Introduction

Quantified measurement of gene transcription is critical to elucidate the mechanisms of cell survival, growth, and differentiation in homeostasis and disease. Oligonucleotide (1,2) and cDNA (3) hybridization microarrays have emerged as the leading analytical tools for de novo discovery of gene expression patterns, due to their ability to record the transcribed messages from many thousands to tens of thousands of genes in a sample simultaneously (4). Typically, only the activity of a subset of genes is of interest in answering a specific biological question; this hypothesis is validated by screening the selected gene set against a larger sample population representing greater biological diversity, or set of test conditions. This secondary screen is performed to overcome inherent microarray deficiencies (5,6,7,8,9) and is implemented as a solution phase, real-time reverse transcriptase quantitative polymerase chain reaction (RT-qPCR) assay in 96- or 384-well microplates. Increases in wait time, labor, and reagent consumption, as well as a more-complex workflow, are fundamental limitations to a wider adoption of RT-qPCR as a primary or secondary screening tool for large sample sizes.

There is, therefore, a need for systems capable of implementing high-throughput RT-qPCR with a simplified workflow and reduced reagent consumption to make scale-up possible for screening the activity of tens to hundreds of genes simultaneously in hundreds to thousands of biological samples. Integral to the development and application of these systems are the methodologies for performance assessment based on standard system metrics of measurement reproducibility, accuracy, precision, sensitivity, and specificity (10,11). Since multiple values are measured simultaneously, it is important to characterize the capabilities of the system for quantitative measurement of gene expression based on such factors as signal magnitude and variability.

Miniaturized systems for high-throughput quantitative PCR

Quantitative or real-time PCR (12) is a derivative of the conventional PCR process (13) wherein the fluorescent signal is recorded at the primer annealing temperature in each temperature cycle. This signal is proportional to the number of DNA or cDNA template copies in the sample and is parameterized by defining the cycle threshold (Ct) as the temperature cycle at which the fluorescent signal is at least three standard deviations above the mean background fluorescence. Calibration by way of a standard curve, which relates Ct to number of transcript copies for a specific gene, makes determining its copy number in a sample possible. RNA quantification requires reverse transcription of RNA into cDNA prior to application of the real-time PCR method.

Strategies for increasing the analytical throughput of RT-qPCR involve either automating the manual workflow developed for microplates, multiplexing several RT-qPCR measurements in a single reaction, or decreasing the PCR assay volume to the microliter- or nanoliter-scale and increasing throughput by either fast serial or highly parallel processing of each miniaturized reaction. Automation of the RT-qPCR process increases the number of genes and samples that can be analyzed and improves the uniformity of data quality compared with a manual workflow, but these gains are offset by increased sample and reagent consumption, capital equipment costs, and the need for specialized labor to run and maintain robotic equipment (14,15). Multiplexed RT-qPCR analyses allow the number of transcript copies for multiple genes to be measured simultaneously in a single sample (16). However, it is a significant challenge to design primer pairs with a high and reproducible specificity to the target sequence and a measurable and reproducible amplification efficiency for each transcript targeted (17).

Miniaturization of PCR volumes without sacrificing data quality allows the number of analyses to increase without consuming more reagents and enables implementation of serial or parallel processing strategies to achieve high throughput. Most reports of microliter, nanoliter or picoliter PCR are for sequence-specific detection and not measurement of the number of expressed gene copies. Continuous-flow PCR devices utilize etched microchannels with fixed temperature zones to reduce reaction volumes to sub-microliter levels and analyze PCR products by hybridization followed by electrochemical detection (18), fluorescence detection (19), or electrophoretic separation with fluorescent detection (20). However, a key drawback in continuous-flow systems is the limited sample throughput and potential for cross-contamination that can result from processing and analyzing samples through a common microchannel. Devices for parallel PCR processing involve thermal cycling PCR reagents and sample template in a high-density array (>1 well/mm²) of micro-, nano- or even picoliter wells. Quantification of transcript copy number for a specific gene in these devices uses endpoint fluorescence of a probe compared with a standard curve (21,22) or digital PCR (23) based on template dilution and counting the number of positive assays for a specific target template (24). The distribution of template molecules at low concentration follows Poisson statistics; therefore, high-density arrays of wells are well-suited for high dynamic range measurement down to single-template copies from partitioning of the dilute sample into many thousands of individual picoliter containers (25,26,27). Despite its utility, the dynamic range of digital PCR is limited by the number of wells in the array and the accuracy of sample dilution.