This is structure in DHPC micelles. A significant rearrangement of C-terminal surface helix was detected by EPR spin-labeling in membrane: FFK segment probably forms a continuation of TM helix, whereas the C-terminal helices bind to the membrane surface and go in direction opposite to the central channel. This model with classic "rows into grooves" packing of TM helices differs from two crystallographic structures (3bkd and 3c9j; CA atom rmsd of 3.8-3.9 A), and from all solid state NMR models (rmsd >3 A; 2h95 is the closest, with rmsd of 2.3 A). 2kih is S31N mutant.
Solid state NMR models are 1mp6, 1nyj, 2h95 and 2kad.