Diffusion dependent cell behavior in microenvironments

PAPER
www.rsc.org/loc | Lab on a Chip
Diffusion dependent cell behavior in microenvironments
Hongmei Yu,a Ivar Meyvantsson,a Irina A. Shkelab and David J. Beebeb
Received 30th March 2005, Accepted 4th July 2005
First published as an Advance Article on the web 11th August 2005
DOI: 10.1039/b504403k
Understanding the interaction between soluble factors and cells in the cellular microenvironment
is critical to understanding a wide range of diseases. Microchannel culture systems provide a tool
for separating diffusion and convection based transport making possible controlled studies of the
effects of soluble factors in the cellular microenvironment. In this paper we compare the
proliferation kinetics of cells in traditional culture flasks to those in microfluidic channels, and
explore the relationship between microchannel geometry and cell proliferation. PDMS
(polydimethylsiloxane) microfluidic channels were fabricated using micromolding methods. Fall
armyworm ovarian cells (Sf9) were homogeneously seeded in a series of different sized
microchannels and cultured under a no flow condition. The proliferation rates of Sf9 cells in all of
the microchannels were slower than in the flask culture over the first 24 h of culture. The
proliferation rates in the microchannels then continuously decreased reaching 5% of that in the
flasks over the next 48 h and maintained this level for 5 days. This growth inhibition was
reversible and influenced only by the cell seeding density and the channel height but not the
channel length or width. One possible explanation for the observed dimension-dependent cell
proliferation is the accumulation of different functional molecules in the diffusion dominant
microchannel environment. This study provides insights into the potential effects of the diffusion
of soluble factors and related effects on cell behavior in microenvironments relevant to the
emerging use of microchannel culture systems.
Introduction
Within multi-cellular organisms, individual cells continuously
receive both endogenous and exogenous signals that regulate
their behavior. For almost a century in vitro cell culture
methods have been used to study these signals.1,2 However, it
is well accepted that this monolayer in vitro culture system
exhibits significant differences from the in vivo environment.
In vivo, cells reside in a tissue specific microenvironment and
interact with various microenvironment factors, such as the
extracellular matrix (ECM), other cells, and various soluble
factors; while cells isolated from tissues and cultured in culture
systems (e.g. Petri dishes) are exposed to a bulk culture
environment very different from the in vivo cellular microenvironment.3,4 Therefore, the development of in vitro culture
systems with more appropriate microenvironment–cell interactions is important to increase the utility and relevance of
in vitro studies. Various cell culture techniques have been
developed in an attempt to restore the in vivo conditions, such
as three-dimensional culture, co-culture, cell patterning and
micro-culture systems.5–7 These methods incorporate components of the cellular microenvironment and improve our
understanding of cell behavior in vivo. Although diffusion and
diffusive factors are important aspects of the in vivo microenvironment, particularly with respect to autocrine or paracrine signaling, diffusion issues have seldom been addressed in
a
Department of Biomedical Engineering, University of WisconsinMadison, Madison WI, 53706, USA
Department of Biochemistry, University of Wisconsin-Madison,
Madison WI, 53706, USA
b
This journal is ß The Royal Society of Chemistry 2005
microscale cell culture systems that typically focus on flowing
systems where convective flow dominates. We are particularly
interested in the diffusion process and the effects of secreted
soluble factors in the cell microenvironment.
In an aqueous solution, such as cell culture media,
spontaneously occurring spatial variability (fluctuations) in
temperature, solute concentration, or dissolved gas concentration can lead to surface tension differences at the gas–solution
interface. These fluctuations in turn cause rapid convection
and mass transfer (Marangoni effect). For this reason, convective mixing dominates over diffusion even in the absence of
flow in macro-scale cell culture devices (e.g. Petri dishes).
Thus, any secreted molecules are rapidly distributed over
the entire volume, impairing both autocrine and paracrine
signaling. To rigorously study phenomena that depend upon
diffusion in the cellular microenvironment we use microchannels as the experimental platform. Due to the small size of
these channels and the absence of free interface of the solution
with air, spontaneous convection becomes insignificant, and
mass transport becomes diffusion dominant.
