基于计算机蛋白质设计的脂肪酶底物活性改造研究

Lipases are the third most commercialized enzymes, and have been wildly utilized in industries including biodiesels, food, and nutraceuticals. One particular application is enrichment of poly-unsaturated fatty acids, which have beneficial effects on human health. However, to our best knowledge, lipases used for this purpose all react on the saturated fatty acids, due to the low activity of lipases on poly-unsaturated fatty acids caused by steric hindrance at the active site. In this project, we will use computational protein design to engineer a lipase LipK107 for enhanced activity toward poly-unsaturated fatty acids. The engineered lipase could enable a new method for enrichment or recovery of poly-unsaturated fatty acids, elucidate the mechanism of substrate specificity, and also deepen our understand of lipase’s substrate recognition process. To increase the success rate of computational design and overcome the limit of current design methods, we will also develop a new scoring function for post-processing designed sequences from Rosetta, based on the multi-body interactions of an amino acid and its environment.

脂肪酶是第三大商业化最广泛的酶，在生物燃料、食品、保健品等行业具有广泛应用。脂肪酶的一个重要应用是富集对人体健康有益的多不饱和脂肪酸。然而，由于多不饱和脂肪酸结构比较刚性，容易在活性位点形成位阻导致酶的活性比较低，目前使用的脂肪酶均是通过选择性的与饱和脂肪酸反应来达到富集多不饱和脂肪酸的目的。在本项目中，我们将通过计算蛋白质设计的方法来对脂肪酶LipK107进行设计和改造，以提高其对多不饱和脂肪酸的活性。新设计的酶不仅可以提供一种新的富集和回收多不饱和脂肪酸的方法，也可阐明脂肪酶底物选择性的机理，加深我们对该酶底物识别过程的认识。为了提高设计的成功率和克服目前蛋白质设计方法中的缺点，我们还将基于氨基酸与其环境之间的多体相互作用，开发一个新的蛋白质设计打分函数，用于对Rosetta软件设计序列的重新打分。

脂肪酶在生物燃料、食品、保健品等行业具有广泛应用。在本项目中，我们利用蛋白质设计的方法对脂肪酶LipK107进行了底物特异性设计，提高了其对多不饱和脂肪酸的活性。为了提高设计的成功率和克服目前蛋白质设计方法中的缺点，我们还基于深度学习发展了蛋白质结构设计方法DenseCPD,该方法的准确性较目前主流方法有大幅提高。另外，还收集了目前最大的蛋白质稳定性实验数据集，并在此基础上开发了蛋白稳定性预测方法DeepDDG，其预测准确性相比其他近10种主流方法有较大提高。