3. Results and Discussion

3.1 The encapsulation and cultivation of single yeast cells in picoliter microdroplets

Similar to the standard method used to measure the average growth rate of bulk populations, the procedure for the microfluidic droplet technique used in this study to measure the growth of single yeast cells includes three main steps: 1) preparation, 2) measurement and 3) analysis (Figs. 1A and 1B). 1) Uniformly distributed microdroplets containing single cells and a small volume of culture medium, subject to a range of different environmental conditions, are generated; 2) the number of cells per droplet and fluorescence intensity of GFP-tagged cells are tracked as cells grow over time; and 3) the growth rate of each cell within the microdroplet is obtained and single-cell growth under varying environmental conditions (with PA, AA and K-ions at different concentrations) is investigated and compared.
Before cell encapsulation within microdroplets, the overnight culture and preculture protocols were successively performed to ensure yeast cells were rapidly proliferating but this also results in an asynchronous culture, containing single cells, single cells with small buds, and cells with large buds. In practise, the budded cells were counted as single cells when the buds did not exceed the one-half the size of the mother cell. Cell density was diluted to an OD600 of 0.1 (3 × 106 cells/mL) with fresh medium and density-matching reagent to ensure the substantial majority (>82%) of cell-laden droplets contained just one cell. According to the Poisson distribution, this concentration of cells theoretically maintains high efficiency of single-cell encapsulation (Liu et al., 2020). It is worth noting that under these conditions the majority (65%) of droplets do not contain any cells. The density-matching reagent used here is 20% OptiPrepTM, which prevents the sedimentation of yeast cells at the inlet and guarantees the neutral buoyancy of cell suspensions for ~30 mins (Allazetta et al., 2015). The asynchronous nature of the cell population combined with the small statistical probability for more than one cell per droplet leads inevitably to some experimental uncertainties in measured cell growth notionally arising from a single cell origin.
The growth of single S. cerevisiae CEN.PK 113-7D cells in ~ 144 pL microdroplets without acid stress is illustrated by the images of Fig. 1C. After culture of 18 hrs, the droplets shrank from ~ 144 pL (empty droplets) to ~ 65 pL (yeast-containing droplets) driven by osmosis. The yeast cells kept consuming the glucose from the medium in droplets and induced water efflux to equalize the solute concentration inside and outside the droplets (Joensson et al., 2011; Siedler et al., 2017). No significant size differences were detected between the same types of droplets. Apart from this, the yeast-containing droplets cells did not experience any noticeable disruption, such as merging and burst, after long-term storage, thus the single yeast growth can be quantitatively and accurately tracked over 24 hours within this type of droplet.
3.2 The effect of acid stress on the growth of single wild-type S. cerevisiae cells in picoliter microdroplets
First, we tracked the growth of the single S. cerevisiae CEN.PK 113-7D cells in ~144 pL microdroplets without acid stress over 24 hours. In Fig. 2A, we plotted the number of cells per droplet at a logarithmic scale based on the first ten hours of culture, assumed during exponential phase. The value of µ for single S. cerevisiae cells in microdroplets was calculated to be 0.23 ± 0.03 h-1, identical to that reported for bulk populations (0.21 ± 0.01 h-1) within experimental uncertainty. Then we plotted the number of cells per droplet at eight selected time points: 0 hr, 2 hrs, 4 hrs, 6 hrs, 8 hrs, 10 hrs, 18 hrs and 24 hrs. (Fig. 2B). In general, the number of cells per droplet increases over time: single yeast cells (at 0 hrs) grow to 3.2 ± 1.4 at 2 hrs, 5.7 ± 2.0 at 4 hrs, 8.7 ± 2.2 at 6 hrs, 11.6 ± 2.7 at 8 hrs, 14.6 ± 3.1 at 10 hrs, 42.2 ± 7.0 at 18 hrs and 50.0 ± 8.0 at 24 hrs. The number of cells per droplet is seen to increase monotonically over time, but with an increasing spread of cell-counts for the later time (18 hrs and 24 hrs). Also, the heterogeneity in the proliferation of single yeast cells is demonstrated. At 24 hrs, a small portion (i.e., 5%) of microdroplets contain more than 60 cells, whereas another small proportion (i.e., 6.7%) of microdroplets has less than 40 cells. This is evidence of subpopulations exhibiting diverse traits that are obscured in bulk assays at the population level. It is noteworthy that after the culture of 18 hrs, the amount of yeast in droplets altered slightly compared to the previous growth, indicating that the yeast cells had reached the stationary phase due to the scarce nutrient and limited space. Droplet size can be tuned to enable different scales of cell culture and has been observed in other studies (Pan et al., 2011; Siedler et al., 2017).
