A knowledge-based scoring function based on residue triplets for protein structure prediction

One of the general paradigms for ab initio protein structure prediction involves sampling the conformational space such that a large set of decoy (candidate) structures are generated and then selecting native-like conformations from those decoys using various scoring functions. In this study, based on a physical/geometric approach first suggested by Banavar and colleagues, we formulate a knowledge-based scoring function, which uses the radii of curvature formed among triplets of residues in a protein conformation. By analyzing its performance on various decoy sets, we determine a good set of parameters--the distance cutoff and the number of distance bins--to use for configuring such a function. Furthermore, we investigate the effect of using various approaches for compiling the prior distribution on the performance of the knowledge-based function. Possible extensions to the current form of the residue triplet scoring function are discussed.     

An evaluation of the performance of an automated procedure for comparative modelling of protein tertiary structure

A 3-D model of a protein can be constructed from its amino acidsequence and the 3-D structures of one or more homologues byannealing three sets of fragments: the structurally conservedregions, structurally variable regions and the side chains.The method encoded in the computer program COMPOSER was assessedby generating 3-D models of eight proteins whose crystal structuresare already known and for which 3-D structures of homologuesare available. In the structurally conserved regions, differencesbetween modelled and X-ray structures are smaller than the differencesbetween the X-ray structures of the modelled protein and thehomologues used to build the model. When several homologuesare used, the contributions of the known structures are weighted,preferably by the square of sequence similarity; this is especiallyimportant when the similarities of the homologues to the modelledstructure differ greatly. The ‘collar’ extensionapproach, in which a similar region of different length in ahomologue is used to extend the framework, can result in a moreaccurate model. If known homologues comprise more than one relatedgroup of proteins and they are both distantly related to theunknown, then alignment of the sequence to be modelled witheach group of homologues facilitates identification of structurallyconserved regions of the unknown and leads to an improved model.Models have root mean square differences (r.m.s.d.s) with thestructures defined by X-ray analysis of between 0.73 and 1.56Å for all C atoms, for seven of the eight models. Forthe model of mucor pepsin, where the closest homologue has 33%sequence identity and 20% of the residues are in structurallyvariable regions, the r.m.s.d. for the framework region is 1.71Å and the r.m.s.d. for all C atoms is 3.47 Â.     

The structure of E.coli soluble inorganic pyrophosphatase at 2.7 A resolution

Hie structure of E.coli soluble inorganic pyrophosphatase hasbeen refined at 2.7 resolution to an R-factor of 20.9. Theoverall fold of the molecule is essentially the same as yeastpyrophosphatase, except that yeast pyrophosphatase is longerat both the N- and C-termini. Escherichia coli pyrophosphataseis a mixed +ß protein with a complicated topology.The active site cavity, which is also very similar to the yeastenzyme, is formed by seven ß-strands and an -helixand has a rather asymmetric distribution of charged residues.Our structure-based alignment extends and improves upon earliersequence alignment studies; it shows that probably no more than14, not 15–17 charged and polar residues are part of theconserved enzyme mechanism of pyrophosphatases. Six of theseconserved residues, at the bottom of the active site cavity,form a tight group centred on Asp70 and probably bind the twoessential Mg⁺ ions. The others, more spreadout and more positivelycharged, presumably bind substrate. Escherichia coli pyrophosphatasehas an extra aspartate residue in the active site cavity, whichmay explain why the two enzymes bind divalent cation differently.Based on the structure, we have identified a sequence motifthat seems to occur only in soluble inorganic pyrophosphatases.     

