Team:UZurich/Engineering
Engineering Success
When a plant is confronted with pathogens, one of the first steps for the plant’s immune system is their recognition which consequently leads to the activation of several defence mechanisms in the plant to resist against the attacker. This is done in a process called PTI, short for pattern-triggered immunity. PTI is triggered when the plants pattern-recognition receptors (PRR’s) recognise conserved pathogen-associated molecular patterns (PAMPs). [1]
For Boom V, we decided to take advantage of this first innate immune response to prime the plants' immune system and therefore prepare them for a potential infection.
In the following text we are going to elaborate on our engineering cycle. Before we dive into the details, we will summarise the aims of each step of our engineering cycle:
- Design: Expression of immunogenic epitopes such as flg22 (BBa_K3286138) and elf18 (BBa_K3989023) on the surface of bacterial outer membrane vesicles, which then can serve as PAMPs to PRRs
- Build: Assemble and introduce the designed plasmid into a chassis where it can be expressed
- Test: Conduct structural and functional tests to ensure our design can induce the PTI
- Learn: Draw conclusions about the results and find ways to improve our product
Figure 1: iGEM engineering cycle
Our Cycles
Cycle 1
1.1 Design
The purpose of our design is the delivery of the immunogenic epitopes to the surface of the bacterium, where they can be recognised by the plants PRRs. We chose a construct containing the gene for the ClyA protein (BBa_K811000), which after translation is integrated into the outer membrane. We linked genes for epitopes such as flg22 (BBa_K3286138) or elf18 (BBa_K3989023) to ClyA, which then places them on the surface of the bacterial membrane. In summary: We implemented ClyA to display epitopes and used the epitopes to trigger the pattern triggered immune response.
We based our construct on previous work done by Prof. Cyril Zipfel, which supports that PAMPs can trigger the PTI of plants [2] and other literature that shows that ClyA can be used as a displaying protein [3].
We wanted to achieve the above with a pre-synthesised plasmid containing following elements: ClyA-3xFLAG and the epitopes flg22 (BBa_K3989010) or elf18 (BBa_K3989011).
Figure 2: Plasmid containing ClyA construct controlled by a constitutive J23119 promoter ligated into a iGEM backbone
1.2 Build
In a first step, we wanted to identify an appropriate chassis. Important was that the chassis would overproduce OMVs, which are buddings of the outer membrane of gram-negative bacteria. We chose an E. coli mutant with a deleted TolB gene for the proof of concept. The tolB strain lacks the periplasmic TolB protein which hence compromises the membrane integrity: The bacterial outer membrane and the inner cytoplasmic membrane become less attached. This leads to an increase in OMV secretion [4]. The tolB strain was chosen based on preliminary experiments that showed that both the OMV secretion rate (determined by quantifying the lipid content in the OMV sample) and growth rate were rather high compared to different deletion strains that overproduce OMVs.
Figure 3: OMV quantification based on lipid quantification of different E.coli strains
We transformed the plasmid from above into the tolB mutant but were faced with difficulties with the transformation. During troubleshooting, we came up with a new design that would be more modular, which would bring us some important advantages over our current design (discussed in Cycle 2, Design). Therefore we decided to start a new design cycle.
Cycle 2
2.1 Design
We decided to change our approach for the second design cycle. Instead of designing and synthesizing the plasmid as a whole, we designed fragments of the plasmid which could later be assembled via a Golden Gate reaction.
The improved design looks as follows: We used an LII backbone [5] and synthesized a gene fragment containing ClyA-3xFLAG, and we synthesized a gene fragment with the epitopes flg22 or elf18. We also used primers to add BsaI or BpiI restriction cutting sites to each end of every individual fragment. Then we performed a Golden Gate reaction to link the two fragments together. We first used a constitutive J23119 promoter (BBa_J23119) (more details in Build section) then switched to a Para promotor (inducible by Arabinose) (BBa_I0500) to achieve an inducible epitope-displaying system. The FLAG tags are used for the Western blot, which can identify whether the construct is expressed in our chassis. The backbone of the plasmid is from our lab and can be changed by a digestion-ligation reaction.
The plasmid map is shown below:
Figure 4: Plasmid containing modular ClyA construct controlled by an inducible Para promoter ligated into an LII backbone
This approach poses one key advantage to the previous approach: It is modular. The epitopes can be changed depending on what plant we want to treat (different plants have different preferences for PAMPs) and the ClyA can be exchanged too if one wanted to use a different outer membrane protein to display the epitopes. This allows us to easily adapt the construct to different experiments.
