. 9, May 1, 2010
and leaks when screwing a cover on the micro cell culture analogs
also entails technical challenges.
CELLULAR PATHWAYS IN MICROFLUIDICS
In addition to the above-mentioned advantages and drawbacks
of microfluidics when working with cells, entirely new types of
experiments become possible when distinct cell populations are
placed within a microfluidic device, which allows one to simulta-
neously carry out a functional analysis on stimulated versus control
cells in a miniaturized, highly controlled microenvironment.
Besides enabling the analysis of cellular behaviors, such as cell
differentiation, motility, or response to external trophic or toxic
stimuli, microfluidics open the possibility of dissecting cell-to-cell
communication in complex intercellular scenarios, such as those
represented by the nervous or the immune systems, where the deep
understanding of the mechanisms involved in the onset of diseases.
which contribute to neuronal death in neurodegenerative diseases.
The latter are among the most devastating diseases, which account
for 30% of healthcare costs in industrialized countries and an
estimated 35% of the disease burden within the seven major pharma
markets. Notably, today there are no drugs available that can arrest
or reverse neurodegeneration.
In the last few years, it has become clear that dysfunctions of
the synapse (the functional contact between neurons) are central
to the etiology and progression of a wide range of neurological
and psychiatric disorders, including neurodegenerative diseases,
schizophrenia, autism, depression, and many others, which can
therefore be collectively regarded as synaptopathies. Besides
specific defects in neuronal proteins, activation of immune mech-
anisms and inflammation play a crucial role in synaptopathies.
Bidirectional functional interactions among neurons, astrocytes,
and microglia, through the release of soluble chemical mediators,
govern both the sequence of inflammatory events (cascades of
inflammatory mediators) and the pathological outcome (damage
or absence of damage to the neurons). However, the pathways of
these inflammatory signaling cascades remain cryptic, the major
mediators unknown, and the sequential flux of molecular informa-
tion among the different cell types still undefined. The scientific
community is currently facing the challenge of dissecting the
neuroinflammatory cascades among microglia, astrocytes, and
neurons by breaking down the cellular networks and controlling
the flow of inflammatory mediators among these cells. This is
crucial for understanding the molecular mechanisms that are at
the basis of neuronal damage, in order to define smarter strategies
for tackling neurodegenerative processes.
CONCEPT OF OVERFLOW MICROFLUIDIC
Here, we describe a concept for microfluidics in which cells
are plated and cultured in an open part of the chip and studied
under “microfluidic conditions” after sealing the open chip with a
lid. Plating and culturing the cells with this chip is extremely
simple: the surface of the chip is treated with cell-adhesion
molecules such as fibronectin or poly-L-lysine; then, a few
microliters of cell suspension and medium are added to the open
chip, and the open chip is placed in a cell incubator to provide
cells with the appropriate temperature and level of CO2to enable
development of their phenotype. Whenever needed, the culture
medium is changed by pipetting. In this way, cells can be
maintained for any duration of time akin to cells cultured in
T-flasks. We first illustrate how to deposit one type of cell on
a chip and then demonstrate that two types of primary cells
can be deposited on a chip and a pathway between the cells
We call the open chip an overflow microfluidic network
(oMFN), the design of which is shown in Figure 2a. The oMFN
is molded in poly(dimethylsiloxane) (PDMS), 150 µm deep, and
possesses a central cell chamber having a volume of ∼0.5 µL.
The chamber is connected to inlet and outlet microchannels via
flow-distributing structures.11The central features of the oMFN
are the “overflow” zones containing capillary microstructures,19,20
which are adjacent to the cell chamber.
After cells had sufficient time to grow and develop their
phenotype in the cell chamber, a Si lid having vias (800 µm in
diameter) and fittings is placed by hand over the oMFN using
alignment marks as visual aids. The lid (4-15 g in weight,
depending on the type of ports and fittings used) is approached
(18) Bjornsson, C. S.; Lin, G.; Al-Kofahi, Y.; Narayanaswamy, A.; Smith, K. L.;
Shain, W.; Roysam, B. J. Neurosci. Methods 2008, 170, 165–178.
