Anal. Chem. 1998, 70, 1164-1170
Single-Cell Measurements of Purine Release Using a Micromachined Electroanalytical Sensor Craig D. T. Bratten,†,‡ Peter H. Cobbold,‡ and Jonathan M. Cooper*,†
Bioelectronics Research Centre, Department of Electronics and Electrical Engineering, University of Glasgow, G12 8QQ, U.K., and Department of Human Anatomy and Cell Biology, University of Liverpool, Liverpool, L69 3EG, U.K.
To study the cellular events surrounding the formation of purines in cardiac ischemia, we have micromachined a micrometer-scale titer chamber containing an integrated electrochemical sensor, capable of measuring analytes produced by a single heart cell. The analytical procedure involves the determination of metabolites via the amperometric detection of enzymically generated hydrogen peroxide, measured at a platinized microelectrode, poised at a suitably oxidizing potential, equivalent to +420 mV vs Ag|AgCl. Signals were recorded as current-time responses and were integrated to give a total charge (Q) attributable to the reaction under investigation. The amount of analyte produced by the cell was subsequently quantified by the addition of a known amount of calibrant. As a consequence, by using a cascade of three enzymes (adenosine deaminase, nucleotide phosphorylase, and xanthine oxidase), we were able to show that, after rigor contracture had been induced in a single myocyte, adenosine (but not inosine) only reached the extracellular space after the cell membrane had been permeabilized by detergent. These data, which could only be obtained unambiguously by using this single-cell methodology, have provided us with information on the origin of ischemic adenosine which challenges the established assumption that purine release is an early retaliatory response from intact anoxic myocytes. The study of single cells has generated data which could not have been obtained using traditional in vitro methodologies. For example, the measurements of cytoplasmic Ca2+ and ATP in single isolated cardiomyocytes have previously revealed new temporal relationships during metabolic inhibition which were not apparent from studies involving cell populations or the whole heart.1-4 In the understanding of myocardial physiology, there is a considerable interest in the role of the release of adenosine, a vasodilatory †
University of Glasgow. University of Liverpool. (1) Cobbold, P. H.; Bourne, P. K. Nature 1984, 312, 444-446. (2) Allshire, A.; Piper, H. M.; Cuthbertson, K.S. B.; Cobbold, P. H. Biochem. J. 1987, 244, 381-385. (3) Bowers, K. C.; Allshire, A. P.; Cobbold, P. H. J. Mol. Cell. Cardiol. 1992, 24, 213-218. (4) Allue, I.; Gandelman, O.; Dementieva, E.; Ugarova, N.; Cobbold, P. H. Biochem. J. 1996, 319, 463-469.
catabolite of ATP5-7 which is believed to impair cell recovery upon reoxygenation by reducing the availability of intracellular purine for the resynthesis of ATP. To investigate more closely the processes governing the release of purines during cardiac ischemia, we have developed a single-cell measurement technology, consisting of an integrated three-electrode microsensor system, microfabricated within a surface micromachined measurement chamber. The reaction volume inside the device is ∼600 pL, defining a titer chamber for the measurement of femtomole quantities of the purines (adenosine and inosine) generated by single rat cardiomyocytes. Previously established single-cell measurements in biomedicine, including both patch clamp and fluorescence techniques, have been used for the measurement of ion flux through membranes. More recently, electrochemical techniques have also been developed for studying single cells using carbon fiber microelectrodes.8-10 In one such case,10 these techniques have been used to complement traditional single-cell fluorescence measurements, providing elegant and corroborative data which can provide “downstream” information on metabolic processes, resulting from changes in ion transport across membranes. Against this background,1-10 many of the arguments both for and against the development of a new single-cell measurement technology involving micromachined sensors have already been established. For example, single-cell analysis has already shown that it has the potential to deconvolute complex patterns of messenger production with a knowledge of the cell’s history, providing the investigator with a high degree of confidence as to the quantitative nature of the dose-response characteristics. Against this, it can be argued that single cells are less relevant as in vitro models, as they are not influenced by their neighbors, and because they are exposed to greater amounts of (substrate) surface area and/or solvent. Currently, there is an interest among pharmaceutical companies in developing micromachined devices, not least because they offer the prospect of obtaining a unique insight into the interactions between cell signals in a manner which is not
