Can correct protein models be identified?

Effect of conformational dynamics on substrate recognition and specificity as probed by the introduction of a de novo disulfide bond into cytochrome P450 2B1

The conformational dynamics of cytochrome P450 2B1 (CYP2B1) were investigated through the introduction of a disulfide bond to link the I- and K-helices by generation of a double Cys variant, Y309C/S360C. The consequences of the disulfide bonding were examined both experimentally and in silico by molecular dynamics simulations. Under high hydrostatic pressures, the partial inactivation volume for the Y309C/S360C variant was determined to be -21 cm3mol(-1), which is more than twice as much as those of the wild type (WT) and single Cys variants (Y309C, S360C). This result indicates that the engineered disulfide bond has substantially reduced the protein plasticity of the Y309C/S360C variant. Under steady-state turnover conditions, the S360C variant catalyzed the N-demethylation of benzphetamine and O-deethylation of 7-ethoxy-trifluoromethylcoumarin as the WT did, whereas the Y309C variant retained only 39% of the N-demethylation activity and 66% of the O-deethylation activity compared with the WT. Interestingly, the Y309C/S360C variant restored the N-demethylation activity to the same level as that of the WT but decreased the O-deethylation activity to only 19% of the WT. Furthermore, the Y309C/S360C variant showed increased substrate specificity for testosterone over androstenedione. Molecular dynamics simulations revealed that the engineered disulfide bond altered substrate access channels. Taken together, these results suggest that protein dynamics play an important role in regulating substrate entry and recognition.

A conserved Na(+) binding site of the sodium-coupled neutral amino acid transporter 2 (SNAT2)

The SLC38 family of solute transporters mediates the coupled transport of amino acids and Na(+) into or out of cells. The structural basis for this coupled transport process is not known. Here, a profile-based sequence analysis approach was used, predicting a distant relationship with the SLC5/6 transporter families. Homology models using the LeuT(Aa) and Mhp1 transporters of known structure as templates were established, predicting the location of a conserved Na(+) binding site in the center of membrane helices 1 and 8. This homology model was tested experimentally in the SLC38 member SNAT2 by analyzing the effect of a mutation to Thr-384, which is predicted to be part of this Na(+) binding site. The results show that the T384A mutation not only inhibits the anion leak current, which requires Na(+) binding to SNAT2, but also dramatically lowers the Na(+) affinity of the transporter. This result is consistent with a previous analysis of the N82A mutant transporter, which has a similar effect on anion leak current and Na(+) binding and which is also expected to form part of the Na(+) binding site. In contrast, random mutations to other sites in the transporter had little or no effect on Na(+) affinity. Our results are consistent with a cation binding site formed by transmembrane helices 1 and 8 that is conserved among the SLC38 transporters as well as among many other bacterial and plant transporter families of unknown structure, which are homologous to SLC38.

The reducible complexity of a mitochondrial molecular machine

Molecular machines drive essential biological processes, with the component parts of these machines each contributing a partial function or structural element. Mitochondria are organelles of eukaryotic cells, and depend for their biogenesis on a set of molecular machines for protein transport. How these molecular machines evolved is a fundamental question. Mitochondria were derived from an alpha-proteobacterial endosymbiont, and we identified in alpha-proteobacteria the component parts of a mitochondrial protein transport machine. In bacteria, the components are found in the inner membrane, topologically equivalent to the mitochondrial proteins. Although the bacterial proteins function in simple assemblies, relatively little mutation would be required to convert them to function as a protein transport machine. This analysis of protein transport provides a blueprint for the evolution of cellular machinery in general.

Conformation of a plasmid replication initiator protein affects its proteolysis by ClpXP system

Proteins from the Rep family of DNA replication initiators exist mainly as dimers, but only monomers can initiate DNA replication by interaction with the replication origin (ori). In this study, we investigated both the activation (monomerization) and the degradation of the broad-host-range plasmid RK2 replication initiation protein TrfA, which we found to be a member of a class of DNA replication initiators containing winged helix (WH) domains. Our in vivo and in vitro experiments demonstrated that the ClpX-dependent activation of TrfA leading to replicationally active protein monomers and mutations affecting TrfA dimer formation, result in the inhibition of TrfA protein degradation by the ClpXP proteolytic system. These data revealed that the TrfA monomers and dimers are degraded at substantially different rates. Our data also show that the plasmid replication initiator activity and stability in E. coli cells are affected by ClpXP system only when the protein sustains dimeric form.

Extracellular acidification exerts opposite actions on TREK1 and TREK2 potassium channels via a single conserved histidine residue

Mechanosensitive K(+) channels TREK1 and TREK2 form a subclass of two P-domain K(+) channels. They are potently activated by polyunsaturated fatty acids and are involved in neuroprotection, anesthesia, and pain perception. Here, we show that acidification of the extracellular medium strongly inhibits TREK1 with an apparent pK near to 7.4 corresponding to the physiological pH. The all-or-none effect of pH variation is steep and is observed within one pH unit. TREK2 is not inhibited but activated by acidification within the same range of pH, despite its close homology with TREK1. A single conserved residue, H126 in TREK1 and H151 in TREK2, is involved in proton sensing. This histidine is located in the M1P1 extracellular loop preceding the first P domain. The differential effect of acidification, that is, activation for TREK2 and inhibition for TREK1, involves other residues located in the P2M4 loop, linking the second P domain and the fourth membrane-spanning segment. Structural modeling of TREK1 and TREK2 and site-directed mutagenesis strongly suggest that attraction or repulsion between the protonated side chain of histidine and closely located negatively or positively charged residues in P2M4 control outer gating of these channels. The differential sensitivity of TREK1 and TREK2 to external pH variations discriminates between these two K(+) channels that otherwise share the same regulations by physical and chemical stimuli, and by hormones and neurotransmitters.

