Direct Manipulation and Observation of the Rotational Motion of Single

Anal. Chem. 2004, 76, 2492-2497

Direct Manipulation and Observation of the Rotational Motion of Single Optically Trapped Microparticles and Biological Cells in Microvortices J. Patrick Shelby, Sarah A. Mutch, and Daniel T. Chiu*

Department of Chemistry, University of Washington, Seattle, Washington 98195-1700

This paper describes a method for manipulating and monitoring the rotational motion of single, optically trapped microparticles and living cells in a microvortex. To induce rotation, we placed the microparticle at the center of rotation of the vortex and used the recirculating fluid flow to drive rotation. We have monitored the rotation of single beads (which ranged in diameter from a few micrometers to tens of micrometers) and living cells in a microvortex. To follow the rotation of a smooth and symmetrically shaped bead, we first ablated a small region (∼1 µm) on the bead. An Ar+ laser was then tightly focused (∼0.5µm spot size) onto the bead, and rotation was tracked by recording changes in the level of backscattered laser light as the ablated region repeatedly transited the laser focus. Using this method, we have followed bead rotation that varied in frequency from 0.15 to 100 Hz and have studied the effect of bead diameter on the rate of rotation at a given fluid flow rate. To monitor the rotation of single living cells, we selectively stained portions of B-lymphocytes with the fluorescent dye DiOC6. We observed rotation by following changes in the fluorescence signal as the dye-stained region transited the laser focal volume. This technique provides a simple and sensitive method for controlling and monitoring the rotational motion of microparticles in a microfluidic environment. The ability to exercise mechanical control at the micro- and nanometer-length scale is technologically intriguing and has widespread practical applications, as exemplified by the family of scanning probe techniques in the nanoscale1,2 and by the host of particle trapping methods in the microscale.3-7 Although control over the translational motions of micro- and nanoparticles can be carried out routinely using a number of different approaches, including ultrasonic,3 dielectric,4 and optical trapping,5-7 methods * To whom correspondence should be addressed. E-mail: chiu@ chem.washington.edu. (1) Gimzewski, J. K.; Joachim, C. Science 1999, 283, 1683-1688. (2) Quate, C. F. Jpn. J. Appl. Phys. 2003, 42, 4777-4779. (3) Yamakoshi, Y.; Noguchi, Y. Ultrasonics 1998, 36, 873-878. (4) Fuhr, G.; Reichle, C.; Mu ¨ ller, T.; Kahlke, K.; Schu ¨ tze, K.; Stuke, M. Appl. Phys. A 1999, 69, 611-616. (5) Ashkin, A.; Dziedzic, J. M.; Yamane, T. Nature 1987, 330, 769-771. (6) Ashkin, A.; Dziedzic, J. M. Science 1987, 235, 1517-1520. (7) Kuyper, C. L.; Chiu, D. T. Appl. Spectrosc. 2002, 56, 300A-312A.

