High-throughput discovery of plant metabolic enzyme function using integrative approaches

多样性是地球上生命最显著的特征之一，自文明诞生以来，多样性一直是人类的魅力之源。这种多样性的一个方面是不同的有机化合物实现不同的功能，如毒药、引诱剂、驱虫剂、信使、能量储存分子等等。植物在所有分类群中产生超过100万种多样而复杂的代谢物，尤其以其产生的化合物的异常多样性而闻名。这种多样性的出现是由代谢酶促进的，其中许多是由串联或全基因组复制产生的大型酶家族的一部分。这些基因家族的成员具有共享蛋白结构域、功能冗余、低底物特异性、混杂性和基因复制后快速功能分化的特点。由于这些原因，计算预测酶家族成员的功能一直是一个困难的努力。例如拟南芥（Arabidopsis thaliana）和番茄茄（Solanum lycopersicum）， 80%以上的基因都是基因家族的成员，其中许多成员的注释很差。这种糟糕的注释为理解植物表型多样性的起源和利用合理的方法为经济目的设计新的植物性状造成了障碍。本研究的总体目标是开发计算和湿实验室方法来预测具有未知功能的酶的假定底物。虽然将对多个酶家族进行生物信息学分析，但我计划将重点放在BAHD家族作为计算建模的模型酶家族上。我将利用比较基因组学的力量，首先编译植物中多种BAHD酶的生化知识，然后开发系统发育指导的预测模型，通过底物相似性预测酶底物。然后，我将确定两个互补的证据线——通过RNA-seq和代谢组学分析获得的转录代谢物相关性，以及酶家族基因的过表达表型——是否有助于补充和验证所开发的计算模型。提出的实验将尝试使用新颖的多学科技术解决植物生物学中一个长期存在的问题。这些实验结果将为理解蛋白质结构的进化和重复基因的进化奠定基础。预测酶功能的能力可以促进通过RNA-seq、QTL定位或GWAS获得的候选代谢基因的注释，从而使非常广泛的植物科学界受益。这些功能的发现有助于作物的合理工程设计和天然产物合成途径的设计。最后，建议的方法将为我自己的多学科培训以及在美国和德国的合作和网络提供重要的机会，以进一步推进我的职业生涯。

Diversity is one of the most remarkable features of life on earth and has been a source of fascination for mankind since the birth of civilization. One aspect of this diversity are the different organic compounds fulfilling various functions as poisons, attractants, repellents, messengers, energy storage molecules, and more. Plants, producing over a million diverse and complex metabolites across all taxa, especially are renowned for their exceptional diversity of produced compounds. The emergence of this diversity is facilitated by metabolic enzymes, many of which are part of large enzyme families generated by tandem or whole genome duplications. Members of these gene families are characterized by shared protein domains, functional redundancy, low substrate specificity, promiscuity, and rapid functional divergence after gene duplication. For these reasons, predicting functions of enzyme family members computationally has been a difficult endeavor. For example, in Arabidopsis thaliana and Solanum lycopersicum, more than 80% of all genes are members of genes families, and many of these members are poorly annotated. Such poor annotation creates obstacles in understanding the origins of plant phenotypic diversity and in utilizing rational approaches to engineer novel plant traits for economic purposes.The overall aim of this study is to develop computational and wet-lab approaches for predicting putative substrates of enzymes with unknown function. Although multiple enzyme families will be analyzed bioinformatically, I plan to focus on the BAHD family as a model enzyme family for computational modeling. I will utilize the power of comparative genomics, by first compiling biochemical knowledge about multiple BAHD enzymes characterized in plants, followed by developing phylogeny-guided predictive models for enzyme substrate prediction by substrate similarity. I will then determine if two complementary lines of evidences – transcript-metabolite correlations obtained through RNA-seq and metabolomic analyses, and overexpression phenotypes of enzyme family genes – help in supplementing as well as validating the computational models developed.The proposed experiments will attempt solving a long-standing problem in plant biology using novel, multi-disciplinary technologies. Results of these experiments will create a foundation for understanding the evolution of protein structure and evolution of duplicate genes. The ability to predict enzyme function can boost annotation of candidate metabolic genes obtained through RNA-seq, QTL mapping or GWAS, benefiting a very broad plant science community. Such functional discovery can aid rational engineering of crops and design of synthetic pathways for natural product synthesis. Finally, the proposed approaches will generate significant opportunities for my own multi-disciplinary training and for collaborations and networking in USA and Germany to further advance my career.