A digital microfluidic system for loop-mediated isothermal amplification and sequence specific pathogen detection

A digital microfluidic (DMF) system has been developed for loop-mediated isothermal amplification (LAMP)-based pathogen nucleic acid detection using specific low melting temperature (Tm) Molecular Beacon DNA probes. A positive-temperature-coefficient heater with a temperature sensor for real-time thermal regulation was integrated into the control unit, which generated actuation signals for droplet manipulation. To enhance the specificity of the LAMP reaction, low-Tm Molecular Beacon probes were designed within the single-stranded loop structures on the LAMP reaction products. In the experiments, only 1 μL of LAMP reaction samples containing purified Trypanosoma brucei DNA were required, which represented over a 10x reduction of reagent consumption when comparing with the conventional off-chip LAMP. On-chip LAMP for unknown sample detection could be accomplished in 40 min with a detection limit of 10 copies/reaction. Also, we accomplished an on-chip melting curve analysis of the Molecular Beacon probe from 30 to 75 °C within 5 min, which was 3x faster than using a commercial qPCR machine. Discrimination of non-specific amplification and lower risk of aerosol contamination for on-chip LAMP also highlight the potential utilization of this system in clinical applications. The entire platform is open for further integration with sample preparation and fluorescence detection towards a total-micro-analysis system.


On-chip fences fabrication
Including the LAMP reaction and MCA, the entire DNA detection process lasted about 1 h, during which droplets might drift in an unpredictable manner from their reaction spots owing to the highly hydrophobic surface of the top and bottom plates. To affix the droplet at a specific position, different methods can be applied including continuously charging specific electrodes or changing the hydrophobicity of the surface on certain electrodes 1,2 . However, these measures can cause electrode breakdown or can reduce reaction efficiency if DNA and protein molecules adhere to the hydrophilic surface and lose amplification activity. In this work, we addressed the issue in a more physical way. On one hand, a spacer of 200 μm height was utilized to increase the surface/height ratio of on-chip droplets, which increased the friction against the top and bottom plate. On the other hand, on-chip SU-8 fences of an approximate 100-μm height were designed and fabricated around the transport and reaction electrodes. These implements successfully prevented the droplets from drifting during sample loading and constrained the droplets at their reaction spots during reactions. Figure S1. 3D schematic of the DMF chip fabrication. The top plate with ITO downward was spin-coated with a thin layer of Teflon (~100 nm). The Cr-patterned bottom plate (wet etched) was spin-coated with 2 layers of SU-8, and 1 thin layer of Teflon. Upon sealing of the whole chip, liquid metal was deposited on the Cr ground pad on the bottom plate (outside the sealed chamber) for connecting the ground pad (bottom plate) with the ITO layer (top plate). Finally, the top plate, PMMA frame spacer (200 μm) and bottom plate were aligned and UV glued.  Figure S3. Operation and experiment setup of the DMF LAMP system. (a) The loading mode (for sample loading) of the system comprises a electronic control system (works as the electrical connection and control electronics for task execution and signal acquisition), a signal generator and a transformer for high voltage generation, plus a operation system (in-house computer software) for actuation control; (b) The measuring mode (for LAMP reaction and MCA) consists of a electronic control system (works as a heater with a temperature sensor for temperature control), a fluorescence microscope for images capture, and an operation system for temperature regulation, data acquisition and capture of fluorescence images. The final reaction volume containing 0.4 μL of DNA template (10,000 copies/reaction) or 0.4 μL of TE for no-template-control (NTC) was adjusted to 10 μL with ddH2O and sealed with a drop of hexadecane (Sigma-Aldrich, Germany) and the PCR tube cap. Each reaction was run in duplicate at 65 °C for 60 min (fluorescence recorded every 1 min), followed by a step at 85 °C for 5 min to inactivate the Bst. MCA was conducted after the stabilization at 30 °C for 5 min, followed by melting from 30 °C to 90 °C with a 1 °C interval for 2 s (fluorescence capture time was approximately 12 s per step).

System setup Supplementary
To demonstrate the inhibition effect of SYBR Green I dye, off-chip LAMP reactions with 2-fold (0.8×), 5-fold (2×), and 10-fold (4×) of the concentration of SYBR Green I dye we used in the manuscript were run. As Figure S6a shows, 2-fold (0.8×) of SYBR Green I caused a delay of the reaction by 5.6 min comparing to 0.4×. Melting peaks in Figure S6b confirmed the products to be true positive. 5-fold (2×) and above of SYBR Green I completely inhibited the LAMP reaction. The NTCs of both samples of 0.4× and 1 NTC of 0.8× showed positive amplification curve but negative Molecular Beacon probe signals ( Figure S6b), indicating non-specific amplification.

Photobleaching effect of SYBR Green I
The reaction mix for on-chip LAMP was identical to that for off-chip LAMP. One 1-μL reaction droplet was loaded on the DMF chip for continuous photobleaching test. Four 1-μL reaction droplets were loaded onto the chip and driven to the reaction spots for LAMP reactions. Two were positive samples containing 1,000 copies of DNA per reaction droplet. Two were NTCs. A fluorescence microscope (Olympus DP80, Japan) was used to monitor the SYBR Green I signals. For continuous photobleaching test, the droplet was exposed under green light from GFP channel for 120 s (signal recorded continuously). For on-chip LAMP reactions, fluorescence signals were captured every 30 s with an exposure time of 0.5 s (GFP channel). Started at 30 °C, the temperature rised up to 67 °C. After reaction for 60 min, temperature was restored to 30 °C. The whole capture time was 85 min. All of the fluorescence signals were captured in a dark environment. Different concentration combination of Mg 2+ and dNTP were tested. We analyzed the melting curve of LF probe and the synthesized single-stranded DNA target binding. As Table S2 and Figure S8 show, Tm raises along with the Mg 2+ concentration increases, whereas Tm goes down when the dNTP concentration increases. The concentrations of Mg 2+ (8 mM), dNTP (1.4 mM) and the product mass in the LAMP reaction are in a high level, so the slight shifting in Tm is reasonable.

(ii) Nucleotide variations in LF probe target
To verify the Tm scattering is caused by the probe-target binding instead of the Molecular Beacon probe-LAMP assay, we used LB probe to do the serial dilutions under the same reaction condition as the LF probe. The target of LB probe has less nucleotide variations than the target of LF probe (Table S1). As Figure S9 shows, the melting peaks are much more centralized (57-58 °C). This result strongly supports that the Tm scattering phenomenon of LF probe is caused by the target nucleotide variations. In conclusion, the Tm scattering phenomenon of LF probe is mainly caused by the nucleotide variations in the target sequence, which is a very unique case of this model system. This can be prevented by staying away from the unconserved sequence in the primer and probe design step. Also, the complexity of the melting solution contributes in the Tm scattering, causing a tolerable small range shifting of the Tm (1-2 °C). These do not affect the MB probe in denoting the specificity of the LAMP product.

Molecular Beacon probe-target melting curve analysis for alternative target
The synthetic target sequences of T. brucei RIME for Molecular Beacon LF probe are listed below: Perfect match target (PM):  Supplementary Movie S1. Sample loading on digital microfluidic chip One microliter of sample was injected through the inlet and pulled into the chamber by the first two energized electrodes. Then, a programmed electrode charging sequence led the droplets to their reaction spots automatically.Video was accelerated by 5 folds.