EMBL@XFEL Background

Brief introduction to biological imaging on FEL

The European XFEL facility in Hamburg will provide high flux coherent pulsed X-ray radiation with properties superior to existing light sources, Figure A.

The beam will be provided in short pulses (100 fs or shorter) at a 4.5 MHz repetition rate (220 ns between the pulses), Figure B. The pulses will be assembled into a 0.6 ms train and there will be 10 trains delivered each second. In the full-load mode there will be up to 27,000 pulses emitted per second. The technical feasibility of providing pulses shorter than 10 fs has already been shown at the LCLS facility.

Simulations on a single lysozyme molecule (Neutze et al. 2000) suggested that radiation damage will not significantly disrupts the initial positions of the atoms within a protein in the first 10 - 20 fs of exposure, Figure C. Over this time, the incident X-ray beam will hit the sample and, travelling at a speed of 300,000 km/s, will travel 3 to 6 µm. All photons scattered from a sample with a size less than this distance will result in a coherent diffraction image measured on a detector. This effect is known as 'diffraction before destruction' since the damage, incurred by the radiation will occur after the diffracted radiation has left the sample.

The collected image is a molecular transform of the sample, 'crystallography without crystals', for which the initial phases can be retrieved numerically, subject to the statistical strength of the recorded signal. Reconstruction of a coherent diffraction image from an object has been shown both theoretically and experimentally (Sayre, 1952; Miao et al. 1999; Chapman et al. 2006; Raines et al. 2010 and references therein). In essence, if the continuous diffraction pattern is sampled at sufficient spacing a real-space image can be reconstructed. For similar objects, repetition of the experiment followed by alignment of the reconstructions, will yield a full 3D model (cf. tomography).

Images for Figures A and B are taken from XFEL Technical Design Report and Figure C from Neutze et al. (2000).

Examples of biological experiments at XFEL

Single-particle diffractive imaging of large multimeric protein complexes and organelles

A great deal of focus in modern biology is the characterisation of large multimeric protein assemblies to provide unprecedented information to understand the complex biological processes that support life. Biological complexes that are rapidly purified using current methods are generally not stable for extended periods of time and are often highly sensitive to freeze-preservation techniques. For example, a recent publication (Akiyoshi et al. 2010) demonstrates that affinity purification methods can yield highly pure kinetochore, a complex of over 90 monomeric units, from natural sources. The recombinant production of such a complex is currently not possible. Membrane proteins that perform the physiological functions of transport of substrates are another class of attractive targets for single-particle diffraction study as they are usually of low abundance and difficult to crystallise. In particular, human ion channels and secondary active transporter proteins are of high biomedical interest. Sub-cellular fractions or organelles, such as microtubules, peroxisomes, mitochondria and nuclei, are often equally dynamic. To make the most efficient use of XFEL beamtime it is essential to adopt a high-throughput approach for the purification of such samples.

For rapid purification of biomolecules (protein complexes from both recombinant and in vivo sources, cell organelles and whole cells) high-throughput approaches suitable for a wide variety of samples are needed. A potential example of such a system is the Dynal-bead system, which can be easily adapted for multiple purification methods. The concept of correlative light and XFEL imaging is particularly attractive. Super-resolution microscopy at 10 nm would be an ideal tool to prescreen well-formed megacomplexes or organelles. Additionally, laboratory facilities with suitable biochemical and proteomic instrumentation will be necessary.

For quality control of purified samples automated capillary electrophoresis or silver staining can be applied. Depending on the experiment the samples may be further characterised by static or dynamic light scattering (SLS/DLS) and purified by fast field flow fractionation (FFF), micro-size exclusion chromatography (µSEC). Immediate access to mass spectrometry could be used for in-line sample characterisation as part of a decision-making process on which samples to pass forward for XFEL investigation and will also allow confirmation that transient complexes have been correctly isolated. Ultimately, we aim to integrate mass spectrometry and other separation methods into sample injectors. The ideal infrastructure would allow all upstream characterisation processes to be linked directly to a sample injection device, allowing characterisation and sample selection immediately prior to measurement. As most of the possible purification and quality control analyses are performed in solution they are directly compatible with both "aerosol" and "liquid jet" methods of sample delivery.

