A microfluidic device for preparing next generation DNA sequencing libraries and for automating other laboratory protocols that require one or more column chromatography steps

Abstract

Library preparation for next-generation DNA sequencing (NGS) remains a key bottleneck in the sequencing process which can be relieved through improved automation and miniaturization. We describe a microfluidic device for automating laboratory protocols that require one or more column chromatography steps and demonstrate its utility for preparing Next Generation sequencing libraries for the Illumina and Ion Torrent platforms. Sixteen different libraries can be generated simultaneously with significantly reduced reagent cost and hands-on time compared to manual library preparation. Using an appropriate column matrix and buffers, size selection can be performed on-chip following end-repair, dA tailing, and linker ligation, so that the libraries eluted from the chip are ready for sequencing. The core architecture of the device ensures uniform, reproducible column packing without user supervision and accommodates multiple routine protocol steps in any sequence, such as reagent mixing and incubation; column packing, loading, washing, elution, and regeneration; capture of eluted material for use as a substrate in a later step of the protocol; and removal of one column matrix so that two or more column matrices with different functional properties can be used in the same protocol. The microfluidic device is mounted on a plastic carrier so that reagents and products can be aliquoted and recovered using standard pipettors and liquid handling robots. The carrier-mounted device is operated using a benchtop controller that seals and operates the device with programmable temperature control, eliminating any requirement for the user to manually attach tubing or connectors. In addition to NGS library preparation, the device and controller are suitable for automating other time-consuming and error-prone laboratory protocols requiring column chromatography steps, such as chromatin immunoprecipitation.

Figure 1. Microfluidic device for automated NGS library preparation.

(a) Automated multi-column device mounted on a plastic carrier that provides wells for loading samples and reagents and for pressurized operation of the device. The wells used to load reagents for NGS library preparation are labeled. Chromatography columns for the selective binding and release of DNA were formed either with ChargeSwitch beads or with carboxylated beads. Reagents that were used exclusively with the carboxylated beads are labeled in green. (b) Schematic of single reactor unit for reaction mixing and DNA purification. The regions denoted in the reaction circuit are as follows: Green, Sample; Orange, Buffer; Blue, Enzyme. Red solid rectangular boxes represent activated valves that partition the individual circuits. (c) Parallelization of 16 reactors on chip for preparation of up to 16 independent libraries. Layout of the entire device without the valve map showing reagent inlets and the design for multiplex library generation. The serpentine metering channel designed to ensure reliable column packing is highlighted in orange. (d) Schematics showing cross-sections of purification columns loaded with either 1) ChargeSwitch beads, which are held in place with a frit layer and a cap layer formed by larger beads, or 2) carboxylated beads.

Figure 2. Quantification of E. coli strain DH10B library DNA after size selection.

(a) Illumina libraries; (b) Ion Torrent libraries. Asterisks indicate sample modules where buffer was loaded instead of genomic DNA. (c) Efficiency of library preparation reactions on the AMCC chip. The percentage of E. coli DNA fragments with Illumina sequencing adapters ligated onto both ends was estimated by RT-qPCR. The amount of E. coli genomic DNA present was determined by RT-qPCR with primer pairs recognizing six regions of the E. coli genome, and the amount of library DNA with adapters ligated onto both ends was determined by RT-qPCR with a primer pair recognizing the Illumina sequencing adapters. RT-qPCR data were converted to nanograms of DNA using standard curves (Materials and Methods). The estimated amounts of E. coli genomic DNA present in each library varied somewhat between the six locus-specific RT-qPCR reactions, so the bar graph indicates the mean values, and the error bars indicate the standard error of the mean.

Figure 3. Coverage depths from sequencing runs using E. coli strain DH10B libraries prepared on the AMCC chip.

(a) Ion Torrent libraries were run on the Ion Torrent PGM using the 100 bp sequencing protocol. (b) Illumina libraries were run on the MiSeq using the 2x25 bp paired-end sequencing protocol. Libraries labeled "Control" were prepared off-chip using the standard benchtop protocols recommended by each manufacturer. Sequencing runs with uniform coverage are expected to yield a Poisson distribution of coverage depths, indicated by the curves labeled "Theoretical limit".

Figure 4. Coverage depth from a sequencing run using an Illumina U-2 OS osteosarcoma cell line library prepared on the AMCC chip.

The library was run on the MiSeq using the 2x150 bp paired-end sequencing protocol. (a) Distribution of sequencing reads across the reference human genome, which has been divided into 1 million bins to assess coverage uniformity. (b) Average coverage across different chromosomes. (c) Average sequencing depth across different chromosomes. The Y chromosome is absent from the U-2 OS osteosarcoma cell line.

Figure 5. Combining NGS library preparation with size selection on the AMCC chip.

Columns were formed on the AMCC chip using carboxylated beads. (A) Electropherogram showing size selection on-chip, illustrating the recovery of DNA fragments in the range of 150-500 bp. The green trace represents the DNA ladder used. GEL image extracted from a high sensitivity chip run on the Bioanalyzer 2100. (B) Quantification of Illumina library DNA eluted from the AMCC chip using RT-qPCR. Coefficient of variation (CV) for this chip run was 0.05. Asterisks indicate sample modules where buffer was loaded instead of genomic DNA. (C) Coverage depths from sequencing runs of libraries prepared on the AMCC chip and using the conventional benchtop protocol recommended by the manufacturer.