For our first meeting, we mainly discussed the OxyR-GFP plasmid that Tsvetan had engineered. He underlined the importance of having a cassette facing separate directions so the promoters would not overlap in their expression (avoid possible leakiness). He gave us cloning tips, notably about Plasmidsaurus to sequence genomic DNA. Moreover, we discussed the sponsorship proposal, he gave us a lot of insights and advice on parts we could improve to make it clearer and more appealing, as he is in close personal contact with the private industry.

Advice on our detection system

Milli-molar reactive oxygen species (ROS) levels are extremely high and the choice of the fluorescent protein should be based on a low basal auto-fluorescence (This is the reason we later switched from GFP to mScarlet expression in yeast). Luminescence is easier to detect than fluorescence.

ROS sensor testing

How could we properly validate our ROS sensors? We should simplify the system as much as possible. He recommends H2O2 as a ROS species, as it is easily accessible and the most stable. Other reactive-oxygen and -nitrogen species can be generated through specific reactions in the lab. He would recommend using a fluorometer.

Directed evolution methods

Concerning FACS, he thinks it is of great use for directed evolution (DE) but that for validation it is overkills. For DE, his lab uses yeast, with Orthorep, but it primarily works for protein engineering rather than promoter sequences. He mentioned the possibility of using an endogenously high-error-rate polymerase as an alternative to varying manganese concentrations with a normal one. Error-prone PCR (epPCR) we ended up doing used a Taq polymerase that did not have proofreading ability.

Protein-DNA interaction models

The Baker lab can help us further with models of protein and DNA interactions. RoseTTAfold All-Atom could be used for shortlisting useful mutation sites for error-prone PCR libraries.

Best method for mutating specific nucleotides

It is best through synthetic DNA ordering and cloning, we can use Twist and IDT offers.

roGFP redox sensors as a control

roGFP is not recommended because it changes the excitation wavelength, not the emission. There is a change in emission intensity based on excitation wavelength.

Having two distinct-ranged sensors

Adding another GFP for general readout - is it worth it? Yes, it would help clarify the range which is detected. The problem is that if the transcription factor (TF) is out of the range due to being too low, there wouldn’t be a second signal.

Current treatments for IBD

There are antibody-based therapies which are widely used (weekly subcutaneous or intravenous administration), a lot targeting the IL-23 cytokine (1), a pro-inflammatory cytokine produced by macrophages. Checkpoint inhibitors or cytokine signaling inhibitors can also be recommended. An S1P1 antagonist (thought to trap lymphocytes in lymph nodes) was tested but did not yield conclusive results. Inhibitors of the TL1A (transmembrane cytokine member of the TNF receptor superfamily) pathway are promising since they show antifibrotic properties which are crucial to prevent strictures formation or worsening (scarring, fibrosis) (2). Most treatments reduce inflammation but not stricture formation directly. The most used treatment is Mesalazine (5-aminosalycilic acid: 5-ASA), thought to be antioxidant (3) and to inhibit the production of cytokines and inflammatory mediators (4).

Would ROS-sensing and an antioxidant output help relieve tissue damage and maintain its integrity?

ROS are thought to over-activate fibroblasts to produce collagen which leads to fibrosis, so yes ROS-targeting could be very relevant.

Are antioxidants commonly used to prevent flares?

Antioxidants are usually part of a diet, so recommendations can be given such as ingesting Vitamin D or curcumin (powerful antioxidants), as there is clear data showing its efficiency (5). However, one needs to be careful since additives like Vitamin C (acidifies, low pH promotes inflammation) or Peparazine (helps absorption, rather useful for rheumatoid arthritis) can impede efficiency. Curcuma has been paired to another antioxidant Indigo in a promising study (6).

On his side, he developed anthocyanin capsules and observed around 50% reduction of calprotectin levels (a marker of inflammation used to detect flares). However, regulations and costs stopped the study from going further. Regulations for methanol extraction are well described but methanol is highly toxic so they tried to develop a safer ethanol extraction, which was too new to get through all regulation hurdles. Right now, he got an SNF grant (Swiss National Science Foundation) to dig into the role of proton-sensing receptors in local pro-inflammation.

