Modeling the Microscopic to Macroscopic Dynamics of Actively Streaming Microtubule Suspensions

Abstract

A number of studies have reported reversible ab-tubulin heterodimer dissociation with K d =10 À8 -10 À7 M, with presumably rapid but undetermined kinetics. However, in a recent study an extremely tight association between subunits (K d <10 À11 M), as well as a slow dissociation rate of the monomers was reported (with a half time for dissociation > 3 hours). In this work we have revisited the thermodynamics and kinetics of rat brain ab-tubulin (RBT) dissociation by taking advantage of sedimentation velocity analytical ultracentrifugation methodologies (SV-AUC) developed in recent years. In particular, a combination of optical detection systems was used to cover a wide range of protein concentrations: pseudo-absorbance optics (pABS-AUC) for micro-and submicromolar concentrations, and the newly developed fluorescence optics (FDS-AUC) for nano-and subnanomolar concentrations. SV-AUC data was modeled using SEDFIT software resulting in a sedimentation coefficient continuous distribution, c(s), from which the weight average sedimentation coefficients (s w ) were obtained. The isotherms obtained from calculated s w as a function of tubulin concentration were analyzed in SEDPHAT software using the heteroassociation model AþB1 AB to compute the limiting sedimentation coefficients of the dimer and monomer species as well as the equilibrium dissociation constant. In our experimental conditions, RBT heterodimer sediments with s w = 5.1 s, while the monomer species co-sediments with s w = 2.9 s. According to the heteroassociation model, the RBT heterodimer dissociates with K d = 5.5 x 10 À8 M. From the analysis of sedimentation profiles we find the characteristic dissociation time constant to be on the order of hours. In conclusion, the ab-tubulin heterodimer displays reversible dissociation, moderate K d and moderate dissociation kinetics. Despite the importance of microtubule dynamic instability for their function in cell division and for the effect of anticancer drugs like taxol, understanding this macroscopic microtubule behavior at the structural level has been so far hindered by the limitations in cryoEM resolution available for this system. To gain mechanistic molecular understanding of the dynamic nature of microtubules, we have used real space refinement helical reconstruction to obtain cryoEM maps of microtubules in various ligand-bound states: GMPCPP, (a non-hydrolyzable GTP analog), GDP (dynamic microtubules), and GDPþtaxol (drug-stabilized). These maps, ranging from 4.5 to 5 Angstrom resolution, represent the most detailed description of alpha and beta tubulin in the microtubule lattice to date. In order analyze the differences between the maps, we performed molecular refinement using the software Rosetta guided by the EM density, as well as information from available crystal structures. With this new approach, we were successful in obtaining well-converged ensembles of models that fit their respective maps significantly better than they fit the maps of other ligand-bound states. We find that the major differences...