(19) Juncker, D.; Schmid, H.; Drechsler, U.; Wolf, H.; Wolf, M.; Michel, B.; de
Rooij, N.; Delamarche, E. Anal. Chem. 2002, 74, 6139–6144.
(20) Zimmermann, M.; Schmid, H.; Hunziker, P.; Delamarche, E. Lab Chip
2007, 7, 119–125.
Figure 1. Different types of cells in the brain. (a) Section of a rat
brain tissue, stained with fluorescence dyes to visualize key cells and
structures. Reprinted with permission from ref 18. Copyright 2008
Elsevier B.V. (b) Simplified illustration of the complex interactions
between some of the cell populations involved in neurodegenerative
diseases (N: neurons, A: astrocytes, M: microglia, O: oligodendro-
cytes, C: blood capillary).
Analytical Chemistry, Vol. 82, No. 9, May 1, 2010
slowly until it touches the droplet of cell suspension located in
the cell chamber. It is then allowed to come into contact with the
oMFN under its own weight. Sometimes, gently pressing the
middle of the lid using tweezers is necessary to trigger contact
between the oMFN and the lid. During this step, excess liquid in
the cell chamber flows over the chamber sidewall and connects
with the overflow areas, Figure 2b. The liquid is pulled away from
the chamber owing to the capillary pressure generated by the
microstructures arrayed in the overflow areas, and the lid
conforms to the elastomeric oMFN. Not having the overflow areas
results in having a thin film of liquid squeezed between the MFN
and the lid, which makes the surfaces slide against each other
unpredictably during assembly and the interstitial liquid will fill
remote microfluidic structures, thereby entrapping air bubbles.
The oblong microstructures visible in Figure 2b have a length of
100 µm, a width of 60 µm, an intrarow spacing of 60 µm, and inter-
row spacing of 30 µm. These structures and their lattice were
selected from earlier work on capillary pumps for microfluidic
diagnostic chips.20,21Excess liquid quickly fills the overflow zone
with a straight filling front, thereby preventing air entrapment in
the overflow area and pushing air toward the periphery of the
oMFN. One overflow area can accommodate 4.35 µL of liquid; by
connecting these areas to the alignment marks using microchan-
nels, venting and draining of the overflow areas are achieved.
Figure 2c,d, respectively, shows a photograph of a PDMS oMFN
and of an oMFN sealed with a Si lid having ports. Si was selected
as a material for the lid because of its mechanical stability, flatness
(the face of the Si wafer in contact with PDMS is polished),
chemical resistance, and compatibility with adhesives such as
those used for bonding ports. The footprint of the ports and
tubings, visible in Figure 2d, can be reduced by soldering metallic
wires directly to the lid if needed.22
Chemicals, Biochemicals, and Buffers. Fibronectin from
human plasma, poly-L-lysine, Trypan blue, trypsin, ethylenedi-
aminetetraacetic acid (EDTA), and phosphate buffered saline
(PBS) were purchased from Sigma Aldrich (Buchs, Switzer-
land). Yo-Pro 1 dye, NBD, and Falloidin T-red (1:200) were
purchased from Invitrogen (Milano, Italy). Beta3 tubulin
monoclonal antibodies (1:400, Promega, Italy), synaptotagmin
(1:100), and v-Glut1 antibodies (1:100) were purchased from
Synaptic Systems (Gottingen Germany); monoclonal anti-glial
fibrillary acidic protein produced in mouse (GFAP) (1:400) was
from Sigma, and ionized calcium binding adaptor molecule 1
antibody from rabbit (IBA) (1:100) was from Duotech. Purified
water was produced using a Simplicity 185 system from
Millipore (Zug, Switzerland).
