Our project aims to inject antiflorigens into the roots of fruit trees using Pseudomonas fluroescens. For this reason, one of the tasks of our project was to investigate the capability of P. fluorescens to attach to the roots of the model plant Arabidopsis thaliana, and release a superfolder Green Fluorescent Protein (GFP) equipped with a Type-III Secretion System-specific secretion signal into the roots.

Round 1: Preliminary studies on colonisation behaviour of P. fluorescens

Design 1:

Our experimental design includes the following steps:

1. Growth of A. thaliana in Fahraeus slides
2. Addition of GFP-secreting P. fluorescens to the plants.
3. Fluorescence microscopy on living plants and cross-sectioned roots.

Build 1:

We want to study the potential of P. fluorescens to inject GFP into roots of A. thaliana. To this end, GFP with a T3SS specific secretion signal was introduced to pSEVA64 vector via Golden Gate cloning and transformed into P. fluorescens to create the plasmid shown in Figure 1. RFP is expressed as a second protein under the control of the same promoter, however without a secretion signal. pSEVA64 was shown effective in Pseudomonas putida, a close relative to P. fluorescens, and was therefore a promising candidate for integration in P. fluorescens [1]. The secretion sequence of the T3SS was adopted from as Jansen et al. (2022) [2].

1. Negative control: GFP is produced but cannot be secreted due to the lack of a T3SS secretion signal.
2. Positive control: gfp was placed under the control of a functional transcriptional promoter + RBS combination sourced from a previously published [3].
3. Test sample: gfp is expressed under the control of a strong promoter (Part: BBa_J23100), in the expectation that this leads to high protein secretion into plants.

Figure 1: Plasmid map of pSEVA64 for integration intoP. fluorescens (test sample). GFP is coupled T3SS-secretion signal under the control of the strong constitutive promoter BBa:J23100. RFP is expressed to counterstain the bacteria, however, RFP has no secretion signal and can therefore not be secreted.

Test 1:

A. thaliana was grown on Fahrraeus slides in 1/2 MS medium (see plant growth protocol). After 6 days, the roots were approximately 2 cm long and had developed their main structures, including the differentiation and elongation zones, as well as lateral roots and root hairs, which was enough to study the basic attachment behaviour of our bacterium.

We conducted a 7-day experiment to investigate bacterial attachment on ten different roots. Inoculation was performed by adding the bacteria samples to the roots, incubating for 1 hour, and then washing the roots to remove unattached bacteria. On the initial day, following inoculation, the bacteria were uniformly distributed across the entire root (Figure 1B).

However, in the subsequent days, it became evident that they exhibited a higher propensity for colonisation of the basal zone of the root, which corresponds to the differentiation zone near the point where the stem initiates, with only a minimal presence of bacteria in the apical region of the root (Figure 1C-D). An overview of different root regions can be found in Figure 1A.

The analysis of the cross-section was hindered by the inability of the vibratome to section the unfixed plant tissue. The tissue was too soft, and bent away from the knife, making the sectioning impossible.

Figure 2: Overview of tissues and root regions in the root system ofA. thaliana and microscopic pictures ofP. fluroescens attached to root ofA. thaliana (100x magnification). A. Overview picture; B:P. fluorescens is evenly attached to the root of A. thaliana (day 0); C: Fluorescence microscopic and merged channel picture ofP. fluorescens attached to the elongation zone ofA. thaliana (day 2); D: Increasing colonialisation towards root-basal region, with a decreased number of bacteria at apical site (day 1).

Learn 1:

First, we gained the knowledge that bacteria mainly colonise the basal zone of the roots where the roots are mature and closer to the surface. This observation cross validates our agent-based model which posits that most bacteria colonise just beneath the soil surface.

Second, we learned that we could not detect GFP secretion into the plant by just looking on the surface of the plant. To get insights of how GFP is potentially secreted into the plant, cross-sections were needed as described below in the Round 2.

Third, from discussions with the plant microscopy expert Norbert de Ruijter, we learned that fixation of the plant material might lead to the fluorescence quenching of GFP. However, we learned that is inevitable to make proper cross-sections of the plant roots with the vibratome.

Round 2: Protein secretion detection on root-cross sections

Design 2:

In this iteration, we improved our design with a fixation step and implemented learnings from round 1. Roots were fixated for 1 h at room temperature in 4% paraformaldehyde, making the root stiffer and allowing proper cross-sectioning of the roots. Assuming that all bacteria secrete similar amounts of GFP into the plant, we anticipated that in densely colonised parts of the root more GFP is secreted into the roots compared to rarely colonised parts such as the root tip.

Build 2:

No further bacteria needed to be built for this round. The used bacteria include the above mentioned GFP-producing negative, positive, and test samples.

Test 2:

After fixation, roots were cut at the beginning and end of the differentiation zone (Figure 3). Cut roots were embedded in the improved embedding medium (6% agarose gel instead of 4%). The embedded roots were directly sectioned on the vibratome (25-50 µm thick sections) and analysed on a fluorescence microscope.

