Anal. Chem. 2003, 75, 918-921
A Biosensing System Based on Extracellular Potential Recording of Ligand-Gated Ion Channel Function Overexpressed in Insect Cells Tetsuya Haruyama,*,† Saknan Bongsebandhu-Phubhakdi,‡ Ibuki Nakamura,‡ David Mottershead,§ Kari Keina 1 nen,§ Eiry Kobatake,‡ and Masuo Aizawa‡
Department of Biological Functions and Engineering, Kyushu Institute of Technology, Kitakyushu Science and Research Park, 2-4 Hibikino, Wakamatsu-ku, Fukuoka 808-0196, Japan, Department of Biological Information, Tokyo Institute of Technology, 4259 Nagatsuta, Midori-ku, Yokohama 226-8501, Japan, and Department of Biosciences, Division of Biochemistry, Viikki Biocenter, 00014, University of Helsinki, P.O. Box 56, Helsinki, Finland
We have used outer cell potential measurement to record agonist-dependent cellular responses in cells engineered to express ligand-gated ion channels and grown on a microelectrode surface. Application of glutamate, a natural agonist, induced a complex and robust potentiometric response in cells expressing homomeric GluR-D glutamate receptor, but not in nonexpressing control cells. The response consisted of an initial decrease in outer potential followed by a transient increase and was not obtained for other amino acids devoid of agonist activity at glutamate receptors. Furthermore, the pharmacological agonist of the GluR-D receptor, kainate, also produced the potentiometric response whereas 6-cyano-7-nitroquinoxaline2,3-dione, a competitive antagonist, was not active in itself but attenuated the responses to glutamate. The time course of the measured changes was slow, which may be partially due to the ligand being applied by free diffusion but may also reflect a contribution by secondary changes in the behavior of the cells. This novel approach should be applicable to other ligand-gated ion channels and holds promise as a cell-based biosensor for high-throughput drug screening and other applications. The highly specific ligand recognition and intrinsic signal amplification by ligand-gated ion channels make them attractive molecules for biosensor development. Both naturally occurring1 and engineered2 ligand-gated channels have been employed as the signal-generating component in both molecular (nonliving)and cell-based biosensors. One obvious application of biosensors that use natural receptors as the element responsible for chemical specificity would be in high-throughput drug screening, as this class of molecules includes a number of important drug targets. * Corresponding author. E-mail:
[email protected]. † Kyushu Institute of Technology. ‡ Tokyo Institute of Technology. § University of Helsinki. (1) Hahnenberger, K. M.; Krystal, M.; Esposito, K.; Tang, W.; Kurtz, S. Nat. Biotechnol. 1996, 14, 880-883. (2) Cornell, B. A.; Braach-Maksvytis, V. L.; King, L. G.; Osman, P. D.; Raguse, B.; Wieczorek, L.; Pace, R. J. Nature 1997, 387, 580-583.
918 Analytical Chemistry, Vol. 75, No. 4, February 15, 2003
For ligand-gated ion channels, a large-scale functional assay may be achieved by coupling the agonist-triggered ion flux in cultured cells to an optical readout, via use of fluorescent indicators sensitive to changes in the concentration of the permeant ions or to changes in the membrane potential.3 Although electrophysiological assays of ion channel function have an exquisite sensitivity and resolution, the level of technical expertise involved and the time-consuming nature of the experiments largely exclude the use of glass electrode implade in cells from drug screening. Noninvasive extracellular measurement of the electrical activity in cells grown on electrode grids would provide a potentially more robust method that may lend itself to automation. Numerous studies have demonstrated measurement of spontaneous and pharmacologically manipulated electric activity (action potentials) in networks of cultured neurons grown on electrode arrays. The measured signal consists of complex spikes that reflect temporal and spatial patterns of activity of voltage-gated ion channels. These patterns can be modified by changes in the external conditions, e.g., by the presence of pharmacological agents, and therefore this kind of system may also hold promise for biosensor applications.4 Although a successful measurement of maxi-K channel function by a field effect transistor was recently reported,5 direct noninvasive electronic measurement of ligand-gated ion channel function has