- Email: [email protected]
Microfluidics for cell-based assays
Microfluidics for cell-based assays Microfluidic systems are powerful tools in chemistry, physics, and biology. The behavior of confined fluids differs from macroscopic systems and allows precise control of the chemical composition of fluids, as well as the temporal and spatial patterns of microscopic fluid elements. Cell-based assays can benefit when adapted to microfluidic formats, since microscale systems can offer low sample consumption, rapid application of well-defined solutions, generation of complex chemical waveforms, and significantly reduced analysis or experiment time. Johan Pihl1, Jon Sinclair2, Mattias Karlsson1, and Owe Orwar2* 1Cellectricon AB, Fabriksgatan 7, SE-412 50 Gothenburg, Sweden 2Department of Chemistry and Bioscience, Chalmers University of Technology, SE-412 96 Gothenburg, Sweden *E-mail: [email protected]
There has been enormous development in the field of microfluidic
systems, while consuming much smaller amounts of chemicals and
systems over the past decade1-3. It has grown from a specialized
solvents. In addition, the low dimensions of microfluidic systems
area of research into a field engaging hundreds of research groups
enable access to phenomena that are not accessible on the macroscopic
and a growing number of companies. The development of
scale under normal conditions, i.e. the absence of inertia results in
microfluidic systems was made possible by fabrication technologies
laminar, turbulence-free flow and all that brings. This results in
developed in the microelectronics industry. This began in the late
entirely new ways of acquiring chemical, biological, and physical
1970s, when Terry et
al.4,5
demonstrated a gas chromatography
system fabricated on a Si wafer, but received little attention until 1990, when Manz et
al.6
published the now classic paper on
The idea of using living cells as sensor elements in microfluidic systems is attractive for obtaining multiparametric information from
miniaturized total analysis systems (µTAS). This resulted in a
biological systems. Besides the general benefits of miniaturization, the
tremendous increase in the research of chip-based microfluidic
superior control and unique flow properties in microfluidic systems
systems, much like the revolution in the electronic industry after
enable the extraction of high-quality data from cell-based biosensors.
the invention of the transistor and the integrated circuit.
Consequently, microfluidics will reduce costs and increase throughput
Correctly designed microfluidic systems possess the capability to execute operations more quickly than conventional, macroscopic
46
information.
DECEMBER 2005 | VOLUME 8 | NUMBER 12
and data quality in cell-based assays in the drug-discovery process, as well as in fundamental biological research.
ISSN:1369 7021 © Elsevier Ltd 2005
Microfluidics for cell-based assays
REVIEW FEATURE
Microfluidics The term microfluidics refers to devices, systems, and methods for the manipulation of fluid flows with characteristic length scales in the micrometer range7. Microfluidic systems display fundamentally different properties from everyday perceptions of how fluids behave. The ratio between the moment of inertia and the viscous forces in a fluid system is described by the Reynolds number Re = ρνl / η , where ρ, ν, l, and η are the fluid density, fluid velocity, characteristic length scale, and dynamic viscosity, respectively. Microfluidic systems are typically low Reynolds number systems, and the fluid flow is governed mainly by viscous forces and pressure gradients, with a low moment of inertia. The result is a truly laminar, turbulence-free flow and, since low Reynolds number flows are time independent, flow patterns are reversible8. The short length scales involved in microfluidic systems result in other properties, such as high surface-to-volume ratios, small diffusion
Fig. 1 The principle of microreplication. (A) The master structure is a negative of the desired structure and is manufactured by standard microfabrication or micromachining techniques. (B) The master is replicated in thermoplastic or elastomeric materials by molding, imprinting, or casting. (C) The replicate is unmolded from the master, and bonded to a flat base substrate (D).
distances, and small heat capacities. The surface tension of the fluid and the wetting properties of the system can also be significant forces9. The behavior and relative influence of these phenomena will differ
techniques resides in the high total cost of the device, dependent on the cost of both the material itself and the method of fabrication.
greatly depending on the size of the system studied and, in microfluidic
Traditional micromachining techniques are not widely used today in the
systems, the result can be highly counterintuitive. Specifically, the
fabrication of microfluidic devices for either high-volume production or
difference in physical behavior between microscopic and macroscopic
research purposes, with some exceptions. The microfluidics community
systems allows for the construction of functionalities that are difficult
has instead driven the development of alternative fabrication
or even impossible to access on the macroscopic scale. Taking all this
technologies and materials, including powderblasting of glass17,18;
into account, one arrives at the conclusion that, rather than trying to
injection molding19,20, hot embossing19,21, and laser ablation22 of
design a microfluidic system that is just a downsized copy of a
polymeric materials such as polycarbonate and polymethylmethacrylate;
macroscopic system, one should use design rules obtained from the
and casting techniques such as rapid prototyping in silicone elastomers
physics of fluid mechanics and diffusion in confined spaces. Examples of
or epoxies23-25. Apart from powderblasting and laser ablation, these
microfluidic systems designed in this manner include on-chip gradient
methods all rely on the replication of a master structure that is the
generation devices, where solutions are diluted in an automated and
negative of the desired structure (Fig. 1). The master structure is
highly reproducible manner10-12. Such systems have been applied in
reusable, enabling replication of microfluidic devices with a relatively
chemotaxis and electrophysiology
studies12,13.
The
T-sensor14-16,
where
low cost. The techniques allow the fabrication of microfluidic devices in
particles and/or molecules with different diffusion coefficients can be
a large selection of thermoplastic and elastomeric materials that have a
separated for analytical or preparative purposes, is an example of a
wide range of chemical, mechanical, and optical properties. Furthermore,
sensor that takes advantage of the predictable diffusion behavior of
the surface of the materials can be modified – the majority of the
molecules and particles in laminar flows.
thermoplastic materials permanently – by physical as well as chemical methods, giving access to an almost unlimited variety of surface
Fabrication of microfluidic devices
properties. This is important in many cases, both for minimizing
Developments in microfluidics can be attributed to the adoption of
adsorption of chemical species such as hydrophobic molecules and
microfabrication technologies from the microelectronics industry,
proteins to surfaces20,26 (a common problem in microfluidic devices
enabling the creation of complex miniaturized systems in materials such
because of their intrinsically high surface-to-volume ratios), as well as
as Si, glass, and metals. As a consequence, microfluidic devices were
tailoring surface chemistries for cell growth and control27,28.
