Developments toward a Microfluidic System for Long-Term Monitoring

Anal. Chem. 2004, 76, 4715-4720

Developments toward a Microfluidic System for Long-Term Monitoring of Dynamic Cellular Events in Immobilized Human Cells Richard Davidsson,† A ° ke Boketoft,‡ Jesper Bristulf,§ Knut Kotarsky,§,⊥ Bjo 1 rn Olde,§ Christer Owman,§ | | ,† Martin Bengtsson, Thomas Laurell, and Jenny Emne´us*

Department of Analytical Chemistry, Lund University, P.O. Box 124, SE-22100 Lund, Sweden, HeptaHelix AB, Ja¨gershillsgatan 15, SE-23175 Malmo¨, Sweden, Department of Physiological Sciences, Division of Molecular Neurobiology, Lund University, BMC A12 Tornava¨gen 10, SE-22184 Lund, Sweden, and Department of Electrical Measurements, Lund Institute of Technology, Lund University, P.O. Box 118, SE-22100 Lund, Sweden

A microfluidic system for long-term real-time monitoring of dynamic cellular events of immobilized human cells was investigated. The luciferase reporter gene activity in the reporter cell line HFF11, based on HeLa cells, was used as the model system. The cells were immobilized on silicon flow-through microchips and continuously supplied with a cell medium at 2 µL/min while maintaining the chip at 37 °C. The HFF11 cell line was designed for high-throughput screening of ligands for seven-transmembrane receptors. When a ligand binds, the receptor is activated and a cascade of intracellular reactions starts, ending with the synthesis of the reporter protein Photinus luciferase. The major goal was to develop a microfluidic system for continuous long-term assaying of the intracellular reporter gene activity in real time and determine the conditions, which could minimize cells stress and hence unspecific expression of the reporter gene. In the resulting microfluidic system and assay protocol, the cell microchip could be kept and assayed for a period up to 30 h. The developed system and data outcome was compared with a corresponding microtiter plate performed with the same cell line to highlight the advantages obtained in the microfluidic format. Cell-based assays are attracting increased attention fueled by advancement in drug discovery for finding novel gene sequences likely to be therapeutic targets1,2 as well as evaluating drug candidates at an early stage.1,3,4 Today, these assays are performed in microtiter plates with 96, 384, or even higher number of wells. The micro well format is identical to the test tube experiment, * Corresponding author. Telephone: +46-46-2224820. Fax: +46-46-2224445. E-mail: [email protected]. † Department of Analytical Chemistry, Lund University. ‡ HeptaHelix AB. § Department of Physiological Sciences, Division of Molecular Neurobiology, Lund University. | Department of Electrical Measurements, Lund Institute of Technology, Lund University. ⊥ Present address: Department of Cell and Molecular Biology, Immunology Section BMC I 13, Lund University, SE-22184 Lund, Sweden. (1) Croston, G. E. Trends Biotechnol. 2002, 20, 110-115. (2) Chanda, S. K.; Caldwell, J. S. Drug Discov. Today 2003, 8, 168-174. (3) Umezawa, Y. Rev. Mol. Biotechnol. 2002, 82, 357-370. 10.1021/ac035249o CCC: $27.50 Published on Web 07/16/2004

