3188 Electrophoresis 2011, 32, 3188–3195

Jane Woods1,2 Research Article

Peter T. Docker1
Charlotte E. Dyer2
Stephen J. Haswell1 On-chip integrated labelling, transport
John Greenman2
and detection of tumour cells
1
Department of Chemistry,
University of Hull, Hull, UK Microflow cytometry represents a promising tool for the investigation of diagnostic and
2
Postgraduate Medical Institute, prognostic cellular cancer markers, particularly if integrated within a device that allows
University of Hull, Hull, UK
primary cells to be freshly isolated from the solid tumour biopsies that more accurately
reflect patient-specific in vivo tissue microenvironments at the time of staining. However,
Received March 16, 2011 current tissue processing techniques involve several sequential stages with concomitant
Revised July 28, 2011 cell losses, and as such are inappropriate for use with small biopsies. Accordingly, we
Accepted August 23, 2011 present a simple method for combined antibody-labelling and dissociation of hetero-
geneous cells from a tumour mass, which reduces the number of processing steps.
Perfusion of ex vivo tissue at 41C with antibodies and enzymes slows cellular activity
while allowing sufficient time for the diffusion of minimally active enzymes. In situ
antibody-labelled cells are then dissociated at 371C from the tumour mass, whereupon
hydrogel-filled channels allow the release of relatively low cell numbers (o1000) into a
biomimetic microenvironment. This novel approach to sample processing is then further
integrated with hydrogel-based electrokinetic transport of the freshly liberated fluor-
escent cells for downstream detection. It is anticipated that this integrated microfluidic
methodology will have wide-ranging biomedical and clinical applications.

Keywords:
Electrokinetic / Head and neck squamous cell carcinoma / Hydrogel / Lab on a
chip / Microflow cytometry DOI 10.1002/elps.201100172

1 Introduction largely restricted to the analysis of cultured cell lines selec-

ted to represent the tissue of origin. However, the risk of
Flow cytometry is an invaluable biomedical tool for research phenotypic alteration brings interpretive problems, i.e.
into the cellular mechanisms of cancer malignancy [1]. insights gained through cell line work cannot be confidently
Microflow cytometry, an emerging technology in cancer extrapolated to explain in vivo tumour behaviour. Even 3-D
research, has advantages over conventional flow cytometry; co-cultures or artificial tissue constructs do not fully repre-
these include compactness, reduced sample size require- sent tissue complexity, in which multiple cell types
ments, prevention of cross-contamination by containment continuously interact with one another and with the extra-
of biohazardous samples in disposable microfluidic devices cellular matrix (ECM [4]). Furthermore, the tumour ECM
and possibilities for integrated analyses [2]. The first and stroma are known to exhibit patient-to-patient hetero-
instance of microflow cytometry on patient-derived solid geneity even among tumours of the same histopathological
tumour cells was reported over four decades ago, where type, and this has a significant influence upon metastatic
fixed cells were detected and sorted according to the size and potential and thus upon patient prognosis [5].
nucleic acid content by differences in light absorption and Consequently, fresh biopsies comprising patient-speci-
scattering properties [3]. fic tumour cells and stromal cells within their native ECM
Despite the inherent promise of this early success, milieu constitute more physiologically relevant samples for
microflow (and flow) cytometry for cancer research remains the exploration of solid tumour biology at the cellular level
[6, 7]. It is postulated that cells directly liberated from this
microenvironment would more reliably retain their pheno-
Correspondence: Professor John Greenman, Centre for Biome- types; however, the labelling of cell-membrane markers
dical Research, Postgraduate Medical Institute, University of entails several sequential steps – tissue disaggregation,
Hull, Cottingham Road, Hull, HU6 7RX, UK
incubation of dissociated cells with fluorescently conjugated
E-mail: [email protected]
Fax: 144-1482-466996 antibody, and removal of unbound antibody – each neces-
sitating cell washing and centrifugation [8]. Inevitable cell
Abbreviations: ECM, extracellular matrix; EPF, losses during these procedures render flow cytometry
electrophoretic flow; HNSCC, head and neck squamous cell
carcinoma Colour online: See the article online to view Figs. 3, 4 and 6 in colour.

