Microfabricated System for Parallel Single-Cell Capillary Electrophoresis

Jul 28, 2004 - trance was manufactured by F2-laser ablation to form a. 20- to 10-µm tapered lysis reservoir, creating an injector geometry effective ...
0 downloads 0 Views 258KB Size
Download PDF
Anal. Chem. 2004, 76, 4983-4989

Microfabricated System for Parallel Single-Cell Capillary Electrophoresis Nigel R. Munce,† Jianzhao Li,‡ Peter R. Herman,‡ and Lothar Lilge*,†,§

Department of Medical Biophysics, University of Toronto, 610 University Avenue, Toronto, Ontario, Canada M5G 2M9

Performing single-cell electrophoresis separations using multiple parallel microchannels offers the possibility of both increasing throughput and eliminating cross-contamination between different separations. The instrumentation for such a system requires spatial and temporal control of both single-cell selection and lysis. To address these problems, a compact platform is presented for single-cell capillary electrophoresis in parallel microchannels that combines optical tweezers for cell selection and electromechanical lysis. Calcein-labeled acute myloid leukemia (AML) cells were selected from an on-chip reservoir and transported by optical tweezers to one of four parallel microfluidic channels. Each channel entrance was manufactured by F2-laser ablation to form a 20- to 10-µm tapered lysis reservoir, creating an injector geometry effective in confining the cellular contents during mechanical shearing of the cell at the 10-µm capillary entrance. The contents of individual cells were simultaneously injected into parallel channels resulting in electrophoretic separation as recorded by laser-induced fluorescence of the labeled cellular contents. The majority of analytical tools and protocols available for molecular cell biology require pooling of a population of cells to obtain a sufficient quantity of analytes. As a result, information regarding analyte variability at the single-cell level and the potential existence of subpopulations within the collective of cells is lost. The ability to examine such variation would allow the study of heterogeneous tissue such as the evolving pathology of cancer or the development of organisms in greater detail. However, this task requires that both a sufficient number of molecular species are resolved from each cell in order to identify variation and a large enough sample of the population is investigated so that potentially rare subpopulations can be detected. Cell by cell protein and morphological analysis using flow cytometry to assess variation1 only partially addresses these needs by sampling thousands of individual cells each labeled with a limited number * Corresponding author. E-mail: [email protected]. † Department of Medical Biophysics, University of Toronto. ‡ Department of Electrical and Computer Engineering, University of Toronto, 10 King’s College Road, Toronto, ON, Canada, M5S 3G4. § Ontario Cancer Institute, University Health Network, 610 University Avenue, Toronto, ON, Canada, M5G 2M9. (1) Chapman, J. W.; Wolman, E.; Wolman, S. R.; Remvikos, Y.; Shackney, S.; Axelrod, D. E.; Baisch, H.; Christensen, I. J.; White, R.A.; Liebovitch, L. S.; Moore, D. H.; Waldman, F. M.; Cornelisse, C. J.; Shankey, T. V. Cytometry 1998, 31, 67-73. 10.1021/ac0496906 CCC: $27.50 Published on Web 07/28/2004