© 2004 American Chemical Society

where samples and reagents are added in discrete portions by manual or automatic pipetting. For cell assays this means that the liquid phase is stagnant and is continuously changing due to accumulation of waste products. Moreover, microtiter plate assays are usually performed by determining the end point after addition of a reagent since the added component cannot be removed without emptying the whole content of the well. In light of this, the emerging microchip technology can offer several features never reached in plate assays. Chip platforms can be fabricated to have spatially and chemically tailored structures in the size of single cells, which has been proposed to increase reliability and accuracy in the cell assay process due to the mimicked in vivolike milieu.5 Moreover, not only the physical structure housing the cells can be precisely controlled, but also the surrounding medium through use of microfluidic liquid handling, which allows a continuous supply of fresh growth medium and precise control over addition and removal of reagents.6 This allows continuous monitoring of the same cell population in real time, which stands in bright contrast to the microtiter plate assays with stagnant liquid and end-point measurements. An additional positive aspect of microfluidic cell handling includes diminished consumption of reagents, for example, receptor ligands, which are expensive and only available in small amounts. In the system reported in this paper we approached the challenge of developing a microfluidic system for long-term monitoring of cellular events in real time. For this purpose, a highly efficient reporter cell system (HFF11), based on genetically modified HeLa cells, specially designed for high-throughput screening of seven-transmembrane receptors7,8 was chosen as a suitable model system. Upon binding of a ligand to a membrane receptor, a cascade of intracellular reactions occurs, which ends up with the expression of Photinus luciferase (LUC). The analytical system was based on these HeLa cells, immobilized on silicon flow-through microchips inserted in a microfluidic system with (4) Giese, K.; Kaufmann, J.; Pronk, G. J.; Klippel, A. Drug Discovery Today 2002, 7, 179-186. (5) Bhadriraju, K.; Chen, C. S. Drug Discov. Today 2002, 7, 612-620. (6) Andersson, H.; van den Berg, A. Sens. Actuators B 2003, 92, 315-325. (7) Kotarsky, K.; Owman, C.; Olde, B. Anal. Biochem. 2001, 288, 209-215. (8) Kotarsky, K.; Antonsson, L.; Owman, C.; Olde, B. Anal. Biochem. 2003, 216, 208-215.

Analytical Chemistry, Vol. 76, No. 16, August 15, 2004 4715

continuous flow of the cell medium, while maintaining the temperature at 37 °C. The receptor-mediated LUC expression was followed in real time by injection of luciferin and the generated chemiluminescent signal was monitored with a photomultiplier tube. The major goal of the work was to address fundamental aspects related to microanalytical work with mammalian cells in a microfluidic system for a prolonged period of time. The system had to fulfill two major requirements: (i) sterile conditions and (ii) conditions that minimized cell stress and hence allowed accurate measurements of the signal. But it also had to evaluate differences in comparison to microtiter plate assays, not only in hardware setup but also in terms of dynamic information in receptor-ligand-mediated reporter gene expression. EXPERIMENTAL SECTION Chemicals and Reagents. The microchip surface modification reagents 3-aminopropyltriethoxysilane (APTS) 98%, glutaraldehyde (GA) 25% grade I, sodium cyanoborohydride >90%, poly-L-lysine (PL, molecular weight 70000-150000 Da), and the receptor stimulating agent adenosine 5′-triphosphate (ATP) were purchased from Sigma Chemicals Co. (St. Louis, MO). The ATP solution for receptor stimulation was prepared in a CO2 independent cell medium. Two different LUC substrate mixtures were used and prepared by dilution according to the manufacturers’ recommendations (BioThema AB, Haninge, Sweden). Mixture I was composed of ATP and luciferin in 0.1 Tris/HCl pH 7.75 with 17% (v/v) addition of a detergent (Reporter lysis buffer, Promega, Madison, WI), which was invasive by rupturing the cell membrane and thus releasing the whole cell content. Substrate mixture II consisted of solely luciferin in 10 mM phosphate buffer saline (PBS) pH 6.0. At this pH luciferin was mainly uncharged and could thereby pass the cell membrane and use the intracellular ATP in the reaction catalyzed by LUC. Microchip Preparation. The silicon microchips were fabricated by chemical wet etching described by Laurell et al.9 and Drott et al.10 To increase the internal surface area, a porous structure of the channels was achieved by anodizing in a 1:1 mixture of 40% hydrofluoric acid and 96% ethanol with a current density of 50 mA/cm2 for 5 min. The microchips had an overall dimension of 13 × 3.1 mm, consisting of 28 parallel V-grooves, 10-mm long, 100-µm wide at the top, and 71-µm deep, with each end falling into inlet and outlet basins, as seen in Figure 1b. The pitch between the grooves was 10 µm and the total volume of one microchip was approximately 1.9 µL. The folded V-structure of the microchip offers an increased attachment area and shortened diffusion distances in comparison to a regular flat surface. To make the microchip surface biocompatible, it was covalently modified with PL. Prior to any surface modification, the microchips were cleaned in a boiling mixture of H2O/NH3/H2O2 (5:1:1 by volume) for 1 min, followed by rinsing with water and then another 1-min boiling in H2O/HCl/H2O2 (5:1:1 by volume), and finally rinsed with water. The microchips were further dehydrated by rinsing in acetone and dried under a stream of air for 1 h, and then they were silanized by immersing the units in 10% (v/v) APTS (9) Laurell, T.; Drott, J.; Rosengren, L.; Lindstro ¨m, K. Sens. Actuators B 1996, 31, 161-166. (10) Drott, J.; Lindstro ¨m, K.; Rosengren, L.; Laurell, T. J. Micromech. Microeng. 1997, 7, 14-23.