‡
1164 Analytical Chemistry, Vol. 70, No. 6, March 15, 1998
(5) Schrader, J.; Haddy, F. J.; Gerlach, E. Eur. J. Physiol. 1977, 369, 1-6. (6) Berne, R. M. Circ. Res. 1980, 47, 807-813. (7) Jennings, R. B.; Steenbergen, C. Annu. Rev. Physiol. 1985, 47, 727-749. (8) Paras, C. D.; Kennedy, R. T. Electroanalysis 1997, 9, 203-208. (9) Swanek, F. D.; Chen, G. Y.; Ewing, A. G. Anal. Chem. 1996, 68, 39123916. (10) Finnegan, J. M.; Wightman, R. M. J. Biol. Chem. 1995, 270, 5353-5359. S0003-2700(97)00982-7 CCC: $15.00
© 1998 American Chemical Society Published on Web 02/18/1998
possible when neighboring cells are influencing the local environment. Importantly, in the case of the study of heart cells in vitro, such local interactions may include the release of chemical species due to mechanical stresses between asynchronously contracting cells. One potential problem which may be associated with using low volumes of cell media, as is the case in these experiments, is that, with time, the concentration of metabolites may increase above physiological levels. However, in the context of the relatively short periods over which the experiments described in this paper were conducted, it was considered unlikely that this would significantly affect the responses of the cell. Notwithstanding this, by using microfabrication and micromachining technologies to reduce the analysis volume, the device that we have developed has a number of advantages over traditional methods for studying pathophysiologies involving either cell culture or whole tissue technologies. In particular, because the diffusion lengths within our device are reduced (i.e., the distance between the cell and the sensor is small), responses become fast and no analyte is lost to bulk solution (with the flux of metabolites, produced by the cell, remaining high). One important innovation has involved the development of a microsystem with a transparent glass base, which has provided us with a means to observe the cell simultaneously by light microscopy during the electroanalytical measurement, thus allowing defined metabolic events to be related directly to observed cellular changes. In the case of the single myocyte used in this experiment, this is of particular relevance, as the cell undergoes significant physical changes in its appearance, which can be correlated to biochemical/pathophysiological changes during ischemia.
MATERIALS AND METHODS Micromachining of Devices. Devices were micromachined using photolithographic techniques, adapted from the semiconductor industry and described by us in a previous publication.11 In brief, photocurable polyimide was spin-coated onto a planar fabricated Au microelectrode array and used to define the micrometer-scale wells with a diameter of 200 µm and a depth of 20 µm (see Figure 2). After fabrication, platinum was electrodeposited onto the gold microelectrodes, providing a stable electrochemical surface for the low-overpotential detection of H2O2.12 Sensors were used either in a standard three electrode configuration, with a silver|silverhalide reference,11 or by using a combined reference and counter electrode, acting as a pseudoreference electrode. Determination of Purines. For the determination of purines, a mixture of adenosine deaminase (AD), nucleoside phosphorylase (NP), and xanthine oxidase (XOD) was microinjected into the device through a thin film of mineral oil in order to prevent the medium from evaporating. In the presence of adenosine, there was an enzymic production of a stoichiometric quantity of H2O2 from adenosine (eqs 1-4), which was subsequently measured (11) Bratten, C. D. T.; Cobbold, P. H.; Cooper, J. M. Anal. Chem. 1997, 69, 253-258. (12) Johnson, K. W.; Bryanpoole, N.; Mastrototaro, J. J. Electroanalysis 1994, 6, 321-326.
