ISSN 1757-9694
1757-9694(2011)3:6;1-R
www.rsc.org/ibiology Volume 3 | Number 6 | June 2011 | Pages 595–708
PAPERKumacheva et al.High-throughput combinatorial cell co-culture using microfl uidics
V
olu
me
3
| N
um
be
r 6
| 2
01
1
Inte
grative B
iolo
gy
Pa
ge
s 5
95
–7
08
Quantitative biosciences from nano to macro
Indexed in
MED
LINE!RSC Advances
An international journal to further the chemical sciences
RSC Advances is a new peer-reviewed journal covering all the chemical sciences, including interdisciplinary fields. Published articles will report high quality, well-conducted research that adds to the development of the field.
Go online today
Submissions now open – first issue mid-2011
An expert editorial team led by Professor Mike Ward, University of Sheffield, UK
Free access to all content throughout 2011 and 2012
Free colour, no page charges
Published online only
RSC Advances
ISSN 2046-2069
www.rsc.org/advances Volume 1 | Number 1 | 2011
www.rsc.org/advances Registered Charity Number 207890
Dow
nloa
ded
by D
alho
usie
Uni
vers
ity o
n 08
/05/
2013
21:
11:0
2.
Publ
ishe
d on
28
Apr
il 20
11 o
n ht
tp://
pubs
.rsc
.org
| do
i:10.
1039
/C1I
B00
002K
View Article Online / Journal Homepage / Table of Contents for this issue
This journal is c The Royal Society of Chemistry 2011 Integr. Biol., 2011, 3, 653–662 653
Cite this: Integr. Biol., 2011, 3, 653–662
High-throughput combinatorial cell co-culture using microfluidicsw
Ethan Tumarkin,aLsan Tzadu,
aElizabeth Csaszar,
bdMinseok Seo,
aHong Zhang,
a
Anna Lee,aRaheem Peerani,
abcKelly Purpura,
bcdPeter W. Zandstra*
bcdand
Eugenia Kumacheva*ab
Received 9th January 2011, Accepted 23rd March 2011
DOI: 10.1039/c1ib00002k
Co-culture strategies are foundational in cell biology. These systems, which serve as mimics of
in vivo tissue niches, are typically poorly defined in terms of cell ratios, local cues and supportive
cell–cell interactions. In the stem cell niche, the ability to screen cell–cell interactions and identify
local supportive microenvironments has a broad range of applications in transplantation, tissue
engineering and wound healing. We present a microfluidic platform for the high-throughput
generation of hydrogel microbeads for cell co-culture. Encapsulation of different cell populations
in microgels was achieved by introducing in a microfluidic device two streams of distinct cell
suspensions, emulsifying the mixed suspension, and gelling the precursor droplets. The cellular
composition in the microgels was controlled by varying the volumetric flow rates of the
corresponding streams. We demonstrate one of the applications of the microfluidic method by
co-encapsulating factor-dependent and responsive blood progenitor cell lines (MBA2 and M07e
cells, respectively) at varying ratios, and show that in-bead paracrine secretion can modulate the
viability of the factor dependent cells. Furthermore, we show the application of the method as a
tool to screen the impact of specific growth factors on a primary human heterogeneous cell
population. Co-encapsulation of IL-3 secreting MBA2 cells with umbilical cord blood cells revealed
differential sub-population responsiveness to paracrine signals (CD14+ cells were particularly
responsive to locally delivered IL-3). This microfluidic co-culture platform should enable high
throughput screening of cell co-culture conditions, leading to new strategies to manipulate cell fate.
Introduction
Local and systemic cell–cell interaction networks modulate
development and homeostasis and underlie ageing and disease
progression.1 In vivo, stem cell (SC) fate may be regulated by
tissue-specific niches such as the bone marrow, vascular, and
hepatic microenvironments.2 Stem cell fate in these niches is
regulated by chemical and biological microenvironments via
the release of cytokines, by cellular contacts via adhesion
molecules, and by interactions with components of the extra-
cellular matrix.3 One of the key functions of SC niches is to
aDepartment of Chemistry, University of Toronto,80 Saint George Street, Toronto, Ontario, M5S 3H6, Canada.E-mail: [email protected]
b Institute of Biomaterials & Biomedical Engineering, University ofToronto, 164 College Street, Toronto, Ontario, M5S 3G9, Canada
c Terrence Donnelly Centre for Cellular and Biomolecular Research,University of Toronto, 160 College Street, Toronto, Ontario,M5S 3E1, Canada
dDepartment of Chemical Engineering and Applied Chemistry,University of Toronto, 200 College Street, Toronto, Ontario,M5S 3E5, Canada
w Electronic supplementary information (ESI) available: Fig. S1–S4,Table S1–S2. See DOI: 10.1039/c1ib00002k
Insight, innovation, integration
This paper reports a microfluidic approach to the fast-
throughput generation of microenvironments for cell
co-culture. The strategy provides a straightforward opera-
tional platform for the co-encapsulation of a controlled
number of cells of different lineages in well-defined
microgel environments. Cells co-encapsulated in each indi-
vidual microgel are separated from cells encapsulated in
different microgels, which enables interrogation of interactions
between two types of cells encapsulated in the same microgel.
The microfluidic approach enables understanding of how cell
fate is affected by their interactions in microenvironments.
One of the applications of the method is demonstrated by
co-encapsulating factor-dependent and responsive blood
progenitor cell lines. We also utilized the method to screen
the impact of specific growth factors on a primary human
heterogeneous cell population.
Integrative Biology Dynamic Article Links
www.rsc.org/ibiology PAPER
Dow
nloa
ded
by D
alho
usie
Uni
vers
ity o
n 08
/05/
2013
21:
11:0
2.
Publ
ishe
d on
28
Apr
il 20
11 o
n ht
tp://
pubs
.rsc
.org
| do
i:10.