& 2011 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.electrophoresis-journal.com
Electrophoresis 2011, 32, 3188–3195 Microfluidics and Miniaturization 3189

impractical for small biopsy samples [9]. Microflow cyto-

metry meanwhile has the capability to analyse small
numbers of cells; nonetheless, there is an acknowledged
need for simplified, robust, integration of pre-analytical
sample processing if such a technology is to be widely
adopted by biomedical researchers [10–12]. Progress in this
area has been demonstrated, for example, by antibody
labelling of primary human blood cells in Lab-on-a-chip (or
micro Total Analysis Systems); accordingly, the ability to
label solid tumour cells on microfluidic-based devices would
represent a further significant step [13].
Other advantageous features of a microflow cytometer
for biomedical research include a reduced footprint and
fewer moving parts; advantages afforded by utilising an
electrokinetic (EK) system for cell movement that can be
rapidly switched for accurate cell sorting. Early work on EK
cytometry explored the manipulation of unlabelled bacteria,
yeasts and red blood cells, employing combinations of
microchannel layouts and voltage switching [14]. Since then,
applications have extended to encompass the labelling of
bacteria on-chip prior to sorting [15] but for tumour cell
manipulation the greater convenience of EK has yet to be
exploited.
We present here an innovative and integrated approach
to pre-analytical tumour tissue disaggregation, single cell
isolation and antibody cell labelling that represents a
departure from conventional protocols such as FACS which
is designed for large cell numbers and high throughput [16].
This methodology allows in situ labelling of cells within a
tumour mass during interactions with neighbouring cells
and with the tumour-specific ECM, prior to cell dissociation.
Moreover, this pre-analytical sample processing is then
integrated with EK transport and detection of relatively
small numbers of freshly liberated tumour cells, as illu-
strated in Fig. 1A which can be used for subsequent
analysis, e.g. polymerase chain reaction (PCR) measure-
ment of changes in gene transcription.

Figure 1. (A) Schematic of integrated microfluidic processes

from biopsy to detection of antibody-labelled tumour cells.
2 Materials and methods Labelling is achieved during in situ interactions with native ECM
and neighbouring cells. Following controlled release, labelled
2.1 Microfluidic devices tissue cells are transported electrokinetically in small numbers
for detection and imaging. (B) HNSCC tissue (E1 mm3) placed
upon cell strainer in tissue chamber, submerged in enzyme/
Glass microfluidic devices were fabricated using standard antibody mixture. At 41C, solid hydrogel acts to contain fluid in
photolithography and wet etching techniques to produce the reservoir. (C) A combination of melting hydrogel and hydrostatic
design shown in Fig. 2. Channels 70 mm deep  150 mm pressure allows in situ antibody-labelled, dissociated cells to
wide were etched in 1 mm glass. A 3-mm diameter tissue pass through the cell strainer and hydrogel macropores in the
microchannel.
chamber and 1.5 mm holes for sample, reagent and
electrode access were drilled in a 3-mm glass top plate
before thermally bonding it to the etched base plate to form
the device. Prior to each use, devices were sterilised by filtered prior to use through a 0.22-mm syringe filter
flushing with ethanol and autoclaving, and fresh reservoirs (Millipore, UK). After filling microchannels with hydrogel
glued around the access ports using epoxy resin (Perma- (formulated as described below), excess hydrogel was
bond, UK). Lids reduced evaporation and housed 0.5 mm removed from the tissue chamber, and to prevent tissue
diameter Pt electrodes. Channels were flushed with debris from entering and clogging the channels, a 50-mm
deionised H2O, then 0.1 M NaOH for 5 min, followed by mesh cell strainer (340632, BD Falcon, UK) was secured
deionised H2O and then finally media. All fluids were above the main channel as shown in Fig. 2C.

& 2011 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.electrophoresis-journal.com
3190 J. Woods et al. Electrophoresis 2011, 32, 3188–3195

Figure 2. (A) Scale drawing of microfluidic

device showing tissue chamber above main
channel, and access ports A–D above
channel ends; (B) photograph of device
with reservoirs in place; (C) photograph
taken on inverted microscope showing
50 mm mesh cell strainer secured in tissue
chamber E100 mm above main channel.

