Integrated sample-to-detection chip for nucleic acid test assays

Biomed Microdevices (2016) 18: 44
DOI 10.1007/s10544-016-0069-8
Integrated sample-to-detection chip for nucleic acid test assays
R. Prakash 1 & K. Pabbaraju 1 & S. Wong 1 & R. Tellier 1,2 & K. V. I. S. Kaler 3
Published online: 11 May 2016
# Springer Science+Business Media New York 2016
Abstract Nucleic acid based diagnostic techniques are routinely used for the detection of infectious agents. Most of these
assays rely on nucleic acid extraction platforms for the extraction and purification of nucleic acids and a separate real-time
PCR platform for quantitative nucleic acid amplification tests
(NATs). Several microfluidic lab on chip (LOC) technologies
have been developed, where mechanical and chemical methods
are used for the extraction and purification of nucleic acids.
Microfluidic technologies have also been effectively utilized
for chip based real-time PCR assays. However, there are few
examples of microfluidic systems which have successfully integrated these two key processes. In this study, we have implemented an electro-actuation based LOC micro-device that leverages multi-frequency actuation of samples and reagents
droplets for chip based nucleic acid extraction and real-time,
reverse transcription (RT) PCR (qRT-PCR) amplification from
clinical samples. Our prototype micro-device combines chemical lysis with electric field assisted isolation of nucleic acid in a
four channel parallel processing scheme. Furthermore, a four
Electronic supplementary material The online version of this article
(doi:10.1007/s10544-016-0069-8) contains supplementary material,
which is available to authorized users.
* R. Tellier
[email protected]
1
Provincial Laboratory for Public Health of Alberta, Calgary, 3030
Hospital Drive NW, Calgary, AB T2N 4W4, Canada
2
Department of Microbiology, Immunology and Infectious Diseases,
Cumming School of Medicine, University of Calgary, 3330 Hospital
Drive NW, Calgary, AB T2N 4N1, Canada
3
Department of Electrical and Computer Engineering, University of
Calgary, 2500 University Drive NW, Calgary, AB T2N1N4, Canada
channel parallel qRT-PCR amplification and detection assay is
integrated to deliver the sample-to-detection NAT chip. The
NAT chip combines dielectrophoresis and electrostatic/
electrowetting actuation methods with resistive micro-heaters
and temperature sensors to perform chip based integrated
NATs. The two chip modules have been validated using different panels of clinical samples and their performance compared
with standard platforms. This study has established that our
integrated NAT chip system has a sensitivity and specificity
comparable to that of the standard platforms while providing
up to 10 fold reduction in sample/reagent volumes.
Keywords Lab-on-a-chip . Nucleic acid tests .
Sample-to-detection . Droplet microfluidics .
Dielectrophoresis . Electrowetting
1 Introduction
The detection and identification of human pathogens is an
important function of medical microbiology laboratories.
Molecular diagnostic techniques such as nucleic acid amplification tests (NATs) provide the necessary sensitivity and specificity for this purpose (Saiki et al. 1988; Logan et al. 2009).
Additional advantages of NATs such as the real-time, reverse
transcription (RT) PCR (qRT-PCR) amplification include rapid turnaround time and improved biosafety profile resulting
from the inactivation of infectious agents during the sample
preparation step. Lastly, when the genomic sequences of
emerging pathogens are determined (van Elden et al. 2003),
nucleic acid assays can be quickly deployed for detection of
such pathogenic agents (Logan et al. 2009; Markoulatos et al.
2002). Investigation of a clinical infectious syndrome often
requires testing for several different agents that have a similar
clinical presentation, even if these agents are not closely
44 Page 2 of 14
related (Markoulatos et al. 2002). This leads to the concept of
a syndromic testing panel where the patient sample is tested
for a multitude of pathogenic agents by deploying a
multiplexed panel assay. Examples include viral gastroenteritis panel, viral encephalitis panel, or respiratory viral panel.
We have shown previously that this concept can be effectively
implemented with a spatially multiplexed qRT-PCR microdevice where parallel and independently controlled reactions
can be accomplished on chip (Prakash et al. 2015). Prior to the
PCR amplification step, NATs require a robust sample preparation step in order to disrupt pathogens or infected cells,
extract and then purify the nucleic acids. Present day sample
preparation platforms generally utilize chemical (chaotropic
salts) or mechanical (grinding, heat, sonication, beadbeating, microwave) disruption techniques and mechanical/
magnetic separation methods to obtain the purified nucleic
acid (Boom et al. 1990; Hourfar et al. 2005; Wu et al. 2014).
To date very few commercial platforms can achieve automated sample preparation integrated with real-time PCR detection assay. Examples include FilmArray® by BioFire
Diagnostics Inc. (Salt Lake City, Utah, USA), GeneXpert®
system by Cepheid (Sunnyvale, CA, USA) and Liat
Analyzer, IQuum (Allston, MA). These commercial platforms
are closed systems which work best when used with proprietary reagents/buffers and PCR assays, whereas our chip based
system is open and can therefore accommodate various commercial nucleic acid extraction kits and in-house assays, designed for other platforms such as the TaqMan® platform.
While NATs have been successfully miniaturized using
several different lab-on-chip (LOC) techniques including conventional closed channel microfluidic methods (Kopp et al.
1998; Saunders et al. 2013; Manz et al. 1990; Lagally et al.
2001; Mathies and Lagally 2004; Burns et al. 1998) and droplet microfluidic techniques (Sista et al. 2008; Chang et al.
2006; Fair et al. 2007; Pollack et al. 2000), the miniaturization
of nucleic acid based sample preparation has proven to be
more challenging. In a record of publications during 1993–
2013, the Web of Science reported more than 3000 articles
reporting different microfluidic implementations of NAT techniques, with almost all of them requiring off-chip sample
preparation (Wu et al. 2014). This is principally due to the
difficult requirement of combining chemical, mechanical
and/or magnetic components on a LOC platform (Wu et al.
2014; Kaler and Prakash 2014). The chemical disruption of
biological samples has been implemented in closed channel
(Price et al. 2009; Liu et al. 2004; Gijs 2004) and droplet
microfluidics (Sista et al. 2008; Chang et al. 2006;
Wijethunga et al. 2011), assisted by off-chip components for
magnetic/mechanical washing and capture of the extracted
DNA/RNA (Kim et al. 2014; Tanriverdi et al. 2010; Lee
et al. 2006).
In examples of LOC implementation of an integrated NAT
scheme, Sista et al. (2008) realized a digital microfluidics
Biomed Microdevices (2016) 18: 44
(DMF) chip based sample preparation and PCR amplification
by combining chemical lysis and magnetic separation/capture
of micro-beads, with electrowetting (EW) based droplet
actuation and PCR thermal cycling. In another example,
Kim et al. (2014) implemented a close channel Lab-on-aDisc system combining chemical lysis with liquid-liquid
phase separation for extraction and purification of nucleic acid
prior to the PCR amplification step.