Automated NMR structure determination of stereo-array isotope labeled ubiquitin from minimal sets of spectra using the SAIL-FLYA system

Stereo-array isotope labeling (SAIL) has been combined with the fully automated NMR structure determination algorithm FLYA to determine the three-dimensional structure of the protein ubiquitin from different sets of input NMR spectra. SAIL provides a complete stereo- and regio-specific pattern of stable isotopes that results in sharper resonance lines and reduced signal overlap, without information loss. Here we show that as a result of the superior quality of the SAIL NMR spectra, reliable, fully automated analyses of the NMR spectra and structure calculations are possible using fewer input spectra than with conventional uniformly 13C/15N-labeled proteins. FLYA calculations with SAIL ubiquitin, using a single three-dimensional "through-bond" spectrum (and 2D HSQC spectra) in addition to the 13C-edited and 15N-edited NOESY spectra for conformational restraints, yielded structures with an accuracy of 0.83-1.15 A for the backbone RMSD to the conventionally determined solution structure of SAIL ubiquitin. NMR structures can thus be determined almost exclusively from the NOESY spectra that yield the conformational restraints, without the need to record many spectra only for determining intermediate, auxiliary data of the chemical shift assignments. The FLYA calculations for this report resulted in 252 ubiquitin structure bundles, obtained with different input data but identical structure calculation and refinement methods. These structures cover the entire range from highly accurate structures to seriously, but not trivially, wrong structures, and thus constitute a valuable database for the substantiation of structure validation methods.

Structural bases of GM1 gangliosidosis and Morquio B disease

Allelic mutations of the lysosomal beta-galactosidase gene cause heterogeneous clinical phenotypes, such as GM1 gangliosidosis and Morquio B disease, the former being further classified into three variants, namely infantile, juvenile and adult forms; and heterogeneous biochemical phenotypes were shown in these forms. We tried to elucidate the bases of these diseases from a structural viewpoint. We first constructed a three-dimensional structural model of human beta-galactosidase by means of homology modeling. The human beta-galactosidase consists of three domains, such as, a TIM barrel fold domain, which functions as a catalytic domain, and two galactose-binding domain-like fold domains. We then constructed structural models of representative mutant beta-galactosidase proteins (G123R, R201C, I51T and Y83H) and predicted the structural change associated with each phenotype by calculating the number of affected atoms, determining the root-mean-square deviation and the solvent-accessible surface area, and by color imaging. The results show that there is a good correlation between the structural changes caused by amino-acid substitutions in the beta-galactosidase molecule, as well as biochemical and clinical phenotypes in these representative cases. Protein structural study is useful for elucidating the bases of these diseases.

QMEANclust: estimation of protein model quality by combining a composite scoring function with structural density information

BACKGROUND

The selection of the most accurate protein model from a set of alternatives is a crucial step in protein structure prediction both in template-based and ab initio approaches. Scoring functions have been developed which can either return a quality estimate for a single model or derive a score from the information contained in the ensemble of models for a given sequence. Local structural features occurring more frequently in the ensemble have a greater probability of being correct. Within the context of the CASP experiment, these so called consensus methods have been shown to perform considerably better in selecting good candidate models, but tend to fail if the best models are far from the dominant structural cluster. In this paper we show that model selection can be improved if both approaches are combined by pre-filtering the models used during the calculation of the structural consensus.

RESULTS

Our recently published QMEAN composite scoring function has been improved by including an all-atom interaction potential term. The preliminary model ranking based on the new QMEAN score is used to select a subset of reliable models against which the structural consensus score is calculated. This scoring function called QMEANclust achieves a correlation coefficient of predicted quality score and GDT_TS of 0.9 averaged over the 98 CASP7 targets and perform significantly better in selecting good models from the ensemble of server models than any other groups participating in the quality estimation category of CASP7. Both scoring functions are also benchmarked on the MOULDER test set consisting of 20 target proteins each with 300 alternatives models generated by MODELLER. QMEAN outperforms all other tested scoring functions operating on individual models, while the consensus method QMEANclust only works properly on decoy sets containing a certain fraction of near-native conformations. We also present a local version of QMEAN for the per-residue estimation of model quality (QMEANlocal) and compare it to a new local consensus-based approach.

CONCLUSION

Improved model selection is obtained by using a composite scoring function operating on single models in order to enrich higher quality models which are subsequently used to calculate the structural consensus. The performance of consensus-based methods such as QMEANclust highly depends on the composition and quality of the model ensemble to be analysed. Therefore, performance estimates for consensus methods based on large meta-datasets (e.g. CASP) might overrate their applicability in more realistic modelling situations with smaller sets of models based on individual methods.

QMEAN server for protein model quality estimation

Model quality estimation is an essential component of protein structure prediction, since ultimately the accuracy of a model determines its usefulness for specific applications. Usually, in the course of protein structure prediction a set of alternative models is produced, from which subsequently the most accurate model has to be selected. The QMEAN server provides access to two scoring functions successfully tested at the eighth round of the community-wide blind test experiment CASP. The user can choose between the composite scoring function QMEAN, which derives a quality estimate on the basis of the geometrical analysis of single models, and the clustering-based scoring function QMEANclust which calculates a global and local quality estimate based on a weighted all-against-all comparison of the models from the ensemble provided by the user. The web server performs a ranking of the input models and highlights potentially problematic regions for each model. The QMEAN server is available at http://swissmodel.expasy.org/qmean.

Solution structure of factor I-like modules from complement C7 reveals a pair of follistatin domains in compact pseudosymmetric arrangement

Factor I-like modules (FIMs) of complement proteins C6, C7, and factor I participate in protein-protein interactions critical to the progress of a complement-mediated immune response to infections and other trauma. For instance, the carboxyl-terminal FIM pair of C7 (C7-FIMs) binds to the C345C domain of C5 and its activated product, C5b, during self-assembly of the cytolytic membrane-attack complex. FIMs share sequence similarity with follistatin domains (FDs) of known three-dimensional structure, suggesting that FIM structures could be reliably modeled. However, conflicting disulfide maps, inconsistent orientations of subdomains within FDs, and the presence of binding partners in all FD structures led us to determine the three-dimensional structure of C7-FIMs by NMR spectroscopy. The solution structure reveals that each FIM within C7 contains a small amino-terminal FOLN subdomain connected to a larger carboxyl-terminal KAZAL domain. The open arrangement of the subdomains within FIMs resembles that of first FDs within structures of tandem FDs but differs from the more compact subdomain arrangement of second or third FDs. Unexpectedly, the two C7-FIMs pack closely together with an approximate 2-fold rotational symmetry that is rarely seen in module pairs and has not been observed in FD-containing proteins. Interfaces between subdomains and between modules include numerous hydrophobic and electrostatic contributions, suggesting that this is a physiologically relevant conformation that persists in the context of the parent protein. Similar interfaces were predicted in a homology-based model of the C6-FIM pair. The C7-FIM structures also facilitated construction of a model of the single FIM of factor I.

