Co-evolution Analysis for Allosteric Network

变构调节是蛋白质界的一种常见机制，但对残留水平上的潜在机制知之甚少。变构调节剂通过在远离活性催化位点的位点结合来调节蛋白质或蛋白质复合物的功能。这种变构信号如何从遥远的变构位点传递到蛋白质的活性位点是一个悬而未决的研究问题。该项目将变构分析与蛋白质残基的共同进化分析联系起来。当两个（或更多）残基之间的相互作用对蛋白质的稳定性或功能至关重要时，就会发生共同进化。如果其中一个相互作用的残基发生突变，另一个也可能发生突变。这些残基-残基耦合可以从多个序列比对中得到，其中提取逐列相关性并由此推断相互作用（接触）。以前的工作表明，接触预测网络中有统计耦合分析（SCA）衍生的变构信息。本文主要研究了一个系统（MutS），并使用了几个任意参数。因此，本研究建立了一个改进的变构网络预测管道。首先，采用基于直接耦合分析（DCA）而非统计耦合分析的CCMpred或metaPSICOV等具有更高接触预测精度的最新协同进化分析方法。这两种基本方法之间的主要区别是直接耦合分析区分真正的进化相关性和传递相关性的能力，这是相关性分析中的一个常见问题。其次，仅使用序列信息来消除先前方法对晶体结构的依赖，这将允许在更大的范围内进行变构预测；第三，通过计算网络中心性度量来自动化变构网络预测和选择用于信号传输的重要残差。可用于分析变构的工具不多，特别是没有高通量的工具，因此开发和改进这些基本工具可能是理解变构信号传输过程的第一步。此外，残留特异性水平的预测可能使研究变构药物以非竞争方式起作用的变构功能的控制成为可能。例如，这可能对联合治疗有益，在联合治疗中，病理相关蛋白不仅被一种药物靶向，而且被两种药物作用于不同的部位，这可能提高治疗效果并避免耐药性。该项目属于EPSRC生物信息学研究领域，由F. Hoffmann-La Roche和UCB这两家医疗保健部门的公司提供支持，这两家公司可以在研究问题上提供宝贵的专业知识，也可以将发现转化为改进药物发现的努力。

Allosteric regulation is a common mechanism in the protein world but little is known about the underlying mechanisms on a residue level. Allosteric regulators modulate a protein's or a protein complex's functionality by binding at a site which is distant to the active catalytic site. How this allosteric signal is transmitted from a distant allosteric site to the proteins' active site is an open research question.This project links the analysis of allostery with the analysis of co-evolution of protein residues. Co-evolution occurs when the interaction between two (or potentially more) residues is crucial for a protein's stability or functionality. If one of the interacting residues mutates, the other is likely to mutate as well. These residue-residue couplings can be derived from multiple sequence alignments where column-wise correlations are extracted and interactions (contacts) are thus inferred.Previous work has shown that there is information about allostery in contact prediction networks derived from statistical coupling analysis (SCA). The main paper on this studied only one system (MutS) and used several arbitrary parameters. Therefore, this research was set up to build an improved allosteric network prediction pipeline.First, by using more recent co-evolution analysis methods that have higher precision in contact predictions such as CCMpred or metaPSICOV, which are based on direct coupling analysis (DCA) instead of statistical coupling analysis. The main difference between these two underlying methodologies is the direct coupling analyses' capability to discriminate between true evolutionary correlations and transitive correlations, a common problem in correlation analysis. Second, using solely sequence information to remove the crystal structure dependence of the previous approach which would allow allostery predictions on a much larger scale and third, automation of the allosteric network prediction and selection of important residues for signal transmission by computing network centrality measures.There are not many tools available to analyse allostery, especially no high-throughput ones, so developing and improving such basic tools could be a first step of in the process of understanding allosteric signal transmission. Furthermore, the prediction on a residue-specific level might enable the control of allosteric functions for the investigation on allosteric drugs acting in a non-competitive way. This, for example, could be beneficial for combination therapies where pathologically relevant proteins are targeted not only by one drug but two drugs acting at different sites which could be improving therapeutical effectiveness and avoiding drug resistance.This project falls within the EPSRC Biological Informatics research area and is supported by F. Hoffmann-La Roche and UCB, two companies of the healthcare sector that can provide valuable expertise in research questions but also in translating findings into improvements of drug discovery efforts.