2492 Analytical Chemistry, Vol. 76, No. 9, May 1, 2004

for rotating trapped particles have been more limited. Nevertheless, both electrorotation8-11 and optical rotation12-18 of a microparticle have been demonstrated. These methods, however, often require intricate microfabricated structures4,19 or complex optical setups.12 In electrically or optically driven rotation, furthermore, the fluid surrounding the rotating object gives rise to viscous retardation that dampens the rotation, which makes rotation at high rates difficult to attain. Here we describe in detail our approach that uses fluid flow to rotate an optically trapped particle and the optical methods we used to monitor in real time the rotation of such trapped microparticles and single cells. To rotate particles, we exploit recirculating flows formed intentionally in microfluidic systems, which we call microvortices,20 to drive the rotation of an optically trapped object, such as a polystyrene bead, a mammalian cell, or a bacterium. To understand and develop microvortices as a platform for rotating a single micro- and nanometer-sized object, it is necessary to follow the rotation of such particles in real time. Yet no suitable methods exist that are capable of monitoring the rotation of isotropic particles (e.g., beads, cells) over a dynamic range of a few hertz to hundreds of hertz. Techniques that employ light-scattering detection, such as forward scattering21 and backscattering,22,23 are well suited for (8) Judy, J. W. Smart Mater. Struct. 2001, 10, 1115-1134. (9) Wilson, C. F.; Wallace, M. I.; Morishima, K.; Simpson, G. J.; Zare, R. N. Anal. Chem. 2002, 74, 5099-5104. (10) Simpson, G. J.; Wilson, C. F.; Gericke, K.-H.; Zare, R. N. ChemPhysChem 2002, 3, 416-423. (11) De Gasperis, G.; Wang, X.; Yang, J.; Becker, F. F.; Gascoyne, P. R. C. Meas. Sci. Technol. 1998, 9, 518-529. (12) Paterson, L.; MacDonald, M. P.; Arlt, J.; Sibbett, W.; Bryant, P. E.; Dholakia, K. Science 2001, 292, 912-914. (13) Sacconi, L.; Rommano, G.; Ballerini, R. B.; Capitanio, M.; De Pas, M.; Giutini, M.; Dunlap, D.; Finzi, L.; Pavone, F. S. Opt. Lett. 2001, 26, 1359-1361. (14) Grier, D. G. Nature 2003, 424, 810-816. (15) Ukita, H.; Kanehira, M. IEEE J. Sel. Top. Quantum Electron. 2002, 8, 111117. (16) Nishioka, M.; Katsura, S.; Hirano, K.; Mizuno, A. IEEE Trans. 1nd. Appl. 1997, 33, 1381-1388. (17) Friese, M. E. J.; Nieminen, T. A.; Heckenberg, N. R.; Rubinsztein-Dunlop, H. Nature 1998, 394, 348-350. (18) Friese, M. E.; Rubinsztein-Dunlop, H.; Gold, J.; Hagberg, P.; Hanstorp, D. Appl. Phys. Lett. 2001, 78, 547-549. (19) Rowe, A. D.; Leake, M. C.; Morgan, H.; Berry, R. M. J. Mod. Opt. 2003, 50, 1539-1554. (20) Shelby, J. P.; Lim, D. S. W.; Kuo, J. S.; Chiu, D. T. Nature 2003, 425, 38. (21) Yamamoto, A.; Yamaguchi, I. Jpn. J. Appl. Phys. 1995, 34, 3104-3108. 10.1021/ac035531v CCC: $27.50

© 2004 American Chemical Society Published on Web 04/02/2004

studying rotation of anisotropically shaped particles. Smooth, symmetrically shaped particles, however, pose a problem because they lack surface anomalies that can scatter light. Most current detection schemes in these light-scattering techniques also lack sufficient sensitivity to follow the rapid rotation (100 Hz) of small particles that are a few micrometers in diameter. In some instances, the poor signal quality often necessitates raw data manipulation, such as Fourier transform.22 Flow profiling techniques such as particle imaging velocimetry24 and flow tagging velocimetry25,26 are capable of mapping fluid movement in microfluidic channels. The presence of the trapped particle or cell, however, may dramatically affect both the fluid movement and the rotation of the object itself. Flow measuring techniques alone thus are insufficient to characterize the behavior of the rotating microparticle. To address the lack of techniques for characterizing the rotation of single microparticles, we have developed two approaches. Both methods employ optical trapping7 to maintain the microparticle at the vortex core during rotation and rely on “marking” the otherwise isotropic object so its motion can be followed optically. This first technique uses a single nanosecond pulse from a UV laser to ablate a small (micrometer) area on the surface of a bead (or any other particles that can be ablated). A visible laser beam is focused onto the bead, and rotation is tracked by recording changes in the level of backscattered light as the “mark” transits the laser focus. The second approach is specific for biological cells, in which a portion of the cell is selectively stained with a fluorescent dye so the rotation can be followed by monitoring changes in the fluorescence signal as the dye-stained region comes in to and out of the laser probe volume. To follow fast rotation (tens to hundreds of hertz), we used confocal detection with single-molecule sensitivity. High sensitivity is critical because of the fast transit time of the “mark” through the small focus of the laser probe, which leads to short signal integration times. For example, a 1-µm (in diameter) “mark” on a 10-µm (in diameter) particle rotating at 100 Hz will transit the probe volume in ∼0.3 ms. Here we present our characterization of these methods for monitoring particle rotation and discuss considerations in using microvortices for controlling rotation of microparticles. EXPERIMENTAL SECTION Fabrication of Microchannels. The microfluidic systems shown in Figure 1 were fabricated in poly(dimethylsiloxane) (PDMS) using rapid prototyping.27 Briefly, a high-resolution mask was generated from a computer-aided drawing file imprinted with the channel design. The mask was used in contact photolithography with SU-8 photoresist (MicroChem, Newton, MA) to create a “master”, which consisted of bas-relief features of SU-8 on a silicon wafer. From the master, PDMS channels were molded and then sealed irreversibly to a borosilicate glass coverslip by (22) Friese, M. E. J.; Enger, J.; Rubinsztein-Dunlop, H.; Heckenberg, N. R. Phys. Rev. A 1996, 54, 1593-1596. (23) Sun, C.-K.; Huang, Y.-C.; Chen, P. C.; Liu, H.-C.; Lin, B.-L. J. Opt. Soc. Am. B 2001, 18, 1483-1489. (24) Adrian, R. J. Annu. Rev. Fluid Mech. 1991, 23, 261-304. (25) Lempert, W. R.; Harris, S. R. Meas. Sci. Technol. 2000, 11, 1251-1258. (26) Shelby, J. P.; Chiu, D. T. Anal. Chem. 2003, 75, 1387-1392. (27) McDonald, J. C.; Duffy, D. C.; Anderson, J. R.; Chiu, D. T.; Wu, H.; Schueller, O. A. J.; Whitesides, G. M. Electrophoresis 2000, 21, 27-40.