Whole cell/tissue imaging and cell culture

In order to address the myriad of biological cellular processes it is necessary to identify those samples that are in a suitable condition for analysis. Local cell culturing and visible light/fluorescence imaging will be used for the characterisation of such samples, and for rapid structural pre-screening a super-resolution microscope will be necessary. A cell culturing facility will allow external researchers to bring seed samples under standard storage/transport conditions and establish an appropriate propagation protocol. Such experiments may typically require days or weeks of initial experiments and could be supported by facilities currently available at the EMBL Hamburg Unit or elsewhere on the DESY campus. It is envisaged that smaller scale culturing facilities and correlative imaging suite directly prior to the experiment will provide a major benefit for FEL analysis of whole cell samples. In the ideal case such imaging should be integrated within suitable XFEL instruments, so that a sample of interest can be directly transferred for analysis. This could be performed through the combination of a microfluidic imaging stage with an adaptation of the "liquid jet" injector. The sample would be identified by microscopy, transferred to a fluid flow that is fed to the "liquid jet" injector, bringing the sample to the XFEL within seconds of identification. The study of spatial/temporal features of cellular structures of pathogens requires their cultivation, ideally in the vicinity of the XFEL instrument.

Nanocrystal production and handling

In many cases large protein complexes only produce small, weakly diffracting crystals that are suboptimal for analysis, even at 3rd generation synchrotron sources. Recent results obtained at LCLS have highlighted the potential for the use of femto-second pulses of FEL radiation in the investigation of protein structures (Chapman et al. 2011). A challenge remains in the preparation and pre-characterisation of sufficiently homogeneous nanocrystals. Here high-throughput crystallisation facilities (HTPX: such as the 10,000-plate storage capacity already available to the international community at EMBL Hamburg), in combination with the "liquid jet" injector technology offer the potential for large numbers of conditions to be screened to generate nanocrystals.

Our experience with the EMBL HPTX for the production of large crystals as needed for MX, indicates that they often result in micro-crystalline material. This material is unsuitable for standard MX but may be suitable for X-FEL. Counter diffusion often produces gradients of crystals sizes. Those gradients could be eventually refined to generate large amount of nanocrystals. This may also be useful to transfer micro-crystalline material into the pulse/beam. Smaller crystals tend to be mechanically weaker and more sensitive to fluctuations in temperature, resulting in an increased risk for the crystals to re-dissolve. An important factor is the necessity to harvest these crystals at the most appropriate moment, as they are often stable only for a certain time, even in the condition under which they are obtained (Mayerhofer et al. 2010).

2D membrane protein crystals can be grown in a lipid bilayer. Although 2D crystals usually produce lower resolution, they have the advantage that less protein is required and that the distinct conformations of the protein, such as open and closed state of an ion channel may be detected. Small size 2D-nanocrystals may be obtained in the presence of ligands, such as regulators of activity, and pharmaceutical agents. The obtained structures of such membrane proteins will provide an important framework for functional characterisation and for the identification of potential drug target sites

Capturing transition state intermediate structure

The presence of ligands in macromolecular crystals and their interactions with proteins and nucleic acids is often crucial to the understanding of a system. There is evidence that the inclusion of protein and ligand dynamics would greatly improve our ability to develop new drugs (Fuentes et al. 2011). We envisage this to be desirable for users both from academia and industry. High flux and ~5 MHz pulse rate of the XFEL, together with next-generation detectors, will allow the analysis of protein-substrate interactions over short (micro-second and shorter) reaction time scales. Some systems (for example those associated with photosynthesis) can be examined by laser activation of 'caged' ligands.

All enzymatic reactions involve the diffusion and interaction of reactants with each other, which is a handicap in time-resolved studies on diffusion-limited reactions. One possibility for lowering diffusion barriers is to significantly reduce the crystal size. With very small crystals, the vast majority of solution kinetic techniques and methodologies will become available, including rapid mixing (e.g. stopped-flow experiments). This brings an entirely new range of biological systems within the scope of detailed time-resolved structural studies. Ultrafast kinetic studies will require a UV-VIS-Near IR and MID-IR pump-probe laser to measure transient absorption and associated 2D/3D IR femtosecond spectroscopy. Any effort in this will require the development of microfluidic devices.

Thus, the main tasks are the preparation and/or suitable storage of samples prior to the experiment, the timely activation of the system and the delivery of the activated complex to interact with the FEL pulses providing high time-resolution snapshots. Suitable "pulse request" modes would provide the appropriate FEL radiation at the appropriate time point.