Digital Microfluidics for ... - University of Toronto
Transcript of Digital Microfluidics for ... - University of Toronto
Digital Microfluidics for Multidimensional Biology
Submitted by:
Irwin Adam Eydelnant
MEng, McGill University, 2008
BEng, McGill University, 2007
A thesis submitted in conformity with the requirements for the degree of Doctor of Philosophy
Institute of Biomaterials and Biomedical Engineering
University of Toronto
© Copyright by Irwin Adam Eydelnant (2013)
ii
Thesis Title: Digital Microfluidics for Multidimensional Biology
Degree and Year: Doctor of Philosophy, 2013
Name: Irwin Adam Eydelnant
Department: Institute of Biomaterials and Biomedical Engineering
University: University of Toronto
Abstract
Digital microfluidics (DMF) has emerged in the past decade as a novel microfluidic
paradigm. As a liquid handling technology, DMF facilitates the electrostatic manipulation
of discrete nano- and micro- litre droplets across open electrode arrays providing the
advantages of single sample addressability, automation, and parallelization. This thesis
presents DMF advances toward improved functionality and compatibility for automated
miniaturized cell culture in two and three dimensions. Through the development and
integration of surface patterning techniques we demonstrate a virtual microwell method
for high precision on-device reagent dispensing in one and two plate DMF geometries.
These methods are shown to be compatible with two-dimensional culture of immortalized
cell lines on ITO, primary cells on coated surfaces, and for co-culture assays. We further
extrapolate this method for the formation of microgels on-demand where form micro scale
hydrogel structures through passive dispensing in a wide array of geometries. With this
system we interrogate three-dimensional cell culture models, specifically for the
recapitulation of kidney epthelialization and the analysis of functional cardiac microgels.
iii
Acknowledgements
A thesis of this nature is produced through the support, guidance, and encouragement of a
community. There are many to whom I am grateful, who I crossed paths over the course of my
PhD tenure, and influenced either myself or my science. As they are too many to list, I
acknowledge a few here who made a particular impact:
Dr. Aaron Wheeler was critical to the success of this thesis, he created a laboratory environment
that knew no barriers for the curious or scientific mind. His unlimited patience allowed for
exploration and learning to occur organically and irrespective of preconceptions.
Dr. William Ryu and his laboratory. They formed my academic home over the course of my PhD
and I’ll always be grateful for being an unofficial part of the group.
Dr. Nathalie Tufenkji for her support and advising during my tenure at McGill University and
the years that followed. She pushed me always to go further and take chances.
My father for his complete support for every decision made at every intersection. For his belief
in the good of people and their ability to affect positive change in this world. His fingers were
crossed for every exam of my academic career including the PhD final defense.
My mother for her belief in my abilities to achieve anything. Her strength and unwavering spirit
have influenced my life greatly. She has always been my biggest fan.
Thank you to you all.
iv
Table of Contents
Acknowledgements ...................................................................................................................................... iii
Table of Contents ......................................................................................................................................... iv
List of Abbreviations .................................................................................................................................. vii
List of Figures ............................................................................................................................................ viii
List of Equations ........................................................................................................................................ xiv
List of Tables .............................................................................................................................................. xv
Overview of chapters ................................................................................................................................. xvi
Overview of author contributions .............................................................................................................. xix
Chapter 1. Introduction ........................................................................................................................... 1
1.1 Historical perspectives on the miniaturization of biology ............................................................ 1
1.2 Microfluidics Paradigms ............................................................................................................... 3
1.3 Digital Microfluidics ..................................................................................................................... 6
1.4 Digital microfluidic theory............................................................................................................ 6
1.4 DMF compatibility with two-dimensional cell culture ............................................................... 14
1.5 Microfluidics for three-dimensional cell culture ......................................................................... 17
1.6 Assays and integration ................................................................................................................ 19
1.7 Future of DMF ............................................................................................................................ 20
Chapter 2. Virtual microwells for digital microfluidic reagent dispensing and cell culture ................. 21
Summary ................................................................................................................................................. 21
2.1 Introduction ................................................................................................................................. 22
2.2 Methods and Materials ................................................................................................................ 25
2.2.1 Reagents .............................................................................................................................. 25
2.2.2 Two-plate DMF bottom-plate fabrication ........................................................................... 25
2.2.3 Two-plate DMF top-plate fabrication ................................................................................. 27
2.2.4 Two-plate DMF device assembly and operation ................................................................ 28
2.2.5 Single-plate DMF device fabrication, assembly, and operation ......................................... 28
2.2.6 DMF dispensing experiments ............................................................................................. 29
2.2.7 Cell Culture and experiments .............................................................................................. 31
2.3 Results and Discussion ............................................................................................................... 31
v
2.3.1 Lift-off patterning ............................................................................................................... 31
2.3.2 Passive dispensing and virtual microwells .......................................................................... 32
2.3.3 Active vs. passive dispensing .............................................................................................. 40
2.3.4 Lift-off vs. protein absorption for passive dispensing......................................................... 41
2.4 Conclusion .................................................................................................................................. 43
Chapter 3. A digital microfluidic platform for primary cell culture and analysis ................................ 44
Summary ................................................................................................................................................. 44
3.1 Introduction ................................................................................................................................. 45
3.2 Methods and Materials ................................................................................................................ 47
3.2.1 Reagents and Materials ....................................................................................................... 47
3.2.2 DMF Device Fabrication and Operation ............................................................................. 47
3.2.3 Primary Cell Isolation and Maintenance ............................................................................. 50
3.2.4 DMF Cell Culture ............................................................................................................... 51
3.2.5 DMF Staining and Microscopy ........................................................................................... 51
3.2.6 DMF Monocyte Adherence Assay ...................................................................................... 52
3.3 Results and Discussion ............................................................................................................... 53
3.3.1 Digital Microfluidic Primary Cell Culture .......................................................................... 53
3.3.2 Digital Microfluidic Microscopy, Fixation, Permeabilization, and Staining ...................... 56
3.3.3 Digital Microfluidic Monocyte Adhesion Assay ................................................................ 58
3.3.4 Conclusions ......................................................................................................................... 59
Chapter 4. Microgels on-demand ......................................................................................................... 62
Summary ................................................................................................................................................. 62
4.1 Introduction ................................................................................................................................. 63
4.2 Methods....................................................................................................................................... 65
4.2.1 Reagents .............................................................................................................................. 65
4.2.2 DMF device fabrication ...................................................................................................... 65
4.2.3 DMF device assembly and operation .................................................................................. 65
4.2.4 Hydrogel pillar formation and addressing........................................................................... 66
4.2.5 Diffusion analysis and modelling ....................................................................................... 68
4.2.6 Composite hydrogel formation ........................................................................................... 69
4.2.7 Cell Culture ......................................................................................................................... 70
4.2.8 Cell viability and cell distribution ....................................................................................... 70
vi
4.2.9 Epithelialization experiments .............................................................................................. 71
4.2.10 Comparison to conventional fluid handling ........................................................................ 71
4.3 Results and discussion ................................................................................................................ 72
4.3.1 Microgels on-demand ......................................................................................................... 72
4.3.2 Reagent exchange and diffusion into hydrogels ................................................................. 77
4.3.3 Pitch, Multi-Component Arrays, and Composite Microgels............................................... 79
4.3.4 DMF recapitulation of epithelialization .............................................................................. 82
4.3.5 Conclusion .......................................................................................................................... 88
Chapter 5. Cardiac microgels ............................................................................................................... 89
Summary ................................................................................................................................................. 89
5.1 Introduction ................................................................................................................................. 90
5.2 Methods....................................................................................................................................... 92
5.2.1 Reagents .............................................................................................................................. 92
5.2.2 DMF device fabrication ...................................................................................................... 93
5.2.3 DMF device assembly and operation .................................................................................. 93
5.2.4 Cell culture and cardiac microgel formation/addressing..................................................... 94
5.2.5 Cardiomyocyte treatment .................................................................................................... 94
5.2.6 Single cell analysis of cardiomyocyte activity .................................................................... 95
5.2.7 Cardiac activity coefficient (CAC) for fast analysis ........................................................... 95
5.3 Results and Discussion ............................................................................................................... 96
5.3.1 DMF device fabrication and design .................................................................................... 96
5.3.2 On-demand cardiac microgel formation and assay ............................................................. 97
5.3.3 Higher efficiency analysis of cardiac microgel activity .................................................... 104
5.3.4 Conclusion ........................................................................................................................ 105
Chapter 6. Concluding Remarks and Perspectives on the Future ....................................................... 109
6.1 Conclusions ............................................................................................................................... 109
6.2 Future perspectives ................................................................................................................... 111
References ................................................................................................................................................. 113
APPENDIX A ............................................................................................................................................. 128
vii
List of Abbreviations
2D Two-dimensional
3D Three-dimensional
CAC Cardiac activity coefficient
CM Cardiac myocyte
DMF Digital microfluidics
FITC Fluorescein isothiocyanate
hESC Human embryonic stem cells
FBS Fetal bovine serum
HTS High-throughput screening
ICAM-1 Intercellular adhesion molecule 1
ITO Indium tin oxide
MDCK Marbin Darby canine kidney epithelial cell
MS Mass spectrometry
NBF Neutral buffered formalin
PAEC Porcine aortic endothelial cell
PAVIC Porcine aortic valve interstitial cell
PAVEC Porcine aortic valve endothelial cell
PCB Printed circuit board
PDMS Polydimethylsiloxane
P-S Penicillin and streptomycin
RPMI Roswell Park Memorial Institute medium
SPR Surface plasmon resonance
TCPS Tissue culture polystyrene
THP-1 Human monocytic cell line from acute monocytic leukemia patient
TNF-α Tumor necrosis factor alpha
VCAM Vascular cell adhesion molecule 1
VM Virtual microwell
viii
List of Figures
Figure 1-1: Microfluidic paradigms. (A) Continuous flow channel microfluidic systems can
exploit the laminar flow properties of micron scale confined flows to generate interesting flow
patterns such as the gradient generator depicted here.11
(B) Two phase droplet-in-channel
systems are capable of high-throughput generation of individual droplet compartments. Here two
strategies for droplet formation are depicted within a T-junction (top) and by flow focusing
(bottom).12
(C) Digital microfluidics allows for the manipulation of discrete droplets across
arrays of electrodes. (Permissions requested). ................................................................................ 5
Figure 1-2 Digital microfluidic device geometry. (A) A DMF device consisting of 144
independent actuation electrodes. (B) Devices are composed of two parallel plates. The bottom
plate is patterned with an array of electrodes and coated by a hydrophobic insulator. The top
plate bears the counter electrode and is covered with a hydrophobic coating. (C) Schematic of
droplet translation principles. Separation of charge occurs across the dielectric layer acting on
charges or dipoles in the droplet thereby driving translation. ......................................................... 8
Figure 1-3: Theoretical framework of DMF. (A) Equivalent circuit analysis of DMF driving
force mechanisms. (B) Force estimation for a two-plate DMF device operating on PBS, DI
water, toluene and methanol. Forces are based on a 1 mm2 electrode size, 6 μm of Parylene-C,
235 nm of Teflon- VRMS for a range of
frequencies (100 Hz to 1 MHz). (Adapted from Choi et al.36
– Permissions requested).............. 12
Figure 1-4: Cell culture on DMF. (A) Virtual microwells: Droplets containing cells suspended
in media are translated across patterned hydrophilic sites where a subdroplet is generated by
surface interaction forces. The device is then flipped to allow for cells to settle and adhere to the
hydrophilic site. Here, cells stained with calcein-AM are imaged by stereomicroscopy
immediately after seeding on device. (B) Primary cells: Aortic interstitial cells isolated from pig
hearts cultured for 48 hours on device were then fixed and stained with Hoescht (blue – nuclei)
and Phalloidin (green – actin). Imaging was performed by epifluorescence microscopy. Scale bar
= 200 µm. (C) Multiplexing: Automation combined with multiplexed devices allows for rapid
screening of multiple conditions. Here 16 conditions are screened simultaneously. ................... 16
Figure 2-1: Two-plate digital microfluidic (DMF) device design and assembly. (A) Exploded
view of a device, comprising a bottom plate with patterned electrodes and a top plate bearing
patterned hydrophilic sites. (B) Side-view, not to scale. (C) Schematic depicting two reagent-
dispensing mechanisms on DMF. Active dispensing (i & ii) involves electrostatic stretching of a
reagent from the reservoir followed by splitting. Passive dispensing (iii & iv) occurs
spontaneously as a source droplet is translated across the hydrophilic site. The inset is a three-
dimensional depiction of a virtual microwell, VM (i.e., a droplet formed by passive dispensing).
VM volume is dictated by the diameter of the hydrophilic site (d) and the distance between top
and bottom plates (h). ................................................................................................................... 23
ix
Figure 2-2: X-ray photoelectron spectroscopy evaluation of patterned surfaces. To evaluate the
chemical composition of exposed hydrophilic sites and adjacent Teflon surfaces XPS
measurements were taken on patterned and unpatterned surfaces for comparison: (A) ITO
composition at hydrophilic sites, (B) Teflon on patterned slides, (C) unprocessed ITO surfaces,
and (D) unprocess Teflon surfaces. .............................................................................................. 33
Figure 2-3: Dry passive dispensing to form virtual microwells. (A) Video stills (top-to-bottom)
depicting dry passive dispensing. The dashed circle in panel (i) indicates the position of the
hydrophilic site. (B) Volumes of droplets dispensed in dry passive dispensing as a function of
spacer height and hydrophilic site diameter (n = 5). Asterisks (*) indicate that source droplets
were formed from two actively dispensed droplets. Error bars are 1 S.D. (C) Parameter NVM
calculated for each experimental condition in (B). The shaded region, NVM < 2, indicates
conditions in which two actively dispensed droplets were required to generate the source droplet
for successful passive dispensing.................................................................................................. 35
Figure 2-4: Wet passive dispensing to exchange fluid in a virtual microwell. (A) Video stills
(top-to-bottom) depicting wet passive dispensing in which the virtual microwell contained blue
dye at the hydrophilic site (i) and a red dye source droplet is actuated across the virtual
microwell displacing the blue droplet (ii-v). (B) Multiple passes of reagent across virtual
microwells with varying diameters for 160 µm spacer height. The gray and white bars represent
devices operated with a surrounding matrix of air and mineral oil, respectively. Error bars are 1
S.D. ............................................................................................................................................... 37
Figure 2-5: Single-plate DMF passive dispensing. (A) Picture of a single-plate device depicting
a source droplet and a passively dispensed droplet. (B) Schematic depicting the single-plate
device geometry. ........................................................................................................................... 39
Figure 2-6: Active and passive dispensing as a function of reagent viscosity. Sucrose solutions
of varying viscosity were dispensed on DMF either by active or passive dispensing onto 1500
µm diameter hydrophilic sites (n = 6). Dispensed volumes are plotted as a function of solution
viscosity. Error bars are ± 1 S.D., 95% confidence intervals are indicated by shaded regions, and
the mean dispensed volume for each dispensing mechanism is indicated by a solid horizontal
line................................................................................................................................................. 39
Figure 2-7: Comparison of hydrophilic sites formed by adsorbed protein (on the bottom plate)
vs. liftoff (on the top plate) for dispensing cells into virtual microwells. (A) Results of five trials
seeding MDCK cells (5 105 cells/mL) by passive dispensing. (B) Bright-field images of
MDCK cells seeded on fibronectin coated Teflon and indium tin oxide after 6 hours. Scale bar =
50 µm. ........................................................................................................................................... 42
Figure 3-1: (A) Photograph of DMF device designed for primary cell culture and analysis. A
series of droplets (coloured with red dye for visualization) are positioned at patterned hydrophilic
sites on a device. (B) Schematic of device geometry. The top plate is patterned by a liftoff
procedure to expose hydrophilic sites. The bottom plate bears an array of individually
addressable electrodes with patterned optical windows for imaging. (C) Top and side view
x
schematic of passive dispensing on hydrophilic sites. (i-ii) A droplet is manipulated to the
hydrophilic site. By actuation of subsequent electrodes the droplet is (iii) stretched then (iv)
passively dispensed, forming a virtual microwell. (D) Side view schematic of device orientation
during experimentation. Devices are maintained right-side up during droplet actuation and are
positioned upside-down during all incubations. ........................................................................... 48
Figure 3-2: Phase contrast images of PAECs, PAVICs, and PAVECs cultured on a DMF device
(top) and in TCPS flasks (bottom). Scale bar = 200 µm. In the DMF images, the bottom plate is
closest to the objective, and the focus is on the layer of cells on the top plate. The cells are
viewed through the circular optical window between two electrodes on the bottom plate (which
are observable but slightly out of focus). ...................................................................................... 57
Figure 3-3: Fluorescent images of PAECs, PAVECs, and PAVICs after fixing, permeabilizing,
and staining on a DMF device. The stains selected for F-actin (FITC-phalloidin, green) and
nuclei (Hoechst, blue). Images were taken at both 10x magnification (top row) and 40x
magnification (bottom row). Scale bar = 200 µm (top row) and 50 µm (bottom row). In these
images, the top plate is closest to the objective. ........................................................................... 57
Figure 3-4: A monocyte adhesion assay performed on DMF-cultured primary PAECs. (A)
Nuclear-stained (Hoechst, red) THP-1 monocytes adhered to PAECs (calcein AM, green).
Representative images of nuclear-stained monocytes adhered to (B) non-stimulated and (C)
TNF-α-stimulated PAECs. In these images, the top plate is closest to the objective. (D)
Monocytes displayed greater adhesion on TNF-α-stimulated PAECs relative to control non-
stimulated PAECs. Data presented as mean ± standard deviation. *P < 0.05. Scale bar = 200 µm.
....................................................................................................................................................... 60
Figure 4-1: Digital microfluidic device geometry. (A) Exploded view of device, comprising a
bottom plate with patterned electrodes and a top plate bearing patterned hydrophilic sites. (B)
Side-view of device, not to scale. (C) Schematic depicting principle of device operation. ......... 67
Figure 4-2: Microgels on-demand. (A-H) Frames from a movie (top) and side-view schematic
(bottom) depicting a sol-state hydrogel droplet containing a fluorescent dye actively dispensed
from a reservoir (A) and then electrostatically manipulated to a patterned hydrophilic site (B &
C). A second droplet is then actively dispensed (D) and passed across the hydrophilic site (E-H).
Scale bars = 2 mm. When the sol-state droplet is passed across the hydrophilic site a sub-droplet
is generated in the shape of the hydrophilic site. Upon crosslinking, each droplet forms a solid
gel pillar. Various geometries can be generated and visualized with epifluorescent
stereomicroscopy (I & J) or confocal microscopy (K & L). A movie depicting the formation of
an array of different microgel shapes can be found online in the supplementary information.
Scale bars = 1 mm. ........................................................................................................................ 73
Figure 4-3: Precision of microgel on-demand formation and reagent exchange. Frames from a
movie (A-D) depicting sol-state Geltrex being passively dispensed to form a sub-droplet. Scale
bar = 2 mm. The diameters of sol-state Geltrex were measured by brightfield microscopy and
compared to the diameters of circular hydrophilic sites (E). The asterisk indicates the single
xi
condition tested for which sub-droplets failed to form. Experiments were performed in triplicate
and error bars indicate ±1 S.D. ..................................................................................................... 75
Figure 4-4: Formation of microgels in varied geometries. Frames from movies depicting the
formation of star (A-D), triangle (E-H), and diamond-shaped microgels (I-L). Microgel density
can be increased by adjusting the pitch of patterned hydrophilic sites on the device top-plate (M).
A movie depicting the formation of an array of microgels in different shapes can be found in the
online supplementary information. Scale bars = 2 mm. ............................................................... 76
Figure 4-5: Off-chip microgel analysis. Agarose (5% w/v) hydrogels were formed on device
then allowed to crosslink at ambient temperature. Top-plates were removed to expose the
hydrogel allowing for imaging by environmental scanning electron microscopy. ....................... 78
Figure 4-6: Reagent exchange in hydrogel pillars. (A-D) Frames from a movie depicting a
droplet of fluorescein being electrostatically manipulated across a (transparent) microgel pillar
and subsequent diffusion of fluorescein into the microgel. Scale bars = 2 mm. (E) Experimental
(red dots), empirical fit (dashed blue line), and simulation (solid black line) of diffusion profiles
for fluorescein into Geltrex microgels. The inset shows a fluoresecent image (left) and heat-map
simulation of concentration (right) of fluorescein at the half-saturation point 1/2). (F) Apparent
diffusion (Da) coefficients measured for fluorescein, 4 kDa FITC-dextran, and 40 kDa FITC-
Dextran into Geltrex. Error bars indicate 1 S.D. ........................................................................... 80
Figure 4-7: Precision of reagent exchange across microgels. Frames from a movie (A-D)
depicting a droplet of reagent passed across a microgel. The diameters of dispensed droplets
were measured by brightfield microscopy and were compared to the diameters of circular
hydrophilic sites (E). The asterisk indicates the single condition tested for which sub-droplet
failed to form. Experiments were performed in triplicate and error bars indicate ±1 S.D. .......... 81
Figure 4-8: Pitch, multicomponent arrays, and composite microgels. 32-plex arrays of microgels
were formed from Geltrex supplemented with fluorescein for visualization (A). Combinatorial
arrays of Geltrex microgels were formed on device containing mixtures of red, yellow, and/or
green microspheres (B). Gradient bars indicate the percent compositions of each respective
microsphere, from 100% red at the left and 100% green at the right. Composite microgels
containing fluorescent microspheres were formed with inner agarose and outer Geltrex layers (C:
green filter, D: red filter, E: composite image). Scale bars = 2 mm. ............................................ 83
Figure 4-9: Higher order tissue formation and handling in microgel pillars. Spheroids are
generated five-plex in microgels on DMF (A). The inset highlights one microgel containing
spheroids. Confocal microscopy image stack of MDCK cells in a Geltrex pillar demonstrates cell
distribution throughout the z-axis (B). Images depicting MDCK spheroid formation in a gel pillar
over four days in culture (C-F). Cells were stained for actin with phalloidin (green) and nuclei
(blue). On day four lumen formation was observed. Hydrogels bearing 4-day MDCK spheroids
were subjected to media exchange five consecutive times either by automated pipetting or DMF.
