Team:UZurich/Results

Results

1. Engineered OMVs that trigger stronger immune response

Key method

In this part, we mainly performed two plant immunity assays: seedling growth inhibition (SGI) assay (Figure 0A, in which lower seedling weight indicates that higher immune responses have been triggered, due to the trade-off between immunity and growth) and ROS burst assay (Figure 0B, in which higher ROS production by leaf discs means higher plant immune response) to measure the immune response (Figure 0). CMVs and OMVs were isolated from different bacteria strains, and their immunogenicities are tested on different plant genotypes (see list of strains and plant genotypes).

Figure 0: Experimental procedures of seedling growth inhibition and ROS burst assay (modified from [1])

Results

Natural bacterial membrane vesicles can elicit a strong plant immune response

First, we wanted to test whether bacteria originated membrane vesicles can trigger plant immune respones. More specifically, which type of vesicle should be used, and what the optimal concentration is. To answer these questions, a seedling growth inhibition assay (SGI, in which lower seedling weight indicates that higher immune responses have been triggered, due to the trade-off between immunity and growth) was performed in A. thaliana Col-0 using both outer membrane vesicles (OMVs, natural bulging of outer membrane, thus containing periplasmic contents) and cytoplasmic membrane vesicles (CMVs, generated by explosive cell lysis, thus containing cytoplasmic contents) deriving from the E. coli BL21 omp8 strain, an OMV-overproducing strain.

Figure 1: Plot of SGI experiment data comparing dosage of vesicles

The performed Two-Way Anova (F(15, 112) = 26.75, p < 2.2e-16, multiple R2 = 0.7818) yielded significance for the concentration (p = 6.23e-10), the vesicle type (p < 2e-16) and the interaction between the two (p = 7.32e-08). This shows that not only the concentration, but also the type of the vesicles have an influence on the strength of the immune response. What's more, the impact of the concentration is dependent on the vesicle type. From a biological viewpoint, E. coli OMVs are more immunogenic than CMVs, therefore OMVs are the better option to use to trigger plant immune responses, hence we used 5μg/mL OMV to trigger plant immune response for subsequent experiments.

After proving that bacterial membrane vesicles, especially OMVs, can trigger plant immune responses, we came up with questions related to the real-world application:

  1. Can we use plant-associated bacteria to produce membrane vesicles in a safer way?
  2. Can our vesicles trigger an immune response in different plant species?

We tested the immunogenicity of OMVs and CMVs from three plant-associated bacteria:

  1. A. tumefaciens GV3101 (disarmed lab strain commonly used for plant transformation)
  2. P. fluorescence CHA0 (a famous biocontrol strain from Switzerland)[3]
  3. P. putida IsoF (a plant growth-promoting strain)[4]

Two A. thaliana genotypes were used to represent different plants:

  1. wildtype Col-0, with an immune receptor called EF-Tu receptor (EFR), representing plants from the Brassicaceae family
  2. efr-1 mutant, without EFR receptor, representing plants that are not from the family Brassicaceae

By conducting a ROS experiment comparing different treatments and their effects on the immune response in different A. thaliana genotypes, the Two-way ANOVA (F(13, 154) = 5.558, p = 3.106e-08, multiple R2 = 0.3193) yielded significance for the treatments (p = 1.679e-8) and the interaction (p = 0.01101). This suggests that treatments do have an effect, but there is no difference between the different plant genotypes in their overall immune response to treatments. However, data also suggests that the absence of the EFR receptor affects the immune response, which could imply that the immunogenicity of OMVs from certain strains could be dependent on EF-Tu. If we look at the R2-value, we can see that there are probably explanatory variables missing in the setup/model. For the real-world application, it means that certain vesicles might not be able to trigger strong immune responses in all crops.