In this work, we have designed a multi frequency electroactuation based LOC system that leverages electric field to
facilitate the handling of samples and reagents for the entire
chip based nucleic acid extraction, purification and qRT-PCR
amplification steps. Our prototype NAT chip integrates chemical lysis with electric field mediated washing and purification
of nucleic acid over four independent parallel sample preparation channels, which are interfaced to four channel parallel
RT-PCR amplification and detection assay region. The sample
preparation chip module presented here is a viable alternative
to the magnetic and mechanical separation techniques which
greatly limit the parallel sample handling and processing aspect of the other LOC systems. Our chip employs electrical
methods to capture/trap micro-beads by leveraging
dielectrophoresis (DEP), which has been widely used to manipulate particles (King et al. 2005; Prakash and Kaler 2012;
Wang et al. 1998), cells (Wang et al. 1998; de la Rosa et al.
2008; Park et al. 2011) and fluidic samples (Kaler et al. 2010;
Jones 2001; Prakash et al. 2010) based on electrical and geometric properties (Wiklund et al. 2006; Schnelle et al. 1996).
The application of DEP in droplet microfluidics has shown
that such effects can also be leveraged in transient liquid jets
(King et al. 2005) and in EW actuated droplets for separation
of particles based on their size and electrical properties (Zhao
et al. 2007). Previously, integration of DEP with EW necessitated a two-surface micro-device where one surface (top) was
covered with patterned EW electrodes and the second surface
(bottom) was patterned with DEP electrodes (Zhao et al.
2007). We have overcome this limitation with a single surface,
multi-layered, micro-electrode design, made feasible by the
herring-bone electrostatic droplet actuation (Kaler and
Prakash 2014; Washizu 1998) and the castellated DEP
micro-electrodes (Wang et al. 1998; Zhao et al. 2007) to generate sufficient DEP field effect, required for the trapping and
washing of micro-beads during the sample preparation
process.
We have demonstrated reliable and efficient nucleic acid
extraction for many different infectious agents and specimen
matrices. The performance of the sample preparation chip (see
Fig. 1) was established through experiments where five different panels of clinical samples were extracted on-chip and
the results compared with a benchmark extraction and RTPCR detection platform. After thorough testing and evaluation
of the sample preparation chip, we implemented and tested the
integrated NAT chip, which was designed as a four channel
Biomed Microdevices (2016) 18: 44
Page 3 of 14 44
Fig. 1 a Photomicrographs of (a)
the four channel sample
preparation chip; b integrated
particle-DEP, droplet-DEP microelectrode architecture c DEP
capture of micro-beads
(diameter ~ 2 μm) over castellated
DEP electrodes (w = g = 10 μm);
DEP capture completed
(t = 120 s); d D-DEP transport
and re-suspension of micro-beads
in wash buffer
sample preparation module interfaced with a spatially
multiplexed RT-PCR assay module (Prakash et al. 2015) on
a single glass substrate. Snapshots of the micro-electrode
structures as well as the different chip modules are shown in
Fig. 2.
Apart from the initial manual sample/reagent loading
steps prior to the sample preparation and the RT-PCR amplification steps, the NAT chip applies DEP, electrostatic/
electrowetting actuation methods, along with resistive
micro-heaters and temperature sensors to perform sampleto-detection assays on a variety of clinical samples, in an
automated fashion.
2 Theory
2.1 Characteristics of the chip micro-electrode
architectures
The planar micro-electrode architecture used to affect bead
capture and washing procedures, as part of the sample preparation chip module, leverages and combines two different
electric field effects (see Fig. 1a). One is low frequency AC
electro-actuation method that relies on the spatial nonuniformity in the electric field to transport individual droplets,
referred to as droplet-DEP (D-DEP) (Vact: 80 Vpp, fact: 40–
60 Hz) (Kaler et al. 2010; Washizu 1998) and the second is an
electrowetting (EW) scheme (Vact: 100 Vpp, fact: 30 Hz)
(Vergauwe et al. 2014; Mugele and Baret 2005), to transport
sample and reagent droplets in the volume range of 10–25 μL.
In addition, we used a higher frequency AC fields, ranging
from 300 kHz to 800 kHz to electro-actuate particles and
achieve the negative DEP capture of micro-beads in the different extraction and wash buffers, during the sample preparation steps (Fig. 1b–d). The underlying principles of D-DEP
and EW have been previously reported and analyzed
(Washizu 1998; Mugele and Baret 2005; Moon et al. 2002;
Choi et al. 2012; Wheeler 2008). In this work we resorted to a
numerical analysis method for elucidating and deriving the
optimized experimental conditions to achieve efficient
negative-DEP capture of micro-beads, suspended in the
different extraction and wash buffer media.
We have previously demonstrated a chip-based parallel
RT-PCR scheme where a D-DEP droplet transport scheme
was integrated with thermostatic zones (micro-heaters and
resistive temperature sensors) (Prakash et al. 2014, 2015),
to enable droplet transport and thermal cycling for parallel
RT-PCR without the requirement of active electrode
switching during the amplification reaction (Fig. 2a, b).
This substantially reduced the electrical overhead, suitable
for the illustrated sample-to-detection NAT system. We
have now combined the sample preparation module with
the chip RT-PCR detection module (Fig. 2a) to achieve
the sample-to-detection NATs on panels of clinical samples with a limit of detection in the order of ~10–100 viral
RNA copies per reaction. An important feature of the
integrated NAT scheme is that it provides four independent sample preparation and RT-PCR detection channels,
which can be leveraged to achieve parallel (up to four
samples) sample-to-detection assays. Furthermore, the integrated NAT chip has a built-in flexibility to accommodate different sample preparation and thermal cycling process parameters for each of the 4 independent chip NAT
channels.
2.2 Manifestation of negative-DEP for capture of nucleic
acid binding micro-beads
DEP is an electrokinetic phenomenon where a spatially nonuniform field can be leveraged to affect the behaviour of
44 Page 4 of 14
Biomed Microdevices (2016) 18: 44
Fig. 2 a Photomicrograph of the four channel NAT assay chip; b CCD
image of the integrated droplet-DEP and thermostatic zone for RT-PCR.
c CCD images of a 10 μL PCR droplet during the 60 °C annealing phase
where the fluorescent signal is recorded; d Image of chip-PCB assembly
with sealed assay chamber during the chip based NAT assay
dielectric particles within a fluidic medium based on differences in their electrical properties (Pohl 1958, 1978). In
microfluidic systems, the DEP field effect is often generated
using electrically energized micro-electrodes, which in the
case of our DEP chips are coplanar micro-electrode structures
fabricated on a glass wafer (see Fig. 1). The nature of DEP
force induced on individual particle, suspended in aqueous
fluidic media, is formulated in terms of the electrical properties of the particle, carrying fluidic medium and size/structure
of the particle. The non-uniform electric field emanating from
the co-planar micro-electrodes when biased with an AC voltage results in generation of DEP force acting on the suspended
particle, expressed as follows (Pohl 1978):
h i !