© 2004 American Chemical Society

of different fluorescently labeled analytes so the dyes can be spectrally resolved. Global genetic analysis of single cells using reverse transcription polymerase chain reaction protocols, combined with laser microdissection, and microarray techniques have also been used to assess heterogeneity at the single-cell level.2 However, this technique suffers from a low throughput rate, and as a result analysis of a statistically significant number of individual cells to obtain a representative sample of the population distribution remains impractical. Capillary electrophoresis of single cells has been examined as a means to separate a large number of analytes from single cells.3,4 Capillary electrophoresis of single cells combined with fluorescence labeling offers the advantage of high sensitivity (attomole-zeptomole5). However, long separation times combined with sequential analysis of each cell preclude applying this technique to a large numbers of cells. To address this shortcoming, electrophoresis systems using multiple capillaries6 or continuous injection and separation of single-cell contents7 have been proposed. Instrumentation for multicapillary single-cell electrophoresis will likely require advanced cell selection schemes such as micromachined vias8 or microfluidic arrays9 for multicell alignment. Continuous sequential analysis of single cells, although enabling increased throughput, is typically limited to resolving only a few analytes to avoid overlap of separation spectra. Additionally, it requires effective cleaning techniques to prevent accumulation of debris in the channel, which may affect the accuracy of subsequent analysis. Thus, high-throughput global analysis at the single-cell level remains difficult. Microfluidic, or “lab-on-a-chip”, devices have attracted interest as a tool for single-cell analysis.10-13 Microfluidic devices have been (2) Kamme, F.; Salunga, R.; Yu, J.; Tran, D.; Zhu, J.; Luo, L.; Bittner, A.; Guo, H.; Miller, N.; Wan, J.; Erlander, M. J. Neurosci. 2003, 23, 3607-3615. (3) Gilman, S. D.; Ewing, A. G. Anal. Chem. 1995, 67, 58-64. (4) Krylov, S. N.; Zhang, Z.; Chan, N. W.; Arriaga, E.; Palcic, M. M.; Dovichi, N. J. Cytometry 1999, 37, 14-20. (5) MacTaylor, C. E.; Ewing, A. G. J. Microcolumn Sep. 2000, 12, 279-284. (6) Zhang, J. Z.; Voss, K. O.; Shaw, D. F.; Roos, K. P.; Lewis, D. F.; Yan, J.; Jiang, R.; Ren, H.; Hou, J. Y.; Fang, Y.; Puyang, X.; Ahmadzadeh, H.; Dovichi, N. J. Nucleic Acids Res. 1999, 27, E36. (7) Chen, S.; Lillard, S. J. Anal. Chem. 2001, 73, 111-118. (8) Herman, P. R.; Yick, A.; Li, J.; Munce, N.; Lilge, L. Jervis, J.; Krylov, S. Conf. on Lasers and Electrooptics Technol. Digest 2003, Paper CFL5. (9) Feng, X.; Tokronova, N.; Xu, B.; Chen, P.; Gillis, K. D.; Castracane, J. Proc. SPIE 2003, 4982, 37-44. (10) Wheeler, A. R.; Morishima, K.; Arnold, D. W.; Rossi, A. B.; Zare, R. N. Proc. µ-TAS 2000, 25-28. (11) Wheeler, A. R.; Throndset, W. R.; Whelan, R. J.; Leach, A. M.; Zare, R. N.; Liao, Y. H.; Farell, K.; Manger, I. D.; Daridon, A. Anal. Chem. 2003, 75, 3581-3586. (12) McClain, M. A.; Culbertson, C. T.; Jacobson, S. C.; Allbritton, N. L.; Sims, C. E.; Ramsey, J. M. Anal. Chem. 2003, 75, 5646-5655.

Analytical Chemistry, Vol. 76, No. 17, September 1, 2004 4983

employed for isolating multiple individual cells11,14 as well as performing high-throughput sequential separations of single cells.12 Microchip designs incorporating multiple independent separation channels and reaction chambers in a small footprint have been presented.15,16 However, combining multiple separation channels for single-cell electrophoretic separations remains challenging due to the need of both selecting and lysing individual cells in preassigned microchannels. To address the problem of isolating single cells, microchip systems may employ passive microfluidic selection,11,14 dielectrophoresis in combination with optical tweezers,17 or optical tweezers for use in a microculture device.18 Passive selection techniques are able to isolate a large number of single cells, but the user loses the ability to identify and manipulate a predetermined cell. Conversely, optical tweezers under microscopic guidance allow a specific cell to be identified, selected, and translated to a desired location, which may be critical in analyzing differences in cells or cell systems, such as the differentiation of stem cells. Optical tweezers have been studied extensively,19,20 and it has been shown that the optical powers required for translating single cells ( 60 °C) caused the seal to conform to protruding objects, such as remaining dust particles, without occluding the microchannels. This sealing method offers the possibility of reusing the device as the seal could be pealed off and the device cleaned as required. Microchannels were examined by both scanning electron microscopy and confocal laser-scanning microscopy. A scanning electron micrograph of the tapered channel is presented in Figure 1b. Channel profiling by confocal microscopy was performed at 5 points per channel across three different devices by filling the anode reservoir with a 30 µM fluorescein (Sigma, St.Louis, MO) solution in dH2O. A total of 120 confocal images were acquired over an axial depth of 30 µm to achieve an oversampled optical slice thickness of 1 mm). Nevertheless, transport of any given cell to a predetermined injection channel was typically accomplished in less than 5 min, with loading of four channels requiring approximately 15-20 min. To observe the dynamics of the nucleus and calcein-labeled cellular contents as well as the change in cell morphology during injection, simultaneous dual channel fluorescence imaging at 385470 nm and 505-550 nm for Hoechst and calcein stains, respectively, and differential interference contrast (DIC) transmission images were collected with the 63× objective under laserscanning imaging mode. A series of 150 images (at 3 Hz repetition) was initiated 2 s prior to the application of the electric field. To demonstrate parallel injection and analysis of calcein-labeled cells, a lower magnification (20×, 0.75 NA Fluar, Zeiss) objective to capture a larger field of view (450 µm × 450 µm) and a fluorescence camera (CoolSnap Pro, Roper Scientific, Duluth, GA) for more uniform image acquisition were used (Figure 4). Exposure time on the camera was set at 100 ms and data collected 4986