initially composed of Si or glass. These materials possess several advantages, such as excellent chemical and physical stability, solvent
Cell-based assays and microfluidics
compatibility, and optical properties, along with generally very desirable
The term ‘cell-based assay’ was introduced in the early 1980s and has
surface properties. The main disadvantage of traditional fabrication
grown to encompass many different areas in assessing the effects of
DECEMBER 2005 | VOLUME 8 | NUMBER 12
47
REVIEW FEATURE
Microfluidics for cell-based assays
chemical stimuli on biological cells. Today, the most widespread use is
which large numbers of compounds can be synthesized by
in the pharmaceutical industry, where cell-based assays are used in
combinatorial chemistry38. If compounds are to be screened with higher
various steps in the drug-discovery process. In identifying leads, large
throughput at lower costs, miniaturization is an obvious route,
compound libraries are screened against targets expressed in
resulting in lower volume consumption of chemicals, solvents, and
immortalized cell lines. For example, G-protein coupled receptor
cells39.
screening can be achieved by monitoring the intracellular Ca2+ concentration using
fluorescence29.
Furthermore, different types of cell-
Caliper Life Sciences has developed a microfluidic system that can be employed in various types of high-throughput screening applications40.
based absorption, distribution, metabolism, excretion, and toxicity
The Caliper platform carries out assays that traditionally used
(ADME/Tox) assays are carried out to determine whether the acquired
conventional, plate-based screening platforms. The key component is a
leads have the necessary pharmacokinetic properties and to assess their
glass microchip with integrated ‘sipper’ capillaries that ‘sips’ fluids,
studies30,
assessment of metabolic
e.g. compounds, dyes, buffers, or negative controls from plate wells,
stability and clearance, and screening for QT interval prolongation using
while at the same time continuously drawing solutions, e.g. enzyme,
toxicity. These include permeability fluorescence-based assays or in vitro
electrophysiology31,32.
The driving force for the development of cell-based assays is a desire
substrate, or cell suspensions from integrated, on-chip wells (Fig. 2). The resulting mixtures are transported downstream in a microchannel, which
to develop tests that provide data representative of higher-level
also serves as an incubation chamber, to a detection point where the
biological responses, i.e. results should ideally be comparable to those
fluorescence signal is recorded. This process is performed in a serial
obtained from animal models or clinical trials. Often, cell-based assays
manner, in the sense that each chip sips a large number of cell
can collect new types of information impossible to obtain by other
suspensions, compounds, and dyes alternated with buffer solution to
means. Importantly, implementation of microfluidics technologies into
flush the system. In order to increase the speed of the system, each chip
cell-based assays can provide significant advantages in throughput,
consists of four or 12 units, and the system is capable of performing
costs, and data quality. However, conventional assays have been vastly
assays with considerably higher throughput and better reproducibility
optimized and miniaturized, e.g. 3456 well plates are routinely used33.
than conventional, plate-based screening platforms. In addition, the
Although several examples of cell-based assays in microfluidic systems
chips consume far smaller amounts of chemicals and solvents than
have been demonstrated34-37, there are only a few applications where
conventional systems. The system integrates a majority of the steps
they have proved to offer advantages over conventional assays: high-
that are typically carried out on different stations in a conventional
throughput screening, ion channel research, and chemotaxis studies.
screening platform, resulting in a smaller footprint. The system has turned out to be a genuine success, and is currently used by a large
High-throughput screening
number of pharmaceutical companies in high-throughput screening
Pharmaceutical companies worldwide own extensive and growing
applications. Examples include everything from enzymatic assays to
libraries of different compounds, mainly because of the ease in
cell-based calcium flux assays41,42.
Fig. 2 (A) Schematic of Caliper Life Sciences’ FS-417 four-sipper cell chip for detecting agonist-induced calcium flux. Dye-loaded platelet samples are drawn from the cell wells and continuously streamed through the chip. Platelets are exposed to agonist aspirated from a 384 well microtiter plate at the sipper junction and mixed by diffusion while the sample flows continuously through the chip past the fluorescence detection zone. The flow rate depends on a combination of the applied vacuum and sample viscosity. (B) The microfluidic chips are designed to flow four samples through the detection zone three times simultaneously, allowing 15 s, 30 s, and 60 s incubation intervals. The time to detection is defined by the flow rate and the distance from the sipper to the detection zone. (Adapted from42. © 2005 Elsevier.)
48
DECEMBER 2005 | VOLUME 8 | NUMBER 12
Microfluidics for cell-based assays
REVIEW FEATURE
Fig. 3 The patch clamp technique. After a cell has been approached by the pipette (A), a high-resistance seal is achieved on application of negative pressure, resulting in the cell-attached configuration (B). Further application of negative pressure ruptures the membrane, resulting in the whole-cell configuration, i.e. electrical contact with the inside of the cell (C). Inside-out and outside-out configurations are achieved by pulling the pipette away from the cell (D and E). (F) An equivalent electrical circuit of a cell in whole-cell configuration during acquisition of data. These physical parameters set the boundary conditions for temporal resolution and the errors obtained during the recording. Em denotes the membrane potential, Rs the series resistance, Cp the pipette capacitance, Cm the membrane capacitance, and Rm the membrane resistance.
Ion channel studies Ion channels are proteins that span the plasma membranes of cells and
place by molecular diffusion alone48. The device is placed on a
organelles and selectively control the diffusive transport of ions across
computer-controlled scanning stage mounted on an inverted
the membrane. Ion channels are one of the fundamental components in
microscope. Cells are patch-clamped in the open volume, translocated
cell-to-cell communication and are involved in everything from the
to the channel exits, and scanned across the different solution
contraction of muscles to the function of the brain, making them highly
environments with full control of transfer and exposure times. The
interesting in many areas of biology. Consequently, the malfunction of
system rapidly generates dose-response curves under extremely stable
ion channels is responsible for a number of diseases, making them
conditions, and several dose-response scans can be obtained from the
important drug targets. The gold standard for investigating ion-channel
same cell. In addition, the well-defined force that originates from the
activity is the patch clamp technique, an electrophysiological method
flow and acts on the cell-pipette system stabilizes the patch-clamped
(Fig. 3)43-45. It can provide single-channel resolution along with a
cell, allowing extended recording times from one cell, in some cases up
temporal resolution in the microsecond range. However, it is labor-
to hours47. The throughput is 10-100 times greater than conventional
intensive, notoriously low in throughput, and limited by the number of cells that can be patch clamped per unit time, making it difficult to generate statistically significant amounts of data. One approach to increase throughput is to increase the number of data points obtained per patch-clamped cell. This can be achieved by adding microfluidics to conventional patch clamp experiments12,46,47, a setup developed by Orwar and coworkers and commercialized by Cellectricon. The system is built around a microfluidic device fabricated in a silicone elastomer and glass (Figs. 4A and 4B). The devices comprise of a number of sample wells connected to an open volume through microfluidic channels. At the exit to the open volume, the channels are tightly packed. Fluids are pumped through the channels and into the open volume by applying pressure to the sample wells. Flows emerging from the channels couple viscously, forming a single laminar flow that propagates in the open volume (Fig. 4C). This flow contains a number of discrete zones with well-defined chemical environments corresponding to the loading of the sample wells and, because of the low Reynolds number of the system, mixing between adjacent environments takes
Fig. 4 Principle of the cell-based bar-code reader. (A) Overview of a 16 channel device, where 16 sample wells are connected to an open volume by individual microchannels. (B) A pipette with a patch-clamped cell situated outside the channel outlets. The cell can be exposed to different solution environments, e.g. different ligands can be alternated with buffer solutions in between exposures by translation of the device using a motorized scanning stage. The different solution environments can be thought of as a bar code that is read by the cell. (C) Fluorescence micrograph showing the collimated stream of different dye solutions at the exit into the open volume.
