Structural trees for protein superfamilies

ROC and confusion analysis of structure comparison methods identify the main causes of divergence from manual protein classification

Background

Current classification of protein folds are based, ultimately, on visual inspection of similarities. Previous attempts to use computerized structure comparison methods show only partial agreement with curated databases, but have failed to provide detailed statistical and structural analysis of the causes of these divergences.

Results

We construct a map of similarities/dissimilarities among manually defined protein folds, using a score cutoff value determined by means of the Receiver Operating Characteristics curve. It identifies folds which appear to overlap or to be "confused" with each other by two distinct similarity measures. It also identifies folds which appear inhomogeneous in that they contain apparently dissimilar domains, as measured by both similarity measures. At a low (1%) false positive rate, 25 to 38% of domain pairs in the same SCOP folds do not appear similar. Our results suggest either that some of these folds are defined using criteria other than purely structural consideration or that the similarity measures used do not recognize some relevant aspects of structural similarity in certain cases. Specifically, variations of the "common core" of some folds are severe enough to defeat attempts to automatically detect structural similarity and/or to lead to false detection of similarity between domains in distinct folds. Structures in some folds vary greatly in size because they contain varying numbers of a repeating unit, while similarity scores are quite sensitive to size differences. Structures in different folds may contain similar substructures, which produce false positives. Finally, the common core within a structure may be too small relative to the entire structure, to be recognized as the basis of similarity to another.

Conclusion

A detailed analysis of the entire available protein fold space by two automated similarity methods reveals the extent and the nature of the divergence between the automatically determined similarity/dissimilarity and the manual fold type classifications. Some of the observed divergences can probably be addressed with better structure comparison methods and better automatic, intelligent classification procedures. Others may be intrinsic to the problem, suggesting a continuous rather than discrete protein fold space.

Comparison of sequence and structure-based datasets for nonredundant structural data mining

Structural data mining studies attempt to deduce general principles of protein structure from solved structures deposited in the protein data bank (PDB). The entire database is unsuitable for such studies because it is not representative of the ensemble of protein folds. Given that novel folds continue to be unearthed, some folds are currently unrepresented in the PDB while other folds are overrepresented. Overrepresentation can easily be avoided by filtering the dataset. PDB_SELECT is a well-used representative subset of the PDB that has been deduced by sequence comparison. Specifically, structures with sequences that exhibit a pairwise sequence identity above a threshold value are weeded from the dataset. Although length criteria for pairwise alignments have a structural basis, this automated method of pruning is essentially sequence-based and runs into problems in the twilight zone, possibly resulting in some folds being overrepresented. The value-added structure databases SCOP and CATH are also a potential source of a nonredundant dataset. Here we compare the sequence-derived dataset PDB_SELECT with the structural databases SCOP (Structural Classification Of Proteins) and CATH (Class-Architecture-Topology-Homology). We show that some folds remain overrepresented in the PDB_SELECT dataset while other folds are not represented at all. However, SCOP and CATH also have their own problems such as the labor-intensiveness of the update process and the problem of determining whether all folds are equally or sufficiently distant. We discuss areas where further work is required.

'Protein Peeling': an approach for splitting a 3D protein structure into compact fragments

MOTIVATION

The object of this study is to propose a new method to identify small compact units that compose protein three-dimensional structures. These fragments, called 'protein units (PU)', are a new level of description to well understand and analyze the organization of protein structures. The method only works from the contact probability matrix, i.e. the inter Calpha-distances translated into probabilities. It uses the principle of conventional hierarchical clustering, leading to a series of nested partitions of the 3D structure. Every step aims at dividing optimally a unit into 2 or 3 subunits according to a criterion called 'partition index' assessing the structural independence of the subunits newly defined. Moreover, an entropy-derived squared correlation R is used for assessing globally the protein structure dissection. The method is compared to other splitting algorithms and shows relevant performance.

