PAPER www.rsc.org/loc | Lab on a Chip

Dow

nloa

ded

by L

akeh

ead

Uni

vers

ity o

n 08

Mar

ch 2

013

Publ

ishe

d on

08

Oct

ober

200

9 on

http

://pu

bs.r

sc.o

rg |

doi:1

0.10

39/B

9118

65A

View Article Online / Journal Homepage / Table of Contents for this issue

Correlating short-term Ca2+ responses with long-term proteinexpression after activation of single T cells†

Michael Kirschbaum, Magnus Sebastian Jaeger* and Claus Duschl

Received 18th June 2009, Accepted 15th September 2009

First published as an Advance Article on the web 8th October 2009

DOI: 10.1039/b911865a

In order to elucidate the dynamics of cellular processes that are induced in context with intercellular

communication, defined events along the signal transduction cascade and subsequent activation steps

have to be analyzed on the level of individual cells and correlated with each other. Here we present an

approach that allows the initiation of cell–cell or cell–particle interactions and the analysis of cellular

reactions within various regimes while the identity of each individual cell is preserved. It utilizes

dielectrophoresis (DEP) and microfluidics in a lab-on-chip system. With high spatial and temporal

precision we contacted single T cells with functionalized microbeads and monitored their immediate

cytosolic Ca2+ response. After this, the cells were released from the chip system and cultivated further.

Expression of the activation marker molecule CD69 was analyzed the next day and correlated with the

previously recorded Ca2+ signal for each individual cell. We found a significant difference in the patterns

of Ca2+ traces between activated and non-activated cells, which shows that Ca2+ signals in T cells can

provide early information about a later reaction of the cell. Although T cells are non-excitable cells, we

also observed irregular Ca2+ transients upon exposure to the DEP field only. These Ca2+ signals depended

on exposure time, electric field strength and field frequency. By minimizing their occurrence rate, we

could identify experimental conditions that caused the least interference with the physiology of the cell.

Introduction

Improving the understanding of biochemical signaling pathways

that are triggered by cellular interactions is at the focus of many

efforts in biological research. As individual signaling events are

interconnected in a very complex manner and occur at a wide

range of time scales, the elucidation of their role along the

signaling cascades is difficult. A key issue in this context is the

correlation of individual signaling events with each other.

However, this cannot be achieved by standard cell culture

methods that pool biological data of thousands of cells because

establishing their averages ignores the complexity of the signal

transduction processes and leads to highly blurred results.

Instead, only a multi-parameter analysis that takes into account

the variability in timing and response patterns of individual cells

is able to exploit the information that is crucially required for

deducing entire signaling pathways.

Here we present a novel approach for the controlled initiation

of cellular interactions and the subsequent investigation of the

signaling cascades induced on a single-cell level. As a model

system, T cell activation has been chosen. It represents a well-

characterized example for cell- or surface-mediated signal

Fraunhofer Institute for Biomedical Engineering (IBMT), AmMuehlenberg 13, 14476 Potsdam, Germany. E-mail: magnus.jaeger@ibmt.fraunhofer.de; Fax: +49/3 31/581 87 - 399; Tel: +49/3 31/581 87 - 305

† Electronic supplementary information (ESI) available: Quantificationof temperature distribution in DEP chip with E-field applied toelectrodes; Ca2+ signals after contact formation with antibody-coatedand uncoated microbeads; influence of Fura-laoding on Jurkat T-cells;ambient temperature experienced by cell during DEP manipulation;and statistical analysis of data on impact of DEP manipulation on Tcells. See DOI: 10.1039/b911865a

This journal is ª The Royal Society of Chemistry 2009

transduction processes and plays a crucial role during the

immune response. We stimulated single T cells with functional-

ized microbeads and analyzed both the short-term response on

the second messenger level and the long-term response on the

level of protein expression. As a parameter reflecting the short-

term response, we analyzed the intracellular Ca2+ mobilization

upon bead stimulation since it plays a central role in many signal

transduction processes and, more specifically, because of current

efforts in immunological research to correlate it to lymphocyte

activation.1 To evaluate the long-term response, we analyzed the

expression of a well-established marker protein for T cell acti-

vation, CD69. Since the cells could be individually addressed

during the entire manipulation procedure, we were able to

correlate both signal types for each cell.

In previous studies, it has been shown that T cells can be

activated by contacting them with antigen-presenting cells

(APC) or antibodies against the T cell receptor (TCR)-associ-

ated kinase CD3 and the costimulatory molecule CD28. Part of

the cellular response to the external stimulus is a rapid rise of

the cytosolic Ca2+ concentration.2 As a consequence, tran-

scription factors are mobilized and modify gene expression,

which, in turn, drives the T cell into activation. In addition,

intracellular Ca2+ signals play a crucial role in cellular prolif-

eration and induce the expansion of activated T cells.2,3 As the

dynamics of the Ca2+ influx upon TCR engagement are modu-

lated by the action of numerous entities, a wide variety of Ca2+

signals has been observed in T cells, ranging from isolated

transient spikes to repetitive oscillations and sustained

plateaus.4 Whether or not T cells are able to decode specific

Ca2+ patterns into a functional outcome is one of the immediate

goals in immunological research.1

Lab Chip, 2009, 9, 3517–3525 | 3517

Dow

nloa

ded

by L

akeh

ead

Uni

vers

ity o

n 08

Mar

ch 2

013

Publ

ishe

d on

08

Oct

ober

200

9 on

http

://pu

bs.r

sc.o

rg |

doi:1

0.10

39/B

9118

65A

View Article Online

For cell handling and manipulation, we employed a dielec-

trophoresis (DEP)-based microfluidic system.5 The electromag-

netic and hydrodynamic forces that are generated in this scheme

enable the manipulation of cells without the need of physically

contacting the cell membrane. The latter is not tolerable when

surface-mediated signal transduction processes are investigated.

