Structural Analysis of Macromolecular Complexes by High

细胞包含成千上万的多分子复合物，它们就像微型工厂机器一样一起工作。对这些分子器件的结构和功能的详细了解是细胞生物学中非常感兴趣的问题。我们的研究工作集中在单粒子的电子冷冻显微镜分析，这是一种确定复杂细胞组装体三维结构的有力方法。我们已经定义和解释的二十面体丙酮酸脱氢酶多酶复合物的结构，一个典型的例子，一个多步骤的催化机器，耦合的活性的三个组成部分的酶（E1，E2和E3）在丙酮酸的氧化脱羧生成乙酰辅酶A的交界处的糖酵解和三羧酸循环。通过将来自电子冷冻显微镜的28结构与先前确定的复合物的各个组分的原子坐标相结合，获得了由60种E2酶和60种E1酶组成的11 MDa、二十面体PDH复合物的三维模型（Milne等人，EMBO J.21，5587，2002）。对该模型的分析为这种分子机器的设计和功能提供了一些新的见解。一个关键特征是E1分子位于外围，其取向允许60个移动的硫辛酰基结构域中的每一个连接到内部E2酶，以从二十面体复合物内部访问多个E1活性位点。这种意想不到的架构提供了一个高效的机制，活性位点耦合和催化速率的增强，我们建议是通过运动的硫辛酰域之间的复杂的内部和外部核心的限制性环形区域。我们目前正在改进由60个E2酶和60个E3酶组成的第二种PDH复合物，以确定为什么在体内60个E2分子的内部二十面体仅被48个E1分子和6个E3分子占据以形成外部蛋白质壳的结构基础。我们的三维重建这个复杂的表明，类似于上述的E1 E2复合物，一个环形间隙75之间存在的内核和外壳的E3同源二聚体。通过用10、20、40或60个E1四聚体或用10、20、40或60个E3同源二聚体装饰E2核心形成的部分占据复合物的图像分析也表明E2和E1或E2和E3密度的75-95分离。因此，蛋白质外壳的酶之间发生的相互作用并不负责维持复合物的大小。相反，连接E2催化结构域与E2外周亚基结合结构域的E2内部连接体必须是相当刚性的放射状辐条，为E1和E3分子的组织提供支架。E3定位在稍微靠近核心的位置，这表明摆动的硫辛酰结构域可以有效地进入所有三种酶的活性位点，而不离开环形空间。这种结构提供了在乙酰辅酶A的生产和随后的酶再生步骤期间活性位点偶联的高效机制的进一步证据，所述酶再生步骤是启动乙酰辅酶A生产的新循环所需的。我们还积极致力于确定导致出色的显微图像的条件，开发选择和准确对齐三维重建的最佳分子图像的方法，可靠地解释这些结构，并开发自动化程序以促进获得高质量大分子复合物三维模型的过程。为此，我们1）开发了在Tecnai系列电子显微镜上自动收集数据的算法，2）表征了4000 x 4000像素数字CCD相机的特性，并评估了从CCD数字图像构建的三维分子模型的质量，3）开发了“核心加权”方法，与网格线程蒙特卡罗方法相结合，以增强可靠地将单个组分的原子坐标的最佳拟合识别到较大复合物的低分辨率图中的能力，所述较大复合物是使用单粒子电子显微镜确定的典型结构，以及4）分子图像计算分析的优化方法。我们已经证明，迭代细化和严格的分子分选相结合，可以是一种有效的方法，以获得显着改善地图质量的1.8 MDa二十面体E2催化核心。从核心复合物的42,945张图像开始，我们已经证明，只使用数据集中最好的139个粒子产生的图比使用更多图像构建的图更上级，并且结构中的许多α螺旋可以在由少至9个粒子构建的图中明确可视化。因此，严格的计算排序的分子图像可能有助于克服显着的限制所造成的图像质量和蛋白质构象的异质性是固有的大数据集的图像。应用这种方法和改进的拟合算法，以促进准确对接的X-射线坐标的单个组分蛋白质的密度图，通过冷冻电子显微镜应提高可视化的重要大分子复合物中存在的正常和恶性细胞。

Cells contain thousands of multimolecular complexes which work together much like miniature factory machines. A detailed understanding of the structure and function of these molecular devices is a problem of great interest in cell biology. Our research efforts focus on electron cryo-microscopic analysis of single particles, a powerful method to determine the three-dimensional architectures of complex cellular assemblies. We have defined and interpreted the structure of an icosahedral pyruvate dehydrogenase multienzyme complex, a prototypical example of a multi-step catalytic machine which couples the activity of three component enzymes (E1, E2, and E3) in the oxidative decarboxylation of pyruvate to generate acetyl CoA at the junction of glycolysis and the tricarboxylic acid cycle. The three-dimensional model for a 11 MDa, icosahedral PDH complex, composed of 60 E2 enzymes and 60 E1 enzymes, was obtained by combining a 28 structure derived from electron cryo-microscopy with previously determined atomic coordinates of the individual components of the complex (Milne et al. EMBO J. 21, 5587, 2002). Analysis of the model provides a number of novel insights into the design and function of this molecular machine. A key feature is that the E1 molecules are located on the periphery in an orientation that allows each of the 60 mobile lipoyl domains tethered to the inner E2 enzyme to access multiple E1 active sites from inside the icosahedral complex. This unanticipated architecture provides a highly efficient mechanism for active site coupling and catalytic rate enhancement, which we propose is achieved by the motion of the lipoyl domain in the restricted annular region between the inner and outer cores of the complex. We are currently refining a second PDH complex comprised of 60 E2 enzymes and 60 E3 enzymes to determine the structural basis of why in vivo the inner icosahedron of 60 E2 molecules