Figure 2: Plot of ROS experiment data in A. thaliana Col-0 and efr-1

Moreover, we wanted to know more about how OMVs trigger the plant immune response, since it can guide us to design OMVs with higher immunogenicity. For this purpose, we tested membrane vesicles with A. thaliana lacking different immune receptors. Comparing six different treatments in four different A. thaliana genotypes, the Two-way ANOVA (F(23, 160) = 3.903, p = 1.549e-07, multiple R2 = 0.3594) yielded significance for the plants genotype (p = 0.00671), the strain (p = 6.67e-07) and the interaction (p = 0.00347). These results indicate that at least some of the response caused by membrane vesicles is dependent on the presence of certain immune receptors. Post-hoc tests (TukeyHSD) imply that plant growth-promoting bacteria P. putida would be a good choice to produce OMVs, because its OMVs induce the strongest immune response among all vesicles tested, and its immunogenicity is EFR-independent, implying its potential to trigger a strong immune response in a wide range of plant species.


Figure 3: Plot of SGI experiment data with 4 phenotypes of A. thaliana

With these preliminary questions answered, let us move on to the data concerning our own engineered vesicles.


Immune response can be further enhanced by displaying certain proteinaceous immune elicitors on the OMV surface

To verify the functionality of our construct, we performed a ROS assay with a negative control (Water/’Mock’), a functional control (ClyA-only, without any peptide elicitor fused), a positive control (200nM flg22 peptide), the construct ClyA-elf18 (BBa_K3989011) and the construct ClyA-flg22 (BBa_K3989010).


Figure 4: Plot of ROS experiment data of immune elicitor display constructs (performed post-hoc tests on One-way ANOVA: non-parametric test, Kruskal Wallis test)

Flg22 (BBa_K3286138) and elf18 (BBa_K3989023) were chosen as the elicitors to be displayed on the OMV surface using ClyA (BBa_K811000), because flg22 is recognized as pathogen-associated molecular pattern (PAMP) in many different plants and elf18 is recognized as a potent PAMP in Brassicaceae, which A. thaliana belongs to. The Two-way ANOVA showed significant differences between OMVs with and without elicitors. This means that the elicitors on our constructs are crucial for the impact they will have on the plant immune response.

Post-hoc tests (TukeyHSD) have also shown that our constructs can elicit an immune response in A. thaliana different from the induced ClyA only, which is a functional negative control that shows if the immune response induced by the constructs is not mediated by ClyA and this suggests that our constructs do in fact work.

To further confirm our results, SGI assays were performed using both E. coli BL21 omp8 and E. coli tolB strains. Two-way ANOVA results indicate that engineered OMVs are indeed capable of inducing a stronger immune response. The post-hoc (TukeyHSD) indicated that this difference becomes visible from 10µg/ml.


Figure 5: Plot of SGI experiment data of immune elicitor display constructs

OMVs with cellulose-binding domain can also trigger stronger immune responses


In addition to the surface display of plant immune elicitors, we also sought for other strategies to improve the performance of our OMVs for agricultural application. During our integrated human practices, a farmer told us that pesticides applied to plants are often washed away by rain. To solve this problem, we came up with the idea to make OMVs better attach to plant tissue by displaying a cellulose-binding domain on the surface of the OMVs. Although we had no time to show that engineered OMVs with dCBD are washed off by rain less extensively, we found that they can trigger stronger immune responses. As shown in Figure 6, the seedling weight of the negative control group gradually decreases as OMV concentration increases. However, for the dCBD group, the seedling weight dramatically decreases between concentrations of 0 and 2.5 but doesn't decrease a lot as the concentration further increases, which means the immune response might get saturated.


One possible explanation for such results is the enrichment of OMVs by cellulose in the cell wall. At low concentrations, the cell wall limits the diffusion of wild-type OMVs, but for engineered OMVs with dCBD, they get enriched in the cell periphery, generating a higher local concentration and facilitating the diffusion. At higher concentrations, the enrichment effect gets saturated, and OMVs attached to cellulose can even block free OMVs from entering, resulting in no difference between ClyA-dCBD and ClyA-only groups.


