News

Microbial biosensors exploit genetically engineered microorganisms to detect the presence of target chemicals. Usually, they are based on the screening of a large library of enzyme or pathway variants in order to identify those that can efficiently generate or detect the desired compound. Such high-throughput screening can be challenging, especially if the tested library is very large.

To address this limitation, the FutureAgriculture team at the MPI-MP proposes the use of metabolic sensor strains which link the bacterial growth to the presence of the relevant compound, thus replacing screening with direct selection. Their method is based on strains with a series of gene-deletions; each strain displays a different dependency on the relevant compound to support growth. In this way, the growth phenotype of the different strains can be used to accurately estimate the availability of the target compound.

In order to design the most suitable metabolic sensor strains, the team at MPI-MP has developed a computational platform that, based on flux coupling, identifies the growth requirement of several combinations of gene deletions.

As a first application, they designed a set of E. coli sensor strains for glycerate – as this compound is one of the key expected metabolic products of the FutureAgriculture pathways.

They constructed several strains, in which glycerate serves as a carbon source for a different fraction of cellular biomass: in each strain, a different set of enzymes is disrupted such that central metabolism is effectively dissected into multiple segments.

The experimentally measured glycerate dependencies on the strains were found to almost perfectly match predicted values, thus confirming the validity of the computational platform. Moreover, they showed that this method can be used to accurately detect a compound concentration across two orders of magnitude. The glycerate biosensors also reveal key phenomena in central metabolism, such as the spontaneous degradation of central metabolites and the importance of metabolic sinks for balancing small metabolic networks. Such valuable information is now used to optimize the pathway design in FutureAgriculture.

Evogene is currently growing and testing the FutureAgriculture plants in its greenhouses under two conditions (normal and drought) with the final aim of validating their growth-advantage.

To evaluate the pathways in vivo, scientists at Evogene have established a set of molecular and phenotypic experiments.  Molecular parameters are tested to ensure the presence of the FutureAgriculture pathway inside the plant’s DNA and its correct expression. Phenotypic parameters are monitored to check the health and growth of the plant, e.g its general appearance, the efficiency of the photosynthetic systems, the biomass, etc.

The results obtained from previous stages of FutureAgriculture were essential for the selection and optimization of potential pathways to be inserted into higher plants. However, higher plants are more complex and differ from the bacteria used until now (E. coli and cyanobacteria). The preceding tests gave a general direction as for what to expect in the engineered plants; therefore, the whole Consortium is eager to see them growing.

In FutureAgriculture, Evogene is responsible for the pathway implementation in higher plants and phenotype characterization. First, they have selected the best sequences to implement the new pathways in vivo based on the results obtained by the other FutureAgriculture partners in silico and in vitro.

Then, in order to insert the necessary components into plants, Evogene takes advantage of the pathogenic machinery of Agrobacterium tumefaciens (A. tumefaciens). In nature, A. tumefaciens is a bacterium responsible for the crown gall disease in many dicot plants: a small segment of its plasmid DNA moves from the bacterium cell into the plant genome and causes the disease. Scientists have succeeded in deactivating the pathogenic machinery and use it to insert desired DNA pieces into dicot plants instead, and later into monocot plants too. The sophisticated machinery of A. tumefaciens is the basis of many scientific studies carried out today in plants as well as of Evogene’s ability at FutureAgriculture to transfer the new metabolic pathway genes into the model plants.

The long-standing experience of Evogene with plant transformation has allowed them to precisely calibrate many parameters, thus mitigating potential difficulties. In the case of B. distachyon, Evogene has even developed unique proprietary transformation protocols for high throughput gene screening.

Nevertheless, some unexpected challenges can still occur during the process. Sometimes even if the gene is correctly transferred into the plant and enters its genome, it doesn’t get expressed, meaning that the plant has silenced it for some unknown reasons. Other difficulties are related to low transformation rates, but internal know-how helps to solve them.

Even if no obstacles arise during the process, it still takes several months from the initial transformation to the first results on the pathway performance in planta. The process takes more time in B. distachyon (up to one year), comparing to A. thaliana (up to six months), due to the transformation method.

In FutureAgriculture, Evogene has chosen to work with two different model plants; Arabidopsis thaliana (A. thaliana) and Brachypodium distachyon (B. distachyon). Together they represent two major types of crop plants: A. thaliana is a model for widespread dicot crops like soybean, tomato, potato, lentil and peanut; whereas, B. distachyon for important monocot crops such as rice, corn, wheat, barley and oat.