Purification and characterization of a wound-inducible thaumatin-like protein from the latex of Carica papaya

A 22.137 kDa protein constituent of fresh latex was isolated both from the latex of regularly damaged papaya trees and from a commercially available papain preparation. The protein was purified up to apparent homogeneity and was shown to be absent in the latex of papaya trees that had never been previously mechanically injured. This suggests that the protein belongs to pathogenesis-related protein family, as expected for several other protein constituents of papaya latex. The protein was identified as a thaumatin-like protein (class 5 of the pathogenesis-related proteins) on the basis of its partial amino acid sequence. By sequence analysis of the Carica genome, three different forms of thaumatin-like protein were identified, where the latex constituent belongs to a well-known form, allowing the molecular modeling of its spatial structure. The papaya latex thaumatin-like protein was further characterized. The protein appears to be stable in the pH interval from 2 to 10 and resistant to chemical denaturation by guanidium chloride, with a DeltaG(water)(0) of 15.2 kcal/mol and to proteolysis by the four papaya cysteine proteinases. The physiological role of this protein is discussed.

Data-driven homology modelling of P-glycoprotein in the ATP-bound state indicates flexibility of the transmembrane domains

Human P-glycoprotein is an ATP-binding cassette transporter that plays an important role in the defence against potentially harmful molecules from the environment. It is involved in conferring resistance against cancer therapeutics and plays an important role for the pharmacokinetics of drugs. The lack of a high resolution structure of P-glycoprotein has hindered its functional understanding and represents an obstacle for structure based drug development. The homologous bacterial exporter Sav1866 has been shown to share a common architecture and overlapping substrate specificity with P-glycoprotein. The structure of Sav1866 suggests that helices in the transmembrane domains diverge at the extracytoplasmic face, whereas cross-link information and a combination of small angle X-ray scattering and cryo-electron crystallography data indicate that helices 6 and 12 of P-glycoprotein are closer in P-glycoprotein than in the crystal structure of Sav1866. Using homology modelling, we present evidence that the protein possesses intrinsic structural flexibility to allow cross-links to occur between helices 6 and 12 of P-glycoprotein, thereby reconciling crystallographic models with available experimental data from cross-linking.

Outcome of a workshop on applications of protein models in biomedical research

We describe the proceedings and conclusions from the "Workshop on Applications of Protein Models in Biomedical Research" (the Workshop) that was held at the University of California, San Francisco on 11 and 12 July, 2008. At the Workshop, international scientists involved with structure modeling explored (i) how models are currently used in biomedical research, (ii) the requirements and challenges for different applications, and (iii) how the interaction between the computational and experimental research communities could be strengthened to advance the field.

Functional analysis of MmeI from methanol utilizer Methylophilus methylotrophus, a subtype IIC restriction-modification enzyme related to type I enzymes

MmeI from Methylophilus methylotrophus belongs to the type II restriction-modification enzymes. It recognizes an asymmetric DNA sequence, 5'-TCCRAC-3' (R indicates G or A), and cuts both strands at fixed positions downstream of the specific site. This particular feature has been exploited in transcript profiling of complex genomes (using serial analysis of gene expression technology). We have shown previously that the endonucleolytic activity of MmeI is strongly dependent on the presence of S-adenosyl-l-methionine (J. Nakonieczna, J. W. Zmijewski, B. Banecki, and A. J. Podhajska, Mol. Biotechnol. 37:127-135, 2007), which puts MmeI in subtype IIG. The same cofactor is used by MmeI as a methyl group donor for modification of an adenine in the upper strand of the recognition site to N(6)-methyladenine. Both enzymatic activities reside in a single polypeptide (919 amino acids [aa]), which puts MmeI also in subtype IIC of the restriction-modification systems. Based on a molecular model, generated with the use of bioinformatic tools and validated by site-directed mutagenesis, we were able to localize three functional domains in the structure of the MmeI enzyme: (i) the N-terminal portion containing the endonucleolytic domain with the catalytic Mg2+-binding motif D(70)-X(9)-EXK(82), characteristic for the PD-(D/E)XK superfamily of nucleases; (ii) a central portion (aa 310 to 610) containing nine sequence motifs conserved among N(6)-adenine gamma-class DNA methyltransferases; (iii) the C-terminal portion (aa 610 to 919) containing a putative target recognition domain. Interestingly, all three domains showed highest similarity to the corresponding elements of type I enzymes rather than to classical type II enzymes. We have found that MmeI variants deficient in restriction activity (D70A, E80A, and K82A) can bind and methylate specific nucleotide sequence. This suggests that domains of MmeI responsible for DNA restriction and modification can act independently. Moreover, we have shown that a single amino acid residue substitution within the putative target recognition domain (S807A) resulted in a MmeI variant with a higher endonucleolytic activity than the wild-type enzyme.

Improving accuracy of multiple sequence alignment algorithms based on alignment of neighboring residues

While most of the recent improvements in multiple sequence alignment accuracy are due to better use of vertical information, which include the incorporation of consistency-based pairwise alignments and the use of profile alignments, we observe that it is possible to further improve accuracy by taking into account alignment of neighboring residues when aligning two residues, thus making better use of horizontal information. By modifying existing multiple alignment algorithms to make use of horizontal information, we show that this strategy is able to consistently improve over existing algorithms on a few sets of benchmark alignments that are commonly used to measure alignment accuracy, and the average improvements in accuracy can be as much as 1–3% on protein sequence alignment and 5–10% on DNA/RNA sequence alignment. Unlike previous algorithms, consistent average improvements can be obtained across all identity levels.

SELECTpro: effective protein model selection using a structure-based energy function resistant to BLUNDERs

Background

Protein tertiary structure prediction is a fundamental problem in computational biology and identifying the most native-like model from a set of predicted models is a key sub-problem. Consensus methods work well when the redundant models in the set are the most native-like, but fail when the most native-like model is unique. In contrast, structure-based methods score models independently and can be applied to model sets of any size and redundancy level. Additionally, structure-based methods have a variety of important applications including analogous fold recognition, refinement of sequence-structure alignments, and de novo prediction. The purpose of this work was to develop a structure-based model selection method based on predicted structural features that could be applied successfully to any set of models.

Results

Here we introduce SELECTpro, a novel structure-based model selection method derived from an energy function comprising physical, statistical, and predicted structural terms. Novel and unique energy terms include predicted secondary structure, predicted solvent accessibility, predicted contact map, β-strand pairing, and side-chain hydrogen bonding.