Microfluidic techniques have been successfully applied to
access, control and study the interactions at the cellular scale
because of the unique physical features the micro scale
provides, such as laminar flow, diffusion and surface tension.8
Examples include studying the response of endothelia cells to
the shear stress exerted by laminar flow,9 the chemotaxis of
neutrophiles in gradients of interleukin-8 produced by diffusion between laminar streams,10 and the ligand-dependent
lateral propagation of EGF signaling in a single cell exposed to
a binary local stimuli field produced by parallel laminar flow.11
Lab Chip, 2005, 5, 1089–1095 | 1089
The intelligent use of microfluidic systems and the dominant
phenomena present at the micro scale have provided
insights into cell biology, especially immediate cell responses
to exogenous stimuli. On the other hand, there are many
important cell processes occurring at longer-time scales, such
as cell–cell interaction, cell differentiation, and proliferation in
which endogenous signaling is important. To study these
longer time scale processes using microfluidics techniques, it is
necessary to develop a long-term cell culture platform with a
‘‘homeostasis’’ microenvironment where diffusion is the major
mass transport mechanism and its influences on cell behavior
can be observed. While micro scale culture systems have
existed for some time,7,12 the relationship between their
geometric properties of the micro scale constructs and cell
behavior is largely unstudied. Understanding the interactions
between geometry and behavior is important if micro scale
culture is to become more widely used for basic cell biology
studies.13,14
Static (no flow) cell culture in microchannels allows us to
focus on the diffusion of soluble factors within the microenvironment and their effects on cell behavior. No flow
conditions and the spatial constraints imparted by the channel
walls have two important consequences for the microchannel
environment. First, in the absence of convection secreted
factors diffuse with predictable profiles creating virtual
interfaces defined by the diffusivity of the factor. Second, the
cellular scale constraints of the microchannel (the shortest
dimension typically only a few cell diameters) allow secreted
factors to accumulate—also in a predictable way. Thus, the
diffusion of molecules is physically limited to in vivo relevant
dimensions leading to accumulation over time periods relevant
to cell proliferation and growth. The ability to vary microchannel dimensions over a wide range enables control of the
accumulation properties. This provides a basis for studying the
effects of soluble factor diffusion and accumulation on cell
proliferation; and an experimental paradigm to investigate a
class of cell biology questions that are difficult to explore in
traditional culture systems.
Previous studies have shown that cell proliferation rates in
static microchannels can differ from rates observed in macroscale culture systems. Murine embryos cultured in microfluidic
channels proliferate more rapidly (with kinetics closer to
in vivo rates) and efficiently than in traditional embryo culture
systems,15 while insect cells (Sf9) proliferate more slowly in
microchannels than in tissue culture flasks.16 In fact, it was
this seeming contradiction that prompted us to investigate
the effect of micro channel geometry on cell behavior in
further detail.
In a previous study we presented data showing that Sf9 cells
proliferate more slowly in microchannels and that channel
width does not affect proliferation.16 Here we perform a
comprehensive study of the relationship between microchannel
geometry and cell proliferation. We found that microchannel
cell cultures differ significantly from conventional macro-scale
cell culture in terms of proliferation kinetics, and that there is a
strong correlation between channel geometry and cell proliferation kinetics. We present a possible explanation for this
dimension dependent cell behavior based on the limited
diffusion and accumulation of soluble factors in the spatially
1090 | Lab Chip, 2005, 5, 1089–1095
constrained microenvironments. Finally, the significance and
potential applications of this approach to studying cell
behavior are discussed.
Methods
System design and fabrication
The microchannels were fabricated using well established
photolithographic and micromolding methods.12 Fig. 1A
shows a representative device. Briefly, SU-8 with a viscosity
of 100 (Microchem Corp., Newton, MA) was spin coated onto
a 3 inch silicon wafer, and baked at 95 uC for 1.5 h. The wafer
was then covered with a transparent mask with designed
patterns (20 mm 6 1 mm rectangles) and exposed to UV at
200 mJ cm22, baked at 150 uC for 3 h and developed
(Microchem Corp., Newton, MA). The master was dried and
hard baked at incremental temperatures up to 95 uC for 0.5 h.
PDMS prepolymer mixed with curing agent (Sylgard 184
silicone elastomer kit, Dow Corning, Midland, MI) at 10 : 1
was poured on the EPON masters and cured at 80 uC for 2 h.
