XTACC3-XMAP215 association reveals an asymmetric interaction promoting microtubule elongation.

chTOG is a conserved microtubule polymerase that catalyses the addition of tubulin dimers to promote microtubule growth. chTOG interacts with TACC3, a member of the transforming acidic coiled-coil (TACC) family. Here we analyse their association using the Xenopus homologues, XTACC3 (TACC3) and XMAP215 (chTOG), dissecting the mechanism by which their interaction promotes microtubule elongation during spindle assembly. Using SAXS, we show that the TACC domain (TD) is an elongated structure that mediates the interaction with the C terminus of XMAP215. Our data suggest that one TD and two XMAP215 molecules associate to form a four-helix coiled-coil complex. A hybrid methods approach was used to define the precise regions of the TACC heptad repeat and the XMAP215 C terminus required for assembly and functioning of the complex. We show that XTACC3 can induce the recruitment of larger amounts of XMAP215 by increasing its local concentration, thereby promoting efficient microtubule elongation during mitosis.

Structure and non-structure of centrosomal proteins.

Here we perform a large-scale study of the structural properties and the expression of proteins that constitute the human Centrosome. Centrosomal proteins tend to be larger than generic human proteins (control set), since their genes contain in average more exons (20.3 versus 14.6). They are rich in predicted disordered regions, which cover 57% of their length, compared to 39% in the general human proteome. They also contain several regions that are dually predicted to be disordered and coiled-coil at the same time: 55 proteins (15%) contain disordered and coiled-coil fragments that cover more than 20% of their length. Helices prevail over strands in regions homologous to known structures (47% predicted helical residues against 17% predicted as strands), and even more in the whole centrosomal proteome (52% against 7%), while for control human proteins 34.5% of the residues are predicted as helical and 12.8% are predicted as strands. This difference is mainly due to residues predicted as disordered and helical (30% in centrosomal and 9.4% in control proteins), which may correspond to alpha-helix forming molecular recognition features (α-MoRFs). We performed expression assays for 120 full-length centrosomal proteins and 72 domain constructs that we have predicted to be globular. These full-length proteins are often insoluble: Only 39 out of 120 expressed proteins (32%) and 19 out of 72 domains (26%) were soluble. We built or retrieved structural models for 277 out of 361 human proteins whose centrosomal localization has been experimentally verified. We could not find any suitable structural template with more than 20% sequence identity for 84 centrosomal proteins (23%), for which around 74% of the residues are predicted to be disordered or coiled-coils. The three-dimensional models that we built are available at http://ub.cbm.uam.es/centrosome/models/index.php.