Representative bright-field images of spheroids pre- and post-manipulation for both systems (G-
J). Scale bars = 20 µm. Graph of spheroid deformation for five trials evaluating 10 spheroids
xii
each using robotic pipetting into a multiwell plate or DMF (K). The deformation value of each
spheroid is indicated by a blue (DMF) or red (conventional reagent delivery) dot with black bars
representing the mean deformation of each trial. Bar graphs show overall deformation averages
across the five trials for each respective system. Error bars show 1 S.D. In three of the trials,
identified by asterisks, the conventional method dislodged the hydrogels from the well, resulting
in material failure. ......................................................................................................................... 85
Figure 4-10: Z-axis distribution of cells in microgels. Graph of distribution of cells in the z-axis
for the data in Figure 5B in the main text. The number of cells was enumerated at each image
plane (~6 µm thick). ...................................................................................................................... 86
Figure 4-11: Spheroid size. Comparison of spheroid size for three-dimensional MDCK culture in
microscale (DMF, white bars) and macroscale (96-well plate, gray bars). .................................. 86
Figure 5-1: Cardiac microgel formation on DMF. A sol-phase hydrogel droplet containing a
suspension of cardiac myocytes is drawn from a reservoir (A). The droplet is translated across
hydrophilic sites that are positioned (on the transparent top plate) above the optical windows on
the bottom plate (B-F). The white arrows indicate the direction of droplet movement. Scale bar =
4 mm. ............................................................................................................................................ 98
Figure 5-2: Cell distributions in microgels as a function of seeding density. Phase contrast
images of micorgels formed with cells seeded at low (A), moderate (B), or high (C) density. The
cells are visible through windows on the bottom plate of the device. At low and modest densities,
cells are homogenously dispersed through the hydrogel. At high density, cell aggregation is
apparent. Scale bar = 250 µm. .................................................................................................... 100
Figure 5-3: Brightfield microscopy of CM microtissues. Images of CMs in a microgel formed
on-demand at 4x (A), 10x (B) and 20x (C) magnification. Images of cells in the same microgel at
........................... 100
Figure 5-4: Confocal images of cardiac microgel on DMF. A cardiac microgel was fixed and
stained (blue for nuclei and red for actin) after five days of culture on DMF. Imaging was
performed at 5× (A), 10× (B), and 20× (C) magnification, with elongated branched cells evident
at each level. Scale bars = 100 µm, 50 µm, and 25 µm respectively. ......................................... 101
Figure 5-5: 3D cross-section of cardiac microgel. Cells were fixed and stained on DMF (blue
for nuclei, red for actin). Maximum combined exposure of an orthographic view in (A) z-x, (B)
x-y, and (C) y-z planes, demonstrates a well-connected network of CMs that are distributed in
3D space. Scale bar = 50 µm. ..................................................................................................... 101
Figure 5-6: Cardiomyocyte function in microgels. Cardiac microtissues were cultured for four
days on DMF and evaluated before and after exposure to epinephrine. The areas of twenty
individual cells were calculated and plotted with respect to time – representative traces (blue-
unstimulated; red – EPI-stimulated) for one cell are shown in (A). Insets microscopy image
demonstrates the contraction measurements performed on single cells. Black line indicates shape
pre-contraction, yellow line indicates shape post-contraction. These data were pooled and
evaluated for contraction frequency (B), contraction distance (C), and duration (D) for control
xiii
and epinephrine stimulated cells. Experiments performed in triplicate and error bars indicate ±1
S.D. ............................................................................................................................................. 103
Figure 5-7: Cardiomyocyte activity coefficient. After five days of culture on DMF, a beating
cardiac microgel was treated with a fixative solution and observed at 10× magnification for 30
minutes (A). A heat map was generated to represent cumulative cardiomyocyte activity
coefficients at 1 (B), 15 (C), and 30 min (D) after exposure to fixative. Heat maps are presented
as overlays on phase contrast microscopic images. Here blue indicates regions of low activity
and red indicates high activity. With increased exposure to fixative over time, reductions of
activity are observed (as shown through reduced red regions in the heat maps). ....................... 107
Figure 5-8: Cardiac activity coefficient of stimulated cells. CACs were calculated for control
and epinephrine-treated cardiac microgels, and the response (A) was significantly different by a
one-tailed t-test (p < 0.01). Heat maps representing cardiomyocyte activity generated for control
and epinephrine-stimulated microgels (B,C respectively), and brightfield images used for
analysis (D, E respectively). In these images red regions indicate higher activity, with blue
regions indicating lower activity. Scale bars = 50 µm. ............................................................... 108
xiv
List of Equations Eqn. 1.1 ........................................................................................................................................... 3
Eqn. 1.2 ........................................................................................................................................... 7
Eqn. 1.3 ......................................................................................................................................... 10
Eqn. 1.4 ......................................................................................................................................... 10
Eqn. 1.5 ......................................................................................................................................... 10
Eqn. 1.6 ......................................................................................................................................... 10
Eqn. 1.7 ......................................................................................................................................... 10
Eqn. 1.8 ......................................................................................................................................... 10
Eqn. 1.9 ......................................................................................................................................... 10
Eqn. 1.10 ....................................................................................................................................... 11
Eqn. 1.11 ....................................................................................................................................... 11
Eqn. 1.12 ....................................................................................................................................... 11
Eqn. 1.13 ....................................................................................................................................... 11
Eqn. 1.14 ....................................................................................................................................... 13
Eqn. 1.15 ....................................................................................................................................... 13
Eqn. 2.1 ......................................................................................................................................... 36
Eqn. 4.1 ......................................................................................................................................... 69
Eqn. 4.2 ......................................................................................................................................... 69
Eqn. 4.3 ......................................................................................................................................... 69
Eqn. 4.4 ......................................................................................................................................... 72
Eqn. 5.1 ......................................................................................................................................... 95
Eqn. 5.2 ......................................................................................................................................... 96
xv
List of Tables
Table 1-1: Comparison of well-plate and microfluidic methods used for cell culture .................. 5
Table 2-1: Elemental identification and quantification of species detected on patterned DMF
surfaces using XPS. Samples labels refer to Figure 2-2. .............................................................. 33
Table 3-1: Comparison of adherent cell culture using DMF between the new methods reported
here and those previously published. ............................................................................................ 61
xvi
Overview of chapters
This thesis is a comprehensive summary of projects related to the development of a digital
microfluidic platform for the culture and analysis of primary cells in two and three dimensional
cell culture.
Chapter one provides an overview of microfluidics in biological applications with an emphasis
on digital microfluidic platforms and theory. Recent developments within the field that have
enabled the implementation of DMF for cell culture and assay are reviewed.
Sections of this introduction were published in the article: Eydelnant, IA, and Wheeler, AR.
Digital microfluidic cell culture. BioTech International. (2012), 24, 20-22.
Chapter two describes the development of ‘virtual microwells’ (VMs) for reagent dispensing and
cell culture on DMF. The hydrophilic patterning technique developed here serves as the basis for
much of the work described in subsequent chapters within this thesis. With this method we were
able to reliably dispense volumes from 80 to 800 nL in air and oil, improve cell culture on DMF,
and perform the first example of reagent dispensing on a single plate DMF device. Further we
demonstrated the first example of passive dispensing in oil filled devices. A quantitative criterion
was also formulated to guide the design of such hydrophilic sites in future DMF systems. The
methods developed in this work are already finding applications in many DMF and non-DMF
applications.
This work resulted in the following publication: Eydelnant IA, Uddayasankar U, Li BB, Liao
MW, Wheeler AR. (2011) Virtual microwells for digital microfluidics. Lab on a Chip. 12, 750-
757.
Chapter three describes the first digital microfluidic method for the culture and analysis of
primary cells. Utilizing the virtual microwell method developed in chapter two, primary porcine
aortic endothelial, valve endothelial, and valve interstitial cells were cultured, fixed, and stained
xvii
in microlitre droplets using electrostatic manipulation of reagents. A monocyte adhesion assay
was performed completely on device supporting the implementation of DMF for cell culture and
assay. This study reinforces the applicability of DMF for the culture of sensitive cell types in
sub-microlitre droplets and looks toward integration of DMF with physiologically relevant
models of biology.
This work resulted in the following publication: Srigunapalan S*, Eydelnant IA*, Simmons C,
Wheeler AR. (2011) A digital microfluidic platform for primary cell culture and analysis. Lab on
a Chip. 12, 369-375. *Equal contribution
And the following proceedings: Eydelnant IA, Li BYB, Chang WC, Stanford W, Wheeler AR.
(2011) Upside-down digital microfluidic based embryonic stem cell culture. Proc. Micro. Tot.
Anal. Sys. (33-35)
Chapter four describes a novel technique for microscale hydrogel formation termed ‘microgels
on-demand’. This work builds on the virtual microwells presented in chapter two and the
determination of biocompatibility of DMF for cell culture in chapter three. This is the first
method capable of generating microscale thermally cross-linked hydrogels in a range of
geometries. Until now, this has only been possible with photo-crosslinking chemistries. Further,
we implement this system for three-dimensional cell culture, specifically to recapitulate a model
of kidney epithelialization. We demonstrate the utility of fully automated DMF operation for the
generation of high density microgel arrays, the formation of combinatorial microgels on-device,
and composite microgels consisting of multiple hydrogel constituents.
This work is the subject of a manuscript that has been submitted for publication: Eydelnant IA,
Li BB, Wheeler AR. (2013) Microgels on-demand (Submitted).
And the following proceedings: Eydelnant IA, Li BB, Wheeler AR. (2012) Virtual microwells
for three-dimensional cell culture on a digital microfluidic platform. Proceedings of the IEEE
25th International Conference on Micro Electro Mechanical Systems (MEMS). 898-901.
xviii
Chapter five presents the implementation of microgels on demand, developed in chapter 4, for
functional drug screening of cardiac microgels. Cardiomyocytes were seeded in sub-microlitre
microgels and cultured on device for up to one week. After three days of culture spontaneous
contractions were observed within the construct and coordinated contractions became evident at
five days after seeding. We perform single cell analysis to determine contraction frequency,
magnitude, and duration with the finding that these constructs are functional and respond to drug
treatments in accordance with previous studies in the literature. To improve throughput of
analysis, an image processing method was developed to analyze broader construct activity
resulting in a metric referred to as the cardiomyocyte activity coefficient (CAC). We present
preliminary evidence that this metric can be correlated to the contraction frequency.
This work is the subject of a manuscript that is currently in preparation: Eydelnant IA,
Thavandiran N, Radisic M, Wheeler AR. (2013) Cardiac microgels (In preparation).
And the following proceedings: Eydelnant IA, Li BB, Wheeler AR. (2012). Digital
microfluidics for on-demand 3D microgel formation and functional myocardial tissue assays.
(2012). Proc. Micro. Tot. Anal. Sys. (112-114).
xix
Overview of author contributions
Chapter two describes the implementation of virtual microwells for DMF. This project was
initiated, designed, and lead by me. Fellow graduate student Uvaraj Uddayasankar provided
helpful discussions that initiated this work. I developed the Teflon liftoff protocol, fabricated
devices, performed dispensing and cell experiments, and developed the non-dimensional number
reported. Undergraduate researchers Bingyu ‘Betty’ Li and Donna Liao reproduced the
dispensing experiments for all conditions, assisted with device fabrication, and prepared reagents
for the experiments.
Chapter three demonstrates culture and analysis of primary cells for the first time on DMF.
These projects resulted in a co-first authored paper shared between graduate student Suthan
Srigunapalan (SS). I designed and fabricated devices then tested reagent compatibility. For the
primary cell work, cells were cultured by SS. Together SS and I performed all experiments for
cell seeding, fixing, staining, and functional analysis. SS imaged the cells and did the analysis for
the adhesion assay.
Chapter four describes a method we term `microgels on-demand`. For this work I selected
appropriate hydrogel materials, tested compatibility with surfactants, assessed the capability of
using virtual microwells to form microgel structures, designed geometries, performed electron
scanning microscopy imaging, confocal imaging, cell seeding, and analaysis. Bingyu `Betty` Li
assisted with kidney epithelialization experiments and was instrumental in the selection of
fixative for this work. She also performed replicate studies on volume dispensing in virtual
microwells.
Chapter five describes the implementation of microgels on demand for cardiac microgels. For
this work I initiated the collaboration with the Radisic laboratory. I seeded cells on device,
cultured them, ran functional assays, and analysis. Both the single cell analysis and
xx
cardiomyocyte activity coefficient were developed by myself. Nimalan Thavandiran (then
graduate student) provided helpful discussions, neonatal cardiomyocytes and media for
experiments.
1
Chapter 1. Introduction
1.1 Historical perspectives on the miniaturization of biology
Prescient scientists in the mid-20th
century set the stage for a future where accessing biology at
the micron scale would become commonplace. In 1944 Schrodinger asked the famous question
‘What is life?’ at his lecture at the Dublin Institute for Advanced Studies, peering into the nano-
and micron scale components of the cell from a purely physical and chemical perspective.1
Feynman inspired with his visionary and speculative 1960 lecture ‘There’s Plenty of Room at the
Bottom’ introducing the untapped field of miniaturization for electronics and biology.2
Concurrently, significant thought was being given to Brownian motion and diffusion by
Einstein3 and Berg
4 in biological systems while Taylor
5 had already begun examining liquid
flows through micron scale channels. The ability to implement these concepts in biological
investigation through microfluidic systems gained significant ground with the development of
miniaturized chromatographics5,6
and ink jet technologies.7 These methods remained inaccessible
to the majority of scientists due to requisite specialized equipment, borrowed from the semi-
conductor industry, which permitted channels to be etched in glass and silicon. With the
development of soft-lithography and subsequent enabling of replica moulding, in cheap
elastomers such as polydimethylsiloxane (PDMS), pioneered by the Whitesides group in 1998,8
the field of microfluidics experienced an immediate explosion of accessibility, spawning
microfluidic applications in nearly every imaginable scientific discipline.
2
Parallel to early works in microfluidics and movement toward miniaturization, bench-top biology
was undergoing rapid change, initiated by Hungarian physician Takatsy in 1955 when he
proposed a move away from traditional glass tubes and dishes by developing the 96-well plate to
increase laboratory throughput.9 The microtitre plate provided a platform for early bench-top
miniaturization which today is driven by the high-throughput demands of contemporary drug
screening and systems approaches to biology. High-throughput screening (HTS) emerged in the
1980s and 1990s as a new standard for drug discovery within the pharmaceutical industry, where
microtitre formats increased to 384 and 1536 wells, thereby maximizing experimental densities
within equivalent footprints and further decreasing reagent consumption. Since then, the
technologies that emerged from this field have spilled over extensively into broad fields of study
including genomics, protein crystallization, materials science, and environmental toxicity
sampling.
Today HTS methods have reached a standstill – throughput with current automated liquid
handling robotics (ALHR) and microtitre plates has plateaued and more attention is now being
paid toward higher content and quality of data. The adoption of high-throughput cell-based
biological screens as standard practice within academic laboratories is furthering the need for
accessible high-throughput systems. Robotic liquid handling – though useful for experimental
automation – is often inaccessible because of high capital costs (e.g., robotics), large volumes of
reagent consumption (e.g., drugs, media, cells) and high turnover of consumables (e.g.,
microwell plates, pipette tips). Data generated in standard HTS assays does not typically include
sufficient content to instruct lead pursuit without further experiments. These obstacles have
motivated the development of miniaturized platforms with the capability to manipulate micro-
3
scale samples. Microfluidic technologies, where liquid manipulation is implemented in micron-
scale confined volumes, potentiate the development of the tools needed for ultra-low volume
reagent handling. With an abundance of research groups developing biologically oriented
microfluidic technologies, the first generation of laboratory protocols implemented in devices
with the footprint of a credit card has been realized enabling improved control over experimental
conditions and higher density experimental footprints.
1.2 Microfluidics Paradigms
Multiple paradigms of microfluidics, including continuous channel microfluidics, droplet-in-
channel microfluidics, and digital microfluidics, have emerged, each with its own advantages
and disadvantages (Table 1-1).10
Continuous channel microfluidics, where reagent flows are
confined within micron scale channels, allow for well controlled serial device operations. Flow
properties of these systems are readily described by the dimensionless Reynold`s number:
vL
ForcesViscous
ForcesInertialRe Eqn. 1.1
where ρ is the fluid density (kg/m3), v is the mean velocity (m/s), L is the characteristic length of
the system, and υ is the kinematic viscosity (m2/s). Flow in channel microfluidic systems is
characterized by viscous dominated flow and low Reynold’s number regimes, where multiple
flow streams can be constrained within a single channel without mixing. These features have
been exploited in biological assays for the formation of well-defined diffusion gradients across
cell monolayers, single-cell capture and analysis (e.g., PCR, fluorescence microscopy), modeling
4
microvasculature, and other studies requiring precise control of the chemical (e.g., growth
factors, cytokines) and physical (e.g., shear stress, flow rate) cellular microenvironment. With
the incorporation of on-device valves, these devices have been scaled into compartmentalized
high-throughput systems. However, because of the inherently complex fabrication protocols
required to form such systems, reliance on external equipment (e.g., pumps, valve manifolds),
and networks of connecting tubing required for reagent transfer, biologists have been slow to
adopt such systems into their routine work-flow.
In droplet-in-channel microfluidics, pico- and nano-litre droplets are generated in a two-phase
flow in microchannels. This allows for the formation of thousands of independent droplets per
second that can be merged, sorted, and reacted. With simple fabrication and high-throughput
operation (10-100 kHz), these devices are well positioned for screening in biological
experiments. Biocompatible surfactants stabilize the emulsions, allowing for encapsulation of
live cells for suspension culture or hydrogel materials for adherent culture. The use of fluorinated
oils in the continuous phase permits sufficient oxygen transfer to maintain cell viability for
multiple days. The throughput of droplet microfluidic systems is unparalleled; however, such
systems are not well suited to multi-step long-term applications involving cells because of the
challenges in addressing individual droplets (for media exchange, reagent addition, staining). In
addition, both continuous and droplet-in-channel methods typically require dedicated specialized
microscopy analysis techniques; this limits their flexibility for integration with standard
analytical laboratory equipment. Over the past decade, a third paradigm, digital microfluidics,
has emerged as a potential solution to these limitations.
5
Figure 1-1: Microfluidic paradigms. (A) Continuous flow channel microfluidic systems can
exploit the laminar flow properties of micron scale confined flows to generate interesting flow
patterns such as the gradient generator depicted here.11
(B) Two phase droplet-in-channel
systems are capable of high-throughput generation of individual droplet compartments. Here two
strategies for droplet formation are depicted within a T-junction (top) and by flow focusing
(bottom).12
(C) Digital microfluidics allows for the manipulation of discrete droplets across
arrays of electrodes.
Table 1-1: Comparison of well-plate and microfluidic methods used for cell culture
Criteria Well-plate Microchannel
Droplets in
Channels
Digital
Microflfuidics
Cost $1-2 per plate <$10 / PDMS
device
<$10 / PDMS
device $10-50 / device
Reagent volumes µL-mL nL- µL pL-nL nL-mL
Throughput High Moderate High Moderate
Automation Yes Yes Yes Yes
No. of scientists >100,000 >1,000 <500 <20
6
1.3 Digital Microfluidics
Digital microfluidics (or DMF) is a liquid handling technology that permits the independent
electrostatic manipulation of individual pico- to micro-litre size droplets across arrays of
electrodes.13
The most common DMF format features droplets sandwiched between two plates,
and typical operations include droplet dispensing, splitting, merging, and mixing.14
The bottom
plate is patterned with electrodes buried beneath a hydrophobic insulator. The top plate
comprises a contiguous electrode coated with a hydrophobic layer. The hydrophobic coatings are
critical to reduce friction forces that can impede droplet movement. When voltages are applied to
a driving electrode on the bottom plate relative to a counter-electrode on the top plate, a
separation of charge occurs across the insulator acting on ions or dipoles within the droplet. The
resulting electrostatic force drives droplet translation. These devices are commonly operated in
air, though filler media such as silicone oil can be used to reduce voltages needed to move
droplets. Device fabrication follows basic photolithography and metal etching protocols on a
range of substrates including glass, silicone, flexible polyimide films, compact discs, and printed
circuit boards (PCBs). Currently a range of techniques for rapid prototyping and mass-
production of devices are being investigated, with an emphasis on multi-layer PCBs15
and high
density thin film transistor (TFT) arrays.16
1.4 Digital microfluidic theory
The physics of DMF droplet actuation has not been completely elucidated, with multiple
mechanisms described in the literature. The forces controlling droplet motion can be divided into
driving and resistive forces. Historical understanding of DMF associated driving forces often
begin with Pellat’s17
original observation of the generic phenomenon related to what we now call
7
DMF over a century ago where he demonstrated that an external voltage can cause an insulating
dielectric liquid to raise upward against gravity when confined between vertical parallel
electrodes. In some instances this phenomenon is described as electrowetting or electrowetting
on dielectric (in the case where electrodes are coated with dielectric material, this is often
described with the acronym, EWOD)18–26
. This follows from the observation that upon
application of a potential across the electrode, droplets of fluid with high surface tension (e.g.,
water) tend to wet the surface (i.e., experience a significant decrease in contact angle) that can be
described within a thermodynamic context by the Young-Lipmann relation27
:
d
V
2coscos
2
r00 Eqn. 1.2
where θ and θo are the static contact angles with and without applied voltage respective, εo is the
permittivity of free space, εr is the relative permittivity of the dielectric, V is the applied voltage,
and γ is the air-liquid surface tension. Here droplet movement is assumed to occur due to
capillary pressure that results from asymmetric contact angles across the droplet. However,
EWOD theory fails to describe two key observations: (1) low surface tension liquids that have no
apparent change in contact angle are movable on such systems28
(and indeed, have been used by
8
Figure 1-2 Digital microfluidic device geometry. (A) A DMF device consisting of 144
independent actuation electrodes. (B) Devices are composed of two parallel plates. The bottom
plate is patterned with an array of electrodes and coated by a hybdrophobic insulator. The top
plate bears the counter electrode and is covered with a hydrophobic coating. (C) Schematic of
droplet translation principles. Separation of charge occurs across the dielectric layer acting on
charges or dipoles in the droplet thereby driving translation.
9
the Wheeler group for many applications), and (2) θE cannot be reduced to zero upon application
of infinitely high voltage (as suggested by the Young-Lippman equation), and in fact saturates at
some intermediate value determined by the particular liquid and dielectric solid being used.
Kang29
proposes that this contact saturation arises from the free-energy contribution of the
electrical double layer in the liquid phase, which would be incapable of driving the observed
changes in contact angle. Buehrle and Mugele30
compute the liquid profile near the contact line
by enforcing electromechanical equilibrium at the interface while minimizing energy. Their
result further contradicts EWOD theory as there exists only an ‘apparent’ change in contact angle
since at a given distance from the dielectric surface the contact angle asymptotically approaches
θo. Jones31
furthers this argument through an electromechanical approach demonstrating contact
angle change as a consequence of a strong electric field near the three-phase contact line. In his
work he proposes a Gedanken experiment modelled after the height of rise experiment described
above, however in his example, he restrains the meniscus shape (and effectively contact angle)
by floating a thin membrane at the top of the column while retaining the height of rise with the
application of a voltage to the electrodes.
Given the uncertainties surrounding the EWOD mechanism, we and others propose that a more
complete illustration of droplet actuation comes from electrodynamic analysis29,31–33
. Under this
treatment droplet movement is explained by the action of electric forces on free charges in the
droplet meniscus of conductive liquids or dipoles inside the droplet of dielectric liquids. As such
this model overcomes the limitations of EWOD demonstrating droplet movement without
requiring wetting and rationalizes contact angle saturation as a balance between electrical and
10
surface tension forces. The theoretical content of electrodynamics is contained within the
Maxwell equations and Lorentz’s Force Law34
:
Gauss’s Law: o
E 1 Eqn. 1.3
Unnamed: 0 B Eqn. 1.4
Faraday’s Law tBE
Eqn. 1.5
Ampere’s Law: tEJB ooo Eqn. 1.6
Lorentz’s Law: )( BEqF Eqn. 1.7
The total electromagnetic force on the charges in a volume V can be determined by Lorentz’s
force law that describes the effects of electromagnetic fields on charges:
VV
BJEBEF )()( Eqn. 1.8
Rewriting this equation as a force per unit volume and substituting Maxwell’s equations (Eqn.
1.3) and (Eqn. 1.6) to eliminate ρ and J, leaving only the field terms we find:
Bt
EBEEf oo
)1()(
Eqn. 1.9
where the first term accounts for Coulombic forces and the second term magnetic forces.
Further manipulation brings this expression to the form:
11
)()1(2
1)()(1)()( 22 BEt
BEBBBBEEEEf oo
oo
o
Eqn. 1.10
Neglecting the magnetic field components, and reverting to the volumetric integral we find:
V
dEEEEEF )(5.0)()( 2 Eqn. 1.11
Simplifying the above volumetric integral to a surface integral by the introduction of the
Maxwell stress tensor facilitates solution described previously for DMF droplet
manipulation29,32,35
:
S
ndsTF Eqn. 1.12
)2
1( 2EEET ijjiij Eqn. 1.13
Here Tij is the stress tensor, where i and j refer to pairs of x, y, and z axes, δij is the Kronecker
delta function, and E is the electric field surrounding the droplet. In this electrostatic
interpretation, Tij is the stress (i.e., force per unit area) of the ith direction on an element of
surface oriented in the jth direction. Here, diagonal elements represent pressures and off-
diagonal elements are shears. The Wheeler group recently used this electrodynamic model to
estimate the driving forces on droplets of water in eight different DMF device geometries, and
the results correlated well with experimental observations32
Equivalent results can be determined through a circuit representation of the DMF device (Figure
1-3A) within an electromechanical framework.31
In this analysis the amount of energy,
12
Figure 1-3: Theoretical framework of DMF. (A) Equivalent circuit analysis of DMF driving
force mechanisms. (B) Force estimation for a two-plate DMF device operating on PBS, DI
water, toluene and methanol. Forces are based on a 1 mm2 electrode size, 6 μm of Parylene-C,
235 nm of Teflon-AF, a gap size of 150 m and an applied voltage of 100 VRMS for a range of
frequencies (100 Hz to 1 MHz). (Adapted from Choi et al.36
)
13
E, stored within the system can be calculated as a function of the applied voltage frequency and
droplet position along the direction of translation. This method assumes that the cross-sectional
area of the drop is readily approximated by a square with sides length L:
i i d
fjVxL
d
fjVx
LxfE
i
filleri,fillerr.i,
i
liquidi,liquidr.i, )2()2(
2),(
Eqn. 1.14
where εr,i,liquid, Vi,liquid, and εr,i,filler, Vi,filler are the relative permittivity and voltage drop for the
liquid and filler fluid portion of the electrode respectively, and di is the thickness of layer i
(corresponding to the dielectric, hydrophobic, liquid, or filler layers). The associated change in
energy as x progresses from 0 to L is equivalent to the work done on the system. Differentiation
of Eqn. 1.12 with respect to x yields the driving force as a function of frequency:
i i d
fjV
d
fjVL
x
xfEfF
i
filleri,fillerr.i,0
i
liquidi,liquidr.i,0 )2()2(
2
),(
Eqn. 1.15
Typical DMF droplet manipulation employs AC frequencies on the order of kHz. At these
frequencies, the majority of the voltage drop occurs across the dielectric layer. Estimates of
forces by Eqn. 1.15 indicate magnitudes in the range of µN (Figure 1-3B). Above a critical
frequency, fc, the electric field shifts from the dielectric to the droplet resulting in a liquid-
dielectrophoretic regime (Figure 1-3B). For these reasons liquids with low conductivity and
permittivity require prohibitively high voltages, which can be remedied through mixing with
reagents with more amenable properties.
14
Droplet movement on DMF must be considered as a balance of electrostatic forces (described
above) and resistive forces. The major resistive forces are friction between the droplet and the
hydrophobic surface, contact line pinning, and viscous drag.22,37–39
The first are determined by
the nano- and micro- scale roughness of the hydrophobic surface, the second is a molecular
adhesion effect occurring at the three-phase (surface, liquid, air) contact line of a droplet that
results in ‘sticking’ thereby impeding droplet movement, while the third are associated directly
with displacement of filler fluid during droplet translation. Contact line pinning is implicated in
DMF observations of contact angle hysteresis and neglecting its effects results in inaccurate
simulations of droplet motion. Modeling this effect is challenging as simulations require
nanometer length scales and nanosecond time scales. For a droplet to move on DMF the electric
field generated across the dielectric must be strong enough to overcome these resistive forces and
can require applied voltages from 10 Vpp to 1000 Vpp depending on dielectric coatings and the
liquid being manipulated.
