Final Report Summary - PCUBE (Infrastructure for Protein Production Platforms)
Executive Summary:
Executive summary
P-CUBE (Infrastructure for Protein Production Platforms) as a “Collaborative Project and Coordination and Support Action” project, was funded by the EU within the Seventh Framework Program from 2009-2013. It comprised research, networking and service activities that aimed at a) offering and granting access to European scientists to all currently available and essential technologies in structural biology and at b) further improving, optimizing and standardizing procedures by sharing expertise and exchange information between the different partners. Thus P-CUBE ensured the highest possible standards for protein production and high-throughput crystallization technologies available in the European Community. In order to enhance the quality and capacity of the platforms available for transnational access, P-CUBE had allocated three work packages to joint research activities that in combination with the transnational access activities guaranteed the fastest possible implementation of new methods and thus helped optimizing the protocols and technologies.
WP6 aimed at optimizing and advancing methods and technologies to increase automation of screening methods thereby ensuring higher throughput and robustness. Thus, the expertise, knowledge and specific skills of each partner were combined and further developed to implement robot friendly-protocols and high-throughput molecular biology methods accelerating the processes of protein selection and also the characterization of proteins. The optimized technologies and methods could be integrated into the infrastructures and thus, served users of the transnational access facilities to faster obtain results. WP7 focused on the automation of DNA purification, the development of light microscopy techniques and the establishment of protein expression systems in eukaryotic, insect and mammalian cells. Again, the specific skills, the expertise of various P-CUBE partners and their technological know-how could be combined for the development of prototypes of robots, tools and of application protocols, which could be successfully integrated into the access platforms. In WP8 emphasis was put on the optimization of methods that allow the crystallization of proteins. The P-CUBE partners have successfully developed low-cost technologies that are robust and accurate and allow easy manipulation of the crystals, thus eliminating errors and mistakes.
The second big pillar of P-CUBE besides research was Transnational Access (TNA). P-CUBE aimed at offering and granting access to all currently available and essential technologies in structural biology to European scientist and provided facilities as well as the expertise and knowhow of experts in the field. P-CUBE welcomed 384 scientists and their projects from 30 different countries all across Europe. Overall, TNA was widely known and happily accepted within the European community, which was mainly due to the broad advertisement such as the distribution of flyers, advertisements in journals, presentations at meetings and the public web page. This web page presented an overview of the project, informing visitors about the main ideas and objectives of the project, explained the scientific approaches, introduced the partners of P-CUBE and provided information on the different platforms offered within the project and on the application procedure. To further reach out to new users, to exchange information, and to introduce and disseminate newly developed techniques trainings and information meetings with former and future users had been organized.
P-CUBE has become a model for bringing together key technologies in cutting edge structural biology research and making these technologies available to the European scientific community in the field. By driving forward technological innovation, providing high precision instrumentation and thus setting up a benchmark in structural biology, P-CUBE has challenged European industry to improve
their capabilities and strengthen the European competitiveness. It furthermore has developed synergies and complementary capabilities in a co-ordinated approach to infrastructure access.
Project Context and Objectives:
Summary description
P-CUBE was a “Collaborative Project and Coordination and Support Action” project and as such it was pursuing the goals to a) offer and grant access to European scientists to all currently available and essential technologies in structural biology such as cloning, expression, protein characterization and crystallisation and b) to further improve such technologies and to optimize and standardize procedures by sharing expertise and exchange information between the different partners thus ensuring the highest possible standards for protein production and high-throughput crystallization technologies available in the European Community.
Different approaches had been envisaged to realize the scientific ambitions. One approach focused on enhancing the automation and throughput of screening methods using bacterial expression by integrating robotic and microfluidic methods. Another aim was to improve automation and throughput of the more demanding eukaryotic expression systems and to develop fluorescent methods of monitoring expression and complex formation in eukaryotic cells. The third objective tackled the development of new crystallization technologies and the automation of the in situ X-ray diffraction of crystals. By integrating such research projects into the access platforms and by ensuring the continuous exchange of information and expertise, the infrastructures available to European users were constantly improved and the access provided optimized.
Project Results:
Joint Research Activities (JRA)
JRA Summary
P-CUBE aimed at improving, optimizing and standardising currently available and essential technologies in structural biology for the efficient production, characterization and finally, the crystallization and the three-dimensional characterization of proteins. To achieve this ambitious goal, P-CUBE first developed and optimized automated high-throughput protein expression methods: Eukaryotic and bacterial cell expression systems had been improved and created to allow the high-throughput production of single proteins (Mammalian und Bacterial cell expression facility) but also of protein complexes (MultiBac facility). In a different approach improved light microscopy techniques were implemented to further characterize proteins inside cells (Advanced Microscopy Facility). In a third approach P-CUBE established a high-throughput technology that allows the screening and the identification of expressible protein domains (Esprit facility). Methods to implement and optimize the screening of binding molecules that serve as chaperon for proteins crystallization have further been established (DARPin technology). Finally, all these techniques served the purpose of producing proteins in sufficient quality and quantity and of enabling the crystallization of proteins (HTX facilities). The knowledge gained and the experience made have been within P-CUBE have been disseminated by all the partners offering workshops, trainings and information meetings.
Grant Agreement no: 227764 P-CUBE Final Report
3
I. Expression Technologies
Bacterial Cell Expression
The objective of the bacterial expression work package was to provide vector construction and expression testing of recombinant proteins in E. coli in order to identify constructs suitable for subsequent functional and structural studies. The platform technology operated by the Oxford Protein Production Facility – UK enables the design, construction and expression testing of multiple vectors in a micro-well plate format. Each project comprises 48 constructs, which can be made up of 48 different targets or 48 variants of a single targets or some combination of both. The workflow is summarized in the figure below.
Eukaryotic Cell Expression
Automation of eukaryotic cell expression
The aim of this project was to develop and benchmark an automated system for preparing large scale DNA preps (of high quality) for transfection into mammalian cells alongside an automated robot for expressing proteins in the mammalian expression system to standards of a quality and consistency suitable for crystallisation and structure determination. In addition, to explore the flexibility to expand the automated system to other eukaryotic expression systems. The automated DNA purification manifold and the expression robot are now in daily use in the UOXF laboratory and routinely used for large scale expression of proteins in mammalian cells in preparation for crystallisation trials.
D7.2 D7.3 and D7.4 described the design, commissioning and implementation of the robot for DNA preparation. The process has been stable and produces high quality DNA for transfection faster and more economically than the previous method of using Giga-Prep columns. Minor modifications in the
OPPF-UK experimental workflow for the construction and expression testing in E. coli of 48 vectors. Each project takes ten working days to carry out.
Grant Agreement no: 227764 P-CUBE Final Report
4
protocol have been made to improve DNA recovery yields and 2 litres of bacterial culture typically produce 18mg DNA. The DNA can be prepared quickly and immediately prior to transfection as the cells become available and transfection efficiencies are also therefore optimised.
Following the development of a transient and scalable expression system for proteins in mammalian cells using standard cell culture methods and large scale expression in roller bottles, a plan for automating some of the processes was co-developed with The Automation Partnership (D7.1 D7.5). Deliverables 7.5 previously reported on the commissioning and implementation of the expression robot using HEK293T cells which underwent significant testing to establish the best robotic parameters and flask brand and style (D7.6). Significant effort was put into benchmarking the performance of Hyperflasks which would provide a significant increase in call capacity, but manufacturing defects made this system unstable and the robot now used Triple flasks routinely. The process was expanded to benchmark HEX293T against another cell line 293S (D7.7) and using a number of different flask brands. Expression time courses were also undertaken to establish variability of these parameters. Results indicated low variation in expression levels, with the major contributing factor being construct-dependent. The expression robot has been a major success in the high volume, large scale mammalian expression of proteins for structure determination and is an integral part of the expression pipeline.
D7.10: A fluorescence-based assay was developed to follow the assembly of protein complexes in a cellular context. Using the ephrin-EphR system, co-localisation and clustering of these molecules in COS cells was followed using localisation microscopy in the form of spectral position determination microscopy (SPDM). Labelling of Ephs with mVenus fluorophore showed different clustering characteristics suggestive of ectodomain-dependent interaction properties in the cell. Fluorescent methods like SPDM are useful for analysis of cellular protein complex formation and dynamics.
Eukaryotic cell expression: Fluorescent light microscopy
Objectives of this work have been the method and infrastructure development for automated eukaryotic expression. Protocols have been developed that allow monitoring the expression and assembly of multi-protein complexes in vivo with the goal to better understand the in vivo assembly process of multi-protein complexes and ultimately overcome bottlenecks in the expression of multi-protein complexes. As part of this work, large collections of protein expression vectors and chemical reagents were assembled that allow specifically labelling proteins and protein complexes for fluorescence light microscopy studies. To study the dynamics of protein interactions and complex assembly in live cells, a technique called Fluorescence Cross Correlation Spectroscopy (FCCS) was applied and proteins tagged with GFP and mCherry (both are small fluorescent proteins and are used to visualize proteins that are otherwise not fluorescent) were used. Using the fluorescence signal emanating from the tagged protein, FCCS can measure how fast a protein is diffusing through a small observation volume and how strong the fluorescence signal is, which is directly related to the concentration of the target protein. By using two different fluorescent tags (colours) on two different proteins (A and B), one can cross-correlate the signal from protein A with the signal from protein B and determine several attributes: what their concentration is in the cell, where they localize (nucleus or cytoplasm), if both proteins interact with each other and how strongly they interact. As a model system, the 6-subunit transcription factor complex TFIIIC was used that is very difficult to overexpress in eukaryotic cells. It could be shows that it is possible to use FCCS in live cells to determine the interaction between individual proteins that together form the complex and obtained
Grant Agreement no: 227764 P-CUBE Final Report
5
first insights into the protein-protein interaction most important for TFIIIC assembly and its cellular localization.
Multiprotein expression in insect cells: Multibac
Almost all important functions in human cells are carried out by molecular machines, which are complexes composed of often more than 10 proteins and other biomolecules. Many of these protein machines could not be studied, as they exist in such small quantities that they could not be extracted from our cells. It is however extremely important to get hold of these protein machines, since they hold the key to understanding cellular activity in health and disease. To overcome the problem, Imre Berger and his colleagues have developed MultiBac, a technology that can produce these important complexes by using a special virus, called baculovirus that can be used to reprogram insect cell cultures to produce even human protein complexes in large quantities. One advantage of MultiBac is that this baculovirus, which can only exist in the laboratory, exclusively infects insect cells, and is entirely harmless to humans. With their patented MultiBac technology, Imre Berger and many others could produce many essential complexes and study their structure and function, to enable future design of drugs to modulate their activity in disease states. MultiBac is one of the PCUBE technology platforms, and many European scientists have accessed the platform through PCUBE TNA to learn the technology and produce their proteins of interest. Recently, the Berger group used MultiBac to produce the core of TFIID, which is an essential multiprotein complex that regulates human gene transcription, i.e. the process that reads our genetic information and is the basis of all functions in our cells. TFIID was discovered more than 20 years ago, but eluded detailed study as it could not be efficiently purified from human cell culture. Production of the TFIID complex MultiBac was of such high quality that the complex could be studied for the first time in detail, providing unprecedented insight into how the machinery that deciphers the hereditary material functions.
PCUBE also resulted in improving the efficiency of the MultiBac platform significantly by RTD. New methods and reagents were developed to integrate genes into the recombinant baculovirus genome, and standard operating protocols were designed and implemented to make the technology also accessible to users not specialised in eukaryotic cell culture techniques, who are interested in producing their multiprotein complexes of interest for research and development.
