Protein Structure Initiative: getting into gear.

The Protein Structure Initiative of the US National Institutes of Health has entered its second year. Its status is reported here, as discussed at December's meeting of the principal investigators.

Use of experimental crystallographic phases to examine the hydration of polar and nonpolar cavities in T4 lysozyme.

There is conflicting evidence as to whether cavities in proteins that are nonpolar and large enough to accommodate solvent are empty or are occupied by disordered water molecules. Here, we use multiple-wavelength x-ray data collected from crystals of the selenomethionine-substituted L99A/M102L mutant of T4 lysozyme to obtain a high-resolution electron density map free of bias that is unavoidably associated with conventional model-based structure determination and refinement. The mutant, L99A/M102L, has four cavities, two being polar in character and the other two nonpolar. Cavity 1 (polar, volume 45.2 A(3)) was expected to contain two well ordered water molecules, and this is confirmed in the experimental electron density map. Likewise, cavity 2 (polar, 16.9 A(3)) is confirmed to contain a single water molecule. Cavity 3 (nonpolar, 21.4 A(3)) was seen to be empty in conventional x-ray refinement, and this is confirmed in the experimental map. Unexpectedly, however, cavity 4 (nonpolar, volume 133.5 A(3)) was seen to contain diffuse electron density equivalent to approximately 1.5 water molecules. Although cavity 4 is largely nonpolar, it does have some polar character, and this apparently contributes to the presence of solvent. The cavity is large enough to accommodate four to five water molecules, and it appears that a hydrogen-bonded chain of three or more solvent molecules could occupy the cavity at a given time. The results are consistent with theoretical predictions that cavities in proteins that are strictly nonpolar will not contain solvent until the volume is large enough to permit mutually satisfying water-water hydrogen bonds.

Use of stabilizing mutations to engineer a charged group within a ligand-binding hydrophobic cavity in T4 lysozyme.

Both large-to-small and nonpolar-to-polar mutations in the hydrophobic core of T4 lysozyme cause significant loss in stability. By including supplementary stabilizing mutations we constructed a variant that combines the cavity-creating substitution Leu99 --> Ala with the buried charge mutant Met102 --> Glu. Crystal structure determination confirmed that this variant has a large cavity with the side chain of Glu102 located within the cavity wall. The cavity includes a large disk-shaped region plus a bulge. The disk-like region is essentially nonpolar, similar to L99A, while the Glu102 substituent is located in the vicinity of the bulge. Three ordered water molecules bind within this part of the cavity and appear to stabilize the conformation of Glu102. Glu102 has an estimated pKa of about 5.5-6.5, suggesting that it is at least partially charged in the crystal structure. The polar ligands pyridine, phenol and aniline bind within the cavity, and crystal structures of the complexes show one or two water molecules to be retained. Nonpolar ligands of appropriate shape can also bind in the cavity and in some cases exclude all three water molecules. This disrupts the hydrogen-bond network and causes the Glu102 side chain to move away from the ligand by up to 0.8 A where it remains buried in a completely nonpolar environment. Isothermal titration calorimetry revealed that the binding of these compounds stabilizes the protein by 4-6 kcal/mol. For both polar and nonpolar ligands the binding is enthalpically driven. Large negative changes in entropy adversely balance the binding of the polar ligands, whereas entropy has little effect on the nonpolar ligand binding.

Structural basis for the unusual specificity of Escherichia coli aminopeptidase N.