DECEMBER 2005 | VOLUME 8 | NUMBER 12
49
REVIEW FEATURE
Microfluidics for cell-based assays
Fig. 5 Patch clamp array on a microfluidic platform. (A) The main chamber containing cells in suspension is connected to a set of recording capillaries. Cell trapping is achieved by applying negative pressure to the recording capillaries. Patch clamp measurements are obtained by placing AgCl electrodes in each of the capillaries, as well as in the main chamber. Signals are fed through a multiplexing circuit and into the data acquisition system. (B) Scanning electron micrograph of three recording capillary orifices as seen from the main chamber. The capillary dimensions are approximately 4 µm x 3 µm, with a center-to-center distance of 20 µm. (C) Fluorescence micrograph of cells trapped at three capillary orifices. The entire device consists of 12 capillaries, with six capillaries arrayed along each side of the main chamber fluidic channel. (Adapted from53. © 2005 National Academy of Sciences.)
Fig. 6 Schematic of a gradient-generating device and the experimental setup. (A) Top view of the entire device consisting of the gradient generator and the observation area. (B) Three-dimensional representation of the observation area, in which cells can be exposed to different gradients of chemoattractants. (Adapted from13. © 2002 Nature Publishing Group.)
environment around cells13 (Fig. 6). A stable time-independent gradient is created by the microfluidic network seen in Fig. 6A. In its simplest form, solutions with and without chemoattractants are introduced into the network. As the two solutions flow down the network, they are
systems, sample consumption is greatly reduced, and there is no risk of
repeatedly split, recombined, and mixed, resulting in an increasing
cross-contamination, since the devices themselves are disposable. The
number of channels and concentrations. At the end of the network, the
system has found use in various stages of the drug-discovery process as
channels are brought together, giving a concentration gradient spanning
well as in studies of receptor
functionality49-51.
a single, wide microchannel. At this point, cells can be introduced and
Another, exclusively microfluidics-based approach to increase throughput has been developed by Lee and
coworkers52,53
(Fig. 5). Cells
their behavior studied. This technique was used by Jeon and coworkers to study chemotaxis of neutrophils in gradients of the chemoattractant
are patch-clamped in a microfluidic channel ~2 µm in width and ~2 µm
interleukin-8. This initial demonstration has resulted in the use of
in height. The cells are delivered from sample reservoirs through
microfluidics to study chemotaxis in different research areas62-66, and
microfluidic channels to an array of small channels where the cells are
offers the possibility of screening for drugs that modulate chemotaxis.
trapped and brought to whole-cell configuration by the application of a vacuum. Drugs are delivered through the same microchannel as the
Emerging technologies
cells, but from different input wells. While this concept does not provide
Apart from the assays mentioned above, there are few examples of
the same data quality as the conventional patch-clamp technique, it has
cell-based assays in a microfluidic format that exhibit unique
potential for massive parallelization and takes full advantage of
advantages over conventional approaches. However, there are a
microfluidics in order to achieve a fully automated patch-clamp device.
number of specialized assays and technologies that are very interesting and could become key components in tomorrow’s microfluidic cell-
Chemotaxis
based assays.
Chemotaxis is the directional migration of cells in response to chemical gradients of molecules called chemoattractants. The process is crucial in numerous biological
50
processes54,55,
and several approaches, e.g. different
Chiu and coworkers67 have demonstrated a microfluidic model for the obstruction of capillary blood vessels by single red blood cells infected with the Plasmodium falciparum parasite, as occurs in severe
types of chambers or puffer pipettes, have been developed to study
malaria. Capillary blood vessels were modeled using microfluidic
chemotaxis56-61. Most of the methods are nonideal in that the
channels fabricated in poly(dimethylsiloxane), and the transport of
generated gradients are created in macroscopic environments, are
infected and uninfected red blood cells through the channels was
nonlinear, and change with time in an uncontrolled manner. By making
studied. An artificial synapse chip that simulates synaptic transmitter
use of chip-based gradient generation10,11, chemotaxis studies can be
release has also been developed68. These models of biological processes
carried out with precise spatial and temporal control of the chemical
would be impossible without the aid of microfluidics.
DECEMBER 2005 | VOLUME 8 | NUMBER 12
Microfluidics for cell-based assays
A multilayer soft lithography technique has been developed by Quake and coworkers69 for fabricating pumps and valves in elastomeric materials. It has been used to make cell
sorters70
and could prove useful
in the construction of powerful cell-based assays in microfluidic formats. Takayama and
coworkers71
have avoided the relatively complex and
REVIEW FEATURE
of functional components, e.g. pumps and valves, as well as new, costeffective fabrication technologies. As the focus now shifts to applications, completely new types of cell-based assays will result in information that was previously impossible to acquire, as shown in the examples of high-throughput screening, patch clamp, and chemotaxis
expensive multistep microfabrication process of multilayer soft
studies. Although there are many good proposals for the use of
lithography in making cell-cultivation devices that operate using
microfluidics in cell-based assays, the field has not yet matured to
elastomeric channels and Braille displays. Their system has been used
present commercial functional devices apart from those mentioned
for long-term cultivation of endothelial cells and observation of their
above. To conclude, the potential of microfluidics will undoubtedly see
behavior in response to shear stress.
the technology assume a larger and more important role in the development of cell-based assays for the creation of more and higher
Future outlook
quality targets and leads in drug discovery. Microfluidics will also
The field of microfluidics is still in its infancy and, until recently, has
generate discoveries unattainable using conventional, macroscopic
been highly technology driven. The focus has been on the development
approaches.