2.2 Preparation of hydrogel for tumour cell electro- (iii) Biocompatible extracellular pH is 6.8–7.8 [25].
kinesis Adequate pH buffering is especially important in
EK, as electrolysis and other interactions can alter the
As fluid flow in tissues occurs within a gel-like matrix local pH, reduce the zeta potential and even reverse
[17, 18], low melting point (LMP) agarose hydrogel (A9419, the direction of flow [26–29]. A 0.025 M HEPES buffer
Sigma-Aldrich, UK) was used in the microchannels was selected as optimal for this application [30, 31].
to provide a biometic microenvironment and also to
suppress the hydrodynamic back pressure during EK A PBS-based hydrogel was also used in all the
transport. 4-(2-Hydroxyethyl)-1-piperazineethanesulfonic comparative EK mobility experiments.
acid (HEPES)/sucrose mixture was first prepared by
dissolving acid and base components of HEPES in deionised
water giving a 0.1 M stock solution, pH 7.4, at 251C (http:// 2.3 Tissue and cell samples
www.liv.ac.uk/buffers/buffercalc.html). This was diluted 1:1
with deionised water and then mixed with 0.50 M sucrose Tumour samples from patients with head and neck squamous
1:1 to give 0.025 M HEPES/0.25 M sucrose (final concentra- cell carcinoma (HNSCC) were selected as representative of
tions). Hydrogel was then prepared by dissolving 0.25 or small clinical biopsies typically consisting of tiny pieces of
0.50% w/v LMP agarose in the above HEPES/sucrose tissue E0.1–3.0 g in mass. Anonymised samples from
buffer, which was specifically formulated to meet the HNSCC lesions were obtained from patients undergoing
following criteria: surgery, having gained ethical approval from Hull and East
Yorkshire Research Ethics Committee (07/H1304) and the
(i) EK linear velocity is proportional to zeta potential z and Hull and East Yorkshire Hospitals NHS Trust (RO568).
field strength. Applied potentials should be kept at Samples were stored in Dulbecco’s modified Eagle’s medium
minimal levels to prevent high currents, Joule heating (DMEM; E15-009, PAA Laboratories, UK) supplemented with
and cell lysis [19]. To maximise bulk flow and cell 10% v/v foetal calf serum (FCS; S1900, Biosera, UK) and
viability at a given voltage, it is desirable to maximise z antibiotic solution [10 000 U/mL penicillin and 10 mg/mL
which is affected by pH and inversely proportional to streptomycin in 0.9% w/v NaCl; P11-010, Sigma, UK], at 41C,
ionic strength [20–22]. Well-buffered fluids at low ionic and if not used immediately, snap frozen and cryopreserved in
strength are thus essential for EK at low voltages in liquid nitrogen at 1961C.
microfluidic cellular applications. K562 human erythroleukaemic cells were selected for
(ii) Biocompatible osmolality for human tumour cells is proof of principle in EK studies, as it was experimentally
E297 mOsm/kg [23]. To prevent artefacts arising from easier to work with cell lines than fresh cells derived from a
hypo- or hyper-tonic shock, osmolality in vitro is tumour. Cells were cultured in Roswell Park Memorial
commonly controlled using salts, e.g. phosphate- Institute medium (RPMI 1640; E15-840, PAA Laboratories)
buffered saline (PBS). The requirement for low ionic supplemented as above and harvested by centrifugation at
strength precludes the use of salts, so an isotonic 400  g. Cell and tissue preparation was carried out in a
sucrose solution was used in place of PBS [24]. class II biological safety cabinet.

& 2011 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.electrophoresis-journal.com
Electrophoresis 2011, 32, 3188–3195 Microfluidics and Miniaturization 3191