Similarly, to animal models, model plants are an ideal ground for testing initial hypothesis rapidly and with high certainty. Compared to crop plants, they are characterized with shorter life cycles, smaller-sized plants and a simpler genome; all features that speed up and simplify the scientific process. Indeed, working with model plants allows scientists to perform a large number of experiments at a relatively low cost while gaining valuable knowledge and process understanding before moving to more complex crop plants.

Another powerful tool of A. thaliana and B. distachyon, is their known genetics. A. thaliana was the first plant genome to be fully sequenced (and the third multicellular organism). Ten years later, in 2010, the genome of B. distachyon – double of the size of A.thaliana’s genome – has been sequenced and published. The knowledge of the plant DNA is a fundamental prerequisite for the success of the FutureAgriculture project as it encodes useful information on the metabolic pathways. The pathways selected in FutureAgriculture are composed both from bacteria genetically modified enzymes as well as plant native ones. Without knowing the enzymes involved in the native network of the plant, it would have been almost impossible to establish a new pathway there. Through new discoveries and insights from the model plants we can take one step closer to better yield productive crop plants.

Introducing new genes into cyanobacteria is quicker than into plants but not as quick as one might expect. For example, the bacteria E. coli can be transformed overnight. The same process for cyanobacteria takes up to 1 month, instead. Its timing is extended because cyanobacteria are polyploid – they have multiple copies of their genome – and a process called segregation needs to happen to ensure that the new DNA propagates in all the copies.

Further care is needed when introducing a complex pathway in cyanobacteria: “We can’t introduce all the necessary genes at once, as there are too many,” explains Patrik, “Therefore, we do it in a stepwise manner: we start from a group of enzymes that operates at the end of the pathway and move forward”. This trick avoids the production of potentially harmful intermediates that otherwise could impair cell growth. “After one month of waiting, there are still chances that the strain rejects the new DNA and we might have to start all-over” warns Patrik, “Still, this method is way faster than the several months needed to generate an engineered plant”.

FutureAgriculture uses cyanobacteria to evaluate the in vivo behaviour of the synthetic photorespiration pathways. Their quick life-cycle and simpler compartmentalisation make cyanobacteria an ideal model organism to work with before moving into plants.

At the laboratory of Patrik Jones at the Imperial College London, scientists evaluate the pathways in a very comprehensive way, addressing their impact on carbon fluxes, metabolism and cell growth: “We study the carbon flux by following the journey inside the cell of carbon atoms taken up during photosynthesis. The CO₂ absorbed by the cell contains the carbon isotope ¹³C, distinguishable in our analytical machines over time. After only 20 seconds we are ready to follow the isotope through its metabolic steps in the natural and synthetic pathway.”

The technique is not trivial as the resulting data need an elaborate computational analysis to be interpreted correctly. But it helps to assess if and how the synthetic pathways are replacing the natural photorespiration pathways, and it might even provide some answers on the role of natural photorespiration.

Who would have known that a tiny unnoticeable organism as a bacterium would be the father of photosynthesis? Plants have taken their ability to convert solar energy into chemical energy from cyanobacterial-like organisms, which ultimately became the chloroplasts in plant cells. Today, cyanobacteria, sometimes referred to as prokaryotic algae, live in a wide variety of environments: moist soils, seawater and lakes, and they still retain the ability to perform photosynthesis. Actually, their efficiency per area of land is even higher than plants.

Like plants, cyanobacteria carry out the detoxifying process called photorespiration to recycle 2-phosphoglycolate (2PG), the toxic side-product of carbon fixation. Unlike plants, they have not one but three photorespiration pathways. “This is quite surprising,” explains Patrik Jones from the Imperial College London, “Cyanobacteria have already a carbon concertation mechanism that elevates the amount of CO₂ near the active site of Rubisco.” This strategy should minimize the need for photorespiration, since, if Rubisco has better access to CO₂, the production of 2PG is lowered. “It is commonly thought that photorespiration is negligible in cyanobacteria. However, when cyanobacteria lose the ability to carry it out, their growth is impaired.” Indeed, as part of the FA studies, the Jones group is also trying to further our understanding of photorespiration in cyanobacteria.

