development of cell culture processes on digital ... · this chapter introduces the fundamental...
TRANSCRIPT
Development of Cell Culture Processes on Digital Microfluidic
Platforms
by
Sam H. Au
A thesis submitted in conformity with the requirements
for the degree of Doctor of Philosophy – Biomedical Engineering
Institute of Biomaterials and Biomedical Engineering
University of Toronto
© Copyright by Sam H. Au (2013)
ii
Development of Cell Culture Processes on Digital Microfluidic
Platforms
Sam H. Au
Doctor of Philosophy – Biomedical Engineering
Institute of Biomaterials and Biomedical Engineering
University of Toronto
2013
Abstract
In vitro microenvironments used for culturing and studying living cells have remained virtually
unchanged for the last five decades. Mammalian cells are routinely seeded as monocultures onto
rigid, homogeneous, two dimensional substrates – systems with limited physiological relevance.
Microfluidics has the potential to significantly improve cell models to better mimic native tissue
or disease states in addition to a host of other benefits such as improved throughput, reduced
consumable requirements and seamless integration with a number of analysis techniques. Digital
microfluidics, a fluid handling technique which manipulates discrete droplets over micro-
electrode patterned surfaces, may be a valuable tool for a number of cell applications
incorporating all of the above potential advantages. Cell culture and analysis is a new application
for digital microfluidics – the first report of such was in 2008. As a result, a number of technical
impediments must be addressed before cells can be effectively and routinely studied on these
microfluidic devices. These impediments include: a) rapid device failure due to protein
biofouling on hydrophobic device surfaces, b) the unexplored possibility of detrimental effects
on cell fitness arising from the electrokinetic manipulation of droplets and c) the lack of robust
iii
systems capable of long term automated culture and integrated analysis of cells. The aim of the
work described in this thesis is to advance digital microfluidics for biologically relevant cellular
culture and analysis by addressing each of these challenges. The technical advancements gained
from these studies can then be used to develop a proof-of-concept digital microfluidic platform
for modeling liver tissue. In summary, this work describes advances towards physiologically-
relevant culture and analysis of living cells on digital microfluidics, a technology which has the
potential to become a valuable tool for the biomedical research community.
iv
To him who has had the experience no explanation is necessary, to him who has not, none is
possible.
- Ram Dass
v
Acknowledgments
First and foremost I wish to express deep gratitude to my many research mentors. I count myself
extremely lucky to have studied under each of these skilled scientists: Aaron Wheeler, whose
enthusiasm, generosity and, most importantly, patience are the best qualities anyone could ask
for in a graduate supervisor. Veronica Carvalhal for teaching me the principles of intelligent
experimental design – like most things in life, what you get out depends on what you put in;
Arindom Sen for teaching me that when it comes to cell culture, nurture beats nature; and Poki
Yuen and Vasiliy Goral for showing me that even though there are an infinite number of
problems to be solved and an infinite ways to solve them, we advance by finding one solution to
one problem at a time.
To my committee members, Professors Jonathan Rocheleau, Alison McGuigan, Christopher Yip
and Eugenia Kumacheva for their guidance and wise counsel. A special thanks to my external
committee member, Professor David Juncker, for making the trip to add his expertise.
For sparking my interest in the sciences, I am eternally grateful to some very special teachers:
Mr. Wereley, Mr. Wrightson, Mr. Cantrill and Mr. Edmiston. Thanks for making math and
science so much fun, I still can’t believe that they actually pay people to do this.
I am especially grateful to some very supportive Wheeler Lab members: Ryan Fobel for
countless discussions on parenthood, caloric-restriction, lock-picking, tax law, get-rich-quick
schemes and fungal mind-control; Alphonsus Ng, who, ever generous with his resources and
time, can be counted to be in the lab 24/7 so that no lab member would ever have to work alone;
Vivienne Luk for her glass always-full optimism; and Andrea Kirby for giving me a life-
threatening addiction to carrot cake. And to the many other Wheelerites who I have had the
vi
awesome privilege of counting not only as colleagues, but also as friends: Mohamed
Abdelgawad, Kihwan Choi, Irwin Eydelnant, Lindsey Fiddes, Lorenzo Gutierrez, Mais Jebrail,
Sydney Kuipers, Paresh Kumar, Nelson Lafreniere, Jared Mudrik, Nauman Mufti, Brendon
Seale, Mahesh Sarvothaman, Motashim Shamsi, Steve Shih, Suthan Srigunapalan and Hao Yang.
There are a number of people outside the Wheeler Lab people who have contributed much
knowledge and inspiration to this academic journey: Dean Chamberlain, Evan Mills and Gary
Mo. Much gratitude to Henry Lee and Yimin Zhou for patiently solving countless equipment
failures in the cleanroom. And to the innumerable other people who have lent me helping hands,
ideas, cells, reagents, solvents, slides, DNA sequences or pipette tips but whose contributions I
may have forgotten, my gratitude is greater than my memory.
For reminding me that there is a world outside the 4th
floor of CCBR, profound thanks to Stefan
Cusi and Dorcas Lam. Finally, I am especially grateful to my brother and parents for lovingly
supporting me in countless ways through over 2 decades of education.
vii
Overview of Chapters
The use of digital microfluidics (DMF) for cell culture and analysis has a number of potential
benefits over traditional macro scale techniques. However, as cell applications are a relatively
new use for this technology, many fundamental challenges must be addressed before DMF can
be reliably used for routine analysis. This thesis describes my work towards addressing several of
these challenges:
Chapter 1 – Microfluidics and Cell Studies
This chapter introduces the fundamental physics of microfluidic devices with a focus on
dimensionless numbers which help describe fluid dynamics on the microscale. Some common
uses of microfluidic devices for cell-based applications are summarized followed by a review of
digital microfluidics and its use for cell applications. Finally, challenges associated with the use
of DMF platforms for cell-based applications are introduced.
Chapter 2 – Pluronic Additives to Inhibit Device Failure
Device failure when using protein-rich solutions required for cell culture is a major impediment
to developing DMF as a useful tool for cellular analysis. The Wheeler Lab previously developed
techniques in which pluronics (block co-polymers with tailorable hydrophobic/hydrophilic chain
lengths) were doped into working solutions to slow device failure. However, the use of pluronics
as an anti-fouling agent had not been optimized, especially in the context of cell culture
applications. The goal of the work described in this chapter was to determine a) which
viii
parameters of pluronics are important in delaying/preventing device failure, b) the mechanisms
behind the anti-fouling properties of the co-polymer and c) how these co-polymers interact with
cells.
Chapter 3 – Effects of Digital Microfluidic Actuation on Cell Fitness
Since DMF uses electric fields to drive droplets, it is important to characterize if and how these
electric fields may influence cell health and fitness. Although previous studies have shown little
affect on cell viability and growth rates, there may be more subtle effects on cells as a result of
DMF manipulation. The goal of the work described in this chapter was to examine the effects of
DMF-operation on the genome-level responses of mammalian cells. A number of responses were
analyzed to this end including heat shock activation, DNA integrity, and genomic expression
profiles.
Chapter 4 – Integrated Microorganism Culture and Analysis
An advantage of microfluidics, and DMF in particular, is the ability to combine a number of
traditionally tedious and labour-intensive experimental steps onto a single integrated platform. A
DMF platform was developed for multi-day culture and analysis of bacteria, algae and yeast in a
highly automated fashion.
ix
Chapter 5 – Microfluidic Liver Organoid Platform
This chapter describes ongoing work-in-progress. The liver is a vital organ for metabolism,
detoxification and hormone production. There is therefore interest in developing in vitro liver
models to study small molecule interactions, pharmacokinetics and disease states. Building upon
the work in Chapters 2-4, the goal of this project is to evaluate the suitability of digital
microfluidics for creating liver models which better mimic the in vivo microenvironment than
models created in traditional cell culture platforms. The use of co-cultured, encapsulated cells in
3D hydrogel matrices were explored to this end.
x
Author Contributions
The work contained within this thesis was made possible by the contributions of many skilled
hands and bright minds.
Chapter 2 – Paresh Kumar (former visiting graduate student from the India Institute of
Technology) and I conducted device longevity experiments and contact angle measurements. He
also aided with device fabrication. Dr. Lindsey Fiddes (former graduate student at the University
of Toronto) provided training on goniometer operation. I conducted critical micelle concentration
determination assays, cell compatibility studies and statistical analyses. Dr. Gary Mo (former
graduate student at the University of Toronto) contributed helpful discussions. This work is
published in Langmuir. Au, S.H.; Kumar, P.; Wheeler, A.R. "A New Angle on Pluronic
Additives: Advancing Droplets and Understanding in Digital Microfluidics" Langmuir, 2011, 27,
8586-8594.
Chapter 3 – Ryan Fobel (graduate student at the University of Toronto) contributed device
modeling expertise, design of droplet temperature measurements systems and insightful
discussions. Dr. Dean Chamberlain (postdoctoral fellow at the University of Toronto) and Dr.
Lindsey Fitzgerald (former graduate student at the University of Toronto) both helped with RT-
qPCR training and operation. Julie Tsao and Carl Virtanen (both employees of the University
Health Network Microarray Centre (Toronto, Canada) ran and analyzed microarray experiments.
Professor Joel Voldman (Massachusetts Institute of Technology) and Dr. Salil Desai (former
graduate student at the Massachusetts Institute of Technology) contributed GFP-HSE cells. I
fabricated devices, cultured cells and exposed cells to DMF manipulation, hypothermia controls
xi
and arsenite controls. I also conducted cell-stress evaluation studies, DNA integrity assays,
RNA isolation and qPCR. This work is published in Integrative Biology, In Press 2013
Chapter 4 – Dr. Steve Shih (former graduate student at the University of Toronto) ran algae
experiments and designed the device automation system. I conducted bacteria and yeast
experiments, cell death assays and bacterial transformation. Dr. Evan Mills (former graduate
student at the University of Toronto) gifted us bacteria, Dawn Edmonds (lab manager at the
University of Toronto) provided us with yeast and training on yeast culture and Dr. Kamlesh
Patel (Sandia National Laboratories) provided training for algae experiments. This work is
published in Biomedical Microdevices: Au, S.H.; Shih, S.C.C.; Wheeler, A.R. "Integrated
Microbioreactor for Culture and Analysis of Bacteria, Algae and Yeast" Biomedical
Microdevices, 2011, 13, 41-50.
Chapter 5 – Dr. Dean Chamberlain (postdoctoral fellow at the University of Toronto) provided
biological guidance. Shruthi Mahesh (former volunteer lab assistant at the University of Toronto)
helped with protocol development. I designed photomasks, fabricated devices and conducted dye
mixing, viability, contractility, albumin and enzymatic activity experiments. A manuscript is in
preparation.
Professor Aaron Wheeler contributed guidance, expertise and direction to all of the work
described in this document.
xii
Table of Contents
Abstract ........................................................................................................................................... ii
Acknowledgments ........................................................................................................................... v
Overview of Chapters ................................................................................................................... vii
Author Contributions ...................................................................................................................... x
List of Figures .............................................................................................................................. xvi
List of Tables ............................................................................................................................. xviii
List of Equations .......................................................................................................................... xix
Abbreviations ................................................................................................................................ xx
List of Foundations and Funding Sources ................................................................................. xxiii
Chapter 1: Microfluidics and Cell Studies ...................................................................................... 1
1.1 Fundamentals of Microfluidics ........................................................................................... 1
1.2 Microfluidics for Cell Applications .................................................................................... 5
1.3 Digital Microfluidics for Cell Applications ........................................................................ 8
1.4 Thesis Objectives .............................................................................................................. 17
Chapter 2: Pluronic Additives to Inhibit Device Failure .............................................................. 18
2.1 Introduction ....................................................................................................................... 19
2.2 Experimental ..................................................................................................................... 21
2.2.1 Reagents and Materials ......................................................................................... 21
2.2.2 Device Fabrication ................................................................................................ 22
2.2.3 DMF Longevity Assay .......................................................................................... 22
2.2.4 Contact Angle Measurements ............................................................................... 25
2.2.5 Cell Growth and Viability Assay .......................................................................... 26
2.2.6 Critical Micelle Concentration Determination ...................................................... 26
2.2.7 Statistical Analysis ................................................................................................ 27
xiii
2.3 Results and Discussion ..................................................................................................... 27
2.3.1 Device Lifetime .................................................................................................... 29
2.3.2 Droplet Wetting .................................................................................................... 35
2.3.3 Compatibility with Cells ....................................................................................... 39
2.4 Conclusions ....................................................................................................................... 41
Chapter 3: Effects of Digital Microfluidic Actuation on Cell Fitness .......................................... 43
3.1 Introduction ....................................................................................................................... 43
3.2 Experimental ..................................................................................................................... 45
3.2.1 Device Fabrication and Operation ........................................................................ 45
3.2.2 Cell Culture and Stress Conditioning ................................................................... 46
3.2.3 Cell Stress Evaluation and Flow Cytometry ......................................................... 47
3.2.4 Single Cell Gel Electrophoresis COMET Assays ................................................. 48
3.2.5 Microarrays and qPCR .......................................................................................... 48
3.2.6 Temperature Measurements .................................................................................. 51
3.3 Results and Discussion ..................................................................................................... 51
3.3.1 Preliminary Experiments and Cell-based Stress Sensors ...................................... 51
3.3.2 DNA Integrity ....................................................................................................... 54
3.3.3 Gene Expression – Microarrays ............................................................................ 56
3.3.4 Gene Expression – qPCR ...................................................................................... 63
3.3.5 Droplet Heating ..................................................................................................... 65
3.4 Conclusions ....................................................................................................................... 68
Chapter 4: Integrated Microorganism Culture and Analysis ........................................................ 69
4.1 Introduction ....................................................................................................................... 69
4.2 Experimental ..................................................................................................................... 71
4.2.1 Macroscale Cultures .............................................................................................. 71
xiv
4.2.2 Device Fabrication ................................................................................................ 72
4.2.3 Device Operation .................................................................................................. 73
4.2.4 Microscale Cultures .............................................................................................. 75
4.2.5 Growth Curve Generation ..................................................................................... 77
4.2.6 Cell Death Assays ................................................................................................. 78
4.2.7 Transformation ...................................................................................................... 78
4.3 Results and Discussions .................................................................................................... 79
4.3.1 Microbioreactor Design ........................................................................................ 79
4.3.2 Microorganism Culture ......................................................................................... 82
4.3.3 Downstream Processing and Analysis .................................................................. 86
4.4 Conclusions ....................................................................................................................... 88
Chapter 5: Microfluidic Liver Organoid Platform ........................................................................ 90
5.1 Introduction ....................................................................................................................... 90
5.2 Experimental ..................................................................................................................... 91
5.2.1 Device and SU-8 Barrier Fabrication ................................................................... 91
5.2.2 Device top and bottom plates were Cell Handling and Preparation ..................... 93
5.2.3 Device Operation Protocols .................................................................................. 94
5.2.4 Mixing Analysis .................................................................................................... 95
5.2.5 Viability and Contractility Assays ........................................................................ 96
5.2.6 Albumin Analysis ................................................................................................. 97
5.2.7 Cytochrome P450 3A4 Activity Assay ................................................................. 97
5.3 Preliminary Results and Discussion .................................................................................. 99
5.3.1 Organoid Confinement, Feeding, and Mixing ...................................................... 99
5.3.2 Organoid Contractility and Viability .................................................................. 102
5.3.3 Albumin Activity ................................................................................................ 105
xv
5.3.4 Cytochrome P450 Enzymatic Activity ............................................................... 106
5.4 Future Work .................................................................................................................... 109
Conclusions and Future Directions ............................................................................................. 110
References ................................................................................................................................... 119
xvi
List of Figures
Figure 1.1 General digital microfluidic device schematic. ............................................................ 9
Figure 1.2 Schematic of hydrophilic adhesion pads .................................................................... 14
Figure 2.1 Schematic device lifetime assay operation. ................................................................ 24
Figure 2.2 Device longevity assay ‒ initial screen. ...................................................................... 31
Figure 2.3 Device longevity assay ‒ concentration dependance. ................................................ 33
Figure 2.4 Non-potentiated contact angles .................................................................................. 37
Figure 2.5 Electrodynamic contact angles ................................................................................... 39
Figure 2.6 Pluronic cytotoxicity ................................................................................................... 41
Figure 3.1 Cell-based stress sensor results. .................................................................................. 53
Figure 3.2 Quantification of DNA integrity. ................................................................................ 55
Figure 3.3 Microarray heat map ................................................................................................... 58
Figure 3.4 Microarray expression comparisons ........................................................................... 59
Figure 3.5 Droplet temperature in digital microfluidics .............................................................. 66
Figure 4.1 Schematic of BAY microbioreactor ........................................................................... 74
Figure 4.2 Operation of BAY microbioreactor ............................................................................ 80
xvii
Figure 4.3 Microorganisms on device .......................................................................................... 83
Figure 4.4 Microorganism growth curves .................................................................................... 84
Figure 4.5 Microorganism viability and transformation .............................................................. 87
Figure 5.1 Digital microfluidic organoid platform ...................................................................... 92
Figure 5.2 General automated droplet exchange procedure ........................................................ 95
Figure 5.3 Dye-mixing study ..................................................................................................... 101
Figure 5.4 Organoid contractility. .............................................................................................. 104
Figure 5.5 Organoid viability. .................................................................................................... 104
Figure 5.6 Organoid albumin secretion assay. ........................................................................... 106
Figure 5.7 Cytochrome P450 3A4 activity. ............................................................................... 108
xviii
List of Tables
Table 2.1 Physical properties of Pluronics ................................................................................... 29
Table 3.1 Stress and apoptosis gene summary ............................................................................. 62
Table 3.2 qPCR validation of Dusp1 ........................................................................................... 64
Table 4.1 BAY microreactor parameters ..................................................................................... 76
Table 4.2 Microorganism doubling time comparison .................................................................. 85
Table 5.1 Collagen-cell suspension components ......................................................................... 94
Table C.1 Current state of digital microfluidics for cell applications. ....................................... 115
xix
List of Equations
Equation 1.1 Reynolds Number ..................................................................................................... 2
Equation 1.2 Péclet Number .......................................................................................................... 3
Equation 1.3 Capillary Number ..................................................................................................... 4
Equation 1.4 Lippman-Young Law ............................................................................................. 10
Equation 1.5 Lippman-Young derived driving force .................................................................. 11
Equation 1.6 Electromechanical Framework ............................................................................... 11
Equation 1.7 Electromechanical Framework derived driving force ............................................ 12
Equation 2.1 Actuation time log-normal curve fit ....................................................................... 25
Equation 4.1 Doubling time ......................................................................................................... 77
Equation 4.2 Growth rate ............................................................................................................. 77
Equation 5.1 Unbiased estimator of standard deviation .............................................................. 96
xx
Abbreviations
AC Alternating current
BAY Bacteria, algae, yeast
Ca Capillary Number
CCD Charge-coupled device
CMC Critical micelle concentration
COMET Single cell gel electrophoresis assay
CS Calf serum
CYP Cytochrome P450
DC Direct current
DEP Dielectrophoresis
DI Deionized
DMF Digital microfluidic(s)
ECM Extracellular matrix
EDTA Ethylenediaminetetraacetic acid
EMF Electromagnetic field(s)
xxi
EthD-1 Ethidium homodimer-1
EWOD Electrowetting on dielectric
FBS Fetal bovine serum
GFP Green fluorescent protein
HLB Hydrophilic-lipophilic balance
HSE Heat shock element
ITO Indium tin oxide
OD Optical density
PBS Phosphate buffered saline
PCB Printed circuit board
PCR Polymerase chain reaction
Pe Péclet Number
PEO Polyethylene oxide
PP Peak-to-peak
PPO Polypropylene oxide
QC Quality control
xxii
qPCR Real time polymerase chain reaction
Re Reynolds Number
RMS Root-mean square
S.D. Standard deviation
UV Ultraviolet
V Voltage
Vis Visible
YFP Yellow fluorescent protein
xxiii
List of Foundations and Funding Sources
National Sciences and Engineering Research Council of Canada
Canadian Institutes of Health Research
1
Sam H. Au Microfluidics and Cell Studies
Chapter 1
Microfluidics and Cell Studies
1.1 Fundamentals of Microfluidics
The premise of microfluidics and micro-total analysis systems are to miniaturize and integrate
bench-top (macro-scale) laboratories processes. Microfluidic devices are commonly defined as
tools with micron-scale features that are capable of manipulating microliter or smaller volumes1,
2.
The most obvious benefits of miniaturization3 are significantly reduced consumables and reagent
use which can reduce costs and may allow for reductions in the volumes of rare and/or patient
samples. Moreover, smaller sample volumes can increase the number of replicates or
experimental conditions which can be conducted at once (parallelization)4. Miniaturization also
significantly decreases the physical size of total analysis systems3, which enables the
development of portable tools such as handheld blood glucose monitors. The reduction of
volume and length scales can also drastically alter the fundamental physics of fluid behaviour1.
This leads to other benefits of microfluidics over traditional macro-scale fluid handling such as
the ability to establish well defined chemical gradients3, shorten analysis times and improve
detection sensitivities5, 6
. These principles are briefly described below; a more in detailed review
of the physics of microscale flows can be found in an excellent review paper by Squires and
Quake1.
2
Sam H. Au Microfluidics and Cell Studies
Surface Area to Volume Ratios
The surface area of an object or liquid is proportional to the square of the characteristic length
scale while the volume is proportional to the cube of that length scale. Therefore the surface area
to volume ratio increases linearly with decreasing length scale. The small length scales in
microfluidic systems result in high surface area to volume ratios which in turn can substantially
increase heat and mass transfer rates, which can be used to speed up the rates of
exothermic/endothermic reactions but may also be detrimental – for example, there are often
increased biofouling rates in microsystems. Addressing this latter consideration is the primary
motivation for Chapter 2.
Inertial and Viscous Forces
Inertia is the resistance of objects/liquids to changing their current state of motion. Viscosity is
the resistance of a fluid to stress-induced deformation. To describe the balance between these
forces, the dimensionless Reynolds number (Re) is used:
L is the characteristic dimension [m]
U is the mean fluid velocity [m/s]
ρ is the fluid density [kg/m3]
µ is the fluid dynamic viscosity [Pa.s]
For fluids in motion, these conflicting forces dictate whether the flow is turbulent (chaotic) or
laminar (deterministic). Since the characteristic dimensions in microfluidics are typically on the
order of microns (10-6
m), fluid flows in most microfluidic systems are usually viscosity
𝑅𝑒 ≝𝐼𝑛𝑒𝑟𝑡𝑖𝑎𝑙 𝑓𝑜𝑟𝑐𝑒𝑠
𝑉𝑖𝑠𝑐𝑜𝑢𝑠 𝑓𝑜𝑟𝑐𝑒𝑠=
𝐿𝑈𝜌
𝜇 (1.1)
3
Sam H. Au Microfluidics and Cell Studies
dominated resulting in laminar flows profiles with Re < 2000. Laminar flow regimes are
characterized by linear deterministic flows with no mixing of streamlines (i.e., fluid flows
parallel to the walls of a straight channel or pipe). In microsystems, the orderly flow results in no
cross-currents, eddies or mixing which is necessary for the formation of linear fluid flow
gradients but can render mixing a challenge.
Convection and Diffusion
The orderly flow profiles in microsystems have substantial impact upon microfluidic mass
transfer. Convective mass transfer in fluid flows occurs when streamlines mix together (bulk
fluid mixing) resulting in increased homogeneity. In most systems, convection is much faster
than diffusive mixing, which relies on the random stochastic motion of particles to reach increase
homogeneity. The relative dominance of convection and diffusion can be described by the
dimensionless Péclet number (Pe):
L is the characteristic dimension for mass transfer [m]
U is the mean fluid velocity [m/s]
D is the diffusion coefficient [m2/s]
Because of the small length scales in microfluidic systems, mass transfer is typically dominated
by diffusion which allows for the formation of well defined chemical gradients. Also in contrast
to macro-scale systems, the small length scales in microfluidic systems often reduce the time
required for diffusion-dominated systems to become well-mixed, meaning that diffusive mass
transfer alone may be sufficient in some microsystems. In the work described in Chapter 4,
diffusion alone is insufficient to properly mix dividing yeast cells due to their relative large size
𝑃𝑒 ≝𝐶𝑜𝑛𝑣𝑒𝑐𝑡𝑖𝑜𝑛
𝐷𝑖𝑓𝑓𝑢𝑠𝑖𝑜𝑛=
𝑈𝐿
D (1.2)
4
Sam H. Au Microfluidics and Cell Studies
and hence low diffusion coefficient – therefore a continuous mixing system was implemented to
improve homogeneity.
Viscous and Interfacial Forces
The increased surface area to volume ratios in microfluidic devices also influence microscale
fluid behaviour. A fluid in contact with a solid (e.g. the surface of a microfluidic device)
experiences both interfacial forces (liquid-solid interaction energy) and viscous forces. The
dimensionless capillary number (Ca) is used to describe the balance between these forces:
µ is the fluid dynamic viscosity [Pa.s]
U is the mean fluid velocity [m/s]
γ is the surface or interfacial tension [N/m]
The capillary number can be used to predict the motion of or control fluids in microfluidic
devices. For example capillary forces can be used to draw fluid through microchannels and the
contact angle/wetting of fluids onto surfaces is an important factor in the biofouling of DMF
devices (described in Chapter 2).
Applications and Driving Forces
The first applications of microfluidics were in analytical chemistry4. Chromatography
7 and
capillary electrophoresis8 both benefit from improved sensitivities and resolutions brought on by
flow through smaller dimensions. Since then, the applications for microfluidics have expanded
many-fold4 to include genomics
9, proteomics
9, polymerase chain reaction
10, drug discovery
11,
biochemical assays12
, crystallization13
, cell culture/analysis,14-16
and many others. In this work I
focus on cell applications of DMF, which is described in more detail in section 1.2.
𝐶𝑎 ≝𝑉𝑖𝑠𝑐𝑜𝑢𝑠 𝑓𝑜𝑟𝑐𝑒𝑠
𝐼𝑛𝑡𝑒𝑟𝑓𝑎𝑐𝑖𝑎𝑙 𝑓𝑜𝑟𝑐𝑒𝑠=
µU
γ (1.3)
5
Sam H. Au Microfluidics and Cell Studies
The driving forces and configurations for fluid handling in microfluidic systems have also
increased many-fold. Early work relied upon pressure-driven7 or electroosmotic
8 flows.
Microfluidic technologies now incorporate many more fluid handling modalities which can be
divided into five main categories2: capillary forces, pressure-driven, centrifugal, electrokinetic
and acoustic. There are a number of trade-offs associated with each modality such as plug flow
liquid-profiles in electroosmotic flow vs. laminar streamline liquid-profiles in pressure-driven
flow2. This thesis focuses on the electrokinetic modality of digital microfluidics (DMF)
(described in more detail in section 1.3).
1.2 Microfluidics for Cell Applications
Mammalian cells, bacteria, algae and yeast are used widely in biotechnology, biopharmaceutical
production, drug discovery, genomics, proteomics and studies in fundamental biology. The
generic advantages of microfluidics, which are applicable to nearly all applications, such as
reduced reagent use and increased throughput are also useful for cell applications. For example,
cell-based microfluidic platforms with 100 chamber perfusion-flow devices17
, automated 96
chamber devices driven by peristaltic pumps18
and 160 chamber devices containing thousands of
encapsulated drops19
have been developed. Moreover, the miniaturization of traditional macro-
scale processes has enabled the integration of microfluidic cell manipulation with sorting and
downstream analyses such as qPCR14, 15
. These micro total analysis systems significantly reduce
the amount of manual sample handling required which reduces labour requirements, user-to-user
variation and experimental error. However, microfluidics also offers a number of additional
benefits which are particularly suited to cell applications.
6
Sam H. Au Microfluidics and Cell Studies
Microenvironment Control
The cellular microenvironment influences all cellular activities including growth, division,
differentiation, motility, survival and phenotypic behaviour14
. Microfluidics offers increased
control of cellular microenvironments14-16
in comparison to traditional cell culture tools such as
flasks and well-plates. This enables the generation of physiologically relevant cell environments
and permits the study of the influence of microenvironmental factors on cell behaviour.
In contrast to traditional cultures where cells are often maintained in static liquid media,
microfluidic systems can maintain cells under highly tunable flow and shear rates. This is
particularly useful for systems which experience complex pulsatile flow profiles in vivo such as
the cardiovascular20
and respiratory21
systems. The development, differentiation and behaviour
of these tissue systems often depend on the presence of shear stress which cannot be accurately
and uniformly applied with tissue culture flask systems.
The ability to create laminar (Re < 2000), diffusion dominated (low Pe) flows in microfluidics
can be used to create well defined gradients14
. One application of gradients is the study of cell
migration in response to differential concentrations of soluble factors or cytokines. In this
manner, microfluidics has been used to study cell migration across well defined chemical22
and
oxygen23
gradients. More complex gradients can be generated on microfluidics as well. For
example, gradients and flows in two or more dimensions can be created in microfluidic systems24
which can be used to study more complex biological systems such as the emergence of antibiotic
resistance in bacterial populations25
. In addition, microfluidic systems can be used to create
multiple liquid streams which are in contact but remain unmixed due to slow diffusion-
dominated mass transfer rates. Signal propagation and dynamics can be studied by exposing
single cell or tissue units to these unmixed streams26
.
7
Sam H. Au Microfluidics and Cell Studies
The micron or nanometer-scale features created by microfluidic fabrication technologies can also
be used to model mammalian physiology. For example, organs are often organized around
basement membranes and basal laminae which are comprised of extracellular matrix and serve to
separate different cell types and fluids. Microfluidic tools have been created with artificial
membrane mimics to better model physiological processes and cellular behaviour15, 27
.
Micrometer-scale control in microfluidics can also be used to create 3D cell culture systems
which are better models of in vivo microenvironments since cells in all multi-cellular organisms
are organized into three dimensions. Synthetic or natural hydrogels are often used in
microsystems as cell scaffolds or extracellular matrix (ECM)3, 14, 28, 29
. The precise spatial control
available in microfluidic systems can be used to create well defined spatial geometries29
which
are important for preventing the development of necrotic cores. Also, for thermosetting
hydrogels such as collagen and agarose, the high surface area to volume ratio in microfluidic
systems can be capitalized on to rapidly set gels with high uniformity29
.
Single Cell and Population Analyses
The micro-scale features of microfluidic devices allows for the capture30, 31
, analysis30, 31
and
manipulation32
of single cells, often in picoliter-volume droplets33
. This enables the study of not
only single isolated cells, but also individual cells within a population to gain a better
understanding of the population as a whole. For example, the majority of traditional cell-based
assays provide a single (mean) read-out for the entire cell population (e.g. mean
fluorescent/luminescent intensity in ELISA assays). This is an appropriate metric for
homogeneous, normally distributed populations. However, a single mean is an inappropriate
measure for many cell applications since even clonal cell populations exposed to ―identical‖
treatments result in significant phenotypic heterogeneity within the popluation34
. In these cases,
8
Sam H. Au Microfluidics and Cell Studies
examining the state of individual cells provides a far better measure of the population. Flow
cytometry is an example of a similar technique which is capable of studying the state of
individual cells within a population, but usually requires the availability of appropriate
fluorescent tags. The wide range of detection modalities that can be coupled with microfluidics
permits the detection of ―secreted antibodies… intracellular, cell-surface or secreted proteins and
for quantifying catalytic or regulatory activities‖ (as aptly phrased by Mazutis et al.35
) of single
cells as well.
1.3 Digital Microfluidics for Cell Applications
Digital microfluidics, also known as electrowetting on dielectric (EWOD), is an electrokinetic
method of microscale fluid manipulation36, 37
. Droplets (of nanoliter to milliliter volumes) can be
mixed, split, merged and dispensed, the four basic operations required for most liquid
experiments. A generic device performing these basic operations is depicted in Figure 1.1A.
9
Sam H. Au Microfluidics and Cell Studies
Figure 1.1 General digital microfluidic device schematic. (A) Photograph of a digital
microfluidic device demonstrating droplet mixing, splitting, merging and dispensing. (B) Side-
view schematic of a two-plate digital microfluidic device38
. Reproduced by permission of The
Royal Society of Chemistry.