FDEP ¼ 2πεm Rp 3 Re K ∇ jErms j2 ^r
ð1Þ
where Rp is the radius of the spherical particle, Erms is the root
mean square value of the non-uniform electric field exerted on
the particle and K is the Clausius-Mossotti (CM) factor
which governs the effect of the actuation frequency on the
Biomed Microdevices (2016) 18: 44
Page 5 of 14 44
resulting DEP force and is dependent on the electrical properties of the particle and the fluidic media (Pohl 1978; Honegger
et al. 2011):
K ð ωÞ ¼
εp −εm
εp þ 2εm
; where ε ¼ ε − jσ =ω
ð2Þ
The quantity ε represents the complex, frequency dependent permittivity of the particle and that of the fluidic media,
ω is the angular frequency of the applied AC electric field
(ω = 2πf ) and ε, σ are the electrical permittivity and conductivity respectively. It is readily apparent from Eq. 1 that the
sign of the real component of the complex CM factor i.e. Re
K determines the salient character of the DEP force, with its
value confined between the limits −0.5 to +1.0. A negative
value of Re K results in a net body force that impels particles
away from regions of high electric field (negative-DEP)
whereas a positive Re K impels particles towards regions
of high field intensity. The five parameters: σp, εp, σm, εm
and ω collectively modulate the DEP force through the CM
factor, resulting in the positive or negative DEP effect on the
subjected particles. This also suggests that for any combination of particle and fluidic medium, one can adjust the AC
frequency to induce the desired DEP effect. This is essentially
achieved by analyzing the cross-over frequency ( fc), at which
point Re K ¼ 0 and the DEP force switches sign, either from
positive to negative or vice versa. This cross-over frequency,
at which |FDEP| = 0 can be calculated as (Pohl 1978; Honegger
et al. 2011):
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
σm − σeq σeq þ 2σm
f c ¼ 1=2π
εeq − εm εeq þ 2εm
ð3Þ
The determination of fc is crucial in applications where the
particles are immersed in different buffers with large variation
in medium conductivity, as is often the case in nucleic acid
extraction assays, since the capture and release of extracted
nucleic acids from the particle is mediated by the pH and ionic
strength of the buffer. Thus, based on this simplified particle
model, there are two regimes of DEP forces separated by the
cross-over frequency ( fc). As reported in the materials and
methods section, the buffer conductivity in our application
ranged from ~20 μS/cm (elution buffer) to ~150 mS/cm (almost three orders of magnitude variation in medium conductivity) while the particle size (diameter: 2 μm), was maintained constant. Hence, in order to elicit a negative DEP response from these specific particles, we opted to numerically
analyze the fc, experimentally validate it and implement it for
solid-liquid separation step, in the different buffer media. We
also re-calculated the effective medium and particle complex
permittivity that correlates to their dielectric permittivity and
conductivity by using Eqs. 4 and 5 respectively. Here, Eq. 4
adjusts the medium permittivity by using the experimentally
calculated solid fraction term (ϕ, vol./vol.%) to account for the
effective medium properties for the colloidal solution using
Maxwell’s mixture theory (Vergauwe et al. 2014) and Eq. 5 is
used to calculate the effective dielectric property of the particle
which has a core-shell structure (magnetite core and silica
shell) as reported in (Pohl 1978; Honegger et al. 2011).
εmix ðωÞ ¼ εm
1 þ 2ϕK ðωÞ
1−ϕK ðωÞ
; ε ¼ ε − j σ =ω
εc −εs
þ2
εc þ 2εs
σeq
3 ε −ε ; εeq ¼ εeq −j =ω
c
s
Rp
−
Rj
εc þ 2εs
ð4Þ
3
εeq ¼ εs
Rp
Rj
ð5Þ
where Rj is the core radius of the particle, εc and εs are the
complex permittivity of the particle core and shell. Finally, the
electrical conductivity of the core-shell particle is evaluated in
terms of the bulk conductivity (σb) and the surface conductance (Ks) of the particle, reported as follows (Honegger et al.
2011):
σ p ¼ σb þ
2K s
Rp
ð6Þ
where K s is the surface conductance of the particle
(Ks = 0.5 nS (Wang et al. 1998)), σb is the bulk conductivity
of the core (σb ~ 5 × 10−4 S/m) and (Rp = 1 μm) is the particle
radius. This numerical analysis helped to establish the experimental parameters required for negative-DEP capture
of the particles (Nucleic acid binding micro-beads) in the
four different buffer media, utilized in the sample preparation
assay (see appendix 1(ESM 1)). The results of the numerical
analysis are reported in the following section. The extracted
values were then incorporated in a COMSOL Multiphysics
simulation to illustrate the gravitational sedimentation assisted
negative-DEP capture of the micro-beads (Online Resource;
Video 1).
2.3 Estimation of negative DEP based micro-bead capture
parameters
The castellated micro-electrodes (Fig. 1b) were used to affect
the negative DEP capture of micro-beads during washing procedure, which was repeated five times per extraction assay, in
four different buffers. During the optimization process, the
electrode dimensions (width/gap values) were initially based
on the micro-bead diameter and the micro-bead concentration,
in order to provide sufficient capture area for efficient collection of nucleic acid during the repeated washing steps. In the
previous section, we established a numerical model to analyze
44 Page 6 of 14
and optimize the negative DEP capture of the nucleic acid
binding beads. Four different micro-electrode designs were
used during this optimization process with electrode width
(and gap) varied from 5 μm to 50 μm for studying the effectiveness of negative DEP capture. The larger micro-electrodes
(width = gap > 20 μm) required a higher actuation voltage and
resulted in relatively longer capture times whereas the smallest
electrode structure (width = gap = 5 μm) resulted in insufficient capture area, leading to the accumulation of particles in
the electrode gaps and formation of unstable pyramid shaped
structures in these regions. We also incorporated the fluidic
and micro-bead properties reported in the appendix 1(ESM 1),
into Eqs. 3–6 in order to determine the variation of DEP force
with medium conductivity and actuation frequency. Fig. 3a
shows the variation of the crossover frequency (fc) with
Fig. 3 a Numerical estimation of
the DEP crossover frequency with
medium conductivity and linear
regression showing confidence of
fit for the numerical plot over the
experimental range of medium
conductivity; b variation ofcross
over frequency ( fc) and Re K
with DEP actuation frequency for
the micro-bead actuation
(diameter ~ 2 μm) in the four
extraction buffers used in the
nucleic acid extraction assays
Biomed Microdevices (2016) 18: 44
medium conductivity, for the four extraction buffers. In
Fig. 3, we have also taken into account the colloidal medium
properties, by using the effective medium permittivity
(Fig. 3a), and the
conductivity (Fig. 3a) parameters
effective
to adjust the Re K in Eqs. 1 and 2 (Fig. 3b). The adjusted
crossover frequency values were found to be in close agreement with the values obtained from Fig. 3, and were used to
select the DEP actuation frequencies to affect negative DEP
capture in the four extraction buffer media, as reported in
Fig. 3b. The negative DEP capture using the castellated DEP
micro-electrodes is illustrated in the attached video file
(Online Resource; Video 2). The findings of the numerical
analysis were extended to a 2-D COMSOL Multiphysics
simulation where the optimized capture parameters were used
in a Monte-Carlo simulation to illustrate real-time capture of
Biomed Microdevices (2016) 18: 44
the micro-beads subjected to both sedimentation and negativeDEP forces (Online Resource; Video 1).