Fabrication of the PDMS oMFNs and Si Lids. The oMFNs
were molded in PDMS using Si molds made from 4-in. Si wafers
(Siltronix, Geneva, Switzerland). The microfabrication of the molds
was done using standard photolithography and photoplotted
polymer masks (64′000 dpi, Zitzmann GmbH, Eching, Germany)
and deep reactive ion etching (STS ICP, Surface Technology
Systems, Newport, UK). The depth of the structures was 150 µm,
and ∼5 nm of a fluorinated material was deposited at the end of
the etching process to act as an antiadhesive layer. Molding the
oMFNs was done by pouring Sylgard 184 prepolymers (Dow
Corning, Midland, MI) onto a Si mold that was placed in a clean
Petri dish. The ratio of polymer to curing agent was 10:1. After a
polymerization time of at least 12 h at 60 °C, the PDMS oMFNs
were peeled off the mold and separated using a scalpel. The
oMFNs were stored in a clean Petri dish with the structures in
contact with the bottom of the dish until they were used for
The Si lids were fabricated using the same method and
equipment than that for the Si molds. The vias and alignment
marks were etched through a 500 µm thick 4 in. Si wafer, and
Nanoport Assemblies from Upchurch Scientific (Ercatech, Bern,
Switzerland) were centered over the vias and glued to the backside
of the Si lids.
Peripheral Equipment. Tubes and standard fittings were
purchased from Upchurch Scientific. High precision motorized
syringe pumps (Cetoni GmbH, Korbussen, Germany) were used
to move liquids through closed oMFNs and were equipped with
50 µL syringes from Hamilton (Bonaduz, Switzerland). The closed
oMFNs were placed onto the stage on an inverted microscope
(Nikon Eclipse TE300, Egg, Switzerland) for perfusion with
medium and staining procedures. The microscope was equipped
with a camcorder (Sony CDR SR100E, Schlieren, Switzerland).
Some images were taken using an upright microscope (Eclipse
90i, Egg, Switzerland) equipped with a color CCD camera (Nikon
DXM 1200C, Egg, Switzerland). Closing procedures with colored
liquids were recorded with the camcorder mounted to a Leica
stereomicroscope. The confocal images from open oMFNs were
acquired using a LSM 510 Meta confocal microscope (Zeiss,
Germany) and a Nikon (Tokyo, Japan) 20× objective with a
Cell Deposition, Cultivation, and Staining. Depending on
the cells used for the experiment, the oMFNs were coated with
fibronectin or poly L-lysine. Fibronectin coating was done by
covering an oMFN with a sterile solution of 50 µg mL-1
fibronectin in PBS for 30 min at room temperature. The oMFN
was then rinsed with PBS and water and dried under a stream
(21) Gervais, L.; Delamarche, E. Lab Chip 2009, 9, 3330–3337.
(22) Murphy, E. R.; Inoue, T.; Sahoo, H. R.; Zaborenko, N.; Jensen, K. F. Lab
Chip 2007, 7, 1309–1314.
Figure 2. Design (a and b) and photographs of a PDMS oMFN
before (c) and after (d) assembly with the Si lid bearing ports, fittings,
Analytical Chemistry, Vol. 82, No. 9, May 1, 2010
of N2. Coating an oMFN with poly L-lysine required a brief
plasma treatment of the oMFN. This was done in a plasma
generator (Technics Plasma 100-E, Florence, KY) in an air
plasma at 200 W (coil power) for 30 s. Directly after the plasma
treatment, the oMFN was coated with a 0.5 mg mL-1solution
of poly L-lysine in a borate buffer and incubated at room
Cell deposition was done by placing a 5 µL drop of fresh
medium onto the chamber of a prepared oMFN and pipetting 2
µL of cell suspension (ca. 1500 cells) into the drop of medium.
The cells sedimented to the bottom of the chamber and were
allowed to attach to the surface by placing the oMFNs in a cell
incubator (37 °C, 5% CO2). To avoid evaporation of the small
drops, 100% humidity was maintained by placing some water
near the oMFNs located in Petri dishes. Primary cells required
an exchange of medium after 24 h.