REFERENCES
36. Wheeler, A. R., et al., Anal. Chem. (2003) 75 (14), 3581
1. Reyes, D. R., et al., Anal. Chem. (2002) 74 (12), 2623
37. Poulsen, C. R., et al., Anal. Chem. (2005) 77 (2), 667
2. Auroux, P.-A., et al., Anal. Chem. (2002) 74 (12), 2637
38. Khandurina, J., and Guttman, A., Curr. Opin. Chem. Biol. (2002) 6 (3), 359
3. Vilkner, T., et al., Anal. Chem. (2004) 76 (12), 3373
39. Sundberg, S. A., Curr. Opin. Biotechnol. (2000) 11 (1), 47
4. Terry, S. C., PhD thesis, Stanford University, Stanford, CA, USA (1975)
40. Johnson, M., et al., J. Assoc. Lab. Automation (2002) 7 (4), 62
5. Terry, S. C., et al., IEEE Trans. Electron Devices (1979) 26 (12), 1880
41. Birkos, S., et al., 10th Ann. Conf. Soc. Biomol. Screening, Orlando, FL, USA, (2004)
6. Manz, A., et al., Sens. Actuators, A (1990) 1, 244
42. Tran, L., et al., Anal. Biochem. (2005) 341 (2), 361
7. Stone, H. A., et al., Annu. Rev. Fluid Mech. (2004) 36, 381
43. Hamill, O. P., et al., Pflügers Arch. (1981) 391 (2), 85
8. Landau, L. D., et al. Fluid Mechanics, 2nd ed., Pergamon Press, 1987
44. González, J. E., et al., Drug Discov. Today (1999) 4 (9), 431
9. Brody, J. P., et al., Biophys. J. (1996) 71, 3430
45. Xu, J., et al., Drug Discov. Today (2001) 6 (24), 1278
10. Jeon, N. L., et al., Langmuir (2000) 16 (22), 8311
46. Sinclair, J., et al., Anal. Chem. (2002) 74 (24), 6133
11. Dertinger, S. K. W., et al., Anal. Chem. (2001) 73 (6), 1240
47. Sinclair, J., et al., Anal. Chem. (2003) 75 (23), 6718
12. Pihl, J., et al., Anal. Chem. (2005) 77 (13), 3897
48. Olofsson, J., et al., Anal. Chem. (2004) 76 (17), 4968
13. Jeon, N. L., et al., Nat. Biotechnol. (2002) 20, 826
49. Persson, F., et al., J. Cardiovasc. Electrophysiol. (2005) 16 (3), 329
14. Weigl, B. H., and Yager, P., Science (1999) 283, 346
50. Persson, F., et al., J. Cardiovasc. Pharmacol. (2005) 46 (1), 7
15. Hatch, A., et al., Nat. Biotechnol. (2001) 19, 461
51. Sinclair, J., et al., (2005), submitted
16. Cho, B. S., et al., Anal. Chem. (2003) 75 (7), 1671
52. Seo, J., et al., Appl. Phys. Lett. (2004) 84 (11), 1973
17. Guijt, R. M., et al., Electrophoresis (2001) 22 (2), 235
53. Ionescu-Zanetti, C., et al., Proc. Natl. Acad. Sci. USA (2005) 102, 9112
18. Brivio, M., et al., Anal. Chem. (2002) 74 (16), 3972
54. Müller, A., et al., Nature (2001) 410, 50
19. Becker, H., and Gärtner, C., Electrophoresis (2000) 21 (1), 12
55. Behar, T. N., et al., J. Neurosci. (1994) 14, 29
20. Becker, H., and Locascio, L. E., Talanta (2002) 56 (2), 267
56. Bignold, L. P., J. Immunol. Methods (1988) 108 (1-2), 1
21. Martynova, L., et al., Anal. Chem. (1997) 69 (23), 4783
57. Zicha, D., et al., J. Cell Sci. (1991) 99, 769
22. Roberts, M. A., et al., Anal. Chem. (1997) 69 (11), 2035
58. Nelson, R. D., et al., J. Immunol. (1975) 115, 1650
23. Duffy, D. C., et al., Anal. Chem. (1998) 70 (23), 4974
59. Boyden, S., J. Exp. Med. (1962) 115, 453
24. McDonald, J. C., et al., Electrophoresis (2000) 21 (1), 27
60. Zigmond, S. H., J. Cell Biol. (1977) 75, 606
25. Sia, S. K., and Whitesides, G. M., Electrophoresis (2003) 24 (21), 3563
61. Entschladen, F., et al., Exp. Cell Res. (2005) 307 (2), 418
26. Makamba, H., et al., Electrophoresis (2003) 24 (21), 3607
62. Mao, H., et al., Proc. Natl. Acad. Sci. USA (2003) 100, 5449
27. Chilkoti, A., et al., Anal. Chem. (1995) 67 (17), 2883
63. Lin, F., et al., Biochem. Biophys. Res. Commun. (2004) 319 (2), 576
28. Chen, C. S., et al., Biotechnol. Prog. (1998) 14 (3), 356
64. Wang, S.-J., et al., Exp. Cell Res. (2004) 300 (1), 180
29. Johnston, P. A., and Johnston, P. A., Drug Discov. Today (2002) 7 (6), 353
65. Chung, B. G., et al., Lab Chip (2005) 5 (4), 401
30. Hidalgo, I. J., et al., Gastroenterology (1989) 96 (3), 736
66. Saadi, W., et al., FASEB J. (2005) 19, A1505
31. Netzer, R., et al., Drug Discov. Today (2001) 6 (2), 78
67. Shelby, J. P., et al., Proc. Natl. Acad. Sci. USA (2003) 100, 14618
32. Fermini, B., and Fossa, A. A., Nat. Rev. Drug Discov. (2003) 2, 439
68. Peterman, M. C., et al., Proc. Natl. Acad. Sci. USA (2004) 101, 9951
33. Wölcke, J., and Ullmann, D., Drug Discov. Today (2001) 6 (12), 637
69. Unger, M. A., et al., Science (2000) 288, 113
34. Li, P. C. H., et al., Anal. Chem. (1997) 69 (8), 1564
70. Fu, A. Y., et al., Anal. Chem. (2002) 74 (11), 2451
35. Andersson, H., and van den Berg, A., Sens. Actuators, B (2003) 92 (3), 315
71. Gu, W., et al., Proc. Natl. Acad. Sci. USA (2004) 101, 15861
DECEMBER 2005 | VOLUME 8 | NUMBER 12
51
There has been enormous development in the field of microfluidic
systems, while consuming much smaller amounts of chemicals and
systems over the past decade1-3. It has grown from a specialized
solvents. In addition, the low dimensions of microfluidic systems
area of research into a field engaging hundreds of research groups
enable access to phenomena that are not accessible on the macroscopic
and a growing number of companies. The development of
scale under normal conditions, i.e. the absence of inertia results in
microfluidic systems was made possible by fabrication technologies
laminar, turbulence-free flow and all that brings. This results in
developed in the microelectronics industry. This began in the late
entirely new ways of acquiring chemical, biological, and physical
1970s, when Terry et
al.4,5
demonstrated a gas chromatography
system fabricated on a Si wafer, but received little attention until 1990, when Manz et
al.6
published the now classic paper on
The idea of using living cells as sensor elements in microfluidic systems is attractive for obtaining multiparametric information from
miniaturized total analysis systems (µTAS). This resulted in a
biological systems. Besides the general benefits of miniaturization, the
tremendous increase in the research of chip-based microfluidic
superior control and unique flow properties in microfluidic systems
systems, much like the revolution in the electronic industry after
enable the extraction of high-quality data from cell-based biosensors.