not been reported. Indirectly, microphysiometry that senses pH changes in the cellular environment, due to a metabolic response to ion channel activation, however, has been used to monitor the activity of ligand-gated channels.1 Ionotropic glutamate receptors (iGluR) are ligand-gated cation channels that mediate the majority of fast excitatory neurotransmission in the brain and comprise three subclasses known as AMPA, kainate, and NMDA receptors, which differ in their molecular composition, biophysical properties, and relative affinities to pharmacological compounds.6 AMPA receptors are hetereoand homomeric assemblies of four homologous subunits, GluR(3) Frostig, R. D.; Lieke, E.; Ts’o, D. Y.; et al. Proc. Natl. Acad. Sci. U.S.A. 1990, 87, 6082-6086. (4) Ziegler, C. J. Anal. Chem. 2000, 366, 552-559. (5) Zeck, G.; Fromherz, P. Proc. Natl. Acad. Sci. U.S.A. 2001, 98, 1045710462. (6) Dingledine, R.; Borges, K.; Bowie, D.; Traynelis, S. F. Pharmacol. Rev. 1999, 51, 7-61. 10.1021/ac025670x CCC: $25.00
© 2003 American Chemical Society Published on Web 01/21/2003
A, -B, -C, and -D (alternatively named as GluR1-4), and form channels that have a relatively low affinity to glutamate (EC50 in the millimolar/submillimolar range) and display rapid kinetics of onset and offset of the response.6 AMPA receptors lacking the GluR-B subunit form channels that are permeable to calcium whereas GluR-B-containing channels are only permeable to monovalent cations. All three glutamate receptor classes are considered as targets for neuroprotective drugs. In this study, we demonstrate the use of outer cell potential (OCP) measurement to monitor agonist-dependent activation of recombinant AMPA receptor overexpressed in insect cells. The system has a remarkable potential for the development of biosensing and drug discovery based on ligand-gated channels. EXPERIMENTAL PROTOCOL Materials. 6-Cyano-7-nitroquinoxaline-2,3-dione (CNQX) were purchased from Tocris. All amino acids were purchased from Wako chemical Co. Ltd. Other chemicals are guaranteed experimental grade. Expression of GluR-D in Insect Cells. Recombinant baculoviruses encoding the GluR-D (flip isoform)7 AMPA receptor subunit and jellyfish green fluorescent protein (GFP) have been described.8,9 Spodoptera frugiperda Sf9 cells were maintained in SF900-II medium (Life Technologies) and infected with the recombinant baculoviruses using standard procedures.10 As an alternative, we used Trichoplusia ni (High Five; Invitrogen) insect cells stably transfected to express GluR-D. A detailed description of the construction of the expression plasmid and of the selection of stably transfected cells will be published elsewhere (D.M. and K.K., in preparation). Outer Cell Potential Recording. A microarray-type electrode (electrode size: 50um × 50 um, 8 × 8 electrode array; Panasonic Co. Ltd.) was employed for outer cell potential measurement. Cells were suspended in phosphate-buffered saline solution (PBS; pH 7.4, 139.68 mM NaCl. 9.57 mM Na2HPO4). The cell suspension was poured into the array electrode part and covered with a thick glass strip in order to obtain a tight contact between the cell and an electrode. Each cell was localized at the same distance from the thick cover edge. The microarray electrode was connected to the computer through the amplifier and noise filter in order to record and analyze the potential output. Electrode potential was measured between a cell-contacted electrode (analysis electrode) and a noncontacted electrode (reference electrode). The potential profile was recorded continuously for detailed investigation. Chemical solutions, i.e., agonist or antagonist, were applied by microsyringe from the top of the meniscus. The application manner was optimized to avoid injection noise. RESULTS For the analysis of agonist-induced changes in the outer cell potential (OCP) of ligand-gated channel overexpressing cells, we used Sf9 insect cells infected with recombinant baculovirus for (7) Sommer, B.; Keina¨nen, K.; Verdoorn, T. A.; Wisden, W.; Burnashev, N.; Herb, A.; Ko ¨hler, M.; Takagi, T.; Sakmann, B.; Seeburg, P. H. Science 1990, 249, 1580-1585. (8) Kuusinen, A.; Abele, R.; Madden, D. R.; Keina¨nen, K. J. Biol. Chem. 1999, 274, 28937-28943. (9) Oker-Blom, C.; Orellana, A.; Keina¨nen, K. FEBS Lett. 1996, 389, 238243. (10) O’Reilly, D. R.; Miller, L. K.; Luckow, V. A. Baculovirus Expression Vectors: A Laboratory Manual; Freeman: New York, 1992.
