生物分子传感中的抗体探针：由碳基计算走向硅基计算

doi:10.12211/2096-8280.2024-085

摘要/Abstract

摘要：

蛋白质等生物大分子在疾病诊断与治疗、基础科学研究中占据核心地位，而抗体探针作为免疫分析的关键工具，其重要性日益凸显。近年来，抗体探针的设计、预测与生成正经历从传统的基于动物免疫的“碳基计算”向人工智能驱动的“硅基计算”的革命性转型。传统的抗体生成技术依赖动物免疫，不仅效率低下，且难以精准控制。人工智能的引入为抗体设计带来了突破，实现了高特异性、高亲和力抗体探针的快速生成及抗原表位的精准预测。这一转变不仅能提高抗体类蛋白探针的性能，也缩短了研发周期。本文介绍并评论了抗体生成技术的演进历程，分析了人工智能在抗体设计中的应用优势与挑战，并展望了抗体类蛋白探针与新一代生物传感器的协同发展前景。蛋白质从头设计随着蛋白结合蛋白（PBP）预测技术的成熟，研究人员有望通过“硅基计算”与“硅基性能表征”，快速生成满足特定需求的探针分子，同时实现抗原表位及分子功能的精确预测。结合新一代高灵敏生物传感技术，人工智能辅助设计的非天然蛋白探针将显著提升免疫分析灵敏度，拓展可分析的分子信息类型，推动免疫分析向多维化方向发展。这一创新不仅为合成生物学研究开辟了新的研究路径，也将为精准医学诊断方法的开发提供有力支撑。

关键词: 免疫分析, 抗体设计, 碳基计算, 人工智能, 生物传感

Abstract:

The precise recognition, detection, and analysis of protein biomarkers are essential for disease diagnosis and fundamental life sciences research. Antibody probes, known for their high specificity and stability, are central to biomolecular sensing assays. Traditionally, antibody development has relied on "carbon-based" approaches using animal immune systems. However, we are currently undergoing a transformative shift toward "silicon-based" methods driven by artificial intelligence (AI). Conventional techniques, such as animal-based antibody production and phage display–based directed evolution, have long been hindered by low efficiency and limited control over epitope specificity and binding affinity. Recent AI advancements, including de novo protein design and deep learning-driven protein binding protein (PBP) generation, are revolutionizing antibody development. These innovations enable the rapid creation of protein-based biosensing probes (e.g., antibodies and nanobodies) with enhanced specificity and affinity, along with accurate predictions of epitopes and structural features. By overcoming the limitations of traditional methods, AI-driven technologies offer unprecedented control over the design and performance of antibody probes. Furthermore, "silicon-based evaluation" plays a crucial role in PBP generation, allowing for quantitative assessment of binding affinity, stability, and robustness.AI-designed biosensing probes offer the potential to capture a broader spectrum of biomolecular information than traditional antibodies. These probes may be able to detect variations in sequence and conformation, post-translational modifications, abnormal polymerization, and shifts in biological activity. In certain diseases, the abnormal dissociation of multimeric proteins can reveal previously concealed antigenic epitopes, creating disease-specific targets. AI-designed probes could play a crucial role in addressing these complex diagnostic challenges, providing more accurate and nuanced insights in the future.Moreover, modern high-performance biomolecular sensing technologies, such as bead-based chemiluminescent immunoassays (CLIA), digital immunoassay, microfluidic immunoassay and single molecule binding kinetics assays, require highly diverse antibody specificity and affinity. AI-based protein design tools can meet these divergent needs, enabling the integration of AI-engineered biosensing probes with next-generation sensors. This integration not only enhances detection sensitivity but also expands the scope of molecular information that can be analyzed. This paradigm shift represents a new era in biomolecular sensing and offers exciting prospects for precision medicine and synthetic biology.

Key words: immunoassay, antibody designing, carbon-based computing, artificial intelligence, biosensors

中图分类号:

Q816

谭骁天, 李睿涵, 杨慧. 生物分子传感中的抗体探针：由碳基计算走向硅基计算[J]. 合成生物学, DOI: 10.12211/2096-8280.2024-085.

Xiaotian TAN, Ruihan LI, Hui YANG. Antibody probes in biomolecular sensing: transitioning from carbon-based computing to silicon-based computing[J]. Synthetic Biology Journal, DOI: 10.12211/2096-8280.2024-085.

