A Three-Dimensional Flow Control Concept for Single-Cell

Anal. Chem. 2004, 76, 5273-5281

A Three-Dimensional Flow Control Concept for Single-Cell Experiments on a Microchip. 1. Cell Selection, Cell Retention, Cell Culture, Cell Balancing, and Cell Scanning Xing Yue (Larry) Peng†,‡ and Paul C. H. Li*,†

Department of Chemistry, Simon Fraser University, Burnaby, British Columbia, Canada V5A 1S6, and Department of Biology, Xiamen University, Xiamen, Fujian, China 361005

An ideal microchip for single-cell experiments should be able to allow us to culture cells, to select any desired single cell from a group, to retain the cell for convenient cellular signal detection, and to deliver any buffer or reagent directly to the cell at any time during continual detection and observation. Most importantly, any negative impact on the live cell should be minimized. To accomplish all these functions, we developed a threedimensional liquid flow control concept and employed special liquid flow fields to manipulate and retain a single yeast cell freely in the chip. A zero-speed point was controlled to retain the cell for three-dimensional cell balancing and cell scanning. A dispersive flow delivered reagents at a high speed to very near the cell and provided them to the cell at a low speed. No force stronger than its gravitational force was exerted on the cell, which could be balanced on different positions on an arc-sloping wall, thus minimizing any negative impact on the cell due to strong liquid flows. Specifically, we demonstrate on-chip single-cell culture, cell wall removal, and reagent delivery. Subsequently, single-cell fluorescence detection was performed, and noise filtering and background correction were applied for data processing. In recent years, microfluidic lab-on-a-chip has widely been applied for biochemical analysis.1-3 In particular, various microchip techniques for cellular biochemical analysis have been developed recently.4-19,42 For on-chip experiments, transport and selection * Corresponding author. Tel: 6042915956. Fax: 6042913765. E-mail: [email protected]. † Simon Fraser University. ‡ Xiamen University. (1) Auroux, P. A.; Lossifidis, D.; Reyes, D. R.; Manz, A. Anal. Chem. 2002, 74, 2637-2652. (2) Landers, J. P. Anal. Chem. 2003, 75, 2919-2927. (3) Northrup, M. A., Jensen, K. F., Harrison, D. J. Eds. Proc. 7th Int. Symp. Micro Total Anal. Syst. Oct 2003, Squaw Valley, CA, Transducer Research Foundation, 2003. (4) 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. (5) Fu, A. Y.; Spence, C.; Scherer, A.; Arnold, F. H.; Quake, S. R. Nat. Biotechnol. 1999, 17, 1109-1111. (6) Li, P. C. H.; Harrison, D. J. Anal. Chem. 1997, 69, 1564-1568. 10.1021/ac049384s CCC: $27.50 Published on Web 08/14/2004

