Automatic Classification and Pattern Discovery in High-throughput Protein Crystallization Trials

Conceptually, protein crystallization can be divided into two phases search and optimization. Robotic protein crystallization screening can speed up the search phase, and has a potential to increase process quality. Automated image classification helps to increase throughput and consistently generate objective results. Although the classification accuracy can always be improved, our image analysis system can classify images from 1536-well plates with high classification accuracy (85%) and ROC score (0.87), as evaluated on 127 human-classified protein screens containing 5600 crystal images and 189472 non-crystal images. Data mining can integrate results from high-throughput screens with information about crystallizing conditions, intrinsic protein properties, and results from crystallization optimization. We apply association mining, a data mining approach that identifies frequently occurring patterns among variables and their values. This approach segregates proteins into groups based on how they react in a broad range of conditions, and clusters cocktails to reflect their potential to achieve crystallization. These results may lead to crystallization screen optimization, and reveal associations between protein properties and crystallization conditions. We also postulate that past experience may lead us to the identification of initial conditions favorable to crystallization for novel proteins.     

Efficient protein crystallization

High-throughput molecular biology and crystallography advances have placed an increasing demand on crystallization, the one remaining bottleneck in macromolecular crystallography. This paper describes three experimental approaches, an incomplete factorial crystallization screen, a high-throughput nanoliter crystallization system, and the use of a neural net to predict crystallization conditions via a small sample (approximately 0.1%) of screening results. The use of these technologies has the potential to reduce time and sample requirements. Initial experimental results indicate that the incomplete factorial design detects initial crystallization conditions not previously discovered using commercial screens. This may be due to the ability of the incomplete factorial screen to sample a broader portion of "crystallization space," using a multidimensional set of components, concentrations, and physical conditions. The incomplete factorial screen is complemented by a neural network program used to model crystallization. This capability is used to help predict new crystallization conditions. An automated, nanoliter crystallization system, with a throughput of up to 400 conditions/h in 40-nl droplets (total volume), accommodates microbatch or traditional "sitting-drop" vapor diffusion experiments. The goal of this research is to develop a fully-automated high-throughput crystallization system that integrates incomplete factorial screen and neural net capabilities.     

Protein crystallization by capillary counterdiffusion for applied crystallographic structure determination

Counterdiffusion crystallization in capillary is a very simple, cost-effective, and practical procedure for obtaining protein crystals suitable for X-ray data analysis. Its principles have been derived using well-known concepts coupling the ideas of precipitation and diffusion mass transport in a restricted geometry. The counterdiffusion process has been used to simultaneously screen for optimal conditions for protein crystal growth, incorporate strong anomalous scattering atoms, and mix in cryogenic solutions in a single capillary tube. The crystals obtained in the capillary have been used in situ for X-ray analysis. The implementation of this technique linked to the advancement of current crystallography software leads to a powerful structure determination method consolidating crystal growth, X-ray data collection, and ab initio phase determination into one without crystal manipulation. We review the historical progress of counterdiffusion crystallization, its application to X-ray crystallography, and ongoing tool development for high-throughput protein structure determination.     

Characterization of lipid matrices for membrane protein crystallization by high-throughput small angle X-ray scattering

The lipidic cubic phase (LCP) has repeatedly proven to serve as a successful membrane-mimetic matrix for a variety of difficult-to-crystallize membrane proteins. While monoolein has been the predominant lipid of choice, there is a growing need for the characterization and use of other LCP host lipids, allowing exploration of a range of structural parameters such as bilayer thickness and curvature for optimal insertion, stability and crystallogenesis of membrane proteins. Here, we describe the development of a high-throughput (HT) pipeline to employ small angle X-ray scattering (SAXS) - the most direct technique to identify lipid mesophases and measure their structural parameters - to interrogate rapidly a large number of lipid samples under a variety of conditions, similar to those encountered during crystallization. Leveraging the identical setup format for LCP crystallization trials, this method allows the quickly assessment of lipid matrices for their utility in membrane protein crystallization, and could inform the tailoring of lipid and precipitant conditions to overcome specific crystallization challenges. As proof of concept, we present HT LCP-SAXS analysis of lipid samples made of monoolein with and without cholesterol, and of monovaccenin, equilibrated with solutions used for crystallization trials and LCP fluorescence recovery after photobleaching (FRAP) experiments.     