图/表 3

图1 抗体生成技术的发展与迭代。在2010年以前，生物传感中使用的抗体生成方式主要依靠动物免疫后人工筛选。2010年之后兴起的循环迭代筛选技术摆脱了抗体生成过程对动物免疫系统的依赖，实现了在体外筛选抗体片段乃至核酸适配体（Aptamer）。2020年以来出现的AI辅助设计则有望通过目标蛋白的氨基酸序列直接生成抗体，实现抗体计算、设计的硅基化。

Fig. 1 Evolution of antibody development approaches.Prior to 2010, antibody generation primarily relied on animal immunization followed by manual screening. From 2010 to 2020, phage display technology facilitated antibody production independent of the animal immune system, enabling in-vitro selection and the iterative generation of antibody fragments, nanobodies and aptamers. Since 2020, AI-assisted design has emerged, opening a gate to directly generate antibodies from the amino acid sequences of target proteins, thereby achieving silicon-based antibody design.

图2 探针蛋白结合蛋白的从头设计（A）两步结合蛋白设计方案的流程。在基于小蛋白文库进行大范围搜索后，与目标区域产生最大相互作用的序列设计会被引导至目标区域对接。潜在的具有相互作用能力的基序可被提取并聚类。筛选出的基序会被整合至小蛋白库并引导下一轮对接和优化。（B）以新冠病毒刺突蛋白RBD为靶点的两种结合蛋白设计方案。包括（1）以ACE2螺旋为起点的抗体设计方案。（2）大规模从头设计小螺旋骨架，并通过旋转异构体相互作用场对接寻找潜在强相互作用蛋白的方案。

Fig.2 De-novo design of protein-binding-proteins (PBP).(A) The workflow of a two-Step PBP design strategy. Following extensive search using a library of small proteins, candidates that exhibit maximal interactions with the target region are directed to dock with the target area. Potential interaction-capable motifs are then extracted and clustered. The selected motifs are subsequently re-integrated into the small protein library, guiding subsequent rounds of docking and optimization. (B) Two PBP design approaches targeting the SARS-CoV-2 Spike-RBD. (1) a protein binder design approach that starts from the ACE2 helix, and (2) a large-scale de-novo design of small helical scaffolds followed by rotamer interaction field docking to identify potential high-affinity binders.

图3 针对非常规/非序列性抗原表位的抗体探针设计方案。（A）针对疾病相关特定修饰位点的抗体组设计。（B）针对异常折叠传感的抗体设计。（C）针对多聚蛋白异常解离的抗体设计。

Fig.3 Antibody designing strategies targeting non-sequential epitopes.(A) A group of antibodies targeting disease-related post-translational modifications. (B) An antibody that detects abnormal protein folding. (C) An antibody that detects epitopes exposed during the abnormal dissociation of a multimeric protein.

参考文献 14

1	MITRA S, TOMAR P C. Hybridoma technology; advancements, clinical significance, and future aspects [J]. Journal of Genetic Engineering and Biotechnology, 2021, 19: 1-12.
2	HOOGENBOOM H R, DE BRUıNE A P, HUFTON S E, et al. Antibody phage display technology and its applications [J]. Immunotechnology, 1998, 4(1): 1-20.
3	STOLTENBURG R, REINEMANN C, STREHLITZ B. SELEX—A (r)evolutionary method to generate high-affinity nucleic acid ligands [J]. Biomolecular Engineering, 2007, 24(4): 381-403.
4	CAO L, GORESHNIK I, COVENTRY B, et al. De novo design of picomolar SARS-CoV-2 miniprotein inhibitors [J]. Science, 2020, 370(6515): 426-31.
5	CAO L, COVENTRY B, GORESHNIK I, et al. Design of protein-binding proteins from the target structure alone [J]. Nature, 2022, 605(7910): 551-60.
6	HIE B L, SHANKER V R, XU D, et al. Efficient evolution of human antibodies from general protein language models [J]. Nature Biotechnology, 2024, 42(2): 275-83.
7	HE H, HE B, GUAN L, et al. De novo generation of SARS-CoV-2 antibody CDRH3 with a pre-trained generative large language model [J]. Nature Communications, 2024, 15(1): 6867.
8	RUFFOLO J A, CHU L-S, MAHAJAN S P, et al. Fast, accurate antibody structure prediction from deep learning on massive set of natural antibodies [J]. Nature Communications, 2023, 14(1): 2389.
9	LI L, GUPTA E, SPAETH J, et al. Machine learning optimization of candidate antibody yields highly diverse sub-nanomolar affinity antibody libraries [J]. Nature Communications, 2023, 14(1): 3454.
10	JUMPER J, EVANS R, PRITZEL A, et al. Highly accurate protein structure prediction with AlphaFold [J]. Nature, 2021, 596(7873): 583-9.
11	ABRAMSON J, ADLER J, DUNGER J, et al. Accurate structure prediction of biomolecular interactions with AlphaFold 3 [J]. Nature, 2024: 1-3.
12	RIPOLL D R, CHAUDHURY S, WALLQVIST A. Using the antibody-antigen binding interface to train image-based deep neural networks for antibody-epitope classification [J]. Computational Biology, 2021, 17(3): e1008864.
13	CONTI S, LAU E Y, OVCHINNIKOV V. On the rapid calculation of binding affinities for antigen and antibody design and affinity maturation simulations [J]. Antibodies, 2022, 11(3): 51.
14	IWANO N, ADACHI T, AOKI K, et al. Generative aptamer discovery using RaptGen [J]. Nature Computational Science, 2022, 2(6): 378-86.

