Enhanced dead-end elimination in the search for the global minimum energy conformation of a collection of protein side chains

Although the conformational states of protein side chains canbe described using a library of rotamers, the determinationof the global minimum energy conformation (GMEC) of a largecollection of side chains, given fixed backbone coordinates,represents a challenging combinatorial problem with importantapplications in the field of homology modelling. Recently, wehave developed a theoretical framework, called the dead-endelimination method, which allows us to identify efficientlyrotamers that cannot be members of the GMEC. Such dead-endingrotamers can be iteratively removed from the system under studythereby tracking down the size of the combinatorial problem.Here we present new developments to the dead-end eliminationmethod that allow us to handle larger proteins and more extensiverotamer libraries. These developments encompass (i) a procedureto determine weight factors in the generalized dead-end eliminationtheorem thereby enhancing the elimination of dead-ending rotamersand (ii) a novel strategy, mainly based on logical argumentsderived from the logic pairs theorem, to use dead-ending rotamerpairs in the efficient elimination of single rotamers. Thesedevelopments are illustrated for proteins of various sizes andthe flow of the current method is discussed in detail. The effectivenessof dead-end elimination is increased by two orders of magnitudeas compared with previous work. In addition, it now becomesfeasible to use extremely detailed libraries. We also providean appendix in which the validity of the generalized dead-endcriterion is shown. Finally, perspectives for further applicationswhich may now become within reach are discussed.     

Studies of the pore-forming domain of a voltage-gated potassium channel protein

Recent mutagenesis studies nave identified a stretch of aminoacid residues which form the ion-selective pore of the voltage-gatedpotassium channel. It has been suggested that this sequenceof amino acids forms a ß-barrel structure making upthe structure of the ion-selective pore [Hartman,H.A., Kirsch,G.E.,DreweJ.A., Taglialatela.M., Joho.R.H. and Brown,A.M. (1991)Science, 251, 942–944; YeUen.G., Jurman,M.E., Abramson,T.and MacKinnon,R. (1991) Science, 251, 939–942; Yool,AJ.and Schwarz.T.L. (1991) Nature, 349, 700–704]. We havesynthesized a polypeptide corresponding to this amino add sequence(residues 431–449 of the ShA potassium channel from Drosophila).A tetrameric version of this sequence was also synthesized byUnking together four of these peptldes onto a branching lysinecore. Fourier transform infrared (FT-LR) and circular dichroism(CD) spectroscopy have been used to investigate the structureof these peptides after their reconstitution into lyso phos-phatidylcholinemicelles and lipid bilayers composed of dimyristoyl phosphatidyfcholineand dimyristoyl phosphatidyl-glycerol. The spectroscopic studiesshow that these peptides are predominantly a-helical in theselipid environments. When Incorporated into planar lipid bilayersboth peptides induce ion channel activity. Molecular modellingstudies based upon the propensity of these peptides to forman -helical secondary structure in a hydrophobfc environmentare described. These results are discussed in the light of recentmutagenesis and binding studies of the Drosophila Shaker potassiumion channel protein     

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.     

Increasing the thermostability of D-xylose isomerase by introduction of a proline into the turn of a random coil

Thermostability can be increased by introducing prolines atsuitable sites in target proteins. Two single (G138P, G247D)mutants and one double (G138P/G247D) mutant of xylose isomerasefrom Streptomyces diastaticus No.7, strain M1033 have been constructedby site-directed mutagenesis. With respect to the wild-typeenzyme, G138P showed about a 100% increase in thermostability,and G247D showed an increased catalytic activity. Significantly,the double mutant, G138P/G247D displayed even higher activitythan G247D and better heat stability than G138P. Its half lifewas about 2.5-fold greater than the wild-type enzyme, usingxylose as a substrate. Molecular modelling suggested that theintroduction of a proline residue in the turn of a random coilmay cause the surrounding conformation to be tightened by reducingthe backbone flexibility. The change in thermostability can,therefore, be explained based on changes in the molecular rigidity.Furthermore, the improvements in the properties of the doublemutant indicated that the advantages of two single mutants canbe combined effectively.     

