a microdevice for multiplexed detection of t-cell-secreted cytokines
TRANSCRIPT
PAPER www.rsc.org/loc | Lab on a Chip
Dow
nloa
ded
by U
nive
rsity
of
Cal
ifor
nia
- D
avis
on
22 N
ovem
ber
2010
Publ
ishe
d on
30
Sept
embe
r 20
08 o
n ht
tp://
pubs
.rsc
.org
| do
i:10.
1039
/B81
0244
AView Online
A microdevice for multiplexed detection of T-cell-secreted cytokines
He Zhu,a Gulnaz Stybayeva,ac Monica Macal,b Erlan Ramanculov,c Michael D. George,b Satya Dandekarb
and Alexander Revzin*a
Received 17th June 2008, Accepted 19th August 2008
First published as an Advance Article on the web 30th September 2008
DOI: 10.1039/b810244a
Cytokines are produced by immune cells in response to viral or bacterial pathogens and therefore have
significant diagnostic value. The goal of the present study was to develop a miniature device for
detection of interleukin (IL)-2 and interferon (IFN)-g cytokines secreted by a small population of CD4
and CD8 T-cells. Microarrays of T-cell- and cytokine-specific Ab spots were printed onto poly(ethylene
glycol) (PEG) hydrogel-coated glass slides and enclosed inside a microfluidic device, creating
a miniature (�3 mL) immunoreaction chamber. Introduction of the red blood cell (RBC) depleted
whole human blood into the microfluidic device followed by washing at a pre-defined shear stress
resulted in isolation of pure CD4 and CD8 T-cells on their respective Ab spots. Importantly, the cells
became localized next to anti-IL-2 and -IFN-g Ab spots. Mitogenic activation of the captured T-cells
was followed by immunofluorescent staining (all steps carried out inside a microfluidic device),
revealing concentration gradients of surface-bound cytokine molecules. A microarray scanner was then
used to quantify the concentration of IFN-g and IL-2 near CD4 and CD8 T-cells. This study represents
one of the first demonstrations of a microdevice for capturing desired T-cell subsets from a small blood
volume and determining, on-chip, cytokine profiles of the isolated cells. Such a microdevice is
envisioned as an immunology tool for multi-parametric analysis of T-cell function with direct
applications in diagnosis/monitoring of HIV and other infectious diseases.
Introduction
Activated T-lymphocytes (T-cells) are central participants of
immune response to viral or bacterial infections. T-helper cells
(CD4 T-cells) and cytotoxic T-lymphocytes (CD8 T-cells) are
major T-cell subsets that regulate immune cell recruitment and
proliferation through a program of cytokine production.1,2Based
on the types of secreted cytokines, CD4 T-helper cells are further
subdivided into T-helper 1 (Th1), Th2, Th17 and other subsets,
making cytokine profiling an important part of determining Th
phenotype. Importantly, cytokines secreted by T-cells have been
reported to correlate with positive or negative disease outcomes
and thus provide important diagnostic information. In human
immunodeficiency virus (HIV) infections, Th1 phenotype is
associated with protection against the pathogen and is preserved
in long-term nonprogressor patients, while Th2 phenotype is
correlated with rapid progression of the infection and onset of
AIDS.3–5 Similarly, vigorous cytotoxic response of CD8 T-cells is
needed to keep HIV viremia in check.6 Therefore, detecting
cytokines (e.g. IFN-g and IL-2) associated with Th1 and cyto-
toxic responses of T-cells is important for HIV diagnosis and
monitoring.3,7
Cytokine production is traditionally determined using several
bioanalytical approaches including flow cytometry coupled
aDepartment of Biomedical Engineering, University of California, Davis,451 East Health Sciences St. #2619, Davis, CA, 95616, USA. E-mail:[email protected]; Fax: +1 530-754-5739; Tel: +1 530-752-2383bMedical Microbiology and Immunology, University of California, Davis,USAcNational Center for Biotechnology, Astana, Republic of Kazakhstan
This journal is ª The Royal Society of Chemistry 2008
with intracellular cytokine staining, enzyme-linked immunospot
(ELISPOT) method, enzyme linked immunosorbent assay
(ELISA) and polymerase chain reaction (PCR).4,8,9 ELISA and
PCR are robust technologies for detecting either cytokines or
cytokine mRNA in the blood stream but they can not be used
to identify specific populations of cytokine producing cells. In
contrast, flow cytometry-based approach allows one to connect
cytokine production, determined by intracytoplasmic staining,
to cell phenotype based on Ab-labeling of leukocyte surface
antigens.9,10 However, this method reports the frequency of
cytokine positive cells and not the cytokine concentration; it
requires a large number of cells for analysis and analyzes fixed
(as opposed to live) cells. ELISPOT is another technology
employed to detect cytokine production in leukocytes.8,11 In
this approach, leukocytes are seeded into micro-titer plates pre-
coated with anti-cytokine Abs and are stimulated to commence
cytokine production. The cells are then washed away and the
numbers of spot forming units are quantified in a manner
analogous to ELISA. While it provides information about the
number of cytokine-secreting cells, the ELISPOT experiment
must be preceded by a sorting step to isolate the desired
leukocyte subset. Such an extra step is required due to
‘‘promiscuity’’ in leukocyte cytokine production where IFN-g,
for example, may be secreted by CD4 and CD8 T-cells. In
addition, the ELISPOT technique does not quantify the
amount of secreted cytokines.
Robotic printing has emerged as an important enabling tech-
nology for proteomics research in general12 and immunoassay
development in particular.13–18 Printed arrays of antibodies (Abs)
have been employed to capture pathogens or proteins such as
Lab Chip, 2008, 8, 2197–2205 | 2197
Dow
nloa
ded
by U
nive
rsity
of
Cal
ifor
nia
- D
avis
on
22 N
ovem
ber
2010
Publ
ishe
d on
30
Sept
embe
r 20
08 o
n ht
tp://
pubs
.rsc
.org
| do
i:10.