1039
/C1I
B00
002K
View Article Online
654 Integr. Biol., 2011, 3, 653–662 This journal is c The Royal Society of Chemistry 2011
maintain tissue homeostasis through direct and indirect
cell–cell interactions. This is achieved by balancing the
proportions of quiescent and activated cells, and the signals
they produce. Stem cells may be activated by specific cues
within the niche to proliferate, self-renew, or differentiate to
regenerate damaged tissue.4 In addition to gaining funda-
mental understanding of cellular interactions, guiding cell fate
by co-culturing specific cell lineages has potential applications
in wound healing,5 tissue engineering,6 and cancer research.7
Cellular co-culture is also important in SC research both in
establishing and maintaining embryonic SC cultures from
mouse or human sources8 and in guiding SC fate. For
instance, co-culture of OP9 stromal cells (derived from
M-CSF deficient mice) with embryonic SCs leads to
primitive and definitive hematopoetic development of the
SCs in vitro, in a manner similar to that occurring in murine
ontogenesis.9
The development of efficient methods which would provide
rapid generation of massive libraries of co-cultured cells has
fundamental and practical importance. These methods should
enable (i) the controlled generation of interactive cellular
microenvironments that closely mimic the native cellular
microenvironment in architecture and chemical and mecha-
nical properties; (ii) the spatio-temporal control over the
relative number of co-cultured cells in the microenvironment;
and (iii) the ability to carry out high-throughput screening
integrated with function and relevant cell outputs. Since
interactions between cells can occur through direct contact
or through soluble factors via paracrine signalling, it is
important to maximize the interaction between neighbouring
cell types by creating microenvironments with an inter-cellular
distance on the scale of cells.
Currently, micropatterning of planar substrates does not
provide the ability to co-culture SCs in 3D environments.10
Microfabrication of scaffolds11 or the generation of sphere-
shaped cell colonies12 produce 3D microenvironments for
co-cultured cells. For example, mixing of cell-laden and cell-
free microgels at varying ratios was utilized to produce
microscale constructs with a controllable ratio of the two
microgels.13 Yet, these approaches have a limited ability to
vary the relative ratios of the co-cultured species in a
continuous high-throughput manner, which counteracts the
capability to create libraries for rapid screening of paracrine
signalling mediated cell–cell interactions.
Emulsification of single cells or cell aggregates14,15 and more
recently, microfluidic (MF) emulsification of suspensions
of cells has offered a new operational platform in cell
biology.16–18 Rapid MF generation of thousands of cell-laden
droplets with precisely controlled dimensions and compositions
provided the ability to create a large number of well-defined
3Dmicroenvironments for studies of cell growth and viability,19,20
gene expression,21 and enzymatic activity.22–24 Encapsulation
of cells in hydrogel microbeads (microgels) derived from
precursor droplets, yielded biocompatible microenvironments,
which allowed exchange of nutrients and oxygen between the
encapsulated cells and the surrounding medium.25–28 Micro-
fluidic encapsulation of cells in droplets and microgels paved
the way for high-throughput generation of combinatorial
libraries of assays,29–31 however, microenvironments for
co-cultured cells with a controlled ratio of the number of cells
of different lineages has not been demonstrated.
Herein we describe a MF strategy for rapid (40 s�1)
generation of 3D microenvironments—agarose microgels—
for cell co-culture. Agarose is a neutral polysaccharide that
undergoes gelation upon cooling.32 Agarose is generally bio-
inert, it exhibits low adsorptivity to proteins and cells, it can be
readily functionalized with cell adhesion proteins, and its
mechanical properties can be tuned by varying the agarose
concentration in the gel.33–36 We demonstrate the practical
advantage of the MF method by (i) achieving a 100%
encapsulation efficiency; (ii) by growing pluripotent cell
derived aggregates in the microgels, (iii) by controlling the
extent of paracrine signalling between M07e human mega-
karyoblastic leukemia cells and MBA2 support cells, which
were encapsulated in varying ratios, and ultimately affecting
the survival of M07e cells37 and (iv) co-encapsulating MBA2
cells in combination with heterogeneous umbilical cord blood
(UCB) cells to discover and determine the rescue effect of
localized IL-3 on the various hematopoietic sub-populations.
The impact of the MBA2 :M07e and the MBA2 :UCB case
studies demonstrates a novel approach to the controlled
generation and characterization of paracrine signalling
amongst cell-populations in an in vitro setting. In comparison
with existing methods used for cell encapsulation, this
approach allows for the continuous high-throughput
generation of microenvironments with a tuneable ratio of
encapsulated species.
Materials and methods
Generation of cell-laden agarose microgels
Two distinct cell suspensions were prepared in 2 wt%
solutions of ultra-low gelling temperature agarose (SeaPrep,
Lonza, Switzerland) in a PBS (Gibco-BRL, Rockville, MD)
buffer pre-heated to 37 1C. The concentration of cells varied in
the range from 2 � 106 to 8 � 106 cells mL�1. The suspensions
with equal concentrations of cells were supplied to two
channels of the MF T-junction droplet generator (Fig. 1a,
Fig. S1w) using two independently controlled syringe pumps
(Harvard Apparatus 33 Dual Syringe Pump, USA). The
temperature of the suspensions was maintained at 37 1C in a
temperature controlled incubator, in order to prevent agarose
gelation. The suspensions were mixed in a serpentine channel
with a length of 250 mm. Mineral oil containing 3 wt% of the
non-ionic surfactant Span 80 was supplied to the horizontal
channel of the MF device, as shown in Fig. 1a.
Droplets compartmentalizing the cells were generated at a
T-junction.38 Control over the encapsulation ratio of different
cells was achieved by tuning the ratio of flow rates of the
individual cell suspensions, while maintaining constant total
flow rate of the suspensions.
The cell-laden droplets travelled towards the outlet of the
MF device and entered the outlet tubing (HPFA tubing,
Upchurch Scientific, USA). The tubing was jacketed with a
hose connected to a water circulator cooled to 2 1C with a 1 : 4
vol glycerol : water mixture. The droplets passed through the
tubing within 5 min and were collected in a 15 mL centrifuge
Dow
nloa
ded
by D
alho
usie
Uni
vers
ity o
n 08
/05/
2013
21:
11:0
2.
Publ
ishe
d on
28
Apr
il 20
11 o
n ht
tp://
pubs
.rsc
.org
| do
i:10.
1039
/C1I
B00
002K
View Article Online
This journal is c The Royal Society of Chemistry 2011 Integr. Biol., 2011, 3, 653–662 655
tube containing Hank’s Buffered Saline Solution (HBSS,
Gibco-BRL). The centrifuge tube was placed in an ice bath
to complete the gelation of the microgels. After the microgels
were maintained for 45 min in the bath, the suspension was
centrifuged, the microgels were separated, washed twice with
HBSS buffer, transferred into HBSS buffer containing 2% v/v
FBS and centrifuged at 1000 rpm for 5 min at 4 1C. The
centrifuged microgels were washed with HBSS buffer before
being transferred to cell culture media.