The PDMS layer was then peeled off, punctured for entries,
exposed to UV in a bio-hood for 20 min to sterilize and then
bonded to a tissue culture dish to construct microchannels.
PDMS channels higher than 1 mm were molded on the
micromachined Lucite masters with the height of patterns
1.0 mm, 1.5 mm and 2.0 mm. Microchannels were equilibrated
in a humidified incubator (27 uC) for several hours before use.
After cell seeding, the space around the PDMS was filled with
sterilized water to maintain local humidity.
Sf9 cell culture and growth analysis
Fall armyworm ovarian cells (Spodoptera frugiperda insect
cells, Sf9) were chosen for initial studies because Sf9 cells are
Fig. 1 The PDMS microfluidic channels and a drawing of the
microchannel cell culture. (A) Patterned PDMS slabs were bonded to a
sterile polystyrene Petri dish to construct a series of microchannels
(each channel is 20 mm 6 1 mm 6 0.25 mm); (B) diagram of the
cross-sectional view of the microchannel seeded with Sf9 cells.
This journal is ß The Royal Society of Chemistry 2005
attachment independent and free of cell–cell contact inhibition, and thus avoid the confounding effects of surface
interactions (extracellular matrix) and contact inhibition, and
instead focus on soluble factor related regulation of cell
proliferation. The distinguishable spherical morphology of
Sf9 cells allows real time observation without staining during
experiment periods. The growth condition required for Sf9
cells in the macrocultures is less strict than most mammalian
cells: Sf9 cells grow in medium with pH 6.2–6.9 and osmolarity
345–380 mOsm kg21 at temperature 26.5–28 uC. There is no
need for extra CO2 supply for Sf9 culture.17 Sf9 cells have
been valuable commercial tools for recombinant protein
production for over 50 years owing to these properties and
their revolution relevance to mammalian cells. Sf9 cells
(Panvera, Madison, WI) were grown in TC-100 (Sigma,
T3160) supplemented with 10% fetal bovine serum and 2 mM
glutamine (Gibco, 25030081). The cells were maintained in T25
tissue culture flasks (BD, NJ) in a 27 uC humidified incubator
and were passed twice a week. Cells in log phase were seeded
into microchannels and incubated in the 27 uC incubator for
5 days. Cells were manually counted every 24 h and the growth
curves were plotted. Flask cultures served as controls. Trypan
Blue test showed that 95% cells were viable in a typical
microchannel (20 mm 6 1 mm 6 0.25 mm) after 5 day
culture. In addition, the cells in all experiments exhibited
normal cell morphology and non-directional distribution as
shown in Fig. 2.
pH measurement
The pH of the medium in the microchannels was monitored
with a microelectrode pH probe (Microelectrodes. Inc.)
during the 5 day culture period. The measurements
were taken every time the cultures were taken out of the
incubator for observations. The pH probe was calibrated,
enzyme treated and sterilized before each use. The reading
was obtained 5–7 s after the probe was inserted into the
channel entrance. The pH of the medium in the flask
cultures and in channels without cells was also measured as
a control.
Fig. 2 Sf9 cells were seeded in PDMS microfluidic channels (20 mm 6
1 mm 6 0.25 mm) and cultured for 3 days. Cells show normal
morphology and 95% viability after a Trypan Blue exclusion test.
This journal is ß The Royal Society of Chemistry 2005
Results
Sf9 cell culture and microchannel geometry
Three geometric parameters of microchannels were investigated: width, length and height. The width is defined as the
shorter dimension of the base of microchannels while length is
the longer dimension of the base, and height is the vertical
dimension. The influences of each of the three parameters on
cell proliferation were studied independently. Additionally, the
effects of seeding density at fixed dimensions were tested. At
least three channels were used in each setting and flask cultures
were used as a control. All the cell proliferation data are
presented as normalized with respect to the cell number on day
1. Data are presented as average ¡ standard deviation. Oneway ANOVA (P , 0.01) and Tukey’s test (for all pairwise
comparisons) were used to determine significant differences
between channel groups for each day (see figure captions).
We performed experiments to test the effect of channel
width and length on cell proliferation. Since these experiments,
in part, duplicate previously reported data, we only summarize
briefly here. To study the effects of width, microchannels with
widths of 0.25 mm, 0.75 mm, 1.5 mm, and 2.0 mm, but having
a fixed length (20 mm) and height (0.25 mm) were tested.