Figure 6: Plot of SGI experiment data of cellulose binding domain display constructs

Conclusions

  1. To induce plant immune responses, OMVs are a better choice than CMVs, due to their high immunogenicity.
  2. Plant growth-promoting bacteria P. putida is a good chassis for OMV production in a real-world application, since it can probably induce a high immune response in most crops.
  3. Our engineered OMVs with either immune elicitor or cellulose-binding domain displayed on surface can trigger stronger plant immune responses than natural OMVs.

2. CMV-based delivery system

Additionally to the main part of our project, we also conducted the fusion experiment. For this experiment we were advised and supervised by Professor Olivier Voinnet, an RNA silencing expert at the ETH.

We wanted to find out if vesicles could fuse with and therefore transfer DNA into plant cells. We used CMVs rather than OMVs since they contain all cell contents, whereas OMVs only contain periplasm. Professor Voinnet suggested using one of his viral vectors containing only GFP and a replicase, which will amplify the GFP signal in a single cell by replicating itself. In addition, we used P19, an RNAi blocker, to inhibit the plant's antiviral response. We infiltrated wild type N. benthamiana with 4 different treatments:

  1. positive control, 1:1 mix of agrobacterium with the viral vector and agrobacterium with P19. 2 leaves were infiltrated.
  2. 1:1 mix of E.coli (BL21 DH10b) with the viral vector and with P19, 2 leaves were infiltrated.
  3. 1:1 mix of CMVs from (BL21 DH10b) with the viral vector and CMVs with P19, the CMVs were prepared using MMC as a mutagen to induce cell lysis. 1 leaf was infiltrated.
  4. We used plain Buffer (MES/MgCl2) as a negative control. We are aware that this is not the most suitable negative control, but we did not have E.coli with an unrelated plasmid prepared. 2 leaves were infiltrated.


We used new gloves and syringes for each infiltration to minimise risk of contamination during infiltration.
Since this viral vector is non-virulent, meaning it can’t spread systemically, we knew even if a fusion happens it would only be the case in isolated cells. Therefore we looked at our plants after 1 week under the confocal microscope (as soon as the positive control was visibly green under UV light).

Figure 7: Positive control from the fusion experiment


This is the positive control, you see the cells are packed with GFP, which makes sense, since agrobacterium is an established DNA delivery system.


Figure 8: E.coli infiltrated cell from the fusion experiment


This is an epidermal cell from an E. coli infiltrated leaf. The GFP illuminates the nucleus and the cytosol, which are pressed to the cell borders by the vacuole. We did not do exact cells per area counts, but we took 4-5 samples of about a 1x2 cm size and saw 5-10 fluorescent single cells.


Figure 9: CMV infiltrated leaf from the fusion experiment


This is from a CMV infiltrated leaf, where we did not find evidence for a fusion in any of the CMV samples. This however does not necessarily mean that a fusion is impossible, it could also be due to our mutagen damaging the plasmids.


In all of the samples there was some autofluorescence, as well as fluorescence around the site of infiltration. However, these were clearly distinguishable from actual fluorescent live cells.

Conclusion of the first experiment

Could E. coli possibly possess a delivery system, allowing the transfer of DNA between plant and E. coli cells? Due to the small sample size and lack of an appropriate negative control, we can not conclude anything. Our result could also stem from contamination. Therefore, we repeated the experiment to see whether we could replicate our results. However, we used a different mutagen (UV rays) to prepare our CMVs, since the chemical mutagen (MMC) stays in the plant and keeps on mutagenising, which could damage our construct. For more information on the general process see our protocol for the first fusion experiment.


Results of the repeated experiment


Figure 10: Positive control (A. tumefaciens)


In this figure we can see clear fluorescence as in the positive control in the first experiment. Thus our construct seems to work.


Figure 11: E. coli BL21


In this figure we can only see some autofluorescence of the plant organelles. The infiltrate was the E. coli strain BL21 which is a common lab strain, which we also used during our project. We cannot see any evidence for DNA delivery in this experiment.


Figure 12: E. coli DH10b


This infiltrate consists of the same E. coli strain (DH10b), that seemed to induce fluorescence in the first experiment, however in this repeat, we could not find any evidence for DNA delivery into the plant through this bacterium.