The refined crystal structure of subtilisin Carlsberg at 2.5 ? resolution

We report here the X-ray crystal structure of native subtilisinCarlsberg, solved at 2.5 ? resolution by molecular replacementand refined by restrained least squares to a crystal-lographicresidual of 0.206. we compare this structure to the crystal structure of subtilisin BPN'. We find that, despite82 amino acid substitutions and one deletion in subtilisin Carlsbergrelative to subtilisin BPN', the structures of these enzymesare remarkably similar. We calculate an r.m.s. difference betweenequivalent a-carbon positions in subtilisin Carlsberg and subtilisinBPN' of only 0.55 ?. This confirms previous reports of extensivestructural bomology between these two subtilisins based on X-raycrystal structures of the complex of eglin-c with subtilisinCarlsberg [McPhalen,C.A., Schnebli.H.P. and James,M.N.G. (1985)FEBS Lett., 188, 55; Bode,W., Papamokos,E. and Musil,D. (1987)Eur. J. Biochem., 166, 673-692]. In addition, we find that thenative active sites of subtilisins Carlsberg and BPN' are virtuallyidentical. While conservative substitutions at residues 217and 156 may have subtle effects on the environments of substrate-bindingsites SI' and SI respectively, we find no obvious structuralcorrelate for reports that subtilisins Carlsberg and BPN' differin their recognition of model substrates. In particular, wefind no evidence that the hydrophobic binding pocket SI in subtilisinCarlsberg is ‘deeper’, ‘narrower’ or'less polar' than the corresponding binding site hi subtilisinBPN' [Karasaki and Ohno (1978) J. Biochem., Tokyo, 84, 531–538].     

The crystal structure of {alpha}-bungarotoxin at 2.5 ? resolution: relation to solution structure and binding to acetylcholine receptor

We report collection of 2.5 ? resolution X-ray diffraction datafrom newly grown crystals of the rare ‘small unit cell’form of the long snake neurotoxin, -bungarotoxin. The previousmodel of the molecule has been rebuilt, and refined using least-squaremethods to a crystallographic residual of 0.24 at 2.5 ? resolution.-Bungarotoxin's crystal structure is compared with the crystalstructures of two other snake neurotoxins (cobratoxin and erabutoxin),and with its solution structure inferred from spectroscopicstudies. Significant differences include less ß-sheetin bungarotoxm's crystal structure than in solution, or in thecrystal structures of other neurotoxins, and an unusual orientationin the crystal of the invariant tryptophan. The functional,binding surface of bungarotoxin is described; it consists primarilyof hydrophobic and hydrogen-bonding groups and only a few chargedside chains. The structure is compared with experimental bindingparameters for neurotoxins.     

A simple approach for protein structure discrimination based on the network pattern of conserved hydrophobic residues

Evolutionarily conserved hydrophobic residues at the core of protein structures are generally assumed to play a structural role in protein folding and stability. Recent studies have implicated that their importance to protein structures is uneven, with a few of them being crucial and the rest of them being secondary. In this work, we explored the possibility of employing this feature of native structures for discriminating non-native structures from native ones. First, we developed a network tool to quantitatively measure the structural contributions of individual amino acid residues. We systematically applied this method to diverse fold-type sets of native proteins. It was confirmed that this method could grasp the essential structural features of native proteins. Next, we applied it to a number of decoy sets of proteins. The results indicate that such an approach indeed identified non-native structures in most test cases. This finding should be of help for the investigation of the fundamental problem of protein structure prediction.     

Skewed distribution of protein secondary structure contents over the conformational triangle

A conformational triangle method is presented to analyze thesecondary structure contents of 1028 structurally known proteinsin the non-redundant data set of the recent 25% PDB_SELECT.The secondary structure contents of each protein are mappedon to a point in the triangle. It was found that the distributionof the 1028 points is strongly skewed in the triangle and about42% of the whole area is empty, which is called the forbiddenarea. The detailed border between the allowable and forbiddenareas was calculated. The possible explanation of the skeweddistribution is discussed. The distributions of the mappingpoints for enzymes and non-enzymes in this non-redundant dataset are compared. It was found that a necessary rather thana sufficient condition for an enzyme molecule is that its coilcontent must be 0.223. It is hoped that the skewed distributionobserved here could be used to test the secondary structureand threading predictions.     

Methods of translating NMR proton distances into their corresponding heavy atom distances for protein structure prediction with limited experimental data

This paper proposes a strategy to translate experimental 1H NMR proton distance restraints into their corresponding heavy atom distance restraints for the purpose of protein structure prediction. The relationships between interproton distances and the corresponding heavy atom distances are determined by studying well-resolved X-ray protein structures. The data from the interproton distances of amide protons, alpha-protons, beta-protons and side chain methyl protons are plotted against the corresponding heavy atoms in scatter plots and then fitted with linear equations for lower bounds, upper bounds and optimal fits. We also transform the scatter plots into two-dimensional heat maps and three-dimensional histograms, which identify the regions where data points concentrate. The common interproton distances between amide protons, alpha-protons, beta-protons in alpha-helices, anti-parallel beta-sheets and parallel beta-sheets are also tabulated. We have found several patterns emerging from the distance relationships between heavy atom pairs and their corresponding proton pairs. All our upper bound, lower bound and optimal fit results for translating the interproton distance into their corresponding heavy atom distances are tabulated.     

