Project
Human Practices
Dry Lab
Wet Lab
Jamboree
Simulating in-situ environmental conditions for studying plant-associated microbiota and their interactions has long been a gold standard for research in biosafety. Observing the behaviour of microorganisms in a controlled environment is a leap in the multistep process required before field application.
Our proposed biological system is an advanced, highly detailed design of our engineered Pseudomonas putida that will ensure the safety of Olive Trees against the pathogen Verticillium Dahliae. To ensure its safety and efficacy, we developed tools to monitor and test it s behaviour under controlled conditions. This collaborative effort addresses a fundamental question in biosafety: "How safe is safe?"
Soil is often regarded as the most complex biomaterial on the planet[2]. Studying the growth of Genetically Modified Organisms (GMOs) in soil is a crucial yet challenging and time-consuming step before their practical use. Our hardware was initially designed to remotely monitor the growth of engineered P. putida in soil, in order to reinforce our biosafety approach. Building on this concept, we propose a standardised experimental method for measuring GMO growth in soil, integrating machine learning to model and better understand the behaviour of microorganisms in this complex environment.
Our hardware is comprised of three parts:
Plant Incubator:
The environmental conditions inside the plant incubator are set through an application we developed where you can find in the repository of this site or externaly in Github. An Arduino Microcontroller is responsible for the operation of a thermal unit, a humidifier unit and an irrigation unit, while photoperiod is set directly by an outlet timer.
The temperature inside the incubator can be set between 15-35°C and can also vary throughout the day. The irrigation unit and mistifier unit provide water, through droplets directly on the plant vessels (pots, magenta boxes etc.) in the former and through mist in the latter. Humidity and air recycling is done through the intake and exhaust fans. Full spectrum growing and flowering LED strips supply light to the plants at specified times in the day.
The Data Logger consists of a series of sensors. A Raspberry Pi handles the temperature, air humidity and the plant culture vessel’s humidity measurements. When running our application the Raspberry Pi logs and plots the sensor measurements.
The Bacteria Interactions Logger consists of olfactory sensors that measure the microbial Volatile Organic Compounds produced on each plant culture vessel. This data informs us about the population and growth stage of the microorganisms inside the plant culture vessel.
By constantly collecting experimental data we can build upon them and in time create a highly accurate prediction model for the growth of engineered bacteria such as our P. putida.
If you are interested in building or expanding this hardware we have a dedicated page on how we built our Proof of Concept and how to Do It Yourself.
If you have any questions, or ideas for improvements do not hesitate to contact our team.
The study of synthetic microorganisms, such as our P. putida, in situ presents significant challenges due to the complexity of natural environments and concerns over ecological impacts. Engineered microorganisms must function effectively in diverse and often unpredictable environmental conditions.
Factors such as soil composition, moisture levels, temperature fluctuations, and interactions with native microbial communities can significantly impact the performance of synthetic microorganisms. Moreover, ensuring the stability and robustness of engineered strains poses a significant consideration. These microorganisms must not only survive but also thrive in their target environments without losing their engineered traits. Biological systems are inherently complex, with numerous interactions at play within microbial communities. Engineering microorganisms to perform specific functions requires a deep understanding of these interactions and the ability to predict how changes will affect overall ecosystem dynamics. This complexity makes it difficult to design reliable systems that can operate as intended in situ.
Our team’s hardware is built on the premise of thoroughly testing microorganisms in-vivo in a way that will eventually lead to in-situ testing. After extensive research and consultation with experts we decided to use olfactory sensors and data processing to correlate the mVOCs produced by a microorganism with its growth stage and population. We then decided to take this approach one step further by constructing a Plant Incubator that will allow testing under various environmental conditions, including stress conditions.
Researchers have managed to assess the growth of microorganisms by measuring their produced microbial Volatile Organic Compounds via olfactory sensors. The earthy smell of damp soil after a rain is actually the microbial Volatile Organic Compound Geosmin, produced by the bacteria Streptomyces when it comes in contact with water. In a similar way to how humans can gauge the intensity or recency of rainfall based on the strength of geosmin's scent, researchers can analyse the unique olfactory fingerprint of mVOCs to estimate the size and activity of microbial populations.
