The XPPAUT ode files for the network in Fig 1D.

Protein synthesis involves translation initiation, elongation, termination, and ribosome recycling, and each step is controlled intricately by many signaling proteins. Translation initiation can be compactly categorized into two mechanisms: primary and secondary. The primary mechanism involves the recruitment of three important eukaryotic initiation factors, eIF2-GDP, eIF5, and eIF2B, and their interactions, followed by the GDP-GTP exchange by eIF2B to form an active dimer eIF2-GTP. The dimer binds with Met-tRNA to form a robust ternary complex (TC). The secondary mechanism closely mirrors the primary reaction mechanism, except that the interactions of eIF2B and eIF5 happen with the TC to form complexes. These interactions happen with high fidelity and precision, failing which fail-safe mechanisms are invoked instantaneously to delay the initiation process. In this work, we build a mathematical model to unravel how the transition between translation initiation and termination occurs at the initiation stage based on the elementary mechanisms we built from the network assembled from experimental observations. We focus only on the dynamics of primary and secondary mechanisms involved in the translation initiation process under normal and integrated stress response (ISR) conditions that act as a fail-safe mechanism by through phosphorylation-dephosphorylation (PdP) reactions. Since the network is huge and has many unknown kinetic parameters, we perform structural analysis using chemical reaction network theory (CRNT) and find hidden positive feedback loops that regulate the initiation mechanism. We apply bifurcation theory to show that the model exhibits ultrasensitivity and bistability under normal conditions, while under ISR, it exhibits both bistability and tristability for the choice of kinetic parameters. We attribute bistability to translation initiation and termination and tristability in ISR to translation recovery and attenuation. We conclude that the translation initiation process is a highly regulated process guided by the threshold and switching mechanisms to make quick decisions on the translation initiation, termination, recovery or attenuation under different conditions.

蛋白质合成过程涵盖翻译起始、延伸、终止与核糖体循环，每一步均受众多信号蛋白的精密调控。翻译起始可被简洁地划分为两种机制：初级机制与次级机制。
初级机制涉及三种关键真核起始因子(eukaryotic initiation factor, eIF)2-GDP、eIF5以及eIF2B的招募与相互作用，随后由eIF2B催化GDP与GTP交换，形成具有活性的二聚体eIF2-GTP。该二聚体与Met-tRNA结合，形成稳定的三元复合物(ternary complex, TC)。
次级机制与初级反应机制高度相似，唯一区别在于eIF2B与eIF5与TC结合形成复合物的过程。上述相互作用具有极高的保真度与精准度，一旦出现差错，机体将立即启动故障安全机制以延迟翻译起始进程。
本研究基于实验观测构建的调控网络，建立了数学模型，以解析翻译起始阶段中翻译起始与终止之间的转变规律。我们仅聚焦于正常条件与整合应激反应(integrated stress response, ISR)条件下，翻译起始过程中初级与次级机制的动力学特征——整合应激反应可通过磷酸化-去磷酸化(phosphorylation-dephosphorylation, PdP)反应作为故障安全机制发挥作用。
由于该调控网络规模庞大且存在诸多未知的动力学参数，我们借助化学反应网络理论(chemical reaction network theory, CRNT)开展结构分析，挖掘出调控翻译起始机制的隐藏正反馈环路。我们应用分歧理论(bifurcation theory)证明，该模型在正常条件下可表现出超敏性与双稳态；而在整合应激反应条件下，针对特定动力学参数选择，模型可同时呈现双稳态与三稳态。
我们将双稳态归因于翻译起始与终止过程，将整合应激反应下的三稳态归因于翻译恢复与翻译衰减过程。本研究最终得出结论：翻译起始过程是受严密调控的过程，其通过阈值与切换机制实现快速决策，以在不同条件下调控翻译起始、终止、恢复或衰减。