Influence of NMR Data Completeness on Structure Determinations of Homodimeric Proteins

The structure determination of homodimeric proteins by NMR using conventional NOESYexperiments is still challenging due to the degeneracy of the chemical shifts in the identical monomers, which causes ambiguity in the NOE assignments. Residues involved in the interface between two monomers provide essential intermolecular NOEs for the structure determinations of homodimeric proteins. Hence NMR data, such as NOE peak lists and chemical shift assignments of these interface residues, play a crucial role for the successful structure determination of homodimeric proteins. This paper extends our previous report (Lin, Y.-J.; Kirchner, D.K.; Guntert, P. J. Magn. Reson. 2012, 222, 96) and investigates the influence of incomplete NOESY peak lists combined with incomplete H-1 chemical shift assignments of the interface residues on the structure determination of homodimeric proteins using the program CYANA. Data incompleteness was simulated by random omission of both NOESY cross peaks and interface H-1 chemical shifts. Our results for three proteins with different percentages of interface residues reveal that the algorithm can tolerate about 40-50% NOESY peak omission with complete interface chemical shift assignments, which indicates that partial NOESY peak omission does not cause severe problems when the interface chemical shifts are completely assigned. Combining NOESY peak omission with incomplete interface chemical shift assignments, the tolerance for interface chemical shift omission decreases with the extent of omitted NOESY peaks. The tolerance for unassigned interface side chain, methyl and aromatic chemical shifts is affected more strongly by NOESY peak omission than that for the omission of general interface H-1 chemical shifts including the backbone. In general about 10-30% peaks omission is tolerated in conjunction with missing chemical shift assignments. If more NOESY peaks are omitted calculations gradually become unstable and tend not to tolerate any missing interface chemical shifts. A large amount of omitted NOESY peaks, for instance 30% omission in our calculations, could decrease the tolerance for missing aromatic or methyl interface H-1 chemical shifts to as few as 2-4 missing chemical shifts, suggesting that complete aromatic and methyl H-1 chemical shift assignments are important when the NOESY peak data is significantly incomplete. Finally, for homodimeric proteins with a low percentage of interface residues, our results reveal that the omission of NOESYpeaks, even at an extent of only 10%, can result in no tolerance against the omission of interface H-1 chemical shifts, suggesting that the completeness of both interface H-1 chemical shift assignments and NOESY peaks are important for the successful structure determination of proteins with a small homodimer interface.