Sample Preparation and Transfer Protocol for In-Vacuum Long-Wavelength Crystallography on Beamline I23 at Diamond Light Source

Long-wavelength X-ray diffraction is used to harness the anomalous scattering properties of specific light atoms natively present in macromolecules. This helps to solve the crystallographic phase problem and to unambiguously confirm the identity and location of such elements within macromolecules. While in the early days of macromolecular crystallography, de novo structures were solved by multiple isomorphous replacement¹, with the advent of tunable X-ray beamlines at synchrotrons, experimental phasing based on multi-wavelength and single-wavelength (SAD) anomalous diffraction techniques have become the dominant methods². Both methods have historically relied on the isomorphous or anomalous signal from heavy metals, which need to be artificially introduced into the crystals by co-crystallization or crystal soaking³. The trial-and-error approach and unpredictable outcome can make these experiments frustratingly time-consuming. The incorporation of seleno-methionine during protein expression⁴ is a very elegant way to overcome these limitations and exploit anomalous diffraction at short wavelengths, although it can be very challenging in eukaryotic protein expression systems.

Long-wavelength MX is extremely appealing for structure determination by native SAD experiments^5,6 due to the convenience of using crystals directly from a successful crystallization trial without further treatment. Additionally, access to the absorption edges of elements of high biological importance, such as calcium, potassium, chlorine, sulfur, and phosphorus, opens the opportunity to directly identify the positions of these elements in macromolecules^7,8,9,10. At medium and low resolution, element assignment based on the 2F_o-F_c electron density and chemical environment can be difficult, particularly for elements with similar number of electrons or weakly bound ions with partial occupancies. These ambiguities can be resolved by collecting data below and above the absorption edge of the element of interest and interpretation of the resulting model-phased anomalous difference Fourier maps^11,12. Locating sulfur atom positions in these maps can also aid model-building into low-resolution electron density maps¹³. The absorption edges of these light elements are observed at wavelengths between λ = 3 and 6 Å (see Figure 1, top). This wavelength range has been well beyond the capabilities of any synchrotron MX beamline, and efficient operation in this range requires overcoming several technical challenges, as outlined below.

Beamline I23 at Diamond Light Source, UK, is a unique instrument, specifically designed to facilitate long-wavelength MX experiments, tunable in a wavelength range between λ = 1.13 and 5.9 Å (energy range between E = 2.1 and 11 keV). By operating in a high-vacuum environment¹⁴, air absorption and scattering are eliminated, consequently enhancing the efficiency of diffraction experiments and the signal-to-noise ratio. A large vacuum endstation encloses all the components of the sample environment, including the semi-cylindrical Pilatus 12M detector, a multi-axis goniometer, the on-line viewing and collimation systems, as well as the bespoke equipment for sample transfer and storage (Figure 2). Each piece of equipment has been optimized to ensure that the best-quality long-wavelength data can be collected. The curved Pilatus 12M detector can collect to diffraction angles of 2θ = ±100°, resulting in sufficiently high-resolution diffraction data even at longest wavelengths (Figure 1, bottom). The 120 detector modules have been specifically selected for low-energy compatibility and calibrations for an additional ultra-high gain mode have been provided.

The lowest possible detector threshold is 1.8 keV, leading to increased corner and edge effects for energies lower than 3.6 keV and compromised data quality at the longest wavelengths, particularly for low mosaicity crystals, can be observed. This effect in combination with the decrease in the detector quantum efficiency¹⁵ needs to be taken into consideration when planning an experiment. The multi-axis goniometer enables reorientation of crystals to allow for data collection strategies that maximize the quality and strength of the anomalous signal, as well as the completeness of the anomalous data collected. Sample absorption is a limiting factor for the experiments, particularly at longest wavelengths. Absorption corrections, as implemented in commonly used MX processing software packages^16,17, are working well to wavelengths around 3 Å. Longer wavelengths will require analytical absorption corrections based on tomographic reconstructions¹⁸ or laser ablation to remove non-diffracting material and cut the crystals into well-defined shapes¹⁹. The latter will also assist in reducing the size of larger crystals as X-ray diffraction experiments at longer wavelengths are more efficient for smaller crystals¹⁴. The challenge of keeping samples at cryogenic temperatures during data collection is addressed by conductive cooling, as using open-flow cold gas stream devices is not compatible with a vacuum environment. Hence, thermally conductive materials, such as copper, are needed for connecting the sample to a pulse tube cryocooler. The stainless steel SPINE standard pins used throughout MX, as well as any other commercially available sample mounts, are not suitable for in-vacuum long-wavelength MX because of their poor thermal conductivity.