Although other methods like optical tweezers6,7 and ultrasonic

manipulation8,9 have also been shown to allow for a competent

manipulation of cells, in our opinion, the features of DEP match

particularly well with those of hydrodynamic manipulation in

microfluidic systems. The combination of hydrodynamic10 and

DEP5,11 forces has been successfully used for sorting,12 sepa-

rating13 and patterning14,15 cells on a single-cell level. Although

contactless manipulation by these forces is minimally invasive, it

still may affect the viability of delicate cells. Thus, the careful

adjustment of the conditions for their in vitro manipulation is of

high relevance. In this context, exploring the suitability of cyto-

solic Ca2+ signals as a sensitive biomarker for studying the impact

of DEP on the cell physiology seemed to us a reasonable

approach.

In the present study, we stimulated individual T cells by con-

tacting them with anti-CD3/anti-CD28-coated microbeads. Due

to the high spatiotemporal resolution of the system, we achieved

a high control over the time point of the bead presentation. The

cytosolic Ca2+ level in the cells was monitored before, during and

after the contact formation procedure and related to the bead

stimulation at sub-second resolution. Diligent fluidic control

enabled us to release each of the manipulated cells from the

microfluidic system and to analyze both the expression of the

activation marker molecule CD69 and the cell division after over

night incubation in a microplate. At the same time, the identity of

each cell was preserved. This allowed us to correlate the analyzed

long-term response with the previously recorded Ca2+ signal for

each individual cell. In addition, the influence of the electric fields

on the cell physiology was investigated by recording cellular Ca2+

spiking prior to stimulation.

Materials and methods

Cells

Jurkat T lymphocytes E6.1 (ATCC, Manassas, VA, USA) were

cultivated at 5% CO2 and 37 �C in RPMI-1640 medium con-

taining 25 mM HEPES buffer, 2 mM stabilized glutamine, 0.1 g

l�1 gentamicin and 10% fetal calf serum (FCS). For single-cell

cultivation, conditioned medium was used that was obtained as

follows. T cells were grown for 3 days under sterile conditions

from an initial concentration of 105 ml�1. After that, the super-

natant was sterile filtered and mixed with fresh medium at

a mixing ratio of 2:1. Finally, 1 mM sodium pyruvate (Sigma)

was added. To examine T cell activation, the cells were stained

with 2 mg ml�1 anti-CD69-FITC (BD Biosciences) for 15 minutes

before they were analyzed in an LSM510 (Zeiss, Oberkochen,

Germany). T cells were counted as activated, if their outlines

were visible after image processing as described previously.5 The

viability of the chip-manipulated single cells was evaluated after

over night incubation in a microwell. Cells were counted as

damaged if they showed a non-spherical morphology or blebs.

Fluorescent antibodies in the cytoplasm made perforated cells

3518 | Lab Chip, 2009, 9, 3517–3525

appear as bright spots in the microscope image after live cell

staining.

Particles and surfaces

Ten-micron polystyrene beads were coated with anti-CD3 and

anti-CD28 antibodies as described before.5 Antibody-coated

microwells (BD Falcon) were made by first incubating them for

1 h with PBS containing 10 mg ml�1 Protein-A. Subsequently,

they were washed two times with PBS and blocked for 1 h with

0.1% BSA. After having been washed again, they were incubated

with 1 mg ml�1 anti-CD3 and anti-CD28 at 4 �C over night, and

washed three times before use.

Chips

The microfluidic systems employed were described earlier.5

Briefly, they were manufactured by sandwiching a 30 mm thick

SU-8 polymer between two glass slides so that the resist formed

the sidewalls of a microchannel (GeSiM mbH, Grosserkmanns-

dorf, Germany). Standard HPLC tubing connected inlets and

outlets of the fluidic system with syringe pumps (WPI, Berlin,

Germany) that allowed the control of the fluid flow in the

microchannels. Fluids, cells and particles were introduced into

the channel system through separate inlets. The dimensions of

the microchannel enabled hydrodynamic transport by laminar

flow. Thus, solutions, cells and particles move in parallel

trajectories.

The inner faces of the top and bottom slides of the channel

were equipped with congruent, 15 mm wide microelectrodes

(Fig. 1a). For dielectrophoretic manipulation, a radio-frequency

generator (Cytocon 400, Evotec Technologies GmbH, Hamburg,

Germany) was used to apply an a.c. voltage of 1–5 MHz to these

electrodes. This created field geometries that prevented cells and

particles from passing between them (Fig. 1b). Depending on the

shape of the electrode structure, particles were held against the

fluid flow (Fig. 1c) or shifted laterally.

Dielectrophoresis

Dielectrophoresis has been described earlier.16–18 In DEP, a force

acts on a dielectric particle in an inhomogeneous electric field.

The force is generated by the interaction of the external field with

induced polarization charges of the suspended object and

depends on particle size, electric field gradient and contrast of

conductivity and permittivity between the particle and the

surrounding medium. DEP forces can either be directed towards

(positive dielectrophoresis, pDEP) or away from the electric field

maxima (negative dielectrophoresis, nDEP). In cell culture

medium with a conductivity of 1.4 S m�1, cells and polystyrene

beads exhibit nDEP when subjected to a non-uniform a.c. field in

the lower MHz range and, thus, are repelled from the electrodes.

Particle manipulation

After having been introduced into the chip system, Fura-loaded

cells and beads were transported hydrodynamically to the central

part of the channel (Fig. 1a). Here, the deflection elements d1 and

d2 were used to guide cells and beads individually to the

successively arranged zigzag elements z1 and z2, where they were

This journal is ª The Royal Society of Chemistry 2009

Fig. 1 Particle manipulation in the chip system and analysis of the cytosolic Ca2+ level. (a) Top view of the central part of the microchannel. Fura-

loaded cells (open arrows) and beads (filled arrows) enter the chip through two separate inlets and move in parallel trajectories according to the fluid flow

from left to right. The active deflection electrodes d1 and d2 direct single beads and cells to the zigzag element z1 or z2, respectively. A subsequent

inactivation of z1 releases the bead from its position and allows contact formation with the T cell. After the manipulation and monitoring of the cytosolic

