Human mammalian cell sorting using a highly integrated micro-fabricated fluorescence-activated cell sorter (microFACS)

Abstract

We demonstrate a high performance microfabricated FACS system with highly integrated microfluidics, optics, acoustics, and electronics. Single cell manipulation at a high speed is made possible by the fast response time (approximately 0.1 ms) of the integrated PZT actuator and the nozzle structure at the sorting junction. A Teflon AF-coated optofluidic waveguide along the microfluidic channel guides the illumination light, enabling multi-spot detection, while a novel space-time coding technology enhances the detection sensitivity of the microFACS system. The real-time control loop system is implemented using a field-programmable-gate-array (FPGA) for automated and accurate sorting. The microFACS achieves a high purification enrichment factor: up to approximately 230 fold for both polystyrene microbeads and suspended human mammalian cells (K562) at a high throughput (>1000 cells s(-1)). The sorting mechanism is independent of cell properties such as size, density, and shape, thus the presented system can be applied to sort out any pure sub-populations. This new lab-on-a-chip FACS system, therefore, holds promise to revolutionize microfluidic cytometers to meet cost, size, and performance goals.

Fig. 1

(a) Device structure. The 250 μm wide main fluidic channel is split into three sub-channels. The center channel is for collecting waste, while the left and the right channels are for collecting samples. The illumination light (488 nm laser) is delivered to the device by the optical fiber and guided by the Teflon AF coated optofluidic waveguide. The PZT actuator is integrated on the device. In the square is the sorting junction of the device made of PDMS. (b) As the PZT actuator bends down, the cell of interest is pushed to the right sorting channel, while the non-targeted cell travels directly to the center waste channel without triggering the PZT. (c) Flow pattern observation. Left: Trace of a fluorescent bead sorted to the right channel by superimposing photos taken every 0.3 ms using a high-speed CMOS camera. Right: The bead trajectory plot for the bead under different voltage magnitudes to the PZT actuator. This helps set the threshold voltage for sufficient deflection.

Fig. 2

Schematic of the optical detection and control/sorting system setup. Specially designed spatial filters (masks) are placed at the image plane to modulate incoming fluorescence signals before those signals are registered by the PMT.

Fig. 3

Spatial filters (masks) are specially designed and placed at the magnified image of the device feature. The input fluorescence pulse signal from stained cells is modulated by different spatial filters before being registered by the PMT, yielding different waveforms of photocurrents in time domain, corresponding to different locations of the cells as they travel through the microfluidics channel, such as (111), (1101) or (1011). This space-time coding technology reduces the size and the cost of the system by using only one PMT to differentiate 3 signals or even more.

Fig. 4

(a) The FPGA implemented real-time process control unit consists of two sub-sections: the detection section and the control section. The timing jitter of the system is less than 10 μs, enabling the real-time control. The match filter is a critical component that enhances the signal to noise ratio significantly. (b) Comparison of the (111) coded raw signal with 3 small peaks with the amplified signal by applying the FIR match filter algorithm.

Fig. 5

Example of space-time coding signals out of 10 μm fluorescent polystyrene beads. The first signal coded as (111) represents the detected fluorescence when the bead passes the detection zone(e.g. three transparent slits). After sorting, the second signal coded as (1011) that is ~5 ms trailing the detection signal indicates that the bead has been correctly switched into the sorting channel, confirming the successful sorting event. Note that the first signal allows one to estimate the velocity of the bead roughly, helping determine the right delay time for the exact triggering of the PZT actuator. Here, the bead travels with a velocity of around 5.5 cm s⁻¹.

Fig. 6

(a) The scattering plots show the result of sorting fluorescent 10 μm beads from non-fluorescent 5 μm beads. (b) The population ratio of the initial bead mixture is 0.67 : 100. After sorting for 30 min, the mixed ratio becomes 1.3 : 1, yielding an enrichment factor of around 200. (c) Before sorting (left), most of cells are non-fluorescent as the mixed ratio of fluorescent cells to non-fluorescent cells is 1 : 150. After sorting, the histogram (right) shows that the population of the sample (i.e. fluorescent cells) is purified with an enrichment factor of 230.