1039
/B81
0244
AView Online
cytokines for subsequent detection with sandwich immuno-
assay.13,15,17,18
In addition to detection of biomolecules, protein microarrays
have been used for high-throughput screening of cell–surface
interactions.19–21 Belov et al. used robotic printing to create
arrays of leukocyte-specific Abs for isolation or ‘‘panning’’ of
multiple leukocyte subsets on the same surface and for subse-
quent correlation of cell binding patterns to blood malignan-
cies.19 More recently functionality of printed microarrays has
been expanded by incorporating sensing elements in addition to
cell-capture molecules. Chen et al. described a microarray with
individual spots containing T-cell specific antigens and cytokine-
specific Abs.22 This method allowed one to detect cytokine
secreting T-cells bound on the spot but was unsuitable for
quantifying levels of secreted cytokines. Another important
recent study described the use DNA-encoded Ab arrays inte-
grated with microfluidics for detection of cells, proteins and
nucleic acids in the same microdevice.23 While highlighting the
potential use of the same microarray for cell sorting and protein
detection, this study employed purified T-cells and stopped short
of detecting cell-secreted cytokines.
Recently, we described the use of Ab microarrays for
capturing pure (>95%) CD4 and CD8 T-cells from whole human
blood and quantifying T-cell proportions based on cell binding
to Ab spots.24 The goal of the present study was to integrate
multiplexed immunoassay for detection of secreted cytokines
with captured T-cells. To achieve this goal, microarrays of cell-
and cytokine-specific Ab spots were printed side-by-side so as to
position T-cells in the immediate vicinity of the immunosensors
for IFN-g and IL-2 (Fig. 1A,B). To further increase sensitivity of
the cytokine immunoassay and to minimize the blood volume
requirement, the microarrays were enclosed inside a microfluidic
chamber using reversible sealing strategy that was non-destruc-
tive to printed biomolecules (Fig. 1C,D). Incubation of Ab
microarrays with RBC-depleted whole human blood inside
a microfluidic chamber resulted in binding of CD4 and CD8
T-cells on their respective Ab spots next to Ab regions for
detection of IL-2 and IFN-g. Activation of T-cells followed by
immunofluorescent staining revealed cytokine concentration
gradients extending from T-cells. Overall, a microdevice enabling
isolation of pure T-cell subsets from a heterogeneous cell sample
and on-chip detection of cytokines has important applications in
vaccine development as well as diagnosis/treatment of HIV and
other infections.
Experimental
Materials
10� Phosphate-buffered saline (PBS) without calcium and
magnesium, paraformaldehyde (PFA), surfactant TWEEN� 20,
rabbit anti-mouse IgG antibody (2nd labeling antibody)
Na4EDTA, KHCO3, NH4Cl, poly(ethylene glycol)diacrylate
(PEG-DA) (MW575), anhydrous toluene (99.9%), sodium azide,
bovine serum albumin (BSA), 0.1% poly-L-lysine were purchased
from Sigma-Aldrich (Saint Louis, MO). Silane adhesion
promoter, 3-acryloxypropyl trichlorosilane, was from Gelest,
Inc. (Morrisville, PA). Monoclonal antibodies used for capturing
T-lymphocytes and cytokines consisted of the following: purified
2198 | Lab Chip, 2008, 8, 2197–2205
mouse anti-human CD4 Abs (13B8.2) and CD8 (B9.11) from
Beckman-Coulter (Fullerton, CA), and purified mouse anti-
human IFNg Ab (clone K3.53), anti-IL2 Ab (clone 5355), bio-
tinylated goat anti-human IFN-g Ab and biotinylated anti-IL2
Ab from R & D Systems (Minneapolis, MN). Mouse IgG2a
(OX34) was purchased from Serotec Antibodies (Raleigh, NC).
Antibodies used for immunostaining of surface bound cells or
flow cytometry analysis listed below: anti-human CD33 -FITC
(UCHT1), anti-CD4-PE (L120) and anti-CD8-PE (RPA-T8)
were all purchased from BD Pharmingen. Streptavidin conju-
gated with Alexa 546 was purchased from Invitrogen (Carlsbad,
CA). Human recombinant IFN-g and IL-2 were from R & D
Systems (Minneapolis, MN) and Endogen (Woburn, MA),
respectively. T-cell activation reagents: Phorbol 12-myristate 13-
acetate (PMA), Phytohemagglutinin (PHA), Staphylococcal
enterotoxin B (from Staphylococcus aureus) and ionomycin were
purchased from Sigma-Aldrich. Cell culture medium RPMI
1640: 1X, with L-glutamine was purchased from VWR. Poly-
(dimethylsiloxane) (PDMS) and its curing agents were purchased
from Dow Corning (Midland, MI).
Fabrication of multi-functional Ab microarrays
Robotic printing of protein microarrays on PEG gel-coated glass
slides was described previously.24 Briefly, a standard microscope
slide (75� 25 mm) was modified with acrylated silane and coated
with poly(ethylene glycol)-diacrylate (PEG-DA)-based prepol-
ymer. Exposure to UV resulted in cross-linking of the pre-poly-
mer and formation of PEG hdyrogel layer on a glass slide. PEG
gel-coated glass slides were lyophilized for 24 h in order to ensure
rapid absorption and uniform distribution of Abmolecules in the
hydrogel upon printing. 8 � 20 microarray of Ab spots (�150–
250 mm diameter, 375 mm center-to-center spacing) was printed
on hydrogel-coated glass slides using contact microarrayer
(GMS 417 Affymetrix, Santa Clara, CA). Prior to printing, anti-
CD4, -CD8, -IFN-g, -IL-2 and mouse IgG (negative control)
Abs were dissolved in DI water at a concentration of 0.2 mg
mL�1 and were supplemented with BSA (0.5% v/v) and Tween20
(0.005% v/v) (anti-CD4 and -CD8 Abs contain only 0.005%
Tween20). Components of the microarray were loaded into
a 384-well plate (Genetix, Boston, MA) and then robotically
imprinted onto a hydrogel-coated glass slide. After printing,
surfaces were air dried and stored in a sealed box at 4 �C prior to
further use.