Ca2+ concentration, the cell–bead pair is directed to the exit of the system and deposited into a well of a microplate. Unprocessed particles are discarded

to the waste outlet. (b) Side view of the microchannel at the position marked with the rectangle in part (a). The top and bottom of the channel carry

corresponding electrode structures. Applying an electric field to both electrode layers prevents cells and particles from passing between them. (c)

Perspective drawing of the electrode structure z2 retaining a cell against the fluid flow (indicated by the white arrows). (d) Time series of micrographs that

show the contact formation procedure between an antibody-coated microbead and a Fura-loaded T cell held in the zigzag element z2 (see part (a)). The

cytosolic Ca2+ level strongly increases approximately 120 s after initiation of the cell–bead contact. Scale bar, 20 mm. (e) A typical course of the cytosolic

Ca2+ level of a bead-contacted T cell. Ca2+ traces were analyzed for the parameters maximum peak height, response latency and rising time (see Analysis of

the calcium traces section).

Dow

nloa

ded

by L

akeh

ead

Uni

vers

ity o

n 08

Mar

ch 2

013

Publ

ishe

d on

08

Oct

ober

200

9 on

http

://pu

bs.r

sc.o

rg |

doi:1

0.10

39/B

9118

65A

View Article Online

held against the fluid flow for 5 min. During that time, the

cytosolic Ca2+ concentration was monitored at 12 frames min�1

to estimate its baseline level. After subsequent deactivation of the

zigzag electrode z1 the bead hydrodynamically relocated towards

the zigzag element z2, where it came into contact with the cell

stored there. The time point of contact formation after the bead

release mainly depended on the flow rate in the microchannel

and, thus, could be controlled with high precision. The cytosolic

Ca2+ concentration was analyzed during and after the contact

formation procedure (at 1–2 frames s�1 over a period of 5 min),

so that it could be related with the bead stimulation at sub-

second resolution.

For manipulating the microscale objects, two different electric

and hydrodynamic parameter configurations were used: a high-

field (100 kV m�1; 14.4 ml h�1) and a low-field (40 kV m�1; 0.7 ml h�1)

condition. Electrodes and fluidics were driven at the high-field

condition to move or deflect cells and beads within the micro-

channel. In contrast, simple retaining of cells or cell–bead pairs in

the zigzag electrode for Ca2+ imaging was performed at the low-

field condition. To record a bead-induced Ca2+ trace, cells and

beads were typically manipulated as follows. For directing them

This journal is ª The Royal Society of Chemistry 2009

to the zigzag electrodes z1 and z2, respectively, they were

manipulated for several seconds at the high-field mode. Here, they

were held for 5 min at the low-field configuration while their

baseline Ca2+ level was recorded. After that, the configuration was

switched for ca. 20 s to the high-field mode in order to initiate the

cell–bead contact, followed by a 5 min observation of the cytosolic

Ca2+ level under low-field condition. Transporting the analyzed

cells to the exit of the microchannel again was performed under

the high-field condition. To test the influence of the electric field

exposure on the viability of the cells, two additional parameter

configurations were used. In these experiments, the cells were

either incubated in the chip without any or under prolonged

electric field exposure (field exposure, 100 kV m�1; 7.2 ml h�1).

Calcium imaging

Loading procedure. 2 mM Fura-2/AM (Invitrogen) were added

to a suspension of Jurkat T cells (ca. 106 ml�1, in culture medium)

followed by a 1 h incubation at room temperature in the dark.

Subsequently, the cells were centrifuged at 500 g for 5 min,

resuspended in culture medium and used within the next 2 h.

Lab Chip, 2009, 9, 3517–3525 | 3519

Fig. 2 Analysis of bead-stimulated T cells after their isolation from the

microfluidic system and over night incubation in a microwell. (a) Fluo-

rescence image of an activated T cell after staining with dye-labeled

antibodies against the activation marker CD69. Binding of the antibodies

to the membrane makes the cell appear as a bright circle in the micro-

scope image. (b) Corresponding brightfield image. (c) Image of a bead-

stimulated T cell that had divided during over night incubation. Scale

bar, 10 mm. (d) Quantification of the viability-, activation- and prolifer-

ation-states of the incubated cells. Note that evaluation of the viability

was done for all manipulated cells, while only viable cells were analyzed

for their activation- and proliferation-state.

Dow

nloa

ded

by L

akeh

ead

Uni

vers

ity o

n 08

Mar

ch 2

013

Publ

ishe

d on

08

Oct

ober

200

9 on

http

://pu

bs.r

sc.o

rg |

doi:1

0.10

39/B

9118

65A

View Article Online

Image acquisition. Fura-loaded cells were analyzed with a 40�Apochromat water immersion objective (UAPO40XW3/340,

Olympus, Hamburg, Germany) on a fully automated microscope

system (Cell-R, Olympus, Hamburg, Germany). Brightfield and

fluorescence images (excitation: 340/13 nm and 380/10 nm,

emission: 510/80 nm) were taken alternately at 0.2, 1 or 2 frames

s�1 for each image type. The UV illumination intensity was set as

low as possible in order to keep irradiation stress for the cells at

a minimum (intensity at the objective: 0.7 mW and 3.7 mW for

340 nm and 380 nm excitation, respectively. Exposure time:

50 ms frame�1).

Image processing. Fluorescence images were background-

subtracted before the F340/F380 ratio picture was calculated. On

the basis of a Fura-2 calibration in vitro (by use of the calcium

calibration kit F6774, Invitrogen), the pictures were further

processed according to Grynkiewicz et al.19 Cytosolic Ca2+ levels

were obtained by averaging all pixels of a cell.

Analysis of the calcium traces

The time point of contact formation was defined as the moment

when both particles acted like one aggregate in the fluid flow, as

observed in brightfield microscopy. For the determination of the

initial calcium rise, the three consecutively increasing values of

the Ca2+ trace with the highest ratio between first to last value

were identified. The interval between contact formation and

initial calcium rise was called response latency. The baseline of the

cytosolic Ca2+ level was detected by averaging the values from the

first five minutes of examination, i.e. before contact formation.