4716 Analytical Chemistry, Vol. 76, No. 16, August 15, 2004

Figure 1. (a) Setup of the microfluidic system. A syringe pump was used to deliver the carrier buffer or cell medium at 2 µL/min. The reagents were injected via a six-port injection valve. A 0.22-µm filter was placed on the inlet used for filling the injection loop, thereby avoiding bacterial contaminants from entering the microfluidic system via the injected reagents. The CL signal was monitored via a computer-connected PMT, mounted right above the microchip with immobilized cells. The entire detection unit (PMT and microchip) was placed in a “black box” to avoid interference from light in the surroundings. (b) The image shows a detailed view of the microchip flow cell with its temperature control. The three insets show SEM images of the microchip at increasing magnification. The microchip V-channels are 71-µm deep, 100-µm wide at the top, and 10-mm long. The channels lead into a basin in each end, which are 1.5-mm long, 3.1-mm wide, and 71-µm deep, which makes the overall etched area 13 × 3.1 mm of the microchips. (c) The picture shows a schematic view of the signal transduction pathway in the HFF11 cell line that leads to reporter gene expression. Upon binding of a ligand, the receptor is activated and this causes a cascade of intracellular reactions, leading to promotor activation followed by DNA transcription and subsequent synthesis of the reporter protein Photinus luciferase (LUC).

in sodium-dried toluene11 for 4 h at room temperature under gentle stirring. After removal of the silanization solution, the microchips were carefully rinsed with acetone followed by 10 mM succinate buffer pH 6.0. The amino residues of the silane layer were then (11) Weetall, H. H. Methods Enzymol. 1976, 44, 134-148.

