Fresenius J Anal Chem (2001) 369 : 30–35
© Springer-Verlag 2001
S P E C I A L I S S U E PA P E R
R. Ehret · W. Baumann · M. Brischwein · M. Lehmann · T. Henning · I. Freund · S. Drechsler · U. Friedrich · M.-L. Hubert · E. Motrescu · A. Kob · H. Palzer · H. Grothe · B. Wolf
Multiparametric microsensor chips for screening applications
Received: 28 July 2000 / Revised: 11 October 2000 / Accepted: 18 October 2000
Abstract The identification of drug targets for pharmaceutical screening can be greatly accelerated by gene databases and expression studies. The identification of leading compounds from growing libraries is realized by high throughput screening platforms. Subsequently, for optimization and validation of identified leading compounds studies of their functionality have to be carried out, and just these functionality tests are a limiting factor. A rigorous preselection of identified compounds by in vitro cellular screening is necessary prior to using the drug candidates for the further time consuming and expensive stage, e.g. in animal models. Our efforts are focused to the parallel development, adaptation and integration of different microelectronic sensors into miniaturized biochips for a multiparametric, functional on-line analysis of living cells in physiologically environments. Parallel and on-line acquisition of data related to different cellular targets is required for advanced stages of drug screening and for economizing animal tests.
Introduction Recently, biochip technologies began to attract an increasing interest from both commercial and scientific institutions. Biochips are usually defined as microscaled devices for automated bioanalytical screening, adapted to the use with fluidic components. Most biochips are designed for high throughput screening in genetic analysis (“DNAR. Ehret · W. Baumann · M. Brischwein1 · M. Lehmann · T. Henning · I. Freund · S. Drechsler · U. Friedrich · M.-L. Hubert · E. Motrescu1 · A. Kob · H. Palzer · B. Wolf1 Universität Rostock, Fachbereich Biowissenschaften (Biophysik), Wismarsche Straße 8, 18057 Rostock, Germany e-mail:
[email protected] M. Brischwein · E. Motrescu · H. Grothe · B. Wolf () Technische Universität München, Heinz-Nixdorf-Lehrstuhl für Medizinische Elektronik, Arcisstraße 21, 80333 München, Germany 1 on
move from Rostock to München
chips”). One reason, why genetic data have to be interpreted with caution is based on the fact, that nearly all regulatory proteins in eukaryotic cells are subjected to delicately tuned and transient patterns of posttranslational modifications and self organization [1]. Subsequent optimization and validation of identified leading compounds by functional studies is usually done by measuring changes of intracellular concentration of Ca2+, intracellular pH or membrane potential in response to applied drugs in cell systems with fluorescent dyes. Nowadays, these optical readout methods are the most important way for in vitro cellular screening. A different approach, however, is the functional online analysis of living cells in physiologically controlled environments for extended periods of time. Our efforts are directed to the parallel development, adaptation and integration of different microelectronic sensors into miniaturized biochips for a multiparametric cellular monitoring with the so-called Cell-Monitoring-System, CMS® [2–4]. Parallel and on-line acquisition of data related to different cellular targets seems to be required for advanced stages of drug screening and for economizing animal tests. In this work, we report about the evaluation of existing sensor structures and current optimization strategies. Thereafter, a stepwise expansion of the system leading to chip arrays for screening applications is envisaged.