Bacterial interactions with the intestinal epithelium

Bacteria are organized in a layered way, so to sense mucosa levels the bacteria would need to sit against the epithelium. A colleague of his tried to secrete recombinant defensins (naturally produced by Paneth cells), but they were secreted in the gut lumen and not in the mucus layer, therefore they did not have any effect in the intestinal epithelium. We mentioned the chitosan-alginate-coating by Zhou et al. (2022) (7) for their engineered probiotic bacteria and he thought it might indeed be interesting. He points out that bacteria adherence can rely on several patient-dependent factors such as blood group and sugars.

Is there an IBD-typical microbiota? Are some strains correlated with flares?

It has not been proved whether there is a clear correlation for a strain, however, it has been shown that stool from an “active” IBD-patient will worsen gut inflammation in germ-free IL-10-deficient mice (called a fecal microbiota transplant, FMT) (8). However, it could be linked to molecules surrounding bacteria in the fecal matter and not bacteria themselves. The question is to understand if dysbiosis is a cause and/or consequence of gut inflammation. Recent studies suggest the role of adherent-invasive E. coli (AIEC) in gut inflammation and fibrosis (9).

Probiotic role of E. coli Nissle 1917 bacteria

Nissle was approved as a maintenance treatment to prevent flares for remission patients (Mutaflor®) instead of 5-ASA. This shows that Nissle had a positive effect on its own. Nissle can also be administered even in acute flares.

Distinction between Crohn’s disease and colitis

Ulcerative colitis is restricted to the colon where Crohn’s is for 50% of cases restricted to the terminal ileum where there are a lot of bile acids. He reckons our bacteria would mostly serve for colitis treatment.

Would engineered bacteria be accepted as they are GMOs?

Phages are, why not bacteria? He knows a lot of people working on modified bacteria. He does say that we should refrain from keeping plasmids (genomic integration would be the next step). There are always risks of mutations and patients the bacteria won't colonize, it is complex to predict. As an example, a treatment induced goblet cell hyperplasia (loss promotes inflammation in Crohn’s) and no one could have predicted it, so in the end, we need to try.

Another potential direction for our ROS-sensing bacteria: improving calprotectin precision as a gut inflammation marker

There are 25% chances of a relapse activity in the 1st year of remission patients, so they usually have calprotectin and ultrasounds checks every 2 months. Calprotectin (S100 protein of neutrophils) is a marker of inflammation generally, meaning bronchitis or otitis also trigger elevated levels of the protein. This leads to false alarms, and some patients present naturally high levels of calprotectin, or inexplicably varying levels. A useful way to use our bacteria would be to have them sense gut inflammation and detach when they do so they can be detected in the stools during flare-ups.

Concept of therapeutic ceiling

It is discussed that the therapeutic ceiling has been attained, where the maximum remission rate of a drug could not exceed 40%. Combination therapies can help improve this ceiling, such as combining anti-TNF and anti-IL3 or anti-P19 treatments, which gives an additional 10%.

Can we predict the affinity of a TF for a sequence? How about using epPCR for random mutagenesis of our sequences?

Prof. Platt said this prediction would be complex. However, he validated the error-prone approach for random mutagenesis of transcription factors and promoters, saying we can tune the Mn (manganese) concentration. He mentioned deep mutation scanning, but we would need to know the binding pocket and create a DNA library. Since our TF do not have a ligand, this would not be possible. He also talked about saturated mutagenesis of an enzyme if we know the sequence.

General advice

He told us to test as many sensors and versions as we can since we can never fully predict which would work. More simply, he encouraged us to test many transformation parameters which can also influence sensitivity. Concerning sorting, he told us to do several sorts of the same batch with different ROS levels to establish better confidence and select the most promising clones (some can have constitutive binding and others can not bind at all). He recommended we freeze part of our libraries each time.

Relevance of our project

More largely, he said there is definitely space for what we aim to do. He told us about calprotectin sensors for IBD (Immune Bowel Disease, symptoms of which we aim to alleviate in humans), since high calprotectin levels are a sign of IBD. We could pair IBD sensing and a local output, but similar approaches have been studied before.

ROS-detection market

First, we discussed current ROS-detection technologies, to understand what is currently on the market and available, and to compare with our biosensor microbiota. Many companies rely on toxicity assays to commercialize their product, for instance, antioxidants effects for skincare products, but also in nutrition and pharma. Aurélie tells us they also collaborate a lot with academia to develop their tools and assist in providing news means of oxidative stress assessment.