Cells were seeded into these channels and T25 flask at 3.5 6
105 cells ml21. No statistically significant difference in cell
proliferation between microchannels having different widths
was observed as reported previously.16
Keeping the width and height fixed (1 mm and 0.25 mm,
respectively), microchannels of lengths 10 mm, 20 mm and
30 mm were designed to explore the effects of the length
dimension. Cells were seeded into channels and a T25 flask at
2.8 6 105 cells ml21. The results are shown in Fig. 3. Like
in the case of the microchannels with varying widths, no
Fig. 3 Comparison of the proliferation of Sf9 cells in microchannels
with various lengths: a. 10 mm, b. 20 mm, c. 30 mm. All the microchannels had the same height (0.25 mm) and width (1 mm). Microchannels were seeded SF9 cells at the surface density 70 cells mm22 or
volume density 280 cells mm23. Cells were cultured for 4 days with the
flask culture as a control (280 cells mm23). (d) The proliferation of
cells in all the microchannels was significantly slower than in the flask
(P , 0.01) and there were no significant differences between different
length channels.
Lab Chip, 2005, 5, 1089–1095 | 1091
significant difference in cell proliferation was observed
between the microchannels with different length. Differences
between the microchannel and flask cultures, however, were
consistent with width experiments above. The population in
the microchannels reached a plateau after 48 h, the population
in the flask continued to increase, expanding 5.5-fold over
the course of the experiment, compared to 2-fold for the
microchannel cultures. Note that there was no significant
change in proliferation beyond day 4.
To explore the influence of channel height on cell proliferation, a series of channels with heights 0.25 mm, 0.5 mm, 1.0 mm,
and 2.0 mm (length 20 mm, width 1 mm) were constructed and
seeded with cells at appropriate volume densities to achieve a
constant surface density of 135 ¡ 10% cells mm22.
As a control, cells were seeded into a T25 flask at 4.4 6
105 cells ml21. Fig. 4 shows that the cell proliferation rates
were significantly lower in the microchannels than in the flask.
However, in the microchannels, the proliferation rates
exhibited a clear dependence on channel height. In the first
24 h, the growth rate is similar between the channels, and does
not differ significantly from that in the flask. Between 24 and
48 h there is a clear trend that taller channels have higher
proliferation rates (60% for 2.0 mm vs. 25% for 0.25 mm)
although still lower than that of the flask. The proliferation
rate then goes down. For the smallest height the cell number
reaches a plateau, but at larger heights the cells continue to
proliferate throughout the experiment (6 days) at a rate which
increases with channel height, but goes down over time. As a
result of the difference in proliferation rate, the net expansion
Fig. 4 Comparison of the proliferation of Sf9 cells cultured in various
height microchannels: a. 0.25 mm, b. 0.5 mm, c. 1.0 mm, d. 2.0 mm. All
the microchannels had the same width (1 mm) and length (20 mm).
Each channel was seeded with cells at the same surface density
120 cells mm22. The proliferation rates of cells gradually increased as
the height of the channels increased from 0.5 mm to 2.0 mm. Similar to
width and length experiments, the net proliferation of cells in
microchannels was much lower than in the flask (not shown). The
differences in proliferation rates between the different channels began
to emerge at day 2. Statistically significant differences were observed
between both 0.25 mm and 0.5 mm microchannels and both 1.0 mm
and 2.0 mm microchannels at day 2 and day 3. The differences between
0.25 and 0.5 mm microchannels and between 1.0 mm and 2.0 mm
micromicrochannels were marginal (P 5 0.045).
1092 | Lab Chip, 2005, 5, 1089–1095
of the cell populations increases with height, such that over the
course of the experiment, it is 2.4, 3.3, 5 and 5.2-fold,
respectively for 0.25 mm, 0.5 mm, 1.0 mm, and 2.0 mm high
channels. For comparison the population expansion in the
flask culture was 9-fold in 6 days.
