Introduction

Our strain will deliver plant hormones through a type III secretion system (T3SS). Since both are heterologous to our strain, they impose an energy burden. Constitutive activation of the T3SS and production of anti-florigens only increases this burden, affecting survival in the biosphere in comparison to native strains and species, and decreasing the efficiency of our protein delivery. To minimize these effects, we set out to create a novel biosensor to enable control of protein production and delivery in a condition-dependent manner, rather than constitutively. The biosensor was developed following the next stages:

Stage 1. Root proximity: Our strain can only effectively deliver proteins to the roots when in close contact with them, making root proximity a crucial permissive condition. The detection of root exudates can signal proximity to the roots. Therefore, inducible expression systems sensitive to different root exudates were introduced in our strain to measure their responses.

Stage 2. Quorum Sensing: When our bacteria colonize the rhizosphere in large numbers, they can deliver more proteins and remain more protected from the environment by means of biofilms. Thus, we want to use high colony density as our second permissive condition for protein production. This will also ensure that if colony size decreases, protein production stops, increasing survival chances in the rhizosphere. To achieve this, we introduced the LuxPR/LuxR quorum sensing module in our strain.

Stage 3. Toehold switch: We want our strain to activate protein production and delivery only when both permissive conditions are present. This integration can be achieved with genetic AND gates such as toehold switches. We will test whether our strain can use this mechanism to activate the expression of the PseuPomona delivery molecules (T3SS and plant hormones).

Stage 1: Root proximity – Taking PseuPomona back to the roots

PLATE READER FLUORESCENCE ASSAYS

Our first goal was to make P. fluorescens able to sense root exudates. To this end, we created four reporter strains, each transformed with a plasmid containing different root exudate inducible expression systems. We then performed a fluorescence assay using a plate reader, testing our strains over a wide range of root exudates to observe their response in biological triplicates.

Figure 1: Corrected fluorescence (GFP/OD₆₀₀) of different root exudate inducible systems in P. fluorescens after 24 hours of exposure to a range of their corresponding inducer. The following systems were used: pBAD/AraC induced by arabinose, pCym/CymR induced by cuminic acid (B), pSal/nahR induced by salicylic acid (C), pTtg/TtgR induced by naringenin.

All of our reporter strains have higher gene activation at higher root exudate concentrations. The results show that the inducible expression systems that we chose are suitable for sensing root exudates. The response curves of pSal/nahR and pCym/CymR show the highest dynamic ranges, both > 9-fold, across a narrow concentration range, which result in highly tuneable switch-like responses to their corresponding root exudates.

IN SITU CHARACTERISATION USING ‘FRUIT’

Our living root exudate biosensors were now characterized and fully functioning. After successfully testing them with a wide range of different root exudates, we wanted to assess their performance with plant roots in situ. pCym/CymR and pSal/nahR showed great performance but since we found the former more suitable for the biosafety part of our project, we decided to proceed with pSal/nahR.

Figure 2: Corrected fluorescence response (GFP/OD₆₀₀) of FRUIT-measured image analysis after 24 hours of P. fluorescens bacteria with pSal/nahR inducible expression system (biological duplicates) cultivated on A. thaliana roots. “Root” samples were exposed to A. thaliana root fragments (Root), as well as a positive control with 1 mM of salicylic acid (+) and a negative control of only M9 medium (-). *** represents a student t test p-value < 0.001

Fluorescence plate reader assays are effective for basic part characterisation. However, they cannot be used to assess how the pSal/nahR inducible expression system in P. fluorescens will respond when the bacteria are attached to live roots. We used FRUIT, our novel fluorescence microscopy image processing protocol, to measure fluorescent output of the modified bacteria growing on live roots. The results show that the bacteria exposed to the roots are more fluorescent than roots exposed to a negative control (roots fixed with 4% paraformaldehyde).

Overall, we have successfully characterised four root exudate inducible expression systems using plate reader fluorescence assays. We even went as far as testing pSal/nahR in situ. This showed us that the system is able to sense root attachment and is therefore a promising candidate to build our biosensor.

Stage 2: Quorum sensing- the more the merrier

With our P. fluorescens strain now able to detect proximity to plant roots, we wanted PseuPomona to detect when it reached a certain colony size threshold. Some bacteria are equipped with quorum sensing, the ability to respond to population density. We introduced the quorum sensing system from Vibrio fischeri into PseuPomona as it is one of the most well-characterised. The system works as follows. The LuxI enzyme produces acyl-homoserine lactone (AHL) quorum molecules. When AHL is present, it binds the LuxR transcription factor which then in turn activates the Lux pR promoter, enabling transcription. AHL is secreted by bacteria but only at high densities does the concentration increase enough to bind LuxR.

Figure 3: A: Schematic repesentatino of Lux pR/LuxR quorum sensing module. B: Bar chart indicating the response (GFP/OD₆₀₀) of luxpr/luxR to increasing concentrations of AHL in P. fluorescens measured in corrected fluorescence (GFP/OD₆₀₀).