Figure 1. Schematic illustration of experimental setup for outer cell potential recording.
expression of Flag-tagged AMPA receptor subunit GluR-D. As control for the specificity of the responses, Sf9 cells infected with a recombinant baculovirus encoding GFP and noninfected Sf9 cells were employed. GluR-D expressing cells produced a 110-kDa band in an anti-Flag western blot and showed intense anti-Flag staining in immunofluorescence microcopy of nonpermebilized cells, indicating robust expression of the receptor on the cell surface (results not shown). This is consistent with our previous work indicating that GluR-D expressing insect cells express 10-20 pmol of high-affinity [3H]AMPA binding sites/mg of membrane proteins (corresponding to binding sites per cell) and produce robust agonist-triggered currents in patch clamp electrophysiology.11,12 For the OCP measurements, Sf9 cells were rinsed extensively with the buffer and transferred on a microelectrode array. The cells were allowed to attach on a microelectrode surface for a few minutes, whereafter a glass coverslip was gently pressed on top of the cells as shown in Figure 1. Using the coverslip, the cells are attached firmly on the electrode surface, which facilitates reliable recording of the OCP. The coverslip isolates the immediate surroundings of the cell from the bulk solution by limiting the diffusion of ions and molecules in and out from the area under recording. We used cell densities that yielded single well-separated cells contacting the electrodes. The cells were stably attached at this point as buffer injection (to simulate to application of test compounds) did not change the position of the cell, as judged by microscopic observation. OCP was measured with single recording electrodes that were in contact with a single GluR-D expressing cell. Without administration of any agonist, the recorded OCP was stable (Figure 2A and B). Injection of buffer did not cause any shifts in the OCP. However, injection of glutamate (final concentration of 2 mM), an AMPA receptor agonist, resulted in dramatic changes in OCP: an initial downward shift to a more negative potential was followed by an upward change which then leveled off back to baseline (Figure 2C-E). The OCP responses were elevated by glutamate concentration increments. The magnitudes of the negative and positive components of the response were similar. Similar injection of two other polar amino acids, aspartate and glutamine, which are structurally and by physicochemical properties closely related to glutamate but do not have agonist activity at AMPA receptor, did not cause any shift in the (11) Keina¨nen, K.; Ko ¨hr, G.; Seeburg, P. H.; Laukkanen, M.-L.; Oker-Blom, C. Bio/Technology 1994, 12, 802-806. (12) Kuusinen, A.; Arvola, M.; Oker-Blom, C.; Keinanen, K. Eur. J. Biochem. 1995, 233, 720-6.
Analytical Chemistry, Vol. 75, No. 4, February 15, 2003
919
Figure 2. Time course OCP profile of GluR-Sf9 cell. (A) Instrumental background level, (B) buffer solution was applied at 0 s point, (C) 1.25 mM glutamic acid solution was applied at 0 s point, (D) 1 mM glutamic acid solution was applied at 0 s point, (E) 2 mM glutamic acid solution was applied at 0 s point, (F) 2 mM glutamic acid was applied at 0 s point to the intact Sf9 cell (GluR -), (G) 4 mM glutamic acid was applied at 0 s point to the GFP-GluR cell (GFP +, GluR -), (H) 2 mM kainic acid was applied at 0 s point, (I) 2 mM glutamic acid was applied at 0 s point. 0.2 mM of CNQX was applied in advance, and (J) CNQX (0.05 mM) was applied at 0 s point (without glutamate).
OCP at 6 mM final concentrations, nor did injection of alanine, an amino acid with a small nonpolar amino acid. A similar specific OCP response could also be obtained in another insect cell line, High Five cells, which were stably transfected to express GluR-D (Figure 3). In the case of High Five cells, however, a somewhat larger concentration of agonist (glutamate) was required. This may be due to the much lower number of receptors expressed on the cell surface as compared to Sf9 cells. To confirm that the response is mediated specifically by GluR-D, we tested OCP responses to injected glutamate in Sf9 cells infected with a recombinant baculovirus encoding GFP. The cells were brightly fluorescent under UV illumination but did not produce any OCP response (Figure 3G). The same lack of glutamate response was observed with noninfected Sf9 cells (Figure 3F). The abovedescribed findings suggest that the OCP response is dependent 920
Analytical Chemistry, Vol. 75, No. 4, February 15, 2003
on the presence of GluR-D receptors in the insect cells. Therefore, one would expect that the responses would display ligand pharmacology typical of AMPA receptors. As an initial test for this, we used kainic acid, a partial agonist giving rise to robust nondesensitizing current responses at AMPA receptors. Application of kainate (at 2 mM final concentration) triggered a strong OCP response in GluR-D expressing Sf9 cells (Figure 2H). Application of CNQX, a competitive antagonist, did not itself induce any OCP response in cells (Figure 2J), but when applied together with glutamate, CNQX diminished the OCP response in a concentration-dependent manner (Figure 2I). DISCUSSION Although OCP measurements have been used extensively to characterize propagation of action potentials in neuronal networks
Figure 3. Time course OCP profile of GluR-High Five cell. (A) Buffer solution was applied at 0 s point, (B) 2.5 mM glutamic acid solution was applied at 0 s point, (C) 3.5 mM glutamic acid solution was applied at 0 s point, (D) 6 mM glutamic acid solution was applied at 0 s point, and (E) 8 mM glutamic acid solution was applied at 0 s point.