© 2004 American Chemical Society

of cells have been mainly achieved by the liquid flow.4-7,9,11,20-22,42 The main technical issues for successful cell biochemical studies include how to retain the cell and how to maintain cell integrity during reagent delivery. To date, the major methods for cell immobilization include (1) cell adhesion,8,23,24,42 (2) physical retention by slit-type filters,25-28 weir-type filters,4,9,11,29,30 or even (7) Voldman, J.; Gray, M. L.; Toner, M.; Schmidt, M. A. Anal. Chem. 2002, 74, 3984-3990. (8) Takayama, S.; Mcdonald, J. C.; Ostuni, E.; Liang, M. N.; Kenis, P. J. A.; Smagilov, R. F.; Whiteside, G. M. Proc. Natl. Acad. Sci. U.S.A. 1999, 96, 5545-5548. (9) Li, P. C. H.; de Camprieu, L.; Cai. J.; Sangar, M. Lab Chip 2004, 4, 174180. (10) Schilling, E. A.; Kamholz, A. E.; Yager, P. Anal. Chem. 2002, 74, 17981804 (11) Yang, M.; Li, C. W.; Yang, J. Anal. Chem. 2002, 74, 3991-4001 (12) Fu, A. Y.; Spence, C.; Scherer, A.; Arnold, F. H.; Quake, S. R. Nat. Biotechnol. 1999, 17, 1109-1111. (13) Lin, Y. C.; Jen, C. M.; Huang, M. Y.; Wu, C. Y.; Lin, X. Z. Sens. Actuators, B 2001, 79, 137-143. (14) Roper, M. G.; Shackman, J. G.; Dahlgren G. M.; Kennedy, R. T. Anal. Chem. 2003, 75, 4711-4717. (15) McClain, M. A.; Culbertson, C. T.; Jacobson, S. C.; Allbritton, N. L.; Sims, C. E.; Ramsey, R. M. Anal. Chem. 2003, 75, 5646-5655. (16) Sohn, L. L.; Saleh, O. A.; Facer, G. R.; Beavis, A. J.; Allan, R. S.; Notterman, D. A. Proc. Natl. Acad Sci. U.S.A. 2000, 97, 10687-10690. (17) Fu, A. Y.; Chou, H.; Spence, C.; Arnold, F. H.; Quake, S. R. Anal. Chem. 2002, 74, 2451-2457. (18) Salimi-Moosavi, H.; Szarka, R.; Andersson, P.; Smith, R.; Harrision, D. J. Proc. Micro Total Anal. Syst. ‘98, Banff, Canada, October 1998; pp, 69-72. (19) Tamaki, E.; Sato, K.; Tokeshi, M.; Sato, K.; Aihara, M.; Kitamori, T. Anal. Chem. 2002, 74, 1560-1564. (20) Dittrich P. S.; Schwille, P. Anal. Chem. 2003, 75, 5767-5774. (21) Carlson, R. H.; Gabel, C.; Chan, S.; Austin, R. H. Biomed. Microdevices 1998, 1, 39-47. (22) Gourley, R. L.; McDonald, A. E.; Hendricks, J. K.; Copeland, G. C.; Hunter, J.; Akhil, O.; Dunne, J. L.; Skirboll, S. L.; Nihlen, L.; Capps, H. Biomed. Microdevices 1999, 2, 111-122. (23) Parce, J. W.; Owicki, J. C.; Kercso, K. M.; Sigal, G. B.; Wada, H. G.; Muir, V. C.; Bousse, L. J.; Ross, K. L.; Sikic, B. I.; McConnell, H. M. Science 1989, 246, 243-247. (24) Tokano, H.; Sul, J.; Mazzanti, M. L.; Doyle, R. T.; Haydon, P. G.; Porter, M. D. Anal. Chem. 2002, 74, 4640-4646. (25) Wilding, P.; Pfahler, J.; Bau, H. H.; Zemel, J. N.; Kricka, L. J. Clin. Chem. 1994, 40, 43-47. (26) Sutton, N.; Tracey, M. C.; Johnston, I. D. Microvasc. Res. 1997, 53, 272281. (27) Stemme, G.; Kittilsland, G. Appl. Phys. Lett. 1998, 53, 1566-1568. (28) He, B.; Tan, L.; Regnier, F. Anal. Chem. 1999, 71, 1464-1468. (29) Brody, J. P.; Osborn, T. D.; Forster, F. K.; Yager, P. Sens. Actuators, A 1996, 54, 704-708.

