Statistical Modeling of Receptor/Ligand Binding Kinetics on the T cell Surface

描述(申请人提供)：T细胞受体(TCR)如何区分主要组织相容性复合体分子(PMHC)提出的不同多肽，是保护人类免受致病病原体攻击的获得性免疫的中心问题。然而，人们对其机制知之甚少，至少部分原因是缺乏适当的工具来分析小至单分子相互作用、小至亚秒的初始识别事件，这些事件超出了标准技术的时间和空间分辨率。粘合频率和热波动分析-两种探测TCR/pMHC相互作用最初几秒钟的技术-以及另外两种单键方法(解结力和键寿命分析)将被用于拟议的研究中，以研究TCR/pMHC、CD8/MHC相互作用的动力学及其串扰。由于单分子相互作用固有的随机性，数据分析需要采用统计建模的方法。具体地说，上述三种化验的数据都是具有固定间隔的二元附着力分数或解粘力或键寿命的连续值的时间序列的形式。热涨落分析的数据以交替键寿命和随机间隔等待时间的形式出现。虽然一些信息可以通过使用描述性统计获得，但更复杂的统计建模将使我们能够极大地提高对数据的理解和实用性。一类新的时间序列模型将被用来量化黏附分数、解结力和键寿命之间的相关性，这表现为记忆效应，即T细胞对前一黏附事件的记忆能力，以及改变发生概率、解结力的概率密度和下一黏附事件的寿命的能力。混合分布将被用来确定测量的单键事件是由单态的同质总体还是由多态混合的异质总体组成。在热波动分析中，将使用变点公式进行统计估计。这些统计模型将进行实验测试，并根据需要进行修改，以提取TCR与不同多肽相互作用的基本特征。

DESCRIPTION (provided by applicant): How T cell receptor (TCR) discriminates different peptides presented by the major histocompatibility complex molecule (pMHC) is a central question in adaptive immunity that defends humans against disease-causing pathogens. Yet its mechanism is poorly understood due at least partly to the lack of appropriate tools to analyze the initial recognition events at scales as small as single molecular interactions and as brief as subseconds, which are beyond the temporal and spatial resolutions of standard techniques. Adhesion frequency and thermal fluctuation assays - two techniques probing the first seconds of TCR/pMHC interactions - as well as two other single-bond methods (unbinding force and bond lifetime assays) will be used in the proposed research to study the dynamics of TCR/pMHC, CD8/MHC interactions and their crosstalk. Because of the inherent stochastic nature of single molecular interactions, statistical modeling approach is required for data analysis. Specifically, the data of three of the above assays are in the form of time series of binary adhesion scores or continuous values of unbinding forces or bond lifetimes with fixed intervals. The data of the thermal fluctuation assay are in the form of alternating bond lifetime and waiting time of random intervals. While some information can be obtained by using descriptive statistics, more sophisticated statistical modeling will enable us to greatly increase the understanding and utilities of the data. New class of time series models will be used to quantify the correlations of the adhesion scores, unbinding forces and bond lifetimes, which are manifested as memory effects, i.e., T cell's ability to "remember" the previous adhesion event and to alter the probability of occurrence and the probability densities of unbinding forces and lifetimes of the next adhesion event. Mixture distribution will be used to determine whether the measured single-bonds events consist of a homogeneous population of single states or a heterogeneous population of multi-state mixture. Change-point formulation will be employ for statistical estimation in the thermal fluctuation assay. These statistical models will be tested experimentally and modified as needed to extract the fundamental characteristics of TCR interaction with different peptides.