Comparison of conservation within and between the Ser/Thr and Tyr protein kinase family: proposed model for the catalytic domain of the epidermal growth factor receptor

The protein kinase family can be subdivided into two main groupsbased on their ability to phosphorylate Ser/Thr or Tyr substrates.In order to understand the basis of this functional difference,we have carried out a comparative analysis of sequence conservationwithin and between the Ser/Thr and Tyr protein kinases. A multiplesequence alignment of 86 protein kinase sequences was generated.For each position in the alignment we have computed the conservationof residue type in the Ser/Thr, in the Tyr and in both of thekinase subfamilies. To understand the structural and/or functionalbasis for the conservation, we have mapped these conservationproperties onto the backbone of the recently determined structureof the cAMP–dependent Ser/Thr kinase. The results showthat the kinase structure can be roughly segregated, based uponconservation, into three zones. The inner zone contains residueshighly conserved in all the kinase family and describes thehydrophobic core of the enzyme together with residues essentialfor substrate and ATP binding and catalysis. The outer zonecontains residues highly variable in all kinases and representsthe solvent–exposed surface of the protein. The thirdzone is comprised of residues conserved in either the Ser/Thror Tyr kinases or in both, but which are not conserved betweenthem. These are sandwiched between the hydrophobic core andthe solvent-exposed surface. In addition to analyzing overallconservation hi the kinase family, we have also looked at conservationof its substrate and ATP binding sites. The ATP site is highlyconserved throughout the kinases, whereas the substrate bindingsite is more variable. The active site contains several positionswhich differ between the Ser/Thr and Tyr kinases and may beresponsible for discriminating between hydroxyl bearing sidechains. Using this information we propose a model for Tyr substratebinding to the catalytic domain of the epidermal growth factorreceptor (EGFR).     

Mutational analysis supports a structural model for the cell cycle protein kinase p34

Structural models for the eukaryotic cell cycle control proteinp34 from human, S.pombe and S.cerevisiae have been derived fromthe crystallographic coordinates of the cAMP-dependent proteinkinase (cAPK) catalytic subunit (active conformation) and comparedwith the structure of Inactive CDK2 apoenzyme. Differences betweenthe p34 and cAPK catalytic sites provide a possible explanationfor their different substrate specificities. The p34 modelslocalize Tyrl5 and Thrl4 close to the sites of catalysis andsubstrate recognition where their phosphorylatlon could inhibitp34 kinase activity either by blocking MgATP or substrate binding.The conserved sequences PSTAIRE and LYLIFEFL are both closeto the catalytic site and accessible on the protein surfaceavailable to mediate interactions with other proteins. It ispredicted that p34 has an active-site cleft composed almostentirely of sequences common to all protein kinases and sequencesunique to the p34 protein family. Genetic and biochemical analysesof p34 have shown that it interacts extensively with a numberof other proteins. The model allows the relative dispositionof these sites of mutation to each other and to the sites ofcatalysis and substrate recognition to be appreciated. Surfaceregions on p34 that are important for function have been identified.These sites identify residues that may interact with p13^SUCL,cydin, plO7^WEEL and p80^cdc25     

Comparative modelling and analysis of amino acid substitutions suggests that the family of pregnancy-associated glycoproteins includes both active and inactive aspartic proteinases

The pregnancy-associated glycoproteins (PAGs) are secretoryproducts synthesized by the outer epithelial cell layer (chorion)of the placentas of various ungulate species. The amino acidsequences of eight PAGs have been inferred from cloned cDNAof cattle and sheep, as well as of the non-ruminant pig andhorse. We compare the PAG sequences and present results of thethree-dimensional models of boPAG-1 and ovPAG-1 that were constructedon the basis of the crystal structures of homologous porcinepepsin and bovine chymosin using a rule-based comparative modellingapproach. Further, we compare peptide binding subsites definedby interactions with pepstatin and a decapeptide inhibitor (CH-66)modelled on the basis of crystal structures of other asparticproteinases. We have extended our analysis of the peptide bindingsubsites to the other PAG molecules of known sequence by aligningthe PAG sequences to the structural template derived from thepepsin family and by making use of the three-dimensional modelsof the boPAG-1 and ovPAG-1. The residues that are likely toaffect peptide binding in the boPAG-1, ovPAG-1 and other PAGmolecules have been identified. Sequence comparisons revealthat all PAG molecules may have evolved from a pepsin-like progenitormolecule with the equine PAG most closely related to the pepsins.The presence of substitutions at the S₁ and other subsites relativeto pepsin make it unlikely that either bovine, ovine or theporcine PAG-1 have catalytic activity. Only two of the eightPAGs examined (porcine PAG-2 and equine PAG-1) retain featuresof active aspartic proteinases with pepsin-like activity. Ourresults indicate that in the PAGs so far characterized the peptidebinding specificities differ significantly from each other andfrom pepsin, despite their high sequence identities. Analysisof the various peptide binding subsites demonstrates why bothbovine and ovine PAG-1 are capable of binding pepstatin. Thestrong negative charge in the binding cleft of boPAG-1 and ovPAG-1indicates a preference for lysine- or arginine-rich peptides.PAGs represent a family where the possible peptide binding functionmay be retained through their binding specificities, but wherethe catalytic activity may be lost in some cases, such as theboPAG-1, ovPAG-1 and the poPAG-1.     