2.4 Antibodies cells had been established, the optimal field strength
required for controlled cell manipulation enabling subse-
Integrin subunits b1 and a6 were selected as ubiquitously quent cell sorting (10–100 V/cm) was determined using
expressed cell membrane targets that are reported to be K562 cells. After functionalising the device by rinsing with
upregulated in metastatic HNSCC cells [32, 33]. Monoclonal NaOH and buffer as described, channels were filled with
antibodies conjugated with red or green fluorophores were hydrogel made with either PBS or HEPES/sucrose buffer,
used: anti-CD29 (integrin b1) PerCPe-Fluors710 (46-0299, and containing either labelled or unlabelled cells suspended
eBioscience, UK) and anti-CD49f (integrin a6) Alexa at E0.5–1.0  106 mL 1. The device was placed on the stage
Fluors488 (313608, Cambridge Bioscience Ltd, UK), each of an inverted fluorescence microscope, and a voltage
at 5 mL per mm3 of tissue (E400 000 cells). applied between access ports A and B (Fig. 2A) using
0.5 mm Pt electrodes connected to a 1-kV DC power supply
(Kingfield Electronics, UK). Fluorescent image sequences
2.5 Integrated antibody labelling and cell release were analysed in ImageJ to derive apparent linear velocity
(vapp) which was converted to apparent mobility (mapp);
A 1 mm3 piece of HNSCC biopsy tissue was placed in the mapp 5 mepf1meof. Cell diameters were measured in ImageJ
tissue reservoir upon the cell strainer and submerged in and plotted against corresponding mapp.
200 mL of enzyme solution. Antibodies as described above Having established cell flow characteristics using the
were added to the collagenase solution (see below) after 3 h, K562 cells, the integrated device was then used to study
and the tissue incubated at 41C for a further 5–21 h as freshly dissociated antibody-labelled tumour cells
shown in Fig. 1B. The reduction in temperature and from a tissue biopsy. Cells were initially observed entering
extended time frame, compared with conventional metho- the channel under gravitational and hydrodynamic
dology, was designed to minimise enzyme and cellular forces, and when this cell movement had subsided,
metabolic activity while permitting sufficient time for the cells were then moved through the microchannel
diffusion of reagents to occur into centre of the tissue. to the detection window by EK with a field strength of
In preliminary experiments, cells were retrieved at 70 V/cm. Again, mapp was derived from vapp using ImageJ
various stages for imaging, to assess the effectiveness of the analysis.
labelling and dissociation method, to optimise the duration
of the cold perfusion required for saturation of tissue with
enzymes and antibodies, and to check whether the enzymes 3 Results
affect antibody labelling. Reservoir contents were gently
pushed through a 50 mm mesh cell strainer using a syringe 3.1 Integrated antibody labelling and cell release
plunger, centrifuged at 400  g for 3 min; 10 mL from the
reservoir was observed on a microscope slide by inverted Antibody labelling and dissociation of cells from tumorous
fluorescence microscopy (Axiovert S100, Carl Zeiss, UK) and normal (peripheral) regions of HNSCC biopsies was
using a 20  objective lens. Identical excitation filters of undertaken both ‘off-chip’ and ‘on-chip’ on a series of
band pass 470/40 nm allowed excitation at 488 nm. To different biopsies, testing different incubation periods
collect the red signal, a 660-nm beam splitter and emission of the enzyme and antibody mixture. An example of an
filter band pass 690/50 nm excluded green fluorescence. To ‘off-chip’ experiment is shown in Fig. 3. It was found that
collect the green signal, a 495 nm beam splitter and emis- incubation periods 420 h were required to elicit relatively
sion filter band pass 540/50 excluded red fluorescence. small clumps of cells and sufficient single cells for
Dual-labelled cells were separately imaged by monochrome downstream analysis; tumour biopsies tended to yield more
CCD camera [Orca ER, Hamamatsu Photonics UK] for single cells than the peripheral tissue although as expected
analysis in ImageJ software [34]. there were difference in the dissociation reflecting tumour
After an allotted time span for incubation in enzyme/ heterogeneity. Greater incubation times were not pursued
antibody mixture, the solution was replaced with HEPES/ as the aim was to label and analyse as contemporaneously as
sucrose buffer and cell release from the tumour mass was possible. This methodology enables minimally disruptive in
instigated by incubating the device at 371C for 30–60 min. situ antibody labelling of cell membrane targets and
The device was then placed upon the microscope stage for moreover is able to reflect the dynamic behaviours of ex
observation of tissue cells passing through the cell strainer, vivo human cancer cells within both normal and malignant
under gravity and hydrodynamic pressure, and into the functional tissues.
microchannel. The rationale behind the cold perfusion approach for
antibody labelling and tissue disaggregation is the total
suffusion of the tissue ECM with minimally active collage-
2.6 EK cell transport nase, and simultaneous saturation of antigen binding sites
with antibody. In contrast to conventional methods, this
Once the optimal conditions for effective in situ labelling of relatively prolonged incubation at 41C preserves the tissue,
cell-membrane integrins and the dissociation of labelled antibodies and enzymes, and slows cellular activity; while

& 2011 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.electrophoresis-journal.com
3192 J. Woods et al. Electrophoresis 2011, 32, 3188–3195

