High-throughput droplet PCR
Amelia L. Markey a
, Stephan Mohr a
, Philip J.R. Day b,c,*
a
School of Chemical Engineering and Analytical Sciences, Manchester Interdisciplinary Biocentre, University of Manchester, UK
b
School of Translational Medicine, Manchester Interdisciplinary Biocentre, University of Manchester, UK
c
Centre for Integrated Genomic Research, Stopford Building, University of Manchester, UK
a r t i c l e i n f o
Article history:
Accepted 27 January 2010
Available online 2 February 2010
Keywords:
Droplets
PCR
Microfabrication
Quantification
High throughput
a b s t r a c t
The polymerase chain reaction has facilitated the ready analysis of nucleic acids. A next challenge
requires the development of means to unravel the complexity of heterogeneous tissues. This has presented
the task of producing massively parallelized quantitative nucleic acid data from the cellular constituents
of tissues. The production of aqueous droplets in a two phase flow is shown to be readily and
routinely facilitated by miniaturized fluidic devices. Droplets serve as ideal means to package a future
generation of PCR, offering an enhanced handling potential by virtue of reactant containment, to concurrently
eliminate both contamination and sample loss. This containment also enables the measurement of
nucleic acids from populations of cells, or molecules by means of high throughput, single cell analysis.
Details are provided for the production of a prototype micro-fluidic device which shows the production
and stable flow of droplets which we suggest will be suitable for droplet-based continuous flow microfluidic
PCR. Suggestions are also made as to the optimal fabrication techniques and the importance of
device calibration.
Ó 2010 Elsevier Inc. All rights reserved.
1. Introduction to the polymerase chain reaction
The polymerase chain reaction (PCR) is an enabling technology
which has allowed major advancements in biomedical research as
a tool for amplifying nucleic acid samples. However despite the
many advantages of conventional PCR there are limitations which
are inherent to both the PCR process itself and the requirement of
multiple sample preparation steps [1–5] This article seeks to
explore these caveats and may provide a novel solution by means
of high-throughput droplet PCR in integrated miniaturized microfluidic
devices [1,2,5]. Furthermore, the choices made in the design
and fabrication of this device are discussed and justified with
reference to some of the advantages and disadvantages of previously
published miniaturized PCR devices. Several vital calibration
experiments are also discussed which are needed in order to establish
an efficient, reliable and reproducible PCR which can be comparable
to that of conventional thermocyclers. Finally some of the
applications are discussed which reflect the rationale behind the
development of this technology.
1.1. Limitations of conventional PCR
The throughput of conventional PCR is limited by both spatial
constraints and a slow thermal ramp rate between temperatures
caused by a large thermal mass [5]. In addition, detection of amplified
species can be achieved at much lower reagent volumes than
are currently used in standard thermocycling machines [5,6]. However
despite these two limitations, the majority of problems are
caused by the steps of sample preparation that are necessary to
isolate purified nucleic acid templates from a population of cells
[4]. The separation of the various steps of sample preparation, from
cell lysis to nucleic acid extraction and purification, can lead to
both loss and contamination of the sample.
The current methods of sample preparation, and the following
qPCR analysis treat heterogeneous sample populations as though
they were homogeneous by applying the averaged result of the
population to represent each individual cell [4]. This resultant
averaged data, caused by bulk sample preparation and analysis
of a heterogeneous population of cells, is also commonly normalised
relative to a chosen reference gene when examining a quantitative
measure of gene expression. Appropriate choice of
reference genes aside [7], this result would carry more meaning
if expressed in absolute units of molecules per cell for each cell
in the sample. The generation of such an unit can only result from
the preparation and analysis of each cell on an individual basis.
However to make single cell analysis a feasible, and not laborious
1046-2023/$ - see front matter Ó 2010 Elsevier Inc. All rights reserved.
doi:10.1016/j.ymeth.2010.01.030
* Corresponding author. Address: School of Translational Medicine, Manchester
Interdisciplinary Biocentre, University of Manchester, UK. Fax: + 44 (0)161 275
1617.