activated by placement in 2.5% GA v/v in 10 mM succinate buffer pH 6.0 for 1 h at room temperature, followed by rinsing with buffer and addition of 0.1 mg/mL PL in succinate buffer. The polymer was allowed to attach for 24 h at room temperature. To block any residual active aldehyde groups, the microchips were immersed in 0.1 M Tris/HCl pH 7.0 for 1 h followed by reduction of the imine bond in 2 mg/mL sodium cyanoborohydride in 10 mM Tris/ HCl pH 7.0. Finally, the microchips were rinsed and stored in water. The microchips were sterilized by immersion in 70% ethanol for 10 min, followed by rinsing and storing in sterile water, using autoclaved test tubes. All work was performed under sterile conditions. Cultivation and Immobilization of HFF11 HeLa Cells. The used HFF11 cell line was based on genetically modified HeLa cells.7,8 The cells were grown in Dulbecco’s modified Eagle’s medium (DMEM) with Glutamax I supplemented with 10% fetal bovine serum (FBS), 0.5% streptomycin, and penicillin at 37 °C in a 7% CO2 atmosphere. One week before chip immobilization, the cells were transferred to a CO2 independent medium (Gibco, Invitrogen Co.) and kept at 37 °C. The DMEM could not be used in the microfluidic system since it needs a CO2 atmosphere for buffering the pH. Before on-chip immobilization a cell suspension was prepared containing 106 cells/mL. The sterile PL-modified microchips were placed in a Petri dish and the cell suspension was poured over the units. The Petri dish was placed in an incubator at 37 °C and the cells were allowed to settle, resulting in a homogeneous coverage all over the microchips. Unless otherwise stated, the cells were allowed to grow on the chip surface for 3 days, after which the chips could be inserted in and assayed in the analytical system described under Microfluidic System Setup and Assay Procedures. Preparation of Chips for SEM Analysis. The cell microchips were prepared for SEM by using standard protocol: The microchips were treated with 2.5% glutaraldehyde for 2 h to fix the cells on the surface. After being rinsed with water, the microchips were dehydrated using 25, 50, 75, and 95% ethanol, 2 × 20 min in each solution, and 99.7% ethanol overnight. The last traces of water were removed by using supercritical carbon dioxide and finally the surfaces were sputtered with gold. Microfluidic System Setup and Assay Procedures. The microfluidic system used for analyzing the cell microchips is shown in Figure 1a. The CO2 independent cell medium as a flow carrier was supplied from a sp260p syringe pump (World Precision Instruments, Ltd., Hertfordshire, United Kingdom) at 2 µL/min. The reagents were introduced into the flow via a Rheodyne sixport injection valve (Rohnert Park, CA) having a 2-µL injection loop. A 0.22-µm filter (Sigma Chemicals Co.) was placed on the inlet used for filling the injection loop to avoid bacteria entering the system when injecting reagents. All units were connected with 0.13-mm i.d. PEEK tubing (Alltech, Deerfield, IL). The microchip was incorporated into the flow system via a specially designed flow cell, where the top unit was fabricated of transparent poly(methyl methacrylate); see Figure 1b. The bottom unit in the flow cell was made of aluminum and had an open interior through which 37 °C tempered water was pumped, ensuring accurate temperature of the immobilized cells. Before insertion of the microchip into the system the flow cell unit was rinsed with 70%

ethanol followed by sterile water. The microchip was placed on the bottom unit and on top a transparent silicone rubber membrane was placed to prevent leakage. Finally, the top cover was placed above and all units of the flow cell were compressed with four screws (not shown in Figure 1b). All this work was performed under sterile conditions. Before the flow cell unit containing the cell chip was connected, the flow system was flushed with 70% ethanol. Then the syringe pump with sterile carrier was connected to the system and ethanol was removed before the immobilized cells chip was connected into the flow system. The chemiluminescent signal that followed injection of LUC substrate was monitored via a photomultiplier tube (PMT, model no. HC135-01 UV to visible, Hamamatsu Photonics K. K., Japan), aligned right above the flow cell unit, collecting the outcoming light from the cell microchip. The PMT and microchip flow cell were placed in a holder to ensure correct positioning of the two units and the entire detection was performed in a “black box” to debar light from the surroundings as seen in Figure 1a. The HFF11 cell model is shown in Figure 1c. The LUC expression was monitored by either injecting substrate mixture I or II into the microfluidic system. Mixture I caused rupture of the cell membrane and corresponded to the assay format used in microtiter plates.7,8 In contrast, substrate mixture II was noninvasive and allowed on-line and continuous monitoring of the same cell microchip over an extended period of time. On-line specific induction of the reporter gene using ATP as a receptor ligand was performed in a repeated stop-flow mode according to the following: 1 mM ATP receptor-stimulating solution was injected and the flow was stopped for 2 min when the reagent reached the microchip surface. This was repeated three times in total with 3-min intervals. The whole LUC expression process (see Figure 1c) was then followed on-line by continuous injections of substrate mixture II, as described above. RESULTS AND DISCUSSION Mammalian cell lines commonly used in molecular biology and medicine research have highly specialized demands on their physical and chemical environment, which must be met in order to obtain reliable data. Most of the mammalian cell lines used in molecular biology are adherent, which means that they need to grow on a solid support in order to promote well. Moreover, the time scale of assays commonly ranges from several hours up to days during which the cells must be continuously supplied with nutrients and oxygen while maintaining a sterile environment and constant temperature. Several elegant microfluidic cell assay systems have been reported12-14 (reviewed recently elsewhere6) but these mainly show short-term assays where the sterility problem might not occur and very often the analysis is based on suspended cells.6 Moreover, electrokinetic liquid control is very popular but might not be compatible with the cells’ need for complex nutrient mediums with high ionic strength.12 The medium must contain all vital components that the cells’ need for living; however, the medium itself is a very good ground for microbial (12) Roper, M. G.; Shackman, J. G.; Dahlgren, G. M.; Kennedy, R. T. Anal. Chem. 2003, 75, 4711-4717. (13) Yang, M.; Li, C.-W.; Yang, J. Anal. Chem. 2002, 74, 3991-4401. (14) Wheeler, A. R.; Throndset, W. R.; Whelan, R. J.; Leach, A. M.; Zare, R. N.; Liao, Y. H.; Farrell, K.; Manger, I. D.; Daridon, A. Anal. Chem. 2003, 75, 3581-3586.