AVAILABILITY

An Internet server with dedicated tools is available at http://www.ebgm.jussieu.fr/~gelly/

The βαβαβ elementary supersecondary structure of the Rossmann fold from porcine lactate dehydrogenase exhibits characteristics of a molten globule

Protein classifications show that the Rossmann fold, which consists of two betaalphabetaalphabeta motifs (BABAB) related by a rough twofold axis, is the most populated alphabeta fold, and that the betaalphabeta submotif (BAB) is a widespread elementary structural arrangement. Herein, we report MD simulations, circular dichroism and NMR analyses on BAB and BABAB from porcine lactate dehydrogenase to evaluate their intrinsic stability. Our results demonstrate that BAB is not stable in solution and is not a folding nucleus. We also find that BABAB, despite its appearance of a functional and structural unit, is not an independent and thermodynamically stable folding unit. Rather, we show that BABAB retains most native secondary structure but very little tertiary structure, thus displaying characteristics of a molten globule.

Detecting local structural similarity in proteins by maximizing number of equivalent residues

A new algorithm for superimposing protein structures based on maximizing the number of spatially equivalent residues is introduced. The algorithm works in three distinct steps. First, the optimal residue map is calculated by structural alignment. By default, the double dynamic programming algorithm, as implemented in the program ASH, was used for the structure alignment step, but we also present results based on alignments imported from three other programs (Dali, CE, and VAST).Second, the structures are spatially superimposed such that the effective number of equivalent residues (NER)--aligned residue pairs that can be spatially overlapped--is maximized. The NER score is an analytic, differentiable similarity function that rewards spatially equivalent residues but ignores non-equivalent ones. Maximization of the NER score results in accurate superpositions in cases where root mean square deviation (RMSD) minimization fails. Third, the NER function is used in conjunction with traditional dynamic programming to realign the structures based on the proximity of residues in the superposition. Results are presented for a wide range of superposition problems and compared to results from Dali, CE, and VAST. In addition, several structure-structure pairs that show only partial similarity are discussed, and results are compared to those from the LGA, SARF2, and ThreeCa programs.

Clustering and diversity of fluctuations for proteins

BACKGROUND

Protein topology plays a key role in various types of interactions. Topological constraints of a protein are defined by a contact map. We studied the fluctuations of proteins with use of a new approach based on contact map.

METHODS

An annealing algorithm is used to generate a 3-dimensional protein structure from the contact map. First, we study the properties of structural elements based on fluctuations by adding individual structures (domains or subdomains). Thereafter, we focus on the building block of proteins in terms of fluctuations.

RESULTS

To verify our hypothesis, we analyzed the pattern of fluctuations for chymotrypsin inhibitor-2 (CI2) by unstructuring (melting) of subregions. The data show different patterns of fluctuations for the unstructured CI2 relative to that calculated for the intact protein.

CONCLUSION

Our approach introduces a new concept for classifying building blocks of proteins based on thermal fluctuations.

A global representation of the protein fold space

One of the principal goals of the structural genomics initiative is to identify the total repertoire of protein folds and obtain a global view of the "protein structure universe." Here, we present a 3D map of the protein fold space in which structurally related folds are represented by spatially adjacent points. Such a representation reveals a high-level organization of the fold space that is intuitively interpretable. The shape of the fold space and the overall distribution of the folds are defined by three dominant trends: secondary structure class, chain topology, and protein domain size. Random coil-like structures of small proteins and peptides are mapped to a region where the three trends converge, offering an interesting perspective on both the demography of fold space and the evolution of protein structures.

Ring and zipper formation is the key to understanding the structural variety in all-beta proteins

A novel structural classification of beta proteins is presented from the viewpoint of the ring-shaped structure and the zipper-like contact pattern, based on the fact that 92% and 60% of beta proteins have the ring topology and the zippered contact pattern, respectively. We discuss the implication of the unexpectedly high preference for the ring and zippered structures in connection with the folding process of beta proteins.

