Feature pubs.acs.org/ac
Teaching Single-Cell Digital Analysis Using Droplet-Based Microfluidics Microfluidics allows the manipulation of small quantities of reagents in a high-throughput manner and is therefore highly amenable to single cell characterization and more generally to digital analysis, with applications in fields as varied as genomics, diagnostics, directed evolution, and drug screening. The growing place of microfluidics in biology laboratories encouraged us to develop a teaching method where advanced undergraduate or first-year graduate-level students are taught to fabricate droplet-based microfluidic devices, characterize them, and finally use them to perform a digital analysis of bacterial samples based on a phenotypic marker. Majdi Najah,†,‡ Andrew D. Griffiths,*,† and Michael Ryckelynck*,† †
Institut de Science et d’Ingénierie Supramoléculaires (ISIS), Université de Strasbourg, CNRS UMR 7006, 8 allée Gaspard Monge, 67083 Strasbourg Cedex, France ‡ Ets J.Soufflet, Quai Sarrail, 10402 Nogent-sur-Seine, France S Supporting Information *
in a population is of great importance. For example, tumors are complex mixtures of cells with different genotypes and phenotypes and to properly understand the development and progression of cancers and to effectively treat them, it is essential to understand tumors at the single-cell level.3 It has also become clear that stochastic variations in gene expression create heterogeneities between cells even when they are genetically identical and in the same environment.4 These cell-to-cell variations play important roles in biological processes such as development or adaptation to the environment (e.g., resistance to antibiotics or chemotherapies) but are overlooked in analogue analyses. To overcome this limitation, the analysis has to be performed with a single-cell resolution. One option is to study single cells using fluorescent probes such as fluorescently labeled antibodies or nucleic acid probes.5,6 Gene expression at the single-cell can also be investigated by expressing genetically encoded fluorescent markers (e.g., Green Fluorescent Protein).6,7 High-throughput fluorescence profiling of individual cells can be performed using conventional flow cytometry8 but does not allow changes in individual cells to be followed over time. A more dynamic view can, however, be achieved using fluorescence microscopy.9 Fluorescence microscopy of cells immobilized in microfluidic devices, which can range from simple flow cells to more sophisticated microfabricated structures,10 opens up many exciting new possibilities for single-cell studies since the environment can be precisely controlled and modified on demand.11 Single cells can also be studied by directly measuring phenotypic markers such as enzymes using flow cytometry or microscopy using fluorogenic assays.12 However, this approach does not allow secreted enzymes to be monitored. Intracellular enzymes can be assayed, but only if the fluorogenic substrate can enter the cell and be transformed into a fluorescent product that stays confined in the cell.
Michael Ryckelynck
M
iniaturization of electronic components has enabled extraordinary advances in electronics and computing. During recent years, a similar phenomenon has begun to emerge in chemical and biological sciences with the development of miniaturized lab-on-chip devices based on microfluidic technology.1,2 Microfluidics allows the manipulation of small volumes of liquid (typically nL to pL) and can be defined as the study and application of fluid flow in microfabricated devices in which at least one length dimension is on the micrometer (10−6 m) scale, such new platforms being frequently reported in Analytical Chemistry. The one thousand to one million-fold reduction of reaction volumes made possible using microfluidics allows the throughput of traditional analytical methods to be increased, but it also represents a new and powerful tool to perform digital analysis at the single cell or single molecule level. Digital assays, in which many single cells or single molecules are assayed in parallel, have a number of advantages over more conventional analogue assays. For example, when analyzing cells in analogue format, a whole population of cells is characterized at once and the readout is an average value of the population. However, the ability to characterize individual cells © 2011 American Chemical Society
Published: December 15, 2011 1202
dx.doi.org/10.1021/ac202645m | Anal. Chem. 2012, 84, 1202−1209
Analytical Chemistry
Feature
An alternative strategy is to split up the assay volume into many small compartments, each of which contains, on average, less than one cell. The single cells can then be lysed to release their content into the compartment. The small volume of the compartments means that the phenotypic markers released are at high concentration and can easily be detected. This is also an exciting strategy for single cell genomics, epigenomics, transcriptomics, proteomics, and metabolomics13 since the low concentration of analytes released from a single cell in a microplate well currently limits the sensitivity of these “omics” techniques. Moreover, if the aim is to identify and quantify rare target cells in a population, while the sensitivity of analogue methods is limited by the dilution of the rare target cells in the whole assay volume, the digital approach has a much higher sensitivity since it is limited only by the number of individual compartments which can be analyzed, and quantification is simply a question of counting the number of compartments in which the target is detected. Even if the target cell is highly diluted and the vast majority of compartments contain no target, in those compartments containing a target, the concentration of the target remains high, making detection straightforward (Figure 1). Digital assays are not, however, limited to analyzing cells. Digital PCR, which is based on the compartmentalization and amplification of single DNA molecules, is a powerful and sensitive technique for the detection and quantification of specific genes.14 For all digital assays, there is a great advantage to use the smallest possible compartments. With smaller compartments, the sensitivity becomes higher since the concentration of a single target in each compartment is higher and more individual compartments can be analyzed