Structure solution of DNA-binding proteins and complexes with ARCIMBOLDO libraries

Fragment-based determination of a proteinase K structure from MicroED data using ARCIMBOLDO_SHREDDER

Structure determination of novel biological macromolecules by X-ray crystallography can be facilitated by the use of small structural fragments, some of only a few residues in length, as effective search models for molecular replacement to overcome the phase problem. Independence from the need for a complete pre-existing model with sequence similarity to the crystallized molecule is the primary appeal of ARCIMBOLDO, a suite of programs which employs this ab initio algorithm for phase determination. Here, the use of ARCIMBOLDO is investigated to overcome the phase problem with the electron cryomicroscopy (cryoEM) method known as microcrystal electron diffraction (MicroED). The results support the use of the ARCIMBOLDO_SHREDDER pipeline to provide phasing solutions for a structure of proteinase K from 1.6 Å resolution data using model fragments derived from the structures of proteins sharing a sequence identity of as low as 20%. ARCIMBOLDO_SHREDDER identified the most accurate polyalanine fragments from a set of distantly related sequence homologues. Alternatively, such templates were extracted in spherical volumes and given internal degrees of freedom to refine towards the target structure. Both modes relied on the rotation function in Phaser to identify or refine fragment models and its translation function to place them. Model completion from the placed fragments proceeded through phase combination of partial solutions and/or density modification and main-chain autotracing using SHELXE. The combined set of fragments was sufficient to arrive at a solution that resembled that determined by conventional molecular replacement using the known target structure as a search model. This approach obviates the need for a single, complete and highly accurate search model when phasing MicroED data, and permits the evaluation of large fragment libraries for this purpose.

ALEPH: a network-oriented approach for the generation of fragment-based libraries and for structure interpretation

The analysis of large structural databases reveals general features and relationships among proteins, providing useful insight. A different approach is required to characterize ubiquitous secondary-structure elements, where flexibility is essential in order to capture small local differences. The ALEPH software is optimized for the analysis and the extraction of small protein folds by relying on their geometry rather than on their sequence. The annotation of the structural variability of a given fold provides valuable information for fragment-based molecular-replacement methods, in which testing alternative model hypotheses can succeed in solving difficult structures when no homology models are available or are successful. ARCIMBOLDO_BORGES combines the use of composite secondary-structure elements as a search model with density modification and tracing to reveal the rest of the structure when both steps are successful. This phasing method relies on general fold libraries describing variations around a given pattern of β-sheets and helices extracted using ALEPH. The program introduces characteristic vectors defined from the main-chain atoms as a way to describe the geometrical properties of the structure. ALEPH encodes structural properties in a graph network, the exploration of which allows secondary-structure annotation, decomposition of a structure into small compact folds, generation of libraries of models representing a variation of a given fold and finally superposition of these folds onto a target structure. These functions are available through a graphical interface designed to interactively show the results of structure manipulation, annotation, fold decomposition, clustering and library generation. ALEPH can produce pictures of the graphs, structures and folds for publication purposes.

How far are we from automatic crystal structure solution via molecular-replacement techniques?

Although the success of molecular-replacement techniques requires the solution of a six-dimensional problem, this is often subdivided into two three-dimensional problems. REMO09 is one of the programs which have adopted this approach. It has been revisited in the light of a new probabilistic approach which is able to directly derive conditional distribution functions without passing through a previous calculation of the joint probability distributions. The conditional distributions take into account various types of prior information: in the rotation step the prior information may concern a non-oriented model molecule alone or together with one or more located model molecules. The formulae thus obtained are used to derive figures of merit for recognizing the correct orientation in the rotation step and the correct location in the translation step. The phases obtained by this new version of REMO09 are used as a starting point for a pipeline which in its first step extends and refines the molecular-replacement phases, and in its second step creates the final electron-density map which is automatically interpreted by CAB, an automatic model-building program for proteins and DNA/RNA structures.