Patrik’s lab uses cyanobacteria as a tool to evaluate the synthetic pathways developed within the FutureAgriculture project and that have been designed to increase the yield of crop plants. “If we could improve the growth of cyanobacteria as well, this would also have positive consequences, as we might even rely directly on them for food” suggests Patrik. “They are extremely good in making proteins: in some strains, proteins account for more than 50% of their dry weight. Few food-sources can boast such a high protein content.” Actually, we already eat cyanobacteria – one example for all, the super-food Spirulina. Get ready, soon cyanobacteria might be part of our diet.

So far, to improve enzyme efficiency only the active site has been taken into account. However, parts that are far away from the active site can still affect the reaction. The further away from the active site, the less predictable it is what will happen by modifying that part and therefore the rational thinking alone won’t help much anymore. The FutureAgriculture team is now using high-throughput mutagenesis to expand the modifications and cover the entire structure of the enzyme.

Lately, some 10.000 variants of the same enzyme have been created and are currently analysed thanks to a novel high throughput method established in collaboration with external partners in Bordeaux that are not part of the consortium. The method consists in creating very small droplets, in the order of 50 picoliters, containing several thousand copies of a modified enzyme. The enzyme activity in the droplets is then tested very quickly with lasers: 200 droplets are processed per second – which means that in only 5 minutes 60.000 enzyme variants can be tested!

This droplet screening method developed some years ago, relies mainly on sensitive instruments with custom setups adapted to each system. To that end the partners at MPI-TM teamed up with Jean-Christoph Baret at the University of Bordeaux who is an expert in placing the enzymes in small droplets, sorting the good drops from the bad ones and assaying the enzyme activity.

As explains Tobias Erb “Among the variants that we are currently testing we are hoping to find a faster and more specific variant of the key-enzyme of the carbon positive pathway, a new CO2 fixing enzyme.  Once we find an improved version, we are going to test it again in the in vitro metabolism and hopefully, it will improve the plant metabolism even more.“

Once identified the best enzymes, it’s time to study them in their context: the carbon positive CO2 fixation pathway. “We want to make sure the enzyme can work well as a team. It’s like in football, you might have great individual players but to score you need a good teamwork, too. Sometimes it makes sense to take out one of the best players and substitute with another option because it interacts better with the rest of the team” explains Tobias Erb (MPI-MP).

To test the interplay of the individual enzymes, the CO2 fixation pathway is first recreated in vitro: a small vessel containing all the ingredients necessary to the pathway i.e. the individual enzymes, the chemical energy (ATP) and its reducing equivalents (NADPH), other cofactors that help the enzyme to exert its function, and a buffer that mimics the cell environment.

The pathway performance is then quantified by measuring the variations in the concentration of the ingredients. This is done by using highly sensitive mass spectrometers that can accurately quantify the amounts of substrates, intermediates and other metabolites for all the reactions that are happening in the vessel.

The team of FutureAgriculture has already tested in vitro the effect of the newly designed pathway onto the CO2 fixing-machinery of the plant. The outcome was extremely encouraging as the pathway showed a beneficial effect on the plant metabolism. This is the very first proof of principle that the pathway is able to improve plant metabolism in the lab condition, and that the pathway if transplanted in vivo has the potential to positively affect plant growth.

“It’s not enough to find an enzyme that can perform a reaction; the enzyme needs to be sufficiently active too” explains Tobias Erb, group leader and director of the MPI-TM in Marburg.
That’s why the partners at the Weizmann Institute and MPI-TM have worked hard to improve the identified enzymes and make them more efficient.

The method used by Erb’s lab starts by looking at the crystal structure of the enzyme – a 3D dimensional model with the position of the atoms. On that model point mutations can be made, usually by modifying the amino acids in active site – where the reaction takes place – to have a positive effect on the enzyme’s efficiency. The changes are not random but according to the scientist’s chemical logic and rational thinking. Two outcomes are important to be achieved:

• A fast enzyme: The faster the enzymes the higher will be the impact on plant growth. If the speed of the enzyme reactions is high then the metabolism of the plant will be very active and can operate with high CO2 fixation efficiency. Thanks to the computer modelling of the photosynthesis developed by the partner at MPI-MP (LINK), the threshold of activity that needs to be reached to ensure high efficiency is well-known.
• An accurate enzyme: The enzyme needs also to be as accurate as possible, which means that it should not have side-reactions. Such side-reaction might be toxic or slow down the metabolism and ultimately affect the plant growth negatively.

Once a set of mutation is decided, the mutations are implemented on the enzyme and their performance is tested in the lab. In Tobias Erb’s words “we try to make sure that the modelling is actually happening, that it becomes real” and therefore there is a lot of trial and error and back and forth, from the model to the experiments. Following this process, the FutureAgriculture team has found all the efficient enzymes necessary to complete a new pathway – the so-called “carbon positive pathway” – that should stimulate CO2 fixation in the plant and increase growth yield. Now it’s time to see how the individual enzymes work together.