Digital microfluidic devices are fabricated with electrodes patterned underneath an insulating
dielectric layer. Surfaces in contact with working liquids are coated in a hydrophobic layer to
reduce droplet-surface interfacial force. A schematic of a generic DMF device is presented in
Figure 1.1B. To achieve control over the small volumes in microfluidics requires the ability to
fabricate features of micron or nanometer-scale dimensions. As a result, DMF (and microfluidics
in general) has benefited from fabrication techniques previously developed for microelectronics2,
4. Some examples of these techniques used to fabricate DMF devices include metal deposition,
10
Sam H. Au Microfluidics and Cell Studies
photolithography, etching and spin-coating39
. The DMF fabrication techniques used in the work
described in this thesis are described in sections 2.2.2, 3.2.1 and 4.2.2.
Physics of Droplet Motion
Multiple models have been developed to describe the forces behind DMF-driven droplet motion.
Two different theoretical approaches will be described here, the electrowetting model and the
electromechanical model, both of which reach consensus on DMF driving forces40
.
In the electrowetting model, charges accumulate at the interface of a charged conductive material
and a non-conductive material (e.g. Fig 1.1B – between the charged electrode and dielectric layer
of a DMF device). The accumulated charges apply an interfacial force which is especially strong
at triple contact lines – that is at solid-liquid-gas interfaces (e.g. Fig 1.1B – at the Teflon-liquid-
air interface of a DMF device). This interfacial force then distorts deformable liquids resulting in
a change in contact angle. The equation governing this contact angle change is the Lippman-
Young law40
:
θ is the deformed contact angle due to applied voltage
θ0 is the contact angle without applied voltage
C is the capacitance of the dielectric layer between the liquid and the electrode [F]
γLG is the liquid-gas interfacial tension [N/m]
V is the applied RMS voltage [V]
ε0 is the permittivity of free space [F/m]
εr is the relative permittivity of dielectric [F/m]
d is the dielectric thickness [m]
Note that the contact angles of Equation 1.4 are static contact angles and do not account for
droplet motion after deformation. In this model, droplet motion occurs due to liquid contact
cosθ = cosθ0 +C
2γLGV2 = cosθ0 +
ε0εr
2γLG dV2 (1.4)
11
Sam H. Au Microfluidics and Cell Studies
angle asymmetry resulting in capillary forces which serve to bring droplets back into symmetry.
The driving force in this model can be expressed as40
:
F is the driving force [N]
L is the length of the triple contact line overlapping the actuated electrode (reduces to the
length of charged electrodes perpendicular to the direction of droplet motion) [m]
Another approach to determining driving forces is the electromechanical framework which
models DMF devices as electric circuits with each component represented as a capacitor and
resistor in parallel (Figure 1.1B). Highly resistive elements (such as the dielectric and
hydrophobic layers) reduce to capacitors. In this model, energy is stored in the capacitors as a
function of frequency and droplet position40
:
E is the energy in the system [J]
f is the applied frequency [Hz]
x is the droplet position along the axis perpendicular to motion [m]
The i subscript in Equation 1.6 refers to each layer of the DMF device (Fig. 1.1B) directly above
or below the liquid or filler portions of the electrode. Note that the voltage drop across each layer
is a function of the magnitude of the applied frequency, layer permittivity and layer thickness.
Differentiating the energy calculated in Equation 1.6 with respect to x yields force:
F = LγLG cosθ − cosθ0 =ε0εrL
2dV2
E(f,𝑥) = L
2 𝑥
ε0εri Vi2(j2πf)
𝑑𝑖𝑖
𝑙𝑖𝑞𝑢𝑖𝑑
+ (𝐿 − 𝑥) ε0εri Vi
2(j2πf)
𝑑𝑖𝑖
𝑓𝑖𝑙𝑙𝑒𝑟
(1.5)
(1.6)
12
Sam H. Au Microfluidics and Cell Studies
Note that the electromechanical model does not describe the fundamental nature of the forces
driving droplet motion, only that stored energy results in an applied force. The advantage of the
electromechanical model is that it takes into account the frequency dependency of the applied
voltage droplets across each layer and portion of the device. This is important if one wishes to
accurately model systems where the voltage drops across layers other than the dielectric are non-
negligible. However, for most DMF systems, the forces calculated by the electrowetting and
electromechanical models reach consensus. For manipulation of conductive liquids (such as cell
media) and air (used for all experiments in this thesis), the energy stored in the filler portion of
the electromechanical model is negligible in comparison to the energy stored in the liquid portion
(εr,liquid >> εr,filler ). Moreover, the energy stored in the liquid layers are negligible in comparison
to the energy stored in the parylene (dielectric) and Teflon (hydrophobic) layers because of the
differences in relative thicknesses (ddielectric ≈ 2-6 µm, dhydrophobic ≈ 235 nm vs. dliquid/filler ≈ 140-280
µm). Therefore, by grouping the dielectric and hydrophobic layers into a general ―dielectric
layer‖ for modeling purposes, Equation 1.7 simplifies to Equation 1.5.
Cell Applications on Digital Microfluidics
A wide range of cell-based applications have been conducted on microfluidic platforms (section
1.2). The majority of these studies were conducted in microflow systems in which an aqueous
fluid is delivered through enclosed micron-scale channels (microchannels). A variation on this
modality, often referred to as droplets in channels28, 41
, is also commonly used for cell
applications. In this mode, one phase (often aqueous) is typically delivered through
F f =∂E(f, 𝑥)
∂𝑥=
L
2
ε0εri Vi2(j2πf)
𝑑𝑖𝑖
𝑙𝑖𝑞𝑢𝑖𝑑
− ε0εri Vi
2(j2πf)
𝑑𝑖𝑖
𝑓𝑖𝑙𝑙𝑒𝑟
(1.7)
13
Sam H. Au Microfluidics and Cell Studies
microchannels as droplets using an immiscible carrier phase (often oil). In comparison to either
of these systems, digital microfluidics manipulates droplets on open planar surfaces typically
filled with air. Each of these modalities (channels, droplets-in-channels, or DMF) is useful for
different cell applications. For example, the establishment of well defined chemical gradients is
difficult in the discrete droplets of DMF or droplet in channel systems but is easily established in
microchannels22
. The encapsulation of thousands or millions of single cells in multiple isolated
serial reactors is difficult in microchannel or DMF systems but can be performed with extremely
high throughput in droplet in channel systems39
. The ability to rapidly reconfigure fluidic
networks for cell and reagent manipulation is difficult to perform in microchannels or droplet in
channel systems but is a trivial operation in DMF systems37
.
Cell studies are an attractive application for DMF because of its ability to precisely manipulate
droplets of different volumes and constituents, rapidly reconfigure fluidic paths, handle cells
with low-shear stress and integrate with numerous analysis modalities. The first report of DMF
used for cell applications was in 2008. Jurkat T-cells, a suspension cell line, were manipulated,
grown and assayed for viability on a DMF platform42
. Since then a number of DMF platforms
have been developed for cells. Cells in suspension were manipulated on hybrid dielectrophoresis
(DEP)-DMF devices43
or combined DMF and optoelectronic tweezer manipulation44
. Adherent
mammalian cell have been grown, subcultured and transfected on dried extracellular matrix
protein spots45
(Figure 1.2). Mammalian cell lines46
and primary cells47
have been cultured and
analyzed on hydrophilic spots of the top-plate by Teflon lift-off and cell lines have been seeded
onto hydrophilic spots of the bottom-plate formed by parylene lift-off48
. DMF platforms have
also been developed to encapsulate NIH-3t3 cells in agarose hydrogel discs49
, to manipulate
yeast and zebrafish embyros51
and for long-term culture and analysis of bacteria, algae and
14
Sam H. Au Microfluidics and Cell Studies
yeast52
(Chapter 3). A multiplexed DMF apoptosis assay was performed on a DMF platform50
which capitalized on the laminar flow profiles of DMF droplet manipulation to permit the
analysis of apoptotic cells delaminating from device surfaces. The explanation for this
phenomenon is that the Reynolds number (Eqn. 1.1) for a typical DMF setup (2 mm diameter
droplet, droplet velocity of 3.33 mm/s) manipulating an aqueous solution at room temperature, is
approximately seven. This is well inside the laminar regime.
Figure 1.2 Schematic of hydrophilic adhesion pads used to culture adherent mammaliancells45
–
Reproduced by permission of The Royal Society of Chemistry
Challenges for Digital Microfluidic Cell Applications
There are a number of challenges associated with the application of DMF for cell-based studies.
Addressing these challenges will be useful to researchers conducting cell-based research on
DMF platforms in the future such that more complex applications can be completed with more
reliability.
One critical challenge is biofouling, which can lead to catastrophic device failure as droplets are
unable to move away from fouled regions. This is a particularly severe problem for cell
applications for a number of reasons. First, serum-containing cell media used to maintain
mammalian cells are protein-rich solutions. The inherent amphiphilic nature of proteins and their
relatively large molecular weights contribute to their rapid adsorption to many material
15
Sam H. Au Microfluidics and Cell Studies
surfaces51
. Second, hydrophobic surfaces such as the Teflon-AF© surfaces used on DMF
devices, are typically fouled more rapidly by proteins than hydrophilic surfaces52
. Third, similar
to most microfluidic systems, DMF platforms have greatly increased surface area to volume
ratios in comparison to macro-scale counterparts. For example, the (liquid-solid) surface area to
volume ratio of a typical DMF system (two-plate, 2 mm diameter droplets, 140 µm spacer) is
approximately 14286 m-1
while for a common macro-scale cell culture format (100 µL in one
well of a flat bottom 96 well plate) it is approximately 952 m-1
, an order of magnitude difference.
Fourth, the accumulation of amphiphilic protein species on DMF devices increases the interfacial
interaction forces between cell media and contact surfaces. The Capillary Number (Eqn. 1.3) for
droplets on DMF is typically well under 1 (for an aqueous droplet moving at 3.33 mm/s, Ca ≈
0.00005) meaning interfacial forces dominate over viscous forces. Therefore increased interfacial
force due to biofouling increases droplet wetting which increases the effective surface area for
biofouling. An increased surface area then leads to an increased rate of biofouling etc., a feed-
forward loop. These four properties contribute to the immovability of droplets of cell media
containing 10% fetal bovine serum on DMF devices without the implementation of anti-
biofouling strategies. Although additives53
and disposable dielectrics54
have been previously
developed to combat biofouling on DMF devices, long-term robust manipulation of cell media
and other protein-rich solutions cannot become a reality without improved anti-fouling
technologies.
A second critical challenge is the potential for detrimental effects of electromagnetic fields
(EMF), which are known to harm cells and living tissue under some circumstances. Although
there is much debate on the mechanisms and the extent of these phenomena, EMF have been
reported to induce cell death 55-57
, cell stress responses 58-61
, DNA damage 62, 63
, tumorigenesis 64-
16
Sam H. Au Microfluidics and Cell Studies
67, and a number of other cellular processes
68. These phenomena have been studied primarily for
―extremely low frequencies‖ (50/60 Hz) commonly used in power transmission and electrical
appliances and ―ultra high frequencies‖ (MHz-GHz) used for mobile phone and wireless
communications. Between these frequency ranges, however, little work has been done on the
potential effects of EMF on cell viability and behavior. Although DMF manipulates droplets by
applying potentials of hundreds of volts at frequencies of approximately 1-18 kHz, little work
has been done to characterize the potential for effects of DMF actuation and associated EMF on
cell fitness. Without a comprehensive study of putative effects of DMF manipulation on cell
fitness, especially subtle genome-level effects, researchers are at risk of mistaking DMF-driven
artifacts as biologically significant results.
A third challenge is long-term device operation. Many cell-based applications such as cell-based
qPCR, immunocytochemistry, metabolomics and cytotoxicity assays require extensive sample
handling and preparation which can increase cost, labour and experimental error with each
additional manual handling step. One of the major advantages of microfluidics is integration of
multiple steps onto a single platform which significantly reduces the required manual handling
steps. A challenge however for many cell applications is phenotypic variation in cell types
(described in section 1.2), which is often complicated by non-uniform in vitro cellular
microenvironments. This is especially true for experiments which require long-term cell culture
processes since cells may be exposed to differential environmental factors for days at a time.
Actively mixing droplets improves the distribution of soluble factors and cells which
homogenizes the cellular microenvironments. Although cell culture and analysis has been
previously integrated onto microfluidic devices42
, many biologically-relevant complex cell
17
Sam H. Au Microfluidics and Cell Studies
applications cannot be conducted on digital microfluidics until robust platforms capable of
continuous mixing over multiple days to maintain homogeneity are developed.
1.4 Thesis Objectives
The primary goal of the work described in this thesis is to address the challenges described above
in section 1.3. Specific aims toward achieving this goal are outlined below:
1) Improve device longevity by delaying or inhibiting the biofouling of protein-rich cell
solutions on DMF surfaces without negatively affecting cell fitness (Chapter 2).
2) Investigate potential genome-level effects of DMF manipulation on mammalian cells
(Chapter 3).
3) Develop a DMF platform capable of robust, multi-day culture and analysis of a number
of microorganisms (Chapter 4).
4) Develop a physiologically relevant in vitro liver model on DMF for pharmacology
applications (Chapter 5).
18
Sam H. Au Pluronic Additives to Inhibit Device Failure
Chapter 2
Pluronic Additives to Inhibit Device Failure
Biofouling in microfluidic devices limits the type of samples which can be handled and the
duration for which samples can be manipulated. Despite the cost of disposing fouled devices,
relatively few strategies have been developed to tackle this problem. Here, we have analyzed a
series of eight amphiphilic droplet additives, Pluronic co-block polymers of poly(propylene
oxide) (PPO) and poly(ethylene oxide) (PEO), as a solution to biofouling in digital microfluidics
using serum-containing cell culture media as a model fluid. Our analysis shows that species with
greater PPO content are superior for enabling droplet motion and reducing biofouling. Two of
the tested species, L92 and P105, were found to lengthen device lifetimes by 2-3 times relative to
additives used previously when used at optimal concentrations. Pluronics with low PEO content
like L92 were found to be cytotoxic to an immortalized mammalian cell line, and therefore, we
recommend that Pluronic additives with high PEO content and greater or equal to 50% PEO
composition, like P105, be used for digital microfluidic applications involving cells. Finally,
contact angle measurements were used to probe the interaction between Pluronic-containing
droplets and device surfaces. Strong correlations were found between various types of contact
angle measurements and the capacity of additives to reduce biofouling, which suggests that
contact angle measurements may be useful as a tool for rapidly screening new candidates for the
potential to reduce biofouling. We propose that the work in this chapter will be useful for
scientists and engineers who are developing digital microfluidic platforms for a wide range of
applications involving protein-containing solutions, and in particular, for applications involving
cells.
19
Sam H. Au Pluronic Additives to Inhibit Device Failure
2.1 Introduction
Biofouling, or unwanted adsorption of biomolecules to surfaces, is a serious problem for a wide
range of biomedical applications including implanted medical devices, bioreactors, and filtration
membranes52, 69-72
. Biofouling is exacerbated in microfluidic devices because of the high surface
area to volume ratios in these systems. A number of strategies exist to combat fouling in
channel-based microfluidics73-77
, yet few strategies have been developed to prevent fouling in
digital microfluidic (DMF) systems. In this chapter, I describe work developing a strategy for
minimizing fouling caused by the use of protein solutions (with a special emphasis on cell
culture media) to maximize the lifetime of DMF devices.
Digital microfluidics is a fluid-handling technique in which droplets are manipulated on an open
surface by applying electrical potentials to an array of electrodes embedded underneath an
insulator37
. Because of its ability to precisely dispense, mix, merge and split discrete droplets,
DMF is becoming an increasingly popular tool for biological and biochemical applications78
,
including cell-based assays42, 45, 79, 80
, enzyme assays81-84
, immunoassays85-87
, processing of
samples for proteomic analysis88-93
, applications involving DNA94-96
, and clinical sample
processing and analysis.97
DMF device surfaces are typically coated with a fluorinated polymer
such as Teflon-AF®91
; unfortunately, these types of surfaces are susceptible to unwanted protein
adsorption98, 99
. This is particularly problematic for digital microfluidics ‒ when proteins adsorb
and accumulate, the hydrophobic device surface becomes hydrophilic, which slows and
eventually stops aqueous droplet motion, resulting in reduced device lifetimes. Cell culture
media is particularly challenging for DMF ‒ the biofouling caused by high concentrations of
serum (a complex mixture of proteins and other factors) in such solutions makes droplets
immobile on many kinds of DMF devices.
20
Sam H. Au Pluronic Additives to Inhibit Device Failure
Previous strategies for reducing the amount of protein adsorption to surfaces of digital
microfluidic devices include immersion in water-immiscible oils100
, careful modulation of
applied voltage polarities101
, the use of replaceable plastic films54
, and operation on
superhydrophobic surfaces102, 103
. The first strategy100
is useful for some applications, but is not a
universal solution, as these oils are incompatible with miscible solvents such as ethanol or
methanol and nonpolar solutes may partition from aqueous droplets into the oil matrix. The
second strategy101
is also useful in certain circumstances, but is less effective for complex
solutions where different protein species may present positive or negative charges at
physiological pH. The third strategy54
is useful for preventing cross-contamination (a new film
can be used for each experiment), but does not solve the problem of biofouling within a given
experiment. The fourth strategy102, 103
is effective at reducing biofouling, but these surfaces are
often difficult to fabricate and cannot tolerate even a small amount of detergent (such as 0.01%
Tween 20)103
.
The Wheeler Lab has recently developed a general strategy for reducing fouling in DMF relying
on the inclusion of Pluronic additives to droplets used in DMF systems53
. Pluronics are tri-block
copolymers of poly(ethylene oxide) (PEO) and poly(propylene oxide) (PPO) and are known to
reduce protein104-106
and cell107
adsorption to surfaces. In earlier work53
, two types of Pluronics
(Pluronic F68 and Pluronic F127) were evaluated for their capacities to limit protein adsorption
and increase device lifetimes. Since that initial report, the pluronic additive strategy has been
applied to a diverse range of applications14-18, 24-26
on digital microfluidics with no indication of
adverse effects; for example, the activity of alkaline phosphatase is unaltered even in high
concentrations of additive18
(0.1% Pluronic F127). Of course, there may be applications in which
21
Sam H. Au Pluronic Additives to Inhibit Device Failure
pluronics is problematic; in such cases, the additive may be removed using digital microfliudic
solid-phase extraction108
.
Although Pluronic additives have been demonstrated to be useful for digital microfluidics, the
recent applications of DMF devices for increasingly complex processes such as cell culture and
assays42, 45, 79, 80
(which necessitate the long-term actuation of solutions containing high
concentrations of proteins such as cell culture media and cell lysate) led us to conduct a more
exhaustive study to find a better solution for biofouling. Here, we have evaluated eight different
Pluronic formulations over a range of concentrations based on their (a) ability to enable the long-
term actuation of protein-containing solutions, (b) effects on surface wettabilities, and (c)
compatibility with mammalian cell adhesion and proliferation. Our objectives were to discover a
superior additive to increase DMF device lifetimes (i.e., to reduce analyte losses during fluid
handling and prevent droplet sticking when working with protein solutions), and to characterize
the mechanism(s) by which Pluronic additives enable the actuation of protein containing
solutions by DMF. We speculate that this study will be useful for scientists and engineers who
are developing digital microfluidic analysis platforms for applications involving protein-
containing solutions, and in particular, for applications involving cells.
2.2 Experimental
2.2.1 Reagents and Materials
Unless specified otherwise, reagents were purchased from Sigma-Aldrich (Oakville, ON). Most
Pluronics (BASF Corp., Germany) were generously donated by Brenntag Canada (Toronto, ON);
Pluronic F-68 was from Sigma-Aldrich. Parylene-C dimer was obtained from Specialty Coating
Systems (Indianapolis, IN). Teflon-AF was from DuPont (Wilmington, DE), and A-174 silane
was from GE Silicones (Albany, NY).
22
Sam H. Au Pluronic Additives to Inhibit Device Failure
2.2.2 Device Fabrication
Digital microfluidic devices were fabricated in the University of Toronto Emerging
Communications Technology Institute (ECTI) fabrication facility. Glass substrates bearing
patterned chromium electrodes (used as bottom plates of DMF devices) were formed by
photolithography and etching as described previously42
using photomasks printed with 20,000
dpi resolution by Pacific Arts and Design (Toronto, ON). After patterning, the substrates were
primed for parylene coating by immersing them in silane solution (isopropanol, DI water, and A-
174, 50:50:1 v/v/v) for 15 min, allowing them to air-dry and then washing with isopropanol.
After priming, substrates were coated with Parylene-C (6.9 µm) and Teflon-AF (235 nm).
Parylene was applied by evaporating 15 g of dimer in a vapor deposition instrument (Specialty
Coating Systems), and Teflon-AF was spin-coated (1% in Fluorinert FC-40, 2000 rpm, 60 s) and
then post-baked on a hot-plate (160 °C, 10 min). To enable the application of driving potentials,
the polymer coatings were locally removed from the contact pads by gentle scraping with a
scalpel. Unpatterned top plates were formed by spin-coating indium tin oxide (ITO) coated glass
substrates (Delta Technologies, Stillwater, MN) with Teflon-AF (235 nm, as above).
2.2.3 DMF Longevity Assay
A longevity assay was developed to evaluate the potential for Pluronic additives to increase
device lifetime. The bottom plate of the device used for this assay featured a linear array of 3
square (4×4 mm) actuation electrodes with inter-electrode gaps of 30 µm. Devices were
assembled with a patterned bottom plate and an unpatterned ITO–glass top plate separated by a
spacer formed from 2 pieces of double-sided tape (total spacer thickness 140 µm). To actuate
droplets, driving potentials (200 VPP) were generated by amplifying the output of a function
generator (Agilent Technologies, Santa Clara, CA) operating at 5 kHz. Droplets were
23
Sam H. Au Pluronic Additives to Inhibit Device Failure
sandwiched between the two plates and actuated by applying driving potentials between the top
electrode (ground) and sequential electrodes on the bottom plate via exposed contact pads on the
bottom plate. Droplet actuation was monitored and recorded by a CCD camera mounted on a
lens.
The longevity assays were used to evaluate RPMI 1640 cell culture medium with 10% fetal
bovine serum (FBS) (Life Technologies/Invitrogen Canada, Burlington, ON) containing one of
eight Pluronic additives at a concentration ranging from 0.0 to 0.15% (w/v). Each concentration
was evaluated 3 times on 3 different devices. During each assay, a 4-µL droplet was actuated in a
five-step process as depicted in Figure 2.1. Briefly, (step 1) with a droplet over electrode 1,
electrode 2 was charged to initiate droplet motion (Figure 2.1A/B). (Step 2) Once the droplet had
travelled to the middle of electrode 2, electrode 3 was charged (while electrode 2 remained
charged) (Figure 2.1C). This ensured smooth droplet transitions between electrodes without
pause. (Step 3) Once the droplet had travelled onto electrode 3, the potential was removed from
electrode 2 (Figure 2.1D). (Step 4) The potential was removed from electrode 3 once the droplet
reached the end of the electrode, and the droplet was moved back to its original position (Figure
2.1E-F). (Step 5) Steps 1-4 were repeated in the same manner reversing directions at the end of
each cycle until the device failed. As droplets containing cell media were actuated across
electrodes in this manner, the speed of the droplets decreased over time. Device failure was
defined as any case in which a droplet required more than 15 seconds to complete a movement
step from one electrode to the next. The number of steps and the time until device failure were
recorded for each condition.
24
Sam H. Au Pluronic Additives to Inhibit Device Failure
Figure 2.1 Schematic device lifetime assay operation. Grey squares represent uncharged (non-
potentiated) electrodes and yellow squares represent charged (potentiated) electrodes. In (A), an
electrical potential is applied to electrode 2 to initiate droplet motion. In (B), the droplet begins
to move onto electrode 2. In (C), when droplet is halfway over electrode 2, a potential is applied
to electrode 3 to ensure continuous droplet motion. In (D), the potential is removed from
electrode 2 once droplet reaches electrode 3. In (E), once droplet has moved to the end of
electrode 3, a potential is applied to electrode 2 to change the direction of movement. In (F), the
droplet begins to move across electrode 2.
25
Sam H. Au Pluronic Additives to Inhibit Device Failure
For concentration dependent studies, the actuation times as a function of Pluronic concentration
in cell media were fit to a lognormal curve:
t is the actuation time [s]
c is the concentration of Pluronic in cell media [mol/L]
A, width are derived constants
2.2.4 Contact Angle Measurements
Contact angle measurements were conducted on single-plate DMF devices (i.e., no top-plate)
with a single 1×1 cm square electrode. In each experiment, a 4-µL droplet of RPMI 1640 cell
culture medium with 10% FBS was positioned on top of the electrode, and the contact angle was
measured using the sessile drop fitting method on a Drop Shape Analysis System (Krüss
DSA100, Hamburg, Germany). Each droplet contained one of eight Pluronic additives at a
concentration ranging from 0 to 0.15% (w/v), and each concentration was evaluated 2 times on 2
different devices. In some experiments, the non-potentiated contact angles (i.e., contact angles of
droplets with no potentials applied) were measured every 2.5 minutes for 20 minutes. In other
experiments, a grounded platinum wire (0.25 mm diameter) was inserted into the top of each
droplet, and the electrodynamic contact angles were measured before, during, and 10 seconds
after the application of a 30 second 200 Vpp 5 kHz potential. Electrodynamic contact angle
hysteresis was defined as the difference between the contact angle observed directly before the
application of potential and that observed after the potential is withdrawn (allowing for a few
seconds for the droplet to stabilize).
𝑡 = 𝑡0 + 𝐴𝑒 (−𝑙𝑛 𝑐 𝑐0
2
𝑤𝑖𝑑𝑡
(2.1)
26
Sam H. Au Pluronic Additives to Inhibit Device Failure
2.2.5 Cell Growth and Viability Assay
Pluronic cytotoxicity experiments were conducted using the Chinese Hamster Ovary (CHO) cell
line. Cells were grown in T-25 flasks in an incubator at 37°C with 5% CO2. At the beginning of
each experiment, cells were detached using a solution of trypsin (0.25% w/v) and EDTA (1 mM)
for 5 minutes and then centrifuged at 173 x g for 5 minutes. The supernatant was removed and
the cells were resuspended at 19,000 cell/cm2 in complete cell culture media (50% DMEM, 40%
Ham’s F12, 10% FBS) containing 0.02% (w/v) Pluronic L62, L64, F68, L92 or P105 and seeded
into 24-well plates. The well plates were stored in an incubator at 37°C with 5% CO2, and each
day for 3 days, cells were collected (using trypsin/EDTA and washing, as above) from wells and
counted using a hemocytometer (Hausser Scientific, Horsham, PA) using the trypan blue
exclusion method. Cells were imaged using a Leica DM2000 microscope (Leica Microsystems
Canada, Richmond Hill, ON). All cell experiments were conducted in triplicate.
2.2.6 Critical Micelle Concentration Determination
Critical micelle concentrations (CMCs) of Pluronic F68, L92 and P105 in RPMI 1640 cell
culture media were determined using the Pyrene solubilization method109
. Briefly, 10 µL aliquots
of 60 µM pyrene in acetone were pipetted into 1.5 mL microcentrifuge tubes and the acetone was
allowed to evaporate. 1 mL aliquots of media containing 0.005%-5.0% (w/v) Pluronics were
added to each tube such that the final pyrene concentration was 6 × 10-7
M and the tubes were
incubated at 65°C for 3 hours and then at 25°C overnight. Samples were transferred to quartz
cuvettes and analyzed using a Fluoromax-3 fluorescence spectrometer (Horiba Jobin Yvon,
Edison, NJ) with excitation at 333 nm and 339 nm. The ratios of the emission intensities at 380
nm resulting from both excitation wavelengths were used for CMC determination. For dilute
solutions of Pluronics this ratio is independent of concentration, but as the concentrations are
27
Sam H. Au Pluronic Additives to Inhibit Device Failure
raised, the ratio is observed to increase. The Pluronic concentrations at which the ratios begin to
increase were taken as the CMCs.109
Media without serum was used for these measurements
because the serum was found to interfere with the analysis.
2.2.7 Statistical Analysis
Statistical analysis was conducted using JMP Statistical Discovery Software (SAS Institute,
Cary, NC). Linear least squares regression was applied to the maximum time and number of
steps as a function of Pluronic molecular weight, PPO chain length, PEO chain length, percent
PEO content, hydrophilic-lipophilic balance, initial non-potentiated contact angle, change in
non-potentiated contact angle over 20 minutes, contact angle during application of potential,
difference in contact angles during and after application of potential, and difference in contact
angles before and after application of potential (i.e., the electrodynamic contact angle hysteresis).
2.3 Results and Discussion
Digital microfluidics (DMF) is a fluid-handling technique in which discrete microdroplets can be
dispensed, merged, mixed and split. As DMF becomes an increasingly popular tool for biological
and biochemical applications, methods for reducing biofouling are imminently needed. This is
particularly the case for applications involving cells42, 45, 79, 80
, which require complex, long-term,
and multi-step experiments. A previous study53
reported the capacity of two solution additives,
Pluronics F68 and Pluronics F127, to reduce the extent of biofouling in digital microfluidics.
Pluronics (also known as poloxamers) are PEO and PPO tri-block copolymers (PEOm – PPOn –
PEOm) with variable PEO and PPO content which controls the degree of hydrophobicity.
The eight Pluronics species evaluated in this work (L35, F38, L44, L62, L64, F68, L92, and
P105), listed in Table 2.1, were chosen to cover a wide range of physical and chemical
28
Sam H. Au Pluronic Additives to Inhibit Device Failure
properties. For example, the PPO chain lengths vary from 16 units (L35 and F38) to 54 units
(P105) and the PEO content varies from 20% (L62 and L92) to 80% (F38 and F68). For each
PEO content percentage, two different PPO lengths were chosen ‒ for example, L62 and L92
each have 20% PEO content, but have average PPO lengths of 30 and 47 units, respectively. To
identify a strategy for preventing biofouling and to gain a better understanding of the
mechanisms behind device fouling and inhibition thereof, we evaluated the eight different
Pluronics over a broad range of concentrations based on their (a) ability to enable the long-term
actuation of complex protein containing solutions by digital microfluidics, (b) effect on droplet
wetting, and (c) compatibility with mammalian cellular adhesion and proliferation.
29
Sam H. Au Pluronic Additives to Inhibit Device Failure
Table 2.1 Physical properties of Pluronics (PEOm – PPOn – PEOm) used in this study
Pluronic Average
Molecular
Weighta
Ave.
PPO
Chain
Length
(n)
Ave.
PEO
Chain
Length
(m)
% PEO
Content
Hydrophilic-
Lipophilic
Balance
(HLB)a
CMC
in
Media
25°C
(%
wt/v)b
Batch/Lot
Numberc
L35 1900 16 11 50 18-23 WPOE579B
F38 4700 16 46 80 >24 WPMD569B
L44 2200 21 11 40 12-18 WPHD524B
L62 2500 30 8 20 1-7 USXW112110
L64 2900 30 13 40 12-18 USXW110132
F68 8400 30 75 80 >24 1.05 097K2410
L92 3650 47 10 20 1-7 0.02 WPIE577B
P105 6500 54 38 50 12-18 0.27 WPIC572B
a Provided by Manufacturer
b Critical Micelle Concentrations in RPMI 1640 cell culture medium were measured using the pyrene solubilization
method as described in the Methods and Materials section.
c Pluronics were obtained from BASF Corp (a generous donation from Brenntag Canada), except for F68 which was
obtained from Sigma Aldrich
2.3.1 Device Lifetime
Eight species of Pluronics were screened at a concentration of 0.02% (wt/v) for the ability to
prolong the motion of droplets of cell culture medium containing 10% fetal bovine serum. This
initial concentration of Pluronics (0.02%) was chosen to balance two factors ‒ on one hand,
concentrations of 0.05% (F6845
) and 0.08% (F12753
) are known to be useful for droplet
manipulation in DMF; on the other hand, some Pluronics have been shown to be toxic to cells at
moderate-to-high concentrations110
that vary depending on which species is used. Thus, 0.02%
30
Sam H. Au Pluronic Additives to Inhibit Device Failure
was used as an initial concentration to balance these two effects ‒ high enough to facilitate
droplet movement but low enough to potentially reduce cell toxicity.