*MultiBac: expanding the research toolbox for multiprotein complexes.*
Bieniossek et al., Trends Biochem Sci. 2012
Bieniossek et al., Nature, 2013
Grant Agreement no: 227764 P-CUBE Final Report
6
II. Library Technologies
ESPRIT
The objectives of this work were to develop screening methods for analysing difficult protein expression or function-related problems. New protocols for definition of soluble expression clones from poorly understood and challenging protein targets have bee developed. Specifically, using the ESPRIT technology, 28,000 randomly truncated DNA fragments of a single target gene are screened for improved expression characteristics thereby removing the necessity to perform unreliable domain prediction; this is especially an advantage for unannotated targets or those with few similar sequences for construction of alignments. Here in P-CUBE this pre-existing technology was developed toward automated protein complex screening resulting in CoESPRIT (An et al. 2011a), a novel method whereby construct libraries of a target are screen for soluble expression in the presence of a fixed bait protein. This has been used by several TNA users during the second half of P-CUBE. The robotic screening capacity was improved as well by first developing a molecular biology-based system to remove all out-of-frame constructs prior to loading the clones onto the robot. This resulted in a 9-fold increase in the throughput of productive clones that could be handled in the same time period; this approach was named ORF-selector ESPRIT (An et al. 2011b). These P-CUBE-funded improvements to the original ESPRIT technology were described in two reviews (Yumerefendi et al. 2011 and Hart & Waldo 2013).
The experience of this partner in automated library screening and colony array analysis assisted partner UZH in their setting up of the same colony picking and arraying technology whereupon they developed it further towards characterisation of DARPins from library selections in a higher throughput format then previously possible.
ESPRIT libraries yield soluble variants of domains that differ slightly in their termini. This facilitated a joint research activity with EMBL HAM into the use of the microfluidic Fluidigm crystallisation platform for prioritisation of constructs using soluble variants NF-κB and DAP kinase for benchmarking. The major advantage of this approach is that only small quantities of purified protein are required, permitting testing of panels of constructs in parallel. The data indicated that crystallisation propensity of constructs could be measured from protein obtained from small volumes of culture (100 ml) and that some variants clearly crystallised better than others.
The approach of construct screening was also applied to FRET microscopy experiments with partner EMBL HD. Here small libraries of FRET donor and acceptor fusions were synthesised that sample different linker lengths and fusion termini of fluorescent tags. By crossing the two together (i.e. 6 donor vectors and 6 acceptor vectors generate 36 permutations) it was possible in a single experiment to define the optimal linker length and fusion orientation for both targets simultaneously. The proof-of-principle study was based upon interactions of influenza polymerase and one permutation of the 36 tested gave a strong and reliable FRET signal.
Selection of specific Ankyrin Repeat Proteins: DARPins
Over the last decade, UZH has developed Designed Ankyrin Repeat Proteins (DARPins) as an alternative to antibodies in many fields of applications. By using protein engineering, libraries consisting of billions of different DARPin molecules have been designed from which specific, high-affinity binding proteins can be selected to various targets. These binders can afterwards be used, for example, as tumour targeting reagents, as inhibitors to study and influence signalling inside the cell,
Grant Agreement no: 227764 P-CUBE Final Report
7
for co-crystallization of proteins for structural studies, or as model proteins to tackle fundamental questions of protein folding and design - just to name a few applications. DARPins in general are very stable, can easily be prepared in very large amounts from the bacterium E. coli and show very strong affinities up to the picomolar range.
In order to select specific DARPins, ribosome display is used, an in vitro evolution technology developed at UZH. This method is based on in vitro translation, but prevents both the newly synthesized protein and the mRNA encoding it from leaving the ribosome. It thereby couples phenotype and genotype which is necessary to perform selections. Since no bacteria need to be transformed within ribosome display (in contrast to e.g. phage display), very large libraries of 2012 different members can be used directly in the selection rounds, providing sufficient diversity for successful selections. This enhanced library size not only boosts the probability of finding a high affinity ligand within the population, it also increases the number of randomized amino acid positions that can be completely encompassed by the library. In summary, ribosome display has made the cell-free Darwinian evolution of proteins over multiple generations a reality and became a robust technology used in academia and industry alike.
Although ribosome display has already resulted in numerous specific binders to various targets, it still is a complex and time-consuming procedure with a limited throughput so far. Therefore, within the P-CUBE program, we wanted to develop reliable methods for the selection and screening of a large number of DARPins in parallel - a prerequisite to provide numerous specific binders for various scientific projects and to reduce the cost per selection. To enable this high-throughput, we focused on the establishment and validation of a semi-automated selection and screening pipeline. Hence, we first developed and optimized individual selection and screening "building blocks" and finally combined them in a final assembly to apply realistic conditions. Using various liquid handling workstations, 48 independent affinity selections could be performed simultaneously, representing a 12-fold improvement over current manual screenings of ribosome display libraries. Next to allowing us to further optimize various stages of this high-throughput setup, all selections also resulted in specific hits against the chosen targets.
In parallel ribosome display (either using a large number of targets with one library or using several libraries on a smaller number of targets), the bottleneck of the binder generation is not the ribosome display per se, but the subsequent characterization of the individual binders. This so called screening is important to identify those binders from the numerous hits which are best suitable for the various downstream applications. Typically, in a single selection, up to 1,000 functional binders can be obtained, for which specificity and selectivity, recognition of the native state, epitope screen and affinity need to be determined. To prevent the need to purify that many binders (which would be quite time- and cost-intensive), we established ELISA assays in the 384-well format that can be performed in different variations with crude extract lysates directly. As no protein purification is required for these analyses, several hundred or even a thousand clones can be easily screened within a reasonable time. Among these variations were both dilution and competition ELISAs, allowing the differentiation of positive hits based on their affinities already at an early stage of the screening. This feature is of great importance for most applications as it indicates the strength with which the DARPins binds to their target. For most targets, several binders displaying strong competition down to 1-5 nM were uncovered, indicating that the selected DARPins have both high affinities and high specificities for their respective targets.
Grant Agreement no: 227764 P-CUBE Final Report
8
To reveal the identity of these DARPins, the coding DNA sequences of the most promising candidates were determined. In order to perform the sequencing reactions in higher throughput, we implemented a "cherry picking" procedure, allowing the combination of positive hits from various (barcoded) master plates onto a few 96-well plates which then were used for setting up the sequencing reactions. The quality of the sequenced population proved to be very satisfactory with ~90% of the DARPins having unique sequences.
Binders with different sequences can, in principle, bind to different sites on the targets (called epitopes); however, this needs to be experimentally verified. This feature surely would be of great interest for various scientific applications as it would allow the target of interest to be bound and stabilized at various sites. This hopefully would e.g. decrease the amount of false-negative hits in many diagnostic assays or enable detailed structural studies. Therefore, multiple assays to analyse the stabilization of the targets by DARPin binders as well as "epitope binning" were established. The determination of the stabilization of the targets by the DARPins was set up as a routine analysis and an additional ELISA-based epitope binning procedure was developed and validated with a dozen targets. Using this setup, DARPins could be identified that bind to different, non-overlapping epitopes on their identical target protein.
Finally, a variety of the selected and screened binders was expressed in larger scale and subsequently analysed biophysically as purified proteins. Performing various assays, the advantageous features of the selected DARPins could be validated. Thus, these binders were provided to many European cooperation partners for being used and tested in a wide range of applications, from chip detection systems to intracellular localization studies.
Taken all of the illustrated developments together, we are now able to generate valuable binders, not just covering a variety of different targets but also meeting the high quality criteria important for most scientific projects: monomeric binders that specifically recognize their targets with high affinities and which can be expressed at high levels in bacterial systems.
III. High throughout Crystallisation (HTX) - Capillary Technologies
The aim of this project was to shorten the time between crystallization and measuring the diffraction quality of crystals for structural analysis. It became apparent in the early stages that for in-situ diffraction and automated imaging the devices should conform to the SBS standard micro plate format.
D8.1 reported on the development of capillary crystallization plates using glass capillaries, revised to flexible fused silica capillaries.
D8.2 reported on the automated imaging of the CrystalHarp plates using a Formulatrix RI182 imaging system and integrating these plates into ‘XtalPiMS’, a laboratory information management system for crystallization experiments. Imaging a whole CrystalHarp plate was found to take considerably longer than standard Greiner plates.
D8.3 reported on crystal growth in CrystalHarp capillary plates and in the UOXF designed capillary plate which used an SBS microplate base. The X-ray scattering of capillaries made from plastic materials was found to be too high and that these types of capillaries were considered not suitable for in-situ diffraction. In all subsequent developments CrystalHarp-type capillaries were used.
Grant Agreement no: 227764 P-CUBE Final Report
9
Tests on the Greiner CrystalQuick X plate, which has crystallization platforms with ultra-thin bottoms, were also reported. These plates have 2 positions for each of the 96 wells and can be used with standard screening blocks. Crystals are grown by the standard method of vapour diffusion.
D8.4 showed comparative background scattering for Greiner ‘CrystalQuick X’ plates, standard Greiner ‘CrystalQuick SW’ plates, capillaries and microfluidic chips ‘DC-10’. Data collected in-house under identical conditions showed a considerably reduced background scattering for CrystalQuick X plates as compared to standard Greiner plates (see figure below). To demonstrate the proof of principle crystals of vaccinia virus D13 were grown in Crystal-Quick X plates, data were collected in-situ to 3.2Å resolution at beamline I24, Diamond and good electron density maps could be obtained.
Figure: Comparison of background scattering from plates, Fluidigm chips and capillaries
D8.5 extended tests of in-situ diffraction with CrystalQuick X plates to the larger Bovine enterovirus type 2 (BEV-2). It was possible to collect data to 2.2Å, refine the structure and calculate excellent electron density maps (see figure below).
Figure: Structure of Bovine enterovirus type 2 (BEV-2) using data collected in-situ with Greiner CrystalQuick X plates.
Grant Agreement no: 227764 P-CUBE Final Report
10
D8.6 – Current availability, further developments and applications.
Crystallization experiments using Greiner CrystalQuick X plates are now fully integrated into the standard high-throughput pipeline and available for all users of the Oxford facility.
Alternatively, crystals can be grown in capillary plates (CrystalHarp) or using microfluidic technology (Fluidigm DC-10 chips used at UOXF) - both of which can be used for in-situ diffraction.
Figure: Typical arrangement of in-situ plate data collection at Diamond Beamline I24.
Crystal diffraction can be evaluated in-house using the PX-Scanner and in-situ data collection is now routinely available on I03, I04-1 and I24 beamlines at Diamond Light Source (Axford et al., Acta Cryst. 2012). Plate mounting and data collection architecture differs slightly between beamlines, but all can accommodate the SBS plate format, and 2 beamlines include robotic plate mounting. Rotation ranges vary from -10o to +38o (I03); -26o to +26o (I04-1); -20o to +20o (I24, see figure above).
Further developments
Crystallization drops set up with the CrystalQuick X plates tended to spread easily and then formed a thin film covering the whole of the platform. This problem could be overcome by working with Greiner Bio-one to prepare hydrophobically treated CrystalQuick X plates. This type of plate is now the preferred in-situ plate used at UOXF and is commercially available from Greiner Bio-one.
