Subtyping the autisms using individualized protein network analysis

项目摘要 自闭症是一种行为定义的诊断，在美国每59名儿童中就有1名受到影响。最近的基因 研究表明，自闭症是一个大家庭的总称， 常见的遗传性疾病，每个单独的基因只占总病例的一小部分。这 对于试图开发基于生物的药物或治疗策略的研究人员来说， 自闭症诊断实体太广泛，没有生物学意义，但大多数个体基因突变太广泛， 很少允许足够的患者招募或商业上可行的药物开发。目前迫切 未满足的需求是开发一种亚型分型策略，可以将患者分配到少数生物学上的 有意义的亚型，可能适合靶向治疗策略。我们最近开发了一种 一种新的蛋白质组学策略，可以对蛋白质-蛋白质相互作用网络进行高维测量 （PIN）。这些措施反映了自闭症发病机制的几个相关特征-突触内容，最近 活动和神经元的发育阶段。我们假设不同的遗传自闭症集中在两个 特定的PIN，并产生网络中断模式，这些模式虽然各自独特，但具有共同的 允许将PIN矩阵聚类为子类型的功能。重要的是，我们的聚类方法允许 识别定义每个亚型的特定信号转导节点， 我们建议的集群的信息。在已发表的概念验证工作中，我们能够将第七组 不同的小鼠模型，并对以前未知的分子病理进行预测。这里我们 我建议将这项工作扩展到人类神经元，使用从基因测序中提取的原代患者细胞， 自闭症研究对象与确定可能致病基因突变，或“特发性”自闭症患者， 进行测序，但没有发现突变，或年龄和性别匹配的典型发育对照。这 这项工作将揭示不同已知和未知自闭症之间新的生物学相关关系。 遗传病因学，并提供了机会，同时确定亚组的患者和潜在的 可以有效治疗每个确定的亚组的药物靶点。

PROJECT SUMMARY Autism is a behaviorally-defined diagnosis that affects approximately 1 in 59 children in the US. Recent genetic studies have revealed that autism is an umbrella term for a large family of individually rare, collectively common genetic disorders, with each individual gene accounting for only a small portion of total cases. This presents a problem for researchers attempting to develop biologically-based drug or treatment strategies: the autism diagnostic entity is too broad to be biologically meaningful, but most individual genetic mutations are too rare to allow for sufficient patient recruitment or for commercially viable drug development. There is an urgent unmet need to develop a subtyping strategy that can assign patients into one of a small number of biologically meaningful subtypes that might be amenable to targeted treatment strategies. We have recently developed a novel proteomic strategy that makes high-dimensional measurements of protein-protein interaction networks (PINs). These measures reflect several relevant features of autism pathogenesis- synaptic content, recent activity, and developmental stage of the neuron. We postulate that different genetic autisms converge on two specific PINs and produce patterns of network disruption that, while individually unique, share common features that will allow clustering of PIN matrices into subtypes. Importantly, our clustering methods allow identification of specific signal transduction nodes that define each sub-type, linking biologically-relevant information with our proposed clusters. In published proof-of-concept work, we were able to cluster seven different mouse models and make predictions about previously unknown molecular pathologies. Here, we propose to extend this work to human neurons, using primary patient cells taken from genetically sequenced autistic research subjects with identified likely causative genetic mutations, or `idiopathic' autism patients who were sequenced but no mutation was identified, or age-and-sex-matched typically developing controls. This work will reveal new, biologically relevant relationships between autisms of different known and unknown genetic etiologies, and offers the opportunity to simultaneously identify sub-groups of patients and potential drug targets that may effectively treat each identified sub-group.