Efficient generation of a reshaped human mAb specific for the {alpha} toxin of Clostridium perfringens

We have used the technique of antibody reshaping to producea humanized antibody specific for the a toxin of Clostridiumperfringens. The starting antibody was from a mouse hybridomafrom which variable (V) region nucleo-tide sequences were determined.The complementarity-determining regions (CDRs) from these Vregions were then inserted into human heavy and light chainV region genes with human constant region gene fragments subsequentlyadded. The insertion of CDRs alone into human frameworks didnot produce a functional reshaped antibody and modificationsto the V region framework were required. With minor frameworkmodifications, the affinity of the original murine mAb was restoredand even exceeded. Where affinity was increased, an alteredbinding profile to overlapping peptides was observed. Computermodelling of the reshaped heavy chain V regions suggested thatamino acids adjacent to CDRs can either contribute to, or distort,CDR loop conformation and must be adjusted to achieve high bindingaffinity.     

Large-scale modelling as a route to multiple surface comparisons of the CCP module family

Numerous mammalian proteins are constructed from a limited repertoire of module-types. Proteins belonging to the regulators of complement activation family--crucial for ensuring a complement-mediated immune response is targeted against infectious agents--are composed solely of complement control protein (CCP) modules. In the current study, CCP module sequences were grouped to allow selection of the most appropriate experimentally determined structures to serve as templates in an automated large-scale structure modelling procedure. The resulting 135 individual CCP module models, valuable in their own right, are available at the online database http://www.bru.ed.ac.uk/~dinesh/ccp-db.html. Comparisons of surface properties within a particular family of modules should be more informative than sequence alignments alone. A comparison of surface electrostatic features was undertaken for the first 28 CCP modules of complement receptor type 1 (CR1). Assignments to clusters based on surface properties differ from assignments to clusters based on sequences. This observation might reflect adaptive evolution of surface-exposed residues involved in protein-protein interactions. This illustrative example of a multiple surface-comparison was indeed able to pinpoint functional sites in CR1.     

Modelling of the lignin peroxidase LIII of Phlebia radiata: use of a sequence template generated from a 3-D structure

A model of the lignin peroxidase LIII of Phlebia radiata wasconstructed on the basis of the structure of cytochrome c peroxidase(CCP). Because of the low percentage of amino acid identitybetween the CCP and the lignin peroxidase LIII of Phlebia radiata,alignment of the sequences was based on the generation of atemplate from a knowledge of the 3-D structure of CCP and consensussequences of lignin peroxidases. This approach gave an alignmentin which all the insertions in the lignin peroxidase were placedat loop regions of CCP, with a 21.1% identity for these twoproteins. The model was constructed using this alignment andthe computer program COMPOSER, which assembles the model asa series of rigid fragments derived from CCP and other proteins.Manual intervention was required for some of the longer loopregions. The -helices forming the structural framework, andespecially the haem environment of CCP, are conserved in theLIII model and the core is close packed without holes. A possiblesite of the substrate oxidation at the haem edge of LIII isdiscussed.     

Contribution of the C-terminal amino acid to the stability of Bacillus subtilis neutral protease

The role of the C-terminal Leu300 in maintaining thermal stabilityof the neutral protease of Bacillus subtilis was investigated.From model building studies based on the three dimensional structureof thermolysin, the neutral protease of B.thermoproteolyticus,it was conduded that this residue is located in a hydrophobicpocket composed of residues located in the C-terminal and themiddle domain. To test the hypothesis that Leu300, by contributingto a stabilizing interaction between these domains, is importantfor enzyme stability, several neutral protease mutants wereconstructed and characterized. The thermostability of the enzymewas lowered by deleting Leu300 or by replacing this residueby a smaller (Ala), a polar (Asn) or a sterically unfavourable(He) amino acid. Thermostabiity was increased upon replacingLeu300 by Phe. These results are in agreement with model-buildingstudies. The effects on thermostability observed after mutatingthe corresponding Val318 in the thermostable neutral proteaseof B.stearothermophilus were less pronounced.     

