Fig. 2
Integrated NMR/EM structure determination approach. a In step 1, 40 different assignment possibilities of the five longest helical stretches in the sequence (TALOS) to cylindrical (helical) densities (“α-helix-to-density assignments”) are used in regular NMR-type structure calculations (CYANA). b Ranking of the 20 solutions from these structure calculations by the CYANA target score and the overlap of the α-helices with the cylindrical density features. Each point represents the average of the 10 lowest-energy structures. c Two example cases are shown with incorrect (top) and correct (bottom) assignments, clearly showing that incorrect α-helix-to-density assignment is not compatible with good map overlap. For simplicity, only the lowest-energy structure is shown. The structure with the correct helix-to-density assignment (labeled as Example 2) has a backbone root-mean-square-deviation, RMSD, to the crystal structure of 7.4 Å, and a backbone bundle-RMSD computed from the 20 lowest-energy models of 11.4 Å. d In step 2, the structure of the TET2 monomeric subunit with the correct α-helix-to-density assignment is iteratively refined by flexible fitting into the EM map (truncated at 8 Å resolution), and CYANA calculations with an increasing number of unambiguous NMR restraints and restraints from the fit in the EM map. After convergence, defined by at least three cycles with an RMSD-difference below 10% (shown in (e)), the structure is refined against the EM map of the entire dodecameric ensemble and all NMR restraints, using XPLOR-NIH, using maps with increasing resolution (8, 6, 4.1 Å; step 3). e Root-mean-square-deviation (RMSD) of the structures at different steps of the protocol, relative to the mean structure