HeLa cells were cultivated in DMEM medium. The cell
suspension used for seeding in oMFNs contained ∼105cells
mL-1. After 1 day of incubation, the oMFN was closed and
perfused with fresh medium at a flow rate of 2 µL min-1for
5 min. The cells were exposed for 2 min to a Trypan Blue
solution perfusing the chamber at a flow rate of 2 µL min-1.
After flushing the cells with fresh medium for another 2 min,
images were taken and the viability of the cells checked
using an upright microscope.
SY5Y neuroblastoma were obtained as a kind gift from Dr.
Fornasari (CNR Institute of Neuroscience, Milano Italy) and
cultivated in RPMI medium (Sigma, Italy) supplemented with 10%
fetal bovine serum (Invitrogen, Italy), 1% penicillin/streptomycin
(Invitrogen Italy), and 1% L-glutamine (Invitrogen, Italy) at 37 °C
and 5% CO2. The cell suspension used for seeding contained
∼105cells mL-1. After 1 day of incubation, the oMFN was
closed and perfused with fresh medium at a flow rate of 2 µL
min-1for 5 min. The cells were then exposed for 2 min to a 1
mM solution of FM 143 (Sigma-Aldrich, Italy) by perfusing the
chamber at a flow rate of 2 µL min-1. After washing the cells
with fresh medium for another 2 min at the same flow rate,
images were taken using an upright microscope.
Primary neuronal cultures were prepared from embryonic
(E18) rat hippocampus (Sprague Dowley rats, Charles River,
Calco, Italy) as previously described.23The medium was ex-
changed after 1 day of incubation. The oMFN was closed after 6
days, and fresh medium perfused at a flow rate of 2 µL min-1for
5 min. The cells were exposed for 2 min to a 1 mM solution of
FM 143 by perfusing the chamber at a flow rate of 2 µL min-1.
After washing the cells with fresh medium for another 2 min
at the same flow rate, images were taken using an upright
Primary astrocytes and microglia cortical mixed-glia cultures
from rat pups (P2) were obtained using previously described
methods.24Briefly, after dissection, the cortices were dissoci-
ated by treatment with trypsin (0.25% for 10 min at 37 °C)
followed by fragmentation with a fire-polished Pasteur pipet.
The dissociated cells were plated onto PDMS oMFNs (∼2000
cells per chamber), and the cultures were grown in Minimum
Essential Medium (Invitrogen, Italy) supplemented with 10%
horse serum (Euroclone Ltd., UK) and glucose at a final
concentration of 5.5 g L-1. Primary microglia were obtained
by shaking the mixed glial cultures 4 times at 150 rpm for
5 min with 2 min pauses in between.
Immunofluorescence: Cells were fixed for 15 min at room
temperature with 4% paraformaldehyde in 0.12 M phosphate buffer
containing 0.12 M sucrose. The fixed cells were detergent-
permeabilized and labeled with primary antibodies followed by
fluorescently labeled conjugated secondary antibodies. The oM-
FNs were placed in 70% glycerol in PBS containing 1 mg mL-1
phenylendiamine. The images were acquired using a BioRad
MRC-1024 confocal microscope equipped with the LaserSharp
RESULTS AND DISCUSSIONS
Overflow Microfluidic Network for One Cell Population.
Figure 3 shows the sealing of an oMFN, the cell chamber of
which was covered with ∼5 µL of water colored in red. The
ports on the lids were connected via tubings to a reservoir (inlet
port) and a high-precision computer-controlled pump (outlet
port, Cetoni GmbH, Korbussen, Germany). The ports and
tubings were filled with water prior to assembling the lid and
oMFN to prevent the trapping of air in the microfluidic flow
paths. Positioning the lid and having all excess liquid moved
to the overflow areas took 5-10 s. After assembly, the liquid
(23) Verderio, C.; Coco, S.; Bacci, A.; Rossetto, O.; De Camilli, P.; Montecucco,
C.; Matteoli, M. J. Neurosci. 1999, 19, 6723–32.