Aminopeptidase N from Escherichia coli is a M1 class aminopeptidase with the active-site region related to that of thermolysin. The enzyme has unusual specificity, cleaving adjacent to the large, nonpolar amino acids Phe and Tyr but also cleaving next to the polar residues Lys and Arg. To try to understand the structural basis for this pattern of hydrolysis, the structure of the enzyme was determined in complex with the amino acids L-arginine, L-lysine, L-phenylalanine, L-tryptophan, and L-tyrosine. These amino acids all bind with their backbone atoms close to the active-site zinc ion and their side chain occupying the S1 subsite. This subsite is in the form of a cylinder, about 10 A in cross-section and 12 A in length. The bottom of the cylinder includes the zinc ion and a number of polar side chains that make multiple hydrogen-bonding and other interactions with the alpha-amino group and the alpha-carboxylate of the bound amino acid. The walls of the S1 cylinder are hydrophobic and accommodate the nonpolar or largely nonpolar side chains of Phe and Tyr. The top of the cylinder is polar in character and includes bound water molecules. The epsilon-amino group of the bound lysine side chain and the guanidinium group of arginine both make multiple hydrogen bonds to this part of the S1 site. At the same time, the hydrocarbon part of the lysine and arginine side chains is accommodated within the nonpolar walls of the S1 cylinder. This combination of hydrophobic and hydrophilic binding surfaces explains the ability of ePepN to cleave Lys, Arg, Phe, and Tyr. Another favored substrate has Ala at the P1 position. The short, nonpolar side chain of this residue can clearly be bound within the hydrophobic part of the S1 cylinder, but the reason for its facile hydrolysis remains uncertain.

Structural and thermodynamic characterization of T4 lysozyme mutants and the contribution of internal cavities to pressure denaturation.

Using small-angle X-ray scattering (SAXS) and tryptophan fluorescence spectroscopy, we have identified multiple compact denatured states of a series of T4 lysozyme mutants that are stabilized by high pressures. Recent studies imply that the mechanism of pressure denaturation is the penetration of water into the protein rather than the transfer of hydrophobic residues into water. To investigate water penetration and the volume change associated with pressure denaturation, we studied the solution behavior of four T4 lysozyme mutants having different cavity volumes at low and neutral pH up to a pressure of 400 MPa (0.1 MPa = 0.9869 atm). At low pH, L99A T4 lysozyme expanded from a compact folded state to a partially unfolded state with a corresponding change in radius of gyration from 17 to 32 A. The volume change upon denaturation correlated well with the total cavity volume, indicating that all of the molecule's major cavities are hydrated with pressure. As a direct comparison to high-pressure crystal structures of L99A T4 lysozyme solved at neutral pH [Collins, M. D., Hummer, G., Quillin, M. L., Matthews, B. W., and Gruner, S. M. (2005) Proc. Natl. Acad. Sci. U.S.A. 102, 16668-16671], pressure denaturation of L99A and the structurally similar L99G/E108V mutant was studied at neutral pH. The pressure-denatured state at neutral pH is even more compact than at low pH, and the small volume changes associated with denaturation suggest that the preferential filling of large cavities is responsible for the compactness of the pressure-denatured state. These results confirm that pressure denaturation is characteristically distinct from thermal or chemical denaturation.

Structural rigidity of a large cavity-containing protein revealed by high-pressure crystallography.

Steric constraints, charged interactions and many other forces important to protein structure and function can be explored by mutagenic experiments. Research of this kind has led to a wealth of knowledge about what stabilizes proteins in their folded states. To gain a more complete picture requires that we perturb these structures in a continuous manner, something mutagenesis cannot achieve. With high pressure crystallographic methods it is now possible to explore the detailed properties of proteins while continuously varying thermodynamic parameters. Here, we detail the structural response of the cavity-containing mutant L99A of T4 lysozyme, as well as its pseudo wild-type (WT*) counterpart, to hydrostatic pressure. Surprisingly, the cavity has almost no effect on the pressure response: virtually the same changes are observed in WT* as in L99A under pressure. The cavity is most rigid, while other regions deform substantially. This implies that while some residues may increase the thermodynamic stability of a protein, they may also be structurally irrelevant. As recently shown, the cavity fills with water at pressures above 100 MPa while retaining its overall size. The resultant picture of the protein is one in which conformationally fluctuating side groups provide a liquid-like environment, but which also contribute to the rigidity of the peptide backbone.

Co-repressor induced order and biotin repressor dimerization: a case for divergent followed by convergent evolution.

