Background

---

As explained further in details in the description, the quagga mussel (Dreissena rostriformis bugensis) is an invasive species from eastern Europe and abundant in Swiss lakes.

We carried out two approaches to control its populations: preventing the establishment of mussel colonies by stopping them to attach to surfaces through zosteric acid production by engineered E. coli BL21 (DE3) (design ZA), and killing them through the overexpression of the FitD toxin in Pseudomonas protegens CHA0 (design FitD).

For the second approach, we wanted to increase production of the toxin either (a) directly, by adding to cells an extra copy of the fitD gene, or (b) indirectly, by overexpressing the fitG gene, whose product is an activator of fitD. To do so, we designed and assembled four different plasmids that encoded one of the two genes under an inducible or a constitutive promoter. We first amplified the plasmids in E. coli DH5α and then transformed them into P. protegens CHA0. All the information about these procedures are found in experiments and results. To test if our plasmids were functional, we first needed to find the best conditions to keep the mussels alive in our laboratory. Secondly, we also had to establish the correct way to accurately measure the effect of our engineered cells on the mussels' survival. Indeed, we pioneered the work with invasive mussels in the competition as we are the first iGEM team working with this multicellular organism. We have developed and written several protocols from scratch about its maintenance and survival measurement techniques. Doing so, we proudly paved the way for future iGEMers to tackle this mollusk. Further information is found in the contribution page. Here, we summarize the work we have carried out towards establishing reliable and reproducible methods to measure the effect of our engineered cells on mussels.

Optimization of culture conditions for maintenance of quagga mussels

---

To set the best conditions to keep the mussels alive, we performed several experiments which required a considerable amount of time and substantial efforts. The animals were taken fresh from Lake Geneva on the same day of the experiment and randomly assigned to the different test conditions. We tried containers of various sizes and materials (e.g. PVC and glass), as well as different allocation conditions (e.g. mussels all together, in a container with separations, or one per container). Also, we tested different volumes of water, ranging from 10 mL to 5 L, with or without current created by either a pump or a shaker. We even tested the influence of light and added nutrients like chlorella (an unicellular algae), as suggested in the scientific literature. When they die, mussels release ammonia, which is toxic and can trigger a chain reaction that leads to the rapid death of nearby individuals (Isadora, n.d.). In addition, we observed that when they were in the same container, they attached to one another, which represented a challenge since we had to check on their viability and to regularly remove the dead ones. If you are interested and would like to have more information about the ideal laboratory conditions for the mussels, we prepared a handy summary of all our experiences, that you can find in contribution. Ultimately, we found out that the best setup was to keep the mussels separated from one another in individual small plastic petri dishes. We were then ready to prepare the bacterial solutions of our toxin to be tested on the mussels.

Preparation of engineered Pseudomonas cells for mussel experiments

---

The isolation of the FitD toxin would have required complex, expensive and time-consuming purification steps, which would make our solution impractical. We therefore decided to apply the entire cell content, being careful that our preparation steps would not damage the protein. This approach is consistent with Zequanox, the product currently available on the market, mainly composed of P. protegens extract (Whitledge et al., 2015). For the preparation of the lysed cells solutions we followed a protocol for cell lysis via bead beating, using glass beads (diameter 100 µm) in a Fastprep-24 machine. Since we did not have a standard to quantify the amount of toxin in solution, we measured the OD of the cell culture as a proxy for the FitD concentration, and diluted accordingly (OD 0.25, 0.5 and 1). To measure the amount of cells per mL, we did serial dilutions starting from a 1mL liquid culture at 1 OD and plated them. We then counted the colonies that grew on the plate and calculated the number of cells in the initial sample. We defined this way the concentrations to test: 2·10⁸ cells per mL for OD 0.25, 4·10⁸ cells per mL for OD 0.5, and 8·10⁸ cells per mL for OD 1. All the details about this procedure can be found in experiment. Our solutions were ready to be tested on mussels.

Preliminary experiments

---

After learning how to maintain the mussels in the laboratory, we moved forward to plan the best conditions for our experiment and ensure that our product had the desired effect. A pilot experiment was initially performed to test 31 conditions, and it was followed by other experiments when we saw it was working. There were 10 mussels per condition for a total of 310 mussels. The conditions were 4 constructs (overexpression of fitG or fitD, under constitutive or inducible promoter), two controls (i.e. lake water alone and wild–type cells), a range of concentrations for each (2·10⁸, 4·10⁸ and 8·10⁸ cells per mL), and with and without cell lysis.

Testing of our engineered Pseudomonas protegens strains

---

The results of the preliminary experiments were in general a little noisy, perhaps due to the small number of mussels tested. Nevertheless, they were promising and deserved better focus. To be able to have a higher statistical power, we scaled up the number of mussels to 45 per condition, while we were working with only the most promising constructs and the relevant controls (16 conditions), for a total of 765 mussels instead of 310. The conditions have been chosen based on different arguments and on the preliminary experiments. This way, we implemented the 3R (replacement, reduction, refinement) framework by reducing the amount of mussels used and refining our experimental design, while still obtaining statistically relevant data.