Tests were carried out to find low scattering sealing tapes. Although the scattering from tapes is generally much lower than from plates it is not insignificant. The ‘ThermalSeal RT 50μm’ tape from web-scientific was found to have very low scattering below 5.7Å, so should be better than the standard Greiner ‘Viewseal’ tape for many applications (see figure below). Images shown in figure 4 were recorded with the in-house X-ray source under identical conditions. The resolution at the edge of the detector was 2.6Å.
In parallel we have reported a significant increase in the lifetime of room temperature macromolecular crystals through the use of a high brilliance X-ray beam, reduced exposure times and a fast readout detector (Owen et al., Acta Cryst., 2012).
Grant Agreement no: 227764 P-CUBE Final Report
11
Figure: Comparison of scattering from CrystalQuick X plate and sealing tapes.
Grant Agreement no: 227764 P-CUBE Final Report
12
b) Dissemination
The ambition to enable dissemination of cutting-edge methodologies.
One of P-CUBE’s central missions has been the efficient training in the state-of-the-art technologies that were available at the transnational access infrastructures. Therefore, P-CUBE training activities primarily targeted users who were granted transnational access to the P- CUBE platforms.
This was achieved by offering or contributing to focused training courses in specific state-of-the-art techniques that were used at the platforms. During the 4 years of the project, P-CUBE organized or participated in ten successful training events covering topics such as production, purification and characterization of protein complexes, mammalian protein expression methods, protein crystallization methods as well as advanced protein engineering technologies. The courses and workshops were very well received by the participants who came from all across Europe and beyond, enthusiastically rated the courses and took home many new techniques and ideas.
Another emphasis was placed on yearly user meetings for platform users and operators. In this setting, former users were given the chance to present their scientific projects and to provide feedback about running the platforms and input on how to improve the services and training. Future users could get detailed information on the various platforms. They were able to learn from the experiences of former users and get in contact with the platform operators to discuss their projects. The platform operators informed the participants on the latest technological developments of their platforms, presented the scientific services that they offered and also gave some insight on successful user projects that had been performed at the respective infrastructures. P-CUBE user meetings were always held in combination with a practical workshop to maximize their attractiveness and outreach.
It was exciting and delightful to experience the quality of the scientific talks and projects by the platform users and the fruitful discussions with the platform operators at yearly P-CUBE user meetings held in Grenoble, Zurich and Heidelberg.
Transnational Access (TNA)
TNA Summary
P-CUBE offered and granted access to European scientists to essential methods and technologies in structural biology with the aim to offer tools and techniques for a better understanding of the three-dimensional structure of proteins and their functions. In particular, P-CUBE opened its infrastructures to the scientific community for (i) high-throughput cloning and expression of proteins in prokaryotic and eukaryotic cells, (ii) high-throughput crystallization, (iii) high throughput selection technologies and (iv) advanced light microscopy. In total 384 scientists from 30 different countries had the possibility to get access to and use the P-CUBE infrastructures for their scientific project. The scientific results obtained improved the general understanding of important processes. The following articles summarize the techniques used and provide examples of projects for which the specific technology was important.
Grant Agreement no: 227764 P-CUBE Final Report
13
Bacterial Cell Expression
During the course of the P-cube project the derivation of a series of expression vectors containing different fusion protein tags was reported (Bird LE. Methods. 2011 Sep; 55(1):29-37). These vectors were used in many of the TNA projects (e.g. figure below).
Eukaryotic cell expression
Automated DNA and protein expression described using the technologies above have been used for several published data;
1. Seiradake E et al 2013, ‘Structurally encoded intraclass differences in EphA clusters drive distinct cell responses’ Nature Struct Mol Biol (in press)
2. Bourhis JM et al 2012, ‘Structural basis of fibrillar collagen trimerization and related genetic disorders.’ Nature Struct Mol Biol 19, 1031-1036 (access visitor result)
3. Banci L et al 2013 ‘Atomic-resolution monitoring of protein maturation in live human cells by NMR.’ Nat Chem Biol. 9(5), 297-9 (access visitor result)
...and other results that are yet unpublished arising from Pcube access visits:
1. CHIT, crystals diffract to 1.2 A, were soaked with inhibitors, and the complex structure was obtained.
2. DBM, native and Seleno-Met labelled crystals diffracted and the structure is in the refinement stage.
3. PAPP-A, the crystals diffracted poorly, about 6A; optimisation in underway.
Expression technologies: Monitoring by microscopy
EMBL Heidelberg offered transnational access to the Protein Expression and Purification (PEPCore) and Advanced Light Microscopy (ALMF) platforms. It provided expert advice and technical support for fluorescence labelling of proteins, either through genetic encoding or by using chemical probes, and subsequent light microscopy experiments. Concerning protein labelling, EMBL HD offered
Figure: Expression of a series of constructs of a LysR family transcription factor from Streptococcus pyogenes. A series of 6 protein fusions were made (panel A) for each of 4 constructs of varying length and the expressed proteins analysed by SDS-PAGE (panel B). Abbreviations: POI = protein of interest; TRX = thioredoxin, MBP = maltose-binding protein, TF =trigger factor. (Project: Izabela Sitkiewicz, National Medicines Institute, Warsaw, POLAND)
Grant Agreement no: 227764 P-CUBE Final Report
14
various techniques including chemical fluorescent labelling with sequence tags (Flash/ReAsh, SNAP) and labelling using click chemistry (with unnatural amino acids). These methods are ideally suitable for studying the behaviour and function of proteins in living cells. Concerning the fluorescent light microscopy imaging, EMBL offered a number of techniques and microscopes, expert advice on planning and executing experiments as well as state-of-the-art methods for data processing and analysis. The techniques that were available included FRAP/FLIP, TIRF, FRET, FLIM and FCS/FCCS. Two examples from our TNA activities are given below:
User project: The user implemented labelling unnatural amino acids with click chemistry. She successfully prepared and genetically encoded a new unnatural amino acid featuring a bicyclononyne moiety that reacts remarkably fast with tetrazine-conjugated dyes. In a recent publication, she could encode the new unnatural amino acid inside mammalian cells and selectively label at low dye concentrations, which has proven to be very useful for imaging studies. This work has been published: Borrmann et al., Chembiochem. 2012.
User project: The user successfully implemented the available resources in her project analysing the study of the maturation pathway of dystroglycan, a member of the glycoprotein complex associated to dystrophin. Dystrophin is part of a protein complex that connects the cytoskeleton of a muscle fibre to the surrounding extracellular matrix through the cell membrane. In a recent publication, the user showed that ERp57, a member of the disulfide isomerase family involved in glycoprotein folding, is associated and colocalizes with the β-subunit of the extra cellular receptor dystroglycan. This work indicates that ERp57 is likely to be involved with the processing and maturation pathway of dystroglycan, which is still largely unknown. This work has been published: Sciandra et al., Exp. Cell Res. 2012.
Multiprotein expression in insect cells
MultiBac is a technology that can produce important protein complexes by using a special virus, called baculovirus. This virus can be used to reprogram insect cell cultures to produce even human protein complexes in large quantities. This technology has been used by various users.
User project: Dr. Dieter Palmberger from the University of Vienna, for example, used a PCUBE TNA visit to Grenoble to engineer SweetBac, a novel variant of the MultiBac virus, which can be used to put certain sugar groups on proteins to make them seem more human-like. This is important for example to produce therapeutic proteins such as antibodies, which need to appear like their human counterparts when they are administered to patients, in order to prevent allergic responses that could cause serious damage to the patient. The work has been published: Palmberger, D., PLoS ONE (2012).
The protocols, results and reagents have all been transferred to the laboratories of project proposers enabling the work begun in the OPPF-UK to be followed up. The training received by the visiting students will help in their scientific and technical development as researchers.
Grant Agreement no: 227764 P-CUBE Final Report
15
Tertiary technologies
ESPRIT
The ESPRIT technology allows scientists to identify solubly expressing forms challenging and poorly understood protein targets. The problem addressed is how to generate truncated sub-constructs of proteins corresponding to high-yielding, well-behaving protein domains (or multidomains). Typically biologists analyse protein sequences and try to predict rationally the domain locations, but this approach is frequently unsuccessful, notably when few similar proteins are available for comparison. ESPRIT takes a completely different approach, cutting the target into tens of thousands of random fragments and testing each one for its individual expression properties. To handle such large numbers of tests, robotics developed originally for genome sequencing have been adapted for expression screening purposes. P-CUBE visitors send their samples to EMBL Grenoble where, after initial preparation work, they then then come for 3 weeks of experimental time during which 28,000 constructs are tested. When successful, the visitor returns home with soluble proteins to continue structural or other studies.
Two highlights (note no-one wants to reveal they’ve expressed particular targets solubly, and the structural work is still in progress, so the target is not named).
A Ph.D. student from Leibniz Institut fuer Molekulare Pharmakologie Berlin had been unable to generate soluble material from a GPCR for NMR studies. The target was truncated randomly on both termini, and 28,000 constructs tested during a 3 week visit. The results included several previously undescribed fragments that were soluble and could be produced at milligram scales. This permitted protein labelling and analysis by 2D NMR, yielding important information on conformational properties that may be useful in understanding receptor activation. As important, the block on his project was resolved, permitting finalisation of his thesis and a manuscript.
A postdoc from Institut de Biomedicina-Universitat de Barcelona (IBUB) working on an orphan nuclear receptor was completely unable to produced soluble ligand binding domain until we applied ESPRIT. The output was a well-behaving construct that has been used in structural studies and in protein-protein interaction analyses with peptides from an interacting hub protein.
DARPINs
The University of Zurich offered its experience and its infrastructure to produce so called Designed Ankyrin Repeat Proteins (Darpin) that bind to specific structures inside a cell and interfere and inhibit the binding of other molecules. Such an approach plays an important role in medical applications as the molecules obtained can bind to proteins thereby inhibiting its normal function. The two examples given below show for what kind of topics the technology has been used.
In this project the aim was to select DARPins against highly dynamic structures within our cells, the tubulin filaments. They belong to a family of proteins, the cytoskeleton, which give cells their structure, but also allow them to migrate and change their shape when a signal from the outside arrives. DARPins specific for tubulin are helpful in studying their function.
The gp120 protein of the human immunodeficiency virus (HIV) is responsible to make contact with a target cell and initiates viral entry. In this study DARPins were selected to bind to a motif of gp120 which inhibited infection.
Grant Agreement no: 227764 P-CUBE Final Report
16
High-throughput Crystallisation technologies
The possibility to use the expertise and knowledge of experts in the field of protein crystallization was given by various P-CUBE partners. By using these infrastructures scientists have been able to find out about the structure of a protein. The structure again provides insights into the function of a protein and therefore reveals information that might be useful for future medical applications. Some projects that have been conducted for other scientists in the P-CUBE labs are given below.
User project: Many bacteria communicate between individual cells through the secretion of small signalling molecules. The signals are used to see if there is a quorum, i.e. a large enough population of bacteria to start colonization. This process can involve the formation of biofilms, and pathogenic bacteria use quorum sensors to initiate the production of virulence factors that destroy host tissue. The small molecules acting as signals are processed by enzymes within the individual bacteria, and this triggers alteration in gene expression. The Streit group from Hamburg University (Germany) determined the crystal structure of a novel class of quorum sensing enzymes. Initial crystallization trials resulted in several conditions were crystalline material was obtained that was not yet of sufficient quality to obtain an interpretable X-ray diffraction pattern. Several rounds of optimization helped to obtain larger well-diffracting crystals at the facility, and the structure helps to explain how this enzyme acts as a quorum sensor (Bijtenhoorn et al., PLoS One. 2011). The structure is now used to investigate if specific inhibitors can interfere with bacterial colonization, which is a promising alternative strategy for the eradication of pathogenic bacteria.
