Computational protocol: Conformational changes in tubulin in GMPCPP and GDP-taxol microtubules observed by cryoelectron microscopy

Similar protocols

Protocol publication

[…] A high-pass–filtered cryo-EM image of a microtubule was cropped into 60 × 60–nm square pieces. The cropping frames were initially centered along the microtubule at 150-nm intervals, closely corresponding to the periodicity of the superhelix of the protofilament. These frames were categorized as class 1. Each of the cropping frames in class 1 was shifted vertically by 8.28 nm, the length of a tubulin dimer, to produce the class 2 set. This process was repeated to create 18 classes (150/8.28 = 18). The positions of the frames were then iteratively adjusted to improve the contrast score of the class-averaged image (). After the simulated annealing, the frame intervals between the neighboring classes were measured to exclude the segments with the lattice failures or distortions. The contrast transfer function (CTF) correction of class averages were then performed with Imagic V (Image Science). The CTF parameters for each micrograph were based on defocus values determined by CTFFIND3 (). After the rotational and translational parameters of each class average was iteratively refined by the modified simulated annealing method (), the 3D structure was reconstructed from this set of CTF-weighted average of microtubule images with different defocus values using the simultaneous iterative reconstruction technique (). The angular interval of each class average was optimized to be 23.82 degrees, which is derived from the supertwist of the microtubule with 15 protofilaments and 2 starts. These iterative processes to refine the in-plane and helical parameters contributed to increasing the resolution of our maps. After the averaging of the whole tubulin dimers within the resulting reconstruction of microtubules, a 3D fast bilateral filter created with MATLAB (MathWorks, Inc.) was applied for noise reduction without blurring (a MATLAB function of bilateral3 written by Igor Solovey). A B-factor of 400 Å2 using the program EM-BFACTOR was then applied to compensate for amplitude attenuation at higher resolutions (). [...] We used the atomic model of the tubulin dimer obtained by electron crystallography (Protein Data Bank accession no. 1JFF; ) as the initial model. The Fit in Map tool in UCSF Chimera () was used for the rigid body fitting of the atomic models into the density maps with the guidance of fitting score (average map value [AMV]). For the first round of the fitting, the atomic model of each monomer was treated as a rigid body to fit into the map. Each tubulin monomer was fitted well to our GDP-taxol microtubule (AMV = 175.4 and 177.4 for α- and β-tubulins, respectively) but did not fit well to the GMPCPP microtubule (AMV = 158.3 and 162.6, respectively), especially in the regions that showed marked structural differences. Therefore, we divided the tubulin monomer atomic model into three subdomains, as previously reported: N domain (aa 1–203), I domain (aa 204–382), and C domain (aa 383–C terminal). The I domain and C domain fitted well to the map by rigid body fitting, but the fitting of the N domain was still poor, providing only modest improvements (AMV = 162.4 and 168.4, respectively). Hence, we further divided the N domain into two domains: N-terminal half (N1 domain, aa 1–94) and C-terminal half (N2 domain, aa 95–203). After division into these four subdomains, the atomic models of both α- and β-tubulins fitted well to the cryo-EM density map (AMV = 168.5 and 170.8, respectively; , S2, and S3 and Video 2). […]

Pipeline specifications

Software tools IMAGIC, CTFFIND, EM-Bfactor, UCSF Chimera
Application cryo-EM
Chemicals Guanosine Diphosphate, Guanosine Triphosphate, Paclitaxel