Regarding the constructs, we focused on the ones that gave the most promising results during the preliminary experiments, and kept all the controls with an addition to increase the reliability of our data:

• P. protegens containing FitD inducible.
• P. protegens containing FitG constitutive.
• The positive control wild-type P. protegens CHA0.
• The negative control with only lake water and without any bacteria nor toxin.
• The new negative control P. protegens containing fitD inducible but not induced with IPTG.

Each of these were tested for lysed and not-lysed cells. The 765 mussels were placed individually in petri dishes with 10 mL of the different solutions. They were kept 3 days in the laboratory and checked at 9 a.m., 12 a.m, 4 p.m. and 9 p.m. In the morning at 9 a.m., 5 mL of lake’s water was added in each box to avoid the sample drying out. To see if they were still alive, they were poked with an inoculation loop to test their responsiveness: staying open meant they were dead and immediately disposed.

Our construct overexpressing FitD has high efficacy against mussels

---

Since we wanted to understand if a low toxin concentration is still efficient compared to the wild-type, we did a selection of some results that we obtained throughout the project.
We believe it was important to test the different effects of the two forms lysed and non-lysed on the mussels, as they can influence the final product. In the end, as we reasoned that applying GMOs in the environment would cause serious safety concerns, and the fact that the results of non-lysed cells were not superior to the lysed ones, we decided to show only the lysed ones, as it is more environmentally friendly and consistent with our mindset.
We did a Gehan-Breslow-Wilcoxon statistical test to see if there was a difference between treatments and controls. This test gives more weight to drops in survival probability observed at early times, and less weight to changes that happen later.

In the plot of figure 1 we can see how the engineered and wild-type P. protegens have a higher success than E. coli BL21(DE3) wild-type in killing the mussels. However, there isn’t any significant difference between our constructs and the wild-type P. protegens. Still, the strain P. protegens pSEVA2313 fitG showed promising trends. We figured out that this outcome could be related to the high bacterial concentration (8·10⁸ cells per mL) and the little number of mussels, thus lowering the statistical power. To overcome this, we scaled up the number of mussels, eliminated the OD 1 concentration and reduced the conditions to the most promising plasmids.

In figure 2 are shown the results for the experiment with 45 mussels at the concentration of 2·10⁸ cells per mL. Here, we can confidently confirm that our constructs overexpressing FitG have a significantly higher killing rate than the wild-type P. protegens (p=0.0117), which had no effect on the mussels. Regarding the overexpression of FitD inducible, there is not any significant difference with the wild-type and the negative control P. protegens pSEVA234 fitD not induced (FitD(-)). Nevertheless, it still has a higher success and it could be in the right direction for further research. The continuation of the results’ discussion is explained in proof of concept.

Taken together, these results prove that at least one of our four constructs works well, while the others generally show similar trends, especially P. protegens pSEVA2313 fitD. More precisely, in both experiments shown here, the construct that overexpresses the activator of the toxin is more efficient in killing mussels than the wild-type bacteria that naturally produce the FitD toxin. The latter is the main component of the current standard treatment method (Zequanox) (Whitledge et al., 2015).

Thanks to the experimental results, we were able to feed relevant data into our model, to estimate the effect of periodic treatments with our constructs in a real life application, namely in infrastructures endangered by mussels invasion, such as pipes. Further information can be found in model.

We were the first iGEM team to test on mussels, and we can say that it was a real challenge, but in the end we made it, even proudly. We are confident about the robustness and reproducibility of our data. During this experience we ran several pilot experiments and we spent a lot of time troubleshooting as well as optimizing our methods. We made all the replication that the time we had allowed us, and we put all our energies and efforts to make it work, until it eventually worked. There is still a lot of research to do, and we hope that our experiments can set a basis for future iGEM teams.

References

1. Isadora. (n.d.). How to Keep and Care for Freshwater Clams in Aquariums. PetHelpful. Retrieved October 7, 2022, from https://pethelpful.com/fish-aquariums/Introduction-Keeping-Clams-In-Aquariums
2. James, B. D., Kimmins, K. M., Nguyen, M.-T., Lausch, A. J., & Sone, E. D. (2021). Attachment of zebra and quagga mussel adhesive plaques to diverse substrates. Scientific Reports, 11(1), Article 1. https://doi.org/10.1038/s41598-021-03227-6 https://doi.org/10.1038/s41598-021-03227-6
3. The Mussel Masquerading as a Clam—Pilsbryoconcha exilis—Asian Gold. (n.d.). Quality Marine. Retrieved October 7, 2022, from https://www.qualitymarine.com/news/the-mussel-masquerading-as-a-clam-pilsbryoconcha-exilis-asian-gold/
4. Whitledge, G., Weber, M., DeMartini, J., Oldenburg, J., Roberts, D., Link, C., Rackl, S., Rude, N., Yung, A., Bock, L., & Oliver, D. (2015). An evaluation Zequanox® efficacy and application strategies for targeted control of zebra mussels in shallow-water habitats in lakes. Management of Biological Invasions, 6(1), 71–82. https://doi.org/10.3391/mbi.2015.6.1.06