oxidizing the PDMS surface in oxygen plasma. Access holes to the channels were formed using a punch made from a 21-gauge needle. Polyethylene tubing (PE20) was inserted into the access holes, which were slightly smaller than the outer diameter of the tubing, to form a pressure seal between the tubing and the hole. The tubing was attached to a 3-mL syringe housed in a syringe pump (kd Scientific, New Hope, PA) for pressure-driven fluid flow. Optical Arrangement. Figure 2A shows the arrangement of the three lasers, the single-molecule confocal detection setup, and the imaging camera with respect to the sample (microfluidic channel) that was placed on an inverted microscope (Nikon TE300). The 337-nm line of a pulsed nitrogen laser (Laser Science Inc., Franklin, MA) was used to create a small ablation on the surface of polystyrene beads; the 1064-nm line of an Nd:YAG laser (Spectra-Physics, Mountain View, CA) was used for the optical trapping and positioning of the microparticle at the center of the microvortex; the 488-nm line of an Ar+ laser (Spectra-Physics) defined the probe region for single-molecule confocal detection. All experiments were carried out using a Nikon 100× Superflaur objective with a 1.3 numerical aperture. The confocal detection was carried out by directing the collected light (e.g., backscattered laser light or fluorescence signal) through a 75-µm pinhole placed at the primary image plane, then onto the avalanche photodiode detector (APD) after the appropriate spectral filtering and spatial focusing. The APD signal was recorded using a multichannel scalar card (PerkinElmer Instruments, Oak Ridge, TN). For widefield imaging, the light was sent to a high-resolution CCD camera (Cohu Electronics, San Diego, CA) on a separate port of the microscope. Selective Staining of Living Cells. Mouse B-lymphocyte cells were cultured in RPMI-1640 media (Cellgro by Mediatech Inc., Herndon, VA) supplemented with 10% fetal calf serum (SigmaAldrich Corp., St. Louis, MO). Cells were selectively stained with the lipophilic 3,3′-dihexyloxacarbocyanine iodide (DiOC6) fluorescent membrane dye (Molecular Probes, Eugene, OR) using the following procedure: Nonadherent cells were removed from the culture dish, and 10 mL of complete media with 5 µL of 1 N DiOC6 (in methanol) was then loaded into the culture dish and allowed to incubate with the adherent cells for ∼1 min at 37 °C. The adherent cells were rinsed with fresh media and resuspended in 1 mL of complete media. Simulation. We used MATLAB (The Mathworks Inc., Natick, MA) and FEMLAB (COSMOL Inc., Burlington, MA) for carrying out the fluidic simulations. RESULTS AND DISCUSSION Design of Microfluidic Systems. We have explored a number of microfluidic designs to produce controlled formation of single microvortices. Figure 1 shows four examples that produce recirculating flows. The streamlines shown in the figures are simulations, which match well with our experimental observations. Figure 1A illustrates how flow past a 90° corner induces flow detachment and causes the fluid to recirculate in the corner area; similar behavior was observed for flow past a microchamber (Figure 1C). Most sharp boundaries cause flow detachment that leads to the formation of a vortex at sufficiently high flow velocity, but the minimum Reynolds number (Re) at which recirculation occurs can vary widely depending on the particular geometry. For example, the minimum Re for the design in Figure 1C was ∼0.5. Analytical Chemistry, Vol. 76, No. 9, May 1, 2004

