Phenotypic diversity and temporal variability in a bacterial signaling network revealed by single-cell FRET.
Phenotypic diversity and temporal variability in a bacterial signaling network revealed by single-cell FRET.
复制标题
DOI：
10.7554/elife.27455
复制
发表时间：
2017-12-12
期刊：
影响因子：
7.7
通讯作者：
Shimizu TS
中科院分区：
文献类型：
--
作者：
Keegstra JM;Kamino K;Anquez F;Lazova MD;Emonet T;Shimizu TS
我们在大肠杆菌趋化系统中进行了体内单细胞 FRET 测量，揭示了随着时间的推移，同基因群体中的细胞之间和单个细胞内普遍存在的信号变化。我们量化了细胞间适应、配体反应以及稳态输出水平的变异性，并分析了网络设计在从基因表达噪声中塑造这种多样性中的作用。在基因表达没有变化的情况下，我们发现单细胞表现出强烈的时间波动。我们提供的证据表明，这种信号噪声至少有两个来源：（i）适应酶的随机活动，以及（ii）在没有适应的情况下受体激酶动力学。我们证明，在某些条件下，（ii）可以产生巨大的波动，驱动整个细胞的信号活动进入随机的两种状态切换状态。我们的研究结果强调了分子噪声的重要性，分子噪声不仅出现在基因表达中，而且出现在蛋白质网络中。许多复杂的计算机程序使用随机数生成器来帮助解决具有挑战性的问题。这些问题的范围从实现互联网上的安全通信到决定如何最好地投资股票市场。近年来的大量研究发现，随机性在活细胞中也很普遍，通常被称为“噪声”。例如，某些基因的活性是如此不可预测，以至于看起来是随机的。然而，人们对这种基因表达噪音如何传播并改变细胞行为的了解相对较少。关于细胞如何利用这些或其他波动来完成复杂的任务，就像人们使用随机数生成器一样，仍然存在许多悬而未决的问题。细菌执行许多复杂的任务。有些细菌会游向暗示潜在奖励的化学物质，例如食物。然而，它们会游走，远离可能导致它们受到伤害的化学物质。这种能力称为趋化性，它依赖于相互作用的酶和其他蛋白质的网络，该网络协调细菌的运动与其感官的输入。基格斯特拉等人。着手寻找可能充当随机数生成器并帮助大肠杆菌最好地执行趋化性的噪声源。当参与趋化网络的两种蛋白质在单个细菌内相互作用时，使用一种称为体内 Förster 共振能量转移（或简称体内 FRET）的技术的改进版本来发出可检测的信号。实验表明，这个蛋白质网络放大了某些基因的基因表达噪音，同时减少了其他基因的基因表达噪音。此外，即使基因表达噪声被消除，基因编码的蛋白质之间的相互作用也会成为额外的噪声源。基格斯特拉等人。发现通过体内 FRET 测量，趋化网络内的信号量随时间变化很大。这揭示了蛋白质信号传导水平上的两个噪音来源。其一是由于调节细胞对其环境变化的敏感性所涉及的酶活性的随机性。另一个原因是作为细胞传感器的大型复合物内的蛋白质相互作用。出乎意料的是，在某些条件下，第二个噪声源可能如此强大，以至于它使细胞信号网络的输出在两种状态之间来回翻转：“开”和“关”。这些发现共同揭示了信号网络如何不仅可以放大或减少基因表达噪音，而且本身也可以成为随机事件的来源。关于此类随机事件如何与活细胞中的复杂特征（即趋化性）相互作用的新知识可能有助于未来的抗菌策略，因为许多细菌利用趋化性来帮助它们建立感染。更一般地说，关于蛋白质网络中噪声的新见解可以帮助工程师寻求构建合成生化网络或在活细胞中生产有用的化合物。
We present in vivo single-cell FRET measurements in the Escherichia coli chemotaxis system that reveal pervasive signaling variability, both across cells in isogenic populations and within individual cells over time. We quantify cell-to-cell variability of adaptation, ligand response, as well as steady-state output level, and analyze the role of network design in shaping this diversity from gene expression noise. In the absence of changes in gene expression, we find that single cells demonstrate strong temporal fluctuations. We provide evidence that such signaling noise can arise from at least two sources: (i) stochastic activities of adaptation enzymes, and (ii) receptor-kinase dynamics in the absence of adaptation. We demonstrate that under certain conditions, (ii) can generate giant fluctuations that drive signaling activity of the entire cell into a stochastic two-state switching regime. Our findings underscore the importance of molecular noise, arising not only in gene expression but also in protein networks. Many sophisticated computer programs use random number generators to help solve challenging problems. These problems range from achieving secure communication across the Internet to deciding how best to invest in the stock market. Much research in recent years has found that randomness is also widespread in living cells, where it is often called “noise”. For example, the activity of some genes is so unpredictable to the extent that it appears random. Yet, relatively little is known about how such gene-expression noise propagates up to change how the cell behaves. Many open questions also remain about how cells might exploit these or other fluctuations to achieve complex tasks, like people use random number generators. Bacteria perform a number of complex tasks. Some bacteria will swim toward chemicals that suggest a potential reward, such as food. Yet they swim away from chemicals that could lead them to harm. This ability is called chemotaxis and it relies on a network of interacting enzymes and other proteins that coordinates a bacterium’s movements with the input from its senses. Keegstra et al. set out to find sources of noise that might act as random number generators and help the bacterium E. coli to best perform chemotaxis. An improved version of a technique called in vivo Förster resonance energy transfer (or in vivo FRET for short) was used to give a detectable signal when two proteins involved in the chemotaxis network interacted inside a single bacterium. The experiments showed that this protein network amplifies gene-expression noise for some genes while lessening it for others. In addition, the interactions between proteins encoded by genes acted as an extra source of noise, even when gene-expression noise was eliminated. Keegstra et al. found that the amount of signaling within the chemotaxis network, as measured by in vivo FRET, varied wildly over time. This revealed two sources of noise at the level of protein signaling. One was due to randomness in the activity of the enzymes involved in tuning the cell’s sensitivity to changes in its environment. The other was due to protein interactions within a large complex that acts as the cell’s sensor. Unexpectedly, this second source of noise under some conditions could be so strong that it flipped the output of the cell’s signaling network back and forth between just two states: “on” and “off”. Together these findings uncover how signaling networks can not only amplify or lessen gene-expression noise, but can themselves become a source of random events. The new knowledge of how such random events interact with a complex trait in a living cell – namely chemotaxis – could aid future antimicrobial strategies, because many bacteria use chemotaxis to help them establish infections. More generally, the new insights about noise in protein networks could help engineers seeking to build synthetic biochemical networks or produce useful compounds in living cells.