the invention of the transistor and the integrated circuit.
Consequently, microfluidics will reduce costs and increase throughput
Correctly designed microfluidic systems possess the capability to execute operations more quickly than conventional, macroscopic
46
information.
DECEMBER 2005 | VOLUME 8 | NUMBER 12
and data quality in cell-based assays in the drug-discovery process, as well as in fundamental biological research.
ISSN:1369 7021 © Elsevier Ltd 2005
Microfluidics for cell-based assays
REVIEW FEATURE
Microfluidics The term microfluidics refers to devices, systems, and methods for the manipulation of fluid flows with characteristic length scales in the micrometer range7. Microfluidic systems display fundamentally different properties from everyday perceptions of how fluids behave. The ratio between the moment of inertia and the viscous forces in a fluid system is described by the Reynolds number Re = ρνl / η , where ρ, ν, l, and η are the fluid density, fluid velocity, characteristic length scale, and dynamic viscosity, respectively. Microfluidic systems are typically low Reynolds number systems, and the fluid flow is governed mainly by viscous forces and pressure gradients, with a low moment of inertia. The result is a truly laminar, turbulence-free flow and, since low Reynolds number flows are time independent, flow patterns are reversible8. The short length scales involved in microfluidic systems result in other properties, such as high surface-to-volume ratios, small diffusion
Fig. 1 The principle of microreplication. (A) The master structure is a negative of the desired structure and is manufactured by standard microfabrication or micromachining techniques. (B) The master is replicated in thermoplastic or elastomeric materials by molding, imprinting, or casting. (C) The replicate is unmolded from the master, and bonded to a flat base substrate (D).
distances, and small heat capacities. The surface tension of the fluid and the wetting properties of the system can also be significant forces9. The behavior and relative influence of these phenomena will differ
techniques resides in the high total cost of the device, dependent on the cost of both the material itself and the method of fabrication.
greatly depending on the size of the system studied and, in microfluidic
Traditional micromachining techniques are not widely used today in the
systems, the result can be highly counterintuitive. Specifically, the
fabrication of microfluidic devices for either high-volume production or
difference in physical behavior between microscopic and macroscopic
research purposes, with some exceptions. The microfluidics community
systems allows for the construction of functionalities that are difficult
has instead driven the development of alternative fabrication
or even impossible to access on the macroscopic scale. Taking all this
technologies and materials, including powderblasting of glass17,18;
into account, one arrives at the conclusion that, rather than trying to
injection molding19,20, hot embossing19,21, and laser ablation22 of
design a microfluidic system that is just a downsized copy of a
polymeric materials such as polycarbonate and polymethylmethacrylate;
macroscopic system, one should use design rules obtained from the
and casting techniques such as rapid prototyping in silicone elastomers
physics of fluid mechanics and diffusion in confined spaces. Examples of
or epoxies23-25. Apart from powderblasting and laser ablation, these
microfluidic systems designed in this manner include on-chip gradient
methods all rely on the replication of a master structure that is the
generation devices, where solutions are diluted in an automated and
negative of the desired structure (Fig. 1). The master structure is
highly reproducible manner10-12. Such systems have been applied in
reusable, enabling replication of microfluidic devices with a relatively
chemotaxis and electrophysiology
studies12,13.
The
T-sensor14-16,
where
low cost. The techniques allow the fabrication of microfluidic devices in
particles and/or molecules with different diffusion coefficients can be
a large selection of thermoplastic and elastomeric materials that have a
separated for analytical or preparative purposes, is an example of a
wide range of chemical, mechanical, and optical properties. Furthermore,
sensor that takes advantage of the predictable diffusion behavior of
the surface of the materials can be modified – the majority of the
molecules and particles in laminar flows.
thermoplastic materials permanently – by physical as well as chemical methods, giving access to an almost unlimited variety of surface
Fabrication of microfluidic devices
properties. This is important in many cases, both for minimizing
Developments in microfluidics can be attributed to the adoption of
adsorption of chemical species such as hydrophobic molecules and
microfabrication technologies from the microelectronics industry,
proteins to surfaces20,26 (a common problem in microfluidic devices
enabling the creation of complex miniaturized systems in materials such
because of their intrinsically high surface-to-volume ratios), as well as
as Si, glass, and metals. As a consequence, microfluidic devices were
tailoring surface chemistries for cell growth and control27,28.
initially composed of Si or glass. These materials possess several advantages, such as excellent chemical and physical stability, solvent
Cell-based assays and microfluidics
compatibility, and optical properties, along with generally very desirable
The term ‘cell-based assay’ was introduced in the early 1980s and has
surface properties. The main disadvantage of traditional fabrication
grown to encompass many different areas in assessing the effects of
DECEMBER 2005 | VOLUME 8 | NUMBER 12
47
REVIEW FEATURE
Microfluidics for cell-based assays
chemical stimuli on biological cells. Today, the most widespread use is
which large numbers of compounds can be synthesized by
in the pharmaceutical industry, where cell-based assays are used in
combinatorial chemistry38. If compounds are to be screened with higher
various steps in the drug-discovery process. In identifying leads, large
throughput at lower costs, miniaturization is an obvious route,
compound libraries are screened against targets expressed in
resulting in lower volume consumption of chemicals, solvents, and
immortalized cell lines. For example, G-protein coupled receptor
cells39.
screening can be achieved by monitoring the intracellular Ca2+ concentration using
fluorescence29.
