Technical Note Cite This: Anal. Chem. XXXX, XXX, XXX−XXX
pubs.acs.org/ac
High-Throughput Generation of Durable Droplet Arrays for SingleCell Encapsulation, Culture, and Monitoring Han Wu,†,‡,# Xinlian Chen,†,# Xinghua Gao,† Mengying Zhang,‡,† Jinbo Wu,*,† and Weijia Wen†,§ †
Materials Genome Institute, Shanghai University, Shanghai 200444, China College of Science, Shanghai University, Shanghai 200444, China § Department of Physics, The Hong Kong University of Science and Technology, Hong Kong, China ‡
S Supporting Information *
ABSTRACT: High-throughput measurements can be achieved using droplet-based assays. In this study, we exploited the principles of wetting behavior and capillarity to guide liquids sliding along a solid surface with hybrid wettability. Oil-covered droplet arrays with uniformly sized and regularly shaped picoliter droplets were successfully generated on hydrophilic-in-hydrophobic patterned substrates. More than ten thousand 31-pL droplets were generated in 5 s without any sophisticated instruments. Covering the droplet arrays with oil during generation not only isolated the droplets from each other but also effectively prevented droplet evaporation. The oil-covered droplet arrays could be stored for more than 2 days with less than 35% volume loss. Single microspheres, microbial cells, or mammalian cells were successfully captured in the droplets. We demonstrate that Escherichia coli could be encapsulated at a certain number (1−4) and cultured for 3 days in droplets. Cell population and morphology were dynamically tracked within individual droplets. Our droplet array generation method enables highthroughput processing and is facile, efficient, and low-cost; in addition, the prepared droplet arrays have enormous potential for applications in chemical and biological assays.
D
the droplets is also controlled, and the volume of the prepared microdroplets can even be minimized to femtoliters. A freestanding picoliter- or femtoliter-volume aqueous droplet would evaporate within a few milliseconds in air.16 Therefore, droplet arrays should be covered with oil as soon as they are generated, to prevent evaporation for long-term analysis or hightemperature operation.17 Iino and Noji et al.11,18 exchanged the aqueous solution on the hydrophobic surface with Fluorinert FC40 oil, which has a higher density than water, while retaining the aqueous solution on the hydrophilic surface; more than 105 oil-covered droplets were formed simultaneously. However, large-scale droplet arrays require additional reagent consumption since the hybrid hydrophobic surface must be covered with a large drop of aqueous solution first. Moreover, the liquid exchange on a hybrid surface induced by density differences can easily get out of hand, and the array always becomes nonuniform, since the oil spreads out and the exchange velocity varies from place to place. In this work, we generated oil-covered uniform droplet arrays in a controlled manner with low reagent consumption. We filled the aqueous solution in a gap between the substrate and a cover glass, loaded the oil, and then guided the oil−water exchange by sliding the cover glass. The patterned substrates were prepared using selective plasma etching with a reusable
roplet arrays are sessile droplets that are confined within affinitive patches. Such arrays are emerging as an open platform that enables a paradigm shift from the conventional, flow-based, lab-on-a-chip philosophy and competes with microplates, in terms of versatility and simplicity of operation.1 These platforms are usually just required to prepare hydrophilic patterns on hydrophobic surfaces2−4 and rely less heavily on sophisticated techniques, yet retain similar merit: low energy and sample consumption, high throughput, and automation.5−7 The open microdroplet arrays have exhibited strong functionality in biological and chemical fields.8−10 For instance, the platforms have been employed in single-molecule enzymatic assays,11 membrane transport analyses,12 nucleation reactions,13 and high-throughput cell screening.14 Song et al.15 presented a facile strategy for microdroplet array preparation. By sliding a 10 μL droplet on a patterned superhydrophilic/superhydrophobic substrate, homogeneous microdroplets were successfully prepared. Superhydrophilic square patterns with different lateral lengths of 5−100 μm were prepared on the superhydrophobic substrates, and the volume of the droplets generated increased from 9.7 fL to 83.5 pL. By employing this droplet splitting strategy, fluorescent polystyrene (PS) spheres and human breast cancer cells were quantitatively separated at designated locations. Such a droplet array offers a unique analytical technology for revealing the life of single cell beyond the traditional bulk culture. By controlling the characteristics of hydrophilic patterns on hydrophobic substrates, the distribution, volume and shape of © XXXX American Chemical Society
Received: January 3, 2018 Accepted: March 18, 2018
A
DOI: 10.1021/acs.analchem.8b00048 Anal. Chem. XXXX, XXX, XXX−XXX
Technical Note
Analytical Chemistry polydimethylsiloxane (PDMS) stencil.19 The sizes of the hydrophilic patterns and the volumes of the generated droplets could be easily controlled by using different PDMS stencils. We characterized and analyzed the three-dimensional (3D) droplet morphology and evaporation mechanism using confocal microscopy. More than 104 pL droplets were generated in 5 s, and they were found to have more than 80% coverage and less than 10% variation in volume. The presence of the oil layer effectively prevented evaporation, and 65% of the volume could be preserved after 60 h of storage. Finally, PS spheres and cells were successfully captured in the droplets; meanwhile, we monitored the division and dynamic changes in morphology of single Escherichia coli (E. coli) in situ and observed the filamentous growth of E. coli in such a confined microenvironment.
