Model-Based Optimization of Fed-Batch In Vitro Transcription

Recent developments in RNA vaccines and therapeutics have motivated the need for process engineering strategies to optimize the in vitro transcription (IVT) reaction for RNA synthesis. Specifically, practitioners seek to maximize the production of RNA and the incorporation of the 5-prime cap to the end of each RNA molecule while minimizing the use of expensive reagents. Fed-batch IVT is a promising technique for achieving these goals but is difficult to optimize by purely experimental means. In this work, we develop a mechanistic model for fed-batch IVT and use it to develop optimal fed-batch protocols which can produce over twice as much RNA as heuristic approaches. In addition, we observe and characterize for the first time the formation of magnesium phosphate crystals during the IVT reaction. We develop strategies informed by thermodynamic modeling to prevent this undesired crystallization during fed-batch IVT. Finally, we incorporate co-transcriptional capping into our model-based optimization approach and develop a strategy to maximize RNA formation while maintaining a high level of 5-prime cap incorporation.