Systematic identification of novel µ-proteins in bacteria using ribosome profiling data

新出现的证据表明，短蛋白（µ蛋白）在生理过程中发挥着更重要的作用。然而，它们的从头识别是困难的，并且由于µ-蛋白的长度较短，通常小于150 nt或小于50个密码子，因此在搜索新的开放阅读框的算法中经常会遗漏它们。在本项目中，我们试图解决目前在系统识别编码微蛋白的细菌中的短orf （sorf）的算法/工具方面的空白。我们将利用RNA-Seq和核糖体分析两种深度测序技术的力量，提取sorf的转录组表达特征，并将其用于设计表达sorf的从头搜索算法。这些特征包括翻译开始，报告蛋白质合成终止的核糖体释放评分，最重要的是，当真正翻译ORF时，核糖体保护片段中的三个核苷酸周期性。为了忠实地确定所有起始位点（包括非规范起始位点），我们将对一种新发现的肽抗生素抑制翻译起始的转录组进行测序。由于细菌通常采用类似的表达规则，因此我们的愿景是使用在各种条件下生长的大肠杆菌MG1655产生的数据集训练算法，并使其适用于所有细菌物种。与仅使用遗传信息的算法相比，利用RNA-Seq和核糖体分析的表达特征可以预测真正表达的sorf，这在检测sorf方面是向前迈出的一步。µ-蛋白只有在一定的生长或环境条件下才能表达。因此，为了解决它们在形成应激反应中的作用，我们的目标是使用核糖体分析和RNA-Seq数据集来探测各种应激（热、氧化和渗透应激）下sORF在翻译和转录水平上的表达。我们将对这些新发现的微蛋白（最好是那些仅在某些胁迫条件下表达的微蛋白）进行进一步的表征，以阐明在蛋白质水平上的表达和使用基因组标记进行下拉实验的相互作用伙伴。

Emerging evidence places short proteins (µ proteins) more centrally in physiological processes. However, their de novo identification is difficult and they are often missed by algorithms searching for new open-reading frames because of the short size of the µ-proteins, often <150 nt or <50 codons. In this project, we seek to address the current void in algorithms/tools for systematic identification of short ORFs (sORFs) in bacteria which encode µ-proteins. We will use the power of two deep-sequencing technologies, RNA-Seq and ribosome profiling, to extract transcriptome-wide expression features of sORFs and use them in designing the algorithm for de novo search of expressed sORFs. These features include translational start, ribosome release score reporting on termination of protein synthesis, and most importantly, the three-nucleotide periodicity in ribosome-protected fragments when truly translating an ORF. To faithfully determine all initiation sites (including non-canonical starts), we will sequence the transcriptome upon inhibition of translation initiation with a newly identified peptide antibiotic. Since bacteria in general employ similar expression rules, it is our vision to train the algorithm with data sets produced in E. coli MG1655 grown at various conditions, and to develop it to work across all bacterial species. Leveraging expression features from RNA-Seq and ribosome profiling, which allow predicting only truly expressed sORFs, is a step forward in detecting sORFs as compared to the algorithms using solely genetic information.µ-proteins might be expressed only under certain growth or environmental conditions. Thus, to address their role in shaping stress response, we aim to probe sORF expression at both translation and transcription level at various stresses (heat, oxidative and osmotic stress) using ribosome profiling and RNA-Seq data sets. We will carry out further characterization of some of these newly discovered µ-proteins (preferably those that are expressed only under certain stress conditions) to elucidate expression on the protein level and interaction partners using genomic tagging for pulldown experiments.