Prior to estimation of the maximum peak height, the traces were

baseline corrected. The rising time was defined as the interval

between the initial calcium rise and the maximum peak (Fig. 1e).

For the quantification of irregular Ca2+ spiking due to the electric

field exposure, the statistical variance of each Ca2+ trace was

calculated. For that, the square deviation from the mean calcium

level of a trace was estimated for each data point before the

average of all deviations was calculated.

Numerical modeling

For the numerical simulation of the electric field within the

microchannel we used the Quasi-Statics, Electromagnetics

module of the finite element software, COMSOL Multiphysics

(COMSOL, Burlington, MA, USA). We drew a section of the

microchannel surrounding zigzag electrode z2 as depicted in

Fig. 1c and set the conductivity and permittivity of the medium

within the box to 1.4 S m�1 and 78, respectively. The 3D distri-

bution of the electric field strength was calculated for voltages of

3.0 and 1.2 V between the electrodes. The mean field strength that

a cell experienced at a given distance d (seen later in Fig. 5b) was

evaluated by integrating the E-field over a 10 mm-diameter

sphere.

Results and discussion

Contacting T cells with antibody-coated beads triggered Ca2+

mobilization in the cells. The results of the contact formation

procedure are shown in Fig. 1. The intracellular Ca2+ concen-

tration was analyzed 5 min before and 5 min after bead

3520 | Lab Chip, 2009, 9, 3517–3525

presentation. A typical course of the cytosolic Ca2+ level is shown

in Fig. 1e: the first 5 min before stimulation, the Ca2+ concen-

tration is at baseline level (ca. 60 nM). After contact formation

with the functionalized bead, the Ca2+ concentration remains at

the baseline level for 1–2 min, until it rapidly rises to a concen-

tration of several hundred nM. After that the concentration

decreases slowly. The Ca2+ traces obtained were analyzed for the

highest value between the initial Ca2+ rise and the subsequent

decrease (maximum peak height), the time between contact

formation and initial calcium rise (response latency) and the time

between the initial calcium rise and the maximum peak (rising

time, see Fig. 1e and Materials and methods section). Contacting

the cells with uncoated beads did not change the cytosolic Ca2+

level (see ESI†).

After the calcium imaging, each of the bead-contacted cells

was transported to the exit of the channel where it was retrieved

from the microfluidic system and separately collected in a well of

a microplate.5 The collected cells were cultivated for 16–24 h and

afterward analyzed for viability.

Preliminary experiments had shown that Fura-loading did

not impair viability- or activation-rates of the T cells (see ESI†).

74% of all manipulated cells were viable the next day (Fig. 2d)

and, thus, were available for further analysis of their prolifer-

ation- and activation-states. For that, the cells were stained with

fluorescently-labeled antibodies against the membrane marker

protein CD69. This made activated cells appear as a bright

circle in the microscope image (Fig. 2a). The proliferation rate

of the cells was determined by counting the number of divided

cells among all analyzed (Fig. 2c). While 50% of all viable cells

were activated, the percentage of dividing cells was 44%

(Fig. 2d).

This journal is ª The Royal Society of Chemistry 2009

Dow

nloa

ded

by L

akeh

ead

Uni

vers

ity o

n 08

Mar

ch 2

013

Publ

ishe

d on

08

Oct

ober

200

9 on

http

://pu

bs.r

sc.o

rg |

doi:1

0.10

39/B

9118

65A

View Article Online

The system offers a high spatiotemporal resolution for the

analysis of the intracellular Ca2+ level of manipulated live cells.

Combining this with the highly efficient isolation of single cells

including their further cultivation allowed correlation of the

measured Ca2+ signal (short-term response) with later activation

events like proliferation and protein expression (long-term

response) on a single-cell level.

After analyzing the proliferation- and activation-states, the

information gained was correlated with the previously recorded

Ca2+ signals to identify an early indicator for the later response.

In this context, we compared the response latency, rising time and

peak height of dividing and activated cells with those of non-

dividing and non-activated cells, respectively. In the case of

activated vs. non-activated cells, a significant difference was

detected in the rising time (80 s vs. 51 s, p < 0.05, Student’s t test)

while no difference could be detected in the parameters response

latency and maximum peak height (Fig. 3a). In the case of

dividing vs. non-dividing cells, only a comparison of the

maximum peak height showed a significant difference between

both populations (500 nM vs. 380 nM, p < 0.05, Student’s t test,

Fig. 3b).

The use of a DEP-based microfluidic system for contacting

single T cells with antibody-coated microbeads guaranteed that

the bead contact represented a biochemically and temporally

Fig. 3 Correlation of short- and long-term responses to the applied bead

stimulus. Mean response latency, rising time and maximum peak height of

the cytosolic Ca2+ traces were evaluated for (a) activated and for non-

activated cells, and for (b) dividing and for non-dividing cells. Error bars,

s.e.m.; n $ 15; * p < 0.05 (Student’s t test).

This journal is ª The Royal Society of Chemistry 2009

highly defined stimulus. The analysis of the long-term response

required further cultivation of the stimulated cells. This was only

possible due to their successful recovery from the microfluidic

system after manipulation, since the prolonged incubation of

single cells in the microchannel caused cell damage (see no field

exposure condition in Fig. 4). Ca2+ traces were analyzed for the

parameters maximum peak height, rising time and response

latency (see Materials and methods section). In particular, the

analysis of the response latency required the control and moni-

toring the time point of stimulus presentation with high temporal

resolution.

We showed that the dynamics of the cytosolic Ca2+ level can

provide useful information about a later reaction of the cell

(Fig. 3): the rising time and the maximum peak height might serve

as an indicator for a later activation and proliferation of the cell,

respectively. In contrast, the response latency did not seem to

provide such information. This parameter has been previously

shown to depend on the conditions of the applied stimulus.20

Since every tested cell had received the same highly defined

stimulus, the mean response latency was in the same range for

activated and non-activated cells or for dividing and non-

dividing cells, respectively. However, in accordance with Wei

et al.,20 we detected a broad spectrum of response latencies

between individual cells (ranging from 4 s to 143 s in our

experiments, data not shown) which suggests that the value of

this parameter is not only a function of the applied bead stimulus

but also depends on the physiological state of the analyzed cell.

