You are on page 1of 3


Integrated multiple patch-clamp array chip via lateral cell

trapping junctions
J. Seo
Berkeley Sensor and Actuator Center, Department of Bioengineering, University of California, Berkeley,
Berkeley, California 94720
C. Ionescu-Zanetti
Neuroscience Research Institute, University of California, Santa Barbara, Santa Barbara, California 93106
J. Diamond
Berkeley Sensor and Actuator Center, Department of Bioengineering, University of California, Berkeley,
Berkeley, California 94720
R. Lal
Neuroscience Research Institute, University of California, Santa Barbara, Santa Barbara, California 93106
L. P. Leea)
Berkeley Sensor and Actuator Center, Department of Bioengineering, University of California, Berkeley,
Berkeley, California 94720
共Received 10 October 2003; accepted 31 December 2003兲
We present an integrated multiple patch-clamp array chip by utilizing lateral cell trapping junctions.
The intersectional design of a microfluidic network provides multiple cell addressing and
manipulation sites for efficient electrophysiological measurements at a number of patch sites. The
patch pores consist of openings in the sidewall of a main fluidic channel, and a membrane patch is
drawn into a smaller horizontal channel. This device geometry not only minimizes capacitive
coupling between the cell reservoir and the patch channel, but also allows simultaneous optical and
electrical measurements of ion channel proteins. Evidence of the hydrodynamic placement of
mammalian cells at the patch sites as well as measurements of patch sealing resistance is presented.
Device fabrication is based on micromolding of polydimethylsiloxane, thus allowing inexpensive
mass production of disposable high-throughput biochips. © 2004 American Institute of Physics.
关DOI: 10.1063/1.1650035兴

Patch-clamp recording has had a profound impact on quartz6 or glass7 substrates. Recently, three-dimensional
electrophysiology, by playing a crucial role in the character- structures more similar to patch pipettes have also been
ization of cellular ion channels. Traditionally, patch-clamp fabricated.8 All chip-based devices developed to date use the
recording is accomplished with a micromanipulator- planar geometry shown in Fig. 1共b兲, where the patch pore is
positioned glass pipette under a microscope.1 As illustrated etched in a horizontal membrane that divides the top cell
in Fig. 1共a兲, a cell membrane patch is sucked into the glass compartment from the recording electrode compartment.
pipette and forms a high electrical resistance seal. Current Here we present the design, fabrication, and character-
that passes through the ion channels in either the membrane ization of an integrated multiple patch-clamp array chip by
patch or the whole cell membrane is then recorded at differ- utilizing lateral cell trapping junctions, as shown in Fig 1共c兲.
ent bias voltages. The properties of ion channels are central This geometry dramatically reduces the capacitive coupling
to nervous systems and often act as targets for drugs.2 between the cell reservoir and the patch channel, an impor-
Despite constant improvements in the traditional patch- tant feature for low noise channel recording. Since the patch
clamp technique, it remains laborious and requires precisely channels are in the horizontal plane, multiplexed parallel
pulled pipettes to be placed in the vicinity of the cell by a patch sites that are only tens of ␮m apart are possible. In the
skillful operator using a micromanipulator under a micro- current design, the distance between patch sites is a few hun-
scope. Because of these requirements, the patch-clamp tech- dred ␮m 关Fig. 1共d兲兴. Channel binding drugs can therefore be
nique has not been widely used in proteomics and drug dis- administered in small volumes, while the effects on channel
covery development, which demand high-throughput activity can be recorded in parallel at a number of patch sites.
automated measurements. An automated patch-clamp setup The whole device is fabricated using micromolding of poly-
for high-throughput measurements using disposable devices dimethylsiloxane 共PDMS兲, a high-throughput, inexpensive
would eliminate the prohibitive investment of time of the procedure.
traditional patch-clamp, while maintaining its advantages The fabrication steps are presented in Figs. 2共a兲–2共f兲. A
over other measurements.2 Consequently, chip-based patch- silicon mold was prepared using surface micromachining
clamp devices have been proposed using silicon oxide coated techniques. First, 3.1 ␮m height patterns were made to define
nitride membranes,3 silicon elastomers,4 polyimide films,5 the narrow patch channels using deep reactive ion etching
关Fig. 2共a兲兴. Second, 50 ␮m high patterns were added for
Electronic mail: wide connection regions using SU-8 negative photoresist

0003-6951/2004/84(11)/1973/3/$22.00 1973 © 2004 American Institute of Physics

Downloaded 06 Dec 2008 to Redistribution subject to AIP license or copyright; see
1974 Appl. Phys. Lett., Vol. 84, No. 11, 15 March 2004 Seo et al.