We introduced the luxR gene and its corresponding promoter controlling the expression of gfp. We then tested the response to increasing concentrations of AHL by means of fluorescence plate reader assays. Our results show that our reporter strain is sensitive to increasing concentrations of AHL. In practice, the gradual increase of AHL represents increasing cell density. When considering this, it becomes clear that our strain is able to effectively detect an increase in colony size and activate gene expression as a result.

Stage 3: Toehold - a key and a lock

We now have established that our bacterium is able to detect root proximity and an increase in colony size. Still, we do not want to start protein production and delivery when PseuPomona receives any of these inputs individually. Since we only want activation when both inputs are present, we implemented a genetic AND gate in the form of an RNA toehold switch to integrate the two signals. We used a toehold switch since they show great orthogonality and programmability. In addition, since they are RNA-based, they pose a low energy burden while offering a high dynamic range [1].

Figure 4: Schematic illustrating the functional parts of a toehold switch and its working mechanism. Adapted from [1]

RNA toeholds switches consist of two parts: the switch RNA and the trigger RNA. The switch RNA can be thought of as the lock. It contains a hairpin loop structure that prevents the ribosome from translating the RNA. This structure blocks the translation of the gene located downstream. The trigger RNA can be thought of as the key. It is a short RNA fragment that is complementary to the RNA that forms the hairpin structure. Only when the key (trigger RNA) is present can the lock open (switch RNA) allowing for gene activation to take place (Figure 4).

Figure 5: Responses (GFP/OD₆₀₀) of toehold with pSal/nahR lux pR/LuxR paired to trigger RNA and switch RNA respectively to salicylic acid and AHL both separate and combined plotted over time (A) and after 24 hours (B). *** represents a student t test p-value < 0.001, n.s. represent a student t test p-value > 0.05.

We used our salicylic acid inducible expression system to control the expression of the trigger RNA and our quorum sensing inducible expression system to drive the expression of the switch RNA on separate plasmids with similar copy numbers. We then exposed our strains to 150 µM of salicylic acid and 0.5 nM of AHL, separately and together. Our results clearly show that our strain is responsive only when both inputs are added (Figure 5 A, B).

Overall, these findings should translate to more specific activation of protein production and delivery by our strain, which in turn should lower the energy burden.

Intoduction

After proving that proteins can not only be produced in low temperatures (Wetlab-colonisation page) but also can be made induced by root exudates and quorum sensing, we wanted to have a look at how proteins are injected in plant roots.

From the literature we know that the etHAn strain of P. fluorecens that we kindly received from Copenhagen University is able to inject toxic effector proteins into leaves of crop species like wheat and maize to elicit a hypersensitive plant response, i.e. the apoptosis of plant cells around the injection site [1]. Now, we want to study the potential of P. fluorescens to inject GFP into plant 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 etHAn (see Figure 1). RFP is expressed as second protein under the same promoter, however without a secretion signal, as a control to.... pSEVA64 was shown effective in Pseudomonas putida, a close relative to P. fluorescens and was therefore a promising candidate for integration in P. fluorescens [2].

Figure 1: Plasmid map of pSEVA64 for integration into P. fluorescens. 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.

Protein detection in the living plant root

From consultations with the plant microscopy-expert Norbert de Ruijter (see attribution site for detailed information) we were advised to first study if we can see injection of GFP from attached bacteria to the epidermis, particularly at root tissues like lateral root entry points and root hairs. Hereby, we make use of an advantageous property of the roots from A. thaliana, which remain transparent when grown in 1/2 MS medium (Figure 2). This allows the detection of intracellular GFP without damaging the roots. Even inner root structures like the stele where xylem and phloem transport takes place can be monitored.

Figure 2: Phase contrast picture of A. thaliana Col-0 root, one week after germination. The transparent cells allow the easy study of epidermal cells with root hairs as well as an emeriging lateral root and the internal stele.

Having this in mind, we wanted to answer the following questions:

Is enough GFP secreted into epidermal cells to see accumulated green fluorescence in root hairs and the epidermis?

Does injected GFP reach the phloem and travel up to possible sink tissues in the apical shoot region?

First, we had a closer look at root hairs, as they seemed to show less autofluorescence than the main root. This characteristic made them suitable to investigated the presence of intracellular - thus secreted - GFP. GFP producing, and potentially secreting, bacteria attach to root hairs (Figure 3). However, no visible accumulation of GFP inside the root hairs is detectable. The stem shows generally more fluorescence, which supports the hypothesis that GFP was secreted into epidermal cells. However, when we look at the same region in with settings adjusted to the RFP without secretion signal, similar patterns of fluorescence are detected in the root (Figure 3 B) . These findings give rise to the hypothesis that the plant produces autofluorescent substances that are nonrelated to the potentially secreted GFP from our engineered bacterium .

