Team:Wageningen UR/Demonstrate



Demonstrate Header

Many separate parts of Xylencer have been successfully created. However, we have not been able to complete Xylencer as an assembled product. We did show the functionality of single and combined parts and present the foundation for meaningful follow-up research regarding Xylella fastidiosa detection and eradication as well as phage therapy in general. On this page you will find a coherent summary of all our scientific achievements. While working hard on the implementation of Xylencer we did comply with all rules and policies approved by the iGEM Safety Comity and the Dutch government.


Detection symbol

Our detection tool is based on insect vectors carrying Xylella fastidiosa being attracted by and feed on our plant mimic. In this way, the pathogen is being released into the reaction tube of the plant mimic, the same way it would be released into the xylem of plants. Once in the tube, detection of X. fastidiosa via LAMP reaction takes place. We showed crucial steps of the Xylencer automated detection tool to be functional.

First, we performed a transmission assay and demonstrated the transfer of bacteria to our plant mimic via insects. Two plant mimics were placed in containers, one of which one was spiked with E. coli carrying GFP and chloramphenicol resistance (Figure 1). Insects were introduced into the containers and allowed to feed for two days. Afterwards, the content of the non-spiked plant mimics and the insects were sampled and plated on LB Cam15 plates (Figure 2). The identity of colonies was confirmed by GFP fluorescence signal. In this way, we demonstrated that the insects acquire the bacteria and release them into the plant mimic during feeding. Successful transmission of the bacteria in our device represents the cornerstone of our automated detection tool, which will render sampling by hand obsolete. Transmission rates are expected to be higher in the final application as the concentration of X. fastidiosa on the insect is likely to be higher than that of the tested bacteria. [1].

Figure 1: Plating results of a transmission assay performed with Cicadella viridis and irradiated with UV light to show GFP. Left: plating of plant mimic feeding solution spiked with E. coli carrying GFP and chloramphenicol resistance. Middle: plating of two insects samples after two days of feeding on spiked plant mimic. Right: Plating of the corresponding non-spiked plant mimic feeding solution.

Second, we determined the limit of detection (LOD) of our detection tool. The in-field application of our detection device relies on applying LAMP on crude samples instead of in-laboratory prepared genomic DNA. By testing our system on E. coli cultures, we show that the LAMP reaction temperature of 65 °C was sufficient to lyse the cells and allow LAMP primers to interact with the DNA. Furthermore, we could determine the LOD of our approach to be 100–1000 cells. These experiments were crucial, as the in-laboratory extraction of genomic DNA would interfere with the automation principle of our detection tool. By determining the LOD for our LAMP mediated cell lysis approach, we came a step closer to automated detection of X. fastidiosa.

Figure 2: Two plant mimics, of which one was spiked with E. coli carrying GFP and chloramphenicol resistance, were placed in containers. Insects were introduced into the containers and allowed to feed for two days.

Due to safety reasons, we performed no direct X. fastidiosa experiments in our facilities. However, to test our system on X. fastidiosa itself, we collaborated with the Dutch Food and Safety Authority (NVWA). Even though, in samples taken by us personally (Figure 3, Lane 1-4), X. fastidiosa could not be detected, we did detect X. fastidiosa in a sample taken by the NVWA. By sampling crude leave extracts of infected plants we proved the functionality of cell lysis and detection of X. fastidiosa via LAMP. This way we could prove the functionality of our system on the pathogen itself. For Xylencer, this represents a major step regarding in-field testing for pathogen presence as no in-laboratory sample preparation was needed. Furthermore, the fact that samples of the same diseased plants show conflicting outcomes strengthens our strategy to sample insects on a large scale instead of individual plants.

Figure 3: LAMP reaction product of plant extracts visualized by gel electrophoresis. Positive samples show a smear, as LAMP produces amplicons of different sizes. Sample 1 & 2 and their 1:10 dilutions derive from samples of X. fastidiosa infected plants. Positive control 1 is a pure culture, positive control 2 genomic DNA of X. fastidiosa. The negative control was extracted from uninfected plants. S1 PE represents a previously extracted sample from the same plant as in Sample 1 (sampled by NVWA employee).


Delivery symbol

The functionality of the Acr-dCas9 gene circuit is crucial for Xylencer in many ways. It ensures the production and release of phages at the right time and location and is therefore fundamental for the effectiveness of phage therapy. Furthermore, it is essential to limit the release of engineered phages to infected plants and fields.