As response latencies are supposed to be mainly determined by

the time a given number of TCR-antibody complexes are formed,

the same approach could also be used for measuring cell-specific

membrane parameters like the TCR density.20

A very important question in context with the DEP manipu-

lation of live cells is whether the electric field exposure impairs

the cell physiology. Especially complex manipulation tasks like

the specific stimulation of single cells with functionalized

microbeads necessitates information about tolerable exposure

times in order to design the optimal experimental procedure. To

clarify this point, we used our DEP chip to direct single cells to

the zigzag electrode and to incubate them for 20 or 60 min under

no field exposure or field exposure condition (see Materials and

methods section for details), before they were moved to the exit

of the system and deposited into an antibody-coated microwell.

Alternatively, the selected cells were immediately directed to the

exit without retention in the zigzag (0 min). The manipulated

cells were cultivated for 16–24 h before their viability-, prolifer-

ation- and activation-states were analyzed.

Without an additional incubation step in the chip system,

DEP-manipulated single cells showed viability-, activation- and

proliferation-rates of 94%, 100% and 83%, respectively (Fig. 4,

0 min, n ¼ 18). After having been incubated for 20 min without

electric field exposure, the cells showed rates that were reduced in

respect to those obtained by just flushing the cells through the

chip (71%, 57% and 29%, n¼ 7). Incubating the cells for the same

duration under field exposure condition (n ¼ 7) did not decrease

these values further, albeit a slightly reduced viability rate (57%)

was detected. A different situation was observed after 60 min

incubation. Without electric field exposure, the tested parameters

(67%, 67% and 33%, n¼ 7) were in the same range like those after

20 min incubation time under no field exposure condition. In

Lab Chip, 2009, 9, 3517–3525 | 3521

Fig. 4 Impact of DEP manipulation on viability-, activation- and proliferation-rates of T cells. Single T cells were brought into the microchip and DEP-

sorted to the exit of the channel (0 min, n ¼ 18), where they were retrieved from the microfluidic system and deposited into antibody-coated microwells.

Where indicated, the sorting procedure within the chip was interrupted by an incubation step at the zigzag electrode under no field exposure or field

exposure condition for durations of 20 min (n ¼ 7) or 60 min (n $ 6). The isolated cells were incubated over night before the viability-, activation- and

proliferation-rates were analyzed. * p < 0.05 (Fisher’s exact test following logistic regression analysis, see ESI† for details).

Dow

nloa

ded

by L

akeh

ead

Uni

vers

ity o

n 08

Mar

ch 2

013

Publ

ishe

d on

08

Oct

ober

200

9 on

http

://pu

bs.r

sc.o

rg |

doi:1

0.10

39/B

9118

65A

View Article Online

contrast, a 60 min exposure to electric fields caused strongly

reduced viability-, activation- and proliferation-rates (29%, 14%

and 0%, n¼ 6). A logistic regression analysis of the data revealed

statistically significant different activation rates between cells

that were manipulated at field exposure or at no field exposure

condition (p < 0.05, see ESI†). Additionally, both activation- and

proliferation-rates were found to be influenced significantly by

the incubation time (p < 0.05, see ESI†). As post-hoc analysis,

Fisher’s exact test was performed to detect differences between

individual groups (see Fig. 4 for details).

Compared to the control group without prolonged electric

field exposure, an influence of DEP manipulation on the

viability-, activation- or proliferation-state of the cells was only

observed at exposure times above 20 min (Fig. 4). Nevertheless,

the manipulation times within the channel should be as short as

possible because the cells showed significantly reduced activa-

tion- and proliferation-rates even after their mere incubation in

the DEP chips. This considerable impact on the cellular physi-

ology may be due to the characteristics of the material used for

chip fabrication and is of serious concern that needs to be

addressed in future studies.

As described above, electric field exposure has been identified

to possibly affect physiological parameters like the viability-,

proliferation- and the activation-state of T cells, depending on

the exposure time. Ca2+ imaging experiments with DEP-manip-

ulated cells reveal if these fields also act on the level of the

cytosolic Ca2+ concentration. Negligible Ca2+ spiking was

observed in cells that had not been exposed to electric fields after

introduction into the microchannel (0 kV m�1; 0 ml h�1, Fig. 5a).

Cells retained at the zigzag electrode against the fluid flow

showed spontaneous Ca2+ spikes as shown in Fig. 5b, albeit these

spikes were clearly different from the signal elicited by contact

formation with an antibody-coated microbead (Fig. 5c). To

investigate the nature of these spontaneous Ca2+ spikes further,

the cytosolic Ca2+ level of Fura-loaded Jurkat T cells was

monitored while they were exposed to electric fields of different

strengths, frequencies and for different exposure times.

3522 | Lab Chip, 2009, 9, 3517–3525

For that, the cells were directed to a zigzag electrode and held

dielectrophoretically against the fluid flow at high-field or low-

field condition (100 kV m�1; 14.4 ml h�1 and 40 kV m�1; 0.7 ml h�1,

respectively) for 5 min. Alternatively, cells that had not been

exposed to electric fields (0 V; 0 ml h�1) were examined. At the

same time, their cytosolic Ca2+ concentration was monitored.

Spiking of the cells was quantified by calculating the statistical

variance of the recorded Ca2+ traces (see Materials and methods

section for details). As shown in Fig. 5d, the mean variance of the

analyzed Ca2+ traces strongly depended on the applied electric

field strength (Kruskal–Wallis test, c2¼ 23.38, df¼ 2, p < 0.001).

Without any electric field exposure, the cells showed only little

spiking, indicated by a mean variance of 12 nM2. In contrast,

manipulation at low- and high-field conditions led to significantly

higher values of 110 nM2 (p < 0.05) and 500 nM2 (p < 0.001),

respectively (multiple comparison post-hoc analysis after Krus-

kal–Wallis according to Siegel and Castellan21).

