In some proteins, a unique class of nonconserved positions is characterized by their ability to generate diverse functional outcomes through single amino acid substitutions. Due to their ability to tune protein function, accurately identifying such “rheostat” positions is crucial for protein design, for understanding the impact of mutations observed in humans, and for predicting the evolution of pathogen drug resistance. However, identifying rheostat positions has been challenging, due—in part—to the absence of a clear structural relationship with binding sites. In this study, experimental data from our previous study of the Escherichia coli lactose repressor protein (LacI) was used to identify rheostat positions for which mutations tune in vivo EC50 for the allosteric ligand “IPTG.” We next used the rheostat assignments to test the hypothesis that rheostat positions have unique dynamic features that will enable their identification. To that end, we integrated all-atom molecular dynamics simulations with perturbation residue response analysis. Results first revealed distinct dynamic behavior in IPTG-bound LacI compared with apo LacI, which was consistent with IPTG’s role as an allosteric inducer. Next, we used a variety of dynamic features to build a classification model that discriminates experimentally characterized rheostat positions in LacI from positions with other types of substitution outcomes. In parallel, we built a second classifier model based on the 3D structural “static” network features of LacI. In comparative studies, the dynamic model better identified rheostat positions that were >8 Å from the binding site. In summary, our study provides insights into the dynamic characteristics of rheostat positions and suggests that models built on dynamic features may be useful for predicting the locations of rheostat positions in a wide range of proteins.

中文翻译：

---

基于动力学的蛋白质网络功能可准确区分中性位和变阻器位置


在一些蛋白质中，一类独特的非保守位置的特点是它们能够通过单个氨基酸取代产生不同的功能结果。由于它们能够调节蛋白质功能，因此准确识别此类“变阻器”位置对于蛋白质设计、了解在人类中观察到的突变的影响以及预测病原体耐药性的演变至关重要。然而，确定变阻器位置一直具有挑战性，部分原因是与结合位点缺乏明确的结构关系。在这项研究中，我们之前对大肠杆菌乳糖抑制蛋白 （LacI） 的研究的实验数据被用于确定突变在体内调节变构配体“IPTG”的 EC50 的变阻器位置。接下来，我们使用变阻器分配来检验变阻器位置具有独特的动态特征的假设，这将使它们能够被识别。为此，我们将全原子分子动力学模拟与扰动残基响应分析相结合。结果首先揭示了与 apo LacI 相比，IPTG 结合的 LacI 具有独特的动态行为，这与 IPTG 作为变构诱导剂的作用一致。接下来，我们使用各种动态特征来构建一个分类模型，该模型将 LacI 中实验表征的变阻器位置与具有其他类型替代结果的位置区分开来。同时，我们基于 LacI 的 3D 结构 “静态” 网络特征构建了第二个分类器模型。在比较研究中，动力学模型更好地识别了距结合位点 >8 Å 的变阻器位置。 总之，我们的研究提供了对变阻器位置的动态特性的见解，并表明基于动态特征构建的模型可能有助于预测变阻器位置在各种蛋白质中的位置。