Figure 13: CMVs


As in the first experiment, we could not find any evidence for DNA delivery into the plant, through our CMVs.


Figure 14: CMVs without p19


We also tested the CMVs without an P19 background. Here as well, one can only see a small amount of autofluorescence but no evidence for DNA delivery into the plant.


Figure 15: Negative control 1 (E. coli lysate)


We included the cell lysate of E. coli as a negative control to see whether in our first experiment, the DNA was released as the bacteria lysed inside the plant’s interstitium and could be taken up even without any living bacteria. Here as well, we couldn’t detect any DNA uptake by the plant epidermal cells.


Figure 16: Negative control 2 (Pure viral vector DNA)


To also test whether the plant epidermal cells could just take up pure viral vector DNA, we also tested a second negative control which was an infiltration, with just the vector. Here we can also see no DNA uptake by the plant.



Conclusion of repeated experiment

After analysing our samples from the second experiment, we could not find any evidence for induced fluorescence in any of our samples, except for the positive control. Thus we conclude that there was probably some kind of contamination in the first experiment that led to the fluorescence. With the experiment repeated, we could not find any evidence for any DNA delivery into the plant epidermal cells through either CMVs, OMVs, E. coli BL21, E. coli DH10b, pure viral vector or E. coli lysate with viral vector.

3. Inducible OMV production & Improvement of an existing part

Characterization of a Quorum Sensing regulator protein EsaR and its variants

The core of our inducible OMV production system (see our Design page) is the repression of TolB expression when engineered bacteria perceive quorum sensing molecules from pathogens. This process is mediated by the EsaR regulator and PesaS promoter.

Previous teams have already introduced the wild type EsaR protein (BBa_K2116001) and its basic information. In our project, we characterized its function of regulating the downstream gene transcription based on the amount of AHL molecule added. Together with the wild-type, we also characterized three variants of this protein: EsaR-I70V (BBa_K3989003), EsaR-D91G (BBa_K3989004) and EsaR-V220A (BBa_K3989005).

The construct we used to characterize these parts with is a combination of the EsaR protein, promoter of the protein PesaS (BBa_3989009) and a GFP protein (BBa_3989025). The promoter we used, once activated, starts the transcription when EsaR binds to it. But when there are AHL molecules in the environment, EsaR will interact with them instead and dislocate from the promoter. Thus, the transcription will be suppressed or even blocked when the concentration of the AHL molecules is high.

From our characterization, we found that the D91G and V220A variants are more sensitive to the AHL molecules and I70V shows a similar sensitivity compared to wild-type.



Figure 17: Fluorescence intensity measured by plate reader (after 7 hours' induction).



Figure 18: Fluorescence intensity measured by flow cytometry. The samples are taken after 7 hours' induction.

go to top

[1] Bredow, M., Sementchoukova, I., Siegel, K., & Monaghan, J. (2019). Pattern-triggered oxidative burst and seedling growth inhibition assays in Arabidopsis thaliana. JoVE (Journal of Visualized Experiments), (147), e59437.

[2] Bahar, Ofir, et al. "Bacterial outer membrane vesicles induce plant immune responses." Molecular Plant-Microbe Interactions 29.5 (2016): 374-384.

[3] Keel, C., Schnider, U., Maurhofer, M., Voisard, C., Laville, J., Burger, U., ... & Défago, G. (1992). Suppression of root diseases by Pseudomonas fluorescens CHA0: importance of the bacterial secondary metabolite 2, 4-diacetylphloroglucinol. Molecular Plant-Microbe Interactions, 5(1), 4-13.

[4] Steidle, A., Allesen-Holm, M., Riedel, K., Berg, G., Givskov, M., Molin, S., & Eberl, L. (2002). Identification and characterization of an N-acylhomoserine lactone-dependent quorum-sensing system in Pseudomonas putida strain IsoF. Applied and Environmental Microbiology, 68(12), 6371-6382.