1.4 DMF compatibility with two-dimensional cell culture
Digital microfluidics is a useful platform for the miniaturization of cell culture and assays. With
the capacity to support both adherent and suspension cell culture, several research groups have
demonstrated long-term culture and passaging on-device. Suspension cell culture is well suited
to DMF as droplets are manipulated across hydrophobic surfaces that are resistant to adhesion.
DMF based viability assays have been performed with comparable results relative to 96-well
plate assays and 100-fold reductions in reagent volumes.40
Droplet mixing by translation allows
for cell growth, and droplet splitting and merging are useful for dilution and passaging. For
adherent cell culture, multiple strategies have been developed for surface functionalization
15
including protein deposition, plasma etching, and a fluorocarbon liftoff technique described in
Chapter 2 of this thesis. Each has been demonstrated for the culture of immortalized cell lines,
while the latter has been successfully implemented in the culture of more sensitive cell types
including primary cells as presented in Chapter 3 (Figure 1-4A & B).41
The introduction of
hydrophilic sites to the hydrophobic coating resulted in the discovery of a novel fluidic
phenomenon termed passive dispensing. As droplets are translated across hydrophilic sites, a
portion of the droplet is pinned to the site and a sub-droplet is formed. Chapter 3 examines
passive dispensing for controlling droplet volumes in cell seeding and subsequent media and
reagent exchange.42
The successful culture of multiple cell types suggests minimal electromagnetic effects on DMF
cultured cells. This has been supported by computational modeling of potentials across the
device, which demonstrated that the majority of the voltage drop occurs in the dielectric layer
and a minimal potential is experienced by the droplet. Recent microarray analysis of cell lines
actuated in DMF-like devices indicate minimal differences in transcriptional profiles at normal
operating conditions.43
Biofouling of device surfaces remains the most significant challenge in
long-term device operation. Protein adsorption to the hydrophobic surfaces results in droplet
pinning thereby restricting droplet translation. The addition of low concentrations of
biocompatible surfactants (e.g., Pluronics) improves device function, but does not allow for
indefinite device operation. Novel surfactants and fouling resistant surface coatings are being
investigated to address these issues.
16
Figure 1-4: Cell culture on DMF. (A) Virtual microwells: Droplets containing cells suspended
in media are translated across patterned hydrophilic sites where a subdroplet is generated by
surface interaction forces. The device is then flipped to allow for cells to settle and adhere to the
hydrophilic site. Here, cells stained with calcein-AM are imaged by stereomicroscopy
immediately after seeding on device. (B) Primary cells: Aortic interstitial cells isolated from pig
hearts cultured for 48 hours on device were then fixed and stained with Hoescht (blue – nuclei)
and Phalloidin (green – actin). Imaging was performed by epifluorescence microscopy. Scale bar
= 200 µm. (C) Multiplexing: Automation combined with multiplexed devices allows for rapid
screening of multiple conditions. Here 16 conditions are screened simultaneously.
17
1.5 Microfluidics for three-dimensional cell culture
Hydrogel based three-dimensional (3D) cell culture is rapidly becoming a fundamentally
important tool in biological research.44
Hydrogel materials can be derived from inert or animal
sources providing for customizable microenvironments to elucidate or direct cell function and
behaviour. Bissell and coworkers45
were the first to demonstrate that in certain models of
epithelialization, cells traditionally cultured in monolayers would form polarized hollow
spheroid structures when dispersed in collagen hydrogel matrices. Microarray analyses of cells
isolated from these spheroid structures confirm that they are transcriptionally more
representative of in vivo conditions than two-dimensional (2D) culture.46
This transition from
monolayer to 3D cell culture bridges the gap between in vitro and in vivo studies. These systems
have allowed for significant strides in cell biology through the reestablishment of critical
microenvironmental factors, particularly cell-cell and cell-ECM interactions, in a range of work
including tumor biology, cell adhesion, migration, and epithelial morphogenesis. To improve
data reliability from drug screening and avoid the pursuit of in vivo studies on false-positive
targets, 3D cell culture presents a potentially ideal system to balance cost and reliability.
Unfortunately, hydrogel culture methods remain under-utilized in part because of high reagent
costs and challenges in the manipulation and handling of delicate hydrogel materials.
A number of strategies relying on microchannels have been proposed to address the challenges
of working with 3D cell structures in hydrogels.47
Microfluidics provides the ability to
manipulate sub-microlitre volumes of liquid thereby reducing reagent consumption.
Furthermore, the associated low Reynolds number flow through microfluidic channels allows for
gentle hydrogel handling and reduces subsequent damage to gels during reagent exchange.
18
Microfluidic devices used for 3D cell culture and handling have ranged in complexity, from
systems with integrated valves48
and off-chip pumps49
to passive perfusion systems50,51
exploiting gravity and surface tension forces to drive reagent exchange. While these systems are
undoubtedly useful, the microchannel modality remains challenged for this application by (1)
lack of flexibility in hydrogel geometry and size, (2) channel clogging and limited perfusion
through hydrogels, and (3) tubing dead-volumes.51
Limitations in hydrogel geometry for 3D cell culture in microchannels are inherent to device
design. In continuous flow systems, the channel is completely filled with the hydrogel material,
resulting in gel structures conformal to the channel geometry. In two-phase flow systems the
majority of demonstrations have been restricted to monodisperse spherical or rod-shaped solids
produced by either photo-initiation or thermal cross-linking at a T-junction or by flow-
focusing.52,53
Recent demonstrations by the Doyle group54 have exploited stop-flow lithography
in the high-throughput formation of microgels in a range on geometries. Though these systems
are proving particularly useful in the conception of drug delivery systems, they are limited to UV
initiated cross-linking which can be detrimental to cells and excludes the many temperature
sensitive hydrogel systems (e.g., collagen, Matrigel) that are commonly used in 3D cell culture
studies. Further these systems typically produce hydrogel geometries with length scales of <100
µm (and volumes of < 1 nL), whereas typical 3D cell construct sizes range from 50-1000 µm in
size. DMF liquid handling is an emerging alternative to the paradigm of enclosed
microchannels.14
Recently, in the unique geometries of DMF have been exploited for handling
and addressing of three-dimensional solids such as paper discs for blood screening,55,56
polymer
monoliths for sample extractions,57
and agarose discs for scaffolding applications.58,59
In
19
chapters 4 and 5 of this thesis DMF liquid handling is combined with the virtual microwells
described in chapter 2 developing a novel platform for ‘microgels on-demand’, that allows for
sub-microlitre 3D cell culture.
1.6 Assays and integration
The real benefits of DMF are realized in the automation of multi-step assays and integration of
devices within existing laboratory analytical infrastructure. Live-cell apoptosis assays with cell
seeding followed by stimulation, washing, and staining steps were recently performed on DMF.5
These assays were performed in microliter droplet volumes and did not suffer from cell loss
during reagent exchange, a common problem for such assays when performed in microwell
plates. Multiplexing of these assays on a single device provides for the ability to quickly and
efficiently screen a range of conditions (Figure 1-4C). The use of fluorescent apoptosis markers
allows for direct device integration with a fluorescent plate reader of the type that is common in
research laboratories. In chapter 3 of this thesis, primary cells were cultured for multiple days on
DMF.3 These cells were subsequently stimulated with cytokines and their functional responses
tested in a monocyte adhesion assay. Further, on-device fixing and staining of these cells
followed by epifluorescent microscopy and imaging in high-content screening equipment,
demonstrated device compatibility with microscopy for the acquisition of high-quality images.
A second area where the small volume reagent handling capabilities of DMF is currently being
exploited is mass spectrometry (MS). Upstream sample preparation on DMF has embodied
liquid/liquid extractions for DNA clean up through aqueous immiscible droplets60
and tissue
processing by extraction in isooctane.61
Liquid/solid sample prep strategies on device have
20
included hydrogel immobilized enzymes for proteolytic digestion,59
and solid-phase extraction in
porous polymer monoliths. These methods have proven critical in separating analytes from salts
and contaminants that can cause ion suppression thereby hindering analysis. Further, methods
for direct sample introduction to MS have explored devices patterned on flexible substrates as
foldable emitters62
and embedding glass capillaries between the device top and bottom plates to
facilitate reagent transport into the MS.56
DMF integration with analytical methods remains an
active area of research with other groups coupling these systems with emerging applications
including surface plasmon resonance (SPR)63
and miniaturized tunable droplet lasing systems64
.
1.7 Future of DMF
One of the visions of digital microfluidics for high throughput biological screening is the
eventual development of low-cost ‘smart’ microwell plates to complement or replace automated
liquid handling robotics. Ideally these would function as self-contained cell culture and analysis
units capable of multiplexed cell based assays. This will require the combination of robust low-
cost devices, novel surfaces that are resistant to biofouling, and automation hardware to drive
droplet translation. In the development of this technology groups are finding capabilities of DMF
systems for applications that are not possible utilizing other technologies – including temporal
studies of cellular responses to cytokines and thermally crosslinked microgel formations in
customizable geometries (described in chapters 4 and 5). With an increasing number of research
groups actively pursuing the development of DMF, there remains great potential for this platform
technology to function as novel foundation for HTS and provide essential tools in gaining greater
insights on fundamental biological processes.
21
Chapter 2. Virtual microwells for digital microfluidic reagent
dispensing and cell culture
Summary
Digital microfluidic (DMF) liquid handling includes active (electrostatic) and passive (surface
tension) mechanisms for reagent dispensing. Here we implement a simple and straightforward
Teflon-AF liftoff protocol for patterning hydrophilic sites on a two-plate device for precise
passive dispensing of reagents forming virtual microwells – an analogy to the wells found on a
microtitre plate. We demonstrate here that devices formed using these methods are capable of
reproducible dispensing of volumes ranging from ~80 to ~800 nL, with CVs of 0.7% to 13.8%
CV. We demonstrate that passive dispensing is compatible with DMF operation in both air and
oil, and provides for improved control of dispensed nano- and micro- litre volumes when
compared to active electrostatic dispensing. Further, the technique is advantageous for cell
culture and we report the first example of reagent dispensing on a single-plate DMF device. This
method has proven useful for DMF based cell culture and analysis. The technology described in
this chapter presents a platform upon which the remainder of this thesis is built upon.
22
2.1 Introduction
Miniaturization of laboratory procedures for lab-on-a-chip technologies requires on-device
methods that are analogous to standard pipette-based reagent dispensing. This has been realized
in microchannel-based systems through on/off device pumps,65
in-line valves,48
electrokinetic
flow,65
and capillary action,66
providing control of femto- to micro- litre volumes. In digital
microfluidics (DMF), a technique in which droplets are manipulated across an array of insulated
electrodes, active14
(electrostatic) and passive67
(surface tension) dispensing modes have been
demonstrated. Here, we introduce an improvement to passive dispensing, with an emphasis on
robustness and reproducibility, and applications in cell culture and analysis.
DMF devices are operated in either single or two-plate geometries. The single plate geometry
typically consists of actuation electrodes with co-planar ground electrodes60
or a suspended
grounded catena.68
In the two-plate format (Figure 2-1A,B), a bottom plate is patterned with
electrodes coated with a dielectric and hydrophobic material, and a top plate comprises a
conductive layer coated with a hydrophobic material. A widely used function of two-plate digital
microfluidics is active dispensing of droplets from reservoirs.14
As shown in Figure 2-1C (panels
i and ii), active dispensing is achieved by actuating a series of electrodes to stretch, neck, and
pinch a droplet off from a reservoir.14
Active dispensing highlights a particularly useful property
of digital microfluidics: reagents and samples can be dispensed reliably and precisely on-
demand. However, as demonstrated in this chapter, inconsistencies arise when dispensing
reagents with varying viscosities.
23
Figure 2-1: Two-plate digital microfluidic (DMF) device design and assembly. (A) Exploded
view of a device, comprising a bottom plate with patterned electrodes and a top plate bearing
patterned hydrophilic sites. (B) Side-view, not to scale. (C) Schematic depicting two reagent-
dispensing mechanisms on DMF. Active dispensing (i & ii) involves electrostatic stretching of a
reagent from the reservoir followed by splitting. Passive dispensing (iii & iv) occurs
spontaneously as a source droplet is translated across the hydrophilic site. The inset is a three-
dimensional depiction of a virtual microwell, VM (i.e., a droplet formed by passive dispensing).
VM volume is dictated by the diameter of the hydrophilic site (d) and the distance between top
and bottom plates (h).
24
An alternative digital microfluidic function called passive dispensing was recently described by
Barbulovic-Nad et al,67
building on similar work by Chen et al.69
Passive dispensing is
implemented using a DMF device surface that is primarily hydrophobic but patterned with
hydrophilic regions. When a source droplet is translated across a hydrophilic site, surface tension
effects result in spontaneous formation of a sub-droplet on the patch (Figure 2-1C, panels iii and
iv). As described previously,41,70–72
passive dispensing is particularly useful for adherent
mammalian cell culture, allowing for cell seeding onto dry hydrophilic sites, as well as for
subsequent media and reagent exchange on droplet-bearing sites. We introduce here a new term
for the cylinder-shaped droplet formed by passive dispensing: a virtual microwell (VM). The
term VM is an analogy to the wells found on a microtitre plate. The "wells" described here are
virtual as they are not confined on the sides like traditional wells, but are defined by the surface
properties of the top and bottom plate. A similar strategy has been described previously73
for
non-microfluidic applications, however this is the first time this concept is being applied within
the context of DMF.
In initial work describing passive dispensing for cell culture,67
hydrophilic patches were formed
by adsorbing extracellular matrix proteins onto Teflon-AF-coated DMF device bottom plate
surfaces. Adaptation of this method for other applications realized several challenges, including:
(1) inconsistent reagent dispensing both initially and during subsequent droplet passages, (2)
protein dissolution and subsequent loss of hydrophilic pad integrity, and (3) difficulty
functionalizing the electrode-bearing surface. Motivated by these challenges, we sought to
develop a simple fabrication protocol for patterning hydrophilic sites directly on device surfaces.
Here we report a new method for forming hydrophilic patches relying on a Teflon lift-off
25
procedure. The method is straightforward and fast, allowing for rapid generation of an array of
individually addressable virtual microwells by passive dispensing. We demonstrate that this
method can be used for two-plate DMF operation in air or oil to (1) reproducibly and precisely
dispense reagents independent of viscosity based solely on device design parameters, (2)
maintain constant droplet volume after subsequent reagent exchanges, and (3) improve cell
seeding when compared with previously published methods. Further we show the first example
of reagent dispensing on a single-plate device. We propose that these new methods will be useful
for a wide range of applications -- particularly those involving adherent cell culture and analysis.
2.2 Methods and Materials
2.2.1 Reagents
Unless stated otherwise, general-use chemicals were from Sigma Aldrich (Oakville, ON,
Canada) or Fisher Scientific Canada (Ottawa, ON, Canada), fluorescent dyes and cell media
components were from Invitrogen/Life Technologies (Burlington, ON, Canada), and
photolithography reagents were from Rohm and Haas (Marlborough, MA). Deionized (DI) water
had a resistivity of 18 MΩ·cm at 25°C.
2.2.2 Two-plate DMF bottom-plate fabrication
Digital microfluidic devices were fabricated in the University of Toronto Emerging
Communications Technology Institute (ECTI) cleanroom facility, using transparent photomasks
printed at 20,000 DPI (Pacific Arts and Designs Inc., Markham, Ontario). Two-plate DMF
device bottom-plates bearing patterned chromium electrodes were formed by photolithography
and etching of commercially available chromium and positive photoresist coated glass slides
(Telic, Valencia, CA). Briefly, substrates were exposed to UV through a mask (8 s, 29.8
26
mW/cm2) and then developed in MF-321 (~2 min). Chromium was etched in CR-4 (~5 min, OM
Group, Cleveland, Ohio), and then substrates were washed with DI water, dried under a stream
of nitrogen. Substrates were then immersed in AZ 300T (3 min) to remove photoresist and then
washed in DI and dried under a stream of nitrogen. This was followed by cleaning in Piranha
solution (10 s, 1:1 conc. sulfuric acid: 30% hydrogen peroxide). Substrates were rinsed in DI
water, then dried under a stream of N2, before dehydrating on a hot plate (165 ºC, 10 min). As
shown in Figure 1A, the bottom-plate design featured an array of 116 actuation electrodes (2.2
2.2 mm ea.) connected to 10 reservoir electrodes (4 4 mm ea.), with inter-electrode gaps of 30-
80 µm. The actuation electrodes were roughly square with 140 µm (peak to peak) sinusoidal
interdigitations. In some experiments, the design also included an array of five 1 mm diameter
optical windows (i.e., circular regions free from chromium) with 9 mm between each window.
Each window straddled the interface between two actuation electrodes. After patterning, the
substrates were immersed for 30 minutes in silanization solution: 3-(Trimethoxysilyl)propyl
methacrylate (Specialty Coating Systems, Indianapolis, IN), diluted to 1% (vol./vol) in 1:1 DI
water:isopropanol. Substrates were air-dried for 30 minutes then washed with isopropanol (IPA)
and dried under a stream of nitrogen. Substrates were then coated with 8 µm of Parylene-C
(Specialty Coating Systems) and 200 nm of Teflon-AF 1600 (DuPont, Wilmington, DE).
Parylene-C was applied using a vapor deposition instrument (Specialty Coating Systems), and
Teflon-AF was spin-coated (1% wt/wt in Fluorinert FC-40, 3000 rpm, 60 s) followed by post-
baking on a hot-plate (165 ºC, 10 min). Each driving electrode and reservoir was connected to a
contact pad on the edge of the substrate. The polymer coatings were removed from contact pads
by gentle scraping with a scalpel to facilitate electrical contact for droplet actuation.
27
For some experiments, hydrophilic sites were formed on ethanol sterilized DMF bottom plates as
reported previously.67
Briefly, 2 µL aliquots of fibronectin (33 µg/mL in DI water) were pipetted
onto the Teflon-AF surface covering optical windows and were allowed to evaporate at room
temperature for ~4 hours. The adsorbed protein spots formed in this manner were roughly
circular with ~1 mm diameter.
2.2.3 Two-plate DMF top-plate fabrication
Two-plate DMF device top-plates were formed from indium tin oxide (ITO) coated glass
substrates (Delta Technologies Ltd, Stillwater, MN). For most experiments, ITO-glass substrates
were coated with Teflon-AF and then were treated with a fluorocarbon lift-off procedure to form
an array of hydrophilic spots (i.e., circular regions of exposed ITO). Briefly, ITO-glass slides
were immersed in RCA solution (6:1:1 DI water: 28% aqueous ammonium hydroxide: 30%
hydrogen peroxide) for 15 minutes at 80°C. After rinsing, drying, and dehydrating, substrates
were spin-coated with Shipley S1811 photoresist (3000 RPM, 60 s) and then post-baked on a hot
plate (2 min, 95°C). The substrates were exposed (10 s, 29.8 mW/cm2) through a mask bearing
an array of five 1.00-, 1.25-, 1.50-, 1.75-, or 2.00-mm-diameter circular features (9 mm between
each feature) and then developed in MF-321. After rinsing and drying, the substrates were flood
exposed (10 s, 29.8 mW/cm2), and then spin-coated with Teflon-AF and post-baked using the
same parameters used for bottom-plate substrates (as above). The substrates were then immersed
in acetone with gentle agitation until the Teflon-AF over the patterned sites was lifted off (5-10
s). After rinsing and drying, the Teflon-AF was reflowed by baking on a hot plate at 165°C,
210°C, and 300°C for 5 minutes at each temperature. For some experiments, unpatterned top-
plates were formed without fluorocarbon liftoff, and were simply spin-coated with Teflon-AF
28
using the same parameters used for bottom-plate substrates (as above).
2.2.4 Two-plate DMF device assembly and operation
Two-plate digital microfluidic devices were assembled with an ITO–glass top plate and a
chromium-glass bottom plate as shown in Figure 2-1B. The two plates were joined by stacking
one, two, or three layers of double-sided tape (each layer ~80 µm), and were aligned such that
the edge of the top plate was adjacent to the outer-edges of the reservoir electrodes on the bottom
plate. In cases in which optical windows and top-plate hydrophilic sites were used, care was
taken to align these features vertically (windows on the bottom plate and hydrophilic sites on the
top plate). Driving potentials, ~300 VRMS for operation in air or ~200 VRMS for operation in oil,
were generated by amplifying the sine wave output of a function generator (Agilent
Technologies, Santa Clara, CA) operating at 18 kHz. Each reagent was loaded onto the device by
pipetting an aliquot onto the bottom plate at the edge of the top plate, and simultaneously
applying driving potential to the closest reservoir electrode (relative to the ITO electrode on the
top plate) to draw the fluid into the reservoir. Thereafter, droplets were actively dispensed,
moved, and merged by applying driving potentials to sequential actuation electrodes on the
bottom plate relative to the ITO electrode on the top plate as described previously.14
In all DMF
experiments, reagent solutions were supplemented with 0.02% Pluronics F68.74
2.2.5 Single-plate DMF device fabrication, assembly, and operation
The one-plate DMF device design consisted of twelve 3 x 3 mm square electrodes adjacent to a
linear 1 mm wide ground electrode with 40 µm between each electrode. Devices were coated
with Parylene-C using the same method described for bottom plates of two-plate devices (as
29
above). Devices were then coated with Teflon-AF bearing 1 to 2 mm diameter circular
hydrophilic sites using a modified liftoff procedure. Briefly, 10 nm of chromium was deposited
onto the Parylene by electron beam evaporation, which was then patterned into circular sites by
photolithography and etching using parameters described for bottom plates of two-plate devices
(as above). Devices were spin coated with Teflon-AF 1600 (1% wt/v in FC-40, 3000 RPM, 1
min), baked for 10 min at 165°C then flood exposed (10 s, 29.8 mW/cm2). Devices were
immersed in acetone with gentle agitation until Teflon-AF lifted off (~5-10 min) revealing a
pattern of circular chromium features. Devices were rinsed with DI, dried under a stream of
nitrogen, then baked on a hot plate for 10 min at 165°C. Single-plate devices were loaded by
pipetting 20 µL of reagent directly onto the outermost driving electrode. Sine wave driving
potentials of ~600 VRMS at 18 kHz were applied with the same amplified function generator
described above. Droplets were made to translate across the device by potentiating sequential
square electrodes relative to the linear electrode (held at ground) as described previously.60
2.2.6 DMF dispensing experiments
Devices bearing droplets were imaged with a CCD camera (Basler, Ahrensburg, Germany)
mounted above the device. For two-plate devices, ImageJ software was used to estimate the
apparent cross-sectional area of each droplet (typically circular but in some cases in the shape of
an irregular polygon) and volume (knowing the intra-plate spacer thickness). For one-plate
devices, each device was weighed on a microbalance before and after dispensing (after removal
of the remainder of the source droplet with a tissue), allowing for estimation of dispensed
volume on the basis of mass. The reagents evaluated included phosphate buffered saline (PBS)
with 0.2% blue food dye, 0-65 wt% sucrose solutions prepared in DI water (with viscosities from
30
literature values75
), and Dulbecco’s Modified Eagle Medium (DMEM) with 10% fetal bovine
serum (FBS). At least three replicates were performed for all conditions.
In each experiment, an aliquot of the appropriate reagent was loaded into a device, and one of a
number of conditions was evaluated. (1) Active dispensing on a two-plate device. A unit droplet
(i.e., a droplet covering one actuation electrode) was actively dispensed onto actuation electrodes
and the volume was estimated. (2) Passive dispensing onto top-plate hydrophilic sites in air on a
two-plate device. A unit droplet was actively dispensed as in (1) above and then translated over a
hydrophilic site, and the volume of the passively dispensed droplet was estimated. In some
instances, two unit droplets were actively dispensed and merged, then the combined droplet was
translated over the hydrophilic site, and the volume of the passively dispensed droplet volume
was estimated. For all spot sizes, dry dispensing (i.e., passive dispensing onto sites not bearing a
droplet), and wet dispensing (i.e., passive dispensing onto sites bearing a droplet from a previous
dispensing experiment) were evaluated. (3) Passive dispensing onto top-plate hydrophilic sites in
oil on a two-plate device. Droplets were first passively dispensed in air (dry dispensing) as in (2),
above. The entire device (i.e., all of the space between the top and bottom plates not occupied by
a droplet) was then filled with light mineral oil. Wet dispensing was then evaluated as in (2),
above. (4) Passive dispensing onto bottom-plate hydrophilic sites in air on a two-plate device. A
unit droplet was actively dispensed as in (1) above, and then translated over the hydrophilic site
(i.e., a patch of adsorbed protein on the bottom plate), and the volume of the passively dispensed
droplet volume was estimated. (5) Passive dispensing on a one-plate device. A 20 µL droplet
was translated across a patterned hydrophilic site. A sub-droplet was generated by passive
dispensing (after which, the volume was estimated), as the main droplet was actuated away.