After synthesis we sequenced the plasmid to verify our Golden Gate reaction was successful. Later we would also test the functionality of the design in a series of experiments testing for the presence of the ClyA-epitope structure in our chassis as well as the immunogenicity of our construct (see Test).
2.2 Build
The chassis in the second cycle was kept the same as in the first design cycle.
The construct described in the Design section was built in fragments, which were assembled via Golden Gate. Thanks to the modular nature of our design we could assemble different variants of the original construct using Golden Gate cloning. For the experiment we had planned, we created variants with ClyA only without epitopes, ClyA with flg22, and ClyA with elf18.
While cloning our construct and its variants, which were under the control of a constitutive J23119 promoter, we ran into problems transforming the construct into our E. coli. In an attempt to fix the problem we decided to change the promoter. The reason for this decision was the fact that ClyA belongs to the α-PFT (pore-forming toxins) which can cause pore formation in the bacterial membrane and hence lead to cell death.[6]
Thanks to the Golden Gate cloning we could easily substitute the constitutive J23119 promoter with an arabinose inducible promoter (pBAD) without having to resynthesize the entire construct. This approach allowed us successful cloning and transformation. To verify that our builds are correct, we used colony PCR to get a selection of possible colonies containing our plasmid. We then sequenced the colonies that showed a plausible band on the gel. We proceeded to sequence the colonies that showed a positive result on the gel to be sure our plasmid is contained in the colony.
It must be noted that we are still unsure if the toxicity of ClyA was the true reason for the unsuccessful transformation of our construct with the constitutive J23119 promoter (previous studies showed that it can be used as a surface display tool without killing the cell [3]) or if the problem stemmed from unsuccessful cloning. The fact that we didn’t manage to grow colonies that we could have sequenced prohibits us from making any definite conclusions.
2.3 Test
For the testing of our system we used two different approaches. Firstly we wanted to make sure that the OMVs we harvested after inducing the ClyA construct with Arabinose truly contained the proteins by running a Western blot (structural confirmation). Secondly, we tested if the OMVs containing the epitopes could induce an immune response by conducting a seedling growth inhibition assay (SGI) and a reactive oxygen species assay (ROS) (functional test). The western blot, the SGI assay, and ROS assays all provide quantitative results while the western blot and SGI assay have absolute units and the ROS assay has relative units. To analyze and represent our gathered data from the SGI assay and ROS assay, we relied on statistical tests and graphs created with R studio. We made a statistical test for each assay looking for a significant difference between the individual OMV treatments and the controls.
The Western blot functions as follows: When we run our engineered OMVs through the gel, we expect to detect a band for ClyA-epitope structure. The 3xFLAG we included in the plasmid serves as a binding site for antibodies. In the western blot, if the OMVs contain the ClyA-3xFLAG-epitope structure they are supposed to express, we should see protein-specific bands. For OMVs containing a ClyA-3xFLAG-flg22, we expect a band at ~40kDa, for OMVs containing ClyA-3xFLAG-elf18 we expect about the same. These expectations were confirmed.
Figure 5: Graphical representation of the bands we received when running a Western Blot on E.coli omp8 derived OMVs: ClyA-FLAG-flg22, ClyA-FLAG-elf18 and ClyA-FLAG all show a band at around 40 kDA which confirms that the ClyA construct was implemented in the OMVs
The SGI is an assay that takes seedling weight as a measure for an immune response. We filled OMV solution (treatment) and MS medium (mock) into a well plate and placed the Arabidopsis (Col-0) seedlings into the wells. After ten days, the seedlings are weighed. The stronger the seedling growth has been stunted, the stronger the immune response was. In the SGI assay, we expected to see a stunted growth of seedlings in OMV treated seedlings if the engineered OMVs induce an immune response. We used 20nM flg22 as positive control and MS medium as mock to which we compared our OMV results to. We used three types of OMVs as treatments: ClyA-flg22 OMVs, ClyA-elf18 OMVs, and Cly-A only OMVs. From the data gathered we can see that this expectation was met: We observe that for both ClyA-elf18 and ClyA-flg22 we can see a stronger immune response compared to ClyA without epitopes and the mock at concentration 20ug/ml.