Fig. 1B shows design of the microarray with columns of anti-
CD4 and -CD8 Ab spots placed next to rows of cytokine
immunoassay spots. This microarray arrangement ensured that
all cytokine immunoassay spots were exposed to the same cyto-
kine concentration. In addition to capture Abs, cytokine
immunoassay microarray included several negative and positive
controls. Specifically, anti-mouse IgG-2a was employed as
a negative control while biotin molecules were printed to provide
a positive control for the binding of the streptavidin-Alexa-546
complex used during immunofluorescent labeling.
Design of a microfluidic platform
A microfluidic platform was developed to enable isolation of
T-cells from a small blood sample and to increase local
This journal is ª The Royal Society of Chemistry 2008
Fig. 1 (A) The conceptual design of microarrays for detection of T-cell-secreted cytokines. Printing of cell- and cytokine-specific Ab spots side-by-side
allowed one to capture T-cells next to IL-2 and IFN-g sensing regions. T-cell-secreted cytokines were detected on the adjacent anti-cytokine Ab spots.
(B) A map of the 8 � 20 microarray for capturing T-cells and detecting T-cell-secreted IL-2 and IFN-g. (C) Design of a microfluidic platform employed
for integration with Ab microarrays. The device containing two reaction chambers was secured on top of microarrays by applying vacuum suction
through a web of auxiliary channels. This approach allowed Ab microarrays to be enclosed inside a microfluidic chamber in a reversible and protein-
friendly manner. (D) An image of a PDMSmicrodevice employed for T-cell capture and cytokine detection experiments with one reaction chamber filled
with unlysed whole blood.
Dow
nloa
ded
by U
nive
rsity
of
Cal
ifor
nia
- D
avis
on
22 N
ovem
ber
2010
Publ
ishe
d on
30
Sept
embe
r 20
08 o
n ht
tp://
pubs
.rsc
.org
| do
i:10.
1039
/B81
0244
AView Online
concentration of secreted cytokines. Standard soft lithography
approaches were used to fabricate poly(dimethyl siloxane)
(PDMS)-based microfluidic devices.25 A transparency photo-
mask (CAD Art Services, Poway, CA) was generated based on
an AutoCAD drawing of the device. This photomask was then
employed to micropattern SU-8 on a 4 inch silicon (Si) wafer in
order to create a negative replica of the fluidic network. Poly(-
dimethylsiloxane) (PDMS) was mixed 10 : 1 with a curing agent,
poured onto a Si wafer containing SU-8 features and cured for 12
h at 60 �C. The elastomer with imbedded channel architecture
was released and inlet/outlet holes were punched with a blunt 16
gauge needle. The microfluidic device, shown in Fig. 1C, con-
tained two immunoreaction chambers with width–length–height
dimensions of 3 � 10 � 0.1 mm as well as a network of inde-
pendently addressed auxiliary channels. The auxiliary channels
were used to apply negative pressure (vacuum suction) to the
PDMS mold and reversibly secure it on top of a glass substrate.
This strategy of reversible sealing of microfluidic devices, first
described by Schaff et al.,26 allowed us to enclose protein
microarrays inside fluidic conduits without damaging or dena-
turing Ab molecules.
A 5 mL syringe was connected to silicone tubing (1/32 inch id,
Fisher) which was then attached to the outlet of the flow chamber
with a metal insert cut from a 20 gauge needle. A blunt shortened
This journal is ª The Royal Society of Chemistry 2008
20 gauge needle carrying a plastic hub with volume of 100 mLwas
inserted in the inlet. A pressure-driven flow in the microdevice
was created by withdrawing the syringe positioned at the outlet
using a precision syringe pump (Harvard Apparatus, Boston,
MA).
The shear stress (t) inside the channel was estimated assuming
an expression for an infinitely wide parallel-plate flow chamber:
t ¼ 6mQ
wh2(1)
where Q is flow rate, m is viscosity of the medium (approximately
10�2 dyn s cm�2), w is channel width and h is channel height. For
a channel 3 mm in width and 100 mm in height, flow rates of 3 and
50 mL min�1 would yield shear stress values of 0.1 and 1.67 dyn
cm�2, respectively. In our experiments, low flow rate was used for
seeding and capturing leukocytes on Ab microarrays, while
higher flow rate was employed for removal of non-specific cells.
Capture of T-cells on Ab microarrays
Red blood cell (RBC) depleted whole human blood was
employed in T-cell isolation experiments. Blood was collected
from healthy adult donors through venipuncture under sterile
conditions with informed consent and approval of the
Lab Chip, 2008, 8, 2197–2205 | 2199
Dow
nloa
ded
by U
nive
rsity
of
Cal
ifor
nia
- D
avis
on
22 N
ovem
ber
2010
Publ
ishe
d on
30
Sept
embe
r 20
08 o
n ht
tp://
pubs
.rsc
.org
| do
i:10.
1039
/B81
0244
AView Online
Institutional Review Board of the University of California at
Davis (protocol number 200311635–6). RBCs were removed
using ammonia chloride based erythrocyte lysis solution (89.9 g
NH4Cl, 10.0 g KHCO3, and 370.0 mg tetrasodium EDTA in 10 L
of deionized water) as described previously.24,27 RBC-depletion
produced blood cell suspension comprised of granulocytes,
peripheral blood mononuclear cells and RBC debris. The cells
were concentrated by centrifugation and re-suspended in
RPMI1640 medium containing 10% FBS and 1% penicillin/
streptomycin but without L-glutamine and Phenol Red (Medi-
atech, Herndon, VA). The leukocyte suspension was used
immediately for microfluidic cytokine detection experiments.