amperometrically, at the platinized working electrode (eq 5), poised at +420 mV vs Ag|AgCl. These enzymes catalyzed the following reaction cascade:
AD: NP:
adenosine f inosine + NH3
inosine f hypoxanthine + ribose-1-phosphate XOD: XOD:
(1) (2)
hypoxanthine f xanthine + H2O2
(3)
xanthine f uric acid + H2O2
(4)
H2O2 f O2 + 2H+ + 2e-
(5)
Data were recorded using a low-current potentiostat with in-house PC-based data acquisition. Responses of the device were measured upon addition of adenosine to an optimized enzyme mixture containing 150 units mL-1 adenosine deaminase, 150 units mL-1 nucleotide phosphorylase, and 75 units mL-1 xanthine oxidase in 10 mM HEPES-buffered Ringer’s solution, pH 7.4. The amount of the purine present was estimated from the total charge passed (measured by integration of the i-t response), as shown in Figure 1. Single-Cell Measurements. Single cell measurements were made in HEPES-buffered Ringer’s solution (125 mM NaCl, 2.6 mM KCl, 1.2 mM KH2PO4, 1.2 mM MgSO4, 1 mM CaCl2, 10 mM HEPES; pH 7.4) covered with paraffin oil and mounted onto a heated Faraday stage (37 °C). The device was viewed under a stereomicroscope (MZ6, Leica, UK, at ×100). Individual myocytes, isolated by collagenase perfusion,13 were selected by micropipet, placed into the droplet, and allowed to settle into the chamber. The enzymes were prepared as a stock solution, in the same buffer, containing 10 mg mL-1 fatty acid-free bovine serum albumin. To compare results, all responses were normalized against a calibrant containing 70 fmol of adenosine injected at the end of each measurement (e.g., see Figures 3-6). As a model for anoxia, the single cell was poisoned by microinjecting 10 mM 2-deoxy-D-glucose (2-DOG) and 10 µM carbonylcyanide p-(trifluoromethoxy)phenylhydrazone (FCCP) into the chamber. In many cases, at the end of the experiment, the cell membrane was permeabilized by the addition of the detergent saponin (∼50 mg mL-1) into the microanalytical chamber to measure the cell’s contents. Reagents. Reagent grade hydrochloric acid, chlorobenzene, diethyl ether, acetic acid, and acetone were from Aldrich (Gillingham, UK). The enzymes adenosine deaminase, nucleotide phosphorylase, and xanthine were used as supplied from Sigma (Poole, UK). Saponin, 2-deoxy-D-glucose (2-DOG), carbonyl cyanide p-(trifluoromethoxy)phenylhydrazone (FCCP), adenosine, inosine, lithium perchlorate, sodium thiosulfate, glacial acetic acid, and Decon 90 were also from Sigma. Probimide 7020 photopatternable polyimide and OCG developer were from OCG (UK). 1400-31 photoresist was from Shipley (UK). Metals, including Au, Cr, Ni, Ti, and Pd, were from Goodfellows (Cambridge, UK). All chemicals were used as received. RESULTS AND DISCUSSION Measurement of Adenosine. Using the stated ratios of adenosine deaminase, nucleotide phosphorylase, and xanthine (13) Powell, T.; Twist, V. W. Biochem. Biophys. Res. Commun. 1976, 72, 327333.
Analytical Chemistry, Vol. 70, No. 6, March 15, 1998
1165
a
b
c
Figure 1. (a) Representative responses of the device to addition of adenosine to 10 mM HEPES-buffered Ringer’s solution, pH 7.4: [1] injection of ∼70 pL of buffer as control for other addition artifacts; [2-4] giving final concentration of adenosine as 1.66, 13.3, and 110 µM, respectively. H2O2, produced enzymically (eqs 1-3), was detected electrochemically according to eq 4, as described in the text. (b) An example of a calibration curve showing a linear response up to 130 fmol (∼220 µM) of adenosine. Responses remained linear over the time course of a cell experiment (r ) 0.98). (c) Trace displaying selectivity of detection protocol: [1] injection of AD/NP/XOD to 150, 150, and 75 units mL-1, and then ∼70-pL additions of 1 mM IMP [2], AMP [3], ADP [4], ATP [5], adenosine [6], and inosine [7]. 1166 Analytical Chemistry, Vol. 70, No. 6, March 15, 1998