Analytical Chemistry, Vol. 76, No. 16, August 15, 2004

4717

Figure 2. Continuous light intensity curves registered by the PMT from nonspecific LUC expression using substrate mixture I (invasive by rupturing the cell membrane). Each curve in the figure represents a different cell chip and the notations indicate how many hours after immobilization the measurements were performed.

growth. When the medium is supplemented with an antibiotic, bacterial growth can in principle be circumvented, however, with the drawback that contamination sources during the cell preparation process might not be discovered. The need for a sterile environment and handling techniques are usually not met in ordinary analytical laboratory praxis. Without taking all the necessary precautions of sterility, the microchip was invaded by bacterial contaminants after 18 h in the microfluidic system (image not shown). However, the introduction of a 0.22-µm filter on the injection valve (as described in section 2.5 and Figure 1a) was sufficient to enable the system to be run in an ordinary laboratory without any contamination problems during 30 h of assaying the cells, as will be demonstrated below. Immobilization-Induced Stress and Cell Attachment during Long-Term Measurements. In general, alterations in the physical or chemical environment of the cells will cause a nonspecific reporter signal, which is an inherent property of all mammalian cell line based reporter gene assays.15 To identify the specific reporter signal, all unspecific activation during handling had to be investigated as far as possible. One such event is the procedure during which the cells are detached from the stock cultivation flask using EDTA treatment and then placed in suspension. This is a standard procedure to enable transfer of adherent cells into, for example, an analysis chamber such as a traditional microtiter plate or, as in this work, a microchip. For analysis of this effect, four microchips were seeded with cells and kept in a Petri dish with the cell medium at 37 °C. At different time intervals a microchip was taken from the batch and analyzed by injecting substrate mixture I (luciferin mixed with ATP and detergent), thus lysing the cells, releasing the entire cell content of LUC. Figure 2 shows the continuous signal readout from the PMT during the analysis of each chip. Four hours after immobilization the LUC activity is still considerable. As time passes, the background signal decreases; that is, the cells return to normal steady-state conditions. The stress event associated with the cell detachment from the cultivation flask could obviously not be avoided, but was dealt with by letting the immobilized cell microchips rest sufficiently long in the incubator to allow the LUC level to decline to steady-state conditions before insertion and (15) Animal cell culture: a practical approach; Freshney, R. I., Ed.; IRL Press: Oxford, 1992.

4718 Analytical Chemistry, Vol. 76, No. 16, August 15, 2004

Figure 3. SEM images of two immobilized cell chips prepared from the same batch, (a) before and (b) after 30 h keeping and assaying in the microfluidic system at 2 µL/min.

