Droplet-based pyrosequencing using digital microfluidics

Abstract

The feasibility of implementing pyrosequencing chemistry within droplets using electrowetting-based digital microfluidics is reported. An array of electrodes patterned on a printed-circuit board was used to control the formation, transportation, merging, mixing, and splitting of submicroliter-sized droplets contained within an oil-filled chamber. A three-enzyme pyrosequencing protocol was implemented in which individual droplets contained enzymes, deoxyribonucleotide triphosphates (dNTPs), and DNA templates. The DNA templates were anchored to magnetic beads which enabled them to be thoroughly washed between nucleotide additions. Reagents and protocols were optimized to maximize signal over background, linearity of response, cycle efficiency, and wash efficiency. As an initial demonstration of feasibility, a portion of a 229 bp Candida parapsilosis template was sequenced using both a de novo protocol and a resequencing protocol. The resequencing protocol generated over 60 bp of sequence with 100% sequence accuracy based on raw pyrogram levels. Excellent linearity was observed for all of the homopolymers (two, three, or four nucleotides) contained in the C. parapsilosis sequence. With improvements in microfluidic design it is expected that longer reads, higher throughput, and improved process integration (i.e., "sample-to-sequence" capability) could eventually be achieved using this low-cost platform.

Figure 1

(A) Photograph of the assembled multiwell-plate-sized PCB-based cartridge, (B) photograph of the sequencing instrument, and (C) schematic illustration of the cartridge showing locations of sample and reagent wells and other features.

Figure 2

Characterization of signal loss (A), wash efficiency (B), and pyrophosphate carryover (C). Variable DNA inputs of a synthetic template ranging from 150 to 460 fmol were used in these experiments. (A) DNA/primer/bead complexes were washed 100, 200, 400, and 800 times prior to incorporation of dTTP. There is little difference in signal intensity indicating minimal DNA, primer, or bead loss. Background signal was collected from beads that were not conjugated with DNA. (B) After incorporation of a dTTP, residual signal was determined during each of 10 wash cycles. Six wash steps were sufficient to attain background signal levels. (C) Alternating high-concentration pyrophosphate (PPi) and wash droplets were passed under the detector and show no change in background signal suggesting no surface fouling. The inset shows wash signal on the appropriate scale for five separate droplets. The high pyrophosphate concentration was approximately 40 times greater than the expected concentration of pyrophosphate generated from a single nucleotide incorporation.

Figure 3

Reproducibility of pyrophosphate detection (A), raw pyrograms from pyrophosphate detection (B), de novo sequencing (C), and “split” protocol resequencing (D) using the same C. parapsilosis template. (A) AUC (photons) for sequential detections of different concentrations of pyrophosphate. Numbers indicate the homopolymer length which is approximately equivalent to the amount of pyrophosphate in the reaction. (B) Pyrogram (AUC) showing 52 cycles of nucleotide addition using a targeted resequencing protocol which included periodic mismatches to evaluate background signals. The nucleotide that was added is indicated by the letter above each bar with uppercase letters indicating expected incorporations and lowercase letters indicating expected mismatches based on the known sequence. A 62 bp sequence can be accurately read from the pyrogram by assigning a threshold level for each number of nucleotide incorporations: TTTGA⁵CTATT¹⁰AAATA¹⁵ATCGG²⁰TTGAC²⁵ATTAA³⁰ATAAA³⁵ATTTG⁴⁰GTTGA⁴⁵GTTTA⁵⁰ATCTC⁵⁵TGGCA⁶⁰GG. (C) Pyrogram (AUC) showing 44 cycles of nucleotide addition for the same template using a de novo pyrosequencing protocol in which the four bases were repeatedly cycled in the same order. (D) Pyrogram (AUC) showing 34 cycles of nucleotide addition for the same template using a resequencing protocol in which the synthesis and detection reactions were performed in separate droplets at different times and locations on the cartridge.