Periodic Reporting for period 2 - BIOMECANET (Integration of the Biochemical and Mechanical Networks of Cell Division)
Reporting period: 2023-01-01 to 2024-06-30
Cells are the functional units of life, and their division ensures that life is propagated across generations. In its outline, cell division is elegant and simple. Chromosomes, which harbor the genetic material, are first replicated in the mother cell. Then, in the following phase, named mitosis (or meiosis, a specialized cell division occurring in the germ line), chromosomes are distributed to the two daughter cells. The orchestrated ballet of chromosomes, with its remarkable precision, is the action of a molecular structure, the mitotic spindle. The simplicity of the task contrasts with the relative complexity of the structure, which is self-assembled from hundreds of different macromolecules. How can we make inroads into the dissection of this mechanism? In our program, called BIOMECANET (which stands for Integration of the Biochemical and Mechanical Networks of Cell Division), we have the ambition to use the building blocks of cell division to reconstitute in vitro the crucial molecular events that underlie this process. At the same time, we will simulate spindle assembly in a computer to make sense of its dynamics and understand how function emerges from structure. For instance, we will reconstitute the mitotic spindle, the kinetochores (the structure that links chromosomes to the mitotic spindle), and the regulatory machinery that coordinates their interactions. We will achieve this goal by combining a minimal – yet very complex – set of components sufficient to mimic the most fundamental dynamics of this process. We will then analyse the behavior of these molecular systems using in silico simulations of unprecedented complexity and realism. Besides investigating a fundamental biological process, our work will also shed light on the circumstances that cause the cell division process to fail, resulting in chromosome imbalances that are associated with many human diseases, most notably cancer.
The Surrey and Musacchio laboratories handle and maintain large collections of proteins involved in spindle assembly and chromosome segregation and that are instrumental for the program. A number of old and new proteins and protein complexes, including NEDD1, TACC3, CDK5RAP2, Aurora A, CENP-E, fully active PLK1, CDK1/Cyclin A, the -tubulin ring complex, augmin, dynein, dynactin were produced for reconstitution studies. Collaborative experiments were started to investigate by in vitro reconstitutions how kinases like CDK1 (with Cyclin A or Cyclin B) and PLK1 control minimal anaphase midzone formation by phosphorylating PRC1 and Dynein-Dynactin motility of the kinetochore corona. The Surrey and Nédélec labs also investigated the interplay of two antagonistic mitotic motors for active microtubule network organization and found that due to its molecular design HSET can both assist or compete with KIF11 (Henkin et al., PNAS, 2022).The Nèdèlec group wrote Evocym, a suite of Python programs implementing Genetic Algorithm and Bayesian Optimization, usable as a front-end to Cytosim, and extensively tested over the past year. They also extended Cytosim to simulate chromosomes as a pair of flexible chromatids with their kinetochores. Such kinetochores can capture, nucleate and stabilize microtubules, and can be used as an input to the spindle assembly checkpoint.
Our understanding of how cell cycle regulation controls active microtubule network organization driven by the combinatorial action of motors, microtubule associated proteins, and chromosomes remains limited. Our current findings are the first steps overcoming this limitation. For the future, we expect to gain a more detailed understanding of how various phosphorylation and dephosphorylation events control spindle assembly, kinetochore function, and reorganization as the cell cycle proceeds. We have now established the key components of future models addressing chromosome segregation, showing that a system with 10 chromosomes and ~1000 microtubules can be simulated sufficiently fast (40s of CPU time for 10s of real time, single core), which demonstrates feasability of our approach.