Analytical Chemistry, Vol. 76, No. 18, September 15, 2004 5273

within polymeric materials,31,32 and (3) dielectrophoresis.33-35 Even if these cell retention strategies might not have any negative effect on a stationary cell, the liquid flow that is essential for transport of buffer and reagents to the cell might damage the cell. It is because the liquid flow always exerts a force on the cell. To balance this force, the cell needs an opposite force. Adhesion or blocking of the cell usually generates a force locally on a little part of the cell’s surface rather than uniformly on the whole cell surface. Therefore, a strong flow might damage the cell. On the other hand, the flow should not be too weak to ensure a sufficient flow for reagent delivery. To solve this problem, we develop the concept of three-dimensional flow control. This flow control combines the cell balancing achieved in the depth dimension (1D) as well as the cell scanning in the channel dimension (2-D). For cell balancing, we make use of the downward residual gravitational force of the cell residing on an arc slope to balance the upward force exerted by the liquid flow of reagents. Interesting enough, the arc slope, which was present in all glass microchips fabricated by isotopic etching, was never exploited for any applications. For cell scanning, we exploit the zero-speed point (ZSP) created by the liquid flow field against a specially shaped microstructure. With the three-dimensional flow controls, we have successfully carried out cell balancing, cell scanning, and single-cell fluorescent measurement on a single yeast cell. The yeast cell was chosen because of its availability and short cell cycle (for cell culture). The fluorescent change in the yeast cell was chosen to be the formation of fluorescein due to hydrolysis of fluorescein diacetate (FDA). Furthermore, with the techniques of cell balancing and cell scanning, culture of a single yeast cell has been accomplished on-chip. Unlike our method, current on-chip culture methods are carried out only in a batch mode, without keeping tracking of a single cell, and only for adherent cells.36-41 Theory (the Three-Dimensional Flow Control Concept). Figure 1 showed the design of the chip’s main cell retention structure. It consists of a V-shaped barrier with a central stretch. The wall of the barrier, which is opposite to the reagent channel c, is in the form of an arc slope shape (Figure 1A). This arc slope is critical for cell balancing and retention (vide infra). (30) Wilding, P.; Kricka, L. J.; Cheng, J.; Hvichia, G. Anal. Biochem. 1998, 257, 95-100. (31) Heo, J.; Thomas, J.; Seong, G. H. Crooks, R. M. Anal. Chem. 2003, 75, 22-26. (32) Koh, W. G.; Itle, L. J.; Pishko, M. V. Anal. Chem. 2003, 75, 57835789. (33) Voldman, J.; Gray, M. L.; Toner, M.; Schmidt, M. A. Anal. Chem. 2002, 74, 3984-3990. (34) Fiedlerr, S.; Shirley, S. G.; Schnelle, T.; Fuhr, G. Anal. Chem. 1998, 70, 1909-1915. (35) De Gasperis, G.; Yang, J.; Becker, F. F.; Gascoyne, P. R. C.; Wang, X. B. Biomed. Microdevices 1999, 2, 41-49. (36) Chang, W. J.; Akin, D.; Sedlak, M.; Ladisch, M. R.; Bashir, R. Biomed. Microdevices 2003, 5, 281-290. (37) Leclerc, E.; Sakai, Y.; Fujii, T. Biomed. Microdevices 2003, 5, 109-114. (38) Gray, B. L.; Lieu.; D. K.; Collins, S.D.; Smith, R. L.; Barakat. A. I. Biomed. Microdevices 2002, 4, 9-16. (39) Walker, G.M.; Ozers, M.S.; Beebe, D. J. Biomed. Microdevices 2002, 4, 161-166. (40) Welle, A.; Gottwald, E. Biomed. Microdevices 2002, 4, 33-41. (41) Pesai, T. A.; Deutsch, J.; Motlagh, D; Tan, W.; Russell, B. Biomed. Microdevices 1999, 2, 123-129. (42) Gao, J.; Yin, X. F.; Fang, Z. L. Lab Chip 2004, 4, 47-52.

5274 Analytical Chemistry, Vol. 76, No. 18, September 15, 2004

Figure 1. Microchip design and cell selection. (A) The cell retention microstructure: it contained a wide channel for cell introduction (from vial a or vial b) and a narrow channel (40 µm wide) for delivery of buffer or reagent solutions (from vial c). Channel depth was 15 µm. The microstructure, which is opposite to the reagent channel c (from vial c), consists of a V-shaped barrier with a central stretch. The fluorescent signal was detected within the detection window (see the white rectangle shown in the inset) by a photomultiplier tube (PMT). The single yeast cell with a bud was actually lying freely on the arc slope of 15-µm radius (see inset) balanced by the liquid flow. (B) Cell introduction: The liquid flow from the left carried a group of cells to the microstructure. (C) Cell selection: The liquid flow from c separated the cells and sent the desired cell downward to the detection window. Liquid flow could be driven by either fluid potential ( 45°, we could prevent this situation from occurring by noting the position of the cell on the slope, and therefore, cell damage by strong liquid flow could be avoided. On the other hand, if the wall was vertical (Figure 3C,D), the cell could not adjust its position and a strong flow could cause a very high reaction force from the vertical wall (PH) and the force balance had no relation to the residual gravity. Analytical Chemistry, Vol. 76, No. 18, September 15, 2004

5275

Figure 3. Analysis of the force balance of the cell. (A) The liquid flow of different directions and strengths near an arc slope wall was shown. (B) The forces exerted on the balanced cell lying on the arc slope: g, The cell’s gravity (buoyancy subtracted); fR, force exerted by the flow at an angle (R); fH, force exerted by the horizontal flow (i.e., R ) 0); PR, reaction force of the arc slope wall to the cell for a flow directed at an angle (R); PH, reaction force of the arc slope wall to the cell for a horizontal flow. (C) The liquid flow near a vertical wall. (D) The forces exerted on the balanced cell against the vertical wall. Pv, reaction force of the bottom to the gravity; PH, reaction force of the vertical wall to the cell for a horizontal flow. (E) The force relationship between g, fR, and PR as given in (B). (F) The force relationship between g, fH, and PH when a cell was balanced on the arc slope wall with an increased angle of the slope (β).