oxidase within the enzyme cascade, described above, the microanalytical device gave responses (Figure 1a) which produced calibration curves (Figure 1b) showing good linearity up to concentrations of ∼220 µM (∼130 fmol of adenosine), an amount which is more than the maximum predicted purine generation from a single myocyte.14 The precision of these measurements was limited by two major sources: errors due either to pipetting or to the surface (electrochemical) condition of the electrodes. The standard error was estimated as (7.9% by comparing a geometrical quantification of the amount of the aliquot injected (measured under a microscope) with a series of successive electrochemical titrations of hydrogen peroxide, made using the same microstructured devices. The detection limit for adenosine was determined as 0.2 µM (measured as 2 SD above baseline noise), equivalent to ∼120 amol of analyte. The selectivity for the measurement of adenosine was confirmed by measuring the responses to the addition of an equivalent amount (70 fmol) of potential interferences, including ATP, ADP, AMP, and IMP (Figure 1c). The broadness of the i-t curves, characteristic of the measurement of adenosine, shown in Figure 1a, was in contrast with the temporally shorter, and hence sharper, “spike” responses which resulted from the addition of a number of potential interferences (Figure 1c) and which may (in part) be attributable to perturbations caused to the electrode double layer, due to the action of pipetting. Additional controls, performed in the presence of cells, were also carried out to exclude the possibility of intracellular electroactive species contributing to the response (see below). Calibration of Responses. Microanalytical chambers which had been fabricated with platinized electrodes proved to give more reproducible results than those with evaporated gold electrodes, although inevitably there was intradevice variability, attributable to the history of the electrochemical surface (including, for example, the electrode’s age, the amount of protein adsorbed, and the electrochemical potentials previously applied). Replatinization of the electrodes was used to regenerate the electrochemical interface, but this process also increased the surface area of the device, hence producing variability in both signal and background currents. To quantify biological measurements, prior to each single-cell analysis, the device being used was calibrated using a series of standard additions of adenosine (a representative example of this process is shown in Figure 1a and b). After cell measurements had been made, the experimentally measured results were quantified by determining the charge generated as a consequence of the addition of a known calibrant, containing 70 fmol of adenosine. By using this method, imprecision could be attributed either to nonlinearity of the sensor over the relevant dynamic range of analyte (in which case the device could be discarded prior to measurement) or to substantial changes to the character of electrochemical surface during the cell measurement (which were minimized due to the relatively short measurement periods). Single-Cell Measurements. The rat cardiomyocyte was placed into the measurement chamber using a micropipet and was manipulated manually into a central position (Figure 2a). As a model for anoxia, the cell was metabolically inhibited by microinjecting 10 mM 2-DOG and 10 µm FCCP into the microme(14) Allen, D. G.; Orchard, C. H. Circ. Res. 1987, 60, 153-168.
Figure 3. Current-time (i-t) traces showing myocyte purine generation. (a) Adenosine generation from the myocyte following rigor contracture is only measurable after saponin (∼50 mg mL-1) has been injected into the microwell, to permeabilize the cell membrane. The trace also shows the addition of a calibrant of adenosine (70 fmol), which is used directly to quantify the size of the recorded coulometric response. (b) Adenosine generation of healthy, rod-shaped cell, after being permeabilized with saponin (∼50 mg mL-1). Again, the trace shows the addition of 70 fmol of the adenosine calibrant.
Figure 2. (a) Use of the nanovolumetric devices to investigate purine generation by a single myocyte. The figure shows a view of the circular platinized working electrode, situated at the bottom of the microelectrochemical chamber (diameter, 200 µm), and encircled by a (larger) counter electrode and a reference (see ref 11 for details). A myocyte was inserted into the chamber using a blunted micropipet and oriented centrally in device. The medium surrounding the cell was then injected with AD/NP/XOD to measure adenosine and inosine generation, or with NP/XOD to measure inosine only. (b) Typical appearance of a cell after rigor contracture, induced by injecting medium containing metabolic inhibitors (see text for details). Addition of saponin (to release the intracellular contents) after rigor did not induce any change in the cell shape. (c) Saponin permeabilization of unpoisoned, rod-shaped cells resulted in their immediate “roundingup”, probably as a result of large amounts of Ca2+ entry.