E-mail address: philip.j.day@manchester.ac.uk (P.J.R. Day).
Methods 50 (2010) 277–281
Contents lists available at ScienceDirect
Methods
journal homepage: www.elsevier.com/locate/ymeth
task, a sensitive and high-throughput technique is needed where
each cell can be enclosed, within an essentially contained analysis
system, thus eliminating factors of sample loss and contamination
[5]. Furthermore, the comprehension of the original position of
specific cells within solid tissues would greatly assist in the
assessment of cell–cell communication and the special context
to cell activity.
1.2. Introduction to droplet-based micro-fluidics
1.2.1. History and rationale behind micro-fluidic miniaturization
Micro-fluidics and microfabrication have the potential to significantly
change the way modern chemistry and biology are
performed. There are several fundamental reasons supporting
the scaling-down of fluid handling. First, in a microchannel, the
reduction of sample size accelerates chemical reactions and heat
transfer due to shorter diffusion length and an increased surface
to volume ratio. Second, the smaller volume size of reactants offers
the opportunity to produce portable devices, particularly relevant
within the field of healthcare, environmental monitoring
and forensics, where point-of-care analysis and remote sampling,
respectively, would be particularly desirable. Other advantages of
smaller reactant volumes are a consequent reduction in the
amount of sample which has major clinical impact, and a prorata
reduction in reagent costs. Miniaturization also offers safety
benefits, including containment, where hazardous materials are
being used.
The fast growing number of applications in micro-fluidics are
driven by so-called ‘Lab-on a Chip’ (LOC) systems, a concept based
on the initial work of Andreas Manz in the early 1990s [8]. Lab-ona-chip
devices are a subset of microelectromechanical systems
(MEMS) devices and integrated processes are often described as
Micro total analysis systems (lTAS). The concept of lTAS eludes
to the fact that integration of laboratory steps, such as sample
preparation and analysis, on the same device can advance conventionally
isolated sample handling procedures towards a complete
laboratory analysis, with the aim of replacing existing laboratory
devices.
1.3. Droplet-based micro-fluidic miniaturization
A subsection of micro-fluidics is the rapidly emergent field of
droplet-based micro-fluidics, with droplets as discrete fluidic volumes
created by two immiscible phases. These micro-fluidic systems
are typically aqueous droplets held within a non-aqueous
carrier fluid such as oil or solvent, and normally are not required
to exchange material at their boundary. However exchanges can
occur which can be eliminated through the use of surfactants.
The use of surfactants have been shown to aid in droplet stability
by preventing the fusion of nearby droplets [9]. In contrast to traditional
micro-fluidics, droplets offer the possibility of serialisation
and isolation of individual samples, therefore offering potentially a
much higher scope of automation.
The produced droplets are normally in the nano to micrometre
range and can be produced at rates of tens of thousands per hour
[10]. Droplets in small channels also allow fluid flows with no dispersion,
which is a general problem with single-phase fluidics. In
addition, when such droplets are surrounded by an immiscible
fluid this can prevent contact between the surfaces of the channels
and the sample within the droplet, eliminating adverse effects due
to the large surface to volume ratios. Since identical droplets are
produced at a very high frequency within one experiment, parallel
processing is achievable to produce large data sets, offering a higher
degree of confidence following analysis. Droplets have also been
manipulated by electric fields to perform merging, sorting [11,12]
or splitting operations[13]. These advantages offer the potential of
higher throughput and the possibility to create a true ‘‘Lab on the
chip”.
Applications of droplet-based micro-fluidics range from studies
of enzyme kinetics [14–16], protein crystallization [17,18] and particle
polymerisation [19–21]. Droplets also offer the possibility to
encapsulate and cultivate biological cells in order to study specific
dynamics between different cells and their metabolites [22–24].