Important for the regulation of the gene circuit is the riboswitch controlling the expression of the Acr. Figure 4 demonstrates the control of GFP expression by the riboswitch in two different systems. The pSEVA23 system shows no fluorescence signal when compared to the negative and positive control. To see whether this effect was due to the gene being unfunctional or the riboswitch controlling its expression tightly, mRNA extraction followed by cDNA synthesis and PCR were performed. Figure 5 shows that even though no GFP is visible, GFP transcripts are present. Therefore, we can conclude that even though transcription of GFP takes place, GFP translation is inhibited by the riboswitch. In this way, we demonstrated the robustness of riboswitch mediated translation inhibition, providing a good foundation for the control of the Acr-dCas9 gene circuit.

Figure 4: Amplification of cDNA with GFP specific primers. The used ladder was 1 Kb+ ladder from NEB. The expected band should be 207 bp. (1-3) pSEVA23 1 Plasmid system. (4-6) pSB1C3 samples with the Vc2 riboswitch + GFP. (7-9) Positive control contained a constitutively expressed GFP. (10-12) Negative control, which is an E. coli Dh5Alpha strain. (13,14) Two positive colony PCR controls with engineered with a GFP sequence. (15) Negative colony PCR control. (16) Positive control on an extracted plasmid containing a GFP sequence.
Figure 5: The four strains tested for GFP transcription were pSEVA23 (top left), pSB1C3 (top right), positive control (bottom left) and a negative control (bottom right). E. coli was grown on LB agar and not induced with DSF. Fluorescence is visualized by a blue light transilluminator.

By constructing the AcrIIA4-dCas9 gene circuit, we could demonstrate two things. First, we were able to repress gene expression significantly by targeting the phage Lambda early transcripts regulatory region (Figure 6). Second, we managed to restore the expression from phage Lambda promoters by induction of Acr expression. The provided library of RBS sequences, characterized for their function in the presented gene circuit, allows for the control of gene expression in different scenarios. Further applications could include phage therapy delivery for other pathogens affecting plants, humans and the biotechnology sector. In this way, we delivered the basis for the controlled phage release, not just for the Xylencer project, but also for future applications involving targeted phage proliferation in chassis organisms and control of gene expression in general.

Figure 6: (Acr-)dCas9 targeting both GFP & mRFP. dCas9 targeting GFP & mRFP expression, either with or without AcrIIA4. Spacers are targeting the early operon regulatory region of phage Lambda controlling gfp & mrfp expression. Different shades of green (GFP) and orange (mRFP) represent different amounts of L-rhamnose added to induce AcrIIA4 expression. E. coli DH5alpha served as a negative control and was subtracted from samples. Fluorescence is relative to (Acr-)dCas with a non-targeting spacer. Data represents the averages of three biological replicates.

To ensure our solution is safe, we required a non-pathogenic bacterium with a high similarity in cell metabolism to X. fastidiosa. Xanthomonas was selected as the most promising candidate, but the genus consists mostly of pathogens, leading us to wonder if characterized non-pathogens might be opportunistic pathogens in reality. We wanted to reinforce their status as non-pathogens by finding a genetic basis for non-pathogenicity. To investigate if such a basis was present, we constructed a dataset of 1372 Xanthomonas genomes and provided them with high-quality annotations. We complemented this dataset with a manually curated dataset of 104 Xanthomonas genomes, of which (non-)pathogenicity was experimentally confirmed. These datasets were used to train a machine learning model, that was able to predict non-pathogens with a 90% ± 0.12 (SD) sensitivity. From this high-performance model, we inferred that both a lack of a type III secretion system and a lack of specific transposons are key for non-pathogenicity. With this information, we were able to select Xanthomonas arboricola CITA 44 as our phage delivery bacterium.

Figure 6: Principal component analysis of the average model performance. Solid dots represent Xanthomonas genomes that are correctly predicted by 95%+ of the models. Color indicates pathogenicity. We can observe two clusters of non-pathogens, that are reliable predicted as such, these compose the group of candidates for our phage delivery bacterium.


Remediation Symbol

The pillar Remediation focuses on enhancing the efficiency of the Xylencer phage therapy by alerting the plant to the presence of the pathogen. This is achieved by releasing PAMPs into the Xylem upon phage mediated lysis. In this way we provide an extra means of combatting X. fastidiosa.