Current ROS-scavenging technologies

Various cellular assays already exist, like DCFHDA, a very common intracellular ROS marker, or reporter cell lines of the Nrf2 oxidative-stress-reactive pathway. They developed a specific test to assess ROS scavenging based on the LUCS (Light Up Cell System) where a photoactivatable ROS-generating DNA intercalator assay can help determine toxicity/antioxidant effects of ROS-scavenging compounds(10). A SOD or catalase effect assay does not exist for the moment and we could consider developing our bacteria to meet this need.

Human research application of our project and GMO considerations

Aurélie thinks our engineered bacteria could be indeed very useful in microbiota applications (notably cosmetics and nutrition), especially in research. The GMO aspect is to be taken into account but she knows many people who are working / worked on GMOs to advance current understanding and develop future technologies, as GMOs regulations are softening more and more and some countries are currently way more open, compared to Europe. She underlines that in cancerology, many bacterial GMOs are being considered such as photosensible bacteria.

Cons of current antioxidants treatments for IBD and our closed-loop alternative

In the scope of IBD, we discussed the cons of current antioxidant symptom management treatments. Risks include unwanted interactions with medications like blood thinners or anti-inflammatory drugs, but also pro-oxidant effects of high doses of certain antioxidants like vitamins C and E. High doses can also lead to gastrointestinal issues (diarrhea etc) and aggravate IBD symptoms. A local and timely response of antioxidants would be very relevant. Considering some compounds, diligence in daily intake and bad taste associated with certain antioxidants (curcumin, fish oil…) might be an obstacle against proper treatment, where a closed-loop system would be fully autonomous and cause no burden. She has not heard of any tolerance or dependance to antioxidants, they should be safe as long as doses are respected.

Fireblight situation in France

Fireblight is still an issue in France and was de-quarantined because it was so widespread that it was unreasonable to cut down every contaminated tree and was becoming taboo. Farmers still keep to themselves and their circles regarding fireblight outbursts (as it seems to hold a similar reputation as HIV in humans, according to our understanding).

Fireblight economical burden

Fireblight is the most costly disease since the tree needs cutting, where 1 hectar (4000 trees) costs 75’000 €, and only profitable after 20 years. Moreover, anti-insects nets can further impede detection. However, “Tavelure” (apple scab) is currently the greatest problem for fruit trees (a fungus).

Further contacts we should talk to

M. Leclanche gave us many contacts, such as Perrine Gravolon, Agroscope researcher in Switzerland who tests potential treatments on apple trees, but also handles the surveillance group for Constance-Sud-Tyrol-Germany fireblight. He also talked to us about Eric Allard, head of the Agriculture Chamber of the PACA region (south-east of France, where weather is very similar to that of central and western Switzerland). They have severe and recurrent outbursts there, representing 20% on the national fruit tree surface. Many sensitive species were banned but are now being reauthorized since 2019, so problems are on the rise again. Cindy Morris at INRAE was also given as contact, as she is part of a group who works on bacteriophages against fireblight (Clara torres).

The support of the Pommes-Poires French Fruit association

Xavier organized funding from the French apple and pear association (1.5k), where there are 1400 adherents.

Urgency of the fireblight situation in France and Italy

Since 2017, fireblight has had some very strong outbreaks in the region. In France, this year they have observed symptoms since the beginning of May. They vary depending on the varieties and are evolutive symptoms, especially noticeable one month after blossoming. Furthermore, in the Italian Piedmont, the situation is very serious but there is also a big taboo around fireblight (quarantined past), and outbreaks are often not communicated to the sanitary authorities. In the US they talk about it more openly now, in Europe it is still very much a taboo, there was a colloquium on fireblight in Bolzano last year which stayed constrained to a very little number of selected attendees.

Preventive measures

The “Blossom protect” is one known biocontrol treatment, composed of yeast and acidic buffer (DGAL), but is not homologated, so many couldn’t protect their trees. There is a strong need and interest in new solutions and technologies.

Uncertainties about how to handle fireblight outbreaks

Should the farmers cut the trees or branches? Leave them? It is controversial, it is usually recommended to cut. The risk of cutting (pruning) is that the tree can get infected more easily, is stressed, and will yield less produce. Carefully disposing of cutting is crucial to avoid further spread.

Natural fireblight-favourable micro-ecosystems

One sort of “aubépine” tree (hawthorn) is very present in the region and is suspected to be a healthy host for Erwinia amylovora which could explain the gravity of the situation in the region and yearly outbreaks.