without the cost becoming prohibitive. The small reaction volumes (nL to pL) achievable in microfluidic systems therefore makes them attractive tools for digital analysis. There are several strategies for compartmentalizing assays in microreactors in microfluidic devices. A first compartmentalization strategy is based on interconnected microreactors directly fabricated into the microfluidic device. Once the device is loaded, each compartment can be sealed either by slipping a cover plate15 or by actuating pneumatic valves.16 Such large scale integration (LSI) microfluidic devices, which can contain as many as 1176 independent microreactors, have proven to be extremely powerful tools to perform digital analysis of uncultivable micro-organisms17,18 or DNA molecules.19,20 An alternative microfluidic compartmentalization strategy is based on confining the objects to analyze (molecules or cells) in aqueous microdroplets (or plugs) of a few nano/picoliters in an immiscible carrier oil. In this format, digital analysis can be performed in >106 compartments (microdroplets) whose volume can easily be adjusted depending on the application. In addition, if required, new reagents can be added to preformed droplets by droplet fusion21 or pico-injection,22 droplet fluorescence can be measured using fluorescence microscopy23 or using systems based on lasers and photomultiplier tubes,24 and droplets can be sorted, triggered on fluorescence.24,25 For a recent review of droplet-based microfluidics see ref 26. Droplets can also be produced and manipulated using electrowetting on dielectric (EWOD) technology.27 However, despite the fact that EWOD is frequently referred to as “Digital Microfluidics”, this technique is not dedicated to digital analyses. The sensitivity of digital PCR in droplet-based microfluidic systems is clearly illustrated by a recent study in which a
Figure 1. Analysis of bacteria in analogue and digital formats. Cells from an overnight culture of bacteria with two different phenotypic markers are collected and washed prior to analysis. In the example described herein, the bacteria are either Gus+/Gal− (white lozenges) or Gus+/Gal+ (blue lozenges). (Left panel) Analogue analysis. The washed cells are pooled in a single compartment (e.g., a tube), and a lysis agent and fluorogenic substrates are added. After cell lysis, enzymes are released and the fluorogenic substrates are converted into fluorescent products: Gus+/Gal− bacteria give rise to a red fluorescent product, and Gus+/Gal+ give rise to both red and green fluorescent products. The major phenotype (Gus+/Gal−) can be observed and quantified, while the minor phenotype (Gus+/Gal+) becomes impossible to detect when the Gus+/Gal+ bacteria are too diluted. (Right panel) Digital analysis. The washed cells are compartmentalized in microcompartments (microdroplets in the example described herein) together with a lysis agent and fluorogenic substrates. Most of the compartments (∼70%) are empty, whereas the majority of the rest are occupied by a single bacterium (either Gus+/Gal− or Gus +/Gal+). The small volume of the compartments ensures that the enzymes released from a single lysed cell are maintained at high concentration, making detection of the phenotype of individual bacteria rapid and simple. Compartmentalization ensures that bacteria with minor phenotype (Gus+/Gal+) can easily be detected, and sensitivity is only limited by the number of compartments that can be analyzed. The quantification of the two phenotypes is simply a question of counting the nonfluorescent droplets (containing no bacterium), red-fluorescent droplets (containing a Gus+/Gal− bacterium), and red plus green (yellow) fluorescent droplets (containing a Gus+/Gal+ bacterium).
mutated KRAS oncogene was detected in the presence of a 200 000-fold excess of unmutated KRAS genes.28 This is at least 3 orders of magnitude more sensitive than equivalent PCR-based analogue assays. However, droplet-based microfluidic systems can also be used to perform digital assays in which individual cells are 1203
dx.doi.org/10.1021/ac202645m | Anal. Chem. 2012, 84, 1202−1209
Analytical Chemistry
Feature
compartmentalized in droplets. Droplet-based microfluidics has already allowed, for example, the quantitative detection of protein expression in single bacterial cells29 and the directed evolution of enzymes displayed on the surface of yeast with a 1000-fold increase in speed and a 1-million-fold reduction in cost compared to state-of-the-art robotic screening systems.30 Compartmentalization of single bacteria into 10 nL plugs has also been used to grow and to recover rare slow-growing bacteria.31 Finally, droplet-based microfluidics has even enabled drug screening at the single cell level.32,33 Given the growing interest of microfluidic-based methods in biological fields, especially for digital characterization, it becomes important to introduce their teaching to undergraduate and graduate-level students. Therefore, we devised a simple experimental method to teach digital analysis using droplet-based microfluidics. The students are taught, using a simple, safe, and inexpensive biological system, how to use a droplet-based microfluidic system to perform a digital assay to rapidly quantify specific strains of the bacteria using a phenotypic marker.
In addition, theoretical exercises were used to familiarize students with basic calculations required for the digital analyses. Digital analysis is based on the idea of performing a limiting dilution of the target cells (or molecules): the diluted target is distributed among many compartments such that each compartment contains, on average, less than one target. If the diluted targets are distributed randomly, the distribution of targets in the compartments is described by the Poisson distribution (eq 1) where P(X = k) is the probability to have k cells (or molecules) per compartment and λ is the mean number of cells (or molecules) per compartment.
e−λ k λ (1) k! Solving eq 1 shows that with λ = 1, 37% of the compartment would be empty and only 37% would contain exactly 1 cell, while the remaining 26% would contain 2 or more cells (Figure S1, Supporting Information). Digital analysis is therefore typically performed at λ values of 0.1−0.337 as this maximizes the number of compartments occupied (9.5−25.9%) while minimizing the multiple encapsulation events (
