Search

Structure comparison is a crucial aspect of structural biology today. The field of structure comparison is developing rapidly, with the development of new algorithms, similarity scores, and statistical scores. The predicted large increase of experimental structures and structural models made possible by high-throughput efforts means that structural comparison and searching of structural databases using automated methods will become increasingly common. This Ways & Means article is meant to guide the structural biologist in the basics of structural alignment, and to provide an overview of the available software tools. The main purpose is to encourage users to gain some understanding of the strengths and limitations of structural alignment, and to take these factors into account when interpreting the results of different programs.

The crystal structures of Family 7 glycohydrolases suggest that a histidine residue near the acid/base catalyst could account for the higher pH optimum of the Humicola insolens endoglucanase Cel7B, than the corresponding Trichoderma reesei enzymes. Modelling studies indicated that introduction of histidine at the homologous position in T. reesei Cel7A (Ala(224)) required additional changes to accommodate the bulkier histidine side chain. X-ray crystallography of the catalytic domain of the E223S/A224H/L225V/T226A/D262G mutant reveals that major differences from the wild-type are confined to the mutations themselves. The introduced histidine residue is in plane with its counterpart in H. insolens Cel7B, but is 1.0 A (=0.1 nm) closer to the acid/base Glu(217) residue, with a 3.1 A contact between N(epsilon2) and O(epsilon1). The pH variation of k(cat)/K(m) for 3,4-dinitrophenyl lactoside hydrolysis was accurately bell-shaped for both wild-type and mutant, with pK(1) shifting from 2.22+/-0.03 in the wild-type to 3.19+/-0.03 in the mutant, and pK(2) shifting from 5.99+/-0.02 to 6.78+/-0.02. With this poor substrate, the ionizations probably represent those of the free enzyme. The relative k(cat) for 2-chloro-4-nitrophenyl lactoside showed similar behaviour. The shift in the mutant pH optimum was associated with lower k(cat)/K(m) values for both lactosides and cellobiosides, and a marginally lower stability. However, k(cat) values for cellobiosides are higher for the mutant. This we attribute to reduced non-productive binding in the +1 and +2 subsites; inhibition by cellobiose is certainly relieved in the mutant. The weaker binding of cellobiose is due to the loss of two water-mediated hydrogen bonds.

Chapter 21.1

Chapter 17.1

Acyl-CoA binding protein (ACBP) maintains a pool of fatty acyl-CoA molecules in the cell and plays a role in fatty acid metabolism. The biochemical properties of Plasmodium falciparum ACBP are described together with the 2.0 A resolution crystal structures of a P. falciparum ACBP-acyl-CoA complex and of bovine ACBP in two crystal forms. Overall, the bovine ACBP crystal structures are similar to the NMR structures published previously; however, the bovine and parasite ACBP structures are less similar. The parasite ACBP is shown to have a different ligand-binding pocket, leading to an acyl-CoA binding specificity different from that of bovine ACBP. Several non-conservative differences in residues that interact with the ligand were identified between the mammalian and parasite ACBPs. These, together with measured binding-specificity differences, suggest that there is a potential for the design of molecules that might selectively block the acyl-CoA binding site.

Since the process of building and refining a model of a biomacromolecule based on crystallographic data is subjective, quality-control techniques are required to assess the validity of such models. During the 1990s, much experience was gained; the methods used and some of the lessons learned are reviewed here. In addition, an extensive compendium of quality criteria and quality-control methods that are or have been used to validate models of biomacromolecules has been compiled. The emphasis in this compendium is on the validation of protein crystal structures.

We report our evaluation of the CASP3 comparative modelling competition. Our analysis covers the accuracy of the over-all fold, the bridging of insertions and deletions, and the adding of side-chains. We describe our attempts at automating aspects of the evaluation.