User project: The finite nature of fossil fuels dictates that we must look for sustainable renewable sources of liquid fuels. Consequently, there is a major thrust towards second-generation biofuel production from lignocellulosic feedstocks, either purpose-grown or as agricultural or municipal wastes, using bacterial fermentation processes. In collaboration with TMO Renewables Ltd, who have developed a fermentative bioethanol production process, the Danson group from the University of Bath, UK, is working on thermostable enzymes that can degrade xylan, namely xylanases and xylosidases.They have isolated a Xylosidase from a thermophilic bacterium and have crystallized and determined the first structure of the class of enzymes to which it belongs, revealing new features. Together with their industrial partner, the group is now analysing the structure to design mutations that could make the enzyme more efficient.
User project: Numerous animal species, from insects (mosquitoes) to mammals (vampire bats), feed primarily on fresh blood from their prey. These parasites produce some of the most potent antagonists of the blood clotting system known. Anopheles mosquitoes are widespread hematophagous parasites and important vectors of malaria, a potentially lethal disease that affects 500 million humans worldwide. Here, the authors show that a salivary polypeptide from Anopheles, anophelin, inhibits thrombin by binding in an orientation opposite to the orientation of substrates and disrupts thrombin’s catalytic machinery. This unique molecular mechanism has implications for the design of synthetic antithrombotics. This work has been published: Figuereido et al. PNAS 2012
User project: DNA Mismatch-repair factors have a prominent role in surveying eukaryotic DNA-replication fidelity. This function depends on the activity of protein complex formed by MutL-homolog and Mlh1. In humans, MLH1 mutations underlie half of hereditary nonpolyposis colorectal
Grant Agreement no: 227764 P-CUBE Final Report
17
cancers (HNPCCs). This work reveals the crystal structures of the MutLα (Mlh1–Pms1 heterodimer) C-terminal domain (CTD) from Saccharomyces cerevisiae, alone and in complex with fragments derived from Mlh1 partners. These structures provide a rationale for the deleterious impact of MLH1 mutations in HNPCCs. This work was published: Egueneau et al., NSMB 2013
User project: A project that benefitted from this approach was on a protein that is involved in bacterial quorum sensing. Some pathogenic bacteria use small molecules as a signal that indicates its population density. These signals are processed by enzymes within the individual bacteria, and this triggers alteration in gene expression and processes such as the formation of biofilms. The Streit group from Hamburg University (Germany) determined the crystal structure of a novel class of quorum sensing enzymes. Initial crystallization trials resulted in several conditions were crystalline material was obtained that was not yet of sufficient quality to obtain an interpretable X-ray diffraction pattern. The Optisalt screens helped to obtain larger well-diffracting crystals in an automatic fashion, and the structure confirmed the role of this enzyme as a quorum sensor. This work has been published: Bijtenhoorn P. et al, PLoS One. 2011.
User project: The C propeptides of fibrillar procollagens have crucial roles in tissue growth and repair by controlling both the intracellular assembly of procollagen molecules and the extracellular assembly of collagen fibrils. Mutations in C propeptides are associated with several, often lethal, genetic disorders affecting bone, cartilage, blood vessels and skin. The crystal structure of a C-propeptide domain from human procollagen III was determined in collaboration with the group of D. Hulmes in Lyon, France (see figure below). The structure reveals an exquisite mechanism of chain recognition during intracellular trimerization of the procollagen molecule. It also gives insights into why some types of collagen consist of three identical polypeptide chains, whereas others do not. The data show striking correlations between the sites of numerous disease-related mutations in different C-propeptide domains and the degree of phenotype severity. The results have broad implications for understanding genetic disorders of connective tissues and designing new therapeutic strategies. This work has been published: Bourhis JM et al. Nat Struct Mol Biol., 2012.
Figure: Structure of the C-propeptide trimer of human procollagen III with secondary structure elements of chain B indicated.
Grant Agreement no: 227764 P-CUBE Final Report
18
User project: It has proved difficult to classify viruses unless they are closely related since their rapid evolution hinders detection of remote evolutionary relationships in their genetic sequences. However, structure varies more slowly than sequence, allowing deeper evolutionary relationships to be detected. Bacteriophage P23-77 is an example of a newly identified viral lineage, with members inhabiting extreme environments. In collaboration with the group of J. Bamford, Jyvaskyla, Finland we have solved multiple crystal structures of the major capsid proteins VP16 and VP17 of bacteriophage P23-77 (see figure below). They fit the 14 Å resolution cryo-electron microscopy reconstruction of the entire virus exquisitely well, allowing us to propose a model for both the capsid architecture and viral assembly, quite different from previously published models. The structures of the capsid proteins and their mode of association to form the viral capsid suggest that the P23-77-like and adeno-PRD1 lineages of viruses share an extremely ancient common ancestor. This work is in press: Rissanen, I et al. Structure 21, in press, May 2013.
Figure: Capsid Organisation of P23-77
(a) Crystal structures of the major capsid proteins VP16 (orange) and VP17 (green) fitted into the cryo-EM electron density map of the P23-77 virion. This work is published: Jaatinen, ST et al., Virology, 2008.
(b) Model of the whole P23-77 capsid with the asymmetric unit and the symmetry axis marked,
(c) Details of the arrangement of VP16 and VP17 within the asymmetric unit.
This work will be published: Ren J, et al Nature Comms 2013 (in press).
Grant Agreement no: 227764 P-CUBE Final Report
19
HTX capillary crystallisation and eukaryotic cell expression applications
CrystalQuick X plates have already been used successfully for the following viruses:
ERAV
EV71 (Wang et al., NSMB 2012)
BEV-1
BEV-2
FMDV (Porta et al., PLoS Pathog., 2013)
CAV16 (Ren et al., Nat. Comms., 2013)
Figure: Crystal structure of FMDV serotype A22 with the pentamers and the α-helix at the two-fold symmetry axis highlighted.
Most recently crystals of empty capsids of Foot and Mouth Disease Virus (FMDV) were grown in CrystalQuick X plates and data were collected in-situ at Diamond, Beamline I24. The structures of A22-wt and A22-H2093C helped in the rational engineering of recombinant picornavirus capsids to produce safe, protective vaccine antigen. The figure shows the structure of A22-wt with His-93 at the interface between the pentamers. In A22-H2093C this histidine is changed to a cysteine thereby forming a disulphide bond between the pentamers and forming a more stable capsid.
Potential Impact:
b) Dissemination
The ambition to enable dissemination of cutting-edge methodologies.
One of P-CUBE’s central missions has been the efficient training in the state-of-the-art technologies that were available at the transnational access infrastructures. Therefore, P-CUBE training activities primarily targeted users who were granted transnational access to the P- CUBE platforms.
This was achieved by offering or contributing to focused training courses in specific state-of-the-art techniques that were used at the platforms. During the 4 years of the project, P-CUBE organized or participated in ten successful training events covering topics such as production, purification and characterization of protein complexes, mammalian protein expression methods, protein crystallization methods as well as advanced protein engineering technologies. The courses and workshops were very well received by the participants who came from all across Europe and beyond, enthusiastically rated the courses and took home many new techniques and ideas.
Another emphasis was placed on yearly user meetings for platform users and operators. In this setting, former users were given the chance to present their scientific projects and to provide feedback about running the platforms and input on how to improve the services and training. Future users could get detailed information on the various platforms. They were able to learn from the experiences of former users and get in contact with the platform operators to discuss their projects. The platform operators informed the participants on the latest technological developments of their platforms, presented the scientific services that they offered and also gave some insight on successful user projects that had been performed at the respective infrastructures. P-CUBE user meetings were always held in combination with a practical workshop to maximize their attractiveness and outreach.
It was exciting and delightful to experience the quality of the scientific talks and projects by the platform users and the fruitful discussions with the platform operators at yearly P-CUBE user meetings held in Grenoble, Zurich and Heidelberg.
c) Impact
All dissemination activities are described in Section b of this report.
In recent years, new pathogens (like viruses or bacteria) frequently emerged that were and are a threat to human or environmental health. Next to the yearly occurrence of new influenza viruses, the outbreak of SARS (Severe acute respiratory syndrome, a pandemic with almost 1,000 deaths in 37 countries in early 2003) or the recent enterohemorrhagic E. coli (EHEC) pandemia clearly showed that the scientific community must be able to react instantly and without delay to these threats, providing binders for better detection and identification of potentially infected patients. Therefore, the impact of our work will be to provide the selected binders against the present targets, but more importantly, to supply a robust selection and screening platform that can be the basis for the prompt selection of highly specific and highly selective DARPin binders against future targets, requiring reduced cost and time investments compared to previous methods. The binders that can be selected with our system will be of great importance for various future projects, ranging from basic biology like e.g. studies of fundamental questions of protein folding or cell signalling cascades to the more applied fields of
diagnostics and therapeutics. Thus, we believe that the true value of our development within the P-CUBE program, despite being already noticeable at the current point, will further grow to its full extent in the years to come.
FCCS is a very promising fluorescence light microscopy technique and is expected to be very useful for precise and detailed analysis and characterization of cellular protein interactions in living cells. Since it is performed in vivo and in live cells, the obtained results provide an unbiased view of cellular interactions that otherwise cannot be obtained. Better understanding the assembly process of multi-protein complexes is critically required to optimize their expression in eukaryotic cells. In general, the FCCS technique has a number of applications, but it is particularly interesting for biomedical fields, where the analysis of quantity, interaction partners or the aggregation state of a protein in the cellular context is important to understand a particular defect and the associated disease.
Applying library strategies to complex problems generates solutions that are difficult to predict rationally. Here we have applied this logic to the generation of soluble variants of challenging and poorly understood protein targets. This has clear applications in generating proteins for pharmaceutical applications (structural biology, high throughput screening) and vaccinology; evidenced by the industrial collaborations performed during this period with world-leading companies on targets such as protein kinases and pathogen proteins for immunisation. The results from this work have been disseminated via 4 peer-reviewed publications and regular invited presentations at industrial and academic conferences.
The automation of eukaryotic expression methodology has been widely adopted as elements in existing pipelines at the UOXF facility. The expression robot is a large high value piece of equipment which is not suitable for every laboratory and therefore, while the principles have been adopted, the scale of automation has not. However, automation in general has produced reproducible, standardised high quality expression in large volume which enables fast progress from construct design to protein expressed. This has been exemplified for several of the access visitors to the UOXF facility, where the aim of the access has been to express a difficult protein or protein complex, and the visitor has been able to go further to express, crystallise and solve the structure of their protein in very short time (<1 week). This is only possible due to automation which allows 24hr cell culture runs and fast purification and crystallisation screening. The societal impact of this work is the improved skills of mainly young researchers; the adoption of new technologies in laboratories following exposure at the access host laboratory; the impact of potentially important biological data derived from the protein structures solved and the knowledge outreach following publication.
The impact of crystallisation developments is mainly in the improved time and quality efficiencies of collecting X-ray data directly from crystals in plates, without the need to mount crystals (a time consuming manual task). This translates into cost efficiencies at the synchrotron. Data quality is maintained and in particular, large numbers of microcrystals can be quickly screened in plates. The societal impact of this work is the improved skills of mainly young researchers; the adoption of new technologies and data capture methods at the synchrotron; the impact of potentially important biological data derived from the protein structures solved and the knowledge outreach following publication.