The data from the initial screen is shown in Figure 2.2 and it leads us to three conclusions. First,
Pluronic PPO chains must be above a threshold of ~30 molecular units to enable motion of
droplets of cell culture media containing 10% serum. As shown, droplets failed to move when
PPO chain lengths were 21 or less (L44, F38 and F35) (identical to the case in which no
additives were present), whereas chain lengths above 30 enabled droplet motion (F64, F68, L62,
L92 and P105). Second, the maximum actuation time and the number of successful droplet
movement steps before device failure generally increased with increasing PPO chain lengths.
Specifically, for a given ratio of PEO to PPO, the longer PPO chain Pluronic had superior droplet
movability ‒ for example, L44 (21 PPO units) did not enable droplet motion while L64 (30 PPO
units) did. Third, PPO chain-length had a greater influence on droplet movability than percent
PEO. For example, for Pluronics with a PPO chain length of 30 molecular units, PEO contents of
20% (L62), 40% (L64) and 80% (F68) all enabled droplet motion. These findings are consistent
with literature reports of the importance of longer PPO lengths104-106
for reducing protein
adsorption to surfaces.
31
Sam H. Au Pluronic Additives to Inhibit Device Failure
Figure 2.2 Device longevity assay ‒ initial screen. Droplets containing cell culture media with or
without one of eight different Pluronic additives at 0.02% (w/v) were actuated repeatedly across
a device until movement failure was observed. The maximum actuation time (left axis) and
maximum number of droplet steps (right axis) for the different Pluronics species are arranged
(left-to-right) by increasing PPO unit length. Error bars are ± 1 S.D.
The effects of the two best-performing additives from the initial screen, Pluronics L92 and P105,
were then investigated as a function of concentration, along with the previous standard for DMF
applications involving cells, F68. As shown in Figure 2.3, device lifetime was concentration-
dependent for all of the Pluronics tested, with optimal concentrations of 0.05%, 0.02% and
0.02% (wt/v) for F68, L92 and P105, respectively. The condition used previously45
for
manipulation of droplets of cell media, 0.05% Pluronic F68, facilitated a maximum actuation
time of 762 ±162 s. Pluronic L92 at 0.02% enabled actuation for 2227 ± 178 s, and Pluronic
P105 at 0.02% enabled actuation for 1470 ±176 s, both of which were statistically significant
2000
1500
1000
500
0
Ma
x A
ctu
atio
n T
ime (
s)
None F38 F35 L44 F68 L64 L62 L92 P105
Pluronic
400
300
200
100
0
Ma
x A
ctu
atio
n S
teps
Time
Steps
32
Sam H. Au Pluronic Additives to Inhibit Device Failure
improvements over F68 at 0.05% (p < 0.05). Interestingly, Pluronics L92 and P105 were most
effective at a narrow distribution of concentrations, while Pluronic F68 was effective over a
broader range of concentrations.
33
Sam H. Au Pluronic Additives to Inhibit Device Failure
Figure 2.3 Device longevity assay ‒ concentration dependance. Droplets containing cell culture
media with Pluronics F68 (A), L92 (B) and P105 (C) at a range of different concentrations were
actuated repeatedly across a device until movement failure was observed, recording the
maximum actuation time (left axis) and maximum number of droplet steps (right axis). The
maximum actuation time data were fit to lognormal curves. Error bars are ± 1 S.D.
L92
B3000
2500
2000
1500
1000
500
0
Max A
ctu
ation T
ime (
s)
0.140.120.100.080.060.040.02
Pluronic Concentration (%)
400
300
200
100
0
Max A
ctu
atio
n S
teps
L92 Time
L92 Steps
AF68
1000
800
600
400
200
0
Max A
ctu
ation T
ime (
s)
0.140.120.100.080.060.040.02Pluronic Concentration (%)
250
200
150
100
50
0
Max A
ctu
atio
n S
teps
F68 Time
F68 Steps
P105C 2000
1500
1000
500
0
Max A
ctu
ation T
ime (
s)
0.140.120.100.080.060.040.02
Pluronic Concentration (%)
300
250
200
150
100
50
0
Max A
ctu
atio
n S
teps
P105 Time
P105 Steps
34
Sam H. Au Pluronic Additives to Inhibit Device Failure
Pluronic micelles have been shown to be effective at encapsulating a number of biomolecules111,
112. Thus, one possibility for the beneficial effects of Pluronic additives on DMF device longevity
is the encapsulation of biomolecules in Pluronic micelles, which may prevent biomolecules from
interacting with the device surface. However, the critical micelle concentrations (CMCs) of
Pluronics F68, L92 and P105 in media were measured to be 1.05%, 0.02% and 0.27% (wt/v),
respectively. As shown in Figure 2.3, the concentrations for F68 and L92 which result in
maximum device longevity are far lower than the CMCs of these species, which suggests that
micelles are not required for improved device performance. Moreover, we propose that the data
in Figure 2.3 suggests that Pluronic-protein interactions (even at sub-CMC concentrations) are
unlikely to be the source of the beneficial effects of Pluronic additives on device longevity. We
estimate the molar concentration of protein in the experimental system to be 45-68 mM
[assuming 3.0-4.5% (wt/v) concentration of protein in serum (as provided by supplier) and 66
kDa average protein molecular weight (of the most abundant protein species, albumin)], while
the molar concentrations of Pluronic F68, L92, and P105 in Figure 2.3 that correlate with the
best device performance are estimated to be 60 µM, 55 µM, and 33 µM, respectively (given the
average molecular weights listed in Table 1). Thus, there are approximately 3 orders of
magnitude more protein molecules than Pluronic molecules in these systems, which suggest that
interactions between proteins and pluronic molecules do not explain the observed effects. Rather,
we propose that the most likely explanation is that Pluronic molecules form a temporary coating
on droplet interfaces, preventing proteins from interacting with the device surfaces. This
assertion is supported by the results of Chen et al.113
, which demonstrated that Pluronic
molecules dissolved in aqueous solvents preferentially form dense ordered layers at solution/air
and solution/solid interfaces.
35
Sam H. Au Pluronic Additives to Inhibit Device Failure
2.3.2 Droplet Wetting
After evaluating the effect of Pluronics on device longevity (described above), we evaluated the
effects of the same panel of Pluronic additives on contact angles measured for droplets
positioned on device surfaces. We note that this is superficially similar to a wide body of
literature114-118
from the early 2000s that sought to model digital microfluidics in terms of
"electrowetting" ‒ i.e., the reduction in contact angle of a droplet upon application of an external
electrical field. We are skeptical of electrowetting as a fluid manipulation model, given that
liquids with no electrowettting behaviour are movable on digital microfluidic devices119
. Thus,
the contact angle measurements presented here were not used to explore the mechanism of
droplet movement, but rather to probe the nature of the effects of Pluronic additives on protein
adsorption to surfaces.
As a first step, the contact angles were measured for droplets of cell culture media with 10% fetal
bovine serum containing 0.02% (wt/v) of each of the eight different Pluronic additives in a non-
potentiated state (i.e., with no voltage applied). A representative picture of such a droplet is
shown in Figure 2.4A. Upon observation of the results, the additives were categorized into two
classes. (1) In some cases (Figure 2.4B), droplet contact angles decrease as a function of time.
This behaviour is typified by media not containing any additives (blue squares in Fig. 2.4B), and
is likely an effect of protein adsorption to the surface as time progresses -- as more protein
adsorbs, the surface becomes more hydrophilic, resulting in lower contact angles. Interestingly,
the additives that exhibit this behaviour (with decreasing contact angles as a function of time)
were incapable of supporting droplet movement (see Figure 2.1) with one exception: Pluronic
F68. (2) In other cases (Figure 2.4C), droplet contact angles are fairly constant as a function of
time. This behaviour is typified by liquids not containing any proteins, like DI water (red upside-
36
Sam H. Au Pluronic Additives to Inhibit Device Failure
down triangles in Figure 2.4C). We propose that the additives that facilitate this behaviour (L64,
L62, L92, and P105) for protein-containing solutions are effective at preventing protein
adsorption on this time scale. Note that the two behaviours (reducing contact angle vs. constant
contact angle as a function of time) correlate approximately with PPO chain length. With the
exception of Pluronic F68, the additives in Figure 2.4B have PPO chain lengths <30, and the
additives in Figure 2.4C have chain lengths ≥30.
The non-potentiated contact angle data in Figure 2.4B and 2.4C show another trend. Immediately
after depositing the droplet on the surface, the non-potentiated contact angles were lower for
additives in Figure 2.4C than for the additives in Figure 2.4B. This is a useful observation, which
we propose may be useful for screening additives for utility for digital microfluidics. In fact,
when examined quantitatively, there is an inverse correlation (R2 = 0.78) between maximum
actuation time on DMF devices (from Figure 2.2) and the initial non-potentiated contact angle,
which is plotted in Figure 2.4D. For example, the two Pluronic additives which enabled the
longest lifetimes were L92 and P105 (from Figure 2.2), and these additives were observed to
have the lowest initial non-potentiated contact angles of 82° and 85°. It is likely that the primary
reason for droplet movement failure in the device longevity assay is the adsorption of proteins to
the device surface. Under this assumption, the data in Figure 2.4D suggests that the more
attracted the Pluronic molecules are to the surface (resulting in lower initial contact angles), the
greater the extent of protection of the surface from protein adsorption. This finding is consistent
with literature on the influence of wettability on protein adsorption – in general, lower contact
angles for aqueous droplets on uncharged surfaces are known to be associated with reduced
protein adhesion.120-122
Pluronics with longer PPO chains assemble more rapidly on hydrophobic
37
Sam H. Au Pluronic Additives to Inhibit Device Failure
device surfaces and also are be more difficult to displace once assembled.51, 123
Both of these
phenomena are likely to be useful for preventing protein adsorption on DMF device surfaces.
Figure 2.4 Non-potentiated contact angles. Picture (A) of a non-potentiated droplet of cell
culture media on a Teflon-AF coated surface. Contact angles of media containing 0.02% (wt/v)
Pluronics were measured for 20 minutes, and then categorized as having large changes over time
(B) or remaining constant (C). Linear least squares regression (D) of maximum droplet actuation
time (from Fig. 2.2) vs. non-potentiated contact angle (R2 = 0.78). Error bars are ± 1 S.D.
After evaluating the relationship between non-potentiated contact angle and device longevity, we
turned our attention to electrodynamic contact angles. In DMF devices, when an aqueous droplet
positioned on an insulator is positioned over a charged electrode (i.e., with potential applied), the
droplet is observed to assume a wetted geometry (Fig. 2.5A). When the electrical potential is
then removed, the droplet is observed to recover to a non-wetted geometry on the now de-
charged electrode (Fig. 2.5B). The electrodynamic contact angles were measured for the series of
C
Initial, Non-potentiated
0
500
1000
1500
2000
Max A
ctu
atio
n T
ime (
s)
F38L35L44
F68
L64
L62
L92
P105
80 85 90 95 100
Initial Non-potentiated Contact Angle
Linear Fit
Max time (s) = 9293.9891 - 92.962056*Passive Starting Angle
RSquare
RSquare Adj
Root Mean Square Error
Mean of Response
Observations (or Sum Wgts)
0.782406
0.746141
403.986
702.5521
8
Summary of Fit
Model
Error
C. Total
Source
1
6
7
DF
3521027.7
979227.9
4500255.6
Sum of
Squares
3521028
163205
Mean Square
21.5743
F Ratio
0.0035*
Prob > F
Analysis of Variance
Intercept
Passive Starting Angle
Term
9293.9891
-92.96206
Estimate
1855.19
20.01416
Std Error
5.01
-4.64
t Ratio
0.0024*
0.0035*
Prob>|t|
Parameter Estimates
Linear Fit
Bivariate Fit of Max time (s) By Passive Starting Angle
R2 = 0.78
D
A BContact Angle Reduces as a Function of Time
110
100
90
80
70
60
Sta
tic C
onta
ct
Angle
20151050
Time (min)
Media
F38
L35
L44
F68
110
100
90
80
70
60
Sta
tic C
onta
ct
Angle
20151050
Time (min)
Water
L64
L62
L92
P105
Contact Angle Stable as a Function of Time
38
Sam H. Au Pluronic Additives to Inhibit Device Failure
cell culture media formulations containing Pluronics; as far as we are aware, this is the first time
electrodynamic contact angles have been measured for Pluronic-containing liquids. As was the
case for non-potentiated contact angles, there was a strong inverse correlation (R2 = 0.85)
between maximum actuation time on DMF devices (from Figure 2.2) and charged contact angle,
which is plotted in Figure 2.5C. For example, the Pluronic additives with the most beneficial
effects on device longevity, L92 and P105, had the lowest charged contact angles of 44.6° and
44.5°. Moreover, there was an even stronger correlation (R2 = 0.88) between the maximum
actuation time on DMF devices and the change in contact angles between the charged and de-
charged states, which is shown in Figure 2.5D. Again, L92 and P105 had the highest observed
contact angle changes of 45.5° and 45.4°. Interestingly, no correlations were found between
electrodynamic contact angle hysteresis (i.e., the difference between non-potentiated contact
angles measured before and after charging the electrode) and maximum actuation time or
PPO/PEO chain lengths. Regardless, the contact angle results presented in Figures 2.4-5 are
useful predictors of the effects of additives on device lifetime, and we suggest that future
investigation of new additives may want to use similar measurements as a screen for beneficial
effects.
39
Sam H. Au Pluronic Additives to Inhibit Device Failure
Figure 2.5 Electrodynamic contact angles. Pictures of droplets on a device during (A) and after
removal (B) application of a 200 VPP potential. Linear least squares regression of maximum
droplet actuation time (from Fig. 2.2) vs. electrodynamic contact angle during charging (R2 =
0.85) (C) and vs. the difference in contact angle between the charged to de-charged states (R2 =
0.88) (D).
2.3.3 Compatibility with Cells
A primary goal for the work in this chapter was to identify Pluronic additives that do not
interfere with cell culture. To this end, Chinese Hamster Ovary (CHO) cells were cultured for 3
days in the presence of the five best performing Pluronics species from the initial screens at
concentrations of 0.02% (wt/v) to characterize their affects on viability (as a marker for toxicity)
and cell density (as a marker for effects on proliferation). Upon observation of the results, the
40
Sam H. Au Pluronic Additives to Inhibit Device Failure
five additives were categorized into two classes, non-toxic and toxic, and the data are
summarized in Figure 2.6. As shown, Pluronics F68 and P105 were found to be non-toxic to
CHO cells and had little or no affect on proliferation (Figure 2.6A). Moreover, as shown in 2.6C,
these two additives had little or no impact on cell morphology or adhesion to the substrates. In
contrast, Pluronics L62, L64 and L92 had significant cytotoxicity (with viabilities decreasing to
less than 20% within 3 days) and in addition prevented cell proliferation (Figure 2.6B). This
observation is supported by previous work110
in which Pluronic L64 was found to be cytotoxic to
epithelial cell lines and primary microphages in vitro. Interestingly, the Pluronics species
categorized as non-toxic had higher PEO content (50% or higher) and higher hydrophilic-
lipophilic balance (HLB) values than the species categorized as toxic (Table 2.1). We speculate
that the more lipophilic Pluronics (with lower HLB) are more likely to interact with and disrupt
phospholipid bilayers, compromising cell membrane integrity. It is clear that Pluronics P105 and
F68 at 0.02% are not detrimental to CHO cells. It is reasonable to assume that many
immortalized cell lines will fare similarly, but we caution that cell phenotypes vary considerably,
so effects should be tested before use.
41
Sam H. Au Pluronic Additives to Inhibit Device Failure
Figure 2.6 Pluronic cytotoxicity. Viability (right axes) and density (left axes) of CHO cells
cultured over 3 days without Pluronics, in Pluronics F68 and P105 at 0.02% (wt/v) (A) and in
Pluronics L62, L64 and L92 at 0.02% (wt/v) (B). Photomicrographs for CHO cells after 3 days
cultured Pluronics-free, in 0.02% F68 and 0.02% P105 (C). Cells show comparable growth rates
and morphologies in the presence of F68 and P105 to standard Pluronics-free conditions.
2.4 Conclusions
We evaluated a series of Pluronic block copolymers as additives for use with digital
microfluidics to reduce protein adsorption and prolong device lifetime. Of the formulations
tested (which included Pluronics F68, which has been used extensively for this purpose), the
additives that yielded the best device performance were Pluronics L92 and P105 at 0.02% (wt/v).
For applications involving mammalian cells, however, P105 is a better choice as L92 is cytotoxic
C
1.5x105
1.0
0.5
0.0
Via
ble
Cell
Concentr
ation (
cell/
mL)
7248240
Time (h)
100
80
60
40
20
0
Via
bility
(%)
L62 Cell Density
L62 Viability
L64 Cell Density
L64 Viability
L92 Cell Density
L92 Viability
1.5x105
1.0
0.5
0.0
Via
ble
Cell
Concentr
ation (
cell/
mL)
7248240
Time (h)
100
80
60
40
20
0
Via
bility
(%)
Pluronics-free Cell Density
Pluronics-free Viability
F68 Cell Density
F68 Viability
P105 Cell Density
P105 Viability
Toxic
Non-Toxic
A
B
Pluronics-free
F68
P105
42
Sam H. Au Pluronic Additives to Inhibit Device Failure
to Chinese hamster ovary cells at 0.02%. More generally, when selecting Pluronic additives to
improve device longevity in DMF devices, we recommend that Pluronic species should have
long PPO chain lengths (30 units or greater). For use with cells, greater PEO content (50% or
more) is likely to be important. To this end, Pluronic species such as F88 and F108 (which were
not evaluated in this chapter) are likely good candidates because they have high PEO content and
PPO chain lengths within the acceptable range determined in this study. In additional
experiments (data not shown), F88 at 0.06% was evaluated to be more effective anti-fouling
additive that is amenable to cells and was used for cell experiments in Chapter 3. It is important
to note that concentration dependence should be evaluated for any future Pluronic additive
candidates since droplet movability is highly dose dependent. Finally, if full device longevity
assays cannot be performed (as may be the case if a large matrix of conditions is being
evaluated), we recommend that contact angles may be a useful screen ‒ lower initial non-
potentiated and electrodynamic/charged contact angles, and higher contact angle differences
between charged and de-charged states correlate with improved performance on device. As we
advance our understanding of the mechanisms behind biofouling and biofouling prevention in
digital microfluidic devices, we will greatly increase its suitability for an ever-greater range of
applications.
43
Sam H. Au Effects of Digital Microfluidic Actuation on Cell Fitness
Chapter 3
Effects of Digital Microfluidic Actuation on Cell Fitness
The potential benefits of using new technologies such as microfluidics for life science
applications are exciting, but it is critical to understand and document potential biases imposed
by these technologies on the observed results. In this chapter, the first study of genome-level
effects on cells manipulated by digital microfluidics is described. These effects were evaluated
using a broad suite of tools: cell-based stress sensors for heat shock activation, single-cell
COMET assays to probe changes in DNA integrity, and DNA microarrays and qPCR to evaluate
changes in genetic expression. The results lead to two key observations. First, most DMF
operating conditions tested, including those that are commonly used in the literature, result in
negligible cell-stress or genome-level effects. Second, for DMF devices operated at high driving
frequency (18 kHz) and with large driving electrodes (10 mm x 10 mm), there are significant
changes in DNA integrity and differential genomic regulation. We hypothesize that these effects
are caused by droplet heating. We recommend that for DMF applications involving mammalian
cells that driving frequencies be kept low (≤10 kHz) and electrode sizes be kept small (≤5 mm)
to avoid detrimental effects.
3.1 Introduction
Microfluidic approaches are growing in popularity; for example, the ―organ on a chip‖ concept
has attracted attention as a potentially disruptive new tool for drug discovery and screening11, 124-
127. But as interest in these techniques grows, so too does the need to better understand the effects
that microfluidic culture conditions have on cell phenotype, fitness, and health. For example,
microchannel-based cell culture is known to cause cells to experience significant changes in
glucose consumption, proliferation, and stress levels128
. Likewise, exposure to
44
Sam H. Au Effects of Digital Microfluidic Actuation on Cell Fitness
polydimethylsiloxane (PDMS), a common material used to form microchannels, has been found
to alter gene expression129
. Thus, we posit that while the potential benefits of using new
technologies (such as microfluidics) for life science applications are exciting, it is critical that we
take the time to understand and document the potential biases imposed by these technologies on
the observed results.
The topic of how the microenvironment in microfluidic systems alters cell fitness is of particular
interest for digital microfluidics (DMF), a fluid-handling technique in which droplets are
manipulated on an open surface by applying electrical potentials to an array of electrodes
embedded under an insulator40
. A typical DMF device is shown in Figure 1.1, which highlights
the capacity to dispense, mix, merge, and split discrete droplets. These operations are attractive
for cell-based applications, and DMF has recently become popular for handling43, 44, 46, 48, 130-132
and culturing42, 45, 50, 133
mammalian cell lines, micro-organisms80, 134
, primary mammalian
cells47
, and 3D cell constructs49
.
Unfortunately, the scientific literature contains little information about the effects of digital
microfluidic droplet actuation on cell health. Barbulovic-Nad et al.42, 45
reported that mammalian
cells exposed to one set of DMF operating conditions had similar viabilities and proliferation
rates when compared to control (non-actuated cells), and Au et al.80
reported similar results for
bacteria, algae, and yeast. However viability and proliferation are crude measures of cell fitness
which may ignore the vast range of subtle effects that may be caused by DMF actuation – stress
responses, DNA damage, differential DNA expression, etc. It is important to catalogue these
potential effects, particularly if DMF becomes useful for screening for cell phenotype differences
in the pharmaceutical industry50
. Moreover, it would be useful to evaluate the potential effects of
45
Sam H. Au Effects of Digital Microfluidic Actuation on Cell Fitness
DMF droplet manipulation over a wide range of operating conditions to determine which should
be used and which should be avoided.
This chapter describes work evaluating the genome-level effects of DMF actuation on cells. A
number of cell fitness indicators were evaluated: heat shock stress response135, 136
, DNA
integrity62, 63
and changes in genetic transcription levels56, 58, 68
. We propose that the results
described here will serve as a useful guide for the rapidly growing number of research groups
that are adopting digital microfluidics as a tool for applications involving cells. Moreover, we
propose that the suite of tests described here (cell stress sensors, COMET assays, and
microarrays/qPCR) represents a useful measuring stick for probing the effects of any new
technology that is applied to applications involving cells.
3.2 Experimental
Unless specified otherwise, reagents used in this chapter were purchased from Sigma-Aldrich
(Oakville, ON). Parylene-C dimer was obtained from Specialty Coating Systems (Indianapolis,
IN). Teflon-AF was from DuPont (Wilmington, DE), and A-174 silane was from GE Silicones
(Albany, NY).
3.2.1 Device Fabrication and Operation
Digital microfluidic devices were fabricated in the University of Toronto Emerging
Communications Technology Institute (ECTI). Bottom plates bearing square electrodes arranged
in 2 x 2 arrays of 10, 5, 2.5, or 2 mm wide square chromium electrodes (used as bottom plates of
DMF devices) were formed by photolithography and coated with a Parylene-C insulating layer
(6.9 µm) and Teflon-AF hydrophobic layer (235 nm) as described previously42, 80
. Unpatterned
top plates were formed by spin-coating indium tin oxide (ITO) coated glass substrates (Delta
46
Sam H. Au Effects of Digital Microfluidic Actuation on Cell Fitness
Technologies, Stillwater, MN) with Teflon-AF (235 nm, as above) and were assembled with
bottom plates by adherent spacers formed from four pieces of double-sided tape (total spacer
thicknesses 280 µm). To drive droplets or expose cells to potentials, square AC potentials were
applied to actuation electrodes on the bottom plate relative to the counter-electrode on the top-
plate in a circular pattern using a custom high voltage switching system (Astinco, Inc., Markham,
ON, Canada).
3.2.2 Cell Culture and Stress Conditioning
WeHi-3B cells (ATCC, Manassas, VA) were grown for 3-4 days at 37°C and 5% CO2 in IMDM
media supplemented with 10% fetal bovine serum (Life Technologies, Inc., Burlington, ON,
Canada), 2 mM L-glutamine (Life Technologies, Inc.), 100 IU/mL penicillin, and 100 µg/mL
streptomycin. After reaching confluency, spent media was exchanged for fresh media, and
thereafter, the media was exchanged every day for three days. Each aliquot of spent media was
centrifuged (300 g, 5 min), the supernatant filtered through a 0.2 µm syringe filter (PALL
Canada Ltd., Saint-Laurent, QC), and the filtrate frozen at -20°C until use. Wild-type Ba/F3 pro-
B murine suspension cells (ATCC) were cultured in complete growth media consisting of RPMI-
1640 supplemented with 10% fetal bovine serum (Life Technologies, Inc.), 10% WeHi-
conditioned media (as above), 100 IU/mL penicillin, 100 µg/mL streptomycin, and 0.06% (wt/v)
pluronic F88 at 37°C and 5% CO2. Stably transformed GFP-HSE Ba/F3 cells137, 138
were treated
similarly with the exception of the addition of 500 µg/mL G418 geneticin (Life Technologies) to
the complete media to maintain selective pressure. Wild-type or GFP-HSE Ba/F3 cells in early
log-phase growth were centrifuged (300 g, 5 min), the supernatant was removed, and the pellet
was resuspended in complete growth media at 2 x 106 cells/mL. In DMF experiments, 70 µL, 18
µL or 5 µL aliquots of the wild-type or GFP-HSE Ba/F3 cell suspension were loaded onto
47
Sam H. Au Effects of Digital Microfluidic Actuation on Cell Fitness
devices and actuated continuously in a circular pattern over 4 electrodes (at approximately 3.33
mm/s) under conditions of varying driving potential (200, 400, 625, or 650 Vpp), driving
frequency (1, 10, or 18 kHz), actuation time (5 or 15 min), and electrode size (2x2, 2.5x2.5, 5x5,
or 10x10 mm).Untreated control cells were handled identically by loading aliquots onto devices,
but without applying driving voltages (i.e., the droplets remained stationary). Heat shock controls
were carried out on 20 or 70 µL aliquots of cell suspension which were loaded onto non-active
devices (as above) and heated on a PMC digital hot plate (Thermo Fisher Scientific, Waltham,
MA) such that droplets were maintained at 42°C, 47°C or 52°C for 5 or 15 minutes. Droplet
temperature was measured with a thermocouple as described in section 3.2.6 information, to
control the actual temperature to within 1°C of set-point. Sodium arsenite controls were carried
out on 100 µL aliquots of cell suspension supplemented with sodium arsenite and incubated for 5
minutes. Several replicates of cells exposed to each condition (DMF actuated, non-actuated
control, heat-shock control, arsenite control) were generated for protein expression/flow
cytometry, COMET assays, oligonucleotide microarrays, and qRT-PCR assays, as described
below.
3.2.3 Cell Stress Evaluation and Flow Cytometry
Aliquots of untreated, DMF-treated, heat-treated, and arsenite-treated GFP-labeled cells were
diluted with 9.9 mL phosphate buffered saline (PBS), centrifuged (300 g, 5 min) and the
supernatant removed. Each aliquot was then resuspended in 1 mL of fresh media (with G418
geneticin), transferred into a well of a 24 well plate, and incubated for 24 hours at 37ºC/5% CO2.
Cells were then collected, centrifuged (300 g, 5 min), the supernatant removed and resuspended
into 2.5 mL PBS. GFP expression was determined using an Epics XL flow cytometer (Beckman
Coulter Canada, Mississauga, ON, Canada) and analyzed using Expo32 Software (Beckman
48
Sam H. Au Effects of Digital Microfluidic Actuation on Cell Fitness
Coulter). The cells were sampled at a rate of ~100 events/s, excited using a 488 nm laser, with
the fluorescent signature detected through a 530/30 nm filter until 5000-20,000 cells were
detected per condition. Histograms of fluorescent intensity were plotted on a logarithmic scale.
3.2.4 Single Cell Gel Electrophoresis COMET Assays
Assays were conducted using a Comet Assay Kit (Trevigen, Inc., Gaithersburg, MD) according
to the manufacturer’s protocols. Briefly, in each assay, a suspension of Ba/F3 cells was diluted to
3 x 105 cells/mL in ice-cold PBS (Mg
2+/Ca
2+ free) immediately after stress or control treatment
(as above). This suspension was combined with agarose solution, spread on a microscope slide
and allowed to gel. Slides were then immersed and incubated in prechilled lysis solution (45
min) and subsequently alkaline unwinding solution (45 min). The slides’ contents were then
electrophoresed in a horizontal gel electrophoresis system (VWR International, LLC, Radnor,
PA) at 0.7 V/cm in a 4°C cold room (60 min) before washing twice in DI water and once in 70%
ethanol. Slides were dried before adding SYBR green and imaging with a Leica DM2000
microscope (Leica Microsystems, Inc., Concord, ON, Canada). Percent fragmented DNA in tail
and Olive moments were quantified using Cometscore™ software (AutoComet, TriTek Corp,
Sumerduck, VA). Each condition was conducted in triplicate with at least 100 cells counted per
sample.
3.2.5 Microarrays and qPCR
For each oligonucleotide microarray or qPCR experiment, a suspension of Ba/F3 cells was
diluted into 1 mL pre-warmed complete growth media in a tissue culture treated well plate (BD,
Franklin Lakes, NJ) immediately after stress or control treatment (as above). The cells were
allowed to recover for 1 hour at 37°C, and then RNA was extracted using Arcturus Picopure
49
Sam H. Au Effects of Digital Microfluidic Actuation on Cell Fitness
RNA Isolation kits (Applied Biosystems, Inc., Foster City, CA) according to the manufacturer’s
guidelines. Purified RNA was stored at -80°C and thawed on ice directly before analysis.
Microarray studies were conducted at the University Health Network Microarray Centre
(Toronto, ON, Canada). Prior to hybridization, each sample was evaluated on an Agilent
Bioanalyzer (Agilent Technologies Canada, Inc., Mississauga, ON) to ensure that it met QC
thresholds (RNA integrity number>9)139
. 60-200 ng of total RNA per sample was labeled using
an Illumina TotalPrep-96 RNA Amplification Kit (Life Technologies) according to the
manufacturer’s protocol. 1.5 ng of the generated cRNA was randomized and hybridized onto
mouse WG-6 v2.0 BeadChip platforms (Illumina, Inc., San Diego, CA) by incubation at 58ºC for
18 hours. Beadchips were washed and stained according to the manufacturer’s protocol and
scanned using an iScan array scanner (Illumina). The dataset is publically available at the Gene
Expression Omnibus (GEO Accession number: GSE43507). Data quality was validated prior to
normalization using Illumina® internal quality control metrics as well the R(v2.10.0)
Bioconductor framework with the Lumi package140
.
Microarray data analysis was conducted using Genespring v.11.5.1 (Agilent Technologies). Two
batches of microarray data were normalized using the Empirical Bayes ComBat accommodation
of batch effects141
using an R script. Each array batch contained 3 untreated control biological
replicates for cross-batch normalization. Data was normalized using a standard quantile method
followed by a ―per probe‖ median centered normalization. A total of 45281 probes were
represented on the mouse array. The data was filtered such that only probes in the upper 80th
percentile of the distribution of intensities were retained so that probes without signal would not
confound subsequent analysis. The filtered set contained 37133 probes. An unsupervised
clustering algorithm using a Pearson centered correlation as a distance metric with average
50
Sam H. Au Effects of Digital Microfluidic Actuation on Cell Fitness
linkage rules was used to build hierarchical trees. Next, an ANOVA was performed using the
Benjamini-Hochberg false discovery rate correction with a multiple testing correction threshold
of p < 0.05. Results were presented in log2 fold change versus untreated controls. Venn diagrams
of significantly expressed probes were created using Venny
(http://bioinfogp.cnb.csic.es/tools/venny/index.html). Putative gene functions were designated
from the AmiGO gene ontology database (http://amigo.geneontology.org) and the Information
Hyperlinked Over Proteins database (http://www.ihop-net.org)142
. Each condition was conducted
in triplicate (n=3) except for non-actuated controls which had a total of 6 replicates (n=6).
