CRYSTALP2: sequence-based protein crystallization propensity prediction

Background  

Current protocols yield crystals for <30% of known proteins, indicating that automatically identifying crystallizable proteins may improve high-throughput structural genomics efforts. We introduce CRYSTALP2, a kernel-based method that predicts the propensity of a given protein sequence to produce diffraction-quality crystals. This method utilizes the composition and collocation of amino acids, isoelectric point, and hydrophobicity, as estimated from the primary sequence, to generate predictions. CRYSTALP2 extends its predecessor, CRYSTALP, by enabling predictions for sequences of unrestricted size and provides improved prediction quality.     

Sequence-based prediction of protein solubility

In order to investigate the relationship between the thermodynamics and kinetics of protein aggregation, we compared the solubility of proteins with their aggregation rates. We found a significant correlation between these two quantities by considering a database of protein solubility values measured using an in vitro reconstituted translation system containing about 70% of Escherichia coli proteins. The existence of such correlation suggests that the thermodynamic stability of the native states of proteins relative to the aggregate states is closely linked with the kinetic barriers that separate them. In order to create the possibility of conducting computational studies at the proteome level to investigate further this concept, we developed a method of predicting the solubility of proteins based on their physicochemical properties.     

Sequence-based prediction of protein domains

Guessing the boundaries of structural domains has been an important and challenging problem in experimental and computational structural biology. Predictions were based on intuition, biochemical properties, statistics, sequence homology and other aspects of predicted protein structure. Here, we introduced CHOPnet, a de novo method that predicts structural domains in the absence of homology to known domains. Our method was based on neural networks and relied exclusively on information available for all proteins. Evaluating sustained performance through rigorous cross-validation on proteins of known structure, we correctly predicted the number of domains in 69% of all proteins. For 50% of the two-domain proteins the centre of the predicted boundary was closer than 20 residues to the boundary assigned from three-dimensional (3D) structures; this was about eight percentage points better than predictions by ‘equal split’. Our results appeared to compare favourably with those from previously published methods. CHOPnet may be useful to restrict the experimental testing of different fragments for structure determination in the context of structural genomics.     

CRYSpred: accurate sequence-based protein crystallization propensity prediction using sequence-derived structural characteristics

Relatively low success rates of X-ray crystallography, which is the most popular method for solving proteins structures, motivate development of novel methods that support selection of tractable protein targets. This aspect is particularly important in the context of the current structural genomics efforts that allow for a certain degree of flexibility in the target selection. We propose CRYSpred, a novel in-silico crystallization propensity predictor that uses a set of 15 novel features which utilize a broad range of inputs including charge, hydrophobicity, and amino acid composition derived from the protein chain, and the solvent accessibility and disorder predicted from the protein sequence. Our method outperforms seven modern crystallization propensity predictors on three, independent from training dataset, benchmark test datasets. The strong predictive performance offered by the CRYSpred is attributed to the careful design of the features, utilization of the comprehensive set of inputs, and the usage of the Support Vector Machine classifier. The inputs utilized by CRYSpred are well-aligned with the existing rules-of-thumb that are used in the structural genomics studies.     

Predicting protein crystallization propensity from protein sequence

The high-throughput structure determination pipelines developed by structural genomics programs offer a unique opportunity for data mining. One important question is how protein properties derived from a primary sequence correlate with the protein’s propensity to yield X-ray quality crystals (crystallizability) and 3D X-ray structures. A set of protein properties were computed for over 1,300 proteins that expressed well but were insoluble, and for ~720 unique proteins that resulted in X-ray structures. The correlation of the protein’s iso-electric point and grand average hydropathy (GRAVY) with crystallizability was analyzed for full length and domain constructs of protein targets. In a second step, several additional properties that can be calculated from the protein sequence were added and evaluated. Using statistical analyses we have identified a set of the attributes correlating with a protein’s propensity to crystallize and implemented a Support Vector Machine (SVM) classifier based on these. We have created applications to analyze and provide optimal boundary information for query sequences and to visualize the data. These tools are available via the web site .     

