Discrepancies between the NMR and X-ray structures of uncomplexed barstar: analysis suggests that packing densities of protein structures determined by NMR are unreliable.

The crystal structure of the C82A mutant of barstar, the intracellular inhibitor of the Bacillus amyloliquefaciens ribonuclease barnase, has been solved to a resolution of 2.8 A. The molecule crystallizes in the space group I41 with a dimer in the asymmetric unit. An identical barstar dimer is also found in the crystal structure of the barnase-barstar complex. This structure of uncomplexed barstar is compared to the structure of barstar bound to barnase and also to the structure of barstar solved using NMR. The free structure is similar to the bound state, and there are no significant main-chain differences in the 27-44 region involved in barstar binding to barnase. The C82A structure shows significant differences from the average NMR structure, both overall and in the binding region. In contrast to the crystal structure, the NMR structure shows an unusually high packing value based on the occluded surface algorithm, indicating errors in the packing of the structure. We show that the NMR structures of homologous proteins generally show large differences in packing value, while the crystal structures of such proteins have very similar packing values, suggesting that protein packing density is not well determined by NMR.