This time, utilizable cross-sections of the root could be made. Under the fluorescence microscope, we detected that the thickness of 50 µm was slightly too thick, which complicated the detection of internal root structures. Reduction to 25 µm allowed an optimal visualisation of main structures.

Finally, we could visualise attached bacteria and investigate whether they could secrete visible amounts of GFP into the plant. As it can be observed in Figure 4B, GFP-producing bacteria can be visualised attaching to the epidermis. Further, green fluorescent structures are visible in the core of the plant. No evident accumulation of green fluorescent particles is observed in the epidermis. The negative control shows a cross section of an A. thaliana root that was not grown in the presence of bacteria. The negative control shows similar patterns of green fluorescence in the cell wall of the epidermis and the stele (Figure 4D).

Figure 3: Tissues and regions ofA. thaliana root system. For vibratome sectioning, roots were cut below the shoot and above the elongation zone, because most bacteria were detected in the intermediate zone.

Figure 4: Phase contrast-(left) and fluorescence microscopy (right) picture ofA. thaliana root cross sections (thickness: 25 µm) A+C: Phase contrast pictures, 20x magnification. Main structures of the root are visible such as root hairs, epidermis, cortex and the slightly brown stele. B: Fluorescence microscopy picture 1, bacteria are visible as green dots surrounding the epidermis, no major accumulation of GFP is visible inside the epidermis. Green fluorescent structures are visible in the stele. D: Fluorescence microscopy picture 2, autofluroescent structures in epidermis and stele.

Learn 2:

The findings of this experiment provide first insights into the injection of proteins from P. fluorescens into roots. However, it has to be critically questioned how GFP might have accumulated in the core of the root while no GFP is detected at the epidermis. From literature we know that oxidative stress can lead to the production of secondary metabolites in plants, which can result in autofluorescence at similar wavelengths to GFP [4, 5]. Therefore, we refrain from making the statement that the fluorescence at the stele is GFP-derived.

In a future experiment, one might consider coapplying a reducing agent like DTT to the fixation procedure to lower the quantity of oxidative metabolites to obtain results that leave less scope for interpretation. Furthermore, one could increase the sensitivity of GFP detection by using GFP-specific antibodies (immunostaining). These modifications would not only enhance the accuracy of the findings but also serve as a validation of the nature of observed fluorescent structures, such as those seen in the root's stele (Figure 4), confirming whether they are GFP or of plant origin.

Conclusion:

Within two iterative cycles of the DBTL, we improved the methodology for detecting GFP originating from Pseudomonas fluorescens in plant roots. First, we had a broad overview of where bacteria attach and tried to already detect injected GFP in the epidermis and root hairs.

Next, we enhanced the procedure for detecting intracellular GFP by analysing cross-sectional samples from root regions where we estimated a high bacterial presence, particularly in the basal portions of the roots.

In the final phase of optimisation, we refined the cross-sectioning procedure for our specific research objectives. This refinement involved the introduction of a root-fixation step and an increase in the concentration of agarose within the embedding medium. These improvements were essential to increase the precision and reliability of our detection method.

As described in the Results section, our biocontainment strategy is based on knocking out an essential gene from the genome of P. fluorescens and introducing it to the cell on a plasmid under the control of an inducible expression system. The envisioned knockout procedure consists of two main parts, the integration of a suicide plasmid, and the subsequent deletion of the gene of interest. Both parts were subjected to DBTL cycles which are presented below. To find an Overview of this method together with the methods used visit our Results page.

Part I - integration of a suicide plasmid into the genome

Round 1: Electroporation

Design 1:

Based on literature research, we chose dapB as suited candidate to knock out in our P. fluorescens host for our biocontainment strategy. dapB is an essential gene encoding for dihydrodipicolinate reductase, an enzyme that plays a key role in the biosynthesis of lysine and diaminopimelate (DAP) which is an essential building block of peptidoglycan of all gram-negative bacteria [6]. Strains devoid of dapB are therefore auxotrophic for both compounds (DAP and lysine). This gene has been knocked out before by Gal et al. (2003) using SOE-PCR and two-step allelic exchange [7]. However, in this project, we chose a slightly different approach which was proposed by Wirth et al. (2020) as a method to modify Pseudomonas putida, a close relative of P. fluorescens [8].

Build 1:

We started by constructing a suicide plasmid based on the pGNW backbone and containing homology arms needed for a first recombination event to happen. The plasmid was assembled via Golden Gate and transformed into E. coli DH5α λpir via chemical transformation. The correctly assembled plasmids were purified and used to transform P. fluorescens via electroporation.

Figure 5:E. coli DH5α λpir containing pGNW-dapB-KO. Fluorescence of the cells, resulting from the expression of agfp transcriptional unit present on pGNW, suggests correct plasmid assembly.

Test 1:

The bacteria were plated on LB with kanamycin, however, despite many tries, no colonies were observed.

