Cheng Zhen, Wu Wenfa, Zhou Xiao, Xu Tingdong, Zhang Guanbin, Chen Hongwei
Department of Automation, Tsinghua University, Beijing 100084, China.
School of Intelligent Medicine, Chengdu University of Traditional Chinese Medicine, Chengdu 611137, China.
ACS Synth Biol. 2025 Jun 20;14(6):2105-2116. doi: 10.1021/acssynbio.5c00025. Epub 2025 May 14.
The artificial design and high-throughput screening (HTS) of synthetic yeast promoters are crucial for gene expression and metabolite biosynthesis in synthetic biology. It is essential to expand the screening scope and enhance the product identification in promoter engineering. Here, we develop a branch convolutional neural network (B-CNN), learning the sequence composition and structural properties of natural promoters, to comprehensively predict their relative strength and redesign the core region. It is combined with a genetic algorithm to identify mutant regions/sites of the alcohol oxidase 1 promoter (P) from (). Consecutive fluorescent recognition and automatic sorting of the GFP-expressing library of core P were finished in a laboratory-designed microfluidic fluorescence-activated cell sorting (μFACS) system. The rigid μFACS chip was fabricated, achieving single-cell sorting and meeting the requirement of complicated sterilization with reduced volumes and improved cell throughput (7000 cells/s). After extensive exploration of the sorting parameters, with high-intensity P (5.3× improvement) was successfully screened and validated by a microplate reader. The mutant P was further applied for the biosynthesis of PD-L1-specific protein with a 30% increase in production, proving its effectiveness and feasibility. This study provided novel insights into the rational design and HTS of yeast functional components with enhanced performance.
合成酵母启动子的人工设计和高通量筛选（HTS）对于合成生物学中的基因表达和代谢物生物合成至关重要。在启动子工程中，扩大筛选范围并增强产物鉴定至关重要。在此，我们开发了一种分支卷积神经网络（B-CNN），学习天然启动子的序列组成和结构特性，以全面预测其相对强度并重新设计核心区域。它与遗传算法相结合，以鉴定来自（）的醇氧化酶1启动子（P）的突变区域/位点。在实验室设计的微流控荧光激活细胞分选（μFACS）系统中完成了核心P的绿色荧光蛋白（GFP）表达文库的连续荧光识别和自动分选。制造了刚性μFACS芯片，实现了单细胞分选，并满足了复杂灭菌的要求，减少了体积并提高了细胞通量（7000个细胞/秒）。在对分选参数进行广泛探索后，通过酶标仪成功筛选并验证了高强度P（提高了5.3倍）。突变型P进一步应用于程序性死亡受体配体1（PD-L1）特异性蛋白的生物合成，产量提高了30%，证明了其有效性和可行性。本研究为酵母功能组件的合理设计和高通量筛选提供了新的见解，其性能得到了增强。