Next, we tested the influence of the exposure time on the

cytosolic Ca2+ level. In order to mimic the manipulation proce-

dure during bead stimulation (see Materials and methods

section), the cells were directed to a zigzag electrode and held

under the high-field condition against the fluid flow. After 15, 30,

or 60 s the mode was changed from high-field to low-field

condition for the rest of the trial period. Alternatively, cells were

held at the zigzag electrode over the complete trial period under

low-field (Fig. 5e, 0 s) or under high-field (Fig. 5e, 300 s) condi-

tion, respectively. Analysis of the obtained Ca2+ traces revealed

highly significant differences across all five groups (Kruskal–

Wallis test, c2¼ 25.52, df¼ 4, p < 0.001). The Ca2+ traces showed

mean variances of 50, 200, 80, 600 and 500 nM2 for 0, 15, 30, 60

and 300 s exposure times at high-field condition, respectively.

Multiple comparison post-hoc analysis according to Siegel and

Castellan21 indicated a statistically significant difference between

the 0 s and the 300 s condition (p < 0.001, Fig. 5e).

Electric fields affect cells in two ways: (i) due to ohmic

warming they induce thermal stress and, (ii) they alter the cellular

membrane potential as a function of the inverse of the field

This journal is ª The Royal Society of Chemistry 2009

Fig. 5 Influence of DEP manipulation on the cytosolic Ca2+ level of T cells. Fura-loaded T lymphocytes were held at the zigzag electrode for 5 min at

different parameter conditions for fluid flow and electric field strength (see text) while their intracellular Ca2+ concentration was monitored. For

quantitative analysis, the variance of each Ca2+ trace was calculated and averaged over all cells kept at a particular condition. (a) Exemplary brightfield

image and corresponding Ca2+ trace of a cell that was not, or (b) was exposed to electric fields. The distance d to the electrode is an important parameter

for estimating the electric field strength experienced by the cell. Distance d varies with the fluid flow and the applied electrode voltage. (c) Image of

a bead-contacted T cell and typical Ca2+ trace due to bead-stimulation. Scale bar, 10 mm. (d–f) Quantification of Ca2+ spiking in cells that were exposed to

electric fields at different strengths (d), for different exposure times (e) and at different frequencies (f) (see text). Error bars, s.e.m.; n $ 14; * p < 0.05; ***

p < 0.001 (Kruskal–Wallis test with multiple comparison post-hoc analysis according to Siegel and Castellan21). (g) Square deviations of the mean Ca2+

signal averaged over all cells that were exposed to high electric fields for 60 s (see part (e), 60 s). Cells were exposed to high electric fields only during the

first minute of the experiment. After that, the field strength was switched to a lower value. Note that variations in the Ca2+ signals mainly occurred after

exposure of the cells to high electric fields was over.

Dow

nloa

ded

by L

akeh

ead

Uni

vers

ity o

n 08

Mar

ch 2

013

Publ

ishe

d on

08

Oct

ober

200

9 on

http

://pu

bs.r

sc.o

rg |

doi:1

0.10

39/B

9118

65A

View Article Online

frequency. To clarify which of these two effects is responsible for

the irregular Ca2+ spiking, we manipulated the cells at different

field frequencies (1, 3 and 5 MHz). The manipulation procedure

was similar to the one used to expose the cells to high electric

fields for 60 s: the cells were directed to a zigzag electrode and

held under the high-field condition against the fluid flow. After

60 s, the parameters were changed from high-field to low-field

condition for the rest of the trial period. Fig. 5f clearly shows

a strong frequency dependency of the Ca2+ spiking. The mean

variance of the Ca2+ traces decreased from 600 nM2 (DEP

manipulation at 1 MHz) to 170 nM2 (3 MHz) and 130 nM2

(5 MHz). Interestingly, spiking of the cells does not occur

simultaneously with the exposure to the high electric fields.

Fig. 5g shows the square deviation from the mean Ca2+ level at

different time points, averaged over all tested cells of the 60 s

condition in Fig. 5e. Spiking of the cells mainly occurred after the

high field had already been switched off.

While short electric field exposure did not impair the viability-,

activation- or proliferation-rates of the cells, they showed irreg-

ular Ca2+ spiking after having been brought into a DEP field. On

This journal is ª The Royal Society of Chemistry 2009

closer examination, these non-specific Ca2+ responses turned out

to depend on the electric field strength, exposure time and field

frequency. In addition to the fact that the microchannel

temperature was not expected to exceed physiological values (see

ESI†), the latter observation suggests that spiking is caused by

E-field-induced shifts in the membrane potential and not by

temperature rises, because only effects on the membrane poten-

tial are frequency-dependent.22,23 Since T lymphocytes are not

supposed to possess voltage-gated Ca2+ channels,24,25 it is no

option that Ca2+ influx is caused by a direct activation of such

channels. In contrast, a plausible reason for transient Ca2+ influx

is electroporation of the cell membrane. Poration of biological

membranes occurs when transmembrane potentials reach a crit-

ical value so that membrane disruptions are induced. While

Glasser and Fuhr22 and Maswiwat et al.26 observed threshold

levels between 0.5 V and 1 V, Weaver reported pore formation at

transmembrane potentials of as low as 200 mV.27

To estimate the induced transmembrane potential in our

experiments, we first numerically calculated the E-field in the

microchannel at high-field and at low-field mode (Fig. 6 a,b).

Lab Chip, 2009, 9, 3517–3525 | 3523

Fig. 6 Quantification of the electric field experienced by the cells during DEP manipulation and its influence on the cellular membrane potential. (a,b)

Side view of the microchannel at zigzag electrode z2 (see also Fig. 1a and 1b). The electric field strength in the channel was numerically calculated for

electrode voltages of 1.2 V (a) and 3.0 V (b), respectively. Bright colors represent high electric field strengths. The distance of a cell (dashed circle) to the

electrode at low-field and at high-field mode was obtained from the analysis of the corresponding brightfield images. The asymmetry in the E-field is due

to the zigzag-shaped electrodes not extending normal through the paper plane. (c) Dependence of the electric field experienced by a cell on the distance

d between the cell and the zigzag electrode z2 (see Materials and methods section). (d) Dependence of the induced membrane potential shift on the field

frequency. The graph was calculated from eqn (1) with the electric field strengths for high-field and for low-field mode obtained from part (c) (see text).