Moreover, sampling mVOCs offers a simple, non-invasive, and non-destructive approach to measuring microbial populations in soil. In contrast, the other methods we explored did not meet these criteria. Many of these techniques either involve the insertion of genes, cumbering the design of the biological system (Bioluminescence methods for example), or rely on destructive sampling (Plate Counting, DNA-based Methods such as qPCR). Advanced imaging methods, such as spectroscopy, while effective, are often prohibitively expensive and complex for most laboratories, making mVOC sampling a far more accessible and practical solution.
In many agricultural biotechnology applications, researchers need to investigate how plants behave under specific environmental conditions or how they grow in association with mutualistic or antagonistic microorganisms. To facilitate this, a plant incubator is often used. A plant incubator is a controlled growing system that allows for the manipulation of various factors such as temperature, humidity, irrigation frequency, photoperiod (day/night cycles), and more. This level of control enables scientists to simulate diverse environments and study plant responses under optimal or stress-inducing conditions.
Our hardware integrates microorganism monitoring through mVOC measuring, offering a non-invasive solution for in-vivo testing. By incorporating this technology into a customizable plant incubator, we enable experimentation under a wide range of environmental conditions, including stress scenarios. This approach brings us closer to achieving in-situ testing, providing a versatile tool for advancing research in biocontainment applications, such as plant-microbe interactions, and at the same time providing useful data for the optimisation of the design of engineered microorganisms.
Over the past year, we successfully built our Proof of Concept, completing a full cycle of the engineering principle Design, Build, Test, Learn (DBTL). In the spring, we focused on designing the hardware, followed by the building and testing phases during the summer. In the fall, we conducted a thorough review of our results, enabling us to refine and expand upon the original concept:
The core of our Proof of Concept involved testing the e-nose. Our goal was not only to test the system but also to construct a complete, step-by-step-guide for other teams or labs to replicate. For the Proof of Concept, we used the MQ-3 gas sensor, which detects ethanol and alcohol. This sensor was chosen for its affordability, accessibility, and the simplicity of testing with ethanol, a readily available household item.
We pipetted small concentrations of ethanol on cotton pads and placed them inside the Plant Incubator. The following video confirms the correct operation of our e-nose.
The first rise in the plot occurred when the fan above the first cotton pad activated, pulling in air and evaporating the ethanol, which was then transported to the sensor. Since the fans operate row by row, the second fan, positioned behind the first, did not have a cotton pad beneath it, resulting in a temporary decline in the plot. The third fan, however, had a cotton pad underneath, causing the plot to rise again. This pattern demonstrates that the system functions as intended, though further testing is required to assess the accuracy of the measurements.
Through extensive literature review and “consultations with biotechnology professors and industry experts, we have identified that the following experiments will allow us to develop a highly sophisticated model of our engineered Pseudomonas putida under a broad range of environmental conditions. This work will lay the foundation for the application of GMOs in the field.
At each stage of the experimental process, the microbial volatile organic compounds (mVOCs) produced by P. putida are measured and correlated with its growth using the e-nose. Specifically, the P. putida KT2442 strain, which we are working with, is known to produce the following mVOCs [3]: 1-Undecene, Carbon Disulfide, Dimethyl Disulfide, Dimethyl Sulfide, Dimethyl Trisulfide, Isoprene, and Methanethiol. To begin, we recommend using an electrochemical Dimethyl Disulfide sensor with a resolution of 0.1ppm in the range of 0-100ppm, which is cost-effective, priced at approximately 50€.
First Experiment: In the first stage the Antagonistic Interactions between Pseudomonas putida and Verticillium dahliae is studied, and in the second stage the interaction studied is between Pseudomonas putida and a model Plant under varying conditions.
What we learn: how engineered P. putida affects the plant.
In this experiment, the engineered Pseudomonas putida is placed in a magenta box with nutrients and then inside the Plant Incubator. We are examining its behaviour and growth under varying temperature, humidity and nutrient concentrations.