Figure 3. Fluorescent pseudo-coloured micrographs of peripheral cells (A–C) and tumour cells (D–F) from a patient with HNSCC. Cells
were dual-labelled and dissociated ‘off-chip’ by the combination method. Antibodies to two integrin subunits were used: (A and D) anti-
CD49f (a6) conjugated with Alexa Fluors488 (green), (B and E) anti-CD29 (b1) conjugated with PerCPe-Fluors710 (red). Monochrome
images were combined to show co-localisation of a6 and b1 (C and F). Pseudopodia are just apparent upon the tumour cells, with
clusters of b1 at the leading edges. Results are representative of three separate experiments.

allowing sufficient time for diffusion to the centre of the collagenase activity. Since this may nonetheless go some
tissue. The diffusion coefficient D in free solution at 201C is way toward preconditioning the tissue to facilitate entry of
dependent upon both medium and solute, and inversely IgG1, a 3-h delay was incorporated before adding the anti-
proportional to solute molecular weight. Diffusion into body. Although 24 h incubation was used for the following
tissue, however, is more complex than in free solution, as proof of concept studies, shorter incubation time periods
extracellular space is non-homogeneous, and tissue porosity also gave effective antibody labelling but less efficient
affects diffusion rate [35]. Tumour tissue is often composed dissociation.
of very densely packed cells, in the region of 2–4  105 cells/ After incubation of tissue biopsies ‘on-chip’ with the
mL (mm 3 [36]). D for IgG1 and collagenase in tumour antibodies and enzymes, the excess mixture was replaced
tissue is typically E1  10 11 m2/s, or 10 mm2/s [36, 37] with HEPES/sucrose buffer. Incubation at 371C then
compared with 4.2  10 11 and 4.5  10 11 m2/s for IgG1 increased enzymatic activity within the tissue, cleaving the
and collagenase in free solution at 201C [38]. For a 1-mm3 ECM and releasing labelled cells. Concurrently, as the
piece of tumour tissue, diffusion distance to the centre is temperature was raised from 4 to 371C, the hydrogel
5  10 4 m, giving an approximate diffusion time of underwent partial melting without reaching the total fluid
E12 500 s or 3.5 h at 201C. However, diffusion is also phase (501C). Heating the microdevice to 371C thus realised
temperature dependent, resulting in significantly longer two desired outcomes – dissociation of labelled cells from
time scales at 41C [39]. the tissue, and flow of cells vertically through the strainer
In the case of antibody diffusion, as the antibody front and laterally into the hydrogel-filled channel by gravitational
progresses, the effective diffusion rate is influenced not only and hydrodynamic forces, as shown in Fig. 1C. The 0.25%
by the above factors but also by antibody–antigen binding hydrogel was found to be superior to the 0.5% hydrogel in
and is thus inversely proportional to antigen density [40]. terms of allowing cells to cross from a fluidic medium into
Pre-treatment with collagenases has been shown previously the hydrogel macropores, whereupon it was found to
to enhance the diffusion of large molecules into in vivo adequately control the hydrodynamic back flow and stabilise
tissue [40]. Accordingly in this current work, collagenase – cell motion in the microchannel. Eight independent
whose activity at 41C is much reduced [41] – was incubated experiments were undertaken to test the labelling and
for 24 h to allow diffusion into the tissue with only residual dissociation aspects of the device.

& 2011 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.electrophoresis-journal.com
Electrophoresis 2011, 32, 3188–3195 Microfluidics and Miniaturization 3193