ParCrys: a Parzen window density estimation approach to protein crystallization propensity prediction

The ability to rank proteins by their likely success in crystallizationis useful in current Structural Biology efforts and in particularin high-throughput Structural Genomics initiatives. We presentParCrys, a Parzen Window approach to estimate a protein's propensityto produce diffraction-quality crystals. The Protein Data Bank(PDB) provided training data whilst the databases TargetDB andPepcDB were used to define feature selection data as well astest data independent of feature selection and training. ParCrysoutperforms the OB-Score, SECRET and CRYSTALP on the data examined,with accuracy and Matthews correlation coefficient values of79.1% and 0.582, respectively (74.0% and 0.227, respectively,on data with a ‘real-world’ ratio of positive:negativeexamples). ParCrys predictions and associated data are availablefrom www.compbio.dundee.ac.uk/parcrys. Contact: geoff{at}compbio.dundee.ac.uk Supplementary information: Supplementary data are availableat Bioinformatics online. Associate Editor: John Quackenbush     

Human recombinant laminin-binding protein: isolation, purification, and crystallization

The mRNA of the precursor of laminin-binding protein (LBP) was isolated from a human embryo kidney cell line and cloned. The determined sequence of the LBP gene showed complete identity with the LBP genes isolated from human lung and large intestine cells. The human LBP was expressed by E. coli cells, and it was purified using Ni-NTA-Sepharose chromatography. The mobility of the homogeneous recombinant human laminin-binding protein on SDS-PAGE was 43 kD. A mixture of eight murine monoclonal antibodies, the MPLR Pool against LBP, reacted with the recombinant LBP in Western blot. The interaction of the antiidiotypical antibodies 10H10 and E6B provided evidence that the epitope binding to protein E of the tick-borne encephalitis (TBE) virus is also preserved on the human recombinant LBP. Enzyme immunoassay confirmed the ability of the recombinant LBP to interact with protein E of TBE virus. The biological activity of the recombinant LBP allowed us to perform X-ray analysis of the spatial arrangement of the LBP molecule using the recombinant protein. For this purpose, crystals of the human LBP were obtained by the standing drop version of the pore diffusion technique. The crystals appropriate for X-ray structural analysis were 0.3 x 0.1 x 0.05 mm in size. The X-ray diffraction field of the crystal extended to 2.5 A.     

Sequence-based prediction of pathological mutations

The development of methods to assess the impact of amino acid mutations on human health has become an important goal in biomedical research, due to the growing number of nonsynonymous SNPs identified. Within this context, computational methods constitute a valuable tool, because they can easily process large amounts of mutations and give useful, almost cost-free, information on their pathological character. In this paper we present a computational approach to the prediction of disease-associated amino acid mutations, using only sequence-based information (amino acid properties, evolutionary information, secondary structure and accessibility predictions, and database annotations) and neural networks, as a model building tool. Mutations are predicted to be either pathological or neutral. Our results show that the method has a good overall success rate, 83%, that can reach 95% when trained for specific proteins. The methodology is fast and flexible enough to provide good estimates of the pathological character of large sets of nonsynonymous SNPs, but can also be easily adapted to give more precise predictions for proteins of special biomedical interest.     

High-throughput protein purification and quality assessment for crystallization

The ultimate goal of structural biology is to understand the structural basis of proteins in cellular processes. In structural biology, the most critical issue is the availability of high-quality samples. "Structural biology-grade" proteins must be generated in the quantity and quality suitable for structure determination using X-ray crystallography or nuclear magnetic resonance (NMR) spectroscopy. The purification procedures must reproducibly yield homogeneous proteins or their derivatives containing marker atom(s) in milligram quantities. The choice of protein purification and handling procedures plays a critical role in obtaining high-quality protein samples. With structural genomics emphasizing a genome-based approach in understanding protein structure and function, a number of unique structures covering most of the protein folding space have been determined and new technologies with high efficiency have been developed. At the Midwest Center for Structural Genomics (MCSG), we have developed semi-automated protocols for high-throughput parallel protein expression and purification. A protein, expressed as a fusion with a cleavable affinity tag, is purified in two consecutive immobilized metal affinity chromatography (IMAC) steps: (i) the first step is an IMAC coupled with buffer-exchange, or size exclusion chromatography (IMAC-I), followed by the cleavage of the affinity tag using the highly specific Tobacco Etch Virus (TEV) protease; the second step is IMAC and buffer exchange (IMAC-II) to remove the cleaved tag and tagged TEV protease. These protocols have been implemented on multidimensional chromatography workstations and, as we have shown, many proteins can be successfully produced in large-scale. All methods and protocols used for purification, some developed by MCSG, others adopted and integrated into the MCSG purification pipeline and more recently the Center for Structural Genomics of Infectious Diseases (CSGID) purification pipeline, are discussed in this chapter.     