1.4. Why dropletize PCR?
PCR has a near omni-presence in the biomedical sciences. Its
applications are therefore unsurprisingly widespread which is
mainly attributable to the ease of execution and the exquisite sensitivity
offered by PCR which encompasses amplification from single
template molecule up to some nine orders of magnitude.
Recent demands of PCR have seen a move away from qualitative
assessment of template sequences to a requirement to perform
accurate, quantitative measurement of analyte.
PCR in droplets would offer substantial advantages compared to
other micro-fluidic-based PCR systems, such as well-based or single-phase
continuous flow devices [25], as they still suffer from
adsorption of reactants such as template, deoxyribonucleotide triphosphates
(dNTPs), and also DNA polymerase enzyme to the
channel walls. Droplet PCR would offer a further sample volume
reduction and shorter amplification times due to the reduction in
thermal mass. Single cell dropletized PCR has the potential to obviate
the problems discussed above relating to sample preparation
such as avoiding the ‘averaging effect’ seen in a heterogeneous
population by permitting the use of the more meaningful units of
molecules per cell. Provided care is taken to ensure that order of
cells entering the stream of droplets is maintained, the original
position of specific cells within solid tissues can be reconstructed
in virtual space which will assist in the assessment of space on cell
activity.
2. Description of methods
Over the last few years a trend is emerging using polymers
for micro-fluidic devices rather than the more traditional materials
such as silicon, glass, ceramics and metal. There are a number
of advantages, where primarily large numbers of polymer
devices can be made quickly and at low costs using mass
fabrication techniques. The most common fabrication methods
applied for the manufacture of polymer microchips are: soft
lithography [26], hot embossing [27], laser ablation [28], injection
moulding [29], and direct Computer Numerical Control
(CNC) machining [19].
2.1. Fabrication and machining
The prototype micro-fluidic PCR device presented here was fabricated
using CNC precision machining and will be briefly
described. The poly(methyl methacrylate) (PMMA) planar chip
shown in Fig. 1 was made from a sheet of 4 mm thickness (RS, Corby,
UK) and cut to 75 Â 75 mm squares. The 3D flow pattern was
designed using AUTODESK INVENTOR (Autodesk Inc., San Rafael,
California, USA) and translated into CNC machine code by EDGECAM
software (EdgeCam, Reading, UK). The micromachining was
carried out using a CAT3D M6 CNC milling machine (Datron Technologies
Ltd., Milton Keynes, UK) with 0.1–0.4 mm diameter tungsten-carbide
milling tools (Toolex Ltd., Trowbridge, UK). The
channels were sealed with a 100 lm thick acetate foil, which
was attached to the PMMA chip using a thin film of Ultraviolet-curable
epoxy (Norland 68, Norland Products Inc., New Brunswick, NJ,
USA). To avoid blocking the channels, a hot roll laminator was used
278 A.L. Markey et al. / Methods 50 (2010) 277–281
to spread the epoxy to a thickness of several microns between two
sheets of acetate. The sheets were then placed on dry ice and
cooled for 5 min before being separated. The solidified epoxy stays
attached to one of the sheets allowing a seal with the chip. The chip
and seal were then brought into contact with the use of the laminator
and finally the epoxy was cured using an Ultraviolet (UV)
light source (60–80 mW/cm2
) for 2 min at a peak wavelength of
365 nm as shown in Fig. 1.
The cylinder device uses narrow bore polytetrafluoroethylene
(PTFE) tubing (Microbore PTFE tubing, Cole-Parmer Instrument
Co. Ltd., UK) coiled around an aluminium cylinder which is divided
into two temperature zones (Fig. 1). This device was designed
using AUTODESK INVENTOR (Autodesk Inc.) and machined
in-house. The cylinder was cut in half to enable these two temperature
zones to be separated. The cylinder was supported by brackets
separating each half by approximately 2 mm air gap, to prevent
the transfer of heat from one half to the other. A thread
(pitch = 1.5 mm, depth = 1.05/2 mm) was cut into the aluminium
cylinder (diameter = 40 mm, length = 65 mm) to accommodate 40
winds of the PTFE tubing (outer diameter = 1.07 mm, inner diameter
= 0.55 mm) with one circumference of the cylinder representing
one cycle of the PCR reaction. The two temperatures were
created by two 50 W cartridge heaters (RS Components Ltd., Corby,
UK) which in turn were controlled by proportional-integral-derivative
(PID) digital controllers (CAL 9900, RS Components Ltd.).