Material and methods As individual transducer elements pH-ISFETs and miniature pH glass electrodes for records of extracellular pH changes due to extracellular acidification were adapted. Miniature clark-type oxygen electrodes were used for monitoring cellular oxygen uptake and interdigitated electrode structures (IDES) for the detection of changes in cellular adhesion and morphology. Along with planar temperature sensors, these three sensor functions are to be incorporated into a multiparametric chip. For continuous long term monitoring of highly sensitive cells the precise maintenance of the environment by fluid exchange systems is necessary. Pump-driven and valve-controlled systems, superfusing the cell cultures inside thermostated chambers were used (Figs. 1, 2). Small, closed cell culture volumes facilitate fast and sensitive determinations of cellular metabolic rates in order to
31 Fig. 1 left: First version of a multiparametric sensor chip; right: 2-channel cell culture unit
Fig. 2 left: Sensor chip with ISFET, IDES, temperature and photo diode; right: chips are mounted in a 40-pin IC-chip case and encapsulated with epoxy
achieve sufficient high cell densities. For a convenient handling, the cells are usually grown on sensor chips attached to small cell culture dishes. Two different test versions for sensor chips have been realized: The first one is a glass chip with outer dimensions of 24.0 × 33.8 × 0.5 mm and a sensor-/culture-area of about 100 mm2. It fits into a 2-channel setup, which is typically mounted on the stage of an inverted light microscope for visual control and documentation of the cell cultures (Fig. 1). This setup was designed for research applications, where an optic access to the cell cultures is important for comparative use of fluorescent probes, e.g. for the determination of cytoplasmic pH, intracellular Ca2+ levels or alterations of cellular membrane potentials. The glass chip was realized by hybrid integration of 3.5 × 7.2 mm silicon chips containing pH-sensitive ISFETs into the 24.0 × 33.8 mm glass chip with IDES, pO2sensors and temperature-sensors. Nowadays, the performances of these sensors on the glass chip are tested in combination with cell cultures. The second test version comprises a silicon sensor chip with outer dimensions of 7.4 × 7.4 mm mounted on a standard 40pin IC socket (Fig. 2) with a sensor-/culture-area of about 15 mm2. This is important, if only small cell numbers are available (e.g. from biopsies). All fabrication steps of this silicon chip are close to CMOS (Complementary Metal-Oxide-Silicon)-processing techniques. The current design essentially includes pH-sensitive ISFETs, IDES, photodiodes and temperature diodes. Extracellular recording of cellular signals by pH-ISFETs is associated with ion fluxes across cell membranes [5–7]. While proton selectivity is achieved by amphoteric gate insulator materials like Si3N4 or Al2O3, selective measurement of other ions like Na+ or Ca2+ requires the deposition of additional (e.g. polymeric) membranes, incorporating dissolved selective ionophores. However, the stability and biocompatibility of such membranes still remain to be improved. The dimensions of sensitive gate areas can be adapted to specific purposes. Up to now, areas reaching from 6 µm2 to 6000 µm2 were used. In the test version shown in Fig. 1, both pH-ISFETs and miniaturized glass electrodes can be used for
records of bulk extracellular acidification. However, information about pH-values in the cellular microenvironment is only obtained by use of planar sensors in direct vicinity of the cells. Records of cellular oxygen uptake were performed using miniaturized clark-type oxygen electrodes with small electrode current (≈50 pA/mg L–1), positioned in the fluid exit port of the cell culture chambers (Fig. 1). Using bypass fluid ports, the electrodes are alternatingly in contact with medium slightly consumed by cellular metabolism and fresh, air-saturated medium having a constant oxygen partial pressure. A simplified recalibration mode of sensors like this one was found to be favorable for long term monitoring. Since the use of conventionally manufactured electrodes is not usable for advanced screening applications, planar oxygen microelectrodes suited for on-chip integration and reproducible mass fabrication are currently adapted for long term monitoring in cell culture media, requiring stable sensor performance for at least one week. Impedimetric analysis on adherently growing cells by interdigitated electrode structures (IDES) provides information related to cell number, cell adhesion and cellular morphology [8, 9]. The flow of alternating current between the electrodes is influenced by the presence and structural properties of living cells growing on the electrode structures. Electrode structures are deposited on materials like glass or silicon. Up to now, sensitive areas ranging from about 4 to 25 mm2 were used. In Fig. 2 (left side) two trapezoid IDES, each with 4.2 mm2 are visible. The width of the electrodes (palladium or platinum) and the distance between was 50 µm. Impedance measurements were performed at 10 kHz and the current was reduced to below 1 µA. Complex impedance values can be specified in several equivalent ways. An equivalent circuit consisting of a conductance and a capacitance in parallel was chosen. For the experiments RPMI-medium (Rosswell Park Memorial Institute) with a weak buffer capacity in order to accelerate extracellular acidification rates (5% fetal calf serum, 1 mm HEPES buffer) was used. At the end of the experiments cells were usually killed by application of Triton X-100 (0.1%) in order to obtain base-line sensor signals.