Page 7 of 14 44
clinical samples prior to the on-chip experiments. Similar to
the validation procedure for the chip based sample preparation
module, four panels of clinical samples were used for validation of the integrated NAT chip (see Table 1b).
3 Materials and methods
A generic nucleic acid extraction kit, Mag-Bind® Viral DNA/
RNA Kit supplied by Omega Bio-Tek (GA, USA), was used to
develop and validate the extraction protocols for the sample
preparation chip. The kit components include micro-beads, lysis, wash and elution buffers. The different physical and chemical attributes of the micro-beads and extraction buffers were
either supplied by the supplier (Omega bio-tek Inc., GA, USA)
or measured in-house (pH, conductivity) using the necessary
laboratory instruments, as reported in appendix 1(ESM 1).
3.1 Characteristics of the clinical panels used for the chip
based sample preparation and NATs
MS2 is a bacteriophage with an RNA genome and it is frequently used as an extraction control in RNA extraction
(Dreier et al. 2005). The chip sample preparation module
was first investigated using serial dilutions of MS2 phages
prepared at the provincial laboratory for public health of
Alberta (ProvLab) Calgary. Details of the MS2 validation experiments are provided separately in the appendix 1(ESM 1).
Following the MS2 validation experiments, four panels of
anonymized clinical samples with a single infection target
and a fifth co-infection panel (with three targets per sample),
in different specimen matrices (see Table 1a) were prepared at
the ProvLab.
The purpose of these different panels was to test the reproducibility and accuracy of the chip based sample preparation
assay in different specimen matrices. The chip extraction was
rigorously compared to the total nucleic acid extraction on the
easyMAG™ system (BioMerieux, QC, Canada) which is
widely used in clinical diagnostic laboratories and served as
the gold standard for the chip extraction assays. All the patient
samples, reported in Table 1 were submitted to the ProvLab
during 2013–2015 and stored in the Universal Transport
Medium (UTM™; Copan Diagnostics Inc., CA, USA). The
patient samples were mixed with the TNA (Total Nucleic
Acid) lysis buffer before being used in the chip based extraction assays. In all of our chip experiments, the mixing of the
sample and the lysis buffer was done Boff-board^ and inside a
biosafety cabinet, prior to loading the sample onto the chip;
this was implemented by design as a biosafety control measure while handling patient samples, as it would inactivate
pathogens and make the method applicable even when dealing
with dangerous pathogens. The addition of the lysis buffer
ensures that the viral RNA is stabilized in a tailored buffer
solution that denatures proteins, resulting in the inactivation
of RNases and DNases, and loss of pathogen infectivity in the
3.2 Primers and probes for the chip and TaqMan®
RT-PCR assays
The primer and probe sequences from previously reported
real-time RT-PCR assays (Pabbaraju et al. 2011; Pabbaraju
et al. 2015) were used for the detection of the viral RNA from
Influenza A, Influenza B, RSV and enterovirus targets. PCR
amplification was performed as one-step RT-PCR using the
TaqMan® Fast Virus One-Step RT-PCR Master Mix, 0.8 μM
each of sense and antisense primers and 0.2 μM of the labeled
probe. Five microliters of the chip extracted RNA target was
combined with 5 μl of this PCR master mix. The one step RTPCR reaction parameters consisted of a reverse transcription
step at 50 °C for 5 min, enzyme activation at 95 °C for 20 s,
and 38 cycles of denaturation at 95 °C for 3 s and annealing/
extension at 60 °C for 30 s.
3.3 Analysis of the on-chip RT-PCR assay
The threshold cycle, Ct is a quantitative value which indicates
the PCR cycle number from the amplification assay at which a
definitive signal, above the background threshold level is observed and quantified (Prakash et al. 2014). In diagnostic RTPCR assays, the Ct value is directly related to the viral load in
the sample, when operating within the dynamic range of the
assay, as seen in the reported panel samples of Table 1. The Ct
values of the panel samples were used to evaluate the extraction efficiency of the sample preparation chip and the overall
performance of the NAT chip compared to that of the gold
standard platform. The extracted nucleic acid from all chip
and easyMAG™ extracted samples were tested in a clinical
diagnostics laboratory, using the Applied Biosystems 7500
Fast Real-Time PCR equipment (Life Technologies, USA).
The Ct values obtained from the ABI 7500 Fast System
were used as the gold standard for comparison of both the
chip and the easyMAG™ extracted samples reported in the
results and discussion section. The chip based NAT performance was also analyzed by comparing the chip RT-PCR
efficiency to the ABI 7500 Fast system (Pabbaraju et al.
2011). For the chip extractions, volumes of different reagents
and extraction buffers were appropriately scaled down in the
ratios recommended by the supplier (see appendix 1(ESM 1)).
3.4 Device fabrication and procedure for chip based NATs
The sample preparation chip and the integrated sample-todetection NAT chip were fabricated at the micro-fabrication
facility, Nanofab (Edmonton, Canada). The micro-chips
44 Page 8 of 14
Table 1 Characteristics of the
different clinical panels used for
validation of the chip based
sample preparation assays and the
integrated sample-to-detection
NATs
Biomed Microdevices (2016) 18: 44
Panel
Sample No.
Sample Type
Target
ProvLab Ct
a. Clinical panels for validation of chip sample preparation
Panel of NP swab samples to test reproducibility of chip sample preparation
1
Flu A
30.30
5, 6, 7, 8
NP Swab
Flu A
Panel of NP swab samples to test reproducibility of chip sample preparation
25.40
2
1, 2, 3, 4
NP Swab
1
NP Swab
N.A.
Neg.
2, 3, 4
NP Swab
Flu A
20.52
5, 6, 7
NP Swab
Flu A
29.77
8
NP Swab
N.A
Neg.
Nucleic acid extraction from plasma matrix (Enterovirus targets)
3
1, 2
Plasma
E18
33.50
3, 4
5
Plasma
Plasma
CVA6
N.A.
27.68
Neg.
Panel of NP swab samples to test chip sample preparation over SH surface
4_SH
1
NP Swab
Flu B
17.49
2
NP Swab
Flu A
27.72
3
NP Swab
RSV
20.35
4
NP Swab
N.A.
Neg.
A,B, RSV
A,B, RSV
16.34, 27.72, 20.35
16.34, 17.49, 23.60
Panel of eight clinical samples with co-infections of respiratory viruses
5
1
2
NP Swab
NP Swab
3
NP Swab
A,B, RSV
26.44, 27.72, 20.35
4
NP Swab
A,B, RSV
26.44, 17.49, 23.60
5
NP Swab
A,B, RSV
16.34, 17.49, 20.35
6
NP Swab
A,B, RSV
26.44, 27.72, 23.60
7
NP Swab
A,B, RSV
16.34, 27.72, 23.60
8
NP Swab
A,B, RSV
26.44, 17.49, 20.35
b. Clinical panels for validation of the integrated NAT chip
Panel of NP swab samples for accuracy of chip based NATs
1
1
NP Swab
Flu A
30.56
2
NP Swab
Flu A
34.54
3
NP Swab
Flu A
25.58
4
NP Swab
N.A.