FIG. 1. Comparison of patch-clamp setups: 共a兲 Traditional patch-clamp

based on a glass micropipette. 共b兲 On-chip planar patch clamp. 共c兲 Microar-
ray design with patch channels on the sides of a large central channel for cell
delivery. A section containing two patch sides is shown. 共d兲 Top view of the
patch-clamp array device 共optical microscope image兲 showing the center
channel and 14 radial patch channels. The connectivity of the reference
electrode and three of the patch electrodes is shown schematically. The
small circle indicates one of the patch sites. FIG. 2. Fabrication of the patch-clamp microarray. The Si etch 共a兲 is used to
define the patch channels 共4 ␮m⫻3.1 ␮m兲, while SU-8 photoresist 共b兲 is
used to define the large access channels 共50 ␮m high兲. After PDMS is cured
共c兲 the devices are detached and mechanically punched. The 共d兲 devices and
关Fig. 2共b兲兴. After a base and a curing agent of PDMS were the 共e兲 glass substrate precoated with a thin PDMS layer are treated with
mixed 共1:10兲, the liquid mixture was then poured onto the oxygen plasma. 共f兲 The devices are bonded to the thin PDMS. SEM images
mold and cured at 80 °C for 1 h. Scanning electron micros- of the overall device geometry before bonding 共upside down兲 and a closeup
of the patch pore after bonding are shown in 共g兲 and 共h兲, respectively. 共i兲
copy 共SEM兲 images of the fabricated device are shown in
SEM image of the mold.
Figs. 2共g兲 and 2共h兲. Despite the fact that fabrication should
result in a square patch orifice, we observe that the top of the
orifice is rounded. Rounding of the top of the orifice is a trode in the main channels and the patch electrode in the
beneficial artifact of mold fabrication, and we observe the lateral patch channel was confirmed by applying a 20 mV
channel top is rounded next to the patch orifice in the mold square pulse and recording the current response. A typical
关Fig. 2共i兲兴. When SU8 is selectively polymerized in order to channel current response is shown in Fig. 4共a兲, indicating a
create the large channels on top of the small patch channel channel resistance of 10–14 M⍀. This is comparable to the
defined in Si, light scattering near the Si surface results in access resistance of traditional micropipettes, but can be re-
deviation from the intended vertical SU8 wall. The resulting duced by altering the channel geometry. After dissociation by
rounded feature at the bottom of the SU8 wall 关Fig. 2共i兲, trypsin treatment, cells were suspended in PBS and injected
arrow A兴 is also present on top of the small Si wall 关Fig. 2共i兲, into the main channel. Gentle pressure 共1 psi兲 was applied to
arrow B兴, resulting in rounding of the top of the patch orifice. the patch channel while cells were loaded into the main flu-
This feature is reproducible since it is observed to be part of idic channel in order to prevent contamination at the patch
the mold geometry at every patch orifice. For fluidic connec- site. A cell can either be trapped randomly or selectively by
tions to outside tubing, 0.5 mm holes were punched me- controlling the flow through the main fluidic channel. A cell
chanically into the cured detached PDMS device. The device found within 100–200 ␮m of the patch channel opening can
was subsequently bonded to a thin PDMS layer which was be trapped within a 1 s time interval by applying 2 psi of
spin cast and then cured onto a glass substrate. Finally, plas- negative pressure to the patch channel 共Fig. 3兲. Right after
tic tubes were connected to the reservoirs, via punched holes, trapping the cell, negative pressure was removed and the cell
to load both cells and electrolyte solutions and to apply suc- was allowed to form a seal with the rim of the patch channel.
tion to the patch channel. The top down view allows effective visualization of the
A human tumor cell line 共HeLa兲, 12–17 ␮m in diameter, membrane protrusion into the patch channel.
was used for seal resistance experiments. Before introducing Patch resistance was recorded by applying a square volt-
the cells, the fluidic network was filled with phosphate buff- age pulse of amplitude 20 mV and 50 ms duration. The cur-
ered saline 共PBS兲, taking care to expel all air bubbles. The rent response was recorded using a standard patch-clamp am-
electrical connection between the reference Ag/AgCl elec- plifier 共Dagan PC-ONE, Minneapolis, Minnesota兲 and low
Downloaded 06 Dec 2008 to Redistribution subject to AIP license or copyright; see
Appl. Phys. Lett., Vol. 84, No. 11, 15 March 2004 Seo et al. 1975

FIG. 3. 共a兲–共c兲 Three sequential images showing a HeLa cell being trapped
by applying negative pressure 共2 psi兲 to the patch channel. The third frame FIG. 4. Device current response to a 20 mV voltage pulse 共a兲 before and 共b兲
is magnified in order to show cell positioning on the patch pore. 共d兲 Real after cell trapping. In 共a兲, the channel impedance is 14 M⍀, while in 共b兲 the
time observation of cell membrane deformation. average seal resistance is 144⫾3 M⍀.

