T.A. Jones

Improved methods for building protein models in electron density maps and the location of errors in these models.

Map interpretation remains a critical step in solving the structure of a macromolecule. Errors introduced at this early stage may persist throughout crystallographic refinement and result in an incorrect structure. The normally quoted crystallographic residual is often a poor description for the quality of the model. Strategies and tools are described that help to alleviate this problem. These simplify the model-building process, quantify the goodness of fit of the model on a per-residue basis and locate possible errors in peptide and side-chain conformations.

A graphics model building and refinement system for macromolecules

A model building and refinement system is described for use with a Vector General 3400 display. The system allows the user to build models using guide atoms and angles to arrive at the final conformation. It has been used to assist in difference Fourier map interpretation at medium and high resolution, and to build a protein molecule into a multiple isomorphous replacement phased electron density map.

Detection, Delineation, Measurement and Display of Cavities in Macromolecular Structures

A computer program, VOIDOO, is described which can be employed in the study of cavities such as they occur in macromolecular structures (in particular, in proteins). The program can be used to detect unknown cavities or to delineate known cavities, either of which may be connected to the outside of the molecule or molecular assembly under study. Optionally, output files can be requested that contain a description of the shape of the cavity which can be displayed by the crystallographic modelling program O. Additionally, VOIDOO can be used to calculate the volume of a molecule and to create a file containing data pertaining to the surface of the molecule which can also be displayed using O. Examples of the use of VOIDOO are given for P2 myelin protein, cellular retinol-binding protein and cellobiohydrolase II. Finally, operational definitions to discern different types of cavity are introduced and guidelines for assessing the accuracy and improving the comparability of cavity calculations are given.

Using known substructures in protein model building and crystallography.

Retinol binding protein can be constructed from a small number of large substructures taken from three unrelated proteins. The known structures are treated as a knowledge base from which one extracts information to be used in molecular modelling when lacking true atomic resolution. This includes the interpretation of electron density maps and modelling homologous proteins. Models can be built into maps more accurately and more quickly. This requires the use of a skeleton representation for the electron density which improves the determination of the initial chain tracing. Fragment‐matching can be used to bridge gaps for inserted residues when modelling homologous proteins.

ELECTRON-DENSITY MAP INTERPRETATION

Publisher Summary Any errors that occur in a crystallographic project usually will be found before publication. Often, if an error has been made, the project will stall and there will be no publication. Introducing a serious error in a model can be different. This chapter discusses the kinds of error that might be made and why these errors are made. It discusses some of the features of the crystallographic model-building program O. Real errors in models occur with frequencies that are, fortunately, inversely proportional to the seriousness of the error. Building a molecular model from electron density is a complicated process. During the interpretation of an electron-density map, the basic function of the molecular graphics program is to assist the scientist in imagining, and then remembering, the three-dimensional folding and features of the structure. Thus, it is important to be able to change the model quickly and not to be interrupted by the details of operating a computer program. To facilitate the rapid building and rebuilding of molecular models, O incorporates autobuild options, allowing the user to create a molecular structure quickly from a rough three-dimensional sketch. This has the drawback of possibly making it even easier to build a wrong structure.

Databases in protein crystallography

Applications of structural databases in the protein crystallographic structure determination process are reviewed, using mostly examples from work carried out by the authors. Four application areas are discussed: model building, model refinement, model validation and model analysis.

Detecting folding motifs and similarities in protein structures

Publisher Summary Detecting similarities at the level of tertiary structure is of interest for at least three reasons—namely, (1) it may provide insight into the modus operandi of proteins that share a common structural and functional trait, (2) it may reveal evolutionary pathways (either divergent or convergent), and (3) it may provide insight into protein folding and stability by revealing that a certain arrangement of helices and strands occurs in unrelated proteins. In all cases, if similarities at the tertiary structure level exist, sequence alignments based on these similarities are important. Such structure-based sequence alignments are expected to correlate with functional similarities. A cluster analysis using several known, high-resolution structures is carried out to find “typical” geometries of consecutive stretches of five Cα atoms in α helices and β strands, respectively.

Structure of a triclinic ternary complex of horse liver alcohol dehydrogenase at 2.9 A resolution.

Abstract The structure of a triclinic complex between liver alcohol dehydrogenase, reduced coenzyme NADH, and the inhibitor dimethylsulfoxide has been determined to 2.9 A resolution using isomorphous replacement methods. The heavy-atom positions were derived by molecular replacement methods using phase angles derived from a model of the orthorhombic apoenzyme structure previously determined to 2.4 A resolution. A model of the present holoenzyme molecule was built on a Vector General 3400 display system using the RING system of programs. This model gave a crystallographic R -value of 37.9%. There are extensive conformational differences between the protein molecules in the two forms. The conformational change involves a rotation of 7.5 ° of the catalytic domains relative to the coenzyme binding domains. A hinge region for this rotation is defined within a hydrophobic core between two helices. The internal structures of the domains are preserved with the exception of a movement of a small loop in the coenzyme binding domain. A cleft between the domains is closed by this coenzyme-induced conformational change, making the active site less accessible from solution and thus more hydrophobic. The two crystallographically independent subunits are very similar and bind both coenzyme and inhibitor in an identical way within the present limits of error. The coenzyme molecule is bound in an extended conformation with the two ends in hydrophobic crevices on opposite sides of the central pleated sheet of the coenzyme binding domain. There are hydrogen bonds to oxygen atoms of the ribose moities from Asp223, Lys228 and His51. The pyrophosphate group is in contact with the side-chains of Arg47 and Arg369. No new residues are brought into the active site compared to the apoenzyme structure. The active site zinc atom is close to the hinge region, where the smallest structural changes occur. Small differences in the co-ordination geometry of the ligands Cys46, His67 and Cysl74 are not excluded and may account for the ordered mechanism. The oxygen atom of the inhibitor dimethylsulfoxide is bound directly to zinc confirming the structural basis for the suggested mechanism of action based on studies of the apoenzyme structure.

The three-dimensional structure of retinol-binding protein.

The complex of retinol with its carrier protein, retinol‐binding protein (RBP) has been crystallized and its three‐dimensional structure determined using X‐ray crystallography. Its most striking feature is an eight‐stranded up‐and‐down beta barrel core that completely encapsulates the retinol molecule. The retinol molecule lies along the axis of the barrel with the beta‐ionone ring innermost and the tip of the isoprene tail close to the surface.

Model building and refinement practice.

Publisher Summary Model refinement has been a personalized affair for which laboratories have their preferred strategies, programs, etc. This has resulted in models with distinctive features of both the groups concerned and the software used. This chapter discusses the way a macromolecule should be refined and argues that the present practices in the community are often far from optimal, especially when only low-resolution data are available. All refinement programs nowadays use empirical restraints or constraints to ensure that a reasonable structure ensues during the refinement steps. This can result in a model with good stereochemical properties and also in a model in which molecules related by non-crystallographic symmetry (NCS) are forced to have similar (restrained) or identical (constrained) conformations. The aim of model building and refinement should be to construct a model that adequately explains the experimental observations, while making physical, chemical, and biological sense. It is a fact that low-resolution data can yield only low-resolution models. The refinement process, in particular, should always be tailored for each problem individually, keeping in mind the amount, resolution, and quality of the data.