Overview
This page only reports our results without delving into the design or background of our thinking. More background information can be found on our Engineering, Experiments, and Notebook pages. Additionally, the main results of our dry lab work are also described on this current page. For more details however, please refer to our Model page.
pET29-linker-sfGFP Assembly
We received the pET29-sfGFP plasmid from our lab. We then proceeded to insert a 21 bp linker upstream of sfGFP sequence by Gibson assembly. The successful integration of the linker sequence is shown in Figure 6 below.
Cloning of Biosurfactant Sequences and transformation of E. coli
We selected several biosurfactants from the literature for expression in E. coli. Specifically, we concentrated on creating expression vectors for MBSP1, HFBI, HFBII, and HGFI. To enhance the expression levels of HFBI and HFBII, we performed in silico mutations on the cysteine residues. For HFBI, we developed two constructs: one with half of the cysteine residues mutated and another with all cysteine residues mutated. Additionally, we retained the natural sequence for comparison in all cases. All of the PCR products showed the desired product length MBSP1 (897 bp), HFBI sequences (231 bp), HFBII sequences (294 bp) and HGFI sequences (324 bp), (Figure 1).
Figure 1: Agarose gel of PCR amplified gBlocks. LMW = low molecular weight DNA ladder (NEB).
Following PCR, the PCR products were ligated into our pET29-linker-sfGFP plasmid through Gibson assembly. The ligated plasmids were transformed in NEB 10-beta Competent E. coli cells. After obtained colonies on selection plates, colony PCR was performed. Colony PCR performed on MBSP1 showed positive results, as well as for our HFBI constructs (natural and mutated sequence) and HFBII construct (mutated sequence).
Figure 2. Colony PCR for Gibson assemblies of MBSP1. LMW = Low Molecular Weight Ladder. Results for 8 different colonies.
Figure 3. Colony PCR for Gibson assemblies of HFBI (natural sequence). LMW = Low Molecular Weight Ladder. Results for 8 different colonies.
Figure 4. Colony PCR for Gibson assemblies of HFBI (mutated sequence). LMW = Low Molecular Weight Ladder. Results for 8 different colonies.
Figure 5. Colony PCR for Gibson assemblies of HFBII (mutated sequence). LMW = Low Molecular Weight Ladder. Results for 8 different colonies.
For all our other Gibson assemblies we obtained colonies but could not obtain positive colony PCR results. These were HFBII (natural sequence) and our three HFGI sequences (natural sequence, half mutated sequence and fully mutated sequence). After confirming the correct sizes of bands for MBSP1, HFBI (natural sequence), HFBI (mutated sequence), and HFBII (mutated sequence), plasmids were extracted and sent for sequencing to Macrogen.
The in silico sequences were aligned with the sequencing results, of which the results are shown below in Figure 6. Ligations for MBSP1, HFBI and mutated HFBI were successful. Sequencing also confirmed the insertion of the linker sequence. However, for the mutated HFBII we could not confirm the insertion of the hydrophobin sequence.
Figure 6. Sequencing results for MBSP1, HFBI (natural sequence) and the HFBI (mutated sequence).
Expression of MBSP1 and HFBI Sequences
We induced recombinant protein expression of MBSP1, HFBI (natural sequence) and our HFBI mutated sequence in E. coli BL21 cells with IPTG. The proteins were extracted, purified and analyzed through SDS-PAGE.
Our main observation here is that the HFBI mutated sequence shows more intense bands than HFBI, which indicates that the mutations introduced in the HFBI sequence increased protein expression. The weights of sfGFP, MBSP1, and HFBI are all about 26.8 kDa, 25 kDa, and 7 kDa, respectively. Therefore, we expected to see a band appearing at around 50 kDa for MBSP1, but we only see one slightly higher than 25 kDa. We suspect MBSP1 was cleaved off during protein purification. The results for HFBI are quite similar: for the natural HFBI protein and mutated HFBI protein, faint bands at different heights can be observed. This might indicate incomplete cleavage of the sfGFP and HFBI proteins.
Figure 7. SDS-PAGE for MBSP1, HFBI (natural sequence) and HFBI (mutated sequence). BPSBR = Blue Protein Standard Broad Range P7706S ladder.
Live-cell Time-lapse Experiment
Time-lapse imaging (Figure 8) reflects the first 20 frames from the microscopy experiment, where images were taken every 10 minutes for the first hour and every 25 minutes for the subsequence measurements. In accordance with hydrophobin physical properties1, green fluorescent aggregates can be seen forming at the poles of the cells. The aggregates are less apparent in the case of the mutated HFBI sequence, where disulfide bridges between cysteine residues were removed.
