Date Published: August 01, 2018
Publisher: International Union of Crystallography
Author(s): Hanna Kwon, Patricia S. Langan, Leighton Coates, Emma L. Raven, Peter C. E. Moody.
The application of the cryogenic data-collection environments used in protein X-ray crystallography to neutron protein crystallography is discussed.
Macromolecular X-ray crystallography provides a wealth of three-dimensional structural information that greatly contributes to our understanding of protein function and provides insight into enzyme reaction mechanisms. The discovery by Dorothy Crowfoot (Bernal & Crowfoot, 1934 ▸) that protein crystals, when kept hydrated by being sealed in thin-walled glass capillaries along with their mother liquor, retain their crystalline order and thus could be used for X-ray diffraction experiments enabled protein crystallography. This method of mounting protein crystals remained the norm until the mid-1990s. Mounting crystals in this way meant that X-ray data collection could be carried out conveniently at room temperature. However, this method meant that any chemical reactions induced by X-rays would be propagated, resulting in damage. Cooling capillary-mounted crystals to below the freezing point of the mother liquor would cause ice-crystal formation, damaging the crystal and giving overwhelming ice diffraction. More modest cooling in the capillary (for example to 258 K with 2.8 M ammonium sulfate) to slow radiation damage (Skarżński et al., 1987 ▸) and to catch labile catalytic intermediates (Moody, 1984 ▸) was not often employed and frequently produced problems with condensation. The susceptibility of crystals to radiation damage often meant that several crystals were needed to collect a complete data set, and each would need to be carefully aligned to avoid unnecessary duplication and waste. For small molecules that did not suffer from loss or change in phase of solvent, a boiled-off liquid-nitrogen apparatus was developed, allowing temperatures near 80 K to be used (Post et al., 1951 ▸). The stability of protein crystals for long periods once sealed in capillaries and the lack of damage from photochemistry from neutron irradiation meant that this was the preferred method to mount protein crystals for neutron crystallography.
Although the collection of neutron data at cryo-temperatures is not normally required to extend the lifetime of the crystal in the beam, it does have particular uses and advantages. The lack of radiation damage and the much more detailed description of the water structure means that neutron crystallography provides an opportunity to properly compare room-temperature and cryo structures. As for X-rays, cryo-cooling of the crystals can reduce atomic movement, thus improving the resolution. The application for neutrons of the techniques used in X-ray work for trapping catalytic intermediates allows the proton (or deuteron) positions to be seen, greatly enhancing the mechanistic information available. Cryogenic data collection allows the study of any proteins and protein complexes (including those with ligands and nucleic acids) that are not stable enough at room temperature for the prolonged data-collection period required for neutron crystallography. To date, very few neutron structures at cryogenic temperatures have been deposited in the Protein Data Bank (PDB). The first pioneering low-resolution neutron cryogenic neutron work was conducted on myoglobin, studying the solvent structure as a function of temperature (Daniels et al., 1996 ▸). A further low-resolution cryogenic structure of myoglobin was published the following year at a resolution of 5 Å (Daniels et al., 1997 ▸). These early studies focused on the hydration layers surrounding the protein in the crystal.