Dow

nloa

ded

by L

akeh

ead

Uni

vers

ity o

n 08

Mar

ch 2

013

Publ

ishe

d on

08

Oct

ober

200

9 on

http

://pu

bs.r

sc.o

rg |

doi:1

0.10

39/B

9118

65A

View Article Online

Since the field strength decays somewhat more than proportional

with the distance to the electrodes, the position of the cell in the

microchannel has to be considered. A dielectrophoretically

retained cell is trapped at the position where hydrodynamic and

DEP forces balance each other. Hence, the distance d between

the cell and the electrode structure (Fig. 5b) depends on the flow

velocity in the microchannel and the applied electrode voltage.

As analysis of the brightfield images revealed, d was 13.3 mm and

18.9 mm at high-field and at low-field mode, respectively. Thus,

the average field strength that a cell experienced was 29.0 kV m�1

at high-field and 7.4 kV m�1 at low-field condition (see Materials

and methods section). For both field strengths, the correspond-

ing shift of membrane potential DJind was calculated according

to Glaser:28

DJind ¼ 1.5Er(1 + (2pfrC(s�1in + 0.5s�1

out))2)�0.5 (1)

with electric field strength E, field frequency f ¼ 1 MHz, cell

radius r ¼ 5 mm, membrane capacitance C ¼ 1 mF cm�2 and the

electric conductivities of the cytosol and the external medium sin

¼ 0.55 S m�1 and sout ¼ 1.4 S m�1, respectively.

The calculated shift in the membrane potential (which has to

be added to or to be subtracted from the resting potential of

approx. �30 mV,29 respectively) at the frequencies used was only

between 26 mV (5 MHz, low field) and 167 mV (1 MHz, high field,

Fig. 6d). Nevertheless, irregular Ca2+ spiking due to electric field

exposure was observed in all conditions. Apparently, the mean

transmembrane potentials were lower than the critical values

proposed above, but local shifts could have been higher,

3524 | Lab Chip, 2009, 9, 3517–3525

especially at areas close to the electrodes and with the field lines

entering the cell perpendicular to the membrane. Interestingly,

Ca2+ spikes due to high field exposure mainly occurred, after the

high-field mode already had been switched to low-field mode

again (Fig. 5g). This suggests that effects with a delayed response

time may be responsible for Ca2+ spiking upon DEP field expo-

sure. Since the formation of membrane pores is reported to occur

on much shorter time scales (i.e. in the range of several ns30), Ca2+

influx could also be due to biochemical signal transduction

processes triggered by voltage-sensitive proteins in the

membrane.31 Besides that, reactive oxygen species (ROS) created

at the electrodes32 could also be responsible for the observed

effects. The addition of catalase to the channel medium could

eliminate possibly created ROS and, thus, might provide more

information about the role of ROS in E-field-induced Ca2+

spiking.

One possibility to minimize the described side effects of DEP

manipulation is to drive the electrodes at high field frequencies.

Since – in the MHz range – this reduces DEP forces, we kept the

duration of the exposure of the cells to high electric fields (in the

bead stimulation experiments) at a minimum (below 30 s). This

ensured that irregular Ca2+ spiking remained moderate and

clearly discriminable from the typical bead-induced Ca2+ signal

(Fig. 5a–c).

In clinical research, working on physiologically relevant cell

models is strongly desirable. It is obvious that the T cells and the

microbeads used can be replaced with arbitrary mammalian cell

types. In addition, contact formation between more than two

cells or particles is possible. Thus, the same approach could be

This journal is ª The Royal Society of Chemistry 2009

Dow

nloa

ded

by L

akeh

ead

Uni

vers

ity o

n 08

Mar

ch 2

013

Publ

ishe

d on

08

Oct

ober

200

9 on

http

://pu

bs.r

sc.o

rg |

doi:1

0.10

39/B

9118

65A

View Article Online

used for the stimulation of, for example, primary T cells with

APC or the presentation of activating or inhibitory stimuli at

a controlled temporal regime while the reaction of the cell of

interest is monitored simultaneously. Additionally, the clustering

of different cells can form a specific microenvironment and

therefore could be useful for the in vitro regulation of stem cell

differentiation.

Conclusions

We presented a novel microfluidic device for triggering and

monitoring cell- or surface-mediated signal transduction

processes on a single-cell level. The system offers contactless

manipulation of cells and particles due to a combination of

hydrodynamic and DEP forces. Unrestricted optical access

provided the observation of cellular reactions like the cytosolic

Ca2+ concentration before, during and after contacting single

T cells with functionalized microbeads. After their manipulation,

the analyzed cells were retrieved from the system which made

them available for further cultivation. This allowed us to also

analyze the cellular response on the protein expression level. The

data obtained were then correlated with the previously recorded

Ca2+ signal on the level of individual cells. As a consequence, we

detected a relationship between the shape of the induced Ca2+

signal upon bead stimulation and the expression of the activation

marker molecule CD69. While a short-term DEP manipulation

did not impair the viability of the cells, their cytosolic Ca2+

concentration was influenced by the applied electric fields in

a frequency-dependent manner.

The possibility to precisely control the time point of stimulus

application, to simultaneously analyze short-term reactions and

to correlate them with late signaling events on the single-cell level

makes this approach unique among previously described appli-

cations based on hydrodynamic33,34 or optical6,35 forces. Hence, it

offers new possibilities to unravel the mechanisms underlying

intercellular communication and helps to understand the related

signal transduction processes in more detail which is of high

interest for example in cellular immunology or stem cell research.