(24) Calegari, F.; Coco, S.; Taverna, E.; Bassetti, M.; Verderio, C.; Corradi, N.;
Matteoli, M.; Rosa, P. J. Biol. Chem. 1999, 274, 22539–22547.
Figure 3. Frames from a video showing (a) the assembly of a Si lid
onto an oMFN containing ∼5 µL of red-colored water, (b) the sealed
chip with water distributed inside the microfluidic flow path and in
some of the overflow areas, and (c) the exchange of the red-colored
water by a green-colored water by pumping liquid through one port
of the sealed chip.
Analytical Chemistry, Vol. 82, No. 9, May 1, 2010
(colored in green) was pumped into the assembled microfluidic
chip at a flow rate of 2 µL min-1. Liquid was pulled through
the chamber to create a negative pressure in the oMFN,
and flow rates between 1 and 3 µL min-1were typically used.
Prior to depositing cells, PDMS oMFNs were treated with
a solution of polylysine (neurons) or fibronectin (other cells).
PDMS oMFNs were exposed for 30 s to an O2-based plasma
to oxidize their surface and create the negative charges
needed to immobilize polylysine via electrostatic interactions.
Cell lines (HeLa and SY5Y neuroblastoma) and primary cells
from the central nervous system (microglia, astrocytes, and
neurons) were deposited in oMFNs. First, 5 µL of culture medium
was placed on the cell chamber, and then, 2 µL of cell suspension
was pipetted into the medium. Cells reached the surface of the
chamber by sedimentation and were allowed to attach. Incubation
of the cells proceeded for 1 (HeLa, SY5Y) to 6 DIV (primary cells).
The oMFN was then closed as described above, and the morphol-
ogy of cells and functionality of the sealed oMFNs were tested
by staining the cells using vital dyes and using an inverted
microscope. Figure 4a is a brightfield image of HeLa cells in the
closed chamber. Images in Figure 4b,c were taken after cell
membrane live staining in the closed microfluidic system. The
images in Figure 4a-c show that cells grow normally in PDMS
chambers. Figures 4d-f are fluorescence microscope images of
primary cells fixed and retrospectively labeled with cell-specific
marker to show that primary brain cells attach and grow normally
on oMFNs. If needed, the lid can also be made from glass25so as
to visualize cells using an upright microscope or differential
interference contrast microscopy. We were not able to close and
remove the lid repeatedly without compromising the viability of
the cells. We think that the shear stress exerted on the cells by
the liquid pulled with the cover and/or the small volume of liquid
(<1 µL) left in the chamber account for this. Separating the lid
from the oMFN, while pumping medium into one port and having
the second port closed, might solve this problem.
Overflow Microfluidic Network for Two Cell Populations.
