Micro Coulter counting device

Activities

We designed and fabricated a two stage micro Coulter counting device to perform cell detection via transit time change induced by magnetic field. The device consists of two identical micro Coulter counting channels, one inlet, one outlet and a chamber between two counters. Cells suspended in medium were loaded to the inlet reservoir and were drive through the channel via a syringe pump. Each cell passing through the device will create two individual resistive pulses in 1st Counter and 2nd Counter. The magnitude of the pulse is proportional to the cell size, while each pulse width represents the cell’s transit time through corresponding counter.

Standard soft lithography process was used to fabricate the device. First, a SU8 mold, consisting patterns for two micro-channels, fluidic chamber, inlet and outlet reservoirs, was created using photolithography. Microchannels, fluidic chamber and the reservoirs were formed by pouring Polydimethylsiloxane (PDMS) on the SU8 layer, followed by degassing and curing. Next, the PDMS was bonded onto a glass substrate after an oxygen plasma treatment (200 mTorr, 50 W, 50 s). The nominal dimensions of the two identical sensing channels are 50 μm wide and 40 μm deep. The channel dimensions measured by a surface profilometer (Dektak 150, Veeco Instrument) are 51.8 ± 0.9 μm wide and 36.9 ± 0.2 μm deep. One pair of Ag/AgCl electrodes (1 mm in diameter) were inserted on each side of each Coulter counting channel via two punched 1-mm holes to finish the fabrication of the resistive pulse sensor. A N40 grade cylindrical rare earth magnet (R125A-DM, Amazing Magnet) was mounted below the glass substrate to create a high gradient magnetic field (∇B^2=67.62 T^2/m) over the 2nd counter.

Fig. 1. Pictures of the 2-stage micro Coulter counting device and its sensing channels.

1.2 Device testing

Fig. 2. Measurement setup

Fig. 2 shows the equivalent circuit to measure the resistive pulse on each sensing channel. The resistances of two sensing channels are modeled as Rchannel_1, Rchannel_2. Each cell passing through the sensing channel created a resistive pulse. The resistive pulse was converted to a voltage pulse when we applied a DC voltage at Vinput. With an operational amplifier AD 620, the original voltage pulses from each channel were amplified to a measurable level, as the voltage output signal Voutput_1 and Voutput_2. In order to get the optimized output signals, R1 was a variable resistor set equal to the initial channel resistances (Rchannel_1, Rchannel_2), R2 = R3 = 500kΩ, Vinput = 2V, and the AD 620 operational amplifier was set at a gain of 184 when the value RG=270Ω. The two output signals Voutput_1 and Voutput_2 containing the voltage pulses information were acquired by a NI-DAQ board and record by a customized LabView program with a sampling rate at 300 KHz. From the measured signals, size and transit time of every cell passing through each counter can be back calculated.

To ensure the microfluidics device can accurately measure the sizes and counts of cells, standard polystyrene micro-particles (72986 FLUKA, Sigma Aldrich, USA) were used to test each micro Coulter counter before the cell testing. In addition, we have tested viability of cells after they were tested under different flow rate.

After device calibration, we tested mixed cells to validate the design concept of the microfluidics magnetic beads assay (as described in section 4.1). Mixture cell groups at six different target cell ratios (2%, 5%, 10%, 50%, 75%, and 100%) were tested and corresponding transit time change for each group were measured and analyzed. Based on the relationship between transit time and target cell ratio, we also estimated the limit of detection for this integrated device.

Findings

After the 2-stage Coulter counting device was fabricated, we calibrated the device using standard polystyrene 10µm micro-particles (density 1.05g/cm³). From the measured pulse magnitudes, the back calculated particle size is 10.00± 0.37 µm, which is close to the manufactured size of 10.00± 0.20 µm. Additionally, cells counts from the 1^st and 2^nd counter are nearly the same with a small difference of 0.5% at various rates. The measured particle concentration, back calculated from number of pulses over a period of time, is the same as the actual particle concentration. This test confirmed the validity of the device for measuring the sizes of and numbers of the microparticles.

Next, mixture cell groups at six different target cell ratios (2%, 5%, 10%, 50%, 75%, and 100%) were tested with the device. Shown in Figure 3, as the target cell ratio was varied from 2% to 100% the average transit time delay increased nearly linearly from 0.98% to 18.11 %. For each fixed tar-get cell ratio, standard deviations of the three replicates were plotted in Figure 3. The standard deviations were likely caused by various surface antigen affinities of cells from different passages or different subpopulations of the same passage.

Fig. 3. Transit time delay measured from mixture cell suspensions with six different target cell ratios. Three replicates were tested at each target cell ratio. Each replicate contains 10⁵ total cells in 200 µL PBS buffer.

Afterwards, we conducted experiments using HUVECs as blank samples to estimate limit of detection (LOD) of the device. The transit time delays were measured and the standard deviation (S) was calculated to be 0.29%. The slope of the calibration curve (in Figure 3), b, is 0.17 from linear curve fit-ting. According to the International Conference on Harmonisation (ICH) guidelines, the LOD = 3.3 × S/b and LOQ= 10 × S/b. Hence the limit of detection (LOD) and limit of quantification (LOQ) of the device in terms of cell ratio were estimated to be 5.6% and 17.1% respectively. We also conducted the same blank experiments using the ferrozine-based colorimetric assay. The LOD and LOQ were estimated to be 7.9% and 23.9%.