D-helix influences dimerization of the ATP-binding cassette (ABC) transporter associated with antigen processing 1 (TAP1) nucleotide-binding domain

Abstract

ATP-binding cassette (ABC) transporters form a large family of transmembrane importers and exporters. Using two nucleotide-binding domains (NBDs), which form a canonical ATP-sandwich dimer at some point within the transport cycle, the transporters harness the energy from ATP binding and hydrolysis to drive substrate transport. However the structural elements that enable and tune the dimerization propensity of the NBDs have not been fully elucidated. Here we compared the biochemical properties of the NBDs of human and rat TAP1, a subunit of the heterodimeric transporter associated with antigen processing (TAP). The isolated human TAP1 NBD was monomeric in solution, in contrast to the previously observed ATP-mediated homodimerization of the isolated rat TAP1 NBD. Using a series of human-rat chimeric constructs, we identified the D-helix, an α-helix N-terminal to the conserved D-loop motif, as an important determinant of NBD dimerization. The ATPase activity of our panel of TAP1 NBD constructs largely correlated with dimerization ability, indicating that the observed dimerization uses the canonical ATP-sandwich interface. The N-terminus of the D-helix from one protomer interacts with the ATP-binding Walker A motif of the second protomer at the ATP-sandwich interface. However, our mutational analysis indicated that residues farther from the interface, within the second and third turn of the D-helix, also influence dimerization. Overall, our data suggest that although the D-helix sequence is not conserved in ABC transporters, its precise positioning within the NBD structure has a critical role in NBD dimerization.

Fig 1. Human TAP1 NBD does not form homodimers.

(A-D) SEC elution profiles for rat (A and B) and human (C and D) WT and Walker B D-to-N NBD variants. Panels A and C show results for proteins preloaded with ATP, while panels B and D show results for proteins preloaded with ADP. The nucleotide present in the SEC elution buffer is indicated in the legends at the top right of each panel. Traces are representatives of at least two replicates, and for at least two independent protein preparations for each NBD. (E-F) Sedimentation coefficient distribution from SV experiments (performed once) on the human and rat Walker B D-to-N NBD variants in the presence of ATP (E) and ADP (F). The data for rat TAP1 NBD are previously published [34] and included here for comparison.

Fig 2. Human-to-rat substitutions in the D-helix enable human TAP1 NBD homodimerization.

The seven sets of substitutions tested are shown in different colors, which are consistent across all panels. (A) At the top is the homodimeric rat TAP1 D645N NBD structure (PDB ID 2IXE; [32]) shown as a cartoon representation viewed from the membrane plane, with one protomer in light grey and the other in dark grey. On the left protomer, the 56 residues differing between human and rat TAP1 are indicated by Cα atom spheres. At the bottom is a surface representation of one protomer, showing the dimerizing surface, with the membrane plane at the top. The cartoon in the middle illustrates how the two views are related. The tested substitutions (colored) were focused on the dimerization interface. (B) Scaled linear representation of the human TAP1 NBD with the tested substitutions indicated above. The inset shows a sequence logo generated by Skylign (www.skylign.org) from 64 eukaryotic TAP1 sequences for the Walker B and D-helix region, highlighting the "D-helix" substitution set. (C) SEC elution profiles of protein samples preloaded with ATP and run in ATP-containing buffer show that only the D-helix variant elutes as a homodimer. Traces are representatives of two replicates from one protein preparation for each NBD.

Fig 3. D-loop D674A mutation impairs engineered human TAP1 NBD dimerization and ATPase activity.

(A) ATPase activity parallels the SEC dimerization data, showing higher ATPase activity for the dimerization-competent human TAP1 D-helix NBD variant compared to WT human TAP1 NBD, and significantly decreased ATPase activity when the D-loop D674A mutation is introduced in the human D-helix variant. ATPase rates are mean ± standard deviation from four separate experiments, each done in triplicate. Significance was evaluated using two-tailed Student’s t-tests, with p < 0.05 indicated by *. (B) The relevant part of the interface from the rat TAP1 NBD structure (PDB ID 2IXE; [32]), showing the D-loop and D-helix from one protomer on the left, and the Walker A and B motifs from the other protomer on the right, forming an interaction network around the ATP γ-phosphate. Note that the corresponding variant D-helix residues in rat/human are: G653/N676, N654/S677, R657/Q680, Q659/E682, and R660/Q683. The grey background delineates the protomer on the right. (C-D) SEC traces of protein samples preloaded with ATP and run in ATP-containing buffer showing that a D-to-A mutation in the D-loop impaired dimerization of the rat TAP1 NBD and its D645N variant (C) and the human TAP1 D-helix NBD variant (D), as evidenced by a right-shift of their elution volumes. (E) Dimerization of the human TAP1 D-helix NBD is ATP dependent as its elution was right-shifted when run in ADP-containing buffer. All SEC traces are representatives of at least two replicates from one protein preparation for each NBD. The SEC traces for the rat D645N and human D-helix variants are reproduced from Figs 1A and 2C, respectively, for comparison.

Fig 4. The full D-helix human-to-rat chimera is required for human TAP1 NBD dimerization.

(A) The human TAP1 S677N (SN) and N676G/S677N (NG/SN) NBD variants showed an SEC elution profile most similar to WT human TAP1 NBD and therefore consistent with a monomeric state. Corresponding rat TAP1 NBD elution volume is shown with a grey line for comparison. Traces are representatives of three replicates from two protein preparation for each NBD. (B) In agreement with their lack of dimerization, the S677N (SN) and N676G/S677N (NG/SN) NBD variants showed little ATPase activity, shown as mean ± standard deviation from four separate experiments, each done in triplicate. Significance was evaluated using two-tailed Student’s t-tests, with p < 0.05 indicated by *.

Fig 5. Introducing the human D-helix sequence into the rat TAP1 NBD impairs homodimerization.

(A) SEC traces show that introducing the single N654S mutation or the five-mutation rat-to-human D-helix swap results in a right shift in the elution volume, indicating impaired dimerization, whereas the G653N/N654S double mutant elutes as an apparent mixture of monomers and dimers. Traces are representatives of two replicates from one protein preparation for each NBD. (B) ATPase activity measurements of the rat-to-human D-helix variants show that ATPase activity, shown as mean ± standard deviation from four separate experiments, each done in triplicate, is sensitive to the D-helix sequence. Significance was evaluated using two-tailed Student’s t-tests, with p < 0.05 indicated by *.