基于神经网络和遗传算法的非天然氨基酸突变蛋白的培养优化

Optimization of Fermentation Medium of Proteins Containing Unnatural Amino Acids by Back-Propagation Neural Network and Genetic Algorithm

非天然氨基酸定点改构技术已被广泛应用于蛋白质结构与功能研究以及新药开发等,然而常用的无机盐培养基营养匮乏导致突变蛋白的产量过低,严重限制了该技术的进一步应用。相比传统的优化方法,文章结合了人工神经网络以及遗传算法,更方便、更准确地对大肠杆菌发酵突变蛋白的培养条件进行了优化。试验以引入对乙酰基苯丙氨酸的尿酸酶作为模式蛋白,以尿酸酶的酶活测定直观反映蛋白表达量,对培养条件中4个因素进行了摸索并优化,最终获得了最佳培养条件为:甘油1.04%、非天然氨基酸溶液1 mmol/L、金属离子综合液0.86×、IPTG 0.5 mmol/L,并且利用优化后的培养条件发酵获得的尿酸酶产量较未优化前提高了14%。并且通过测定尿酸酶酶活的精确比较,验证了ANN方法较响应面法在优化复杂的非线性生物工艺方面的优势。本试验的顺利完成,不仅提高了引入对乙酰基苯丙氨酸的尿酸酶的产量,同时也为其他非天然氨基酸定点引入蛋白的发酵培养基优化提供了重要参考。

The low level of protein expression has severely limited the application of genetic code expanding technique in the field of biochemistry. This article succeeded in increasing the production by optimizing the fermentation conditions. The Artificial Neural Network and Genetic Algorithm were used and the activity of uricase was measured to intuitively reflect the amount of protein expressed in different fermentation conditions. After a series of calculations and experiments,the best fermentation condition was intended to be cultured with glycerol 1. 04%,unnatural amino acid 1 mmol / L,metal ions mixture 0. 86 × and IPTG 0. 5 mmol / L. The production of uricase was increased by 14% in the optimized culture. This study not only increased the production of the mutant uricase,but also proposed a new method to optimize the culture in which the proteins incorporated with unnatural amino acids were expressed.