Secondly, we investigated the effect of acid stress on cell growth by tracking and comparing the growth of single S. cerevisiae cells with the addition of PA (at 7.5 mM and 35 mM, Fig. 2C) and AA (at 50 mM and 67 mM, Fig. 2D). The study of yeast growth and responses to PA and AA is of great importance, because PA is a valuable organic acid produced by yeast during fermentation, and AA is a main growth inhibitor found in lignocellulose hydrolysate for lignocellulose-based biofuel production. Experimental results show that microdroplets enable cell growth in all conditions. In more detail, at 7.5 mM PA, single cells in microdroplets grow to 31.0 ± 5.5 and 40.0 ± 4.8, respectively, at 18 hrs and 24 hrs; while at 35 mM PA, single cells grow to 3.1 ± 1.4 at 18 hrs and 3.8 ± 1.4 at 24 hrs (Fig. 2C). At 50 mM AA, single cells in microdroplets grow to 13.9 ± 2.6 and 22.0 ± 5.84, respectively, at 18 hrs and 24 hrs; while at 67 mM AA, single cells grow to 8.6 ± 2.7 at 18 hrs and 12.4 ± 3.1 at 24 hrs (Fig. 2D). Moreover, we found that growth of single S. cerevisiae cells responds sensitively to both acids and decreases as the concentration of acids increases. At 24 hrs, the number of yeast cells per microdroplet under 7.5 mM PA (40.0 ± 4.8) and 35 mM PA (3.8 ± 1.4) declined, respectively, to 80.0% and 7.6% of that for no acid control (50.0 ± 8.0) (Fig. 2C). Additionally, at 24 hrs, the number of yeast cells per microdroplet decreased to 44.0% (22.0 ± 5.84) and 24.8% (12.4 ± 3.1) of cell number of the control group (no AA), respectively, when the concentration of AA increased to 50 mM and 67 mM (Fig. 2D). It is noteworthy that only the cell counts per droplet with the addition of 7.5 mM PA can be well fitted to sigmoid growth curve, indicating the small volume of PA, like 7.5 mM, did not affect yeast growth much but the large volume of that and AA will impose negative effect since the beginning of cell growth. These results indicate that the growth and physiology of single cells in microdroplet are the same as those of yeast populations grown in bulk, although microdroplet culture reveals subpopulation phenomena that are obscured by population average measurements.
3.3 The effect of K+on the growth of single GFP-tagged S. cerevisiae cells in picoliter microdroplets
Since biochemical assays are typically measured using fluorescence detection techniques, we investigated the growth of fluorescent GFP-tagged S. cerevisiae strain (CEN.PK2-1C) to demonstrate the capability of our platform for fluorescence-based quantification and detection of single-cell features. Both bright-field and fluorescence images show that the number of fluorescent cells per droplet increases over time (Figs. 3A and 3B). We counted the number of cells per droplet at five selected time points: 2 hrs, 6 hrs, 10 hrs, 18 hrs and 24 hrs (Fig. 3C). The data shows that growth from single cells has a high degree of variability: although the average cell number per droplet is 14.4 ± 3.3 at 24 hrs, a few microdroplets (i.e., 3.3%) contain more than 20 cells, whereas some microdroplets (i.e., 15%) contain less or equal to 10 cells. This is further evidence of cellular subpopulation with different growth growth rates which are obscured in bulk assays.
In S. cerevisiae , potassium uptake has been shown to stabilize membrane potential, and mediate intracellular pH, protein synthesis and function (Arino et al., 2010; Kahm et al., 2012; Yenush et al., 2002). In previous studies, potassium supplementation was also demonstrated to be beneficial to PA-tolerance behaviours of S. cerevisiae (Xu et al., 2019). We firstly tracked the growth of single CEN.PK2-1C cells in microdroplets under a fixed potassium defined condition (10 mM K+) with or without 25 mM PA (Fig. 3D). The results show that yeast growth in microdroplets was inhibited under PA stress condition when the medium contains 10 mM K+.
We then applied another two concentrations of K+, excessive supply of 50 mM and a scanty supply of 1 mM, when the concentration of PA is fixed at 25 mM (Fig. 3E). Compared to the µ under 10 mM K+ at 24 hrs, there is a 38.9% increase when 50 mM K+ was used, and no significant decrease when the concentration of K+ reduces to 1 mM. These results agree with the previous findings that extracellular supplementation of K+ can increase PA tolerance in yeast, and potassium influx is important to increase organic acid tolerance in S. cerevisiae By using the GFP-labelled strain and supplementing K+ under PA stress conditions, we have shown that single-cell culture in microdroplets demonstrate the same phenotype and the similar growth profiles as bulk cultures, although cell-to-cell variations in proliferation are observed. We conclude therefore that the microdroplet platform can reliably quantify the effects of external factors on cell growth and complex physiology under varying conditions.

