Megahertz serial crystallography at the European XFEL

The first experiment at the European XFEL proved it can provide high quality structures

The new European X-ray free electron laser (XFEL) is an international collaboration between 12 countries: Denmark, France, Germany, Hungary, Italy, Poland, Russia, Slovakia, Spain, Sweden, Switzerland, and the United Kingdom. Construction on the project began in 2009, and the facility welcomed its first users in September 2017. Housed in 3.4 km of tunnels underneath Germany, the European XFEL is the first XFEL designed to deliver X-ray pulses with a megahertz inter-pulse spacing, more than four orders of magnitude higher than previously possible. However, it was unclear whether it would be possible to measure high-quality data at megahertz pulse repetition rates because of the damaging effects of the X-ray pulse on the liquid jet delivering fresh sample into the interaction region. Allen Orville, Henry Chapman, and Anton Barty were asked to run community-based experiments for the first beamtime sessions, and they spearheaded ground-state and time-resolved experiments in structural biology. Some of their results, recently published in Nature Communications, show that high-quality structures can be obtained using the 1.1 MHz repetition rate at the European XFEL. 

X-ray serial crystallography at the European XFEL

Megahertz serial crystallography. Pulses from the European XFEL were focused on the interaction region using a set of Beryllium lenses. Protein crystals in crystallization solution were introduced into the focused XFEL beam using a liquid jet of 1.8 µm diameter moving at speeds between 50 m/s and 100 m/s. Diffraction from the sample was measured using an AGIPD, which is capable of measuring up to 3520 pulses per second at megahertz frame rates. In-situ jet imaging (inset) showed that the liquid column does explode under the X-ray illumination conditions of this experiment using a jet with a speed of 100 m/s, but that the liquid jet recovered in less than 1 μs to deliver fresh sample in time for arrival of the next X-ray pulse. Images and movies of jets at different speeds are included in the supplementary material
Megahertz serial crystallography. Pulses from the European XFEL were focused on the interaction region using a set of Beryllium lenses. Protein crystals in crystallization solution were introduced into the focused XFEL beam using a liquid jet of 1.8 µm diameter moving at speeds between 50 m/s and 100 m/s. Diffraction from the sample was measured using an AGIPD, which is capable of measuring up to 3520 pulses per second at megahertz frame rates. In-situ jet imaging (inset) showed that the liquid column does explode under the X-ray illumination conditions of this experiment using a jet with a speed of 100 m/s, but that the liquid jet recovered in less than 1 μs to deliver fresh sample in time for arrival of the next X-ray pulse. Images and movies of jets at different speeds are included in the supplementary material
Nature Communications volume 9, Article number: 4025 (2018)
The European XFEL is the world's largest X-ray laser, generating tens of femtosecond (fs)-long X-ray flashes that repeat at 27,000 times per second overall, and with up to a 4.4 MHz intra-train pulse frequency. The XFEL brilliance is a billion times higher than the best conventional X-ray radiation sources. The high brilliance and tight focus of the beam make it possible to collect single shot diffraction images from slurries of micron-size crystals, a technique first demonstrated at the LCLS in Stanford, USA and termed Serial Femtosecond Crystallography (SFX). SFX is particularly useful for measurements at physiologically-relevant temperatures or time-resolved studies of protein crystals. 
 
The fs pulse duration of XFELs allow researchers to investigate the atomic and electronic structures from nano- to micron-size crystals. However, the X-ray pulse causes the sample to explode, and therefore, one of the challenges of SFX is to deliver new samples to the beam rapidly enough for every pulse to be used. Another is efficiently measuring diffraction data from the large number of individual crystals required for the SFX approach - a complete dataset requires collecting the structure of tens of thousands of crystals.
 

 

The European XFEL produces bursts of X-ray pulses at a megahertz repetition rate within a train, repeating at 10 Hz frequency. At the current intra-bunch repetition rate of 1.1 MHz the pulse spacing is less than 1 μs, nearly four orders of magnitude shorter than previously available. The decreased time between X-ray pulses delivers more pulses per second while maintaining the same X-ray peak power, but simultaneously poses several challenges for SFX. 

 

Megahertz serial crystallography proof-of-concept

The first experiment at the European XFEL was also the first ever to collect SFX data at over 1 MHz, a rate that would be equivalent to 1.1 million pulses if the pulse rate was continuous. Typically, samples are injected into the interaction region as liquid jets; how fast must the jet be to feed this new source? The X-ray beam causes a shockwave to travel up the liquid jet; does this damage the protein crystals on their way to the interaction region before SFX data can be measured? When you’re conducting an experiment with a microcrystal ‘slurry’ and a pulsed X-ray beam just 10-15 microns in size, there are obvious challenges in making sure they coincide in both time and space.