Conserved key amino acid positions (CKAAPs) derived from the analysis of common substructures in proteins

An all-against-all protein structure comparison using the Combinatorial Extension (CE) algorithm applied to a representative set of PDB structures revealed a gallery of common substructures in proteins (http://cl.sdsc.edu/ce.html). These substructures represent commonly identified folds, domains, or components thereof. Most of the subsequences forming these similar substructures have no significant sequence similarity. We present a method to identify conserved amino acid positions and residue-dependent property clusters within these subsequences starting with structure alignments. Each of the subsequences is aligned to its homologues in SWALL, a nonredundant protein sequence database. The most similar sequences are purged into a common frequency matrix, and weighted homologues of each one of the subsequences are used in scoring for conserved key amino acid positions (CKAAPs). We have set the top 20% of the high-scoring positions in each substructure to be CKAAPs. It is hypothesized that CKAAPs may be responsible for the common folding patterns in either a local or global view of the protein-folding pathway. Where a significant number of structures exist, CKAAPs have also been identified in structure alignments of complete polypeptide chains from the same protein family or superfamily. Evidence to support the presence of CKAAPs comes from other computational approaches and experimental studies of mutation and protein-folding experiments, notably the Paracelsus challenge. Finally, the structural environment of CKAAPs versus non-CKAAPs is examined for solvent accessibility, hydrogen bonding, and secondary structure. The identification of CKAAPs has important implications for protein engineering, fold recognition, modeling, and structure prediction studies and is dependent on the availability of structures and an accurate structure alignment methodology. Proteins 2001;42:148-163.

A comprehensive analysis of the Greek key motifs in protein beta-barrels and beta-sandwiches

The Greek key motifs are the topological signature of many beta-barrels and a majority of beta-sandwich structures. An updated survey of these structures integrates many early observations and newly emerging patterns and provides a better understanding of the unique role of Greek keys in protein structures. A stereotypical Greek key beta-barrel accommodates five or six strands and can have 12 possible topologies. All except one six-stranded topologies have been observed, and only one five-stranded topologies have been seen in actual structures. Of the representative beta-barrel structures analyzed here, half have left-handed Greek keys. This result challenges the empirical claim of the handedness regularity of Greek keys in beta-barrels. One of the five-stranded topologies that has not been observed in beta-barrels comprises two overlapping Greek keys. The two three-dimensional forms of this topology constitute a structural unit that is present in a vast majority of known beta-sandwich structures. Using this unit as the root, we have built a new taxonomy tree for the beta-sandwich folds and deduced a set of rules that appear to constrain how other beta-strands adjoin the unit to form a larger double-layered structure. These rules, though derived from a larger data set, are essentially the same as those drawn from earlier studies, suggesting that they may reflect the true topological constraints in the design of beta-sandwich structures. Finally, a novel variant of the Greek key motif (defined here as the twisted Greek key) has emerged which introduces loop crossings into the folded structures. Proteins 2000;40:409-419.

Analysis and prediction of inter-strand packing distances between beta-sheets of globular proteins

Any two beta-strands belonging to two different beta-sheets in a protein structure are considered to pack interactively if each beta-strand has at least one residue that undergoes a loss of one tenth or more of its solvent contact surface area upon packing. A data set of protein 3-D structures (determined at 2.5 A resolution or better), corresponding to 428 protein chains, contains 1986 non-identical pairs of beta-strands involved in interactive packing. The inter-axial distance between these is significantly correlated to the weighted sum of the volumes of the interacting residues at the packing interface. This correlation can be used to predict the changes in the inter-sheet distances in equivalent beta-sheets in homologous proteins and, therefore, is of value in comparative modelling of proteins.

