Targeting tumor-intrinsic neural vulnerabilities of glioblastoma

The neural behavior of glioblastoma, including the formation of tumor microtubes and synaptic circuitry, is increasingly understood to be pivotal for disease manifestation (Osswald et al. 2015; Venkatesh et al. 2015; Weil et al. 2017; Venkataramani et al. 2019; Venkatesh et al. 2019; Alcantara Llaguno et al. 2019; Venkataramani et al. 2022). Nonetheless, the few approved treatments for glioblastoma target its oncological nature, while its neural vulnerabilities remain incompletely mapped and clinically unexploited. Here, we systematically survey the neural molecular dependencies and cellular heterogeneity across 27 glioblastoma patients and diverse model systems. In patient tumor samples taken directly after surgery, we identify a spectrum of neural stem cell morphologies indicative of poor prognosis, and discover a set of repurposable neuroactive drugs with unexpected and consistent anti-glioma efficacy. Glioblastoma cells exhibit functional dependencies on highly expressed drug targets including neurological ion channels and receptors, while interpretable molecular machine learning reveals downstream convergence on secondary drug targets (COSTAR) involving AP-1-driven tumor suppression. COSTAR enables in silico drug screening on >1 million compounds that are validated with high accuracy. Multi-omic profiling of drug-treated glioblastoma cells confirms rapid Ca2+-driven AP-1 pathway induction to represent a tumor-intrinsic vulnerability at the intersection of oncogenesis and neural activity-dependent signaling. Finally, the consistent anti-glioma activity across patients and model systems is epitomized by the antidepressant Vortioxetine, which synergizes in vivo with approved glioblastoma chemotherapies. In all, our global analysis reveals that the neural vulnerabilities of glioblastoma converge on an AP-1 mediated gene regulatory network with direct translatable potential.

胶质母细胞瘤（glioblastoma）的神经生物学行为——包括肿瘤微管（tumor microtubes）形成与突触回路构建——现已被越来越多的研究证实与疾病发生发展密切相关（Osswald等人，2015；Venkatesh等人，2015；Weil等人，2017；Venkataramani等人，2019；Venkatesh等人，2019；Alcantara Llaguno等人，2019；Venkataramani等人，2022）。然而，目前获批的胶质母细胞瘤治疗手段均仅针对其肿瘤学特性，而其神经层面的脆弱靶点仍未被完全阐明，且尚未在临床中得到开发利用。本研究系统性调研了27名胶质母细胞瘤患者及多种模型系统中的神经分子依赖特征与细胞异质性。在术后即刻采集的患者肿瘤样本中，我们鉴定出一系列与不良预后相关的神经干细胞形态特征，并发现了一批可实现老药新用的神经活性药物，其具备意外且稳定的抗胶质瘤功效。胶质母细胞瘤细胞对高表达的药物靶点（包括神经离子通道与受体）存在功能依赖；同时，可解释性分子机器学习模型揭示，其下游汇聚于涉及AP-1介导肿瘤抑制的次级药物靶点（COSTAR）。COSTAR可实现超过100万种化合物的计算机虚拟筛选，且筛选结果具有较高验证准确率。对药物处理后的胶质母细胞瘤细胞进行多组学分析证实，钙依赖的AP-1通路快速激活，代表了肿瘤发生与神经活性依赖信号通路交汇点处的肿瘤固有脆弱靶点。最终，在所有患者与模型系统中均展现出一致抗胶质瘤活性的代表药物为抗抑郁药沃替西汀（Vortioxetine），其与获批的胶质母细胞瘤化疗药物在体内具有协同作用。综上，本研究的全局分析揭示，胶质母细胞瘤的神经脆弱靶点汇聚于AP-1介导的基因调控网络，具备直接的转化应用潜力。