It is worthwhile to mention that either the arc slope wall or vertical wall will give rise to a ZSP due to the split two-dimensional flow. However, only the arc slope wall has the ability to let the cell adjust its position to prevent itself from damage by a strong flow. The arc slope wall actually serves as a buffer zone. When a cell recedes to near the ZSP, the cell can escape from the strong flow. So the arc slope is very effective for protecting the cell. EXPERIMENTAL SECTION The glass microchip was fabricated through the ProtoChip Program of Canadian Microelectronic Corp. Borofloat glass wafers were used to fabricate the channel and cover plates (16 mm × 95 mm). Then, the two pieces of glass plate were thermally bonded together to produce the finished chip. The layout of the cell retention microstructure has been depicted in Figure 1A. The chip was determined to contain 15-µm-deep channels. The radius of the arc slope was 15 µm, which should be greater than the diameter of the cell. The central stretch is normally flat for uniformity in scanning the cell to obtain the cell signal. The same chip was easily washed and reused and has survived many hours (∼200 h) of experiments. In all experiments of yeast cells, no surface coating on the glass chip was necessary. For optical measurements, the microchip was placed on the translation stage of an inverted microscope (Nikon TE 300) with a dual-image module (Nikon), which was coupled to both a CCD 5276 Analytical Chemistry, Vol. 76, No. 18, September 15, 2004

Figure 4. Schematic diagram of the optical measurement setup that included an inverted microscope and the associated optics: a, dichroic filter 1 (495 nm); b, dichroic filter 2 (540 nm); c, band-pass filter (470 nm/40 nm); d, long-pass filter (645 nm); e, band-pass filter (525 nm/50 nm); f, microscope objective (ELWD, 40×/0.60); g, mirror. The first optical path (red light, d, chip, f, a, g, b, CCD camera) was used for bright-field optical observation. The second optical path (xenon arc lamp excitation light, c, a, f, chip, f, g, b, e, PMT) was used for fluorescent measurement. The chip as shown in the image has been used in all single-cell experiments within 9 months. The width of the chip is 16 mm.

video camera (JVC TKC 1380) and a photomultiplier tube (PMT) (Photon Technology Int., PTI) (Figure 4). Simultaneous optical observation and fluorescent measurement of the single cell was achieved using this special optical measurement setup. Specifically, the red light (>645 nm) was used to observe the cells using the video camera. The motions of any cells were continually displayed on a television monitor and recorded by a videotape recorder (JVC HR-S7500U). A xenon arc lamp was employed to excite the fluorophore. Green fluorescent signals due to intracellular fluorescein formed (520 nm) were not able to reach the camera and could only be detected by the PMT. Fluorescence signals from the PMT were recorded by a computer using the Felix software (PTI). The PMT only recorded the fluorescent signal within the detection window (Figure 1A). If the yeast cell (mother and daughter) was within the window, the signal represented the cellular fluorescence plus the fluorescent background. If not, only the fluorescent background was detected. The three-dimensional liquid flows could be driven by electric potentials. To create a downward flow of reagents, a high voltage (50-500V) was applied to c, and both a and b were at ground (see Figure 1A for the locations of a, b, and c). To create a lateral flow to the right, a high voltage was applied to a with b at ground, and vice versa. When a high-conductivity electrolyte buffer was required, e.g., in cell culture experiments, voltage control could not be used, and only fluid potential (2.5 Hz). (C) The use of a narrower detection window on another yeast cell could distinguish the mother yeast cell from its daughter yeast cell by showing the peak and the shoulder, respectively.