3.4 The growth of wild-type and PA evolved mutant S. cerevisiae strains in picoliter microdroplets

ALE has previously been employed to improve PA tolerance in yeast, and PA-3 is one of the isolated strains with increased PA tolerance after performing ALE. The non-synonymous mutation in potassium transporter encoding gene TRK1 , has been confirmed to be the cause of the increased PA tolerance (Xu et al., 2019). To demonstrate that microdroplets could be used to track the growth of yeast mutant strains, we monitored and compared the growth of PA evolved mutant strain (PA-3) and its parental strain (CEN.PK 113-7D) when 15 mM PA was applied (Fig.4). The experimental data confirms that PA-3 grows faster and reaches a significantly higher average number of cells per droplet, i.e., 18.0 ± 3.0 at 24 hrs, whereas the average number of cells per droplet for wild-type strain is 5.2 ± 1.3 at 24 hrs (Fig. 4). This result demonstrates that the microdroplet reactor approach is effective for both normal and mutant strains of S. cerevisiae and holds their difference in cell growth and physiology at the population level when single cells are tracked in microdroplets.

3.5 The growth of single P. pastoris cells in picoliter microdroplets

In order to demonstrate that this platform can be applied to species other than S. cerevisiae , we tracked the growth of GFP-taggedP. pastoris strain (CBS7435-GFP) at a single-cell level in ~144 pL microdroplets. P. pastoris has a similar cell size to S. cerevisiae , but the proliferation behaviour is different. The CBS7435-GFP used here tended to aggregate in the center of the microdroplets due to the lack of motility. The bright-field and fluorescence images (stacks of eight slices) show that single P. pastoris cells are able to grow in microdroplets over time (Fig. 5A). The distribution of fluorescence intensity of cells within the microdroplets at five selected time points demonstrated the variations between individual cells (Fig. 5B). Although some outliers exist, the growth curve of P. pastoris shows a similar profile over 24 hours to that of S. cerevisiae under normal conditions. This indicates that the fluorescence measurement can quantitively indicate the growth of single cells in microdroplets and demonstrates that the microdroplet bioreactors used in this study can maintain and screen of growth of single yeast cells of different species.

3.6 Viability assays of S. cerevisiae and P. pastoris grown in microdroplets

We used the cell staining live/dead kit to investigate whether 24 hours or a prolonged period of culture will affect the viability of S. cerevisiae and P. pastoris . This is to ensure that encapsulation and cultivation of cells in microdroplets is a feasible and stable method for long-time single-cell assays.
The viability tests were performed and compared at three time points: before encapsulation, after 24 hrs and 48 hrs of encapsulation. The bright-field and fluorescence images show that both S. cerevisiaeand P. pastoris cells maintain a high level of viability after 24 hours of culture (Fig. 6A). For S. cerevisiae , 94.4 ± 1.3% cells remain alive after 24 hours of culture, and cell viability slightly decreases to 93.6% ± 1.7% after 48 hours of culture; while forP. pastoris , 97.8 ± 0.8% and 95.5% ± 1.1% cells remain alive after 24 hours and 48 hours of culture, respectively (Fig. 6B).
Considering that the oil-removing reagent, pico-breakTM, contains PFOH (1H,1H,2H,2H-Perfluoro-1-octanol) which is a potential chemical hazard for yeast cells, the measured viability of encapsulated cells may represent an underestimate of the true viability. Moreover, the result of one-way repeated measures ANOVA (P <0.005) shows that there is no significant difference in cell viability among that before encapsulation, that for 24 hours and 48 hours of culture. This demonstrates that the viability of yeasts cultured in microdroplets is not significantly affected and the method is capable of prolonged assays of live yeast cells.
In this study, we explored the feasibility of using microdroplets as bioreactors to screen cell-to-cell variations in growth. Cell encapsulation in microdroplets is a random process limited by the Poisson distribution but affected by cell sedimentation, leading to a majority of droplets that are empty. To maximize the proportion of single cell-encapsulated microdroplets without any noticeable damage, we used a non-ionic solution of 60% iodixanol, OptiPrep™, which has proved to be biocompatible, has low osmotic pressure and low intrinsic viscosity suitable for the culture of cells in microdroplets (Allazetta et al., 2015; Ma et al., 2017). Here, we used the addition of 20% OptiPrep™ to reduce the effect of cell sedimentation (the density of yeast cells is 1.1g/mL, which is higher than that of culture medium), and to temporarily create neutrally buoyant cell suspensions without noticeable adverse effects. This concentration of OptiPrep™ (i.e., 20%) enables the generation of a total of 830,000 microdroplets (~ 28.0% containing single yeast cells) in 30 mins. We note, however, that for studies that require a continuous generation of large amounts of cell-laden microdroplets, a higher concentration of OptiPrep™ or an alternative density-matching reagent of higher density may be necessary.
Moreover, we demonstrated the capability of droplet microfluidic platform for quantitatively tracking of single yeast cell growth of different species, genotypes and phenotypes, and also under different environmental conditions. When single cells are contained in isolated environments, not only can the growth rate of cells be screened, but also the phenotypes to secrete multiple high-value bioproducts (e.g., organic acids, antibodies and cellulases), since all the secreted products are confined within the microdroplet compartments. We can also obtain further understanding of genetic and molecular mechanisms underpinning beneficial phenotypes due to the genotype-phenotype linkages provided by the microdroplets. By combining with high-throughput screening and sorting technologies, e.g., fluorescence-activated cell sorting (FACS) and image-activated cell sorting (IACS) (Nitta et al., 2018), this platform can accelerate the progress of development of yeast strains with desirable properties (e.g., high yield of valuable products, high environmental tolerance and high growth rate) for industrial applications.