The surface density of cells is constant in each of the
experiments described above for the microchannel width,
length and height. To explore the importance of surface
seeding density, microchannels of equal size (length 20 mm,
width 1 mm and height 0.25 mm) were seeded with different
concentrations of cells (9.0 6 105 cells ml21). Flasks seeded
with 2.0 6 105 cells ml21 served as controls. Fig. 5 shows the
cell proliferation rate for each of these seeding densities. Cell
proliferation rates were significantly lower in the microchannels than in the flask. However, the proliferation rates
exhibited a clear dependence on surface seeding density. In the
first 24 h, the growth rate is similar between channels, but
between 24 and 48 h there is a clear trend that microchannels
with lower seeding density have higher proliferation rates
(75% for 25 cell mm22 vs. 18% for 400 cells mm22). The
proliferation rate then goes down, and the cells in the
microchannels with a higher seeding density reach a plateau
earlier (day 4) than those with lower seeding density (day 5).
These different proliferation rates result in a 3.5-fold net
expansion of the cell populations in channels seeded at
25 cell mm22, 1.8-fold in channels seeded with 400 cells mm22
and 9-fold for the flask culture in 6 days.
To determine if the cell behavior was permanently altered by
culture in microchannels, cells were moved from the microchannel culture to a flask culture after day 4. The cells
immediately resumed typical flask proliferation profiles.
Fig. 5 Sf9 cells from the same exponential flask culture were diluted
accordingly and seeded in microchannels of the same size (length 6
width 6 height: 20 mm 6 1 mm 6 0.25 mm). The surface seeding
density in each channel was: a. 400 cells mm22, b. 140 cells mm22,
c. 50 cells mm22, d. 25 cells mm22 (and a flask control at
280 cells mm22 , not shown). Similar to all previous experiments, the
proliferation rates of cells in all microchannels were slower than of
flask cultures (not shown). The proliferation rates gradually increased
as the seeding density decreased from 400 cells mm22 to 25 cells mm22.
The differences between both 400 cells mm22 and 140 cells mm22
and both 50 cells mm22 and 25 cells mm22 were significant (P , 0.01)
after day 2.
This journal is ß The Royal Society of Chemistry 2005
pH measurements
We recorded the pH of the medium in microchannel cultures
over each day throughout the culture period (5 days, data not
shown). The pH remained relatively constant at 6.2–6.9 in Sf9
microchannel culture. There were no significant differences in
pH between channels with cells and without cells (medium
only) during culture periods.
Discussion
The studies presented here provide insights into the relationship between cell proliferation and the spatially constrained
environment of microchannels. There are three main conclusions that may be drawn from the data. First, cells cultured in
microchannels in the absence of flow proliferate significantly
more slowly than cells cultured in traditional macro-scale
culture systems, and enter a quiescent state. Second, the
inhibition of cell proliferation in microchannels is removed
when cells are returned to a macro-scale culture environment.
Third, proliferation rates in microchannels are not influenced
by channel width or length but are increased with increasing
microchannel height or decreasing seeding density.
Importantly, these phenomena are not seen in traditional
macro culture systems or in other microfluidic culture
conditions (e.g. culture at very high surface density,18 or under
flow or frequent medium changes,19 or in microdevices without
constraining physical walls and immersed into bulk medium).
