Figure 1: Schematic overview of FRUIT image analysis. Our novel protocol uses fluorescence microscopy images of bacteria on solid surfaces and transforms qualitative data into quantitative data.

Motivation: sometimes plate readers are just not enough...

The characterisation of genetic parts is crucial in synthetic biology. Plate reader assays are a common choice to for the study of inducible systems in synthetic biology. However, a prerequisite for many plate reader experiments is that bacteria remain in a transparent suspension. Those conditions do often not represent the in-situ behaviour of bacteria. The limitation to suspensions ultimately means that plate reader assays are restricted in measuring bacterial behaviour on surfaces.

We ran into this problem ourselves during our project. We wanted to confer our bacteria the ability to sense root proximity through root exudate inducible expression systems. We were able to give a basic characterization of these systems through plate reader assays by adding known concentration of exudates. However, we realised that these results did not give us any information on how our bacteria would behave on the surface of live roots. Obtaining data of bacteria fixed on surfaces is important in areas such as therapy development, food safety and wastewater treatments. Given the magnitude of this problem, we set out to create accurate and informative method to measure induction in situ on fixed substrates.

FRUIT: Fluorescence Root-exudate Unbiased In-situ Test

Experimental set-up

• A. thaliana seedlings preparation

A. thaliana seedlings were grown in sterile 1/2 MS10 Agar in a plant growth chamber to optimise root development for 1 week (16-hour photoperiod). After one week, 9 out of 12 seedlings were fixed in 4% paraformaldehyde at room temperature for one hour. This halts the metabolic activity in the plants and prevented the secretion of root exudates while still providing a similar surface for bacterial attachment as the test samples [5]. The root seedlings were then divided in three groups. Our sample group consisted of the 1-week-old non-fixated roots. The seedlings were then gently transferred onto a microscopic slide.

• P. fluorescens root inoculation

Biological duplicates of our P. fluorescens containing the pSal/nahR inducible expression system and a strain containing the corresponding empty vector strains were grown as overnight culture in 5mL LB-medium with its appropriate antibiotic (15 µM/ml Kan.). To maintain uniformity in bacterial inoculation, we diluted the overnight cultures to an OD600 of 0.1 before introducing 500 µl to each group of one-week-old A. thaliana live seedlings. After one hour, we washed away the cells with 1/2 MS medium and allowed overnight cultivation at room temperature.

• Fluorescence microscopy:

After 24 hours, we captured fluorescent microscopic images using the following parameters: 1- second exposure time, excitation wavelength: λ=485 nm, emission wavelength: λ=510 nm, 100x magnification, pH1 lens. For each fluorescence picture taken, a phase contrast picture was taken as well. All images were taken at the lower part of the differentiation zone (see Figure 2).

Figure 2: Main structures and tissues of the root system of A. thaliana

Control samples

Negative control: For the negative control, we tried to decrease the concentration of root exudate secretion as far as possible by root fixation with 4% paraformaldehyde. Paraformaldehyde is a widely used cross linking agent that stops all metabolic activity by creating covalent cross links between molecules, effectively gluing them together 2. Fixed samples are expected to not further secrete root exudate, while still allowing natural attachment to the root epidermis.

Positive control: Our positive control group consisted of fixed roots, that were incubated with 1 mM of salicylic acid. In a previous plate reader experiment, we established that the inducible pSAL/nahR expression system is activated in response to a concentration of 1 mM salicylic acid (for further information visit our “result, toehold” page). Our objective was to replicate the conditions of the plate reader experiment as closely as possible by employing fixated, non-secreting root samples and introducing the well-documented amount of inducer in a controlled manner.

Image processing protocol

1. Subtraction of the background/noise fluorescence:

   Background fluorescence was determined as the mean fluorescence of a root picture with P. flourescens harbouring the empty vector (subtracted value: -450). Everything that showed higher fluorescence values was recognised as “truly fluorescent”.