Expression, purification, and crystallization of the adipocyte lipid binding protein.

The murine adipocyte lipid binding protein (ALBP/aP2) has been cloned and expressed in Escherichia coli, purified to homogeneity, biochemically characterized, and crystallized for x-ray diffraction study. In the cloning, the ALBP coding region was placed under control of the recA promoter and downstream of the phage T7 g-10 translation enhancer sequence. Nalidixic acid (50 micrograms/ml) induced the expression of ALBP 20-fold over that attained using the pT7 system previously reported (Chinander, L. L., and Bernlohr, D. A. (1989) J. Biol. Chem. 264, 19564-19572). Recombinant ALBP was purified to homogeneity using a combination of pH fractionation, gel filtration, and immobilized metal affinity chromatography. The fluorescent affinity ligand 12-(9-anthroyloxy)oleic acid bound to homogeneous ALBP with an apparent Kd of 0.5 microM. rALBP was devoid of endogenous fatty acid, and oleic acid inhibited cysteine 117 modification by 5,5' -dithiobis-(2-nitrobenzoic acid) indicating integrity of the binding domain. Recombinant ALBP was phosphorylated by the soluble kinase domain of the insulin receptor with a Vmax of 11 nmol.min.mg of kinase and an apparent Km of 270 microM. Purified protein was crystallized using the hanging drop method with seeding. Crystalline ALBP was orthorhombic with cell dimensions of a = 34.4 A, b = 54.8 A, and c = 76.3 A. The space group was P212121, and there was one molecule per asymmetric unit.     

Sequence-based feature prediction and annotation of proteins

A recent trend in computational methods for annotation of protein function is that many prediction tools are combined in complex workflows and pipelines to facilitate the analysis of feature combinations, for example, the entire repertoire of kinase-binding motifs in the human proteome.     

Renaissance of protein crystallization and precipitation in biopharmaceuticals purification

The current chromatographic approaches used in protein purification are not keeping pace with the increasing biopharmaceutical market demand. With the upstream improvements, the bottleneck shifted towards the downstream process. New approaches rely in Anything But Chromatography methodologies and revisiting former techniques with a bioprocess perspective. Protein crystallization and precipitation methods are already implemented in the downstream process of diverse therapeutic biological macromolecules, overcoming the current chromatographic bottlenecks. Promising work is being developed in order to implement crystallization and precipitation in the purification pipeline of high value therapeutic molecules. This review focuses in the role of these two methodologies in current industrial purification processes, and highlights their potential implementation in the purification pipeline of high value therapeutic molecules, overcoming chromatographic holdups.     

Expression, purification, crystallization and preliminary x-ray analysis of Cyan fluorescent protein CyPet

The technique of fluorescence (or F?rster) resonance energy transfer (FRET) is widely used to observe bimolecular interaction in living cells. Cyan and yellow fluorescent proteins are the most widely used pair in FRET analysis. CyPet and YPet are two newly optimized fluorescent proteins that have much better dynamic range and sensitivity than CFP/YFP pair, although the crystallographic structure and the mechanism of better fluorescent characteristics of CyPet are still unknown. We have expressed the cyan fluorescent protein CyPet using pT7 prokaryocyte expression system in Escherichia coli strain Rosetta (DE3) pLysS by auto-induction. After purification, the recombinant CyPet protein was crystallized by hanging drop vapor diffusion technique and could diffract to 2.55A resolution. The data showed that the orthorhombic CyPet crystal was in space group P212121 with unit cell parameters (51.55, 61.53, 63.36) and contained one molecule in one asymmetric unit.     

A simple strategy towards membrane protein purification and crystallization

A simple and cost-efficient detergent screening strategy has been developed, by which a number of detergents were screened for their efficiency to extract and purify the recombinant ammonium/ammonia channel, AmtB, from Escherichia coli, hence selecting the most efficient detergents prior to large-scale protein production and crystallization. The method requires 1 ml cell culture and is a combination of immobilized metal ion affinity chromatography and filtration steps in 96-well plates. Large-scale protein purification and subsequent crystallization screening resulted in AmtB crystals diffracting to low resolution with three detergents. This strategy allows exclusion of detergents with the lowest probability in yielding protein crystals and selecting those with higher probability, hence, reducing the number of detergents to be screened prior to large-scale membrane protein purification and perhaps also crystallization.     