Prior to introduction of cells, PDMS molds containing fluidic
and vacuum channels were sterilized by 15 min UV exposure in
a tissue culture hood. Sterile 1� PBS was first injected into the
flow chamber to remove air bubbles, then 50 mL of RBC-lysed
blood resuspended in phenol red-free RPMI1640 was added into
the inlet reservoir (hub of a 20 gauge needle) and drawn into
a microfluidic channel at shear stress of 0.3 dyn cm�2 (9 mL
min�1). Upon entry of cells into a channel the shear stress was
decreased to 0.1 dyn cm�2 (3 mL min�1) and was maintained for
10 min allowing cells to interact with Ab microarrays. This time
was found by us to be sufficient for binding of cells on Ab spots.27
In order to wash away nonspecific cells while retaining T-cells
captured on Ab microarrays flow rate was increased to 50 mL
min�1, creating a shear stress of 1.7 dyn cm�2. The shear stress
window of 1–2 dyn cm�2 was previously found to be optimal for
removal of nonspecifically bound cells as well as contaminating
monocytes, while retaining CD4 and CD8 T-cells.24,28
Detection of T-cell secreted cytokines in a microfluidic device
After capturing T-cells, a microfluidic device was flushed with 1�PBS and filled with mitogenic agents to commence cell activation
and cytokine production. The mitogenic solution consisted of
PMA and ionomycin dissolved to concentrations of 50 ng mL�1
and 2 mM respectively in phenol red-free RPMI1640 media
supplemented with 10% FBS. A surgical clamp was secured
around the inlet/outlet tubing to eliminate convective mixing and
to establish diffusion gradient of cytokine concentration. A
microfluidic chamber with captured T-cell was then placed into
a tissue culture incubator (37 �C, 5% CO2 and 90% humidity) for
periods of time ranging from 1 to 6 h.
At the end of the desired activation period T-cells and anti-
cytokine Ab spots enclosed in a microfluidic chamber were
exposed to reagents for immunofluorescent staining of cells and
cytokines. After flushing away mitogenic solution with 1� PBS
fluidic chambers were filled with biotinylated anti-IFN-g and
-IL-2 Abs (5 mg mL�1 in 1� PBS) and incubated for 1 h at room
temperature. The chambers were then flushed with 1� PBS again
and filled with streptavidin-Alexa546 (10 mg mL�1 in 1� PBS) for
30 min in order to reveal the presence of biotin-Ab-cytokine-Ab
sandwich complex. Immunoassay strategy employed in this
study is described pictorially in Fig. 1A.
The immunofluorescently labeled anti-cytokine Ab spots were
visualized and imaged inside a fluidic chamber using a confocal
microscope (Zeiss LSM 5 Pascal, Carl Zeiss, Inc.). Quantifying
cytokine signals using a microarray scanner required removal of
a PDMS chamber. Prior to opening a microfluidic device cells
2200 | Lab Chip, 2008, 8, 2197–2205
adherent on Ab spots were fixed by exposure to 1% para-
formaldehyde (PFA) for 15 min. Glass slides containing T-cell
arrays and cytokine immunoassays were air dried and analyzed
using a microarray scanner.
Calibration curves were constructed to translate fluorescence
signal to concentration of IFN-g and IL-2. To construct cali-
bration curves, microarrays of anti-Il-2 and anti-IFN-gAbs were
incubated for 30 min with recombinant human IFN-g and IL-2
dissolved in 1� PBS supplemented with 0.5% v/v BSA and
0.005% Tween20. Cytokine concentration range tested was 1 to
400 ng mL�1. Importantly, printed arrays contained both types
of anti-cytokine Abs but were exposed to one type recombinant
cytokine to ensure lack of cross-reactivity between the assays.
These microarrays were incubated with biotinylated anti-cyto-
kine Abs and streptavidin-Alexa546 to reveal cytokine-concen-
tration dependent fluorescence signal. In order to create
a quantitative readout of fluorescence signal emanating from the
array, the laser microarray scanner was employed to scan the
glass slides at a spot pixel resolution of 5 mm (Agilent, Santa
Clara, CA). The fluorescence intensity of each array element was
determined using GenePix Pro 6.0 data analysis software
(Molecular Devices, Downingtown, PA).
Flow cytometry analysis of T-cell cytokine production
RBC-lysed whole blood was stimulated with single mitogens or
media alone (negative control) for 6 h at 37 �C. Brefeldin A
(Sigma, St. Louis, MO) was added for the final 5 h to prevent
Golgi transport of cytokines out of cells. Specifically the cells
were stimulated in a fashion identical to that described in the
previous section using 50 ng mL�1 PMA, or 2 mM mL�1 Ion-
omycin and incubated at 37 �C with 5% CO2 for 6 h.
Following stimulation, cells were washed in 1 mL of 1� PBS
for 5 min at 1600 rpm and incubated with Aqua (Amcyan) Live/
Dead Dye (Invitrogen, Carlsbad, CA) at a concentration of 0.5
mL million cells�1 in 1 mL of 1� PBS for 30 min at 4 �C. Cells
were washed with staining buffer (filter sterilized 1� PBS with
3% heat-inactivated FBS (Gibco) and 0.1% sodium azide
(Sigma)). Cells were re-suspended in staining buffer and incu-
bated with Abs for surface markers CD3 APC Cy7, CD4 APC
(Beckton Dickinson (BD), Mountain View, CA), and CD8 FITC
(Ebioscience, San Diego, CA) for 30 min 4 �C. Cells were then
fixed in 1% PFA for 15 min at room temperature. Fixative was
washed out and cells were permeabilized with Caltag Per-
meabilization Solution B (Invitrogen, Carlsbad, CA) and stained
with monoclonal antibodies IFN-g PE Cy7 (Ebioscience) and
IL-2 PE (BD) for 25 min at room temperature. Samples were
washed and fixed with 1% PFA before analysis with LSR II (BD)
at the UC Davis Optical Core Facility. A minimum of 300 000
events were collected for each sample and analyzed using Flow Jo
software (Treestar, Inc, San Carlos, CA).