32
Results All materials and sensor chips in direct contact to the cells have been tested for biocompatibility. There have been no problems concerning growing and adhesion of the cells. This behavior has been characterized by ISO 10993–5 and scanning electron microscopy. Figure 3 shows an example of a colon adenocarcinoma cell line (LS 174T) directly growing on the gate of an ISFET. Figure 4 shows two records from a silicon sensor chip being in contact to a cell culture of LS 174T: One record of a pH-sensitive field effect transistor and the other one of an IDES (see Fig. 2). Due to the switching cycle of the fluid perfusion system, the ISFET plot displays periodicity and the amplitudes are taken as a proportional mea-
Fig. 5 Application of cytochalasin B (A) and D (B) on LS 174T cells. Impedance and metabolic data were obtained in parallel. At times a), b) and c) photographs of the cell culture on the chips were taken (see Fig. 6) Fig. 3 Scanning electron micrograph of LS 174T colon adenocarcinoma cell line on gate of an ISFET
Fig. 4 Records of pH and impedance on the silicon chip (Fig. 2) in contact to a monolayer of LS 174T cells. Effects on the inhibition of the glycolysis was tested with 100 µM iodoacetate
surement of extracellular acidification in the corresponding time interval. At the time indicated, iodoacetate (an inhibitor of the enzyme glycerinaldehyde-3-phosphatedehydrogenase which is involved in glycolysis) has been added. The ISFET senses a marked step of pH, proving an acidic microenvironment in the small extracellular space between cell and sensor surface. This acidic compartment (pH about 6.8) disappeared rapidly, when glycolytic acid formation was inhibited. The IDES-record shows significant changes in cell adhesion/cell morphology induced by iodoacetate, which might contribute to the origin of the observed pH-step as well. The following experiments have been performed using the setup shown in Fig. 1. Drugs acting on the cytoskeleton demonstrate the connection between cellular morphology and impedance records. Cultures of LS 174T cells
33 (a) just before drug addition
(b) 50 minutes after addition
Fig. 6 Photograph of cell culture unit at times indicated with a), b) and c) in Fig. 5. (a) shows a complete monolayer of cells before drug addition; (b) shows the same detail 50 min after drug addition. A rounding up of the cells and a shrinking of the cytoplasma is visible. 120 min after drug withdrawal (c) the situation is nearly the same as before drug addition. The microscopical images correspond well with the data of Fig. 5
have been exposed to 1 µg/mL cytochalasine B and D (Fig. 5). The impact on the actin cytoskeleton is accompanied by a remarkable change of cellular morphology, which was monitored by IDES-sensors and confirmed by simultaneous video microscopy (Fig. 6). Apart from acting on the cytoskeleton, cytochalasine B is known to inhibit the glucose transport system across the cell membrane. Therefore, the observed strong effects on extracellular acidification and respiration are easily explained by a shift of cellular energy metabolism from glycolysis to, e.g. oxidative utilization of glutamine. Surprisingly, the results clearly show comparable side effects of cytochalasine D on energy metabolism, although less pronounced. Additionally, the dynamic responses of metabolic rates (in contrast to the effects sensed by IDES) are different with regard to the two subtypes of cytochalasine, suggesting different kinetics of binding to and dissociation from the cellular target, presumably the Na+/glucose transport protein. Phorbol 12-myristate 13-acetate (PMA) is an agonist of protein kinase C (PKC), which is known to act on the organization of tight junctional structures [10, 11]. While effects on metabolic activity have not been evident, very pronounced changes of the impedimetric signal occurred (Fig. 7). This demonstrates a strong influence of intercellular junctional structures in epithelial cell sheets on impedance values monitored by IDES-sensors. They are in
Fig. 7 Application of PMA on LS 174T cells. Only the impedance showed significant effects in this experiment. There were no effects on the cellular acidification and respiration
(c) 120 minutes after drug withdrawal
accordance with an initial increase of transepithelial resistance, induced by PKC-mediated protein phosphorylation. It should be noted, however, that IDES-sensors not only respond to changes in transepithelial resistance, but also to changes in cell-substrate adhesion. Another typical cellular behavior detectable by IDES and worth mentioning is the response to periodic medium exchange (5 min flow/5 min stop), as can be seen most distinctly on the Rpar-record. Although not completely investigated, this behavior probably reflects a cellular response to fresh medium, which is not solely linked to small variations of temperature and fluid streaming.