For qPCR experiments, total RNA was extracted and purified as above. Reverse transcription
was completed using a Quantitect Reverse Transcription Kit (Qiagen, Inc., Toronto, ON)
according to the manufacturer’s guidelines. The expression stability of beta-2-microgobulin
(B2m) and glyceraldehyde 3-phosphate dehydrogenase (Gapdh) were evaluated in pilot qPCR
runs. Primers (Life Technologies, Inc.) were as follows: B2m forward and reverse primers -
TTCTGGTGCTTGTCTCACTGA and CAGTATGTTCGGCTTCCCATTC; Gapdh forward and
reverse primers - AGGTCGGTGTGAACGGATTTG and
TGTAGACCATGTAGTTGAGGTCA; Dusp1 forward and reverse primers -
GTTGTTGGATTGTCGCTCCTT and TTGGGCACGATATGCTCCAG respectively. qPCR
was conducted on a 7900HT qRT-PCR system (Applied Biosystems, Inc.) using Quantifast
SYBR green PCR kits (Qiagen, Inc.). Three biological replicates and at least two technical
replicates were conducted for cells exposed to each condition. Relative quantification was
conducted according to Pfaffl143
and Rieu and Powers144
with the expression of Dusp1
normalized to both B2m and Gapdh housekeeping genes. Baselines, windows of linearity (by
amplicon group) and Cq thresholds (by sample group) were detected by LinRegPCR145
. The
51
Sam H. Au Effects of Digital Microfluidic Actuation on Cell Fitness
RNA used in qPCR experiments was from the same pool for RNA used for microarray
experiments, whenever possible. Two-tailed student t-tests were conducted to determine p-
values.
3.2.6 Temperature Measurements
70 µL, 18 µL or 5 µL droplets of RPMI-1640 cell culture media supplemented with 10% fetal
bovine serum and 0.06% (wt/v) Pluronic F88 (Brenntag Canada, Toronto, ON) were loaded onto
DMF devices bearing 10 mm x 10 mm, 5 x 5 mm or 2.5 x 2.5 mm electrodes respectively. The
electrodes were then charged for 15 minutes continuously using the methods and system
described in section 3.2.1, with 400 VPP driving potentials at 1, 10, or 18 kHz. A K-type
KMQSS-010U-6 thermocouple (Omega Engineering, Inc., Laval, Canada) was inserted between
the two plates to measure the temperature in each droplet as a function of time. Three replicate
measurements were collected for each of the nine electrode size/frequency combinations.
3.3 Results and Discussion
3.3.1 Preliminary Experiments and Cell-based Stress Sensors
In preliminary experiments, Ba/F3 pro-B murine cells that had been stably transfected to express
green fluorescent protein (GFP) under the transcriptional control of heat shock element (HSE)
promoter137, 138
were used as cell-based sensors to probe for stressful stimuli. GFP-HSE cell
sensors were exposed to a range of DMF operating conditions, including driving potentials of
200-650 Vpp and frequencies of 1-18 kHz applied to droplets actuated continuously for five
minutes on devices with 2 x 2 mm driving electrodes. Stress responses (reported by the
expression of GFP) were evaluated by flow cytometry (Fig. 3.1). As shown, the stress response
of cells actuated by digital microfluidics across all of the conditions (Fig. 3.1B-F) was similar to
untreated control cells (Fig. 3.1A) (mean fluorescent intensities of 1.0 or less). In contrast, cells
52
Sam H. Au Effects of Digital Microfluidic Actuation on Cell Fitness
exposed to chemical or heat shock controls were observed to have increased expression of GFP
relative to untreated controls (Fig. 3.1G-H). This is a useful finding, as heat shock proteins
(many of which are under transcriptional control of HSE) are known to be upregulated under
many different types of stress135, 136
, and it appears that DMF operation in the conditions
described above does not trigger this pathway. There may be other cell responses that might not
be captured by these experiments, which led us to select a sub-set of operating conditions for
further experiments. Specifically, a protocol was developed in which one driving potential (400
Vpp) and three different driving frequencies (1, 10, and 18 kHz), were applied to droplets
containing cells that were continuously actuated for 15 minutes. To enable the use of microarray
analysis (which requires relatively large numbers of cells), we tested the effects of devices with
larger actuation electrodes (up to 10 mm x 10 mm, useful for actuating 70 L droplets) than
those used in the preliminary experiments. These conditions were tested on non-transfected wild-
type Ba/F3 cells and were used for all experiments described below.
53
Sam H. Au Effects of Digital Microfluidic Actuation on Cell Fitness
Figure 3.1 Cell-based stress sensor results. Flow cytometry histograms of GFP-HSE Ba/F3 cells
(GFP transcription under transcriptional control of heat shock element promoter). Cells were (A)
untreated for 5 min, manipulated continuously by DMF on devices with 2 x 2 mm electrodes for
5 min at: (B) 200 Vpp and 10 kHz, (C) 400 Vpp and 1 kHz, (D) 400 Vpp and 18 kHz, (E) 625 Vpp
and 1 kHz, or (F) 650 Vpp and 15 kHz, or (G) heat-shock treated at 42ºC for 5 min, or (H)
exposed to sodium arsenite at 200 µg/mL for 5 min.
Untreated
A
Arsenite control
Co
unts
DMF 200 Vpp 10 kHz
42ºC control
Co
unts
GFP Intensity GFP Intensity
B
DMF 400 Vpp 1 kHz
C
DMF 400 Vpp 18 kHz
D
G H
Co
unts
Co
unts
Co
unts
Co
unts
DMF 625 Vpp 1 kHz
E
Co
unts
DMF 650 Vpp 15 kHz
F
Co
unts
54
Sam H. Au Effects of Digital Microfluidic Actuation on Cell Fitness
Interestingly, as described in the section 3.3.5, some of the DMF operating conditions used in
these experiments resulted in increased droplet temperatures. This observation was unexpected
(and is the first report of this phenomenon), but it added to our motivation to evaluate the effects
of DMF actuation on the cells in a series of assays for DNA damage and expression.
3.3.2 DNA Integrity
To evaluate the effects of DMF actuation on DNA damage, we conducted COMET assays, an
established method of evaluating the extent of DNA damage146
by subjecting individual cells
fixed within a hydrogel to electrophoresis. Intact (undamaged) DNA remains immobilized in the
fixed cells, while fragmented (damaged) DNA electrophoreses out of the cells, forming a
characteristic ―comet-like‖ pattern. As shown in Figure 3.2, DNA damage was observed for cells
actuated by DMF at 18 kHz (Fig. 3.2C), but not at 1 kHz (Fig. 3.2B). Heat-shock controls at
42°C, 47°C and 52°C also showed significant fragmentation (Fig. 3.2D-F). Quantification by
percent DNA found in the tails and the Olive moment (the product of the length and the fraction
of DNA in each tail) indicate that the DNA damage observed for cells actuated at 18 kHz or in
any of the three heat-shock controls was significantly higher than that observed for the untreated
controls (p < 0.01) while cells actuated on DMF at 1 kHz and untreated controls were not
significantly different (Figure 3.2G). Since cells were assayed immediately after DMF actuation,
it is unlikely that the observed DNA damage was a result of apoptotic fragmentation. Heat shock,
on the other hand, has been reported to be a cause of double stranded DNA breaks62
, potentially
through protein damage147
or the bystander effect, in which dying cells release factors lethal to
neighbouring cells148
. Since droplets actuated on DMF devices with 10 mm x 10 mm electrodes
at 18 kHz were (unintentionally) heated (section 3.3.5), the observed DNA fragmentation for the
18 kHz case may be a result of droplet heating.
55
Sam H. Au Effects of Digital Microfluidic Actuation on Cell Fitness
Figure 3.2 Quantification of DNA integrity. Representative photomicrographs of cells assayed
for DNA damage using the single cell gel electrophoresis COMET assay for (A) untreated
controls, DMF-treated for 15 min at 400 Vpp on devices with 10 x 10 mm electrodes at
frequencies of (B) 1 kHz or (C) 18 kHz, or heat-shock treated at (D) 42°C, (E) 47°C or (F) 52°C.
Scale bars represent 50 µm. (G) Percent fragmented DNA (solid red bars) and Olive moment
(striped blue bars) were quantified by CometScore™. Error bars represent one standard deviation
(n=3).
G
Untreated control
A
42 C control
B
DMF - 1 kHz
C
DMF - 18 kHz
D
E
47 C control
F
52 C control
56
Sam H. Au Effects of Digital Microfluidic Actuation on Cell Fitness
In summary, the data in Figure 3.2 suggests that the DNA integrity of cells exposed to DMF
manipulation at 400 Vpp on devices with 10 x 10 mm electrodes was strongly frequency
dependent, with DNA fragmentation ranging from negligible at 1 kHz, to significant at 18 kHz.
3.3.3 Gene Expression – Microarrays
To evaluate if DMF actuation causes changes in DNA expression, oligonucleotide microarrays
were conducted; the complete data set is publically available at the Gene Expression Omnibus
(GEO Accession number: GSE43507). Differentially expressed genes were identified using an
ANOVA with false discovery rate (FDR) q-value cut-off of 0.05. Using a 2-fold change
threshold, 196 unique genes were differentially expressed among the various conditions, and a
heat map of these genes is shown in Figure 3.3. The numbers of differentially expressed genes
observed for cells exposed to the different conditions relative to control cells were 3, 85, 65, 48,
and 82, for DMF/1 kHz, DMF/18 kHz, 42°C control, 47°C control, and 52°C control,
respectively.
Figure 3.3 suggests that there were few effects on gene expression caused by DMF actuation at
1 kHz. For example, the hierarchical clustering of conditions (the top axis in Figure 3.3)
indicates that cells manipulated by DMF at 1 kHz group with untreated controls, in contrast to
cells manipulated by DMF at 18 kHz, which group closely with the externally heated controls.
The three genes that were differentially expressed relative to untreated controls under actuation
by DMF at 1 kHz are ubiquitin-conjugating enzyme E2C (Ube2c), destrin (Dstn) and a predicted
pseudogene (ECG635570). The functions of these genes (Ube2c: cell division; Dstn: cytoskeletal
organization; ECG635570: unknown) are not related to known forms of cell stress (heat,
oxidation, osmotic pressure, etc.) or apoptosis. And as shown in the Venn diagrams in Figure
57
Sam H. Au Effects of Digital Microfluidic Actuation on Cell Fitness
3.4A, these three genes were not differentially expressed in cells exposed to heat-shock. Finally
the magnitudes of the differential expression for these genes were modest (+2.3, +2.2 and +2.1
fold changes for Ube2c, Dstn, and ECG635570); in some analyses, much higher thresholds are
used to identify differentially expressed genes149, 150
.
58
Sam H. Au Effects of Digital Microfluidic Actuation on Cell Fitness
Figure 3.3 Microarray heat map. Each gene shown has a ≥2-fold difference determined by an
ANOVA using the FDR Benjamini and Hochberg multiple testing correction (p<0.05). Cells
were either untreated (n=6), manipulated by DMF for 15 min at 400 Vpp on devices with 10 x 10
mm electrodes at 18 kHz or 1 kHz (n=3 ea.), or heat-shock treated at 42°C, 47°C or 52°C (n=3
ea.). Hierarchies were generated with a Pearson centered correlation tree building algorithm as a
distance metric with average linkage rules. Green shaded cells and red shaded cells represent log
fold-change down-regulated or up-regulated expression versus the mean, respectively.
-4 log fold +4 log fold
DMF 18 kHz
47 C control
52 C control
42 C control
DMF1 kHz
Untreated Control
Hierarchial Clustering (conditions)
Hie
rarc
hia
lC
lust
eri
ng (
genes
)
59
Sam H. Au Effects of Digital Microfluidic Actuation on Cell Fitness
Figure 3.4 Microarray expression comparisons. Venn diagrams comparing the number of
significant probes with ≥2-fold (absolute) change relative to untreated controls common
between: (A) cells manipulated on DMF for 15 min at 400 Vpp on devices with 10 x 10 or 5 x 5
mm electrodes at 1 kHz or 18 kHz frequencies and heat-shock controls (42°C, 47°C or 52°C);
and (B) heat-shock control cells.
52°C control 47°C control 42°C control
DMF –1 kHz
DMF –18 kHz
A
B
60
Sam H. Au Effects of Digital Microfluidic Actuation on Cell Fitness
In contrast to actuation at 1 kHz frequencies, the data in Figure 3.3 clearly shows that cells
exposed to 18 kHz DMF actuation at 400 Vpp on devices with 10 x 10 mm electrodes experience
dramatic changes in gene expression relative to untreated controls. In fact, the hierarchical
clustering of conditions (the top axis in Figure 3.3) suggests that the DMF/18 kHz expression
profile is most closely related to those of the 52°C and 47°C controls. This is consistent with the
observations described in section 3.3.5, which suggest that the conditions imposed on these cells
(i.e., 70 L droplets manipulated on 10 x 10 mm electrodes with 400 Vpp at 18 kHz for 15 min)
are associated with temperature increases from ambient to the 47°C-52°C range. The similarity
between DMF/18 kHz and the 52°C and 47°C controls is further highlighted in the Venn
diagrams in Figure 3.4A—there are 23 and 29 gene expression overlaps with the 52°C and 47°C
controls, respectively. These numbers are large relative to the gene expression overlaps between
the three temperature controls themselves (Fig. 3.4B). Thus, we hypothesize that electrically
driven heating is one of the sources of the differential expression observed for the cells actuated
at 400 Vpp and 18 kHz for 15 min.
Many of the 85 genes that are differentially expressed for cells actuated at 18 kHz are known to
be related to stress and/or apoptosis – these genes are listed in Table 3.1. The magnitudes of the
differential expression of many of these genes are large – for example Fos, Egr2, and Dusp1 are
expressed at -30.5, -7.6, and -7.5-fold relative to untreated controls, respectively (note that the
corresponding fold changes for cells exposed to DMF actuation at 1 kHz are near unity). Eight of
the genes in Table 3.1 are involved in MAPK signaling pathways, which are known to be
activated by heat, oxidation, osmotic pressure, DNA damage and ischemic stresses151
.
Specifically (from Table 3.1), Fos152, 153
, Egr2154
, Dusp1155-158
, Osm159
, Mcm7160
, Dusp2161
,
Ier3162
, and Zfp36163
are all components of or interact with the MAPK pathway. Interestingly,
61
Sam H. Au Effects of Digital Microfluidic Actuation on Cell Fitness
seven of these eight genes are down-regulated for DMF/18 kHz actuation rather than up-
regulated, which is the expected response. Note that heat shocked control cells also showed
down-regulation of these stress genes as well (Table 3.1), suggesting that this phenomenon is a
result of the cells’ inherent heat shock response and not simply an artifact of DMF manipulation.
However, this phenomenon merits future study.
62
Sam H. Au Effects of Digital Microfluidic Actuation on Cell Fitness
Table 3.1 Stress and apoptosis gene summary. Sub-set of differentially expressed genes for
DMF/18 kHz operation on devices with 10 x 10 mm electrodes (with a ≥2 absolute fold change
relative to untreated controls) related to stress and/or apoptosis. Putative functions designated by
AmiGO gene ontology database and Information Hyperlinked Over Proteins (iHOP)142
.
-30
-20
-10 0
10
Fos E
gr2
Dusp1
Axud1
Osm
M
cm
7
Dusp2
Ier3 Z
fp36
Prf1 P
lk3
Nisch
Myd116
Hspd1
Hspa8
Fosl2 Atf4
D
MF
- 18 kH
z
D
MF
- 1 K
Hz
Symbol Entrez
Gene ID
Annotation DMF -
18 kHz
FC
DMF -
1 kHz
FC
52 C
control
FC
47 C
control
FC
42 C
control
FC
Putative Function
Fos 14281 FBJ osteosarcoma
oncogene
-30.5 -1.3 -9.8 -5.4 -2.1 Response to oxidative stress and
extracellular stimulus, positive regulation
of transcription
Egr2 13654 early growth response 2 -7.6 -1.5 -1.0 -1.1 -1.7 Response to stress and DNA damage,
positive regulation of transcription,
myelination, cellular response to organic
substance
Dusp1 19252 dual specificity
phosphatase 1
-7.5 -1.1 -6.9 -6.3 1.0 Response to oxidative stress, cell cycle,
MAPK activity, regulation of apoptosis
Axud1 215418 cysteine-serine-rich
nuclear protein 1
-.5.3 -1.3 -4.3 -3.6 -2.1 Apoptotic process, regulation of
transcription
Osm 18413 oncostatin M -4.9 -1.3 -4.1 -2.6 -3.6 Response to heat, apoptosis, positive
regulation of cell proliferation and
MAPK cascade
Mcm7 17220 minichromosome
maintenance deficient 7
4.0 1.0 -1.1 -1.0 -1.1 Response to DNA damage stimulus, cell
cycle, cell proliferation
Dusp2 13537 dual specificity
phosphatase 2
-3.9 -1.3 -2.0 -1.6 -2.0 Response to oxidative stress, inactivation
of MAPK activity, regulation of
apoptotic process
Ier3 15937 immediate early
response 3
-3.8 -1.1 -3.1 -2.1 -3.0 Response to stress, regulation of
apoptosis and ROS metabolic process
Zfp36 22695 zinc finger protein 36 -3.6 -1.2 -3.8 -3.5 -2.0 Response to stress, mRNA catabolism
Prf1 18646 perforin 1 3.1 1.6 1.5 4.1 8.5 Apoptotic process, cytolysis, immune
response to tumor cell, defense response
to virus
Plk3 12795 polo-like kinase 3 -3.0 -1.3 -2.9 -2..7 -2.1 Response to DNA damage stimulus,
response to ROS, response to osmotic
stress, G1/S transition
Nisch 64652 nischarin 2.6 1.2 1.4 2.0 1.0 Apoptotic process, cell communication,
negative regulation of cell migration
Myd116 17872 protein phosphatase 1,
regulatory (inhibitor)
subunit 15A
-2.4 -1.1 -2.2 -1.7 1.1 Response to stress, apoptotic process,
regulation of translation
Hspd1 15510 heat shock protein 1
(chaperonin)
-2.3 -1.2 -1.7 -2.3 -1.1 Response to stress, B-cell activation, B
cell proliferation, protein folding,
regulation of apoptosis
Hspa8 15481 heat shock protein 8 -2.3 -1.2 1.0 -2.1 1.3 Response to stress, protein folding,
regulation of cell cycle, regulation of
transcription
Fosl2 14284 fos-like antigen 2 -2.1 -1.5 -2.4 -2.5 -1.5 Apoptosis, regulation of fibroblast
proliferation, regulation of transcription,
response to hypoxia
Atf4 11911 activating transcription
factor 4
-2.1 1.2 -1.3 -1.5 -1.2 Response to ER stress, GABA signalling
pathway, regulation of transcription
-30
-20
-10
0
10
Fos Egr2 Dusp1 Axud1 Osm Mcm7 Dusp2 Ier3 Zfp36 Prf1 Plk3 Nisch Myd116 Hspd1 Hspa8 Fosl2 Atf4
DMF - 18 kHz
DMF - 1 KHz
Fold Change
-30 -20 -10 0 +10
63
Sam H. Au Effects of Digital Microfluidic Actuation on Cell Fitness
In summary, it is evident that for actuation for 15 minutes at 400 Vpp on devices with 10x10 mm
electrodes, the frequency of the applied electric field plays a pivotal role in the magnitude and
nature of cell responses to actuation. Low frequency (1 kHz) operation led to the up-regulation of
only 3 probes (with modest magnitudes), none of which are involved in cell stress or death;
while high frequency (18 kHz) operation led to the significant modulation of 85 probes, many of
which are implicated in stress or death responses.
3.3.4 Gene Expression – qPCR
qPCR of dual specificity phosphatase 1 (Dusp1) was chosen as an orthogonal test because it
exhibited strong differential expression from microarray results (described above), and because it
is known to be a gene that responds rapidly to heat stress155, 164, 165
. In addition, Dusp1 is a key
element regulating the MAPK pathway155-158
, which (as described above) appears to be a vital
component of the cellular responses observed in the microarray data. B2m and Gapdh were
chosen as reference genes for these experiments because they demonstrated high stability across
the different conditions in preliminary qPCR validation experiments (stability values less than
0.03 as determined by NormFinder software)166
.
In qPCR experiments, cells were treated on DMF devices with 5 x 5 mm electrodes (in addition
to 10 mm x 10 mm), and with 10 kHz driving frequency (in addition to 18 kHz and 1 kHz). As
shown in Table 3.2, of all the DMF conditions, only cells actuated at the highest frequency (18
kHz) and on the largest (10 mm x 10 mm) electrodes led to a statistically significant (p<0.05)
modulation of Dusp1 expression while cells actuated on smaller electrodes (5 mm x 5 mm) at the
same high frequency and cells actuated at lower frequencies (10 kHz and 1 kHz) on large
electrodes did not have a statistically significant difference relative to untreated controls. This
finding is consistent with the observations of droplet temperature described in section 3.3.5. In
64
Sam H. Au Effects of Digital Microfluidic Actuation on Cell Fitness
addition, the qPCR data is consistent with the microarray data in that Dusp1 expression was
down-regulated by treatment at 47°C and 52°C, but not at 42°C or with any of the other DMF
operating conditions. This finding reinforces the discovery that significant stress responses are
only observed for large electrodes and high actuation frequencies (i.e., those conditions that are
associated with large temperature changes).
Table 3.2 qPCR validation of Dusp1. Microarray and qPCR fold changes (relative to untreated
controls) of dual specificity phosphatase 1 (Dusp1) for cells treated by DMF manipulation at
different frequencies and electrode sizes or without manipulation at various temperatures (n=3).
Conditions with Dusp1 PCR fold-changes significantly different (p<0.05) than untreated controls
are shaded.
ConditionDusp1 Microarray
Fold Change
Dusp 1 PCR Fold
ChangePCR p-value
10 mm 18 kHz -7.5 -3.5 <0.01
5 mm 18 kHz - 1.4 >0.05
10 mm 10 kHz - -1.0 >0.05
5 mm 10 kHz - 1.4 >0.05
10 mm 1 kHz -1.1 -1.3 >0.05
5 mm 1 kHz - 1.0 >0.05
52°C control -6.9 -5.4 <0.025
47°C control -6.3 -2.3 <0.01
42°C control 1.0 1.0 >0.05
65
Sam H. Au Effects of Digital Microfluidic Actuation on Cell Fitness
In summary, the qPCR data agrees with the microarray results – the transcription levels of a
model stress gene for cells exposed to DMF operation at 400 Vpp for 15 min have strong
frequency dependence. This is consistent with previous observations of frequency-dependent cell
stress and death for cells manipulated on dielectrophoresis (DEP) systems138, 167, 168
. This is
notable and interesting, given that DEP systems, which typically include modest or no electrical
insulation and much higher (~MHz) frequencies, are quite different than the DMF system
described here. Furthermore, the qPCR results (supported by the cell-based stress sensor results)
indicate a strong dependence on electrode size – smaller DMF driving electrodes result in
negligible changes in DNA expression over a range of different operating conditions. This is
consistent with the hypothesis that the changes observed in cell fitness (both in transcription
profiles and in DNA integrity) are caused by droplet heating. Similar hypotheses have been
proposed for DEP-driven effects169
.
3.3.5 Droplet Heating
The results of the temperature measurement experiments are shown in Figure 3.5. In some cases,
droplet temperatures increased significantly– up to 25°C above ambient – during application of
DMF driving potentials. While such effects are purposefully generated in specialized DMF
devices modified to include resistive heaters (typically driven by DC potentials),170, 171
the data
shown in Figure 3.5 were generated from droplets positioned on standard devices (with no
heaters) driven by standard AC driving potentials. As far as we are aware, this phenomenon has
never before been reported.
66
Sam H. Au Effects of Digital Microfluidic Actuation on Cell Fitness
Figure 3.5 Droplet temperature in digital microfluidics. Images of 70 µL (top left), 18 µL (top
right) and 5 µL (bottom) droplets on digital microfluidic devices (A). Graphs of the temperatures
of droplets subjected to driving potentials on devices bearing 10.0 x 10.0 mm (red triangles), 5.0
x 5.0 mm (blue circles) and 2.5 x 2.5 mm (green squares) square electrodes at 400 Vpp at (B) 18
kHz, (C) 10 kHz and (D) 1 kHz frequencies. Error bars represent one standard deviation (n=3),
and curves were added to guide the eye.
B
C
D
A
55
50
45
40
35
30
25
Tem
pera
ture
(°C
)
8006004002000
Time (s)
18 kHz 10 mm electrode
18 khz 5 mm electrode
18 kHz 2.5 mm electrode
38
36
34
32
30
28
26
24
Tem
pera
ture
(°C
)
8006004002000
Time (s)
10 kHz 10 mm electrode
10 khz 5 mm electrode
10 kHz 2.5 mm electrode
27.5
27.0
26.5
26.0
25.5
25.0
24.5
24.0
Tem
pera
ture
(°C
)
8006004002000
Time (s)
1 kHz 10 mm electrode
1 khz 5 mm electrode
1 kHz 2.5 mm electrode
67
Sam H. Au Effects of Digital Microfluidic Actuation on Cell Fitness
The data in Figure 3.5 suggest several trends. First, the conditions tested that are closest to those
used regularly for digital microfluidics (i.e., 2.5 mm x 2.5 mm electrodes at 1 kHz or 10 kHz
frequencies) have near-negligible effects on droplet temperature, with average temperature
increases of 0°C and 2.3°C respectively. Second, elevated frequency or electrode size alone
results in minor effects – e.g., 2.5 mm electrode/18 kHz and 10 mm electrode/1 kHz result in
average temperature increases of 3.9°C and 2.8°C, respectively. Third, the large heating effects
(i.e., temperature increases greater than 5°C) were only observed for conditions with both
elevated electrode size and frequency.
It should be noted that the droplet temperatures in Figure 3.5 were measured in stationary
droplets (not moving droplets, as in actual DMF experiments). We expect that the temperatures
recorded in Figure 3.5 are an over-estimation of the temperatures of droplets in motion, as
mobile droplets such as those used for manipulating cells, were repeatedly moved to device
regions which have had time to cool, allowing heat to be dissipated more rapidly.
A potential mechanism for the temperature increases represented in Figure 3.5 is resistive (Joule)
heating, which is commonly observed in MEMS devices169
. As frequency increases, the
impedance of the digital microfluidic circuit decreases172
, resulting in an increase in current
which will increase the Joule heating. Another candidate is dielectric heating, which is caused by
the frictional loss of energy of rotating dipoles in the presence of an applied electric field173
. The
power generated by dielectric heating scales with the square of electric field strength and linearly
with applied frequency. Although DMF operation uses a much lower frequency range than those
which are typically used for dielectric heating (MHz-GHz), the physical scale of DMF devices
may render dielectric heating significant because large field strengths can be achieved over very
short distances. Both mechanisms of heating are consistent with the frequency-dependent
68
Sam H. Au Effects of Digital Microfluidic Actuation on Cell Fitness
increases in temperature shown in Figure 3.5, and in other experiments, much higher frequency
waveforms led to even greater temperature changes (data not shown). In on-going work, we are
evaluating these effects in more detail, but these observations are not the main focus of the work
presented here.
3.4 Conclusions
The goal of the work described in this chapter was to determine if the manipulation of
mammalian cells by digital microfluidics causes measurable genomic effects. In an attempt to
discover the boundaries of such (putative) effects, a range of different DMF operating conditions
were evaluated, including driving voltage and frequency, electrode size, and actuation time. The
results indicate that cells manipulated by DMF under such conditions exhibit a broad range of
responses, ranging from negligible to significant. In particular, for cells actuated on small
electrodes (5 mm x 5 mm or less), negligible detrimental effects were observed for all operating
conditions tested. In contrast, for cells actuated on large electrodes (10 x 10 mm), DNA damage
and protein expression changes in stress-related genes were observed for high-frequency (18
kHz) DMF operation, but not for low-frequency operation (10 or 1 kHz). The genomic
expression pattern for cells exposed to high-frequency operation/large electrodes showed
similarity to heat shocked cells, which was consistent with observations that DMF operation
under these conditions causes droplet heating. Future study is merited for alternate conditions
and different cell types, but the results described here suggest that DMF experiments involving
cells are best suited for operation at low frequencies on devices with small electrodes to avoid
excessive droplet heating, DNA damage and/or changes in gene expression. Also, the droplet
heating phenomenon may have potential uses which require rapid heating such as qPCR.
69
Sam H. Au Integrated Microorganism Culture and Analysis
Chapter 4
Integrated Microorganism Culture and Analysis
This chapter describes the development of a micro-scale bioreactor for automated culture and
density analysis of microorganisms. The microbioreactor is powered by digital microfluidics
(DMF) and because it is used with bacteria, algae and yeast, we call it the BAY microbioreactor.
Previous miniaturized bioreactors have relied on microchannels which often require valves,
mixers and complex optical systems. In contrast, the BAY microbioreactor is capable of
culturing microorganisms in distinct droplets on a format compatible with conventional bench-
top analyzers without the use of valves, mixers or pumps. Bacteria, algae and yeast were grown
for up to 5 days with automated semi-continuous mixing and temperature control. Cell densities
were determined by measuring absorbances through transparent regions of the devices, and
growth profiles were shown to be comparable to those generated in conventional, macro-scale
systems. Cell growth and density measurements were integrated in the microbioreactor with a
fluorescent viability assay and transformation of bacteria with a fluorescent reporter gene. These
results suggest that DMF may be a useful tool in automated culture and analysis of
microorganisms for a wide range of applications.
4.1 Introduction
Microorganisms such as bacteria, algae, and yeast are important for a wide range of applications.
For example, bacteria and yeast are used extensively for protein production174-176
and genomic
studies177, 178
, and algae is a potential source of biofuel production179, 180
. These types of cells are
cultured in specialized growth media, often accompanied by active mixing and temperature
control, and algae cultures have an added requirement of light as an energy source. A common
70
Sam H. Au Integrated Microorganism Culture and Analysis
method for monitoring growth profiles is to measure the absorbance of the culture at a specific
wavelength. As biomass accumulates, the absorbance increases in a manner that is predictable
and correlated with the density of cells in suspension.
In commercial applications, microorganisms are often grown in bioreactors with volumes up to
thousands of liters, but prior to large-scale culture, smaller systems (for example, microwell
plates bearing hundreds of microliters) are used to screen for optimum conditions for growth and
analyte production181
. There is great interest in developing miniaturized culture systems to
further reduce the costs of consumables, increase throughput and reduce manual labour
requirements. Most such efforts have relied on enclosed networks of microchannels; for
example, microfluidic devices have been developed to grow bacteria and yeast with integrated
sensors182, 183
and/or the ability to precisely control media delivery rates184, 185
. Moreover, there is
great potential for integrating channel-based microorganism culture systems with other
operations such as dielectrophoretic sorting186, 187
or even single-cell analysis188, 189
. However, a
disadvantage of microchannel-based culture systems is that they are not well-suited to
absorbance-based cell density measurements because of short path-lengths. Although devices
with integrated optics190-193
or systems that allow for direct counting of cells 194, 195
offer some
relief from these problems, fabrication of such devices is complicated and time-consuming, and
can lead to high fabrication costs. In addition, parallelization in microchannel-based systems is
challenging, especially for perfusion systems183-185
.
This chapter describes work developing a proof-of-principle microbioreactor relying on digital
microfluidics (DMF)37
. In DMF, fluid droplets are controlled in parallel on an open surface by
applying electrical potentials to an array of electrodes coated with a hydrophobic insulator (for a
comprehensive review of device geometries and fabrication techniques, see196
). DMF has
71
Sam H. Au Integrated Microorganism Culture and Analysis
become a popular tool for biochemical applications, including mammalian cell-based assays42, 79,
197, enzyme assays
100, 198, 199, immunoassays
85, 86, protein processing
88, 89, 91, 200-202, the polymerase
chain reaction203
, and clinical sample processing and analysis204
. However, at the time that this
work was completed, there was only one report of the use of DMF with microorganisms – Son
and Garrell demonstrated that droplets containing yeast could be moved on a DMF system134
.