SELECTpro participated in the new model quality assessment (QA) category in CASP7, submitting predictions for all 95 targets and achieved top results. The average difference in GDT-TS between models ranked first by SELECTpro and the most native-like model was 5.07. This GDT-TS difference was less than 1% of the GDT-TS of the most native-like model for 18 targets, and less than 10% for 66 targets. SELECTpro also ranked the single most native-like first for 15 targets, in the top five for 39 targets, and in the top ten for 53 targets, more often than any other method. Because the ranking metric is skewed by model redundancy and ignores poor models with a better ranking than the most native-like model, the BLUNDER metric is introduced to overcome these limitations. SELECTpro is also evaluated on a recent benchmark set of 16 small proteins with large decoy sets of 12500 to 20000 models for each protein, where it outperforms the benchmarked method (I-TASSER).

Conclusion

SELECTpro is an effective model selection method that scores models independently and is appropriate for use on any model set. SELECTpro is available for download as a stand alone application at: . SELECTpro is also available as a public server at the same site.

Modeling of Escherichia coli Endonuclease V structure in complex with DNA

Endonuclease V (EndoV) is a metal-dependent DNA repair enzyme involved in removal of deaminated bases (e.g., deoxyuridine, deoxyinosine, and deoxyxanthosine), with pairing specificities different from the original bases. Homologs of EndoV are present in all major phyla from bacteria to humans and their function is quite well analyzed. EndoV has been combined with DNA ligase to develop an enzymatic method for mutation scanning and has been engineered to obtain variants with different substrate specificities that serve as improved tools in mutation recognition and cancer mutation scanning. However, little is known about the structure and mechanism of substrate DNA binding by EndoV. Here, we present the results of a bioinformatic analysis and a structural model of EndoV from Escherichia coli in complex with DNA. The structure was obtained by a combination of fold-recognition, comparative modeling, de novo modeling and docking methods. The modeled structure provides a convenient tool to study protein sequence-structure-function relationships in EndoV and to engineer its further variants.

An interrupted beta-propeller and protein disorder: structural bioinformatics insights into the N-terminus of alsin

Defects in the human ALS2 gene, which encodes the 1,657-amino-acid residue protein alsin, are linked to several related motor neuron diseases. We created a structural model for the N-terminal 690-residue region of alsin through comparative modelling based on regulator of chromosome condensation 1 (RCC1). We propose that this alsin region contains seven RCC1-like repeats in a seven-bladed beta-propeller structure. The propeller is formed by a double clasp arrangement containing two segments (residues 1-218 and residues 525-690). The 306-residue insert region, predicted to lie within blade 5 and to be largely disordered, is poorly conserved across species. Surface patches of evolutionary conservation probably indicate locations of binding sites. Both disease-causing missense mutations-Cys157Tyr and Gly540Glu-are buried in the propeller and likely to be structurally disruptive. This study aids design of experimental studies by highlighting the importance of construct length, will enhance interpretation of protein-protein interactions, and enable rational site-directed mutagenesis.

Type II restriction endonuclease R.Hpy188I belongs to the GIY-YIG nuclease superfamily, but exhibits an unusual active site

BACKGROUND

Catalytic domains of Type II restriction endonucleases (REases) belong to a few unrelated three-dimensional folds. While the PD-(D/E)XK fold is most common among these enzymes, crystal structures have been also determined for single representatives of two other folds: PLD (R.BfiI) and half-pipe (R.PabI). Bioinformatics analyses supported by mutagenesis experiments suggested that some REases belong to the HNH fold (e.g. R.KpnI), and that a small group represented by R.Eco29kI belongs to the GIY-YIG fold. However, for a large fraction of REases with known sequences, the three-dimensional fold and the architecture of the active site remain unknown, mostly due to extreme sequence divergence that hampers detection of homology to enzymes with known folds.

RESULTS

R.Hpy188I is a Type II REase with unknown structure. PSI-BLAST searches of the non-redundant protein sequence database reveal only 1 homolog (R.HpyF17I, with nearly identical amino acid sequence and the same DNA sequence specificity). Standard application of state-of-the-art protein fold-recognition methods failed to predict the relationship of R.Hpy188I to proteins with known structure or to other protein families. In order to increase the amount of evolutionary information in the multiple sequence alignment, we have expanded our sequence database searches to include sequences from metagenomics projects. This search resulted in identification of 23 further members of R.Hpy188I family, both from metagenomics and the non-redundant database. Moreover, fold-recognition analysis of the extended R.Hpy188I family revealed its relationship to the GIY-YIG domain and allowed for computational modeling of the R.Hpy188I structure. Analysis of the R.Hpy188I model in the light of sequence conservation among its homologs revealed an unusual variant of the active site, in which the typical Tyr residue of the YIG half-motif had been substituted by a Lys residue. Moreover, some of its homologs have the otherwise invariant Arg residue in a non-homologous position in sequence that nonetheless allows for spatial conservation of the guanidino group potentially involved in phosphate binding.

CONCLUSION

The present study eliminates a significant "white spot" on the structural map of REases. It also provides important insight into sequence-structure-function relationships in the GIY-YIG nuclease superfamily. Our results reveal that in the case of proteins with no or few detectable homologs in the standard "non-redundant" database, it is useful to expand this database by adding the metagenomic sequences, which may provide evolutionary linkage to detect more remote homologs.

How well can the accuracy of comparative protein structure models be predicted?

Comparative structure models are available for two orders of magnitude more protein sequences than are experimentally determined structures. These models, however, suffer from two limitations that experimentally determined structures do not: They frequently contain significant errors, and their accuracy cannot be readily assessed. We have addressed the latter limitation by developing a protocol optimized specifically for predicting the Calpha root-mean-squared deviation (RMSD) and native overlap (NO3.5A) errors of a model in the absence of its native structure. In contrast to most traditional assessment scores that merely predict one model is more accurate than others, this approach quantifies the error in an absolute sense, thus helping to determine whether or not the model is suitable for intended applications. The assessment relies on a model-specific scoring function constructed by a support vector machine. This regression optimizes the weights of up to nine features, including various sequence similarity measures and statistical potentials, extracted from a tailored training set of models unique to the model being assessed: If possible, we use similarly sized models with the same fold; otherwise, we use similarly sized models with the same secondary structure composition. This protocol predicts the RMSD and NO3.5A errors for a diverse set of 580,317 comparative models of 6174 sequences with correlation coefficients (r) of 0.84 and 0.86, respectively, to the actual errors. This scoring function achieves the best correlation compared to 13 other tested assessment criteria that achieved correlations ranging from 0.35 to 0.71.