31
2.2.7 Cell Culture and experiments
Marbin Darby canine kidney (MDCK) epithelial cells were kindly provided by Dr. N. Tufenkji
(McGill University). MDCKs were cultured in DMEM supplemented with 10% FBS, 100 U/mL
penicillin, 100 ug/mL streptomycin. Cells were incubated at 37°C in a humidified incubator
containing 5% CO2. 5 µL aliquots of cell suspensions (1 106 cells/mL) in media were pipetted
onto reservoir electrodes of two-plate devices bearing optical windows on the bottom plate. Unit
droplets were actively dispensed onto the electrode array, and translated across hydrophilic sites,
(formed either by liftoff on the top-plate or fibronectin absorption on the bottom-plate), resulting
in passive dispensing. Devices were then incubated at 37°C in a humidified incubator containing
5% CO2 for 18 hours. Images of cells were acquired through the optical windows by light
microscopy and the numbers of cells dispensed were enumerated using ImageJ.
2.3 Results and Discussion
2.3.1 Lift-off patterning
DMF devices are typically coated with fluorocarbon (FC) polymers such as Teflon-AF (DuPont),
CYTOP (Asahi), or Fluorad (3M). These materials have desirable properties including low
surface energy, broad chemical resistance, thermal stability, and biocompatibility. Early DMF
applications were implemented in devices bearing homogenous FC surfaces for applications
including suspension cell culture,40
PCR,76
enzymatic assays,77
and DNA sequencing.78
More
recently, DMF devices have been combined with heterogeneous surfaces (bearing different
chemical functionalities) for more sophisticated applications. These methods can be sub-divided
into those relying on modifications of the device surface itself67
or by incorporation of external
materials with heterogeneous surface properties such as magnetic beads79
or polymer plugs.57
In
32
this paper, we focus on the former -- heterogeneous patterned device surfaces. As described in
the introduction (and depicted in Figure 2-1C), DMF devices with surface modifications are
particularly useful for a form of fluidic manipulation called passive dispensing.40
We report here a technique to form DMF devices that are globally coated with Teflon-AF, but
periodically patterned with hydrophilic spots. In most of the work reported here, the hydrophilic
spots were formed from exposed indium tin oxide (ITO) on the top plate of a DMF device as
illustrated in Figure 2-1B. Our new technique is similar to that described by Chen et al. and
Malic et al.80
for forming patterned Teflon-AF on ITO and gold surfaces, respectively.
Significant trial and error was required to develop techniques that were reproducible with
particularly important results being inclusion of an RCA cleaning step for improved adhesion of
Teflon-AF to ITO and an extra UV exposure step to assist in photoresist removal. We
characterized these surfaces by x-ray photoelectron spectroscopy finding similar indium, tin, and
oxygen compositions on the liftoff sites as compared to untreated ITO (Figure 2-2 and Table
1-1). These measures and others (described in detail in the experimental section) form a robust
and repeatable method that we have now used to pattern hundreds of substrates bearing circular
structures with near-perfect pattern fidelity.
2.3.2 Passive dispensing and virtual microwells
The new methods for patterning surfaces described above were developed to facilitate robust
formation of virtual microwells by passive dispensing. As shown in Figure 2-3A, the simplest
form of passive dispensing can be called "dry" passive dispensing, in which a VM is formed on
an empty hydrophilic site. We evaluated the effects of varying gap spacing between
33
Figure 2-2: X-ray photoelectron spectroscopy evaluation of patterned surfaces. To evaluate the
chemical composition of exposed hydrophilic sites and adjacent Teflon surfaces XPS
measurements were taken on patterned and unpatterned surfaces for comparison: (A) ITO
composition at hydrophilic sites, (B) Teflon on patterned slides, (C) unprocessed ITO surfaces,
and (D) unprocess Teflon surfaces.
Table 2-1: Elemental identification and quantification of species detected on patterned DMF
surfaces using XPS. Samples labels refer to Figure 2-2. Sample (Atomic %)
Species A B C D
C 40.73 38.30 36.30 31.00
F 0.31 50.26 52.38 0.00
N 0.68 0.11 0.01 0.10
S 1.26 0.01 0.00 0.00
Sn 2.61 0.00 0.00 4.14
In 17.22 0.00 0.00 26.77
O 35.83 11.12 11.29 37.86
Si 0.34 0.03 0.01 0.13
Ca 1.02 0.16 0.01 0.00
34
the top and bottom plates and the diameter of the hydrophilic site on dry passive dispensing. As
shown in Figure 2B, by varying the hydrophilic site diameter from 1 mm to 2 mm and the inter-
plate gap height from 80 µm to 240 µm, VMs with volumes ranging from ~80 nL to ~800 nL
were formed. The precision of these volumes varied within the range of 0.7% to 13.8% for all
conditions tested. The CVs increased with greater dispensed volume (either higher gap spacing
or larger hydrophilic surface area).
In initial experiments, we observed that a single actively dispensed droplet (i.e., Figure 2-1C,
frames i-ii) was not always sufficiently large to serve as the source droplet for dry passive
dispensing. In such cases, the VM did not properly separate from the source droplet, or the
remainder of the source droplet (after forming the VM) was too small to actuate away. Thus, for
the conditions in Figure 2-3B labeled with an asterisk (*), two droplets were actively dispensed
and subsequently merged, and this combined volume served as the source droplet for passive
dispensing. This observation led us to develop a quantitative criterion predictive of passive
dispensing success, which we call the "virtual microwell number," Nvm, which is defined in terms
of the area of the square actuation electrodes (Ae), the area of the circular hydrophilic site (Ahs),
and the distance between top and bottom plates (h):
35
Figure 2-3: Dry passive dispensing to form virtual microwells. (A) Video stills (top-to-bottom)
depicting dry passive dispensing. The dashed circle in panel (i) indicates the position of the
hydrophilic site. (B) Volumes of droplets dispensed in dry passive dispensing as a function of
spacer height and hydrophilic site diameter (n = 5). Asterisks (*) indicate that source droplets were
formed from two actively dispensed droplets. Error bars are 1 S.D. (C) Parameter NVM calculated
for each experimental condition in (B). The shaded region, NVM < 2, indicates conditions in which
two actively dispensed droplets were required to generate the source droplet for successful passive
dispensing.
36
hA
AN
hs
evm Eqn. 2.1
Figure 2-3C summarizes Nvm across a range of device and feature parameters. We observe for all
experiments where Nvm > 2, a single unit droplet actively dispensed from the reservoir was
sufficient for successful generation of the VM. For Nvm < 2, a single unit droplet was insufficient
for successful passive dispensing. This is mostly consistent with the observations described by
Chen et al.,69
with a discrepancy observed for cases when Nvm is close to 2. We propose that this
discrepancy may be attributed to differences in device design and operation. Regardless, we
anticipate that Nvm will be a useful heuristic in the design of VMs on DMF devices in the future.
For the majority of applications it is of interest to exchange reagents in VMs. We term this type
of exchange "wet" passive dispensing, which is implemented when a source droplet is actively
dispensed and then translated across a previously formed VM, displacing its original contents
(Figure 2-4A). Barbulovic-Nad et al.67
demonstrated that after three such exchanges, 100% of the
content of the original VM is replaced. We evaluated VMs for hydrophilic sites with diameters
ranging from 1000 to 2000 µm for the ability to repeatedly dispense volumes to sites bearing
VMs. As shown in the gray bars in Figure 2-4B, the precision in wet dispensing is very high for
small sites, with CVs of 1.8%, 1.5%, and 0.7% for 1000 µm (126 nL), 1250 µm (196 nL), and
1500 µm (283 nL) diameter hydrophilic sites, respectively. Larger hydrophilic sites were
associated with lower precision, with CVs of 12% and 7% for 1750 µm (385 nL) and 2000 µm
(502 nL) diameter hydrophilic sites, respectively. Regardless, these data indicate that given
37
Figure 2-4: Wet passive dispensing to exchange fluid in a virtual microwell. (A) Video stills
(top-to-bottom) depicting wet passive dispensing in which the virtual microwell contained blue
dye at the hydrophilic site (i) and a red dye source droplet is actuated across the virtual
microwell displacing the blue droplet (ii-v). (B) Multiple passes of reagent across virtual
microwells with varying diameters for 160 µm spacer height. The gray and white bars represent
devices operated with a surrounding matrix of air and mineral oil, respectively. Error bars are 1
S.D.
38
volumes can be repeatedly dispensed to a given site multiple times with good (and for low
volumes, excellent) precision.
The data above (and in most of the experiments described here) were generated using devices in
which droplets were surrounded by a matrix of air. An alternative format is to fill devices such
that droplets are surrounded by a matrix of oil, which has the benefit of lower voltage
requirements, reduced surface fouling, and decreased droplet evaporation.80
We demonstrate
here the implementation of passive dispensing in oil-filled DMF devices. Interestingly, we found
dry passive dispensing in oil to be impossible. We speculate that this is because a thin film of oil
film forms over the hydrophilic site and prevents hydrophilic interactions with aqueous droplets.
In contrast, we found that wet passive dispensing in oil was straightforward when VMs were first
loaded in air by dry passive dispensing and the devices were then filled with light mineral oil
(which did not displace the aqueous droplets from the VMs) for wet dispensing, as shown in the
white bars in Figure 2-4B. The VMs formed in oil were in 1.1-fold to 1.25-fold larger the initial
dry dispensed volume in air. We attribute this phenomenon to the increased viscous forces
between the droplet and the filler medium.22
With multiple passes, the oil-associated volume was
maintained consistently, with CVs ranging from 2.8% to 9.4% for volumes of 240 nL to 810 nL.
We propose that the capacity to combine passive dispensing with oil-filled devices will be useful
for a range of different applications benefitting from the use of oil to facilitate droplet motion
and reduce the effects of potential evaporation.
The data above (and in most of the experiments described here) were generated using two-plate
DMF devices in which droplets are sandwiched between a top and bottom plate (Figure 2-1). In
39
Figure 2-5: Single-plate DMF passive dispensing. (A) Picture of a single-plate device depicting
a source droplet and a passively dispensed droplet. (B) Schematic depicting the single-plate
device geometry.
Figure 2-6: Active and passive dispensing as a function of reagent viscosity. Sucrose solutions
of varying viscosity were dispensed on DMF either by active or passive dispensing onto 1500
µm diameter hydrophilic sites (n = 6). Dispensed volumes are plotted as a function of solution
viscosity. Error bars are ± 1 S.D., 95% confidence intervals are indicated by shaded regions, and
the mean dispensed volume for each dispensing mechanism is indicated by a solid horizontal
line.
40
the alternative single-plate device format, the larger droplet volume-to-electrode-area ratio
results in lower actuation forces relative to two-plate DMF; thus as far as we are aware, there
have been no reports of reagent dispensing (of any kind) on single-plate DMF devices. Here, we
report the extension of the concept of passive dispensing on hydrophilic sites to single-plate
DMF devices (Figure 2-5). Applying fluorocarbon lift-off to single-plate DMF devices provides
the ability to dispense reagents in this format. The hydrophilic sites in such systems were formed
by evaporation of chromium (and subsequent patterning) on the top of the Teflon surface of a
complete single-plate device. For hydrophilic sites with diameters of 1250 µm and 1500 µm, the
droplets dispensed from 20 µL source droplets had volumes of 330 nL ± 35 nL and 420 nL ± 55
nL (CVs of 11 and 13%, respectively).
2.3.3 Active vs. passive dispensing
Active dispensing is regarded as the standard technique for reagent dispensing on DMF. In active
dispensing, the liquid is stretched from a reservoir by electrostatic manipulation and then necked
prior to splitting. Fouillet et al.81
reported a CV of below 4% for active droplet dispensing of
reagent into oil-filled DMF devices, similar to pipettes, where precision is reported at under 6%
CV for 0.1 to 2.5 uL. Actively dispensed volumes are limited by device geometry; different
electrode sizes are required to achieve dispensing of different volumes.57
Further, when we
examined the active dispensing of a range of sucrose solutions with varying viscosity from 0 cP
to ~150 cP, we found poor repeatability, with errors of up to 30% and a 95% confidence interval
across all viscosities of 0.2 mm3 (red circles in Figure 2-6). In comparison, VM volumes formed
by passive dispensing were relatively independent of viscosity, with calculated 95% confidence
41
intervals for dispensed volume of 0.02 mm
3. In the future, we propose that the integration of
different sizes of hydrophilic sites on device will make it straightforward to access a broad range
of reagent volumes, and improve precision and accuracy of reagent dispensing independent of
viscosity.
2.3.4 Lift-off vs. protein absorption for passive dispensing
The liftoff-based techniques for forming hydrophilic patches on device top plates described here
were developed as a result of our dissatisfaction with methods relying on hydrophilic patches
formed from adsorbed proteins,67
for the reasons listed in the introduction. Here, we report a
comparison of the two systems for the ability to passively dispense droplets containing
suspended cells, and the ability of the cells to spread on the device surface. After performing
independent trials of dispensing MDCK cells suspended in cell culture media, we found lower
cell numbers and greater variability in the case of sites formed by protein spotting (cell number
mean = 24 cells, cell number CV = 79%) as compared to those formed by fluorocarbon liftoff
(cell number mean = 70 cells, cell number CV = 19%) (Figure 2-7A). Furthermore, as shown in
Figure 2-7B, there were no significant differences in cell morphology for the two types of
systems, which supports literature reports of ITO as being a suitable surface for cell culture.82
In addition to improved reproducibility in dispensing, the new technique reported here has
additional benefits for cell culture in DMF, including increased electrical isolation of cells from
the actuation electrodes and compatibility with long-term culture. In DMF systems such as those
reported here, the majority of the voltage drop occurs across the dielectric coating of the bottom-
plate;40
however, localized charge densities at the device surface may vary, which might result in
42
Figure 2-7: Comparison of hydrophilic sites formed by adsorbed protein (on the bottom plate)
vs. liftoff (on the top plate) for dispensing cells into virtual microwells. (A) Results of five trials
seeding MDCK cells (5 105 cells/mL) by passive dispensing. (B) Bright-field images of
MDCK cells seeded on fibronectin coated Teflon and indium tin oxide after 6 hours. Scale bar =
50 µm.
43
augmented transcriptional profiles within cultured cells. Further, localized heating in the
dielectric layer might result in potentially deleterious cellular effects. Decoupling the cell culture
site (by moving it to the top plate) from the electrode-bearing substrate (on the bottom plate) may
dampen these effects. In the case of long-term culture, dielectric coatings are prone to failure due
to accumulation of charge and moisture infiltration during incubation. Cell culture on the top-
plate allows for the replacement of defective bottom-plates (with cells grown continuously on
top-plates) without compromising the experiment being performed. We propose that this
arrangement will be useful for the long-term culture of sensitive cell types, particularly stem and
primary cells.
2.4 Conclusion
We present the utility of Teflon liftoff for improving digital microfluidic functionality. This
precise method for patterning hydrophilic sites on hydrophobic DMF device surfaces resulted in
multiple advances, including: (1) formation of virtual microwells for precise reagent dispensing,
(2) passive dispensing in air and oil filled devices, (3) the first demonstration of passive
dispensing on a single-plate device, and (4) improved surfaces for cell culture and other
heterogeneous assays. We anticipate this new method will be useful for DMF-based techniques
applied to a broad range of applications.
44
Chapter 3. A digital microfluidic platform for primary cell
culture and analysis
Summary
Digital microfluidics (DMF) is a technology that facilitates electrostatic manipulation of discrete
nano- and micro- litre droplets across an array of electrodes, which provides the advantages of
single sample addressability, automation, and parallelization. There has been considerable
interest in recent years in using DMF for cell culture and analysis, but previous studies have used
immortalized cell lines. We report here the first digital microfluidic method for primary cell
culture and analysis. A new mode of “upside-down” cell culture was implemented in by
patterning the top plate of a device using a fluorocarbon liftoff technique. This method was
useful for culturing three different primary cell types for up to one week, as well as
implementing a fixation, permeabilization, and staining procedure for F-actin and nuclei. A
multistep assay for monocyte adhesion to endothelial cells (ECs) was performed to evaluate
functionality in DMF-cultured primary cells and to demonstrate co-culture using a DMF
platform. Monocytes were observed to adhere in significantly greater numbers to ECs exposed to
tumor necrosis factor (TNF)-α than those that were not, confirming that ECs cultured in this
format maintain in-vivo-like properties. The ability to manipulate, maintain, and assay primary
cells, demonstrates a useful application for DMF in studies involving precious samples of cells
from small animals or human patients.
45
3.1 Introduction
There are two types of mammalian cells that are commonly used in biomedical research:
immortalized cell lines and primary cells. Immortalized cell lines can be grown in vitro for many
generations, spanning many months-to-years. These cells are straight-forward to grow and
maintain, but often have phenotypes that differ significantly from those of cells in vivo. In
contrast, primary cells are used immediately after isolation from animal tissue, and therefore are
much closer to in vivo phenotype. Unfortunately, primary cells have several limitations for
regular use in the laboratory. In long-term studies involving animal models of disease, primary
cells are typically available only in limited quantities (e.g., with monthly or yearly isolations).
The process of primary cell isolation can be laborious and costly, requiring expensive reagents
and hours-to-days of work depending on the cell type. Furthermore, due to their limited number
of population doublings, primary cells can only be used for a short period of time in the
laboratory. These factors make primary cells an attractive target for miniaturized tools to reduce
costs and for automated cell culture and analysis.
Microfluidic channels are the most popular technology used for miniaturization. Primary cell
culture in microfluidic channels has been demonstrated repeatedly with applications including
cell migration,1-3
adhesion,4-6
shear stress,7-9
cell sorting,10
and cell-based screening assays.11
However, microchannel-based systems often require pumps or other external apparatus (with
noted exceptions12
) for applications involving cells. This increases reagent/sample consumption,
as such systems require macro-scale tubing and interconnects, which inherently contributes
unwanted dead volumes. An additional problem associated with interconnects and other world-
to-chip interfaces is the presence of bubbles, which can disturb the local fluid flow within
46
microchannels and can damage cells as a result of the high interfacial energy at the gas-liquid
interface. Removing bubbles can be difficult, requiring complex degassing mechanisms or
bubble traps.13
Digital microfluidics (DMF) is an alternative platform to conventional enclosed microchannels
that is capable of manipulating discrete liquid droplets on an array of patterned electrodes.14
In
DMF, droplets can be controlled individually or in parallel to provide precise spatial and
temporal control of reagents. Typical volumes for droplets can range from nanolitres to
microlitres, and because there is no dead volume, these systems are well suited for minimal
reagent/sample consumption. Moreover, because there are no open reservoirs or tubes and
interconnects, devices can be readily flipped, allowing for convenient use of both sides of each
device for imaging. Finally, unlike enclosed microchannels, in non-oil-filled DMF systems,
bubble nucleation and growth are non-existent. Previous studies15-21
have demonstrated that
mammalian cells can be cultured and/or analyzed on DMF platforms, but all of the previous
work used immortalized cell lines.
Here, we report the first application of DMF to the culture and analysis of primary cells. Three
phenotypically different cell types isolated from pig blood vessels (aortic endothelial cells) and
heart valves (aortic valve endothelial cells and aortic valve interstitial cells) were cultured and
analyzed on a DMF platform. The devices and methods reported here use a new mode of
"upside-down" culture in virtual microwells22
formed by a patterned DMF top plate. Cells were
cultured on multiple sites per device for up to one week. With minimal reagent use, primary
mammalian cells were fixed, permeabilized and stained on a DMF device. Furthermore, a co-
47
culture system for growing an analyzing endothelial cells and monocytes was developed; this is
the first co-culture system that we are aware of in DMF. The co-culture system was used to
implement a monocyte adhesion assay, which confirmed that intricate signaling mechanisms
were retained by primary cells cultured on this new digital microfluidic platform.
3.2 Methods and Materials
3.2.1 Reagents and Materials
Unless stated otherwise, materials were purchased from Fisher Scientific Canada (Ottawa, ON,
Canada). General-use chemicals were from Sigma Aldrich (Oakville, ON, Canada), fluorescent
dyes were from Invitrogen/Life Technologies (Burlington, ON, Canada), and photolithography
reagents were from Rohm and Haas (Marlborough, MA). Deionized (DI) water had a resistivity
of 18 MΩ·cm at 25°C.
3.2.2 DMF Device Fabrication and Operation
Digital microfluidic devices were fabricated in the University of Toronto Emerging
Communications Technology Institute (ECTI) cleanroom facility, using transparent photomasks
printed at 20,000 DPI (Pacific Arts and Designs Inc., Markham, Ontario). Glass DMF device
bottom-plates bearing patterned chromium electrodes were formed by photolithography and
etching as described previously.15
As shown in Figure 3-1, the design featured an array of 116
actuation electrodes (2.2 mm x 2.2 mm ea.) connected to 10 reservoir electrodes (4 mm x 4 mm
ea.), with inter-electrode gaps of 30-80 µm. The actuation electrodes were roughly square with
interdigitated borders (140 µm peak to peak sinusoids). The design also included an array of five
1 mm diameter optical windows (i.e., circular regions free from chromium) with 9 mm between
48
Figure 3-1: (A) Photograph of DMF device designed for primary cell culture and analysis. A
series of droplets (coloured with red dye for visualization) are positioned at patterned hydrophilic
sites on a device. (B) Schematic of device geometry. The top plate is patterned by a liftoff
procedure to expose hydrophilic sites. The bottom plate bears an array of individually
addressable electrodes with patterned optical windows for imaging. (C) Top and side view
schematic of passive dispensing on hydrophilic sites. (i-ii) A droplet is manipulated to the
hydrophilic site. By actuation of subsequent electrodes the droplet is (iii) stretched then (iv)
passively dispensed, forming a virtual microwell. (D) Side view schematic of device orientation
during experimentation. Devices are maintained right-side up during droplet actuation and are
positioned upside-down during all incubations.
49
each window. As illustrated in Figure 3-1C, each window straddled two actuation electrodes.
After patterning the electrodes, the substrates were coated with 7 µm of Parylene-C (Specialty
Coating Systems, Indianapolis, IN) and 200 nm of Teflon-AF (DuPont, Wilmington, DE).
Parylene-C was applied using a vapor deposition instrument (Specialty Coating Systems), and
Teflon-AF was spin-coated (1% wt/wt in Fluorinert FC-40, 3000 rpm, 60 s) followed by post-
baking on a hot-plate (165 ºC, 10 min). The polymer coatings were removed from contact pads
by gentle scraping with a scalpel to facilitate electrical contact for droplet actuation.
DMF device top-plates were formed from indium tin oxide (ITO) coated glass substrates (Delta
Technologies Ltd, Stillwater, MN) that were coated with Teflon-AF (200 nm, as above). A lift-
off process was used to form an array of 1.5 mm diameter openings of exposed ITO (9 mm
between each opening) through the Teflon-AF using methods developed for this purpose.22
Digital microfluidic devices were assembled with an ITO–glass top plate and a chromium-glass
bottom plate. Prior to assembly, the two plates were sterilized by immersing in 70% ethanol (10
min) and then air dried. The hydrophilic sites (exposed ITO) on the top plate were aligned
visually to the optical windows on the bottom plate, and the two plates were joined by a spacer
formed from four pieces of double-sided tape (total space between plates ~280 µm). Driving
potentials (~280 VRMS) were generated by amplifying the sine wave output of a function
generator (Agilent Technologies, Santa Clara, CA) operating at 18 kHz and were applied
between the top plate (ground) and sequential electrodes on the bottom plate via the exposed
contact pads. Pluronics F68 (0.02% wt/vol) was added to all reagents used with digital
microfluidics (excluding solutions of Triton X-100) to facilitate droplet movement.23
A
50
In addition to the standard digital microfluidic operations24
(i.e., active droplet translation, active
droplet dispensing from reservoirs, etc.), the devices supported a phenomenon known as passive
dispensing.15
As illustrated in Figure 3-1C, in passive dispensing, a source droplet is translated
across a hydrophilic site, and surface tension effects result in spontaneous formation of a sub-
droplet. In the devices with the dimensions described here, source droplets were 1.4 µL and
passively dispensed droplets were 0.5 µL; as reported elsewhere,22
the volumetric reproducibility
for passive dispensing for these dimensions is excellent, with a CV of ~1.2%. As described
below, passive dispensing was used for all DMF operations for primary cell culture and analysis.
3.2.3 Primary Cell Isolation and Maintenance
Porcine aortic endothelial cells (PAECs) isolated from pig thoracic aortas were kindly donated
from Lowell Langille (University of Toronto).25
Porcine aortic valve endothelial cells (PAVECs)
and porcine aortic valvular interstitial cells (PAVICs) were isolated as described previously.26, 27
PAECs were cultured in M199 (Wisent, St. Bruno, QC, Canada) supplemented with 5% cosmic
calf serum (Fisher Scientific Canada), 5% fetal bovine serum (FBS) (Fisher Scientific Canada),
and 1% penicillin-streptomycin (P-S) (Sigma Aldrich). PAVECs and PAVICs were cultured in
M199 and Dulbecco’s modified eagle’s medium (DMEM) (Wisent), respectively, each
supplemented with 10% FBS and 1% P-S. Cells were cultured in T75 flasks until 80% confluent,
then trypsinized, centrifuged, and resuspended at approximately 105-10
6 cells/mL in the
appropriate completed culture medium (with M199 or DMEM, as above) to form a cell
suspension for use with DMF.