Results and discussion
The goal of this study was to develop a strategy for on-chip
detection of cytokines secreted from captured T-cells. An
approach utilizing a printed microarray of cell- and cytokine-
specific Ab spots enclosed inside a microfluidic device enabled
placement of T-cells next to cytokine immunosensors inside
This journal is ª The Royal Society of Chemistry 2008
Dow
nloa
ded
by U
nive
rsity
of
Cal
ifor
nia
- D
avis
on
22 N
ovem
ber
2010
Publ
ishe
d on
30
Sept
embe
r 20
08 o
n ht
tp://
pubs
.rsc
.org
| do
i:10.
1039
/B81
0244
AView Online
a miniature fluidic chamber. Mitogenic stimulation of T-cells
then resulted in detection of IL-2 and IFN-g on the adjacent anti-
cytokine Ab spots (Fig. 1A). A miniature cytometry platform
described here may have a wide range of applications in immu-
nology research including monitoring/diagnostics of infectious
diseases and development of vaccines.
Integrating Ab microarrays into microfluidic devices
In order to minimize blood sample required for T-cell analysis
and to increase local concentration of secreted cytokines it was
important to enclose Ab microarrays inside a miniature flow-
through cell. Microfluidic devices have been used extensively for
cell sorting and manipulation25,29,30 and have recently found an
application in blood cell analysis.31 Several studies described the
use of Ab-modified microfluidic chambers for capturing model
immune cells as well as primary T-lymphocytes from heteroge-
neous cell suspensions.28,32,33 In these studies, microfluidic devices
were irreversibly sealed using oxygen plasma treatment and then
modified with a uniform layer of Ab molecules, thus, allowing
one to avoid exposure of Ab molecules to potentially destructive/
denaturing effects of oxygen plasma. In contrast, Abmicroarrays
employed for cell capture in our studies had to be printed prior to
assembly of microfluidics and had to be removed after cell
capture experiment for scanning of cytokine arrays. Therefore,
a reversible and protein-friendly method for sealing a micro-
fluidic chamber was required. Such a reversible seal was created
by applying negative pressure (vacuum suction) to PDMS mold
through a network of auxiliary channels, in a manner similar to
that described recently by Schaff et al.26 This method entailed
creating two networks of channels in PDMS, one for blood
Fig. 2 Immunofluorescent staining of cells captured on anti-CD4 and -CD8
clusters formed on 150 mm diameter anti-CD4 spots showing that captured c
identifying these cells as CD4+ T-cells. (C,D) Immunofluorescent image of a 3�showing that the majority of captured cells are stained with anti-CD3 FITC
T-lymphocyte phenotype.
This journal is ª The Royal Society of Chemistry 2008
sample analysis, another for applying vacuum, addressed with
separate inlet/outlet ports (see Fig. 1C for pictorial description).
An example of a microfluidic device containing two immuno-
reaction chambers (one of which is filled with blood) as well as
a network of auxiliary vacuum channels used to achieve revers-
ible seal between a PDMS device and a glass substrate is shown
in Fig. 1D. This microfluidic platform was employed for all cell
seeding, washing and immunofluorescent staining steps
described in the paper.
Capturing T-cells and detection of secreted cytokines on Ab
microarrays
It was important to ensure purity of the captured T-cell pop-
ulation in order to correctly assign a cytokine profile to a desired
cell subset. This requirement was motivated by the fact that
production of IL-2 and IFN-g may be associated with Th1
response of CD4 T-cells or cytotoxic function of CD8 T-cells.
Attaining high purity of the desired leukocyte type was highly
dependent on optimization of washing protocols in order to
remove loosely bound non-specific cells without dislodging CD4
or CD8 T-cells. Employing a microfluidic device for T-cell
capture experiments allowed to precisely control shear stress
inside a fluidic conduit. T-cell panning step was performed at
a low flow rate of 3 mL min�1 (0.1 dyn cm�2) while the washing
step was carried out at 50 mL min�1 (1.7 dyn cm�2). The latter
value of shear stress was previously shown to result in removal of
loosely bound non-specific cells and elimination of CD4-
expressing monocytes.24,27,28 Coupled with nonfouling properties
of PEG hydrogel coating, the washing procedure described
above allowed to isolate T-cells on Ab spots with minimal
Ab microarrays. (A,B) Fluorescent image of the 3 � 4 array of leukocyte
ells are stained with both anti-CD3 FITC (green) and anti-CD4 PE (red)
4 array of leukocyte clusters formed on 150 mm diameter anti-CD8 spots
(green) but not with anti-CD4 PE (red) pointing to CD3+CD4� or CD8+
Lab Chip, 2008, 8, 2197–2205 | 2201
Fig. 3 Detection of secreted cytokines using Ab microarrays. (A) Cells
residing on 300 mm diameter anti-CD4 Ab spots next to cytokine-sensing
spots after 6 h mitogenic activation. Fluorescent staining of the micro-
arrays with CD3 FITC (green), Dapi (blue) and anti-cyotkine Ab-biotin/
streptavidin-Alexa 546 (red) shows that CD3+ T-cells were captured on
cell-specific Abs and minimal cell attachment was observed on anti-
cytokine spots or elsewhere on a PEG hydrogel coated glass surface.