The work in this chapter demonstrates that DMF is capable of automated growth and density
analysis of several different types of microorganisms. To validate the new technique, the growth
characteristics of bacteria, algae, and yeast were measured and compared to those of
microorganisms grown and analyzed using conventional macroscale techniques. Furthermore, a
viability assay and a genetic transformation were implemented on-chip to illustrate how the
platform can be integrated with down-stream analyses after up-stream culture and density
measurement.
4.2 Experimental
Unless specified otherwise, reagents were purchased from Sigma-Aldrich (Oakville, ON).
Escherichia coli DH5α were generously donated by Prof. Kevin Truong (Institute of
Biomaterials and Biomedical Engineering, University of Toronto). Saccharomyces cerevisiae
BY4741 (S288C Background) were generously donated by Prof. Igor Stagjlar (Department of
Medical Genetics and Microbiology, University of Toronto). Cyclotella cryptica (CCMP 332)
algae and associated culture reagents were purchased from the Center for Culture of Marine
Phytoplankton (Maine, NE).
4.2.1 Macroscale Cultures
Bacteria and yeast were grown in 3 mL aliquots of media (LB broth and YPD broth,
respectively) in vented tubes in a shaking incubator (37°C/225 rpm and 30°C/200 rpm/45°
72
Sam H. Au Integrated Microorganism Culture and Analysis
inclination, respectively). To generate growth curves, 0.3 mL aliquots of saturated culture (OD600
= 2.76 ± 0.02 for bacteria and OD600 = 6.95 ± 0.14 for yeast) were inoculated into 2.7 mL fresh
broth, and absorbance at 600 nm of diluted aliquots were measured periodically using a UV/Vis
spectrophotometer (Eppendorf, Westbury, NY). Algae was grown in 30-60 mL aliquots of f/2
medium (CCMP, Maine, NE) supplemented with biotin and cyanocobalamin (2 nM final
concentration ea., CCMP) in vented bottles at 14°C with agitation by magnetic stir bar (60 rpm),
with continuous illumination by a 60 W lamp positioned 20 cm from the culture. Algae were
maintained by weekly subculture at inoculation densities of ~9.0 x 105 cells/mL. To initiate
growth curves, exponentially proliferating algae were harvested by centrifugation (2000 g, 12
min) and inoculated in medium at a density of 7.0 x 104
cells/mL, and absorbance at 660 nm was
measured periodically using a UV/Vis spectrophotometer (Shimadzu, Burlington, ON). All
cultures were evaluated by microscopy (Leica DM2000, Leica Microsystems Canada, Richmond
Hill, ON) and were grown and evaluated in triplicate.
4.2.2 Device Fabrication
Devices were fabricated in the University of Toronto Emerging Communications Technology
Institute (ECTI) fabrication facility. Fabrication supplies included parylene-C dimer from
Specialty Coating Systems (Indianapolis, IN), Teflon-AF from DuPont (Wilmington, DE), and
A-174 silane from GE Silicones (Albany, NY). Silane solution comprised isopropanol, DI water,
and A-174 solution (50:50:1 v/v/v).
Glass substrates bearing patterned chromium electrodes (used as bottom plates of DMF devices)
were formed by photolithography and etching as described previously42
using photomasks
printed with 20,000 dpi resolution by Pacific Arts and Design (Toronto, ON). After patterning,
devices were primed for parylene coating by immersing them in silane solution for 15 min,
73
Sam H. Au Integrated Microorganism Culture and Analysis
allowing them to air-dry and then washing with isopropanol. After priming, devices were coated
with Parylene-C (6.9 µm) and Teflon-AF (235 nm). Parylene was applied by evaporating 15 g of
dimer in a vapor deposition instrument (Model PDS 2010 LABCOATER® 2, Specialty Coating
Systems, Indianapolis, IN), and Teflon-AF was spin-coated (1% w/w in Fluorinert FC-40, 2000
rpm, 60 s) and then post-baked on a hot-plate (160 °C, 10 min). To facilitate the application of
driving potentials , the polymer coatings were locally removed from the contact pads by gentle
scraping with a scalpel. Unpatterned top plates were formed by spin-coating indium tin oxide
(ITO) coated glass substrates (Delta Technologies, Stillwater, MN) with Teflon-AF (235 nm, as
above).
4.2.3 Device Operation
As depicted in Figure 4.1, the BAY microbioreactor comprises a reactor region (four 10.5 x 9.5
mm electrodes arranged in a 2 x 2 array) mated to a sample region (three rows of eleven 3 x 3
mm electrodes) and a reservoir region (three 6 x 6 mm electrodes, one for each sample row).
Each of the droplet actuation electrodes (shown) are connected by microfabricated wires to
contact pads (not shown for clarity) to facilitate application of driving potentials. Each of the
three rows includes an L-shaped electrode which defines a 1.5 x 1.5 mm transparent window for
absorbance measurements. Prior to an experiment, reagents were pipetted onto the appropriate
electrodes on a bottom plate, and then an unpatterned, transparent top ITO-coated plate was
positioned onto the device, sandwiching the droplets between the two plates. The spacing
between the two plates was defined by 350-µm thick spacers formed from five-high stacks of
double-sided tape. Driving potentials of 400-500 Vrms were generated by amplifying the output
of a function generator (Agilent Technologies, Santa Clara, CA) operating at 1-5 kHz, and
74
Sam H. Au Integrated Microorganism Culture and Analysis
droplets were actuated by applying driving potentials between the top electrode (ground) and
sequential electrodes on the bottom plate.
Figure 4.1 Schematic of BAY microbioreactor. A reactor region contains the mother drop, from
which daughter droplets are dispensed for analysis in the sample region or mixed with reagents
dispensed from the reservoir region. L-shaped electrodes in the sample region define 1.5 x 1.5
mm transparent windows which are used for absorbance measurements
Droplet motion was managed using an automated control system. Briefly, a computer running a
custom LabVIEW (National Instruments, Austin, TX) program interfaced to a DAQPAD 6507
(National Instruments, Austin, TX) controls the states of a network of high-voltage relays
(RT424012F, Tyco Electronics, Berwyn, PA). The inputs of the relays are connected to the
function generator/amplifier (see above), and the outputs of the relays mated to the contact pads
on the bottom plate of a device via a 40-pin connector. In practice, the user manually loads
75
Sam H. Au Integrated Microorganism Culture and Analysis
reagents into the microbioreactor and then inputs a series of desired droplet movement steps,
after which all droplet actuation is controlled automatically by the system.
4.2.4 Microscale Cultures
The media used for microscale culture were identical to those used for macroscale culture
(section 4.2.1), but supplemented with Pluronic F68 (bacteria and yeast 0.1 % w/v, algae 0.02%).
Pluronic additives reduce the adhesion of cells42, 197
and proteins53
to DMF device substrates and
have additives species with high hydrophilic-lipophilic balance (such as F68) have no
detrimental effects on cell vitality or proliferation as described in Chapter 2. Prior to use, devices
were sterilized by rinsing in 70% ethanol, and microorganisms were grown in ~70 µL aliquots
termed ―mother drops‖ in the reactor region. During culture, the devices were stored in a
humidified chamber (a sealed Petri dish saturated with water vapor), and the mother drops were
actuated in a circular pattern at programmed intervals. Temperatures were controlled by means
of a digital hot-plate (bacteria and yeast) or by storage in a chilled room (algae). Algae cultures
were positioned under a 60 W lamp (at a distance of 20 cm) for continuous illumination. The
parameters for each type of culture are listed in Table 4.1.
76
Sam H. Au Integrated Microorganism Culture and Analysis
Table 4.1 BAY microreactor parameters. Media, temperature, mixing frequency and absorbance
measurement frequecny used for microfluidic culture and analysis of bacteria, algae, and yeast
E. coli S. cerevisiae C. cryptica
Growth Media
LB broth YPD broth f/2
supplemented
with biotin and
cyanocobalamin
Temperature (°C)
37 30 14
Mixing Frequency
(min)
2.5
2.5 120
Absorbance
Measurement
Frequency (h)
1 2 24
To generate growth curves, microbial cultures were initialized by inoculating culture fluid into
fresh media, using identical procedures and densities to those used in the macroscale (see section
3.2.1). Mother drops containing bacteria, algae, or yeast were then grown with automated semi-
continuous mixing. For absorbance measurements, three ~7 µL daughter droplets were dispensed
from the mother drop onto the sample region and were driven to the L-shaped electrodes at
designated intervals (see Table 4.1). The microbioreactors were then positioned onto the tops of
transparent 96 well-plates and inserted into a PHERAstar microplate reader (BMG Labtech,
Durham, NC) for absorbance measurements at 600 nm for bacteria/yeast and 660 nm for algae.
The absorbances were collected using a well-scanning program, in which 8 separate
measurements were collected from pre-determined spots in a ~2.25 mm2 area. The absorbances
of the three daughter droplets were averaged together and were background-corrected by
subtracting the average absorbance (collected once at the beginning of each experiment)
measured from droplets containing only media on the same devices. After measuring the optical
77
Sam H. Au Integrated Microorganism Culture and Analysis
densities, the daughter droplets were translated by DMF actuation back to the reactor region
where they were re-combined with the mother drop for continued culture.
4.2.5 Growth Curve Generation
To generate growth curves for each microorganism, OD measurements from the well plate reader
and benchtop spectrometers were plotted in natural log scale. The data were then baseline
corrected (subtracting the lowest value) and re-scaled (dividing by the highest, corrected
macroscale value) to generate growth curves in the range of 0-1 for comparison between
macroscale and microscale profiles. For each data point, three replicate measurements were
obtained and the average and standard deviations were plotted as a function of time. Doubling
times were calculated as follows:
𝑇
Td is the doubling time [s]
K is the growth rate [s-1
]
Cn is the cell density (as determined by absorbance) at timepoint n [cell/L]
tn is the elapsed time at timepoint n [s]
Doubling time values were determined during early log phase growth and compared using a two-
tailed t-test.
=
𝑙𝑜𝑔 𝐶2
𝐶1
(𝑡2 − 𝑡1)
(4.1)
(4.2)
78
Sam H. Au Integrated Microorganism Culture and Analysis
4.2.6 Cell Death Assays
The viability of S. cerevisiae yeast grown in BAY microbioreactors was assayed using the
nucleic acid dye Ethidium homodimer-1 (EthD-1) (Invitrogen Molecular Probes, Eugene, OR).
Prior to operation, a 20 µL mixture of 2 µM EthD-1 in PBS supplemented with 0.05% F68 was
added to one reservoir and another mixture of 2 µM EthD-1 in PBS supplemented with 0.05%
F68 and 0.05% (v/v) Triton X-100 was added to another. Yeast were then inoculated as
described above and incubated at 30°C with automated mixing for 4 hours before the assay was
started. The assay was completed by dispensing daughter droplets of yeast from the mother drop
and merging each of them with droplets containing Triton X-100 and dye dispensed from the
reservoirs. The combined droplets were mixed in the sample region by actuation along the linear
path 10 times. The microbioreactor was incubated at 30°C for 1 hour and then visualized for
fluorescence over a square sample electrode.
4.2.7 Transformation
E. coli bacteria were transformed in a BAY microbioreactor with a pTriEx vector encoding
yellow fluorescent protein (YFP) and ampicillin resistance (generously donated by Kevin
Truong, University of Toronto). Prior to operation, a 20 µL mixture of 200 ng plasmid DNA and
0.20 M CaCl2 in LB broth without antibiotic supplemented with 0.05% F68 was added to a
reservoir. Bacteria (without ampicillin resistance) were then inoculated as described above and
incubated at 37°C with automated mixing for 1 hour before transformation. After confirming that
the cultures were at early log phase densities (as above), the microbioreactor was placed on ice
for 5 minutes, after which a daughter droplet was combined with an equal volume droplet
dispensed from the reservoir and mixed in the sample region by actuating the droplet in the
sample region approximately 10 times. The microbioreactor was chilled on ice for 1 hour,
79
Sam H. Au Integrated Microorganism Culture and Analysis
rapidly heated on a hot plate at 42°C for 50 seconds and then cooled on ice for an additional 1
minute. The microbioreactor was incubated at 37°C for 1 hour with automated mixing after
which the droplet containing transformed bacteria was spread on an LB agar plate containing
ampicillin at 100 mg/L and incubated overnight at 37°C to allow colony formation.
4.3 Results and Discussions
4.3.1 Microbioreactor Design
A wide range of applications for microorganisms, particularly those involving screening of
different conditions, would benefit from automated, micro-scale culture techniques. The work in
this chapter describes the development of an automated microbioreactor using digital
microfluidics that is capable of culture, analysis and transformation of microorganisms in
droplets. This device is called the ―BAY‖ microbioreactor, after the three microorganism species
used here: bacteria (E. coli), algae (C. cryptica) and yeast (S. cerevisiae). The primary function
of the BAY microbioreactor is long-term cell culture. As shown in Figure 4.1, the device was
designed such that culture takes place in a ~70 L aliquot of media called a ―mother drop.‖ In
conventional bioreactors, cultured cells are gently and continuously mixed to ensure uniform
distribution of dissolved gases, nutrients, and the cells themselves185, 186
. In the BAY
microbioreactor, this function was accomplished by manipulating the mother drop in a circular
path at regular intervals. As has been reported elsewhere205-207
, when droplets are actuated in
similar paths in array-based DMF systems, droplet contents are mixed at rates up to 10-50x faster
than by diffusion alone. As shown in Figure 4.2A, in the current work, each circular mix step
comprised a sequence of four movements between adjacent electrodes. While future designs may
be developed for more efficient mixing (using more complex DMF movement paths such as
figure-eights206
), in this work, a simple circular path was observed to be adequate.
80
Sam H. Au Integrated Microorganism Culture and Analysis
Figure 4.2 Operation of BAY microbioreactor. In (A), the mother drop was mixed by an
automated control mechanism. The drop was moved in a circular pattern on the four large
electrodes (which facilitates active mixing) at specific time intervals (i.e., every 2.5 minutes for
bacteria and yeast and every 2 hours for algae). In (B), daughter droplets were dispensed from
the mother drop to facilitate absorption measurements and were returned to the mother drop after
measurement
81
Sam H. Au Integrated Microorganism Culture and Analysis
To facilitate automated microorganism culture, the BAY microbioreactor was designed to be
compatible with absorbance measurements, which serve as an indicator of cell density. As shown
in Figure 4.2B, in each such measurement, three daughter droplets were dispensed successively
from the mother drop and were driven onto L-shaped electrodes for analysis using a well plate
reader. The frequency of these measurements (i.e., every 1 h for bacteria, 2 h for yeast and 24 h
for algae) was determined by the growth rates of the organism. After the measurements, the
daughter droplets were returned to mother drops to continue incubation (thus maintaining the
volume of the culture). The L-shaped electrodes were designed with transparent regions that are
one quarter (1/4) of the area of the square electrode. Other ratios of window to electrode area
(e.g. 1/2 and 1/8) were evaluated, but were found to be sub-optimal, as droplets either were not
reliably moved over the window (1/2) or the window was too small for reproducible
measurements in the plate reader (1/8). While the strategy of using an L-shaped electrode
worked well in the current design, future devices might be formed using transparent driving
electrodes for simultaneous incubation and analysis.
A significant advantage of the new technique is the simplicity of the analysis, especially when
evaluation of many different conditions is required. The L-shaped electrodes in BAY devices
were designed to be 9 mm from each other, matching the pitch of a 96-well plate, and
absorbances were measured by inserting devices into a multiwell plate reader. As described
elsewhere for other applications42, 89, 198
, we propose that compatibility with off-the-shelf optical
detectors is an attractive feature of DMF-based systems. The optical path-length for absorbance
measurements in the BAY devices is determined by the spacer thickness between top and bottom
plates in the device. A 350 µm spacer was used for the work reported here, but DMF is
compatible with inter-plate spacers of up to several millimeters (data not shown).
82
Sam H. Au Integrated Microorganism Culture and Analysis
Other advantages of this method are the ease with which active mixing can be incorporated and
the inherent batch mode of operation. For the former, many microscale reactors rely on diffusion
for mass transport, which is inefficient for cell-sized particles208
. Specially designed
micromechanical mixers can provide some relief for this limitation in microchannels183, 209
; in
contrast, simple repetitive manipulation of droplets in the DMF microbioreactor is sufficient for
active mixing without added complexity. Although the current generation of BAY devices was
designed for a single culture, the format, which matches the pitch and dimensions of multiwell
plates and detectors, is likely scalable in future generations for analysis of different culture and
analysis conditions in parallel. We propose that future generations of BAY systems may be
useful for growth and screening of many populations of organisms (e.g. S. cerevisiae gene
deletion libraries210
).
4.3.2 Microorganism Culture
To compare the growth rates of microorganisms grown in BAY microbioreactors to those
cultured by conventional means, bacteria, yeast and algae grown in both systems (micro- and
macro-scale) were interrogated with absorbance measurements over 8, 12 or 120 hours of
culture. As described in the section 4.2, the culture conditions in both systems (macro-scale or
BAY) were similar. Cell proliferation with minimal clumping was observed in both systems for
all three microorganisms over the course of the experiments (Figure 4.3). As shown in Figure 4.4
and Table 4.2, the growth rates of bacteria, algae, and yeast were similar for the macro- and
micro-scale; this is striking, given the significant differences between the systems (volumes,
electrostatic actuation, detectors, etc.).
83
Sam H. Au Integrated Microorganism Culture and Analysis
Figure 4.3 Microorganisms on device. Photomicrographs of daughter droplets positioned on L-
shaped electrodes containing (A) E. coli at 0 and 8 hours, (B) C. cryptica at 0 and 5 days, and (C)
S. cerevisiae at 0 and 12 hours in BAY microbioreactors. Scale bars are 500 µm
84
Sam H. Au Integrated Microorganism Culture and Analysis
Figure 4.4 Microorganism growth curves. Representative growth curves of (A) E. coli, (B) C.
cryptica, and (C) S. cerevisiae grown in macroscale (circles) or in BAY microbioreactors
(triangles). Absorbance measurements were taken at 600 nm for bacteria/yeast and 660 nm for
algae and were normalized to the highest value. Macroscale measurements were conducted using
benchtop UV/Vis spectrophotometers while microscale measurements were conducted on BAY
microbioreactors using a well-plate reader. Samples were evaluated in triplicate and error bars
represent one standard deviation.
A - Bacteria
B - Algae
C - Yeast
85
Sam H. Au Integrated Microorganism Culture and Analysis
Table 4.2 Microorganism doubling time comparison. Exponential doubling times of bacteria,
yeast and algae in macro and micro-scale formats. P-values compare the difference between
macro- and micro-scale.
E. coli S. cerevisiae C. cryptica
Macro-scale
Doubling Time (h)
0.79±0.06 1.80±0.24 37.0±1.2
Micro-scale
Doubling Time (h)
1.23 ±0.43 1.88±0.15 42.6±2.4
P-value 0.08 0.31 0.02
We speculate that the variations in the growth rates between the micro- and macro-scale systems
may be caused by a number of factors. The most likely is temperature differences since small
fluctuations in incubation temperature can result in vastly different growth rates in bacteria, algae
and yeast211-213
. This difference between the macro- and micro-scale systems is most relevant
during absorbance measurements – in the macro-scale system, small aliquots were collected
from the main culture, measured for density, and then disposed (while the main culture remained
at temperature). In the micro-scale system, the entire device (including the main culture)
remained at room temperature (in the plate reader) during density measurements. In the future, a
plate reader with temperature control might be used to correct for this. Other potential sources of
variation may include differences in mixing efficiency and imprecise temperature control on hot
plates in comparison to incubators. The greater variances in the optical density measurements in
the microscale are most likely a result of the combination of shorter path lengths in the
microscale (350 µm versus 1 cm) and differences in the analysis tools (well-plate reader versus
benchtop spectrophotometer).
86
Sam H. Au Integrated Microorganism Culture and Analysis
4.3.3 Downstream Processing and Analysis
A key advantage of microfluidic systems is the potential for integration of multiple processes
onto a single platform. To illustrate this point with the BAY microbioreactor, two different
downstream processes were integrated: a fluorescent viability assay of yeast and genetic
transformation of bacteria. In the former, the viability of S. cerevisiae grown in the BAY
microbioreactor was assayed on-chip with a fluorescent nucleic acid stain (Ethidium homodimer-
1). Dye with or without the surfactant, Triton X-100, was loaded into device reservoirs, the yeast
were grown for 4 hours and then daughter droplets dispensed and mixed with reagents dispensed
from reservoirs. Figures 4.5A-C are representative images collected in this assay, which
demonstrates the toxicity of Triton-X 100 at this concentration. Here, fluorescence was used a
read-out for cell death; in the future, we propose that many other probes or assays relying on
fluorescence or luminescence are likely compatible with the BAY microbioreactor.
87
Sam H. Au Integrated Microorganism Culture and Analysis
Figure 4.5 Microorganism viability and transformation. Representative images of yeast viability
assay and bacterial transformation. Yeast grown for 4 hours in BAY microbioreactor were
incubated with 2 µM Ethidium homodimer-1. (A) and (B) are brightfield and fluorescence
images of yeast not exposed to Triton X-100, and (C) is a fluorescence image of yeast exposed
to0.05% (v/v) Triton X-100. Brightfield images of yeast exposed to Triton X-100 were similar to
(A) and are not shown. Comparison of the images reveals that over 99% of the yeast were non-
viable after treatment with Triton X-100. (D) is a fluorescent image of an LB agar plate spread
with YFP-transformed bacteria. Scale bars represent 60 µm.
Yeast (-Triton) Fluorescence
Yeast (+Triton) Fluorescence
Yeast (-Triton) Bright Field
A
Transformed Bacteria
B
C D
88
Sam H. Au Integrated Microorganism Culture and Analysis
In the latter application, genetic manipulation was implemented on-chip; this is particularly
relevant for the BAY reactor, as this process is typically performed in early to mid-log phase
growth214
. To demonstrate this integrated process, E. coli were grown and their optical densities
were measured to confirm that they were in early to mid-log phase. The droplets containing the
bacteria were then transformed with a YFP reporter gene in a step-wise process involving several
thermal cycling steps and exposure to calcium chloride (which facilitates gene uptake). After
transformation, the bacteria were spread on an ampicillin agar plate overnight to confirm
successful transformation (Figure 4.5D).
The work in this chapter demonstrates a device architecture that can be used for microliter-scale
culture of bacteria, algae, and yeast, and integration with down-stream processing. But the
principle of using digital microfluidics for microorganism culture, analysis and manipulation
could extend to an even wider range of organisms and high-throughput technologies (e.g. two-
hybrid screens). Devices can be configured to accommodate droplets ranging from nanoliters to
milliliters215
, and electrodes can be arranged into virtually unlimited numbers of spatial
configurations. For example, future designs might be developed for self-contained culture of
limited supply or dangerous species, yielding information on gene expression, protein
interactions, and biological interactions within living cells, microarray or parallel-scale culture
(e.g., on 384-well or 1536-well formats)216-218
, or integration with microchannels219, 220
for other
types of analyses188, 189, 204
.
4.4 Conclusions
A platform for the integrated growth and cell density analysis of microorganisms in distinct
microdroplets has been developed using digital microfluidics (DMF). These microbioreactors
operate with long term automated semi-continuous DMF-driven mixing and are compatible with
89
Sam H. Au Integrated Microorganism Culture and Analysis
a diverse range of organisms and processes. The new technique may be beneficial for microbial
applications that require miniaturization or parallelization in highly customizable formats,
especially those which require complex, multi-step, multi-day processes.
90
Sam H. Au Microfluidic Liver Organoid Platform
Chapter 5
Microfluidic Liver Organoid Platform
5.1 Introduction
The liver is a vital organ responsible for metabolism, protein synthesis, bile production,
detoxification and drug clearance. Understanding, modeling and predicting the rate and scope of
liver processing of pharmaceutical candidates are critical components of drug discovery and
development. An ideal liver model for the pharmaceutical industry would be simple,
inexpensive, scalable and most importantly, strongly mimic liver function. Unfortunately, the
two-dimensional in vitro models that have been developed to study liver function and activity
often poorly mimic three-dimensional in vivo tissue. This is widely recognized as a bottleneck in
the pharmaceutical industry, one which has led to the requirement for exhaustive and expensive
multi-phase clinical testing of drug candidates221, 222
. We hypothesized that the unique attributes
of digital microfluidics for cell applications (described in section 1.3) might permit the
development of new systems capable of generating and evaluating liver models which are more
physiologically relevant than current best practices. Specifically, in contrast to other microfluidic
liver models223, 224
, we proposed that DMF would be useful to develop methods for the
formation, maintenance and analysis of cell-dense 3D constructs with precise control over
environmental parameters such as extracellular stiffness, cell density and biochemical stimuli of
individual tissue constructs. We further speculated that the new system would be useful for
systematically evaluating parameters such as construct size and matrix stiffness, to maximize
hepatic function.
In this chapter, I describe efforts to develop a digital microfluidic system capable of growing
three-dimensional liver organoid cultures, which were evaluated in terms of viability,
91
Sam H. Au Microfluidic Liver Organoid Platform
contractility, albumin production and enzymatic activity, towards the final goal of improving
upon current in vitro liver models. This work builds on the findings developed in the main text of
this thesis in the areas of anti-biofouling strategies (chapter 2), bias-free cell manipulation on
DMF (chapter 3) and long-term integrated analysis (chapter 4). While digital microfluidics has
been used previously to grow cells in three dimensions in a hydrogel matrices49, 225
, to our
knowledge, this represents both the first three-dimensional co-culture system and first ―organ on
a chip‖ work on digital microfluidics. Although work is on-going in some sections of this
project, preliminary results suggest that this method could become a useful tool for in vitro
screening for liver activity.
5.2 Experimental
Unless specified otherwise, reagents were purchased from Sigma-Aldrich. Parylene-C dimer was
obtained from Specialty Coating Systems. Teflon-AF was from DuPont, and A-174 silane was
from GE Silicones. All working solutions in sections 5.2.2-5.2.5 were supplemented with 0.06%
(wt/v) Pluronic F88 (BASF Corp.) to inhibit droplet fouling. All experiments were conducted
with three or more replicates.
5.2.1 Device and SU-8 Barrier Fabrication
Device top and bottom plates were fabricated at the University of Toronto Emerging
Communications Technology Institute in the same manner as the work described in sections
2.2.2, 3.2.1 and 4.2.2 except that additional steps were required to create SU-8 organoid retention
barriers. Briefly, before spin-coating Teflon® onto parylene-coated bottom plates, substrates
were pre-heated on a hot-plate at 95ºC for 5 minutes before spin coating ~5 mL SU-8 3035
(Microchem Corp.) for 10 s at 500 rpm followed immediately by a second 30 s spin at 1000 rpm.
SU-8 coated substrates were ramp heated (~3ºC/min) on hotplates from 65ºC to 95 ºC for 20 min
92
Sam H. Au Microfluidic Liver Organoid Platform
before ramp cooling (~3ºC/min) to 65ºC. Substrates were exposed through a photomask (formed
with ―negative‖ features) with 20,000 dpi resolution (Pacific Arts and Design) for 10 seconds
and then ramp heated on hotplates from 65ºC to 95 ºC for 5 min before ramp cooling to 65ºC.
Substrates were developed for 10 min in SU-8 developer, washed with isopropanol, dried with
nitrogen gas and baked at 170ºC for 10 min before Teflon® coating as described in section 2.2.2.
Top and bottom plates were separated by two pieces of double sided tape for a gap spacing of
140 µm.
Figure 5.1 depicts the device design and geometry. Briefly, the bottom-plate device design
comprises 65 electrodes, including a 2 x17 array of 2.2 x 2.2 mm electrodes, five ―large‖
reservoirs (10.0 x 6.5 mm) and four ―small‖ reservoirs (8.4 x 4.0 mm). Each large reservoir was
connected to the array by two 2.2 x 2.2 mm electrodes, while each small reservoir was connected
to the array by four 1.5 x 1.5 mm electrodes. The 1.5 x 1.5 mm electrodes served as ―organoid
culture regions‖, each with an SU-8 retention barrier. Each retention barrier featured fourteen
200 x 50 x 70 µm oval or rectangular pillars separated by 50 µm gaps.
Figure 5.1 Digital microfluidic organoid platform for construct creation, maintenance and
evaluation. (A) Photograph of device showing reservoirs and 4-plex design, (B)
Photomicrograph of SU-8 retention barrier and retained organoid, (C) Top- view (top) and side-
view (bottom) schematics of DMF device.
Large Reagent
Reservoirs
Organoid DMF Device Organoid Culture RegionSmall Reagent
Reservoir
SU-8 Retention
Barrier
Liver
OrganoidDigital Microfluidic
Electrodes
A
BC
Extraction Port
93
Sam H. Au Microfluidic Liver Organoid Platform
5.2.2 Device top and bottom plates were Cell Handling and Preparation
HepG2 cells and NIH-3t3 cells were maintained separately in feed media (50/50 DMEM/F12
with 8% fetal bovine serum (FBS), 2% calf serum (CS), 100 IU/mL penicillin and 100 µg/mL
streptomycin) by passaging every 3-4 days in T-25 flasks (Corning, Inc.). For use in forming
organoids, flasks containing the two cell types were trypsinized with 0.25% trypsin-EDTA for 5
minutes at 37ºC followed by resuspension in separate centrifuge tubes in feed media at 4.0 x 107
cell/mL concentrations. Collagen-cell suspensions were prepared on ice in 1.5 mL
microcentrifuge tubes by combining the solutions listed in Table 5.1 using 3D collagen cell
culture kits (Millipore, Inc.) where required.
94
Sam H. Au Microfluidic Liver Organoid Platform
Table 5.1 Collagen-cell suspension components. Volumes (µL) of components used to create
collagen-cell suspensions which gel to become organoids.
Component Co-culture
0.9 mg/mL
Collagen
Mono-culture
0.9 mg/mL
Collagen
Co-culture
1.5 mg/mL
Collagen
Mono-culture
1.5 mg/mL
Collagen
Collagen I 80 80 80 80
5X DMEM 20 20 20 20
Feed Media 166 181 61 70
10% (wt/v) F88 1.8 1.8 1.1 1.1
Neutralization buffer 2.5 2.5 2.5 2.5
4.0 x 107 HepG2/mL 15 15 9.1 9.1
4.0 x 107 NIH-3t3/mL 15 0 9.1 0
5.2.3 Device Operation Protocols
Droplets were manipulated by applying 220 Vpp, 5 kHz sinusoidal potentials to electrodes using
an automated high voltage switching system226
. To create liver organoids, device top and bottom
plates were washed with 70% ethanol and allowed to air dry in a laminar flow hood before
assembly with double sided tape. Then 6.0 µL aliquots of collagen-cell suspensions were
electrokinetically loaded onto 8.4 x 4.0 mm reservoirs and 315 nL droplets were dispensed onto
1.5 x 1.5 mm square electrodes adjacent to SU-8 retention barriers. The droplets were allowed to
gel (forming organoids) for 1 hour at 37ºC/5% CO2. The organoids were then ―fed‖ with feed
media using the general automated droplet exchange procedure, depicted in Figure 5.2. The
general automated droplet exchange procedure is a standard protocol designed to bring fresh
media or reagents to organoids and to remove spent media from devices for subsequent analysis.
Briefly, 12 µL aliquots of feed media (or other reagents, as described below) were loaded into
10.0 x 6.5 mm reservoirs and then 1.36 µL droplets were dispensed onto the 2.2 x 2.2 mm
electrode array. Up to four of these droplets were independently delivered to organoid-containing
95
Sam H. Au Microfluidic Liver Organoid Platform
droplets, and the merged contents were mixed by actuation across five linear electrodes in the
organoid culture region. Media in excess of 630 nL (equivalent to the volume associated with
two 1.5 x 1.5 mm electrodes) were then dispensed from merged droplet for extraction either to
waste or for subsequent analysis from the edge of the device using a blunt tip 24 gauge needle
connected to a 1 mL syringe.
Figure 5.2 General automated droplet exchange procedure for reagent/dye exchange and sample
extraction: (1) Feed droplets dispensed from large reservoir, (2) Feed drops aligned with
organoids droplets, (3) Feed and organoids droplets merged, (4) Merged droplets mixed, (5)
Excess volume split from merged droplets, (6) Excess volume/waste removed at removal port.