Certain culture conditions are known to inhibit growth in
macro-scale culture20 and it is important to determine their
possible effects in the microchannel experiments. One of these
is pH change. We measured the pH each time the cell number
was evaluated. It was found to be consistently within the
normal range for each cell type we studied. Another condition
is non-physiologic osmolarity. Because the microchannels have
very small volumes, the local humidity was carefully maintained by creating a stable high humidity environment. The
devices were placed within covered dishes with fluid surrounding the devices. This creates a humid environment that
minimizes disruptions (e.g. when the incubator door is
opened). Under these conditions no visible volume change of
medium was observed during culture. Additionally, we
assessed the morphology of cells in channel culture and found
it to be indistinguishable from that of the control cells cultured
in flasks, suggesting that the culture media did indeed have
appropriate osmolarity. A third condition associated with
impaired proliferation is nutrient depletion. Let us consider the
amount of media per cell in macro-scale culture on one hand
and microfluidic culture on the other. Looking at the data
presented in Fig. 5, cells were seeded in flasks at a density of
4.4 6 105 cells ml21 and they expanded over 9-fold. The
expansion of microchannel cell cultures with same volume
concentration of cells at time of seeding is only a fraction of
this (less than 1 6 106 cells ml21 after five days), despite them
having an equal amount of nutrients available per cell. Now let
us consider the amount of nutrients that the cells in the macroscale culture mentioned above require during the first four
days. During that time, their growth is exponential, indicating
that nutrients are plentiful. The volume concentration of cells
averaged over this time is greater than that for the highest
This journal is ß The Royal Society of Chemistry 2005
volume concentration of cells seeded in microchannels in our
experiments, which was 1.6 6 106 cells ml21. So, if we assume
a constant nutrient consumption per cell per day, the total
nutrient consumption will have been greater in the macro-scale
culture. Thus, nutrient depletion is highly unlikely to have
taken place under our experimental conditions. A fourth
condition is hypoxia. PDMS, the material used for construction of the microchannels is highly permeable to oxygen,
ensuring sufficient gas supply, and the high surface area/
volume ratio in microchannels promotes efficient gas delivery
to cells.21 We have thus ruled out the possibility that known
macro-scale proliferation-inhibiting conditions are the underlying cause of the observed growth kinetics. Furthermore,
there was no evidence of necrosis or apoptosis in any of the
experiments, suggesting that in the microchannels the cells are
entering a quiescent state.
In our studies we compared cell growth in microchannels to
that in flasks at equal volume concentrations (Fig. 3, all
curves). The results were conclusive, that in microchannels, the
proliferation is slower and plateau sooner than in the flask. We
also compared situations where the microchannel cultures
spanned surface densities both lower and higher than that in
the flask culture (Fig. 5, b–d and a, respectively). We observed
a much slower proliferation, and quiescence was reached
earlier in the microchannels than in the flask. If the observed
phenomena were an artifact of cell density, proliferation in
microchannels would have reached the same rate in microfluidic channels as in flasks when cells were cultured at the
same density in each system. This is not the case. Neither
surface nor volume density can explain the disparities between
the microchannel and flask culture data.
We propose that the proliferation kinetics observed in
microchannels may be a result of the physical properties of
the microchannel environment, specifically, the dominance
of diffusion over convection. While in the flask, secreted
molecules are swept away by random convection; these same
molecules are transported from their source by diffusion alone
in the microchannels. This enables the formation of gradients,
and the retention of these molecules in proximity of the cells.
The relevant time scale of this retention depends on the size
of each molecule involved. From preliminary modeling
(Shkel and Beebe, to be published) we have learned that
over a period of several hours, molecules with molecular
weights similar to that of many chemical messengers (on the
order of 50 kDa) accumulate around the secreting cell at
concentrations significantly higher than what would be the
concentration in a mixed solution, other parameters being
equal. If we hypothesize the existence of a growth inhibitory
influence by one or more molecules secreted by the cells, the
accumulation of these molecules might explain the observed
proliferation kinetics.
When the surface density of cells is increased, the average
distance between them is reduced, and the overlapping local
gradients present these secreted molecules at concentrations
that increase with the surface density of cells. This would
exert a stronger growth inhibiting influence, in agreement
with the proliferation kinetics we observed for different
seeding density, as presented in Fig. 6. We have observed
similar dimension dependence with mouse mammary gland
Lab Chip, 2005, 5, 1089–1095 | 1093
Fig. 6 Schematic of the local cell environment in the microchannel environment and the macro-scale culture environment (flask). (left) Cells
cultured in microchannels secrete various biomolecules. These molecules with different diffusivities will distribute around cells and into the
surrounding medium by diffusion. The local cell microenvironment can be adequately maintained by diffusion. (right) Cells cultured in flasks also
secrete biomolecules, but these factors are dissipated in the bulk medium by both convection and diffusion; the cell local environment is not stable.
epithelial cells (NMuMG) suggesting it is not a cell type
specific phenomenon.
The elucidation of the mechanism behind the altered
proliferation in microchannels is complicated by the multitude
of pathways that may be involved, and the way in which the
latency of the response may vary depending on whether it is
direct or involves changes in gene expression. The explanation
of the proliferation dependence on channel height is not
immediately apparent and may require further knowledge on
the mechanism responsible for the modified growth kinetics.