2. Remove autofluorescence of a highly autofluorescent root part (stele):

   The autofluorescence emitted by the stele of the roots was removed. The stele is the inner region of the root, containing main vascular tissues like the phloem and xylem and was found to show high autofluorescence after fixation with 4% paraformaldehyde. We manually selected the stele of the roots using phase contrast pictures of the roots, transferred the selection to the fluorescence pictures, and subtracted the autofluorescence for fixated and non-fixated samples, respectively (see Figure 4).



Figure 4: Microscopy pictures of P. fluorescens attaching to the roots of A. thaliana (negative control 1, 100x magnification, 1 day after inoculation). (A) Phase contrast picture with stele in selection, the stele is easily detectable as a, darker structure going across the length of the root. B: The selection from the phase contrast picture is transferred to the fluorescence microscopic picture and the autofluorescence is subsequently subtracted.

3. Bacterial fluorescence measurements:

   First, we created a mask to include fluorescence signals from bacteria but exclude remaining autofluorescence from the roots: we set a threshold that was above the mean autofluorescence of roots but below the average fluorescence of induced bacteria (fluorescence value of 125-200 after subtraction of background). Then, we analysed particles above a minimal size of 3 µm because smaller particles are not likely to be bacteria (the mean size of P. fluorescens is 3 μm [6]). Finally, the mean fluorescence of particles larger than 3 uM with more than 125 fluorescence was measured to represent bacterial presence. We created an image J macro that included these steps and can be applied to a picture tfor processing in a similar way:

   Image J Macro:

• run("Subtract...", "value=450");
• run("Duplicate...", " ");
• setMinAndMax(125, 2189);
• setAutoThreshold("Default dark no-reset");
• //run("Threshold...");
• setOption("BlackBackground", true);
• run("Convert to Mask");
• run("Close");
• run("Analyze Particles...", "size=3-Infinity clear add");
• close();
• roiManager("Measure");
• String.copyResults();

Results

We performed an initial qualitative assessment after having removed the background fluorescence from all samples. This revealed that the stele of fixed roots showed higher fluorescence than live roots as clearly seen in the roots inoculated by our empty vector strains (Figure 5). An analysis of the fluorescence values in the stele region shows that the fixed roots are 50x more autofluorescent than living roots (see Table 1). To prevent inherent differences in autofluorescence between experimental and control samples from affecting our measurements, we systematically removed the respective autofluorescence values from the stele region in the corresponding samples, as outlined in Methods, section 2.

Table 1: Mean autofluorescence in the stele of living and fixated A. thaliana roots. Fixation of roots in 4% paraformaldehyde resulted in a high autofluorescence in the stele of the roots

Figure 4: Fluorescence microscopic picture of P. fluorescens (non-gfp expressing) attached to roots of A. thaliana (1 day after first attachment). A: Root that was fixated in 4% paraformaldehyde for 1 hour before addition of P. fluorescens. B: Non-fixated, living root of A. thaliana.

Figure 5: Fluorescence microscopy picture of P. fluorescens (pSAL/nahR) root-sensitive bacteria with roots of A. thaliana (100x magnification, 1 day after inoculation). Test: Bacteria with living, exudate-secreting roots. Neg.: Bacteria with fixated, non-secreting, roots. Pos: Bacteria with fixated roots, in presence of 1 mM salicylic acid.

Our fluorescent microscopy pictures revealed that the strains harbouring the inducible expression system induced respond to live A. thaliana roots as shown in the upper panel of Figure 6. For the negative control, we opted to decrease the concentration of root exudate secretion as far as possible by root fixationfixating roots with 4% paraformaldehyde. In the middle panel of Figure 6, one can see that the reduction of root exudates results in the decrease of bacterial fluorescence..

In the positive control group, induction occurred not only in bacteria directly attached to the root but also in the surrounding medium. This aligns with the fact that the synthetic inducer was added to the 1/2 MS medium and is not directly secreted by the root itself.