Cell culture
Cells were cultured under sterile conditions and maintained in
a 5% CO2 humidified incubator at 37 1C. Mouse embryonic
stem (mES) cells, R1 and YC5-YFP-NEO,39 were maintained
in cell culture media composed of Dulbecco’s Modified Eagle
Media (DMEM, Gibco-BRL) supplemented with 15% (v/v)
Fetal bovine serum (FBS, Gibco-BRL), 2 mM L-glutamine,
0.1 mM b-mercaptoethanol (BME, Sigma, St. Louise, MO),
0.1 mM non-essential amino acids (NEAA, Gibco-BRL),
1 mM sodium pyruvate (Gibco-BRL), 50 mg mL�1 penicillin
(Gibco-BRL), 50 mg mL�1 streptomycin (Gibco-BRL), and
500 pM Leukemia inhibitory factor (LIF, Chemicon,
Temecula, CA). Confluent dishes of mES cells were fed and
passaged on alternate days. M07e and MBA2 human mega-
karyoblastic leukemia cells were maintained in cell culture
media composed of Iscoves Modified Dulbecco’s Media
(IMDM, Gibco-BRL), 10% (v/v) FBS, 50 mg mL�1 penicillin,
50 mg mL�1 streptomycin, 2 mM L-glutamine, and 0.1 mM
BME. Cells were passaged every three days (1 : 6 split ratio)
and M07e cells were supplemented with 10 ng mL�1 of IL-3.
Embryoid body formation was induced by transfer of the
microgels laden with mES cells to the culture media comprising
DMEM supplemented with 15% (v/v) FBS, 2 mM L-glutamine,
0.1 mM BME, 0.1 mM NEAA, 50 mg mL�1 penicillin,
50 mg mL�1 streptomycin. The samples were maintained under
sterile conditions in a 5% CO2 humidified incubator at 37 1C
for 4.5 days.
Umbilical cord blood cells were obtained from consenting
donors according to procedures accepted by the ethics board
of Mt. Sinai hospital (Toronto, ON, Canada). Mononuclear
cells were processed as reported elsewhere.40 Progenitor cells
were isolated from the mononuclear cell fraction using the
EasySep human hematopoietic progenitor cell enrichment kit
(Stem Cell Technologies, Vancouver, BC, Canada), following
the manufacturer’s protocol. Lineage-depleted cells (lin-) were
cultured in serum-free conditions, as previously described40 for
six days to achieve required cell numbers.
Fabrication of microfluidic devices
Photolithographic masters were prepared from SU-8 50
photoresist (MicroChem, USA) in base-relief on silicon wafers.
The MF devices were fabricated in poly(dimethyl siloxane)
(Sylgard 184, Dow Corning, USA), using a standard soft
lithography procedure.41 After fabrication, the devices were
post-hydrophobized by maintaining the sealed MF device for
12 h in an oven at 140 1C and subsequently, by exposing it to
the vapor of 1,1,1,3,3,3-hexamethyldisilazane. To achieve
silanization, N2 gas was bubbled into a sealed 25 mL vial
containing 1,1,1,3,3,3-hexamethyldisilazane (99.9% pure,
Sigma, USA). During silanization, the MF devices were placed
on a hot plate at 80 1C for 1 h. The vial was sealed with a sleeve
style rubber stopper (Wheaton Science Products, USA).
Bubbling of N2 gas resulted in the vaporization of the
1,1,1,3,3,3-hexamethyldisilazane. Polyethylene tubing (Small
Parts, USA) was placed into the rubber stopper to supply the
1,1,1,3,3,3-hexamethyldisilazane gas into the MF device.
Characterization of cell-laden agarose microgels
Distributions of the dimensions of precursor agarose droplets
and the corresponding microgels were determined by
analyzing optical microscopy images of 300 droplets or micro-
gels using Image Pro 5.0 (Media Cybernetics, USA) software.
Fluorescence microscopy images of encapsulated fluores-
cently-labelled cells were captured using a Zeiss Microscope
(Axio Observer D1, USA) coupled with a digital camera
(Axio Cam HRm, Zeiss, USA). Ratios between different
dye-labelled encapsulated cells were determined by analyzing
images obtained by optical fluorescence microscopy. Approxi-
mately 8000 cell-laden microgels were included in the analysis,
in order to determine the optimized conditions for the
generation of cell-laden agarose microgels.
Flow Cytometry Analysis of cell-laden agarose microgels
Encapsulation ratios of R1 mES cells were monitored using a
Biosort flow cytometer (Union Biometrica, USA) with a bore
diameter of 250 mm. The cytometer was equipped with 488 and
561 nm excitation lasers. Prior to encapsulation, the cells were
labelled with CFDA SE (green) or CMTMR (red) dyes.
Microgel samples were introduced into the flow cytometer
following the manufacturer’s instructions (Union Biometrica).
Settings for sorting of cell-laden microgels
The sorting process was performed according to the manu-
facturer’s instructions (Union Biometrica). The cell-laden
microgels were gated to exclude 1.5% of cell-free microgels
and cell debris by using the gating parameters Time of Flight vs.
Extinction. Time of Flight provided the time required for an
object to pass through the instrument detector, that is, the
length of an object. Extinction reflected the internal complexity
Fig. 1 Microfluidic encapsulation of cells in agarose microgels.
(a) Schematic of MF cell co-encapsulation system. A serpentine
downstream channel with a length of 250 mm is not shown in the
figure. (b) The distribution of sizes of cell-laden droplets of 2 wt%
agarose solution (dotted curve) and of the corresponding agarose
microgel (solid curve).
Dow
nloa
ded
by D
alho
usie
Uni
vers
ity o
n 08
/05/
2013
21:
11:0
2.
Publ
ishe
d on
28
Apr
il 20
11 o
n ht
tp://
pubs
.rsc
.org
| do
i:10.
1039
/C1I
B00
002K
View Article Online
656 Integr. Biol., 2011, 3, 653–662 This journal is c The Royal Society of Chemistry 2011
of an object similar to side scatter in conventional flow
cytometry instruments. Microgels were then sorted based on
fluorescence intensities, that is, Red Peak Height vs. Green
Peak Height. We set the parameters for the fluorescence signal
amplification as follows: Full scale: Tof 256, Ext 256, RedPH
65536, GreenPH 65536; Gains (Signal): Ext 3, Green 2, Red 2;
Trigger: Ext Thresholds: Signal 30, TOF Minimum: 10; PMT
Control: Green 500, Red 600.
Characterization of cell viability as a function of encapsulation
ratio
M07e cells labelled with CFDA SE were monitored by
fluorescence-activated cell sorting (FACS). Following 4.5 days
incubation of the cell-laden microgels in a co-culture environ-
ment, agarose microgels were digested by adding a 0.1% v/v
aqueous solution of agarase enzyme (Sigma) and maintaining
the system for 30 min at 37 1C. In the control experiments
conducted with non-encapsulated M07e cells it was established
that no significant change in cell viability occurred before and
after the digestion of agarose with agarase (82.3 � 1.5% and
81.8 � 1.1%, respectively). The released cells were washed
with HBSS buffer and centrifuged at 1000 rpm for 5 min at r.t.
to remove excess agarose. 7-Aminoactinomycin D (7-AAD)
dye (Molecular Probes, USA) was added at 1 mg ml�1 to cell
suspensions to label non-viable cells. The dispersion of
cells was transferred to an appropriate tube for FACS analysis
and the viability of M07e cells was determined as a function
of the encapsulation ratio of the MBA2 and M07e cell
populations.
