School of Plant Sciences and Food Security, The George S. Wise Faculty of Life Sciences, Tel Aviv University, Ramat Aviv, Tel Aviv 69978, Israel.
Department of Biomedical Engineering, The Iby and Aladar Fleischman Faculty of Engineering, Tel Aviv University, Tel Aviv 6997801, Israel.
Nucleic Acids Res. 2019 Apr 23;47(7):3344-3352. doi: 10.1093/nar/gkz151.
While bacterial operons have been thoroughly studied, few analyses of chloroplast operons exist, limiting the ability to study fundamental elements of these structures and utilize them for synthetic biology. Here, we describe the creation of a plastome-specific operon database (link provided below) achieved by combining experimental tools and predictive modeling. Using a Reverse-Transcription-PCR based method and published data, we determined the transcription-state of 213 gene pairs from four plastomes of evolutionary distinct organisms. By analyzing sequence-based features computed for our dataset, we were able to highlight fundamental characteristics differentiating between operon pairs and non-operon pairs. These include an interesting tendency toward maintaining similar messenger RNA-folding profiles in operon gene pairs, a feature that failed to yield any informative separation in cyanobacteria, suggesting that it catches unique traits of operon gene expression, which have evolved post-endosymbiosis. Subsequently, we used this feature set to train a random-forest classifier for operon prediction. As our results demonstrate the ability of our predictor to obtain accurate (84%) and robust predictions on unlabeled datasets, we proceeded to building operon maps for 2018 sequenced plastids. Our database may now present new opportunities for promoting metabolic engineering and synthetic biology in chloroplasts.
虽然细菌操纵子已经得到了深入研究，但对叶绿体操纵子的分析却很少，这限制了对这些结构基本要素的研究，并限制了它们在合成生物学中的应用。在这里，我们通过结合实验工具和预测模型，描述了一个质体特异性操纵子数据库（见下文链接）的创建。我们使用基于逆转录聚合酶链反应的方法和已发表的数据，确定了来自四个进化上不同生物体的 213 对质体基因对的转录状态。通过分析为我们的数据集计算的基于序列的特征，我们能够突出区分操纵子对和非操纵子对的基本特征。这些特征包括在操纵子基因对中保持相似信使 RNA 折叠谱的有趣趋势，这一特征在蓝藻中未能产生任何有信息性的分离，表明它捕捉到了操纵子基因表达的独特特征，这些特征是在内共生后进化而来的。随后，我们使用这个特征集来训练一个随机森林分类器进行操纵子预测。正如我们的结果所证明的，我们的预测器能够在未标记的数据集中获得准确（84%）和稳健的预测，因此我们继续为 2018 年测序的质体构建操纵子图谱。我们的数据库现在可能为促进叶绿体中的代谢工程和合成生物学提供新的机会。