Discussion For the interpretation of drug effects on living cells (particularly those of new drug candidates) it is important to acquire as much data as possible. Primary cellular targets of drugs under investigation may not be closely related to any single parameter used for convenience. On the other hand, side effects are probably detected more easily by multiparametric assays. Monitoring dynamic features of drug effects is particularly helpful in cellular pharmakokinetics, e.g. in analyzing drug uptake, early events of drug action or reversibility of drug effects. In contrast to endpoint methods, it is feasible to evaluate relative changes in cellular behavior in the course of the experiment, making absolute cell numbers less important. Miniaturization is favorable for applications, where only small numbers of cells or limited amounts of drugs are available. On the other hand, analyzing single, dispersed cells deprived of intercellular contacts means an extremely artificial situation for many cell types derived from multicellular organisms.
34
ton translocating systems located in the basolateral membrane (e.g. H+/lactate symport) [18, 19].
Conclusion
Fig. 8 Outlook: Concept of a 96-well-microtiterplate equipped with multiparametric sensor chips for parallel data acquisition
The relationship between microsensor signal changes (as well as the endpoints of numerous cytologic assays) and the exact cellular event is rarely straightforward. Multiparametric data recording proves to be useful for the interpretation, particularly in combination with specifically acting biochemical inhibitors. Monitoring of cellular oxygen exchange can be employed for the assessment of both mitochondrial and photosynthetic activities. Other sinks or sources of oxygen do not contribute significantly in most cell types. Mitochondria, apart from supplying the cell with energy and membrane lipids, are suggested to be involved in cell aging, accelerated by reactive oxygen species [12, 13], in apoptosis [14] and in the toxic mechanisms of numerous anticancer drugs [15]. Monitoring mitochondrial activity by lipophilic dyes such as rhodamine 123 suffers from the risk of potential artifacts [14] and therefore, non-toxic and non-invasive techniques are particularly valuable. Microsensor-based pH-records are predominantly used for determinations of extracellular acidification rates. Several metabolic pathways contribute to extracellular acidification and thus, unlike respiration, it can be regarded as a global indicator of metabolic activity [2, 4, 16, 17]. In this sense, fast changes (time range of minutes) tend to reflect cellular activation events (e.g. receptor mediated signaling) while slow changes (time range of hours or longer) are usually attributable to cell growth/cell proliferation or cell death, respectively. By growing epithelial cells on solid surfaces, a basolateral compartment is formed that is separated from the bulk (“apical”) compartment by intercellular connections (e.g. tight junctions). If the surface includes pH-sensitive structures, variations of proton diffusional resistance across these connections can be monitored. Sufficiently small pH-sensitive structures (e.g. ISFETs with gate areas of 6 × 1 µm) should admit the detection of pH variations between a few cells and the surface. The acidity of the basolateral compartment is predominantly influenced by pro-
It could be shown that pH-ISFETs are sensor devices suited for on-line cellular acidification measurements. Changes in the extracellular acidification rates caused by the addition of drugs to the medium can be measured on-line and non-invasively. Variations of cell adhesion, cell morphology and intercellular junctions are detectable by impedance measurements with IDES. In this work, the cointegration of ISFETs and IDES on the same chip is shown to be useful for the detection of both cell metabolic and cell physiological responses to drugs. For the interpretation of drug effects on living cells the correlation of different parameters is important. On-line monitoring with microsensors for cellular acidification, respiration and adhesion seems to be particularly helpful in cellular pharmakokinetics, e.g. in analyzing drug uptake, early events of drug action or reversibility of drug effects. Interestingly, cells undergoing apoptosis in vivo are reported to loose attachment to their neighbors and to their substratum in an early stage of the process [20] while in vitro, this change was demonstrated by the loss of attachment of adherent cells to the tissue culture surface [21], which should be detectable by IDES-structures. Since no metabolic activity is involved in the process of signal generation, it does not require any microchamber and fluid perfusion system. For numerous cell-based assays this might be a substantial advantage. However, the principle of measurement restricts applications to adherently growing cells. After successful completion of single, multiparametric sensor chips, the realization of chip arrays is a logic continuation. In cell biology, the widespread use of 96-wellmicrotiterplates is certainly due to the ease and efficiency of their handling in experimental protocols. Figure 8 shows a first draft of a microtiterplate equipped with sensor chips for cellular screening. Parallel performance of the same experiment is the only way to analyze data by statistical methods, and thus to prove reproducibility of results. Technically demanding fluidic and data acquisition systems might complicate the setup of multiplexed microsensor arrays, making them less suited for massive high throughput screening applications. However, the technology might become an attractive way to reduce expenditure in settings of late drug screening (i.e. functional screening) in pharmaceutical research, which is currently the domain of animal tests. Acknowledgement We want to thank Dr. W. Oelssner, Dr. S. Herrmann, Prof. Dr. H. Kaden (Kurt-Schwabe-Institut für Messund Sensortechnik e.V., Meinsberg, Germany), G. Sulz, Ch. Vogel, Dr. K. Seibert (Institut für physikalische Messtechnik der Fraunhofer-Gesellschaft, Freiburg, Germany), Dr. H. G. Gahle, G. Igel, Dr. J. Giehl, Dr. U. Sieben (Micronas GmbH) for technical cooperation. These investigations were kindly supported by grants of the “Bundesministerium für Bildung und Forschung” (0310856) and by grants of Micronas GmbH.