Neg.
Panel of plasma matrix for chip based NATs
2
1
Plasma
CVA6
27.68
2
3
Plasma
Plasma
E18
E18
33.19
35.00
4
Plasma
N.A.
Neg.
Panel of NP swab samples to test the dynamic range of chip NATs
3
1
NP Swab
Flu A
19.45
2
NP Swab
N.A.
Neg.
3
NP Swab
Flu A
25.58
4
NP Swab
Flu A
31.12
Panel of NP swab samples to test for possible contamination issues
4
1
NP Swab
N.A.
Neg.
2
NP Swab
N.A.
Neg.
3
NP Swab
N.A.
Neg.
4
NP Swab
Flu B
17.05
NP Swab Nasopharyngeal Swab, A, Flu A Influenza A virus, E18 Echovirus 18, CVA6 Coxsackievirus A6; B, Flu
B Influenza B virus, RSV Respiratory Syncytial virus, Neg. no target Ct, ProvLab Ct: Threshold PCR cycle value
from the original diagnostic RT-PCR assay on the patient sample when submitted to ProvLab, N.A. Negative
control (negative patient sample or, UTM™). For panel 5, two different patient samples with different initial viral
loads for each of A, B and RSV were combinatorially mixed for the eight panel co-infection samples
Biomed Microdevices (2016) 18: 44
measuring 6.5 cm × 3.0 cm were fabricated using a 10 cm
square glass (Borofloat) wafer. The sample preparation chip
module, shown in Fig. 1, consists of three different electrode
structures: (1) castellated electrodes with different width/gap
sizes for DEP capture of the nucleic acid binding micro-beads
(Fig. 1a, c), (2) a herring-bone shaped D-DEP electrodes for
cycling of sample and buffer droplets during the DEP capture
and, (3) an EW electrode structure for loading wash, elution
buffers onto one or more nucleic acid extraction/purification
tracks. The micro-electrode structures were fabricated in two
different metallization layers (Cr/Al: 200 nm). The DEP electrodes and the droplet actuation electrodes were electrically
isolated and passivated by a thin 500 nm silicon nitride layer,
to prevent sample electrolysis. For panel experiments 1, 2, 3
and 5 of the sample preparation assay, the top dielectric layer
was furthermore rendered hydrophobic by depositing a composite fluorocarbon (FC) coating (composed of 30 nm of plasma deposited FC and 30 nm of spin coated Teflon AF® 2400),
as previously reported (Prakash et al. 2012). For panel 4_SH
experiment on the sample preparation chip and for all sampleto-detection NAT chips, the top dielectric layer (Si3N4) was
nano-textured to create a super hydrophobic (SH) top surface,
utilizing a soft lithography technique previously reported by
Prakash et al. (2013). The SH surface provided a large droplet
contact angle (CA ∼ 156o) during the device application and
significantly reduces the bio-sample adsorption.
The RT-PCR chip module of the NAT chip, shown in
Fig. 2, consists of: (1) an array of photo lithographically patterned chromium (Cr thickness: 200 nm) micro-heaters and
resistance temperature detectors (RTDs) to create the two thermostatic zones (thermal zones 1 and 2) required during the
thermal cycling, (2) a photo lithographically patterned aluminum (200 nm) layer for D-DEP electrodes and for the EW
tracks, utilized for loading the PCR primers, probes and reagent mix droplets to the thermal cycler electrodes. The different metal layers were electrically isolated from each other
and passivated using dielectric stacks of silicon nitride (thickness: 500 nm), to prevent sample electrolysis during electroactuations.
In each of the panel NATs, TNA lysis buffer was added offchip to the patient sample. Also, prior to the chip assay, microbeads and other reagents were added to create the 75 μL sample aliquot which contained 20 μL of the patient sample. The
clinical specimens were chemically lysed and the nucleic acid
bound to the surface of micro-beads. Volumetric details for the
different washing and elution steps are provided in Table A.1
of the appendix 1 document (ESM 1).
Photomicrographs of droplet actuation over the three different segments of the sample preparation micro-electrodes
are shown in Fig. 1c, d. Following the off-chip mixing, the
resultant volume of ~75 μL is serially pipetted into three equal
aliquots of 25 μL onto the EW electrode array and carried over
to the herringbone shaped D-DEP electrode (AC voltage:
Page 9 of 14 44
100 Vpp, AC frequency: 60 Hz; see Fig. 1). The D-DEP tracks
(AC voltage: 80 Vpp, AC frequency: 40 Hz; see Fig. 2a) were
designed to handle droplets in the volume range of 10–25 μL
(Prakash et al. 2014). The droplet track is fragmented into the
DEP capture zones (Fig. 1b) which allows for positioning of
the droplet into the DEP capture zone where the D-DEP actuation is turned off and the high frequency negative DEP capture field is turned on (AC voltage: ~100 Vpp; frequency:
750 kHz). The droplet is unperturbed for the duration of
micro-bead capture (Fig. 2b). Once the micro-bead capture
is complete, the D-DEP actuation is re-established to gently
pull the droplet out of the capture zone, leaving behind
~500 nL of solution trapped with micro-beads immobilized
on the chip surface (Fig. 2b, c). This capture and release step is
repeated three times to trap all the micro-beads from the 75 μL
sample aliquot (Fig. 2a, c). For the washing step, wash buffer
droplets of 20 μL are actuated over the D-DEP tracks onto the
captured micro-bead cluster which results in the re-suspension
and washing of micro-beads (Fig. 2d). The micro-beads were
re-captured on the surface (AC frequency: 350 kHz) after each
washing and finally following the elution step (see Online
Resource; Video 2). In the prototype version of the sample
preparation chip, the eluted nucleic acid samples are recovered
manually with a micro-pipetter. All panel sample preparation
assays were carried out using the four channel sample preparation chip, shown in Fig. 1.
3.5 Experimental set-up
Several on-chip experiments were designed to illustrate the
extraction and purification of viral RNA from patient samples.
The experimental set-up was built around the microfluidic
chip-printed circuit board (PCB) assembly (Fig. 2d). The
set-up, which was housed on an Olympus BX-51 fluorescence
microscope stage, was actuated and position controlled by the
NI PXIe digital controller (National Instrument, Austin, TX,
USA). The optical components, currently housed on the platform include: micro-photomultiplier tube (microPMT;
H12400–00-01, Hamamatsu, Japan) for read-out during onchip, RT-PCR assays; a color charge-coupled device (CCD)
camera (QImaging, Surrey, Canada) and a high speed complementary metal oxide semiconductor (CMOS) camera
(Canadian Photonics Lab, Manitoba, Canada) for visual inspection and video/image capturing; a motorized xyz stage,
controlled by an OptiScan unit (Prior Scientific, USA) via
NI LabVIEW program for real-time scanning, optical imaging
and fluorescent signal read-outs during the parallel RT-PCR
amplification step.
