Current methods for screening libraries of compounds for biological activity are rather cumbersome, slow and imprecise. A method that breaks up a continuous flow of a compound's solution into droplets offers radical improvements.
Finding new leads for drug discovery increasingly depends on high-throughput methods that allow efficient biological screening of chemical libraries. Reporting in Proceedings of the National Academy of Sciences, Miller et al.1 describe a method for screening large numbers of compounds —thousands or more — for their activity as enzyme inhibitors. This protocol generates immense amounts of high-precision dose–response data in ultrashort times of just a few minutes per compound. The authors' approach could vastly improve the efficiency of current screening processes, which represent a major bottleneck for drug-discovery programmes.
Conventional routes to screening compounds typically begin with a qualitative primary screen of a diverse compound library to identify which compounds are active at a biological target. This is supplemented by a dose–response analysis of selected active compounds, to relate the target's activity to each compound's concentration. Unfortunately, dose–response screening of libraries is a time-consuming and complex procedure, with many potential pitfalls. Not least of these is the amount of time required to gather sufficient data to probe complex pharmacology, and the large number of false positives and false negatives generated from standard assays2.
Library screening to find enzyme inhibitors in particular can be fraught with difficulties because of the high number of compounds that must be screened and the rather cumbersome methods used — typically involving robots running assays in tiny wells on plates. In practice, every compound in a primary screen is normally assayed at a single concentration; the vast majority of candidates are almost inevitably inactive and are thus eliminated. The first screen is then followed by a more complex, dose-dependent assay of the remaining candidates. But although secondary assays are more detailed, they are frequently run at only a limited number of compound concentrations, because they face similar constraints to the primary screen3.
Assays based on microfluidics — the precise control of fluids constrained within submillimetre-scale channels — could transform high-throughput screening. Microfluidic technology is being increasingly used in chemistry and biology because of its ability to rapidly perform complex analytical procedures on minute sample volumes, with greater efficiency than traditional macroscale approaches4. Such systems normally operate either by passing a continuous flow of a solution through channels, or by breaking a flow into segments or droplets.
Continuous-flow systems offer some advantages over those involving segmented flow — for example, they allow chemical gradients to be established in a flow. But they are also hampered by sample dispersion, which spreads a compound sample out from an initial, focused volume to a wider, more dilute region. Continuous-flow systems also suffer from residence-time distributions, in which some parts of a sample move faster than others, thus adding to sample spreading and dilution. Segmented (or droplet-based) flows, on the other hand, offer excellent control over sample dispersion and residence time, although gradients are more difficult to establish.
Miller et al.1 have exploited key features of both continuous- and segmented-flow microfluidics to develop a robust system capable of generating high-quality dose–response data in screens for enzyme inhibitors. At the heart of their approach is the generation of a concentration gradient over time in a flow of a solution of the compound using Taylor–Aris dispersion (Fig. 1a) — a phenomenon that occurs in continuous flows as a result of a combination of molecular diffusion and the variation of flow velocity over the cross-section of a microchannel. The authors segment this flow into a series of droplets dispersed in a continuous flow of oil (Fig. 1b), in a process similar to the analog-to-digital conversion of an electrical signal.
Each droplet has a volume of just 140 picolitres (1 picolitre is 10−12 litres), and defines a unique reactor containing a concentration of the candidate inhibitor that differs from the concentrations in droplets preceding or following it. No dispersion of the dissolved compound can occur between droplets, and so the concentration gradient established in the original continuous flow is preserved. By incorporating the target enzyme and its substrate within each droplet and measuring the enzyme inhibition by the candidate, the authors were able to produce dose–response curves containing 10,000 data points for a compound from a single sample of solution, within a few minutes.
The large amount of data extracted from each experiment means that the resulting dose–response curves can be analysed with high precision using complex kinetic models. This greatly reduces the occurrence of false negatives and false positives, and generates much more helpful data than are produced by currently available secondary assays. Given that the authors' microfluidic device can screen each compound so effectively within two to three minutes, their approach also opens the way to more thorough and effective primary screening.
As Miller et al. highlight1, their system still has some issues that must be resolved, such as ensuring that each droplet remains in the microfluidic system long enough for enzyme inhibition to occur. The authors currently achieve this by incorporating channels (known as delay lines) in which the flow is slowed. Although these lines allow reactions to be assayed over a period of 3.5 minutes, they also introduce residence-time distributions because of droplet disordering. This is clearly problematic, especially when screening slower enzymatic processes, but could be overcome by changing delay-line geometry and channel lengths. Another concern is that the range of inhibitor concentrations that can be accessed by the authors' system is rather limited. However, this issue could easily be resolved by incorporating passive droplet-dilution modules5 — devices that would dilute each sample-containing droplet into a stream of new, more-dilute droplets.
Despite these limitations, the impact of Miller and colleagues' approach on the drug-discovery process will be considerable. By enabling large amounts of dose–response information to be obtained from small sample volumes, it could facilitate the direct, quantitative dose–response screening of whole libraries in short times, eliminating the need for second-pass screening of active compounds.
Miller, O. J. et al. Proc. Natl Acad. Sci. USA 109, 378–383 (2012).
Brian, K. S. Drug Discov. Today 11, 607–615 (2006).
Inglese, J. et al. Proc. Natl Acad. Sci. USA 103, 11473–11478 (2006).
Hong, J., Edel, J. B. & deMello, A. J. Drug Discov. Today 14, 134–146 (2009).
Niu, X., Gielen, F., Edel, J. B. & deMello, A. J. Nature Chem. 3, 437–442 (2011).
About this article
Miniaturized technologies for high-throughput drug screening enzymatic assays and diagnostics – A review
TrAC Trends in Analytical Chemistry (2020)
Dynamic characterization of nanoparticles production in a droplet-based continuous flow microreactor
Chemical Engineering Research and Design (2019)
Lab on a Chip (2019)
International Journal of Heat and Mass Transfer (2019)