2493

Figure 1. Generation of microvortices in microfluidic systems. Recirculation flows occur under a wide range of conditions and channel geometries, including sharp corners (A), circular cavities that we call “rotors” (B), side chambers (C), and interconnect chambers (D). For rotation of microparticles, vortices with stable centers of rotation are required, which are present in the geometries shown in (C) and (D) but not in (A) and (B). (C) Although the straight channel velocity is high (red), the recirculation flow around the bead is low (blue). The upper inset is a fluorescence micrograph that shows the flow profile inside the diamond-shaped cavity as traced by 1 µm in diameter fluorescent beads. (D) By placing two microchambers adjacent to each other, counterrotating flows that we call “fluidic gears” can be created (counterclockwise in lower diamond and clockwise in the upper diamond). ([) Center of rotation of lower diamond. (9) Center of rotation of upper diamond. The Re at the opening of each microchamber was (A) 75, (B) 285, (C) 30, and (D) 40. Scale bars: (A) and (C) 30 µm; (B) and (D) 60 µm.

More importantly, most of these vortices were unsuitable for our experiments because they had poor stability in their center of rotation. For example, the center of rotation in the 90° corner (Figure 1A) moved laterally along the direction of flow (x-direction), as the fluid velocity varied. As the vortex core moved in relation to the rotating microparticle, the particle may be easily swept away by the flow. Figure 1B shows another example. This “rotor” design produced high rates of rotation where the small inlet channel fed fluid into the circular area where a vortex was formed, which had a rotational velocity that was ∼40% of the average inlet channel velocity. Again, the center of rotation in this design was unstable, moving a substantial distance from the inlet (at relatively low flow rates) to the center of the circular chamber with increasing flow velocity, which made stable particle rotation difficult. Panels C and D of Figure 1 show two designs that had stable centers of rotation, which enabled controlled rotation of the microparticle. The drawback of these designs was their high surface area-to-volume ratio, which tended to dampen the rotational velocity of the microvortex. The simulated three-dimensional velocity plot in Figure 1C shows high flow velocity in the center of the straight channel, as expected from pressure-driven flow, but the fluid velocity that drove recirculation at the chamber opening was only ∼20% of the average flow velocity in the straight channel.28 Here the average straight channel velocity was ∼1 m/s 2494 Analytical Chemistry, Vol. 76, No. 9, May 1, 2004

with a Re at the cavity opening of ∼6, and the velocity of fluid rotation in the chamber peaked at ∼0.2 m/s. The upper inset in Figure 1C is a fluorescence image showing the flow profile in the chamber, which was traced using 1-µm-diameter fluorescent beads. The design shown in Figure 1Dswhich we termed “fluidic gears” because the larger vortex adjacent to the straight channel was used to drive a counterrotating vortex in the inner chambers had an especially stable center of rotation. Particles located in the inner chamber were isolated from slight perturbations in flow in the straight channel, such as rapid changes in fluid velocity when syringes in the syringe pump were exchanged. The drawback with this design was again the dampened fluid velocity, which was even more pronounced in this design than compared with the single-chamber design (Figure 1C). The advantage of a stable center of rotation, however, outweighs the loss of fluid rotational velocity because the straight channel velocity may be increased to compensate for the dampened fluid velocity in the chamber. Optical Marking and Rotation of Microparticles. Parts B and C of Figure 2 outline our procedure for marking and monitoring the rotation of a single bead. We first applied a single pulse from a highly focused N2 laser to create a small pockmark on the periphery of the optically trapped particle. Once the bead was (28) Lim, D. S. W.; Shelby, J. P.; Kuo, J. S.; Chiu, D. T. Appl. Phys. Lett. 2003, 83, 1145-1147.