growing on a microelectrode array,6,13-15 the ability of this system to monitor activation of ligand-gated channels in single cells has not been demonstrated until now. The OCP response to glutamate of the GluR-D-expressing cells in the present study is most likely due to the ligand-gated channel function (see below) of GluR-D AMPA receptor, and therefore, our work provides a proof of the principle for a new type of cell-based biosensor. Several findings are consistent with the conclusion that the OCP shifts are due to cation channel activity of GluR-D. First, only glutamate and not other, even closely related amino acids was capable of triggering the response. Second, the absence of any responses in baculovirusinfected Sf9 cells expressing GFP, and in noninfected cells, argues against any endogenous glutamate-dependent mechanism, like metabolism or amino acid transport. Third, similar responses were produced by kainate, a pharmacological agonist at AMPA receptors, whereas CNQX, a specific antagonist, was not effective as such but inhibited the response to glutamate. At present, the physical basis of the biphasic nature of the OCP response is difficult to explain from the known properties of the system’s electronic and biological components and, therefore, remains a subject of ongoing studies. Binding of glutamate to GluR-D will lead to transient opening of an intrinsic transmembrane cation channel and movement of cations toward their lower electrochemical potential.13-15 Under the recording condition, glutamate will cause the channels expressed on the cell surface to open, which will trigger a membrane potential change due to sodium influx. In the case of the engineered cells, GluRs are overexpressed and are much exhibited on the cell surface. The OCP response represents a time-averaged summed response from all channels expressed on the target cell. The time course is determined by the rate of diffusion of glutamate, which largely varies between different experimental systems, by the spatial distribution of the channels, and by the intrinsic kinetic properties of the channels themselves. In our system, the very slow onset (13) Eichenbaum, H. B., Davis, J. L., Eds. Neuronal Ensembles: Strategies for Recording and Decoding; Wiley-Liss: New York, 1998. (14) Najafi, K. IEEE EMBS 1994, 13, 375-385. (15) Oka, H.; Shimono, K.; Ogawa, R.; Sugihara, H.; Taketani, M. J. Neurosci. Methods 1999, 93, 61-67. (16) Mano, I.; Lamed, Y.; Teichberg, V. I. J. Biol. Chem. 1996, 271, 1529915302.
of the response (or the rather long dead time) is most likely due to the slow diffusion-driven application of the agonist. Given the rapidly desensitizing character of AMPA receptor responses, the downward OCP shift in the recorded responses may reflect the nondesensitizing component of the GluR-D-mediated ion currents. Although in patch clamp electrophysiological experiments with ultrarapid agonist application, the peak response is reached within millisecond, substantially longer values (microseconds to seconds) are typically obtained in electrophysiological recording systems, which use slower agonist application and are capable of recording only the nondesensitizing component.16 Currently, the origin of the upward shift is unclear and a subject of further studies. Metabolic changes in the cells leading to efflux of positive ions or influx of negative ions are among the possibilities. Whatever the exact mechanisms leading to the two-phased response, the present results clearly show that the OCP response is crucially dependent on the presence of GluR-D receptor and follows a ligand specificity typical of AMPA-type glutamate receptors. In the present work, we did not assess the sensitivity of the system to glutamate or other ligand substances. The functional affinity of AMPA receptors in terms of EC50 values (agonist concentration to have a half-maximal response) measured in electrophysiological assays is the low-millimolar/submillimolar range and, therefore, consistent with the responses we recorded at 2 mM final concentrations. One important factor contributing to the successful recording of OCP responses to glutamate in the present study must be the highly efficient expression of the glutamate-gated ion channel (GluR-D) on insect cells’ surface. An obvious application for the present system would be in highthroughput screening of drugs targeted on the functions of ligandgated channels. ACKNOWLEDGMENT This work was partly supported by grants from the Ministry of Education, Sports and Science, Japan to M.A. and T.H., and from the National Technology Agency of Finland (TEKES) to K.K. Received for review April 1, 2002. Accepted November 20, 2002. AC025670X Analytical Chemistry, Vol. 75, No. 4, February 15, 2003
921