chip via reliable lateral cell trapping junctions is presented.

pass filtered at 1 kHz. The current response presented con- The intersectional design of the microfluidic network pro-
tains no capacitance compensation. The resistance of the vides instantaneous multiple cell addressing. The geometry
open patch channel was measured to be 14⫾4 M⍀. The of this device not only minimizes capacitive coupling be-
channel geometry 共4 ␮m⫻3.1 ␮m⫻200 ␮m兲 and the con- tween the cell reservoir and the patch channel, but also al-
ductivity of the electrolyte used 共␴⫽1 S/m兲 yield a calcu- lows simultaneous optical and electrical characterizations for
lated resistance of 17 M⍀, in reasonable agreement with the studying roles of ion channels with respect to cellular func-
measurement. Capacitive coupling leads to a spike in current tions. The device geometry, together with the low dielectric
when the bias voltage is first applied 共Fig. 4兲. Integrating constant of PDMS, results in very low capacitive coupling
spike currents gives an approximation of the charge stored in between the cell reservoir and the patch channel: C predicted
the capacitor, q⫽ 兰 Idt. Capacitance can then be calculated ⫽0.4 fF and C measured⭐1 pF. The lateral design also allows
by using C⫽q/V. This capacitance measurement method efficient multiplexing of patch measurements, exchange of
yielded apacitance of 10⫾1 pF for connections between the intracellular electrolytes while the cell is attached to the
device and the patch-clamp amplifier input, but showed no patch pore, and optical observation of membrane deforma-
further increase in capacitance when the device itself was tion. While measured seal resistances are not yet high
attached. We can conclude that the device capacitance is enough for single channel measurements (R⫽140
within measurement error, or C dev⭐1 pF. Our calculations, ⫾20 M⍀), they should be sufficient for loose patch mea-
using the device geometry and ␧ PDMS⫽2.46,9 yielded a pre- surements. Future improvements in PDMS surface treatment
dicted device capacitance of C dev⫽0.5 fF. By comparison, and patch pore geometry should lead to increases in seal
capacitances for micromachined patch-clamp devices are 30 resistance. This device has the potential for high-throughput,
pF for micronozzle devices8 and 1 pF for glass substrates,6 low cost cell-based patch-clamp measurements.
while micropipette capacitances are in the range of 2 pF
⭐C pipette⭐20 pF.
Cell trapping by suction was described earlier in this 1
B. Sakmann and E. Neher, Single Channel Recording 共Plenum, New York,
letter. The current response from the cell by a 20 mV/50 ms 1983兲.
current pulse is shown in Fig. 4共b兲. The calculated sealing J. Xu, X. B. Wang, B. Ensgn, M. Li, A. Guia, and J. Q. Xu, Drug Dis-
covery Today 6, 1278 共2001兲.
resistance after attaching the cell was 140 M⍀. Typical seal 3
N. Fertig, A. Tilke, R. H. Blick, J. P. Kotthaus, J. C. Behrends, and G. ten
resistances were in the range of 140⫾20 M⍀, while 200 M⍀ Bruggencate, Appl. Phys. Lett. 77, 1218 共2000兲.
was the highest seal resistance obtained. By applying posi- K. G. Klemic, J. F. Klemic, M. A. Reed, and F. J. Sigworth, Biosens.
Bioelectron. 17, 597 共2002兲.
tive pressure to the patch-clamp channel, the trapped cell was 5
A. Stett, V. Bucher, C. Burkhardt, U. Weber, and W. Nisch, Med. Biol.
expelled from the channel. As soon as the cell was expelled, eng. Comput. 41, 233 共2003兲.
the current response returned to that of the open channel. 6
N. Fertig, R. H. Blick, and J. C. Behrends, Biophys. J. 82, 3056 共2002兲.
Subsequent cell trapping in the same channel resulted in N. Fertig, M. Klau, M. George, R. H. Blick, and J. C. Behrends, Appl.
Phys. Lett. 81, 4865 共2002兲.
lower seal resistance, presumably due to contamination at the 8
T. Lehnert, M. A. M. Gijs, R. Netzer, and U. Bischoff, Appl. Phys. Lett.
opening of the patch channel. 81, 5063 共2002兲.
In summary, an integrated multiple patch-clamp array 9
Z. Lin, T. Kerle, and T. P. Russel, Macromolecules 35, 3971 共2002兲.

Downloaded 06 Dec 2008 to Redistribution subject to AIP license or copyright; see