Predicted secondary structure and membrane topology of the scrapie prion protein

The integral membrane sialoglycoprotein PrP^Sc is the only identifiablecomponent of the scrapie prion. Scrapie in animals and Creutzfeldt-Jakobdisease in humans are transmissible, degenerative neurologicaldiseases caused by prions. Standard predictive strategies havebeen used to analyze the secondary structure of the prion proteinin conjunction with Fourier analysis of the primary sequencehydrophobicities to detect potential amphipathic regions. Severalhydrophobic segments, a proline- and glycine-rich repeat regionand putative glycosylation sites are incorporated into a modelfor the integral membrane topology of PrP. The complete aminoacid sequences of the hamster, human and mouse prion proteinsare compared and the effects of residue substitutions upon thepredicted conformation of the polypeptide chain are discussed.While PrP has a unique primary structure, its predicted secondarystructure shares some interesting features with the serum amyloidA proteins. These proteins undergo a post-translational modificationto yield amyloid A, molecules that share with PrP the abilityto polymerize into birefringent filaments. Our analyses mayexplain some experimental observations on PrP, and suggest furtherstudies on the properties of the scrapie and cellular PrP isoforms.     

Toxicity-based selection of Escherichia coli mutants for functional recombinant protein production: application to an antibody fragment

We propose a novel approach to the selection of Escherichia coli bacterial strains improved for the production of recombinant functional proteins. This approach is based on aggregation-induced toxicity of recombinant proteins. We show that selection of clones displaying a reduced toxicity is an efficient means of isolating bacteria producing recombinant protein with reduced aggregation in favour of correct folding. For an efficient selection, we found that time of toxicity induction must be precisely determined and recombinant protein must be expressed as a fusion with a protein whose activity is easily detectable on plates, thus allowing elimination of non-productive mutants. Choosing the expression to the periplasmic space of an scFv fragment fused to the N-terminus of alkaline phosphatase as a model, we selected chromosomal mutations that reduce aggregation-induced toxicity and showed that they concomitantly improve production of a functional recombinant hybrid. The effects of the mutations isolated could then be cumulated with those of other strategies used for recombinant scFv production. Thus, we could ensure a 6- to 16-fold increase in production of a functional scFv-PhoA hybrid. This is the first report demonstrating the possibility of directly selecting on agar plates E.coli strains improved for functional recombinant protein production from a large bacterial mutant library.     

Modeling the three-dimensional structure of the monocyte chemo-attractant and activating protein MCAF/MCP-1 on the basis of the solution structure of interleukin-8

A model of the three-dimensional structure of the monocyte chemo-attractantand activating protein MCAF/MCP-1 is presented. The model ispredicted based on the previously determined solution structureof interleukin-8 (IL-8/NAP-1) [Clore, G.M., Appella, E., Yamada,M., Matsushima, K. and Gronenborn, A.M. (1990) Biochemistry29, 1689–1696]. Both proteins belong to a superfamilyof cytokine proteins involved in cell-specific chemotaxis, hostdefense and the inflammatory response. The amino acid sequenceidentity between the two proteins is 24%. It is shown that theregular secondary structure elements of the parent structurecan be retained in the modeled structure, such that the backbonehydrogen bonding pattern is very similar in the two structures.The polypeptide backbone is superimposable with an atomic r.m.s.difference of 0.9 Å and all side chains can be modeledby transferring the parent side chain conformation to the newstructure. Thus, the deduced structure, like the parent one,is a dimer and consists of a six-stranded antiparallel /3-sheet,formed by two three-stranded Greek keys, one from each monomer,upon which lie two symmetry-related antiparallel a-helices,24 Å long and separated by 14 Å. All amino acidsequence changes can be accommodated within the parent polypeptideframework without major rearrangements. This is borne out bythe fact that the IL-8/NAP-1 and modeled MCAF/MCP-1 structureshave similar non-bonding energies. These results strongly suggestthat both proteins and all other members of the superfamilymost likely have the same tertiary structure. Analysis of thedistribution of the solvent-exposed residues can be interpretedin the context of the different receptors involved in mediatingthe specific responses to both proteins and suggests that thedifferent activities of the two proteins, namely neutrophil(IL-8) versus monocyte (MCAF/MCP-1) activation and chemotaxis,reside in the specific arrangements of amino acid side chainspointing outwards from and lying in the cleft between the twoexposed long a-helices.     