List of Websites:
www.p-cube.eu
Executive summary
P-CUBE (Infrastructure for Protein Production Platforms) as a “Collaborative Project and Coordination and Support Action” project, was funded by the EU within the Seventh Framework Program from 2009-2013. It comprised research, networking and service activities that aimed at a) offering and granting access to European scientists to all currently available and essential technologies in structural biology and at b) further improving, optimizing and standardizing procedures by sharing expertise and exchange information between the different partners. Thus P-CUBE ensured the highest possible standards for protein production and high-throughput crystallization technologies available in the European Community. In order to enhance the quality and capacity of the platforms available for transnational access, P-CUBE had allocated three work packages to joint research activities that in combination with the transnational access activities guaranteed the fastest possible implementation of new methods and thus helped optimizing the protocols and technologies.
WP6 aimed at optimizing and advancing methods and technologies to increase automation of screening methods thereby ensuring higher throughput and robustness. Thus, the expertise, knowledge and specific skills of each partner were combined and further developed to implement robot friendly-protocols and high-throughput molecular biology methods accelerating the processes of protein selection and also the characterization of proteins. The optimized technologies and methods could be integrated into the infrastructures and thus, served users of the transnational access facilities to faster obtain results. WP7 focused on the automation of DNA purification, the development of light microscopy techniques and the establishment of protein expression systems in eukaryotic, insect and mammalian cells. Again, the specific skills, the expertise of various P-CUBE partners and their technological know-how could be combined for the development of prototypes of robots, tools and of application protocols, which could be successfully integrated into the access platforms. In WP8 emphasis was put on the optimization of methods that allow the crystallization of proteins. The P-CUBE partners have successfully developed low-cost technologies that are robust and accurate and allow easy manipulation of the crystals, thus eliminating errors and mistakes.
The second big pillar of P-CUBE besides research was Transnational Access (TNA). P-CUBE aimed at offering and granting access to all currently available and essential technologies in structural biology to European scientist and provided facilities as well as the expertise and knowhow of experts in the field. P-CUBE welcomed 384 scientists and their projects from 30 different countries all across Europe. Overall, TNA was widely known and happily accepted within the European community, which was mainly due to the broad advertisement such as the distribution of flyers, advertisements in journals, presentations at meetings and the public web page. This web page presented an overview of the project, informing visitors about the main ideas and objectives of the project, explained the scientific approaches, introduced the partners of P-CUBE and provided information on the different platforms offered within the project and on the application procedure. To further reach out to new users, to exchange information, and to introduce and disseminate newly developed techniques trainings and information meetings with former and future users had been organized.
P-CUBE has become a model for bringing together key technologies in cutting edge structural biology research and making these technologies available to the European scientific community in the field. By driving forward technological innovation, providing high precision instrumentation and thus setting up a benchmark in structural biology, P-CUBE has challenged European industry to improve
their capabilities and strengthen the European competitiveness. It furthermore has developed synergies and complementary capabilities in a co-ordinated approach to infrastructure access.
Project Context and Objectives:
Summary description
P-CUBE was a “Collaborative Project and Coordination and Support Action” project and as such it was pursuing the goals to a) offer and grant access to European scientists to all currently available and essential technologies in structural biology such as cloning, expression, protein characterization and crystallisation and b) to further improve such technologies and to optimize and standardize procedures by sharing expertise and exchange information between the different partners thus ensuring the highest possible standards for protein production and high-throughput crystallization technologies available in the European Community.
Different approaches had been envisaged to realize the scientific ambitions. One approach focused on enhancing the automation and throughput of screening methods using bacterial expression by integrating robotic and microfluidic methods. Another aim was to improve automation and throughput of the more demanding eukaryotic expression systems and to develop fluorescent methods of monitoring expression and complex formation in eukaryotic cells. The third objective tackled the development of new crystallization technologies and the automation of the in situ X-ray diffraction of crystals. By integrating such research projects into the access platforms and by ensuring the continuous exchange of information and expertise, the infrastructures available to European users were constantly improved and the access provided optimized.
Project Results:
Joint Research Activities (JRA)
JRA Summary
P-CUBE aimed at improving, optimizing and standardising currently available and essential technologies in structural biology for the efficient production, characterization and finally, the crystallization and the three-dimensional characterization of proteins. To achieve this ambitious goal, P-CUBE first developed and optimized automated high-throughput protein expression methods: Eukaryotic and bacterial cell expression systems had been improved and created to allow the high-throughput production of single proteins (Mammalian und Bacterial cell expression facility) but also of protein complexes (MultiBac facility). In a different approach improved light microscopy techniques were implemented to further characterize proteins inside cells (Advanced Microscopy Facility). In a third approach P-CUBE established a high-throughput technology that allows the screening and the identification of expressible protein domains (Esprit facility). Methods to implement and optimize the screening of binding molecules that serve as chaperon for proteins crystallization have further been established (DARPin technology). Finally, all these techniques served the purpose of producing proteins in sufficient quality and quantity and of enabling the crystallization of proteins (HTX facilities). The knowledge gained and the experience made have been within P-CUBE have been disseminated by all the partners offering workshops, trainings and information meetings.
Grant Agreement no: 227764 P-CUBE Final Report
3
I. Expression Technologies
Bacterial Cell Expression
The objective of the bacterial expression work package was to provide vector construction and expression testing of recombinant proteins in E. coli in order to identify constructs suitable for subsequent functional and structural studies. The platform technology operated by the Oxford Protein Production Facility – UK enables the design, construction and expression testing of multiple vectors in a micro-well plate format. Each project comprises 48 constructs, which can be made up of 48 different targets or 48 variants of a single targets or some combination of both. The workflow is summarized in the figure below.
Eukaryotic Cell Expression
Automation of eukaryotic cell expression
The aim of this project was to develop and benchmark an automated system for preparing large scale DNA preps (of high quality) for transfection into mammalian cells alongside an automated robot for expressing proteins in the mammalian expression system to standards of a quality and consistency suitable for crystallisation and structure determination. In addition, to explore the flexibility to expand the automated system to other eukaryotic expression systems. The automated DNA purification manifold and the expression robot are now in daily use in the UOXF laboratory and routinely used for large scale expression of proteins in mammalian cells in preparation for crystallisation trials.
D7.2 D7.3 and D7.4 described the design, commissioning and implementation of the robot for DNA preparation. The process has been stable and produces high quality DNA for transfection faster and more economically than the previous method of using Giga-Prep columns. Minor modifications in the
OPPF-UK experimental workflow for the construction and expression testing in E. coli of 48 vectors. Each project takes ten working days to carry out.
Grant Agreement no: 227764 P-CUBE Final Report
4
protocol have been made to improve DNA recovery yields and 2 litres of bacterial culture typically produce 18mg DNA. The DNA can be prepared quickly and immediately prior to transfection as the cells become available and transfection efficiencies are also therefore optimised.
Following the development of a transient and scalable expression system for proteins in mammalian cells using standard cell culture methods and large scale expression in roller bottles, a plan for automating some of the processes was co-developed with The Automation Partnership (D7.1 D7.5). Deliverables 7.5 previously reported on the commissioning and implementation of the expression robot using HEK293T cells which underwent significant testing to establish the best robotic parameters and flask brand and style (D7.6). Significant effort was put into benchmarking the performance of Hyperflasks which would provide a significant increase in call capacity, but manufacturing defects made this system unstable and the robot now used Triple flasks routinely. The process was expanded to benchmark HEX293T against another cell line 293S (D7.7) and using a number of different flask brands. Expression time courses were also undertaken to establish variability of these parameters. Results indicated low variation in expression levels, with the major contributing factor being construct-dependent. The expression robot has been a major success in the high volume, large scale mammalian expression of proteins for structure determination and is an integral part of the expression pipeline.
D7.10: A fluorescence-based assay was developed to follow the assembly of protein complexes in a cellular context. Using the ephrin-EphR system, co-localisation and clustering of these molecules in COS cells was followed using localisation microscopy in the form of spectral position determination microscopy (SPDM). Labelling of Ephs with mVenus fluorophore showed different clustering characteristics suggestive of ectodomain-dependent interaction properties in the cell. Fluorescent methods like SPDM are useful for analysis of cellular protein complex formation and dynamics.
Eukaryotic cell expression: Fluorescent light microscopy
Objectives of this work have been the method and infrastructure development for automated eukaryotic expression. Protocols have been developed that allow monitoring the expression and assembly of multi-protein complexes in vivo with the goal to better understand the in vivo assembly process of multi-protein complexes and ultimately overcome bottlenecks in the expression of multi-protein complexes. As part of this work, large collections of protein expression vectors and chemical reagents were assembled that allow specifically labelling proteins and protein complexes for fluorescence light microscopy studies. To study the dynamics of protein interactions and complex assembly in live cells, a technique called Fluorescence Cross Correlation Spectroscopy (FCCS) was applied and proteins tagged with GFP and mCherry (both are small fluorescent proteins and are used to visualize proteins that are otherwise not fluorescent) were used. Using the fluorescence signal emanating from the tagged protein, FCCS can measure how fast a protein is diffusing through a small observation volume and how strong the fluorescence signal is, which is directly related to the concentration of the target protein. By using two different fluorescent tags (colours) on two different proteins (A and B), one can cross-correlate the signal from protein A with the signal from protein B and determine several attributes: what their concentration is in the cell, where they localize (nucleus or cytoplasm), if both proteins interact with each other and how strongly they interact. As a model system, the 6-subunit transcription factor complex TFIIIC was used that is very difficult to overexpress in eukaryotic cells. It could be shows that it is possible to use FCCS in live cells to determine the interaction between individual proteins that together form the complex and obtained
Grant Agreement no: 227764 P-CUBE Final Report
5
first insights into the protein-protein interaction most important for TFIIIC assembly and its cellular localization.
Multiprotein expression in insect cells: Multibac
Almost all important functions in human cells are carried out by molecular machines, which are complexes composed of often more than 10 proteins and other biomolecules. Many of these protein machines could not be studied, as they exist in such small quantities that they could not be extracted from our cells. It is however extremely important to get hold of these protein machines, since they hold the key to understanding cellular activity in health and disease. To overcome the problem, Imre Berger and his colleagues have developed MultiBac, a technology that can produce these important complexes by using a special virus, called baculovirus that can be used to reprogram insect cell cultures to produce even human protein complexes in large quantities. One advantage of MultiBac is that this baculovirus, which can only exist in the laboratory, exclusively infects insect cells, and is entirely harmless to humans. With their patented MultiBac technology, Imre Berger and many others could produce many essential complexes and study their structure and function, to enable future design of drugs to modulate their activity in disease states. MultiBac is one of the PCUBE technology platforms, and many European scientists have accessed the platform through PCUBE TNA to learn the technology and produce their proteins of interest. Recently, the Berger group used MultiBac to produce the core of TFIID, which is an essential multiprotein complex that regulates human gene transcription, i.e. the process that reads our genetic information and is the basis of all functions in our cells. TFIID was discovered more than 20 years ago, but eluded detailed study as it could not be efficiently purified from human cell culture. Production of the TFIID complex MultiBac was of such high quality that the complex could be studied for the first time in detail, providing unprecedented insight into how the machinery that deciphers the hereditary material functions.
PCUBE also resulted in improving the efficiency of the MultiBac platform significantly by RTD. New methods and reagents were developed to integrate genes into the recombinant baculovirus genome, and standard operating protocols were designed and implemented to make the technology also accessible to users not specialised in eukaryotic cell culture techniques, who are interested in producing their multiprotein complexes of interest for research and development.
