《生命科学》 2026, 38(7): 1284-1300
时空单细胞多组学解码大脑衰老：分子机制、干预主线与智能因果展望
摘 要：
大脑衰老是多细胞、多分子层级的复杂过程，同时也是阿尔茨海默病、帕金森病等神经退行性疾病的核心风险 因素。传统组织水平研究受细胞异质性掩盖，长期难以揭示关键分子事件，形成认知瓶颈，这一局限推动了技术层面的 创新探索。近年来，单细胞多组学与空间转录组学（spatial transcriptomics，ST）相继取得突破：前者清晰解析了不同细胞 群体的衰老动态轨迹，明确少突胶质前体细胞（OPCs）、小胶质细胞的衰老易感性，神经元突触相关基因失调及染色质- 转录网络重构规律，后者则进一步揭示非神经元细胞空间紊乱与免疫微环境失衡机制，为大脑衰老的多维度认知提供 技术支撑。在干预领域，单细胞多组学技术的应用价值进一步拓展，不仅为潜在干预靶点筛选提供精准工具，也为药物 研发、细胞疗法及多维度联合干预策略搭建验证平台。衰老细胞清除（senolytics）、NAD+补充介导的能量代谢重塑、神 经-免疫-血管单元稳态恢复等方向，已借助该技术构建“靶点筛选-机制验证-临床前转化”的完整研究链条，为大脑衰老 的精准干预奠定坚实基础。展望未来，人工智能与因果扰动技术（如CRISPR-perturb-seq、谱系追踪）将与单细胞多组学 深度融合：一方面，AI可通过多模态数据整合构建更精准的大脑“衰老时钟”及个体化风险预测模型；另一方面，因果扰 动实验推动研究从单纯的相关性描述，迈向机制验证与干预效果模拟的新阶段，进而实现从静态图谱绘制到动态干预 设计的跨越；同时，跨物种模型与类器官系统将成为关键验证环节，助力缩短基础研究到临床应用的转化周期。本综述 系统总结截至2025年大脑衰老领域的最新研究进展，重点阐释了单细胞多组学技术在机制解析、干预靶点筛选及策略 评估中的核心价值，同时展望人工智能与因果扰动技术结合的研究新范式。这些进展将共同推动该领域迈入精准化、 智能化与个性化的新时代，为健康老龄化推进及神经退行性疾病防治提供重要理论依据与技术支撑。
通讯作者：王亚云 , Email:wangyy@fmmu.edu.cn 阮彩莲 , Email:rcl1157@163.com
Abstract:
Brain aging is a complex process involving interrelated changes at cellular and molecular levels, and it represents a core risk factor for neurodegenerative diseases such as Alzheimer′s disease and Parkinson′s disease. Conventional tissue studies are constrained by cellular heterogeneity, making it difficult to uncover cell-specific molecular events and key regulatory mechanisms associated with brain aging. Single-cell multi-omics and spatial transcriptomics (ST) serve as transformative technologies that provide the foundation for overcoming this challenge. This review summarizes advances in brain aging research through to 2025, elucidating the core value of these technologies in deciphering aging mechanisms, identifying intervention targets, and evaluating therapeutic strategies. It explores emerging paradigms integrating AI with causal perturbation techniques, providing a comprehensive overview on related fields. In mechanism analysis, single-cell multi-omics overcomes the limitations of bulk tissue analysis, enabling high-resolution profiling of aging dynamics across distinct brain cell populations. Studies confirm impaired differentiation capacity in oligodendrocyte precursor cells (OPCs), altered expression of myelin-related genes, and a shift toward a pro-inflammatory phenotype with phagocytic dysfunction in microglia. In neurons, dysregulation of synapse-related genes, along with chromatin accessibility and transcriptional regulatory network reorganization, drives aging. Spatial transcriptomics further elucidates spatial disorganization in non- neuronal cells and imbalances in the brain′s immune microenvironment, highlighting the importance of region-specific aging patterns. In intervention studies, this technology provides precise tools for target screening, establishing a validation platform for drug development, cell therapies, and combined interventions. It builds a chain from “target discovery → mechanism validation → preclinical translation”, laying the foundation for precision interventions. Future advancements will be driven by the convergence of AI, causal perturbation techniques (e.g., CRISPR-perturb-seq, lineage tracing), and single-cell multi- omics. AI can integrate multimodal data to construct precise “aging clocks” and risk prediction models for early identification. Causal perturbation techniques will advance research from correlation studies toward mechanism validation and intervention simulation. Cross-species models and organoid systems will bridge the gap between basic research and clinical applications, accelerating translation. Future research should focus on building integrated technology platforms, fostering interdisciplinary collaboration, and promoting translational outcomes to address the global challenge of brain aging-related diseases.
Communication Author：WANG Ya-Yun , Email:wangyy@fmmu.edu.cn RUAN Cai-Lian , Email:rcl1157@163.com
SciCloud投稿系统