51
3.2.4 DMF Cell Culture
Five 5 µL aliquots of cell suspensions were pipetted onto the reservoir electrodes, and then five
1.4 µL droplets (one per reservoir) were actively dispensed by applying potentials to a series of
actuation electrodes adjacent to each reservoir. These 1.4 µL cell-containing droplets were
driven to the hydrophilic spots patterned on the top plate such that 0.5 µL droplets were
generated by passive dispensing (Figure 3-1C). The devices were then inverted (with the top
plate on the bottom) (Figure 3-1D) and were placed in a homemade humidified chamber (a Petri
dish containing dampened Kimwipes to prevent evaporation) in an incubator at 37 oC and 5%
CO2 for 12 h. This “incubation state” (i.e., top plate on the bottom in a humidified chamber in a
cell culture incubator) was used for all incubation steps described herein. Periodically, devices
were removed from the incubator, flipped to orient each device with the ITO top-plate on the top
(such that the device was upright) and used for droplet movement. Afterwards, devices were
returned to the incubation state. For cell culture, new droplets of media were delivered to cells
every 12-16 hours until cells were ~70-80% confluent.
3.2.5 DMF Staining and Microscopy
For imaging without staining, primary cells cultured on DMF were imaged using an inverted
CKX41 microscope (Olympus, Markham, ON, Canada) in phase-contrast mode. For comparison,
cells were also cultured on tissue culture treated polystyrene (TCPS) flasks and imaged. For
imaging of stained cells, after ~70-80% confluence was reached on DMF devices, primary cells
were washed by dispensing at least two 1.4 µL droplets of phosphate buffered saline (PBS)
across the virtual microwell sites (displacing the existing droplets with fresh 0.5 µL volumes).
Cells were fixed and permeabilized by dispensing and actuating three 1.4 µL droplets across the
52
cells (in series) of (a) 10% (v/v in DI water) neutral buffered formalin (NBF) for 5 minutes, (b)
PBS, and (c) 0.01% (v/v in PBS) Triton X-100 for 5 minutes. The cells were then washed (two
droplets of PBS as above), and 1.4 µL droplets containing FITC-labeled phalloidin (0.1 mg/mL
in PBS) were actively dispensed from reservoirs and actuated across the cell culture site such
that 0.5 µL sub-droplets were passively dispensed and then incubated for 45 minutes at room
temperature. The cells were then washed in PBS (as above), and 1.4µL droplets containing
Hoechst (1 µg/mL in PBS) were driven across the c -droplets
were passively dispensed and then incubated for 5 minutes at room temperature, and then
washed again with PBS (as above). Cells on DMF devices were imaged by flipping them (such
that the top plate was on the bottom) using an IX-71 microscope (Olympus) in fluorescence
mode.
3.2.6 DMF Monocyte Adherence Assay
THP-1 monocytes (ATCC, Manassas, WA) were cultured in suspension off-chip in RPMI 1640
medium (Invitrogen/Life Technologies) completed with 10% FBS and 1% P-S. Prior to
experiments, monocytes were centrifuged, resuspended in media containing Hoechst (0.2 µg/ml
in complete medium), incubated for 30 minutes, and then centrifuged and resuspended in fresh
complete medium at 106
cells/mL. PAECs grown to confluence on DMF devices were incubated
with passively dispensed 0.5 µL droplets containing 0 or 25 ng/mL tumour necrosis factor alpha
(TNF)-α (Invitrogen/Life Technologies) in complete medium for 4 hours in the incubation state
(see above). Cells were then rinsed by passively dispensing two 0.5 µL droplets of PBS,
followed by passive dispensing of one 0.5 µL droplet containing calcein AM (2 µM in PBS
containing Ca2+
and Mg2+
) and storing for 15 minutes in the incubation state. Cells were then
53
rinsed by passively dispensing two 0.5 µL droplets of PBS, followed by passive dispensing of
one 0.5 µL droplet of complete culture medium and incubating for 30 min in the incubation state.
0.5 µL droplets containing Hoechst-labeled monocytes were then delivered to the PAECs by
passive dispensing and stored for 10 min in the incubation state. Two droplets of PBS were used
to wash the cells (as above), and the cells were then evaluated using an IX-71 microscope for
monocyte adhesion. One central image per hydrophilic spot was collected and images were
analyzed for monocyte number. Briefly, IMAGEJ software was used to convert images to binary
and the “analyze particles” function was used to count the cells.
3.3 Results and Discussion
3.3.1 Digital Microfluidic Primary Cell Culture
We present here the first digital microfluidic platform capable of culturing and analyzing
primary cells, shown in Figure 3-1. PAECs, PAVECs, and PAVICs were chosen as model cell
types because of their importance in cardiovascular biology.4, 26-29
Although these cell types are
found in close proximity anatomically, they represent three distinctly different phenotypes.30
Moreover, PAVECs are an especially interesting target because they are challenging to isolate
and culture in vitro; under improper culture conditions, they display altered morphologies,
function and short-term viability.26, 29, 31
We hypothesize that if DMF is useful for culturing,
handling, and analyzing these different types of cells (particularly, the sensitive PAVECs),
similar methods may be applicable to cells derived from a wide range of tissue types.
PAECs, PAVECs, and PAVICs are adherent cells -- that is, they attach, spread, and grow on
solid surfaces. There have been three previous reports15, 20, 21
of culture of adherent cells on DMF
54
platforms. As listed in Table 1, the new methods reported here share a number of similarities and
differences with those reported previously. The most notable similarity is that each of these
systems is capable of supporting a phenomenon known as passive dispensing. Passive dispensing
is represented in Figure 3-1C; when an aqueous droplet is driven across a hydrophilic site, a
smaller droplet, which we call a "virtual microwell,"22
is spontaneously formed and left behind.
Passive dispensing to form virtual microwells is a unique feature of digital microfluidics, and
serves as a convenient mechanism to seed, culture, and analyze adherent cells.
The most important difference between the current system and those reported previously15, 20, 21
is the new device format and orientation. The methods reported here rely on hydrophilic sites
formed on the device top plate, which led us to implement a new method of "upside-down" cell
culture in virtual microwells (Figure 3-1D). In this scheme, devices are stored for most of the
time upside-down (i.e., top plate on the bottom) which allows the cells to adhere, spread, and
proliferate. At designated periods, devices are flipped to standard configuration (i.e., ITO plate
on the top) for droplet manipulation, but after experiments, the devices are returned to the
inverted state. This arrangement is advantageous for a number of reasons. First, it allows for cell
growth on hydrophilic sites formed from regions of exposed ITO22
rather than the adsorbed
proteins15
or peptides20, 21
used previously. In initial experiments with primary cells grown on
adsorbed fibronectin on DMF device substrates, we observed that the cells had unexpected
morphologies, whereas on ITO surfaces, cells had morphologies that are similar to those grown
on conventional TCPS substrates. Second, this device arrangement de-couples the active portion
of the digital microfluidic device (i.e., the insulating layer on the bottom plate which allows for
the buildup of charge necessary for droplet movement32
) from the cells. The insulating layers on
DMF devices are prone to failure over time because of dielectric breakdown, and the upside-
55
down culture arrangement allows for the possibility of replacing a used/defective bottom plate
with a fresh one between experiments (note that this putative feature was not used in any
experiments reported here). We propose that this arrangement will be useful for a variety of
applications for cell culture and other applications.
Using the methods described here, PAECs, PAVECs, and PAVICs can be reproducibly seeded
and grown with high viability. A significant amount of trial-and-error was required for this level
of performance, however, and some of the key points are described here. Factors such as cell
seeding density and media exchange frequency were critical in maintaining primary cell viability
and morphology on device. Seeding densities between 2 x 105 – 1 x 10
6 cells/mL coupled with a
media exchange frequency of every 12-16 hours maintained viable primary cells with
appropriate morphologies. Depending on the assay, the cell seeding densities were altered to
vary the duration of culture on device. For example, to demonstrate long-term cell culture,
PAECs were cultured for up to 1 week with an initial seeding density of 2 x 105 cells/mL. For
shorter experiments (e.g., those in which microscopy was performed 24 h after staining), primary
cells were seeded at 5 x 105 - 1 x 10
6 cells/mL, to achieve the desired level of confluence within
24 hours. At densities greater than 2 x 106 cells/mL, cells displayed rounded morphologies with
little spreading, possibly due to overpopulation of the hydrophilic sites and rapid accumulation of
cellular waste products. In all experiments, devices were stored in an incubator in humidified
chambers with no appreciable evaporation.
56
3.3.2 Digital Microfluidic Microscopy, Fixation, Permeabilization, and Staining
As shown in Figure 3-2, DMF devices proved to be a useful platform for microscopic imaging
of primary cells (in this case, using an inverted microscope). For imaging, devices were either
positioned with the bottom plate on the bottom (such that the bottom plate was adjacent to the
objective) as was the case for the images in Figure 3-2, or with top plate on the bottom (such
that the top plate was adjacent to the objective). The capacity to use and flip devices to either
orientation for imaging is a unique property of digital microfluidic devices, which have no open
reservoirs or tubing interconnects that might otherwise interfere. Figure 3-2 shows
representative phase contrast images of PAECs, PAVECs, and PAVICs grown on DMF devices
and for comparison, cells grown on conventional TCPS substrates. As shown, the morphologies
of cultured primary cells were similar on the two surfaces.
Microscopic imaging of cells is often enhanced by staining with fluorescent dyes, which reveals
information about cell state and phenotype. Prior to staining, cells are often fixed to preserve cell
state (by exposure to fixatives such as NBF), and permeabilized to allow for deep penetration by
dyes and other reagents (by exposure to mild surfactants such as Triton X-100). Before and after
these steps and others, the specimen must be repeatedly rinsed as the various reagents can
interfere with each other. Here, as demonstrated in Figure 3-3, we report the first combination of
all of these steps (cell growth, fixation, permeabilization, staining, and rinsing) by DMF. As
shown, at 40x magnification, individual actin stress fibers can be observed, demonstrating the
compatibility of DMF with high-resolution fluorescent microscopy.
57
Figure 3-2: Phase contrast images of PAECs, PAVICs, and PAVECs cultured on a DMF device
(top) and in TCPS flasks (bottom). Scale bar = 200 µm. In the DMF images, the bottom plate is
closest to the objective, and the focus is on the layer of cells on the top plate. The cells are
viewed through the circular optical window between two electrodes on the bottom plate (which
are observable but slightly out of focus).
Figure 3-3: Fluorescent images of PAECs, PAVECs, and PAVICs after fixing, permeabilizing,
and staining on a DMF device. The stains selected for F-actin (FITC-phalloidin, green) and
nuclei (Hoechst, blue). Images were taken at both 10x magnification (top row) and 40x
magnification (bottom row). Scale bar = 200 µm (top row) and 50 µm (bottom row). In these
images, the top plate is closest to the objective.
58
PAECs were cultured on DMF devices and then incubated either with or without TNF-α for 4
hours. Monocytes pre-labeled with Hoechst were then dispensed from reservoirs and delivered to
endothelial cells, which were then rinsed to remove monocytes that did not adhere. As shown in
Figure 3-4, monocytes had greater adhesion to TNF-α-stimulated PAECs compared to non-
stimulated controls, which is consistent with previous studies.9,36-38
These results demonstrate
compatibility of DMF with a fourth cell type (monocytes) and show that primary PAECs
cultured using DMF retain in vivo-like responses to TNF-α. Moreover, this is the first
demonstration of co-culture on a DMF platform. The ability to detect a response with monocytes
(i.e. adhesion) as a result of endothelial cell activation highlights the potential of DMF to
investigate cell-cell interactions.
3.3.3 Digital Microfluidic Monocyte Adhesion Assay
To evaluate the potential for using digital microfluidic systems for co-culture and multistep
assays, we probed its compatibility with endothelial cell/monocyte adhesion experiments.
Monocyte adhesion to endothelial cells is an important initiating event in the inflammatory
process. Endothelial cells are generally activated prior to adhesion, and this state can be induced
by exposure to cytokines such as TNF-α. TNF-α increases monocyte adhesion through
upregulation of EC receptors such as E-selectin,33
intercellular cell adhesion molecule-1 (ICAM-
1)34
and vascular cell adhesion molecule-1 (VCAM-1).34, 35
The assay represented in Figure 3-4
required only a 1.4 µL droplet of reagent and cell suspension for each virtual microwell. In
comparison, macroscale39, 40
and some microchannel-based adhesion assays9 require working
volumes of tens to hundreds of microlitres, such that the DMF system facilitates a 10-100-fold
reduction in reagents used. The capacity to reduce reagent consumption and increase throughput
59
with DMF is desirable in monocyte adhesion assays or other cases in which precious sample or
expensive reagents are used. The potential for combining automated imaging and analysis with
DMF in the future is an attractive vision, as such a system would likely be useful for applications
ranging from basic biology to drug discovery.
3.3.4 Conclusions
We present the first demonstration of primary cell culture using digital microfluidics. A new
mode of “upside-down” culture in virtual microwells was developed to enable primary cell
growth with appropriate morphologies and to decouple the cell growth sites from the digital
microfluidic driving electrodes. Multi-step cell fixation, permeabilization, and staining processes
were demonstrated for the first time on a DMF platform. A monocyte adhesion assay was
performed to demonstrate functionality in DMF-cultured primary ECs and to highlight the co-
culture capabilities of the device. The combination of DMF and primary cell culture/analysis
presented here provides a basis for future studies involving co-culture, high resolution
microscopy, and multiplexed experimentation.
60
Figure 3-4: A monocyte adhesion assay performed on DMF-cultured primary PAECs. (A)
Nuclear-stained (Hoechst, red) THP-1 monocytes adhered to PAECs (calcein AM, green).
Representative images of nuclear-stained monocytes adhered to (B) non-stimulated and (C)
TNF-α-stimulated PAECs. In these images, the top plate is closest to the objective. (D)
Monocytes displayed greater adhesion on TNF-α-stimulated PAECs relative to control non-
stimulated PAECs. Data presented as mean ± standard deviation. *P < 0.05. Scale bar = 200 µm.
61
Table 3-1: Comparison of adherent cell culture using DMF between the new methods reported
here and those previously published.
New Methods
Reported Here
Methods Reported by
Barbulovic-Nad et al.15
Methods Reported by
Lammertyn and
colleagues20,21
Type of cells cultured Primary cells Immortalized cell lines Immortalized cell lines
Pattern of hydrophilic
sites useful for passive
dispensing?
Yes Yes Yes
Location of
hydrophilic sites
Top plate Bottom plate Bottom plate
Hydrophilic site
format
Exposed regions of
ITO surrounded by
Teflon-AF
Spots of adsorbed
fibronectin on a Teflon-
AF surface
Spots of adsorbed poly-
L-lysine on a Teflon AF
surface
Device format for
droplet movement
Right-side up Right-side up Right-side up
Device format for cell
culture
Upside down Right-side up Right-side up
Maximum duration of
cell culture
1 week 2 weeks 3 days
Demonstration of co-
culture
Yes No No
62
Chapter 4. Microgels on-demand
Summary
Three-dimensional (3D) hydrogel particles are enabling technologies in disparate fields ranging
from medical diagnostics to photonics, and they are finding use in fundamental studies in self-
assembly, rheology, and 3D cell culture. Unfortunately, most techniques used for forming 3D
hydrogels are limited to spherical particles and are ‘single pot’ methods in which individual gels
are not addressable after formation. Furthermore, for many applications, it would be useful to be
able to form arrays of gel particles bearing mixtures of constituents and/or are formed from
composites of different gel materials. In response to this challenge, we introduce a digital
microfluidic method for ‘on demand’ formation of arrays of microgels bearing arbitrary
geometries, contents, and shapes. Upon formation of the gels, each particle is individually
addressable for reagent delivery and analysis. We demonstrate the utility of the method for 3D
cell culture and higher order tissue formation by implementing the first sub-microlitre
recapitulation of 3D kidney epithelialization. The new method allows for culture and analysis of
these delicate structures with high success rates relative to conventional techniques relying on
multiwell plates. We anticipate that this platform will provide novel opportunities exploiting
arrays of individually addressable hydrogels of arbitrary shape and size for a wide range of
applications.
63
4.1 Introduction
Precision polymer microgels with complex geometries can support self-assembly for tissue
engineering,83
bar-coding for chemistry,84
and flexible geometric arrangements for forming
photonic crystals.85
A number of strategies relying on microchannels have been developed for
microgel formation including enclosed channels with continuous flows86–88
and two-phase
systems consisting of droplets in a carrier fluid.89,90
Surface tension effects have restricted two-
phase flow systems to the formation of monodisperse spherical, disc, or rod-shaped solids
produced by either photo-initiation or thermal cross-linking at a T-junction or by flow-
focusing.52,53,91
After formation, phase separation requires compatible chemistries to isolate
microgels from the immiscible phase. Recently the Doyle group54,92
improved upon these
techniques, exploiting single-phase stop-flow lithography in the formation of microgels in a
range of geometries. This is important, as arbitrarily shaped microgels (rather than spheres) are
useful for providing novel insights on of the relationship between the microenvironment and cell
fate and behavior.93
The Doyle method has proven particularly useful for UV initiated cross-
linked polymers; however, the method excludes chemically and thermally cross-linked polymer
systems which are important for three-dimensional cell culture. Further, all of the systems
described above are ‘single pot’ methods, such that after formation, the microgels are not
individually addressable.
Motivated by the need to overcome the challenges of polymer compatibility and individual
microgel addressability, we sought to exploit recent developments in reagent dispensing and the
handling of solids on digital microfluidic (DMF) devices for microgel formation. DMF liquid
handling is an emerging alternative to the paradigm of enclosed microchannels.14
This
64
technology facilitates electrostatic manipulation of discrete nano- and micro-litre droplets across
open electrode arrays providing the advantages of single sample addressability, automation, and
parallelization. Variations of this platform have been demonstrated for a broad range of
applications including proteomics,59,94
cell culture and analysis,41,42,72
immunoassays,95,96
chemical synthesis,97,98
and on-chip lasing.64
Recently, the unique geometry of DMF has been
exploited for handling and addressing of three-dimensional solids such as paper discs for genetic
screening,55,56
polymer monoliths for sample extractions,57
and agarose discs for scaffolding
applications.59,99
Here we recognized the potential for DMF to address obstacles to microgel
formation: (1) on-demand hydrogel formation, (2) flexible hydrogel geometries, (3) single
hydrogel addressability, and (4) compatibility with UV, chemical, or thermally cross-linked
polymer systems. Finally, we propose that the combination of digital microfluidics and hydrogels
may represent a useful new tool for three dimensional cell culture, an important emerging
technique for in vitro biology44
that is not widely used because of experimental complexity
(diffusion into 3D matrices is slow and inconvenient for automation) and hydrogel fragility (3D
matrix materials break down upon repeated handling and fluid exchange).
We introduce here the first method for forming “microgels on-demand,” in which individually
addressable three-dimensional (3D) gel structures of varying sizes, shapes, and compositions can
be formed in situ. These advances have the potential to overcome current challenges in complex
microgel formation and addressability, with particular emphasis on 3D cell culture for drug-
screening with improved physiological relevance.
65
4.2 Methods
4.2.1 Reagents
Unless stated otherwise, general-use chemicals were from Sigma Aldrich (Oakville, ON,
Canada) or Fisher Scientific Canada (Ottawa, ON, Canada), antibodies, fluorescent dyes, and,
cell media components were from Invitrogen/Life Technologies (Burlington, ON, Canada), and
photolithography reagents were from Rohm and Haas (Marlborough, MA). Deionized (DI) water
had a resistivity of 18 MΩ·cm at 25°C.
4.2.2 DMF device fabrication
Digital microfluidic devices were fabricated using standard photolithography and metal etching
as detailed previously.42
The bottom-plate device design featured an array of 2.2 mm × 2.2 mm
chromium actuation electrodes and also included an array of five 1 mm diameter optical
windows (i.e., circular regions free from chromium) with 9 mm between each window. Each
window straddled the interface between two actuation electrodes. DMF device top-plates bearing
hydrophilic sites were formed by performing a Teflon liftoff procedure on ITO coated glass
substrates as detailed in chapter 242
The sites were 1-2 mm in diameter and were either circular,
star, heart, or diamond in geometry. The diameter for non-circular shapes was defined as the
diameter of the smallest circle that would enclose the feature.
4.2.3 DMF device assembly and operation
Digital microfluidic devices were assembled with an ITO–glass top plate and a chromium-glass
bottom plate as shown in
Figure 4-1. The two plates were joined by stacking two, three, or four layers of double-sided
tape (each layer ~80 µm), and were aligned such that the edge of the top plate was adjacent to the
66
outer-edges of the reservoir electrodes on the bottom plate. Care was taken to align top and
bottom plate features vertically (windows on the bottom plate and hydrophilic sites on the top
plate). A driving potential of 120 VRMS was generated by amplifying the square wave output of a
function generator (Agilent Technologies, Santa Clara, CA) operating at 10 kHz. Reagents were
loaded and dispensed, moved, and merged as described previously.42
In all DMF experiments,
reagent solutions were supplemented with 0.02% Pluronics F68 except for sol-state Geltrex that
was supplemented with 0.02% Pluronics F127.74
Agarose was not supplemented with Pluronics.
4.2.4 Hydrogel pillar formation and addressing
Devices bearing droplets were imaged with a CCD camera (Basler, Ahrensburg, Germany)
mounted on a fluorescence equipped stereomicroscope (Leica, Wetzlar, Germany). Geltrex was
prepared by 1:1 dilution in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with
10% FBS and 0.02% F127. The solution was maintained on ice until device loading. Low gelling
temperature agarose hydrogels were prepared as 1-6% w/v solutions in DI water by microwaving
the solution for 30 seconds prior to device loading. In some cases sol-state hydrogel solutions
were supplemented with 10 µM fluorescein or with a suspension of fluorescently labelled (green,
yellow, or red) 10 µm dia. microspheres. To form hydrogel pillars, 5 µL of sol-state gel solutions
were loaded into device reservoirs, then one or two droplets were actively dispensed and then
manipulated across the patterned hydrophilic sites where sub-droplets were generated by
hydrophobic-hydrophilic interactions.42
Devices containing Geltrex and agarose sub-droplets
were cross-linked to form pillars by incubation for 1-4 hours at 37 ºC in a humidified chamber,
or at room temperature, respectively. After formation, individual reagents (e.g., fixatives,
permeabilizers, dyes, etc., as described below) were delivered to individual gel pillars by loading
67
Figure 4-1: Digital microfluidic device geometry. (A) Exploded view of device, comprising a
bottom plate with patterned electrodes and a top plate bearing patterned hydrophilic sites. (B)
Side-view of device, not to scale. (C) Schematic depicting principle of device operation.
68
the appropriate mixture into a reservoir, actively dispensing a 0.4 L to 2.2 L droplet onto the
array of electrodes, and then passing the droplet across the pillars (passively exchanging the
contents of the pillar).
4.2.5 Diffusion analysis and modelling
Hydrogel pillars were formed from Geltrex and agarose in 1.5 mm dia. circular virtual
microwells. After two hours of cross-linking at room temperature in a humidified flask, a droplet
of either fluorescein (10 µM in DI water) or fluorescein isothiocyanate conjugated dextran (4
kDa or 40 kDa, 25 mg/mL in DI water) was passed across each hydrogel in a temperature-
controlled imaging suite at 23 ºC. Each condition was repeated in triplicate. Fluorescent images
were recorded at five frames per second (FPS) for up to three hours. Fifty frames were selected
for analysis from each experiment between the onset of diffusion and saturation. Image analysis
was performed in ImageJ (http://rsb.info.nih.gov/ij/) by integrating the pixel intensity in a ~1.3
mm dia. circular region of interest (ROI) within the hydrogelfor each frame over the
experimental observation time. The apparent diffusion coefficient (Da) was determined by an
adaptation of the method reported by Axelrod et al.100
for fluorescence photobleaching recovery.
This model assumes no source density, uniform pore size, and no evaporation. Briefly, integrated
pixel intensities were normalized to the initial fluorscence intensity of the carrier droplet. These
were then plotted with respect to time and the plateau intensity (A) was determined. A
flourescence saturation half-point (τ1/2) was defined as the time required for the normalized
intensity to reach ½A. The time constant for diffusion (τ) was determined by:
69
2/1
)2
1ln(
Eqn. 4.1
An empirical curve was fit to the experimental data:
))exp(1()(
tAxtf Eqn. 4.2
And the apparent diffusion coefficient was set to, where r is the radius of the hydrogel:
2/1
2
4
88.0
rDa Eqn. 4.3
A numerical simulation was performed using COMSOL Multiphysics (http://www.comsol.com/).
A two-dimensional diffusion model was implemented with boundary conditions set to a constant
concentration and the respective diffusion coefficients from Eq. 3. These experiments were
performed with ~160 µm spacers between plates.
4.2.6 Composite hydrogel formation
To build composite hydrogels a cross-linked microgel was formed initially as described above
from either Geltrex or agarose. Secondary hydrogel structures were then formed by manipulating
sol-phase hydrogel droplets to surround the initial structures. These were then cross-linked by
room temperature incubation and imaged under UV illumination with a digital camera. These
experiments were performed ~160 µm spacers between plates.
70
4.2.7 Cell Culture
Marbin Darby canine kidney (MDCK) epithelial cells were kindly provided by Dr. N. Tufenkji
(McGill University). MDCKs were cultured in DMEM supplemented with 10% fetal bovine
serum (FBS), 100 U∙mL-1
penicillin, and 100 µg∙mL-1
streptomycin.