patterned area. The cover glass was pressed on the solution to uniformly spread out the solution on the patterned area, and then, Fluorinert FC40 was added near the patterned area. Afterward, the cover glass was slid toward the edge of the patterned area, and a droplet array covered with oil was formed as a result. To prevent the oil from flowing away, a PDMS barrier was utilized. The fluorescence images of the droplet arrays were obtained using an inverted microscope (Model IX73, Olympus), and the droplet diameters were measured using computer software (cellSens, Olympus). The wavelength of the fluorescent light used was 488 nm. Sectional views of the droplets were obtained using confocal laser scanning microscopy (Model FV3000, Olympus). Then, 3D images of the droplets were constructed, and the heights of the droplets were measured using computer software (FV31S-SW, Olympus) that was applied to analyze image slices taken along the Zaxis with intervals of 1.0 μm. A uniform polystyrene microsphere (PS spheres, 5 μm in diameter) solution (2.5% w/v) was obtained from Aladdin Industrial Corporation (Shanghai, China). The PS-sphere solution was diluted by factors of 50, 100, 200, 400, and 800, using fluorescein sodium solution. The diluted PS-sphere solutions were used to prepare the oil-covered droplet arrays, and the PS spheres were captured by the forming droplets. Fluorescent images and phase contrast images of the PS spheres capture by the droplets were obtained using an inverted microscope (Model IX73, Olympus). The PS spheres in the droplets were counted manually. A549 human nonsmall cell lung cancer cells (Stem Cell Bank, Chinese Academy of Sciences) were cultured in a cell culture medium containing 90% F-12K (Invitrogen) and 10% fetal bovine serum (FBS, Gibco) in a CO2 incubator at 37 °C with a 5% CO2 level. After digestion, in order to stain the cells, 1 μL of Green 5-chloromethylfluorescein diacetate (Green CMFDA, Molecular Probes) solution (1 mmol/L) was added to 1 mL of the cell suspension without FBS and then cultured for 30 min at 37 °C. After cell staining, the cell suspension was centrifuged for 3 min at 1000 rpm. After removing the supernatant, a cell suspension with a concentration of 5 × 105/ mL was obtained. The cell suspension was diluted 10 times with F-12K and employed for droplet array preparation. Similarly, the cells were captured by the droplets. Fluorescent images and phase contrast images of the cells captured by the droplets were obtained using the inverted microscope (Model IX73, Olympus) with polarizing components. Surface of the clean glass was spin-coated with prepolymer solution of PDMS at a speed of 3000 rpm and then the formed substrate was placed into an air oven at a temperature of 80 °C for ∼2 days to completely remove residual reagents in PDMS. Such a substrate provided an artificial hydrophobic surface without the self-assembled monolayers(SAMs) of POTS. This substrate was combined with the LPS and treated with plasma for 1 min to form hydrophilic spots on the PDMS substrate. E. coli were cultured in standard Luria−Bertani broth (LB) media at 37 °C. The LB medium was composed with LB-medium (powder) (Solarbio-Life Sciences) and ultrapure water, composed of yeast, tryptone, and sodium chloride, using a ratio of 5 g, 10 g, and 10 g, respectively, per liter of medium. Droplets containing E. coli cells were formed using the method mentioned in this article (Figure 1). By regulating the volume of bulk cell medium, the number of cells was controlled within a certain range, according with Poisson law. After the droplets were formed with an oil covering (M8410, Sigma−Aldrich), the
■
EXPERIMENTAL SECTION Silicon wafers with micropillars were fabricated by Rdmicro company (Suzhou, China). 1H,1H,2H,2H-Perfluorooctyltriethoxysilane (POTS) was obtained from Sigma−Aldrich (USA). The silicon wafers were cleaned using a plasma cleaner (PDC002, Harrick Plasma) and then modified with POTS by baking the wafers for 1 h at 120 °C in a sealed Crisper containing an open vial of POTS. PDMS was produced using a Sylgard 184 silicone elastomer mixture (Dow Corning Corporation, Midland, MI, USA) at a base:curing agent weight ratio of 10:1. The bubbles produced during mixing were eliminated under vacuum. After vacuum treatment, the PDMS was poured onto the micropillar array domain of the silicon wafer and scraped to homogeneously cover the micropillars. Afterward, the silicon wafer with PDMS was cured on a hot plate (Model C-MAG HP7, IKA) at a constant temperature of 100 °C for 10 min. The cured PDMS, which is referenced as the PDMS stencil (2.5 cm × 2.5 cm), could be easily peeled off from the silicon wafer, and the thickness of the PDMS stencil was ∼120 μm, as measured using a microscope (Model LV100ND, Nikon). In addition, the diameters of the round through holes in the PDMS stencil were measured using the microscope. Glass substrates (7.5 cm × 7.5 cm) were cleaned using plasma treatment and treated to be made hydrophobic by baking them with POTS for 2 h at 120 °C. Then, the PDMS stencil was placed tightly on the hydrophobic glass surface, and the combination was treated with plasma for 10 s. In this way, the hydrophobic glass side was selectively etched. The exposed part of the glass was etched and became hydrophilic, while the covered part remained