Comparison of systematic search and database methods for constructing segments of protein structure

Two principal methods of determining the conformation of shortpieces of polypeptide backbone in proteins have been developed:using a database of known structures and systematicallygeneratingall conformations. In this paper, we compare the effectivenessof these two techniques. The completeness of the database forsegments of different lengths is examined and it is found tocontain most conformations for segments seven residues long,but to deteriorate rapidly for longer regions. When the databasesegment is to be incorporated into the rest of a structure,at least seven residues are required to build four new residues,because of the need to positionthe segment relative to the restof the structure.It is found that such positioning using flankingresidues results in large errors in the inserted region. Weconclude that the database method is currently not effectivefor comparative modeling, even for short segments. The systematicsearchprocedure is found to generate almost all structures of shortsegments found in proteinsIn contrast to the database method,low root mean square error structures are obtained for a setof trial segments embedded in the rest of a protein structure.Thus, it should be considered the method of choice.     

An intein-based genetic selection allows the construction of a high-quality library of binary patterned de novo protein sequences

Combinatorial libraries of synthetic DNA are increasingly being used to identify and evolve proteins with novel folds and functions. An effective strategy for maximizing the diversity of these libraries relies on the assembly of large genes from smaller fragments of synthetic DNA. To optimize library assembly and screening, it is desirable to remove from the synthetic libraries any sequences that contain unintended frameshifts or stop codons. Although genetic selection systems can be used to accomplish this task, the tendency of individual segments to yield misfolded or aggregated products can decrease the effectiveness of these selections. Furthermore, individual protein domains may misfold when removed from their native context. We report the development and characterization of an in vivo system to preselect sequences that encode uninterrupted gene segments regardless of the foldedness of the encoded polypeptide. In this system, the inserted synthetic gene segment is separated from an intein/thymidylate synthase (TS) reporter domain by a polyasparagine linker, thereby permitting the TS reporter to fold and function independently of the folding and function of the segment-encoded polypeptide. TS-deficient Escherichia coli host cells survive on selective medium only if the insert is uninterrupted and in-frame, thereby allowing selection and amplification of desired sequences. We demonstrate that this system can be used as a highly effective preselection tool for the production of large, diverse and high-quality libraries of de novo protein sequences.     

Prediction of protein secondary structure based on physical theory. Histones

Secondary structures of histones H1, H2A, H2B, H3, H4 and H5have been calculated by the computer program ALB based on amolecular theory of protein secondary structure. The predictedsecondary structures of all histones are predominantly -helical.The calculated secondary structure of linker histones H1 andH5 is close to that previously obtained from two-dimensionalNMR data. For each of the core histones (H2A, H2B, H3, H4) onelong -helix and several short ones have been predicted. Theselong helices can be identified with rods in the low-resolutionelectron density map.     

When awaiting 'Bio' Champollion: dynamic programming regularization of the protein secondary structure predictions

Predictions of protein secondary structure using current methodsare often unrealistic, i.e. the predicted -helices or ß-strandsare too short. To improve the realism, various heuristic ‘filtering’or ‘smoothing’ methods are used. They are more orless intuitive and are based on ad hoc corrections. We presenta regularization method to obtain a realistic secondary structurefrom predicted propensities. It is based on the known dynamicprogramming algorithm and is quite objective. It can be usedwith any prediction method which yields propensities. The regularizedpredictions conserve well the overall prediction accuracy andimprove the ‘protein-likeness’ of the prediction.     