4.2.8 Cell viability and cell distribution
MDCK cells were prepared at 1 106 cells∙mL
-1 in culture media then diluted by 50% in 4ºC
Geltrex solution. 5 µL aliquots of this suspension were loaded and droplets were actively
dispensed from reservoirs and then driven across 1.5 mm circular hydrophilic sites to form
hydrogel pillars bearing suspended cells. Each device was then inverted, placed within a petri
dish containing kimwipes saturated with DI, and incubated at 37 °C in a humidified incubator
containing 5% CO2 for two hours. Cells were then stained by manipulating PBS droplets
supplemented with Live/Dead reagents (Life Technologies) and 5 µg/mL Hoeschst 33342 to
exchange the solution with the pillars. Cells in the hydrogel pillar were imaged by confocal
microscopy (Zeiss LSM700, Carl Zeiss, Toronto, Canada) at 6 µm increments along the z-axis.
Viability after cross-linking was enumerating the number of live (green) fluorescent cells, and
dead (red) fluorescent cells. Viable spheroids were evaluated on days two through four by light
microscopy, where viable spheroids have obvious spheroid shapes and non-viable spheroids
appear disorganized and/or blebbed. Cell distribution was determined by recording the z position
of the largest diameter portion of each individual nucleus throughout the hydrogel. Image
analysis was performed using Zen Light Edition software (Carl Zeiss, Toronto, Canada). These
experiments were performed with ~240 µm spacers between plates.
71
4.2.9 Epithelialization experiments
MDCK cells were seeded in pillars on devices as described above. Cells seeded on separate
devices were fixed at 24, 48, and 96 hours after initial seeding by passing three droplets of 10%
v/v Histochoice Tissue Fixative in DI water to each hydrogel pillar then incubating for 30
minutes at room temperature. Cells were then permeabilized by passing three droplets of
permeabilization solution (PS, 0.5% Tween 20 diluted in PBS) across each hydrogel pillar then
incubating for 90 minutes at room temperature. Prior to staining, three droplets of PS were
driven across the hydrogel with 10 minute incubation at room temperature between each droplet.
Each gel pillar was addressed with a droplet containing staining reagents and then incubated at
room temperature for up to 3 hours in the dark. Prior to imaging, three droplets of PS were
passed across each gel pillar with 10 minute incubation at room temperature between each
droplet. Spheroid diameter was determined by light microscopy and cell number per spheroid
was determined by enumerating nuclei per spheroid by epifluorescent microscopy. Confocal
microscopy was used to image cells stained with Alexa Fluor 488® phalloidin and Hoescht
33342. These experiments were performed with ~240 µm spacers between plates.
4.2.10 Comparison to conventional fluid handling
MDCK spheroids were formed in 3D culture as described above on DMF or in 96-well plates.
On day four of culture ten randomly selected spheroids for each condition were imaged by light
microscopy, followed by reagent exchange. For DMF this comprised manipulation of five
separate 1.2 µL droplets of PBS across the microgel containing spheroids. For 96-well plates, a
Biomek FX system (Beckman Coulter, Inc., Fullerton, CA, USA) was used to deliver 100 µL of
72
PBS and then remove 100 µL of PBS at a rate of 50 µL/s from each well five times. Spheroids
were then re-imaged post manipulation and compared to spheroids pre-manipulation by a
measure of circularity using ImageJ to indicate the level of spheroid deformation. Spheroids
were identified by automated microscopy homing (to pre-selected x, y coordinates) and the use
of other local structures as fiduciary marks. Circularity was determined by drawing an ROI along
the mid-point of each spheroid (before and after manipulation) and calculated as:
24
perimeter
areayciruclarit Eqn. 4.4
Deformation was determined as the absolute percent change in circularity before and after
manipulation, where 100% would indicate complete spheroid disintegration. These experiments
were performed with ~240 µm spacers between plates.
4.3 Results and discussion
4.3.1 Microgels on-demand
To form microgels on-demand, droplets of sol-phase hydrogel material are manipulated across
hydrophilic sites on a DMF device (
Figure 4-1,Figure 4-2), where sub-droplets are generated by line-pinning. This technique has
been used previously for non-gelling fluids (known in DMF as “passive dispensing”42
and in
non-DMF methods as “surface energy traps”101
), but this is the first application to exploit this
technique for hydrogels. As shown in Figure 4-2, in this method, sub-droplets of sol-form
hydrogel materials deposited onto hydrophilic sites can be thermally cross-linked to form
73
Figure 4-2: Microgels on-demand. (A-H) Frames from a movie (top) and side-view schematic
(bottom) depicting a sol-state hydrogel droplet containing a fluorescent dye actively dispensed
from a reservoir (A) and then electrostatically manipulated to a patterned hydrophilic site (B &
C). A second droplet is then actively dispensed (D) and passed across the hydrophilic site (E-H).
Scale bars = 2 mm. When the sol-state droplet is passed across the hydrophilic site a sub-droplet
is generated in the shape of the hydrophilic site. Upon crosslinking, each droplet forms a solid
gel pillar. Various geometries can be generated and visualized with epifluorescent
stereomicroscopy (I & J) or confocal microscopy (K & L). A movie depicting the formation of
an array of different microgel shapes can be found online in the supplementary information.
Scale bars = 1 mm.
74
microgel pillars. The two materials used here were Geltrex (a reconstituted basement membrane
complex of extracellular matrix (ECM) proteins102
that remains in sol-phase at temperatures
below 4°C) and agarose (a gel that remains in sol-phase at temperatures above 42°C), but we
propose that similar methods should be applicable to a wide range of hydrogel materials. In the
only previous work that we are aware of describing the use of hydrogels on DMF, gel discs were
formed off-device then manually positioned on device (without passive dispensing).59,99
Though
useful, the previously described method is limited to circular structures, cannot be implemented
on-demand, and is incompatible with soft gel materials such as those that are based on collagen
(like Geltrex). We found that the microgel on-demand method could repeatedly produce
columnar hydrogels formed from both agarose and Geltrex with dimensions ranging from 1000-
2000 µm diameter and 75-225 µm in height (~60-700 nL) with volumetric precision varying
from 0.3% to 8.1% (Figure 4-3).
The formation of gels with arbitrary shapes has previously only been possible using UV cross-
linkable hydrogels; however, UV initiators are often cytotoxic and therefore are typically not
ideal for cell applications.54
Here, we report that by simply altering the shape of the hydrophilic
site for passive dispensing, star-, heart- triangle-, and diamond-shaped microgels were readily
formed using the microgel on-demand technique (Figure 4-2,I-J, Figure 4-4). We further
confirmed that these microgels were conformal through the vertical axis by confocal microscopy
(Figure 4-2,K-L). Given the extensive evidence in the literature demonstrating the effects of two-
dimensional geometry on cell function and the emerging evidence that 3D microenvironmental
geometry plays a role in tissue morphogenesis,93
we propose that the microgel on demand
method will be a useful new tool for probing the role of microenvironment geometry on cell and
75
Figure 4-3: Precision of microgel on-demand formation and reagent exchange. Frames from a
movie (A-D) depicting sol-state Geltrex being passively dispensed to form a sub-droplet. Scale
bar = 2 mm. The diameters of sol-state Geltrex were measured by brightfield microscopy and
compared to the diameters of circular hydrophilic sites (E). The asterisk indicates the single
condition tested for which sub-droplets failed to form. Experiments were performed in triplicate
and error bars indicate ±1 S.D.
76
Figure 4-4: Formation of microgels in varied geometries. Frames from movies depicting the
formation of star (A-D), triangle (E-H), and diamond-shaped microgels (I-L). Microgel density
can be increased by adjusting the pitch of patterned hydrophilic sites on the device top-plate (M).
A movie depicting the formation of an array of microgels in different shapes can be found in the
online supplementary information. Scale bars = 2 mm.
77
tissue behavior and function.93,103
Finally, microgels (with sufficient rigidity) formed on demand
are accessible for off-chip handling and analysis. Agarose gels prepared in this manner were
imaged by environmental scanning electron microscopy (ESEM) (Figure 4-5), revealing that
they possess the honeycomb structure typical of agarose microgels formed by other means.104
In
contrast, Geltrex microgels were not sufficiently rigid to maintain their form after top-plate
removal.
4.3.2 Reagent exchange and diffusion into hydrogels
A key benefit of DMF liquid handling is the independent addressability of electrodes and thereby
individual droplets. This feature allows for the targeted delivery of reagents, an important
attribute for the maintenance of microgel hydration and cell viability, in the case of three-
dimensional cell culture, through media exchange. As shown in Figure 4-6A-D, droplets of
reagents were individually delivered to (and removed from) microgels by applying an
appropriate sequence of driving potentials. This dispensing mechanism is reproducible and
precise, ensuring that each passage delivers equal amounts of fresh reagent with volumetric
precision varying from 0.29% to 3.29% (Figure 4-7).
To examine diffusion into DMF-generated microgels we used fluorescein (0.3 kDa), 4-, and 40-
kDa FITC-dextrans as surrogates for the small molecules, growth factors, and cytokines often
used in cell culture and stimulation.105
Specifically, droplets containing these tracers were
78
Figure 4-5: Off-chip microgel analysis. Agarose (5% w/v) hydrogels were formed on device
then allowed to crosslink at ambient temperature. Top-plates were removed to expose the
hydrogel allowing for imaging by environmental scanning electron microscopy.
79
delivered to Geltrex microgels formed on-demand and diffusion was tracked by fluorescence
imaging. Upon delivery of a droplet, a film of tracer was observed to form around the microgel
and diffuse towards the centre. As shown in Figure 4-6E-F, the three tracers had apparent
diffusion coefficients (Da) in Geltrex of Da = 10×10-10
m2/s (fluorescein), 7.2×10
-11 m
2/s (4-kDa
FITC-dextran), and 3.1×10-11
m2/s (40-kDa FITC-dextran), respectively. For comparison,
diffusion was modeled numerically, which was found to be consistent with the experimental
results (Figure 4-6E). The deviation between modeled and experimental results could be
attributed to the depleting level of Co within the absorbed ring of tracer over time. Further, the
experimental values are consistent with the limited number of diffusion coefficients, D, reported
in the literature for similar (but not identical) systems: fluorescein in water106
(D = 4.9×10-10
m2/s) and matrigel107 (D = 4.2×10-10
m2/s), 8-kDa FITC-dextran in agarose108
(D = 8×10-11
m2/s),
20 kDa FITC-dextran in water109 (D = 8×10-11
m2/s) and agarose105 (D = 4.2×10-11
m2/s). The Da
values determined here suggest that delivery of small molecules, growth factors, or cytokines for
hydrogel based cell culture will be fast, requiring only ~minutes to reach cells embedded within
the gels (in comparison to ~hours for macro-scale 3D cell culture in multiwell plates).
4.3.3 Pitch, Multi-Component Arrays, and Composite Microgels
In most of the data described herein, the spacing between gels (or “pitch”) was 9 mm, which
enabled the formation of five gels at a time on the ~microscope slide sized substrates used here.
But the spatial density of hydrogel structures can be readily increased by reduced pitch and size
of individual hydrogel structures (Figure 4-8A). In addition, the new technique facilitates the
formation of combinatorial arrays of microgels containing unique constituents. To highlight this
80
Figure 4-6: Reagent exchange in hydrogel pillars. (A-D) Frames from a movie depicting a
droplet of fluorescein being electrostatically manipulated across a (transparent) microgel pillar
and subsequent diffusion of fluorescein into the microgel. Scale bars = 2 mm. (E) Experimental
(red dots), empirical fit (dashed blue line), and simulation (solid black line) of diffusion profiles
for fluorescein into Geltrex microgels. The inset shows a fluoresecent image (left) and heat-map
simulation of concentration (right) of fluorescein at the half-saturation point (1/2). (F) Apparent
diffusion (Da) coefficients measured for fluorescein, 4 kDa FITC-dextran, and 40 kDa FITC-
Dextran into Geltrex. Error bars indicate 1 S.D.
81
Figure 4-7: Precision of reagent exchange across microgels. Frames from a movie (A-D)
depicting a droplet of reagent passed across a microgel. The diameters of dispensed droplets
were measured by brightfield microscopy and were compared to the diameters of circular
hydrophilic sites (E). The asterisk indicates the single condition tested for which sub-droplet
failed to form. Experiments were performed in triplicate and error bars indicate ±1 S.D.
82
capability, we generated arrays of hydrogel mixtures of red, yellow, and green constituents,
forming a gradient of compositions (Figure 4-8B). These arrays were generated combinatorially
on device from three stock sol-phase solutions (each containing only one colour of microsphere)
through a series of droplet merging, mixing, and splitting steps. We propose that this technique
may be useful in the future to screen arrays of different hydrogels compositions and multiple cell
types for studies in tissue engineering and cell-ECM interactions.110,111
Finally, the technique can
be used to individually address microgels to form composite materials by forming a first
structure from either Geltrex or agarose, followed by the formation of a secondary hydrogel
structure engulfing the initial structure (Figure 4-8C-E). Composite microgel structures have
been reported previously, but they have been limited to spherical geometries.112
We propose that
the new capacity to form arbritrarily shaped composite gels may be useful for forming unique
environments for multi-scaffold chemistries and cell culture.110,113
4.3.4 DMF recapitulation of epithelialization
With the ability to form microgels on-demand by DMF and to address them independently for
reagent exchange, we tested this system for three dimensional cell culture using the well-
characterized model of kidney epithelialization.44,45,114
MDCK cells were suspended in sol-phase
Geltrex and this suspension was used to form microgels on-demand, as above (Figure 4-9A).
Cells cultured in microgels remained viable (~100%, data not shown) for up to 5 days, with
delivery of fresh media droplets at 24-h intervals. The effects of electrostatic actuation on cell
health were not explicitly evaluated here, but previous studies with 2D adherent or suspension
cell culture have reported no or negligible effects on cell viability/morphology67,72
or gene
expression115
when compared with non-actuated cells (because the electrical field drops across
83
Figure 4-8: Pitch, multicomponent arrays, and composite microgels. 32-plex arrays of microgels
were formed from Geltrex supplemented with fluorescein for visualization (A). Combinatorial
arrays of Geltrex microgels were formed on device containing mixtures of red, yellow, and/or
green microspheres (B). Gradient bars indicate the percent compositions of each respective
microsphere, from 100% red at the left and 100% green at the right. Composite microgels
containing fluorescent microspheres were formed with inner agarose and outer Geltrex layers (C:
green filter, D: red filter, E: composite image). Scale bars = 2 mm.
84
the insulating layer rather than droplets containing cells). As shown in Figure 4-9B and Figure
4-10, the cells evaluated here were found to be suspended within the hydrogel matrix with a
slight bias to the bottom of the device, with no cells adhered to either top or bottom plates.
MDCKs are known to form hollow spheroid structures when cultured in collagen or matrigel
matrices, making it possible to study epithelialization and primitive tissue formation in vitro.45
To evaluate whether cells cultured in microgel pillars on DMF could recapitulate this model,
MDCKs were maintained on-device with daily media exchange for one through four days. On
each day microgels were fixed, permeabilized and stained for actin and nuclei (all steps
implemented as in Figure 4-6, allowing for delivery of reagents in minutes). The inherent
addressability of DMF makes this process straightforward, allowing for one gel to be stained and
imaged on day one (while maintaining cell growth in the other gels), a second gel to be stained
and imaged on day two, and so on. Cells formed multicellular clusters in DMF-microgels over
the first 72 hours (Figure 4-9C-F) with a visible lumen observed after 96 hours. Nuclei were
distributed around well-formed lumina with actin accumulation at the luminal edge, an indication
of cellular polarization. The sizes of spheroids cultured within sub-microlitre gel pillars were
similar to those grown in 96-well plates in 100 L aliquots (Figure 4-11).
The results in Figure 4-9 represent the first demonstration of MDCK spheroid formation in sub-
microlitre hydrogels (using microfluidics or any other format). Note that these spheroids are
higher-order structures that form spontaneously as a result of cell directed assembly; these
structures are quite distinct from those in which cells are grown on pre-formed hydrogel targets
85
Figure 4-9: Higher order tissue formation and handling in microgel pillars. Spheroids are
generated five-plex in microgels on DMF (A). The inset highlights one microgel containing
spheroids. Confocal microscopy image stack of MDCK cells in a Geltrex pillar demonstrates cell
distribution throughout the z-axis (B). Images depicting MDCK spheroid formation in a gel pillar
over four days in culture (C-F). Cells were stained for actin with phalloidin (green) and nuclei
(blue). On day four lumen formation was observed. Hydrogels bearing 4-day MDCK spheroids
were subjected to media exchange five consecutive times either by automated pipetting or DMF.
Representative bright-field images of spheroids pre- and post-manipulation for both systems (G-
J). Scale bars = 20 µm. Graph of spheroid deformation for five trials evaluating 10 spheroids
each using robotic pipetting into a multiwell plate or DMF (K). The deformation value of each
spheroid is indicated by a blue (DMF) or red (conventional reagent delivery) dot with black bars
representing the mean deformation of each trial. Bar graphs show overall deformation averages
across the five trials for each respective system. Error bars show 1 S.D. In three of the trials,
identified by asterisks, the conventional method dislodged the hydrogels from the well, resulting
in material failure.
86
Figure 4-10: Z-axis distribution of cells in microgels. Graph of distribution of cells in the z-axis
for the data in Figure 4-9B. The number of cells was enumerated at each image plane (~6 µm
thick).
Figure 4-11: Spheroid size. Comparison of spheroid size for three-dimensional MDCK culture in
microscale (DMF, white bars) and macroscale (96-well plate, gray bars).
87
or molds.116,117
One reason for the relative dearth of this type of work in the literature is that
higher order tissue structures are very fragile. The simple act of delivering reagents to 3D cell
structures is known to cause morphological damage51,118
; this is particularly problematic for cells
grown in hydrogel scaffolds with low viscoelastic storage modulus (~10 Pa for 50% Matrigel119
).
We hypothesized that digital microfluidics, which has been previously reported to be useful for
gentle handling of weakly adhered apoptotic cells,72
might also be useful for non-disruptive
reagent delivery to cell constructs grown in 3D microgels.
To characterize and compare the damage induced to MDCK spheroids by reagent exchange, we
performed assays in which day-4 spheroids (cultured either in microgels on DMF devices or in
wells on microtitre plates) were imaged and then five aliquots of PBS were added and removed.
The spheroids were then imaged again, and deformation was determined by calculating the
change in circularity of each spheroid’s two-dimensional projection imaged by light microscopy
(Figure 4-9G-J). As shown in Figure 4-9K, in five separate trials (evaluating 10 spheroids in
each trial), DMF handling of hydrogel bearing spheroids resulted in a nearly ten-fold reduction
in deformation when compared to conventional fluid delivery to cells in mutliwell plates (DMF
7.6%.and conventional 75.9%). Of particular note are three trials in which conventional handling
of the microgels resulted in complete hydrogel displacement (quantified as 100% deformation),
in which spheroids were disrupted to the extent that they could no longer be identified. This
evidence supports the hypothesis that DMF is superior to conventional techniques for gentle
handling of microgels, an important factor in terms of the cost-benefit of 3D cell culture.
88
4.3.5 Conclusion
Digital microfluidic (DMF) microgels on-demand is a new tool for the formation of microgels in
a range of geometries and compositions. Importantly, this is the first method capable of forming
complex and composite hydrogel geometries from thermally cross-linked hydrogels, with
individual addressability of microgel structures. With demonstrated utility for three-dimensional
cell culture, we anticipate this technology will enable future microgel studies across a range of
applications in chemistry, biology, physics, and beyond.
89
Chapter 5. Cardiac microgels
Summary
Three-dimensional (3D) cell culture is attractive because of the ability to better reproduce in vivo
biology in vitro. Unfortunately, 3D cell culture techniques are much less common than their 2D
analogues because the fragility of matrix materials makes them difficult to manipulate using
conventional tools, and because the reagents are expensive. Moreover, most techniques used to
form 3D cell constructs form them and address them in bulk dispersions; ideally, methods could
be developed to evaluate and address each 3D tissue one-at-a-time. Here we describe the
application of ‘microgels on-demand’, a flexible method for micro-scale hydrogel formation for
3D cell culture and analysis, for the culture of functional cardiomyocyte (CM) microtissues. In
this technique, the unique fluid handling capabilities of digital microfluidics (DMF) is leveraged
to generate, address, and maintain sub-microlitre gel pillars that support the growth of 3D
neonatal rat CM structures. Over the course of 5 days, these ‘cardiac microgels’ initiate
spontaneous beating and are independently addressable and responsive when stimulated with
epinephrine. Further we report a new label-free method for analyzing CM activity which may
someday facilitate integration with high throughput processing for screening experiments. This
DMF based technique supports cardiac microgel formation and we anticipate access to this
technology to enable improved biological readouts from in vitro assays at significantly reduced
costs.
90
5.1 Introduction
Cardiac tissue presents a complex three-dimensional environment composed primarily of
contractile muscle cells and fibroblasts. The interaction of cardiomyocytes (CMs) with the
surrounding extracellular matrix is critical for the formation of anchor points that transduce the
mechanical signals needed to promote CM formation of myofibrils and sarcomeres. Cultured
primary CMs are important for the study of cardiac function in vitro. Since the initial isolation of
neonatal rat CMs by Harary and Farley,120
these cells have been well characterized in terms of
their morphological, biochemical, and electrophysiological properties and are commonly used to
evaluate cardiac toxicities of drugs. To date, the majority of studies have focused on CMs
cultured on planar two-dimensional (2D) plastic or glass surfaces. Though the data generated
from these studies has proven invaluable, there remains a significant disparity between 2D in
vitro model systems and observed in vivo responses. For this reason microtissue models for in
vitro assays are being developed to bridge the gap between traditional two-dimensional cell
culture systems and in vivo biology.121,122
Hydrogel-based 3D cell culture is rapidly becoming an important tool in life-science research.44
Hydrogel materials can be derived from inert or animal sources providing for customizable
microenvironments to elucidate or direct cell function and behaviour. These systems have
allowed for significant strides in cell biology through the reestablishment of critical
microenvironmental factors, particularly cell-cell and cell-matrix interactions. To improve the
reliability of data from drug screens and avoid the pursuit of in vivo follow-up studies on false-
positive targets, 3D cell culture presents a potentially ideal system to balance cost and biological
relevance. Unfortunately, hydrogel culture methods remain under-utilized in part because of high
91
reagent costs and challenges in the manipulation and handling of delicate hydrogel materials.
A number of strategies relying on microchannels have been developed to address the challenges
of working with 3D cell structures in hydrogels,47
including enclosed channels with continuous
flows86,87
and two-phase systems consisting of droplets in a carrier fluid.89,90
Microfluidics is
useful for this goal, as it provides the ability to manipulate sub-microlitre volumes of liquid
thereby reducing reagent consumption. Devices used for 3D cell culture and handling have
ranged in complexity from systems with integrated valves48
and off-chip pumps52,123
to passive
perfusion systems50,51
exploiting gravity and surface tension forces to drive reagent exchange.
Microchannel devices have been demonstrated for 3D culture of primary CMs, to pattern
CMs,124
to form two-dimensional culture systems on hydrogel coated surfaces,125–127
or to enable
3D spheroid culture.128
Though these examples successfully form functional cardiac microtissues
in their respective formats, they are ‘single pot’ systems, and inherently lack the ability to
individually address each microtissue during experimentation.
To address this challenge we recently developed a digital microfluidic method (DMF) called
‘microgels on-demand’, introduced in Chapter 4, that allows for the formation of microgels in
customizable geometries from a broad range of hydrogel materials. This method was
successfully implemented in the recapitulation of a model of kidney epithelialization. DMF
based biological assays facilitate the manipulation of sub-microlitre volumes of reagent across
arrays of electrodes and have been used for a broad range of cell based assays with immortalized
and primary cells cultured in both 2D and 3D formats (several examples of which are presented
in chapters 2, 3, and 4).41,58,67,72
While DMF actuation has been demonstrated to exert little if
92
any transcriptional effect on cells,115
the small volumes required (< 1 µL) are particularly
attractive for primary cell culture because of the inherently limited amounts of available material.
Contraction frequency is well-known property that is used classically to evaluate CM function –
with an emphasis in microtissue models to achieve physiological rates of contraction in
vitro.121,122,124,126
Unfortunately, the measurement of contraction frequency is significantly more
challenging in 3D culture than in 2D culture. Moreover, traditional methods extrapolate the
behaviour of tissue constructs from the morophological changes in single cells. To circumvent
these challenges, we developed a whole frame method to quickly ascertain ‘CM activity’ within
3D constructs. This metric can be correlated with CM beat frequency, providing a rapid, simple
and novel high-level analysis technique in the determination of 3D CM activity. We anticipate
the methods reported here, if widely applied, will enable laboratories to improve efficiency in
screening CM based culture systems, while reducing the amount of material required (and thus
the numbers of animals to be sacrificed).
5.2 Methods
5.2.1 Reagents
Unless stated otherwise, general-use chemicals were from Sigma Aldrich (Oakville, ON,
Canada) or Fisher Scientific Canada (Ottawa, ON, Canada), antibodies, fluorescent dyes, and,
cell media components were from Invitrogen/Life Technologies (Burlington, ON, Canada), and
photolithography reagents were from Rohm and Haas (Marlborough, MA). Deionized (DI) water
had a resistivity of 18 MΩ·cm at 25°C.