SDBP-Pred: Prediction of single-stranded and double-stranded DNA-binding proteins by extending consensus sequence and K-segmentation strategies into PSSM

Identification of DNA-binding proteins (DNA-BPs) is a hot issue in protein science due to its key role in various biological processes. These processes are highly concerned with DNA-binding protein types. DNA-BPs are classified into single-stranded DNA-binding proteins (SSBs) and double-stranded DNA-binding proteins (DSBs). SSBs mainly involved in DNA recombination, replication, and repair, while DSBs regulate transcription process, DNA cleavage, and chromosome packaging. In spite of the aforementioned significance, few methods have been proposed for discrimination of SSBs and DSBs. Therefore, more predictors with favorable performance are indispensable. In this work, we present an innovative predictor, called SDBP-Pred with a novel feature descriptor, named consensus sequence-based K-segmentation position-specific scoring matrix (CSKS-PSSM). We encoded the local discriminative features concealed in PSSM via K-segmentation strategy and the global potential features by applying the notion of the consensus sequence. The obtained feature vector then input to support vector machine (SVM) with linear, polynomial and radial base function (RBF) kernels. Our model with SVM-RBF achieved the highest accuracies on three tests namely jackknife, 10-fold, and independent tests, respectively than the recent method. The obtained prediction results illustrate the superlative prediction performance of SDBP-Pred over existing studies in the literature so far.

DP-BINDER: machine learning model for prediction of DNA-binding proteins by fusing evolutionary and physicochemical information

DNA-binding proteins (DBPs) participate in various biological processes including DNA replication, recombination, and repair. In the human genome, about 6-7% of these proteins are utilized for genes encoding. DBPs shape the DNA into a compact structure known chromatin while some of these proteins regulate the chromosome packaging and transcription process. In the pharmaceutical industry, DBPs are used as a key component of antibiotics, steroids, and cancer drugs. These proteins also involve in biophysical, biological, and biochemical studies of DNA. Due to the crucial role in various biological activities, identification of DBPs is a hot issue in protein science. A series of experimental and computational methods have been proposed, however, some methods didn't achieve the desired results while some are inadequate in its accuracy and authenticity. Still, it is highly desired to present more intelligent computational predictors. In this work, we introduce an innovative computational method namely DP-BINDER based on physicochemical and evolutionary information. We captured local highly decisive features from physicochemical properties of primary protein sequences via normalized Moreau-Broto autocorrelation (NMBAC) and evolutionary information by position specific scoring matrix-transition probability composition (PSSM-TPC) and pseudo-position specific scoring matrix (PsePSSM) using training and independent datasets. The optimal features were selected by the support vector machine-recursive feature elimination and correlation bias reduction (SVM-RFE + CBR) from fused features and were fed into random forest (RF) and support vector machine (SVM). Our method attained 92.46% and 89.58% accuracy with jackknife and ten-fold cross-validation, respectively on the training dataset, while 81.17% accuracy on the independent dataset for prediction of DBPs. These results demonstrate that our method attained the highest success rate in the literature. The superiority of DP-BINDER over existing approaches due to several reasons including abstraction of local dominant features via effective feature descriptors, utilization of appropriate feature selection algorithms and effective classifier.

DNA-protein interaction: identification, prediction and data analysis

Life in living organisms is dependent on specific and purposeful interaction between other molecules. Such purposeful interactions make the various processes inside the cells and the bodies of living organisms possible. DNA-protein interactions, among all the types of interactions between different molecules, are of considerable importance. Currently, with the development of numerous experimental techniques, diverse methods are convenient for recognition and investigating such interactions. While the traditional experimental techniques to identify DNA-protein complexes are time-consuming and are unsuitable for genome-scale studies, the current high throughput approaches are more efficient in determining such interaction at a large-scale, but they are clearly too costly to be practice for daily applications. Hence, according to the availability of much information related to different biological sequences and clearing different dimensions of conditions in which such interactions are formed, with the developments related to the computer, mathematics, and statistics motivate scientists to develop bioinformatics tools for prediction the interaction site(s). Until now, there has been much progress in this field. In this review, the factors and conditions governing the interaction and the laboratory techniques for examining such interactions are addressed. In addition, developed bioinformatics tools are introduced and compared for this reason and, in the end, several suggestions are offered for the promotion of such tools in prediction with much more precision.

Ab initio structure solution of a proteolytic fragment using ARCIMBOLDO

Crystal structure determination requires solving the phase problem. This can be accomplished using ab initio direct methods for small molecules and macromolecules at resolutions higher than 1.2 Å, whereas macromolecular structure determination at lower resolution requires either molecular replacement using a homologous structure or experimental phases using a derivative such as covalent labeling (for example selenomethionine or mercury derivatization) or heavy-atom soaking (for example iodide ions). Here, a case is presented in which crystals were obtained from a 30.8 kDa protein sample and yielded a 1.6 Å resolution data set with a unit cell that could accommodate approximately 8 kDa of protein. Thus, it was unclear what had been crystallized. Molecular replacement with pieces of homologous proteins and attempts at iodide ion soaking failed to yield a solution. The crystals could not be reproduced. Sequence-independent molecular replacement using the structures available in the Protein Data Bank also failed to yield a solution. Ultimately, ab initio structure solution proved successful using the program ARCIMBOLDO, which identified two α-helical elements and yielded interpretable maps. The structure was the C-terminal dimerization domain of the intended target from Mycobacterium smegmatis. This structure is presented as a user-friendly test case in which an unknown protein fragment could be determined using ARCIMBOLDO.