3.2 EK cell transport

In preparation for transporting dissociated cancer cells, the

interplay between electroosmotic flow (EOF) and electro-
phoretic flow (EPF) in hydrogel-based EK was explored.
Figure 4A shows a scatter plot for mapp versus cell diameter
for unlabelled K562 cells in PBS-based hydrogel in a field of
70 V/cm. Figure 4B shows the experimental set-up in which
a range of EK field strengths were studied. The levels below
50 V/cm did not reproducibly cause movement with all
cell types studied (data not shown). The levels in excess of
70 V/cm, although causing more rapid movement,
commonly induced extensive cell lysis, particularly in the
freshly isolated tumour cells. The data show a positive
correlation between cell diameter and EK mobility indicat-
ing that in this application, mapp is generally higher for larger
cells than for smaller ones. These results suggest that cells
can be discriminated on the basis of size alone, a factor that
could be useful in future applications of this technology.
Figure 5 shows mapp for fluorescent antibody-labelled
Figure 4. (A) Scatter plot of EK mobility (mapp) versus cell
versus unlabelled cells in HEPES/sucrose hydrogel.
diameter for unlabelled K562 cells in PBS-based hydrogel. Each
This result indicates that mapp is much higher for point represents an individual cell, data shown are representa-
unlabelled cells (42.0  10 8 m2/Vs) than for labelled cells tive of over 150 cells. Mean vappE9.0  10 5 m/s. Mean
(o1.0  10 8 m2/Vs ). These variations in overall EK mappE1.34  10 8 m2/Vs; (B) K562 cells moving by EK in a
mobilities for cells of varying size and for labelled and hydrogel-filled microchannel. All observed cell velocities are
toward the anode.
unlabelled cells reveal the part played by EPF. The nega-
tively charged membrane attracts cations to form an electric
double layer and cellular zeta potential zc. Simultaneous to
cell movement within the EOF bulk flow, cell movement
relative to the fluid occurs by EPF, with linear velocity vepf
dependent upon charge-to-size ratio. Smaller cells possess
higher charge-to-size ratios and thus experience greater
cathodic EPF. In the case of labelled cells, antibody charge is
pH dependent and governed by the isoelectric point. For
monoclonal IgG1, the isoelectric point is highly hetero-
geneous but lies in the range 7.0–9.0, and so IgG1 has a net
positive charge at pH 7.4 [42]. The overall charge on Figure 5. EK mobility (mapp) for antibody-labelled and unlabelled
conjugated antibodies should include fluorophore charge; cells in HEPES/sucrose hydrogel. Error bars are SE for four
however, this information is not readily available. However, repeats.
what is known is that cells with bound conjugated antibody
have increased zc and greater charge-to-size ratios than 0.25% hydrogel-filled channel, just prior to application of an
unlabelled cells [43]. Labelled cells are consequently more electric field. The fluorescent signal is sufficiently bright to
electrophoretically mobile, a phenomenon used for the be easily detectable, showing that antibody remains bound
characterisation of lymphocyte phenotypes, zeta potential to cells while in motion within the hydrogel, and that the
calculation and immunoelectrophoresis of erythrocytes hydrogel is a suitably transparent medium. In addition, the
[43, 44]. levels of unbound antibody in the hydrogel are low in
If mepf and meof are in the same direction, mapp is contrast with antibody bound to the cells.
increased; conversely, if they are in opposite directions mapp Figure 6B shows a plot of intensity versus time for the
is decreased. As a consequence of microchannel fouling, same cell (representative of many cells) traversing a region
due to the deposition and adsorption of proteins and other of interest (ROI; marked on Fig. 6C) upon the application of
cellular debris, it is known that normally negative channel 45 V/cm. The fluorescence peak for the cell is distinct from
surfaces become positively charged resulting in reverse that of the background, indicating that washing off excess
EOF, i.e. bulk flow is towards the anode [45, 46]. Following antibody is un-necessary, and also that the selected fluor-
release into the microchannel as described above, antibody- ophores are sufficiently bright and photostable to give
labelled dissociated tumour cells were transported by prolonged, detectable signals from moving cells.
hydrogel-based EK for detection by fluorescence microscopy. The sequence in Fig. 6C shows cells moving by hydro-
Figure 6A shows a stationary integrin b1-labelled cell in the gel-based EK, with dominant anodic EOF. The overall

& 2011 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.electrophoresis-journal.com
3194 J. Woods et al. Electrophoresis 2011, 32, 3188–3195

Figure 6. Electrokinesis of antibody-

labelled and dissociated HNSCC tumour
cells in 0.25% hydrogel-filled microchannel:
(A) stationary b1-labelled cell immediately
prior to application of electric field; 45 V/cm
applied at 14 s; (B) plot of intensity versus
time for same cell traversing ROI marked on
(C). [(B) t 5 0–4 s corresponds approxi-
mately to (C), t 5 14–18 s]; (C) fluorescent
micrograph sequence of EK cell transport.
vappE5.9  10 5 m/s; mappE1.3  10 8 m2/Vs.

