Large Scale Single Cell Analysis Using High Density Hydrodynamic Trapping Arrays for Quantitative System Biology and Rapid Medical Diagnostics

This paper reports a novel approach for performing large scale single cell analysis which has significant advantages over miniaturized flow cytometry and laser scanning cytometry (LSC). Regular trapping arrays allow for high density analysis and ease image processing. Moreover, on-chip sample preparation (e.g. fluorescent labeling, washing) is performed, as opposed to manual intensive operations of incubation, centrifugation, and resuspension in previous techniques - saving time and reagents. Additionally, time-dependent phenomena of a large number of single cells over different time scales are characterized using this device. This method is well-suited to quantitative systems biology and rapid medical diagnostics.

Previously, microfluidic traps and dams have been used to trap multiple single cells [1,2] without the control over the number of cells trapped, or precise cell position necessary for quantitative analysis of large numbers of cells. Single cells have also been trapped [3,4] and analyzed, but not in a high density manner as is presented. Others have successfully used dielectrophoretic trapping of single cells in arrays [5] in a similar approach towards array cytometry.

Here we employ a purely hydrodynamic method of cell trapping where single to multiple cells are trapped by varying trap dimensions (Fig. 1). The U-shaped PDMS traps are suspended above the substrate with a 2 micrometer gap to allow flow into the trap.

High-density single-cell isolation. (a,b) A schematic diagram is shown to describe the mechanism of cell trapping using flow-through arrayed suspended obstacles. Two-layer (40 and 2 ím) cup-shaped PDMS trapping sites allow a fraction of fluid streamlines to enter the traps. After a cell is trapped and partially occludes the 2-ím open region, the fraction of streamlines through the barred trap decreases, leading to the self-sealing quality of the traps and a high quantity of single-cell isolates. Drawing is not to scale. (c) A phase constrast image of an array of single trapped cells is shown. The scale bar is 30 um.

Figure 2. Statistics of single-cell isolation (a-d) Phase contrast micrographs of cell trapping in varying geometry cell isolation traps are shown. From (a) to (d), trap depth varied as 10, 15, 30, and 60 um. The number of cells trapped scales with the trap size, with more trapping of single cells observed as the trap size decreases. (e) The distribution of trapped cells for the geometry shown in (a) is plotted along with a Poisson distribution for the same average value. If the probability of trapping was independent of the amount of previously trapped cells, one would expect a Poisson distribution. In this case, an enhancement of single-cell-containing traps and a reduction of zero and greater than two cell-containing traps is observed above the random process. Here data from four separate loadings of 100 íL of cell solution containing 3 million cells mL-1 flowed through the device before data were collected.

Figure 3. Single-cell array enzymology. (a) For the average of a number of single 293T (closed circles), HeLa cells (open circles), and Jurkat cells (open squares), the amount of calcein AM converted to fluorescent product is plotted as a function of time. Error bars represent plus and minus one standard deviation. The model predictions with best-fit intracellular carboxylesterase concentrations are also plotted (solid line). The average amount of carboxylesterase for 293T cells was 123 ( 32 nM (N ) 21) and for Jurkat cells was 100 ( 11 nM (N ) 14), while for HeLa cells it was much lower: 47.7 ( 9.5 nM (N ) 15). (b) A histogram of the predicted enzyme concentrations for 293T, Jurkat, and HeLa cells is shown with Gaussian fits.

References:

1. M. Yang et al. Anal. Chem. 74, 3991 (2002).
2. A. Valero et al. Lab Chip. 5, 49 (2005).
3. M. Khine et al. Lab Chip. 5, 38 (2005).
4. A. R. Wheeler et al. Anal. Chem. 75, 3581 (2003).
5. J. Voldman et al. Anal. Chem. 74, 3984 (2002).