2. Materials and Methods

2.1 Fabrication of microfluidic devices

The T-junction microfluidic device used in this study consists of two inlets for infusing the disperse (aqueous) phase and continuous (oil) phase, respectively, one outlet for transporting microdroplets into the collection tube and one rectangular chamber for observation of cell-laden microdroplets. A corona-shaped filter was designed at the inlets to prevent any possible dust entry. The connecting microchannels have an aspect ratio of height/width = 4:5 (height: ~40 µm; width: ~50 µm), and the rectangular observation chamber dimensions of 1.4 × 0.65 cm (Fig. S1 & Fig. S2).
This droplet-based microfluidic device was fabricated from a silicon wafer patterned with SU-8 mould (SU-8 2035, MicroChem, Newton, MA, USA) using standard soft-lithography techniques (Xia and Whitesides, 1998). Degassed poly(dimethyl siloxane) (PDMS, Sylgard 184, Dow Corning, Midland, MI, USA) in liquid form prepared by mixing the base and curing agent at a ratio of 10:1 was poured onto the SU-8 mould and cured in an oven at 80 °C for two hours. Then the PDMS slab with microchannels was peeled off from the mould and the fluidic access holes were created using PDMS biopsy puncher with an outer diameter of 1.5 mm. After cleaning the channel side of PDMS with scotch tape, isopropanol and DI water in order, the PDMS slab and a standard glass slide was treated using an oxygen plasma cleaner to increase the surface energy and immediately pressured to each other for an irreversible bonding. Lastly, the channel was rendered hydrophobic by infusing with 0.02 % Trichloro(octadecyl)silane (OTS, 104817, Sigma-Aldrich, St. Louis, MO, USA) in isopropanol for 5 mins and drying with nitrogen gas followed by drying in an oven at 100 °C for 10 mins.

2.2 Cell preparation

We used a total of four yeast strains, the haploid S. cerevisiaestrain (CEN.PK113-7D), the PA evolved mutant S. cerevisiae strain (PA-3), GFP-tagged S. cerevisiae strain (CEN.PK2-1C-GFP) and GFP-tagged P. pastoris strain (CBS7435-GFP) in the single-cell growth assays (Table S1).
CEN.PK113-7D and PA-3 were grown overnight at 30 °C, 200 rpm in 5 mL buffered minimal medium (Xu et al., 2019). The overnight culture was washed twice and re-inoculated into 5 mL buffered minimal medium at three different concentrations of PA: 15 mM, 25 mM and 35 mM. To investigate the effect of K-ions on the tolerance to PA in yeast, CEN.PK2-1C-GFP was pre-cultured overnight at 30 °C, 200 rpm in synthetic drop-out medium, without uracil (contains 1× yeast nitrogen base (YNB) without amino acids mix (Y0626, Sigma-Aldrich, St.Louis, MO, USA), 1% glucose and Yeast Synthetic Drop-out Medium Supplements without uracil (Y1501, Sigma-Aldrich, St.Louis, MO, USA)). Cells were washed twice and re-inoculated into 5 mL uracil drop-out medium containing defined concentrations of potassium (1× translucent K+ free YNB, 1% glucose, and 0, 1, 10 and 50 mM potassium chloride), supplemented with or without 25 mM PA. CBS7435-GFP was pre-cultured overnight in the same drop-out medium as the one used for CEN.PK2-1C-GFP at 25 °C, 200 rpm, and the culture was washed twice and re-inoculated into 5 mL uracil drop-out medium. All the cultures were reinoculated at an initial OD600 of 0.2 before being encapsulated into microdroplets.