In this pillar, we demonstrated that triggering the immune response of plants is a successful method to treat diseased plants. Xylem inoculation of flg22 significantly decreases symptoms seen in leaves infected with Xanthomonas (figure 7). Even though this effect is temporary, it proves the functionality of the Xylencer approach to trigger a systemic plant immune response. The temporary effect does not pose a major problem, as within Xylencer flg22 will be supplied as long as X. fastidiosa cells are being lysed.

Figure 7: The lesion size of leaves of Brassica oleracea on day 3, 4 and 5 after inoculation with flg22, compared to a control (H2O).


Spread Symbol

An important part of Xylencer is the self-spreading phage, that enables the therapy to autonomously apply itself to neighbouring plants. This enables the Xylencer phages to reach plants that would otherwise evade treatment. Spreading of the phages is achieved by engineering the phages with proteins that bind to the same insect vectors as X. fastidiosa.

The attachment to the insects is mediated via chitin-binding proteins, which adhere to the insect stylet. The insect uses the stylet to feed on plants. We demonstrated that PD1764, an adhesion protein derived from X. fastidiosa, shows high binding abilities to chitin as compared to a bovine serum albumin (BSA) control (Figure 8). Since PD1764 is a chitin binding protein from X. fastidiosa it shows the potential of this protein for effective transmission of Xylencer phages.

Figure 8: PD1764sh and BSA were incubated with and without chitin for 1 h at constant agitation. After centrifugation, the protein concentration in the supernatant was determined using a Bradford assay. The protein concentration of the sample with chitin was corrected for the concentration of the sample without chitin, to calculate the percentage of protein that bound to chitin.

In a next step, we generated constructs encoding for the fusion of phage Lambda capsid protein gpD to our adhesion protein candidates. As the phage capsid integrity can be negatively influenced by high amounts of capsid fusion protein [2], we decided to integrate a ribosomal frame shift (RFS) site between the gpd and adhesion protein coding sequence. The RFS should create a mix of WT and fusion protein. By western blot we could confirm the transcription and translation of our fusion proteins in E. coli (Figure 9). The first lane shows the expected fragment size of the gpd and adhesion proteins connected by a flexible linker (CGSGSGSG). In the second lane, single gpD and gpD fusion protein with RFS sites are visible. In this way we demonstrated the successful fusion of our adhesion protein candidates to a phage capsid protein.

Figure 9: Western blot of the lysates of the bacterial cells transformed with: 1) the plasmid containing the gpd + chitinase a1 fusion gene with a linker (14.65 kDa), 2) the plasmid containing the gpd + chitinase a1 fusion gene with a RFS sequence (16.8 kDa without GFP fusion, 13.4 kDa with GFP fusion).  Ladder: PageRulerTM Prestained Protein Ladder. 

With the adhesion assay (figure 8) we could demonstrate the binding ability of our protein to structures found in the insects stylet. Through the western blot we show the fusion of adhesion proteins to phage capsid proteins. Together, these successes represent two major steps towards implementing the Xylencer phage spreading mechanisms via X. fastidiosa insect vectors.

Finally, we produced a model describing the spread of Xylencer phages via insects and their effectiveness in eradicating X. fastidiosa. We assumed a 10 x 10 km wide area and trees placed 10 m apart from each other. Upon inoculation of three individual trees at a distance of 3.3 km from each other, Xylencer phages spread and eradicated X. fastidiosa populations within 40 days in an area of 100 km2 (Figure 10). The modeled spreading ability shows the efficiency Xylencer would have in combating X. fastidiosa. However, the rapid and efficient spread also brings up biosafety issues which we address on our Biosafety page.

Figure 10: Animation of the effectiveness of the Xylencer phages over time. Phages were applied in a field infected with X. fastidiosa. The scales show bacterial load in cells ml-1 and the number of insects carrying X. fastidiosa per m2.
  • References arrow_downward
    1. Labroussaa, F., Zeilinger, A. R., & Almeida, R. P. (2016). Blocking the transmission of a noncirculative vector-borne plant pathogenic bacterium. Molecular Plant-Microbe Interactions, 29(7), 535-544
    2. Pavoni, E., Vaccaro, P., D’Alessio, V., De Santis, R., & Minenkova, O. (2013). Simultaneous display of two large proteins on the head and tail of bacteriophage lambda. BMC biotechnology, 13(1), 79