ter-scale chamber. After a variable length of time, the poisoned cell shortened to between 75 and 50% of its original length, but still maintained its rod-shaped morphology, in a rigor-mediated contracture process (Figure 2b). The cell membrane integrity could be breached by permeabilization with saponin, after which time the signal due to the generated adenosine could be measured (Figure 3a). In all cases, the current-time (i-t) responses, generated by formation of purines by the cell, were integrated to give the charge (Q), which was quantified by direct comparison with the integral of a signal from the addition of 70 fmol of purine calibrant. Control experiments on unpoisoned, rod-shaped cells were performed by permeabilizing the membrane with saponin (Figure 3b). This caused the cell to round-up immediately, as shown in Figure 2c, and showed that a significantly reduced amount of adenosine was released from the unpoisoned cell than was the case for the poisoned cell (Figure 3a) and Table 1. It was also noted that both of the adenosine responses (Figure 3a,b), had higher peak currents and were broader than the “spike” responses, recorded on addition of potential interferences (Figure 1c), including the response from addition of the detergent saponin, shown in Figure 4a. Although some control responses had significant peak currents, their absolute coulometric values were small compared with the experimental measurements. An additional control using poisoned, contracted single cells (Figure 4b), permeabilized in the absence of all three enzymes (AD, NP, and XOD), did not result in an electrochemical response. This result confirmed that other electroactive species, including intracellular ascorbate, as well as any other species which may be present as a consequences of metabolic inhibition, such as Analytical Chemistry, Vol. 70, No. 6, March 15, 1998
1167
Table 1. Purine Generation by Single Myocytes experiment (i) inosine generation by unpoisoned rod-shaped cell (ii) inosine generation by poisoned cell after rigor contracture (iii) adenosine generation by unpoisoned rod-shaped cell (iv) adenosine generation by poisoned cell before rigor contracture (v) adenosine generation by poisoned cell after rigor contracture
mean purine generation/fmol 3.8 ( 0.8 (n ) 4) 7.5 ( 2.5 (n ) 4) 38.0 ( 10 (n ) 8) 50.1 ( 15 (n ) 4) 160.3 ( 29 (n ) 11)
Figure 4. Current-time (i-t) traces showing myocyte purine generation. (a) Control measurement for the additin of saponin to a normal rod-shaped cell in the absence of all three enzymes (AD, NP, and XOD), indicating that the responses in Figure 3 are not due to interference from the detergent, nor from the release of electroactive species from the cytoplasm. (b) A comparable control for a poisoned contracted single cell, permeabilized by saponin in the absence of all three enzymes (AD, NP, and XOD), which again did not result in an electrochemical response due to interference from the cell’s contents.
intracellular acidosis or buildup of NADH, did not interfere with the electrochemical measurement. Inosine generation was measured using the same protocol as above (Figure 5a,b), although adenosine deaminase (but not NP and XOD) was omitted from the enzyme cascade, such that detection was via eqs 2-5 only. Data obtained showed that a greatly reduced amount of inosine was generated during the process of cellular metabolic inhibition, (Table 1, (i) and (ii)), either by a healthy cell (Figure 5b) or by a cell after rigor contracture (Figure 5a). This result indicated that the AMP deamination pathway to IMP and subsequent inosine generation by 5′-nucleotidase was not active under the conditions for cell measurement, a finding which is consistent with the belief that most purine catabolism to inosine takes place in the heart endothelia and not within the myocardial cells.15,16 Adenosine generation by isolated healthy, rod-shaped cells, when permeabilized by saponin, is considerably higher (38 ( 10 (15) Rubio, V. F.; Wiedmeier, T.; Berne, R. M. Am. J. Physiol. 1972, 222, 550555. (16) Meghji, P.; Pearson J. D.; Slakey, L. L. Am. J. Physiol. 1992, 263, H40H47.
1168 Analytical Chemistry, Vol. 70, No. 6, March 15, 1998
Figure 5. Current-time (i-t) traces showing myocyte purine generation. (a) Response of a poisoned myocyte after cell permeabilization by saponin, made in the presence of the enzymes NP and XOD (but not AD). The addition of a standard containing 70 fmol of inosine demonstrates that cellular inosine would have been measured, if present. (b) A representative trace from a control experiment showing the lack of inosine generationg by a healthy myocyte permeabilized with saponin (∼50 mg mL-1). Again, the addition of a standard containing 70 fmol of inosine demonstrated that cellular inosine would have been measured, if present.
fmol, n ) 8) than that predicted (2.6 fmol) from whole tissue studies15,17 (Table 1, (iii)). Given the constrained volume in which cells were permeabilized, we could not discount a significant amount of purine generation, either by membrane-bound ectoenzymes16 or by released cytosolic enzymes.18 Further, we would have expected that permeabilization in the presence of millimolar concentration of extracellular Ca2+ will lead to a maximal activation of myosin ATPase and may induce a catastrophic degradation of ATP (via ADP) to adenosine. Evidence supporting this was obtained when cell permeabilization was performed in a Ca2+-free medium, achieved by adding both saponin and ethylene glycol bis(β-aminoethyl ether) tetraacetic acid (EGTA) at a concentration of 10 mM. Under these conditions, there was no immediate shortening of the cell, but a Ca2+-independent rigor contracture occurred after between 30 s and 1 min, probably as a result of falling ATP levels.14 This was accompanied by a coincident start of a rise in the adenosine signal (69 fmol ( 8.1, n ) 4, e.g., Figure 6). In all experiments, there is a large variation in the time between initial metabolic inhibition and the onset of rigor contracture,19 which makes the estimation of adenosine formation during this time difficult. In two cases, however, the cell was permeabilized precisely at the point of rigor contracture. The mean adenosine produced (83 fmol) suggests a relatively small increase in (17) Frangakis, C. J.; Bahl, J. J.; McDaniel, H.; Bressler, R. Life Sci. 1980, 27, 815-825.