5.2.4 Mixing Analysis
An experiment using dyes was devised to characterize mixing in the organoid culture region
(Figure 5.3A). 630 nL PBS droplets containing blue food dye (representing media containing
organoids) were positioned adjacent to SU-8 barriers. 1.36 µL droplets of PBS were then
dispensed and merged with the dye-containing droplets following the general automated droplet
exchange procedure. Linear actuation of the merged droplet from the nearest 2.2 x 2.2 mm
electrodes to the small (8.4 x 4.0 mm) reservoirs and back again constituted one mix cycle. A
total of 6 mix cycles were conducted with photos taken immediately after mixing and at the end
96
Sam H. Au Microfluidic Liver Organoid Platform
of each mix cycle (as depicted in Figure 5.3.A) for analysis with ImageJ software. Four regions
of interest were defined encompassing the majority of each of four merged droplets at the end of
each mix cycle to maintain consistency. Then, images were split into red, green and blue
channels, and the histogram function was used within the regions of interest in each red channel
image to determine the standard deviation of the dye intensity as an estimate of unmixed
heterogeneity. The unbiased estimate of the standard deviation was used to estimate the standard
deviation of the sample deviation according to equation 5.1227
.
SD is the unbiased estimate of the standard deviation of the population standard deviation
s is the sample standard deviation
Γ(·) is the gamma function
n is the number of sample elements
5.2.5 Viability and Contractility Assays
Liver organoids were formed on device, incubated at 37ºC/5% CO2 and maintained by feeding
with feed media using the general automated droplet exchange procedure (as described above)
every day for four days. On the fourth day, PBS droplets containing 5.86 µM calcein AM and
11.72 µM ethidium homodimer-1 were merged with organoids droplets (merged droplet final
concentrations of 4 µM and 8 µM respectively). Merged droplets were mixed and excess media
split from organoid cultures using the general automated droplet exchange procedure (described
above). Organoids were incubated at room temperature for 30 minutes before washing with PBS
droplets using the general automated droplet exchange procedure and analysis using microscopy
(Leica DM2000, Leica Microsystems Canada).
𝑆𝐷 𝑠 = 𝑠𝛤
𝑛 − 12
𝛤 𝑛 2 𝑛 − 1
2−
𝛤 𝑛 2
𝛤 𝑛 − 1
2
2
(5.1)
97
Sam H. Au Microfluidic Liver Organoid Platform
5.2.6 Albumin Analysis
Liver organoids were formed and maintained as above (section 5.2.3) except that droplets
containing spent media were removed from devices during the general automated droplet
exchange procedure on days 1-4 and frozen in 0.6 mL microcentrifuge tubes at -80ºC until
analysis. Human albumin levels were quantified using a Human Albumin ELISA Kit (Abnova
Corporation, Taipei, Taiwan) following the manufacturer’s recommended guidelines. The
measured albumin levels were dilution-adjusted by multiplying the values by 3.16 (the ratio of
the merged droplet volume [1.99 µL] to that of the culture organoids volume [0.63 µL]) to obtain
the concentration of albumin in organoid culture droplets.
5.2.7 Cytochrome P450 3A4 Activity Assay
Liver organoids were formed in the same manner as above (section 5.2.3) to form HepG2 &
NIH-3t3 co-culture constructs. The general automated droplet exchange procedure was used to
introduce feed media containing reagents and remove an equal volume of excess liquid to
organoid cultures daily. Three populations of organoids were treated for three consecutive days:
control, induced, and induced-inhibited. Control organoids were fed on days one and two with
1.36 µL feed droplets containing 1.46% (v/v) ethanol (to a final concentration in the organoid
droplet of 1.0 %). Induced and induced-inhibited organoids were fed on day one with 1.36 µL
feed droplets containing 14.6 mM dexamethasone and 1.46% (v/v) ethanol (to concentrations in
the organoid droplet of 10.0 mM and 1.0 %, respectively) and on day two with feed droplets
containing 10.0 mM dexamethasone and 1.0% ethanol. Control and induced organoids were fed
on day three with 1.36 µL feed droplets containing 0.146% (v/v) ethanol (to 0.10% final
concentrations in the organoid droplets). Induced-inhibited organoids were fed on day three
with1.36 µL feed droplets containing 14.6 mM ketoconazole and 0.146% (v/v) ethanol (to
98
Sam H. Au Microfluidic Liver Organoid Platform
concentrations in the organoid droplet of 10.0 mM and 0.10%, respectively). For all conditions,
one hour after the feeding on day three, 1.36 µL droplets of feed media containing 14.6 mM
Vivid® BOMR dye (Life Technologies, Inc.) were added to each organoid culture to obtain a
final dye concentration of 10.0 mM.
For comparison to two-dimensional formats, on day zero, 50 µL of PBS containing 0.1 mg/mL
neutralized collagen I were dispensed into each well of tissue culture treated polystyrene flat-
bottom 96 well plates (Corning, Inc.), incubated at 37ºC/5% CO2, aspirated dry and allowed to
air dry for 30 minutes in a laminar biosafety cabinet. 1.0 x 105 HepG2 cells and 3.0 x 10
4 NIH-
3t3 cells were seeded into 100µL feed media per well and incubated at 37ºC/5% CO2. Analogous
three-day control, induced, and induced-inhibited conditions were defined and implemented as
for DMF (as above). In place of the general automated reagent exchange procedure, each feed
was implemented by aspirating out well contents and replacing them with 100 L aliquots of the
new contents (to the same final concentrations as described above).
For both microscale and macroscale cultures, BOMR dye was metabolized into a fluorescent
substrate and the intensity was determined immediately upon adding the dye and every 15
minutes afterwards for 1 hour, with incubation at 37ºC/5% CO2 between time-points, using a
Pherastar multiwell plate reader (BMG Labtech) at 530/620 nm wavelength excitation/emission.
Fluorescent intensity was normalized to the starting intensity for each organoid culture droplet or
culture well and enzymatic activity was estimated by the rate at which the fluorescent intensity
increased over time.
99
Sam H. Au Microfluidic Liver Organoid Platform
5.3 Preliminary Results and Discussion
5.3.1 Organoid Confinement, Feeding, and Mixing
As a first step towards generating a physiologically relevant liver model, we explored the use of
digital microfluidics in creating liver ―organoids‖ using hepatocytes in a hydrogel matrix. A new
digital microfluidic platform was developed to this end (Figure 5.1). Since organoids were
suspended rather than adherent to surfaces, a means to confine the three dimension constructs
was required. As shown (Figure 5.1B/C), arrays of SU-8 features formed on DMF bottom plates
served as retention barriers to spatially localize organoids while allowing for media exchange.
The only other use of SU-8 barriers in a DMF system that we are aware of was reported by
Mousa et al.97
, who used them for a different purpose (to aid in partitioning non-mixing solvents
for liquid-liquid extraction). We propose that this strategy is a useful new method for integrating
digital microfluidics with the culture of cells in 3D hydrogels.
As shown in Figure 5.2, a general automated droplet exchange procedure was developed for
organoids cultured adjacent to the retention barrier. As described in the experimental section, in
this procedure, droplets of feed media or other reagents were driven to the organoid culture
region and merged (and mixed) with the organoid-containing droplet. Excess media was then
driven away from the organoid, either to waste or for subsequent analysis. Reagent
concentrations in feed droplets were thus diluted to 68.5% of original concentrations once
merged with organoid-containing droplets. In typical experiments, organoids were maintained
for multiple days with a general automated droplet exchange of feed media every 24 hours.
A critical aspect of the general automated droplet exchange procedure is mixing. Nutrients must
be effectively delivered to organoids, waste products sufficiently removed and soluble analytes
such as albumin predictably diluted into extracted droplets. To examine this process, a dye
100
Sam H. Au Microfluidic Liver Organoid Platform
mixing experiment was devised. Figure 5.3A shows sample frames of the dye mixing experiment
and Figure 5.3B plots the dye distribution as quantified by the variance of dye intensity within
the merged droplets at the end of each mix cycle. The red channel was used to determine the
mixing efficiency since the blue dye absorbs most strongly in the red spectrum allowing for
strong contrast versus the bright device background. The merged droplets were well mixed
within only 2 cycles (4 total paths across 5 linear electrodes) since subsequent mix cycles failed
to reduce the variance in dye intensity. This suggests that the mixing procedure incorporated
sufficient convective/advective mixing. The presence of SU-8 features may have contributed to
the mixing efficiency by introducing hydrodynamic instabilities228
. Nonetheless, 5 cycles (10
paths) was chosen for all experiments described here to ensure adequate mixing.
101
Sam H. Au Microfluidic Liver Organoid Platform
Figure 5.3 Dye-mixing study to characterize the mixing efficiency of the automated droplet
exchange procedure. (A) Sample frames of the mixing experiment. 1. Feed droplets dispensed
and aligned with dye drops, 2. Feed and dye droplets merged, 3. First mix cycle begun by
actuating merged droplets towards small reservoirs, 4. First mix cycle ended by actuating merged
droplets onto 2.2 x 2.2 mm electrode, 5. Second mix cycle begun, 6. Second mix cycle ended.
(B) Standard deviation of red channel intensity within merged droplets (red channel) at the end
of each mix cycle. Error bars represent the unbiased estimate of the standard deviation of the
sample standard deviation. Conducted in quadruplicate.
A
B60
50
40
30
20
10
Sta
ndard
Devia
tion -
Red C
hannel
6543210
Mix Cycles
102
Sam H. Au Microfluidic Liver Organoid Platform
5.3.2 Organoid Contractility and Viability
Liver organoids were formed, cultured, and analyzed using the system described above. In initial
work, organoids were formed from HepG2 cells suspended in collagen, which may direct cells
into phenotypes that more closely resemble those found in vivo229
. But a simple suspension of
immortalized hepatocytes in collagen may lack the microenvironmental cues required for liver-
like phenotypic activity. To increase physiological relevance, organoids should (a) possess the
means to remodel the extracellular matrix (ECM) by generating additional matrix components
such as collagen, elastin, fibronectin and proteoglycans to supplement the (initially homogenous)
collagen matrix and (b) include stromal cells which may provide biochemical signals required
for hepatocyte activity224
. Here we chose to evaluate the use of NIH-3t3 fibroblasts as a
component of the DMF liver organoids, as they are known to secrete ECM proteins and to
actively remodel and contract hydrogels29
in three-dimensional cell culture.
Matrix remodeling and contraction are likely important components of organoids for a number of
reasons. First, even a modest hydrogel contraction can significantly increase cell densities. For
example, an isotropic contraction to half the original length scale results in an 8-fold reduction in
total volume, or an 8-fold increase in cell density before taking in account cell division or other
processes. This allows for the study of cell densities that are close to those of native tissue (~109
cell/cm3 in liver
230). Second, hydrogel contraction coupled with matrix protein remodeling can
bring cells into physical contact with each other, which is important because cell-cell contact of
hepatocytes is known to inhibit division related processes while increasing liver-specific
functions231
. Third, hydrogel contraction also increases matrix stiffness, which affects a wide
range of cellular processes including growth, morphology and migration232
.
103
Sam H. Au Microfluidic Liver Organoid Platform
The effects of two parameters during DMF organoid formation were evaluated: the presence of
NIH-3t3 fibroblasts and the concentration of collagen I in pre-gelled organoid suspensions.
Figure 5.4 depicts liver organoids seeded with or without 2 x 106 cell/mL NIH-3t3 and at low
(0.9 mg/mL) or high (1.5 mg/mL) collagen I concentrations on day zero and after four days in
culture. As expected, the presence of NIH-3t3 fibroblasts substantially increased the contraction
of organoids over 4 days relative to organoids without fibroblasts. Collagen density also played a
role in this process, with high collagen density inhibiting the magnitude of contraction. The
presence or absence of fibroblasts at the concentrations used in this study had a greater impact on
the contraction than did the change in collagen density. Importantly, the diameter of organoids,
even when seeded with NIH-3t3 cells in low density collagen, did not decrease to smaller than
the gaps in the retention barrier (~50 µm). In the near future, the degree of contraction for each
of the conditions will be quantified using photoanalysis software. Similar levels of contraction in
hydrogels seeded with 3t3 fibroblasts have been previously reported29
in pooled hydrogel
systems (i.e. many hydrogels in a chamber). However, as far as we are aware this is the first
report of work in which the contraction of individually addressable hydrogel constructs can be
monitored over time. DMF is uniquely suited for creating individually suspended hydrogels on
an open platform which are free to contract in three dimensions.
The viabilities of liver organoids over four days were also assayed to determine the effects of
culture conditions, including the degree of contraction, on cell health. As shown in Figure 5.5,
the vast majority of cells remained viable after a week as determined by calcein-AM staining
with very few dead cells (determined by ethidium homodimer-1staining) in any of the tested
conditions. This suggests that there is adequate diffusion of nutrients into and adequate diffusion
of waste products out of hydrogel organoids.
104
Sam H. Au Microfluidic Liver Organoid Platform
Figure 5.4 Organoid contractility. Photomicrographs of representative organoids cultured on
DMF platform on Day 0 (top) and Day 4 (bottom) after gel formation. Organoids were seeded
with HepG2 cells (2 x 106
cell/mL) with or without NIH-3T3 fibroblasts (2 x 106 cell/mL each)
in low (0.9 mg/mL) or high (1.5 mg/mL) density collagen. Scale bar represents 200 µm.
Figure 5.5 Organoid viability. Organoids cultured on a DMF platform 4 days after gel formation
in brightfield (top), stained for viability with calcein-AM (middle) or stained for cell death with
ethidium homodimer-1 (bottom). Organoids were seeded with HepG2 cells (2 x 106 cell/mL)
with or without NIH-3T3 fibroblasts (2 x 106 cell/mL each) in low (0.9 mg/mL) or high (1.5
mg/mL) density collagen. Scale bar represents 100 µm.
SU-8 Barriers
105
Sam H. Au Microfluidic Liver Organoid Platform
5.3.3 Albumin Activity
The production of albumin, a class of protein which constitutes approximately half of all proteins
in human plasma233
, is a primary function of hepatocytes and the secretion of albumin is
commonly used as a measure of liver-specific phenotypic activity. For example, HepG2 cells
cultured in three dimensional hydrogel matrices (similar to the liver organoids developed in this
work) are known to have increased levels of albumin secretion relative to two dimensional
formats223
. To assay the amount of albumin secreted by liver organoids formed and cultured on
DMF devices, media was collected during daily feeds and assayed for human albumin using an
ELISA kit (Figure 5.6). For the first 3 days, no significant difference in albumin levels was
observed between HepG2 organoids and HepG2/NIH-3t3 organoids, but by day 4, the co-
cultured organoids significantly outperformed mono-culture organoids. This is consistent with
previous work234-236
which suggests that co-cultured with fibroblasts improve the functional
activity of hepatocytes. Many of the factors discussed in section 5.3.1, including fibroblast-
induced hydrogel contraction, may have contributed to these results. Interestingly, there were no
statistical differences in albumin levels between organoids cultured in low or high collagen
densities. Because of the superior function in co-cultured organoids, Cytochrome P450 activity
studies (section 5.3.4) were conducted with co-cultured systems only.
106
Sam H. Au Microfluidic Liver Organoid Platform
Figure 5.6 Organoid albumin secretion assay. Concentration of secreted albumin in liver
organoid media collected during daily feeds determined using a human albumin ELISA kit.
Organoids were created with HepG2, with (closed symbols) or without (open symbols) NIH-3t3
fibroblasts and in either low (0.9 mg/mL) (purple/blue circles) or high (2.9 mg/mL) (red/green
squares) collagen. Error bars represent 1 standard deviation (n=3).
5.3.4 Cytochrome P450 Enzymatic Activity
Cytochrome P450 (CYP) is a superfamily of proteins found primarily in the liver which are
responsible for the catalysis of organic substances237
. These enzymes are of particular interest to
the pharmaceutical industry because they metabolize many drugs and antibiotics. In addition,
some small molecules are known to interfere with CYP enzymatic activity, delaying the
clearance of other drugs or toxins in vivo238
. In this work, the activity of human Cytochrome
P450 3A4 (CYP3A4), a CYP isoform, was evaluated after incubation of liver organoids with
compounds known to induce or inhibit CYP3A4 enzymatic activity. Dexamethasone, an anti-
inflammatory and immunosuppressant drug was used as a CYP3A4 inducer, and ketoconazole,
an anti-fungal drug, was used as a CYP3A4 inhibitor. CYP3A4 activity was monitored using a
12x103
10x103
8x103
6x103
4x103
2x103
Alb
um
in C
oncentr
ation (
ng/m
L)
4321
Day
HepG2 - Low Collagen
HepG2 - High Collagen
HepG2 + NIH3t3 - Low Collagen
HepG2 + NIH3t3 - High Collagen
107
Sam H. Au Microfluidic Liver Organoid Platform
fluorogenic substrate (BOMR) with specificity to that isoform239, 240
in control, induced, or
induced-inhibited HepG2/NIH-3t3 co-cultures grown either as 2D monolayers in well plates or
as 3D organoids in DMF devices. As shown in Figure 5.7A, the rates of substrate metabolism by
HepG2/NIH-3t3 co-cultured cells in well-plates were indistinguishable, regardless of the
condition. In contrast, the rates of substrate metabolism were clearly distinguishable in DMF-
cultured organoids (Figure 5.7B) with dexamethasone-treated organoids having higher activity
than control cells and dexamethasone and ketoconazole-treated organoids showing the lowest
level of activity. Hepatocyte CYP activity has been shown to be higher in three dimensional
systems than in traditional two dimensional formats241
, which may explain the higher HepG2
liver-specific function in the DMF platform relative to conventional 2D cultures. Another
attribute of the DMF system which may contribute to this difference may be that the detection
limits for fluorescent read-outs on DMF devices are often superior to those of comparable assays
implemented on macroscale well plates42
; a phenomenon that is likely a result of increased signal
from the reflective metal layer on devices.
The data in Figure 5.7 suggests that the new DMF/organoid system may represent an
inexpensive option for screening pharmaceutical candidates for CYP metabolism. Interestingly,
HepG2 cells are typically eschewed by the pharmaceutical industry (in favour of much more
expensive primary hepatocytes) for metabolism tests because of the inability (in conventional
formats) of HepG2 cells to model CYP activity21, 242
. For example, there are no significant
differences in transcriptional regulation for human CYP3A4 in HepG2 cells in response to doses
of Beta-naphtoflavone, Phenobarbitol or Rifampicin242
, which is consistent with the data in
Figure 5.7A. But as shown in Figure 5.7B, the CYP activity of DMF-cultured organoids can be
both induced and repressed by small molecules. If similar responses can be observed for other
108
Sam H. Au Microfluidic Liver Organoid Platform
compounds, this will be a particularly attractive feature of the DMF organoid model system,
without requiring the use of expensive primary cells.
Figure 5.7 Cytochrome P450 3A4 activity. Cells were untreated (blue circles), incubated with 10
mM dexamethasone for 48 hours prior to assay (red squares), or incubated with 10 mM
dexamethasone for 48 hours plus 10 mM ketoconazole for 1 hour (green diamonds) prior to
assay. Assays and cultures were conducted on HepG2/NIH-3t3 co-cultures in (A) two-
dimensional format in 96 well plates or (B) three-dimensional organoids on DMF device.
400
300
200
100
0
-100Norm
aliz
ed
Flu
ore
sce
nce
In
ten
sity (
Arb
itra
ry)
6050403020100
Time (hr)
Untreated Control
Dexamethasone
Dexamethasone & Ketoconazole
A
B400
300
200
100
0
Norm
aliz
ed
Flu
ore
sce
nt
Inte
nsity (
Arb
itra
ry)
6050403020100
Time (hr)
Untreated Control
Dexamethasone
Dexamethasone & Ketoconazole
109
Sam H. Au
5.4 Future Work
Liver organoids consisting of co-cultured HepG2 and NIH-3t3 cells in 3D collagen matrices
have been successfully created, maintained and assayed for albumin secretion and enzymatic
activity on DMF. In the immediate future, organoid contractility will be quantified. A second
potential future experiment is to probe the response of more physiologically representative cells
such as primary human hepatocytes or engineered cell lines242
grown in organoids. Liver
organoids created with more active cells more may further improve physiologically relevance
versus current best practices, and the relatively fewer numbers of cells used to form organoids
relative to well-plate cultures may make this a cost-effective option for screening. A third
potential direction of study would be to examine the cytotoxicity of small molecules to liver
organoids as a measure of hepatotoxicity. Acetominophen would be a prime candidate for these
studies since it has been suggested that acetaminophen overdose causes about 39% of acute liver
failure cases in the United States243
.
110
Sam H. Au Conclusions and Future Directions
Conclusions and Future Directions
Digital microfluidics is emerging as a useful tool for a wide range of applications including cell-
based studies. There are however a number of unresolved impediments to implementing
experiments with living cells on DMF. This thesis describes work addressing some of these
impediments including device failure caused by protein biofouling, potential effects of DMF
manipulation on cell fitness, and a lack of robust microfluidic platforms for long-term integrated
cell culture and analysis. This section outlines advances made in these key areas which should
benefit digital microfluidic researchers and the biomedical community as a whole. Future
directions for not only these areas but also for other challenges facing DMF for cell applications
are also discussed.
Pluronic Additives to Inhibit Device Failure (Chapter 2)
DMF device failure is a significant problem when using protein-rich solutions such as complete
cell media. This work advances the field in a number of directions. First, a panel of anti-fouling
additives was screened at different concentrations. Additives were found which prolonged device
lifetimes when manipulating serum-containing cell media by 2-3 times relative to previous best
practices. Second, the hydrophilic-lipophilic balance of the additive was found to be an
important factor affecting cell viability and growth rates. Third, a novel technique for rapidly
screening the biofouling rates of a number of candidate additives was developed such that the
work of researchers addressing this impediment in the future can be expedited. This work
enables researchers to reliably operate DMF devices for cell applications for longer periods
without undesired cell death. This is a key development since many potentially groundbreaking
111
Sam H. Au Conclusions and Future Directions
cell culture applications on DMF may require long-term manipulation of protein-solutions (e.g.
proteomics and metabolomics). Future work in this area should approach this problem from
multiple angles. For example, other researchers have examined the anti-fouling potential of
graphene oxides on DMF using the same screening methods that we outlined in chapter 2244
.
Other anti-fouling strategies such as alternate device surfaces comprised of anti-fouling block-
copolymers but which still exhibit a low interfacial interaction with aqueous solutions245
should
be investigated. Superhydrophobic surfaces, which often present significantly reduced contact
surface areas, may also significantly reduce biofouling rates246
. Another potential direction
would be thermally-switchable polymers which have been developed to release adsorbed
proteins from the surfaces of microfluidic devices247
.
Effects of Digital Microfluidics on Cell Fitness (Chapter 3)
The goal of the work described in chapter 3 was to identify DMF operating parameters which
may have detrimental effects on cell fitness such that these parameters can be avoided in the
future. A cell-based stress sensor sensitive to heat shock induction was employed to screen a
wide range of operating voltages and frequencies. None of the tested conditions, even at voltages
far greater than those typically used for droplet manipulation (625-650 Vpp), activated the sensor.
To search for other effects, genome-wide microarray studies and DNA integrity assays were
conducted for low and high frequency operation at high voltages. The majority of operating
conditions showed no or negligible effects on gene regulation or DNA fragmentation vs.
untreated controls. Only operation at high voltages (400 Vpp), high frequencies (18 kHz) and on
large electrodes (10 mm x 10 mm) resulted in substantial changes in gene expression and DNA
fragmentation (the presence of only two of these three parameters was insufficient to induce a
112
Sam H. Au Conclusions and Future Directions
response). It was determined that droplet heating (above 37ºC) due to DMF actuation was the
most likely cause of these effects. This work demonstrates that DMF operation at moderate
voltages, frequencies and electrode sizes and induced EMF result in no detrimental effects on
cells. However, at higher voltages, frequencies and electrode sizes, researchers should evaluate
the potential of droplet heating and cell effects. The validation that the vast majority of DMF
operating parameters are compatible with cells is an important finding since without this work,
DMF researchers performing cell applications would be unable to determine if experimental
results were a result of manipulated variables or an simply artifact of DMF manipulation. In the
future, researchers should examine the mechanism(s) behind droplet heating in DMF devices.
Adequate control over the heating mechanism(s) may be potentially useful for increasing
reaction kinetics, studying hyperthermia or cell incubation applications on DMF.
Integrated Microorganism Culture and Analysis (Chapter 4)
Many cell applications require the integration of well mixed cell cultures with long-term analysis
tools. The work described in chapter 4 represents the first digital microfluidic platform for the
multi-day culture and analysis of microorganisms including bacteria, algae and yeast. Semi-
continuous mixing was incorporated into the system with an automated control system. Cell
densities were determined using on-chip absorbance measurements and growth rates for all
organisms were found to be similar to those of counterparts cultured in the macro-scale. The
viability of yeast cells was analyzed and a proof-of-concept toxicity assay was conducted. This
work, the first demonstration of microorganism growth and analysis on DMF, paves the way for
more complex long-term cell applications on DMF systems, especially those which require
multi-day active mixing; examples include protein-protein interactions using yeast two-hybrid
113
Sam H. Au Conclusions and Future Directions
systems248
or multi-day metabolite-tracking experiments for metabolomics249
. In addition, the
system could be improved by miniaturizing absorbance or fluorescent detection tools such that
they can be incorporated into a complete micro total analysis system. The high intensity and low
size of light emitting diodes may be useful to this end250
.
Microfluidic Liver Organoid Platform (Chapter 5)
Building upon the work of chapters 2-4, a DMF platform was developed for creation of ―liver-
on-a-chip‖ organ models. In the first demonstration of 3D co-culture on DMF, hepatocyte and
fibroblast cells were encapsulated into neutralized collagen hydrogels, which were formed,
maintained and assayed on device as ―organoid‖ liver models. Extracellular albumin levels (a
measure of metabolic activity) increased daily for three days for both mono-culture (hepatocyte
only) and co-culture organoids. However, by day 4, mono-culture organoids had reduced
albumin secretion while co-culture organoids continued to increase albumin levels suggesting
that stromal cells improved the phenotypic activity of hepatocytes. For use as in physiologically-
relevant in vitro pharmacology studies, the activity of liver-specific enzyme Cytochrome P450
3a1 in DMF-cultured organoids was found to be induced and inhibited by tested small molecule
compounds. When similar macro-scale well-plate cultures were assayed, the induction and
inhibition of the same enzyme were undetectable. Thus, the DMF platform is uniquely
compatible with the formation and maintenance of physiologically relevant liver-specific tissue
constructs with relatively low-activity liver cell lines. In the future, I propose that similar systems
could be developed to accommodate constructs modeling other organ systems such as the
cardiovascular, neural and respiratory systems27
. I propose that the ease splitting, transporting
114
Sam H. Au Conclusions and Future Directions
and mixing discrete droplets on DMF makes it a promising tool for developing a ―human-on-
chip‖ platform for modeling systemic multi-organ interactions251
.
State of the Art and Future Directions
As a new technology, there are a great number of challenges facing DMF for cell applications.
Table C.1 summarizes the current state of DMF technology for cell applications and potential
directions for advancement.
115
Sam H. Au Conclusions and Future Directions
Table C.1 Current state of digital microfluidics for cell applications. Challenges, current
progress and future research directions for cell applications in digital microfluidics.
Challenge Description State of the Technology Potential Directions
2D Culture
Formats
Robust methods of
culturing suspension or
adherent cell types on
DMF platforms for routine
analysis must be
developed.
A wide range of cell types
cultured on native
hydrophobic surfaces 42-44,
80 and on hydrophilic
patches45-48
.
Co-culture of different
cell types in well-
defined geometries252
e.g. with
micropatterned
ECM253
.
3D Culture
Formats
Device features and
protocols must be
developed to generate
better DMF mimics of in
vivo microenvironments.
Cell encapsulated in
hydrogels49
and on-chip
evaluation of phenotypic
activity of organoids
(Chapter 5).
Hydrogel formation on
hydrophilic patches for
cell culture225
.
DMF-based models of
other organ systems
may be of interest to
biologists27
.
The incorporation of
systemic interactions
e.g. ―human-on-chip‖
DMF platform251
.
Biofouling
Device longevity is
severely inhibited by
biomolecule adsorption
onto DMF device surfaces.
Additives such as
Pluronics block co-
polymers53, 254
(Chapter 2)
and Graphene oxides244
.
Replaceable films254
and
filling devices with water-
immiscible oils100
.
Modifications to
device surfaces e.g.
fouling resistant block
co-polymer245
and
superhydrophobic
surfaces246
Thermally-switchable
polymers for release
of adsorbed
proteins247
Detrimental
Cell Effects
The electrokinetic
manipulation of droplets
may influence cell fitness
and phenotypic behaviour.
Only one study examining
the detrimental effects of
DMF actuation beyond
crude measures of growth
and viability38
(Chapter 3).
Investigate droplet
heating mechanisms
e.g. Joule172
or
dielectric173
heating.
Study genome-level
effects on other cell
types e.g. primary,
stem or neuronal.
116
Sam H. Au Conclusions and Future Directions
Challenge Description State of the Technology Potential Directions
Integrated
Analysis
The incorporation of more
analysis modalities enables
more complex cellular
analysis on DMF micro
total analysis systems.
Integrated absorbance80
(chapter 4) and
fluorescent42, 50
detection
systems for cell
applications.
Impedance-based systems
for on-chip cell
quantification255
.
Other analysis tools
developed for DMF
but as of yet not
applied to cell
applications e.g.
surface plasmon
resonance256
and
electrochemistry257
.
Consolidation of
current tools for more
complex studies of
cell states (e.g.
integrated
manipulation, lysis
and qPCR96
).
Humidity
Humid environments,
especially long durations
and at elevated
temperatures (e.g. inside
cell incubators) can be
catastrophic to DMF
devices. Water creeps
underneath the dielectric
layer (e.g. entering devices
at parylene-glass
boundaries) leading to
electrical shorts and
electrolysis when
operated.
Parylene C is a superior
conformal coating for
preventing water
penetration258
but adhesion
to glass can be poor.
A silane solution is used to
improve the bonding of
parylene C to the glass
substrates259
. A mitigation
strategy is the use water-
resistant tape on parylene-
glass boundaries.
Improve parylene
adhesion to glass e.g.
modifying
vaporization
temperature259
.
Switch to dielectrics
with stronger glass
adhesion e.g., spin on
glass260
.
Seal parylene-glass
boundaries e.g.
waterproof epoxies.
Temperature
Control
Temperature control on
DMF devices may be
beneficial for a number of
cell applications e.g.
incubation,
hyperthermia/hypothermia
studies, cell lysis,
cryopreservation,
thermotaxis.
Microfabricated resistive
heaters170, 171
.
Device operation on
simple heating elements
(Chapter 4).
Peltier systems similar
to those previously
integrated into other
microfluidic
formats261
.
Droplet heating by
DMF actuation
(Chapter 3).
117
Sam H. Au Conclusions and Future Directions
Challenge Description State of the Technology Potential Directions
Throughput
The inherent variability
and of cell systems and the
interest in screening
thousands of small
molecule targets in
addition to
genomic/proteomic studies
requires significant
parallelization .
Connecting (busing)
electrodes increases the
number of activatable
electrodes given voltage
switch limitations262
.
Cross-referencing (NxM
grid arrays)263, 264
e.g. a
15 x 15 array provides 225
effective grid
electrodes264
.
Thin film transistors can
be used to create large
arrays of individually
addressable electrodes
(e.g. 64 x 64 = 4096
electrodes)265
.
Complex multi-layer
PCB design permits
direct wiring to
potentially thousands
of individually
addressable
electrodes266-268
.
Incorporation of thin
film transistor
fabrication and
control systems for
routine DMF use.
Cost
If DMF devices are to be
adopted for routine use,
consumable costs must be
comparable to current low-
cost fabrication techniques
used for competitive
technologies e.g. injection
molding.
Disposable dielectric
coatings so that bottom
plates can be re-used54
.
Rapid copper prototyping
by photolithography and
laser printing269
.
PCB fabrication266-268
.