The PTFE tubing was interfaced with a planar chip, in which
droplets could be produced. A rectangular bedding groove, which
would accommodate the tubing, was machined into the PMMA
planar chip (RS Components Ltd.). Ultra Violet (UV)-curable epoxy
(Norland 68, Norland Products Inc., New Brunswick, NJ, USA) was
used to fill the gap between the bedding groove and the tubing.
Finally a hole was drilled into the centre of the PTFE tubing with
a small drill (100 lm diameter) to allow the micro-fluidic channels
to join the PTFE tubing.
2.2. Justification of design
The design of the aluminium cylinder core with PTFE tubing
used to carry out the PCR amplification and analysis stage is
unconventional compared to most previously published devices.
The design of the aluminium cylinder and PTFE tubing device
was intended to circumvent some of the problems seen in more
conventional micro-machined devices. Devices based on reaction
wells [30], oscillatory systems [31–33], or closed-loop systems
[34,35] are restricted to a fixed throughput which is only a marginal
improvement over the conventional thermal cycler. However
devices, similar to the one described here, that are based on a
fixed-loop system [5,36–38] have the advantage of being high
throughput where samples can be continuously injected and collected
in an automated fashion. This fixed-loop system does
restrict the flexibility of the cycle number and the timings of the
cycles within an individual experiment. However, the cycle timings
can be directly controlled by adjusting the flow rate in order to
optimise the PCR for a particular experiment.
PTFE has a low coefficient of friction allowing the oil carrier
fluid and aqueous droplets to pass smoothly through the tubing
with minimal resistance or backpressure. Not only does this allow
for faster flow rates and thus a higher throughput of reaction but
breakages and adsorption of reagents onto the inner walls of the
channels is also minimised. Adsorption of samples onto the inner
walls of channels is a potential problem of many of the micro-machined
devices that are based on a single-phase flow of the reaction
mixture [31,32,34,36,37]. To overcome the problems of sample
adsorption onto the inner walls of the channels, and the possible
inhibition of the channel materials, many groups have used the
technique of silanization [39–42] or the inclusion of passivating
reagents such as PEG or BSA in the PCR reaction mixture [2]. Such
techniques are not required with the use of the hydrophobic,
chemically inert and virtually non-porous PTFE tubing. Similarly
PTFE tubing can withstand extreme temperatures whilst being
crack-resistant and stress-resistant making this material ideal for
PCR. Tubing of similar material has been effectively used to contain
samples in other miniaturized PCR devices [43] and has been
shown to be a favourable material for use in micro-fluidic
devices [44].
The use of a fluoro-carbon oil (FC-40, 3M) as an immiscible carrier
fluid further improves the design of this device as this oil has a
good heat transfer (specific heat = 1050 J kgÀ1
°CÀ1
) [45] The low
Fig. 1. Design of the device. This figure shows the schematic designs for both the aluminium cylinder device (far left) and the PMMA planar chip (far right, with zoomed
image of the channels in the centre) which were both at 90°. Due to the shear forces the flowing aqueous phase is sheared off into droplets which are then suspended in the
immiscible oil carrier fluid.
A.L. Markey et al. / Methods 50 (2010) 277–281 279
viscosity of FC-40 (3.4 centipoise) also aids in the generation of faster
flow rates and reduced resistance. [45] An additional benefit of
using this design of device, where PTFE tubing is wrapped around a
central core, is that the PTFE tubing can be easily removed and
replaced with new tubing if required. However the isolation of
both the sample and the reagents within an individual droplet
should greatly reduce, if not eliminate, sample loss and
contamination.