93
5.2.2 DMF device fabrication
Digital microfluidic devices were fabricated using standard photolithography and metal etching
as detailed previously.42
The bottom-plate device design featured an array of 2.2 mm × 2.2 mm
chromium actuation electrodes and also included an array of four 1.5 mm diameter optical
windows (i.e., circular regions free from chromium) with 9 mm between each window. Each
window straddled the interface between two actuation electrodes. DMF device top-plates bearing
1.5 mm dia. circular hydrophilic sites were formed by performing a Teflon liftoff procedure on
ITO coated glass substrates as detailed in Chapter 2.42
5.2.3 DMF device assembly and operation
Prior to experiments, digital microfluidic top and bottom plates were sterilized in 70% ethanol
for 10 minutes. Excess ethanol was shaken off and devices were permitted to air dry for 30
minutes within a biosafety cabinet. DMF devices were assembled with an ITO–glass top plate
and a chromium-glass bottom plate. The two plates were joined by three layers of double-sided
tape (each layer ~80 µm), and aligned such that the edge of the top plate was adjacent to the
outer-edges of the reservoir electrodes on the bottom plate. Care was taken to align top and
bottom plate features vertically (windows on the bottom plate and hydrophilic sites on the top
plate). An open-source, automated DMF actuation system called “DropBot” (described in detail
elsewhere129
) was used to program and manage the application of driving potentials of 120-140
VRMS generated by amplifying the sine wave output of a built-in function generator operating at
10 kHz. Reagents were loaded and dispensed, moved, and merged as described previously.42
In
all DMF experiments, reagent solutions were supplemented with 0.02% Pluronics F68 except for
sol-state Geltrex that was supplemented with 0.02% Pluronics F127.74
Imaging during droplet
manipulation was performed with a built-in webcam.129
94
5.2.4 Cell culture and cardiac microgel formation/addressing
CMs were isolated from neonatal (24-48 hour old) Sprague-Dawley rat hearts as described
previously130
and cultured in DMEM supplemented with 10% FBS, 1% HEPES, 100 U/mL
penicillin-streptomycin, 0.02 U/mL insulin, and 5 µg/mL vitamin C. For microscale 3D cell
culture experiments, cell suspensions (5-20×106 cells/mL) in sol-state gel were prepared in a
50% Geltrex solution in CM media. To form microgels, an 8 µL aliquot of sol-state gel/cell
suspension was loaded into device reservoirs. The volume of the entire media reservoir was
dispensed onto the platform and then translated across the patterned hydrophilic sites to generate
sub-droplets by hydrophobic-hydrophilic interactions.42
Devices were inverted then placed into a
Petri dish in a 150 cm2
tissue culture flask with re-closable lid containing DI water. The flask
was then incubated for 4 hours at 37 ºC in a humidified chamber to allow for cross-linking. After
cross-linking, devices were returned to upright state, reconnected to DropBot, and fresh media
was delivered to the microgels by single droplet passive exchange. Subsequent media exchanges
were performed every 24 hours for up to five days (with devices stored inverted in the incubator
between exchanges). For macroscale 2D cell culture experiments, cells were seeded at 50,000
cells per well on a 96-well plate that had been coated with Geltrex (1:30 dilution in CM media).
Media was exchange at 24 hour intervals after seeding for up to five days.
5.2.5 Cardiomyocyte treatment
Media was exchanged by passive dispensing and cells were maintained in an incubator for 30
minutes prior to collecting a movie of CM activity for 20 seconds (30 frames per second, fps) at
4×, 10×, or 20× magnification with a brightfield microscope (Motic AE31, Japan) equipped with
a black and white camera (Basler A640, Exton, PA) controlled by a custom Labview program
95
(National Instruments, Toronto, Canada). Cells were then treated with either the stimulant
epinephrine (10 µg/mL in CM media) or the fixative histochoice (10% in DI water) by passive
reagent exchange. In the case of epinephrine stimulation, cells were incubated for 20 minutes
followed by imaging as described above. Cells were then washed with fresh media, incubated for
20 minutes, then fixed. In certain experiments, cells were fixed in histochoice (10% in DI), then
imaged for 30 minutes (30 fps) at 10× magnification. Cells cultured in 2D macroscale culture
were treated with similar conditions.
5.2.6 Single cell analysis of cardiomyocyte activity
Cells were imaged by light microscopy at 10× and 20× magnification. Single cell function was
determined by evaluating cell body displacement, contraction frequency, and contraction length
from videos of beating tissues. Specifically, ImageJ was used to determine the regions where
cells were contracted. Subsequent relaxation was determined by cell movement relative to the
original region of interest. The distance of the contraction was measured as the greatest distance
travelled by any single edge of the cell. Twenty cells were evaluated in three independent micro
gels to determine overall functionality.
5.2.7 Cardiac activity coefficient (CAC) for fast analysis
Cells were imaged by light microscopy at 10× and 20× magnification. A custom Matlab
(Mathworks, Nattick, MA) script was written to evaluate changes in pixel intensity across image
sequences. With a maximum observed beat rate of less than 120 BPM, we selected a sampling
frequency to avoid aliasing using the Nyquist criterion:
cs ff 2 Eqn. 5.1
96
where fs is the sampling frequency and fc is the cutoff frequency. To meet this criterion, we
analyzed image frames at a frequency of 6 fps. In the analysis, the Matlab script determined the
mean change in 8 bit pixel intensity between every sample frame (every 0.167 s); we termed this
the cardiac activity coefficient (CAC). This is described by the following equation:
N
II
CAC
N
i
ii
1
Eqn. 5.2
Where I is the image intensity and N is the total number of images. In certain experiments this
was presented as a function of time. In others, the mean pixel intensity over a twenty second
video at 0.167 intervals was determined. These were either collected as a sum of pixel change
over the course of a twenty second sequence, or the average over the course of a longer
experiment. High and low thresholds were determined by finding maximum and minimum
values in each condition. These thresholds were equal for all conditions compared to one
another.
5.3 Results and Discussion
5.3.1 DMF device fabrication and design
Digital microfluidics provides for the electrical manipulation of liquid,14
gas,131
plasma,132
and
sol and gel phase droplets59,99
(presented in chapter 4) across electrode array surfaces. For this
work we implemented a variation of a previously published device design42
with several notable
features: (1) the ground electrode of the top-plate composed of transparent indium tin oxide
97
permits imaging on-device, (2) interdigitated electrodes facilitate droplet movement during
actuation, (3) optical windows patterned on the bottom plate permit light microscopy imaging of
cells cultured on-device, and (4) the hydrophilic sites patterned on the device top-plate allow the
formation of microgels by passive dispensing (Figure 4-1).
5.3.2 On-demand cardiac microgel formation and assay
Two mechanisms of droplet splitting on DMF are described in the literature – active and passive
dispensing, the former relying on electrode actuation to split droplets,14
while the latter exploits
surface heterogeneities to generate sub-droplets by pinning through hydrophilic interactions (as
presented in chapter 2).42
In addition to being useful for high precision dispensing of reagents on
device, surface heterogeneities introduced by protein spotting,67
liftoff patterning,41,42
or
chemical functionalization63,80
have enabled applications including cell culture and DNA
immobilization. In chapter 4 of this thesis, we demonstrated biocompatibility of DMF and
passive dispensing for the formation of microgels on-demand (applied to a representative model
of epithelial morphogenesis). Here, we evaluated the compatibility of a similar method with
screening of functionality in 3D microtissues.
We selected a CM model system as there is great interest in evaluating the functional
characteristics of these cells, including contraction frequency and magnitude.133,134
Myocardial
tissue is organized in three-dimensions in vivo, but the vast majority of in vitro CM studies use
the conventional two-dimensional format.135
In the experiments reported here, we form 3D
microgels on-demand by passive dispensing sol-phase hydrogel material across hydrophilic sites,
98
Figure 5-1: Cardiac microgel formation on DMF. A sol-phase hydrogel droplet containing a
suspension of cardiac myocytes is drawn from a reservoir (A). The droplet is translated across
hydrophilic sites that are positioned (on the transparent top plate) above the optical windows on
the bottom plate (B-F). The white arrows indicate the direction of droplet movement. Scale bar =
4 mm.
99
followed by thermal cross-linking of the hydrogel (Figure 5-1). In this geometry, each microgel
is approximately 400 nL. When cells were seeded in varying densities, we observed two modes
of cellular arrangement throughout the construct. For lower initial seeding densities (up to 10 ×
106 cells/mL), cells were well dispersed throughout the 3D matrix (Figure 5-2A,B); however, at
higher initial seeding densities (20 × 106 cells/mL and above), cells formed aggregates within the
hydrogel matrix (Figure 5-2C).
At 3 or 4 days post-seeding, spontaneous contractions of 3D CM tissue formed from moderate-
density suspensions (10 × 106 cells/mL) were observed, followed by coordinated contractions on
day 5. Further, these constructs exhibited well connected 3D branched CM morphologies
throughout z (or vertical) axis of the microgel as observed by brightfield (Figure 5-3) and
confocal laser scanning microscopy (Figure 5-4, Figure 5-5). When seeded at lower densities (5
× 106 cells/mL) or higher densities (20 × 10
6 cells/mL), no or minimal spontaneous contractions
were observed. Thus, 3D CM constructs formed at 10 × 106 cells/mL were used for the
remainder of the experiments described here.
A functional response assay was performed to determine whether the cardiac microgel pillars
formed on demand were responsive to drug treatment. Specifically, cardiac microgels were
monitored by microscopy with contraction magnitude, duration, and frequency determined for
single cells within the construct (Figure 5-6). CMs evaluated in this manner exhibited a mean (±
1 S.D.) contraction magnitude of 2.4 µm ± 0.3 µm, a contraction duration of 0.11 s ± 0.02 s, and
a frequency of 50 beats per minute (BPM) ± 4.2 BPM. These data are consistent with literature
values for CM beating in vitro, which range from ~30 BPM for CMs cultured in 3D
100
Figure 5-2: Cell distributions in microgels as a function of seeding density. Phase contrast
images of micorgels formed with cells seeded at low (A), moderate (B), or high (C) density. The
cells are visible through windows on the bottom plate of the device. At low and modest densities,
cells are homogenously dispersed through the hydrogel. At high density, cell aggregation is
apparent. Scale bar = 250 µm.
Figure 5-3: Brightfield microscopy of CM microtissues. Images of CMs in a microgel formed
on-demand at 4x (A), 10x (B) and 20x (C) magnification. Images of cells in the same microgel at
20x magnification at z=40 (D), 120 (E), or 200 m (F).Scale bars = 50 µm.
101
Figure 5-4: Confocal images of cardiac microgel on DMF. A cardiac microgel was fixed and
stained (blue for nuclei and red for actin) after five days of culture on DMF. Imaging was
performed at 5× (A), 10× (B), and 20× (C) magnification, with elongated branched cells evident
at each level. Scale bars = 100 µm, 50 µm, and 25 µm respectively.
Figure 5-5: 3D cross-section of cardiac microgel. Cells were fixed and stained on DMF (blue
for nuclei, red for actin). Maximum combined exposure of an orthographic view in (A) z-x, (B)
x-y, and (C) y-z planes, demonstrates a well-connected network of CMs that are distributed in
3D space. Scale bar = 50 µm.
102
collagen/fibrin matrices130
to ~70-90 BPM for 2D cultured CMs.136,137
As cardiomyocytes are
known to be sensitive to the hormone and neurotransmitter epinephrine (EPI), we challenged
cardiac microgels with droplets of EPI and characterized their response: the magnitudes of
contraction, duration, and frequency increased by mean fold-values ± 1 S.D. of 1.9-fold ± 0.3,
1.8-fold ± 0.5, and 2-fold ± 0.2, respectively (Figure 5-6). These observations suggest that DMF
based cardiac microgel assays may be a useful new tool in understanding the functional
consequences of pharmacologic agents.
The microgel-on-demand method seems well-suited for parallel screening of conditions in 3D
cell constructs, as highlighted in Chapter 4. The new CM microgel-on-demand technique
described here joins a small group of other techniques directed to this purpose. For example,
Parker and colleagues122,126
recently described methods to form CM films that bend as a function
of the degree of CM contractions, a technique that should be readily adaptable to multiplexed
screening (by measuring bend frequency). This method is intriguing, but it is limited in that it is
not designed for 3D culture and it does not allow for each film to be addressed individually.
Likewise, Radisic and colleagues121
recently demonstrated a creative wire-based seeding strategy
to form functional cardiac ‘biowires’, but the protocol (as used now) is likely not amenable to
automation of high throughput tissue generation or individual ‘biowire’ treatment and analysis.
We posit that the CM microgel on-demand system presented here is unique in that it permits the
potential formation of multiple cardiac microtissues (e.g., 32 microgels, as in Figure 4-8A),
each of which can be individually addressed with independent treatments. Further, the footprint
of the microgels per unit area is equivalent to the conventional 96-well plate format, which
should facilitate translation to multiplexed screening. But unfortunately, the analysis
103
Figure 5-6: Cardiomyocyte function in microgels. Cardiac microtissues were cultured for four
days on DMF and evaluated before and after exposure to epinephrine. The areas of twenty
individual cells were calculated and plotted with respect to time – representative traces (blue-
unstimulated; red – EPI-stimulated) for one cell are shown in (A). Insets brightfield image shows
the contraction measurements performed on single cells. Black line indicates shape pre-
contraction, yellow line indicates shape post-contraction. These data were pooled and evaluated
for contraction frequency (B), contraction distance (C), and duration (D) for control and
epinephrine stimulated cells. Experiments performed in triplicate and error bars indicate ±1 S.D.
104
methods described here are too slow – new analytical tehcniques are needed to facilitate
efficient, multiplexed screening (as described below).
5.3.3 Higher efficiency analysis of cardiac microgel activity
The data in Figure 5-6 was collected using a standard single-cell image analysis technique that is
widely in CM research,138
but it requires excessive manual intervention for analysis. Specifically,
in our work, evaluating contractions on a cell-to-cell basis required ten minutes of analysis per
cell, which required >6 h to evaluate the two conditions represented in Figure 5-6. With
increasing numbers of conditions this method is prohibitive in terms of time required for
analysis. More importantly, the data is not representative of the tissue as a whole – evaluating the
activities of single cells misses the big-picture understanding of the functional behaviour of the
CM gel as a unit. For this reason we developed a faster method compatible with tissue-level
analysis to be combined with the novel DMF based cell handling technique. To this end, we
examined the ability to analyze CM tissue activity based on simple analysis of bright field
whole-image sequences. In an initial experiment we observed an initially beating cardiac
microgel during and after treatment with fixative. By calculating changes in pixel intensity
between frames, we generated an average intensity change for each image, which we termed the
cardiac activity coefficient (CAC). When CACs were plotted against time (Figure 5-7A) it was
clear that the morphological changes decreased, as is expected for the process of fixation.
Qualitative observation of image sequences demonstrated that with time, contraction frequency
across the microgel decreased until there was no longer any observable movement at 30 minutes
after treatment with fixative. This can be visualized as heat maps of activity corresponding to
105
time points at the beginning, middle, and end of the experiment (Figure 5-7B-D) that confirm
decreases and eventual cessation of contractile activity with exposure to fixative.
To test the CAC as a proxy for average CM contraction frequency, we applied it to the same data
used to generate the single-cell analysis shown in Figure 5-6. Images of unstimulated CM
microgels collected at 10× magnification had a mean (±R.S.D.) CAC of 1.44 ±.02, while
corresponding images of epinephrine-stimulated CM microgels had a mean (±R.S.D.) CAC of
1.58 ±.04 (Figure 5-8A). Each image evaluated the contributions of approximately 400-500 cells
(some perfectly in-focus, many slightly out of focus), and thus represented the behaviour of each
CM microgel as functional unit. The difference between the CACs for these two conditions was
statistically significant (p < 0.01). Heat maps representing activity generated from this data
demonstrate more regions of high activity in epinephrine treated cells than control (Figure
5-8B). Most importantly, this analysis was generated in a matter of seconds; to our knowledge,
this is the first report of such a tool. Moreover, this method allows for the evaluation of 3D tissue
activity as a whole. More work is required to validate and test the CAC method, but we propose
that this preliminary data is promising for the development of CAC as a tool for screening tissue-
level morphological changes.
5.3.4 Conclusion
Here, we demonstrated a DMF based microgel-on-demand method for high efficiency studies of
3D cardiac myocytes. This is the first microfluidic method that permits automated individual
addressability of microtissues for parallel and multiplexed experiments. Further, we developed a
new rapid analysis technique for evaluating the functional beading response of CMs grown in
106
3D. We expect that the implementation of DMF technology within high throughput screening
regimes will not only provide the benefit of reduced reagent consumption and improved
experimental workflow – but will enable physiologically relevant in vitro models of biology.
107
Figure 5-7: Cardiomyocyte activity coefficient. After five days of culture on DMF, a beating cardiac
microgel was treated with a fixative solution and observed at 10× magnification for 30 minutes (A). A heat
map was generated to represent cumulative cardiomyocyte activity coefficients at 1 (B), 15 (C), and 30 min
(D) after exposure to fixative. Heat maps are presented as overlays on phase contrast microscopic images.
Here blue indicates regions of low activity and red indicates high activity. With increased exposure to
fixative over time, reductions of activity are observed (as shown through reduced red regions in the heat
maps).
108
Figure 5-8: Cardiac activity coefficient of stimulated cells. CACs were calculated for control
and epinephrine-treated cardiac microgels, and the response (A) was significantly different by a
one-tailed t-test (p < 0.01). Heat maps representing cardiomyocyte activity generated for control
and epinephrine-stimulated microgels (B,C respectively), and brightfield images used for
analysis (D, E respectively). In these images red regions indicate higher activity, with blue
regions indicating lower activity. Scale bars = 50 µm.
109
Chapter 6. Concluding Remarks and
Perspectives on the Future
6.1 Conclusions
At the outset of carrying out the work described in this thesis, digital microfluidics as applied to
cell culture and analysis was very much in the “proof-of-principle” stage. Moreover, the
techniques being developed were constrained to specialized microfluidics laboratories such as
our own. Through the work presented here we have made significant contributions to (1)
improving DMF device functionality, (2) the development of DMF as a robust platform for 2D
cell culture, and (3) improving in vitro tools for the formation and analysis of 3D microtissues.
Chapter 2 describes the foundational development of hydrophilic site patterning within DMF
devices through the implementation of a fluorocarbon liftoff technique for improved device
functionality. This method overcomes limitations associated with inconsistent and incompatible
protein spotting techniques described in the initial reports of DMF based cell culture. We showed
this method was useful for the implementation of passive dispensing with high precision and
accuracy in both air and oil. The success of this technique is being appropriated in the Wheeler
Laboratory for a diverse array of applications ranging including impedance based cell
quantification139
, single cell analysis,140
immobilized immunoassays, janus nanoparticle
formation,141
and portable systems for in-the-field urinalysis by mass spectrometry.142
Further,
others have appropriated the method to study 2D constraints on cell morphology143
and for
increasing throughput of C. elegans droplet based behavioural assays.144
110
The method presented in Chapter 2 was fundamental to the development of robust DMF based
2D culture of primary cells, as described in Chapter 3. Previously published techniques required
protein spotting of extracellular matrix proteins; however, the previous techniques were
incompatible with these sensitive cell types. The implementation of the fluorocarbon liftoff
technique from Chapter 2 enabled the culture of primary cells and preliminary studies for the
culture of human embryonic stem cells in sub-microlitre droplets. Further we found that these
devices were amenable to integration with standard imaging equipment, permitting brightfield
microscopy and high resolution epifluorescent microscopy. This was the first platform to
integrate cell culture, fixing, and analysis completely on a single device. We demonstrated the
ability to run co-culture assays, furthering the versatility of DMF as a platform for cell biology.
Building on the ability to form discrete liquid structures on device (Chapter 2) combined with
the validation of biocompatibility with sensitive cell types (Chapter 3), we became interested in
exploring DMF as a platform for improved in vitro 3D cell culture models. In Chapter 4 we
present a novel technique termed ‘microgels-on-demand’ that for the first time permits the
formation of complex microscale hydrogel structures from thermally cross-linked hydrogel
materials. In this instance the unique properties of DMF enable a method that has not been
demonstrated with any other technique. With automation we demonstrate the ability to generate
high density individually addressable microgel structures in a wide range of geometries. Further,
the multiplexed liquid handling capabilities of DMF were used to demonstrate the formation of
combinatorial and composite microgels. Finally this system was used to recapitulate a model of
kidney epithelialization within a sub-microlitre droplet, and imaged on-device with confocal
microscopy.
111
The findings of Chapter 4 suggested to us that DMF microgels-on-demand could be used to
produce micro-scale functional tissues. In Chapter 5, we implemented this method for the
formation and assay of cardiac microgels. The tissues formed in this manner require minimal cell
material and presented spontaneous coordinated contractions after only a few days in culture.
Treatment with known accelerants demonstrated that these tissues were responsive and
representative of in vivo cardiomyocytes. Further we developeda new method for analyzing
whole tissue functional data.
6.2 Future perspectives
The specific contributions of this thesis build toward a broader vision of fully integrated micro-
scale systems for biological research. In parallel with other innovations in device fabrication and
automation, we are approaching the reality of autonomous bench top cell culture systems that are
accessible to the basic biologist. These technologies have to potential to disrupt the current state
of biological research workflow as they will enable improved consistency, by reducing manual
intervention, higher throughput, and lower-cost as a function of decreased reagent consumption.
We predict that the microgels-on-demand method presented in this thesis will enable the study of
3D microenvironment geometry influence on cell phenotype and behaviour, thereby improving
knowledge in tissue engineering and bridging the gap between in vitro and in vivo systems.
We are witnessing an ever increasing number of researchers joining the field of digital
microfluidics. Major and emerging players in the pharmaceutical, diagnostics, and electronics
industries are increasingly investing in the development of DMF. Life Technologies has invested
in cell culture applications, Abbot Diagnostics in immunoassays,96
and Sharp Laboratories in
thin film transitor based devices,16
Sofie Biosciences in radiotracer synthesis,98
and Advanced
112
Liquid Logic for dry blood spot analysis145
and PCR.146
The increasing interaction between
academic researchers and industrial players is driving rapid evolution in DMF technology and is
promoting its adoption within the research and clinical environments. With the development of
lower cost PCB15
and paper based devices147
combined with automation strategies, this
technology is well positioned to become accessible to a broad range of users in fields from
chemistry to biology.
With significant progress and interest over the past few decades, microfluidic technologies and
their application to biological questions are developing at an exponential rate. As those outside
the microfluidics community begin adopting these technologies we will find increasingly novel
biological questions to be answered at the micro scale, and find the adoption of these
technologies to have a broad impact at the level of basic research, drug discovery, and clinical
applications.
113
References
1. Schrödinger, E. What is life? The physical aspect of the living cell. (Cambridge
University Press, 1948).
2. Feynman, R. There is plenty of room at the bottom. Engineering and Science 23, 22–36
(1960).
3. Einstein, A. Investigations on the theory of brownian movement. (Dover Publications,
1956).
4. Berg, H. Random walks in biology. (Princeton University Press, 1993).
5. Taylor, G. Dispersion of soluble matter in solvent flowing slowly through a tube. Proc.
R. Soc. Lond. A Math. Phys. Sci. 219, 186–203 (1953).
6. Manz, A., Graber, N. & Widmer, H. M. Miniaturized total chemical analysis systems: A
novel concept for chemical sensing. Sensors and Actuators B: Chemical 1, 244–248 (1990).
7. Lemmo, A. V., Rose, D. J. & Tisone, T. C. Inkjet dispensing technology: applications in
drug discovery. Curr. Opin. Biotechnol. 9, 615–617 (1998).
8. Duffy, D. C., McDonald, J. C., Schueller, O. J. A. & Whitesides, G. M. Rapid
Prototyping of Microfluidic Systems in Poly(dimethylsiloxane). Analytical Chemistry 70, 4974–
4984 (1998).
9. Takatsy, G. The use of spiral loops in serological and virological micro-methods. Acta
Microbiol Acad Sci Hung 3, 191–202 (1955).
10. Squires, T. M. & Quake, S. R. Microfluidics: Fluid physics at the nanoliter scale. Rev.
Mod. Phys. 77, 977–1026 (2005).
114
11. Grimes, A. et al. Shrinky-Dink microfluidics: rapid generation of deep and rounded
patterns. Lab Chip 8, 170–172 (2007).
12. Hong, J., deMello, A. J. & Jayasinghe, S. N. Bio-electrospraying and droplet-based
microfluidics: control of cell numbers within living residues. Biomed Mater 5, 21001 (2010).
13. Choi, K. Digital Microfluidics.
14. Sung Kwon Cho, Hyejin Moon & Chang-Jin Kim. Creating, transporting, cutting, and
merging liquid droplets by electrowetting-based actuation for digital microfluidic circuits. J.
Microelectromech. Syst. 12, 70–80 (2003).
15. Jian Gong. Direct-Referencing Two-Dimensional-Array Digital Microfluidics Using
Multilayer Printed Circuit Board. Journal of Microelectromechanical Systems 17, 257–264
(2008).