*MultiBac: expanding the research toolbox for multiprotein complexes.*
Bieniossek et al., Trends Biochem Sci. 2012
Bieniossek et al., Nature, 2013
Grant Agreement no: 227764 P-CUBE Final Report
6
II. Library Technologies
ESPRIT
The objectives of this work were to develop screening methods for analysing difficult protein expression or function-related problems. New protocols for definition of soluble expression clones from poorly understood and challenging protein targets have bee developed. Specifically, using the ESPRIT technology, 28,000 randomly truncated DNA fragments of a single target gene are screened for improved expression characteristics thereby removing the necessity to perform unreliable domain prediction; this is especially an advantage for unannotated targets or those with few similar sequences for construction of alignments. Here in P-CUBE this pre-existing technology was developed toward automated protein complex screening resulting in CoESPRIT (An et al. 2011a), a novel method whereby construct libraries of a target are screen for soluble expression in the presence of a fixed bait protein. This has been used by several TNA users during the second half of P-CUBE. The robotic screening capacity was improved as well by first developing a molecular biology-based system to remove all out-of-frame constructs prior to loading the clones onto the robot. This resulted in a 9-fold increase in the throughput of productive clones that could be handled in the same time period; this approach was named ORF-selector ESPRIT (An et al. 2011b). These P-CUBE-funded improvements to the original ESPRIT technology were described in two reviews (Yumerefendi et al. 2011 and Hart & Waldo 2013).
The experience of this partner in automated library screening and colony array analysis assisted partner UZH in their setting up of the same colony picking and arraying technology whereupon they developed it further towards characterisation of DARPins from library selections in a higher throughput format then previously possible.
ESPRIT libraries yield soluble variants of domains that differ slightly in their termini. This facilitated a joint research activity with EMBL HAM into the use of the microfluidic Fluidigm crystallisation platform for prioritisation of constructs using soluble variants NF-κB and DAP kinase for benchmarking. The major advantage of this approach is that only small quantities of purified protein are required, permitting testing of panels of constructs in parallel. The data indicated that crystallisation propensity of constructs could be measured from protein obtained from small volumes of culture (100 ml) and that some variants clearly crystallised better than others.
The approach of construct screening was also applied to FRET microscopy experiments with partner EMBL HD. Here small libraries of FRET donor and acceptor fusions were synthesised that sample different linker lengths and fusion termini of fluorescent tags. By crossing the two together (i.e. 6 donor vectors and 6 acceptor vectors generate 36 permutations) it was possible in a single experiment to define the optimal linker length and fusion orientation for both targets simultaneously. The proof-of-principle study was based upon interactions of influenza polymerase and one permutation of the 36 tested gave a strong and reliable FRET signal.
Selection of specific Ankyrin Repeat Proteins: DARPins
Over the last decade, UZH has developed Designed Ankyrin Repeat Proteins (DARPins) as an alternative to antibodies in many fields of applications. By using protein engineering, libraries consisting of billions of different DARPin molecules have been designed from which specific, high-affinity binding proteins can be selected to various targets. These binders can afterwards be used, for example, as tumour targeting reagents, as inhibitors to study and influence signalling inside the cell,
Grant Agreement no: 227764 P-CUBE Final Report
7
for co-crystallization of proteins for structural studies, or as model proteins to tackle fundamental questions of protein folding and design - just to name a few applications. DARPins in general are very stable, can easily be prepared in very large amounts from the bacterium E. coli and show very strong affinities up to the picomolar range.
In order to select specific DARPins, ribosome display is used, an in vitro evolution technology developed at UZH. This method is based on in vitro translation, but prevents both the newly synthesized protein and the mRNA encoding it from leaving the ribosome. It thereby couples phenotype and genotype which is necessary to perform selections. Since no bacteria need to be transformed within ribosome display (in contrast to e.g. phage display), very large libraries of 2012 different members can be used directly in the selection rounds, providing sufficient diversity for successful selections. This enhanced library size not only boosts the probability of finding a high affinity ligand within the population, it also increases the number of randomized amino acid positions that can be completely encompassed by the library. In summary, ribosome display has made the cell-free Darwinian evolution of proteins over multiple generations a reality and became a robust technology used in academia and industry alike.
Although ribosome display has already resulted in numerous specific binders to various targets, it still is a complex and time-consuming procedure with a limited throughput so far. Therefore, within the P-CUBE program, we wanted to develop reliable methods for the selection and screening of a large number of DARPins in parallel - a prerequisite to provide numerous specific binders for various scientific projects and to reduce the cost per selection. To enable this high-throughput, we focused on the establishment and validation of a semi-automated selection and screening pipeline. Hence, we first developed and optimized individual selection and screening "building blocks" and finally combined them in a final assembly to apply realistic conditions. Using various liquid handling workstations, 48 independent affinity selections could be performed simultaneously, representing a 12-fold improvement over current manual screenings of ribosome display libraries. Next to allowing us to further optimize various stages of this high-throughput setup, all selections also resulted in specific hits against the chosen targets.
In parallel ribosome display (either using a large number of targets with one library or using several libraries on a smaller number of targets), the bottleneck of the binder generation is not the ribosome display per se, but the subsequent characterization of the individual binders. This so called screening is important to identify those binders from the numerous hits which are best suitable for the various downstream applications. Typically, in a single selection, up to 1,000 functional binders can be obtained, for which specificity and selectivity, recognition of the native state, epitope screen and affinity need to be determined. To prevent the need to purify that many binders (which would be quite time- and cost-intensive), we established ELISA assays in the 384-well format that can be performed in different variations with crude extract lysates directly. As no protein purification is required for these analyses, several hundred or even a thousand clones can be easily screened within a reasonable time. Among these variations were both dilution and competition ELISAs, allowing the differentiation of positive hits based on their affinities already at an early stage of the screening. This feature is of great importance for most applications as it indicates the strength with which the DARPins binds to their target. For most targets, several binders displaying strong competition down to 1-5 nM were uncovered, indicating that the selected DARPins have both high affinities and high specificities for their respective targets.
Grant Agreement no: 227764 P-CUBE Final Report
8
To reveal the identity of these DARPins, the coding DNA sequences of the most promising candidates were determined. In order to perform the sequencing reactions in higher throughput, we implemented a "cherry picking" procedure, allowing the combination of positive hits from various (barcoded) master plates onto a few 96-well plates which then were used for setting up the sequencing reactions. The quality of the sequenced population proved to be very satisfactory with ~90% of the DARPins having unique sequences.
Binders with different sequences can, in principle, bind to different sites on the targets (called epitopes); however, this needs to be experimentally verified. This feature surely would be of great interest for various scientific applications as it would allow the target of interest to be bound and stabilized at various sites. This hopefully would e.g. decrease the amount of false-negative hits in many diagnostic assays or enable detailed structural studies. Therefore, multiple assays to analyse the stabilization of the targets by DARPin binders as well as "epitope binning" were established. The determination of the stabilization of the targets by the DARPins was set up as a routine analysis and an additional ELISA-based epitope binning procedure was developed and validated with a dozen targets. Using this setup, DARPins could be identified that bind to different, non-overlapping epitopes on their identical target protein.
Finally, a variety of the selected and screened binders was expressed in larger scale and subsequently analysed biophysically as purified proteins. Performing various assays, the advantageous features of the selected DARPins could be validated. Thus, these binders were provided to many European cooperation partners for being used and tested in a wide range of applications, from chip detection systems to intracellular localization studies.
Taken all of the illustrated developments together, we are now able to generate valuable binders, not just covering a variety of different targets but also meeting the high quality criteria important for most scientific projects: monomeric binders that specifically recognize their targets with high affinities and which can be expressed at high levels in bacterial systems.
III. High throughout Crystallisation (HTX) - Capillary Technologies
The aim of this project was to shorten the time between crystallization and measuring the diffraction quality of crystals for structural analysis. It became apparent in the early stages that for in-situ diffraction and automated imaging the devices should conform to the SBS standard micro plate format.
D8.1 reported on the development of capillary crystallization plates using glass capillaries, revised to flexible fused silica capillaries.
D8.2 reported on the automated imaging of the CrystalHarp plates using a Formulatrix RI182 imaging system and integrating these plates into ‘XtalPiMS’, a laboratory information management system for crystallization experiments. Imaging a whole CrystalHarp plate was found to take considerably longer than standard Greiner plates.
D8.3 reported on crystal growth in CrystalHarp capillary plates and in the UOXF designed capillary plate which used an SBS microplate base. The X-ray scattering of capillaries made from plastic materials was found to be too high and that these types of capillaries were considered not suitable for in-situ diffraction. In all subsequent developments CrystalHarp-type capillaries were used.
Grant Agreement no: 227764 P-CUBE Final Report
9
Tests on the Greiner CrystalQuick X plate, which has crystallization platforms with ultra-thin bottoms, were also reported. These plates have 2 positions for each of the 96 wells and can be used with standard screening blocks. Crystals are grown by the standard method of vapour diffusion.
D8.4 showed comparative background scattering for Greiner ‘CrystalQuick X’ plates, standard Greiner ‘CrystalQuick SW’ plates, capillaries and microfluidic chips ‘DC-10’. Data collected in-house under identical conditions showed a considerably reduced background scattering for CrystalQuick X plates as compared to standard Greiner plates (see figure below). To demonstrate the proof of principle crystals of vaccinia virus D13 were grown in Crystal-Quick X plates, data were collected in-situ to 3.2Å resolution at beamline I24, Diamond and good electron density maps could be obtained.
Figure: Comparison of background scattering from plates, Fluidigm chips and capillaries
D8.5 extended tests of in-situ diffraction with CrystalQuick X plates to the larger Bovine enterovirus type 2 (BEV-2). It was possible to collect data to 2.2Å, refine the structure and calculate excellent electron density maps (see figure below).
Figure: Structure of Bovine enterovirus type 2 (BEV-2) using data collected in-situ with Greiner CrystalQuick X plates.
Grant Agreement no: 227764 P-CUBE Final Report
10
D8.6 – Current availability, further developments and applications.
Crystallization experiments using Greiner CrystalQuick X plates are now fully integrated into the standard high-throughput pipeline and available for all users of the Oxford facility.
Alternatively, crystals can be grown in capillary plates (CrystalHarp) or using microfluidic technology (Fluidigm DC-10 chips used at UOXF) - both of which can be used for in-situ diffraction.
Figure: Typical arrangement of in-situ plate data collection at Diamond Beamline I24.
Crystal diffraction can be evaluated in-house using the PX-Scanner and in-situ data collection is now routinely available on I03, I04-1 and I24 beamlines at Diamond Light Source (Axford et al., Acta Cryst. 2012). Plate mounting and data collection architecture differs slightly between beamlines, but all can accommodate the SBS plate format, and 2 beamlines include robotic plate mounting. Rotation ranges vary from -10o to +38o (I03); -26o to +26o (I04-1); -20o to +20o (I24, see figure above).
Further developments
Crystallization drops set up with the CrystalQuick X plates tended to spread easily and then formed a thin film covering the whole of the platform. This problem could be overcome by working with Greiner Bio-one to prepare hydrophobically treated CrystalQuick X plates. This type of plate is now the preferred in-situ plate used at UOXF and is commercially available from Greiner Bio-one.