Characterization of % rescue of specific lineage of UCB cells
Flow cytometry analysis was performed on cells prior to their
encapsulation in microgels, using a FACSCanto flow
cytometer (BD Biosciences, San Jose, CA, USA), to assess
for cell type frequencies. Cells were stained with the following
conjugated antibodies: CD34-APC (BD Biosciences),
CD133-PE (Miltenyi Biotec, Auburn, CA, USA), CD14-PE
(BD Biosciences), GyA-PE (BD Biosciences), CD41-PE
(BD biosciences), and CD11b-APC (BD Biosciences).
7-AAD dye was added to assess cell viability and isolate live
cells for quantification. UCB cells were then encapsulated in
microgels either in the absence of MBA2 cells (1 : 0 ratio), or in
the presence of MBA2 cells in a 1 : 1 ratio. MBA2 cells were
labelled with CFDA SE dye prior to encapsulation, in order to
distinguish MBA2 cells from UCB cells in subsequent analysis.
The cells encapsulated within the microgels were cultured in
Iscove’s Modified Dulbecco’s Medium (IMDM, Gibco) with
no addition of growth factors or serum substitutes. Following
three days of co-culture incubation, the agarose was digested,
counts of viable cells were performed, and flow cytometry
analysis was repeated. Cells labelled positive in the FITC
channel were excluded to remove all MBA2 cells from
analysis, and viable hematopoietic cell sub-populations were
quantified as
%Rescue¼ # Viable cellsð3 days post-encapsulationÞ# Viable cellsðpre-encapsulationÞ �100%:
ð1Þ
Results and discussion
A schematic of the MF method for producing 3D micro-
environments for cell co-culture is shown in Fig. 1a. Agarose
solutions carrying two distinct populations of cells were
supplied at 37 1C to the MF droplet generator (streams R
and G) at the total constant flow rate of 0.09 mL h�1. Prior to
reaching the T-junction, mixing of streams R and G was
enhanced by passing them through a serpentine channel.
A stream of mineral oil (viscosity 30 cP) was supplied to the
MF device perpendicular to the stream of the mixed cell
suspensions. At the T-junction, the shear stress imposed by
the stream of the mineral oil onto the stream of the mixed cell
suspension led to the generation of droplets of cell suspension.
During the encapsulation, no cells were released into the
continuous phase.
The dimensions of droplets were tuned from 70 to 110 mmby varying the flow rate of the continuous oil phase, Qc, from
to 1.2 to 2.5 mL h�1. The relative standard deviation in droplet
dimensions did not exceed 2.5%. At the end of the down-
stream channel the mineral oil was cooled to 2 1C. Gelation of
the agarose solution led to the transformation of droplets into
agarose particles (microgels). The microgels were collected in
15 mL centrifuge tubes containing HBSS buffer cooled to 4 1C.
Uniform and fast gelation of the droplets yielded microgels
with polydispersity below 3% (Fig. 1b). The final size of the
microgels in the HBSS buffer was approximately 10% smaller
than the diameter of the precursor droplets, owing to droplet
shrinkage upon gelation. Cell-laden microgels were centri-
fuged, washed with the HBSS buffer and transferred to the
appropriate cell media culture (described in Materials and
methods).
To achieve a 100% encapsulation efficiency of cells within
droplets, we utilized the Poisson distribution for the compart-
mentalized cells as in ref. 42
PðxÞ ¼ e�llx
x!ð2Þ
where P(x) is the fraction of droplets expected to contain x
cells, and l is the average number of cells per droplet. The
theoretical value of l was determined by the ratio between the
concentration of cells in the droplet phase and the average
volume of droplets (Fig. S2w). By setting the concentration of
cells to e.g., 8�106 cells mL�1 we used eqn (2) to establish that
for 110 mm-diameter droplets the encapsulation efficiency was
99.7%, sufficiently close to a 100% encapsulation rate.
In the encapsulation experiments, the concentration of cells
in the feed suspension varied from 2�106 to 8�106 cells mL�1.
We controlled the encapsulation efficiency by tuning the
dimensions of droplets from 70 to 110 mm. The theoretical
and experimental results for the total number of cells per
droplet—without differentiating between the different types of
cells—were in excellent agreement. For example, for
70 mm-diameter precursor droplets and cell concentration of
8�106 cells mL�1, we achieved 78.0% encapsulation efficiency
vs. the predicted value of 76.2%, demonstrating concordance
between theory and experiment. An encapsulation efficiency
of 98.5% was achieved for 110 mm-diameter droplets at a
cell concentration of 8�106 cells mL�1 (Fig. 2). At a cell
Dow
nloa
ded
by D
alho
usie
Uni
vers
ity o
n 08
/05/
2013
21:
11:0
2.
Publ
ishe
d on
28
Apr
il 20
11 o
n ht
tp://
pubs
.rsc
.org
| do
i:10.