The sample holders (SHs) for in-vacuum MX must be an essential part of the heat removal thermal pathway (Figure 3A). As such, they consist of a thermally conductive copper body and pin and include two important features: a strong magnet base to ensure an adequate thermal link to the cold goniometer head, and a sample mount, made from polyimide, to minimize X-ray absorption and scattering²⁰. Efforts were made to ensure that the user experience of crystal-harvesting and flash-cooling is almost identical to that associated with standard MX practices. As the dedicated I23 SHs are not directly compatible with other synchrotron beamlines, a stainless steel adaptor is used for compatibility with the crystal-harvesting magnetic wands and existing goniometer interfaces on other MX beamlines (Figure 3B). The adaptor is also important for making use of the automation facilities on other Diamond MX beamlines, which are based on ALS-type robot gripper heads²¹ and unipuck-style base layouts²², if sample variation requires fast pre-screening for selection of the best diffracting crystals. The sample preparation and loading protocol can be broken down into two stages:

Stage 1: Harvesting crystals and flash-freezing performed by users in their own laboratories

Following assessment of the project suitability for I23 data collection, sample holders with loops matching the crystal sizes (pre-assembled with adaptors) are sent to user labs for crystal harvesting. To prevent any damage, SHs and adaptors should not be separated and are to be used as one unit for the purpose of fishing crystals with appropriately sized loops using standard crystal-harvesting magnetic wands. As is common in MX, this task is performed manually under the microscope, and crystals are immediately flash-cooled in a foam dewar with liquid nitrogen²³. Due to a mismatch of magnetic forces, the SHs are currently not compatible with unipucks. Storage and shipping are realized using combipucks (see the Table of Materials), which are available to users upon request, along with the compatible dry shipper inserts (Figure 3C). These pucks share the same base plate with the widely used unipucks and allow fast pre-screening of samples at other Diamond MX beamlines. Loaning this equipment to users is currently the best arrangement, until the bespoke sample holders are commercially available. Transport to the beamline requires the standard dry shippers used in the MX community.

Stage 2: Transfer of cryo-cooled samples into the vacuum endstation

Once the samples arrive on the beamline, they are prepared for transfer into the vacuum endstation. This involves removal of SHs from combipucks and separation from adaptors. Introducing biological samples to vacuum is routinely performed in the field of cryo-electron microscopy. Some of the well-established concepts were adapted for the I23 sample transfer. In short, SHs are transferred under liquid nitrogen onto transfer blocks (Figure 3D). These blocks have excellent thermal conductivity and a significant thermal mass, preventing the crystals from reaching the glass-transition temperature when in vacuum. Up to four blocks, with a capacity of four samples each, are loaded under liquid nitrogen into a block puck (Figure 3H), which is used either for transferring samples to the Cryogenic Transfer System (CTS) or for storage in liquid nitrogen dewars between experiments.  

The Cryogenic Transfer System developed at Diamond Light Source comprises of two sub-assemblies, the Sample Station and the Shuttle (Figure 4A). The Sample Station consists of a liquid nitrogen bath for temporary storage of protein crystals and has specific features to ensure safety and allow a user-friendly experience (Figure 5). The CTS is controlled by a programmable logic controller via a user-friendly touchscreen interface. The Sample Station has light-emitting diodes built in for better visualization and a set of heaters controlled in close-loop to automate the drying of the liquid nitrogen bath once the samples have been transferred. It also has a variety of sensors to ensure safety and efficient functioning of the system. The Sample Station has bespoke hardware to provide a reliable electrical interface to interact with the shuttle for operations, such as pumping down to rough vacuum for sample transfer, as well as monitoring of liquid nitrogen levels and the temperature inside the shuttle.

The Shuttle (Figure 6) is a portable device used to pick up a transfer block from the Sample Station liquid nitrogen bath and transfer it inside a cryogenic and vacuum environment to the endstation. It includes a liquid nitrogen dewar to keep the samples cold during transfer, liquid level monitoring in the dewar, and a variety of sensors for operation and user safety. The transfer arm is equipped with a magnetic drive and includes machined grooves to guide users in safely loading and unloading transfer blocks into the endstation. Transfer from the shuttle to the vacuum vessel is conducted via an airlock. The airlock is an interface for the shuttle on the endstation used to evacuate the interspace between the shuttle and the endstation, before opening the shuttle and endstation vacuum valves. The pumping and venting sequences are fully automated and can be operated via a large touchscreen with a user-friendly interface (Figure 4C). The current protocol is used to transfer a thaumatin crystal to the vacuum endstation for data collection.