Retinoids play important roles in diverse cellular processes including growth, cell differentiation and vision. Many natural and synthetic retinoids are used as drugs in dermatology and oncology. A large amount of data has been accumulated on the cellular activity of different synthetic retinoids. They are stabilized and transported inside the cell cytoplasm by binding and transport proteins, such as cellular retinol-binding proteins and cellular retinoic acid binding proteins (CRABPs). The structures of human CRABP II in complex with two different synthetic retinoids, Ro13-6307 and Ro12--7310 (at 2.1 and 2.0 A resolution, respectively) and of bovine CRABP I in complex with a retinobenzoic acid, Am80 (at 2.8 A resolution) are described. The binding affinities of human CRABP I and II for the retinoids studied here have been determined. All these compounds have comparable binding affinities (nanomolar range) for both CRABPs. Apart from the particular interactions of the carboxylate group of the retinoids with specific protein groups, each structure reveals characteristic interactions. Studying the atomic details of the interaction of retinoids with retinoid-binding proteins facilitates the understanding of the kinetics of retinoid trafficking inside the cytoplasm.

Prior to attaching any biological significance to differences between two related protein crystal structures, it must be established that such differences are genuine, rather than artefacts of the structure-determination protocol. This will be all the more important as more and more related protein structures are solved and comparative structural biology attempts to correlate structural differences with variations in biological function, activity or affinity. A method has been developed which enables unbiased assessment of differences between the structures of related biomacromolecules using experimental crystallographic information alone. It is based on the use of local density-correlation maps, which contain information regarding the similarity of the experimental electron density for corresponding parts of different copies of a molecule. The method can be used to assess a priori which parts of two or more molecules are likely to be structurally similar; this information can then be employed during structure refinement. Alternatively, the method can be used a posteriori to verify that differences observed in two or more models are supported by the experimental information. Several examples are discussed which validate the notion that local conformational variability is highly correlated to differences in the local experimental electron density.

BACKGROUND: Cel6A is one of the two cellobiohydrolases produced by Trichoderma reesei. The catalytic core has a structure that is a variation of the classic TIM barrel. The active site is located inside a tunnel, the roof of which is formed mainly by a pair of loops. RESULTS: We describe three new ligand complexes. One is the structure of the wild-type enzyme in complex with a nonhydrolysable cello-oligosaccharide, methyl 4-S-beta-cellobiosyl-4-thio-beta-cellobioside (Glc)(2)-S-(Glc)(2), which differs from a cellotetraose in the nature of the central glycosidic linkage where a sulphur atom replaces an oxygen atom. The second structure is a mutant, Y169F, in complex with the same ligand, and the third is the wild-type enzyme in complex with m-iodobenzyl beta-D-glucopyranosyl-beta(1,4)-D-xylopyranoside (IBXG). CONCLUSIONS: The (Glc)(2)-S-(Glc)(2) ligand binds in the -2 to +2 sites in both the wild-type and mutant enzymes. The glucosyl unit in the -1 site is distorted from the usual chair conformation in both structures. The IBXG ligand binds in the -2 to +1 sites, with the xylosyl unit in the -1 site where it adopts the energetically favourable chair conformation. The -1 site glucosyl of the (Glc)(2)-S-(Glc)(2) ligand is unable to take on this conformation because of steric clashes with the protein. The crystallographic results show that one of the tunnel-forming loops in Cel6A is sensitive to modifications at the active site, and is able to take on a number of different conformations. One of the conformational changes disrupts a set of interactions at the active site that we propose is an integral part of the reaction mechanism.

As the structural database continues to expand, new methods are required to analyse and compare protein structures. Whereas the recognition, comparison, and classification of folds is now more or less a solved problem, tools for the study of constellations of small numbers of residues are few and far between. In this paper, two programs are described for the analysis of spatial motifs in protein structures. The first, SPASM, can be used to find the occurrence of a motif consisting of arbitrary main-chain and/or side-chains in a database of protein structures. The program also has a unique capability to carry out "fuzzy pattern matching" with relaxed requirements on the types of some or all of the matching residues. The second program, RIGOR, scans a single protein structure for the occurrence of any of a set of pre-defined motifs from a database. In one application, spatial motif recognition combined with profile analysis enabled the assignment of the structural and functional class of an uncharacterised hypothetical protein in the sequence database. In another application, the occurrence of short left-handed helical segments in protein structures was investigated, and such segments were found to be fairly common. Potential applications of the techniques presented here lie in the analysis of (newly determined) structures, in comparative structural analysis, in the design and engineering of novel functional sites, and in the prediction of structure and function of uncharacterised proteins. Copyright 1999 Academic Press.

