Multipronged Modeling of Subcellular Self-Organization

我们体内的细胞有一定的结构：细胞核和其他细胞器不是随机放置的，而是定位在特定的位置，以优化细胞内的运输和通信。这些细胞器的位置不是预先编程的；细胞器通过机械的自组织过程找到它们的位置。具体来说，细胞器延伸出由细胞骨架蛋白构成的动态“手臂”，并利用这些“手臂”来抓住、推动和拉动彼此。随后的拉锯战导致了复杂的机械平衡，决定了细胞器的位置。这些位置可以用显微图像测量，而潜在的机械力是不可能测量的。这个项目将通过对细胞内细胞器位置的力进行逆向工程来解决一个逆向问题。这个项目也将从多个方向解决这个问题：考虑细胞器是受远程力相互作用的微小刚性粒子，考虑由细胞器组成的抽象粘性流体，最后在计算机模拟中模拟所有基本分子和细胞器的动力学和力学。该项目将使用这些模拟来筛选数十万种可能的力，并使用实验数据来提取细胞中正确的力。数学模型将被开发，测试和完善两个基本的细胞系统：有丝分裂纺锤体（细胞分裂中分离染色体的分子机器）和大肌肉细胞的多核。这些模型将帮助实验学家了解哪些分子负责健康细胞的适当结构，以及老化肌肉和分裂癌细胞的缺陷。在此过程中，跨学科研究人员将得到培训，生物系统建模的新课程将被开发，K-12学生将被引入定量生物学。细胞生物学的基本挑战之一是定义细胞的空间组织原理。细胞器定位本质上是一种机械现象：一组细胞内的力负责放置细胞器。这些力是由细胞骨架的活动产生的，细胞骨架是一种浸入粘性细胞质中的弹性纤维和分子运动蛋白的动态支架，更具体地说，是微管-动力蛋白-动力蛋白力产生系统。主要目标是使用实验数据来逆向工程细胞内力和了解潜在的分子机制。为了实现这一目标，将开发新的方法来生成多个模型并根据实验数据自动筛选模型，并将其应用于两种现象：多中心体细胞中两极有丝分裂纺锤体的自组织和多核胚胎肌细胞中的动态核定位。具体来说，将开发一种计算机代码，该代码将产生广泛的力，并通过检测导致观察到的细胞结构的一小部分力来筛选力。多个模型（细胞器作为由两两距离依赖力驱动的粒子，能量最小化模型，连续积分微分和基于智能体的模型）将被开发，比较和测试，机器学习将用于加速模型筛选。与两个实验实验室合作，将发现发育中的肌肉细胞核定位的分子机制和有丝分裂纺锤体的动态结构。该奖项反映了美国国家科学基金会的法定使命，并通过使用基金会的知识价值和更广泛的影响审查标准进行评估，被认为值得支持。

Cells in our body have a certain structure: nucleus and other organelles are not placed at random, but rather positioned in certain locations to optimize transport and communication inside the cell. These organelles’ positions are not pre-programmed; the organelles find their positions through a process of mechanical self-organization. Specifically, the organelles extend dynamic ‘arms’ made of cytoskeletal proteins, and use these arms to grab, push and pull each other. The ensuing tug-of-war leads to a complex mechanical equilibrium that determines organelles’ positions. These positions can be measured in microscopic images, while underlying mechanical forces are impossible to measure. This project will solve an inverse problem by reverse-engineering the forces from the positions of organelles inside the cell. This project will also address this problem from multiple directions: considering organelles to be tiny rigid particles interacting by long-range forces, considering an abstract viscous fluid made of organelles, and finally mimicking dynamics and mechanics of all essential molecules and organelles in a computer simulation. The project will use these simulations to screen hundreds of thousands of possible forces, and experimental data to distill the correct forces in the cell. The mathematical models will be developed, tested and refined for two fundamental cellular systems: mitotic spindle (molecular machine segregating chromosomes in cell division) and multiple nuclei in large muscle cells. The models will help experimentalists to understand which molecules are responsible for proper architecture of healthy cells and for defects in aging muscles and dividing cancer cells. In the process, interdisciplinary researchers will be trained, novel courses on modeling biosystems will be developed, and K-12 students will be introduced to quantitative biology.One of the fundamental challenges of cell biology is to define principles of spatial organization of the cell. Organelle positioning is essentially a mechanical phenomenon: a set of intracellular forces is responsible for placing the organelles. These forces are generated by activities of cytoskeleton, a dynamic scaffold of elastic fibers and molecular motor proteins immersed into viscous cytosol, more specifically, the microtubule-kinesin-dynein force-generating system. The main goal is to use experimental data to reverse-engineer the intracellular forces and understand underlying molecular mechanisms. To achieve this goal, novel methods to generate multiple models and automatically screen the models against the experimental data will be developed and applied to two phenomena: self-organization of a bipolar mitotic spindle in multi-centrosomal cells, and dynamic nuclear positioning in multinucleated embryonic muscle cells. Specifically, a computer code will be developed that will generate a wide class of forces and screen the forces by detecting a small minority of the forces leading to the observed cell architectures. Multiple models (organelles as particles driven by pair-wise distance-dependent forces, energy-minimization models, continuous integro-differential and agent-based models) will be developed, compared and tested, machine learning will be used to accelerate the model screening. Collaboration with two experimental labs will result in discovery of molecular mechanisms of nuclei positioning in developing muscle cells and of dynamic architecture of mitotic spindle.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.