中国神经再生研究(英文版)

中国神经再生研究(英文版) ›› 2026, Vol. 21 ›› Issue (8): 3754-3768.doi: 10.4103/NRR.NRR-D-24-01650

• 原著:退行性病与再生 • 上一篇    下一篇

揭示阿尔茨海默病免疫微环境中RNA甲基化调控模式：基于机器学习RNA测序与单细胞分析

  

  • 出版日期:2026-08-18 发布日期:2026-04-27
  • 基金资助:


 Integrated machine learning–based RNA sequencing and single-cell analysis reveal RNA methylation regulation patterns in the immune microenvironment of Alzheimer’s disease

Shuguang Wu1, #, Ting Guo2, #, Xingyongpei Zheng3, #, Caihong Gu4, Yujie Hu5, Xinru Gu3, Xinyu Zhou3, 4, 6, *   

  1. 1Department of Anesthesiology, The Second Affiliated Hospital, Zhejiang University School of Medicine, Hangzhou, Zhejiang Province, China; 
    2Department of Geriatric Medicine, Shanghai Sixth People’s Hospital Affiliated to Shanghai Jiao Tong University School of Medicine, Shanghai, China; 
    3Department of Neurology, The Affiliated Lianyungang Hospital of Xuzhou Medical University, Lianyungang, Jiangsu Province, China; 
    4Department of Neurology, The First Affiliated Hospital of Kangda College of Nanjing Medical University, Lianyungang, Jiangsu Province, China; 
    5Department of Neurology, Nanjing Drum Tower Hospital, Affiliated Hospital of Medical School, Nanjing University, Nanjing, Jiangsu Province, China; 
    6Department of Neurology, Lianyungang Clinical College of Nanjing Medical University, Lianyungang, Jiangsu Province, China
  • Online:2026-08-18 Published:2026-04-27
  • Contact: Xinyu Zhou, MD, zhouxy0712@126.com.
  • Supported by:
    This study was supported by the Elderly Health Research Project of Jiangsu Province, No. LKM2023043 (to XZ).

PDF

6

摘要：

RNA甲基化水平的改变可能影响阿尔茨海默病的发生和发展。然而，RNA甲基化与阿尔茨海默病之间的确切关系尚不明确。实验采用基于机器学习的RNA测序和单细胞RNA测序分析了RNA甲基化水平。通过分子生物学技术验证了RNA甲基化调节因子的表达水平。采用共表达网络分析识别相关长非编码RNA。对与RNA甲基化相关的分子亚型进行分类，并评估了不同亚型在临床特征、生物学行为和免疫特征方面的差异。机器学习方法被用于识别与甲基化相关的长非编码RNA，这些甲基化相关的长非编码RNA被用于构建阿尔茨海默病的风险模型和 nomogram。预测了不同风险组的潜在治疗药物，并通过体外实验评估了关键RNA甲基化的表达。单细胞分析显示，阿尔茨海默病患者的RNA甲基化谱更为显著，尤其在T细胞、B细胞和NK细胞中。随后通过定量逆转录聚合酶链反应和Western blot验证，证实了Aβ寡聚体诱导的神经元中RNA甲基化调节因子的改变。这一证据支持将阿尔茨海默病患者分类为异质性亚型。具体而言，亚型1被识别为免疫活性亚型，而亚型2则表现出代谢表型。机器学习算法识别出五种与甲基化相关的长非编码RNA——LINC01007、MAP4K3-DT、MIR302CHG、VAC14-AS1和TGFB2-OT1——这些基因能准确预测阿尔茨海默病患者的临床预后。这些患者被分为低风险和高风险两组；后者表现出更高的免疫浸润、上调的免疫调节基因表达、升高的免疫评分，并对花生四烯酸-三氟乙烷治疗有更好的反应。这些发现表明，RNA甲基化的紊乱会改变阿尔茨海默病的免疫微环境，并与疾病进展密切相关。这一现象为针对RNA甲基化的阿尔茨海默病潜在治疗策略提供了新的见解。


https://orcid.org/0000-0002-9647-1100 (Xinyu Zhou)

关键词: 阿尔茨海默病, RNA甲基化, 长非编码RNA, 免疫, 机器学习, 风险模型

Abstract: Alterations in RNA methylation may affect the initiation and development of Alzheimer’s disease. However, the exact nature of the relationship between RNA methylation and Alzheimer’s disease remains unclear. In this study, RNA methylation levels were analyzed by bulk transcriptomic and single-cell RNA sequencing. The expression levels of RNA methylation regulators were confirmed using molecular biology techniques. Co-expression network analysis was used to identify relevant long non-coding RNAs. Molecular subtypes related to RNA methylation were classified, and variations in clinical characteristics, biological behavior, and immune signatures between subtypes were assessed. Machine learning approaches were applied to identify methylation-associated long non-coding RNAs, which were used to construct a risk model and nomogram for Alzheimer’s disease. Potential therapeutic agents for different risk groups were predicted, and in vitro experiments were conducted to identify key RNA methylation events. Single-cell analysis demonstrated enhanced RNA methylation in patients with Alzheimer’s disease, particularly within T cells, B cells, and NK cells. Quantitative reverse transcription-polymerase chain reaction and western blot confirmed alterations in RNA methylation regulators in neurons treated with amyloid-β oligomers in vitro. This evidence supported the classification of patients with Alzheimer’s disease into heterogeneous subtypes. Specifically, subtype 1 was identified as the immune-active subtype, while subtype 2 was characterized by a metabolic phenotype. Machine learning algorithms identified five significant methylation-associated long non-coding RNAs —LINC01007, MAP4K3-DT, MIR302CHG, VAC14-AS1, and TGFB2-OT1—that accurately predict clinical outcomes for patients with Alzheimer’s disease. These patients were classified into low- and high-risk categories; the latter group displayed higher immune infiltration, upregulated immune regulatory gene expression, and elevated immune scores and responded better to treatment with arachidonic-trifluoroethane. These findings suggest that dysregulated RNA methylation alters the immune microenvironment in Alzheimer’s disease and is closely associated with its progression. This phenomenon provides novel insights into potential therapeutic strategies for Alzheimer’s disease that target RNA methylation.

Key words: Alzheimer’s disease, immunity, long non-coding RNAs, machine learning, nerve regeneration, risk model, RNA methylation