Macromolecular phase-refinement and phase-extension calculations using real-space electron-density averaging techniques require accurate envelopes (or masks) to define the boundaries of each domain or molecule whose density is to be averaged. An extensive set of tools, implemented in four computer programs (O, MAMA, COMA and MASKIT) are described which can be used to generate such envelopes (either from an atomic model or based on local density-correlation maps), to improve them, to remove overlap owing to crystallographic or non-crystallographic symmetry, to display them and to manipulate them in a variety of manners.

Alpha 2u-globulin (A2U) is the major urinary protein excreted by adult male rats. The structure of a monoclinic crystal form of A2U was reported in 1992 [Bocskei et al. (1992). Nature (London), 360, 186-188]. The structures of an orthorhombic crystal form of A2U at 2. 5 A resolution (refined to an R factor of 0.248; Rfree = 0.264) and of a complex between A2U and d-limonene 1,2-epoxide (DLO) at 2.9 A resolution (R factor = 0.248; Rfree = 0.260) are presented here. DLO is one of a diverse group of chemicals which cause a male rat-specific renal carcinogenesis called hyaline-droplet nephropathy. The rate-determining step in the development of this disorder is the binding of the toxin to A2U. Comparison of the cavities in A2U and in the corresponding mouse urinary protein (MUP) reveal that the former is tailor-made for small oval hydrophobic ligands such as DLO. The cavity in MUP is more shallow and elongated and cannot easily accommodate such ligands.

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.

A conceptually simple real-space convolution method has been developed which can be used to detect or enhance structural features in experimental macromolecular electron-density maps. The method has been implemented in a computer program (ESSENS). One application of the method is in selectively visualizing secondary-structure elements in multiple isomorphous replacement (MIR) maps of proteins, prior to map interpretation. This application is demonstrated for MIR maps of P2 myelin protein [Jones, Bergfors, Sedzik & Unge (1988). EMBO J. 7, 1597-1604; Cowan, Newcomer & Jones (1993). J. Mol. Biol. 230, 1225-1246] and glyoxalase I [Cameron, Olin, Ridderstrom, Mannervik & Jones (1997). In preparation]. Another application is in finding the optimal orientation and position of a known structural fragment (e.g. a protein domain or a ligand) in any type of electron-density map (real-space or phased molecular replacement). This application is demonstrated for the complex of acetylcholinesterase and the snake toxin fasciculin II [Harel, Kleywegt, Ravelli, Silman & Sussman (1995). Structure, 3, 1355-1366] where the toxin was located in a map phased using the molecular-replacement solution for the acetylcholinesterase alone.

Cellulose is the most abundant polymer in the biosphere. Although generally resistant to degradation, it may be hydrolysed by cellulolytic organisms that have evolved a variety of structurally distinct enzymes, cellobiohydrolases and endoglucanases, for this purpose. Endoglucanase I (EG I) is the major endoglucanase produced by the cellulolytic fungus Trichoderma reesei, accounting for 5 to 10% of the total amount of cellulases produced by this organism. Together with EG I from Humicola insolens and T. reesei cellobiohydrolase I (CBH I), the enzyme is classified into family 7 of the glycosyl hydrolases, and it catalyses hydrolysis with a net retention of the anomeric configuration.The structure of the catalytic core domain (residues 1 to 371) of EG I from T. reesei has been determined at 3.6 A resolution by the molecular replacement method using the structures of T. reesei CBH I and H. insolens EG I as search models. By employing the 2-fold non-crystallographic symmetry (NCS), the structure was refined successfully, despite the limited resolution. The final model has an R-factor of 0.201 (Rfree 0.258).The structure of EG I reveals an extended, open substrate-binding cleft, rather than a tunnel as found in the homologous cellobiohydrolase CBH I. This confirms the earlier proposal that the tunnel-forming loops in CBH I have been deleted in EG I, which has resulted in an open active site in EG I, enabling it to function as an endoglucanase. Comparison of the structure of EG I with several related enzymes reveals structural similarities, and differences that relate to their biological function in degrading particular substrates. A possible structural explanation of the drastically different pH profiles of T. reesei and H. insolens EG I is proposed.