Estimating quality of template-based protein models by alignment stability

The error in protein tertiary structure prediction is unavoidable, but it is not explicitly shown in most of the current prediction algorithms. Estimated error of a predicted structure is crucial information for experimental biologists to use the prediction model for design and interpretation of experiments. Here, we propose a method to estimate errors in predicted structures based on the stability of the optimal target-template alignment when compared with a set of suboptimal alignments. The stability of the optimal alignment is quantified by an index named the SuboPtimal Alignment Diversity (SPAD). We implemented SPAD in a profile-based threading algorithm and investigated how well SPAD can indicate errors in threading models using a large benchmark dataset of 5232 alignments. SPAD shows a very good correlation not only to alignment shift errors but also structure-level errors, the root mean square deviation (RMSD) of predicted structure models to the native structures (i.e. global errors), and local errors at each residue position. We have further compared SPAD with seven other quality measures, six from sequence alignment-based measures and one atomic statistical potential, discrete optimized protein energy (DOPE), in terms of the correlation coefficient to the global and local structure-level errors. In terms of the correlation to the RMSD of structure models, when a target and a template are in the same SCOP family, the sequence identity showed a best correlation to the RMSD; in the superfamily level, SPAD was the best; and in the fold level, DOPE was best. However, in a head-to-head comparison, SPAD wins over the other measures. Next, SPAD is compared with three other measures of local errors. In this comparison, SPAD was best in all of the family, the superfamily and the fold levels. Using the discovered correlation, we have also predicted the global and local error of our predicted structures of CASP7 targets by the SPAD. Finally, we proposed a sausage representation of predicted tertiary structures which intuitively indicate the predicted structure and the estimated error range of the structure simultaneously.

Protein model quality assessment prediction by combining fragment comparisons and a consensus C(alpha) contact potential

In this work, we develop a fully automated method for the quality assessment prediction of protein structural models generated by structure prediction approaches such as fold recognition servers, or ab initio methods. The approach is based on fragment comparisons and a consensus C(alpha) contact potential derived from the set of models to be assessed and was tested on CASP7 server models. The average Pearson linear correlation coefficient between predicted quality and model GDT-score per target is 0.83 for the 98 targets, which is better than those of other quality assessment methods that participated in CASP7. Our method also outperforms the other methods by about 3% as assessed by the total GDT-score of the selected top models.

Sequence-structure-function analysis of the bifunctional enzyme MnmC that catalyses the last two steps in the biosynthesis of hypermodified nucleoside mnm5s2U in tRNA

MnmC catalyses the last two steps in the biosynthesis of 5-methylaminomethyl-2-thiouridine (mnm(5)s(2)U) in tRNA. Previously, we reported that this bifunctional enzyme is encoded by the yfcK open reading frame in the Escherichia coli K12 genome. However, the mechanism of its activity, in particular the potential structural and functional dependence of the domains responsible for catalyzing the two modification reactions, remains unknown. With the aid of the protein fold-recognition method, we constructed a structural model of MnmC in complex with the ligands and target nucleosides and studied the role of individual amino acids and entire domains by site-directed and deletion mutagenesis, respectively. We found out that the N-terminal domain contains residues responsible for binding of the S-adenosylmethionine cofactor and catalyzing the methylation of nm(5)s(2)U to form mnm(5)s(2)U, while the C-terminal domain contains residues responsible for binding of the FAD cofactor. Further, point mutants with compromised activity of either domain can complement each other to restore a fully functional enzyme. Thus, in the conserved fusion protein MnmC, the individual domains retain independence as enzymes. Interestingly, the N-terminal domain is capable of independent folding, while the isolated C-terminal domain is incapable of folding on its own, a situation similar to the one reported recently for the rRNA modification enzyme RsmC.

Alternating evolutionary pressure in a genetic algorithm facilitates protein model selection

Background

Automatic protein modelling pipelines are becoming ever more accurate; this has come hand in hand with an increasingly complicated interplay between all components involved. Nevertheless, there are still potential improvements to be made in template selection, refinement and protein model selection.

Results

In the context of an automatic modelling pipeline, we analysed each step separately, revealing several non-intuitive trends and explored a new strategy for protein conformation sampling using Genetic Algorithms (GA). We apply the concept of alternating evolutionary pressure (AEP), i.e. intermediate rounds within the GA runs where unrestrained, linear growth of the model populations is allowed.

Conclusion

This approach improves the overall performance of the GA by allowing models to overcome local energy barriers. AEP enabled the selection of the best models in 40% of all targets; compared to 25% for a normal GA.

Using multiple templates to improve quality of homology models in automated homology modeling

When researchers build high-quality models of protein structure from sequence homology, it is today common to use several alternative target-template alignments. Several methods can, at least in theory, utilize information from multiple templates, and many examples of improved model quality have been reported. However, to our knowledge, thus far no study has shown that automatic inclusion of multiple alignments is guaranteed to improve models without artifacts. Here, we have carried out a systematic investigation of the potential of multiple templates to improving homology model quality. We have used test sets consisting of targets from both recent CASP experiments and a larger reference set. In addition to Modeller and Nest, a new method (Pfrag) for multiple template-based modeling is used, based on the segment-matching algorithm from Levitt's SegMod program. Our results show that all programs can produce multi-template models better than any of the single-template models, but a large part of the improvement is simply due to extension of the models. Most of the remaining improved cases were produced by Modeller. The most important factor is the existence of high-quality single-sequence input alignments. Because of the existence of models that are worse than any of the top single-template models, the average model quality does not improve significantly. However, by ranking models with a model quality assessment program such as ProQ, the average quality is improved by approximately 5% in the CASP7 test set.