1039
/C1I
B00
002K
View Article Online
This journal is c The Royal Society of Chemistry 2011 Integr. Biol., 2011, 3, 653–662 657
concentration of 2�106 cells mL�1, a maximum encapsulation
efficiency of 74.6%, in agreement with a theoretical encapsula-
tion efficiency of 75.2%. We attained excellent correlation
between experimental results and theoretical values of Poisson
distribution for each discrete number of cells per droplet in
experiments conducted at a cell concentration of 4�106 and
8�106 cells mL�1 (Fig. S3 and Table S1 and S2w).Co-encapsulation of cells was tested for mES (murine
embryonic stem) cells labelled with Vybrant Cell Tracer
(CFDA SE ‘‘green’’ (Invitrogen)) and CellTracker Orange
(CMTMR ‘‘red’’ (Invitrogen)). Two suspensions, each con-
taining red or green mES cells (streams R and G, respectively),
at a concentration of cells of 8�106 cells mL�1, were supplied
to the MF device at equal volumetric flow rates QR = QG =
0.045 mL h�1. Fig. 3a shows a representative image of red and
green mES cells encapsulated in agarose microgels. Following
the encapsulation of the cell in the microgels and their
subsequent transfer to culture media, cell viability was
79.6 � 2.5% and embryoid bodies were observed 4.5 days
after encapsulation (Fig. 3b). We note that the viability of
individual mESC cells is low (typically, in the range from 20 to
50%) and is dependent on the cell line used.43
In co-encapsulation experiments, the concentration of cells
in each stream was 8�106 cells mL�1 and the total flow rate of
the R and G streams was 0.09 mL h�1. The ratios of flow rates
of the R-to-G streams were 1 : 0, 4 : 1, 1 : 1, 1 : 4, 0 : 1, where the
flow rate ratios of 1 : 0 and 0 : 1 corresponded to the generation
of droplets from a single stream (R or G, respectively), thereby
encapsulating a single cell population. We note that when
the two cell suspensions were mixed, the concentration of the
individual cell populations reduced in proportion to the
relative flow rates. The effect of dilution on the encapsulation
concentration of each cell population was described by the
Poisson distribution expected for 110 mm-diameter droplets
and the corresponding cell concentration in the mixed suspen-
sion (Fig. S4w).The encapsulation of red and green mES cells in the micro-
gels was analyzed using optical fluorescence microscopy
(Zeiss Axio Observer D1) and flow cytometry (Biosort, Union
Biometrica, 250 mm-diameter bore size) (Fig. 4a–e). The
utilization of flow cytometry provided a high-throughput
approach to analyzing cell-laden microgels. In Fig. 4 encap-
sulation of red and green mES cells at various ratios is
reflected in flow cytometry plots with populations in three
separate gating regions, which were determined by the ratio of
the red-to-green cells in the entire microgel population. As
expected, the samples comprising microgels laden solely with
green (R :G= 0 : 1) or solely with red mES cells (R :G= 1 : 0)
(Fig. 4a and e, respectively) were characterized by signals
appearing exclusively in their respective gating regions. Signals
appearing in the third region (double positive for both red and
green cells) were a direct result of a red cell and a green cell
residing in the same path length of the flow cytometer excita-
tion laser. This effect would result in the instrument detector
registering a signal comprised of both a red and green
component. Signals in the double positive region could not
be quantitatively assessed, due to the uncertainty in the
number of red or green cells which contributed to the signal.
To quantitatively compare the results of flow cytometry
experiments, only signals appearing exclusively in the red
region or exclusively in the green region were compared
(Flowjo 7.5). Signals collected exclusively in the red-gated
and the green-gated regions were used for analysis, and signals
outside of either the red or green gate were excluded from the
analysis. The microgels generated at the 1 : 4 flow rate ratio of
the R-to-G streams (Fig. 4b) contained B80% of the cell
population in the green gating region and B20% in the red-
gating region. The microgels produced at equal flow rates of
the R and G streams (Fig. 4c) had statistically equal ratios of
red-to-green cells of 47.7% and 52.3%, respectively, appearing
in the pure red and pure green regions. The microgels formed
at the R-to-G flow rate ratios of 4 : 1 (Fig. 4d) showed
an encapsulation trend that was inverse to that shown in
Fig. 4b, with populations of red and green cells of 79.7%
Fig. 2 Encapsulation efficiency at varying cell concentration and
droplet diameter. Experimental (blue) and theoretical (grey) fractions
of droplets encapsulating at least one cell are plotted for 70 and
110 mm-diameter droplets at cell concentrations of 2�106 and
8�106 cells mL�1. The theoretical fraction of cell-laden droplets was
calculated using eqn (2).
Fig. 3 Cell-laden agarose microgels. (a) Fluorescence optical micro-
scopy image of 100 mm-diameter agarose microgels encapsulating mES
cells. Encapsulation occurred at a total cell concentration of
8�106 cells mL�1 in the droplet phase. The cells appearing to reside
outside the microgel are out of the focus plane. (b) Optical microscopy
image of embryoid bodies formed from mES cells encapsulated in
agarose microgels after 4.5 days after encapsulation. The flow rate of
the oil and aqueous phases were 1.2 and 0.09 mL h�1, respectively.
Scale bar is 100 mm.
Dow
nloa
ded
by D
alho
usie
Uni
vers
ity o
n 08
/05/
2013
21:
11:0
2.
Publ
ishe
d on
28
Apr
il 20
11 o
n ht
tp://
pubs
.rsc
.org
| do
i:10.
1039
/C1I
B00
002K
View Article Online
658 Integr. Biol., 2011, 3, 653–662 This journal is c The Royal Society of Chemistry 2011
and 20.3%, respectively. The results of the analysis of cell
co-encapsulation by flow cytometry and optical microscopy
images were in excellent agreement (Fig. 4f).
One utility of the MF cell co-culture method was demon-
strated by uncovering cellular interaction networks. We
encapsulated in the microgels in different ratios the MBA2
cell line (secreting IL-3) and the IL-3-dependent M07e human
megakaryoblastic leukemia cell line.37 The M07e cells are a
sub-line of the M07 human megakaryoblastic leukemia cell
lines that require interleukin-3 (IL-3) for proliferation and
survival. MBA2, a cell line transfected with the P210 gene,
secretes IL-3. We hypothesized that the survival and prolifera-
tion of encapsulated M07e cells could be controlled by varying
the relative number of MBA2 cells in the microgels (Fig. 5a).
The IL-3 secreted by the MBA2 cells would be used in both an
autocrine and paracrine fashion.
Experiments were carried out by co-encapsulating M07e
and MBA2 cells in the microgels at the flow rate ratios of the
corresponding suspensions of 0 : 1, 1 : 9, 1 : 4, 1 : 1, and 4 : 1 and
by determining the viability of M07e cells after 4.5 days
(Fig. 5b). As expected, low survival of M07e was observed
in the presence of a small relative fraction of MBA2 cells. With
increasing fractions of MBA2 cells in the microgels, M07e
viability increased, reaching a maximum of 65.9 � 8.4% at the
ratio of the encapsulated cell lines of 1 : 1. To verify that higher
M07e viability results from the higher relative number of
MBA2 cells secreting IL-3, we conducted control experiments
with non-encapsulated M07e cells in the presence of exogenous
IL-3. At the IL-3 concentration of 10 ng mL�1, the viability of
M07e cells was 76.3 � 3.5% after 4.5 days. A modestly higher
value than that achieved for a 1 : 1 ratio of encapsulated
MBA2 :M07e cells was presumably caused by the higher levels
of IL-3 in the control system.
Since M07e cells are a parental line to the MBA2 cells,
standard approaches to distinguish the two populations by
side scatter (SSC) vs. front scatter (FSC) could not be applied.
