导　　师： 曾耀英;韩博平

学科专业： G1004

授予学位： 博士

作　　者： ;

机构地区： 暨南大学

摘　　要： 本文对tcr，pmhc和辅助分子cd2，cd58等与免疫识别有关的分子进行了结构研究，试图从结构上理解t细胞特异性识别的形成机制；另外本文还提出了涡旋驱动模型，通过解释t细胞受体在免疫突触形成过程中的重定向(reorientation)运动，来试图理解t细胞特异性识别的动力学机制。最后，在t细胞表位预测方面，本文利用神经网络集成预测了mhc-Ⅰ类分子结合肽。研究结果如下： 1.tcr识别不同性质的pmhc复合物时，其结构变化只是表现在可变区的cdr上，这种变化是局部的，并不能影响其近膜端的结构变化，这表明通过tcr传导的信息似乎不由tcr的结构变化决定；而且，受体/配体对的耦合作用是由各种弱相互作用介导的，这些弱相互作用在分子作用界面内的分布是各向异性的，这将导致受体/配体对的相互作用具有确定的取向偏好。 2.本文提出的涡旋驱动模型证明突触内受体/配体对耦合时通过将其结合自由能转化为细胞内外流体涡旋运动的机械能可能直接提供了tcr重定向运动的驱动力。模型计算的结果表明，在强度及作用频率同时具备一定范围的涡旋连续驱动下，tcr重定向运动速度可达到实验测定的范围(0.04～0.1μm/s)。同时，本模型还指出，涡旋的强度及作用频率依赖于突触内各种受体/配体对相互作用的动力学及其数量水平的协同作用。考虑到胸腺对t细胞的正、负选择作用，经过选择的t细胞表面tcr重定向机制可能皆处于一种临界状态，即只有具有特定范围的动力学参数及数量水平的tcr/pmhc相互作用才能促使成熟的免疫突触的形成，进而标志着t细胞活化的开始。因此，本模型可能有助于我们理解t细胞在抗原识别过程中表现出的高特异性和高灵敏性的共存机制，理解的焦点将直接集中在tcr/pmhc相互作用的动力学特征及其数量水平而不是集中在后续的信号级联过程。 3.对于hla-a*0201编码的mhc-Ⅰ类分子结合肽数据库(含有628个9聚物)及其结合能力分类(无、低、中和高4类)，集成数为12的神经网络集成的平均分类预测命中率可达0.8；而且该神经网络集成对潜在t细胞表位的预测能力也较高，利用t细胞真实表位集(含50个表位)所进行的评估表明，约84％的真实表位归于高和中等亲合性的潜在抗原肽一类。因此，我们可以利用神经网络集成预测mhc-Ⅰ类分子结合肽，并进而预测相应的t细胞表位。经适当修改，基于神经网络集成的预测工具可扩展为能涵盖任意长度的Ⅰ类分子结合肽甚至可扩展到Ⅱ类分子结合肽的预测。 以上这些结果表明，要解决t细胞识别抗原过程中同时表现出的高灵敏性和特异性这一矛盾，可能需要整合结构(空间)决定论和时间决定论的观点，并扩大结构和时间的外延，即所说的“结构”应指细胞层面上的超分子结构群，即免疫突触，而非指单个分子或分子复合物的结构；而所谓的“时间”则应是包含结构内容的动力学演化过程，即免疫突触内各种受体/配体对动力学过程的协同作用。 在t细胞表位的预测方法上，神经网络集成是值得深入研究的备选方案。神经网络集成考虑了抗原肽每一个位置氨基酸残基对亲合性的贡献，并能够通过训练神经网络组件和整合它们的预测能力来显著改善学习系统的泛化能力，同时神经网络集成具有高通量的数据处理特点。 T cell antigen recognition is the initial behavior of cellular immune system against dangerous signals. There are two specific sources in T cell antigen recognition: one is MHC molecule and the other one is the presented antigenic peptide by MHC, while the antigenic peptide is the most important deterministic factor. T cells are documented to be extremely sensitive to antigens and simultaneously with high specificity, while the underlying mechanism has not been worked out. To date, there are two ways to the understanding of this mechanism, namely, the structure /(space/)-dependent mechanism and the time-dependent mechanism. The structure /(space/)-dependent mechanism argues that the signaling of T cell specific recognition is achieved by the conformational changes of the antigen recognition module /(mainly the complex combining pMHC, TCR//CD3, and coreceptor CD4 or CD8 molecules/); while the time-dependent mechanism insists that T cell antigen recognition is dominated by the dynamics which depends on the interaction kinetics of various receptor//ligand pairs /(esp., the antigenic pMHC//TCR pair/) between T cell and APC.One important objective of studying T cell specific recognition is to provide the theoretical basis for the design of the T cell epitope-based vaccines. As T cell epitope is the specific condition for the activation of the T cell, then, how to find the epitope presented by specific MHC molecule from the pathogens is critical to the design of the T cell epitope-based vaccines. Because T cell can only cognize the peptides presented by the MHC molecule, the predictors for T cell epitopes are usually based on the prediction of the peptides binding to a specific MHC molecule.This paper studied the structural mechanisms underlying the TCR recognition of pMHC as well as the coupling of CD2 with CD58, and tried to provide an approach to the understanding of the mechanism that underlies the T cell specific recognition based on the structure. In addition, we proposed a model based on the vortex-driven mechanism to interpret the reorientation of T cell receptor during the formation of immunological synapse, and thereby tried to understand the T cell specific recognition based on the kinetics. Finally, on the prediction of T cell epitopes, we employed the neural network ensemble to predict the peptides binding to MHC class I molecule. The results are as follows;1. When TCR binding to pMHC, the slight structural adjustment only occurs on the CDRs and as this change is just local, it will not affect the close-to-membrane structure of the TCR, showing that TCR signaling is not determined by the structure change of the TCR; In addition, receptor//ligand coupling