First, we studied the effects of different TAG lipase (TGL3, TGL4, TGL5) and ZWF1 gene deletions on lipid production in a batch culture experiment. The strains were inoculated in a liquid medium at equal densities and the lipid accumulation was monitored by Nile Red staining after 24 and 72 hours of cultivation. Nile red is a fluorescent dye that interacts specifically with neutral lipids in lipid droplets (LDs) and it does not stain the cellular membranes (Greenspan et al., 1985). In 24 hours after dilution, the cells from wild-type strain showed very low Nile red staining, indicative of low intracellular lipid levels (Fig 1, 2A). Single deletions of TGL3 and TGL4 resulted in a considerable increase in lipid accumulation, with tgl3Δ having a slightly greater effect than tgl4Δ (Fig 1, 2A). Interestingly, in the cells with TAG lipase deletions, clearly visible LDs could be detected in both the bright-field image and the Nile Red fluorescent image (Fig 1). Although tgl3Δ tgl4Δ double deletion did not further increase the lipid levels in LDs, a triple deletion of all major TAG lipases (tgl3Δ tgl4Δ tgl5Δ) had a profound positive effect on lipid build-up. In the 24-hour time point, the Nile Red staining intensity of the triple TAG lipase deletion strain was not affected by the deletion of ZWF1 (Fig 1, 2A).
The lipid accumulation was also measured at 72 hours following dilution. In this time point, the culture had reached the stationary phase, and LDs were clearly detectable with Nile red staining for all tested strains, including the wild-type (Fig 2B). Lipid accumulation is known to increase in S. cerevisiae cells upon nutrient depletion (Werner-Washburne et al., 1993), and our results reveal that the difference between wild-type and TAG lipase deletion strains is considerably reduced in the stationary phase cells (Fig 2B). In the 72-hour time point, we did not observe a statistically significant difference between wild-type and the single tgl4Δ or double tgl3Δ tgl4Δ deletion cells. Single deletion of TGL3, which had the largest effect in the 24-hour time point, also led to increased LD staining intensity at 72 hours. We observed the highest intracellular lipid levels with the triple tgl3Δ tgl4Δ tgl5Δ deletion strain, and interestingly, in this time point, the zwf1Δ strain had considerably lower LD staining, comparable to the level of the wild-type strain (Fig 2B). Although the lipid levels of zwf1Δ strain were not lower in the 24-hour time point, ZWF1 deletion could be expected to result in decreased lipid synthesis, as Zwf1 is required to regenerate NADPH, a critical cofactor in fatty acid synthesis. Taken together, by preventing TAG degradation with triple deletion of TAG lipases, we have achieved a considerable increase in lipid production compared to the wild-type strain.
Figure 1. Nile red staining shows build-up of lipids in LDs upon TAG lipase deletions. Microscopy images showing the cells in brightfield image and the fluorescent signal of lipids stained with Nile red. The cells are from 24 hour time point after inoculation.
Figure 2. TAG lipase deletions result in greatly increased lipid accumulation in LDs. (A, B) Plots showing the mean Nile red fluorescence intensities in single cells from indicated strains. The samples were analyzed at 24 hour time point (A) and 72 hour time point (B). The bars show median and its 95% confidence intervals. ****, **, *, and ns note p-values <0.0001, <0.01, <0.05, and >0.05, respectively, of pair-wise comparisons with wild-type using Mann-Whitney U test.
We aimed to optimize the production process so that the growth phase (biomass accumulation) would not be negatively affected by simultaneous product synthesis. This would allow rapid biomass growth and shorter production cycles. Thus, we were interested to test whether the modifications in our production strain affect the growth rate. For this, we tested the effect of TAG lipase and ZWF1 deletions in a serial dilution viability assay. The experiment showed that while TAG lipase deletions did not significantly affect the growth rate, the ZWF1 deletion did reduce the growth rate (Fig 3). This suggests that the production process could be made more efficient by triggering inducible shutoff of Zwf1 at a time when sufficient biomass has been reached. In this case, Zwf1 would be present during the growth phase, allowing normal growth rate, but would be inactivated at the same time when the lipid synthesizing enzymes are induced.
Figure 3. Viability assay shows a decreased growth rate of zwf1Δ strain.
As our next step to improve lipid bioproduction, we tested whether expression of human perilipin Plin3 could further increase the lipid bioproduction in the tgl3Δ tgl4Δ tgl5Δ zwf1Δ strain. For this, we constructed and transformed to these cells a plasmid that contains the light-inducible transcription factor VP-EL222 expression cassette and the PLIN3 expression cassette with the light-regulated promoter. The cells containing this plasmid were set to grow at 30 °C either under constant illumination or in the dark. In 24 hours, the lipid accumulation in LDs was monitored by Nile red staining. Interestingly, we detected significantly higher lipid levels in the cells that were grown under illumination, which activates PLIN3 expression via VP-EL222 (Fig 4). As Plin3 is expected to promote LD assembly, the observed effect could be due to stabilization of the synthesized TAGs. We did not observe any noticeable changes in LD morphology between cells that either express or do not express Plin3 (Fig 4A). These results show that the lipid bioproduction in S. cerevisiae could be further improved by expression of human perilipin Plin3.
