Microfluidic platform combining droplets and magnetic tweezers: application to HER2 expression in cancer diagnosis

Abstract

The development of precision medicine, together with the multiplication of targeted therapies and associated molecular biomarkers, call for major progress in genetic analysis methods, allowing increased multiplexing and the implementation of more complex decision trees, without cost increase or loss of robustness. We present a platform combining droplet microfluidics and magnetic tweezers, performing RNA purification, reverse transcription and amplification in a fully automated and programmable way, in droplets of 250nL directly sampled from a microtiter-plate. This platform decreases sample consumption about 100 fold as compared to current robotized platforms and it reduces human manipulations and contamination risk. The platform's performance was first evaluated on cell lines, showing robust operation on RNA quantities corresponding to less than one cell, and then clinically validated with a cohort of 21 breast cancer samples, for the determination of their HER2 expression status, in a blind comparison with an established routine clinical analysis.

Figure 1. Magnetic tweezers technology and experimental workflow for RT-PCR experiment.

(a) Picture of the magnetic tweezers with the capillary (highlighted with a red liquid). A second passive magnetic tip, placed opposite to the first with regards to the capillary, contributes to field lines shaping and to the optimization of the magnetic force. (b–h) Sequence of images showing the extraction and the redispersion of magnetic beads from one droplet to another one (colored in orange), switching ON and OFF the electric current in the coil. (i) Scheme of a typical train of droplets used for the RT-qPCR analysis: 1) oligo-dT magnetic beads, 2) total RNA sample, 3) washing buffer and 4) RT-PCR mix. (j–o) Workflow of beads and droplets manipulation during the protocol.

Figure 2. Microfluidic platform for the generation of trains of droplets.

(a) Scheme of the droplet microfluidic generator module, which combines a MTP plate placed under a motorized pipettor arm (Cetoni), two syringes, two pinch valves (b) and two PDMS chips (c,d). The connections between the PDMS chips inlets and the different parts of the platform are numbered from 1 to 6 and the same numbers are reported on the scheme (a). Syringe 1 is connected with inlet 1 of PDMS chip 1 (Fig. 2c) while syringe 2 is connected with PDMS chip 2 (Fig. 2d) at inlet 5; inlets 2 and 3 are connected respectively to the pipettor head and to the PDMS device 2 (inlet 4), thanks to tubing passing through the two valves (V1 and V2). Finally, inlet 6 of the second PDMS chip is connected with the capillary in which droplets are driven between the magnetic tweezers. In order to generate droplets trains, while other droplets are flowing in the rest of the platform, homemade software switches alternatively between two conditions represented by red and blue lines. In condition 1, V1 is open and V2 is closed; syringe 1 is in aspiration mode generating and storing droplets in the PDMS device 1 while syringe 2 is pushing oil in the rest of the circuit. Then, in condition 2 (V1 closed, V2 open), syringe 2 is stopped and syringe 1 pushes the generated droplet train over PDMS device 2; after that, condition 1 is reactivated, and a new droplet train can be generated while the previous one is flowing in the processing system. In order to avoid pinching of a droplet by a valve, 10 mm oil spacers were introduced between droplet trains, and pipetting and valving were synchronized to ensure that valves close only onto these spacers. Each train in the series of trains can be prepared from any arbitrary combination of samples and reagents. The water and oil phase solutions are placed in the MTP, which is covered by a mineral oil film of 2–3 mm for avoiding contaminations or solution evaporations. The length of the scale bar of (b–d) is 1 cm.

Figure 3. qPCR calibration curves.

Calibration curves for (a) MCF7 and (b) SKBR3 cell lines for genes HER2 and TBP, represented in blue and green, respectively. Cts are plotted in logarithmic scale as a function of the total RNA concentration. The qPCR efficiency Eff, reported below the fit parameters, was extracted from the slope of the linear fits. (c) Calibration curves for MCF7 without (blue and green points and fits) and with (red and black points and fits) 15 PCR pre-amplifications cycles: as expected, the curves are shifted of about 15 Cts. Error bars are not visible when they are smaller than the point.

Figure 4. Correlation between hospital and microfluidic platform results.

Graph in Logarithm scale (base 2) of the obtained by the Saint-Louis Hospital platform versus the obtained by the droplet microfluidic platform. Samples disclosed a posteriori as HER2- samples are plotted in red, and HER2+ ones in black. The dashed lines help to identify the excluded zone between the two populations: blue for the droplet platform (10.48–39.67), green for the hospital (1.5–11.71).