When the cell was out of the window, the PMT measured the background fluorescence. When the cell entered the window, the PMT measured the signal together with the background fluorescence. It is shown that the fluorescent intensity of a yeast cell as given by the peak height began to rise due to increased FDA metabolism. Because of the noise, fluorescent intensity was clearly seen only after 75 s (Figure 8A). Since the data collection rate was 50 Hz and we normally controlled the peak width from 1 s to over 10 s (i.e., 0.1-1 Hz), we performed filtering of noise in the frequency range of 2.5-50 Hz. After filtering the noise in the data represented in Figure 8A, the results are shown in Figure 8B. After noise filtering, even the weak cellular signal became very clear, especially during the time of 25-75 s (Figure 8B). If the detection window was larger than the cell, the peak height represented the total fluorescence of the whole cell regardless of the scanning rate. If we wanted to know the fluorescent distribution of the cell, we could narrow down the detection window. This strategy could differentiate between the larger mother cell and its smaller budding daughter cell, as shown in the fluorescent data after noise filtering (Figure 8C). The high peak came from the mother cell and the shoulder peak was caused by the daughter cell. Scanning the cell back and forth generated pairs of mirror peaks. Background Correction. In single-cell experiments using different reagents to stimulate the cell, the background fluorescence might not be a constant due to the different fluorescent backgrounds of the reagents or buffers. In our experiments, there were always gradual increases in the fluorescent intensity of G7 and H4 buffers. This increase was caused by the slow hydrolysis of FDA in aqueous solutions. Using the cell scanning technique, the fluorescent background was recorded as a baseline and the cell fluorescent signal as peaks. 5280 Analytical Chemistry, Vol. 76, No. 18, September 15, 2004

Figure 9. Background correction applied to an experiment with a yeast cell (cell 5). (A) Peaks due to cell fluorescence plus background. (B) Background baseline extracted from (A). (C) Cell fluorescence peaks after background subtraction, (D) Peak envelope of all fluorescence peaks. Two reagent scales showed the medium types and the FDA concentrations. In addition, one excitation light scale shows when the excitation light was shut off or turned on. Further discussion on this experiment is given in the companion paper.

In a complex experiment using various reagents at different time points, the fluorescent data appeared to be very strange and were hard to interpret (Figure 9A). However, the baseline was easy to be separated (Figure 9B). After background correction was performed using these baseline data, the peak-only signals were obtained (Figure 9C). This background correction method enabled us to grasp the real dynamic information from the cell, thus assisting data interpretation. Furthermore, the baseline provided us additional information. We could know whether the switching of buffers between G7 and H4 did occur successfully by examining the baseline (Figure 9B, 1, 2, 12, and 13 ks). Moreover, we shut off the excitation light three times (Figure 9A-C, 10-12 ks) to determine whether the photobleaching had any significant effect on cell fluorescence. The fast-decaying baseline showed that the fluorescent background was indeed affected by the photobleaching effect (Figure 9B, 1012 ks). Nevertheless, after the background correction, the cell showed no apparent decrease in signal (Figure 9C, 10-12 ks). Finally, the peak envelope (Figure 9D) was generated, which was significant for curve-fitting to our proposed model of FDA metabolism; see the companion paper. CONCLUSION The concept of three-dimensional flow control with a ZSP at the top of the arc slope wall was successfully applied for carrying out single-cell experiments. All microfluidic operations such as cell selection, cell scanning, cell retention, and reagent delivery could be achieved in the chip. Moreover, yeast cell culture and cell wall removal were achieved on-chip. Complex experiments involving changing of different reagents and buffers could easily be performed on the same single cell retained in the chip. Fluorescent signals from the cell could be easily extracted after

noise filtering and background correction. The three-dimensional flow control, which has been applied to perform successful experiments on yeast cells, could also be applied for mammalian cells experiments. ACKNOWLEDGMENT We thank Natural Sciences and Engineering Research Council of Canada, Canada Foundation for Innovation, and British Columbia Knowledge Development Fund for financial support. We are grateful for the ProtoChip Program of Canadian Microelectron-

ics Corporation for microchip fabrication. X.Y.P. thanks the China Scholarship Council for financial support. SUPPORTING INFORMATION AVAILABLE Eight QuickTime movies. This material is available free of charge via the Internet at http://pubs.acs.org. Received for review April 25, 2004. Accepted July 2, 2004. AC049384S

Analytical Chemistry, Vol. 76, No. 18, September 15, 2004

5281