Figure 7. Transverse section of a schematic diagram of the physical processes occurring within a microelectroanalytical chamber containing a counter (C) and working (W) electrode, in which an enzyme cascade (E) converts substrate (S) to product (P). See text for full details. Figure 6. Delay of contracture of healthy cell due to saponin injection in Ca2+-free medium (achieved by addition of saponin with 10 mM EGTA), with the onset of adenosine generation coincident with rigor contracture.
adenosine formation, when viewed in the context of the data for pre-rigor content (50 fmol ( 15, n ) 4, Table 1, (iv)). In contrast, the generation of adenosine after rigor contracture was considerable, with the mean (160 fmol ( 29, n ) 11, Table 1, (v)) correlating closely with estimates of ATP content in normal cells.14,20 In a minority of these experiments (n ) 2), the membrane of a poisoned cell ruptured spontaneously, and a large adenosine response (mean 116 fmol) was obtained precisely coincident with contracture; addition of saponin did not yield any further signal. Thus, we have shown that the majority of adenosine generation occurs rapidly and accompanies rigor contracture, in an amount which is equivalent to the cell’s ATP content. Despite the high sensitivity of this device, no increase in baseline current was detected from intact cells, even up to 10 min after rigor contracture prior to permeabilization of the cell membrane, indicating unambiguously that the continuous release of adenosine from intact cells is very slow. Over a similar time period, whole heart perfusates already shows a significant rise in adenosine content.20-22 The amount of adenosine generated by the cell after rigor contracture correlates well to values obtained from whole heart studies, in which the tissue content of purine nucleosides has been shown to increase commensurably with the decline in ATP levels.20,24 To date, the assumption has been that the purine nucleosides, which appear in the coronary perfusate during cardiac ischemia,18,24,25 are released from intact cells via the membrane purine transporter.18,21,26,27 Here we show the unexpected and (18) Schu ¨ tz, W.; Schrader, J.; Gerlach, E. Am. J. Physiol. 1981, 240, H963H970. (19) Haworth, R. A.; Hunter, D. R.; Berkoff, H. A. Circ. Res. 1981, 49, 11191128. (20) Geisbuhler, T.; Altschuld, R. A.; Trewyn, R. W.; Ansel, A. Z.; Lamka, K.; Brierley, G. P. Circ. Res. 1984, 54, 536-546. (21) Van Belle, H.; Goosens, F.; Wynants, J. Am. J. Physiol. 1987, 252, H886H893. (22) Vanwylen, D. G. L.; Schmit, T. J.; Lasley, R. D.; Gingell, R. L.; Mentzer, R. M. Am. J. Physiol. 1992, 262, H1934-H1938. (23) Jennings, R. B.; Reimer, K. A.; Hill, M. L.; Mayer, S. E. Circ. Res. 1981, 49, 892-900. (24) Katori, M.; Berne, R. M. Circ. Res. 1966, 19, 420-425. (25) Olsson, R. A. Circ. Res. 1997, 26, 301-306.