The YqfN protein of Bacillus subtilis is the tRNA: m1A22 methyltransferase (TrmK)

N¹-methylation of adenosine to m¹A occurs in several different positions in tRNAs from various organisms. A methyl group at position N¹ prevents Watson–Crick-type base pairing by adenosine and is therefore important for regulation of structure and stability of tRNA molecules. Thus far, only one family of genes encoding enzymes responsible for m¹A methylation at position 58 has been identified, while other m¹A methyltransferases (MTases) remain elusive. Here, we show that Bacillus subtilis open reading frame yqfN is necessary and sufficient for N¹-adenosine methylation at position 22 of bacterial tRNA. Thus, we propose to rename YqfN as TrmK, according to the traditional nomenclature for bacterial tRNA MTases, or TrMet(m¹A22) according to the nomenclature from the MODOMICS database of RNA modification enzymes. tRNAs purified from a ΔtrmK strain are a good substrate in vitro for the recombinant TrmK protein, which is sufficient for m¹A methylation at position 22 as are tRNAs from Escherichia coli, which natively lacks m¹A22. TrmK is conserved in Gram-positive bacteria and present in some Gram-negative bacteria, but its orthologs are apparently absent from archaea and eukaryota. Protein structure prediction indicates that the active site of TrmK does not resemble the active site of the m¹A58 MTase TrmI, suggesting that these two enzymatic activities evolved independently.

Evolutionary potentials: structure specific knowledge-based potentials exploiting the evolutionary record of sequence homologs

So-called ‘Evolutionary potentials’ for protein structure prediction are derived using a single experimental protein structure and all three-dimensional models of its homologous sequences.

We introduce a new type of knowledge-based potentials for protein structure prediction, called 'evolutionary potentials', which are derived using a single experimental protein structure and all three-dimensional models of its homologous sequences. The new potentials have been benchmarked against other knowledge-based potentials, resulting in a significant increase in accuracy for model assessment. In contrast to standard knowledge-based potentials, we propose that evolutionary potentials capture key determinants of thermodynamic stability and specific sequence constraints required for fast folding.

QMEAN: A comprehensive scoring function for model quality assessment

In protein structure prediction, a considerable number of alternative models are usually produced from which subsequently the final model has to be selected. Thus, a scoring function for the identification of the best model within an ensemble of alternative models is a key component of most protein structure prediction pipelines. QMEAN, which stands for Qualitative Model Energy ANalysis, is a composite scoring function describing the major geometrical aspects of protein structures. Five different structural descriptors are used. The local geometry is analyzed by a new kind of torsion angle potential over three consecutive amino acids. A secondary structure-specific distance-dependent pairwise residue-level potential is used to assess long-range interactions. A solvation potential describes the burial status of the residues. Two simple terms describing the agreement of predicted and calculated secondary structure and solvent accessibility, respectively, are also included. A variety of different implementations are investigated and several approaches to combine and optimize them are discussed. QMEAN was tested on several standard decoy sets including a molecular dynamics simulation decoy set as well as on a comprehensive data set of totally 22,420 models from server predictions for the 95 targets of CASP7. In a comparison to five well-established model quality assessment programs, QMEAN shows a statistically significant improvement over nearly all quality measures describing the ability of the scoring function to identify the native structure and to discriminate good from bad models. The three-residue torsion angle potential turned out to be very effective in recognizing the native fold.

Bud23 methylates G1575 of 18S rRNA and is required for efficient nuclear export of pre-40S subunits

BUD23 was identified from a bioinformatics analysis of Saccharomyces cerevisiae genes involved in ribosome biogenesis. Deletion of BUD23 leads to severely impaired growth, reduced levels of the small (40S) ribosomal subunit, and a block in processing 20S rRNA to 18S rRNA, a late step in 40S maturation. Bud23 belongs to the S-adenosylmethionine-dependent Rossmann-fold methyltransferase superfamily and is related to small-molecule methyltransferases. Nevertheless, we considered that Bud23 methylates rRNA. Methylation of G1575 is the only mapped modification for which the methylase has not been assigned. Here, we show that this modification is lost in bud23 mutants. The nuclear accumulation of the small-subunit reporters Rps2-green fluorescent protein (GFP) and Rps3-GFP, as well as the rRNA processing intermediate, the 5' internal transcribed spacer 1, indicate that bud23 mutants are defective for small-subunit export. Mutations in Bud23 that inactivated its methyltransferase activity complemented a bud23Delta mutant. In addition, mutant ribosomes in which G1575 was changed to adenosine supported growth comparable to that of cells with wild-type ribosomes. Thus, Bud23 protein, but not its methyltransferase activity, is important for biogenesis and export of the 40S subunit in yeast.

HsdR subunit of the type I restriction-modification enzyme EcoR124I: biophysical characterisation and structural modelling

Type I restriction-modification (RM) systems are large, multifunctional enzymes composed of three different subunits. HsdS and HsdM form a complex in which HsdS recognizes the target DNA sequence, and HsdM carries out methylation of adenosine residues. The HsdR subunit, when associated with the HsdS-HsdM complex, translocates DNA in an ATP-dependent process and cleaves unmethylated DNA at a distance of several thousand base-pairs from the recognition site. The molecular mechanism by which these enzymes translocate the DNA is not fully understood, in part because of the absence of crystal structures. To date, crystal structures have been determined for the individual HsdS and HsdM subunits and models have been built for the HsdM-HsdS complex with the DNA. However, no structure is available for the HsdR subunit. In this work, the gene coding for the HsdR subunit of EcoR124I was re-sequenced, which showed that there was an error in the published sequence. This changed the position of the stop codon and altered the last 17 amino acid residues of the protein sequence. An improved purification procedure was developed to enable HsdR to be purified efficiently for biophysical and structural analysis. Analytical ultracentrifugation shows that HsdR is monomeric in solution, and the frictional ratio of 1.21 indicates that the subunit is globular and fairly compact. Small angle neutron-scattering of the HsdR subunit indicates a radius of gyration of 3.4 nm and a maximum dimension of 10 nm. We constructed a model of the HsdR using protein fold-recognition and homology modelling to model individual domains, and small-angle neutron scattering data as restraints to combine them into a single molecule. The model reveals an ellipsoidal shape of the enzymatic core comprising the N-terminal and central domains, and suggests conformational heterogeneity of the C-terminal region implicated in binding of HsdR to the HsdS-HsdM complex.

Pseudodominant inheritance of goitrous congenital hypothyroidism caused by TPO mutations: molecular and in silico studies

CONTEXT AND OBJECTIVE

Most cases of goitrous congenital hypothyroidism (CH) from thyroid dyshormonogenesis 1) follow a recessive mode of inheritance and 2) are due to mutations in the thyroid peroxidase gene (TPO). We report the genetic mechanism underlying the apparently dominant inheritance of goitrous CH in a nonconsanguineous family of French Canadian origin.