is mediated by the various weak interactions, which distribute anisotropically in the extensive interface of two reactants. This will lead to the preferential docking orientation during the interaction of the receptor//ligand.2. We demonstrated that the couplings of synaptic receptor//ligand pairs per se have driven the directed motion of T cell receptors. During the coupling, the membrane-tethered receptor//ligand pairs are assumed to transform the binding free energies into the rotational energies of the reactants, thereby leading to the vortexes of the surrounding water continuum inside and outside the T cell. These resulting vortexes can recruit TCRs /(as well as other synaptic molecules/) into the synapse and ma> also orient the active cytoskeletal transport of receptors toward the junction. The model results indicated that efficient TCR reorientation requires sufficiently high values of both strengths and acting frequencies of the consecutive driving vortexes, which are finally determined by the coordination of kinetics and quantitative levels of various receptor//ligand interactions within the synapse. With the thymic selections on the T cells, TCR reorientation mechanisms in the selected T cells are all at a critical point where only TCR//pMHC interactions with specific range of kinetic parameters and quantitative levels can trigger the formation of the mature synapse - a precursor for T cell activation. This may direct the understanding of the coexistence of high sensitivity with high specificity in the T cell antigen recognition just based on the kinetics and quantitative levels of TCR//pMHC interactions themselves rather than on the subsequent signaling cascades.3. On a database of 628 nonamers and their classified binding capacities to MHC class I molecule encoded by gene HLA-A*0201, the generalized neural network ensemble /(NNE/) with 12 component neural networks achieved an average predictive hit rate of 0.8 for the classifications of the peptides respectively with non, low, moderate and high binding capacities. In addition, NNE was also efficient in the prediction of the potential T-cell epitopes. The predictive power of NNE was further evaluated by running generalized NNE on a set of actual T-cell epitopes /(including 50 actual epitopes/) and showed that about 84/% of the actual T-cell epitopes are among the potentially antigenic peptides with high and moderate affinities. Therefore, the neural network ensemble can be applied in the prediction of MHC class I binding peptides and moreover, after proper modifications, they can be conveniently extended to cover peptides with any length and thus suitable for the prediction of peptides binding to other MHC class I or even class II molecules.The above results suggested thai in order to crack the paradox of the coexistence of high specificity with high sensitivity during T cell antigen recognition, it is required that the structure /(space/(-dependent mechanism should be integrated with the time-dependent mechanism and that the denotations of structure and time should be extended. In other words, the so-called 'structure' should not refer to the structure of single molecule or molecular complex, but to the super-molecular cluster at the cellular level, i.e. the immunological synapse; while the concept 'time' should be defined by the evolving dynamics that involves the structural content, namely, the coordination of the kinetics and quantitative levels of various receptor//ligand interactions within the synapse.On the predictors for the T cell epitopes, neural network ensemble is a valuable choice worthy of advanced studies. NNE considers the contribution of each amino acid residue at the presented peptide to the affinity for the MHC molecule, and remarkably improves the generalization capacity of the learning system through training the component neural networks and integrating their predictive abilities. In addition, neural network ensembles are characterized by their high-throughput data processing capacities.

关 键 词： 生物细胞学 结合肽 免疫识别 免疫突触

分 类 号： [Q172]

领　　域： [生物学] [生物学]