Tests were carried out to find low scattering sealing tapes. Although the scattering from tapes is generally much lower than from plates it is not insignificant. The ‘ThermalSeal RT 50μm’ tape from web-scientific was found to have very low scattering below 5.7Å, so should be better than the standard Greiner ‘Viewseal’ tape for many applications (see figure below). Images shown in figure 4 were recorded with the in-house X-ray source under identical conditions. The resolution at the edge of the detector was 2.6Å.
In parallel we have reported a significant increase in the lifetime of room temperature macromolecular crystals through the use of a high brilliance X-ray beam, reduced exposure times and a fast readout detector (Owen et al., Acta Cryst., 2012).
Grant Agreement no: 227764 P-CUBE Final Report
11
Figure: Comparison of scattering from CrystalQuick X plate and sealing tapes.
Grant Agreement no: 227764 P-CUBE Final Report
12
b) Dissemination
The ambition to enable dissemination of cutting-edge methodologies.
One of P-CUBE’s central missions has been the efficient training in the state-of-the-art technologies that were available at the transnational access infrastructures. Therefore, P-CUBE training activities primarily targeted users who were granted transnational access to the P- CUBE platforms.
This was achieved by offering or contributing to focused training courses in specific state-of-the-art techniques that were used at the platforms. During the 4 years of the project, P-CUBE organized or participated in ten successful training events covering topics such as production, purification and characterization of protein complexes, mammalian protein expression methods, protein crystallization methods as well as advanced protein engineering technologies. The courses and workshops were very well received by the participants who came from all across Europe and beyond, enthusiastically rated the courses and took home many new techniques and ideas.
Another emphasis was placed on yearly user meetings for platform users and operators. In this setting, former users were given the chance to present their scientific projects and to provide feedback about running the platforms and input on how to improve the services and training. Future users could get detailed information on the various platforms. They were able to learn from the experiences of former users and get in contact with the platform operators to discuss their projects. The platform operators informed the participants on the latest technological developments of their platforms, presented the scientific services that they offered and also gave some insight on successful user projects that had been performed at the respective infrastructures. P-CUBE user meetings were always held in combination with a practical workshop to maximize their attractiveness and outreach.
It was exciting and delightful to experience the quality of the scientific talks and projects by the platform users and the fruitful discussions with the platform operators at yearly P-CUBE user meetings held in Grenoble, Zurich and Heidelberg.
Transnational Access (TNA)
TNA Summary
P-CUBE offered and granted access to European scientists to essential methods and technologies in structural biology with the aim to offer tools and techniques for a better understanding of the three-dimensional structure of proteins and their functions. In particular, P-CUBE opened its infrastructures to the scientific community for (i) high-throughput cloning and expression of proteins in prokaryotic and eukaryotic cells, (ii) high-throughput crystallization, (iii) high throughput selection technologies and (iv) advanced light microscopy. In total 384 scientists from 30 different countries had the possibility to get access to and use the P-CUBE infrastructures for their scientific project. The scientific results obtained improved the general understanding of important processes. The following articles summarize the techniques used and provide examples of projects for which the specific technology was important.
Grant Agreement no: 227764 P-CUBE Final Report
13
Bacterial Cell Expression
During the course of the P-cube project the derivation of a series of expression vectors containing different fusion protein tags was reported (Bird LE. Methods. 2011 Sep; 55(1):29-37). These vectors were used in many of the TNA projects (e.g. figure below).
Eukaryotic cell expression
Automated DNA and protein expression described using the technologies above have been used for several published data;
1. Seiradake E et al 2013, ‘Structurally encoded intraclass differences in EphA clusters drive distinct cell responses’ Nature Struct Mol Biol (in press)
2. Bourhis JM et al 2012, ‘Structural basis of fibrillar collagen trimerization and related genetic disorders.’ Nature Struct Mol Biol 19, 1031-1036 (access visitor result)
3. Banci L et al 2013 ‘Atomic-resolution monitoring of protein maturation in live human cells by NMR.’ Nat Chem Biol. 9(5), 297-9 (access visitor result)
...and other results that are yet unpublished arising from Pcube access visits:
1. CHIT, crystals diffract to 1.2 A, were soaked with inhibitors, and the complex structure was obtained.
2. DBM, native and Seleno-Met labelled crystals diffracted and the structure is in the refinement stage.
3. PAPP-A, the crystals diffracted poorly, about 6A; optimisation in underway.
Expression technologies: Monitoring by microscopy
EMBL Heidelberg offered transnational access to the Protein Expression and Purification (PEPCore) and Advanced Light Microscopy (ALMF) platforms. It provided expert advice and technical support for fluorescence labelling of proteins, either through genetic encoding or by using chemical probes, and subsequent light microscopy experiments. Concerning protein labelling, EMBL HD offered
Figure: Expression of a series of constructs of a LysR family transcription factor from Streptococcus pyogenes. A series of 6 protein fusions were made (panel A) for each of 4 constructs of varying length and the expressed proteins analysed by SDS-PAGE (panel B). Abbreviations: POI = protein of interest; TRX = thioredoxin, MBP = maltose-binding protein, TF =trigger factor. (Project: Izabela Sitkiewicz, National Medicines Institute, Warsaw, POLAND)
Grant Agreement no: 227764 P-CUBE Final Report
14
various techniques including chemical fluorescent labelling with sequence tags (Flash/ReAsh, SNAP) and labelling using click chemistry (with unnatural amino acids). These methods are ideally suitable for studying the behaviour and function of proteins in living cells. Concerning the fluorescent light microscopy imaging, EMBL offered a number of techniques and microscopes, expert advice on planning and executing experiments as well as state-of-the-art methods for data processing and analysis. The techniques that were available included FRAP/FLIP, TIRF, FRET, FLIM and FCS/FCCS. Two examples from our TNA activities are given below:
User project: The user implemented labelling unnatural amino acids with click chemistry. She successfully prepared and genetically encoded a new unnatural amino acid featuring a bicyclononyne moiety that reacts remarkably fast with tetrazine-conjugated dyes. In a recent publication, she could encode the new unnatural amino acid inside mammalian cells and selectively label at low dye concentrations, which has proven to be very useful for imaging studies. This work has been published: Borrmann et al., Chembiochem. 2012.
User project: The user successfully implemented the available resources in her project analysing the study of the maturation pathway of dystroglycan, a member of the glycoprotein complex associated to dystrophin. Dystrophin is part of a protein complex that connects the cytoskeleton of a muscle fibre to the surrounding extracellular matrix through the cell membrane. In a recent publication, the user showed that ERp57, a member of the disulfide isomerase family involved in glycoprotein folding, is associated and colocalizes with the β-subunit of the extra cellular receptor dystroglycan. This work indicates that ERp57 is likely to be involved with the processing and maturation pathway of dystroglycan, which is still largely unknown. This work has been published: Sciandra et al., Exp. Cell Res. 2012.
Multiprotein expression in insect cells
MultiBac is a technology that can produce important protein complexes by using a special virus, called baculovirus. This virus can be used to reprogram insect cell cultures to produce even human protein complexes in large quantities. This technology has been used by various users.
User project: Dr. Dieter Palmberger from the University of Vienna, for example, used a PCUBE TNA visit to Grenoble to engineer SweetBac, a novel variant of the MultiBac virus, which can be used to put certain sugar groups on proteins to make them seem more human-like. This is important for example to produce therapeutic proteins such as antibodies, which need to appear like their human counterparts when they are administered to patients, in order to prevent allergic responses that could cause serious damage to the patient. The work has been published: Palmberger, D., PLoS ONE (2012).
The protocols, results and reagents have all been transferred to the laboratories of project proposers enabling the work begun in the OPPF-UK to be followed up. The training received by the visiting students will help in their scientific and technical development as researchers.
Grant Agreement no: 227764 P-CUBE Final Report
15
Tertiary technologies
ESPRIT
The ESPRIT technology allows scientists to identify solubly expressing forms challenging and poorly understood protein targets. The problem addressed is how to generate truncated sub-constructs of proteins corresponding to high-yielding, well-behaving protein domains (or multidomains). Typically biologists analyse protein sequences and try to predict rationally the domain locations, but this approach is frequently unsuccessful, notably when few similar proteins are available for comparison. ESPRIT takes a completely different approach, cutting the target into tens of thousands of random fragments and testing each one for its individual expression properties. To handle such large numbers of tests, robotics developed originally for genome sequencing have been adapted for expression screening purposes. P-CUBE visitors send their samples to EMBL Grenoble where, after initial preparation work, they then then come for 3 weeks of experimental time during which 28,000 constructs are tested. When successful, the visitor returns home with soluble proteins to continue structural or other studies.
Two highlights (note no-one wants to reveal they’ve expressed particular targets solubly, and the structural work is still in progress, so the target is not named).
A Ph.D. student from Leibniz Institut fuer Molekulare Pharmakologie Berlin had been unable to generate soluble material from a GPCR for NMR studies. The target was truncated randomly on both termini, and 28,000 constructs tested during a 3 week visit. The results included several previously undescribed fragments that were soluble and could be produced at milligram scales. This permitted protein labelling and analysis by 2D NMR, yielding important information on conformational properties that may be useful in understanding receptor activation. As important, the block on his project was resolved, permitting finalisation of his thesis and a manuscript.
A postdoc from Institut de Biomedicina-Universitat de Barcelona (IBUB) working on an orphan nuclear receptor was completely unable to produced soluble ligand binding domain until we applied ESPRIT. The output was a well-behaving construct that has been used in structural studies and in protein-protein interaction analyses with peptides from an interacting hub protein.
DARPINs
The University of Zurich offered its experience and its infrastructure to produce so called Designed Ankyrin Repeat Proteins (Darpin) that bind to specific structures inside a cell and interfere and inhibit the binding of other molecules. Such an approach plays an important role in medical applications as the molecules obtained can bind to proteins thereby inhibiting its normal function. The two examples given below show for what kind of topics the technology has been used.
In this project the aim was to select DARPins against highly dynamic structures within our cells, the tubulin filaments. They belong to a family of proteins, the cytoskeleton, which give cells their structure, but also allow them to migrate and change their shape when a signal from the outside arrives. DARPins specific for tubulin are helpful in studying their function.
The gp120 protein of the human immunodeficiency virus (HIV) is responsible to make contact with a target cell and initiates viral entry. In this study DARPins were selected to bind to a motif of gp120 which inhibited infection.
Grant Agreement no: 227764 P-CUBE Final Report
16
High-throughput Crystallisation technologies
The possibility to use the expertise and knowledge of experts in the field of protein crystallization was given by various P-CUBE partners. By using these infrastructures scientists have been able to find out about the structure of a protein. The structure again provides insights into the function of a protein and therefore reveals information that might be useful for future medical applications. Some projects that have been conducted for other scientists in the P-CUBE labs are given below.
User project: Many bacteria communicate between individual cells through the secretion of small signalling molecules. The signals are used to see if there is a quorum, i.e. a large enough population of bacteria to start colonization. This process can involve the formation of biofilms, and pathogenic bacteria use quorum sensors to initiate the production of virulence factors that destroy host tissue. The small molecules acting as signals are processed by enzymes within the individual bacteria, and this triggers alteration in gene expression. The Streit group from Hamburg University (Germany) determined the crystal structure of a novel class of quorum sensing enzymes. Initial crystallization trials resulted in several conditions were crystalline material was obtained that was not yet of sufficient quality to obtain an interpretable X-ray diffraction pattern. Several rounds of optimization helped to obtain larger well-diffracting crystals at the facility, and the structure helps to explain how this enzyme acts as a quorum sensor (Bijtenhoorn et al., PLoS One. 2011). The structure is now used to investigate if specific inhibitors can interfere with bacterial colonization, which is a promising alternative strategy for the eradication of pathogenic bacteria.
