The efforts to increase the sustainability of our everyday lives take place on many platforms. One method that can protect the planet from the consequences of energy waste and pollution is to convert microorganisms into small factories for producing food, textiles and medicine. Since this production is not natural for the microbes, we need to make it attractive for them to help us. Researchers have now joined forces with biosensors to get to know the microorganisms so well that they can persuade them to work even harder for us.
Bacteria and fungi have always been instrumental in helping people to produce everything from bread to medicine. In recent decades, researchers have increasingly succeeded in getting microorganisms to not only produce more of what they already produce but also to help us produce substances that they do not naturally produce. The challenge is not usually to get microorganisms to produce new things but rather in getting them to continue production long term, since continuing production of substances new to bacteria or fungi often confer an evolutionary disadvantage, and for this reason they lose this ability again. Researchers have now succeeded in tackling this problem.
“The trick is to create the evolutionary advantage that makes it worthwhile for the microorganisms to produce the new substances. To achieve this, we have enlisted the help of biosensors to analyse their metabolism. For the first time, we have now managed to create such an evolutionary advantage in yeast, so that the cells in the 55 generations we followed retained their ability to produce vanillin and thereby produced 92% more of it than cells without the built-in evolutionary advantage. The same method can be used in revolutionizing the manufacturing of food, textiles and medicine,” explains Michael Krogh Jensen, Senior Researcher at the Novo Nordisk Foundation Center for Biosustainability, Technical University of Denmark.
An evolutionary slap in the face
Vanilla production is a good example of how microorganisms can promote sustainability. The production of natural vanilla from the beans of vanilla orchids strongly affects the environment and leads to the felling of large areas of forest in Madagascar and other places. If yeast can instead produce vanillin, which comprises the largest proportion of natural vanilla, vanillin can be produced in fermentation tanks, and this avoids felling trees. The researchers can easily get yeast to produce vanillin in the laboratory, but when the production is scaled up, yields decrease.
“The problem is that the yeasts do not naturally produce vanillin, so they do not get any energy benefit by producing it. There is therefore evolutionary pressure to stop and the yeast cells simply drop production after a few generations of growth because it is a strain in terms of energy and because they do not depend on vanillin. We have been trying to create this dependence,” says Michael Krogh Jensen.
To get the yeast cells to produce vanillin, the researchers spliced genes into the yeast that encode several enzymes, each of which brings the production of vanillin one step closer to the end product. By thoroughly analysing the yeast cells’ production, the researchers determined which production step imposed the greatest strain on the yeast cells and then decided to link that step to a special benefit for the cells.
“When the yeast cells produce vanillin as the end product, this has required such a huge quantity of energy that they often dump genes to avoid this. We therefore decided to link the production of vanillin biosynthesis intermediates to essential cellular functions of yeast cells. When the cells start accumulating the intermediates of vanillin, they express a set of other genes that are necessary for the yeast cells to survive. Yeast cells that drop the production of vanillin therefore get an evolutionary slap in the face so that they do not survive,” explains Michael Krogh Jensen.
In technical terms, this benefit is induced by biosensors, which works by sensing the accumulation of the intermediates inside the yeast cells and then coupling this accumulation with activating vital genes, thereby creating an evolutionary advantage. In addition, researchers can use biosensors to measure whether production is proceeding as it should.
“Using this method, we managed to maintain production for more than 50 generations and with a yield that was more than 90% higher than cells without biosensors,” says Michael Krogh Jensen.
Libraries with DNA elements
This biosensor trick previously worked on Escherichia coli bacteria, but this is the first time it has succeeded in yeast. This is a significant result because yeast is an essential organism in industrial cell factories. E. coli can often be used to make bulk chemicals and fuels, but yeast provides the finer and more advanced chemistry, such as producing insulin.
“In addition to this being the first time this has succeeded in yeast, another almost equally important result is that the biosensor is not connected to the end product but is instead an intermediate product. This results in more rapid feedback, but even more importantly, the intermediate product is also part of synthesizing acids used to produce such products as nylon. We have therefore developed a tool that can be used in other contexts,” explains Michael Krogh Jensen.
To become even better at optimizing the microorganisms’ production of various substances, the researchers have now adopted even more radical tools. In a new study, the researchers succeeded in combining the increasingly advanced mechanistic models of the microorganisms’ metabolism with computer models and artificial intelligence based on data from biosensors and showed that they can construct algorithms using this method that can predict the design of metabolic processes.
“We initially used machine learning to study the production of some of the complex aromatic amino acids in yeast and thus succeeded in building algorithms that could predict new aromatic amino acid pathway designs, improving production by up to 74% compared with the best manual design used for training the algorithms. So this reveals enormous potential, and combined with our results on the vanillin production in yeast, we have high hopes that, within a few years, we will have more libraries with DNA elements to build these biosynthetic pathways in ways that ensure better and cheaper production stability ,and much greater sustainability,” says Michael Krogh Jensen.
“Combining mechanistic and machine learning models for predictive engineering and optimization of tryptophan metabolism” has been published in Nature Communications. “Regulatory control circuits for stabilizing long-term anabolic product formation in yeast” has been published in Metabolic Engineering. Main author Michael Krogh Jensen is a senior researcher at the Novo Nordisk Foundation Center for Biosustainability, Technical University of Denmark, Kongens Lyngby.