Analytical Chemistry, Vol. 76, No. 17, September 1, 2004

RESULTS AND DISCUSSION Single-Cell Injection and Lysis. Selection and transport of nonstained cells by tweezers and up to 30 min inside the injector did not lead to cell morphological changes or uptake of membrane integrity markers, such as propidium iodide, indicating that neither the physical movement by tweezers nor the diffusion of oxygen to the injectors compromised survival. Oxygen diffusion is facilitated not only via the capillary core as during capillary electrophoresis but also via the 35-µm PET film. Hence, the oxygen diffusion distance is always less than 50 µm. A principal aim of this study was to demonstrate spatial and temporal control of cell lysis and injection to allow the use of multiple separation channels. Upon initiation of the electric field, cells not in contact with the opening of the channel would be drawn to the anode but would remain intact. Cells at the interface of the two diameter channels, however, would be lysed as the electric field would pull the cell into an opening smaller than its diameter putting stress on the membrane and drawing charged molecules out. All cells from multicellular organisms possess negatively charged surfaces at physiological pH28 and a negative internal charge due to nucleic acid and proteins. These properties have been used previously for cell electrophoresis in capillaries29 as well as in microchannels.30 This spatial selectivity is shown in Video 2 in the Supporting Information. The time delay between the application of the electric field and cell lysis was less than 300 ms. Cell lysis and injection of cellular contents into the microchannel could be observed by the decrease in fluorescence intensity of the cell and the corresponding increase in the fluorescence signal from the ROI inside the microchannel as plotted in Figure S3 of the Supporting Information as a function of time. The profile shows a sharp peak with a fwhm of 800 ( 150 ms. Accurate evaluation of this temporal dispersion was not (28) Mehrishi, J. N.; Bauer, J. Electrophoresis 2002, 23, 1984-1994. (29) Kitagawa, S.; Nozaki, O.; Tsuda, T. Electrophoresis 1999, 20, 2560-2565. (30) Ichiki, T.; Ujiie, T.; Shinbashi, S.; Okuda, T.; Horiike, Y. Electrophoresis 2002, 23, 2029-203.

possible due to the slow frame rate of the fluorescence imaging compared to the injection process. The injection plug is observed to have a normal spatial distribution with a length of 30 µm at fwhm (see Figure S4 of the Supporting Information). This dispersion of a 10-µm cell to a 30-µm plug upon injection is most likely caused by the variation in the field at the injector, resulting in molecules in the cell closer to the sides of the channel experiencing less force. An injection plug of 30 µm at fwhm is comparable to other recently published work using either capillary electrophoresis (Krylov et al.4) or on-chip electrophoresis (McClain et al.12 and Fu et al.31). Additionally, the molecular size distribution of the injected content due to calcein binding to molecules in the cell contributes further to the dispersion of the injected contents. Calcein binding to subcellular structures in a cell, postinjection, can be seen in the fluorescence microscope image in Figure S5 of the Supporting Information. By comparing the sum of the fluorescence intensity of the pixels from the cell prior to injection to the integrated fluorescence from the ROI one can estimate an injection efficiency of 30%. Analyte loss is attributed to the dye binding to subcellular organelles and the fact that positively charged contents of the cell were propelled to the anode. Additionally, the slow frame rate may have resulted in unbound dye molecules remaining undetected. Morphological changes of the cell, taking place in less than 300 ms upon injection, were observed in DIC imaging. These include the appearance of large vacuole-like structures in the cell and increased cellular membrane roughness and cell diameter, all indicative of membrane integrity loss. Observing the Hoechst 33342-based signal in Figure 3a-c, indicates nuclear swelling. The fluorescence intensity appears to decrease and to localize on the nuclear membrane upon injection. These observations suggest that, while the membrane bound organelle is not drawn into the channel, minor DNA fragments may be injected. Retaining the large organelles within the cell membrane is advantageous as they may clog channels with diameters of 10 µm or less. It also opens the possibility of harvesting the nucleus for subsequent analysis. Microchip Single-Cell Electrophoretic Separations. From the separations performed (n ) 20 spectra), four peaks, consisting of two small sharp peaks followed by two broad peaks, were consistently observed (sample spectra are shown in Figure 5). In some instances, these peaks appeared to be composed of at least one subpeak, indicating that we had not completely resolved all of the molecular species. The average fwhm of the peaks were found to be 1.2 ( 0.2, 1.4 ( 0.3, 5.8 (1.0, and 8.7 ( 4.5 s at times of 16.8 ( 2.4, 20.0 ( 2.7, 31.8 ( 5.1, 45.2 ( 8.8 s for peaks 1-4, respectively. Increased resolution could be accomplished by using a longer separation distances than 10 mm. The appearance of multiple peaks is likely attributed to the hydrolysis of the calcein dye upon uptake by the cell. This has been reported with similar dyes (Oregon green) upon electrophoretic separation.12 Additionally, the dye could be seen to bind to subcellular structures after cell lysis, leading to the assumption that the injected fluorescent analytes were heterogeneous. Comparison of the separations obtained in this microchip to those performed in a conventional capillary electrophoresis system was (31) Fu, L.-M.; Yang; R.-J.; Lee, G.-B.; Liu, H.-H. Anal. Chem. 2002, 74, 50845091.