2.3 Generation and storage of microdroplets

In the T-shaped droplet generator of 50 × 50 µm (width × depth), continuous phase flowed to the observation chamber and the disperse phase flowed and sheared at the interface to generate monodisperse microdroplets with a diameter of ~65 µm. Continuous phase used here was Novec™ 7500 Engineered Fluid (3M, St. Paul, MN, USA) containing 2% Pico-Surf™ 1 (Sphere Fluidics, Cambridge, UK). Disperse phase used here was culture medium added with 20% OptiPrepTM (D1556, Sigma-Aldrich, St. Louis, MO, USA). Cells were diluted to an OD600 of 0.1 for the disperse phase. Two syringe pumps (Fusion 100, Chemyx, Stafford, TX, USA) were used to inject the two phases, respectively. When the ratio of the flow rates of the two phases reached 4:1 (continuous phase: 16 µL/min vs disperse phase: 4 µL/min), ~ 144 pL monodisperse microdroplets were created in a high-throughput fashion (~116 droplets per second). By Poisson distribution, this size of droplet can ensure substantial droplets containing single cells (28.0%) and maintain relatively low ratio of droplets containing two cells (6.0%) and less than 1.0% droplets containing more than two cells, and the rest 65.0% droplets are empty. A fluorinated ethylene propylene (FEP) tubing (IDEX, Lake Forest, IL, USA) with an inner diameter of 0.5 mm was used to transfer microdroplets into a 2 mL EppendorfTM safe-lock tube (Hamburg, Germany) pre-filled with 100 µl continuous (Fig. S3).

2.4 Image acquisition and analysis

An initial observation of microdroplets containing single cells was performed at a rectangular chamber of the microdevice before microdroplets were transported into the collection tube. Bright-field snapshots of the generated microdroplets were captured by a digital camera (DS-Qi1Mc, Nikon, Tokyo, Japan) installed on an inverted microscope (Eclipse Ti-U, Nikon, Tokyo, Japan). After images were taken, the number of cells per droplet was counted using ImageJ® (National Institutes of Health (NIH), Bethesda, MD, USA).
To quantify growth of single yeast cells, the values for the specific growth rate μ were determined based on the number of cells per droplet at hourly time points up to 10 hours. The number of cells per droplet (N) was converted into the logarithmic scale as ln(N), and the estimation of the biokinetic constant, μ, over time of culture, t, was obtained by the equation below:
\begin{equation} \begin{matrix}\mu=\frac{\text{dln}\left(N\right)}{\text{dt}}\#\left(1\right)\\ \end{matrix}\nonumber \\ \end{equation}
Fluorescence images of GFP-tagged cells were obtained at fluorescein isothiocyanate (FITC) channel by a confocal microscope FV3000 (Olympus, Tokyo, Japan). The number of yeast cells in a complex cluster were more easily counted from the fluorescence images compared to bright-field images. Regarding uncountable agglomeration, images of eight slices were stacked up and the corresponding fluorescence intensity was measured by ImageJ®. The image processing includes the following four main steps: 1) the bit depth of images was reduced to 8 bits; 2) area of each droplet was recognized; 3) The threshold was set by Yen’s algorithm to remove noise; and 4) “limit to threshold” was chosen and the mean fluorescence intensity within the area of each droplet was measured.

2.5 Cell viability test

To measure the viability of yeast cells in microdroplets, live/dead staining tests were performed at three selected time points: before encapsulation, at 24 hours of culture and 48 hours of culture. At each time point, 50 µl emulsion microdroplets in oil were collected into a centrifuge tube, and 2 μL Pico-Break™ (Sphere Fluidics, Cambridge, UK) was added subsequently to release yeast cells from the microdroplets. After a short centrifugation at 2000 rpm for 30 s, the oil phase was kept at the bottom and the supernatant was transferred into a new centrifuge tube. 30 µl staining solution consisting of 2 µM SYTO 9 (Thermo Fisher Scientific, Waltham, MA, USA) and 4 µM EthD-III (Biotium, Hayward, Ca, USA) was added into the suspension of released yeast and co-cultured for 20 mins. Fluorescence images were taken when excited at a wavelength of ~495 nm and ~530 nm, respectively: live cells showed green fluorescence in 515 nm channel while dead cells showed red fluorescence in ~635 nm channel and yellow colour in merged images. The cell viability over time was tested based on three random frames for each measurement, and a total of 100 cells were tested for each measurement. The viability test results were analysed by one-way repeated measures analysis of variance (ANOVA) to determine whether a significant difference was existed among different time points.