striking observation that, for single cells, significant adenosine only reaches the extracellular space after the cell has been lysed, regardless of its metabolic condition. Not withstanding this latter point, it should be emphasized that single-cell studies provide a highly appropriate technology for conducting these investigations into purine formation, as the myocyte is free of the mechanical stresses between asynchronously (neighboring) contractions,28 which may contribute to values observed for purine generation from the whole heart. Indeed, we now speculate that a large part of the adenosine release observed in whole heart or in vitro cell population studies may originate from lysed cells. The Nature of the Temporal Response. In analyzing the temporal response of the single cells within the device, it is important to consider carefully the many processes which are occurring in the micromachined structure, and how these relate to the measurements being made. These are illustrated schematically in Figure 7, which represents a myocyte contained within a microanalytical electrochemical cell, in which hydrogen peroxide, oxygen, and the enzyme substrates all are in solution (with diffusion coefficients, DH2O2, DO2, and DS, all of which are all similar). Both the cell and the enzyme XOD (within the oxidase enzyme cascade) require oxygen, which must be partitioned across the mineral oil (KO2) into the electrochemical cell. The cell will obtain this oxygen, first by diffusion through the buffered medium (DO2) and then by transfer across its plasma membrane (K′O2). The fact that the cell can survive within the device for periods up to 2 h (the maximum tested) would indicate that the partitioning of oxygen into the chamber is greater than or equal to the equivalent term for the partitioning of oxygen across into the cell membrane (i.e., KO2 ) K′O2). In alternative assay schemes, there is the possibility that, when the detection enzyme is an oxidase, it may compete with the cell for the available oxygen. However, in the measurements that we describe in this paper, adenosine is only present after cell lysis, (26) He, M.-X.; Gorman, M. W.; Romig, G. D.; Meyer, R. A.; Sparks, H. V. Am. J. Physiol. 1991, 260, H917-H926. (27) Chen, W. N.; Hoerter, J.; Gueron, M. J. Mol. Cell. Cardiol. 1996, 28, 21632174. (28) Siegmund, B.; Koop, A.; Klietz, T.; Schwartz, P.; Piper, H. M. Am. J. Physiol. 1990, 258, H285-H291. (29) Cass, A. E. G.; Davis G.; Francis, G. D.; Hill, H. A. O.; Aston, W. J.; Higgins, I. J.; Plotkin, E. V.; Scott, L. D. L.; Turner, A. P. F. Anal. Chem. 1984, 56, 667-671. (30) Lemmo, A. V.; Fisher, J. T.; Geysen, H. M.; Rose, D. R. Anal. Chem. 1997, 69, 543-551.
Analytical Chemistry, Vol. 70, No. 6, March 15, 1998
1169
and thus it can be assumed that the cell is not competing with the enzyme for oxygen and that the detection reaction is able to go to completion. That the enzyme cascade can be used to measure adenosine while the cell is still alive, as evidenced through a standard addition of analyte, supports the general conclusion of this paper, namely that adenosine is only released from the myocyte after its plasma membrane is breached. The advantage of using a chronocoulometric “end-point” measurement, integrated over the whole time course of a reaction, is that it is possible to quantify the purine signal, regardless of which of a number of processes is the rate-determining step (for example, either the electrochemical oxidation rate constant for the hydrogen peroxide, ks, or the kcat for the enzyme cascade, E, may be rate determining, although in either case the experimental result should be identical, as no analyte is lost to the bulk solution). That this “end-point” is reached relatively quickly (Figures 3-5) is due to the fact that the diffusion distances in the device are short. Despite the constraints imposed on the types of measurements which can be made using these devices (see above), it should be noted that this technology has a potential to further extend the range of compounds that can be detected from single cells to include the purine nucleosides. By incorporating different enzyme systems, these microsystems can be readily adapted to measure a wider range of analytes from single cells, including glucose, lactate, and glutamate. Although, in the future, there is the possibility of adding a mediator29 to negate oxygen dependency,
1170 Analytical Chemistry, Vol. 70, No. 6, March 15, 1998
it was felt that, at this stage, this would detract from the overall simplicity of this detection method described. Since the microelectroanalytical chambers are fabricated using protocols adapted from standard photolithographic methods, there is the potential to reduce the dimensions of the device further and thus provide structures that are more appropriately sized for smaller cell types, giving higher fluxes of analyte (perhaps enabling faster response times). There is also the possibility of having more than one sensor incorporated within each microstructure, to allow for detection of more than one analyte, simultaneously. And finally, in the longer term, we will produce arrays of these devices and interface them with technologies for the dispensation of low volumes of fluid; e.g., “ink-jet” technology has already been used in order to develop high-throughput secondary screens for drug discovery.30 Using such microfluidic dispensations, there will be an ability to control more reproducibly the position of the microcapillary tip when adding fluids and hence minimize (and quantify) the nonFaradaic changes in the electrode double layer (seen as “spikes”) resulting from perturbances in the medium during pipetting. Acknowledgment is given to both the Wellcome Trust and to Glaxo-Wellcome for supporting aspects of this research.
Received for review September 8, 1997. January 4, 1998. AC970982Z
Accepted