Our ability to test promising pathways in vitro and in vivo is quite limited. The testing of every candidate pathway would become an endless quest – worth of decades and billions of investments.  To select only a few promising pathways we need the support of computational models that predict how each pathway will affect the carbon fixation rate in plants. We have generated a model of plant photosynthesis that takes into account both the central carbon fixation pathway (the Calvin cycle) as well as the photorespiration pathway – either the natural pathway or a synthetic alternative route.

This model enables us to estimate which synthetic pathway will result in the highest enhancement of carbon fixation rate, under different conditions such high/low illumination or high/low CO2-availability due to the opening and closing of the stomata. Only the most promising pathways will undergo a more extensive testing in vitro and in vivo.

The model also feeds both the in vivo and in vitro testing by setting important parameters thresholds and by giving suggestions on how to reach them.  For example, the model can estimate the needed quantity of each pathway component to reach the best performance possible. Therefore it directs the enzyme engineering phase by specifying the minimal activity that an engineered enzyme needs to reach, or it warns us against toxic or reactive compounds that might accumulate during the activity of the pathway within the cells. We then can fine-tune the expression of the pathway components to avoid such deleterious accumulations.

The model is also continuously improved by integrating the data from the in vivo and in vitro testing i.e. the enzymatic activity, the growth rate, etc.  Models are not completely finished yet but they are already well productive in giving us valuable information regarding the expected pathway. They are expected to be fully operative during the 3rd year.

After the identification of the pathways, we need to make sure that all the pathways’ components are available. Our pathways involve both existing and novel reaction – reactions that are not known to be catalysed by any enzyme in nature. We engineer existing enzymes to catalyze such novel reactions in order to sustain the activity of the pathway.

First, we need to find existing enzymes in nature that catalyze similar reactions or that can catalyze the novel reactions promiscuously. The concept of promiscuous enzymes is fundamental for this phase: enzymes generally have evolved to catalyze one primary reaction that represents their main task, but promiscuous enzymes can also catalyze, besides the primary reaction, side-reactions at a lower rate.  During the 1st year of the project, we were able to identify the promiscuous enzymes that catalyze all the novel reactions of four of our candidate pathways, although the catalysis rate is quite low as expected.  However, they represent the starting seed for the second stage where we are going to enhance the low-rate reactions of interest by three methods:

1. Rational design – by applying biochemical knowledge while looking at the enzyme active site and its structure, we can predict amino-acid residues substitutions in order for the enzyme to accept our substrate and support the novel reactions.
2. Library of changes in protein sequences – By using a large collection of alteration in the protein sequence, we can systematically screen the changes that result in better activity. To easily identify the best change within the library we have designed in vitro assays that couples the target activity to a measurable property i.e. fluoresce.
3. In vivo selection – by creating E. coli strains whose growth depends on the novel reactions, we can directly select for enzymes that catalyze it efficiently. Thanks to this method, we can directly select for higher enzymatic activity by simply selecting the cells that grow faster.

The backbone of our project is the identifications of novel pathways that increase agricultural productivity by enhancing carbon fixation rate and efficiency in plants. The pathway design was completed in the first year, during which we identified more than 100 candidate pathways that can potentially bypass the natural photorespiration without releasing CO₂.

We have considered all known enzymes and all known enzymatic mechanism to systematically search for all the possible routes that recycle 2-phosphoglycolate, the product of Rubisco oxygenation, back to the Calvin cycles (that supports carbon fixation). Our candidate pathways contained both reactions catalyzed by existing enzymes as well as plausible reactions, i.e.  reactions that potentially can be catalyzed by well-characterized enzymes or that follow a well-known mechanism. We compared the candidate pathways according to various physicochemical properties including thermodynamics, kinetics, resources consumptions, and overlap with endogenous metabolism. Our analysis takes into account also how easy it will be to evolve the novel reactions from existing enzyme and mechanism (i.e. the number of novel enzymes required in the candidate pathway and hints in the scientific literature on existing enzymes that could support such new activity). This approach enabled us to select the most promising pathways in terms of such properties and test them in vitro, reconstructing the pathway from its enzymatic components. We are now implementing the pathways in E. coli, using it as a platform for the pathway selection, before finally moving to cyanobacteria and plants.