Influenza virus M2 protein: a molecular modelling study of the ion channel

The influenza A M₂ protein forms cation-selective ion channelswhich are blocked by the anti-influenza drug amantadine. A molecularmodel of the M₂ channel is presented in which a bundle of fourparallel M₂ transbilayer helices surrounds a central ion-permeablepore. Analysis of helix amphipathicity was used to aid determinationof the orientation of the helices about their long axes. Thehelices are tilted such that the N-terminal mouth of the poreis wider than the C-terminal mouth. The channel is lined byresidues V27, S31 and I42. Residues D24 and D44 are locatedat opposite mouths of the pore, which is narrowest in the vicinityof I42. Energy profiles for interaction of the channel withNa⁺, amantadine-H⁺ and cyclopentylamine-H⁺ are evaluated. Theinteraction profile for Na⁺ exhibits three minima, one at eachmouth of the pore, and one in the region of residue S31. Theamantadine-H⁺ profile exhibits a minimum close to S31 and abarrier near residue I42. This provides a molecular model foramantadine-H⁺ block of M₂ channels. The profile for cyclopentylamine-H⁺does not exhibit such a barrier. It is predicted that cyclopentyl-amine-H⁺will not act as an M₂ channel blocker.     

Replacing the glutamate ligand in the structural zinc site of Sulfolobus solfataricus alcohol dehydrogenase with a cysteine decreases thermostability

The alcohol dehydrogenase gene from the thermophilic archaeumSulfolobus solfataricus has been subcloned and expressed inEscherichia coli under the control of the T7 inducible promoter.Therecombinant protein shows properties analogous to those of thenative enzyme, including thermostability, despite the fact thatE.coli does not post-translationally modify two lysine residueswhich are N--methylated in the native enzyme. We constructeda 3-D model of the S.solfataricus alcohol dehydrogenase usingthe known structure of its isozyme from horse liver as a template.Our analysis of the structural zinc binding site suggested thatthis site is present andfunctional in the S.solfataricus enzymeand that a glutamate ligand can contribute to thermostabilityby influencing electrostatic interactions around the metal centre.To investigate thishypothesis, we constructed, expressed andcharacterized a mutant where the glutamate is replaced by acysteine, thus restoring the zinc binding site of mesophilicalcohol dehydrogenases. Themutant shows the same activity buta reduced thermostability with respect to the wild-type recombinantprotein, as suggested by our model.     

Binding of the globular domain of linker histones H5/H1 to the nucleosome: a hypothesis

The amino acid sequence of the central globular domain of histoneH1/H5 family members is highly homologous. Twenty-four suchsequences have been compared to establish the conserved andvariable residues. Fitting this to the tertiary structure ofthe H5 globular domain shows which of the conserved and variableresidues are peripheral and which internal. Particular attentionis paid to conserved basic residues on the surface, which wetake to be DNA binding. Variable regions and conserved acidicresidues are assumed not to be sites of contact with DNA. Weconclude that one face of the domain, containing a cluster ofbasic residues, is the principal DNA binding site whilst twoopposing faces, orthogonal to the principal site and also containingconserved basic residues, are subsidiary DNA binding sites.Since the DNA binding surface of the domain covers a full 180°arc, we propose that it contacts a ‘cage’ of threeDNA strands on the 2-fold axis of the chromatosome.     

Modeling membrane proteins based on low-resolution electron microscopy maps: a template for the TM domains of the oxalate transporter OxlT

The availability of both EM and high-resolution crystallographic data for several membrane proteins (MPs) permits a detailed evaluation of the ability of molecular modeling techniques to complement EM data in the development of models of MPs. A protocol for this purpose is presented, consisting of (1) identifying transmembrane (TM) domains from sequence; (2) assigning buried and lipid-exposed faces of the TM domains; and (3) assembling the TM domains into a bundle, based on geometric restraints obtained from the EM data. The protocol is validated by predicting the structures of several 7- and 12-TM MPs to within 3-5 A r.m.s.d. from their crystal structures. The protocol is applied to generate a model of the oxalate transporter OxlT, for which a high-resolution structure is not yet available.     