35
References 1. Kraus M, Wolf B (1995) Structured Biological Modelling – A New Approach to Biophysical Cell Biology. CRC Press Inc., Boca Raton, Florida 2. Wolf B, Brischwein M, Baumann W, Ehret R, Kraus M (1998) Biosensors and Bioelectronics 13: 501–9 3. Wolf B, Brischwein M, Baumann W, Ehret R, Henning T, Lehmann M, Schwinde A (1998) Tumor Biology 19: 374–383 4. Brischwein M, Baumann W, Ehret R, Kraus M, Lehmann M, Wolf B (1998) In: Ehrfeld W (ed) Microreaction Technology: Proceedings of the first international conference on microreaction technology, Springer, Berlin, p 251–258 5. Baumann W, Lehmann M, Bitzenhofer M, Schwinde A, Brischwein M, Ehret R, Wolf B (1999) Sensors and Actuators B 55: 77–89 6. Lehmann M, Baumann, W, Brischwein M, Ehret R, Kraus M, Schwinde A, Bitzenhofer M, Freund I, Wolf B (2000) Biosensors and Bioelectronics 15: 117–124 7. Schubnell D, Lehmann M, Baumann W, Rott F, Wolf B, Beck C (1999) Biosensors and Bioelectronics 14: 465–472 8. Ehret R, Baumann W, Brischwein M, Schwinde A, Stegbauer K, Wolf B (1997) Biosensors and Bioelectronics 12: 29–41 9. Ehret R, Baumann W, Brischwein M, Schwinde A, Wolf B (1998) Med and Biol Engin and Computing 36: 365–370
10. Citi S, Denisenko N (1995) J Cell Sci 108: 2917–2926 11. Tai YH, Flick J, Levine SA, Madara JL, Sharp GW, Donowitz M (1996) J Membrane Biol 149: 71–79 12. Sohal RS, Weindruch R (1996) Science 273: 59–63 13. Yakes FM, Van Houten B (1997) Proc Natl Acad Sci USA 94: 514–519 14. Green DR, Reed JC (1998) Science 281: 1309–1312 15. Mancini M, Sedghinasab M, Knowlton K, Tam A, Hockenbery D, Anderson BO (1998) Annals of Surgical Oncology 5: 287– 295 16. Owicki JC, Parce JW (1992) Biosensors and Bioelectronics 7: 255–272 17. McConnel HM, Owicki JC, Parce JW, Miller DL, Baxter GT, Wada HG, Pitchford S (1992) Science 257: 1906–19012 18. Poole RC, Halestrap AP (1993) American J Physiol 264: C761C782 19. Bischof G, Cosentini E, Hamilton G, Riegler M, Zacherl J, Teleky B, Feil W, Schiessel R, Machen TE, Wenzl E (1996) Biochim Biophys Acta 1282: 131–139 20. Kerr JF, Winterford CM, Harmon BV (1994) Cancer 73: 2013– 2026 21. Lock RB, Galperina OV, Feldhoff RC, Rhodes LJ (1994) Cancer Research 54: 4933–4939