Figure 2. Optical marking of optically trapped microparticle and monitoring of the rotation of the particle with backscattered laser light. (A) The setup uses a Nd:YAG laser for optical trapping, a N2 laser for ablation, an Ar+ laser for illuminating the optically trapped and marked bead, and confocal single-molecule detection for monitoring the backscattered laser light from the bead. The incident power of the Ar+ laser was 5 mW, and the pulse energy of the N2 laser was 45 µJ, both measured prior to the objective. M, mirror; APD, avalanche photodiode. (B) An optically trapped bead (trap not shown for clarity) in the microchamber is first ablated with a single pulse from the N2 laser, etching a mark (∼1 µm) on the bead (black circle). (C) The focus of the Ar+ laser is then positioned onto the mark, and the bead is rotated (black arrow). The rotation of the bead is monitored by following the changes in intensity of the backscattered laser light (gray arrow), which is collected by the microscope objective and directed to the confocal pinhole and the APD. The presence of the mark increases the amount of backscattered light, thereby giving rise to an increase in the number of detected photons.

marked, the focus of an Ar+ laser, which was aligned collinearly with the N2 laser, was used to illuminate the pockmark so differences in the backscattered light as the pockmark left and reentered the laser focus could be monitored using the confocal setup. The presence of the pockmark generated a photon spike in our detected optical signal. For clarity, Figure 2B and C omitted the optical trapping beam. The average size of an optical mark on a polystyrene bead was 1-3 µm. The range of bead size we used was 10-25 µm in diameter, but we have marked beads as small as 4 µm. The width of our detected peak caused by the backscattered laser light was a function of (1) the width of the optical mark, (2) the width of the beam waist of the Ar+ laser focus, and (3) the frequency of rotation of the microparticle. The width of the beam waist was measured to be ∼500 nm, well below the average diameter of

Figure 3. (A) Schematic of a four-diamond microchamber, which gives rise to three counterrotating flows. The open arrow indicates the position of the bead in the chamber. (B) A 15-µm bead (inset) is rotated in a microvortex at 0.15 Hz. (C) A 10-µm bead is rotated in a microvortex at 100 Hz. The Re at the opening of the microchamber was (B) 1.5 and (C) 150. Scale bar in (B) represents 15 µm.

the mark; therefore, the dominant factors in determining peak width were the width of the mark and the rotational frequency of the bead. Characterizing the width of a peak is thus one method of identifying the physical markings on the microparticle that produce a signal. For example, in Figure 3B, the width of the peak at half-height is 0.48 s in duration. The mark on the 15-µm bead is 3.5 µm in width (inset). At a measured rotational frequency of 0.15 Hz, the radial velocity of the bead is 7.1 µm/s. The predicted duration of the peak based upon the measured width of the optical mark is therefore 0.49 s, which is in excellent agreement with the measured peak duration (2% error). Dynamic Range. Figure 3B shows a slow, controlled rotation of a 15-µm bead (inset). The bead was marked in one location with the pulse N2 laser and spun in a four-diamond gear structure (Figure 3A), a design we used for all experiments shown in subsequent figures. The rotational frequency of the bead was 0.15 Hz, which we could maintain consistently until the fluid supply in the syringe was exhausted (∼10-12 min for a 3-mL syringe). Figure 3C shows a relatively fast rotation of a 10 µm in diameter bead, for which the rotational frequency was ∼100 Hz. The rate at which rotation occurs at a given flow rate (or Re) also depends on the trapping power of the laser, which tends to dampen microparticle rotation, as well as on the exact placement of the trapped particle with respect to the vortex core. The limiting factor to high rotational rates in our experiments is currently determined by the ability of our optical trap to hold the particle at the center of rotation at high flow velocities. With further optimization and Analytical Chemistry, Vol. 76, No. 9, May 1, 2004