BirA catalyzes the adenylation and subsequent covalent attachment of biotin to the biotin carboxyl carrier protein (BCCP). In the absence of apo-BCCP, biotin-5'-AMP acts as a co-repressor that induces BirA dimerization and binding to the bio operator to repress biotin biosynthesis. The crystal structures of apo-BirA, and BirA in complex with biotin have been reported. We here describe the 2.8A resolution crystal structure of BirA in complex with the co-repressor analog biotinol-5'-AMP. It was previously shown that the structure of apo-BirA is monomeric and that binding of biotin weakly induces a dimeric structure in which three disordered surface loops become organized to form the dimer interface. The structure of the co-repressor complex is also a dimer, clearly related to the BirA.biotin structure, but with several significant conformational changes. A hitherto disordered "adenylate binding loop" forms a well-defined structure covering the co-repressor. The co-repressor buttresses the dimer interface, resulting in improved packing and a 12 degrees change in the hinge-bending angle along the dimer interface relative to the BirA.biotin structure. This helps explain why the binding of the co-repressor is necessary to optimize the binding of BirA to the bioO operator. The structure reveals an unexpected use of the nucleotide-binding motif GXGXXG in binding adenylate and controlling the repressor function. Finally, based on structural analysis we propose that the class of adenylating enzymes represented by BirA, lipoate protein ligase and class II tRNA synthetases diverged early and were selected based on their ability to sequester co-factors or amino acid residues, and adenylation activity arose independently through functional convergence.

Structural basis of Prospero-DNA interaction: implications for transcription regulation in developing cells.

The crystal structure of a complex between the novel homeodomain of the neural transcription factor Prospero and DNA shows that the invariant residues Lys1290, Asn1294, and Asp1297 make specific contacts with the noncanonical DNA binding site. The overall structure includes the homeodomain and the adjacent Prospero domain and confirms that they act as a single structural unit, a Homeo-Prospero domain. The Prospero domain facilitates the proper alignment of the protein on the DNA. Knowledge of the structure reconciles two different DNA sequences that have been proposed as transcriptional targets for Prospero. As in the apo structure, the C terminus of the Prospero domain shields a short helix within the homeodomain that includes a nuclear export signal (NES). The structural results suggest that exposure of the NES is not coupled directly to DNA binding. We propose a DNA recognition mechanism specific to Prospero-type homeodomains in developing cells.

Structure of the angiogenesis inhibitor ovalicin bound to its noncognate target, human Type 1 methionine aminopeptidase.

Methionine aminopeptidases (MetAPs) remove the initiator methionine during protein biosynthesis. They exist in two isoforms, MetAP1 and MetAP2. The anti-angiogenic compound fumagillin binds tightly to the Type 2 MetAPs but only weakly to Type 1. High-affinity complexes of fumagillin and its relative ovalicin with Type 2 human MetAP have been reported. Here we describe the crystallographic structure of the low-affinity complex between ovalicin and Type 1 human MetAP at 1.1 A resolution. This provides the first opportunity to compare the structures of ovalicin or fumagillin bound to a Type 1 and a Type 2 MetAP. For both Type 1 and Type 2 human MetAPs the inhibitor makes a covalent adduct with a corresponding histidine. At the same time there are significant differences in the alignment of the inhibitors within the respective active sites. It has been argued that the lower affinity of ovalicin and fumagillin for the Type 1 MetAPs is due to the smaller size of their active sites relative to the Type 2 enzymes. Comparison with the uncomplexed structure of human Type 1 MetAP indicates that there is some truth to this. Several active site residues have to move "outward" by 0.5 Angstroms or so to accommodate the inhibitor. Other residues move "inward." There are, however, other factors that come into play. In particular, the side chain of His310 rotates by 134 degrees into a different position where (together with Glu128 and Tyr195) it coordinates a metal ion not seen at this site in the native enzyme.

Sequential reorganization of beta-sheet topology by insertion of a single strand.