Expression, purification and crystallization of cysteine-rich human protein muskelin in Escherichia coli

The increasing interest in the structural arrangements and functional interdependencies of individual modules within large multidomain proteins requires the development of new methods allowing efficient production and purification of large human proteins. Heterologous expression in bacteria is still the most convenient and widely-used approach. However, most of the existing tools are not well suited to expression of cysteine-rich proteins in a native-like soluble form, and with the increasing protein size refolding may result in obtaining non-native conformations or improper disulfide bridging pattern. Here, we present an efficient method of expression and purification of muskelin, a large, multidomain, cysteine-rich eukaryotic protein involved in cell adhesion and regulation of cytoskeleton dynamics. Using a broad range of purification and solubility tags, expression strains and conditions we optimized the procedure to acquire a natively folded protein of crystallization-scale quantity and purity. The correct protein conformation and disulfide bonding were anticipated from the results of circular dichroism spectra and Ellman’s assay. Successful crystallization trials are a step towards muskelin crystal-structure determination, while the optimized expression and purification procedure can easily be applied to produce other eukaryotic proteins in the bacterial expression system.     

Expression, purification and crystallization of the human UL16-binding protein ULBP1

UL16-binding proteins (ULBPs) are markers of cellular stress which are upregulated on the surface of virus-infected and tumor cells. Recognition of ULBP1 by the activating receptor NKG2D on the surface of cytotoxic natural killer (NK) and T cells promotes lysis of cells expressing ULBP1 and is an important mechanism of immune surveillance. We report a robust method for the generation of large quantities of crystal-grade recombinant ULBP1 protein. The extracellular portion of human ULBP1 was cloned into a T7 expression vector for expression in Escherichia coli. Unpaired cysteines in the sequence which are predicted not to be involved in the intramolecular disulfide bond formation were mutated to serine. ULBP1 was expressed in E. coli BL21 (DE3) pLysS cells as inclusion bodies. Purified inclusion bodies were solubilized by denaturation in guanidine, and refolded by slow dilution. The refolded protein was purified by size exclusion gel filtration and anion exchange chromatography. Furthermore, we have identified conditions optimal for the crystallization of this protein and have obtained initial diffraction data to 4.6? from these crystals.     

High-level expression, purification, kinetic characterization and crystallization of protein farnesyltransferase beta-subunit C-terminal mutants.

Protein farnesyltransferase (FPT) is a 97 000 Da heterodimeric enzyme that catalyzes post-translational farnesylation of many cellular regulatory proteins including p21 Ras. To facilitate the construction of site-directed mutants, a novel translationally coupled, two-cistron Escherichia coli expression system for rat FPT has been developed. This expression system enabled yields of >5 mg of purified protein per liter of E.coli culture to be obtained. The E.coli-derived FPT demonstrated an activity comparable to that of protein isolated from other sources. The reported expression system was used to construct three beta-subunit C-terminal truncation mutants, Delta5, Delta10 and Delta14, which were designed to eliminate a lattice interaction between the beta-subunit C-terminus of one molecule and the active site of a symmetry-related molecule. Steady-state kinetic analyses of these mutants showed that deletion up to 14 residues at the C-terminus did not reduce the value of kcat; however, Km values for both peptide and FPP increased 2-3-fold. A new crystalline form of FPT was obtained for the Delta10 C-terminal mutant grown in the presence of the substrate analogs acetyl-Cys-Val-Ile-Met-COOH peptide and alpha-hydroxyfarnesylphosphonic acid. The crystals diffract to beyond 2.0 A resolution. The refined structure clearly shows that both substrate analogs adopt extended conformations within the FPT active site cavity.     

Expression, purification, and crystallization of the Escherichia coli selenomethionyl beta-ketoacyl-acyl carrier protein synthase III

Bacterial beta-ketoacyl-acyl carrier protein (ACP) synthase III (KAS III, also called FabH) catalyzes the condensation and transacylation of acetyl-CoA with malonyl-ACP. In order to understand the mode of enzyme/substrate interaction and design small molecule inhibitors, we have expressed, purified, and crystallized a selenomethionyl-derivative of E. coli KAS III. Several lines of evidence confirmed that purified selenomethionyl KAS III was homogenous, stably folded, and enzymatically active. Dynamic light scattering, size exclusion chromatography, and mass spectrometry results indicated that selenomethionyl KAS III is a noncovalent homodimer. Diffraction quality crystals of selenomethionyl KAS III/acetyl-CoA complex, which grew overnight to a size of 0.2 mm(3), belonged to the tetragonal space group P4(1)2(1)2.