Current status on innovation

France is very much behind in terms of allowing innovative solutions for fireblight. We should discuss with Belgian experts (PC Fruit is equivalent to the Swiss Agroscope), they are currently doing some testing with fireblight-targeting bacteriophages and have already tested them in many European countries (unauthorized in France). It is very hard to obtain a marketing authorization (“AMM: autorisation de mise sur le marché”) in France, so firms develop their products and commercialize them through other countries like Switzerland, Belgium or Italy. Apple scab is also a main concern and he reckons it triggers oxidative stress as well.

Drone use in French agriculture

Camera technology is already being used by apple producers but they put it on their vehicles mainly to count the numbers of flowers. It is promising since they are already familiar with the technology and instruments.

Further contacts

M. Allard gave us names of people to contact like Stéphane Kunz in Konstanz that developed the Blossom Protect, but also Vincent Feillon who brought it to the American continent, and finally, Perrine Gravolon who is a Swiss researcher working on fireblight. He told us to look at lake Balaton in Hungary, where they had a lot of problems with fireblight and first talked about bacteriophages.

Current biocontrol agents

We discussed current means of protection against fireblight. Bacillus subtilis can be sprayed on flowers during the blooming period. Pantoea agglomerans C9-15 and P. fluorescens A506 are the main components of the “Serenade” biocontrol solution (talk to Cosima Pelludat). He confirmed that PCR can be used for fireblight detection, but it is not a high-throughput detection method.

Regulation of GMOs

Cisgenesis (engineered species which could be obtained by selection and breeding) is accepted outside Europe (proposition studied in Europe). Currently, generated species having resistance genes are being tested, while being very close to GMOs. He mentioned the 37A article of the joint EPA / ENCA Interest Group “Definition of Genome Editing and Relevant Terminology” report from 2023 (Switzerland, France, Finland, Italy, Austria, Germany). The Swiss Gene Technology Act includes “provisions regarding a transition period for placing on the market of certain GMOs (the so-called GMO moratorium). Until 31 December 2025 GM plants (including plant parts, seeds and other propagation material) as well as GM animals for agricultural, horticultural or silvicultural purposes are not to be authorised for placing on the market (Gene Technology Act, 2022).”

In vitro flower testing possibilities

For engineered bacteria testing, he mentioned the flower on agar in vitro test we could do. He said we could also test inoculation on an Arabidopsis thaliana leaf with a ROS inducer.

Fireblight situation in Switzerland

Ea is not quarantined anymore which leads to a lack in funding, linked to the fact that infections were weaker in recent years. This is because there were not ideal conditions for E. amylovora infection; high heat and high humidity need to be combined during blooming.

Most susceptible species to fireblight

Gala and Golden delicious are the most susceptible apple tree species.

Xylella fastidiosa in Switzerland

We also asked about Xylella fastidiosa since it also triggers a high ROS burst, but X. fastidiosa has not been found in Switzerland yet.

Strain choice for engineering

She would also use Pantoea for engineering. P. vagans C9-1 (reclassified P. agglomerans) ideally since it has 2 antibiotics and 2 siderophores which can act against E. amylovora infection.

Tips for experiments

• As Giovanni said, we could use flowers stored in the cold room for tests.
• For Pantoea culture: we can use a media made by Pusey (2008) that mimics nectar content (flower media basically, adding nicotinic acid, amino acids). It is not LB but TSB in Pusey media.
• We need to culture Pantoea on the flower to keep its virulence. It can also be epiphytic.
• We can use the P. vagrans white variant (lacking a 530 kb mega plasmid called pPag3 for thiamine sufficiency), it can not survive more than a few days without supplementation (auxotrophy), but still keeps its virulence. Theo Smits should have it in his collection.

The fireblight taboo

There is a big taboo around the theme of fireblight. Before, when E. amylovora was a quarantined organism, there was more money for research, and when farmers had a fireblight infection they could announce it to the sanitary authorities and receive compensation money. Nowadays, this system does not exist anymore and the farmers who have fireblight on their trees prefer to keep it for themselves since they would have to cut their trees.

Her research at Agroscope

She is the main active researcher of her group, specialized in field tests for solutions against fireblight. Solutions include testing new products and also new apple tree varieties which might be less susceptible to fireblight. The varieties might have a different test, so her end application is making sure the relevant varieties still make tasty apple juice.

How are these field tests performed?

The trees used are 3 years old pot trees. Fireblight bacteria are sprayed on one tree directly on the blossoms, at a high concentration to make sure infection happens. Symptoms sometimes appear after a few days only. Bumble bees are used to transmit the diseases to the other control trees (non-infected), so they also include bumble bee transmission assays in their routine. This does not reflect the condition in nature but is a good compromise between practicality for testing and comparability with trees in nature.