Acknowledgements

We acknowledge financial support from the European

Commission in the framework of the integrated Project ‘Cell-

PROM’ (NMP4-CT-2004-500039) and from the German

Research Foundation (DFG, FU 345/12-1). We thank Richard

Kroczek (Robert Koch Institute, Berlin, Germany) for scientific

advice, Steffen Howitz (GeSiM, Grosserkmannsdorf, Germany)

for chip processing support and Beate Morgenstern (Fraunhofer

IBMT, Potsdam, Germany) for cell culture assistance.

References

1 A. Quintana, D. Griesemer, E. C. Schwarz and M. Hoth, PfluegersArch., 2005, 450, 1–12.

2 F. Macian, Nat. Rev. Immunol., 2005, 5, 472–484.

This journal is ª The Royal Society of Chemistry 2009

3 M. J. Berridge, BioEssays, 1995, 17, 491–500.4 R. S. Lewis, Annu. Rev. Immunol., 2001, 19, 497–521.5 M. Kirschbaum, M. S. Jaeger, T. Schenkel, T. Breinig, A. Meyerhans

and C. Duschl, J. Chromatogr., A, 2008, 1202, 83–89.6 E. Eriksson, J. Scrimgeour, A. Graneli, K. Ramser, R. Wellander,

J. Enger, D. Hanstrop and M. Goksor, J. Opt. A: Pure Appl. Opt.,2007, 9, S113–S121.

7 C. Reichle, K. Sparbier, T. Muller, T. Schnelle, P. Walden andG. Fuhr, Electrophoresis, 2001, 22, 272–282.

8 M. Wiklund, J. Toivonen, M. Tirri, P. Hanninen and H. M. Hertz,J. Appl. Phys., 2004, 96, 1242–1248.

9 M. Wiklund, C. Gunther, R. Lemor, M. Jager, G. Fuhr andH. M. Hertz, Lab Chip, 2006, 6, 1537–1544.

10 A. R. Wheeler, W. R. Throndset, R. J. Whelan, A. M. Leach,R. N. Zare, Y. H. Liao, K. Farrell, I. D. Manger and A. Daridon,Anal. Chem., 2003, 75, 3581–3586.

11 M. Stuke, K. Mueller, T. Mueller, R. Hagedorn, M. Jaeger andG. Fuhr, Appl. Phys. A: Mater. Sci. Process., 2005, 81, 915–922.

12 X. Y. Hu, P. H. Bessette, J. R. Qian, C. D. Meinhart, P. S. Daughertyand H. T. Soh, Proc. Natl. Acad. Sci. U. S. A., 2005, 102, 15757–15761.

13 M. Boettcher, M. Jaeger, M. Kirschbaum, T. Mueller, T. Schnelle andC. Duschl, Anal. Bioanal. Chem., 2008, 390, 857–863.

14 A. Rosenthal and J. Voldman, Biophys. J., 2005, 88, 2193–2205.15 N. Mittal, A. Rosenthal and J. Voldman, Lab Chip, 2007, 7, 1146–

1153.16 J. Kentsch, M. Durr, T. Schnelle, G. Gradl, T. Muller, M. Jager,

A. Normann and M. Stelzle, IEE Proc.: Nanobiotechnol., 2003, 150,82–89.

17 A. Ramos, H. Morgan, N. G. Green and A. Castellanos, J. Phys. D:Appl. Phys., 1998, 31, 2338–2353.

18 G. H. Markx, P. A. Dyda and R. Pethig, J. Biotechnol., 1996, 51, 175–180.

19 G. Grynkiewicz, M. Poenie and R. Y. Tsien, J. Biol. Chem., 1985, 260,3440–3450.

20 X. Wei, B. J. Tromberg and M. D. Cahalan, Proc. Natl. Acad. Sci.U. S. A., 1999, 96, 8471–8476.

21 S. Siegel and N. J. Castellan, Non parametric statistics for thebehavioural sciences, MacGraw Hill Int., New York, 1988.

22 H. Glasser and G. Fuhr, Bioelectrochem. Bioenerg., 1998, 47, 301–310.

23 M. S. Jaeger, T. Mueller and T. Schnelle, J. Phys. D: Appl. Phys.,2007, 40, 95–105.

24 L. Stokes, J. Gordon and G. Grafton, J. Biol. Chem., 2004, 279,19566–19573.

25 A. Badou, M. K. Jha, D. Matza, W. Z. Mehal, M. Freichel,V. Flockerzi and R. A. Flavell, Proc. Natl. Acad. Sci. U. S. A.,2006, 103, 15529–15534.

26 K. Maswiwat, D. Wachner and J. Gimsa, Bioelectrochemistry, 2008,74, 130–141.

27 J. C. Weaver, IEEE Trans. Dielectr. Electr. Insul., 2003, 10, 754–768.

28 R. Glaser, Biophysics, Springer-Verlag, Berlin, 2001.29 S. P. Fraser, J. K. Diss, L. J. Lloyd, F. Pani, A. M. Chioni,

A. J. George and M. B. Djamgoz, FEBS Lett., 2004, 569, 191–194.30 C. Chen, S. W. Smye, M. P. Robinson and J. A. Evans, Med. Biol.

Eng. Comput., 2006, 44, 5–14.31 F. Bezanilla, Nat. Rev. Mol. Cell Biol., 2008, 9, 323–332.32 X. J. Wang, J. Yang and P. R. C. Gascoyne, Biochim. Biophys. Acta,

Gen. Subj., 1999, 1426, 53–68.33 A. M. Skelley, O. Kirak, H. Suh, R. Jaenisch and J. Voldman, Nat.

Methods, 2009, 6, 147–152.34 S. Faley, K. Seale, J. Hughey, D. K. Schaffer, S. VanCompernolle,

B. McKinney, F. Baudenbacher, D. Unutmaz and J. P. Wikswo,Lab Chip, 2008, 8, 1700–1712.

35 S. Oddos, C. Dunsby, M. A. Purbhoo, A. Chauveau, D. M. Owen,M. A. Neil, D. M. Davis and P. M. French, Biophys. J., 2008, 95,L66–68.

Lab Chip, 2009, 9, 3517–3525 | 3525