BRITE-Eye: An integrated discovery engine for CNS therapeutic targets driven by high throughput genetic screens, functional readouts in human neurons, and machine learning

项目摘要 神经系统疾病影响着全世界数百万患者，是一个主要的未满足的医疗需求。最近 开发新种类的中枢神经系统（CNS）治疗剂的进展相对滞后， 其他疾病领域。中枢神经系统药物发现过程中的一个关键障碍是需要细胞模型， 分析和技术，可以更可靠地评估疾病相关的神经生理参数， 在单个神经元和突触水平上的人类细胞背景，具有捕获的规模和分辨率 这些系统的复杂性和可变性。我们建议通过以下三个方面的整合来满足这一需求： 关键技术-（i）我们用于人类神经元全光学生理学的高通量BRITETM平台， 实现单细胞和单动作电位分辨率，每天的吞吐量约为500，000个神经元， 仪器;（ii）使用CRISPR核酸酶破坏基因功能的基因组筛选;（iii）机器学习 以单细胞分辨率鉴定代表复杂生理表型的指纹。这 第二阶段计划包括四个主要目标。1)在人类神经元中建立CRISPRn筛选条件。 我们将选择20个候选靶基因，包括癫痫和神经发育靶点，进一步优化 与全光学生理表型相容的测定条件，包括遗传破坏的时间， CRISPRn/gRNA组分的浓度以有效敲低基因靶标。2)构建深度学习- 用于单细胞表型分析的动力分析工具。我们将使用深度神经网络来学习一个紧凑的向量 利用我们的单细胞分辨率，在药物干预后呈现神经元行为 测量和适应神经元群体中的潜在异质性。3)鉴定遗传 使用全基因组CRISPRn筛选神经元功能的调节剂。我们将联合收割机 在目标1-2中建立的条件和分析模型，以进行全基因组CRISPRn筛选（> 18，000 基因靶标）与野生型人iPSC-兴奋性神经元中的阵列gRNA文库。我们会找出基因 其下调导致功能参数显著变化的靶点。潜在命中和特异性 将使用单一gRNA和qPCR以及免疫印迹在独立轮次中确认靶敲低 测定。4)在脆性X综合征的人iPSC-神经元模型中预测和验证表型拯救。 最后，我们将评估目标3中开发的功能指纹的预测价值，以生成一个 可以拯救（抑制）我们在人类中鉴定的表型参数的基因靶点候选列表 神经发育障碍的细胞模型，脆性X综合征。我们将调节 在FMR 1-/y iPSC-神经元和基准表型拯救中， 使用FMRP的基因再引入。成功完成拟议的工作有可能产生一个新的 了解人类神经生理学的分子结构，并为新的治疗方法提供平台。 靶点鉴定集中在调节神经生理疾病机制的分子基础上。

Project Summary Neurological disorders affect millions of patients worldwide and represent a major unmet medical need. Recent progress on developing new classes of central nervous system (CNS) therapeutics has lagged compared to other disease areas. A key obstacle in the CNS drug discovery process has been a need for cellular models, assays, and technologies that can more reliably assess disease-relevant neurophysiological parameters in a human cellular context at the level of individual neurons and synapses, with the scale and resolution to capture the complexity and variability of these systems. We propose to address this need through the integration of three key technologies – (i) our high throughput BRITETM platform for all-optical physiology in human neurons, which achieves single-cell and single-action-potential resolution with a throughput of ~500,000 neurons per day per instrument; (ii) genomic screens using CRISPR nuclease to disrupt gene function; (iii) machine learning for identification of fingerprints that represent complex physiological phenotypes with single-cell resolution. This Phase II program includes four key objectives. 1) Establish CRISPRn screening conditions in human neurons. We will select 20 candidate target genes, including epilepsy and neurodevelopmental targets to further optimize assay conditions compatible with all-optical physiology phenotyping, including timing of genetic disruption and concentration of CRISPRn/gRNA components for effective knockdown of gene targets. 2) Build deep-learning- powered analytical tools for single-cell phenotyping. We will use deep neural networks to learn a compact vector representation of neuronal behavior after pharmacological intervention that leverages our single cell resolution measurements and accommodates potential heterogeneity in the population of neurons. 3) Identify genetic modulators of neuronal function using a genome-wide CRISPRn screen. We will combine experimental conditions and analytical models established in Aims 1-2 to carry out a genome-wide CRISPRn screen (>18,000 gene targets) with arrayed gRNA libraries in wild-type human iPSC-excitatory neurons. We will identify gene targets whose downregulation leads to significant changes in functional parameters. Potential hits and specificity of target knockdown will be confirmed in independent rounds using single gRNA and qPCR and immunoblotting assays. 4) Predict and validate phenotypic rescue in a human iPSC-neuronal model of Fragile X Syndrome. Finally, we will assess the predictive value of the functional fingerprints developed in Aim 3 to generate a candidate list of gene targets that can rescue (suppress) phenotypic parameters we have identified in a human cellular model of the neurodevelopmental disorder, Fragile X syndrome. We will modulate the expression of these potential genetic suppressors with CRISPRn in FMR1-/y iPSC-neurons and benchmark phenotypic rescue using genetic re-introduction of FMRP. Successful completion of the proposed work has potential to yield a new understanding of the molecular architecture of human neurophysiology and a platform for novel therapeutic target identification focused on the molecular basis for modulation of neurophysiological disease mechanisms.