Figure 6: Plotted data from an SGI. We treated A.thaliana (Col-0) with our OMV variants ClyA only, ClyA-elf18 and ClyA-flg22. The ability of the ClyA-construct to reduce seedling weight is significantly influenced by the OMV variant we treated the plant with and the concentration of OMV variant used
The ROS assay takes the amount of reactive oxygen species in a leaf disc as the measure for an immune response. For this assay, we punched leaf discs and let them sit in water overnight to make sure the immune reaction from the punch wound could subside. Then we placed the leaf discs into a well plate. We used water as a mock, flg22 at 200nM as positive control, and three treatments: ClyA-flg22 OMVs, ClyA-elf18 OMVs, and ClyA-only OMVs. The quantification of the ROS burst was done by a plate reader, which detects the luminescence of the ROS reacting with luminol that is generated in the leaf disc. This luminescence will be depicted as a peak (measured in RLU).
In the ROS assay, we expected to see a larger ROS burst in leaf discs treated with OMVs expressing epitopes compared to the mock (water) and the OMVs with a ClyA without any attached epitopes. This was truly the case with Arabidopsis (Col-0) leave discs treated with OMVs that expressed elf18 on their surface. The ROS burst is clearly higher in discs treated with ClyA-elf18 OMVs than in the leaf discs treated with the mock (water) or ClyA only OMVs. The reason why the ClyA-flg22 OMV didn’t induce a higher immune response is due to the fact that elf18 is more immunogenic than flg22 in Brassicaceae, to which our model organism belongs.
However it must be noted that the ClyA-elf18 expressing OMVs did not reach quite the levels of pure flg22. We could try and solve this problem by using a different membrane protein with more extracellular domains to display more epitopes at the surface of the vesicles (see Learn).
Figure 7: Kinetics of ROS response in A.thaliana (Col-0). As a treatment we used the tolB (E.coli) derived OMV variants ClyA-elf18, ClyA-flg22 and ClyA without epitopes (induced ClyA). Flg22 was used as a positive control, water as a mock and ClyA without epitopes as a negative control. We can see that ClyA-elf18 OMVs trigger a high ROS burst measured in RLU
Since we didn’t use any metadata for our experiments, our experiments can be repeated by others as long as they have the plasmid used to express the epitopes. The individual parts of our construct, namely the epitopes and the promoter, can also be tested separately. Since we are using an inducible promoter (induced by arabinose), we used different concentrations of arabinose to test the expression level. This provides valuable data for other teams in the future. The epitopes we mainly tested for their immune triggering capability. Different epitopes elicit different immune responses in different plants, which can be quantified in immune assays. However, the outputs from the SGI and ROS assays are influenced by the environment and the plant, therefore the generated data may vary depending on the lab and location. Having said that, we will provide the R code for the analysis of the ROS assay data on the Measurement page which can be used by other teams.
2.4 Learn
From our data gathered from the SGI assay and the ROS assay we learned that at least some of our engineered OMVs do trigger a higher immune response compared to the controls and mocks. We could however try and increase the immune response further. We hypothesize that we could achieve this by using an omp8 E. coli strain instead of the tolB strain. omp8 expresses low Omp levels in the outer membrane which creates more surface area into which ClyA can be incorporated [7]. This would result in more epitopes being displayed on the bacterial surface and therefore also on the OMVs.
We also learned that we need to take into account what plants we wish to treat with our OMVs since different epitopes would be required to maximize the immune response due to different preferences for different PAMPs among plants. Thanks to the modular nature of our construct, this could be easily achieved.
Cycle 3
3.1 Design
We used the same construct as in the design cycle 2, except that we transferred our construct into an iGEM backbone with lower expression levels than the backbone from cycle 2. Reason being is that a lower expression causes less disturbance to the bacteria cell membrane and therefore increases the viability of our bacteria.
3.2 Build
We changed our chassis for this cycle in an attempt to improve the immune response that our OMVs would elicit. Therefore we transformed our ClyA-epitope variants (ClyA only without epitopes, ClyA with flg22, and ClyA with elf18) from the second cycle into an Omp8 deleted E. coli strain (omp8). We again sequenced the colonies to verify that our plasmid had been taken up during the transformation.
3.3 Test
We tested the engineered omp8 derived OMVs in a ROS assay. We had two questions in mind that we wanted to answer: Firstly we wanted to know if the ClyA-construct can elicit an immune response when they are transformed into the new chassis. Secondly, we wanted to know if the immunogenicity differs between engineered OMVs derived from tolB and omp8.