The team showed with direct imaging of the liquid jet using a stroboscopic laser that the XFEL pulse initially vaporises the jet, but the liquid column recovers in time for the next X-ray pulse for jets with a diameter of less than 2 μm and speeds between 50 and 100 m/s, while jets with a speed of 25 m/s do not recover in time.

The researchers were able to produce two complete data sets, one from the well-known model system lysozyme and the other from a previously unknown complex of a β-lactamase from K. pneumoniae. This enzyme belongs to the extended spectrum β-lactamases (ESBLs) that play an important role in emerging multi-antibiotic resistance mechanisms. These result open up megahertz SFX as a tool for reliable structure determination, substrate screening and the efficient measurement of the evolution and dynamics of molecular structures using megahertz repetition rate pulses.

The European XFEL has already performed experiments at 600 pulses per second at 1.1 MHz pulse rates. The instrument has been tested using pulse trains enabling 1760 pulses per second and 3520 pulses per second. It is planned to be available for users in 2019. When the number of pulses in the pulse train is increased to match the maximum AGIPD detector frame rate of 3520 frames per second, the β-lactamase measurements presented here could be completed in under a minute assuming a high crystal hit ratio.

The UK XFEL Hub

The planned LCLS-II-HE facility promises up to 106 equally-spaced pulses per second, increasing the rate of structure determination even further and enhancing the opportunities for rapid screening for drug targets using on-the-fly substrate mixing. Indeed, rapid data acquisition has the potential for generalising time-resolved movies of macromolecules catalysing their native reaction(s) at physiological temperatures.

Images of interaction of the EuXFEL liquid jet for the first 5 pulses in a train. Jets in the range of 50–100 m/s recover in time for the next pulse (first three rows), whereas slower jets of the type commonly used at LCLS do not recover in time for the next XFEL pulse at MHz repetition rates (bottom row). The bottom line provides linkage back to the results presented in presented in Chapman, et al. Nature 470, 73–77 (2011). Red line shows the intersection point with X-ray pulses. Images obtained by synchronised laser back illumination. Movies with finer time steps are included as supplementary material
Images of interaction of the EuXFEL liquid jet for the first 5 pulses in a train. Jets in the range of 50–100 m/s recover in time for the next pulse (first three rows), whereas slower jets of the type commonly used at LCLS do not recover in time for the next XFEL pulse at MHz repetition rates (bottom row). The bottom line provides linkage back to the results presented in presented in Chapman, et al. Nature 470, 73–77 (2011). Red line shows the intersection point with X-ray pulses. Images obtained by synchronised laser back illumination. Movies with finer time steps are included as supplementary material
Nature Communications volume 9, Article number: 4025 (2018)

However, unlike at a synchrotron, one SFX experiment uses the whole XFEL facility. Consequently, beamtime is both expensive and hard to get (with beamtime granted after rigorous international peer-review). It’s vitally important for users to maximise the data they get from their visit, and yet few users have experience conducting XFEL experiments.

One of the roles of the XFEL Hub at Diamond is to help the UK user community put together the best proposals for XFEL beamtime. Almost all of the proposals received at all XFEL facilies are very good, but the success rate is only about 20% because of the limited beamtime; some facilities are reporting a more than 10-fold oversubscription demand for their instruments. The XFEL Hub works to help users create a successful proposal.
 
The Hub can also help users to carry out preliminary, supporting and/or complementary experiments at Diamond. Hub staff often go to the XFEL with users to help with data collection and analysis, and can even provide support right through to publication. It’s up to the user groups to decide what support they want and how deeply to collaborate with Hub staff.
 
As Principal Scientist Allen M. Orville explains, “When users try to grow crystals of a new protein, it’s a process of trial and error. Their first crystals are usually a shower of microcrystals, which weren’t previously useful. It was necessary to optimise the conditions in order to produce one large crystal, which could be mounted in a synchrotron beam and rotated to give one complete dataset. Serial crystallography is different, and “easier” in the respect that it can use the shower of microcrystals and deliver high quality results. Passing that information on to users can save them weeks, or even months, of unnecessary work.”

To find out more about the XFEL-Hub, or to discuss potential applications, please contact Principal Scientist Allen Orville: allen.orville@diamond.ac.uk 

Related Publications:

 Wiedorn MO et al. Megahertz serial crystallography. Nature Communications 9, 4025 (2018). DOI:10.1038/s41467-018-06156-7.