Although cryogenic data-collection capabilities have been available at previous generations of neutron protein crystallography (NPX) instruments, for example, use of the cryogenic data-collection facilities at the recently decommissioned Protein Crystallography Station (PCS) at Los Alamos National Laboratory (Schoenborn & Langan, 2004 ▸; Langan et al., 2008 ▸) was never reported during its 15 years of operation (Chen & Unkefer, 2017 ▸). One possible reason for this is the difficulty associated with successfully cryocooling the larger crystals needed for neutron diffraction and maintaining them at 100 K over the many days required to collect a complete neutron data set. More recently, the deployment of equipment and facilities for 100 K cryogenic neutron data collection has been a common theme in upgrades at NPX instruments around the world (Coates et al., 2014 ▸, 2015 ▸). Although NPX instruments at spallation and reactor sources use significantly different instrument designs and detector technologies, they share several characteristics and it is worth reviewing them here. Owing to the lower flux and longer wavelengths used for neutron diffraction compared with X-ray diffraction, a large solid-angle detector coverage is highly desirable to reduce the number of crystal orientations needed for a complete data set. Therefore, a nitrogen cryostream is often positioned above a vertically mounted goniostat and rotation axis such that the sample mount makes an angle close to 180° with the cryostream. This has been found to help to reduce ice formation on the sample pin during the extended time required for neutron data collection. The incoming neutron beam is then horizontal to the sample position to form a 90° angle with the cryostream. One benefit of large angular detector coverage is that the detectors and the infrastructure used to support them often help to isolate the sample from the surrounding atmosphere, preventing drafts and humid air from reaching the sample position. The upgraded LADI-III instrument (Blakeley et al., 2010 ▸) at the at the Institut Laue–Langevin, the iBIX instrument at the Japanese Particle Research Accelerator Complex (J-PARC; Kusaka et al., 2013 ▸) and the BIODIFF instrument at the FRM II research reactor have deployed similar equipment for cryogenic neutron data collection (Coates et al., 2014 ▸), which has been utilized in recent studies (Casadei et al., 2014 ▸; Kwon et al., 2016 ▸). A different approach using a liquid-helium Displex cryorefrigerator is also possible. This was used originally on LADI-I (Myles et al., 2012 ▸; retained as a 15 K option on LADI-III) and has recently been deployed on the IMAGINE instrument (Meilleur et al., 2013 ▸). It enables lower temperatures to be reached (4 K) but requires the use of an aluminium container that blocks direct viewing of the sample, complicating the alignment of the crystal into the neutron beam. The cooling of the crystal by conduction also prevents the use of standard X-ray mounts. However, it does enable different areas of science besides NPX to be served by the same instrument.
Haem peroxidases catalyse the reduction of peroxide to water, using high-oxidation-state intermediates (compound I and compound II) to activate peroxide and oxygen. It has been difficult to study this activation using X-ray crystallography as the photoelectrons generated during X-ray exposure reduce many of the chemically interesting high-valence species. In order to understand the mechanisms of the peroxidase, it is necessary to see the protonation states of these intermediates (Fe=O or Fe—OH). Early work on haem peroxidases using X-ray crystallography and various spectroscopic methods reported varying results on the identity of the haem ligand extrapolated from bond order based on the Fe—O distances, rather than direct visualization, owing to the limitations of X-ray crystallography (Fülöp et al., 1994 ▸; Bonagura et al., 2003 ▸; Gumiero et al., 2011 ▸). These distances were difficult to establish with confidence, as well as the strongly electropositive iron being susceptible to reduction by photoelectrons (an effect often ignored in these and other redox enzymes). Furthermore, the close proximity of a relatively electron-dense Fe atom and a lighter O atom means that series-termination errors in the Fourier transform are likely to distort the distance between these atoms (Fülöp et al., 2000 ▸). When the data are incomplete, of low resolution or of poor quality, and if the models are incomplete or have high B values, the uncertainties in atomic positions are also increased (Murshudov & Dodson, 1997 ▸). This difficulty in measuring the distances with the accuracy required (∼0.2 Å) meant that it was difficult to identify the ligand species conclusively.
The first cryogenic NPX experiments were conducted in the 1990s (Daniels et al., 1996 ▸) at low resolution, followed by a 15 K structure in 2004 (Blakeley et al., 2004 ▸). In the last four years the collection of cryogenic neutron data has become more feasible in part owing to improved sources and instrumentation, which enable faster data-collection times. Perdeuteration of the protein increases the scattering power of the crystal and reduces the background emanating from the sample, enabling the use of smaller crystals for data collection. Small crystals are more easily cryocooled successfully and cryogenic data collection will become more routine as more advanced and powerful NPX instruments are constructed such as the Ewald instrument at the second target station at the SNS (Coates & Robertson, 2017 ▸) and the NMX diffractometer at the European Spallation Source. The lack of radiation damage by neutrons makes them ideal for collecting data at room temperature, but for a subset of experiments the ability to freeze-trap a short-lived intermediate enables NPX to address interesting scientific questions.