Molecular approaches in pre-clinical models to identify and validate biomarker signatures that predict treatment response in prostate cancer patients

背景：选择性剪接（AS）是一种重要的基因调控机制，极大地扩展了蛋白质组的复杂性。前列腺癌（PC）每年导致超过11，000名英国男性死亡。失调的AS在PC进展中至关重要，特别是通过响应激素消融治疗产生组成性活性雄激素受体（AR）变体而明显。大规模RNA测序项目（包括TCGA）提供了将基因表达与临床信息联系起来的日益丰富的数据。AS事件在这些数据库中的调查很差。在100 - 1000例PC患者中确定异常AS事件将揭示预测治疗反应和治疗抗性发展的关键AS基因改变。这些新的生物标志物的签名将促进最佳的个性化治疗和疾病monitoring.Objectives：最初，我们将应用分子方法，使用临床前模型，以确定候选AS事件与PC进展到治疗耐药。在第二个目标中，机器学习工具将询问公开可用的RNA测序的PC患者队列，以识别预测对治疗的反应和治疗抵抗的AS事件的生物特征。这些生物特征将使用几种高度相关的临床前PC模型进行验证。实验方法：除了已建立的分子生物学技术，学生将应用siRNA，CRISPR技术和慢病毒感染来修改PC细胞系/类器官中的基因表达。将细胞暴露于AR拮抗剂并进行RNA测序。学生将在生物信息学分析中得到支持，以识别与细胞对药物治疗的敏感性/耐药性相关的多种AS事件，并在大型RNA测序PC患者队列中应用机器学习工具进行培训，以识别与患者治疗反应相关的AS事件的潜在特征（生化复发，生存）。将在我们的离体组织切片模型中使用多重、基于PCR的测定来检查AS事件特征的验证，该模型包括来自暴露于AR拮抗剂的未经治疗和治疗抗性组织的患者活检。这个跨学科的转化研究项目将提供癌细胞生物学、分子生物学技术、先进成像方法、生物信息学和机器学习方面的培训。新奇：通过访问由Supervisor-1和合作者开发的一套独特的临床前工具提供（组织切片，类器官，iPSC，CRISPR修饰的细胞）能够在细胞条件下彻底询问AS事件，该细胞条件非常模拟PC患者的治疗和治疗耐药性的出现。来源于对PC患者的可访问和内部（通过欧洲合作），注释良好的RNA测序临床队列的访问，包括治疗前和治疗后匹配的活检和最近可用的生物信息学软件包，这些软件包能够对AS事件进行稳健的评估，以及Supervisor-2的机器学习工具和AI专业知识。

Background: Alternative splicing (AS) of precursor-mRNAs is an important gene regulatory mechanism that massively expands proteome complexity. Prostate cancer (PC) accounts for >11,000 UK male deaths annually. Dysregulated AS is critically important in PC progression, notably evident by the generation of constitutively-active androgen receptor (AR) variants in response to hormone ablation therapy. Large-scale RNA-sequencing projects (including TCGA) provide an ever-expanding wealth of data linking gene expression to clinical information. AS events are poorly investigated in these databases. Determining aberrant AS events in 100s-1000s PC patients will reveal critical AS gene alterations that predict treatment response and development of therapy-resistance. These novel biomarker signatures will facilitate optimal personalized treatment and disease monitoring.Objectives: Initially we will apply molecular approaches using pre-clinical models to identify candidate AS events associated with PC progression to therapy-resistance. In the second Objective, machine learning tools will interrogate publically-available RNA-sequenced PC patient cohorts to identify bio-signatures of AS events that predict response to therapy and treatment-resistance. These bio-signatures will be validated using several highly relevant pre-clinical PC models. Experimental approach: Incorporating established molecular biology techniques, the student will apply siRNA, CRISPR technology and lentiviral infections to modify gene expression in PC cell lines/organoids. Cells will be exposed to AR antagonists and subjected to RNA-sequencing. The student will be supported in bioinformatics analysis to identify multiple AS events associated with cellular sensitivity/resistance to drug treatments and trained in the application of machine learning tools on large RNA-sequenced PC patient cohorts to identify potential signatures of AS events linked to patient treatment response (biochemical recurrence, survival). Validation of AS event signatures will be examined using multiplex, PCR-based assays in our ex vivo tissue slice model incorporating patient biopsies from treatment-naïve and therapy-resistant tissue exposed to AR antagonists. This interdisciplinary translational research project will provide training in cancer cell biology, molecular biology techniques, advanced imaging approaches, bioinformatics and machine learning.Novelty: provided by accessing a unique suite of pre-clinical tools developed by Supervisor-1 and collaborators (tissue slice, organoid, iPSCs, CRISPR-modified cells) enabling a thorough interrogation of AS events under cellular conditions that closely mimic treatment and emergence of therapy-resistance for PC patients.Timeliness: derived from access to publically accessible and in-house (through European collaboratives), well-annotated, RNA sequenced clinical cohorts of PC patients, including pre- and post-treatment matched biopsies and recently available bioinformatics packages that enable robust evaluation of AS events, alongside the machine learning tools and AI expertise of Supervisor-2.