Figure 3: Fluorescence microscopic picture ofP. fluorescens attaching to root hair orA. thaliana (1 day after first attachment). A: GFP producing, and because of the secretion signal also potentially secreting, colony of P. fluorescens attachted to root hair (exposure time 1sec, ext: λ=485, ems: 510); B: Same picture showing fluorescence settings for RFP (exposure time: 3 sec, ext: λ=555, ems: 583). Both pictures show similar fluorescent patterns, GFP and RFP expressing bacteria are fluorescent, the plant tissue is fluorescent to a lower extent.

Conclusion 1

We developed a novel approach to investigate the protein injection via the T3SS into plant root cell. Therefore, we used an in vivo fluorescence microscopic approach in which undamaged, living roots could be analysed under the fluorescence microscope to detect the live secretion of GFP from P. fluorescence into A. thaliana. We got insights of how bacteria attach on the hair roots but could not make qualitative statements on the possible secretion of GFP into the cells.

Protein detection in fixated vibratome sectionings

To obtain more informative results regarding the actual transfer of GFP from bacteria to plant cells, we opted to create vibratome sections, which are 25-50 µm slices , of the basal root regions that were coated with bacteria. Using cross-sectional and longitudinal sections, we aimed to closely examine the internal structures of the plant, which would enable us to detect GFP that may have been injected into the plant.

For a detailed description of the experimental procedure and our findings, have a look at our Engineering site.

Both the fluorescence and the multichannel images in Figure 4 depict green P. fluorescens colonies adhering to the root epidermis of A. thaliana. However, it can be noted that there is no observable secretion of GFP from the bacteria into the epidermis, nor is there any evident accumulation of green fluorescent signal within the phloem of the root.

Figure 4: Longitudinal vibratome section ofA. thaliana root with GFP producing P. fluorescens (thickness: 50 µm). Left: Multi-channel picture with green bacteria attached to root-epidermis. Right: fluorescence microscopy picture.

Cross-sections of the roots show once again high quantities of bacteria coated to the epidermis, but no or too low levels of GFP to be seen in the epidermis (Figure 5). Interestingly, green fluorescent structures were visible in the stele of the root (Figure 11 B). However, it remains uncertain whether the green fluorescent structures observed at the outermost layer of the plant's vascular tissue are a result of GFP being secreted into this region or if they originate from natural plant compounds. We critically question how GFP might have reached the stele without accumulating in the epidermis of the cell. Cross sections of a negative control (A. thaliana roots without bacteria) show similar patterns of fluorescence in the stele (Figure 5 D).

Existing scientific literature indicates that oxidative stress can trigger the production of secondary metabolites in plants, and these metabolites can exhibit autofluorescence properties at wavelengths similar to GFP [2,3]. Knowing this, we consider it probable that the fluorescence at the stele is not GFP-derived but is plant-originated.

Figure 5: Phase contrast-(left) and fluorescence microscopy (right) picture of A. thaliana root cross sections (thickness: 25µm) A and 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: 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: Autofluroescent structures in epidermis and stele.

Conclusion 2

The vibratome sections gave interesting insights of intracellular root structures and showed a new perspective on how bacteria attach to the root surface. Still, bacterial derived GFP was secreted in no or too low quantities to be detectable inside the epidermal cells. The investigation was complicated by autofluorescent structures that were mainly visible in the stele of the plant.

Summary and outlook for future experiments:

We employed various fluorescent microscopy techniques to gain insights into the secretion of GFP from Pseudomonas fluorescens etHAn into the roots of Arabidopsis thaliana. However, we did not see detectable levels of GFP inside the plant roots. We considered two possible explanations for the nonappearance of GFP inside the root cells:

1. Choice of Protein: While GFP is an excellent protein for fluorescence experiments both under the microscope and on plate readers it might have been an imperfect protein for the here described study. In this study, superfolder GFP was used. As the name already implies, superfolder GFP (in this text referred to as: GFP) has a very stable 3D-conformation [5]. For injection into plant cells, proteins are secreted through the 25-Å channel – enough space for proteins in their unfolded, but not tertiary, globular conformation [6]. Therefore, specific chaperons can bind to proteins with the T3SS secretion signal, facilitating the unfolding process and allowing passage through the T3SS [8]. In a future experiment, we should use a less stable protein that is easier unfolded and thus secreted by the T3SS.

2. Sensitivity of detection: Fluorescence microscopy is a great tool to detect GFP. However, in our case, we were not able to clearly differentiate between fluorescent structures from the plant and those caused by GFP. One can overcome these challenges by implementing more sensitive protein detection methods such as Western Blotting or silver staining (see future experiments).

For future experiments, we plan to enhance the sensitivity of GFP detection within the plant by employing western blotting to a root lysate. In the initial step, we will incubate the roots as we did in previous experiments with bacteria and subsequently remove the bacteria. Following the removal of bacteria, we will proceed to lyse the roots and perform a western blot analysis on the resulting cell lysate. We already integrated the protocol for Western Blotting on our protocol site. In a first iteration, we will try this, using our GFP-secreting bacteria. Later, we might use other proteins that are less stable and therefore more suitable for secretion. We hope to present the results of these experiments at the Jamboree in Paris!