16. Hadwen, B. et al. Programmable large area digital microfluidic array with integrated
droplet sensing for bioassays. Lab Chip 12, 3305–3313 (2012).
17. Pellat, H. Mesure de la force agissant sur les diélectriques liquides non électrisés placés
dans un champ électrique. C.R.Acad.Sci.Paris 119, 691–694
18. Lu, H.-W. et al. A Study of EWOD-Driven Droplets by PIV Investigation. Lab Chip 8,
456–461 (2008).
19. Poulos, J. L., Nelson, W. C., Jeon, T.-J., Kim, C.-J. ‘CJ’ & Schmidt, J. J. Electrowetting
on dielectric-based microfluidics for integrated lipid bilayer formation and measurement. Appl.
Phys. Lett. 95, 013706 (2009).
20. Wheeler, A. R., Moon, H., Kim, C.-J. ‘CJ’, Loo, J. A. & Garrell, R. L. Electrowetting-
Based Microfluidics for Analysis of Peptides and Proteins by Matrix-Assisted Laser
Desorption/Ionization Mass Spectrometry. Analytical Chemistry 76, 4833–4838 (2004).
115
21. Shah, G. J. EWOD-driven droplet microfluidic device integrated with optoelectronic
tweezers as an automated platform for cellular isolation and analysis. Lab on a Chip 9, 1732–
1739 (2009).
22. Song, J. H., Evans, R., Lin, Y.-Y., Hsu, B.-N. & Fair, R. B. A scaling model for
electrowetting-on-dielectric microfluidic actuators. Microfluid Nanofluid 7, 75–89 (2008).
23. Pollack, M. G., Fair, R. B. & Shenderov, A. D. Electrowetting-based actuation of liquid
droplets for microfluidic applications. Applied Physics Letters 77, 1725 (2000).
24. Paik, P., Pamula, V. K., Pollack, M. G. & Fair, R. B. Electrowetting-based droplet mixers
for microfluidic systemsElectronic supplementary information (ESI) available: six mpeg videos
showing some mixing schemes used in Fig. 7. See http://www.rsc.org/suppdata/lc/b2/b210825a/.
Lab Chip 3, 28 (2003).
25. Zhou, J. et al. Electrowetting-based multi-microfluidics array printing of high resolution
tissue construct with embedded cells and growth factors. Virtual & Phys. Prototyping 2, 217–223
(2007).
26. Bahadur, V. An energy-based model for electrowetting-induced droplet actuation.
Journal of Micromechanics and Microengineering 16, 1494–1503 (2006).
27. Verheijen, H. J. J. & Prins, M. W. J. Reversible Electrowetting and Trapping of Charge:
Model and Experiments. Langmuir 15, 6616–6620 (1999).
28. Chatterjee, D. Droplet-based microfluidics with nonaqueous solvents and solutions. Lab
on a Chip 6, 199–206 (2006).
29. Kang, K. H. How Electrostatic Fields Change Contact Angle in Electrowetting. Langmuir
18, 10318–10322 (2002).
116
30. Buehrle, J., Herminghaus, S. & Mugele, F. Interface Profiles near Three-Phase Contact
Lines in Electric Fields. Phys. Rev. Lett. 91, (2003).
31. Jones, T. B. An electromechanical interpretation of electrowetting. Journal of
Micromechanics and Microengineering 15, 1184–1187 (2005).
32. Abdelgawad, M., Park, P. & Wheeler, A. R. Optimization of device geometry in single-
plate digital microfluidics. J. Appl. Phys. 105, 094506 (2009).
33. Kang, K. H., Kang, I. S. & Lee, C. M. Wetting Tension Due to Coulombic Interaction in
Charge-Related Wetting Phenomena. Langmuir 19, 5407–5412 (2003).
34. Griffiths, D. J. Introduction to Electrodynamics. (Prentice-Hall, 1999).
35. Jones, T. B., Wang, K.-L. & Yao, D.-J. Frequency-Dependent Electromechanics of
Aqueous Liquids: Electrowetting and Dielectrophoresis. Langmuir 20, 2813–2818 (2004).
36. Choi, K., Ng, A. H. C., Fobel, R. & Wheeler, A. R. Digital microfluidics. Annu Rev Anal
Chem (Palo Alto Calif) 5, 413–440 (2012).
37. Walker, S. W., Shapiro, B. & Nochetto, R. H. Electrowetting with contact line pinning:
Computational modeling and comparisons with experiments. Phys. Fluids 21, 102103 (2009).
38. Sen, P. & Kim, C.-J. Capillary Spreading Dynamics of Electrowetted Sessile Droplets in
Air. Langmuir 25, 4302–4305 (2009).
39. Monroe, C. W., Urbakh, M. & Kornyshev, A. A. Double-Layer Effects in Electrowetting
with Two Conductive Liquids. J. Electrochem. Soc. 156, P21–P28 (2009).
40. Barbulovic-Nad, I., Yang, H., Park, P. S. & Wheeler, A. R. Digital microfluidics for cell-
based assays. Lab Chip 8, 519 (2008).
41. Srigunapalan, S., Eydelnant, I. A., Simmons, C. A. & Wheeler, A. R. A digital
microfluidic platform for primary cell culture and analysis. Lab on a Chip 12, 369 (2012).
117
42. Eydelnant, I. A., Uddayasankar, U., Li, B., Liao, M. W. & Wheeler, A. R. Virtual
microwells for digital microfluidic reagent dispensing and cell culture. Lab Chip 12, 750–757
(2012).
43. Au, S. H., Fobel, R., Desai, S. P., Voldman, J. & Wheeler, A. R. Cellular bias on the
microscale: probing the effects of digital microfluidic actuation on mammalian cell health,
fitness and phenotype. Integr. Biol. (2013). doi:10.1039/C3IB40104A
44. Pampaloni, F., Reynaud, E. G. & Stelzer, E. H. K. The third dimension bridges the gap
between cell culture and live tissue. Nature Reviews Molecular Cell Biology 8, 839–845 (2007).
45. Hall, H. G., Farson, D. A. & Bissell, M. J. Lumen formation by epithelial cell lines in
response to collagen overlay: a morphogenetic model in culture. Proceedings of the National
Academy of Sciences of the United States of America 79, 4672 (1982).
46. Töyli, M., Rosberg-Kulha, L., Capra, J., Vuoristo, J. & Eskelinen, S. Different responses
in transformation of MDCK cells in 2D and 3D culture by v-Src as revealed by microarray
techniques, RT-PCR and functional assays. Laboratory Investigation 90, 915–928 (2010).
47. Huh, D., Hamilton, G. A. & Ingber, D. E. From 3D cell culture to organs-on-chips.
Trends in Cell Biology 21, 745–754 (2011).
48. Unger, M. A., Chou, H.-P., Thorsen, T., Scherer, A. & Quake, S. R. Monolithic
Microfabricated Valves and Pumps by Multilayer Soft Lithography. Science 288, 113–116
(2000).
49. Tan, W. & Desai, T. A. Microfluidic Patterning of Cells in Extracellular Matrix
Biopolymers: Effects of Channel Size, Cell Type, and Matrix Composition on Pattern Integrity.
Tissue Engineering 9, 255–267 (2003).
118
50. Bauer, M., Su, G., Beebe, D. J. & Friedl, A. 3D microchannel co-culture: method and
biological validation. Integr. Biol. 2, 371 (2010).
51. Montanez-Sauri, S. I., Sung, K. E., Puccinelli, J. P., Pehlke, C. & Beebe, D. J.
Automation of three-dimensional cell culture in arrayed microfluidic devices. J Lab Autom 16,
171–185 (2011).
52. Tan, W.-H. & Takeuchi, S. Monodisperse Alginate Hydrogel Microbeads for Cell
Encapsulation. Advanced Materials 19, 2696–2701 (2007).
53. Tumarkin, E. et al. High-throughput combinatorial cell co-culture using microfluidics.
Integr Biol (Camb) 3, 653–662 (2011).
54. An, H. Z., Helgeson, M. E. & Doyle, P. S. Nanoemulsion Composite Microgels for
Orthogonal Encapsulation and Release. Advanced Materials 24, 3838–3844 (2012).
55. Jebrail, M. J. et al. A digital microfluidic method for dried blood spot analysis. Lab on a
Chip 11, 3218 (2011).
56. Shih, S. C. C. et al. Dried Blood Spot Analysis by Digital Microfluidics Coupled to
Nanoelectrospray Ionization Mass Spectrometry. Anal. Chem. 84, 3731–3738 (2012).
57. Yang, H., Mudrik, J. M., Jebrail, M. J. & Wheeler, A. R. A Digital Microfluidic Method
for in Situ Formation of Porous Polymer Monoliths with Application to Solid-Phase Extraction.
Anal. Chem. 83, 3824–3830 (2011).
58. Fiddes, L. K. et al. Hydrogel discs for digital microfluidics. Biomicrofluidics 6, 014112–
014112–11 (2012).
59. Luk, V. N., Fiddes, L. K., Luk, V. M., Kumacheva, E. & Wheeler, A. R. Digital
microfluidic hydrogel microreactors for proteomics. Proteomics 12, 1310–1318 (2012).
119
60. Abdelgawad, M., Freire, S. L. S., Yang, H. & Wheeler, A. R. All-terrain droplet
actuation. Lab Chip 8, 672–677 (2008).
61. Mousa, N. A. et al. Droplet-scale estrogen assays in breast tissue, blood, and serum. Sci
Transl Med 1, 1ra2 (2009).
62. Kirby, A. E. & Wheeler, A. R. Microfluidic origami: a new device format for in-line
reaction monitoring by nanoelectrospray ionization mass spectrometry. Lab Chip 13, 2533–2540
(2013).
63. Malic, L., Veres, T. & Tabrizian, M. Two-dimensional droplet-based surface plasmon
resonance imaging using electrowetting-on-dielectric microfluidics. Lab Chip 9, 473 (2009).
64. Kuehne, A. J. C. et al. A switchable digital microfluidic droplet dye-laser. Lab on a Chip
11, 3716 (2011).
65. Sia, S. K. & Whitesides, G. M. Microfluidic devices fabricated in Poly(dimethylsiloxane)
for biological studies. Electrophoresis 24, 3563–3576 (2003).
66. Delamarche, E., Bernard, A., Schmid, H., Michel, B. & Biebuyck, H. Patterned Delivery
of Immunoglobulins to Surfaces Using Microfluidic Networks. Science 276, 779 –781 (1997).
67. Barbulovic-Nad, I., Au, S. H. & Wheeler, A. R. A microfluidic platform for complete
mammalian cell culture. Lab Chip 10, 1536–1542 (2010).
68. Fouillet, Y. & Achard, J.-L. Microfluidique discrète et biotechnologie. Comptes Rendus
Physique 5, 577–588 (2004).
69. Chen, T., Su, C., Chih, H. & Yang, C. Selective wettability assisted nanoliter sample
generation via electrowetting-based transportation. ASME Conf. Proc. 147–153 (2007).
120
70. Vergauwe, N. et al. A versatile electrowetting-based digital microfluidic platform for
quantitative homogeneous and heterogeneous bio-assays. J. Micromech. Microeng. 21, 054026
(2011).
71. Witters, D. et al. Biofunctionalization of electrowetting-on-dielectric digital microfluidic
chips for miniaturized cell-based applications. Lab Chip 11, 2790–2794 (2011).
72. Bogojevic, D., Chamberlain, M. D., Barbulovic-Nad, I. & Wheeler, A. R. A digital
microfluidic method for multiplexed cell-based apoptosis assays. Lab on a Chip 12, 627 (2012).
73. Garyantes, T. Virtual wells for use in high throughput screening assays. (2003). at
<http://www.freepatentsonline.com/6565813.html>
74. Luk, V. N., Mo, G. C. & Wheeler, A. R. Pluronic Additives: A Solution to Sticky
Problems in Digital Microfluidics. Langmuir 24, 6382–6389 (2008).
75. Knovel Critical Tables. (Knovel, 2008).
76. Chang, Y.-H., Lee, G.-B., Huang, F.-C., Chen, Y.-Y. & Lin, J.-L. Integrated polymerase
chain reaction chips utilizing digital microfluidics. Biomed Microdevices 8, 215–225 (2006).
77. Miller, E. M. & Wheeler, A. R. A Digital Microfluidic Approach to Homogeneous
Enzyme Assays. Analytical Chemistry 80, 1614–1619 (2008).
78. Fair, R. B. et al. Chemical and biological applications of digital-microfluidic devices.
IEEE Design & Test of Computers 24, 10–24 (2007).
79. Wang, Y., Zhao, Y. & Cho, S. K. Efficient in-droplet separation of magnetic particles for
digital microfluidics. J. Micromech. Microeng. 17, 2148 (2007).
80. Malic, L., Veres, T. & Tabrizian, M. Biochip functionalization using electrowetting-on-
dielectric digital microfluidics for surface plasmon resonance imaging detection of DNA
hybridization. Biosensors and Bioelectronics 24, 2218–2224 (2009).
121
81. Fouillet, Y., Jary, D., Chabrol, C., Claustre, P. & Peponnet, C. Digital microfluidic design
and optimization of classic and new fluidic functions for lab on a chip systems. Microfluid
Nanofluid 4, 159–165 (2007).
82. Qiu, Q., Sayer, M., Kawaja, M., Shen, X. & Davies, J. E. Attachment, morphology, and
protein expression of rat marrow stromal cells cultured on charged substrate surfaces. J. Biomed.
Mater. Res. 42, 117–127 (1998).
83. Khademhosseini, A., Langer, R., Borenstein, J. & Vacanti, J. P. Microscale technologies
for tissue engineering and biology. PNAS 103, 2480–2487 (2006).
84. Finkel, N. H., Lou, X., Wang, C. & He, L. Barcoding the microworld. Anal. Chem. 76,
352A–359A (2004).
85. Lu, Y., Yin, Y. & Xia, Y. Three-Dimensional Photonic Crystals with Non-spherical
Colloids as Building Blocks. Advanced Materials 13, 415–420 (2001).
86. Balagaddé, F. K., You, L., Hansen, C. L., Arnold, F. H. & Quake, S. R. Long-term
monitoring of bacteria undergoing programmed population control in a microchemostat. Science
309, 137–140 (2005).
87. Liu, T. et al. A microfluidic device for characterizing the invasion of cancer cells in 3-D
matrix. Electrophoresis 30, 4285–4291 (2009).
88. Eddington, D. T. & Beebe, D. J. Flow control with hydrogels. Advanced Drug Delivery
Reviews 56, 199–210 (2004).
89. Agresti, J. J. et al. Ultrahigh-throughput screening in drop-based microfluidics for
directed evolution. Proceedings of the National Academy of Sciences 107, 4004–4009 (2010).
122
90. Leung, K. et al. A programmable droplet-based microfluidic device applied to
multiparameter analysis of single microbes and microbial communities. PNAS (2012).
doi:10.1073/pnas.1106752109
91. Edd, J. F. et al. Controlled encapsulation of single-cells into monodisperse picolitre
drops. Lab Chip 8, 1262–1264 (2008).
92. Dendukuri, D., Pregibon, D. C., Collins, J., Hatton, T. A. & Doyle, P. S. Continuous-flow
lithography for high-throughput microparticle synthesis. Nat Mater 5, 365–369 (2006).
93. Kloxin, A. M. et al. Responsive culture platform to examine the influence of
microenvironmental geometry on cell function in 3D. Integr Biol (Camb) 4, 1540–1549 (2012).
94. Nelson, W. C. et al. Incubated protein reduction and digestion on an electrowetting-on-
dielectric digital microfluidic chip for MALDI-MS. Anal. Chem. 82, 9932–9937 (2010).
95. Sista, R. S. et al. Heterogeneous Immunoassays Using Magnetic beads On a Digital
Microfluidic Platform. Lab Chip 8, 2188–2196 (2008).
96. Ng, A. H. C., Choi, K., Luoma, R. P., Robinson, J. M. & Wheeler, A. R. Digital
Microfluidic Magnetic Separation for Particle-Based Immunoassays. Analytical Chemistry 84,
8805–8812 (2012).
97. Jebrail, M. J. et al. Synchronized synthesis of peptide-based macrocycles by digital
microfluidics. Angew. Chem. Int. Ed. Engl. 49, 8625–8629 (2010).
98. Keng, P. Y. et al. Micro-chemical synthesis of molecular probes on an electronic
microfluidic device. PNAS 109, 690–695 (2012).
99. Fiddes, L. K. et al. Hydrogel discs for digital microfluidics. Biomicrofluidics 6, 14112–
1411211 (2012).
123
100. Axelrod, D., Koppel, D. E., Schlessinger, J., Elson, E. & Webb, W. W. Mobility
measurement by analysis of fluorescence photobleaching recovery kinetics. Biophys J 16, 1055–
1069 (1976).
101. Zhang, Y. & Wang, T.-H. Full-Range Magnetic Manipulation of Droplets via Surface
Energy Traps Enables Complex Bioassays. Advanced Materials n/a–n/a (2013).
doi:10.1002/adma.201300383
102. Kleinman, H. K. & Martin, G. R. Matrigel: Basement membrane matrix with biological
activity. Seminars in Cancer Biology 15, 378–386 (2005).
103. Albrecht, D. R., Underhill, G. H., Wassermann, T. B., Sah, R. L. & Bhatia, S. N. Probing
the role of multicellular organization in three-dimensional microenvironments. Nat Meth 3, 369–
375 (2006).
104. Gel-forming structures and stages of red algal galactans of different sulfation levels -
Springer. at <http://link.springer.com/article/10.1007/s10811-007-9229-9/fulltext.html>
105. Park, J. H. et al. Microporous cell-laden hydrogels for engineered tissue constructs.
Biotechnology and Bioengineering 106, 138–148 (2010).
106. Rani, S. A., Pitts, B. & Stewart, P. S. Rapid Diffusion of Fluorescent Tracers into
Staphylococcus epidermidis Biofilms Visualized by Time Lapse Microscopy. Antimicrob Agents
Chemother 49, 728–732 (2005).
107. Kim, B. et al. Tuning Payload Delivery in Tumour Cylindroids using Gold Nanoparticles.
Nat Nanotechnol 5, 465–472 (2010).
108. Walker, G. M. et al. Effects of flow and diffusion on chemotaxis studies in a
microfabricated gradient generator. Lab Chip 5, 611–618 (2005).
124
109. Cornelissen, L. H., Bronneberg, D., Oomens, C. W. J. & Baaijens, F. P. T. Diffusion
measurements in epidermal tissues with fluorescent recovery after photobleaching. Skin Res
Technol 14, 462–467 (2008).
110. Tsang, V. L. et al. Fabrication of 3D hepatic tissues by additive photopatterning of
cellular hydrogels. FASEB J 21, 790–801 (2007).
111. Hui, E. E. & Bhatia, S. N. Micromechanical control of cell–cell interactions. PNAS 104,
5722–5726 (2007).
112. Ma, M. et al. Core–Shell Hydrogel Microcapsules for Improved Islets Encapsulation.
Advanced Healthcare Materials n/a–n/a (2012). doi:10.1002/adhm.201200341
113. Liu, T., Lin, B. & Qin, J. Carcinoma-associated fibroblasts promoted tumor spheroid
invasion on a microfluidic 3D co-culture device. Lab Chip 10, 1671–1677 (2010).
114. Elia, N. & Lippincott-Schwartz, J. Culturing Three Dimensional MDCK cells for
Analyzing Intracellular Dynamics. Curr Protoc Cell Biol Unit–4.22 (2009).
115. Au, S., Fobel, R. & Wheeler, A. Genome-level Cell Responses to Digital Microfluidic
Manipulation. Submitted
116. Matsunaga, Y. T., Morimoto, Y. & Takeuchi, S. Molding Cell Beads for Rapid
Construction of Macroscopic 3D Tissue Architecture. Advanced Materials 23, H90–H94 (2011).
117. Du, Y., Lo, E., Vidula, M. K., Khabiry, M. & Khademhosseini, A. Method of Bottom-Up
Directed Assembly of Cell-Laden Microgels. Cell Mol Bioeng 1, 157–162 (2008).
118. Gilbert, P. Personal communication.
119. Zaman, M. H. et al. Migration of tumor cells in 3D matrices is governed by matrix
stiffness along with cell-matrix adhesion and proteolysis. PNAS 103, 10889–10894 (2006).
125
120. Harary, I. & Farley, B. In vitro studies on single beating rat heart cells: I. Growth and
organization. Experimental Cell Research 29, 451–465 (1963).
121. Nunes, S. S. et al. Biowire: a platform for maturation of human pluripotent stem cell-
derived cardiomyocytes. Nat Meth advance online publication, (2013).
122. Agarwal, A., Goss, J. A., Cho, A., McCain, M. L. & Parker, K. K. Microfluidic heart on a
chip for higher throughput pharmacological studies. Lab Chip (2013). doi:10.1039/C3LC50350J
123. Köster, S. et al. Drop-based microfluidic devices for encapsulation of single cells. Lab
Chip 8, 1110–1115 (2008).
124. Khademhosseini, A. et al. Microfluidic patterning for fabrication of contractile cardiac
organoids. Biomed Microdevices 9, 149–157 (2007).
125. Khanal, G., Chung, K., Solis-Wever, X., Johnson, B. & Pappas, D. Ischemia/reperfusion
injury of primary porcine cardiomyocytes in a low-shear microfluidic culture and analysis
device. Analyst 136, 3519–3526 (2011).
126. Grosberg, A., Alford, P. W., McCain, M. L. & Parker, K. K. Ensembles of engineered
cardiac tissues for physiological and pharmacological study: Heart on a chip. Lab Chip 11,
4165–4173 (2011).
127. Annabi, N. et al. Hydrogel-coated microfluidic channels for cardiomyocyte culture. Lab
Chip (2013). doi:10.1039/C3LC50252J
128. Kloß, D. et al. Microcavity array (MCA)-based biosensor chip for functional drug
screening of 3D tissue models. Biosensors and Bioelectronics 23, 1473–1480 (2008).
129. Fobel, R., Fobel, C. & Wheeler, A. R. DropBot: An open-source digital microfluidic
control system with precise control of electrostatic driving force and instantaneous drop velocity
measurement. Applied Physics Letters 102, 193513–193513–5 (2013).
126
130. Boudou, T. et al. A Microfabricated Platform to Measure and Manipulate the Mechanics
of Engineered Cardiac Microtissues. Tissue Engineering Part A 18, 910–919 (2012).
131. Zhao, Y. & Cho, S. K. Micro air bubble manipulation by electrowetting on dielectric
(EWOD): transporting, splitting, merging and eliminating of bubbles. Lab Chip 7, 273–280
(2007).
132. Fan, S.-K. et al. Atmospheric-pressure microplasma in dielectrophoresis-driven bubbles
for optical emission spectroscopy. Lab on a Chip 12, 3694 (2012).
133. Hansen, A. et al. Development of a Drug Screening Platform Based on Engineered Heart
Tissue. Circulation Research 107, 35–44 (2010).
134. Iyer, R. K., Chiu, L. L., Reis, L. A. & Radisic, M. Engineered cardiac tissues. Current
Opinion in Biotechnology 22, 706–714 (2011).
135. Zimmermann, W. H. et al. Three-dimensional engineered heart tissue from neonatal rat
cardiac myocytes. Biotechnology and Bioengineering 68, 106–114 (2000).
136. Fukuhara, S. et al. Direct cell-cell interaction of cardiomyocytes is key for bone marrow
stromal cells to go into cardiac lineage in vitro. J Thorac Cardiovasc Surg 125, 1470–1480
(2003).
137. Webster, D. R. & Patrick, D. L. Beating rate of isolated neonatal cardiomyocytes is
regulated by the stable microtubule subset. Am J Physiol Heart Circ Physiol 278, H1653–H1661
(2000).
138. Hersch, N. et al. The constant beat: cardiomyocytes adapt their forces by equal
contraction upon environmental stiffening. Biology Open (2013). doi:10.1242/bio.20133830
127
139. Shih, S. C. C., Barbulovic-Nad, I., Yang, X., Fobel, R. & Wheeler, A. R. Digital
microfluidics with impedance sensing for integrated cell culture andanalysis. Biosensors and
Bioelectronics 42, 314–320 (2013).
140. Ng, A. H. & Chamberlain, D. Personal communication.
141. Ng, A. & Sykes, E. Personal communication.
142. Lafrenière, N. M. & Kirby, A. E. Personal communication.
143. Pelletier, L. Personal communication.
144. Palanski, K. & Ryu, W. Personal communication.
145. Sista, R. S. et al. Multiplex newborn screening for Pompe, Fabry, Hunter, Gaucher, and
Hurler diseases using a digital microfluidic platform. Clin. Chim. Acta 424C, 12–18 (2013).
146. Schell, W. A. et al. Evaluation of a digital microfluidic real-time PCR platform to detect
DNA of Candida albicans in blood. Eur. J. Clin. Microbiol. Infect. Dis. 31, 2237–2245 (2012).
147. Fobel, R. Personal communication.
128
APPENDIX A
Figure A.1: Spheroid identification after manipulation. For deformation studies
individual spheroids were imaged before manipulation by ALHR or DMF by light
microscopy (indicated with arrows). To identify spheroids post-manipulation the
automated position coordinates on the automated microscope were used to find the x-y
coordinates. Then local fiduciary markers (particularly other spheroid structures,
indicated by asterisks in the figure) were used to determine exact spheroid location.