In the first variation of the experiment, Verticillium dahliae is introduced into the magenta box to study the effect of the fungus on the olfactory fingerprint and bacterial population dynamics, by comparing to the control.
P. putida is co-cultured with a model plant that does not produce volatile organic compounds (VOCs) similar to those produced by the bacteria, such as Solanum lycopersicum L. (tomato plant)[4].
In this second variation the effect of the bacteria on the phenotype of the model plant is examined. The colonisation on the root of the model plant by the P. putida plays a major role in this interaction and should also be compared both the control and to the agent-based model we built that examines root colonisation based on data taken from literarily sources.
In both of the above variations the mVOCs produced by the P. putida are measured and the olfactory fingerprint contributes in establishing a model of the growth of the bacteria under various situations.
Second Experiment: Studying the effects of the P. putida on an infected plant.
In this experiment, a model plant inside a magenta box is placed inside the Plant Incubator and is infected with Verticillium dahliae. The engineered Pseudomonas putida is introduced into the same environment to study its effects on the infection's phenotype. As in previous experiments, the olfactory fingerprint of the bacteria will be analysed, contributing to the development of a more sophisticated model of its behaviour and growth under these specific conditions.
Third Experiment: Studying the interactions between the P. putida and other microorganisms.
Up to this point, P. putida has been studied and modelled under a wide range of conditions. In this stage, a diverse selection of microorganisms commonly found in soil will be grown alongside P. putida, inside plant culturing vessels that are then placed in the Plant Incubator. These microorganisms should be chosen based on the criterion that they do not produce mVOCs in concentrations similar to those produced by P. putida. The publicly available database mVOC 4.0 [3] can be utilised for this selection. If the model developed in the previous steps is accurate and comprehensive, any observable differences in the growth of P. putida should mostly be attributed to its interactions with the introduced microorganisms.
Fourth Experiment: Studying the growth of the P. putida on field collected soil.
Field-collected soil is placed in plant culture vessels, and the growth of P. putida is examined under controlled conditions in the plant incubator, similar to the previous step. This phase introduces significant "noise" to the e-nose, as the multitude of microorganisms in the soil will produce a wide range of mVOCs in varying concentrations.
To improve the accuracy of the model, additional VOC sensors specific to P. putida should be employed, providing a clearer picture of its growth amidst the background noise. Additionally, conducting a Next Generation Sequencing (NGS) test on the soil sample will help identify the microbial species present. By cross-referencing this data with the mVOC 4.0 database, it will be possible to determine which microorganisms produce specific mVOCs and at what concentrations, allowing for more precise calculations of the differences in measured mVOCs.
Fifth Experiment: Expanding to mesocosms.
By now, our prediction model for the growth of engineered Pseudomonas putida has been refined through a series of experiments. In this phase, we transition from controlled lab experiments to mesocosms, such as EcoTron[5], intermediate-scale ecosystems that simulate natural environments more closely than laboratory setups. VOC sensors and microbial monitoring tools will be used to observe real-time changes, comparing actual bacterial behaviour to the predictions made by our model. The purpose of this experiment is to validate the model’s performance in a more complex environment.
If the model predictions and observed outcomes are different, we will revisit earlier experiments, in order to refine both the model and our understanding of P. putida's behaviour under these more complex conditions. At the point where the model aligns closely with the observed data, this will confirm its robustness, allowing us to proceed to in-situ testing.
Sixth Experiment: In-situ testing.
After multiple rounds of successful tests in controlled environments and mesocosms, along with external validation by other research teams employing various methods, we are prepared to move to full-scale field testing. In this stage, the engineered Pseudomonas putida will be introduced into a natural ecosystem, where it will interact with wild soil communities, plants, and environmental variables such as temperature, humidity, and nutrient availability in real-world conditions. The goal of in-situ testing is to observe the bacteria’s growth and interactions in an open field setting while verifying that the model predictions are accurate. Again, the growth of the P. putida will be achieved by a -mature by now- e-nose system, that will study mVOCs, as well as other microbial monitoring tools.
Studying in the field.