assaying in the analytical system in Figure 1a. Thus, to allow sufficient attachment of cells to the microchips and ensuring a steady-state background LUC expression from the immobilized cells before analysis, the resting time was set to 3 days as standard for all further cell microchip preparations. An important aspect is the loss of cells due to the continuous flowing environment. This was qualitatively analyzed by taking SEM images of immobilized cell microchips (prepared as described under Cultivation and Immobilization of HFF11 HeLa Cells) before and after measurements (i.e., using the substrate mixture II, leaving the cells intact). Figure 3 shows the results from one immobilization, before (Figure 3a) and after continuous assaying for approximately 30 h (Figure 3b). No major differences could be detected from these images, and this was confirmed by repeating the experiment with a second cell chip (results not shown). We conclude that the cells attach firmly to the surface and the majority of the initial cell population remains during the whole assay, which is a prerequisite for developing a reliable system. Real-Time Monitoring of the Dynamic LUC Expression Level. To investigate the possibility of long-term real-time monitoring of the reporter gene expression, a cell chip was prepared as described under Cultivation and Immobilization of HFF11 HeLa Cells and inserted into the microfluidic system. No specific receptor stimulation with ATP was performed, but with injection of substrate mixture II once every hour the background

Figure 4. The curve shows the dynamic changes of the LUC expression level of HFF11 cells on-chip using substrate mixture II (noninvasively, leaving the cells intact) directly after insertion into the microfluidic system and 13 h forward. The y axis represents the height of the PMT signal peak that followed each injection of substrate mixture II.

expression of LUC was assayed, as seen in Figure 4. It is obvious that the insertion of the cell chip into the flow system is a stressful event for the cells since it results in elevated LUC activity. This is very likely due to changes in the physicochemical conditions imposed by moving the cell chips from a Petri dish in batch culture into the microfluidic system in which the cells are continuously supplied with a fresh cell medium, avoiding accumulation and exposure to metabolic waste products. Another influencing factor could be the exposure to fluidic shear stress as the cells are moved from a stagnant environment in the Petri dish to continuous perfusion in the microfluidic system. As seen, the maximum background level was obtained at 7-8 h (Figure 4) and then declines. The variation in signal probably stems from the cells and is not related to the PMT detector; however, the exact source is difficult to decipher. There will always be variations among individual cells in luciferase expression level, and since we have not performed single-cell analysis, the output signal will always be a mean value of the whole cell population. To avoid interference from the nonspecific LUC expression, the cell microchip was allowed to adapt in the microfluidic system for approximately 1718 h before any receptor-ligand assay was performed. Thus, the microfluidic system with an inserted cell chip was started in the afternoon by continuous pumping of the cell medium and the analysis of the cell microchip was started in the morning after by continuous injection of luciferin. For all further assays this overnight handling procedure was applied. We now turned to investigate on-line receptor stimulation followed by on-line monitoring of the LUC expression. In these experiments two cell chips were compared: one for the specific response signal and another for the background activity of LUC. Thus, after one of the cell chips had spent 17 h in the microfluidic system, the background level of LUC was analyzed for 3 h, by injecting substrate mixture II. The reporter activation (see Figure 1c) was then accomplished as described under Microfluidic System Setup and Assay Procedures, by stop-flow injections of 1 mM ATP. Since the system was flow-based, the ligand could only be in contact with the immobilized cells for a very short time. This time was thus extended using a stop-flow technique for 2 min, repeated three times in total. The reason for repeating it three times was to a higher degree maintain a continuous flow format rather than injecting once and stopping the flow for a prolonged

Figure 5. Dynamic changes of specific (S) and nonspecific (NS) LUC expression levels of two cell microchips, using substrate mixture II. LUC assaying was initiated after 17 h of adaptation of the cell chips to the microfluidic system. Specific induction of the LUC reporter was effectuated on one chip by stimulating three times with 1 mM ATP, each with 2-min stop-flow (the dashed line indicates when all three stimulation injections had been performed, 20 h after insertion into the microfluidic system). Nonspecific induction of the LUC reporter was tested on the other chip with the same stop flow procedure, however, by injecting only a fresh cell medium. Since the cell microchips were prepared on different days, the obtained peak heights were made relative to the last data point (at t ) 19 h) in each set before the stop-flow injections were performed.