[1]	雷茹, 陶慧, 刘天罡. 基因组深度挖掘驱动微生物萜类化合物高效发现[J]. 合成生物学, 2024, 5(3): 507-526.
[2]	奚萌宇, 胡逸灵, 顾玉诚, 戈惠明. 基因组挖掘指导天然药物分子的发现[J]. 合成生物学, 2024, 5(3): 447-473.
[3]	杜瑶, 高宏丹, 刘家坤, 刘孝荣, 邢志浩, 张涛, 马东礼. CRISPR-Cas系统在病原核酸检测中的研究进展[J]. 合成生物学, 2024, 5(1): 202-216.
[4]	朱景勇, 李钧翔, 李旭辉, 张瑾, 毋文静. 深度学习在基于序列的蛋白质互作预测中的应用进展[J]. 合成生物学, 2024, 5(1): 88-106.
[5]	赵静宇, 张健, 祁庆生, 王倩. 基于细菌双组分系统的生物传感器的研究进展[J]. 合成生物学, 2024, 5(1): 38-52.
[6]	王晟, 王泽琛, 陈威华, 陈珂, 彭向达, 欧发芬, 郑良振, 孙瑨原, 沈涛, 赵国屏. 基于人工智能和计算生物学的合成生物学元件设计[J]. 合成生物学, 2023, 4(3): 422-443.
[7]	康里奇, 谈攀, 洪亮. 人工智能时代下的酶工程[J]. 合成生物学, 2023, 4(3): 524-534.
[8]	孟巧珍, 郭菲. “可折叠性”在酶智能设计改造中的应用研究——以AlphaFold2为例[J]. 合成生物学, 2023, 4(3): 571-589.
[9]	陈志航, 季梦麟, 戚逸飞. 人工智能蛋白质结构设计算法研究进展[J]. 合成生物学, 2023, 4(3): 464-487.
[10]	赖奇龙, 姚帅, 查毓国, 白虹, 宁康. 微生物组生物合成基因簇发掘方法及应用前景[J]. 合成生物学, 2023, 4(3): 611-627.
[11]	杨璐, 吴楠, 白茸茸, 董维亮, 周杰, 姜岷. 基因回路型全细胞微生物传感器的设计、优化与应用[J]. 合成生物学, 2022, 3(6): 1061-1080.
[12]	卞佳豪, 杨广宇. 人工智能辅助的蛋白质工程[J]. 合成生物学, 2022, 3(3): 429-444.
[13]	梁晓声, 郭永超, 门冬, 张先恩. 病毒-纳米金杂合导电网络结构在电化学分析的应用[J]. 合成生物学, 2022, 3(2): 415-427.
[14]	张萍, 魏文平, 周英, 叶邦策. 解脂耶氏酵母中光控表达系统的构建及其应用研究[J]. 合成生物学, 2021, 2(5): 778-791.
[15]	于慧敏, 郑煜堃, 杜岩, 王苗苗, 梁有向. 合成生物学研究中的微生物启动子工程策略[J]. 合成生物学, 2021, 2(4): 598-611.