What fireblight biocontrol research do they focus on?

One approach consists in finding new varieties of apple trees that are more resistant to fireblight infection (done by another group). It is unlikely that a variety will be found that does not get infected at all, but in some varieties, the infection spreads slower or does not spread further than the blossom. The other approaches are the different strategies for the treatment of fireblight on infected trees. There are 3 main strategies: chemicals, yeasts and bacteriophages. Chemicals are simple to use, products containing living organisms are not as easy because the bacteria have to proliferate when you spray them first. They are dependent on environmental factors and also on the other bacteria that are already present on the trees, such as E. amylovora in this case.

Potential obstacles to our approach

Our product would be sprayed on the trees, however, spraying the blossoms increases the humidity locally and thus promotes infection, so the farmers prefer to spray only when it is really necessary. We could mix our bacteria with other products to only have 1 spray. Finally, when the field is very wet, sometimes the vehicles are not even able to go into the fields and the spraying is not possible at all.

E. amylovora infection dynamics

The last few years, trees were not very affected by fireblight: the weather needs to be humid and hot.

Strain engineering choice

Pantoea strains would be most relevant since the bacteria has many natural antimicrobials to fight E. amylovoraa. For E. amylovora, he would use Cspp1430 (less infectious) or Faw611.

About other pathogenic bacteria

He mentioned that Pseudomonas syringae is a problem in Latvia, where one of our teammates is from.

Phages against fireblight

We talked about her research and the fireblight situation in France. She engineers phage cocktails against fireblight. Phages can stay on the plant for up to a week if there is no rain or sun, used as a preventive measure.

Has resistance from E. amylovora been observed against these phages?

Erwinia amylovora has shown little resistance to Ea phages in vivo, since if they do develop a resistant phenotype, it impedes their fitness and pathogenicity too much. Cocktails can also help target Ea strains for which we do not have a particular specific phage.

Fireblight situation in France and integration of phages

She also said that fireblight was controversial since trees had to be cut down, and told us to talk to Eric Allard. The French fruit cultivators are trying to get a Belgian phage cocktail treatment approved (DCM company) but French regulations are too strict. For now, they mainly use copper or Blossom protect (yeast and acid buffer).

Synergy between phages and our engineered microbiota

She said the phages could use our bacteria as vectors since Pantoea (a potential plant epiphytic bacteria for us to engineer) can be a host for Ea phages.

What kind of fluorescence would be adapted for plants?

We should not have red fluorescence (RFP) since there is significant autofluorescence in the red spectrum in plants.

Random mutagenesis considerations on the promoter sequence

For P. fluorescens engineering (use a non-fluorescent strain), there is a plasmid with a dominant-negative MutL (K35E) to have a higher mutation rate (x1000). He could give us this mutagenesis plasmid called pORTmage. He warned us about mutations leading to decreased fluorescence not linked to less sensitivity but impaired intensity. Therefore, we would need to sort systematically (second round) for bright cells at a lower concentration, to see if the promoter is still functional. He also recommended we sequence a few colonies after each sorting.

Is having a two-step detection signal relevant?

We discussed having 2 fluorescent signals, one for pre-warning (lower threshold) and one for the higher range. He thought this would be quite interesting to have a more precise output.

Is fluorescence the right way to go? Would luminescence be better?

A luminescent output would be too weak. He mentioned a luciferase+fluo FRET.

Fireblight biocontrol brainstorming

He had the idea about adding a competing siderophore / siderophore transporter to take the iron from Erwinia amylovora (would be a bacteriostatic activity).

How is wheat blight infection approached in their lab?

We discussed his domains of expertise, notably the wheat infection by Zymoseptoria tritici. He talked about a dual approach to treat it: GMO and biocontrol through the experimental evolution of Myxococcus xanthus, a bacteria which would be competing with the fungus on the wheat stubble that remains after harvest. For the directed evolution, they test a panel of M. xanthus against a panel of Z. tritici. For strain engineering, they use non-self replicating plasmids and genomic insertions.

Which bacterial strain models are best for directed evolution?

Prof. Velicer said that Pseudomonas fluorescens was very used.

What is the fireblight situation and management in Switzerland?

Pests need to have a requirement for quarantine to be monitored, and fireblight is not quarantined anymore, which, sadly, leads to lack of money in the sector which impacts biocontrol development.