Novel protein structural motifs containing two-turn and longer 3(10)-helices

The 3(10)-helix constitutes a small but significant fraction of secondary structural elements in proteins. Protein data base surveys have shown these helices to be present as alpha-helical extensions, in loops and as connectors between beta-strands. The present work focuses on two-turn and longer 3(10)-helices where we establish that two-turn and longer 3(10) helices, unlike the more abundant single-turn 3(10)-helices, frequently occur independent of any other contiguous secondary structural elements. More importantly, a large fraction of these independent two-turn and longer 3(10)-helices, along with alpha-helices and beta-strands, are found to form novel super-secondary structural motifs in several proteins with possible implications for protein folding, local conformational relaxation and biological functions.

Polymerization of IgA and IgM: Roles of Cys309/Cys414 and the Secretory Tailpiece

We have investigated how the secretory tailpiece (tp), Cys414 and the amino acids flanking Cys414 or Cys309 are involved in regulating the different polymerization of IgM and IgA to pentamers and dimers/monomers, respectively. Whereas changing the tp of IgM to that of IgA has little effect on IgM polymerization, introducing the μtp to IgA leads to the formation of larger than wild-type IgA polymers, including pentamers and hexamer. This shows that the secretory tp can differentially regulate polymerization depending on the heavy chain context. Cys414, which is engaged in intermonomeric disulfide bonds in IgM, is not crucial for the difference in IgM and IgA polymerization; IgM with a C414S mutation forms more large polymers than IgA. Also, IgA with IgM-like mutations in the five amino acids flanking Cys309, which is homologous to Cys414, oligomerize similarly as IgA wild type. Thus, IgA appears to have an inherent tendency to form monomers and dimers that is partially regulated by the tp, while the Cys309 region has only a minor effect. We also show that complement activation by IgM is sensitive to alterations in the polymeric structure, while IgA is inactive in classical complement activation even for polymers such as pentamers and hexamers.

A structural tree for proteins containing S-like beta-sheets

A structural tree for proteins and domains containing S-like beta-sheets has been constructed. An S-like beta-sheet is taken as a starting structure in modelling or as a root structure of the tree. Larger structures are obtained by a stepwise addition of beta-strands and/or alpha-helices to the root S-like beta-sheet in accordance with a restricted set of rules inferred from known principles of protein structure. Applications of the structural tree to structure comparison, protein classification and protein folding are described.

Cation binding to the integrin CD11b I domain and activation model assessment

BACKGROUND

The integrin family of cell-surface receptors mediate cell adhesion through interactions with the extracellular matrix or other cell-surface receptors. The alpha chain of some integrin heterodimers includes an inserted 'I domain' of about 200 amino acids which binds divalent metal ions and is essential for integrin function. Lee et al. proposed that the I domain of the integrin CD11b adopts a unique 'active' conformation when bound to its counter receptor. In addition, they proposed that the lack of adhesion in the presence of Ca2+ ion reflected the stabilization of an 'inactive' I-domain conformation. We set out to independently determine the structure of the CD11 b I domain and to evaluate the structural effects of divalent ion binding to this protein.

RESULTS

We have determined the X-ray structure of a new crystal form of the CD11 b I domain in the absence of added metal ions by multiple isomorphous replacement (MIR). Metal ions were easily introduced into this crystal form allowing the straight-forward assessment of the structural effects of divalent cation binding at the metal ion dependent adhesion site (MIDAS). The equilibrium binding constants for these ions were determined by titration calorimetry. The overall protein conformation and metal-ion coordination of the I domain is the same as that observed for all previously reported CD11 a I-domain structures and a CD11 b I-domain complex with Mn2+. These structures define a majority conformation.

CONCLUSIONS

Addition of the cations Mg2+, Mn2+ and Cd2+ to the metal-free I domain does not induce conformational changes in the crystalline environment. Moreover, we find that Ca2+ binds poorly to the I domain which serves to explain its failure to support adhesion. We show that the active conformation proposed by Lee et al, is likely to be a construct artifact and we propose that the currently available data do not support a dramatic structural transition for the I domain during counter-receptor binding.