A more conventional PMMA chip has been designed for the
stage of droplet production. The micro-machined channels have
been created to carry the aqueous phase which will go onto form
the droplets. These channels meet the oil phase, which is carried
in the PTFE tubing, at 90°. The device was made to integrate the
PTFE tubing into the PMMA chip so that the oil could remain in
the PTFE tubing and move smoothly from one part of the device
to the next. Previously the oil channel was also micro-machined
into the PMMA substrate and then transferred to the PTFE tubing
via a 90° connector. This sharp angle caused a slight change in
the flow rate which led to the merging of droplets in the oil carrier
fluid at this point. The current design of the device, including both
the PMMA chip and the aluminium cylinder, is ideal for a smooth
and consistent flow rate as there are no sharp turns or connections.
The PMMA chip was also designed with two inlets so that the chip
would have the potential of acting as a droplet merging device. The
fusion of two droplets, one containing the sample and the other
containing reagents, by the application of an electric field [46],
would form the basis of sample preparation on this device. This
fused droplet can then be passed across a heater, which can be
incorporated onto the PMMA planar chip, to both lyse the cell
and activate the polymerase enzyme thus forming the basis of single
cell integrated preparation and analysis. The droplet will then
be transported, in the oil carrier fluid, through the thermocycles
of the cylinder device allowing the generation of data, expressed
in an absolute unit of molecules, for each individual cell in the
sample.
2.3. Droplet formation
A feature of this miniaturized device that differs from many
previously developed devices is that it is based on a system of
aqueous droplets suspended in an immiscible oil carrier fluid. This
system allows both the sample and reagents to be contained within
the droplet isolating them from factors such as sample loss by
adsorption onto the channel walls, or cross-contamination from
adsorption of previous samples. The aqueous droplets are formed
in this device at a T-junction where the inlet carrying the aqueous
phase meets the oil flow at 90°. Due to the shear force acting on
this aqueous phase, by the flow of the immiscible oil carrier fluid,
the aqueous solution is sheared off into droplets that quickly
become spherical due to surface tension.
This can be controlled in a passive manner by adjusting the flow
rates of the oil carrier fluid and the aqueous phase, which will form
the droplets, until a balance is reached such that droplets can be
produced. However adjustments to the flow rates of the oil and
aqueous phase also have an impact on various other factors such
as the rate of droplet production, the size of droplets produced,
the amount of time each droplet spends in each cycle of the PCR,
and the throughput of the reaction.
2.4. Impact of flow rates
Changing the flow rates of both the oil carrier fluid and the
aqueous phase in a systematic way will permit the optimisation
of droplet sizes allowing reagent costs to be reduced to a minimum.
Optimisation of the rate of droplet production is also necessary
so that a balance between throughput of the reaction and
prevention of cross-contamination by the merging of droplets
can be met. A thorough understanding of the effect of flow rates
on the time a droplet spends in each cycle of the cylinder is also
important for optimisation of the PCR to maintain efficient amplification
whilst increasing the throughput of sample analysis. The
consistency of the flow rates, measured by the time each droplet
spends in each cycle of the device, is also vital to ensure homogeneity
across the entire device.
2.5. Calibration of temperatures
Oneof the main limitations of conventional PCR is the throughput
of the technique as this is hindered by both spatial constraints and
the slow temperature ramp rate. The temperature ramp rate of this
device was explored using Chromazone slurry (Thermographic Measurement
Ltd., Flintshire, UK) which undergoes a colour change from
black to white at 67.3 ± 2.1 °C. This chromazone slurry was injected
into the PTFE tubing of the device and allowed to break into small
droplets which were dispersed throughout the tubing. The temperatures
were set to 75 and 40 °C and a video was recorded at 30
frames, per second using a Lumenera LU125 M-IO camera, of one
of these droplets passing between the two temperature zones. This
video was then edited into single frame images using Adobe Premier
Pro software. The number of frames taken for the droplet to change
from black to white at a known frame rate can then be used to calculate
the ramp rate of the device. The temperature ramp rate of this
device (5.83 °C/s) was found to be superior to that of a conventional
thermal cycler (4.8 °C/s). Also of importance is that this temperature
change is consistent across the entire device so that each cycle of the
PCR is equivalent. This can be seen by allowing the Chromazone slurry
to fill the entire cylinder and observing whether the colour change
is homogeneous across each cycle of the device.