DESIGN, SETTING, AND PARTICIPANTS

Two brothers identified by newborn TSH screening had severe hypothyroidism and a goiter with increased (99m)Tc uptake. The mother was euthyroid, but the father and two paternal uncles had also been diagnosed with goitrous CH. After having excluded PAX8 gene mutations, we hypothesized that the underlying defect could be TPO mutations.

RESULTS

Both compound heterozygous siblings had inherited a mutant TPO allele carried by their mother (c.1496delC; p.Pro499Argfs2X), and from their father, one brother had inherited a missense mutation (c.1978C-->G; p.Gln660Glu) and the other an insertion (c.1955insT; p.Phe653Valfs15X). The thyroid gland of one uncle who is a compound heterozygote for TPO mutations (p.Phe653Valfs15X/p.Gln660Glu) was removed because of concurrent multiple endocrine neoplasia type 2A. Immunohistochemistry revealed normal TPO staining, implying that Gln660Glu TPO is expressed properly. Modeling of this mutant in silico suggests that its three-dimensional structure is conserved, whereas the electrostatic binding energy between the Gln660Glu TPO and its heme group becomes repulsive.

CONCLUSION

We report a pedigree presenting with pseudodominant goitrous CH due to segregation of three different TPO mutations. Although goitrous CH generally follows a recessive mode of inheritance, the high frequency of TPO mutations carriers may lead to pseudodominant inheritance.

Validation of protein models by a neural network approach

Background

The development and improvement of reliable computational methods designed to evaluate the quality of protein models is relevant in the context of protein structure refinement, which has been recently identified as one of the bottlenecks limiting the quality and usefulness of protein structure prediction.

Results

In this contribution, we present a computational method (Artificial Intelligence Decoys Evaluator: AIDE) which is able to consistently discriminate between correct and incorrect protein models. In particular, the method is based on neural networks that use as input 15 structural parameters, which include energy, solvent accessible surface, hydrophobic contacts and secondary structure content. The results obtained with AIDE on a set of decoy structures were evaluated using statistical indicators such as Pearson correlation coefficients, Z_nat, fraction enrichment, as well as ROC plots. It turned out that AIDE performances are comparable and often complementary to available state-of-the-art learning-based methods.

Conclusion

In light of the results obtained with AIDE, as well as its comparison with available learning-based methods, it can be concluded that AIDE can be successfully used to evaluate the quality of protein structures. The use of AIDE in combination with other evaluation tools is expected to further enhance protein refinement efforts.

Prediction of global and local model quality in CASP7 using Pcons and ProQ

The ability to rank and select the best model is important in protein structure prediction. Model Quality Assessment Programs (MQAPs) are programs developed to perform this task. They can be divided into three categories based on the information they use. Consensus based methods use the similarity to other models, structure-based methods use features calculated from the structure and evolutionary based methods use the sequence similarity between a model and a template. These methods can be trained to predict the overall global quality of a model, that is, how much a model is likely to differ from the native structure. The methods can also be trained to pinpoint which local regions in a model are likely to be incorrect. In CASP7, we participated with three predictors of global and four of local quality using information from the three categories described above. The result shows that the MQAP using consensus, Pcons, was significantly better at predicting both global and local quality compared with MQAPs using only structure or sequence based information.

Fams-ace: a combined method to select the best model after remodeling all server models

During Critical Assessment of Protein Structure Prediction (CASP7, Pacific Grove, CA, 2006), fams-ace was entered in the 3D coordinate prediction category as a human expert group. The procedure can be summarized by the following three steps. (1) All the server models were refined and rebuilt utilizing our homology modeling method. (2) Representative structures were selected from each server, according to a model quality evaluation, based on a 3D1D profile score (like Verify3D). (3) The top five models were selected and submitted in the order of the consensus-based score (like 3D-Jury). Fams-ace is a fully automated server and does not require human intervention. In this article, we introduce the methodology of fams-ace and discuss the successes and failures of this approach during CASP7. In addition, we discuss possible improvements for the next CASP.

Improvement of cyclophosphamide activation by CYP2B6 mutants: from in silico to ex vivo

Cyclophosphamide (CPA) is a chemotherapeutic agent that is primarily activated in the liver by cytochrome P4502B6 (CYP2B6) and then transported to the tumor via blood flow. To prevent deleterious secondary effects, P450-based gene-directed enzyme prodrug therapy (GDEPT) consists of expressing CYP2B6 in tumor cells before CPA treatment. Given the relatively low affinity of CYP2B6 for CPA, the aim of our work was to modify CYP2B6 to increase its catalytic efficiency (V(max)/K(m)) to metabolize CPA into 4'-OH CPA. A molecular model of CYP2B6 was built, and four residues in close contact with the substrate were subjected to mutagenesis. Canine CYP2B11 exhibiting a particularly low K(m) to CPA, the amino acids exclusively present in the CYP2B11 substrate recognition sequences were substituted in human CYP2B6. All mutants (n = 26) were expressed in Saccharomyces cerevisiae and their enzymatic constants (K(m), V(max)) evaluated using CPA as substrate. Five mutants exhibited a 2- to 3-fold higher catalytic efficiency than wild-type CYP2B6. A double mutant, comprising the two most effective mutations, showed a 4-fold increase in K(m)/V(max). Molecular dynamic simulations of several mutants were found to be consistent with the observed modifications in catalytic efficiency. Finally, expression of the CYP2B6 114V/477W double mutant, contrary to wt CYP2B6, allowed switching of a resistant human head and neck cancer cell line (A-253) into a sensitive cell line toward CPA. Thus, we were able to obtain a new efficient CYP2B6 mutant able to metabolize CPA, an important step in the GDEPT strategy for human cancer treatment.