time period. Next, the LUC activity was assayed every hour by injecting substrate mixture II. The second cell chip was treated in exactly the same way except that ATP was not included in the stop-flow injection procedure. This would reveal whether the stopflow injections had any effect on the LUC activity or to the exposure to PBS pH 6.0 when injecting luciferin. Figure 5 shows the two response curves obtained for receptor-mediated and background expression of LUC. The first 3 h show the decline of the background activity of LUC in both curves, but after stopflow injections the cell chip treated with ATP shows a significant activity increase, while the other chip exposed to injections of pure medium shows a continuous decrease. From Figure 5 we conclude that on-line receptor stimulation and measurement of the intracellular activity can be performed successfully in the microfluidic system and that the stop-flow handling or substrate mixture II injections did not affect the cells. The time for ligand stimulation was very short (3 × 2 min) compared to the experiments performed in corresponding micro-well assays8 in which the ligand is present during the whole assay until the LUC activity is assessed by lysing the cells and addition of luciferin. In this context a feature of the microfluidic system was the ability to time control addition and removal of ligand and luciferin (substrate mixture II) by the continuous carrier flow, which allowed monitoring of the dynamic changes over an extended time. The expression was detected approximately 30 min after stimulation with ligand (see the rise of curve S after finished stimulation, dashed line in Figure 5), which was faster compared to the corresponding microtiter plate assays (expression detected after 6 h).7,8 A major difference between the developed microfluidic system and corresponding microtiter plate assay is the continuous supply of fresh cell medium in the former, which could be a factor that influences the fast kinetics in Figure 5. Finally, the instrumental setup of the PMT and the smaller diffusion pathways in the microchip compared to microtiter plates (see Figure 1a,b) could be additional factors influencing the fast response. From the experiments Analytical Chemistry, Vol. 76, No. 16, August 15, 2004

4719

(Figure 5) it seems that only a short ligand stimulation time is necessary to obtain a strong activation of the reporter system and the microfluidic system could thereby be useful for kinetic studies to follow intracellular kinetics in real time. CONCLUSIONS This paper presents the development toward a microfluidic system for long-term monitoring of a dynamic cellular event in real time. The human reporter cell line HFF11, based on adherent HeLa cells immobilized on flow-through silicon microchips, was used as the model system. The results and system design reveal the potential of microfluidic cell based assays in which the cells can easily be manipulated, for example, change of cell medium and introduction of different chemical reagents, compared to ordinary micro-well assays. The work emphasizes that special care must be taken to minimize cell stress as it leads to nonspecifically expressed reporter activity. Unspecific activation was provoked by detachment and movement of adherent cells from culture plate/flask to the microchip, changes in composition of the chemical environment (addition of fresh cell medium), and change from batch to flow environment. As demonstrated, unspecific cell activation can be overcome by adapting the cells to the new conditions for a certain period of time. Another crucial matter addressed is the issue of microbial contamination during long-term handling and assaying of mammalian cells under microfluidic conditions. A correct design and the performance of critical steps under sterile conditions were shown to circumvent this type of problem.

4720

Analytical Chemistry, Vol. 76, No. 16, August 15, 2004

Fulfilling the above-mentioned requirements, we were able to follow the ATP receptor-mediated specific reporter gene expression of LUC in real time. In comparison to corresponding microtiter plate assays,7,8 the microfluidic format allows continuous monitoring of the changes in reporter gene activity of the same cell population both before and after stimulation with ligand. The miniature format consumes very small amounts of ligands, which usually are expensive and available in limited amounts. The latter can easily be added or removed in the microfluidic system via the carrier flow, which also supplies the cells continuously with a fresh cell medium. Consequently, consecutive additions of different ligand concentrations or even different ligands are possible. ACKNOWLEDGMENT The authors would like to thank Ulla-Britt Andersson at the Division of Molecular Biology for cell culturing and technical assistance. Financial support from the Swedish Science Council (Vetenskapsrådet) and the Swedish Foundation for Strategic Environmental Research (MISTRA) is kindly acknowledged. A travel scholarship for R. Davidsson from Bokelund Foundation and The Swedish Chemical Society are kindly acknowledged. The authors also wish to thank Sven Ha¨gg (Department of Analytical Chemistry, Lund University) for providing us with homemade software for data acquisition. Received for review October 23, 2003. Accepted June 1, 2004. AC035249O