Whilst a rapid and consistent temperature ramp rate is important
to generate a high-throughput PCR device this temperature
change should not disrupt the flow of droplets within the oil carrier
fluid. In order to test this the time taken for a droplet to pass
through one temperature zone can be recorded, in triplicate, at five
cycle intervals in the presence and absence of heat. This experiment
was carried out using two sample injection systems. First,
both the aqueous phase and oil were allowed to flow into the chip
via gravity from two 3 mL open syringes connected to the inlets. A
Gilson Minipuls 3 peristaltic pump was connected to the outlet and
set at 2 rpm. This reduced the resistive pressure allowing the oil
and aqueous phase to flow into the tubing via gravity. The second
system involved both the removal of the peristaltic pump and the
connection of the two syringes to a Harvard ‘‘33” syringe pump
system. The flow rate of the oil and aqueous phase were set to
6 lL/min and 3 lL/min, respectively, which were arbitrarily chosen
for ease of droplet counting. Although the peristaltic pump was
shown to aid in decreasing the resistance and backpressure in
the system the pump had an adverse effect on the reproducibility
of the flow of droplets. The flow of the droplets was inconsistent
both within and between cycles with the first system and was
affected by the heat of thermocycling. However no change in the
flow rate of the droplets was seen between or within cycles when
using the Harvard syringe pump system and this consistent flow
remained unaffected by the thermocycling process.
2.6. Limitations and future work
The device presented here is a prototype, designed to potentially
address some limitations of conventional PCR, such as sample
loss and contamination, through the integration of sample preparation
and analysis steps and the isolation of the samples within
individual droplets. Furthermore dropletized PCR may allow a
ready solution to the ‘‘averaging” problem seen when measuring
280 A.L. Markey et al. / Methods 50 (2010) 277–281
analytes within heterogeneous biological samples, by creating single
cell high-throughput PCR. This will generate a quantitative
result for each individual cell of a population, and simultaneously
enable the cell to be used as the denominator for assessing
expression activity and predict likely function. The described calibration
experiments showed that the flow of droplets within this
system is both stable and responsive to the changes in flow rate
yet unaffected by the thermal cycling process. An improved temperature
ramp rate has also been achieved on this device which
will directly increase the throughput of PCRs by allowing a faster
transition between the two temperatures of each PCR cycle. However,
despite the potential advantages of this novel device, additional
work is required to fully optimise the device for an
efficient PCR along with eventual improvements to enhance its
automation and ease of use.
3. Concluding remarks
The increased density of microtitre plate formats for PCR are
positively correlated to the evolving applications of PCR. Specifically,
the heightened requirement for throughput has been attributed
to both a genuine requirement to increase throughput of
routine gene-based assays, and also to enable the generation of
high clarity measurements of nucleic acid copies. To a greater
extent, the huge increase in throughput is necessitated to overcome
heterogeneity of cells in tissues, causing the bulk sample
preparation ‘averaging’ phenomena that preclude accurate measurement
and inability to closely associate multiple gene regulation
and decipher pathogenesis. A similar scenario exists for the
measurement of nucleic acids within a single cell. The gigantic
advances seen recently in 3rd Generation Sequencing platforms
(such as Genome Sequencer FLX System, Roche; SOLiD™ System,
AB, and Genome AnalyzerIIx, Illumina) permit the accurate assessment
of nucleic acid copy numbers by simply counting the number
of individually sequenced molecules. Indeed, whilst the dropletized
lTAS devices are under development using similar approaches
to those described in this article, it is most noteworthy
to reflect that the new sequencing platforms are based on emulsified
droplets in which PCR is performed. Therefore, the justification
and evidence that PCR can be encapsulated in droplets is well
founded (as seen with RainDance Technologies). The rationale for
the production of streams of PCRs in a micro-fluidic device is more
a function of enhancing sample handling, to enable PCR-based
nucleic acid measurements from complex biological matter.