mobility of these cells showed no difference from that directly from this vital and information-rich tumour
observed in initial experiments characterising the flow microenvironment provide unique data on individual
system with the K562 cell line. Agarose hydrogel is patient tumours. Finally, the use of EOF and EPF to
mechanically similar to ECM and supports the cells in transport antibody-labelled and unlabelled heterogeneous
motion, preventing settling out, while permitting continued tumour cells suggests that through careful selection of field
paracrine interactions via soluble signalling factors strength and flow direction, some degree of cell separation
[18, 47, 48]. may also be achievable. Future studies using multiple
fluorescently labelled antibodies to validate the detection of
known diagnostic/prognostic markers on tumour samples
4 Discussion in the microfluidic device are planned, e.g. detection of
epidermal growth factor receptor on HNSCC, in parallel
This study has demonstrated the feasibility of using ex vivo with technological developments of the integrated micro-
human tumour tissue as a source of primary cells for EK fluidic device.
microflow cytometry. Furthermore, it has enabled the
integration of solid tumour cell dissociation, antibody The authors thank Dr. S. Clark, Dr. S. Hattersley,
labelling, cell transportation and fluorescence-based cell L. Houghton, M. Park, Dr. K. Shaw, M. Tarn, and Dr. X.
detection to be achieved on a single microfluidic platform. Zhang for technical assistance, Professor N. Stafford for HNSCC
The demonstration that these procedures can be success- tumour samples and the BBSRC (BB/E002722) and the
fully assimilated will lead to future devices containing these University of Hull for financial support.
processes in portable units as the existing components can
all be miniaturised without loss of function or sensitivity [2]. The authors have declared no conflict of interest.
The innovative tissue processing approach reduces cell
handling, simplifies the pre-analytical stage of microflow
cytometry and thus minimises the potential for generating
experimental artefacts. In addition, cell labelling can be 5 References
achieved alongside complex dynamic signalling and adhe-
sion events with neighbouring cells and with the ECM. The [1] Bonetta, L., Nat. Methods 2005, 2, 785–794.
work demonstrates the value of microfluidics as a tool for [2] Chung, T. D., Kim, H. C., Electrophoresis 2007, 28,
solid tumour research at the cellular level. Cells released 4511–4520.

& 2011 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.electrophoresis-journal.com
Electrophoresis 2011, 32, 3188–3195 Microfluidics and Miniaturization 3195

[3] Kamentsky, L. A., Melamed, M. R., Science 1967, 156, [25] Wu, M. H., Lin, J. L., Wang, J., Cui, Z. F., Cui, Z., Biomed.
1364–1365. Microdevices 2009, 11, 265–273.
[4] Schwartz, M. P., Fairbanks, B. D., Rogers, R. E., [26] Persat, A., Suss, M. E., Santiago, J. G., Lab Chip 2009, 9,
Rangarajan, R., Zaman, M. H., Anseth, K. S., Integr. Biol. 2454–2469.
2010, 2, 32–40. [27] Hsieh, S. S., Lin, H. C., Lin, C. Y., Colloid Polym. Sci.
[5] Joyce, J. A., Pollard, J. W., Nat. Rev. Cancer 2009, 9, 2006, 284, 1275–1286.
239–252. [28] Minerick, A. R., Ostafin, A. E., Chang, H. C., Electro-
[6] Webster, A., Dyer, C. E., Haswell, S. J., Greenman, J., phoresis 2002, 23, 2165–2173.
Anal. Methods 2010, 2, 1005–1007. [29] Xuan, X., Li, D., J. Micromech. Microeng. 2004, 14,
[7] Hattersley, S. M., Dyer, C. E., Greenman, J., Haswell, 1171–1180.
S. J., Lab Chip 2008, 8, 1842–1846. [30] Good, N. E., Winget, G. D., Winter, W., Connolly, T. N.,
[8] Ailles, L., Prince, M., in: Yu, J. S. (Ed.), Cancer Stem Izawa, S., Singh, R. M. M., Biochemistry (NY) 1966, 5,
Cells: Methods and Protocols, Humana Press, London 467–477.
2009, pp. 175. [31] Beynon, R. J., Easterby, J. S., Buffer Solutions, IRL
[9] Fitzpatrick, E., McBride, S., Yavelow, J., Najmi, S., Press, Oxford 1996.
Zanzucchi, P., Wieder, R., Clin. Chem. 2006, 52, [32] Shinohara, M., Nakamura, S., Sasaki, M., Kurahara, S.,
1080–1088. Ikebe, T., Harada, T., Shirasuna, K., Am. J. Clin. Pathol.
[10] Ateya, D. A., Erickson, J. S., Howell, P. B., Jr., Hilliard, 1999, 111, 75–88.
L. R., Golden, J. P., Ligler, F. S., Anal. Bioanal. Chem. [33] Zhang, X., Liu, Y., Gilcrease, M. Z., Yuan, X. H.,
2008, 391, 1485–1498. Clayman, G. L., Adler-Storthz, K., Chen, Z., Cancer 2002,
[11] Sims, C. E., Allbritton, N. L., Lab Chip 2007, 7, 95, 1663–1672.
423–440. [34] Rasband, W. S., ImageJUSA, U.S. National Institutes of
[12] Whitesides, G. M., Woodford, L., Flowers, A. A., Lab Health, Bethesda, MD 1997–2009.
Chip 2010, 10, 2317–2318. [35] O’Sullivan, D. G., J. Theor. Biol. 1962, 2, 117–128.
[13] Lancaster, C., Kokoris, M., Nabavi, M., Clemmens, J., [36] Thurber, G. M., Schmidt, M. M., Wittrup, K. D., Adv.
Maloney, P., Capadanno, J., Gerdes, J., Battrell, C. F., Drug Deliv. Rev. 2008, 60, 1421–1434.
Methods 2005, 37, 120–127.
[37] Ackerman, M. E., Pawlowski, D., Wittrup, K. D., Mol.
[14] Li, P. C. H., Harrison, D. J., Anal. Chem. 1997, 69, Cancer Ther. 2008, 7, 2233–2240.
1564–1566.
[38] Freitas, R. A., Jr, Nanomedicine Volume 1: Basic
[15] Dittrich, P. S., Schwille, P., Anal. Chem. 2003, 75, Capabilities, Landes Bioscience, Georgetown 1999.
5767–5774.
[39] McCabe, M., Adam, K., Maguire, D., J. Theor. Biol. 1979,
[16] Wang, M. M., Tu, E., Raymond, D. E., Yang, J. M., 78, 51–59.
Zhang, H. C., Hagen, N., Dees, B., Mercer, E. M., Forster,
A. H., Kariv, I., Marchand, P. J., Butler, W. F., Nat. [40] Magzoub, M., Jin, S., Verkman, A. S., FASEB J. 2008, 22,
Biotechnol. 2005, 23, 83–87. 276–284.
[41] Dono, K., Gotoh, M., Monden, M., Kanai, T., Fukuzaki,
[17] Rutkowski, J. M., Swartz, M. A., Trends Cell Biol. 2007,
17, 44–50. T., Mori, T., Transplantation 1994, 57, 22–26.

