Fig. 2

Modeling of the mutant GAT-1 protein with machine learning tools. a-b. Tertiary structures of both the wildtype (a) and P361T mutant (b) GAT-1 protein are predicted by I-TASSER and DynaMut. The proline at residue 361 is mutated to threonine, both highlighted in light green, alongside with the surrounding residues. The interatomic interactions were predicted by DynaMut, where halogen bonds are depicted in blue and hydrogen bonds are colored in red. The P361T mutation results in the loss of two hydrogen bonds, those between residues 361 and 365 (yellow arrow with red border) and between 361 and 364 (yellow arrow with blue border). This supports the result in Table 1 that this mutation destabilized the global conformation of the GAT-1 protein. c. Machine learning tools predicted ΔΔG (Kcal/mol) of the mutant GAT-1 protein. Bars in the positive direction are predicted as stabilizing while bars in the negative direction are predicted as destabilizing