Structural bioinformatics analysis of enzymes involved in the biosynthesis pathway of the hypermodified nucleoside ms(2)io(6)A37 in tRNA

TRNAs from all organisms contain posttranscriptionally modified nucleosides, which are derived from the four canonical nucleosides. In most tRNAs that read codons beginning with U, adenosine in the position 37 adjacent to the 3' position of the anticodon is modified to N(6)-(Delta(2)-isopentenyl) adenosine (i(6)A). In many bacteria, such as Escherichia coli, this residue is typically hypermodified to N(6)-isopentenyl-2-thiomethyladenosine (ms(2)i(6)A). In a few bacteria, such as Salmonella typhimurium, ms(2)i(6)A can be further hydroxylated to N(6)-(cis-4-hydroxyisopentenyl)-2-thiomethyladenosine (ms(2)io(6)A). Although the enzymes that introduce the respective modifications (prenyltransferase MiaA, methylthiotransferase MiaB, and hydroxylase MiaE) have been identified, their structures remain unknown and sequence-function relationships remain obscure. We carried out sequence analysis and structure prediction of MiaA, MiaB, and MiaE, using the protein fold-recognition approach. Three-dimensional models of all three proteins were then built using a new modeling protocol designed to overcome uncertainties in the alignments and divergence between the templates. For MiaA and MiaB, the catalytic core was built based on the templates from the P-loop NTPase and Radical-SAM superfamilies, respectively. For MiaB, we have also modeled the C-terminal TRAM domain and the newly predicted N-terminal flavodoxin-fold domain. For MiaE, we confidently predict that it shares the three-dimensional fold with the ferritin-like four-helix bundle proteins and that it has a similar active site and mechanism of action to diiron carboxylate enzymes, in particular, methane monooxygenase (E.C.1.14.13.25) that catalyses the biological hydroxylation of alkanes. Our models provide the first structural platform for enzymes involved in the biosynthesis of i(6)A, ms(2)i(6)A, and ms(2)io(6)A, explain the data available from the literature and will help to design further experiments and interpret their results.

Protein structure prediction using threading

This chapter discusses the protocol for computational protein structure prediction by protein threading. First, we present a general procedure and summarize some typical ideas for each step of protein threading. Then, we describe the design and implementation of RAPTOR, a protein structure prediction program based on threading. The major focuses are three key components of RAPTOR: a linear programming approach to protein threading, two machine learning approaches (SVM and Gradient Boosting) to fold recognition, and evaluation of the statistical significance of the prediction results. The first part of this chapter is a brief review of protein threading, and the second part contains original research results. Some key ideas and results have been previously published.

A model of restriction endonuclease MvaI in complex with DNA: a template for interpretation of experimental data and a guide for specificity engineering

R.MvaI is a Type II restriction enzyme (REase), which specifically recognizes the pentanucleotide DNA sequence 5'-CCWGG-3' (W indicates A or T). It belongs to a family of enzymes, which recognize related sequences, including 5'-CCSGG-3' (S indicates G or C) in the case of R.BcnI, or 5'-CCNGG-3' (where N indicates any nucleoside) in the case of R.ScrFI. REases from this family hydrolyze the phosphodiester bond in the DNA between the 2nd and 3rd base in both strands, thereby generating a double strand break with 5'-protruding single nucleotides. So far, no crystal structures of REases with similar cleavage patterns have been solved. Characterization of sequence-structure-function relationships in this family would facilitate understanding of evolution of sequence specificity among REases and could aid in engineering of enzymes with new specificities. However, sequences of R.MvaI or its homologs show no significant similarity to any proteins with known structures, thus precluding straightforward comparative modeling. We used a fold recognition approach to identify a remote relationship between R.MvaI and the structure of DNA repair enzyme MutH, which belongs to the PD-(D/E)XK superfamily together with many other REases. We constructed a homology model of R.MvaI and used it to predict functionally important amino acid residues and the mode of interaction with the DNA. In particular, we predict that only one active site of R.MvaI interacts with the DNA target at a time, and the cleavage of both strands (5'-CCAGG-3' and 5'-CCTGG-3') is achieved by two independent catalytic events. The model is in good agreement with the available experimental data and will serve as a template for further analyses of R.MvaI, R.BcnI, R.ScrFI and other related enzymes.

The sequence and model structure analysis of three Polish peanut stunt virus strains

Peanut stunt virus (PSV) belongs to the Cucumovirus genus of the family Bromoviridae and is widely distributed worldwide, also in Poland. PSV is a common pathogen of a wide range of economically important plants. Its coat protein (CP), similarly as in other viruses, plays an important role in many processes during viral life cycle and has great impact on the infectivity. In this study, we present the results of sequence-structure analysis of CP derived from three Polish strains of PSV: PSV Ag, G, and P. Sequences were determined using RT-PCR amplification followed by sequencing and compared with each other and also with CP from other known PSV viruses. We analyzed their phylogenetic relationship, based on CP sequence, using bioinformatic tools as well as their spatial model using homology-modeling approach with combination of ROSETTA algorithm for de novo modeling. We compared our model with those recently obtained for other cucumoviruses including PSV-Er. Our results have shown that all Polish strains probably belong to the first subgroup of PSV viruses. Homology level between strains Ag and G proved very high. Using theoretical modeling approach we obtained a model very similar to the one resolved previously with the differences caused by slightly different amino acid sequence. We have also undertaken an attempt to analyze its distant regions; however, results are not unequivocal. Analysis of symptoms and their correlation with specific amino acid position was also performed on the basis of results published elsewhere. The definite interpretation is impeded by the presence of satellite RNAs in Ag and P strains modulating symptoms' severity, though.

Structure-Function Analysis of the Endoplasmic Reticulum Oxidoreductase TMX3 Reveals Interdomain Stabilization of the N-terminal Redox-active Domain

Disulfide bond formation in the endoplasmic reticulum is catalyzed by enzymes of the protein disulfide-isomerase family that harbor one or more thioredoxin-like domains. We recently discovered the transmembrane protein TMX3, a thiol-disulfide oxidoreductase of the protein disulfide-isomerase family. Here, we show that the endoplasmic reticulum-luminal region of TMX3 contains three thioredoxin-like domains, an N-terminal redox-active domain (named a) followed by two enzymatically inactive domains (b and b'). Using the recombinantly expressed TMX3 domain constructs a, ab, and abb', we compared structural stability and enzymatic properties. By structural and biophysical methods, we demonstrate that the reduced a domain has features typical of a globular folded domain that is, however, greatly destabilized upon oxidization. Importantly, interdomain stabilization by the b domain renders the a domain more resistant toward chemical denaturation and proteolysis in both the oxidized and reduced form. In combination with molecular modeling studies of TMX3 abb', the experimental results provide a new understanding of the relationship between the multidomain structure of TMX3 and its function as a redox enzyme. Overall, the data indicate that in addition to their role as substrate and co-factor binding domains, redox-inactive thioredoxin-like domains also function in stabilizing neighboring redox-active domains.