A deliberate approach to screening for initial crystallization conditions of biological macromolecules

A method to rationally predict crystallization conditions for a previously uncrystallized macromolecule has not yet been developed. One way around this problem is to determine initial crystallization conditions by casting a wide net, surveying a large number of chemical and physical conditions to locate crystallization leads. A facility that executes the rapid survey of crystallization lead conditions is described in detail. Results and guidelines for the initial screening of crystallization conditions, applicable to both manual and robotic setups, are discussed.     

Thermofluor-based high-throughput stability optimization of proteins for structural studies

Production of proteins well suited for structural studies is inherently difficult and time-consuming. Protein sample homogeneity, stability, and solubility are strongly correlated with the proteins' probability of yielding crystals, and optimization of these properties will improve success rates of crystallization. In the current study, we applied the thermofluor method as a high-throughput approach for identifying optimal protein formulation for crystallization. The method also allowed optimal stabilizing buffer compositions to be rapidly identified for each protein. Furthermore, the method allowed the identification of potential ligands, physiological or non-physiological, that can be used in subsequent crystallization trials. For this study, the thermally induced melting points were determined in different buffers as well as with additives for a total of 25 Escherichia coli proteins. Crystallization trials were set up together with stabilizing and destabilizing additives identified using thermofluor screening. A twofold increase in the number of crystallization leads was observed when the proteins were cocrystallized with stabilizing additives as compared with experiments without these additives. This suggests that thermofluor constitutes an efficient generic high-throughput method for identification of protein properties predictive of crystallizability.     

Protein crystallization: virtual screening and optimization

Advances in genomics have yielded entire genetic sequences for a variety of prokaryotic and eukaryotic organisms. This accumulating information has escalated the demands for three-dimensional protein structure determinations. As a result, high-throughput structural genomics has become a major international research focus. This effort has already led to several significant improvements in X-ray crystallographic and nuclear magnetic resonance methodologies. Crystallography is currently the major contributor to three-dimensional protein structure information. However, the production of soluble, purified protein and diffraction-quality crystals are clearly the major roadblocks preventing the realization of high-throughput structure determination.

This paper discusses a novel approach that may improve the efficiency and success rate for protein crystallization. An automated nanodispensing system is used to rapidly prepare crystallization conditions using minimal sample. Proteins are subjected to an incomplete factorial screen (balanced parameter screen), thereby efficiently searching the entire “crystallization space” for suitable conditions. The screen conditions and scored experimental results are subsequently analyzed using a neural network algorithm to predict new conditions likely to yield improved crystals. Results based on a small number of proteins suggest that the combination of a balanced incomplete factorial screen and neural network analysis may provide an efficient method for producing diffraction-quality protein crystals.     

Systematic Improvement of Protein Crystals by Determining the Supersolubility Curves of Phase Diagrams

A systematic approach for improving protein crystals by growing them in the metastable zone using the vapor diffusion technique is described. This is a simple technique for optimization of crystallization conditions. Screening around known conditions is performed to establish a working phase diagram for the crystallization of the protein. Dilutions of the crystallization drops across the supersolubility curve into the metastable zone are then carried out as follows: the coverslips holding the hanging drops are transferred, after being incubated for some time at conditions normally giving many small crystals, over reservoirs at concentrations which normally yield clear drops. Fewer, much larger crystals are obtained when the incubation times are optimized, compared with conventional crystallization at similar conditions. This systematic approach has led to the structure determination of the light-harvesting protein C-phycocyanin to the highest-ever resolution of 1.45 Å.     

Current trends in α-helical membrane protein crystallization: An update

α-Helical membrane proteins (MPs) are the targets for many pharmaceutical drugs and play important roles in human physiology. In recent years, significant progress has been made in determining their atomic structure using X-ray crystallography. However, a major bottleneck in MP crystallography still remains, namely, the identification of conditions that give crystals that are suitable for structural determination. In 2008, we undertook an analysis of the crystallization conditions for 121 α-helical MPs to design a rationalized sparse matrix crystallization screen, MemGold. We now report an updated analysis that includes a further 133 conditions. The results reveal the current trends in α-helical MP crystallization with notable differences since 2008. The updated information has been used to design new crystallization and additive screens that should prove useful for both initial crystallization scouting and subsequent crystal optimization.     