It is true that the growth inhibiting cell-derived factors we
hypothesize being responsible for the altered growth kinetics in
microchannels, will eventually accumulate in flasks and other
macro-scale culture systems as well. The main difference lies in
the time at which this takes place. In the microchannels,
accumulation may occur very early, within hours of cell
seeding. Conversely, in any macro-scale culture system,
secreted molecules are mixed by random convection and their
concentration rises linearly with time. Therefore, the threshold
level required for them to exert their effect may not be reached
until after the cells have proliferated substantially. The cells
may even reach confluency, and cease to grow due to contact
inhibition before this happens. To put the results in context
with previous experiments, the phenomena observed here
would not be expected in cell culture under constant flow.17
The same applies to microchannels open to bulk medium, as
random convection will take place, much like in other macroscale culture environments.
The microchannel cell culture platform facilitates cell
communication through soluble factors, which could be the
reason for the observed proliferation kinetics and their
difference from those in the flask cell culture. If this is true,
the microfluidic cell culture platform will be of practical and
theoretical importance for the investigation of cell biology.
In vivo, cells interact intimately with the microenvironment
around them, and their behavior is very sensitive to any change
1094 | Lab Chip, 2005, 5, 1089–1095
in the microenvironment. While other technologies have
facilitated the study of various aspects of these interactions
as discussed above, soluble factors have largely been neglected.
Nonetheless, soluble factors play important roles in cell
proliferation, differentiation and pathological behavior. Our
work presents an approach to address a variety of questions
about the effects of cell-derived soluble factors on cell behavior
(e.g. autocrine/paracrine signaling).22 For example, Sf9 cells
have been found to autocrine IGF-I in macro-scale culture
systems, which could stimulate the proliferation of Sf9 cells.23
Shaping cell microenvironment through microchannel geometry allows predictable control of the accumulation of these
factors, emphasizing their role in regulating cell behavior.
Another interesting observation is the temporary nature of
inhibition of cell proliferation in microchannels. When
placed back into flask culture the cells resume normal flask
proliferation rates. This suggests that the inhibition that occurs
within the microchannel might be via a cell cycle check point
mechanism. Although we have not explored this phenomenon
further, there is potential application in stem cell biology
where effective control of self-renewal and differentiation is
important.24
The situation in vivo is much more complex where, for
example, the availability of soluble factors can be controlled
by sequestration and release from the ECM.25 This in vitro
model focuses only on the effects of diffusion in the
microenvironment, and provides a framework for studying
such phenomena in vitro.
Conclusion
Microfluidic systems have great potential to enhance our
ability to study cell behavior. However, there are important
and fundamental differences between micro-scale culture
systems and macro-scale culture systems. In order to
effectively utilize emerging micro-scale culture systems, it is
This journal is ß The Royal Society of Chemistry 2005
important to understand how the physical attributes of microscale culture systems influence cell behavior. The work
presented here begins to explore the relationship between the
physical characteristics of the culture system, the local cell
microenvironment and the mechanisms controlling cell proliferation. We observed clear differences between proliferation
rates in macro-scale and microfluidic channel cell cultures.
Furthermore, we found that the proliferation of cells in the
microchannels depended on the channel dimensions as well as
the surface seeding density. It is clear that a great deal of work
remains to fully understand the interaction between microchannel culture systems and cell behavior. However, we believe
that this work provides initial insights that will prove useful in
guiding the design of future microchannel culture systems.
Acknowledgements
The authors would like to acknowledge funding from NIH and
ARMY-BCRP. The authors would like to thank Sarah
Michaels for her assistance in performing culture experiments
and Caroline Alexander for many helpful discussions.
References
1 R. G. Harrison, Observations on the living developing nerve fiber,
Proc. Soc. Exp. Biol. Med., 1907, 4, 140–143.
2 A. Carrel, On the permanent life of tissues outside of the organism,
J. Exp. Med., 1912, 15, 516–528.
3 F. Watt and B. Hogan, Out of Eden: stem cells and their niches,
Science, 2000, 287, 5457, 1427–1430.
4 M. J. Bissel and M. A. Labarge, Context, tissue plasticity and
cancer: are tumor stem cells also regulated by the microenvironment? Cancer Cell, 2005, 7, 1, 17–23.