The Effectors and Sensory Sites of Formaldehyde-responsive Regulator FrmR and Metal-sensing Variant

The DUF156 family of DNA-binding transcriptional regulators includes metal sensors that respond to cobalt and/or nickel (RcnR, InrS) or copper (CsoR) plus CstR, which responds to persulfide, and formaldehyde-responsive FrmR. Unexpectedly, the allosteric mechanism of FrmR from Salmonella enterica serovar Typhimurium is triggered by metals in vitro, and variant FrmR(E64H) gains responsiveness to Zn(II) and cobalt in vivo Here we establish that the allosteric mechanism of FrmR is triggered directly by formaldehyde in vitro Sensitivity to formaldehyde requires a cysteine (Cys(35) in FrmR) conserved in all DUF156 proteins. A crystal structure of metal- and formaldehyde-sensing FrmR(E64H) reveals that an FrmR-specific amino-terminal Pro(2) is proximal to Cys(35), and these residues form the deduced formaldehyde-sensing site. Evidence is presented that implies that residues spatially close to the conserved cysteine tune the sensitivities of DUF156 proteins above or below critical thresholds for different effectors, generating the semblance of specificity within cells. Relative to FrmR, RcnR is less responsive to formaldehyde in vitro, and RcnR does not sense formaldehyde in vivo, but reciprocal mutations FrmR(P2S) and RcnR(S2P), respectively, impair and enhance formaldehyde reactivity in vitro Formaldehyde detoxification by FrmA requires S-(hydroxymethyl)glutathione, yet glutathione inhibits formaldehyde detection by FrmR in vivo and in vitro Quantifying the number of FrmR molecules per cell and modeling formaldehyde modification as a function of [formaldehyde] demonstrates that FrmR reactivity is optimized such that FrmR is modified and frmRA is derepressed at lower [formaldehyde] than required to generate S-(hydroxymethyl)glutathione. Expression of FrmA is thereby coordinated with the accumulation of its substrate.

Combining phase information in reciprocal space for molecular replacement with partial models

ARCIMBOLDO allows ab initio phasing of macromolecular structures below atomic resolution by exploiting the location of small model fragments combined with density modification in a multisolution frame. The model fragments can be either secondary-structure elements predicted from the sequence or tertiary-structure fragments. The latter can be derived from libraries of typical local folds or from related structures, such as a low-homology model that is unsuccessful in molecular replacement. In all ARCIMBOLDO applications, fragments are searched for sequentially. Correct partial solutions obtained after each fragment-search stage but lacking the necessary phasing power can, if combined, succeed. Here, an analysis is presented of the clustering of partial solutions in reciprocal space and of its application to a set of different cases. In practice, the task of combining model fragments from an ARCIMBOLDO run requires their referral to a common origin and is complicated by the presence of correct and incorrect solutions as well as by their not being independent. The F-weighted mean phase difference has been used as a figure of merit. Clustering perfect, non-overlapping fragments dismembered from test structures in polar and nonpolar space groups shows that density modification before determining the relative origin shift enhances its discrimination. In the case of nonpolar space groups, clustering of ARCIMBOLDO solutions from secondary-structure models is feasible. The use of partially overlapping search fragments provides a more favourable circumstance and was assessed on a test case. Applying the devised strategy, a previously unknown structure was solved from clustered correct partial solutions.

ARCIMBOLDO_LITE: single-workstation implementation and use

ARCIMBOLDO solves the phase problem at resolutions of around 2 Å or better through massive combination of small fragments and density modification. For complex structures, this imposes a need for a powerful grid where calculations can be distributed, but for structures with up to 200 amino acids in the asymmetric unit a single workstation may suffice. The use and performance of the single-workstation implementation, ARCIMBOLDO_LITE, on a pool of test structures with 40-120 amino acids and resolutions between 0.54 and 2.2 Å is described. Inbuilt polyalanine helices and iron cofactors are used as search fragments. ARCIMBOLDO_BORGES can also run on a single workstation to solve structures in this test set using precomputed libraries of local folds. The results of this study have been incorporated into an automated, resolution- and hardware-dependent parameterization. ARCIMBOLDO has been thoroughly rewritten and three binaries are now available: ARCIMBOLDO_LITE, ARCIMBOLDO_SHREDDER and ARCIMBOLDO_BORGES. The programs and libraries can be downloaded from http://chango.ibmb.csic.es/ARCIMBOLDO_LITE.