The answer to the first question is, that ClyA-elf18 OMVs and ClyA-flg22 OMVs both elicit an immune response in A. thaliana (Col-0), while the ClyA-elf18 OMVs again elicit a stronger immune response.
Figure 8: Kinetics of ROS response in A.thaliana (Col-0). As a treatment we used the omp8 (E.coli) derived OMV variants ClyA-elf18, ClyA-flg22 and ClyA without epitopes (induced ClyA). Flg22 was used as a positive control, water as a mock and ClyA without epitopes as a negative control. We can see that ClyA-elf18 OMVs trigger a higher ROS burst measured in RLU compared to the mock
As for the second question, we conducted a ROS assay with ClyA-only, ClyA-elf18, and ClyA-flg22 OMVs derived from tolB and omp8 strains as treatments. The results show that our omp8 derived ClyA-elf18 OMVs indeed manage to elicit a higher immune response compared to tolB derived ClyA-elf18 OMVs. With this result, we can confirm that a change in chassis from tolB to omp8 does improve the immunogenicity of the OMVs.
Figure 9: Comparison of the induced ROS bust of omp8 and tolB derived OMV variants ClyA-elf18, ClyA-flg22 and ClyA without epitopes (induced ClyA) in A.thaliana (Col-0). We can see that the strain used significantly influences the ability of the OMV to elicit a ROS burst. Here all the omp8 derived OMV (except for ClyA induced which is the construct with only ClyA and no epitopes) variants elicit the stronger ROS burst compared to the tolB derived counterparts
3.4 Learn
From the experiments conducted in the Test section we could learn that our plasmid can be transformed into different chassis and that by doing so, the OMVs don’t lose immunogenicity and can even become more immunogenic, if the chassis is chosen correctly.
These results provide necessary data that enables us to implement our design into a bigger picture: An inducible system. This inducible system would include our ClyA-construct that delivers epitopes to the surface and two other plasmids: A TEV-construct and a TolB-construct with a TEV cutting site. All three would be introduced into a tolB mutant strain of choice (e.g. a plant-associated soil bacterium such as P. putida), which lacks the TolB membrane protein that normally helps to hold the two membranes together. With this system, we would be able to produce OMVs in an inducible manner by both repressing the TolB gene in the TolB-construct and by activating the TEV protease gene with a quorum-sensing molecule. This would have the following as an effect: The protease will cut the produced TolB at the inserted TEV cutting site and additionally no new TolB is produced since the gene is repressed. For a more detailed description of this inducible system, go to Design. If implemented in a plant-associated soil bacterium such as P. putida, the engineered bacteria could be released into a field where they could sense pathogens attacking the plant via their quorum-sensing molecules and secrete the immunogenic OMVs as a response to the attack.
For complete scientific names of the strains used see: list strains.
[1] Bigeard, J., Colcombet, J. and Hirt, H., 2015. Signaling mechanisms in pattern-triggered immunity (PTI). Molecular Plant, 8(4), pp.521-539.
[2] Monaghan, J., & Zipfel, C. (2012). Plant pattern recognition receptor complexes at the plasma membrane. Current opinion in plant biology, 15(4), 349–357.
[3] Kim, J. Y., Doody, A. M., Chen, D. J., Cremona, G. H., Shuler, M. L., Putnam, D., & DeLisa, M. P. (2008). Engineered bacterial outer membrane vesicles with enhanced functionality. Journal of molecular biology, 380(1), 51-66.
[4] Turner, L., Praszkier, J., Hutton, M.L., Steer, D., Ramm, G., Kaparakis‐Liaskos, M. and Ferrero, R.L., 2015. Increased outer membrane vesicle formation in a Helicobacter pylori tolB mutant. Helicobacter, 20(4), pp.269-283.
[5] Chiasson, D., Giménez-Oya, V., Bircheneder, M., Bachmaier, S., Studtrucker, T., Ryan, J., ... & Parniske, M. (2019). A unified multi-kingdom Golden Gate cloning platform. Scientific reports, 9(1), 1-12
[6] Pore-Forming Toxin
[7] Thoma, J., Manioglu, S., Kalbermatter, D., Bosshart, P.D., Fotiadis, D. and Müller, D.J., 2018. Protein-enriched outer membrane vesicles as a native platform for outer membrane protein studies. Communications Biology, 1(1), pp.1-9. (Figure 1a)