Mitogenic activation resulted in cytokine production detected with IFN-
g and IL-2 sensing spots (red). (B) CD4+ human T-cells captured next to
anti-IL-2 and –IFN-g Ab spots. Mitogenic stimulation of T-cells with
PMA and ionomycin was followed by incubation with biotinylated anti-
cytokine Abs, followed by incubation with streptavidin-Alexa546. There
2202 | Lab Chip, 2008, 8, 2197–2205
Dow
nloa
ded
by U
nive
rsity
of
Cal
ifor
nia
- D
avis
on
22 N
ovem
ber
2010
Publ
ishe
d on
30
Sept
embe
r 20
08 o
n ht
tp://
pubs
.rsc
.org
| do
i:10.
1039
/B81
0244
AView Online
nonspecific cell binding. Phenotypic purity of the captured cells
was analyzed by immunofluorescent staining. Representative
images shown in Fig. 2A,B demonstrate that cells captured on
anti-CD4 Ab spots were CD3+CD4+, definitely identifying them
as CD4 T-cells. Conversely, cells captured on anti-CD8 Ab spots
were CD3+CD4� which is logical given that CD4 antigen is not
expressed on CD8 T-cells (Fig. 2C,D). Because the majority
(over 95%) of mature T-cells express either CD4 or CD8 antigen,
the CD3+CD4� cells were assigned CD8 T-cell phenotype.
Direct immunostaining for CD8 antigen was performed by us
previously, confirming the presence of CD8+ T-cells on the Ab
spots.24
The capture of pure T-cell subsets next to cytokine immuno-
sensors inside a miniature (�3 mL) immunoreaction chamber was
followed by cell activation and cytokine detection steps. T-cells
were mitogenically stimulated using a combination of PMA and
ionomycin for a period of time ranging from 1 to 6 h under
physiological conditions (37 �C and 5% CO2). Activation with
PMA and ionomycin was previously reported to result in robust
production of IL-2 and IFN-g as verified by intracellular staining
and flow cytometry.34 During T-cell activation, flow in the
microfluidic channel was stopped and care was taken to eliminate
convective mixing, and ensure that diffusion was the sole mass
transport mechanism. Fig. 3A demonstrates immunofluorescent
labeling of CD4 T-cell- and cytokine-capture microarray after 6h
activation inside a microfluidic device. T-cells were stained with
anti-CD3 FITC (green) and Dapi (blue) whereas immnuno-
sensing spots were stained with biotinylated anti-cytokine Abs,
followed by streptavidin-Alexa 546 (red). Fig. 3A highlights
several important aspects of the cytokine detection platform: (1)
the majority of cells panned from RBC-depleted whole blood on
anti-CD4 Ab spots were CD3+, suggesting T-cell phenotype, (2)
minimal cell adhesion was observed on anti-cytokine Abs or
PEG hydrogel-coated glass substrate, underscoring the possi-
bility of placing pure cell population next to immunosensing
regions, (3) IL-2 and IFN-g could be detected using microarray
immunoassay format.
Fig. 3B,C demonstrates another cytokine detection experiment
where CD4 and CD8 T-cells were captured in different chambers
of a microfluidic device and were activated for 4 h. As seen from
Fig. 3B, activation of CD4 T-cells resulted in cytokine secretion
and binding on anti-IL-2 and anti-IFN-g Ab spots. A concen-
tration gradient in signal intensity was observed with signal
becoming undetectable � 2 mm away from the cytokine-
secreting T-cells. This underscores the importance of placing
immunosensing domains next to T-cells and miniaturizing the
reaction volume. Fig. 3B,C also illustrates a difference in cyto-
kine profiles of CD4 T-cells and CD8 T-cells, with former
producing both cytokines and latter producing only IFN-g. It
should be also be noted that cytokine production was observed
only when T-cells were mitogenically activated (data not shown).
The detection of IL-2 and IFN-g after 4 h mitogenic stimu-
lation described in Fig. 3B,C was confirmed by intracellular
was a clear fluorescence signature for both IL-2 and IFN-g secreted by
CD4 T-cells. (C) CD8+ human T-cells captured next to anti-IL-2 and
–IFN-g Ab spots. Mitogenic stimulation of these cells pointed to the
presence of IFN-g and lack of IL-2 production.
This journal is ª The Royal Society of Chemistry 2008
Fig. 4 Analysis of cytokine production using flow cytometry. RBC-depleted whole blood was mitogenically activated and cytokine production was
analyzed by intracellular staining and flow cytometry. Lymphocyte population determined by forward and side scatter was gated to single cells and then
further gated to CD3+ T-cells. (A) Flow cytometry analysis of IFN-g and IL-2 in CD4+ T-cell population points to robust production of both cytokines
by this leukocyte subset. (B) Flow cytometry analysis showed that CD8+ T-cells robustly produced IFN-g but not IL-2. These data corroborate results
observed with our microarray immunosensing strategy.
Dow
nloa
ded
by U
nive
rsity
of
Cal
ifor
nia
- D
avis
on
22 N
ovem
ber
2010
Publ
ishe
d on
30
Sept
embe
r 20
08 o
n ht
tp://
pubs
.rsc
.org
| do
i:10.
1039
/B81
0244
AView Online
staining and flow cytometry of the same blood sample. As shown
in Fig. 4A, 45% of CD4 T-cells stained positive for intracellular
IL-2 while 28% were positive for IFN-g. In contrast, as seen from
Fig. 4B, CD8 cells were observed to have a robust IFN-g
response (39%) but limited IL-2 production (4%). Therefore,
microarray- and flow cytometry-based detection methods
showed similar patterns in cytokine production by CD4 and CD8
T-cells. However, the flow cytometry reported the frequency of
appearance of cytokine-producing cells and did not reveal the
amount of cytokines produced by the cells. On the other hand,
microarray-based cytokine immunoassays enabled us to quantify
the concentration of cytokines produced by the captured T-cells.
These experiments are described in the following section.
