No abstract

The molecular structures of two human transthyretin (hTTR, prealbumin) complexes, co-crystallized with thyroxine (3,5,3',5'-tetraiodo-L-thyronine; T(4)), and with 3',5'-dinitro-N-acetyl-LL-thyronine (DNNAT), were determined by X-ray diffraction methods. Crystals of both structures are orthorhombic, space group P2(1)2(1)2, and have two independent monomers in the asymmetric unit of the crystal lattice. These structures have been refined to 17.0% for 8-2.0 A resolution data for the T(4) complex (I), and to R = 18.4% for 8-2.2 A resolution data for the DNNAT structure (II). This report provides a detailed description of T(4) binding to wild-type hTTR at 2.0 A resolution, as well as DNNAT. In both structures, the two independent hormone-binding sites of the TTR tetramer are occupied by ligand. A 50% statistical disorder model was applied to account for the crystallographic twofold symmetry along the binding channel and the lack of such symmetry for the ligands. Results for the co-crystallized T(4) complex show that T(4) binds deep in the hormone-binding channel and displaces the bound water previously reported for T(4) soaked into a native transthyretin crystal [Blake & Oatley (1977). Nature (London), 268, 115-120]. DNNAT also binds deeper in the channel toward the tetramer center than T(4) with the nitro groups occupying the symmetrical innermost halogen pockets. The N-acetyl moiety does not form polar contacts with the protein side chains as it is oriented toward the center of the channel. The weak binding affinity of DNNAT results from the loss of hydrophobic interactions with the halogen binding pockets as observed in T(4) binding. These data suggest that the halogen-binding sites toward the tetramer center are of primary importance as they are occupied by analogues with weak affinity to TTR, and are therefore selected over the other halogen sites which contribute more strongly to the overall binding affinity.

Crystallization has proven to be the most significant bottleneck to high-throughput protein structure determination using diffraction methods. We have used the large-scale, systematically generated experimental results of the Northeast Structural Genomics Consortium to characterize the biophysical properties that control protein crystallization. Datamining of crystallization results combined with explicit folding studies lead to the conclusion that crystallization propensity is controlled primarily by the prevalence of well-ordered surface epitopes capable of mediating interprotein interactions and is not strongly influenced by overall thermodynamic stability. These analyses identify specific sequence features correlating with crystallization propensity that can be used to estimate the crystallization probability of a given construct. Analyses of entire predicted proteomes demonstrate substantial differences in the bulk amino acid sequence properties of human versus eubacterial proteins that reflect likely differences in their biophysical properties including crystallization propensity. Finally, our thermodynamic measurements enable critical evaluation of previous claims regarding correlations between protein stability and bulk sequence properties, which generally are not supported by our dataset. NIH Public Access Author ManuscriptNat Biotechnol. Author manuscript; available in PMC 2010 January 1. Published in final edited form as:Nat Biotechnol. 2009 January ; 27(1): 51-57. doi:10.1038/nbt.1514. NIH-PA Author ManuscriptNIH-PA Author Manuscript NIH-PA Author ManuscriptThe ability to determine the atomic structures of macromolecules represents a great achievement in molecular biology because of the unparalleled value of this information in understanding the fundamental chemistry of life [1][2][3][4][5] . While nuclear magnetic resonance represents an invaluable source of structural information, especially for small proteins, most macromolecular structures are determined using x-ray crystallography. Capitalizing on the recent proliferation of genomic sequence data, "structural genomics" consortia have been organized worldwide to develop methods and infrastructure for high-throughput protein structure determination. These groups have contributed to improvements in expression and structure determination methods 6 , and the four largest U.S. consortia accounted for 45% of all novel structures deposited in the Protein Data Bank (PDB) in 2007 7 . While these efforts contribute to the impressive progress of the structural biology community in characterizing the full repertoire of protein structures, the rate of growth in sequence information nonetheless far out-paces that of structural information. Given the ongoing acceleration of whole-genome sequencing, the gap between the two will continue to expand without a breakthrough in macromolecular structure determination methods.The systematic efforts of structural genomics projects show that crystallization is the major bottleneck to protein structure determinati...