5 G. S. Themistocleous, H. Katopodis and M. Koutsilieris, Threedimensional type I collagen cell culture systems for the study of
bone pathophysiology, In Vivo, 2004, 18, 6, 687–96.
6 J. A. Hall, N. J. Maitland, M. Stower and S. H. Lang, Primary
prostate stromal cells modulate the morphology and migration of
primary prostate epithelial cells in type 1 collagen gels, Cancer
Res., 2002, 62, 1, 58–62.
7 T. H. Park and M. L. Shuler, Integration of cell culture and
microfabrication technology, Biotechnol. Prog., 2003, 19, 243–253.
8 A. Folch and M. Toner, Microengineering of cellular interactions,
Annu. Rev. Biomed. Eng., 2000, 2, 227–256.
This journal is ß The Royal Society of Chemistry 2005
9 B. L. Gray, D. K. Lieu and A. Barakat, Microchannel flatform
for the study of endothelial cell shape and function, Biomed.
Microdev., 2002, 4, 9–16.
10 N. L. Jeon, H. Baskaran, G. M. Whitesides and M. Toner,
Neutrophil chemotaxis in linear and complex gradients
of interleukin-8 formed in a microfabricated device, Nat.
Biotechnol., 2002, 20, 8, 826–830.
11 A. Sawano and S. Takayama, Lateral propagation of EGF
signaling after local stimulation is dependent on receptor density,
Dev. Cell, 2002, 3, 245–257.
12 K. Viravaidya, A. Sin and M. Shuler, Development of a microscale
cell culture analog to probe naphthalene toxicity, Biotechnol. Prog.,
2004, 20, 1, 316–323.
13 D. Beebe, G. Mensing and G. Walker, Physics and applications
of microfluidics in biology, Annu. Rev. Biomed. Eng., 2002, 4,
261–286.
14 G. Walker, H. Zeringue and D. Beebe, Microenvironment design
considerations for cellular scale studies, Lab Chip, 2004, 4, 2,
91–97.
15 S. Raty, J. Davis, D. Beebe and M. Wheeler, Culture in
microchannels enhances in vitro embryonic development of
preimplantation mouse embryos, Theriogenology, 2001, 55, 241.
16 G. Walker, M. Ozers and D. Beebe, Insect Cell culture in
microfluidic channels, Biomed. Microdev., 2002, 3, 161–166.
17 T. D. Grace, Establishment of four strains of cells from insect
tissues grown in vitro, Nature, 1962, 195, 758–759.
18 A. Tourovskaia, X. Figueroa-Masot and A. Folch, Differentiationon-a-chip: A microfluidic platform for long-term cell culture
studies, Lab Chip, 2005, 5, 14–19.
19 S. H. Cartmell, B. D. Porter and R. E. Guldberg, Effects of
medium perfudion rate on cell-seeded three dimensional bone
constructs in vitro, Tissue Eng., 2003, 9, 6, 1197–1203.
20 C. Bedard, R. Tom and A. Kamen, Growth, nutrient consumption,
and end-product accumulation in Sf-9 and BTI-EAA insect cell
cultures: insights into growth limitation and metabolism,
Biotechnol. Prog., 1993, 9, 615–624.
21 C. Stern, Diffusion of gases in silicone polymer: molecular
dynamics simulations, Macromolecules, 1998, 31, 5529–5535.
22 P. Carmeliet, Mechanisms of angiogenesis and arteriogenesis,
Nat. Med., 2000, 6, 4, 389–395.
23 M. Doverskog, M. Tally and L. Häggström, Constitutive secretion
of an endogenous insulin-like peptide binding protein with high
affinity for insulin in Spodoptera frugiperda (Sf9) cell cultures,
Biochem. Biophys. Res. Commun., 1999, 265, 674–679.
24 V. V. Abhyankar and D. J. Beebe, Human embryonic stem cells &
microfluidics, in Lab-on-a-chips for Cellomics, ed. H. Andersson
and A. van den Berg, Kluwer, 2004, pp. 257–272.
25 C. R. Ozawa, A. Banfi and H. M. Blau, Microenvironmental
VEGF concentration, not total dose, determines a threshold
between normal and aberrant angiogenesis, J. Clin. Invest., 2004,
113, 516–527.
Lab Chip, 2005, 5, 1089–1095 | 1095