Furthermore, different types of cell-
Caliper Life Sciences has developed a microfluidic system that can be employed in various types of high-throughput screening applications40.
based absorption, distribution, metabolism, excretion, and toxicity
The Caliper platform carries out assays that traditionally used
(ADME/Tox) assays are carried out to determine whether the acquired
conventional, plate-based screening platforms. The key component is a
leads have the necessary pharmacokinetic properties and to assess their
glass microchip with integrated ‘sipper’ capillaries that ‘sips’ fluids,
studies30,
assessment of metabolic
e.g. compounds, dyes, buffers, or negative controls from plate wells,
stability and clearance, and screening for QT interval prolongation using
while at the same time continuously drawing solutions, e.g. enzyme,
toxicity. These include permeability fluorescence-based assays or in vitro
electrophysiology31,32.
The driving force for the development of cell-based assays is a desire
substrate, or cell suspensions from integrated, on-chip wells (Fig. 2). The resulting mixtures are transported downstream in a microchannel, which
to develop tests that provide data representative of higher-level
also serves as an incubation chamber, to a detection point where the
biological responses, i.e. results should ideally be comparable to those
fluorescence signal is recorded. This process is performed in a serial
obtained from animal models or clinical trials. Often, cell-based assays
manner, in the sense that each chip sips a large number of cell
can collect new types of information impossible to obtain by other
suspensions, compounds, and dyes alternated with buffer solution to
means. Importantly, implementation of microfluidics technologies into
flush the system. In order to increase the speed of the system, each chip
cell-based assays can provide significant advantages in throughput,
consists of four or 12 units, and the system is capable of performing
costs, and data quality. However, conventional assays have been vastly
assays with considerably higher throughput and better reproducibility
optimized and miniaturized, e.g. 3456 well plates are routinely used33.
than conventional, plate-based screening platforms. In addition, the
Although several examples of cell-based assays in microfluidic systems
chips consume far smaller amounts of chemicals and solvents than
have been demonstrated34-37, there are only a few applications where
conventional systems. The system integrates a majority of the steps
they have proved to offer advantages over conventional assays: high-
that are typically carried out on different stations in a conventional
throughput screening, ion channel research, and chemotaxis studies.
screening platform, resulting in a smaller footprint. The system has turned out to be a genuine success, and is currently used by a large
High-throughput screening
number of pharmaceutical companies in high-throughput screening
Pharmaceutical companies worldwide own extensive and growing
applications. Examples include everything from enzymatic assays to
libraries of different compounds, mainly because of the ease in
cell-based calcium flux assays41,42.
Fig. 2 (A) Schematic of Caliper Life Sciences’ FS-417 four-sipper cell chip for detecting agonist-induced calcium flux. Dye-loaded platelet samples are drawn from the cell wells and continuously streamed through the chip. Platelets are exposed to agonist aspirated from a 384 well microtiter plate at the sipper junction and mixed by diffusion while the sample flows continuously through the chip past the fluorescence detection zone. The flow rate depends on a combination of the applied vacuum and sample viscosity. (B) The microfluidic chips are designed to flow four samples through the detection zone three times simultaneously, allowing 15 s, 30 s, and 60 s incubation intervals. The time to detection is defined by the flow rate and the distance from the sipper to the detection zone. (Adapted from42. © 2005 Elsevier.)
48
DECEMBER 2005 | VOLUME 8 | NUMBER 12
Microfluidics for cell-based assays
REVIEW FEATURE
Fig. 3 The patch clamp technique. After a cell has been approached by the pipette (A), a high-resistance seal is achieved on application of negative pressure, resulting in the cell-attached configuration (B). Further application of negative pressure ruptures the membrane, resulting in the whole-cell configuration, i.e. electrical contact with the inside of the cell (C). Inside-out and outside-out configurations are achieved by pulling the pipette away from the cell (D and E). (F) An equivalent electrical circuit of a cell in whole-cell configuration during acquisition of data. These physical parameters set the boundary conditions for temporal resolution and the errors obtained during the recording. Em denotes the membrane potential, Rs the series resistance, Cp the pipette capacitance, Cm the membrane capacitance, and Rm the membrane resistance.
Ion channel studies Ion channels are proteins that span the plasma membranes of cells and
place by molecular diffusion alone48. The device is placed on a
organelles and selectively control the diffusive transport of ions across
computer-controlled scanning stage mounted on an inverted
the membrane. Ion channels are one of the fundamental components in
microscope. Cells are patch-clamped in the open volume, translocated
cell-to-cell communication and are involved in everything from the
to the channel exits, and scanned across the different solution
contraction of muscles to the function of the brain, making them highly
environments with full control of transfer and exposure times. The
interesting in many areas of biology. Consequently, the malfunction of
system rapidly generates dose-response curves under extremely stable
ion channels is responsible for a number of diseases, making them
conditions, and several dose-response scans can be obtained from the
important drug targets. The gold standard for investigating ion-channel
same cell. In addition, the well-defined force that originates from the
activity is the patch clamp technique, an electrophysiological method
flow and acts on the cell-pipette system stabilizes the patch-clamped
(Fig. 3)43-45. It can provide single-channel resolution along with a
cell, allowing extended recording times from one cell, in some cases up
temporal resolution in the microsecond range. However, it is labor-
to hours47. The throughput is 10-100 times greater than conventional
intensive, notoriously low in throughput, and limited by the number of cells that can be patch clamped per unit time, making it difficult to generate statistically significant amounts of data. One approach to increase throughput is to increase the number of data points obtained per patch-clamped cell. This can be achieved by adding microfluidics to conventional patch clamp experiments12,46,47, a setup developed by Orwar and coworkers and commercialized by Cellectricon. The system is built around a microfluidic device fabricated in a silicone elastomer and glass (Figs. 4A and 4B). The devices comprise of a number of sample wells connected to an open volume through microfluidic channels. At the exit to the open volume, the channels are tightly packed. Fluids are pumped through the channels and into the open volume by applying pressure to the sample wells. Flows emerging from the channels couple viscously, forming a single laminar flow that propagates in the open volume (Fig. 4C). This flow contains a number of discrete zones with well-defined chemical environments corresponding to the loading of the sample wells and, because of the low Reynolds number of the system, mixing between adjacent environments takes
Fig. 4 Principle of the cell-based bar-code reader. (A) Overview of a 16 channel device, where 16 sample wells are connected to an open volume by individual microchannels. (B) A pipette with a patch-clamped cell situated outside the channel outlets. The cell can be exposed to different solution environments, e.g. different ligands can be alternated with buffer solutions in between exposures by translation of the device using a motorized scanning stage. The different solution environments can be thought of as a bar code that is read by the cell. (C) Fluorescence micrograph showing the collimated stream of different dye solutions at the exit into the open volume.