hydrophobic. After removing the PDMS stencil from the glass substrate, the hydrophilic-in-hydrophobic micropatterned surface was successful achieved. The contact angles of water and oil (Fluorinert FC40, 3 M) on the hydrophilic and hydrophobic surfaces were measured using a contact-angle measuring instrument (Model XG-CAMC3, XYCXIE) at ambient temperature. The glass substrates could be recycled by cleaning them with a mixed solution of saturated sodium hydroxide and ethanol. The volume ratio of the sodium hydroxide solution to ethanol was 1:10. Fluorescein sodium salt was purchased from Aladdin Industrial Corporation (Shanghai, China), and a fluorescein sodium solution of 0.1 mmol/L was prepared for the following experiments. Cover glass and a gasket glass were needed to prepare the oil-covered droplet array, and their thicknesses were 1 and 0.15 mm, respectively. Two pieces of gasket glass were adhered to the sides of the patterned area of the substrate, and 85 μL of a fluorescein sodium solution was added to the B
DOI: 10.1021/acs.analchem.8b00048 Anal. Chem. XXXX, XXX, XXX−XXX
Technical Note
Analytical Chemistry
thick formed droplet arrays with different sizes are shown in Figures S3(a), S3(b), and S3(g) in the Supporting Information. Fluorescent images and 3D images of the oil-covered droplet arrays are shown in Figure 2, as well as Figure S3. Using
Figure 1. Schematic drawing of the oil-covered droplet array preparation procedure.
assembly was placed onto the objective table of the microscope, at an environmental temperature of 22 °C. Images of cells in the droplets were taken at intervals of a certain number of hours every time, according to the growth rate of cells.
■
RESULTS AND DISCUSSION The cleaned glass substrate was rendered hydrophobic and combined with the PDMS stencil. The combination was treated with plasma, and selective etching on the hydrophobic glass substrate was realized using the optimized parameters. After plasma treatment, neatly arranged and regularly shaped hydrophilic circular-patterned arrays were prepared on the hydrophobic glass substrate. The size of the hydrophilic circular patterns was closely related to the PDMS stencil dimensions and plasma treatment time. The size of the pattern directly determined the sizes of the droplets, which will be described in the next section. We made three different PDMS stencils: SPS (small-hole-sized PDMS stencil, hole diameter = 78.7 ± 1.9 μm), MPS (medium-hole-sized PDMS stencil, hole diameter = 108.7 ± 3.2 μm), and LPS (large-hole-sized PDMS stencil, hole diameter = 139.0 ± 1.9 μm). Therefore, the volume of the droplets was precisely controlled by the dimensions of the PDMS stencil. In addition, the PDMS stencil can be used multiple times, greatly simplifying the experimental procedure. Figure 1 shows a schematic drawing of the oil-covered droplet array preparation procedure. In our method, the droplet array is formed in a facile and rapid manner. Meanwhile, the volume of the solution consumed during the preparation of the droplet array is only 85 μL; such a small volume is significant when expensive reagents are required. The oil-covered droplet array preparation procedure involves four steps, as shown in Figure 1. The cover glass that we employed is hydrophilic, and the thickness of the gasket glass is 150 μm. Without the gasket glass, the volume of the solution consumed during the droplet array preparation process is ∼2 mL. The aqueous solution is forced to fill the space between the cover glass and the patterned area of the substrate when the cover glass is pressed onto the solution. In addition, the oil (Fluorinert FC40) added to the substrate has a higher density than water. When sliding the cover glass in the direction indicated by the yellow arrow, as shown in the third step in Figure 1, the oil/water interface also moves. The hydrophilic patterned surface pins the aqueous solution, while the solution on the hydrophobic surface is easily replaced by the oil. To maintain the oil on the substrate, 1-cm-
Figure 2. (a) Sectional views of the droplets formed on circular patterns prepared by LPS. Images were taken by a fluorescence confocal microscopy. (b) Three-dimensional (3D) images of droplets that are 145.1 μm in diameter. (c) The diameter, height, and volume of the droplet varies with storage time.
different PDMS stencils, we prepared droplet arrays with different sizes and numbers of droplets. Highly ordered, uniformly sized and regularly shaped droplet arrays with different sizes are shown in Figures S3(a), S3(b), and S3(g). On the substrates, ∼15000, ∼11500, and ∼9500 droplets were formed using the SPS, MPS, and LPS size levels, respectively. The droplet array formation process is very fast and requires only 5 s. Therefore, the droplet preparation throughput using our method can reach up to 3 kHz, which is comparable to that of existing microfluidic technologies, and the number and density of droplets can be easily controlled using different PDMS stencils. The bottom diameters of 144 droplets were measured in each droplet array and the average diameters of the droplets prepared using the SPS, MPS, and LPS were 88.8 ± 7.7, 123.7 ± 5.5, and 145.1 ± 7.7 μm, respectively. Fairly monodisperse lateral diameter distributions of the droplets were obtained with an average diameter deviation of