Figure 8a: HFBI natural sequence
Figure 8b: HFBI mutated sequence
Figure 8c: MBSP1
Analyzing the Growth of Yarrowia lipolytica
Figure 9: Optical density (OD) measurements of Y. lipolytica on
a logarithmic scale as a function of time. Each row represents a different carbon
source used in the growth of the yeast (sunflower oil or glucose at concentrations
of 3.7% and 5%, respectively). The OD values were measured at two different dilution
factors (1:10 and 1:100), as highlighted by the 2 separate columns in the panel.
Our project implies that Y. lipolytica would be grown on waste cooking oil (WCO) at some point. Thus, we proceeded to evaluate the growth of Y. lipolytica on sunflower seed oil, a commonly used oil for cooking purposes. We then compared this to the growth on glucose. We know that sunflower oil is not the oil most commonly used for cooking fries in Belgium, but it was available in the lab and still serves as a good starting point.
Two replicates of glucose and four of oil culture were prepared for additional accuracy and OD600 was measured from the cultures every 2 hours for 60 hours.
Based on previous research done on Y. lipolytica's growth on oil we were expecting a slower growth on oil compared to glucose2. However, as seen on our growth curve, we observed a faster growth on oil. This could perhaps be because of an initial glucose starvation as Y. lipolytica was inoculated overnight in 3 ml YPD. This may have led to expression of enzymes to harvest alternative carbon sources.
Figure 10. Growth on 2% glucose medium and 3.7% oil.
Figure 11. Growth on 2% glucose medium and 3.7% oil.
Alternatively, we hypothesized that 5% glucose could also cause osmotic stress to the cells. However, when comparing the growth to growth on 2% glucose (YPD), the cells reached the end OD of around 24 in both cases and the growth was faster in the case of 5% glucose. Therefore, this hypothesis was dismissed. Growth curves are shown in Figure 10 and Figure 11.
The potential effect of oil on the accuracy of the measurements was also a concern. However, when cells were spun down and the size of pellets were compared between the two cultures, we saw that there was in fact a significant difference between the number of cells present in each culture. This allowed us to also rule out this hypothesis.
In conclusion, since none of the previous research comparing these growth conditions was performed on our strain, we concluded that our strain grows faster and reaches a higher OD on oil compared to glucose. This further supports us choosing Y. lipolytica as the strain for our project.
Figure 12. Growth on oil medium modelled by our growth model (Baranyi model).
Figure 13. Growth on glucose medium modelled by our growth model (Baranyi model).
Figures 12 and 13 show the growth curves fitted by the Baranyi model. When comparing the two carbon sources, it's notable that the maximum growth rate of oil is two times smaller than that of glucose (Table 1). Additionally, the maximum yield for oil is larger than that for glucose.
| Parameters | Glucose Medium | Oil Medium |
|---|---|---|
| Initial cell density | 0.599 | 0.631 |
| Maximum growth rate | 0.169 | 0.081 |
| Maximum yield | 3.374 | 4.060 |
Table 1. Maximum growth rates and maximum yields calculated with the Baranyi model for Y. lipolytica growth on glucose and oil medium.
Cloning of Plasmids for Transformation in S. cerevisiae
After successful transformation of E. coli and biosurfactant production, we proceeded to create constructs for transformation in S. cerevisiae as the next step preceding Y. lipolytica’s transformation.
For that purpose, we transformed E. coli with a pBEVY vector assembled with HFBI, HFBII or MBSP1 sequences coupled with the native Mating factor alpha-1 secretion signal as our signal for protein export.
pBEVY, Biosurfactant and Secretion Factor Amplification
Biosurfactants and the secretion factor were successfully amplified using the Q5® High-Fidelity DNA Polymerase (Figure 14). However, the pBEVY amplification was not successful using this enzyme. Therefore, TaKaRa Ex Taq DNA Polymerase was used instead for the plasmid and amplification was carried out successfully (Figure 15). Used primers are shown in Table 2.
| pBEVY Fwd | CACCATCACCATTAATACCGAGCTCGAATTCGACAC |
| pBEVY Rev | GCAGTAAAAATTGAAGGAAATCTCATCCCGGGGAGTTGATTGTATGC |
| MF Fwd | GCATACAATCAACTCCCCGGGATGAGATTTCCTTCAATTTTTACTGCAG |
| MF Rev | TGAACCTCCAGAACCGCCACTTCTTTTATCCAAAGATACCCCTTCTTC |
| HFBI Fwd | AGTGGCGGTTCTGGAGGTTCATCTAACGGCAATGGTAACGTTTGTCC |
| HFBI Rev | CGAGCTCGGTATTAATGGTGATGGTGATGGTGCGCCCCAACCGCTGTTTG |
| HFBII Fwd | AGTGGCGGTTCTGGAGGTTCAAAGTTCTTCACTGCAGCAGCCC |
| HFBII Rev | CGAGCTCGGTATTAATGGTGATGGTGATGGTGAGCAGCACCAACAGCAGC |
| MBSP1 Fwd | AGTGGCGGTTCTGGAGGTTCATCGGACCAATACCTTGACTTTG |
| MBSP1 Rev | CGAGCTCGGTATTAATGGTGATGGTGATGGTGCGTGGAATCCTGGCCC |
Table 2: Primers used in the amplification of vector, biosurfactants, and secretion signal.