Paper substrates used
for other microfluidic
formats270
.
Increase demand to
capitalize on economy
of scale.
Automation
The adoption of DMF
technology by the
biomedical community
requires that end users
need only minimal training
to operate devices.
Automated control
systems have been
developed271
including an
open source feedback
system capable of
handling 320 simultaneous
outputs226
.
Automated droplet
generator uses feedback to
reduce droplet splitting
error272
.
Further reduce
required user input
e.g. implement code
to automatically route
droplets in the
minimum number of
steps262
such that
cross-contamination
can be prevented273
.
118
Sam H. Au Conclusions and Future Directions
Digital microfluidics has the potential to improve the speed, relevance and cost of biomedical
cell-based research. However, a number of major impediments exist for this new technological
application. This thesis describes work to advance DMF technology in a number of key areas
including device biofouling, potential cell effects of DMF manipulation and integration of long-
term cell culture and analysis. The discoveries described in this thesis permits biomedical
researchers to conduct complex, long-term, multi-step, on-chip cell-based analyses using digital
microfluidics without severe biofouling or undesired changes to genomic expression. However, a
significant amount of work still remains to be done not only in these areas but in others as well.
Only through the concerted efforts of many researchers can robust, routine cell application on
DMF platforms become a reality.
119
Sam H. Au References
References
1. T. M. Squires, S. R. Quake, Microfluidics: Fluid physics at the nanoliter scale. Rev. Mod.
Phys. 2005, 77. 977-1026.
2. D. Mark, S. Haeberle, G. Roth, F. von Stetten, R. Zengerle, Microfluidic lab-on-a-chip
platforms: requirements, characteristics and applications. Chem. Soc. Rev. 2010, 39. 1153-1182.
3. D. B. Weibel, G. M. Whitesides, Applications of microfluidics in chemical biology.
Current Opinion in Chemical Biology 2006, 10. 584-591.
4. G. M. Whitesides, The origins and the future of microfluidics. Nature 2006, 442. 368-
373.
5. J. J. Burbaum, Miniaturization technologies in HTS: how fast, how small, how soon?
Drug Discovery Today 1998, 3. 313-322.
6. J. W. Hong, S. R. Quake, Integrated nanoliter systems. Nature Biotechnology 2003, 21.
1179 - 1183.
7. A. Manz, Y. Miyahara, J. Miura, Y. Watanabe, H. Miyagi, K. Sato, Design of an Open-
tubular Column Liquid Chromatograph using Silicon Chip Technology. Sens. Actuator B-Chem.
1990, 1. 249-255.
8. D. J. Harrison, A. Manz, Z. H. Fan, H. Ludi, H. M. Widmer, Capillary Electrophoresis
and Sample Injection Systems Integrated on a Planar Glass Chip. Analytical Chemistry 1992, 64.
1926-1932.
9. G. H. W. Sanders, A. Manz, Chip-based microsystems for genomic and proteomic
analysis. Trac-Trends Anal. Chem. 2000, 19. 364-378.
10. C. S. Zhang, J. L. Xu, W. L. Ma, W. L. Zheng, PCR microfluidic devices for DNA
amplification. Biotechnol. Adv. 2006, 24. 243-284.
11. P. Neuzil, S. Giselbrecht, K. Lange, T. J. Huang, A. Manz, Revisiting lab-on-a-chip
technology for drug discovery. Nat. Rev. Drug Discov. 2012, 11. 620-632.
12. E. Delamarche, D. Juncker, H. Schmid, Microfluidics for processing surfaces and
miniaturizing biological assays. Adv. Mater. 2005, 17. 2911-2933.
13. J. Leng, J. B. Salmon, Microfluidic crystallization. Lab on a Chip 2009, 9. 24-34.
14. I. Meyvantsson, D. J. Beebe, in Annual Review of Analytical Chemistry. Annual Reviews:
Palo Alto, 2008, vol. 1, pp 423-449.
15. E. W. K. Young, D. J. Beebe, Fundamentals of microfluidic cell culture in controlled
microenvironments. Chem. Soc. Rev. 2010, 39. 1036-1048.
120
Sam H. Au References
16. C. Q. Yi, C. W. Li, S. L. Ji, M. S. Yang, Microfluidics technology for manipulation and
analysis of biological cells. Anal. Chim. Acta 2006, 560. 1-23.
17. P. J. Hung, P. J. Lee, P. Sabounchi, R. Lin, L. P. Lee, Continuous Perfusion Microfluidic
Cell Culture Array for High-Throughput Cell-Based Assays. Biotechnology and Bioengineering
2004, 89. 1-8.
18. R. Gómez-Sjöberg, A. A. Leyrat, D. M. Pirone, C. S. Chen, S. R. Quake, Versatile, Fully
Automated, Microfluidic Cell Culture System. Analytical Chemistry 2007, 79. 8557-8563.
19. C. H. J. Schmitz, A. C. Rowat, S. Ko¨ster, D. A. Weitz, Dropspots: a picoliter array in a
microfluidic device. Lab Chip 2008, 9. 44-49.
20. E. W. K. Young, C. A. Simmons, Macro- and microscale fluid flow systems for
endothelial cell biology. Lab on a Chip 2010, 10. 143-160.
21. P. M. van Midwoud, E. Verpoorte, G. M. M. Groothuis, Microfluidic devices for in vitro
studies on liver drug metabolism and toxicity. Integr. Biol. 2011, 3. 509-521.
22. G. M. Walker, J. Sai, A. Richmond, M. Stremler, C. Y. Chung, J. P. Wikswo, Effects of
flow and diffusion on chemotaxis studies in a microfabricated gradient generator. Lab Chip
2005, 5. 611-618.
23. N. Li, C. X. Luo, X. J. Zhu, Y. Chen, O. Y. Qi, L. P. Zhou, Microfluidic generation and
dynamically switching of oxygen gradients applied to the observation of cell aerotactic
behaviour. Microelectron. Eng. 2011, 88. 1698-1701.
24. M. A. Qasaimeh, S. G. Ricoult, D. Juncker, Microfluidic probes for use in life sciences
and medicine. Lab on a Chip 2013, 13. 40-50.
25. Q. C. Zhang, G. Lambert, D. Liao, H. Kim, K. Robin, C. K. Tung, N. Pourmand, R. H.
Austin, Acceleration of Emergence of Bacterial Antibiotic Resistance in Connected
Microenvironments. Science 2011, 333. 1764-1767.
26. J. V. Rocheleau, G. M. Walker, W. S. Head, O. P. McGuinness, D. W. Piston,
Microfluidic glucose stimulation reveals limited coordination of intracellular Ca2+ activity
oscillations in pancreatic islets. Proc. Natl. Acad. Sci. U. S. A. 2004, 101. 12899-12903.
27. H. Andersson, A. van den Berg, Microfabrication and microfluidics for tissue
engineering: state of the art and future opportunities. Lab on a Chip 2004, 4. 98-103.
28. D. Velasco, E. Tumarkin, E. Kumacheva, Microfluidic Encapsulation of Cells in Polymer
Microgels. Small 2012, 8. 1633-1642.
29. A. P. McGuigan, D. A. Bruzewicz, A. Glavan, M. Butte, G. M. Whitesides, Cell
Encapsulation in Sub-mm Sized Gel Modules Using Replica Molding. PLoS One 2008, 3. 11.
30. R. N. Zare, S. Kim, Microfluidic Platforms for Single-Cell Analysis. Annual Review of
Biomedical Engineering 2010, 12. 187-201.
121
Sam H. Au References
31. A. R. Wheeler, W. R. Throndset, R. J. Whelan, A. M. Leach, R. N. Zare, Y. H. Liao, K.
Farrell, I. D. Manger, A. Daridon, Microfluidic Device for Single-Cell Analysis. Analytical
Chemistry 2003, 75. 3581-3586.
32. M. Khine, A. Lau, C. Ionescu-Zanetti, J. Seo, L. P. Lee, A single cell electroporation chip.
Lab Chip 2004, 5. 38-43.
33. S. Koester, F. E. Angile, H. Duan, J. J. Agresti, A. Wintner, C. Schmitz, A. C. Rowat, C.
A. Merten, D. Pisignano, A. D. Griffiths, D. A. Weitz, Drop-based microfluidic devices for
encapsulation of single cells. Lab on a Chip 2008, 8. 1110-1115.
34. D. Stockholm, R. Benchaouir, J. Picot, P. Rameau, T. M. A. Neildez, G. Landini, C.
Laplace-Builhé, A. Paldi, The Origin of Phenotypic Heterogeneity in a Clonal Cell Population
In Vitro. PLoS One 2007, 2. e394.
35. L. Mazutis, J. Gilbert, W. L. Ung, D. A. Weitz, A. D. Griffiths, J. A. Heyman, Single-cell
analysis and sorting using droplet-based microfluidics. Nature protocols 2013, 8. 870-91.
36. F. Mugele, J. C. Baret, Electrowetting: From basics to applications. J. Phys.-Condes.
Matter 2005, 17. R705-R774.
37. A. R. Wheeler, Chemistry. Putting electrowetting to work. Science 2008, 322. 539-40.
38. S. H. Au, R. Fobel, S. P. Desai, J. Voldman, A. R. Wheeler, Cellular Bias on the
Microscale: Probing the Effects of Digital Microfluidic Actuation on Mammalian Cell Health,
Fitness and Phenotype. Integr. Biol. In Press 2013.
39. H. Becker, C. Gartner, Polymer microfabrication technologies for microfluidic systems.
Analytical and Bioanalytical Chemistry 2008, 390. 89-111.
40. K. Choi, A. H. C. Ng, R. Fobel, A. R. Wheeler, in Annual Review of Analytical
Chemistry, ed. R. G. Cooks, E. S. Yeung. Annual Reviews: Palo Alto, 2012, vol. 5, pp 413-440.
41. H. Song, D. L. Chen, R. F. Ismagilov, Reactions in droplets in microfluidic channels.
Angew Chem Int Ed Engl 2006, 45. 7336-56.
42. I. Barbulovic-Nad, H. Yang, P. S. Park, A. R. Wheeler, Digital microfluidics for cell-
based assays. Lab Chip 2008, 8. 519-26.
43. S. K. Fan, P. W. Huang, T. T. Wang, Y. H. Peng, Cross-scale electric manipulations of
cells and droplets by frequency-modulated dielectrophoresis and electrowetting. Lab on a Chip
2008, 8. 1325-1331.
44. G. J. Shah, A. T. Ohta, E. P. Y. Chiou, M. C. Wu, C. J. Kim, EWOD-driven droplet
microfluidic device integrated with optoelectronic tweezers as an automated platform for cellular
isolation and analysis. Lab on a Chip 2009, 9. 1732-1739.
45. I. Barbulovic-Nad, S. H. Au, A. R. Wheeler, A microfluidic platform for complete
mammalian cell culture. Lab on a Chip 2010, 10. -1536-1542.
122
Sam H. Au References
46. I. A. Eydelnant, U. Uddayasankar, B. B. Li, M. W. Liao, A. R. Wheeler, Virtual
microwells for digital microfluidic reagent dispensing and cell culture. Lab on a Chip 2012, 12.
750-757.
47. S. Srigunapalan, I. A. Eydelnant, C. A. Simmons, A. R. Wheeler, A digital microfluidic
platform for primary cell culture and analysis. Lab on a Chip 2012, 12. 369-375.
48. D. Witters, N. Vergauwe, S. Vermeir, F. Ceyssens, S. Liekens, R. Puers, J. Lammertyn,
Biofunctionalization of electrowetting-on-dielectric digital microfluidic chips for miniaturized
cell-based applications. Lab on a Chip 2011, 11. 2790-2794.
49. L. K. Fiddes, V. N. Luk, S. H. Au, A. H. C. Ng, V. Luk, E. Kumacheva, A. R. Wheeler,
Hydrogel discs for digital microfluidics. Biomicrofluidics 2012, 6.
50. D. Bogojevic, M. D. Chamberlain, I. Barbulovic-Nad, A. R. Wheeler, A digital
microfluidic method for multiplexed cell-based apoptosis assays. Lab on a Chip 2012, 12. 627-
634.
51. J. Andrade, V. Hlady, in Biopolymers/Non-Exclusion HPLC, ed. Springer. Heidelberg,
Germany 1986, pp 1-63.
52. M. Morra, On the molecular basis of fouling resistance. J. Biomater. Sci.-Polym. Ed.
2000, 11. 547-569.
53. V. N. Luk, G. Mo, A. R. Wheeler, Pluronic additives: a solution to sticky problems in
digital microfluidics. Langmuir 2008, 24. 6382-9.
54. H. Yang, V. N. Luk, M. Abelgawad, I. Barbulovic-Nad, A. R. Wheeler, A world-to-chip
interface for digital microfluidics. Anal Chem 2009, 81. 1061-7.
55. M. Emre, S. Cetiner, S. Zencir, I. Unlukurt, I. Kahraman, Z. Topcu, Oxidative Stress and
Apoptosis in Relation to Exposure to Magnetic Field. Cell Biochem. Biophys. 2011, 59. 71-77.
56. F. Marinelli, D. La Sala, G. Cicciotti, L. Cattini, C. Trimarchi, S. Putti, A. Zamparelli, L.
Giuliani, G. Tomassetti, C. Cinti, Exposure to 900 MHz electromagnetic field induces an
unbalance between pro-apoptotic and pro-survival signals in T-lymphoblastoid leukemia CCRF-
CEM cells. Journal of Cellular Physiology 2004, 198. 324-332.
57. I. Y. Belyaev, L. Hillert, M. Protopopova, C. Tamm, L. O. G. Malmgren, B. R. R.
Persson, G. Selivanova, M. Harms-Ringdahl, 915 MHz microwaves and 50 Hz magnetic field
affect chromatin conformation and 53BP1 foci in human lymphocytes from hypersensitive and
healthy persons. Bioelectromagnetics 2005, 26. 173-184.
58. H. Lin, M. Opler, M. Head, M. Blank, R. Goodman, Electromagnetic field exposure
induces rapid, transitory heat shock factor activation in human cells. Journal of Cellular
Biochemistry 1997, 66. 482-488.
59. C. A. Morehouse, R. D. Owen, Exposure to low-frequency electromagnetic fields does
not alter HSP70 expression or HSF-HSE binding in HL60 cells. Radiat. Res. 2000, 153. 658-662.
123
Sam H. Au References
60. R. Goodman, M. Blank, Insights into electromagnetic interaction mechanisms. Journal of
Cellular Physiology 2002, 192. 16-22.
61. I. A. Cotgreave, Biological stress responses to radio frequency electromagnetic radiation:
are mobile phones really so (heat) shocking? Archives of Biochemistry and Biophysics 2005,
435. 227-240.
62. M. J. Ruiz-Gomez, M. Martinez-Morillo, Electromagnetic Fields and the Induction of
DNA Strand Breaks. Electromagn. Biol. Med. 2009, 28. 201-214.
63. M. Blank, R. Goodman, Electromagnetic fields may act directly on DNA. Journal of
Cellular Biochemistry 1999, 75. 369-374.
64. V. G. Khurana, C. Teo, M. Kundi, L. Hardell, M. Carlberg, Cell phones and brain
tumors: a review including the long-term epidemiologic data. Surg. Neurol. 2009, 72. 205-214.
65. P. Frei, A. H. Poulsen, C. Johansen, J. H. Olsen, M. Steding-Jessen, J. Schuez, Use of
mobile phones and risk of brain tumours: update of Danish cohort study. British Medical Journal
2011, 343.
66. H. J. Lee, Y. B. Jin, J. S. Lee, J. I. Choi, J. W. Lee, S. H. Myung, Y. S. Lee, Combined
effects of 60 Hz electromagnetic field exposure with various stress factors on cellular
transformation in NIH3T3 cells. Bioelectromagnetics 2012, 33. 207-214.
67. L. Kheifets, D. Renew, G. Sias, J. Swanson, Extremely Low Frequency Electric Fields
and Cancer: Assessing the Evidence. Bioelectromagnetics 2010, 31. 89-101.
68. P. Kovacic, R. Somanathan, Electromagnetic fields: mechanism, cell signaling, other
bioprocesses, toxicity, radicals, antioxidants and beneficial effects. J. Recept. Signal Transduct.
2010, 30. 214-226.
69. P. Le-Clech, V. Chen, T. A. G. Fane, Fouling in membrane bioreactors used in
wastewater treatment. J. Membr. Sci. 2006, 284. 17-53.
70. M. Nystrom, L. Kaipia, S. Luque, Fouling and retention of nanofiltration membranes. J.
Membr. Sci. 1995, 98. 249-262.
71. R. Mukhopadhyay, When microfluidic devices go bad - How does fouling occur in
microfluidic devices, and what can be done about it? Analytical Chemistry 2005, 77. 429A-
432A.
72. M. Wahlgren, T. Arnebrant, Protein adsorption to solid-surfaces. Trends in
Biotechnology 1991, 9. 201-208.
73. A. Khademhosseini, K. Y. Suh, S. Jon, G. Eng, J. Yeh, G. J. Chen, R. Langer, A soft
lithographic approach to fabricate patterned microfluidic channels. Analytical Chemistry 2004,
76. 3675-3681.
124
Sam H. Au References
74. K. C. Popat, T. A. Desai, Poly(ethylene glycol) interfaces: an approach for enhanced
performance of microfluidic systems. Biosens. Bioelectron. 2004, 19. 1037-1044.
75. V. Linder, E. Verpoorte, W. Thormann, N. F. de Rooij, M. Sigrist, Surface biopassivation
of replicated poly(dimethylsiloxane) microfluidic channels and application to heterogeneous
immunoreaction with on-chip fluorescence detection. Analytical Chemistry 2001, 73. 4181-4189.
76. S. W. Hu, X. Q. Ren, M. Bachman, C. E. Sims, G. P. Li, N. Allbritton, Surface
modification of poly(dimethylsiloxane) microfluidic devices by ultraviolet polymer grafting.
Analytical Chemistry 2002, 74. 4117-4123.
77. M. E. Vlachopoulou, P. S. Petrou, S. E. Kakabakos, A. Tserepi, K. Beltsios, E.
Gogolides, Effect of surface nanostructuring of PDMS on wetting properties, hydrophobic
recovery and protein adsorption. Microelectron. Eng. 2009, 86. 1321-1324.
78. M. J. Jebrail, A. R. Wheeler, Let's get digital: digitizing chemical biology with
microfluidics. Current Opinion in Chemical Biology 2010, 14. 574-581.
79. G. J. Shah, A. T. Ohta, E. P. Chiou, M. C. Wu, C. J. Kim, EWOD-driven droplet
microfluidic device integrated with optoelectronic tweezers as an automated platform for cellular
isolation and analysis. Lab Chip 2009, 9. 1732-9.
80. S. H. S. Au, S.C.C.; Wheeler, A.R. , Integrated Microbioreactor for Culture and Analysis
of Bacteria, Algae and Yeast. Biomed. Microdevices 2011, 13. 41-50.
81. E. M. Miller, A. R. Wheeler, A digital microfluidic approach to homogeneous enzyme
assays. Anal. Chem. 2008, 80. 1614-1619.
82. V. Srinivasan, V. K. Pamula, R. B. Fair, Droplet-based microfluidic lab-on-a-chip for
glucose detection. Anal. Chim. Acta 2004, 507. 145-150.
83. V. Srinivasan, V. K. Pamula, R. B. Fair, An integrated digital microfluidic lab-on-a-chip
for clinical diagnostics on human physiological fluids. Lab Chip 2004, 4. 310-315.
84. J. G. Martin, M. Gupta, Y. M. Xu, S. Akella, J. Liu, J. S. Dordick, R. J. Linhardt, Toward
an Artificial Golgi: Redesigning the Biological Activities of Heparan Sulfate on a Digital
Microfluidic Chip. J. Am. Chem. Soc. 2009, 131. 11041-11048.
85. R. Sista, Z. Hua, P. Thwar, A. Sudarsan, V. Srinivasan, A. Eckhardt, M. Pollack, V.
Pamula, Development of a digital microfluidic platform for point of care testing. Lab Chip 2008,
8. 2091-104.
86. R. S. Sista, A. E. Eckhardt, V. Srinivasan, M. G. Pollack, S. Palanki, V. K. Pamula,
Heterogeneous immunoassays using magnetic beads on a digital microfluidic platform. Lab Chip
2008, 8. 2188-96.
87. E. M. N. Miller, A.H.C.; Uddayasankar, U.; Wheeler, A.R., A Digital Microfluidic
Approach to Heterogeneous Immunoassays. Anal Bioanal Chem 2010, 339. 337-345.
125
Sam H. Au References
88. M. J. Jebrail, A. R. Wheeler, Digital microfluidic method for protein extraction by
precipitation. Anal Chem 2009, 81. 330-5.
89. V. N. Luk, A. R. Wheeler, A Digital Microfluidic Approach to Proteomic Sample
Processing. Analytical Chemistry 2009, 81. 4524-4530.
90. H. Moon, A. R. Wheeler, R. L. Garrell, J. A. Loo, C. J. Kim, An integrated digital
microfluidic chip for multiplexed proteomic sample preparation and analysis by MALDI-MS.
Lab on a Chip 2006, 6. 1213-1219.
91. A. R. Wheeler, H. Moon, C. A. Bird, R. R. O. Loo, C. J. Kim, J. A. Loo, R. L. Garrell,
Digital microfluidics with in-line sample purification for proteomics analyses with MALDI-MS.
Analytical Chemistry 2005, 77. 534-540.
92. M. J. Jebrail, V. N. Luk, S. C. C. Shih, R. Fobel, A. Ng, H. Yang, S. L. S. Freire, A. R.
Wheeler, Digital Microfluidics for Automated Proteomic Processing. Journal of Visualized
Experiments 2009, 6. 1603.
93. D. Chatterjee, A. J. Ytterberg, S. U. Son, J. A. Loo, R. L. Garrell, Integration of Protein
Processing Steps on a Droplet Microfluidics Platform for MALDI-MS Analysis. Analytical
Chemistry 82. 2095-2101.
94. L. Malic, D. Brassard, T. Veres, M. Tabrizian, Integration and detection of biochemical
assays in digital microfluidic LOC devices. Lab on a Chip 2010, 10. 418-431.
95. Z. S. Hua, J. L. Rouse, A. E. Eckhardt, V. Srinivasan, V. K. Pamula, W. A. Schell, J. L.
Benton, T. G. Mitchell, M. G. Pollack, Multiplexed Real-Time Polymerase Chain Reaction on a
Digital Microfluidic Platform. Analytical Chemistry 82. 2310-2316.
96. Y. H. Chang, G. B. Lee, F. C. Huang, Y. Y. Chen, J. L. Lin, Integrated polymerase chain
reaction chips utilizing digital microfluidics. Biomed. Microdevices 2006, 8. 215-225.
97. N. A. Mousa, M. J. Jebrail, H. Yang, M. Abdegawad, P. Metalnikov, J. Chen, A. R.
Wheeler, R. F. Casper, Droplet-Scale Estrogen Assays in Breast Tissue, Blood, and Serum.
Science Translational Medicine 2009, 1. 1ra2.
98. S. O. Lumsdon, J. Green, B. Stieglitz, Adsorption of hydrophobin proteins at
hydrophobic and hydrophilic interfaces. Colloid Surf. B-Biointerfaces 2005, 44. 172-178.
99. A. W. P. Vermeer, C. E. Giacomelli, W. Norde, Adsorption of IgG onto hydrophobic
teflon. Differences between the F-ab and F-c domains. Biochim. Biophys. Acta-Gen. Subj. 2001,
1526. 61-69.
100. V. Srinivasan, V. K. Pamula, R. B. Fair, An integrated digital microfluidic lab-on-a-chip
for clinical diagnostics on human physiological fluids. Lab Chip 2004, 4. 310-5.
101. J. Y. Yoon, R. L. Garrell, Preventing biomolecular adsorption in electrowetting-based
biofluidic chips. Analytical Chemistry 2003, 75. 5097-5102.
126
Sam H. Au References
102. Y. Koc, A. J. de Mello, G. McHale, M. I. Newton, P. Roach, N. J. Shirtcliffe, Nano-scale
superhydrophobicity: suppression of protein adsorption and promotion of flow-induced
detachment. Lab on a Chip 2008, 8. 582-586.
103. M. Jonsson-Niedziolka, F. Lapierre, Y. Coffinier, S. J. Parry, F. Zoueshtiagh, T. Foat, V.
Thomy, R. Boukherroub, EWOD driven cleaning of bioparticles on hydrophobic and
superhydrophobic surfaces. Lab on a Chip 2011, 11. 490-496.
104. R. J. Green, M. C. Davies, C. J. Roberts, S. J. B. Tendler, A surface plasmon resonance
study of albumin adsorption to PEO-PPO-PEO triblock copolymers. J. Biomed. Mater. Res.
1998, 42. 165-171.
105. R. J. Green, S. Tasker, J. Davies, M. C. Davies, C. J. Roberts, S. J. B. Tendler,
Adsorption of PEO-PPO-PEO Triblock Copolymers at the Solid/Liquid Interface: A Surface
Plasmon Resonance Study. Langmuir 1997, 13. 6510-6515.
106. M. Amiji, K. Park, Prevention of protien adsorption and platelet-adhesion on surfaces by
PEO PPO PEO triblock copolymers. Biomaterials 1992, 13. 682-692.
107. M. A. McClain, C. T. Culbertson, S. C. Jacobson, N. L. Allbritton, C. E. Sims, J. M.
Ramsey, Microfluidic Devices for the High-Throughput Chemical Analysis of Cells. Analytical
Chemistry 2003, 75. 5646-5655.
108. H. M. Yang, 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
2011, 83. 3824-3830.
109. M. Wilhelm, C. L. Zhao, Y. C. Wang, R. L. Xu, M. A. Winnik, J. L. Mura, G. Riess, M.
D. Croucher, Polymer micelle formation .3. poly(styrene-ethylene oxide) block copolymer
micelle formation in water - a fluorescence probe study. Macromolecules 1991, 24. 1033-1040.
110. L. D. Kier, L. M. Wagner, T. V. Wilson, A. P. Li, R. D. Short, G. L. Kennedy,
Cytotoxicity of ethylene oxide/propylene oxide copolymers in cultured-mammalian-cells Drug
Chem. Toxicol. 1995, 18. 29-41.
111. A. V. Kabanov, E. V. Batrakova, V. Y. Alakhov, Pluronic (R) block copolymers as novel
polymer therapeutics for drug and gene delivery. J. Control. Release 2002, 82. 189-212.
112. J. H. Choi, J. Y. Jang, Y. K. Joung, M. H. Kwon, K. D. Park, Intracellular delivery and
anti-cancer effect of self-assembled heparin-Pluronic nanogels with RNase A. J. Control.
Release 2010, 147. 420-427.
113. C. Y. Chen, M. A. Even, Z. Chen, Detecting molecular-level chemical structure and
group orientation of amphiphilic PEO-PPO-PEO copolymers at solution/air and solid/solution
interfaces by SFG vibrational spectroscopy. Macromolecules 2003, 36. 4478-4484.
114. J. Lee, H. Moon, J. Fowler, T. Schoellhammer, C. J. Kim, Electrowetting and
electrowetting-on-dielectric for microscale liquid handling. Sens. Actuator A-Phys. 2002, 95.
259-268.
127
Sam H. Au References
115. S. K. Cho, H. J. Moon, C. J. Kim, Creating, transporting, cutting, and merging liquid
droplets by electrowetting-based actuation for digital microfluidic circuits. J. Microelectromech.
Syst. 2003, 12. 70-80.
116. H. Moon, S. K. Cho, R. L. Garrell, C. J. Kim, Low voltage electrowetting-on-dielectric.
Journal of Applied Physics 2002, 92. 4080-4087.
117. M. G. Pollack, R. B. Fair, A. D. Shenderov, Electrowetting-based actuation of liquid
droplets for microfluidic applications. Applied Physics Letters 2000, 77. 1725-1726.
118. M. G. Pollack, A. D. Shenderov, R. B. Fair, Electrowetting-based actuation of droplets
for integrated microfluidics. Lab on a Chip 2002, 2. 96-101.
119. D. Chatterjee, B. Hetayothin, A. R. Wheeler, D. J. King, R. L. Garrell, Droplet-based
microfluidics with nonaqueous solvents and solutions. Lab on a Chip 2006, 6. 199-206.
120. G. B. Sigal, M. Mrksich, G. M. Whitesides, Effect of surface wettability on the
adsorption of proteins and detergents. J. Am. Chem. Soc. 1998, 120. 3464-3473.
121. A. Sethuraman, M. Han, R. S. Kane, G. Belfort, Effect of surface wettability on the
adhesion of proteins. Langmuir 2004, 20. 7779-7788.
122. H. Elwing, S. Welin, A. Askendal, U. Nilsson, I. Lundstrom, A wettability gradient-
method for studies of macromolecular interactions at the liquid solid interface. J. Colloid
Interface Sci. 1987, 119. 203-210.
123. J. H. Lee, H. B. Lee, J. D. Andrade, Blood compatibility of polyethylene oxide surfaces.
Prog. Polym. Sci. 1995, 20. 1043-1079.
124. J. H. Sung, C. Kam, M. L. Shuler, A microfluidic device for a pharmacokinetic-
pharmacodynamic (PK-PD) model on a chip. Lab on a Chip 2010, 10. 446-455.
125. D. Huh, B. D. Matthews, A. Mammoto, M. Montoya-Zavala, H. Y. Hsin, D. E. Ingber,
Reconstituting Organ-Level Lung Functions on a Chip. Science 2010, 328. 1662-1668.
126. A. M. Ghaemmaghami, M. J. Hancock, H. Harrington, H. Kaji, A. Khademhosseini,
Biomimetic tissues on a chip for drug discovery. Drug Discovery Today 2012, 17. 173-181.
127. C. Moraes, G. Mehta, S. C. Lesher-Perez, S. Takayama, Organs-on-a-Chip: A Focus on
Compartmentalized Microdevices. Ann. Biomed. Eng. 2012, 40. 1211-1227.
128. A. L. Paguirigan, D. J. Beebe, From the cellular perspective: exploring differences in the
cellular baseline in macroscale and microfluidic cultures. Integr. Biol. 2009, 1. 182-195.
129. J. M. Lopacinska, J. Emneus, M. Dufva, Poly(Dimethylsiloxane) (PDMS) Affects Gene
Expression in PC12 Cells Differentiating into Neuronal-Like Cells. PLoS One 2013, 8. e53107.
130. P. W. Huang, T. T. Wang, S. W. Lin, Y. C. Chang, S. K. Fan, Ieee, Dielectrophoretic cell
concentrator on EWOD-Based chips. IEEE: New York, 2006; p 1418-1421.
128
Sam H. Au References
131. N. Vergauwe, D. Witters, F. Ceyssens, S. Vermeir, B. Verbruggen, R. Puers, J.
Lammertyn, A versatile electrowetting-based digital microfluidic platform for quantitative
homogeneous and heterogeneous bio-assays. J. Micromech. Microeng. 2011, 21. 11.
132. R. B. Fair, A. Khlystov, T. D. Tailor, V. Ivanov, R. D. Evans, V. Srinivasan, V. K.
Pamula, M. G. Pollack, P. B. Griffin, J. Zhou, Chemical and Biological Applications of Digital-
Microfluidic Devices. IEEE Des. Test 2007, 24. 10-24.
133. S. C. Shih, I. Barbulovic-Nad, X. Yang, R. Fobel, A. R. Wheeler, Digital microfluidics
with impedance sensing for integrated cell culture and analysis. Biosens Bioelectron 2012, 42C.
314-320.
134. S. Son, R. Garrell, Transport of live yeast and zebrafish embryo on a droplet (digital)
microfluidic platform. Lab Chip 2009, 9. 2398-2401.
135. C. Jolly, R. I. Morimoto, Role of the heat shock response and molecular chaperones in
oncogenesis and cell death. Journal of the National Cancer Institute 2000, 92. 1564-1572.
136. K. Richter, M. Haslbeck, J. Buchner, The Heat Shock Response: Life on the Verge of
Death. Mol. Cell 2010, 40. 253-266.
137. S. P. Desai, J. Voldman, in Twelfth International Conference on Miniaturized Systems for
Chemistry and Life Sciences. San Diego, California, USA, 2008, pp 1308-1310.