2495

Figure 4. Analysis of the rotation of an anisotropically marked microparticle. Three distinct marks on a 15-µm bead (1-3 in the inset) can be detected during bead rotation. The frequency of rotation measured from all three sets of marks (f1-f3) was ∼3.8 Hz. The length of the arcs between the marks (6.8, 9.6, and 18.2 µm starting with mark 1) was determined both by measuring the corresponding distances in the CCD image (inset) and more accurately by the corresponding time delays between the peaks detected by the APD (t1-t3). The Re at the opening of the microchamber was 30. Scale bar represents 15 µm.

better designs for maintaining a fast rotating particle, significantly higher single-particle rotation rates may be possible. Monitoring Rotation of a Microparticle with Multiple Facets. Figure 4 shows a 15-µm bead with three distinct marks (inset). The distance separating the marks, which are designated 1-3, as measured using the image from the CCD camera matches well the values obtained using the signals from the confocal detection (1-2 ) 6.8 µm, 2-3 ) 9.6 µm, 3-1 ) 18.2 µm). The rotational frequency measured using the three different marks (f1-f3) all resulted in 3.84 Hz, as anticipated. The width and intensity of the detected peaks also correlate well with the width of the mark and the path through which the laser focus traversed the mark. This excellent correlation demonstrates the potential of the technique to provide insight into the physical structure of the microparticle during rotation. Detection of Effects Caused by Wall Damping. In pressuredriven flow, fluid movement is dampened by the presence of immobile surfaces owing to the no-slip boundary condition. In the diamond-shaped cavity (Figure 1C), the fluid encounters four vertical walls, in addition to the ceiling and floor of the chamber, all of which slow fluid flow. Because the rate at which particles rotate is driven by the recirculation of the surrounding fluid, which in turn is affected by the walls of the chamber and the available space between the walls and the particle, it can be expected that the size of the bead will influence its rotation frequency at a given applied flow rate in the straight channel. Figure 5 shows a characterization of the effect of the size of the bead on its rotational rate. The beads varied from 10 to 25 µm in diameter and were rotated in one of the small inner chambers located adjacent to the large primary chamber (open arrow in Figure 3A). The rotational frequency increased as the diameter of the bead decreased because the flow that drove bead rotation was further away from the wall and was less dampened. This increase in rotational frequency from one bead size to another, however, was not linear, owing to the nonlinear parabolic velocity profile that drove bead rotation. Figure 5 also shows the onset of rotation occurs at a Re of ∼1. In general, the smallest Reynolds number 2496 Analytical Chemistry, Vol. 76, No. 9, May 1, 2004

Figure 5. Rotational frequency of the bead versus the Re at the opening of the chamber. The beads varied in diameter: (b) 10, (0) 15, ([) 20, and (]) 25 µm. The inset shows the top view of the chamber used to rotate the bead; black arrows indicate the direction of fluid flow.

Figure 6. Selective staining of living cells. (A) B-Lymphocytes resting on the surface of a tissue culture flask were subjected to a timed staining with the membrane dye, DiOC6, in which portions of the cells were stained (bright fluorescent regions). Bright-field (B) and fluorescent images (C) of a selectively stained B-lymphocyte. Scale bar in (B) represents 15 µm.

that is required for rotation depends strongly on the fluidic design as well as on the power of the trapping laser that is used to maintain the rotating particle at the vortex core. Spatially Selective Staining of Living Cells. Irreversible marking using the pulse N2 laser is obviously not practical for living cells. One method of selectively marking cells is the spatially defined delivery of fluorescent beads or molecules that bind to the cell surface receptors, which can then be used to track cell movement.29 More recent research has explored the use of quantum dot bioconjugates to track localized areas of living cells.30 These methods, although effective in labeling a select area of the cell, can be complex and time-consuming. Here we describe a (29) Wang, Y.-l.; Silverman, J. D.; Cao, L.-g. J. Cell Biol. 1994, 127, 963-971. (30) Jaiswal, J. K.; Mattoussi, H.; Mauro, J. M.; Simon, S. M. Nat. Biotechnol. 2003, 21, 47-51.