The SSC vs. FSC was utilized in distinguishing between viable
and non-viable M07e cells, based on the change in cell
Fig. 4 Co-encapsulation of mES cells in microgels achieved at varying flow rate ratios of the corresponding cell suspensions. (a)–(e) Left: Relative
fluorescence intensity plots of red and green channels for sorted microgels laden with R1 mES cells labelled with Vybrant CFDA or CellTrackert
Orange CMTMR (green and red cells, respectively). Right: Optical microscopy images of the corresponding microgels. The microgels were
produced at the respective flow rate ratios of the R and G streams: (a) QR :QG = 0 : 1, (b) QR :QG = 1 : 4, (c) QR :QG = 1 : 1. (d) QR :QG = 4 : 1,
(e) QR :QG = 1 : 0. Gating was determined by positive controls comprising cells labelled with only one dye (R :G = 1 : 0 and 0 : 1). (f) Fraction, a,of green and red cells encapsulated at different ratios of QR :QG. Light and dark green bars show the fraction of encapsulated green cells,
determined by image analysis and flow cytometry, respectively. Light and dark red bars represent the fraction of encapsulated red cells, determined
by image analysis and flow cytometry, respectively. Fluorescence intensity scale is defined by the sorting parameters (ESIw). Scale bar is 100 mm.
Dow
nloa
ded
by D
alho
usie
Uni
vers
ity o
n 08
/05/
2013
21:
11:0
2.
Publ
ishe
d on
28
Apr
il 20
11 o
n ht
tp://
pubs
.rsc
.org
| do
i:10.
1039
/C1I
B00
002K
View Article Online
This journal is c The Royal Society of Chemistry 2011 Integr. Biol., 2011, 3, 653–662 659
morphology upon cell death. In order to differentiate between
the two cell lines, M07e cells were marked with the CFDA dye
prior to encapsulation in the microgels. The dye was retained
by the cells throughout cellular development, inherited by
daughter cells, and was not transferred to adjacent cells.44–46
Furthermore, to quantify the viability of M07e cells after their
encapsulation in microgels, 7-Aminoactinomycin D (7-AAD)
dye was added to all cell samples prior to FACS analysis.
7-AAD cannot readily pass through intact cell membranes
and is commonly used for DNA staining of dead cells. The
fluorescence intensity as a result of cellular uptake of 7-AAD
was determined by flow cytometry.
Typical histograms of the cellular uptake of 7-AAD dye by
CFDA-labelled cells at various co-culture ratios of M07e-to-
MBA2 cells are shown in Fig. 5c.
To verify that endogenously produced IL-3 resulted from
cell co-encapsulation within the microgels, rather than
diffusion of IL-3 through the medium, we carried out control
experiments. We encapsulated either MBA2, or M07e cells in
agarose microbeads and mixed cell-laden microgels in a 1 : 1
ratio in the media with varying volume. Encapsulation of each
cell line in the microgels was carried out at a cell concentration
of 8�106 cells mL�1 Fig. 6 shows that regardless of the volume
of the media, M07e cells had viability not exceeding 20%. We
attribute this effect to the limited diffusion of IL-3 from the
MBA2-laden microgels into the M07e-laden microgels, there-
by resulting in low concentrations of endogenously produced
IL-3 in the bulk media.
In order to explore another application of the proposed MF
platform, we investigated the impact of a growth factor on
Fig. 5 Variation in the viability of M07e human megakaryoblastic leukemia cells examined at varying encapsulation ratio of MBA2 to M07e
cells. (a) Schematic of co-culture of MBA2 and M07e cells in different number ratios. MBA2 cells secrete IL-3 which is required for the survival of
both cell populations. Increase in the relative number of MBA2 cells results in an increase in M07e survival rate, due to the local increase in IL-3.
(b) Viability of M07e cells plotted as a function of the number ratio of MBA2 cells-to-M07e cells encapsulated in 100 mm-diameter agarose
microgels and analyzed by flow cytometry. Asterisk denotes non-encapsulated M07e cells in the presence of exogenous IL-3. (c) Representative
flow cytometry histograms of cellular uptake of 7-AAD fluorescence intensity for varying M07e co-culture conditions 4.5 days after encapsulation.
X-axis represents fluorescence intensity of cells stained with 7-AAD. Numerical percentages represent the mean and standard error of M07e
cellular uptake of 7-AAD dye from at least three independent experiments.
Dow
nloa
ded
by D
alho
usie
Uni
vers
ity o
n 08
/05/
2013
21:
11:0
2.
Publ
ishe
d on
28
Apr
il 20
11 o
n ht
tp://
pubs
.rsc
.org
| do
i:10.
1039
/C1I
B00
002K
View Article Online
660 Integr. Biol., 2011, 3, 653–662 This journal is c The Royal Society of Chemistry 2011
specific cell types within the primary human hematopoietic
SC culture system (Fig. 7a). Hematopoietic cell culture is
highly complex and heterogeneous. The heterogeneity is
compounded by system dynamics and paracrine signalling
cascades that act to regulate the system. A study of the role
of specific ligand-cell interactions within the context of this
system would enable a greater understanding of signalling
networks and provide a means of manipulating the culture
system, however, currently this task is biologically and tech-
nically challenging. Fig. 6 shows that cells encapsulated in
each microgel are isolated from each other. This feature
enables high throughput (parallel) interrogation of inter-
actions between two types of cells, when they are interacting
within the microgel, but are isolated from each other between
different microgels. An important aspect is the co-localization
of two types of cells within a small sample volume. In
comparison to bulk encapsulation in a macroscale agarose
gel, the probability of putting two types of cell populations in
close spatial proximity to each other would be very low and
highly variable, making it difficult to examine interactions and
their effects across a highly heterogeneous cell hierarchy in a
controlled manner.