Figure 4. Light-induced expression of perilipin greatly promotes lipid accumulation. (A) Microscopy images of exemplary cells showing the accumulation of lipids in LDs detected by Nile red staining. tgl3Δ tgl4Δ tgl5Δ zwf1Δ cells with PLIN3 expression plasmid were grown with or without illumination. (B) Plot showing the average fluorescence intensities of Nile red staining. The bars show median values and its 95% confidence intervals. **** indicates p-value <0.0001 in pair-wise comparison by Mann-Whitney U test.
To maximize the carbon yield in lipid production, we set out to engineer yeast biohybrids that could take use of solar power to drive their metabolism. For this, we used indium phosphide nanoparticles, which, as previously shown, when assembled on the cell, can provide electrons for NADPH regeneration (Guo et al., 2018). As NADPH is a critical cofactor for fatty acid synthesis, we hypothesized that these nanoparticles could also support lipid synthesis. To test this, we coated the cells from tgl3Δ tgl4Δ tgl5Δ zwf1Δ strain with the indium phosphide nanoparticles and monitored their lipid accumulation when grown either in darkness or under constant illumination. At 24 hours after inoculation, we measured the lipid levels by Nile red staining. We found that the biohybrids grown under light contained substantially more intracellular lipids than the ones grown in darkness (Fig 5). This indicates that our production strain is able to harvest the light energy and direct it to lipid synthesis. Thus, we have made significant improvements to TAG accumulation in yeast by combining three different engineering approaches. First, we showed that deletion of TAG lipases results in increased lipid levels in LDs. Secondly, we found that expression of human perilipin Plin3 in yeast leads to lipid accumulation, presumably by promoting LD assembly and stabilizing TAGs. Thirdly, we used nanoparticles to give our production strain a new quality – the ability to use light energy to drive the lipid synthesis.
Figure 5. Biohybrids use light energy to drive lipid synthesis. (A) Microscopy images of exemplary cells showing the accumulation of lipids in LDs detected by Nile red staining. The cells from tgl3Δ tgl4Δ tgl5Δ zwf1Δ background and functionalized with indium phosphide (InP) nanoparticles were grown with or without illumination. (B) Plot showing the average fluorescence intensities of Nile red staining. The bars show median values and its 95% confidence intervals. **** indicates p-value <0.0001 in pair-wise comparison by Mann-Whitney U test.
To further increase the lipid production, we aim to introduce cassettes for light-inducible expression of PAH1 and DGA1 . These genes encode for enzymes that are involved in TAG biosynthesis and it has been shown previously that their overexpression leads to increased TAG accumulation (Ferreira et al., 2018) . Further, we plan to increase the precursor supply by removing the feedback inhibition of Acc1, which catalyzes the formation of malonyl-CoA, a precursor for fatty acid synthesis (Chen et al., 2018) . In our experiments, we found that ZWF1 deletion caused decreased growth rate (Fig 3). To optimize the production process, we plan to use an inducible depletion strategy to remove Zwf1 protein from the cells at the same time as the metabolic switch to trigger lipid synthesis is induced. Recent studies have developed light-triggered depletion methods (Deng et al., 2020) , which would be an efficient solution for our strain. After introduction of every modification to our production strain, we will measure the lipid production to test if the implemented designs have the expected effects. We observed a beneficial effect of functionalizing the cells with light-harvesting indium phosphide nanoparticles (Fig 4). Although using the yeast-nanoparticle biohybrids enables maximization of carbon yields, we find that at the moment there are several limitations to using the indium phosphide nanoparticles in a large-scale production. First, the process of creating the biohybrids is rather complex. Secondly, indium phosphide is an expensive compound, so it would be necessary to recycle the nanoparticles to decrease the production costs. In fact, we have discussed this issue with Prof. Guo, who confirmed that the nanoparticles could be recycled. In order to increase the competitiveness of this approach, we plan to further study different methods that enable functionalization of cells with nanoparticles, keeping in mind that the method should be easily upscaled. Finally, as the product extraction often makes up 85% of total production costs in biosynthesis, we plan to engineer our production strain cells to be able to autolyse upon induction. For this, we aim to improve our last years project, where inducible autolysis was acheived by expression of bacterial glucanases. We hypothesize that by targeting the glucanases directly to the cell wall, the local concentration of glucanases will reach higher levels, necessary for the breakdown of yeast cell wall.
Chen, X., Yang, X., Shen, Y., Hou, J., and Bao, X. (2018). Screening phosphorylation site mutations in yeast acetyl-CoA carboxylase using malonyl-CoA sensor to improve malonyl-CoA- derived product. Front. Microbiol. 9. Deng, W., Bates, J.A., Wei, H., Bartoschek, M.D., Conradt, B., and Leonhardt, H. (2020). Tunable light and drug induced depletion of target proteins. Nat. Commun. 11, 1–13. Ferreira, R., Teixeira, P.G., Gossing, M., David, F., Siewers, V., and Nielsen, J. (2018). Metabolic engineering of Saccharomyces cerevisiae for overproduction of triacylglycerols. Metab. Eng. Commun. 6, 22–27. Greenspan, P., Mayer, E.P., and Fowler, S.D. (1985). Nile red: A selective fluorescent stain for intracellular lipid droplets. J. Cell Biol. 100, 965–973. Guo, J., Suástegui, M., Sakimoto, K.K., Moody, V.M., Xiao, G., Nocera, D.G., and Joshi, N.S. (2018). Light-driven fine chemical production in yeast biohybrids. Science (80-. ). 362, 813–816. Werner-Washburne, M., Braun, E., Johnston, G.C., and Singer, R.A. (1993). Stationary Phase in the Yeast Saccharomyces cerevisiae.