[18] Ling, Y., Rubin, J., Deng, Y., Huang, C., Demirci, U., [42] Prin, C., Bene, M. C., Gobert, B., Montagne, P., Faure,
G. C., Biochim. Biophys. Acta 1995, 1243, 287–290.
Karp, J. M., Khademhosseini, A., Lab Chip 2007, 7,
756–762. [43] Wall, J., Ayoub, F., O’Shea, P., J. Cell Sci. 1995, 108,
2673–2682.
[19] Gabi, M., Sannomiya, T., Larmagnac, A., Puttaswamy,
M., Voros, J., Integr. Biol. 2009, 1, 108–115. [44] Ichiki, T., Ujiie, T., Shinbashi, S., Okuda, T., Horiike, Y.,
[20] Barz, D. P. J., Ehrhard, P., Lab Chip 2005, 5, 949–958. Electrophoresis 2002, 23, 2029–2034.

[21] Sparks, D. L., Environmental Soil Chemistry, Academic [45] Mukhopadhyay, R., Anal. Chem. 2005, 77, 429A–432A.
Press, San Diego 2003. [46] Kirby, B. J., Wheeler, A. R., Zare, R. N., Fruetel, J. A.,
Shepodd, T. J., Lab Chip 2003, 3, 5–10.
[22] Li, D., Electrokinetics in Microfluidics, Elsevier, London
2004. [47] Walker, G. M., Zeringue, H. C., Beebe, D. J., Lab Chip
2004, 4, 91–97.
[23] You, J., Aznavoorian, S., Liotta, L. A., Dong, C., J. Cell.
Physiol. 1996, 167, 156–163. [48] Domenech, M., Yu, H., Warrick, J., Badders, N. M.,
[24] Shier, W. T., Olsen, S. G., In Vitro Cell Dev. Biol. Anim. Meyvantsson, I., Alexander, C. M., Beebe, D. J., Integr.
Biol. 2009, 1, 267–274.
1995, 31, 336–337.

& 2011 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.electrophoresis-journal.com