In our work, after co-culturing the IL-3-secreting MBA2
cells with heterogeneous UCB cells, the UCB cells were
assessed to determine cell type specific effect of localized
IL-3 secretion on viability. By assessing the cell phenotype
of the viable UCB cells, the rescue (relative to control condi-
tions where the paracrine cell was not present) of specific
hematopoietic cell types was quantified. Fig. 7b shows poor
survival of individually encapsulated UCB cells when no
growth factor (IL-3) was added. Furthermore, when bulk
exogenous IL-3 was added to the culture, weak enhancement
of UCB cell survival was observed, likely due to the limited
IL-3 diffusion across the agarose microgel matrix to the UCB
cells. When the UCB cells were cultured in the presence of IL-3
secreting MBA2 cells, a dramatic increase in cell survival was
observed, indicating that the local (in-bead) secretion of IL-3
enabled survival and detection of responsive phenotypes from
a heterogeneous input cell population. Importantly, the
effectiveness of the rescue of the UCB cells varied among the
hematopoietic cell types that were present in the hetero-
geneous culture. Fig. 7b illustrates that monocytes (CD14+)
were most impacted by the co-culture, achieving 480%
rescue, which suggested the importance of IL-3 to the viability
of this cell type. Intermediate IL-3 dependent viability was
observed in the hematopoietic progenitors (CD34+, CD133+),
erythroid cells (GyA+), and granulocytes (CD11b+). Minimal
rescue (effect on viability) was observed in the megakaryocyte
population (CD41+), indicating that IL-3 did not play a role
Fig. 7 Co-encapsulation of MBA2 cells with UCB cells were used to determine the survival effect of IL-3 on sub-populations within a
hematopoietic culture system. (a) Schematic of co-culture of MBA2 cells and UCB cells. (i) Encapsulation of solely UCB cells without exogenous
IL-3 results in poor cell rescue. (ii) Addition of exogenous IL-3 (10 ng mL�1) results in low rescue of several UCB phenotypes.
(iii) Co-encapsulation of MBA2 cells and UCB cells results in low, moderate, and high rescue of specific UCB phenotypes. (b) % Rescue of specific
lineages of hematopoietic UCB cells co-encapsulated with MBA2 cells. % rescue of specific UCB phenotypes were measured 3 days after encapsulation
in a non-supportive media using eqn (1). This is a representative experiment conducted from at least two independent experiments (n = 2).
Fig. 6 Viability of encapsulated M07e cells at varying media volume.
Microgels containing only MBA2 cells and only M07e cells mixed in a
1 : 1 ratio at varying media volume. Asterisk denotes non-encapsulated
M07e cells in the presence of exogenous IL-3.
Dow
nloa
ded
by D
alho
usie
Uni
vers
ity o
n 08
/05/
2013
21:
11:0
2.
Publ
ishe
d on
28
Apr
il 20
11 o
n ht
tp://
pubs
.rsc
.org
| do
i:10.
1039
/C1I
B00
002K
View Article Online
This journal is c The Royal Society of Chemistry 2011 Integr. Biol., 2011, 3, 653–662 661
in the survival of this cell type under these conditions. These
results show the differential effect of IL-3 among cell types
within a hematopoietic culture system and provides a high-
throughput means of screening signalling effects in stem
cell systems. This experiment also suggests that there are
qualitative differences in biological response seen by local
(cell mediated—in bead) delivery of soluble factors versus
global (media added) soluble factor delivery, in part due to
higher effective concentrations likely achievable through con-
tinuous local delivery.
Given the complex heterogeneity of hematopoietic cell
cultures, it has been difficult to discern the varying effects of
individual growth factors on specific cell types within the
heterogeneous culture. Paracrine signalling cascades and
culture dynamics further complicate traditional methods of
analysis. The co-encapsulation of the MBA2 cells and the
UCB cells allows one to determine the hematopoietic cell types
that can be rescued by the localized presence of IL-3. Thus, a
MF platform provides a strategy for intricate co-culture
studies, while minimizing the barriers of complex paracrine
interactions that are seen in bulk cultures and eliminating
difficulties in isolating individual cell types from a hetero-
geneous culture. By doing so, specific ligand-cell interactions
can be precisely interrogated and new biological hypotheses
can be developed and investigated.
Conclusions
The MF platform provides a fast, efficient, and easy to
implement method for the high-throughput generation of cell
co-culture compound libraries which, after incubation, can be
analyzed by optical microscopy and flow cytometry. The MF
strategy offers the speed and the control over the total number
of compartmentalized cells and the relative numbers of
different cell populations. Co-encapsulation and co-culture
of MBA2 and M07e cells at varying ratios demonstrated the
ability of the MF approach to modulate paracrine signalling
amongst cell populations in a well-defined microenvironment.
Furthermore, the MF method can be utilized as a tool to
investigate the impact of specific growth factors on a hetero-
geneous cell population. Co-encapsulation of MBA2 cells with
UCB cells was used to determine sub-populations of the
hematopoietic cell types that are largely IL-3 dependent. This
approach offers two important abilities: (1) the use of isolated
micro-environments provides the capability for studying
biological networks which otherwise could not be easily
isolated in a bulk environment. (2) The MF platform offers
the ability to control the ratio of the two cell populations
(a secreted factor population and a heterogeneous cell population).
As was demonstrated, the ratio of the encapsulated species is
important as it determines the dose of the signal given by the
secreting cells and received by the receptor cells (and its
consequent biological response). Our findings indicate that
direct co-culture had different biological effects from soluble
factor addition.
Further development and improvement of the MF platform
will increase the breath of its applications. The ability to screen
the phenotypic and functional effects of cell–cell interactions
in-bead, ideally with live cells, would prevent the need to
remove cells from the beads for analysis. Additionally,
co-encapsulation of various test cell populations combined
with microfluidically mediated variation in properties of the
microenvironment (i.e., growth factors, physical properties
(stiffness), or other ‘‘niche’’ parameters would add to the
range of conditions that can be queried. Given the rapid
development of microfluidic encapsulation technologies, these
next steps should emerge shortly.
Acknowledgements
The authors acknowledge financial support of this work by the
Natural Sciences and Engineering Research Council of
Canada and the Canadian Institutes for Health Research.
Biosort flow cytometry results were acquired by Dr Rock
Pulak (Union Biometrica).
References
1 D. E. Discher, D. J. Mooney and P. W. Zandstra, Science, 2009,324, 1673–1677.
2 J. Voog and D. L. Jones, Cell Stem Cell, 2010, 6, 103–115.3 R. Peerani and P. W. Zandstra, J. Clin. Invest., 2010, 120, 60–70.4 M. P. Lutolf, P. M. Gilbert and H. M. Blau, Nature, 2009, 462,433–441.
5 F. Grinnell, J. Cell Sci., 1992, 101, 1–5.6 A. F. Black, F. Berthod, N. L’Heureux, L. Germain andF. A. Auger, FASEB J., 1998, 12, 1331–1340.
7 J. L. Camps, S. M. Chang, T. C. Hsu, M. R. Freeman, S. J. Hong,H. E. Zhau, A. C. Voneschenbach and L. W. K. Chung, Proc.Natl. Acad. Sci. U. S. A., 1990, 87, 75–79.
8 H. J. Rippon and A. E. Bishop, Cell Proliferation, 2004, 37, 23–34.9 T. Nakano, H. Kodama and T. Honjo, Science, 1996, 272,722–724.