What are the biocontrol approaches currently put in place?

Concerning prevention, the Wallis canton recommends what is approved, mainly organic (like Blossom protect which is limited in the number of applications), and closely follows Agroscope’s research and recommendations. For management, surveillance of orchards is critical, most notably of 1st symptoms on the flower and shoots which need to be cut, and this is an obligation. This obligation extends to private owners who have an apple/pear tree in their garden (which can be complex to handle due to attachment towards the tree), where awareness is raised in each town during the blooming period and an app for reporting cases exists.

Are there collaborations between cantons and countries?

Wallis was one of the last cantons affected by fireblight (in 2019) and collaborated a lot with German-speaking cantons (who were affected first) to know how to handle the situation. They are also part of a big collaboration group with other countries like Germany and Italy to discuss the situation and treatment updates (the next one taking place in November). Erwinia amylovora bacteria strain monitoring is done in Germany.

How do producers feel about new technologies, are they responsive and interested about current advancements?

Producers actually push for new technologies to be developed and closely follow advancements. Our early means of detection would be highly relevant as, the earliest the detection can be, the better.

Are drones currently being utilized? How would they be received?

Drones are mainly used in vineyards at the moment, and also apricot trees. Machinery is often shared and group controls could be envisaged. Right now, surveillance takes up a lot of time and is stressful.

Would you see another use for our sensor?

Ms. Gilli mentioned it would be interesting to have a way to follow fireblight dynamics on the Rhone riverbanks where hawthorn is a great spread factor, creating a microenvironment which favors fireblight’s anchoring and yearly return.

What is the usage of drones in agriculture currently?

It depends on the country. In Switzerland, they are mainly used for research purposes and observations, but also applied research i.e. biodiversity monitoring. There are some niche applications in the praxis as well, one of them being in maize cultures, where they use drones to distribute products against insects when the plants are too tall to go into the field themselves. They are also being used in some vineyards because of the steep slopes. In conclusion, they are being deployed more and more, but the applications remain very specialized. Moreover, most farmers who do use drones often will have their own drones, but rather employ a company. In the Netherlands, he has seen very large farms using drones to monitor their fields on a regular basis.

Are farmers interested in these new technologies?

Yes, they are interested in such technologies because they see the limitations of the tools they use currently. There are still a lot of issues, and are eager for innovative technologies.

How expensive would the drone service be for farmers?

The prices are available on the websites of drone companies. They would typically be a few hundred swiss francs for a field, which is usually also in the range of the price you would pay a person to do the management in your field if you don’t have the machinery (for spraying or harvesting for example). There are also associations of farmers who share equipment, it could be likely that they could afford a drone they would share.

Is special training needed to fly a drone, and are there restrictions? Such as environmental concerns?

The approval is quite easy to obtain. There are restrictions on where it is allowed to fly drones but fields are often quite isolated so this should not be a major issue. There are areas where flying drones is not allowed due to planes or in bird protection zones. There is a public map for approved drone flight zones.

How would the detection of the fluorescence work?

It is not trivial. Two different devices would be needed. One that would trigger our fluorescent protein and one receptor. The light emitted by the protein is in the visible range so a visible light camera would be needed, and the light would be filtered out in the specific range of the signal emitted by our fluorescent protein. A camera and a filter would have to be bought, and are not difficult to find, this is also an established method of detection with drones and works from a certain distance. Exciting the molecules and triggering the signal would be the trickier part. Something intense such as lights (you do not want to have huge lights on the drone for practical reasons) or a laser would be needed. There are strong regulations on lasers, it is theoretically not allowed to put them on devices you can not control, to eye damage. Our laser would not have to be very strong but it is something to consider. Also, our laser would be only strong enough to induce fluorescence and targeted at the flowers. However, there is an existing device called LIFT that induces fluorescence to measure photosynthesis in fields, with a distance of about 1 meter(11), so this is doable.

Do you think of any hurdles we might face in our application, except for regulations?

It is experimental. Would it work? There will be issues with sunlight because fields are outdoors, but we could measure the fluorescence at night, in the shadow or during a cloudy day. Also, assembling the components on the drones, the one that would trigger fluorescence and the receptor would be a technical challenge, because this is not standard drone equipment.

Would you be open to collaborating with us to build such drones?

Yes of course, and also other labs at ETH show a big interest in real-life applications for drones. It would be nice to have a meaningful collaboration between our project and people who are specialized in drones.