High-Resolution Structure of P-glycoprotein

X-ray diffraction data were collected at 100 K at either the Stanford Synchrotron Radiation Laboratory (SSRL; BL11-1) or the Canadian Light Source (CLS; 08ID-1). Fluorescence scans were taken on P-gp–cyclopeptide co-crystals to maximize the anomalous signal contribution from the incorporated selenium (Table 1 ▶). All diffraction data were processed with MOSFLM (Battye et al., 2011 ▶ ) and reduced with SCALA (Evans, 2006 ▶ ) within the CCP4 suite of programs (Winn et al., 2011 ▶ ). In the case of QZ-Ala, the data from three isomorphous crystals were scaled together to maximize the completeness (Table 1 ▶). The 3.4 Å resolution structure of P-gp was initially solved by molecular replacement (MR) with Phaser (McCoy et al., 2007 ▶ ) using a previously determined P-gp structure (PDB entry 4ksc; Ward et al., 2013 ▶ ) as a search model with no modifications. Commensurate with the improved resolution, the new electron-density features guided adjustments of our model when compared with the same more ‘open’ crystal form that we reported in 2013 (Ward et al., 2013 ▶ ) and are summarized in Supplementary Fig. S2. Residues 30–32 were located in the electron density, and resulted in a subsequent shift in the registration of residues in the first helix (residues 30–43) preceding TM1. Amendments were made to the topology of intracellular helix 1 (IH1; residues 154–168), extracellular loop 3 (ECL3; residues 318–338) and a portion of TM6 leading into the first NBD (residues 358–387). Within NBD1, residues 398–404, 424–427, 520–526 and 597–602 were rebuilt. Elbow helix 2 (EH2) was rebuilt from residues 689 to 708. A registry issue was amended from ECL4 (residue 738) to TM8 (residue 760) and another that constitutes segments of TM9, ECL5 and the beginning of TM10 (residues 826–855). The topology of IH3 was adjusted (residues 795–806), as was ECL6 (residues 961–967) and a portion of TM12 (residues 972–984). Further modifications were made in the region leading into and contributing to NBD2 (residues 1010–1028, 1042–1047, 1129–1137 and 1165–1172). Residues 1272 and 1273 were also located in the electron-density maps at the C-terminus. As for all structures of P-gp determined to date, the ‘linker’ region (residues 627–688) was not located in the electron density. Many of the structural adjustments are in general agreement with the recent corrections (Li et al., 2014 ▶ ) made to the model of the more ‘closed’ conformation of P-gp first reported in 2009 (Aller et al., 2009 ▶ ). During the refinement process, the model underwent rigid-body and restrained positional refinement, with H atoms applied in their riding positions, using phenix.refine (Afonine et al., 2012 ▶ ) against a maximum-likelihood target function with grouped B factors, secondary-structure restraints, reference-model restraints and TLS. Rounds of refinement were interspersed with manual inspection and correction against σ_A-weighted electron-density maps in Coot (Emsley et al., 2010 ▶ ) and improvements to model geometry and stereochemistry were monitored using MolProbity (Chen et al., 2010 ▶ ). Subsequent cyclopeptide co-crystal structures were solved by either MR or rigid-body refinement using the refined 3.4 Å resolution model with residues from TM4 (218–243) and EH2 (689–694) removed to avoid biasing their placement within the electron-density maps. These structures were then refined in a similar fashion to the 3.4 Å resolution structure with an additional round of positional refinement with ligand B factors set to the Wilson B value. Ligand description dictionaries were determined using phenix.elbow (Adams et al., 2010 ▶ ) and the crystallographic positions of the incorporated seleniums were validated using anomalous scattering methods. The refined structures were judged to have excellent geometry as determined by MolProbity (Chen et al., 2010 ▶ ). The resulting refinement statistics are listed in Table 1 ▶.