is occupied with only 48 E1 molecules and 6 E3 molecules to form the outer protein shell. Our three-dimensional reconstruction this complex indicates that, similar to the E1E2 complex described above, an annular gap of 75 exists between the inner core and the outer shell of E3 homodimers. Image analysis of partial occupancy complexes, formed by decorating the E2 core with 10, 20, 40 or 60 E1 tetramers or with 10, 20, 40 or 60 E3 homodimers, also indicates a 75-95 separation of the E2 and E1 or E2 and E3 densities. Thus, interactions occurring between the enzymes of the outer protein shell are not responsible for maintenance of the size of the complex. Rather, the E2 inner linkers that connect the E2 catalytic domains to the E2 peripheral-subunit binding domains must be fairly rigid radial spokes that provide a scaffold for an organization of E1 and E3 molecules. E3 localizes slightly closer to the core, suggesting that the swinging lipoyl domain can effectively access the active sites of all three enzymes, without leaving the annular space. This architecture provides further evidence of the highly efficient mechanism for active site coupling during both the production of acetyl CoA and the subsequent enzyme regeneration step that is required to initiate a new cycle of acetyl CoA production. We are also working actively to identify conditions that lead to outstanding microscopic images, to develop methods to select and accurately align the best molecular images for three-dimensional reconstructions, to reliably interpret these structures, and to develop automated procedures to facilitate the process of obtaining high quality three dimensional models of macromolecular complexes. To this end, we have 1) developed algorithms to collect data automatically on the Tecnai series of electron microscopes, 2) characterized the properties of a 4000 x 4000 pixel digital CCD camera, and assessed the quality of the three-dimensional molecular models constructed from CCD digital images, 3) developed a "core-weighting" method, combined with a grid-threading Monte Carlo approach to enhance the ability to reliably identify the best fit of atomic coordinates of individual components into low resolution maps of larger complexes that are typical of structures determined with the use of single particle electron microscopy and 4) optimized methods for the computational analysis of molecular images . We have demonstrated that the combination of iterative refinement and stringent molecular sorting can be an effective method to obtain substantial improvements in map quality of the 1.8 MDa icosahedral E2 catalytic core. From a starting set of 42,945 images of the core complex, we have shown that using only the best 139 particles in the data set produces a map that is superior to those constructed with greater numbers of images, and that many of the alpha-helices in the structure can be unambiguously visualized in a map constructed from as few as 9 particles. Thus, stringent computational sorting of molecular images may help overcome the significant limitations posed by heterogeneity in image quality and in protein conformation that are inherent to large data sets of images. Application of this method and improved fitting algorithms to facilitate accurate docking of X-ray coordinates of individual component proteins into the density maps obtained by cryo electron microscopy should enhance visualization of important macromolecular complexes present in normal and malignant cells.