Monitoring the Rotation of Single Cells. We rotated the selectively stained cells in the four-diamond gear design (Figure 3A). To follow single-cell rotation, it was necessary to optimize the laser power because excessive power often resulted in the rapid photobleaching of the cell stain. Figure 7A shows the rapid decrease in signal intensity as the fluorescent label was photobleached. The rotational signal, however, can still be extracted from this plot; the cell was rotating at a rate of ∼1 Hz. The peak widths are greater here than with the bead rotation (Figures 3-5) because the bead marks were approximately 1-3 µm in diameter, but the cell stain encompassed approximately half of the cell volume. Figure 7B shows a cell at a steady-state rotational rate of 5 Hz. Similar to the analyses of the bead markings (Figures 3 and 4), the correlation between the width and frequency of the observed peaks, the rotation rate, and the observed area of the cell stain from the CCD image are all in good agreement (2% error).

Figure 7. Monitoring of the rotation of selectively stained lymphocytes. (A) The use of high laser power will cause photobleaching of the stained region of the cell, as indicated by the decrease in the detected fluorescence signal of a rotating (1 Hz) cell. (B) Signal from a steady-state rotation of a B-lymphocyte at 5 Hz. The Re at the opening of the microchamber in (B) was 25.

method for staining parts of the cell, which is easy to implement and is especially suitable for tracking the rotation of single cells. This technique relies on the timed staining of cells attached to the surface of the culture dish and in part on the cellular selectivity of the stain. The lipophilic dye, DiOC6, is a selective label for the mitochondria at low concentrations31,32 but can stain the nuclear envelope and other lipid membranes at high concentrations. Upon reaching the surface of the cell that is exposed to solution, DiOC6 will diffuse into the cell and stain the intracellular membranes. The diffusion of a sufficient amount of DiOC6 through the cell from the surface that is exposed to solution to the nuclear envelope is slow, and disruption of this process by removal of the dye from solution after a timed period (e.g., 1 min or less) will result in cells that are selectively stained (Figure 6). Figure 6B shows a close up view of a B-lymphocyte under bright-field illumination. The nucleus is faintly visible on the right-hand side of the cell. Figure 6C shows the same cell under fluorescence imaging, where the stained organelles are localized on the left-hand side of the cell adjacent to the nucleus. Using this procedure, greater than 90% of the cells were selectively stained. (31) Koning, A. J.; Lum, P. Y.; Williams, J. M.; Wright, R. Cell Motil. the Cytoskeleton 1993, 25, 111-128. (32) Voeltz, G. K.; Rolls, M. M.; Rapoport, T. A. EMBO J. 2002, 3, 944-950.

CONCLUSIONS This paper describes a method of marking, rotating, and monitoring the rotation of microparticles in microvortices using sensitive optical methods. The drawback of this approach is the semidestructive nature of the marking technique used to etch solid microparticles, but the strategy for the selective staining of living cells is nondestructive. The advantage of this approach lies in its excellent sensitivity and good spatial and temporal resolution, as demonstrated by following rotation rates over 3 orders of magnitude and the ability to identify structural facets present in the microparticle. We anticipate this technique will be generally useful for monitoring the rotation of micrometer-size particles in real time. In addition, the use of microvortices may prove to be a flexible and convenient platform for inducing controlled rotation of micro- and nanoparticles. In principle, the rotational signal detected optically can be used as a feedback signal to control the flow rate in the channel and thus the rotation rate of the particle. The ability to precisely manipulate and monitor the rotation of microparticles is anticipated to be broadly useful for studying the effects of rotation on biological systems, such as individual living cells, in particular, the effect of high shear stress and radial acceleration that is present in microvortices on the behavior of suspension cells, since such cells are constantly subjected to similar hydrodynamic forces in the body. ACKNOWLEDGMENT We gratefully acknowledge funding of this work from the NSF and the NIH. Received for review December 22, 2003. Accepted February 24, 2004. AC035531V

Analytical Chemistry, Vol. 76, No. 9, May 1, 2004

2497