Learn 1:

The most likely reason explaining why no transformants were obtained was the failure of electroporation to deliver the suicide plasmid to P. fluorescens. Therefore, we decided to try a different method of transformation

Round 2: Tri-parental mating

Design 2:

We chose another approach proposed by Wirth et al. (2020) [8] which is based on tri-parental mating, a form of conjugation in which a helper strain facilitates the transfer of a plasmid between two other bacteria. This method uses E. coli DH5α λpir containing the pGNW-based plasmid as a donor, P. fluorescens as an acceptor, and E. coli HB101 as a helper.

Build 2:

We followed the procedure of tri-parental mating according to the protocol.

Test 2:

We managed to obtain fluorescent P. fluorescens colonies indicating successful transformation. This was further confirmed by colony PCR and subsequent electrophoresis which showed bands of the expected size indicating integration into the genome

Figure 6: A:P. fluorescens colonies after transformation with the pGNW-based plasmid showing fluorescence. B: Agarose gel visualising the results of colony PCR ofP. fluorescens after transformation with pGNW. Two bands of the expected size proved that the plasmid was successfully incorporated into the genome.

Learn 2:

Tri-parental mating turned out to be a successful method of delivery of a suicide plasmid to P. fluorescens SBW25. A single recombination event resulted in the incorporation of pGNW into the genome.

Part II - creating the knockout

Round 1: DAP lysine supplementation from the medium

Design 1:

In order to knock out the gene of interest, P. fluorescens containing pGNW must be transformed with a pQURE6-H plasmid encoding I-SceI endonuclease. The enzyme recognises I-SceI sites flanking the homology arms introduced with the suicide pGNW plasmid and cuts the DNA. These double-strand breaks can be fixed by the cell by a second recombination event which can either lead to restoration of the wild-type genotype or a knock-out.

Build 1:

We delivered pQURE6-H via chemical transformation, a method with which other teammates had successfully introduced different plasmids into P. fluorescens in the meantime. The transformants were plated on LB with the addition of DAP and lysine in the concentrations determined by Gal et al., as well as 2 mM toluic acid to induce the expression of the meganuclease I-SceI.

Learn 1:

Red colonies appearing on the plates indicated that chemical transformation successfully delivered pQURE6-H to the cell because a red reporter unit is cloned into this plasmid’s backbone (Figure 7). However, colony PCR proved that all the tested colonies had a wild-type genotype (Figure 8). Despite several trials, the result was always unsuccessful.

Figure 7: Red colonies ofP. fluorescens after transformation with pQURE6-H.

Figure 8: Agarose gel visualising the results of a colony PCR after transformation ofP. fluorescens with pGNW incorporated into the genome with pQURE6-H. The bands correspond to the length of 1850 bp, which means that the gene of interest was not deleted.

Learn 1:

The inability to obtain a knockout despite the successful transformation with pQURE6-H was intriguing. After analysing all the steps, we concluded that the most likely reason was the fact that the cells with successful deletion of dapB could not survive in the medium used due to insufficient concentrations of DAP and lysine.

Round 2: overexpression of dapB

Design 2:

Therefore, we redesigned our approach. Rather than supplying DAP and lysine from the medium, we decided to transform the cells with the overexpression plasmid first, and then carry out the knockout procedure. Since before the transformation with pQURE6-H, the cells would already contain the overexpression plasmid, we decided to change the method of pQURE6-H delivery. Instead of using chemical transformation, we chose to deliver it via tri-parental mating, which had proven to be more efficient.

Build 2:

We constructed the overexpression plasmid based on the pSEVAb33 backbone with dapB placed under the control of the strong constitutive promoter J23100 (BBa_J23100). After confirmation of the correct assembly, the plasmid was delivered to P. fluorescens containing pGNW via chemical transformation. Afterwards, the cells were transformed with pQURE6-H via tri-parental mating using P. fluorescens with pGNW incorporated into the genome as an acceptor, E. coli DH10β containing pQURE6-H as a donor, and E. coli HB101 as a helper strain.

Test 2:

The cells were successfully transformed with the overexpression plasmid, which was checked by colony PCR. Transformation with pQURE6-H was successful. However, after induction, the colony PCR once again indicated that all the bacteria had a wild-type genotype.

Learn 2 and final conclusions:

In the end, we did not manage to knock out the dapB gene. One possible reason is that the I-SceI meganuclease was not activated properly or was inactive. In that case, however, the genome would not be cut at all, and the suicide plasmid would still be in the genome. This was however not the case. The fact that we could not obtain a knockout even when introducing dapB on a plasmid suggests that the genomic copy remained essential for the survival of the cell. One possible explanation is that the high levels of enzyme resulting from overexpression were too burdensome for the cells. This could force the plasmid to mutate making it unable to sustain the growth of cells with a knockout.

Therefore, the next step of this cycle would be to sequence the overexpression plasmid. If the results indeed showed the presence of mutations affecting dapB expression, we would repeat the procedure described in Part II Round 2 but using a promoter and RBS that give lower expression levels.