Practical aspects of using the microbatch method in screening conditions for protein crystallization

The microbatch technique is a simple and efficient method for screening for protein crystallization conditions both by hand and using automated systems. Many of the problems associated with more commonly used methods such as vapour diffusion can be overcome using this method. Despite its promise microbatch has not been widely utilized as a viable screening procedure. This review aims to describe the method from a practical point of view with an emphasis on screening crystallization conditions, outlining current progress and discussion of the advantages compared to other methods.     

Managing and mining protein crystallization data

The crystallization of macromolecules remains a major bottleneck in structural biology. The routine screening of more than one thousand crystallization conditions and subsequent optimization by fine screening presents a challenge to conventional laboratory notebook keeping. In addition, the development of high-throughput robotic crystallization and imaging systems presents a pressing need for low-cost laboratory information management system (LIMS). Here we describe CLIMS2, a crystallization LIMS that features a simple, user-friendly graphical interface, allowing the storage, management, retrieval and mining of crystallization data. The CLIMS2 executable and documentation is freely available at http://clims.med.monash.edu.au.     

Lysine methylation as a routine rescue strategy for protein crystallization

Crystallization remains a critical step in X-ray structure determination. Because it is not generally possible to rationally predict crystallization conditions, commercial screens have been developed which sample a wide range of crystallization space. While this approach has proved successful in many cases, a significant number of proteins fail to crystallize despite being soluble and monodispersed. It is established that chemical modification can facilitate the crystallization of otherwise intractable proteins. Here we describe a method for the reductive methylation of lysine residues which is simple, inexpensive, and efficient, and report on its application to ten proteins. We describe the effect of methylation on the physico-chemical properties of these proteins, and show that it led to diffraction-quality crystals from four proteins and structures for three that had hitherto proved refractory to crystallization. The method is suited to both low- and high-throughput laboratories.     

Quantitative Modeling of the High-Throughput Production and In Vivo Kinetics of (Drug-Encapsulating) Liposomes

In developing liposomes for in vivo use, it is important to design the liposomes to have optimal in vivo kinetics, and it is also necessary to identify optimal high-throughput production conditions for these liposomes. Previous work has not definitively established the general relationship between liposomes'' configuration and composition, and their in vivo kinetics. Also, no straightforward method exists to calculate optimal liposome high-throughput production conditions for specific liposome compositions. This work presents first-principles quantitative correlations describing liposomes'' in vivo drug leakage and vascular mass transfer kinetics. This work further presents a simple quantitative model relating specific liposome compositions to ideal high-throughput production parameters. The results have implications for the identification of promising liposome compositions via high-throughput screening methodologies, as well as the design and optimization of high-throughput reactors for liposome production.     

xtalPiMS: a PiMS-based web application for the management and monitoring of crystallization trials

A major advance in protein structure determination has been the advent of nanolitre-scale crystallization and (in a high-throughput environment) the development of robotic systems for storing and imaging crystallization trials. Most of these trials are carried out in 96-well (or higher density) plates and managing them is a significant information management challenge. We describe xtalPiMS, a web-based application for the management and monitoring of crystallization trials. xtalPiMS has a user-interface layer based on the standards of the Protein Information Management System (PiMS) and a database layer which links the crystallization trial images to the meta-data associated with a particular crystallization trial. The user interface has been optimized for the efficient monitoring of high-throughput environments with three different automated imagers and work to support a fourth imager is in progress, but it can even be of use without robotics. The database can either be a PiMS database or a legacy database for which a suitable mapping layer has been developed.     

The promise of macromolecular crystallization in microfluidic chips

Microfluidics, or lab-on-a-chip technology, is proving to be a powerful, rapid, and efficient approach to a wide variety of bioanalytical and microscale biopreparative needs. The low materials consumption, combined with the potential for packing a large number of experiments in a few cubic centimeters, makes it an attractive technique for both initial screening and subsequent optimization of macromolecular crystallization conditions. Screening operations, which require a macromolecule solution with a standard set of premixed solutions, are relatively straightforward and have been successfully demonstrated in a microfluidics platform. Optimization methods, in which crystallization solutions are independently formulated from a range of stock solutions, are considerably more complex and have yet to be demonstrated. To be competitive with either approach, a microfluidics system must offer ease of operation, be able to maintain a sealed environment over several weeks to months, and give ready access for the observation and harvesting of crystals as they are grown.     