The experiments were carried out with the chip secured on
a printed circuit board (PCB) and a plexiglass fixture secured
on top with silicone sealant and a top glass cover-slide, creating a sealed enclosure for the chip assays, as shown in Fig. 2d.
The glass cover-slide is removed only during the sample
44 Page 10 of 14
loading, collection of the purified nucleic acid samples and
removal of waste reagents.
4 Results and discussion
4.1 Chip based nucleic acid sample preparation assays
using panels of clinical samples
The sample preparation chip (Fig. 1) was used to extract nucleic
acids from five different clinical panels. Characteristics of these
panels are reported in Table 1a. Each panel was designed to
investigate and validate key performance parameters of the
sample preparation chip in comparison to the benchmark
extractor.
For characterizing the reproducibility and accuracy of the
sample preparation chip, nasopharyngeal swab samples from
patients (Table 1a) were extracted in quadruplicates for panel
1 and triplicate for panel 2, in parallel over the four channel
sample preparation electrodes, shown in Fig. 1.
The Ct values obtained by amplifying the target Influenza
A virus from the chip extracts were compared to the
easyMAG™ extracted samples and reported in Fig. 4. The
reported Ct values show reproducible performance on the chip
with extraction efficiency comparable to the benchmark
platform.
All panel 2 samples (see Table 1a) were extracted in triplicate using three separate four channel sample preparation
chips. Each chip processed one aliquot from each of the four
panel 2 samples. The Ct outcomes from the TaqMan® RTPCR of the chip and the easyMAG™ extracts for panel 2
samples are summarized in Fig. 4. The extracted aliquots were
also tested for MS2 (Internal control) (Fig. A.1;
Appendix 1(ESM 1)) to evaluate the effects of potential
PCR inhibitors that may have been co-extracted from the
specimen. All the Influenza A RNA was extracted with high
efficiency (RT-PCR efficiency ~94.8 %) and reproducibility,
comparable to the validation platform.
Having benchmarked the efficiency and reproducibility
of the chip extraction assay for several different nasopharyngeal swab samples, the chip performance was further
investigated for more complex human specimen matrices.
To facilitate this, a third panel of clinical samples
(Table 1a: panel 3), consisting of two positive plasma samples and a negative control (blood plasma) were extracted
using the sample preparation chip. The plasma samples
were extracted in duplicate, on two four channel sample
extraction chips. The Ct values obtained for the different
viral targets from the chip extracted samples were compared to those from the commercial extractors as reported
in Fig. 4a. The chip performed equally well for the plasma
samples and the extraction of the two enterovirus targets
was sensitive and reproducible.
Biomed Microdevices (2016) 18: 44
Since the surface topology of the sample-to-detection NAT
chip was designed to be nano-textured and SH, a fourth panel
(Table 1a: panel 4_SH) was subjected to a sample preparation
chip with the SH top surface. Three of the four NP swab
samples on this panel contained a single yet different viral
target, with a wide range of viral loads (up to 3 logs) in an
effort to establish that the performance of the sample preparation chip module remained consistent over both hydrophobic
and SH surfaces. As the outcomes of the panel 4_SH sample
preparation assay in Fig. 4a illustrate, we found minimal variation in chip sensitivity and extraction efficiency with the
changed surface topology.
4.2 Chip based sample preparation assay using clinical
samples with co-infecting viruses
Co-infections with more than one virus species can occur in
some diseases; for example this is a common occurrence in
viral upper respiratory tract infections. It is therefore important to ensure that the chip based extraction can recover all
the co-infecting viruses present in the patient samples.
Extraction of co-infection samples allowed us to test for
diagnostic scenarios where more than one viral species
may be present at different viral loads. Samples were prepared at ProvLab where reference patient samples with two
different viral loads (high Ct and low Ct) of three different
respiratory pathogens (Flu A, Flu B and RSV) were combinatorially mixed to generate an eight sample co-infection
panel (panel 5, Table 1a).
Each sample was extracted in duplicate (using 4 separate
sample preparation chips), resulting in up to ~25 μL of extract
per panel sample, which were analyzed using the TaqMan®
RT-PCR platform for detection and quantification of each of
the three viral targets. Results were then compared to the adjusted values of the reference patient samples, as reported in
Fig. 4b. The Ct values measured from chip and easyMAG™
extracts were in excellent agreement for the different viral
loads in all co-infection samples.
4.3 Chip based integrated sample-to-detection NATs
To validate the sample-to-detection NAT chip shown in Fig. 2,
we used four panels of patient samples, in two different sample matrices. For each of the four panel assays, samples were
mixed off-chip with the lysis buffer, reagents and microbeads, as described in the experimental section. The mixture
was loaded on each of the four sample preparation channels
using EW droplet actuation (see Fig. 2). Upon completion of
the extraction step, approximately 7.5 μL of each eluted sample was manually pipetted out for testing on the TaqMan®
platform. For each of the four panel chip NATs, the extraction
step was repeated twice in order to produce sufficient volume
for the validation experiments. The chip based sample
Biomed Microdevices (2016) 18: 44
Page 11 of 14 44
Fig. 4 a Bar chart comparing the
performance in terms of the Ct
values of sample preparation chip
used in the different panel
extraction assays, to that of the
validation platform (EasyMAG™
extraction platform); b results of
chip extraction of the panel 5 coinfection samples and comparison
to the validation platform
preparation step required ~2.5 h for completion, from sample
loading to the elution of the purified RNA extracts.
The remaining 5 μL extract volume from each sample
was mixed serially using EW droplet actuation (Fig. 2a)
with an off-chip prepared RT-PCR mastermix, consisting
of RT-PCR primers and probes required for the amplification. The resultant four 10 μL PCR mixture droplets
were then subjected to 38 RT-PCR amplification cycles
and the fluorescent signal from each of the four PCR
droplets was recorded during the annealing phase of each
PCR cycle by using the scanning micro-photomultiplier tube
sensor.
In order to demonstrate the versatility of viral targets, viral
loads and reporter dyes that can be used, we investigated different panels of clinical samples, shown in Table 1b, which
contained different initial concentrations of Influenza A,
Influenza B and enterovirus (E18 and CVA6) RNA. The molecular probes used for the RT-PCR detection of the different RNA
targets were labelled with FAM™ (λex./λem.: 492 nm/520 nm),
VIC™ (λex./λem.: 538 nm/554 nm) and NED™ (λex./λem.:
546 nm/576 nm) fluorophores respectively.
The entire RT-PCR amplification assay (RT reaction and 38
PCR cycles) was completed within 45 min following the chip
based extraction step and mixing/loading of the PCR
mastermix to the acquisition of the RT-PCR amplification
curves (and the corresponding Ct values).
The measured Ct data are reported in Fig. 5a and the
resulting qRT-PCR curves are shown in Fig. 5b. The RTPCR amplification curves and the corresponding Ct values
(Fig. 5b) show that the integrated chip based sample-to-
44 Page 12 of 14
Biomed Microdevices (2016) 18: 44
Fig. 5 a Bar chart showing the
relative performance of the
integrated sample-to-detection
NAT assay chip for the four
anonymized panel assays,
compared to that of validation
platforms (EasyMAG™ extractor
and ABI 7500 Fast RT-PCR
system); b Real-time PCR
amplification curves for each of
the four clinical panels used for
validation of the integrated
sample-to-detection NAT chip.
