**3.1 Sequence analysis and homology modeling of neuronal specific tubulin isotypes**

The residue composition of different β-tubulin isotypes mostly varies at the carboxy-terminal tail region as revealed by the multiple sequence alignment (**Figure 2**). The βI and βIII tubulin isotypes have longer C-terminal tail regions when compared with the βIIb tubulin isotype. The β-tubulin sequence in the template structure i.e., 6CVN (chain A) and human βIIb tubulin isotypes show 98.65% sequence identity. These sequence variations in the tubulin isotypes are reported to regulate number of protofilaments in the MT and their stability [73]. These β-tubulin isotypes sequences were used to generate three-dimensional homology models using 6CVN as the template. The structures of βI, βIIb, βIII tubulin isotypes were modeled using Modeler 9v20 [38]. The best homology model generated is selected using DOPE score. The DOPE score value for A and C chain of (i) βI subunits are −54299.89 and − 54291.24 (ii) βIIb subunits are −53487.13 and − 53054.42 and (iii) βIII subunits are −53725.86 and − 53054.42. The quality of these models is accessed using GMQE score [74], Verify3D [40], Errat score [41], Z-score [75] and Ramachandran plot [76, 77]. The parameters describing the overall quality of the modeled neuronal specific β-tubulin subunits are shorn in **Table 2**. The GMQE score provides an estimate of the accuracy of the modeled tertiary structure of neuronal specific β-tubulins. Here, GMQE score for all the modeled β-subunits is 0.98, which represents the accuracy of the generated model. Further, verify3D and ERRAT score also validates the quality of the generated models (**Table 1**). Ramachandran plots for all the modeled β-tubulin isotypes represents more than 98% of the residues occupy a favored region. The occupancy of amino acid residues in the Ramachandran plot is given in **Table 3**. These modeled structures of neuronal specific β-tubulin isotypes were used further to build the tubulin and TauR2 complexes such as βI/α/βI-TauR2, βIIb/α/βIIb-TauR2 and βIII/α/βIII-TauR2 using 6CVN.pdb as a template structure. These modeled complexes were used as starting structures to perform MD simulations.