Structure of a 13-fold superhelix (almost) determined from first principles

Nuclear hormone receptors are cytoplasm-based transcription factors that bind a ligand, translate to the nucleus and initiate gene transcription in complex with a co-activator such as TIF2 (transcriptional intermediary factor 2). For structural studies the co-activator is usually mimicked by a peptide of circa 13 residues, which for the largest part forms an α-helix when bound to the receptor. The aim was to co-crystallize the glucocorticoid receptor in complex with a ligand and the TIF2 co-activator peptide. The 1.82 Å resolution diffraction data obtained from the crystal could not be phased by molecular replacement using the known receptor structures. HPLC analysis of the crystals revealed the absence of the receptor and indicated that only the co-activator peptide was present. The self-rotation function displayed 13-fold rotational symmetry, which initiated an exhaustive but unsuccessful molecular-replacement approach using motifs of 13-fold symmetry such as α- and β-barrels in various geometries. The structure was ultimately determined by using a single α-helix and the software ARCIMBOLDO, which assembles fragments placed by PHASER before using them as seeds for density modification model building in SHELXE. Systematic variation of the helix length revealed upper and lower size limits for successful structure determination. A beautiful but unanticipated structure was obtained that forms superhelices with left-handed twist throughout the crystal, stabilized by ligand interactions. Together with the increasing diversity of structural elements in the Protein Data Bank the results from TIF2 confirm the potential of fragment-based molecular replacement to significantly accelerate the phasing step for native diffraction data at around 2 Å resolution.

A fragmentation and reassembly method for ab initio phasing

Ab initio phasing with de novo models has become a viable approach for structural solution from protein crystallographic diffraction data. This approach takes advantage of the known protein sequence information, predicts de novo models and uses them for structure determination by molecular replacement. However, even the current state-of-the-art de novo modelling method has a limit as to the accuracy of the model predicted, which is sometimes insufficient to be used as a template for successful molecular replacement. A fragment-assembly phasing method has been developed that starts from an ensemble of low-accuracy de novo models, disassembles them into fragments, places them independently in the crystallographic unit cell by molecular replacement and then reassembles them into a whole structure that can provide sufficient phase information to enable complete structure determination by automated model building. Tests on ten protein targets showed that the method could solve structures for eight of these targets, although the predicted de novo models cannot be used as templates for successful molecular replacement since the best model for each target is on average more than 4.0 Å away from the native structure. The method has extended the applicability of the ab initio phasing by de novo models approach. The method can be used to solve structures when the best de novo models are still of low accuracy.

Macromolecular ab initio phasing enforcing secondary and tertiary structure

Ab initio phasing of macromolecular structures, from the native intensities alone with no experimental phase information or previous particular structural knowledge, has been the object of a long quest, limited by two main barriers: structure size and resolution of the data. Current approaches to extend the scope of ab initio phasing include use of the Patterson function, density modification and data extrapolation. The authors' approach relies on the combination of locating model fragments such as polyalanine α-helices with the program PHASER and density modification with the program SHELXE. Given the difficulties in discriminating correct small substructures, many putative groups of fragments have to be tested in parallel; thus calculations are performed in a grid or supercomputer. The method has been named after the Italian painter Arcimboldo, who used to compose portraits out of fruit and vegetables. With ARCIMBOLDO, most collections of fragments remain a 'still-life', but some are correct enough for density modification and main-chain tracing to reveal the protein's true portrait. Beyond α-helices, other fragments can be exploited in an analogous way: libraries of helices with modelled side chains, β-strands, predictable fragments such as DNA-binding folds or fragments selected from distant homologues up to libraries of small local folds that are used to enforce nonspecific tertiary structure; thus restoring the ab initio nature of the method. Using these methods, a number of unknown macromolecules with a few thousand atoms and resolutions around 2 Å have been solved. In the 2014 release, use of the program has been simplified. The software mediates the use of massive computing to automate the grid access required in difficult cases but may also run on a single multicore workstation (http://chango.ibmb.csic.es/ARCIMBOLDO_LITE) to solve straightforward cases.