User project: The finite nature of fossil fuels dictates that we must look for sustainable renewable sources of liquid fuels. Consequently, there is a major thrust towards second-generation biofuel production from lignocellulosic feedstocks, either purpose-grown or as agricultural or municipal wastes, using bacterial fermentation processes. In collaboration with TMO Renewables Ltd, who have developed a fermentative bioethanol production process, the Danson group from the University of Bath, UK, is working on thermostable enzymes that can degrade xylan, namely xylanases and xylosidases.They have isolated a Xylosidase from a thermophilic bacterium and have crystallized and determined the first structure of the class of enzymes to which it belongs, revealing new features. Together with their industrial partner, the group is now analysing the structure to design mutations that could make the enzyme more efficient.
User project: Numerous animal species, from insects (mosquitoes) to mammals (vampire bats), feed primarily on fresh blood from their prey. These parasites produce some of the most potent antagonists of the blood clotting system known. Anopheles mosquitoes are widespread hematophagous parasites and important vectors of malaria, a potentially lethal disease that affects 500 million humans worldwide. Here, the authors show that a salivary polypeptide from Anopheles, anophelin, inhibits thrombin by binding in an orientation opposite to the orientation of substrates and disrupts thrombin’s catalytic machinery. This unique molecular mechanism has implications for the design of synthetic antithrombotics. This work has been published: Figuereido et al. PNAS 2012
User project: DNA Mismatch-repair factors have a prominent role in surveying eukaryotic DNA-replication fidelity. This function depends on the activity of protein complex formed by MutL-homolog and Mlh1. In humans, MLH1 mutations underlie half of hereditary nonpolyposis colorectal
Grant Agreement no: 227764 P-CUBE Final Report
17
cancers (HNPCCs). This work reveals the crystal structures of the MutLα (Mlh1–Pms1 heterodimer) C-terminal domain (CTD) from Saccharomyces cerevisiae, alone and in complex with fragments derived from Mlh1 partners. These structures provide a rationale for the deleterious impact of MLH1 mutations in HNPCCs. This work was published: Egueneau et al., NSMB 2013
User project: A project that benefitted from this approach was on a protein that is involved in bacterial quorum sensing. Some pathogenic bacteria use small molecules as a signal that indicates its population density. These signals are processed by enzymes within the individual bacteria, and this triggers alteration in gene expression and processes such as the formation of biofilms. The Streit group from Hamburg University (Germany) determined the crystal structure of a novel class of quorum sensing enzymes. Initial crystallization trials resulted in several conditions were crystalline material was obtained that was not yet of sufficient quality to obtain an interpretable X-ray diffraction pattern. The Optisalt screens helped to obtain larger well-diffracting crystals in an automatic fashion, and the structure confirmed the role of this enzyme as a quorum sensor. This work has been published: Bijtenhoorn P. et al, PLoS One. 2011.
User project: The C propeptides of fibrillar procollagens have crucial roles in tissue growth and repair by controlling both the intracellular assembly of procollagen molecules and the extracellular assembly of collagen fibrils. Mutations in C propeptides are associated with several, often lethal, genetic disorders affecting bone, cartilage, blood vessels and skin. The crystal structure of a C-propeptide domain from human procollagen III was determined in collaboration with the group of D. Hulmes in Lyon, France (see figure below). The structure reveals an exquisite mechanism of chain recognition during intracellular trimerization of the procollagen molecule. It also gives insights into why some types of collagen consist of three identical polypeptide chains, whereas others do not. The data show striking correlations between the sites of numerous disease-related mutations in different C-propeptide domains and the degree of phenotype severity. The results have broad implications for understanding genetic disorders of connective tissues and designing new therapeutic strategies. This work has been published: Bourhis JM et al. Nat Struct Mol Biol., 2012.
Figure: Structure of the C-propeptide trimer of human procollagen III with secondary structure elements of chain B indicated.
Grant Agreement no: 227764 P-CUBE Final Report
18
User project: It has proved difficult to classify viruses unless they are closely related since their rapid evolution hinders detection of remote evolutionary relationships in their genetic sequences. However, structure varies more slowly than sequence, allowing deeper evolutionary relationships to be detected. Bacteriophage P23-77 is an example of a newly identified viral lineage, with members inhabiting extreme environments. In collaboration with the group of J. Bamford, Jyvaskyla, Finland we have solved multiple crystal structures of the major capsid proteins VP16 and VP17 of bacteriophage P23-77 (see figure below). They fit the 14 Å resolution cryo-electron microscopy reconstruction of the entire virus exquisitely well, allowing us to propose a model for both the capsid architecture and viral assembly, quite different from previously published models. The structures of the capsid proteins and their mode of association to form the viral capsid suggest that the P23-77-like and adeno-PRD1 lineages of viruses share an extremely ancient common ancestor. This work is in press: Rissanen, I et al. Structure 21, in press, May 2013.
Figure: Capsid Organisation of P23-77
(a) Crystal structures of the major capsid proteins VP16 (orange) and VP17 (green) fitted into the cryo-EM electron density map of the P23-77 virion. This work is published: Jaatinen, ST et al., Virology, 2008.
(b) Model of the whole P23-77 capsid with the asymmetric unit and the symmetry axis marked,
(c) Details of the arrangement of VP16 and VP17 within the asymmetric unit.
This work will be published: Ren J, et al Nature Comms 2013 (in press).
Grant Agreement no: 227764 P-CUBE Final Report
19
HTX capillary crystallisation and eukaryotic cell expression applications
CrystalQuick X plates have already been used successfully for the following viruses:
ERAV
EV71 (Wang et al., NSMB 2012)
BEV-1
BEV-2
FMDV (Porta et al., PLoS Pathog., 2013)
CAV16 (Ren et al., Nat. Comms., 2013)
Figure: Crystal structure of FMDV serotype A22 with the pentamers and the α-helix at the two-fold symmetry axis highlighted.
Most recently crystals of empty capsids of Foot and Mouth Disease Virus (FMDV) were grown in CrystalQuick X plates and data were collected in-situ at Diamond, Beamline I24. The structures of A22-wt and A22-H2093C helped in the rational engineering of recombinant picornavirus capsids to produce safe, protective vaccine antigen. The figure shows the structure of A22-wt with His-93 at the interface between the pentamers. In A22-H2093C this histidine is changed to a cysteine thereby forming a disulphide bond between the pentamers and forming a more stable capsid.
Potential Impact:
b) Dissemination
The ambition to enable dissemination of cutting-edge methodologies.
One of P-CUBE’s central missions has been the efficient training in the state-of-the-art technologies that were available at the transnational access infrastructures. Therefore, P-CUBE training activities primarily targeted users who were granted transnational access to the P- CUBE platforms.
This was achieved by offering or contributing to focused training courses in specific state-of-the-art techniques that were used at the platforms. During the 4 years of the project, P-CUBE organized or participated in ten successful training events covering topics such as production, purification and characterization of protein complexes, mammalian protein expression methods, protein crystallization methods as well as advanced protein engineering technologies. The courses and workshops were very well received by the participants who came from all across Europe and beyond, enthusiastically rated the courses and took home many new techniques and ideas.
Another emphasis was placed on yearly user meetings for platform users and operators. In this setting, former users were given the chance to present their scientific projects and to provide feedback about running the platforms and input on how to improve the services and training. Future users could get detailed information on the various platforms. They were able to learn from the experiences of former users and get in contact with the platform operators to discuss their projects. The platform operators informed the participants on the latest technological developments of their platforms, presented the scientific services that they offered and also gave some insight on successful user projects that had been performed at the respective infrastructures. P-CUBE user meetings were always held in combination with a practical workshop to maximize their attractiveness and outreach.
It was exciting and delightful to experience the quality of the scientific talks and projects by the platform users and the fruitful discussions with the platform operators at yearly P-CUBE user meetings held in Grenoble, Zurich and Heidelberg.
c) Impact
All dissemination activities are described in Section b of this report.
In recent years, new pathogens (like viruses or bacteria) frequently emerged that were and are a threat to human or environmental health. Next to the yearly occurrence of new influenza viruses, the outbreak of SARS (Severe acute respiratory syndrome, a pandemic with almost 1,000 deaths in 37 countries in early 2003) or the recent enterohemorrhagic E. coli (EHEC) pandemia clearly showed that the scientific community must be able to react instantly and without delay to these threats, providing binders for better detection and identification of potentially infected patients. Therefore, the impact of our work will be to provide the selected binders against the present targets, but more importantly, to supply a robust selection and screening platform that can be the basis for the prompt selection of highly specific and highly selective DARPin binders against future targets, requiring reduced cost and time investments compared to previous methods. The binders that can be selected with our system will be of great importance for various future projects, ranging from basic biology like e.g. studies of fundamental questions of protein folding or cell signalling cascades to the more applied fields of
diagnostics and therapeutics. Thus, we believe that the true value of our development within the P-CUBE program, despite being already noticeable at the current point, will further grow to its full extent in the years to come.
FCCS is a very promising fluorescence light microscopy technique and is expected to be very useful for precise and detailed analysis and characterization of cellular protein interactions in living cells. Since it is performed in vivo and in live cells, the obtained results provide an unbiased view of cellular interactions that otherwise cannot be obtained. Better understanding the assembly process of multi-protein complexes is critically required to optimize their expression in eukaryotic cells. In general, the FCCS technique has a number of applications, but it is particularly interesting for biomedical fields, where the analysis of quantity, interaction partners or the aggregation state of a protein in the cellular context is important to understand a particular defect and the associated disease.
Applying library strategies to complex problems generates solutions that are difficult to predict rationally. Here we have applied this logic to the generation of soluble variants of challenging and poorly understood protein targets. This has clear applications in generating proteins for pharmaceutical applications (structural biology, high throughput screening) and vaccinology; evidenced by the industrial collaborations performed during this period with world-leading companies on targets such as protein kinases and pathogen proteins for immunisation. The results from this work have been disseminated via 4 peer-reviewed publications and regular invited presentations at industrial and academic conferences.
The automation of eukaryotic expression methodology has been widely adopted as elements in existing pipelines at the UOXF facility. The expression robot is a large high value piece of equipment which is not suitable for every laboratory and therefore, while the principles have been adopted, the scale of automation has not. However, automation in general has produced reproducible, standardised high quality expression in large volume which enables fast progress from construct design to protein expressed. This has been exemplified for several of the access visitors to the UOXF facility, where the aim of the access has been to express a difficult protein or protein complex, and the visitor has been able to go further to express, crystallise and solve the structure of their protein in very short time (<1 week). This is only possible due to automation which allows 24hr cell culture runs and fast purification and crystallisation screening. The societal impact of this work is the improved skills of mainly young researchers; the adoption of new technologies in laboratories following exposure at the access host laboratory; the impact of potentially important biological data derived from the protein structures solved and the knowledge outreach following publication.
The impact of crystallisation developments is mainly in the improved time and quality efficiencies of collecting X-ray data directly from crystals in plates, without the need to mount crystals (a time consuming manual task). This translates into cost efficiencies at the synchrotron. Data quality is maintained and in particular, large numbers of microcrystals can be quickly screened in plates. The societal impact of this work is the improved skills of mainly young researchers; the adoption of new technologies and data capture methods at the synchrotron; the impact of potentially important biological data derived from the protein structures solved and the knowledge outreach following publication.
List of Websites:
www.p-cube.eu