Quantifying T-cell cytokine production
The use of the microarray scanner allowed us to rapidly and
quantitatively analyze fluorescence signals emanating from
immunosensing regions. Fig. 5A shows a region of the T-cell/Ab
microarray with a tracer line across cytokine sensing spots. These
line scans were converted into fluorescence intensity profiles
using a microarray scanner. Fig. 5B demonstrates scanned
fluorescence intensity data from anti-mouse IgG (negative
control), anti-Il-2 and anti-IFN-g spots located next to cells in
experiments where mitogenic activation of CD4 T-cells was
varied from 1 to 6 h. As seen from these data, IFN-g and IL-2
signal intensity increased as a function of activation time;
This journal is ª The Royal Society of Chemistry 2008
however, no IL-2 signal was observed after 1 h of mitogenic
activation. It should be noted that the negative control (mouse
IgG) spots had a low background signal contributing to high
signal-to-noise ratio of immunosensing microarrays.
In order to convert arbitrary fluorescence intensity units into
cytokine concentration, calibration curves using known amounts
of recombinant IFN-g and IL-2 were constructed. Fig. 5C,D
shows concentration dependent modulation in fluorescence
intensity of anti-IFN-g and anti-IL-2 spots. The higher sensi-
tivity of IFN-g immunosensor compared to IL-2 may be due to
differences in association/dissociation kinetics of antibody–
antigen interactions. Calibration curve for IFN-g was fitted with
a logarithmic best fit while IL-2 concentration was a linear
function of fluorescence with R2 ¼ 0.96 for both curves.
Employing calibration curves, the fluorescence intensity profiles
presented in Fig. 5B were converted into cytokine concentration,
thus allowing us to establish the amount of IL-2 and IFN-g
secreted by CD4 T-cells after varying extents of mitogenic acti-
vation. As shown in Fig. 5E, IFN-g concentration was estimated
to be �18 ng mL�1 after 1 and 4 h of mitogenic activation, and
increased further to �230 ng mL�1 after 6 h of stimulation inside
a microdevice. No IL-2 production was observed after 1 h
stimulation; however, this cytokine was detected after 4 h of
activation (70 ng mL�1) and was strongly expressed after 6 h of
stimulation (360 ng mL�1). The data presented in Fig. 5 point to
the possibility of using microarray immunoassays placed next to
arrays of T-cells for on-chip detection of secreted cytokines.
Lab Chip, 2008, 8, 2197–2205 | 2203
Fig. 5 Quantifying cytokine production by captured T-cells. (A) Staining of microarrays with biotinylated anti-cytokine Abs followed by streptavidin-
Alexa 546 revealed strong fluorescence signals from anti-IL-2 and anti-IFN-gAb spots and no signal from mouse IgG serving as a negative control. The
white trace shows a line scan from a microarray scanner. (B) Scans of fluorescence intensity profiles for different times of mitogenic activation of CD4
T-cells. (C–D) Calibration curves for converting fluorescence intensity into concentration of IFN-g and IL-2. (E) Concentration of IL-2 and IFN-g
detected from CD4 T-cells activated for different time periods. The concentration values correspond to fluorescence intensity scans shown in part (B).
Dow
nloa
ded
by U
nive
rsity
of
Cal
ifor
nia
- D
avis
on
22 N
ovem
ber
2010
Publ
ishe
d on
30
Sept
embe
r 20
08 o
n ht
tp://
pubs
.rsc
.org
| do
i:10.
1039
/B81
0244
AView Online
Another advantage of using microarray format for cytokine
detection is the simplicity of the signal readout. In our strategy,
anti-cytokine Abs, positive/negative controls and cytokine refer-
ence points were printed in discrete locations of the substrate so
that only one fluorescence color was needed to discern all signals.
In addition, using cell-specific microarrays allowed us to capture
different leukocyte subsets on defined Ab spots and to identify
a cell subset solely based on ‘‘location’’ within the microarray
without the use of fluorescence. In contrast, flow cytometry
analysis of multiple cytokines produced by different leukocyte
subsets requires each cell subset to have surface antigen- and
cytokine-specific fluorescence signature, necessitating the use of
complex instrumentation for multi-parametric analysis.
Conclusion
This paper describes the development of a microdevice for
isolating pure T-cell subsets and detecting cell-secreted cytokines:
IL-2 and IFN-g. Printedmicroarrays of cell- and cytokine-specific
Ab spots were enclosed inside a microfluidic device and were
employed to capture pure T-cell subsets next to cytokine-sensing
domains. Mitogenic activation of T-cells followed by immuno-
fluorescent staining of the microarrays revealed fluorescence
signals due to binding of IL-2 and IFN-g on anti-cytokine Ab
spots. An immunosensing strategy described here offers several
2204 | Lab Chip, 2008, 8, 2197–2205
advantages: (1) Ab microarrays allowed us to perform T-cell
sorting and cytokine detection in the same microdevice, elimi-
nating an off-chip leukocyte purification step employed by
conventional immunological methods, (2) placement of cells near
cytokine immunosensors inside a microfluidic chamber ensured
high local concentration of the analyte, (3) capturing pure
leukocyte subsets on specific Ab spots within the microarray
allowed us to identify cell phenotype solely based on location and
eliminated the need for fluorescent staining of cells, (4) printing
cytokine-sensing Abs in discrete locations of the surface enabled
the use of single fluorophore for microarray analysis. In the
future, a miniature cytometry platform described here may be
employed for multi-parametric analysis of T-cell function in
diagnosis and monitoring of infectious disease such as HIV.
Acknowledgements
We thank Prof. Louie’s lab in the Department of Biomedical
Engineering at UC Davis for providing assistance with confocal
microscopy. Scanning of the microarrays was performed at
Expression Analysis Facility at UC Davis. Financial support for
this work was provided in part by the California Research Center
for the Biology of HIV in Minorities, California HIV/AIDS
Research Program #CH05-D-606. Additional financial support
was provided in part by an NIH grant DK61297 awarded to SD.