DECEMBER 2005 | VOLUME 8 | NUMBER 12
49
REVIEW FEATURE
Microfluidics for cell-based assays
Fig. 5 Patch clamp array on a microfluidic platform. (A) The main chamber containing cells in suspension is connected to a set of recording capillaries. Cell trapping is achieved by applying negative pressure to the recording capillaries. Patch clamp measurements are obtained by placing AgCl electrodes in each of the capillaries, as well as in the main chamber. Signals are fed through a multiplexing circuit and into the data acquisition system. (B) Scanning electron micrograph of three recording capillary orifices as seen from the main chamber. The capillary dimensions are approximately 4 µm x 3 µm, with a center-to-center distance of 20 µm. (C) Fluorescence micrograph of cells trapped at three capillary orifices. The entire device consists of 12 capillaries, with six capillaries arrayed along each side of the main chamber fluidic channel. (Adapted from53. © 2005 National Academy of Sciences.)
Fig. 6 Schematic of a gradient-generating device and the experimental setup. (A) Top view of the entire device consisting of the gradient generator and the observation area. (B) Three-dimensional representation of the observation area, in which cells can be exposed to different gradients of chemoattractants. (Adapted from13. © 2002 Nature Publishing Group.)
environment around cells13 (Fig. 6). A stable time-independent gradient is created by the microfluidic network seen in Fig. 6A. In its simplest form, solutions with and without chemoattractants are introduced into the network. As the two solutions flow down the network, they are
systems, sample consumption is greatly reduced, and there is no risk of
repeatedly split, recombined, and mixed, resulting in an increasing
cross-contamination, since the devices themselves are disposable. The
number of channels and concentrations. At the end of the network, the
system has found use in various stages of the drug-discovery process as
channels are brought together, giving a concentration gradient spanning
well as in studies of receptor
functionality49-51.
a single, wide microchannel. At this point, cells can be introduced and
Another, exclusively microfluidics-based approach to increase throughput has been developed by Lee and
coworkers52,53
(Fig. 5). Cells
their behavior studied. This technique was used by Jeon and coworkers to study chemotaxis of neutrophils in gradients of the chemoattractant
are patch-clamped in a microfluidic channel ~2 µm in width and ~2 µm
interleukin-8. This initial demonstration has resulted in the use of
in height. The cells are delivered from sample reservoirs through
microfluidics to study chemotaxis in different research areas62-66, and
microfluidic channels to an array of small channels where the cells are
offers the possibility of screening for drugs that modulate chemotaxis.
trapped and brought to whole-cell configuration by the application of a vacuum. Drugs are delivered through the same microchannel as the
Emerging technologies
cells, but from different input wells. While this concept does not provide
Apart from the assays mentioned above, there are few examples of
the same data quality as the conventional patch-clamp technique, it has
cell-based assays in a microfluidic format that exhibit unique
potential for massive parallelization and takes full advantage of
advantages over conventional approaches. However, there are a
microfluidics in order to achieve a fully automated patch-clamp device.
number of specialized assays and technologies that are very interesting and could become key components in tomorrow’s microfluidic cell-
Chemotaxis
based assays.
Chemotaxis is the directional migration of cells in response to chemical gradients of molecules called chemoattractants. The process is crucial in numerous biological
50
processes54,55,
and several approaches, e.g. different
Chiu and coworkers67 have demonstrated a microfluidic model for the obstruction of capillary blood vessels by single red blood cells infected with the Plasmodium falciparum parasite, as occurs in severe
types of chambers or puffer pipettes, have been developed to study
malaria. Capillary blood vessels were modeled using microfluidic
chemotaxis56-61. Most of the methods are nonideal in that the
channels fabricated in poly(dimethylsiloxane), and the transport of
generated gradients are created in macroscopic environments, are
infected and uninfected red blood cells through the channels was
nonlinear, and change with time in an uncontrolled manner. By making
studied. An artificial synapse chip that simulates synaptic transmitter
use of chip-based gradient generation10,11, chemotaxis studies can be
release has also been developed68. These models of biological processes
carried out with precise spatial and temporal control of the chemical
would be impossible without the aid of microfluidics.
DECEMBER 2005 | VOLUME 8 | NUMBER 12
Microfluidics for cell-based assays
A multilayer soft lithography technique has been developed by Quake and coworkers69 for fabricating pumps and valves in elastomeric materials. It has been used to make cell
sorters70
and could prove useful
in the construction of powerful cell-based assays in microfluidic formats. Takayama and
coworkers71
have avoided the relatively complex and
REVIEW FEATURE
of functional components, e.g. pumps and valves, as well as new, costeffective fabrication technologies. As the focus now shifts to applications, completely new types of cell-based assays will result in information that was previously impossible to acquire, as shown in the examples of high-throughput screening, patch clamp, and chemotaxis
expensive multistep microfabrication process of multilayer soft
studies. Although there are many good proposals for the use of
lithography in making cell-cultivation devices that operate using
microfluidics in cell-based assays, the field has not yet matured to
elastomeric channels and Braille displays. Their system has been used
present commercial functional devices apart from those mentioned
for long-term cultivation of endothelial cells and observation of their
above. To conclude, the potential of microfluidics will undoubtedly see
behavior in response to shear stress.
the technology assume a larger and more important role in the development of cell-based assays for the creation of more and higher
Future outlook
quality targets and leads in drug discovery. Microfluidics will also
The field of microfluidics is still in its infancy and, until recently, has
generate discoveries unattainable using conventional, macroscopic
been highly technology driven. The focus has been on the development
approaches.