Insertions, duplications, and deletions of sequence segments are thought to be major evolutionary mechanisms that increase the structural and functional diversity of proteins. Alternative splicing, for example, is an intracellular editing mechanism that is thought to generate isoforms for 30%-50% of all human genes. Whereas the inserted sequences usually display only minor structural rearrangements at the insertion site, recent observations indicate that they may also cause more dramatic structural displacements of adjacent structures. In the present study we test how artificially inserted sequences change the structure of the beta-sheet region in T4 lysozyme. Copies of two different beta-strands were inserted into two different loops of the beta-sheet, and the structures were determined. Not surprisingly, one insert "loops out" at its insertion site and forms a new small beta-hairpin structure. Unexpectedly, however, the second insertion leads to displacement of adjacent strands and a sequential reorganization of the beta-sheet topology. Even though the insertions were performed at two different sites, looping out occurred at the C-terminal end of the same beta-strand. Reasons as to why a non-native sequence would be recruited to replace that which occurs in the native protein are discussed. Our results illustrate how sequence insertions can facilitate protein evolution through both local and nonlocal changes in structure.

Guanidinium derivatives bind preferentially and trigger long-distance conformational changes in an engineered T4 lysozyme.

The binding of guanidinium ion has been shown to promote a large-scale translation of a tandemly duplicated helix in an engineered mutant of T4 lysozyme. The guanidinium ion acts as a surrogate for the guanidino group of an arginine side chain. Here we determine whether methyl- and ethylguanidinium provide better mimics. The results show that addition of the hydrophobic moieties to the ligand enhances the binding affinity concomitant with reduction in ligand solubility. Crystallographic analysis confirms that binding of the alternative ligands to the engineered site still drives the large-scale conformational change. Thermal analysis and NMR data show, in comparison to guanidinium, an increase in protein stability and in ligand affinity. This is presumably due to the successive increase in hydrophobicity in going from guanidinium to ethylguanidinium. A fluorescence-based optical method was developed to sense the ligand-triggered helix translation in solution. The results are a first step in the de novo design of a molecular switch that is not related to the normal function of the protein.

Structural analysis of the group II intron splicing factor CRS2 yields insights into its protein and RNA interaction surfaces.

Chloroplast RNA splicing 2 (CRS2) is a nuclear-encoded protein required for the splicing of nine group II introns in maize chloroplasts. CRS2 functions in the context of splicing complexes that include one of two CRS2-associated factors (CAF1 and CAF2). The CRS2-CAF1 and CRS2-CAF2 complexes are required for the splicing of different subsets of CRS2-dependent introns, and they bind tightly and specifically to their genetically defined intron targets in vivo. The CRS2 amino acid sequence is closely related to those of bacterial peptidyl-tRNA hydrolases (PTHs). To identify the structural differences between CRS2 and bacterial PTHs responsible for CRS2's gains of CAF binding and intron splicing functions, we determined the structure of CRS2 by X-ray crystallography. The fold of CRS2 is the same as that of Escherichia coli PTH, but CRS2 has two surfaces that differ from the corresponding surfaces in PTH. One of these is more hydrophobic in CRS2 than in PTH. Site-directed mutagenesis of this surface blocked CRS2-CAF complex formation, indicating that it is the CAF binding site. The CRS2 surface corresponding to the putative tRNA binding face of PTH is considerably more basic than in PTH, suggesting that CRS2 interacts with group II intron substrates via this surface. Both the sequence and the structural context of the amino acid residues essential for peptidyl-tRNA hydrolase activity are conserved in CRS2, yet expression of CRS2 is incapable of rescuing a pth(ts)E.coli strain.

Structure of the DNA binding region of prospero reveals a novel homeo-prospero domain.

The Prospero transcription factor promotes neural differentiation in Drosophila, and its activity is tightly regulated by modulating its subcellular localization. Prospero is exported from the nucleus of neural precursors but imported into the nucleus of daughter cells, which is necessary for their proper differentiation. Prospero has a highly divergent putative homeodomain adjacent to a conserved Prospero domain; both are required for sequence-specific DNA binding. Here we show that the structure of these two regions consists of a single structural unit (a homeo-prospero domain), in which the Prospero domain region is in position to contribute to DNA binding and also to mask a defined nuclear export signal that is within the putative homeodomain region. We propose that the homeo-prospero domain coordinately regulates Prospero nuclear localization and DNA binding specificity.