Improvement of the crystallizability and expression of an RNA crystallization chaperone

Crystallizing RNA has been an imperative and challenging task in the world of RNA research. Assistive methods such as chaperone-assisted RNA crystallography (CARC), employing monoclonal antibody fragments (Fabs) as crystallization chaperones have enabled us to obtain RNA crystal structures by forming crystal contacts and providing initial phasing information. Despite the early successes, the crystallization of large RNA-Fab complex remains a challenge in practice. The possible reason for this difficulty is that the Fab scaffold has not been optimized for crystallization in complex with RNA. Here, we have used the surface entropy reduction (SER) technique for the optimization of ΔC209 P4-P6/Fab2 model system. Protruding lysine and glutamate residues were mutated to a set of alanines or serines to construct Fab2SMA or Fab2SMS. Expression with the shake flask approach was optimized to allow large scale production for crystallization. Crystal screening shows that significantly higher crystal-forming ratio was observed for the mutant complexes. As the chosen SER residues are far away from the CDR regions of the Fab, the same set of mutations can now be directly applied to other Fabs binding to a variety of ribozymes and riboswitches to improve the crystallizability of Fab-RNA complex.     

Predictive crystallization of ribonuclease A via rapid screening of osmotic second virial coefficients

Important progress has been made in recent years toward developing a molecular-level understanding of protein phase behavior in terms of the osmotic second virial coefficient, a thermodynamic parameter that characterizes pairwise protein interactions. Yet there has been little practical application of this knowledge to the field of protein crystallization, largely because of the difficult and time-consuming nature of traditional techniques for characterizing protein interactions. Self-interaction chromatography has recently been proposed as a highly efficient method for measuring the osmotic second virial coefficient. The utility of the technique is examined in this work by characterizing virial coefficients for ribonuclease A under 59 solution conditions using several crystallization additives, including PEG, sodium chloride, ammonium sulfate, and propanol. The virial coefficient measurements show some counterintuitive trends and shed light on the previous difficulties in crystallizing ribonuclease A. Crystallization experiments at the corresponding solution conditions were conducted by using ultracentrifugal crystallization. Using this methodology, ribonuclease A crystals were obtained under conditions for which the virial coefficients fell within the "crystallization slot." Crystallographic characterization showed that the crystals diffract to high resolution. Metastable crystals were also obtained for conditions outside, but near, the "crystallization slot," and they could also be frozen and used to collect structural information.     

Development and evaluation of real competitive PCR for high-throughput quantitative applications

Real competitive PCR (rcPCR) has been shown to have high sensitivity, reproducibility, and high-throughput potential. We describe further development and evaluation of this methodology as a tool for measuring nucleic acid abundance within a cell. Modifications to the original protocol allow analysis of gene expression levels using standard conditions regardless of mRNA abundance and assay type, thereby increasing throughput and ease of reaction setup while decreasing optimization time. In addition, we have developed a software package, TITAN, to automatically analyze the results. The details are relevant to researchers performing competitive PCR using any detection technique. The effectiveness of the described developments is demonstrated using 12 genes known to have differential expression in cell lines grown under normal and hypoxic conditions. Quantitative and qualitative comparisons to real-time PCR are presented. It is also demonstrated that the technique is capable of detecting submicroscopic chromosomal DNA deletions.     

Maximum-likelihood crystallization

The crystallization facility of the TB Structural Genomics Consortium, one of nine NIH-sponsored structural genomics pilot projects, employs a combinatorial random sampling technique in high-throughput crystallization screening. Although data are still sparse and a comprehensive analysis cannot be performed at this stage, preliminary results appear to validate the random-screening concept. A discussion of statistical crystallization data analysis aims to draw attention to the need for comprehensive and valid sampling protocols. In view of limited overlap in techniques and sampling parameters between the publicly funded high-throughput crystallography initiatives, exchange of information should be encouraged, aiming to effectively integrate data mining efforts into a comprehensive predictive framework for protein crystallization.     

High-throughput crystallography to enhance drug discovery

Protein crystallography has traditionally been regarded as a resource-intensive, time-consuming technique that, with some notable exceptions, has not made a significant impact on drug discovery. However, inspired by successes in the genome-sequencing initiatives, recent years have seen major changes in X-ray crystallography methodologies and the concept of high-throughput crystallography has emerged. Advances have been made in all phases of the process, including improved molecular biology, protein expression, crystallization and structure determination. This transformation has allowed X-ray crystallography to impact more broadly in the drug-discovery process, extending its utility from structure-based lead optimisation to novel fragment-based lead generation approaches.