Figure 7: Fluorescence response 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. Test: bacteria with living, exudate secreting roots. Neg: bacteria with fixed, non-secreting roots. Pos: bacteria with fixated roots, in presence of 1mM of salicylic acid. *** represents a student t test p-value < 0.001.

• FRUIT quantitively shows that the pSal/nahR inducible expression system is activated by live roots. Our assay can detect and quantify higher fluorescence in the sample containing live roots compared to the negative control while remaining lower than the positive control, matching the qualitative observations from figure 7. This gives FRUIT potential as a tool to characterize more accurately genetic parts in situ.
• By measuring the fluorescence of single cells, FRUIT allows for statistical analysis of the quantitative results produced from microscopy images. This shows that the results obtained with FRUIT are statistically significant.
• FRUIT is automated, and publicly available in the form of an Image J. This makes it easy to use, speeds up image processing and allows for reproducible fluorescent microscopy image analysis from different samples.

Conclusions

1. We characterised the pSal/nahR inducible expression system in depth. We did this using both plate reader fluorescence assays and our novel FRUIT in-vivo fluorescent microscopy image analysis protocol which is more likely to represent in-situ conditions than the former. Our microscopy images produced qualitative data that show that our strain senses root proximity in situ. FRUIT suggests that that the increase in GFP expression between the test group and negative control is statistically significant (see Figure 3, p-value <0,001). Still, we acknowledge that even if results were significant using biological duplicates, more samples must be processed to make stronger statements.
2. We developed an automated, standardised, reproducible image analysis protocol to be used by future iGEM teams looking to characterise parts while cells are attached to surfaces in situ
3. We present a simple tool by which the GFP-expression of bacteria in the rhizosphere can be studied and measured. Our new tool offers several benefits over plate reader data:
   • a. Qualitative fluorescence microscopy studies on living roots: Coinoculation of bacteria to living roots of A. thaliana allow us to see the induction of single cells in vivo under the microscope. In contrast to plate reader experiments, one can not only make statements of general induction but also about patterns of local inductions. In our case we detected that bacteria are only induced when being directly attached to the root surface with less or no induction in the rhizosphere.
   • b. Picture processing protocol for single cells: We present a method to measure the mean fluorescence of single particles bigger than 3 µm with the aim to get insights about the induction capacity of root exudates on the level of particles as small as single bacteria. In contrast to the previous part (mean fluorescence value of the picture) the disruptive impact of differences in bacterial quantities are mitigated, allowing measurements on the level of a single cell. For this part, we recommend subtracting the background fluorescence (in our case, a fluorescence value of 450) and mitigating the autofluorescence of the stele.

The shortcomings of plate reader assays and the need to better characterize parts in situ, motivated us to developed FRUIT, a new image processing and analysis protocol. The here described measurement approach is tailored to our project, using our own PseuPomona strain and measuring the induction capacity of root exudates from A. thaliana. However, we are convinced that the utility of FRUIT extends beyond our current project and holds promise.

The 2020 TU Darmstadt iGEM team found a way to reduce wastewater toxicity using B. subtilis. They used quorum sensing and biofilms in a beneficial way, to control their kill switch. Our FRUIT assay could have been used to look at biofilms up closely and provide quantitative data, improving the projects biosafety. In the food safety sector, the 2017 GlasgowiGEM team created a biosensor against a pathogen causing food poisoning. They characterize different quorum sensors in different bacterial strains. Some of the differences between them were not entirely clear from their plate reader experiments. With FRUIT, they could give more insight and detail of the quorum sensing dynamics between constructs and strains. FRUIT could be useful for iGEM teams developing medical diagnostics and therapeutics. The 2023 Groningen team developed a biofilm sensor to detect their presence in patients after surgery. With the use of FRUIT, more information about their biosensor could be obtained by looking at its response to biofilm structures in-situ.

Overall, FRUIT has potential to offer a new way of characterising parts for multiple synthetic biology applications localising in in situ. We believe that future iGEM teams that want to gain deep understanding of how the parts that they use would perform outside of the lab, should use and improve FRUIT.