Crystal structure of E. coli YhbY: a representative of a novel class of RNA binding proteins.

E. coli YhbY belongs to a conserved family of hypothetical proteins represented in eubacteria, archaea, and plants (Pfam code UPF0044). Three maize proteins harboring UPF0044-like domains are required for chloroplast group II intron splicing, and bioinformatic data suggest a role for prokaryotic UPF0044 members in translation. The crystal structure of YhbY has been determined. YhbY has a fold similar to that of the C-terminal domain of translation initiation factor 3 (IF3C), which binds to 16S rRNA in the 30S ribosome. Modeling studies indicate that the same surface is highly basic in all members of UPF0044, suggesting a conserved RNA binding surface. Taken together, the evidence suggests that members of UPF0044 constitute a previously unrecognized class of RNA binding domain.

Crystal structures of a T4-lysozyme duplication-extension mutant demonstrate that the highly conserved beta-sheet region has low intrinsic folding propensity.

Residues 24 to 35 of T4 lysozyme correspond to the second and third strands of a region of beta-sheet that is highly conserved in all known lysozyme and chitinase structures. To evaluate the intrinsic propensity of these amino acid residues to form a defined structure they were added at the C terminus of the native protein, together with a dipeptide linker. Two crystal structures of this active, mutant protein were obtained, to 1.9A and 2.3A resolution, respectively. Even though the crystal conditions are similar, the appended sequence adopts very different secondary structures. In one case it is weakly structured and appears to extend through the active-site cleft, perhaps in part adding an extra strand to the original beta-sheet. In the other crystal form the extension is largely alpha-helical. The formation of these alternative structures shows that the sequence does not have a strong intrinsic propensity to form a unique fold (either beta-sheet or otherwise). The results also suggest that structural conservation during evolution does not necessarily depend on sequence conservation or the conservation of folding propensity.

Testing a flexible-receptor docking algorithm in a model binding site.

Sampling receptor flexibility is challenging for database docking. We consider a method that treats multiple flexible regions of the binding site independently, recombining them to generate different discrete conformations. This algorithm scales linearly rather than exponentially with the receptor's degrees of freedom. The method was first evaluated for its ability to identify known ligands of a hydrophobic cavity mutant of T4 lysozyme (L99A). Some 200000 molecules of the Available Chemical Directory (ACD) were docked against an ensemble of cavity conformations. Surprisingly, the enrichment of known ligands from among a much larger number of decoys in the ACD was worse than simply docking to the apo conformation alone. Large decoys, accommodated in the larger cavity conformations sampled in the ensemble, were ranked better than known small ligands. The calculation was redone with an energy correction term that considered the cost of forming the larger cavity conformations. Enrichment improved, as did the balance between high-ranking large and small ligands. In a second retrospective test, the ACD was docked against a conformational ensemble of thymidylate synthase. Compared to docking against individual enzyme conformations, the flexible receptor docking approach improved enrichment of known ligands. Including a receptor conformational energy weighting term improved enrichment further. To test the method prospectively, the ACD database was docked against another cavity mutant of lysozyme (L99A/M102Q). A total of 18 new compounds predicted to bind this polar cavity and to change its conformation were tested experimentally; 14 were found to bind. The bound structures for seven ligands were determined by X-ray crystallography. The predicted geometries of these ligands all corresponded to the observed geometries to within 0.7A RMSD or better. Significant conformational changes of the cavity were observed in all seven complexes. In five structures, part of the observed accommodations were correctly predicted; in two structures, the receptor conformational changes were unanticipated and thus never sampled. These results suggest that although sampling receptor flexibility can lead to novel ligands that would have been missed when docking a rigid structure, it is also important to consider receptor conformational energy.