The chip RT-PCR assay data was
recorded in real-time and
subsequently normalized to
extract the Ct values
detection assays correlate very well with the routinely used
diagnostic assays.
All positive viral targets were successfully extracted
and amplified in each panel assay, with fairly high PCR
amplification efficiencies, ranging from 94 to 97 %. The
negative control samples from each panel did not show
any contamination from the adjacent sample preparation
and RT-PCR amplification channels. Also, the parallel
Biomed Microdevices (2016) 18: 44
sample preparation and RT-PCR amplification steps were
equally sensitive in the integrated NAT chip. The complete sample-to-detection assay on the chip was completed in less than 4 h. The entire NAT experiment, including
the off-chip mixing and off-board chemical lysis steps
(< 2 h), was completed within ~5 h. The chip based assay
completion time is comparable to the current commercial
sample-to-detection NAT platforms.
4.4 Overall performance of the nucleic acid sample
preparation chip and the integrated sample-to-detection
NAT chip
In this study, we tested over 50 samples and serial dilutions of MS2 bacteriophage. The patient samples were
anonymized into different panels of single and coinfection samples which were used to test the reliability,
reproducibility and the extraction efficiency over a wide
range of viral loads. Out of these, 37 samples were used
for the sample preparation chip experiments and 16 samples for the integrated NAT chip experiments. The samples comprised of 49 clinical samples and 4 serial dilutions of MS2 in UTM™. There were 12 negative controls
(all negative patient samples), 29 positive samples with
single but different viral targets and 8 positive samples
with viral co-infections. The sample preparation chip efficiently extracted different nucleic acid targets from complex patient sample matrices, over a broad range of viral
loads. The parallel arrangement of the DEP based nucleic
acid extraction electrodes in the four channel structure
was also shown to be beneficial for applications where
larger extract volumes are necessary for simultaneous testing of a range of pathogens. All samples were scored
correctly during the four panel tests using the integrated
NAT chip. Comparison of the chip performance to that of
the easyMAG™ extraction and the TaqMan® RT-PCR
platforms showed that the chip based NATs achieved a
100 % concordance (sensitivity and specificity) in the
detection of different viral targets, in a variety of patient
sample matrices. The efficiency of the RT-PCR validation
assays was analyzed using the method outlined in the
appendix 1. Comparison of the analyzed RT-PCR efficiency for the chip extracted nucleic acid sample and the
easyMAG™ extracted samples resulted in fairly comparable values for both validation experiments which furthermore indicates comparable high purity of the extracted
nucleic acid in the on-chip tested samples.
Our in-house influenza A and influenza B TaqMan® assays
have a reported limit of detection (LOD) of ~293 and 52.8
copies/reaction respectively (validation performed at
ProvLab). Based on the comparable sensitivity and the overall
performance of the integrated NAT chip (Fig. 5a), a similar
LOD is estimated for the chip NAT assay.
Page 13 of 14 44
5 Conclusion
In this study, we have developed a chip based sample preparation module for nucleic acid extraction in different specimen
matrices (NP swabs, blood plasma etc.), an essential preliminary step for NATs. We have validated chip extraction of
nucleic acid in the different clinical samples for different species of RNA viruses, including enveloped and non-enveloped
viruses as well as co-infection samples. To the best of our
knowledge, this is a first demonstration of integrating
particle-DEP with electro-actuation of droplets on a single
substrate to deliver a LOC micro-device for sample preparation (nucleic acid washing and purification). The performance
of the chip was compared to a standard platform and a ten-fold
reduction of reagent/buffer volumes was achieved on the chip,
with similar processing time (up to 2 h following off-board
lysis) while handling up to four different samples in parallel.
The extraction efficiency, measured using TaqMan® RT-PCR
platform for the extracted viral RNA and the extraction control
(MS2) has been found to be comparable and in several cases
perhaps slightly better than the commercial extractor.
We have furthermore demonstrated that by developing parallel sample preparation channels, one can achieve rapid, parallel nucleic acid extraction reactions, coupled with an on-chip
RT-PCR amplification module to realize a compact costeffective LOC NAT based diagnostic microsystem. The reported integrated NAT chip can accommodate up to four parallel, real-time RT-PCR reactions. The chip was tested using
four panels of patient samples where successful extraction and
accurate identification were carried out using the NAT chip.
The outcomes of the various panel experiments confirm that
the developed sample preparation and NAT chips can successfully handle multiple pathogens in different sample matrices,
in a parallel fashion that is highly suitable for syndromic panel
testing. The efficiency of chip based RT-PCR assays were
reasonably within the accepted range (Logan et al. 2009)
(PCR efficiency ~94 %) and the completion time for the sample loading, nucleic acid extraction, purification and RT-PCR
thermal cycling were also comparable to that achieved by
conventional sample-to-detection NAT platforms. The NAT
chip furthermore offers integration of both spatial (parallel
RT-PCR reactions with differed targets) and spectral (multiple
fluorophore markers in same qPCR reaction) multiplexing to
screen for larger panels of infectious agents. Throughout the
chip validation experiments, we have not found any crosscontamination between the parallel sample preparation and
RT-PCR channels. With the large viral load variation amongst
the different clinical panels, we have demonstrated a wide
dynamic range for the chip assay, from very low (~10
copies/reaction) to high viral loads (~105 copies/reaction) in
the tested samples. Unlike other illustrations of LOC sampleto-detection systems, which rely upon micro-machined setups to house the different sample preparation and RT-PCR
44 Page 14 of 14
amplification modules (Dineva et al. 2007; Easley et al. 2006)
we have designed our LOC system entirely as a single chip
module, interfaced with electrical/electronic components for
all the electro-actuation requirements.
Acknowledgments The authors acknowledge the financial support
from Natural Sciences and Engineering Research Council of Canada
(NSERC) for the research presented in this paper. We furthermore
acknowledge the support from CMC Microsystems for the fabrication
of the microchips used in this work.