The concept of the oMFN is here extended to a chip having two
cell chambers. Figure 5a shows the photograph of a two-chamber
oMFN assembled with a Si lid having 6 ports for fluidic connec-
tion. The oMFN has a footprint of 32 × 26 mm2, and its two cell
chambers are only 3 mm apart to enable visualization of both
chambers simultaneously using a 4× microscope objective. The
layout of the channels and corresponding ports allows liquids
to be drawn sequentially, or independently if needed, through
the chambers.26The entire overflow zone can accommodate
up to 48 µL of excess liquid. Typically, ∼7 µL of cell suspension
is used to plate and culture the cells in each chamber. Then,
the chip is closed with the lid, and cell culture medium is
pumped through the chambers using ports 1 (inlet) and 6
A known biochemical pathway of intercellular communication
between primary astrocytes and microglia was performed to
validate the system: primary cortical astrocytes in the first
chamber were exposed to 50 µM glutamate, a condition which is
known to induce the regulated release of adenosine triphosphate
(ATP). Upon transfer into the second chamber, the gliotransmitter
activated purinergic receptors on microglia cells. For these
experiments, astrocytes were cultivated in the left cell chamber
for 3 DIV, while microglia were plated in the right chamber for 1
DIV. The oMFN was then closed using the lid, and regular
morphology of the plated cells was assessed by inverted micro-
scope observation. A growth medium was perfused for ∼3 min at
a flow rate of 1 µL min-1from port 1 to 6 before introducing a
buffered solution containing 50 µM glutamate and 20 µg mL-1
propidium iodide (PI) dye for 20 min at a flow rate of 1 µL
min-1. It is well-known that glutamate-induced release of ATP
results in activation of purinergic receptors on microglia cells
and, in particular, activation of ionotropic P2X7 receptor.27
Prolonged activation of P2X7 is known to induce the formation
of an aspecific pore, measured by dye uptake, which is known
to induce sustained intracellular calcium levels in microglia
leading to the release of neuroinflammatory mediators.28In
our experiment, we show that ATP conveying in the second
chamber as a consequence of astrocyte stimulation with
glutamate leads to an uptake of PI by microglia. The micro-
graphs in Figure 5b-g show images (10× objective) of microglia
after three independent experiments. Microglial dye uptake in
Figure 5c indicates cell exposure to ATP, which was released by
astrocytes and convected to the second chamber. Direct PI uptake
by the microglia following glutamate exposure is excluded because
omitting the astrocytes in the first chamber did not lead to staining
of the microglia by PI, Figure 5e. As an additional control
experiment, the astrocytes were stimulated with glutamate, this
time having an addition of oxidized ATP (oATP, 100 µg mL-1) to
the glutamate/PI solution. Oxidized ATP inhibits P2X7 receptor
of microglia, and, as expected, a strong reduction of dye uptake
by microglia is observed (compare Figure 5g with Figure 5c).
Figure 5h shows the percentage of PI positive cells relative to
the number of cells observed in brightfield for the two experi-
ments (with and without receptor inhibition). Moreover, the
oMFN is not permanently bonded to the Si lid, so cells can be
fixed after the experiments by perfusing a 4% paraformaldehyde
(25) Madou, M. Fundamentals of Microfabrication; CRC Press: Boca Raton, FL,
(26) Lovchik, R. D.; Tonna, N.; Bianco, F.; Matteoli, M.; Delamarche, E. Biomed.
Microdevices 2010, 12, 275–282.
(27) Verderio, C.; Matteoli, M. J. Immunol. 2001, 166, 6383–6391.
(28) Bianco, F.; Pravettoni, E.; Colombo, A.; Schenk, U.; Moller, T.; Matteoli,
M.; Verderio, C. J. Immunol. 2005, 174, 7268–7277.
Figure 4. Brightfield and fluorescence microscope images of various
types of cells cultured in oMFNs for 1 DIV (a, b, and d) up to 6 DIV
(c, e, and f).
Analytical Chemistry, Vol. 82, No. 9, May 1, 2010
solution for 15 min followed with PBS and, then, further analyzed
after removal of the lid. The astrocytes and microglia used in the
last experiment (glutamate + oATP, Figure 5g) were stained
specifically for either GFAP or IBA after removal of the lid and
imaged using a fluorescence microscope, Figure 5i,j.
These experiments are significant because they demonstrate
that a pathway between two primary cells can be studied using
an oMFN in the presence of, for example, inhibitors of cellular
receptors. An obvious application of this method is to carefully
screen for neuroprotective action of selected compounds in the
context of synaptopathies. For more complex pathways or experi-
ments requiring very precise stimulation of cell populations, flow
tracers, such as polystyrene beads, can be added.
