Design

Although spinosad is safer, more environmentally friendly and more efficient than traditional pesticides, the main biosynthesis method that use \( Saccharopolyspora\:spinosa \) to produce spinosa through aerobic fermentation is still expensive due to factors such as low unit yield and so on, which restricts the promotion and application of spinosad.

During our team's research, we identified that one of the major limiting factors of low efficiency is the gap between the optimal temperature for enzymes within the biological system and the fermentation temperature or Optimal Growth Temperature. According to relevant research, the correlation coefficient between enzyme optimal temperature and the growth temperature of the host organism is only 0.48. This discrepancy may lead to reduced reaction speed involving critical enzymes, limiting overall product formation.

As we went deeper, we found out that it is the lack of comprehensive enzyme optimal temperature data that hampers our ability to address this issue. Enzyme optimal temperature, a fundamental property, has long been a focus of enzyme research. However, even yeast, as a chassis cell, lacks optimal temperature data for more than a quarter of its metabolic network, let alone actinomycetes which are far less general.

And this lack of enzyme optimal temperature is caused by the complex experimental measurement , and limited existing databases that provide information on key enzymes. Based on these facts, by collecting and analyzing the relevant data of \( Saccharopolyspora\:spinosa \), we developed a deep learning model to predict the optimal temperature of each enzyme in the process of spinosad synthesis in vivo, and compared it with the actual industrial fermentation temperature to find the enzyme with a large difference between the optimal temperature and the actual temperature. It is the fact that the activity of these enzymes is not fully realized during industrial fermentation that results in a decrease in yield.

In order to further optimize, we added enzyme constraints to the existing GEM model and constructed the ecGEM model to identify the key enzymes that were both limited by temperature and affected by yield by analyzing the rate-limiting enzymes in the metabolic pathway.

For these key enzymes, we will combine the prediction results of the deep learning model to propose an optimization strategy and conduct directed evolution experiments to make the optimal temperature close to the actual fermentation temperature, so as to improve their catalytic efficiency.

[1]  Li G, Rabe KS, Nielsen J, Engqvist MKM. Machine Learning Applied to Predicting Microorganism Growth Temperatures and Enzyme Catalytic Optima. ACS Synth Biol. 2019 Jun 21;8(6):1411-1420. doi: 10.1021/acssynbio.9b00099. Epub 2019 Jun 7. PMID: 31117361.

[2]  Sparks TC, Crouse GD, Benko Z, Demeter D, Giampietro NC, Lambert W, Brown AV. The spinosyns, spinosad, spinetoram, and synthetic spinosyn mimics - discovery, exploration, and evolution of a natural product chemistry and the impact of computational tools. Pest Manag Sci. 2021 Aug;77(8):3637-3649. doi: 10.1002/ps.6073. Epub 2020 Sep 28. PMID: 32893433.