Main structural and functional features of the basic cytosolic bovine 21 kDa protein delineated through Hydrophobic Cluster Analysis and molecular modelling

A 21 kDa protein purified from bovine brain cytosol was previouslydescribed as a hydrophobic ngand binding protein; however, itsaccurate biological function remained still uncertain. In orderto get further information about its potential biological role,an extended prediction of its secondary and three dimensionalstructures was undertaken. We describe here a process whichpermitted us to discover a structural homology between the 21kDa protein and the N-domain of yeast phosphoglycerate kinase(PGK). This process is based on comparing the 21 kDa proteinwith all the proteins presenting a slight homology, by usingthe Hydrophobic Cluster Analysis (HCA) method. According tothe observed similarity between the N-domaln of yeast PGK andthe 21 kDa protein, we built a model which was shown to possessa potential binding site for nucleotides. Moreover, the modelobtained presents three-dimensional (3D) structure similaritywith adenylate kinase. These results suggest two main hypotheses:(0 the 21 kDa protein may belong to the kinase family; (ii)the binding of a nucleotide could imply a modification of the3D structure of the 21 kDa protein that can promote the transferof hydrophobic ligands to the plasma membrane. Meanwhile, verificationof these hypotheses has been in part performed experimentally:the 21 kDa protein binds MgATP as well as, to a lesser extent,phosphoglycerate.     

Homology modelling of rat kallikrein rK9, a member of the tissue kallikrein family: implications for substrate specificity and inhibitor binding

The rat kallikrein rK9 is one of the six members of the rattissue kallikrein family isolated to date. It is 84% identicalto rK2 (tonin), and both proteinases are thought to have vasoconstrictiveproperties. Recently we have shown that rK9 and rK2 have distinctsubstrate specificities and sensitivities to inhibitors, despitetheir similar sequences. Unlike all other mammalian kallikrein-relatedproteinases, rK9 is resistant to inhibition by aprotinin. Wehave developed a 3-D model of rK9, based on the known X-raystructures of rK2, porcine kallikrein and bovine trypsin, toidentify the structural features underlying this functionaldiversity. The final rK9 model is structurally similar to rK2,but variable regions surrounding the active site differ quitemarkedly from the reference proteins. The kallikrein loop, whichdiffers from that in porcine kallikrein by a seven-residue insertion,has been generated de novo and subjected to simulated annealingto assess its influence on the restricted substrate specificityof these proteinases. The proposed conformation of the specificitypocket in rK9 differs from that of other serine proteinases,but it can still accommodate both aromatic and basic amino acidside chains at the substrate P₁ position, thus explaining thedual chymotrypsin and trypsin-like activity of rK9. The electrostaticpotentials of rK9 and aprotinin were calculated using the finitedifference Poisson–Boltzmann method. They indicated alarge positive region near the active site of rK9 not foundin related proteinases because of positively charged residuesat positions 61 and 65 in rK9. They generate a positive region,which overlaps a positive region in aprotinin, and may preventaprotinin binding. A single mutation in aprotinin is suggestedthat might allow kallikrein rK9 inhibition by aprotinin. Thismodel contributes significantly to our understanding of thestructure-function relationships among proteinases of the tissuekallikrein family.     

On the use of machine learning to identify topological rules in the packing of {beta}-strands

The machine learning program GOLEM was applied to discover topologicalrules in the packing ofß-sheets in /ß-domainproteins. Rules (constraints) were determined for four featuresof ß-sheet packing: (i) whether a ß-strandis at an edge; (ii) whether two consecutive ß-strandspack parallel or anti-parallel; (iii) whether twoß-strandspack adjacently; and (iv) the winding direction of two consecutiveß-strands. Rules were found with high predictive accuracyand coverage. The errors were generally associated with complicationsin domain folds, especially in one doubly wound domains. Investigationof the rules revealed interesting patterns, some of which wereknown previously, others that are novel. Novel features include(i) the relationship between pairs of sequential strands isin general one of decreasing size; (ii) more sequential pairsof strands wind in the direction out than in; and (iii) it takesa larger alteration in hydrophobicity to change a strand fromwinding in the direction out than in. These patterns in thedata may be the result of folding pathways in the domains. Therules found are of predictive value and could be used in thecombinatorial prediction of protein structure, or as a generaltest of model structures, e.g. those produced by threading.We conclude that machine learning has a useful role in the analysisof protein structures.     

Predicting the point at which transmembrane helices protrude from the bilayer: a model of the antenna complexes from photosynthetic bacteria

We describe a method for predicting the point at which a transmembranehelix leaves the bilayer and enters the more polar region ofthe aqueous exterior. This is achieved by comparing the relativedirections of the hydrophobic and internal faces of the transmembranehelices which should be opposite for the regions within thebilayer but equivalent for the regions on the outside. Thisinformation provides a strong constraint in the process of modellingmembrane proteins. We go on to use the approach to model themonomers of the bacterial light-harvesting antenna complexes.This information is then combined with some preliminary crystallographkdata and biochemical results to produce a 3-D model of a tetramer.