Hydrophobic regions on protein surfaces: definition based on hydration shell structure and a quick method for their computation

The hydrophobic part of the solvent-accessible surface of atypical monomeric globular protein consists of a single, largeinterconnected region formed from faces of apolar atoms andconstituting –60% of the solvent-accessible surface area.Therefore, the direct delineation of the hydrophobic surfacepatches on an atom-wise basis is impossible. Experimental dataindicate that, in a two-state hydration model, a protein canbe considered to be unified with its first hydration shell inits interaction with bulk water. We show that, if the surfacearea occupied by water molecules bound at polar protein atomsas generated by AUTOSOL is removed, only about two-thirds ofthe hydrophobic part of the protein surface remains accessibleto bulk solvent. Moreover, the organization of the hydrophobicpart of the solvent-accessible surface experiences a drasticchange, such that the single interconnected hydrophobic regiondisintegrates into many smaller patches, i.e. the physical definitionof a hydrophobic surface region as unoccupied by first hydrationshell water molecules can distinguish between hydrophobic surfaceclusters and small interconnecting channels. It is these remaininghydrophobic surface pieces that probably play an important rolein intraand intermolecular recognition processes such as ligandbinding, protein folding and protein–protein associationin solution conditions. These observations have led to the developmentof an accurate and quick analytical technique for the automaticdetermination of hydrophobic surface patches of proteins. Thistechnique is not aggravated by the limiting assumptions of themethods for generating explicit water hydration positions. Formationof the hydrophobic surface regions owing to the structure ofthe first hydration shell can be computationally simulated bya small radial increment in solvent-accessible polar atoms,followed by calculation of the remaining exposed hydrophobicpatches. We demonstrate that a radial increase of 0.35–0.50Å resembles the effect of tightly bound water on the organizationof the hydrophobic part of the solvent-accessible surface.     

The effect of cavity-filling mutations on the thermostability of Bacillus stearothermophilus neutral protease

Cavities in the hydrophobic core of the neutral protease ofBacillus stearothermophilus were analyzed using a threedimensionalmodel that was inferred from the crystal structure of thermolysin,the highly homologous neutral protease of B.thermoproteolyticus(85% sequence identity). Site–directed mutagenesis wasused to fill some of these cavities, thereby improving hydrophobicpacking in the protein interior. The mutations had small effectson the thermostability, even after drastic changes, such asLeu284Trp and Met168Trp. The effects on T50, the temperatureat which 50% of the enzyme is irreversibly inactivated in 30min, ranged from 0.0 to +0.4°C. These results can be explainedby assuming that the mutations have positive and negative structuraleffects of approximately the same magnitude. Alternatively,it could be envisaged that the local unfolding steps, whichrender the enzyme susceptible towards autolysis and which arerate limiting in the process of thermal inactivation, are onlyslightly affected by alterations in the hydrophobic core.     

Conformational properties of the guanine-binding site of ribonuclease T1 inferred from the X-ray structure and protein engineering

Recognition by ribonuclease T₁ of guanine bases via multidentatehydrogen bonding and stacking interactions appears to be mediatedmainly by a short peptide segment formed by one stretch of aheptapeptide, Tyr⁴²-Asn⁴³-Asn⁴⁴-Tyr⁴⁵-Gly⁴⁶-Gly⁴⁷-Phe⁴⁸. Thesegment displays a unique folding of the polypeptide chain—consistingof a reverse turn, Asn⁴⁴-Tyr⁴⁵-Glu⁴⁶-Gly⁴⁷, stabilized by ahydrogen-bond network involving the side chain of Asn⁴⁴, themain-chain atoms of Asn⁴⁴, Gly⁴⁷ and Phe⁴⁸ and one water molecule.The segment is connected to the C terminus of a ß-strandand expands into a loop region between Asn⁴³ and Ser⁵⁴. Lowvalues for the crystallographic thermal parameters of the segmentindicate that the structure has a rigidity comparable to thatof a ß-pleated sheet. Replacement of Asn⁴⁴ with alanineleads to a far lower enzymatic activity and demonstrates thatthe side chain of Asn⁴⁴ plays a key role in polypeptide foldingin addition to a role in maintaining the segment structure.Substitution of Asn⁴³ by alanine to remove a weak hydrogen bondto the guanine base destabilized the transition state of thecomplex by 6.3 kJ/mol at 37°C. In contrast, mutation ofGlu⁴⁶ to alanine to remove a strong hydrogen bond to the guaninebase caused a destabilization of the complex by 14.0 kJ/mol.A double-mutant enzyme with substitutions of Asn⁴³ by a histidineand Asn⁴⁴ by an aspartic acid, to reproduce the natural substitutionsfound in ribonuclease M_s, showed an activity and base specificitysimilar to that of the wild-type ribonuclease M_s. The segmenttherefore appears to be well conserved in several fungal ribonucleases.