10 E. J. Lee, E. W. L. Chan and M. N. Yousaf, ChemBioChem, 2009,10, 1648–1653.
11 A. P. McGuigan and M. V. Sefton, Tissue Eng., 2007, 13,1079–1089.
12 A. Y. Hsiao, Y. S. Torisawa, Y. C. Tung, S. Sud, R. S. Taichman,K. J. Pienta and S. Takayama, Biomaterials, 2009, 30, 3020–3027.
13 Y. A. Du, E. Lo, S. Ali and A. Khademhosseini, Proc. Natl. Acad.Sci. U. S. A., 2008, 105, 9522–9527.
14 G. Karoubi, M. L. Ormiston, D. J. Stewart and D. W. Courtman,Biomaterials, 2009, 30, 5445–5455.
15 S. M. Dang, S. Gerecht-Nir, J. Chen, J. Itskovitz-Eldor andP. W. Zandstra, Stem Cells, 2004, 22, 275–282.
16 E. Tumarkin and E. Kumacheva, Chem. Soc. Rev., 2009, 38,2161–2168.
17 A. Huebner, S. Sharma, M. Srisa-Art, F. Hollfelder, J. B. Edel andA. J. deMello, Lab Chip, 2008, 8, 1244–1254.
18 M. E. Vincent, W. S. Liu, E. B. Haney and R. F. Ismagilov, Chem.Soc. Rev., 2010, 39, 974–984.
19 J. Clausell-Tormos, D. Lieber, J. C. Baret, A. El-Harrak,O. J. Miller, L. Frenz, J. Blouwolff, K. J. Humphry, S. Koster,H. Duan, C. Holtze, D. A. Weitz, A. D. Griffiths andC. A. Merten, Chem. Biol., 2008, 15, 427–437.
20 E. Brouzes, M. Medkova, N. Savenelli, D. Marran,M. Twardowski, J. B. Hutchison, J. M. Rothberg, D. R. Link,N. Perrimon and M. L. Samuels, Proc. Natl. Acad. Sci. U. S. A.,2009, 106, 14195–14200.
21 J. U. Shim, L. F. Olguin, G. Whyte, D. Scott, A. Babtie, C. Abell,W. T. S. Huck and F. Hollfelder, J. Am. Chem. Soc., 2009, 131,15251–15256.
22 A. Huebner, D. Bratton, G. Whyte, M. Yang, A. J. deMello,C. Abell and F. Hollfelder, Lab Chip, 2009, 9, 692–698.
23 A. Huebner, L. F. Olguin, D. Bratton, G. Whyte, W. T. S. Huck,A. J. deMello, J. B. Edel, C. Abell and F. Hollfelder, Anal. Chem.,2008, 80, 3890–3896.
24 S. Y. Teh, R. Lin, L. H. Hung and A. P. Lee, Lab Chip, 2008, 8,198–220.
Dow
nloa
ded
by D
alho
usie
Uni
vers
ity o
n 08
/05/
2013
21:
11:0
2.
Publ
ishe
d on
28
Apr
il 20
11 o
n ht
tp://
pubs
.rsc
.org
| do
i:10.
1039
/C1I
B00
002K
View Article Online
662 Integr. Biol., 2011, 3, 653–662 This journal is c The Royal Society of Chemistry 2011
25 S. Sugiura, T. Oda, Y. Izumida, Y. Aoyagi, M. Satake, A. Ochiai,N. Ohkohchi and M. Nakajima, Biomaterials, 2005, 26, 3327–3331.
26 W. H. Tan and S. Takeuchi, Adv. Mater., 2007, 19, 2696–2701.27 P. Panda, S. Ali, E. Lo, B. G. Chung, T. A. Hatton,
A. Khademhosseini and P. S. Doyle, Lab Chip, 2008, 8, 1056–1061.28 D. Chicheportiche, S. Darquy, J. Lepeintre, P. Chenard,
C. Khosrovaninejad, H. Briandet, J. Rouchette and G. Reach,Diabetologia, 1988, 31, A479–A479.
29 A. D. Griffiths and D. S. Tawfik, Trends Biotechnol., 2006, 24,395–402.
30 J. J. Agresti, E. Antipov, A. R. Abate, K. Ahn, A. C. Rowat,J. C. Baret, M. Marquez, A. M. Klibanov, A. D. Griffiths andD. A. Weitz, Proc. Natl. Acad. Sci. U. S. A., 2010, 107, 4004–4009.
31 S. Vyawahare, A. D. Griffiths and C. A. Merten, Chem. Biol., 2010,17, 1052–1065.
32 A. Hayashi and T. Kanzaki, Food Hydrocolloids, 1987, 1, 317–325.33 C. A. Agudelo, Y. Teramura and H. Iwata, Transplantation, 2009,
87, 29–34.34 D. Pelaez, C. Y. C. Huang and H. S. Cheung, Stem Cells Dev.,
2009, 18, 93–102.
35 A. Kumachev, J. Greener, E. Tumarkin, E. Eiser, P. W. Zandstraand E. Kumacheva, Biomaterials, 2011, 32, 1477–1483.
36 S. Lahooti and M. V. Sefton, Cell Transplant., 2000, 9, 785–796.37 C. Sirard, P. Laneuville and J. E. Dick, Blood, 1994, 83, 1575–1585.38 T. Thorsen, R. W. Roberts, F. H. Arnold and S. R. Quake, Phys.
Rev. Lett., 2001, 86, 4163–4166.39 A. K. Hadjantonakis and A. Nagy, Genesis, 2000, 27, 95–98.40 D. C. Kirouac, G. J. Madlambayan, M. Yu, E. A. Sykes, C. Ito
and P. W. Zandstra, Mol. Syst. Biol., 2009, 5, 293–313.41 Y. N. Xia and G. M. Whitesides, Annu. Rev. Mater. Sci., 1998, 28,
153–184.42 F. A. Haight,Handbook of the Poisson distribution, ed. F. A. Haight,
Wiley, New York, 1967.43 S. M. Dang, M. Kyba, R. Perlingeiro, G. Q. Daley and
P. W. Zandstra, Biotechnol. Bioeng., 2002, 78, 442–453.44 A. Nose and M. Takeichi, J. Cell Biol., 1986, 103, 2649–2658.45 A. B. Lyons and C. R. Parish, J. Immunol. Methods, 1994, 171,
131–137.46 P. D. Hodgkin, J. H. Lee and A. B. Lyons, J. Exp. Med., 1996, 184,
277–281.
Dow
nloa
ded
by D
alho
usie
Uni
vers
ity o
n 08
/05/
2013
21:
11:0
2.
Publ
ishe
d on
28
Apr
il 20
11 o
n ht
tp://
pubs
.rsc
.org
| do
i:10.
1039
/C1I
B00
002K
View Article Online