The new computational model can predict metabolic profile of yeast based on the measured protein concentrations, even when the enzyme kinetic parameters are missing. The model, abbreviated ecRhtoGEM, is the first of a kind to predict metabolism of yeast called Rhodotorula toruloides in the recovery of local organic waste.

Fulfilling expectations of people who care about the climate change

“Woody biomass is one of the richest sources of natural biomass the main components of which – hemicellulose, cellulose and lignin - can be used as raw materials for biotechnological processes,” explains Alīna Reķēna, PhD candidate and junior researcher at Tallinn University of Technology. It is the best alternative to petroleum oil for people who care about the climate change have been waiting for a long time.

While there has been a lot of exciting research about the use of lignin into applications for bitumen and biofuels, one might ask, but what about the sugars from hemicellulose? The simple answer is: where there are sugars, there is yeast. The latter historically have been very effective in processing sugars. If you think about beer or bread, yeasts change sugars into other molecules that we like in these products.

For many years, the so-called biotechnological problem of classical yeast was that, unlike bacteria, yeasts could not naturally process hemicellulosic sugars with 5 carbon atoms (C5), such as xylose. Therefore, attempts have been made in recent decades to make yeast consume it by genetic modification.

Red yeast comes to the rescue

Alternatively, a yeast has been discovered that naturally has the genes for processing C5 sugars and therefore can “consume” mixtures of hemicellulosic sugar mixtures and wastes. This red yeast named Rhodotorula toruloides, originally isolated in 1957 from tree leaves in China, can convert cellulosic sugars into lipids and oils, which can also be used as a substitute to palm oil, for example. Because it contains carotenoids (that is why this yeast is red), it can add antioxidant properties to products. The model also helps to design novel years to produce biofuels, biosurfactants and similar materials by adding to the yeast few new genes from some other organisms found in nature.

How to explain the “capability” of cell factories?

With the help of synthetic biology and metabolic engineering, it is possible to create all kinds of engineered microbial cell factories, but so far researchers have little information about how functional of “capable” these engineered microbes could be.

“To get more information, we grew this yeast in bioreactors specifically in a lipid inducing environment (known from early studies to be a secondary nutrient starvation) and used this data to generate advanced computational models,” explains Alīna Reķēna. According to her, models that try to predict metabolism of R. toruloides generally rely on the laws of physics, such as the law of conservation of mass and energy and environmental conditions. “Unfortunately, they don’t work well for fermentation data.”

According to Reķēna, ecRhtoGEM is the first model that uses absolute protein concentration data to predict metabolism. “We found that this model works well with the real experimental data. The reason is that the cell’s ability to support the metabolic flow is limited by the distribution of its resources and the enzymes that catalyze most metabolic reactions.” According to Reķēna, it is not possible to guarantee that the use of each new substrate will be correctly predicted, but it is possible get correct predictions on three different carbon sources present, for example, in wood sugars produced by the Estonian company Fibenol.

The new model is available to all researchers

The novel computatonal model was published in PLOS Computational Biology and ecRhtoGEM is made accessible online.

As stated, the established model is able to predict the metabolism even when R. toruloides specific enzyme kinetic data are not available. This has two implications. On the one hand, it shows the power of the model, so that the algorithm is able to make meaningful predictions in the absence of data. On the other hand, this situation reduces the predictive power of final model. Nevertheless, the model can still make reasonable predictions on possible candidates for the design of metabolic pathways.

“We made the ecRhtoGEM coding accessible online to other researchers working in the same field, and we hope that the scientific community will also use this tool. From Github, researchers can just download the models and complementary scripts, and run model simulations.”

According to Reķēna, the desire is to increase the predictive power of the model, for which the plan is to apply the latest machine-learning algorithms that predict the kinetic parameters of enzymes based on their structural and substrate information, so that users could obtain more accurate metabolic predictions even without proteomics data.

The research was supported by the Estonian Research Council through the national program for funding fundamental and applied research. Food Tech and Bioengineering lab was established at TalTech in 2021 and aims to develop novel processes for sustainable food and feed, biochemicals and materials (https://bioeng.taltech.ee/).