New ABCG2 homology model reveals importance of C‐terminus in protein trafficking

Abstract

Recent genetic studies have implicated ABCG transporters in human disease and identified potential causal variants. However interpretation of the resulting molecular defect is often limited by the lack of high‐resolution crystal structures. This has been the case for ABCG2. Here we present a new homology model of ABCG2 using a recent ABCG5/G8 crystal structure (PDB:5do7) as a template in an effort to better understanding the implications of human ABCG2 gout causing mutations. Our homology model is composed of the ABCG2 NBD domains (residues 29–288) and TMD domains (residues 289–655) and was built using the Swiss‐Model server. To avoid forcing the ABCG2 into a heterodimer, the ABCG5 served as template for both ABCG2 copies because its NBD structure is closer to the canonical NBD structure. The loop regions of the model were further refined by GalaxyLoop. The final model complied with the modular multi‐domain architecture of ABC proteins, and had RMSD values with ABCG5 NBD, ABCG5 TMD, ABCG8 NBD, ABCG8 TMD of 0.2Å, 0.5Å, 1.7Å, and 2.4Å, respectively. The new ABCG2 homology model allows for the visualization of parts of the protein previously undefined. One intriguing finding was that the loop between the first and second β‐strands of the NBD (residues 42–63) is in close proximity with the TMD C‐terminus S655. Notably, the corresponding loop (residues 47–65) and the C‐terminus R651 were not resolved in the ABCG5/G8 crystal structure, suggesting a dynamic interaction in this region. We explored the significance of the C‐terminal tail (resides 652–655; KKYS) on protein abundance and trafficking. The addition of a C‐terminal Flag‐tag or the deletion of the last 4 residues (651X) led to a profound decrease in protein abundance. The 651X protein is not glycosylated and localizes to the ER only. Interestingly, a YS653AA (KKAA) substitution is less deleterious and results in protein that reaches the Golgi compartment, highlighting the dilysine motif, a potential COPI target, as key in trafficking ABCG2. Finally, we found the S655A substitution (KKYA) results is significantly greater protein abundance than wild type and partially rescues the abundance of the Q141K human gout mutation, implicating the Y654 as a key trafficking residue. In conclusion, our new ABCG2 homology model suggests the C‐terminus is in close proximity to the NBD and may be a critical feature in regulating the trafficking of the ABCG2 protein.Support or Funding InformationAmerican Heart Association 14SDG18060004