The Protein Data Bank contains a number of structures for which only the coordinates of the Calpha atoms have been deposited. Although many tools are available for the validation of all-atom protein models, hardly any of these can be used to assess the quality of models for which only Calpha coordinates are available. Two rapid and simple tests to assess the quality of the Calpha backbone of a protein model are described, one based on the distribution of Calpha-Calpha distances, and the other on the two-dimensional distribution of the angles and dihedrals formed by sequential Calpha atoms. Expected distributions were derived by analysing a set of 1343 high-resolution, all-atom protein models. The distance criterion is useful to discriminate between refined and unrefined models, whereas the angle/dihedral criterion can be used to discriminate between normal and possibly problematic Calpha models. The method has been applied to a set of 88 Calpha-only models from the Protein Data Bank. The tracing of two of the models that are outliers in this analysis has recently been shown to be incorrect. Other applications of the method are discussed.

Trichoderma reesei cellobiohydrolase II (CBHII) is an exoglucanase cleaving primarily cellobiose units from the non-reducing end of cellulose chains. The beta-1,4 glycosidic bond is cleaved by acid catalysis with an aspartic acid, D221, as the likely proton donor, and another aspartate, D175, probably ensuring its protonation and stabilizing charged reaction intermediates. The catalytic base has not yet been identified experimentally. The refined crystal structure of CBHII also shows a tyrosine residue, Y169, located close enough to the scissile bond to be involved in catalysis. The role of this residue has been studied by introducing a mutation Y169F, and analysing the kinetic and binding behavior of the mutated CBHII. The crystal structure of the mutated enzyme was determined to 2.0 A resolution showing no changes when compared with the structure of native CBHII. However, the association constants of the mutant enzyme for cellobiose and cellotriose are increased threefold and for 4-methylumbelliferyl cellobioside over 50-fold. The catalytic constants towards cellotriose and cellotetraose are four times lower for the mutant. These data suggest that Y169, on interacting with a glucose ring entering the second subsite in a narrow tunnel, helps to distort the glucose ring into a more reactive conformation. In addition, a change in the pH activity profile was observed. This indicates that Y169 may have a second role in the catalysis, namely to affect the protonation state of the active site carboxylates, D175 and D221.

A computer program, called OOPS, is described which facilitates and speeds up the process of rebuilding a protein structure inside its electron density and reduces the chances of local errors persevering throughout the crystallographic protein structure determination process. The program uses a set of criteria to judge how reasonable each protein residue is and it generates macros for the macromolecular crystallographic model-building program O [Jones, Zou, Cowan & Kjeldgaard (1991). Acta Cryst. A47, 110-119] which, when executed, will take the crystallographer on a journey along all suspect residues.

Two user-friendly computer programs are described for use in macromolecular X-ray crystallography, xdlMAPMAN provides an interface for electron-density map exchange between some of the most commonly used phase refinement, structure refinement and model- building programs. In addition, it contains several options to analyse and abstract such maps. xdlDATAMAN provides similar functionality for the analysis and manipulation of macromolecular reflection data sets. Both programs have a simple graphical user interface, and their source code has been put into the public domain.

Several methods to assess the (dis)similarity of protein structures objectively are described, some of which, when applied to non-crystallographically related protein models, are able to discriminate between significant differences and 'random noise'. Some of these methods have been used to investigate a sample of several hundred protein structures which have been solved by means of X-ray crystallography in order to investigate the extent to which non-crystallographically related protein models differ from one another. It is shown that the extent of such differences is largely dependent on the resolution of the data used for the determination and refinement of the structure and, measured by some statistics, even varies essentially linearly with the resolution. The implications of these findings for the strategies used to refine structures with non-crystallographic symmetry, in particular at low resolution, are discussed. Finally, two examples are given of recent structure determinations from this laboratory in which the presence (and employment) of non-crystallographic symmetry was crucial to the solution and refinement of the structure.