The scaling process of our hardware will initially support experimenting on our engineered P. putida in-vitro, within the controlled environment of a plant incubator. Upon completion of the final design, we envision that the system will enable research to be conducted in-situ on multiple microorganisms, as described in the Experimental Process section. This will evolve the hardware from its current prototype stage into a sophisticated, and cost-effective system capable of accurately and automatically quantifying microbial growth in soil, without the need for external intervention.
Improving measurement accuracy.
To achieve higher accuracy in mVOC measurements, improvements can be made to the existing e-nose system. In the current prototype, gas buildup in the top compartment may introduce noise over time. Furthermore, the aluminium chimney system can inadvertently draw air from multiple plant vessels simultaneously, affecting measurement precision.
A potential solution is to drill a small hole in each magenta box, insert a narrow-diameter tube, and connect each tube directly to a solenoid gang valve. This valve has multiple air inlets, each individually controlled by electronics, allowing air to flow from only one magenta box at a time. The gang valve’s outlet is connected to an air pump, which directs the air to the sensor. After each measurement, the air pump draws in clean air from outside the incubator to recalibrate the sensor. This setup ensures a direct air path from each magenta box to the sensor, eliminating gas buildup and isolating gas samples, resulting in significantly improved measurement accuracy.
Deep Learning model studying the Plant’s health.
One or multiple cameras could be placed inside the main chamber. The footage will be processed by a computer, such as a Raspberry Pi, by a Deep Learning model that will be trained on plant phenotype. Useful information can then be automatically collected from the footage, such as the effect of the engineered P. putida on the plant phenotype, or its effects on the infection, hopefully detecting that the infection is contained.
Material Sustainability.
Our team acknowledges the importance of sustainability and environmental responsibility, and we propose using durable, recyclable materials. While our Proof of Concept was constructed with polystyrene foam—chosen for its affordability, accessibility, excellent insulation, anti-oxidative properties, and ease of handling without the need for specialised tools—this material is neither recyclable nor durable enough for long-term use.
We propose using polylactic acid (PLA), a material used in 3D printing. PLA is derived from renewable, organic resources such as corn starch or sugarcane, and is both affordable and accessible. It offers excellent thermal insulation and eliminates the need for manual construction, as a 3D printer can fabricate the required components. Although PLA may degrade over time due to moisture exposure, protective coatings are available in the market, to extend its lifespan, making it a more sustainable and long-lasting alternative.
[1] Ke, J., Wang, B., & Yoshikuni, Y. (2021). Microbiome Engineering: Synthetic Biology of Plant-Associated Microbiomes in Sustainable Agriculture. Trends in biotechnology, 39(3), 244–261. https://doi.org/10.1016/j.tibtech.2020.07.008
[2] Young, Iain & Crawford, John. (2004). Interactions and Self-Organization in the Soil-Microbe Complex. Science (New York, N.Y.). 304. 1634-7. 10.1126/science.1097394.
[3] Marie C. Lemfack, Bjoern-Oliver Gohlke, Serge M. T. Toguem, Saskia Preissner, Birgit Piechulla and Robert Preissner. mVOC 2.0: a database of microbial volatiles Nucleic Acids Res. 2017 Nov 2; doi: gkx1016
[4] Nawrocka, Justyna & Szymczak, Kamil & Skwarek-Fadecka, Monika & Małolepsza, Urszula. (2023). Toward the Analysis of Volatile Organic Compounds from Tomato Plants (Solanum lycopersicum L.) Treated with Trichoderma virens or/and Botrytis cinerea. Cells. 12. 1271. 10.3390/cells12091271.
[5] Roy, J., Rineau, F., De Boeck, H. J., Nijs, I., Pütz, T., Abiven, S., Arnone, J. A., 3rd, Barton, C. V. M., Beenaerts, N., Brüggemann, N., Dainese, M., Domisch, T., Eisenhauer, N., Garré, S., Gebler, A., Ghirardo, A., Jasoni, R. L., Kowalchuk, G., Landais, D., Larsen, S. H., … Milcu, A. (2021). Ecotrons: Powerful and versatile ecosystem analysers for ecology, agronomy and environmental science. Global change biology, 27(7), 1387–1407. https://doi.org/10.1111/gcb.15471