References
R. Boom, C. J. Sol, M. M. Salimans, C. L. Jansen, P. M. W. van Dillen, J.
van der Noordaa, J. Clin. Microbiol. 28(3), 495–503 (1990)
M. A. Burns et al., Science 282(5388), 484–487 (1998)
Y. H. Chang, G. B. Lee, F. C. Huang, Y. Y. Chen, J. L. Lin, Biomed.
Microdevices 8(3), 215–225 (2006)
K. Choi, A. H. Ng, R. Fobel, A. R. Wheeler, Annual review of anal.
Chemistry 5, 413–440 (2012)
C. de la Rosa, P. A. Tilley, J. D. Fox, K. V. I. S. Kaler, IEEE Trans.
Biomed. Eng. 55(10), 2426–2432 (2008)
M. A. Dineva, L. M. Tapay, H. Lee, Analyst 132, 1193–1199 (2007)
J. Dreier, M. Stormer, K. Kleesiek, J. Clin. Microbiol. 43, 4551–4557
(2005)
C. J. Easley et al., Proc. Natl. Acad. Sci. U. S. A. 103, 19272 (2006)
R. B. Fair, A. Khlystov, T. D. Tailor, V. Ivanov, R. D. Evans, P. B. Griffin,
V. Srinivasan, V. K. Pamula, M. G. Pollock, J. Zhou, IEEE Des.
Com. 24, 10–24 (2007)
M. A. M. Gijs, Microfluid. Nanofluid. 1, 22–40 (2004)
T. Honegger, K. Berton, E. Picard, D. Peyrade, Appl. Phys. Lett. 98,
181906 (2011)
M. K. Hourfar, U. Michelsen, M. Schmidt, A. Berger, E. Seifried, W. Kurt
Roth, Clin. Chem. 51(7), 1217–1222 (2005)
T. B. Jones, J. Electrost. 51–52, 290–299 (2001)
K. V. I. S. Kaler, R. Prakash, Sensors 14(12), 23283–23306 (2014)
K.V.I.S. Kaler, R. Prakash, D. Chugh. Biomicrofluidics. 4(2), (022805)
1–17 (2010).
T.-H. Kim, J. Park, C.-J. Kim, Y.-K. Cho, Anal. Chem. 86(8), 3841–3848
(2014)
M. R. King, O. A. Lomakin, R. Ahmed, T. B. Jones, J. Appl. Physiol.
97(1–7), 054902 (2005)
M. U. Kopp, A. J. de Mello, A. Manz, Science 280, 1046–1048 (1998)
E. T. Lagally, C. A. Emrich, R. A. Mathies, Lab Chip 1-2, 102–107
(2001)
J.-G. Lee, K. H. Cheong, N. Huh, S. Kim, J.-W. Choi, C. Ko, Lab Chip 6,
886–895 (2006)
R. H. Liu, J. Yang, R. Lenigk, J. Bonanno, P. Grodzinski, J. Anal. Chem.
76(7), 1824–1831 (2004)
J. Logan, K.J. Edwards, N. Saunders, Real-Time PCR: Current
Technology and Applications. in: Applied and Functional
Genomics, Health Protection Agency, London, UK (2009)
A. Manz, N. Graber, H. M. Widmer, Sensors Actuators B1, 244–248
(1990)
Biomed Microdevices (2016) 18: 44
P. Markoulatos, N. Siafakas, M. Moncany, J. Clin. Lab. Anal. 16(1), 47–
51 (2002)
R. A. Mathies, E. T. Lagally, J. Phys. D. Appl. Phys. 37, R245–R261
(2004)
H. Moon, S. K. Cho, R. L. Garrell, C. J. Kim, J. Appl. Phys. 92(7), 4080
(2002)
F. Mugele, J. C. Baret, J. Phys. Condens. Matter 17, R705–R774 (2005)
K. Pabbaraju, S. Wong, B. Lee, R. Tellier, K. Fonseca, M. Louie, S. J.
Drews, Influenza Other Respir. Viruses 5(2), 99–103 (2011)
K. Pabbaraju, S. Wong, A. Wong, R. Tellier, Mol. Cell. Probes 29, 81–85
(2015)
S. Park, Y. Zhang, T. H. Wang, S. Yang, Lab Chip 11, 2893–2900 (2011)
H. A. Pohl, J. Appl. Phys. 29(8), 1182–1188 (1958)
H. A. Pohl, Dielectrophoresis: The Behavior of Neutral Matter in
Nonuniform Electric Fields (Cambridge University Press,
Cambridge, 1978)
M. G. Pollack, R. B. Fair, A. D. Shenderov, Appl. Phys. Lett. 77, 1725–
1726 (2000)
R. Prakash, K. V. I. S. Kaler, Sensors Actuators B Chem. 169, 274–283
(2012)
R. Prakash, R. Paul, K. V. I. S. Kaler, Lab Chip 10, 3094–3102 (2010)
R. Prakash, K. V. I. S. Kaler, D. P. Papageorgiou, A. G. Papathanasiou,
Microfluid. Nanofluid. 13, 309–318 (2012)
R. Prakash, D. P. Papageorgiou, A. G. Papathanasiou, K. V. I. S. Kaler,
Sensors Actuators B Chem. 182, 351–361 (2013)
R. Prakash, K. Pabbaraju, S. Wong, A. Wong, R. Tellier, K. V. I. S. Kaler,
J. Electrochem. Soc. 161(2), 3083–3093 (2014)
R. Prakash, K. Pabbaraju, S. Wong, A. Wong, R. Tellier, K. V. I. S. Kaler,
Micromachines 6, 63–79 (2015)
C. W. Price, D. C. Leslie, J. P. Landers, Lab Chip 9, 2484–2494 (2009)
D. H. Saiki, S. Gelfand, S. Stoffel, S. J. Scharf, R. Higuchi, G. T. Horn, K.
B. Mullis, H. A. Erlich, Science 239, 487–491 (1988)
D. C. Saunders, G. L. Holst, C. R. Phaneuf, N. Pak, M. Marchese, N.
Sondej, M. McKinnon, C. R. Forest, Biosens. Bioelectron. 44, 222–
228 (2013)
T. Schnelle, T. Moiler, S. Fiedler, S. G. Shirley, K. Ludwig, A. Herrmann,
G. Fuhr, Naturwissenschaften 83, 172–176 (1996)
R. S. Sista, A. E. Eckhardt, V. Srinivasan, M. G. Pollack, S. Palanki, V. K.
Pamula, Lab Chip 8(12), 2091–2104 (2008)
S. Tanriverdi, L. Chen, S. Chen, J. Infect. Dis. 201(S1), 52–58 (2010)
L. J. R. Van Elden, A. M. van Loon, A. van der Beek, K. A. W.
Hendriksen, A. I. M. Hoepelman, M. G. J. van Kraaij, P. Schipper,
M. Nijhuis, J. Clin. Microbiol. 41(9), 4378–4381 (2003)
N. Vergauwe et al., Sensors Actuators B Chem. 196, 282–291 (2014)
X. B. Wang, J. Vykoukal, F. F. Becker, P. R. C. Gascoyne, J. Biophys. 74,
2689–2701 (1998)
M. Washizu, IEEE Trans. Ind. Appl. 34, 732–737 (1998)
A. R. Wheeler, Science 322, 539–540 (2008)
P. A. L. Wijethunga, Y. S. Nanayakkara, P. Kunchala, D. W. Armstrong,
H. Moon, Anal. Chem. 83, 1658–1664 (2011)
M. Wiklund, C. Gunther, R. Lemor, M. Jager, G. Fuhr, H. M. Hertz, Lab
Chip 6, 1537–1544 (2006)
J. Wu, R. Kodzius, W. Cao, W. Wen, Microchim. Acta 181(13), 1611–
1631 (2014)
Y. Zhao, U. C. Yi, S. K. Cho, J. Microelectromech. Syst. 16, 1472–1481
(2007)