References
[1] L. Chen, A. Manz, P.J. Day, Lab Chip 7 (2007) 1413–1423.
[2] P.A. Auroux, Y. Koc, A. deMello, A. Manz, P.J. Day, Lab Chip 4 (2004) 534–546.
[3] P.J.R. Day, Expert Rev. Mol. Diagn. 6 (2006) 23–28.
[4] R. Doshi, P.J. Day, Lab Chip 8 (2008) 1774–1778.
[5] S. Mohr, Y.H. Zhang, A. Macaskill, P.J.R. Day, R.W. Barber, N.J. Goddard, D.R.
Emerson, P.R. Fielden, Microfluid. Nanofluid. 3 (2007) 611–621.
[6] M.M. Kiss, L. Ortoleva-Donnelly, N.R. Beer, J. Warner, C.G. Bailey, B.W. Colston,
J.M. Rothberg, D.R. Link, J.H. Leamon, Anal. Chem. (2008).
[7] S.A. Bustin, V. Benes, J.A. Garson, J. Hellemans, J. Huggett, M. Kubista, R.
Mueller, T. Nolan, M.W. Pfaffl, G.L. Shipley, J. Vandesompele, C.T. Wittwer, Clin.
Chem. 55 (2009) 611–622.
[8] A. Manz, N. Graber, H.M. Widner, Sens. Actuator B 1 (1990) 244–248.
[9] J.H. Xu, G.S. Luo, S.W. Li, G.G. Chen, Lab Chip 6 (2006) 131–136.
[10] I. Kobayashi, K. Uemura, M. Nakajima, Colloids Surf. A 296 (2007)
285–289.
[11] D.R. Link, E. Grasland-Mongrain, A. Duri, F. Sarrazin, Z. Cheng, G. Cristobal, D.A.
Weitz, Angew. Chem. Int. Ed. 118 (2006) 2618–2622.
[12] A. Macaskill, P.R. Fielden, N.J. Goddard, S. Mohr, B.J.T. Brown, Proc. MicroTAS
2004 (2004) 207–209.
[13] L.Y. Yeo, R.V. Craster, O.K. Matar, J. Colloid Interface Sci. 306 (2007) 368–
378.
[14] H. Song, R.F. Ismagilov, J. Am. Chem. Soc. 125 (2003) 14613–14619.
[15] L.S. Roach, H. Song, R.F. Ismagilov, Anal. Chem. 77 (2005) 785–796.
[16] A. Liau, R. Karnik, A. Majumdar, J.H.D. Cate, Anal.Chem. 77 (2005) 7618–
7625.
[17] B. Zheng, J.D. Tice, L.S. Roach, R.F. Ismagilov, Angew. Chem. Int. Ed. 43 (2004)
2508–2511.
[18] B. Zheng, L.S. Roach, R.F. Ismagilov, J. Am, Chem. Soc 125 (2003)
11170–11171.
[19] M. Zourob, S. Mohr, A.G. Mayes, A. Macaskill, N. Perez-Moral, P.R. Fielden, N.J.
Goddard, Lab Chip 6 (2006) 296–301.
[20] W.J. Jeong, J.Y. Kim, J. Choo, E.K. Lee, C.S. Han, D.J. Beebe, G.H. Seong, Langmuir
21 (2005) 3738–3741.
[21] S.E. Barnes, Z.T. Cygan, J.K. Yates, K.L. Beers, E.J. Amis, Analyst 131 (2006)
1027–1033.