Primary cultures from the central nervous system, in
contrast to cell lines, represent a physiologically more relevant
cell model, given that they largely maintain the features of their
correlates in situ and, therefore, represent a more reliable
system for investigating neuronal development and molecular
processes of brain diseases. For this reason, leading neurobi-
ology laboratories worldwide use primary cultures to investigate
neuronal development, cell-to-cell signaling, and molecular
mechanisms of brain diseases. Also, primary neuronal cultures
are increasingly used for preclinical studies on potential drug
candidates.29In a recent study, Millet et al. showed that
noncross-linked PDMS oligomers affect the viability of primary
hippocampal neurons that are plated at low density in a closed
microfluidic.15For this reason, culturing cells in an open PDMS
chamber brings an additional benefit when working with
neurons. Overflow MFNs represent a bridge between the
typical workflow in cell biology and the field of microfluidics,
as they combine the convenience of working on primary cell
cultures with the capabilities of microfluidics for high-
throughput, low-volumes, automated, and accurate assays. The
possibility of plating cells in a sterile environment, keeping
them in standard culturing conditions within incubators, and
retrieving them at the right developmental state to use them
in a microfluidic device may serve a broad range of purposes,
such as performing toxicity, motility, and adhesion assays, as
well as cell differentiation studies. We currently are scaling the
concept of oMFNs to having three or more interconnected cell
chambers, in view of depositing different cell populations in
each chamber for dissecting the intercellular flows of informa-
tion occurring among brain cells.27,28This approach might help
(29) Shaughnessy, L.; Chamblin, B.; McMahon, L.; Nair, A.; Thomas, M. B.;
Wakefield, J.; Koentgen, F.; Ramabhadran, R. J. Mol. Neurosci. 2004, 24,
Figure 5. Communication studies between two primary cell populations. (a) Photograph of a two-chamber oMFN sealed with a Si lid (bottom)
having 6 ports. Primary astrocytes (left chamber) and microglia (right chamber) were plated and cultured for 3 and 1 DIV, respectively; the lid
was then assembled and experiments were performed using several chips. (b, c) Brightfield and corresponding fluorescence microscope images
of microglia activated by ATP released from glutamate-stimulated astrocytes in the first chamber; ATP activation of microglia is revealed by the
uptake of fluorescent PI (c). (d, e) Control experiment showing the lack of PI uptake by microglia exposed to glutamate in the absence of
astrocytes in the first chamber. (f, g) Control experiment showing that PI uptake by microglia is strongly reduced by concomitant flushing of the
P2 × 7 receptor blocker oATP. (h) Quantitation of the percentage of cells labeled by fluorescent PI upon the glutamate-ATP astrocyte-microglia
pathway, with and without the P2 × 7 receptor inhibitor oATP in the second chamber. (i, j) Fluorescence microscope images of the primary
astrocytes and microglia used in the control in (f) and (g), after the cells were fixed, the lid removed, and the cells stained using dyes for cell
specific markers (GFAP for astrocytes, IBA for microglia).
Analytical Chemistry, Vol. 82, No. 9, May 1, 2010
identify the mediators and the cellular sources of molecules
involved in neuroinflammation, a process that plays a very active
role in the pathophysiology of progressive neurodegenerative
disorders, such as Alzheimer’s and Parkinson’s diseases, and
synaptopathies in general.
We thank R. Stutz for his help with the fabrication of the Si
lids and molds for the PDMS oMFNs, L. Gervais and C. Verderio
for discussions, and W. Riess and M. Despont for their continuous
support. We are grateful to C. Bjornsson and B. Roysam, who
provided Figure 1a, and E. Lovchik, who drew Figure 1b. This
work was partially supported by CARIPLO No. 2006-0948 to M.M.
and by MIUR art. 11 D.M. No. 593/2000 to M.M. and F.B. The
research leading to these results has received funding from the
European Union Seventh Framework Programme under grant
agreement No. HEALTH-F2-2009-241498 (“EUROSPIN” project)
Received for review March 25, 2010. Accepted April 3,
Analytical Chemistry, Vol. 82, No. 9, May 1, 2010