Results

Initial testing

The basis of our project revolves around counteracting the toxic effects of the mevalonate pathway products. As such, we first performed a growth assay to better quantify the effects of PA on cell survival. We found the most resilient of the strains we had access to (DH5α, BL21, DH10b, C41 and C43). Figure 1. shows that all cells have no growth under 1024 µg/ml perillyl alcohol (PA) concentration but at all of the lower concentrations they all grow. BL21 had the highest growth and DH5α having the lowest growth, shown with the final readings of each graph having little to no error bar overlap. We chose to use DH5α cells to perform our initial cloning experiments (due to its higher insert stability) and then once we confirmed the plasmid are correct, we transform it into the more PA resistant and faster growing BL21 cells.

With our project revolving around mevalonate pathway plasmids that produce limonene and PA purchased from Addgene (more info on the contributions page.), the next step was to transform them into our cells (Alonso-Gutiérrez et al., 2013). After determining the cell types we want to use, we next confirmed the plasmids were as expected. To do this, we performed a restriction digest using enzymes EcoRI and SacI (as shown in Figure 2). For the most part, the digest was very successful however we did see an unexpected band that persisted after numerous digests with different enzymes. It did not affect the plasmids function, however we contacted the Addgene who provided it to us for more information (see more information on the contributions page. We then transformed the plasmids into homemade chemically competent E. coli BL21 and DH5α through heat-shock.

Using these cells, we next tested to visualise the burden of inducing our pathway on cell growth and survival. As you can see in Figure 3, more notably in the case of BL21 cells, the more IPTG present (and the more the plasmid is induced) the lower the cell growth. This is because the cell is spending more resources to induce the plasmid, which increases the amount of toxic products being produced by the mevalonate pathway enzymes (also lowering growth). As the IPTG concentration decreases, the BL21+ cells spend fewer resources on the plasmids and due to lower induction, less toxic products will be present to slow growth - because of both these reasons the BL21+ cells grow more. At 0µM IPTG, the BL21+ cells containing mevalonate pathway plasmids have very similar growth to the normal BL21 cells and the error bars overlap, this is expected as it doesn’t have any induction of the plasmid. DH5α on the other hand shows significantly less growth with and without the plasmid, and even with no IPTG the DH5α cells with plasmid grow less than those without. Confirming that for our final tests, we should use BL21 cells to host our plasmids and synthesise our product.

Meanwhile, we also saw fit to test for the production of PA. Without a distinct absorbance wavelength, the method we used for analysing the presence of perillyl alcohol was gas chromatography-mass spectrometry (GCMS). Using the literature to see the most common fragmentation patterns of limonene and PA we could find the retention time in each sample by looking at the chromatograms (Figure 4). Limonene has a retention time of 4.849 minutes within our runs with m/z 57, 69 and 98 peaks being the identifiable fragments showing in the chromatograms. Looking at the standard, PA has a retention time of 6.028 minutes within our runs and its most notable fragments showing in the chromatograms are m/z 67, 79 and 91. With the intensity on the Y axis, it is clear that the BL21+ sample has a higher PA concentration than DH5α+ and PA is produced in large amounts compared to limonene.

Testing our idea

After laying the foundation for our project with the previous tests, the next step was to edit our plasmids and test our modifications. Using primers designed on Benchling and ordered from IDT/Twist, our goal was to add a pump and tet protein sequence to the pJBEI-6411 plasmid and add a short Ptet promoter sequence to the pJBEI-6410 plasmid (to introduce a pump that removes the toxic products alongside a feedback system that downregulates enzymes in response to stress that upregulates tet production). After the Gibson assembly reaction was performed, the reaction was transformed into chemically competent DH5α cells and grown on an antibiotic selective plate. After this, the plasmid was checked for the presence of the insert using colony PCR followed by plasmid purification, restriction digest and full-plasmid sequencing.

After checking over 50 colonies for potential pJBEI-6411* (containing modified plasmid insert) with predesigned primers, we chose 3 colonies (18, 54, 55) that had a band at 709bp (shown in Figure 5). The purified plasmids were then digested using AvrII and EcoRI enzymes to further confirm that they are correct (shown in Figure 5).

With the modified pJBEI-6411 plasmid (6411*) complete much earlier than 6410*, we wanted to assess the effectiveness of our pump without any input from the feedback system that reduces enzyme expression. Similar to before, we performed an assay to test the growth of cells in different concentrations of IPTG (which induces the plasmid). With more IPTG, the plasmid will be induced at higher levels and as such more enzymes will be made to contribute to the production of limonene and PA. As you can see in Figure 6, at very large IPTG concentrations (100uM), expression of the pump is very effective at increasing cell survival compared to without but is still lower than the cells without plasmid. As the IPTG concentration is lowered, the cells containing plasmid are induced less and grow more (closing the gap between growth of the three BL21 types).

Eventually, after changing the cloning method from overlap extension PCR to Gibson assembly (see Engineering success for more detail), we obtained 6410* plasmid (containing the Ptet insert). This was first confirmed by doing colony PCR on ~50 colonies with an expected band for 6410* being 768bp using a predesigned primer. The first colony PCR (Figure 7) had a primer annealing temperature too low so results are inconclusive, though promising colonies were still kept for a restriction digest test (8, 11, 14 and 16). In the second gel things (Figure 7) were clearer and we thought that colony 38 looked the most accurate, we also took colonies 39, 40 and 48 that we could test later.

Using these 8 colonies, we then set up a restriction digest of the purified plasmids using enzymes EcoRI and ApaI that will give a 6410* band at 1061bp and the unmodified pJBEI-6410 will give a band at 1007bp. As seen in Figure 8 (on the left gel), despite the pJBEI-6410 band not showing up, the colony 11 and 14 bands appeared higher than the others so we decided to repeat the digest using a higher concentration of control plasmid. We also included colony 39. On the second gel, all three of the colony bands appear above the unmodified pJBEI-6410 control so we were confident that they had the insert. Nevertheless, using a primer that binds in the insert region, we sent the plasmids off for Sanger sequencing to do one final check and results were positive. Using this data, after the wiki freeze on the 12th we plan to test out the construct effectiveness by doing another IPTG assay and another GCMS experiment (if possible).

Overall, after running out of time to put more work into our project, there are still plenty of experiments that we would have liked to have done. In particular, testing each stage of the system individually using fluorescence to monitor membrane stress, production of our membrane pumps and production of tet would be very helpful to confirm that our system is working. Doing even more repeats of both growth curves and GCMS would also be helpful to ensure our results are all valid and accurately quantify production. Having more data for our models would also give us many more insights.

Conclusion

We began the summer with the goal of increasing production of perillyl alcohol (PA) and limonene by mitigating the toxic effects of said products. We initially planned to modify both mevalonate plasmids to contain not only a pump system but also a potential feedback system that would downregulate enzymes. Despite not performing as many tests as we had hoped, we still laid a strong foundation for our project to work off by showing the limitations of producing PA/limonene intracellularly. At the end of the project, we successfully modified both plasmids and, with quite promising preliminary results, showed that our pump system in particular resulted in higher cell growth. While our feedback system wasn’t tested (before the wiki deadline) and we couldn’t accurately quantify the PA produced with and without each component of our project, we are pleased with our progress and can say without a doubt that we enjoyed our time in the lab and learning just a fraction of what synthetic biology is capable of.

References