REFERENCES
36. Wheeler, A. R., et al., Anal. Chem. (2003) 75 (14), 3581
1. Reyes, D. R., et al., Anal. Chem. (2002) 74 (12), 2623
37. Poulsen, C. R., et al., Anal. Chem. (2005) 77 (2), 667
2. Auroux, P.-A., et al., Anal. Chem. (2002) 74 (12), 2637
38. Khandurina, J., and Guttman, A., Curr. Opin. Chem. Biol. (2002) 6 (3), 359
3. Vilkner, T., et al., Anal. Chem. (2004) 76 (12), 3373
39. Sundberg, S. A., Curr. Opin. Biotechnol. (2000) 11 (1), 47
4. Terry, S. C., PhD thesis, Stanford University, Stanford, CA, USA (1975)
40. Johnson, M., et al., J. Assoc. Lab. Automation (2002) 7 (4), 62
5. Terry, S. C., et al., IEEE Trans. Electron Devices (1979) 26 (12), 1880
41. Birkos, S., et al., 10th Ann. Conf. Soc. Biomol. Screening, Orlando, FL, USA, (2004)
6. Manz, A., et al., Sens. Actuators, A (1990) 1, 244
42. Tran, L., et al., Anal. Biochem. (2005) 341 (2), 361
7. Stone, H. A., et al., Annu. Rev. Fluid Mech. (2004) 36, 381
43. Hamill, O. P., et al., Pflügers Arch. (1981) 391 (2), 85
8. Landau, L. D., et al. Fluid Mechanics, 2nd ed., Pergamon Press, 1987
44. González, J. E., et al., Drug Discov. Today (1999) 4 (9), 431
9. Brody, J. P., et al., Biophys. J. (1996) 71, 3430
45. Xu, J., et al., Drug Discov. Today (2001) 6 (24), 1278
10. Jeon, N. L., et al., Langmuir (2000) 16 (22), 8311
46. Sinclair, J., et al., Anal. Chem. (2002) 74 (24), 6133
11. Dertinger, S. K. W., et al., Anal. Chem. (2001) 73 (6), 1240
47. Sinclair, J., et al., Anal. Chem. (2003) 75 (23), 6718
12. Pihl, J., et al., Anal. Chem. (2005) 77 (13), 3897
48. Olofsson, J., et al., Anal. Chem. (2004) 76 (17), 4968
13. Jeon, N. L., et al., Nat. Biotechnol. (2002) 20, 826
49. Persson, F., et al., J. Cardiovasc. Electrophysiol. (2005) 16 (3), 329
14. Weigl, B. H., and Yager, P., Science (1999) 283, 346
50. Persson, F., et al., J. Cardiovasc. Pharmacol. (2005) 46 (1), 7
15. Hatch, A., et al., Nat. Biotechnol. (2001) 19, 461
51. Sinclair, J., et al., (2005), submitted
16. Cho, B. S., et al., Anal. Chem. (2003) 75 (7), 1671
52. Seo, J., et al., Appl. Phys. Lett. (2004) 84 (11), 1973
17. Guijt, R. M., et al., Electrophoresis (2001) 22 (2), 235
53. Ionescu-Zanetti, C., et al., Proc. Natl. Acad. Sci. USA (2005) 102, 9112
18. Brivio, M., et al., Anal. Chem. (2002) 74 (16), 3972
54. Müller, A., et al., Nature (2001) 410, 50
19. Becker, H., and Gärtner, C., Electrophoresis (2000) 21 (1), 12
55. Behar, T. N., et al., J. Neurosci. (1994) 14, 29
20. Becker, H., and Locascio, L. E., Talanta (2002) 56 (2), 267
56. Bignold, L. P., J. Immunol. Methods (1988) 108 (1-2), 1
21. Martynova, L., et al., Anal. Chem. (1997) 69 (23), 4783
57. Zicha, D., et al., J. Cell Sci. (1991) 99, 769
22. Roberts, M. A., et al., Anal. Chem. (1997) 69 (11), 2035
58. Nelson, R. D., et al., J. Immunol. (1975) 115, 1650
23. Duffy, D. C., et al., Anal. Chem. (1998) 70 (23), 4974
59. Boyden, S., J. Exp. Med. (1962) 115, 453
24. McDonald, J. C., et al., Electrophoresis (2000) 21 (1), 27
60. Zigmond, S. H., J. Cell Biol. (1977) 75, 606
25. Sia, S. K., and Whitesides, G. M., Electrophoresis (2003) 24 (21), 3563
61. Entschladen, F., et al., Exp. Cell Res. (2005) 307 (2), 418
26. Makamba, H., et al., Electrophoresis (2003) 24 (21), 3607
62. Mao, H., et al., Proc. Natl. Acad. Sci. USA (2003) 100, 5449
27. Chilkoti, A., et al., Anal. Chem. (1995) 67 (17), 2883
63. Lin, F., et al., Biochem. Biophys. Res. Commun. (2004) 319 (2), 576
28. Chen, C. S., et al., Biotechnol. Prog. (1998) 14 (3), 356
64. Wang, S.-J., et al., Exp. Cell Res. (2004) 300 (1), 180
29. Johnston, P. A., and Johnston, P. A., Drug Discov. Today (2002) 7 (6), 353
65. Chung, B. G., et al., Lab Chip (2005) 5 (4), 401
30. Hidalgo, I. J., et al., Gastroenterology (1989) 96 (3), 736
66. Saadi, W., et al., FASEB J. (2005) 19, A1505
31. Netzer, R., et al., Drug Discov. Today (2001) 6 (2), 78
67. Shelby, J. P., et al., Proc. Natl. Acad. Sci. USA (2003) 100, 14618
32. Fermini, B., and Fossa, A. A., Nat. Rev. Drug Discov. (2003) 2, 439
68. Peterman, M. C., et al., Proc. Natl. Acad. Sci. USA (2004) 101, 9951
33. Wölcke, J., and Ullmann, D., Drug Discov. Today (2001) 6 (12), 637
69. Unger, M. A., et al., Science (2000) 288, 113
34. Li, P. C. H., et al., Anal. Chem. (1997) 69 (8), 1564
70. Fu, A. Y., et al., Anal. Chem. (2002) 74 (11), 2451
35. Andersson, H., and van den Berg, A., Sens. Actuators, B (2003) 92 (3), 315
71. Gu, W., et al., Proc. Natl. Acad. Sci. USA (2004) 101, 15861
DECEMBER 2005 | VOLUME 8 | NUMBER 12
51