This journal is ª The Royal Society of Chemistry 2008
Dow
nloa
ded
by U
nive
rsity
of
Cal
ifor
nia
- D
avis
on
22 N
ovem
ber
2010
Publ
ishe
d on
30
Sept
embe
r 20
08 o
n ht
tp://
pubs
.rsc
.org
| do
i:10.
1039
/B81
0244
AView Online
HZ was supported through an NIH Training Grant (EB003827).
GS was supported through a Biotechnology Fellowship from
National Center for Biotechnology, Republic of Kazakhstan.
References
1 S. Romagnani, Clin. Immunol. Immunopathol., 1996, 80(3), 225–235.2 P. Salgame, J. S. Abrams, C. Clayberger, H. Goldstein, J. Convit,R. L. Modlin and B. R. Bloom, Science, 1991, 254(5029), 279–282.
3 G. Pantaleo and R. A. Koup, Nat. Med., 2004, 10, 806–810.4 M. Clerici, F. T. Hakim, D. J. Venzon, S. Blatt, C. W. Hendrix,T. A. Wynn and G. M. Shearer, J. Clin. Invest., 1993, 91(3), 759–765.
5 M. Clerici and G. M. Shearer, Immunol. Lett., 1996, 51, 69–73.6 P. Shankar, M. Russo, B. Harnisch, M. Patterson, P. Skolnik andJ. Lieberman, Blood, 2000, 96, 3094–3101.
7 G. Pantaleo and A. Harari, Nat. Rev. Immunol., 2006, 6, 417–423.8 J. H. Cox, G. Ferrari and S. Janetzki, Methods, 2006, 38, 274–282.9 A. C. Karlsson, J. N. Martin, S. R. Younger, B. M. Bredt, L. Epling,R. Ronquillo, A. Varma, S. C. Deeks, J. M. McCune, D. F. Nixonand E. Sinclair, J. Immunol. Methods, 2003, 283, 141–153.
10 V. C. Maino and L. J. Picker, Cytometry, 1998, 34(5), 207–215.11 R. E. M. Guerkov, O. S. Targoni, C. R. Kreher, B. O. Boehm,
M. T. Herrrera, M. Tary-Lehmann, P. V. Lehmann andS. K. Schwander, J. Immunol. Methods, 2003, 279, 111–121.
12 G. MacBeath and S. L. Schreiber, Science, 2000, 289, 1760–1763.13 J. B. Delehanty and F. S. Ligler, Anal. Chem., 2002, 74, 5681–5687.14 C. R. Taitt, G. P. Anderson, B. M. Lingerfelt, M. J. Feldstein and
F. S. Ligler, Anal. Chem., 2002, 74(23), 6114–6120.15 J. R. Falsey, M. Renil, S. Park, S. Li and K. S. Lam, Bioconjugate
Chem., 2001, 12, 346–353.16 Y. Li, N. Nath and W. M. Reichert, Anal. Chem., 2003, 75, 5274.
This journal is ª The Royal Society of Chemistry 2008
17 Y. Li and W. M. Reichert, Langmuir, 2003, 19, 1557–1566.18 C. C. Wang, J. Proteome Res., 2002, 1, 337–343.19 L. Belov, O. de la Vega, C. G. dos Remedios, S. P. Mulligan and
R. I. Christopherson, Cancer Res., 2001, 61, 4483–4489.20 A. Revzin, P. Rajagopalan, A. W. Tilles, F. Berthiaume,
M. L. Yarmush and M. Toner, Langmuir, 2004, 20, 2999–3005.21 C. J. Flaim, S. Chien and S. N. Bhatia, Nat. Methods, 2005, 3, 119.22 D. S. Chen, Y. Soen, T. B. Stuge, P. P. Lee, J. S. Weber, P. O. Brown
and M. M. Davis, PLoS Med., 2005, 2(10), 1018–1030.23 R. C. Bailey, G. A. Kwong, C. G. Radu, O. N. Witte and J. R. Heath,
J. Am. Chem. Soc., 2007, 129, 1959–1967.24 H. Zhu, M. Macal, M. D. George, S. Dandekar and A. Revzin, Anal.
Chim. Acta, 2008, 608, 186–196.25 G.M.Whitesides, E. Ostuni, S. Takayama, X. Jiang andD. E. Ingber,
Annu. Rev. Biomed. Eng., 2001, 3, 335–373.26 U. Y. Schaff, M. M. Q. Xing, K. K. Lin, N. Pan, N. L. Jeon and
S. I. Simon, Lab Chip, 2007, 7(4), 448–456.27 K. Sekine, A. Revzin, R. G. Tompkins and M. Toner, J. Immunol.
Methods, 2006, 313, 96–109.28 X. H. Cheng, D. Irimia, M. Dixon, K. Sekine, U. Demirci, L. Zamir,
R. G. Tompkins, W. Rodriguez and M. Toner, Lab Chip, 2007, 7(2),170–178.
29 D. Irimia and M. Toner, Lab Chip, 2006, 6, 345–352.30 C. E. Sims and N. L. Allbritton, Lab Chip, 2007, 7, 423–440.31 M. Toner and D. Irimia, Annu. Rev. Biomed. Eng., 2005, 7, 77–103.32 S. K. Murthy, A. Sin, R. G. Tompkins and M. Toner, Langmuir,
2004, 20, 11649–11655.33 A. Sin, S. K. Murthy, A. Revzin, R. G. Tompkins and M. Toner,
Biotechnol. Bioeng., 2005, 91, 816–826.34 A. Rodriguez-Caballero, A. Garcia-Montero, C. Bueno, J. Almeida,
R. Varro, R. Chen, A. Pandiella and A. Orfao, Lab. Invest., 2004,84, 1387–1398.
Lab Chip, 2008, 8, 2197–2205 | 2205