Figure 3. Cell lysis of a calcein (green) and Hoechst (blue) labeled AML cell at the opening of the tapered channel (a) immediately prior to the application of the electric field, (b) 290 ms, and (c) 580 ms after the application of the electric field.

complicated by the lack of an available analogous nondenaturating lysis method in conventional capillary electrophoresis. Throughput of the device presented here was approximately 24 cells/h. This rate could be greatly increased with the addition of more channels per device. On the basis of the largest field of view of the detection optics used (20×, 0.75 NA) a maximum of eight parallel separations may be run simultaneously. Thus, work will focus on the development of suitable parallel detection schemes such as high NA fiber-optic arrays for multiple singlepoint detection. CONCLUSIONS This study demonstrates the novel combination of tools for selecting, transporting, and lysing cells in a microchip platform. The system uses optical tweezers to select predetermined single cells and transports them to an injection region with the mechanical shearing of the cell against the opening of the 10-µm channel used for lysis, allowing a selected cell to be lysed in a specified microchannel and its fluorescently labeled cytoplasmic contents to be separated. To our knowledge, this work is the first to demonstrate parallel injection and separation of single cells in a multichannel microchip platform. The use of PET film for sealing of microchannels avoids the limitations due to restricted oxygen diffusion on cell viability. Performing separations in parallel allows more complex analyte mixtures to be separated, thus significantly increasing throughput compared to conventional single-cell capillary electrophoresis systems. The use of individual separation Analytical Chemistry, Vol. 76, No. 17, September 1, 2004

4987

Figure 4. Simultaneous injection of two calcein-labeled cells in parallel tapered channels: (a) two cells at the opening of adjacent microchannels, (b) approximately 300 ms after the activation of the electric field the cells, (c) at 600 ms parallel injection plugs are seen, and in (d) residual fluorescence in the cells is seen at 900 ms.

Figure 5. Three example electrophoretic spectra of calcein-labeled single cells.

channels for each cell separation eliminates possible crosscontamination from different single-cell separations. A potential drawback of using optical tweezers as a selection method in the system is the slow selection time of each cell compared to systems employing passive selection. However, with optical tweezers the user gains the ability to examine, select, and analyze predetermined individual cells. 4988

Analytical Chemistry, Vol. 76, No. 17, September 1, 2004

ACKNOWLEDGMENT Acute myloid leukemia cells were provided courtesy of Dr. Mark Minden, at the Ontario Cancer Institute. The authors also acknowledge the Advanced Optical Microscopy Facility at the Ontario Cancer Institute for assistance with this project. Funding for this work was provided by Photonics Research Ontario, the

Canadian Institute for Photonics Innovation, and the National Science and Engineering Council of Canada. SUPPORTING INFORMATION AVAILABLE Tiled microscope image of tapered microchannels; confocal “z-stack” profiles of 10- and 20-µm channels; graph of temporal width of injection plug at the opening of the channel; spatial intensity profile of the injection plug; image of a cell postinjection, illustrating calcein binding to subcellular structures; movie of

single-cell selection into a microchannel with optical tweezers. This material is available free of charge via the Internet at http://pubs.acs.org.

Received for review February 25, 2004. Accepted May 27, 2004. AC0496906

Analytical Chemistry, Vol. 76, No. 17, September 1, 2004

4989