[22] G. Axel, K.F. Jenson, Lab Chip 6 (2006) 1487–1503.
[23] B.T. Kelly, J.-C. Baret, V. Taly, A.D. Griffiths, Chem. Commun. (2007) 1773–
1788.
[24] J.M. Kohler, T. Henkel, Appl. Microbiol. Biotechnol. 69 (2005) 113–125.
[25] A.J.d. Mello, Nature 422 (2003) 28–29.
[26] Y. Xia, G.M. Whitesides, Annu. Rev. Mater. Sci. 28 (1998) 153–184.
[27] L. Martynova, L.E. Locascio, M. Gaitan, G.W. Kramer, R.G. Christensen, W.A.
MacCrehan, Anal. Chem. 69 (1997) 4783–4789.
[28] M.A. Roberts, J.S. Rossier, P. Bercier, H. Girault, Anal. Chem. 69 (1997) 2035–
2042.
[29] R.M. McCormick, R.J. Nelson, M.G. AlonsoAmigo, J. Benvegnu, H.H. Hooper,
Anal. Chem. 69 (1997) 2626–2630.
[30] P. Wilding, M.A. Shoffner, L.J. Kricka, Clin. Chem. 40 (1994) 1815–1818.
[31] L.J. Chien, J.H. Wang, T.M. Hsieh, P.H. Chen, P.J. Chen, D.S. Lee, et al., Biomed.
Microdevices 11 (2009) 359–367.
[32] O. Frey, S. Bonneick, A. Hierlemann, J. Lichtenberg, Biomed. Microdevices 9
(2007) 711–718.
[33] T. Ohashi, H. Kuyama, N. Hanafusa, Y. Togawa, Biomed. Microdevices 9 (2007)
695–702.
[34] Y. Sun, Y.C. Kwok, N.T. Nguyen, Lab Chip 7 (2007) 1012–1017.
[35] A. Gonzalez, D. Ciobanu, M. Sayers, N. Sirr, T. Dalton, M. Davies, Biomed.
Microdevices 9 (2007) 729–736.
[36] S.F. Li, D.Y. Fozdar, M.F. Ali, H. Li, D.B. Shao, D.M. Vykoukal, J. Vykoukal, P.N.
Floriano, M. Olsen, J.T. McDevitt, P.R.C. Gascoyne, S.C. Chen, J.
Microelectromechan. Syst. 15 (2006) 223–236.
[37] N. Crews, C. Wittwer, B. Gale, Biomed. Microdevices 10 (2008) 187–195.
[38] Y. Schaerli, R.C. Wootton, T. Robinson, V. Stein, C. Dunsby, M.A.A. Neil,
P.M.W. French, A.J. deMello, C. Abell, F. Hollfelder, Anal. Chem. 81 (2009)
302–306.
[39] M.A. Shoffner, J. Cheng, G.E. Hvichia, L.J. Kricka, P. Wilding, Nucleic Acids Res.
24 (1996) 375–379.
[40] B.C. Giordano, E.R. Copeland, J.P. Landers, Electrophoresis 22 (2001) 334–340.
[41] N.J. Munro, A.F. Huhmer, J.P. Landers, Anal. Chem. 73 (2001) 1784–1794.
[42] M.C. Carles, N.J. Sucher, Methods Mol. Biol. 321 (2006) 131–140.
[43] K.D. Dorfman, M. Chabert, J.-H. Codarbox, G. Rousseau, P. De Cremoux, J.-L.
Viovy, Anal. Chem. 77 (2005) 3700–3704.
[44] H. Xu, M.L. Shuler, Biotechnol. Prog. 25 (2009) 543–551.
[45] L. Acota, Battlefield Enterprise Park, Shrewsbury, Shropshire.
[46] A. Macaskill, in: School of Chemical Engineering and Analytical Science, Vol.
Doctor of Philosophy, University of Manchester, Manchester, 2006, pp. 207.
A.L. Markey et al. / Methods 50 (2010) 277–281 281