Figure 14. Q5 PCR result: 1= ladder, 2=HFBI, 3= HFBII, 4= MBSP1, 5=pBEVY, 6= MF.
Figure 15. Ex Taq PCR result: 1 & 3= DNA ladder (1kb), 2= pBEVY.
Colony PCR
Colony PCR was performed on eight colonies from the plate with the lesser number of colonies using TaKaRa Ex Taq DNA Polymerase.
However, the result of the colony PCR showed an unsuccessful cloning (Figures 16, 17, and 18) for the most part. Further analysis did still show possible successful transformation on colony 7 of HFBII and colony 2 of MBSP1. However, as we were approaching the deadline, the S. cerevisiae transformation was cut short. Table 3 shows the primers that were used in the colony PCR.
| Forward primer | 5' GTTGTTGTCTCACCATATCC 3' |
| Reverse primer | 5' CTGTTTGTTGGAGGATGCCGTA 3' |
Table 3: Primers used in the colony PCR.
Figure 16. HFBII colony PCR result.
Figure 17. HFBI colony PCR result.
Figure 18. MBSP1 colony PCR result.
Enzyme Screening
The results obtained from molecular docking and MD simulations deviated from our initial expectations. In many of the analyzed organisms, the atoms that were anticipated to form hydrogen bonds did not come into close proximity after simulation, even when they appeared to establish hydrogen bonds during the initial docking phase. Of particular surprise is the case of Danio rerio, as the DHCR7 enzyme from this organism had previously demonstrated the highest campesterol yield in the literature3. The docking results indicated strong ligand binding and hydrogen bond formation, yet after conducting MD simulations, the hydroxyl group of ergosta-5.7-dienol and the binding site residues moved apart. This trend was observed in several organisms. We have an idea what could have caused the results to be suboptimal and we elaborate on that on our Model page.
After analyzing the results from all stages of our enzyme screening, we have identified few promising DHCR7 enzymes that we can introduce to Yarrowia lipolytica to facilitate the transformation of ergosta-5.7-dienol into campesterol. In our study, DHCR7 enzymes from Xenopus laevis and Waddlia chondrophila displayed the most promising behavior with our substrate. However, to validate their efficacy, experimental testing is still required. Additionally, we recommend testing DHCR7 from Danio rerio alongside the previously mentioned enzymes, as prior literature4 has indicated its potential for achieving the best campesterol yield.
Hydrophobin Design
We obtained a mutated sequence of HFBI that exhibits a lower isoelectric point than the unmodified HFBI. The isoelectric point decreased from 5.74 to 3.17 due to the mutations K32D, R35D, and K50D. We attempted to employ AlphaFold to investigate the impact of these mutations on the protein structure; however, we encountered a limitation with the tool. The algorithm heavily relies on sequence similarity and cannot effectively detect the influence of point mutations on protein stability.
Furthermore, we employed another tool, PredictSNP, which did not indicate that the introduced mutations would disrupt the stability of the hydrophobin. Nevertheless, this study represents a preliminary assessment, and a more comprehensive analysis, along with wet lab experiments, will be necessary to determine the true success of our design.
- Torkkeli, M., Serimaa, R., Ikkala, O. & Linder, M. Aggregation and self-assembly of hydrophobins from trichoderma reesei: Low-resolution structural models. Biophysical Journal 83, 2240–2247 (2002).
- Du HX, Xiao WH, Wang Y, Zhou X, Zhang Y, Liu D, Yuan YJ. Engineering Yarrowia lipolytica for Campesterol Overproduction. PLoS One. 2016 Jan 11;11(1):e0146773.
- Zhang, Y. et al. Improved campesterol production in engineered Yarrowia lipolytica strains. Biotechnol Lett 39, (2017).
- Ryan Sestric, Garret Munch, Nazim Cicek, Richard Sparling, David B. Levin, Growth and neutral lipid synthesis by Yarrowia lipolytica on various carbon substrates under nutrient-sufficient and nutrient-limited conditions, Bioresource Technology, 164, 2014, 41-46, ISSN 0960-8524.