138. S. P. Desai, J. Voldman, Cell-based sensors for quantifying the physiological impact of
microsystems. Integr. Biol. 2011, 3. 48-56.
139. A. Schroeder, O. Mueller, S. Stocker, R. Salowsky, M. Leiber, M. Gassmann, S.
Lightfoot, W. Menzel, M. Granzow, T. Ragg, The RIN: an RNA integrity number for assigning
integrity values to RNA measurements. BMC Mol. Biol. 2006, 7. 14.
140. P. Du, W. A. Kibbe, S. M. Lin, lumi: a pipeline for processing Illumina microarray.
Bioinformatics 2008, 24. 1547-1548.
141. W. E. Johnson, C. Li, A. Rabinovic, Adjusting batch effects in microarray expression
data using empirical Bayes methods. Biostatistics 2007, 8. 118-127.
142. R. Hoffmann, A. Valencia, A gene network for navigating the literature. Nature Genetics
2004, 36. 664-664.
143. M. W. Pfaffl, A new mathematical model for relative quantification in real-time RT-PCR.
Nucleic Acids Res. 2001, 29. 6.
144. I. Rieu, S. J. Powers, Real-Time Quantitative RT-PCR: Design, Calculations, and
Statistics. Plant Cell 2009, 21. 1031-1033.
145. C. Ramakers, J. M. Ruijter, R. H. L. Deprez, A. F. M. Moorman, Assumption-free
analysis of quantitative real-time polymerase chain reaction (PCR) data. Neurosci. Lett. 2003,
339. 62-66.
129
Sam H. Au References
146. A. R. Collins, The comet assay for DNA damage and repair - Principles, applications,
and limitations. Molecular Biotechnology 2004, 26. 249-261.
147. A. Takahashi, T. Ohnishi, Does yH2AX foci formation depend on the presence of DNA
double strand breaks? Cancer Letters 2005, 229. 171-179.
148. M. Purschke, H.-J. Laubach, R. R. Anderson, D. Manstein, Thermal Injury Causes DNA
Damage and Lethality in Unheated Surrounding Cells: Active Thermal Bystander Effect. Journal
of Investigative Dermatology 2010, 130. 86-92.
149. G. Dimopoulos, G. K. Christophides, S. Meister, J. r. Schultz, K. P. White, C. Barillas-
Mury, F. C. Kafatos, Genome expression analysis of Anopheles gambiae: Responses to injury,
bacterial challenge, and malaria infection. Proceedings of the National Academy of Sciences
2002, 99. 8814-8819.
150. P. J. Maxwell, D. B. Longley, T. Latif, J. Boyer, W. Allen, M. Lynch, U. McDermott, D.
P. Harkin, C. J. Allegra, P. G. Johnston, Identification of 5-fluorouracil-inducible Target Genes
Using cDNA Microarray Profiling. Cancer Res. 2003, 63. 4602-4606.
151. T. Wada, J. M. Penninger, Mitogen-activated protein kinases in apoptosis regulation.
Oncogene 2004, 23. 2838-2849.
152. A. O'Donnell, Z. Odrowaz, A. D. Sharrocks, Immediate-early gene activation by the
MAPK pathways: what do and don't we know? Biochem. Soc. Trans. 2012, 40. 58-66.
153. H. Zhou, J. Gao, Z. Y. Lu, L. Lu, W. Dai, M. Xu, Role of c-Fos/JunD in protecting
stress-induced cell death. Cell Prolif. 2007, 40. 431-444.
154. X. R. Zeng, Y. B. Sun, L. Wenger, H. S. Cheung, Induction of early growth response
gene Egr2 by basic calcium phosphate crystals through a calcium-dependent protein kinase C-
independent p44/42 mitogen-activated protein kinase pathway. Cells Tissues Organs 2003, 174.
63-72.
155. R. Lang, M. Hammer, J. Mages, DUSP meet immunology: Dual specificity MAPK
phosphatases in control of the inflammatory response. Journal of Immunology 2006, 177. 7497-
7504.
156. E. D. Borkamo, O. Dahl, O. Bruland, O. Fluge, Global gene expression analyses reveal
changes in biological processes after hyperthermia in a rat glioma model. International Journal
of Hyperthermia 2008, 24. 425-441.
157. M. Hammer, J. Mages, H. Dietrich, A. Servatius, N. Howells, A. C. B. Cato, R. Lang,
Dual specificity phosphatase 1 (DUSP1) regulates a subset of LPS-induced genes and protects
mice from lethal endotoxin shock. J. Exp. Med. 2006, 203. 15-20.
158. T. Obata, G. E. Brown, M. B. Yaffe, MAP kinase pathways activated by stress: The p38
MAPK pathway. Critical Care Medicine 2000, 28. N67-N77.
130
Sam H. Au References
159. I. Boing, C. Stross, S. Radtke, B. E. Lippok, P. C. Heinrich, H. M. Hermanns, Oncostatin
M-induced activation of stress-activated MAP kinases depends on tyrosine 861 in the OSM
receptor and requires Jak1 but not Src kinases. Cell. Signal. 2006, 18. 50-61.
160. D. Bruemmer, F. Yin, J. Liu, T. Kiyono, E. Fleck, A. J. Van Herle, R. E. Law,
Expression of minichromosome maintenance proteins in vascular smooth muscle cells is
ERK/MAPK dependent. Exp. Cell Res. 2003, 290. 28-37.
161. S. M. Keyse, Dual-specificity MAP kinase phosphatases (MKPs) and cancer. Cancer
Metastasis Rev. 2008, 27. 253-261.
162. A. Arlt, H. Schafer, Role of the immediate early response 3 (IER3) gene in cellular stress
response, inflammation and tumorigenesis. Eur. J. Cell Biol. 2010, 90. 545-552.
163. Y. L. Chen, Y. L. Huang, N. Y. Lin, H. C. Chen, W. C. Chiu, C. J. Chang, Differential
regulation of ARE-mediated TNF alpha and IL-1 beta mRNA stability by lipopolysaccharide in
RAW264.7 cells. Biochem. Biophys. Res. Commun. 2006, 346. 160-168.
164. E. A. Emslie, T. A. Jones, D. Sheer, S. M. Keyse, The CL100 gene, which encodes a dual
specificity (Tyr/Thr) MAP kinase phosphatase, is highly conserved and maps to human
chromosome 5q34. Hum. Genet. 1994, 93. 513-516.
165. B. Li, P. Yi, B. Zhang, C. Xu, Q. Liu, Z. Pi, X. Xu, E. Chevet, J. Liu, Differences in
endoplasmic reticulum stress signalling kinetics determine cell survival outcome through
activation of MKP-1. Cell. Signal. 2010, 23. 35-45.
166. C. L. Andersen, J. L. Jensen, T. F. Ørntoft, Normalization of Real-Time Quantitative
Reverse Transcription-PCR Data: A Model-Based Variance Estimation Approach to Identify
Genes Suited for Normalization, Applied to Bladder and Colon Cancer Data Sets. Cancer Res.
2004, 64. 5245-5250.
167. Y. Huang, S. Joo, M. Duhon, M. Heller, B. Wallace, X. Xu, Dielectrophoretic cell
separation and gene expression profiling on microelectronic chip arrays. Analytical Chemistry
2002, 74. 3362-3371.
168. A. Menachery, R. Pethig, Controlling cell destruction using dielectrophoretic forces. IEE
Proc-Nanobiotechnol. 2005, 152. 145-149.
169. U. Seger, M. Panayiotou, S. Schnydrig, M. Jordan, P. Renaud, Temperature
measurements in microfluidic systems: Heat dissipation of negative dielectrophoresis barriers.
Electrophoresis 2005, 26. 2239-2246.
170. J. H. Wei, W. S. Hsu, S. K. Fan, Realizing temperature-controlled digital microfluidic
chips with versatile microelectrodes. IEEE: New York, 2007; p 741-744.
171. J. T. Cheng, C. L. Chen, Active thermal management of on-chip hot spots using EWOD-
driven droplet microfluidics. Exp. Fluids 2010, 49. 1349-1357.
131
Sam H. Au References
172. D. Chatterjee, H. Shepherd, R. L. Garrell, Electromechanical model for actuating liquids
in a two-plate droplet microfluidic device. Lab on a Chip 2009, 9. 1219-1229.
173. C. Gabriel, S. Gabriel, E. H. Grant, B. S. J. Halstead, D. Michael P. Mingos, Dielectric
parameters relevant to microwave dielectric heating. Chem. Soc. Rev. 1998, 27. 213-224.
174. G. P. L. Cereghino, J. M. Cregg, Applications of yeast in biotechnology: protein
production and genetic analysis. Current Opinion in Biotechnology 1999, 10. 422-427.
175. A. L. Demain, J. L. Adrio, Contributions of microorganisms to industrial biology. Mol
Biotechnol 2008, 38. 41-55.
176. J. R. Swartz, Advances in Escherichia coli production of therapeutic proteins. Current
Opinion in Biotechnology 2001, 12. 195-201.
177. E. V. Koonin, M. Y. Galperin, Prokaryotic genomes: the emerging paradigm of genome-
based microbiology. Current Opinion in Genetics & Development 1997, 7. 757-763.
178. J. Piškur, R. B. Langkjær, Yeast genome sequencing: the power of comparative
genomics. Molecular Microbiology 2004, 53. 381-389.
179. Y. Chisti, Biodiesel from microalgae beats bioethanol. Trends Biotechnol 2008, 26. 126-
31.
180. E. Yu, F. Zendejas, P. Lane, S. Gaucher, B. Simmons, T. Lane, in Jour App. Phycol.
2009.
181. A. Humphrey, Shake Flask to Fermentor: What Have We Learned? Biotechnology
Progress 2008, 14. 3-7.
182. I. Walther, B. v. d. Schootb, M. Boillatb, A. Cogoli, Performance of a miniaturized
bioreactor in space flight: microtechnology at the service of space biology. Enzyme and
Microbial Technology 2000, 27. 778-783.
183. Z. Zhang, P. Boccazzi, H. G. Choi, G. Perozziello, A. J. Sinskey, K. F. Jensen,
Microchemostat-microbial continuous culture in a polymer-based, instrumented microbioreactor.
Lab Chip 2006, 6. 906-13.
184. F. K. Balagadde, L. You, C. L. Hansen, F. H. Arnold, S. R. Quake, Long-term
monitoring of bacteria undergoing programmed population control in a microchemostat. Science
2005, 309. 137-40.
185. A. Groisman, C. Lobo, H. Cho, J. K. Campbell, Y. S. Dufour, A. M. Stevens, A.
Levchenko, A microfluidic chemostat for experiments with bacterial and yeast cells. Nat
Methods 2005, 2. 685-9.
186. B. H. Lapizco-Encinas, B. A. Simmons, E. B. Cummings, Y. Fintschenko,
Dielectrophoretic Concentration and Separation of Live and Dead Bacteria in an Array of
Insulators. Analytical Chemistry 2004, 76. 1571-1579.
132
Sam H. Au References
187. M. D. Vahey, J. Voldman, An Equilibrium Method for Continuous-Flow Cell Sorting
Using Dielectrophoresis. Analytical Chemistry 2008, 80. 3135-3143.
188. X. Y. Peng, P. C. H. Li, A Three-Dimensional Flow Control Concept for Single-Cell
Experiments on a Microchip. 2. Fluorescein Diacetate Metabolism and Calcium Mobilization in
a Single Yeast Cell As Stimulated by Glucose and pH Changes. Analytical Chemistry 2004, 76.
5282-5292.
189. J. Ryley, O. M. Pereira-Smith, Microfluidics device for single cell gene expression
analysis in Saccharomyces cerevisiae. Yeast 2006, 23. 1065-1073.
190. J. S. Kee, D. P. Poenar, P. Neuzil, L. Yobas, Monolithic integration of
poly(dimethylsiloxane) waveguides and microfluidics for on chip absorbance measurements.
Sens and Act. B 2008, 134. 532-538.
191. A. Llobera, S. Demming, R. Wilke, S. Buttgenbach, Multiple internal reflection
poly(dimethylsiloxane) systems for optical sensing. Lab Chip 2007, 7. 1560-6.
192. E. Verpoorte, A. Manz, H. Ludi, A. E. Bruno, F. Maystre, B. Krattiger, H. M. Widmer,
B. H. van der Schoot, N. F. de Roojj, A silicon flow cell for optical detection in miniaturized
total chemical analysis. Sens and Act B. 1992, 6. 66-70.
193. Z. Liang, N. Chiem, G. Ocvirk, T. Tang, K. Fluri, D. J. Harrison, Microfabrication of a
Planar Absorbance and Fluorescence Cell for Integrated Capillary Electrophoresis Devices. Anal
Chem 1996, 68. 1040-1046.
194. S. Faley, K. Seale, J. Hughey, D. K. Schaffer, S. VanCompernolle, B. McKinney, F.
Baudenbacher, D. Unutmaz, J. P. Wikswo, Microfluidic platform for real-time signaling analysis
of multiple single T cells in parallel. Lab Chip 2008, 8. 1700-12.
195. S. L. Faley, M. Copland, D. Wlodkowic, W. Kolch, K. T. Seale, J. P. Wikswo, J. M.
Cooper, Microfluidic single cell arrays to interrogate signalling dynamics of individual, patient-
derived hematopoietic stem cells. Lab Chip 2009, 9. 2659-64.
196. M. Abdelgawad, A. R. Wheeler, The Digital Revolution: A New Paradigm for
Microfluidics. Adv. Mat 2009, 21. 920-925.
197. I. Barbulovic-Nad, S. H. Au, A. R. Wheeler, A microfluidic platform for complete
mammalian cell culture. Lab Chip 2010, 10. 1536-42.
198. E. M. Miller, A. R. Wheeler, A digital microfluidic approach to homogeneous enzyme
assays. Anal Chem 2008, 80. 1614-9.
199. V. Srinivasan, V. K. Pamula, R. B. Fair, Droplet-based microfluidic lab-on-a-chip for
glucose detection. Anal. Chim. Acta 2004, 507. 145-150.
200. M. J. Jebrail, V. N. Luk, S. C. C. Shih, R. Fobel, A. H. Ng, H. Yang, S. L. Freire, A. R.
Wheeler, in J Vis Exp. 2009/11/10 edn., 2009.
133
Sam H. Au References
201. H. Moon, A. R. Wheeler, R. L. Garrell, J. A. Loo, C. J. Kim, An integrated digital
microfluidic chip for multiplexed proteomic sample preparation and analysis by MALDI-MS.
Lab Chip 2006, 6. 1213-9.
202. D. Chatterjee, A. J. Ytterberg, S. U. Son, J. A. Loo, R. L. Garrell, Integration of protein
processing steps on a droplet microfluidics platform for MALDI-MS analysis. Anal Chem 2010,
82. 2095-101.
203. Y. H. Chang, G. B. Lee, F. C. Huang, Y. Y. Chen, J. L. Lin, Integrated polymerase chain
reaction chips utilizing digital microfluidics. Biomed Microdevices 2006, 8. 215-25.
204. N. A. Mousa, M. J. Jebrail, H. Yang, M. Abdelgawad, P. Metalnikov, J. Chen, A. R.
Wheeler, R. F. Casper, Droplet-scale estrogen assays in breast tissue, blood, and serum. Sci
Transl Med 2009, 1. 1ra2.
205. H. W. Lu, F. Bottausci, J. D. Fowler, A. L. Bertozzi, C. Meinhart, C. J. Kim, A study of
EWOD-driven droplets by PIV investigation. Lab on a Chip 2008, 8. 456-461.
206. P. Paik, V. K. Pamula, R. B. Fair, Rapid droplet mixers for digital microfluidic systems.
Lab Chip 2003, 3. 253-259.
207. P. Paik, V. K. Pamula, M. G. Pollack, R. B. Fair, Electrowetting-based droplet mixers for
microfluidic systems. Lab on a Chip 2003, 3. 28-33.
208. J. Fowler, M. Hyejin, K. Chang-Jin, in Micro Electro Mechanical Systems, 2002. The
Fifteenth IEEE International Conference on. 2002, pp 97-100.
209. H. L. Lee, P. Boccazzi, R. J. Ram, A. J. Sinskey, Microbioreactor arrays with integrated
mixers and fluid injectors for high-throughput experimentation with pH and dissolved oxygen
control. Lab Chip 2006, 6. 1229-35.
210. G. Giaever, A. M. Chu, L. Ni, C. Connelly, L. Riles, S. Veronneau, S. Dow, A. Lucau-
Danila, K. Anderson, B. Andre, A. P. Arkin, A. Astromoff, M. El Bakkoury, R. Bangham, R.
Benito, S. Brachat, S. Campanaro, M. Curtiss, K. Davis, A. Deutschbauer, K.-D. Entian, P.
Flaherty, F. Foury, D. J. Garfinkel, M. Gerstein, D. Gotte, U. Guldener, J. H. Hegemann, S.
Hempel, Z. Herman, D. F. Jaramillo, D. E. Kelly, S. L. Kelly, P. Kotter, D. LaBonte, D. C.
Lamb, N. Lan, H. Liang, H. Liao, L. Liu, C. Luo, M. Lussier, R. Mao, P. Menard, S. L. Ooi, J. L.
Revuelta, C. J. Roberts, M. Rose, P. Ross-Macdonald, B. Scherens, G. Schimmack, B. Shafer, D.
D. Shoemaker, S. Sookhai-Mahadeo, R. K. Storms, J. N. Strathern, G. Valle, M. Voet, G.
Volckaert, C.-y. Wang, T. R. Ward, J. Wilhelmy, E. A. Winzeler, Y. Yang, G. Yen, E.
Youngman, K. Yu, H. Bussey, J. D. Boeke, M. Snyder, P. Philippsen, R. W. Davis, M. Johnston,
Functional profiling of the Saccharomyces cerevisiae genome. Nature 2002, 418. 387-391.
211. G. B. Fogel, C. F. Brunk, Temperature gradient chamber for relative growth rate analysis
of yeast. Anal Biochem 1998, 260. 80-4.
212. D. J. S. Montagnes, D. J. Franklin, Effect of temperature on diatom volume, growth rate,
and carbon and nitrogen content: Reconsidering some paradigms. Limnol. Oceangr. 2001, 46.
2008-2018.
134
Sam H. Au References
213. D. A. Ratkowsky, J. Olley, T. A. McMeekin, A. Ball, Relationship between temperature
and growth rate of bacterial cultures. J Bacteriol 1982, 149. 1-5.
214. M. Mandel, A. Higa, Calcium-dependent bacteriophage DNA infection. J Mol Biol 1970,
53. 159-62.
215. M. Abdelgawad, S. L. Freire, H. Yang, A. R. Wheeler, All-terrain droplet actuation. Lab
Chip 2008, 8. 672-7.
216. J. Ziauddin, D. M. Sabatini, Microarrays of cells expressing defined cDNAs. Nature
2001, 411. 107-10.
217. S. N. Bailey, D. M. Sabatini, B. R. Stockwell, Microarrays of small molecules embedded
in biodegradable polymers for use in mammalian cell-based screens. Proc Natl Acad Sci U S A
2004, 101. 16144-9.
218. S. N. Bailey, R. Z. Wu, D. M. Sabatini, Applications of transfected cell microarrays in
high-throughput drug discovery. Drug Discov Today 2002, 7. S113-8.
219. M. Abdelgawad, M. W. Watson, A. R. Wheeler, Hybrid microfluidics: a digital-to-
channel interface for in-line sample processing and chemical separations. Lab Chip 2009, 9.
1046-51.
220. M. W. Watson, M. J. Jebrail, A. R. Wheeler, Multilayer hybrid microfluidics: a digital-to-
channel interface for sample processing and separations. Anal Chem 2010, 82. 6680-6.
221. C. Guguen-Guillouzo, A. Guillouzo, in Hepatocytes: Methods and Protocols, ed. P.
Maurel. 2010, vol. 640, pp 1-40.
222. D. A. Groneberg, C. Grosse-Siestrup, A. Fischer, In vitro models to study hepatotoxicity.
Toxicologic Pathology 2002, 30. 394-399.
223. C. M. Hwang, S. Sant, M. Masaeli, N. N. Kachouie, B. Zamanian, S.-H. Lee, A.
Khademhosseini, Fabrication of three-dimensional porous cell-laden hydrogel for tissue
engineering. Biofabrication 2010, 2. Article No.: 035003.
224. V. L. Tsang, A. A. Chen, L. M. Cho, K. D. Jadin, R. L. Sah, S. DeLong, J. L. West, S. N.
Bhatia, Fabrication of 3D hepatic tissues by additive photopatterning of cellular hydrogels.
Faseb J. 2007, 21. 790-801.
225. I. A. Eydelnant, B. B. Li, A. R. Wheeler, Microgels On-Demand. Submitted 2013.
226. R. Fobel, C. Fobel, A. R. Wheeler, DropBot: An open-source digital microfluidic control
system with precise control of electrostatic driving force and instantaneous drop velocity
measurement. Applied Physics Letters 2013, 102. 193513-193513-5.
227. W. H. Holtzman, The Unbiased Estimate of the Population Variance and Standard
Deviation. Am. J. Psychol. 1950, 63. 615-617.
135
Sam H. Au References
228. J. Jung, C. J. Kuo, Y. Peles, M. Amitay, The flow field around a micropillar confined in a
microchannel. Int. J. Heat Fluid Flow 2012, 36. 118-132.
229. M. W. Tibbitt, K. S. Anseth, Hydrogels as Extracellular Matrix Mimics for 3D Cell
Culture. Biotechnology and Bioengineering 2009, 103. 655-663.
230. A. P. McGuigan, M. V. Sefton, Vascularized organoid engineered by modular assembly
enables blood perfusion. Proc. Natl. Acad. Sci. U. S. A. 2006, 103. 11461-11466.
231. T. Nakamura, K. Yoshimoto, Y. Nakayama, Y. Tomita, A. Ichihara, Reciprocal
modulation of growth and differentiated functions of mature rat hepatocytes in primary culture
by cell-cell contact and cell membranes. Proceedings of the National Academy of Sciences of the
United States of America-Biological Sciences 1983, 80. 7229-7233.
232. R. G. Wells, The role of matrix stiffness in regulating cell behavior. Hepatology 2008,
47. 1394-1400.
233. A. Farrugia, Albumin Usage in Clinical Medicine: Tradition or Therapeutic? Transfusion
Medicine Reviews 2010, 24. 53-63.
234. E. E. Hui, S. N. Bhatia, Micromechanical control of cell-cell interactions. Proc. Natl.
Acad. Sci. U. S. A. 2007, 104. 5722-5726.
235. A. Ito, H. Jitsunobu, Y. Kawabe, M. Karnihira, Construction of heterotypic cell sheets by
magnetic force-based 3-D coculture of HepG2 and NIH3T3 cells. J. Biosci. Bioeng. 2007, 104.
371-378.
236. S. J. Seo, I. Y. Kim, Y. J. Choi, T. Akaike, C. S. Cho, Enhanced liver functions of
hepatocytes cocultured with NIH 3T3 in the alginate/galactosylated chitosan scaffold.
Biomaterials 2006, 27. 1487-1495.
237. S. A. Wrighton, J. C. Stevens, The Human Hepatic Cytochromes P450 Involved in Drug
Metabolism. Crit. Rev. Toxicol. 1992, 22. 1-21.
238. J. H. Lin, A. Y. H. Lu, Inhibition and induction of cytochrome P450 and the clinical
implications. Clin. Pharmacokinet. 1998, 35. 361-390.
239. N. J. Liptrott, M. Penny, P. G. Bray, J. Sathish, S. H. Khoo, D. J. Back, A. Owen, The
impact of cytokines on the expression of drug transporters, cytochrome P450 enzymes and
chemokine receptors in human PBMC. Br. J. Pharmacol. 2009, 156. 497-508.
240. O. Trubetskoy, B. Marks, T. Zielinski, M. F. Yueh, J. Raucy, A simultaneous assessment
of CYP3A4 metabolism and induction in the DPX-2 cell line. Aaps Journal 2005, 7. E6-E13.
241. K. Nakamura, R. Mizutani, A. Sanbe, S. Enosawa, M. Kasahara, A. Nakagawa, Y. Ejiri,
N. Murayama, Y. Miyamoto, T. Torii, S. Kusakawa, J. Yamauchi, M. Fukuda, H. Yamazaki, A.
Tanoue, Evaluation of drug toxicity with hepatocytes cultured in a micro-space cell culture
system. J. Biosci. Bioeng. 2011, 111. 78-84.
136
Sam H. Au References
242. H. H. Gerets, K. Tilmant, B. Gerin, H. Chanteux, B. O. Depelchin, S. Dhalluin, F. A.
Atienzar, Characterization of primary human hepatocytes, HepG2 cells, and HepaRG cells at the
mRNA level and CYP activity in response to inducers and their predictivity for the detection of
human hepatotoxins. Cell Biol Toxicol 2012, 28. 69-87.
243. G. Ostapowicz, R. J. Fontana, F. V. Schiodt, A. Larson, T. J. Davern, S. H. B. Han, T. M.
McCashland, A. O. Shakil, J. E. Hay, L. Hynan, J. S. Crippin, A. T. Blei, G. Samuel, J. Reisch,
W. M. Lee, U. S. A. L. F. S. Grp, Results of a prospective study of acute liver failure at 17
tertiary care centers in the United States. Annals of Internal Medicine 2002, 137. 947-954.
244. G. Perry, V. Thomy, M. R. Das, Y. Coffinier, R. Boukherroub, Inhibiting protein
biofouling using graphene oxide in droplet-based microfluidic microsystems. Lab on a Chip
2012, 12. 1601-1604.
245. S. Krishnan, R. Ayothi, A. Hexemer, J. A. Finlay, K. E. Sohn, R. Perry, C. K. Ober, E. J.
Kramer, M. E. Callow, J. A. Callow, D. A. Fischer, Anti-biofouling properties of comblike block
copolymers with amphiphilic side chains. Langmuir 2006, 22. 5075-5086.
246. J. Genzer, K. Efimenko, Recent developments in superhydrophobic surfaces and their
relevance to marine fouling: a review. Biofouling 2006, 22. 339-360.
247. D. L. Huber, R. P. Manginell, M. A. Samara, B. I. Kim, B. C. Bunker, Programmed
adsorption and release of proteins in a microfluidic device. Science 2003, 301. 352-354.
248. T. Ito, K. Tashiro, S. Muta, R. Ozawa, T. Chiba, M. Nishizawa, K. Yamamoto, S.
Kuhara, Y. Sakaki, Toward a protein-protein interaction map of the budding yeast: A
comprehensive system to examine two-hybrid interactions in all possible combinations between
the yeast proteins. Proc. Natl. Acad. Sci. U. S. A. 2000, 97. 1143-1147.
249. R. Goodacre, S. Vaidyanathan, W. B. Dunn, G. G. Harrigan, D. B. Kell, Metabolomics
by numbers: acquiring and understanding global metabolite data. Trends in Biotechnology 2004,
22. 245-252.
250. M. O'Toole, D. Diamond, Absorbance based light emitting diode optical sensors and
sensing devices. Sensors 2008, 8. 2453-2479.
251. C. Zhang, Z. Zhao, N. A. A. Rahim, D. van Noort, H. Yu, Towards a human-on-chip:
Culturing multiple cell types on a chip with compartmentalized microenvironments. Lab on a
Chip 2009, 9. 3185-3192.
252. C. A. Goubko, X. D. Cao, Patterning multiple cell types in co-cultures: A review. Mater.
Sci. Eng. C-Mater. Biol. Appl. 2009, 29. 1855-1868.
253. J. Fukuda, A. Khademhosseini, J. Yeh, G. Eng, J. J. Cheng, O. C. Farokhzad, R. Langer,
Micropatterned cell co-cultures using layer-by-layer deposition of extracellular matrix
components. Biomaterials 2006, 27. 1479-1486.
254. S. H. Au, P. Kumar, A. R. Wheeler, A New Angle on Pluronic Additives: Advancing
Droplets and Understanding in Digital Microfluidics. Langmuir 2011, 27. 8586-8594.
137
Sam H. Au References
255. S. C. C. Shih, I. Barbulovic-Nad, X. Yang, R. Fobel, A. R. Wheeler, Digital microfluidics
with impedance sensing for integrated cell culture and analysis. Biosens. Bioelectron. 2013, 42.
314-20.
256. L. Malic, T. Veres, M. Tabrizian, Two-dimensional droplet-based surface plasmon
resonance imaging using electrowetting-on-dielectric microfluidics. Lab on a Chip 2009, 9. 473-
475.
257. C. Karuwan, K. Sukthang, A. Wisitsoraat, D. Phokharatkul, V. Patthanasettakul, W.
Wechsatol, A. Tuantranont, Electrochemical detection on electrowetting-on-dielectric digital
microfluidic chip. Talanta 2011, 84. 1384-1389.
258. S. Han, M. Osterman, S. Meschter, M. Pecht, Evaluation of Effectiveness of Conformal
Coatings as Tin Whisker Mitigation. Journal of Electronic Materials 2012, 41. 2508-2518.
259. J. M. Hsu, L. Rieth, S. Kammer, M. Orthner, F. Solzbacher, Effect of thermal and
deposition processes on surface morphology, crystallinity, and adhesion of Parylene-C. Sens.
Mater. 2008, 20. 87-102.
260. R. F. Cook, E. G. Liniger, Stress-corrosion cracking of low-dielectric-constant spin-on-
glass thin films. J. Electrochem. Soc. 1999, 146. 4439-4448.
261. G. Maltezos, M. Johnston, A. Scherer, Thermal management in microfluidics using
micro-Peltier junctions. Applied Physics Letters 2005, 87. 3.
262. M. Gupta, S. Akella, in 2007 Ieee/Rsj International Conference on Intelligent Robots and
Systems, Vols 1-9. 2007, pp 3150-3156.
263. S. K. Fan, C. Hashi, C. J. Kim, I. Ieee, in Mems-03: Ieee the Sixteenth Annual
International Conference on Micro Electro Mechanical Systems. 2003, pp 694-697.
264. X. Tao, K. Chakrabarty, in Design, Automation & Test in Europe Conference &
Exhibition, 2007. DATE '07. 2007, pp 1-6.
265. B. Hadwen, G. R. Broder, D. Morganti, A. Jacobs, C. Brown, J. R. Hector, Y. Kubota, H.
Morgan, Programmable large area digital microfluidic array with integrated droplet sensing for
bioassays. Lab on a Chip 2012, 12. 3305-3313.
266. J. Gong, C.-J. Kim, Direct-referencing two-dimensional-array digital microfluidics using
multilayer printed circuit board. J. Microelectromech. Syst. 2008, 17. 257-264.
267. J. Gong, C. J. Kim, Ieee, in MEMS 2005 Miami: Technical Digest. 2005, pp 726-729.
268. Z. Hua, J. L. Rouse, A. E. Eckhardt, V. Srinivasan, V. K. Pamula, W. A. Schell, J. L.
Benton, T. G. Mitchell, M. G. Pollack, Multiplexed Real-Time Polymerase Chain Reaction on a
Digital Microfluidic Platform. Analytical Chemistry 2010, 82. 2310-2316.
269. M. Abdelgawad, A. R. Wheeler, Rapid prototyping in copper substrates for digital
microfluidics. Adv. Mater. 2007, 19. 133-+.
138
Sam H. Au References
270. A. W. Martinez, S. T. Phillips, M. J. Butte, G. M. Whitesides, Patterned paper as a
platform for inexpensive, low-volume, portable bioassays. Angew. Chem.-Int. Edit. 2007, 46.
1318-1320.
271. S. C. C. Shih, R. Fobel, P. Kumar, A. R. Wheeler, A feedback control system for high-
fidelity digital microfluidics. Lab on a Chip 2011, 11. 535-540.
272. J. Gong, C.-J. Kim, All-electronic droplet generation on-chip with real-time feedback
control for EWOD digital microfluidics. Lab on a Chip 2008, 8. 898-906.
273. Y. Zhao, K. Chakrabarty, Cross-Contamination Avoidance for Droplet Routing in Digital
Microfluidic Biochips. IEEE Trans. Comput-Aided Des. Integr. Circuits Syst. 2012, 31. 817-830.