It is shown here that the reported 3 A crystal structure of chloromuconate cycloisomerase from Alcaligenes eutrophus [Hoier, Schlomann, Hammer, Glusker, Carrell, Goldman, Stezowski & Heinemann (1994). Acta Cryst. D50, 75-84] was refined in the incorrect space group I4. In addition, a stretch of about 25 residues near the N-terminus is out-of-register with the density in the original structure. From the coordinates and structure factors deposited in the Protein Data Bank (PDB), it was possible to determine the correct space group to be I422. The structure was then re-refined, using the original data reduced to I422, to a crystallographic free R factor of 0.264 at 3 A resolution (conventional R factor 0.189). With conservative refinement and rebuilding methods, the errors in the chain tracing could be identified and remedied. Since the two molecules per asymmetric unit in the original structure are actually related by crystallographic symmetry, the observed differences between them are artefacts. In particular, the differences between, and peculiarities of the metal-binding sites are unreal. This case shows the dangers of crystallographic refinement in cases with unfavourable data-to-parameter ratios, and the importance of reducing the number of parameters in such cases to prevent gross errors (for instance, by using NCS constraints). It also demonstrates how the evaluation and monitoring of model quality during the entire refinement and rebuilding process can be used to detect and remedy serious errors. Finally, it presents a strong case in favour of depositing not only model coordinates, but also experimental data (preferably, both merged and unmerged data).

BACKGROUND: Streptococcal protein G comprises two or three domains that bind to the constant Fc region of most mammalian immunoglobulin Gs (IgGs). Protein G is functionally related to staphylococcal protein A, with which it shares neither sequence nor structural homology. RESULTS: To understand the competitive binding of these two proteins to the Fc region, the crystal structure of a single Ig-binding domain of streptococcal protein G was determined at 3.5 A resolution in complex with the Fc fragment of human IgG and compared with the structures of protein A:Fc and protein G:Fab complexes. Protein G binds to the interface between the second and third heavy chain constant domains of Fc, which is roughly the same binding site used by protein A. Protein G comprises one alpha-helix packed onto a four-stranded beta-sheet. Residues from protein G that are involved in binding are situated within the C-terminal part of the alpha-helix, the N-terminal part of the third beta-strand and the loop region connecting these two structural elements. The identified Fc-binding region of protein G agrees well with both biochemical and NMR spectroscopic data. However, the Fc-binding helices of protein G and protein A are not superimposable. CONCLUSIONS: Protein G and protein A have developed different strategies for binding to Fc. The protein G:Fc complex involves mainly charged and polar contacts, whereas protein A and Fc are held together through non-specific hydrophobic interactions and a few polar interactions. Several residues of Fc are involved in both the protein G:Fc and the protein A:Fc interaction, which explains the competitive binding of the two proteins. The apparent differences in their Fc-binding activities result from additional unique interactions.

BACKGROUND: Retinoic acid (RA) plays a fundamental role in diverse cellular activities. Cellular RA binding proteins (CRABPs) are thought to act by modulating the amount of RA available to nuclear RA receptors. CRABPs and cellular retinol-binding proteins (CRBPs) share a unique fold of two orthogonal beta-sheets that encapsulate their ligands. It has been suggested that a trio of residues are the prime determinants defining the high specificity of CRBPs and CRABPs for their physiological ligands. RESULTS: Bovine/murine CRABP I and human CRABP II have been crystallized in complex with their natural ligand, all-trans-RA. Human CRABP II has also been crystallized in complex with a synthetic retinoid, 'compound 19'. Their structures have been determined and refined at resolutions of 2.9 A, 1.8 A and 2.2 A, respectively. CONCLUSIONS: The retinoid-binding site in CRABPs differs significantly from that observed in CRBP. Structural changes in three juxtaposed areas of the protein create a new, displaced binding site for RA. The carboxylate of the ligand interacts with the expected trio of residues (Arg132, Tyr134 and Arg111; CRABP II numbering). The RA ligand is almost flat with the beta-ionone ring showing a significant deviation (-33 degrees) from a cis conformation relative to the isoprene tail. The edge atoms of the beta-ionone ring are accessible to solvent in a suitable orientation for presentation to metabolizing enzymes. The bulkier synthetic retinoid causes small conformational changes in the protein structure.